diff options
Diffstat (limited to 'net/sunrpc')
60 files changed, 29559 insertions, 7698 deletions
diff --git a/net/sunrpc/Kconfig b/net/sunrpc/Kconfig new file mode 100644 index 00000000000..0754d0f466d --- /dev/null +++ b/net/sunrpc/Kconfig @@ -0,0 +1,72 @@ +config SUNRPC + tristate + +config SUNRPC_GSS + tristate + select OID_REGISTRY + +config SUNRPC_BACKCHANNEL + bool + depends on SUNRPC + +config SUNRPC_SWAP + bool + depends on SUNRPC + +config RPCSEC_GSS_KRB5 + tristate "Secure RPC: Kerberos V mechanism" + depends on SUNRPC && CRYPTO + depends on CRYPTO_MD5 && CRYPTO_DES && CRYPTO_CBC && CRYPTO_CTS + depends on CRYPTO_ECB && CRYPTO_HMAC && CRYPTO_SHA1 && CRYPTO_AES + depends on CRYPTO_ARC4 + default y + select SUNRPC_GSS + help + Choose Y here to enable Secure RPC using the Kerberos version 5 + GSS-API mechanism (RFC 1964). + + Secure RPC calls with Kerberos require an auxiliary user-space + daemon which may be found in the Linux nfs-utils package + available from http://linux-nfs.org/. In addition, user-space + Kerberos support should be installed. + + If unsure, say Y. + +config SUNRPC_DEBUG + bool "RPC: Enable dprintk debugging" + depends on SUNRPC && SYSCTL + help + This option enables a sysctl-based debugging interface + that is be used by the 'rpcdebug' utility to turn on or off + logging of different aspects of the kernel RPC activity. + + Disabling this option will make your kernel slightly smaller, + but makes troubleshooting NFS issues significantly harder. + + If unsure, say Y. + +config SUNRPC_XPRT_RDMA_CLIENT + tristate "RPC over RDMA Client Support" + depends on SUNRPC && INFINIBAND && INFINIBAND_ADDR_TRANS + default SUNRPC && INFINIBAND + help + This option allows the NFS client to support an RDMA-enabled + transport. + + To compile RPC client RDMA transport support as a module, + choose M here: the module will be called xprtrdma. + + If unsure, say N. + +config SUNRPC_XPRT_RDMA_SERVER + tristate "RPC over RDMA Server Support" + depends on SUNRPC && INFINIBAND && INFINIBAND_ADDR_TRANS + default SUNRPC && INFINIBAND + help + This option allows the NFS server to support an RDMA-enabled + transport. + + To compile RPC server RDMA transport support as a module, + choose M here: the module will be called svcrdma. + + If unsure, say N. diff --git a/net/sunrpc/Makefile b/net/sunrpc/Makefile index cdcab9ca4c6..e5a7a1cac8f 100644 --- a/net/sunrpc/Makefile +++ b/net/sunrpc/Makefile @@ -6,10 +6,14 @@ obj-$(CONFIG_SUNRPC) += sunrpc.o obj-$(CONFIG_SUNRPC_GSS) += auth_gss/ +obj-y += xprtrdma/ + sunrpc-y := clnt.o xprt.o socklib.o xprtsock.o sched.o \ - auth.o auth_null.o auth_unix.o \ + auth.o auth_null.o auth_unix.o auth_generic.o \ svc.o svcsock.o svcauth.o svcauth_unix.o \ - pmap_clnt.o timer.o xdr.o \ - sunrpc_syms.o cache.o rpc_pipe.o + addr.o rpcb_clnt.o timer.o xdr.o \ + sunrpc_syms.o cache.o rpc_pipe.o \ + svc_xprt.o +sunrpc-$(CONFIG_SUNRPC_BACKCHANNEL) += backchannel_rqst.o bc_svc.o sunrpc-$(CONFIG_PROC_FS) += stats.o sunrpc-$(CONFIG_SYSCTL) += sysctl.o diff --git a/net/sunrpc/addr.c b/net/sunrpc/addr.c new file mode 100644 index 00000000000..a622ad64acd --- /dev/null +++ b/net/sunrpc/addr.c @@ -0,0 +1,361 @@ +/* + * Copyright 2009, Oracle. All rights reserved. + * + * Convert socket addresses to presentation addresses and universal + * addresses, and vice versa. + * + * Universal addresses are introduced by RFC 1833 and further refined by + * recent RFCs describing NFSv4. The universal address format is part + * of the external (network) interface provided by rpcbind version 3 + * and 4, and by NFSv4. Such an address is a string containing a + * presentation format IP address followed by a port number in + * "hibyte.lobyte" format. + * + * IPv6 addresses can also include a scope ID, typically denoted by + * a '%' followed by a device name or a non-negative integer. Refer to + * RFC 4291, Section 2.2 for details on IPv6 presentation formats. + */ + +#include <net/ipv6.h> +#include <linux/sunrpc/addr.h> +#include <linux/sunrpc/msg_prot.h> +#include <linux/slab.h> +#include <linux/export.h> + +#if IS_ENABLED(CONFIG_IPV6) + +static size_t rpc_ntop6_noscopeid(const struct sockaddr *sap, + char *buf, const int buflen) +{ + const struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)sap; + const struct in6_addr *addr = &sin6->sin6_addr; + + /* + * RFC 4291, Section 2.2.2 + * + * Shorthanded ANY address + */ + if (ipv6_addr_any(addr)) + return snprintf(buf, buflen, "::"); + + /* + * RFC 4291, Section 2.2.2 + * + * Shorthanded loopback address + */ + if (ipv6_addr_loopback(addr)) + return snprintf(buf, buflen, "::1"); + + /* + * RFC 4291, Section 2.2.3 + * + * Special presentation address format for mapped v4 + * addresses. + */ + if (ipv6_addr_v4mapped(addr)) + return snprintf(buf, buflen, "::ffff:%pI4", + &addr->s6_addr32[3]); + + /* + * RFC 4291, Section 2.2.1 + */ + return snprintf(buf, buflen, "%pI6c", addr); +} + +static size_t rpc_ntop6(const struct sockaddr *sap, + char *buf, const size_t buflen) +{ + const struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)sap; + char scopebuf[IPV6_SCOPE_ID_LEN]; + size_t len; + int rc; + + len = rpc_ntop6_noscopeid(sap, buf, buflen); + if (unlikely(len == 0)) + return len; + + if (!(ipv6_addr_type(&sin6->sin6_addr) & IPV6_ADDR_LINKLOCAL)) + return len; + if (sin6->sin6_scope_id == 0) + return len; + + rc = snprintf(scopebuf, sizeof(scopebuf), "%c%u", + IPV6_SCOPE_DELIMITER, sin6->sin6_scope_id); + if (unlikely((size_t)rc > sizeof(scopebuf))) + return 0; + + len += rc; + if (unlikely(len > buflen)) + return 0; + + strcat(buf, scopebuf); + return len; +} + +#else /* !IS_ENABLED(CONFIG_IPV6) */ + +static size_t rpc_ntop6_noscopeid(const struct sockaddr *sap, + char *buf, const int buflen) +{ + return 0; +} + +static size_t rpc_ntop6(const struct sockaddr *sap, + char *buf, const size_t buflen) +{ + return 0; +} + +#endif /* !IS_ENABLED(CONFIG_IPV6) */ + +static int rpc_ntop4(const struct sockaddr *sap, + char *buf, const size_t buflen) +{ + const struct sockaddr_in *sin = (struct sockaddr_in *)sap; + + return snprintf(buf, buflen, "%pI4", &sin->sin_addr); +} + +/** + * rpc_ntop - construct a presentation address in @buf + * @sap: socket address + * @buf: construction area + * @buflen: size of @buf, in bytes + * + * Plants a %NUL-terminated string in @buf and returns the length + * of the string, excluding the %NUL. Otherwise zero is returned. + */ +size_t rpc_ntop(const struct sockaddr *sap, char *buf, const size_t buflen) +{ + switch (sap->sa_family) { + case AF_INET: + return rpc_ntop4(sap, buf, buflen); + case AF_INET6: + return rpc_ntop6(sap, buf, buflen); + } + + return 0; +} +EXPORT_SYMBOL_GPL(rpc_ntop); + +static size_t rpc_pton4(const char *buf, const size_t buflen, + struct sockaddr *sap, const size_t salen) +{ + struct sockaddr_in *sin = (struct sockaddr_in *)sap; + u8 *addr = (u8 *)&sin->sin_addr.s_addr; + + if (buflen > INET_ADDRSTRLEN || salen < sizeof(struct sockaddr_in)) + return 0; + + memset(sap, 0, sizeof(struct sockaddr_in)); + + if (in4_pton(buf, buflen, addr, '\0', NULL) == 0) + return 0; + + sin->sin_family = AF_INET; + return sizeof(struct sockaddr_in); +} + +#if IS_ENABLED(CONFIG_IPV6) +static int rpc_parse_scope_id(struct net *net, const char *buf, + const size_t buflen, const char *delim, + struct sockaddr_in6 *sin6) +{ + char *p; + size_t len; + + if ((buf + buflen) == delim) + return 1; + + if (*delim != IPV6_SCOPE_DELIMITER) + return 0; + + if (!(ipv6_addr_type(&sin6->sin6_addr) & IPV6_ADDR_LINKLOCAL)) + return 0; + + len = (buf + buflen) - delim - 1; + p = kstrndup(delim + 1, len, GFP_KERNEL); + if (p) { + unsigned long scope_id = 0; + struct net_device *dev; + + dev = dev_get_by_name(net, p); + if (dev != NULL) { + scope_id = dev->ifindex; + dev_put(dev); + } else { + if (strict_strtoul(p, 10, &scope_id) == 0) { + kfree(p); + return 0; + } + } + + kfree(p); + + sin6->sin6_scope_id = scope_id; + return 1; + } + + return 0; +} + +static size_t rpc_pton6(struct net *net, const char *buf, const size_t buflen, + struct sockaddr *sap, const size_t salen) +{ + struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)sap; + u8 *addr = (u8 *)&sin6->sin6_addr.in6_u; + const char *delim; + + if (buflen > (INET6_ADDRSTRLEN + IPV6_SCOPE_ID_LEN) || + salen < sizeof(struct sockaddr_in6)) + return 0; + + memset(sap, 0, sizeof(struct sockaddr_in6)); + + if (in6_pton(buf, buflen, addr, IPV6_SCOPE_DELIMITER, &delim) == 0) + return 0; + + if (!rpc_parse_scope_id(net, buf, buflen, delim, sin6)) + return 0; + + sin6->sin6_family = AF_INET6; + return sizeof(struct sockaddr_in6); +} +#else +static size_t rpc_pton6(struct net *net, const char *buf, const size_t buflen, + struct sockaddr *sap, const size_t salen) +{ + return 0; +} +#endif + +/** + * rpc_pton - Construct a sockaddr in @sap + * @net: applicable network namespace + * @buf: C string containing presentation format IP address + * @buflen: length of presentation address in bytes + * @sap: buffer into which to plant socket address + * @salen: size of buffer in bytes + * + * Returns the size of the socket address if successful; otherwise + * zero is returned. + * + * Plants a socket address in @sap and returns the size of the + * socket address, if successful. Returns zero if an error + * occurred. + */ +size_t rpc_pton(struct net *net, const char *buf, const size_t buflen, + struct sockaddr *sap, const size_t salen) +{ + unsigned int i; + + for (i = 0; i < buflen; i++) + if (buf[i] == ':') + return rpc_pton6(net, buf, buflen, sap, salen); + return rpc_pton4(buf, buflen, sap, salen); +} +EXPORT_SYMBOL_GPL(rpc_pton); + +/** + * rpc_sockaddr2uaddr - Construct a universal address string from @sap. + * @sap: socket address + * @gfp_flags: allocation mode + * + * Returns a %NUL-terminated string in dynamically allocated memory; + * otherwise NULL is returned if an error occurred. Caller must + * free the returned string. + */ +char *rpc_sockaddr2uaddr(const struct sockaddr *sap, gfp_t gfp_flags) +{ + char portbuf[RPCBIND_MAXUADDRPLEN]; + char addrbuf[RPCBIND_MAXUADDRLEN]; + unsigned short port; + + switch (sap->sa_family) { + case AF_INET: + if (rpc_ntop4(sap, addrbuf, sizeof(addrbuf)) == 0) + return NULL; + port = ntohs(((struct sockaddr_in *)sap)->sin_port); + break; + case AF_INET6: + if (rpc_ntop6_noscopeid(sap, addrbuf, sizeof(addrbuf)) == 0) + return NULL; + port = ntohs(((struct sockaddr_in6 *)sap)->sin6_port); + break; + default: + return NULL; + } + + if (snprintf(portbuf, sizeof(portbuf), + ".%u.%u", port >> 8, port & 0xff) > (int)sizeof(portbuf)) + return NULL; + + if (strlcat(addrbuf, portbuf, sizeof(addrbuf)) > sizeof(addrbuf)) + return NULL; + + return kstrdup(addrbuf, gfp_flags); +} + +/** + * rpc_uaddr2sockaddr - convert a universal address to a socket address. + * @net: applicable network namespace + * @uaddr: C string containing universal address to convert + * @uaddr_len: length of universal address string + * @sap: buffer into which to plant socket address + * @salen: size of buffer + * + * @uaddr does not have to be '\0'-terminated, but strict_strtoul() and + * rpc_pton() require proper string termination to be successful. + * + * Returns the size of the socket address if successful; otherwise + * zero is returned. + */ +size_t rpc_uaddr2sockaddr(struct net *net, const char *uaddr, + const size_t uaddr_len, struct sockaddr *sap, + const size_t salen) +{ + char *c, buf[RPCBIND_MAXUADDRLEN + sizeof('\0')]; + unsigned long portlo, porthi; + unsigned short port; + + if (uaddr_len > RPCBIND_MAXUADDRLEN) + return 0; + + memcpy(buf, uaddr, uaddr_len); + + buf[uaddr_len] = '\0'; + c = strrchr(buf, '.'); + if (unlikely(c == NULL)) + return 0; + if (unlikely(strict_strtoul(c + 1, 10, &portlo) != 0)) + return 0; + if (unlikely(portlo > 255)) + return 0; + + *c = '\0'; + c = strrchr(buf, '.'); + if (unlikely(c == NULL)) + return 0; + if (unlikely(strict_strtoul(c + 1, 10, &porthi) != 0)) + return 0; + if (unlikely(porthi > 255)) + return 0; + + port = (unsigned short)((porthi << 8) | portlo); + + *c = '\0'; + if (rpc_pton(net, buf, strlen(buf), sap, salen) == 0) + return 0; + + switch (sap->sa_family) { + case AF_INET: + ((struct sockaddr_in *)sap)->sin_port = htons(port); + return sizeof(struct sockaddr_in); + case AF_INET6: + ((struct sockaddr_in6 *)sap)->sin6_port = htons(port); + return sizeof(struct sockaddr_in6); + } + + return 0; +} +EXPORT_SYMBOL_GPL(rpc_uaddr2sockaddr); diff --git a/net/sunrpc/auth.c b/net/sunrpc/auth.c index 8c7756036e9..f7736671742 100644 --- a/net/sunrpc/auth.c +++ b/net/sunrpc/auth.c @@ -11,72 +11,279 @@ #include <linux/module.h> #include <linux/slab.h> #include <linux/errno.h> +#include <linux/hash.h> #include <linux/sunrpc/clnt.h> +#include <linux/sunrpc/gss_api.h> #include <linux/spinlock.h> #ifdef RPC_DEBUG # define RPCDBG_FACILITY RPCDBG_AUTH #endif -static struct rpc_authops * auth_flavors[RPC_AUTH_MAXFLAVOR] = { +#define RPC_CREDCACHE_DEFAULT_HASHBITS (4) +struct rpc_cred_cache { + struct hlist_head *hashtable; + unsigned int hashbits; + spinlock_t lock; +}; + +static unsigned int auth_hashbits = RPC_CREDCACHE_DEFAULT_HASHBITS; + +static DEFINE_SPINLOCK(rpc_authflavor_lock); +static const struct rpc_authops *auth_flavors[RPC_AUTH_MAXFLAVOR] = { &authnull_ops, /* AUTH_NULL */ &authunix_ops, /* AUTH_UNIX */ NULL, /* others can be loadable modules */ }; +static LIST_HEAD(cred_unused); +static unsigned long number_cred_unused; + +#define MAX_HASHTABLE_BITS (14) +static int param_set_hashtbl_sz(const char *val, const struct kernel_param *kp) +{ + unsigned long num; + unsigned int nbits; + int ret; + + if (!val) + goto out_inval; + ret = strict_strtoul(val, 0, &num); + if (ret == -EINVAL) + goto out_inval; + nbits = fls(num); + if (num > (1U << nbits)) + nbits++; + if (nbits > MAX_HASHTABLE_BITS || nbits < 2) + goto out_inval; + *(unsigned int *)kp->arg = nbits; + return 0; +out_inval: + return -EINVAL; +} + +static int param_get_hashtbl_sz(char *buffer, const struct kernel_param *kp) +{ + unsigned int nbits; + + nbits = *(unsigned int *)kp->arg; + return sprintf(buffer, "%u", 1U << nbits); +} + +#define param_check_hashtbl_sz(name, p) __param_check(name, p, unsigned int); + +static struct kernel_param_ops param_ops_hashtbl_sz = { + .set = param_set_hashtbl_sz, + .get = param_get_hashtbl_sz, +}; + +module_param_named(auth_hashtable_size, auth_hashbits, hashtbl_sz, 0644); +MODULE_PARM_DESC(auth_hashtable_size, "RPC credential cache hashtable size"); + static u32 pseudoflavor_to_flavor(u32 flavor) { - if (flavor >= RPC_AUTH_MAXFLAVOR) + if (flavor > RPC_AUTH_MAXFLAVOR) return RPC_AUTH_GSS; return flavor; } int -rpcauth_register(struct rpc_authops *ops) +rpcauth_register(const struct rpc_authops *ops) { rpc_authflavor_t flavor; + int ret = -EPERM; if ((flavor = ops->au_flavor) >= RPC_AUTH_MAXFLAVOR) return -EINVAL; - if (auth_flavors[flavor] != NULL) - return -EPERM; /* what else? */ - auth_flavors[flavor] = ops; - return 0; + spin_lock(&rpc_authflavor_lock); + if (auth_flavors[flavor] == NULL) { + auth_flavors[flavor] = ops; + ret = 0; + } + spin_unlock(&rpc_authflavor_lock); + return ret; } +EXPORT_SYMBOL_GPL(rpcauth_register); int -rpcauth_unregister(struct rpc_authops *ops) +rpcauth_unregister(const struct rpc_authops *ops) { rpc_authflavor_t flavor; + int ret = -EPERM; if ((flavor = ops->au_flavor) >= RPC_AUTH_MAXFLAVOR) return -EINVAL; - if (auth_flavors[flavor] != ops) - return -EPERM; /* what else? */ - auth_flavors[flavor] = NULL; - return 0; + spin_lock(&rpc_authflavor_lock); + if (auth_flavors[flavor] == ops) { + auth_flavors[flavor] = NULL; + ret = 0; + } + spin_unlock(&rpc_authflavor_lock); + return ret; +} +EXPORT_SYMBOL_GPL(rpcauth_unregister); + +/** + * rpcauth_get_pseudoflavor - check if security flavor is supported + * @flavor: a security flavor + * @info: a GSS mech OID, quality of protection, and service value + * + * Verifies that an appropriate kernel module is available or already loaded. + * Returns an equivalent pseudoflavor, or RPC_AUTH_MAXFLAVOR if "flavor" is + * not supported locally. + */ +rpc_authflavor_t +rpcauth_get_pseudoflavor(rpc_authflavor_t flavor, struct rpcsec_gss_info *info) +{ + const struct rpc_authops *ops; + rpc_authflavor_t pseudoflavor; + + ops = auth_flavors[flavor]; + if (ops == NULL) + request_module("rpc-auth-%u", flavor); + spin_lock(&rpc_authflavor_lock); + ops = auth_flavors[flavor]; + if (ops == NULL || !try_module_get(ops->owner)) { + spin_unlock(&rpc_authflavor_lock); + return RPC_AUTH_MAXFLAVOR; + } + spin_unlock(&rpc_authflavor_lock); + + pseudoflavor = flavor; + if (ops->info2flavor != NULL) + pseudoflavor = ops->info2flavor(info); + + module_put(ops->owner); + return pseudoflavor; +} +EXPORT_SYMBOL_GPL(rpcauth_get_pseudoflavor); + +/** + * rpcauth_get_gssinfo - find GSS tuple matching a GSS pseudoflavor + * @pseudoflavor: GSS pseudoflavor to match + * @info: rpcsec_gss_info structure to fill in + * + * Returns zero and fills in "info" if pseudoflavor matches a + * supported mechanism. + */ +int +rpcauth_get_gssinfo(rpc_authflavor_t pseudoflavor, struct rpcsec_gss_info *info) +{ + rpc_authflavor_t flavor = pseudoflavor_to_flavor(pseudoflavor); + const struct rpc_authops *ops; + int result; + + if (flavor >= RPC_AUTH_MAXFLAVOR) + return -EINVAL; + + ops = auth_flavors[flavor]; + if (ops == NULL) + request_module("rpc-auth-%u", flavor); + spin_lock(&rpc_authflavor_lock); + ops = auth_flavors[flavor]; + if (ops == NULL || !try_module_get(ops->owner)) { + spin_unlock(&rpc_authflavor_lock); + return -ENOENT; + } + spin_unlock(&rpc_authflavor_lock); + + result = -ENOENT; + if (ops->flavor2info != NULL) + result = ops->flavor2info(pseudoflavor, info); + + module_put(ops->owner); + return result; +} +EXPORT_SYMBOL_GPL(rpcauth_get_gssinfo); + +/** + * rpcauth_list_flavors - discover registered flavors and pseudoflavors + * @array: array to fill in + * @size: size of "array" + * + * Returns the number of array items filled in, or a negative errno. + * + * The returned array is not sorted by any policy. Callers should not + * rely on the order of the items in the returned array. + */ +int +rpcauth_list_flavors(rpc_authflavor_t *array, int size) +{ + rpc_authflavor_t flavor; + int result = 0; + + spin_lock(&rpc_authflavor_lock); + for (flavor = 0; flavor < RPC_AUTH_MAXFLAVOR; flavor++) { + const struct rpc_authops *ops = auth_flavors[flavor]; + rpc_authflavor_t pseudos[4]; + int i, len; + + if (result >= size) { + result = -ENOMEM; + break; + } + + if (ops == NULL) + continue; + if (ops->list_pseudoflavors == NULL) { + array[result++] = ops->au_flavor; + continue; + } + len = ops->list_pseudoflavors(pseudos, ARRAY_SIZE(pseudos)); + if (len < 0) { + result = len; + break; + } + for (i = 0; i < len; i++) { + if (result >= size) { + result = -ENOMEM; + break; + } + array[result++] = pseudos[i]; + } + } + spin_unlock(&rpc_authflavor_lock); + + dprintk("RPC: %s returns %d\n", __func__, result); + return result; } +EXPORT_SYMBOL_GPL(rpcauth_list_flavors); struct rpc_auth * -rpcauth_create(rpc_authflavor_t pseudoflavor, struct rpc_clnt *clnt) +rpcauth_create(struct rpc_auth_create_args *args, struct rpc_clnt *clnt) { struct rpc_auth *auth; - struct rpc_authops *ops; - u32 flavor = pseudoflavor_to_flavor(pseudoflavor); + const struct rpc_authops *ops; + u32 flavor = pseudoflavor_to_flavor(args->pseudoflavor); - if (flavor >= RPC_AUTH_MAXFLAVOR || !(ops = auth_flavors[flavor])) - return ERR_PTR(-EINVAL); - auth = ops->create(clnt, pseudoflavor); + auth = ERR_PTR(-EINVAL); + if (flavor >= RPC_AUTH_MAXFLAVOR) + goto out; + + if ((ops = auth_flavors[flavor]) == NULL) + request_module("rpc-auth-%u", flavor); + spin_lock(&rpc_authflavor_lock); + ops = auth_flavors[flavor]; + if (ops == NULL || !try_module_get(ops->owner)) { + spin_unlock(&rpc_authflavor_lock); + goto out; + } + spin_unlock(&rpc_authflavor_lock); + auth = ops->create(args, clnt); + module_put(ops->owner); if (IS_ERR(auth)) return auth; if (clnt->cl_auth) - rpcauth_destroy(clnt->cl_auth); + rpcauth_release(clnt->cl_auth); clnt->cl_auth = auth; + +out: return auth; } +EXPORT_SYMBOL_GPL(rpcauth_create); void -rpcauth_destroy(struct rpc_auth *auth) +rpcauth_release(struct rpc_auth *auth) { if (!atomic_dec_and_test(&auth->au_count)) return; @@ -85,37 +292,88 @@ rpcauth_destroy(struct rpc_auth *auth) static DEFINE_SPINLOCK(rpc_credcache_lock); +static void +rpcauth_unhash_cred_locked(struct rpc_cred *cred) +{ + hlist_del_rcu(&cred->cr_hash); + smp_mb__before_atomic(); + clear_bit(RPCAUTH_CRED_HASHED, &cred->cr_flags); +} + +static int +rpcauth_unhash_cred(struct rpc_cred *cred) +{ + spinlock_t *cache_lock; + int ret; + + cache_lock = &cred->cr_auth->au_credcache->lock; + spin_lock(cache_lock); + ret = atomic_read(&cred->cr_count) == 0; + if (ret) + rpcauth_unhash_cred_locked(cred); + spin_unlock(cache_lock); + return ret; +} + /* * Initialize RPC credential cache */ int -rpcauth_init_credcache(struct rpc_auth *auth, unsigned long expire) +rpcauth_init_credcache(struct rpc_auth *auth) { struct rpc_cred_cache *new; - int i; + unsigned int hashsize; - new = (struct rpc_cred_cache *)kmalloc(sizeof(*new), GFP_KERNEL); + new = kmalloc(sizeof(*new), GFP_KERNEL); if (!new) - return -ENOMEM; - for (i = 0; i < RPC_CREDCACHE_NR; i++) - INIT_HLIST_HEAD(&new->hashtable[i]); - new->expire = expire; - new->nextgc = jiffies + (expire >> 1); + goto out_nocache; + new->hashbits = auth_hashbits; + hashsize = 1U << new->hashbits; + new->hashtable = kcalloc(hashsize, sizeof(new->hashtable[0]), GFP_KERNEL); + if (!new->hashtable) + goto out_nohashtbl; + spin_lock_init(&new->lock); auth->au_credcache = new; return 0; +out_nohashtbl: + kfree(new); +out_nocache: + return -ENOMEM; } +EXPORT_SYMBOL_GPL(rpcauth_init_credcache); + +/* + * Setup a credential key lifetime timeout notification + */ +int +rpcauth_key_timeout_notify(struct rpc_auth *auth, struct rpc_cred *cred) +{ + if (!cred->cr_auth->au_ops->key_timeout) + return 0; + return cred->cr_auth->au_ops->key_timeout(auth, cred); +} +EXPORT_SYMBOL_GPL(rpcauth_key_timeout_notify); + +bool +rpcauth_cred_key_to_expire(struct rpc_cred *cred) +{ + if (!cred->cr_ops->crkey_to_expire) + return false; + return cred->cr_ops->crkey_to_expire(cred); +} +EXPORT_SYMBOL_GPL(rpcauth_cred_key_to_expire); /* * Destroy a list of credentials */ static inline -void rpcauth_destroy_credlist(struct hlist_head *head) +void rpcauth_destroy_credlist(struct list_head *head) { struct rpc_cred *cred; - while (!hlist_empty(head)) { - cred = hlist_entry(head->first, struct rpc_cred, cr_hash); - hlist_del_init(&cred->cr_hash); + while (!list_empty(head)) { + cred = list_entry(head->next, struct rpc_cred, cr_lru); + list_del_init(&cred->cr_lru); put_rpccred(cred); } } @@ -125,58 +383,125 @@ void rpcauth_destroy_credlist(struct hlist_head *head) * that are not referenced. */ void -rpcauth_free_credcache(struct rpc_auth *auth) +rpcauth_clear_credcache(struct rpc_cred_cache *cache) { - struct rpc_cred_cache *cache = auth->au_credcache; - HLIST_HEAD(free); - struct hlist_node *pos, *next; + LIST_HEAD(free); + struct hlist_head *head; struct rpc_cred *cred; + unsigned int hashsize = 1U << cache->hashbits; int i; spin_lock(&rpc_credcache_lock); - for (i = 0; i < RPC_CREDCACHE_NR; i++) { - hlist_for_each_safe(pos, next, &cache->hashtable[i]) { - cred = hlist_entry(pos, struct rpc_cred, cr_hash); - __hlist_del(&cred->cr_hash); - hlist_add_head(&cred->cr_hash, &free); + spin_lock(&cache->lock); + for (i = 0; i < hashsize; i++) { + head = &cache->hashtable[i]; + while (!hlist_empty(head)) { + cred = hlist_entry(head->first, struct rpc_cred, cr_hash); + get_rpccred(cred); + if (!list_empty(&cred->cr_lru)) { + list_del(&cred->cr_lru); + number_cred_unused--; + } + list_add_tail(&cred->cr_lru, &free); + rpcauth_unhash_cred_locked(cred); } } + spin_unlock(&cache->lock); spin_unlock(&rpc_credcache_lock); rpcauth_destroy_credlist(&free); } -static void -rpcauth_prune_expired(struct rpc_auth *auth, struct rpc_cred *cred, struct hlist_head *free) -{ - if (atomic_read(&cred->cr_count) != 1) - return; - if (time_after(jiffies, cred->cr_expire + auth->au_credcache->expire)) - cred->cr_flags &= ~RPCAUTH_CRED_UPTODATE; - if (!(cred->cr_flags & RPCAUTH_CRED_UPTODATE)) { - __hlist_del(&cred->cr_hash); - hlist_add_head(&cred->cr_hash, free); +/* + * Destroy the RPC credential cache + */ +void +rpcauth_destroy_credcache(struct rpc_auth *auth) +{ + struct rpc_cred_cache *cache = auth->au_credcache; + + if (cache) { + auth->au_credcache = NULL; + rpcauth_clear_credcache(cache); + kfree(cache->hashtable); + kfree(cache); } } +EXPORT_SYMBOL_GPL(rpcauth_destroy_credcache); + + +#define RPC_AUTH_EXPIRY_MORATORIUM (60 * HZ) /* * Remove stale credentials. Avoid sleeping inside the loop. */ -static void -rpcauth_gc_credcache(struct rpc_auth *auth, struct hlist_head *free) +static long +rpcauth_prune_expired(struct list_head *free, int nr_to_scan) { - struct rpc_cred_cache *cache = auth->au_credcache; - struct hlist_node *pos, *next; - struct rpc_cred *cred; - int i; + spinlock_t *cache_lock; + struct rpc_cred *cred, *next; + unsigned long expired = jiffies - RPC_AUTH_EXPIRY_MORATORIUM; + long freed = 0; + + list_for_each_entry_safe(cred, next, &cred_unused, cr_lru) { - dprintk("RPC: gc'ing RPC credentials for auth %p\n", auth); - for (i = 0; i < RPC_CREDCACHE_NR; i++) { - hlist_for_each_safe(pos, next, &cache->hashtable[i]) { - cred = hlist_entry(pos, struct rpc_cred, cr_hash); - rpcauth_prune_expired(auth, cred, free); + if (nr_to_scan-- == 0) + break; + /* + * Enforce a 60 second garbage collection moratorium + * Note that the cred_unused list must be time-ordered. + */ + if (time_in_range(cred->cr_expire, expired, jiffies) && + test_bit(RPCAUTH_CRED_HASHED, &cred->cr_flags) != 0) + break; + + list_del_init(&cred->cr_lru); + number_cred_unused--; + freed++; + if (atomic_read(&cred->cr_count) != 0) + continue; + + cache_lock = &cred->cr_auth->au_credcache->lock; + spin_lock(cache_lock); + if (atomic_read(&cred->cr_count) == 0) { + get_rpccred(cred); + list_add_tail(&cred->cr_lru, free); + rpcauth_unhash_cred_locked(cred); } + spin_unlock(cache_lock); } - cache->nextgc = jiffies + cache->expire; + return freed; +} + +/* + * Run memory cache shrinker. + */ +static unsigned long +rpcauth_cache_shrink_scan(struct shrinker *shrink, struct shrink_control *sc) + +{ + LIST_HEAD(free); + unsigned long freed; + + if ((sc->gfp_mask & GFP_KERNEL) != GFP_KERNEL) + return SHRINK_STOP; + + /* nothing left, don't come back */ + if (list_empty(&cred_unused)) + return SHRINK_STOP; + + spin_lock(&rpc_credcache_lock); + freed = rpcauth_prune_expired(&free, sc->nr_to_scan); + spin_unlock(&rpc_credcache_lock); + rpcauth_destroy_credlist(&free); + + return freed; +} + +static unsigned long +rpcauth_cache_shrink_count(struct shrinker *shrink, struct shrink_control *sc) + +{ + return (number_cred_unused / 100) * sysctl_vfs_cache_pressure; } /* @@ -184,191 +509,289 @@ rpcauth_gc_credcache(struct rpc_auth *auth, struct hlist_head *free) */ struct rpc_cred * rpcauth_lookup_credcache(struct rpc_auth *auth, struct auth_cred * acred, - int taskflags) + int flags) { + LIST_HEAD(free); struct rpc_cred_cache *cache = auth->au_credcache; - HLIST_HEAD(free); - struct hlist_node *pos, *next; - struct rpc_cred *new = NULL, - *cred = NULL; - int nr = 0; - - if (!(taskflags & RPC_TASK_ROOTCREDS)) - nr = acred->uid & RPC_CREDCACHE_MASK; -retry: - spin_lock(&rpc_credcache_lock); - if (time_before(cache->nextgc, jiffies)) - rpcauth_gc_credcache(auth, &free); - hlist_for_each_safe(pos, next, &cache->hashtable[nr]) { - struct rpc_cred *entry; - entry = hlist_entry(pos, struct rpc_cred, cr_hash); - if (entry->cr_ops->crmatch(acred, entry, taskflags)) { - hlist_del(&entry->cr_hash); - cred = entry; - break; + struct rpc_cred *cred = NULL, + *entry, *new; + unsigned int nr; + + nr = hash_long(from_kuid(&init_user_ns, acred->uid), cache->hashbits); + + rcu_read_lock(); + hlist_for_each_entry_rcu(entry, &cache->hashtable[nr], cr_hash) { + if (!entry->cr_ops->crmatch(acred, entry, flags)) + continue; + spin_lock(&cache->lock); + if (test_bit(RPCAUTH_CRED_HASHED, &entry->cr_flags) == 0) { + spin_unlock(&cache->lock); + continue; } - rpcauth_prune_expired(auth, entry, &free); + cred = get_rpccred(entry); + spin_unlock(&cache->lock); + break; } - if (new) { - if (cred) - hlist_add_head(&new->cr_hash, &free); - else - cred = new; - } - if (cred) { - hlist_add_head(&cred->cr_hash, &cache->hashtable[nr]); - get_rpccred(cred); - } - spin_unlock(&rpc_credcache_lock); + rcu_read_unlock(); - rpcauth_destroy_credlist(&free); + if (cred != NULL) + goto found; - if (!cred) { - new = auth->au_ops->crcreate(auth, acred, taskflags); - if (!IS_ERR(new)) { -#ifdef RPC_DEBUG - new->cr_magic = RPCAUTH_CRED_MAGIC; -#endif - goto retry; - } else - cred = new; + new = auth->au_ops->crcreate(auth, acred, flags); + if (IS_ERR(new)) { + cred = new; + goto out; } - return (struct rpc_cred *) cred; + spin_lock(&cache->lock); + hlist_for_each_entry(entry, &cache->hashtable[nr], cr_hash) { + if (!entry->cr_ops->crmatch(acred, entry, flags)) + continue; + cred = get_rpccred(entry); + break; + } + if (cred == NULL) { + cred = new; + set_bit(RPCAUTH_CRED_HASHED, &cred->cr_flags); + hlist_add_head_rcu(&cred->cr_hash, &cache->hashtable[nr]); + } else + list_add_tail(&new->cr_lru, &free); + spin_unlock(&cache->lock); +found: + if (test_bit(RPCAUTH_CRED_NEW, &cred->cr_flags) && + cred->cr_ops->cr_init != NULL && + !(flags & RPCAUTH_LOOKUP_NEW)) { + int res = cred->cr_ops->cr_init(auth, cred); + if (res < 0) { + put_rpccred(cred); + cred = ERR_PTR(res); + } + } + rpcauth_destroy_credlist(&free); +out: + return cred; } +EXPORT_SYMBOL_GPL(rpcauth_lookup_credcache); struct rpc_cred * -rpcauth_lookupcred(struct rpc_auth *auth, int taskflags) +rpcauth_lookupcred(struct rpc_auth *auth, int flags) { - struct auth_cred acred = { - .uid = current->fsuid, - .gid = current->fsgid, - .group_info = current->group_info, - }; + struct auth_cred acred; struct rpc_cred *ret; + const struct cred *cred = current_cred(); - dprintk("RPC: looking up %s cred\n", + dprintk("RPC: looking up %s cred\n", auth->au_ops->au_name); - get_group_info(acred.group_info); - ret = auth->au_ops->lookup_cred(auth, &acred, taskflags); + + memset(&acred, 0, sizeof(acred)); + acred.uid = cred->fsuid; + acred.gid = cred->fsgid; + acred.group_info = get_group_info(((struct cred *)cred)->group_info); + + ret = auth->au_ops->lookup_cred(auth, &acred, flags); put_group_info(acred.group_info); return ret; } +EXPORT_SYMBOL_GPL(rpcauth_lookupcred); + +void +rpcauth_init_cred(struct rpc_cred *cred, const struct auth_cred *acred, + struct rpc_auth *auth, const struct rpc_credops *ops) +{ + INIT_HLIST_NODE(&cred->cr_hash); + INIT_LIST_HEAD(&cred->cr_lru); + atomic_set(&cred->cr_count, 1); + cred->cr_auth = auth; + cred->cr_ops = ops; + cred->cr_expire = jiffies; +#ifdef RPC_DEBUG + cred->cr_magic = RPCAUTH_CRED_MAGIC; +#endif + cred->cr_uid = acred->uid; +} +EXPORT_SYMBOL_GPL(rpcauth_init_cred); struct rpc_cred * -rpcauth_bindcred(struct rpc_task *task) +rpcauth_generic_bind_cred(struct rpc_task *task, struct rpc_cred *cred, int lookupflags) { - struct rpc_auth *auth = task->tk_auth; + dprintk("RPC: %5u holding %s cred %p\n", task->tk_pid, + cred->cr_auth->au_ops->au_name, cred); + return get_rpccred(cred); +} +EXPORT_SYMBOL_GPL(rpcauth_generic_bind_cred); + +static struct rpc_cred * +rpcauth_bind_root_cred(struct rpc_task *task, int lookupflags) +{ + struct rpc_auth *auth = task->tk_client->cl_auth; struct auth_cred acred = { - .uid = current->fsuid, - .gid = current->fsgid, - .group_info = current->group_info, + .uid = GLOBAL_ROOT_UID, + .gid = GLOBAL_ROOT_GID, }; - struct rpc_cred *ret; - dprintk("RPC: %4d looking up %s cred\n", - task->tk_pid, task->tk_auth->au_ops->au_name); - get_group_info(acred.group_info); - ret = auth->au_ops->lookup_cred(auth, &acred, task->tk_flags); - if (!IS_ERR(ret)) - task->tk_msg.rpc_cred = ret; - else - task->tk_status = PTR_ERR(ret); - put_group_info(acred.group_info); - return ret; + dprintk("RPC: %5u looking up %s cred\n", + task->tk_pid, task->tk_client->cl_auth->au_ops->au_name); + return auth->au_ops->lookup_cred(auth, &acred, lookupflags); } -void -rpcauth_holdcred(struct rpc_task *task) +static struct rpc_cred * +rpcauth_bind_new_cred(struct rpc_task *task, int lookupflags) { - dprintk("RPC: %4d holding %s cred %p\n", - task->tk_pid, task->tk_auth->au_ops->au_name, task->tk_msg.rpc_cred); - if (task->tk_msg.rpc_cred) - get_rpccred(task->tk_msg.rpc_cred); + struct rpc_auth *auth = task->tk_client->cl_auth; + + dprintk("RPC: %5u looking up %s cred\n", + task->tk_pid, auth->au_ops->au_name); + return rpcauth_lookupcred(auth, lookupflags); +} + +static int +rpcauth_bindcred(struct rpc_task *task, struct rpc_cred *cred, int flags) +{ + struct rpc_rqst *req = task->tk_rqstp; + struct rpc_cred *new; + int lookupflags = 0; + + if (flags & RPC_TASK_ASYNC) + lookupflags |= RPCAUTH_LOOKUP_NEW; + if (cred != NULL) + new = cred->cr_ops->crbind(task, cred, lookupflags); + else if (flags & RPC_TASK_ROOTCREDS) + new = rpcauth_bind_root_cred(task, lookupflags); + else + new = rpcauth_bind_new_cred(task, lookupflags); + if (IS_ERR(new)) + return PTR_ERR(new); + if (req->rq_cred != NULL) + put_rpccred(req->rq_cred); + req->rq_cred = new; + return 0; } void put_rpccred(struct rpc_cred *cred) { - cred->cr_expire = jiffies; - if (!atomic_dec_and_test(&cred->cr_count)) + /* Fast path for unhashed credentials */ + if (test_bit(RPCAUTH_CRED_HASHED, &cred->cr_flags) == 0) { + if (atomic_dec_and_test(&cred->cr_count)) + cred->cr_ops->crdestroy(cred); + return; + } + + if (!atomic_dec_and_lock(&cred->cr_count, &rpc_credcache_lock)) return; + if (!list_empty(&cred->cr_lru)) { + number_cred_unused--; + list_del_init(&cred->cr_lru); + } + if (test_bit(RPCAUTH_CRED_HASHED, &cred->cr_flags) != 0) { + if (test_bit(RPCAUTH_CRED_UPTODATE, &cred->cr_flags) != 0) { + cred->cr_expire = jiffies; + list_add_tail(&cred->cr_lru, &cred_unused); + number_cred_unused++; + goto out_nodestroy; + } + if (!rpcauth_unhash_cred(cred)) { + /* We were hashed and someone looked us up... */ + goto out_nodestroy; + } + } + spin_unlock(&rpc_credcache_lock); cred->cr_ops->crdestroy(cred); + return; +out_nodestroy: + spin_unlock(&rpc_credcache_lock); } +EXPORT_SYMBOL_GPL(put_rpccred); -void -rpcauth_unbindcred(struct rpc_task *task) +__be32 * +rpcauth_marshcred(struct rpc_task *task, __be32 *p) { - struct rpc_cred *cred = task->tk_msg.rpc_cred; + struct rpc_cred *cred = task->tk_rqstp->rq_cred; - dprintk("RPC: %4d releasing %s cred %p\n", - task->tk_pid, task->tk_auth->au_ops->au_name, cred); + dprintk("RPC: %5u marshaling %s cred %p\n", + task->tk_pid, cred->cr_auth->au_ops->au_name, cred); - put_rpccred(cred); - task->tk_msg.rpc_cred = NULL; + return cred->cr_ops->crmarshal(task, p); } -u32 * -rpcauth_marshcred(struct rpc_task *task, u32 *p) +__be32 * +rpcauth_checkverf(struct rpc_task *task, __be32 *p) { - struct rpc_cred *cred = task->tk_msg.rpc_cred; + struct rpc_cred *cred = task->tk_rqstp->rq_cred; - dprintk("RPC: %4d marshaling %s cred %p\n", - task->tk_pid, task->tk_auth->au_ops->au_name, cred); + dprintk("RPC: %5u validating %s cred %p\n", + task->tk_pid, cred->cr_auth->au_ops->au_name, cred); - return cred->cr_ops->crmarshal(task, p); + return cred->cr_ops->crvalidate(task, p); } -u32 * -rpcauth_checkverf(struct rpc_task *task, u32 *p) +static void rpcauth_wrap_req_encode(kxdreproc_t encode, struct rpc_rqst *rqstp, + __be32 *data, void *obj) { - struct rpc_cred *cred = task->tk_msg.rpc_cred; + struct xdr_stream xdr; - dprintk("RPC: %4d validating %s cred %p\n", - task->tk_pid, task->tk_auth->au_ops->au_name, cred); - - return cred->cr_ops->crvalidate(task, p); + xdr_init_encode(&xdr, &rqstp->rq_snd_buf, data); + encode(rqstp, &xdr, obj); } int -rpcauth_wrap_req(struct rpc_task *task, kxdrproc_t encode, void *rqstp, - u32 *data, void *obj) +rpcauth_wrap_req(struct rpc_task *task, kxdreproc_t encode, void *rqstp, + __be32 *data, void *obj) { - struct rpc_cred *cred = task->tk_msg.rpc_cred; + struct rpc_cred *cred = task->tk_rqstp->rq_cred; - dprintk("RPC: %4d using %s cred %p to wrap rpc data\n", + dprintk("RPC: %5u using %s cred %p to wrap rpc data\n", task->tk_pid, cred->cr_ops->cr_name, cred); if (cred->cr_ops->crwrap_req) return cred->cr_ops->crwrap_req(task, encode, rqstp, data, obj); /* By default, we encode the arguments normally. */ - return encode(rqstp, data, obj); + rpcauth_wrap_req_encode(encode, rqstp, data, obj); + return 0; +} + +static int +rpcauth_unwrap_req_decode(kxdrdproc_t decode, struct rpc_rqst *rqstp, + __be32 *data, void *obj) +{ + struct xdr_stream xdr; + + xdr_init_decode(&xdr, &rqstp->rq_rcv_buf, data); + return decode(rqstp, &xdr, obj); } int -rpcauth_unwrap_resp(struct rpc_task *task, kxdrproc_t decode, void *rqstp, - u32 *data, void *obj) +rpcauth_unwrap_resp(struct rpc_task *task, kxdrdproc_t decode, void *rqstp, + __be32 *data, void *obj) { - struct rpc_cred *cred = task->tk_msg.rpc_cred; + struct rpc_cred *cred = task->tk_rqstp->rq_cred; - dprintk("RPC: %4d using %s cred %p to unwrap rpc data\n", + dprintk("RPC: %5u using %s cred %p to unwrap rpc data\n", task->tk_pid, cred->cr_ops->cr_name, cred); if (cred->cr_ops->crunwrap_resp) return cred->cr_ops->crunwrap_resp(task, decode, rqstp, data, obj); /* By default, we decode the arguments normally. */ - return decode(rqstp, data, obj); + return rpcauth_unwrap_req_decode(decode, rqstp, data, obj); } int rpcauth_refreshcred(struct rpc_task *task) { - struct rpc_cred *cred = task->tk_msg.rpc_cred; + struct rpc_cred *cred; int err; - dprintk("RPC: %4d refreshing %s cred %p\n", - task->tk_pid, task->tk_auth->au_ops->au_name, cred); + cred = task->tk_rqstp->rq_cred; + if (cred == NULL) { + err = rpcauth_bindcred(task, task->tk_msg.rpc_cred, task->tk_flags); + if (err < 0) + goto out; + cred = task->tk_rqstp->rq_cred; + } + dprintk("RPC: %5u refreshing %s cred %p\n", + task->tk_pid, cred->cr_auth->au_ops->au_name, cred); err = cred->cr_ops->crrefresh(task); +out: if (err < 0) task->tk_status = err; return err; @@ -377,17 +800,50 @@ rpcauth_refreshcred(struct rpc_task *task) void rpcauth_invalcred(struct rpc_task *task) { - dprintk("RPC: %4d invalidating %s cred %p\n", - task->tk_pid, task->tk_auth->au_ops->au_name, task->tk_msg.rpc_cred); - spin_lock(&rpc_credcache_lock); - if (task->tk_msg.rpc_cred) - task->tk_msg.rpc_cred->cr_flags &= ~RPCAUTH_CRED_UPTODATE; - spin_unlock(&rpc_credcache_lock); + struct rpc_cred *cred = task->tk_rqstp->rq_cred; + + dprintk("RPC: %5u invalidating %s cred %p\n", + task->tk_pid, cred->cr_auth->au_ops->au_name, cred); + if (cred) + clear_bit(RPCAUTH_CRED_UPTODATE, &cred->cr_flags); } int rpcauth_uptodatecred(struct rpc_task *task) { - return !(task->tk_msg.rpc_cred) || - (task->tk_msg.rpc_cred->cr_flags & RPCAUTH_CRED_UPTODATE); + struct rpc_cred *cred = task->tk_rqstp->rq_cred; + + return cred == NULL || + test_bit(RPCAUTH_CRED_UPTODATE, &cred->cr_flags) != 0; +} + +static struct shrinker rpc_cred_shrinker = { + .count_objects = rpcauth_cache_shrink_count, + .scan_objects = rpcauth_cache_shrink_scan, + .seeks = DEFAULT_SEEKS, +}; + +int __init rpcauth_init_module(void) +{ + int err; + + err = rpc_init_authunix(); + if (err < 0) + goto out1; + err = rpc_init_generic_auth(); + if (err < 0) + goto out2; + register_shrinker(&rpc_cred_shrinker); + return 0; +out2: + rpc_destroy_authunix(); +out1: + return err; +} + +void rpcauth_remove_module(void) +{ + rpc_destroy_authunix(); + rpc_destroy_generic_auth(); + unregister_shrinker(&rpc_cred_shrinker); } diff --git a/net/sunrpc/auth_generic.c b/net/sunrpc/auth_generic.c new file mode 100644 index 00000000000..ed04869b2d4 --- /dev/null +++ b/net/sunrpc/auth_generic.c @@ -0,0 +1,284 @@ +/* + * Generic RPC credential + * + * Copyright (C) 2008, Trond Myklebust <Trond.Myklebust@netapp.com> + */ + +#include <linux/err.h> +#include <linux/slab.h> +#include <linux/types.h> +#include <linux/module.h> +#include <linux/sched.h> +#include <linux/sunrpc/auth.h> +#include <linux/sunrpc/clnt.h> +#include <linux/sunrpc/debug.h> +#include <linux/sunrpc/sched.h> + +#ifdef RPC_DEBUG +# define RPCDBG_FACILITY RPCDBG_AUTH +#endif + +#define RPC_MACHINE_CRED_USERID GLOBAL_ROOT_UID +#define RPC_MACHINE_CRED_GROUPID GLOBAL_ROOT_GID + +struct generic_cred { + struct rpc_cred gc_base; + struct auth_cred acred; +}; + +static struct rpc_auth generic_auth; +static const struct rpc_credops generic_credops; + +/* + * Public call interface + */ +struct rpc_cred *rpc_lookup_cred(void) +{ + return rpcauth_lookupcred(&generic_auth, 0); +} +EXPORT_SYMBOL_GPL(rpc_lookup_cred); + +/* + * Public call interface for looking up machine creds. + */ +struct rpc_cred *rpc_lookup_machine_cred(const char *service_name) +{ + struct auth_cred acred = { + .uid = RPC_MACHINE_CRED_USERID, + .gid = RPC_MACHINE_CRED_GROUPID, + .principal = service_name, + .machine_cred = 1, + }; + + dprintk("RPC: looking up machine cred for service %s\n", + service_name); + return generic_auth.au_ops->lookup_cred(&generic_auth, &acred, 0); +} +EXPORT_SYMBOL_GPL(rpc_lookup_machine_cred); + +static struct rpc_cred *generic_bind_cred(struct rpc_task *task, + struct rpc_cred *cred, int lookupflags) +{ + struct rpc_auth *auth = task->tk_client->cl_auth; + struct auth_cred *acred = &container_of(cred, struct generic_cred, gc_base)->acred; + + return auth->au_ops->lookup_cred(auth, acred, lookupflags); +} + +/* + * Lookup generic creds for current process + */ +static struct rpc_cred * +generic_lookup_cred(struct rpc_auth *auth, struct auth_cred *acred, int flags) +{ + return rpcauth_lookup_credcache(&generic_auth, acred, flags); +} + +static struct rpc_cred * +generic_create_cred(struct rpc_auth *auth, struct auth_cred *acred, int flags) +{ + struct generic_cred *gcred; + + gcred = kmalloc(sizeof(*gcred), GFP_KERNEL); + if (gcred == NULL) + return ERR_PTR(-ENOMEM); + + rpcauth_init_cred(&gcred->gc_base, acred, &generic_auth, &generic_credops); + gcred->gc_base.cr_flags = 1UL << RPCAUTH_CRED_UPTODATE; + + gcred->acred.uid = acred->uid; + gcred->acred.gid = acred->gid; + gcred->acred.group_info = acred->group_info; + gcred->acred.ac_flags = 0; + if (gcred->acred.group_info != NULL) + get_group_info(gcred->acred.group_info); + gcred->acred.machine_cred = acred->machine_cred; + gcred->acred.principal = acred->principal; + + dprintk("RPC: allocated %s cred %p for uid %d gid %d\n", + gcred->acred.machine_cred ? "machine" : "generic", + gcred, + from_kuid(&init_user_ns, acred->uid), + from_kgid(&init_user_ns, acred->gid)); + return &gcred->gc_base; +} + +static void +generic_free_cred(struct rpc_cred *cred) +{ + struct generic_cred *gcred = container_of(cred, struct generic_cred, gc_base); + + dprintk("RPC: generic_free_cred %p\n", gcred); + if (gcred->acred.group_info != NULL) + put_group_info(gcred->acred.group_info); + kfree(gcred); +} + +static void +generic_free_cred_callback(struct rcu_head *head) +{ + struct rpc_cred *cred = container_of(head, struct rpc_cred, cr_rcu); + generic_free_cred(cred); +} + +static void +generic_destroy_cred(struct rpc_cred *cred) +{ + call_rcu(&cred->cr_rcu, generic_free_cred_callback); +} + +static int +machine_cred_match(struct auth_cred *acred, struct generic_cred *gcred, int flags) +{ + if (!gcred->acred.machine_cred || + gcred->acred.principal != acred->principal || + !uid_eq(gcred->acred.uid, acred->uid) || + !gid_eq(gcred->acred.gid, acred->gid)) + return 0; + return 1; +} + +/* + * Match credentials against current process creds. + */ +static int +generic_match(struct auth_cred *acred, struct rpc_cred *cred, int flags) +{ + struct generic_cred *gcred = container_of(cred, struct generic_cred, gc_base); + int i; + + if (acred->machine_cred) + return machine_cred_match(acred, gcred, flags); + + if (!uid_eq(gcred->acred.uid, acred->uid) || + !gid_eq(gcred->acred.gid, acred->gid) || + gcred->acred.machine_cred != 0) + goto out_nomatch; + + /* Optimisation in the case where pointers are identical... */ + if (gcred->acred.group_info == acred->group_info) + goto out_match; + + /* Slow path... */ + if (gcred->acred.group_info->ngroups != acred->group_info->ngroups) + goto out_nomatch; + for (i = 0; i < gcred->acred.group_info->ngroups; i++) { + if (!gid_eq(GROUP_AT(gcred->acred.group_info, i), + GROUP_AT(acred->group_info, i))) + goto out_nomatch; + } +out_match: + return 1; +out_nomatch: + return 0; +} + +int __init rpc_init_generic_auth(void) +{ + return rpcauth_init_credcache(&generic_auth); +} + +void rpc_destroy_generic_auth(void) +{ + rpcauth_destroy_credcache(&generic_auth); +} + +/* + * Test the the current time (now) against the underlying credential key expiry + * minus a timeout and setup notification. + * + * The normal case: + * If 'now' is before the key expiry minus RPC_KEY_EXPIRE_TIMEO, set + * the RPC_CRED_NOTIFY_TIMEOUT flag to setup the underlying credential + * rpc_credops crmatch routine to notify this generic cred when it's key + * expiration is within RPC_KEY_EXPIRE_TIMEO, and return 0. + * + * The error case: + * If the underlying cred lookup fails, return -EACCES. + * + * The 'almost' error case: + * If 'now' is within key expiry minus RPC_KEY_EXPIRE_TIMEO, but not within + * key expiry minus RPC_KEY_EXPIRE_FAIL, set the RPC_CRED_EXPIRE_SOON bit + * on the acred ac_flags and return 0. + */ +static int +generic_key_timeout(struct rpc_auth *auth, struct rpc_cred *cred) +{ + struct auth_cred *acred = &container_of(cred, struct generic_cred, + gc_base)->acred; + struct rpc_cred *tcred; + int ret = 0; + + + /* Fast track for non crkey_timeout (no key) underlying credentials */ + if (test_bit(RPC_CRED_NO_CRKEY_TIMEOUT, &acred->ac_flags)) + return 0; + + /* Fast track for the normal case */ + if (test_bit(RPC_CRED_NOTIFY_TIMEOUT, &acred->ac_flags)) + return 0; + + /* lookup_cred either returns a valid referenced rpc_cred, or PTR_ERR */ + tcred = auth->au_ops->lookup_cred(auth, acred, 0); + if (IS_ERR(tcred)) + return -EACCES; + + if (!tcred->cr_ops->crkey_timeout) { + set_bit(RPC_CRED_NO_CRKEY_TIMEOUT, &acred->ac_flags); + ret = 0; + goto out_put; + } + + /* Test for the almost error case */ + ret = tcred->cr_ops->crkey_timeout(tcred); + if (ret != 0) { + set_bit(RPC_CRED_KEY_EXPIRE_SOON, &acred->ac_flags); + ret = 0; + } else { + /* In case underlying cred key has been reset */ + if (test_and_clear_bit(RPC_CRED_KEY_EXPIRE_SOON, + &acred->ac_flags)) + dprintk("RPC: UID %d Credential key reset\n", + from_kuid(&init_user_ns, tcred->cr_uid)); + /* set up fasttrack for the normal case */ + set_bit(RPC_CRED_NOTIFY_TIMEOUT, &acred->ac_flags); + } + +out_put: + put_rpccred(tcred); + return ret; +} + +static const struct rpc_authops generic_auth_ops = { + .owner = THIS_MODULE, + .au_name = "Generic", + .lookup_cred = generic_lookup_cred, + .crcreate = generic_create_cred, + .key_timeout = generic_key_timeout, +}; + +static struct rpc_auth generic_auth = { + .au_ops = &generic_auth_ops, + .au_count = ATOMIC_INIT(0), +}; + +static bool generic_key_to_expire(struct rpc_cred *cred) +{ + struct auth_cred *acred = &container_of(cred, struct generic_cred, + gc_base)->acred; + bool ret; + + get_rpccred(cred); + ret = test_bit(RPC_CRED_KEY_EXPIRE_SOON, &acred->ac_flags); + put_rpccred(cred); + + return ret; +} + +static const struct rpc_credops generic_credops = { + .cr_name = "Generic cred", + .crdestroy = generic_destroy_cred, + .crbind = generic_bind_cred, + .crmatch = generic_match, + .crkey_to_expire = generic_key_to_expire, +}; diff --git a/net/sunrpc/auth_gss/Makefile b/net/sunrpc/auth_gss/Makefile index f3431a7e33d..14e9e53e63d 100644 --- a/net/sunrpc/auth_gss/Makefile +++ b/net/sunrpc/auth_gss/Makefile @@ -4,15 +4,11 @@ obj-$(CONFIG_SUNRPC_GSS) += auth_rpcgss.o -auth_rpcgss-objs := auth_gss.o gss_generic_token.o \ - gss_mech_switch.o svcauth_gss.o gss_krb5_crypto.o +auth_rpcgss-y := auth_gss.o gss_generic_token.o \ + gss_mech_switch.o svcauth_gss.o \ + gss_rpc_upcall.o gss_rpc_xdr.o obj-$(CONFIG_RPCSEC_GSS_KRB5) += rpcsec_gss_krb5.o -rpcsec_gss_krb5-objs := gss_krb5_mech.o gss_krb5_seal.o gss_krb5_unseal.o \ - gss_krb5_seqnum.o gss_krb5_wrap.o - -obj-$(CONFIG_RPCSEC_GSS_SPKM3) += rpcsec_gss_spkm3.o - -rpcsec_gss_spkm3-objs := gss_spkm3_mech.o gss_spkm3_seal.o gss_spkm3_unseal.o \ - gss_spkm3_token.o +rpcsec_gss_krb5-y := gss_krb5_mech.o gss_krb5_seal.o gss_krb5_unseal.o \ + gss_krb5_seqnum.o gss_krb5_wrap.o gss_krb5_crypto.o gss_krb5_keys.o diff --git a/net/sunrpc/auth_gss/auth_gss.c b/net/sunrpc/auth_gss/auth_gss.c index f44f46f1d8e..b6e440baccc 100644 --- a/net/sunrpc/auth_gss/auth_gss.c +++ b/net/sunrpc/auth_gss/auth_gss.c @@ -1,8 +1,8 @@ /* - * linux/net/sunrpc/auth_gss.c + * linux/net/sunrpc/auth_gss/auth_gss.c * * RPCSEC_GSS client authentication. - * + * * Copyright (c) 2000 The Regents of the University of Michigan. * All rights reserved. * @@ -33,8 +33,6 @@ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING * NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS * SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. - * - * $Id$ */ @@ -53,87 +51,68 @@ #include <linux/sunrpc/rpc_pipe_fs.h> #include <linux/sunrpc/gss_api.h> #include <asm/uaccess.h> +#include <linux/hashtable.h> + +#include "../netns.h" + +static const struct rpc_authops authgss_ops; -static struct rpc_authops authgss_ops; +static const struct rpc_credops gss_credops; +static const struct rpc_credops gss_nullops; -static struct rpc_credops gss_credops; +#define GSS_RETRY_EXPIRED 5 +static unsigned int gss_expired_cred_retry_delay = GSS_RETRY_EXPIRED; + +#define GSS_KEY_EXPIRE_TIMEO 240 +static unsigned int gss_key_expire_timeo = GSS_KEY_EXPIRE_TIMEO; #ifdef RPC_DEBUG # define RPCDBG_FACILITY RPCDBG_AUTH #endif -#define NFS_NGROUPS 16 - -#define GSS_CRED_EXPIRE (60 * HZ) /* XXX: reasonable? */ -#define GSS_CRED_SLACK 1024 /* XXX: unused */ +#define GSS_CRED_SLACK (RPC_MAX_AUTH_SIZE * 2) /* length of a krb5 verifier (48), plus data added before arguments when * using integrity (two 4-byte integers): */ -#define GSS_VERF_SLACK 56 - -/* XXX this define must match the gssd define -* as it is passed to gssd to signal the use of -* machine creds should be part of the shared rpc interface */ +#define GSS_VERF_SLACK 100 -#define CA_RUN_AS_MACHINE 0x00000200 +static DEFINE_HASHTABLE(gss_auth_hash_table, 4); +static DEFINE_SPINLOCK(gss_auth_hash_lock); -/* dump the buffer in `emacs-hexl' style */ -#define isprint(c) ((c > 0x1f) && (c < 0x7f)) - -static DEFINE_RWLOCK(gss_ctx_lock); +struct gss_pipe { + struct rpc_pipe_dir_object pdo; + struct rpc_pipe *pipe; + struct rpc_clnt *clnt; + const char *name; + struct kref kref; +}; struct gss_auth { + struct kref kref; + struct hlist_node hash; struct rpc_auth rpc_auth; struct gss_api_mech *mech; enum rpc_gss_svc service; - struct list_head upcalls; struct rpc_clnt *client; - struct dentry *dentry; - char path[48]; - spinlock_t lock; + struct net *net; + /* + * There are two upcall pipes; dentry[1], named "gssd", is used + * for the new text-based upcall; dentry[0] is named after the + * mechanism (for example, "krb5") and exists for + * backwards-compatibility with older gssd's. + */ + struct gss_pipe *gss_pipe[2]; + const char *target_name; }; -static void gss_destroy_ctx(struct gss_cl_ctx *); -static struct rpc_pipe_ops gss_upcall_ops; - -void -print_hexl(u32 *p, u_int length, u_int offset) -{ - u_int i, j, jm; - u8 c, *cp; - - dprintk("RPC: print_hexl: length %d\n",length); - dprintk("\n"); - cp = (u8 *) p; - - for (i = 0; i < length; i += 0x10) { - dprintk(" %04x: ", (u_int)(i + offset)); - jm = length - i; - jm = jm > 16 ? 16 : jm; - - for (j = 0; j < jm; j++) { - if ((j % 2) == 1) - dprintk("%02x ", (u_int)cp[i+j]); - else - dprintk("%02x", (u_int)cp[i+j]); - } - for (; j < 16; j++) { - if ((j % 2) == 1) - dprintk(" "); - else - dprintk(" "); - } - dprintk(" "); - - for (j = 0; j < jm; j++) { - c = cp[i+j]; - c = isprint(c) ? c : '.'; - dprintk("%c", c); - } - dprintk("\n"); - } -} +/* pipe_version >= 0 if and only if someone has a pipe open. */ +static DEFINE_SPINLOCK(pipe_version_lock); +static struct rpc_wait_queue pipe_version_rpc_waitqueue; +static DECLARE_WAIT_QUEUE_HEAD(pipe_version_waitqueue); +static void gss_put_auth(struct gss_auth *gss_auth); -EXPORT_SYMBOL(print_hexl); +static void gss_free_ctx(struct gss_cl_ctx *); +static const struct rpc_pipe_ops gss_upcall_ops_v0; +static const struct rpc_pipe_ops gss_upcall_ops_v1; static inline struct gss_cl_ctx * gss_get_ctx(struct gss_cl_ctx *ctx) @@ -146,34 +125,26 @@ static inline void gss_put_ctx(struct gss_cl_ctx *ctx) { if (atomic_dec_and_test(&ctx->count)) - gss_destroy_ctx(ctx); + gss_free_ctx(ctx); } +/* gss_cred_set_ctx: + * called by gss_upcall_callback and gss_create_upcall in order + * to set the gss context. The actual exchange of an old context + * and a new one is protected by the pipe->lock. + */ static void gss_cred_set_ctx(struct rpc_cred *cred, struct gss_cl_ctx *ctx) { struct gss_cred *gss_cred = container_of(cred, struct gss_cred, gc_base); - struct gss_cl_ctx *old; - write_lock(&gss_ctx_lock); - old = gss_cred->gc_ctx; - gss_cred->gc_ctx = ctx; - cred->cr_flags |= RPCAUTH_CRED_UPTODATE; - write_unlock(&gss_ctx_lock); - if (old) - gss_put_ctx(old); -} - -static int -gss_cred_is_uptodate_ctx(struct rpc_cred *cred) -{ - struct gss_cred *gss_cred = container_of(cred, struct gss_cred, gc_base); - int res = 0; - read_lock(&gss_ctx_lock); - if ((cred->cr_flags & RPCAUTH_CRED_UPTODATE) && gss_cred->gc_ctx) - res = 1; - read_unlock(&gss_ctx_lock); - return res; + if (!test_bit(RPCAUTH_CRED_NEW, &cred->cr_flags)) + return; + gss_get_ctx(ctx); + rcu_assign_pointer(gss_cred->gc_ctx, ctx); + set_bit(RPCAUTH_CRED_UPTODATE, &cred->cr_flags); + smp_mb__before_atomic(); + clear_bit(RPCAUTH_CRED_NEW, &cred->cr_flags); } static const void * @@ -198,11 +169,10 @@ simple_get_netobj(const void *p, const void *end, struct xdr_netobj *dest) q = (const void *)((const char *)p + len); if (unlikely(q > end || q < p)) return ERR_PTR(-EFAULT); - dest->data = kmalloc(len, GFP_KERNEL); + dest->data = kmemdup(p, len, GFP_NOFS); if (unlikely(dest->data == NULL)) return ERR_PTR(-ENOMEM); dest->len = len; - memcpy(dest->data, p, len); return q; } @@ -212,10 +182,10 @@ gss_cred_get_ctx(struct rpc_cred *cred) struct gss_cred *gss_cred = container_of(cred, struct gss_cred, gc_base); struct gss_cl_ctx *ctx = NULL; - read_lock(&gss_ctx_lock); + rcu_read_lock(); if (gss_cred->gc_ctx) ctx = gss_get_ctx(gss_cred->gc_ctx); - read_unlock(&gss_ctx_lock); + rcu_read_unlock(); return ctx; } @@ -224,9 +194,8 @@ gss_alloc_context(void) { struct gss_cl_ctx *ctx; - ctx = kmalloc(sizeof(*ctx), GFP_KERNEL); + ctx = kzalloc(sizeof(*ctx), GFP_NOFS); if (ctx != NULL) { - memset(ctx, 0, sizeof(*ctx)); ctx->gc_proc = RPC_GSS_PROC_DATA; ctx->gc_seq = 1; /* NetApp 6.4R1 doesn't accept seq. no. 0 */ spin_lock_init(&ctx->gc_seq_lock); @@ -242,25 +211,37 @@ gss_fill_context(const void *p, const void *end, struct gss_cl_ctx *ctx, struct const void *q; unsigned int seclen; unsigned int timeout; + unsigned long now = jiffies; u32 window_size; int ret; - /* First unsigned int gives the lifetime (in seconds) of the cred */ + /* First unsigned int gives the remaining lifetime in seconds of the + * credential - e.g. the remaining TGT lifetime for Kerberos or + * the -t value passed to GSSD. + */ p = simple_get_bytes(p, end, &timeout, sizeof(timeout)); if (IS_ERR(p)) goto err; if (timeout == 0) timeout = GSSD_MIN_TIMEOUT; - ctx->gc_expiry = jiffies + (unsigned long)timeout * HZ * 3 / 4; - /* Sequence number window. Determines the maximum number of simultaneous requests */ + ctx->gc_expiry = now + ((unsigned long)timeout * HZ); + /* Sequence number window. Determines the maximum number of + * simultaneous requests + */ p = simple_get_bytes(p, end, &window_size, sizeof(window_size)); if (IS_ERR(p)) goto err; ctx->gc_win = window_size; /* gssd signals an error by passing ctx->gc_win = 0: */ if (ctx->gc_win == 0) { - /* in which case, p points to an error code which we ignore */ - p = ERR_PTR(-EACCES); + /* + * in which case, p points to an error code. Anything other + * than -EKEYEXPIRED gets converted to -EACCES. + */ + p = simple_get_bytes(p, end, &ret, sizeof(ret)); + if (!IS_ERR(p)) + p = (ret == -EKEYEXPIRED) ? ERR_PTR(-EKEYEXPIRED) : + ERR_PTR(-EACCES); goto err; } /* copy the opaque wire context */ @@ -276,80 +257,113 @@ gss_fill_context(const void *p, const void *end, struct gss_cl_ctx *ctx, struct p = ERR_PTR(-EFAULT); goto err; } - ret = gss_import_sec_context(p, seclen, gm, &ctx->gc_gss_ctx); + ret = gss_import_sec_context(p, seclen, gm, &ctx->gc_gss_ctx, NULL, GFP_NOFS); if (ret < 0) { p = ERR_PTR(ret); goto err; } + dprintk("RPC: %s Success. gc_expiry %lu now %lu timeout %u\n", + __func__, ctx->gc_expiry, now, timeout); return q; err: - dprintk("RPC: gss_fill_context returning %ld\n", -PTR_ERR(p)); + dprintk("RPC: %s returns error %ld\n", __func__, -PTR_ERR(p)); return p; } +#define UPCALL_BUF_LEN 128 struct gss_upcall_msg { atomic_t count; - uid_t uid; + kuid_t uid; struct rpc_pipe_msg msg; struct list_head list; struct gss_auth *auth; + struct rpc_pipe *pipe; struct rpc_wait_queue rpc_waitqueue; wait_queue_head_t waitqueue; struct gss_cl_ctx *ctx; + char databuf[UPCALL_BUF_LEN]; }; +static int get_pipe_version(struct net *net) +{ + struct sunrpc_net *sn = net_generic(net, sunrpc_net_id); + int ret; + + spin_lock(&pipe_version_lock); + if (sn->pipe_version >= 0) { + atomic_inc(&sn->pipe_users); + ret = sn->pipe_version; + } else + ret = -EAGAIN; + spin_unlock(&pipe_version_lock); + return ret; +} + +static void put_pipe_version(struct net *net) +{ + struct sunrpc_net *sn = net_generic(net, sunrpc_net_id); + + if (atomic_dec_and_lock(&sn->pipe_users, &pipe_version_lock)) { + sn->pipe_version = -1; + spin_unlock(&pipe_version_lock); + } +} + static void gss_release_msg(struct gss_upcall_msg *gss_msg) { + struct net *net = gss_msg->auth->net; if (!atomic_dec_and_test(&gss_msg->count)) return; + put_pipe_version(net); BUG_ON(!list_empty(&gss_msg->list)); if (gss_msg->ctx != NULL) gss_put_ctx(gss_msg->ctx); + rpc_destroy_wait_queue(&gss_msg->rpc_waitqueue); + gss_put_auth(gss_msg->auth); kfree(gss_msg); } static struct gss_upcall_msg * -__gss_find_upcall(struct gss_auth *gss_auth, uid_t uid) +__gss_find_upcall(struct rpc_pipe *pipe, kuid_t uid) { struct gss_upcall_msg *pos; - list_for_each_entry(pos, &gss_auth->upcalls, list) { - if (pos->uid != uid) + list_for_each_entry(pos, &pipe->in_downcall, list) { + if (!uid_eq(pos->uid, uid)) continue; atomic_inc(&pos->count); - dprintk("RPC: gss_find_upcall found msg %p\n", pos); + dprintk("RPC: %s found msg %p\n", __func__, pos); return pos; } - dprintk("RPC: gss_find_upcall found nothing\n"); + dprintk("RPC: %s found nothing\n", __func__); return NULL; } -/* Try to add a upcall to the pipefs queue. +/* Try to add an upcall to the pipefs queue. * If an upcall owned by our uid already exists, then we return a reference * to that upcall instead of adding the new upcall. */ static inline struct gss_upcall_msg * -gss_add_msg(struct gss_auth *gss_auth, struct gss_upcall_msg *gss_msg) +gss_add_msg(struct gss_upcall_msg *gss_msg) { + struct rpc_pipe *pipe = gss_msg->pipe; struct gss_upcall_msg *old; - spin_lock(&gss_auth->lock); - old = __gss_find_upcall(gss_auth, gss_msg->uid); + spin_lock(&pipe->lock); + old = __gss_find_upcall(pipe, gss_msg->uid); if (old == NULL) { atomic_inc(&gss_msg->count); - list_add(&gss_msg->list, &gss_auth->upcalls); + list_add(&gss_msg->list, &pipe->in_downcall); } else gss_msg = old; - spin_unlock(&gss_auth->lock); + spin_unlock(&pipe->lock); return gss_msg; } static void __gss_unhash_msg(struct gss_upcall_msg *gss_msg) { - if (list_empty(&gss_msg->list)) - return; list_del_init(&gss_msg->list); rpc_wake_up_status(&gss_msg->rpc_waitqueue, gss_msg->msg.errno); wake_up_all(&gss_msg->waitqueue); @@ -359,63 +373,159 @@ __gss_unhash_msg(struct gss_upcall_msg *gss_msg) static void gss_unhash_msg(struct gss_upcall_msg *gss_msg) { - struct gss_auth *gss_auth = gss_msg->auth; + struct rpc_pipe *pipe = gss_msg->pipe; - spin_lock(&gss_auth->lock); - __gss_unhash_msg(gss_msg); - spin_unlock(&gss_auth->lock); + if (list_empty(&gss_msg->list)) + return; + spin_lock(&pipe->lock); + if (!list_empty(&gss_msg->list)) + __gss_unhash_msg(gss_msg); + spin_unlock(&pipe->lock); +} + +static void +gss_handle_downcall_result(struct gss_cred *gss_cred, struct gss_upcall_msg *gss_msg) +{ + switch (gss_msg->msg.errno) { + case 0: + if (gss_msg->ctx == NULL) + break; + clear_bit(RPCAUTH_CRED_NEGATIVE, &gss_cred->gc_base.cr_flags); + gss_cred_set_ctx(&gss_cred->gc_base, gss_msg->ctx); + break; + case -EKEYEXPIRED: + set_bit(RPCAUTH_CRED_NEGATIVE, &gss_cred->gc_base.cr_flags); + } + gss_cred->gc_upcall_timestamp = jiffies; + gss_cred->gc_upcall = NULL; + rpc_wake_up_status(&gss_msg->rpc_waitqueue, gss_msg->msg.errno); } static void gss_upcall_callback(struct rpc_task *task) { - struct gss_cred *gss_cred = container_of(task->tk_msg.rpc_cred, + struct gss_cred *gss_cred = container_of(task->tk_rqstp->rq_cred, struct gss_cred, gc_base); struct gss_upcall_msg *gss_msg = gss_cred->gc_upcall; + struct rpc_pipe *pipe = gss_msg->pipe; - BUG_ON(gss_msg == NULL); - if (gss_msg->ctx) - gss_cred_set_ctx(task->tk_msg.rpc_cred, gss_get_ctx(gss_msg->ctx)); - else - task->tk_status = gss_msg->msg.errno; - spin_lock(&gss_msg->auth->lock); - gss_cred->gc_upcall = NULL; - rpc_wake_up_status(&gss_msg->rpc_waitqueue, gss_msg->msg.errno); - spin_unlock(&gss_msg->auth->lock); + spin_lock(&pipe->lock); + gss_handle_downcall_result(gss_cred, gss_msg); + spin_unlock(&pipe->lock); + task->tk_status = gss_msg->msg.errno; gss_release_msg(gss_msg); } -static inline struct gss_upcall_msg * -gss_alloc_msg(struct gss_auth *gss_auth, uid_t uid) +static void gss_encode_v0_msg(struct gss_upcall_msg *gss_msg) { - struct gss_upcall_msg *gss_msg; + uid_t uid = from_kuid(&init_user_ns, gss_msg->uid); + memcpy(gss_msg->databuf, &uid, sizeof(uid)); + gss_msg->msg.data = gss_msg->databuf; + gss_msg->msg.len = sizeof(uid); + + BUILD_BUG_ON(sizeof(uid) > sizeof(gss_msg->databuf)); +} - gss_msg = kmalloc(sizeof(*gss_msg), GFP_KERNEL); - if (gss_msg != NULL) { - memset(gss_msg, 0, sizeof(*gss_msg)); - INIT_LIST_HEAD(&gss_msg->list); - rpc_init_wait_queue(&gss_msg->rpc_waitqueue, "RPCSEC_GSS upcall waitq"); - init_waitqueue_head(&gss_msg->waitqueue); - atomic_set(&gss_msg->count, 1); - gss_msg->msg.data = &gss_msg->uid; - gss_msg->msg.len = sizeof(gss_msg->uid); - gss_msg->uid = uid; - gss_msg->auth = gss_auth; +static int gss_encode_v1_msg(struct gss_upcall_msg *gss_msg, + const char *service_name, + const char *target_name) +{ + struct gss_api_mech *mech = gss_msg->auth->mech; + char *p = gss_msg->databuf; + size_t buflen = sizeof(gss_msg->databuf); + int len; + + len = scnprintf(p, buflen, "mech=%s uid=%d ", mech->gm_name, + from_kuid(&init_user_ns, gss_msg->uid)); + buflen -= len; + p += len; + gss_msg->msg.len = len; + if (target_name) { + len = scnprintf(p, buflen, "target=%s ", target_name); + buflen -= len; + p += len; + gss_msg->msg.len += len; + } + if (service_name != NULL) { + len = scnprintf(p, buflen, "service=%s ", service_name); + buflen -= len; + p += len; + gss_msg->msg.len += len; + } + if (mech->gm_upcall_enctypes) { + len = scnprintf(p, buflen, "enctypes=%s ", + mech->gm_upcall_enctypes); + buflen -= len; + p += len; + gss_msg->msg.len += len; } + len = scnprintf(p, buflen, "\n"); + if (len == 0) + goto out_overflow; + gss_msg->msg.len += len; + + gss_msg->msg.data = gss_msg->databuf; + return 0; +out_overflow: + WARN_ON_ONCE(1); + return -ENOMEM; +} + +static struct gss_upcall_msg * +gss_alloc_msg(struct gss_auth *gss_auth, + kuid_t uid, const char *service_name) +{ + struct gss_upcall_msg *gss_msg; + int vers; + int err = -ENOMEM; + + gss_msg = kzalloc(sizeof(*gss_msg), GFP_NOFS); + if (gss_msg == NULL) + goto err; + vers = get_pipe_version(gss_auth->net); + err = vers; + if (err < 0) + goto err_free_msg; + gss_msg->pipe = gss_auth->gss_pipe[vers]->pipe; + INIT_LIST_HEAD(&gss_msg->list); + rpc_init_wait_queue(&gss_msg->rpc_waitqueue, "RPCSEC_GSS upcall waitq"); + init_waitqueue_head(&gss_msg->waitqueue); + atomic_set(&gss_msg->count, 1); + gss_msg->uid = uid; + gss_msg->auth = gss_auth; + switch (vers) { + case 0: + gss_encode_v0_msg(gss_msg); + break; + default: + err = gss_encode_v1_msg(gss_msg, service_name, gss_auth->target_name); + if (err) + goto err_put_pipe_version; + }; + kref_get(&gss_auth->kref); return gss_msg; +err_put_pipe_version: + put_pipe_version(gss_auth->net); +err_free_msg: + kfree(gss_msg); +err: + return ERR_PTR(err); } static struct gss_upcall_msg * -gss_setup_upcall(struct rpc_clnt *clnt, struct gss_auth *gss_auth, struct rpc_cred *cred) +gss_setup_upcall(struct gss_auth *gss_auth, struct rpc_cred *cred) { + struct gss_cred *gss_cred = container_of(cred, + struct gss_cred, gc_base); struct gss_upcall_msg *gss_new, *gss_msg; + kuid_t uid = cred->cr_uid; - gss_new = gss_alloc_msg(gss_auth, cred->cr_uid); - if (gss_new == NULL) - return ERR_PTR(-ENOMEM); - gss_msg = gss_add_msg(gss_auth, gss_new); + gss_new = gss_alloc_msg(gss_auth, uid, gss_cred->gc_principal); + if (IS_ERR(gss_new)) + return gss_new; + gss_msg = gss_add_msg(gss_new); if (gss_msg == gss_new) { - int res = rpc_queue_upcall(gss_auth->dentry->d_inode, &gss_new->msg); + int res = rpc_queue_upcall(gss_new->pipe, &gss_new->msg); if (res) { gss_unhash_msg(gss_new); gss_msg = ERR_PTR(res); @@ -425,103 +535,125 @@ gss_setup_upcall(struct rpc_clnt *clnt, struct gss_auth *gss_auth, struct rpc_cr return gss_msg; } +static void warn_gssd(void) +{ + dprintk("AUTH_GSS upcall failed. Please check user daemon is running.\n"); +} + static inline int gss_refresh_upcall(struct rpc_task *task) { - struct rpc_cred *cred = task->tk_msg.rpc_cred; - struct gss_auth *gss_auth = container_of(task->tk_client->cl_auth, + struct rpc_cred *cred = task->tk_rqstp->rq_cred; + struct gss_auth *gss_auth = container_of(cred->cr_auth, struct gss_auth, rpc_auth); struct gss_cred *gss_cred = container_of(cred, struct gss_cred, gc_base); struct gss_upcall_msg *gss_msg; + struct rpc_pipe *pipe; int err = 0; - dprintk("RPC: %4u gss_refresh_upcall for uid %u\n", task->tk_pid, cred->cr_uid); - gss_msg = gss_setup_upcall(task->tk_client, gss_auth, cred); + dprintk("RPC: %5u %s for uid %u\n", + task->tk_pid, __func__, from_kuid(&init_user_ns, cred->cr_uid)); + gss_msg = gss_setup_upcall(gss_auth, cred); + if (PTR_ERR(gss_msg) == -EAGAIN) { + /* XXX: warning on the first, under the assumption we + * shouldn't normally hit this case on a refresh. */ + warn_gssd(); + task->tk_timeout = 15*HZ; + rpc_sleep_on(&pipe_version_rpc_waitqueue, task, NULL); + return -EAGAIN; + } if (IS_ERR(gss_msg)) { err = PTR_ERR(gss_msg); goto out; } - spin_lock(&gss_auth->lock); + pipe = gss_msg->pipe; + spin_lock(&pipe->lock); if (gss_cred->gc_upcall != NULL) - rpc_sleep_on(&gss_cred->gc_upcall->rpc_waitqueue, task, NULL, NULL); + rpc_sleep_on(&gss_cred->gc_upcall->rpc_waitqueue, task, NULL); else if (gss_msg->ctx == NULL && gss_msg->msg.errno >= 0) { task->tk_timeout = 0; gss_cred->gc_upcall = gss_msg; /* gss_upcall_callback will release the reference to gss_upcall_msg */ atomic_inc(&gss_msg->count); - rpc_sleep_on(&gss_msg->rpc_waitqueue, task, gss_upcall_callback, NULL); - } else + rpc_sleep_on(&gss_msg->rpc_waitqueue, task, gss_upcall_callback); + } else { + gss_handle_downcall_result(gss_cred, gss_msg); err = gss_msg->msg.errno; - spin_unlock(&gss_auth->lock); + } + spin_unlock(&pipe->lock); gss_release_msg(gss_msg); out: - dprintk("RPC: %4u gss_refresh_upcall for uid %u result %d\n", task->tk_pid, - cred->cr_uid, err); + dprintk("RPC: %5u %s for uid %u result %d\n", + task->tk_pid, __func__, + from_kuid(&init_user_ns, cred->cr_uid), err); return err; } static inline int gss_create_upcall(struct gss_auth *gss_auth, struct gss_cred *gss_cred) { + struct net *net = gss_auth->net; + struct sunrpc_net *sn = net_generic(net, sunrpc_net_id); + struct rpc_pipe *pipe; struct rpc_cred *cred = &gss_cred->gc_base; struct gss_upcall_msg *gss_msg; DEFINE_WAIT(wait); - int err = 0; + int err; - dprintk("RPC: gss_upcall for uid %u\n", cred->cr_uid); - gss_msg = gss_setup_upcall(gss_auth->client, gss_auth, cred); + dprintk("RPC: %s for uid %u\n", + __func__, from_kuid(&init_user_ns, cred->cr_uid)); +retry: + err = 0; + /* if gssd is down, just skip upcalling altogether */ + if (!gssd_running(net)) { + warn_gssd(); + return -EACCES; + } + gss_msg = gss_setup_upcall(gss_auth, cred); + if (PTR_ERR(gss_msg) == -EAGAIN) { + err = wait_event_interruptible_timeout(pipe_version_waitqueue, + sn->pipe_version >= 0, 15 * HZ); + if (sn->pipe_version < 0) { + warn_gssd(); + err = -EACCES; + } + if (err < 0) + goto out; + goto retry; + } if (IS_ERR(gss_msg)) { err = PTR_ERR(gss_msg); goto out; } + pipe = gss_msg->pipe; for (;;) { - prepare_to_wait(&gss_msg->waitqueue, &wait, TASK_INTERRUPTIBLE); - spin_lock(&gss_auth->lock); + prepare_to_wait(&gss_msg->waitqueue, &wait, TASK_KILLABLE); + spin_lock(&pipe->lock); if (gss_msg->ctx != NULL || gss_msg->msg.errno < 0) { - spin_unlock(&gss_auth->lock); break; } - spin_unlock(&gss_auth->lock); - if (signalled()) { + spin_unlock(&pipe->lock); + if (fatal_signal_pending(current)) { err = -ERESTARTSYS; goto out_intr; } schedule(); } if (gss_msg->ctx) - gss_cred_set_ctx(cred, gss_get_ctx(gss_msg->ctx)); + gss_cred_set_ctx(cred, gss_msg->ctx); else err = gss_msg->msg.errno; + spin_unlock(&pipe->lock); out_intr: finish_wait(&gss_msg->waitqueue, &wait); gss_release_msg(gss_msg); out: - dprintk("RPC: gss_create_upcall for uid %u result %d\n", cred->cr_uid, err); + dprintk("RPC: %s for uid %u result %d\n", + __func__, from_kuid(&init_user_ns, cred->cr_uid), err); return err; } -static ssize_t -gss_pipe_upcall(struct file *filp, struct rpc_pipe_msg *msg, - char __user *dst, size_t buflen) -{ - char *data = (char *)msg->data + msg->copied; - ssize_t mlen = msg->len; - ssize_t left; - - if (mlen > buflen) - mlen = buflen; - left = copy_to_user(dst, data, mlen); - if (left < 0) { - msg->errno = left; - return left; - } - mlen -= left; - msg->copied += mlen; - msg->errno = 0; - return mlen; -} - #define MSG_BUF_MAXSIZE 1024 static ssize_t @@ -529,181 +661,386 @@ gss_pipe_downcall(struct file *filp, const char __user *src, size_t mlen) { const void *p, *end; void *buf; - struct rpc_clnt *clnt; - struct gss_auth *gss_auth; - struct rpc_cred *cred; struct gss_upcall_msg *gss_msg; + struct rpc_pipe *pipe = RPC_I(file_inode(filp))->pipe; struct gss_cl_ctx *ctx; - uid_t uid; - int err = -EFBIG; + uid_t id; + kuid_t uid; + ssize_t err = -EFBIG; if (mlen > MSG_BUF_MAXSIZE) goto out; err = -ENOMEM; - buf = kmalloc(mlen, GFP_KERNEL); + buf = kmalloc(mlen, GFP_NOFS); if (!buf) goto out; - clnt = RPC_I(filp->f_dentry->d_inode)->private; err = -EFAULT; if (copy_from_user(buf, src, mlen)) goto err; end = (const void *)((char *)buf + mlen); - p = simple_get_bytes(buf, end, &uid, sizeof(uid)); + p = simple_get_bytes(buf, end, &id, sizeof(id)); if (IS_ERR(p)) { err = PTR_ERR(p); goto err; } + uid = make_kuid(&init_user_ns, id); + if (!uid_valid(uid)) { + err = -EINVAL; + goto err; + } + err = -ENOMEM; ctx = gss_alloc_context(); if (ctx == NULL) goto err; - err = 0; - gss_auth = container_of(clnt->cl_auth, struct gss_auth, rpc_auth); - p = gss_fill_context(p, end, ctx, gss_auth->mech); + + err = -ENOENT; + /* Find a matching upcall */ + spin_lock(&pipe->lock); + gss_msg = __gss_find_upcall(pipe, uid); + if (gss_msg == NULL) { + spin_unlock(&pipe->lock); + goto err_put_ctx; + } + list_del_init(&gss_msg->list); + spin_unlock(&pipe->lock); + + p = gss_fill_context(p, end, ctx, gss_msg->auth->mech); if (IS_ERR(p)) { err = PTR_ERR(p); - if (err != -EACCES) - goto err_put_ctx; - } - spin_lock(&gss_auth->lock); - gss_msg = __gss_find_upcall(gss_auth, uid); - if (gss_msg) { - if (err == 0 && gss_msg->ctx == NULL) - gss_msg->ctx = gss_get_ctx(ctx); - gss_msg->msg.errno = err; - __gss_unhash_msg(gss_msg); - spin_unlock(&gss_auth->lock); - gss_release_msg(gss_msg); - } else { - struct auth_cred acred = { .uid = uid }; - spin_unlock(&gss_auth->lock); - cred = rpcauth_lookup_credcache(clnt->cl_auth, &acred, 0); - if (IS_ERR(cred)) { - err = PTR_ERR(cred); - goto err_put_ctx; + switch (err) { + case -EACCES: + case -EKEYEXPIRED: + gss_msg->msg.errno = err; + err = mlen; + break; + case -EFAULT: + case -ENOMEM: + case -EINVAL: + case -ENOSYS: + gss_msg->msg.errno = -EAGAIN; + break; + default: + printk(KERN_CRIT "%s: bad return from " + "gss_fill_context: %zd\n", __func__, err); + BUG(); } - gss_cred_set_ctx(cred, gss_get_ctx(ctx)); + goto err_release_msg; } - gss_put_ctx(ctx); - kfree(buf); - dprintk("RPC: gss_pipe_downcall returning length %Zu\n", mlen); - return mlen; + gss_msg->ctx = gss_get_ctx(ctx); + err = mlen; + +err_release_msg: + spin_lock(&pipe->lock); + __gss_unhash_msg(gss_msg); + spin_unlock(&pipe->lock); + gss_release_msg(gss_msg); err_put_ctx: gss_put_ctx(ctx); err: kfree(buf); out: - dprintk("RPC: gss_pipe_downcall returning %d\n", err); + dprintk("RPC: %s returning %Zd\n", __func__, err); return err; } +static int gss_pipe_open(struct inode *inode, int new_version) +{ + struct net *net = inode->i_sb->s_fs_info; + struct sunrpc_net *sn = net_generic(net, sunrpc_net_id); + int ret = 0; + + spin_lock(&pipe_version_lock); + if (sn->pipe_version < 0) { + /* First open of any gss pipe determines the version: */ + sn->pipe_version = new_version; + rpc_wake_up(&pipe_version_rpc_waitqueue); + wake_up(&pipe_version_waitqueue); + } else if (sn->pipe_version != new_version) { + /* Trying to open a pipe of a different version */ + ret = -EBUSY; + goto out; + } + atomic_inc(&sn->pipe_users); +out: + spin_unlock(&pipe_version_lock); + return ret; + +} + +static int gss_pipe_open_v0(struct inode *inode) +{ + return gss_pipe_open(inode, 0); +} + +static int gss_pipe_open_v1(struct inode *inode) +{ + return gss_pipe_open(inode, 1); +} + static void gss_pipe_release(struct inode *inode) { - struct rpc_inode *rpci = RPC_I(inode); - struct rpc_clnt *clnt; - struct rpc_auth *auth; - struct gss_auth *gss_auth; + struct net *net = inode->i_sb->s_fs_info; + struct rpc_pipe *pipe = RPC_I(inode)->pipe; + struct gss_upcall_msg *gss_msg; - clnt = rpci->private; - auth = clnt->cl_auth; - gss_auth = container_of(auth, struct gss_auth, rpc_auth); - spin_lock(&gss_auth->lock); - while (!list_empty(&gss_auth->upcalls)) { - struct gss_upcall_msg *gss_msg; +restart: + spin_lock(&pipe->lock); + list_for_each_entry(gss_msg, &pipe->in_downcall, list) { - gss_msg = list_entry(gss_auth->upcalls.next, - struct gss_upcall_msg, list); + if (!list_empty(&gss_msg->msg.list)) + continue; gss_msg->msg.errno = -EPIPE; atomic_inc(&gss_msg->count); __gss_unhash_msg(gss_msg); - spin_unlock(&gss_auth->lock); + spin_unlock(&pipe->lock); gss_release_msg(gss_msg); - spin_lock(&gss_auth->lock); + goto restart; } - spin_unlock(&gss_auth->lock); + spin_unlock(&pipe->lock); + + put_pipe_version(net); } static void gss_pipe_destroy_msg(struct rpc_pipe_msg *msg) { struct gss_upcall_msg *gss_msg = container_of(msg, struct gss_upcall_msg, msg); - static unsigned long ratelimit; if (msg->errno < 0) { - dprintk("RPC: gss_pipe_destroy_msg releasing msg %p\n", - gss_msg); + dprintk("RPC: %s releasing msg %p\n", + __func__, gss_msg); atomic_inc(&gss_msg->count); gss_unhash_msg(gss_msg); - if (msg->errno == -ETIMEDOUT || msg->errno == -EPIPE) { - unsigned long now = jiffies; - if (time_after(now, ratelimit)) { - printk(KERN_WARNING "RPC: AUTH_GSS upcall timed out.\n" - "Please check user daemon is running!\n"); - ratelimit = now + 15*HZ; - } - } + if (msg->errno == -ETIMEDOUT) + warn_gssd(); gss_release_msg(gss_msg); } } -/* - * NOTE: we have the opportunity to use different +static void gss_pipe_dentry_destroy(struct dentry *dir, + struct rpc_pipe_dir_object *pdo) +{ + struct gss_pipe *gss_pipe = pdo->pdo_data; + struct rpc_pipe *pipe = gss_pipe->pipe; + + if (pipe->dentry != NULL) { + rpc_unlink(pipe->dentry); + pipe->dentry = NULL; + } +} + +static int gss_pipe_dentry_create(struct dentry *dir, + struct rpc_pipe_dir_object *pdo) +{ + struct gss_pipe *p = pdo->pdo_data; + struct dentry *dentry; + + dentry = rpc_mkpipe_dentry(dir, p->name, p->clnt, p->pipe); + if (IS_ERR(dentry)) + return PTR_ERR(dentry); + p->pipe->dentry = dentry; + return 0; +} + +static const struct rpc_pipe_dir_object_ops gss_pipe_dir_object_ops = { + .create = gss_pipe_dentry_create, + .destroy = gss_pipe_dentry_destroy, +}; + +static struct gss_pipe *gss_pipe_alloc(struct rpc_clnt *clnt, + const char *name, + const struct rpc_pipe_ops *upcall_ops) +{ + struct gss_pipe *p; + int err = -ENOMEM; + + p = kmalloc(sizeof(*p), GFP_KERNEL); + if (p == NULL) + goto err; + p->pipe = rpc_mkpipe_data(upcall_ops, RPC_PIPE_WAIT_FOR_OPEN); + if (IS_ERR(p->pipe)) { + err = PTR_ERR(p->pipe); + goto err_free_gss_pipe; + } + p->name = name; + p->clnt = clnt; + kref_init(&p->kref); + rpc_init_pipe_dir_object(&p->pdo, + &gss_pipe_dir_object_ops, + p); + return p; +err_free_gss_pipe: + kfree(p); +err: + return ERR_PTR(err); +} + +struct gss_alloc_pdo { + struct rpc_clnt *clnt; + const char *name; + const struct rpc_pipe_ops *upcall_ops; +}; + +static int gss_pipe_match_pdo(struct rpc_pipe_dir_object *pdo, void *data) +{ + struct gss_pipe *gss_pipe; + struct gss_alloc_pdo *args = data; + + if (pdo->pdo_ops != &gss_pipe_dir_object_ops) + return 0; + gss_pipe = container_of(pdo, struct gss_pipe, pdo); + if (strcmp(gss_pipe->name, args->name) != 0) + return 0; + if (!kref_get_unless_zero(&gss_pipe->kref)) + return 0; + return 1; +} + +static struct rpc_pipe_dir_object *gss_pipe_alloc_pdo(void *data) +{ + struct gss_pipe *gss_pipe; + struct gss_alloc_pdo *args = data; + + gss_pipe = gss_pipe_alloc(args->clnt, args->name, args->upcall_ops); + if (!IS_ERR(gss_pipe)) + return &gss_pipe->pdo; + return NULL; +} + +static struct gss_pipe *gss_pipe_get(struct rpc_clnt *clnt, + const char *name, + const struct rpc_pipe_ops *upcall_ops) +{ + struct net *net = rpc_net_ns(clnt); + struct rpc_pipe_dir_object *pdo; + struct gss_alloc_pdo args = { + .clnt = clnt, + .name = name, + .upcall_ops = upcall_ops, + }; + + pdo = rpc_find_or_alloc_pipe_dir_object(net, + &clnt->cl_pipedir_objects, + gss_pipe_match_pdo, + gss_pipe_alloc_pdo, + &args); + if (pdo != NULL) + return container_of(pdo, struct gss_pipe, pdo); + return ERR_PTR(-ENOMEM); +} + +static void __gss_pipe_free(struct gss_pipe *p) +{ + struct rpc_clnt *clnt = p->clnt; + struct net *net = rpc_net_ns(clnt); + + rpc_remove_pipe_dir_object(net, + &clnt->cl_pipedir_objects, + &p->pdo); + rpc_destroy_pipe_data(p->pipe); + kfree(p); +} + +static void __gss_pipe_release(struct kref *kref) +{ + struct gss_pipe *p = container_of(kref, struct gss_pipe, kref); + + __gss_pipe_free(p); +} + +static void gss_pipe_free(struct gss_pipe *p) +{ + if (p != NULL) + kref_put(&p->kref, __gss_pipe_release); +} + +/* + * NOTE: we have the opportunity to use different * parameters based on the input flavor (which must be a pseudoflavor) */ -static struct rpc_auth * -gss_create(struct rpc_clnt *clnt, rpc_authflavor_t flavor) +static struct gss_auth * +gss_create_new(struct rpc_auth_create_args *args, struct rpc_clnt *clnt) { + rpc_authflavor_t flavor = args->pseudoflavor; struct gss_auth *gss_auth; + struct gss_pipe *gss_pipe; struct rpc_auth * auth; int err = -ENOMEM; /* XXX? */ - dprintk("RPC: creating GSS authenticator for client %p\n",clnt); + dprintk("RPC: creating GSS authenticator for client %p\n", clnt); if (!try_module_get(THIS_MODULE)) return ERR_PTR(err); if (!(gss_auth = kmalloc(sizeof(*gss_auth), GFP_KERNEL))) goto out_dec; + INIT_HLIST_NODE(&gss_auth->hash); + gss_auth->target_name = NULL; + if (args->target_name) { + gss_auth->target_name = kstrdup(args->target_name, GFP_KERNEL); + if (gss_auth->target_name == NULL) + goto err_free; + } gss_auth->client = clnt; + gss_auth->net = get_net(rpc_net_ns(clnt)); err = -EINVAL; gss_auth->mech = gss_mech_get_by_pseudoflavor(flavor); if (!gss_auth->mech) { - printk(KERN_WARNING "%s: Pseudoflavor %d not found!", - __FUNCTION__, flavor); - goto err_free; + dprintk("RPC: Pseudoflavor %d not found!\n", flavor); + goto err_put_net; } gss_auth->service = gss_pseudoflavor_to_service(gss_auth->mech, flavor); if (gss_auth->service == 0) goto err_put_mech; - INIT_LIST_HEAD(&gss_auth->upcalls); - spin_lock_init(&gss_auth->lock); + if (!gssd_running(gss_auth->net)) + goto err_put_mech; auth = &gss_auth->rpc_auth; auth->au_cslack = GSS_CRED_SLACK >> 2; auth->au_rslack = GSS_VERF_SLACK >> 2; auth->au_ops = &authgss_ops; auth->au_flavor = flavor; atomic_set(&auth->au_count, 1); + kref_init(&gss_auth->kref); - err = rpcauth_init_credcache(auth, GSS_CRED_EXPIRE); + err = rpcauth_init_credcache(auth); if (err) goto err_put_mech; + /* + * Note: if we created the old pipe first, then someone who + * examined the directory at the right moment might conclude + * that we supported only the old pipe. So we instead create + * the new pipe first. + */ + gss_pipe = gss_pipe_get(clnt, "gssd", &gss_upcall_ops_v1); + if (IS_ERR(gss_pipe)) { + err = PTR_ERR(gss_pipe); + goto err_destroy_credcache; + } + gss_auth->gss_pipe[1] = gss_pipe; - snprintf(gss_auth->path, sizeof(gss_auth->path), "%s/%s", - clnt->cl_pathname, - gss_auth->mech->gm_name); - gss_auth->dentry = rpc_mkpipe(gss_auth->path, clnt, &gss_upcall_ops, RPC_PIPE_WAIT_FOR_OPEN); - if (IS_ERR(gss_auth->dentry)) { - err = PTR_ERR(gss_auth->dentry); - goto err_put_mech; + gss_pipe = gss_pipe_get(clnt, gss_auth->mech->gm_name, + &gss_upcall_ops_v0); + if (IS_ERR(gss_pipe)) { + err = PTR_ERR(gss_pipe); + goto err_destroy_pipe_1; } + gss_auth->gss_pipe[0] = gss_pipe; - return auth; + return gss_auth; +err_destroy_pipe_1: + gss_pipe_free(gss_auth->gss_pipe[1]); +err_destroy_credcache: + rpcauth_destroy_credcache(auth); err_put_mech: gss_mech_put(gss_auth->mech); +err_put_net: + put_net(gss_auth->net); err_free: + kfree(gss_auth->target_name); kfree(gss_auth); out_dec: module_put(THIS_MODULE); @@ -711,123 +1048,369 @@ out_dec: } static void +gss_free(struct gss_auth *gss_auth) +{ + gss_pipe_free(gss_auth->gss_pipe[0]); + gss_pipe_free(gss_auth->gss_pipe[1]); + gss_mech_put(gss_auth->mech); + put_net(gss_auth->net); + kfree(gss_auth->target_name); + + kfree(gss_auth); + module_put(THIS_MODULE); +} + +static void +gss_free_callback(struct kref *kref) +{ + struct gss_auth *gss_auth = container_of(kref, struct gss_auth, kref); + + gss_free(gss_auth); +} + +static void +gss_put_auth(struct gss_auth *gss_auth) +{ + kref_put(&gss_auth->kref, gss_free_callback); +} + +static void gss_destroy(struct rpc_auth *auth) { + struct gss_auth *gss_auth = container_of(auth, + struct gss_auth, rpc_auth); + + dprintk("RPC: destroying GSS authenticator %p flavor %d\n", + auth, auth->au_flavor); + + if (hash_hashed(&gss_auth->hash)) { + spin_lock(&gss_auth_hash_lock); + hash_del(&gss_auth->hash); + spin_unlock(&gss_auth_hash_lock); + } + + gss_pipe_free(gss_auth->gss_pipe[0]); + gss_auth->gss_pipe[0] = NULL; + gss_pipe_free(gss_auth->gss_pipe[1]); + gss_auth->gss_pipe[1] = NULL; + rpcauth_destroy_credcache(auth); + + gss_put_auth(gss_auth); +} + +/* + * Auths may be shared between rpc clients that were cloned from a + * common client with the same xprt, if they also share the flavor and + * target_name. + * + * The auth is looked up from the oldest parent sharing the same + * cl_xprt, and the auth itself references only that common parent + * (which is guaranteed to last as long as any of its descendants). + */ +static struct gss_auth * +gss_auth_find_or_add_hashed(struct rpc_auth_create_args *args, + struct rpc_clnt *clnt, + struct gss_auth *new) +{ struct gss_auth *gss_auth; + unsigned long hashval = (unsigned long)clnt; + + spin_lock(&gss_auth_hash_lock); + hash_for_each_possible(gss_auth_hash_table, + gss_auth, + hash, + hashval) { + if (gss_auth->client != clnt) + continue; + if (gss_auth->rpc_auth.au_flavor != args->pseudoflavor) + continue; + if (gss_auth->target_name != args->target_name) { + if (gss_auth->target_name == NULL) + continue; + if (args->target_name == NULL) + continue; + if (strcmp(gss_auth->target_name, args->target_name)) + continue; + } + if (!atomic_inc_not_zero(&gss_auth->rpc_auth.au_count)) + continue; + goto out; + } + if (new) + hash_add(gss_auth_hash_table, &new->hash, hashval); + gss_auth = new; +out: + spin_unlock(&gss_auth_hash_lock); + return gss_auth; +} - dprintk("RPC: destroying GSS authenticator %p flavor %d\n", - auth, auth->au_flavor); +static struct gss_auth * +gss_create_hashed(struct rpc_auth_create_args *args, struct rpc_clnt *clnt) +{ + struct gss_auth *gss_auth; + struct gss_auth *new; - gss_auth = container_of(auth, struct gss_auth, rpc_auth); - rpc_unlink(gss_auth->path); - gss_mech_put(gss_auth->mech); + gss_auth = gss_auth_find_or_add_hashed(args, clnt, NULL); + if (gss_auth != NULL) + goto out; + new = gss_create_new(args, clnt); + if (IS_ERR(new)) + return new; + gss_auth = gss_auth_find_or_add_hashed(args, clnt, new); + if (gss_auth != new) + gss_destroy(&new->rpc_auth); +out: + return gss_auth; +} - rpcauth_free_credcache(auth); - kfree(gss_auth); - module_put(THIS_MODULE); +static struct rpc_auth * +gss_create(struct rpc_auth_create_args *args, struct rpc_clnt *clnt) +{ + struct gss_auth *gss_auth; + struct rpc_xprt *xprt = rcu_access_pointer(clnt->cl_xprt); + + while (clnt != clnt->cl_parent) { + struct rpc_clnt *parent = clnt->cl_parent; + /* Find the original parent for this transport */ + if (rcu_access_pointer(parent->cl_xprt) != xprt) + break; + clnt = parent; + } + + gss_auth = gss_create_hashed(args, clnt); + if (IS_ERR(gss_auth)) + return ERR_CAST(gss_auth); + return &gss_auth->rpc_auth; +} + +/* + * gss_destroying_context will cause the RPCSEC_GSS to send a NULL RPC call + * to the server with the GSS control procedure field set to + * RPC_GSS_PROC_DESTROY. This should normally cause the server to release + * all RPCSEC_GSS state associated with that context. + */ +static int +gss_destroying_context(struct rpc_cred *cred) +{ + struct gss_cred *gss_cred = container_of(cred, struct gss_cred, gc_base); + struct gss_auth *gss_auth = container_of(cred->cr_auth, struct gss_auth, rpc_auth); + struct rpc_task *task; + + if (gss_cred->gc_ctx == NULL || + test_bit(RPCAUTH_CRED_UPTODATE, &cred->cr_flags) == 0) + return 0; + + gss_cred->gc_ctx->gc_proc = RPC_GSS_PROC_DESTROY; + cred->cr_ops = &gss_nullops; + + /* Take a reference to ensure the cred will be destroyed either + * by the RPC call or by the put_rpccred() below */ + get_rpccred(cred); + + task = rpc_call_null(gss_auth->client, cred, RPC_TASK_ASYNC|RPC_TASK_SOFT); + if (!IS_ERR(task)) + rpc_put_task(task); + + put_rpccred(cred); + return 1; } -/* gss_destroy_cred (and gss_destroy_ctx) are used to clean up after failure +/* gss_destroy_cred (and gss_free_ctx) are used to clean up after failure * to create a new cred or context, so they check that things have been * allocated before freeing them. */ static void -gss_destroy_ctx(struct gss_cl_ctx *ctx) +gss_do_free_ctx(struct gss_cl_ctx *ctx) { - dprintk("RPC: gss_destroy_ctx\n"); - - if (ctx->gc_gss_ctx) - gss_delete_sec_context(&ctx->gc_gss_ctx); + dprintk("RPC: %s\n", __func__); + gss_delete_sec_context(&ctx->gc_gss_ctx); kfree(ctx->gc_wire_ctx.data); kfree(ctx); } static void -gss_destroy_cred(struct rpc_cred *rc) +gss_free_ctx_callback(struct rcu_head *head) +{ + struct gss_cl_ctx *ctx = container_of(head, struct gss_cl_ctx, gc_rcu); + gss_do_free_ctx(ctx); +} + +static void +gss_free_ctx(struct gss_cl_ctx *ctx) +{ + call_rcu(&ctx->gc_rcu, gss_free_ctx_callback); +} + +static void +gss_free_cred(struct gss_cred *gss_cred) +{ + dprintk("RPC: %s cred=%p\n", __func__, gss_cred); + kfree(gss_cred); +} + +static void +gss_free_cred_callback(struct rcu_head *head) +{ + struct gss_cred *gss_cred = container_of(head, struct gss_cred, gc_base.cr_rcu); + gss_free_cred(gss_cred); +} + +static void +gss_destroy_nullcred(struct rpc_cred *cred) { - struct gss_cred *cred = container_of(rc, struct gss_cred, gc_base); + struct gss_cred *gss_cred = container_of(cred, struct gss_cred, gc_base); + struct gss_auth *gss_auth = container_of(cred->cr_auth, struct gss_auth, rpc_auth); + struct gss_cl_ctx *ctx = gss_cred->gc_ctx; + + RCU_INIT_POINTER(gss_cred->gc_ctx, NULL); + call_rcu(&cred->cr_rcu, gss_free_cred_callback); + if (ctx) + gss_put_ctx(ctx); + gss_put_auth(gss_auth); +} - dprintk("RPC: gss_destroy_cred \n"); +static void +gss_destroy_cred(struct rpc_cred *cred) +{ - if (cred->gc_ctx) - gss_put_ctx(cred->gc_ctx); - kfree(cred); + if (gss_destroying_context(cred)) + return; + gss_destroy_nullcred(cred); } /* * Lookup RPCSEC_GSS cred for the current process */ static struct rpc_cred * -gss_lookup_cred(struct rpc_auth *auth, struct auth_cred *acred, int taskflags) +gss_lookup_cred(struct rpc_auth *auth, struct auth_cred *acred, int flags) { - return rpcauth_lookup_credcache(auth, acred, taskflags); + return rpcauth_lookup_credcache(auth, acred, flags); } static struct rpc_cred * -gss_create_cred(struct rpc_auth *auth, struct auth_cred *acred, int taskflags) +gss_create_cred(struct rpc_auth *auth, struct auth_cred *acred, int flags) { struct gss_auth *gss_auth = container_of(auth, struct gss_auth, rpc_auth); struct gss_cred *cred = NULL; int err = -ENOMEM; - dprintk("RPC: gss_create_cred for uid %d, flavor %d\n", - acred->uid, auth->au_flavor); + dprintk("RPC: %s for uid %d, flavor %d\n", + __func__, from_kuid(&init_user_ns, acred->uid), + auth->au_flavor); - if (!(cred = kmalloc(sizeof(*cred), GFP_KERNEL))) + if (!(cred = kzalloc(sizeof(*cred), GFP_NOFS))) goto out_err; - memset(cred, 0, sizeof(*cred)); - atomic_set(&cred->gc_count, 1); - cred->gc_uid = acred->uid; + rpcauth_init_cred(&cred->gc_base, acred, auth, &gss_credops); /* * Note: in order to force a call to call_refresh(), we deliberately * fail to flag the credential as RPCAUTH_CRED_UPTODATE. */ - cred->gc_flags = 0; - cred->gc_base.cr_ops = &gss_credops; + cred->gc_base.cr_flags = 1UL << RPCAUTH_CRED_NEW; cred->gc_service = gss_auth->service; - err = gss_create_upcall(gss_auth, cred); - if (err < 0) - goto out_err; - + cred->gc_principal = NULL; + if (acred->machine_cred) + cred->gc_principal = acred->principal; + kref_get(&gss_auth->kref); return &cred->gc_base; out_err: - dprintk("RPC: gss_create_cred failed with error %d\n", err); - if (cred) gss_destroy_cred(&cred->gc_base); + dprintk("RPC: %s failed with error %d\n", __func__, err); return ERR_PTR(err); } static int -gss_match(struct auth_cred *acred, struct rpc_cred *rc, int taskflags) +gss_cred_init(struct rpc_auth *auth, struct rpc_cred *cred) +{ + struct gss_auth *gss_auth = container_of(auth, struct gss_auth, rpc_auth); + struct gss_cred *gss_cred = container_of(cred,struct gss_cred, gc_base); + int err; + + do { + err = gss_create_upcall(gss_auth, gss_cred); + } while (err == -EAGAIN); + return err; +} + +/* + * Returns -EACCES if GSS context is NULL or will expire within the + * timeout (miliseconds) + */ +static int +gss_key_timeout(struct rpc_cred *rc) +{ + struct gss_cred *gss_cred = container_of(rc, struct gss_cred, gc_base); + unsigned long now = jiffies; + unsigned long expire; + + if (gss_cred->gc_ctx == NULL) + return -EACCES; + + expire = gss_cred->gc_ctx->gc_expiry - (gss_key_expire_timeo * HZ); + + if (time_after(now, expire)) + return -EACCES; + return 0; +} + +static int +gss_match(struct auth_cred *acred, struct rpc_cred *rc, int flags) { struct gss_cred *gss_cred = container_of(rc, struct gss_cred, gc_base); + int ret; + if (test_bit(RPCAUTH_CRED_NEW, &rc->cr_flags)) + goto out; /* Don't match with creds that have expired. */ - if (gss_cred->gc_ctx && time_after(jiffies, gss_cred->gc_ctx->gc_expiry)) + if (time_after(jiffies, gss_cred->gc_ctx->gc_expiry)) + return 0; + if (!test_bit(RPCAUTH_CRED_UPTODATE, &rc->cr_flags)) return 0; - return (rc->cr_uid == acred->uid); +out: + if (acred->principal != NULL) { + if (gss_cred->gc_principal == NULL) + return 0; + ret = strcmp(acred->principal, gss_cred->gc_principal) == 0; + goto check_expire; + } + if (gss_cred->gc_principal != NULL) + return 0; + ret = uid_eq(rc->cr_uid, acred->uid); + +check_expire: + if (ret == 0) + return ret; + + /* Notify acred users of GSS context expiration timeout */ + if (test_bit(RPC_CRED_NOTIFY_TIMEOUT, &acred->ac_flags) && + (gss_key_timeout(rc) != 0)) { + /* test will now be done from generic cred */ + test_and_clear_bit(RPC_CRED_NOTIFY_TIMEOUT, &acred->ac_flags); + /* tell NFS layer that key will expire soon */ + set_bit(RPC_CRED_KEY_EXPIRE_SOON, &acred->ac_flags); + } + return ret; } /* * Marshal credentials. * Maybe we should keep a cached credential for performance reasons. */ -static u32 * -gss_marshal(struct rpc_task *task, u32 *p) +static __be32 * +gss_marshal(struct rpc_task *task, __be32 *p) { - struct rpc_cred *cred = task->tk_msg.rpc_cred; + struct rpc_rqst *req = task->tk_rqstp; + struct rpc_cred *cred = req->rq_cred; struct gss_cred *gss_cred = container_of(cred, struct gss_cred, gc_base); struct gss_cl_ctx *ctx = gss_cred_get_ctx(cred); - u32 *cred_len; - struct rpc_rqst *req = task->tk_rqstp; + __be32 *cred_len; u32 maj_stat = 0; struct xdr_netobj mic; struct kvec iov; struct xdr_buf verf_buf; - dprintk("RPC: %4u gss_marshal\n", task->tk_pid); + dprintk("RPC: %5u %s\n", task->tk_pid, __func__); *p++ = htonl(RPC_AUTH_GSS); cred_len = p++; @@ -845,7 +1428,7 @@ gss_marshal(struct rpc_task *task, u32 *p) /* We compute the checksum for the verifier over the xdr-encoded bytes * starting with the xid and ending at the end of the credential: */ - iov.iov_base = xprt_skip_transport_header(task->tk_xprt, + iov.iov_base = xprt_skip_transport_header(req->rq_xprt, req->rq_snd_buf.head[0].iov_base); iov.iov_len = (u8 *)p - (u8 *)iov.iov_base; xdr_buf_from_iov(&iov, &verf_buf); @@ -856,7 +1439,7 @@ gss_marshal(struct rpc_task *task, u32 *p) mic.data = (u8 *)(p + 1); maj_stat = gss_get_mic(ctx->gc_gss_ctx, &verf_buf, &mic); if (maj_stat == GSS_S_CONTEXT_EXPIRED) { - cred->cr_flags &= ~RPCAUTH_CRED_UPTODATE; + clear_bit(RPCAUTH_CRED_UPTODATE, &cred->cr_flags); } else if (maj_stat != 0) { printk("gss_marshal: gss_get_mic FAILED (%d)\n", maj_stat); goto out_put_ctx; @@ -869,35 +1452,96 @@ out_put_ctx: return NULL; } +static int gss_renew_cred(struct rpc_task *task) +{ + struct rpc_cred *oldcred = task->tk_rqstp->rq_cred; + struct gss_cred *gss_cred = container_of(oldcred, + struct gss_cred, + gc_base); + struct rpc_auth *auth = oldcred->cr_auth; + struct auth_cred acred = { + .uid = oldcred->cr_uid, + .principal = gss_cred->gc_principal, + .machine_cred = (gss_cred->gc_principal != NULL ? 1 : 0), + }; + struct rpc_cred *new; + + new = gss_lookup_cred(auth, &acred, RPCAUTH_LOOKUP_NEW); + if (IS_ERR(new)) + return PTR_ERR(new); + task->tk_rqstp->rq_cred = new; + put_rpccred(oldcred); + return 0; +} + +static int gss_cred_is_negative_entry(struct rpc_cred *cred) +{ + if (test_bit(RPCAUTH_CRED_NEGATIVE, &cred->cr_flags)) { + unsigned long now = jiffies; + unsigned long begin, expire; + struct gss_cred *gss_cred; + + gss_cred = container_of(cred, struct gss_cred, gc_base); + begin = gss_cred->gc_upcall_timestamp; + expire = begin + gss_expired_cred_retry_delay * HZ; + + if (time_in_range_open(now, begin, expire)) + return 1; + } + return 0; +} + /* * Refresh credentials. XXX - finish */ static int gss_refresh(struct rpc_task *task) { + struct rpc_cred *cred = task->tk_rqstp->rq_cred; + int ret = 0; + + if (gss_cred_is_negative_entry(cred)) + return -EKEYEXPIRED; + + if (!test_bit(RPCAUTH_CRED_NEW, &cred->cr_flags) && + !test_bit(RPCAUTH_CRED_UPTODATE, &cred->cr_flags)) { + ret = gss_renew_cred(task); + if (ret < 0) + goto out; + cred = task->tk_rqstp->rq_cred; + } + + if (test_bit(RPCAUTH_CRED_NEW, &cred->cr_flags)) + ret = gss_refresh_upcall(task); +out: + return ret; +} - if (!gss_cred_is_uptodate_ctx(task->tk_msg.rpc_cred)) - return gss_refresh_upcall(task); +/* Dummy refresh routine: used only when destroying the context */ +static int +gss_refresh_null(struct rpc_task *task) +{ return 0; } -static u32 * -gss_validate(struct rpc_task *task, u32 *p) +static __be32 * +gss_validate(struct rpc_task *task, __be32 *p) { - struct rpc_cred *cred = task->tk_msg.rpc_cred; + struct rpc_cred *cred = task->tk_rqstp->rq_cred; struct gss_cl_ctx *ctx = gss_cred_get_ctx(cred); - u32 seq; + __be32 seq; struct kvec iov; struct xdr_buf verf_buf; struct xdr_netobj mic; u32 flav,len; u32 maj_stat; + __be32 *ret = ERR_PTR(-EIO); - dprintk("RPC: %4u gss_validate\n", task->tk_pid); + dprintk("RPC: %5u %s\n", task->tk_pid, __func__); flav = ntohl(*p++); if ((len = ntohl(*p++)) > RPC_MAX_AUTH_SIZE) - goto out_bad; + goto out_bad; if (flav != RPC_AUTH_GSS) goto out_bad; seq = htonl(task->tk_rqstp->rq_seqno); @@ -907,33 +1551,49 @@ gss_validate(struct rpc_task *task, u32 *p) mic.data = (u8 *)p; mic.len = len; + ret = ERR_PTR(-EACCES); maj_stat = gss_verify_mic(ctx->gc_gss_ctx, &verf_buf, &mic); if (maj_stat == GSS_S_CONTEXT_EXPIRED) - cred->cr_flags &= ~RPCAUTH_CRED_UPTODATE; - if (maj_stat) + clear_bit(RPCAUTH_CRED_UPTODATE, &cred->cr_flags); + if (maj_stat) { + dprintk("RPC: %5u %s: gss_verify_mic returned error 0x%08x\n", + task->tk_pid, __func__, maj_stat); goto out_bad; + } /* We leave it to unwrap to calculate au_rslack. For now we just * calculate the length of the verifier: */ - task->tk_auth->au_verfsize = XDR_QUADLEN(len) + 2; + cred->cr_auth->au_verfsize = XDR_QUADLEN(len) + 2; gss_put_ctx(ctx); - dprintk("RPC: %4u GSS gss_validate: gss_verify_mic succeeded.\n", - task->tk_pid); + dprintk("RPC: %5u %s: gss_verify_mic succeeded.\n", + task->tk_pid, __func__); return p + XDR_QUADLEN(len); out_bad: gss_put_ctx(ctx); - dprintk("RPC: %4u gss_validate failed.\n", task->tk_pid); - return NULL; + dprintk("RPC: %5u %s failed ret %ld.\n", task->tk_pid, __func__, + PTR_ERR(ret)); + return ret; +} + +static void gss_wrap_req_encode(kxdreproc_t encode, struct rpc_rqst *rqstp, + __be32 *p, void *obj) +{ + struct xdr_stream xdr; + + xdr_init_encode(&xdr, &rqstp->rq_snd_buf, p); + encode(rqstp, &xdr, obj); } static inline int gss_wrap_req_integ(struct rpc_cred *cred, struct gss_cl_ctx *ctx, - kxdrproc_t encode, struct rpc_rqst *rqstp, u32 *p, void *obj) + kxdreproc_t encode, struct rpc_rqst *rqstp, + __be32 *p, void *obj) { struct xdr_buf *snd_buf = &rqstp->rq_snd_buf; struct xdr_buf integ_buf; - u32 *integ_len = NULL; + __be32 *integ_len = NULL; struct xdr_netobj mic; - u32 offset, *q; + u32 offset; + __be32 *q; struct kvec *iov; u32 maj_stat = 0; int status = -EIO; @@ -942,9 +1602,7 @@ gss_wrap_req_integ(struct rpc_cred *cred, struct gss_cl_ctx *ctx, offset = (u8 *)p - (u8 *)snd_buf->head[0].iov_base; *p++ = htonl(rqstp->rq_seqno); - status = encode(rqstp, p, obj); - if (status) - return status; + gss_wrap_req_encode(encode, rqstp, p, obj); if (xdr_buf_subsegment(snd_buf, &integ_buf, offset, snd_buf->len - offset)) @@ -952,7 +1610,7 @@ gss_wrap_req_integ(struct rpc_cred *cred, struct gss_cl_ctx *ctx, *integ_len = htonl(integ_buf.len); /* guess whether we're in the head or the tail: */ - if (snd_buf->page_len || snd_buf->tail[0].iov_len) + if (snd_buf->page_len || snd_buf->tail[0].iov_len) iov = snd_buf->tail; else iov = snd_buf->head; @@ -962,7 +1620,7 @@ gss_wrap_req_integ(struct rpc_cred *cred, struct gss_cl_ctx *ctx, maj_stat = gss_get_mic(ctx->gc_gss_ctx, &integ_buf, &mic); status = -EIO; /* XXX? */ if (maj_stat == GSS_S_CONTEXT_EXPIRED) - cred->cr_flags &= ~RPCAUTH_CRED_UPTODATE; + clear_bit(RPCAUTH_CRED_UPTODATE, &cred->cr_flags); else if (maj_stat) return status; q = xdr_encode_opaque(p, NULL, mic.len); @@ -1010,22 +1668,22 @@ alloc_enc_pages(struct rpc_rqst *rqstp) rqstp->rq_release_snd_buf = priv_release_snd_buf; return 0; out_free: - for (i--; i >= 0; i--) { - __free_page(rqstp->rq_enc_pages[i]); - } + rqstp->rq_enc_pages_num = i; + priv_release_snd_buf(rqstp); out: return -EAGAIN; } static inline int gss_wrap_req_priv(struct rpc_cred *cred, struct gss_cl_ctx *ctx, - kxdrproc_t encode, struct rpc_rqst *rqstp, u32 *p, void *obj) + kxdreproc_t encode, struct rpc_rqst *rqstp, + __be32 *p, void *obj) { struct xdr_buf *snd_buf = &rqstp->rq_snd_buf; u32 offset; u32 maj_stat; int status; - u32 *opaque_len; + __be32 *opaque_len; struct page **inpages; int first; int pad; @@ -1036,9 +1694,7 @@ gss_wrap_req_priv(struct rpc_cred *cred, struct gss_cl_ctx *ctx, offset = (u8 *)p - (u8 *)snd_buf->head[0].iov_base; *p++ = htonl(rqstp->rq_seqno); - status = encode(rqstp, p, obj); - if (status) - return status; + gss_wrap_req_encode(encode, rqstp, p, obj); status = alloc_enc_pages(rqstp); if (status) @@ -1047,21 +1703,27 @@ gss_wrap_req_priv(struct rpc_cred *cred, struct gss_cl_ctx *ctx, inpages = snd_buf->pages + first; snd_buf->pages = rqstp->rq_enc_pages; snd_buf->page_base -= first << PAGE_CACHE_SHIFT; - /* Give the tail its own page, in case we need extra space in the - * head when wrapping: */ + /* + * Give the tail its own page, in case we need extra space in the + * head when wrapping: + * + * call_allocate() allocates twice the slack space required + * by the authentication flavor to rq_callsize. + * For GSS, slack is GSS_CRED_SLACK. + */ if (snd_buf->page_len || snd_buf->tail[0].iov_len) { tmp = page_address(rqstp->rq_enc_pages[rqstp->rq_enc_pages_num - 1]); memcpy(tmp, snd_buf->tail[0].iov_base, snd_buf->tail[0].iov_len); snd_buf->tail[0].iov_base = tmp; } maj_stat = gss_wrap(ctx->gc_gss_ctx, offset, snd_buf, inpages); - /* RPC_SLACK_SPACE should prevent this ever happening: */ + /* slack space should prevent this ever happening: */ BUG_ON(snd_buf->len > snd_buf->buflen); - status = -EIO; + status = -EIO; /* We're assuming that when GSS_S_CONTEXT_EXPIRED, the encryption was * done anyway, so it's safe to put the request on the wire: */ if (maj_stat == GSS_S_CONTEXT_EXPIRED) - cred->cr_flags &= ~RPCAUTH_CRED_UPTODATE; + clear_bit(RPCAUTH_CRED_UPTODATE, &cred->cr_flags); else if (maj_stat) return status; @@ -1082,44 +1744,44 @@ gss_wrap_req_priv(struct rpc_cred *cred, struct gss_cl_ctx *ctx, static int gss_wrap_req(struct rpc_task *task, - kxdrproc_t encode, void *rqstp, u32 *p, void *obj) + kxdreproc_t encode, void *rqstp, __be32 *p, void *obj) { - struct rpc_cred *cred = task->tk_msg.rpc_cred; + struct rpc_cred *cred = task->tk_rqstp->rq_cred; struct gss_cred *gss_cred = container_of(cred, struct gss_cred, gc_base); struct gss_cl_ctx *ctx = gss_cred_get_ctx(cred); int status = -EIO; - dprintk("RPC: %4u gss_wrap_req\n", task->tk_pid); + dprintk("RPC: %5u %s\n", task->tk_pid, __func__); if (ctx->gc_proc != RPC_GSS_PROC_DATA) { /* The spec seems a little ambiguous here, but I think that not * wrapping context destruction requests makes the most sense. */ - status = encode(rqstp, p, obj); + gss_wrap_req_encode(encode, rqstp, p, obj); + status = 0; goto out; } switch (gss_cred->gc_service) { - case RPC_GSS_SVC_NONE: - status = encode(rqstp, p, obj); - break; - case RPC_GSS_SVC_INTEGRITY: - status = gss_wrap_req_integ(cred, ctx, encode, - rqstp, p, obj); - break; - case RPC_GSS_SVC_PRIVACY: - status = gss_wrap_req_priv(cred, ctx, encode, - rqstp, p, obj); - break; + case RPC_GSS_SVC_NONE: + gss_wrap_req_encode(encode, rqstp, p, obj); + status = 0; + break; + case RPC_GSS_SVC_INTEGRITY: + status = gss_wrap_req_integ(cred, ctx, encode, rqstp, p, obj); + break; + case RPC_GSS_SVC_PRIVACY: + status = gss_wrap_req_priv(cred, ctx, encode, rqstp, p, obj); + break; } out: gss_put_ctx(ctx); - dprintk("RPC: %4u gss_wrap_req returning %d\n", task->tk_pid, status); + dprintk("RPC: %5u %s returning %d\n", task->tk_pid, __func__, status); return status; } static inline int gss_unwrap_resp_integ(struct rpc_cred *cred, struct gss_cl_ctx *ctx, - struct rpc_rqst *rqstp, u32 **p) + struct rpc_rqst *rqstp, __be32 **p) { struct xdr_buf *rcv_buf = &rqstp->rq_rcv_buf; struct xdr_buf integ_buf; @@ -1148,7 +1810,7 @@ gss_unwrap_resp_integ(struct rpc_cred *cred, struct gss_cl_ctx *ctx, maj_stat = gss_verify_mic(ctx->gc_gss_ctx, &integ_buf, &mic); if (maj_stat == GSS_S_CONTEXT_EXPIRED) - cred->cr_flags &= ~RPCAUTH_CRED_UPTODATE; + clear_bit(RPCAUTH_CRED_UPTODATE, &cred->cr_flags); if (maj_stat != GSS_S_COMPLETE) return status; return 0; @@ -1156,7 +1818,7 @@ gss_unwrap_resp_integ(struct rpc_cred *cred, struct gss_cl_ctx *ctx, static inline int gss_unwrap_resp_priv(struct rpc_cred *cred, struct gss_cl_ctx *ctx, - struct rpc_rqst *rqstp, u32 **p) + struct rpc_rqst *rqstp, __be32 **p) { struct xdr_buf *rcv_buf = &rqstp->rq_rcv_buf; u32 offset; @@ -1173,7 +1835,7 @@ gss_unwrap_resp_priv(struct rpc_cred *cred, struct gss_cl_ctx *ctx, maj_stat = gss_unwrap(ctx->gc_gss_ctx, offset, rcv_buf); if (maj_stat == GSS_S_CONTEXT_EXPIRED) - cred->cr_flags &= ~RPCAUTH_CRED_UPTODATE; + clear_bit(RPCAUTH_CRED_UPTODATE, &cred->cr_flags); if (maj_stat != GSS_S_COMPLETE) return status; if (ntohl(*(*p)++) != rqstp->rq_seqno) @@ -1182,16 +1844,25 @@ gss_unwrap_resp_priv(struct rpc_cred *cred, struct gss_cl_ctx *ctx, return 0; } +static int +gss_unwrap_req_decode(kxdrdproc_t decode, struct rpc_rqst *rqstp, + __be32 *p, void *obj) +{ + struct xdr_stream xdr; + + xdr_init_decode(&xdr, &rqstp->rq_rcv_buf, p); + return decode(rqstp, &xdr, obj); +} static int gss_unwrap_resp(struct rpc_task *task, - kxdrproc_t decode, void *rqstp, u32 *p, void *obj) + kxdrdproc_t decode, void *rqstp, __be32 *p, void *obj) { - struct rpc_cred *cred = task->tk_msg.rpc_cred; + struct rpc_cred *cred = task->tk_rqstp->rq_cred; struct gss_cred *gss_cred = container_of(cred, struct gss_cred, gc_base); struct gss_cl_ctx *ctx = gss_cred_get_ctx(cred); - u32 *savedp = p; + __be32 *savedp = p; struct kvec *head = ((struct rpc_rqst *)rqstp)->rq_rcv_buf.head; int savedlen = head->iov_len; int status = -EIO; @@ -1199,61 +1870,101 @@ gss_unwrap_resp(struct rpc_task *task, if (ctx->gc_proc != RPC_GSS_PROC_DATA) goto out_decode; switch (gss_cred->gc_service) { - case RPC_GSS_SVC_NONE: - break; - case RPC_GSS_SVC_INTEGRITY: - status = gss_unwrap_resp_integ(cred, ctx, rqstp, &p); - if (status) - goto out; - break; - case RPC_GSS_SVC_PRIVACY: - status = gss_unwrap_resp_priv(cred, ctx, rqstp, &p); - if (status) - goto out; - break; + case RPC_GSS_SVC_NONE: + break; + case RPC_GSS_SVC_INTEGRITY: + status = gss_unwrap_resp_integ(cred, ctx, rqstp, &p); + if (status) + goto out; + break; + case RPC_GSS_SVC_PRIVACY: + status = gss_unwrap_resp_priv(cred, ctx, rqstp, &p); + if (status) + goto out; + break; } /* take into account extra slack for integrity and privacy cases: */ - task->tk_auth->au_rslack = task->tk_auth->au_verfsize + (p - savedp) + cred->cr_auth->au_rslack = cred->cr_auth->au_verfsize + (p - savedp) + (savedlen - head->iov_len); out_decode: - status = decode(rqstp, p, obj); + status = gss_unwrap_req_decode(decode, rqstp, p, obj); out: gss_put_ctx(ctx); - dprintk("RPC: %4u gss_unwrap_resp returning %d\n", task->tk_pid, - status); + dprintk("RPC: %5u %s returning %d\n", + task->tk_pid, __func__, status); return status; } - -static struct rpc_authops authgss_ops = { + +static const struct rpc_authops authgss_ops = { .owner = THIS_MODULE, .au_flavor = RPC_AUTH_GSS, -#ifdef RPC_DEBUG .au_name = "RPCSEC_GSS", -#endif .create = gss_create, .destroy = gss_destroy, .lookup_cred = gss_lookup_cred, - .crcreate = gss_create_cred + .crcreate = gss_create_cred, + .list_pseudoflavors = gss_mech_list_pseudoflavors, + .info2flavor = gss_mech_info2flavor, + .flavor2info = gss_mech_flavor2info, }; -static struct rpc_credops gss_credops = { +static const struct rpc_credops gss_credops = { .cr_name = "AUTH_GSS", .crdestroy = gss_destroy_cred, + .cr_init = gss_cred_init, + .crbind = rpcauth_generic_bind_cred, .crmatch = gss_match, .crmarshal = gss_marshal, .crrefresh = gss_refresh, .crvalidate = gss_validate, .crwrap_req = gss_wrap_req, .crunwrap_resp = gss_unwrap_resp, + .crkey_timeout = gss_key_timeout, }; -static struct rpc_pipe_ops gss_upcall_ops = { - .upcall = gss_pipe_upcall, +static const struct rpc_credops gss_nullops = { + .cr_name = "AUTH_GSS", + .crdestroy = gss_destroy_nullcred, + .crbind = rpcauth_generic_bind_cred, + .crmatch = gss_match, + .crmarshal = gss_marshal, + .crrefresh = gss_refresh_null, + .crvalidate = gss_validate, + .crwrap_req = gss_wrap_req, + .crunwrap_resp = gss_unwrap_resp, +}; + +static const struct rpc_pipe_ops gss_upcall_ops_v0 = { + .upcall = rpc_pipe_generic_upcall, + .downcall = gss_pipe_downcall, + .destroy_msg = gss_pipe_destroy_msg, + .open_pipe = gss_pipe_open_v0, + .release_pipe = gss_pipe_release, +}; + +static const struct rpc_pipe_ops gss_upcall_ops_v1 = { + .upcall = rpc_pipe_generic_upcall, .downcall = gss_pipe_downcall, .destroy_msg = gss_pipe_destroy_msg, + .open_pipe = gss_pipe_open_v1, .release_pipe = gss_pipe_release, }; +static __net_init int rpcsec_gss_init_net(struct net *net) +{ + return gss_svc_init_net(net); +} + +static __net_exit void rpcsec_gss_exit_net(struct net *net) +{ + gss_svc_shutdown_net(net); +} + +static struct pernet_operations rpcsec_gss_net_ops = { + .init = rpcsec_gss_init_net, + .exit = rpcsec_gss_exit_net, +}; + /* * Initialize RPCSEC_GSS module */ @@ -1267,7 +1978,13 @@ static int __init init_rpcsec_gss(void) err = gss_svc_init(); if (err) goto out_unregister; + err = register_pernet_subsys(&rpcsec_gss_net_ops); + if (err) + goto out_svc_exit; + rpc_init_wait_queue(&pipe_version_rpc_waitqueue, "gss pipe version"); return 0; +out_svc_exit: + gss_svc_shutdown(); out_unregister: rpcauth_unregister(&authgss_ops); out: @@ -1276,10 +1993,26 @@ out: static void __exit exit_rpcsec_gss(void) { + unregister_pernet_subsys(&rpcsec_gss_net_ops); gss_svc_shutdown(); rpcauth_unregister(&authgss_ops); + rcu_barrier(); /* Wait for completion of call_rcu()'s */ } +MODULE_ALIAS("rpc-auth-6"); MODULE_LICENSE("GPL"); +module_param_named(expired_cred_retry_delay, + gss_expired_cred_retry_delay, + uint, 0644); +MODULE_PARM_DESC(expired_cred_retry_delay, "Timeout (in seconds) until " + "the RPC engine retries an expired credential"); + +module_param_named(key_expire_timeo, + gss_key_expire_timeo, + uint, 0644); +MODULE_PARM_DESC(key_expire_timeo, "Time (in seconds) at the end of a " + "credential keys lifetime where the NFS layer cleans up " + "prior to key expiration"); + module_init(init_rpcsec_gss) module_exit(exit_rpcsec_gss) diff --git a/net/sunrpc/auth_gss/gss_generic_token.c b/net/sunrpc/auth_gss/gss_generic_token.c index 826df44e7fc..c586e92bcf7 100644 --- a/net/sunrpc/auth_gss/gss_generic_token.c +++ b/net/sunrpc/auth_gss/gss_generic_token.c @@ -11,7 +11,7 @@ /* * Copyright 1993 by OpenVision Technologies, Inc. - * + * * Permission to use, copy, modify, distribute, and sell this software * and its documentation for any purpose is hereby granted without fee, * provided that the above copyright notice appears in all copies and @@ -21,7 +21,7 @@ * without specific, written prior permission. OpenVision makes no * representations about the suitability of this software for any * purpose. It is provided "as is" without express or implied warranty. - * + * * OPENVISION DISCLAIMS ALL WARRANTIES WITH REGARD TO THIS SOFTWARE, * INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS, IN NO * EVENT SHALL OPENVISION BE LIABLE FOR ANY SPECIAL, INDIRECT OR @@ -33,7 +33,6 @@ #include <linux/types.h> #include <linux/module.h> -#include <linux/slab.h> #include <linux/string.h> #include <linux/sunrpc/sched.h> #include <linux/sunrpc/gss_asn1.h> @@ -77,19 +76,19 @@ static int der_length_size( int length) { if (length < (1<<7)) - return(1); + return 1; else if (length < (1<<8)) - return(2); + return 2; #if (SIZEOF_INT == 2) else - return(3); + return 3; #else else if (length < (1<<16)) - return(3); + return 3; else if (length < (1<<24)) - return(4); + return 4; else - return(5); + return 5; #endif } @@ -122,14 +121,14 @@ der_read_length(unsigned char **buf, int *bufsize) int ret; if (*bufsize < 1) - return(-1); + return -1; sf = *(*buf)++; (*bufsize)--; if (sf & 0x80) { if ((sf &= 0x7f) > ((*bufsize)-1)) - return(-1); + return -1; if (sf > SIZEOF_INT) - return (-1); + return -1; ret = 0; for (; sf; sf--) { ret = (ret<<8) + (*(*buf)++); @@ -139,7 +138,7 @@ der_read_length(unsigned char **buf, int *bufsize) ret = sf; } - return(ret); + return ret; } /* returns the length of a token, given the mech oid and the body size */ @@ -148,11 +147,11 @@ int g_token_size(struct xdr_netobj *mech, unsigned int body_size) { /* set body_size to sequence contents size */ - body_size += 4 + (int) mech->len; /* NEED overflow check */ - return(1 + der_length_size(body_size) + body_size); + body_size += 2 + (int) mech->len; /* NEED overflow check */ + return 1 + der_length_size(body_size) + body_size; } -EXPORT_SYMBOL(g_token_size); +EXPORT_SYMBOL_GPL(g_token_size); /* fills in a buffer with the token header. The buffer is assumed to be the right size. buf is advanced past the token header */ @@ -161,13 +160,13 @@ void g_make_token_header(struct xdr_netobj *mech, int body_size, unsigned char **buf) { *(*buf)++ = 0x60; - der_write_length(buf, 4 + mech->len + body_size); + der_write_length(buf, 2 + mech->len + body_size); *(*buf)++ = 0x06; *(*buf)++ = (unsigned char) mech->len; TWRITE_STR(*buf, mech->data, ((int) mech->len)); } -EXPORT_SYMBOL(g_make_token_header); +EXPORT_SYMBOL_GPL(g_make_token_header); /* * Given a buffer containing a token, reads and verifies the token, @@ -187,49 +186,49 @@ g_verify_token_header(struct xdr_netobj *mech, int *body_size, int ret = 0; if ((toksize-=1) < 0) - return(G_BAD_TOK_HEADER); + return G_BAD_TOK_HEADER; if (*buf++ != 0x60) - return(G_BAD_TOK_HEADER); + return G_BAD_TOK_HEADER; if ((seqsize = der_read_length(&buf, &toksize)) < 0) - return(G_BAD_TOK_HEADER); + return G_BAD_TOK_HEADER; if (seqsize != toksize) - return(G_BAD_TOK_HEADER); + return G_BAD_TOK_HEADER; if ((toksize-=1) < 0) - return(G_BAD_TOK_HEADER); + return G_BAD_TOK_HEADER; if (*buf++ != 0x06) - return(G_BAD_TOK_HEADER); - + return G_BAD_TOK_HEADER; + if ((toksize-=1) < 0) - return(G_BAD_TOK_HEADER); + return G_BAD_TOK_HEADER; toid.len = *buf++; if ((toksize-=toid.len) < 0) - return(G_BAD_TOK_HEADER); + return G_BAD_TOK_HEADER; toid.data = buf; buf+=toid.len; - if (! g_OID_equal(&toid, mech)) + if (! g_OID_equal(&toid, mech)) ret = G_WRONG_MECH; - + /* G_WRONG_MECH is not returned immediately because it's more important to return G_BAD_TOK_HEADER if the token header is in fact bad */ if ((toksize-=2) < 0) - return(G_BAD_TOK_HEADER); + return G_BAD_TOK_HEADER; if (ret) - return(ret); + return ret; if (!ret) { *buf_in = buf; *body_size = toksize; } - return(ret); + return ret; } -EXPORT_SYMBOL(g_verify_token_header); +EXPORT_SYMBOL_GPL(g_verify_token_header); diff --git a/net/sunrpc/auth_gss/gss_krb5_crypto.c b/net/sunrpc/auth_gss/gss_krb5_crypto.c index 97c981fa6b8..0f43e894bc0 100644 --- a/net/sunrpc/auth_gss/gss_krb5_crypto.c +++ b/net/sunrpc/auth_gss/gss_krb5_crypto.c @@ -1,7 +1,7 @@ /* * linux/net/sunrpc/gss_krb5_crypto.c * - * Copyright (c) 2000 The Regents of the University of Michigan. + * Copyright (c) 2000-2008 The Regents of the University of Michigan. * All rights reserved. * * Andy Adamson <andros@umich.edu> @@ -34,14 +34,16 @@ * WARRANTIES OF MERCHANTIBILITY AND FITNESS FOR A PARTICULAR PURPOSE. */ +#include <linux/err.h> #include <linux/types.h> #include <linux/mm.h> -#include <linux/slab.h> #include <linux/scatterlist.h> #include <linux/crypto.h> #include <linux/highmem.h> #include <linux/pagemap.h> +#include <linux/random.h> #include <linux/sunrpc/gss_krb5.h> +#include <linux/sunrpc/xdr.h> #ifdef RPC_DEBUG # define RPCDBG_FACILITY RPCDBG_AUTH @@ -49,48 +51,41 @@ u32 krb5_encrypt( - struct crypto_tfm *tfm, + struct crypto_blkcipher *tfm, void * iv, void * in, void * out, int length) { u32 ret = -EINVAL; - struct scatterlist sg[1]; - u8 local_iv[16] = {0}; - - dprintk("RPC: krb5_encrypt: input data:\n"); - print_hexl((u32 *)in, length, 0); + struct scatterlist sg[1]; + u8 local_iv[GSS_KRB5_MAX_BLOCKSIZE] = {0}; + struct blkcipher_desc desc = { .tfm = tfm, .info = local_iv }; - if (length % crypto_tfm_alg_blocksize(tfm) != 0) + if (length % crypto_blkcipher_blocksize(tfm) != 0) goto out; - if (crypto_tfm_alg_ivsize(tfm) > 16) { - dprintk("RPC: gss_k5encrypt: tfm iv size to large %d\n", - crypto_tfm_alg_ivsize(tfm)); + if (crypto_blkcipher_ivsize(tfm) > GSS_KRB5_MAX_BLOCKSIZE) { + dprintk("RPC: gss_k5encrypt: tfm iv size too large %d\n", + crypto_blkcipher_ivsize(tfm)); goto out; } if (iv) - memcpy(local_iv, iv, crypto_tfm_alg_ivsize(tfm)); + memcpy(local_iv, iv, crypto_blkcipher_ivsize(tfm)); memcpy(out, in, length); - sg_set_buf(sg, out, length); + sg_init_one(sg, out, length); - ret = crypto_cipher_encrypt_iv(tfm, sg, sg, length, local_iv); - - dprintk("RPC: krb5_encrypt: output data:\n"); - print_hexl((u32 *)out, length, 0); + ret = crypto_blkcipher_encrypt_iv(&desc, sg, sg, length); out: - dprintk("RPC: krb5_encrypt returns %d\n",ret); - return(ret); + dprintk("RPC: krb5_encrypt returns %d\n", ret); + return ret; } -EXPORT_SYMBOL(krb5_encrypt); - u32 krb5_decrypt( - struct crypto_tfm *tfm, + struct crypto_blkcipher *tfm, void * iv, void * in, void * out, @@ -98,154 +93,301 @@ krb5_decrypt( { u32 ret = -EINVAL; struct scatterlist sg[1]; - u8 local_iv[16] = {0}; - - dprintk("RPC: krb5_decrypt: input data:\n"); - print_hexl((u32 *)in, length, 0); + u8 local_iv[GSS_KRB5_MAX_BLOCKSIZE] = {0}; + struct blkcipher_desc desc = { .tfm = tfm, .info = local_iv }; - if (length % crypto_tfm_alg_blocksize(tfm) != 0) + if (length % crypto_blkcipher_blocksize(tfm) != 0) goto out; - if (crypto_tfm_alg_ivsize(tfm) > 16) { - dprintk("RPC: gss_k5decrypt: tfm iv size to large %d\n", - crypto_tfm_alg_ivsize(tfm)); + if (crypto_blkcipher_ivsize(tfm) > GSS_KRB5_MAX_BLOCKSIZE) { + dprintk("RPC: gss_k5decrypt: tfm iv size too large %d\n", + crypto_blkcipher_ivsize(tfm)); goto out; } if (iv) - memcpy(local_iv,iv, crypto_tfm_alg_ivsize(tfm)); + memcpy(local_iv,iv, crypto_blkcipher_ivsize(tfm)); memcpy(out, in, length); - sg_set_buf(sg, out, length); - - ret = crypto_cipher_decrypt_iv(tfm, sg, sg, length, local_iv); + sg_init_one(sg, out, length); - dprintk("RPC: krb5_decrypt: output_data:\n"); - print_hexl((u32 *)out, length, 0); + ret = crypto_blkcipher_decrypt_iv(&desc, sg, sg, length); out: - dprintk("RPC: gss_k5decrypt returns %d\n",ret); - return(ret); + dprintk("RPC: gss_k5decrypt returns %d\n",ret); + return ret; } -EXPORT_SYMBOL(krb5_decrypt); +static int +checksummer(struct scatterlist *sg, void *data) +{ + struct hash_desc *desc = data; + + return crypto_hash_update(desc, sg, sg->length); +} static int -process_xdr_buf(struct xdr_buf *buf, int offset, int len, - int (*actor)(struct scatterlist *, void *), void *data) +arcfour_hmac_md5_usage_to_salt(unsigned int usage, u8 salt[4]) { - int i, page_len, thislen, page_offset, ret = 0; - struct scatterlist sg[1]; + unsigned int ms_usage; + + switch (usage) { + case KG_USAGE_SIGN: + ms_usage = 15; + break; + case KG_USAGE_SEAL: + ms_usage = 13; + break; + default: + return -EINVAL; + } + salt[0] = (ms_usage >> 0) & 0xff; + salt[1] = (ms_usage >> 8) & 0xff; + salt[2] = (ms_usage >> 16) & 0xff; + salt[3] = (ms_usage >> 24) & 0xff; - if (offset >= buf->head[0].iov_len) { - offset -= buf->head[0].iov_len; - } else { - thislen = buf->head[0].iov_len - offset; - if (thislen > len) - thislen = len; - sg_set_buf(sg, buf->head[0].iov_base + offset, thislen); - ret = actor(sg, data); - if (ret) - goto out; - offset = 0; - len -= thislen; + return 0; +} + +static u32 +make_checksum_hmac_md5(struct krb5_ctx *kctx, char *header, int hdrlen, + struct xdr_buf *body, int body_offset, u8 *cksumkey, + unsigned int usage, struct xdr_netobj *cksumout) +{ + struct hash_desc desc; + struct scatterlist sg[1]; + int err; + u8 checksumdata[GSS_KRB5_MAX_CKSUM_LEN]; + u8 rc4salt[4]; + struct crypto_hash *md5; + struct crypto_hash *hmac_md5; + + if (cksumkey == NULL) + return GSS_S_FAILURE; + + if (cksumout->len < kctx->gk5e->cksumlength) { + dprintk("%s: checksum buffer length, %u, too small for %s\n", + __func__, cksumout->len, kctx->gk5e->name); + return GSS_S_FAILURE; } - if (len == 0) - goto out; - if (offset >= buf->page_len) { - offset -= buf->page_len; - } else { - page_len = buf->page_len - offset; - if (page_len > len) - page_len = len; - len -= page_len; - page_offset = (offset + buf->page_base) & (PAGE_CACHE_SIZE - 1); - i = (offset + buf->page_base) >> PAGE_CACHE_SHIFT; - thislen = PAGE_CACHE_SIZE - page_offset; - do { - if (thislen > page_len) - thislen = page_len; - sg->page = buf->pages[i]; - sg->offset = page_offset; - sg->length = thislen; - ret = actor(sg, data); - if (ret) - goto out; - page_len -= thislen; - i++; - page_offset = 0; - thislen = PAGE_CACHE_SIZE; - } while (page_len != 0); - offset = 0; + if (arcfour_hmac_md5_usage_to_salt(usage, rc4salt)) { + dprintk("%s: invalid usage value %u\n", __func__, usage); + return GSS_S_FAILURE; } - if (len == 0) - goto out; - if (offset < buf->tail[0].iov_len) { - thislen = buf->tail[0].iov_len - offset; - if (thislen > len) - thislen = len; - sg_set_buf(sg, buf->tail[0].iov_base + offset, thislen); - ret = actor(sg, data); - len -= thislen; + md5 = crypto_alloc_hash("md5", 0, CRYPTO_ALG_ASYNC); + if (IS_ERR(md5)) + return GSS_S_FAILURE; + + hmac_md5 = crypto_alloc_hash(kctx->gk5e->cksum_name, 0, + CRYPTO_ALG_ASYNC); + if (IS_ERR(hmac_md5)) { + crypto_free_hash(md5); + return GSS_S_FAILURE; } - if (len != 0) - ret = -EINVAL; -out: - return ret; -} -static int -checksummer(struct scatterlist *sg, void *data) -{ - struct crypto_tfm *tfm = (struct crypto_tfm *)data; + desc.tfm = md5; + desc.flags = CRYPTO_TFM_REQ_MAY_SLEEP; + + err = crypto_hash_init(&desc); + if (err) + goto out; + sg_init_one(sg, rc4salt, 4); + err = crypto_hash_update(&desc, sg, 4); + if (err) + goto out; - crypto_digest_update(tfm, sg, 1); + sg_init_one(sg, header, hdrlen); + err = crypto_hash_update(&desc, sg, hdrlen); + if (err) + goto out; + err = xdr_process_buf(body, body_offset, body->len - body_offset, + checksummer, &desc); + if (err) + goto out; + err = crypto_hash_final(&desc, checksumdata); + if (err) + goto out; - return 0; + desc.tfm = hmac_md5; + desc.flags = CRYPTO_TFM_REQ_MAY_SLEEP; + + err = crypto_hash_init(&desc); + if (err) + goto out; + err = crypto_hash_setkey(hmac_md5, cksumkey, kctx->gk5e->keylength); + if (err) + goto out; + + sg_init_one(sg, checksumdata, crypto_hash_digestsize(md5)); + err = crypto_hash_digest(&desc, sg, crypto_hash_digestsize(md5), + checksumdata); + if (err) + goto out; + + memcpy(cksumout->data, checksumdata, kctx->gk5e->cksumlength); + cksumout->len = kctx->gk5e->cksumlength; +out: + crypto_free_hash(md5); + crypto_free_hash(hmac_md5); + return err ? GSS_S_FAILURE : 0; } -/* checksum the plaintext data and hdrlen bytes of the token header */ -s32 -make_checksum(s32 cksumtype, char *header, int hdrlen, struct xdr_buf *body, - int body_offset, struct xdr_netobj *cksum) +/* + * checksum the plaintext data and hdrlen bytes of the token header + * The checksum is performed over the first 8 bytes of the + * gss token header and then over the data body + */ +u32 +make_checksum(struct krb5_ctx *kctx, char *header, int hdrlen, + struct xdr_buf *body, int body_offset, u8 *cksumkey, + unsigned int usage, struct xdr_netobj *cksumout) { - char *cksumname; - struct crypto_tfm *tfm = NULL; /* XXX add to ctx? */ + struct hash_desc desc; struct scatterlist sg[1]; - u32 code = GSS_S_FAILURE; - - switch (cksumtype) { - case CKSUMTYPE_RSA_MD5: - cksumname = "md5"; - break; - default: - dprintk("RPC: krb5_make_checksum:" - " unsupported checksum %d", cksumtype); + int err; + u8 checksumdata[GSS_KRB5_MAX_CKSUM_LEN]; + unsigned int checksumlen; + + if (kctx->gk5e->ctype == CKSUMTYPE_HMAC_MD5_ARCFOUR) + return make_checksum_hmac_md5(kctx, header, hdrlen, + body, body_offset, + cksumkey, usage, cksumout); + + if (cksumout->len < kctx->gk5e->cksumlength) { + dprintk("%s: checksum buffer length, %u, too small for %s\n", + __func__, cksumout->len, kctx->gk5e->name); + return GSS_S_FAILURE; + } + + desc.tfm = crypto_alloc_hash(kctx->gk5e->cksum_name, 0, CRYPTO_ALG_ASYNC); + if (IS_ERR(desc.tfm)) + return GSS_S_FAILURE; + desc.flags = CRYPTO_TFM_REQ_MAY_SLEEP; + + checksumlen = crypto_hash_digestsize(desc.tfm); + + if (cksumkey != NULL) { + err = crypto_hash_setkey(desc.tfm, cksumkey, + kctx->gk5e->keylength); + if (err) goto out; } - if (!(tfm = crypto_alloc_tfm(cksumname, CRYPTO_TFM_REQ_MAY_SLEEP))) + + err = crypto_hash_init(&desc); + if (err) goto out; - cksum->len = crypto_tfm_alg_digestsize(tfm); - if ((cksum->data = kmalloc(cksum->len, GFP_KERNEL)) == NULL) + sg_init_one(sg, header, hdrlen); + err = crypto_hash_update(&desc, sg, hdrlen); + if (err) + goto out; + err = xdr_process_buf(body, body_offset, body->len - body_offset, + checksummer, &desc); + if (err) + goto out; + err = crypto_hash_final(&desc, checksumdata); + if (err) goto out; - crypto_digest_init(tfm); - sg_set_buf(sg, header, hdrlen); - crypto_digest_update(tfm, sg, 1); - process_xdr_buf(body, body_offset, body->len - body_offset, - checksummer, tfm); - crypto_digest_final(tfm, cksum->data); - code = 0; + switch (kctx->gk5e->ctype) { + case CKSUMTYPE_RSA_MD5: + err = kctx->gk5e->encrypt(kctx->seq, NULL, checksumdata, + checksumdata, checksumlen); + if (err) + goto out; + memcpy(cksumout->data, + checksumdata + checksumlen - kctx->gk5e->cksumlength, + kctx->gk5e->cksumlength); + break; + case CKSUMTYPE_HMAC_SHA1_DES3: + memcpy(cksumout->data, checksumdata, kctx->gk5e->cksumlength); + break; + default: + BUG(); + break; + } + cksumout->len = kctx->gk5e->cksumlength; out: - crypto_free_tfm(tfm); - return code; + crypto_free_hash(desc.tfm); + return err ? GSS_S_FAILURE : 0; } -EXPORT_SYMBOL(make_checksum); +/* + * checksum the plaintext data and hdrlen bytes of the token header + * Per rfc4121, sec. 4.2.4, the checksum is performed over the data + * body then over the first 16 octets of the MIC token + * Inclusion of the header data in the calculation of the + * checksum is optional. + */ +u32 +make_checksum_v2(struct krb5_ctx *kctx, char *header, int hdrlen, + struct xdr_buf *body, int body_offset, u8 *cksumkey, + unsigned int usage, struct xdr_netobj *cksumout) +{ + struct hash_desc desc; + struct scatterlist sg[1]; + int err; + u8 checksumdata[GSS_KRB5_MAX_CKSUM_LEN]; + unsigned int checksumlen; + + if (kctx->gk5e->keyed_cksum == 0) { + dprintk("%s: expected keyed hash for %s\n", + __func__, kctx->gk5e->name); + return GSS_S_FAILURE; + } + if (cksumkey == NULL) { + dprintk("%s: no key supplied for %s\n", + __func__, kctx->gk5e->name); + return GSS_S_FAILURE; + } + + desc.tfm = crypto_alloc_hash(kctx->gk5e->cksum_name, 0, + CRYPTO_ALG_ASYNC); + if (IS_ERR(desc.tfm)) + return GSS_S_FAILURE; + checksumlen = crypto_hash_digestsize(desc.tfm); + desc.flags = CRYPTO_TFM_REQ_MAY_SLEEP; + + err = crypto_hash_setkey(desc.tfm, cksumkey, kctx->gk5e->keylength); + if (err) + goto out; + + err = crypto_hash_init(&desc); + if (err) + goto out; + err = xdr_process_buf(body, body_offset, body->len - body_offset, + checksummer, &desc); + if (err) + goto out; + if (header != NULL) { + sg_init_one(sg, header, hdrlen); + err = crypto_hash_update(&desc, sg, hdrlen); + if (err) + goto out; + } + err = crypto_hash_final(&desc, checksumdata); + if (err) + goto out; + + cksumout->len = kctx->gk5e->cksumlength; + + switch (kctx->gk5e->ctype) { + case CKSUMTYPE_HMAC_SHA1_96_AES128: + case CKSUMTYPE_HMAC_SHA1_96_AES256: + /* note that this truncates the hash */ + memcpy(cksumout->data, checksumdata, kctx->gk5e->cksumlength); + break; + default: + BUG(); + break; + } +out: + crypto_free_hash(desc.tfm); + return err ? GSS_S_FAILURE : 0; +} struct encryptor_desc { - u8 iv[8]; /* XXX hard-coded blocksize */ - struct crypto_tfm *tfm; + u8 iv[GSS_KRB5_MAX_BLOCKSIZE]; + struct blkcipher_desc desc; int pos; struct xdr_buf *outbuf; struct page **pages; @@ -268,8 +410,6 @@ encryptor(struct scatterlist *sg, void *data) /* Worst case is 4 fragments: head, end of page 1, start * of page 2, tail. Anything more is a bug. */ BUG_ON(desc->fragno > 3); - desc->infrags[desc->fragno] = *sg; - desc->outfrags[desc->fragno] = *sg; page_pos = desc->pos - outbuf->head[0].iov_len; if (page_pos >= 0 && page_pos < outbuf->page_len) { @@ -277,29 +417,38 @@ encryptor(struct scatterlist *sg, void *data) int i = (page_pos + outbuf->page_base) >> PAGE_CACHE_SHIFT; in_page = desc->pages[i]; } else { - in_page = sg->page; + in_page = sg_page(sg); } - desc->infrags[desc->fragno].page = in_page; + sg_set_page(&desc->infrags[desc->fragno], in_page, sg->length, + sg->offset); + sg_set_page(&desc->outfrags[desc->fragno], sg_page(sg), sg->length, + sg->offset); desc->fragno++; desc->fraglen += sg->length; desc->pos += sg->length; - fraglen = thislen & 7; /* XXX hardcoded blocksize */ + fraglen = thislen & (crypto_blkcipher_blocksize(desc->desc.tfm) - 1); thislen -= fraglen; if (thislen == 0) return 0; - ret = crypto_cipher_encrypt_iv(desc->tfm, desc->outfrags, desc->infrags, - thislen, desc->iv); + sg_mark_end(&desc->infrags[desc->fragno - 1]); + sg_mark_end(&desc->outfrags[desc->fragno - 1]); + + ret = crypto_blkcipher_encrypt_iv(&desc->desc, desc->outfrags, + desc->infrags, thislen); if (ret) return ret; + + sg_init_table(desc->infrags, 4); + sg_init_table(desc->outfrags, 4); + if (fraglen) { - desc->outfrags[0].page = sg->page; - desc->outfrags[0].offset = sg->offset + sg->length - fraglen; - desc->outfrags[0].length = fraglen; + sg_set_page(&desc->outfrags[0], sg_page(sg), fraglen, + sg->offset + sg->length - fraglen); desc->infrags[0] = desc->outfrags[0]; - desc->infrags[0].page = in_page; + sg_assign_page(&desc->infrags[0], in_page); desc->fragno = 1; desc->fraglen = fraglen; } else { @@ -310,31 +459,34 @@ encryptor(struct scatterlist *sg, void *data) } int -gss_encrypt_xdr_buf(struct crypto_tfm *tfm, struct xdr_buf *buf, int offset, - struct page **pages) +gss_encrypt_xdr_buf(struct crypto_blkcipher *tfm, struct xdr_buf *buf, + int offset, struct page **pages) { int ret; struct encryptor_desc desc; - BUG_ON((buf->len - offset) % crypto_tfm_alg_blocksize(tfm) != 0); + BUG_ON((buf->len - offset) % crypto_blkcipher_blocksize(tfm) != 0); memset(desc.iv, 0, sizeof(desc.iv)); - desc.tfm = tfm; + desc.desc.tfm = tfm; + desc.desc.info = desc.iv; + desc.desc.flags = 0; desc.pos = offset; desc.outbuf = buf; desc.pages = pages; desc.fragno = 0; desc.fraglen = 0; - ret = process_xdr_buf(buf, offset, buf->len - offset, encryptor, &desc); + sg_init_table(desc.infrags, 4); + sg_init_table(desc.outfrags, 4); + + ret = xdr_process_buf(buf, offset, buf->len - offset, encryptor, &desc); return ret; } -EXPORT_SYMBOL(gss_encrypt_xdr_buf); - struct decryptor_desc { - u8 iv[8]; /* XXX hard-coded blocksize */ - struct crypto_tfm *tfm; + u8 iv[GSS_KRB5_MAX_BLOCKSIZE]; + struct blkcipher_desc desc; struct scatterlist frags[4]; int fragno; int fraglen; @@ -350,24 +502,29 @@ decryptor(struct scatterlist *sg, void *data) /* Worst case is 4 fragments: head, end of page 1, start * of page 2, tail. Anything more is a bug. */ BUG_ON(desc->fragno > 3); - desc->frags[desc->fragno] = *sg; + sg_set_page(&desc->frags[desc->fragno], sg_page(sg), sg->length, + sg->offset); desc->fragno++; desc->fraglen += sg->length; - fraglen = thislen & 7; /* XXX hardcoded blocksize */ + fraglen = thislen & (crypto_blkcipher_blocksize(desc->desc.tfm) - 1); thislen -= fraglen; if (thislen == 0) return 0; - ret = crypto_cipher_decrypt_iv(desc->tfm, desc->frags, desc->frags, - thislen, desc->iv); + sg_mark_end(&desc->frags[desc->fragno - 1]); + + ret = crypto_blkcipher_decrypt_iv(&desc->desc, desc->frags, + desc->frags, thislen); if (ret) return ret; + + sg_init_table(desc->frags, 4); + if (fraglen) { - desc->frags[0].page = sg->page; - desc->frags[0].offset = sg->offset + sg->length - fraglen; - desc->frags[0].length = fraglen; + sg_set_page(&desc->frags[0], sg_page(sg), fraglen, + sg->offset + sg->length - fraglen); desc->fragno = 1; desc->fraglen = fraglen; } else { @@ -378,18 +535,459 @@ decryptor(struct scatterlist *sg, void *data) } int -gss_decrypt_xdr_buf(struct crypto_tfm *tfm, struct xdr_buf *buf, int offset) +gss_decrypt_xdr_buf(struct crypto_blkcipher *tfm, struct xdr_buf *buf, + int offset) { struct decryptor_desc desc; /* XXXJBF: */ - BUG_ON((buf->len - offset) % crypto_tfm_alg_blocksize(tfm) != 0); + BUG_ON((buf->len - offset) % crypto_blkcipher_blocksize(tfm) != 0); memset(desc.iv, 0, sizeof(desc.iv)); - desc.tfm = tfm; + desc.desc.tfm = tfm; + desc.desc.info = desc.iv; + desc.desc.flags = 0; desc.fragno = 0; desc.fraglen = 0; - return process_xdr_buf(buf, offset, buf->len - offset, decryptor, &desc); + + sg_init_table(desc.frags, 4); + + return xdr_process_buf(buf, offset, buf->len - offset, decryptor, &desc); +} + +/* + * This function makes the assumption that it was ultimately called + * from gss_wrap(). + * + * The client auth_gss code moves any existing tail data into a + * separate page before calling gss_wrap. + * The server svcauth_gss code ensures that both the head and the + * tail have slack space of RPC_MAX_AUTH_SIZE before calling gss_wrap. + * + * Even with that guarantee, this function may be called more than + * once in the processing of gss_wrap(). The best we can do is + * verify at compile-time (see GSS_KRB5_SLACK_CHECK) that the + * largest expected shift will fit within RPC_MAX_AUTH_SIZE. + * At run-time we can verify that a single invocation of this + * function doesn't attempt to use more the RPC_MAX_AUTH_SIZE. + */ + +int +xdr_extend_head(struct xdr_buf *buf, unsigned int base, unsigned int shiftlen) +{ + u8 *p; + + if (shiftlen == 0) + return 0; + + BUILD_BUG_ON(GSS_KRB5_MAX_SLACK_NEEDED > RPC_MAX_AUTH_SIZE); + BUG_ON(shiftlen > RPC_MAX_AUTH_SIZE); + + p = buf->head[0].iov_base + base; + + memmove(p + shiftlen, p, buf->head[0].iov_len - base); + + buf->head[0].iov_len += shiftlen; + buf->len += shiftlen; + + return 0; +} + +static u32 +gss_krb5_cts_crypt(struct crypto_blkcipher *cipher, struct xdr_buf *buf, + u32 offset, u8 *iv, struct page **pages, int encrypt) +{ + u32 ret; + struct scatterlist sg[1]; + struct blkcipher_desc desc = { .tfm = cipher, .info = iv }; + u8 data[GSS_KRB5_MAX_BLOCKSIZE * 2]; + struct page **save_pages; + u32 len = buf->len - offset; + + if (len > ARRAY_SIZE(data)) { + WARN_ON(0); + return -ENOMEM; + } + + /* + * For encryption, we want to read from the cleartext + * page cache pages, and write the encrypted data to + * the supplied xdr_buf pages. + */ + save_pages = buf->pages; + if (encrypt) + buf->pages = pages; + + ret = read_bytes_from_xdr_buf(buf, offset, data, len); + buf->pages = save_pages; + if (ret) + goto out; + + sg_init_one(sg, data, len); + + if (encrypt) + ret = crypto_blkcipher_encrypt_iv(&desc, sg, sg, len); + else + ret = crypto_blkcipher_decrypt_iv(&desc, sg, sg, len); + + if (ret) + goto out; + + ret = write_bytes_to_xdr_buf(buf, offset, data, len); + +out: + return ret; +} + +u32 +gss_krb5_aes_encrypt(struct krb5_ctx *kctx, u32 offset, + struct xdr_buf *buf, int ec, struct page **pages) +{ + u32 err; + struct xdr_netobj hmac; + u8 *cksumkey; + u8 *ecptr; + struct crypto_blkcipher *cipher, *aux_cipher; + int blocksize; + struct page **save_pages; + int nblocks, nbytes; + struct encryptor_desc desc; + u32 cbcbytes; + unsigned int usage; + + if (kctx->initiate) { + cipher = kctx->initiator_enc; + aux_cipher = kctx->initiator_enc_aux; + cksumkey = kctx->initiator_integ; + usage = KG_USAGE_INITIATOR_SEAL; + } else { + cipher = kctx->acceptor_enc; + aux_cipher = kctx->acceptor_enc_aux; + cksumkey = kctx->acceptor_integ; + usage = KG_USAGE_ACCEPTOR_SEAL; + } + blocksize = crypto_blkcipher_blocksize(cipher); + + /* hide the gss token header and insert the confounder */ + offset += GSS_KRB5_TOK_HDR_LEN; + if (xdr_extend_head(buf, offset, kctx->gk5e->conflen)) + return GSS_S_FAILURE; + gss_krb5_make_confounder(buf->head[0].iov_base + offset, kctx->gk5e->conflen); + offset -= GSS_KRB5_TOK_HDR_LEN; + + if (buf->tail[0].iov_base != NULL) { + ecptr = buf->tail[0].iov_base + buf->tail[0].iov_len; + } else { + buf->tail[0].iov_base = buf->head[0].iov_base + + buf->head[0].iov_len; + buf->tail[0].iov_len = 0; + ecptr = buf->tail[0].iov_base; + } + + memset(ecptr, 'X', ec); + buf->tail[0].iov_len += ec; + buf->len += ec; + + /* copy plaintext gss token header after filler (if any) */ + memcpy(ecptr + ec, buf->head[0].iov_base + offset, + GSS_KRB5_TOK_HDR_LEN); + buf->tail[0].iov_len += GSS_KRB5_TOK_HDR_LEN; + buf->len += GSS_KRB5_TOK_HDR_LEN; + + /* Do the HMAC */ + hmac.len = GSS_KRB5_MAX_CKSUM_LEN; + hmac.data = buf->tail[0].iov_base + buf->tail[0].iov_len; + + /* + * When we are called, pages points to the real page cache + * data -- which we can't go and encrypt! buf->pages points + * to scratch pages which we are going to send off to the + * client/server. Swap in the plaintext pages to calculate + * the hmac. + */ + save_pages = buf->pages; + buf->pages = pages; + + err = make_checksum_v2(kctx, NULL, 0, buf, + offset + GSS_KRB5_TOK_HDR_LEN, + cksumkey, usage, &hmac); + buf->pages = save_pages; + if (err) + return GSS_S_FAILURE; + + nbytes = buf->len - offset - GSS_KRB5_TOK_HDR_LEN; + nblocks = (nbytes + blocksize - 1) / blocksize; + cbcbytes = 0; + if (nblocks > 2) + cbcbytes = (nblocks - 2) * blocksize; + + memset(desc.iv, 0, sizeof(desc.iv)); + + if (cbcbytes) { + desc.pos = offset + GSS_KRB5_TOK_HDR_LEN; + desc.fragno = 0; + desc.fraglen = 0; + desc.pages = pages; + desc.outbuf = buf; + desc.desc.info = desc.iv; + desc.desc.flags = 0; + desc.desc.tfm = aux_cipher; + + sg_init_table(desc.infrags, 4); + sg_init_table(desc.outfrags, 4); + + err = xdr_process_buf(buf, offset + GSS_KRB5_TOK_HDR_LEN, + cbcbytes, encryptor, &desc); + if (err) + goto out_err; + } + + /* Make sure IV carries forward from any CBC results. */ + err = gss_krb5_cts_crypt(cipher, buf, + offset + GSS_KRB5_TOK_HDR_LEN + cbcbytes, + desc.iv, pages, 1); + if (err) { + err = GSS_S_FAILURE; + goto out_err; + } + + /* Now update buf to account for HMAC */ + buf->tail[0].iov_len += kctx->gk5e->cksumlength; + buf->len += kctx->gk5e->cksumlength; + +out_err: + if (err) + err = GSS_S_FAILURE; + return err; +} + +u32 +gss_krb5_aes_decrypt(struct krb5_ctx *kctx, u32 offset, struct xdr_buf *buf, + u32 *headskip, u32 *tailskip) +{ + struct xdr_buf subbuf; + u32 ret = 0; + u8 *cksum_key; + struct crypto_blkcipher *cipher, *aux_cipher; + struct xdr_netobj our_hmac_obj; + u8 our_hmac[GSS_KRB5_MAX_CKSUM_LEN]; + u8 pkt_hmac[GSS_KRB5_MAX_CKSUM_LEN]; + int nblocks, blocksize, cbcbytes; + struct decryptor_desc desc; + unsigned int usage; + + if (kctx->initiate) { + cipher = kctx->acceptor_enc; + aux_cipher = kctx->acceptor_enc_aux; + cksum_key = kctx->acceptor_integ; + usage = KG_USAGE_ACCEPTOR_SEAL; + } else { + cipher = kctx->initiator_enc; + aux_cipher = kctx->initiator_enc_aux; + cksum_key = kctx->initiator_integ; + usage = KG_USAGE_INITIATOR_SEAL; + } + blocksize = crypto_blkcipher_blocksize(cipher); + + + /* create a segment skipping the header and leaving out the checksum */ + xdr_buf_subsegment(buf, &subbuf, offset + GSS_KRB5_TOK_HDR_LEN, + (buf->len - offset - GSS_KRB5_TOK_HDR_LEN - + kctx->gk5e->cksumlength)); + + nblocks = (subbuf.len + blocksize - 1) / blocksize; + + cbcbytes = 0; + if (nblocks > 2) + cbcbytes = (nblocks - 2) * blocksize; + + memset(desc.iv, 0, sizeof(desc.iv)); + + if (cbcbytes) { + desc.fragno = 0; + desc.fraglen = 0; + desc.desc.info = desc.iv; + desc.desc.flags = 0; + desc.desc.tfm = aux_cipher; + + sg_init_table(desc.frags, 4); + + ret = xdr_process_buf(&subbuf, 0, cbcbytes, decryptor, &desc); + if (ret) + goto out_err; + } + + /* Make sure IV carries forward from any CBC results. */ + ret = gss_krb5_cts_crypt(cipher, &subbuf, cbcbytes, desc.iv, NULL, 0); + if (ret) + goto out_err; + + + /* Calculate our hmac over the plaintext data */ + our_hmac_obj.len = sizeof(our_hmac); + our_hmac_obj.data = our_hmac; + + ret = make_checksum_v2(kctx, NULL, 0, &subbuf, 0, + cksum_key, usage, &our_hmac_obj); + if (ret) + goto out_err; + + /* Get the packet's hmac value */ + ret = read_bytes_from_xdr_buf(buf, buf->len - kctx->gk5e->cksumlength, + pkt_hmac, kctx->gk5e->cksumlength); + if (ret) + goto out_err; + + if (memcmp(pkt_hmac, our_hmac, kctx->gk5e->cksumlength) != 0) { + ret = GSS_S_BAD_SIG; + goto out_err; + } + *headskip = kctx->gk5e->conflen; + *tailskip = kctx->gk5e->cksumlength; +out_err: + if (ret && ret != GSS_S_BAD_SIG) + ret = GSS_S_FAILURE; + return ret; +} + +/* + * Compute Kseq given the initial session key and the checksum. + * Set the key of the given cipher. + */ +int +krb5_rc4_setup_seq_key(struct krb5_ctx *kctx, struct crypto_blkcipher *cipher, + unsigned char *cksum) +{ + struct crypto_hash *hmac; + struct hash_desc desc; + struct scatterlist sg[1]; + u8 Kseq[GSS_KRB5_MAX_KEYLEN]; + u32 zeroconstant = 0; + int err; + + dprintk("%s: entered\n", __func__); + + hmac = crypto_alloc_hash(kctx->gk5e->cksum_name, 0, CRYPTO_ALG_ASYNC); + if (IS_ERR(hmac)) { + dprintk("%s: error %ld, allocating hash '%s'\n", + __func__, PTR_ERR(hmac), kctx->gk5e->cksum_name); + return PTR_ERR(hmac); + } + + desc.tfm = hmac; + desc.flags = 0; + + err = crypto_hash_init(&desc); + if (err) + goto out_err; + + /* Compute intermediate Kseq from session key */ + err = crypto_hash_setkey(hmac, kctx->Ksess, kctx->gk5e->keylength); + if (err) + goto out_err; + + sg_init_table(sg, 1); + sg_set_buf(sg, &zeroconstant, 4); + + err = crypto_hash_digest(&desc, sg, 4, Kseq); + if (err) + goto out_err; + + /* Compute final Kseq from the checksum and intermediate Kseq */ + err = crypto_hash_setkey(hmac, Kseq, kctx->gk5e->keylength); + if (err) + goto out_err; + + sg_set_buf(sg, cksum, 8); + + err = crypto_hash_digest(&desc, sg, 8, Kseq); + if (err) + goto out_err; + + err = crypto_blkcipher_setkey(cipher, Kseq, kctx->gk5e->keylength); + if (err) + goto out_err; + + err = 0; + +out_err: + crypto_free_hash(hmac); + dprintk("%s: returning %d\n", __func__, err); + return err; +} + +/* + * Compute Kcrypt given the initial session key and the plaintext seqnum. + * Set the key of cipher kctx->enc. + */ +int +krb5_rc4_setup_enc_key(struct krb5_ctx *kctx, struct crypto_blkcipher *cipher, + s32 seqnum) +{ + struct crypto_hash *hmac; + struct hash_desc desc; + struct scatterlist sg[1]; + u8 Kcrypt[GSS_KRB5_MAX_KEYLEN]; + u8 zeroconstant[4] = {0}; + u8 seqnumarray[4]; + int err, i; + + dprintk("%s: entered, seqnum %u\n", __func__, seqnum); + + hmac = crypto_alloc_hash(kctx->gk5e->cksum_name, 0, CRYPTO_ALG_ASYNC); + if (IS_ERR(hmac)) { + dprintk("%s: error %ld, allocating hash '%s'\n", + __func__, PTR_ERR(hmac), kctx->gk5e->cksum_name); + return PTR_ERR(hmac); + } + + desc.tfm = hmac; + desc.flags = 0; + + err = crypto_hash_init(&desc); + if (err) + goto out_err; + + /* Compute intermediate Kcrypt from session key */ + for (i = 0; i < kctx->gk5e->keylength; i++) + Kcrypt[i] = kctx->Ksess[i] ^ 0xf0; + + err = crypto_hash_setkey(hmac, Kcrypt, kctx->gk5e->keylength); + if (err) + goto out_err; + + sg_init_table(sg, 1); + sg_set_buf(sg, zeroconstant, 4); + + err = crypto_hash_digest(&desc, sg, 4, Kcrypt); + if (err) + goto out_err; + + /* Compute final Kcrypt from the seqnum and intermediate Kcrypt */ + err = crypto_hash_setkey(hmac, Kcrypt, kctx->gk5e->keylength); + if (err) + goto out_err; + + seqnumarray[0] = (unsigned char) ((seqnum >> 24) & 0xff); + seqnumarray[1] = (unsigned char) ((seqnum >> 16) & 0xff); + seqnumarray[2] = (unsigned char) ((seqnum >> 8) & 0xff); + seqnumarray[3] = (unsigned char) ((seqnum >> 0) & 0xff); + + sg_set_buf(sg, seqnumarray, 4); + + err = crypto_hash_digest(&desc, sg, 4, Kcrypt); + if (err) + goto out_err; + + err = crypto_blkcipher_setkey(cipher, Kcrypt, kctx->gk5e->keylength); + if (err) + goto out_err; + + err = 0; + +out_err: + crypto_free_hash(hmac); + dprintk("%s: returning %d\n", __func__, err); + return err; } -EXPORT_SYMBOL(gss_decrypt_xdr_buf); diff --git a/net/sunrpc/auth_gss/gss_krb5_keys.c b/net/sunrpc/auth_gss/gss_krb5_keys.c new file mode 100644 index 00000000000..24589bd2a4b --- /dev/null +++ b/net/sunrpc/auth_gss/gss_krb5_keys.c @@ -0,0 +1,327 @@ +/* + * COPYRIGHT (c) 2008 + * The Regents of the University of Michigan + * ALL RIGHTS RESERVED + * + * Permission is granted to use, copy, create derivative works + * and redistribute this software and such derivative works + * for any purpose, so long as the name of The University of + * Michigan is not used in any advertising or publicity + * pertaining to the use of distribution of this software + * without specific, written prior authorization. If the + * above copyright notice or any other identification of the + * University of Michigan is included in any copy of any + * portion of this software, then the disclaimer below must + * also be included. + * + * THIS SOFTWARE IS PROVIDED AS IS, WITHOUT REPRESENTATION + * FROM THE UNIVERSITY OF MICHIGAN AS TO ITS FITNESS FOR ANY + * PURPOSE, AND WITHOUT WARRANTY BY THE UNIVERSITY OF + * MICHIGAN OF ANY KIND, EITHER EXPRESS OR IMPLIED, INCLUDING + * WITHOUT LIMITATION THE IMPLIED WARRANTIES OF + * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE. THE + * REGENTS OF THE UNIVERSITY OF MICHIGAN SHALL NOT BE LIABLE + * FOR ANY DAMAGES, INCLUDING SPECIAL, INDIRECT, INCIDENTAL, OR + * CONSEQUENTIAL DAMAGES, WITH RESPECT TO ANY CLAIM ARISING + * OUT OF OR IN CONNECTION WITH THE USE OF THE SOFTWARE, EVEN + * IF IT HAS BEEN OR IS HEREAFTER ADVISED OF THE POSSIBILITY OF + * SUCH DAMAGES. + */ + +/* + * Copyright (C) 1998 by the FundsXpress, INC. + * + * All rights reserved. + * + * Export of this software from the United States of America may require + * a specific license from the United States Government. It is the + * responsibility of any person or organization contemplating export to + * obtain such a license before exporting. + * + * WITHIN THAT CONSTRAINT, permission to use, copy, modify, and + * distribute this software and its documentation for any purpose and + * without fee is hereby granted, provided that the above copyright + * notice appear in all copies and that both that copyright notice and + * this permission notice appear in supporting documentation, and that + * the name of FundsXpress. not be used in advertising or publicity pertaining + * to distribution of the software without specific, written prior + * permission. FundsXpress makes no representations about the suitability of + * this software for any purpose. It is provided "as is" without express + * or implied warranty. + * + * THIS SOFTWARE IS PROVIDED ``AS IS'' AND WITHOUT ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, WITHOUT LIMITATION, THE IMPLIED + * WARRANTIES OF MERCHANTIBILITY AND FITNESS FOR A PARTICULAR PURPOSE. + */ + +#include <linux/err.h> +#include <linux/types.h> +#include <linux/crypto.h> +#include <linux/sunrpc/gss_krb5.h> +#include <linux/sunrpc/xdr.h> +#include <linux/lcm.h> + +#ifdef RPC_DEBUG +# define RPCDBG_FACILITY RPCDBG_AUTH +#endif + +/* + * This is the n-fold function as described in rfc3961, sec 5.1 + * Taken from MIT Kerberos and modified. + */ + +static void krb5_nfold(u32 inbits, const u8 *in, + u32 outbits, u8 *out) +{ + unsigned long ulcm; + int byte, i, msbit; + + /* the code below is more readable if I make these bytes + instead of bits */ + + inbits >>= 3; + outbits >>= 3; + + /* first compute lcm(n,k) */ + ulcm = lcm(inbits, outbits); + + /* now do the real work */ + + memset(out, 0, outbits); + byte = 0; + + /* this will end up cycling through k lcm(k,n)/k times, which + is correct */ + for (i = ulcm-1; i >= 0; i--) { + /* compute the msbit in k which gets added into this byte */ + msbit = ( + /* first, start with the msbit in the first, + * unrotated byte */ + ((inbits << 3) - 1) + /* then, for each byte, shift to the right + * for each repetition */ + + (((inbits << 3) + 13) * (i/inbits)) + /* last, pick out the correct byte within + * that shifted repetition */ + + ((inbits - (i % inbits)) << 3) + ) % (inbits << 3); + + /* pull out the byte value itself */ + byte += (((in[((inbits - 1) - (msbit >> 3)) % inbits] << 8)| + (in[((inbits) - (msbit >> 3)) % inbits])) + >> ((msbit & 7) + 1)) & 0xff; + + /* do the addition */ + byte += out[i % outbits]; + out[i % outbits] = byte & 0xff; + + /* keep around the carry bit, if any */ + byte >>= 8; + + } + + /* if there's a carry bit left over, add it back in */ + if (byte) { + for (i = outbits - 1; i >= 0; i--) { + /* do the addition */ + byte += out[i]; + out[i] = byte & 0xff; + + /* keep around the carry bit, if any */ + byte >>= 8; + } + } +} + +/* + * This is the DK (derive_key) function as described in rfc3961, sec 5.1 + * Taken from MIT Kerberos and modified. + */ + +u32 krb5_derive_key(const struct gss_krb5_enctype *gk5e, + const struct xdr_netobj *inkey, + struct xdr_netobj *outkey, + const struct xdr_netobj *in_constant, + gfp_t gfp_mask) +{ + size_t blocksize, keybytes, keylength, n; + unsigned char *inblockdata, *outblockdata, *rawkey; + struct xdr_netobj inblock, outblock; + struct crypto_blkcipher *cipher; + u32 ret = EINVAL; + + blocksize = gk5e->blocksize; + keybytes = gk5e->keybytes; + keylength = gk5e->keylength; + + if ((inkey->len != keylength) || (outkey->len != keylength)) + goto err_return; + + cipher = crypto_alloc_blkcipher(gk5e->encrypt_name, 0, + CRYPTO_ALG_ASYNC); + if (IS_ERR(cipher)) + goto err_return; + if (crypto_blkcipher_setkey(cipher, inkey->data, inkey->len)) + goto err_return; + + /* allocate and set up buffers */ + + ret = ENOMEM; + inblockdata = kmalloc(blocksize, gfp_mask); + if (inblockdata == NULL) + goto err_free_cipher; + + outblockdata = kmalloc(blocksize, gfp_mask); + if (outblockdata == NULL) + goto err_free_in; + + rawkey = kmalloc(keybytes, gfp_mask); + if (rawkey == NULL) + goto err_free_out; + + inblock.data = (char *) inblockdata; + inblock.len = blocksize; + + outblock.data = (char *) outblockdata; + outblock.len = blocksize; + + /* initialize the input block */ + + if (in_constant->len == inblock.len) { + memcpy(inblock.data, in_constant->data, inblock.len); + } else { + krb5_nfold(in_constant->len * 8, in_constant->data, + inblock.len * 8, inblock.data); + } + + /* loop encrypting the blocks until enough key bytes are generated */ + + n = 0; + while (n < keybytes) { + (*(gk5e->encrypt))(cipher, NULL, inblock.data, + outblock.data, inblock.len); + + if ((keybytes - n) <= outblock.len) { + memcpy(rawkey + n, outblock.data, (keybytes - n)); + break; + } + + memcpy(rawkey + n, outblock.data, outblock.len); + memcpy(inblock.data, outblock.data, outblock.len); + n += outblock.len; + } + + /* postprocess the key */ + + inblock.data = (char *) rawkey; + inblock.len = keybytes; + + BUG_ON(gk5e->mk_key == NULL); + ret = (*(gk5e->mk_key))(gk5e, &inblock, outkey); + if (ret) { + dprintk("%s: got %d from mk_key function for '%s'\n", + __func__, ret, gk5e->encrypt_name); + goto err_free_raw; + } + + /* clean memory, free resources and exit */ + + ret = 0; + +err_free_raw: + memset(rawkey, 0, keybytes); + kfree(rawkey); +err_free_out: + memset(outblockdata, 0, blocksize); + kfree(outblockdata); +err_free_in: + memset(inblockdata, 0, blocksize); + kfree(inblockdata); +err_free_cipher: + crypto_free_blkcipher(cipher); +err_return: + return ret; +} + +#define smask(step) ((1<<step)-1) +#define pstep(x, step) (((x)&smask(step))^(((x)>>step)&smask(step))) +#define parity_char(x) pstep(pstep(pstep((x), 4), 2), 1) + +static void mit_des_fixup_key_parity(u8 key[8]) +{ + int i; + for (i = 0; i < 8; i++) { + key[i] &= 0xfe; + key[i] |= 1^parity_char(key[i]); + } +} + +/* + * This is the des3 key derivation postprocess function + */ +u32 gss_krb5_des3_make_key(const struct gss_krb5_enctype *gk5e, + struct xdr_netobj *randombits, + struct xdr_netobj *key) +{ + int i; + u32 ret = EINVAL; + + if (key->len != 24) { + dprintk("%s: key->len is %d\n", __func__, key->len); + goto err_out; + } + if (randombits->len != 21) { + dprintk("%s: randombits->len is %d\n", + __func__, randombits->len); + goto err_out; + } + + /* take the seven bytes, move them around into the top 7 bits of the + 8 key bytes, then compute the parity bits. Do this three times. */ + + for (i = 0; i < 3; i++) { + memcpy(key->data + i*8, randombits->data + i*7, 7); + key->data[i*8+7] = (((key->data[i*8]&1)<<1) | + ((key->data[i*8+1]&1)<<2) | + ((key->data[i*8+2]&1)<<3) | + ((key->data[i*8+3]&1)<<4) | + ((key->data[i*8+4]&1)<<5) | + ((key->data[i*8+5]&1)<<6) | + ((key->data[i*8+6]&1)<<7)); + + mit_des_fixup_key_parity(key->data + i*8); + } + ret = 0; +err_out: + return ret; +} + +/* + * This is the aes key derivation postprocess function + */ +u32 gss_krb5_aes_make_key(const struct gss_krb5_enctype *gk5e, + struct xdr_netobj *randombits, + struct xdr_netobj *key) +{ + u32 ret = EINVAL; + + if (key->len != 16 && key->len != 32) { + dprintk("%s: key->len is %d\n", __func__, key->len); + goto err_out; + } + if (randombits->len != 16 && randombits->len != 32) { + dprintk("%s: randombits->len is %d\n", + __func__, randombits->len); + goto err_out; + } + if (randombits->len != key->len) { + dprintk("%s: randombits->len is %d, key->len is %d\n", + __func__, randombits->len, key->len); + goto err_out; + } + memcpy(key->data, randombits->data, key->len); + ret = 0; +err_out: + return ret; +} + diff --git a/net/sunrpc/auth_gss/gss_krb5_mech.c b/net/sunrpc/auth_gss/gss_krb5_mech.c index 5f1f806a0b1..0d3c158ef8f 100644 --- a/net/sunrpc/auth_gss/gss_krb5_mech.c +++ b/net/sunrpc/auth_gss/gss_krb5_mech.c @@ -1,7 +1,7 @@ /* * linux/net/sunrpc/gss_krb5_mech.c * - * Copyright (c) 2001 The Regents of the University of Michigan. + * Copyright (c) 2001-2008 The Regents of the University of Michigan. * All rights reserved. * * Andy Adamson <andros@umich.edu> @@ -34,6 +34,7 @@ * */ +#include <linux/err.h> #include <linux/module.h> #include <linux/init.h> #include <linux/types.h> @@ -42,11 +43,149 @@ #include <linux/sunrpc/gss_krb5.h> #include <linux/sunrpc/xdr.h> #include <linux/crypto.h> +#include <linux/sunrpc/gss_krb5_enctypes.h> #ifdef RPC_DEBUG # define RPCDBG_FACILITY RPCDBG_AUTH #endif +static struct gss_api_mech gss_kerberos_mech; /* forward declaration */ + +static const struct gss_krb5_enctype supported_gss_krb5_enctypes[] = { + /* + * DES (All DES enctypes are mapped to the same gss functionality) + */ + { + .etype = ENCTYPE_DES_CBC_RAW, + .ctype = CKSUMTYPE_RSA_MD5, + .name = "des-cbc-crc", + .encrypt_name = "cbc(des)", + .cksum_name = "md5", + .encrypt = krb5_encrypt, + .decrypt = krb5_decrypt, + .mk_key = NULL, + .signalg = SGN_ALG_DES_MAC_MD5, + .sealalg = SEAL_ALG_DES, + .keybytes = 7, + .keylength = 8, + .blocksize = 8, + .conflen = 8, + .cksumlength = 8, + .keyed_cksum = 0, + }, + /* + * RC4-HMAC + */ + { + .etype = ENCTYPE_ARCFOUR_HMAC, + .ctype = CKSUMTYPE_HMAC_MD5_ARCFOUR, + .name = "rc4-hmac", + .encrypt_name = "ecb(arc4)", + .cksum_name = "hmac(md5)", + .encrypt = krb5_encrypt, + .decrypt = krb5_decrypt, + .mk_key = NULL, + .signalg = SGN_ALG_HMAC_MD5, + .sealalg = SEAL_ALG_MICROSOFT_RC4, + .keybytes = 16, + .keylength = 16, + .blocksize = 1, + .conflen = 8, + .cksumlength = 8, + .keyed_cksum = 1, + }, + /* + * 3DES + */ + { + .etype = ENCTYPE_DES3_CBC_RAW, + .ctype = CKSUMTYPE_HMAC_SHA1_DES3, + .name = "des3-hmac-sha1", + .encrypt_name = "cbc(des3_ede)", + .cksum_name = "hmac(sha1)", + .encrypt = krb5_encrypt, + .decrypt = krb5_decrypt, + .mk_key = gss_krb5_des3_make_key, + .signalg = SGN_ALG_HMAC_SHA1_DES3_KD, + .sealalg = SEAL_ALG_DES3KD, + .keybytes = 21, + .keylength = 24, + .blocksize = 8, + .conflen = 8, + .cksumlength = 20, + .keyed_cksum = 1, + }, + /* + * AES128 + */ + { + .etype = ENCTYPE_AES128_CTS_HMAC_SHA1_96, + .ctype = CKSUMTYPE_HMAC_SHA1_96_AES128, + .name = "aes128-cts", + .encrypt_name = "cts(cbc(aes))", + .cksum_name = "hmac(sha1)", + .encrypt = krb5_encrypt, + .decrypt = krb5_decrypt, + .mk_key = gss_krb5_aes_make_key, + .encrypt_v2 = gss_krb5_aes_encrypt, + .decrypt_v2 = gss_krb5_aes_decrypt, + .signalg = -1, + .sealalg = -1, + .keybytes = 16, + .keylength = 16, + .blocksize = 16, + .conflen = 16, + .cksumlength = 12, + .keyed_cksum = 1, + }, + /* + * AES256 + */ + { + .etype = ENCTYPE_AES256_CTS_HMAC_SHA1_96, + .ctype = CKSUMTYPE_HMAC_SHA1_96_AES256, + .name = "aes256-cts", + .encrypt_name = "cts(cbc(aes))", + .cksum_name = "hmac(sha1)", + .encrypt = krb5_encrypt, + .decrypt = krb5_decrypt, + .mk_key = gss_krb5_aes_make_key, + .encrypt_v2 = gss_krb5_aes_encrypt, + .decrypt_v2 = gss_krb5_aes_decrypt, + .signalg = -1, + .sealalg = -1, + .keybytes = 32, + .keylength = 32, + .blocksize = 16, + .conflen = 16, + .cksumlength = 12, + .keyed_cksum = 1, + }, +}; + +static const int num_supported_enctypes = + ARRAY_SIZE(supported_gss_krb5_enctypes); + +static int +supported_gss_krb5_enctype(int etype) +{ + int i; + for (i = 0; i < num_supported_enctypes; i++) + if (supported_gss_krb5_enctypes[i].etype == etype) + return 1; + return 0; +} + +static const struct gss_krb5_enctype * +get_gss_krb5_enctype(int etype) +{ + int i; + for (i = 0; i < num_supported_enctypes; i++) + if (supported_gss_krb5_enctypes[i].etype == etype) + return &supported_gss_krb5_enctypes[i]; + return NULL; +} + static const void * simple_get_bytes(const void *p, const void *end, void *res, int len) { @@ -69,47 +208,62 @@ simple_get_netobj(const void *p, const void *end, struct xdr_netobj *res) q = (const void *)((const char *)p + len); if (unlikely(q > end || q < p)) return ERR_PTR(-EFAULT); - res->data = kmalloc(len, GFP_KERNEL); + res->data = kmemdup(p, len, GFP_NOFS); if (unlikely(res->data == NULL)) return ERR_PTR(-ENOMEM); - memcpy(res->data, p, len); res->len = len; return q; } static inline const void * -get_key(const void *p, const void *end, struct crypto_tfm **res) +get_key(const void *p, const void *end, + struct krb5_ctx *ctx, struct crypto_blkcipher **res) { struct xdr_netobj key; - int alg, alg_mode; - char *alg_name; + int alg; p = simple_get_bytes(p, end, &alg, sizeof(alg)); if (IS_ERR(p)) goto out_err; + + switch (alg) { + case ENCTYPE_DES_CBC_CRC: + case ENCTYPE_DES_CBC_MD4: + case ENCTYPE_DES_CBC_MD5: + /* Map all these key types to ENCTYPE_DES_CBC_RAW */ + alg = ENCTYPE_DES_CBC_RAW; + break; + } + + if (!supported_gss_krb5_enctype(alg)) { + printk(KERN_WARNING "gss_kerberos_mech: unsupported " + "encryption key algorithm %d\n", alg); + p = ERR_PTR(-EINVAL); + goto out_err; + } p = simple_get_netobj(p, end, &key); if (IS_ERR(p)) goto out_err; - switch (alg) { - case ENCTYPE_DES_CBC_RAW: - alg_name = "des"; - alg_mode = CRYPTO_TFM_MODE_CBC; - break; - default: - dprintk("RPC: get_key: unsupported algorithm %d\n", alg); - goto out_err_free_key; - } - if (!(*res = crypto_alloc_tfm(alg_name, alg_mode))) + *res = crypto_alloc_blkcipher(ctx->gk5e->encrypt_name, 0, + CRYPTO_ALG_ASYNC); + if (IS_ERR(*res)) { + printk(KERN_WARNING "gss_kerberos_mech: unable to initialize " + "crypto algorithm %s\n", ctx->gk5e->encrypt_name); + *res = NULL; goto out_err_free_key; - if (crypto_cipher_setkey(*res, key.data, key.len)) + } + if (crypto_blkcipher_setkey(*res, key.data, key.len)) { + printk(KERN_WARNING "gss_kerberos_mech: error setting key for " + "crypto algorithm %s\n", ctx->gk5e->encrypt_name); goto out_err_free_tfm; + } kfree(key.data); return p; out_err_free_tfm: - crypto_free_tfm(*res); + crypto_free_blkcipher(*res); out_err_free_key: kfree(key.data); p = ERR_PTR(-EINVAL); @@ -118,45 +272,59 @@ out_err: } static int -gss_import_sec_context_kerberos(const void *p, - size_t len, - struct gss_ctx *ctx_id) +gss_import_v1_context(const void *p, const void *end, struct krb5_ctx *ctx) { - const void *end = (const void *)((const char *)p + len); - struct krb5_ctx *ctx; - - if (!(ctx = kmalloc(sizeof(*ctx), GFP_KERNEL))) - goto out_err; - memset(ctx, 0, sizeof(*ctx)); + int tmp; p = simple_get_bytes(p, end, &ctx->initiate, sizeof(ctx->initiate)); if (IS_ERR(p)) - goto out_err_free_ctx; - p = simple_get_bytes(p, end, &ctx->seed_init, sizeof(ctx->seed_init)); - if (IS_ERR(p)) - goto out_err_free_ctx; - p = simple_get_bytes(p, end, ctx->seed, sizeof(ctx->seed)); - if (IS_ERR(p)) - goto out_err_free_ctx; - p = simple_get_bytes(p, end, &ctx->signalg, sizeof(ctx->signalg)); + goto out_err; + + /* Old format supports only DES! Any other enctype uses new format */ + ctx->enctype = ENCTYPE_DES_CBC_RAW; + + ctx->gk5e = get_gss_krb5_enctype(ctx->enctype); + if (ctx->gk5e == NULL) { + p = ERR_PTR(-EINVAL); + goto out_err; + } + + /* The downcall format was designed before we completely understood + * the uses of the context fields; so it includes some stuff we + * just give some minimal sanity-checking, and some we ignore + * completely (like the next twenty bytes): */ + if (unlikely(p + 20 > end || p + 20 < p)) { + p = ERR_PTR(-EFAULT); + goto out_err; + } + p += 20; + p = simple_get_bytes(p, end, &tmp, sizeof(tmp)); if (IS_ERR(p)) - goto out_err_free_ctx; - p = simple_get_bytes(p, end, &ctx->sealalg, sizeof(ctx->sealalg)); + goto out_err; + if (tmp != SGN_ALG_DES_MAC_MD5) { + p = ERR_PTR(-ENOSYS); + goto out_err; + } + p = simple_get_bytes(p, end, &tmp, sizeof(tmp)); if (IS_ERR(p)) - goto out_err_free_ctx; + goto out_err; + if (tmp != SEAL_ALG_DES) { + p = ERR_PTR(-ENOSYS); + goto out_err; + } p = simple_get_bytes(p, end, &ctx->endtime, sizeof(ctx->endtime)); if (IS_ERR(p)) - goto out_err_free_ctx; + goto out_err; p = simple_get_bytes(p, end, &ctx->seq_send, sizeof(ctx->seq_send)); if (IS_ERR(p)) - goto out_err_free_ctx; + goto out_err; p = simple_get_netobj(p, end, &ctx->mech_used); if (IS_ERR(p)) - goto out_err_free_ctx; - p = get_key(p, end, &ctx->enc); + goto out_err; + p = get_key(p, end, ctx, &ctx->enc); if (IS_ERR(p)) goto out_err_free_mech; - p = get_key(p, end, &ctx->seq); + p = get_key(p, end, ctx, &ctx->seq); if (IS_ERR(p)) goto out_err_free_key1; if (p != end) { @@ -164,33 +332,395 @@ gss_import_sec_context_kerberos(const void *p, goto out_err_free_key2; } - ctx_id->internal_ctx_id = ctx; - dprintk("RPC: Succesfully imported new context.\n"); return 0; out_err_free_key2: - crypto_free_tfm(ctx->seq); + crypto_free_blkcipher(ctx->seq); out_err_free_key1: - crypto_free_tfm(ctx->enc); + crypto_free_blkcipher(ctx->enc); out_err_free_mech: kfree(ctx->mech_used.data); -out_err_free_ctx: - kfree(ctx); out_err: return PTR_ERR(p); } +static struct crypto_blkcipher * +context_v2_alloc_cipher(struct krb5_ctx *ctx, const char *cname, u8 *key) +{ + struct crypto_blkcipher *cp; + + cp = crypto_alloc_blkcipher(cname, 0, CRYPTO_ALG_ASYNC); + if (IS_ERR(cp)) { + dprintk("gss_kerberos_mech: unable to initialize " + "crypto algorithm %s\n", cname); + return NULL; + } + if (crypto_blkcipher_setkey(cp, key, ctx->gk5e->keylength)) { + dprintk("gss_kerberos_mech: error setting key for " + "crypto algorithm %s\n", cname); + crypto_free_blkcipher(cp); + return NULL; + } + return cp; +} + +static inline void +set_cdata(u8 cdata[GSS_KRB5_K5CLENGTH], u32 usage, u8 seed) +{ + cdata[0] = (usage>>24)&0xff; + cdata[1] = (usage>>16)&0xff; + cdata[2] = (usage>>8)&0xff; + cdata[3] = usage&0xff; + cdata[4] = seed; +} + +static int +context_derive_keys_des3(struct krb5_ctx *ctx, gfp_t gfp_mask) +{ + struct xdr_netobj c, keyin, keyout; + u8 cdata[GSS_KRB5_K5CLENGTH]; + u32 err; + + c.len = GSS_KRB5_K5CLENGTH; + c.data = cdata; + + keyin.data = ctx->Ksess; + keyin.len = ctx->gk5e->keylength; + keyout.len = ctx->gk5e->keylength; + + /* seq uses the raw key */ + ctx->seq = context_v2_alloc_cipher(ctx, ctx->gk5e->encrypt_name, + ctx->Ksess); + if (ctx->seq == NULL) + goto out_err; + + ctx->enc = context_v2_alloc_cipher(ctx, ctx->gk5e->encrypt_name, + ctx->Ksess); + if (ctx->enc == NULL) + goto out_free_seq; + + /* derive cksum */ + set_cdata(cdata, KG_USAGE_SIGN, KEY_USAGE_SEED_CHECKSUM); + keyout.data = ctx->cksum; + err = krb5_derive_key(ctx->gk5e, &keyin, &keyout, &c, gfp_mask); + if (err) { + dprintk("%s: Error %d deriving cksum key\n", + __func__, err); + goto out_free_enc; + } + + return 0; + +out_free_enc: + crypto_free_blkcipher(ctx->enc); +out_free_seq: + crypto_free_blkcipher(ctx->seq); +out_err: + return -EINVAL; +} + +/* + * Note that RC4 depends on deriving keys using the sequence + * number or the checksum of a token. Therefore, the final keys + * cannot be calculated until the token is being constructed! + */ +static int +context_derive_keys_rc4(struct krb5_ctx *ctx) +{ + struct crypto_hash *hmac; + char sigkeyconstant[] = "signaturekey"; + int slen = strlen(sigkeyconstant) + 1; /* include null terminator */ + struct hash_desc desc; + struct scatterlist sg[1]; + int err; + + dprintk("RPC: %s: entered\n", __func__); + /* + * derive cksum (aka Ksign) key + */ + hmac = crypto_alloc_hash(ctx->gk5e->cksum_name, 0, CRYPTO_ALG_ASYNC); + if (IS_ERR(hmac)) { + dprintk("%s: error %ld allocating hash '%s'\n", + __func__, PTR_ERR(hmac), ctx->gk5e->cksum_name); + err = PTR_ERR(hmac); + goto out_err; + } + + err = crypto_hash_setkey(hmac, ctx->Ksess, ctx->gk5e->keylength); + if (err) + goto out_err_free_hmac; + + sg_init_table(sg, 1); + sg_set_buf(sg, sigkeyconstant, slen); + + desc.tfm = hmac; + desc.flags = 0; + + err = crypto_hash_init(&desc); + if (err) + goto out_err_free_hmac; + + err = crypto_hash_digest(&desc, sg, slen, ctx->cksum); + if (err) + goto out_err_free_hmac; + /* + * allocate hash, and blkciphers for data and seqnum encryption + */ + ctx->enc = crypto_alloc_blkcipher(ctx->gk5e->encrypt_name, 0, + CRYPTO_ALG_ASYNC); + if (IS_ERR(ctx->enc)) { + err = PTR_ERR(ctx->enc); + goto out_err_free_hmac; + } + + ctx->seq = crypto_alloc_blkcipher(ctx->gk5e->encrypt_name, 0, + CRYPTO_ALG_ASYNC); + if (IS_ERR(ctx->seq)) { + crypto_free_blkcipher(ctx->enc); + err = PTR_ERR(ctx->seq); + goto out_err_free_hmac; + } + + dprintk("RPC: %s: returning success\n", __func__); + + err = 0; + +out_err_free_hmac: + crypto_free_hash(hmac); +out_err: + dprintk("RPC: %s: returning %d\n", __func__, err); + return err; +} + +static int +context_derive_keys_new(struct krb5_ctx *ctx, gfp_t gfp_mask) +{ + struct xdr_netobj c, keyin, keyout; + u8 cdata[GSS_KRB5_K5CLENGTH]; + u32 err; + + c.len = GSS_KRB5_K5CLENGTH; + c.data = cdata; + + keyin.data = ctx->Ksess; + keyin.len = ctx->gk5e->keylength; + keyout.len = ctx->gk5e->keylength; + + /* initiator seal encryption */ + set_cdata(cdata, KG_USAGE_INITIATOR_SEAL, KEY_USAGE_SEED_ENCRYPTION); + keyout.data = ctx->initiator_seal; + err = krb5_derive_key(ctx->gk5e, &keyin, &keyout, &c, gfp_mask); + if (err) { + dprintk("%s: Error %d deriving initiator_seal key\n", + __func__, err); + goto out_err; + } + ctx->initiator_enc = context_v2_alloc_cipher(ctx, + ctx->gk5e->encrypt_name, + ctx->initiator_seal); + if (ctx->initiator_enc == NULL) + goto out_err; + + /* acceptor seal encryption */ + set_cdata(cdata, KG_USAGE_ACCEPTOR_SEAL, KEY_USAGE_SEED_ENCRYPTION); + keyout.data = ctx->acceptor_seal; + err = krb5_derive_key(ctx->gk5e, &keyin, &keyout, &c, gfp_mask); + if (err) { + dprintk("%s: Error %d deriving acceptor_seal key\n", + __func__, err); + goto out_free_initiator_enc; + } + ctx->acceptor_enc = context_v2_alloc_cipher(ctx, + ctx->gk5e->encrypt_name, + ctx->acceptor_seal); + if (ctx->acceptor_enc == NULL) + goto out_free_initiator_enc; + + /* initiator sign checksum */ + set_cdata(cdata, KG_USAGE_INITIATOR_SIGN, KEY_USAGE_SEED_CHECKSUM); + keyout.data = ctx->initiator_sign; + err = krb5_derive_key(ctx->gk5e, &keyin, &keyout, &c, gfp_mask); + if (err) { + dprintk("%s: Error %d deriving initiator_sign key\n", + __func__, err); + goto out_free_acceptor_enc; + } + + /* acceptor sign checksum */ + set_cdata(cdata, KG_USAGE_ACCEPTOR_SIGN, KEY_USAGE_SEED_CHECKSUM); + keyout.data = ctx->acceptor_sign; + err = krb5_derive_key(ctx->gk5e, &keyin, &keyout, &c, gfp_mask); + if (err) { + dprintk("%s: Error %d deriving acceptor_sign key\n", + __func__, err); + goto out_free_acceptor_enc; + } + + /* initiator seal integrity */ + set_cdata(cdata, KG_USAGE_INITIATOR_SEAL, KEY_USAGE_SEED_INTEGRITY); + keyout.data = ctx->initiator_integ; + err = krb5_derive_key(ctx->gk5e, &keyin, &keyout, &c, gfp_mask); + if (err) { + dprintk("%s: Error %d deriving initiator_integ key\n", + __func__, err); + goto out_free_acceptor_enc; + } + + /* acceptor seal integrity */ + set_cdata(cdata, KG_USAGE_ACCEPTOR_SEAL, KEY_USAGE_SEED_INTEGRITY); + keyout.data = ctx->acceptor_integ; + err = krb5_derive_key(ctx->gk5e, &keyin, &keyout, &c, gfp_mask); + if (err) { + dprintk("%s: Error %d deriving acceptor_integ key\n", + __func__, err); + goto out_free_acceptor_enc; + } + + switch (ctx->enctype) { + case ENCTYPE_AES128_CTS_HMAC_SHA1_96: + case ENCTYPE_AES256_CTS_HMAC_SHA1_96: + ctx->initiator_enc_aux = + context_v2_alloc_cipher(ctx, "cbc(aes)", + ctx->initiator_seal); + if (ctx->initiator_enc_aux == NULL) + goto out_free_acceptor_enc; + ctx->acceptor_enc_aux = + context_v2_alloc_cipher(ctx, "cbc(aes)", + ctx->acceptor_seal); + if (ctx->acceptor_enc_aux == NULL) { + crypto_free_blkcipher(ctx->initiator_enc_aux); + goto out_free_acceptor_enc; + } + } + + return 0; + +out_free_acceptor_enc: + crypto_free_blkcipher(ctx->acceptor_enc); +out_free_initiator_enc: + crypto_free_blkcipher(ctx->initiator_enc); +out_err: + return -EINVAL; +} + +static int +gss_import_v2_context(const void *p, const void *end, struct krb5_ctx *ctx, + gfp_t gfp_mask) +{ + int keylen; + + p = simple_get_bytes(p, end, &ctx->flags, sizeof(ctx->flags)); + if (IS_ERR(p)) + goto out_err; + ctx->initiate = ctx->flags & KRB5_CTX_FLAG_INITIATOR; + + p = simple_get_bytes(p, end, &ctx->endtime, sizeof(ctx->endtime)); + if (IS_ERR(p)) + goto out_err; + p = simple_get_bytes(p, end, &ctx->seq_send64, sizeof(ctx->seq_send64)); + if (IS_ERR(p)) + goto out_err; + /* set seq_send for use by "older" enctypes */ + ctx->seq_send = ctx->seq_send64; + if (ctx->seq_send64 != ctx->seq_send) { + dprintk("%s: seq_send64 %lx, seq_send %x overflow?\n", __func__, + (unsigned long)ctx->seq_send64, ctx->seq_send); + p = ERR_PTR(-EINVAL); + goto out_err; + } + p = simple_get_bytes(p, end, &ctx->enctype, sizeof(ctx->enctype)); + if (IS_ERR(p)) + goto out_err; + /* Map ENCTYPE_DES3_CBC_SHA1 to ENCTYPE_DES3_CBC_RAW */ + if (ctx->enctype == ENCTYPE_DES3_CBC_SHA1) + ctx->enctype = ENCTYPE_DES3_CBC_RAW; + ctx->gk5e = get_gss_krb5_enctype(ctx->enctype); + if (ctx->gk5e == NULL) { + dprintk("gss_kerberos_mech: unsupported krb5 enctype %u\n", + ctx->enctype); + p = ERR_PTR(-EINVAL); + goto out_err; + } + keylen = ctx->gk5e->keylength; + + p = simple_get_bytes(p, end, ctx->Ksess, keylen); + if (IS_ERR(p)) + goto out_err; + + if (p != end) { + p = ERR_PTR(-EINVAL); + goto out_err; + } + + ctx->mech_used.data = kmemdup(gss_kerberos_mech.gm_oid.data, + gss_kerberos_mech.gm_oid.len, gfp_mask); + if (unlikely(ctx->mech_used.data == NULL)) { + p = ERR_PTR(-ENOMEM); + goto out_err; + } + ctx->mech_used.len = gss_kerberos_mech.gm_oid.len; + + switch (ctx->enctype) { + case ENCTYPE_DES3_CBC_RAW: + return context_derive_keys_des3(ctx, gfp_mask); + case ENCTYPE_ARCFOUR_HMAC: + return context_derive_keys_rc4(ctx); + case ENCTYPE_AES128_CTS_HMAC_SHA1_96: + case ENCTYPE_AES256_CTS_HMAC_SHA1_96: + return context_derive_keys_new(ctx, gfp_mask); + default: + return -EINVAL; + } + +out_err: + return PTR_ERR(p); +} + +static int +gss_import_sec_context_kerberos(const void *p, size_t len, + struct gss_ctx *ctx_id, + time_t *endtime, + gfp_t gfp_mask) +{ + const void *end = (const void *)((const char *)p + len); + struct krb5_ctx *ctx; + int ret; + + ctx = kzalloc(sizeof(*ctx), gfp_mask); + if (ctx == NULL) + return -ENOMEM; + + if (len == 85) + ret = gss_import_v1_context(p, end, ctx); + else + ret = gss_import_v2_context(p, end, ctx, gfp_mask); + + if (ret == 0) { + ctx_id->internal_ctx_id = ctx; + if (endtime) + *endtime = ctx->endtime; + } else + kfree(ctx); + + dprintk("RPC: %s: returning %d\n", __func__, ret); + return ret; +} + static void gss_delete_sec_context_kerberos(void *internal_ctx) { struct krb5_ctx *kctx = internal_ctx; - crypto_free_tfm(kctx->seq); - crypto_free_tfm(kctx->enc); + crypto_free_blkcipher(kctx->seq); + crypto_free_blkcipher(kctx->enc); + crypto_free_blkcipher(kctx->acceptor_enc); + crypto_free_blkcipher(kctx->initiator_enc); + crypto_free_blkcipher(kctx->acceptor_enc_aux); + crypto_free_blkcipher(kctx->initiator_enc_aux); kfree(kctx->mech_used.data); kfree(kctx); } -static struct gss_api_ops gss_kerberos_ops = { +static const struct gss_api_ops gss_kerberos_ops = { .gss_import_sec_context = gss_import_sec_context_kerberos, .gss_get_mic = gss_get_mic_kerberos, .gss_verify_mic = gss_verify_mic_kerberos, @@ -202,27 +732,40 @@ static struct gss_api_ops gss_kerberos_ops = { static struct pf_desc gss_kerberos_pfs[] = { [0] = { .pseudoflavor = RPC_AUTH_GSS_KRB5, + .qop = GSS_C_QOP_DEFAULT, .service = RPC_GSS_SVC_NONE, .name = "krb5", }, [1] = { .pseudoflavor = RPC_AUTH_GSS_KRB5I, + .qop = GSS_C_QOP_DEFAULT, .service = RPC_GSS_SVC_INTEGRITY, .name = "krb5i", }, [2] = { .pseudoflavor = RPC_AUTH_GSS_KRB5P, + .qop = GSS_C_QOP_DEFAULT, .service = RPC_GSS_SVC_PRIVACY, .name = "krb5p", }, }; +MODULE_ALIAS("rpc-auth-gss-krb5"); +MODULE_ALIAS("rpc-auth-gss-krb5i"); +MODULE_ALIAS("rpc-auth-gss-krb5p"); +MODULE_ALIAS("rpc-auth-gss-390003"); +MODULE_ALIAS("rpc-auth-gss-390004"); +MODULE_ALIAS("rpc-auth-gss-390005"); +MODULE_ALIAS("rpc-auth-gss-1.2.840.113554.1.2.2"); + static struct gss_api_mech gss_kerberos_mech = { .gm_name = "krb5", .gm_owner = THIS_MODULE, + .gm_oid = { 9, "\x2a\x86\x48\x86\xf7\x12\x01\x02\x02" }, .gm_ops = &gss_kerberos_ops, .gm_pf_num = ARRAY_SIZE(gss_kerberos_pfs), .gm_pfs = gss_kerberos_pfs, + .gm_upcall_enctypes = KRB5_SUPPORTED_ENCTYPES, }; static int __init init_kerberos_module(void) diff --git a/net/sunrpc/auth_gss/gss_krb5_seal.c b/net/sunrpc/auth_gss/gss_krb5_seal.c index 13f8ae97945..62ae3273186 100644 --- a/net/sunrpc/auth_gss/gss_krb5_seal.c +++ b/net/sunrpc/auth_gss/gss_krb5_seal.c @@ -3,7 +3,7 @@ * * Adapted from MIT Kerberos 5-1.2.1 lib/gssapi/krb5/k5seal.c * - * Copyright (c) 2000 The Regents of the University of Michigan. + * Copyright (c) 2000-2008 The Regents of the University of Michigan. * All rights reserved. * * Andy Adamson <andros@umich.edu> @@ -59,90 +59,165 @@ */ #include <linux/types.h> -#include <linux/slab.h> #include <linux/jiffies.h> #include <linux/sunrpc/gss_krb5.h> #include <linux/random.h> -#include <asm/scatterlist.h> #include <linux/crypto.h> #ifdef RPC_DEBUG # define RPCDBG_FACILITY RPCDBG_AUTH #endif -u32 -gss_get_mic_kerberos(struct gss_ctx *gss_ctx, struct xdr_buf *text, +DEFINE_SPINLOCK(krb5_seq_lock); + +static char * +setup_token(struct krb5_ctx *ctx, struct xdr_netobj *token) +{ + __be16 *ptr, *krb5_hdr; + int body_size = GSS_KRB5_TOK_HDR_LEN + ctx->gk5e->cksumlength; + + token->len = g_token_size(&ctx->mech_used, body_size); + + ptr = (__be16 *)token->data; + g_make_token_header(&ctx->mech_used, body_size, (unsigned char **)&ptr); + + /* ptr now at start of header described in rfc 1964, section 1.2.1: */ + krb5_hdr = ptr; + *ptr++ = KG_TOK_MIC_MSG; + *ptr++ = cpu_to_le16(ctx->gk5e->signalg); + *ptr++ = SEAL_ALG_NONE; + *ptr++ = 0xffff; + + return (char *)krb5_hdr; +} + +static void * +setup_token_v2(struct krb5_ctx *ctx, struct xdr_netobj *token) +{ + __be16 *ptr, *krb5_hdr; + u8 *p, flags = 0x00; + + if ((ctx->flags & KRB5_CTX_FLAG_INITIATOR) == 0) + flags |= 0x01; + if (ctx->flags & KRB5_CTX_FLAG_ACCEPTOR_SUBKEY) + flags |= 0x04; + + /* Per rfc 4121, sec 4.2.6.1, there is no header, + * just start the token */ + krb5_hdr = ptr = (__be16 *)token->data; + + *ptr++ = KG2_TOK_MIC; + p = (u8 *)ptr; + *p++ = flags; + *p++ = 0xff; + ptr = (__be16 *)p; + *ptr++ = 0xffff; + *ptr++ = 0xffff; + + token->len = GSS_KRB5_TOK_HDR_LEN + ctx->gk5e->cksumlength; + return krb5_hdr; +} + +static u32 +gss_get_mic_v1(struct krb5_ctx *ctx, struct xdr_buf *text, struct xdr_netobj *token) { - struct krb5_ctx *ctx = gss_ctx->internal_ctx_id; - s32 checksum_type; - struct xdr_netobj md5cksum = {.len = 0, .data = NULL}; - unsigned char *ptr, *krb5_hdr, *msg_start; + char cksumdata[GSS_KRB5_MAX_CKSUM_LEN]; + struct xdr_netobj md5cksum = {.len = sizeof(cksumdata), + .data = cksumdata}; + void *ptr; s32 now; + u32 seq_send; + u8 *cksumkey; - dprintk("RPC: gss_krb5_seal\n"); + dprintk("RPC: %s\n", __func__); + BUG_ON(ctx == NULL); now = get_seconds(); - switch (ctx->signalg) { - case SGN_ALG_DES_MAC_MD5: - checksum_type = CKSUMTYPE_RSA_MD5; - break; - default: - dprintk("RPC: gss_krb5_seal: ctx->signalg %d not" - " supported\n", ctx->signalg); - goto out_err; - } - if (ctx->sealalg != SEAL_ALG_NONE && ctx->sealalg != SEAL_ALG_DES) { - dprintk("RPC: gss_krb5_seal: ctx->sealalg %d not supported\n", - ctx->sealalg); - goto out_err; - } - - token->len = g_token_size(&ctx->mech_used, 22); + ptr = setup_token(ctx, token); - ptr = token->data; - g_make_token_header(&ctx->mech_used, 22, &ptr); + if (ctx->gk5e->keyed_cksum) + cksumkey = ctx->cksum; + else + cksumkey = NULL; - *ptr++ = (unsigned char) ((KG_TOK_MIC_MSG>>8)&0xff); - *ptr++ = (unsigned char) (KG_TOK_MIC_MSG&0xff); + if (make_checksum(ctx, ptr, 8, text, 0, cksumkey, + KG_USAGE_SIGN, &md5cksum)) + return GSS_S_FAILURE; - /* ptr now at byte 2 of header described in rfc 1964, section 1.2.1: */ - krb5_hdr = ptr - 2; - msg_start = krb5_hdr + 24; + memcpy(ptr + GSS_KRB5_TOK_HDR_LEN, md5cksum.data, md5cksum.len); - *(u16 *)(krb5_hdr + 2) = htons(ctx->signalg); - memset(krb5_hdr + 4, 0xff, 4); + spin_lock(&krb5_seq_lock); + seq_send = ctx->seq_send++; + spin_unlock(&krb5_seq_lock); - if (make_checksum(checksum_type, krb5_hdr, 8, text, 0, &md5cksum)) - goto out_err; + if (krb5_make_seq_num(ctx, ctx->seq, ctx->initiate ? 0 : 0xff, + seq_send, ptr + GSS_KRB5_TOK_HDR_LEN, ptr + 8)) + return GSS_S_FAILURE; - switch (ctx->signalg) { - case SGN_ALG_DES_MAC_MD5: - if (krb5_encrypt(ctx->seq, NULL, md5cksum.data, - md5cksum.data, md5cksum.len)) - goto out_err; - memcpy(krb5_hdr + 16, - md5cksum.data + md5cksum.len - KRB5_CKSUM_LENGTH, - KRB5_CKSUM_LENGTH); + return (ctx->endtime < now) ? GSS_S_CONTEXT_EXPIRED : GSS_S_COMPLETE; +} - dprintk("RPC: make_seal_token: cksum data: \n"); - print_hexl((u32 *) (krb5_hdr + 16), KRB5_CKSUM_LENGTH, 0); - break; - default: - BUG(); +static u32 +gss_get_mic_v2(struct krb5_ctx *ctx, struct xdr_buf *text, + struct xdr_netobj *token) +{ + char cksumdata[GSS_KRB5_MAX_CKSUM_LEN]; + struct xdr_netobj cksumobj = { .len = sizeof(cksumdata), + .data = cksumdata}; + void *krb5_hdr; + s32 now; + u64 seq_send; + u8 *cksumkey; + unsigned int cksum_usage; + + dprintk("RPC: %s\n", __func__); + + krb5_hdr = setup_token_v2(ctx, token); + + /* Set up the sequence number. Now 64-bits in clear + * text and w/o direction indicator */ + spin_lock(&krb5_seq_lock); + seq_send = ctx->seq_send64++; + spin_unlock(&krb5_seq_lock); + *((u64 *)(krb5_hdr + 8)) = cpu_to_be64(seq_send); + + if (ctx->initiate) { + cksumkey = ctx->initiator_sign; + cksum_usage = KG_USAGE_INITIATOR_SIGN; + } else { + cksumkey = ctx->acceptor_sign; + cksum_usage = KG_USAGE_ACCEPTOR_SIGN; } - kfree(md5cksum.data); + if (make_checksum_v2(ctx, krb5_hdr, GSS_KRB5_TOK_HDR_LEN, + text, 0, cksumkey, cksum_usage, &cksumobj)) + return GSS_S_FAILURE; - if ((krb5_make_seq_num(ctx->seq, ctx->initiate ? 0 : 0xff, - ctx->seq_send, krb5_hdr + 16, krb5_hdr + 8))) - goto out_err; + memcpy(krb5_hdr + GSS_KRB5_TOK_HDR_LEN, cksumobj.data, cksumobj.len); - ctx->seq_send++; + now = get_seconds(); - return ((ctx->endtime < now) ? GSS_S_CONTEXT_EXPIRED : GSS_S_COMPLETE); -out_err: - if (md5cksum.data) kfree(md5cksum.data); - return GSS_S_FAILURE; + return (ctx->endtime < now) ? GSS_S_CONTEXT_EXPIRED : GSS_S_COMPLETE; } + +u32 +gss_get_mic_kerberos(struct gss_ctx *gss_ctx, struct xdr_buf *text, + struct xdr_netobj *token) +{ + struct krb5_ctx *ctx = gss_ctx->internal_ctx_id; + + switch (ctx->enctype) { + default: + BUG(); + case ENCTYPE_DES_CBC_RAW: + case ENCTYPE_DES3_CBC_RAW: + case ENCTYPE_ARCFOUR_HMAC: + return gss_get_mic_v1(ctx, text, token); + case ENCTYPE_AES128_CTS_HMAC_SHA1_96: + case ENCTYPE_AES256_CTS_HMAC_SHA1_96: + return gss_get_mic_v2(ctx, text, token); + } +} + diff --git a/net/sunrpc/auth_gss/gss_krb5_seqnum.c b/net/sunrpc/auth_gss/gss_krb5_seqnum.c index c53ead39118..62ac90c62cb 100644 --- a/net/sunrpc/auth_gss/gss_krb5_seqnum.c +++ b/net/sunrpc/auth_gss/gss_krb5_seqnum.c @@ -11,7 +11,7 @@ /* * Copyright 1993 by OpenVision Technologies, Inc. - * + * * Permission to use, copy, modify, distribute, and sell this software * and its documentation for any purpose is hereby granted without fee, * provided that the above copyright notice appears in all copies and @@ -21,7 +21,7 @@ * without specific, written prior permission. OpenVision makes no * representations about the suitability of this software for any * purpose. It is provided "as is" without express or implied warranty. - * + * * OPENVISION DISCLAIMS ALL WARRANTIES WITH REGARD TO THIS SOFTWARE, * INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS, IN NO * EVENT SHALL OPENVISION BE LIABLE FOR ANY SPECIAL, INDIRECT OR @@ -32,7 +32,6 @@ */ #include <linux/types.h> -#include <linux/slab.h> #include <linux/sunrpc/gss_krb5.h> #include <linux/crypto.h> @@ -40,14 +39,51 @@ # define RPCDBG_FACILITY RPCDBG_AUTH #endif +static s32 +krb5_make_rc4_seq_num(struct krb5_ctx *kctx, int direction, s32 seqnum, + unsigned char *cksum, unsigned char *buf) +{ + struct crypto_blkcipher *cipher; + unsigned char plain[8]; + s32 code; + + dprintk("RPC: %s:\n", __func__); + cipher = crypto_alloc_blkcipher(kctx->gk5e->encrypt_name, 0, + CRYPTO_ALG_ASYNC); + if (IS_ERR(cipher)) + return PTR_ERR(cipher); + + plain[0] = (unsigned char) ((seqnum >> 24) & 0xff); + plain[1] = (unsigned char) ((seqnum >> 16) & 0xff); + plain[2] = (unsigned char) ((seqnum >> 8) & 0xff); + plain[3] = (unsigned char) ((seqnum >> 0) & 0xff); + plain[4] = direction; + plain[5] = direction; + plain[6] = direction; + plain[7] = direction; + + code = krb5_rc4_setup_seq_key(kctx, cipher, cksum); + if (code) + goto out; + + code = krb5_encrypt(cipher, cksum, plain, buf, 8); +out: + crypto_free_blkcipher(cipher); + return code; +} s32 -krb5_make_seq_num(struct crypto_tfm *key, +krb5_make_seq_num(struct krb5_ctx *kctx, + struct crypto_blkcipher *key, int direction, - s32 seqnum, + u32 seqnum, unsigned char *cksum, unsigned char *buf) { unsigned char plain[8]; + if (kctx->enctype == ENCTYPE_ARCFOUR_HMAC) + return krb5_make_rc4_seq_num(kctx, direction, seqnum, + cksum, buf); + plain[0] = (unsigned char) (seqnum & 0xff); plain[1] = (unsigned char) ((seqnum >> 8) & 0xff); plain[2] = (unsigned char) ((seqnum >> 16) & 0xff); @@ -61,22 +97,64 @@ krb5_make_seq_num(struct crypto_tfm *key, return krb5_encrypt(key, cksum, plain, buf, 8); } +static s32 +krb5_get_rc4_seq_num(struct krb5_ctx *kctx, unsigned char *cksum, + unsigned char *buf, int *direction, s32 *seqnum) +{ + struct crypto_blkcipher *cipher; + unsigned char plain[8]; + s32 code; + + dprintk("RPC: %s:\n", __func__); + cipher = crypto_alloc_blkcipher(kctx->gk5e->encrypt_name, 0, + CRYPTO_ALG_ASYNC); + if (IS_ERR(cipher)) + return PTR_ERR(cipher); + + code = krb5_rc4_setup_seq_key(kctx, cipher, cksum); + if (code) + goto out; + + code = krb5_decrypt(cipher, cksum, buf, plain, 8); + if (code) + goto out; + + if ((plain[4] != plain[5]) || (plain[4] != plain[6]) + || (plain[4] != plain[7])) { + code = (s32)KG_BAD_SEQ; + goto out; + } + + *direction = plain[4]; + + *seqnum = ((plain[0] << 24) | (plain[1] << 16) | + (plain[2] << 8) | (plain[3])); +out: + crypto_free_blkcipher(cipher); + return code; +} + s32 -krb5_get_seq_num(struct crypto_tfm *key, +krb5_get_seq_num(struct krb5_ctx *kctx, unsigned char *cksum, unsigned char *buf, - int *direction, s32 * seqnum) + int *direction, u32 *seqnum) { s32 code; unsigned char plain[8]; + struct crypto_blkcipher *key = kctx->seq; + + dprintk("RPC: krb5_get_seq_num:\n"); - dprintk("RPC: krb5_get_seq_num:\n"); + if (kctx->enctype == ENCTYPE_ARCFOUR_HMAC) + return krb5_get_rc4_seq_num(kctx, cksum, buf, + direction, seqnum); if ((code = krb5_decrypt(key, cksum, buf, plain, 8))) return code; - if ((plain[4] != plain[5]) || (plain[4] != plain[6]) - || (plain[4] != plain[7])) + if ((plain[4] != plain[5]) || (plain[4] != plain[6]) || + (plain[4] != plain[7])) return (s32)KG_BAD_SEQ; *direction = plain[4]; @@ -84,5 +162,5 @@ krb5_get_seq_num(struct crypto_tfm *key, *seqnum = ((plain[0]) | (plain[1] << 8) | (plain[2] << 16) | (plain[3] << 24)); - return (0); + return 0; } diff --git a/net/sunrpc/auth_gss/gss_krb5_unseal.c b/net/sunrpc/auth_gss/gss_krb5_unseal.c index 2030475d98e..6c981ddc19f 100644 --- a/net/sunrpc/auth_gss/gss_krb5_unseal.c +++ b/net/sunrpc/auth_gss/gss_krb5_unseal.c @@ -3,7 +3,7 @@ * * Adapted from MIT Kerberos 5-1.2.1 lib/gssapi/krb5/k5unseal.c * - * Copyright (c) 2000 The Regents of the University of Michigan. + * Copyright (c) 2000-2008 The Regents of the University of Michigan. * All rights reserved. * * Andy Adamson <andros@umich.edu> @@ -58,7 +58,6 @@ */ #include <linux/types.h> -#include <linux/slab.h> #include <linux/jiffies.h> #include <linux/sunrpc/gss_krb5.h> #include <linux/crypto.h> @@ -71,111 +70,157 @@ /* read_token is a mic token, and message_buffer is the data that the mic was * supposedly taken over. */ -u32 -gss_verify_mic_kerberos(struct gss_ctx *gss_ctx, +static u32 +gss_verify_mic_v1(struct krb5_ctx *ctx, struct xdr_buf *message_buffer, struct xdr_netobj *read_token) { - struct krb5_ctx *ctx = gss_ctx->internal_ctx_id; int signalg; int sealalg; - s32 checksum_type; - struct xdr_netobj md5cksum = {.len = 0, .data = NULL}; + char cksumdata[GSS_KRB5_MAX_CKSUM_LEN]; + struct xdr_netobj md5cksum = {.len = sizeof(cksumdata), + .data = cksumdata}; s32 now; int direction; - s32 seqnum; + u32 seqnum; unsigned char *ptr = (unsigned char *)read_token->data; int bodysize; - u32 ret = GSS_S_DEFECTIVE_TOKEN; + u8 *cksumkey; - dprintk("RPC: krb5_read_token\n"); + dprintk("RPC: krb5_read_token\n"); if (g_verify_token_header(&ctx->mech_used, &bodysize, &ptr, read_token->len)) - goto out; + return GSS_S_DEFECTIVE_TOKEN; - if ((*ptr++ != ((KG_TOK_MIC_MSG>>8)&0xff)) || - (*ptr++ != ( KG_TOK_MIC_MSG &0xff)) ) - goto out; + if ((ptr[0] != ((KG_TOK_MIC_MSG >> 8) & 0xff)) || + (ptr[1] != (KG_TOK_MIC_MSG & 0xff))) + return GSS_S_DEFECTIVE_TOKEN; /* XXX sanity-check bodysize?? */ - /* get the sign and seal algorithms */ + signalg = ptr[2] + (ptr[3] << 8); + if (signalg != ctx->gk5e->signalg) + return GSS_S_DEFECTIVE_TOKEN; - signalg = ptr[0] + (ptr[1] << 8); - sealalg = ptr[2] + (ptr[3] << 8); + sealalg = ptr[4] + (ptr[5] << 8); + if (sealalg != SEAL_ALG_NONE) + return GSS_S_DEFECTIVE_TOKEN; - /* Sanity checks */ + if ((ptr[6] != 0xff) || (ptr[7] != 0xff)) + return GSS_S_DEFECTIVE_TOKEN; - if ((ptr[4] != 0xff) || (ptr[5] != 0xff)) - goto out; + if (ctx->gk5e->keyed_cksum) + cksumkey = ctx->cksum; + else + cksumkey = NULL; - if (sealalg != 0xffff) - goto out; + if (make_checksum(ctx, ptr, 8, message_buffer, 0, + cksumkey, KG_USAGE_SIGN, &md5cksum)) + return GSS_S_FAILURE; - /* there are several mappings of seal algorithms to sign algorithms, - but few enough that we can try them all. */ + if (memcmp(md5cksum.data, ptr + GSS_KRB5_TOK_HDR_LEN, + ctx->gk5e->cksumlength)) + return GSS_S_BAD_SIG; - if ((ctx->sealalg == SEAL_ALG_NONE && signalg > 1) || - (ctx->sealalg == SEAL_ALG_1 && signalg != SGN_ALG_3) || - (ctx->sealalg == SEAL_ALG_DES3KD && - signalg != SGN_ALG_HMAC_SHA1_DES3_KD)) - goto out; + /* it got through unscathed. Make sure the context is unexpired */ - /* compute the checksum of the message */ + now = get_seconds(); - /* initialize the the cksum */ - switch (signalg) { - case SGN_ALG_DES_MAC_MD5: - checksum_type = CKSUMTYPE_RSA_MD5; - break; - default: - ret = GSS_S_DEFECTIVE_TOKEN; - goto out; + if (now > ctx->endtime) + return GSS_S_CONTEXT_EXPIRED; + + /* do sequencing checks */ + + if (krb5_get_seq_num(ctx, ptr + GSS_KRB5_TOK_HDR_LEN, ptr + 8, + &direction, &seqnum)) + return GSS_S_FAILURE; + + if ((ctx->initiate && direction != 0xff) || + (!ctx->initiate && direction != 0)) + return GSS_S_BAD_SIG; + + return GSS_S_COMPLETE; +} + +static u32 +gss_verify_mic_v2(struct krb5_ctx *ctx, + struct xdr_buf *message_buffer, struct xdr_netobj *read_token) +{ + char cksumdata[GSS_KRB5_MAX_CKSUM_LEN]; + struct xdr_netobj cksumobj = {.len = sizeof(cksumdata), + .data = cksumdata}; + s32 now; + u8 *ptr = read_token->data; + u8 *cksumkey; + u8 flags; + int i; + unsigned int cksum_usage; + + dprintk("RPC: %s\n", __func__); + + if (be16_to_cpu(*((__be16 *)ptr)) != KG2_TOK_MIC) + return GSS_S_DEFECTIVE_TOKEN; + + flags = ptr[2]; + if ((!ctx->initiate && (flags & KG2_TOKEN_FLAG_SENTBYACCEPTOR)) || + (ctx->initiate && !(flags & KG2_TOKEN_FLAG_SENTBYACCEPTOR))) + return GSS_S_BAD_SIG; + + if (flags & KG2_TOKEN_FLAG_SEALED) { + dprintk("%s: token has unexpected sealed flag\n", __func__); + return GSS_S_FAILURE; } - switch (signalg) { - case SGN_ALG_DES_MAC_MD5: - ret = make_checksum(checksum_type, ptr - 2, 8, - message_buffer, 0, &md5cksum); - if (ret) - goto out; - - ret = krb5_encrypt(ctx->seq, NULL, md5cksum.data, - md5cksum.data, 16); - if (ret) - goto out; - - if (memcmp(md5cksum.data + 8, ptr + 14, 8)) { - ret = GSS_S_BAD_SIG; - goto out; - } - break; - default: - ret = GSS_S_DEFECTIVE_TOKEN; - goto out; + for (i = 3; i < 8; i++) + if (ptr[i] != 0xff) + return GSS_S_DEFECTIVE_TOKEN; + + if (ctx->initiate) { + cksumkey = ctx->acceptor_sign; + cksum_usage = KG_USAGE_ACCEPTOR_SIGN; + } else { + cksumkey = ctx->initiator_sign; + cksum_usage = KG_USAGE_INITIATOR_SIGN; } - /* it got through unscathed. Make sure the context is unexpired */ + if (make_checksum_v2(ctx, ptr, GSS_KRB5_TOK_HDR_LEN, message_buffer, 0, + cksumkey, cksum_usage, &cksumobj)) + return GSS_S_FAILURE; - now = get_seconds(); + if (memcmp(cksumobj.data, ptr + GSS_KRB5_TOK_HDR_LEN, + ctx->gk5e->cksumlength)) + return GSS_S_BAD_SIG; - ret = GSS_S_CONTEXT_EXPIRED; + /* it got through unscathed. Make sure the context is unexpired */ + now = get_seconds(); if (now > ctx->endtime) - goto out; + return GSS_S_CONTEXT_EXPIRED; - /* do sequencing checks */ + /* + * NOTE: the sequence number at ptr + 8 is skipped, rpcsec_gss + * doesn't want it checked; see page 6 of rfc 2203. + */ - ret = GSS_S_BAD_SIG; - if ((ret = krb5_get_seq_num(ctx->seq, ptr + 14, ptr + 6, &direction, - &seqnum))) - goto out; + return GSS_S_COMPLETE; +} - if ((ctx->initiate && direction != 0xff) || - (!ctx->initiate && direction != 0)) - goto out; +u32 +gss_verify_mic_kerberos(struct gss_ctx *gss_ctx, + struct xdr_buf *message_buffer, + struct xdr_netobj *read_token) +{ + struct krb5_ctx *ctx = gss_ctx->internal_ctx_id; - ret = GSS_S_COMPLETE; -out: - if (md5cksum.data) kfree(md5cksum.data); - return ret; + switch (ctx->enctype) { + default: + BUG(); + case ENCTYPE_DES_CBC_RAW: + case ENCTYPE_DES3_CBC_RAW: + case ENCTYPE_ARCFOUR_HMAC: + return gss_verify_mic_v1(ctx, message_buffer, read_token); + case ENCTYPE_AES128_CTS_HMAC_SHA1_96: + case ENCTYPE_AES256_CTS_HMAC_SHA1_96: + return gss_verify_mic_v2(ctx, message_buffer, read_token); + } } + diff --git a/net/sunrpc/auth_gss/gss_krb5_wrap.c b/net/sunrpc/auth_gss/gss_krb5_wrap.c index af777cf9f25..42560e55d97 100644 --- a/net/sunrpc/auth_gss/gss_krb5_wrap.c +++ b/net/sunrpc/auth_gss/gss_krb5_wrap.c @@ -1,10 +1,38 @@ +/* + * COPYRIGHT (c) 2008 + * The Regents of the University of Michigan + * ALL RIGHTS RESERVED + * + * Permission is granted to use, copy, create derivative works + * and redistribute this software and such derivative works + * for any purpose, so long as the name of The University of + * Michigan is not used in any advertising or publicity + * pertaining to the use of distribution of this software + * without specific, written prior authorization. If the + * above copyright notice or any other identification of the + * University of Michigan is included in any copy of any + * portion of this software, then the disclaimer below must + * also be included. + * + * THIS SOFTWARE IS PROVIDED AS IS, WITHOUT REPRESENTATION + * FROM THE UNIVERSITY OF MICHIGAN AS TO ITS FITNESS FOR ANY + * PURPOSE, AND WITHOUT WARRANTY BY THE UNIVERSITY OF + * MICHIGAN OF ANY KIND, EITHER EXPRESS OR IMPLIED, INCLUDING + * WITHOUT LIMITATION THE IMPLIED WARRANTIES OF + * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE. THE + * REGENTS OF THE UNIVERSITY OF MICHIGAN SHALL NOT BE LIABLE + * FOR ANY DAMAGES, INCLUDING SPECIAL, INDIRECT, INCIDENTAL, OR + * CONSEQUENTIAL DAMAGES, WITH RESPECT TO ANY CLAIM ARISING + * OUT OF OR IN CONNECTION WITH THE USE OF THE SOFTWARE, EVEN + * IF IT HAS BEEN OR IS HEREAFTER ADVISED OF THE POSSIBILITY OF + * SUCH DAMAGES. + */ + #include <linux/types.h> -#include <linux/slab.h> #include <linux/jiffies.h> #include <linux/sunrpc/gss_krb5.h> #include <linux/random.h> #include <linux/pagemap.h> -#include <asm/scatterlist.h> #include <linux/crypto.h> #ifdef RPC_DEBUG @@ -14,10 +42,7 @@ static inline int gss_krb5_padding(int blocksize, int length) { - /* Most of the code is block-size independent but currently we - * use only 8: */ - BUG_ON(blocksize != 8); - return 8 - (length & 7); + return blocksize - (length % blocksize); } static inline void @@ -42,7 +67,7 @@ gss_krb5_remove_padding(struct xdr_buf *buf, int blocksize) { u8 *ptr; u8 pad; - int len = buf->len; + size_t len = buf->len; if (len <= buf->head[0].iov_len) { pad = *(u8 *)(buf->head[0].iov_base + len - 1); @@ -53,13 +78,13 @@ gss_krb5_remove_padding(struct xdr_buf *buf, int blocksize) } else len -= buf->head[0].iov_len; if (len <= buf->page_len) { - int last = (buf->page_base + len - 1) + unsigned int last = (buf->page_base + len - 1) >>PAGE_CACHE_SHIFT; - int offset = (buf->page_base + len - 1) + unsigned int offset = (buf->page_base + len - 1) & (PAGE_CACHE_SIZE - 1); - ptr = kmap_atomic(buf->pages[last], KM_SKB_SUNRPC_DATA); + ptr = kmap_atomic(buf->pages[last]); pad = *(ptr + offset); - kunmap_atomic(ptr, KM_SKB_SUNRPC_DATA); + kunmap_atomic(ptr); goto out; } else len -= buf->page_len; @@ -88,8 +113,8 @@ out: return 0; } -static inline void -make_confounder(char *p, int blocksize) +void +gss_krb5_make_confounder(char *p, u32 conflen) { static u64 i = 0; u64 *q = (u64 *)p; @@ -103,8 +128,22 @@ make_confounder(char *p, int blocksize) * uniqueness would mean worrying about atomicity and rollover, and I * don't care enough. */ - BUG_ON(blocksize != 8); - *q = i++; + /* initialize to random value */ + if (i == 0) { + i = prandom_u32(); + i = (i << 32) | prandom_u32(); + } + + switch (conflen) { + case 16: + *q++ = i++; + /* fall through */ + case 8: + *q++ = i++; + break; + default: + BUG(); + } } /* Assumptions: the head and tail of inbuf are ours to play with. @@ -115,249 +154,467 @@ make_confounder(char *p, int blocksize) /* XXX factor out common code with seal/unseal. */ -u32 -gss_wrap_kerberos(struct gss_ctx *ctx, int offset, +static u32 +gss_wrap_kerberos_v1(struct krb5_ctx *kctx, int offset, struct xdr_buf *buf, struct page **pages) { - struct krb5_ctx *kctx = ctx->internal_ctx_id; - s32 checksum_type; - struct xdr_netobj md5cksum = {.len = 0, .data = NULL}; + char cksumdata[GSS_KRB5_MAX_CKSUM_LEN]; + struct xdr_netobj md5cksum = {.len = sizeof(cksumdata), + .data = cksumdata}; int blocksize = 0, plainlen; - unsigned char *ptr, *krb5_hdr, *msg_start; + unsigned char *ptr, *msg_start; s32 now; int headlen; struct page **tmp_pages; + u32 seq_send; + u8 *cksumkey; + u32 conflen = kctx->gk5e->conflen; - dprintk("RPC: gss_wrap_kerberos\n"); + dprintk("RPC: %s\n", __func__); now = get_seconds(); - switch (kctx->signalg) { - case SGN_ALG_DES_MAC_MD5: - checksum_type = CKSUMTYPE_RSA_MD5; - break; - default: - dprintk("RPC: gss_krb5_seal: kctx->signalg %d not" - " supported\n", kctx->signalg); - goto out_err; - } - if (kctx->sealalg != SEAL_ALG_NONE && kctx->sealalg != SEAL_ALG_DES) { - dprintk("RPC: gss_krb5_seal: kctx->sealalg %d not supported\n", - kctx->sealalg); - goto out_err; - } - - blocksize = crypto_tfm_alg_blocksize(kctx->enc); + blocksize = crypto_blkcipher_blocksize(kctx->enc); gss_krb5_add_padding(buf, offset, blocksize); BUG_ON((buf->len - offset) % blocksize); - plainlen = blocksize + buf->len - offset; + plainlen = conflen + buf->len - offset; - headlen = g_token_size(&kctx->mech_used, 22 + plainlen) - - (buf->len - offset); + headlen = g_token_size(&kctx->mech_used, + GSS_KRB5_TOK_HDR_LEN + kctx->gk5e->cksumlength + plainlen) - + (buf->len - offset); ptr = buf->head[0].iov_base + offset; /* shift data to make room for header. */ + xdr_extend_head(buf, offset, headlen); + /* XXX Would be cleverer to encrypt while copying. */ - /* XXX bounds checking, slack, etc. */ - memmove(ptr + headlen, ptr, buf->head[0].iov_len - offset); - buf->head[0].iov_len += headlen; - buf->len += headlen; BUG_ON((buf->len - offset - headlen) % blocksize); - g_make_token_header(&kctx->mech_used, 22 + plainlen, &ptr); + g_make_token_header(&kctx->mech_used, + GSS_KRB5_TOK_HDR_LEN + + kctx->gk5e->cksumlength + plainlen, &ptr); + + /* ptr now at header described in rfc 1964, section 1.2.1: */ + ptr[0] = (unsigned char) ((KG_TOK_WRAP_MSG >> 8) & 0xff); + ptr[1] = (unsigned char) (KG_TOK_WRAP_MSG & 0xff); - *ptr++ = (unsigned char) ((KG_TOK_WRAP_MSG>>8)&0xff); - *ptr++ = (unsigned char) (KG_TOK_WRAP_MSG&0xff); + msg_start = ptr + GSS_KRB5_TOK_HDR_LEN + kctx->gk5e->cksumlength; - /* ptr now at byte 2 of header described in rfc 1964, section 1.2.1: */ - krb5_hdr = ptr - 2; - msg_start = krb5_hdr + 24; - /* XXXJBF: */ BUG_ON(buf->head[0].iov_base + offset + headlen != msg_start + blocksize); + *(__be16 *)(ptr + 2) = cpu_to_le16(kctx->gk5e->signalg); + memset(ptr + 4, 0xff, 4); + *(__be16 *)(ptr + 4) = cpu_to_le16(kctx->gk5e->sealalg); - *(u16 *)(krb5_hdr + 2) = htons(kctx->signalg); - memset(krb5_hdr + 4, 0xff, 4); - *(u16 *)(krb5_hdr + 4) = htons(kctx->sealalg); + gss_krb5_make_confounder(msg_start, conflen); - make_confounder(msg_start, blocksize); + if (kctx->gk5e->keyed_cksum) + cksumkey = kctx->cksum; + else + cksumkey = NULL; /* XXXJBF: UGH!: */ tmp_pages = buf->pages; buf->pages = pages; - if (make_checksum(checksum_type, krb5_hdr, 8, buf, - offset + headlen - blocksize, &md5cksum)) - goto out_err; + if (make_checksum(kctx, ptr, 8, buf, offset + headlen - conflen, + cksumkey, KG_USAGE_SEAL, &md5cksum)) + return GSS_S_FAILURE; buf->pages = tmp_pages; - switch (kctx->signalg) { - case SGN_ALG_DES_MAC_MD5: - if (krb5_encrypt(kctx->seq, NULL, md5cksum.data, - md5cksum.data, md5cksum.len)) - goto out_err; - memcpy(krb5_hdr + 16, - md5cksum.data + md5cksum.len - KRB5_CKSUM_LENGTH, - KRB5_CKSUM_LENGTH); - - dprintk("RPC: make_seal_token: cksum data: \n"); - print_hexl((u32 *) (krb5_hdr + 16), KRB5_CKSUM_LENGTH, 0); - break; - default: - BUG(); - } + memcpy(ptr + GSS_KRB5_TOK_HDR_LEN, md5cksum.data, md5cksum.len); - kfree(md5cksum.data); + spin_lock(&krb5_seq_lock); + seq_send = kctx->seq_send++; + spin_unlock(&krb5_seq_lock); /* XXX would probably be more efficient to compute checksum * and encrypt at the same time: */ - if ((krb5_make_seq_num(kctx->seq, kctx->initiate ? 0 : 0xff, - kctx->seq_send, krb5_hdr + 16, krb5_hdr + 8))) - goto out_err; - - if (gss_encrypt_xdr_buf(kctx->enc, buf, offset + headlen - blocksize, - pages)) - goto out_err; - - kctx->seq_send++; + if ((krb5_make_seq_num(kctx, kctx->seq, kctx->initiate ? 0 : 0xff, + seq_send, ptr + GSS_KRB5_TOK_HDR_LEN, ptr + 8))) + return GSS_S_FAILURE; + + if (kctx->enctype == ENCTYPE_ARCFOUR_HMAC) { + struct crypto_blkcipher *cipher; + int err; + cipher = crypto_alloc_blkcipher(kctx->gk5e->encrypt_name, 0, + CRYPTO_ALG_ASYNC); + if (IS_ERR(cipher)) + return GSS_S_FAILURE; + + krb5_rc4_setup_enc_key(kctx, cipher, seq_send); + + err = gss_encrypt_xdr_buf(cipher, buf, + offset + headlen - conflen, pages); + crypto_free_blkcipher(cipher); + if (err) + return GSS_S_FAILURE; + } else { + if (gss_encrypt_xdr_buf(kctx->enc, buf, + offset + headlen - conflen, pages)) + return GSS_S_FAILURE; + } - return ((kctx->endtime < now) ? GSS_S_CONTEXT_EXPIRED : GSS_S_COMPLETE); -out_err: - if (md5cksum.data) kfree(md5cksum.data); - return GSS_S_FAILURE; + return (kctx->endtime < now) ? GSS_S_CONTEXT_EXPIRED : GSS_S_COMPLETE; } -u32 -gss_unwrap_kerberos(struct gss_ctx *ctx, int offset, struct xdr_buf *buf) +static u32 +gss_unwrap_kerberos_v1(struct krb5_ctx *kctx, int offset, struct xdr_buf *buf) { - struct krb5_ctx *kctx = ctx->internal_ctx_id; int signalg; int sealalg; - s32 checksum_type; - struct xdr_netobj md5cksum = {.len = 0, .data = NULL}; + char cksumdata[GSS_KRB5_MAX_CKSUM_LEN]; + struct xdr_netobj md5cksum = {.len = sizeof(cksumdata), + .data = cksumdata}; s32 now; int direction; s32 seqnum; unsigned char *ptr; int bodysize; - u32 ret = GSS_S_DEFECTIVE_TOKEN; void *data_start, *orig_start; int data_len; int blocksize; + u32 conflen = kctx->gk5e->conflen; + int crypt_offset; + u8 *cksumkey; - dprintk("RPC: gss_unwrap_kerberos\n"); + dprintk("RPC: gss_unwrap_kerberos\n"); ptr = (u8 *)buf->head[0].iov_base + offset; if (g_verify_token_header(&kctx->mech_used, &bodysize, &ptr, buf->len - offset)) - goto out; + return GSS_S_DEFECTIVE_TOKEN; - if ((*ptr++ != ((KG_TOK_WRAP_MSG>>8)&0xff)) || - (*ptr++ != (KG_TOK_WRAP_MSG &0xff)) ) - goto out; + if ((ptr[0] != ((KG_TOK_WRAP_MSG >> 8) & 0xff)) || + (ptr[1] != (KG_TOK_WRAP_MSG & 0xff))) + return GSS_S_DEFECTIVE_TOKEN; /* XXX sanity-check bodysize?? */ /* get the sign and seal algorithms */ - signalg = ptr[0] + (ptr[1] << 8); - sealalg = ptr[2] + (ptr[3] << 8); - - /* Sanity checks */ - - if ((ptr[4] != 0xff) || (ptr[5] != 0xff)) - goto out; - - if (sealalg == 0xffff) - goto out; + signalg = ptr[2] + (ptr[3] << 8); + if (signalg != kctx->gk5e->signalg) + return GSS_S_DEFECTIVE_TOKEN; - /* in the current spec, there is only one valid seal algorithm per - key type, so a simple comparison is ok */ + sealalg = ptr[4] + (ptr[5] << 8); + if (sealalg != kctx->gk5e->sealalg) + return GSS_S_DEFECTIVE_TOKEN; - if (sealalg != kctx->sealalg) - goto out; + if ((ptr[6] != 0xff) || (ptr[7] != 0xff)) + return GSS_S_DEFECTIVE_TOKEN; - /* there are several mappings of seal algorithms to sign algorithms, - but few enough that we can try them all. */ + /* + * Data starts after token header and checksum. ptr points + * to the beginning of the token header + */ + crypt_offset = ptr + (GSS_KRB5_TOK_HDR_LEN + kctx->gk5e->cksumlength) - + (unsigned char *)buf->head[0].iov_base; - if ((kctx->sealalg == SEAL_ALG_NONE && signalg > 1) || - (kctx->sealalg == SEAL_ALG_1 && signalg != SGN_ALG_3) || - (kctx->sealalg == SEAL_ALG_DES3KD && - signalg != SGN_ALG_HMAC_SHA1_DES3_KD)) - goto out; + /* + * Need plaintext seqnum to derive encryption key for arcfour-hmac + */ + if (krb5_get_seq_num(kctx, ptr + GSS_KRB5_TOK_HDR_LEN, + ptr + 8, &direction, &seqnum)) + return GSS_S_BAD_SIG; - if (gss_decrypt_xdr_buf(kctx->enc, buf, - ptr + 22 - (unsigned char *)buf->head[0].iov_base)) - goto out; + if ((kctx->initiate && direction != 0xff) || + (!kctx->initiate && direction != 0)) + return GSS_S_BAD_SIG; + + if (kctx->enctype == ENCTYPE_ARCFOUR_HMAC) { + struct crypto_blkcipher *cipher; + int err; + + cipher = crypto_alloc_blkcipher(kctx->gk5e->encrypt_name, 0, + CRYPTO_ALG_ASYNC); + if (IS_ERR(cipher)) + return GSS_S_FAILURE; + + krb5_rc4_setup_enc_key(kctx, cipher, seqnum); + + err = gss_decrypt_xdr_buf(cipher, buf, crypt_offset); + crypto_free_blkcipher(cipher); + if (err) + return GSS_S_DEFECTIVE_TOKEN; + } else { + if (gss_decrypt_xdr_buf(kctx->enc, buf, crypt_offset)) + return GSS_S_DEFECTIVE_TOKEN; + } - /* compute the checksum of the message */ + if (kctx->gk5e->keyed_cksum) + cksumkey = kctx->cksum; + else + cksumkey = NULL; - /* initialize the the cksum */ - switch (signalg) { - case SGN_ALG_DES_MAC_MD5: - checksum_type = CKSUMTYPE_RSA_MD5; - break; - default: - ret = GSS_S_DEFECTIVE_TOKEN; - goto out; - } + if (make_checksum(kctx, ptr, 8, buf, crypt_offset, + cksumkey, KG_USAGE_SEAL, &md5cksum)) + return GSS_S_FAILURE; - switch (signalg) { - case SGN_ALG_DES_MAC_MD5: - ret = make_checksum(checksum_type, ptr - 2, 8, buf, - ptr + 22 - (unsigned char *)buf->head[0].iov_base, &md5cksum); - if (ret) - goto out; - - ret = krb5_encrypt(kctx->seq, NULL, md5cksum.data, - md5cksum.data, md5cksum.len); - if (ret) - goto out; - - if (memcmp(md5cksum.data + 8, ptr + 14, 8)) { - ret = GSS_S_BAD_SIG; - goto out; - } - break; - default: - ret = GSS_S_DEFECTIVE_TOKEN; - goto out; - } + if (memcmp(md5cksum.data, ptr + GSS_KRB5_TOK_HDR_LEN, + kctx->gk5e->cksumlength)) + return GSS_S_BAD_SIG; /* it got through unscathed. Make sure the context is unexpired */ now = get_seconds(); - ret = GSS_S_CONTEXT_EXPIRED; if (now > kctx->endtime) - goto out; + return GSS_S_CONTEXT_EXPIRED; /* do sequencing checks */ - ret = GSS_S_BAD_SIG; - if ((ret = krb5_get_seq_num(kctx->seq, ptr + 14, ptr + 6, &direction, - &seqnum))) - goto out; - - if ((kctx->initiate && direction != 0xff) || - (!kctx->initiate && direction != 0)) - goto out; - /* Copy the data back to the right position. XXX: Would probably be * better to copy and encrypt at the same time. */ - blocksize = crypto_tfm_alg_blocksize(kctx->enc); - data_start = ptr + 22 + blocksize; + blocksize = crypto_blkcipher_blocksize(kctx->enc); + data_start = ptr + (GSS_KRB5_TOK_HDR_LEN + kctx->gk5e->cksumlength) + + conflen; orig_start = buf->head[0].iov_base + offset; data_len = (buf->head[0].iov_base + buf->head[0].iov_len) - data_start; memmove(orig_start, data_start, data_len); buf->head[0].iov_len -= (data_start - orig_start); buf->len -= (data_start - orig_start); - ret = GSS_S_DEFECTIVE_TOKEN; if (gss_krb5_remove_padding(buf, blocksize)) - goto out; + return GSS_S_DEFECTIVE_TOKEN; - ret = GSS_S_COMPLETE; -out: - if (md5cksum.data) kfree(md5cksum.data); - return ret; + return GSS_S_COMPLETE; } + +/* + * We can shift data by up to LOCAL_BUF_LEN bytes in a pass. If we need + * to do more than that, we shift repeatedly. Kevin Coffman reports + * seeing 28 bytes as the value used by Microsoft clients and servers + * with AES, so this constant is chosen to allow handling 28 in one pass + * without using too much stack space. + * + * If that proves to a problem perhaps we could use a more clever + * algorithm. + */ +#define LOCAL_BUF_LEN 32u + +static void rotate_buf_a_little(struct xdr_buf *buf, unsigned int shift) +{ + char head[LOCAL_BUF_LEN]; + char tmp[LOCAL_BUF_LEN]; + unsigned int this_len, i; + + BUG_ON(shift > LOCAL_BUF_LEN); + + read_bytes_from_xdr_buf(buf, 0, head, shift); + for (i = 0; i + shift < buf->len; i += LOCAL_BUF_LEN) { + this_len = min(LOCAL_BUF_LEN, buf->len - (i + shift)); + read_bytes_from_xdr_buf(buf, i+shift, tmp, this_len); + write_bytes_to_xdr_buf(buf, i, tmp, this_len); + } + write_bytes_to_xdr_buf(buf, buf->len - shift, head, shift); +} + +static void _rotate_left(struct xdr_buf *buf, unsigned int shift) +{ + int shifted = 0; + int this_shift; + + shift %= buf->len; + while (shifted < shift) { + this_shift = min(shift - shifted, LOCAL_BUF_LEN); + rotate_buf_a_little(buf, this_shift); + shifted += this_shift; + } +} + +static void rotate_left(u32 base, struct xdr_buf *buf, unsigned int shift) +{ + struct xdr_buf subbuf; + + xdr_buf_subsegment(buf, &subbuf, base, buf->len - base); + _rotate_left(&subbuf, shift); +} + +static u32 +gss_wrap_kerberos_v2(struct krb5_ctx *kctx, u32 offset, + struct xdr_buf *buf, struct page **pages) +{ + int blocksize; + u8 *ptr, *plainhdr; + s32 now; + u8 flags = 0x00; + __be16 *be16ptr, ec = 0; + __be64 *be64ptr; + u32 err; + + dprintk("RPC: %s\n", __func__); + + if (kctx->gk5e->encrypt_v2 == NULL) + return GSS_S_FAILURE; + + /* make room for gss token header */ + if (xdr_extend_head(buf, offset, GSS_KRB5_TOK_HDR_LEN)) + return GSS_S_FAILURE; + + /* construct gss token header */ + ptr = plainhdr = buf->head[0].iov_base + offset; + *ptr++ = (unsigned char) ((KG2_TOK_WRAP>>8) & 0xff); + *ptr++ = (unsigned char) (KG2_TOK_WRAP & 0xff); + + if ((kctx->flags & KRB5_CTX_FLAG_INITIATOR) == 0) + flags |= KG2_TOKEN_FLAG_SENTBYACCEPTOR; + if ((kctx->flags & KRB5_CTX_FLAG_ACCEPTOR_SUBKEY) != 0) + flags |= KG2_TOKEN_FLAG_ACCEPTORSUBKEY; + /* We always do confidentiality in wrap tokens */ + flags |= KG2_TOKEN_FLAG_SEALED; + + *ptr++ = flags; + *ptr++ = 0xff; + be16ptr = (__be16 *)ptr; + + blocksize = crypto_blkcipher_blocksize(kctx->acceptor_enc); + *be16ptr++ = cpu_to_be16(ec); + /* "inner" token header always uses 0 for RRC */ + *be16ptr++ = cpu_to_be16(0); + + be64ptr = (__be64 *)be16ptr; + spin_lock(&krb5_seq_lock); + *be64ptr = cpu_to_be64(kctx->seq_send64++); + spin_unlock(&krb5_seq_lock); + + err = (*kctx->gk5e->encrypt_v2)(kctx, offset, buf, ec, pages); + if (err) + return err; + + now = get_seconds(); + return (kctx->endtime < now) ? GSS_S_CONTEXT_EXPIRED : GSS_S_COMPLETE; +} + +static u32 +gss_unwrap_kerberos_v2(struct krb5_ctx *kctx, int offset, struct xdr_buf *buf) +{ + s32 now; + u8 *ptr; + u8 flags = 0x00; + u16 ec, rrc; + int err; + u32 headskip, tailskip; + u8 decrypted_hdr[GSS_KRB5_TOK_HDR_LEN]; + unsigned int movelen; + + + dprintk("RPC: %s\n", __func__); + + if (kctx->gk5e->decrypt_v2 == NULL) + return GSS_S_FAILURE; + + ptr = buf->head[0].iov_base + offset; + + if (be16_to_cpu(*((__be16 *)ptr)) != KG2_TOK_WRAP) + return GSS_S_DEFECTIVE_TOKEN; + + flags = ptr[2]; + if ((!kctx->initiate && (flags & KG2_TOKEN_FLAG_SENTBYACCEPTOR)) || + (kctx->initiate && !(flags & KG2_TOKEN_FLAG_SENTBYACCEPTOR))) + return GSS_S_BAD_SIG; + + if ((flags & KG2_TOKEN_FLAG_SEALED) == 0) { + dprintk("%s: token missing expected sealed flag\n", __func__); + return GSS_S_DEFECTIVE_TOKEN; + } + + if (ptr[3] != 0xff) + return GSS_S_DEFECTIVE_TOKEN; + + ec = be16_to_cpup((__be16 *)(ptr + 4)); + rrc = be16_to_cpup((__be16 *)(ptr + 6)); + + /* + * NOTE: the sequence number at ptr + 8 is skipped, rpcsec_gss + * doesn't want it checked; see page 6 of rfc 2203. + */ + + if (rrc != 0) + rotate_left(offset + 16, buf, rrc); + + err = (*kctx->gk5e->decrypt_v2)(kctx, offset, buf, + &headskip, &tailskip); + if (err) + return GSS_S_FAILURE; + + /* + * Retrieve the decrypted gss token header and verify + * it against the original + */ + err = read_bytes_from_xdr_buf(buf, + buf->len - GSS_KRB5_TOK_HDR_LEN - tailskip, + decrypted_hdr, GSS_KRB5_TOK_HDR_LEN); + if (err) { + dprintk("%s: error %u getting decrypted_hdr\n", __func__, err); + return GSS_S_FAILURE; + } + if (memcmp(ptr, decrypted_hdr, 6) + || memcmp(ptr + 8, decrypted_hdr + 8, 8)) { + dprintk("%s: token hdr, plaintext hdr mismatch!\n", __func__); + return GSS_S_FAILURE; + } + + /* do sequencing checks */ + + /* it got through unscathed. Make sure the context is unexpired */ + now = get_seconds(); + if (now > kctx->endtime) + return GSS_S_CONTEXT_EXPIRED; + + /* + * Move the head data back to the right position in xdr_buf. + * We ignore any "ec" data since it might be in the head or + * the tail, and we really don't need to deal with it. + * Note that buf->head[0].iov_len may indicate the available + * head buffer space rather than that actually occupied. + */ + movelen = min_t(unsigned int, buf->head[0].iov_len, buf->len); + movelen -= offset + GSS_KRB5_TOK_HDR_LEN + headskip; + BUG_ON(offset + GSS_KRB5_TOK_HDR_LEN + headskip + movelen > + buf->head[0].iov_len); + memmove(ptr, ptr + GSS_KRB5_TOK_HDR_LEN + headskip, movelen); + buf->head[0].iov_len -= GSS_KRB5_TOK_HDR_LEN + headskip; + buf->len -= GSS_KRB5_TOK_HDR_LEN + headskip; + + /* Trim off the trailing "extra count" and checksum blob */ + xdr_buf_trim(buf, ec + GSS_KRB5_TOK_HDR_LEN + tailskip); + return GSS_S_COMPLETE; +} + +u32 +gss_wrap_kerberos(struct gss_ctx *gctx, int offset, + struct xdr_buf *buf, struct page **pages) +{ + struct krb5_ctx *kctx = gctx->internal_ctx_id; + + switch (kctx->enctype) { + default: + BUG(); + case ENCTYPE_DES_CBC_RAW: + case ENCTYPE_DES3_CBC_RAW: + case ENCTYPE_ARCFOUR_HMAC: + return gss_wrap_kerberos_v1(kctx, offset, buf, pages); + case ENCTYPE_AES128_CTS_HMAC_SHA1_96: + case ENCTYPE_AES256_CTS_HMAC_SHA1_96: + return gss_wrap_kerberos_v2(kctx, offset, buf, pages); + } +} + +u32 +gss_unwrap_kerberos(struct gss_ctx *gctx, int offset, struct xdr_buf *buf) +{ + struct krb5_ctx *kctx = gctx->internal_ctx_id; + + switch (kctx->enctype) { + default: + BUG(); + case ENCTYPE_DES_CBC_RAW: + case ENCTYPE_DES3_CBC_RAW: + case ENCTYPE_ARCFOUR_HMAC: + return gss_unwrap_kerberos_v1(kctx, offset, buf); + case ENCTYPE_AES128_CTS_HMAC_SHA1_96: + case ENCTYPE_AES256_CTS_HMAC_SHA1_96: + return gss_unwrap_kerberos_v2(kctx, offset, buf); + } +} + diff --git a/net/sunrpc/auth_gss/gss_mech_switch.c b/net/sunrpc/auth_gss/gss_mech_switch.c index b048bf672da..92d5ab99fbf 100644 --- a/net/sunrpc/auth_gss/gss_mech_switch.c +++ b/net/sunrpc/auth_gss/gss_mech_switch.c @@ -6,14 +6,14 @@ * * J. Bruce Fields <bfields@umich.edu> * - * Redistribution and use in source and binary forms, with or without + * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright - * notice, this list of conditions and the following disclaimer in the + * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 3. Neither the name of the University nor the names of its * contributors may be used to endorse or promote products derived @@ -36,6 +36,7 @@ #include <linux/types.h> #include <linux/slab.h> #include <linux/module.h> +#include <linux/oid_registry.h> #include <linux/sunrpc/msg_prot.h> #include <linux/sunrpc/gss_asn1.h> #include <linux/sunrpc/auth_gss.h> @@ -60,8 +61,7 @@ gss_mech_free(struct gss_api_mech *gm) for (i = 0; i < gm->gm_pf_num; i++) { pf = &gm->gm_pfs[i]; - if (pf->auth_domain_name) - kfree(pf->auth_domain_name); + kfree(pf->auth_domain_name); pf->auth_domain_name = NULL; } } @@ -103,8 +103,13 @@ out: return status; } -int -gss_mech_register(struct gss_api_mech *gm) +/** + * gss_mech_register - register a GSS mechanism + * @gm: GSS mechanism handle + * + * Returns zero if successful, or a negative errno. + */ +int gss_mech_register(struct gss_api_mech *gm) { int status; @@ -114,35 +119,35 @@ gss_mech_register(struct gss_api_mech *gm) spin_lock(®istered_mechs_lock); list_add(&gm->gm_list, ®istered_mechs); spin_unlock(®istered_mechs_lock); - dprintk("RPC: registered gss mechanism %s\n", gm->gm_name); + dprintk("RPC: registered gss mechanism %s\n", gm->gm_name); return 0; } +EXPORT_SYMBOL_GPL(gss_mech_register); -EXPORT_SYMBOL(gss_mech_register); - -void -gss_mech_unregister(struct gss_api_mech *gm) +/** + * gss_mech_unregister - release a GSS mechanism + * @gm: GSS mechanism handle + * + */ +void gss_mech_unregister(struct gss_api_mech *gm) { spin_lock(®istered_mechs_lock); list_del(&gm->gm_list); spin_unlock(®istered_mechs_lock); - dprintk("RPC: unregistered gss mechanism %s\n", gm->gm_name); + dprintk("RPC: unregistered gss mechanism %s\n", gm->gm_name); gss_mech_free(gm); } +EXPORT_SYMBOL_GPL(gss_mech_unregister); -EXPORT_SYMBOL(gss_mech_unregister); - -struct gss_api_mech * -gss_mech_get(struct gss_api_mech *gm) +struct gss_api_mech *gss_mech_get(struct gss_api_mech *gm) { __module_get(gm->gm_owner); return gm; } - EXPORT_SYMBOL(gss_mech_get); -struct gss_api_mech * -gss_mech_get_by_name(const char *name) +static struct gss_api_mech * +_gss_mech_get_by_name(const char *name) { struct gss_api_mech *pos, *gm = NULL; @@ -159,7 +164,41 @@ gss_mech_get_by_name(const char *name) } -EXPORT_SYMBOL(gss_mech_get_by_name); +struct gss_api_mech * gss_mech_get_by_name(const char *name) +{ + struct gss_api_mech *gm = NULL; + + gm = _gss_mech_get_by_name(name); + if (!gm) { + request_module("rpc-auth-gss-%s", name); + gm = _gss_mech_get_by_name(name); + } + return gm; +} + +struct gss_api_mech *gss_mech_get_by_OID(struct rpcsec_gss_oid *obj) +{ + struct gss_api_mech *pos, *gm = NULL; + char buf[32]; + + if (sprint_oid(obj->data, obj->len, buf, sizeof(buf)) < 0) + return NULL; + dprintk("RPC: %s(%s)\n", __func__, buf); + request_module("rpc-auth-gss-%s", buf); + + spin_lock(®istered_mechs_lock); + list_for_each_entry(pos, ®istered_mechs, gm_list) { + if (obj->len == pos->gm_oid.len) { + if (0 == memcmp(obj->data, pos->gm_oid.data, obj->len)) { + if (try_module_get(pos->gm_owner)) + gm = pos; + break; + } + } + } + spin_unlock(®istered_mechs_lock); + return gm; +} static inline int mech_supports_pseudoflavor(struct gss_api_mech *gm, u32 pseudoflavor) @@ -173,17 +212,14 @@ mech_supports_pseudoflavor(struct gss_api_mech *gm, u32 pseudoflavor) return 0; } -struct gss_api_mech * -gss_mech_get_by_pseudoflavor(u32 pseudoflavor) +static struct gss_api_mech *_gss_mech_get_by_pseudoflavor(u32 pseudoflavor) { - struct gss_api_mech *pos, *gm = NULL; + struct gss_api_mech *gm = NULL, *pos; spin_lock(®istered_mechs_lock); list_for_each_entry(pos, ®istered_mechs, gm_list) { - if (!mech_supports_pseudoflavor(pos, pseudoflavor)) { - module_put(pos->gm_owner); + if (!mech_supports_pseudoflavor(pos, pseudoflavor)) continue; - } if (try_module_get(pos->gm_owner)) gm = pos; break; @@ -192,7 +228,125 @@ gss_mech_get_by_pseudoflavor(u32 pseudoflavor) return gm; } -EXPORT_SYMBOL(gss_mech_get_by_pseudoflavor); +struct gss_api_mech * +gss_mech_get_by_pseudoflavor(u32 pseudoflavor) +{ + struct gss_api_mech *gm; + + gm = _gss_mech_get_by_pseudoflavor(pseudoflavor); + + if (!gm) { + request_module("rpc-auth-gss-%u", pseudoflavor); + gm = _gss_mech_get_by_pseudoflavor(pseudoflavor); + } + return gm; +} + +/** + * gss_mech_list_pseudoflavors - Discover registered GSS pseudoflavors + * @array: array to fill in + * @size: size of "array" + * + * Returns the number of array items filled in, or a negative errno. + * + * The returned array is not sorted by any policy. Callers should not + * rely on the order of the items in the returned array. + */ +int gss_mech_list_pseudoflavors(rpc_authflavor_t *array_ptr, int size) +{ + struct gss_api_mech *pos = NULL; + int j, i = 0; + + spin_lock(®istered_mechs_lock); + list_for_each_entry(pos, ®istered_mechs, gm_list) { + for (j = 0; j < pos->gm_pf_num; j++) { + if (i >= size) { + spin_unlock(®istered_mechs_lock); + return -ENOMEM; + } + array_ptr[i++] = pos->gm_pfs[j].pseudoflavor; + } + } + spin_unlock(®istered_mechs_lock); + return i; +} + +/** + * gss_svc_to_pseudoflavor - map a GSS service number to a pseudoflavor + * @gm: GSS mechanism handle + * @qop: GSS quality-of-protection value + * @service: GSS service value + * + * Returns a matching security flavor, or RPC_AUTH_MAXFLAVOR if none is found. + */ +rpc_authflavor_t gss_svc_to_pseudoflavor(struct gss_api_mech *gm, u32 qop, + u32 service) +{ + int i; + + for (i = 0; i < gm->gm_pf_num; i++) { + if (gm->gm_pfs[i].qop == qop && + gm->gm_pfs[i].service == service) { + return gm->gm_pfs[i].pseudoflavor; + } + } + return RPC_AUTH_MAXFLAVOR; +} + +/** + * gss_mech_info2flavor - look up a pseudoflavor given a GSS tuple + * @info: a GSS mech OID, quality of protection, and service value + * + * Returns a matching pseudoflavor, or RPC_AUTH_MAXFLAVOR if the tuple is + * not supported. + */ +rpc_authflavor_t gss_mech_info2flavor(struct rpcsec_gss_info *info) +{ + rpc_authflavor_t pseudoflavor; + struct gss_api_mech *gm; + + gm = gss_mech_get_by_OID(&info->oid); + if (gm == NULL) + return RPC_AUTH_MAXFLAVOR; + + pseudoflavor = gss_svc_to_pseudoflavor(gm, info->qop, info->service); + + gss_mech_put(gm); + return pseudoflavor; +} + +/** + * gss_mech_flavor2info - look up a GSS tuple for a given pseudoflavor + * @pseudoflavor: GSS pseudoflavor to match + * @info: rpcsec_gss_info structure to fill in + * + * Returns zero and fills in "info" if pseudoflavor matches a + * supported mechanism. Otherwise a negative errno is returned. + */ +int gss_mech_flavor2info(rpc_authflavor_t pseudoflavor, + struct rpcsec_gss_info *info) +{ + struct gss_api_mech *gm; + int i; + + gm = gss_mech_get_by_pseudoflavor(pseudoflavor); + if (gm == NULL) + return -ENOENT; + + for (i = 0; i < gm->gm_pf_num; i++) { + if (gm->gm_pfs[i].pseudoflavor == pseudoflavor) { + memcpy(info->oid.data, gm->gm_oid.data, gm->gm_oid.len); + info->oid.len = gm->gm_oid.len; + info->qop = gm->gm_pfs[i].qop; + info->service = gm->gm_pfs[i].service; + gss_mech_put(gm); + return 0; + } + } + + gss_mech_put(gm); + return -ENOENT; +} u32 gss_pseudoflavor_to_service(struct gss_api_mech *gm, u32 pseudoflavor) @@ -205,7 +359,6 @@ gss_pseudoflavor_to_service(struct gss_api_mech *gm, u32 pseudoflavor) } return 0; } - EXPORT_SYMBOL(gss_pseudoflavor_to_service); char * @@ -220,14 +373,12 @@ gss_service_to_auth_domain_name(struct gss_api_mech *gm, u32 service) return NULL; } -EXPORT_SYMBOL(gss_service_to_auth_domain_name); - void gss_mech_put(struct gss_api_mech * gm) { - module_put(gm->gm_owner); + if (gm) + module_put(gm->gm_owner); } - EXPORT_SYMBOL(gss_mech_put); /* The mech could probably be determined from the token instead, but it's just @@ -235,15 +386,16 @@ EXPORT_SYMBOL(gss_mech_put); int gss_import_sec_context(const void *input_token, size_t bufsize, struct gss_api_mech *mech, - struct gss_ctx **ctx_id) + struct gss_ctx **ctx_id, + time_t *endtime, + gfp_t gfp_mask) { - if (!(*ctx_id = kmalloc(sizeof(**ctx_id), GFP_KERNEL))) - return GSS_S_FAILURE; - memset(*ctx_id, 0, sizeof(**ctx_id)); + if (!(*ctx_id = kzalloc(sizeof(**ctx_id), gfp_mask))) + return -ENOMEM; (*ctx_id)->mech_type = gss_mech_get(mech); - return mech->gm_ops - ->gss_import_sec_context(input_token, bufsize, *ctx_id); + return mech->gm_ops->gss_import_sec_context(input_token, bufsize, + *ctx_id, endtime, gfp_mask); } /* gss_get_mic: compute a mic over message and return mic_token. */ @@ -272,6 +424,20 @@ gss_verify_mic(struct gss_ctx *context_handle, mic_token); } +/* + * This function is called from both the client and server code. + * Each makes guarantees about how much "slack" space is available + * for the underlying function in "buf"'s head and tail while + * performing the wrap. + * + * The client and server code allocate RPC_MAX_AUTH_SIZE extra + * space in both the head and tail which is available for use by + * the wrap function. + * + * Underlying functions should verify they do not use more than + * RPC_MAX_AUTH_SIZE of extra space in either the head or tail + * when performing the wrap. + */ u32 gss_wrap(struct gss_ctx *ctx_id, int offset, @@ -299,17 +465,16 @@ gss_unwrap(struct gss_ctx *ctx_id, u32 gss_delete_sec_context(struct gss_ctx **context_handle) { - dprintk("RPC: gss_delete_sec_context deleting %p\n", + dprintk("RPC: gss_delete_sec_context deleting %p\n", *context_handle); if (!*context_handle) - return(GSS_S_NO_CONTEXT); - if ((*context_handle)->internal_ctx_id != 0) + return GSS_S_NO_CONTEXT; + if ((*context_handle)->internal_ctx_id) (*context_handle)->mech_type->gm_ops ->gss_delete_sec_context((*context_handle) ->internal_ctx_id); - if ((*context_handle)->mech_type) - gss_mech_put((*context_handle)->mech_type); + gss_mech_put((*context_handle)->mech_type); kfree(*context_handle); *context_handle=NULL; return GSS_S_COMPLETE; diff --git a/net/sunrpc/auth_gss/gss_rpc_upcall.c b/net/sunrpc/auth_gss/gss_rpc_upcall.c new file mode 100644 index 00000000000..abbb7dcd168 --- /dev/null +++ b/net/sunrpc/auth_gss/gss_rpc_upcall.c @@ -0,0 +1,382 @@ +/* + * linux/net/sunrpc/gss_rpc_upcall.c + * + * Copyright (C) 2012 Simo Sorce <simo@redhat.com> + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation; either version 2 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. + */ + +#include <linux/types.h> +#include <linux/un.h> + +#include <linux/sunrpc/svcauth.h> +#include "gss_rpc_upcall.h" + +#define GSSPROXY_SOCK_PATHNAME "/var/run/gssproxy.sock" + +#define GSSPROXY_PROGRAM (400112u) +#define GSSPROXY_VERS_1 (1u) + +/* + * Encoding/Decoding functions + */ + +enum { + GSSX_NULL = 0, /* Unused */ + GSSX_INDICATE_MECHS = 1, + GSSX_GET_CALL_CONTEXT = 2, + GSSX_IMPORT_AND_CANON_NAME = 3, + GSSX_EXPORT_CRED = 4, + GSSX_IMPORT_CRED = 5, + GSSX_ACQUIRE_CRED = 6, + GSSX_STORE_CRED = 7, + GSSX_INIT_SEC_CONTEXT = 8, + GSSX_ACCEPT_SEC_CONTEXT = 9, + GSSX_RELEASE_HANDLE = 10, + GSSX_GET_MIC = 11, + GSSX_VERIFY = 12, + GSSX_WRAP = 13, + GSSX_UNWRAP = 14, + GSSX_WRAP_SIZE_LIMIT = 15, +}; + +#define PROC(proc, name) \ +[GSSX_##proc] = { \ + .p_proc = GSSX_##proc, \ + .p_encode = (kxdreproc_t)gssx_enc_##name, \ + .p_decode = (kxdrdproc_t)gssx_dec_##name, \ + .p_arglen = GSSX_ARG_##name##_sz, \ + .p_replen = GSSX_RES_##name##_sz, \ + .p_statidx = GSSX_##proc, \ + .p_name = #proc, \ +} + +static struct rpc_procinfo gssp_procedures[] = { + PROC(INDICATE_MECHS, indicate_mechs), + PROC(GET_CALL_CONTEXT, get_call_context), + PROC(IMPORT_AND_CANON_NAME, import_and_canon_name), + PROC(EXPORT_CRED, export_cred), + PROC(IMPORT_CRED, import_cred), + PROC(ACQUIRE_CRED, acquire_cred), + PROC(STORE_CRED, store_cred), + PROC(INIT_SEC_CONTEXT, init_sec_context), + PROC(ACCEPT_SEC_CONTEXT, accept_sec_context), + PROC(RELEASE_HANDLE, release_handle), + PROC(GET_MIC, get_mic), + PROC(VERIFY, verify), + PROC(WRAP, wrap), + PROC(UNWRAP, unwrap), + PROC(WRAP_SIZE_LIMIT, wrap_size_limit), +}; + + + +/* + * Common transport functions + */ + +static const struct rpc_program gssp_program; + +static int gssp_rpc_create(struct net *net, struct rpc_clnt **_clnt) +{ + static const struct sockaddr_un gssp_localaddr = { + .sun_family = AF_LOCAL, + .sun_path = GSSPROXY_SOCK_PATHNAME, + }; + struct rpc_create_args args = { + .net = net, + .protocol = XPRT_TRANSPORT_LOCAL, + .address = (struct sockaddr *)&gssp_localaddr, + .addrsize = sizeof(gssp_localaddr), + .servername = "localhost", + .program = &gssp_program, + .version = GSSPROXY_VERS_1, + .authflavor = RPC_AUTH_NULL, + /* + * Note we want connection to be done in the caller's + * filesystem namespace. We therefore turn off the idle + * timeout, which would result in reconnections being + * done without the correct namespace: + */ + .flags = RPC_CLNT_CREATE_NOPING | + RPC_CLNT_CREATE_NO_IDLE_TIMEOUT + }; + struct rpc_clnt *clnt; + int result = 0; + + clnt = rpc_create(&args); + if (IS_ERR(clnt)) { + dprintk("RPC: failed to create AF_LOCAL gssproxy " + "client (errno %ld).\n", PTR_ERR(clnt)); + result = PTR_ERR(clnt); + *_clnt = NULL; + goto out; + } + + dprintk("RPC: created new gssp local client (gssp_local_clnt: " + "%p)\n", clnt); + *_clnt = clnt; + +out: + return result; +} + +void init_gssp_clnt(struct sunrpc_net *sn) +{ + mutex_init(&sn->gssp_lock); + sn->gssp_clnt = NULL; +} + +int set_gssp_clnt(struct net *net) +{ + struct sunrpc_net *sn = net_generic(net, sunrpc_net_id); + struct rpc_clnt *clnt; + int ret; + + mutex_lock(&sn->gssp_lock); + ret = gssp_rpc_create(net, &clnt); + if (!ret) { + if (sn->gssp_clnt) + rpc_shutdown_client(sn->gssp_clnt); + sn->gssp_clnt = clnt; + } + mutex_unlock(&sn->gssp_lock); + return ret; +} + +void clear_gssp_clnt(struct sunrpc_net *sn) +{ + mutex_lock(&sn->gssp_lock); + if (sn->gssp_clnt) { + rpc_shutdown_client(sn->gssp_clnt); + sn->gssp_clnt = NULL; + } + mutex_unlock(&sn->gssp_lock); +} + +static struct rpc_clnt *get_gssp_clnt(struct sunrpc_net *sn) +{ + struct rpc_clnt *clnt; + + mutex_lock(&sn->gssp_lock); + clnt = sn->gssp_clnt; + if (clnt) + atomic_inc(&clnt->cl_count); + mutex_unlock(&sn->gssp_lock); + return clnt; +} + +static int gssp_call(struct net *net, struct rpc_message *msg) +{ + struct sunrpc_net *sn = net_generic(net, sunrpc_net_id); + struct rpc_clnt *clnt; + int status; + + clnt = get_gssp_clnt(sn); + if (!clnt) + return -EIO; + status = rpc_call_sync(clnt, msg, 0); + if (status < 0) { + dprintk("gssp: rpc_call returned error %d\n", -status); + switch (status) { + case -EPROTONOSUPPORT: + status = -EINVAL; + break; + case -ECONNREFUSED: + case -ETIMEDOUT: + case -ENOTCONN: + status = -EAGAIN; + break; + case -ERESTARTSYS: + if (signalled ()) + status = -EINTR; + break; + default: + break; + } + } + rpc_release_client(clnt); + return status; +} + +static void gssp_free_receive_pages(struct gssx_arg_accept_sec_context *arg) +{ + int i; + + for (i = 0; i < arg->npages && arg->pages[i]; i++) + __free_page(arg->pages[i]); +} + +static int gssp_alloc_receive_pages(struct gssx_arg_accept_sec_context *arg) +{ + arg->npages = DIV_ROUND_UP(NGROUPS_MAX * 4, PAGE_SIZE); + arg->pages = kzalloc(arg->npages * sizeof(struct page *), GFP_KERNEL); + /* + * XXX: actual pages are allocated by xdr layer in + * xdr_partial_copy_from_skb. + */ + if (!arg->pages) + return -ENOMEM; + return 0; +} + +/* + * Public functions + */ + +/* numbers somewhat arbitrary but large enough for current needs */ +#define GSSX_MAX_OUT_HANDLE 128 +#define GSSX_MAX_SRC_PRINC 256 +#define GSSX_KMEMBUF (GSSX_max_output_handle_sz + \ + GSSX_max_oid_sz + \ + GSSX_max_princ_sz + \ + sizeof(struct svc_cred)) + +int gssp_accept_sec_context_upcall(struct net *net, + struct gssp_upcall_data *data) +{ + struct gssx_ctx ctxh = { + .state = data->in_handle + }; + struct gssx_arg_accept_sec_context arg = { + .input_token = data->in_token, + }; + struct gssx_ctx rctxh = { + /* + * pass in the max length we expect for each of these + * buffers but let the xdr code kmalloc them: + */ + .exported_context_token.len = GSSX_max_output_handle_sz, + .mech.len = GSS_OID_MAX_LEN, + .src_name.display_name.len = GSSX_max_princ_sz + }; + struct gssx_res_accept_sec_context res = { + .context_handle = &rctxh, + .output_token = &data->out_token + }; + struct rpc_message msg = { + .rpc_proc = &gssp_procedures[GSSX_ACCEPT_SEC_CONTEXT], + .rpc_argp = &arg, + .rpc_resp = &res, + .rpc_cred = NULL, /* FIXME ? */ + }; + struct xdr_netobj client_name = { 0 , NULL }; + int ret; + + if (data->in_handle.len != 0) + arg.context_handle = &ctxh; + res.output_token->len = GSSX_max_output_token_sz; + + ret = gssp_alloc_receive_pages(&arg); + if (ret) + return ret; + + /* use nfs/ for targ_name ? */ + + ret = gssp_call(net, &msg); + + gssp_free_receive_pages(&arg); + + /* we need to fetch all data even in case of error so + * that we can free special strctures is they have been allocated */ + data->major_status = res.status.major_status; + data->minor_status = res.status.minor_status; + if (res.context_handle) { + data->out_handle = rctxh.exported_context_token; + data->mech_oid.len = rctxh.mech.len; + if (rctxh.mech.data) + memcpy(data->mech_oid.data, rctxh.mech.data, + data->mech_oid.len); + client_name = rctxh.src_name.display_name; + } + + if (res.options.count == 1) { + gssx_buffer *value = &res.options.data[0].value; + /* Currently we only decode CREDS_VALUE, if we add + * anything else we'll have to loop and match on the + * option name */ + if (value->len == 1) { + /* steal group info from struct svc_cred */ + data->creds = *(struct svc_cred *)value->data; + data->found_creds = 1; + } + /* whether we use it or not, free data */ + kfree(value->data); + } + + if (res.options.count != 0) { + kfree(res.options.data); + } + + /* convert to GSS_NT_HOSTBASED_SERVICE form and set into creds */ + if (data->found_creds && client_name.data != NULL) { + char *c; + + data->creds.cr_principal = kstrndup(client_name.data, + client_name.len, GFP_KERNEL); + if (data->creds.cr_principal) { + /* terminate and remove realm part */ + c = strchr(data->creds.cr_principal, '@'); + if (c) { + *c = '\0'; + + /* change service-hostname delimiter */ + c = strchr(data->creds.cr_principal, '/'); + if (c) *c = '@'; + } + if (!c) { + /* not a service principal */ + kfree(data->creds.cr_principal); + data->creds.cr_principal = NULL; + } + } + } + kfree(client_name.data); + + return ret; +} + +void gssp_free_upcall_data(struct gssp_upcall_data *data) +{ + kfree(data->in_handle.data); + kfree(data->out_handle.data); + kfree(data->out_token.data); + free_svc_cred(&data->creds); +} + +/* + * Initialization stuff + */ + +static const struct rpc_version gssp_version1 = { + .number = GSSPROXY_VERS_1, + .nrprocs = ARRAY_SIZE(gssp_procedures), + .procs = gssp_procedures, +}; + +static const struct rpc_version *gssp_version[] = { + NULL, + &gssp_version1, +}; + +static struct rpc_stat gssp_stats; + +static const struct rpc_program gssp_program = { + .name = "gssproxy", + .number = GSSPROXY_PROGRAM, + .nrvers = ARRAY_SIZE(gssp_version), + .version = gssp_version, + .stats = &gssp_stats, +}; diff --git a/net/sunrpc/auth_gss/gss_rpc_upcall.h b/net/sunrpc/auth_gss/gss_rpc_upcall.h new file mode 100644 index 00000000000..1e542aded90 --- /dev/null +++ b/net/sunrpc/auth_gss/gss_rpc_upcall.h @@ -0,0 +1,48 @@ +/* + * linux/net/sunrpc/gss_rpc_upcall.h + * + * Copyright (C) 2012 Simo Sorce <simo@redhat.com> + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation; either version 2 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. + */ + +#ifndef _GSS_RPC_UPCALL_H +#define _GSS_RPC_UPCALL_H + +#include <linux/sunrpc/gss_api.h> +#include <linux/sunrpc/auth_gss.h> +#include "gss_rpc_xdr.h" +#include "../netns.h" + +struct gssp_upcall_data { + struct xdr_netobj in_handle; + struct gssp_in_token in_token; + struct xdr_netobj out_handle; + struct xdr_netobj out_token; + struct rpcsec_gss_oid mech_oid; + struct svc_cred creds; + int found_creds; + int major_status; + int minor_status; +}; + +int gssp_accept_sec_context_upcall(struct net *net, + struct gssp_upcall_data *data); +void gssp_free_upcall_data(struct gssp_upcall_data *data); + +void init_gssp_clnt(struct sunrpc_net *); +int set_gssp_clnt(struct net *); +void clear_gssp_clnt(struct sunrpc_net *); +#endif /* _GSS_RPC_UPCALL_H */ diff --git a/net/sunrpc/auth_gss/gss_rpc_xdr.c b/net/sunrpc/auth_gss/gss_rpc_xdr.c new file mode 100644 index 00000000000..1ec19f6f0c2 --- /dev/null +++ b/net/sunrpc/auth_gss/gss_rpc_xdr.c @@ -0,0 +1,839 @@ +/* + * GSS Proxy upcall module + * + * Copyright (C) 2012 Simo Sorce <simo@redhat.com> + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation; either version 2 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. + */ + +#include <linux/sunrpc/svcauth.h> +#include "gss_rpc_xdr.h" + +static int gssx_enc_bool(struct xdr_stream *xdr, int v) +{ + __be32 *p; + + p = xdr_reserve_space(xdr, 4); + if (unlikely(p == NULL)) + return -ENOSPC; + *p = v ? xdr_one : xdr_zero; + return 0; +} + +static int gssx_dec_bool(struct xdr_stream *xdr, u32 *v) +{ + __be32 *p; + + p = xdr_inline_decode(xdr, 4); + if (unlikely(p == NULL)) + return -ENOSPC; + *v = be32_to_cpu(*p); + return 0; +} + +static int gssx_enc_buffer(struct xdr_stream *xdr, + gssx_buffer *buf) +{ + __be32 *p; + + p = xdr_reserve_space(xdr, sizeof(u32) + buf->len); + if (!p) + return -ENOSPC; + xdr_encode_opaque(p, buf->data, buf->len); + return 0; +} + +static int gssx_enc_in_token(struct xdr_stream *xdr, + struct gssp_in_token *in) +{ + __be32 *p; + + p = xdr_reserve_space(xdr, 4); + if (!p) + return -ENOSPC; + *p = cpu_to_be32(in->page_len); + + /* all we need to do is to write pages */ + xdr_write_pages(xdr, in->pages, in->page_base, in->page_len); + + return 0; +} + + +static int gssx_dec_buffer(struct xdr_stream *xdr, + gssx_buffer *buf) +{ + u32 length; + __be32 *p; + + p = xdr_inline_decode(xdr, 4); + if (unlikely(p == NULL)) + return -ENOSPC; + + length = be32_to_cpup(p); + p = xdr_inline_decode(xdr, length); + if (unlikely(p == NULL)) + return -ENOSPC; + + if (buf->len == 0) { + /* we intentionally are not interested in this buffer */ + return 0; + } + if (length > buf->len) + return -ENOSPC; + + if (!buf->data) { + buf->data = kmemdup(p, length, GFP_KERNEL); + if (!buf->data) + return -ENOMEM; + } else { + memcpy(buf->data, p, length); + } + buf->len = length; + return 0; +} + +static int gssx_enc_option(struct xdr_stream *xdr, + struct gssx_option *opt) +{ + int err; + + err = gssx_enc_buffer(xdr, &opt->option); + if (err) + return err; + err = gssx_enc_buffer(xdr, &opt->value); + return err; +} + +static int gssx_dec_option(struct xdr_stream *xdr, + struct gssx_option *opt) +{ + int err; + + err = gssx_dec_buffer(xdr, &opt->option); + if (err) + return err; + err = gssx_dec_buffer(xdr, &opt->value); + return err; +} + +static int dummy_enc_opt_array(struct xdr_stream *xdr, + struct gssx_option_array *oa) +{ + __be32 *p; + + if (oa->count != 0) + return -EINVAL; + + p = xdr_reserve_space(xdr, 4); + if (!p) + return -ENOSPC; + *p = 0; + + return 0; +} + +static int dummy_dec_opt_array(struct xdr_stream *xdr, + struct gssx_option_array *oa) +{ + struct gssx_option dummy; + u32 count, i; + __be32 *p; + + p = xdr_inline_decode(xdr, 4); + if (unlikely(p == NULL)) + return -ENOSPC; + count = be32_to_cpup(p++); + memset(&dummy, 0, sizeof(dummy)); + for (i = 0; i < count; i++) { + gssx_dec_option(xdr, &dummy); + } + + oa->count = 0; + oa->data = NULL; + return 0; +} + +static int get_host_u32(struct xdr_stream *xdr, u32 *res) +{ + __be32 *p; + + p = xdr_inline_decode(xdr, 4); + if (!p) + return -EINVAL; + /* Contents of linux creds are all host-endian: */ + memcpy(res, p, sizeof(u32)); + return 0; +} + +static int gssx_dec_linux_creds(struct xdr_stream *xdr, + struct svc_cred *creds) +{ + u32 length; + __be32 *p; + u32 tmp; + u32 N; + int i, err; + + p = xdr_inline_decode(xdr, 4); + if (unlikely(p == NULL)) + return -ENOSPC; + + length = be32_to_cpup(p); + + if (length > (3 + NGROUPS_MAX) * sizeof(u32)) + return -ENOSPC; + + /* uid */ + err = get_host_u32(xdr, &tmp); + if (err) + return err; + creds->cr_uid = make_kuid(&init_user_ns, tmp); + + /* gid */ + err = get_host_u32(xdr, &tmp); + if (err) + return err; + creds->cr_gid = make_kgid(&init_user_ns, tmp); + + /* number of additional gid's */ + err = get_host_u32(xdr, &tmp); + if (err) + return err; + N = tmp; + if ((3 + N) * sizeof(u32) != length) + return -EINVAL; + creds->cr_group_info = groups_alloc(N); + if (creds->cr_group_info == NULL) + return -ENOMEM; + + /* gid's */ + for (i = 0; i < N; i++) { + kgid_t kgid; + err = get_host_u32(xdr, &tmp); + if (err) + goto out_free_groups; + err = -EINVAL; + kgid = make_kgid(&init_user_ns, tmp); + if (!gid_valid(kgid)) + goto out_free_groups; + GROUP_AT(creds->cr_group_info, i) = kgid; + } + + return 0; +out_free_groups: + groups_free(creds->cr_group_info); + return err; +} + +static int gssx_dec_option_array(struct xdr_stream *xdr, + struct gssx_option_array *oa) +{ + struct svc_cred *creds; + u32 count, i; + __be32 *p; + int err; + + p = xdr_inline_decode(xdr, 4); + if (unlikely(p == NULL)) + return -ENOSPC; + count = be32_to_cpup(p++); + if (!count) + return 0; + + /* we recognize only 1 currently: CREDS_VALUE */ + oa->count = 1; + + oa->data = kmalloc(sizeof(struct gssx_option), GFP_KERNEL); + if (!oa->data) + return -ENOMEM; + + creds = kmalloc(sizeof(struct svc_cred), GFP_KERNEL); + if (!creds) { + kfree(oa->data); + return -ENOMEM; + } + + oa->data[0].option.data = CREDS_VALUE; + oa->data[0].option.len = sizeof(CREDS_VALUE); + oa->data[0].value.data = (void *)creds; + oa->data[0].value.len = 0; + + for (i = 0; i < count; i++) { + gssx_buffer dummy = { 0, NULL }; + u32 length; + + /* option buffer */ + p = xdr_inline_decode(xdr, 4); + if (unlikely(p == NULL)) + return -ENOSPC; + + length = be32_to_cpup(p); + p = xdr_inline_decode(xdr, length); + if (unlikely(p == NULL)) + return -ENOSPC; + + if (length == sizeof(CREDS_VALUE) && + memcmp(p, CREDS_VALUE, sizeof(CREDS_VALUE)) == 0) { + /* We have creds here. parse them */ + err = gssx_dec_linux_creds(xdr, creds); + if (err) + return err; + oa->data[0].value.len = 1; /* presence */ + } else { + /* consume uninteresting buffer */ + err = gssx_dec_buffer(xdr, &dummy); + if (err) + return err; + } + } + return 0; +} + +static int gssx_dec_status(struct xdr_stream *xdr, + struct gssx_status *status) +{ + __be32 *p; + int err; + + /* status->major_status */ + p = xdr_inline_decode(xdr, 8); + if (unlikely(p == NULL)) + return -ENOSPC; + p = xdr_decode_hyper(p, &status->major_status); + + /* status->mech */ + err = gssx_dec_buffer(xdr, &status->mech); + if (err) + return err; + + /* status->minor_status */ + p = xdr_inline_decode(xdr, 8); + if (unlikely(p == NULL)) + return -ENOSPC; + p = xdr_decode_hyper(p, &status->minor_status); + + /* status->major_status_string */ + err = gssx_dec_buffer(xdr, &status->major_status_string); + if (err) + return err; + + /* status->minor_status_string */ + err = gssx_dec_buffer(xdr, &status->minor_status_string); + if (err) + return err; + + /* status->server_ctx */ + err = gssx_dec_buffer(xdr, &status->server_ctx); + if (err) + return err; + + /* we assume we have no options for now, so simply consume them */ + /* status->options */ + err = dummy_dec_opt_array(xdr, &status->options); + + return err; +} + +static int gssx_enc_call_ctx(struct xdr_stream *xdr, + struct gssx_call_ctx *ctx) +{ + struct gssx_option opt; + __be32 *p; + int err; + + /* ctx->locale */ + err = gssx_enc_buffer(xdr, &ctx->locale); + if (err) + return err; + + /* ctx->server_ctx */ + err = gssx_enc_buffer(xdr, &ctx->server_ctx); + if (err) + return err; + + /* we always want to ask for lucid contexts */ + /* ctx->options */ + p = xdr_reserve_space(xdr, 4); + *p = cpu_to_be32(2); + + /* we want a lucid_v1 context */ + opt.option.data = LUCID_OPTION; + opt.option.len = sizeof(LUCID_OPTION); + opt.value.data = LUCID_VALUE; + opt.value.len = sizeof(LUCID_VALUE); + err = gssx_enc_option(xdr, &opt); + + /* ..and user creds */ + opt.option.data = CREDS_OPTION; + opt.option.len = sizeof(CREDS_OPTION); + opt.value.data = CREDS_VALUE; + opt.value.len = sizeof(CREDS_VALUE); + err = gssx_enc_option(xdr, &opt); + + return err; +} + +static int gssx_dec_name_attr(struct xdr_stream *xdr, + struct gssx_name_attr *attr) +{ + int err; + + /* attr->attr */ + err = gssx_dec_buffer(xdr, &attr->attr); + if (err) + return err; + + /* attr->value */ + err = gssx_dec_buffer(xdr, &attr->value); + if (err) + return err; + + /* attr->extensions */ + err = dummy_dec_opt_array(xdr, &attr->extensions); + + return err; +} + +static int dummy_enc_nameattr_array(struct xdr_stream *xdr, + struct gssx_name_attr_array *naa) +{ + __be32 *p; + + if (naa->count != 0) + return -EINVAL; + + p = xdr_reserve_space(xdr, 4); + if (!p) + return -ENOSPC; + *p = 0; + + return 0; +} + +static int dummy_dec_nameattr_array(struct xdr_stream *xdr, + struct gssx_name_attr_array *naa) +{ + struct gssx_name_attr dummy = { .attr = {.len = 0} }; + u32 count, i; + __be32 *p; + + p = xdr_inline_decode(xdr, 4); + if (unlikely(p == NULL)) + return -ENOSPC; + count = be32_to_cpup(p++); + for (i = 0; i < count; i++) { + gssx_dec_name_attr(xdr, &dummy); + } + + naa->count = 0; + naa->data = NULL; + return 0; +} + +static struct xdr_netobj zero_netobj = {}; + +static struct gssx_name_attr_array zero_name_attr_array = {}; + +static struct gssx_option_array zero_option_array = {}; + +static int gssx_enc_name(struct xdr_stream *xdr, + struct gssx_name *name) +{ + int err; + + /* name->display_name */ + err = gssx_enc_buffer(xdr, &name->display_name); + if (err) + return err; + + /* name->name_type */ + err = gssx_enc_buffer(xdr, &zero_netobj); + if (err) + return err; + + /* name->exported_name */ + err = gssx_enc_buffer(xdr, &zero_netobj); + if (err) + return err; + + /* name->exported_composite_name */ + err = gssx_enc_buffer(xdr, &zero_netobj); + if (err) + return err; + + /* leave name_attributes empty for now, will add once we have any + * to pass up at all */ + /* name->name_attributes */ + err = dummy_enc_nameattr_array(xdr, &zero_name_attr_array); + if (err) + return err; + + /* leave options empty for now, will add once we have any options + * to pass up at all */ + /* name->extensions */ + err = dummy_enc_opt_array(xdr, &zero_option_array); + + return err; +} + + +static int gssx_dec_name(struct xdr_stream *xdr, + struct gssx_name *name) +{ + struct xdr_netobj dummy_netobj = { .len = 0 }; + struct gssx_name_attr_array dummy_name_attr_array = { .count = 0 }; + struct gssx_option_array dummy_option_array = { .count = 0 }; + int err; + + /* name->display_name */ + err = gssx_dec_buffer(xdr, &name->display_name); + if (err) + return err; + + /* name->name_type */ + err = gssx_dec_buffer(xdr, &dummy_netobj); + if (err) + return err; + + /* name->exported_name */ + err = gssx_dec_buffer(xdr, &dummy_netobj); + if (err) + return err; + + /* name->exported_composite_name */ + err = gssx_dec_buffer(xdr, &dummy_netobj); + if (err) + return err; + + /* we assume we have no attributes for now, so simply consume them */ + /* name->name_attributes */ + err = dummy_dec_nameattr_array(xdr, &dummy_name_attr_array); + if (err) + return err; + + /* we assume we have no options for now, so simply consume them */ + /* name->extensions */ + err = dummy_dec_opt_array(xdr, &dummy_option_array); + + return err; +} + +static int dummy_enc_credel_array(struct xdr_stream *xdr, + struct gssx_cred_element_array *cea) +{ + __be32 *p; + + if (cea->count != 0) + return -EINVAL; + + p = xdr_reserve_space(xdr, 4); + if (!p) + return -ENOSPC; + *p = 0; + + return 0; +} + +static int gssx_enc_cred(struct xdr_stream *xdr, + struct gssx_cred *cred) +{ + int err; + + /* cred->desired_name */ + err = gssx_enc_name(xdr, &cred->desired_name); + if (err) + return err; + + /* cred->elements */ + err = dummy_enc_credel_array(xdr, &cred->elements); + if (err) + return err; + + /* cred->cred_handle_reference */ + err = gssx_enc_buffer(xdr, &cred->cred_handle_reference); + if (err) + return err; + + /* cred->needs_release */ + err = gssx_enc_bool(xdr, cred->needs_release); + + return err; +} + +static int gssx_enc_ctx(struct xdr_stream *xdr, + struct gssx_ctx *ctx) +{ + __be32 *p; + int err; + + /* ctx->exported_context_token */ + err = gssx_enc_buffer(xdr, &ctx->exported_context_token); + if (err) + return err; + + /* ctx->state */ + err = gssx_enc_buffer(xdr, &ctx->state); + if (err) + return err; + + /* ctx->need_release */ + err = gssx_enc_bool(xdr, ctx->need_release); + if (err) + return err; + + /* ctx->mech */ + err = gssx_enc_buffer(xdr, &ctx->mech); + if (err) + return err; + + /* ctx->src_name */ + err = gssx_enc_name(xdr, &ctx->src_name); + if (err) + return err; + + /* ctx->targ_name */ + err = gssx_enc_name(xdr, &ctx->targ_name); + if (err) + return err; + + /* ctx->lifetime */ + p = xdr_reserve_space(xdr, 8+8); + if (!p) + return -ENOSPC; + p = xdr_encode_hyper(p, ctx->lifetime); + + /* ctx->ctx_flags */ + p = xdr_encode_hyper(p, ctx->ctx_flags); + + /* ctx->locally_initiated */ + err = gssx_enc_bool(xdr, ctx->locally_initiated); + if (err) + return err; + + /* ctx->open */ + err = gssx_enc_bool(xdr, ctx->open); + if (err) + return err; + + /* leave options empty for now, will add once we have any options + * to pass up at all */ + /* ctx->options */ + err = dummy_enc_opt_array(xdr, &ctx->options); + + return err; +} + +static int gssx_dec_ctx(struct xdr_stream *xdr, + struct gssx_ctx *ctx) +{ + __be32 *p; + int err; + + /* ctx->exported_context_token */ + err = gssx_dec_buffer(xdr, &ctx->exported_context_token); + if (err) + return err; + + /* ctx->state */ + err = gssx_dec_buffer(xdr, &ctx->state); + if (err) + return err; + + /* ctx->need_release */ + err = gssx_dec_bool(xdr, &ctx->need_release); + if (err) + return err; + + /* ctx->mech */ + err = gssx_dec_buffer(xdr, &ctx->mech); + if (err) + return err; + + /* ctx->src_name */ + err = gssx_dec_name(xdr, &ctx->src_name); + if (err) + return err; + + /* ctx->targ_name */ + err = gssx_dec_name(xdr, &ctx->targ_name); + if (err) + return err; + + /* ctx->lifetime */ + p = xdr_inline_decode(xdr, 8+8); + if (unlikely(p == NULL)) + return -ENOSPC; + p = xdr_decode_hyper(p, &ctx->lifetime); + + /* ctx->ctx_flags */ + p = xdr_decode_hyper(p, &ctx->ctx_flags); + + /* ctx->locally_initiated */ + err = gssx_dec_bool(xdr, &ctx->locally_initiated); + if (err) + return err; + + /* ctx->open */ + err = gssx_dec_bool(xdr, &ctx->open); + if (err) + return err; + + /* we assume we have no options for now, so simply consume them */ + /* ctx->options */ + err = dummy_dec_opt_array(xdr, &ctx->options); + + return err; +} + +static int gssx_enc_cb(struct xdr_stream *xdr, struct gssx_cb *cb) +{ + __be32 *p; + int err; + + /* cb->initiator_addrtype */ + p = xdr_reserve_space(xdr, 8); + if (!p) + return -ENOSPC; + p = xdr_encode_hyper(p, cb->initiator_addrtype); + + /* cb->initiator_address */ + err = gssx_enc_buffer(xdr, &cb->initiator_address); + if (err) + return err; + + /* cb->acceptor_addrtype */ + p = xdr_reserve_space(xdr, 8); + if (!p) + return -ENOSPC; + p = xdr_encode_hyper(p, cb->acceptor_addrtype); + + /* cb->acceptor_address */ + err = gssx_enc_buffer(xdr, &cb->acceptor_address); + if (err) + return err; + + /* cb->application_data */ + err = gssx_enc_buffer(xdr, &cb->application_data); + + return err; +} + +void gssx_enc_accept_sec_context(struct rpc_rqst *req, + struct xdr_stream *xdr, + struct gssx_arg_accept_sec_context *arg) +{ + int err; + + err = gssx_enc_call_ctx(xdr, &arg->call_ctx); + if (err) + goto done; + + /* arg->context_handle */ + if (arg->context_handle) + err = gssx_enc_ctx(xdr, arg->context_handle); + else + err = gssx_enc_bool(xdr, 0); + if (err) + goto done; + + /* arg->cred_handle */ + if (arg->cred_handle) + err = gssx_enc_cred(xdr, arg->cred_handle); + else + err = gssx_enc_bool(xdr, 0); + if (err) + goto done; + + /* arg->input_token */ + err = gssx_enc_in_token(xdr, &arg->input_token); + if (err) + goto done; + + /* arg->input_cb */ + if (arg->input_cb) + err = gssx_enc_cb(xdr, arg->input_cb); + else + err = gssx_enc_bool(xdr, 0); + if (err) + goto done; + + err = gssx_enc_bool(xdr, arg->ret_deleg_cred); + if (err) + goto done; + + /* leave options empty for now, will add once we have any options + * to pass up at all */ + /* arg->options */ + err = dummy_enc_opt_array(xdr, &arg->options); + + xdr_inline_pages(&req->rq_rcv_buf, + PAGE_SIZE/2 /* pretty arbitrary */, + arg->pages, 0 /* page base */, arg->npages * PAGE_SIZE); +done: + if (err) + dprintk("RPC: gssx_enc_accept_sec_context: %d\n", err); +} + +int gssx_dec_accept_sec_context(struct rpc_rqst *rqstp, + struct xdr_stream *xdr, + struct gssx_res_accept_sec_context *res) +{ + u32 value_follows; + int err; + + /* res->status */ + err = gssx_dec_status(xdr, &res->status); + if (err) + return err; + + /* res->context_handle */ + err = gssx_dec_bool(xdr, &value_follows); + if (err) + return err; + if (value_follows) { + err = gssx_dec_ctx(xdr, res->context_handle); + if (err) + return err; + } else { + res->context_handle = NULL; + } + + /* res->output_token */ + err = gssx_dec_bool(xdr, &value_follows); + if (err) + return err; + if (value_follows) { + err = gssx_dec_buffer(xdr, res->output_token); + if (err) + return err; + } else { + res->output_token = NULL; + } + + /* res->delegated_cred_handle */ + err = gssx_dec_bool(xdr, &value_follows); + if (err) + return err; + if (value_follows) { + /* we do not support upcall servers sending this data. */ + return -EINVAL; + } + + /* res->options */ + err = gssx_dec_option_array(xdr, &res->options); + + return err; +} diff --git a/net/sunrpc/auth_gss/gss_rpc_xdr.h b/net/sunrpc/auth_gss/gss_rpc_xdr.h new file mode 100644 index 00000000000..685a688f3d8 --- /dev/null +++ b/net/sunrpc/auth_gss/gss_rpc_xdr.h @@ -0,0 +1,267 @@ +/* + * GSS Proxy upcall module + * + * Copyright (C) 2012 Simo Sorce <simo@redhat.com> + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation; either version 2 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. + */ + +#ifndef _LINUX_GSS_RPC_XDR_H +#define _LINUX_GSS_RPC_XDR_H + +#include <linux/sunrpc/xdr.h> +#include <linux/sunrpc/clnt.h> +#include <linux/sunrpc/xprtsock.h> + +#ifdef RPC_DEBUG +# define RPCDBG_FACILITY RPCDBG_AUTH +#endif + +#define LUCID_OPTION "exported_context_type" +#define LUCID_VALUE "linux_lucid_v1" +#define CREDS_OPTION "exported_creds_type" +#define CREDS_VALUE "linux_creds_v1" + +typedef struct xdr_netobj gssx_buffer; +typedef struct xdr_netobj utf8string; +typedef struct xdr_netobj gssx_OID; + +enum gssx_cred_usage { + GSSX_C_INITIATE = 1, + GSSX_C_ACCEPT = 2, + GSSX_C_BOTH = 3, +}; + +struct gssx_option { + gssx_buffer option; + gssx_buffer value; +}; + +struct gssx_option_array { + u32 count; + struct gssx_option *data; +}; + +struct gssx_status { + u64 major_status; + gssx_OID mech; + u64 minor_status; + utf8string major_status_string; + utf8string minor_status_string; + gssx_buffer server_ctx; + struct gssx_option_array options; +}; + +struct gssx_call_ctx { + utf8string locale; + gssx_buffer server_ctx; + struct gssx_option_array options; +}; + +struct gssx_name_attr { + gssx_buffer attr; + gssx_buffer value; + struct gssx_option_array extensions; +}; + +struct gssx_name_attr_array { + u32 count; + struct gssx_name_attr *data; +}; + +struct gssx_name { + gssx_buffer display_name; +}; +typedef struct gssx_name gssx_name; + +struct gssx_cred_element { + gssx_name MN; + gssx_OID mech; + u32 cred_usage; + u64 initiator_time_rec; + u64 acceptor_time_rec; + struct gssx_option_array options; +}; + +struct gssx_cred_element_array { + u32 count; + struct gssx_cred_element *data; +}; + +struct gssx_cred { + gssx_name desired_name; + struct gssx_cred_element_array elements; + gssx_buffer cred_handle_reference; + u32 needs_release; +}; + +struct gssx_ctx { + gssx_buffer exported_context_token; + gssx_buffer state; + u32 need_release; + gssx_OID mech; + gssx_name src_name; + gssx_name targ_name; + u64 lifetime; + u64 ctx_flags; + u32 locally_initiated; + u32 open; + struct gssx_option_array options; +}; + +struct gssx_cb { + u64 initiator_addrtype; + gssx_buffer initiator_address; + u64 acceptor_addrtype; + gssx_buffer acceptor_address; + gssx_buffer application_data; +}; + + +/* This structure is not defined in the protocol. + * It is used in the kernel to carry around a big buffer + * as a set of pages */ +struct gssp_in_token { + struct page **pages; /* Array of contiguous pages */ + unsigned int page_base; /* Start of page data */ + unsigned int page_len; /* Length of page data */ +}; + +struct gssx_arg_accept_sec_context { + struct gssx_call_ctx call_ctx; + struct gssx_ctx *context_handle; + struct gssx_cred *cred_handle; + struct gssp_in_token input_token; + struct gssx_cb *input_cb; + u32 ret_deleg_cred; + struct gssx_option_array options; + struct page **pages; + unsigned int npages; +}; + +struct gssx_res_accept_sec_context { + struct gssx_status status; + struct gssx_ctx *context_handle; + gssx_buffer *output_token; + /* struct gssx_cred *delegated_cred_handle; not used in kernel */ + struct gssx_option_array options; +}; + + + +#define gssx_enc_indicate_mechs NULL +#define gssx_dec_indicate_mechs NULL +#define gssx_enc_get_call_context NULL +#define gssx_dec_get_call_context NULL +#define gssx_enc_import_and_canon_name NULL +#define gssx_dec_import_and_canon_name NULL +#define gssx_enc_export_cred NULL +#define gssx_dec_export_cred NULL +#define gssx_enc_import_cred NULL +#define gssx_dec_import_cred NULL +#define gssx_enc_acquire_cred NULL +#define gssx_dec_acquire_cred NULL +#define gssx_enc_store_cred NULL +#define gssx_dec_store_cred NULL +#define gssx_enc_init_sec_context NULL +#define gssx_dec_init_sec_context NULL +void gssx_enc_accept_sec_context(struct rpc_rqst *req, + struct xdr_stream *xdr, + struct gssx_arg_accept_sec_context *args); +int gssx_dec_accept_sec_context(struct rpc_rqst *rqstp, + struct xdr_stream *xdr, + struct gssx_res_accept_sec_context *res); +#define gssx_enc_release_handle NULL +#define gssx_dec_release_handle NULL +#define gssx_enc_get_mic NULL +#define gssx_dec_get_mic NULL +#define gssx_enc_verify NULL +#define gssx_dec_verify NULL +#define gssx_enc_wrap NULL +#define gssx_dec_wrap NULL +#define gssx_enc_unwrap NULL +#define gssx_dec_unwrap NULL +#define gssx_enc_wrap_size_limit NULL +#define gssx_dec_wrap_size_limit NULL + +/* non implemented calls are set to 0 size */ +#define GSSX_ARG_indicate_mechs_sz 0 +#define GSSX_RES_indicate_mechs_sz 0 +#define GSSX_ARG_get_call_context_sz 0 +#define GSSX_RES_get_call_context_sz 0 +#define GSSX_ARG_import_and_canon_name_sz 0 +#define GSSX_RES_import_and_canon_name_sz 0 +#define GSSX_ARG_export_cred_sz 0 +#define GSSX_RES_export_cred_sz 0 +#define GSSX_ARG_import_cred_sz 0 +#define GSSX_RES_import_cred_sz 0 +#define GSSX_ARG_acquire_cred_sz 0 +#define GSSX_RES_acquire_cred_sz 0 +#define GSSX_ARG_store_cred_sz 0 +#define GSSX_RES_store_cred_sz 0 +#define GSSX_ARG_init_sec_context_sz 0 +#define GSSX_RES_init_sec_context_sz 0 + +#define GSSX_default_in_call_ctx_sz (4 + 4 + 4 + \ + 8 + sizeof(LUCID_OPTION) + sizeof(LUCID_VALUE) + \ + 8 + sizeof(CREDS_OPTION) + sizeof(CREDS_VALUE)) +#define GSSX_default_in_ctx_hndl_sz (4 + 4+8 + 4 + 4 + 6*4 + 6*4 + 8 + 8 + \ + 4 + 4 + 4) +#define GSSX_default_in_cred_sz 4 /* we send in no cred_handle */ +#define GSSX_default_in_token_sz 4 /* does *not* include token data */ +#define GSSX_default_in_cb_sz 4 /* we do not use channel bindings */ +#define GSSX_ARG_accept_sec_context_sz (GSSX_default_in_call_ctx_sz + \ + GSSX_default_in_ctx_hndl_sz + \ + GSSX_default_in_cred_sz + \ + GSSX_default_in_token_sz + \ + GSSX_default_in_cb_sz + \ + 4 /* no deleg creds boolean */ + \ + 4) /* empty options */ + +/* somewhat arbitrary numbers but large enough (we ignore some of the data + * sent down, but it is part of the protocol so we need enough space to take + * it in) */ +#define GSSX_default_status_sz 8 + 24 + 8 + 256 + 256 + 16 + 4 +#define GSSX_max_output_handle_sz 128 +#define GSSX_max_oid_sz 16 +#define GSSX_max_princ_sz 256 +#define GSSX_default_ctx_sz (GSSX_max_output_handle_sz + \ + 16 + 4 + GSSX_max_oid_sz + \ + 2 * GSSX_max_princ_sz + \ + 8 + 8 + 4 + 4 + 4) +#define GSSX_max_output_token_sz 1024 +/* grouplist not included; we allocate separate pages for that: */ +#define GSSX_max_creds_sz (4 + 4 + 4 /* + NGROUPS_MAX*4 */) +#define GSSX_RES_accept_sec_context_sz (GSSX_default_status_sz + \ + GSSX_default_ctx_sz + \ + GSSX_max_output_token_sz + \ + 4 + GSSX_max_creds_sz) + +#define GSSX_ARG_release_handle_sz 0 +#define GSSX_RES_release_handle_sz 0 +#define GSSX_ARG_get_mic_sz 0 +#define GSSX_RES_get_mic_sz 0 +#define GSSX_ARG_verify_sz 0 +#define GSSX_RES_verify_sz 0 +#define GSSX_ARG_wrap_sz 0 +#define GSSX_RES_wrap_sz 0 +#define GSSX_ARG_unwrap_sz 0 +#define GSSX_RES_unwrap_sz 0 +#define GSSX_ARG_wrap_size_limit_sz 0 +#define GSSX_RES_wrap_size_limit_sz 0 + + + +#endif /* _LINUX_GSS_RPC_XDR_H */ diff --git a/net/sunrpc/auth_gss/gss_spkm3_mech.c b/net/sunrpc/auth_gss/gss_spkm3_mech.c deleted file mode 100644 index 39b3edc1469..00000000000 --- a/net/sunrpc/auth_gss/gss_spkm3_mech.c +++ /dev/null @@ -1,291 +0,0 @@ -/* - * linux/net/sunrpc/gss_spkm3_mech.c - * - * Copyright (c) 2003 The Regents of the University of Michigan. - * All rights reserved. - * - * Andy Adamson <andros@umich.edu> - * J. Bruce Fields <bfields@umich.edu> - * - * Redistribution and use in source and binary forms, with or without - * modification, are permitted provided that the following conditions - * are met: - * - * 1. Redistributions of source code must retain the above copyright - * notice, this list of conditions and the following disclaimer. - * 2. Redistributions in binary form must reproduce the above copyright - * notice, this list of conditions and the following disclaimer in the - * documentation and/or other materials provided with the distribution. - * 3. Neither the name of the University nor the names of its - * contributors may be used to endorse or promote products derived - * from this software without specific prior written permission. - * - * THIS SOFTWARE IS PROVIDED ``AS IS'' AND ANY EXPRESS OR IMPLIED - * WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF - * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE - * DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE - * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR - * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF - * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR - * BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF - * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING - * NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS - * SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. - * - */ - -#include <linux/module.h> -#include <linux/init.h> -#include <linux/types.h> -#include <linux/slab.h> -#include <linux/sunrpc/auth.h> -#include <linux/in.h> -#include <linux/sunrpc/svcauth_gss.h> -#include <linux/sunrpc/gss_spkm3.h> -#include <linux/sunrpc/xdr.h> -#include <linux/crypto.h> - -#ifdef RPC_DEBUG -# define RPCDBG_FACILITY RPCDBG_AUTH -#endif - -static const void * -simple_get_bytes(const void *p, const void *end, void *res, int len) -{ - const void *q = (const void *)((const char *)p + len); - if (unlikely(q > end || q < p)) - return ERR_PTR(-EFAULT); - memcpy(res, p, len); - return q; -} - -static const void * -simple_get_netobj(const void *p, const void *end, struct xdr_netobj *res) -{ - const void *q; - unsigned int len; - p = simple_get_bytes(p, end, &len, sizeof(len)); - if (IS_ERR(p)) - return p; - res->len = len; - if (len == 0) { - res->data = NULL; - return p; - } - q = (const void *)((const char *)p + len); - if (unlikely(q > end || q < p)) - return ERR_PTR(-EFAULT); - res->data = kmalloc(len, GFP_KERNEL); - if (unlikely(res->data == NULL)) - return ERR_PTR(-ENOMEM); - memcpy(res->data, p, len); - return q; -} - -static inline const void * -get_key(const void *p, const void *end, struct crypto_tfm **res, int *resalg) -{ - struct xdr_netobj key = { 0 }; - int alg_mode,setkey = 0; - char *alg_name; - - p = simple_get_bytes(p, end, resalg, sizeof(*resalg)); - if (IS_ERR(p)) - goto out_err; - p = simple_get_netobj(p, end, &key); - if (IS_ERR(p)) - goto out_err; - - switch (*resalg) { - case NID_des_cbc: - alg_name = "des"; - alg_mode = CRYPTO_TFM_MODE_CBC; - setkey = 1; - break; - case NID_md5: - if (key.len == 0) { - dprintk("RPC: SPKM3 get_key: NID_md5 zero Key length\n"); - } - alg_name = "md5"; - alg_mode = 0; - setkey = 0; - break; - default: - dprintk("RPC: SPKM3 get_key: unsupported algorithm %d", *resalg); - goto out_err_free_key; - } - if (!(*res = crypto_alloc_tfm(alg_name, alg_mode))) - goto out_err_free_key; - if (setkey) { - if (crypto_cipher_setkey(*res, key.data, key.len)) - goto out_err_free_tfm; - } - - if(key.len > 0) - kfree(key.data); - return p; - -out_err_free_tfm: - crypto_free_tfm(*res); -out_err_free_key: - if(key.len > 0) - kfree(key.data); - p = ERR_PTR(-EINVAL); -out_err: - return p; -} - -static int -gss_import_sec_context_spkm3(const void *p, size_t len, - struct gss_ctx *ctx_id) -{ - const void *end = (const void *)((const char *)p + len); - struct spkm3_ctx *ctx; - - if (!(ctx = kmalloc(sizeof(*ctx), GFP_KERNEL))) - goto out_err; - memset(ctx, 0, sizeof(*ctx)); - - p = simple_get_netobj(p, end, &ctx->ctx_id); - if (IS_ERR(p)) - goto out_err_free_ctx; - - p = simple_get_bytes(p, end, &ctx->qop, sizeof(ctx->qop)); - if (IS_ERR(p)) - goto out_err_free_ctx_id; - - p = simple_get_netobj(p, end, &ctx->mech_used); - if (IS_ERR(p)) - goto out_err_free_mech; - - p = simple_get_bytes(p, end, &ctx->ret_flags, sizeof(ctx->ret_flags)); - if (IS_ERR(p)) - goto out_err_free_mech; - - p = simple_get_bytes(p, end, &ctx->req_flags, sizeof(ctx->req_flags)); - if (IS_ERR(p)) - goto out_err_free_mech; - - p = simple_get_netobj(p, end, &ctx->share_key); - if (IS_ERR(p)) - goto out_err_free_s_key; - - p = get_key(p, end, &ctx->derived_conf_key, &ctx->conf_alg); - if (IS_ERR(p)) - goto out_err_free_s_key; - - p = get_key(p, end, &ctx->derived_integ_key, &ctx->intg_alg); - if (IS_ERR(p)) - goto out_err_free_key1; - - p = simple_get_bytes(p, end, &ctx->keyestb_alg, sizeof(ctx->keyestb_alg)); - if (IS_ERR(p)) - goto out_err_free_key2; - - p = simple_get_bytes(p, end, &ctx->owf_alg, sizeof(ctx->owf_alg)); - if (IS_ERR(p)) - goto out_err_free_key2; - - if (p != end) - goto out_err_free_key2; - - ctx_id->internal_ctx_id = ctx; - - dprintk("Succesfully imported new spkm context.\n"); - return 0; - -out_err_free_key2: - crypto_free_tfm(ctx->derived_integ_key); -out_err_free_key1: - crypto_free_tfm(ctx->derived_conf_key); -out_err_free_s_key: - kfree(ctx->share_key.data); -out_err_free_mech: - kfree(ctx->mech_used.data); -out_err_free_ctx_id: - kfree(ctx->ctx_id.data); -out_err_free_ctx: - kfree(ctx); -out_err: - return PTR_ERR(p); -} - -static void -gss_delete_sec_context_spkm3(void *internal_ctx) { - struct spkm3_ctx *sctx = internal_ctx; - - crypto_free_tfm(sctx->derived_integ_key); - crypto_free_tfm(sctx->derived_conf_key); - kfree(sctx->share_key.data); - kfree(sctx->mech_used.data); - kfree(sctx); -} - -static u32 -gss_verify_mic_spkm3(struct gss_ctx *ctx, - struct xdr_buf *signbuf, - struct xdr_netobj *checksum) -{ - u32 maj_stat = 0; - struct spkm3_ctx *sctx = ctx->internal_ctx_id; - - dprintk("RPC: gss_verify_mic_spkm3 calling spkm3_read_token\n"); - maj_stat = spkm3_read_token(sctx, checksum, signbuf, SPKM_MIC_TOK); - - dprintk("RPC: gss_verify_mic_spkm3 returning %d\n", maj_stat); - return maj_stat; -} - -static u32 -gss_get_mic_spkm3(struct gss_ctx *ctx, - struct xdr_buf *message_buffer, - struct xdr_netobj *message_token) -{ - u32 err = 0; - struct spkm3_ctx *sctx = ctx->internal_ctx_id; - - dprintk("RPC: gss_get_mic_spkm3\n"); - - err = spkm3_make_token(sctx, message_buffer, - message_token, SPKM_MIC_TOK); - return err; -} - -static struct gss_api_ops gss_spkm3_ops = { - .gss_import_sec_context = gss_import_sec_context_spkm3, - .gss_get_mic = gss_get_mic_spkm3, - .gss_verify_mic = gss_verify_mic_spkm3, - .gss_delete_sec_context = gss_delete_sec_context_spkm3, -}; - -static struct pf_desc gss_spkm3_pfs[] = { - {RPC_AUTH_GSS_SPKM, RPC_GSS_SVC_NONE, "spkm3"}, - {RPC_AUTH_GSS_SPKMI, RPC_GSS_SVC_INTEGRITY, "spkm3i"}, -}; - -static struct gss_api_mech gss_spkm3_mech = { - .gm_name = "spkm3", - .gm_owner = THIS_MODULE, - .gm_ops = &gss_spkm3_ops, - .gm_pf_num = ARRAY_SIZE(gss_spkm3_pfs), - .gm_pfs = gss_spkm3_pfs, -}; - -static int __init init_spkm3_module(void) -{ - int status; - - status = gss_mech_register(&gss_spkm3_mech); - if (status) - printk("Failed to register spkm3 gss mechanism!\n"); - return 0; -} - -static void __exit cleanup_spkm3_module(void) -{ - gss_mech_unregister(&gss_spkm3_mech); -} - -MODULE_LICENSE("GPL"); -module_init(init_spkm3_module); -module_exit(cleanup_spkm3_module); diff --git a/net/sunrpc/auth_gss/gss_spkm3_seal.c b/net/sunrpc/auth_gss/gss_spkm3_seal.c deleted file mode 100644 index 148201e929d..00000000000 --- a/net/sunrpc/auth_gss/gss_spkm3_seal.c +++ /dev/null @@ -1,130 +0,0 @@ -/* - * linux/net/sunrpc/gss_spkm3_seal.c - * - * Copyright (c) 2003 The Regents of the University of Michigan. - * All rights reserved. - * - * Andy Adamson <andros@umich.edu> - * - * Redistribution and use in source and binary forms, with or without - * modification, are permitted provided that the following conditions - * are met: - * - * 1. Redistributions of source code must retain the above copyright - * notice, this list of conditions and the following disclaimer. - * 2. Redistributions in binary form must reproduce the above copyright - * notice, this list of conditions and the following disclaimer in the - * documentation and/or other materials provided with the distribution. - * 3. Neither the name of the University nor the names of its - * contributors may be used to endorse or promote products derived - * from this software without specific prior written permission. - * - * THIS SOFTWARE IS PROVIDED ``AS IS'' AND ANY EXPRESS OR IMPLIED - * WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF - * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE - * DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE - * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR - * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF - * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR - * BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF - * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING - * NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS - * SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. - * - */ - -#include <linux/types.h> -#include <linux/slab.h> -#include <linux/jiffies.h> -#include <linux/sunrpc/gss_spkm3.h> -#include <linux/random.h> -#include <linux/crypto.h> - -#ifdef RPC_DEBUG -# define RPCDBG_FACILITY RPCDBG_AUTH -#endif - -/* - * spkm3_make_token() - * - * Only SPKM_MIC_TOK with md5 intg-alg is supported - */ - -u32 -spkm3_make_token(struct spkm3_ctx *ctx, - struct xdr_buf * text, struct xdr_netobj * token, - int toktype) -{ - s32 checksum_type; - char tokhdrbuf[25]; - struct xdr_netobj md5cksum = {.len = 0, .data = NULL}; - struct xdr_netobj mic_hdr = {.len = 0, .data = tokhdrbuf}; - int tmsglen, tokenlen = 0; - unsigned char *ptr; - s32 now; - int ctxelen = 0, ctxzbit = 0; - int md5elen = 0, md5zbit = 0; - - dprintk("RPC: spkm3_make_token\n"); - - now = jiffies; - - if (ctx->ctx_id.len != 16) { - dprintk("RPC: spkm3_make_token BAD ctx_id.len %d\n", - ctx->ctx_id.len); - goto out_err; - } - - switch (ctx->intg_alg) { - case NID_md5: - checksum_type = CKSUMTYPE_RSA_MD5; - break; - default: - dprintk("RPC: gss_spkm3_seal: ctx->signalg %d not" - " supported\n", ctx->intg_alg); - goto out_err; - } - /* XXX since we don't support WRAP, perhaps we don't care... */ - if (ctx->conf_alg != NID_cast5_cbc) { - dprintk("RPC: gss_spkm3_seal: ctx->sealalg %d not supported\n", - ctx->conf_alg); - goto out_err; - } - - if (toktype == SPKM_MIC_TOK) { - tmsglen = 0; - /* Calculate checksum over the mic-header */ - asn1_bitstring_len(&ctx->ctx_id, &ctxelen, &ctxzbit); - spkm3_mic_header(&mic_hdr.data, &mic_hdr.len, ctx->ctx_id.data, - ctxelen, ctxzbit); - - if (make_checksum(checksum_type, mic_hdr.data, mic_hdr.len, - text, &md5cksum)) - goto out_err; - - asn1_bitstring_len(&md5cksum, &md5elen, &md5zbit); - tokenlen = 10 + ctxelen + 1 + 2 + md5elen + 1; - - /* Create token header using generic routines */ - token->len = g_token_size(&ctx->mech_used, tokenlen + tmsglen); - - ptr = token->data; - g_make_token_header(&ctx->mech_used, tokenlen + tmsglen, &ptr); - - spkm3_make_mic_token(&ptr, tokenlen, &mic_hdr, &md5cksum, md5elen, md5zbit); - } else if (toktype == SPKM_WRAP_TOK) { /* Not Supported */ - dprintk("RPC: gss_spkm3_seal: SPKM_WRAP_TOK not supported\n"); - goto out_err; - } - kfree(md5cksum.data); - - /* XXX need to implement sequence numbers, and ctx->expired */ - - return GSS_S_COMPLETE; -out_err: - if (md5cksum.data) - kfree(md5cksum.data); - token->data = NULL; - token->len = 0; - return GSS_S_FAILURE; -} diff --git a/net/sunrpc/auth_gss/gss_spkm3_token.c b/net/sunrpc/auth_gss/gss_spkm3_token.c deleted file mode 100644 index 46c08a0710f..00000000000 --- a/net/sunrpc/auth_gss/gss_spkm3_token.c +++ /dev/null @@ -1,266 +0,0 @@ -/* - * linux/net/sunrpc/gss_spkm3_token.c - * - * Copyright (c) 2003 The Regents of the University of Michigan. - * All rights reserved. - * - * Andy Adamson <andros@umich.edu> - * - * Redistribution and use in source and binary forms, with or without - * modification, are permitted provided that the following conditions - * are met: - * - * 1. Redistributions of source code must retain the above copyright - * notice, this list of conditions and the following disclaimer. - * 2. Redistributions in binary form must reproduce the above copyright - * notice, this list of conditions and the following disclaimer in the - * documentation and/or other materials provided with the distribution. - * 3. Neither the name of the University nor the names of its - * contributors may be used to endorse or promote products derived - * from this software without specific prior written permission. - * - * THIS SOFTWARE IS PROVIDED ``AS IS'' AND ANY EXPRESS OR IMPLIED - * WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF - * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE - * DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE - * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR - * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF - * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR - * BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF - * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING - * NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS - * SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. - * - */ - -#include <linux/types.h> -#include <linux/slab.h> -#include <linux/jiffies.h> -#include <linux/sunrpc/gss_spkm3.h> -#include <linux/random.h> -#include <linux/crypto.h> - -#ifdef RPC_DEBUG -# define RPCDBG_FACILITY RPCDBG_AUTH -#endif - -/* - * asn1_bitstring_len() - * - * calculate the asn1 bitstring length of the xdr_netobject - */ -void -asn1_bitstring_len(struct xdr_netobj *in, int *enclen, int *zerobits) -{ - int i, zbit = 0,elen = in->len; - char *ptr; - - ptr = &in->data[in->len -1]; - - /* count trailing 0's */ - for(i = in->len; i > 0; i--) { - if (*ptr == 0) { - ptr--; - elen--; - } else - break; - } - - /* count number of 0 bits in final octet */ - ptr = &in->data[elen - 1]; - for(i = 0; i < 8; i++) { - short mask = 0x01; - - if (!((mask << i) & *ptr)) - zbit++; - else - break; - } - *enclen = elen; - *zerobits = zbit; -} - -/* - * decode_asn1_bitstring() - * - * decode a bitstring into a buffer of the expected length. - * enclen = bit string length - * explen = expected length (define in rfc) - */ -int -decode_asn1_bitstring(struct xdr_netobj *out, char *in, int enclen, int explen) -{ - if (!(out->data = kmalloc(explen,GFP_KERNEL))) - return 0; - out->len = explen; - memset(out->data, 0, explen); - memcpy(out->data, in, enclen); - return 1; -} - -/* - * SPKMInnerContextToken choice SPKM_MIC asn1 token layout - * - * contextid is always 16 bytes plain data. max asn1 bitstring len = 17. - * - * tokenlen = pos[0] to end of token (max pos[45] with MD5 cksum) - * - * pos value - * ---------- - * [0] a4 SPKM-MIC tag - * [1] ?? innertoken length (max 44) - * - * - * tok_hdr piece of checksum data starts here - * - * the maximum mic-header len = 9 + 17 = 26 - * mic-header - * ---------- - * [2] 30 SEQUENCE tag - * [3] ?? mic-header length: (max 23) = TokenID + ContextID - * - * TokenID - all fields constant and can be hardcoded - * ------- - * [4] 02 Type 2 - * [5] 02 Length 2 - * [6][7] 01 01 TokenID (SPKM_MIC_TOK) - * - * ContextID - encoded length not constant, calculated - * --------- - * [8] 03 Type 3 - * [9] ?? encoded length - * [10] ?? ctxzbit - * [11] contextid - * - * mic_header piece of checksum data ends here. - * - * int-cksum - encoded length not constant, calculated - * --------- - * [??] 03 Type 3 - * [??] ?? encoded length - * [??] ?? md5zbit - * [??] int-cksum (NID_md5 = 16) - * - * maximum SPKM-MIC innercontext token length = - * 10 + encoded contextid_size(17 max) + 2 + encoded - * cksum_size (17 maxfor NID_md5) = 46 - */ - -/* - * spkm3_mic_header() - * - * Prepare the SPKM_MIC_TOK mic-header for check-sum calculation - * elen: 16 byte context id asn1 bitstring encoded length - */ -void -spkm3_mic_header(unsigned char **hdrbuf, unsigned int *hdrlen, unsigned char *ctxdata, int elen, int zbit) -{ - char *hptr = *hdrbuf; - char *top = *hdrbuf; - - *(u8 *)hptr++ = 0x30; - *(u8 *)hptr++ = elen + 7; /* on the wire header length */ - - /* tokenid */ - *(u8 *)hptr++ = 0x02; - *(u8 *)hptr++ = 0x02; - *(u8 *)hptr++ = 0x01; - *(u8 *)hptr++ = 0x01; - - /* coniextid */ - *(u8 *)hptr++ = 0x03; - *(u8 *)hptr++ = elen + 1; /* add 1 to include zbit */ - *(u8 *)hptr++ = zbit; - memcpy(hptr, ctxdata, elen); - hptr += elen; - *hdrlen = hptr - top; -} - -/* - * spkm3_mic_innercontext_token() - * - * *tokp points to the beginning of the SPKM_MIC token described - * in rfc 2025, section 3.2.1: - * - */ -void -spkm3_make_mic_token(unsigned char **tokp, int toklen, struct xdr_netobj *mic_hdr, struct xdr_netobj *md5cksum, int md5elen, int md5zbit) -{ - unsigned char *ict = *tokp; - - *(u8 *)ict++ = 0xa4; - *(u8 *)ict++ = toklen - 2; - memcpy(ict, mic_hdr->data, mic_hdr->len); - ict += mic_hdr->len; - - *(u8 *)ict++ = 0x03; - *(u8 *)ict++ = md5elen + 1; /* add 1 to include zbit */ - *(u8 *)ict++ = md5zbit; - memcpy(ict, md5cksum->data, md5elen); -} - -u32 -spkm3_verify_mic_token(unsigned char **tokp, int *mic_hdrlen, unsigned char **cksum) -{ - struct xdr_netobj spkm3_ctx_id = {.len =0, .data = NULL}; - unsigned char *ptr = *tokp; - int ctxelen; - u32 ret = GSS_S_DEFECTIVE_TOKEN; - - /* spkm3 innercontext token preamble */ - if ((ptr[0] != 0xa4) || (ptr[2] != 0x30)) { - dprintk("RPC: BAD SPKM ictoken preamble\n"); - goto out; - } - - *mic_hdrlen = ptr[3]; - - /* token type */ - if ((ptr[4] != 0x02) || (ptr[5] != 0x02)) { - dprintk("RPC: BAD asn1 SPKM3 token type\n"); - goto out; - } - - /* only support SPKM_MIC_TOK */ - if((ptr[6] != 0x01) || (ptr[7] != 0x01)) { - dprintk("RPC: ERROR unsupported SPKM3 token \n"); - goto out; - } - - /* contextid */ - if (ptr[8] != 0x03) { - dprintk("RPC: BAD SPKM3 asn1 context-id type\n"); - goto out; - } - - ctxelen = ptr[9]; - if (ctxelen > 17) { /* length includes asn1 zbit octet */ - dprintk("RPC: BAD SPKM3 contextid len %d\n", ctxelen); - goto out; - } - - /* ignore ptr[10] */ - - if(!decode_asn1_bitstring(&spkm3_ctx_id, &ptr[11], ctxelen - 1, 16)) - goto out; - - /* - * in the current implementation: the optional int-alg is not present - * so the default int-alg (md5) is used the optional snd-seq field is - * also not present - */ - - if (*mic_hdrlen != 6 + ctxelen) { - dprintk("RPC: BAD SPKM_ MIC_TOK header len %d: we only support default int-alg (should be absent) and do not support snd-seq\n", *mic_hdrlen); - goto out; - } - /* checksum */ - *cksum = (&ptr[10] + ctxelen); /* ctxelen includes ptr[10] */ - - ret = GSS_S_COMPLETE; -out: - if (spkm3_ctx_id.data) - kfree(spkm3_ctx_id.data); - return ret; -} - diff --git a/net/sunrpc/auth_gss/gss_spkm3_unseal.c b/net/sunrpc/auth_gss/gss_spkm3_unseal.c deleted file mode 100644 index c3c0d958610..00000000000 --- a/net/sunrpc/auth_gss/gss_spkm3_unseal.c +++ /dev/null @@ -1,128 +0,0 @@ -/* - * linux/net/sunrpc/gss_spkm3_unseal.c - * - * Copyright (c) 2003 The Regents of the University of Michigan. - * All rights reserved. - * - * Andy Adamson <andros@umich.edu> - * - * Redistribution and use in source and binary forms, with or without - * modification, are permitted provided that the following conditions - * are met: - * - * 1. Redistributions of source code must retain the above copyright - * notice, this list of conditions and the following disclaimer. - * 2. Redistributions in binary form must reproduce the above copyright - * notice, this list of conditions and the following disclaimer in the - * documentation and/or other materials provided with the distribution. - * 3. Neither the name of the University nor the names of its - * contributors may be used to endorse or promote products derived - * from this software without specific prior written permission. - * - * THIS SOFTWARE IS PROVIDED ``AS IS'' AND ANY EXPRESS OR IMPLIED - * WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF - * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE - * DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE - * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR - * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF - * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR - * BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF - * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING - * NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS - * SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. - * - */ - -#include <linux/types.h> -#include <linux/slab.h> -#include <linux/jiffies.h> -#include <linux/sunrpc/gss_spkm3.h> -#include <linux/crypto.h> - -#ifdef RPC_DEBUG -# define RPCDBG_FACILITY RPCDBG_AUTH -#endif - -/* - * spkm3_read_token() - * - * only SPKM_MIC_TOK with md5 intg-alg is supported - */ -u32 -spkm3_read_token(struct spkm3_ctx *ctx, - struct xdr_netobj *read_token, /* checksum */ - struct xdr_buf *message_buffer, /* signbuf */ - int toktype) -{ - s32 code; - struct xdr_netobj wire_cksum = {.len =0, .data = NULL}; - struct xdr_netobj md5cksum = {.len = 0, .data = NULL}; - unsigned char *ptr = (unsigned char *)read_token->data; - unsigned char *cksum; - int bodysize, md5elen; - int mic_hdrlen; - u32 ret = GSS_S_DEFECTIVE_TOKEN; - - dprintk("RPC: spkm3_read_token read_token->len %d\n", read_token->len); - - if (g_verify_token_header((struct xdr_netobj *) &ctx->mech_used, - &bodysize, &ptr, read_token->len)) - goto out; - - /* decode the token */ - - if (toktype == SPKM_MIC_TOK) { - - if ((ret = spkm3_verify_mic_token(&ptr, &mic_hdrlen, &cksum))) - goto out; - - if (*cksum++ != 0x03) { - dprintk("RPC: spkm3_read_token BAD checksum type\n"); - goto out; - } - md5elen = *cksum++; - cksum++; /* move past the zbit */ - - if(!decode_asn1_bitstring(&wire_cksum, cksum, md5elen - 1, 16)) - goto out; - - /* HARD CODED FOR MD5 */ - - /* compute the checksum of the message. - * ptr + 2 = start of header piece of checksum - * mic_hdrlen + 2 = length of header piece of checksum - */ - ret = GSS_S_DEFECTIVE_TOKEN; - code = make_checksum(CKSUMTYPE_RSA_MD5, ptr + 2, - mic_hdrlen + 2, - message_buffer, &md5cksum); - - if (code) - goto out; - - dprintk("RPC: spkm3_read_token: digest wire_cksum.len %d:\n", - wire_cksum.len); - dprintk(" md5cksum.data\n"); - print_hexl((u32 *) md5cksum.data, 16, 0); - dprintk(" cksum.data:\n"); - print_hexl((u32 *) wire_cksum.data, wire_cksum.len, 0); - - ret = GSS_S_BAD_SIG; - code = memcmp(md5cksum.data, wire_cksum.data, wire_cksum.len); - if (code) - goto out; - - } else { - dprintk("RPC: BAD or UNSUPPORTED SPKM3 token type: %d\n",toktype); - goto out; - } - - /* XXX: need to add expiration and sequencing */ - ret = GSS_S_COMPLETE; -out: - if (md5cksum.data) - kfree(md5cksum.data); - if (wire_cksum.data) - kfree(wire_cksum.data); - return ret; -} diff --git a/net/sunrpc/auth_gss/svcauth_gss.c b/net/sunrpc/auth_gss/svcauth_gss.c index e4ada15ed85..4ce5eccec1f 100644 --- a/net/sunrpc/auth_gss/svcauth_gss.c +++ b/net/sunrpc/auth_gss/svcauth_gss.c @@ -37,16 +37,19 @@ * */ +#include <linux/slab.h> #include <linux/types.h> #include <linux/module.h> #include <linux/pagemap.h> +#include <linux/user_namespace.h> #include <linux/sunrpc/auth_gss.h> -#include <linux/sunrpc/svcauth.h> #include <linux/sunrpc/gss_err.h> #include <linux/sunrpc/svcauth.h> #include <linux/sunrpc/svcauth_gss.h> #include <linux/sunrpc/cache.h> +#include "gss_rpc_upcall.h" + #ifdef RPC_DEBUG # define RPCDBG_FACILITY RPCDBG_AUTH @@ -67,7 +70,6 @@ static int netobj_equal(struct xdr_netobj *a, struct xdr_netobj *b) #define RSI_HASHBITS 6 #define RSI_HASHMAX (1<<RSI_HASHBITS) -#define RSI_HASHMASK (RSI_HASHMAX-1) struct rsi { struct cache_head h; @@ -76,9 +78,8 @@ struct rsi { int major_status, minor_status; }; -static struct cache_head *rsi_table[RSI_HASHMAX]; -static struct cache_detail rsi_cache; -static struct rsi *rsi_lookup(struct rsi *item, int set); +static struct rsi *rsi_update(struct cache_detail *cd, struct rsi *new, struct rsi *old); +static struct rsi *rsi_lookup(struct cache_detail *cd, struct rsi *item); static void rsi_free(struct rsi *rsii) { @@ -88,13 +89,11 @@ static void rsi_free(struct rsi *rsii) kfree(rsii->out_token.data); } -static void rsi_put(struct cache_head *item, struct cache_detail *cd) +static void rsi_put(struct kref *ref) { - struct rsi *rsii = container_of(item, struct rsi, h); - if (cache_put(item, cd)) { - rsi_free(rsii); - kfree(rsii); - } + struct rsi *rsii = container_of(ref, struct rsi, h.ref); + rsi_free(rsii); + kfree(rsii); } static inline int rsi_hash(struct rsi *item) @@ -103,18 +102,18 @@ static inline int rsi_hash(struct rsi *item) ^ hash_mem(item->in_token.data, item->in_token.len, RSI_HASHBITS); } -static inline int rsi_match(struct rsi *item, struct rsi *tmp) +static int rsi_match(struct cache_head *a, struct cache_head *b) { - return netobj_equal(&item->in_handle, &tmp->in_handle) - && netobj_equal(&item->in_token, &tmp->in_token); + struct rsi *item = container_of(a, struct rsi, h); + struct rsi *tmp = container_of(b, struct rsi, h); + return netobj_equal(&item->in_handle, &tmp->in_handle) && + netobj_equal(&item->in_token, &tmp->in_token); } static int dup_to_netobj(struct xdr_netobj *dst, char *src, int len) { dst->len = len; - dst->data = (len ? kmalloc(len, GFP_KERNEL) : NULL); - if (dst->data) - memcpy(dst->data, src, len); + dst->data = (len ? kmemdup(src, len, GFP_KERNEL) : NULL); if (len && !dst->data) return -ENOMEM; return 0; @@ -125,8 +124,11 @@ static inline int dup_netobj(struct xdr_netobj *dst, struct xdr_netobj *src) return dup_to_netobj(dst, src->data, src->len); } -static inline void rsi_init(struct rsi *new, struct rsi *item) +static void rsi_init(struct cache_head *cnew, struct cache_head *citem) { + struct rsi *new = container_of(cnew, struct rsi, h); + struct rsi *item = container_of(citem, struct rsi, h); + new->out_handle.data = NULL; new->out_handle.len = 0; new->out_token.data = NULL; @@ -141,8 +143,11 @@ static inline void rsi_init(struct rsi *new, struct rsi *item) item->in_token.data = NULL; } -static inline void rsi_update(struct rsi *new, struct rsi *item) +static void update_rsi(struct cache_head *cnew, struct cache_head *citem) { + struct rsi *new = container_of(cnew, struct rsi, h); + struct rsi *item = container_of(citem, struct rsi, h); + BUG_ON(new->out_handle.data || new->out_token.data); new->out_handle.len = item->out_handle.len; item->out_handle.len = 0; @@ -157,9 +162,18 @@ static inline void rsi_update(struct rsi *new, struct rsi *item) new->minor_status = item->minor_status; } +static struct cache_head *rsi_alloc(void) +{ + struct rsi *rsii = kmalloc(sizeof(*rsii), GFP_KERNEL); + if (rsii) + return &rsii->h; + else + return NULL; +} + static void rsi_request(struct cache_detail *cd, - struct cache_head *h, - char **bpp, int *blen) + struct cache_head *h, + char **bpp, int *blen) { struct rsi *rsii = container_of(h, struct rsi, h); @@ -168,9 +182,8 @@ static void rsi_request(struct cache_detail *cd, (*bpp)[-1] = '\n'; } - static int rsi_parse(struct cache_detail *cd, - char *mesg, int mlen) + char *mesg, int mlen) { /* context token expiry major minor context token */ char *buf = mesg; @@ -198,6 +211,10 @@ static int rsi_parse(struct cache_detail *cd, if (dup_to_netobj(&rsii.in_token, buf, len)) goto out; + rsip = rsi_lookup(cd, &rsii); + if (!rsip) + goto out; + rsii.h.flags = 0; /* expiry */ expiry = get_expiry(&mesg); @@ -207,59 +224,84 @@ static int rsi_parse(struct cache_detail *cd, /* major/minor */ len = qword_get(&mesg, buf, mlen); - if (len < 0) + if (len <= 0) goto out; - if (len == 0) { + rsii.major_status = simple_strtoul(buf, &ep, 10); + if (*ep) + goto out; + len = qword_get(&mesg, buf, mlen); + if (len <= 0) + goto out; + rsii.minor_status = simple_strtoul(buf, &ep, 10); + if (*ep) goto out; - } else { - rsii.major_status = simple_strtoul(buf, &ep, 10); - if (*ep) - goto out; - len = qword_get(&mesg, buf, mlen); - if (len <= 0) - goto out; - rsii.minor_status = simple_strtoul(buf, &ep, 10); - if (*ep) - goto out; - /* out_handle */ - len = qword_get(&mesg, buf, mlen); - if (len < 0) - goto out; - status = -ENOMEM; - if (dup_to_netobj(&rsii.out_handle, buf, len)) - goto out; + /* out_handle */ + len = qword_get(&mesg, buf, mlen); + if (len < 0) + goto out; + status = -ENOMEM; + if (dup_to_netobj(&rsii.out_handle, buf, len)) + goto out; - /* out_token */ - len = qword_get(&mesg, buf, mlen); - status = -EINVAL; - if (len < 0) - goto out; - status = -ENOMEM; - if (dup_to_netobj(&rsii.out_token, buf, len)) - goto out; - } + /* out_token */ + len = qword_get(&mesg, buf, mlen); + status = -EINVAL; + if (len < 0) + goto out; + status = -ENOMEM; + if (dup_to_netobj(&rsii.out_token, buf, len)) + goto out; rsii.h.expiry_time = expiry; - rsip = rsi_lookup(&rsii, 1); + rsip = rsi_update(cd, &rsii, rsip); status = 0; out: rsi_free(&rsii); if (rsip) - rsi_put(&rsip->h, &rsi_cache); + cache_put(&rsip->h, cd); + else + status = -ENOMEM; return status; } -static struct cache_detail rsi_cache = { +static struct cache_detail rsi_cache_template = { .owner = THIS_MODULE, .hash_size = RSI_HASHMAX, - .hash_table = rsi_table, .name = "auth.rpcsec.init", .cache_put = rsi_put, .cache_request = rsi_request, .cache_parse = rsi_parse, + .match = rsi_match, + .init = rsi_init, + .update = update_rsi, + .alloc = rsi_alloc, }; -static DefineSimpleCacheLookup(rsi, 0) +static struct rsi *rsi_lookup(struct cache_detail *cd, struct rsi *item) +{ + struct cache_head *ch; + int hash = rsi_hash(item); + + ch = sunrpc_cache_lookup(cd, &item->h, hash); + if (ch) + return container_of(ch, struct rsi, h); + else + return NULL; +} + +static struct rsi *rsi_update(struct cache_detail *cd, struct rsi *new, struct rsi *old) +{ + struct cache_head *ch; + int hash = rsi_hash(new); + + ch = sunrpc_cache_update(cd, &new->h, + &old->h, hash); + if (ch) + return container_of(ch, struct rsi, h); + else + return NULL; +} + /* * The rpcsec_context cache is used to store a context that is @@ -270,7 +312,6 @@ static DefineSimpleCacheLookup(rsi, 0) #define RSC_HASHBITS 10 #define RSC_HASHMAX (1<<RSC_HASHBITS) -#define RSC_HASHMASK (RSC_HASHMAX-1) #define GSS_SEQ_WIN 128 @@ -291,27 +332,23 @@ struct rsc { struct gss_ctx *mechctx; }; -static struct cache_head *rsc_table[RSC_HASHMAX]; -static struct cache_detail rsc_cache; -static struct rsc *rsc_lookup(struct rsc *item, int set); +static struct rsc *rsc_update(struct cache_detail *cd, struct rsc *new, struct rsc *old); +static struct rsc *rsc_lookup(struct cache_detail *cd, struct rsc *item); static void rsc_free(struct rsc *rsci) { kfree(rsci->handle.data); if (rsci->mechctx) gss_delete_sec_context(&rsci->mechctx); - if (rsci->cred.cr_group_info) - put_group_info(rsci->cred.cr_group_info); + free_svc_cred(&rsci->cred); } -static void rsc_put(struct cache_head *item, struct cache_detail *cd) +static void rsc_put(struct kref *ref) { - struct rsc *rsci = container_of(item, struct rsc, h); + struct rsc *rsci = container_of(ref, struct rsc, h.ref); - if (cache_put(item, cd)) { - rsc_free(rsci); - kfree(rsci); - } + rsc_free(rsci); + kfree(rsci); } static inline int @@ -320,32 +357,51 @@ rsc_hash(struct rsc *rsci) return hash_mem(rsci->handle.data, rsci->handle.len, RSC_HASHBITS); } -static inline int -rsc_match(struct rsc *new, struct rsc *tmp) +static int +rsc_match(struct cache_head *a, struct cache_head *b) { + struct rsc *new = container_of(a, struct rsc, h); + struct rsc *tmp = container_of(b, struct rsc, h); + return netobj_equal(&new->handle, &tmp->handle); } -static inline void -rsc_init(struct rsc *new, struct rsc *tmp) +static void +rsc_init(struct cache_head *cnew, struct cache_head *ctmp) { + struct rsc *new = container_of(cnew, struct rsc, h); + struct rsc *tmp = container_of(ctmp, struct rsc, h); + new->handle.len = tmp->handle.len; tmp->handle.len = 0; new->handle.data = tmp->handle.data; tmp->handle.data = NULL; new->mechctx = NULL; - new->cred.cr_group_info = NULL; + init_svc_cred(&new->cred); } -static inline void -rsc_update(struct rsc *new, struct rsc *tmp) +static void +update_rsc(struct cache_head *cnew, struct cache_head *ctmp) { + struct rsc *new = container_of(cnew, struct rsc, h); + struct rsc *tmp = container_of(ctmp, struct rsc, h); + new->mechctx = tmp->mechctx; tmp->mechctx = NULL; memset(&new->seqdata, 0, sizeof(new->seqdata)); spin_lock_init(&new->seqdata.sd_lock); new->cred = tmp->cred; - tmp->cred.cr_group_info = NULL; + init_svc_cred(&tmp->cred); +} + +static struct cache_head * +rsc_alloc(void) +{ + struct rsc *rsci = kmalloc(sizeof(*rsci), GFP_KERNEL); + if (rsci) + return &rsci->h; + else + return NULL; } static int rsc_parse(struct cache_detail *cd, @@ -353,10 +409,12 @@ static int rsc_parse(struct cache_detail *cd, { /* contexthandle expiry [ uid gid N <n gids> mechname ...mechdata... ] */ char *buf = mesg; + int id; int len, rv; struct rsc rsci, *rscp = NULL; time_t expiry; int status = -EINVAL; + struct gss_api_mech *gm = NULL; memset(&rsci, 0, sizeof(rsci)); /* context handle */ @@ -373,19 +431,34 @@ static int rsc_parse(struct cache_detail *cd, if (expiry == 0) goto out; + rscp = rsc_lookup(cd, &rsci); + if (!rscp) + goto out; + /* uid, or NEGATIVE */ - rv = get_int(&mesg, &rsci.cred.cr_uid); + rv = get_int(&mesg, &id); if (rv == -EINVAL) goto out; if (rv == -ENOENT) set_bit(CACHE_NEGATIVE, &rsci.h.flags); else { int N, i; - struct gss_api_mech *gm; + + /* + * NOTE: we skip uid_valid()/gid_valid() checks here: + * instead, * -1 id's are later mapped to the + * (export-specific) anonymous id by nfsd_setuser. + * + * (But supplementary gid's get no such special + * treatment so are checked for validity here.) + */ + /* uid */ + rsci.cred.cr_uid = make_kuid(&init_user_ns, id); /* gid */ - if (get_int(&mesg, &rsci.cred.cr_gid)) + if (get_int(&mesg, &id)) goto out; + rsci.cred.cr_gid = make_kgid(&init_user_ns, id); /* number of additional gid's */ if (get_int(&mesg, &N)) @@ -398,17 +471,20 @@ static int rsc_parse(struct cache_detail *cd, /* gid's */ status = -EINVAL; for (i=0; i<N; i++) { - gid_t gid; - if (get_int(&mesg, &gid)) + kgid_t kgid; + if (get_int(&mesg, &id)) goto out; - GROUP_AT(rsci.cred.cr_group_info, i) = gid; + kgid = make_kgid(&init_user_ns, id); + if (!gid_valid(kgid)) + goto out; + GROUP_AT(rsci.cred.cr_group_info, i) = kgid; } /* mech name */ len = qword_get(&mesg, buf, mlen); if (len < 0) goto out; - gm = gss_mech_get_by_name(buf); + gm = rsci.cred.cr_gss_mech = gss_mech_get_by_name(buf); status = -EOPNOTSUPP; if (!gm) goto out; @@ -416,39 +492,76 @@ static int rsc_parse(struct cache_detail *cd, status = -EINVAL; /* mech-specific data: */ len = qword_get(&mesg, buf, mlen); - if (len < 0) { - gss_mech_put(gm); + if (len < 0) goto out; - } - if (gss_import_sec_context(buf, len, gm, &rsci.mechctx)) { - gss_mech_put(gm); + status = gss_import_sec_context(buf, len, gm, &rsci.mechctx, + NULL, GFP_KERNEL); + if (status) goto out; + + /* get client name */ + len = qword_get(&mesg, buf, mlen); + if (len > 0) { + rsci.cred.cr_principal = kstrdup(buf, GFP_KERNEL); + if (!rsci.cred.cr_principal) { + status = -ENOMEM; + goto out; + } } - gss_mech_put(gm); + } rsci.h.expiry_time = expiry; - rscp = rsc_lookup(&rsci, 1); + rscp = rsc_update(cd, &rsci, rscp); status = 0; out: rsc_free(&rsci); if (rscp) - rsc_put(&rscp->h, &rsc_cache); + cache_put(&rscp->h, cd); + else + status = -ENOMEM; return status; } -static struct cache_detail rsc_cache = { +static struct cache_detail rsc_cache_template = { .owner = THIS_MODULE, .hash_size = RSC_HASHMAX, - .hash_table = rsc_table, .name = "auth.rpcsec.context", .cache_put = rsc_put, .cache_parse = rsc_parse, + .match = rsc_match, + .init = rsc_init, + .update = update_rsc, + .alloc = rsc_alloc, }; -static DefineSimpleCacheLookup(rsc, 0); +static struct rsc *rsc_lookup(struct cache_detail *cd, struct rsc *item) +{ + struct cache_head *ch; + int hash = rsc_hash(item); + + ch = sunrpc_cache_lookup(cd, &item->h, hash); + if (ch) + return container_of(ch, struct rsc, h); + else + return NULL; +} + +static struct rsc *rsc_update(struct cache_detail *cd, struct rsc *new, struct rsc *old) +{ + struct cache_head *ch; + int hash = rsc_hash(new); + + ch = sunrpc_cache_update(cd, &new->h, + &old->h, hash); + if (ch) + return container_of(ch, struct rsc, h); + else + return NULL; +} + static struct rsc * -gss_svc_searchbyctx(struct xdr_netobj *handle) +gss_svc_searchbyctx(struct cache_detail *cd, struct xdr_netobj *handle) { struct rsc rsci; struct rsc *found; @@ -456,11 +569,11 @@ gss_svc_searchbyctx(struct xdr_netobj *handle) memset(&rsci, 0, sizeof(rsci)); if (dup_to_netobj(&rsci.handle, handle->data, handle->len)) return NULL; - found = rsc_lookup(&rsci, 0); + found = rsc_lookup(cd, &rsci); rsc_free(&rsci); if (!found) return NULL; - if (cache_check(&rsc_cache, &found->h, NULL)) + if (cache_check(cd, &found->h, NULL)) return NULL; return found; } @@ -508,7 +621,7 @@ svc_safe_getnetobj(struct kvec *argv, struct xdr_netobj *o) if (argv->iov_len < 4) return -1; - o->len = ntohl(svc_getu32(argv)); + o->len = svc_getnl(argv); l = round_up_to_quad(o->len); if (argv->iov_len < l) return -1; @@ -521,27 +634,28 @@ svc_safe_getnetobj(struct kvec *argv, struct xdr_netobj *o) static inline int svc_safe_putnetobj(struct kvec *resv, struct xdr_netobj *o) { - u32 *p; + u8 *p; if (resv->iov_len + 4 > PAGE_SIZE) return -1; - svc_putu32(resv, htonl(o->len)); + svc_putnl(resv, o->len); p = resv->iov_base + resv->iov_len; resv->iov_len += round_up_to_quad(o->len); if (resv->iov_len > PAGE_SIZE) return -1; memcpy(p, o->data, o->len); - memset((u8 *)p + o->len, 0, round_up_to_quad(o->len) - o->len); + memset(p + o->len, 0, round_up_to_quad(o->len) - o->len); return 0; } -/* Verify the checksum on the header and return SVC_OK on success. +/* + * Verify the checksum on the header and return SVC_OK on success. * Otherwise, return SVC_DROP (in the case of a bad sequence number) * or return SVC_DENIED and indicate error in authp. */ static int gss_verify_header(struct svc_rqst *rqstp, struct rsc *rsci, - u32 *rpcstart, struct rpc_gss_wire_cred *gc, u32 *authp) + __be32 *rpcstart, struct rpc_gss_wire_cred *gc, __be32 *authp) { struct gss_ctx *ctx_id = rsci->mechctx; struct xdr_buf rpchdr; @@ -558,7 +672,7 @@ gss_verify_header(struct svc_rqst *rqstp, struct rsc *rsci, *authp = rpc_autherr_badverf; if (argv->iov_len < 4) return SVC_DENIED; - flavor = ntohl(svc_getu32(argv)); + flavor = svc_getnl(argv); if (flavor != RPC_AUTH_GSS) return SVC_DENIED; if (svc_safe_getnetobj(argv, &checksum)) @@ -572,30 +686,44 @@ gss_verify_header(struct svc_rqst *rqstp, struct rsc *rsci, } if (gc->gc_seq > MAXSEQ) { - dprintk("RPC: svcauth_gss: discarding request with large sequence number %d\n", - gc->gc_seq); + dprintk("RPC: svcauth_gss: discarding request with " + "large sequence number %d\n", gc->gc_seq); *authp = rpcsec_gsserr_ctxproblem; return SVC_DENIED; } if (!gss_check_seq_num(rsci, gc->gc_seq)) { - dprintk("RPC: svcauth_gss: discarding request with old sequence number %d\n", - gc->gc_seq); + dprintk("RPC: svcauth_gss: discarding request with " + "old sequence number %d\n", gc->gc_seq); return SVC_DROP; } return SVC_OK; } static int +gss_write_null_verf(struct svc_rqst *rqstp) +{ + __be32 *p; + + svc_putnl(rqstp->rq_res.head, RPC_AUTH_NULL); + p = rqstp->rq_res.head->iov_base + rqstp->rq_res.head->iov_len; + /* don't really need to check if head->iov_len > PAGE_SIZE ... */ + *p++ = 0; + if (!xdr_ressize_check(rqstp, p)) + return -1; + return 0; +} + +static int gss_write_verf(struct svc_rqst *rqstp, struct gss_ctx *ctx_id, u32 seq) { - u32 xdr_seq; + __be32 xdr_seq; u32 maj_stat; struct xdr_buf verf_data; struct xdr_netobj mic; - u32 *p; + __be32 *p; struct kvec iov; - svc_putu32(rqstp->rq_res.head, htonl(RPC_AUTH_GSS)); + svc_putnl(rqstp->rq_res.head, RPC_AUTH_GSS); xdr_seq = htonl(seq); iov.iov_base = &xdr_seq; @@ -630,6 +758,17 @@ find_gss_auth_domain(struct gss_ctx *ctx, u32 svc) return auth_domain_find(name); } +static struct auth_ops svcauthops_gss; + +u32 svcauth_gss_flavor(struct auth_domain *dom) +{ + struct gss_domain *gd = container_of(dom, struct gss_domain, h); + + return gd->pseudoflavor; +} + +EXPORT_SYMBOL_GPL(svcauth_gss_flavor); + int svcauth_gss_register_pseudoflavor(u32 pseudoflavor, char * name) { @@ -640,21 +779,19 @@ svcauth_gss_register_pseudoflavor(u32 pseudoflavor, char * name) new = kmalloc(sizeof(*new), GFP_KERNEL); if (!new) goto out; - cache_init(&new->h.h); - new->h.name = kmalloc(strlen(name) + 1, GFP_KERNEL); + kref_init(&new->h.ref); + new->h.name = kstrdup(name, GFP_KERNEL); if (!new->h.name) goto out_free_dom; - strcpy(new->h.name, name); - new->h.flavour = RPC_AUTH_GSS; + new->h.flavour = &svcauthops_gss; new->pseudoflavor = pseudoflavor; - new->h.h.expiry_time = NEVER; - test = auth_domain_lookup(&new->h, 1); - if (test == &new->h) { - BUG_ON(atomic_dec_and_test(&new->h.h.refcnt)); - } else { /* XXX Duplicate registration? */ - auth_domain_put(&new->h); - goto out; + stat = 0; + test = auth_domain_lookup(name, &new->h); + if (test != &new->h) { /* Duplicate registration */ + auth_domain_put(test); + kfree(new->h.name); + goto out_free_dom; } return 0; @@ -664,12 +801,12 @@ out: return stat; } -EXPORT_SYMBOL(svcauth_gss_register_pseudoflavor); +EXPORT_SYMBOL_GPL(svcauth_gss_register_pseudoflavor); static inline int read_u32_from_xdr_buf(struct xdr_buf *buf, int base, u32 *obj) { - u32 raw; + __be32 raw; int status; status = read_bytes_from_xdr_buf(buf, base, &raw, sizeof(*obj)); @@ -685,46 +822,121 @@ read_u32_from_xdr_buf(struct xdr_buf *buf, int base, u32 *obj) * The server uses base of head iovec as read pointer, while the * client uses separate pointer. */ static int -unwrap_integ_data(struct xdr_buf *buf, u32 seq, struct gss_ctx *ctx) +unwrap_integ_data(struct svc_rqst *rqstp, struct xdr_buf *buf, u32 seq, struct gss_ctx *ctx) { int stat = -EINVAL; u32 integ_len, maj_stat; struct xdr_netobj mic; struct xdr_buf integ_buf; - integ_len = ntohl(svc_getu32(&buf->head[0])); + /* Did we already verify the signature on the original pass through? */ + if (rqstp->rq_deferred) + return 0; + + integ_len = svc_getnl(&buf->head[0]); if (integ_len & 3) - goto out; + return stat; if (integ_len > buf->len) - goto out; + return stat; if (xdr_buf_subsegment(buf, &integ_buf, 0, integ_len)) BUG(); /* copy out mic... */ if (read_u32_from_xdr_buf(buf, integ_len, &mic.len)) BUG(); if (mic.len > RPC_MAX_AUTH_SIZE) - goto out; + return stat; mic.data = kmalloc(mic.len, GFP_KERNEL); if (!mic.data) - goto out; + return stat; if (read_bytes_from_xdr_buf(buf, integ_len + 4, mic.data, mic.len)) goto out; maj_stat = gss_verify_mic(ctx, &integ_buf, &mic); if (maj_stat != GSS_S_COMPLETE) goto out; - if (ntohl(svc_getu32(&buf->head[0])) != seq) + if (svc_getnl(&buf->head[0]) != seq) goto out; + /* trim off the mic at the end before returning */ + xdr_buf_trim(buf, mic.len + 4); stat = 0; out: + kfree(mic.data); return stat; } +static inline int +total_buf_len(struct xdr_buf *buf) +{ + return buf->head[0].iov_len + buf->page_len + buf->tail[0].iov_len; +} + +static void +fix_priv_head(struct xdr_buf *buf, int pad) +{ + if (buf->page_len == 0) { + /* We need to adjust head and buf->len in tandem in this + * case to make svc_defer() work--it finds the original + * buffer start using buf->len - buf->head[0].iov_len. */ + buf->head[0].iov_len -= pad; + } +} + +static int +unwrap_priv_data(struct svc_rqst *rqstp, struct xdr_buf *buf, u32 seq, struct gss_ctx *ctx) +{ + u32 priv_len, maj_stat; + int pad, saved_len, remaining_len, offset; + + rqstp->rq_splice_ok = 0; + + priv_len = svc_getnl(&buf->head[0]); + if (rqstp->rq_deferred) { + /* Already decrypted last time through! The sequence number + * check at out_seq is unnecessary but harmless: */ + goto out_seq; + } + /* buf->len is the number of bytes from the original start of the + * request to the end, where head[0].iov_len is just the bytes + * not yet read from the head, so these two values are different: */ + remaining_len = total_buf_len(buf); + if (priv_len > remaining_len) + return -EINVAL; + pad = remaining_len - priv_len; + buf->len -= pad; + fix_priv_head(buf, pad); + + /* Maybe it would be better to give gss_unwrap a length parameter: */ + saved_len = buf->len; + buf->len = priv_len; + maj_stat = gss_unwrap(ctx, 0, buf); + pad = priv_len - buf->len; + buf->len = saved_len; + buf->len -= pad; + /* The upper layers assume the buffer is aligned on 4-byte boundaries. + * In the krb5p case, at least, the data ends up offset, so we need to + * move it around. */ + /* XXX: This is very inefficient. It would be better to either do + * this while we encrypt, or maybe in the receive code, if we can peak + * ahead and work out the service and mechanism there. */ + offset = buf->head[0].iov_len % 4; + if (offset) { + buf->buflen = RPCSVC_MAXPAYLOAD; + xdr_shift_buf(buf, offset); + fix_priv_head(buf, pad); + } + if (maj_stat != GSS_S_COMPLETE) + return -EINVAL; +out_seq: + if (svc_getnl(&buf->head[0]) != seq) + return -EINVAL; + return 0; +} + struct gss_svc_data { /* decoded gss client cred: */ struct rpc_gss_wire_cred clcred; - /* pointer to the beginning of the procedure-specific results, - * which may be encrypted/checksummed in svcauth_gss_release: */ - u32 *body_start; + /* save a pointer to the beginning of the encoded verifier, + * for use in encryption/checksumming in svcauth_gss_release: */ + __be32 *verf_start; struct rsc *rsci; }; @@ -734,13 +946,443 @@ svcauth_gss_set_client(struct svc_rqst *rqstp) struct gss_svc_data *svcdata = rqstp->rq_auth_data; struct rsc *rsci = svcdata->rsci; struct rpc_gss_wire_cred *gc = &svcdata->clcred; + int stat; - rqstp->rq_client = find_gss_auth_domain(rsci->mechctx, gc->gc_svc); - if (rqstp->rq_client == NULL) + /* + * A gss export can be specified either by: + * export *(sec=krb5,rw) + * or by + * export gss/krb5(rw) + * The latter is deprecated; but for backwards compatibility reasons + * the nfsd code will still fall back on trying it if the former + * doesn't work; so we try to make both available to nfsd, below. + */ + rqstp->rq_gssclient = find_gss_auth_domain(rsci->mechctx, gc->gc_svc); + if (rqstp->rq_gssclient == NULL) return SVC_DENIED; + stat = svcauth_unix_set_client(rqstp); + if (stat == SVC_DROP || stat == SVC_CLOSE) + return stat; return SVC_OK; } +static inline int +gss_write_init_verf(struct cache_detail *cd, struct svc_rqst *rqstp, + struct xdr_netobj *out_handle, int *major_status) +{ + struct rsc *rsci; + int rc; + + if (*major_status != GSS_S_COMPLETE) + return gss_write_null_verf(rqstp); + rsci = gss_svc_searchbyctx(cd, out_handle); + if (rsci == NULL) { + *major_status = GSS_S_NO_CONTEXT; + return gss_write_null_verf(rqstp); + } + rc = gss_write_verf(rqstp, rsci->mechctx, GSS_SEQ_WIN); + cache_put(&rsci->h, cd); + return rc; +} + +static inline int +gss_read_common_verf(struct rpc_gss_wire_cred *gc, + struct kvec *argv, __be32 *authp, + struct xdr_netobj *in_handle) +{ + /* Read the verifier; should be NULL: */ + *authp = rpc_autherr_badverf; + if (argv->iov_len < 2 * 4) + return SVC_DENIED; + if (svc_getnl(argv) != RPC_AUTH_NULL) + return SVC_DENIED; + if (svc_getnl(argv) != 0) + return SVC_DENIED; + /* Martial context handle and token for upcall: */ + *authp = rpc_autherr_badcred; + if (gc->gc_proc == RPC_GSS_PROC_INIT && gc->gc_ctx.len != 0) + return SVC_DENIED; + if (dup_netobj(in_handle, &gc->gc_ctx)) + return SVC_CLOSE; + *authp = rpc_autherr_badverf; + + return 0; +} + +static inline int +gss_read_verf(struct rpc_gss_wire_cred *gc, + struct kvec *argv, __be32 *authp, + struct xdr_netobj *in_handle, + struct xdr_netobj *in_token) +{ + struct xdr_netobj tmpobj; + int res; + + res = gss_read_common_verf(gc, argv, authp, in_handle); + if (res) + return res; + + if (svc_safe_getnetobj(argv, &tmpobj)) { + kfree(in_handle->data); + return SVC_DENIED; + } + if (dup_netobj(in_token, &tmpobj)) { + kfree(in_handle->data); + return SVC_CLOSE; + } + + return 0; +} + +/* Ok this is really heavily depending on a set of semantics in + * how rqstp is set up by svc_recv and pages laid down by the + * server when reading a request. We are basically guaranteed that + * the token lays all down linearly across a set of pages, starting + * at iov_base in rq_arg.head[0] which happens to be the first of a + * set of pages stored in rq_pages[]. + * rq_arg.head[0].iov_base will provide us the page_base to pass + * to the upcall. + */ +static inline int +gss_read_proxy_verf(struct svc_rqst *rqstp, + struct rpc_gss_wire_cred *gc, __be32 *authp, + struct xdr_netobj *in_handle, + struct gssp_in_token *in_token) +{ + struct kvec *argv = &rqstp->rq_arg.head[0]; + u32 inlen; + int res; + + res = gss_read_common_verf(gc, argv, authp, in_handle); + if (res) + return res; + + inlen = svc_getnl(argv); + if (inlen > (argv->iov_len + rqstp->rq_arg.page_len)) + return SVC_DENIED; + + in_token->pages = rqstp->rq_pages; + in_token->page_base = (ulong)argv->iov_base & ~PAGE_MASK; + in_token->page_len = inlen; + + return 0; +} + +static inline int +gss_write_resv(struct kvec *resv, size_t size_limit, + struct xdr_netobj *out_handle, struct xdr_netobj *out_token, + int major_status, int minor_status) +{ + if (resv->iov_len + 4 > size_limit) + return -1; + svc_putnl(resv, RPC_SUCCESS); + if (svc_safe_putnetobj(resv, out_handle)) + return -1; + if (resv->iov_len + 3 * 4 > size_limit) + return -1; + svc_putnl(resv, major_status); + svc_putnl(resv, minor_status); + svc_putnl(resv, GSS_SEQ_WIN); + if (svc_safe_putnetobj(resv, out_token)) + return -1; + return 0; +} + +/* + * Having read the cred already and found we're in the context + * initiation case, read the verifier and initiate (or check the results + * of) upcalls to userspace for help with context initiation. If + * the upcall results are available, write the verifier and result. + * Otherwise, drop the request pending an answer to the upcall. + */ +static int svcauth_gss_legacy_init(struct svc_rqst *rqstp, + struct rpc_gss_wire_cred *gc, __be32 *authp) +{ + struct kvec *argv = &rqstp->rq_arg.head[0]; + struct kvec *resv = &rqstp->rq_res.head[0]; + struct rsi *rsip, rsikey; + int ret; + struct sunrpc_net *sn = net_generic(rqstp->rq_xprt->xpt_net, sunrpc_net_id); + + memset(&rsikey, 0, sizeof(rsikey)); + ret = gss_read_verf(gc, argv, authp, + &rsikey.in_handle, &rsikey.in_token); + if (ret) + return ret; + + /* Perform upcall, or find upcall result: */ + rsip = rsi_lookup(sn->rsi_cache, &rsikey); + rsi_free(&rsikey); + if (!rsip) + return SVC_CLOSE; + if (cache_check(sn->rsi_cache, &rsip->h, &rqstp->rq_chandle) < 0) + /* No upcall result: */ + return SVC_CLOSE; + + ret = SVC_CLOSE; + /* Got an answer to the upcall; use it: */ + if (gss_write_init_verf(sn->rsc_cache, rqstp, + &rsip->out_handle, &rsip->major_status)) + goto out; + if (gss_write_resv(resv, PAGE_SIZE, + &rsip->out_handle, &rsip->out_token, + rsip->major_status, rsip->minor_status)) + goto out; + + ret = SVC_COMPLETE; +out: + cache_put(&rsip->h, sn->rsi_cache); + return ret; +} + +static int gss_proxy_save_rsc(struct cache_detail *cd, + struct gssp_upcall_data *ud, + uint64_t *handle) +{ + struct rsc rsci, *rscp = NULL; + static atomic64_t ctxhctr; + long long ctxh; + struct gss_api_mech *gm = NULL; + time_t expiry; + int status = -EINVAL; + + memset(&rsci, 0, sizeof(rsci)); + /* context handle */ + status = -ENOMEM; + /* the handle needs to be just a unique id, + * use a static counter */ + ctxh = atomic64_inc_return(&ctxhctr); + + /* make a copy for the caller */ + *handle = ctxh; + + /* make a copy for the rsc cache */ + if (dup_to_netobj(&rsci.handle, (char *)handle, sizeof(uint64_t))) + goto out; + rscp = rsc_lookup(cd, &rsci); + if (!rscp) + goto out; + + /* creds */ + if (!ud->found_creds) { + /* userspace seem buggy, we should always get at least a + * mapping to nobody */ + dprintk("RPC: No creds found!\n"); + goto out; + } else { + + /* steal creds */ + rsci.cred = ud->creds; + memset(&ud->creds, 0, sizeof(struct svc_cred)); + + status = -EOPNOTSUPP; + /* get mech handle from OID */ + gm = gss_mech_get_by_OID(&ud->mech_oid); + if (!gm) + goto out; + rsci.cred.cr_gss_mech = gm; + + status = -EINVAL; + /* mech-specific data: */ + status = gss_import_sec_context(ud->out_handle.data, + ud->out_handle.len, + gm, &rsci.mechctx, + &expiry, GFP_KERNEL); + if (status) + goto out; + } + + rsci.h.expiry_time = expiry; + rscp = rsc_update(cd, &rsci, rscp); + status = 0; +out: + rsc_free(&rsci); + if (rscp) + cache_put(&rscp->h, cd); + else + status = -ENOMEM; + return status; +} + +static int svcauth_gss_proxy_init(struct svc_rqst *rqstp, + struct rpc_gss_wire_cred *gc, __be32 *authp) +{ + struct kvec *resv = &rqstp->rq_res.head[0]; + struct xdr_netobj cli_handle; + struct gssp_upcall_data ud; + uint64_t handle; + int status; + int ret; + struct net *net = rqstp->rq_xprt->xpt_net; + struct sunrpc_net *sn = net_generic(net, sunrpc_net_id); + + memset(&ud, 0, sizeof(ud)); + ret = gss_read_proxy_verf(rqstp, gc, authp, + &ud.in_handle, &ud.in_token); + if (ret) + return ret; + + ret = SVC_CLOSE; + + /* Perform synchronous upcall to gss-proxy */ + status = gssp_accept_sec_context_upcall(net, &ud); + if (status) + goto out; + + dprintk("RPC: svcauth_gss: gss major status = %d\n", + ud.major_status); + + switch (ud.major_status) { + case GSS_S_CONTINUE_NEEDED: + cli_handle = ud.out_handle; + break; + case GSS_S_COMPLETE: + status = gss_proxy_save_rsc(sn->rsc_cache, &ud, &handle); + if (status) + goto out; + cli_handle.data = (u8 *)&handle; + cli_handle.len = sizeof(handle); + break; + default: + ret = SVC_CLOSE; + goto out; + } + + /* Got an answer to the upcall; use it: */ + if (gss_write_init_verf(sn->rsc_cache, rqstp, + &cli_handle, &ud.major_status)) + goto out; + if (gss_write_resv(resv, PAGE_SIZE, + &cli_handle, &ud.out_token, + ud.major_status, ud.minor_status)) + goto out; + + ret = SVC_COMPLETE; +out: + gssp_free_upcall_data(&ud); + return ret; +} + +/* + * Try to set the sn->use_gss_proxy variable to a new value. We only allow + * it to be changed if it's currently undefined (-1). If it's any other value + * then return -EBUSY unless the type wouldn't have changed anyway. + */ +static int set_gss_proxy(struct net *net, int type) +{ + struct sunrpc_net *sn = net_generic(net, sunrpc_net_id); + int ret; + + WARN_ON_ONCE(type != 0 && type != 1); + ret = cmpxchg(&sn->use_gss_proxy, -1, type); + if (ret != -1 && ret != type) + return -EBUSY; + return 0; +} + +static bool use_gss_proxy(struct net *net) +{ + struct sunrpc_net *sn = net_generic(net, sunrpc_net_id); + + /* If use_gss_proxy is still undefined, then try to disable it */ + if (sn->use_gss_proxy == -1) + set_gss_proxy(net, 0); + return sn->use_gss_proxy; +} + +#ifdef CONFIG_PROC_FS + +static ssize_t write_gssp(struct file *file, const char __user *buf, + size_t count, loff_t *ppos) +{ + struct net *net = PDE_DATA(file_inode(file)); + char tbuf[20]; + unsigned long i; + int res; + + if (*ppos || count > sizeof(tbuf)-1) + return -EINVAL; + if (copy_from_user(tbuf, buf, count)) + return -EFAULT; + + tbuf[count] = 0; + res = kstrtoul(tbuf, 0, &i); + if (res) + return res; + if (i != 1) + return -EINVAL; + res = set_gssp_clnt(net); + if (res) + return res; + res = set_gss_proxy(net, 1); + if (res) + return res; + return count; +} + +static ssize_t read_gssp(struct file *file, char __user *buf, + size_t count, loff_t *ppos) +{ + struct net *net = PDE_DATA(file_inode(file)); + struct sunrpc_net *sn = net_generic(net, sunrpc_net_id); + unsigned long p = *ppos; + char tbuf[10]; + size_t len; + + snprintf(tbuf, sizeof(tbuf), "%d\n", sn->use_gss_proxy); + len = strlen(tbuf); + if (p >= len) + return 0; + len -= p; + if (len > count) + len = count; + if (copy_to_user(buf, (void *)(tbuf+p), len)) + return -EFAULT; + *ppos += len; + return len; +} + +static const struct file_operations use_gss_proxy_ops = { + .open = nonseekable_open, + .write = write_gssp, + .read = read_gssp, +}; + +static int create_use_gss_proxy_proc_entry(struct net *net) +{ + struct sunrpc_net *sn = net_generic(net, sunrpc_net_id); + struct proc_dir_entry **p = &sn->use_gssp_proc; + + sn->use_gss_proxy = -1; + *p = proc_create_data("use-gss-proxy", S_IFREG|S_IRUSR|S_IWUSR, + sn->proc_net_rpc, + &use_gss_proxy_ops, net); + if (!*p) + return -ENOMEM; + init_gssp_clnt(sn); + return 0; +} + +static void destroy_use_gss_proxy_proc_entry(struct net *net) +{ + struct sunrpc_net *sn = net_generic(net, sunrpc_net_id); + + if (sn->use_gssp_proc) { + remove_proc_entry("use-gss-proxy", sn->proc_net_rpc); + clear_gssp_clnt(sn); + } +} +#else /* CONFIG_PROC_FS */ + +static int create_use_gss_proxy_proc_entry(struct net *net) +{ + return 0; +} + +static void destroy_use_gss_proxy_proc_entry(struct net *net) {} + +#endif /* CONFIG_PROC_FS */ + /* * Accept an rpcsec packet. * If context establishment, punt to user space @@ -750,21 +1392,21 @@ svcauth_gss_set_client(struct svc_rqst *rqstp) * response here and return SVC_COMPLETE. */ static int -svcauth_gss_accept(struct svc_rqst *rqstp, u32 *authp) +svcauth_gss_accept(struct svc_rqst *rqstp, __be32 *authp) { struct kvec *argv = &rqstp->rq_arg.head[0]; struct kvec *resv = &rqstp->rq_res.head[0]; u32 crlen; - struct xdr_netobj tmpobj; struct gss_svc_data *svcdata = rqstp->rq_auth_data; struct rpc_gss_wire_cred *gc; struct rsc *rsci = NULL; - struct rsi *rsip, rsikey; - u32 *rpcstart; - u32 *reject_stat = resv->iov_base + resv->iov_len; + __be32 *rpcstart; + __be32 *reject_stat = resv->iov_base + resv->iov_len; int ret; + struct sunrpc_net *sn = net_generic(rqstp->rq_xprt->xpt_net, sunrpc_net_id); - dprintk("RPC: svcauth_gss: argv->iov_len = %zd\n",argv->iov_len); + dprintk("RPC: svcauth_gss: argv->iov_len = %zd\n", + argv->iov_len); *authp = rpc_autherr_badcred; if (!svcdata) @@ -772,7 +1414,7 @@ svcauth_gss_accept(struct svc_rqst *rqstp, u32 *authp) if (!svcdata) goto auth_err; rqstp->rq_auth_data = svcdata; - svcdata->body_start = NULL; + svcdata->verf_start = NULL; svcdata->rsci = NULL; gc = &svcdata->clcred; @@ -784,17 +1426,17 @@ svcauth_gss_accept(struct svc_rqst *rqstp, u32 *authp) /* credential is: * version(==1), proc(0,1,2,3), seq, service (1,2,3), handle - * at least 5 u32s, and is preceeded by length, so that makes 6. + * at least 5 u32s, and is preceded by length, so that makes 6. */ if (argv->iov_len < 5 * 4) goto auth_err; - crlen = ntohl(svc_getu32(argv)); - if (ntohl(svc_getu32(argv)) != RPC_GSS_VERSION) + crlen = svc_getnl(argv); + if (svc_getnl(argv) != RPC_GSS_VERSION) goto auth_err; - gc->gc_proc = ntohl(svc_getu32(argv)); - gc->gc_seq = ntohl(svc_getu32(argv)); - gc->gc_svc = ntohl(svc_getu32(argv)); + gc->gc_proc = svc_getnl(argv); + gc->gc_seq = svc_getnl(argv); + gc->gc_svc = svc_getnl(argv); if (svc_safe_getnetobj(argv, &gc->gc_ctx)) goto auth_err; if (crlen != round_up_to_quad(gc->gc_ctx.len) + 5 * 4) @@ -803,32 +1445,19 @@ svcauth_gss_accept(struct svc_rqst *rqstp, u32 *authp) if ((gc->gc_proc != RPC_GSS_PROC_DATA) && (rqstp->rq_proc != 0)) goto auth_err; - /* - * We've successfully parsed the credential. Let's check out the - * verifier. An AUTH_NULL verifier is allowed (and required) for - * INIT and CONTINUE_INIT requests. AUTH_RPCSEC_GSS is required for - * PROC_DATA and PROC_DESTROY. - * - * AUTH_NULL verifier is 0 (AUTH_NULL), 0 (length). - * AUTH_RPCSEC_GSS verifier is: - * 6 (AUTH_RPCSEC_GSS), length, checksum. - * checksum is calculated over rpcheader from xid up to here. - */ *authp = rpc_autherr_badverf; switch (gc->gc_proc) { case RPC_GSS_PROC_INIT: case RPC_GSS_PROC_CONTINUE_INIT: - if (argv->iov_len < 2 * 4) - goto auth_err; - if (ntohl(svc_getu32(argv)) != RPC_AUTH_NULL) - goto auth_err; - if (ntohl(svc_getu32(argv)) != 0) - goto auth_err; - break; + if (use_gss_proxy(SVC_NET(rqstp))) + return svcauth_gss_proxy_init(rqstp, gc, authp); + else + return svcauth_gss_legacy_init(rqstp, gc, authp); case RPC_GSS_PROC_DATA: case RPC_GSS_PROC_DESTROY: + /* Look up the context, and check the verifier: */ *authp = rpcsec_gsserr_credproblem; - rsci = gss_svc_searchbyctx(&gc->gc_ctx); + rsci = gss_svc_searchbyctx(sn->rsc_cache, &gc->gc_ctx); if (!rsci) goto auth_err; switch (gss_verify_header(rqstp, rsci, rpcstart, gc, authp)) { @@ -847,64 +1476,18 @@ svcauth_gss_accept(struct svc_rqst *rqstp, u32 *authp) /* now act upon the command: */ switch (gc->gc_proc) { - case RPC_GSS_PROC_INIT: - case RPC_GSS_PROC_CONTINUE_INIT: - *authp = rpc_autherr_badcred; - if (gc->gc_proc == RPC_GSS_PROC_INIT && gc->gc_ctx.len != 0) - goto auth_err; - memset(&rsikey, 0, sizeof(rsikey)); - if (dup_netobj(&rsikey.in_handle, &gc->gc_ctx)) - goto drop; - *authp = rpc_autherr_badverf; - if (svc_safe_getnetobj(argv, &tmpobj)) { - kfree(rsikey.in_handle.data); - goto auth_err; - } - if (dup_netobj(&rsikey.in_token, &tmpobj)) { - kfree(rsikey.in_handle.data); - goto drop; - } - - rsip = rsi_lookup(&rsikey, 0); - rsi_free(&rsikey); - if (!rsip) { - goto drop; - } - switch(cache_check(&rsi_cache, &rsip->h, &rqstp->rq_chandle)) { - case -EAGAIN: - goto drop; - case -ENOENT: - goto drop; - case 0: - rsci = gss_svc_searchbyctx(&rsip->out_handle); - if (!rsci) { - goto drop; - } - if (gss_write_verf(rqstp, rsci->mechctx, GSS_SEQ_WIN)) - goto drop; - if (resv->iov_len + 4 > PAGE_SIZE) - goto drop; - svc_putu32(resv, rpc_success); - if (svc_safe_putnetobj(resv, &rsip->out_handle)) - goto drop; - if (resv->iov_len + 3 * 4 > PAGE_SIZE) - goto drop; - svc_putu32(resv, htonl(rsip->major_status)); - svc_putu32(resv, htonl(rsip->minor_status)); - svc_putu32(resv, htonl(GSS_SEQ_WIN)); - if (svc_safe_putnetobj(resv, &rsip->out_token)) - goto drop; - rqstp->rq_client = NULL; - } - goto complete; case RPC_GSS_PROC_DESTROY: + if (gss_write_verf(rqstp, rsci->mechctx, gc->gc_seq)) + goto auth_err; + rsci->h.expiry_time = get_seconds(); set_bit(CACHE_NEGATIVE, &rsci->h.flags); if (resv->iov_len + 4 > PAGE_SIZE) goto drop; - svc_putu32(resv, rpc_success); + svc_putnl(resv, RPC_SUCCESS); goto complete; case RPC_GSS_PROC_DATA: *authp = rpcsec_gsserr_ctxproblem; + svcdata->verf_start = resv->iov_base + resv->iov_len; if (gss_write_verf(rqstp, rsci->mechctx, gc->gc_seq)) goto auth_err; rqstp->rq_cred = rsci->cred; @@ -914,26 +1497,40 @@ svcauth_gss_accept(struct svc_rqst *rqstp, u32 *authp) case RPC_GSS_SVC_NONE: break; case RPC_GSS_SVC_INTEGRITY: - if (unwrap_integ_data(&rqstp->rq_arg, - gc->gc_seq, rsci->mechctx)) - goto auth_err; /* placeholders for length and seq. number: */ - svcdata->body_start = resv->iov_base + resv->iov_len; - svc_putu32(resv, 0); - svc_putu32(resv, 0); + svc_putnl(resv, 0); + svc_putnl(resv, 0); + if (unwrap_integ_data(rqstp, &rqstp->rq_arg, + gc->gc_seq, rsci->mechctx)) + goto garbage_args; + rqstp->rq_auth_slack = RPC_MAX_AUTH_SIZE; break; case RPC_GSS_SVC_PRIVACY: - /* currently unsupported */ + /* placeholders for length and seq. number: */ + svc_putnl(resv, 0); + svc_putnl(resv, 0); + if (unwrap_priv_data(rqstp, &rqstp->rq_arg, + gc->gc_seq, rsci->mechctx)) + goto garbage_args; + rqstp->rq_auth_slack = RPC_MAX_AUTH_SIZE * 2; + break; default: goto auth_err; } svcdata->rsci = rsci; cache_get(&rsci->h); + rqstp->rq_cred.cr_flavor = gss_svc_to_pseudoflavor( + rsci->mechctx->mech_type, + GSS_C_QOP_DEFAULT, + gc->gc_svc); ret = SVC_OK; goto out; } +garbage_args: + ret = SVC_GARBAGE; + goto out; auth_err: - /* Restore write pointer to original value: */ + /* Restore write pointer to its original value: */ xdr_ressize_check(rqstp, reject_stat); ret = SVC_DENIED; goto out; @@ -944,12 +1541,39 @@ drop: ret = SVC_DROP; out: if (rsci) - rsc_put(&rsci->h, &rsc_cache); + cache_put(&rsci->h, sn->rsc_cache); return ret; } -static int -svcauth_gss_release(struct svc_rqst *rqstp) +static __be32 * +svcauth_gss_prepare_to_wrap(struct xdr_buf *resbuf, struct gss_svc_data *gsd) +{ + __be32 *p; + u32 verf_len; + + p = gsd->verf_start; + gsd->verf_start = NULL; + + /* If the reply stat is nonzero, don't wrap: */ + if (*(p-1) != rpc_success) + return NULL; + /* Skip the verifier: */ + p += 1; + verf_len = ntohl(*p++); + p += XDR_QUADLEN(verf_len); + /* move accept_stat to right place: */ + memcpy(p, p + 2, 4); + /* Also don't wrap if the accept stat is nonzero: */ + if (*p != rpc_success) { + resbuf->head[0].iov_len -= 2 * 4; + return NULL; + } + p++; + return p; +} + +static inline int +svcauth_gss_wrap_resp_integ(struct svc_rqst *rqstp) { struct gss_svc_data *gsd = (struct gss_svc_data *)rqstp->rq_auth_data; struct rpc_gss_wire_cred *gc = &gsd->clcred; @@ -957,73 +1581,144 @@ svcauth_gss_release(struct svc_rqst *rqstp) struct xdr_buf integ_buf; struct xdr_netobj mic; struct kvec *resv; - u32 *p; + __be32 *p; int integ_offset, integ_len; int stat = -EINVAL; + p = svcauth_gss_prepare_to_wrap(resbuf, gsd); + if (p == NULL) + goto out; + integ_offset = (u8 *)(p + 1) - (u8 *)resbuf->head[0].iov_base; + integ_len = resbuf->len - integ_offset; + BUG_ON(integ_len % 4); + *p++ = htonl(integ_len); + *p++ = htonl(gc->gc_seq); + if (xdr_buf_subsegment(resbuf, &integ_buf, integ_offset, integ_len)) + BUG(); + if (resbuf->tail[0].iov_base == NULL) { + if (resbuf->head[0].iov_len + RPC_MAX_AUTH_SIZE > PAGE_SIZE) + goto out_err; + resbuf->tail[0].iov_base = resbuf->head[0].iov_base + + resbuf->head[0].iov_len; + resbuf->tail[0].iov_len = 0; + } + resv = &resbuf->tail[0]; + mic.data = (u8 *)resv->iov_base + resv->iov_len + 4; + if (gss_get_mic(gsd->rsci->mechctx, &integ_buf, &mic)) + goto out_err; + svc_putnl(resv, mic.len); + memset(mic.data + mic.len, 0, + round_up_to_quad(mic.len) - mic.len); + resv->iov_len += XDR_QUADLEN(mic.len) << 2; + /* not strictly required: */ + resbuf->len += XDR_QUADLEN(mic.len) << 2; + BUG_ON(resv->iov_len > PAGE_SIZE); +out: + stat = 0; +out_err: + return stat; +} + +static inline int +svcauth_gss_wrap_resp_priv(struct svc_rqst *rqstp) +{ + struct gss_svc_data *gsd = (struct gss_svc_data *)rqstp->rq_auth_data; + struct rpc_gss_wire_cred *gc = &gsd->clcred; + struct xdr_buf *resbuf = &rqstp->rq_res; + struct page **inpages = NULL; + __be32 *p, *len; + int offset; + int pad; + + p = svcauth_gss_prepare_to_wrap(resbuf, gsd); + if (p == NULL) + return 0; + len = p++; + offset = (u8 *)p - (u8 *)resbuf->head[0].iov_base; + *p++ = htonl(gc->gc_seq); + inpages = resbuf->pages; + /* XXX: Would be better to write some xdr helper functions for + * nfs{2,3,4}xdr.c that place the data right, instead of copying: */ + + /* + * If there is currently tail data, make sure there is + * room for the head, tail, and 2 * RPC_MAX_AUTH_SIZE in + * the page, and move the current tail data such that + * there is RPC_MAX_AUTH_SIZE slack space available in + * both the head and tail. + */ + if (resbuf->tail[0].iov_base) { + BUG_ON(resbuf->tail[0].iov_base >= resbuf->head[0].iov_base + + PAGE_SIZE); + BUG_ON(resbuf->tail[0].iov_base < resbuf->head[0].iov_base); + if (resbuf->tail[0].iov_len + resbuf->head[0].iov_len + + 2 * RPC_MAX_AUTH_SIZE > PAGE_SIZE) + return -ENOMEM; + memmove(resbuf->tail[0].iov_base + RPC_MAX_AUTH_SIZE, + resbuf->tail[0].iov_base, + resbuf->tail[0].iov_len); + resbuf->tail[0].iov_base += RPC_MAX_AUTH_SIZE; + } + /* + * If there is no current tail data, make sure there is + * room for the head data, and 2 * RPC_MAX_AUTH_SIZE in the + * allotted page, and set up tail information such that there + * is RPC_MAX_AUTH_SIZE slack space available in both the + * head and tail. + */ + if (resbuf->tail[0].iov_base == NULL) { + if (resbuf->head[0].iov_len + 2*RPC_MAX_AUTH_SIZE > PAGE_SIZE) + return -ENOMEM; + resbuf->tail[0].iov_base = resbuf->head[0].iov_base + + resbuf->head[0].iov_len + RPC_MAX_AUTH_SIZE; + resbuf->tail[0].iov_len = 0; + } + if (gss_wrap(gsd->rsci->mechctx, offset, resbuf, inpages)) + return -ENOMEM; + *len = htonl(resbuf->len - offset); + pad = 3 - ((resbuf->len - offset - 1)&3); + p = (__be32 *)(resbuf->tail[0].iov_base + resbuf->tail[0].iov_len); + memset(p, 0, pad); + resbuf->tail[0].iov_len += pad; + resbuf->len += pad; + return 0; +} + +static int +svcauth_gss_release(struct svc_rqst *rqstp) +{ + struct gss_svc_data *gsd = (struct gss_svc_data *)rqstp->rq_auth_data; + struct rpc_gss_wire_cred *gc = &gsd->clcred; + struct xdr_buf *resbuf = &rqstp->rq_res; + int stat = -EINVAL; + struct sunrpc_net *sn = net_generic(rqstp->rq_xprt->xpt_net, sunrpc_net_id); + if (gc->gc_proc != RPC_GSS_PROC_DATA) goto out; /* Release can be called twice, but we only wrap once. */ - if (gsd->body_start == NULL) + if (gsd->verf_start == NULL) goto out; /* normally not set till svc_send, but we need it here: */ - resbuf->len = resbuf->head[0].iov_len - + resbuf->page_len + resbuf->tail[0].iov_len; + /* XXX: what for? Do we mess it up the moment we call svc_putu32 + * or whatever? */ + resbuf->len = total_buf_len(resbuf); switch (gc->gc_svc) { case RPC_GSS_SVC_NONE: break; case RPC_GSS_SVC_INTEGRITY: - p = gsd->body_start; - gsd->body_start = NULL; - /* move accept_stat to right place: */ - memcpy(p, p + 2, 4); - /* don't wrap in failure case: */ - /* Note: counting on not getting here if call was not even - * accepted! */ - if (*p != rpc_success) { - resbuf->head[0].iov_len -= 2 * 4; - goto out; - } - p++; - integ_offset = (u8 *)(p + 1) - (u8 *)resbuf->head[0].iov_base; - integ_len = resbuf->len - integ_offset; - BUG_ON(integ_len % 4); - *p++ = htonl(integ_len); - *p++ = htonl(gc->gc_seq); - if (xdr_buf_subsegment(resbuf, &integ_buf, integ_offset, - integ_len)) - BUG(); - if (resbuf->page_len == 0 - && resbuf->tail[0].iov_len + RPC_MAX_AUTH_SIZE - < PAGE_SIZE) { - BUG_ON(resbuf->tail[0].iov_len); - /* Use head for everything */ - resv = &resbuf->head[0]; - } else if (resbuf->tail[0].iov_base == NULL) { - /* copied from nfsd4_encode_read */ - svc_take_page(rqstp); - resbuf->tail[0].iov_base = page_address(rqstp - ->rq_respages[rqstp->rq_resused-1]); - rqstp->rq_restailpage = rqstp->rq_resused-1; - resbuf->tail[0].iov_len = 0; - resv = &resbuf->tail[0]; - } else { - resv = &resbuf->tail[0]; - } - mic.data = (u8 *)resv->iov_base + resv->iov_len + 4; - if (gss_get_mic(gsd->rsci->mechctx, &integ_buf, &mic)) + stat = svcauth_gss_wrap_resp_integ(rqstp); + if (stat) goto out_err; - svc_putu32(resv, htonl(mic.len)); - memset(mic.data + mic.len, 0, - round_up_to_quad(mic.len) - mic.len); - resv->iov_len += XDR_QUADLEN(mic.len) << 2; - /* not strictly required: */ - resbuf->len += XDR_QUADLEN(mic.len) << 2; - BUG_ON(resv->iov_len > PAGE_SIZE); break; case RPC_GSS_SVC_PRIVACY: - default: - goto out_err; + stat = svcauth_gss_wrap_resp_priv(rqstp); + if (stat) + goto out_err; + break; + /* + * For any other gc_svc value, svcauth_gss_accept() already set + * the auth_error appropriately; just fall through: + */ } out: @@ -1032,11 +1727,14 @@ out_err: if (rqstp->rq_client) auth_domain_put(rqstp->rq_client); rqstp->rq_client = NULL; + if (rqstp->rq_gssclient) + auth_domain_put(rqstp->rq_gssclient); + rqstp->rq_gssclient = NULL; if (rqstp->rq_cred.cr_group_info) put_group_info(rqstp->rq_cred.cr_group_info); rqstp->rq_cred.cr_group_info = NULL; if (gsd->rsci) - rsc_put(&gsd->rsci->h, &rsc_cache); + cache_put(&gsd->rsci->h, sn->rsc_cache); gsd->rsci = NULL; return stat; @@ -1061,23 +1759,102 @@ static struct auth_ops svcauthops_gss = { .set_client = svcauth_gss_set_client, }; -int -gss_svc_init(void) +static int rsi_cache_create_net(struct net *net) +{ + struct sunrpc_net *sn = net_generic(net, sunrpc_net_id); + struct cache_detail *cd; + int err; + + cd = cache_create_net(&rsi_cache_template, net); + if (IS_ERR(cd)) + return PTR_ERR(cd); + err = cache_register_net(cd, net); + if (err) { + cache_destroy_net(cd, net); + return err; + } + sn->rsi_cache = cd; + return 0; +} + +static void rsi_cache_destroy_net(struct net *net) { - int rv = svc_auth_register(RPC_AUTH_GSS, &svcauthops_gss); - if (rv == 0) { - cache_register(&rsc_cache); - cache_register(&rsi_cache); + struct sunrpc_net *sn = net_generic(net, sunrpc_net_id); + struct cache_detail *cd = sn->rsi_cache; + + sn->rsi_cache = NULL; + cache_purge(cd); + cache_unregister_net(cd, net); + cache_destroy_net(cd, net); +} + +static int rsc_cache_create_net(struct net *net) +{ + struct sunrpc_net *sn = net_generic(net, sunrpc_net_id); + struct cache_detail *cd; + int err; + + cd = cache_create_net(&rsc_cache_template, net); + if (IS_ERR(cd)) + return PTR_ERR(cd); + err = cache_register_net(cd, net); + if (err) { + cache_destroy_net(cd, net); + return err; } + sn->rsc_cache = cd; + return 0; +} + +static void rsc_cache_destroy_net(struct net *net) +{ + struct sunrpc_net *sn = net_generic(net, sunrpc_net_id); + struct cache_detail *cd = sn->rsc_cache; + + sn->rsc_cache = NULL; + cache_purge(cd); + cache_unregister_net(cd, net); + cache_destroy_net(cd, net); +} + +int +gss_svc_init_net(struct net *net) +{ + int rv; + + rv = rsc_cache_create_net(net); + if (rv) + return rv; + rv = rsi_cache_create_net(net); + if (rv) + goto out1; + rv = create_use_gss_proxy_proc_entry(net); + if (rv) + goto out2; + return 0; +out2: + destroy_use_gss_proxy_proc_entry(net); +out1: + rsc_cache_destroy_net(net); return rv; } void +gss_svc_shutdown_net(struct net *net) +{ + destroy_use_gss_proxy_proc_entry(net); + rsi_cache_destroy_net(net); + rsc_cache_destroy_net(net); +} + +int +gss_svc_init(void) +{ + return svc_auth_register(RPC_AUTH_GSS, &svcauthops_gss); +} + +void gss_svc_shutdown(void) { - if (cache_unregister(&rsc_cache)) - printk(KERN_ERR "auth_rpcgss: failed to unregister rsc cache\n"); - if (cache_unregister(&rsi_cache)) - printk(KERN_ERR "auth_rpcgss: failed to unregister rsi cache\n"); svc_auth_unregister(RPC_AUTH_GSS); } diff --git a/net/sunrpc/auth_null.c b/net/sunrpc/auth_null.c index f56767aaa92..f0ebe07978a 100644 --- a/net/sunrpc/auth_null.c +++ b/net/sunrpc/auth_null.c @@ -8,9 +8,7 @@ #include <linux/types.h> #include <linux/module.h> -#include <linux/utsname.h> #include <linux/sunrpc/clnt.h> -#include <linux/sched.h> #ifdef RPC_DEBUG # define RPCDBG_FACILITY RPCDBG_AUTH @@ -20,7 +18,7 @@ static struct rpc_auth null_auth; static struct rpc_cred null_cred; static struct rpc_auth * -nul_create(struct rpc_clnt *clnt, rpc_authflavor_t flavor) +nul_create(struct rpc_auth_create_args *args, struct rpc_clnt *clnt) { atomic_inc(&null_auth.au_count); return &null_auth; @@ -60,8 +58,8 @@ nul_match(struct auth_cred *acred, struct rpc_cred *cred, int taskflags) /* * Marshal credential. */ -static u32 * -nul_marshal(struct rpc_task *task, u32 *p) +static __be32 * +nul_marshal(struct rpc_task *task, __be32 *p) { *p++ = htonl(RPC_AUTH_NULL); *p++ = 0; @@ -77,12 +75,12 @@ nul_marshal(struct rpc_task *task, u32 *p) static int nul_refresh(struct rpc_task *task) { - task->tk_msg.rpc_cred->cr_flags |= RPCAUTH_CRED_UPTODATE; + set_bit(RPCAUTH_CRED_UPTODATE, &task->tk_rqstp->rq_cred->cr_flags); return 0; } -static u32 * -nul_validate(struct rpc_task *task, u32 *p) +static __be32 * +nul_validate(struct rpc_task *task, __be32 *p) { rpc_authflavor_t flavor; u32 size; @@ -90,24 +88,22 @@ nul_validate(struct rpc_task *task, u32 *p) flavor = ntohl(*p++); if (flavor != RPC_AUTH_NULL) { printk("RPC: bad verf flavor: %u\n", flavor); - return NULL; + return ERR_PTR(-EIO); } size = ntohl(*p++); if (size != 0) { printk("RPC: bad verf size: %u\n", size); - return NULL; + return ERR_PTR(-EIO); } return p; } -struct rpc_authops authnull_ops = { +const struct rpc_authops authnull_ops = { .owner = THIS_MODULE, .au_flavor = RPC_AUTH_NULL, -#ifdef RPC_DEBUG .au_name = "NULL", -#endif .create = nul_create, .destroy = nul_destroy, .lookup_cred = nul_lookup_cred, @@ -118,12 +114,15 @@ struct rpc_auth null_auth = { .au_cslack = 4, .au_rslack = 2, .au_ops = &authnull_ops, + .au_flavor = RPC_AUTH_NULL, + .au_count = ATOMIC_INIT(0), }; static -struct rpc_credops null_credops = { +const struct rpc_credops null_credops = { .cr_name = "AUTH_NULL", .crdestroy = nul_destroy_cred, + .crbind = rpcauth_generic_bind_cred, .crmatch = nul_match, .crmarshal = nul_marshal, .crrefresh = nul_refresh, @@ -132,9 +131,11 @@ struct rpc_credops null_credops = { static struct rpc_cred null_cred = { + .cr_lru = LIST_HEAD_INIT(null_cred.cr_lru), + .cr_auth = &null_auth, .cr_ops = &null_credops, .cr_count = ATOMIC_INIT(1), - .cr_flags = RPCAUTH_CRED_UPTODATE, + .cr_flags = 1UL << RPCAUTH_CRED_UPTODATE, #ifdef RPC_DEBUG .cr_magic = RPCAUTH_CRED_MAGIC, #endif diff --git a/net/sunrpc/auth_unix.c b/net/sunrpc/auth_unix.c index 890fb5ea0dc..d5d69236629 100644 --- a/net/sunrpc/auth_unix.c +++ b/net/sunrpc/auth_unix.c @@ -6,25 +6,22 @@ * Copyright (C) 1996, Olaf Kirch <okir@monad.swb.de> */ +#include <linux/slab.h> #include <linux/types.h> #include <linux/sched.h> #include <linux/module.h> #include <linux/sunrpc/clnt.h> #include <linux/sunrpc/auth.h> +#include <linux/user_namespace.h> #define NFS_NGROUPS 16 struct unx_cred { struct rpc_cred uc_base; - gid_t uc_gid; - gid_t uc_gids[NFS_NGROUPS]; + kgid_t uc_gid; + kgid_t uc_gids[NFS_NGROUPS]; }; #define uc_uid uc_base.cr_uid -#define uc_count uc_base.cr_count -#define uc_flags uc_base.cr_flags -#define uc_expire uc_base.cr_expire - -#define UNX_CRED_EXPIRE (60 * HZ) #define UNX_WRITESLACK (21 + (UNX_MAXNODENAME >> 2)) @@ -33,23 +30,22 @@ struct unx_cred { #endif static struct rpc_auth unix_auth; -static struct rpc_cred_cache unix_cred_cache; -static struct rpc_credops unix_credops; +static const struct rpc_credops unix_credops; static struct rpc_auth * -unx_create(struct rpc_clnt *clnt, rpc_authflavor_t flavor) +unx_create(struct rpc_auth_create_args *args, struct rpc_clnt *clnt) { - dprintk("RPC: creating UNIX authenticator for client %p\n", clnt); - if (atomic_inc_return(&unix_auth.au_count) == 0) - unix_cred_cache.nextgc = jiffies + (unix_cred_cache.expire >> 1); + dprintk("RPC: creating UNIX authenticator for client %p\n", + clnt); + atomic_inc(&unix_auth.au_count); return &unix_auth; } static void unx_destroy(struct rpc_auth *auth) { - dprintk("RPC: destroying UNIX authenticator %p\n", auth); - rpcauth_free_credcache(auth); + dprintk("RPC: destroying UNIX authenticator %p\n", auth); + rpcauth_clear_credcache(auth->au_credcache); } /* @@ -65,41 +61,51 @@ static struct rpc_cred * unx_create_cred(struct rpc_auth *auth, struct auth_cred *acred, int flags) { struct unx_cred *cred; - int i; + unsigned int groups = 0; + unsigned int i; - dprintk("RPC: allocating UNIX cred for uid %d gid %d\n", - acred->uid, acred->gid); + dprintk("RPC: allocating UNIX cred for uid %d gid %d\n", + from_kuid(&init_user_ns, acred->uid), + from_kgid(&init_user_ns, acred->gid)); - if (!(cred = (struct unx_cred *) kmalloc(sizeof(*cred), GFP_KERNEL))) + if (!(cred = kmalloc(sizeof(*cred), GFP_NOFS))) return ERR_PTR(-ENOMEM); - atomic_set(&cred->uc_count, 1); - cred->uc_flags = RPCAUTH_CRED_UPTODATE; - if (flags & RPC_TASK_ROOTCREDS) { - cred->uc_uid = 0; - cred->uc_gid = 0; - cred->uc_gids[0] = NOGROUP; - } else { - int groups = acred->group_info->ngroups; - if (groups > NFS_NGROUPS) - groups = NFS_NGROUPS; - - cred->uc_uid = acred->uid; - cred->uc_gid = acred->gid; - for (i = 0; i < groups; i++) - cred->uc_gids[i] = GROUP_AT(acred->group_info, i); - if (i < NFS_NGROUPS) - cred->uc_gids[i] = NOGROUP; - } - cred->uc_base.cr_ops = &unix_credops; + rpcauth_init_cred(&cred->uc_base, acred, auth, &unix_credops); + cred->uc_base.cr_flags = 1UL << RPCAUTH_CRED_UPTODATE; + + if (acred->group_info != NULL) + groups = acred->group_info->ngroups; + if (groups > NFS_NGROUPS) + groups = NFS_NGROUPS; + + cred->uc_gid = acred->gid; + for (i = 0; i < groups; i++) + cred->uc_gids[i] = GROUP_AT(acred->group_info, i); + if (i < NFS_NGROUPS) + cred->uc_gids[i] = INVALID_GID; - return (struct rpc_cred *) cred; + return &cred->uc_base; +} + +static void +unx_free_cred(struct unx_cred *unx_cred) +{ + dprintk("RPC: unx_free_cred %p\n", unx_cred); + kfree(unx_cred); +} + +static void +unx_free_cred_callback(struct rcu_head *head) +{ + struct unx_cred *unx_cred = container_of(head, struct unx_cred, uc_base.cr_rcu); + unx_free_cred(unx_cred); } static void unx_destroy_cred(struct rpc_cred *cred) { - kfree(cred); + call_rcu(&cred->cr_rcu, unx_free_cred_callback); } /* @@ -108,41 +114,38 @@ unx_destroy_cred(struct rpc_cred *cred) * request root creds (e.g. for NFS swapping). */ static int -unx_match(struct auth_cred *acred, struct rpc_cred *rcred, int taskflags) +unx_match(struct auth_cred *acred, struct rpc_cred *rcred, int flags) { - struct unx_cred *cred = (struct unx_cred *) rcred; - int i; + struct unx_cred *cred = container_of(rcred, struct unx_cred, uc_base); + unsigned int groups = 0; + unsigned int i; - if (!(taskflags & RPC_TASK_ROOTCREDS)) { - int groups; - if (cred->uc_uid != acred->uid - || cred->uc_gid != acred->gid) - return 0; + if (!uid_eq(cred->uc_uid, acred->uid) || !gid_eq(cred->uc_gid, acred->gid)) + return 0; + if (acred->group_info != NULL) groups = acred->group_info->ngroups; - if (groups > NFS_NGROUPS) - groups = NFS_NGROUPS; - for (i = 0; i < groups ; i++) - if (cred->uc_gids[i] != GROUP_AT(acred->group_info, i)) - return 0; - return 1; - } - return (cred->uc_uid == 0 - && cred->uc_gid == 0 - && cred->uc_gids[0] == (gid_t) NOGROUP); + if (groups > NFS_NGROUPS) + groups = NFS_NGROUPS; + for (i = 0; i < groups ; i++) + if (!gid_eq(cred->uc_gids[i], GROUP_AT(acred->group_info, i))) + return 0; + if (groups < NFS_NGROUPS && gid_valid(cred->uc_gids[groups])) + return 0; + return 1; } /* * Marshal credentials. * Maybe we should keep a cached credential for performance reasons. */ -static u32 * -unx_marshal(struct rpc_task *task, u32 *p) +static __be32 * +unx_marshal(struct rpc_task *task, __be32 *p) { struct rpc_clnt *clnt = task->tk_client; - struct unx_cred *cred = (struct unx_cred *) task->tk_msg.rpc_cred; - u32 *base, *hold; + struct unx_cred *cred = container_of(task->tk_rqstp->rq_cred, struct unx_cred, uc_base); + __be32 *base, *hold; int i; *p++ = htonl(RPC_AUTH_UNIX); @@ -154,11 +157,11 @@ unx_marshal(struct rpc_task *task, u32 *p) */ p = xdr_encode_array(p, clnt->cl_nodename, clnt->cl_nodelen); - *p++ = htonl((u32) cred->uc_uid); - *p++ = htonl((u32) cred->uc_gid); + *p++ = htonl((u32) from_kuid(&init_user_ns, cred->uc_uid)); + *p++ = htonl((u32) from_kgid(&init_user_ns, cred->uc_gid)); hold = p++; - for (i = 0; i < 16 && cred->uc_gids[i] != (gid_t) NOGROUP; i++) - *p++ = htonl((u32) cred->uc_gids[i]); + for (i = 0; i < 16 && gid_valid(cred->uc_gids[i]); i++) + *p++ = htonl((u32) from_kgid(&init_user_ns, cred->uc_gids[i])); *hold = htonl(p - hold - 1); /* gid array length */ *base = htonl((p - base - 1) << 2); /* cred length */ @@ -174,12 +177,12 @@ unx_marshal(struct rpc_task *task, u32 *p) static int unx_refresh(struct rpc_task *task) { - task->tk_msg.rpc_cred->cr_flags |= RPCAUTH_CRED_UPTODATE; + set_bit(RPCAUTH_CRED_UPTODATE, &task->tk_rqstp->rq_cred->cr_flags); return 0; } -static u32 * -unx_validate(struct rpc_task *task, u32 *p) +static __be32 * +unx_validate(struct rpc_task *task, __be32 *p) { rpc_authflavor_t flavor; u32 size; @@ -189,26 +192,34 @@ unx_validate(struct rpc_task *task, u32 *p) flavor != RPC_AUTH_UNIX && flavor != RPC_AUTH_SHORT) { printk("RPC: bad verf flavor: %u\n", flavor); - return NULL; + return ERR_PTR(-EIO); } size = ntohl(*p++); if (size > RPC_MAX_AUTH_SIZE) { printk("RPC: giant verf size: %u\n", size); - return NULL; + return ERR_PTR(-EIO); } - task->tk_auth->au_rslack = (size >> 2) + 2; + task->tk_rqstp->rq_cred->cr_auth->au_rslack = (size >> 2) + 2; p += (size >> 2); return p; } -struct rpc_authops authunix_ops = { +int __init rpc_init_authunix(void) +{ + return rpcauth_init_credcache(&unix_auth); +} + +void rpc_destroy_authunix(void) +{ + rpcauth_destroy_credcache(&unix_auth); +} + +const struct rpc_authops authunix_ops = { .owner = THIS_MODULE, .au_flavor = RPC_AUTH_UNIX, -#ifdef RPC_DEBUG .au_name = "UNIX", -#endif .create = unx_create, .destroy = unx_destroy, .lookup_cred = unx_lookup_cred, @@ -216,23 +227,19 @@ struct rpc_authops authunix_ops = { }; static -struct rpc_cred_cache unix_cred_cache = { - .expire = UNX_CRED_EXPIRE, -}; - -static struct rpc_auth unix_auth = { .au_cslack = UNX_WRITESLACK, .au_rslack = 2, /* assume AUTH_NULL verf */ .au_ops = &authunix_ops, + .au_flavor = RPC_AUTH_UNIX, .au_count = ATOMIC_INIT(0), - .au_credcache = &unix_cred_cache, }; static -struct rpc_credops unix_credops = { +const struct rpc_credops unix_credops = { .cr_name = "AUTH_UNIX", .crdestroy = unx_destroy_cred, + .crbind = rpcauth_generic_bind_cred, .crmatch = unx_match, .crmarshal = unx_marshal, .crrefresh = unx_refresh, diff --git a/net/sunrpc/backchannel_rqst.c b/net/sunrpc/backchannel_rqst.c new file mode 100644 index 00000000000..9761a0da964 --- /dev/null +++ b/net/sunrpc/backchannel_rqst.c @@ -0,0 +1,322 @@ +/****************************************************************************** + +(c) 2007 Network Appliance, Inc. All Rights Reserved. +(c) 2009 NetApp. All Rights Reserved. + +NetApp provides this source code under the GPL v2 License. +The GPL v2 license is available at +http://opensource.org/licenses/gpl-license.php. + +THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS +"AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT +LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR +A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR +CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, +EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, +PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR +PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF +LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING +NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS +SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + +******************************************************************************/ + +#include <linux/tcp.h> +#include <linux/slab.h> +#include <linux/sunrpc/xprt.h> +#include <linux/export.h> +#include <linux/sunrpc/bc_xprt.h> + +#ifdef RPC_DEBUG +#define RPCDBG_FACILITY RPCDBG_TRANS +#endif + +/* + * Helper routines that track the number of preallocation elements + * on the transport. + */ +static inline int xprt_need_to_requeue(struct rpc_xprt *xprt) +{ + return xprt->bc_alloc_count > 0; +} + +static inline void xprt_inc_alloc_count(struct rpc_xprt *xprt, unsigned int n) +{ + xprt->bc_alloc_count += n; +} + +static inline int xprt_dec_alloc_count(struct rpc_xprt *xprt, unsigned int n) +{ + return xprt->bc_alloc_count -= n; +} + +/* + * Free the preallocated rpc_rqst structure and the memory + * buffers hanging off of it. + */ +static void xprt_free_allocation(struct rpc_rqst *req) +{ + struct xdr_buf *xbufp; + + dprintk("RPC: free allocations for req= %p\n", req); + WARN_ON_ONCE(test_bit(RPC_BC_PA_IN_USE, &req->rq_bc_pa_state)); + xbufp = &req->rq_private_buf; + free_page((unsigned long)xbufp->head[0].iov_base); + xbufp = &req->rq_snd_buf; + free_page((unsigned long)xbufp->head[0].iov_base); + kfree(req); +} + +/* + * Preallocate up to min_reqs structures and related buffers for use + * by the backchannel. This function can be called multiple times + * when creating new sessions that use the same rpc_xprt. The + * preallocated buffers are added to the pool of resources used by + * the rpc_xprt. Anyone of these resources may be used used by an + * incoming callback request. It's up to the higher levels in the + * stack to enforce that the maximum number of session slots is not + * being exceeded. + * + * Some callback arguments can be large. For example, a pNFS server + * using multiple deviceids. The list can be unbound, but the client + * has the ability to tell the server the maximum size of the callback + * requests. Each deviceID is 16 bytes, so allocate one page + * for the arguments to have enough room to receive a number of these + * deviceIDs. The NFS client indicates to the pNFS server that its + * callback requests can be up to 4096 bytes in size. + */ +int xprt_setup_backchannel(struct rpc_xprt *xprt, unsigned int min_reqs) +{ + struct page *page_rcv = NULL, *page_snd = NULL; + struct xdr_buf *xbufp = NULL; + struct rpc_rqst *req, *tmp; + struct list_head tmp_list; + int i; + + dprintk("RPC: setup backchannel transport\n"); + + /* + * We use a temporary list to keep track of the preallocated + * buffers. Once we're done building the list we splice it + * into the backchannel preallocation list off of the rpc_xprt + * struct. This helps minimize the amount of time the list + * lock is held on the rpc_xprt struct. It also makes cleanup + * easier in case of memory allocation errors. + */ + INIT_LIST_HEAD(&tmp_list); + for (i = 0; i < min_reqs; i++) { + /* Pre-allocate one backchannel rpc_rqst */ + req = kzalloc(sizeof(struct rpc_rqst), GFP_KERNEL); + if (req == NULL) { + printk(KERN_ERR "Failed to create bc rpc_rqst\n"); + goto out_free; + } + + /* Add the allocated buffer to the tmp list */ + dprintk("RPC: adding req= %p\n", req); + list_add(&req->rq_bc_pa_list, &tmp_list); + + req->rq_xprt = xprt; + INIT_LIST_HEAD(&req->rq_list); + INIT_LIST_HEAD(&req->rq_bc_list); + + /* Preallocate one XDR receive buffer */ + page_rcv = alloc_page(GFP_KERNEL); + if (page_rcv == NULL) { + printk(KERN_ERR "Failed to create bc receive xbuf\n"); + goto out_free; + } + xbufp = &req->rq_rcv_buf; + xbufp->head[0].iov_base = page_address(page_rcv); + xbufp->head[0].iov_len = PAGE_SIZE; + xbufp->tail[0].iov_base = NULL; + xbufp->tail[0].iov_len = 0; + xbufp->page_len = 0; + xbufp->len = PAGE_SIZE; + xbufp->buflen = PAGE_SIZE; + + /* Preallocate one XDR send buffer */ + page_snd = alloc_page(GFP_KERNEL); + if (page_snd == NULL) { + printk(KERN_ERR "Failed to create bc snd xbuf\n"); + goto out_free; + } + + xbufp = &req->rq_snd_buf; + xbufp->head[0].iov_base = page_address(page_snd); + xbufp->head[0].iov_len = 0; + xbufp->tail[0].iov_base = NULL; + xbufp->tail[0].iov_len = 0; + xbufp->page_len = 0; + xbufp->len = 0; + xbufp->buflen = PAGE_SIZE; + } + + /* + * Add the temporary list to the backchannel preallocation list + */ + spin_lock_bh(&xprt->bc_pa_lock); + list_splice(&tmp_list, &xprt->bc_pa_list); + xprt_inc_alloc_count(xprt, min_reqs); + spin_unlock_bh(&xprt->bc_pa_lock); + + dprintk("RPC: setup backchannel transport done\n"); + return 0; + +out_free: + /* + * Memory allocation failed, free the temporary list + */ + list_for_each_entry_safe(req, tmp, &tmp_list, rq_bc_pa_list) { + list_del(&req->rq_bc_pa_list); + xprt_free_allocation(req); + } + + dprintk("RPC: setup backchannel transport failed\n"); + return -ENOMEM; +} +EXPORT_SYMBOL_GPL(xprt_setup_backchannel); + +/** + * xprt_destroy_backchannel - Destroys the backchannel preallocated structures. + * @xprt: the transport holding the preallocated strucures + * @max_reqs the maximum number of preallocated structures to destroy + * + * Since these structures may have been allocated by multiple calls + * to xprt_setup_backchannel, we only destroy up to the maximum number + * of reqs specified by the caller. + */ +void xprt_destroy_backchannel(struct rpc_xprt *xprt, unsigned int max_reqs) +{ + struct rpc_rqst *req = NULL, *tmp = NULL; + + dprintk("RPC: destroy backchannel transport\n"); + + if (max_reqs == 0) + goto out; + + spin_lock_bh(&xprt->bc_pa_lock); + xprt_dec_alloc_count(xprt, max_reqs); + list_for_each_entry_safe(req, tmp, &xprt->bc_pa_list, rq_bc_pa_list) { + dprintk("RPC: req=%p\n", req); + list_del(&req->rq_bc_pa_list); + xprt_free_allocation(req); + if (--max_reqs == 0) + break; + } + spin_unlock_bh(&xprt->bc_pa_lock); + +out: + dprintk("RPC: backchannel list empty= %s\n", + list_empty(&xprt->bc_pa_list) ? "true" : "false"); +} +EXPORT_SYMBOL_GPL(xprt_destroy_backchannel); + +static struct rpc_rqst *xprt_alloc_bc_request(struct rpc_xprt *xprt, __be32 xid) +{ + struct rpc_rqst *req = NULL; + + dprintk("RPC: allocate a backchannel request\n"); + if (list_empty(&xprt->bc_pa_list)) + goto not_found; + + req = list_first_entry(&xprt->bc_pa_list, struct rpc_rqst, + rq_bc_pa_list); + req->rq_reply_bytes_recvd = 0; + req->rq_bytes_sent = 0; + memcpy(&req->rq_private_buf, &req->rq_rcv_buf, + sizeof(req->rq_private_buf)); + req->rq_xid = xid; + req->rq_connect_cookie = xprt->connect_cookie; +not_found: + dprintk("RPC: backchannel req=%p\n", req); + return req; +} + +/* + * Return the preallocated rpc_rqst structure and XDR buffers + * associated with this rpc_task. + */ +void xprt_free_bc_request(struct rpc_rqst *req) +{ + struct rpc_xprt *xprt = req->rq_xprt; + + dprintk("RPC: free backchannel req=%p\n", req); + + req->rq_connect_cookie = xprt->connect_cookie - 1; + smp_mb__before_atomic(); + WARN_ON_ONCE(!test_bit(RPC_BC_PA_IN_USE, &req->rq_bc_pa_state)); + clear_bit(RPC_BC_PA_IN_USE, &req->rq_bc_pa_state); + smp_mb__after_atomic(); + + if (!xprt_need_to_requeue(xprt)) { + /* + * The last remaining session was destroyed while this + * entry was in use. Free the entry and don't attempt + * to add back to the list because there is no need to + * have anymore preallocated entries. + */ + dprintk("RPC: Last session removed req=%p\n", req); + xprt_free_allocation(req); + return; + } + + /* + * Return it to the list of preallocations so that it + * may be reused by a new callback request. + */ + spin_lock_bh(&xprt->bc_pa_lock); + list_add_tail(&req->rq_bc_pa_list, &xprt->bc_pa_list); + spin_unlock_bh(&xprt->bc_pa_lock); +} + +/* + * One or more rpc_rqst structure have been preallocated during the + * backchannel setup. Buffer space for the send and private XDR buffers + * has been preallocated as well. Use xprt_alloc_bc_request to allocate + * to this request. Use xprt_free_bc_request to return it. + * + * We know that we're called in soft interrupt context, grab the spin_lock + * since there is no need to grab the bottom half spin_lock. + * + * Return an available rpc_rqst, otherwise NULL if non are available. + */ +struct rpc_rqst *xprt_lookup_bc_request(struct rpc_xprt *xprt, __be32 xid) +{ + struct rpc_rqst *req; + + spin_lock(&xprt->bc_pa_lock); + list_for_each_entry(req, &xprt->bc_pa_list, rq_bc_pa_list) { + if (req->rq_connect_cookie != xprt->connect_cookie) + continue; + if (req->rq_xid == xid) + goto found; + } + req = xprt_alloc_bc_request(xprt, xid); +found: + spin_unlock(&xprt->bc_pa_lock); + return req; +} + +/* + * Add callback request to callback list. The callback + * service sleeps on the sv_cb_waitq waiting for new + * requests. Wake it up after adding enqueing the + * request. + */ +void xprt_complete_bc_request(struct rpc_rqst *req, uint32_t copied) +{ + struct rpc_xprt *xprt = req->rq_xprt; + struct svc_serv *bc_serv = xprt->bc_serv; + + req->rq_private_buf.len = copied; + set_bit(RPC_BC_PA_IN_USE, &req->rq_bc_pa_state); + + dprintk("RPC: add callback request to list\n"); + spin_lock(&bc_serv->sv_cb_lock); + list_del(&req->rq_bc_pa_list); + list_add(&req->rq_bc_list, &bc_serv->sv_cb_list); + wake_up(&bc_serv->sv_cb_waitq); + spin_unlock(&bc_serv->sv_cb_lock); +} + diff --git a/net/sunrpc/bc_svc.c b/net/sunrpc/bc_svc.c new file mode 100644 index 00000000000..15c7a8a1c24 --- /dev/null +++ b/net/sunrpc/bc_svc.c @@ -0,0 +1,63 @@ +/****************************************************************************** + +(c) 2007 Network Appliance, Inc. All Rights Reserved. +(c) 2009 NetApp. All Rights Reserved. + +NetApp provides this source code under the GPL v2 License. +The GPL v2 license is available at +http://opensource.org/licenses/gpl-license.php. + +THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS +"AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT +LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR +A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR +CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, +EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, +PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR +PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF +LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING +NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS +SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + +******************************************************************************/ + +/* + * The NFSv4.1 callback service helper routines. + * They implement the transport level processing required to send the + * reply over an existing open connection previously established by the client. + */ + +#include <linux/module.h> + +#include <linux/sunrpc/xprt.h> +#include <linux/sunrpc/sched.h> +#include <linux/sunrpc/bc_xprt.h> + +#define RPCDBG_FACILITY RPCDBG_SVCDSP + +/* Empty callback ops */ +static const struct rpc_call_ops nfs41_callback_ops = { +}; + + +/* + * Send the callback reply + */ +int bc_send(struct rpc_rqst *req) +{ + struct rpc_task *task; + int ret; + + dprintk("RPC: bc_send req= %p\n", req); + task = rpc_run_bc_task(req, &nfs41_callback_ops); + if (IS_ERR(task)) + ret = PTR_ERR(task); + else { + WARN_ON_ONCE(atomic_read(&task->tk_count) != 1); + ret = task->tk_status; + rpc_put_task(task); + } + dprintk("RPC: bc_send ret= %d\n", ret); + return ret; +} + diff --git a/net/sunrpc/cache.c b/net/sunrpc/cache.c index f509e999276..06636214113 100644 --- a/net/sunrpc/cache.c +++ b/net/sunrpc/cache.c @@ -26,28 +26,212 @@ #include <linux/proc_fs.h> #include <linux/net.h> #include <linux/workqueue.h> +#include <linux/mutex.h> +#include <linux/pagemap.h> #include <asm/ioctls.h> #include <linux/sunrpc/types.h> #include <linux/sunrpc/cache.h> #include <linux/sunrpc/stats.h> +#include <linux/sunrpc/rpc_pipe_fs.h> +#include "netns.h" #define RPCDBG_FACILITY RPCDBG_CACHE -static void cache_defer_req(struct cache_req *req, struct cache_head *item); +static bool cache_defer_req(struct cache_req *req, struct cache_head *item); static void cache_revisit_request(struct cache_head *item); -void cache_init(struct cache_head *h) +static void cache_init(struct cache_head *h) { - time_t now = get_seconds(); + time_t now = seconds_since_boot(); h->next = NULL; h->flags = 0; - atomic_set(&h->refcnt, 1); + kref_init(&h->ref); h->expiry_time = now + CACHE_NEW_EXPIRY; h->last_refresh = now; } +struct cache_head *sunrpc_cache_lookup(struct cache_detail *detail, + struct cache_head *key, int hash) +{ + struct cache_head **head, **hp; + struct cache_head *new = NULL, *freeme = NULL; + + head = &detail->hash_table[hash]; + + read_lock(&detail->hash_lock); + + for (hp=head; *hp != NULL ; hp = &(*hp)->next) { + struct cache_head *tmp = *hp; + if (detail->match(tmp, key)) { + if (cache_is_expired(detail, tmp)) + /* This entry is expired, we will discard it. */ + break; + cache_get(tmp); + read_unlock(&detail->hash_lock); + return tmp; + } + } + read_unlock(&detail->hash_lock); + /* Didn't find anything, insert an empty entry */ + + new = detail->alloc(); + if (!new) + return NULL; + /* must fully initialise 'new', else + * we might get lose if we need to + * cache_put it soon. + */ + cache_init(new); + detail->init(new, key); + + write_lock(&detail->hash_lock); + + /* check if entry appeared while we slept */ + for (hp=head; *hp != NULL ; hp = &(*hp)->next) { + struct cache_head *tmp = *hp; + if (detail->match(tmp, key)) { + if (cache_is_expired(detail, tmp)) { + *hp = tmp->next; + tmp->next = NULL; + detail->entries --; + freeme = tmp; + break; + } + cache_get(tmp); + write_unlock(&detail->hash_lock); + cache_put(new, detail); + return tmp; + } + } + new->next = *head; + *head = new; + detail->entries++; + cache_get(new); + write_unlock(&detail->hash_lock); + + if (freeme) + cache_put(freeme, detail); + return new; +} +EXPORT_SYMBOL_GPL(sunrpc_cache_lookup); + + +static void cache_dequeue(struct cache_detail *detail, struct cache_head *ch); + +static void cache_fresh_locked(struct cache_head *head, time_t expiry) +{ + head->expiry_time = expiry; + head->last_refresh = seconds_since_boot(); + smp_wmb(); /* paired with smp_rmb() in cache_is_valid() */ + set_bit(CACHE_VALID, &head->flags); +} + +static void cache_fresh_unlocked(struct cache_head *head, + struct cache_detail *detail) +{ + if (test_and_clear_bit(CACHE_PENDING, &head->flags)) { + cache_revisit_request(head); + cache_dequeue(detail, head); + } +} + +struct cache_head *sunrpc_cache_update(struct cache_detail *detail, + struct cache_head *new, struct cache_head *old, int hash) +{ + /* The 'old' entry is to be replaced by 'new'. + * If 'old' is not VALID, we update it directly, + * otherwise we need to replace it + */ + struct cache_head **head; + struct cache_head *tmp; + + if (!test_bit(CACHE_VALID, &old->flags)) { + write_lock(&detail->hash_lock); + if (!test_bit(CACHE_VALID, &old->flags)) { + if (test_bit(CACHE_NEGATIVE, &new->flags)) + set_bit(CACHE_NEGATIVE, &old->flags); + else + detail->update(old, new); + cache_fresh_locked(old, new->expiry_time); + write_unlock(&detail->hash_lock); + cache_fresh_unlocked(old, detail); + return old; + } + write_unlock(&detail->hash_lock); + } + /* We need to insert a new entry */ + tmp = detail->alloc(); + if (!tmp) { + cache_put(old, detail); + return NULL; + } + cache_init(tmp); + detail->init(tmp, old); + head = &detail->hash_table[hash]; + + write_lock(&detail->hash_lock); + if (test_bit(CACHE_NEGATIVE, &new->flags)) + set_bit(CACHE_NEGATIVE, &tmp->flags); + else + detail->update(tmp, new); + tmp->next = *head; + *head = tmp; + detail->entries++; + cache_get(tmp); + cache_fresh_locked(tmp, new->expiry_time); + cache_fresh_locked(old, 0); + write_unlock(&detail->hash_lock); + cache_fresh_unlocked(tmp, detail); + cache_fresh_unlocked(old, detail); + cache_put(old, detail); + return tmp; +} +EXPORT_SYMBOL_GPL(sunrpc_cache_update); + +static int cache_make_upcall(struct cache_detail *cd, struct cache_head *h) +{ + if (cd->cache_upcall) + return cd->cache_upcall(cd, h); + return sunrpc_cache_pipe_upcall(cd, h); +} + +static inline int cache_is_valid(struct cache_head *h) +{ + if (!test_bit(CACHE_VALID, &h->flags)) + return -EAGAIN; + else { + /* entry is valid */ + if (test_bit(CACHE_NEGATIVE, &h->flags)) + return -ENOENT; + else { + /* + * In combination with write barrier in + * sunrpc_cache_update, ensures that anyone + * using the cache entry after this sees the + * updated contents: + */ + smp_rmb(); + return 0; + } + } +} + +static int try_to_negate_entry(struct cache_detail *detail, struct cache_head *h) +{ + int rv; + + write_lock(&detail->hash_lock); + rv = cache_is_valid(h); + if (rv == -EAGAIN) { + set_bit(CACHE_NEGATIVE, &h->flags); + cache_fresh_locked(h, seconds_since_boot()+CACHE_NEW_EXPIRY); + rv = -ENOENT; + } + write_unlock(&detail->hash_lock); + cache_fresh_unlocked(h, detail); + return rv; +} -static int cache_make_upcall(struct cache_detail *detail, struct cache_head *h); /* * This is the generic cache management routine for all * the authentication caches. @@ -56,7 +240,10 @@ static int cache_make_upcall(struct cache_detail *detail, struct cache_head *h); * * * Returns 0 if the cache_head can be used, or cache_puts it and returns - * -EAGAIN if upcall is pending, + * -EAGAIN if upcall is pending and request has been queued + * -ETIMEDOUT if upcall failed or request could not be queue or + * upcall completed but item is still invalid (implying that + * the cache item has been replaced with a newer one). * -ENOENT if cache entry was negative */ int cache_check(struct cache_detail *detail, @@ -66,67 +253,47 @@ int cache_check(struct cache_detail *detail, long refresh_age, age; /* First decide return status as best we can */ - if (!test_bit(CACHE_VALID, &h->flags) || - h->expiry_time < get_seconds()) - rv = -EAGAIN; - else if (detail->flush_time > h->last_refresh) - rv = -EAGAIN; - else { - /* entry is valid */ - if (test_bit(CACHE_NEGATIVE, &h->flags)) - rv = -ENOENT; - else rv = 0; - } + rv = cache_is_valid(h); /* now see if we want to start an upcall */ refresh_age = (h->expiry_time - h->last_refresh); - age = get_seconds() - h->last_refresh; + age = seconds_since_boot() - h->last_refresh; if (rqstp == NULL) { if (rv == -EAGAIN) rv = -ENOENT; - } else if (rv == -EAGAIN || age > refresh_age/2) { - dprintk("Want update, refage=%ld, age=%ld\n", refresh_age, age); + } else if (rv == -EAGAIN || + (h->expiry_time != 0 && age > refresh_age/2)) { + dprintk("RPC: Want update, refage=%ld, age=%ld\n", + refresh_age, age); if (!test_and_set_bit(CACHE_PENDING, &h->flags)) { switch (cache_make_upcall(detail, h)) { case -EINVAL: - clear_bit(CACHE_PENDING, &h->flags); - if (rv == -EAGAIN) { - set_bit(CACHE_NEGATIVE, &h->flags); - cache_fresh(detail, h, get_seconds()+CACHE_NEW_EXPIRY); - rv = -ENOENT; - } + rv = try_to_negate_entry(detail, h); break; - case -EAGAIN: - clear_bit(CACHE_PENDING, &h->flags); - cache_revisit_request(h); + cache_fresh_unlocked(h, detail); break; } } } - if (rv == -EAGAIN) - cache_defer_req(rqstp, h); - - if (rv && h) - detail->cache_put(h, detail); + if (rv == -EAGAIN) { + if (!cache_defer_req(rqstp, h)) { + /* + * Request was not deferred; handle it as best + * we can ourselves: + */ + rv = cache_is_valid(h); + if (rv == -EAGAIN) + rv = -ETIMEDOUT; + } + } + if (rv) + cache_put(h, detail); return rv; } - -static void queue_loose(struct cache_detail *detail, struct cache_head *ch); - -void cache_fresh(struct cache_detail *detail, - struct cache_head *head, time_t expiry) -{ - - head->expiry_time = expiry; - head->last_refresh = get_seconds(); - if (!test_and_set_bit(CACHE_VALID, &head->flags)) - cache_revisit_request(head); - if (test_and_clear_bit(CACHE_PENDING, &head->flags)) - queue_loose(detail, head); -} +EXPORT_SYMBOL_GPL(cache_check); /* * caches need to be periodically cleaned. @@ -134,7 +301,7 @@ void cache_fresh(struct cache_detail *detail, * a current pointer into that list and into the table * for that entry. * - * Each time clean_cache is called it finds the next non-empty entry + * Each time cache_clean is called it finds the next non-empty entry * in the current table and walks the list in that entry * looking for entries that can be removed. * @@ -157,7 +324,7 @@ void cache_fresh(struct cache_detail *detail, * * A table is then only scanned if the current time is at least * the nextcheck time. - * + * */ static LIST_HEAD(cache_list); @@ -165,51 +332,11 @@ static DEFINE_SPINLOCK(cache_list_lock); static struct cache_detail *current_detail; static int current_index; -static struct file_operations cache_file_operations; -static struct file_operations content_file_operations; -static struct file_operations cache_flush_operations; - -static void do_cache_clean(void *data); -static DECLARE_WORK(cache_cleaner, do_cache_clean, NULL); - -void cache_register(struct cache_detail *cd) -{ - cd->proc_ent = proc_mkdir(cd->name, proc_net_rpc); - if (cd->proc_ent) { - struct proc_dir_entry *p; - cd->proc_ent->owner = cd->owner; - cd->channel_ent = cd->content_ent = NULL; - - p = create_proc_entry("flush", S_IFREG|S_IRUSR|S_IWUSR, - cd->proc_ent); - cd->flush_ent = p; - if (p) { - p->proc_fops = &cache_flush_operations; - p->owner = cd->owner; - p->data = cd; - } - - if (cd->cache_request || cd->cache_parse) { - p = create_proc_entry("channel", S_IFREG|S_IRUSR|S_IWUSR, - cd->proc_ent); - cd->channel_ent = p; - if (p) { - p->proc_fops = &cache_file_operations; - p->owner = cd->owner; - p->data = cd; - } - } - if (cd->cache_show) { - p = create_proc_entry("content", S_IFREG|S_IRUSR|S_IWUSR, - cd->proc_ent); - cd->content_ent = p; - if (p) { - p->proc_fops = &content_file_operations; - p->owner = cd->owner; - p->data = cd; - } - } - } +static void do_cache_clean(struct work_struct *work); +static struct delayed_work cache_cleaner; + +void sunrpc_init_cache_detail(struct cache_detail *cd) +{ rwlock_init(&cd->hash_lock); INIT_LIST_HEAD(&cd->queue); spin_lock(&cache_list_lock); @@ -222,10 +349,11 @@ void cache_register(struct cache_detail *cd) spin_unlock(&cache_list_lock); /* start the cleaning process */ - schedule_work(&cache_cleaner); + schedule_delayed_work(&cache_cleaner, 0); } +EXPORT_SYMBOL_GPL(sunrpc_init_cache_detail); -int cache_unregister(struct cache_detail *cd) +void sunrpc_destroy_cache_detail(struct cache_detail *cd) { cache_purge(cd); spin_lock(&cache_list_lock); @@ -233,31 +361,22 @@ int cache_unregister(struct cache_detail *cd) if (cd->entries || atomic_read(&cd->inuse)) { write_unlock(&cd->hash_lock); spin_unlock(&cache_list_lock); - return -EBUSY; + goto out; } if (current_detail == cd) current_detail = NULL; list_del_init(&cd->others); write_unlock(&cd->hash_lock); spin_unlock(&cache_list_lock); - if (cd->proc_ent) { - if (cd->flush_ent) - remove_proc_entry("flush", cd->proc_ent); - if (cd->channel_ent) - remove_proc_entry("channel", cd->proc_ent); - if (cd->content_ent) - remove_proc_entry("content", cd->proc_ent); - - cd->proc_ent = NULL; - remove_proc_entry(cd->name, proc_net_rpc); - } if (list_empty(&cache_list)) { /* module must be being unloaded so its safe to kill the worker */ - cancel_delayed_work(&cache_cleaner); - flush_scheduled_work(); + cancel_delayed_work_sync(&cache_cleaner); } - return 0; + return; +out: + printk(KERN_ERR "RPC: failed to unregister %s cache\n", cd->name); } +EXPORT_SYMBOL_GPL(sunrpc_destroy_cache_detail); /* clean cache tries to find something to clean * and cleans it. @@ -285,11 +404,11 @@ static int cache_clean(void) return -1; } current_detail = list_entry(next, struct cache_detail, others); - if (current_detail->nextcheck > get_seconds()) + if (current_detail->nextcheck > seconds_since_boot()) current_index = current_detail->hash_size; else { current_index = 0; - current_detail->nextcheck = get_seconds()+30*60; + current_detail->nextcheck = seconds_since_boot()+30*60; } } @@ -300,43 +419,39 @@ static int cache_clean(void) current_index++; /* find a cleanable entry in the bucket and clean it, or set to next bucket */ - + if (current_detail && current_index < current_detail->hash_size) { struct cache_head *ch, **cp; struct cache_detail *d; - + write_lock(¤t_detail->hash_lock); /* Ok, now to clean this strand */ - + cp = & current_detail->hash_table[current_index]; - ch = *cp; - for (; ch; cp= & ch->next, ch= *cp) { + for (ch = *cp ; ch ; cp = & ch->next, ch = *cp) { if (current_detail->nextcheck > ch->expiry_time) current_detail->nextcheck = ch->expiry_time+1; - if (ch->expiry_time >= get_seconds() - && ch->last_refresh >= current_detail->flush_time - ) + if (!cache_is_expired(current_detail, ch)) continue; - if (test_and_clear_bit(CACHE_PENDING, &ch->flags)) - queue_loose(current_detail, ch); - if (atomic_read(&ch->refcnt) == 1) - break; - } - if (ch) { *cp = ch->next; ch->next = NULL; current_detail->entries--; rv = 1; + break; } + write_unlock(¤t_detail->hash_lock); d = current_detail; if (!ch) current_index ++; spin_unlock(&cache_list_lock); - if (ch) - d->cache_put(ch, d); + if (ch) { + set_bit(CACHE_CLEANED, &ch->flags); + cache_fresh_unlocked(ch, d); + cache_put(ch, d); + } } else spin_unlock(&cache_list_lock); @@ -346,11 +461,11 @@ static int cache_clean(void) /* * We want to regularly clean the cache, so we need to schedule some work ... */ -static void do_cache_clean(void *data) +static void do_cache_clean(struct work_struct *work) { int delay = 5; if (cache_clean() == -1) - delay = 30*HZ; + delay = round_jiffies_relative(30*HZ); if (list_empty(&cache_list)) delay = 0; @@ -360,9 +475,9 @@ static void do_cache_clean(void *data) } -/* +/* * Clean all caches promptly. This just calls cache_clean - * repeatedly until we are sure that every cache has had a chance to + * repeatedly until we are sure that every cache has had a chance to * be fully cleaned */ void cache_flush(void) @@ -372,15 +487,16 @@ void cache_flush(void) while (cache_clean() != -1) cond_resched(); } +EXPORT_SYMBOL_GPL(cache_flush); void cache_purge(struct cache_detail *detail) { detail->flush_time = LONG_MAX; - detail->nextcheck = get_seconds(); + detail->nextcheck = seconds_since_boot(); cache_flush(); detail->flush_time = 1; } - +EXPORT_SYMBOL_GPL(cache_purge); /* @@ -391,7 +507,7 @@ void cache_purge(struct cache_detail *detail) * All deferred requests are stored in a hash table, * indexed by "struct cache_head *". * As it may be wasteful to store a whole request - * structure, we allow the request to provide a + * structure, we allow the request to provide a * deferred form, which must contain a * 'struct cache_deferred_req' * This cache_deferred_req contains a method to allow @@ -405,82 +521,157 @@ void cache_purge(struct cache_detail *detail) static DEFINE_SPINLOCK(cache_defer_lock); static LIST_HEAD(cache_defer_list); -static struct list_head cache_defer_hash[DFR_HASHSIZE]; +static struct hlist_head cache_defer_hash[DFR_HASHSIZE]; static int cache_defer_cnt; -static void cache_defer_req(struct cache_req *req, struct cache_head *item) +static void __unhash_deferred_req(struct cache_deferred_req *dreq) +{ + hlist_del_init(&dreq->hash); + if (!list_empty(&dreq->recent)) { + list_del_init(&dreq->recent); + cache_defer_cnt--; + } +} + +static void __hash_deferred_req(struct cache_deferred_req *dreq, struct cache_head *item) { - struct cache_deferred_req *dreq; int hash = DFR_HASH(item); - dreq = req->defer(req); - if (dreq == NULL) - return; + INIT_LIST_HEAD(&dreq->recent); + hlist_add_head(&dreq->hash, &cache_defer_hash[hash]); +} + +static void setup_deferral(struct cache_deferred_req *dreq, + struct cache_head *item, + int count_me) +{ dreq->item = item; - dreq->recv_time = get_seconds(); spin_lock(&cache_defer_lock); - list_add(&dreq->recent, &cache_defer_list); + __hash_deferred_req(dreq, item); + + if (count_me) { + cache_defer_cnt++; + list_add(&dreq->recent, &cache_defer_list); + } + + spin_unlock(&cache_defer_lock); - if (cache_defer_hash[hash].next == NULL) - INIT_LIST_HEAD(&cache_defer_hash[hash]); - list_add(&dreq->hash, &cache_defer_hash[hash]); +} - /* it is in, now maybe clean up */ - dreq = NULL; - if (++cache_defer_cnt > DFR_MAX) { - /* too much in the cache, randomly drop - * first or last +struct thread_deferred_req { + struct cache_deferred_req handle; + struct completion completion; +}; + +static void cache_restart_thread(struct cache_deferred_req *dreq, int too_many) +{ + struct thread_deferred_req *dr = + container_of(dreq, struct thread_deferred_req, handle); + complete(&dr->completion); +} + +static void cache_wait_req(struct cache_req *req, struct cache_head *item) +{ + struct thread_deferred_req sleeper; + struct cache_deferred_req *dreq = &sleeper.handle; + + sleeper.completion = COMPLETION_INITIALIZER_ONSTACK(sleeper.completion); + dreq->revisit = cache_restart_thread; + + setup_deferral(dreq, item, 0); + + if (!test_bit(CACHE_PENDING, &item->flags) || + wait_for_completion_interruptible_timeout( + &sleeper.completion, req->thread_wait) <= 0) { + /* The completion wasn't completed, so we need + * to clean up */ - if (net_random()&1) - dreq = list_entry(cache_defer_list.next, - struct cache_deferred_req, - recent); + spin_lock(&cache_defer_lock); + if (!hlist_unhashed(&sleeper.handle.hash)) { + __unhash_deferred_req(&sleeper.handle); + spin_unlock(&cache_defer_lock); + } else { + /* cache_revisit_request already removed + * this from the hash table, but hasn't + * called ->revisit yet. It will very soon + * and we need to wait for it. + */ + spin_unlock(&cache_defer_lock); + wait_for_completion(&sleeper.completion); + } + } +} + +static void cache_limit_defers(void) +{ + /* Make sure we haven't exceed the limit of allowed deferred + * requests. + */ + struct cache_deferred_req *discard = NULL; + + if (cache_defer_cnt <= DFR_MAX) + return; + + spin_lock(&cache_defer_lock); + + /* Consider removing either the first or the last */ + if (cache_defer_cnt > DFR_MAX) { + if (prandom_u32() & 1) + discard = list_entry(cache_defer_list.next, + struct cache_deferred_req, recent); else - dreq = list_entry(cache_defer_list.prev, - struct cache_deferred_req, - recent); - list_del(&dreq->recent); - list_del(&dreq->hash); - cache_defer_cnt--; + discard = list_entry(cache_defer_list.prev, + struct cache_deferred_req, recent); + __unhash_deferred_req(discard); } spin_unlock(&cache_defer_lock); + if (discard) + discard->revisit(discard, 1); +} - if (dreq) { - /* there was one too many */ - dreq->revisit(dreq, 1); +/* Return true if and only if a deferred request is queued. */ +static bool cache_defer_req(struct cache_req *req, struct cache_head *item) +{ + struct cache_deferred_req *dreq; + + if (req->thread_wait) { + cache_wait_req(req, item); + if (!test_bit(CACHE_PENDING, &item->flags)) + return false; } - if (test_bit(CACHE_VALID, &item->flags)) { - /* must have just been validated... */ + dreq = req->defer(req); + if (dreq == NULL) + return false; + setup_deferral(dreq, item, 1); + if (!test_bit(CACHE_PENDING, &item->flags)) + /* Bit could have been cleared before we managed to + * set up the deferral, so need to revisit just in case + */ cache_revisit_request(item); - } + + cache_limit_defers(); + return true; } static void cache_revisit_request(struct cache_head *item) { struct cache_deferred_req *dreq; struct list_head pending; - - struct list_head *lp; + struct hlist_node *tmp; int hash = DFR_HASH(item); INIT_LIST_HEAD(&pending); spin_lock(&cache_defer_lock); - - lp = cache_defer_hash[hash].next; - if (lp) { - while (lp != &cache_defer_hash[hash]) { - dreq = list_entry(lp, struct cache_deferred_req, hash); - lp = lp->next; - if (dreq->item == item) { - list_del(&dreq->hash); - list_move(&dreq->recent, &pending); - cache_defer_cnt--; - } + + hlist_for_each_entry_safe(dreq, tmp, &cache_defer_hash[hash], hash) + if (dreq->item == item) { + __unhash_deferred_req(dreq); + list_add(&dreq->recent, &pending); } - } + spin_unlock(&cache_defer_lock); while (!list_empty(&pending)) { @@ -498,12 +689,11 @@ void cache_clean_deferred(void *owner) INIT_LIST_HEAD(&pending); spin_lock(&cache_defer_lock); - + list_for_each_entry_safe(dreq, tmp, &cache_defer_list, recent) { if (dreq->owner == owner) { - list_del(&dreq->hash); - list_move(&dreq->recent, &pending); - cache_defer_cnt--; + __unhash_deferred_req(dreq); + list_add(&dreq->recent, &pending); } } spin_unlock(&cache_defer_lock); @@ -518,13 +708,13 @@ void cache_clean_deferred(void *owner) /* * communicate with user-space * - * We have a magic /proc file - /proc/sunrpc/cache - * On read, you get a full request, or block - * On write, an update request is processed - * Poll works if anything to read, and always allows write + * We have a magic /proc file - /proc/sunrpc/<cachename>/channel. + * On read, you get a full request, or block. + * On write, an update request is processed. + * Poll works if anything to read, and always allows write. * - * Implemented by linked list of requests. Each open file has - * a ->private that also exists in this list. New request are added + * Implemented by linked list of requests. Each open file has + * a ->private that also exists in this list. New requests are added * to the end and may wakeup and preceding readers. * New readers are added to the head. If, on read, an item is found with * CACHE_UPCALLING clear, we free it from the list. @@ -532,7 +722,7 @@ void cache_clean_deferred(void *owner) */ static DEFINE_SPINLOCK(queue_lock); -static DECLARE_MUTEX(queue_io_sem); +static DEFINE_MUTEX(queue_io_mutex); struct cache_queue { struct list_head list; @@ -550,18 +740,30 @@ struct cache_reader { int offset; /* if non-0, we have a refcnt on next request */ }; -static ssize_t -cache_read(struct file *filp, char __user *buf, size_t count, loff_t *ppos) +static int cache_request(struct cache_detail *detail, + struct cache_request *crq) +{ + char *bp = crq->buf; + int len = PAGE_SIZE; + + detail->cache_request(detail, crq->item, &bp, &len); + if (len < 0) + return -EAGAIN; + return PAGE_SIZE - len; +} + +static ssize_t cache_read(struct file *filp, char __user *buf, size_t count, + loff_t *ppos, struct cache_detail *cd) { struct cache_reader *rp = filp->private_data; struct cache_request *rq; - struct cache_detail *cd = PDE(filp->f_dentry->d_inode)->data; + struct inode *inode = file_inode(filp); int err; if (count == 0) return 0; - down(&queue_io_sem); /* protect against multiple concurrent + mutex_lock(&inode->i_mutex); /* protect against multiple concurrent * readers on this file */ again: spin_lock(&queue_lock); @@ -574,17 +776,23 @@ cache_read(struct file *filp, char __user *buf, size_t count, loff_t *ppos) } if (rp->q.list.next == &cd->queue) { spin_unlock(&queue_lock); - up(&queue_io_sem); - if (rp->offset) - BUG(); + mutex_unlock(&inode->i_mutex); + WARN_ON_ONCE(rp->offset); return 0; } rq = container_of(rp->q.list.next, struct cache_request, q.list); - if (rq->q.reader) BUG(); + WARN_ON_ONCE(rq->q.reader); if (rp->offset == 0) rq->readers++; spin_unlock(&queue_lock); + if (rq->len == 0) { + err = cache_request(cd, rq); + if (err < 0) + goto out; + rq->len = err; + } + if (rp->offset == 0 && !test_bit(CACHE_PENDING, &rq->item->flags)) { err = -EAGAIN; spin_lock(&queue_lock); @@ -614,7 +822,7 @@ cache_read(struct file *filp, char __user *buf, size_t count, loff_t *ppos) !test_bit(CACHE_PENDING, &rq->item->flags)) { list_del(&rq->q.list); spin_unlock(&queue_lock); - cd->cache_put(rq->item, cd); + cache_put(rq->item, cd); kfree(rq->buf); kfree(rq); } else @@ -622,49 +830,92 @@ cache_read(struct file *filp, char __user *buf, size_t count, loff_t *ppos) } if (err == -EAGAIN) goto again; - up(&queue_io_sem); + mutex_unlock(&inode->i_mutex); return err ? err : count; } -static char write_buf[8192]; /* protected by queue_io_sem */ - -static ssize_t -cache_write(struct file *filp, const char __user *buf, size_t count, - loff_t *ppos) +static ssize_t cache_do_downcall(char *kaddr, const char __user *buf, + size_t count, struct cache_detail *cd) { - int err; - struct cache_detail *cd = PDE(filp->f_dentry->d_inode)->data; + ssize_t ret; if (count == 0) - return 0; - if (count >= sizeof(write_buf)) return -EINVAL; + if (copy_from_user(kaddr, buf, count)) + return -EFAULT; + kaddr[count] = '\0'; + ret = cd->cache_parse(cd, kaddr, count); + if (!ret) + ret = count; + return ret; +} - down(&queue_io_sem); +static ssize_t cache_slow_downcall(const char __user *buf, + size_t count, struct cache_detail *cd) +{ + static char write_buf[8192]; /* protected by queue_io_mutex */ + ssize_t ret = -EINVAL; - if (copy_from_user(write_buf, buf, count)) { - up(&queue_io_sem); - return -EFAULT; - } - write_buf[count] = '\0'; - if (cd->cache_parse) - err = cd->cache_parse(cd, write_buf, count); - else - err = -EINVAL; + if (count >= sizeof(write_buf)) + goto out; + mutex_lock(&queue_io_mutex); + ret = cache_do_downcall(write_buf, buf, count, cd); + mutex_unlock(&queue_io_mutex); +out: + return ret; +} + +static ssize_t cache_downcall(struct address_space *mapping, + const char __user *buf, + size_t count, struct cache_detail *cd) +{ + struct page *page; + char *kaddr; + ssize_t ret = -ENOMEM; + + if (count >= PAGE_CACHE_SIZE) + goto out_slow; + + page = find_or_create_page(mapping, 0, GFP_KERNEL); + if (!page) + goto out_slow; + + kaddr = kmap(page); + ret = cache_do_downcall(kaddr, buf, count, cd); + kunmap(page); + unlock_page(page); + page_cache_release(page); + return ret; +out_slow: + return cache_slow_downcall(buf, count, cd); +} - up(&queue_io_sem); - return err ? err : count; +static ssize_t cache_write(struct file *filp, const char __user *buf, + size_t count, loff_t *ppos, + struct cache_detail *cd) +{ + struct address_space *mapping = filp->f_mapping; + struct inode *inode = file_inode(filp); + ssize_t ret = -EINVAL; + + if (!cd->cache_parse) + goto out; + + mutex_lock(&inode->i_mutex); + ret = cache_downcall(mapping, buf, count, cd); + mutex_unlock(&inode->i_mutex); +out: + return ret; } static DECLARE_WAIT_QUEUE_HEAD(queue_wait); -static unsigned int -cache_poll(struct file *filp, poll_table *wait) +static unsigned int cache_poll(struct file *filp, poll_table *wait, + struct cache_detail *cd) { unsigned int mask; struct cache_reader *rp = filp->private_data; struct cache_queue *cq; - struct cache_detail *cd = PDE(filp->f_dentry->d_inode)->data; poll_wait(filp, &queue_wait, wait); @@ -686,14 +937,13 @@ cache_poll(struct file *filp, poll_table *wait) return mask; } -static int -cache_ioctl(struct inode *ino, struct file *filp, - unsigned int cmd, unsigned long arg) +static int cache_ioctl(struct inode *ino, struct file *filp, + unsigned int cmd, unsigned long arg, + struct cache_detail *cd) { int len = 0; struct cache_reader *rp = filp->private_data; struct cache_queue *cq; - struct cache_detail *cd = PDE(ino)->data; if (cmd != FIONREAD || !rp) return -EINVAL; @@ -716,18 +966,20 @@ cache_ioctl(struct inode *ino, struct file *filp, return put_user(len, (int __user *)arg); } -static int -cache_open(struct inode *inode, struct file *filp) +static int cache_open(struct inode *inode, struct file *filp, + struct cache_detail *cd) { struct cache_reader *rp = NULL; + if (!cd || !try_module_get(cd->owner)) + return -EACCES; nonseekable_open(inode, filp); if (filp->f_mode & FMODE_READ) { - struct cache_detail *cd = PDE(inode)->data; - rp = kmalloc(sizeof(*rp), GFP_KERNEL); - if (!rp) + if (!rp) { + module_put(cd->owner); return -ENOMEM; + } rp->offset = 0; rp->q.reader = 1; atomic_inc(&cd->readers); @@ -739,11 +991,10 @@ cache_open(struct inode *inode, struct file *filp) return 0; } -static int -cache_release(struct inode *inode, struct file *filp) +static int cache_release(struct inode *inode, struct file *filp, + struct cache_detail *cd) { struct cache_reader *rp = filp->private_data; - struct cache_detail *cd = PDE(inode)->data; if (rp) { spin_lock(&queue_lock); @@ -764,45 +1015,43 @@ cache_release(struct inode *inode, struct file *filp) filp->private_data = NULL; kfree(rp); - cd->last_close = get_seconds(); + cd->last_close = seconds_since_boot(); atomic_dec(&cd->readers); } + module_put(cd->owner); return 0; } -static struct file_operations cache_file_operations = { - .owner = THIS_MODULE, - .llseek = no_llseek, - .read = cache_read, - .write = cache_write, - .poll = cache_poll, - .ioctl = cache_ioctl, /* for FIONREAD */ - .open = cache_open, - .release = cache_release, -}; - - -static void queue_loose(struct cache_detail *detail, struct cache_head *ch) +static void cache_dequeue(struct cache_detail *detail, struct cache_head *ch) { - struct cache_queue *cq; + struct cache_queue *cq, *tmp; + struct cache_request *cr; + struct list_head dequeued; + + INIT_LIST_HEAD(&dequeued); spin_lock(&queue_lock); - list_for_each_entry(cq, &detail->queue, list) + list_for_each_entry_safe(cq, tmp, &detail->queue, list) if (!cq->reader) { - struct cache_request *cr = container_of(cq, struct cache_request, q); + cr = container_of(cq, struct cache_request, q); if (cr->item != ch) continue; - if (cr->readers != 0) + if (test_bit(CACHE_PENDING, &ch->flags)) + /* Lost a race and it is pending again */ break; - list_del(&cr->q.list); - spin_unlock(&queue_lock); - detail->cache_put(cr->item, detail); - kfree(cr->buf); - kfree(cr); - return; + if (cr->readers != 0) + continue; + list_move(&cr->q.list, &dequeued); } spin_unlock(&queue_lock); + while (!list_empty(&dequeued)) { + cr = list_entry(dequeued.next, struct cache_request, q.list); + list_del(&cr->q.list); + cache_put(cr->item, detail); + kfree(cr->buf); + kfree(cr); + } } /* @@ -848,6 +1097,7 @@ void qword_add(char **bpp, int *lp, char *str) *bpp = bp; *lp = len; } +EXPORT_SYMBOL_GPL(qword_add); void qword_addhex(char **bpp, int *lp, char *buf, int blen) { @@ -861,9 +1111,7 @@ void qword_addhex(char **bpp, int *lp, char *buf, int blen) *bp++ = 'x'; len -= 2; while (blen && len >= 2) { - unsigned char c = *buf++; - *bp++ = '0' + ((c&0xf0)>>4) + (c>=0xa0)*('a'-'9'-1); - *bp++ = '0' + (c&0x0f) + ((c&0x0f)>=0x0a)*('a'-'9'-1); + bp = hex_byte_pack(bp, *buf++); len -= 2; blen--; } @@ -876,36 +1124,57 @@ void qword_addhex(char **bpp, int *lp, char *buf, int blen) *bpp = bp; *lp = len; } +EXPORT_SYMBOL_GPL(qword_addhex); static void warn_no_listener(struct cache_detail *detail) { if (detail->last_warn != detail->last_close) { detail->last_warn = detail->last_close; if (detail->warn_no_listener) - detail->warn_no_listener(detail); + detail->warn_no_listener(detail, detail->last_close != 0); } } +static bool cache_listeners_exist(struct cache_detail *detail) +{ + if (atomic_read(&detail->readers)) + return true; + if (detail->last_close == 0) + /* This cache was never opened */ + return false; + if (detail->last_close < seconds_since_boot() - 30) + /* + * We allow for the possibility that someone might + * restart a userspace daemon without restarting the + * server; but after 30 seconds, we give up. + */ + return false; + return true; +} + /* - * register an upcall request to user-space. + * register an upcall request to user-space and queue it up for read() by the + * upcall daemon. + * * Each request is at most one page long. */ -static int cache_make_upcall(struct cache_detail *detail, struct cache_head *h) +int sunrpc_cache_pipe_upcall(struct cache_detail *detail, struct cache_head *h) { char *buf; struct cache_request *crq; - char *bp; - int len; + int ret = 0; - if (detail->cache_request == NULL) + if (!detail->cache_request) return -EINVAL; - if (atomic_read(&detail->readers) == 0 && - detail->last_close < get_seconds() - 30) { - warn_no_listener(detail); - return -EINVAL; + if (!cache_listeners_exist(detail)) { + warn_no_listener(detail); + return -EINVAL; } + if (test_bit(CACHE_CLEANED, &h->flags)) + /* Too late to make an upcall */ + return -EAGAIN; buf = kmalloc(PAGE_SIZE, GFP_KERNEL); if (!buf) @@ -917,26 +1186,26 @@ static int cache_make_upcall(struct cache_detail *detail, struct cache_head *h) return -EAGAIN; } - bp = buf; len = PAGE_SIZE; - - detail->cache_request(detail, h, &bp, &len); - - if (len < 0) { - kfree(buf); - kfree(crq); - return -EAGAIN; - } crq->q.reader = 0; crq->item = cache_get(h); crq->buf = buf; - crq->len = PAGE_SIZE - len; + crq->len = 0; crq->readers = 0; spin_lock(&queue_lock); - list_add_tail(&crq->q.list, &detail->queue); + if (test_bit(CACHE_PENDING, &h->flags)) + list_add_tail(&crq->q.list, &detail->queue); + else + /* Lost a race, no longer PENDING, so don't enqueue */ + ret = -EAGAIN; spin_unlock(&queue_lock); wake_up(&queue_wait); - return 0; + if (ret == -EAGAIN) { + kfree(buf); + kfree(crq); + } + return ret; } +EXPORT_SYMBOL_GPL(sunrpc_cache_pipe_upcall); /* * parse a message from user-space and pass it @@ -944,13 +1213,12 @@ static int cache_make_upcall(struct cache_detail *detail, struct cache_head *h) * Messages are, like requests, separated into fields by * spaces and dequotes as \xHEXSTRING or embedded \nnn octal * - * Message is + * Message is * reply cachename expiry key ... content.... * - * key and content are both parsed by cache + * key and content are both parsed by cache */ -#define isodigit(c) (isdigit(c) && c <= '7') int qword_get(char **bpp, char *dest, int bufsize) { /* return bytes copied, or -1 on error */ @@ -962,13 +1230,19 @@ int qword_get(char **bpp, char *dest, int bufsize) if (bp[0] == '\\' && bp[1] == 'x') { /* HEX STRING */ bp += 2; - while (isxdigit(bp[0]) && isxdigit(bp[1]) && len < bufsize) { - int byte = isdigit(*bp) ? *bp-'0' : toupper(*bp)-'A'+10; - bp++; - byte <<= 4; - byte |= isdigit(*bp) ? *bp-'0' : toupper(*bp)-'A'+10; - *dest++ = byte; - bp++; + while (len < bufsize) { + int h, l; + + h = hex_to_bin(bp[0]); + if (h < 0) + break; + + l = hex_to_bin(bp[1]); + if (l < 0) + break; + + *dest++ = (h << 4) | l; + bp += 2; len++; } } else { @@ -998,6 +1272,7 @@ int qword_get(char **bpp, char *dest, int bufsize) *dest = '\0'; return len; } +EXPORT_SYMBOL_GPL(qword_get); /* @@ -1012,12 +1287,13 @@ struct handle { }; static void *c_start(struct seq_file *m, loff_t *pos) + __acquires(cd->hash_lock) { loff_t n = *pos; - unsigned hash, entry; + unsigned int hash, entry; struct cache_head *ch; struct cache_detail *cd = ((struct handle*)m->private)->cd; - + read_lock(&cd->hash_lock); if (!n--) @@ -1032,7 +1308,7 @@ static void *c_start(struct seq_file *m, loff_t *pos) do { hash++; n += 1LL<<32; - } while(hash < cd->hash_size && + } while(hash < cd->hash_size && cd->hash_table[hash]==NULL); if (hash >= cd->hash_size) return NULL; @@ -1068,6 +1344,7 @@ static void *c_next(struct seq_file *m, void *p, loff_t *pos) } static void c_stop(struct seq_file *m, void *p) + __releases(cd->hash_lock) { struct cache_detail *cd = ((struct handle*)m->private)->cd; read_unlock(&cd->hash_lock); @@ -1082,108 +1359,480 @@ static int c_show(struct seq_file *m, void *p) return cd->cache_show(m, cd, NULL); ifdebug(CACHE) - seq_printf(m, "# expiry=%ld refcnt=%d\n", - cp->expiry_time, atomic_read(&cp->refcnt)); + seq_printf(m, "# expiry=%ld refcnt=%d flags=%lx\n", + convert_to_wallclock(cp->expiry_time), + atomic_read(&cp->ref.refcount), cp->flags); cache_get(cp); if (cache_check(cd, cp, NULL)) /* cache_check does a cache_put on failure */ seq_printf(m, "# "); - else + else { + if (cache_is_expired(cd, cp)) + seq_printf(m, "# "); cache_put(cp, cd); + } return cd->cache_show(m, cd, cp); } -static struct seq_operations cache_content_op = { +static const struct seq_operations cache_content_op = { .start = c_start, .next = c_next, .stop = c_stop, .show = c_show, }; -static int content_open(struct inode *inode, struct file *file) +static int content_open(struct inode *inode, struct file *file, + struct cache_detail *cd) { - int res; struct handle *han; - struct cache_detail *cd = PDE(inode)->data; - han = kmalloc(sizeof(*han), GFP_KERNEL); - if (han == NULL) + if (!cd || !try_module_get(cd->owner)) + return -EACCES; + han = __seq_open_private(file, &cache_content_op, sizeof(*han)); + if (han == NULL) { + module_put(cd->owner); return -ENOMEM; + } han->cd = cd; + return 0; +} - res = seq_open(file, &cache_content_op); - if (res) - kfree(han); - else - ((struct seq_file *)file->private_data)->private = han; - - return res; +static int content_release(struct inode *inode, struct file *file, + struct cache_detail *cd) +{ + int ret = seq_release_private(inode, file); + module_put(cd->owner); + return ret; } -static int content_release(struct inode *inode, struct file *file) + +static int open_flush(struct inode *inode, struct file *file, + struct cache_detail *cd) { - struct seq_file *m = (struct seq_file *)file->private_data; - struct handle *han = m->private; - kfree(han); - m->private = NULL; - return seq_release(inode, file); + if (!cd || !try_module_get(cd->owner)) + return -EACCES; + return nonseekable_open(inode, file); } -static struct file_operations content_file_operations = { - .open = content_open, - .read = seq_read, - .llseek = seq_lseek, - .release = content_release, -}; +static int release_flush(struct inode *inode, struct file *file, + struct cache_detail *cd) +{ + module_put(cd->owner); + return 0; +} static ssize_t read_flush(struct file *file, char __user *buf, - size_t count, loff_t *ppos) + size_t count, loff_t *ppos, + struct cache_detail *cd) { - struct cache_detail *cd = PDE(file->f_dentry->d_inode)->data; - char tbuf[20]; + char tbuf[22]; unsigned long p = *ppos; - int len; + size_t len; - sprintf(tbuf, "%lu\n", cd->flush_time); + snprintf(tbuf, sizeof(tbuf), "%lu\n", convert_to_wallclock(cd->flush_time)); len = strlen(tbuf); if (p >= len) return 0; len -= p; - if (len > count) len = count; + if (len > count) + len = count; if (copy_to_user(buf, (void*)(tbuf+p), len)) - len = -EFAULT; - else - *ppos += len; + return -EFAULT; + *ppos += len; return len; } -static ssize_t write_flush(struct file * file, const char __user * buf, - size_t count, loff_t *ppos) +static ssize_t write_flush(struct file *file, const char __user *buf, + size_t count, loff_t *ppos, + struct cache_detail *cd) { - struct cache_detail *cd = PDE(file->f_dentry->d_inode)->data; char tbuf[20]; - char *ep; - long flushtime; + char *bp, *ep; + if (*ppos || count > sizeof(tbuf)-1) return -EINVAL; if (copy_from_user(tbuf, buf, count)) return -EFAULT; tbuf[count] = 0; - flushtime = simple_strtoul(tbuf, &ep, 0); + simple_strtoul(tbuf, &ep, 0); if (*ep && *ep != '\n') return -EINVAL; - cd->flush_time = flushtime; - cd->nextcheck = get_seconds(); + bp = tbuf; + cd->flush_time = get_expiry(&bp); + cd->nextcheck = seconds_since_boot(); cache_flush(); *ppos += count; return count; } -static struct file_operations cache_flush_operations = { - .open = nonseekable_open, - .read = read_flush, - .write = write_flush, +static ssize_t cache_read_procfs(struct file *filp, char __user *buf, + size_t count, loff_t *ppos) +{ + struct cache_detail *cd = PDE_DATA(file_inode(filp)); + + return cache_read(filp, buf, count, ppos, cd); +} + +static ssize_t cache_write_procfs(struct file *filp, const char __user *buf, + size_t count, loff_t *ppos) +{ + struct cache_detail *cd = PDE_DATA(file_inode(filp)); + + return cache_write(filp, buf, count, ppos, cd); +} + +static unsigned int cache_poll_procfs(struct file *filp, poll_table *wait) +{ + struct cache_detail *cd = PDE_DATA(file_inode(filp)); + + return cache_poll(filp, wait, cd); +} + +static long cache_ioctl_procfs(struct file *filp, + unsigned int cmd, unsigned long arg) +{ + struct inode *inode = file_inode(filp); + struct cache_detail *cd = PDE_DATA(inode); + + return cache_ioctl(inode, filp, cmd, arg, cd); +} + +static int cache_open_procfs(struct inode *inode, struct file *filp) +{ + struct cache_detail *cd = PDE_DATA(inode); + + return cache_open(inode, filp, cd); +} + +static int cache_release_procfs(struct inode *inode, struct file *filp) +{ + struct cache_detail *cd = PDE_DATA(inode); + + return cache_release(inode, filp, cd); +} + +static const struct file_operations cache_file_operations_procfs = { + .owner = THIS_MODULE, + .llseek = no_llseek, + .read = cache_read_procfs, + .write = cache_write_procfs, + .poll = cache_poll_procfs, + .unlocked_ioctl = cache_ioctl_procfs, /* for FIONREAD */ + .open = cache_open_procfs, + .release = cache_release_procfs, }; + +static int content_open_procfs(struct inode *inode, struct file *filp) +{ + struct cache_detail *cd = PDE_DATA(inode); + + return content_open(inode, filp, cd); +} + +static int content_release_procfs(struct inode *inode, struct file *filp) +{ + struct cache_detail *cd = PDE_DATA(inode); + + return content_release(inode, filp, cd); +} + +static const struct file_operations content_file_operations_procfs = { + .open = content_open_procfs, + .read = seq_read, + .llseek = seq_lseek, + .release = content_release_procfs, +}; + +static int open_flush_procfs(struct inode *inode, struct file *filp) +{ + struct cache_detail *cd = PDE_DATA(inode); + + return open_flush(inode, filp, cd); +} + +static int release_flush_procfs(struct inode *inode, struct file *filp) +{ + struct cache_detail *cd = PDE_DATA(inode); + + return release_flush(inode, filp, cd); +} + +static ssize_t read_flush_procfs(struct file *filp, char __user *buf, + size_t count, loff_t *ppos) +{ + struct cache_detail *cd = PDE_DATA(file_inode(filp)); + + return read_flush(filp, buf, count, ppos, cd); +} + +static ssize_t write_flush_procfs(struct file *filp, + const char __user *buf, + size_t count, loff_t *ppos) +{ + struct cache_detail *cd = PDE_DATA(file_inode(filp)); + + return write_flush(filp, buf, count, ppos, cd); +} + +static const struct file_operations cache_flush_operations_procfs = { + .open = open_flush_procfs, + .read = read_flush_procfs, + .write = write_flush_procfs, + .release = release_flush_procfs, + .llseek = no_llseek, +}; + +static void remove_cache_proc_entries(struct cache_detail *cd, struct net *net) +{ + struct sunrpc_net *sn; + + if (cd->u.procfs.proc_ent == NULL) + return; + if (cd->u.procfs.flush_ent) + remove_proc_entry("flush", cd->u.procfs.proc_ent); + if (cd->u.procfs.channel_ent) + remove_proc_entry("channel", cd->u.procfs.proc_ent); + if (cd->u.procfs.content_ent) + remove_proc_entry("content", cd->u.procfs.proc_ent); + cd->u.procfs.proc_ent = NULL; + sn = net_generic(net, sunrpc_net_id); + remove_proc_entry(cd->name, sn->proc_net_rpc); +} + +#ifdef CONFIG_PROC_FS +static int create_cache_proc_entries(struct cache_detail *cd, struct net *net) +{ + struct proc_dir_entry *p; + struct sunrpc_net *sn; + + sn = net_generic(net, sunrpc_net_id); + cd->u.procfs.proc_ent = proc_mkdir(cd->name, sn->proc_net_rpc); + if (cd->u.procfs.proc_ent == NULL) + goto out_nomem; + cd->u.procfs.channel_ent = NULL; + cd->u.procfs.content_ent = NULL; + + p = proc_create_data("flush", S_IFREG|S_IRUSR|S_IWUSR, + cd->u.procfs.proc_ent, + &cache_flush_operations_procfs, cd); + cd->u.procfs.flush_ent = p; + if (p == NULL) + goto out_nomem; + + if (cd->cache_request || cd->cache_parse) { + p = proc_create_data("channel", S_IFREG|S_IRUSR|S_IWUSR, + cd->u.procfs.proc_ent, + &cache_file_operations_procfs, cd); + cd->u.procfs.channel_ent = p; + if (p == NULL) + goto out_nomem; + } + if (cd->cache_show) { + p = proc_create_data("content", S_IFREG|S_IRUSR, + cd->u.procfs.proc_ent, + &content_file_operations_procfs, cd); + cd->u.procfs.content_ent = p; + if (p == NULL) + goto out_nomem; + } + return 0; +out_nomem: + remove_cache_proc_entries(cd, net); + return -ENOMEM; +} +#else /* CONFIG_PROC_FS */ +static int create_cache_proc_entries(struct cache_detail *cd, struct net *net) +{ + return 0; +} +#endif + +void __init cache_initialize(void) +{ + INIT_DEFERRABLE_WORK(&cache_cleaner, do_cache_clean); +} + +int cache_register_net(struct cache_detail *cd, struct net *net) +{ + int ret; + + sunrpc_init_cache_detail(cd); + ret = create_cache_proc_entries(cd, net); + if (ret) + sunrpc_destroy_cache_detail(cd); + return ret; +} +EXPORT_SYMBOL_GPL(cache_register_net); + +void cache_unregister_net(struct cache_detail *cd, struct net *net) +{ + remove_cache_proc_entries(cd, net); + sunrpc_destroy_cache_detail(cd); +} +EXPORT_SYMBOL_GPL(cache_unregister_net); + +struct cache_detail *cache_create_net(struct cache_detail *tmpl, struct net *net) +{ + struct cache_detail *cd; + + cd = kmemdup(tmpl, sizeof(struct cache_detail), GFP_KERNEL); + if (cd == NULL) + return ERR_PTR(-ENOMEM); + + cd->hash_table = kzalloc(cd->hash_size * sizeof(struct cache_head *), + GFP_KERNEL); + if (cd->hash_table == NULL) { + kfree(cd); + return ERR_PTR(-ENOMEM); + } + cd->net = net; + return cd; +} +EXPORT_SYMBOL_GPL(cache_create_net); + +void cache_destroy_net(struct cache_detail *cd, struct net *net) +{ + kfree(cd->hash_table); + kfree(cd); +} +EXPORT_SYMBOL_GPL(cache_destroy_net); + +static ssize_t cache_read_pipefs(struct file *filp, char __user *buf, + size_t count, loff_t *ppos) +{ + struct cache_detail *cd = RPC_I(file_inode(filp))->private; + + return cache_read(filp, buf, count, ppos, cd); +} + +static ssize_t cache_write_pipefs(struct file *filp, const char __user *buf, + size_t count, loff_t *ppos) +{ + struct cache_detail *cd = RPC_I(file_inode(filp))->private; + + return cache_write(filp, buf, count, ppos, cd); +} + +static unsigned int cache_poll_pipefs(struct file *filp, poll_table *wait) +{ + struct cache_detail *cd = RPC_I(file_inode(filp))->private; + + return cache_poll(filp, wait, cd); +} + +static long cache_ioctl_pipefs(struct file *filp, + unsigned int cmd, unsigned long arg) +{ + struct inode *inode = file_inode(filp); + struct cache_detail *cd = RPC_I(inode)->private; + + return cache_ioctl(inode, filp, cmd, arg, cd); +} + +static int cache_open_pipefs(struct inode *inode, struct file *filp) +{ + struct cache_detail *cd = RPC_I(inode)->private; + + return cache_open(inode, filp, cd); +} + +static int cache_release_pipefs(struct inode *inode, struct file *filp) +{ + struct cache_detail *cd = RPC_I(inode)->private; + + return cache_release(inode, filp, cd); +} + +const struct file_operations cache_file_operations_pipefs = { + .owner = THIS_MODULE, + .llseek = no_llseek, + .read = cache_read_pipefs, + .write = cache_write_pipefs, + .poll = cache_poll_pipefs, + .unlocked_ioctl = cache_ioctl_pipefs, /* for FIONREAD */ + .open = cache_open_pipefs, + .release = cache_release_pipefs, +}; + +static int content_open_pipefs(struct inode *inode, struct file *filp) +{ + struct cache_detail *cd = RPC_I(inode)->private; + + return content_open(inode, filp, cd); +} + +static int content_release_pipefs(struct inode *inode, struct file *filp) +{ + struct cache_detail *cd = RPC_I(inode)->private; + + return content_release(inode, filp, cd); +} + +const struct file_operations content_file_operations_pipefs = { + .open = content_open_pipefs, + .read = seq_read, + .llseek = seq_lseek, + .release = content_release_pipefs, +}; + +static int open_flush_pipefs(struct inode *inode, struct file *filp) +{ + struct cache_detail *cd = RPC_I(inode)->private; + + return open_flush(inode, filp, cd); +} + +static int release_flush_pipefs(struct inode *inode, struct file *filp) +{ + struct cache_detail *cd = RPC_I(inode)->private; + + return release_flush(inode, filp, cd); +} + +static ssize_t read_flush_pipefs(struct file *filp, char __user *buf, + size_t count, loff_t *ppos) +{ + struct cache_detail *cd = RPC_I(file_inode(filp))->private; + + return read_flush(filp, buf, count, ppos, cd); +} + +static ssize_t write_flush_pipefs(struct file *filp, + const char __user *buf, + size_t count, loff_t *ppos) +{ + struct cache_detail *cd = RPC_I(file_inode(filp))->private; + + return write_flush(filp, buf, count, ppos, cd); +} + +const struct file_operations cache_flush_operations_pipefs = { + .open = open_flush_pipefs, + .read = read_flush_pipefs, + .write = write_flush_pipefs, + .release = release_flush_pipefs, + .llseek = no_llseek, +}; + +int sunrpc_cache_register_pipefs(struct dentry *parent, + const char *name, umode_t umode, + struct cache_detail *cd) +{ + struct dentry *dir = rpc_create_cache_dir(parent, name, umode, cd); + if (IS_ERR(dir)) + return PTR_ERR(dir); + cd->u.pipefs.dir = dir; + return 0; +} +EXPORT_SYMBOL_GPL(sunrpc_cache_register_pipefs); + +void sunrpc_cache_unregister_pipefs(struct cache_detail *cd) +{ + rpc_remove_cache_dir(cd->u.pipefs.dir); + cd->u.pipefs.dir = NULL; +} +EXPORT_SYMBOL_GPL(sunrpc_cache_unregister_pipefs); + diff --git a/net/sunrpc/clnt.c b/net/sunrpc/clnt.c index 702ede309b0..2e6ab10734f 100644 --- a/net/sunrpc/clnt.c +++ b/net/sunrpc/clnt.c @@ -13,35 +13,47 @@ * and need to be refreshed, or when a packet was damaged in transit. * This may be have to be moved to the VFS layer. * - * NB: BSD uses a more intelligent approach to guessing when a request - * or reply has been lost by keeping the RTO estimate for each procedure. - * We currently make do with a constant timeout value. - * * Copyright (C) 1992,1993 Rick Sladkey <jrs@world.std.com> * Copyright (C) 1995,1996 Olaf Kirch <okir@monad.swb.de> */ -#include <asm/system.h> #include <linux/module.h> #include <linux/types.h> +#include <linux/kallsyms.h> #include <linux/mm.h> +#include <linux/namei.h> +#include <linux/mount.h> #include <linux/slab.h> +#include <linux/rcupdate.h> #include <linux/utsname.h> +#include <linux/workqueue.h> +#include <linux/in.h> +#include <linux/in6.h> +#include <linux/un.h> #include <linux/sunrpc/clnt.h> -#include <linux/workqueue.h> +#include <linux/sunrpc/addr.h> #include <linux/sunrpc/rpc_pipe_fs.h> +#include <linux/sunrpc/metrics.h> +#include <linux/sunrpc/bc_xprt.h> +#include <trace/events/sunrpc.h> -#include <linux/nfs.h> - - -#define RPC_SLACK_SPACE (1024) /* total overkill */ +#include "sunrpc.h" +#include "netns.h" #ifdef RPC_DEBUG # define RPCDBG_FACILITY RPCDBG_CALL #endif +#define dprint_status(t) \ + dprintk("RPC: %5u %s (status %d)\n", t->tk_pid, \ + __func__, t->tk_status) + +/* + * All RPC clients are linked into this list + */ + static DECLARE_WAIT_QUEUE_HEAD(destroy_wait); @@ -49,480 +61,1259 @@ static void call_start(struct rpc_task *task); static void call_reserve(struct rpc_task *task); static void call_reserveresult(struct rpc_task *task); static void call_allocate(struct rpc_task *task); -static void call_encode(struct rpc_task *task); static void call_decode(struct rpc_task *task); static void call_bind(struct rpc_task *task); static void call_bind_status(struct rpc_task *task); static void call_transmit(struct rpc_task *task); +#if defined(CONFIG_SUNRPC_BACKCHANNEL) +static void call_bc_transmit(struct rpc_task *task); +#endif /* CONFIG_SUNRPC_BACKCHANNEL */ static void call_status(struct rpc_task *task); +static void call_transmit_status(struct rpc_task *task); static void call_refresh(struct rpc_task *task); static void call_refreshresult(struct rpc_task *task); static void call_timeout(struct rpc_task *task); static void call_connect(struct rpc_task *task); static void call_connect_status(struct rpc_task *task); -static u32 * call_header(struct rpc_task *task); -static u32 * call_verify(struct rpc_task *task); +static __be32 *rpc_encode_header(struct rpc_task *task); +static __be32 *rpc_verify_header(struct rpc_task *task); +static int rpc_ping(struct rpc_clnt *clnt); -static int -rpc_setup_pipedir(struct rpc_clnt *clnt, char *dir_name) +static void rpc_register_client(struct rpc_clnt *clnt) { - static uint32_t clntid; - int error; + struct net *net = rpc_net_ns(clnt); + struct sunrpc_net *sn = net_generic(net, sunrpc_net_id); - if (dir_name == NULL) - return 0; + spin_lock(&sn->rpc_client_lock); + list_add(&clnt->cl_clients, &sn->all_clients); + spin_unlock(&sn->rpc_client_lock); +} + +static void rpc_unregister_client(struct rpc_clnt *clnt) +{ + struct net *net = rpc_net_ns(clnt); + struct sunrpc_net *sn = net_generic(net, sunrpc_net_id); + + spin_lock(&sn->rpc_client_lock); + list_del(&clnt->cl_clients); + spin_unlock(&sn->rpc_client_lock); +} + +static void __rpc_clnt_remove_pipedir(struct rpc_clnt *clnt) +{ + rpc_remove_client_dir(clnt); +} + +static void rpc_clnt_remove_pipedir(struct rpc_clnt *clnt) +{ + struct net *net = rpc_net_ns(clnt); + struct super_block *pipefs_sb; + + pipefs_sb = rpc_get_sb_net(net); + if (pipefs_sb) { + __rpc_clnt_remove_pipedir(clnt); + rpc_put_sb_net(net); + } +} + +static struct dentry *rpc_setup_pipedir_sb(struct super_block *sb, + struct rpc_clnt *clnt) +{ + static uint32_t clntid; + const char *dir_name = clnt->cl_program->pipe_dir_name; + char name[15]; + struct dentry *dir, *dentry; + + dir = rpc_d_lookup_sb(sb, dir_name); + if (dir == NULL) { + pr_info("RPC: pipefs directory doesn't exist: %s\n", dir_name); + return dir; + } for (;;) { - snprintf(clnt->cl_pathname, sizeof(clnt->cl_pathname), - "%s/clnt%x", dir_name, - (unsigned int)clntid++); - clnt->cl_pathname[sizeof(clnt->cl_pathname) - 1] = '\0'; - clnt->cl_dentry = rpc_mkdir(clnt->cl_pathname, clnt); - if (!IS_ERR(clnt->cl_dentry)) - return 0; - error = PTR_ERR(clnt->cl_dentry); - if (error != -EEXIST) { - printk(KERN_INFO "RPC: Couldn't create pipefs entry %s, error %d\n", - clnt->cl_pathname, error); - return error; - } + snprintf(name, sizeof(name), "clnt%x", (unsigned int)clntid++); + name[sizeof(name) - 1] = '\0'; + dentry = rpc_create_client_dir(dir, name, clnt); + if (!IS_ERR(dentry)) + break; + if (dentry == ERR_PTR(-EEXIST)) + continue; + printk(KERN_INFO "RPC: Couldn't create pipefs entry" + " %s/%s, error %ld\n", + dir_name, name, PTR_ERR(dentry)); + break; } + dput(dir); + return dentry; } -/* - * Create an RPC client - * FIXME: This should also take a flags argument (as in task->tk_flags). - * It's called (among others) from pmap_create_client, which may in - * turn be called by an async task. In this case, rpciod should not be - * made to sleep too long. - */ -struct rpc_clnt * -rpc_new_client(struct rpc_xprt *xprt, char *servname, - struct rpc_program *program, u32 vers, - rpc_authflavor_t flavor) +static int +rpc_setup_pipedir(struct super_block *pipefs_sb, struct rpc_clnt *clnt) +{ + struct dentry *dentry; + + if (clnt->cl_program->pipe_dir_name != NULL) { + dentry = rpc_setup_pipedir_sb(pipefs_sb, clnt); + if (IS_ERR(dentry)) + return PTR_ERR(dentry); + } + return 0; +} + +static int rpc_clnt_skip_event(struct rpc_clnt *clnt, unsigned long event) +{ + if (clnt->cl_program->pipe_dir_name == NULL) + return 1; + + switch (event) { + case RPC_PIPEFS_MOUNT: + if (clnt->cl_pipedir_objects.pdh_dentry != NULL) + return 1; + if (atomic_read(&clnt->cl_count) == 0) + return 1; + break; + case RPC_PIPEFS_UMOUNT: + if (clnt->cl_pipedir_objects.pdh_dentry == NULL) + return 1; + break; + } + return 0; +} + +static int __rpc_clnt_handle_event(struct rpc_clnt *clnt, unsigned long event, + struct super_block *sb) +{ + struct dentry *dentry; + int err = 0; + + switch (event) { + case RPC_PIPEFS_MOUNT: + dentry = rpc_setup_pipedir_sb(sb, clnt); + if (!dentry) + return -ENOENT; + if (IS_ERR(dentry)) + return PTR_ERR(dentry); + break; + case RPC_PIPEFS_UMOUNT: + __rpc_clnt_remove_pipedir(clnt); + break; + default: + printk(KERN_ERR "%s: unknown event: %ld\n", __func__, event); + return -ENOTSUPP; + } + return err; +} + +static int __rpc_pipefs_event(struct rpc_clnt *clnt, unsigned long event, + struct super_block *sb) +{ + int error = 0; + + for (;; clnt = clnt->cl_parent) { + if (!rpc_clnt_skip_event(clnt, event)) + error = __rpc_clnt_handle_event(clnt, event, sb); + if (error || clnt == clnt->cl_parent) + break; + } + return error; +} + +static struct rpc_clnt *rpc_get_client_for_event(struct net *net, int event) +{ + struct sunrpc_net *sn = net_generic(net, sunrpc_net_id); + struct rpc_clnt *clnt; + + spin_lock(&sn->rpc_client_lock); + list_for_each_entry(clnt, &sn->all_clients, cl_clients) { + if (rpc_clnt_skip_event(clnt, event)) + continue; + spin_unlock(&sn->rpc_client_lock); + return clnt; + } + spin_unlock(&sn->rpc_client_lock); + return NULL; +} + +static int rpc_pipefs_event(struct notifier_block *nb, unsigned long event, + void *ptr) +{ + struct super_block *sb = ptr; + struct rpc_clnt *clnt; + int error = 0; + + while ((clnt = rpc_get_client_for_event(sb->s_fs_info, event))) { + error = __rpc_pipefs_event(clnt, event, sb); + if (error) + break; + } + return error; +} + +static struct notifier_block rpc_clients_block = { + .notifier_call = rpc_pipefs_event, + .priority = SUNRPC_PIPEFS_RPC_PRIO, +}; + +int rpc_clients_notifier_register(void) +{ + return rpc_pipefs_notifier_register(&rpc_clients_block); +} + +void rpc_clients_notifier_unregister(void) +{ + return rpc_pipefs_notifier_unregister(&rpc_clients_block); +} + +static struct rpc_xprt *rpc_clnt_set_transport(struct rpc_clnt *clnt, + struct rpc_xprt *xprt, + const struct rpc_timeout *timeout) +{ + struct rpc_xprt *old; + + spin_lock(&clnt->cl_lock); + old = rcu_dereference_protected(clnt->cl_xprt, + lockdep_is_held(&clnt->cl_lock)); + + if (!xprt_bound(xprt)) + clnt->cl_autobind = 1; + + clnt->cl_timeout = timeout; + rcu_assign_pointer(clnt->cl_xprt, xprt); + spin_unlock(&clnt->cl_lock); + + return old; +} + +static void rpc_clnt_set_nodename(struct rpc_clnt *clnt, const char *nodename) +{ + clnt->cl_nodelen = strlen(nodename); + if (clnt->cl_nodelen > UNX_MAXNODENAME) + clnt->cl_nodelen = UNX_MAXNODENAME; + memcpy(clnt->cl_nodename, nodename, clnt->cl_nodelen); +} + +static int rpc_client_register(struct rpc_clnt *clnt, + rpc_authflavor_t pseudoflavor, + const char *client_name) +{ + struct rpc_auth_create_args auth_args = { + .pseudoflavor = pseudoflavor, + .target_name = client_name, + }; + struct rpc_auth *auth; + struct net *net = rpc_net_ns(clnt); + struct super_block *pipefs_sb; + int err; + + pipefs_sb = rpc_get_sb_net(net); + if (pipefs_sb) { + err = rpc_setup_pipedir(pipefs_sb, clnt); + if (err) + goto out; + } + + rpc_register_client(clnt); + if (pipefs_sb) + rpc_put_sb_net(net); + + auth = rpcauth_create(&auth_args, clnt); + if (IS_ERR(auth)) { + dprintk("RPC: Couldn't create auth handle (flavor %u)\n", + pseudoflavor); + err = PTR_ERR(auth); + goto err_auth; + } + return 0; +err_auth: + pipefs_sb = rpc_get_sb_net(net); + rpc_unregister_client(clnt); + __rpc_clnt_remove_pipedir(clnt); +out: + if (pipefs_sb) + rpc_put_sb_net(net); + return err; +} + +static DEFINE_IDA(rpc_clids); + +static int rpc_alloc_clid(struct rpc_clnt *clnt) +{ + int clid; + + clid = ida_simple_get(&rpc_clids, 0, 0, GFP_KERNEL); + if (clid < 0) + return clid; + clnt->cl_clid = clid; + return 0; +} + +static void rpc_free_clid(struct rpc_clnt *clnt) +{ + ida_simple_remove(&rpc_clids, clnt->cl_clid); +} + +static struct rpc_clnt * rpc_new_client(const struct rpc_create_args *args, + struct rpc_xprt *xprt, + struct rpc_clnt *parent) { - struct rpc_version *version; - struct rpc_clnt *clnt = NULL; - struct rpc_auth *auth; + const struct rpc_program *program = args->program; + const struct rpc_version *version; + struct rpc_clnt *clnt = NULL; + const struct rpc_timeout *timeout; int err; - int len; - dprintk("RPC: creating %s client for %s (xprt %p)\n", - program->name, servname, xprt); + /* sanity check the name before trying to print it */ + dprintk("RPC: creating %s client for %s (xprt %p)\n", + program->name, args->servername, xprt); + + err = rpciod_up(); + if (err) + goto out_no_rpciod; err = -EINVAL; - if (!xprt) + if (args->version >= program->nrvers) goto out_err; - if (vers >= program->nrvers || !(version = program->version[vers])) + version = program->version[args->version]; + if (version == NULL) goto out_err; err = -ENOMEM; - clnt = (struct rpc_clnt *) kmalloc(sizeof(*clnt), GFP_KERNEL); + clnt = kzalloc(sizeof(*clnt), GFP_KERNEL); if (!clnt) goto out_err; - memset(clnt, 0, sizeof(*clnt)); - atomic_set(&clnt->cl_users, 0); - atomic_set(&clnt->cl_count, 1); - clnt->cl_parent = clnt; + clnt->cl_parent = parent ? : clnt; - clnt->cl_server = clnt->cl_inline_name; - len = strlen(servname) + 1; - if (len > sizeof(clnt->cl_inline_name)) { - char *buf = kmalloc(len, GFP_KERNEL); - if (buf != 0) - clnt->cl_server = buf; - else - len = sizeof(clnt->cl_inline_name); - } - strlcpy(clnt->cl_server, servname, len); + err = rpc_alloc_clid(clnt); + if (err) + goto out_no_clid; - clnt->cl_xprt = xprt; clnt->cl_procinfo = version->procs; clnt->cl_maxproc = version->nrprocs; - clnt->cl_protname = program->name; - clnt->cl_pmap = &clnt->cl_pmap_default; - clnt->cl_port = xprt->addr.sin_port; - clnt->cl_prog = program->number; + clnt->cl_prog = args->prognumber ? : program->number; clnt->cl_vers = version->number; - clnt->cl_prot = xprt->prot; clnt->cl_stats = program->stats; - rpc_init_wait_queue(&clnt->cl_pmap_default.pm_bindwait, "bindwait"); + clnt->cl_metrics = rpc_alloc_iostats(clnt); + rpc_init_pipe_dir_head(&clnt->cl_pipedir_objects); + err = -ENOMEM; + if (clnt->cl_metrics == NULL) + goto out_no_stats; + clnt->cl_program = program; + INIT_LIST_HEAD(&clnt->cl_tasks); + spin_lock_init(&clnt->cl_lock); + + timeout = xprt->timeout; + if (args->timeout != NULL) { + memcpy(&clnt->cl_timeout_default, args->timeout, + sizeof(clnt->cl_timeout_default)); + timeout = &clnt->cl_timeout_default; + } - if (!clnt->cl_port) - clnt->cl_autobind = 1; + rpc_clnt_set_transport(clnt, xprt, timeout); clnt->cl_rtt = &clnt->cl_rtt_default; - rpc_init_rtt(&clnt->cl_rtt_default, xprt->timeout.to_initval); - - err = rpc_setup_pipedir(clnt, program->pipe_dir_name); - if (err < 0) - goto out_no_path; + rpc_init_rtt(&clnt->cl_rtt_default, clnt->cl_timeout->to_initval); - auth = rpcauth_create(flavor, clnt); - if (IS_ERR(auth)) { - printk(KERN_INFO "RPC: Couldn't create auth handle (flavor %u)\n", - flavor); - err = PTR_ERR(auth); - goto out_no_auth; - } + atomic_set(&clnt->cl_count, 1); /* save the nodename */ - clnt->cl_nodelen = strlen(system_utsname.nodename); - if (clnt->cl_nodelen > UNX_MAXNODENAME) - clnt->cl_nodelen = UNX_MAXNODENAME; - memcpy(clnt->cl_nodename, system_utsname.nodename, clnt->cl_nodelen); + rpc_clnt_set_nodename(clnt, utsname()->nodename); + + err = rpc_client_register(clnt, args->authflavor, args->client_name); + if (err) + goto out_no_path; + if (parent) + atomic_inc(&parent->cl_count); return clnt; -out_no_auth: - rpc_rmdir(clnt->cl_pathname); out_no_path: - if (clnt->cl_server != clnt->cl_inline_name) - kfree(clnt->cl_server); + rpc_free_iostats(clnt->cl_metrics); +out_no_stats: + rpc_free_clid(clnt); +out_no_clid: kfree(clnt); out_err: - xprt_destroy(xprt); + rpciod_down(); +out_no_rpciod: + xprt_put(xprt); return ERR_PTR(err); } +struct rpc_clnt *rpc_create_xprt(struct rpc_create_args *args, + struct rpc_xprt *xprt) +{ + struct rpc_clnt *clnt = NULL; + + clnt = rpc_new_client(args, xprt, NULL); + if (IS_ERR(clnt)) + return clnt; + + if (!(args->flags & RPC_CLNT_CREATE_NOPING)) { + int err = rpc_ping(clnt); + if (err != 0) { + rpc_shutdown_client(clnt); + return ERR_PTR(err); + } + } + + clnt->cl_softrtry = 1; + if (args->flags & RPC_CLNT_CREATE_HARDRTRY) + clnt->cl_softrtry = 0; + + if (args->flags & RPC_CLNT_CREATE_AUTOBIND) + clnt->cl_autobind = 1; + if (args->flags & RPC_CLNT_CREATE_DISCRTRY) + clnt->cl_discrtry = 1; + if (!(args->flags & RPC_CLNT_CREATE_QUIET)) + clnt->cl_chatty = 1; + + return clnt; +} +EXPORT_SYMBOL_GPL(rpc_create_xprt); + /** - * Create an RPC client - * @xprt - pointer to xprt struct - * @servname - name of server - * @info - rpc_program - * @version - rpc_program version - * @authflavor - rpc_auth flavour to use + * rpc_create - create an RPC client and transport with one call + * @args: rpc_clnt create argument structure * - * Creates an RPC client structure, then pings the server in order to - * determine if it is up, and if it supports this program and version. + * Creates and initializes an RPC transport and an RPC client. * - * This function should never be called by asynchronous tasks such as - * the portmapper. + * It can ping the server in order to determine if it is up, and to see if + * it supports this program and version. RPC_CLNT_CREATE_NOPING disables + * this behavior so asynchronous tasks can also use rpc_create. */ -struct rpc_clnt *rpc_create_client(struct rpc_xprt *xprt, char *servname, - struct rpc_program *info, u32 version, rpc_authflavor_t authflavor) +struct rpc_clnt *rpc_create(struct rpc_create_args *args) { - struct rpc_clnt *clnt; - int err; - - clnt = rpc_new_client(xprt, servname, info, version, authflavor); - if (IS_ERR(clnt)) - return clnt; - err = rpc_ping(clnt, RPC_TASK_SOFT|RPC_TASK_NOINTR); - if (err == 0) - return clnt; - rpc_shutdown_client(clnt); - return ERR_PTR(err); + struct rpc_xprt *xprt; + struct xprt_create xprtargs = { + .net = args->net, + .ident = args->protocol, + .srcaddr = args->saddress, + .dstaddr = args->address, + .addrlen = args->addrsize, + .servername = args->servername, + .bc_xprt = args->bc_xprt, + }; + char servername[48]; + + if (args->flags & RPC_CLNT_CREATE_INFINITE_SLOTS) + xprtargs.flags |= XPRT_CREATE_INFINITE_SLOTS; + if (args->flags & RPC_CLNT_CREATE_NO_IDLE_TIMEOUT) + xprtargs.flags |= XPRT_CREATE_NO_IDLE_TIMEOUT; + /* + * If the caller chooses not to specify a hostname, whip + * up a string representation of the passed-in address. + */ + if (xprtargs.servername == NULL) { + struct sockaddr_un *sun = + (struct sockaddr_un *)args->address; + struct sockaddr_in *sin = + (struct sockaddr_in *)args->address; + struct sockaddr_in6 *sin6 = + (struct sockaddr_in6 *)args->address; + + servername[0] = '\0'; + switch (args->address->sa_family) { + case AF_LOCAL: + snprintf(servername, sizeof(servername), "%s", + sun->sun_path); + break; + case AF_INET: + snprintf(servername, sizeof(servername), "%pI4", + &sin->sin_addr.s_addr); + break; + case AF_INET6: + snprintf(servername, sizeof(servername), "%pI6", + &sin6->sin6_addr); + break; + default: + /* caller wants default server name, but + * address family isn't recognized. */ + return ERR_PTR(-EINVAL); + } + xprtargs.servername = servername; + } + + xprt = xprt_create_transport(&xprtargs); + if (IS_ERR(xprt)) + return (struct rpc_clnt *)xprt; + + /* + * By default, kernel RPC client connects from a reserved port. + * CAP_NET_BIND_SERVICE will not be set for unprivileged requesters, + * but it is always enabled for rpciod, which handles the connect + * operation. + */ + xprt->resvport = 1; + if (args->flags & RPC_CLNT_CREATE_NONPRIVPORT) + xprt->resvport = 0; + + return rpc_create_xprt(args, xprt); } +EXPORT_SYMBOL_GPL(rpc_create); /* * This function clones the RPC client structure. It allows us to share the * same transport while varying parameters such as the authentication * flavour. */ -struct rpc_clnt * -rpc_clone_client(struct rpc_clnt *clnt) +static struct rpc_clnt *__rpc_clone_client(struct rpc_create_args *args, + struct rpc_clnt *clnt) { + struct rpc_xprt *xprt; struct rpc_clnt *new; + int err; + + err = -ENOMEM; + rcu_read_lock(); + xprt = xprt_get(rcu_dereference(clnt->cl_xprt)); + rcu_read_unlock(); + if (xprt == NULL) + goto out_err; + args->servername = xprt->servername; + + new = rpc_new_client(args, xprt, clnt); + if (IS_ERR(new)) { + err = PTR_ERR(new); + goto out_err; + } - new = (struct rpc_clnt *)kmalloc(sizeof(*new), GFP_KERNEL); - if (!new) - goto out_no_clnt; - memcpy(new, clnt, sizeof(*new)); - atomic_set(&new->cl_count, 1); - atomic_set(&new->cl_users, 0); - new->cl_parent = clnt; - atomic_inc(&clnt->cl_count); - /* Duplicate portmapper */ - rpc_init_wait_queue(&new->cl_pmap_default.pm_bindwait, "bindwait"); /* Turn off autobind on clones */ new->cl_autobind = 0; - new->cl_oneshot = 0; - new->cl_dead = 0; - rpc_init_rtt(&new->cl_rtt_default, clnt->cl_xprt->timeout.to_initval); - if (new->cl_auth) - atomic_inc(&new->cl_auth->au_count); - new->cl_pmap = &new->cl_pmap_default; - rpc_init_wait_queue(&new->cl_pmap_default.pm_bindwait, "bindwait"); + new->cl_softrtry = clnt->cl_softrtry; + new->cl_discrtry = clnt->cl_discrtry; + new->cl_chatty = clnt->cl_chatty; return new; -out_no_clnt: - printk(KERN_INFO "RPC: out of memory in %s\n", __FUNCTION__); - return ERR_PTR(-ENOMEM); + +out_err: + dprintk("RPC: %s: returned error %d\n", __func__, err); + return ERR_PTR(err); +} + +/** + * rpc_clone_client - Clone an RPC client structure + * + * @clnt: RPC client whose parameters are copied + * + * Returns a fresh RPC client or an ERR_PTR. + */ +struct rpc_clnt *rpc_clone_client(struct rpc_clnt *clnt) +{ + struct rpc_create_args args = { + .program = clnt->cl_program, + .prognumber = clnt->cl_prog, + .version = clnt->cl_vers, + .authflavor = clnt->cl_auth->au_flavor, + }; + return __rpc_clone_client(&args, clnt); +} +EXPORT_SYMBOL_GPL(rpc_clone_client); + +/** + * rpc_clone_client_set_auth - Clone an RPC client structure and set its auth + * + * @clnt: RPC client whose parameters are copied + * @flavor: security flavor for new client + * + * Returns a fresh RPC client or an ERR_PTR. + */ +struct rpc_clnt * +rpc_clone_client_set_auth(struct rpc_clnt *clnt, rpc_authflavor_t flavor) +{ + struct rpc_create_args args = { + .program = clnt->cl_program, + .prognumber = clnt->cl_prog, + .version = clnt->cl_vers, + .authflavor = flavor, + }; + return __rpc_clone_client(&args, clnt); } +EXPORT_SYMBOL_GPL(rpc_clone_client_set_auth); + +/** + * rpc_switch_client_transport: switch the RPC transport on the fly + * @clnt: pointer to a struct rpc_clnt + * @args: pointer to the new transport arguments + * @timeout: pointer to the new timeout parameters + * + * This function allows the caller to switch the RPC transport for the + * rpc_clnt structure 'clnt' to allow it to connect to a mirrored NFS + * server, for instance. It assumes that the caller has ensured that + * there are no active RPC tasks by using some form of locking. + * + * Returns zero if "clnt" is now using the new xprt. Otherwise a + * negative errno is returned, and "clnt" continues to use the old + * xprt. + */ +int rpc_switch_client_transport(struct rpc_clnt *clnt, + struct xprt_create *args, + const struct rpc_timeout *timeout) +{ + const struct rpc_timeout *old_timeo; + rpc_authflavor_t pseudoflavor; + struct rpc_xprt *xprt, *old; + struct rpc_clnt *parent; + int err; + + xprt = xprt_create_transport(args); + if (IS_ERR(xprt)) { + dprintk("RPC: failed to create new xprt for clnt %p\n", + clnt); + return PTR_ERR(xprt); + } + + pseudoflavor = clnt->cl_auth->au_flavor; + + old_timeo = clnt->cl_timeout; + old = rpc_clnt_set_transport(clnt, xprt, timeout); + + rpc_unregister_client(clnt); + __rpc_clnt_remove_pipedir(clnt); + + /* + * A new transport was created. "clnt" therefore + * becomes the root of a new cl_parent tree. clnt's + * children, if it has any, still point to the old xprt. + */ + parent = clnt->cl_parent; + clnt->cl_parent = clnt; + + /* + * The old rpc_auth cache cannot be re-used. GSS + * contexts in particular are between a single + * client and server. + */ + err = rpc_client_register(clnt, pseudoflavor, NULL); + if (err) + goto out_revert; + + synchronize_rcu(); + if (parent != clnt) + rpc_release_client(parent); + xprt_put(old); + dprintk("RPC: replaced xprt for clnt %p\n", clnt); + return 0; + +out_revert: + rpc_clnt_set_transport(clnt, old, old_timeo); + clnt->cl_parent = parent; + rpc_client_register(clnt, pseudoflavor, NULL); + xprt_put(xprt); + dprintk("RPC: failed to switch xprt for clnt %p\n", clnt); + return err; +} +EXPORT_SYMBOL_GPL(rpc_switch_client_transport); /* - * Properly shut down an RPC client, terminating all outstanding - * requests. Note that we must be certain that cl_oneshot and - * cl_dead are cleared, or else the client would be destroyed - * when the last task releases it. + * Kill all tasks for the given client. + * XXX: kill their descendants as well? */ -int -rpc_shutdown_client(struct rpc_clnt *clnt) +void rpc_killall_tasks(struct rpc_clnt *clnt) { - dprintk("RPC: shutting down %s client for %s, tasks=%d\n", - clnt->cl_protname, clnt->cl_server, - atomic_read(&clnt->cl_users)); + struct rpc_task *rovr; - while (atomic_read(&clnt->cl_users) > 0) { - /* Don't let rpc_release_client destroy us */ - clnt->cl_oneshot = 0; - clnt->cl_dead = 0; - rpc_killall_tasks(clnt); - sleep_on_timeout(&destroy_wait, 1*HZ); + + if (list_empty(&clnt->cl_tasks)) + return; + dprintk("RPC: killing all tasks for client %p\n", clnt); + /* + * Spin lock all_tasks to prevent changes... + */ + spin_lock(&clnt->cl_lock); + list_for_each_entry(rovr, &clnt->cl_tasks, tk_task) { + if (!RPC_IS_ACTIVATED(rovr)) + continue; + if (!(rovr->tk_flags & RPC_TASK_KILLED)) { + rovr->tk_flags |= RPC_TASK_KILLED; + rpc_exit(rovr, -EIO); + if (RPC_IS_QUEUED(rovr)) + rpc_wake_up_queued_task(rovr->tk_waitqueue, + rovr); + } } + spin_unlock(&clnt->cl_lock); +} +EXPORT_SYMBOL_GPL(rpc_killall_tasks); - if (atomic_read(&clnt->cl_users) < 0) { - printk(KERN_ERR "RPC: rpc_shutdown_client clnt %p tasks=%d\n", - clnt, atomic_read(&clnt->cl_users)); -#ifdef RPC_DEBUG - rpc_show_tasks(); -#endif - BUG(); +/* + * Properly shut down an RPC client, terminating all outstanding + * requests. + */ +void rpc_shutdown_client(struct rpc_clnt *clnt) +{ + might_sleep(); + + dprintk_rcu("RPC: shutting down %s client for %s\n", + clnt->cl_program->name, + rcu_dereference(clnt->cl_xprt)->servername); + + while (!list_empty(&clnt->cl_tasks)) { + rpc_killall_tasks(clnt); + wait_event_timeout(destroy_wait, + list_empty(&clnt->cl_tasks), 1*HZ); } - return rpc_destroy_client(clnt); + rpc_release_client(clnt); } +EXPORT_SYMBOL_GPL(rpc_shutdown_client); /* - * Delete an RPC client + * Free an RPC client */ -int -rpc_destroy_client(struct rpc_clnt *clnt) +static struct rpc_clnt * +rpc_free_client(struct rpc_clnt *clnt) { - if (!atomic_dec_and_test(&clnt->cl_count)) - return 1; - BUG_ON(atomic_read(&clnt->cl_users) != 0); - - dprintk("RPC: destroying %s client for %s\n", - clnt->cl_protname, clnt->cl_server); - if (clnt->cl_auth) { - rpcauth_destroy(clnt->cl_auth); - clnt->cl_auth = NULL; - } - if (clnt->cl_parent != clnt) { - rpc_destroy_client(clnt->cl_parent); - goto out_free; - } - if (clnt->cl_pathname[0]) - rpc_rmdir(clnt->cl_pathname); - if (clnt->cl_xprt) { - xprt_destroy(clnt->cl_xprt); - clnt->cl_xprt = NULL; - } - if (clnt->cl_server != clnt->cl_inline_name) - kfree(clnt->cl_server); -out_free: + struct rpc_clnt *parent = NULL; + + dprintk_rcu("RPC: destroying %s client for %s\n", + clnt->cl_program->name, + rcu_dereference(clnt->cl_xprt)->servername); + if (clnt->cl_parent != clnt) + parent = clnt->cl_parent; + rpc_clnt_remove_pipedir(clnt); + rpc_unregister_client(clnt); + rpc_free_iostats(clnt->cl_metrics); + clnt->cl_metrics = NULL; + xprt_put(rcu_dereference_raw(clnt->cl_xprt)); + rpciod_down(); + rpc_free_clid(clnt); kfree(clnt); - return 0; + return parent; } /* - * Release an RPC client + * Free an RPC client + */ +static struct rpc_clnt * +rpc_free_auth(struct rpc_clnt *clnt) +{ + if (clnt->cl_auth == NULL) + return rpc_free_client(clnt); + + /* + * Note: RPCSEC_GSS may need to send NULL RPC calls in order to + * release remaining GSS contexts. This mechanism ensures + * that it can do so safely. + */ + atomic_inc(&clnt->cl_count); + rpcauth_release(clnt->cl_auth); + clnt->cl_auth = NULL; + if (atomic_dec_and_test(&clnt->cl_count)) + return rpc_free_client(clnt); + return NULL; +} + +/* + * Release reference to the RPC client */ void rpc_release_client(struct rpc_clnt *clnt) { - dprintk("RPC: rpc_release_client(%p, %d)\n", - clnt, atomic_read(&clnt->cl_users)); + dprintk("RPC: rpc_release_client(%p)\n", clnt); - if (!atomic_dec_and_test(&clnt->cl_users)) - return; - wake_up(&destroy_wait); - if (clnt->cl_oneshot || clnt->cl_dead) - rpc_destroy_client(clnt); + do { + if (list_empty(&clnt->cl_tasks)) + wake_up(&destroy_wait); + if (!atomic_dec_and_test(&clnt->cl_count)) + break; + clnt = rpc_free_auth(clnt); + } while (clnt != NULL); } +EXPORT_SYMBOL_GPL(rpc_release_client); /** * rpc_bind_new_program - bind a new RPC program to an existing client - * @old - old rpc_client - * @program - rpc program to set - * @vers - rpc program version + * @old: old rpc_client + * @program: rpc program to set + * @vers: rpc program version * * Clones the rpc client and sets up a new RPC program. This is mainly * of use for enabling different RPC programs to share the same transport. * The Sun NFSv2/v3 ACL protocol can do this. */ struct rpc_clnt *rpc_bind_new_program(struct rpc_clnt *old, - struct rpc_program *program, - int vers) + const struct rpc_program *program, + u32 vers) { + struct rpc_create_args args = { + .program = program, + .prognumber = program->number, + .version = vers, + .authflavor = old->cl_auth->au_flavor, + }; struct rpc_clnt *clnt; - struct rpc_version *version; int err; - BUG_ON(vers >= program->nrvers || !program->version[vers]); - version = program->version[vers]; - clnt = rpc_clone_client(old); + clnt = __rpc_clone_client(&args, old); if (IS_ERR(clnt)) goto out; - clnt->cl_procinfo = version->procs; - clnt->cl_maxproc = version->nrprocs; - clnt->cl_protname = program->name; - clnt->cl_prog = program->number; - clnt->cl_vers = version->number; - clnt->cl_stats = program->stats; - err = rpc_ping(clnt, RPC_TASK_SOFT|RPC_TASK_NOINTR); + err = rpc_ping(clnt); if (err != 0) { rpc_shutdown_client(clnt); clnt = ERR_PTR(err); } -out: +out: return clnt; } +EXPORT_SYMBOL_GPL(rpc_bind_new_program); + +void rpc_task_release_client(struct rpc_task *task) +{ + struct rpc_clnt *clnt = task->tk_client; + + if (clnt != NULL) { + /* Remove from client task list */ + spin_lock(&clnt->cl_lock); + list_del(&task->tk_task); + spin_unlock(&clnt->cl_lock); + task->tk_client = NULL; + + rpc_release_client(clnt); + } +} + +static +void rpc_task_set_client(struct rpc_task *task, struct rpc_clnt *clnt) +{ + if (clnt != NULL) { + rpc_task_release_client(task); + task->tk_client = clnt; + atomic_inc(&clnt->cl_count); + if (clnt->cl_softrtry) + task->tk_flags |= RPC_TASK_SOFT; + if (clnt->cl_noretranstimeo) + task->tk_flags |= RPC_TASK_NO_RETRANS_TIMEOUT; + if (sk_memalloc_socks()) { + struct rpc_xprt *xprt; + + rcu_read_lock(); + xprt = rcu_dereference(clnt->cl_xprt); + if (xprt->swapper) + task->tk_flags |= RPC_TASK_SWAPPER; + rcu_read_unlock(); + } + /* Add to the client's list of all tasks */ + spin_lock(&clnt->cl_lock); + list_add_tail(&task->tk_task, &clnt->cl_tasks); + spin_unlock(&clnt->cl_lock); + } +} + +void rpc_task_reset_client(struct rpc_task *task, struct rpc_clnt *clnt) +{ + rpc_task_release_client(task); + rpc_task_set_client(task, clnt); +} +EXPORT_SYMBOL_GPL(rpc_task_reset_client); + + +static void +rpc_task_set_rpc_message(struct rpc_task *task, const struct rpc_message *msg) +{ + if (msg != NULL) { + task->tk_msg.rpc_proc = msg->rpc_proc; + task->tk_msg.rpc_argp = msg->rpc_argp; + task->tk_msg.rpc_resp = msg->rpc_resp; + if (msg->rpc_cred != NULL) + task->tk_msg.rpc_cred = get_rpccred(msg->rpc_cred); + } +} /* * Default callback for async RPC calls */ static void -rpc_default_callback(struct rpc_task *task) +rpc_default_callback(struct rpc_task *task, void *data) { } -/* - * Export the signal mask handling for synchronous code that - * sleeps on RPC calls +static const struct rpc_call_ops rpc_default_ops = { + .rpc_call_done = rpc_default_callback, +}; + +/** + * rpc_run_task - Allocate a new RPC task, then run rpc_execute against it + * @task_setup_data: pointer to task initialisation data */ -#define RPC_INTR_SIGNALS (sigmask(SIGINT) | sigmask(SIGQUIT) | sigmask(SIGKILL)) - -static void rpc_save_sigmask(sigset_t *oldset, int intr) +struct rpc_task *rpc_run_task(const struct rpc_task_setup *task_setup_data) { - unsigned long sigallow = 0; - sigset_t sigmask; + struct rpc_task *task; + + task = rpc_new_task(task_setup_data); + if (IS_ERR(task)) + goto out; - /* Block all signals except those listed in sigallow */ - if (intr) - sigallow |= RPC_INTR_SIGNALS; - siginitsetinv(&sigmask, sigallow); - sigprocmask(SIG_BLOCK, &sigmask, oldset); + rpc_task_set_client(task, task_setup_data->rpc_client); + rpc_task_set_rpc_message(task, task_setup_data->rpc_message); + + if (task->tk_action == NULL) + rpc_call_start(task); + + atomic_inc(&task->tk_count); + rpc_execute(task); +out: + return task; } +EXPORT_SYMBOL_GPL(rpc_run_task); -static inline void rpc_task_sigmask(struct rpc_task *task, sigset_t *oldset) +/** + * rpc_call_sync - Perform a synchronous RPC call + * @clnt: pointer to RPC client + * @msg: RPC call parameters + * @flags: RPC call flags + */ +int rpc_call_sync(struct rpc_clnt *clnt, const struct rpc_message *msg, int flags) { - rpc_save_sigmask(oldset, !RPC_TASK_UNINTERRUPTIBLE(task)); + struct rpc_task *task; + struct rpc_task_setup task_setup_data = { + .rpc_client = clnt, + .rpc_message = msg, + .callback_ops = &rpc_default_ops, + .flags = flags, + }; + int status; + + WARN_ON_ONCE(flags & RPC_TASK_ASYNC); + if (flags & RPC_TASK_ASYNC) { + rpc_release_calldata(task_setup_data.callback_ops, + task_setup_data.callback_data); + return -EINVAL; + } + + task = rpc_run_task(&task_setup_data); + if (IS_ERR(task)) + return PTR_ERR(task); + status = task->tk_status; + rpc_put_task(task); + return status; } +EXPORT_SYMBOL_GPL(rpc_call_sync); -static inline void rpc_restore_sigmask(sigset_t *oldset) +/** + * rpc_call_async - Perform an asynchronous RPC call + * @clnt: pointer to RPC client + * @msg: RPC call parameters + * @flags: RPC call flags + * @tk_ops: RPC call ops + * @data: user call data + */ +int +rpc_call_async(struct rpc_clnt *clnt, const struct rpc_message *msg, int flags, + const struct rpc_call_ops *tk_ops, void *data) { - sigprocmask(SIG_SETMASK, oldset, NULL); + struct rpc_task *task; + struct rpc_task_setup task_setup_data = { + .rpc_client = clnt, + .rpc_message = msg, + .callback_ops = tk_ops, + .callback_data = data, + .flags = flags|RPC_TASK_ASYNC, + }; + + task = rpc_run_task(&task_setup_data); + if (IS_ERR(task)) + return PTR_ERR(task); + rpc_put_task(task); + return 0; } +EXPORT_SYMBOL_GPL(rpc_call_async); -void rpc_clnt_sigmask(struct rpc_clnt *clnt, sigset_t *oldset) +#if defined(CONFIG_SUNRPC_BACKCHANNEL) +/** + * rpc_run_bc_task - Allocate a new RPC task for backchannel use, then run + * rpc_execute against it + * @req: RPC request + * @tk_ops: RPC call ops + */ +struct rpc_task *rpc_run_bc_task(struct rpc_rqst *req, + const struct rpc_call_ops *tk_ops) { - rpc_save_sigmask(oldset, clnt->cl_intr); + struct rpc_task *task; + struct xdr_buf *xbufp = &req->rq_snd_buf; + struct rpc_task_setup task_setup_data = { + .callback_ops = tk_ops, + }; + + dprintk("RPC: rpc_run_bc_task req= %p\n", req); + /* + * Create an rpc_task to send the data + */ + task = rpc_new_task(&task_setup_data); + if (IS_ERR(task)) { + xprt_free_bc_request(req); + goto out; + } + task->tk_rqstp = req; + + /* + * Set up the xdr_buf length. + * This also indicates that the buffer is XDR encoded already. + */ + xbufp->len = xbufp->head[0].iov_len + xbufp->page_len + + xbufp->tail[0].iov_len; + + task->tk_action = call_bc_transmit; + atomic_inc(&task->tk_count); + WARN_ON_ONCE(atomic_read(&task->tk_count) != 2); + rpc_execute(task); + +out: + dprintk("RPC: rpc_run_bc_task: task= %p\n", task); + return task; } +#endif /* CONFIG_SUNRPC_BACKCHANNEL */ -void rpc_clnt_sigunmask(struct rpc_clnt *clnt, sigset_t *oldset) +void +rpc_call_start(struct rpc_task *task) { - rpc_restore_sigmask(oldset); + task->tk_action = call_start; } +EXPORT_SYMBOL_GPL(rpc_call_start); -/* - * New rpc_call implementation +/** + * rpc_peeraddr - extract remote peer address from clnt's xprt + * @clnt: RPC client structure + * @buf: target buffer + * @bufsize: length of target buffer + * + * Returns the number of bytes that are actually in the stored address. */ -int rpc_call_sync(struct rpc_clnt *clnt, struct rpc_message *msg, int flags) +size_t rpc_peeraddr(struct rpc_clnt *clnt, struct sockaddr *buf, size_t bufsize) { - struct rpc_task *task; - sigset_t oldset; - int status; + size_t bytes; + struct rpc_xprt *xprt; - /* If this client is slain all further I/O fails */ - if (clnt->cl_dead) - return -EIO; + rcu_read_lock(); + xprt = rcu_dereference(clnt->cl_xprt); - BUG_ON(flags & RPC_TASK_ASYNC); + bytes = xprt->addrlen; + if (bytes > bufsize) + bytes = bufsize; + memcpy(buf, &xprt->addr, bytes); + rcu_read_unlock(); - status = -ENOMEM; - task = rpc_new_task(clnt, NULL, flags); - if (task == NULL) - goto out; - - /* Mask signals on RPC calls _and_ GSS_AUTH upcalls */ - rpc_task_sigmask(task, &oldset); + return bytes; +} +EXPORT_SYMBOL_GPL(rpc_peeraddr); - rpc_call_setup(task, msg, 0); +/** + * rpc_peeraddr2str - return remote peer address in printable format + * @clnt: RPC client structure + * @format: address format + * + * NB: the lifetime of the memory referenced by the returned pointer is + * the same as the rpc_xprt itself. As long as the caller uses this + * pointer, it must hold the RCU read lock. + */ +const char *rpc_peeraddr2str(struct rpc_clnt *clnt, + enum rpc_display_format_t format) +{ + struct rpc_xprt *xprt; - /* Set up the call info struct and execute the task */ - if (task->tk_status == 0) { - status = rpc_execute(task); - } else { - status = task->tk_status; - rpc_release_task(task); - } + xprt = rcu_dereference(clnt->cl_xprt); - rpc_restore_sigmask(&oldset); -out: - return status; + if (xprt->address_strings[format] != NULL) + return xprt->address_strings[format]; + else + return "unprintable"; } +EXPORT_SYMBOL_GPL(rpc_peeraddr2str); + +static const struct sockaddr_in rpc_inaddr_loopback = { + .sin_family = AF_INET, + .sin_addr.s_addr = htonl(INADDR_ANY), +}; + +static const struct sockaddr_in6 rpc_in6addr_loopback = { + .sin6_family = AF_INET6, + .sin6_addr = IN6ADDR_ANY_INIT, +}; /* - * New rpc_call implementation + * Try a getsockname() on a connected datagram socket. Using a + * connected datagram socket prevents leaving a socket in TIME_WAIT. + * This conserves the ephemeral port number space. + * + * Returns zero and fills in "buf" if successful; otherwise, a + * negative errno is returned. */ -int -rpc_call_async(struct rpc_clnt *clnt, struct rpc_message *msg, int flags, - rpc_action callback, void *data) +static int rpc_sockname(struct net *net, struct sockaddr *sap, size_t salen, + struct sockaddr *buf, int buflen) { - struct rpc_task *task; - sigset_t oldset; - int status; - - /* If this client is slain all further I/O fails */ - if (clnt->cl_dead) - return -EIO; + struct socket *sock; + int err; - flags |= RPC_TASK_ASYNC; + err = __sock_create(net, sap->sa_family, + SOCK_DGRAM, IPPROTO_UDP, &sock, 1); + if (err < 0) { + dprintk("RPC: can't create UDP socket (%d)\n", err); + goto out; + } - /* Create/initialize a new RPC task */ - if (!callback) - callback = rpc_default_callback; - status = -ENOMEM; - if (!(task = rpc_new_task(clnt, callback, flags))) + switch (sap->sa_family) { + case AF_INET: + err = kernel_bind(sock, + (struct sockaddr *)&rpc_inaddr_loopback, + sizeof(rpc_inaddr_loopback)); + break; + case AF_INET6: + err = kernel_bind(sock, + (struct sockaddr *)&rpc_in6addr_loopback, + sizeof(rpc_in6addr_loopback)); + break; + default: + err = -EAFNOSUPPORT; goto out; - task->tk_calldata = data; + } + if (err < 0) { + dprintk("RPC: can't bind UDP socket (%d)\n", err); + goto out_release; + } - /* Mask signals on GSS_AUTH upcalls */ - rpc_task_sigmask(task, &oldset); + err = kernel_connect(sock, sap, salen, 0); + if (err < 0) { + dprintk("RPC: can't connect UDP socket (%d)\n", err); + goto out_release; + } - rpc_call_setup(task, msg, 0); + err = kernel_getsockname(sock, buf, &buflen); + if (err < 0) { + dprintk("RPC: getsockname failed (%d)\n", err); + goto out_release; + } - /* Set up the call info struct and execute the task */ - status = task->tk_status; - if (status == 0) - rpc_execute(task); - else - rpc_release_task(task); + err = 0; + if (buf->sa_family == AF_INET6) { + struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)buf; + sin6->sin6_scope_id = 0; + } + dprintk("RPC: %s succeeded\n", __func__); - rpc_restore_sigmask(&oldset); +out_release: + sock_release(sock); out: - return status; + return err; } +/* + * Scraping a connected socket failed, so we don't have a useable + * local address. Fallback: generate an address that will prevent + * the server from calling us back. + * + * Returns zero and fills in "buf" if successful; otherwise, a + * negative errno is returned. + */ +static int rpc_anyaddr(int family, struct sockaddr *buf, size_t buflen) +{ + switch (family) { + case AF_INET: + if (buflen < sizeof(rpc_inaddr_loopback)) + return -EINVAL; + memcpy(buf, &rpc_inaddr_loopback, + sizeof(rpc_inaddr_loopback)); + break; + case AF_INET6: + if (buflen < sizeof(rpc_in6addr_loopback)) + return -EINVAL; + memcpy(buf, &rpc_in6addr_loopback, + sizeof(rpc_in6addr_loopback)); + default: + dprintk("RPC: %s: address family not supported\n", + __func__); + return -EAFNOSUPPORT; + } + dprintk("RPC: %s: succeeded\n", __func__); + return 0; +} -void -rpc_call_setup(struct rpc_task *task, struct rpc_message *msg, int flags) +/** + * rpc_localaddr - discover local endpoint address for an RPC client + * @clnt: RPC client structure + * @buf: target buffer + * @buflen: size of target buffer, in bytes + * + * Returns zero and fills in "buf" and "buflen" if successful; + * otherwise, a negative errno is returned. + * + * This works even if the underlying transport is not currently connected, + * or if the upper layer never previously provided a source address. + * + * The result of this function call is transient: multiple calls in + * succession may give different results, depending on how local + * networking configuration changes over time. + */ +int rpc_localaddr(struct rpc_clnt *clnt, struct sockaddr *buf, size_t buflen) { - task->tk_msg = *msg; - task->tk_flags |= flags; - /* Bind the user cred */ - if (task->tk_msg.rpc_cred != NULL) - rpcauth_holdcred(task); - else - rpcauth_bindcred(task); + struct sockaddr_storage address; + struct sockaddr *sap = (struct sockaddr *)&address; + struct rpc_xprt *xprt; + struct net *net; + size_t salen; + int err; - if (task->tk_status == 0) - task->tk_action = call_start; - else - task->tk_action = NULL; + rcu_read_lock(); + xprt = rcu_dereference(clnt->cl_xprt); + salen = xprt->addrlen; + memcpy(sap, &xprt->addr, salen); + net = get_net(xprt->xprt_net); + rcu_read_unlock(); + + rpc_set_port(sap, 0); + err = rpc_sockname(net, sap, salen, buf, buflen); + put_net(net); + if (err != 0) + /* Couldn't discover local address, return ANYADDR */ + return rpc_anyaddr(sap->sa_family, buf, buflen); + return 0; } +EXPORT_SYMBOL_GPL(rpc_localaddr); void rpc_setbufsize(struct rpc_clnt *clnt, unsigned int sndsize, unsigned int rcvsize) { - struct rpc_xprt *xprt = clnt->cl_xprt; + struct rpc_xprt *xprt; + + rcu_read_lock(); + xprt = rcu_dereference(clnt->cl_xprt); if (xprt->ops->set_buffer_size) xprt->ops->set_buffer_size(xprt, sndsize, rcvsize); + rcu_read_unlock(); } +EXPORT_SYMBOL_GPL(rpc_setbufsize); -/* - * Return size of largest payload RPC client can support, in bytes +/** + * rpc_protocol - Get transport protocol number for an RPC client + * @clnt: RPC client to query + * + */ +int rpc_protocol(struct rpc_clnt *clnt) +{ + int protocol; + + rcu_read_lock(); + protocol = rcu_dereference(clnt->cl_xprt)->prot; + rcu_read_unlock(); + return protocol; +} +EXPORT_SYMBOL_GPL(rpc_protocol); + +/** + * rpc_net_ns - Get the network namespace for this RPC client + * @clnt: RPC client to query + * + */ +struct net *rpc_net_ns(struct rpc_clnt *clnt) +{ + struct net *ret; + + rcu_read_lock(); + ret = rcu_dereference(clnt->cl_xprt)->xprt_net; + rcu_read_unlock(); + return ret; +} +EXPORT_SYMBOL_GPL(rpc_net_ns); + +/** + * rpc_max_payload - Get maximum payload size for a transport, in bytes + * @clnt: RPC client to query * * For stream transports, this is one RPC record fragment (see RFC * 1831), as we don't support multi-record requests yet. For datagram @@ -531,22 +1322,91 @@ rpc_setbufsize(struct rpc_clnt *clnt, unsigned int sndsize, unsigned int rcvsize */ size_t rpc_max_payload(struct rpc_clnt *clnt) { - return clnt->cl_xprt->max_payload; + size_t ret; + + rcu_read_lock(); + ret = rcu_dereference(clnt->cl_xprt)->max_payload; + rcu_read_unlock(); + return ret; } -EXPORT_SYMBOL(rpc_max_payload); +EXPORT_SYMBOL_GPL(rpc_max_payload); + +/** + * rpc_get_timeout - Get timeout for transport in units of HZ + * @clnt: RPC client to query + */ +unsigned long rpc_get_timeout(struct rpc_clnt *clnt) +{ + unsigned long ret; + + rcu_read_lock(); + ret = rcu_dereference(clnt->cl_xprt)->timeout->to_initval; + rcu_read_unlock(); + return ret; +} +EXPORT_SYMBOL_GPL(rpc_get_timeout); + +/** + * rpc_force_rebind - force transport to check that remote port is unchanged + * @clnt: client to rebind + * + */ +void rpc_force_rebind(struct rpc_clnt *clnt) +{ + if (clnt->cl_autobind) { + rcu_read_lock(); + xprt_clear_bound(rcu_dereference(clnt->cl_xprt)); + rcu_read_unlock(); + } +} +EXPORT_SYMBOL_GPL(rpc_force_rebind); + +/* + * Restart an (async) RPC call from the call_prepare state. + * Usually called from within the exit handler. + */ +int +rpc_restart_call_prepare(struct rpc_task *task) +{ + if (RPC_ASSASSINATED(task)) + return 0; + task->tk_action = call_start; + task->tk_status = 0; + if (task->tk_ops->rpc_call_prepare != NULL) + task->tk_action = rpc_prepare_task; + return 1; +} +EXPORT_SYMBOL_GPL(rpc_restart_call_prepare); /* * Restart an (async) RPC call. Usually called from within the * exit handler. */ -void +int rpc_restart_call(struct rpc_task *task) { if (RPC_ASSASSINATED(task)) - return; - + return 0; task->tk_action = call_start; + task->tk_status = 0; + return 1; +} +EXPORT_SYMBOL_GPL(rpc_restart_call); + +#ifdef RPC_DEBUG +static const char *rpc_proc_name(const struct rpc_task *task) +{ + const struct rpc_procinfo *proc = task->tk_msg.rpc_proc; + + if (proc) { + if (proc->p_name) + return proc->p_name; + else + return "NULL"; + } else + return "no proc"; } +#endif /* * 0. Initial state @@ -559,9 +1419,10 @@ call_start(struct rpc_task *task) { struct rpc_clnt *clnt = task->tk_client; - dprintk("RPC: %4d call_start %s%d proc %d (%s)\n", task->tk_pid, - clnt->cl_protname, clnt->cl_vers, task->tk_msg.rpc_proc->p_proc, - (RPC_IS_ASYNC(task) ? "async" : "sync")); + dprintk("RPC: %5u call_start %s%d proc %s (%s)\n", task->tk_pid, + clnt->cl_program->name, clnt->cl_vers, + rpc_proc_name(task), + (RPC_IS_ASYNC(task) ? "async" : "sync")); /* Increment call count */ task->tk_msg.rpc_proc->p_count++; @@ -575,18 +1436,15 @@ call_start(struct rpc_task *task) static void call_reserve(struct rpc_task *task) { - dprintk("RPC: %4d call_reserve\n", task->tk_pid); - - if (!rpcauth_uptodatecred(task)) { - task->tk_action = call_refresh; - return; - } + dprint_status(task); task->tk_status = 0; task->tk_action = call_reserveresult; xprt_reserve(task); } +static void call_retry_reserve(struct rpc_task *task); + /* * 1b. Grok the result of xprt_reserve() */ @@ -595,8 +1453,7 @@ call_reserveresult(struct rpc_task *task) { int status = task->tk_status; - dprintk("RPC: %4d call_reserveresult (status %d)\n", - task->tk_pid, task->tk_status); + dprint_status(task); /* * After a call to xprt_reserve(), we must have either @@ -605,12 +1462,12 @@ call_reserveresult(struct rpc_task *task) task->tk_status = 0; if (status >= 0) { if (task->tk_rqstp) { - task->tk_action = call_allocate; + task->tk_action = call_refresh; return; } printk(KERN_ERR "%s: status=%d, but no request slot, exiting\n", - __FUNCTION__, status); + __func__, status); rpc_exit(task, -EIO); return; } @@ -621,50 +1478,136 @@ call_reserveresult(struct rpc_task *task) */ if (task->tk_rqstp) { printk(KERN_ERR "%s: status=%d, request allocated anyway\n", - __FUNCTION__, status); + __func__, status); xprt_release(task); } switch (status) { + case -ENOMEM: + rpc_delay(task, HZ >> 2); case -EAGAIN: /* woken up; retry */ - task->tk_action = call_reserve; + task->tk_action = call_retry_reserve; return; case -EIO: /* probably a shutdown */ break; default: printk(KERN_ERR "%s: unrecognized error %d, exiting\n", - __FUNCTION__, status); + __func__, status); break; } rpc_exit(task, status); } /* - * 2. Allocate the buffer. For details, see sched.c:rpc_malloc. - * (Note: buffer memory is freed in rpc_task_release). + * 1c. Retry reserving an RPC call slot + */ +static void +call_retry_reserve(struct rpc_task *task) +{ + dprint_status(task); + + task->tk_status = 0; + task->tk_action = call_reserveresult; + xprt_retry_reserve(task); +} + +/* + * 2. Bind and/or refresh the credentials + */ +static void +call_refresh(struct rpc_task *task) +{ + dprint_status(task); + + task->tk_action = call_refreshresult; + task->tk_status = 0; + task->tk_client->cl_stats->rpcauthrefresh++; + rpcauth_refreshcred(task); +} + +/* + * 2a. Process the results of a credential refresh + */ +static void +call_refreshresult(struct rpc_task *task) +{ + int status = task->tk_status; + + dprint_status(task); + + task->tk_status = 0; + task->tk_action = call_refresh; + switch (status) { + case 0: + if (rpcauth_uptodatecred(task)) { + task->tk_action = call_allocate; + return; + } + /* Use rate-limiting and a max number of retries if refresh + * had status 0 but failed to update the cred. + */ + case -ETIMEDOUT: + rpc_delay(task, 3*HZ); + case -EAGAIN: + status = -EACCES; + case -EKEYEXPIRED: + if (!task->tk_cred_retry) + break; + task->tk_cred_retry--; + dprintk("RPC: %5u %s: retry refresh creds\n", + task->tk_pid, __func__); + return; + } + dprintk("RPC: %5u %s: refresh creds failed with error %d\n", + task->tk_pid, __func__, status); + rpc_exit(task, status); +} + +/* + * 2b. Allocate the buffer. For details, see sched.c:rpc_malloc. + * (Note: buffer memory is freed in xprt_release). */ static void call_allocate(struct rpc_task *task) { - unsigned int bufsiz; + unsigned int slack = task->tk_rqstp->rq_cred->cr_auth->au_cslack; + struct rpc_rqst *req = task->tk_rqstp; + struct rpc_xprt *xprt = req->rq_xprt; + struct rpc_procinfo *proc = task->tk_msg.rpc_proc; - dprintk("RPC: %4d call_allocate (status %d)\n", - task->tk_pid, task->tk_status); + dprint_status(task); + + task->tk_status = 0; task->tk_action = call_bind; - if (task->tk_buffer) + + if (req->rq_buffer) return; - /* FIXME: compute buffer requirements more exactly using - * auth->au_wslack */ - bufsiz = task->tk_msg.rpc_proc->p_bufsiz + RPC_SLACK_SPACE; + if (proc->p_proc != 0) { + BUG_ON(proc->p_arglen == 0); + if (proc->p_decode != NULL) + BUG_ON(proc->p_replen == 0); + } - if (rpc_malloc(task, bufsiz << 1) != NULL) + /* + * Calculate the size (in quads) of the RPC call + * and reply headers, and convert both values + * to byte sizes. + */ + req->rq_callsize = RPC_CALLHDRSIZE + (slack << 1) + proc->p_arglen; + req->rq_callsize <<= 2; + req->rq_rcvsize = RPC_REPHDRSIZE + slack + proc->p_replen; + req->rq_rcvsize <<= 2; + + req->rq_buffer = xprt->ops->buf_alloc(task, + req->rq_callsize + req->rq_rcvsize); + if (req->rq_buffer != NULL) return; - printk(KERN_INFO "RPC: buffer allocation failed for task %p\n", task); - if (RPC_IS_ASYNC(task) || !signalled()) { - xprt_release(task); - task->tk_action = call_reserve; + dprintk("RPC: %5u rpc_buffer allocation failed\n", task->tk_pid); + + if (RPC_IS_ASYNC(task) || !fatal_signal_pending(current)) { + task->tk_action = call_allocate; rpc_delay(task, HZ>>4); return; } @@ -672,54 +1615,63 @@ call_allocate(struct rpc_task *task) rpc_exit(task, -ERESTARTSYS); } +static inline int +rpc_task_need_encode(struct rpc_task *task) +{ + return task->tk_rqstp->rq_snd_buf.len == 0; +} + +static inline void +rpc_task_force_reencode(struct rpc_task *task) +{ + task->tk_rqstp->rq_snd_buf.len = 0; + task->tk_rqstp->rq_bytes_sent = 0; +} + +static inline void +rpc_xdr_buf_init(struct xdr_buf *buf, void *start, size_t len) +{ + buf->head[0].iov_base = start; + buf->head[0].iov_len = len; + buf->tail[0].iov_len = 0; + buf->page_len = 0; + buf->flags = 0; + buf->len = 0; + buf->buflen = len; +} + /* * 3. Encode arguments of an RPC call */ static void -call_encode(struct rpc_task *task) +rpc_xdr_encode(struct rpc_task *task) { struct rpc_rqst *req = task->tk_rqstp; - struct xdr_buf *sndbuf = &req->rq_snd_buf; - struct xdr_buf *rcvbuf = &req->rq_rcv_buf; - unsigned int bufsiz; - kxdrproc_t encode; - u32 *p; + kxdreproc_t encode; + __be32 *p; - dprintk("RPC: %4d call_encode (status %d)\n", - task->tk_pid, task->tk_status); + dprint_status(task); - /* Default buffer setup */ - bufsiz = task->tk_bufsize >> 1; - sndbuf->head[0].iov_base = (void *)task->tk_buffer; - sndbuf->head[0].iov_len = bufsiz; - sndbuf->tail[0].iov_len = 0; - sndbuf->page_len = 0; - sndbuf->len = 0; - sndbuf->buflen = bufsiz; - rcvbuf->head[0].iov_base = (void *)((char *)task->tk_buffer + bufsiz); - rcvbuf->head[0].iov_len = bufsiz; - rcvbuf->tail[0].iov_len = 0; - rcvbuf->page_len = 0; - rcvbuf->len = 0; - rcvbuf->buflen = bufsiz; - - /* Encode header and provided arguments */ - encode = task->tk_msg.rpc_proc->p_encode; - if (!(p = call_header(task))) { - printk(KERN_INFO "RPC: call_header failed, exit EIO\n"); + rpc_xdr_buf_init(&req->rq_snd_buf, + req->rq_buffer, + req->rq_callsize); + rpc_xdr_buf_init(&req->rq_rcv_buf, + (char *)req->rq_buffer + req->rq_callsize, + req->rq_rcvsize); + + p = rpc_encode_header(task); + if (p == NULL) { + printk(KERN_INFO "RPC: couldn't encode RPC header, exit EIO\n"); rpc_exit(task, -EIO); return; } + + encode = task->tk_msg.rpc_proc->p_encode; if (encode == NULL) return; task->tk_status = rpcauth_wrap_req(task, encode, req, p, task->tk_msg.rpc_argp); - if (task->tk_status == -ENOMEM) { - /* XXX: Is this sane? */ - rpc_delay(task, 3*HZ); - task->tk_status = -EAGAIN; - } } /* @@ -728,16 +1680,15 @@ call_encode(struct rpc_task *task) static void call_bind(struct rpc_task *task) { - struct rpc_clnt *clnt = task->tk_client; + struct rpc_xprt *xprt = task->tk_rqstp->rq_xprt; - dprintk("RPC: %4d call_bind (status %d)\n", - task->tk_pid, task->tk_status); + dprint_status(task); task->tk_action = call_connect; - if (!clnt->cl_port) { + if (!xprt_bound(xprt)) { task->tk_action = call_bind_status; - task->tk_timeout = task->tk_xprt->bind_timeout; - rpc_getport(task, clnt); + task->tk_timeout = xprt->bind_timeout; + xprt->ops->rpcbind(task); } } @@ -747,52 +1698,74 @@ call_bind(struct rpc_task *task) static void call_bind_status(struct rpc_task *task) { - int status = -EACCES; + int status = -EIO; if (task->tk_status >= 0) { - dprintk("RPC: %4d call_bind_status (status %d)\n", - task->tk_pid, task->tk_status); + dprint_status(task); task->tk_status = 0; task->tk_action = call_connect; return; } + trace_rpc_bind_status(task); switch (task->tk_status) { + case -ENOMEM: + dprintk("RPC: %5u rpcbind out of memory\n", task->tk_pid); + rpc_delay(task, HZ >> 2); + goto retry_timeout; case -EACCES: - dprintk("RPC: %4d remote rpcbind: RPC program/version unavailable\n", - task->tk_pid); + dprintk("RPC: %5u remote rpcbind: RPC program/version " + "unavailable\n", task->tk_pid); + /* fail immediately if this is an RPC ping */ + if (task->tk_msg.rpc_proc->p_proc == 0) { + status = -EOPNOTSUPP; + break; + } + if (task->tk_rebind_retry == 0) + break; + task->tk_rebind_retry--; rpc_delay(task, 3*HZ); - goto retry_bind; + goto retry_timeout; case -ETIMEDOUT: - dprintk("RPC: %4d rpcbind request timed out\n", + dprintk("RPC: %5u rpcbind request timed out\n", task->tk_pid); - if (RPC_IS_SOFT(task)) { - status = -EIO; - break; - } - goto retry_bind; + goto retry_timeout; case -EPFNOSUPPORT: - dprintk("RPC: %4d remote rpcbind service unavailable\n", + /* server doesn't support any rpcbind version we know of */ + dprintk("RPC: %5u unrecognized remote rpcbind service\n", task->tk_pid); break; case -EPROTONOSUPPORT: - dprintk("RPC: %4d remote rpcbind version 2 unavailable\n", + dprintk("RPC: %5u remote rpcbind version unavailable, retrying\n", task->tk_pid); + goto retry_timeout; + case -ECONNREFUSED: /* connection problems */ + case -ECONNRESET: + case -ECONNABORTED: + case -ENOTCONN: + case -EHOSTDOWN: + case -EHOSTUNREACH: + case -ENETUNREACH: + case -EPIPE: + dprintk("RPC: %5u remote rpcbind unreachable: %d\n", + task->tk_pid, task->tk_status); + if (!RPC_IS_SOFTCONN(task)) { + rpc_delay(task, 5*HZ); + goto retry_timeout; + } + status = task->tk_status; break; default: - dprintk("RPC: %4d unrecognized rpcbind error (%d)\n", + dprintk("RPC: %5u unrecognized rpcbind error (%d)\n", task->tk_pid, -task->tk_status); - status = -EIO; - break; } rpc_exit(task, status); return; -retry_bind: +retry_timeout: task->tk_status = 0; - task->tk_action = call_bind; - return; + task->tk_action = call_timeout; } /* @@ -801,9 +1774,9 @@ retry_bind: static void call_connect(struct rpc_task *task) { - struct rpc_xprt *xprt = task->tk_xprt; + struct rpc_xprt *xprt = task->tk_rqstp->rq_xprt; - dprintk("RPC: %4d call_connect xprt %p %s connected\n", + dprintk("RPC: %5u call_connect xprt %p %s connected\n", task->tk_pid, xprt, (xprt_connected(xprt) ? "is" : "is not")); @@ -812,6 +1785,10 @@ call_connect(struct rpc_task *task) task->tk_action = call_connect_status; if (task->tk_status < 0) return; + if (task->tk_flags & RPC_TASK_NOCONNECT) { + rpc_exit(task, -ENOTCONN); + return; + } xprt_connect(task); } } @@ -825,30 +1802,31 @@ call_connect_status(struct rpc_task *task) struct rpc_clnt *clnt = task->tk_client; int status = task->tk_status; - dprintk("RPC: %5u call_connect_status (status %d)\n", - task->tk_pid, task->tk_status); + dprint_status(task); + trace_rpc_connect_status(task, status); task->tk_status = 0; - if (status >= 0) { + switch (status) { + case -ECONNREFUSED: + case -ECONNRESET: + case -ECONNABORTED: + case -ENETUNREACH: + case -EHOSTUNREACH: + if (RPC_IS_SOFTCONN(task)) + break; + /* retry with existing socket, after a delay */ + rpc_delay(task, 3*HZ); + case -EAGAIN: + /* Check for timeouts before looping back to call_bind */ + case -ETIMEDOUT: + task->tk_action = call_timeout; + return; + case 0: clnt->cl_stats->netreconn++; task->tk_action = call_transmit; return; } - - /* Something failed: remote service port may have changed */ - if (clnt->cl_autobind) - clnt->cl_port = 0; - - switch (status) { - case -ENOTCONN: - case -ETIMEDOUT: - case -EAGAIN: - task->tk_action = call_bind; - break; - default: - rpc_exit(task, -EIO); - break; - } + rpc_exit(task, status); } /* @@ -857,34 +1835,157 @@ call_connect_status(struct rpc_task *task) static void call_transmit(struct rpc_task *task) { - dprintk("RPC: %4d call_transmit (status %d)\n", - task->tk_pid, task->tk_status); + int is_retrans = RPC_WAS_SENT(task); + + dprint_status(task); task->tk_action = call_status; if (task->tk_status < 0) return; - task->tk_status = xprt_prepare_transmit(task); - if (task->tk_status != 0) + if (!xprt_prepare_transmit(task)) return; + task->tk_action = call_transmit_status; /* Encode here so that rpcsec_gss can use correct sequence number. */ - if (task->tk_rqstp->rq_bytes_sent == 0) { - call_encode(task); + if (rpc_task_need_encode(task)) { + rpc_xdr_encode(task); /* Did the encode result in an error condition? */ - if (task->tk_status != 0) - goto out_nosend; + if (task->tk_status != 0) { + /* Was the error nonfatal? */ + if (task->tk_status == -EAGAIN) + rpc_delay(task, HZ >> 4); + else + rpc_exit(task, task->tk_status); + return; + } } xprt_transmit(task); if (task->tk_status < 0) return; - if (!task->tk_msg.rpc_proc->p_decode) { - task->tk_action = NULL; - rpc_wake_up_task(task); + if (is_retrans) + task->tk_client->cl_stats->rpcretrans++; + /* + * On success, ensure that we call xprt_end_transmit() before sleeping + * in order to allow access to the socket to other RPC requests. + */ + call_transmit_status(task); + if (rpc_reply_expected(task)) + return; + task->tk_action = rpc_exit_task; + rpc_wake_up_queued_task(&task->tk_rqstp->rq_xprt->pending, task); +} + +/* + * 5a. Handle cleanup after a transmission + */ +static void +call_transmit_status(struct rpc_task *task) +{ + task->tk_action = call_status; + + /* + * Common case: success. Force the compiler to put this + * test first. + */ + if (task->tk_status == 0) { + xprt_end_transmit(task); + rpc_task_force_reencode(task); + return; } - return; -out_nosend: - /* release socket write lock before attempting to handle error */ - xprt_abort_transmit(task); + + switch (task->tk_status) { + case -EAGAIN: + break; + default: + dprint_status(task); + xprt_end_transmit(task); + rpc_task_force_reencode(task); + break; + /* + * Special cases: if we've been waiting on the + * socket's write_space() callback, or if the + * socket just returned a connection error, + * then hold onto the transport lock. + */ + case -ECONNREFUSED: + case -EHOSTDOWN: + case -EHOSTUNREACH: + case -ENETUNREACH: + if (RPC_IS_SOFTCONN(task)) { + xprt_end_transmit(task); + rpc_exit(task, task->tk_status); + break; + } + case -ECONNRESET: + case -ECONNABORTED: + case -ENOTCONN: + case -EPIPE: + rpc_task_force_reencode(task); + } +} + +#if defined(CONFIG_SUNRPC_BACKCHANNEL) +/* + * 5b. Send the backchannel RPC reply. On error, drop the reply. In + * addition, disconnect on connectivity errors. + */ +static void +call_bc_transmit(struct rpc_task *task) +{ + struct rpc_rqst *req = task->tk_rqstp; + + if (!xprt_prepare_transmit(task)) { + /* + * Could not reserve the transport. Try again after the + * transport is released. + */ + task->tk_status = 0; + task->tk_action = call_bc_transmit; + return; + } + + task->tk_action = rpc_exit_task; + if (task->tk_status < 0) { + printk(KERN_NOTICE "RPC: Could not send backchannel reply " + "error: %d\n", task->tk_status); + return; + } + + xprt_transmit(task); + xprt_end_transmit(task); + dprint_status(task); + switch (task->tk_status) { + case 0: + /* Success */ + break; + case -EHOSTDOWN: + case -EHOSTUNREACH: + case -ENETUNREACH: + case -ETIMEDOUT: + /* + * Problem reaching the server. Disconnect and let the + * forechannel reestablish the connection. The server will + * have to retransmit the backchannel request and we'll + * reprocess it. Since these ops are idempotent, there's no + * need to cache our reply at this time. + */ + printk(KERN_NOTICE "RPC: Could not send backchannel reply " + "error: %d\n", task->tk_status); + xprt_conditional_disconnect(req->rq_xprt, + req->rq_connect_cookie); + break; + default: + /* + * We were unable to reply and will have to drop the + * request. The server should reconnect and retransmit. + */ + WARN_ON_ONCE(task->tk_status == -EAGAIN); + printk(KERN_NOTICE "RPC: Could not send backchannel reply " + "error: %d\n", task->tk_status); + break; + } + rpc_wake_up_queued_task(&req->rq_xprt->pending, task); } +#endif /* CONFIG_SUNRPC_BACKCHANNEL */ /* * 6. Sort out the RPC call status @@ -896,11 +1997,10 @@ call_status(struct rpc_task *task) struct rpc_rqst *req = task->tk_rqstp; int status; - if (req->rq_received > 0 && !req->rq_bytes_sent) - task->tk_status = req->rq_received; + if (req->rq_reply_bytes_recvd > 0 && !req->rq_bytes_sent) + task->tk_status = req->rq_reply_bytes_recvd; - dprintk("RPC: %4d call_status (status %d)\n", - task->tk_pid, task->tk_status); + dprint_status(task); status = task->tk_status; if (status >= 0) { @@ -908,16 +2008,35 @@ call_status(struct rpc_task *task) return; } + trace_rpc_call_status(task); task->tk_status = 0; switch(status) { + case -EHOSTDOWN: + case -EHOSTUNREACH: + case -ENETUNREACH: + if (RPC_IS_SOFTCONN(task)) { + rpc_exit(task, status); + break; + } + /* + * Delay any retries for 3 seconds, then handle as if it + * were a timeout. + */ + rpc_delay(task, 3*HZ); case -ETIMEDOUT: task->tk_action = call_timeout; + if (!(task->tk_flags & RPC_TASK_NO_RETRANS_TIMEOUT) + && task->tk_client->cl_discrtry) + xprt_conditional_disconnect(req->rq_xprt, + req->rq_connect_cookie); break; case -ECONNREFUSED: + case -ECONNRESET: + case -ECONNABORTED: + rpc_force_rebind(clnt); + rpc_delay(task, 3*HZ); + case -EPIPE: case -ENOTCONN: - req->rq_bytes_sent = 0; - if (clnt->cl_autobind) - clnt->cl_port = 0; task->tk_action = call_bind; break; case -EAGAIN: @@ -930,9 +2049,8 @@ call_status(struct rpc_task *task) default: if (clnt->cl_chatty) printk("%s: RPC call returned error %d\n", - clnt->cl_protname, -status); + clnt->cl_program->name, -status); rpc_exit(task, status); - break; } } @@ -947,29 +2065,50 @@ call_timeout(struct rpc_task *task) struct rpc_clnt *clnt = task->tk_client; if (xprt_adjust_timeout(task->tk_rqstp) == 0) { - dprintk("RPC: %4d call_timeout (minor)\n", task->tk_pid); + dprintk("RPC: %5u call_timeout (minor)\n", task->tk_pid); goto retry; } - dprintk("RPC: %4d call_timeout (major)\n", task->tk_pid); + dprintk("RPC: %5u call_timeout (major)\n", task->tk_pid); + task->tk_timeouts++; + + if (RPC_IS_SOFTCONN(task)) { + rpc_exit(task, -ETIMEDOUT); + return; + } if (RPC_IS_SOFT(task)) { - if (clnt->cl_chatty) + if (clnt->cl_chatty) { + rcu_read_lock(); printk(KERN_NOTICE "%s: server %s not responding, timed out\n", - clnt->cl_protname, clnt->cl_server); - rpc_exit(task, -EIO); + clnt->cl_program->name, + rcu_dereference(clnt->cl_xprt)->servername); + rcu_read_unlock(); + } + if (task->tk_flags & RPC_TASK_TIMEOUT) + rpc_exit(task, -ETIMEDOUT); + else + rpc_exit(task, -EIO); return; } - if (clnt->cl_chatty && !(task->tk_flags & RPC_CALL_MAJORSEEN)) { + if (!(task->tk_flags & RPC_CALL_MAJORSEEN)) { task->tk_flags |= RPC_CALL_MAJORSEEN; - printk(KERN_NOTICE "%s: server %s not responding, still trying\n", - clnt->cl_protname, clnt->cl_server); + if (clnt->cl_chatty) { + rcu_read_lock(); + printk(KERN_NOTICE "%s: server %s not responding, still trying\n", + clnt->cl_program->name, + rcu_dereference(clnt->cl_xprt)->servername); + rcu_read_unlock(); + } } - if (clnt->cl_autobind) - clnt->cl_port = 0; + rpc_force_rebind(clnt); + /* + * Did our request time out due to an RPCSEC_GSS out-of-sequence + * event? RFC2203 requires the server to drop all such requests. + */ + rpcauth_invalcred(task); retry: - clnt->cl_stats->rpcretrans++; task->tk_action = call_bind; task->tk_status = 0; } @@ -982,108 +2121,81 @@ call_decode(struct rpc_task *task) { struct rpc_clnt *clnt = task->tk_client; struct rpc_rqst *req = task->tk_rqstp; - kxdrproc_t decode = task->tk_msg.rpc_proc->p_decode; - u32 *p; - - dprintk("RPC: %4d call_decode (status %d)\n", - task->tk_pid, task->tk_status); - - if (clnt->cl_chatty && (task->tk_flags & RPC_CALL_MAJORSEEN)) { - printk(KERN_NOTICE "%s: server %s OK\n", - clnt->cl_protname, clnt->cl_server); - task->tk_flags &= ~RPC_CALL_MAJORSEEN; - } - - if (task->tk_status < 12) { - if (!RPC_IS_SOFT(task)) { - task->tk_action = call_bind; - clnt->cl_stats->rpcretrans++; - goto out_retry; + kxdrdproc_t decode = task->tk_msg.rpc_proc->p_decode; + __be32 *p; + + dprint_status(task); + + if (task->tk_flags & RPC_CALL_MAJORSEEN) { + if (clnt->cl_chatty) { + rcu_read_lock(); + printk(KERN_NOTICE "%s: server %s OK\n", + clnt->cl_program->name, + rcu_dereference(clnt->cl_xprt)->servername); + rcu_read_unlock(); } - printk(KERN_WARNING "%s: too small RPC reply size (%d bytes)\n", - clnt->cl_protname, task->tk_status); - rpc_exit(task, -EIO); - return; + task->tk_flags &= ~RPC_CALL_MAJORSEEN; } + /* + * Ensure that we see all writes made by xprt_complete_rqst() + * before it changed req->rq_reply_bytes_recvd. + */ + smp_rmb(); req->rq_rcv_buf.len = req->rq_private_buf.len; /* Check that the softirq receive buffer is valid */ WARN_ON(memcmp(&req->rq_rcv_buf, &req->rq_private_buf, sizeof(req->rq_rcv_buf)) != 0); - /* Verify the RPC header */ - if (!(p = call_verify(task))) { - if (task->tk_action == NULL) - return; + if (req->rq_rcv_buf.len < 12) { + if (!RPC_IS_SOFT(task)) { + task->tk_action = call_bind; + goto out_retry; + } + dprintk("RPC: %s: too small RPC reply size (%d bytes)\n", + clnt->cl_program->name, task->tk_status); + task->tk_action = call_timeout; goto out_retry; } - task->tk_action = NULL; + p = rpc_verify_header(task); + if (IS_ERR(p)) { + if (p == ERR_PTR(-EAGAIN)) + goto out_retry; + return; + } + + task->tk_action = rpc_exit_task; - if (decode) + if (decode) { task->tk_status = rpcauth_unwrap_resp(task, decode, req, p, task->tk_msg.rpc_resp); - dprintk("RPC: %4d call_decode result %d\n", task->tk_pid, - task->tk_status); + } + dprintk("RPC: %5u call_decode result %d\n", task->tk_pid, + task->tk_status); return; out_retry: - req->rq_received = req->rq_private_buf.len = 0; - task->tk_status = 0; -} - -/* - * 8. Refresh the credentials if rejected by the server - */ -static void -call_refresh(struct rpc_task *task) -{ - dprintk("RPC: %4d call_refresh\n", task->tk_pid); - - xprt_release(task); /* Must do to obtain new XID */ - task->tk_action = call_refreshresult; task->tk_status = 0; - task->tk_client->cl_stats->rpcauthrefresh++; - rpcauth_refreshcred(task); -} - -/* - * 8a. Process the results of a credential refresh - */ -static void -call_refreshresult(struct rpc_task *task) -{ - int status = task->tk_status; - dprintk("RPC: %4d call_refreshresult (status %d)\n", - task->tk_pid, task->tk_status); - - task->tk_status = 0; - task->tk_action = call_reserve; - if (status >= 0 && rpcauth_uptodatecred(task)) - return; - if (status == -EACCES) { - rpc_exit(task, -EACCES); - return; + /* Note: rpc_verify_header() may have freed the RPC slot */ + if (task->tk_rqstp == req) { + req->rq_reply_bytes_recvd = req->rq_rcv_buf.len = 0; + if (task->tk_client->cl_discrtry) + xprt_conditional_disconnect(req->rq_xprt, + req->rq_connect_cookie); } - task->tk_action = call_refresh; - if (status != -ETIMEDOUT) - rpc_delay(task, 3*HZ); - return; } -/* - * Call header serialization - */ -static u32 * -call_header(struct rpc_task *task) +static __be32 * +rpc_encode_header(struct rpc_task *task) { struct rpc_clnt *clnt = task->tk_client; struct rpc_rqst *req = task->tk_rqstp; - u32 *p = req->rq_svec[0].iov_base; + __be32 *p = req->rq_svec[0].iov_base; /* FIXME: check buffer size? */ - p = xprt_skip_transport_header(task->tk_xprt, p); + p = xprt_skip_transport_header(req->rq_xprt, p); *p++ = req->rq_xid; /* XID */ *p++ = htonl(RPC_CALL); /* CALL */ *p++ = htonl(RPC_VERSION); /* RPC version */ @@ -1095,38 +2207,56 @@ call_header(struct rpc_task *task) return p; } -/* - * Reply header verification - */ -static u32 * -call_verify(struct rpc_task *task) +static __be32 * +rpc_verify_header(struct rpc_task *task) { + struct rpc_clnt *clnt = task->tk_client; struct kvec *iov = &task->tk_rqstp->rq_rcv_buf.head[0]; int len = task->tk_rqstp->rq_rcv_buf.len >> 2; - u32 *p = iov->iov_base, n; + __be32 *p = iov->iov_base; + u32 n; int error = -EACCES; + if ((task->tk_rqstp->rq_rcv_buf.len & 3) != 0) { + /* RFC-1014 says that the representation of XDR data must be a + * multiple of four bytes + * - if it isn't pointer subtraction in the NFS client may give + * undefined results + */ + dprintk("RPC: %5u %s: XDR representation not a multiple of" + " 4 bytes: 0x%x\n", task->tk_pid, __func__, + task->tk_rqstp->rq_rcv_buf.len); + error = -EIO; + goto out_err; + } if ((len -= 3) < 0) goto out_overflow; - p += 1; /* skip XID */ + p += 1; /* skip XID */ if ((n = ntohl(*p++)) != RPC_REPLY) { - printk(KERN_WARNING "call_verify: not an RPC reply: %x\n", n); - goto out_retry; + dprintk("RPC: %5u %s: not an RPC reply: %x\n", + task->tk_pid, __func__, n); + error = -EIO; + goto out_garbage; } + if ((n = ntohl(*p++)) != RPC_MSG_ACCEPTED) { if (--len < 0) goto out_overflow; switch ((n = ntohl(*p++))) { - case RPC_AUTH_ERROR: - break; - case RPC_MISMATCH: - dprintk("%s: RPC call version mismatch!\n", __FUNCTION__); - error = -EPROTONOSUPPORT; - goto out_err; - default: - dprintk("%s: RPC call rejected, unknown error: %x\n", __FUNCTION__, n); - goto out_eio; + case RPC_AUTH_ERROR: + break; + case RPC_MISMATCH: + dprintk("RPC: %5u %s: RPC call version mismatch!\n", + task->tk_pid, __func__); + error = -EPROTONOSUPPORT; + goto out_err; + default: + dprintk("RPC: %5u %s: RPC call rejected, " + "unknown error: %x\n", + task->tk_pid, __func__, n); + error = -EIO; + goto out_err; } if (--len < 0) goto out_overflow; @@ -1138,97 +2268,112 @@ call_verify(struct rpc_task *task) if (!task->tk_cred_retry) break; task->tk_cred_retry--; - dprintk("RPC: %4d call_verify: retry stale creds\n", - task->tk_pid); + dprintk("RPC: %5u %s: retry stale creds\n", + task->tk_pid, __func__); rpcauth_invalcred(task); - task->tk_action = call_refresh; - return NULL; + /* Ensure we obtain a new XID! */ + xprt_release(task); + task->tk_action = call_reserve; + goto out_retry; case RPC_AUTH_BADCRED: case RPC_AUTH_BADVERF: /* possibly garbled cred/verf? */ if (!task->tk_garb_retry) break; task->tk_garb_retry--; - dprintk("RPC: %4d call_verify: retry garbled creds\n", - task->tk_pid); + dprintk("RPC: %5u %s: retry garbled creds\n", + task->tk_pid, __func__); task->tk_action = call_bind; - return NULL; + goto out_retry; case RPC_AUTH_TOOWEAK: - printk(KERN_NOTICE "call_verify: server requires stronger " - "authentication.\n"); + rcu_read_lock(); + printk(KERN_NOTICE "RPC: server %s requires stronger " + "authentication.\n", + rcu_dereference(clnt->cl_xprt)->servername); + rcu_read_unlock(); break; default: - printk(KERN_WARNING "call_verify: unknown auth error: %x\n", n); + dprintk("RPC: %5u %s: unknown auth error: %x\n", + task->tk_pid, __func__, n); error = -EIO; } - dprintk("RPC: %4d call_verify: call rejected %d\n", - task->tk_pid, n); + dprintk("RPC: %5u %s: call rejected %d\n", + task->tk_pid, __func__, n); goto out_err; } - if (!(p = rpcauth_checkverf(task, p))) { - printk(KERN_WARNING "call_verify: auth check failed\n"); - goto out_retry; /* bad verifier, retry */ + p = rpcauth_checkverf(task, p); + if (IS_ERR(p)) { + error = PTR_ERR(p); + dprintk("RPC: %5u %s: auth check failed with %d\n", + task->tk_pid, __func__, error); + goto out_garbage; /* bad verifier, retry */ } - len = p - (u32 *)iov->iov_base - 1; + len = p - (__be32 *)iov->iov_base - 1; if (len < 0) goto out_overflow; switch ((n = ntohl(*p++))) { case RPC_SUCCESS: return p; case RPC_PROG_UNAVAIL: - dprintk("RPC: call_verify: program %u is unsupported by server %s\n", - (unsigned int)task->tk_client->cl_prog, - task->tk_client->cl_server); + dprintk_rcu("RPC: %5u %s: program %u is unsupported " + "by server %s\n", task->tk_pid, __func__, + (unsigned int)clnt->cl_prog, + rcu_dereference(clnt->cl_xprt)->servername); error = -EPFNOSUPPORT; goto out_err; case RPC_PROG_MISMATCH: - dprintk("RPC: call_verify: program %u, version %u unsupported by server %s\n", - (unsigned int)task->tk_client->cl_prog, - (unsigned int)task->tk_client->cl_vers, - task->tk_client->cl_server); + dprintk_rcu("RPC: %5u %s: program %u, version %u unsupported " + "by server %s\n", task->tk_pid, __func__, + (unsigned int)clnt->cl_prog, + (unsigned int)clnt->cl_vers, + rcu_dereference(clnt->cl_xprt)->servername); error = -EPROTONOSUPPORT; goto out_err; case RPC_PROC_UNAVAIL: - dprintk("RPC: call_verify: proc %p unsupported by program %u, version %u on server %s\n", - task->tk_msg.rpc_proc, - task->tk_client->cl_prog, - task->tk_client->cl_vers, - task->tk_client->cl_server); + dprintk_rcu("RPC: %5u %s: proc %s unsupported by program %u, " + "version %u on server %s\n", + task->tk_pid, __func__, + rpc_proc_name(task), + clnt->cl_prog, clnt->cl_vers, + rcu_dereference(clnt->cl_xprt)->servername); error = -EOPNOTSUPP; goto out_err; case RPC_GARBAGE_ARGS: - dprintk("RPC: %4d %s: server saw garbage\n", task->tk_pid, __FUNCTION__); + dprintk("RPC: %5u %s: server saw garbage\n", + task->tk_pid, __func__); break; /* retry */ default: - printk(KERN_WARNING "call_verify: server accept status: %x\n", n); + dprintk("RPC: %5u %s: server accept status: %x\n", + task->tk_pid, __func__, n); /* Also retry */ } -out_retry: - task->tk_client->cl_stats->rpcgarbage++; +out_garbage: + clnt->cl_stats->rpcgarbage++; if (task->tk_garb_retry) { task->tk_garb_retry--; - dprintk("RPC %s: retrying %4d\n", __FUNCTION__, task->tk_pid); + dprintk("RPC: %5u %s: retrying\n", + task->tk_pid, __func__); task->tk_action = call_bind; - return NULL; +out_retry: + return ERR_PTR(-EAGAIN); } - printk(KERN_WARNING "RPC %s: retry failed, exit EIO\n", __FUNCTION__); -out_eio: - error = -EIO; out_err: rpc_exit(task, error); - return NULL; + dprintk("RPC: %5u %s: call failed with error %d\n", task->tk_pid, + __func__, error); + return ERR_PTR(error); out_overflow: - printk(KERN_WARNING "RPC %s: server reply was truncated.\n", __FUNCTION__); - goto out_retry; + dprintk("RPC: %5u %s: server reply was truncated.\n", task->tk_pid, + __func__); + goto out_garbage; } -static int rpcproc_encode_null(void *rqstp, u32 *data, void *obj) +static void rpcproc_encode_null(void *rqstp, struct xdr_stream *xdr, void *obj) { - return 0; } -static int rpcproc_decode_null(void *rqstp, u32 *data, void *obj) +static int rpcproc_decode_null(void *rqstp, struct xdr_stream *xdr, void *obj) { return 0; } @@ -1238,14 +2383,75 @@ static struct rpc_procinfo rpcproc_null = { .p_decode = rpcproc_decode_null, }; -int rpc_ping(struct rpc_clnt *clnt, int flags) +static int rpc_ping(struct rpc_clnt *clnt) { struct rpc_message msg = { .rpc_proc = &rpcproc_null, }; int err; msg.rpc_cred = authnull_ops.lookup_cred(NULL, NULL, 0); - err = rpc_call_sync(clnt, &msg, flags); + err = rpc_call_sync(clnt, &msg, RPC_TASK_SOFT | RPC_TASK_SOFTCONN); put_rpccred(msg.rpc_cred); return err; } + +struct rpc_task *rpc_call_null(struct rpc_clnt *clnt, struct rpc_cred *cred, int flags) +{ + struct rpc_message msg = { + .rpc_proc = &rpcproc_null, + .rpc_cred = cred, + }; + struct rpc_task_setup task_setup_data = { + .rpc_client = clnt, + .rpc_message = &msg, + .callback_ops = &rpc_default_ops, + .flags = flags, + }; + return rpc_run_task(&task_setup_data); +} +EXPORT_SYMBOL_GPL(rpc_call_null); + +#ifdef RPC_DEBUG +static void rpc_show_header(void) +{ + printk(KERN_INFO "-pid- flgs status -client- --rqstp- " + "-timeout ---ops--\n"); +} + +static void rpc_show_task(const struct rpc_clnt *clnt, + const struct rpc_task *task) +{ + const char *rpc_waitq = "none"; + + if (RPC_IS_QUEUED(task)) + rpc_waitq = rpc_qname(task->tk_waitqueue); + + printk(KERN_INFO "%5u %04x %6d %8p %8p %8ld %8p %sv%u %s a:%ps q:%s\n", + task->tk_pid, task->tk_flags, task->tk_status, + clnt, task->tk_rqstp, task->tk_timeout, task->tk_ops, + clnt->cl_program->name, clnt->cl_vers, rpc_proc_name(task), + task->tk_action, rpc_waitq); +} + +void rpc_show_tasks(struct net *net) +{ + struct rpc_clnt *clnt; + struct rpc_task *task; + int header = 0; + struct sunrpc_net *sn = net_generic(net, sunrpc_net_id); + + spin_lock(&sn->rpc_client_lock); + list_for_each_entry(clnt, &sn->all_clients, cl_clients) { + spin_lock(&clnt->cl_lock); + list_for_each_entry(task, &clnt->cl_tasks, tk_task) { + if (!header) { + rpc_show_header(); + header++; + } + rpc_show_task(clnt, task); + } + spin_unlock(&clnt->cl_lock); + } + spin_unlock(&sn->rpc_client_lock); +} +#endif diff --git a/net/sunrpc/netns.h b/net/sunrpc/netns.h new file mode 100644 index 00000000000..df582687653 --- /dev/null +++ b/net/sunrpc/netns.h @@ -0,0 +1,42 @@ +#ifndef __SUNRPC_NETNS_H__ +#define __SUNRPC_NETNS_H__ + +#include <net/net_namespace.h> +#include <net/netns/generic.h> + +struct cache_detail; + +struct sunrpc_net { + struct proc_dir_entry *proc_net_rpc; + struct cache_detail *ip_map_cache; + struct cache_detail *unix_gid_cache; + struct cache_detail *rsc_cache; + struct cache_detail *rsi_cache; + + struct super_block *pipefs_sb; + struct rpc_pipe *gssd_dummy; + struct mutex pipefs_sb_lock; + + struct list_head all_clients; + spinlock_t rpc_client_lock; + + struct rpc_clnt *rpcb_local_clnt; + struct rpc_clnt *rpcb_local_clnt4; + spinlock_t rpcb_clnt_lock; + unsigned int rpcb_users; + unsigned int rpcb_is_af_local : 1; + + struct mutex gssp_lock; + struct rpc_clnt *gssp_clnt; + int use_gss_proxy; + int pipe_version; + atomic_t pipe_users; + struct proc_dir_entry *use_gssp_proc; +}; + +extern int sunrpc_net_id; + +int ip_map_cache_create(struct net *); +void ip_map_cache_destroy(struct net *); + +#endif diff --git a/net/sunrpc/pmap_clnt.c b/net/sunrpc/pmap_clnt.c deleted file mode 100644 index a398575f94b..00000000000 --- a/net/sunrpc/pmap_clnt.c +++ /dev/null @@ -1,301 +0,0 @@ -/* - * linux/net/sunrpc/pmap.c - * - * Portmapper client. - * - * Copyright (C) 1996, Olaf Kirch <okir@monad.swb.de> - */ - -#include <linux/config.h> -#include <linux/types.h> -#include <linux/socket.h> -#include <linux/kernel.h> -#include <linux/errno.h> -#include <linux/uio.h> -#include <linux/in.h> -#include <linux/sunrpc/clnt.h> -#include <linux/sunrpc/xprt.h> -#include <linux/sunrpc/sched.h> - -#ifdef RPC_DEBUG -# define RPCDBG_FACILITY RPCDBG_PMAP -#endif - -#define PMAP_SET 1 -#define PMAP_UNSET 2 -#define PMAP_GETPORT 3 - -static struct rpc_procinfo pmap_procedures[]; -static struct rpc_clnt * pmap_create(char *, struct sockaddr_in *, int, int); -static void pmap_getport_done(struct rpc_task *); -static struct rpc_program pmap_program; -static DEFINE_SPINLOCK(pmap_lock); - -/* - * Obtain the port for a given RPC service on a given host. This one can - * be called for an ongoing RPC request. - */ -void -rpc_getport(struct rpc_task *task, struct rpc_clnt *clnt) -{ - struct rpc_portmap *map = clnt->cl_pmap; - struct sockaddr_in *sap = &clnt->cl_xprt->addr; - struct rpc_message msg = { - .rpc_proc = &pmap_procedures[PMAP_GETPORT], - .rpc_argp = map, - .rpc_resp = &clnt->cl_port, - .rpc_cred = NULL - }; - struct rpc_clnt *pmap_clnt; - struct rpc_task *child; - - dprintk("RPC: %4d rpc_getport(%s, %d, %d, %d)\n", - task->tk_pid, clnt->cl_server, - map->pm_prog, map->pm_vers, map->pm_prot); - - /* Autobind on cloned rpc clients is discouraged */ - BUG_ON(clnt->cl_parent != clnt); - - spin_lock(&pmap_lock); - if (map->pm_binding) { - rpc_sleep_on(&map->pm_bindwait, task, NULL, NULL); - spin_unlock(&pmap_lock); - return; - } - map->pm_binding = 1; - spin_unlock(&pmap_lock); - - pmap_clnt = pmap_create(clnt->cl_server, sap, map->pm_prot, 0); - if (IS_ERR(pmap_clnt)) { - task->tk_status = PTR_ERR(pmap_clnt); - goto bailout; - } - task->tk_status = 0; - - /* - * Note: rpc_new_child will release client after a failure. - */ - if (!(child = rpc_new_child(pmap_clnt, task))) - goto bailout; - - /* Setup the call info struct */ - rpc_call_setup(child, &msg, 0); - - /* ... and run the child task */ - rpc_run_child(task, child, pmap_getport_done); - return; - -bailout: - spin_lock(&pmap_lock); - map->pm_binding = 0; - rpc_wake_up(&map->pm_bindwait); - spin_unlock(&pmap_lock); - task->tk_status = -EIO; - task->tk_action = NULL; -} - -#ifdef CONFIG_ROOT_NFS -int -rpc_getport_external(struct sockaddr_in *sin, __u32 prog, __u32 vers, int prot) -{ - struct rpc_portmap map = { - .pm_prog = prog, - .pm_vers = vers, - .pm_prot = prot, - .pm_port = 0 - }; - struct rpc_clnt *pmap_clnt; - char hostname[32]; - int status; - - dprintk("RPC: rpc_getport_external(%u.%u.%u.%u, %d, %d, %d)\n", - NIPQUAD(sin->sin_addr.s_addr), prog, vers, prot); - - sprintf(hostname, "%u.%u.%u.%u", NIPQUAD(sin->sin_addr.s_addr)); - pmap_clnt = pmap_create(hostname, sin, prot, 0); - if (IS_ERR(pmap_clnt)) - return PTR_ERR(pmap_clnt); - - /* Setup the call info struct */ - status = rpc_call(pmap_clnt, PMAP_GETPORT, &map, &map.pm_port, 0); - - if (status >= 0) { - if (map.pm_port != 0) - return map.pm_port; - status = -EACCES; - } - return status; -} -#endif - -static void -pmap_getport_done(struct rpc_task *task) -{ - struct rpc_clnt *clnt = task->tk_client; - struct rpc_portmap *map = clnt->cl_pmap; - - dprintk("RPC: %4d pmap_getport_done(status %d, port %d)\n", - task->tk_pid, task->tk_status, clnt->cl_port); - if (task->tk_status < 0) { - /* Make the calling task exit with an error */ - task->tk_action = NULL; - } else if (clnt->cl_port == 0) { - /* Program not registered */ - task->tk_status = -EACCES; - task->tk_action = NULL; - } else { - /* byte-swap port number first */ - clnt->cl_port = htons(clnt->cl_port); - clnt->cl_xprt->addr.sin_port = clnt->cl_port; - } - spin_lock(&pmap_lock); - map->pm_binding = 0; - rpc_wake_up(&map->pm_bindwait); - spin_unlock(&pmap_lock); -} - -/* - * Set or unset a port registration with the local portmapper. - * port == 0 means unregister, port != 0 means register. - */ -int -rpc_register(u32 prog, u32 vers, int prot, unsigned short port, int *okay) -{ - struct sockaddr_in sin; - struct rpc_portmap map; - struct rpc_clnt *pmap_clnt; - int error = 0; - - dprintk("RPC: registering (%d, %d, %d, %d) with portmapper.\n", - prog, vers, prot, port); - - sin.sin_family = AF_INET; - sin.sin_addr.s_addr = htonl(INADDR_LOOPBACK); - pmap_clnt = pmap_create("localhost", &sin, IPPROTO_UDP, 1); - if (IS_ERR(pmap_clnt)) { - error = PTR_ERR(pmap_clnt); - dprintk("RPC: couldn't create pmap client. Error = %d\n", error); - return error; - } - - map.pm_prog = prog; - map.pm_vers = vers; - map.pm_prot = prot; - map.pm_port = port; - - error = rpc_call(pmap_clnt, port? PMAP_SET : PMAP_UNSET, - &map, okay, 0); - - if (error < 0) { - printk(KERN_WARNING - "RPC: failed to contact portmap (errno %d).\n", - error); - } - dprintk("RPC: registration status %d/%d\n", error, *okay); - - /* Client deleted automatically because cl_oneshot == 1 */ - return error; -} - -static struct rpc_clnt * -pmap_create(char *hostname, struct sockaddr_in *srvaddr, int proto, int privileged) -{ - struct rpc_xprt *xprt; - struct rpc_clnt *clnt; - - /* printk("pmap: create xprt\n"); */ - xprt = xprt_create_proto(proto, srvaddr, NULL); - if (IS_ERR(xprt)) - return (struct rpc_clnt *)xprt; - xprt->addr.sin_port = htons(RPC_PMAP_PORT); - if (!privileged) - xprt->resvport = 0; - - /* printk("pmap: create clnt\n"); */ - clnt = rpc_new_client(xprt, hostname, - &pmap_program, RPC_PMAP_VERSION, - RPC_AUTH_UNIX); - if (!IS_ERR(clnt)) { - clnt->cl_softrtry = 1; - clnt->cl_chatty = 1; - clnt->cl_oneshot = 1; - } - return clnt; -} - -/* - * XDR encode/decode functions for PMAP - */ -static int -xdr_encode_mapping(struct rpc_rqst *req, u32 *p, struct rpc_portmap *map) -{ - dprintk("RPC: xdr_encode_mapping(%d, %d, %d, %d)\n", - map->pm_prog, map->pm_vers, map->pm_prot, map->pm_port); - *p++ = htonl(map->pm_prog); - *p++ = htonl(map->pm_vers); - *p++ = htonl(map->pm_prot); - *p++ = htonl(map->pm_port); - - req->rq_slen = xdr_adjust_iovec(req->rq_svec, p); - return 0; -} - -static int -xdr_decode_port(struct rpc_rqst *req, u32 *p, unsigned short *portp) -{ - *portp = (unsigned short) ntohl(*p++); - return 0; -} - -static int -xdr_decode_bool(struct rpc_rqst *req, u32 *p, unsigned int *boolp) -{ - *boolp = (unsigned int) ntohl(*p++); - return 0; -} - -static struct rpc_procinfo pmap_procedures[] = { -[PMAP_SET] = { - .p_proc = PMAP_SET, - .p_encode = (kxdrproc_t) xdr_encode_mapping, - .p_decode = (kxdrproc_t) xdr_decode_bool, - .p_bufsiz = 4, - .p_count = 1, - }, -[PMAP_UNSET] = { - .p_proc = PMAP_UNSET, - .p_encode = (kxdrproc_t) xdr_encode_mapping, - .p_decode = (kxdrproc_t) xdr_decode_bool, - .p_bufsiz = 4, - .p_count = 1, - }, -[PMAP_GETPORT] = { - .p_proc = PMAP_GETPORT, - .p_encode = (kxdrproc_t) xdr_encode_mapping, - .p_decode = (kxdrproc_t) xdr_decode_port, - .p_bufsiz = 4, - .p_count = 1, - }, -}; - -static struct rpc_version pmap_version2 = { - .number = 2, - .nrprocs = 4, - .procs = pmap_procedures -}; - -static struct rpc_version * pmap_version[] = { - NULL, - NULL, - &pmap_version2 -}; - -static struct rpc_stat pmap_stats; - -static struct rpc_program pmap_program = { - .name = "portmap", - .number = RPC_PMAP_PROGRAM, - .nrvers = ARRAY_SIZE(pmap_version), - .version = pmap_version, - .stats = &pmap_stats, -}; diff --git a/net/sunrpc/rpc_pipe.c b/net/sunrpc/rpc_pipe.c index 4f188d0a5d1..b1855489856 100644 --- a/net/sunrpc/rpc_pipe.c +++ b/net/sunrpc/rpc_pipe.c @@ -8,18 +8,18 @@ * Copyright (c) 2002, Trond Myklebust <trond.myklebust@fys.uio.no> * */ -#include <linux/config.h> #include <linux/module.h> #include <linux/slab.h> #include <linux/string.h> #include <linux/pagemap.h> #include <linux/mount.h> #include <linux/namei.h> -#include <linux/dnotify.h> +#include <linux/fsnotify.h> #include <linux/kernel.h> +#include <linux/rcupdate.h> +#include <linux/utsname.h> #include <asm/ioctls.h> -#include <linux/fs.h> #include <linux/poll.h> #include <linux/wait.h> #include <linux/seq_file.h> @@ -27,77 +27,137 @@ #include <linux/sunrpc/clnt.h> #include <linux/workqueue.h> #include <linux/sunrpc/rpc_pipe_fs.h> +#include <linux/sunrpc/cache.h> +#include <linux/nsproxy.h> +#include <linux/notifier.h> -static struct vfsmount *rpc_mount __read_mostly; -static int rpc_mount_count; +#include "netns.h" +#include "sunrpc.h" -static struct file_system_type rpc_pipe_fs_type; +#define RPCDBG_FACILITY RPCDBG_DEBUG + +#define NET_NAME(net) ((net == &init_net) ? " (init_net)" : "") +static struct file_system_type rpc_pipe_fs_type; +static const struct rpc_pipe_ops gssd_dummy_pipe_ops; -static kmem_cache_t *rpc_inode_cachep __read_mostly; +static struct kmem_cache *rpc_inode_cachep __read_mostly; #define RPC_UPCALL_TIMEOUT (30*HZ) -static void -__rpc_purge_upcall(struct inode *inode, int err) +static BLOCKING_NOTIFIER_HEAD(rpc_pipefs_notifier_list); + +int rpc_pipefs_notifier_register(struct notifier_block *nb) +{ + return blocking_notifier_chain_cond_register(&rpc_pipefs_notifier_list, nb); +} +EXPORT_SYMBOL_GPL(rpc_pipefs_notifier_register); + +void rpc_pipefs_notifier_unregister(struct notifier_block *nb) +{ + blocking_notifier_chain_unregister(&rpc_pipefs_notifier_list, nb); +} +EXPORT_SYMBOL_GPL(rpc_pipefs_notifier_unregister); + +static void rpc_purge_list(wait_queue_head_t *waitq, struct list_head *head, + void (*destroy_msg)(struct rpc_pipe_msg *), int err) { - struct rpc_inode *rpci = RPC_I(inode); struct rpc_pipe_msg *msg; - while (!list_empty(&rpci->pipe)) { - msg = list_entry(rpci->pipe.next, struct rpc_pipe_msg, list); + if (list_empty(head)) + return; + do { + msg = list_entry(head->next, struct rpc_pipe_msg, list); list_del_init(&msg->list); msg->errno = err; - rpci->ops->destroy_msg(msg); - } - while (!list_empty(&rpci->in_upcall)) { - msg = list_entry(rpci->pipe.next, struct rpc_pipe_msg, list); - list_del_init(&msg->list); - msg->errno = err; - rpci->ops->destroy_msg(msg); - } - rpci->pipelen = 0; - wake_up(&rpci->waitq); + destroy_msg(msg); + } while (!list_empty(head)); + + if (waitq) + wake_up(waitq); } static void -rpc_timeout_upcall_queue(void *data) +rpc_timeout_upcall_queue(struct work_struct *work) +{ + LIST_HEAD(free_list); + struct rpc_pipe *pipe = + container_of(work, struct rpc_pipe, queue_timeout.work); + void (*destroy_msg)(struct rpc_pipe_msg *); + struct dentry *dentry; + + spin_lock(&pipe->lock); + destroy_msg = pipe->ops->destroy_msg; + if (pipe->nreaders == 0) { + list_splice_init(&pipe->pipe, &free_list); + pipe->pipelen = 0; + } + dentry = dget(pipe->dentry); + spin_unlock(&pipe->lock); + rpc_purge_list(dentry ? &RPC_I(dentry->d_inode)->waitq : NULL, + &free_list, destroy_msg, -ETIMEDOUT); + dput(dentry); +} + +ssize_t rpc_pipe_generic_upcall(struct file *filp, struct rpc_pipe_msg *msg, + char __user *dst, size_t buflen) { - struct rpc_inode *rpci = (struct rpc_inode *)data; - struct inode *inode = &rpci->vfs_inode; + char *data = (char *)msg->data + msg->copied; + size_t mlen = min(msg->len - msg->copied, buflen); + unsigned long left; + + left = copy_to_user(dst, data, mlen); + if (left == mlen) { + msg->errno = -EFAULT; + return -EFAULT; + } - down(&inode->i_sem); - if (rpci->nreaders == 0 && !list_empty(&rpci->pipe)) - __rpc_purge_upcall(inode, -ETIMEDOUT); - up(&inode->i_sem); + mlen -= left; + msg->copied += mlen; + msg->errno = 0; + return mlen; } +EXPORT_SYMBOL_GPL(rpc_pipe_generic_upcall); +/** + * rpc_queue_upcall - queue an upcall message to userspace + * @pipe: upcall pipe on which to queue given message + * @msg: message to queue + * + * Call with an @inode created by rpc_mkpipe() to queue an upcall. + * A userspace process may then later read the upcall by performing a + * read on an open file for this inode. It is up to the caller to + * initialize the fields of @msg (other than @msg->list) appropriately. + */ int -rpc_queue_upcall(struct inode *inode, struct rpc_pipe_msg *msg) +rpc_queue_upcall(struct rpc_pipe *pipe, struct rpc_pipe_msg *msg) { - struct rpc_inode *rpci = RPC_I(inode); int res = -EPIPE; + struct dentry *dentry; - down(&inode->i_sem); - if (rpci->ops == NULL) - goto out; - if (rpci->nreaders) { - list_add_tail(&msg->list, &rpci->pipe); - rpci->pipelen += msg->len; + spin_lock(&pipe->lock); + if (pipe->nreaders) { + list_add_tail(&msg->list, &pipe->pipe); + pipe->pipelen += msg->len; res = 0; - } else if (rpci->flags & RPC_PIPE_WAIT_FOR_OPEN) { - if (list_empty(&rpci->pipe)) - schedule_delayed_work(&rpci->queue_timeout, + } else if (pipe->flags & RPC_PIPE_WAIT_FOR_OPEN) { + if (list_empty(&pipe->pipe)) + queue_delayed_work(rpciod_workqueue, + &pipe->queue_timeout, RPC_UPCALL_TIMEOUT); - list_add_tail(&msg->list, &rpci->pipe); - rpci->pipelen += msg->len; + list_add_tail(&msg->list, &pipe->pipe); + pipe->pipelen += msg->len; res = 0; } -out: - up(&inode->i_sem); - wake_up(&rpci->waitq); + dentry = dget(pipe->dentry); + spin_unlock(&pipe->lock); + if (dentry) { + wake_up(&RPC_I(dentry->d_inode)->waitq); + dput(dentry); + } return res; } +EXPORT_SYMBOL_GPL(rpc_queue_upcall); static inline void rpc_inode_setowner(struct inode *inode, void *private) @@ -108,186 +168,233 @@ rpc_inode_setowner(struct inode *inode, void *private) static void rpc_close_pipes(struct inode *inode) { - struct rpc_inode *rpci = RPC_I(inode); - - cancel_delayed_work(&rpci->queue_timeout); - flush_scheduled_work(); - down(&inode->i_sem); - if (rpci->ops != NULL) { - rpci->nreaders = 0; - __rpc_purge_upcall(inode, -EPIPE); - rpci->nwriters = 0; - if (rpci->ops->release_pipe) - rpci->ops->release_pipe(inode); - rpci->ops = NULL; - } + struct rpc_pipe *pipe = RPC_I(inode)->pipe; + int need_release; + LIST_HEAD(free_list); + + mutex_lock(&inode->i_mutex); + spin_lock(&pipe->lock); + need_release = pipe->nreaders != 0 || pipe->nwriters != 0; + pipe->nreaders = 0; + list_splice_init(&pipe->in_upcall, &free_list); + list_splice_init(&pipe->pipe, &free_list); + pipe->pipelen = 0; + pipe->dentry = NULL; + spin_unlock(&pipe->lock); + rpc_purge_list(&RPC_I(inode)->waitq, &free_list, pipe->ops->destroy_msg, -EPIPE); + pipe->nwriters = 0; + if (need_release && pipe->ops->release_pipe) + pipe->ops->release_pipe(inode); + cancel_delayed_work_sync(&pipe->queue_timeout); rpc_inode_setowner(inode, NULL); - up(&inode->i_sem); + RPC_I(inode)->pipe = NULL; + mutex_unlock(&inode->i_mutex); } static struct inode * rpc_alloc_inode(struct super_block *sb) { struct rpc_inode *rpci; - rpci = (struct rpc_inode *)kmem_cache_alloc(rpc_inode_cachep, SLAB_KERNEL); + rpci = (struct rpc_inode *)kmem_cache_alloc(rpc_inode_cachep, GFP_KERNEL); if (!rpci) return NULL; return &rpci->vfs_inode; } static void -rpc_destroy_inode(struct inode *inode) +rpc_i_callback(struct rcu_head *head) { + struct inode *inode = container_of(head, struct inode, i_rcu); kmem_cache_free(rpc_inode_cachep, RPC_I(inode)); } +static void +rpc_destroy_inode(struct inode *inode) +{ + call_rcu(&inode->i_rcu, rpc_i_callback); +} + static int rpc_pipe_open(struct inode *inode, struct file *filp) { - struct rpc_inode *rpci = RPC_I(inode); + struct rpc_pipe *pipe; + int first_open; int res = -ENXIO; - down(&inode->i_sem); - if (rpci->ops != NULL) { - if (filp->f_mode & FMODE_READ) - rpci->nreaders ++; - if (filp->f_mode & FMODE_WRITE) - rpci->nwriters ++; - res = 0; + mutex_lock(&inode->i_mutex); + pipe = RPC_I(inode)->pipe; + if (pipe == NULL) + goto out; + first_open = pipe->nreaders == 0 && pipe->nwriters == 0; + if (first_open && pipe->ops->open_pipe) { + res = pipe->ops->open_pipe(inode); + if (res) + goto out; } - up(&inode->i_sem); + if (filp->f_mode & FMODE_READ) + pipe->nreaders++; + if (filp->f_mode & FMODE_WRITE) + pipe->nwriters++; + res = 0; +out: + mutex_unlock(&inode->i_mutex); return res; } static int rpc_pipe_release(struct inode *inode, struct file *filp) { - struct rpc_inode *rpci = RPC_I(filp->f_dentry->d_inode); + struct rpc_pipe *pipe; struct rpc_pipe_msg *msg; + int last_close; - down(&inode->i_sem); - if (rpci->ops == NULL) + mutex_lock(&inode->i_mutex); + pipe = RPC_I(inode)->pipe; + if (pipe == NULL) goto out; - msg = (struct rpc_pipe_msg *)filp->private_data; + msg = filp->private_data; if (msg != NULL) { - msg->errno = -EPIPE; + spin_lock(&pipe->lock); + msg->errno = -EAGAIN; list_del_init(&msg->list); - rpci->ops->destroy_msg(msg); + spin_unlock(&pipe->lock); + pipe->ops->destroy_msg(msg); } if (filp->f_mode & FMODE_WRITE) - rpci->nwriters --; - if (filp->f_mode & FMODE_READ) - rpci->nreaders --; - if (!rpci->nreaders) - __rpc_purge_upcall(inode, -EPIPE); - if (rpci->ops->release_pipe) - rpci->ops->release_pipe(inode); + pipe->nwriters --; + if (filp->f_mode & FMODE_READ) { + pipe->nreaders --; + if (pipe->nreaders == 0) { + LIST_HEAD(free_list); + spin_lock(&pipe->lock); + list_splice_init(&pipe->pipe, &free_list); + pipe->pipelen = 0; + spin_unlock(&pipe->lock); + rpc_purge_list(&RPC_I(inode)->waitq, &free_list, + pipe->ops->destroy_msg, -EAGAIN); + } + } + last_close = pipe->nwriters == 0 && pipe->nreaders == 0; + if (last_close && pipe->ops->release_pipe) + pipe->ops->release_pipe(inode); out: - up(&inode->i_sem); + mutex_unlock(&inode->i_mutex); return 0; } static ssize_t rpc_pipe_read(struct file *filp, char __user *buf, size_t len, loff_t *offset) { - struct inode *inode = filp->f_dentry->d_inode; - struct rpc_inode *rpci = RPC_I(inode); + struct inode *inode = file_inode(filp); + struct rpc_pipe *pipe; struct rpc_pipe_msg *msg; int res = 0; - down(&inode->i_sem); - if (rpci->ops == NULL) { + mutex_lock(&inode->i_mutex); + pipe = RPC_I(inode)->pipe; + if (pipe == NULL) { res = -EPIPE; goto out_unlock; } msg = filp->private_data; if (msg == NULL) { - if (!list_empty(&rpci->pipe)) { - msg = list_entry(rpci->pipe.next, + spin_lock(&pipe->lock); + if (!list_empty(&pipe->pipe)) { + msg = list_entry(pipe->pipe.next, struct rpc_pipe_msg, list); - list_move(&msg->list, &rpci->in_upcall); - rpci->pipelen -= msg->len; + list_move(&msg->list, &pipe->in_upcall); + pipe->pipelen -= msg->len; filp->private_data = msg; msg->copied = 0; } + spin_unlock(&pipe->lock); if (msg == NULL) goto out_unlock; } /* NOTE: it is up to the callback to update msg->copied */ - res = rpci->ops->upcall(filp, msg, buf, len); + res = pipe->ops->upcall(filp, msg, buf, len); if (res < 0 || msg->len == msg->copied) { filp->private_data = NULL; + spin_lock(&pipe->lock); list_del_init(&msg->list); - rpci->ops->destroy_msg(msg); + spin_unlock(&pipe->lock); + pipe->ops->destroy_msg(msg); } out_unlock: - up(&inode->i_sem); + mutex_unlock(&inode->i_mutex); return res; } static ssize_t rpc_pipe_write(struct file *filp, const char __user *buf, size_t len, loff_t *offset) { - struct inode *inode = filp->f_dentry->d_inode; - struct rpc_inode *rpci = RPC_I(inode); + struct inode *inode = file_inode(filp); int res; - down(&inode->i_sem); + mutex_lock(&inode->i_mutex); res = -EPIPE; - if (rpci->ops != NULL) - res = rpci->ops->downcall(filp, buf, len); - up(&inode->i_sem); + if (RPC_I(inode)->pipe != NULL) + res = RPC_I(inode)->pipe->ops->downcall(filp, buf, len); + mutex_unlock(&inode->i_mutex); return res; } static unsigned int rpc_pipe_poll(struct file *filp, struct poll_table_struct *wait) { - struct rpc_inode *rpci; - unsigned int mask = 0; + struct inode *inode = file_inode(filp); + struct rpc_inode *rpci = RPC_I(inode); + unsigned int mask = POLLOUT | POLLWRNORM; - rpci = RPC_I(filp->f_dentry->d_inode); poll_wait(filp, &rpci->waitq, wait); - mask = POLLOUT | POLLWRNORM; - if (rpci->ops == NULL) + mutex_lock(&inode->i_mutex); + if (rpci->pipe == NULL) mask |= POLLERR | POLLHUP; - if (!list_empty(&rpci->pipe)) + else if (filp->private_data || !list_empty(&rpci->pipe->pipe)) mask |= POLLIN | POLLRDNORM; + mutex_unlock(&inode->i_mutex); return mask; } -static int -rpc_pipe_ioctl(struct inode *ino, struct file *filp, - unsigned int cmd, unsigned long arg) +static long +rpc_pipe_ioctl(struct file *filp, unsigned int cmd, unsigned long arg) { - struct rpc_inode *rpci = RPC_I(filp->f_dentry->d_inode); + struct inode *inode = file_inode(filp); + struct rpc_pipe *pipe; int len; switch (cmd) { case FIONREAD: - if (rpci->ops == NULL) + mutex_lock(&inode->i_mutex); + pipe = RPC_I(inode)->pipe; + if (pipe == NULL) { + mutex_unlock(&inode->i_mutex); return -EPIPE; - len = rpci->pipelen; + } + spin_lock(&pipe->lock); + len = pipe->pipelen; if (filp->private_data) { struct rpc_pipe_msg *msg; - msg = (struct rpc_pipe_msg *)filp->private_data; + msg = filp->private_data; len += msg->len - msg->copied; } + spin_unlock(&pipe->lock); + mutex_unlock(&inode->i_mutex); return put_user(len, (int __user *)arg); default: return -EINVAL; } } -static struct file_operations rpc_pipe_fops = { +static const struct file_operations rpc_pipe_fops = { .owner = THIS_MODULE, .llseek = no_llseek, .read = rpc_pipe_read, .write = rpc_pipe_write, .poll = rpc_pipe_poll, - .ioctl = rpc_pipe_ioctl, + .unlocked_ioctl = rpc_pipe_ioctl, .open = rpc_pipe_open, .release = rpc_pipe_release, }; @@ -297,34 +404,38 @@ rpc_show_info(struct seq_file *m, void *v) { struct rpc_clnt *clnt = m->private; - seq_printf(m, "RPC server: %s\n", clnt->cl_server); - seq_printf(m, "service: %s (%d) version %d\n", clnt->cl_protname, + rcu_read_lock(); + seq_printf(m, "RPC server: %s\n", + rcu_dereference(clnt->cl_xprt)->servername); + seq_printf(m, "service: %s (%d) version %d\n", clnt->cl_program->name, clnt->cl_prog, clnt->cl_vers); - seq_printf(m, "address: %u.%u.%u.%u\n", - NIPQUAD(clnt->cl_xprt->addr.sin_addr.s_addr)); - seq_printf(m, "protocol: %s\n", - clnt->cl_xprt->prot == IPPROTO_UDP ? "udp" : "tcp"); + seq_printf(m, "address: %s\n", rpc_peeraddr2str(clnt, RPC_DISPLAY_ADDR)); + seq_printf(m, "protocol: %s\n", rpc_peeraddr2str(clnt, RPC_DISPLAY_PROTO)); + seq_printf(m, "port: %s\n", rpc_peeraddr2str(clnt, RPC_DISPLAY_PORT)); + rcu_read_unlock(); return 0; } static int rpc_info_open(struct inode *inode, struct file *file) { - struct rpc_clnt *clnt; + struct rpc_clnt *clnt = NULL; int ret = single_open(file, rpc_show_info, NULL); if (!ret) { struct seq_file *m = file->private_data; - down(&inode->i_sem); - clnt = RPC_I(inode)->private; - if (clnt) { - atomic_inc(&clnt->cl_users); + + spin_lock(&file->f_path.dentry->d_lock); + if (!d_unhashed(file->f_path.dentry)) + clnt = RPC_I(inode)->private; + if (clnt != NULL && atomic_inc_not_zero(&clnt->cl_count)) { + spin_unlock(&file->f_path.dentry->d_lock); m->private = clnt; } else { + spin_unlock(&file->f_path.dentry->d_lock); single_release(inode, file); ret = -EINVAL; } - up(&inode->i_sem); } return ret; } @@ -340,7 +451,7 @@ rpc_info_release(struct inode *inode, struct file *file) return single_release(inode, file); } -static struct file_operations rpc_info_operations = { +static const struct file_operations rpc_info_operations = { .owner = THIS_MODULE, .open = rpc_info_open, .read = seq_read, @@ -350,404 +461,683 @@ static struct file_operations rpc_info_operations = { /* - * We have a single directory with 1 node in it. - */ -enum { - RPCAUTH_Root = 1, - RPCAUTH_lockd, - RPCAUTH_mount, - RPCAUTH_nfs, - RPCAUTH_portmap, - RPCAUTH_statd, - RPCAUTH_RootEOF -}; - -/* * Description of fs contents. */ struct rpc_filelist { - char *name; - struct file_operations *i_fop; - int mode; + const char *name; + const struct file_operations *i_fop; + umode_t mode; }; -static struct rpc_filelist files[] = { - [RPCAUTH_lockd] = { - .name = "lockd", - .mode = S_IFDIR | S_IRUGO | S_IXUGO, - }, - [RPCAUTH_mount] = { - .name = "mount", - .mode = S_IFDIR | S_IRUGO | S_IXUGO, - }, - [RPCAUTH_nfs] = { - .name = "nfs", - .mode = S_IFDIR | S_IRUGO | S_IXUGO, - }, - [RPCAUTH_portmap] = { - .name = "portmap", - .mode = S_IFDIR | S_IRUGO | S_IXUGO, - }, - [RPCAUTH_statd] = { - .name = "statd", - .mode = S_IFDIR | S_IRUGO | S_IXUGO, - }, -}; +static struct inode * +rpc_get_inode(struct super_block *sb, umode_t mode) +{ + struct inode *inode = new_inode(sb); + if (!inode) + return NULL; + inode->i_ino = get_next_ino(); + inode->i_mode = mode; + inode->i_atime = inode->i_mtime = inode->i_ctime = CURRENT_TIME; + switch (mode & S_IFMT) { + case S_IFDIR: + inode->i_fop = &simple_dir_operations; + inode->i_op = &simple_dir_inode_operations; + inc_nlink(inode); + default: + break; + } + return inode; +} -enum { - RPCAUTH_info = 2, - RPCAUTH_EOF -}; +static int __rpc_create_common(struct inode *dir, struct dentry *dentry, + umode_t mode, + const struct file_operations *i_fop, + void *private) +{ + struct inode *inode; -static struct rpc_filelist authfiles[] = { - [RPCAUTH_info] = { - .name = "info", - .i_fop = &rpc_info_operations, - .mode = S_IFREG | S_IRUSR, - }, -}; + d_drop(dentry); + inode = rpc_get_inode(dir->i_sb, mode); + if (!inode) + goto out_err; + inode->i_ino = iunique(dir->i_sb, 100); + if (i_fop) + inode->i_fop = i_fop; + if (private) + rpc_inode_setowner(inode, private); + d_add(dentry, inode); + return 0; +out_err: + printk(KERN_WARNING "%s: %s failed to allocate inode for dentry %pd\n", + __FILE__, __func__, dentry); + dput(dentry); + return -ENOMEM; +} -static int -rpc_get_mount(void) +static int __rpc_create(struct inode *dir, struct dentry *dentry, + umode_t mode, + const struct file_operations *i_fop, + void *private) +{ + int err; + + err = __rpc_create_common(dir, dentry, S_IFREG | mode, i_fop, private); + if (err) + return err; + fsnotify_create(dir, dentry); + return 0; +} + +static int __rpc_mkdir(struct inode *dir, struct dentry *dentry, + umode_t mode, + const struct file_operations *i_fop, + void *private) { - return simple_pin_fs("rpc_pipefs", &rpc_mount, &rpc_mount_count); + int err; + + err = __rpc_create_common(dir, dentry, S_IFDIR | mode, i_fop, private); + if (err) + return err; + inc_nlink(dir); + fsnotify_mkdir(dir, dentry); + return 0; } static void -rpc_put_mount(void) +init_pipe(struct rpc_pipe *pipe) { - simple_release_fs(&rpc_mount, &rpc_mount_count); + pipe->nreaders = 0; + pipe->nwriters = 0; + INIT_LIST_HEAD(&pipe->in_upcall); + INIT_LIST_HEAD(&pipe->in_downcall); + INIT_LIST_HEAD(&pipe->pipe); + pipe->pipelen = 0; + INIT_DELAYED_WORK(&pipe->queue_timeout, + rpc_timeout_upcall_queue); + pipe->ops = NULL; + spin_lock_init(&pipe->lock); + pipe->dentry = NULL; } -static int -rpc_lookup_parent(char *path, struct nameidata *nd) -{ - if (path[0] == '\0') - return -ENOENT; - if (rpc_get_mount()) { - printk(KERN_WARNING "%s: %s failed to mount " - "pseudofilesystem \n", __FILE__, __FUNCTION__); - return -ENODEV; - } - nd->mnt = mntget(rpc_mount); - nd->dentry = dget(rpc_mount->mnt_root); - nd->last_type = LAST_ROOT; - nd->flags = LOOKUP_PARENT; - nd->depth = 0; - - if (path_walk(path, nd)) { - printk(KERN_WARNING "%s: %s failed to find path %s\n", - __FILE__, __FUNCTION__, path); - rpc_put_mount(); - return -ENOENT; - } +void rpc_destroy_pipe_data(struct rpc_pipe *pipe) +{ + kfree(pipe); +} +EXPORT_SYMBOL_GPL(rpc_destroy_pipe_data); + +struct rpc_pipe *rpc_mkpipe_data(const struct rpc_pipe_ops *ops, int flags) +{ + struct rpc_pipe *pipe; + + pipe = kzalloc(sizeof(struct rpc_pipe), GFP_KERNEL); + if (!pipe) + return ERR_PTR(-ENOMEM); + init_pipe(pipe); + pipe->ops = ops; + pipe->flags = flags; + return pipe; +} +EXPORT_SYMBOL_GPL(rpc_mkpipe_data); + +static int __rpc_mkpipe_dentry(struct inode *dir, struct dentry *dentry, + umode_t mode, + const struct file_operations *i_fop, + void *private, + struct rpc_pipe *pipe) +{ + struct rpc_inode *rpci; + int err; + + err = __rpc_create_common(dir, dentry, S_IFIFO | mode, i_fop, private); + if (err) + return err; + rpci = RPC_I(dentry->d_inode); + rpci->private = private; + rpci->pipe = pipe; + fsnotify_create(dir, dentry); return 0; } -static void -rpc_release_path(struct nameidata *nd) +static int __rpc_rmdir(struct inode *dir, struct dentry *dentry) { - path_release(nd); - rpc_put_mount(); + int ret; + + dget(dentry); + ret = simple_rmdir(dir, dentry); + d_delete(dentry); + dput(dentry); + return ret; } -static struct inode * -rpc_get_inode(struct super_block *sb, int mode) +int rpc_rmdir(struct dentry *dentry) { - struct inode *inode = new_inode(sb); - if (!inode) - return NULL; - inode->i_mode = mode; - inode->i_uid = inode->i_gid = 0; - inode->i_blksize = PAGE_CACHE_SIZE; - inode->i_blocks = 0; - inode->i_atime = inode->i_mtime = inode->i_ctime = CURRENT_TIME; - switch(mode & S_IFMT) { - case S_IFDIR: - inode->i_fop = &simple_dir_operations; - inode->i_op = &simple_dir_inode_operations; - inode->i_nlink++; - default: - break; + struct dentry *parent; + struct inode *dir; + int error; + + parent = dget_parent(dentry); + dir = parent->d_inode; + mutex_lock_nested(&dir->i_mutex, I_MUTEX_PARENT); + error = __rpc_rmdir(dir, dentry); + mutex_unlock(&dir->i_mutex); + dput(parent); + return error; +} +EXPORT_SYMBOL_GPL(rpc_rmdir); + +static int __rpc_unlink(struct inode *dir, struct dentry *dentry) +{ + int ret; + + dget(dentry); + ret = simple_unlink(dir, dentry); + d_delete(dentry); + dput(dentry); + return ret; +} + +static int __rpc_rmpipe(struct inode *dir, struct dentry *dentry) +{ + struct inode *inode = dentry->d_inode; + + rpc_close_pipes(inode); + return __rpc_unlink(dir, dentry); +} + +static struct dentry *__rpc_lookup_create_exclusive(struct dentry *parent, + const char *name) +{ + struct qstr q = QSTR_INIT(name, strlen(name)); + struct dentry *dentry = d_hash_and_lookup(parent, &q); + if (!dentry) { + dentry = d_alloc(parent, &q); + if (!dentry) + return ERR_PTR(-ENOMEM); } - return inode; + if (dentry->d_inode == NULL) + return dentry; + dput(dentry); + return ERR_PTR(-EEXIST); } /* * FIXME: This probably has races. */ -static void -rpc_depopulate(struct dentry *parent) +static void __rpc_depopulate(struct dentry *parent, + const struct rpc_filelist *files, + int start, int eof) { struct inode *dir = parent->d_inode; - struct list_head *pos, *next; - struct dentry *dentry, *dvec[10]; - int n = 0; - - down(&dir->i_sem); -repeat: - spin_lock(&dcache_lock); - list_for_each_safe(pos, next, &parent->d_subdirs) { - dentry = list_entry(pos, struct dentry, d_child); - spin_lock(&dentry->d_lock); - if (!d_unhashed(dentry)) { - dget_locked(dentry); - __d_drop(dentry); - spin_unlock(&dentry->d_lock); - dvec[n++] = dentry; - if (n == ARRAY_SIZE(dvec)) - break; - } else - spin_unlock(&dentry->d_lock); - } - spin_unlock(&dcache_lock); - if (n) { - do { - dentry = dvec[--n]; - if (dentry->d_inode) { - rpc_close_pipes(dentry->d_inode); - simple_unlink(dir, dentry); - } - dput(dentry); - } while (n); - goto repeat; - } - up(&dir->i_sem); -} - -static int -rpc_populate(struct dentry *parent, - struct rpc_filelist *files, - int start, int eof) -{ - struct inode *inode, *dir = parent->d_inode; - void *private = RPC_I(dir)->private; struct dentry *dentry; - int mode, i; + struct qstr name; + int i; - down(&dir->i_sem); for (i = start; i < eof; i++) { - dentry = d_alloc_name(parent, files[i].name); - if (!dentry) - goto out_bad; - mode = files[i].mode; - inode = rpc_get_inode(dir->i_sb, mode); - if (!inode) { - dput(dentry); - goto out_bad; + name.name = files[i].name; + name.len = strlen(files[i].name); + dentry = d_hash_and_lookup(parent, &name); + + if (dentry == NULL) + continue; + if (dentry->d_inode == NULL) + goto next; + switch (dentry->d_inode->i_mode & S_IFMT) { + default: + BUG(); + case S_IFREG: + __rpc_unlink(dir, dentry); + break; + case S_IFDIR: + __rpc_rmdir(dir, dentry); } - inode->i_ino = i; - if (files[i].i_fop) - inode->i_fop = files[i].i_fop; - if (private) - rpc_inode_setowner(inode, private); - if (S_ISDIR(mode)) - dir->i_nlink++; - d_add(dentry, inode); +next: + dput(dentry); } - up(&dir->i_sem); - return 0; -out_bad: - up(&dir->i_sem); - printk(KERN_WARNING "%s: %s failed to populate directory %s\n", - __FILE__, __FUNCTION__, parent->d_name.name); - return -ENOMEM; } -static int -__rpc_mkdir(struct inode *dir, struct dentry *dentry) +static void rpc_depopulate(struct dentry *parent, + const struct rpc_filelist *files, + int start, int eof) { - struct inode *inode; + struct inode *dir = parent->d_inode; - inode = rpc_get_inode(dir->i_sb, S_IFDIR | S_IRUSR | S_IXUSR); - if (!inode) - goto out_err; - inode->i_ino = iunique(dir->i_sb, 100); - d_instantiate(dentry, inode); - dir->i_nlink++; - inode_dir_notify(dir, DN_CREATE); - rpc_get_mount(); - return 0; -out_err: - printk(KERN_WARNING "%s: %s failed to allocate inode for dentry %s\n", - __FILE__, __FUNCTION__, dentry->d_name.name); - return -ENOMEM; + mutex_lock_nested(&dir->i_mutex, I_MUTEX_CHILD); + __rpc_depopulate(parent, files, start, eof); + mutex_unlock(&dir->i_mutex); } -static int -__rpc_rmdir(struct inode *dir, struct dentry *dentry) +static int rpc_populate(struct dentry *parent, + const struct rpc_filelist *files, + int start, int eof, + void *private) { - int error; + struct inode *dir = parent->d_inode; + struct dentry *dentry; + int i, err; - shrink_dcache_parent(dentry); - if (dentry->d_inode) - rpc_close_pipes(dentry->d_inode); - if ((error = simple_rmdir(dir, dentry)) != 0) - return error; - if (!error) { - inode_dir_notify(dir, DN_DELETE); - d_drop(dentry); - rpc_put_mount(); + mutex_lock(&dir->i_mutex); + for (i = start; i < eof; i++) { + dentry = __rpc_lookup_create_exclusive(parent, files[i].name); + err = PTR_ERR(dentry); + if (IS_ERR(dentry)) + goto out_bad; + switch (files[i].mode & S_IFMT) { + default: + BUG(); + case S_IFREG: + err = __rpc_create(dir, dentry, + files[i].mode, + files[i].i_fop, + private); + break; + case S_IFDIR: + err = __rpc_mkdir(dir, dentry, + files[i].mode, + NULL, + private); + } + if (err != 0) + goto out_bad; } + mutex_unlock(&dir->i_mutex); return 0; +out_bad: + __rpc_depopulate(parent, files, start, eof); + mutex_unlock(&dir->i_mutex); + printk(KERN_WARNING "%s: %s failed to populate directory %pd\n", + __FILE__, __func__, parent); + return err; } -static struct dentry * -rpc_lookup_negative(char *path, struct nameidata *nd) +static struct dentry *rpc_mkdir_populate(struct dentry *parent, + const char *name, umode_t mode, void *private, + int (*populate)(struct dentry *, void *), void *args_populate) { struct dentry *dentry; - struct inode *dir; + struct inode *dir = parent->d_inode; int error; - if ((error = rpc_lookup_parent(path, nd)) != 0) - return ERR_PTR(error); - dir = nd->dentry->d_inode; - down(&dir->i_sem); - dentry = lookup_hash(&nd->last, nd->dentry); + mutex_lock_nested(&dir->i_mutex, I_MUTEX_PARENT); + dentry = __rpc_lookup_create_exclusive(parent, name); if (IS_ERR(dentry)) + goto out; + error = __rpc_mkdir(dir, dentry, mode, NULL, private); + if (error != 0) goto out_err; - if (dentry->d_inode) { - dput(dentry); - dentry = ERR_PTR(-EEXIST); - goto out_err; + if (populate != NULL) { + error = populate(dentry, args_populate); + if (error) + goto err_rmdir; } - return dentry; -out_err: - up(&dir->i_sem); - rpc_release_path(nd); - return dentry; -} - - -struct dentry * -rpc_mkdir(char *path, struct rpc_clnt *rpc_client) -{ - struct nameidata nd; - struct dentry *dentry; - struct inode *dir; - int error; - - dentry = rpc_lookup_negative(path, &nd); - if (IS_ERR(dentry)) - return dentry; - dir = nd.dentry->d_inode; - if ((error = __rpc_mkdir(dir, dentry)) != 0) - goto err_dput; - RPC_I(dentry->d_inode)->private = rpc_client; - error = rpc_populate(dentry, authfiles, - RPCAUTH_info, RPCAUTH_EOF); - if (error) - goto err_depopulate; out: - up(&dir->i_sem); - rpc_release_path(&nd); + mutex_unlock(&dir->i_mutex); return dentry; -err_depopulate: - rpc_depopulate(dentry); +err_rmdir: __rpc_rmdir(dir, dentry); -err_dput: - dput(dentry); - printk(KERN_WARNING "%s: %s() failed to create directory %s (errno = %d)\n", - __FILE__, __FUNCTION__, path, error); +out_err: dentry = ERR_PTR(error); goto out; } -int -rpc_rmdir(char *path) +static int rpc_rmdir_depopulate(struct dentry *dentry, + void (*depopulate)(struct dentry *)) { - struct nameidata nd; - struct dentry *dentry; + struct dentry *parent; struct inode *dir; int error; - if ((error = rpc_lookup_parent(path, &nd)) != 0) - return error; - dir = nd.dentry->d_inode; - down(&dir->i_sem); - dentry = lookup_hash(&nd.last, nd.dentry); - if (IS_ERR(dentry)) { - error = PTR_ERR(dentry); - goto out_release; - } - rpc_depopulate(dentry); + parent = dget_parent(dentry); + dir = parent->d_inode; + mutex_lock_nested(&dir->i_mutex, I_MUTEX_PARENT); + if (depopulate != NULL) + depopulate(dentry); error = __rpc_rmdir(dir, dentry); - dput(dentry); -out_release: - up(&dir->i_sem); - rpc_release_path(&nd); + mutex_unlock(&dir->i_mutex); + dput(parent); return error; } -struct dentry * -rpc_mkpipe(char *path, void *private, struct rpc_pipe_ops *ops, int flags) +/** + * rpc_mkpipe - make an rpc_pipefs file for kernel<->userspace communication + * @parent: dentry of directory to create new "pipe" in + * @name: name of pipe + * @private: private data to associate with the pipe, for the caller's use + * @pipe: &rpc_pipe containing input parameters + * + * Data is made available for userspace to read by calls to + * rpc_queue_upcall(). The actual reads will result in calls to + * @ops->upcall, which will be called with the file pointer, + * message, and userspace buffer to copy to. + * + * Writes can come at any time, and do not necessarily have to be + * responses to upcalls. They will result in calls to @msg->downcall. + * + * The @private argument passed here will be available to all these methods + * from the file pointer, via RPC_I(file_inode(file))->private. + */ +struct dentry *rpc_mkpipe_dentry(struct dentry *parent, const char *name, + void *private, struct rpc_pipe *pipe) { - struct nameidata nd; struct dentry *dentry; - struct inode *dir, *inode; - struct rpc_inode *rpci; + struct inode *dir = parent->d_inode; + umode_t umode = S_IFIFO | S_IRUSR | S_IWUSR; + int err; + + if (pipe->ops->upcall == NULL) + umode &= ~S_IRUGO; + if (pipe->ops->downcall == NULL) + umode &= ~S_IWUGO; - dentry = rpc_lookup_negative(path, &nd); + mutex_lock_nested(&dir->i_mutex, I_MUTEX_PARENT); + dentry = __rpc_lookup_create_exclusive(parent, name); if (IS_ERR(dentry)) - return dentry; - dir = nd.dentry->d_inode; - inode = rpc_get_inode(dir->i_sb, S_IFSOCK | S_IRUSR | S_IWUSR); - if (!inode) - goto err_dput; - inode->i_ino = iunique(dir->i_sb, 100); - inode->i_fop = &rpc_pipe_fops; - d_instantiate(dentry, inode); - rpci = RPC_I(inode); - rpci->private = private; - rpci->flags = flags; - rpci->ops = ops; - inode_dir_notify(dir, DN_CREATE); + goto out; + err = __rpc_mkpipe_dentry(dir, dentry, umode, &rpc_pipe_fops, + private, pipe); + if (err) + goto out_err; out: - up(&dir->i_sem); - rpc_release_path(&nd); + mutex_unlock(&dir->i_mutex); return dentry; -err_dput: - dput(dentry); - dentry = ERR_PTR(-ENOMEM); - printk(KERN_WARNING "%s: %s() failed to create pipe %s (errno = %d)\n", - __FILE__, __FUNCTION__, path, -ENOMEM); +out_err: + dentry = ERR_PTR(err); + printk(KERN_WARNING "%s: %s() failed to create pipe %pd/%s (errno = %d)\n", + __FILE__, __func__, parent, name, + err); goto out; } +EXPORT_SYMBOL_GPL(rpc_mkpipe_dentry); +/** + * rpc_unlink - remove a pipe + * @dentry: dentry for the pipe, as returned from rpc_mkpipe + * + * After this call, lookups will no longer find the pipe, and any + * attempts to read or write using preexisting opens of the pipe will + * return -EPIPE. + */ int -rpc_unlink(char *path) +rpc_unlink(struct dentry *dentry) { - struct nameidata nd; - struct dentry *dentry; + struct dentry *parent; struct inode *dir; - int error; + int error = 0; + + parent = dget_parent(dentry); + dir = parent->d_inode; + mutex_lock_nested(&dir->i_mutex, I_MUTEX_PARENT); + error = __rpc_rmpipe(dir, dentry); + mutex_unlock(&dir->i_mutex); + dput(parent); + return error; +} +EXPORT_SYMBOL_GPL(rpc_unlink); + +/** + * rpc_init_pipe_dir_head - initialise a struct rpc_pipe_dir_head + * @pdh: pointer to struct rpc_pipe_dir_head + */ +void rpc_init_pipe_dir_head(struct rpc_pipe_dir_head *pdh) +{ + INIT_LIST_HEAD(&pdh->pdh_entries); + pdh->pdh_dentry = NULL; +} +EXPORT_SYMBOL_GPL(rpc_init_pipe_dir_head); + +/** + * rpc_init_pipe_dir_object - initialise a struct rpc_pipe_dir_object + * @pdo: pointer to struct rpc_pipe_dir_object + * @pdo_ops: pointer to const struct rpc_pipe_dir_object_ops + * @pdo_data: pointer to caller-defined data + */ +void rpc_init_pipe_dir_object(struct rpc_pipe_dir_object *pdo, + const struct rpc_pipe_dir_object_ops *pdo_ops, + void *pdo_data) +{ + INIT_LIST_HEAD(&pdo->pdo_head); + pdo->pdo_ops = pdo_ops; + pdo->pdo_data = pdo_data; +} +EXPORT_SYMBOL_GPL(rpc_init_pipe_dir_object); + +static int +rpc_add_pipe_dir_object_locked(struct net *net, + struct rpc_pipe_dir_head *pdh, + struct rpc_pipe_dir_object *pdo) +{ + int ret = 0; + + if (pdh->pdh_dentry) + ret = pdo->pdo_ops->create(pdh->pdh_dentry, pdo); + if (ret == 0) + list_add_tail(&pdo->pdo_head, &pdh->pdh_entries); + return ret; +} + +static void +rpc_remove_pipe_dir_object_locked(struct net *net, + struct rpc_pipe_dir_head *pdh, + struct rpc_pipe_dir_object *pdo) +{ + if (pdh->pdh_dentry) + pdo->pdo_ops->destroy(pdh->pdh_dentry, pdo); + list_del_init(&pdo->pdo_head); +} + +/** + * rpc_add_pipe_dir_object - associate a rpc_pipe_dir_object to a directory + * @net: pointer to struct net + * @pdh: pointer to struct rpc_pipe_dir_head + * @pdo: pointer to struct rpc_pipe_dir_object + * + */ +int +rpc_add_pipe_dir_object(struct net *net, + struct rpc_pipe_dir_head *pdh, + struct rpc_pipe_dir_object *pdo) +{ + int ret = 0; - if ((error = rpc_lookup_parent(path, &nd)) != 0) - return error; - dir = nd.dentry->d_inode; - down(&dir->i_sem); - dentry = lookup_hash(&nd.last, nd.dentry); - if (IS_ERR(dentry)) { - error = PTR_ERR(dentry); - goto out_release; + if (list_empty(&pdo->pdo_head)) { + struct sunrpc_net *sn = net_generic(net, sunrpc_net_id); + + mutex_lock(&sn->pipefs_sb_lock); + ret = rpc_add_pipe_dir_object_locked(net, pdh, pdo); + mutex_unlock(&sn->pipefs_sb_lock); } - d_drop(dentry); - if (dentry->d_inode) { - rpc_close_pipes(dentry->d_inode); - error = simple_unlink(dir, dentry); + return ret; +} +EXPORT_SYMBOL_GPL(rpc_add_pipe_dir_object); + +/** + * rpc_remove_pipe_dir_object - remove a rpc_pipe_dir_object from a directory + * @net: pointer to struct net + * @pdh: pointer to struct rpc_pipe_dir_head + * @pdo: pointer to struct rpc_pipe_dir_object + * + */ +void +rpc_remove_pipe_dir_object(struct net *net, + struct rpc_pipe_dir_head *pdh, + struct rpc_pipe_dir_object *pdo) +{ + if (!list_empty(&pdo->pdo_head)) { + struct sunrpc_net *sn = net_generic(net, sunrpc_net_id); + + mutex_lock(&sn->pipefs_sb_lock); + rpc_remove_pipe_dir_object_locked(net, pdh, pdo); + mutex_unlock(&sn->pipefs_sb_lock); } - dput(dentry); - inode_dir_notify(dir, DN_DELETE); -out_release: - up(&dir->i_sem); - rpc_release_path(&nd); - return error; +} +EXPORT_SYMBOL_GPL(rpc_remove_pipe_dir_object); + +/** + * rpc_find_or_alloc_pipe_dir_object + * @net: pointer to struct net + * @pdh: pointer to struct rpc_pipe_dir_head + * @match: match struct rpc_pipe_dir_object to data + * @alloc: allocate a new struct rpc_pipe_dir_object + * @data: user defined data for match() and alloc() + * + */ +struct rpc_pipe_dir_object * +rpc_find_or_alloc_pipe_dir_object(struct net *net, + struct rpc_pipe_dir_head *pdh, + int (*match)(struct rpc_pipe_dir_object *, void *), + struct rpc_pipe_dir_object *(*alloc)(void *), + void *data) +{ + struct sunrpc_net *sn = net_generic(net, sunrpc_net_id); + struct rpc_pipe_dir_object *pdo; + + mutex_lock(&sn->pipefs_sb_lock); + list_for_each_entry(pdo, &pdh->pdh_entries, pdo_head) { + if (!match(pdo, data)) + continue; + goto out; + } + pdo = alloc(data); + if (!pdo) + goto out; + rpc_add_pipe_dir_object_locked(net, pdh, pdo); +out: + mutex_unlock(&sn->pipefs_sb_lock); + return pdo; +} +EXPORT_SYMBOL_GPL(rpc_find_or_alloc_pipe_dir_object); + +static void +rpc_create_pipe_dir_objects(struct rpc_pipe_dir_head *pdh) +{ + struct rpc_pipe_dir_object *pdo; + struct dentry *dir = pdh->pdh_dentry; + + list_for_each_entry(pdo, &pdh->pdh_entries, pdo_head) + pdo->pdo_ops->create(dir, pdo); +} + +static void +rpc_destroy_pipe_dir_objects(struct rpc_pipe_dir_head *pdh) +{ + struct rpc_pipe_dir_object *pdo; + struct dentry *dir = pdh->pdh_dentry; + + list_for_each_entry(pdo, &pdh->pdh_entries, pdo_head) + pdo->pdo_ops->destroy(dir, pdo); +} + +enum { + RPCAUTH_info, + RPCAUTH_EOF +}; + +static const struct rpc_filelist authfiles[] = { + [RPCAUTH_info] = { + .name = "info", + .i_fop = &rpc_info_operations, + .mode = S_IFREG | S_IRUSR, + }, +}; + +static int rpc_clntdir_populate(struct dentry *dentry, void *private) +{ + return rpc_populate(dentry, + authfiles, RPCAUTH_info, RPCAUTH_EOF, + private); +} + +static void rpc_clntdir_depopulate(struct dentry *dentry) +{ + rpc_depopulate(dentry, authfiles, RPCAUTH_info, RPCAUTH_EOF); +} + +/** + * rpc_create_client_dir - Create a new rpc_client directory in rpc_pipefs + * @dentry: the parent of new directory + * @name: the name of new directory + * @rpc_client: rpc client to associate with this directory + * + * This creates a directory at the given @path associated with + * @rpc_clnt, which will contain a file named "info" with some basic + * information about the client, together with any "pipes" that may + * later be created using rpc_mkpipe(). + */ +struct dentry *rpc_create_client_dir(struct dentry *dentry, + const char *name, + struct rpc_clnt *rpc_client) +{ + struct dentry *ret; + + ret = rpc_mkdir_populate(dentry, name, S_IRUGO | S_IXUGO, NULL, + rpc_clntdir_populate, rpc_client); + if (!IS_ERR(ret)) { + rpc_client->cl_pipedir_objects.pdh_dentry = ret; + rpc_create_pipe_dir_objects(&rpc_client->cl_pipedir_objects); + } + return ret; +} + +/** + * rpc_remove_client_dir - Remove a directory created with rpc_create_client_dir() + * @rpc_client: rpc_client for the pipe + */ +int rpc_remove_client_dir(struct rpc_clnt *rpc_client) +{ + struct dentry *dentry = rpc_client->cl_pipedir_objects.pdh_dentry; + + if (dentry == NULL) + return 0; + rpc_destroy_pipe_dir_objects(&rpc_client->cl_pipedir_objects); + rpc_client->cl_pipedir_objects.pdh_dentry = NULL; + return rpc_rmdir_depopulate(dentry, rpc_clntdir_depopulate); +} + +static const struct rpc_filelist cache_pipefs_files[3] = { + [0] = { + .name = "channel", + .i_fop = &cache_file_operations_pipefs, + .mode = S_IFREG|S_IRUSR|S_IWUSR, + }, + [1] = { + .name = "content", + .i_fop = &content_file_operations_pipefs, + .mode = S_IFREG|S_IRUSR, + }, + [2] = { + .name = "flush", + .i_fop = &cache_flush_operations_pipefs, + .mode = S_IFREG|S_IRUSR|S_IWUSR, + }, +}; + +static int rpc_cachedir_populate(struct dentry *dentry, void *private) +{ + return rpc_populate(dentry, + cache_pipefs_files, 0, 3, + private); +} + +static void rpc_cachedir_depopulate(struct dentry *dentry) +{ + rpc_depopulate(dentry, cache_pipefs_files, 0, 3); +} + +struct dentry *rpc_create_cache_dir(struct dentry *parent, const char *name, + umode_t umode, struct cache_detail *cd) +{ + return rpc_mkdir_populate(parent, name, umode, NULL, + rpc_cachedir_populate, cd); +} + +void rpc_remove_cache_dir(struct dentry *dentry) +{ + rpc_rmdir_depopulate(dentry, rpc_cachedir_depopulate); } /* * populate the filesystem */ -static struct super_operations s_ops = { +static const struct super_operations s_ops = { .alloc_inode = rpc_alloc_inode, .destroy_inode = rpc_destroy_inode, .statfs = simple_statfs, @@ -755,85 +1145,383 @@ static struct super_operations s_ops = { #define RPCAUTH_GSSMAGIC 0x67596969 +/* + * We have a single directory with 1 node in it. + */ +enum { + RPCAUTH_lockd, + RPCAUTH_mount, + RPCAUTH_nfs, + RPCAUTH_portmap, + RPCAUTH_statd, + RPCAUTH_nfsd4_cb, + RPCAUTH_cache, + RPCAUTH_nfsd, + RPCAUTH_gssd, + RPCAUTH_RootEOF +}; + +static const struct rpc_filelist files[] = { + [RPCAUTH_lockd] = { + .name = "lockd", + .mode = S_IFDIR | S_IRUGO | S_IXUGO, + }, + [RPCAUTH_mount] = { + .name = "mount", + .mode = S_IFDIR | S_IRUGO | S_IXUGO, + }, + [RPCAUTH_nfs] = { + .name = "nfs", + .mode = S_IFDIR | S_IRUGO | S_IXUGO, + }, + [RPCAUTH_portmap] = { + .name = "portmap", + .mode = S_IFDIR | S_IRUGO | S_IXUGO, + }, + [RPCAUTH_statd] = { + .name = "statd", + .mode = S_IFDIR | S_IRUGO | S_IXUGO, + }, + [RPCAUTH_nfsd4_cb] = { + .name = "nfsd4_cb", + .mode = S_IFDIR | S_IRUGO | S_IXUGO, + }, + [RPCAUTH_cache] = { + .name = "cache", + .mode = S_IFDIR | S_IRUGO | S_IXUGO, + }, + [RPCAUTH_nfsd] = { + .name = "nfsd", + .mode = S_IFDIR | S_IRUGO | S_IXUGO, + }, + [RPCAUTH_gssd] = { + .name = "gssd", + .mode = S_IFDIR | S_IRUGO | S_IXUGO, + }, +}; + +/* + * This call can be used only in RPC pipefs mount notification hooks. + */ +struct dentry *rpc_d_lookup_sb(const struct super_block *sb, + const unsigned char *dir_name) +{ + struct qstr dir = QSTR_INIT(dir_name, strlen(dir_name)); + return d_hash_and_lookup(sb->s_root, &dir); +} +EXPORT_SYMBOL_GPL(rpc_d_lookup_sb); + +int rpc_pipefs_init_net(struct net *net) +{ + struct sunrpc_net *sn = net_generic(net, sunrpc_net_id); + + sn->gssd_dummy = rpc_mkpipe_data(&gssd_dummy_pipe_ops, 0); + if (IS_ERR(sn->gssd_dummy)) + return PTR_ERR(sn->gssd_dummy); + + mutex_init(&sn->pipefs_sb_lock); + sn->pipe_version = -1; + return 0; +} + +void rpc_pipefs_exit_net(struct net *net) +{ + struct sunrpc_net *sn = net_generic(net, sunrpc_net_id); + + rpc_destroy_pipe_data(sn->gssd_dummy); +} + +/* + * This call will be used for per network namespace operations calls. + * Note: Function will be returned with pipefs_sb_lock taken if superblock was + * found. This lock have to be released by rpc_put_sb_net() when all operations + * will be completed. + */ +struct super_block *rpc_get_sb_net(const struct net *net) +{ + struct sunrpc_net *sn = net_generic(net, sunrpc_net_id); + + mutex_lock(&sn->pipefs_sb_lock); + if (sn->pipefs_sb) + return sn->pipefs_sb; + mutex_unlock(&sn->pipefs_sb_lock); + return NULL; +} +EXPORT_SYMBOL_GPL(rpc_get_sb_net); + +void rpc_put_sb_net(const struct net *net) +{ + struct sunrpc_net *sn = net_generic(net, sunrpc_net_id); + + WARN_ON(sn->pipefs_sb == NULL); + mutex_unlock(&sn->pipefs_sb_lock); +} +EXPORT_SYMBOL_GPL(rpc_put_sb_net); + +static const struct rpc_filelist gssd_dummy_clnt_dir[] = { + [0] = { + .name = "clntXX", + .mode = S_IFDIR | S_IRUGO | S_IXUGO, + }, +}; + +static ssize_t +dummy_downcall(struct file *filp, const char __user *src, size_t len) +{ + return -EINVAL; +} + +static const struct rpc_pipe_ops gssd_dummy_pipe_ops = { + .upcall = rpc_pipe_generic_upcall, + .downcall = dummy_downcall, +}; + +/* + * Here we present a bogus "info" file to keep rpc.gssd happy. We don't expect + * that it will ever use this info to handle an upcall, but rpc.gssd expects + * that this file will be there and have a certain format. + */ +static int +rpc_show_dummy_info(struct seq_file *m, void *v) +{ + seq_printf(m, "RPC server: %s\n", utsname()->nodename); + seq_printf(m, "service: foo (1) version 0\n"); + seq_printf(m, "address: 127.0.0.1\n"); + seq_printf(m, "protocol: tcp\n"); + seq_printf(m, "port: 0\n"); + return 0; +} + +static int +rpc_dummy_info_open(struct inode *inode, struct file *file) +{ + return single_open(file, rpc_show_dummy_info, NULL); +} + +static const struct file_operations rpc_dummy_info_operations = { + .owner = THIS_MODULE, + .open = rpc_dummy_info_open, + .read = seq_read, + .llseek = seq_lseek, + .release = single_release, +}; + +static const struct rpc_filelist gssd_dummy_info_file[] = { + [0] = { + .name = "info", + .i_fop = &rpc_dummy_info_operations, + .mode = S_IFREG | S_IRUSR, + }, +}; + +/** + * rpc_gssd_dummy_populate - create a dummy gssd pipe + * @root: root of the rpc_pipefs filesystem + * @pipe_data: pipe data created when netns is initialized + * + * Create a dummy set of directories and a pipe that gssd can hold open to + * indicate that it is up and running. + */ +static struct dentry * +rpc_gssd_dummy_populate(struct dentry *root, struct rpc_pipe *pipe_data) +{ + int ret = 0; + struct dentry *gssd_dentry; + struct dentry *clnt_dentry = NULL; + struct dentry *pipe_dentry = NULL; + struct qstr q = QSTR_INIT(files[RPCAUTH_gssd].name, + strlen(files[RPCAUTH_gssd].name)); + + /* We should never get this far if "gssd" doesn't exist */ + gssd_dentry = d_hash_and_lookup(root, &q); + if (!gssd_dentry) + return ERR_PTR(-ENOENT); + + ret = rpc_populate(gssd_dentry, gssd_dummy_clnt_dir, 0, 1, NULL); + if (ret) { + pipe_dentry = ERR_PTR(ret); + goto out; + } + + q.name = gssd_dummy_clnt_dir[0].name; + q.len = strlen(gssd_dummy_clnt_dir[0].name); + clnt_dentry = d_hash_and_lookup(gssd_dentry, &q); + if (!clnt_dentry) { + pipe_dentry = ERR_PTR(-ENOENT); + goto out; + } + + ret = rpc_populate(clnt_dentry, gssd_dummy_info_file, 0, 1, NULL); + if (ret) { + __rpc_depopulate(gssd_dentry, gssd_dummy_clnt_dir, 0, 1); + pipe_dentry = ERR_PTR(ret); + goto out; + } + + pipe_dentry = rpc_mkpipe_dentry(clnt_dentry, "gssd", NULL, pipe_data); + if (IS_ERR(pipe_dentry)) { + __rpc_depopulate(clnt_dentry, gssd_dummy_info_file, 0, 1); + __rpc_depopulate(gssd_dentry, gssd_dummy_clnt_dir, 0, 1); + } +out: + dput(clnt_dentry); + dput(gssd_dentry); + return pipe_dentry; +} + +static void +rpc_gssd_dummy_depopulate(struct dentry *pipe_dentry) +{ + struct dentry *clnt_dir = pipe_dentry->d_parent; + struct dentry *gssd_dir = clnt_dir->d_parent; + + __rpc_rmpipe(clnt_dir->d_inode, pipe_dentry); + __rpc_depopulate(clnt_dir, gssd_dummy_info_file, 0, 1); + __rpc_depopulate(gssd_dir, gssd_dummy_clnt_dir, 0, 1); + dput(pipe_dentry); +} + static int rpc_fill_super(struct super_block *sb, void *data, int silent) { struct inode *inode; - struct dentry *root; + struct dentry *root, *gssd_dentry; + struct net *net = data; + struct sunrpc_net *sn = net_generic(net, sunrpc_net_id); + int err; sb->s_blocksize = PAGE_CACHE_SIZE; sb->s_blocksize_bits = PAGE_CACHE_SHIFT; sb->s_magic = RPCAUTH_GSSMAGIC; sb->s_op = &s_ops; + sb->s_d_op = &simple_dentry_operations; sb->s_time_gran = 1; - inode = rpc_get_inode(sb, S_IFDIR | 0755); - if (!inode) + inode = rpc_get_inode(sb, S_IFDIR | S_IRUGO | S_IXUGO); + sb->s_root = root = d_make_root(inode); + if (!root) return -ENOMEM; - root = d_alloc_root(inode); - if (!root) { - iput(inode); + if (rpc_populate(root, files, RPCAUTH_lockd, RPCAUTH_RootEOF, NULL)) return -ENOMEM; + + gssd_dentry = rpc_gssd_dummy_populate(root, sn->gssd_dummy); + if (IS_ERR(gssd_dentry)) { + __rpc_depopulate(root, files, RPCAUTH_lockd, RPCAUTH_RootEOF); + return PTR_ERR(gssd_dentry); } - if (rpc_populate(root, files, RPCAUTH_Root + 1, RPCAUTH_RootEOF)) - goto out; - sb->s_root = root; + + dprintk("RPC: sending pipefs MOUNT notification for net %p%s\n", + net, NET_NAME(net)); + mutex_lock(&sn->pipefs_sb_lock); + sn->pipefs_sb = sb; + err = blocking_notifier_call_chain(&rpc_pipefs_notifier_list, + RPC_PIPEFS_MOUNT, + sb); + if (err) + goto err_depopulate; + sb->s_fs_info = get_net(net); + mutex_unlock(&sn->pipefs_sb_lock); return 0; -out: - d_genocide(root); - dput(root); - return -ENOMEM; + +err_depopulate: + rpc_gssd_dummy_depopulate(gssd_dentry); + blocking_notifier_call_chain(&rpc_pipefs_notifier_list, + RPC_PIPEFS_UMOUNT, + sb); + sn->pipefs_sb = NULL; + __rpc_depopulate(root, files, RPCAUTH_lockd, RPCAUTH_RootEOF); + mutex_unlock(&sn->pipefs_sb_lock); + return err; } -static struct super_block * -rpc_get_sb(struct file_system_type *fs_type, +bool +gssd_running(struct net *net) +{ + struct sunrpc_net *sn = net_generic(net, sunrpc_net_id); + struct rpc_pipe *pipe = sn->gssd_dummy; + + return pipe->nreaders || pipe->nwriters; +} +EXPORT_SYMBOL_GPL(gssd_running); + +static struct dentry * +rpc_mount(struct file_system_type *fs_type, int flags, const char *dev_name, void *data) { - return get_sb_single(fs_type, flags, data, rpc_fill_super); + return mount_ns(fs_type, flags, current->nsproxy->net_ns, rpc_fill_super); +} + +static void rpc_kill_sb(struct super_block *sb) +{ + struct net *net = sb->s_fs_info; + struct sunrpc_net *sn = net_generic(net, sunrpc_net_id); + + mutex_lock(&sn->pipefs_sb_lock); + if (sn->pipefs_sb != sb) { + mutex_unlock(&sn->pipefs_sb_lock); + goto out; + } + sn->pipefs_sb = NULL; + dprintk("RPC: sending pipefs UMOUNT notification for net %p%s\n", + net, NET_NAME(net)); + blocking_notifier_call_chain(&rpc_pipefs_notifier_list, + RPC_PIPEFS_UMOUNT, + sb); + mutex_unlock(&sn->pipefs_sb_lock); + put_net(net); +out: + kill_litter_super(sb); } static struct file_system_type rpc_pipe_fs_type = { .owner = THIS_MODULE, .name = "rpc_pipefs", - .get_sb = rpc_get_sb, - .kill_sb = kill_litter_super, + .mount = rpc_mount, + .kill_sb = rpc_kill_sb, }; +MODULE_ALIAS_FS("rpc_pipefs"); +MODULE_ALIAS("rpc_pipefs"); static void -init_once(void * foo, kmem_cache_t * cachep, unsigned long flags) +init_once(void *foo) { struct rpc_inode *rpci = (struct rpc_inode *) foo; - if ((flags & (SLAB_CTOR_VERIFY|SLAB_CTOR_CONSTRUCTOR)) == - SLAB_CTOR_CONSTRUCTOR) { - inode_init_once(&rpci->vfs_inode); - rpci->private = NULL; - rpci->nreaders = 0; - rpci->nwriters = 0; - INIT_LIST_HEAD(&rpci->in_upcall); - INIT_LIST_HEAD(&rpci->pipe); - rpci->pipelen = 0; - init_waitqueue_head(&rpci->waitq); - INIT_WORK(&rpci->queue_timeout, rpc_timeout_upcall_queue, rpci); - rpci->ops = NULL; - } + inode_init_once(&rpci->vfs_inode); + rpci->private = NULL; + rpci->pipe = NULL; + init_waitqueue_head(&rpci->waitq); } int register_rpc_pipefs(void) { + int err; + rpc_inode_cachep = kmem_cache_create("rpc_inode_cache", - sizeof(struct rpc_inode), - 0, SLAB_HWCACHE_ALIGN|SLAB_RECLAIM_ACCOUNT, - init_once, NULL); + sizeof(struct rpc_inode), + 0, (SLAB_HWCACHE_ALIGN|SLAB_RECLAIM_ACCOUNT| + SLAB_MEM_SPREAD), + init_once); if (!rpc_inode_cachep) return -ENOMEM; - register_filesystem(&rpc_pipe_fs_type); + err = rpc_clients_notifier_register(); + if (err) + goto err_notifier; + err = register_filesystem(&rpc_pipe_fs_type); + if (err) + goto err_register; return 0; + +err_register: + rpc_clients_notifier_unregister(); +err_notifier: + kmem_cache_destroy(rpc_inode_cachep); + return err; } void unregister_rpc_pipefs(void) { - if (kmem_cache_destroy(rpc_inode_cachep)) - printk(KERN_WARNING "RPC: unable to free inode cache\n"); + rpc_clients_notifier_unregister(); + kmem_cache_destroy(rpc_inode_cachep); unregister_filesystem(&rpc_pipe_fs_type); } diff --git a/net/sunrpc/rpcb_clnt.c b/net/sunrpc/rpcb_clnt.c new file mode 100644 index 00000000000..1891a1022c1 --- /dev/null +++ b/net/sunrpc/rpcb_clnt.c @@ -0,0 +1,1147 @@ +/* + * In-kernel rpcbind client supporting versions 2, 3, and 4 of the rpcbind + * protocol + * + * Based on RFC 1833: "Binding Protocols for ONC RPC Version 2" and + * RFC 3530: "Network File System (NFS) version 4 Protocol" + * + * Original: Gilles Quillard, Bull Open Source, 2005 <gilles.quillard@bull.net> + * Updated: Chuck Lever, Oracle Corporation, 2007 <chuck.lever@oracle.com> + * + * Descended from net/sunrpc/pmap_clnt.c, + * Copyright (C) 1996, Olaf Kirch <okir@monad.swb.de> + */ + +#include <linux/module.h> + +#include <linux/types.h> +#include <linux/socket.h> +#include <linux/un.h> +#include <linux/in.h> +#include <linux/in6.h> +#include <linux/kernel.h> +#include <linux/errno.h> +#include <linux/mutex.h> +#include <linux/slab.h> +#include <net/ipv6.h> + +#include <linux/sunrpc/clnt.h> +#include <linux/sunrpc/addr.h> +#include <linux/sunrpc/sched.h> +#include <linux/sunrpc/xprtsock.h> + +#include "netns.h" + +#ifdef RPC_DEBUG +# define RPCDBG_FACILITY RPCDBG_BIND +#endif + +#define RPCBIND_SOCK_PATHNAME "/var/run/rpcbind.sock" + +#define RPCBIND_PROGRAM (100000u) +#define RPCBIND_PORT (111u) + +#define RPCBVERS_2 (2u) +#define RPCBVERS_3 (3u) +#define RPCBVERS_4 (4u) + +enum { + RPCBPROC_NULL, + RPCBPROC_SET, + RPCBPROC_UNSET, + RPCBPROC_GETPORT, + RPCBPROC_GETADDR = 3, /* alias for GETPORT */ + RPCBPROC_DUMP, + RPCBPROC_CALLIT, + RPCBPROC_BCAST = 5, /* alias for CALLIT */ + RPCBPROC_GETTIME, + RPCBPROC_UADDR2TADDR, + RPCBPROC_TADDR2UADDR, + RPCBPROC_GETVERSADDR, + RPCBPROC_INDIRECT, + RPCBPROC_GETADDRLIST, + RPCBPROC_GETSTAT, +}; + +/* + * r_owner + * + * The "owner" is allowed to unset a service in the rpcbind database. + * + * For AF_LOCAL SET/UNSET requests, rpcbind treats this string as a + * UID which it maps to a local user name via a password lookup. + * In all other cases it is ignored. + * + * For SET/UNSET requests, user space provides a value, even for + * network requests, and GETADDR uses an empty string. We follow + * those precedents here. + */ +#define RPCB_OWNER_STRING "0" +#define RPCB_MAXOWNERLEN sizeof(RPCB_OWNER_STRING) + +/* + * XDR data type sizes + */ +#define RPCB_program_sz (1) +#define RPCB_version_sz (1) +#define RPCB_protocol_sz (1) +#define RPCB_port_sz (1) +#define RPCB_boolean_sz (1) + +#define RPCB_netid_sz (1 + XDR_QUADLEN(RPCBIND_MAXNETIDLEN)) +#define RPCB_addr_sz (1 + XDR_QUADLEN(RPCBIND_MAXUADDRLEN)) +#define RPCB_ownerstring_sz (1 + XDR_QUADLEN(RPCB_MAXOWNERLEN)) + +/* + * XDR argument and result sizes + */ +#define RPCB_mappingargs_sz (RPCB_program_sz + RPCB_version_sz + \ + RPCB_protocol_sz + RPCB_port_sz) +#define RPCB_getaddrargs_sz (RPCB_program_sz + RPCB_version_sz + \ + RPCB_netid_sz + RPCB_addr_sz + \ + RPCB_ownerstring_sz) + +#define RPCB_getportres_sz RPCB_port_sz +#define RPCB_setres_sz RPCB_boolean_sz + +/* + * Note that RFC 1833 does not put any size restrictions on the + * address string returned by the remote rpcbind database. + */ +#define RPCB_getaddrres_sz RPCB_addr_sz + +static void rpcb_getport_done(struct rpc_task *, void *); +static void rpcb_map_release(void *data); +static const struct rpc_program rpcb_program; + +struct rpcbind_args { + struct rpc_xprt * r_xprt; + + u32 r_prog; + u32 r_vers; + u32 r_prot; + unsigned short r_port; + const char * r_netid; + const char * r_addr; + const char * r_owner; + + int r_status; +}; + +static struct rpc_procinfo rpcb_procedures2[]; +static struct rpc_procinfo rpcb_procedures3[]; +static struct rpc_procinfo rpcb_procedures4[]; + +struct rpcb_info { + u32 rpc_vers; + struct rpc_procinfo * rpc_proc; +}; + +static const struct rpcb_info rpcb_next_version[]; +static const struct rpcb_info rpcb_next_version6[]; + +static const struct rpc_call_ops rpcb_getport_ops = { + .rpc_call_done = rpcb_getport_done, + .rpc_release = rpcb_map_release, +}; + +static void rpcb_wake_rpcbind_waiters(struct rpc_xprt *xprt, int status) +{ + xprt_clear_binding(xprt); + rpc_wake_up_status(&xprt->binding, status); +} + +static void rpcb_map_release(void *data) +{ + struct rpcbind_args *map = data; + + rpcb_wake_rpcbind_waiters(map->r_xprt, map->r_status); + xprt_put(map->r_xprt); + kfree(map->r_addr); + kfree(map); +} + +static int rpcb_get_local(struct net *net) +{ + int cnt; + struct sunrpc_net *sn = net_generic(net, sunrpc_net_id); + + spin_lock(&sn->rpcb_clnt_lock); + if (sn->rpcb_users) + sn->rpcb_users++; + cnt = sn->rpcb_users; + spin_unlock(&sn->rpcb_clnt_lock); + + return cnt; +} + +void rpcb_put_local(struct net *net) +{ + struct sunrpc_net *sn = net_generic(net, sunrpc_net_id); + struct rpc_clnt *clnt = sn->rpcb_local_clnt; + struct rpc_clnt *clnt4 = sn->rpcb_local_clnt4; + int shutdown = 0; + + spin_lock(&sn->rpcb_clnt_lock); + if (sn->rpcb_users) { + if (--sn->rpcb_users == 0) { + sn->rpcb_local_clnt = NULL; + sn->rpcb_local_clnt4 = NULL; + } + shutdown = !sn->rpcb_users; + } + spin_unlock(&sn->rpcb_clnt_lock); + + if (shutdown) { + /* + * cleanup_rpcb_clnt - remove xprtsock's sysctls, unregister + */ + if (clnt4) + rpc_shutdown_client(clnt4); + if (clnt) + rpc_shutdown_client(clnt); + } +} + +static void rpcb_set_local(struct net *net, struct rpc_clnt *clnt, + struct rpc_clnt *clnt4, + bool is_af_local) +{ + struct sunrpc_net *sn = net_generic(net, sunrpc_net_id); + + /* Protected by rpcb_create_local_mutex */ + sn->rpcb_local_clnt = clnt; + sn->rpcb_local_clnt4 = clnt4; + sn->rpcb_is_af_local = is_af_local ? 1 : 0; + smp_wmb(); + sn->rpcb_users = 1; + dprintk("RPC: created new rpcb local clients (rpcb_local_clnt: " + "%p, rpcb_local_clnt4: %p) for net %p%s\n", + sn->rpcb_local_clnt, sn->rpcb_local_clnt4, + net, (net == &init_net) ? " (init_net)" : ""); +} + +/* + * Returns zero on success, otherwise a negative errno value + * is returned. + */ +static int rpcb_create_local_unix(struct net *net) +{ + static const struct sockaddr_un rpcb_localaddr_rpcbind = { + .sun_family = AF_LOCAL, + .sun_path = RPCBIND_SOCK_PATHNAME, + }; + struct rpc_create_args args = { + .net = net, + .protocol = XPRT_TRANSPORT_LOCAL, + .address = (struct sockaddr *)&rpcb_localaddr_rpcbind, + .addrsize = sizeof(rpcb_localaddr_rpcbind), + .servername = "localhost", + .program = &rpcb_program, + .version = RPCBVERS_2, + .authflavor = RPC_AUTH_NULL, + /* + * We turn off the idle timeout to prevent the kernel + * from automatically disconnecting the socket. + * Otherwise, we'd have to cache the mount namespace + * of the caller and somehow pass that to the socket + * reconnect code. + */ + .flags = RPC_CLNT_CREATE_NO_IDLE_TIMEOUT, + }; + struct rpc_clnt *clnt, *clnt4; + int result = 0; + + /* + * Because we requested an RPC PING at transport creation time, + * this works only if the user space portmapper is rpcbind, and + * it's listening on AF_LOCAL on the named socket. + */ + clnt = rpc_create(&args); + if (IS_ERR(clnt)) { + dprintk("RPC: failed to create AF_LOCAL rpcbind " + "client (errno %ld).\n", PTR_ERR(clnt)); + result = PTR_ERR(clnt); + goto out; + } + + clnt4 = rpc_bind_new_program(clnt, &rpcb_program, RPCBVERS_4); + if (IS_ERR(clnt4)) { + dprintk("RPC: failed to bind second program to " + "rpcbind v4 client (errno %ld).\n", + PTR_ERR(clnt4)); + clnt4 = NULL; + } + + rpcb_set_local(net, clnt, clnt4, true); + +out: + return result; +} + +/* + * Returns zero on success, otherwise a negative errno value + * is returned. + */ +static int rpcb_create_local_net(struct net *net) +{ + static const struct sockaddr_in rpcb_inaddr_loopback = { + .sin_family = AF_INET, + .sin_addr.s_addr = htonl(INADDR_LOOPBACK), + .sin_port = htons(RPCBIND_PORT), + }; + struct rpc_create_args args = { + .net = net, + .protocol = XPRT_TRANSPORT_TCP, + .address = (struct sockaddr *)&rpcb_inaddr_loopback, + .addrsize = sizeof(rpcb_inaddr_loopback), + .servername = "localhost", + .program = &rpcb_program, + .version = RPCBVERS_2, + .authflavor = RPC_AUTH_UNIX, + .flags = RPC_CLNT_CREATE_NOPING, + }; + struct rpc_clnt *clnt, *clnt4; + int result = 0; + + clnt = rpc_create(&args); + if (IS_ERR(clnt)) { + dprintk("RPC: failed to create local rpcbind " + "client (errno %ld).\n", PTR_ERR(clnt)); + result = PTR_ERR(clnt); + goto out; + } + + /* + * This results in an RPC ping. On systems running portmapper, + * the v4 ping will fail. Proceed anyway, but disallow rpcb + * v4 upcalls. + */ + clnt4 = rpc_bind_new_program(clnt, &rpcb_program, RPCBVERS_4); + if (IS_ERR(clnt4)) { + dprintk("RPC: failed to bind second program to " + "rpcbind v4 client (errno %ld).\n", + PTR_ERR(clnt4)); + clnt4 = NULL; + } + + rpcb_set_local(net, clnt, clnt4, false); + +out: + return result; +} + +/* + * Returns zero on success, otherwise a negative errno value + * is returned. + */ +int rpcb_create_local(struct net *net) +{ + static DEFINE_MUTEX(rpcb_create_local_mutex); + int result = 0; + + if (rpcb_get_local(net)) + return result; + + mutex_lock(&rpcb_create_local_mutex); + if (rpcb_get_local(net)) + goto out; + + if (rpcb_create_local_unix(net) != 0) + result = rpcb_create_local_net(net); + +out: + mutex_unlock(&rpcb_create_local_mutex); + return result; +} + +static struct rpc_clnt *rpcb_create(struct net *net, const char *hostname, + struct sockaddr *srvaddr, size_t salen, + int proto, u32 version) +{ + struct rpc_create_args args = { + .net = net, + .protocol = proto, + .address = srvaddr, + .addrsize = salen, + .servername = hostname, + .program = &rpcb_program, + .version = version, + .authflavor = RPC_AUTH_UNIX, + .flags = (RPC_CLNT_CREATE_NOPING | + RPC_CLNT_CREATE_NONPRIVPORT), + }; + + switch (srvaddr->sa_family) { + case AF_INET: + ((struct sockaddr_in *)srvaddr)->sin_port = htons(RPCBIND_PORT); + break; + case AF_INET6: + ((struct sockaddr_in6 *)srvaddr)->sin6_port = htons(RPCBIND_PORT); + break; + default: + return ERR_PTR(-EAFNOSUPPORT); + } + + return rpc_create(&args); +} + +static int rpcb_register_call(struct sunrpc_net *sn, struct rpc_clnt *clnt, struct rpc_message *msg, bool is_set) +{ + int flags = RPC_TASK_NOCONNECT; + int error, result = 0; + + if (is_set || !sn->rpcb_is_af_local) + flags = RPC_TASK_SOFTCONN; + msg->rpc_resp = &result; + + error = rpc_call_sync(clnt, msg, flags); + if (error < 0) { + dprintk("RPC: failed to contact local rpcbind " + "server (errno %d).\n", -error); + return error; + } + + if (!result) + return -EACCES; + return 0; +} + +/** + * rpcb_register - set or unset a port registration with the local rpcbind svc + * @net: target network namespace + * @prog: RPC program number to bind + * @vers: RPC version number to bind + * @prot: transport protocol to register + * @port: port value to register + * + * Returns zero if the registration request was dispatched successfully + * and the rpcbind daemon returned success. Otherwise, returns an errno + * value that reflects the nature of the error (request could not be + * dispatched, timed out, or rpcbind returned an error). + * + * RPC services invoke this function to advertise their contact + * information via the system's rpcbind daemon. RPC services + * invoke this function once for each [program, version, transport] + * tuple they wish to advertise. + * + * Callers may also unregister RPC services that are no longer + * available by setting the passed-in port to zero. This removes + * all registered transports for [program, version] from the local + * rpcbind database. + * + * This function uses rpcbind protocol version 2 to contact the + * local rpcbind daemon. + * + * Registration works over both AF_INET and AF_INET6, and services + * registered via this function are advertised as available for any + * address. If the local rpcbind daemon is listening on AF_INET6, + * services registered via this function will be advertised on + * IN6ADDR_ANY (ie available for all AF_INET and AF_INET6 + * addresses). + */ +int rpcb_register(struct net *net, u32 prog, u32 vers, int prot, unsigned short port) +{ + struct rpcbind_args map = { + .r_prog = prog, + .r_vers = vers, + .r_prot = prot, + .r_port = port, + }; + struct rpc_message msg = { + .rpc_argp = &map, + }; + struct sunrpc_net *sn = net_generic(net, sunrpc_net_id); + bool is_set = false; + + dprintk("RPC: %sregistering (%u, %u, %d, %u) with local " + "rpcbind\n", (port ? "" : "un"), + prog, vers, prot, port); + + msg.rpc_proc = &rpcb_procedures2[RPCBPROC_UNSET]; + if (port != 0) { + msg.rpc_proc = &rpcb_procedures2[RPCBPROC_SET]; + is_set = true; + } + + return rpcb_register_call(sn, sn->rpcb_local_clnt, &msg, is_set); +} + +/* + * Fill in AF_INET family-specific arguments to register + */ +static int rpcb_register_inet4(struct sunrpc_net *sn, + const struct sockaddr *sap, + struct rpc_message *msg) +{ + const struct sockaddr_in *sin = (const struct sockaddr_in *)sap; + struct rpcbind_args *map = msg->rpc_argp; + unsigned short port = ntohs(sin->sin_port); + bool is_set = false; + int result; + + map->r_addr = rpc_sockaddr2uaddr(sap, GFP_KERNEL); + + dprintk("RPC: %sregistering [%u, %u, %s, '%s'] with " + "local rpcbind\n", (port ? "" : "un"), + map->r_prog, map->r_vers, + map->r_addr, map->r_netid); + + msg->rpc_proc = &rpcb_procedures4[RPCBPROC_UNSET]; + if (port != 0) { + msg->rpc_proc = &rpcb_procedures4[RPCBPROC_SET]; + is_set = true; + } + + result = rpcb_register_call(sn, sn->rpcb_local_clnt4, msg, is_set); + kfree(map->r_addr); + return result; +} + +/* + * Fill in AF_INET6 family-specific arguments to register + */ +static int rpcb_register_inet6(struct sunrpc_net *sn, + const struct sockaddr *sap, + struct rpc_message *msg) +{ + const struct sockaddr_in6 *sin6 = (const struct sockaddr_in6 *)sap; + struct rpcbind_args *map = msg->rpc_argp; + unsigned short port = ntohs(sin6->sin6_port); + bool is_set = false; + int result; + + map->r_addr = rpc_sockaddr2uaddr(sap, GFP_KERNEL); + + dprintk("RPC: %sregistering [%u, %u, %s, '%s'] with " + "local rpcbind\n", (port ? "" : "un"), + map->r_prog, map->r_vers, + map->r_addr, map->r_netid); + + msg->rpc_proc = &rpcb_procedures4[RPCBPROC_UNSET]; + if (port != 0) { + msg->rpc_proc = &rpcb_procedures4[RPCBPROC_SET]; + is_set = true; + } + + result = rpcb_register_call(sn, sn->rpcb_local_clnt4, msg, is_set); + kfree(map->r_addr); + return result; +} + +static int rpcb_unregister_all_protofamilies(struct sunrpc_net *sn, + struct rpc_message *msg) +{ + struct rpcbind_args *map = msg->rpc_argp; + + dprintk("RPC: unregistering [%u, %u, '%s'] with " + "local rpcbind\n", + map->r_prog, map->r_vers, map->r_netid); + + map->r_addr = ""; + msg->rpc_proc = &rpcb_procedures4[RPCBPROC_UNSET]; + + return rpcb_register_call(sn, sn->rpcb_local_clnt4, msg, false); +} + +/** + * rpcb_v4_register - set or unset a port registration with the local rpcbind + * @net: target network namespace + * @program: RPC program number of service to (un)register + * @version: RPC version number of service to (un)register + * @address: address family, IP address, and port to (un)register + * @netid: netid of transport protocol to (un)register + * + * Returns zero if the registration request was dispatched successfully + * and the rpcbind daemon returned success. Otherwise, returns an errno + * value that reflects the nature of the error (request could not be + * dispatched, timed out, or rpcbind returned an error). + * + * RPC services invoke this function to advertise their contact + * information via the system's rpcbind daemon. RPC services + * invoke this function once for each [program, version, address, + * netid] tuple they wish to advertise. + * + * Callers may also unregister RPC services that are registered at a + * specific address by setting the port number in @address to zero. + * They may unregister all registered protocol families at once for + * a service by passing a NULL @address argument. If @netid is "" + * then all netids for [program, version, address] are unregistered. + * + * This function uses rpcbind protocol version 4 to contact the + * local rpcbind daemon. The local rpcbind daemon must support + * version 4 of the rpcbind protocol in order for these functions + * to register a service successfully. + * + * Supported netids include "udp" and "tcp" for UDP and TCP over + * IPv4, and "udp6" and "tcp6" for UDP and TCP over IPv6, + * respectively. + * + * The contents of @address determine the address family and the + * port to be registered. The usual practice is to pass INADDR_ANY + * as the raw address, but specifying a non-zero address is also + * supported by this API if the caller wishes to advertise an RPC + * service on a specific network interface. + * + * Note that passing in INADDR_ANY does not create the same service + * registration as IN6ADDR_ANY. The former advertises an RPC + * service on any IPv4 address, but not on IPv6. The latter + * advertises the service on all IPv4 and IPv6 addresses. + */ +int rpcb_v4_register(struct net *net, const u32 program, const u32 version, + const struct sockaddr *address, const char *netid) +{ + struct rpcbind_args map = { + .r_prog = program, + .r_vers = version, + .r_netid = netid, + .r_owner = RPCB_OWNER_STRING, + }; + struct rpc_message msg = { + .rpc_argp = &map, + }; + struct sunrpc_net *sn = net_generic(net, sunrpc_net_id); + + if (sn->rpcb_local_clnt4 == NULL) + return -EPROTONOSUPPORT; + + if (address == NULL) + return rpcb_unregister_all_protofamilies(sn, &msg); + + switch (address->sa_family) { + case AF_INET: + return rpcb_register_inet4(sn, address, &msg); + case AF_INET6: + return rpcb_register_inet6(sn, address, &msg); + } + + return -EAFNOSUPPORT; +} + +static struct rpc_task *rpcb_call_async(struct rpc_clnt *rpcb_clnt, struct rpcbind_args *map, struct rpc_procinfo *proc) +{ + struct rpc_message msg = { + .rpc_proc = proc, + .rpc_argp = map, + .rpc_resp = map, + }; + struct rpc_task_setup task_setup_data = { + .rpc_client = rpcb_clnt, + .rpc_message = &msg, + .callback_ops = &rpcb_getport_ops, + .callback_data = map, + .flags = RPC_TASK_ASYNC | RPC_TASK_SOFTCONN, + }; + + return rpc_run_task(&task_setup_data); +} + +/* + * In the case where rpc clients have been cloned, we want to make + * sure that we use the program number/version etc of the actual + * owner of the xprt. To do so, we walk back up the tree of parents + * to find whoever created the transport and/or whoever has the + * autobind flag set. + */ +static struct rpc_clnt *rpcb_find_transport_owner(struct rpc_clnt *clnt) +{ + struct rpc_clnt *parent = clnt->cl_parent; + struct rpc_xprt *xprt = rcu_dereference(clnt->cl_xprt); + + while (parent != clnt) { + if (rcu_dereference(parent->cl_xprt) != xprt) + break; + if (clnt->cl_autobind) + break; + clnt = parent; + parent = parent->cl_parent; + } + return clnt; +} + +/** + * rpcb_getport_async - obtain the port for a given RPC service on a given host + * @task: task that is waiting for portmapper request + * + * This one can be called for an ongoing RPC request, and can be used in + * an async (rpciod) context. + */ +void rpcb_getport_async(struct rpc_task *task) +{ + struct rpc_clnt *clnt; + struct rpc_procinfo *proc; + u32 bind_version; + struct rpc_xprt *xprt; + struct rpc_clnt *rpcb_clnt; + struct rpcbind_args *map; + struct rpc_task *child; + struct sockaddr_storage addr; + struct sockaddr *sap = (struct sockaddr *)&addr; + size_t salen; + int status; + + rcu_read_lock(); + do { + clnt = rpcb_find_transport_owner(task->tk_client); + xprt = xprt_get(rcu_dereference(clnt->cl_xprt)); + } while (xprt == NULL); + rcu_read_unlock(); + + dprintk("RPC: %5u %s(%s, %u, %u, %d)\n", + task->tk_pid, __func__, + xprt->servername, clnt->cl_prog, clnt->cl_vers, xprt->prot); + + /* Put self on the wait queue to ensure we get notified if + * some other task is already attempting to bind the port */ + rpc_sleep_on(&xprt->binding, task, NULL); + + if (xprt_test_and_set_binding(xprt)) { + dprintk("RPC: %5u %s: waiting for another binder\n", + task->tk_pid, __func__); + xprt_put(xprt); + return; + } + + /* Someone else may have bound if we slept */ + if (xprt_bound(xprt)) { + status = 0; + dprintk("RPC: %5u %s: already bound\n", + task->tk_pid, __func__); + goto bailout_nofree; + } + + /* Parent transport's destination address */ + salen = rpc_peeraddr(clnt, sap, sizeof(addr)); + + /* Don't ever use rpcbind v2 for AF_INET6 requests */ + switch (sap->sa_family) { + case AF_INET: + proc = rpcb_next_version[xprt->bind_index].rpc_proc; + bind_version = rpcb_next_version[xprt->bind_index].rpc_vers; + break; + case AF_INET6: + proc = rpcb_next_version6[xprt->bind_index].rpc_proc; + bind_version = rpcb_next_version6[xprt->bind_index].rpc_vers; + break; + default: + status = -EAFNOSUPPORT; + dprintk("RPC: %5u %s: bad address family\n", + task->tk_pid, __func__); + goto bailout_nofree; + } + if (proc == NULL) { + xprt->bind_index = 0; + status = -EPFNOSUPPORT; + dprintk("RPC: %5u %s: no more getport versions available\n", + task->tk_pid, __func__); + goto bailout_nofree; + } + + dprintk("RPC: %5u %s: trying rpcbind version %u\n", + task->tk_pid, __func__, bind_version); + + rpcb_clnt = rpcb_create(xprt->xprt_net, xprt->servername, sap, salen, + xprt->prot, bind_version); + if (IS_ERR(rpcb_clnt)) { + status = PTR_ERR(rpcb_clnt); + dprintk("RPC: %5u %s: rpcb_create failed, error %ld\n", + task->tk_pid, __func__, PTR_ERR(rpcb_clnt)); + goto bailout_nofree; + } + + map = kzalloc(sizeof(struct rpcbind_args), GFP_ATOMIC); + if (!map) { + status = -ENOMEM; + dprintk("RPC: %5u %s: no memory available\n", + task->tk_pid, __func__); + goto bailout_release_client; + } + map->r_prog = clnt->cl_prog; + map->r_vers = clnt->cl_vers; + map->r_prot = xprt->prot; + map->r_port = 0; + map->r_xprt = xprt; + map->r_status = -EIO; + + switch (bind_version) { + case RPCBVERS_4: + case RPCBVERS_3: + map->r_netid = xprt->address_strings[RPC_DISPLAY_NETID]; + map->r_addr = rpc_sockaddr2uaddr(sap, GFP_ATOMIC); + map->r_owner = ""; + break; + case RPCBVERS_2: + map->r_addr = NULL; + break; + default: + BUG(); + } + + child = rpcb_call_async(rpcb_clnt, map, proc); + rpc_release_client(rpcb_clnt); + if (IS_ERR(child)) { + /* rpcb_map_release() has freed the arguments */ + dprintk("RPC: %5u %s: rpc_run_task failed\n", + task->tk_pid, __func__); + return; + } + + xprt->stat.bind_count++; + rpc_put_task(child); + return; + +bailout_release_client: + rpc_release_client(rpcb_clnt); +bailout_nofree: + rpcb_wake_rpcbind_waiters(xprt, status); + task->tk_status = status; + xprt_put(xprt); +} +EXPORT_SYMBOL_GPL(rpcb_getport_async); + +/* + * Rpcbind child task calls this callback via tk_exit. + */ +static void rpcb_getport_done(struct rpc_task *child, void *data) +{ + struct rpcbind_args *map = data; + struct rpc_xprt *xprt = map->r_xprt; + int status = child->tk_status; + + /* Garbage reply: retry with a lesser rpcbind version */ + if (status == -EIO) + status = -EPROTONOSUPPORT; + + /* rpcbind server doesn't support this rpcbind protocol version */ + if (status == -EPROTONOSUPPORT) + xprt->bind_index++; + + if (status < 0) { + /* rpcbind server not available on remote host? */ + xprt->ops->set_port(xprt, 0); + } else if (map->r_port == 0) { + /* Requested RPC service wasn't registered on remote host */ + xprt->ops->set_port(xprt, 0); + status = -EACCES; + } else { + /* Succeeded */ + xprt->ops->set_port(xprt, map->r_port); + xprt_set_bound(xprt); + status = 0; + } + + dprintk("RPC: %5u rpcb_getport_done(status %d, port %u)\n", + child->tk_pid, status, map->r_port); + + map->r_status = status; +} + +/* + * XDR functions for rpcbind + */ + +static void rpcb_enc_mapping(struct rpc_rqst *req, struct xdr_stream *xdr, + const struct rpcbind_args *rpcb) +{ + __be32 *p; + + dprintk("RPC: %5u encoding PMAP_%s call (%u, %u, %d, %u)\n", + req->rq_task->tk_pid, + req->rq_task->tk_msg.rpc_proc->p_name, + rpcb->r_prog, rpcb->r_vers, rpcb->r_prot, rpcb->r_port); + + p = xdr_reserve_space(xdr, RPCB_mappingargs_sz << 2); + *p++ = cpu_to_be32(rpcb->r_prog); + *p++ = cpu_to_be32(rpcb->r_vers); + *p++ = cpu_to_be32(rpcb->r_prot); + *p = cpu_to_be32(rpcb->r_port); +} + +static int rpcb_dec_getport(struct rpc_rqst *req, struct xdr_stream *xdr, + struct rpcbind_args *rpcb) +{ + unsigned long port; + __be32 *p; + + rpcb->r_port = 0; + + p = xdr_inline_decode(xdr, 4); + if (unlikely(p == NULL)) + return -EIO; + + port = be32_to_cpup(p); + dprintk("RPC: %5u PMAP_%s result: %lu\n", req->rq_task->tk_pid, + req->rq_task->tk_msg.rpc_proc->p_name, port); + if (unlikely(port > USHRT_MAX)) + return -EIO; + + rpcb->r_port = port; + return 0; +} + +static int rpcb_dec_set(struct rpc_rqst *req, struct xdr_stream *xdr, + unsigned int *boolp) +{ + __be32 *p; + + p = xdr_inline_decode(xdr, 4); + if (unlikely(p == NULL)) + return -EIO; + + *boolp = 0; + if (*p != xdr_zero) + *boolp = 1; + + dprintk("RPC: %5u RPCB_%s call %s\n", + req->rq_task->tk_pid, + req->rq_task->tk_msg.rpc_proc->p_name, + (*boolp ? "succeeded" : "failed")); + return 0; +} + +static void encode_rpcb_string(struct xdr_stream *xdr, const char *string, + const u32 maxstrlen) +{ + __be32 *p; + u32 len; + + len = strlen(string); + WARN_ON_ONCE(len > maxstrlen); + if (len > maxstrlen) + /* truncate and hope for the best */ + len = maxstrlen; + p = xdr_reserve_space(xdr, 4 + len); + xdr_encode_opaque(p, string, len); +} + +static void rpcb_enc_getaddr(struct rpc_rqst *req, struct xdr_stream *xdr, + const struct rpcbind_args *rpcb) +{ + __be32 *p; + + dprintk("RPC: %5u encoding RPCB_%s call (%u, %u, '%s', '%s')\n", + req->rq_task->tk_pid, + req->rq_task->tk_msg.rpc_proc->p_name, + rpcb->r_prog, rpcb->r_vers, + rpcb->r_netid, rpcb->r_addr); + + p = xdr_reserve_space(xdr, (RPCB_program_sz + RPCB_version_sz) << 2); + *p++ = cpu_to_be32(rpcb->r_prog); + *p = cpu_to_be32(rpcb->r_vers); + + encode_rpcb_string(xdr, rpcb->r_netid, RPCBIND_MAXNETIDLEN); + encode_rpcb_string(xdr, rpcb->r_addr, RPCBIND_MAXUADDRLEN); + encode_rpcb_string(xdr, rpcb->r_owner, RPCB_MAXOWNERLEN); +} + +static int rpcb_dec_getaddr(struct rpc_rqst *req, struct xdr_stream *xdr, + struct rpcbind_args *rpcb) +{ + struct sockaddr_storage address; + struct sockaddr *sap = (struct sockaddr *)&address; + __be32 *p; + u32 len; + + rpcb->r_port = 0; + + p = xdr_inline_decode(xdr, 4); + if (unlikely(p == NULL)) + goto out_fail; + len = be32_to_cpup(p); + + /* + * If the returned universal address is a null string, + * the requested RPC service was not registered. + */ + if (len == 0) { + dprintk("RPC: %5u RPCB reply: program not registered\n", + req->rq_task->tk_pid); + return 0; + } + + if (unlikely(len > RPCBIND_MAXUADDRLEN)) + goto out_fail; + + p = xdr_inline_decode(xdr, len); + if (unlikely(p == NULL)) + goto out_fail; + dprintk("RPC: %5u RPCB_%s reply: %s\n", req->rq_task->tk_pid, + req->rq_task->tk_msg.rpc_proc->p_name, (char *)p); + + if (rpc_uaddr2sockaddr(req->rq_xprt->xprt_net, (char *)p, len, + sap, sizeof(address)) == 0) + goto out_fail; + rpcb->r_port = rpc_get_port(sap); + + return 0; + +out_fail: + dprintk("RPC: %5u malformed RPCB_%s reply\n", + req->rq_task->tk_pid, + req->rq_task->tk_msg.rpc_proc->p_name); + return -EIO; +} + +/* + * Not all rpcbind procedures described in RFC 1833 are implemented + * since the Linux kernel RPC code requires only these. + */ + +static struct rpc_procinfo rpcb_procedures2[] = { + [RPCBPROC_SET] = { + .p_proc = RPCBPROC_SET, + .p_encode = (kxdreproc_t)rpcb_enc_mapping, + .p_decode = (kxdrdproc_t)rpcb_dec_set, + .p_arglen = RPCB_mappingargs_sz, + .p_replen = RPCB_setres_sz, + .p_statidx = RPCBPROC_SET, + .p_timer = 0, + .p_name = "SET", + }, + [RPCBPROC_UNSET] = { + .p_proc = RPCBPROC_UNSET, + .p_encode = (kxdreproc_t)rpcb_enc_mapping, + .p_decode = (kxdrdproc_t)rpcb_dec_set, + .p_arglen = RPCB_mappingargs_sz, + .p_replen = RPCB_setres_sz, + .p_statidx = RPCBPROC_UNSET, + .p_timer = 0, + .p_name = "UNSET", + }, + [RPCBPROC_GETPORT] = { + .p_proc = RPCBPROC_GETPORT, + .p_encode = (kxdreproc_t)rpcb_enc_mapping, + .p_decode = (kxdrdproc_t)rpcb_dec_getport, + .p_arglen = RPCB_mappingargs_sz, + .p_replen = RPCB_getportres_sz, + .p_statidx = RPCBPROC_GETPORT, + .p_timer = 0, + .p_name = "GETPORT", + }, +}; + +static struct rpc_procinfo rpcb_procedures3[] = { + [RPCBPROC_SET] = { + .p_proc = RPCBPROC_SET, + .p_encode = (kxdreproc_t)rpcb_enc_getaddr, + .p_decode = (kxdrdproc_t)rpcb_dec_set, + .p_arglen = RPCB_getaddrargs_sz, + .p_replen = RPCB_setres_sz, + .p_statidx = RPCBPROC_SET, + .p_timer = 0, + .p_name = "SET", + }, + [RPCBPROC_UNSET] = { + .p_proc = RPCBPROC_UNSET, + .p_encode = (kxdreproc_t)rpcb_enc_getaddr, + .p_decode = (kxdrdproc_t)rpcb_dec_set, + .p_arglen = RPCB_getaddrargs_sz, + .p_replen = RPCB_setres_sz, + .p_statidx = RPCBPROC_UNSET, + .p_timer = 0, + .p_name = "UNSET", + }, + [RPCBPROC_GETADDR] = { + .p_proc = RPCBPROC_GETADDR, + .p_encode = (kxdreproc_t)rpcb_enc_getaddr, + .p_decode = (kxdrdproc_t)rpcb_dec_getaddr, + .p_arglen = RPCB_getaddrargs_sz, + .p_replen = RPCB_getaddrres_sz, + .p_statidx = RPCBPROC_GETADDR, + .p_timer = 0, + .p_name = "GETADDR", + }, +}; + +static struct rpc_procinfo rpcb_procedures4[] = { + [RPCBPROC_SET] = { + .p_proc = RPCBPROC_SET, + .p_encode = (kxdreproc_t)rpcb_enc_getaddr, + .p_decode = (kxdrdproc_t)rpcb_dec_set, + .p_arglen = RPCB_getaddrargs_sz, + .p_replen = RPCB_setres_sz, + .p_statidx = RPCBPROC_SET, + .p_timer = 0, + .p_name = "SET", + }, + [RPCBPROC_UNSET] = { + .p_proc = RPCBPROC_UNSET, + .p_encode = (kxdreproc_t)rpcb_enc_getaddr, + .p_decode = (kxdrdproc_t)rpcb_dec_set, + .p_arglen = RPCB_getaddrargs_sz, + .p_replen = RPCB_setres_sz, + .p_statidx = RPCBPROC_UNSET, + .p_timer = 0, + .p_name = "UNSET", + }, + [RPCBPROC_GETADDR] = { + .p_proc = RPCBPROC_GETADDR, + .p_encode = (kxdreproc_t)rpcb_enc_getaddr, + .p_decode = (kxdrdproc_t)rpcb_dec_getaddr, + .p_arglen = RPCB_getaddrargs_sz, + .p_replen = RPCB_getaddrres_sz, + .p_statidx = RPCBPROC_GETADDR, + .p_timer = 0, + .p_name = "GETADDR", + }, +}; + +static const struct rpcb_info rpcb_next_version[] = { + { + .rpc_vers = RPCBVERS_2, + .rpc_proc = &rpcb_procedures2[RPCBPROC_GETPORT], + }, + { + .rpc_proc = NULL, + }, +}; + +static const struct rpcb_info rpcb_next_version6[] = { + { + .rpc_vers = RPCBVERS_4, + .rpc_proc = &rpcb_procedures4[RPCBPROC_GETADDR], + }, + { + .rpc_vers = RPCBVERS_3, + .rpc_proc = &rpcb_procedures3[RPCBPROC_GETADDR], + }, + { + .rpc_proc = NULL, + }, +}; + +static const struct rpc_version rpcb_version2 = { + .number = RPCBVERS_2, + .nrprocs = ARRAY_SIZE(rpcb_procedures2), + .procs = rpcb_procedures2 +}; + +static const struct rpc_version rpcb_version3 = { + .number = RPCBVERS_3, + .nrprocs = ARRAY_SIZE(rpcb_procedures3), + .procs = rpcb_procedures3 +}; + +static const struct rpc_version rpcb_version4 = { + .number = RPCBVERS_4, + .nrprocs = ARRAY_SIZE(rpcb_procedures4), + .procs = rpcb_procedures4 +}; + +static const struct rpc_version *rpcb_version[] = { + NULL, + NULL, + &rpcb_version2, + &rpcb_version3, + &rpcb_version4 +}; + +static struct rpc_stat rpcb_stats; + +static const struct rpc_program rpcb_program = { + .name = "rpcbind", + .number = RPCBIND_PROGRAM, + .nrvers = ARRAY_SIZE(rpcb_version), + .version = rpcb_version, + .stats = &rpcb_stats, +}; diff --git a/net/sunrpc/sched.c b/net/sunrpc/sched.c index 54e60a65750..c0365c14b85 100644 --- a/net/sunrpc/sched.c +++ b/net/sunrpc/sched.c @@ -4,7 +4,7 @@ * Scheduling for synchronous and asynchronous RPC requests. * * Copyright (C) 1996 Olaf Kirch, <okir@monad.swb.de> - * + * * TCP NFS related read + write fixes * (C) 1999 Dave Airlie, University of Limerick, Ireland <airlied@linux.ie> */ @@ -16,147 +16,139 @@ #include <linux/slab.h> #include <linux/mempool.h> #include <linux/smp.h> -#include <linux/smp_lock.h> #include <linux/spinlock.h> +#include <linux/mutex.h> +#include <linux/freezer.h> #include <linux/sunrpc/clnt.h> -#include <linux/sunrpc/xprt.h> + +#include "sunrpc.h" #ifdef RPC_DEBUG #define RPCDBG_FACILITY RPCDBG_SCHED -#define RPC_TASK_MAGIC_ID 0xf00baa -static int rpc_task_id; #endif +#define CREATE_TRACE_POINTS +#include <trace/events/sunrpc.h> + /* * RPC slabs and memory pools */ #define RPC_BUFFER_MAXSIZE (2048) #define RPC_BUFFER_POOLSIZE (8) #define RPC_TASK_POOLSIZE (8) -static kmem_cache_t *rpc_task_slabp __read_mostly; -static kmem_cache_t *rpc_buffer_slabp __read_mostly; +static struct kmem_cache *rpc_task_slabp __read_mostly; +static struct kmem_cache *rpc_buffer_slabp __read_mostly; static mempool_t *rpc_task_mempool __read_mostly; static mempool_t *rpc_buffer_mempool __read_mostly; -static void __rpc_default_timer(struct rpc_task *task); -static void rpciod_killall(void); -static void rpc_free(struct rpc_task *task); - -static void rpc_async_schedule(void *); - -/* - * RPC tasks that create another task (e.g. for contacting the portmapper) - * will wait on this queue for their child's completion - */ -static RPC_WAITQ(childq, "childq"); +static void rpc_async_schedule(struct work_struct *); +static void rpc_release_task(struct rpc_task *task); +static void __rpc_queue_timer_fn(unsigned long ptr); /* * RPC tasks sit here while waiting for conditions to improve. */ -static RPC_WAITQ(delay_queue, "delayq"); - -/* - * All RPC tasks are linked into this list - */ -static LIST_HEAD(all_tasks); +static struct rpc_wait_queue delay_queue; /* * rpciod-related stuff */ -static DECLARE_MUTEX(rpciod_sema); -static unsigned int rpciod_users; -static struct workqueue_struct *rpciod_workqueue; - -/* - * Spinlock for other critical sections of code. - */ -static DEFINE_SPINLOCK(rpc_sched_lock); +struct workqueue_struct *rpciod_workqueue; /* * Disable the timer for a given RPC task. Should be called with * queue->lock and bh_disabled in order to avoid races within * rpc_run_timer(). */ -static inline void -__rpc_disable_timer(struct rpc_task *task) +static void +__rpc_disable_timer(struct rpc_wait_queue *queue, struct rpc_task *task) { - dprintk("RPC: %4d disabling timer\n", task->tk_pid); - task->tk_timeout_fn = NULL; + if (task->tk_timeout == 0) + return; + dprintk("RPC: %5u disabling timer\n", task->tk_pid); task->tk_timeout = 0; + list_del(&task->u.tk_wait.timer_list); + if (list_empty(&queue->timer_list.list)) + del_timer(&queue->timer_list.timer); } -/* - * Run a timeout function. - * We use the callback in order to allow __rpc_wake_up_task() - * and friends to disable the timer synchronously on SMP systems - * without calling del_timer_sync(). The latter could cause a - * deadlock if called while we're holding spinlocks... - */ -static void rpc_run_timer(struct rpc_task *task) +static void +rpc_set_queue_timer(struct rpc_wait_queue *queue, unsigned long expires) { - void (*callback)(struct rpc_task *); - - callback = task->tk_timeout_fn; - task->tk_timeout_fn = NULL; - if (callback && RPC_IS_QUEUED(task)) { - dprintk("RPC: %4d running timer\n", task->tk_pid); - callback(task); - } - smp_mb__before_clear_bit(); - clear_bit(RPC_TASK_HAS_TIMER, &task->tk_runstate); - smp_mb__after_clear_bit(); + queue->timer_list.expires = expires; + mod_timer(&queue->timer_list.timer, expires); } /* * Set up a timer for the current task. */ -static inline void -__rpc_add_timer(struct rpc_task *task, rpc_action timer) +static void +__rpc_add_timer(struct rpc_wait_queue *queue, struct rpc_task *task) { if (!task->tk_timeout) return; - dprintk("RPC: %4d setting alarm for %lu ms\n", + dprintk("RPC: %5u setting alarm for %lu ms\n", task->tk_pid, task->tk_timeout * 1000 / HZ); - if (timer) - task->tk_timeout_fn = timer; - else - task->tk_timeout_fn = __rpc_default_timer; - set_bit(RPC_TASK_HAS_TIMER, &task->tk_runstate); - mod_timer(&task->tk_timer, jiffies + task->tk_timeout); + task->u.tk_wait.expires = jiffies + task->tk_timeout; + if (list_empty(&queue->timer_list.list) || time_before(task->u.tk_wait.expires, queue->timer_list.expires)) + rpc_set_queue_timer(queue, task->u.tk_wait.expires); + list_add(&task->u.tk_wait.timer_list, &queue->timer_list.list); } -/* - * Delete any timer for the current task. Because we use del_timer_sync(), - * this function should never be called while holding queue->lock. - */ -static void -rpc_delete_timer(struct rpc_task *task) +static void rpc_rotate_queue_owner(struct rpc_wait_queue *queue) { - if (RPC_IS_QUEUED(task)) - return; - if (test_and_clear_bit(RPC_TASK_HAS_TIMER, &task->tk_runstate)) { - del_singleshot_timer_sync(&task->tk_timer); - dprintk("RPC: %4d deleting timer\n", task->tk_pid); + struct list_head *q = &queue->tasks[queue->priority]; + struct rpc_task *task; + + if (!list_empty(q)) { + task = list_first_entry(q, struct rpc_task, u.tk_wait.list); + if (task->tk_owner == queue->owner) + list_move_tail(&task->u.tk_wait.list, q); + } +} + +static void rpc_set_waitqueue_priority(struct rpc_wait_queue *queue, int priority) +{ + if (queue->priority != priority) { + /* Fairness: rotate the list when changing priority */ + rpc_rotate_queue_owner(queue); + queue->priority = priority; } } +static void rpc_set_waitqueue_owner(struct rpc_wait_queue *queue, pid_t pid) +{ + queue->owner = pid; + queue->nr = RPC_BATCH_COUNT; +} + +static void rpc_reset_waitqueue_priority(struct rpc_wait_queue *queue) +{ + rpc_set_waitqueue_priority(queue, queue->maxpriority); + rpc_set_waitqueue_owner(queue, 0); +} + /* * Add new request to a priority queue. */ -static void __rpc_add_wait_queue_priority(struct rpc_wait_queue *queue, struct rpc_task *task) +static void __rpc_add_wait_queue_priority(struct rpc_wait_queue *queue, + struct rpc_task *task, + unsigned char queue_priority) { struct list_head *q; struct rpc_task *t; INIT_LIST_HEAD(&task->u.tk_wait.links); - q = &queue->tasks[task->tk_priority]; - if (unlikely(task->tk_priority > queue->maxpriority)) - q = &queue->tasks[queue->maxpriority]; + if (unlikely(queue_priority > queue->maxpriority)) + queue_priority = queue->maxpriority; + if (queue_priority > queue->priority) + rpc_set_waitqueue_priority(queue, queue_priority); + q = &queue->tasks[queue_priority]; list_for_each_entry(t, q, u.tk_wait.list) { - if (t->tk_cookie == task->tk_cookie) { + if (t->tk_owner == task->tk_owner) { list_add_tail(&task->u.tk_wait.list, &t->u.tk_wait.links); return; } @@ -172,21 +164,28 @@ static void __rpc_add_wait_queue_priority(struct rpc_wait_queue *queue, struct r * improve overall performance. * Everyone else gets appended to the queue to ensure proper FIFO behavior. */ -static void __rpc_add_wait_queue(struct rpc_wait_queue *queue, struct rpc_task *task) +static void __rpc_add_wait_queue(struct rpc_wait_queue *queue, + struct rpc_task *task, + unsigned char queue_priority) { - BUG_ON (RPC_IS_QUEUED(task)); + WARN_ON_ONCE(RPC_IS_QUEUED(task)); + if (RPC_IS_QUEUED(task)) + return; if (RPC_IS_PRIORITY(queue)) - __rpc_add_wait_queue_priority(queue, task); + __rpc_add_wait_queue_priority(queue, task, queue_priority); else if (RPC_IS_SWAPPER(task)) list_add(&task->u.tk_wait.list, &queue->tasks[0]); else list_add_tail(&task->u.tk_wait.list, &queue->tasks[0]); - task->u.tk_wait.rpc_waitq = queue; + task->tk_waitqueue = queue; + queue->qlen++; + /* barrier matches the read in rpc_wake_up_task_queue_locked() */ + smp_wmb(); rpc_set_queued(task); - dprintk("RPC: %4d added to queue %p \"%s\"\n", - task->tk_pid, queue, rpc_qname(queue)); + dprintk("RPC: %5u added to queue %p \"%s\"\n", + task->tk_pid, queue, rpc_qname(queue)); } /* @@ -201,232 +200,277 @@ static void __rpc_remove_wait_queue_priority(struct rpc_task *task) list_move(&t->u.tk_wait.list, &task->u.tk_wait.list); list_splice_init(&task->u.tk_wait.links, &t->u.tk_wait.links); } - list_del(&task->u.tk_wait.list); } /* * Remove request from queue. * Note: must be called with spin lock held. */ -static void __rpc_remove_wait_queue(struct rpc_task *task) +static void __rpc_remove_wait_queue(struct rpc_wait_queue *queue, struct rpc_task *task) { - struct rpc_wait_queue *queue; - queue = task->u.tk_wait.rpc_waitq; - + __rpc_disable_timer(queue, task); if (RPC_IS_PRIORITY(queue)) __rpc_remove_wait_queue_priority(task); - else - list_del(&task->u.tk_wait.list); - dprintk("RPC: %4d removed from queue %p \"%s\"\n", - task->tk_pid, queue, rpc_qname(queue)); + list_del(&task->u.tk_wait.list); + queue->qlen--; + dprintk("RPC: %5u removed from queue %p \"%s\"\n", + task->tk_pid, queue, rpc_qname(queue)); } -static inline void rpc_set_waitqueue_priority(struct rpc_wait_queue *queue, int priority) +static void __rpc_init_priority_wait_queue(struct rpc_wait_queue *queue, const char *qname, unsigned char nr_queues) { - queue->priority = priority; - queue->count = 1 << (priority * 2); + int i; + + spin_lock_init(&queue->lock); + for (i = 0; i < ARRAY_SIZE(queue->tasks); i++) + INIT_LIST_HEAD(&queue->tasks[i]); + queue->maxpriority = nr_queues - 1; + rpc_reset_waitqueue_priority(queue); + queue->qlen = 0; + setup_timer(&queue->timer_list.timer, __rpc_queue_timer_fn, (unsigned long)queue); + INIT_LIST_HEAD(&queue->timer_list.list); + rpc_assign_waitqueue_name(queue, qname); } -static inline void rpc_set_waitqueue_cookie(struct rpc_wait_queue *queue, unsigned long cookie) +void rpc_init_priority_wait_queue(struct rpc_wait_queue *queue, const char *qname) { - queue->cookie = cookie; - queue->nr = RPC_BATCH_COUNT; + __rpc_init_priority_wait_queue(queue, qname, RPC_NR_PRIORITY); } +EXPORT_SYMBOL_GPL(rpc_init_priority_wait_queue); -static inline void rpc_reset_waitqueue_priority(struct rpc_wait_queue *queue) +void rpc_init_wait_queue(struct rpc_wait_queue *queue, const char *qname) { - rpc_set_waitqueue_priority(queue, queue->maxpriority); - rpc_set_waitqueue_cookie(queue, 0); + __rpc_init_priority_wait_queue(queue, qname, 1); } +EXPORT_SYMBOL_GPL(rpc_init_wait_queue); -static void __rpc_init_priority_wait_queue(struct rpc_wait_queue *queue, const char *qname, int maxprio) +void rpc_destroy_wait_queue(struct rpc_wait_queue *queue) { - int i; + del_timer_sync(&queue->timer_list.timer); +} +EXPORT_SYMBOL_GPL(rpc_destroy_wait_queue); - spin_lock_init(&queue->lock); - for (i = 0; i < ARRAY_SIZE(queue->tasks); i++) - INIT_LIST_HEAD(&queue->tasks[i]); - queue->maxpriority = maxprio; - rpc_reset_waitqueue_priority(queue); -#ifdef RPC_DEBUG - queue->name = qname; +static int rpc_wait_bit_killable(void *word) +{ + if (fatal_signal_pending(current)) + return -ERESTARTSYS; + freezable_schedule_unsafe(); + return 0; +} + +#if defined(RPC_DEBUG) || defined(RPC_TRACEPOINTS) +static void rpc_task_set_debuginfo(struct rpc_task *task) +{ + static atomic_t rpc_pid; + + task->tk_pid = atomic_inc_return(&rpc_pid); +} +#else +static inline void rpc_task_set_debuginfo(struct rpc_task *task) +{ +} #endif + +static void rpc_set_active(struct rpc_task *task) +{ + trace_rpc_task_begin(task->tk_client, task, NULL); + + rpc_task_set_debuginfo(task); + set_bit(RPC_TASK_ACTIVE, &task->tk_runstate); } -void rpc_init_priority_wait_queue(struct rpc_wait_queue *queue, const char *qname) +/* + * Mark an RPC call as having completed by clearing the 'active' bit + * and then waking up all tasks that were sleeping. + */ +static int rpc_complete_task(struct rpc_task *task) { - __rpc_init_priority_wait_queue(queue, qname, RPC_PRIORITY_HIGH); + void *m = &task->tk_runstate; + wait_queue_head_t *wq = bit_waitqueue(m, RPC_TASK_ACTIVE); + struct wait_bit_key k = __WAIT_BIT_KEY_INITIALIZER(m, RPC_TASK_ACTIVE); + unsigned long flags; + int ret; + + trace_rpc_task_complete(task->tk_client, task, NULL); + + spin_lock_irqsave(&wq->lock, flags); + clear_bit(RPC_TASK_ACTIVE, &task->tk_runstate); + ret = atomic_dec_and_test(&task->tk_count); + if (waitqueue_active(wq)) + __wake_up_locked_key(wq, TASK_NORMAL, &k); + spin_unlock_irqrestore(&wq->lock, flags); + return ret; } -void rpc_init_wait_queue(struct rpc_wait_queue *queue, const char *qname) +/* + * Allow callers to wait for completion of an RPC call + * + * Note the use of out_of_line_wait_on_bit() rather than wait_on_bit() + * to enforce taking of the wq->lock and hence avoid races with + * rpc_complete_task(). + */ +int __rpc_wait_for_completion_task(struct rpc_task *task, int (*action)(void *)) { - __rpc_init_priority_wait_queue(queue, qname, 0); + if (action == NULL) + action = rpc_wait_bit_killable; + return out_of_line_wait_on_bit(&task->tk_runstate, RPC_TASK_ACTIVE, + action, TASK_KILLABLE); } -EXPORT_SYMBOL(rpc_init_wait_queue); +EXPORT_SYMBOL_GPL(__rpc_wait_for_completion_task); /* * Make an RPC task runnable. * - * Note: If the task is ASYNC, this must be called with - * the spinlock held to protect the wait queue operation. + * Note: If the task is ASYNC, and is being made runnable after sitting on an + * rpc_wait_queue, this must be called with the queue spinlock held to protect + * the wait queue operation. + * Note the ordering of rpc_test_and_set_running() and rpc_clear_queued(), + * which is needed to ensure that __rpc_execute() doesn't loop (due to the + * lockless RPC_IS_QUEUED() test) before we've had a chance to test + * the RPC_TASK_RUNNING flag. */ static void rpc_make_runnable(struct rpc_task *task) { - int do_ret; + bool need_wakeup = !rpc_test_and_set_running(task); - BUG_ON(task->tk_timeout_fn); - do_ret = rpc_test_and_set_running(task); rpc_clear_queued(task); - if (do_ret) + if (!need_wakeup) return; if (RPC_IS_ASYNC(task)) { - int status; - - INIT_WORK(&task->u.tk_work, rpc_async_schedule, (void *)task); - status = queue_work(task->tk_workqueue, &task->u.tk_work); - if (status < 0) { - printk(KERN_WARNING "RPC: failed to add task to queue: error: %d!\n", status); - task->tk_status = status; - return; - } + INIT_WORK(&task->u.tk_work, rpc_async_schedule); + queue_work(rpciod_workqueue, &task->u.tk_work); } else wake_up_bit(&task->tk_runstate, RPC_TASK_QUEUED); } /* - * Place a newly initialized task on the workqueue. - */ -static inline void -rpc_schedule_run(struct rpc_task *task) -{ - /* Don't run a child twice! */ - if (RPC_IS_ACTIVATED(task)) - return; - task->tk_active = 1; - rpc_make_runnable(task); -} - -/* * Prepare for sleeping on a wait queue. * By always appending tasks to the list we ensure FIFO behavior. * NB: An RPC task will only receive interrupt-driven events as long * as it's on a wait queue. */ -static void __rpc_sleep_on(struct rpc_wait_queue *q, struct rpc_task *task, - rpc_action action, rpc_action timer) +static void __rpc_sleep_on_priority(struct rpc_wait_queue *q, + struct rpc_task *task, + rpc_action action, + unsigned char queue_priority) { - dprintk("RPC: %4d sleep_on(queue \"%s\" time %ld)\n", task->tk_pid, - rpc_qname(q), jiffies); - - if (!RPC_IS_ASYNC(task) && !RPC_IS_ACTIVATED(task)) { - printk(KERN_ERR "RPC: Inactive synchronous task put to sleep!\n"); - return; - } + dprintk("RPC: %5u sleep_on(queue \"%s\" time %lu)\n", + task->tk_pid, rpc_qname(q), jiffies); - /* Mark the task as being activated if so needed */ - if (!RPC_IS_ACTIVATED(task)) - task->tk_active = 1; + trace_rpc_task_sleep(task->tk_client, task, q); - __rpc_add_wait_queue(q, task); + __rpc_add_wait_queue(q, task, queue_priority); - BUG_ON(task->tk_callback != NULL); + WARN_ON_ONCE(task->tk_callback != NULL); task->tk_callback = action; - __rpc_add_timer(task, timer); + __rpc_add_timer(q, task); } void rpc_sleep_on(struct rpc_wait_queue *q, struct rpc_task *task, - rpc_action action, rpc_action timer) + rpc_action action) { + /* We shouldn't ever put an inactive task to sleep */ + WARN_ON_ONCE(!RPC_IS_ACTIVATED(task)); + if (!RPC_IS_ACTIVATED(task)) { + task->tk_status = -EIO; + rpc_put_task_async(task); + return; + } + + /* + * Protect the queue operations. + */ + spin_lock_bh(&q->lock); + __rpc_sleep_on_priority(q, task, action, task->tk_priority); + spin_unlock_bh(&q->lock); +} +EXPORT_SYMBOL_GPL(rpc_sleep_on); + +void rpc_sleep_on_priority(struct rpc_wait_queue *q, struct rpc_task *task, + rpc_action action, int priority) +{ + /* We shouldn't ever put an inactive task to sleep */ + WARN_ON_ONCE(!RPC_IS_ACTIVATED(task)); + if (!RPC_IS_ACTIVATED(task)) { + task->tk_status = -EIO; + rpc_put_task_async(task); + return; + } + /* * Protect the queue operations. */ spin_lock_bh(&q->lock); - __rpc_sleep_on(q, task, action, timer); + __rpc_sleep_on_priority(q, task, action, priority - RPC_PRIORITY_LOW); spin_unlock_bh(&q->lock); } +EXPORT_SYMBOL_GPL(rpc_sleep_on_priority); /** * __rpc_do_wake_up_task - wake up a single rpc_task + * @queue: wait queue * @task: task to be woken up * * Caller must hold queue->lock, and have cleared the task queued flag. */ -static void __rpc_do_wake_up_task(struct rpc_task *task) +static void __rpc_do_wake_up_task(struct rpc_wait_queue *queue, struct rpc_task *task) { - dprintk("RPC: %4d __rpc_wake_up_task (now %ld)\n", task->tk_pid, jiffies); + dprintk("RPC: %5u __rpc_wake_up_task (now %lu)\n", + task->tk_pid, jiffies); -#ifdef RPC_DEBUG - BUG_ON(task->tk_magic != RPC_TASK_MAGIC_ID); -#endif /* Has the task been executed yet? If not, we cannot wake it up! */ if (!RPC_IS_ACTIVATED(task)) { printk(KERN_ERR "RPC: Inactive task (%p) being woken up!\n", task); return; } - __rpc_disable_timer(task); - __rpc_remove_wait_queue(task); + trace_rpc_task_wakeup(task->tk_client, task, queue); + + __rpc_remove_wait_queue(queue, task); rpc_make_runnable(task); - dprintk("RPC: __rpc_wake_up_task done\n"); + dprintk("RPC: __rpc_wake_up_task done\n"); } /* - * Wake up the specified task + * Wake up a queued task while the queue lock is being held */ -static void __rpc_wake_up_task(struct rpc_task *task) +static void rpc_wake_up_task_queue_locked(struct rpc_wait_queue *queue, struct rpc_task *task) { - if (rpc_start_wakeup(task)) { - if (RPC_IS_QUEUED(task)) - __rpc_do_wake_up_task(task); - rpc_finish_wakeup(task); + if (RPC_IS_QUEUED(task)) { + smp_rmb(); + if (task->tk_waitqueue == queue) + __rpc_do_wake_up_task(queue, task); } } /* - * Default timeout handler if none specified by user - */ -static void -__rpc_default_timer(struct rpc_task *task) -{ - dprintk("RPC: %d timeout (default timer)\n", task->tk_pid); - task->tk_status = -ETIMEDOUT; - rpc_wake_up_task(task); -} - -/* - * Wake up the specified task + * Wake up a task on a specific queue */ -void rpc_wake_up_task(struct rpc_task *task) +void rpc_wake_up_queued_task(struct rpc_wait_queue *queue, struct rpc_task *task) { - if (rpc_start_wakeup(task)) { - if (RPC_IS_QUEUED(task)) { - struct rpc_wait_queue *queue = task->u.tk_wait.rpc_waitq; - - spin_lock_bh(&queue->lock); - __rpc_do_wake_up_task(task); - spin_unlock_bh(&queue->lock); - } - rpc_finish_wakeup(task); - } + spin_lock_bh(&queue->lock); + rpc_wake_up_task_queue_locked(queue, task); + spin_unlock_bh(&queue->lock); } +EXPORT_SYMBOL_GPL(rpc_wake_up_queued_task); /* * Wake up the next task on a priority queue. */ -static struct rpc_task * __rpc_wake_up_next_priority(struct rpc_wait_queue *queue) +static struct rpc_task *__rpc_find_next_queued_priority(struct rpc_wait_queue *queue) { struct list_head *q; struct rpc_task *task; /* - * Service a batch of tasks from a single cookie. + * Service a batch of tasks from a single owner. */ q = &queue->tasks[queue->priority]; if (!list_empty(q)) { task = list_entry(q->next, struct rpc_task, u.tk_wait.list); - if (queue->cookie == task->tk_cookie) { + if (queue->owner == task->tk_owner) { if (--queue->nr) goto out; list_move_tail(&task->u.tk_wait.list, q); @@ -434,8 +478,7 @@ static struct rpc_task * __rpc_wake_up_next_priority(struct rpc_wait_queue *queu /* * Check if we need to switch queues. */ - if (--queue->count) - goto new_cookie; + goto new_owner; } /* @@ -457,32 +500,58 @@ static struct rpc_task * __rpc_wake_up_next_priority(struct rpc_wait_queue *queu new_queue: rpc_set_waitqueue_priority(queue, (unsigned int)(q - &queue->tasks[0])); -new_cookie: - rpc_set_waitqueue_cookie(queue, task->tk_cookie); +new_owner: + rpc_set_waitqueue_owner(queue, task->tk_owner); out: - __rpc_wake_up_task(task); return task; } +static struct rpc_task *__rpc_find_next_queued(struct rpc_wait_queue *queue) +{ + if (RPC_IS_PRIORITY(queue)) + return __rpc_find_next_queued_priority(queue); + if (!list_empty(&queue->tasks[0])) + return list_first_entry(&queue->tasks[0], struct rpc_task, u.tk_wait.list); + return NULL; +} + /* - * Wake up the next task on the wait queue. + * Wake up the first task on the wait queue. */ -struct rpc_task * rpc_wake_up_next(struct rpc_wait_queue *queue) +struct rpc_task *rpc_wake_up_first(struct rpc_wait_queue *queue, + bool (*func)(struct rpc_task *, void *), void *data) { struct rpc_task *task = NULL; - dprintk("RPC: wake_up_next(%p \"%s\")\n", queue, rpc_qname(queue)); + dprintk("RPC: wake_up_first(%p \"%s\")\n", + queue, rpc_qname(queue)); spin_lock_bh(&queue->lock); - if (RPC_IS_PRIORITY(queue)) - task = __rpc_wake_up_next_priority(queue); - else { - task_for_first(task, &queue->tasks[0]) - __rpc_wake_up_task(task); + task = __rpc_find_next_queued(queue); + if (task != NULL) { + if (func(task, data)) + rpc_wake_up_task_queue_locked(queue, task); + else + task = NULL; } spin_unlock_bh(&queue->lock); return task; } +EXPORT_SYMBOL_GPL(rpc_wake_up_first); + +static bool rpc_wake_up_next_func(struct rpc_task *task, void *data) +{ + return true; +} + +/* + * Wake up the next task on the wait queue. +*/ +struct rpc_task *rpc_wake_up_next(struct rpc_wait_queue *queue) +{ + return rpc_wake_up_first(queue, rpc_wake_up_next_func, NULL); +} +EXPORT_SYMBOL_GPL(rpc_wake_up_next); /** * rpc_wake_up - wake up all rpc_tasks @@ -492,15 +561,17 @@ struct rpc_task * rpc_wake_up_next(struct rpc_wait_queue *queue) */ void rpc_wake_up(struct rpc_wait_queue *queue) { - struct rpc_task *task; - struct list_head *head; + spin_lock_bh(&queue->lock); head = &queue->tasks[queue->maxpriority]; for (;;) { while (!list_empty(head)) { - task = list_entry(head->next, struct rpc_task, u.tk_wait.list); - __rpc_wake_up_task(task); + struct rpc_task *task; + task = list_first_entry(head, + struct rpc_task, + u.tk_wait.list); + rpc_wake_up_task_queue_locked(queue, task); } if (head == &queue->tasks[0]) break; @@ -508,6 +579,7 @@ void rpc_wake_up(struct rpc_wait_queue *queue) } spin_unlock_bh(&queue->lock); } +EXPORT_SYMBOL_GPL(rpc_wake_up); /** * rpc_wake_up_status - wake up all rpc_tasks and set their status value. @@ -519,15 +591,17 @@ void rpc_wake_up(struct rpc_wait_queue *queue) void rpc_wake_up_status(struct rpc_wait_queue *queue, int status) { struct list_head *head; - struct rpc_task *task; spin_lock_bh(&queue->lock); head = &queue->tasks[queue->maxpriority]; for (;;) { while (!list_empty(head)) { - task = list_entry(head->next, struct rpc_task, u.tk_wait.list); + struct rpc_task *task; + task = list_first_entry(head, + struct rpc_task, + u.tk_wait.list); task->tk_status = status; - __rpc_wake_up_task(task); + rpc_wake_up_task_queue_locked(queue, task); } if (head == &queue->tasks[0]) break; @@ -535,131 +609,177 @@ void rpc_wake_up_status(struct rpc_wait_queue *queue, int status) } spin_unlock_bh(&queue->lock); } +EXPORT_SYMBOL_GPL(rpc_wake_up_status); + +static void __rpc_queue_timer_fn(unsigned long ptr) +{ + struct rpc_wait_queue *queue = (struct rpc_wait_queue *)ptr; + struct rpc_task *task, *n; + unsigned long expires, now, timeo; + + spin_lock(&queue->lock); + expires = now = jiffies; + list_for_each_entry_safe(task, n, &queue->timer_list.list, u.tk_wait.timer_list) { + timeo = task->u.tk_wait.expires; + if (time_after_eq(now, timeo)) { + dprintk("RPC: %5u timeout\n", task->tk_pid); + task->tk_status = -ETIMEDOUT; + rpc_wake_up_task_queue_locked(queue, task); + continue; + } + if (expires == now || time_after(expires, timeo)) + expires = timeo; + } + if (!list_empty(&queue->timer_list.list)) + rpc_set_queue_timer(queue, expires); + spin_unlock(&queue->lock); +} + +static void __rpc_atrun(struct rpc_task *task) +{ + if (task->tk_status == -ETIMEDOUT) + task->tk_status = 0; +} /* * Run a task at a later time */ -static void __rpc_atrun(struct rpc_task *); -void -rpc_delay(struct rpc_task *task, unsigned long delay) +void rpc_delay(struct rpc_task *task, unsigned long delay) { task->tk_timeout = delay; - rpc_sleep_on(&delay_queue, task, NULL, __rpc_atrun); + rpc_sleep_on(&delay_queue, task, __rpc_atrun); +} +EXPORT_SYMBOL_GPL(rpc_delay); + +/* + * Helper to call task->tk_ops->rpc_call_prepare + */ +void rpc_prepare_task(struct rpc_task *task) +{ + task->tk_ops->rpc_call_prepare(task, task->tk_calldata); } static void -__rpc_atrun(struct rpc_task *task) +rpc_init_task_statistics(struct rpc_task *task) { - task->tk_status = 0; - rpc_wake_up_task(task); + /* Initialize retry counters */ + task->tk_garb_retry = 2; + task->tk_cred_retry = 2; + task->tk_rebind_retry = 2; + + /* starting timestamp */ + task->tk_start = ktime_get(); +} + +static void +rpc_reset_task_statistics(struct rpc_task *task) +{ + task->tk_timeouts = 0; + task->tk_flags &= ~(RPC_CALL_MAJORSEEN|RPC_TASK_KILLED|RPC_TASK_SENT); + + rpc_init_task_statistics(task); } /* - * Helper that calls task->tk_exit if it exists and then returns - * true if we should exit __rpc_execute. + * Helper that calls task->tk_ops->rpc_call_done if it exists */ -static inline int __rpc_do_exit(struct rpc_task *task) +void rpc_exit_task(struct rpc_task *task) { - if (task->tk_exit != NULL) { - lock_kernel(); - task->tk_exit(task); - unlock_kernel(); - /* If tk_action is non-null, we should restart the call */ + task->tk_action = NULL; + if (task->tk_ops->rpc_call_done != NULL) { + task->tk_ops->rpc_call_done(task, task->tk_calldata); if (task->tk_action != NULL) { - if (!RPC_ASSASSINATED(task)) { - /* Release RPC slot and buffer memory */ - xprt_release(task); - rpc_free(task); - return 0; - } - printk(KERN_ERR "RPC: dead task tried to walk away.\n"); + WARN_ON(RPC_ASSASSINATED(task)); + /* Always release the RPC slot and buffer memory */ + xprt_release(task); + rpc_reset_task_statistics(task); } } - return 1; } -static int rpc_wait_bit_interruptible(void *word) +void rpc_exit(struct rpc_task *task, int status) { - if (signal_pending(current)) - return -ERESTARTSYS; - schedule(); - return 0; + task->tk_status = status; + task->tk_action = rpc_exit_task; + if (RPC_IS_QUEUED(task)) + rpc_wake_up_queued_task(task->tk_waitqueue, task); +} +EXPORT_SYMBOL_GPL(rpc_exit); + +void rpc_release_calldata(const struct rpc_call_ops *ops, void *calldata) +{ + if (ops->rpc_release != NULL) + ops->rpc_release(calldata); } /* * This is the RPC `scheduler' (or rather, the finite state machine). */ -static int __rpc_execute(struct rpc_task *task) +static void __rpc_execute(struct rpc_task *task) { - int status = 0; + struct rpc_wait_queue *queue; + int task_is_async = RPC_IS_ASYNC(task); + int status = 0; - dprintk("RPC: %4d rpc_execute flgs %x\n", - task->tk_pid, task->tk_flags); + dprintk("RPC: %5u __rpc_execute flags=0x%x\n", + task->tk_pid, task->tk_flags); - BUG_ON(RPC_IS_QUEUED(task)); + WARN_ON_ONCE(RPC_IS_QUEUED(task)); + if (RPC_IS_QUEUED(task)) + return; for (;;) { - /* - * Garbage collection of pending timers... - */ - rpc_delete_timer(task); + void (*do_action)(struct rpc_task *); /* - * Execute any pending callback. + * Execute any pending callback first. */ - if (RPC_DO_CALLBACK(task)) { - /* Define a callback save pointer */ - void (*save_callback)(struct rpc_task *); - - /* - * If a callback exists, save it, reset it, - * call it. - * The save is needed to stop from resetting - * another callback set within the callback handler - * - Dave + do_action = task->tk_callback; + task->tk_callback = NULL; + if (do_action == NULL) { + /* + * Perform the next FSM step. + * tk_action may be NULL if the task has been killed. + * In particular, note that rpc_killall_tasks may + * do this at any time, so beware when dereferencing. */ - save_callback=task->tk_callback; - task->tk_callback=NULL; - lock_kernel(); - save_callback(task); - unlock_kernel(); - } - - /* - * Perform the next FSM step. - * tk_action may be NULL when the task has been killed - * by someone else. - */ - if (!RPC_IS_QUEUED(task)) { - if (task->tk_action != NULL) { - lock_kernel(); - task->tk_action(task); - unlock_kernel(); - } else if (__rpc_do_exit(task)) + do_action = task->tk_action; + if (do_action == NULL) break; } + trace_rpc_task_run_action(task->tk_client, task, task->tk_action); + do_action(task); /* * Lockless check for whether task is sleeping or not. */ if (!RPC_IS_QUEUED(task)) continue; - rpc_clear_running(task); - if (RPC_IS_ASYNC(task)) { - /* Careful! we may have raced... */ - if (RPC_IS_QUEUED(task)) - return 0; - if (rpc_test_and_set_running(task)) - return 0; + /* + * The queue->lock protects against races with + * rpc_make_runnable(). + * + * Note that once we clear RPC_TASK_RUNNING on an asynchronous + * rpc_task, rpc_make_runnable() can assign it to a + * different workqueue. We therefore cannot assume that the + * rpc_task pointer may still be dereferenced. + */ + queue = task->tk_waitqueue; + spin_lock_bh(&queue->lock); + if (!RPC_IS_QUEUED(task)) { + spin_unlock_bh(&queue->lock); continue; } + rpc_clear_running(task); + spin_unlock_bh(&queue->lock); + if (task_is_async) + return; /* sync task: sleep here */ - dprintk("RPC: %4d sync task going to sleep\n", task->tk_pid); - /* Note: Caller should be using rpc_clnt_sigmask() */ + dprintk("RPC: %5u sync task going to sleep\n", task->tk_pid); status = out_of_line_wait_on_bit(&task->tk_runstate, - RPC_TASK_QUEUED, rpc_wait_bit_interruptible, - TASK_INTERRUPTIBLE); + RPC_TASK_QUEUED, rpc_wait_bit_killable, + TASK_KILLABLE); if (status == -ERESTARTSYS) { /* * When a sync task receives a signal, it exits with @@ -667,21 +787,17 @@ static int __rpc_execute(struct rpc_task *task) * clean up after sleeping on some queue, we don't * break the loop here, but go around once more. */ - dprintk("RPC: %4d got signal\n", task->tk_pid); + dprintk("RPC: %5u got signal\n", task->tk_pid); task->tk_flags |= RPC_TASK_KILLED; rpc_exit(task, -ERESTARTSYS); - rpc_wake_up_task(task); } - rpc_set_running(task); - dprintk("RPC: %4d sync task resuming\n", task->tk_pid); + dprintk("RPC: %5u sync task resuming\n", task->tk_pid); } - dprintk("RPC: %4d exit() = %d\n", task->tk_pid, task->tk_status); - status = task->tk_status; - + dprintk("RPC: %5u return %d, status %d\n", task->tk_pid, status, + task->tk_status); /* Release all resources associated with the task */ rpc_release_task(task); - return status; } /* @@ -693,435 +809,330 @@ static int __rpc_execute(struct rpc_task *task) * released. In particular note that tk_release() will have * been called, so your task memory may have been freed. */ -int -rpc_execute(struct rpc_task *task) +void rpc_execute(struct rpc_task *task) { - BUG_ON(task->tk_active); + bool is_async = RPC_IS_ASYNC(task); - task->tk_active = 1; - rpc_set_running(task); - return __rpc_execute(task); + rpc_set_active(task); + rpc_make_runnable(task); + if (!is_async) + __rpc_execute(task); } -static void rpc_async_schedule(void *arg) +static void rpc_async_schedule(struct work_struct *work) { - __rpc_execute((struct rpc_task *)arg); + current->flags |= PF_FSTRANS; + __rpc_execute(container_of(work, struct rpc_task, u.tk_work)); + current->flags &= ~PF_FSTRANS; } -/* - * Allocate memory for RPC purposes. +/** + * rpc_malloc - allocate an RPC buffer + * @task: RPC task that will use this buffer + * @size: requested byte size + * + * To prevent rpciod from hanging, this allocator never sleeps, + * returning NULL and suppressing warning if the request cannot be serviced + * immediately. + * The caller can arrange to sleep in a way that is safe for rpciod. + * + * Most requests are 'small' (under 2KiB) and can be serviced from a + * mempool, ensuring that NFS reads and writes can always proceed, + * and that there is good locality of reference for these buffers. * - * We try to ensure that some NFS reads and writes can always proceed - * by using a mempool when allocating 'small' buffers. * In order to avoid memory starvation triggering more writebacks of - * NFS requests, we use GFP_NOFS rather than GFP_KERNEL. + * NFS requests, we avoid using GFP_KERNEL. */ -void * -rpc_malloc(struct rpc_task *task, size_t size) +void *rpc_malloc(struct rpc_task *task, size_t size) { - gfp_t gfp; + struct rpc_buffer *buf; + gfp_t gfp = GFP_NOWAIT | __GFP_NOWARN; - if (task->tk_flags & RPC_TASK_SWAPPER) - gfp = GFP_ATOMIC; + if (RPC_IS_SWAPPER(task)) + gfp |= __GFP_MEMALLOC; + + size += sizeof(struct rpc_buffer); + if (size <= RPC_BUFFER_MAXSIZE) + buf = mempool_alloc(rpc_buffer_mempool, gfp); else - gfp = GFP_NOFS; + buf = kmalloc(size, gfp); - if (size > RPC_BUFFER_MAXSIZE) { - task->tk_buffer = kmalloc(size, gfp); - if (task->tk_buffer) - task->tk_bufsize = size; - } else { - task->tk_buffer = mempool_alloc(rpc_buffer_mempool, gfp); - if (task->tk_buffer) - task->tk_bufsize = RPC_BUFFER_MAXSIZE; - } - return task->tk_buffer; + if (!buf) + return NULL; + + buf->len = size; + dprintk("RPC: %5u allocated buffer of size %zu at %p\n", + task->tk_pid, size, buf); + return &buf->data; } +EXPORT_SYMBOL_GPL(rpc_malloc); -static void -rpc_free(struct rpc_task *task) +/** + * rpc_free - free buffer allocated via rpc_malloc + * @buffer: buffer to free + * + */ +void rpc_free(void *buffer) { - if (task->tk_buffer) { - if (task->tk_bufsize == RPC_BUFFER_MAXSIZE) - mempool_free(task->tk_buffer, rpc_buffer_mempool); - else - kfree(task->tk_buffer); - task->tk_buffer = NULL; - task->tk_bufsize = 0; - } + size_t size; + struct rpc_buffer *buf; + + if (!buffer) + return; + + buf = container_of(buffer, struct rpc_buffer, data); + size = buf->len; + + dprintk("RPC: freeing buffer of size %zu at %p\n", + size, buf); + + if (size <= RPC_BUFFER_MAXSIZE) + mempool_free(buf, rpc_buffer_mempool); + else + kfree(buf); } +EXPORT_SYMBOL_GPL(rpc_free); /* * Creation and deletion of RPC task structures */ -void rpc_init_task(struct rpc_task *task, struct rpc_clnt *clnt, rpc_action callback, int flags) +static void rpc_init_task(struct rpc_task *task, const struct rpc_task_setup *task_setup_data) { memset(task, 0, sizeof(*task)); - init_timer(&task->tk_timer); - task->tk_timer.data = (unsigned long) task; - task->tk_timer.function = (void (*)(unsigned long)) rpc_run_timer; - task->tk_client = clnt; - task->tk_flags = flags; - task->tk_exit = callback; - - /* Initialize retry counters */ - task->tk_garb_retry = 2; - task->tk_cred_retry = 2; + atomic_set(&task->tk_count, 1); + task->tk_flags = task_setup_data->flags; + task->tk_ops = task_setup_data->callback_ops; + task->tk_calldata = task_setup_data->callback_data; + INIT_LIST_HEAD(&task->tk_task); - task->tk_priority = RPC_PRIORITY_NORMAL; - task->tk_cookie = (unsigned long)current; + task->tk_priority = task_setup_data->priority - RPC_PRIORITY_LOW; + task->tk_owner = current->tgid; /* Initialize workqueue for async tasks */ - task->tk_workqueue = rpciod_workqueue; - - if (clnt) { - atomic_inc(&clnt->cl_users); - if (clnt->cl_softrtry) - task->tk_flags |= RPC_TASK_SOFT; - if (!clnt->cl_intr) - task->tk_flags |= RPC_TASK_NOINTR; - } + task->tk_workqueue = task_setup_data->workqueue; -#ifdef RPC_DEBUG - task->tk_magic = RPC_TASK_MAGIC_ID; - task->tk_pid = rpc_task_id++; -#endif - /* Add to global list of all tasks */ - spin_lock(&rpc_sched_lock); - list_add_tail(&task->tk_task, &all_tasks); - spin_unlock(&rpc_sched_lock); + if (task->tk_ops->rpc_call_prepare != NULL) + task->tk_action = rpc_prepare_task; + + rpc_init_task_statistics(task); - dprintk("RPC: %4d new task procpid %d\n", task->tk_pid, - current->pid); + dprintk("RPC: new task initialized, procpid %u\n", + task_pid_nr(current)); } static struct rpc_task * rpc_alloc_task(void) { - return (struct rpc_task *)mempool_alloc(rpc_task_mempool, GFP_NOFS); + return (struct rpc_task *)mempool_alloc(rpc_task_mempool, GFP_NOIO); } -static void -rpc_default_free_task(struct rpc_task *task) +/* + * Create a new task for the specified client. + */ +struct rpc_task *rpc_new_task(const struct rpc_task_setup *setup_data) { - dprintk("RPC: %4d freeing task\n", task->tk_pid); - mempool_free(task, rpc_task_mempool); + struct rpc_task *task = setup_data->task; + unsigned short flags = 0; + + if (task == NULL) { + task = rpc_alloc_task(); + if (task == NULL) { + rpc_release_calldata(setup_data->callback_ops, + setup_data->callback_data); + return ERR_PTR(-ENOMEM); + } + flags = RPC_TASK_DYNAMIC; + } + + rpc_init_task(task, setup_data); + task->tk_flags |= flags; + dprintk("RPC: allocated task %p\n", task); + return task; } /* - * Create a new task for the specified client. We have to - * clean up after an allocation failure, as the client may - * have specified "oneshot". + * rpc_free_task - release rpc task and perform cleanups + * + * Note that we free up the rpc_task _after_ rpc_release_calldata() + * in order to work around a workqueue dependency issue. + * + * Tejun Heo states: + * "Workqueue currently considers two work items to be the same if they're + * on the same address and won't execute them concurrently - ie. it + * makes a work item which is queued again while being executed wait + * for the previous execution to complete. + * + * If a work function frees the work item, and then waits for an event + * which should be performed by another work item and *that* work item + * recycles the freed work item, it can create a false dependency loop. + * There really is no reliable way to detect this short of verifying + * every memory free." + * */ -struct rpc_task * -rpc_new_task(struct rpc_clnt *clnt, rpc_action callback, int flags) +static void rpc_free_task(struct rpc_task *task) { - struct rpc_task *task; - - task = rpc_alloc_task(); - if (!task) - goto cleanup; - - rpc_init_task(task, clnt, callback, flags); + unsigned short tk_flags = task->tk_flags; - /* Replace tk_release */ - task->tk_release = rpc_default_free_task; - - dprintk("RPC: %4d allocated task\n", task->tk_pid); - task->tk_flags |= RPC_TASK_DYNAMIC; -out: - return task; + rpc_release_calldata(task->tk_ops, task->tk_calldata); -cleanup: - /* Check whether to release the client */ - if (clnt) { - printk("rpc_new_task: failed, users=%d, oneshot=%d\n", - atomic_read(&clnt->cl_users), clnt->cl_oneshot); - atomic_inc(&clnt->cl_users); /* pretend we were used ... */ - rpc_release_client(clnt); + if (tk_flags & RPC_TASK_DYNAMIC) { + dprintk("RPC: %5u freeing task\n", task->tk_pid); + mempool_free(task, rpc_task_mempool); } - goto out; } -void rpc_release_task(struct rpc_task *task) +static void rpc_async_release(struct work_struct *work) { - dprintk("RPC: %4d release task\n", task->tk_pid); - -#ifdef RPC_DEBUG - BUG_ON(task->tk_magic != RPC_TASK_MAGIC_ID); -#endif + rpc_free_task(container_of(work, struct rpc_task, u.tk_work)); +} - /* Remove from global task list */ - spin_lock(&rpc_sched_lock); - list_del(&task->tk_task); - spin_unlock(&rpc_sched_lock); - - BUG_ON (RPC_IS_QUEUED(task)); - task->tk_active = 0; - - /* Synchronously delete any running timer */ - rpc_delete_timer(task); - - /* Release resources */ - if (task->tk_rqstp) - xprt_release(task); - if (task->tk_msg.rpc_cred) - rpcauth_unbindcred(task); - rpc_free(task); - if (task->tk_client) { - rpc_release_client(task->tk_client); - task->tk_client = NULL; +static void rpc_release_resources_task(struct rpc_task *task) +{ + xprt_release(task); + if (task->tk_msg.rpc_cred) { + put_rpccred(task->tk_msg.rpc_cred); + task->tk_msg.rpc_cred = NULL; } - -#ifdef RPC_DEBUG - task->tk_magic = 0; -#endif - if (task->tk_release) - task->tk_release(task); + rpc_task_release_client(task); } -/** - * rpc_find_parent - find the parent of a child task. - * @child: child task - * - * Checks that the parent task is still sleeping on the - * queue 'childq'. If so returns a pointer to the parent. - * Upon failure returns NULL. - * - * Caller must hold childq.lock - */ -static inline struct rpc_task *rpc_find_parent(struct rpc_task *child) +static void rpc_final_put_task(struct rpc_task *task, + struct workqueue_struct *q) { - struct rpc_task *task, *parent; - struct list_head *le; - - parent = (struct rpc_task *) child->tk_calldata; - task_for_each(task, le, &childq.tasks[0]) - if (task == parent) - return parent; - - return NULL; + if (q != NULL) { + INIT_WORK(&task->u.tk_work, rpc_async_release); + queue_work(q, &task->u.tk_work); + } else + rpc_free_task(task); } -static void rpc_child_exit(struct rpc_task *child) +static void rpc_do_put_task(struct rpc_task *task, struct workqueue_struct *q) { - struct rpc_task *parent; - - spin_lock_bh(&childq.lock); - if ((parent = rpc_find_parent(child)) != NULL) { - parent->tk_status = child->tk_status; - __rpc_wake_up_task(parent); + if (atomic_dec_and_test(&task->tk_count)) { + rpc_release_resources_task(task); + rpc_final_put_task(task, q); } - spin_unlock_bh(&childq.lock); } -/* - * Note: rpc_new_task releases the client after a failure. - */ -struct rpc_task * -rpc_new_child(struct rpc_clnt *clnt, struct rpc_task *parent) +void rpc_put_task(struct rpc_task *task) { - struct rpc_task *task; - - task = rpc_new_task(clnt, NULL, RPC_TASK_ASYNC | RPC_TASK_CHILD); - if (!task) - goto fail; - task->tk_exit = rpc_child_exit; - task->tk_calldata = parent; - return task; - -fail: - parent->tk_status = -ENOMEM; - return NULL; + rpc_do_put_task(task, NULL); } +EXPORT_SYMBOL_GPL(rpc_put_task); -void rpc_run_child(struct rpc_task *task, struct rpc_task *child, rpc_action func) +void rpc_put_task_async(struct rpc_task *task) { - spin_lock_bh(&childq.lock); - /* N.B. Is it possible for the child to have already finished? */ - __rpc_sleep_on(&childq, task, func, NULL); - rpc_schedule_run(child); - spin_unlock_bh(&childq.lock); + rpc_do_put_task(task, task->tk_workqueue); } +EXPORT_SYMBOL_GPL(rpc_put_task_async); -/* - * Kill all tasks for the given client. - * XXX: kill their descendants as well? - */ -void rpc_killall_tasks(struct rpc_clnt *clnt) +static void rpc_release_task(struct rpc_task *task) { - struct rpc_task *rovr; - struct list_head *le; + dprintk("RPC: %5u release task\n", task->tk_pid); + + WARN_ON_ONCE(RPC_IS_QUEUED(task)); - dprintk("RPC: killing all tasks for client %p\n", clnt); + rpc_release_resources_task(task); /* - * Spin lock all_tasks to prevent changes... + * Note: at this point we have been removed from rpc_clnt->cl_tasks, + * so it should be safe to use task->tk_count as a test for whether + * or not any other processes still hold references to our rpc_task. */ - spin_lock(&rpc_sched_lock); - alltask_for_each(rovr, le, &all_tasks) { - if (! RPC_IS_ACTIVATED(rovr)) - continue; - if (!clnt || rovr->tk_client == clnt) { - rovr->tk_flags |= RPC_TASK_KILLED; - rpc_exit(rovr, -EIO); - rpc_wake_up_task(rovr); - } + if (atomic_read(&task->tk_count) != 1 + !RPC_IS_ASYNC(task)) { + /* Wake up anyone who may be waiting for task completion */ + if (!rpc_complete_task(task)) + return; + } else { + if (!atomic_dec_and_test(&task->tk_count)) + return; } - spin_unlock(&rpc_sched_lock); + rpc_final_put_task(task, task->tk_workqueue); } -static DECLARE_MUTEX_LOCKED(rpciod_running); - -static void rpciod_killall(void) +int rpciod_up(void) { - unsigned long flags; - - while (!list_empty(&all_tasks)) { - clear_thread_flag(TIF_SIGPENDING); - rpc_killall_tasks(NULL); - flush_workqueue(rpciod_workqueue); - if (!list_empty(&all_tasks)) { - dprintk("rpciod_killall: waiting for tasks to exit\n"); - yield(); - } - } + return try_module_get(THIS_MODULE) ? 0 : -EINVAL; +} - spin_lock_irqsave(¤t->sighand->siglock, flags); - recalc_sigpending(); - spin_unlock_irqrestore(¤t->sighand->siglock, flags); +void rpciod_down(void) +{ + module_put(THIS_MODULE); } /* - * Start up the rpciod process if it's not already running. + * Start up the rpciod workqueue. */ -int -rpciod_up(void) +static int rpciod_start(void) { struct workqueue_struct *wq; - int error = 0; - down(&rpciod_sema); - dprintk("rpciod_up: users %d\n", rpciod_users); - rpciod_users++; - if (rpciod_workqueue) - goto out; - /* - * If there's no pid, we should be the first user. - */ - if (rpciod_users > 1) - printk(KERN_WARNING "rpciod_up: no workqueue, %d users??\n", rpciod_users); /* * Create the rpciod thread and wait for it to start. */ - error = -ENOMEM; - wq = create_workqueue("rpciod"); - if (wq == NULL) { - printk(KERN_WARNING "rpciod_up: create workqueue failed, error=%d\n", error); - rpciod_users--; - goto out; - } + dprintk("RPC: creating workqueue rpciod\n"); + wq = alloc_workqueue("rpciod", WQ_MEM_RECLAIM, 1); rpciod_workqueue = wq; - error = 0; -out: - up(&rpciod_sema); - return error; + return rpciod_workqueue != NULL; } -void -rpciod_down(void) +static void rpciod_stop(void) { - down(&rpciod_sema); - dprintk("rpciod_down sema %d\n", rpciod_users); - if (rpciod_users) { - if (--rpciod_users) - goto out; - } else - printk(KERN_WARNING "rpciod_down: no users??\n"); + struct workqueue_struct *wq = NULL; - if (!rpciod_workqueue) { - dprintk("rpciod_down: Nothing to do!\n"); - goto out; - } - rpciod_killall(); + if (rpciod_workqueue == NULL) + return; + dprintk("RPC: destroying workqueue rpciod\n"); - destroy_workqueue(rpciod_workqueue); + wq = rpciod_workqueue; rpciod_workqueue = NULL; - out: - up(&rpciod_sema); + destroy_workqueue(wq); } -#ifdef RPC_DEBUG -void rpc_show_tasks(void) -{ - struct list_head *le; - struct rpc_task *t; - - spin_lock(&rpc_sched_lock); - if (list_empty(&all_tasks)) { - spin_unlock(&rpc_sched_lock); - return; - } - printk("-pid- proc flgs status -client- -prog- --rqstp- -timeout " - "-rpcwait -action- --exit--\n"); - alltask_for_each(t, le, &all_tasks) { - const char *rpc_waitq = "none"; - - if (RPC_IS_QUEUED(t)) - rpc_waitq = rpc_qname(t->u.tk_wait.rpc_waitq); - - printk("%05d %04d %04x %06d %8p %6d %8p %08ld %8s %8p %8p\n", - t->tk_pid, - (t->tk_msg.rpc_proc ? t->tk_msg.rpc_proc->p_proc : -1), - t->tk_flags, t->tk_status, - t->tk_client, - (t->tk_client ? t->tk_client->cl_prog : 0), - t->tk_rqstp, t->tk_timeout, - rpc_waitq, - t->tk_action, t->tk_exit); - } - spin_unlock(&rpc_sched_lock); -} -#endif - void rpc_destroy_mempool(void) { + rpciod_stop(); if (rpc_buffer_mempool) mempool_destroy(rpc_buffer_mempool); if (rpc_task_mempool) mempool_destroy(rpc_task_mempool); - if (rpc_task_slabp && kmem_cache_destroy(rpc_task_slabp)) - printk(KERN_INFO "rpc_task: not all structures were freed\n"); - if (rpc_buffer_slabp && kmem_cache_destroy(rpc_buffer_slabp)) - printk(KERN_INFO "rpc_buffers: not all structures were freed\n"); + if (rpc_task_slabp) + kmem_cache_destroy(rpc_task_slabp); + if (rpc_buffer_slabp) + kmem_cache_destroy(rpc_buffer_slabp); + rpc_destroy_wait_queue(&delay_queue); } int rpc_init_mempool(void) { + /* + * The following is not strictly a mempool initialisation, + * but there is no harm in doing it here + */ + rpc_init_wait_queue(&delay_queue, "delayq"); + if (!rpciod_start()) + goto err_nomem; + rpc_task_slabp = kmem_cache_create("rpc_tasks", sizeof(struct rpc_task), 0, SLAB_HWCACHE_ALIGN, - NULL, NULL); + NULL); if (!rpc_task_slabp) goto err_nomem; rpc_buffer_slabp = kmem_cache_create("rpc_buffers", RPC_BUFFER_MAXSIZE, 0, SLAB_HWCACHE_ALIGN, - NULL, NULL); + NULL); if (!rpc_buffer_slabp) goto err_nomem; - rpc_task_mempool = mempool_create(RPC_TASK_POOLSIZE, - mempool_alloc_slab, - mempool_free_slab, - rpc_task_slabp); + rpc_task_mempool = mempool_create_slab_pool(RPC_TASK_POOLSIZE, + rpc_task_slabp); if (!rpc_task_mempool) goto err_nomem; - rpc_buffer_mempool = mempool_create(RPC_BUFFER_POOLSIZE, - mempool_alloc_slab, - mempool_free_slab, - rpc_buffer_slabp); + rpc_buffer_mempool = mempool_create_slab_pool(RPC_BUFFER_POOLSIZE, + rpc_buffer_slabp); if (!rpc_buffer_mempool) goto err_nomem; return 0; diff --git a/net/sunrpc/socklib.c b/net/sunrpc/socklib.c index 8f97e90f36c..2df87f78e51 100644 --- a/net/sunrpc/socklib.c +++ b/net/sunrpc/socklib.c @@ -6,14 +6,19 @@ * Copyright (C) 1995, 1996 Olaf Kirch <okir@monad.swb.de> */ +#include <linux/compiler.h> +#include <linux/netdevice.h> +#include <linux/gfp.h> +#include <linux/skbuff.h> #include <linux/types.h> #include <linux/pagemap.h> #include <linux/udp.h> #include <linux/sunrpc/xdr.h> +#include <linux/export.h> /** - * skb_read_bits - copy some data bits from skb to internal buffer + * xdr_skb_read_bits - copy some data bits from skb to internal buffer * @desc: sk_buff copy helper * @to: copy destination * @len: number of bytes to copy @@ -21,28 +26,30 @@ * Possibly called several times to iterate over an sk_buff and copy * data out of it. */ -static size_t skb_read_bits(skb_reader_t *desc, void *to, size_t len) +size_t xdr_skb_read_bits(struct xdr_skb_reader *desc, void *to, size_t len) { if (len > desc->count) len = desc->count; - if (skb_copy_bits(desc->skb, desc->offset, to, len)) + if (unlikely(skb_copy_bits(desc->skb, desc->offset, to, len))) return 0; desc->count -= len; desc->offset += len; return len; } +EXPORT_SYMBOL_GPL(xdr_skb_read_bits); /** - * skb_read_and_csum_bits - copy and checksum from skb to buffer + * xdr_skb_read_and_csum_bits - copy and checksum from skb to buffer * @desc: sk_buff copy helper * @to: copy destination * @len: number of bytes to copy * * Same as skb_read_bits, but calculate a checksum at the same time. */ -static size_t skb_read_and_csum_bits(skb_reader_t *desc, void *to, size_t len) +static size_t xdr_skb_read_and_csum_bits(struct xdr_skb_reader *desc, void *to, size_t len) { - unsigned int csum2, pos; + unsigned int pos; + __wsum csum2; if (len > desc->count) len = desc->count; @@ -62,12 +69,12 @@ static size_t skb_read_and_csum_bits(skb_reader_t *desc, void *to, size_t len) * @copy_actor: virtual method for copying data * */ -ssize_t xdr_partial_copy_from_skb(struct xdr_buf *xdr, unsigned int base, skb_reader_t *desc, skb_read_actor_t copy_actor) +ssize_t xdr_partial_copy_from_skb(struct xdr_buf *xdr, unsigned int base, struct xdr_skb_reader *desc, xdr_skb_read_actor copy_actor) { struct page **ppage = xdr->pages; unsigned int len, pglen = xdr->page_len; ssize_t copied = 0; - int ret; + size_t ret; len = xdr->head[0].iov_len; if (base < len) { @@ -107,7 +114,7 @@ ssize_t xdr_partial_copy_from_skb(struct xdr_buf *xdr, unsigned int base, skb_re } len = PAGE_CACHE_SIZE; - kaddr = kmap_atomic(*ppage, KM_SKB_SUNRPC_DATA); + kaddr = kmap_atomic(*ppage); if (base) { len -= base; if (pglen < len) @@ -120,7 +127,7 @@ ssize_t xdr_partial_copy_from_skb(struct xdr_buf *xdr, unsigned int base, skb_re ret = copy_actor(desc, kaddr, len); } flush_dcache_page(*ppage); - kunmap_atomic(kaddr, KM_SKB_SUNRPC_DATA); + kunmap_atomic(kaddr); copied += ret; if (ret != len || !desc->count) goto out; @@ -133,6 +140,7 @@ copy_tail: out: return copied; } +EXPORT_SYMBOL_GPL(xdr_partial_copy_from_skb); /** * csum_partial_copy_to_xdr - checksum and copy data @@ -144,32 +152,36 @@ out: */ int csum_partial_copy_to_xdr(struct xdr_buf *xdr, struct sk_buff *skb) { - skb_reader_t desc; + struct xdr_skb_reader desc; desc.skb = skb; desc.offset = sizeof(struct udphdr); desc.count = skb->len - desc.offset; - if (skb->ip_summed == CHECKSUM_UNNECESSARY) + if (skb_csum_unnecessary(skb)) goto no_checksum; desc.csum = csum_partial(skb->data, desc.offset, skb->csum); - if (xdr_partial_copy_from_skb(xdr, 0, &desc, skb_read_and_csum_bits) < 0) + if (xdr_partial_copy_from_skb(xdr, 0, &desc, xdr_skb_read_and_csum_bits) < 0) return -1; if (desc.offset != skb->len) { - unsigned int csum2; + __wsum csum2; csum2 = skb_checksum(skb, desc.offset, skb->len - desc.offset, 0); desc.csum = csum_block_add(desc.csum, csum2, desc.offset); } if (desc.count) return -1; - if ((unsigned short)csum_fold(desc.csum)) + if (csum_fold(desc.csum)) return -1; + if (unlikely(skb->ip_summed == CHECKSUM_COMPLETE) && + !skb->csum_complete_sw) + netdev_rx_csum_fault(skb->dev); return 0; no_checksum: - if (xdr_partial_copy_from_skb(xdr, 0, &desc, skb_read_bits) < 0) + if (xdr_partial_copy_from_skb(xdr, 0, &desc, xdr_skb_read_bits) < 0) return -1; if (desc.count) return -1; return 0; } +EXPORT_SYMBOL_GPL(csum_partial_copy_to_xdr); diff --git a/net/sunrpc/stats.c b/net/sunrpc/stats.c index 4979f226e28..54530490944 100644 --- a/net/sunrpc/stats.c +++ b/net/sunrpc/stats.c @@ -13,18 +13,20 @@ */ #include <linux/module.h> +#include <linux/slab.h> #include <linux/init.h> #include <linux/kernel.h> -#include <linux/sched.h> #include <linux/proc_fs.h> #include <linux/seq_file.h> #include <linux/sunrpc/clnt.h> #include <linux/sunrpc/svcsock.h> +#include <linux/sunrpc/metrics.h> +#include <linux/rcupdate.h> -#define RPCDBG_FACILITY RPCDBG_MISC +#include "netns.h" -struct proc_dir_entry *proc_net_rpc = NULL; +#define RPCDBG_FACILITY RPCDBG_MISC /* * Get RPC client stats @@ -32,7 +34,7 @@ struct proc_dir_entry *proc_net_rpc = NULL; static int rpc_proc_show(struct seq_file *seq, void *v) { const struct rpc_stat *statp = seq->private; const struct rpc_program *prog = statp->program; - int i, j; + unsigned int i, j; seq_printf(seq, "net %u %u %u %u\n", @@ -62,10 +64,10 @@ static int rpc_proc_show(struct seq_file *seq, void *v) { static int rpc_proc_open(struct inode *inode, struct file *file) { - return single_open(file, rpc_proc_show, PDE(inode)->data); + return single_open(file, rpc_proc_show, PDE_DATA(inode)); } -static struct file_operations rpc_proc_fops = { +static const struct file_operations rpc_proc_fops = { .owner = THIS_MODULE, .open = rpc_proc_open, .read = seq_read, @@ -80,7 +82,7 @@ void svc_seq_show(struct seq_file *seq, const struct svc_stat *statp) { const struct svc_program *prog = statp->program; const struct svc_procedure *proc; const struct svc_version *vers; - int i, j; + unsigned int i, j; seq_printf(seq, "net %u %u %u %u\n", @@ -105,71 +107,176 @@ void svc_seq_show(struct seq_file *seq, const struct svc_stat *statp) { seq_putc(seq, '\n'); } } +EXPORT_SYMBOL_GPL(svc_seq_show); + +/** + * rpc_alloc_iostats - allocate an rpc_iostats structure + * @clnt: RPC program, version, and xprt + * + */ +struct rpc_iostats *rpc_alloc_iostats(struct rpc_clnt *clnt) +{ + return kcalloc(clnt->cl_maxproc, sizeof(struct rpc_iostats), GFP_KERNEL); +} +EXPORT_SYMBOL_GPL(rpc_alloc_iostats); + +/** + * rpc_free_iostats - release an rpc_iostats structure + * @stats: doomed rpc_iostats structure + * + */ +void rpc_free_iostats(struct rpc_iostats *stats) +{ + kfree(stats); +} +EXPORT_SYMBOL_GPL(rpc_free_iostats); + +/** + * rpc_count_iostats - tally up per-task stats + * @task: completed rpc_task + * @stats: array of stat structures + * + * Relies on the caller for serialization. + */ +void rpc_count_iostats(const struct rpc_task *task, struct rpc_iostats *stats) +{ + struct rpc_rqst *req = task->tk_rqstp; + struct rpc_iostats *op_metrics; + ktime_t delta; + + if (!stats || !req) + return; + + op_metrics = &stats[task->tk_msg.rpc_proc->p_statidx]; + + op_metrics->om_ops++; + op_metrics->om_ntrans += req->rq_ntrans; + op_metrics->om_timeouts += task->tk_timeouts; + + op_metrics->om_bytes_sent += req->rq_xmit_bytes_sent; + op_metrics->om_bytes_recv += req->rq_reply_bytes_recvd; + + delta = ktime_sub(req->rq_xtime, task->tk_start); + op_metrics->om_queue = ktime_add(op_metrics->om_queue, delta); + + op_metrics->om_rtt = ktime_add(op_metrics->om_rtt, req->rq_rtt); + + delta = ktime_sub(ktime_get(), task->tk_start); + op_metrics->om_execute = ktime_add(op_metrics->om_execute, delta); +} +EXPORT_SYMBOL_GPL(rpc_count_iostats); + +static void _print_name(struct seq_file *seq, unsigned int op, + struct rpc_procinfo *procs) +{ + if (procs[op].p_name) + seq_printf(seq, "\t%12s: ", procs[op].p_name); + else if (op == 0) + seq_printf(seq, "\t NULL: "); + else + seq_printf(seq, "\t%12u: ", op); +} + +void rpc_print_iostats(struct seq_file *seq, struct rpc_clnt *clnt) +{ + struct rpc_iostats *stats = clnt->cl_metrics; + struct rpc_xprt *xprt; + unsigned int op, maxproc = clnt->cl_maxproc; + + if (!stats) + return; + + seq_printf(seq, "\tRPC iostats version: %s ", RPC_IOSTATS_VERS); + seq_printf(seq, "p/v: %u/%u (%s)\n", + clnt->cl_prog, clnt->cl_vers, clnt->cl_program->name); + + rcu_read_lock(); + xprt = rcu_dereference(clnt->cl_xprt); + if (xprt) + xprt->ops->print_stats(xprt, seq); + rcu_read_unlock(); + + seq_printf(seq, "\tper-op statistics\n"); + for (op = 0; op < maxproc; op++) { + struct rpc_iostats *metrics = &stats[op]; + _print_name(seq, op, clnt->cl_procinfo); + seq_printf(seq, "%lu %lu %lu %Lu %Lu %Lu %Lu %Lu\n", + metrics->om_ops, + metrics->om_ntrans, + metrics->om_timeouts, + metrics->om_bytes_sent, + metrics->om_bytes_recv, + ktime_to_ms(metrics->om_queue), + ktime_to_ms(metrics->om_rtt), + ktime_to_ms(metrics->om_execute)); + } +} +EXPORT_SYMBOL_GPL(rpc_print_iostats); /* * Register/unregister RPC proc files */ static inline struct proc_dir_entry * -do_register(const char *name, void *data, struct file_operations *fops) +do_register(struct net *net, const char *name, void *data, + const struct file_operations *fops) { - struct proc_dir_entry *ent; + struct sunrpc_net *sn; - rpc_proc_init(); - dprintk("RPC: registering /proc/net/rpc/%s\n", name); - - ent = create_proc_entry(name, 0, proc_net_rpc); - if (ent) { - ent->proc_fops = fops; - ent->data = data; - } - return ent; + dprintk("RPC: registering /proc/net/rpc/%s\n", name); + sn = net_generic(net, sunrpc_net_id); + return proc_create_data(name, 0, sn->proc_net_rpc, fops, data); } struct proc_dir_entry * -rpc_proc_register(struct rpc_stat *statp) +rpc_proc_register(struct net *net, struct rpc_stat *statp) { - return do_register(statp->program->name, statp, &rpc_proc_fops); + return do_register(net, statp->program->name, statp, &rpc_proc_fops); } +EXPORT_SYMBOL_GPL(rpc_proc_register); void -rpc_proc_unregister(const char *name) +rpc_proc_unregister(struct net *net, const char *name) { - remove_proc_entry(name, proc_net_rpc); + struct sunrpc_net *sn; + + sn = net_generic(net, sunrpc_net_id); + remove_proc_entry(name, sn->proc_net_rpc); } +EXPORT_SYMBOL_GPL(rpc_proc_unregister); struct proc_dir_entry * -svc_proc_register(struct svc_stat *statp, struct file_operations *fops) +svc_proc_register(struct net *net, struct svc_stat *statp, const struct file_operations *fops) { - return do_register(statp->program->pg_name, statp, fops); + return do_register(net, statp->program->pg_name, statp, fops); } +EXPORT_SYMBOL_GPL(svc_proc_register); void -svc_proc_unregister(const char *name) +svc_proc_unregister(struct net *net, const char *name) { - remove_proc_entry(name, proc_net_rpc); + struct sunrpc_net *sn; + + sn = net_generic(net, sunrpc_net_id); + remove_proc_entry(name, sn->proc_net_rpc); } +EXPORT_SYMBOL_GPL(svc_proc_unregister); -void -rpc_proc_init(void) +int rpc_proc_init(struct net *net) { - dprintk("RPC: registering /proc/net/rpc\n"); - if (!proc_net_rpc) { - struct proc_dir_entry *ent; - ent = proc_mkdir("rpc", proc_net); - if (ent) { - ent->owner = THIS_MODULE; - proc_net_rpc = ent; - } - } + struct sunrpc_net *sn; + + dprintk("RPC: registering /proc/net/rpc\n"); + sn = net_generic(net, sunrpc_net_id); + sn->proc_net_rpc = proc_mkdir("rpc", net->proc_net); + if (sn->proc_net_rpc == NULL) + return -ENOMEM; + + return 0; } -void -rpc_proc_exit(void) +void rpc_proc_exit(struct net *net) { - dprintk("RPC: unregistering /proc/net/rpc\n"); - if (proc_net_rpc) { - proc_net_rpc = NULL; - remove_proc_entry("net/rpc", NULL); - } + dprintk("RPC: unregistering /proc/net/rpc\n"); + remove_proc_entry("rpc", net->proc_net); } diff --git a/net/sunrpc/sunrpc.h b/net/sunrpc/sunrpc.h new file mode 100644 index 00000000000..f2b7cb540e6 --- /dev/null +++ b/net/sunrpc/sunrpc.h @@ -0,0 +1,66 @@ +/****************************************************************************** + +(c) 2008 NetApp. All Rights Reserved. + +NetApp provides this source code under the GPL v2 License. +The GPL v2 license is available at +http://opensource.org/licenses/gpl-license.php. + +THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS +"AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT +LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR +A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR +CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, +EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, +PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR +PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF +LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING +NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS +SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + +******************************************************************************/ + +/* + * Functions and macros used internally by RPC + */ + +#ifndef _NET_SUNRPC_SUNRPC_H +#define _NET_SUNRPC_SUNRPC_H + +#include <linux/net.h> + +/* + * Header for dynamically allocated rpc buffers. + */ +struct rpc_buffer { + size_t len; + char data[]; +}; + +static inline int rpc_reply_expected(struct rpc_task *task) +{ + return (task->tk_msg.rpc_proc != NULL) && + (task->tk_msg.rpc_proc->p_decode != NULL); +} + +static inline int sock_is_loopback(struct sock *sk) +{ + struct dst_entry *dst; + int loopback = 0; + rcu_read_lock(); + dst = rcu_dereference(sk->sk_dst_cache); + if (dst && dst->dev && + (dst->dev->features & NETIF_F_LOOPBACK)) + loopback = 1; + rcu_read_unlock(); + return loopback; +} + +int svc_send_common(struct socket *sock, struct xdr_buf *xdr, + struct page *headpage, unsigned long headoffset, + struct page *tailpage, unsigned long tailoffset); + +int rpc_clients_notifier_register(void); +void rpc_clients_notifier_unregister(void); +#endif /* _NET_SUNRPC_SUNRPC_H */ + diff --git a/net/sunrpc/sunrpc_syms.c b/net/sunrpc/sunrpc_syms.c index a03d4b600c9..cd30120de9e 100644 --- a/net/sunrpc/sunrpc_syms.c +++ b/net/sunrpc/sunrpc_syms.c @@ -6,11 +6,9 @@ * Copyright (C) 1997 Olaf Kirch <okir@monad.swb.de> */ -#include <linux/config.h> #include <linux/module.h> #include <linux/types.h> -#include <linux/sched.h> #include <linux/uio.h> #include <linux/unistd.h> #include <linux/init.h> @@ -22,148 +20,96 @@ #include <linux/sunrpc/auth.h> #include <linux/workqueue.h> #include <linux/sunrpc/rpc_pipe_fs.h> +#include <linux/sunrpc/xprtsock.h> +#include "netns.h" -/* RPC scheduler */ -EXPORT_SYMBOL(rpc_execute); -EXPORT_SYMBOL(rpc_init_task); -EXPORT_SYMBOL(rpc_sleep_on); -EXPORT_SYMBOL(rpc_wake_up_next); -EXPORT_SYMBOL(rpc_wake_up_task); -EXPORT_SYMBOL(rpc_new_child); -EXPORT_SYMBOL(rpc_run_child); -EXPORT_SYMBOL(rpciod_down); -EXPORT_SYMBOL(rpciod_up); -EXPORT_SYMBOL(rpc_new_task); -EXPORT_SYMBOL(rpc_wake_up_status); -EXPORT_SYMBOL(rpc_release_task); - -/* RPC client functions */ -EXPORT_SYMBOL(rpc_create_client); -EXPORT_SYMBOL(rpc_new_client); -EXPORT_SYMBOL(rpc_clone_client); -EXPORT_SYMBOL(rpc_bind_new_program); -EXPORT_SYMBOL(rpc_destroy_client); -EXPORT_SYMBOL(rpc_shutdown_client); -EXPORT_SYMBOL(rpc_release_client); -EXPORT_SYMBOL(rpc_killall_tasks); -EXPORT_SYMBOL(rpc_call_sync); -EXPORT_SYMBOL(rpc_call_async); -EXPORT_SYMBOL(rpc_call_setup); -EXPORT_SYMBOL(rpc_clnt_sigmask); -EXPORT_SYMBOL(rpc_clnt_sigunmask); -EXPORT_SYMBOL(rpc_delay); -EXPORT_SYMBOL(rpc_restart_call); -EXPORT_SYMBOL(rpc_setbufsize); -EXPORT_SYMBOL(rpc_unlink); -EXPORT_SYMBOL(rpc_wake_up); -EXPORT_SYMBOL(rpc_queue_upcall); -EXPORT_SYMBOL(rpc_mkpipe); - -/* Client transport */ -EXPORT_SYMBOL(xprt_create_proto); -EXPORT_SYMBOL(xprt_set_timeout); - -/* Client credential cache */ -EXPORT_SYMBOL(rpcauth_register); -EXPORT_SYMBOL(rpcauth_unregister); -EXPORT_SYMBOL(rpcauth_create); -EXPORT_SYMBOL(rpcauth_lookupcred); -EXPORT_SYMBOL(rpcauth_lookup_credcache); -EXPORT_SYMBOL(rpcauth_free_credcache); -EXPORT_SYMBOL(rpcauth_init_credcache); -EXPORT_SYMBOL(put_rpccred); - -/* RPC server stuff */ -EXPORT_SYMBOL(svc_create); -EXPORT_SYMBOL(svc_create_thread); -EXPORT_SYMBOL(svc_exit_thread); -EXPORT_SYMBOL(svc_destroy); -EXPORT_SYMBOL(svc_drop); -EXPORT_SYMBOL(svc_process); -EXPORT_SYMBOL(svc_recv); -EXPORT_SYMBOL(svc_wake_up); -EXPORT_SYMBOL(svc_makesock); -EXPORT_SYMBOL(svc_reserve); -EXPORT_SYMBOL(svc_auth_register); -EXPORT_SYMBOL(auth_domain_lookup); -EXPORT_SYMBOL(svc_authenticate); -EXPORT_SYMBOL(svc_set_client); - -/* RPC statistics */ -#ifdef CONFIG_PROC_FS -EXPORT_SYMBOL(rpc_proc_register); -EXPORT_SYMBOL(rpc_proc_unregister); -EXPORT_SYMBOL(svc_proc_register); -EXPORT_SYMBOL(svc_proc_unregister); -EXPORT_SYMBOL(svc_seq_show); -#endif +int sunrpc_net_id; +EXPORT_SYMBOL_GPL(sunrpc_net_id); -/* caching... */ -EXPORT_SYMBOL(auth_domain_find); -EXPORT_SYMBOL(auth_domain_put); -EXPORT_SYMBOL(auth_unix_add_addr); -EXPORT_SYMBOL(auth_unix_forget_old); -EXPORT_SYMBOL(auth_unix_lookup); -EXPORT_SYMBOL(cache_check); -EXPORT_SYMBOL(cache_flush); -EXPORT_SYMBOL(cache_purge); -EXPORT_SYMBOL(cache_fresh); -EXPORT_SYMBOL(cache_init); -EXPORT_SYMBOL(cache_register); -EXPORT_SYMBOL(cache_unregister); -EXPORT_SYMBOL(qword_add); -EXPORT_SYMBOL(qword_addhex); -EXPORT_SYMBOL(qword_get); -EXPORT_SYMBOL(svcauth_unix_purge); -EXPORT_SYMBOL(unix_domain_find); - -/* Generic XDR */ -EXPORT_SYMBOL(xdr_encode_string); -EXPORT_SYMBOL(xdr_decode_string); -EXPORT_SYMBOL(xdr_decode_string_inplace); -EXPORT_SYMBOL(xdr_decode_netobj); -EXPORT_SYMBOL(xdr_encode_netobj); -EXPORT_SYMBOL(xdr_encode_pages); -EXPORT_SYMBOL(xdr_inline_pages); -EXPORT_SYMBOL(xdr_shift_buf); -EXPORT_SYMBOL(xdr_encode_word); -EXPORT_SYMBOL(xdr_decode_word); -EXPORT_SYMBOL(xdr_encode_array2); -EXPORT_SYMBOL(xdr_decode_array2); -EXPORT_SYMBOL(xdr_buf_from_iov); -EXPORT_SYMBOL(xdr_buf_subsegment); -EXPORT_SYMBOL(xdr_buf_read_netobj); -EXPORT_SYMBOL(read_bytes_from_xdr_buf); - -/* Debugging symbols */ -#ifdef RPC_DEBUG -EXPORT_SYMBOL(rpc_debug); -EXPORT_SYMBOL(nfs_debug); -EXPORT_SYMBOL(nfsd_debug); -EXPORT_SYMBOL(nlm_debug); -#endif +static __net_init int sunrpc_init_net(struct net *net) +{ + int err; + struct sunrpc_net *sn = net_generic(net, sunrpc_net_id); + + err = rpc_proc_init(net); + if (err) + goto err_proc; + + err = ip_map_cache_create(net); + if (err) + goto err_ipmap; + + err = unix_gid_cache_create(net); + if (err) + goto err_unixgid; + + err = rpc_pipefs_init_net(net); + if (err) + goto err_pipefs; + + INIT_LIST_HEAD(&sn->all_clients); + spin_lock_init(&sn->rpc_client_lock); + spin_lock_init(&sn->rpcb_clnt_lock); + return 0; + +err_pipefs: + unix_gid_cache_destroy(net); +err_unixgid: + ip_map_cache_destroy(net); +err_ipmap: + rpc_proc_exit(net); +err_proc: + return err; +} + +static __net_exit void sunrpc_exit_net(struct net *net) +{ + rpc_pipefs_exit_net(net); + unix_gid_cache_destroy(net); + ip_map_cache_destroy(net); + rpc_proc_exit(net); +} -extern int register_rpc_pipefs(void); -extern void unregister_rpc_pipefs(void); +static struct pernet_operations sunrpc_net_ops = { + .init = sunrpc_init_net, + .exit = sunrpc_exit_net, + .id = &sunrpc_net_id, + .size = sizeof(struct sunrpc_net), +}; static int __init init_sunrpc(void) { - int err = register_rpc_pipefs(); + int err = rpc_init_mempool(); if (err) goto out; - err = rpc_init_mempool() != 0; + err = rpcauth_init_module(); if (err) - goto out; + goto out2; + + cache_initialize(); + + err = register_pernet_subsys(&sunrpc_net_ops); + if (err) + goto out3; + + err = register_rpc_pipefs(); + if (err) + goto out4; #ifdef RPC_DEBUG rpc_register_sysctl(); #endif -#ifdef CONFIG_PROC_FS - rpc_proc_init(); -#endif - cache_register(&auth_domain_cache); - cache_register(&ip_map_cache); + svc_init_xprt_sock(); /* svc sock transport */ + init_socket_xprt(); /* clnt sock transport */ + return 0; + +out4: + unregister_pernet_subsys(&sunrpc_net_ops); +out3: + rpcauth_remove_module(); +out2: + rpc_destroy_mempool(); out: return err; } @@ -171,19 +117,17 @@ out: static void __exit cleanup_sunrpc(void) { + rpcauth_remove_module(); + cleanup_socket_xprt(); + svc_cleanup_xprt_sock(); unregister_rpc_pipefs(); rpc_destroy_mempool(); - if (cache_unregister(&auth_domain_cache)) - printk(KERN_ERR "sunrpc: failed to unregister auth_domain cache\n"); - if (cache_unregister(&ip_map_cache)) - printk(KERN_ERR "sunrpc: failed to unregister ip_map cache\n"); + unregister_pernet_subsys(&sunrpc_net_ops); #ifdef RPC_DEBUG rpc_unregister_sysctl(); #endif -#ifdef CONFIG_PROC_FS - rpc_proc_exit(); -#endif + rcu_barrier(); /* Wait for completion of call_rcu()'s */ } MODULE_LICENSE("GPL"); -module_init(init_sunrpc); +fs_initcall(init_sunrpc); /* Ensure we're initialised before nfs */ module_exit(cleanup_sunrpc); diff --git a/net/sunrpc/svc.c b/net/sunrpc/svc.c index e9bd91265f7..5de6801cd92 100644 --- a/net/sunrpc/svc.c +++ b/net/sunrpc/svc.c @@ -4,6 +4,10 @@ * High-level RPC service routines * * Copyright (C) 1995, 1996 Olaf Kirch <okir@monad.swb.de> + * + * Multiple threads pools and NUMAisation + * Copyright (c) 2006 Silicon Graphics, Inc. + * by Greg Banks <gnb@melbourne.sgi.com> */ #include <linux/linkage.h> @@ -12,34 +16,429 @@ #include <linux/net.h> #include <linux/in.h> #include <linux/mm.h> +#include <linux/interrupt.h> +#include <linux/module.h> +#include <linux/kthread.h> +#include <linux/slab.h> #include <linux/sunrpc/types.h> #include <linux/sunrpc/xdr.h> #include <linux/sunrpc/stats.h> #include <linux/sunrpc/svcsock.h> #include <linux/sunrpc/clnt.h> +#include <linux/sunrpc/bc_xprt.h> #define RPCDBG_FACILITY RPCDBG_SVCDSP -#define RPC_PARANOIA 1 + +static void svc_unregister(const struct svc_serv *serv, struct net *net); + +#define svc_serv_is_pooled(serv) ((serv)->sv_function) + +/* + * Mode for mapping cpus to pools. + */ +enum { + SVC_POOL_AUTO = -1, /* choose one of the others */ + SVC_POOL_GLOBAL, /* no mapping, just a single global pool + * (legacy & UP mode) */ + SVC_POOL_PERCPU, /* one pool per cpu */ + SVC_POOL_PERNODE /* one pool per numa node */ +}; +#define SVC_POOL_DEFAULT SVC_POOL_GLOBAL + +/* + * Structure for mapping cpus to pools and vice versa. + * Setup once during sunrpc initialisation. + */ +static struct svc_pool_map { + int count; /* How many svc_servs use us */ + int mode; /* Note: int not enum to avoid + * warnings about "enumeration value + * not handled in switch" */ + unsigned int npools; + unsigned int *pool_to; /* maps pool id to cpu or node */ + unsigned int *to_pool; /* maps cpu or node to pool id */ +} svc_pool_map = { + .count = 0, + .mode = SVC_POOL_DEFAULT +}; +static DEFINE_MUTEX(svc_pool_map_mutex);/* protects svc_pool_map.count only */ + +static int +param_set_pool_mode(const char *val, struct kernel_param *kp) +{ + int *ip = (int *)kp->arg; + struct svc_pool_map *m = &svc_pool_map; + int err; + + mutex_lock(&svc_pool_map_mutex); + + err = -EBUSY; + if (m->count) + goto out; + + err = 0; + if (!strncmp(val, "auto", 4)) + *ip = SVC_POOL_AUTO; + else if (!strncmp(val, "global", 6)) + *ip = SVC_POOL_GLOBAL; + else if (!strncmp(val, "percpu", 6)) + *ip = SVC_POOL_PERCPU; + else if (!strncmp(val, "pernode", 7)) + *ip = SVC_POOL_PERNODE; + else + err = -EINVAL; + +out: + mutex_unlock(&svc_pool_map_mutex); + return err; +} + +static int +param_get_pool_mode(char *buf, struct kernel_param *kp) +{ + int *ip = (int *)kp->arg; + + switch (*ip) + { + case SVC_POOL_AUTO: + return strlcpy(buf, "auto", 20); + case SVC_POOL_GLOBAL: + return strlcpy(buf, "global", 20); + case SVC_POOL_PERCPU: + return strlcpy(buf, "percpu", 20); + case SVC_POOL_PERNODE: + return strlcpy(buf, "pernode", 20); + default: + return sprintf(buf, "%d", *ip); + } +} + +module_param_call(pool_mode, param_set_pool_mode, param_get_pool_mode, + &svc_pool_map.mode, 0644); + +/* + * Detect best pool mapping mode heuristically, + * according to the machine's topology. + */ +static int +svc_pool_map_choose_mode(void) +{ + unsigned int node; + + if (nr_online_nodes > 1) { + /* + * Actually have multiple NUMA nodes, + * so split pools on NUMA node boundaries + */ + return SVC_POOL_PERNODE; + } + + node = first_online_node; + if (nr_cpus_node(node) > 2) { + /* + * Non-trivial SMP, or CONFIG_NUMA on + * non-NUMA hardware, e.g. with a generic + * x86_64 kernel on Xeons. In this case we + * want to divide the pools on cpu boundaries. + */ + return SVC_POOL_PERCPU; + } + + /* default: one global pool */ + return SVC_POOL_GLOBAL; +} + +/* + * Allocate the to_pool[] and pool_to[] arrays. + * Returns 0 on success or an errno. + */ +static int +svc_pool_map_alloc_arrays(struct svc_pool_map *m, unsigned int maxpools) +{ + m->to_pool = kcalloc(maxpools, sizeof(unsigned int), GFP_KERNEL); + if (!m->to_pool) + goto fail; + m->pool_to = kcalloc(maxpools, sizeof(unsigned int), GFP_KERNEL); + if (!m->pool_to) + goto fail_free; + + return 0; + +fail_free: + kfree(m->to_pool); + m->to_pool = NULL; +fail: + return -ENOMEM; +} + +/* + * Initialise the pool map for SVC_POOL_PERCPU mode. + * Returns number of pools or <0 on error. + */ +static int +svc_pool_map_init_percpu(struct svc_pool_map *m) +{ + unsigned int maxpools = nr_cpu_ids; + unsigned int pidx = 0; + unsigned int cpu; + int err; + + err = svc_pool_map_alloc_arrays(m, maxpools); + if (err) + return err; + + for_each_online_cpu(cpu) { + BUG_ON(pidx > maxpools); + m->to_pool[cpu] = pidx; + m->pool_to[pidx] = cpu; + pidx++; + } + /* cpus brought online later all get mapped to pool0, sorry */ + + return pidx; +}; + + +/* + * Initialise the pool map for SVC_POOL_PERNODE mode. + * Returns number of pools or <0 on error. + */ +static int +svc_pool_map_init_pernode(struct svc_pool_map *m) +{ + unsigned int maxpools = nr_node_ids; + unsigned int pidx = 0; + unsigned int node; + int err; + + err = svc_pool_map_alloc_arrays(m, maxpools); + if (err) + return err; + + for_each_node_with_cpus(node) { + /* some architectures (e.g. SN2) have cpuless nodes */ + BUG_ON(pidx > maxpools); + m->to_pool[node] = pidx; + m->pool_to[pidx] = node; + pidx++; + } + /* nodes brought online later all get mapped to pool0, sorry */ + + return pidx; +} + + +/* + * Add a reference to the global map of cpus to pools (and + * vice versa). Initialise the map if we're the first user. + * Returns the number of pools. + */ +static unsigned int +svc_pool_map_get(void) +{ + struct svc_pool_map *m = &svc_pool_map; + int npools = -1; + + mutex_lock(&svc_pool_map_mutex); + + if (m->count++) { + mutex_unlock(&svc_pool_map_mutex); + return m->npools; + } + + if (m->mode == SVC_POOL_AUTO) + m->mode = svc_pool_map_choose_mode(); + + switch (m->mode) { + case SVC_POOL_PERCPU: + npools = svc_pool_map_init_percpu(m); + break; + case SVC_POOL_PERNODE: + npools = svc_pool_map_init_pernode(m); + break; + } + + if (npools < 0) { + /* default, or memory allocation failure */ + npools = 1; + m->mode = SVC_POOL_GLOBAL; + } + m->npools = npools; + + mutex_unlock(&svc_pool_map_mutex); + return m->npools; +} + + +/* + * Drop a reference to the global map of cpus to pools. + * When the last reference is dropped, the map data is + * freed; this allows the sysadmin to change the pool + * mode using the pool_mode module option without + * rebooting or re-loading sunrpc.ko. + */ +static void +svc_pool_map_put(void) +{ + struct svc_pool_map *m = &svc_pool_map; + + mutex_lock(&svc_pool_map_mutex); + + if (!--m->count) { + kfree(m->to_pool); + m->to_pool = NULL; + kfree(m->pool_to); + m->pool_to = NULL; + m->npools = 0; + } + + mutex_unlock(&svc_pool_map_mutex); +} + + +static int svc_pool_map_get_node(unsigned int pidx) +{ + const struct svc_pool_map *m = &svc_pool_map; + + if (m->count) { + if (m->mode == SVC_POOL_PERCPU) + return cpu_to_node(m->pool_to[pidx]); + if (m->mode == SVC_POOL_PERNODE) + return m->pool_to[pidx]; + } + return NUMA_NO_NODE; +} +/* + * Set the given thread's cpus_allowed mask so that it + * will only run on cpus in the given pool. + */ +static inline void +svc_pool_map_set_cpumask(struct task_struct *task, unsigned int pidx) +{ + struct svc_pool_map *m = &svc_pool_map; + unsigned int node = m->pool_to[pidx]; + + /* + * The caller checks for sv_nrpools > 1, which + * implies that we've been initialized. + */ + WARN_ON_ONCE(m->count == 0); + if (m->count == 0) + return; + + switch (m->mode) { + case SVC_POOL_PERCPU: + { + set_cpus_allowed_ptr(task, cpumask_of(node)); + break; + } + case SVC_POOL_PERNODE: + { + set_cpus_allowed_ptr(task, cpumask_of_node(node)); + break; + } + } +} + +/* + * Use the mapping mode to choose a pool for a given CPU. + * Used when enqueueing an incoming RPC. Always returns + * a non-NULL pool pointer. + */ +struct svc_pool * +svc_pool_for_cpu(struct svc_serv *serv, int cpu) +{ + struct svc_pool_map *m = &svc_pool_map; + unsigned int pidx = 0; + + /* + * An uninitialised map happens in a pure client when + * lockd is brought up, so silently treat it the + * same as SVC_POOL_GLOBAL. + */ + if (svc_serv_is_pooled(serv)) { + switch (m->mode) { + case SVC_POOL_PERCPU: + pidx = m->to_pool[cpu]; + break; + case SVC_POOL_PERNODE: + pidx = m->to_pool[cpu_to_node(cpu)]; + break; + } + } + return &serv->sv_pools[pidx % serv->sv_nrpools]; +} + +int svc_rpcb_setup(struct svc_serv *serv, struct net *net) +{ + int err; + + err = rpcb_create_local(net); + if (err) + return err; + + /* Remove any stale portmap registrations */ + svc_unregister(serv, net); + return 0; +} +EXPORT_SYMBOL_GPL(svc_rpcb_setup); + +void svc_rpcb_cleanup(struct svc_serv *serv, struct net *net) +{ + svc_unregister(serv, net); + rpcb_put_local(net); +} +EXPORT_SYMBOL_GPL(svc_rpcb_cleanup); + +static int svc_uses_rpcbind(struct svc_serv *serv) +{ + struct svc_program *progp; + unsigned int i; + + for (progp = serv->sv_program; progp; progp = progp->pg_next) { + for (i = 0; i < progp->pg_nvers; i++) { + if (progp->pg_vers[i] == NULL) + continue; + if (progp->pg_vers[i]->vs_hidden == 0) + return 1; + } + } + + return 0; +} + +int svc_bind(struct svc_serv *serv, struct net *net) +{ + if (!svc_uses_rpcbind(serv)) + return 0; + return svc_rpcb_setup(serv, net); +} +EXPORT_SYMBOL_GPL(svc_bind); /* * Create an RPC service */ -struct svc_serv * -svc_create(struct svc_program *prog, unsigned int bufsize) +static struct svc_serv * +__svc_create(struct svc_program *prog, unsigned int bufsize, int npools, + void (*shutdown)(struct svc_serv *serv, struct net *net)) { struct svc_serv *serv; - int vers; + unsigned int vers; unsigned int xdrsize; + unsigned int i; - if (!(serv = (struct svc_serv *) kmalloc(sizeof(*serv), GFP_KERNEL))) + if (!(serv = kzalloc(sizeof(*serv), GFP_KERNEL))) return NULL; - memset(serv, 0, sizeof(*serv)); serv->sv_name = prog->pg_name; serv->sv_program = prog; serv->sv_nrthreads = 1; serv->sv_stats = prog->pg_stats; - serv->sv_bufsz = bufsize? bufsize : 4096; + if (bufsize > RPCSVC_MAXPAYLOAD) + bufsize = RPCSVC_MAXPAYLOAD; + serv->sv_max_payload = bufsize? bufsize : 4096; + serv->sv_max_mesg = roundup(serv->sv_max_payload + PAGE_SIZE, PAGE_SIZE); + serv->sv_shutdown = shutdown; xdrsize = 0; while (prog) { prog->pg_lovers = prog->pg_nvers-1; @@ -54,27 +453,83 @@ svc_create(struct svc_program *prog, unsigned int bufsize) prog = prog->pg_next; } serv->sv_xdrsize = xdrsize; - INIT_LIST_HEAD(&serv->sv_threads); - INIT_LIST_HEAD(&serv->sv_sockets); INIT_LIST_HEAD(&serv->sv_tempsocks); INIT_LIST_HEAD(&serv->sv_permsocks); + init_timer(&serv->sv_temptimer); spin_lock_init(&serv->sv_lock); - /* Remove any stale portmap registrations */ - svc_register(serv, 0, 0); + serv->sv_nrpools = npools; + serv->sv_pools = + kcalloc(serv->sv_nrpools, sizeof(struct svc_pool), + GFP_KERNEL); + if (!serv->sv_pools) { + kfree(serv); + return NULL; + } + + for (i = 0; i < serv->sv_nrpools; i++) { + struct svc_pool *pool = &serv->sv_pools[i]; + + dprintk("svc: initialising pool %u for %s\n", + i, serv->sv_name); + + pool->sp_id = i; + INIT_LIST_HEAD(&pool->sp_threads); + INIT_LIST_HEAD(&pool->sp_sockets); + INIT_LIST_HEAD(&pool->sp_all_threads); + spin_lock_init(&pool->sp_lock); + } + + if (svc_uses_rpcbind(serv) && (!serv->sv_shutdown)) + serv->sv_shutdown = svc_rpcb_cleanup; return serv; } +struct svc_serv * +svc_create(struct svc_program *prog, unsigned int bufsize, + void (*shutdown)(struct svc_serv *serv, struct net *net)) +{ + return __svc_create(prog, bufsize, /*npools*/1, shutdown); +} +EXPORT_SYMBOL_GPL(svc_create); + +struct svc_serv * +svc_create_pooled(struct svc_program *prog, unsigned int bufsize, + void (*shutdown)(struct svc_serv *serv, struct net *net), + svc_thread_fn func, struct module *mod) +{ + struct svc_serv *serv; + unsigned int npools = svc_pool_map_get(); + + serv = __svc_create(prog, bufsize, npools, shutdown); + + if (serv != NULL) { + serv->sv_function = func; + serv->sv_module = mod; + } + + return serv; +} +EXPORT_SYMBOL_GPL(svc_create_pooled); + +void svc_shutdown_net(struct svc_serv *serv, struct net *net) +{ + svc_close_net(serv, net); + + if (serv->sv_shutdown) + serv->sv_shutdown(serv, net); +} +EXPORT_SYMBOL_GPL(svc_shutdown_net); + /* - * Destroy an RPC service + * Destroy an RPC service. Should be called with appropriate locking to + * protect the sv_nrthreads, sv_permsocks and sv_tempsocks. */ void svc_destroy(struct svc_serv *serv) { - struct svc_sock *svsk; - - dprintk("RPC: svc_destroy(%s, %d)\n", + dprintk("svc: svc_destroy(%s, %d)\n", serv->sv_program->pg_name, serv->sv_nrthreads); @@ -86,53 +541,53 @@ svc_destroy(struct svc_serv *serv) } else printk("svc_destroy: no threads for serv=%p!\n", serv); - while (!list_empty(&serv->sv_tempsocks)) { - svsk = list_entry(serv->sv_tempsocks.next, - struct svc_sock, - sk_list); - svc_delete_socket(svsk); - } - while (!list_empty(&serv->sv_permsocks)) { - svsk = list_entry(serv->sv_permsocks.next, - struct svc_sock, - sk_list); - svc_delete_socket(svsk); - } - + del_timer_sync(&serv->sv_temptimer); + + /* + * The last user is gone and thus all sockets have to be destroyed to + * the point. Check this. + */ + BUG_ON(!list_empty(&serv->sv_permsocks)); + BUG_ON(!list_empty(&serv->sv_tempsocks)); + cache_clean_deferred(serv); - /* Unregister service with the portmapper */ - svc_register(serv, 0, 0); + if (svc_serv_is_pooled(serv)) + svc_pool_map_put(); + + kfree(serv->sv_pools); kfree(serv); } +EXPORT_SYMBOL_GPL(svc_destroy); /* * Allocate an RPC server's buffer space. * We allocate pages and place them in rq_argpages. */ static int -svc_init_buffer(struct svc_rqst *rqstp, unsigned int size) -{ - int pages; - int arghi; - - if (size > RPCSVC_MAXPAYLOAD) - size = RPCSVC_MAXPAYLOAD; - pages = 2 + (size+ PAGE_SIZE -1) / PAGE_SIZE; - rqstp->rq_argused = 0; - rqstp->rq_resused = 0; +svc_init_buffer(struct svc_rqst *rqstp, unsigned int size, int node) +{ + unsigned int pages, arghi; + + /* bc_xprt uses fore channel allocated buffers */ + if (svc_is_backchannel(rqstp)) + return 1; + + pages = size / PAGE_SIZE + 1; /* extra page as we hold both request and reply. + * We assume one is at most one page + */ arghi = 0; + WARN_ON_ONCE(pages > RPCSVC_MAXPAGES); if (pages > RPCSVC_MAXPAGES) - BUG(); + pages = RPCSVC_MAXPAGES; while (pages) { - struct page *p = alloc_page(GFP_KERNEL); + struct page *p = alloc_pages_node(node, GFP_KERNEL, 0); if (!p) break; - rqstp->rq_argpages[arghi++] = p; + rqstp->rq_pages[arghi++] = p; pages--; } - rqstp->rq_arghi = arghi; - return ! pages; + return pages == 0; } /* @@ -141,178 +596,527 @@ svc_init_buffer(struct svc_rqst *rqstp, unsigned int size) static void svc_release_buffer(struct svc_rqst *rqstp) { - while (rqstp->rq_arghi) - put_page(rqstp->rq_argpages[--rqstp->rq_arghi]); - while (rqstp->rq_resused) { - if (rqstp->rq_respages[--rqstp->rq_resused] == NULL) - continue; - put_page(rqstp->rq_respages[rqstp->rq_resused]); - } - rqstp->rq_argused = 0; + unsigned int i; + + for (i = 0; i < ARRAY_SIZE(rqstp->rq_pages); i++) + if (rqstp->rq_pages[i]) + put_page(rqstp->rq_pages[i]); } -/* - * Create a server thread - */ -int -svc_create_thread(svc_thread_fn func, struct svc_serv *serv) +struct svc_rqst * +svc_prepare_thread(struct svc_serv *serv, struct svc_pool *pool, int node) { struct svc_rqst *rqstp; - int error = -ENOMEM; - rqstp = kmalloc(sizeof(*rqstp), GFP_KERNEL); + rqstp = kzalloc_node(sizeof(*rqstp), GFP_KERNEL, node); if (!rqstp) - goto out; + goto out_enomem; - memset(rqstp, 0, sizeof(*rqstp)); init_waitqueue_head(&rqstp->rq_wait); - if (!(rqstp->rq_argp = (u32 *) kmalloc(serv->sv_xdrsize, GFP_KERNEL)) - || !(rqstp->rq_resp = (u32 *) kmalloc(serv->sv_xdrsize, GFP_KERNEL)) - || !svc_init_buffer(rqstp, serv->sv_bufsz)) - goto out_thread; - serv->sv_nrthreads++; + spin_lock_bh(&pool->sp_lock); + pool->sp_nrthreads++; + list_add(&rqstp->rq_all, &pool->sp_all_threads); + spin_unlock_bh(&pool->sp_lock); rqstp->rq_server = serv; - error = kernel_thread((int (*)(void *)) func, rqstp, 0); - if (error < 0) + rqstp->rq_pool = pool; + + rqstp->rq_argp = kmalloc_node(serv->sv_xdrsize, GFP_KERNEL, node); + if (!rqstp->rq_argp) + goto out_thread; + + rqstp->rq_resp = kmalloc_node(serv->sv_xdrsize, GFP_KERNEL, node); + if (!rqstp->rq_resp) goto out_thread; - svc_sock_update_bufs(serv); - error = 0; -out: - return error; + if (!svc_init_buffer(rqstp, serv->sv_max_mesg, node)) + goto out_thread; + + return rqstp; out_thread: svc_exit_thread(rqstp); - goto out; +out_enomem: + return ERR_PTR(-ENOMEM); +} +EXPORT_SYMBOL_GPL(svc_prepare_thread); + +/* + * Choose a pool in which to create a new thread, for svc_set_num_threads + */ +static inline struct svc_pool * +choose_pool(struct svc_serv *serv, struct svc_pool *pool, unsigned int *state) +{ + if (pool != NULL) + return pool; + + return &serv->sv_pools[(*state)++ % serv->sv_nrpools]; +} + +/* + * Choose a thread to kill, for svc_set_num_threads + */ +static inline struct task_struct * +choose_victim(struct svc_serv *serv, struct svc_pool *pool, unsigned int *state) +{ + unsigned int i; + struct task_struct *task = NULL; + + if (pool != NULL) { + spin_lock_bh(&pool->sp_lock); + } else { + /* choose a pool in round-robin fashion */ + for (i = 0; i < serv->sv_nrpools; i++) { + pool = &serv->sv_pools[--(*state) % serv->sv_nrpools]; + spin_lock_bh(&pool->sp_lock); + if (!list_empty(&pool->sp_all_threads)) + goto found_pool; + spin_unlock_bh(&pool->sp_lock); + } + return NULL; + } + +found_pool: + if (!list_empty(&pool->sp_all_threads)) { + struct svc_rqst *rqstp; + + /* + * Remove from the pool->sp_all_threads list + * so we don't try to kill it again. + */ + rqstp = list_entry(pool->sp_all_threads.next, struct svc_rqst, rq_all); + list_del_init(&rqstp->rq_all); + task = rqstp->rq_task; + } + spin_unlock_bh(&pool->sp_lock); + + return task; +} + +/* + * Create or destroy enough new threads to make the number + * of threads the given number. If `pool' is non-NULL, applies + * only to threads in that pool, otherwise round-robins between + * all pools. Caller must ensure that mutual exclusion between this and + * server startup or shutdown. + * + * Destroying threads relies on the service threads filling in + * rqstp->rq_task, which only the nfs ones do. Assumes the serv + * has been created using svc_create_pooled(). + * + * Based on code that used to be in nfsd_svc() but tweaked + * to be pool-aware. + */ +int +svc_set_num_threads(struct svc_serv *serv, struct svc_pool *pool, int nrservs) +{ + struct svc_rqst *rqstp; + struct task_struct *task; + struct svc_pool *chosen_pool; + int error = 0; + unsigned int state = serv->sv_nrthreads-1; + int node; + + if (pool == NULL) { + /* The -1 assumes caller has done a svc_get() */ + nrservs -= (serv->sv_nrthreads-1); + } else { + spin_lock_bh(&pool->sp_lock); + nrservs -= pool->sp_nrthreads; + spin_unlock_bh(&pool->sp_lock); + } + + /* create new threads */ + while (nrservs > 0) { + nrservs--; + chosen_pool = choose_pool(serv, pool, &state); + + node = svc_pool_map_get_node(chosen_pool->sp_id); + rqstp = svc_prepare_thread(serv, chosen_pool, node); + if (IS_ERR(rqstp)) { + error = PTR_ERR(rqstp); + break; + } + + __module_get(serv->sv_module); + task = kthread_create_on_node(serv->sv_function, rqstp, + node, "%s", serv->sv_name); + if (IS_ERR(task)) { + error = PTR_ERR(task); + module_put(serv->sv_module); + svc_exit_thread(rqstp); + break; + } + + rqstp->rq_task = task; + if (serv->sv_nrpools > 1) + svc_pool_map_set_cpumask(task, chosen_pool->sp_id); + + svc_sock_update_bufs(serv); + wake_up_process(task); + } + /* destroy old threads */ + while (nrservs < 0 && + (task = choose_victim(serv, pool, &state)) != NULL) { + send_sig(SIGINT, task, 1); + nrservs++; + } + + return error; } +EXPORT_SYMBOL_GPL(svc_set_num_threads); /* - * Destroy an RPC server thread + * Called from a server thread as it's exiting. Caller must hold the BKL or + * the "service mutex", whichever is appropriate for the service. */ void svc_exit_thread(struct svc_rqst *rqstp) { struct svc_serv *serv = rqstp->rq_server; + struct svc_pool *pool = rqstp->rq_pool; svc_release_buffer(rqstp); - if (rqstp->rq_resp) - kfree(rqstp->rq_resp); - if (rqstp->rq_argp) - kfree(rqstp->rq_argp); - if (rqstp->rq_auth_data) - kfree(rqstp->rq_auth_data); + kfree(rqstp->rq_resp); + kfree(rqstp->rq_argp); + kfree(rqstp->rq_auth_data); + + spin_lock_bh(&pool->sp_lock); + pool->sp_nrthreads--; + list_del(&rqstp->rq_all); + spin_unlock_bh(&pool->sp_lock); + kfree(rqstp); /* Release the server */ if (serv) svc_destroy(serv); } +EXPORT_SYMBOL_GPL(svc_exit_thread); /* - * Register an RPC service with the local portmapper. - * To unregister a service, call this routine with - * proto and port == 0. + * Register an "inet" protocol family netid with the local + * rpcbind daemon via an rpcbind v4 SET request. + * + * No netconfig infrastructure is available in the kernel, so + * we map IP_ protocol numbers to netids by hand. + * + * Returns zero on success; a negative errno value is returned + * if any error occurs. */ -int -svc_register(struct svc_serv *serv, int proto, unsigned short port) +static int __svc_rpcb_register4(struct net *net, const u32 program, + const u32 version, + const unsigned short protocol, + const unsigned short port) { - struct svc_program *progp; - unsigned long flags; - int i, error = 0, dummy; + const struct sockaddr_in sin = { + .sin_family = AF_INET, + .sin_addr.s_addr = htonl(INADDR_ANY), + .sin_port = htons(port), + }; + const char *netid; + int error; + + switch (protocol) { + case IPPROTO_UDP: + netid = RPCBIND_NETID_UDP; + break; + case IPPROTO_TCP: + netid = RPCBIND_NETID_TCP; + break; + default: + return -ENOPROTOOPT; + } - progp = serv->sv_program; + error = rpcb_v4_register(net, program, version, + (const struct sockaddr *)&sin, netid); - dprintk("RPC: svc_register(%s, %s, %d)\n", - progp->pg_name, proto == IPPROTO_UDP? "udp" : "tcp", port); + /* + * User space didn't support rpcbind v4, so retry this + * registration request with the legacy rpcbind v2 protocol. + */ + if (error == -EPROTONOSUPPORT) + error = rpcb_register(net, program, version, protocol, port); - if (!port) - clear_thread_flag(TIF_SIGPENDING); + return error; +} - for (i = 0; i < progp->pg_nvers; i++) { - if (progp->pg_vers[i] == NULL) - continue; - error = rpc_register(progp->pg_prog, i, proto, port, &dummy); - if (error < 0) - break; - if (port && !dummy) { - error = -EACCES; - break; - } +#if IS_ENABLED(CONFIG_IPV6) +/* + * Register an "inet6" protocol family netid with the local + * rpcbind daemon via an rpcbind v4 SET request. + * + * No netconfig infrastructure is available in the kernel, so + * we map IP_ protocol numbers to netids by hand. + * + * Returns zero on success; a negative errno value is returned + * if any error occurs. + */ +static int __svc_rpcb_register6(struct net *net, const u32 program, + const u32 version, + const unsigned short protocol, + const unsigned short port) +{ + const struct sockaddr_in6 sin6 = { + .sin6_family = AF_INET6, + .sin6_addr = IN6ADDR_ANY_INIT, + .sin6_port = htons(port), + }; + const char *netid; + int error; + + switch (protocol) { + case IPPROTO_UDP: + netid = RPCBIND_NETID_UDP6; + break; + case IPPROTO_TCP: + netid = RPCBIND_NETID_TCP6; + break; + default: + return -ENOPROTOOPT; + } + + error = rpcb_v4_register(net, program, version, + (const struct sockaddr *)&sin6, netid); + + /* + * User space didn't support rpcbind version 4, so we won't + * use a PF_INET6 listener. + */ + if (error == -EPROTONOSUPPORT) + error = -EAFNOSUPPORT; + + return error; +} +#endif /* IS_ENABLED(CONFIG_IPV6) */ + +/* + * Register a kernel RPC service via rpcbind version 4. + * + * Returns zero on success; a negative errno value is returned + * if any error occurs. + */ +static int __svc_register(struct net *net, const char *progname, + const u32 program, const u32 version, + const int family, + const unsigned short protocol, + const unsigned short port) +{ + int error = -EAFNOSUPPORT; + + switch (family) { + case PF_INET: + error = __svc_rpcb_register4(net, program, version, + protocol, port); + break; +#if IS_ENABLED(CONFIG_IPV6) + case PF_INET6: + error = __svc_rpcb_register6(net, program, version, + protocol, port); +#endif } - if (!port) { - spin_lock_irqsave(¤t->sighand->siglock, flags); - recalc_sigpending(); - spin_unlock_irqrestore(¤t->sighand->siglock, flags); + return error; +} + +/** + * svc_register - register an RPC service with the local portmapper + * @serv: svc_serv struct for the service to register + * @net: net namespace for the service to register + * @family: protocol family of service's listener socket + * @proto: transport protocol number to advertise + * @port: port to advertise + * + * Service is registered for any address in the passed-in protocol family + */ +int svc_register(const struct svc_serv *serv, struct net *net, + const int family, const unsigned short proto, + const unsigned short port) +{ + struct svc_program *progp; + struct svc_version *vers; + unsigned int i; + int error = 0; + + WARN_ON_ONCE(proto == 0 && port == 0); + if (proto == 0 && port == 0) + return -EINVAL; + + for (progp = serv->sv_program; progp; progp = progp->pg_next) { + for (i = 0; i < progp->pg_nvers; i++) { + vers = progp->pg_vers[i]; + if (vers == NULL) + continue; + + dprintk("svc: svc_register(%sv%d, %s, %u, %u)%s\n", + progp->pg_name, + i, + proto == IPPROTO_UDP? "udp" : "tcp", + port, + family, + vers->vs_hidden ? + " (but not telling portmap)" : ""); + + if (vers->vs_hidden) + continue; + + error = __svc_register(net, progp->pg_name, progp->pg_prog, + i, family, proto, port); + + if (vers->vs_rpcb_optnl) { + error = 0; + continue; + } + + if (error < 0) { + printk(KERN_WARNING "svc: failed to register " + "%sv%u RPC service (errno %d).\n", + progp->pg_name, i, -error); + break; + } + } } return error; } /* - * Process the RPC request. + * If user space is running rpcbind, it should take the v4 UNSET + * and clear everything for this [program, version]. If user space + * is running portmap, it will reject the v4 UNSET, but won't have + * any "inet6" entries anyway. So a PMAP_UNSET should be sufficient + * in this case to clear all existing entries for [program, version]. */ -int -svc_process(struct svc_serv *serv, struct svc_rqst *rqstp) +static void __svc_unregister(struct net *net, const u32 program, const u32 version, + const char *progname) +{ + int error; + + error = rpcb_v4_register(net, program, version, NULL, ""); + + /* + * User space didn't support rpcbind v4, so retry this + * request with the legacy rpcbind v2 protocol. + */ + if (error == -EPROTONOSUPPORT) + error = rpcb_register(net, program, version, 0, 0); + + dprintk("svc: %s(%sv%u), error %d\n", + __func__, progname, version, error); +} + +/* + * All netids, bind addresses and ports registered for [program, version] + * are removed from the local rpcbind database (if the service is not + * hidden) to make way for a new instance of the service. + * + * The result of unregistration is reported via dprintk for those who want + * verification of the result, but is otherwise not important. + */ +static void svc_unregister(const struct svc_serv *serv, struct net *net) +{ + struct svc_program *progp; + unsigned long flags; + unsigned int i; + + clear_thread_flag(TIF_SIGPENDING); + + for (progp = serv->sv_program; progp; progp = progp->pg_next) { + for (i = 0; i < progp->pg_nvers; i++) { + if (progp->pg_vers[i] == NULL) + continue; + if (progp->pg_vers[i]->vs_hidden) + continue; + + dprintk("svc: attempting to unregister %sv%u\n", + progp->pg_name, i); + __svc_unregister(net, progp->pg_prog, i, progp->pg_name); + } + } + + spin_lock_irqsave(¤t->sighand->siglock, flags); + recalc_sigpending(); + spin_unlock_irqrestore(¤t->sighand->siglock, flags); +} + +/* + * dprintk the given error with the address of the client that caused it. + */ +#ifdef RPC_DEBUG +static __printf(2, 3) +void svc_printk(struct svc_rqst *rqstp, const char *fmt, ...) +{ + struct va_format vaf; + va_list args; + char buf[RPC_MAX_ADDRBUFLEN]; + + va_start(args, fmt); + + vaf.fmt = fmt; + vaf.va = &args; + + dprintk("svc: %s: %pV", svc_print_addr(rqstp, buf, sizeof(buf)), &vaf); + + va_end(args); +} +#else +static __printf(2,3) void svc_printk(struct svc_rqst *rqstp, const char *fmt, ...) {} +#endif + +/* + * Common routine for processing the RPC request. + */ +static int +svc_process_common(struct svc_rqst *rqstp, struct kvec *argv, struct kvec *resv) { struct svc_program *progp; struct svc_version *versp = NULL; /* compiler food */ struct svc_procedure *procp = NULL; - struct kvec * argv = &rqstp->rq_arg.head[0]; - struct kvec * resv = &rqstp->rq_res.head[0]; + struct svc_serv *serv = rqstp->rq_server; kxdrproc_t xdr; - u32 *statp; - u32 dir, prog, vers, proc, - auth_stat, rpc_stat; + __be32 *statp; + u32 prog, vers, proc; + __be32 auth_stat, rpc_stat; int auth_res; - u32 *accept_statp; + __be32 *reply_statp; rpc_stat = rpc_success; if (argv->iov_len < 6*4) goto err_short_len; - /* setup response xdr_buf. - * Initially it has just one page - */ - svc_take_page(rqstp); /* must succeed */ - resv->iov_base = page_address(rqstp->rq_respages[0]); - resv->iov_len = 0; - rqstp->rq_res.pages = rqstp->rq_respages+1; - rqstp->rq_res.len = 0; - rqstp->rq_res.page_base = 0; - rqstp->rq_res.page_len = 0; - rqstp->rq_res.buflen = PAGE_SIZE; - rqstp->rq_res.tail[0].iov_len = 0; - /* tcp needs a space for the record length... */ - if (rqstp->rq_prot == IPPROTO_TCP) - svc_putu32(resv, 0); + /* Will be turned off only in gss privacy case: */ + rqstp->rq_splice_ok = 1; + /* Will be turned off only when NFSv4 Sessions are used */ + rqstp->rq_usedeferral = 1; + rqstp->rq_dropme = false; + + /* Setup reply header */ + rqstp->rq_xprt->xpt_ops->xpo_prep_reply_hdr(rqstp); - rqstp->rq_xid = svc_getu32(argv); svc_putu32(resv, rqstp->rq_xid); - dir = ntohl(svc_getu32(argv)); - vers = ntohl(svc_getu32(argv)); + vers = svc_getnl(argv); /* First words of reply: */ - svc_putu32(resv, xdr_one); /* REPLY */ + svc_putnl(resv, 1); /* REPLY */ - if (dir != 0) /* direction != CALL */ - goto err_bad_dir; if (vers != 2) /* RPC version number */ goto err_bad_rpc; /* Save position in case we later decide to reject: */ - accept_statp = resv->iov_base + resv->iov_len; + reply_statp = resv->iov_base + resv->iov_len; + + svc_putnl(resv, 0); /* ACCEPT */ - svc_putu32(resv, xdr_zero); /* ACCEPT */ + rqstp->rq_prog = prog = svc_getnl(argv); /* program number */ + rqstp->rq_vers = vers = svc_getnl(argv); /* version number */ + rqstp->rq_proc = proc = svc_getnl(argv); /* procedure number */ - rqstp->rq_prog = prog = ntohl(svc_getu32(argv)); /* program number */ - rqstp->rq_vers = vers = ntohl(svc_getu32(argv)); /* version number */ - rqstp->rq_proc = proc = ntohl(svc_getu32(argv)); /* procedure number */ + for (progp = serv->sv_program; progp; progp = progp->pg_next) + if (prog == progp->pg_prog) + break; - progp = serv->sv_program; /* * Decode auth data, and add verifier to reply buffer. * We do this before anything else in order to get a decent @@ -320,7 +1124,7 @@ svc_process(struct svc_serv *serv, struct svc_rqst *rqstp) */ auth_res = svc_authenticate(rqstp, &auth_stat); /* Also give the program a chance to reject this call: */ - if (auth_res == SVC_OK) { + if (auth_res == SVC_OK && progp) { auth_stat = rpc_autherr_badcred; auth_res = progp->pg_authenticate(rqstp); } @@ -328,22 +1132,21 @@ svc_process(struct svc_serv *serv, struct svc_rqst *rqstp) case SVC_OK: break; case SVC_GARBAGE: - rpc_stat = rpc_garbage_args; - goto err_bad; + goto err_garbage; case SVC_SYSERR: rpc_stat = rpc_system_err; goto err_bad; case SVC_DENIED: goto err_bad_auth; + case SVC_CLOSE: + if (test_bit(XPT_TEMP, &rqstp->rq_xprt->xpt_flags)) + svc_close_xprt(rqstp->rq_xprt); case SVC_DROP: goto dropit; case SVC_COMPLETE: goto sendit; } - - for (progp = serv->sv_program; progp; progp = progp->pg_next) - if (prog == progp->pg_prog) - break; + if (progp == NULL) goto err_bad_prog; @@ -354,7 +1157,6 @@ svc_process(struct svc_serv *serv, struct svc_rqst *rqstp) procp = versp->vs_proc + proc; if (proc >= versp->vs_nproc || !procp->pc_func) goto err_bad_proc; - rqstp->rq_server = serv; rqstp->rq_procinfo = procp; /* Syntactic check complete */ @@ -362,7 +1164,7 @@ svc_process(struct svc_serv *serv, struct svc_rqst *rqstp) /* Build the reply header. */ statp = resv->iov_base +resv->iov_len; - svc_putu32(resv, rpc_success); /* RPC_SUCCESS */ + svc_putnl(resv, RPC_SUCCESS); /* Bump per-procedure stats counter */ procp->pc_count++; @@ -371,11 +1173,11 @@ svc_process(struct svc_serv *serv, struct svc_rqst *rqstp) memset(rqstp->rq_argp, 0, procp->pc_argsize); memset(rqstp->rq_resp, 0, procp->pc_ressize); - /* un-reserve some of the out-queue now that we have a + /* un-reserve some of the out-queue now that we have a * better idea of reply size */ if (procp->pc_xdrressize) - svc_reserve(rqstp, procp->pc_xdrressize<<2); + svc_reserve_auth(rqstp, procp->pc_xdrressize<<2); /* Call the function that processes the request. */ if (!versp->vs_dispatch) { @@ -387,8 +1189,14 @@ svc_process(struct svc_serv *serv, struct svc_rqst *rqstp) *statp = procp->pc_func(rqstp, rqstp->rq_argp, rqstp->rq_resp); /* Encode reply */ - if (*statp == rpc_success && (xdr = procp->pc_encode) - && !xdr(rqstp, resv->iov_base+resv->iov_len, rqstp->rq_resp)) { + if (rqstp->rq_dropme) { + if (procp->pc_release) + procp->pc_release(rqstp, NULL, rqstp->rq_resp); + goto dropit; + } + if (*statp == rpc_success && + (xdr = procp->pc_encode) && + !xdr(rqstp, resv->iov_base+resv->iov_len, rqstp->rq_resp)) { dprintk("svc: failed to encode reply\n"); /* serv->sv_stats->rpcsystemerr++; */ *statp = rpc_system_err; @@ -417,76 +1225,178 @@ svc_process(struct svc_serv *serv, struct svc_rqst *rqstp) sendit: if (svc_authorise(rqstp)) goto dropit; - return svc_send(rqstp); + return 1; /* Caller can now send it */ dropit: svc_authorise(rqstp); /* doesn't hurt to call this twice */ dprintk("svc: svc_process dropit\n"); - svc_drop(rqstp); return 0; err_short_len: -#ifdef RPC_PARANOIA - printk("svc: short len %Zd, dropping request\n", argv->iov_len); -#endif - goto dropit; /* drop request */ + svc_printk(rqstp, "short len %Zd, dropping request\n", + argv->iov_len); -err_bad_dir: -#ifdef RPC_PARANOIA - printk("svc: bad direction %d, dropping request\n", dir); -#endif - serv->sv_stats->rpcbadfmt++; goto dropit; /* drop request */ err_bad_rpc: serv->sv_stats->rpcbadfmt++; - svc_putu32(resv, xdr_one); /* REJECT */ - svc_putu32(resv, xdr_zero); /* RPC_MISMATCH */ - svc_putu32(resv, xdr_two); /* Only RPCv2 supported */ - svc_putu32(resv, xdr_two); + svc_putnl(resv, 1); /* REJECT */ + svc_putnl(resv, 0); /* RPC_MISMATCH */ + svc_putnl(resv, 2); /* Only RPCv2 supported */ + svc_putnl(resv, 2); goto sendit; err_bad_auth: dprintk("svc: authentication failed (%d)\n", ntohl(auth_stat)); serv->sv_stats->rpcbadauth++; /* Restore write pointer to location of accept status: */ - xdr_ressize_check(rqstp, accept_statp); - svc_putu32(resv, xdr_one); /* REJECT */ - svc_putu32(resv, xdr_one); /* AUTH_ERROR */ - svc_putu32(resv, auth_stat); /* status */ + xdr_ressize_check(rqstp, reply_statp); + svc_putnl(resv, 1); /* REJECT */ + svc_putnl(resv, 1); /* AUTH_ERROR */ + svc_putnl(resv, ntohl(auth_stat)); /* status */ goto sendit; err_bad_prog: dprintk("svc: unknown program %d\n", prog); serv->sv_stats->rpcbadfmt++; - svc_putu32(resv, rpc_prog_unavail); + svc_putnl(resv, RPC_PROG_UNAVAIL); goto sendit; err_bad_vers: -#ifdef RPC_PARANOIA - printk("svc: unknown version (%d)\n", vers); -#endif + svc_printk(rqstp, "unknown version (%d for prog %d, %s)\n", + vers, prog, progp->pg_name); + serv->sv_stats->rpcbadfmt++; - svc_putu32(resv, rpc_prog_mismatch); - svc_putu32(resv, htonl(progp->pg_lovers)); - svc_putu32(resv, htonl(progp->pg_hivers)); + svc_putnl(resv, RPC_PROG_MISMATCH); + svc_putnl(resv, progp->pg_lovers); + svc_putnl(resv, progp->pg_hivers); goto sendit; err_bad_proc: -#ifdef RPC_PARANOIA - printk("svc: unknown procedure (%d)\n", proc); -#endif + svc_printk(rqstp, "unknown procedure (%d)\n", proc); + serv->sv_stats->rpcbadfmt++; - svc_putu32(resv, rpc_proc_unavail); + svc_putnl(resv, RPC_PROC_UNAVAIL); goto sendit; err_garbage: -#ifdef RPC_PARANOIA - printk("svc: failed to decode args\n"); -#endif + svc_printk(rqstp, "failed to decode args\n"); + rpc_stat = rpc_garbage_args; err_bad: serv->sv_stats->rpcbadfmt++; - svc_putu32(resv, rpc_stat); + svc_putnl(resv, ntohl(rpc_stat)); goto sendit; } +EXPORT_SYMBOL_GPL(svc_process); + +/* + * Process the RPC request. + */ +int +svc_process(struct svc_rqst *rqstp) +{ + struct kvec *argv = &rqstp->rq_arg.head[0]; + struct kvec *resv = &rqstp->rq_res.head[0]; + struct svc_serv *serv = rqstp->rq_server; + u32 dir; + + /* + * Setup response xdr_buf. + * Initially it has just one page + */ + rqstp->rq_next_page = &rqstp->rq_respages[1]; + resv->iov_base = page_address(rqstp->rq_respages[0]); + resv->iov_len = 0; + rqstp->rq_res.pages = rqstp->rq_respages + 1; + rqstp->rq_res.len = 0; + rqstp->rq_res.page_base = 0; + rqstp->rq_res.page_len = 0; + rqstp->rq_res.buflen = PAGE_SIZE; + rqstp->rq_res.tail[0].iov_base = NULL; + rqstp->rq_res.tail[0].iov_len = 0; + + rqstp->rq_xid = svc_getu32(argv); + + dir = svc_getnl(argv); + if (dir != 0) { + /* direction != CALL */ + svc_printk(rqstp, "bad direction %d, dropping request\n", dir); + serv->sv_stats->rpcbadfmt++; + svc_drop(rqstp); + return 0; + } + + /* Returns 1 for send, 0 for drop */ + if (svc_process_common(rqstp, argv, resv)) + return svc_send(rqstp); + else { + svc_drop(rqstp); + return 0; + } +} + +#if defined(CONFIG_SUNRPC_BACKCHANNEL) +/* + * Process a backchannel RPC request that arrived over an existing + * outbound connection + */ +int +bc_svc_process(struct svc_serv *serv, struct rpc_rqst *req, + struct svc_rqst *rqstp) +{ + struct kvec *argv = &rqstp->rq_arg.head[0]; + struct kvec *resv = &rqstp->rq_res.head[0]; + + /* Build the svc_rqst used by the common processing routine */ + rqstp->rq_xprt = serv->sv_bc_xprt; + rqstp->rq_xid = req->rq_xid; + rqstp->rq_prot = req->rq_xprt->prot; + rqstp->rq_server = serv; + + rqstp->rq_addrlen = sizeof(req->rq_xprt->addr); + memcpy(&rqstp->rq_addr, &req->rq_xprt->addr, rqstp->rq_addrlen); + memcpy(&rqstp->rq_arg, &req->rq_rcv_buf, sizeof(rqstp->rq_arg)); + memcpy(&rqstp->rq_res, &req->rq_snd_buf, sizeof(rqstp->rq_res)); + + /* reset result send buffer "put" position */ + resv->iov_len = 0; + + if (rqstp->rq_prot != IPPROTO_TCP) { + printk(KERN_ERR "No support for Non-TCP transports!\n"); + BUG(); + } + + /* + * Skip the next two words because they've already been + * processed in the trasport + */ + svc_getu32(argv); /* XID */ + svc_getnl(argv); /* CALLDIR */ + + /* Returns 1 for send, 0 for drop */ + if (svc_process_common(rqstp, argv, resv)) { + memcpy(&req->rq_snd_buf, &rqstp->rq_res, + sizeof(req->rq_snd_buf)); + return bc_send(req); + } else { + /* drop request */ + xprt_free_bc_request(req); + return 0; + } +} +EXPORT_SYMBOL_GPL(bc_svc_process); +#endif /* CONFIG_SUNRPC_BACKCHANNEL */ + +/* + * Return (transport-specific) limit on the rpc payload. + */ +u32 svc_max_payload(const struct svc_rqst *rqstp) +{ + u32 max = rqstp->rq_xprt->xpt_class->xcl_max_payload; + + if (rqstp->rq_server->sv_max_payload < max) + max = rqstp->rq_server->sv_max_payload; + return max; +} +EXPORT_SYMBOL_GPL(svc_max_payload); diff --git a/net/sunrpc/svc_xprt.c b/net/sunrpc/svc_xprt.c new file mode 100644 index 00000000000..b4737fbdec1 --- /dev/null +++ b/net/sunrpc/svc_xprt.c @@ -0,0 +1,1333 @@ +/* + * linux/net/sunrpc/svc_xprt.c + * + * Author: Tom Tucker <tom@opengridcomputing.com> + */ + +#include <linux/sched.h> +#include <linux/errno.h> +#include <linux/freezer.h> +#include <linux/kthread.h> +#include <linux/slab.h> +#include <net/sock.h> +#include <linux/sunrpc/stats.h> +#include <linux/sunrpc/svc_xprt.h> +#include <linux/sunrpc/svcsock.h> +#include <linux/sunrpc/xprt.h> +#include <linux/module.h> + +#define RPCDBG_FACILITY RPCDBG_SVCXPRT + +static struct svc_deferred_req *svc_deferred_dequeue(struct svc_xprt *xprt); +static int svc_deferred_recv(struct svc_rqst *rqstp); +static struct cache_deferred_req *svc_defer(struct cache_req *req); +static void svc_age_temp_xprts(unsigned long closure); +static void svc_delete_xprt(struct svc_xprt *xprt); + +/* apparently the "standard" is that clients close + * idle connections after 5 minutes, servers after + * 6 minutes + * http://www.connectathon.org/talks96/nfstcp.pdf + */ +static int svc_conn_age_period = 6*60; + +/* List of registered transport classes */ +static DEFINE_SPINLOCK(svc_xprt_class_lock); +static LIST_HEAD(svc_xprt_class_list); + +/* SMP locking strategy: + * + * svc_pool->sp_lock protects most of the fields of that pool. + * svc_serv->sv_lock protects sv_tempsocks, sv_permsocks, sv_tmpcnt. + * when both need to be taken (rare), svc_serv->sv_lock is first. + * BKL protects svc_serv->sv_nrthread. + * svc_sock->sk_lock protects the svc_sock->sk_deferred list + * and the ->sk_info_authunix cache. + * + * The XPT_BUSY bit in xprt->xpt_flags prevents a transport being + * enqueued multiply. During normal transport processing this bit + * is set by svc_xprt_enqueue and cleared by svc_xprt_received. + * Providers should not manipulate this bit directly. + * + * Some flags can be set to certain values at any time + * providing that certain rules are followed: + * + * XPT_CONN, XPT_DATA: + * - Can be set or cleared at any time. + * - After a set, svc_xprt_enqueue must be called to enqueue + * the transport for processing. + * - After a clear, the transport must be read/accepted. + * If this succeeds, it must be set again. + * XPT_CLOSE: + * - Can set at any time. It is never cleared. + * XPT_DEAD: + * - Can only be set while XPT_BUSY is held which ensures + * that no other thread will be using the transport or will + * try to set XPT_DEAD. + */ + +int svc_reg_xprt_class(struct svc_xprt_class *xcl) +{ + struct svc_xprt_class *cl; + int res = -EEXIST; + + dprintk("svc: Adding svc transport class '%s'\n", xcl->xcl_name); + + INIT_LIST_HEAD(&xcl->xcl_list); + spin_lock(&svc_xprt_class_lock); + /* Make sure there isn't already a class with the same name */ + list_for_each_entry(cl, &svc_xprt_class_list, xcl_list) { + if (strcmp(xcl->xcl_name, cl->xcl_name) == 0) + goto out; + } + list_add_tail(&xcl->xcl_list, &svc_xprt_class_list); + res = 0; +out: + spin_unlock(&svc_xprt_class_lock); + return res; +} +EXPORT_SYMBOL_GPL(svc_reg_xprt_class); + +void svc_unreg_xprt_class(struct svc_xprt_class *xcl) +{ + dprintk("svc: Removing svc transport class '%s'\n", xcl->xcl_name); + spin_lock(&svc_xprt_class_lock); + list_del_init(&xcl->xcl_list); + spin_unlock(&svc_xprt_class_lock); +} +EXPORT_SYMBOL_GPL(svc_unreg_xprt_class); + +/* + * Format the transport list for printing + */ +int svc_print_xprts(char *buf, int maxlen) +{ + struct svc_xprt_class *xcl; + char tmpstr[80]; + int len = 0; + buf[0] = '\0'; + + spin_lock(&svc_xprt_class_lock); + list_for_each_entry(xcl, &svc_xprt_class_list, xcl_list) { + int slen; + + sprintf(tmpstr, "%s %d\n", xcl->xcl_name, xcl->xcl_max_payload); + slen = strlen(tmpstr); + if (len + slen > maxlen) + break; + len += slen; + strcat(buf, tmpstr); + } + spin_unlock(&svc_xprt_class_lock); + + return len; +} + +static void svc_xprt_free(struct kref *kref) +{ + struct svc_xprt *xprt = + container_of(kref, struct svc_xprt, xpt_ref); + struct module *owner = xprt->xpt_class->xcl_owner; + if (test_bit(XPT_CACHE_AUTH, &xprt->xpt_flags)) + svcauth_unix_info_release(xprt); + put_net(xprt->xpt_net); + /* See comment on corresponding get in xs_setup_bc_tcp(): */ + if (xprt->xpt_bc_xprt) + xprt_put(xprt->xpt_bc_xprt); + xprt->xpt_ops->xpo_free(xprt); + module_put(owner); +} + +void svc_xprt_put(struct svc_xprt *xprt) +{ + kref_put(&xprt->xpt_ref, svc_xprt_free); +} +EXPORT_SYMBOL_GPL(svc_xprt_put); + +/* + * Called by transport drivers to initialize the transport independent + * portion of the transport instance. + */ +void svc_xprt_init(struct net *net, struct svc_xprt_class *xcl, + struct svc_xprt *xprt, struct svc_serv *serv) +{ + memset(xprt, 0, sizeof(*xprt)); + xprt->xpt_class = xcl; + xprt->xpt_ops = xcl->xcl_ops; + kref_init(&xprt->xpt_ref); + xprt->xpt_server = serv; + INIT_LIST_HEAD(&xprt->xpt_list); + INIT_LIST_HEAD(&xprt->xpt_ready); + INIT_LIST_HEAD(&xprt->xpt_deferred); + INIT_LIST_HEAD(&xprt->xpt_users); + mutex_init(&xprt->xpt_mutex); + spin_lock_init(&xprt->xpt_lock); + set_bit(XPT_BUSY, &xprt->xpt_flags); + rpc_init_wait_queue(&xprt->xpt_bc_pending, "xpt_bc_pending"); + xprt->xpt_net = get_net(net); +} +EXPORT_SYMBOL_GPL(svc_xprt_init); + +static struct svc_xprt *__svc_xpo_create(struct svc_xprt_class *xcl, + struct svc_serv *serv, + struct net *net, + const int family, + const unsigned short port, + int flags) +{ + struct sockaddr_in sin = { + .sin_family = AF_INET, + .sin_addr.s_addr = htonl(INADDR_ANY), + .sin_port = htons(port), + }; +#if IS_ENABLED(CONFIG_IPV6) + struct sockaddr_in6 sin6 = { + .sin6_family = AF_INET6, + .sin6_addr = IN6ADDR_ANY_INIT, + .sin6_port = htons(port), + }; +#endif + struct sockaddr *sap; + size_t len; + + switch (family) { + case PF_INET: + sap = (struct sockaddr *)&sin; + len = sizeof(sin); + break; +#if IS_ENABLED(CONFIG_IPV6) + case PF_INET6: + sap = (struct sockaddr *)&sin6; + len = sizeof(sin6); + break; +#endif + default: + return ERR_PTR(-EAFNOSUPPORT); + } + + return xcl->xcl_ops->xpo_create(serv, net, sap, len, flags); +} + +/* + * svc_xprt_received conditionally queues the transport for processing + * by another thread. The caller must hold the XPT_BUSY bit and must + * not thereafter touch transport data. + * + * Note: XPT_DATA only gets cleared when a read-attempt finds no (or + * insufficient) data. + */ +static void svc_xprt_received(struct svc_xprt *xprt) +{ + WARN_ON_ONCE(!test_bit(XPT_BUSY, &xprt->xpt_flags)); + if (!test_bit(XPT_BUSY, &xprt->xpt_flags)) + return; + /* As soon as we clear busy, the xprt could be closed and + * 'put', so we need a reference to call svc_xprt_enqueue with: + */ + svc_xprt_get(xprt); + clear_bit(XPT_BUSY, &xprt->xpt_flags); + svc_xprt_enqueue(xprt); + svc_xprt_put(xprt); +} + +void svc_add_new_perm_xprt(struct svc_serv *serv, struct svc_xprt *new) +{ + clear_bit(XPT_TEMP, &new->xpt_flags); + spin_lock_bh(&serv->sv_lock); + list_add(&new->xpt_list, &serv->sv_permsocks); + spin_unlock_bh(&serv->sv_lock); + svc_xprt_received(new); +} + +int svc_create_xprt(struct svc_serv *serv, const char *xprt_name, + struct net *net, const int family, + const unsigned short port, int flags) +{ + struct svc_xprt_class *xcl; + + dprintk("svc: creating transport %s[%d]\n", xprt_name, port); + spin_lock(&svc_xprt_class_lock); + list_for_each_entry(xcl, &svc_xprt_class_list, xcl_list) { + struct svc_xprt *newxprt; + unsigned short newport; + + if (strcmp(xprt_name, xcl->xcl_name)) + continue; + + if (!try_module_get(xcl->xcl_owner)) + goto err; + + spin_unlock(&svc_xprt_class_lock); + newxprt = __svc_xpo_create(xcl, serv, net, family, port, flags); + if (IS_ERR(newxprt)) { + module_put(xcl->xcl_owner); + return PTR_ERR(newxprt); + } + svc_add_new_perm_xprt(serv, newxprt); + newport = svc_xprt_local_port(newxprt); + return newport; + } + err: + spin_unlock(&svc_xprt_class_lock); + dprintk("svc: transport %s not found\n", xprt_name); + + /* This errno is exposed to user space. Provide a reasonable + * perror msg for a bad transport. */ + return -EPROTONOSUPPORT; +} +EXPORT_SYMBOL_GPL(svc_create_xprt); + +/* + * Copy the local and remote xprt addresses to the rqstp structure + */ +void svc_xprt_copy_addrs(struct svc_rqst *rqstp, struct svc_xprt *xprt) +{ + memcpy(&rqstp->rq_addr, &xprt->xpt_remote, xprt->xpt_remotelen); + rqstp->rq_addrlen = xprt->xpt_remotelen; + + /* + * Destination address in request is needed for binding the + * source address in RPC replies/callbacks later. + */ + memcpy(&rqstp->rq_daddr, &xprt->xpt_local, xprt->xpt_locallen); + rqstp->rq_daddrlen = xprt->xpt_locallen; +} +EXPORT_SYMBOL_GPL(svc_xprt_copy_addrs); + +/** + * svc_print_addr - Format rq_addr field for printing + * @rqstp: svc_rqst struct containing address to print + * @buf: target buffer for formatted address + * @len: length of target buffer + * + */ +char *svc_print_addr(struct svc_rqst *rqstp, char *buf, size_t len) +{ + return __svc_print_addr(svc_addr(rqstp), buf, len); +} +EXPORT_SYMBOL_GPL(svc_print_addr); + +/* + * Queue up an idle server thread. Must have pool->sp_lock held. + * Note: this is really a stack rather than a queue, so that we only + * use as many different threads as we need, and the rest don't pollute + * the cache. + */ +static void svc_thread_enqueue(struct svc_pool *pool, struct svc_rqst *rqstp) +{ + list_add(&rqstp->rq_list, &pool->sp_threads); +} + +/* + * Dequeue an nfsd thread. Must have pool->sp_lock held. + */ +static void svc_thread_dequeue(struct svc_pool *pool, struct svc_rqst *rqstp) +{ + list_del(&rqstp->rq_list); +} + +static bool svc_xprt_has_something_to_do(struct svc_xprt *xprt) +{ + if (xprt->xpt_flags & ((1<<XPT_CONN)|(1<<XPT_CLOSE))) + return true; + if (xprt->xpt_flags & ((1<<XPT_DATA)|(1<<XPT_DEFERRED))) + return xprt->xpt_ops->xpo_has_wspace(xprt); + return false; +} + +/* + * Queue up a transport with data pending. If there are idle nfsd + * processes, wake 'em up. + * + */ +void svc_xprt_enqueue(struct svc_xprt *xprt) +{ + struct svc_pool *pool; + struct svc_rqst *rqstp; + int cpu; + + if (!svc_xprt_has_something_to_do(xprt)) + return; + + cpu = get_cpu(); + pool = svc_pool_for_cpu(xprt->xpt_server, cpu); + put_cpu(); + + spin_lock_bh(&pool->sp_lock); + + if (!list_empty(&pool->sp_threads) && + !list_empty(&pool->sp_sockets)) + printk(KERN_ERR + "svc_xprt_enqueue: " + "threads and transports both waiting??\n"); + + pool->sp_stats.packets++; + + /* Mark transport as busy. It will remain in this state until + * the provider calls svc_xprt_received. We update XPT_BUSY + * atomically because it also guards against trying to enqueue + * the transport twice. + */ + if (test_and_set_bit(XPT_BUSY, &xprt->xpt_flags)) { + /* Don't enqueue transport while already enqueued */ + dprintk("svc: transport %p busy, not enqueued\n", xprt); + goto out_unlock; + } + + if (!list_empty(&pool->sp_threads)) { + rqstp = list_entry(pool->sp_threads.next, + struct svc_rqst, + rq_list); + dprintk("svc: transport %p served by daemon %p\n", + xprt, rqstp); + svc_thread_dequeue(pool, rqstp); + if (rqstp->rq_xprt) + printk(KERN_ERR + "svc_xprt_enqueue: server %p, rq_xprt=%p!\n", + rqstp, rqstp->rq_xprt); + rqstp->rq_xprt = xprt; + svc_xprt_get(xprt); + pool->sp_stats.threads_woken++; + wake_up(&rqstp->rq_wait); + } else { + dprintk("svc: transport %p put into queue\n", xprt); + list_add_tail(&xprt->xpt_ready, &pool->sp_sockets); + pool->sp_stats.sockets_queued++; + } + +out_unlock: + spin_unlock_bh(&pool->sp_lock); +} +EXPORT_SYMBOL_GPL(svc_xprt_enqueue); + +/* + * Dequeue the first transport. Must be called with the pool->sp_lock held. + */ +static struct svc_xprt *svc_xprt_dequeue(struct svc_pool *pool) +{ + struct svc_xprt *xprt; + + if (list_empty(&pool->sp_sockets)) + return NULL; + + xprt = list_entry(pool->sp_sockets.next, + struct svc_xprt, xpt_ready); + list_del_init(&xprt->xpt_ready); + + dprintk("svc: transport %p dequeued, inuse=%d\n", + xprt, atomic_read(&xprt->xpt_ref.refcount)); + + return xprt; +} + +/** + * svc_reserve - change the space reserved for the reply to a request. + * @rqstp: The request in question + * @space: new max space to reserve + * + * Each request reserves some space on the output queue of the transport + * to make sure the reply fits. This function reduces that reserved + * space to be the amount of space used already, plus @space. + * + */ +void svc_reserve(struct svc_rqst *rqstp, int space) +{ + space += rqstp->rq_res.head[0].iov_len; + + if (space < rqstp->rq_reserved) { + struct svc_xprt *xprt = rqstp->rq_xprt; + atomic_sub((rqstp->rq_reserved - space), &xprt->xpt_reserved); + rqstp->rq_reserved = space; + + svc_xprt_enqueue(xprt); + } +} +EXPORT_SYMBOL_GPL(svc_reserve); + +static void svc_xprt_release(struct svc_rqst *rqstp) +{ + struct svc_xprt *xprt = rqstp->rq_xprt; + + rqstp->rq_xprt->xpt_ops->xpo_release_rqst(rqstp); + + kfree(rqstp->rq_deferred); + rqstp->rq_deferred = NULL; + + svc_free_res_pages(rqstp); + rqstp->rq_res.page_len = 0; + rqstp->rq_res.page_base = 0; + + /* Reset response buffer and release + * the reservation. + * But first, check that enough space was reserved + * for the reply, otherwise we have a bug! + */ + if ((rqstp->rq_res.len) > rqstp->rq_reserved) + printk(KERN_ERR "RPC request reserved %d but used %d\n", + rqstp->rq_reserved, + rqstp->rq_res.len); + + rqstp->rq_res.head[0].iov_len = 0; + svc_reserve(rqstp, 0); + rqstp->rq_xprt = NULL; + + svc_xprt_put(xprt); +} + +/* + * External function to wake up a server waiting for data + * This really only makes sense for services like lockd + * which have exactly one thread anyway. + */ +void svc_wake_up(struct svc_serv *serv) +{ + struct svc_rqst *rqstp; + unsigned int i; + struct svc_pool *pool; + + for (i = 0; i < serv->sv_nrpools; i++) { + pool = &serv->sv_pools[i]; + + spin_lock_bh(&pool->sp_lock); + if (!list_empty(&pool->sp_threads)) { + rqstp = list_entry(pool->sp_threads.next, + struct svc_rqst, + rq_list); + dprintk("svc: daemon %p woken up.\n", rqstp); + /* + svc_thread_dequeue(pool, rqstp); + rqstp->rq_xprt = NULL; + */ + wake_up(&rqstp->rq_wait); + } else + pool->sp_task_pending = 1; + spin_unlock_bh(&pool->sp_lock); + } +} +EXPORT_SYMBOL_GPL(svc_wake_up); + +int svc_port_is_privileged(struct sockaddr *sin) +{ + switch (sin->sa_family) { + case AF_INET: + return ntohs(((struct sockaddr_in *)sin)->sin_port) + < PROT_SOCK; + case AF_INET6: + return ntohs(((struct sockaddr_in6 *)sin)->sin6_port) + < PROT_SOCK; + default: + return 0; + } +} + +/* + * Make sure that we don't have too many active connections. If we have, + * something must be dropped. It's not clear what will happen if we allow + * "too many" connections, but when dealing with network-facing software, + * we have to code defensively. Here we do that by imposing hard limits. + * + * There's no point in trying to do random drop here for DoS + * prevention. The NFS clients does 1 reconnect in 15 seconds. An + * attacker can easily beat that. + * + * The only somewhat efficient mechanism would be if drop old + * connections from the same IP first. But right now we don't even + * record the client IP in svc_sock. + * + * single-threaded services that expect a lot of clients will probably + * need to set sv_maxconn to override the default value which is based + * on the number of threads + */ +static void svc_check_conn_limits(struct svc_serv *serv) +{ + unsigned int limit = serv->sv_maxconn ? serv->sv_maxconn : + (serv->sv_nrthreads+3) * 20; + + if (serv->sv_tmpcnt > limit) { + struct svc_xprt *xprt = NULL; + spin_lock_bh(&serv->sv_lock); + if (!list_empty(&serv->sv_tempsocks)) { + /* Try to help the admin */ + net_notice_ratelimited("%s: too many open connections, consider increasing the %s\n", + serv->sv_name, serv->sv_maxconn ? + "max number of connections" : + "number of threads"); + /* + * Always select the oldest connection. It's not fair, + * but so is life + */ + xprt = list_entry(serv->sv_tempsocks.prev, + struct svc_xprt, + xpt_list); + set_bit(XPT_CLOSE, &xprt->xpt_flags); + svc_xprt_get(xprt); + } + spin_unlock_bh(&serv->sv_lock); + + if (xprt) { + svc_xprt_enqueue(xprt); + svc_xprt_put(xprt); + } + } +} + +static int svc_alloc_arg(struct svc_rqst *rqstp) +{ + struct svc_serv *serv = rqstp->rq_server; + struct xdr_buf *arg; + int pages; + int i; + + /* now allocate needed pages. If we get a failure, sleep briefly */ + pages = (serv->sv_max_mesg + PAGE_SIZE) / PAGE_SIZE; + WARN_ON_ONCE(pages >= RPCSVC_MAXPAGES); + if (pages >= RPCSVC_MAXPAGES) + /* use as many pages as possible */ + pages = RPCSVC_MAXPAGES - 1; + for (i = 0; i < pages ; i++) + while (rqstp->rq_pages[i] == NULL) { + struct page *p = alloc_page(GFP_KERNEL); + if (!p) { + set_current_state(TASK_INTERRUPTIBLE); + if (signalled() || kthread_should_stop()) { + set_current_state(TASK_RUNNING); + return -EINTR; + } + schedule_timeout(msecs_to_jiffies(500)); + } + rqstp->rq_pages[i] = p; + } + rqstp->rq_page_end = &rqstp->rq_pages[i]; + rqstp->rq_pages[i++] = NULL; /* this might be seen in nfs_read_actor */ + + /* Make arg->head point to first page and arg->pages point to rest */ + arg = &rqstp->rq_arg; + arg->head[0].iov_base = page_address(rqstp->rq_pages[0]); + arg->head[0].iov_len = PAGE_SIZE; + arg->pages = rqstp->rq_pages + 1; + arg->page_base = 0; + /* save at least one page for response */ + arg->page_len = (pages-2)*PAGE_SIZE; + arg->len = (pages-1)*PAGE_SIZE; + arg->tail[0].iov_len = 0; + return 0; +} + +static struct svc_xprt *svc_get_next_xprt(struct svc_rqst *rqstp, long timeout) +{ + struct svc_xprt *xprt; + struct svc_pool *pool = rqstp->rq_pool; + DECLARE_WAITQUEUE(wait, current); + long time_left; + + /* Normally we will wait up to 5 seconds for any required + * cache information to be provided. + */ + rqstp->rq_chandle.thread_wait = 5*HZ; + + spin_lock_bh(&pool->sp_lock); + xprt = svc_xprt_dequeue(pool); + if (xprt) { + rqstp->rq_xprt = xprt; + svc_xprt_get(xprt); + + /* As there is a shortage of threads and this request + * had to be queued, don't allow the thread to wait so + * long for cache updates. + */ + rqstp->rq_chandle.thread_wait = 1*HZ; + pool->sp_task_pending = 0; + } else { + if (pool->sp_task_pending) { + pool->sp_task_pending = 0; + spin_unlock_bh(&pool->sp_lock); + return ERR_PTR(-EAGAIN); + } + /* No data pending. Go to sleep */ + svc_thread_enqueue(pool, rqstp); + + /* + * We have to be able to interrupt this wait + * to bring down the daemons ... + */ + set_current_state(TASK_INTERRUPTIBLE); + + /* + * checking kthread_should_stop() here allows us to avoid + * locking and signalling when stopping kthreads that call + * svc_recv. If the thread has already been woken up, then + * we can exit here without sleeping. If not, then it + * it'll be woken up quickly during the schedule_timeout + */ + if (kthread_should_stop()) { + set_current_state(TASK_RUNNING); + spin_unlock_bh(&pool->sp_lock); + return ERR_PTR(-EINTR); + } + + add_wait_queue(&rqstp->rq_wait, &wait); + spin_unlock_bh(&pool->sp_lock); + + time_left = schedule_timeout(timeout); + + try_to_freeze(); + + spin_lock_bh(&pool->sp_lock); + remove_wait_queue(&rqstp->rq_wait, &wait); + if (!time_left) + pool->sp_stats.threads_timedout++; + + xprt = rqstp->rq_xprt; + if (!xprt) { + svc_thread_dequeue(pool, rqstp); + spin_unlock_bh(&pool->sp_lock); + dprintk("svc: server %p, no data yet\n", rqstp); + if (signalled() || kthread_should_stop()) + return ERR_PTR(-EINTR); + else + return ERR_PTR(-EAGAIN); + } + } + spin_unlock_bh(&pool->sp_lock); + return xprt; +} + +static void svc_add_new_temp_xprt(struct svc_serv *serv, struct svc_xprt *newxpt) +{ + spin_lock_bh(&serv->sv_lock); + set_bit(XPT_TEMP, &newxpt->xpt_flags); + list_add(&newxpt->xpt_list, &serv->sv_tempsocks); + serv->sv_tmpcnt++; + if (serv->sv_temptimer.function == NULL) { + /* setup timer to age temp transports */ + setup_timer(&serv->sv_temptimer, svc_age_temp_xprts, + (unsigned long)serv); + mod_timer(&serv->sv_temptimer, + jiffies + svc_conn_age_period * HZ); + } + spin_unlock_bh(&serv->sv_lock); + svc_xprt_received(newxpt); +} + +static int svc_handle_xprt(struct svc_rqst *rqstp, struct svc_xprt *xprt) +{ + struct svc_serv *serv = rqstp->rq_server; + int len = 0; + + if (test_bit(XPT_CLOSE, &xprt->xpt_flags)) { + dprintk("svc_recv: found XPT_CLOSE\n"); + svc_delete_xprt(xprt); + /* Leave XPT_BUSY set on the dead xprt: */ + return 0; + } + if (test_bit(XPT_LISTENER, &xprt->xpt_flags)) { + struct svc_xprt *newxpt; + /* + * We know this module_get will succeed because the + * listener holds a reference too + */ + __module_get(xprt->xpt_class->xcl_owner); + svc_check_conn_limits(xprt->xpt_server); + newxpt = xprt->xpt_ops->xpo_accept(xprt); + if (newxpt) + svc_add_new_temp_xprt(serv, newxpt); + else + module_put(xprt->xpt_class->xcl_owner); + } else if (xprt->xpt_ops->xpo_has_wspace(xprt)) { + /* XPT_DATA|XPT_DEFERRED case: */ + dprintk("svc: server %p, pool %u, transport %p, inuse=%d\n", + rqstp, rqstp->rq_pool->sp_id, xprt, + atomic_read(&xprt->xpt_ref.refcount)); + rqstp->rq_deferred = svc_deferred_dequeue(xprt); + if (rqstp->rq_deferred) + len = svc_deferred_recv(rqstp); + else + len = xprt->xpt_ops->xpo_recvfrom(rqstp); + dprintk("svc: got len=%d\n", len); + rqstp->rq_reserved = serv->sv_max_mesg; + atomic_add(rqstp->rq_reserved, &xprt->xpt_reserved); + } + /* clear XPT_BUSY: */ + svc_xprt_received(xprt); + return len; +} + +/* + * Receive the next request on any transport. This code is carefully + * organised not to touch any cachelines in the shared svc_serv + * structure, only cachelines in the local svc_pool. + */ +int svc_recv(struct svc_rqst *rqstp, long timeout) +{ + struct svc_xprt *xprt = NULL; + struct svc_serv *serv = rqstp->rq_server; + int len, err; + + dprintk("svc: server %p waiting for data (to = %ld)\n", + rqstp, timeout); + + if (rqstp->rq_xprt) + printk(KERN_ERR + "svc_recv: service %p, transport not NULL!\n", + rqstp); + if (waitqueue_active(&rqstp->rq_wait)) + printk(KERN_ERR + "svc_recv: service %p, wait queue active!\n", + rqstp); + + err = svc_alloc_arg(rqstp); + if (err) + return err; + + try_to_freeze(); + cond_resched(); + if (signalled() || kthread_should_stop()) + return -EINTR; + + xprt = svc_get_next_xprt(rqstp, timeout); + if (IS_ERR(xprt)) + return PTR_ERR(xprt); + + len = svc_handle_xprt(rqstp, xprt); + + /* No data, incomplete (TCP) read, or accept() */ + if (len <= 0) + goto out; + + clear_bit(XPT_OLD, &xprt->xpt_flags); + + rqstp->rq_secure = xprt->xpt_ops->xpo_secure_port(rqstp); + rqstp->rq_chandle.defer = svc_defer; + + if (serv->sv_stats) + serv->sv_stats->netcnt++; + return len; +out: + rqstp->rq_res.len = 0; + svc_xprt_release(rqstp); + return -EAGAIN; +} +EXPORT_SYMBOL_GPL(svc_recv); + +/* + * Drop request + */ +void svc_drop(struct svc_rqst *rqstp) +{ + dprintk("svc: xprt %p dropped request\n", rqstp->rq_xprt); + svc_xprt_release(rqstp); +} +EXPORT_SYMBOL_GPL(svc_drop); + +/* + * Return reply to client. + */ +int svc_send(struct svc_rqst *rqstp) +{ + struct svc_xprt *xprt; + int len; + struct xdr_buf *xb; + + xprt = rqstp->rq_xprt; + if (!xprt) + return -EFAULT; + + /* release the receive skb before sending the reply */ + rqstp->rq_xprt->xpt_ops->xpo_release_rqst(rqstp); + + /* calculate over-all length */ + xb = &rqstp->rq_res; + xb->len = xb->head[0].iov_len + + xb->page_len + + xb->tail[0].iov_len; + + /* Grab mutex to serialize outgoing data. */ + mutex_lock(&xprt->xpt_mutex); + if (test_bit(XPT_DEAD, &xprt->xpt_flags) + || test_bit(XPT_CLOSE, &xprt->xpt_flags)) + len = -ENOTCONN; + else + len = xprt->xpt_ops->xpo_sendto(rqstp); + mutex_unlock(&xprt->xpt_mutex); + rpc_wake_up(&xprt->xpt_bc_pending); + svc_xprt_release(rqstp); + + if (len == -ECONNREFUSED || len == -ENOTCONN || len == -EAGAIN) + return 0; + return len; +} + +/* + * Timer function to close old temporary transports, using + * a mark-and-sweep algorithm. + */ +static void svc_age_temp_xprts(unsigned long closure) +{ + struct svc_serv *serv = (struct svc_serv *)closure; + struct svc_xprt *xprt; + struct list_head *le, *next; + + dprintk("svc_age_temp_xprts\n"); + + if (!spin_trylock_bh(&serv->sv_lock)) { + /* busy, try again 1 sec later */ + dprintk("svc_age_temp_xprts: busy\n"); + mod_timer(&serv->sv_temptimer, jiffies + HZ); + return; + } + + list_for_each_safe(le, next, &serv->sv_tempsocks) { + xprt = list_entry(le, struct svc_xprt, xpt_list); + + /* First time through, just mark it OLD. Second time + * through, close it. */ + if (!test_and_set_bit(XPT_OLD, &xprt->xpt_flags)) + continue; + if (atomic_read(&xprt->xpt_ref.refcount) > 1 || + test_bit(XPT_BUSY, &xprt->xpt_flags)) + continue; + list_del_init(le); + set_bit(XPT_CLOSE, &xprt->xpt_flags); + set_bit(XPT_DETACHED, &xprt->xpt_flags); + dprintk("queuing xprt %p for closing\n", xprt); + + /* a thread will dequeue and close it soon */ + svc_xprt_enqueue(xprt); + } + spin_unlock_bh(&serv->sv_lock); + + mod_timer(&serv->sv_temptimer, jiffies + svc_conn_age_period * HZ); +} + +static void call_xpt_users(struct svc_xprt *xprt) +{ + struct svc_xpt_user *u; + + spin_lock(&xprt->xpt_lock); + while (!list_empty(&xprt->xpt_users)) { + u = list_first_entry(&xprt->xpt_users, struct svc_xpt_user, list); + list_del(&u->list); + u->callback(u); + } + spin_unlock(&xprt->xpt_lock); +} + +/* + * Remove a dead transport + */ +static void svc_delete_xprt(struct svc_xprt *xprt) +{ + struct svc_serv *serv = xprt->xpt_server; + struct svc_deferred_req *dr; + + /* Only do this once */ + if (test_and_set_bit(XPT_DEAD, &xprt->xpt_flags)) + BUG(); + + dprintk("svc: svc_delete_xprt(%p)\n", xprt); + xprt->xpt_ops->xpo_detach(xprt); + + spin_lock_bh(&serv->sv_lock); + if (!test_and_set_bit(XPT_DETACHED, &xprt->xpt_flags)) + list_del_init(&xprt->xpt_list); + WARN_ON_ONCE(!list_empty(&xprt->xpt_ready)); + if (test_bit(XPT_TEMP, &xprt->xpt_flags)) + serv->sv_tmpcnt--; + spin_unlock_bh(&serv->sv_lock); + + while ((dr = svc_deferred_dequeue(xprt)) != NULL) + kfree(dr); + + call_xpt_users(xprt); + svc_xprt_put(xprt); +} + +void svc_close_xprt(struct svc_xprt *xprt) +{ + set_bit(XPT_CLOSE, &xprt->xpt_flags); + if (test_and_set_bit(XPT_BUSY, &xprt->xpt_flags)) + /* someone else will have to effect the close */ + return; + /* + * We expect svc_close_xprt() to work even when no threads are + * running (e.g., while configuring the server before starting + * any threads), so if the transport isn't busy, we delete + * it ourself: + */ + svc_delete_xprt(xprt); +} +EXPORT_SYMBOL_GPL(svc_close_xprt); + +static int svc_close_list(struct svc_serv *serv, struct list_head *xprt_list, struct net *net) +{ + struct svc_xprt *xprt; + int ret = 0; + + spin_lock(&serv->sv_lock); + list_for_each_entry(xprt, xprt_list, xpt_list) { + if (xprt->xpt_net != net) + continue; + ret++; + set_bit(XPT_CLOSE, &xprt->xpt_flags); + svc_xprt_enqueue(xprt); + } + spin_unlock(&serv->sv_lock); + return ret; +} + +static struct svc_xprt *svc_dequeue_net(struct svc_serv *serv, struct net *net) +{ + struct svc_pool *pool; + struct svc_xprt *xprt; + struct svc_xprt *tmp; + int i; + + for (i = 0; i < serv->sv_nrpools; i++) { + pool = &serv->sv_pools[i]; + + spin_lock_bh(&pool->sp_lock); + list_for_each_entry_safe(xprt, tmp, &pool->sp_sockets, xpt_ready) { + if (xprt->xpt_net != net) + continue; + list_del_init(&xprt->xpt_ready); + spin_unlock_bh(&pool->sp_lock); + return xprt; + } + spin_unlock_bh(&pool->sp_lock); + } + return NULL; +} + +static void svc_clean_up_xprts(struct svc_serv *serv, struct net *net) +{ + struct svc_xprt *xprt; + + while ((xprt = svc_dequeue_net(serv, net))) { + set_bit(XPT_CLOSE, &xprt->xpt_flags); + svc_delete_xprt(xprt); + } +} + +/* + * Server threads may still be running (especially in the case where the + * service is still running in other network namespaces). + * + * So we shut down sockets the same way we would on a running server, by + * setting XPT_CLOSE, enqueuing, and letting a thread pick it up to do + * the close. In the case there are no such other threads, + * threads running, svc_clean_up_xprts() does a simple version of a + * server's main event loop, and in the case where there are other + * threads, we may need to wait a little while and then check again to + * see if they're done. + */ +void svc_close_net(struct svc_serv *serv, struct net *net) +{ + int delay = 0; + + while (svc_close_list(serv, &serv->sv_permsocks, net) + + svc_close_list(serv, &serv->sv_tempsocks, net)) { + + svc_clean_up_xprts(serv, net); + msleep(delay++); + } +} + +/* + * Handle defer and revisit of requests + */ + +static void svc_revisit(struct cache_deferred_req *dreq, int too_many) +{ + struct svc_deferred_req *dr = + container_of(dreq, struct svc_deferred_req, handle); + struct svc_xprt *xprt = dr->xprt; + + spin_lock(&xprt->xpt_lock); + set_bit(XPT_DEFERRED, &xprt->xpt_flags); + if (too_many || test_bit(XPT_DEAD, &xprt->xpt_flags)) { + spin_unlock(&xprt->xpt_lock); + dprintk("revisit canceled\n"); + svc_xprt_put(xprt); + kfree(dr); + return; + } + dprintk("revisit queued\n"); + dr->xprt = NULL; + list_add(&dr->handle.recent, &xprt->xpt_deferred); + spin_unlock(&xprt->xpt_lock); + svc_xprt_enqueue(xprt); + svc_xprt_put(xprt); +} + +/* + * Save the request off for later processing. The request buffer looks + * like this: + * + * <xprt-header><rpc-header><rpc-pagelist><rpc-tail> + * + * This code can only handle requests that consist of an xprt-header + * and rpc-header. + */ +static struct cache_deferred_req *svc_defer(struct cache_req *req) +{ + struct svc_rqst *rqstp = container_of(req, struct svc_rqst, rq_chandle); + struct svc_deferred_req *dr; + + if (rqstp->rq_arg.page_len || !rqstp->rq_usedeferral) + return NULL; /* if more than a page, give up FIXME */ + if (rqstp->rq_deferred) { + dr = rqstp->rq_deferred; + rqstp->rq_deferred = NULL; + } else { + size_t skip; + size_t size; + /* FIXME maybe discard if size too large */ + size = sizeof(struct svc_deferred_req) + rqstp->rq_arg.len; + dr = kmalloc(size, GFP_KERNEL); + if (dr == NULL) + return NULL; + + dr->handle.owner = rqstp->rq_server; + dr->prot = rqstp->rq_prot; + memcpy(&dr->addr, &rqstp->rq_addr, rqstp->rq_addrlen); + dr->addrlen = rqstp->rq_addrlen; + dr->daddr = rqstp->rq_daddr; + dr->argslen = rqstp->rq_arg.len >> 2; + dr->xprt_hlen = rqstp->rq_xprt_hlen; + + /* back up head to the start of the buffer and copy */ + skip = rqstp->rq_arg.len - rqstp->rq_arg.head[0].iov_len; + memcpy(dr->args, rqstp->rq_arg.head[0].iov_base - skip, + dr->argslen << 2); + } + svc_xprt_get(rqstp->rq_xprt); + dr->xprt = rqstp->rq_xprt; + rqstp->rq_dropme = true; + + dr->handle.revisit = svc_revisit; + return &dr->handle; +} + +/* + * recv data from a deferred request into an active one + */ +static int svc_deferred_recv(struct svc_rqst *rqstp) +{ + struct svc_deferred_req *dr = rqstp->rq_deferred; + + /* setup iov_base past transport header */ + rqstp->rq_arg.head[0].iov_base = dr->args + (dr->xprt_hlen>>2); + /* The iov_len does not include the transport header bytes */ + rqstp->rq_arg.head[0].iov_len = (dr->argslen<<2) - dr->xprt_hlen; + rqstp->rq_arg.page_len = 0; + /* The rq_arg.len includes the transport header bytes */ + rqstp->rq_arg.len = dr->argslen<<2; + rqstp->rq_prot = dr->prot; + memcpy(&rqstp->rq_addr, &dr->addr, dr->addrlen); + rqstp->rq_addrlen = dr->addrlen; + /* Save off transport header len in case we get deferred again */ + rqstp->rq_xprt_hlen = dr->xprt_hlen; + rqstp->rq_daddr = dr->daddr; + rqstp->rq_respages = rqstp->rq_pages; + return (dr->argslen<<2) - dr->xprt_hlen; +} + + +static struct svc_deferred_req *svc_deferred_dequeue(struct svc_xprt *xprt) +{ + struct svc_deferred_req *dr = NULL; + + if (!test_bit(XPT_DEFERRED, &xprt->xpt_flags)) + return NULL; + spin_lock(&xprt->xpt_lock); + if (!list_empty(&xprt->xpt_deferred)) { + dr = list_entry(xprt->xpt_deferred.next, + struct svc_deferred_req, + handle.recent); + list_del_init(&dr->handle.recent); + } else + clear_bit(XPT_DEFERRED, &xprt->xpt_flags); + spin_unlock(&xprt->xpt_lock); + return dr; +} + +/** + * svc_find_xprt - find an RPC transport instance + * @serv: pointer to svc_serv to search + * @xcl_name: C string containing transport's class name + * @net: owner net pointer + * @af: Address family of transport's local address + * @port: transport's IP port number + * + * Return the transport instance pointer for the endpoint accepting + * connections/peer traffic from the specified transport class, + * address family and port. + * + * Specifying 0 for the address family or port is effectively a + * wild-card, and will result in matching the first transport in the + * service's list that has a matching class name. + */ +struct svc_xprt *svc_find_xprt(struct svc_serv *serv, const char *xcl_name, + struct net *net, const sa_family_t af, + const unsigned short port) +{ + struct svc_xprt *xprt; + struct svc_xprt *found = NULL; + + /* Sanity check the args */ + if (serv == NULL || xcl_name == NULL) + return found; + + spin_lock_bh(&serv->sv_lock); + list_for_each_entry(xprt, &serv->sv_permsocks, xpt_list) { + if (xprt->xpt_net != net) + continue; + if (strcmp(xprt->xpt_class->xcl_name, xcl_name)) + continue; + if (af != AF_UNSPEC && af != xprt->xpt_local.ss_family) + continue; + if (port != 0 && port != svc_xprt_local_port(xprt)) + continue; + found = xprt; + svc_xprt_get(xprt); + break; + } + spin_unlock_bh(&serv->sv_lock); + return found; +} +EXPORT_SYMBOL_GPL(svc_find_xprt); + +static int svc_one_xprt_name(const struct svc_xprt *xprt, + char *pos, int remaining) +{ + int len; + + len = snprintf(pos, remaining, "%s %u\n", + xprt->xpt_class->xcl_name, + svc_xprt_local_port(xprt)); + if (len >= remaining) + return -ENAMETOOLONG; + return len; +} + +/** + * svc_xprt_names - format a buffer with a list of transport names + * @serv: pointer to an RPC service + * @buf: pointer to a buffer to be filled in + * @buflen: length of buffer to be filled in + * + * Fills in @buf with a string containing a list of transport names, + * each name terminated with '\n'. + * + * Returns positive length of the filled-in string on success; otherwise + * a negative errno value is returned if an error occurs. + */ +int svc_xprt_names(struct svc_serv *serv, char *buf, const int buflen) +{ + struct svc_xprt *xprt; + int len, totlen; + char *pos; + + /* Sanity check args */ + if (!serv) + return 0; + + spin_lock_bh(&serv->sv_lock); + + pos = buf; + totlen = 0; + list_for_each_entry(xprt, &serv->sv_permsocks, xpt_list) { + len = svc_one_xprt_name(xprt, pos, buflen - totlen); + if (len < 0) { + *buf = '\0'; + totlen = len; + } + if (len <= 0) + break; + + pos += len; + totlen += len; + } + + spin_unlock_bh(&serv->sv_lock); + return totlen; +} +EXPORT_SYMBOL_GPL(svc_xprt_names); + + +/*----------------------------------------------------------------------------*/ + +static void *svc_pool_stats_start(struct seq_file *m, loff_t *pos) +{ + unsigned int pidx = (unsigned int)*pos; + struct svc_serv *serv = m->private; + + dprintk("svc_pool_stats_start, *pidx=%u\n", pidx); + + if (!pidx) + return SEQ_START_TOKEN; + return (pidx > serv->sv_nrpools ? NULL : &serv->sv_pools[pidx-1]); +} + +static void *svc_pool_stats_next(struct seq_file *m, void *p, loff_t *pos) +{ + struct svc_pool *pool = p; + struct svc_serv *serv = m->private; + + dprintk("svc_pool_stats_next, *pos=%llu\n", *pos); + + if (p == SEQ_START_TOKEN) { + pool = &serv->sv_pools[0]; + } else { + unsigned int pidx = (pool - &serv->sv_pools[0]); + if (pidx < serv->sv_nrpools-1) + pool = &serv->sv_pools[pidx+1]; + else + pool = NULL; + } + ++*pos; + return pool; +} + +static void svc_pool_stats_stop(struct seq_file *m, void *p) +{ +} + +static int svc_pool_stats_show(struct seq_file *m, void *p) +{ + struct svc_pool *pool = p; + + if (p == SEQ_START_TOKEN) { + seq_puts(m, "# pool packets-arrived sockets-enqueued threads-woken threads-timedout\n"); + return 0; + } + + seq_printf(m, "%u %lu %lu %lu %lu\n", + pool->sp_id, + pool->sp_stats.packets, + pool->sp_stats.sockets_queued, + pool->sp_stats.threads_woken, + pool->sp_stats.threads_timedout); + + return 0; +} + +static const struct seq_operations svc_pool_stats_seq_ops = { + .start = svc_pool_stats_start, + .next = svc_pool_stats_next, + .stop = svc_pool_stats_stop, + .show = svc_pool_stats_show, +}; + +int svc_pool_stats_open(struct svc_serv *serv, struct file *file) +{ + int err; + + err = seq_open(file, &svc_pool_stats_seq_ops); + if (!err) + ((struct seq_file *) file->private_data)->private = serv; + return err; +} +EXPORT_SYMBOL(svc_pool_stats_open); + +/*----------------------------------------------------------------------------*/ diff --git a/net/sunrpc/svcauth.c b/net/sunrpc/svcauth.c index dda4f0c6351..79c0f3459b5 100644 --- a/net/sunrpc/svcauth.c +++ b/net/sunrpc/svcauth.c @@ -2,7 +2,7 @@ * linux/net/sunrpc/svcauth.c * * The generic interface for RPC authentication on the server side. - * + * * Copyright (C) 1995, 1996 Olaf Kirch <okir@monad.swb.de> * * CHANGES @@ -10,7 +10,6 @@ */ #include <linux/types.h> -#include <linux/sched.h> #include <linux/module.h> #include <linux/sunrpc/types.h> #include <linux/sunrpc/xdr.h> @@ -35,37 +34,41 @@ static struct auth_ops *authtab[RPC_AUTH_MAXFLAVOR] = { }; int -svc_authenticate(struct svc_rqst *rqstp, u32 *authp) +svc_authenticate(struct svc_rqst *rqstp, __be32 *authp) { rpc_authflavor_t flavor; struct auth_ops *aops; *authp = rpc_auth_ok; - flavor = ntohl(svc_getu32(&rqstp->rq_arg.head[0])); + flavor = svc_getnl(&rqstp->rq_arg.head[0]); dprintk("svc: svc_authenticate (%d)\n", flavor); spin_lock(&authtab_lock); - if (flavor >= RPC_AUTH_MAXFLAVOR || !(aops = authtab[flavor]) - || !try_module_get(aops->owner)) { + if (flavor >= RPC_AUTH_MAXFLAVOR || !(aops = authtab[flavor]) || + !try_module_get(aops->owner)) { spin_unlock(&authtab_lock); *authp = rpc_autherr_badcred; return SVC_DENIED; } spin_unlock(&authtab_lock); + rqstp->rq_auth_slack = 0; + rqstp->rq_authop = aops; return aops->accept(rqstp, authp); } +EXPORT_SYMBOL_GPL(svc_authenticate); int svc_set_client(struct svc_rqst *rqstp) { return rqstp->rq_authop->set_client(rqstp); } +EXPORT_SYMBOL_GPL(svc_set_client); /* A request, which was authenticated, has now executed. - * Time to finalise the the credentials and verifier + * Time to finalise the credentials and verifier * and release and resources */ int svc_authorise(struct svc_rqst *rqstp) @@ -74,7 +77,7 @@ int svc_authorise(struct svc_rqst *rqstp) int rv = 0; rqstp->rq_authop = NULL; - + if (aops) { rv = aops->release(rqstp); module_put(aops->owner); @@ -94,6 +97,7 @@ svc_auth_register(rpc_authflavor_t flavor, struct auth_ops *aops) spin_unlock(&authtab_lock); return rv; } +EXPORT_SYMBOL_GPL(svc_auth_register); void svc_auth_unregister(rpc_authflavor_t flavor) @@ -103,115 +107,60 @@ svc_auth_unregister(rpc_authflavor_t flavor) authtab[flavor] = NULL; spin_unlock(&authtab_lock); } -EXPORT_SYMBOL(svc_auth_unregister); +EXPORT_SYMBOL_GPL(svc_auth_unregister); /************************************************** - * cache for domain name to auth_domain - * Entries are only added by flavours which will normally - * have a structure that 'inherits' from auth_domain. - * e.g. when an IP -> domainname is given to auth_unix, - * and the domain name doesn't exist, it will create a - * auth_unix_domain and add it to this hash table. - * If it finds the name does exist, but isn't AUTH_UNIX, - * it will complain. + * 'auth_domains' are stored in a hash table indexed by name. + * When the last reference to an 'auth_domain' is dropped, + * the object is unhashed and freed. + * If auth_domain_lookup fails to find an entry, it will return + * it's second argument 'new'. If this is non-null, it will + * have been atomically linked into the table. */ -/* - * Auth auth_domain cache is somewhat different to other caches, - * largely because the entries are possibly of different types: - * each auth flavour has it's own type. - * One consequence of this that DefineCacheLookup cannot - * allocate a new structure as it cannot know the size. - * Notice that the "INIT" code fragment is quite different - * from other caches. When auth_domain_lookup might be - * creating a new domain, the new domain is passed in - * complete and it is used as-is rather than being copied into - * another structure. - */ #define DN_HASHBITS 6 #define DN_HASHMAX (1<<DN_HASHBITS) -#define DN_HASHMASK (DN_HASHMAX-1) - -static struct cache_head *auth_domain_table[DN_HASHMAX]; -static void auth_domain_drop(struct cache_head *item, struct cache_detail *cd) -{ - struct auth_domain *dom = container_of(item, struct auth_domain, h); - if (cache_put(item,cd)) - authtab[dom->flavour]->domain_release(dom); -} - - -struct cache_detail auth_domain_cache = { - .owner = THIS_MODULE, - .hash_size = DN_HASHMAX, - .hash_table = auth_domain_table, - .name = "auth.domain", - .cache_put = auth_domain_drop, -}; +static struct hlist_head auth_domain_table[DN_HASHMAX]; +static spinlock_t auth_domain_lock = + __SPIN_LOCK_UNLOCKED(auth_domain_lock); void auth_domain_put(struct auth_domain *dom) { - auth_domain_drop(&dom->h, &auth_domain_cache); -} - -static inline int auth_domain_hash(struct auth_domain *item) -{ - return hash_str(item->name, DN_HASHBITS); -} -static inline int auth_domain_match(struct auth_domain *tmp, struct auth_domain *item) -{ - return strcmp(tmp->name, item->name) == 0; + if (atomic_dec_and_lock(&dom->ref.refcount, &auth_domain_lock)) { + hlist_del(&dom->hash); + dom->flavour->domain_release(dom); + spin_unlock(&auth_domain_lock); + } } +EXPORT_SYMBOL_GPL(auth_domain_put); struct auth_domain * -auth_domain_lookup(struct auth_domain *item, int set) +auth_domain_lookup(char *name, struct auth_domain *new) { - struct auth_domain *tmp = NULL; - struct cache_head **hp, **head; - head = &auth_domain_cache.hash_table[auth_domain_hash(item)]; - - if (set) - write_lock(&auth_domain_cache.hash_lock); - else - read_lock(&auth_domain_cache.hash_lock); - for (hp=head; *hp != NULL; hp = &tmp->h.next) { - tmp = container_of(*hp, struct auth_domain, h); - if (!auth_domain_match(tmp, item)) - continue; - if (!set) { - cache_get(&tmp->h); - goto out_noset; + struct auth_domain *hp; + struct hlist_head *head; + + head = &auth_domain_table[hash_str(name, DN_HASHBITS)]; + + spin_lock(&auth_domain_lock); + + hlist_for_each_entry(hp, head, hash) { + if (strcmp(hp->name, name)==0) { + kref_get(&hp->ref); + spin_unlock(&auth_domain_lock); + return hp; } - *hp = tmp->h.next; - tmp->h.next = NULL; - auth_domain_drop(&tmp->h, &auth_domain_cache); - goto out_set; } - /* Didn't find anything */ - if (!set) - goto out_nada; - auth_domain_cache.entries++; -out_set: - item->h.next = *head; - *head = &item->h; - cache_get(&item->h); - write_unlock(&auth_domain_cache.hash_lock); - cache_fresh(&auth_domain_cache, &item->h, item->h.expiry_time); - cache_get(&item->h); - return item; -out_nada: - tmp = NULL; -out_noset: - read_unlock(&auth_domain_cache.hash_lock); - return tmp; + if (new) + hlist_add_head(&new->hash, head); + spin_unlock(&auth_domain_lock); + return new; } +EXPORT_SYMBOL_GPL(auth_domain_lookup); struct auth_domain *auth_domain_find(char *name) { - struct auth_domain *rv, ad; - - ad.name = name; - rv = auth_domain_lookup(&ad, 0); - return rv; + return auth_domain_lookup(name, NULL); } +EXPORT_SYMBOL_GPL(auth_domain_find); diff --git a/net/sunrpc/svcauth_unix.c b/net/sunrpc/svcauth_unix.c index cac2e774dd8..621ca7b4a15 100644 --- a/net/sunrpc/svcauth_unix.c +++ b/net/sunrpc/svcauth_unix.c @@ -5,14 +5,22 @@ #include <linux/sunrpc/xdr.h> #include <linux/sunrpc/svcsock.h> #include <linux/sunrpc/svcauth.h> +#include <linux/sunrpc/gss_api.h> +#include <linux/sunrpc/addr.h> #include <linux/err.h> #include <linux/seq_file.h> #include <linux/hash.h> #include <linux/string.h> - +#include <linux/slab.h> +#include <net/sock.h> +#include <net/ipv6.h> +#include <linux/kernel.h> +#include <linux/user_namespace.h> #define RPCDBG_FACILITY RPCDBG_AUTH +#include "netns.h" + /* * AUTHUNIX and AUTHNULL credentials are both handled here. * AUTHNULL is treated just like AUTHUNIX except that the uid/gid @@ -23,46 +31,11 @@ struct unix_domain { struct auth_domain h; - int addr_changes; /* other stuff later */ }; -struct auth_domain *unix_domain_find(char *name) -{ - struct auth_domain *rv, ud; - struct unix_domain *new; - - ud.name = name; - - rv = auth_domain_lookup(&ud, 0); - - foundit: - if (rv && rv->flavour != RPC_AUTH_UNIX) { - auth_domain_put(rv); - return NULL; - } - if (rv) - return rv; - - new = kmalloc(sizeof(*new), GFP_KERNEL); - if (new == NULL) - return NULL; - cache_init(&new->h.h); - new->h.name = kstrdup(name, GFP_KERNEL); - new->h.flavour = RPC_AUTH_UNIX; - new->addr_changes = 0; - new->h.h.expiry_time = NEVER; - - rv = auth_domain_lookup(&new->h, 2); - if (rv == &new->h) { - if (atomic_dec_and_test(&new->h.h.refcnt)) BUG(); - } else { - auth_domain_put(&new->h); - goto foundit; - } - - return rv; -} +extern struct auth_ops svcauth_null; +extern struct auth_ops svcauth_unix; static void svcauth_unix_domain_release(struct auth_domain *dom) { @@ -72,6 +45,39 @@ static void svcauth_unix_domain_release(struct auth_domain *dom) kfree(ud); } +struct auth_domain *unix_domain_find(char *name) +{ + struct auth_domain *rv; + struct unix_domain *new = NULL; + + rv = auth_domain_lookup(name, NULL); + while(1) { + if (rv) { + if (new && rv != &new->h) + svcauth_unix_domain_release(&new->h); + + if (rv->flavour != &svcauth_unix) { + auth_domain_put(rv); + return NULL; + } + return rv; + } + + new = kmalloc(sizeof(*new), GFP_KERNEL); + if (new == NULL) + return NULL; + kref_init(&new->h.ref); + new->h.name = kstrdup(name, GFP_KERNEL); + if (new->h.name == NULL) { + kfree(new); + return NULL; + } + new->h.flavour = &svcauth_unix; + rv = auth_domain_lookup(name, &new->h); + } +} +EXPORT_SYMBOL_GPL(unix_domain_find); + /************************************************** * cache for IP address to unix_domain @@ -79,70 +85,80 @@ static void svcauth_unix_domain_release(struct auth_domain *dom) */ #define IP_HASHBITS 8 #define IP_HASHMAX (1<<IP_HASHBITS) -#define IP_HASHMASK (IP_HASHMAX-1) struct ip_map { struct cache_head h; char m_class[8]; /* e.g. "nfsd" */ - struct in_addr m_addr; + struct in6_addr m_addr; struct unix_domain *m_client; - int m_add_change; }; -static struct cache_head *ip_table[IP_HASHMAX]; -static void ip_map_put(struct cache_head *item, struct cache_detail *cd) +static void ip_map_put(struct kref *kref) { + struct cache_head *item = container_of(kref, struct cache_head, ref); struct ip_map *im = container_of(item, struct ip_map,h); - if (cache_put(item, cd)) { - if (test_bit(CACHE_VALID, &item->flags) && - !test_bit(CACHE_NEGATIVE, &item->flags)) - auth_domain_put(&im->m_client->h); - kfree(im); - } + + if (test_bit(CACHE_VALID, &item->flags) && + !test_bit(CACHE_NEGATIVE, &item->flags)) + auth_domain_put(&im->m_client->h); + kfree(im); } -static inline int ip_map_hash(struct ip_map *item) +static inline int hash_ip6(const struct in6_addr *ip) { - return hash_str(item->m_class, IP_HASHBITS) ^ - hash_long((unsigned long)item->m_addr.s_addr, IP_HASHBITS); + return hash_32(ipv6_addr_hash(ip), IP_HASHBITS); } -static inline int ip_map_match(struct ip_map *item, struct ip_map *tmp) +static int ip_map_match(struct cache_head *corig, struct cache_head *cnew) { - return strcmp(tmp->m_class, item->m_class) == 0 - && tmp->m_addr.s_addr == item->m_addr.s_addr; + struct ip_map *orig = container_of(corig, struct ip_map, h); + struct ip_map *new = container_of(cnew, struct ip_map, h); + return strcmp(orig->m_class, new->m_class) == 0 && + ipv6_addr_equal(&orig->m_addr, &new->m_addr); } -static inline void ip_map_init(struct ip_map *new, struct ip_map *item) +static void ip_map_init(struct cache_head *cnew, struct cache_head *citem) { + struct ip_map *new = container_of(cnew, struct ip_map, h); + struct ip_map *item = container_of(citem, struct ip_map, h); + strcpy(new->m_class, item->m_class); - new->m_addr.s_addr = item->m_addr.s_addr; + new->m_addr = item->m_addr; } -static inline void ip_map_update(struct ip_map *new, struct ip_map *item) +static void update(struct cache_head *cnew, struct cache_head *citem) { - cache_get(&item->m_client->h.h); + struct ip_map *new = container_of(cnew, struct ip_map, h); + struct ip_map *item = container_of(citem, struct ip_map, h); + + kref_get(&item->m_client->h.ref); new->m_client = item->m_client; - new->m_add_change = item->m_add_change; +} +static struct cache_head *ip_map_alloc(void) +{ + struct ip_map *i = kmalloc(sizeof(*i), GFP_KERNEL); + if (i) + return &i->h; + else + return NULL; } static void ip_map_request(struct cache_detail *cd, struct cache_head *h, char **bpp, int *blen) { - char text_addr[20]; + char text_addr[40]; struct ip_map *im = container_of(h, struct ip_map, h); - __u32 addr = im->m_addr.s_addr; - - snprintf(text_addr, 20, "%u.%u.%u.%u", - ntohl(addr) >> 24 & 0xff, - ntohl(addr) >> 16 & 0xff, - ntohl(addr) >> 8 & 0xff, - ntohl(addr) >> 0 & 0xff); + if (ipv6_addr_v4mapped(&(im->m_addr))) { + snprintf(text_addr, 20, "%pI4", &im->m_addr.s6_addr32[3]); + } else { + snprintf(text_addr, 40, "%pI6", &im->m_addr); + } qword_add(bpp, blen, im->m_class); qword_add(bpp, blen, text_addr); (*bpp)[-1] = '\n'; } -static struct ip_map *ip_map_lookup(struct ip_map *, int); +static struct ip_map *__ip_map_lookup(struct cache_detail *cd, char *class, struct in6_addr *addr); +static int __ip_map_update(struct cache_detail *cd, struct ip_map *ipm, struct unix_domain *udom, time_t expiry); static int ip_map_parse(struct cache_detail *cd, char *mesg, int mlen) @@ -152,9 +168,16 @@ static int ip_map_parse(struct cache_detail *cd, * for scratch: */ char *buf = mesg; int len; - int b1,b2,b3,b4; - char c; - struct ip_map ipm, *ipmp; + char class[8]; + union { + struct sockaddr sa; + struct sockaddr_in s4; + struct sockaddr_in6 s6; + } address; + struct sockaddr_in6 sin6; + int err; + + struct ip_map *ipmp; struct auth_domain *dom; time_t expiry; @@ -163,16 +186,31 @@ static int ip_map_parse(struct cache_detail *cd, mesg[mlen-1] = 0; /* class */ - len = qword_get(&mesg, ipm.m_class, sizeof(ipm.m_class)); + len = qword_get(&mesg, class, sizeof(class)); if (len <= 0) return -EINVAL; /* ip address */ len = qword_get(&mesg, buf, mlen); if (len <= 0) return -EINVAL; - if (sscanf(buf, "%u.%u.%u.%u%c", &b1, &b2, &b3, &b4, &c) != 4) + if (rpc_pton(cd->net, buf, len, &address.sa, sizeof(address)) == 0) + return -EINVAL; + switch (address.sa.sa_family) { + case AF_INET: + /* Form a mapped IPv4 address in sin6 */ + sin6.sin6_family = AF_INET6; + ipv6_addr_set_v4mapped(address.s4.sin_addr.s_addr, + &sin6.sin6_addr); + break; +#if IS_ENABLED(CONFIG_IPV6) + case AF_INET6: + memcpy(&sin6, &address.s6, sizeof(sin6)); + break; +#endif + default: return -EINVAL; - + } + expiry = get_expiry(&mesg); if (expiry ==0) return -EINVAL; @@ -188,25 +226,20 @@ static int ip_map_parse(struct cache_detail *cd, } else dom = NULL; - ipm.m_addr.s_addr = - htonl((((((b1<<8)|b2)<<8)|b3)<<8)|b4); - ipm.h.flags = 0; - if (dom) { - ipm.m_client = container_of(dom, struct unix_domain, h); - ipm.m_add_change = ipm.m_client->addr_changes; + /* IPv6 scope IDs are ignored for now */ + ipmp = __ip_map_lookup(cd, class, &sin6.sin6_addr); + if (ipmp) { + err = __ip_map_update(cd, ipmp, + container_of(dom, struct unix_domain, h), + expiry); } else - set_bit(CACHE_NEGATIVE, &ipm.h.flags); - ipm.h.expiry_time = expiry; + err = -ENOMEM; - ipmp = ip_map_lookup(&ipm, 1); - if (ipmp) - ip_map_put(&ipmp->h, &ip_map_cache); if (dom) auth_domain_put(dom); - if (!ipmp) - return -ENOMEM; + cache_flush(); - return 0; + return err; } static int ip_map_show(struct seq_file *m, @@ -214,7 +247,7 @@ static int ip_map_show(struct seq_file *m, struct cache_head *h) { struct ip_map *im; - struct in_addr addr; + struct in6_addr addr; char *dom = "-no-domain-"; if (h == NULL) { @@ -225,172 +258,506 @@ static int ip_map_show(struct seq_file *m, /* class addr domain */ addr = im->m_addr; - if (test_bit(CACHE_VALID, &h->flags) && + if (test_bit(CACHE_VALID, &h->flags) && !test_bit(CACHE_NEGATIVE, &h->flags)) dom = im->m_client->h.name; - seq_printf(m, "%s %d.%d.%d.%d %s\n", - im->m_class, - htonl(addr.s_addr) >> 24 & 0xff, - htonl(addr.s_addr) >> 16 & 0xff, - htonl(addr.s_addr) >> 8 & 0xff, - htonl(addr.s_addr) >> 0 & 0xff, - dom - ); + if (ipv6_addr_v4mapped(&addr)) { + seq_printf(m, "%s %pI4 %s\n", + im->m_class, &addr.s6_addr32[3], dom); + } else { + seq_printf(m, "%s %pI6 %s\n", im->m_class, &addr, dom); + } return 0; } - -struct cache_detail ip_map_cache = { - .owner = THIS_MODULE, - .hash_size = IP_HASHMAX, - .hash_table = ip_table, - .name = "auth.unix.ip", - .cache_put = ip_map_put, - .cache_request = ip_map_request, - .cache_parse = ip_map_parse, - .cache_show = ip_map_show, -}; -static DefineSimpleCacheLookup(ip_map, 0) +static struct ip_map *__ip_map_lookup(struct cache_detail *cd, char *class, + struct in6_addr *addr) +{ + struct ip_map ip; + struct cache_head *ch; + + strcpy(ip.m_class, class); + ip.m_addr = *addr; + ch = sunrpc_cache_lookup(cd, &ip.h, + hash_str(class, IP_HASHBITS) ^ + hash_ip6(addr)); + + if (ch) + return container_of(ch, struct ip_map, h); + else + return NULL; +} + +static inline struct ip_map *ip_map_lookup(struct net *net, char *class, + struct in6_addr *addr) +{ + struct sunrpc_net *sn; + sn = net_generic(net, sunrpc_net_id); + return __ip_map_lookup(sn->ip_map_cache, class, addr); +} -int auth_unix_add_addr(struct in_addr addr, struct auth_domain *dom) +static int __ip_map_update(struct cache_detail *cd, struct ip_map *ipm, + struct unix_domain *udom, time_t expiry) { - struct unix_domain *udom; - struct ip_map ip, *ipmp; + struct ip_map ip; + struct cache_head *ch; - if (dom->flavour != RPC_AUTH_UNIX) - return -EINVAL; - udom = container_of(dom, struct unix_domain, h); - strcpy(ip.m_class, "nfsd"); - ip.m_addr = addr; ip.m_client = udom; - ip.m_add_change = udom->addr_changes+1; ip.h.flags = 0; - ip.h.expiry_time = NEVER; - - ipmp = ip_map_lookup(&ip, 1); - - if (ipmp) { - ip_map_put(&ipmp->h, &ip_map_cache); - return 0; - } else + if (!udom) + set_bit(CACHE_NEGATIVE, &ip.h.flags); + ip.h.expiry_time = expiry; + ch = sunrpc_cache_update(cd, &ip.h, &ipm->h, + hash_str(ipm->m_class, IP_HASHBITS) ^ + hash_ip6(&ipm->m_addr)); + if (!ch) return -ENOMEM; + cache_put(ch, cd); + return 0; } -int auth_unix_forget_old(struct auth_domain *dom) +static inline int ip_map_update(struct net *net, struct ip_map *ipm, + struct unix_domain *udom, time_t expiry) { - struct unix_domain *udom; - - if (dom->flavour != RPC_AUTH_UNIX) - return -EINVAL; - udom = container_of(dom, struct unix_domain, h); - udom->addr_changes++; - return 0; + struct sunrpc_net *sn; + + sn = net_generic(net, sunrpc_net_id); + return __ip_map_update(sn->ip_map_cache, ipm, udom, expiry); } -struct auth_domain *auth_unix_lookup(struct in_addr addr) +void svcauth_unix_purge(struct net *net) { - struct ip_map key, *ipm; - struct auth_domain *rv; + struct sunrpc_net *sn; - strcpy(key.m_class, "nfsd"); - key.m_addr = addr; + sn = net_generic(net, sunrpc_net_id); + cache_purge(sn->ip_map_cache); +} +EXPORT_SYMBOL_GPL(svcauth_unix_purge); - ipm = ip_map_lookup(&key, 0); +static inline struct ip_map * +ip_map_cached_get(struct svc_xprt *xprt) +{ + struct ip_map *ipm = NULL; + struct sunrpc_net *sn; + + if (test_bit(XPT_CACHE_AUTH, &xprt->xpt_flags)) { + spin_lock(&xprt->xpt_lock); + ipm = xprt->xpt_auth_cache; + if (ipm != NULL) { + sn = net_generic(xprt->xpt_net, sunrpc_net_id); + if (cache_is_expired(sn->ip_map_cache, &ipm->h)) { + /* + * The entry has been invalidated since it was + * remembered, e.g. by a second mount from the + * same IP address. + */ + xprt->xpt_auth_cache = NULL; + spin_unlock(&xprt->xpt_lock); + cache_put(&ipm->h, sn->ip_map_cache); + return NULL; + } + cache_get(&ipm->h); + } + spin_unlock(&xprt->xpt_lock); + } + return ipm; +} - if (!ipm) - return NULL; - if (cache_check(&ip_map_cache, &ipm->h, NULL)) +static inline void +ip_map_cached_put(struct svc_xprt *xprt, struct ip_map *ipm) +{ + if (test_bit(XPT_CACHE_AUTH, &xprt->xpt_flags)) { + spin_lock(&xprt->xpt_lock); + if (xprt->xpt_auth_cache == NULL) { + /* newly cached, keep the reference */ + xprt->xpt_auth_cache = ipm; + ipm = NULL; + } + spin_unlock(&xprt->xpt_lock); + } + if (ipm) { + struct sunrpc_net *sn; + + sn = net_generic(xprt->xpt_net, sunrpc_net_id); + cache_put(&ipm->h, sn->ip_map_cache); + } +} + +void +svcauth_unix_info_release(struct svc_xprt *xpt) +{ + struct ip_map *ipm; + + ipm = xpt->xpt_auth_cache; + if (ipm != NULL) { + struct sunrpc_net *sn; + + sn = net_generic(xpt->xpt_net, sunrpc_net_id); + cache_put(&ipm->h, sn->ip_map_cache); + } +} + +/**************************************************************************** + * auth.unix.gid cache + * simple cache to map a UID to a list of GIDs + * because AUTH_UNIX aka AUTH_SYS has a max of 16 + */ +#define GID_HASHBITS 8 +#define GID_HASHMAX (1<<GID_HASHBITS) + +struct unix_gid { + struct cache_head h; + kuid_t uid; + struct group_info *gi; +}; + +static int unix_gid_hash(kuid_t uid) +{ + return hash_long(from_kuid(&init_user_ns, uid), GID_HASHBITS); +} + +static void unix_gid_put(struct kref *kref) +{ + struct cache_head *item = container_of(kref, struct cache_head, ref); + struct unix_gid *ug = container_of(item, struct unix_gid, h); + if (test_bit(CACHE_VALID, &item->flags) && + !test_bit(CACHE_NEGATIVE, &item->flags)) + put_group_info(ug->gi); + kfree(ug); +} + +static int unix_gid_match(struct cache_head *corig, struct cache_head *cnew) +{ + struct unix_gid *orig = container_of(corig, struct unix_gid, h); + struct unix_gid *new = container_of(cnew, struct unix_gid, h); + return uid_eq(orig->uid, new->uid); +} +static void unix_gid_init(struct cache_head *cnew, struct cache_head *citem) +{ + struct unix_gid *new = container_of(cnew, struct unix_gid, h); + struct unix_gid *item = container_of(citem, struct unix_gid, h); + new->uid = item->uid; +} +static void unix_gid_update(struct cache_head *cnew, struct cache_head *citem) +{ + struct unix_gid *new = container_of(cnew, struct unix_gid, h); + struct unix_gid *item = container_of(citem, struct unix_gid, h); + + get_group_info(item->gi); + new->gi = item->gi; +} +static struct cache_head *unix_gid_alloc(void) +{ + struct unix_gid *g = kmalloc(sizeof(*g), GFP_KERNEL); + if (g) + return &g->h; + else return NULL; +} - if ((ipm->m_client->addr_changes - ipm->m_add_change) >0) { - if (test_and_set_bit(CACHE_NEGATIVE, &ipm->h.flags) == 0) - auth_domain_put(&ipm->m_client->h); - rv = NULL; - } else { - rv = &ipm->m_client->h; - cache_get(&rv->h); +static void unix_gid_request(struct cache_detail *cd, + struct cache_head *h, + char **bpp, int *blen) +{ + char tuid[20]; + struct unix_gid *ug = container_of(h, struct unix_gid, h); + + snprintf(tuid, 20, "%u", from_kuid(&init_user_ns, ug->uid)); + qword_add(bpp, blen, tuid); + (*bpp)[-1] = '\n'; +} + +static struct unix_gid *unix_gid_lookup(struct cache_detail *cd, kuid_t uid); + +static int unix_gid_parse(struct cache_detail *cd, + char *mesg, int mlen) +{ + /* uid expiry Ngid gid0 gid1 ... gidN-1 */ + int id; + kuid_t uid; + int gids; + int rv; + int i; + int err; + time_t expiry; + struct unix_gid ug, *ugp; + + if (mesg[mlen - 1] != '\n') + return -EINVAL; + mesg[mlen-1] = 0; + + rv = get_int(&mesg, &id); + if (rv) + return -EINVAL; + uid = make_kuid(&init_user_ns, id); + ug.uid = uid; + + expiry = get_expiry(&mesg); + if (expiry == 0) + return -EINVAL; + + rv = get_int(&mesg, &gids); + if (rv || gids < 0 || gids > 8192) + return -EINVAL; + + ug.gi = groups_alloc(gids); + if (!ug.gi) + return -ENOMEM; + + for (i = 0 ; i < gids ; i++) { + int gid; + kgid_t kgid; + rv = get_int(&mesg, &gid); + err = -EINVAL; + if (rv) + goto out; + kgid = make_kgid(&init_user_ns, gid); + if (!gid_valid(kgid)) + goto out; + GROUP_AT(ug.gi, i) = kgid; } - ip_map_put(&ipm->h, &ip_map_cache); - return rv; + + ugp = unix_gid_lookup(cd, uid); + if (ugp) { + struct cache_head *ch; + ug.h.flags = 0; + ug.h.expiry_time = expiry; + ch = sunrpc_cache_update(cd, + &ug.h, &ugp->h, + unix_gid_hash(uid)); + if (!ch) + err = -ENOMEM; + else { + err = 0; + cache_put(ch, cd); + } + } else + err = -ENOMEM; + out: + if (ug.gi) + put_group_info(ug.gi); + return err; } -void svcauth_unix_purge(void) +static int unix_gid_show(struct seq_file *m, + struct cache_detail *cd, + struct cache_head *h) { - cache_purge(&ip_map_cache); - cache_purge(&auth_domain_cache); + struct user_namespace *user_ns = &init_user_ns; + struct unix_gid *ug; + int i; + int glen; + + if (h == NULL) { + seq_puts(m, "#uid cnt: gids...\n"); + return 0; + } + ug = container_of(h, struct unix_gid, h); + if (test_bit(CACHE_VALID, &h->flags) && + !test_bit(CACHE_NEGATIVE, &h->flags)) + glen = ug->gi->ngroups; + else + glen = 0; + + seq_printf(m, "%u %d:", from_kuid_munged(user_ns, ug->uid), glen); + for (i = 0; i < glen; i++) + seq_printf(m, " %d", from_kgid_munged(user_ns, GROUP_AT(ug->gi, i))); + seq_printf(m, "\n"); + return 0; } -static int +static struct cache_detail unix_gid_cache_template = { + .owner = THIS_MODULE, + .hash_size = GID_HASHMAX, + .name = "auth.unix.gid", + .cache_put = unix_gid_put, + .cache_request = unix_gid_request, + .cache_parse = unix_gid_parse, + .cache_show = unix_gid_show, + .match = unix_gid_match, + .init = unix_gid_init, + .update = unix_gid_update, + .alloc = unix_gid_alloc, +}; + +int unix_gid_cache_create(struct net *net) +{ + struct sunrpc_net *sn = net_generic(net, sunrpc_net_id); + struct cache_detail *cd; + int err; + + cd = cache_create_net(&unix_gid_cache_template, net); + if (IS_ERR(cd)) + return PTR_ERR(cd); + err = cache_register_net(cd, net); + if (err) { + cache_destroy_net(cd, net); + return err; + } + sn->unix_gid_cache = cd; + return 0; +} + +void unix_gid_cache_destroy(struct net *net) +{ + struct sunrpc_net *sn = net_generic(net, sunrpc_net_id); + struct cache_detail *cd = sn->unix_gid_cache; + + sn->unix_gid_cache = NULL; + cache_purge(cd); + cache_unregister_net(cd, net); + cache_destroy_net(cd, net); +} + +static struct unix_gid *unix_gid_lookup(struct cache_detail *cd, kuid_t uid) +{ + struct unix_gid ug; + struct cache_head *ch; + + ug.uid = uid; + ch = sunrpc_cache_lookup(cd, &ug.h, unix_gid_hash(uid)); + if (ch) + return container_of(ch, struct unix_gid, h); + else + return NULL; +} + +static struct group_info *unix_gid_find(kuid_t uid, struct svc_rqst *rqstp) +{ + struct unix_gid *ug; + struct group_info *gi; + int ret; + struct sunrpc_net *sn = net_generic(rqstp->rq_xprt->xpt_net, + sunrpc_net_id); + + ug = unix_gid_lookup(sn->unix_gid_cache, uid); + if (!ug) + return ERR_PTR(-EAGAIN); + ret = cache_check(sn->unix_gid_cache, &ug->h, &rqstp->rq_chandle); + switch (ret) { + case -ENOENT: + return ERR_PTR(-ENOENT); + case -ETIMEDOUT: + return ERR_PTR(-ESHUTDOWN); + case 0: + gi = get_group_info(ug->gi); + cache_put(&ug->h, sn->unix_gid_cache); + return gi; + default: + return ERR_PTR(-EAGAIN); + } +} + +int svcauth_unix_set_client(struct svc_rqst *rqstp) { - struct ip_map key, *ipm; + struct sockaddr_in *sin; + struct sockaddr_in6 *sin6, sin6_storage; + struct ip_map *ipm; + struct group_info *gi; + struct svc_cred *cred = &rqstp->rq_cred; + struct svc_xprt *xprt = rqstp->rq_xprt; + struct net *net = xprt->xpt_net; + struct sunrpc_net *sn = net_generic(net, sunrpc_net_id); + + switch (rqstp->rq_addr.ss_family) { + case AF_INET: + sin = svc_addr_in(rqstp); + sin6 = &sin6_storage; + ipv6_addr_set_v4mapped(sin->sin_addr.s_addr, &sin6->sin6_addr); + break; + case AF_INET6: + sin6 = svc_addr_in6(rqstp); + break; + default: + BUG(); + } rqstp->rq_client = NULL; if (rqstp->rq_proc == 0) return SVC_OK; - strcpy(key.m_class, rqstp->rq_server->sv_program->pg_class); - key.m_addr = rqstp->rq_addr.sin_addr; - - ipm = ip_map_lookup(&key, 0); + ipm = ip_map_cached_get(xprt); + if (ipm == NULL) + ipm = __ip_map_lookup(sn->ip_map_cache, rqstp->rq_server->sv_program->pg_class, + &sin6->sin6_addr); if (ipm == NULL) return SVC_DENIED; - switch (cache_check(&ip_map_cache, &ipm->h, &rqstp->rq_chandle)) { + switch (cache_check(sn->ip_map_cache, &ipm->h, &rqstp->rq_chandle)) { default: BUG(); + case -ETIMEDOUT: + return SVC_CLOSE; case -EAGAIN: return SVC_DROP; case -ENOENT: return SVC_DENIED; case 0: rqstp->rq_client = &ipm->m_client->h; - cache_get(&rqstp->rq_client->h); - ip_map_put(&ipm->h, &ip_map_cache); + kref_get(&rqstp->rq_client->ref); + ip_map_cached_put(xprt, ipm); break; } + + gi = unix_gid_find(cred->cr_uid, rqstp); + switch (PTR_ERR(gi)) { + case -EAGAIN: + return SVC_DROP; + case -ESHUTDOWN: + return SVC_CLOSE; + case -ENOENT: + break; + default: + put_group_info(cred->cr_group_info); + cred->cr_group_info = gi; + } return SVC_OK; } +EXPORT_SYMBOL_GPL(svcauth_unix_set_client); + static int -svcauth_null_accept(struct svc_rqst *rqstp, u32 *authp) +svcauth_null_accept(struct svc_rqst *rqstp, __be32 *authp) { struct kvec *argv = &rqstp->rq_arg.head[0]; struct kvec *resv = &rqstp->rq_res.head[0]; struct svc_cred *cred = &rqstp->rq_cred; cred->cr_group_info = NULL; + cred->cr_principal = NULL; rqstp->rq_client = NULL; if (argv->iov_len < 3*4) return SVC_GARBAGE; - if (svc_getu32(argv) != 0) { + if (svc_getu32(argv) != 0) { dprintk("svc: bad null cred\n"); *authp = rpc_autherr_badcred; return SVC_DENIED; } - if (svc_getu32(argv) != RPC_AUTH_NULL || svc_getu32(argv) != 0) { + if (svc_getu32(argv) != htonl(RPC_AUTH_NULL) || svc_getu32(argv) != 0) { dprintk("svc: bad null verf\n"); *authp = rpc_autherr_badverf; return SVC_DENIED; } /* Signal that mapping to nobody uid/gid is required */ - cred->cr_uid = (uid_t) -1; - cred->cr_gid = (gid_t) -1; + cred->cr_uid = INVALID_UID; + cred->cr_gid = INVALID_GID; cred->cr_group_info = groups_alloc(0); if (cred->cr_group_info == NULL) - return SVC_DROP; /* kmalloc failure - client must retry */ + return SVC_CLOSE; /* kmalloc failure - client must retry */ /* Put NULL verifier */ - svc_putu32(resv, RPC_AUTH_NULL); - svc_putu32(resv, 0); + svc_putnl(resv, RPC_AUTH_NULL); + svc_putnl(resv, 0); + rqstp->rq_cred.cr_flavor = RPC_AUTH_NULL; return SVC_OK; } @@ -419,7 +786,7 @@ struct auth_ops svcauth_null = { static int -svcauth_unix_accept(struct svc_rqst *rqstp, u32 *authp) +svcauth_unix_accept(struct svc_rqst *rqstp, __be32 *authp) { struct kvec *argv = &rqstp->rq_arg.head[0]; struct kvec *resv = &rqstp->rq_res.head[0]; @@ -428,6 +795,7 @@ svcauth_unix_accept(struct svc_rqst *rqstp, u32 *authp) int len = argv->iov_len; cred->cr_group_info = NULL; + cred->cr_principal = NULL; rqstp->rq_client = NULL; if ((len -= 3*4) < 0) @@ -435,32 +803,40 @@ svcauth_unix_accept(struct svc_rqst *rqstp, u32 *authp) svc_getu32(argv); /* length */ svc_getu32(argv); /* time stamp */ - slen = XDR_QUADLEN(ntohl(svc_getu32(argv))); /* machname length */ + slen = XDR_QUADLEN(svc_getnl(argv)); /* machname length */ if (slen > 64 || (len -= (slen + 3)*4) < 0) goto badcred; - argv->iov_base = (void*)((u32*)argv->iov_base + slen); /* skip machname */ + argv->iov_base = (void*)((__be32*)argv->iov_base + slen); /* skip machname */ argv->iov_len -= slen*4; - - cred->cr_uid = ntohl(svc_getu32(argv)); /* uid */ - cred->cr_gid = ntohl(svc_getu32(argv)); /* gid */ - slen = ntohl(svc_getu32(argv)); /* gids length */ + /* + * Note: we skip uid_valid()/gid_valid() checks here for + * backwards compatibility with clients that use -1 id's. + * Instead, -1 uid or gid is later mapped to the + * (export-specific) anonymous id by nfsd_setuser. + * Supplementary gid's will be left alone. + */ + cred->cr_uid = make_kuid(&init_user_ns, svc_getnl(argv)); /* uid */ + cred->cr_gid = make_kgid(&init_user_ns, svc_getnl(argv)); /* gid */ + slen = svc_getnl(argv); /* gids length */ if (slen > 16 || (len -= (slen + 2)*4) < 0) goto badcred; cred->cr_group_info = groups_alloc(slen); if (cred->cr_group_info == NULL) - return SVC_DROP; - for (i = 0; i < slen; i++) - GROUP_AT(cred->cr_group_info, i) = ntohl(svc_getu32(argv)); - - if (svc_getu32(argv) != RPC_AUTH_NULL || svc_getu32(argv) != 0) { + return SVC_CLOSE; + for (i = 0; i < slen; i++) { + kgid_t kgid = make_kgid(&init_user_ns, svc_getnl(argv)); + GROUP_AT(cred->cr_group_info, i) = kgid; + } + if (svc_getu32(argv) != htonl(RPC_AUTH_NULL) || svc_getu32(argv) != 0) { *authp = rpc_autherr_badverf; return SVC_DENIED; } /* Put NULL verifier */ - svc_putu32(resv, RPC_AUTH_NULL); - svc_putu32(resv, 0); + svc_putnl(resv, RPC_AUTH_NULL); + svc_putnl(resv, 0); + rqstp->rq_cred.cr_flavor = RPC_AUTH_UNIX; return SVC_OK; badcred: @@ -494,3 +870,45 @@ struct auth_ops svcauth_unix = { .set_client = svcauth_unix_set_client, }; +static struct cache_detail ip_map_cache_template = { + .owner = THIS_MODULE, + .hash_size = IP_HASHMAX, + .name = "auth.unix.ip", + .cache_put = ip_map_put, + .cache_request = ip_map_request, + .cache_parse = ip_map_parse, + .cache_show = ip_map_show, + .match = ip_map_match, + .init = ip_map_init, + .update = update, + .alloc = ip_map_alloc, +}; + +int ip_map_cache_create(struct net *net) +{ + struct sunrpc_net *sn = net_generic(net, sunrpc_net_id); + struct cache_detail *cd; + int err; + + cd = cache_create_net(&ip_map_cache_template, net); + if (IS_ERR(cd)) + return PTR_ERR(cd); + err = cache_register_net(cd, net); + if (err) { + cache_destroy_net(cd, net); + return err; + } + sn->ip_map_cache = cd; + return 0; +} + +void ip_map_cache_destroy(struct net *net) +{ + struct sunrpc_net *sn = net_generic(net, sunrpc_net_id); + struct cache_detail *cd = sn->ip_map_cache; + + sn->ip_map_cache = NULL; + cache_purge(cd); + cache_unregister_net(cd, net); + cache_destroy_net(cd, net); +} diff --git a/net/sunrpc/svcsock.c b/net/sunrpc/svcsock.c index f16e7cdd615..b507cd327d9 100644 --- a/net/sunrpc/svcsock.c +++ b/net/sunrpc/svcsock.c @@ -5,7 +5,7 @@ * * The server scheduling algorithm does not always distribute the load * evenly when servicing a single client. May need to modify the - * svc_sock_enqueue procedure... + * svc_xprt_enqueue procedure... * * TCP support is largely untested and may be a little slow. The problem * is that we currently do two separate recvfrom's, one for the 4-byte @@ -19,7 +19,9 @@ * Copyright (C) 1995, 1996 Olaf Kirch <okir@monad.swb.de> */ +#include <linux/kernel.h> #include <linux/sched.h> +#include <linux/module.h> #include <linux/errno.h> #include <linux/fcntl.h> #include <linux/net.h> @@ -31,364 +33,165 @@ #include <linux/slab.h> #include <linux/netdevice.h> #include <linux/skbuff.h> +#include <linux/file.h> +#include <linux/freezer.h> #include <net/sock.h> #include <net/checksum.h> #include <net/ip.h> +#include <net/ipv6.h> +#include <net/tcp.h> #include <net/tcp_states.h> #include <asm/uaccess.h> #include <asm/ioctls.h> +#include <trace/events/skb.h> #include <linux/sunrpc/types.h> +#include <linux/sunrpc/clnt.h> #include <linux/sunrpc/xdr.h> +#include <linux/sunrpc/msg_prot.h> #include <linux/sunrpc/svcsock.h> #include <linux/sunrpc/stats.h> +#include <linux/sunrpc/xprt.h> -/* SMP locking strategy: - * - * svc_serv->sv_lock protects most stuff for that service. - * - * Some flags can be set to certain values at any time - * providing that certain rules are followed: - * - * SK_BUSY can be set to 0 at any time. - * svc_sock_enqueue must be called afterwards - * SK_CONN, SK_DATA, can be set or cleared at any time. - * after a set, svc_sock_enqueue must be called. - * after a clear, the socket must be read/accepted - * if this succeeds, it must be set again. - * SK_CLOSE can set at any time. It is never cleared. - * - */ +#include "sunrpc.h" -#define RPCDBG_FACILITY RPCDBG_SVCSOCK +#define RPCDBG_FACILITY RPCDBG_SVCXPRT static struct svc_sock *svc_setup_socket(struct svc_serv *, struct socket *, - int *errp, int pmap_reg); -static void svc_udp_data_ready(struct sock *, int); + int flags); +static void svc_udp_data_ready(struct sock *); static int svc_udp_recvfrom(struct svc_rqst *); static int svc_udp_sendto(struct svc_rqst *); - -static struct svc_deferred_req *svc_deferred_dequeue(struct svc_sock *svsk); -static int svc_deferred_recv(struct svc_rqst *rqstp); -static struct cache_deferred_req *svc_defer(struct cache_req *req); - -/* - * Queue up an idle server thread. Must have serv->sv_lock held. - * Note: this is really a stack rather than a queue, so that we only - * use as many different threads as we need, and the rest don't polute - * the cache. - */ -static inline void -svc_serv_enqueue(struct svc_serv *serv, struct svc_rqst *rqstp) +static void svc_sock_detach(struct svc_xprt *); +static void svc_tcp_sock_detach(struct svc_xprt *); +static void svc_sock_free(struct svc_xprt *); + +static struct svc_xprt *svc_create_socket(struct svc_serv *, int, + struct net *, struct sockaddr *, + int, int); +#if defined(CONFIG_SUNRPC_BACKCHANNEL) +static struct svc_xprt *svc_bc_create_socket(struct svc_serv *, int, + struct net *, struct sockaddr *, + int, int); +static void svc_bc_sock_free(struct svc_xprt *xprt); +#endif /* CONFIG_SUNRPC_BACKCHANNEL */ + +#ifdef CONFIG_DEBUG_LOCK_ALLOC +static struct lock_class_key svc_key[2]; +static struct lock_class_key svc_slock_key[2]; + +static void svc_reclassify_socket(struct socket *sock) { - list_add(&rqstp->rq_list, &serv->sv_threads); -} - -/* - * Dequeue an nfsd thread. Must have serv->sv_lock held. - */ -static inline void -svc_serv_dequeue(struct svc_serv *serv, struct svc_rqst *rqstp) -{ - list_del(&rqstp->rq_list); -} - -/* - * Release an skbuff after use - */ -static inline void -svc_release_skb(struct svc_rqst *rqstp) -{ - struct sk_buff *skb = rqstp->rq_skbuff; - struct svc_deferred_req *dr = rqstp->rq_deferred; - - if (skb) { - rqstp->rq_skbuff = NULL; - - dprintk("svc: service %p, releasing skb %p\n", rqstp, skb); - skb_free_datagram(rqstp->rq_sock->sk_sk, skb); - } - if (dr) { - rqstp->rq_deferred = NULL; - kfree(dr); - } -} - -/* - * Any space to write? - */ -static inline unsigned long -svc_sock_wspace(struct svc_sock *svsk) -{ - int wspace; - - if (svsk->sk_sock->type == SOCK_STREAM) - wspace = sk_stream_wspace(svsk->sk_sk); - else - wspace = sock_wspace(svsk->sk_sk); + struct sock *sk = sock->sk; - return wspace; -} - -/* - * Queue up a socket with data pending. If there are idle nfsd - * processes, wake 'em up. - * - */ -static void -svc_sock_enqueue(struct svc_sock *svsk) -{ - struct svc_serv *serv = svsk->sk_server; - struct svc_rqst *rqstp; - - if (!(svsk->sk_flags & - ( (1<<SK_CONN)|(1<<SK_DATA)|(1<<SK_CLOSE)|(1<<SK_DEFERRED)) )) - return; - if (test_bit(SK_DEAD, &svsk->sk_flags)) + WARN_ON_ONCE(sock_owned_by_user(sk)); + if (sock_owned_by_user(sk)) return; - spin_lock_bh(&serv->sv_lock); - - if (!list_empty(&serv->sv_threads) && - !list_empty(&serv->sv_sockets)) - printk(KERN_ERR - "svc_sock_enqueue: threads and sockets both waiting??\n"); - - if (test_bit(SK_DEAD, &svsk->sk_flags)) { - /* Don't enqueue dead sockets */ - dprintk("svc: socket %p is dead, not enqueued\n", svsk->sk_sk); - goto out_unlock; - } - - if (test_bit(SK_BUSY, &svsk->sk_flags)) { - /* Don't enqueue socket while daemon is receiving */ - dprintk("svc: socket %p busy, not enqueued\n", svsk->sk_sk); - goto out_unlock; - } - - set_bit(SOCK_NOSPACE, &svsk->sk_sock->flags); - if (((svsk->sk_reserved + serv->sv_bufsz)*2 - > svc_sock_wspace(svsk)) - && !test_bit(SK_CLOSE, &svsk->sk_flags) - && !test_bit(SK_CONN, &svsk->sk_flags)) { - /* Don't enqueue while not enough space for reply */ - dprintk("svc: socket %p no space, %d*2 > %ld, not enqueued\n", - svsk->sk_sk, svsk->sk_reserved+serv->sv_bufsz, - svc_sock_wspace(svsk)); - goto out_unlock; - } - clear_bit(SOCK_NOSPACE, &svsk->sk_sock->flags); - - /* Mark socket as busy. It will remain in this state until the - * server has processed all pending data and put the socket back - * on the idle list. - */ - set_bit(SK_BUSY, &svsk->sk_flags); - - if (!list_empty(&serv->sv_threads)) { - rqstp = list_entry(serv->sv_threads.next, - struct svc_rqst, - rq_list); - dprintk("svc: socket %p served by daemon %p\n", - svsk->sk_sk, rqstp); - svc_serv_dequeue(serv, rqstp); - if (rqstp->rq_sock) - printk(KERN_ERR - "svc_sock_enqueue: server %p, rq_sock=%p!\n", - rqstp, rqstp->rq_sock); - rqstp->rq_sock = svsk; - svsk->sk_inuse++; - rqstp->rq_reserved = serv->sv_bufsz; - svsk->sk_reserved += rqstp->rq_reserved; - wake_up(&rqstp->rq_wait); - } else { - dprintk("svc: socket %p put into queue\n", svsk->sk_sk); - list_add_tail(&svsk->sk_ready, &serv->sv_sockets); + switch (sk->sk_family) { + case AF_INET: + sock_lock_init_class_and_name(sk, "slock-AF_INET-NFSD", + &svc_slock_key[0], + "sk_xprt.xpt_lock-AF_INET-NFSD", + &svc_key[0]); + break; + + case AF_INET6: + sock_lock_init_class_and_name(sk, "slock-AF_INET6-NFSD", + &svc_slock_key[1], + "sk_xprt.xpt_lock-AF_INET6-NFSD", + &svc_key[1]); + break; + + default: + BUG(); } - -out_unlock: - spin_unlock_bh(&serv->sv_lock); } - -/* - * Dequeue the first socket. Must be called with the serv->sv_lock held. - */ -static inline struct svc_sock * -svc_sock_dequeue(struct svc_serv *serv) +#else +static void svc_reclassify_socket(struct socket *sock) { - struct svc_sock *svsk; - - if (list_empty(&serv->sv_sockets)) - return NULL; - - svsk = list_entry(serv->sv_sockets.next, - struct svc_sock, sk_ready); - list_del_init(&svsk->sk_ready); - - dprintk("svc: socket %p dequeued, inuse=%d\n", - svsk->sk_sk, svsk->sk_inuse); - - return svsk; } +#endif /* - * Having read something from a socket, check whether it - * needs to be re-enqueued. - * Note: SK_DATA only gets cleared when a read-attempt finds - * no (or insufficient) data. - */ -static inline void -svc_sock_received(struct svc_sock *svsk) -{ - clear_bit(SK_BUSY, &svsk->sk_flags); - svc_sock_enqueue(svsk); -} - - -/** - * svc_reserve - change the space reserved for the reply to a request. - * @rqstp: The request in question - * @space: new max space to reserve - * - * Each request reserves some space on the output queue of the socket - * to make sure the reply fits. This function reduces that reserved - * space to be the amount of space used already, plus @space. - * + * Release an skbuff after use */ -void svc_reserve(struct svc_rqst *rqstp, int space) +static void svc_release_skb(struct svc_rqst *rqstp) { - space += rqstp->rq_res.head[0].iov_len; + struct sk_buff *skb = rqstp->rq_xprt_ctxt; - if (space < rqstp->rq_reserved) { - struct svc_sock *svsk = rqstp->rq_sock; - spin_lock_bh(&svsk->sk_server->sv_lock); - svsk->sk_reserved -= (rqstp->rq_reserved - space); - rqstp->rq_reserved = space; - spin_unlock_bh(&svsk->sk_server->sv_lock); + if (skb) { + struct svc_sock *svsk = + container_of(rqstp->rq_xprt, struct svc_sock, sk_xprt); + rqstp->rq_xprt_ctxt = NULL; - svc_sock_enqueue(svsk); + dprintk("svc: service %p, releasing skb %p\n", rqstp, skb); + skb_free_datagram_locked(svsk->sk_sk, skb); } } -/* - * Release a socket after use. - */ -static inline void -svc_sock_put(struct svc_sock *svsk) -{ - struct svc_serv *serv = svsk->sk_server; - - spin_lock_bh(&serv->sv_lock); - if (!--(svsk->sk_inuse) && test_bit(SK_DEAD, &svsk->sk_flags)) { - spin_unlock_bh(&serv->sv_lock); - dprintk("svc: releasing dead socket\n"); - sock_release(svsk->sk_sock); - kfree(svsk); - } - else - spin_unlock_bh(&serv->sv_lock); -} +union svc_pktinfo_u { + struct in_pktinfo pkti; + struct in6_pktinfo pkti6; +}; +#define SVC_PKTINFO_SPACE \ + CMSG_SPACE(sizeof(union svc_pktinfo_u)) -static void -svc_sock_release(struct svc_rqst *rqstp) +static void svc_set_cmsg_data(struct svc_rqst *rqstp, struct cmsghdr *cmh) { - struct svc_sock *svsk = rqstp->rq_sock; - - svc_release_skb(rqstp); - - svc_free_allpages(rqstp); - rqstp->rq_res.page_len = 0; - rqstp->rq_res.page_base = 0; - - - /* Reset response buffer and release - * the reservation. - * But first, check that enough space was reserved - * for the reply, otherwise we have a bug! - */ - if ((rqstp->rq_res.len) > rqstp->rq_reserved) - printk(KERN_ERR "RPC request reserved %d but used %d\n", - rqstp->rq_reserved, - rqstp->rq_res.len); - - rqstp->rq_res.head[0].iov_len = 0; - svc_reserve(rqstp, 0); - rqstp->rq_sock = NULL; - - svc_sock_put(svsk); -} + struct svc_sock *svsk = + container_of(rqstp->rq_xprt, struct svc_sock, sk_xprt); + switch (svsk->sk_sk->sk_family) { + case AF_INET: { + struct in_pktinfo *pki = CMSG_DATA(cmh); + + cmh->cmsg_level = SOL_IP; + cmh->cmsg_type = IP_PKTINFO; + pki->ipi_ifindex = 0; + pki->ipi_spec_dst.s_addr = + svc_daddr_in(rqstp)->sin_addr.s_addr; + cmh->cmsg_len = CMSG_LEN(sizeof(*pki)); + } + break; -/* - * External function to wake up a server waiting for data - */ -void -svc_wake_up(struct svc_serv *serv) -{ - struct svc_rqst *rqstp; + case AF_INET6: { + struct in6_pktinfo *pki = CMSG_DATA(cmh); + struct sockaddr_in6 *daddr = svc_daddr_in6(rqstp); - spin_lock_bh(&serv->sv_lock); - if (!list_empty(&serv->sv_threads)) { - rqstp = list_entry(serv->sv_threads.next, - struct svc_rqst, - rq_list); - dprintk("svc: daemon %p woken up.\n", rqstp); - /* - svc_serv_dequeue(serv, rqstp); - rqstp->rq_sock = NULL; - */ - wake_up(&rqstp->rq_wait); + cmh->cmsg_level = SOL_IPV6; + cmh->cmsg_type = IPV6_PKTINFO; + pki->ipi6_ifindex = daddr->sin6_scope_id; + pki->ipi6_addr = daddr->sin6_addr; + cmh->cmsg_len = CMSG_LEN(sizeof(*pki)); + } + break; } - spin_unlock_bh(&serv->sv_lock); } /* - * Generic sendto routine + * send routine intended to be shared by the fore- and back-channel */ -static int -svc_sendto(struct svc_rqst *rqstp, struct xdr_buf *xdr) +int svc_send_common(struct socket *sock, struct xdr_buf *xdr, + struct page *headpage, unsigned long headoffset, + struct page *tailpage, unsigned long tailoffset) { - struct svc_sock *svsk = rqstp->rq_sock; - struct socket *sock = svsk->sk_sock; - int slen; - char buffer[CMSG_SPACE(sizeof(struct in_pktinfo))]; - struct cmsghdr *cmh = (struct cmsghdr *)buffer; - struct in_pktinfo *pki = (struct in_pktinfo *)CMSG_DATA(cmh); - int len = 0; int result; int size; struct page **ppage = xdr->pages; size_t base = xdr->page_base; unsigned int pglen = xdr->page_len; unsigned int flags = MSG_MORE; + int slen; + int len = 0; slen = xdr->len; - if (rqstp->rq_prot == IPPROTO_UDP) { - /* set the source and destination */ - struct msghdr msg; - msg.msg_name = &rqstp->rq_addr; - msg.msg_namelen = sizeof(rqstp->rq_addr); - msg.msg_iov = NULL; - msg.msg_iovlen = 0; - msg.msg_flags = MSG_MORE; - - msg.msg_control = cmh; - msg.msg_controllen = sizeof(buffer); - cmh->cmsg_len = CMSG_LEN(sizeof(*pki)); - cmh->cmsg_level = SOL_IP; - cmh->cmsg_type = IP_PKTINFO; - pki->ipi_ifindex = 0; - pki->ipi_spec_dst.s_addr = rqstp->rq_daddr; - - if (sock_sendmsg(sock, &msg, 0) < 0) - goto out; - } - /* send head */ if (slen == xdr->head[0].iov_len) flags = 0; - len = sock->ops->sendpage(sock, rqstp->rq_respages[0], 0, xdr->head[0].iov_len, flags); + len = kernel_sendpage(sock, headpage, headoffset, + xdr->head[0].iov_len, flags); if (len != xdr->head[0].iov_len) goto out; slen -= xdr->head[0].iov_len; @@ -400,7 +203,7 @@ svc_sendto(struct svc_rqst *rqstp, struct xdr_buf *xdr) while (pglen > 0) { if (slen == size) flags = 0; - result = sock->ops->sendpage(sock, *ppage, base, size, flags); + result = kernel_sendpage(sock, *ppage, base, size, flags); if (result > 0) len += result; if (result != size) @@ -411,36 +214,112 @@ svc_sendto(struct svc_rqst *rqstp, struct xdr_buf *xdr) base = 0; ppage++; } + /* send tail */ if (xdr->tail[0].iov_len) { - result = sock->ops->sendpage(sock, rqstp->rq_respages[rqstp->rq_restailpage], - ((unsigned long)xdr->tail[0].iov_base)& (PAGE_SIZE-1), - xdr->tail[0].iov_len, 0); - + result = kernel_sendpage(sock, tailpage, tailoffset, + xdr->tail[0].iov_len, 0); if (result > 0) len += result; } + +out: + return len; +} + + +/* + * Generic sendto routine + */ +static int svc_sendto(struct svc_rqst *rqstp, struct xdr_buf *xdr) +{ + struct svc_sock *svsk = + container_of(rqstp->rq_xprt, struct svc_sock, sk_xprt); + struct socket *sock = svsk->sk_sock; + union { + struct cmsghdr hdr; + long all[SVC_PKTINFO_SPACE / sizeof(long)]; + } buffer; + struct cmsghdr *cmh = &buffer.hdr; + int len = 0; + unsigned long tailoff; + unsigned long headoff; + RPC_IFDEBUG(char buf[RPC_MAX_ADDRBUFLEN]); + + if (rqstp->rq_prot == IPPROTO_UDP) { + struct msghdr msg = { + .msg_name = &rqstp->rq_addr, + .msg_namelen = rqstp->rq_addrlen, + .msg_control = cmh, + .msg_controllen = sizeof(buffer), + .msg_flags = MSG_MORE, + }; + + svc_set_cmsg_data(rqstp, cmh); + + if (sock_sendmsg(sock, &msg, 0) < 0) + goto out; + } + + tailoff = ((unsigned long)xdr->tail[0].iov_base) & (PAGE_SIZE-1); + headoff = 0; + len = svc_send_common(sock, xdr, rqstp->rq_respages[0], headoff, + rqstp->rq_respages[0], tailoff); + out: - dprintk("svc: socket %p sendto([%p %Zu... ], %d) = %d (addr %x)\n", - rqstp->rq_sock, xdr->head[0].iov_base, xdr->head[0].iov_len, xdr->len, len, - rqstp->rq_addr.sin_addr.s_addr); + dprintk("svc: socket %p sendto([%p %Zu... ], %d) = %d (addr %s)\n", + svsk, xdr->head[0].iov_base, xdr->head[0].iov_len, + xdr->len, len, svc_print_addr(rqstp, buf, sizeof(buf))); return len; } /* + * Report socket names for nfsdfs + */ +static int svc_one_sock_name(struct svc_sock *svsk, char *buf, int remaining) +{ + const struct sock *sk = svsk->sk_sk; + const char *proto_name = sk->sk_protocol == IPPROTO_UDP ? + "udp" : "tcp"; + int len; + + switch (sk->sk_family) { + case PF_INET: + len = snprintf(buf, remaining, "ipv4 %s %pI4 %d\n", + proto_name, + &inet_sk(sk)->inet_rcv_saddr, + inet_sk(sk)->inet_num); + break; +#if IS_ENABLED(CONFIG_IPV6) + case PF_INET6: + len = snprintf(buf, remaining, "ipv6 %s %pI6 %d\n", + proto_name, + &sk->sk_v6_rcv_saddr, + inet_sk(sk)->inet_num); + break; +#endif + default: + len = snprintf(buf, remaining, "*unknown-%d*\n", + sk->sk_family); + } + + if (len >= remaining) { + *buf = '\0'; + return -ENAMETOOLONG; + } + return len; +} + +/* * Check input queue length */ -static int -svc_recv_available(struct svc_sock *svsk) +static int svc_recv_available(struct svc_sock *svsk) { - mm_segment_t oldfs; struct socket *sock = svsk->sk_sock; int avail, err; - oldfs = get_fs(); set_fs(KERNEL_DS); - err = sock->ops->ioctl(sock, TIOCINQ, (unsigned long) &avail); - set_fs(oldfs); + err = kernel_sock_ioctl(sock, TIOCINQ, (unsigned long) &avail); return (err >= 0)? avail : err; } @@ -448,43 +327,58 @@ svc_recv_available(struct svc_sock *svsk) /* * Generic recvfrom routine. */ -static int -svc_recvfrom(struct svc_rqst *rqstp, struct kvec *iov, int nr, int buflen) +static int svc_recvfrom(struct svc_rqst *rqstp, struct kvec *iov, int nr, + int buflen) { - struct msghdr msg; - struct socket *sock; - int len, alen; - - rqstp->rq_addrlen = sizeof(rqstp->rq_addr); - sock = rqstp->rq_sock->sk_sock; + struct svc_sock *svsk = + container_of(rqstp->rq_xprt, struct svc_sock, sk_xprt); + struct msghdr msg = { + .msg_flags = MSG_DONTWAIT, + }; + int len; - msg.msg_name = &rqstp->rq_addr; - msg.msg_namelen = sizeof(rqstp->rq_addr); - msg.msg_control = NULL; - msg.msg_controllen = 0; + rqstp->rq_xprt_hlen = 0; - msg.msg_flags = MSG_DONTWAIT; - - len = kernel_recvmsg(sock, &msg, iov, nr, buflen, MSG_DONTWAIT); - - /* sock_recvmsg doesn't fill in the name/namelen, so we must.. - * possibly we should cache this in the svc_sock structure - * at accept time. FIXME - */ - alen = sizeof(rqstp->rq_addr); - sock->ops->getname(sock, (struct sockaddr *)&rqstp->rq_addr, &alen, 1); + len = kernel_recvmsg(svsk->sk_sock, &msg, iov, nr, buflen, + msg.msg_flags); dprintk("svc: socket %p recvfrom(%p, %Zu) = %d\n", - rqstp->rq_sock, iov[0].iov_base, iov[0].iov_len, len); - + svsk, iov[0].iov_base, iov[0].iov_len, len); return len; } +static int svc_partial_recvfrom(struct svc_rqst *rqstp, + struct kvec *iov, int nr, + int buflen, unsigned int base) +{ + size_t save_iovlen; + void *save_iovbase; + unsigned int i; + int ret; + + if (base == 0) + return svc_recvfrom(rqstp, iov, nr, buflen); + + for (i = 0; i < nr; i++) { + if (iov[i].iov_len > base) + break; + base -= iov[i].iov_len; + } + save_iovlen = iov[i].iov_len; + save_iovbase = iov[i].iov_base; + iov[i].iov_len -= base; + iov[i].iov_base += base; + ret = svc_recvfrom(rqstp, &iov[i], nr - i, buflen); + iov[i].iov_len = save_iovlen; + iov[i].iov_base = save_iovbase; + return ret; +} + /* * Set socket snd and rcv buffer lengths */ -static inline void -svc_sock_setbufsize(struct socket *sock, unsigned int snd, unsigned int rcv) +static void svc_sock_setbufsize(struct socket *sock, unsigned int snd, + unsigned int rcv) { #if 0 mm_segment_t oldfs; @@ -502,111 +396,192 @@ svc_sock_setbufsize(struct socket *sock, unsigned int snd, unsigned int rcv) lock_sock(sock->sk); sock->sk->sk_sndbuf = snd * 2; sock->sk->sk_rcvbuf = rcv * 2; - sock->sk->sk_userlocks |= SOCK_SNDBUF_LOCK|SOCK_RCVBUF_LOCK; + sock->sk->sk_write_space(sock->sk); release_sock(sock->sk); #endif } + +static int svc_sock_secure_port(struct svc_rqst *rqstp) +{ + return svc_port_is_privileged(svc_addr(rqstp)); +} + /* * INET callback when data has been received on the socket. */ -static void -svc_udp_data_ready(struct sock *sk, int count) +static void svc_udp_data_ready(struct sock *sk) { struct svc_sock *svsk = (struct svc_sock *)sk->sk_user_data; + wait_queue_head_t *wq = sk_sleep(sk); if (svsk) { - dprintk("svc: socket %p(inet %p), count=%d, busy=%d\n", - svsk, sk, count, test_bit(SK_BUSY, &svsk->sk_flags)); - set_bit(SK_DATA, &svsk->sk_flags); - svc_sock_enqueue(svsk); + dprintk("svc: socket %p(inet %p), busy=%d\n", + svsk, sk, + test_bit(XPT_BUSY, &svsk->sk_xprt.xpt_flags)); + set_bit(XPT_DATA, &svsk->sk_xprt.xpt_flags); + svc_xprt_enqueue(&svsk->sk_xprt); } - if (sk->sk_sleep && waitqueue_active(sk->sk_sleep)) - wake_up_interruptible(sk->sk_sleep); + if (wq && waitqueue_active(wq)) + wake_up_interruptible(wq); } /* * INET callback when space is newly available on the socket. */ -static void -svc_write_space(struct sock *sk) +static void svc_write_space(struct sock *sk) { struct svc_sock *svsk = (struct svc_sock *)(sk->sk_user_data); + wait_queue_head_t *wq = sk_sleep(sk); if (svsk) { dprintk("svc: socket %p(inet %p), write_space busy=%d\n", - svsk, sk, test_bit(SK_BUSY, &svsk->sk_flags)); - svc_sock_enqueue(svsk); + svsk, sk, test_bit(XPT_BUSY, &svsk->sk_xprt.xpt_flags)); + svc_xprt_enqueue(&svsk->sk_xprt); } - if (sk->sk_sleep && waitqueue_active(sk->sk_sleep)) { + if (wq && waitqueue_active(wq)) { dprintk("RPC svc_write_space: someone sleeping on %p\n", svsk); - wake_up_interruptible(sk->sk_sleep); + wake_up_interruptible(wq); } } +static void svc_tcp_write_space(struct sock *sk) +{ + struct socket *sock = sk->sk_socket; + + if (sk_stream_is_writeable(sk) && sock) + clear_bit(SOCK_NOSPACE, &sock->flags); + svc_write_space(sk); +} + +/* + * See net/ipv6/ip_sockglue.c : ip_cmsg_recv_pktinfo + */ +static int svc_udp_get_dest_address4(struct svc_rqst *rqstp, + struct cmsghdr *cmh) +{ + struct in_pktinfo *pki = CMSG_DATA(cmh); + struct sockaddr_in *daddr = svc_daddr_in(rqstp); + + if (cmh->cmsg_type != IP_PKTINFO) + return 0; + + daddr->sin_family = AF_INET; + daddr->sin_addr.s_addr = pki->ipi_spec_dst.s_addr; + return 1; +} + +/* + * See net/ipv6/datagram.c : ip6_datagram_recv_ctl + */ +static int svc_udp_get_dest_address6(struct svc_rqst *rqstp, + struct cmsghdr *cmh) +{ + struct in6_pktinfo *pki = CMSG_DATA(cmh); + struct sockaddr_in6 *daddr = svc_daddr_in6(rqstp); + + if (cmh->cmsg_type != IPV6_PKTINFO) + return 0; + + daddr->sin6_family = AF_INET6; + daddr->sin6_addr = pki->ipi6_addr; + daddr->sin6_scope_id = pki->ipi6_ifindex; + return 1; +} + +/* + * Copy the UDP datagram's destination address to the rqstp structure. + * The 'destination' address in this case is the address to which the + * peer sent the datagram, i.e. our local address. For multihomed + * hosts, this can change from msg to msg. Note that only the IP + * address changes, the port number should remain the same. + */ +static int svc_udp_get_dest_address(struct svc_rqst *rqstp, + struct cmsghdr *cmh) +{ + switch (cmh->cmsg_level) { + case SOL_IP: + return svc_udp_get_dest_address4(rqstp, cmh); + case SOL_IPV6: + return svc_udp_get_dest_address6(rqstp, cmh); + } + + return 0; +} + /* * Receive a datagram from a UDP socket. */ -static int -svc_udp_recvfrom(struct svc_rqst *rqstp) +static int svc_udp_recvfrom(struct svc_rqst *rqstp) { - struct svc_sock *svsk = rqstp->rq_sock; - struct svc_serv *serv = svsk->sk_server; + struct svc_sock *svsk = + container_of(rqstp->rq_xprt, struct svc_sock, sk_xprt); + struct svc_serv *serv = svsk->sk_xprt.xpt_server; struct sk_buff *skb; - int err, len; - - if (test_and_clear_bit(SK_CHNGBUF, &svsk->sk_flags)) + union { + struct cmsghdr hdr; + long all[SVC_PKTINFO_SPACE / sizeof(long)]; + } buffer; + struct cmsghdr *cmh = &buffer.hdr; + struct msghdr msg = { + .msg_name = svc_addr(rqstp), + .msg_control = cmh, + .msg_controllen = sizeof(buffer), + .msg_flags = MSG_DONTWAIT, + }; + size_t len; + int err; + + if (test_and_clear_bit(XPT_CHNGBUF, &svsk->sk_xprt.xpt_flags)) /* udp sockets need large rcvbuf as all pending * requests are still in that buffer. sndbuf must * also be large enough that there is enough space - * for one reply per thread. + * for one reply per thread. We count all threads + * rather than threads in a particular pool, which + * provides an upper bound on the number of threads + * which will access the socket. */ svc_sock_setbufsize(svsk->sk_sock, - (serv->sv_nrthreads+3) * serv->sv_bufsz, - (serv->sv_nrthreads+3) * serv->sv_bufsz); - - if ((rqstp->rq_deferred = svc_deferred_dequeue(svsk))) { - svc_sock_received(svsk); - return svc_deferred_recv(rqstp); - } - - clear_bit(SK_DATA, &svsk->sk_flags); - while ((skb = skb_recv_datagram(svsk->sk_sk, 0, 1, &err)) == NULL) { - if (err == -EAGAIN) { - svc_sock_received(svsk); - return err; + (serv->sv_nrthreads+3) * serv->sv_max_mesg, + (serv->sv_nrthreads+3) * serv->sv_max_mesg); + + clear_bit(XPT_DATA, &svsk->sk_xprt.xpt_flags); + skb = NULL; + err = kernel_recvmsg(svsk->sk_sock, &msg, NULL, + 0, 0, MSG_PEEK | MSG_DONTWAIT); + if (err >= 0) + skb = skb_recv_datagram(svsk->sk_sk, 0, 1, &err); + + if (skb == NULL) { + if (err != -EAGAIN) { + /* possibly an icmp error */ + dprintk("svc: recvfrom returned error %d\n", -err); + set_bit(XPT_DATA, &svsk->sk_xprt.xpt_flags); } - /* possibly an icmp error */ - dprintk("svc: recvfrom returned error %d\n", -err); + return 0; } - if (skb->tstamp.off_sec == 0) { - struct timeval tv; - - tv.tv_sec = xtime.tv_sec; - tv.tv_usec = xtime.tv_nsec / NSEC_PER_USEC; - skb_set_timestamp(skb, &tv); - /* Don't enable netstamp, sunrpc doesn't + len = svc_addr_len(svc_addr(rqstp)); + rqstp->rq_addrlen = len; + if (skb->tstamp.tv64 == 0) { + skb->tstamp = ktime_get_real(); + /* Don't enable netstamp, sunrpc doesn't need that much accuracy */ } - skb_get_timestamp(skb, &svsk->sk_sk->sk_stamp); - set_bit(SK_DATA, &svsk->sk_flags); /* there may be more data... */ - - /* - * Maybe more packets - kick another thread ASAP. - */ - svc_sock_received(svsk); + svsk->sk_sk->sk_stamp = skb->tstamp; + set_bit(XPT_DATA, &svsk->sk_xprt.xpt_flags); /* there may be more data... */ len = skb->len - sizeof(struct udphdr); rqstp->rq_arg.len = len; - rqstp->rq_prot = IPPROTO_UDP; + rqstp->rq_prot = IPPROTO_UDP; - /* Get sender address */ - rqstp->rq_addr.sin_family = AF_INET; - rqstp->rq_addr.sin_port = skb->h.uh->source; - rqstp->rq_addr.sin_addr.s_addr = skb->nh.iph->saddr; - rqstp->rq_daddr = skb->nh.iph->daddr; + if (!svc_udp_get_dest_address(rqstp, cmh)) { + net_warn_ratelimited("svc: received unknown control message %d/%d; dropping RPC reply datagram\n", + cmh->cmsg_level, cmh->cmsg_type); + goto out_free; + } + rqstp->rq_daddrlen = svc_addr_len(svc_daddr(rqstp)); if (skb_is_nonlinear(skb)) { /* we have to copy */ @@ -614,38 +589,40 @@ svc_udp_recvfrom(struct svc_rqst *rqstp) if (csum_partial_copy_to_xdr(&rqstp->rq_arg, skb)) { local_bh_enable(); /* checksum error */ - skb_free_datagram(svsk->sk_sk, skb); - return 0; + goto out_free; } local_bh_enable(); - skb_free_datagram(svsk->sk_sk, skb); + skb_free_datagram_locked(svsk->sk_sk, skb); } else { /* we can use it in-place */ - rqstp->rq_arg.head[0].iov_base = skb->data + sizeof(struct udphdr); + rqstp->rq_arg.head[0].iov_base = skb->data + + sizeof(struct udphdr); rqstp->rq_arg.head[0].iov_len = len; - if (skb->ip_summed != CHECKSUM_UNNECESSARY) { - if ((unsigned short)csum_fold(skb_checksum(skb, 0, skb->len, skb->csum))) { - skb_free_datagram(svsk->sk_sk, skb); - return 0; - } - skb->ip_summed = CHECKSUM_UNNECESSARY; - } - rqstp->rq_skbuff = skb; + if (skb_checksum_complete(skb)) + goto out_free; + rqstp->rq_xprt_ctxt = skb; } rqstp->rq_arg.page_base = 0; if (len <= rqstp->rq_arg.head[0].iov_len) { rqstp->rq_arg.head[0].iov_len = len; rqstp->rq_arg.page_len = 0; + rqstp->rq_respages = rqstp->rq_pages+1; } else { rqstp->rq_arg.page_len = len - rqstp->rq_arg.head[0].iov_len; - rqstp->rq_argused += (rqstp->rq_arg.page_len + PAGE_SIZE - 1)/ PAGE_SIZE; + rqstp->rq_respages = rqstp->rq_pages + 1 + + DIV_ROUND_UP(rqstp->rq_arg.page_len, PAGE_SIZE); } + rqstp->rq_next_page = rqstp->rq_respages+1; if (serv->sv_stats) serv->sv_stats->netudpcnt++; return len; +out_free: + trace_kfree_skb(skb, svc_udp_recvfrom); + skb_free_datagram_locked(svsk->sk_sk, skb); + return 0; } static int @@ -661,34 +638,110 @@ svc_udp_sendto(struct svc_rqst *rqstp) return error; } -static void -svc_udp_init(struct svc_sock *svsk) +static void svc_udp_prep_reply_hdr(struct svc_rqst *rqstp) +{ +} + +static int svc_udp_has_wspace(struct svc_xprt *xprt) +{ + struct svc_sock *svsk = container_of(xprt, struct svc_sock, sk_xprt); + struct svc_serv *serv = xprt->xpt_server; + unsigned long required; + + /* + * Set the SOCK_NOSPACE flag before checking the available + * sock space. + */ + set_bit(SOCK_NOSPACE, &svsk->sk_sock->flags); + required = atomic_read(&svsk->sk_xprt.xpt_reserved) + serv->sv_max_mesg; + if (required*2 > sock_wspace(svsk->sk_sk)) + return 0; + clear_bit(SOCK_NOSPACE, &svsk->sk_sock->flags); + return 1; +} + +static struct svc_xprt *svc_udp_accept(struct svc_xprt *xprt) +{ + BUG(); + return NULL; +} + +static struct svc_xprt *svc_udp_create(struct svc_serv *serv, + struct net *net, + struct sockaddr *sa, int salen, + int flags) +{ + return svc_create_socket(serv, IPPROTO_UDP, net, sa, salen, flags); +} + +static struct svc_xprt_ops svc_udp_ops = { + .xpo_create = svc_udp_create, + .xpo_recvfrom = svc_udp_recvfrom, + .xpo_sendto = svc_udp_sendto, + .xpo_release_rqst = svc_release_skb, + .xpo_detach = svc_sock_detach, + .xpo_free = svc_sock_free, + .xpo_prep_reply_hdr = svc_udp_prep_reply_hdr, + .xpo_has_wspace = svc_udp_has_wspace, + .xpo_accept = svc_udp_accept, + .xpo_secure_port = svc_sock_secure_port, +}; + +static struct svc_xprt_class svc_udp_class = { + .xcl_name = "udp", + .xcl_owner = THIS_MODULE, + .xcl_ops = &svc_udp_ops, + .xcl_max_payload = RPCSVC_MAXPAYLOAD_UDP, +}; + +static void svc_udp_init(struct svc_sock *svsk, struct svc_serv *serv) { + int err, level, optname, one = 1; + + svc_xprt_init(sock_net(svsk->sk_sock->sk), &svc_udp_class, + &svsk->sk_xprt, serv); + clear_bit(XPT_CACHE_AUTH, &svsk->sk_xprt.xpt_flags); svsk->sk_sk->sk_data_ready = svc_udp_data_ready; svsk->sk_sk->sk_write_space = svc_write_space; - svsk->sk_recvfrom = svc_udp_recvfrom; - svsk->sk_sendto = svc_udp_sendto; /* initialise setting must have enough space to - * receive and respond to one request. + * receive and respond to one request. * svc_udp_recvfrom will re-adjust if necessary */ svc_sock_setbufsize(svsk->sk_sock, - 3 * svsk->sk_server->sv_bufsz, - 3 * svsk->sk_server->sv_bufsz); - - set_bit(SK_DATA, &svsk->sk_flags); /* might have come in before data_ready set up */ - set_bit(SK_CHNGBUF, &svsk->sk_flags); + 3 * svsk->sk_xprt.xpt_server->sv_max_mesg, + 3 * svsk->sk_xprt.xpt_server->sv_max_mesg); + + /* data might have come in before data_ready set up */ + set_bit(XPT_DATA, &svsk->sk_xprt.xpt_flags); + set_bit(XPT_CHNGBUF, &svsk->sk_xprt.xpt_flags); + + /* make sure we get destination address info */ + switch (svsk->sk_sk->sk_family) { + case AF_INET: + level = SOL_IP; + optname = IP_PKTINFO; + break; + case AF_INET6: + level = SOL_IPV6; + optname = IPV6_RECVPKTINFO; + break; + default: + BUG(); + } + err = kernel_setsockopt(svsk->sk_sock, level, optname, + (char *)&one, sizeof(one)); + dprintk("svc: kernel_setsockopt returned %d\n", err); } /* * A data_ready event on a listening socket means there's a connection * pending. Do not use state_change as a substitute for it. */ -static void -svc_tcp_listen_data_ready(struct sock *sk, int count_unused) +static void svc_tcp_listen_data_ready(struct sock *sk) { struct svc_sock *svsk = (struct svc_sock *)sk->sk_user_data; + wait_queue_head_t *wq; dprintk("svc: socket %p TCP (listen) state change %d\n", sk, sk->sk_state); @@ -705,23 +758,24 @@ svc_tcp_listen_data_ready(struct sock *sk, int count_unused) */ if (sk->sk_state == TCP_LISTEN) { if (svsk) { - set_bit(SK_CONN, &svsk->sk_flags); - svc_sock_enqueue(svsk); + set_bit(XPT_CONN, &svsk->sk_xprt.xpt_flags); + svc_xprt_enqueue(&svsk->sk_xprt); } else printk("svc: socket %p: no user data\n", sk); } - if (sk->sk_sleep && waitqueue_active(sk->sk_sleep)) - wake_up_interruptible_all(sk->sk_sleep); + wq = sk_sleep(sk); + if (wq && waitqueue_active(wq)) + wake_up_interruptible_all(wq); } /* * A state change on a connected socket means it's dying or dead. */ -static void -svc_tcp_state_change(struct sock *sk) +static void svc_tcp_state_change(struct sock *sk) { struct svc_sock *svsk = (struct svc_sock *)sk->sk_user_data; + wait_queue_head_t *wq = sk_sleep(sk); dprintk("svc: socket %p TCP (connected) state change %d (svsk %p)\n", sk, sk->sk_state, sk->sk_user_data); @@ -729,210 +783,180 @@ svc_tcp_state_change(struct sock *sk) if (!svsk) printk("svc: socket %p: no user data\n", sk); else { - set_bit(SK_CLOSE, &svsk->sk_flags); - svc_sock_enqueue(svsk); + set_bit(XPT_CLOSE, &svsk->sk_xprt.xpt_flags); + svc_xprt_enqueue(&svsk->sk_xprt); } - if (sk->sk_sleep && waitqueue_active(sk->sk_sleep)) - wake_up_interruptible_all(sk->sk_sleep); + if (wq && waitqueue_active(wq)) + wake_up_interruptible_all(wq); } -static void -svc_tcp_data_ready(struct sock *sk, int count) +static void svc_tcp_data_ready(struct sock *sk) { struct svc_sock *svsk = (struct svc_sock *)sk->sk_user_data; + wait_queue_head_t *wq = sk_sleep(sk); dprintk("svc: socket %p TCP data ready (svsk %p)\n", sk, sk->sk_user_data); if (svsk) { - set_bit(SK_DATA, &svsk->sk_flags); - svc_sock_enqueue(svsk); + set_bit(XPT_DATA, &svsk->sk_xprt.xpt_flags); + svc_xprt_enqueue(&svsk->sk_xprt); } - if (sk->sk_sleep && waitqueue_active(sk->sk_sleep)) - wake_up_interruptible(sk->sk_sleep); + if (wq && waitqueue_active(wq)) + wake_up_interruptible(wq); } /* * Accept a TCP connection */ -static void -svc_tcp_accept(struct svc_sock *svsk) +static struct svc_xprt *svc_tcp_accept(struct svc_xprt *xprt) { - struct sockaddr_in sin; - struct svc_serv *serv = svsk->sk_server; + struct svc_sock *svsk = container_of(xprt, struct svc_sock, sk_xprt); + struct sockaddr_storage addr; + struct sockaddr *sin = (struct sockaddr *) &addr; + struct svc_serv *serv = svsk->sk_xprt.xpt_server; struct socket *sock = svsk->sk_sock; struct socket *newsock; - struct proto_ops *ops; struct svc_sock *newsvsk; int err, slen; + RPC_IFDEBUG(char buf[RPC_MAX_ADDRBUFLEN]); dprintk("svc: tcp_accept %p sock %p\n", svsk, sock); if (!sock) - return; + return NULL; - err = sock_create_lite(PF_INET, SOCK_STREAM, IPPROTO_TCP, &newsock); - if (err) { + clear_bit(XPT_CONN, &svsk->sk_xprt.xpt_flags); + err = kernel_accept(sock, &newsock, O_NONBLOCK); + if (err < 0) { if (err == -ENOMEM) printk(KERN_WARNING "%s: no more sockets!\n", serv->sv_name); - return; - } - - dprintk("svc: tcp_accept %p allocated\n", newsock); - newsock->ops = ops = sock->ops; - - clear_bit(SK_CONN, &svsk->sk_flags); - if ((err = ops->accept(sock, newsock, O_NONBLOCK)) < 0) { - if (err != -EAGAIN && net_ratelimit()) - printk(KERN_WARNING "%s: accept failed (err %d)!\n", - serv->sv_name, -err); - goto failed; /* aborted connection or whatever */ + else if (err != -EAGAIN) + net_warn_ratelimited("%s: accept failed (err %d)!\n", + serv->sv_name, -err); + return NULL; } - set_bit(SK_CONN, &svsk->sk_flags); - svc_sock_enqueue(svsk); + set_bit(XPT_CONN, &svsk->sk_xprt.xpt_flags); - slen = sizeof(sin); - err = ops->getname(newsock, (struct sockaddr *) &sin, &slen, 1); + err = kernel_getpeername(newsock, sin, &slen); if (err < 0) { - if (net_ratelimit()) - printk(KERN_WARNING "%s: peername failed (err %d)!\n", - serv->sv_name, -err); + net_warn_ratelimited("%s: peername failed (err %d)!\n", + serv->sv_name, -err); goto failed; /* aborted connection or whatever */ } /* Ideally, we would want to reject connections from unauthorized - * hosts here, but when we get encription, the IP of the host won't - * tell us anything. For now just warn about unpriv connections. + * hosts here, but when we get encryption, the IP of the host won't + * tell us anything. For now just warn about unpriv connections. */ - if (ntohs(sin.sin_port) >= 1024) { - dprintk(KERN_WARNING - "%s: connect from unprivileged port: %u.%u.%u.%u:%d\n", - serv->sv_name, - NIPQUAD(sin.sin_addr.s_addr), ntohs(sin.sin_port)); + if (!svc_port_is_privileged(sin)) { + dprintk("%s: connect from unprivileged port: %s\n", + serv->sv_name, + __svc_print_addr(sin, buf, sizeof(buf))); } - - dprintk("%s: connect from %u.%u.%u.%u:%04x\n", serv->sv_name, - NIPQUAD(sin.sin_addr.s_addr), ntohs(sin.sin_port)); + dprintk("%s: connect from %s\n", serv->sv_name, + __svc_print_addr(sin, buf, sizeof(buf))); /* make sure that a write doesn't block forever when * low on memory */ newsock->sk->sk_sndtimeo = HZ*30; - if (!(newsvsk = svc_setup_socket(serv, newsock, &err, 0))) + newsvsk = svc_setup_socket(serv, newsock, + (SVC_SOCK_ANONYMOUS | SVC_SOCK_TEMPORARY)); + if (IS_ERR(newsvsk)) goto failed; - - - /* make sure that we don't have too many active connections. - * If we have, something must be dropped. - * - * There's no point in trying to do random drop here for - * DoS prevention. The NFS clients does 1 reconnect in 15 - * seconds. An attacker can easily beat that. - * - * The only somewhat efficient mechanism would be if drop - * old connections from the same IP first. But right now - * we don't even record the client IP in svc_sock. - */ - if (serv->sv_tmpcnt > (serv->sv_nrthreads+3)*20) { - struct svc_sock *svsk = NULL; - spin_lock_bh(&serv->sv_lock); - if (!list_empty(&serv->sv_tempsocks)) { - if (net_ratelimit()) { - /* Try to help the admin */ - printk(KERN_NOTICE "%s: too many open TCP " - "sockets, consider increasing the " - "number of nfsd threads\n", - serv->sv_name); - printk(KERN_NOTICE "%s: last TCP connect from " - "%u.%u.%u.%u:%d\n", - serv->sv_name, - NIPQUAD(sin.sin_addr.s_addr), - ntohs(sin.sin_port)); - } - /* - * Always select the oldest socket. It's not fair, - * but so is life - */ - svsk = list_entry(serv->sv_tempsocks.prev, - struct svc_sock, - sk_list); - set_bit(SK_CLOSE, &svsk->sk_flags); - svsk->sk_inuse ++; - } - spin_unlock_bh(&serv->sv_lock); - - if (svsk) { - svc_sock_enqueue(svsk); - svc_sock_put(svsk); - } - + svc_xprt_set_remote(&newsvsk->sk_xprt, sin, slen); + err = kernel_getsockname(newsock, sin, &slen); + if (unlikely(err < 0)) { + dprintk("svc_tcp_accept: kernel_getsockname error %d\n", -err); + slen = offsetof(struct sockaddr, sa_data); } + svc_xprt_set_local(&newsvsk->sk_xprt, sin, slen); + if (sock_is_loopback(newsock->sk)) + set_bit(XPT_LOCAL, &newsvsk->sk_xprt.xpt_flags); + else + clear_bit(XPT_LOCAL, &newsvsk->sk_xprt.xpt_flags); if (serv->sv_stats) serv->sv_stats->nettcpconn++; - return; + return &newsvsk->sk_xprt; failed: sock_release(newsock); - return; + return NULL; } -/* - * Receive data from a TCP socket. - */ -static int -svc_tcp_recvfrom(struct svc_rqst *rqstp) +static unsigned int svc_tcp_restore_pages(struct svc_sock *svsk, struct svc_rqst *rqstp) { - struct svc_sock *svsk = rqstp->rq_sock; - struct svc_serv *serv = svsk->sk_server; - int len; - struct kvec vec[RPCSVC_MAXPAGES]; - int pnum, vlen; + unsigned int i, len, npages; - dprintk("svc: tcp_recv %p data %d conn %d close %d\n", - svsk, test_bit(SK_DATA, &svsk->sk_flags), - test_bit(SK_CONN, &svsk->sk_flags), - test_bit(SK_CLOSE, &svsk->sk_flags)); - - if ((rqstp->rq_deferred = svc_deferred_dequeue(svsk))) { - svc_sock_received(svsk); - return svc_deferred_recv(rqstp); + if (svsk->sk_datalen == 0) + return 0; + len = svsk->sk_datalen; + npages = (len + PAGE_SIZE - 1) >> PAGE_SHIFT; + for (i = 0; i < npages; i++) { + if (rqstp->rq_pages[i] != NULL) + put_page(rqstp->rq_pages[i]); + BUG_ON(svsk->sk_pages[i] == NULL); + rqstp->rq_pages[i] = svsk->sk_pages[i]; + svsk->sk_pages[i] = NULL; } + rqstp->rq_arg.head[0].iov_base = page_address(rqstp->rq_pages[0]); + return len; +} - if (test_bit(SK_CLOSE, &svsk->sk_flags)) { - svc_delete_socket(svsk); - return 0; +static void svc_tcp_save_pages(struct svc_sock *svsk, struct svc_rqst *rqstp) +{ + unsigned int i, len, npages; + + if (svsk->sk_datalen == 0) + return; + len = svsk->sk_datalen; + npages = (len + PAGE_SIZE - 1) >> PAGE_SHIFT; + for (i = 0; i < npages; i++) { + svsk->sk_pages[i] = rqstp->rq_pages[i]; + rqstp->rq_pages[i] = NULL; } +} - if (test_bit(SK_CONN, &svsk->sk_flags)) { - svc_tcp_accept(svsk); - svc_sock_received(svsk); - return 0; +static void svc_tcp_clear_pages(struct svc_sock *svsk) +{ + unsigned int i, len, npages; + + if (svsk->sk_datalen == 0) + goto out; + len = svsk->sk_datalen; + npages = (len + PAGE_SIZE - 1) >> PAGE_SHIFT; + for (i = 0; i < npages; i++) { + if (svsk->sk_pages[i] == NULL) { + WARN_ON_ONCE(1); + continue; + } + put_page(svsk->sk_pages[i]); + svsk->sk_pages[i] = NULL; } +out: + svsk->sk_tcplen = 0; + svsk->sk_datalen = 0; +} - if (test_and_clear_bit(SK_CHNGBUF, &svsk->sk_flags)) - /* sndbuf needs to have room for one request - * per thread, otherwise we can stall even when the - * network isn't a bottleneck. - * rcvbuf just needs to be able to hold a few requests. - * Normally they will be removed from the queue - * as soon a a complete request arrives. - */ - svc_sock_setbufsize(svsk->sk_sock, - (serv->sv_nrthreads+3) * serv->sv_bufsz, - 3 * serv->sv_bufsz); +/* + * Receive fragment record header. + * If we haven't gotten the record length yet, get the next four bytes. + */ +static int svc_tcp_recv_record(struct svc_sock *svsk, struct svc_rqst *rqstp) +{ + struct svc_serv *serv = svsk->sk_xprt.xpt_server; + unsigned int want; + int len; - clear_bit(SK_DATA, &svsk->sk_flags); + clear_bit(XPT_DATA, &svsk->sk_xprt.xpt_flags); - /* Receive data. If we haven't got the record length yet, get - * the next four bytes. Otherwise try to gobble up as much as - * possible up to the complete record length. - */ - if (svsk->sk_tcplen < 4) { - unsigned long want = 4 - svsk->sk_tcplen; + if (svsk->sk_tcplen < sizeof(rpc_fraghdr)) { struct kvec iov; + want = sizeof(rpc_fraghdr) - svsk->sk_tcplen; iov.iov_base = ((char *) &svsk->sk_reclen) + svsk->sk_tcplen; iov.iov_len = want; if ((len = svc_recvfrom(rqstp, &iov, 1, want)) < 0) @@ -940,110 +964,205 @@ svc_tcp_recvfrom(struct svc_rqst *rqstp) svsk->sk_tcplen += len; if (len < want) { - dprintk("svc: short recvfrom while reading record length (%d of %lu)\n", - len, want); - svc_sock_received(svsk); - return -EAGAIN; /* record header not complete */ + dprintk("svc: short recvfrom while reading record " + "length (%d of %d)\n", len, want); + return -EAGAIN; } - svsk->sk_reclen = ntohl(svsk->sk_reclen); - if (!(svsk->sk_reclen & 0x80000000)) { - /* FIXME: technically, a record can be fragmented, - * and non-terminal fragments will not have the top - * bit set in the fragment length header. - * But apparently no known nfs clients send fragmented - * records. */ - printk(KERN_NOTICE "RPC: bad TCP reclen 0x%08lx (non-terminal)\n", - (unsigned long) svsk->sk_reclen); - goto err_delete; - } - svsk->sk_reclen &= 0x7fffffff; - dprintk("svc: TCP record, %d bytes\n", svsk->sk_reclen); - if (svsk->sk_reclen > serv->sv_bufsz) { - printk(KERN_NOTICE "RPC: bad TCP reclen 0x%08lx (large)\n", - (unsigned long) svsk->sk_reclen); + dprintk("svc: TCP record, %d bytes\n", svc_sock_reclen(svsk)); + if (svc_sock_reclen(svsk) + svsk->sk_datalen > + serv->sv_max_mesg) { + net_notice_ratelimited("RPC: fragment too large: %d\n", + svc_sock_reclen(svsk)); goto err_delete; } } - /* Check whether enough data is available */ - len = svc_recv_available(svsk); - if (len < 0) - goto error; + return svc_sock_reclen(svsk); +error: + dprintk("RPC: TCP recv_record got %d\n", len); + return len; +err_delete: + set_bit(XPT_CLOSE, &svsk->sk_xprt.xpt_flags); + return -EAGAIN; +} - if (len < svsk->sk_reclen) { - dprintk("svc: incomplete TCP record (%d of %d)\n", - len, svsk->sk_reclen); - svc_sock_received(svsk); - return -EAGAIN; /* record not complete */ +static int receive_cb_reply(struct svc_sock *svsk, struct svc_rqst *rqstp) +{ + struct rpc_xprt *bc_xprt = svsk->sk_xprt.xpt_bc_xprt; + struct rpc_rqst *req = NULL; + struct kvec *src, *dst; + __be32 *p = (__be32 *)rqstp->rq_arg.head[0].iov_base; + __be32 xid; + __be32 calldir; + + xid = *p++; + calldir = *p; + + if (bc_xprt) + req = xprt_lookup_rqst(bc_xprt, xid); + + if (!req) { + printk(KERN_NOTICE + "%s: Got unrecognized reply: " + "calldir 0x%x xpt_bc_xprt %p xid %08x\n", + __func__, ntohl(calldir), + bc_xprt, xid); + return -EAGAIN; } - len = svsk->sk_reclen; - set_bit(SK_DATA, &svsk->sk_flags); - - vec[0] = rqstp->rq_arg.head[0]; - vlen = PAGE_SIZE; - pnum = 1; - while (vlen < len) { - vec[pnum].iov_base = page_address(rqstp->rq_argpages[rqstp->rq_argused++]); - vec[pnum].iov_len = PAGE_SIZE; - pnum++; - vlen += PAGE_SIZE; + + memcpy(&req->rq_private_buf, &req->rq_rcv_buf, sizeof(struct xdr_buf)); + /* + * XXX!: cheating for now! Only copying HEAD. + * But we know this is good enough for now (in fact, for any + * callback reply in the forseeable future). + */ + dst = &req->rq_private_buf.head[0]; + src = &rqstp->rq_arg.head[0]; + if (dst->iov_len < src->iov_len) + return -EAGAIN; /* whatever; just giving up. */ + memcpy(dst->iov_base, src->iov_base, src->iov_len); + xprt_complete_rqst(req->rq_task, rqstp->rq_arg.len); + rqstp->rq_arg.len = 0; + return 0; +} + +static int copy_pages_to_kvecs(struct kvec *vec, struct page **pages, int len) +{ + int i = 0; + int t = 0; + + while (t < len) { + vec[i].iov_base = page_address(pages[i]); + vec[i].iov_len = PAGE_SIZE; + i++; + t += PAGE_SIZE; } + return i; +} - /* Now receive data */ - len = svc_recvfrom(rqstp, vec, pnum, len); +static void svc_tcp_fragment_received(struct svc_sock *svsk) +{ + /* If we have more data, signal svc_xprt_enqueue() to try again */ + if (svc_recv_available(svsk) > sizeof(rpc_fraghdr)) + set_bit(XPT_DATA, &svsk->sk_xprt.xpt_flags); + dprintk("svc: TCP %s record (%d bytes)\n", + svc_sock_final_rec(svsk) ? "final" : "nonfinal", + svc_sock_reclen(svsk)); + svsk->sk_tcplen = 0; + svsk->sk_reclen = 0; +} + +/* + * Receive data from a TCP socket. + */ +static int svc_tcp_recvfrom(struct svc_rqst *rqstp) +{ + struct svc_sock *svsk = + container_of(rqstp->rq_xprt, struct svc_sock, sk_xprt); + struct svc_serv *serv = svsk->sk_xprt.xpt_server; + int len; + struct kvec *vec; + unsigned int want, base; + __be32 *p; + __be32 calldir; + int pnum; + + dprintk("svc: tcp_recv %p data %d conn %d close %d\n", + svsk, test_bit(XPT_DATA, &svsk->sk_xprt.xpt_flags), + test_bit(XPT_CONN, &svsk->sk_xprt.xpt_flags), + test_bit(XPT_CLOSE, &svsk->sk_xprt.xpt_flags)); + + len = svc_tcp_recv_record(svsk, rqstp); if (len < 0) goto error; - dprintk("svc: TCP complete record (%d bytes)\n", len); - rqstp->rq_arg.len = len; + base = svc_tcp_restore_pages(svsk, rqstp); + want = svc_sock_reclen(svsk) - (svsk->sk_tcplen - sizeof(rpc_fraghdr)); + + vec = rqstp->rq_vec; + + pnum = copy_pages_to_kvecs(&vec[0], &rqstp->rq_pages[0], + svsk->sk_datalen + want); + + rqstp->rq_respages = &rqstp->rq_pages[pnum]; + rqstp->rq_next_page = rqstp->rq_respages + 1; + + /* Now receive data */ + len = svc_partial_recvfrom(rqstp, vec, pnum, want, base); + if (len >= 0) { + svsk->sk_tcplen += len; + svsk->sk_datalen += len; + } + if (len != want || !svc_sock_final_rec(svsk)) { + svc_tcp_save_pages(svsk, rqstp); + if (len < 0 && len != -EAGAIN) + goto err_delete; + if (len == want) + svc_tcp_fragment_received(svsk); + else + dprintk("svc: incomplete TCP record (%d of %d)\n", + (int)(svsk->sk_tcplen - sizeof(rpc_fraghdr)), + svc_sock_reclen(svsk)); + goto err_noclose; + } + + if (svsk->sk_datalen < 8) { + svsk->sk_datalen = 0; + goto err_delete; /* client is nuts. */ + } + + rqstp->rq_arg.len = svsk->sk_datalen; rqstp->rq_arg.page_base = 0; - if (len <= rqstp->rq_arg.head[0].iov_len) { - rqstp->rq_arg.head[0].iov_len = len; + if (rqstp->rq_arg.len <= rqstp->rq_arg.head[0].iov_len) { + rqstp->rq_arg.head[0].iov_len = rqstp->rq_arg.len; rqstp->rq_arg.page_len = 0; - } else { - rqstp->rq_arg.page_len = len - rqstp->rq_arg.head[0].iov_len; - } + } else + rqstp->rq_arg.page_len = rqstp->rq_arg.len - rqstp->rq_arg.head[0].iov_len; - rqstp->rq_skbuff = NULL; + rqstp->rq_xprt_ctxt = NULL; rqstp->rq_prot = IPPROTO_TCP; + rqstp->rq_local = !!test_bit(XPT_LOCAL, &svsk->sk_xprt.xpt_flags); + + p = (__be32 *)rqstp->rq_arg.head[0].iov_base; + calldir = p[1]; + if (calldir) + len = receive_cb_reply(svsk, rqstp); /* Reset TCP read info */ - svsk->sk_reclen = 0; - svsk->sk_tcplen = 0; + svsk->sk_datalen = 0; + svc_tcp_fragment_received(svsk); + + if (len < 0) + goto error; - svc_sock_received(svsk); + svc_xprt_copy_addrs(rqstp, &svsk->sk_xprt); if (serv->sv_stats) serv->sv_stats->nettcpcnt++; - return len; - - err_delete: - svc_delete_socket(svsk); - return -EAGAIN; - - error: - if (len == -EAGAIN) { - dprintk("RPC: TCP recvfrom got EAGAIN\n"); - svc_sock_received(svsk); - } else { - printk(KERN_NOTICE "%s: recvfrom returned errno %d\n", - svsk->sk_server->sv_name, -len); - svc_sock_received(svsk); - } - - return len; + return rqstp->rq_arg.len; + +error: + if (len != -EAGAIN) + goto err_delete; + dprintk("RPC: TCP recvfrom got EAGAIN\n"); + return 0; +err_delete: + printk(KERN_NOTICE "%s: recvfrom returned errno %d\n", + svsk->sk_xprt.xpt_server->sv_name, -len); + set_bit(XPT_CLOSE, &svsk->sk_xprt.xpt_flags); +err_noclose: + return 0; /* record not complete */ } /* * Send out data on TCP socket. */ -static int -svc_tcp_sendto(struct svc_rqst *rqstp) +static int svc_tcp_sendto(struct svc_rqst *rqstp) { struct xdr_buf *xbufp = &rqstp->rq_res; int sent; - u32 reclen; + __be32 reclen; /* Set up the first element of the reply kvec. * Any other kvecs that may be in use have been taken @@ -1052,528 +1171,491 @@ svc_tcp_sendto(struct svc_rqst *rqstp) reclen = htonl(0x80000000|((xbufp->len ) - 4)); memcpy(xbufp->head[0].iov_base, &reclen, 4); - if (test_bit(SK_DEAD, &rqstp->rq_sock->sk_flags)) - return -ENOTCONN; - sent = svc_sendto(rqstp, &rqstp->rq_res); if (sent != xbufp->len) { - printk(KERN_NOTICE "rpc-srv/tcp: %s: %s %d when sending %d bytes - shutting down socket\n", - rqstp->rq_sock->sk_server->sv_name, + printk(KERN_NOTICE + "rpc-srv/tcp: %s: %s %d when sending %d bytes " + "- shutting down socket\n", + rqstp->rq_xprt->xpt_server->sv_name, (sent<0)?"got error":"sent only", sent, xbufp->len); - svc_delete_socket(rqstp->rq_sock); + set_bit(XPT_CLOSE, &rqstp->rq_xprt->xpt_flags); + svc_xprt_enqueue(rqstp->rq_xprt); sent = -EAGAIN; } return sent; } -static void -svc_tcp_init(struct svc_sock *svsk) +/* + * Setup response header. TCP has a 4B record length field. + */ +static void svc_tcp_prep_reply_hdr(struct svc_rqst *rqstp) { - struct sock *sk = svsk->sk_sk; - struct tcp_sock *tp = tcp_sk(sk); + struct kvec *resv = &rqstp->rq_res.head[0]; + + /* tcp needs a space for the record length... */ + svc_putnl(resv, 0); +} + +static int svc_tcp_has_wspace(struct svc_xprt *xprt) +{ + struct svc_sock *svsk = container_of(xprt, struct svc_sock, sk_xprt); + struct svc_serv *serv = svsk->sk_xprt.xpt_server; + int required; + + if (test_bit(XPT_LISTENER, &xprt->xpt_flags)) + return 1; + required = atomic_read(&xprt->xpt_reserved) + serv->sv_max_mesg; + if (sk_stream_wspace(svsk->sk_sk) >= required || + (sk_stream_min_wspace(svsk->sk_sk) == 0 && + atomic_read(&xprt->xpt_reserved) == 0)) + return 1; + set_bit(SOCK_NOSPACE, &svsk->sk_sock->flags); + return 0; +} + +static struct svc_xprt *svc_tcp_create(struct svc_serv *serv, + struct net *net, + struct sockaddr *sa, int salen, + int flags) +{ + return svc_create_socket(serv, IPPROTO_TCP, net, sa, salen, flags); +} + +#if defined(CONFIG_SUNRPC_BACKCHANNEL) +static struct svc_xprt *svc_bc_create_socket(struct svc_serv *, int, + struct net *, struct sockaddr *, + int, int); +static void svc_bc_sock_free(struct svc_xprt *xprt); - svsk->sk_recvfrom = svc_tcp_recvfrom; - svsk->sk_sendto = svc_tcp_sendto; +static struct svc_xprt *svc_bc_tcp_create(struct svc_serv *serv, + struct net *net, + struct sockaddr *sa, int salen, + int flags) +{ + return svc_bc_create_socket(serv, IPPROTO_TCP, net, sa, salen, flags); +} + +static void svc_bc_tcp_sock_detach(struct svc_xprt *xprt) +{ +} + +static struct svc_xprt_ops svc_tcp_bc_ops = { + .xpo_create = svc_bc_tcp_create, + .xpo_detach = svc_bc_tcp_sock_detach, + .xpo_free = svc_bc_sock_free, + .xpo_prep_reply_hdr = svc_tcp_prep_reply_hdr, + .xpo_secure_port = svc_sock_secure_port, +}; + +static struct svc_xprt_class svc_tcp_bc_class = { + .xcl_name = "tcp-bc", + .xcl_owner = THIS_MODULE, + .xcl_ops = &svc_tcp_bc_ops, + .xcl_max_payload = RPCSVC_MAXPAYLOAD_TCP, +}; + +static void svc_init_bc_xprt_sock(void) +{ + svc_reg_xprt_class(&svc_tcp_bc_class); +} + +static void svc_cleanup_bc_xprt_sock(void) +{ + svc_unreg_xprt_class(&svc_tcp_bc_class); +} +#else /* CONFIG_SUNRPC_BACKCHANNEL */ +static void svc_init_bc_xprt_sock(void) +{ +} + +static void svc_cleanup_bc_xprt_sock(void) +{ +} +#endif /* CONFIG_SUNRPC_BACKCHANNEL */ + +static struct svc_xprt_ops svc_tcp_ops = { + .xpo_create = svc_tcp_create, + .xpo_recvfrom = svc_tcp_recvfrom, + .xpo_sendto = svc_tcp_sendto, + .xpo_release_rqst = svc_release_skb, + .xpo_detach = svc_tcp_sock_detach, + .xpo_free = svc_sock_free, + .xpo_prep_reply_hdr = svc_tcp_prep_reply_hdr, + .xpo_has_wspace = svc_tcp_has_wspace, + .xpo_accept = svc_tcp_accept, + .xpo_secure_port = svc_sock_secure_port, +}; + +static struct svc_xprt_class svc_tcp_class = { + .xcl_name = "tcp", + .xcl_owner = THIS_MODULE, + .xcl_ops = &svc_tcp_ops, + .xcl_max_payload = RPCSVC_MAXPAYLOAD_TCP, +}; + +void svc_init_xprt_sock(void) +{ + svc_reg_xprt_class(&svc_tcp_class); + svc_reg_xprt_class(&svc_udp_class); + svc_init_bc_xprt_sock(); +} + +void svc_cleanup_xprt_sock(void) +{ + svc_unreg_xprt_class(&svc_tcp_class); + svc_unreg_xprt_class(&svc_udp_class); + svc_cleanup_bc_xprt_sock(); +} + +static void svc_tcp_init(struct svc_sock *svsk, struct svc_serv *serv) +{ + struct sock *sk = svsk->sk_sk; + svc_xprt_init(sock_net(svsk->sk_sock->sk), &svc_tcp_class, + &svsk->sk_xprt, serv); + set_bit(XPT_CACHE_AUTH, &svsk->sk_xprt.xpt_flags); if (sk->sk_state == TCP_LISTEN) { dprintk("setting up TCP socket for listening\n"); + set_bit(XPT_LISTENER, &svsk->sk_xprt.xpt_flags); sk->sk_data_ready = svc_tcp_listen_data_ready; - set_bit(SK_CONN, &svsk->sk_flags); + set_bit(XPT_CONN, &svsk->sk_xprt.xpt_flags); } else { dprintk("setting up TCP socket for reading\n"); sk->sk_state_change = svc_tcp_state_change; sk->sk_data_ready = svc_tcp_data_ready; - sk->sk_write_space = svc_write_space; + sk->sk_write_space = svc_tcp_write_space; svsk->sk_reclen = 0; svsk->sk_tcplen = 0; + svsk->sk_datalen = 0; + memset(&svsk->sk_pages[0], 0, sizeof(svsk->sk_pages)); - tp->nonagle = 1; /* disable Nagle's algorithm */ + tcp_sk(sk)->nonagle |= TCP_NAGLE_OFF; - /* initialise setting must have enough space to - * receive and respond to one request. - * svc_tcp_recvfrom will re-adjust if necessary - */ - svc_sock_setbufsize(svsk->sk_sock, - 3 * svsk->sk_server->sv_bufsz, - 3 * svsk->sk_server->sv_bufsz); - - set_bit(SK_CHNGBUF, &svsk->sk_flags); - set_bit(SK_DATA, &svsk->sk_flags); - if (sk->sk_state != TCP_ESTABLISHED) - set_bit(SK_CLOSE, &svsk->sk_flags); + set_bit(XPT_DATA, &svsk->sk_xprt.xpt_flags); + if (sk->sk_state != TCP_ESTABLISHED) + set_bit(XPT_CLOSE, &svsk->sk_xprt.xpt_flags); } } -void -svc_sock_update_bufs(struct svc_serv *serv) +void svc_sock_update_bufs(struct svc_serv *serv) { /* * The number of server threads has changed. Update * rcvbuf and sndbuf accordingly on all sockets */ - struct list_head *le; - - spin_lock_bh(&serv->sv_lock); - list_for_each(le, &serv->sv_permsocks) { - struct svc_sock *svsk = - list_entry(le, struct svc_sock, sk_list); - set_bit(SK_CHNGBUF, &svsk->sk_flags); - } - list_for_each(le, &serv->sv_tempsocks) { - struct svc_sock *svsk = - list_entry(le, struct svc_sock, sk_list); - set_bit(SK_CHNGBUF, &svsk->sk_flags); - } - spin_unlock_bh(&serv->sv_lock); -} - -/* - * Receive the next request on any socket. - */ -int -svc_recv(struct svc_serv *serv, struct svc_rqst *rqstp, long timeout) -{ - struct svc_sock *svsk =NULL; - int len; - int pages; - struct xdr_buf *arg; - DECLARE_WAITQUEUE(wait, current); - - dprintk("svc: server %p waiting for data (to = %ld)\n", - rqstp, timeout); - - if (rqstp->rq_sock) - printk(KERN_ERR - "svc_recv: service %p, socket not NULL!\n", - rqstp); - if (waitqueue_active(&rqstp->rq_wait)) - printk(KERN_ERR - "svc_recv: service %p, wait queue active!\n", - rqstp); - - /* Initialize the buffers */ - /* first reclaim pages that were moved to response list */ - svc_pushback_allpages(rqstp); - - /* now allocate needed pages. If we get a failure, sleep briefly */ - pages = 2 + (serv->sv_bufsz + PAGE_SIZE -1) / PAGE_SIZE; - while (rqstp->rq_arghi < pages) { - struct page *p = alloc_page(GFP_KERNEL); - if (!p) { - schedule_timeout_uninterruptible(msecs_to_jiffies(500)); - continue; - } - rqstp->rq_argpages[rqstp->rq_arghi++] = p; - } - - /* Make arg->head point to first page and arg->pages point to rest */ - arg = &rqstp->rq_arg; - arg->head[0].iov_base = page_address(rqstp->rq_argpages[0]); - arg->head[0].iov_len = PAGE_SIZE; - rqstp->rq_argused = 1; - arg->pages = rqstp->rq_argpages + 1; - arg->page_base = 0; - /* save at least one page for response */ - arg->page_len = (pages-2)*PAGE_SIZE; - arg->len = (pages-1)*PAGE_SIZE; - arg->tail[0].iov_len = 0; - - try_to_freeze(); - if (signalled()) - return -EINTR; + struct svc_sock *svsk; spin_lock_bh(&serv->sv_lock); - if (!list_empty(&serv->sv_tempsocks)) { - svsk = list_entry(serv->sv_tempsocks.next, - struct svc_sock, sk_list); - /* apparently the "standard" is that clients close - * idle connections after 5 minutes, servers after - * 6 minutes - * http://www.connectathon.org/talks96/nfstcp.pdf - */ - if (get_seconds() - svsk->sk_lastrecv < 6*60 - || test_bit(SK_BUSY, &svsk->sk_flags)) - svsk = NULL; - } - if (svsk) { - set_bit(SK_BUSY, &svsk->sk_flags); - set_bit(SK_CLOSE, &svsk->sk_flags); - rqstp->rq_sock = svsk; - svsk->sk_inuse++; - } else if ((svsk = svc_sock_dequeue(serv)) != NULL) { - rqstp->rq_sock = svsk; - svsk->sk_inuse++; - rqstp->rq_reserved = serv->sv_bufsz; - svsk->sk_reserved += rqstp->rq_reserved; - } else { - /* No data pending. Go to sleep */ - svc_serv_enqueue(serv, rqstp); - - /* - * We have to be able to interrupt this wait - * to bring down the daemons ... - */ - set_current_state(TASK_INTERRUPTIBLE); - add_wait_queue(&rqstp->rq_wait, &wait); - spin_unlock_bh(&serv->sv_lock); - - schedule_timeout(timeout); - - try_to_freeze(); - - spin_lock_bh(&serv->sv_lock); - remove_wait_queue(&rqstp->rq_wait, &wait); - - if (!(svsk = rqstp->rq_sock)) { - svc_serv_dequeue(serv, rqstp); - spin_unlock_bh(&serv->sv_lock); - dprintk("svc: server %p, no data yet\n", rqstp); - return signalled()? -EINTR : -EAGAIN; - } - } + list_for_each_entry(svsk, &serv->sv_permsocks, sk_xprt.xpt_list) + set_bit(XPT_CHNGBUF, &svsk->sk_xprt.xpt_flags); spin_unlock_bh(&serv->sv_lock); - - dprintk("svc: server %p, socket %p, inuse=%d\n", - rqstp, svsk, svsk->sk_inuse); - len = svsk->sk_recvfrom(rqstp); - dprintk("svc: got len=%d\n", len); - - /* No data, incomplete (TCP) read, or accept() */ - if (len == 0 || len == -EAGAIN) { - rqstp->rq_res.len = 0; - svc_sock_release(rqstp); - return -EAGAIN; - } - svsk->sk_lastrecv = get_seconds(); - if (test_bit(SK_TEMP, &svsk->sk_flags)) { - /* push active sockets to end of list */ - spin_lock_bh(&serv->sv_lock); - if (!list_empty(&svsk->sk_list)) - list_move_tail(&svsk->sk_list, &serv->sv_tempsocks); - spin_unlock_bh(&serv->sv_lock); - } - - rqstp->rq_secure = ntohs(rqstp->rq_addr.sin_port) < 1024; - rqstp->rq_chandle.defer = svc_defer; - - if (serv->sv_stats) - serv->sv_stats->netcnt++; - return len; -} - -/* - * Drop request - */ -void -svc_drop(struct svc_rqst *rqstp) -{ - dprintk("svc: socket %p dropped request\n", rqstp->rq_sock); - svc_sock_release(rqstp); -} - -/* - * Return reply to client. - */ -int -svc_send(struct svc_rqst *rqstp) -{ - struct svc_sock *svsk; - int len; - struct xdr_buf *xb; - - if ((svsk = rqstp->rq_sock) == NULL) { - printk(KERN_WARNING "NULL socket pointer in %s:%d\n", - __FILE__, __LINE__); - return -EFAULT; - } - - /* release the receive skb before sending the reply */ - svc_release_skb(rqstp); - - /* calculate over-all length */ - xb = & rqstp->rq_res; - xb->len = xb->head[0].iov_len + - xb->page_len + - xb->tail[0].iov_len; - - /* Grab svsk->sk_sem to serialize outgoing data. */ - down(&svsk->sk_sem); - if (test_bit(SK_DEAD, &svsk->sk_flags)) - len = -ENOTCONN; - else - len = svsk->sk_sendto(rqstp); - up(&svsk->sk_sem); - svc_sock_release(rqstp); - - if (len == -ECONNREFUSED || len == -ENOTCONN || len == -EAGAIN) - return 0; - return len; } +EXPORT_SYMBOL_GPL(svc_sock_update_bufs); /* * Initialize socket for RPC use and create svc_sock struct * XXX: May want to setsockopt SO_SNDBUF and SO_RCVBUF. */ -static struct svc_sock * -svc_setup_socket(struct svc_serv *serv, struct socket *sock, - int *errp, int pmap_register) +static struct svc_sock *svc_setup_socket(struct svc_serv *serv, + struct socket *sock, + int flags) { struct svc_sock *svsk; struct sock *inet; + int pmap_register = !(flags & SVC_SOCK_ANONYMOUS); + int err = 0; dprintk("svc: svc_setup_socket %p\n", sock); - if (!(svsk = kmalloc(sizeof(*svsk), GFP_KERNEL))) { - *errp = -ENOMEM; - return NULL; - } - memset(svsk, 0, sizeof(*svsk)); + svsk = kzalloc(sizeof(*svsk), GFP_KERNEL); + if (!svsk) + return ERR_PTR(-ENOMEM); inet = sock->sk; /* Register socket with portmapper */ - if (*errp >= 0 && pmap_register) - *errp = svc_register(serv, inet->sk_protocol, - ntohs(inet_sk(inet)->sport)); + if (pmap_register) + err = svc_register(serv, sock_net(sock->sk), inet->sk_family, + inet->sk_protocol, + ntohs(inet_sk(inet)->inet_sport)); - if (*errp < 0) { + if (err < 0) { kfree(svsk); - return NULL; + return ERR_PTR(err); } - set_bit(SK_BUSY, &svsk->sk_flags); inet->sk_user_data = svsk; svsk->sk_sock = sock; svsk->sk_sk = inet; svsk->sk_ostate = inet->sk_state_change; svsk->sk_odata = inet->sk_data_ready; svsk->sk_owspace = inet->sk_write_space; - svsk->sk_server = serv; - svsk->sk_lastrecv = get_seconds(); - INIT_LIST_HEAD(&svsk->sk_deferred); - INIT_LIST_HEAD(&svsk->sk_ready); - sema_init(&svsk->sk_sem, 1); /* Initialize the socket */ if (sock->type == SOCK_DGRAM) - svc_udp_init(svsk); - else - svc_tcp_init(svsk); - - spin_lock_bh(&serv->sv_lock); - if (!pmap_register) { - set_bit(SK_TEMP, &svsk->sk_flags); - list_add(&svsk->sk_list, &serv->sv_tempsocks); - serv->sv_tmpcnt++; - } else { - clear_bit(SK_TEMP, &svsk->sk_flags); - list_add(&svsk->sk_list, &serv->sv_permsocks); + svc_udp_init(svsk, serv); + else { + /* initialise setting must have enough space to + * receive and respond to one request. + */ + svc_sock_setbufsize(svsk->sk_sock, 4 * serv->sv_max_mesg, + 4 * serv->sv_max_mesg); + svc_tcp_init(svsk, serv); } - spin_unlock_bh(&serv->sv_lock); dprintk("svc: svc_setup_socket created %p (inet %p)\n", svsk, svsk->sk_sk); - clear_bit(SK_BUSY, &svsk->sk_flags); - svc_sock_enqueue(svsk); return svsk; } +bool svc_alien_sock(struct net *net, int fd) +{ + int err; + struct socket *sock = sockfd_lookup(fd, &err); + bool ret = false; + + if (!sock) + goto out; + if (sock_net(sock->sk) != net) + ret = true; + sockfd_put(sock); +out: + return ret; +} +EXPORT_SYMBOL_GPL(svc_alien_sock); + +/** + * svc_addsock - add a listener socket to an RPC service + * @serv: pointer to RPC service to which to add a new listener + * @fd: file descriptor of the new listener + * @name_return: pointer to buffer to fill in with name of listener + * @len: size of the buffer + * + * Fills in socket name and returns positive length of name if successful. + * Name is terminated with '\n'. On error, returns a negative errno + * value. + */ +int svc_addsock(struct svc_serv *serv, const int fd, char *name_return, + const size_t len) +{ + int err = 0; + struct socket *so = sockfd_lookup(fd, &err); + struct svc_sock *svsk = NULL; + struct sockaddr_storage addr; + struct sockaddr *sin = (struct sockaddr *)&addr; + int salen; + + if (!so) + return err; + err = -EAFNOSUPPORT; + if ((so->sk->sk_family != PF_INET) && (so->sk->sk_family != PF_INET6)) + goto out; + err = -EPROTONOSUPPORT; + if (so->sk->sk_protocol != IPPROTO_TCP && + so->sk->sk_protocol != IPPROTO_UDP) + goto out; + err = -EISCONN; + if (so->state > SS_UNCONNECTED) + goto out; + err = -ENOENT; + if (!try_module_get(THIS_MODULE)) + goto out; + svsk = svc_setup_socket(serv, so, SVC_SOCK_DEFAULTS); + if (IS_ERR(svsk)) { + module_put(THIS_MODULE); + err = PTR_ERR(svsk); + goto out; + } + if (kernel_getsockname(svsk->sk_sock, sin, &salen) == 0) + svc_xprt_set_local(&svsk->sk_xprt, sin, salen); + svc_add_new_perm_xprt(serv, &svsk->sk_xprt); + return svc_one_sock_name(svsk, name_return, len); +out: + sockfd_put(so); + return err; +} +EXPORT_SYMBOL_GPL(svc_addsock); + /* * Create socket for RPC service. */ -static int -svc_create_socket(struct svc_serv *serv, int protocol, struct sockaddr_in *sin) +static struct svc_xprt *svc_create_socket(struct svc_serv *serv, + int protocol, + struct net *net, + struct sockaddr *sin, int len, + int flags) { struct svc_sock *svsk; struct socket *sock; int error; int type; + struct sockaddr_storage addr; + struct sockaddr *newsin = (struct sockaddr *)&addr; + int newlen; + int family; + int val; + RPC_IFDEBUG(char buf[RPC_MAX_ADDRBUFLEN]); - dprintk("svc: svc_create_socket(%s, %d, %u.%u.%u.%u:%d)\n", - serv->sv_program->pg_name, protocol, - NIPQUAD(sin->sin_addr.s_addr), - ntohs(sin->sin_port)); + dprintk("svc: svc_create_socket(%s, %d, %s)\n", + serv->sv_program->pg_name, protocol, + __svc_print_addr(sin, buf, sizeof(buf))); if (protocol != IPPROTO_UDP && protocol != IPPROTO_TCP) { printk(KERN_WARNING "svc: only UDP and TCP " "sockets supported\n"); - return -EINVAL; + return ERR_PTR(-EINVAL); } + type = (protocol == IPPROTO_UDP)? SOCK_DGRAM : SOCK_STREAM; + switch (sin->sa_family) { + case AF_INET6: + family = PF_INET6; + break; + case AF_INET: + family = PF_INET; + break; + default: + return ERR_PTR(-EINVAL); + } - if ((error = sock_create_kern(PF_INET, type, protocol, &sock)) < 0) - return error; + error = __sock_create(net, family, type, protocol, &sock, 1); + if (error < 0) + return ERR_PTR(error); - if (sin != NULL) { - if (type == SOCK_STREAM) - sock->sk->sk_reuse = 1; /* allow address reuse */ - error = sock->ops->bind(sock, (struct sockaddr *) sin, - sizeof(*sin)); - if (error < 0) - goto bummer; - } + svc_reclassify_socket(sock); + + /* + * If this is an PF_INET6 listener, we want to avoid + * getting requests from IPv4 remotes. Those should + * be shunted to a PF_INET listener via rpcbind. + */ + val = 1; + if (family == PF_INET6) + kernel_setsockopt(sock, SOL_IPV6, IPV6_V6ONLY, + (char *)&val, sizeof(val)); + + if (type == SOCK_STREAM) + sock->sk->sk_reuse = SK_CAN_REUSE; /* allow address reuse */ + error = kernel_bind(sock, sin, len); + if (error < 0) + goto bummer; + + newlen = len; + error = kernel_getsockname(sock, newsin, &newlen); + if (error < 0) + goto bummer; if (protocol == IPPROTO_TCP) { - if ((error = sock->ops->listen(sock, 64)) < 0) + if ((error = kernel_listen(sock, 64)) < 0) goto bummer; } - if ((svsk = svc_setup_socket(serv, sock, &error, 1)) != NULL) - return 0; - + svsk = svc_setup_socket(serv, sock, flags); + if (IS_ERR(svsk)) { + error = PTR_ERR(svsk); + goto bummer; + } + svc_xprt_set_local(&svsk->sk_xprt, newsin, newlen); + return (struct svc_xprt *)svsk; bummer: dprintk("svc: svc_create_socket error = %d\n", -error); sock_release(sock); - return error; + return ERR_PTR(error); } /* - * Remove a dead socket + * Detach the svc_sock from the socket so that no + * more callbacks occur. */ -void -svc_delete_socket(struct svc_sock *svsk) +static void svc_sock_detach(struct svc_xprt *xprt) { - struct svc_serv *serv; - struct sock *sk; + struct svc_sock *svsk = container_of(xprt, struct svc_sock, sk_xprt); + struct sock *sk = svsk->sk_sk; + wait_queue_head_t *wq; - dprintk("svc: svc_delete_socket(%p)\n", svsk); - - serv = svsk->sk_server; - sk = svsk->sk_sk; + dprintk("svc: svc_sock_detach(%p)\n", svsk); + /* put back the old socket callbacks */ sk->sk_state_change = svsk->sk_ostate; sk->sk_data_ready = svsk->sk_odata; sk->sk_write_space = svsk->sk_owspace; - spin_lock_bh(&serv->sv_lock); + wq = sk_sleep(sk); + if (wq && waitqueue_active(wq)) + wake_up_interruptible(wq); +} - list_del_init(&svsk->sk_list); - list_del_init(&svsk->sk_ready); - if (!test_and_set_bit(SK_DEAD, &svsk->sk_flags)) - if (test_bit(SK_TEMP, &svsk->sk_flags)) - serv->sv_tmpcnt--; +/* + * Disconnect the socket, and reset the callbacks + */ +static void svc_tcp_sock_detach(struct svc_xprt *xprt) +{ + struct svc_sock *svsk = container_of(xprt, struct svc_sock, sk_xprt); - if (!svsk->sk_inuse) { - spin_unlock_bh(&serv->sv_lock); - sock_release(svsk->sk_sock); - kfree(svsk); - } else { - spin_unlock_bh(&serv->sv_lock); - dprintk(KERN_NOTICE "svc: server socket destroy delayed\n"); - /* svsk->sk_server = NULL; */ + dprintk("svc: svc_tcp_sock_detach(%p)\n", svsk); + + svc_sock_detach(xprt); + + if (!test_bit(XPT_LISTENER, &xprt->xpt_flags)) { + svc_tcp_clear_pages(svsk); + kernel_sock_shutdown(svsk->sk_sock, SHUT_RDWR); } } /* - * Make a socket for nfsd and lockd + * Free the svc_sock's socket resources and the svc_sock itself. */ -int -svc_makesock(struct svc_serv *serv, int protocol, unsigned short port) +static void svc_sock_free(struct svc_xprt *xprt) { - struct sockaddr_in sin; + struct svc_sock *svsk = container_of(xprt, struct svc_sock, sk_xprt); + dprintk("svc: svc_sock_free(%p)\n", svsk); - dprintk("svc: creating socket proto = %d\n", protocol); - sin.sin_family = AF_INET; - sin.sin_addr.s_addr = INADDR_ANY; - sin.sin_port = htons(port); - return svc_create_socket(serv, protocol, &sin); + if (svsk->sk_sock->file) + sockfd_put(svsk->sk_sock); + else + sock_release(svsk->sk_sock); + kfree(svsk); } +#if defined(CONFIG_SUNRPC_BACKCHANNEL) /* - * Handle defer and revisit of requests + * Create a back channel svc_xprt which shares the fore channel socket. */ - -static void svc_revisit(struct cache_deferred_req *dreq, int too_many) +static struct svc_xprt *svc_bc_create_socket(struct svc_serv *serv, + int protocol, + struct net *net, + struct sockaddr *sin, int len, + int flags) { - struct svc_deferred_req *dr = container_of(dreq, struct svc_deferred_req, handle); - struct svc_serv *serv = dreq->owner; struct svc_sock *svsk; + struct svc_xprt *xprt; - if (too_many) { - svc_sock_put(dr->svsk); - kfree(dr); - return; + if (protocol != IPPROTO_TCP) { + printk(KERN_WARNING "svc: only TCP sockets" + " supported on shared back channel\n"); + return ERR_PTR(-EINVAL); } - dprintk("revisit queued\n"); - svsk = dr->svsk; - dr->svsk = NULL; - spin_lock_bh(&serv->sv_lock); - list_add(&dr->handle.recent, &svsk->sk_deferred); - spin_unlock_bh(&serv->sv_lock); - set_bit(SK_DEFERRED, &svsk->sk_flags); - svc_sock_enqueue(svsk); - svc_sock_put(svsk); -} -static struct cache_deferred_req * -svc_defer(struct cache_req *req) -{ - struct svc_rqst *rqstp = container_of(req, struct svc_rqst, rq_chandle); - int size = sizeof(struct svc_deferred_req) + (rqstp->rq_arg.len); - struct svc_deferred_req *dr; - - if (rqstp->rq_arg.page_len) - return NULL; /* if more than a page, give up FIXME */ - if (rqstp->rq_deferred) { - dr = rqstp->rq_deferred; - rqstp->rq_deferred = NULL; - } else { - int skip = rqstp->rq_arg.len - rqstp->rq_arg.head[0].iov_len; - /* FIXME maybe discard if size too large */ - dr = kmalloc(size, GFP_KERNEL); - if (dr == NULL) - return NULL; - - dr->handle.owner = rqstp->rq_server; - dr->prot = rqstp->rq_prot; - dr->addr = rqstp->rq_addr; - dr->argslen = rqstp->rq_arg.len >> 2; - memcpy(dr->args, rqstp->rq_arg.head[0].iov_base-skip, dr->argslen<<2); - } - spin_lock_bh(&rqstp->rq_server->sv_lock); - rqstp->rq_sock->sk_inuse++; - dr->svsk = rqstp->rq_sock; - spin_unlock_bh(&rqstp->rq_server->sv_lock); + svsk = kzalloc(sizeof(*svsk), GFP_KERNEL); + if (!svsk) + return ERR_PTR(-ENOMEM); + + xprt = &svsk->sk_xprt; + svc_xprt_init(net, &svc_tcp_bc_class, xprt, serv); + + serv->sv_bc_xprt = xprt; - dr->handle.revisit = svc_revisit; - return &dr->handle; + return xprt; } /* - * recv data from a deferred request into an active one + * Free a back channel svc_sock. */ -static int svc_deferred_recv(struct svc_rqst *rqstp) +static void svc_bc_sock_free(struct svc_xprt *xprt) { - struct svc_deferred_req *dr = rqstp->rq_deferred; - - rqstp->rq_arg.head[0].iov_base = dr->args; - rqstp->rq_arg.head[0].iov_len = dr->argslen<<2; - rqstp->rq_arg.page_len = 0; - rqstp->rq_arg.len = dr->argslen<<2; - rqstp->rq_prot = dr->prot; - rqstp->rq_addr = dr->addr; - return dr->argslen<<2; -} - - -static struct svc_deferred_req *svc_deferred_dequeue(struct svc_sock *svsk) -{ - struct svc_deferred_req *dr = NULL; - struct svc_serv *serv = svsk->sk_server; - - if (!test_bit(SK_DEFERRED, &svsk->sk_flags)) - return NULL; - spin_lock_bh(&serv->sv_lock); - clear_bit(SK_DEFERRED, &svsk->sk_flags); - if (!list_empty(&svsk->sk_deferred)) { - dr = list_entry(svsk->sk_deferred.next, - struct svc_deferred_req, - handle.recent); - list_del_init(&dr->handle.recent); - set_bit(SK_DEFERRED, &svsk->sk_flags); - } - spin_unlock_bh(&serv->sv_lock); - return dr; + if (xprt) + kfree(container_of(xprt, struct svc_sock, sk_xprt)); } +#endif /* CONFIG_SUNRPC_BACKCHANNEL */ diff --git a/net/sunrpc/sysctl.c b/net/sunrpc/sysctl.c index 1065904841f..c99c58e2ee6 100644 --- a/net/sunrpc/sysctl.c +++ b/net/sunrpc/sysctl.c @@ -7,7 +7,6 @@ * impossible at the moment. */ -#include <linux/config.h> #include <linux/types.h> #include <linux/linkage.h> #include <linux/ctype.h> @@ -19,32 +18,35 @@ #include <linux/sunrpc/types.h> #include <linux/sunrpc/sched.h> #include <linux/sunrpc/stats.h> -#include <linux/sunrpc/xprt.h> +#include <linux/sunrpc/svc_xprt.h> + +#include "netns.h" /* * Declare the debug flags here */ unsigned int rpc_debug; +EXPORT_SYMBOL_GPL(rpc_debug); + unsigned int nfs_debug; +EXPORT_SYMBOL_GPL(nfs_debug); + unsigned int nfsd_debug; +EXPORT_SYMBOL_GPL(nfsd_debug); + unsigned int nlm_debug; +EXPORT_SYMBOL_GPL(nlm_debug); #ifdef RPC_DEBUG static struct ctl_table_header *sunrpc_table_header; -static ctl_table sunrpc_table[]; +static struct ctl_table sunrpc_table[]; void rpc_register_sysctl(void) { - if (!sunrpc_table_header) { - sunrpc_table_header = register_sysctl_table(sunrpc_table, 1); -#ifdef CONFIG_PROC_FS - if (sunrpc_table[0].de) - sunrpc_table[0].de->owner = THIS_MODULE; -#endif - } - + if (!sunrpc_table_header) + sunrpc_table_header = register_sysctl_table(sunrpc_table); } void @@ -56,8 +58,22 @@ rpc_unregister_sysctl(void) } } +static int proc_do_xprt(struct ctl_table *table, int write, + void __user *buffer, size_t *lenp, loff_t *ppos) +{ + char tmpbuf[256]; + size_t len; + + if ((*ppos && !write) || !*lenp) { + *lenp = 0; + return 0; + } + len = svc_print_xprts(tmpbuf, sizeof(tmpbuf)); + return simple_read_from_buffer(buffer, *lenp, ppos, tmpbuf, len); +} + static int -proc_dodebug(ctl_table *table, int write, struct file *file, +proc_dodebug(struct ctl_table *table, int write, void __user *buffer, size_t *lenp, loff_t *ppos) { char tmpbuf[20], c, *s; @@ -95,9 +111,8 @@ proc_dodebug(ctl_table *table, int write, struct file *file, left--, s++; *(unsigned int *) table->data = value; /* Display the RPC tasks on writing to rpc_debug */ - if (table->ctl_name == CTL_RPCDEBUG) { - rpc_show_tasks(); - } + if (strcmp(table->procname, "rpc_debug") == 0) + rpc_show_tasks(&init_net); } else { if (!access_ok(VERIFY_WRITE, buffer, left)) return -EFAULT; @@ -120,99 +135,51 @@ done: } -static unsigned int min_slot_table_size = RPC_MIN_SLOT_TABLE; -static unsigned int max_slot_table_size = RPC_MAX_SLOT_TABLE; -static unsigned int xprt_min_resvport_limit = RPC_MIN_RESVPORT; -static unsigned int xprt_max_resvport_limit = RPC_MAX_RESVPORT; - -static ctl_table debug_table[] = { +static struct ctl_table debug_table[] = { { - .ctl_name = CTL_RPCDEBUG, .procname = "rpc_debug", .data = &rpc_debug, .maxlen = sizeof(int), .mode = 0644, - .proc_handler = &proc_dodebug - }, + .proc_handler = proc_dodebug + }, { - .ctl_name = CTL_NFSDEBUG, .procname = "nfs_debug", .data = &nfs_debug, .maxlen = sizeof(int), .mode = 0644, - .proc_handler = &proc_dodebug - }, + .proc_handler = proc_dodebug + }, { - .ctl_name = CTL_NFSDDEBUG, .procname = "nfsd_debug", .data = &nfsd_debug, .maxlen = sizeof(int), .mode = 0644, - .proc_handler = &proc_dodebug - }, + .proc_handler = proc_dodebug + }, { - .ctl_name = CTL_NLMDEBUG, .procname = "nlm_debug", .data = &nlm_debug, .maxlen = sizeof(int), .mode = 0644, - .proc_handler = &proc_dodebug - }, - { - .ctl_name = CTL_SLOTTABLE_UDP, - .procname = "udp_slot_table_entries", - .data = &xprt_udp_slot_table_entries, - .maxlen = sizeof(unsigned int), - .mode = 0644, - .proc_handler = &proc_dointvec_minmax, - .strategy = &sysctl_intvec, - .extra1 = &min_slot_table_size, - .extra2 = &max_slot_table_size + .proc_handler = proc_dodebug }, { - .ctl_name = CTL_SLOTTABLE_TCP, - .procname = "tcp_slot_table_entries", - .data = &xprt_tcp_slot_table_entries, - .maxlen = sizeof(unsigned int), - .mode = 0644, - .proc_handler = &proc_dointvec_minmax, - .strategy = &sysctl_intvec, - .extra1 = &min_slot_table_size, - .extra2 = &max_slot_table_size - }, - { - .ctl_name = CTL_MIN_RESVPORT, - .procname = "min_resvport", - .data = &xprt_min_resvport, - .maxlen = sizeof(unsigned int), - .mode = 0644, - .proc_handler = &proc_dointvec_minmax, - .strategy = &sysctl_intvec, - .extra1 = &xprt_min_resvport_limit, - .extra2 = &xprt_max_resvport_limit - }, - { - .ctl_name = CTL_MAX_RESVPORT, - .procname = "max_resvport", - .data = &xprt_max_resvport, - .maxlen = sizeof(unsigned int), - .mode = 0644, - .proc_handler = &proc_dointvec_minmax, - .strategy = &sysctl_intvec, - .extra1 = &xprt_min_resvport_limit, - .extra2 = &xprt_max_resvport_limit + .procname = "transports", + .maxlen = 256, + .mode = 0444, + .proc_handler = proc_do_xprt, }, - { .ctl_name = 0 } + { } }; -static ctl_table sunrpc_table[] = { +static struct ctl_table sunrpc_table[] = { { - .ctl_name = CTL_SUNRPC, .procname = "sunrpc", .mode = 0555, .child = debug_table }, - { .ctl_name = 0 } + { } }; #endif diff --git a/net/sunrpc/timer.c b/net/sunrpc/timer.c index bcbdf6430d5..08881d0c967 100644 --- a/net/sunrpc/timer.c +++ b/net/sunrpc/timer.c @@ -17,20 +17,24 @@ #include <linux/types.h> #include <linux/unistd.h> +#include <linux/module.h> #include <linux/sunrpc/clnt.h> -#include <linux/sunrpc/xprt.h> -#include <linux/sunrpc/timer.h> #define RPC_RTO_MAX (60*HZ) #define RPC_RTO_INIT (HZ/5) #define RPC_RTO_MIN (HZ/10) -void -rpc_init_rtt(struct rpc_rtt *rt, unsigned long timeo) +/** + * rpc_init_rtt - Initialize an RPC RTT estimator context + * @rt: context to initialize + * @timeo: initial timeout value, in jiffies + * + */ +void rpc_init_rtt(struct rpc_rtt *rt, unsigned long timeo) { unsigned long init = 0; - unsigned i; + unsigned int i; rt->timeo = timeo; @@ -42,13 +46,18 @@ rpc_init_rtt(struct rpc_rtt *rt, unsigned long timeo) rt->ntimeouts[i] = 0; } } +EXPORT_SYMBOL_GPL(rpc_init_rtt); -/* +/** + * rpc_update_rtt - Update an RPC RTT estimator context + * @rt: context to update + * @timer: timer array index (request type) + * @m: recent actual RTT, in jiffies + * * NB: When computing the smoothed RTT and standard deviation, * be careful not to produce negative intermediate results. */ -void -rpc_update_rtt(struct rpc_rtt *rt, unsigned timer, long m) +void rpc_update_rtt(struct rpc_rtt *rt, unsigned int timer, long m) { long *srtt, *sdrtt; @@ -77,22 +86,27 @@ rpc_update_rtt(struct rpc_rtt *rt, unsigned timer, long m) if (*sdrtt < RPC_RTO_MIN) *sdrtt = RPC_RTO_MIN; } +EXPORT_SYMBOL_GPL(rpc_update_rtt); -/* - * Estimate rto for an nfs rpc sent via. an unreliable datagram. - * Use the mean and mean deviation of rtt for the appropriate type of rpc - * for the frequent rpcs and a default for the others. - * The justification for doing "other" this way is that these rpcs - * happen so infrequently that timer est. would probably be stale. - * Also, since many of these rpcs are - * non-idempotent, a conservative timeout is desired. +/** + * rpc_calc_rto - Provide an estimated timeout value + * @rt: context to use for calculation + * @timer: timer array index (request type) + * + * Estimate RTO for an NFS RPC sent via an unreliable datagram. Use + * the mean and mean deviation of RTT for the appropriate type of RPC + * for frequently issued RPCs, and a fixed default for the others. + * + * The justification for doing "other" this way is that these RPCs + * happen so infrequently that timer estimation would probably be + * stale. Also, since many of these RPCs are non-idempotent, a + * conservative timeout is desired. + * * getattr, lookup, * read, write, commit - A+4D * other - timeo */ - -unsigned long -rpc_calc_rto(struct rpc_rtt *rt, unsigned timer) +unsigned long rpc_calc_rto(struct rpc_rtt *rt, unsigned int timer) { unsigned long res; @@ -105,3 +119,4 @@ rpc_calc_rto(struct rpc_rtt *rt, unsigned timer) return res; } +EXPORT_SYMBOL_GPL(rpc_calc_rto); diff --git a/net/sunrpc/xdr.c b/net/sunrpc/xdr.c index 32df43372ee..23fb4e75e24 100644 --- a/net/sunrpc/xdr.c +++ b/net/sunrpc/xdr.c @@ -7,6 +7,7 @@ */ #include <linux/module.h> +#include <linux/slab.h> #include <linux/types.h> #include <linux/string.h> #include <linux/kernel.h> @@ -18,28 +19,30 @@ /* * XDR functions for basic NFS types */ -u32 * -xdr_encode_netobj(u32 *p, const struct xdr_netobj *obj) +__be32 * +xdr_encode_netobj(__be32 *p, const struct xdr_netobj *obj) { unsigned int quadlen = XDR_QUADLEN(obj->len); p[quadlen] = 0; /* zero trailing bytes */ - *p++ = htonl(obj->len); + *p++ = cpu_to_be32(obj->len); memcpy(p, obj->data, obj->len); return p + XDR_QUADLEN(obj->len); } +EXPORT_SYMBOL_GPL(xdr_encode_netobj); -u32 * -xdr_decode_netobj(u32 *p, struct xdr_netobj *obj) +__be32 * +xdr_decode_netobj(__be32 *p, struct xdr_netobj *obj) { unsigned int len; - if ((len = ntohl(*p++)) > XDR_MAX_NETOBJ) + if ((len = be32_to_cpu(*p++)) > XDR_MAX_NETOBJ) return NULL; obj->len = len; obj->data = (u8 *) p; return p + XDR_QUADLEN(len); } +EXPORT_SYMBOL_GPL(xdr_decode_netobj); /** * xdr_encode_opaque_fixed - Encode fixed length opaque data @@ -55,7 +58,7 @@ xdr_decode_netobj(u32 *p, struct xdr_netobj *obj) * Returns the updated current XDR buffer position * */ -u32 *xdr_encode_opaque_fixed(u32 *p, const void *ptr, unsigned int nbytes) +__be32 *xdr_encode_opaque_fixed(__be32 *p, const void *ptr, unsigned int nbytes) { if (likely(nbytes != 0)) { unsigned int quadlen = XDR_QUADLEN(nbytes); @@ -69,7 +72,7 @@ u32 *xdr_encode_opaque_fixed(u32 *p, const void *ptr, unsigned int nbytes) } return p; } -EXPORT_SYMBOL(xdr_encode_opaque_fixed); +EXPORT_SYMBOL_GPL(xdr_encode_opaque_fixed); /** * xdr_encode_opaque - Encode variable length opaque data @@ -79,78 +82,51 @@ EXPORT_SYMBOL(xdr_encode_opaque_fixed); * * Returns the updated current XDR buffer position */ -u32 *xdr_encode_opaque(u32 *p, const void *ptr, unsigned int nbytes) +__be32 *xdr_encode_opaque(__be32 *p, const void *ptr, unsigned int nbytes) { - *p++ = htonl(nbytes); + *p++ = cpu_to_be32(nbytes); return xdr_encode_opaque_fixed(p, ptr, nbytes); } -EXPORT_SYMBOL(xdr_encode_opaque); +EXPORT_SYMBOL_GPL(xdr_encode_opaque); -u32 * -xdr_encode_string(u32 *p, const char *string) +__be32 * +xdr_encode_string(__be32 *p, const char *string) { return xdr_encode_array(p, string, strlen(string)); } +EXPORT_SYMBOL_GPL(xdr_encode_string); -u32 * -xdr_decode_string(u32 *p, char **sp, int *lenp, int maxlen) +__be32 * +xdr_decode_string_inplace(__be32 *p, char **sp, + unsigned int *lenp, unsigned int maxlen) { - unsigned int len; - char *string; - - if ((len = ntohl(*p++)) > maxlen) - return NULL; - if (lenp) - *lenp = len; - if ((len % 4) != 0) { - string = (char *) p; - } else { - string = (char *) (p - 1); - memmove(string, p, len); - } - string[len] = '\0'; - *sp = string; - return p + XDR_QUADLEN(len); -} - -u32 * -xdr_decode_string_inplace(u32 *p, char **sp, int *lenp, int maxlen) -{ - unsigned int len; + u32 len; - if ((len = ntohl(*p++)) > maxlen) + len = be32_to_cpu(*p++); + if (len > maxlen) return NULL; *lenp = len; *sp = (char *) p; return p + XDR_QUADLEN(len); } +EXPORT_SYMBOL_GPL(xdr_decode_string_inplace); +/** + * xdr_terminate_string - '\0'-terminate a string residing in an xdr_buf + * @buf: XDR buffer where string resides + * @len: length of string, in bytes + * + */ void -xdr_encode_pages(struct xdr_buf *xdr, struct page **pages, unsigned int base, - unsigned int len) +xdr_terminate_string(struct xdr_buf *buf, const u32 len) { - struct kvec *tail = xdr->tail; - u32 *p; - - xdr->pages = pages; - xdr->page_base = base; - xdr->page_len = len; + char *kaddr; - p = (u32 *)xdr->head[0].iov_base + XDR_QUADLEN(xdr->head[0].iov_len); - tail->iov_base = p; - tail->iov_len = 0; - - if (len & 3) { - unsigned int pad = 4 - (len & 3); - - *p = 0; - tail->iov_base = (char *)p + (len & 3); - tail->iov_len = pad; - len += pad; - } - xdr->buflen += len; - xdr->len += len; + kaddr = kmap_atomic(buf->pages[0]); + kaddr[buf->page_base + len] = '\0'; + kunmap_atomic(kaddr); } +EXPORT_SYMBOL_GPL(xdr_terminate_string); void xdr_inline_pages(struct xdr_buf *xdr, unsigned int offset, @@ -172,11 +148,13 @@ xdr_inline_pages(struct xdr_buf *xdr, unsigned int offset, xdr->buflen += len; } - +EXPORT_SYMBOL_GPL(xdr_inline_pages); /* * Helper routines for doing 'memmove' like operations on a struct xdr_buf - * + */ + +/** * _shift_data_right_pages * @pages: vector of pages containing both the source and dest memory area. * @pgto_base: page vector address of destination @@ -212,7 +190,6 @@ _shift_data_right_pages(struct page **pages, size_t pgto_base, do { /* Are any pointers crossing a page boundary? */ if (pgto_base == 0) { - flush_dcache_page(*pgto); pgto_base = PAGE_CACHE_SIZE; pgto--; } @@ -229,17 +206,20 @@ _shift_data_right_pages(struct page **pages, size_t pgto_base, pgto_base -= copy; pgfrom_base -= copy; - vto = kmap_atomic(*pgto, KM_USER0); - vfrom = kmap_atomic(*pgfrom, KM_USER1); - memmove(vto + pgto_base, vfrom + pgfrom_base, copy); - kunmap_atomic(vfrom, KM_USER1); - kunmap_atomic(vto, KM_USER0); + vto = kmap_atomic(*pgto); + if (*pgto != *pgfrom) { + vfrom = kmap_atomic(*pgfrom); + memcpy(vto + pgto_base, vfrom + pgfrom_base, copy); + kunmap_atomic(vfrom); + } else + memmove(vto + pgto_base, vto + pgfrom_base, copy); + flush_dcache_page(*pgto); + kunmap_atomic(vto); } while ((len -= copy) != 0); - flush_dcache_page(*pgto); } -/* +/** * _copy_to_pages * @pages: array of pages * @pgbase: page vector address of destination @@ -259,14 +239,18 @@ _copy_to_pages(struct page **pages, size_t pgbase, const char *p, size_t len) pgto = pages + (pgbase >> PAGE_CACHE_SHIFT); pgbase &= ~PAGE_CACHE_MASK; - do { + for (;;) { copy = PAGE_CACHE_SIZE - pgbase; if (copy > len) copy = len; - vto = kmap_atomic(*pgto, KM_USER0); + vto = kmap_atomic(*pgto); memcpy(vto + pgbase, p, copy); - kunmap_atomic(vto, KM_USER0); + kunmap_atomic(vto); + + len -= copy; + if (len == 0) + break; pgbase += copy; if (pgbase == PAGE_CACHE_SIZE) { @@ -275,12 +259,11 @@ _copy_to_pages(struct page **pages, size_t pgbase, const char *p, size_t len) pgto++; } p += copy; - - } while ((len -= copy) != 0); + } flush_dcache_page(*pgto); } -/* +/** * _copy_from_pages * @p: pointer to destination * @pages: array of pages @@ -290,7 +273,7 @@ _copy_to_pages(struct page **pages, size_t pgbase, const char *p, size_t len) * Copies data into an arbitrary memory location from an array of pages * The copy is assumed to be non-overlapping. */ -static void +void _copy_from_pages(char *p, struct page **pages, size_t pgbase, size_t len) { struct page **pgfrom; @@ -305,9 +288,9 @@ _copy_from_pages(char *p, struct page **pages, size_t pgbase, size_t len) if (copy > len) copy = len; - vfrom = kmap_atomic(*pgfrom, KM_USER0); + vfrom = kmap_atomic(*pgfrom); memcpy(p, vfrom + pgbase, copy); - kunmap_atomic(vfrom, KM_USER0); + kunmap_atomic(vfrom); pgbase += copy; if (pgbase == PAGE_CACHE_SIZE) { @@ -318,13 +301,14 @@ _copy_from_pages(char *p, struct page **pages, size_t pgbase, size_t len) } while ((len -= copy) != 0); } +EXPORT_SYMBOL_GPL(_copy_from_pages); -/* +/** * xdr_shrink_bufhead * @buf: xdr_buf * @len: bytes to remove from buf->head[0] * - * Shrinks XDR buffer's header kvec buf->head[0] by + * Shrinks XDR buffer's header kvec buf->head[0] by * 'len' bytes. The extra data is not lost, but is instead * moved into the inlined pages and/or the tail. */ @@ -337,7 +321,10 @@ xdr_shrink_bufhead(struct xdr_buf *buf, size_t len) tail = buf->tail; head = buf->head; - BUG_ON (len > head->iov_len); + + WARN_ON_ONCE(len > head->iov_len); + if (len > head->iov_len) + len = head->iov_len; /* Shift the tail first */ if (tail->iov_len != 0) { @@ -392,12 +379,12 @@ xdr_shrink_bufhead(struct xdr_buf *buf, size_t len) buf->len = buf->buflen; } -/* +/** * xdr_shrink_pagelen * @buf: xdr_buf * @len: bytes to remove from buf->pages * - * Shrinks XDR buffer's page array buf->pages by + * Shrinks XDR buffer's page array buf->pages by * 'len' bytes. The extra data is not lost, but is instead * moved into the tail. */ @@ -406,24 +393,29 @@ xdr_shrink_pagelen(struct xdr_buf *buf, size_t len) { struct kvec *tail; size_t copy; - char *p; unsigned int pglen = buf->page_len; + unsigned int tailbuf_len; tail = buf->tail; BUG_ON (len > pglen); + tailbuf_len = buf->buflen - buf->head->iov_len - buf->page_len; + /* Shift the tail first */ - if (tail->iov_len != 0) { - p = (char *)tail->iov_base + len; + if (tailbuf_len != 0) { + unsigned int free_space = tailbuf_len - tail->iov_len; + + if (len < free_space) + free_space = len; + tail->iov_len += free_space; + + copy = len; if (tail->iov_len > len) { - copy = tail->iov_len - len; - memmove(p, tail->iov_base, copy); + char *p = (char *)tail->iov_base + len; + memmove(p, tail->iov_base, tail->iov_len - len); } else - buf->buflen -= len; - /* Copy from the inlined pages into the tail */ - copy = len; - if (copy > tail->iov_len) copy = tail->iov_len; + /* Copy from the inlined pages into the tail */ _copy_from_pages((char *)tail->iov_base, buf->pages, buf->page_base + pglen - len, copy); @@ -440,6 +432,17 @@ xdr_shift_buf(struct xdr_buf *buf, size_t len) { xdr_shrink_bufhead(buf, len); } +EXPORT_SYMBOL_GPL(xdr_shift_buf); + +/** + * xdr_stream_pos - Return the current offset from the start of the xdr_stream + * @xdr: pointer to struct xdr_stream + */ +unsigned int xdr_stream_pos(const struct xdr_stream *xdr) +{ + return (unsigned int)(XDR_QUADLEN(xdr->buf->len) - xdr->nwords) << 2; +} +EXPORT_SYMBOL_GPL(xdr_stream_pos); /** * xdr_init_encode - Initialize a struct xdr_stream for sending data. @@ -454,16 +457,17 @@ xdr_shift_buf(struct xdr_buf *buf, size_t len) * of the buffer length, and takes care of adjusting the kvec * length for us. */ -void xdr_init_encode(struct xdr_stream *xdr, struct xdr_buf *buf, uint32_t *p) +void xdr_init_encode(struct xdr_stream *xdr, struct xdr_buf *buf, __be32 *p) { struct kvec *iov = buf->head; int scratch_len = buf->buflen - buf->page_len - buf->tail[0].iov_len; + xdr_set_scratch_buffer(xdr, NULL, 0); BUG_ON(scratch_len < 0); xdr->buf = buf; xdr->iov = iov; - xdr->p = (uint32_t *)((char *)iov->iov_base + iov->iov_len); - xdr->end = (uint32_t *)((char *)iov->iov_base + scratch_len); + xdr->p = (__be32 *)((char *)iov->iov_base + iov->iov_len); + xdr->end = (__be32 *)((char *)iov->iov_base + scratch_len); BUG_ON(iov->iov_len > scratch_len); if (p != xdr->p && p != NULL) { @@ -476,7 +480,74 @@ void xdr_init_encode(struct xdr_stream *xdr, struct xdr_buf *buf, uint32_t *p) iov->iov_len += len; } } -EXPORT_SYMBOL(xdr_init_encode); +EXPORT_SYMBOL_GPL(xdr_init_encode); + +/** + * xdr_commit_encode - Ensure all data is written to buffer + * @xdr: pointer to xdr_stream + * + * We handle encoding across page boundaries by giving the caller a + * temporary location to write to, then later copying the data into + * place; xdr_commit_encode does that copying. + * + * Normally the caller doesn't need to call this directly, as the + * following xdr_reserve_space will do it. But an explicit call may be + * required at the end of encoding, or any other time when the xdr_buf + * data might be read. + */ +void xdr_commit_encode(struct xdr_stream *xdr) +{ + int shift = xdr->scratch.iov_len; + void *page; + + if (shift == 0) + return; + page = page_address(*xdr->page_ptr); + memcpy(xdr->scratch.iov_base, page, shift); + memmove(page, page + shift, (void *)xdr->p - page); + xdr->scratch.iov_len = 0; +} +EXPORT_SYMBOL_GPL(xdr_commit_encode); + +__be32 *xdr_get_next_encode_buffer(struct xdr_stream *xdr, size_t nbytes) +{ + static __be32 *p; + int space_left; + int frag1bytes, frag2bytes; + + if (nbytes > PAGE_SIZE) + return NULL; /* Bigger buffers require special handling */ + if (xdr->buf->len + nbytes > xdr->buf->buflen) + return NULL; /* Sorry, we're totally out of space */ + frag1bytes = (xdr->end - xdr->p) << 2; + frag2bytes = nbytes - frag1bytes; + if (xdr->iov) + xdr->iov->iov_len += frag1bytes; + else + xdr->buf->page_len += frag1bytes; + xdr->page_ptr++; + xdr->iov = NULL; + /* + * If the last encode didn't end exactly on a page boundary, the + * next one will straddle boundaries. Encode into the next + * page, then copy it back later in xdr_commit_encode. We use + * the "scratch" iov to track any temporarily unused fragment of + * space at the end of the previous buffer: + */ + xdr->scratch.iov_base = xdr->p; + xdr->scratch.iov_len = frag1bytes; + p = page_address(*xdr->page_ptr); + /* + * Note this is where the next encode will start after we've + * shifted this one back: + */ + xdr->p = (void *)p + frag2bytes; + space_left = xdr->buf->buflen - xdr->buf->len; + xdr->end = (void *)p + min_t(int, space_left, PAGE_SIZE); + xdr->buf->page_len += frag2bytes; + xdr->buf->len += nbytes; + return p; +} /** * xdr_reserve_space - Reserve buffer space for sending @@ -487,23 +558,125 @@ EXPORT_SYMBOL(xdr_init_encode); * bytes of data. If so, update the total xdr_buf length, and * adjust the length of the current kvec. */ -uint32_t * xdr_reserve_space(struct xdr_stream *xdr, size_t nbytes) +__be32 * xdr_reserve_space(struct xdr_stream *xdr, size_t nbytes) { - uint32_t *p = xdr->p; - uint32_t *q; + __be32 *p = xdr->p; + __be32 *q; + xdr_commit_encode(xdr); /* align nbytes on the next 32-bit boundary */ nbytes += 3; nbytes &= ~3; q = p + (nbytes >> 2); if (unlikely(q > xdr->end || q < p)) - return NULL; + return xdr_get_next_encode_buffer(xdr, nbytes); xdr->p = q; - xdr->iov->iov_len += nbytes; + if (xdr->iov) + xdr->iov->iov_len += nbytes; + else + xdr->buf->page_len += nbytes; xdr->buf->len += nbytes; return p; } -EXPORT_SYMBOL(xdr_reserve_space); +EXPORT_SYMBOL_GPL(xdr_reserve_space); + +/** + * xdr_truncate_encode - truncate an encode buffer + * @xdr: pointer to xdr_stream + * @len: new length of buffer + * + * Truncates the xdr stream, so that xdr->buf->len == len, + * and xdr->p points at offset len from the start of the buffer, and + * head, tail, and page lengths are adjusted to correspond. + * + * If this means moving xdr->p to a different buffer, we assume that + * that the end pointer should be set to the end of the current page, + * except in the case of the head buffer when we assume the head + * buffer's current length represents the end of the available buffer. + * + * This is *not* safe to use on a buffer that already has inlined page + * cache pages (as in a zero-copy server read reply), except for the + * simple case of truncating from one position in the tail to another. + * + */ +void xdr_truncate_encode(struct xdr_stream *xdr, size_t len) +{ + struct xdr_buf *buf = xdr->buf; + struct kvec *head = buf->head; + struct kvec *tail = buf->tail; + int fraglen; + int new, old; + + if (len > buf->len) { + WARN_ON_ONCE(1); + return; + } + xdr_commit_encode(xdr); + + fraglen = min_t(int, buf->len - len, tail->iov_len); + tail->iov_len -= fraglen; + buf->len -= fraglen; + if (tail->iov_len && buf->len == len) { + xdr->p = tail->iov_base + tail->iov_len; + /* xdr->end, xdr->iov should be set already */ + return; + } + WARN_ON_ONCE(fraglen); + fraglen = min_t(int, buf->len - len, buf->page_len); + buf->page_len -= fraglen; + buf->len -= fraglen; + + new = buf->page_base + buf->page_len; + old = new + fraglen; + xdr->page_ptr -= (old >> PAGE_SHIFT) - (new >> PAGE_SHIFT); + + if (buf->page_len && buf->len == len) { + xdr->p = page_address(*xdr->page_ptr); + xdr->end = (void *)xdr->p + PAGE_SIZE; + xdr->p = (void *)xdr->p + (new % PAGE_SIZE); + /* xdr->iov should already be NULL */ + return; + } + if (fraglen) { + xdr->end = head->iov_base + head->iov_len; + xdr->page_ptr--; + } + /* (otherwise assume xdr->end is already set) */ + head->iov_len = len; + buf->len = len; + xdr->p = head->iov_base + head->iov_len; + xdr->iov = buf->head; +} +EXPORT_SYMBOL(xdr_truncate_encode); + +/** + * xdr_restrict_buflen - decrease available buffer space + * @xdr: pointer to xdr_stream + * @newbuflen: new maximum number of bytes available + * + * Adjust our idea of how much space is available in the buffer. + * If we've already used too much space in the buffer, returns -1. + * If the available space is already smaller than newbuflen, returns 0 + * and does nothing. Otherwise, adjusts xdr->buf->buflen to newbuflen + * and ensures xdr->end is set at most offset newbuflen from the start + * of the buffer. + */ +int xdr_restrict_buflen(struct xdr_stream *xdr, int newbuflen) +{ + struct xdr_buf *buf = xdr->buf; + int left_in_this_buf = (void *)xdr->end - (void *)xdr->p; + int end_offset = buf->len + left_in_this_buf; + + if (newbuflen < 0 || newbuflen < buf->len) + return -1; + if (newbuflen > buf->buflen) + return 0; + if (newbuflen < end_offset) + xdr->end = (void *)xdr->end + newbuflen - end_offset; + buf->buflen = newbuflen; + return 0; +} +EXPORT_SYMBOL(xdr_restrict_buflen); /** * xdr_write_pages - Insert a list of pages into an XDR buffer for sending @@ -538,7 +711,73 @@ void xdr_write_pages(struct xdr_stream *xdr, struct page **pages, unsigned int b buf->buflen += len; buf->len += len; } -EXPORT_SYMBOL(xdr_write_pages); +EXPORT_SYMBOL_GPL(xdr_write_pages); + +static void xdr_set_iov(struct xdr_stream *xdr, struct kvec *iov, + unsigned int len) +{ + if (len > iov->iov_len) + len = iov->iov_len; + xdr->p = (__be32*)iov->iov_base; + xdr->end = (__be32*)(iov->iov_base + len); + xdr->iov = iov; + xdr->page_ptr = NULL; +} + +static int xdr_set_page_base(struct xdr_stream *xdr, + unsigned int base, unsigned int len) +{ + unsigned int pgnr; + unsigned int maxlen; + unsigned int pgoff; + unsigned int pgend; + void *kaddr; + + maxlen = xdr->buf->page_len; + if (base >= maxlen) + return -EINVAL; + maxlen -= base; + if (len > maxlen) + len = maxlen; + + base += xdr->buf->page_base; + + pgnr = base >> PAGE_SHIFT; + xdr->page_ptr = &xdr->buf->pages[pgnr]; + kaddr = page_address(*xdr->page_ptr); + + pgoff = base & ~PAGE_MASK; + xdr->p = (__be32*)(kaddr + pgoff); + + pgend = pgoff + len; + if (pgend > PAGE_SIZE) + pgend = PAGE_SIZE; + xdr->end = (__be32*)(kaddr + pgend); + xdr->iov = NULL; + return 0; +} + +static void xdr_set_next_page(struct xdr_stream *xdr) +{ + unsigned int newbase; + + newbase = (1 + xdr->page_ptr - xdr->buf->pages) << PAGE_SHIFT; + newbase -= xdr->buf->page_base; + + if (xdr_set_page_base(xdr, newbase, PAGE_SIZE) < 0) + xdr_set_iov(xdr, xdr->buf->tail, xdr->buf->len); +} + +static bool xdr_set_next_buffer(struct xdr_stream *xdr) +{ + if (xdr->page_ptr != NULL) + xdr_set_next_page(xdr); + else if (xdr->iov == xdr->buf->head) { + if (xdr_set_page_base(xdr, 0, PAGE_SIZE) < 0) + xdr_set_iov(xdr, xdr->buf->tail, xdr->buf->len); + } + return xdr->p != xdr->end; +} /** * xdr_init_decode - Initialize an xdr_stream for decoding data. @@ -546,22 +785,95 @@ EXPORT_SYMBOL(xdr_write_pages); * @buf: pointer to XDR buffer from which to decode data * @p: current pointer inside XDR buffer */ -void xdr_init_decode(struct xdr_stream *xdr, struct xdr_buf *buf, uint32_t *p) +void xdr_init_decode(struct xdr_stream *xdr, struct xdr_buf *buf, __be32 *p) { - struct kvec *iov = buf->head; - unsigned int len = iov->iov_len; - - if (len > buf->len) - len = buf->len; xdr->buf = buf; - xdr->iov = iov; - xdr->p = p; - xdr->end = (uint32_t *)((char *)iov->iov_base + len); + xdr->scratch.iov_base = NULL; + xdr->scratch.iov_len = 0; + xdr->nwords = XDR_QUADLEN(buf->len); + if (buf->head[0].iov_len != 0) + xdr_set_iov(xdr, buf->head, buf->len); + else if (buf->page_len != 0) + xdr_set_page_base(xdr, 0, buf->len); + if (p != NULL && p > xdr->p && xdr->end >= p) { + xdr->nwords -= p - xdr->p; + xdr->p = p; + } +} +EXPORT_SYMBOL_GPL(xdr_init_decode); + +/** + * xdr_init_decode - Initialize an xdr_stream for decoding data. + * @xdr: pointer to xdr_stream struct + * @buf: pointer to XDR buffer from which to decode data + * @pages: list of pages to decode into + * @len: length in bytes of buffer in pages + */ +void xdr_init_decode_pages(struct xdr_stream *xdr, struct xdr_buf *buf, + struct page **pages, unsigned int len) +{ + memset(buf, 0, sizeof(*buf)); + buf->pages = pages; + buf->page_len = len; + buf->buflen = len; + buf->len = len; + xdr_init_decode(xdr, buf, NULL); +} +EXPORT_SYMBOL_GPL(xdr_init_decode_pages); + +static __be32 * __xdr_inline_decode(struct xdr_stream *xdr, size_t nbytes) +{ + unsigned int nwords = XDR_QUADLEN(nbytes); + __be32 *p = xdr->p; + __be32 *q = p + nwords; + + if (unlikely(nwords > xdr->nwords || q > xdr->end || q < p)) + return NULL; + xdr->p = q; + xdr->nwords -= nwords; + return p; } -EXPORT_SYMBOL(xdr_init_decode); /** - * xdr_inline_decode - Retrieve non-page XDR data to decode + * xdr_set_scratch_buffer - Attach a scratch buffer for decoding data. + * @xdr: pointer to xdr_stream struct + * @buf: pointer to an empty buffer + * @buflen: size of 'buf' + * + * The scratch buffer is used when decoding from an array of pages. + * If an xdr_inline_decode() call spans across page boundaries, then + * we copy the data into the scratch buffer in order to allow linear + * access. + */ +void xdr_set_scratch_buffer(struct xdr_stream *xdr, void *buf, size_t buflen) +{ + xdr->scratch.iov_base = buf; + xdr->scratch.iov_len = buflen; +} +EXPORT_SYMBOL_GPL(xdr_set_scratch_buffer); + +static __be32 *xdr_copy_to_scratch(struct xdr_stream *xdr, size_t nbytes) +{ + __be32 *p; + void *cpdest = xdr->scratch.iov_base; + size_t cplen = (char *)xdr->end - (char *)xdr->p; + + if (nbytes > xdr->scratch.iov_len) + return NULL; + memcpy(cpdest, xdr->p, cplen); + cpdest += cplen; + nbytes -= cplen; + if (!xdr_set_next_buffer(xdr)) + return NULL; + p = __xdr_inline_decode(xdr, nbytes); + if (p == NULL) + return NULL; + memcpy(cpdest, p, nbytes); + return xdr->scratch.iov_base; +} + +/** + * xdr_inline_decode - Retrieve XDR data to decode * @xdr: pointer to xdr_stream struct * @nbytes: number of bytes of data to decode * @@ -570,17 +882,50 @@ EXPORT_SYMBOL(xdr_init_decode); * If so return the current pointer, then update the current * pointer position. */ -uint32_t * xdr_inline_decode(struct xdr_stream *xdr, size_t nbytes) +__be32 * xdr_inline_decode(struct xdr_stream *xdr, size_t nbytes) { - uint32_t *p = xdr->p; - uint32_t *q = p + XDR_QUADLEN(nbytes); + __be32 *p; - if (unlikely(q > xdr->end || q < p)) + if (nbytes == 0) + return xdr->p; + if (xdr->p == xdr->end && !xdr_set_next_buffer(xdr)) return NULL; - xdr->p = q; - return p; + p = __xdr_inline_decode(xdr, nbytes); + if (p != NULL) + return p; + return xdr_copy_to_scratch(xdr, nbytes); +} +EXPORT_SYMBOL_GPL(xdr_inline_decode); + +static unsigned int xdr_align_pages(struct xdr_stream *xdr, unsigned int len) +{ + struct xdr_buf *buf = xdr->buf; + struct kvec *iov; + unsigned int nwords = XDR_QUADLEN(len); + unsigned int cur = xdr_stream_pos(xdr); + + if (xdr->nwords == 0) + return 0; + /* Realign pages to current pointer position */ + iov = buf->head; + if (iov->iov_len > cur) { + xdr_shrink_bufhead(buf, iov->iov_len - cur); + xdr->nwords = XDR_QUADLEN(buf->len - cur); + } + + if (nwords > xdr->nwords) { + nwords = xdr->nwords; + len = nwords << 2; + } + if (buf->page_len <= len) + len = buf->page_len; + else if (nwords < xdr->nwords) { + /* Truncate page data and move it into the tail */ + xdr_shrink_pagelen(buf, buf->page_len - len); + xdr->nwords = XDR_QUADLEN(buf->len - cur); + } + return len; } -EXPORT_SYMBOL(xdr_inline_decode); /** * xdr_read_pages - Ensure page-based XDR data to decode is aligned at current pointer position @@ -589,43 +934,62 @@ EXPORT_SYMBOL(xdr_inline_decode); * * Moves data beyond the current pointer position from the XDR head[] buffer * into the page list. Any data that lies beyond current position + "len" - * bytes is moved into the XDR tail[]. The current pointer is then - * repositioned at the beginning of the XDR tail. + * bytes is moved into the XDR tail[]. + * + * Returns the number of XDR encoded bytes now contained in the pages */ -void xdr_read_pages(struct xdr_stream *xdr, unsigned int len) +unsigned int xdr_read_pages(struct xdr_stream *xdr, unsigned int len) { struct xdr_buf *buf = xdr->buf; struct kvec *iov; - ssize_t shift; + unsigned int nwords; unsigned int end; - int padding; + unsigned int padding; - /* Realign pages to current pointer position */ - iov = buf->head; - shift = iov->iov_len + (char *)iov->iov_base - (char *)xdr->p; - if (shift > 0) - xdr_shrink_bufhead(buf, shift); - - /* Truncate page data and move it into the tail */ - if (buf->page_len > len) - xdr_shrink_pagelen(buf, buf->page_len - len); - padding = (XDR_QUADLEN(len) << 2) - len; + len = xdr_align_pages(xdr, len); + if (len == 0) + return 0; + nwords = XDR_QUADLEN(len); + padding = (nwords << 2) - len; xdr->iov = iov = buf->tail; /* Compute remaining message length. */ - end = iov->iov_len; - shift = buf->buflen - buf->len; - if (shift < end) - end -= shift; - else if (shift > 0) - end = 0; + end = ((xdr->nwords - nwords) << 2) + padding; + if (end > iov->iov_len) + end = iov->iov_len; + /* * Position current pointer at beginning of tail, and * set remaining message length. */ - xdr->p = (uint32_t *)((char *)iov->iov_base + padding); - xdr->end = (uint32_t *)((char *)iov->iov_base + end); + xdr->p = (__be32 *)((char *)iov->iov_base + padding); + xdr->end = (__be32 *)((char *)iov->iov_base + end); + xdr->page_ptr = NULL; + xdr->nwords = XDR_QUADLEN(end - padding); + return len; } -EXPORT_SYMBOL(xdr_read_pages); +EXPORT_SYMBOL_GPL(xdr_read_pages); + +/** + * xdr_enter_page - decode data from the XDR page + * @xdr: pointer to xdr_stream struct + * @len: number of bytes of page data + * + * Moves data beyond the current pointer position from the XDR head[] buffer + * into the page list. Any data that lies beyond current position + "len" + * bytes is moved into the XDR tail[]. The current pointer is then + * repositioned at the beginning of the first XDR page. + */ +void xdr_enter_page(struct xdr_stream *xdr, unsigned int len) +{ + len = xdr_align_pages(xdr, len); + /* + * Position current pointer at beginning of tail, and + * set remaining message length. + */ + if (len != 0) + xdr_set_page_base(xdr, 0, len); +} +EXPORT_SYMBOL_GPL(xdr_enter_page); static struct kvec empty_iov = {.iov_base = NULL, .iov_len = 0}; @@ -637,42 +1001,43 @@ xdr_buf_from_iov(struct kvec *iov, struct xdr_buf *buf) buf->page_len = 0; buf->buflen = buf->len = iov->iov_len; } +EXPORT_SYMBOL_GPL(xdr_buf_from_iov); -/* Sets subiov to the intersection of iov with the buffer of length len - * starting base bytes after iov. Indicates empty intersection by setting - * length of subiov to zero. Decrements len by length of subiov, sets base - * to zero (or decrements it by length of iov if subiov is empty). */ -static void -iov_subsegment(struct kvec *iov, struct kvec *subiov, int *base, int *len) -{ - if (*base > iov->iov_len) { - subiov->iov_base = NULL; - subiov->iov_len = 0; - *base -= iov->iov_len; - } else { - subiov->iov_base = iov->iov_base + *base; - subiov->iov_len = min(*len, (int)iov->iov_len - *base); - *base = 0; - } - *len -= subiov->iov_len; -} - -/* Sets subbuf to the portion of buf of length len beginning base bytes - * from the start of buf. Returns -1 if base of length are out of bounds. */ +/** + * xdr_buf_subsegment - set subbuf to a portion of buf + * @buf: an xdr buffer + * @subbuf: the result buffer + * @base: beginning of range in bytes + * @len: length of range in bytes + * + * sets @subbuf to an xdr buffer representing the portion of @buf of + * length @len starting at offset @base. + * + * @buf and @subbuf may be pointers to the same struct xdr_buf. + * + * Returns -1 if base of length are out of bounds. + */ int xdr_buf_subsegment(struct xdr_buf *buf, struct xdr_buf *subbuf, - int base, int len) + unsigned int base, unsigned int len) { - int i; - subbuf->buflen = subbuf->len = len; - iov_subsegment(buf->head, subbuf->head, &base, &len); + if (base < buf->head[0].iov_len) { + subbuf->head[0].iov_base = buf->head[0].iov_base + base; + subbuf->head[0].iov_len = min_t(unsigned int, len, + buf->head[0].iov_len - base); + len -= subbuf->head[0].iov_len; + base = 0; + } else { + base -= buf->head[0].iov_len; + subbuf->head[0].iov_len = 0; + } if (base < buf->page_len) { - i = (base + buf->page_base) >> PAGE_CACHE_SHIFT; - subbuf->pages = &buf->pages[i]; - subbuf->page_base = (base + buf->page_base) & ~PAGE_CACHE_MASK; - subbuf->page_len = min((int)buf->page_len - base, len); + subbuf->page_len = min(buf->page_len - base, len); + base += buf->page_base; + subbuf->page_base = base & ~PAGE_CACHE_MASK; + subbuf->pages = &buf->pages[base >> PAGE_CACHE_SHIFT]; len -= subbuf->page_len; base = 0; } else { @@ -680,128 +1045,186 @@ xdr_buf_subsegment(struct xdr_buf *buf, struct xdr_buf *subbuf, subbuf->page_len = 0; } - iov_subsegment(buf->tail, subbuf->tail, &base, &len); + if (base < buf->tail[0].iov_len) { + subbuf->tail[0].iov_base = buf->tail[0].iov_base + base; + subbuf->tail[0].iov_len = min_t(unsigned int, len, + buf->tail[0].iov_len - base); + len -= subbuf->tail[0].iov_len; + base = 0; + } else { + base -= buf->tail[0].iov_len; + subbuf->tail[0].iov_len = 0; + } + if (base || len) return -1; return 0; } +EXPORT_SYMBOL_GPL(xdr_buf_subsegment); -/* obj is assumed to point to allocated memory of size at least len: */ -int -read_bytes_from_xdr_buf(struct xdr_buf *buf, int base, void *obj, int len) +/** + * xdr_buf_trim - lop at most "len" bytes off the end of "buf" + * @buf: buf to be trimmed + * @len: number of bytes to reduce "buf" by + * + * Trim an xdr_buf by the given number of bytes by fixing up the lengths. Note + * that it's possible that we'll trim less than that amount if the xdr_buf is + * too small, or if (for instance) it's all in the head and the parser has + * already read too far into it. + */ +void xdr_buf_trim(struct xdr_buf *buf, unsigned int len) { - struct xdr_buf subbuf; - int this_len; - int status; + size_t cur; + unsigned int trim = len; + + if (buf->tail[0].iov_len) { + cur = min_t(size_t, buf->tail[0].iov_len, trim); + buf->tail[0].iov_len -= cur; + trim -= cur; + if (!trim) + goto fix_len; + } - status = xdr_buf_subsegment(buf, &subbuf, base, len); - if (status) - goto out; - this_len = min(len, (int)subbuf.head[0].iov_len); - memcpy(obj, subbuf.head[0].iov_base, this_len); + if (buf->page_len) { + cur = min_t(unsigned int, buf->page_len, trim); + buf->page_len -= cur; + trim -= cur; + if (!trim) + goto fix_len; + } + + if (buf->head[0].iov_len) { + cur = min_t(size_t, buf->head[0].iov_len, trim); + buf->head[0].iov_len -= cur; + trim -= cur; + } +fix_len: + buf->len -= (len - trim); +} +EXPORT_SYMBOL_GPL(xdr_buf_trim); + +static void __read_bytes_from_xdr_buf(struct xdr_buf *subbuf, void *obj, unsigned int len) +{ + unsigned int this_len; + + this_len = min_t(unsigned int, len, subbuf->head[0].iov_len); + memcpy(obj, subbuf->head[0].iov_base, this_len); len -= this_len; obj += this_len; - this_len = min(len, (int)subbuf.page_len); + this_len = min_t(unsigned int, len, subbuf->page_len); if (this_len) - _copy_from_pages(obj, subbuf.pages, subbuf.page_base, this_len); + _copy_from_pages(obj, subbuf->pages, subbuf->page_base, this_len); len -= this_len; obj += this_len; - this_len = min(len, (int)subbuf.tail[0].iov_len); - memcpy(obj, subbuf.tail[0].iov_base, this_len); -out: - return status; + this_len = min_t(unsigned int, len, subbuf->tail[0].iov_len); + memcpy(obj, subbuf->tail[0].iov_base, this_len); } /* obj is assumed to point to allocated memory of size at least len: */ -int -write_bytes_to_xdr_buf(struct xdr_buf *buf, int base, void *obj, int len) +int read_bytes_from_xdr_buf(struct xdr_buf *buf, unsigned int base, void *obj, unsigned int len) { struct xdr_buf subbuf; - int this_len; int status; status = xdr_buf_subsegment(buf, &subbuf, base, len); - if (status) - goto out; - this_len = min(len, (int)subbuf.head[0].iov_len); - memcpy(subbuf.head[0].iov_base, obj, this_len); + if (status != 0) + return status; + __read_bytes_from_xdr_buf(&subbuf, obj, len); + return 0; +} +EXPORT_SYMBOL_GPL(read_bytes_from_xdr_buf); + +static void __write_bytes_to_xdr_buf(struct xdr_buf *subbuf, void *obj, unsigned int len) +{ + unsigned int this_len; + + this_len = min_t(unsigned int, len, subbuf->head[0].iov_len); + memcpy(subbuf->head[0].iov_base, obj, this_len); len -= this_len; obj += this_len; - this_len = min(len, (int)subbuf.page_len); + this_len = min_t(unsigned int, len, subbuf->page_len); if (this_len) - _copy_to_pages(subbuf.pages, subbuf.page_base, obj, this_len); + _copy_to_pages(subbuf->pages, subbuf->page_base, obj, this_len); len -= this_len; obj += this_len; - this_len = min(len, (int)subbuf.tail[0].iov_len); - memcpy(subbuf.tail[0].iov_base, obj, this_len); -out: - return status; + this_len = min_t(unsigned int, len, subbuf->tail[0].iov_len); + memcpy(subbuf->tail[0].iov_base, obj, this_len); } +/* obj is assumed to point to allocated memory of size at least len: */ +int write_bytes_to_xdr_buf(struct xdr_buf *buf, unsigned int base, void *obj, unsigned int len) +{ + struct xdr_buf subbuf; + int status; + + status = xdr_buf_subsegment(buf, &subbuf, base, len); + if (status != 0) + return status; + __write_bytes_to_xdr_buf(&subbuf, obj, len); + return 0; +} +EXPORT_SYMBOL_GPL(write_bytes_to_xdr_buf); + int -xdr_decode_word(struct xdr_buf *buf, int base, u32 *obj) +xdr_decode_word(struct xdr_buf *buf, unsigned int base, u32 *obj) { - u32 raw; + __be32 raw; int status; status = read_bytes_from_xdr_buf(buf, base, &raw, sizeof(*obj)); if (status) return status; - *obj = ntohl(raw); + *obj = be32_to_cpu(raw); return 0; } +EXPORT_SYMBOL_GPL(xdr_decode_word); int -xdr_encode_word(struct xdr_buf *buf, int base, u32 obj) +xdr_encode_word(struct xdr_buf *buf, unsigned int base, u32 obj) { - u32 raw = htonl(obj); + __be32 raw = cpu_to_be32(obj); return write_bytes_to_xdr_buf(buf, base, &raw, sizeof(obj)); } +EXPORT_SYMBOL_GPL(xdr_encode_word); /* If the netobj starting offset bytes from the start of xdr_buf is contained * entirely in the head or the tail, set object to point to it; otherwise * try to find space for it at the end of the tail, copy it there, and * set obj to point to it. */ -int -xdr_buf_read_netobj(struct xdr_buf *buf, struct xdr_netobj *obj, int offset) +int xdr_buf_read_netobj(struct xdr_buf *buf, struct xdr_netobj *obj, unsigned int offset) { - u32 tail_offset = buf->head[0].iov_len + buf->page_len; - u32 obj_end_offset; + struct xdr_buf subbuf; if (xdr_decode_word(buf, offset, &obj->len)) - goto out; - obj_end_offset = offset + 4 + obj->len; - - if (obj_end_offset <= buf->head[0].iov_len) { - /* The obj is contained entirely in the head: */ - obj->data = buf->head[0].iov_base + offset + 4; - } else if (offset + 4 >= tail_offset) { - if (obj_end_offset - tail_offset - > buf->tail[0].iov_len) - goto out; - /* The obj is contained entirely in the tail: */ - obj->data = buf->tail[0].iov_base - + offset - tail_offset + 4; - } else { - /* use end of tail as storage for obj: - * (We don't copy to the beginning because then we'd have - * to worry about doing a potentially overlapping copy. - * This assumes the object is at most half the length of the - * tail.) */ - if (obj->len > buf->tail[0].iov_len) - goto out; - obj->data = buf->tail[0].iov_base + buf->tail[0].iov_len - - obj->len; - if (read_bytes_from_xdr_buf(buf, offset + 4, - obj->data, obj->len)) - goto out; + return -EFAULT; + if (xdr_buf_subsegment(buf, &subbuf, offset + 4, obj->len)) + return -EFAULT; - } + /* Is the obj contained entirely in the head? */ + obj->data = subbuf.head[0].iov_base; + if (subbuf.head[0].iov_len == obj->len) + return 0; + /* ..or is the obj contained entirely in the tail? */ + obj->data = subbuf.tail[0].iov_base; + if (subbuf.tail[0].iov_len == obj->len) + return 0; + + /* use end of tail as storage for obj: + * (We don't copy to the beginning because then we'd have + * to worry about doing a potentially overlapping copy. + * This assumes the object is at most half the length of the + * tail.) */ + if (obj->len > buf->buflen - buf->len) + return -ENOMEM; + if (buf->tail[0].iov_len != 0) + obj->data = buf->tail[0].iov_base + buf->tail[0].iov_len; + else + obj->data = buf->head[0].iov_base + buf->head[0].iov_len; + __read_bytes_from_xdr_buf(&subbuf, obj->data, obj->len); return 0; -out: - return -1; } +EXPORT_SYMBOL_GPL(xdr_buf_read_netobj); /* Returns 0 on success, or else a negative error code. */ static int @@ -992,8 +1415,7 @@ xdr_xcode_array2(struct xdr_buf *buf, unsigned int base, err = 0; out: - if (elem) - kfree(elem); + kfree(elem); if (ppages) kunmap(*ppages); return err; @@ -1008,6 +1430,7 @@ xdr_decode_array2(struct xdr_buf *buf, unsigned int base, return xdr_xcode_array2(buf, base, desc, 0); } +EXPORT_SYMBOL_GPL(xdr_decode_array2); int xdr_encode_array2(struct xdr_buf *buf, unsigned int base, @@ -1019,3 +1442,72 @@ xdr_encode_array2(struct xdr_buf *buf, unsigned int base, return xdr_xcode_array2(buf, base, desc, 1); } +EXPORT_SYMBOL_GPL(xdr_encode_array2); + +int +xdr_process_buf(struct xdr_buf *buf, unsigned int offset, unsigned int len, + int (*actor)(struct scatterlist *, void *), void *data) +{ + int i, ret = 0; + unsigned int page_len, thislen, page_offset; + struct scatterlist sg[1]; + + sg_init_table(sg, 1); + + if (offset >= buf->head[0].iov_len) { + offset -= buf->head[0].iov_len; + } else { + thislen = buf->head[0].iov_len - offset; + if (thislen > len) + thislen = len; + sg_set_buf(sg, buf->head[0].iov_base + offset, thislen); + ret = actor(sg, data); + if (ret) + goto out; + offset = 0; + len -= thislen; + } + if (len == 0) + goto out; + + if (offset >= buf->page_len) { + offset -= buf->page_len; + } else { + page_len = buf->page_len - offset; + if (page_len > len) + page_len = len; + len -= page_len; + page_offset = (offset + buf->page_base) & (PAGE_CACHE_SIZE - 1); + i = (offset + buf->page_base) >> PAGE_CACHE_SHIFT; + thislen = PAGE_CACHE_SIZE - page_offset; + do { + if (thislen > page_len) + thislen = page_len; + sg_set_page(sg, buf->pages[i], thislen, page_offset); + ret = actor(sg, data); + if (ret) + goto out; + page_len -= thislen; + i++; + page_offset = 0; + thislen = PAGE_CACHE_SIZE; + } while (page_len != 0); + offset = 0; + } + if (len == 0) + goto out; + if (offset < buf->tail[0].iov_len) { + thislen = buf->tail[0].iov_len - offset; + if (thislen > len) + thislen = len; + sg_set_buf(sg, buf->tail[0].iov_base + offset, thislen); + ret = actor(sg, data); + len -= thislen; + } + if (len != 0) + ret = -EINVAL; +out: + return ret; +} +EXPORT_SYMBOL_GPL(xdr_process_buf); + diff --git a/net/sunrpc/xprt.c b/net/sunrpc/xprt.c index 6dda3860351..c3b2b3369e5 100644 --- a/net/sunrpc/xprt.c +++ b/net/sunrpc/xprt.c @@ -12,8 +12,9 @@ * - Next, the caller puts together the RPC message, stuffs it into * the request struct, and calls xprt_transmit(). * - xprt_transmit sends the message and installs the caller on the - * transport's wait list. At the same time, it installs a timer that - * is run after the packet's timeout has expired. + * transport's wait list. At the same time, if a reply is expected, + * it installs a timer that is run after the packet's timeout has + * expired. * - When a packet arrives, the data_ready handler walks the list of * pending requests for that transport. If a matching XID is found, the * caller is woken up, and the timer removed. @@ -41,83 +42,182 @@ #include <linux/types.h> #include <linux/interrupt.h> #include <linux/workqueue.h> -#include <linux/random.h> +#include <linux/net.h> +#include <linux/ktime.h> #include <linux/sunrpc/clnt.h> +#include <linux/sunrpc/metrics.h> +#include <linux/sunrpc/bc_xprt.h> + +#include "sunrpc.h" /* * Local variables */ #ifdef RPC_DEBUG -# undef RPC_DEBUG_DATA # define RPCDBG_FACILITY RPCDBG_XPRT #endif /* * Local functions */ +static void xprt_init(struct rpc_xprt *xprt, struct net *net); static void xprt_request_init(struct rpc_task *, struct rpc_xprt *); -static inline void do_xprt_reserve(struct rpc_task *); static void xprt_connect_status(struct rpc_task *task); static int __xprt_get_cong(struct rpc_xprt *, struct rpc_task *); +static void xprt_destroy(struct rpc_xprt *xprt); -/* - * The transport code maintains an estimate on the maximum number of out- - * standing RPC requests, using a smoothed version of the congestion - * avoidance implemented in 44BSD. This is basically the Van Jacobson - * congestion algorithm: If a retransmit occurs, the congestion window is - * halved; otherwise, it is incremented by 1/cwnd when +static DEFINE_SPINLOCK(xprt_list_lock); +static LIST_HEAD(xprt_list); + +/** + * xprt_register_transport - register a transport implementation + * @transport: transport to register * - * - a reply is received and - * - a full number of requests are outstanding and - * - the congestion window hasn't been updated recently. + * If a transport implementation is loaded as a kernel module, it can + * call this interface to make itself known to the RPC client. + * + * Returns: + * 0: transport successfully registered + * -EEXIST: transport already registered + * -EINVAL: transport module being unloaded */ -#define RPC_CWNDSHIFT (8U) -#define RPC_CWNDSCALE (1U << RPC_CWNDSHIFT) -#define RPC_INITCWND RPC_CWNDSCALE -#define RPC_MAXCWND(xprt) ((xprt)->max_reqs << RPC_CWNDSHIFT) +int xprt_register_transport(struct xprt_class *transport) +{ + struct xprt_class *t; + int result; + + result = -EEXIST; + spin_lock(&xprt_list_lock); + list_for_each_entry(t, &xprt_list, list) { + /* don't register the same transport class twice */ + if (t->ident == transport->ident) + goto out; + } + + list_add_tail(&transport->list, &xprt_list); + printk(KERN_INFO "RPC: Registered %s transport module.\n", + transport->name); + result = 0; -#define RPCXPRT_CONGESTED(xprt) ((xprt)->cong >= (xprt)->cwnd) +out: + spin_unlock(&xprt_list_lock); + return result; +} +EXPORT_SYMBOL_GPL(xprt_register_transport); + +/** + * xprt_unregister_transport - unregister a transport implementation + * @transport: transport to unregister + * + * Returns: + * 0: transport successfully unregistered + * -ENOENT: transport never registered + */ +int xprt_unregister_transport(struct xprt_class *transport) +{ + struct xprt_class *t; + int result; + + result = 0; + spin_lock(&xprt_list_lock); + list_for_each_entry(t, &xprt_list, list) { + if (t == transport) { + printk(KERN_INFO + "RPC: Unregistered %s transport module.\n", + transport->name); + list_del_init(&transport->list); + goto out; + } + } + result = -ENOENT; + +out: + spin_unlock(&xprt_list_lock); + return result; +} +EXPORT_SYMBOL_GPL(xprt_unregister_transport); + +/** + * xprt_load_transport - load a transport implementation + * @transport_name: transport to load + * + * Returns: + * 0: transport successfully loaded + * -ENOENT: transport module not available + */ +int xprt_load_transport(const char *transport_name) +{ + struct xprt_class *t; + int result; + + result = 0; + spin_lock(&xprt_list_lock); + list_for_each_entry(t, &xprt_list, list) { + if (strcmp(t->name, transport_name) == 0) { + spin_unlock(&xprt_list_lock); + goto out; + } + } + spin_unlock(&xprt_list_lock); + result = request_module("xprt%s", transport_name); +out: + return result; +} +EXPORT_SYMBOL_GPL(xprt_load_transport); /** * xprt_reserve_xprt - serialize write access to transports * @task: task that is requesting access to the transport + * @xprt: pointer to the target transport * * This prevents mixing the payload of separate requests, and prevents * transport connects from colliding with writes. No congestion control * is provided. */ -int xprt_reserve_xprt(struct rpc_task *task) +int xprt_reserve_xprt(struct rpc_xprt *xprt, struct rpc_task *task) { - struct rpc_xprt *xprt = task->tk_xprt; struct rpc_rqst *req = task->tk_rqstp; + int priority; if (test_and_set_bit(XPRT_LOCKED, &xprt->state)) { if (task == xprt->snd_task) return 1; - if (task == NULL) - return 0; goto out_sleep; } xprt->snd_task = task; - if (req) { - req->rq_bytes_sent = 0; + if (req != NULL) req->rq_ntrans++; - } + return 1; out_sleep: - dprintk("RPC: %4d failed to lock transport %p\n", + dprintk("RPC: %5u failed to lock transport %p\n", task->tk_pid, xprt); task->tk_timeout = 0; task->tk_status = -EAGAIN; - if (req && req->rq_ntrans) - rpc_sleep_on(&xprt->resend, task, NULL, NULL); + if (req == NULL) + priority = RPC_PRIORITY_LOW; + else if (!req->rq_ntrans) + priority = RPC_PRIORITY_NORMAL; else - rpc_sleep_on(&xprt->sending, task, NULL, NULL); + priority = RPC_PRIORITY_HIGH; + rpc_sleep_on_priority(&xprt->sending, task, NULL, priority); return 0; } +EXPORT_SYMBOL_GPL(xprt_reserve_xprt); + +static void xprt_clear_locked(struct rpc_xprt *xprt) +{ + xprt->snd_task = NULL; + if (!test_bit(XPRT_CLOSE_WAIT, &xprt->state)) { + smp_mb__before_atomic(); + clear_bit(XPRT_LOCKED, &xprt->state); + smp_mb__after_atomic(); + } else + queue_work(rpciod_workqueue, &xprt->task_cleanup); +} /* * xprt_reserve_xprt_cong - serialize write access to transports @@ -127,104 +227,101 @@ out_sleep: * integrated into the decision of whether a request is allowed to be * woken up and given access to the transport. */ -int xprt_reserve_xprt_cong(struct rpc_task *task) +int xprt_reserve_xprt_cong(struct rpc_xprt *xprt, struct rpc_task *task) { - struct rpc_xprt *xprt = task->tk_xprt; struct rpc_rqst *req = task->tk_rqstp; + int priority; if (test_and_set_bit(XPRT_LOCKED, &xprt->state)) { if (task == xprt->snd_task) return 1; goto out_sleep; } + if (req == NULL) { + xprt->snd_task = task; + return 1; + } if (__xprt_get_cong(xprt, task)) { xprt->snd_task = task; - if (req) { - req->rq_bytes_sent = 0; - req->rq_ntrans++; - } + req->rq_ntrans++; return 1; } - smp_mb__before_clear_bit(); - clear_bit(XPRT_LOCKED, &xprt->state); - smp_mb__after_clear_bit(); + xprt_clear_locked(xprt); out_sleep: - dprintk("RPC: %4d failed to lock transport %p\n", task->tk_pid, xprt); + dprintk("RPC: %5u failed to lock transport %p\n", task->tk_pid, xprt); task->tk_timeout = 0; task->tk_status = -EAGAIN; - if (req && req->rq_ntrans) - rpc_sleep_on(&xprt->resend, task, NULL, NULL); + if (req == NULL) + priority = RPC_PRIORITY_LOW; + else if (!req->rq_ntrans) + priority = RPC_PRIORITY_NORMAL; else - rpc_sleep_on(&xprt->sending, task, NULL, NULL); + priority = RPC_PRIORITY_HIGH; + rpc_sleep_on_priority(&xprt->sending, task, NULL, priority); return 0; } +EXPORT_SYMBOL_GPL(xprt_reserve_xprt_cong); static inline int xprt_lock_write(struct rpc_xprt *xprt, struct rpc_task *task) { int retval; spin_lock_bh(&xprt->transport_lock); - retval = xprt->ops->reserve_xprt(task); + retval = xprt->ops->reserve_xprt(xprt, task); spin_unlock_bh(&xprt->transport_lock); return retval; } -static void __xprt_lock_write_next(struct rpc_xprt *xprt) +static bool __xprt_lock_write_func(struct rpc_task *task, void *data) { - struct rpc_task *task; + struct rpc_xprt *xprt = data; struct rpc_rqst *req; + req = task->tk_rqstp; + xprt->snd_task = task; + if (req) + req->rq_ntrans++; + return true; +} + +static void __xprt_lock_write_next(struct rpc_xprt *xprt) +{ if (test_and_set_bit(XPRT_LOCKED, &xprt->state)) return; - task = rpc_wake_up_next(&xprt->resend); - if (!task) { - task = rpc_wake_up_next(&xprt->sending); - if (!task) - goto out_unlock; - } + if (rpc_wake_up_first(&xprt->sending, __xprt_lock_write_func, xprt)) + return; + xprt_clear_locked(xprt); +} + +static bool __xprt_lock_write_cong_func(struct rpc_task *task, void *data) +{ + struct rpc_xprt *xprt = data; + struct rpc_rqst *req; req = task->tk_rqstp; - xprt->snd_task = task; - if (req) { - req->rq_bytes_sent = 0; + if (req == NULL) { + xprt->snd_task = task; + return true; + } + if (__xprt_get_cong(xprt, task)) { + xprt->snd_task = task; req->rq_ntrans++; + return true; } - return; - -out_unlock: - smp_mb__before_clear_bit(); - clear_bit(XPRT_LOCKED, &xprt->state); - smp_mb__after_clear_bit(); + return false; } static void __xprt_lock_write_next_cong(struct rpc_xprt *xprt) { - struct rpc_task *task; - if (test_and_set_bit(XPRT_LOCKED, &xprt->state)) return; if (RPCXPRT_CONGESTED(xprt)) goto out_unlock; - task = rpc_wake_up_next(&xprt->resend); - if (!task) { - task = rpc_wake_up_next(&xprt->sending); - if (!task) - goto out_unlock; - } - if (__xprt_get_cong(xprt, task)) { - struct rpc_rqst *req = task->tk_rqstp; - xprt->snd_task = task; - if (req) { - req->rq_bytes_sent = 0; - req->rq_ntrans++; - } + if (rpc_wake_up_first(&xprt->sending, __xprt_lock_write_cong_func, xprt)) return; - } out_unlock: - smp_mb__before_clear_bit(); - clear_bit(XPRT_LOCKED, &xprt->state); - smp_mb__after_clear_bit(); + xprt_clear_locked(xprt); } /** @@ -237,13 +334,16 @@ out_unlock: void xprt_release_xprt(struct rpc_xprt *xprt, struct rpc_task *task) { if (xprt->snd_task == task) { - xprt->snd_task = NULL; - smp_mb__before_clear_bit(); - clear_bit(XPRT_LOCKED, &xprt->state); - smp_mb__after_clear_bit(); + if (task != NULL) { + struct rpc_rqst *req = task->tk_rqstp; + if (req != NULL) + req->rq_bytes_sent = 0; + } + xprt_clear_locked(xprt); __xprt_lock_write_next(xprt); } } +EXPORT_SYMBOL_GPL(xprt_release_xprt); /** * xprt_release_xprt_cong - allow other requests to use a transport @@ -256,13 +356,16 @@ void xprt_release_xprt(struct rpc_xprt *xprt, struct rpc_task *task) void xprt_release_xprt_cong(struct rpc_xprt *xprt, struct rpc_task *task) { if (xprt->snd_task == task) { - xprt->snd_task = NULL; - smp_mb__before_clear_bit(); - clear_bit(XPRT_LOCKED, &xprt->state); - smp_mb__after_clear_bit(); + if (task != NULL) { + struct rpc_rqst *req = task->tk_rqstp; + if (req != NULL) + req->rq_bytes_sent = 0; + } + xprt_clear_locked(xprt); __xprt_lock_write_next_cong(xprt); } } +EXPORT_SYMBOL_GPL(xprt_release_xprt_cong); static inline void xprt_release_write(struct rpc_xprt *xprt, struct rpc_task *task) { @@ -282,7 +385,7 @@ __xprt_get_cong(struct rpc_xprt *xprt, struct rpc_task *task) if (req->rq_cong) return 1; - dprintk("RPC: %4d xprt_cwnd_limited cong = %ld cwnd = %ld\n", + dprintk("RPC: %5u xprt_cwnd_limited cong = %lu cwnd = %lu\n", task->tk_pid, xprt->cong, xprt->cwnd); if (RPCXPRT_CONGESTED(xprt)) return 0; @@ -313,20 +416,31 @@ __xprt_put_cong(struct rpc_xprt *xprt, struct rpc_rqst *req) */ void xprt_release_rqst_cong(struct rpc_task *task) { - __xprt_put_cong(task->tk_xprt, task->tk_rqstp); + struct rpc_rqst *req = task->tk_rqstp; + + __xprt_put_cong(req->rq_xprt, req); } +EXPORT_SYMBOL_GPL(xprt_release_rqst_cong); /** * xprt_adjust_cwnd - adjust transport congestion window + * @xprt: pointer to xprt * @task: recently completed RPC request used to adjust window * @result: result code of completed RPC request * - * We use a time-smoothed congestion estimator to avoid heavy oscillation. + * The transport code maintains an estimate on the maximum number of out- + * standing RPC requests, using a smoothed version of the congestion + * avoidance implemented in 44BSD. This is basically the Van Jacobson + * congestion algorithm: If a retransmit occurs, the congestion window is + * halved; otherwise, it is incremented by 1/cwnd when + * + * - a reply is received and + * - a full number of requests are outstanding and + * - the congestion window hasn't been updated recently. */ -void xprt_adjust_cwnd(struct rpc_task *task, int result) +void xprt_adjust_cwnd(struct rpc_xprt *xprt, struct rpc_task *task, int result) { struct rpc_rqst *req = task->tk_rqstp; - struct rpc_xprt *xprt = task->tk_xprt; unsigned long cwnd = xprt->cwnd; if (result >= 0 && cwnd <= xprt->cong) { @@ -341,11 +455,12 @@ void xprt_adjust_cwnd(struct rpc_task *task, int result) if (cwnd < RPC_CWNDSCALE) cwnd = RPC_CWNDSCALE; } - dprintk("RPC: cong %ld, cwnd was %ld, now %ld\n", + dprintk("RPC: cong %ld, cwnd was %ld, now %ld\n", xprt->cong, xprt->cwnd, cwnd); xprt->cwnd = cwnd; __xprt_put_cong(xprt, req); } +EXPORT_SYMBOL_GPL(xprt_adjust_cwnd); /** * xprt_wake_pending_tasks - wake all tasks on a transport's pending queue @@ -360,20 +475,26 @@ void xprt_wake_pending_tasks(struct rpc_xprt *xprt, int status) else rpc_wake_up(&xprt->pending); } +EXPORT_SYMBOL_GPL(xprt_wake_pending_tasks); /** * xprt_wait_for_buffer_space - wait for transport output buffer to clear * @task: task to be put to sleep + * @action: function pointer to be executed after wait * + * Note that we only set the timer for the case of RPC_IS_SOFT(), since + * we don't in general want to force a socket disconnection due to + * an incomplete RPC call transmission. */ -void xprt_wait_for_buffer_space(struct rpc_task *task) +void xprt_wait_for_buffer_space(struct rpc_task *task, rpc_action action) { struct rpc_rqst *req = task->tk_rqstp; struct rpc_xprt *xprt = req->rq_xprt; - task->tk_timeout = req->rq_timeout; - rpc_sleep_on(&xprt->pending, task, NULL, NULL); + task->tk_timeout = RPC_IS_SOFT(task) ? req->rq_timeout : 0; + rpc_sleep_on(&xprt->pending, task, action); } +EXPORT_SYMBOL_GPL(xprt_wait_for_buffer_space); /** * xprt_write_space - wake the task waiting for transport output buffer space @@ -383,17 +504,15 @@ void xprt_wait_for_buffer_space(struct rpc_task *task) */ void xprt_write_space(struct rpc_xprt *xprt) { - if (unlikely(xprt->shutdown)) - return; - spin_lock_bh(&xprt->transport_lock); if (xprt->snd_task) { - dprintk("RPC: write space: waking waiting task on xprt %p\n", - xprt); - rpc_wake_up_task(xprt->snd_task); + dprintk("RPC: write space: waking waiting task on " + "xprt %p\n", xprt); + rpc_wake_up_queued_task(&xprt->pending, xprt->snd_task); } spin_unlock_bh(&xprt->transport_lock); } +EXPORT_SYMBOL_GPL(xprt_write_space); /** * xprt_set_retrans_timeout_def - set a request's retransmit timeout @@ -407,29 +526,32 @@ void xprt_set_retrans_timeout_def(struct rpc_task *task) { task->tk_timeout = task->tk_rqstp->rq_timeout; } +EXPORT_SYMBOL_GPL(xprt_set_retrans_timeout_def); -/* +/** * xprt_set_retrans_timeout_rtt - set a request's retransmit timeout * @task: task whose timeout is to be set - * + * * Set a request's retransmit timeout using the RTT estimator. */ void xprt_set_retrans_timeout_rtt(struct rpc_task *task) { int timer = task->tk_msg.rpc_proc->p_timer; - struct rpc_rtt *rtt = task->tk_client->cl_rtt; + struct rpc_clnt *clnt = task->tk_client; + struct rpc_rtt *rtt = clnt->cl_rtt; struct rpc_rqst *req = task->tk_rqstp; - unsigned long max_timeout = req->rq_xprt->timeout.to_maxval; + unsigned long max_timeout = clnt->cl_timeout->to_maxval; task->tk_timeout = rpc_calc_rto(rtt, timer); task->tk_timeout <<= rpc_ntimeo(rtt, timer) + req->rq_retries; if (task->tk_timeout > max_timeout || task->tk_timeout == 0) task->tk_timeout = max_timeout; } +EXPORT_SYMBOL_GPL(xprt_set_retrans_timeout_rtt); static void xprt_reset_majortimeo(struct rpc_rqst *req) { - struct rpc_timeout *to = &req->rq_xprt->timeout; + const struct rpc_timeout *to = req->rq_task->tk_client->cl_timeout; req->rq_majortimeo = req->rq_timeout; if (to->to_exponential) @@ -449,7 +571,7 @@ static void xprt_reset_majortimeo(struct rpc_rqst *req) int xprt_adjust_timeout(struct rpc_rqst *req) { struct rpc_xprt *xprt = req->rq_xprt; - struct rpc_timeout *to = &xprt->timeout; + const struct rpc_timeout *to = req->rq_task->tk_client->cl_timeout; int status = 0; if (time_before(jiffies, req->rq_majortimeo)) { @@ -460,7 +582,6 @@ int xprt_adjust_timeout(struct rpc_rqst *req) if (to->to_maxval && req->rq_timeout >= to->to_maxval) req->rq_timeout = to->to_maxval; req->rq_retries++; - pprintk("RPC: %lu retrans\n", jiffies); } else { req->rq_timeout = to->to_initval; req->rq_retries = 0; @@ -469,7 +590,6 @@ int xprt_adjust_timeout(struct rpc_rqst *req) spin_lock_bh(&xprt->transport_lock); rpc_init_rtt(req->rq_task->tk_client->cl_rtt, to->to_initval); spin_unlock_bh(&xprt->transport_lock); - pprintk("RPC: %lu timeout\n", jiffies); status = -ETIMEDOUT; } @@ -480,26 +600,73 @@ int xprt_adjust_timeout(struct rpc_rqst *req) return status; } -static void xprt_autoclose(void *args) +static void xprt_autoclose(struct work_struct *work) { - struct rpc_xprt *xprt = (struct rpc_xprt *)args; + struct rpc_xprt *xprt = + container_of(work, struct rpc_xprt, task_cleanup); - xprt_disconnect(xprt); xprt->ops->close(xprt); + clear_bit(XPRT_CLOSE_WAIT, &xprt->state); xprt_release_write(xprt, NULL); } /** - * xprt_disconnect - mark a transport as disconnected + * xprt_disconnect_done - mark a transport as disconnected * @xprt: transport to flag for disconnect * */ -void xprt_disconnect(struct rpc_xprt *xprt) +void xprt_disconnect_done(struct rpc_xprt *xprt) { - dprintk("RPC: disconnected transport %p\n", xprt); + dprintk("RPC: disconnected transport %p\n", xprt); spin_lock_bh(&xprt->transport_lock); xprt_clear_connected(xprt); - xprt_wake_pending_tasks(xprt, -ENOTCONN); + xprt_wake_pending_tasks(xprt, -EAGAIN); + spin_unlock_bh(&xprt->transport_lock); +} +EXPORT_SYMBOL_GPL(xprt_disconnect_done); + +/** + * xprt_force_disconnect - force a transport to disconnect + * @xprt: transport to disconnect + * + */ +void xprt_force_disconnect(struct rpc_xprt *xprt) +{ + /* Don't race with the test_bit() in xprt_clear_locked() */ + spin_lock_bh(&xprt->transport_lock); + set_bit(XPRT_CLOSE_WAIT, &xprt->state); + /* Try to schedule an autoclose RPC call */ + if (test_and_set_bit(XPRT_LOCKED, &xprt->state) == 0) + queue_work(rpciod_workqueue, &xprt->task_cleanup); + xprt_wake_pending_tasks(xprt, -EAGAIN); + spin_unlock_bh(&xprt->transport_lock); +} + +/** + * xprt_conditional_disconnect - force a transport to disconnect + * @xprt: transport to disconnect + * @cookie: 'connection cookie' + * + * This attempts to break the connection if and only if 'cookie' matches + * the current transport 'connection cookie'. It ensures that we don't + * try to break the connection more than once when we need to retransmit + * a batch of RPC requests. + * + */ +void xprt_conditional_disconnect(struct rpc_xprt *xprt, unsigned int cookie) +{ + /* Don't race with the test_bit() in xprt_clear_locked() */ + spin_lock_bh(&xprt->transport_lock); + if (cookie != xprt->connect_cookie) + goto out; + if (test_bit(XPRT_CLOSING, &xprt->state) || !xprt_connected(xprt)) + goto out; + set_bit(XPRT_CLOSE_WAIT, &xprt->state); + /* Try to schedule an autoclose RPC call */ + if (test_and_set_bit(XPRT_LOCKED, &xprt->state) == 0) + queue_work(rpciod_workqueue, &xprt->task_cleanup); + xprt_wake_pending_tasks(xprt, -EAGAIN); +out: spin_unlock_bh(&xprt->transport_lock); } @@ -509,15 +676,13 @@ xprt_init_autodisconnect(unsigned long data) struct rpc_xprt *xprt = (struct rpc_xprt *)data; spin_lock(&xprt->transport_lock); - if (!list_empty(&xprt->recv) || xprt->shutdown) + if (!list_empty(&xprt->recv)) goto out_abort; if (test_and_set_bit(XPRT_LOCKED, &xprt->state)) goto out_abort; spin_unlock(&xprt->transport_lock); - if (xprt_connecting(xprt)) - xprt_release_write(xprt, NULL); - else - schedule_work(&xprt->task_cleanup); + set_bit(XPRT_CONNECTION_CLOSE, &xprt->state); + queue_work(rpciod_workqueue, &xprt->task_cleanup); return; out_abort: spin_unlock(&xprt->transport_lock); @@ -530,40 +695,45 @@ out_abort: */ void xprt_connect(struct rpc_task *task) { - struct rpc_xprt *xprt = task->tk_xprt; + struct rpc_xprt *xprt = task->tk_rqstp->rq_xprt; - dprintk("RPC: %4d xprt_connect xprt %p %s connected\n", task->tk_pid, + dprintk("RPC: %5u xprt_connect xprt %p %s connected\n", task->tk_pid, xprt, (xprt_connected(xprt) ? "is" : "is not")); - if (xprt->shutdown) { - task->tk_status = -EIO; - return; - } - if (!xprt->addr.sin_port) { - task->tk_status = -EIO; + if (!xprt_bound(xprt)) { + task->tk_status = -EAGAIN; return; } if (!xprt_lock_write(xprt, task)) return; + + if (test_and_clear_bit(XPRT_CLOSE_WAIT, &xprt->state)) + xprt->ops->close(xprt); + if (xprt_connected(xprt)) xprt_release_write(xprt, task); else { - if (task->tk_rqstp) - task->tk_rqstp->rq_bytes_sent = 0; - - task->tk_timeout = xprt->connect_timeout; - rpc_sleep_on(&xprt->pending, task, xprt_connect_status, NULL); - xprt->ops->connect(task); + task->tk_rqstp->rq_bytes_sent = 0; + task->tk_timeout = task->tk_rqstp->rq_timeout; + rpc_sleep_on(&xprt->pending, task, xprt_connect_status); + + if (test_bit(XPRT_CLOSING, &xprt->state)) + return; + if (xprt_test_and_set_connecting(xprt)) + return; + xprt->stat.connect_start = jiffies; + xprt->ops->connect(xprt, task); } - return; } static void xprt_connect_status(struct rpc_task *task) { - struct rpc_xprt *xprt = task->tk_xprt; + struct rpc_xprt *xprt = task->tk_rqstp->rq_xprt; - if (task->tk_status >= 0) { - dprintk("RPC: %4d xprt_connect_status: connection established\n", + if (task->tk_status == 0) { + xprt->stat.connect_count++; + xprt->stat.connect_time += (long)jiffies - xprt->stat.connect_start; + dprintk("RPC: %5u xprt_connect_status: connection established\n", task->tk_pid); return; } @@ -571,27 +741,20 @@ static void xprt_connect_status(struct rpc_task *task) switch (task->tk_status) { case -ECONNREFUSED: case -ECONNRESET: - dprintk("RPC: %4d xprt_connect_status: server %s refused connection\n", - task->tk_pid, task->tk_client->cl_server); - break; - case -ENOTCONN: - dprintk("RPC: %4d xprt_connect_status: connection broken\n", - task->tk_pid); + case -ECONNABORTED: + case -ENETUNREACH: + case -EHOSTUNREACH: + case -EAGAIN: + dprintk("RPC: %5u xprt_connect_status: retrying\n", task->tk_pid); break; case -ETIMEDOUT: - dprintk("RPC: %4d xprt_connect_status: connect attempt timed out\n", - task->tk_pid); + dprintk("RPC: %5u xprt_connect_status: connect attempt timed " + "out\n", task->tk_pid); break; default: - dprintk("RPC: %4d xprt_connect_status: error %d connecting to server %s\n", - task->tk_pid, -task->tk_status, task->tk_client->cl_server); - xprt_release_write(xprt, task); - task->tk_status = -EIO; - return; - } - - /* if soft mounted, just cause this RPC to fail */ - if (RPC_IS_SOFT(task)) { + dprintk("RPC: %5u xprt_connect_status: error %d connecting to " + "server %s\n", task->tk_pid, -task->tk_status, + xprt->servername); xprt_release_write(xprt, task); task->tk_status = -EIO; } @@ -603,36 +766,31 @@ static void xprt_connect_status(struct rpc_task *task) * @xid: RPC XID of incoming reply * */ -struct rpc_rqst *xprt_lookup_rqst(struct rpc_xprt *xprt, u32 xid) +struct rpc_rqst *xprt_lookup_rqst(struct rpc_xprt *xprt, __be32 xid) { - struct list_head *pos; - struct rpc_rqst *req = NULL; + struct rpc_rqst *entry; - list_for_each(pos, &xprt->recv) { - struct rpc_rqst *entry = list_entry(pos, struct rpc_rqst, rq_list); - if (entry->rq_xid == xid) { - req = entry; - break; - } - } - return req; + list_for_each_entry(entry, &xprt->recv, rq_list) + if (entry->rq_xid == xid) + return entry; + + dprintk("RPC: xprt_lookup_rqst did not find xid %08x\n", + ntohl(xid)); + xprt->stat.bad_xids++; + return NULL; } +EXPORT_SYMBOL_GPL(xprt_lookup_rqst); -/** - * xprt_update_rtt - update an RPC client's RTT state after receiving a reply - * @task: RPC request that recently completed - * - */ -void xprt_update_rtt(struct rpc_task *task) +static void xprt_update_rtt(struct rpc_task *task) { struct rpc_rqst *req = task->tk_rqstp; struct rpc_rtt *rtt = task->tk_client->cl_rtt; - unsigned timer = task->tk_msg.rpc_proc->p_timer; + unsigned int timer = task->tk_msg.rpc_proc->p_timer; + long m = usecs_to_jiffies(ktime_to_us(req->rq_rtt)); if (timer) { if (req->rq_ntrans == 1) - rpc_update_rtt(rtt, timer, - (long)jiffies - req->rq_xtime); + rpc_update_rtt(rtt, timer, m); rpc_set_timeo(rtt, timer, req->rq_ntrans - 1); } } @@ -647,31 +805,47 @@ void xprt_update_rtt(struct rpc_task *task) void xprt_complete_rqst(struct rpc_task *task, int copied) { struct rpc_rqst *req = task->tk_rqstp; + struct rpc_xprt *xprt = req->rq_xprt; dprintk("RPC: %5u xid %08x complete (%d bytes received)\n", task->tk_pid, ntohl(req->rq_xid), copied); + xprt->stat.recvs++; + req->rq_rtt = ktime_sub(ktime_get(), req->rq_xtime); + if (xprt->ops->timer != NULL) + xprt_update_rtt(task); + list_del_init(&req->rq_list); - req->rq_received = req->rq_private_buf.len = copied; - rpc_wake_up_task(task); + req->rq_private_buf.len = copied; + /* Ensure all writes are done before we update */ + /* req->rq_reply_bytes_recvd */ + smp_wmb(); + req->rq_reply_bytes_recvd = copied; + rpc_wake_up_queued_task(&xprt->pending, task); } +EXPORT_SYMBOL_GPL(xprt_complete_rqst); static void xprt_timer(struct rpc_task *task) { struct rpc_rqst *req = task->tk_rqstp; struct rpc_xprt *xprt = req->rq_xprt; - dprintk("RPC: %4d xprt_timer\n", task->tk_pid); + if (task->tk_status != -ETIMEDOUT) + return; + dprintk("RPC: %5u xprt_timer\n", task->tk_pid); - spin_lock(&xprt->transport_lock); - if (!req->rq_received) { + spin_lock_bh(&xprt->transport_lock); + if (!req->rq_reply_bytes_recvd) { if (xprt->ops->timer) - xprt->ops->timer(task); - task->tk_status = -ETIMEDOUT; - } - task->tk_timeout = 0; - rpc_wake_up_task(task); - spin_unlock(&xprt->transport_lock); + xprt->ops->timer(xprt, task); + } else + task->tk_status = 0; + spin_unlock_bh(&xprt->transport_lock); +} + +static inline int xprt_has_timer(struct rpc_xprt *xprt) +{ + return xprt->idle_timeout != 0; } /** @@ -679,42 +853,41 @@ static void xprt_timer(struct rpc_task *task) * @task: RPC task about to send a request * */ -int xprt_prepare_transmit(struct rpc_task *task) +bool xprt_prepare_transmit(struct rpc_task *task) { struct rpc_rqst *req = task->tk_rqstp; struct rpc_xprt *xprt = req->rq_xprt; - int err = 0; + bool ret = false; - dprintk("RPC: %4d xprt_prepare_transmit\n", task->tk_pid); - - if (xprt->shutdown) - return -EIO; + dprintk("RPC: %5u xprt_prepare_transmit\n", task->tk_pid); spin_lock_bh(&xprt->transport_lock); - if (req->rq_received && !req->rq_bytes_sent) { - err = req->rq_received; - goto out_unlock; - } - if (!xprt->ops->reserve_xprt(task)) { - err = -EAGAIN; - goto out_unlock; + if (!req->rq_bytes_sent) { + if (req->rq_reply_bytes_recvd) { + task->tk_status = req->rq_reply_bytes_recvd; + goto out_unlock; + } + if ((task->tk_flags & RPC_TASK_NO_RETRANS_TIMEOUT) + && xprt_connected(xprt) + && req->rq_connect_cookie == xprt->connect_cookie) { + xprt->ops->set_retrans_timeout(task); + rpc_sleep_on(&xprt->pending, task, xprt_timer); + goto out_unlock; + } } - - if (!xprt_connected(xprt)) { - err = -ENOTCONN; + if (!xprt->ops->reserve_xprt(xprt, task)) { + task->tk_status = -EAGAIN; goto out_unlock; } + ret = true; out_unlock: spin_unlock_bh(&xprt->transport_lock); - return err; + return ret; } -void -xprt_abort_transmit(struct rpc_task *task) +void xprt_end_transmit(struct rpc_task *task) { - struct rpc_xprt *xprt = task->tk_xprt; - - xprt_release_write(xprt, task); + xprt_release_write(task->tk_rqstp->rq_xprt, task); } /** @@ -727,13 +900,15 @@ void xprt_transmit(struct rpc_task *task) { struct rpc_rqst *req = task->tk_rqstp; struct rpc_xprt *xprt = req->rq_xprt; - int status; + int status, numreqs; - dprintk("RPC: %4d xprt_transmit(%u)\n", task->tk_pid, req->rq_slen); + dprintk("RPC: %5u xprt_transmit(%u)\n", task->tk_pid, req->rq_slen); - smp_rmb(); - if (!req->rq_received) { - if (list_empty(&req->rq_list)) { + if (!req->rq_reply_bytes_recvd) { + if (list_empty(&req->rq_list) && rpc_reply_expected(task)) { + /* + * Add to the list only if we're expecting a reply + */ spin_lock_bh(&xprt->transport_lock); /* Update the softirq receive buffer */ memcpy(&req->rq_private_buf, &req->rq_rcv_buf, @@ -748,99 +923,288 @@ void xprt_transmit(struct rpc_task *task) } else if (!req->rq_bytes_sent) return; + req->rq_xtime = ktime_get(); status = xprt->ops->send_request(task); - if (status == 0) { - dprintk("RPC: %4d xmit complete\n", task->tk_pid); - spin_lock_bh(&xprt->transport_lock); - xprt->ops->set_retrans_timeout(task); - /* Don't race with disconnect */ - if (!xprt_connected(xprt)) - task->tk_status = -ENOTCONN; - else if (!req->rq_received) - rpc_sleep_on(&xprt->pending, task, NULL, xprt_timer); - xprt->ops->release_xprt(xprt, task); - spin_unlock_bh(&xprt->transport_lock); + if (status != 0) { + task->tk_status = status; return; } - /* Note: at this point, task->tk_sleeping has not yet been set, - * hence there is no danger of the waking up task being put on - * schedq, and being picked up by a parallel run of rpciod(). - */ - task->tk_status = status; + dprintk("RPC: %5u xmit complete\n", task->tk_pid); + task->tk_flags |= RPC_TASK_SENT; + spin_lock_bh(&xprt->transport_lock); - switch (status) { - case -ECONNREFUSED: - rpc_sleep_on(&xprt->sending, task, NULL, NULL); + xprt->ops->set_retrans_timeout(task); + + numreqs = atomic_read(&xprt->num_reqs); + if (numreqs > xprt->stat.max_slots) + xprt->stat.max_slots = numreqs; + xprt->stat.sends++; + xprt->stat.req_u += xprt->stat.sends - xprt->stat.recvs; + xprt->stat.bklog_u += xprt->backlog.qlen; + xprt->stat.sending_u += xprt->sending.qlen; + xprt->stat.pending_u += xprt->pending.qlen; + + /* Don't race with disconnect */ + if (!xprt_connected(xprt)) + task->tk_status = -ENOTCONN; + else { + /* + * Sleep on the pending queue since + * we're expecting a reply. + */ + if (!req->rq_reply_bytes_recvd && rpc_reply_expected(task)) + rpc_sleep_on(&xprt->pending, task, xprt_timer); + req->rq_connect_cookie = xprt->connect_cookie; + } + spin_unlock_bh(&xprt->transport_lock); +} + +static void xprt_add_backlog(struct rpc_xprt *xprt, struct rpc_task *task) +{ + set_bit(XPRT_CONGESTED, &xprt->state); + rpc_sleep_on(&xprt->backlog, task, NULL); +} + +static void xprt_wake_up_backlog(struct rpc_xprt *xprt) +{ + if (rpc_wake_up_next(&xprt->backlog) == NULL) + clear_bit(XPRT_CONGESTED, &xprt->state); +} + +static bool xprt_throttle_congested(struct rpc_xprt *xprt, struct rpc_task *task) +{ + bool ret = false; + + if (!test_bit(XPRT_CONGESTED, &xprt->state)) + goto out; + spin_lock(&xprt->reserve_lock); + if (test_bit(XPRT_CONGESTED, &xprt->state)) { + rpc_sleep_on(&xprt->backlog, task, NULL); + ret = true; + } + spin_unlock(&xprt->reserve_lock); +out: + return ret; +} + +static struct rpc_rqst *xprt_dynamic_alloc_slot(struct rpc_xprt *xprt, gfp_t gfp_flags) +{ + struct rpc_rqst *req = ERR_PTR(-EAGAIN); + + if (!atomic_add_unless(&xprt->num_reqs, 1, xprt->max_reqs)) + goto out; + req = kzalloc(sizeof(struct rpc_rqst), gfp_flags); + if (req != NULL) + goto out; + atomic_dec(&xprt->num_reqs); + req = ERR_PTR(-ENOMEM); +out: + return req; +} + +static bool xprt_dynamic_free_slot(struct rpc_xprt *xprt, struct rpc_rqst *req) +{ + if (atomic_add_unless(&xprt->num_reqs, -1, xprt->min_reqs)) { + kfree(req); + return true; + } + return false; +} + +void xprt_alloc_slot(struct rpc_xprt *xprt, struct rpc_task *task) +{ + struct rpc_rqst *req; + + spin_lock(&xprt->reserve_lock); + if (!list_empty(&xprt->free)) { + req = list_entry(xprt->free.next, struct rpc_rqst, rq_list); + list_del(&req->rq_list); + goto out_init_req; + } + req = xprt_dynamic_alloc_slot(xprt, GFP_NOWAIT|__GFP_NOWARN); + if (!IS_ERR(req)) + goto out_init_req; + switch (PTR_ERR(req)) { + case -ENOMEM: + dprintk("RPC: dynamic allocation of request slot " + "failed! Retrying\n"); + task->tk_status = -ENOMEM; + break; case -EAGAIN: - case -ENOTCONN: - return; + xprt_add_backlog(xprt, task); + dprintk("RPC: waiting for request slot\n"); default: - break; + task->tk_status = -EAGAIN; } - xprt_release_write(xprt, task); + spin_unlock(&xprt->reserve_lock); return; +out_init_req: + task->tk_status = 0; + task->tk_rqstp = req; + xprt_request_init(task, xprt); + spin_unlock(&xprt->reserve_lock); } +EXPORT_SYMBOL_GPL(xprt_alloc_slot); -static inline void do_xprt_reserve(struct rpc_task *task) +void xprt_lock_and_alloc_slot(struct rpc_xprt *xprt, struct rpc_task *task) { - struct rpc_xprt *xprt = task->tk_xprt; + /* Note: grabbing the xprt_lock_write() ensures that we throttle + * new slot allocation if the transport is congested (i.e. when + * reconnecting a stream transport or when out of socket write + * buffer space). + */ + if (xprt_lock_write(xprt, task)) { + xprt_alloc_slot(xprt, task); + xprt_release_write(xprt, task); + } +} +EXPORT_SYMBOL_GPL(xprt_lock_and_alloc_slot); + +static void xprt_free_slot(struct rpc_xprt *xprt, struct rpc_rqst *req) +{ + spin_lock(&xprt->reserve_lock); + if (!xprt_dynamic_free_slot(xprt, req)) { + memset(req, 0, sizeof(*req)); /* mark unused */ + list_add(&req->rq_list, &xprt->free); + } + xprt_wake_up_backlog(xprt); + spin_unlock(&xprt->reserve_lock); +} + +static void xprt_free_all_slots(struct rpc_xprt *xprt) +{ + struct rpc_rqst *req; + while (!list_empty(&xprt->free)) { + req = list_first_entry(&xprt->free, struct rpc_rqst, rq_list); + list_del(&req->rq_list); + kfree(req); + } +} + +struct rpc_xprt *xprt_alloc(struct net *net, size_t size, + unsigned int num_prealloc, + unsigned int max_alloc) +{ + struct rpc_xprt *xprt; + struct rpc_rqst *req; + int i; + + xprt = kzalloc(size, GFP_KERNEL); + if (xprt == NULL) + goto out; + + xprt_init(xprt, net); + + for (i = 0; i < num_prealloc; i++) { + req = kzalloc(sizeof(struct rpc_rqst), GFP_KERNEL); + if (!req) + goto out_free; + list_add(&req->rq_list, &xprt->free); + } + if (max_alloc > num_prealloc) + xprt->max_reqs = max_alloc; + else + xprt->max_reqs = num_prealloc; + xprt->min_reqs = num_prealloc; + atomic_set(&xprt->num_reqs, num_prealloc); + + return xprt; + +out_free: + xprt_free(xprt); +out: + return NULL; +} +EXPORT_SYMBOL_GPL(xprt_alloc); + +void xprt_free(struct rpc_xprt *xprt) +{ + put_net(xprt->xprt_net); + xprt_free_all_slots(xprt); + kfree(xprt); +} +EXPORT_SYMBOL_GPL(xprt_free); + +/** + * xprt_reserve - allocate an RPC request slot + * @task: RPC task requesting a slot allocation + * + * If the transport is marked as being congested, or if no more + * slots are available, place the task on the transport's + * backlog queue. + */ +void xprt_reserve(struct rpc_task *task) +{ + struct rpc_xprt *xprt; task->tk_status = 0; - if (task->tk_rqstp) + if (task->tk_rqstp != NULL) return; - if (!list_empty(&xprt->free)) { - struct rpc_rqst *req = list_entry(xprt->free.next, struct rpc_rqst, rq_list); - list_del_init(&req->rq_list); - task->tk_rqstp = req; - xprt_request_init(task, xprt); - return; - } - dprintk("RPC: waiting for request slot\n"); - task->tk_status = -EAGAIN; + task->tk_timeout = 0; - rpc_sleep_on(&xprt->backlog, task, NULL, NULL); + task->tk_status = -EAGAIN; + rcu_read_lock(); + xprt = rcu_dereference(task->tk_client->cl_xprt); + if (!xprt_throttle_congested(xprt, task)) + xprt->ops->alloc_slot(xprt, task); + rcu_read_unlock(); } /** - * xprt_reserve - allocate an RPC request slot + * xprt_retry_reserve - allocate an RPC request slot * @task: RPC task requesting a slot allocation * * If no more slots are available, place the task on the transport's * backlog queue. + * Note that the only difference with xprt_reserve is that we now + * ignore the value of the XPRT_CONGESTED flag. */ -void xprt_reserve(struct rpc_task *task) +void xprt_retry_reserve(struct rpc_task *task) { - struct rpc_xprt *xprt = task->tk_xprt; + struct rpc_xprt *xprt; - task->tk_status = -EIO; - if (!xprt->shutdown) { - spin_lock(&xprt->reserve_lock); - do_xprt_reserve(task); - spin_unlock(&xprt->reserve_lock); - } + task->tk_status = 0; + if (task->tk_rqstp != NULL) + return; + + task->tk_timeout = 0; + task->tk_status = -EAGAIN; + rcu_read_lock(); + xprt = rcu_dereference(task->tk_client->cl_xprt); + xprt->ops->alloc_slot(xprt, task); + rcu_read_unlock(); } -static inline u32 xprt_alloc_xid(struct rpc_xprt *xprt) +static inline __be32 xprt_alloc_xid(struct rpc_xprt *xprt) { - return xprt->xid++; + return (__force __be32)xprt->xid++; } static inline void xprt_init_xid(struct rpc_xprt *xprt) { - get_random_bytes(&xprt->xid, sizeof(xprt->xid)); + xprt->xid = prandom_u32(); } static void xprt_request_init(struct rpc_task *task, struct rpc_xprt *xprt) { struct rpc_rqst *req = task->tk_rqstp; - req->rq_timeout = xprt->timeout.to_initval; + INIT_LIST_HEAD(&req->rq_list); + req->rq_timeout = task->tk_client->cl_timeout->to_initval; req->rq_task = task; req->rq_xprt = xprt; + req->rq_buffer = NULL; req->rq_xid = xprt_alloc_xid(xprt); + req->rq_connect_cookie = xprt->connect_cookie - 1; + req->rq_bytes_sent = 0; + req->rq_snd_buf.len = 0; + req->rq_snd_buf.buflen = 0; + req->rq_rcv_buf.len = 0; + req->rq_rcv_buf.buflen = 0; req->rq_release_snd_buf = NULL; - dprintk("RPC: %4d reserved req %p xid %08x\n", task->tk_pid, + xprt_reset_majortimeo(req); + dprintk("RPC: %5u reserved req %p xid %08x\n", task->tk_pid, req, ntohl(req->rq_xid)); } @@ -851,11 +1215,25 @@ static void xprt_request_init(struct rpc_task *task, struct rpc_xprt *xprt) */ void xprt_release(struct rpc_task *task) { - struct rpc_xprt *xprt = task->tk_xprt; - struct rpc_rqst *req; + struct rpc_xprt *xprt; + struct rpc_rqst *req = task->tk_rqstp; - if (!(req = task->tk_rqstp)) + if (req == NULL) { + if (task->tk_client) { + rcu_read_lock(); + xprt = rcu_dereference(task->tk_client->cl_xprt); + if (xprt->snd_task == task) + xprt_release_write(xprt, task); + rcu_read_unlock(); + } return; + } + + xprt = req->rq_xprt; + if (task->tk_ops->rpc_count_stats != NULL) + task->tk_ops->rpc_count_stats(task, task->tk_calldata); + else if (task->tk_client) + rpc_count_iostats(task, task->tk_client->cl_metrics); spin_lock_bh(&xprt->transport_lock); xprt->ops->release_xprt(xprt, task); if (xprt->ops->release_request) @@ -863,125 +1241,104 @@ void xprt_release(struct rpc_task *task) if (!list_empty(&req->rq_list)) list_del(&req->rq_list); xprt->last_used = jiffies; - if (list_empty(&xprt->recv) && !xprt->shutdown) + if (list_empty(&xprt->recv) && xprt_has_timer(xprt)) mod_timer(&xprt->timer, xprt->last_used + xprt->idle_timeout); spin_unlock_bh(&xprt->transport_lock); + if (req->rq_buffer) + xprt->ops->buf_free(req->rq_buffer); + if (req->rq_cred != NULL) + put_rpccred(req->rq_cred); task->tk_rqstp = NULL; if (req->rq_release_snd_buf) req->rq_release_snd_buf(req); - memset(req, 0, sizeof(*req)); /* mark unused */ - dprintk("RPC: %4d release request %p\n", task->tk_pid, req); - - spin_lock(&xprt->reserve_lock); - list_add(&req->rq_list, &xprt->free); - rpc_wake_up_next(&xprt->backlog); - spin_unlock(&xprt->reserve_lock); + dprintk("RPC: %5u release request %p\n", task->tk_pid, req); + if (likely(!bc_prealloc(req))) + xprt_free_slot(xprt, req); + else + xprt_free_bc_request(req); } -/** - * xprt_set_timeout - set constant RPC timeout - * @to: RPC timeout parameters to set up - * @retr: number of retries - * @incr: amount of increase after each retry - * - */ -void xprt_set_timeout(struct rpc_timeout *to, unsigned int retr, unsigned long incr) +static void xprt_init(struct rpc_xprt *xprt, struct net *net) { - to->to_initval = - to->to_increment = incr; - to->to_maxval = to->to_initval + (incr * retr); - to->to_retries = retr; - to->to_exponential = 0; -} - -static struct rpc_xprt *xprt_setup(int proto, struct sockaddr_in *ap, struct rpc_timeout *to) -{ - int result; - struct rpc_xprt *xprt; - struct rpc_rqst *req; - - if ((xprt = kmalloc(sizeof(struct rpc_xprt), GFP_KERNEL)) == NULL) - return ERR_PTR(-ENOMEM); - memset(xprt, 0, sizeof(*xprt)); /* Nnnngh! */ - - xprt->addr = *ap; - - switch (proto) { - case IPPROTO_UDP: - result = xs_setup_udp(xprt, to); - break; - case IPPROTO_TCP: - result = xs_setup_tcp(xprt, to); - break; - default: - printk(KERN_ERR "RPC: unrecognized transport protocol: %d\n", - proto); - result = -EIO; - break; - } - if (result) { - kfree(xprt); - return ERR_PTR(result); - } + atomic_set(&xprt->count, 1); spin_lock_init(&xprt->transport_lock); spin_lock_init(&xprt->reserve_lock); INIT_LIST_HEAD(&xprt->free); INIT_LIST_HEAD(&xprt->recv); - INIT_WORK(&xprt->task_cleanup, xprt_autoclose, xprt); - init_timer(&xprt->timer); - xprt->timer.function = xprt_init_autodisconnect; - xprt->timer.data = (unsigned long) xprt; +#if defined(CONFIG_SUNRPC_BACKCHANNEL) + spin_lock_init(&xprt->bc_pa_lock); + INIT_LIST_HEAD(&xprt->bc_pa_list); +#endif /* CONFIG_SUNRPC_BACKCHANNEL */ + xprt->last_used = jiffies; xprt->cwnd = RPC_INITCWND; + xprt->bind_index = 0; + rpc_init_wait_queue(&xprt->binding, "xprt_binding"); rpc_init_wait_queue(&xprt->pending, "xprt_pending"); - rpc_init_wait_queue(&xprt->sending, "xprt_sending"); - rpc_init_wait_queue(&xprt->resend, "xprt_resend"); + rpc_init_priority_wait_queue(&xprt->sending, "xprt_sending"); rpc_init_priority_wait_queue(&xprt->backlog, "xprt_backlog"); - /* initialize free list */ - for (req = &xprt->slot[xprt->max_reqs-1]; req >= &xprt->slot[0]; req--) - list_add(&req->rq_list, &xprt->free); - xprt_init_xid(xprt); - dprintk("RPC: created transport %p with %u slots\n", xprt, - xprt->max_reqs); - - return xprt; + xprt->xprt_net = get_net(net); } /** - * xprt_create_proto - create an RPC client transport - * @proto: requested transport protocol - * @sap: remote peer's address - * @to: timeout parameters for new transport + * xprt_create_transport - create an RPC transport + * @args: rpc transport creation arguments * */ -struct rpc_xprt *xprt_create_proto(int proto, struct sockaddr_in *sap, struct rpc_timeout *to) +struct rpc_xprt *xprt_create_transport(struct xprt_create *args) { struct rpc_xprt *xprt; + struct xprt_class *t; - xprt = xprt_setup(proto, sap, to); - if (IS_ERR(xprt)) - dprintk("RPC: xprt_create_proto failed\n"); + spin_lock(&xprt_list_lock); + list_for_each_entry(t, &xprt_list, list) { + if (t->ident == args->ident) { + spin_unlock(&xprt_list_lock); + goto found; + } + } + spin_unlock(&xprt_list_lock); + printk(KERN_ERR "RPC: transport (%d) not supported\n", args->ident); + return ERR_PTR(-EIO); + +found: + xprt = t->setup(args); + if (IS_ERR(xprt)) { + dprintk("RPC: xprt_create_transport: failed, %ld\n", + -PTR_ERR(xprt)); + goto out; + } + if (args->flags & XPRT_CREATE_NO_IDLE_TIMEOUT) + xprt->idle_timeout = 0; + INIT_WORK(&xprt->task_cleanup, xprt_autoclose); + if (xprt_has_timer(xprt)) + setup_timer(&xprt->timer, xprt_init_autodisconnect, + (unsigned long)xprt); else - dprintk("RPC: xprt_create_proto created xprt %p\n", xprt); - return xprt; -} + init_timer(&xprt->timer); -static void xprt_shutdown(struct rpc_xprt *xprt) -{ - xprt->shutdown = 1; - rpc_wake_up(&xprt->sending); - rpc_wake_up(&xprt->resend); - xprt_wake_pending_tasks(xprt, -EIO); - rpc_wake_up(&xprt->backlog); - del_timer_sync(&xprt->timer); + if (strlen(args->servername) > RPC_MAXNETNAMELEN) { + xprt_destroy(xprt); + return ERR_PTR(-EINVAL); + } + xprt->servername = kstrdup(args->servername, GFP_KERNEL); + if (xprt->servername == NULL) { + xprt_destroy(xprt); + return ERR_PTR(-ENOMEM); + } + + dprintk("RPC: created transport %p with %u slots\n", xprt, + xprt->max_reqs); +out: + return xprt; } /** @@ -989,12 +1346,30 @@ static void xprt_shutdown(struct rpc_xprt *xprt) * @xprt: transport to destroy * */ -int xprt_destroy(struct rpc_xprt *xprt) +static void xprt_destroy(struct rpc_xprt *xprt) { - dprintk("RPC: destroying transport %p\n", xprt); - xprt_shutdown(xprt); + dprintk("RPC: destroying transport %p\n", xprt); + del_timer_sync(&xprt->timer); + + rpc_destroy_wait_queue(&xprt->binding); + rpc_destroy_wait_queue(&xprt->pending); + rpc_destroy_wait_queue(&xprt->sending); + rpc_destroy_wait_queue(&xprt->backlog); + cancel_work_sync(&xprt->task_cleanup); + kfree(xprt->servername); + /* + * Tear down transport state and free the rpc_xprt + */ xprt->ops->destroy(xprt); - kfree(xprt); +} - return 0; +/** + * xprt_put - release a reference to an RPC transport. + * @xprt: pointer to the transport + * + */ +void xprt_put(struct rpc_xprt *xprt) +{ + if (atomic_dec_and_test(&xprt->count)) + xprt_destroy(xprt); } diff --git a/net/sunrpc/xprtrdma/Makefile b/net/sunrpc/xprtrdma/Makefile new file mode 100644 index 00000000000..da5136fd569 --- /dev/null +++ b/net/sunrpc/xprtrdma/Makefile @@ -0,0 +1,8 @@ +obj-$(CONFIG_SUNRPC_XPRT_RDMA_CLIENT) += xprtrdma.o + +xprtrdma-y := transport.o rpc_rdma.o verbs.o + +obj-$(CONFIG_SUNRPC_XPRT_RDMA_SERVER) += svcrdma.o + +svcrdma-y := svc_rdma.o svc_rdma_transport.o \ + svc_rdma_marshal.o svc_rdma_sendto.o svc_rdma_recvfrom.o diff --git a/net/sunrpc/xprtrdma/rpc_rdma.c b/net/sunrpc/xprtrdma/rpc_rdma.c new file mode 100644 index 00000000000..693966d3f33 --- /dev/null +++ b/net/sunrpc/xprtrdma/rpc_rdma.c @@ -0,0 +1,872 @@ +/* + * Copyright (c) 2003-2007 Network Appliance, Inc. All rights reserved. + * + * This software is available to you under a choice of one of two + * licenses. You may choose to be licensed under the terms of the GNU + * General Public License (GPL) Version 2, available from the file + * COPYING in the main directory of this source tree, or the BSD-type + * license below: + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * + * Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * + * Redistributions in binary form must reproduce the above + * copyright notice, this list of conditions and the following + * disclaimer in the documentation and/or other materials provided + * with the distribution. + * + * Neither the name of the Network Appliance, Inc. nor the names of + * its contributors may be used to endorse or promote products + * derived from this software without specific prior written + * permission. + * + * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS + * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT + * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR + * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT + * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, + * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT + * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE + * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +/* + * rpc_rdma.c + * + * This file contains the guts of the RPC RDMA protocol, and + * does marshaling/unmarshaling, etc. It is also where interfacing + * to the Linux RPC framework lives. + */ + +#include "xprt_rdma.h" + +#include <linux/highmem.h> + +#ifdef RPC_DEBUG +# define RPCDBG_FACILITY RPCDBG_TRANS +#endif + +enum rpcrdma_chunktype { + rpcrdma_noch = 0, + rpcrdma_readch, + rpcrdma_areadch, + rpcrdma_writech, + rpcrdma_replych +}; + +#ifdef RPC_DEBUG +static const char transfertypes[][12] = { + "pure inline", /* no chunks */ + " read chunk", /* some argument via rdma read */ + "*read chunk", /* entire request via rdma read */ + "write chunk", /* some result via rdma write */ + "reply chunk" /* entire reply via rdma write */ +}; +#endif + +/* + * Chunk assembly from upper layer xdr_buf. + * + * Prepare the passed-in xdr_buf into representation as RPC/RDMA chunk + * elements. Segments are then coalesced when registered, if possible + * within the selected memreg mode. + * + * Returns positive number of segments converted, or a negative errno. + */ + +static int +rpcrdma_convert_iovs(struct xdr_buf *xdrbuf, unsigned int pos, + enum rpcrdma_chunktype type, struct rpcrdma_mr_seg *seg, int nsegs) +{ + int len, n = 0, p; + int page_base; + struct page **ppages; + + if (pos == 0 && xdrbuf->head[0].iov_len) { + seg[n].mr_page = NULL; + seg[n].mr_offset = xdrbuf->head[0].iov_base; + seg[n].mr_len = xdrbuf->head[0].iov_len; + ++n; + } + + len = xdrbuf->page_len; + ppages = xdrbuf->pages + (xdrbuf->page_base >> PAGE_SHIFT); + page_base = xdrbuf->page_base & ~PAGE_MASK; + p = 0; + while (len && n < nsegs) { + if (!ppages[p]) { + /* alloc the pagelist for receiving buffer */ + ppages[p] = alloc_page(GFP_ATOMIC); + if (!ppages[p]) + return -ENOMEM; + } + seg[n].mr_page = ppages[p]; + seg[n].mr_offset = (void *)(unsigned long) page_base; + seg[n].mr_len = min_t(u32, PAGE_SIZE - page_base, len); + if (seg[n].mr_len > PAGE_SIZE) + return -EIO; + len -= seg[n].mr_len; + ++n; + ++p; + page_base = 0; /* page offset only applies to first page */ + } + + /* Message overflows the seg array */ + if (len && n == nsegs) + return -EIO; + + if (xdrbuf->tail[0].iov_len) { + /* the rpcrdma protocol allows us to omit any trailing + * xdr pad bytes, saving the server an RDMA operation. */ + if (xdrbuf->tail[0].iov_len < 4 && xprt_rdma_pad_optimize) + return n; + if (n == nsegs) + /* Tail remains, but we're out of segments */ + return -EIO; + seg[n].mr_page = NULL; + seg[n].mr_offset = xdrbuf->tail[0].iov_base; + seg[n].mr_len = xdrbuf->tail[0].iov_len; + ++n; + } + + return n; +} + +/* + * Create read/write chunk lists, and reply chunks, for RDMA + * + * Assume check against THRESHOLD has been done, and chunks are required. + * Assume only encoding one list entry for read|write chunks. The NFSv3 + * protocol is simple enough to allow this as it only has a single "bulk + * result" in each procedure - complicated NFSv4 COMPOUNDs are not. (The + * RDMA/Sessions NFSv4 proposal addresses this for future v4 revs.) + * + * When used for a single reply chunk (which is a special write + * chunk used for the entire reply, rather than just the data), it + * is used primarily for READDIR and READLINK which would otherwise + * be severely size-limited by a small rdma inline read max. The server + * response will come back as an RDMA Write, followed by a message + * of type RDMA_NOMSG carrying the xid and length. As a result, reply + * chunks do not provide data alignment, however they do not require + * "fixup" (moving the response to the upper layer buffer) either. + * + * Encoding key for single-list chunks (HLOO = Handle32 Length32 Offset64): + * + * Read chunklist (a linked list): + * N elements, position P (same P for all chunks of same arg!): + * 1 - PHLOO - 1 - PHLOO - ... - 1 - PHLOO - 0 + * + * Write chunklist (a list of (one) counted array): + * N elements: + * 1 - N - HLOO - HLOO - ... - HLOO - 0 + * + * Reply chunk (a counted array): + * N elements: + * 1 - N - HLOO - HLOO - ... - HLOO + * + * Returns positive RPC/RDMA header size, or negative errno. + */ + +static ssize_t +rpcrdma_create_chunks(struct rpc_rqst *rqst, struct xdr_buf *target, + struct rpcrdma_msg *headerp, enum rpcrdma_chunktype type) +{ + struct rpcrdma_req *req = rpcr_to_rdmar(rqst); + struct rpcrdma_xprt *r_xprt = rpcx_to_rdmax(rqst->rq_xprt); + int n, nsegs, nchunks = 0; + unsigned int pos; + struct rpcrdma_mr_seg *seg = req->rl_segments; + struct rpcrdma_read_chunk *cur_rchunk = NULL; + struct rpcrdma_write_array *warray = NULL; + struct rpcrdma_write_chunk *cur_wchunk = NULL; + __be32 *iptr = headerp->rm_body.rm_chunks; + + if (type == rpcrdma_readch || type == rpcrdma_areadch) { + /* a read chunk - server will RDMA Read our memory */ + cur_rchunk = (struct rpcrdma_read_chunk *) iptr; + } else { + /* a write or reply chunk - server will RDMA Write our memory */ + *iptr++ = xdr_zero; /* encode a NULL read chunk list */ + if (type == rpcrdma_replych) + *iptr++ = xdr_zero; /* a NULL write chunk list */ + warray = (struct rpcrdma_write_array *) iptr; + cur_wchunk = (struct rpcrdma_write_chunk *) (warray + 1); + } + + if (type == rpcrdma_replych || type == rpcrdma_areadch) + pos = 0; + else + pos = target->head[0].iov_len; + + nsegs = rpcrdma_convert_iovs(target, pos, type, seg, RPCRDMA_MAX_SEGS); + if (nsegs < 0) + return nsegs; + + do { + n = rpcrdma_register_external(seg, nsegs, + cur_wchunk != NULL, r_xprt); + if (n <= 0) + goto out; + if (cur_rchunk) { /* read */ + cur_rchunk->rc_discrim = xdr_one; + /* all read chunks have the same "position" */ + cur_rchunk->rc_position = htonl(pos); + cur_rchunk->rc_target.rs_handle = htonl(seg->mr_rkey); + cur_rchunk->rc_target.rs_length = htonl(seg->mr_len); + xdr_encode_hyper( + (__be32 *)&cur_rchunk->rc_target.rs_offset, + seg->mr_base); + dprintk("RPC: %s: read chunk " + "elem %d@0x%llx:0x%x pos %u (%s)\n", __func__, + seg->mr_len, (unsigned long long)seg->mr_base, + seg->mr_rkey, pos, n < nsegs ? "more" : "last"); + cur_rchunk++; + r_xprt->rx_stats.read_chunk_count++; + } else { /* write/reply */ + cur_wchunk->wc_target.rs_handle = htonl(seg->mr_rkey); + cur_wchunk->wc_target.rs_length = htonl(seg->mr_len); + xdr_encode_hyper( + (__be32 *)&cur_wchunk->wc_target.rs_offset, + seg->mr_base); + dprintk("RPC: %s: %s chunk " + "elem %d@0x%llx:0x%x (%s)\n", __func__, + (type == rpcrdma_replych) ? "reply" : "write", + seg->mr_len, (unsigned long long)seg->mr_base, + seg->mr_rkey, n < nsegs ? "more" : "last"); + cur_wchunk++; + if (type == rpcrdma_replych) + r_xprt->rx_stats.reply_chunk_count++; + else + r_xprt->rx_stats.write_chunk_count++; + r_xprt->rx_stats.total_rdma_request += seg->mr_len; + } + nchunks++; + seg += n; + nsegs -= n; + } while (nsegs); + + /* success. all failures return above */ + req->rl_nchunks = nchunks; + + /* + * finish off header. If write, marshal discrim and nchunks. + */ + if (cur_rchunk) { + iptr = (__be32 *) cur_rchunk; + *iptr++ = xdr_zero; /* finish the read chunk list */ + *iptr++ = xdr_zero; /* encode a NULL write chunk list */ + *iptr++ = xdr_zero; /* encode a NULL reply chunk */ + } else { + warray->wc_discrim = xdr_one; + warray->wc_nchunks = htonl(nchunks); + iptr = (__be32 *) cur_wchunk; + if (type == rpcrdma_writech) { + *iptr++ = xdr_zero; /* finish the write chunk list */ + *iptr++ = xdr_zero; /* encode a NULL reply chunk */ + } + } + + /* + * Return header size. + */ + return (unsigned char *)iptr - (unsigned char *)headerp; + +out: + for (pos = 0; nchunks--;) + pos += rpcrdma_deregister_external( + &req->rl_segments[pos], r_xprt); + return n; +} + +/* + * Copy write data inline. + * This function is used for "small" requests. Data which is passed + * to RPC via iovecs (or page list) is copied directly into the + * pre-registered memory buffer for this request. For small amounts + * of data, this is efficient. The cutoff value is tunable. + */ +static int +rpcrdma_inline_pullup(struct rpc_rqst *rqst, int pad) +{ + int i, npages, curlen; + int copy_len; + unsigned char *srcp, *destp; + struct rpcrdma_xprt *r_xprt = rpcx_to_rdmax(rqst->rq_xprt); + int page_base; + struct page **ppages; + + destp = rqst->rq_svec[0].iov_base; + curlen = rqst->rq_svec[0].iov_len; + destp += curlen; + /* + * Do optional padding where it makes sense. Alignment of write + * payload can help the server, if our setting is accurate. + */ + pad -= (curlen + 36/*sizeof(struct rpcrdma_msg_padded)*/); + if (pad < 0 || rqst->rq_slen - curlen < RPCRDMA_INLINE_PAD_THRESH) + pad = 0; /* don't pad this request */ + + dprintk("RPC: %s: pad %d destp 0x%p len %d hdrlen %d\n", + __func__, pad, destp, rqst->rq_slen, curlen); + + copy_len = rqst->rq_snd_buf.page_len; + + if (rqst->rq_snd_buf.tail[0].iov_len) { + curlen = rqst->rq_snd_buf.tail[0].iov_len; + if (destp + copy_len != rqst->rq_snd_buf.tail[0].iov_base) { + memmove(destp + copy_len, + rqst->rq_snd_buf.tail[0].iov_base, curlen); + r_xprt->rx_stats.pullup_copy_count += curlen; + } + dprintk("RPC: %s: tail destp 0x%p len %d\n", + __func__, destp + copy_len, curlen); + rqst->rq_svec[0].iov_len += curlen; + } + r_xprt->rx_stats.pullup_copy_count += copy_len; + + page_base = rqst->rq_snd_buf.page_base; + ppages = rqst->rq_snd_buf.pages + (page_base >> PAGE_SHIFT); + page_base &= ~PAGE_MASK; + npages = PAGE_ALIGN(page_base+copy_len) >> PAGE_SHIFT; + for (i = 0; copy_len && i < npages; i++) { + curlen = PAGE_SIZE - page_base; + if (curlen > copy_len) + curlen = copy_len; + dprintk("RPC: %s: page %d destp 0x%p len %d curlen %d\n", + __func__, i, destp, copy_len, curlen); + srcp = kmap_atomic(ppages[i]); + memcpy(destp, srcp+page_base, curlen); + kunmap_atomic(srcp); + rqst->rq_svec[0].iov_len += curlen; + destp += curlen; + copy_len -= curlen; + page_base = 0; + } + /* header now contains entire send message */ + return pad; +} + +/* + * Marshal a request: the primary job of this routine is to choose + * the transfer modes. See comments below. + * + * Uses multiple RDMA IOVs for a request: + * [0] -- RPC RDMA header, which uses memory from the *start* of the + * preregistered buffer that already holds the RPC data in + * its middle. + * [1] -- the RPC header/data, marshaled by RPC and the NFS protocol. + * [2] -- optional padding. + * [3] -- if padded, header only in [1] and data here. + * + * Returns zero on success, otherwise a negative errno. + */ + +int +rpcrdma_marshal_req(struct rpc_rqst *rqst) +{ + struct rpc_xprt *xprt = rqst->rq_xprt; + struct rpcrdma_xprt *r_xprt = rpcx_to_rdmax(xprt); + struct rpcrdma_req *req = rpcr_to_rdmar(rqst); + char *base; + size_t rpclen, padlen; + ssize_t hdrlen; + enum rpcrdma_chunktype rtype, wtype; + struct rpcrdma_msg *headerp; + + /* + * rpclen gets amount of data in first buffer, which is the + * pre-registered buffer. + */ + base = rqst->rq_svec[0].iov_base; + rpclen = rqst->rq_svec[0].iov_len; + + /* build RDMA header in private area at front */ + headerp = (struct rpcrdma_msg *) req->rl_base; + /* don't htonl XID, it's already done in request */ + headerp->rm_xid = rqst->rq_xid; + headerp->rm_vers = xdr_one; + headerp->rm_credit = htonl(r_xprt->rx_buf.rb_max_requests); + headerp->rm_type = htonl(RDMA_MSG); + + /* + * Chunks needed for results? + * + * o If the expected result is under the inline threshold, all ops + * return as inline (but see later). + * o Large non-read ops return as a single reply chunk. + * o Large read ops return data as write chunk(s), header as inline. + * + * Note: the NFS code sending down multiple result segments implies + * the op is one of read, readdir[plus], readlink or NFSv4 getacl. + */ + + /* + * This code can handle read chunks, write chunks OR reply + * chunks -- only one type. If the request is too big to fit + * inline, then we will choose read chunks. If the request is + * a READ, then use write chunks to separate the file data + * into pages; otherwise use reply chunks. + */ + if (rqst->rq_rcv_buf.buflen <= RPCRDMA_INLINE_READ_THRESHOLD(rqst)) + wtype = rpcrdma_noch; + else if (rqst->rq_rcv_buf.page_len == 0) + wtype = rpcrdma_replych; + else if (rqst->rq_rcv_buf.flags & XDRBUF_READ) + wtype = rpcrdma_writech; + else + wtype = rpcrdma_replych; + + /* + * Chunks needed for arguments? + * + * o If the total request is under the inline threshold, all ops + * are sent as inline. + * o Large non-write ops are sent with the entire message as a + * single read chunk (protocol 0-position special case). + * o Large write ops transmit data as read chunk(s), header as + * inline. + * + * Note: the NFS code sending down multiple argument segments + * implies the op is a write. + * TBD check NFSv4 setacl + */ + if (rqst->rq_snd_buf.len <= RPCRDMA_INLINE_WRITE_THRESHOLD(rqst)) + rtype = rpcrdma_noch; + else if (rqst->rq_snd_buf.page_len == 0) + rtype = rpcrdma_areadch; + else + rtype = rpcrdma_readch; + + /* The following simplification is not true forever */ + if (rtype != rpcrdma_noch && wtype == rpcrdma_replych) + wtype = rpcrdma_noch; + if (rtype != rpcrdma_noch && wtype != rpcrdma_noch) { + dprintk("RPC: %s: cannot marshal multiple chunk lists\n", + __func__); + return -EIO; + } + + hdrlen = 28; /*sizeof *headerp;*/ + padlen = 0; + + /* + * Pull up any extra send data into the preregistered buffer. + * When padding is in use and applies to the transfer, insert + * it and change the message type. + */ + if (rtype == rpcrdma_noch) { + + padlen = rpcrdma_inline_pullup(rqst, + RPCRDMA_INLINE_PAD_VALUE(rqst)); + + if (padlen) { + headerp->rm_type = htonl(RDMA_MSGP); + headerp->rm_body.rm_padded.rm_align = + htonl(RPCRDMA_INLINE_PAD_VALUE(rqst)); + headerp->rm_body.rm_padded.rm_thresh = + htonl(RPCRDMA_INLINE_PAD_THRESH); + headerp->rm_body.rm_padded.rm_pempty[0] = xdr_zero; + headerp->rm_body.rm_padded.rm_pempty[1] = xdr_zero; + headerp->rm_body.rm_padded.rm_pempty[2] = xdr_zero; + hdrlen += 2 * sizeof(u32); /* extra words in padhdr */ + if (wtype != rpcrdma_noch) { + dprintk("RPC: %s: invalid chunk list\n", + __func__); + return -EIO; + } + } else { + headerp->rm_body.rm_nochunks.rm_empty[0] = xdr_zero; + headerp->rm_body.rm_nochunks.rm_empty[1] = xdr_zero; + headerp->rm_body.rm_nochunks.rm_empty[2] = xdr_zero; + /* new length after pullup */ + rpclen = rqst->rq_svec[0].iov_len; + /* + * Currently we try to not actually use read inline. + * Reply chunks have the desirable property that + * they land, packed, directly in the target buffers + * without headers, so they require no fixup. The + * additional RDMA Write op sends the same amount + * of data, streams on-the-wire and adds no overhead + * on receive. Therefore, we request a reply chunk + * for non-writes wherever feasible and efficient. + */ + if (wtype == rpcrdma_noch) + wtype = rpcrdma_replych; + } + } + + /* + * Marshal chunks. This routine will return the header length + * consumed by marshaling. + */ + if (rtype != rpcrdma_noch) { + hdrlen = rpcrdma_create_chunks(rqst, + &rqst->rq_snd_buf, headerp, rtype); + wtype = rtype; /* simplify dprintk */ + + } else if (wtype != rpcrdma_noch) { + hdrlen = rpcrdma_create_chunks(rqst, + &rqst->rq_rcv_buf, headerp, wtype); + } + if (hdrlen < 0) + return hdrlen; + + dprintk("RPC: %s: %s: hdrlen %zd rpclen %zd padlen %zd" + " headerp 0x%p base 0x%p lkey 0x%x\n", + __func__, transfertypes[wtype], hdrlen, rpclen, padlen, + headerp, base, req->rl_iov.lkey); + + /* + * initialize send_iov's - normally only two: rdma chunk header and + * single preregistered RPC header buffer, but if padding is present, + * then use a preregistered (and zeroed) pad buffer between the RPC + * header and any write data. In all non-rdma cases, any following + * data has been copied into the RPC header buffer. + */ + req->rl_send_iov[0].addr = req->rl_iov.addr; + req->rl_send_iov[0].length = hdrlen; + req->rl_send_iov[0].lkey = req->rl_iov.lkey; + + req->rl_send_iov[1].addr = req->rl_iov.addr + (base - req->rl_base); + req->rl_send_iov[1].length = rpclen; + req->rl_send_iov[1].lkey = req->rl_iov.lkey; + + req->rl_niovs = 2; + + if (padlen) { + struct rpcrdma_ep *ep = &r_xprt->rx_ep; + + req->rl_send_iov[2].addr = ep->rep_pad.addr; + req->rl_send_iov[2].length = padlen; + req->rl_send_iov[2].lkey = ep->rep_pad.lkey; + + req->rl_send_iov[3].addr = req->rl_send_iov[1].addr + rpclen; + req->rl_send_iov[3].length = rqst->rq_slen - rpclen; + req->rl_send_iov[3].lkey = req->rl_iov.lkey; + + req->rl_niovs = 4; + } + + return 0; +} + +/* + * Chase down a received write or reply chunklist to get length + * RDMA'd by server. See map at rpcrdma_create_chunks()! :-) + */ +static int +rpcrdma_count_chunks(struct rpcrdma_rep *rep, unsigned int max, int wrchunk, __be32 **iptrp) +{ + unsigned int i, total_len; + struct rpcrdma_write_chunk *cur_wchunk; + + i = ntohl(**iptrp); /* get array count */ + if (i > max) + return -1; + cur_wchunk = (struct rpcrdma_write_chunk *) (*iptrp + 1); + total_len = 0; + while (i--) { + struct rpcrdma_segment *seg = &cur_wchunk->wc_target; + ifdebug(FACILITY) { + u64 off; + xdr_decode_hyper((__be32 *)&seg->rs_offset, &off); + dprintk("RPC: %s: chunk %d@0x%llx:0x%x\n", + __func__, + ntohl(seg->rs_length), + (unsigned long long)off, + ntohl(seg->rs_handle)); + } + total_len += ntohl(seg->rs_length); + ++cur_wchunk; + } + /* check and adjust for properly terminated write chunk */ + if (wrchunk) { + __be32 *w = (__be32 *) cur_wchunk; + if (*w++ != xdr_zero) + return -1; + cur_wchunk = (struct rpcrdma_write_chunk *) w; + } + if ((char *) cur_wchunk > rep->rr_base + rep->rr_len) + return -1; + + *iptrp = (__be32 *) cur_wchunk; + return total_len; +} + +/* + * Scatter inline received data back into provided iov's. + */ +static void +rpcrdma_inline_fixup(struct rpc_rqst *rqst, char *srcp, int copy_len, int pad) +{ + int i, npages, curlen, olen; + char *destp; + struct page **ppages; + int page_base; + + curlen = rqst->rq_rcv_buf.head[0].iov_len; + if (curlen > copy_len) { /* write chunk header fixup */ + curlen = copy_len; + rqst->rq_rcv_buf.head[0].iov_len = curlen; + } + + dprintk("RPC: %s: srcp 0x%p len %d hdrlen %d\n", + __func__, srcp, copy_len, curlen); + + /* Shift pointer for first receive segment only */ + rqst->rq_rcv_buf.head[0].iov_base = srcp; + srcp += curlen; + copy_len -= curlen; + + olen = copy_len; + i = 0; + rpcx_to_rdmax(rqst->rq_xprt)->rx_stats.fixup_copy_count += olen; + page_base = rqst->rq_rcv_buf.page_base; + ppages = rqst->rq_rcv_buf.pages + (page_base >> PAGE_SHIFT); + page_base &= ~PAGE_MASK; + + if (copy_len && rqst->rq_rcv_buf.page_len) { + npages = PAGE_ALIGN(page_base + + rqst->rq_rcv_buf.page_len) >> PAGE_SHIFT; + for (; i < npages; i++) { + curlen = PAGE_SIZE - page_base; + if (curlen > copy_len) + curlen = copy_len; + dprintk("RPC: %s: page %d" + " srcp 0x%p len %d curlen %d\n", + __func__, i, srcp, copy_len, curlen); + destp = kmap_atomic(ppages[i]); + memcpy(destp + page_base, srcp, curlen); + flush_dcache_page(ppages[i]); + kunmap_atomic(destp); + srcp += curlen; + copy_len -= curlen; + if (copy_len == 0) + break; + page_base = 0; + } + } + + if (copy_len && rqst->rq_rcv_buf.tail[0].iov_len) { + curlen = copy_len; + if (curlen > rqst->rq_rcv_buf.tail[0].iov_len) + curlen = rqst->rq_rcv_buf.tail[0].iov_len; + if (rqst->rq_rcv_buf.tail[0].iov_base != srcp) + memmove(rqst->rq_rcv_buf.tail[0].iov_base, srcp, curlen); + dprintk("RPC: %s: tail srcp 0x%p len %d curlen %d\n", + __func__, srcp, copy_len, curlen); + rqst->rq_rcv_buf.tail[0].iov_len = curlen; + copy_len -= curlen; ++i; + } else + rqst->rq_rcv_buf.tail[0].iov_len = 0; + + if (pad) { + /* implicit padding on terminal chunk */ + unsigned char *p = rqst->rq_rcv_buf.tail[0].iov_base; + while (pad--) + p[rqst->rq_rcv_buf.tail[0].iov_len++] = 0; + } + + if (copy_len) + dprintk("RPC: %s: %d bytes in" + " %d extra segments (%d lost)\n", + __func__, olen, i, copy_len); + + /* TBD avoid a warning from call_decode() */ + rqst->rq_private_buf = rqst->rq_rcv_buf; +} + +void +rpcrdma_connect_worker(struct work_struct *work) +{ + struct rpcrdma_ep *ep = + container_of(work, struct rpcrdma_ep, rep_connect_worker.work); + struct rpc_xprt *xprt = ep->rep_xprt; + + spin_lock_bh(&xprt->transport_lock); + if (++xprt->connect_cookie == 0) /* maintain a reserved value */ + ++xprt->connect_cookie; + if (ep->rep_connected > 0) { + if (!xprt_test_and_set_connected(xprt)) + xprt_wake_pending_tasks(xprt, 0); + } else { + if (xprt_test_and_clear_connected(xprt)) + xprt_wake_pending_tasks(xprt, -ENOTCONN); + } + spin_unlock_bh(&xprt->transport_lock); +} + +/* + * This function is called when an async event is posted to + * the connection which changes the connection state. All it + * does at this point is mark the connection up/down, the rpc + * timers do the rest. + */ +void +rpcrdma_conn_func(struct rpcrdma_ep *ep) +{ + schedule_delayed_work(&ep->rep_connect_worker, 0); +} + +/* + * Called as a tasklet to do req/reply match and complete a request + * Errors must result in the RPC task either being awakened, or + * allowed to timeout, to discover the errors at that time. + */ +void +rpcrdma_reply_handler(struct rpcrdma_rep *rep) +{ + struct rpcrdma_msg *headerp; + struct rpcrdma_req *req; + struct rpc_rqst *rqst; + struct rpc_xprt *xprt = rep->rr_xprt; + struct rpcrdma_xprt *r_xprt = rpcx_to_rdmax(xprt); + __be32 *iptr; + int rdmalen, status; + unsigned long cwnd; + + /* Check status. If bad, signal disconnect and return rep to pool */ + if (rep->rr_len == ~0U) { + rpcrdma_recv_buffer_put(rep); + if (r_xprt->rx_ep.rep_connected == 1) { + r_xprt->rx_ep.rep_connected = -EIO; + rpcrdma_conn_func(&r_xprt->rx_ep); + } + return; + } + if (rep->rr_len < 28) { + dprintk("RPC: %s: short/invalid reply\n", __func__); + goto repost; + } + headerp = (struct rpcrdma_msg *) rep->rr_base; + if (headerp->rm_vers != xdr_one) { + dprintk("RPC: %s: invalid version %d\n", + __func__, ntohl(headerp->rm_vers)); + goto repost; + } + + /* Get XID and try for a match. */ + spin_lock(&xprt->transport_lock); + rqst = xprt_lookup_rqst(xprt, headerp->rm_xid); + if (rqst == NULL) { + spin_unlock(&xprt->transport_lock); + dprintk("RPC: %s: reply 0x%p failed " + "to match any request xid 0x%08x len %d\n", + __func__, rep, headerp->rm_xid, rep->rr_len); +repost: + r_xprt->rx_stats.bad_reply_count++; + rep->rr_func = rpcrdma_reply_handler; + if (rpcrdma_ep_post_recv(&r_xprt->rx_ia, &r_xprt->rx_ep, rep)) + rpcrdma_recv_buffer_put(rep); + + return; + } + + /* get request object */ + req = rpcr_to_rdmar(rqst); + if (req->rl_reply) { + spin_unlock(&xprt->transport_lock); + dprintk("RPC: %s: duplicate reply 0x%p to RPC " + "request 0x%p: xid 0x%08x\n", __func__, rep, req, + headerp->rm_xid); + goto repost; + } + + dprintk("RPC: %s: reply 0x%p completes request 0x%p\n" + " RPC request 0x%p xid 0x%08x\n", + __func__, rep, req, rqst, headerp->rm_xid); + + /* from here on, the reply is no longer an orphan */ + req->rl_reply = rep; + xprt->reestablish_timeout = 0; + + /* check for expected message types */ + /* The order of some of these tests is important. */ + switch (headerp->rm_type) { + case htonl(RDMA_MSG): + /* never expect read chunks */ + /* never expect reply chunks (two ways to check) */ + /* never expect write chunks without having offered RDMA */ + if (headerp->rm_body.rm_chunks[0] != xdr_zero || + (headerp->rm_body.rm_chunks[1] == xdr_zero && + headerp->rm_body.rm_chunks[2] != xdr_zero) || + (headerp->rm_body.rm_chunks[1] != xdr_zero && + req->rl_nchunks == 0)) + goto badheader; + if (headerp->rm_body.rm_chunks[1] != xdr_zero) { + /* count any expected write chunks in read reply */ + /* start at write chunk array count */ + iptr = &headerp->rm_body.rm_chunks[2]; + rdmalen = rpcrdma_count_chunks(rep, + req->rl_nchunks, 1, &iptr); + /* check for validity, and no reply chunk after */ + if (rdmalen < 0 || *iptr++ != xdr_zero) + goto badheader; + rep->rr_len -= + ((unsigned char *)iptr - (unsigned char *)headerp); + status = rep->rr_len + rdmalen; + r_xprt->rx_stats.total_rdma_reply += rdmalen; + /* special case - last chunk may omit padding */ + if (rdmalen &= 3) { + rdmalen = 4 - rdmalen; + status += rdmalen; + } + } else { + /* else ordinary inline */ + rdmalen = 0; + iptr = (__be32 *)((unsigned char *)headerp + 28); + rep->rr_len -= 28; /*sizeof *headerp;*/ + status = rep->rr_len; + } + /* Fix up the rpc results for upper layer */ + rpcrdma_inline_fixup(rqst, (char *)iptr, rep->rr_len, rdmalen); + break; + + case htonl(RDMA_NOMSG): + /* never expect read or write chunks, always reply chunks */ + if (headerp->rm_body.rm_chunks[0] != xdr_zero || + headerp->rm_body.rm_chunks[1] != xdr_zero || + headerp->rm_body.rm_chunks[2] != xdr_one || + req->rl_nchunks == 0) + goto badheader; + iptr = (__be32 *)((unsigned char *)headerp + 28); + rdmalen = rpcrdma_count_chunks(rep, req->rl_nchunks, 0, &iptr); + if (rdmalen < 0) + goto badheader; + r_xprt->rx_stats.total_rdma_reply += rdmalen; + /* Reply chunk buffer already is the reply vector - no fixup. */ + status = rdmalen; + break; + +badheader: + default: + dprintk("%s: invalid rpcrdma reply header (type %d):" + " chunks[012] == %d %d %d" + " expected chunks <= %d\n", + __func__, ntohl(headerp->rm_type), + headerp->rm_body.rm_chunks[0], + headerp->rm_body.rm_chunks[1], + headerp->rm_body.rm_chunks[2], + req->rl_nchunks); + status = -EIO; + r_xprt->rx_stats.bad_reply_count++; + break; + } + + cwnd = xprt->cwnd; + xprt->cwnd = atomic_read(&r_xprt->rx_buf.rb_credits) << RPC_CWNDSHIFT; + if (xprt->cwnd > cwnd) + xprt_release_rqst_cong(rqst->rq_task); + + dprintk("RPC: %s: xprt_complete_rqst(0x%p, 0x%p, %d)\n", + __func__, xprt, rqst, status); + xprt_complete_rqst(rqst->rq_task, status); + spin_unlock(&xprt->transport_lock); +} diff --git a/net/sunrpc/xprtrdma/svc_rdma.c b/net/sunrpc/xprtrdma/svc_rdma.c new file mode 100644 index 00000000000..c1b6270262c --- /dev/null +++ b/net/sunrpc/xprtrdma/svc_rdma.c @@ -0,0 +1,302 @@ +/* + * Copyright (c) 2005-2006 Network Appliance, Inc. All rights reserved. + * + * This software is available to you under a choice of one of two + * licenses. You may choose to be licensed under the terms of the GNU + * General Public License (GPL) Version 2, available from the file + * COPYING in the main directory of this source tree, or the BSD-type + * license below: + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * + * Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * + * Redistributions in binary form must reproduce the above + * copyright notice, this list of conditions and the following + * disclaimer in the documentation and/or other materials provided + * with the distribution. + * + * Neither the name of the Network Appliance, Inc. nor the names of + * its contributors may be used to endorse or promote products + * derived from this software without specific prior written + * permission. + * + * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS + * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT + * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR + * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT + * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, + * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT + * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE + * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + * + * Author: Tom Tucker <tom@opengridcomputing.com> + */ +#include <linux/module.h> +#include <linux/init.h> +#include <linux/slab.h> +#include <linux/fs.h> +#include <linux/sysctl.h> +#include <linux/workqueue.h> +#include <linux/sunrpc/clnt.h> +#include <linux/sunrpc/sched.h> +#include <linux/sunrpc/svc_rdma.h> +#include "xprt_rdma.h" + +#define RPCDBG_FACILITY RPCDBG_SVCXPRT + +/* RPC/RDMA parameters */ +unsigned int svcrdma_ord = RPCRDMA_ORD; +static unsigned int min_ord = 1; +static unsigned int max_ord = 4096; +unsigned int svcrdma_max_requests = RPCRDMA_MAX_REQUESTS; +static unsigned int min_max_requests = 4; +static unsigned int max_max_requests = 16384; +unsigned int svcrdma_max_req_size = RPCRDMA_MAX_REQ_SIZE; +static unsigned int min_max_inline = 4096; +static unsigned int max_max_inline = 65536; + +atomic_t rdma_stat_recv; +atomic_t rdma_stat_read; +atomic_t rdma_stat_write; +atomic_t rdma_stat_sq_starve; +atomic_t rdma_stat_rq_starve; +atomic_t rdma_stat_rq_poll; +atomic_t rdma_stat_rq_prod; +atomic_t rdma_stat_sq_poll; +atomic_t rdma_stat_sq_prod; + +/* Temporary NFS request map and context caches */ +struct kmem_cache *svc_rdma_map_cachep; +struct kmem_cache *svc_rdma_ctxt_cachep; + +struct workqueue_struct *svc_rdma_wq; + +/* + * This function implements reading and resetting an atomic_t stat + * variable through read/write to a proc file. Any write to the file + * resets the associated statistic to zero. Any read returns it's + * current value. + */ +static int read_reset_stat(struct ctl_table *table, int write, + void __user *buffer, size_t *lenp, + loff_t *ppos) +{ + atomic_t *stat = (atomic_t *)table->data; + + if (!stat) + return -EINVAL; + + if (write) + atomic_set(stat, 0); + else { + char str_buf[32]; + char *data; + int len = snprintf(str_buf, 32, "%d\n", atomic_read(stat)); + if (len >= 32) + return -EFAULT; + len = strlen(str_buf); + if (*ppos > len) { + *lenp = 0; + return 0; + } + data = &str_buf[*ppos]; + len -= *ppos; + if (len > *lenp) + len = *lenp; + if (len && copy_to_user(buffer, str_buf, len)) + return -EFAULT; + *lenp = len; + *ppos += len; + } + return 0; +} + +static struct ctl_table_header *svcrdma_table_header; +static struct ctl_table svcrdma_parm_table[] = { + { + .procname = "max_requests", + .data = &svcrdma_max_requests, + .maxlen = sizeof(unsigned int), + .mode = 0644, + .proc_handler = proc_dointvec_minmax, + .extra1 = &min_max_requests, + .extra2 = &max_max_requests + }, + { + .procname = "max_req_size", + .data = &svcrdma_max_req_size, + .maxlen = sizeof(unsigned int), + .mode = 0644, + .proc_handler = proc_dointvec_minmax, + .extra1 = &min_max_inline, + .extra2 = &max_max_inline + }, + { + .procname = "max_outbound_read_requests", + .data = &svcrdma_ord, + .maxlen = sizeof(unsigned int), + .mode = 0644, + .proc_handler = proc_dointvec_minmax, + .extra1 = &min_ord, + .extra2 = &max_ord, + }, + + { + .procname = "rdma_stat_read", + .data = &rdma_stat_read, + .maxlen = sizeof(atomic_t), + .mode = 0644, + .proc_handler = read_reset_stat, + }, + { + .procname = "rdma_stat_recv", + .data = &rdma_stat_recv, + .maxlen = sizeof(atomic_t), + .mode = 0644, + .proc_handler = read_reset_stat, + }, + { + .procname = "rdma_stat_write", + .data = &rdma_stat_write, + .maxlen = sizeof(atomic_t), + .mode = 0644, + .proc_handler = read_reset_stat, + }, + { + .procname = "rdma_stat_sq_starve", + .data = &rdma_stat_sq_starve, + .maxlen = sizeof(atomic_t), + .mode = 0644, + .proc_handler = read_reset_stat, + }, + { + .procname = "rdma_stat_rq_starve", + .data = &rdma_stat_rq_starve, + .maxlen = sizeof(atomic_t), + .mode = 0644, + .proc_handler = read_reset_stat, + }, + { + .procname = "rdma_stat_rq_poll", + .data = &rdma_stat_rq_poll, + .maxlen = sizeof(atomic_t), + .mode = 0644, + .proc_handler = read_reset_stat, + }, + { + .procname = "rdma_stat_rq_prod", + .data = &rdma_stat_rq_prod, + .maxlen = sizeof(atomic_t), + .mode = 0644, + .proc_handler = read_reset_stat, + }, + { + .procname = "rdma_stat_sq_poll", + .data = &rdma_stat_sq_poll, + .maxlen = sizeof(atomic_t), + .mode = 0644, + .proc_handler = read_reset_stat, + }, + { + .procname = "rdma_stat_sq_prod", + .data = &rdma_stat_sq_prod, + .maxlen = sizeof(atomic_t), + .mode = 0644, + .proc_handler = read_reset_stat, + }, + { }, +}; + +static struct ctl_table svcrdma_table[] = { + { + .procname = "svc_rdma", + .mode = 0555, + .child = svcrdma_parm_table + }, + { }, +}; + +static struct ctl_table svcrdma_root_table[] = { + { + .procname = "sunrpc", + .mode = 0555, + .child = svcrdma_table + }, + { }, +}; + +void svc_rdma_cleanup(void) +{ + dprintk("SVCRDMA Module Removed, deregister RPC RDMA transport\n"); + destroy_workqueue(svc_rdma_wq); + if (svcrdma_table_header) { + unregister_sysctl_table(svcrdma_table_header); + svcrdma_table_header = NULL; + } + svc_unreg_xprt_class(&svc_rdma_class); + kmem_cache_destroy(svc_rdma_map_cachep); + kmem_cache_destroy(svc_rdma_ctxt_cachep); +} + +int svc_rdma_init(void) +{ + dprintk("SVCRDMA Module Init, register RPC RDMA transport\n"); + dprintk("\tsvcrdma_ord : %d\n", svcrdma_ord); + dprintk("\tmax_requests : %d\n", svcrdma_max_requests); + dprintk("\tsq_depth : %d\n", + svcrdma_max_requests * RPCRDMA_SQ_DEPTH_MULT); + dprintk("\tmax_inline : %d\n", svcrdma_max_req_size); + + svc_rdma_wq = alloc_workqueue("svc_rdma", 0, 0); + if (!svc_rdma_wq) + return -ENOMEM; + + if (!svcrdma_table_header) + svcrdma_table_header = + register_sysctl_table(svcrdma_root_table); + + /* Create the temporary map cache */ + svc_rdma_map_cachep = kmem_cache_create("svc_rdma_map_cache", + sizeof(struct svc_rdma_req_map), + 0, + SLAB_HWCACHE_ALIGN, + NULL); + if (!svc_rdma_map_cachep) { + printk(KERN_INFO "Could not allocate map cache.\n"); + goto err0; + } + + /* Create the temporary context cache */ + svc_rdma_ctxt_cachep = + kmem_cache_create("svc_rdma_ctxt_cache", + sizeof(struct svc_rdma_op_ctxt), + 0, + SLAB_HWCACHE_ALIGN, + NULL); + if (!svc_rdma_ctxt_cachep) { + printk(KERN_INFO "Could not allocate WR ctxt cache.\n"); + goto err1; + } + + /* Register RDMA with the SVC transport switch */ + svc_reg_xprt_class(&svc_rdma_class); + return 0; + err1: + kmem_cache_destroy(svc_rdma_map_cachep); + err0: + unregister_sysctl_table(svcrdma_table_header); + destroy_workqueue(svc_rdma_wq); + return -ENOMEM; +} +MODULE_AUTHOR("Tom Tucker <tom@opengridcomputing.com>"); +MODULE_DESCRIPTION("SVC RDMA Transport"); +MODULE_LICENSE("Dual BSD/GPL"); +module_init(svc_rdma_init); +module_exit(svc_rdma_cleanup); diff --git a/net/sunrpc/xprtrdma/svc_rdma_marshal.c b/net/sunrpc/xprtrdma/svc_rdma_marshal.c new file mode 100644 index 00000000000..65b146297f5 --- /dev/null +++ b/net/sunrpc/xprtrdma/svc_rdma_marshal.c @@ -0,0 +1,386 @@ +/* + * Copyright (c) 2005-2006 Network Appliance, Inc. All rights reserved. + * + * This software is available to you under a choice of one of two + * licenses. You may choose to be licensed under the terms of the GNU + * General Public License (GPL) Version 2, available from the file + * COPYING in the main directory of this source tree, or the BSD-type + * license below: + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * + * Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * + * Redistributions in binary form must reproduce the above + * copyright notice, this list of conditions and the following + * disclaimer in the documentation and/or other materials provided + * with the distribution. + * + * Neither the name of the Network Appliance, Inc. nor the names of + * its contributors may be used to endorse or promote products + * derived from this software without specific prior written + * permission. + * + * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS + * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT + * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR + * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT + * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, + * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT + * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE + * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + * + * Author: Tom Tucker <tom@opengridcomputing.com> + */ + +#include <linux/sunrpc/xdr.h> +#include <linux/sunrpc/debug.h> +#include <asm/unaligned.h> +#include <linux/sunrpc/rpc_rdma.h> +#include <linux/sunrpc/svc_rdma.h> + +#define RPCDBG_FACILITY RPCDBG_SVCXPRT + +/* + * Decodes a read chunk list. The expected format is as follows: + * descrim : xdr_one + * position : u32 offset into XDR stream + * handle : u32 RKEY + * . . . + * end-of-list: xdr_zero + */ +static u32 *decode_read_list(u32 *va, u32 *vaend) +{ + struct rpcrdma_read_chunk *ch = (struct rpcrdma_read_chunk *)va; + + while (ch->rc_discrim != xdr_zero) { + if (((unsigned long)ch + sizeof(struct rpcrdma_read_chunk)) > + (unsigned long)vaend) { + dprintk("svcrdma: vaend=%p, ch=%p\n", vaend, ch); + return NULL; + } + ch++; + } + return (u32 *)&ch->rc_position; +} + +/* + * Determine number of chunks and total bytes in chunk list. The chunk + * list has already been verified to fit within the RPCRDMA header. + */ +void svc_rdma_rcl_chunk_counts(struct rpcrdma_read_chunk *ch, + int *ch_count, int *byte_count) +{ + /* compute the number of bytes represented by read chunks */ + *byte_count = 0; + *ch_count = 0; + for (; ch->rc_discrim != 0; ch++) { + *byte_count = *byte_count + ntohl(ch->rc_target.rs_length); + *ch_count = *ch_count + 1; + } +} + +/* + * Decodes a write chunk list. The expected format is as follows: + * descrim : xdr_one + * nchunks : <count> + * handle : u32 RKEY ---+ + * length : u32 <len of segment> | + * offset : remove va + <count> + * . . . | + * ---+ + */ +static u32 *decode_write_list(u32 *va, u32 *vaend) +{ + unsigned long start, end; + int nchunks; + + struct rpcrdma_write_array *ary = + (struct rpcrdma_write_array *)va; + + /* Check for not write-array */ + if (ary->wc_discrim == xdr_zero) + return (u32 *)&ary->wc_nchunks; + + if ((unsigned long)ary + sizeof(struct rpcrdma_write_array) > + (unsigned long)vaend) { + dprintk("svcrdma: ary=%p, vaend=%p\n", ary, vaend); + return NULL; + } + nchunks = ntohl(ary->wc_nchunks); + + start = (unsigned long)&ary->wc_array[0]; + end = (unsigned long)vaend; + if (nchunks < 0 || + nchunks > (SIZE_MAX - start) / sizeof(struct rpcrdma_write_chunk) || + (start + (sizeof(struct rpcrdma_write_chunk) * nchunks)) > end) { + dprintk("svcrdma: ary=%p, wc_nchunks=%d, vaend=%p\n", + ary, nchunks, vaend); + return NULL; + } + /* + * rs_length is the 2nd 4B field in wc_target and taking its + * address skips the list terminator + */ + return (u32 *)&ary->wc_array[nchunks].wc_target.rs_length; +} + +static u32 *decode_reply_array(u32 *va, u32 *vaend) +{ + unsigned long start, end; + int nchunks; + struct rpcrdma_write_array *ary = + (struct rpcrdma_write_array *)va; + + /* Check for no reply-array */ + if (ary->wc_discrim == xdr_zero) + return (u32 *)&ary->wc_nchunks; + + if ((unsigned long)ary + sizeof(struct rpcrdma_write_array) > + (unsigned long)vaend) { + dprintk("svcrdma: ary=%p, vaend=%p\n", ary, vaend); + return NULL; + } + nchunks = ntohl(ary->wc_nchunks); + + start = (unsigned long)&ary->wc_array[0]; + end = (unsigned long)vaend; + if (nchunks < 0 || + nchunks > (SIZE_MAX - start) / sizeof(struct rpcrdma_write_chunk) || + (start + (sizeof(struct rpcrdma_write_chunk) * nchunks)) > end) { + dprintk("svcrdma: ary=%p, wc_nchunks=%d, vaend=%p\n", + ary, nchunks, vaend); + return NULL; + } + return (u32 *)&ary->wc_array[nchunks]; +} + +int svc_rdma_xdr_decode_req(struct rpcrdma_msg **rdma_req, + struct svc_rqst *rqstp) +{ + struct rpcrdma_msg *rmsgp = NULL; + u32 *va; + u32 *vaend; + u32 hdr_len; + + rmsgp = (struct rpcrdma_msg *)rqstp->rq_arg.head[0].iov_base; + + /* Verify that there's enough bytes for header + something */ + if (rqstp->rq_arg.len <= RPCRDMA_HDRLEN_MIN) { + dprintk("svcrdma: header too short = %d\n", + rqstp->rq_arg.len); + return -EINVAL; + } + + /* Decode the header */ + rmsgp->rm_xid = ntohl(rmsgp->rm_xid); + rmsgp->rm_vers = ntohl(rmsgp->rm_vers); + rmsgp->rm_credit = ntohl(rmsgp->rm_credit); + rmsgp->rm_type = ntohl(rmsgp->rm_type); + + if (rmsgp->rm_vers != RPCRDMA_VERSION) + return -ENOSYS; + + /* Pull in the extra for the padded case and bump our pointer */ + if (rmsgp->rm_type == RDMA_MSGP) { + int hdrlen; + rmsgp->rm_body.rm_padded.rm_align = + ntohl(rmsgp->rm_body.rm_padded.rm_align); + rmsgp->rm_body.rm_padded.rm_thresh = + ntohl(rmsgp->rm_body.rm_padded.rm_thresh); + + va = &rmsgp->rm_body.rm_padded.rm_pempty[4]; + rqstp->rq_arg.head[0].iov_base = va; + hdrlen = (u32)((unsigned long)va - (unsigned long)rmsgp); + rqstp->rq_arg.head[0].iov_len -= hdrlen; + if (hdrlen > rqstp->rq_arg.len) + return -EINVAL; + return hdrlen; + } + + /* The chunk list may contain either a read chunk list or a write + * chunk list and a reply chunk list. + */ + va = &rmsgp->rm_body.rm_chunks[0]; + vaend = (u32 *)((unsigned long)rmsgp + rqstp->rq_arg.len); + va = decode_read_list(va, vaend); + if (!va) + return -EINVAL; + va = decode_write_list(va, vaend); + if (!va) + return -EINVAL; + va = decode_reply_array(va, vaend); + if (!va) + return -EINVAL; + + rqstp->rq_arg.head[0].iov_base = va; + hdr_len = (unsigned long)va - (unsigned long)rmsgp; + rqstp->rq_arg.head[0].iov_len -= hdr_len; + + *rdma_req = rmsgp; + return hdr_len; +} + +int svc_rdma_xdr_decode_deferred_req(struct svc_rqst *rqstp) +{ + struct rpcrdma_msg *rmsgp = NULL; + struct rpcrdma_read_chunk *ch; + struct rpcrdma_write_array *ary; + u32 *va; + u32 hdrlen; + + dprintk("svcrdma: processing deferred RDMA header on rqstp=%p\n", + rqstp); + rmsgp = (struct rpcrdma_msg *)rqstp->rq_arg.head[0].iov_base; + + /* Pull in the extra for the padded case and bump our pointer */ + if (rmsgp->rm_type == RDMA_MSGP) { + va = &rmsgp->rm_body.rm_padded.rm_pempty[4]; + rqstp->rq_arg.head[0].iov_base = va; + hdrlen = (u32)((unsigned long)va - (unsigned long)rmsgp); + rqstp->rq_arg.head[0].iov_len -= hdrlen; + return hdrlen; + } + + /* + * Skip all chunks to find RPC msg. These were previously processed + */ + va = &rmsgp->rm_body.rm_chunks[0]; + + /* Skip read-list */ + for (ch = (struct rpcrdma_read_chunk *)va; + ch->rc_discrim != xdr_zero; ch++); + va = (u32 *)&ch->rc_position; + + /* Skip write-list */ + ary = (struct rpcrdma_write_array *)va; + if (ary->wc_discrim == xdr_zero) + va = (u32 *)&ary->wc_nchunks; + else + /* + * rs_length is the 2nd 4B field in wc_target and taking its + * address skips the list terminator + */ + va = (u32 *)&ary->wc_array[ary->wc_nchunks].wc_target.rs_length; + + /* Skip reply-array */ + ary = (struct rpcrdma_write_array *)va; + if (ary->wc_discrim == xdr_zero) + va = (u32 *)&ary->wc_nchunks; + else + va = (u32 *)&ary->wc_array[ary->wc_nchunks]; + + rqstp->rq_arg.head[0].iov_base = va; + hdrlen = (unsigned long)va - (unsigned long)rmsgp; + rqstp->rq_arg.head[0].iov_len -= hdrlen; + + return hdrlen; +} + +int svc_rdma_xdr_encode_error(struct svcxprt_rdma *xprt, + struct rpcrdma_msg *rmsgp, + enum rpcrdma_errcode err, u32 *va) +{ + u32 *startp = va; + + *va++ = htonl(rmsgp->rm_xid); + *va++ = htonl(rmsgp->rm_vers); + *va++ = htonl(xprt->sc_max_requests); + *va++ = htonl(RDMA_ERROR); + *va++ = htonl(err); + if (err == ERR_VERS) { + *va++ = htonl(RPCRDMA_VERSION); + *va++ = htonl(RPCRDMA_VERSION); + } + + return (int)((unsigned long)va - (unsigned long)startp); +} + +int svc_rdma_xdr_get_reply_hdr_len(struct rpcrdma_msg *rmsgp) +{ + struct rpcrdma_write_array *wr_ary; + + /* There is no read-list in a reply */ + + /* skip write list */ + wr_ary = (struct rpcrdma_write_array *) + &rmsgp->rm_body.rm_chunks[1]; + if (wr_ary->wc_discrim) + wr_ary = (struct rpcrdma_write_array *) + &wr_ary->wc_array[ntohl(wr_ary->wc_nchunks)]. + wc_target.rs_length; + else + wr_ary = (struct rpcrdma_write_array *) + &wr_ary->wc_nchunks; + + /* skip reply array */ + if (wr_ary->wc_discrim) + wr_ary = (struct rpcrdma_write_array *) + &wr_ary->wc_array[ntohl(wr_ary->wc_nchunks)]; + else + wr_ary = (struct rpcrdma_write_array *) + &wr_ary->wc_nchunks; + + return (unsigned long) wr_ary - (unsigned long) rmsgp; +} + +void svc_rdma_xdr_encode_write_list(struct rpcrdma_msg *rmsgp, int chunks) +{ + struct rpcrdma_write_array *ary; + + /* no read-list */ + rmsgp->rm_body.rm_chunks[0] = xdr_zero; + + /* write-array discrim */ + ary = (struct rpcrdma_write_array *) + &rmsgp->rm_body.rm_chunks[1]; + ary->wc_discrim = xdr_one; + ary->wc_nchunks = htonl(chunks); + + /* write-list terminator */ + ary->wc_array[chunks].wc_target.rs_handle = xdr_zero; + + /* reply-array discriminator */ + ary->wc_array[chunks].wc_target.rs_length = xdr_zero; +} + +void svc_rdma_xdr_encode_reply_array(struct rpcrdma_write_array *ary, + int chunks) +{ + ary->wc_discrim = xdr_one; + ary->wc_nchunks = htonl(chunks); +} + +void svc_rdma_xdr_encode_array_chunk(struct rpcrdma_write_array *ary, + int chunk_no, + __be32 rs_handle, + __be64 rs_offset, + u32 write_len) +{ + struct rpcrdma_segment *seg = &ary->wc_array[chunk_no].wc_target; + seg->rs_handle = rs_handle; + seg->rs_offset = rs_offset; + seg->rs_length = htonl(write_len); +} + +void svc_rdma_xdr_encode_reply_header(struct svcxprt_rdma *xprt, + struct rpcrdma_msg *rdma_argp, + struct rpcrdma_msg *rdma_resp, + enum rpcrdma_proc rdma_type) +{ + rdma_resp->rm_xid = htonl(rdma_argp->rm_xid); + rdma_resp->rm_vers = htonl(rdma_argp->rm_vers); + rdma_resp->rm_credit = htonl(xprt->sc_max_requests); + rdma_resp->rm_type = htonl(rdma_type); + + /* Encode <nul> chunks lists */ + rdma_resp->rm_body.rm_chunks[0] = xdr_zero; + rdma_resp->rm_body.rm_chunks[1] = xdr_zero; + rdma_resp->rm_body.rm_chunks[2] = xdr_zero; +} diff --git a/net/sunrpc/xprtrdma/svc_rdma_recvfrom.c b/net/sunrpc/xprtrdma/svc_rdma_recvfrom.c new file mode 100644 index 00000000000..8f92a61ee2d --- /dev/null +++ b/net/sunrpc/xprtrdma/svc_rdma_recvfrom.c @@ -0,0 +1,586 @@ +/* + * Copyright (c) 2014 Open Grid Computing, Inc. All rights reserved. + * Copyright (c) 2005-2006 Network Appliance, Inc. All rights reserved. + * + * This software is available to you under a choice of one of two + * licenses. You may choose to be licensed under the terms of the GNU + * General Public License (GPL) Version 2, available from the file + * COPYING in the main directory of this source tree, or the BSD-type + * license below: + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * + * Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * + * Redistributions in binary form must reproduce the above + * copyright notice, this list of conditions and the following + * disclaimer in the documentation and/or other materials provided + * with the distribution. + * + * Neither the name of the Network Appliance, Inc. nor the names of + * its contributors may be used to endorse or promote products + * derived from this software without specific prior written + * permission. + * + * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS + * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT + * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR + * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT + * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, + * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT + * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE + * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + * + * Author: Tom Tucker <tom@opengridcomputing.com> + */ + +#include <linux/sunrpc/debug.h> +#include <linux/sunrpc/rpc_rdma.h> +#include <linux/spinlock.h> +#include <asm/unaligned.h> +#include <rdma/ib_verbs.h> +#include <rdma/rdma_cm.h> +#include <linux/sunrpc/svc_rdma.h> + +#define RPCDBG_FACILITY RPCDBG_SVCXPRT + +/* + * Replace the pages in the rq_argpages array with the pages from the SGE in + * the RDMA_RECV completion. The SGL should contain full pages up until the + * last one. + */ +static void rdma_build_arg_xdr(struct svc_rqst *rqstp, + struct svc_rdma_op_ctxt *ctxt, + u32 byte_count) +{ + struct page *page; + u32 bc; + int sge_no; + + /* Swap the page in the SGE with the page in argpages */ + page = ctxt->pages[0]; + put_page(rqstp->rq_pages[0]); + rqstp->rq_pages[0] = page; + + /* Set up the XDR head */ + rqstp->rq_arg.head[0].iov_base = page_address(page); + rqstp->rq_arg.head[0].iov_len = + min_t(size_t, byte_count, ctxt->sge[0].length); + rqstp->rq_arg.len = byte_count; + rqstp->rq_arg.buflen = byte_count; + + /* Compute bytes past head in the SGL */ + bc = byte_count - rqstp->rq_arg.head[0].iov_len; + + /* If data remains, store it in the pagelist */ + rqstp->rq_arg.page_len = bc; + rqstp->rq_arg.page_base = 0; + rqstp->rq_arg.pages = &rqstp->rq_pages[1]; + sge_no = 1; + while (bc && sge_no < ctxt->count) { + page = ctxt->pages[sge_no]; + put_page(rqstp->rq_pages[sge_no]); + rqstp->rq_pages[sge_no] = page; + bc -= min_t(u32, bc, ctxt->sge[sge_no].length); + rqstp->rq_arg.buflen += ctxt->sge[sge_no].length; + sge_no++; + } + rqstp->rq_respages = &rqstp->rq_pages[sge_no]; + rqstp->rq_next_page = rqstp->rq_respages + 1; + + /* We should never run out of SGE because the limit is defined to + * support the max allowed RPC data length + */ + BUG_ON(bc && (sge_no == ctxt->count)); + BUG_ON((rqstp->rq_arg.head[0].iov_len + rqstp->rq_arg.page_len) + != byte_count); + BUG_ON(rqstp->rq_arg.len != byte_count); + + /* If not all pages were used from the SGL, free the remaining ones */ + bc = sge_no; + while (sge_no < ctxt->count) { + page = ctxt->pages[sge_no++]; + put_page(page); + } + ctxt->count = bc; + + /* Set up tail */ + rqstp->rq_arg.tail[0].iov_base = NULL; + rqstp->rq_arg.tail[0].iov_len = 0; +} + +static int rdma_read_max_sge(struct svcxprt_rdma *xprt, int sge_count) +{ + if (rdma_node_get_transport(xprt->sc_cm_id->device->node_type) == + RDMA_TRANSPORT_IWARP) + return 1; + else + return min_t(int, sge_count, xprt->sc_max_sge); +} + +typedef int (*rdma_reader_fn)(struct svcxprt_rdma *xprt, + struct svc_rqst *rqstp, + struct svc_rdma_op_ctxt *head, + int *page_no, + u32 *page_offset, + u32 rs_handle, + u32 rs_length, + u64 rs_offset, + int last); + +/* Issue an RDMA_READ using the local lkey to map the data sink */ +static int rdma_read_chunk_lcl(struct svcxprt_rdma *xprt, + struct svc_rqst *rqstp, + struct svc_rdma_op_ctxt *head, + int *page_no, + u32 *page_offset, + u32 rs_handle, + u32 rs_length, + u64 rs_offset, + int last) +{ + struct ib_send_wr read_wr; + int pages_needed = PAGE_ALIGN(*page_offset + rs_length) >> PAGE_SHIFT; + struct svc_rdma_op_ctxt *ctxt = svc_rdma_get_context(xprt); + int ret, read, pno; + u32 pg_off = *page_offset; + u32 pg_no = *page_no; + + ctxt->direction = DMA_FROM_DEVICE; + ctxt->read_hdr = head; + pages_needed = + min_t(int, pages_needed, rdma_read_max_sge(xprt, pages_needed)); + read = min_t(int, pages_needed << PAGE_SHIFT, rs_length); + + for (pno = 0; pno < pages_needed; pno++) { + int len = min_t(int, rs_length, PAGE_SIZE - pg_off); + + head->arg.pages[pg_no] = rqstp->rq_arg.pages[pg_no]; + head->arg.page_len += len; + head->arg.len += len; + if (!pg_off) + head->count++; + rqstp->rq_respages = &rqstp->rq_arg.pages[pg_no+1]; + rqstp->rq_next_page = rqstp->rq_respages + 1; + ctxt->sge[pno].addr = + ib_dma_map_page(xprt->sc_cm_id->device, + head->arg.pages[pg_no], pg_off, + PAGE_SIZE - pg_off, + DMA_FROM_DEVICE); + ret = ib_dma_mapping_error(xprt->sc_cm_id->device, + ctxt->sge[pno].addr); + if (ret) + goto err; + atomic_inc(&xprt->sc_dma_used); + + /* The lkey here is either a local dma lkey or a dma_mr lkey */ + ctxt->sge[pno].lkey = xprt->sc_dma_lkey; + ctxt->sge[pno].length = len; + ctxt->count++; + + /* adjust offset and wrap to next page if needed */ + pg_off += len; + if (pg_off == PAGE_SIZE) { + pg_off = 0; + pg_no++; + } + rs_length -= len; + } + + if (last && rs_length == 0) + set_bit(RDMACTXT_F_LAST_CTXT, &ctxt->flags); + else + clear_bit(RDMACTXT_F_LAST_CTXT, &ctxt->flags); + + memset(&read_wr, 0, sizeof(read_wr)); + read_wr.wr_id = (unsigned long)ctxt; + read_wr.opcode = IB_WR_RDMA_READ; + ctxt->wr_op = read_wr.opcode; + read_wr.send_flags = IB_SEND_SIGNALED; + read_wr.wr.rdma.rkey = rs_handle; + read_wr.wr.rdma.remote_addr = rs_offset; + read_wr.sg_list = ctxt->sge; + read_wr.num_sge = pages_needed; + + ret = svc_rdma_send(xprt, &read_wr); + if (ret) { + pr_err("svcrdma: Error %d posting RDMA_READ\n", ret); + set_bit(XPT_CLOSE, &xprt->sc_xprt.xpt_flags); + goto err; + } + + /* return current location in page array */ + *page_no = pg_no; + *page_offset = pg_off; + ret = read; + atomic_inc(&rdma_stat_read); + return ret; + err: + svc_rdma_unmap_dma(ctxt); + svc_rdma_put_context(ctxt, 0); + return ret; +} + +/* Issue an RDMA_READ using an FRMR to map the data sink */ +static int rdma_read_chunk_frmr(struct svcxprt_rdma *xprt, + struct svc_rqst *rqstp, + struct svc_rdma_op_ctxt *head, + int *page_no, + u32 *page_offset, + u32 rs_handle, + u32 rs_length, + u64 rs_offset, + int last) +{ + struct ib_send_wr read_wr; + struct ib_send_wr inv_wr; + struct ib_send_wr fastreg_wr; + u8 key; + int pages_needed = PAGE_ALIGN(*page_offset + rs_length) >> PAGE_SHIFT; + struct svc_rdma_op_ctxt *ctxt = svc_rdma_get_context(xprt); + struct svc_rdma_fastreg_mr *frmr = svc_rdma_get_frmr(xprt); + int ret, read, pno; + u32 pg_off = *page_offset; + u32 pg_no = *page_no; + + if (IS_ERR(frmr)) + return -ENOMEM; + + ctxt->direction = DMA_FROM_DEVICE; + ctxt->frmr = frmr; + pages_needed = min_t(int, pages_needed, xprt->sc_frmr_pg_list_len); + read = min_t(int, pages_needed << PAGE_SHIFT, rs_length); + + frmr->kva = page_address(rqstp->rq_arg.pages[pg_no]); + frmr->direction = DMA_FROM_DEVICE; + frmr->access_flags = (IB_ACCESS_LOCAL_WRITE|IB_ACCESS_REMOTE_WRITE); + frmr->map_len = pages_needed << PAGE_SHIFT; + frmr->page_list_len = pages_needed; + + for (pno = 0; pno < pages_needed; pno++) { + int len = min_t(int, rs_length, PAGE_SIZE - pg_off); + + head->arg.pages[pg_no] = rqstp->rq_arg.pages[pg_no]; + head->arg.page_len += len; + head->arg.len += len; + if (!pg_off) + head->count++; + rqstp->rq_respages = &rqstp->rq_arg.pages[pg_no+1]; + rqstp->rq_next_page = rqstp->rq_respages + 1; + frmr->page_list->page_list[pno] = + ib_dma_map_page(xprt->sc_cm_id->device, + head->arg.pages[pg_no], 0, + PAGE_SIZE, DMA_FROM_DEVICE); + ret = ib_dma_mapping_error(xprt->sc_cm_id->device, + frmr->page_list->page_list[pno]); + if (ret) + goto err; + atomic_inc(&xprt->sc_dma_used); + + /* adjust offset and wrap to next page if needed */ + pg_off += len; + if (pg_off == PAGE_SIZE) { + pg_off = 0; + pg_no++; + } + rs_length -= len; + } + + if (last && rs_length == 0) + set_bit(RDMACTXT_F_LAST_CTXT, &ctxt->flags); + else + clear_bit(RDMACTXT_F_LAST_CTXT, &ctxt->flags); + + /* Bump the key */ + key = (u8)(frmr->mr->lkey & 0x000000FF); + ib_update_fast_reg_key(frmr->mr, ++key); + + ctxt->sge[0].addr = (unsigned long)frmr->kva + *page_offset; + ctxt->sge[0].lkey = frmr->mr->lkey; + ctxt->sge[0].length = read; + ctxt->count = 1; + ctxt->read_hdr = head; + + /* Prepare FASTREG WR */ + memset(&fastreg_wr, 0, sizeof(fastreg_wr)); + fastreg_wr.opcode = IB_WR_FAST_REG_MR; + fastreg_wr.send_flags = IB_SEND_SIGNALED; + fastreg_wr.wr.fast_reg.iova_start = (unsigned long)frmr->kva; + fastreg_wr.wr.fast_reg.page_list = frmr->page_list; + fastreg_wr.wr.fast_reg.page_list_len = frmr->page_list_len; + fastreg_wr.wr.fast_reg.page_shift = PAGE_SHIFT; + fastreg_wr.wr.fast_reg.length = frmr->map_len; + fastreg_wr.wr.fast_reg.access_flags = frmr->access_flags; + fastreg_wr.wr.fast_reg.rkey = frmr->mr->lkey; + fastreg_wr.next = &read_wr; + + /* Prepare RDMA_READ */ + memset(&read_wr, 0, sizeof(read_wr)); + read_wr.send_flags = IB_SEND_SIGNALED; + read_wr.wr.rdma.rkey = rs_handle; + read_wr.wr.rdma.remote_addr = rs_offset; + read_wr.sg_list = ctxt->sge; + read_wr.num_sge = 1; + if (xprt->sc_dev_caps & SVCRDMA_DEVCAP_READ_W_INV) { + read_wr.opcode = IB_WR_RDMA_READ_WITH_INV; + read_wr.wr_id = (unsigned long)ctxt; + read_wr.ex.invalidate_rkey = ctxt->frmr->mr->lkey; + } else { + read_wr.opcode = IB_WR_RDMA_READ; + read_wr.next = &inv_wr; + /* Prepare invalidate */ + memset(&inv_wr, 0, sizeof(inv_wr)); + inv_wr.wr_id = (unsigned long)ctxt; + inv_wr.opcode = IB_WR_LOCAL_INV; + inv_wr.send_flags = IB_SEND_SIGNALED | IB_SEND_FENCE; + inv_wr.ex.invalidate_rkey = frmr->mr->lkey; + } + ctxt->wr_op = read_wr.opcode; + + /* Post the chain */ + ret = svc_rdma_send(xprt, &fastreg_wr); + if (ret) { + pr_err("svcrdma: Error %d posting RDMA_READ\n", ret); + set_bit(XPT_CLOSE, &xprt->sc_xprt.xpt_flags); + goto err; + } + + /* return current location in page array */ + *page_no = pg_no; + *page_offset = pg_off; + ret = read; + atomic_inc(&rdma_stat_read); + return ret; + err: + svc_rdma_unmap_dma(ctxt); + svc_rdma_put_context(ctxt, 0); + svc_rdma_put_frmr(xprt, frmr); + return ret; +} + +static int rdma_read_chunks(struct svcxprt_rdma *xprt, + struct rpcrdma_msg *rmsgp, + struct svc_rqst *rqstp, + struct svc_rdma_op_ctxt *head) +{ + int page_no, ch_count, ret; + struct rpcrdma_read_chunk *ch; + u32 page_offset, byte_count; + u64 rs_offset; + rdma_reader_fn reader; + + /* If no read list is present, return 0 */ + ch = svc_rdma_get_read_chunk(rmsgp); + if (!ch) + return 0; + + svc_rdma_rcl_chunk_counts(ch, &ch_count, &byte_count); + if (ch_count > RPCSVC_MAXPAGES) + return -EINVAL; + + /* The request is completed when the RDMA_READs complete. The + * head context keeps all the pages that comprise the + * request. + */ + head->arg.head[0] = rqstp->rq_arg.head[0]; + head->arg.tail[0] = rqstp->rq_arg.tail[0]; + head->arg.pages = &head->pages[head->count]; + head->hdr_count = head->count; + head->arg.page_base = 0; + head->arg.page_len = 0; + head->arg.len = rqstp->rq_arg.len; + head->arg.buflen = rqstp->rq_arg.buflen; + + /* Use FRMR if supported */ + if (xprt->sc_dev_caps & SVCRDMA_DEVCAP_FAST_REG) + reader = rdma_read_chunk_frmr; + else + reader = rdma_read_chunk_lcl; + + page_no = 0; page_offset = 0; + for (ch = (struct rpcrdma_read_chunk *)&rmsgp->rm_body.rm_chunks[0]; + ch->rc_discrim != 0; ch++) { + + xdr_decode_hyper((__be32 *)&ch->rc_target.rs_offset, + &rs_offset); + byte_count = ntohl(ch->rc_target.rs_length); + + while (byte_count > 0) { + ret = reader(xprt, rqstp, head, + &page_no, &page_offset, + ntohl(ch->rc_target.rs_handle), + byte_count, rs_offset, + ((ch+1)->rc_discrim == 0) /* last */ + ); + if (ret < 0) + goto err; + byte_count -= ret; + rs_offset += ret; + head->arg.buflen += ret; + } + } + ret = 1; + err: + /* Detach arg pages. svc_recv will replenish them */ + for (page_no = 0; + &rqstp->rq_pages[page_no] < rqstp->rq_respages; page_no++) + rqstp->rq_pages[page_no] = NULL; + + return ret; +} + +static int rdma_read_complete(struct svc_rqst *rqstp, + struct svc_rdma_op_ctxt *head) +{ + int page_no; + int ret; + + BUG_ON(!head); + + /* Copy RPC pages */ + for (page_no = 0; page_no < head->count; page_no++) { + put_page(rqstp->rq_pages[page_no]); + rqstp->rq_pages[page_no] = head->pages[page_no]; + } + /* Point rq_arg.pages past header */ + rqstp->rq_arg.pages = &rqstp->rq_pages[head->hdr_count]; + rqstp->rq_arg.page_len = head->arg.page_len; + rqstp->rq_arg.page_base = head->arg.page_base; + + /* rq_respages starts after the last arg page */ + rqstp->rq_respages = &rqstp->rq_arg.pages[page_no]; + rqstp->rq_next_page = rqstp->rq_respages + 1; + + /* Rebuild rq_arg head and tail. */ + rqstp->rq_arg.head[0] = head->arg.head[0]; + rqstp->rq_arg.tail[0] = head->arg.tail[0]; + rqstp->rq_arg.len = head->arg.len; + rqstp->rq_arg.buflen = head->arg.buflen; + + /* Free the context */ + svc_rdma_put_context(head, 0); + + /* XXX: What should this be? */ + rqstp->rq_prot = IPPROTO_MAX; + svc_xprt_copy_addrs(rqstp, rqstp->rq_xprt); + + ret = rqstp->rq_arg.head[0].iov_len + + rqstp->rq_arg.page_len + + rqstp->rq_arg.tail[0].iov_len; + dprintk("svcrdma: deferred read ret=%d, rq_arg.len =%d, " + "rq_arg.head[0].iov_base=%p, rq_arg.head[0].iov_len = %zd\n", + ret, rqstp->rq_arg.len, rqstp->rq_arg.head[0].iov_base, + rqstp->rq_arg.head[0].iov_len); + + return ret; +} + +/* + * Set up the rqstp thread context to point to the RQ buffer. If + * necessary, pull additional data from the client with an RDMA_READ + * request. + */ +int svc_rdma_recvfrom(struct svc_rqst *rqstp) +{ + struct svc_xprt *xprt = rqstp->rq_xprt; + struct svcxprt_rdma *rdma_xprt = + container_of(xprt, struct svcxprt_rdma, sc_xprt); + struct svc_rdma_op_ctxt *ctxt = NULL; + struct rpcrdma_msg *rmsgp; + int ret = 0; + int len; + + dprintk("svcrdma: rqstp=%p\n", rqstp); + + spin_lock_bh(&rdma_xprt->sc_rq_dto_lock); + if (!list_empty(&rdma_xprt->sc_read_complete_q)) { + ctxt = list_entry(rdma_xprt->sc_read_complete_q.next, + struct svc_rdma_op_ctxt, + dto_q); + list_del_init(&ctxt->dto_q); + spin_unlock_bh(&rdma_xprt->sc_rq_dto_lock); + return rdma_read_complete(rqstp, ctxt); + } else if (!list_empty(&rdma_xprt->sc_rq_dto_q)) { + ctxt = list_entry(rdma_xprt->sc_rq_dto_q.next, + struct svc_rdma_op_ctxt, + dto_q); + list_del_init(&ctxt->dto_q); + } else { + atomic_inc(&rdma_stat_rq_starve); + clear_bit(XPT_DATA, &xprt->xpt_flags); + ctxt = NULL; + } + spin_unlock_bh(&rdma_xprt->sc_rq_dto_lock); + if (!ctxt) { + /* This is the EAGAIN path. The svc_recv routine will + * return -EAGAIN, the nfsd thread will go to call into + * svc_recv again and we shouldn't be on the active + * transport list + */ + if (test_bit(XPT_CLOSE, &xprt->xpt_flags)) + goto close_out; + + goto out; + } + dprintk("svcrdma: processing ctxt=%p on xprt=%p, rqstp=%p, status=%d\n", + ctxt, rdma_xprt, rqstp, ctxt->wc_status); + BUG_ON(ctxt->wc_status != IB_WC_SUCCESS); + atomic_inc(&rdma_stat_recv); + + /* Build up the XDR from the receive buffers. */ + rdma_build_arg_xdr(rqstp, ctxt, ctxt->byte_len); + + /* Decode the RDMA header. */ + len = svc_rdma_xdr_decode_req(&rmsgp, rqstp); + rqstp->rq_xprt_hlen = len; + + /* If the request is invalid, reply with an error */ + if (len < 0) { + if (len == -ENOSYS) + svc_rdma_send_error(rdma_xprt, rmsgp, ERR_VERS); + goto close_out; + } + + /* Read read-list data. */ + ret = rdma_read_chunks(rdma_xprt, rmsgp, rqstp, ctxt); + if (ret > 0) { + /* read-list posted, defer until data received from client. */ + goto defer; + } else if (ret < 0) { + /* Post of read-list failed, free context. */ + svc_rdma_put_context(ctxt, 1); + return 0; + } + + ret = rqstp->rq_arg.head[0].iov_len + + rqstp->rq_arg.page_len + + rqstp->rq_arg.tail[0].iov_len; + svc_rdma_put_context(ctxt, 0); + out: + dprintk("svcrdma: ret = %d, rq_arg.len =%d, " + "rq_arg.head[0].iov_base=%p, rq_arg.head[0].iov_len = %zd\n", + ret, rqstp->rq_arg.len, + rqstp->rq_arg.head[0].iov_base, + rqstp->rq_arg.head[0].iov_len); + rqstp->rq_prot = IPPROTO_MAX; + svc_xprt_copy_addrs(rqstp, xprt); + return ret; + + close_out: + if (ctxt) + svc_rdma_put_context(ctxt, 1); + dprintk("svcrdma: transport %p is closing\n", xprt); + /* + * Set the close bit and enqueue it. svc_recv will see the + * close bit and call svc_xprt_delete + */ + set_bit(XPT_CLOSE, &xprt->xpt_flags); +defer: + return 0; +} diff --git a/net/sunrpc/xprtrdma/svc_rdma_sendto.c b/net/sunrpc/xprtrdma/svc_rdma_sendto.c new file mode 100644 index 00000000000..49fd21a5c21 --- /dev/null +++ b/net/sunrpc/xprtrdma/svc_rdma_sendto.c @@ -0,0 +1,563 @@ +/* + * Copyright (c) 2014 Open Grid Computing, Inc. All rights reserved. + * Copyright (c) 2005-2006 Network Appliance, Inc. All rights reserved. + * + * This software is available to you under a choice of one of two + * licenses. You may choose to be licensed under the terms of the GNU + * General Public License (GPL) Version 2, available from the file + * COPYING in the main directory of this source tree, or the BSD-type + * license below: + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * + * Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * + * Redistributions in binary form must reproduce the above + * copyright notice, this list of conditions and the following + * disclaimer in the documentation and/or other materials provided + * with the distribution. + * + * Neither the name of the Network Appliance, Inc. nor the names of + * its contributors may be used to endorse or promote products + * derived from this software without specific prior written + * permission. + * + * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS + * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT + * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR + * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT + * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, + * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT + * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE + * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + * + * Author: Tom Tucker <tom@opengridcomputing.com> + */ + +#include <linux/sunrpc/debug.h> +#include <linux/sunrpc/rpc_rdma.h> +#include <linux/spinlock.h> +#include <asm/unaligned.h> +#include <rdma/ib_verbs.h> +#include <rdma/rdma_cm.h> +#include <linux/sunrpc/svc_rdma.h> + +#define RPCDBG_FACILITY RPCDBG_SVCXPRT + +static int map_xdr(struct svcxprt_rdma *xprt, + struct xdr_buf *xdr, + struct svc_rdma_req_map *vec) +{ + int sge_no; + u32 sge_bytes; + u32 page_bytes; + u32 page_off; + int page_no; + + BUG_ON(xdr->len != + (xdr->head[0].iov_len + xdr->page_len + xdr->tail[0].iov_len)); + + /* Skip the first sge, this is for the RPCRDMA header */ + sge_no = 1; + + /* Head SGE */ + vec->sge[sge_no].iov_base = xdr->head[0].iov_base; + vec->sge[sge_no].iov_len = xdr->head[0].iov_len; + sge_no++; + + /* pages SGE */ + page_no = 0; + page_bytes = xdr->page_len; + page_off = xdr->page_base; + while (page_bytes) { + vec->sge[sge_no].iov_base = + page_address(xdr->pages[page_no]) + page_off; + sge_bytes = min_t(u32, page_bytes, (PAGE_SIZE - page_off)); + page_bytes -= sge_bytes; + vec->sge[sge_no].iov_len = sge_bytes; + + sge_no++; + page_no++; + page_off = 0; /* reset for next time through loop */ + } + + /* Tail SGE */ + if (xdr->tail[0].iov_len) { + vec->sge[sge_no].iov_base = xdr->tail[0].iov_base; + vec->sge[sge_no].iov_len = xdr->tail[0].iov_len; + sge_no++; + } + + dprintk("svcrdma: map_xdr: sge_no %d page_no %d " + "page_base %u page_len %u head_len %zu tail_len %zu\n", + sge_no, page_no, xdr->page_base, xdr->page_len, + xdr->head[0].iov_len, xdr->tail[0].iov_len); + + vec->count = sge_no; + return 0; +} + +static dma_addr_t dma_map_xdr(struct svcxprt_rdma *xprt, + struct xdr_buf *xdr, + u32 xdr_off, size_t len, int dir) +{ + struct page *page; + dma_addr_t dma_addr; + if (xdr_off < xdr->head[0].iov_len) { + /* This offset is in the head */ + xdr_off += (unsigned long)xdr->head[0].iov_base & ~PAGE_MASK; + page = virt_to_page(xdr->head[0].iov_base); + } else { + xdr_off -= xdr->head[0].iov_len; + if (xdr_off < xdr->page_len) { + /* This offset is in the page list */ + xdr_off += xdr->page_base; + page = xdr->pages[xdr_off >> PAGE_SHIFT]; + xdr_off &= ~PAGE_MASK; + } else { + /* This offset is in the tail */ + xdr_off -= xdr->page_len; + xdr_off += (unsigned long) + xdr->tail[0].iov_base & ~PAGE_MASK; + page = virt_to_page(xdr->tail[0].iov_base); + } + } + dma_addr = ib_dma_map_page(xprt->sc_cm_id->device, page, xdr_off, + min_t(size_t, PAGE_SIZE, len), dir); + return dma_addr; +} + +/* Assumptions: + * - The specified write_len can be represented in sc_max_sge * PAGE_SIZE + */ +static int send_write(struct svcxprt_rdma *xprt, struct svc_rqst *rqstp, + u32 rmr, u64 to, + u32 xdr_off, int write_len, + struct svc_rdma_req_map *vec) +{ + struct ib_send_wr write_wr; + struct ib_sge *sge; + int xdr_sge_no; + int sge_no; + int sge_bytes; + int sge_off; + int bc; + struct svc_rdma_op_ctxt *ctxt; + + BUG_ON(vec->count > RPCSVC_MAXPAGES); + dprintk("svcrdma: RDMA_WRITE rmr=%x, to=%llx, xdr_off=%d, " + "write_len=%d, vec->sge=%p, vec->count=%lu\n", + rmr, (unsigned long long)to, xdr_off, + write_len, vec->sge, vec->count); + + ctxt = svc_rdma_get_context(xprt); + ctxt->direction = DMA_TO_DEVICE; + sge = ctxt->sge; + + /* Find the SGE associated with xdr_off */ + for (bc = xdr_off, xdr_sge_no = 1; bc && xdr_sge_no < vec->count; + xdr_sge_no++) { + if (vec->sge[xdr_sge_no].iov_len > bc) + break; + bc -= vec->sge[xdr_sge_no].iov_len; + } + + sge_off = bc; + bc = write_len; + sge_no = 0; + + /* Copy the remaining SGE */ + while (bc != 0) { + sge_bytes = min_t(size_t, + bc, vec->sge[xdr_sge_no].iov_len-sge_off); + sge[sge_no].length = sge_bytes; + sge[sge_no].addr = + dma_map_xdr(xprt, &rqstp->rq_res, xdr_off, + sge_bytes, DMA_TO_DEVICE); + xdr_off += sge_bytes; + if (ib_dma_mapping_error(xprt->sc_cm_id->device, + sge[sge_no].addr)) + goto err; + atomic_inc(&xprt->sc_dma_used); + sge[sge_no].lkey = xprt->sc_dma_lkey; + ctxt->count++; + sge_off = 0; + sge_no++; + xdr_sge_no++; + BUG_ON(xdr_sge_no > vec->count); + bc -= sge_bytes; + } + + /* Prepare WRITE WR */ + memset(&write_wr, 0, sizeof write_wr); + ctxt->wr_op = IB_WR_RDMA_WRITE; + write_wr.wr_id = (unsigned long)ctxt; + write_wr.sg_list = &sge[0]; + write_wr.num_sge = sge_no; + write_wr.opcode = IB_WR_RDMA_WRITE; + write_wr.send_flags = IB_SEND_SIGNALED; + write_wr.wr.rdma.rkey = rmr; + write_wr.wr.rdma.remote_addr = to; + + /* Post It */ + atomic_inc(&rdma_stat_write); + if (svc_rdma_send(xprt, &write_wr)) + goto err; + return 0; + err: + svc_rdma_unmap_dma(ctxt); + svc_rdma_put_context(ctxt, 0); + /* Fatal error, close transport */ + return -EIO; +} + +static int send_write_chunks(struct svcxprt_rdma *xprt, + struct rpcrdma_msg *rdma_argp, + struct rpcrdma_msg *rdma_resp, + struct svc_rqst *rqstp, + struct svc_rdma_req_map *vec) +{ + u32 xfer_len = rqstp->rq_res.page_len + rqstp->rq_res.tail[0].iov_len; + int write_len; + int max_write; + u32 xdr_off; + int chunk_off; + int chunk_no; + struct rpcrdma_write_array *arg_ary; + struct rpcrdma_write_array *res_ary; + int ret; + + arg_ary = svc_rdma_get_write_array(rdma_argp); + if (!arg_ary) + return 0; + res_ary = (struct rpcrdma_write_array *) + &rdma_resp->rm_body.rm_chunks[1]; + + max_write = xprt->sc_max_sge * PAGE_SIZE; + + /* Write chunks start at the pagelist */ + for (xdr_off = rqstp->rq_res.head[0].iov_len, chunk_no = 0; + xfer_len && chunk_no < arg_ary->wc_nchunks; + chunk_no++) { + struct rpcrdma_segment *arg_ch; + u64 rs_offset; + + arg_ch = &arg_ary->wc_array[chunk_no].wc_target; + write_len = min(xfer_len, ntohl(arg_ch->rs_length)); + + /* Prepare the response chunk given the length actually + * written */ + xdr_decode_hyper((__be32 *)&arg_ch->rs_offset, &rs_offset); + svc_rdma_xdr_encode_array_chunk(res_ary, chunk_no, + arg_ch->rs_handle, + arg_ch->rs_offset, + write_len); + chunk_off = 0; + while (write_len) { + int this_write; + this_write = min(write_len, max_write); + ret = send_write(xprt, rqstp, + ntohl(arg_ch->rs_handle), + rs_offset + chunk_off, + xdr_off, + this_write, + vec); + if (ret) { + dprintk("svcrdma: RDMA_WRITE failed, ret=%d\n", + ret); + return -EIO; + } + chunk_off += this_write; + xdr_off += this_write; + xfer_len -= this_write; + write_len -= this_write; + } + } + /* Update the req with the number of chunks actually used */ + svc_rdma_xdr_encode_write_list(rdma_resp, chunk_no); + + return rqstp->rq_res.page_len + rqstp->rq_res.tail[0].iov_len; +} + +static int send_reply_chunks(struct svcxprt_rdma *xprt, + struct rpcrdma_msg *rdma_argp, + struct rpcrdma_msg *rdma_resp, + struct svc_rqst *rqstp, + struct svc_rdma_req_map *vec) +{ + u32 xfer_len = rqstp->rq_res.len; + int write_len; + int max_write; + u32 xdr_off; + int chunk_no; + int chunk_off; + int nchunks; + struct rpcrdma_segment *ch; + struct rpcrdma_write_array *arg_ary; + struct rpcrdma_write_array *res_ary; + int ret; + + arg_ary = svc_rdma_get_reply_array(rdma_argp); + if (!arg_ary) + return 0; + /* XXX: need to fix when reply lists occur with read-list and or + * write-list */ + res_ary = (struct rpcrdma_write_array *) + &rdma_resp->rm_body.rm_chunks[2]; + + max_write = xprt->sc_max_sge * PAGE_SIZE; + + /* xdr offset starts at RPC message */ + nchunks = ntohl(arg_ary->wc_nchunks); + for (xdr_off = 0, chunk_no = 0; + xfer_len && chunk_no < nchunks; + chunk_no++) { + u64 rs_offset; + ch = &arg_ary->wc_array[chunk_no].wc_target; + write_len = min(xfer_len, htonl(ch->rs_length)); + + /* Prepare the reply chunk given the length actually + * written */ + xdr_decode_hyper((__be32 *)&ch->rs_offset, &rs_offset); + svc_rdma_xdr_encode_array_chunk(res_ary, chunk_no, + ch->rs_handle, ch->rs_offset, + write_len); + chunk_off = 0; + while (write_len) { + int this_write; + + this_write = min(write_len, max_write); + ret = send_write(xprt, rqstp, + ntohl(ch->rs_handle), + rs_offset + chunk_off, + xdr_off, + this_write, + vec); + if (ret) { + dprintk("svcrdma: RDMA_WRITE failed, ret=%d\n", + ret); + return -EIO; + } + chunk_off += this_write; + xdr_off += this_write; + xfer_len -= this_write; + write_len -= this_write; + } + } + /* Update the req with the number of chunks actually used */ + svc_rdma_xdr_encode_reply_array(res_ary, chunk_no); + + return rqstp->rq_res.len; +} + +/* This function prepares the portion of the RPCRDMA message to be + * sent in the RDMA_SEND. This function is called after data sent via + * RDMA has already been transmitted. There are three cases: + * - The RPCRDMA header, RPC header, and payload are all sent in a + * single RDMA_SEND. This is the "inline" case. + * - The RPCRDMA header and some portion of the RPC header and data + * are sent via this RDMA_SEND and another portion of the data is + * sent via RDMA. + * - The RPCRDMA header [NOMSG] is sent in this RDMA_SEND and the RPC + * header and data are all transmitted via RDMA. + * In all three cases, this function prepares the RPCRDMA header in + * sge[0], the 'type' parameter indicates the type to place in the + * RPCRDMA header, and the 'byte_count' field indicates how much of + * the XDR to include in this RDMA_SEND. NB: The offset of the payload + * to send is zero in the XDR. + */ +static int send_reply(struct svcxprt_rdma *rdma, + struct svc_rqst *rqstp, + struct page *page, + struct rpcrdma_msg *rdma_resp, + struct svc_rdma_op_ctxt *ctxt, + struct svc_rdma_req_map *vec, + int byte_count) +{ + struct ib_send_wr send_wr; + int sge_no; + int sge_bytes; + int page_no; + int pages; + int ret; + + /* Post a recv buffer to handle another request. */ + ret = svc_rdma_post_recv(rdma); + if (ret) { + printk(KERN_INFO + "svcrdma: could not post a receive buffer, err=%d." + "Closing transport %p.\n", ret, rdma); + set_bit(XPT_CLOSE, &rdma->sc_xprt.xpt_flags); + svc_rdma_put_context(ctxt, 0); + return -ENOTCONN; + } + + /* Prepare the context */ + ctxt->pages[0] = page; + ctxt->count = 1; + + /* Prepare the SGE for the RPCRDMA Header */ + ctxt->sge[0].lkey = rdma->sc_dma_lkey; + ctxt->sge[0].length = svc_rdma_xdr_get_reply_hdr_len(rdma_resp); + ctxt->sge[0].addr = + ib_dma_map_page(rdma->sc_cm_id->device, page, 0, + ctxt->sge[0].length, DMA_TO_DEVICE); + if (ib_dma_mapping_error(rdma->sc_cm_id->device, ctxt->sge[0].addr)) + goto err; + atomic_inc(&rdma->sc_dma_used); + + ctxt->direction = DMA_TO_DEVICE; + + /* Map the payload indicated by 'byte_count' */ + for (sge_no = 1; byte_count && sge_no < vec->count; sge_no++) { + int xdr_off = 0; + sge_bytes = min_t(size_t, vec->sge[sge_no].iov_len, byte_count); + byte_count -= sge_bytes; + ctxt->sge[sge_no].addr = + dma_map_xdr(rdma, &rqstp->rq_res, xdr_off, + sge_bytes, DMA_TO_DEVICE); + xdr_off += sge_bytes; + if (ib_dma_mapping_error(rdma->sc_cm_id->device, + ctxt->sge[sge_no].addr)) + goto err; + atomic_inc(&rdma->sc_dma_used); + ctxt->sge[sge_no].lkey = rdma->sc_dma_lkey; + ctxt->sge[sge_no].length = sge_bytes; + } + BUG_ON(byte_count != 0); + + /* Save all respages in the ctxt and remove them from the + * respages array. They are our pages until the I/O + * completes. + */ + pages = rqstp->rq_next_page - rqstp->rq_respages; + for (page_no = 0; page_no < pages; page_no++) { + ctxt->pages[page_no+1] = rqstp->rq_respages[page_no]; + ctxt->count++; + rqstp->rq_respages[page_no] = NULL; + /* + * If there are more pages than SGE, terminate SGE + * list so that svc_rdma_unmap_dma doesn't attempt to + * unmap garbage. + */ + if (page_no+1 >= sge_no) + ctxt->sge[page_no+1].length = 0; + } + rqstp->rq_next_page = rqstp->rq_respages + 1; + + BUG_ON(sge_no > rdma->sc_max_sge); + memset(&send_wr, 0, sizeof send_wr); + ctxt->wr_op = IB_WR_SEND; + send_wr.wr_id = (unsigned long)ctxt; + send_wr.sg_list = ctxt->sge; + send_wr.num_sge = sge_no; + send_wr.opcode = IB_WR_SEND; + send_wr.send_flags = IB_SEND_SIGNALED; + + ret = svc_rdma_send(rdma, &send_wr); + if (ret) + goto err; + + return 0; + + err: + svc_rdma_unmap_dma(ctxt); + svc_rdma_put_context(ctxt, 1); + return -EIO; +} + +void svc_rdma_prep_reply_hdr(struct svc_rqst *rqstp) +{ +} + +/* + * Return the start of an xdr buffer. + */ +static void *xdr_start(struct xdr_buf *xdr) +{ + return xdr->head[0].iov_base - + (xdr->len - + xdr->page_len - + xdr->tail[0].iov_len - + xdr->head[0].iov_len); +} + +int svc_rdma_sendto(struct svc_rqst *rqstp) +{ + struct svc_xprt *xprt = rqstp->rq_xprt; + struct svcxprt_rdma *rdma = + container_of(xprt, struct svcxprt_rdma, sc_xprt); + struct rpcrdma_msg *rdma_argp; + struct rpcrdma_msg *rdma_resp; + struct rpcrdma_write_array *reply_ary; + enum rpcrdma_proc reply_type; + int ret; + int inline_bytes; + struct page *res_page; + struct svc_rdma_op_ctxt *ctxt; + struct svc_rdma_req_map *vec; + + dprintk("svcrdma: sending response for rqstp=%p\n", rqstp); + + /* Get the RDMA request header. */ + rdma_argp = xdr_start(&rqstp->rq_arg); + + /* Build an req vec for the XDR */ + ctxt = svc_rdma_get_context(rdma); + ctxt->direction = DMA_TO_DEVICE; + vec = svc_rdma_get_req_map(); + ret = map_xdr(rdma, &rqstp->rq_res, vec); + if (ret) + goto err0; + inline_bytes = rqstp->rq_res.len; + + /* Create the RDMA response header */ + res_page = svc_rdma_get_page(); + rdma_resp = page_address(res_page); + reply_ary = svc_rdma_get_reply_array(rdma_argp); + if (reply_ary) + reply_type = RDMA_NOMSG; + else + reply_type = RDMA_MSG; + svc_rdma_xdr_encode_reply_header(rdma, rdma_argp, + rdma_resp, reply_type); + + /* Send any write-chunk data and build resp write-list */ + ret = send_write_chunks(rdma, rdma_argp, rdma_resp, + rqstp, vec); + if (ret < 0) { + printk(KERN_ERR "svcrdma: failed to send write chunks, rc=%d\n", + ret); + goto err1; + } + inline_bytes -= ret; + + /* Send any reply-list data and update resp reply-list */ + ret = send_reply_chunks(rdma, rdma_argp, rdma_resp, + rqstp, vec); + if (ret < 0) { + printk(KERN_ERR "svcrdma: failed to send reply chunks, rc=%d\n", + ret); + goto err1; + } + inline_bytes -= ret; + + ret = send_reply(rdma, rqstp, res_page, rdma_resp, ctxt, vec, + inline_bytes); + svc_rdma_put_req_map(vec); + dprintk("svcrdma: send_reply returns %d\n", ret); + return ret; + + err1: + put_page(res_page); + err0: + svc_rdma_put_req_map(vec); + svc_rdma_put_context(ctxt, 0); + return ret; +} diff --git a/net/sunrpc/xprtrdma/svc_rdma_transport.c b/net/sunrpc/xprtrdma/svc_rdma_transport.c new file mode 100644 index 00000000000..e7323fbbd34 --- /dev/null +++ b/net/sunrpc/xprtrdma/svc_rdma_transport.c @@ -0,0 +1,1369 @@ +/* + * Copyright (c) 2014 Open Grid Computing, Inc. All rights reserved. + * Copyright (c) 2005-2007 Network Appliance, Inc. All rights reserved. + * + * This software is available to you under a choice of one of two + * licenses. You may choose to be licensed under the terms of the GNU + * General Public License (GPL) Version 2, available from the file + * COPYING in the main directory of this source tree, or the BSD-type + * license below: + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * + * Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * + * Redistributions in binary form must reproduce the above + * copyright notice, this list of conditions and the following + * disclaimer in the documentation and/or other materials provided + * with the distribution. + * + * Neither the name of the Network Appliance, Inc. nor the names of + * its contributors may be used to endorse or promote products + * derived from this software without specific prior written + * permission. + * + * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS + * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT + * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR + * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT + * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, + * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT + * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE + * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + * + * Author: Tom Tucker <tom@opengridcomputing.com> + */ + +#include <linux/sunrpc/svc_xprt.h> +#include <linux/sunrpc/debug.h> +#include <linux/sunrpc/rpc_rdma.h> +#include <linux/interrupt.h> +#include <linux/sched.h> +#include <linux/slab.h> +#include <linux/spinlock.h> +#include <linux/workqueue.h> +#include <rdma/ib_verbs.h> +#include <rdma/rdma_cm.h> +#include <linux/sunrpc/svc_rdma.h> +#include <linux/export.h> +#include "xprt_rdma.h" + +#define RPCDBG_FACILITY RPCDBG_SVCXPRT + +static struct svc_xprt *svc_rdma_create(struct svc_serv *serv, + struct net *net, + struct sockaddr *sa, int salen, + int flags); +static struct svc_xprt *svc_rdma_accept(struct svc_xprt *xprt); +static void svc_rdma_release_rqst(struct svc_rqst *); +static void dto_tasklet_func(unsigned long data); +static void svc_rdma_detach(struct svc_xprt *xprt); +static void svc_rdma_free(struct svc_xprt *xprt); +static int svc_rdma_has_wspace(struct svc_xprt *xprt); +static int svc_rdma_secure_port(struct svc_rqst *); +static void rq_cq_reap(struct svcxprt_rdma *xprt); +static void sq_cq_reap(struct svcxprt_rdma *xprt); + +static DECLARE_TASKLET(dto_tasklet, dto_tasklet_func, 0UL); +static DEFINE_SPINLOCK(dto_lock); +static LIST_HEAD(dto_xprt_q); + +static struct svc_xprt_ops svc_rdma_ops = { + .xpo_create = svc_rdma_create, + .xpo_recvfrom = svc_rdma_recvfrom, + .xpo_sendto = svc_rdma_sendto, + .xpo_release_rqst = svc_rdma_release_rqst, + .xpo_detach = svc_rdma_detach, + .xpo_free = svc_rdma_free, + .xpo_prep_reply_hdr = svc_rdma_prep_reply_hdr, + .xpo_has_wspace = svc_rdma_has_wspace, + .xpo_accept = svc_rdma_accept, + .xpo_secure_port = svc_rdma_secure_port, +}; + +struct svc_xprt_class svc_rdma_class = { + .xcl_name = "rdma", + .xcl_owner = THIS_MODULE, + .xcl_ops = &svc_rdma_ops, + .xcl_max_payload = RPCSVC_MAXPAYLOAD_TCP, +}; + +struct svc_rdma_op_ctxt *svc_rdma_get_context(struct svcxprt_rdma *xprt) +{ + struct svc_rdma_op_ctxt *ctxt; + + while (1) { + ctxt = kmem_cache_alloc(svc_rdma_ctxt_cachep, GFP_KERNEL); + if (ctxt) + break; + schedule_timeout_uninterruptible(msecs_to_jiffies(500)); + } + ctxt->xprt = xprt; + INIT_LIST_HEAD(&ctxt->dto_q); + ctxt->count = 0; + ctxt->frmr = NULL; + atomic_inc(&xprt->sc_ctxt_used); + return ctxt; +} + +void svc_rdma_unmap_dma(struct svc_rdma_op_ctxt *ctxt) +{ + struct svcxprt_rdma *xprt = ctxt->xprt; + int i; + for (i = 0; i < ctxt->count && ctxt->sge[i].length; i++) { + /* + * Unmap the DMA addr in the SGE if the lkey matches + * the sc_dma_lkey, otherwise, ignore it since it is + * an FRMR lkey and will be unmapped later when the + * last WR that uses it completes. + */ + if (ctxt->sge[i].lkey == xprt->sc_dma_lkey) { + atomic_dec(&xprt->sc_dma_used); + ib_dma_unmap_page(xprt->sc_cm_id->device, + ctxt->sge[i].addr, + ctxt->sge[i].length, + ctxt->direction); + } + } +} + +void svc_rdma_put_context(struct svc_rdma_op_ctxt *ctxt, int free_pages) +{ + struct svcxprt_rdma *xprt; + int i; + + BUG_ON(!ctxt); + xprt = ctxt->xprt; + if (free_pages) + for (i = 0; i < ctxt->count; i++) + put_page(ctxt->pages[i]); + + kmem_cache_free(svc_rdma_ctxt_cachep, ctxt); + atomic_dec(&xprt->sc_ctxt_used); +} + +/* + * Temporary NFS req mappings are shared across all transport + * instances. These are short lived and should be bounded by the number + * of concurrent server threads * depth of the SQ. + */ +struct svc_rdma_req_map *svc_rdma_get_req_map(void) +{ + struct svc_rdma_req_map *map; + while (1) { + map = kmem_cache_alloc(svc_rdma_map_cachep, GFP_KERNEL); + if (map) + break; + schedule_timeout_uninterruptible(msecs_to_jiffies(500)); + } + map->count = 0; + return map; +} + +void svc_rdma_put_req_map(struct svc_rdma_req_map *map) +{ + kmem_cache_free(svc_rdma_map_cachep, map); +} + +/* ib_cq event handler */ +static void cq_event_handler(struct ib_event *event, void *context) +{ + struct svc_xprt *xprt = context; + dprintk("svcrdma: received CQ event id=%d, context=%p\n", + event->event, context); + set_bit(XPT_CLOSE, &xprt->xpt_flags); +} + +/* QP event handler */ +static void qp_event_handler(struct ib_event *event, void *context) +{ + struct svc_xprt *xprt = context; + + switch (event->event) { + /* These are considered benign events */ + case IB_EVENT_PATH_MIG: + case IB_EVENT_COMM_EST: + case IB_EVENT_SQ_DRAINED: + case IB_EVENT_QP_LAST_WQE_REACHED: + dprintk("svcrdma: QP event %d received for QP=%p\n", + event->event, event->element.qp); + break; + /* These are considered fatal events */ + case IB_EVENT_PATH_MIG_ERR: + case IB_EVENT_QP_FATAL: + case IB_EVENT_QP_REQ_ERR: + case IB_EVENT_QP_ACCESS_ERR: + case IB_EVENT_DEVICE_FATAL: + default: + dprintk("svcrdma: QP ERROR event %d received for QP=%p, " + "closing transport\n", + event->event, event->element.qp); + set_bit(XPT_CLOSE, &xprt->xpt_flags); + break; + } +} + +/* + * Data Transfer Operation Tasklet + * + * Walks a list of transports with I/O pending, removing entries as + * they are added to the server's I/O pending list. Two bits indicate + * if SQ, RQ, or both have I/O pending. The dto_lock is an irqsave + * spinlock that serializes access to the transport list with the RQ + * and SQ interrupt handlers. + */ +static void dto_tasklet_func(unsigned long data) +{ + struct svcxprt_rdma *xprt; + unsigned long flags; + + spin_lock_irqsave(&dto_lock, flags); + while (!list_empty(&dto_xprt_q)) { + xprt = list_entry(dto_xprt_q.next, + struct svcxprt_rdma, sc_dto_q); + list_del_init(&xprt->sc_dto_q); + spin_unlock_irqrestore(&dto_lock, flags); + + rq_cq_reap(xprt); + sq_cq_reap(xprt); + + svc_xprt_put(&xprt->sc_xprt); + spin_lock_irqsave(&dto_lock, flags); + } + spin_unlock_irqrestore(&dto_lock, flags); +} + +/* + * Receive Queue Completion Handler + * + * Since an RQ completion handler is called on interrupt context, we + * need to defer the handling of the I/O to a tasklet + */ +static void rq_comp_handler(struct ib_cq *cq, void *cq_context) +{ + struct svcxprt_rdma *xprt = cq_context; + unsigned long flags; + + /* Guard against unconditional flush call for destroyed QP */ + if (atomic_read(&xprt->sc_xprt.xpt_ref.refcount)==0) + return; + + /* + * Set the bit regardless of whether or not it's on the list + * because it may be on the list already due to an SQ + * completion. + */ + set_bit(RDMAXPRT_RQ_PENDING, &xprt->sc_flags); + + /* + * If this transport is not already on the DTO transport queue, + * add it + */ + spin_lock_irqsave(&dto_lock, flags); + if (list_empty(&xprt->sc_dto_q)) { + svc_xprt_get(&xprt->sc_xprt); + list_add_tail(&xprt->sc_dto_q, &dto_xprt_q); + } + spin_unlock_irqrestore(&dto_lock, flags); + + /* Tasklet does all the work to avoid irqsave locks. */ + tasklet_schedule(&dto_tasklet); +} + +/* + * rq_cq_reap - Process the RQ CQ. + * + * Take all completing WC off the CQE and enqueue the associated DTO + * context on the dto_q for the transport. + * + * Note that caller must hold a transport reference. + */ +static void rq_cq_reap(struct svcxprt_rdma *xprt) +{ + int ret; + struct ib_wc wc; + struct svc_rdma_op_ctxt *ctxt = NULL; + + if (!test_and_clear_bit(RDMAXPRT_RQ_PENDING, &xprt->sc_flags)) + return; + + ib_req_notify_cq(xprt->sc_rq_cq, IB_CQ_NEXT_COMP); + atomic_inc(&rdma_stat_rq_poll); + + while ((ret = ib_poll_cq(xprt->sc_rq_cq, 1, &wc)) > 0) { + ctxt = (struct svc_rdma_op_ctxt *)(unsigned long)wc.wr_id; + ctxt->wc_status = wc.status; + ctxt->byte_len = wc.byte_len; + svc_rdma_unmap_dma(ctxt); + if (wc.status != IB_WC_SUCCESS) { + /* Close the transport */ + dprintk("svcrdma: transport closing putting ctxt %p\n", ctxt); + set_bit(XPT_CLOSE, &xprt->sc_xprt.xpt_flags); + svc_rdma_put_context(ctxt, 1); + svc_xprt_put(&xprt->sc_xprt); + continue; + } + spin_lock_bh(&xprt->sc_rq_dto_lock); + list_add_tail(&ctxt->dto_q, &xprt->sc_rq_dto_q); + spin_unlock_bh(&xprt->sc_rq_dto_lock); + svc_xprt_put(&xprt->sc_xprt); + } + + if (ctxt) + atomic_inc(&rdma_stat_rq_prod); + + set_bit(XPT_DATA, &xprt->sc_xprt.xpt_flags); + /* + * If data arrived before established event, + * don't enqueue. This defers RPC I/O until the + * RDMA connection is complete. + */ + if (!test_bit(RDMAXPRT_CONN_PENDING, &xprt->sc_flags)) + svc_xprt_enqueue(&xprt->sc_xprt); +} + +/* + * Process a completion context + */ +static void process_context(struct svcxprt_rdma *xprt, + struct svc_rdma_op_ctxt *ctxt) +{ + svc_rdma_unmap_dma(ctxt); + + switch (ctxt->wr_op) { + case IB_WR_SEND: + BUG_ON(ctxt->frmr); + svc_rdma_put_context(ctxt, 1); + break; + + case IB_WR_RDMA_WRITE: + BUG_ON(ctxt->frmr); + svc_rdma_put_context(ctxt, 0); + break; + + case IB_WR_RDMA_READ: + case IB_WR_RDMA_READ_WITH_INV: + svc_rdma_put_frmr(xprt, ctxt->frmr); + if (test_bit(RDMACTXT_F_LAST_CTXT, &ctxt->flags)) { + struct svc_rdma_op_ctxt *read_hdr = ctxt->read_hdr; + BUG_ON(!read_hdr); + spin_lock_bh(&xprt->sc_rq_dto_lock); + set_bit(XPT_DATA, &xprt->sc_xprt.xpt_flags); + list_add_tail(&read_hdr->dto_q, + &xprt->sc_read_complete_q); + spin_unlock_bh(&xprt->sc_rq_dto_lock); + svc_xprt_enqueue(&xprt->sc_xprt); + } + svc_rdma_put_context(ctxt, 0); + break; + + default: + BUG_ON(1); + printk(KERN_ERR "svcrdma: unexpected completion type, " + "opcode=%d\n", + ctxt->wr_op); + break; + } +} + +/* + * Send Queue Completion Handler - potentially called on interrupt context. + * + * Note that caller must hold a transport reference. + */ +static void sq_cq_reap(struct svcxprt_rdma *xprt) +{ + struct svc_rdma_op_ctxt *ctxt = NULL; + struct ib_wc wc_a[6]; + struct ib_wc *wc; + struct ib_cq *cq = xprt->sc_sq_cq; + int ret; + + memset(wc_a, 0, sizeof(wc_a)); + + if (!test_and_clear_bit(RDMAXPRT_SQ_PENDING, &xprt->sc_flags)) + return; + + ib_req_notify_cq(xprt->sc_sq_cq, IB_CQ_NEXT_COMP); + atomic_inc(&rdma_stat_sq_poll); + while ((ret = ib_poll_cq(cq, ARRAY_SIZE(wc_a), wc_a)) > 0) { + int i; + + for (i = 0; i < ret; i++) { + wc = &wc_a[i]; + if (wc->status != IB_WC_SUCCESS) { + dprintk("svcrdma: sq wc err status %d\n", + wc->status); + + /* Close the transport */ + set_bit(XPT_CLOSE, &xprt->sc_xprt.xpt_flags); + } + + /* Decrement used SQ WR count */ + atomic_dec(&xprt->sc_sq_count); + wake_up(&xprt->sc_send_wait); + + ctxt = (struct svc_rdma_op_ctxt *) + (unsigned long)wc->wr_id; + if (ctxt) + process_context(xprt, ctxt); + + svc_xprt_put(&xprt->sc_xprt); + } + } + + if (ctxt) + atomic_inc(&rdma_stat_sq_prod); +} + +static void sq_comp_handler(struct ib_cq *cq, void *cq_context) +{ + struct svcxprt_rdma *xprt = cq_context; + unsigned long flags; + + /* Guard against unconditional flush call for destroyed QP */ + if (atomic_read(&xprt->sc_xprt.xpt_ref.refcount)==0) + return; + + /* + * Set the bit regardless of whether or not it's on the list + * because it may be on the list already due to an RQ + * completion. + */ + set_bit(RDMAXPRT_SQ_PENDING, &xprt->sc_flags); + + /* + * If this transport is not already on the DTO transport queue, + * add it + */ + spin_lock_irqsave(&dto_lock, flags); + if (list_empty(&xprt->sc_dto_q)) { + svc_xprt_get(&xprt->sc_xprt); + list_add_tail(&xprt->sc_dto_q, &dto_xprt_q); + } + spin_unlock_irqrestore(&dto_lock, flags); + + /* Tasklet does all the work to avoid irqsave locks. */ + tasklet_schedule(&dto_tasklet); +} + +static struct svcxprt_rdma *rdma_create_xprt(struct svc_serv *serv, + int listener) +{ + struct svcxprt_rdma *cma_xprt = kzalloc(sizeof *cma_xprt, GFP_KERNEL); + + if (!cma_xprt) + return NULL; + svc_xprt_init(&init_net, &svc_rdma_class, &cma_xprt->sc_xprt, serv); + INIT_LIST_HEAD(&cma_xprt->sc_accept_q); + INIT_LIST_HEAD(&cma_xprt->sc_dto_q); + INIT_LIST_HEAD(&cma_xprt->sc_rq_dto_q); + INIT_LIST_HEAD(&cma_xprt->sc_read_complete_q); + INIT_LIST_HEAD(&cma_xprt->sc_frmr_q); + init_waitqueue_head(&cma_xprt->sc_send_wait); + + spin_lock_init(&cma_xprt->sc_lock); + spin_lock_init(&cma_xprt->sc_rq_dto_lock); + spin_lock_init(&cma_xprt->sc_frmr_q_lock); + + cma_xprt->sc_ord = svcrdma_ord; + + cma_xprt->sc_max_req_size = svcrdma_max_req_size; + cma_xprt->sc_max_requests = svcrdma_max_requests; + cma_xprt->sc_sq_depth = svcrdma_max_requests * RPCRDMA_SQ_DEPTH_MULT; + atomic_set(&cma_xprt->sc_sq_count, 0); + atomic_set(&cma_xprt->sc_ctxt_used, 0); + + if (listener) + set_bit(XPT_LISTENER, &cma_xprt->sc_xprt.xpt_flags); + + return cma_xprt; +} + +struct page *svc_rdma_get_page(void) +{ + struct page *page; + + while ((page = alloc_page(GFP_KERNEL)) == NULL) { + /* If we can't get memory, wait a bit and try again */ + printk(KERN_INFO "svcrdma: out of memory...retrying in 1s\n"); + schedule_timeout_uninterruptible(msecs_to_jiffies(1000)); + } + return page; +} + +int svc_rdma_post_recv(struct svcxprt_rdma *xprt) +{ + struct ib_recv_wr recv_wr, *bad_recv_wr; + struct svc_rdma_op_ctxt *ctxt; + struct page *page; + dma_addr_t pa; + int sge_no; + int buflen; + int ret; + + ctxt = svc_rdma_get_context(xprt); + buflen = 0; + ctxt->direction = DMA_FROM_DEVICE; + for (sge_no = 0; buflen < xprt->sc_max_req_size; sge_no++) { + BUG_ON(sge_no >= xprt->sc_max_sge); + page = svc_rdma_get_page(); + ctxt->pages[sge_no] = page; + pa = ib_dma_map_page(xprt->sc_cm_id->device, + page, 0, PAGE_SIZE, + DMA_FROM_DEVICE); + if (ib_dma_mapping_error(xprt->sc_cm_id->device, pa)) + goto err_put_ctxt; + atomic_inc(&xprt->sc_dma_used); + ctxt->sge[sge_no].addr = pa; + ctxt->sge[sge_no].length = PAGE_SIZE; + ctxt->sge[sge_no].lkey = xprt->sc_dma_lkey; + ctxt->count = sge_no + 1; + buflen += PAGE_SIZE; + } + recv_wr.next = NULL; + recv_wr.sg_list = &ctxt->sge[0]; + recv_wr.num_sge = ctxt->count; + recv_wr.wr_id = (u64)(unsigned long)ctxt; + + svc_xprt_get(&xprt->sc_xprt); + ret = ib_post_recv(xprt->sc_qp, &recv_wr, &bad_recv_wr); + if (ret) { + svc_rdma_unmap_dma(ctxt); + svc_rdma_put_context(ctxt, 1); + svc_xprt_put(&xprt->sc_xprt); + } + return ret; + + err_put_ctxt: + svc_rdma_unmap_dma(ctxt); + svc_rdma_put_context(ctxt, 1); + return -ENOMEM; +} + +/* + * This function handles the CONNECT_REQUEST event on a listening + * endpoint. It is passed the cma_id for the _new_ connection. The context in + * this cma_id is inherited from the listening cma_id and is the svc_xprt + * structure for the listening endpoint. + * + * This function creates a new xprt for the new connection and enqueues it on + * the accept queue for the listent xprt. When the listen thread is kicked, it + * will call the recvfrom method on the listen xprt which will accept the new + * connection. + */ +static void handle_connect_req(struct rdma_cm_id *new_cma_id, size_t client_ird) +{ + struct svcxprt_rdma *listen_xprt = new_cma_id->context; + struct svcxprt_rdma *newxprt; + struct sockaddr *sa; + + /* Create a new transport */ + newxprt = rdma_create_xprt(listen_xprt->sc_xprt.xpt_server, 0); + if (!newxprt) { + dprintk("svcrdma: failed to create new transport\n"); + return; + } + newxprt->sc_cm_id = new_cma_id; + new_cma_id->context = newxprt; + dprintk("svcrdma: Creating newxprt=%p, cm_id=%p, listenxprt=%p\n", + newxprt, newxprt->sc_cm_id, listen_xprt); + + /* Save client advertised inbound read limit for use later in accept. */ + newxprt->sc_ord = client_ird; + + /* Set the local and remote addresses in the transport */ + sa = (struct sockaddr *)&newxprt->sc_cm_id->route.addr.dst_addr; + svc_xprt_set_remote(&newxprt->sc_xprt, sa, svc_addr_len(sa)); + sa = (struct sockaddr *)&newxprt->sc_cm_id->route.addr.src_addr; + svc_xprt_set_local(&newxprt->sc_xprt, sa, svc_addr_len(sa)); + + /* + * Enqueue the new transport on the accept queue of the listening + * transport + */ + spin_lock_bh(&listen_xprt->sc_lock); + list_add_tail(&newxprt->sc_accept_q, &listen_xprt->sc_accept_q); + spin_unlock_bh(&listen_xprt->sc_lock); + + set_bit(XPT_CONN, &listen_xprt->sc_xprt.xpt_flags); + svc_xprt_enqueue(&listen_xprt->sc_xprt); +} + +/* + * Handles events generated on the listening endpoint. These events will be + * either be incoming connect requests or adapter removal events. + */ +static int rdma_listen_handler(struct rdma_cm_id *cma_id, + struct rdma_cm_event *event) +{ + struct svcxprt_rdma *xprt = cma_id->context; + int ret = 0; + + switch (event->event) { + case RDMA_CM_EVENT_CONNECT_REQUEST: + dprintk("svcrdma: Connect request on cma_id=%p, xprt = %p, " + "event=%d\n", cma_id, cma_id->context, event->event); + handle_connect_req(cma_id, + event->param.conn.initiator_depth); + break; + + case RDMA_CM_EVENT_ESTABLISHED: + /* Accept complete */ + dprintk("svcrdma: Connection completed on LISTEN xprt=%p, " + "cm_id=%p\n", xprt, cma_id); + break; + + case RDMA_CM_EVENT_DEVICE_REMOVAL: + dprintk("svcrdma: Device removal xprt=%p, cm_id=%p\n", + xprt, cma_id); + if (xprt) + set_bit(XPT_CLOSE, &xprt->sc_xprt.xpt_flags); + break; + + default: + dprintk("svcrdma: Unexpected event on listening endpoint %p, " + "event=%d\n", cma_id, event->event); + break; + } + + return ret; +} + +static int rdma_cma_handler(struct rdma_cm_id *cma_id, + struct rdma_cm_event *event) +{ + struct svc_xprt *xprt = cma_id->context; + struct svcxprt_rdma *rdma = + container_of(xprt, struct svcxprt_rdma, sc_xprt); + switch (event->event) { + case RDMA_CM_EVENT_ESTABLISHED: + /* Accept complete */ + svc_xprt_get(xprt); + dprintk("svcrdma: Connection completed on DTO xprt=%p, " + "cm_id=%p\n", xprt, cma_id); + clear_bit(RDMAXPRT_CONN_PENDING, &rdma->sc_flags); + svc_xprt_enqueue(xprt); + break; + case RDMA_CM_EVENT_DISCONNECTED: + dprintk("svcrdma: Disconnect on DTO xprt=%p, cm_id=%p\n", + xprt, cma_id); + if (xprt) { + set_bit(XPT_CLOSE, &xprt->xpt_flags); + svc_xprt_enqueue(xprt); + svc_xprt_put(xprt); + } + break; + case RDMA_CM_EVENT_DEVICE_REMOVAL: + dprintk("svcrdma: Device removal cma_id=%p, xprt = %p, " + "event=%d\n", cma_id, xprt, event->event); + if (xprt) { + set_bit(XPT_CLOSE, &xprt->xpt_flags); + svc_xprt_enqueue(xprt); + } + break; + default: + dprintk("svcrdma: Unexpected event on DTO endpoint %p, " + "event=%d\n", cma_id, event->event); + break; + } + return 0; +} + +/* + * Create a listening RDMA service endpoint. + */ +static struct svc_xprt *svc_rdma_create(struct svc_serv *serv, + struct net *net, + struct sockaddr *sa, int salen, + int flags) +{ + struct rdma_cm_id *listen_id; + struct svcxprt_rdma *cma_xprt; + struct svc_xprt *xprt; + int ret; + + dprintk("svcrdma: Creating RDMA socket\n"); + if (sa->sa_family != AF_INET) { + dprintk("svcrdma: Address family %d is not supported.\n", sa->sa_family); + return ERR_PTR(-EAFNOSUPPORT); + } + cma_xprt = rdma_create_xprt(serv, 1); + if (!cma_xprt) + return ERR_PTR(-ENOMEM); + xprt = &cma_xprt->sc_xprt; + + listen_id = rdma_create_id(rdma_listen_handler, cma_xprt, RDMA_PS_TCP, + IB_QPT_RC); + if (IS_ERR(listen_id)) { + ret = PTR_ERR(listen_id); + dprintk("svcrdma: rdma_create_id failed = %d\n", ret); + goto err0; + } + + ret = rdma_bind_addr(listen_id, sa); + if (ret) { + dprintk("svcrdma: rdma_bind_addr failed = %d\n", ret); + goto err1; + } + cma_xprt->sc_cm_id = listen_id; + + ret = rdma_listen(listen_id, RPCRDMA_LISTEN_BACKLOG); + if (ret) { + dprintk("svcrdma: rdma_listen failed = %d\n", ret); + goto err1; + } + + /* + * We need to use the address from the cm_id in case the + * caller specified 0 for the port number. + */ + sa = (struct sockaddr *)&cma_xprt->sc_cm_id->route.addr.src_addr; + svc_xprt_set_local(&cma_xprt->sc_xprt, sa, salen); + + return &cma_xprt->sc_xprt; + + err1: + rdma_destroy_id(listen_id); + err0: + kfree(cma_xprt); + return ERR_PTR(ret); +} + +static struct svc_rdma_fastreg_mr *rdma_alloc_frmr(struct svcxprt_rdma *xprt) +{ + struct ib_mr *mr; + struct ib_fast_reg_page_list *pl; + struct svc_rdma_fastreg_mr *frmr; + + frmr = kmalloc(sizeof(*frmr), GFP_KERNEL); + if (!frmr) + goto err; + + mr = ib_alloc_fast_reg_mr(xprt->sc_pd, RPCSVC_MAXPAGES); + if (IS_ERR(mr)) + goto err_free_frmr; + + pl = ib_alloc_fast_reg_page_list(xprt->sc_cm_id->device, + RPCSVC_MAXPAGES); + if (IS_ERR(pl)) + goto err_free_mr; + + frmr->mr = mr; + frmr->page_list = pl; + INIT_LIST_HEAD(&frmr->frmr_list); + return frmr; + + err_free_mr: + ib_dereg_mr(mr); + err_free_frmr: + kfree(frmr); + err: + return ERR_PTR(-ENOMEM); +} + +static void rdma_dealloc_frmr_q(struct svcxprt_rdma *xprt) +{ + struct svc_rdma_fastreg_mr *frmr; + + while (!list_empty(&xprt->sc_frmr_q)) { + frmr = list_entry(xprt->sc_frmr_q.next, + struct svc_rdma_fastreg_mr, frmr_list); + list_del_init(&frmr->frmr_list); + ib_dereg_mr(frmr->mr); + ib_free_fast_reg_page_list(frmr->page_list); + kfree(frmr); + } +} + +struct svc_rdma_fastreg_mr *svc_rdma_get_frmr(struct svcxprt_rdma *rdma) +{ + struct svc_rdma_fastreg_mr *frmr = NULL; + + spin_lock_bh(&rdma->sc_frmr_q_lock); + if (!list_empty(&rdma->sc_frmr_q)) { + frmr = list_entry(rdma->sc_frmr_q.next, + struct svc_rdma_fastreg_mr, frmr_list); + list_del_init(&frmr->frmr_list); + frmr->map_len = 0; + frmr->page_list_len = 0; + } + spin_unlock_bh(&rdma->sc_frmr_q_lock); + if (frmr) + return frmr; + + return rdma_alloc_frmr(rdma); +} + +static void frmr_unmap_dma(struct svcxprt_rdma *xprt, + struct svc_rdma_fastreg_mr *frmr) +{ + int page_no; + for (page_no = 0; page_no < frmr->page_list_len; page_no++) { + dma_addr_t addr = frmr->page_list->page_list[page_no]; + if (ib_dma_mapping_error(frmr->mr->device, addr)) + continue; + atomic_dec(&xprt->sc_dma_used); + ib_dma_unmap_page(frmr->mr->device, addr, PAGE_SIZE, + frmr->direction); + } +} + +void svc_rdma_put_frmr(struct svcxprt_rdma *rdma, + struct svc_rdma_fastreg_mr *frmr) +{ + if (frmr) { + frmr_unmap_dma(rdma, frmr); + spin_lock_bh(&rdma->sc_frmr_q_lock); + BUG_ON(!list_empty(&frmr->frmr_list)); + list_add(&frmr->frmr_list, &rdma->sc_frmr_q); + spin_unlock_bh(&rdma->sc_frmr_q_lock); + } +} + +/* + * This is the xpo_recvfrom function for listening endpoints. Its + * purpose is to accept incoming connections. The CMA callback handler + * has already created a new transport and attached it to the new CMA + * ID. + * + * There is a queue of pending connections hung on the listening + * transport. This queue contains the new svc_xprt structure. This + * function takes svc_xprt structures off the accept_q and completes + * the connection. + */ +static struct svc_xprt *svc_rdma_accept(struct svc_xprt *xprt) +{ + struct svcxprt_rdma *listen_rdma; + struct svcxprt_rdma *newxprt = NULL; + struct rdma_conn_param conn_param; + struct ib_qp_init_attr qp_attr; + struct ib_device_attr devattr; + int uninitialized_var(dma_mr_acc); + int need_dma_mr; + int ret; + int i; + + listen_rdma = container_of(xprt, struct svcxprt_rdma, sc_xprt); + clear_bit(XPT_CONN, &xprt->xpt_flags); + /* Get the next entry off the accept list */ + spin_lock_bh(&listen_rdma->sc_lock); + if (!list_empty(&listen_rdma->sc_accept_q)) { + newxprt = list_entry(listen_rdma->sc_accept_q.next, + struct svcxprt_rdma, sc_accept_q); + list_del_init(&newxprt->sc_accept_q); + } + if (!list_empty(&listen_rdma->sc_accept_q)) + set_bit(XPT_CONN, &listen_rdma->sc_xprt.xpt_flags); + spin_unlock_bh(&listen_rdma->sc_lock); + if (!newxprt) + return NULL; + + dprintk("svcrdma: newxprt from accept queue = %p, cm_id=%p\n", + newxprt, newxprt->sc_cm_id); + + ret = ib_query_device(newxprt->sc_cm_id->device, &devattr); + if (ret) { + dprintk("svcrdma: could not query device attributes on " + "device %p, rc=%d\n", newxprt->sc_cm_id->device, ret); + goto errout; + } + + /* Qualify the transport resource defaults with the + * capabilities of this particular device */ + newxprt->sc_max_sge = min((size_t)devattr.max_sge, + (size_t)RPCSVC_MAXPAGES); + newxprt->sc_max_requests = min((size_t)devattr.max_qp_wr, + (size_t)svcrdma_max_requests); + newxprt->sc_sq_depth = RPCRDMA_SQ_DEPTH_MULT * newxprt->sc_max_requests; + + /* + * Limit ORD based on client limit, local device limit, and + * configured svcrdma limit. + */ + newxprt->sc_ord = min_t(size_t, devattr.max_qp_rd_atom, newxprt->sc_ord); + newxprt->sc_ord = min_t(size_t, svcrdma_ord, newxprt->sc_ord); + + newxprt->sc_pd = ib_alloc_pd(newxprt->sc_cm_id->device); + if (IS_ERR(newxprt->sc_pd)) { + dprintk("svcrdma: error creating PD for connect request\n"); + goto errout; + } + newxprt->sc_sq_cq = ib_create_cq(newxprt->sc_cm_id->device, + sq_comp_handler, + cq_event_handler, + newxprt, + newxprt->sc_sq_depth, + 0); + if (IS_ERR(newxprt->sc_sq_cq)) { + dprintk("svcrdma: error creating SQ CQ for connect request\n"); + goto errout; + } + newxprt->sc_rq_cq = ib_create_cq(newxprt->sc_cm_id->device, + rq_comp_handler, + cq_event_handler, + newxprt, + newxprt->sc_max_requests, + 0); + if (IS_ERR(newxprt->sc_rq_cq)) { + dprintk("svcrdma: error creating RQ CQ for connect request\n"); + goto errout; + } + + memset(&qp_attr, 0, sizeof qp_attr); + qp_attr.event_handler = qp_event_handler; + qp_attr.qp_context = &newxprt->sc_xprt; + qp_attr.cap.max_send_wr = newxprt->sc_sq_depth; + qp_attr.cap.max_recv_wr = newxprt->sc_max_requests; + qp_attr.cap.max_send_sge = newxprt->sc_max_sge; + qp_attr.cap.max_recv_sge = newxprt->sc_max_sge; + qp_attr.sq_sig_type = IB_SIGNAL_REQ_WR; + qp_attr.qp_type = IB_QPT_RC; + qp_attr.send_cq = newxprt->sc_sq_cq; + qp_attr.recv_cq = newxprt->sc_rq_cq; + dprintk("svcrdma: newxprt->sc_cm_id=%p, newxprt->sc_pd=%p\n" + " cm_id->device=%p, sc_pd->device=%p\n" + " cap.max_send_wr = %d\n" + " cap.max_recv_wr = %d\n" + " cap.max_send_sge = %d\n" + " cap.max_recv_sge = %d\n", + newxprt->sc_cm_id, newxprt->sc_pd, + newxprt->sc_cm_id->device, newxprt->sc_pd->device, + qp_attr.cap.max_send_wr, + qp_attr.cap.max_recv_wr, + qp_attr.cap.max_send_sge, + qp_attr.cap.max_recv_sge); + + ret = rdma_create_qp(newxprt->sc_cm_id, newxprt->sc_pd, &qp_attr); + if (ret) { + /* + * XXX: This is a hack. We need a xx_request_qp interface + * that will adjust the qp_attr's with a best-effort + * number + */ + qp_attr.cap.max_send_sge -= 2; + qp_attr.cap.max_recv_sge -= 2; + ret = rdma_create_qp(newxprt->sc_cm_id, newxprt->sc_pd, + &qp_attr); + if (ret) { + dprintk("svcrdma: failed to create QP, ret=%d\n", ret); + goto errout; + } + newxprt->sc_max_sge = qp_attr.cap.max_send_sge; + newxprt->sc_max_sge = qp_attr.cap.max_recv_sge; + newxprt->sc_sq_depth = qp_attr.cap.max_send_wr; + newxprt->sc_max_requests = qp_attr.cap.max_recv_wr; + } + newxprt->sc_qp = newxprt->sc_cm_id->qp; + + /* + * Use the most secure set of MR resources based on the + * transport type and available memory management features in + * the device. Here's the table implemented below: + * + * Fast Global DMA Remote WR + * Reg LKEY MR Access + * Sup'd Sup'd Needed Needed + * + * IWARP N N Y Y + * N Y Y Y + * Y N Y N + * Y Y N - + * + * IB N N Y N + * N Y N - + * Y N Y N + * Y Y N - + * + * NB: iWARP requires remote write access for the data sink + * of an RDMA_READ. IB does not. + */ + if (devattr.device_cap_flags & IB_DEVICE_MEM_MGT_EXTENSIONS) { + newxprt->sc_frmr_pg_list_len = + devattr.max_fast_reg_page_list_len; + newxprt->sc_dev_caps |= SVCRDMA_DEVCAP_FAST_REG; + } + + /* + * Determine if a DMA MR is required and if so, what privs are required + */ + switch (rdma_node_get_transport(newxprt->sc_cm_id->device->node_type)) { + case RDMA_TRANSPORT_IWARP: + newxprt->sc_dev_caps |= SVCRDMA_DEVCAP_READ_W_INV; + if (!(newxprt->sc_dev_caps & SVCRDMA_DEVCAP_FAST_REG)) { + need_dma_mr = 1; + dma_mr_acc = + (IB_ACCESS_LOCAL_WRITE | + IB_ACCESS_REMOTE_WRITE); + } else if (!(devattr.device_cap_flags & IB_DEVICE_LOCAL_DMA_LKEY)) { + need_dma_mr = 1; + dma_mr_acc = IB_ACCESS_LOCAL_WRITE; + } else + need_dma_mr = 0; + break; + case RDMA_TRANSPORT_IB: + if (!(newxprt->sc_dev_caps & SVCRDMA_DEVCAP_FAST_REG)) { + need_dma_mr = 1; + dma_mr_acc = IB_ACCESS_LOCAL_WRITE; + } else if (!(devattr.device_cap_flags & + IB_DEVICE_LOCAL_DMA_LKEY)) { + need_dma_mr = 1; + dma_mr_acc = IB_ACCESS_LOCAL_WRITE; + } else + need_dma_mr = 0; + break; + default: + goto errout; + } + + /* Create the DMA MR if needed, otherwise, use the DMA LKEY */ + if (need_dma_mr) { + /* Register all of physical memory */ + newxprt->sc_phys_mr = + ib_get_dma_mr(newxprt->sc_pd, dma_mr_acc); + if (IS_ERR(newxprt->sc_phys_mr)) { + dprintk("svcrdma: Failed to create DMA MR ret=%d\n", + ret); + goto errout; + } + newxprt->sc_dma_lkey = newxprt->sc_phys_mr->lkey; + } else + newxprt->sc_dma_lkey = + newxprt->sc_cm_id->device->local_dma_lkey; + + /* Post receive buffers */ + for (i = 0; i < newxprt->sc_max_requests; i++) { + ret = svc_rdma_post_recv(newxprt); + if (ret) { + dprintk("svcrdma: failure posting receive buffers\n"); + goto errout; + } + } + + /* Swap out the handler */ + newxprt->sc_cm_id->event_handler = rdma_cma_handler; + + /* + * Arm the CQs for the SQ and RQ before accepting so we can't + * miss the first message + */ + ib_req_notify_cq(newxprt->sc_sq_cq, IB_CQ_NEXT_COMP); + ib_req_notify_cq(newxprt->sc_rq_cq, IB_CQ_NEXT_COMP); + + /* Accept Connection */ + set_bit(RDMAXPRT_CONN_PENDING, &newxprt->sc_flags); + memset(&conn_param, 0, sizeof conn_param); + conn_param.responder_resources = 0; + conn_param.initiator_depth = newxprt->sc_ord; + ret = rdma_accept(newxprt->sc_cm_id, &conn_param); + if (ret) { + dprintk("svcrdma: failed to accept new connection, ret=%d\n", + ret); + goto errout; + } + + dprintk("svcrdma: new connection %p accepted with the following " + "attributes:\n" + " local_ip : %pI4\n" + " local_port : %d\n" + " remote_ip : %pI4\n" + " remote_port : %d\n" + " max_sge : %d\n" + " sq_depth : %d\n" + " max_requests : %d\n" + " ord : %d\n", + newxprt, + &((struct sockaddr_in *)&newxprt->sc_cm_id-> + route.addr.src_addr)->sin_addr.s_addr, + ntohs(((struct sockaddr_in *)&newxprt->sc_cm_id-> + route.addr.src_addr)->sin_port), + &((struct sockaddr_in *)&newxprt->sc_cm_id-> + route.addr.dst_addr)->sin_addr.s_addr, + ntohs(((struct sockaddr_in *)&newxprt->sc_cm_id-> + route.addr.dst_addr)->sin_port), + newxprt->sc_max_sge, + newxprt->sc_sq_depth, + newxprt->sc_max_requests, + newxprt->sc_ord); + + return &newxprt->sc_xprt; + + errout: + dprintk("svcrdma: failure accepting new connection rc=%d.\n", ret); + /* Take a reference in case the DTO handler runs */ + svc_xprt_get(&newxprt->sc_xprt); + if (newxprt->sc_qp && !IS_ERR(newxprt->sc_qp)) + ib_destroy_qp(newxprt->sc_qp); + rdma_destroy_id(newxprt->sc_cm_id); + /* This call to put will destroy the transport */ + svc_xprt_put(&newxprt->sc_xprt); + return NULL; +} + +static void svc_rdma_release_rqst(struct svc_rqst *rqstp) +{ +} + +/* + * When connected, an svc_xprt has at least two references: + * + * - A reference held by the cm_id between the ESTABLISHED and + * DISCONNECTED events. If the remote peer disconnected first, this + * reference could be gone. + * + * - A reference held by the svc_recv code that called this function + * as part of close processing. + * + * At a minimum one references should still be held. + */ +static void svc_rdma_detach(struct svc_xprt *xprt) +{ + struct svcxprt_rdma *rdma = + container_of(xprt, struct svcxprt_rdma, sc_xprt); + dprintk("svc: svc_rdma_detach(%p)\n", xprt); + + /* Disconnect and flush posted WQE */ + rdma_disconnect(rdma->sc_cm_id); +} + +static void __svc_rdma_free(struct work_struct *work) +{ + struct svcxprt_rdma *rdma = + container_of(work, struct svcxprt_rdma, sc_work); + dprintk("svcrdma: svc_rdma_free(%p)\n", rdma); + + /* We should only be called from kref_put */ + BUG_ON(atomic_read(&rdma->sc_xprt.xpt_ref.refcount) != 0); + + /* + * Destroy queued, but not processed read completions. Note + * that this cleanup has to be done before destroying the + * cm_id because the device ptr is needed to unmap the dma in + * svc_rdma_put_context. + */ + while (!list_empty(&rdma->sc_read_complete_q)) { + struct svc_rdma_op_ctxt *ctxt; + ctxt = list_entry(rdma->sc_read_complete_q.next, + struct svc_rdma_op_ctxt, + dto_q); + list_del_init(&ctxt->dto_q); + svc_rdma_put_context(ctxt, 1); + } + + /* Destroy queued, but not processed recv completions */ + while (!list_empty(&rdma->sc_rq_dto_q)) { + struct svc_rdma_op_ctxt *ctxt; + ctxt = list_entry(rdma->sc_rq_dto_q.next, + struct svc_rdma_op_ctxt, + dto_q); + list_del_init(&ctxt->dto_q); + svc_rdma_put_context(ctxt, 1); + } + + /* Warn if we leaked a resource or under-referenced */ + WARN_ON(atomic_read(&rdma->sc_ctxt_used) != 0); + WARN_ON(atomic_read(&rdma->sc_dma_used) != 0); + + /* De-allocate fastreg mr */ + rdma_dealloc_frmr_q(rdma); + + /* Destroy the QP if present (not a listener) */ + if (rdma->sc_qp && !IS_ERR(rdma->sc_qp)) + ib_destroy_qp(rdma->sc_qp); + + if (rdma->sc_sq_cq && !IS_ERR(rdma->sc_sq_cq)) + ib_destroy_cq(rdma->sc_sq_cq); + + if (rdma->sc_rq_cq && !IS_ERR(rdma->sc_rq_cq)) + ib_destroy_cq(rdma->sc_rq_cq); + + if (rdma->sc_phys_mr && !IS_ERR(rdma->sc_phys_mr)) + ib_dereg_mr(rdma->sc_phys_mr); + + if (rdma->sc_pd && !IS_ERR(rdma->sc_pd)) + ib_dealloc_pd(rdma->sc_pd); + + /* Destroy the CM ID */ + rdma_destroy_id(rdma->sc_cm_id); + + kfree(rdma); +} + +static void svc_rdma_free(struct svc_xprt *xprt) +{ + struct svcxprt_rdma *rdma = + container_of(xprt, struct svcxprt_rdma, sc_xprt); + INIT_WORK(&rdma->sc_work, __svc_rdma_free); + queue_work(svc_rdma_wq, &rdma->sc_work); +} + +static int svc_rdma_has_wspace(struct svc_xprt *xprt) +{ + struct svcxprt_rdma *rdma = + container_of(xprt, struct svcxprt_rdma, sc_xprt); + + /* + * If there are already waiters on the SQ, + * return false. + */ + if (waitqueue_active(&rdma->sc_send_wait)) + return 0; + + /* Otherwise return true. */ + return 1; +} + +static int svc_rdma_secure_port(struct svc_rqst *rqstp) +{ + return 1; +} + +/* + * Attempt to register the kvec representing the RPC memory with the + * device. + * + * Returns: + * NULL : The device does not support fastreg or there were no more + * fastreg mr. + * frmr : The kvec register request was successfully posted. + * <0 : An error was encountered attempting to register the kvec. + */ +int svc_rdma_fastreg(struct svcxprt_rdma *xprt, + struct svc_rdma_fastreg_mr *frmr) +{ + struct ib_send_wr fastreg_wr; + u8 key; + + /* Bump the key */ + key = (u8)(frmr->mr->lkey & 0x000000FF); + ib_update_fast_reg_key(frmr->mr, ++key); + + /* Prepare FASTREG WR */ + memset(&fastreg_wr, 0, sizeof fastreg_wr); + fastreg_wr.opcode = IB_WR_FAST_REG_MR; + fastreg_wr.send_flags = IB_SEND_SIGNALED; + fastreg_wr.wr.fast_reg.iova_start = (unsigned long)frmr->kva; + fastreg_wr.wr.fast_reg.page_list = frmr->page_list; + fastreg_wr.wr.fast_reg.page_list_len = frmr->page_list_len; + fastreg_wr.wr.fast_reg.page_shift = PAGE_SHIFT; + fastreg_wr.wr.fast_reg.length = frmr->map_len; + fastreg_wr.wr.fast_reg.access_flags = frmr->access_flags; + fastreg_wr.wr.fast_reg.rkey = frmr->mr->lkey; + return svc_rdma_send(xprt, &fastreg_wr); +} + +int svc_rdma_send(struct svcxprt_rdma *xprt, struct ib_send_wr *wr) +{ + struct ib_send_wr *bad_wr, *n_wr; + int wr_count; + int i; + int ret; + + if (test_bit(XPT_CLOSE, &xprt->sc_xprt.xpt_flags)) + return -ENOTCONN; + + BUG_ON(wr->send_flags != IB_SEND_SIGNALED); + wr_count = 1; + for (n_wr = wr->next; n_wr; n_wr = n_wr->next) + wr_count++; + + /* If the SQ is full, wait until an SQ entry is available */ + while (1) { + spin_lock_bh(&xprt->sc_lock); + if (xprt->sc_sq_depth < atomic_read(&xprt->sc_sq_count) + wr_count) { + spin_unlock_bh(&xprt->sc_lock); + atomic_inc(&rdma_stat_sq_starve); + + /* See if we can opportunistically reap SQ WR to make room */ + sq_cq_reap(xprt); + + /* Wait until SQ WR available if SQ still full */ + wait_event(xprt->sc_send_wait, + atomic_read(&xprt->sc_sq_count) < + xprt->sc_sq_depth); + if (test_bit(XPT_CLOSE, &xprt->sc_xprt.xpt_flags)) + return -ENOTCONN; + continue; + } + /* Take a transport ref for each WR posted */ + for (i = 0; i < wr_count; i++) + svc_xprt_get(&xprt->sc_xprt); + + /* Bump used SQ WR count and post */ + atomic_add(wr_count, &xprt->sc_sq_count); + ret = ib_post_send(xprt->sc_qp, wr, &bad_wr); + if (ret) { + set_bit(XPT_CLOSE, &xprt->sc_xprt.xpt_flags); + atomic_sub(wr_count, &xprt->sc_sq_count); + for (i = 0; i < wr_count; i ++) + svc_xprt_put(&xprt->sc_xprt); + dprintk("svcrdma: failed to post SQ WR rc=%d, " + "sc_sq_count=%d, sc_sq_depth=%d\n", + ret, atomic_read(&xprt->sc_sq_count), + xprt->sc_sq_depth); + } + spin_unlock_bh(&xprt->sc_lock); + if (ret) + wake_up(&xprt->sc_send_wait); + break; + } + return ret; +} + +void svc_rdma_send_error(struct svcxprt_rdma *xprt, struct rpcrdma_msg *rmsgp, + enum rpcrdma_errcode err) +{ + struct ib_send_wr err_wr; + struct page *p; + struct svc_rdma_op_ctxt *ctxt; + u32 *va; + int length; + int ret; + + p = svc_rdma_get_page(); + va = page_address(p); + + /* XDR encode error */ + length = svc_rdma_xdr_encode_error(xprt, rmsgp, err, va); + + ctxt = svc_rdma_get_context(xprt); + ctxt->direction = DMA_FROM_DEVICE; + ctxt->count = 1; + ctxt->pages[0] = p; + + /* Prepare SGE for local address */ + ctxt->sge[0].addr = ib_dma_map_page(xprt->sc_cm_id->device, + p, 0, length, DMA_FROM_DEVICE); + if (ib_dma_mapping_error(xprt->sc_cm_id->device, ctxt->sge[0].addr)) { + put_page(p); + svc_rdma_put_context(ctxt, 1); + return; + } + atomic_inc(&xprt->sc_dma_used); + ctxt->sge[0].lkey = xprt->sc_dma_lkey; + ctxt->sge[0].length = length; + + /* Prepare SEND WR */ + memset(&err_wr, 0, sizeof err_wr); + ctxt->wr_op = IB_WR_SEND; + err_wr.wr_id = (unsigned long)ctxt; + err_wr.sg_list = ctxt->sge; + err_wr.num_sge = 1; + err_wr.opcode = IB_WR_SEND; + err_wr.send_flags = IB_SEND_SIGNALED; + + /* Post It */ + ret = svc_rdma_send(xprt, &err_wr); + if (ret) { + dprintk("svcrdma: Error %d posting send for protocol error\n", + ret); + svc_rdma_unmap_dma(ctxt); + svc_rdma_put_context(ctxt, 1); + } +} diff --git a/net/sunrpc/xprtrdma/transport.c b/net/sunrpc/xprtrdma/transport.c new file mode 100644 index 00000000000..66f91f0d071 --- /dev/null +++ b/net/sunrpc/xprtrdma/transport.c @@ -0,0 +1,746 @@ +/* + * Copyright (c) 2003-2007 Network Appliance, Inc. All rights reserved. + * + * This software is available to you under a choice of one of two + * licenses. You may choose to be licensed under the terms of the GNU + * General Public License (GPL) Version 2, available from the file + * COPYING in the main directory of this source tree, or the BSD-type + * license below: + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * + * Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * + * Redistributions in binary form must reproduce the above + * copyright notice, this list of conditions and the following + * disclaimer in the documentation and/or other materials provided + * with the distribution. + * + * Neither the name of the Network Appliance, Inc. nor the names of + * its contributors may be used to endorse or promote products + * derived from this software without specific prior written + * permission. + * + * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS + * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT + * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR + * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT + * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, + * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT + * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE + * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +/* + * transport.c + * + * This file contains the top-level implementation of an RPC RDMA + * transport. + * + * Naming convention: functions beginning with xprt_ are part of the + * transport switch. All others are RPC RDMA internal. + */ + +#include <linux/module.h> +#include <linux/init.h> +#include <linux/slab.h> +#include <linux/seq_file.h> +#include <linux/sunrpc/addr.h> + +#include "xprt_rdma.h" + +#ifdef RPC_DEBUG +# define RPCDBG_FACILITY RPCDBG_TRANS +#endif + +MODULE_LICENSE("Dual BSD/GPL"); + +MODULE_DESCRIPTION("RPC/RDMA Transport for Linux kernel NFS"); +MODULE_AUTHOR("Network Appliance, Inc."); + +/* + * tunables + */ + +static unsigned int xprt_rdma_slot_table_entries = RPCRDMA_DEF_SLOT_TABLE; +static unsigned int xprt_rdma_max_inline_read = RPCRDMA_DEF_INLINE; +static unsigned int xprt_rdma_max_inline_write = RPCRDMA_DEF_INLINE; +static unsigned int xprt_rdma_inline_write_padding; +static unsigned int xprt_rdma_memreg_strategy = RPCRDMA_FRMR; + int xprt_rdma_pad_optimize = 0; + +#ifdef RPC_DEBUG + +static unsigned int min_slot_table_size = RPCRDMA_MIN_SLOT_TABLE; +static unsigned int max_slot_table_size = RPCRDMA_MAX_SLOT_TABLE; +static unsigned int zero; +static unsigned int max_padding = PAGE_SIZE; +static unsigned int min_memreg = RPCRDMA_BOUNCEBUFFERS; +static unsigned int max_memreg = RPCRDMA_LAST - 1; + +static struct ctl_table_header *sunrpc_table_header; + +static struct ctl_table xr_tunables_table[] = { + { + .procname = "rdma_slot_table_entries", + .data = &xprt_rdma_slot_table_entries, + .maxlen = sizeof(unsigned int), + .mode = 0644, + .proc_handler = proc_dointvec_minmax, + .extra1 = &min_slot_table_size, + .extra2 = &max_slot_table_size + }, + { + .procname = "rdma_max_inline_read", + .data = &xprt_rdma_max_inline_read, + .maxlen = sizeof(unsigned int), + .mode = 0644, + .proc_handler = proc_dointvec, + }, + { + .procname = "rdma_max_inline_write", + .data = &xprt_rdma_max_inline_write, + .maxlen = sizeof(unsigned int), + .mode = 0644, + .proc_handler = proc_dointvec, + }, + { + .procname = "rdma_inline_write_padding", + .data = &xprt_rdma_inline_write_padding, + .maxlen = sizeof(unsigned int), + .mode = 0644, + .proc_handler = proc_dointvec_minmax, + .extra1 = &zero, + .extra2 = &max_padding, + }, + { + .procname = "rdma_memreg_strategy", + .data = &xprt_rdma_memreg_strategy, + .maxlen = sizeof(unsigned int), + .mode = 0644, + .proc_handler = proc_dointvec_minmax, + .extra1 = &min_memreg, + .extra2 = &max_memreg, + }, + { + .procname = "rdma_pad_optimize", + .data = &xprt_rdma_pad_optimize, + .maxlen = sizeof(unsigned int), + .mode = 0644, + .proc_handler = proc_dointvec, + }, + { }, +}; + +static struct ctl_table sunrpc_table[] = { + { + .procname = "sunrpc", + .mode = 0555, + .child = xr_tunables_table + }, + { }, +}; + +#endif + +#define RPCRDMA_BIND_TO (60U * HZ) +#define RPCRDMA_INIT_REEST_TO (5U * HZ) +#define RPCRDMA_MAX_REEST_TO (30U * HZ) +#define RPCRDMA_IDLE_DISC_TO (5U * 60 * HZ) + +static struct rpc_xprt_ops xprt_rdma_procs; /* forward reference */ + +static void +xprt_rdma_format_addresses(struct rpc_xprt *xprt) +{ + struct sockaddr *sap = (struct sockaddr *) + &rpcx_to_rdmad(xprt).addr; + struct sockaddr_in *sin = (struct sockaddr_in *)sap; + char buf[64]; + + (void)rpc_ntop(sap, buf, sizeof(buf)); + xprt->address_strings[RPC_DISPLAY_ADDR] = kstrdup(buf, GFP_KERNEL); + + snprintf(buf, sizeof(buf), "%u", rpc_get_port(sap)); + xprt->address_strings[RPC_DISPLAY_PORT] = kstrdup(buf, GFP_KERNEL); + + xprt->address_strings[RPC_DISPLAY_PROTO] = "rdma"; + + snprintf(buf, sizeof(buf), "%08x", ntohl(sin->sin_addr.s_addr)); + xprt->address_strings[RPC_DISPLAY_HEX_ADDR] = kstrdup(buf, GFP_KERNEL); + + snprintf(buf, sizeof(buf), "%4hx", rpc_get_port(sap)); + xprt->address_strings[RPC_DISPLAY_HEX_PORT] = kstrdup(buf, GFP_KERNEL); + + /* netid */ + xprt->address_strings[RPC_DISPLAY_NETID] = "rdma"; +} + +static void +xprt_rdma_free_addresses(struct rpc_xprt *xprt) +{ + unsigned int i; + + for (i = 0; i < RPC_DISPLAY_MAX; i++) + switch (i) { + case RPC_DISPLAY_PROTO: + case RPC_DISPLAY_NETID: + continue; + default: + kfree(xprt->address_strings[i]); + } +} + +static void +xprt_rdma_connect_worker(struct work_struct *work) +{ + struct rpcrdma_xprt *r_xprt = + container_of(work, struct rpcrdma_xprt, rdma_connect.work); + struct rpc_xprt *xprt = &r_xprt->xprt; + int rc = 0; + + current->flags |= PF_FSTRANS; + xprt_clear_connected(xprt); + + dprintk("RPC: %s: %sconnect\n", __func__, + r_xprt->rx_ep.rep_connected != 0 ? "re" : ""); + rc = rpcrdma_ep_connect(&r_xprt->rx_ep, &r_xprt->rx_ia); + if (rc) + xprt_wake_pending_tasks(xprt, rc); + + dprintk("RPC: %s: exit\n", __func__); + xprt_clear_connecting(xprt); + current->flags &= ~PF_FSTRANS; +} + +/* + * xprt_rdma_destroy + * + * Destroy the xprt. + * Free all memory associated with the object, including its own. + * NOTE: none of the *destroy methods free memory for their top-level + * objects, even though they may have allocated it (they do free + * private memory). It's up to the caller to handle it. In this + * case (RDMA transport), all structure memory is inlined with the + * struct rpcrdma_xprt. + */ +static void +xprt_rdma_destroy(struct rpc_xprt *xprt) +{ + struct rpcrdma_xprt *r_xprt = rpcx_to_rdmax(xprt); + + dprintk("RPC: %s: called\n", __func__); + + cancel_delayed_work_sync(&r_xprt->rdma_connect); + + xprt_clear_connected(xprt); + + rpcrdma_buffer_destroy(&r_xprt->rx_buf); + rpcrdma_ep_destroy(&r_xprt->rx_ep, &r_xprt->rx_ia); + rpcrdma_ia_close(&r_xprt->rx_ia); + + xprt_rdma_free_addresses(xprt); + + xprt_free(xprt); + + dprintk("RPC: %s: returning\n", __func__); + + module_put(THIS_MODULE); +} + +static const struct rpc_timeout xprt_rdma_default_timeout = { + .to_initval = 60 * HZ, + .to_maxval = 60 * HZ, +}; + +/** + * xprt_setup_rdma - Set up transport to use RDMA + * + * @args: rpc transport arguments + */ +static struct rpc_xprt * +xprt_setup_rdma(struct xprt_create *args) +{ + struct rpcrdma_create_data_internal cdata; + struct rpc_xprt *xprt; + struct rpcrdma_xprt *new_xprt; + struct rpcrdma_ep *new_ep; + struct sockaddr_in *sin; + int rc; + + if (args->addrlen > sizeof(xprt->addr)) { + dprintk("RPC: %s: address too large\n", __func__); + return ERR_PTR(-EBADF); + } + + xprt = xprt_alloc(args->net, sizeof(struct rpcrdma_xprt), + xprt_rdma_slot_table_entries, + xprt_rdma_slot_table_entries); + if (xprt == NULL) { + dprintk("RPC: %s: couldn't allocate rpcrdma_xprt\n", + __func__); + return ERR_PTR(-ENOMEM); + } + + /* 60 second timeout, no retries */ + xprt->timeout = &xprt_rdma_default_timeout; + xprt->bind_timeout = RPCRDMA_BIND_TO; + xprt->reestablish_timeout = RPCRDMA_INIT_REEST_TO; + xprt->idle_timeout = RPCRDMA_IDLE_DISC_TO; + + xprt->resvport = 0; /* privileged port not needed */ + xprt->tsh_size = 0; /* RPC-RDMA handles framing */ + xprt->max_payload = RPCRDMA_MAX_DATA_SEGS * PAGE_SIZE; + xprt->ops = &xprt_rdma_procs; + + /* + * Set up RDMA-specific connect data. + */ + + /* Put server RDMA address in local cdata */ + memcpy(&cdata.addr, args->dstaddr, args->addrlen); + + /* Ensure xprt->addr holds valid server TCP (not RDMA) + * address, for any side protocols which peek at it */ + xprt->prot = IPPROTO_TCP; + xprt->addrlen = args->addrlen; + memcpy(&xprt->addr, &cdata.addr, xprt->addrlen); + + sin = (struct sockaddr_in *)&cdata.addr; + if (ntohs(sin->sin_port) != 0) + xprt_set_bound(xprt); + + dprintk("RPC: %s: %pI4:%u\n", + __func__, &sin->sin_addr.s_addr, ntohs(sin->sin_port)); + + /* Set max requests */ + cdata.max_requests = xprt->max_reqs; + + /* Set some length limits */ + cdata.rsize = RPCRDMA_MAX_SEGS * PAGE_SIZE; /* RDMA write max */ + cdata.wsize = RPCRDMA_MAX_SEGS * PAGE_SIZE; /* RDMA read max */ + + cdata.inline_wsize = xprt_rdma_max_inline_write; + if (cdata.inline_wsize > cdata.wsize) + cdata.inline_wsize = cdata.wsize; + + cdata.inline_rsize = xprt_rdma_max_inline_read; + if (cdata.inline_rsize > cdata.rsize) + cdata.inline_rsize = cdata.rsize; + + cdata.padding = xprt_rdma_inline_write_padding; + + /* + * Create new transport instance, which includes initialized + * o ia + * o endpoint + * o buffers + */ + + new_xprt = rpcx_to_rdmax(xprt); + + rc = rpcrdma_ia_open(new_xprt, (struct sockaddr *) &cdata.addr, + xprt_rdma_memreg_strategy); + if (rc) + goto out1; + + /* + * initialize and create ep + */ + new_xprt->rx_data = cdata; + new_ep = &new_xprt->rx_ep; + new_ep->rep_remote_addr = cdata.addr; + + rc = rpcrdma_ep_create(&new_xprt->rx_ep, + &new_xprt->rx_ia, &new_xprt->rx_data); + if (rc) + goto out2; + + /* + * Allocate pre-registered send and receive buffers for headers and + * any inline data. Also specify any padding which will be provided + * from a preregistered zero buffer. + */ + rc = rpcrdma_buffer_create(&new_xprt->rx_buf, new_ep, &new_xprt->rx_ia, + &new_xprt->rx_data); + if (rc) + goto out3; + + /* + * Register a callback for connection events. This is necessary because + * connection loss notification is async. We also catch connection loss + * when reaping receives. + */ + INIT_DELAYED_WORK(&new_xprt->rdma_connect, xprt_rdma_connect_worker); + new_ep->rep_func = rpcrdma_conn_func; + new_ep->rep_xprt = xprt; + + xprt_rdma_format_addresses(xprt); + + if (!try_module_get(THIS_MODULE)) + goto out4; + + return xprt; + +out4: + xprt_rdma_free_addresses(xprt); + rc = -EINVAL; +out3: + rpcrdma_ep_destroy(new_ep, &new_xprt->rx_ia); +out2: + rpcrdma_ia_close(&new_xprt->rx_ia); +out1: + xprt_free(xprt); + return ERR_PTR(rc); +} + +/* + * Close a connection, during shutdown or timeout/reconnect + */ +static void +xprt_rdma_close(struct rpc_xprt *xprt) +{ + struct rpcrdma_xprt *r_xprt = rpcx_to_rdmax(xprt); + + dprintk("RPC: %s: closing\n", __func__); + if (r_xprt->rx_ep.rep_connected > 0) + xprt->reestablish_timeout = 0; + xprt_disconnect_done(xprt); + (void) rpcrdma_ep_disconnect(&r_xprt->rx_ep, &r_xprt->rx_ia); +} + +static void +xprt_rdma_set_port(struct rpc_xprt *xprt, u16 port) +{ + struct sockaddr_in *sap; + + sap = (struct sockaddr_in *)&xprt->addr; + sap->sin_port = htons(port); + sap = (struct sockaddr_in *)&rpcx_to_rdmad(xprt).addr; + sap->sin_port = htons(port); + dprintk("RPC: %s: %u\n", __func__, port); +} + +static void +xprt_rdma_connect(struct rpc_xprt *xprt, struct rpc_task *task) +{ + struct rpcrdma_xprt *r_xprt = rpcx_to_rdmax(xprt); + + if (r_xprt->rx_ep.rep_connected != 0) { + /* Reconnect */ + schedule_delayed_work(&r_xprt->rdma_connect, + xprt->reestablish_timeout); + xprt->reestablish_timeout <<= 1; + if (xprt->reestablish_timeout > RPCRDMA_MAX_REEST_TO) + xprt->reestablish_timeout = RPCRDMA_MAX_REEST_TO; + else if (xprt->reestablish_timeout < RPCRDMA_INIT_REEST_TO) + xprt->reestablish_timeout = RPCRDMA_INIT_REEST_TO; + } else { + schedule_delayed_work(&r_xprt->rdma_connect, 0); + if (!RPC_IS_ASYNC(task)) + flush_delayed_work(&r_xprt->rdma_connect); + } +} + +/* + * The RDMA allocate/free functions need the task structure as a place + * to hide the struct rpcrdma_req, which is necessary for the actual send/recv + * sequence. For this reason, the recv buffers are attached to send + * buffers for portions of the RPC. Note that the RPC layer allocates + * both send and receive buffers in the same call. We may register + * the receive buffer portion when using reply chunks. + */ +static void * +xprt_rdma_allocate(struct rpc_task *task, size_t size) +{ + struct rpc_xprt *xprt = task->tk_rqstp->rq_xprt; + struct rpcrdma_req *req, *nreq; + + req = rpcrdma_buffer_get(&rpcx_to_rdmax(xprt)->rx_buf); + if (req == NULL) + return NULL; + + if (size > req->rl_size) { + dprintk("RPC: %s: size %zd too large for buffer[%zd]: " + "prog %d vers %d proc %d\n", + __func__, size, req->rl_size, + task->tk_client->cl_prog, task->tk_client->cl_vers, + task->tk_msg.rpc_proc->p_proc); + /* + * Outgoing length shortage. Our inline write max must have + * been configured to perform direct i/o. + * + * This is therefore a large metadata operation, and the + * allocate call was made on the maximum possible message, + * e.g. containing long filename(s) or symlink data. In + * fact, while these metadata operations *might* carry + * large outgoing payloads, they rarely *do*. However, we + * have to commit to the request here, so reallocate and + * register it now. The data path will never require this + * reallocation. + * + * If the allocation or registration fails, the RPC framework + * will (doggedly) retry. + */ + if (task->tk_flags & RPC_TASK_SWAPPER) + nreq = kmalloc(sizeof *req + size, GFP_ATOMIC); + else + nreq = kmalloc(sizeof *req + size, GFP_NOFS); + if (nreq == NULL) + goto outfail; + + if (rpcrdma_register_internal(&rpcx_to_rdmax(xprt)->rx_ia, + nreq->rl_base, size + sizeof(struct rpcrdma_req) + - offsetof(struct rpcrdma_req, rl_base), + &nreq->rl_handle, &nreq->rl_iov)) { + kfree(nreq); + goto outfail; + } + rpcx_to_rdmax(xprt)->rx_stats.hardway_register_count += size; + nreq->rl_size = size; + nreq->rl_niovs = 0; + nreq->rl_nchunks = 0; + nreq->rl_buffer = (struct rpcrdma_buffer *)req; + nreq->rl_reply = req->rl_reply; + memcpy(nreq->rl_segments, + req->rl_segments, sizeof nreq->rl_segments); + /* flag the swap with an unused field */ + nreq->rl_iov.length = 0; + req->rl_reply = NULL; + req = nreq; + } + dprintk("RPC: %s: size %zd, request 0x%p\n", __func__, size, req); + req->rl_connect_cookie = 0; /* our reserved value */ + return req->rl_xdr_buf; + +outfail: + rpcrdma_buffer_put(req); + rpcx_to_rdmax(xprt)->rx_stats.failed_marshal_count++; + return NULL; +} + +/* + * This function returns all RDMA resources to the pool. + */ +static void +xprt_rdma_free(void *buffer) +{ + struct rpcrdma_req *req; + struct rpcrdma_xprt *r_xprt; + struct rpcrdma_rep *rep; + int i; + + if (buffer == NULL) + return; + + req = container_of(buffer, struct rpcrdma_req, rl_xdr_buf[0]); + if (req->rl_iov.length == 0) { /* see allocate above */ + r_xprt = container_of(((struct rpcrdma_req *) req->rl_buffer)->rl_buffer, + struct rpcrdma_xprt, rx_buf); + } else + r_xprt = container_of(req->rl_buffer, struct rpcrdma_xprt, rx_buf); + rep = req->rl_reply; + + dprintk("RPC: %s: called on 0x%p%s\n", + __func__, rep, (rep && rep->rr_func) ? " (with waiter)" : ""); + + /* + * Finish the deregistration. The process is considered + * complete when the rr_func vector becomes NULL - this + * was put in place during rpcrdma_reply_handler() - the wait + * call below will not block if the dereg is "done". If + * interrupted, our framework will clean up. + */ + for (i = 0; req->rl_nchunks;) { + --req->rl_nchunks; + i += rpcrdma_deregister_external( + &req->rl_segments[i], r_xprt); + } + + if (req->rl_iov.length == 0) { /* see allocate above */ + struct rpcrdma_req *oreq = (struct rpcrdma_req *)req->rl_buffer; + oreq->rl_reply = req->rl_reply; + (void) rpcrdma_deregister_internal(&r_xprt->rx_ia, + req->rl_handle, + &req->rl_iov); + kfree(req); + req = oreq; + } + + /* Put back request+reply buffers */ + rpcrdma_buffer_put(req); +} + +/* + * send_request invokes the meat of RPC RDMA. It must do the following: + * 1. Marshal the RPC request into an RPC RDMA request, which means + * putting a header in front of data, and creating IOVs for RDMA + * from those in the request. + * 2. In marshaling, detect opportunities for RDMA, and use them. + * 3. Post a recv message to set up asynch completion, then send + * the request (rpcrdma_ep_post). + * 4. No partial sends are possible in the RPC-RDMA protocol (as in UDP). + */ + +static int +xprt_rdma_send_request(struct rpc_task *task) +{ + struct rpc_rqst *rqst = task->tk_rqstp; + struct rpc_xprt *xprt = rqst->rq_xprt; + struct rpcrdma_req *req = rpcr_to_rdmar(rqst); + struct rpcrdma_xprt *r_xprt = rpcx_to_rdmax(xprt); + int rc; + + if (req->rl_niovs == 0) { + rc = rpcrdma_marshal_req(rqst); + if (rc < 0) + goto failed_marshal; + } + + if (req->rl_reply == NULL) /* e.g. reconnection */ + rpcrdma_recv_buffer_get(req); + + if (req->rl_reply) { + req->rl_reply->rr_func = rpcrdma_reply_handler; + /* this need only be done once, but... */ + req->rl_reply->rr_xprt = xprt; + } + + /* Must suppress retransmit to maintain credits */ + if (req->rl_connect_cookie == xprt->connect_cookie) + goto drop_connection; + req->rl_connect_cookie = xprt->connect_cookie; + + if (rpcrdma_ep_post(&r_xprt->rx_ia, &r_xprt->rx_ep, req)) + goto drop_connection; + + rqst->rq_xmit_bytes_sent += rqst->rq_snd_buf.len; + rqst->rq_bytes_sent = 0; + return 0; + +failed_marshal: + r_xprt->rx_stats.failed_marshal_count++; + dprintk("RPC: %s: rpcrdma_marshal_req failed, status %i\n", + __func__, rc); + if (rc == -EIO) + return -EIO; +drop_connection: + xprt_disconnect_done(xprt); + return -ENOTCONN; /* implies disconnect */ +} + +static void xprt_rdma_print_stats(struct rpc_xprt *xprt, struct seq_file *seq) +{ + struct rpcrdma_xprt *r_xprt = rpcx_to_rdmax(xprt); + long idle_time = 0; + + if (xprt_connected(xprt)) + idle_time = (long)(jiffies - xprt->last_used) / HZ; + + seq_printf(seq, + "\txprt:\trdma %u %lu %lu %lu %ld %lu %lu %lu %Lu %Lu " + "%lu %lu %lu %Lu %Lu %Lu %Lu %lu %lu %lu\n", + + 0, /* need a local port? */ + xprt->stat.bind_count, + xprt->stat.connect_count, + xprt->stat.connect_time, + idle_time, + xprt->stat.sends, + xprt->stat.recvs, + xprt->stat.bad_xids, + xprt->stat.req_u, + xprt->stat.bklog_u, + + r_xprt->rx_stats.read_chunk_count, + r_xprt->rx_stats.write_chunk_count, + r_xprt->rx_stats.reply_chunk_count, + r_xprt->rx_stats.total_rdma_request, + r_xprt->rx_stats.total_rdma_reply, + r_xprt->rx_stats.pullup_copy_count, + r_xprt->rx_stats.fixup_copy_count, + r_xprt->rx_stats.hardway_register_count, + r_xprt->rx_stats.failed_marshal_count, + r_xprt->rx_stats.bad_reply_count); +} + +/* + * Plumbing for rpc transport switch and kernel module + */ + +static struct rpc_xprt_ops xprt_rdma_procs = { + .reserve_xprt = xprt_reserve_xprt_cong, + .release_xprt = xprt_release_xprt_cong, /* sunrpc/xprt.c */ + .alloc_slot = xprt_alloc_slot, + .release_request = xprt_release_rqst_cong, /* ditto */ + .set_retrans_timeout = xprt_set_retrans_timeout_def, /* ditto */ + .rpcbind = rpcb_getport_async, /* sunrpc/rpcb_clnt.c */ + .set_port = xprt_rdma_set_port, + .connect = xprt_rdma_connect, + .buf_alloc = xprt_rdma_allocate, + .buf_free = xprt_rdma_free, + .send_request = xprt_rdma_send_request, + .close = xprt_rdma_close, + .destroy = xprt_rdma_destroy, + .print_stats = xprt_rdma_print_stats +}; + +static struct xprt_class xprt_rdma = { + .list = LIST_HEAD_INIT(xprt_rdma.list), + .name = "rdma", + .owner = THIS_MODULE, + .ident = XPRT_TRANSPORT_RDMA, + .setup = xprt_setup_rdma, +}; + +static void __exit xprt_rdma_cleanup(void) +{ + int rc; + + dprintk("RPCRDMA Module Removed, deregister RPC RDMA transport\n"); +#ifdef RPC_DEBUG + if (sunrpc_table_header) { + unregister_sysctl_table(sunrpc_table_header); + sunrpc_table_header = NULL; + } +#endif + rc = xprt_unregister_transport(&xprt_rdma); + if (rc) + dprintk("RPC: %s: xprt_unregister returned %i\n", + __func__, rc); +} + +static int __init xprt_rdma_init(void) +{ + int rc; + + rc = xprt_register_transport(&xprt_rdma); + + if (rc) + return rc; + + dprintk("RPCRDMA Module Init, register RPC RDMA transport\n"); + + dprintk("Defaults:\n"); + dprintk("\tSlots %d\n" + "\tMaxInlineRead %d\n\tMaxInlineWrite %d\n", + xprt_rdma_slot_table_entries, + xprt_rdma_max_inline_read, xprt_rdma_max_inline_write); + dprintk("\tPadding %d\n\tMemreg %d\n", + xprt_rdma_inline_write_padding, xprt_rdma_memreg_strategy); + +#ifdef RPC_DEBUG + if (!sunrpc_table_header) + sunrpc_table_header = register_sysctl_table(sunrpc_table); +#endif + return 0; +} + +module_init(xprt_rdma_init); +module_exit(xprt_rdma_cleanup); diff --git a/net/sunrpc/xprtrdma/verbs.c b/net/sunrpc/xprtrdma/verbs.c new file mode 100644 index 00000000000..13dbd1c389f --- /dev/null +++ b/net/sunrpc/xprtrdma/verbs.c @@ -0,0 +1,1811 @@ +/* + * Copyright (c) 2003-2007 Network Appliance, Inc. All rights reserved. + * + * This software is available to you under a choice of one of two + * licenses. You may choose to be licensed under the terms of the GNU + * General Public License (GPL) Version 2, available from the file + * COPYING in the main directory of this source tree, or the BSD-type + * license below: + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * + * Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * + * Redistributions in binary form must reproduce the above + * copyright notice, this list of conditions and the following + * disclaimer in the documentation and/or other materials provided + * with the distribution. + * + * Neither the name of the Network Appliance, Inc. nor the names of + * its contributors may be used to endorse or promote products + * derived from this software without specific prior written + * permission. + * + * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS + * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT + * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR + * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT + * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, + * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT + * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE + * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +/* + * verbs.c + * + * Encapsulates the major functions managing: + * o adapters + * o endpoints + * o connections + * o buffer memory + */ + +#include <linux/interrupt.h> +#include <linux/slab.h> +#include <asm/bitops.h> + +#include "xprt_rdma.h" + +/* + * Globals/Macros + */ + +#ifdef RPC_DEBUG +# define RPCDBG_FACILITY RPCDBG_TRANS +#endif + +/* + * internal functions + */ + +/* + * handle replies in tasklet context, using a single, global list + * rdma tasklet function -- just turn around and call the func + * for all replies on the list + */ + +static DEFINE_SPINLOCK(rpcrdma_tk_lock_g); +static LIST_HEAD(rpcrdma_tasklets_g); + +static void +rpcrdma_run_tasklet(unsigned long data) +{ + struct rpcrdma_rep *rep; + void (*func)(struct rpcrdma_rep *); + unsigned long flags; + + data = data; + spin_lock_irqsave(&rpcrdma_tk_lock_g, flags); + while (!list_empty(&rpcrdma_tasklets_g)) { + rep = list_entry(rpcrdma_tasklets_g.next, + struct rpcrdma_rep, rr_list); + list_del(&rep->rr_list); + func = rep->rr_func; + rep->rr_func = NULL; + spin_unlock_irqrestore(&rpcrdma_tk_lock_g, flags); + + if (func) + func(rep); + else + rpcrdma_recv_buffer_put(rep); + + spin_lock_irqsave(&rpcrdma_tk_lock_g, flags); + } + spin_unlock_irqrestore(&rpcrdma_tk_lock_g, flags); +} + +static DECLARE_TASKLET(rpcrdma_tasklet_g, rpcrdma_run_tasklet, 0UL); + +static inline void +rpcrdma_schedule_tasklet(struct rpcrdma_rep *rep) +{ + unsigned long flags; + + spin_lock_irqsave(&rpcrdma_tk_lock_g, flags); + list_add_tail(&rep->rr_list, &rpcrdma_tasklets_g); + spin_unlock_irqrestore(&rpcrdma_tk_lock_g, flags); + tasklet_schedule(&rpcrdma_tasklet_g); +} + +static void +rpcrdma_qp_async_error_upcall(struct ib_event *event, void *context) +{ + struct rpcrdma_ep *ep = context; + + dprintk("RPC: %s: QP error %X on device %s ep %p\n", + __func__, event->event, event->device->name, context); + if (ep->rep_connected == 1) { + ep->rep_connected = -EIO; + ep->rep_func(ep); + wake_up_all(&ep->rep_connect_wait); + } +} + +static void +rpcrdma_cq_async_error_upcall(struct ib_event *event, void *context) +{ + struct rpcrdma_ep *ep = context; + + dprintk("RPC: %s: CQ error %X on device %s ep %p\n", + __func__, event->event, event->device->name, context); + if (ep->rep_connected == 1) { + ep->rep_connected = -EIO; + ep->rep_func(ep); + wake_up_all(&ep->rep_connect_wait); + } +} + +static void +rpcrdma_sendcq_process_wc(struct ib_wc *wc) +{ + struct rpcrdma_mw *frmr = (struct rpcrdma_mw *)(unsigned long)wc->wr_id; + + dprintk("RPC: %s: frmr %p status %X opcode %d\n", + __func__, frmr, wc->status, wc->opcode); + + if (wc->wr_id == 0ULL) + return; + if (wc->status != IB_WC_SUCCESS) + return; + + if (wc->opcode == IB_WC_FAST_REG_MR) + frmr->r.frmr.state = FRMR_IS_VALID; + else if (wc->opcode == IB_WC_LOCAL_INV) + frmr->r.frmr.state = FRMR_IS_INVALID; +} + +static int +rpcrdma_sendcq_poll(struct ib_cq *cq, struct rpcrdma_ep *ep) +{ + struct ib_wc *wcs; + int budget, count, rc; + + budget = RPCRDMA_WC_BUDGET / RPCRDMA_POLLSIZE; + do { + wcs = ep->rep_send_wcs; + + rc = ib_poll_cq(cq, RPCRDMA_POLLSIZE, wcs); + if (rc <= 0) + return rc; + + count = rc; + while (count-- > 0) + rpcrdma_sendcq_process_wc(wcs++); + } while (rc == RPCRDMA_POLLSIZE && --budget); + return 0; +} + +/* + * Handle send, fast_reg_mr, and local_inv completions. + * + * Send events are typically suppressed and thus do not result + * in an upcall. Occasionally one is signaled, however. This + * prevents the provider's completion queue from wrapping and + * losing a completion. + */ +static void +rpcrdma_sendcq_upcall(struct ib_cq *cq, void *cq_context) +{ + struct rpcrdma_ep *ep = (struct rpcrdma_ep *)cq_context; + int rc; + + rc = rpcrdma_sendcq_poll(cq, ep); + if (rc) { + dprintk("RPC: %s: ib_poll_cq failed: %i\n", + __func__, rc); + return; + } + + rc = ib_req_notify_cq(cq, + IB_CQ_NEXT_COMP | IB_CQ_REPORT_MISSED_EVENTS); + if (rc == 0) + return; + if (rc < 0) { + dprintk("RPC: %s: ib_req_notify_cq failed: %i\n", + __func__, rc); + return; + } + + rpcrdma_sendcq_poll(cq, ep); +} + +static void +rpcrdma_recvcq_process_wc(struct ib_wc *wc) +{ + struct rpcrdma_rep *rep = + (struct rpcrdma_rep *)(unsigned long)wc->wr_id; + + dprintk("RPC: %s: rep %p status %X opcode %X length %u\n", + __func__, rep, wc->status, wc->opcode, wc->byte_len); + + if (wc->status != IB_WC_SUCCESS) { + rep->rr_len = ~0U; + goto out_schedule; + } + if (wc->opcode != IB_WC_RECV) + return; + + rep->rr_len = wc->byte_len; + ib_dma_sync_single_for_cpu(rdmab_to_ia(rep->rr_buffer)->ri_id->device, + rep->rr_iov.addr, rep->rr_len, DMA_FROM_DEVICE); + + if (rep->rr_len >= 16) { + struct rpcrdma_msg *p = (struct rpcrdma_msg *)rep->rr_base; + unsigned int credits = ntohl(p->rm_credit); + + if (credits == 0) + credits = 1; /* don't deadlock */ + else if (credits > rep->rr_buffer->rb_max_requests) + credits = rep->rr_buffer->rb_max_requests; + atomic_set(&rep->rr_buffer->rb_credits, credits); + } + +out_schedule: + rpcrdma_schedule_tasklet(rep); +} + +static int +rpcrdma_recvcq_poll(struct ib_cq *cq, struct rpcrdma_ep *ep) +{ + struct ib_wc *wcs; + int budget, count, rc; + + budget = RPCRDMA_WC_BUDGET / RPCRDMA_POLLSIZE; + do { + wcs = ep->rep_recv_wcs; + + rc = ib_poll_cq(cq, RPCRDMA_POLLSIZE, wcs); + if (rc <= 0) + return rc; + + count = rc; + while (count-- > 0) + rpcrdma_recvcq_process_wc(wcs++); + } while (rc == RPCRDMA_POLLSIZE && --budget); + return 0; +} + +/* + * Handle receive completions. + * + * It is reentrant but processes single events in order to maintain + * ordering of receives to keep server credits. + * + * It is the responsibility of the scheduled tasklet to return + * recv buffers to the pool. NOTE: this affects synchronization of + * connection shutdown. That is, the structures required for + * the completion of the reply handler must remain intact until + * all memory has been reclaimed. + */ +static void +rpcrdma_recvcq_upcall(struct ib_cq *cq, void *cq_context) +{ + struct rpcrdma_ep *ep = (struct rpcrdma_ep *)cq_context; + int rc; + + rc = rpcrdma_recvcq_poll(cq, ep); + if (rc) { + dprintk("RPC: %s: ib_poll_cq failed: %i\n", + __func__, rc); + return; + } + + rc = ib_req_notify_cq(cq, + IB_CQ_NEXT_COMP | IB_CQ_REPORT_MISSED_EVENTS); + if (rc == 0) + return; + if (rc < 0) { + dprintk("RPC: %s: ib_req_notify_cq failed: %i\n", + __func__, rc); + return; + } + + rpcrdma_recvcq_poll(cq, ep); +} + +#ifdef RPC_DEBUG +static const char * const conn[] = { + "address resolved", + "address error", + "route resolved", + "route error", + "connect request", + "connect response", + "connect error", + "unreachable", + "rejected", + "established", + "disconnected", + "device removal" +}; +#endif + +static int +rpcrdma_conn_upcall(struct rdma_cm_id *id, struct rdma_cm_event *event) +{ + struct rpcrdma_xprt *xprt = id->context; + struct rpcrdma_ia *ia = &xprt->rx_ia; + struct rpcrdma_ep *ep = &xprt->rx_ep; +#ifdef RPC_DEBUG + struct sockaddr_in *addr = (struct sockaddr_in *) &ep->rep_remote_addr; +#endif + struct ib_qp_attr attr; + struct ib_qp_init_attr iattr; + int connstate = 0; + + switch (event->event) { + case RDMA_CM_EVENT_ADDR_RESOLVED: + case RDMA_CM_EVENT_ROUTE_RESOLVED: + ia->ri_async_rc = 0; + complete(&ia->ri_done); + break; + case RDMA_CM_EVENT_ADDR_ERROR: + ia->ri_async_rc = -EHOSTUNREACH; + dprintk("RPC: %s: CM address resolution error, ep 0x%p\n", + __func__, ep); + complete(&ia->ri_done); + break; + case RDMA_CM_EVENT_ROUTE_ERROR: + ia->ri_async_rc = -ENETUNREACH; + dprintk("RPC: %s: CM route resolution error, ep 0x%p\n", + __func__, ep); + complete(&ia->ri_done); + break; + case RDMA_CM_EVENT_ESTABLISHED: + connstate = 1; + ib_query_qp(ia->ri_id->qp, &attr, + IB_QP_MAX_QP_RD_ATOMIC | IB_QP_MAX_DEST_RD_ATOMIC, + &iattr); + dprintk("RPC: %s: %d responder resources" + " (%d initiator)\n", + __func__, attr.max_dest_rd_atomic, attr.max_rd_atomic); + goto connected; + case RDMA_CM_EVENT_CONNECT_ERROR: + connstate = -ENOTCONN; + goto connected; + case RDMA_CM_EVENT_UNREACHABLE: + connstate = -ENETDOWN; + goto connected; + case RDMA_CM_EVENT_REJECTED: + connstate = -ECONNREFUSED; + goto connected; + case RDMA_CM_EVENT_DISCONNECTED: + connstate = -ECONNABORTED; + goto connected; + case RDMA_CM_EVENT_DEVICE_REMOVAL: + connstate = -ENODEV; +connected: + dprintk("RPC: %s: %s: %pI4:%u (ep 0x%p event 0x%x)\n", + __func__, + (event->event <= 11) ? conn[event->event] : + "unknown connection error", + &addr->sin_addr.s_addr, + ntohs(addr->sin_port), + ep, event->event); + atomic_set(&rpcx_to_rdmax(ep->rep_xprt)->rx_buf.rb_credits, 1); + dprintk("RPC: %s: %sconnected\n", + __func__, connstate > 0 ? "" : "dis"); + ep->rep_connected = connstate; + ep->rep_func(ep); + wake_up_all(&ep->rep_connect_wait); + break; + default: + dprintk("RPC: %s: unexpected CM event %d\n", + __func__, event->event); + break; + } + +#ifdef RPC_DEBUG + if (connstate == 1) { + int ird = attr.max_dest_rd_atomic; + int tird = ep->rep_remote_cma.responder_resources; + printk(KERN_INFO "rpcrdma: connection to %pI4:%u " + "on %s, memreg %d slots %d ird %d%s\n", + &addr->sin_addr.s_addr, + ntohs(addr->sin_port), + ia->ri_id->device->name, + ia->ri_memreg_strategy, + xprt->rx_buf.rb_max_requests, + ird, ird < 4 && ird < tird / 2 ? " (low!)" : ""); + } else if (connstate < 0) { + printk(KERN_INFO "rpcrdma: connection to %pI4:%u closed (%d)\n", + &addr->sin_addr.s_addr, + ntohs(addr->sin_port), + connstate); + } +#endif + + return 0; +} + +static struct rdma_cm_id * +rpcrdma_create_id(struct rpcrdma_xprt *xprt, + struct rpcrdma_ia *ia, struct sockaddr *addr) +{ + struct rdma_cm_id *id; + int rc; + + init_completion(&ia->ri_done); + + id = rdma_create_id(rpcrdma_conn_upcall, xprt, RDMA_PS_TCP, IB_QPT_RC); + if (IS_ERR(id)) { + rc = PTR_ERR(id); + dprintk("RPC: %s: rdma_create_id() failed %i\n", + __func__, rc); + return id; + } + + ia->ri_async_rc = -ETIMEDOUT; + rc = rdma_resolve_addr(id, NULL, addr, RDMA_RESOLVE_TIMEOUT); + if (rc) { + dprintk("RPC: %s: rdma_resolve_addr() failed %i\n", + __func__, rc); + goto out; + } + wait_for_completion_interruptible_timeout(&ia->ri_done, + msecs_to_jiffies(RDMA_RESOLVE_TIMEOUT) + 1); + rc = ia->ri_async_rc; + if (rc) + goto out; + + ia->ri_async_rc = -ETIMEDOUT; + rc = rdma_resolve_route(id, RDMA_RESOLVE_TIMEOUT); + if (rc) { + dprintk("RPC: %s: rdma_resolve_route() failed %i\n", + __func__, rc); + goto out; + } + wait_for_completion_interruptible_timeout(&ia->ri_done, + msecs_to_jiffies(RDMA_RESOLVE_TIMEOUT) + 1); + rc = ia->ri_async_rc; + if (rc) + goto out; + + return id; + +out: + rdma_destroy_id(id); + return ERR_PTR(rc); +} + +/* + * Drain any cq, prior to teardown. + */ +static void +rpcrdma_clean_cq(struct ib_cq *cq) +{ + struct ib_wc wc; + int count = 0; + + while (1 == ib_poll_cq(cq, 1, &wc)) + ++count; + + if (count) + dprintk("RPC: %s: flushed %d events (last 0x%x)\n", + __func__, count, wc.opcode); +} + +/* + * Exported functions. + */ + +/* + * Open and initialize an Interface Adapter. + * o initializes fields of struct rpcrdma_ia, including + * interface and provider attributes and protection zone. + */ +int +rpcrdma_ia_open(struct rpcrdma_xprt *xprt, struct sockaddr *addr, int memreg) +{ + int rc, mem_priv; + struct ib_device_attr devattr; + struct rpcrdma_ia *ia = &xprt->rx_ia; + + ia->ri_id = rpcrdma_create_id(xprt, ia, addr); + if (IS_ERR(ia->ri_id)) { + rc = PTR_ERR(ia->ri_id); + goto out1; + } + + ia->ri_pd = ib_alloc_pd(ia->ri_id->device); + if (IS_ERR(ia->ri_pd)) { + rc = PTR_ERR(ia->ri_pd); + dprintk("RPC: %s: ib_alloc_pd() failed %i\n", + __func__, rc); + goto out2; + } + + /* + * Query the device to determine if the requested memory + * registration strategy is supported. If it isn't, set the + * strategy to a globally supported model. + */ + rc = ib_query_device(ia->ri_id->device, &devattr); + if (rc) { + dprintk("RPC: %s: ib_query_device failed %d\n", + __func__, rc); + goto out2; + } + + if (devattr.device_cap_flags & IB_DEVICE_LOCAL_DMA_LKEY) { + ia->ri_have_dma_lkey = 1; + ia->ri_dma_lkey = ia->ri_id->device->local_dma_lkey; + } + + if (memreg == RPCRDMA_FRMR) { + /* Requires both frmr reg and local dma lkey */ + if ((devattr.device_cap_flags & + (IB_DEVICE_MEM_MGT_EXTENSIONS|IB_DEVICE_LOCAL_DMA_LKEY)) != + (IB_DEVICE_MEM_MGT_EXTENSIONS|IB_DEVICE_LOCAL_DMA_LKEY)) { + dprintk("RPC: %s: FRMR registration " + "not supported by HCA\n", __func__); + memreg = RPCRDMA_MTHCAFMR; + } else { + /* Mind the ia limit on FRMR page list depth */ + ia->ri_max_frmr_depth = min_t(unsigned int, + RPCRDMA_MAX_DATA_SEGS, + devattr.max_fast_reg_page_list_len); + } + } + if (memreg == RPCRDMA_MTHCAFMR) { + if (!ia->ri_id->device->alloc_fmr) { + dprintk("RPC: %s: MTHCAFMR registration " + "not supported by HCA\n", __func__); +#if RPCRDMA_PERSISTENT_REGISTRATION + memreg = RPCRDMA_ALLPHYSICAL; +#else + rc = -ENOMEM; + goto out2; +#endif + } + } + + /* + * Optionally obtain an underlying physical identity mapping in + * order to do a memory window-based bind. This base registration + * is protected from remote access - that is enabled only by binding + * for the specific bytes targeted during each RPC operation, and + * revoked after the corresponding completion similar to a storage + * adapter. + */ + switch (memreg) { + case RPCRDMA_FRMR: + break; +#if RPCRDMA_PERSISTENT_REGISTRATION + case RPCRDMA_ALLPHYSICAL: + mem_priv = IB_ACCESS_LOCAL_WRITE | + IB_ACCESS_REMOTE_WRITE | + IB_ACCESS_REMOTE_READ; + goto register_setup; +#endif + case RPCRDMA_MTHCAFMR: + if (ia->ri_have_dma_lkey) + break; + mem_priv = IB_ACCESS_LOCAL_WRITE; +#if RPCRDMA_PERSISTENT_REGISTRATION + register_setup: +#endif + ia->ri_bind_mem = ib_get_dma_mr(ia->ri_pd, mem_priv); + if (IS_ERR(ia->ri_bind_mem)) { + printk(KERN_ALERT "%s: ib_get_dma_mr for " + "phys register failed with %lX\n", + __func__, PTR_ERR(ia->ri_bind_mem)); + rc = -ENOMEM; + goto out2; + } + break; + default: + printk(KERN_ERR "RPC: Unsupported memory " + "registration mode: %d\n", memreg); + rc = -ENOMEM; + goto out2; + } + dprintk("RPC: %s: memory registration strategy is %d\n", + __func__, memreg); + + /* Else will do memory reg/dereg for each chunk */ + ia->ri_memreg_strategy = memreg; + + return 0; +out2: + rdma_destroy_id(ia->ri_id); + ia->ri_id = NULL; +out1: + return rc; +} + +/* + * Clean up/close an IA. + * o if event handles and PD have been initialized, free them. + * o close the IA + */ +void +rpcrdma_ia_close(struct rpcrdma_ia *ia) +{ + int rc; + + dprintk("RPC: %s: entering\n", __func__); + if (ia->ri_bind_mem != NULL) { + rc = ib_dereg_mr(ia->ri_bind_mem); + dprintk("RPC: %s: ib_dereg_mr returned %i\n", + __func__, rc); + } + if (ia->ri_id != NULL && !IS_ERR(ia->ri_id)) { + if (ia->ri_id->qp) + rdma_destroy_qp(ia->ri_id); + rdma_destroy_id(ia->ri_id); + ia->ri_id = NULL; + } + if (ia->ri_pd != NULL && !IS_ERR(ia->ri_pd)) { + rc = ib_dealloc_pd(ia->ri_pd); + dprintk("RPC: %s: ib_dealloc_pd returned %i\n", + __func__, rc); + } +} + +/* + * Create unconnected endpoint. + */ +int +rpcrdma_ep_create(struct rpcrdma_ep *ep, struct rpcrdma_ia *ia, + struct rpcrdma_create_data_internal *cdata) +{ + struct ib_device_attr devattr; + struct ib_cq *sendcq, *recvcq; + int rc, err; + + rc = ib_query_device(ia->ri_id->device, &devattr); + if (rc) { + dprintk("RPC: %s: ib_query_device failed %d\n", + __func__, rc); + return rc; + } + + /* check provider's send/recv wr limits */ + if (cdata->max_requests > devattr.max_qp_wr) + cdata->max_requests = devattr.max_qp_wr; + + ep->rep_attr.event_handler = rpcrdma_qp_async_error_upcall; + ep->rep_attr.qp_context = ep; + /* send_cq and recv_cq initialized below */ + ep->rep_attr.srq = NULL; + ep->rep_attr.cap.max_send_wr = cdata->max_requests; + switch (ia->ri_memreg_strategy) { + case RPCRDMA_FRMR: { + int depth = 7; + + /* Add room for frmr register and invalidate WRs. + * 1. FRMR reg WR for head + * 2. FRMR invalidate WR for head + * 3. N FRMR reg WRs for pagelist + * 4. N FRMR invalidate WRs for pagelist + * 5. FRMR reg WR for tail + * 6. FRMR invalidate WR for tail + * 7. The RDMA_SEND WR + */ + + /* Calculate N if the device max FRMR depth is smaller than + * RPCRDMA_MAX_DATA_SEGS. + */ + if (ia->ri_max_frmr_depth < RPCRDMA_MAX_DATA_SEGS) { + int delta = RPCRDMA_MAX_DATA_SEGS - + ia->ri_max_frmr_depth; + + do { + depth += 2; /* FRMR reg + invalidate */ + delta -= ia->ri_max_frmr_depth; + } while (delta > 0); + + } + ep->rep_attr.cap.max_send_wr *= depth; + if (ep->rep_attr.cap.max_send_wr > devattr.max_qp_wr) { + cdata->max_requests = devattr.max_qp_wr / depth; + if (!cdata->max_requests) + return -EINVAL; + ep->rep_attr.cap.max_send_wr = cdata->max_requests * + depth; + } + break; + } + default: + break; + } + ep->rep_attr.cap.max_recv_wr = cdata->max_requests; + ep->rep_attr.cap.max_send_sge = (cdata->padding ? 4 : 2); + ep->rep_attr.cap.max_recv_sge = 1; + ep->rep_attr.cap.max_inline_data = 0; + ep->rep_attr.sq_sig_type = IB_SIGNAL_REQ_WR; + ep->rep_attr.qp_type = IB_QPT_RC; + ep->rep_attr.port_num = ~0; + + dprintk("RPC: %s: requested max: dtos: send %d recv %d; " + "iovs: send %d recv %d\n", + __func__, + ep->rep_attr.cap.max_send_wr, + ep->rep_attr.cap.max_recv_wr, + ep->rep_attr.cap.max_send_sge, + ep->rep_attr.cap.max_recv_sge); + + /* set trigger for requesting send completion */ + ep->rep_cqinit = ep->rep_attr.cap.max_send_wr/2 - 1; + if (ep->rep_cqinit <= 2) + ep->rep_cqinit = 0; + INIT_CQCOUNT(ep); + ep->rep_ia = ia; + init_waitqueue_head(&ep->rep_connect_wait); + INIT_DELAYED_WORK(&ep->rep_connect_worker, rpcrdma_connect_worker); + + sendcq = ib_create_cq(ia->ri_id->device, rpcrdma_sendcq_upcall, + rpcrdma_cq_async_error_upcall, ep, + ep->rep_attr.cap.max_send_wr + 1, 0); + if (IS_ERR(sendcq)) { + rc = PTR_ERR(sendcq); + dprintk("RPC: %s: failed to create send CQ: %i\n", + __func__, rc); + goto out1; + } + + rc = ib_req_notify_cq(sendcq, IB_CQ_NEXT_COMP); + if (rc) { + dprintk("RPC: %s: ib_req_notify_cq failed: %i\n", + __func__, rc); + goto out2; + } + + recvcq = ib_create_cq(ia->ri_id->device, rpcrdma_recvcq_upcall, + rpcrdma_cq_async_error_upcall, ep, + ep->rep_attr.cap.max_recv_wr + 1, 0); + if (IS_ERR(recvcq)) { + rc = PTR_ERR(recvcq); + dprintk("RPC: %s: failed to create recv CQ: %i\n", + __func__, rc); + goto out2; + } + + rc = ib_req_notify_cq(recvcq, IB_CQ_NEXT_COMP); + if (rc) { + dprintk("RPC: %s: ib_req_notify_cq failed: %i\n", + __func__, rc); + ib_destroy_cq(recvcq); + goto out2; + } + + ep->rep_attr.send_cq = sendcq; + ep->rep_attr.recv_cq = recvcq; + + /* Initialize cma parameters */ + + /* RPC/RDMA does not use private data */ + ep->rep_remote_cma.private_data = NULL; + ep->rep_remote_cma.private_data_len = 0; + + /* Client offers RDMA Read but does not initiate */ + ep->rep_remote_cma.initiator_depth = 0; + if (devattr.max_qp_rd_atom > 32) /* arbitrary but <= 255 */ + ep->rep_remote_cma.responder_resources = 32; + else + ep->rep_remote_cma.responder_resources = devattr.max_qp_rd_atom; + + ep->rep_remote_cma.retry_count = 7; + ep->rep_remote_cma.flow_control = 0; + ep->rep_remote_cma.rnr_retry_count = 0; + + return 0; + +out2: + err = ib_destroy_cq(sendcq); + if (err) + dprintk("RPC: %s: ib_destroy_cq returned %i\n", + __func__, err); +out1: + return rc; +} + +/* + * rpcrdma_ep_destroy + * + * Disconnect and destroy endpoint. After this, the only + * valid operations on the ep are to free it (if dynamically + * allocated) or re-create it. + */ +void +rpcrdma_ep_destroy(struct rpcrdma_ep *ep, struct rpcrdma_ia *ia) +{ + int rc; + + dprintk("RPC: %s: entering, connected is %d\n", + __func__, ep->rep_connected); + + cancel_delayed_work_sync(&ep->rep_connect_worker); + + if (ia->ri_id->qp) { + rc = rpcrdma_ep_disconnect(ep, ia); + if (rc) + dprintk("RPC: %s: rpcrdma_ep_disconnect" + " returned %i\n", __func__, rc); + rdma_destroy_qp(ia->ri_id); + ia->ri_id->qp = NULL; + } + + /* padding - could be done in rpcrdma_buffer_destroy... */ + if (ep->rep_pad_mr) { + rpcrdma_deregister_internal(ia, ep->rep_pad_mr, &ep->rep_pad); + ep->rep_pad_mr = NULL; + } + + rpcrdma_clean_cq(ep->rep_attr.recv_cq); + rc = ib_destroy_cq(ep->rep_attr.recv_cq); + if (rc) + dprintk("RPC: %s: ib_destroy_cq returned %i\n", + __func__, rc); + + rpcrdma_clean_cq(ep->rep_attr.send_cq); + rc = ib_destroy_cq(ep->rep_attr.send_cq); + if (rc) + dprintk("RPC: %s: ib_destroy_cq returned %i\n", + __func__, rc); +} + +/* + * Connect unconnected endpoint. + */ +int +rpcrdma_ep_connect(struct rpcrdma_ep *ep, struct rpcrdma_ia *ia) +{ + struct rdma_cm_id *id; + int rc = 0; + int retry_count = 0; + + if (ep->rep_connected != 0) { + struct rpcrdma_xprt *xprt; +retry: + dprintk("RPC: %s: reconnecting...\n", __func__); + rc = rpcrdma_ep_disconnect(ep, ia); + if (rc && rc != -ENOTCONN) + dprintk("RPC: %s: rpcrdma_ep_disconnect" + " status %i\n", __func__, rc); + + rpcrdma_clean_cq(ep->rep_attr.recv_cq); + rpcrdma_clean_cq(ep->rep_attr.send_cq); + + xprt = container_of(ia, struct rpcrdma_xprt, rx_ia); + id = rpcrdma_create_id(xprt, ia, + (struct sockaddr *)&xprt->rx_data.addr); + if (IS_ERR(id)) { + rc = -EHOSTUNREACH; + goto out; + } + /* TEMP TEMP TEMP - fail if new device: + * Deregister/remarshal *all* requests! + * Close and recreate adapter, pd, etc! + * Re-determine all attributes still sane! + * More stuff I haven't thought of! + * Rrrgh! + */ + if (ia->ri_id->device != id->device) { + printk("RPC: %s: can't reconnect on " + "different device!\n", __func__); + rdma_destroy_id(id); + rc = -ENETUNREACH; + goto out; + } + /* END TEMP */ + rc = rdma_create_qp(id, ia->ri_pd, &ep->rep_attr); + if (rc) { + dprintk("RPC: %s: rdma_create_qp failed %i\n", + __func__, rc); + rdma_destroy_id(id); + rc = -ENETUNREACH; + goto out; + } + rdma_destroy_qp(ia->ri_id); + rdma_destroy_id(ia->ri_id); + ia->ri_id = id; + } else { + dprintk("RPC: %s: connecting...\n", __func__); + rc = rdma_create_qp(ia->ri_id, ia->ri_pd, &ep->rep_attr); + if (rc) { + dprintk("RPC: %s: rdma_create_qp failed %i\n", + __func__, rc); + /* do not update ep->rep_connected */ + return -ENETUNREACH; + } + } + + ep->rep_connected = 0; + + rc = rdma_connect(ia->ri_id, &ep->rep_remote_cma); + if (rc) { + dprintk("RPC: %s: rdma_connect() failed with %i\n", + __func__, rc); + goto out; + } + + wait_event_interruptible(ep->rep_connect_wait, ep->rep_connected != 0); + + /* + * Check state. A non-peer reject indicates no listener + * (ECONNREFUSED), which may be a transient state. All + * others indicate a transport condition which has already + * undergone a best-effort. + */ + if (ep->rep_connected == -ECONNREFUSED && + ++retry_count <= RDMA_CONNECT_RETRY_MAX) { + dprintk("RPC: %s: non-peer_reject, retry\n", __func__); + goto retry; + } + if (ep->rep_connected <= 0) { + /* Sometimes, the only way to reliably connect to remote + * CMs is to use same nonzero values for ORD and IRD. */ + if (retry_count++ <= RDMA_CONNECT_RETRY_MAX + 1 && + (ep->rep_remote_cma.responder_resources == 0 || + ep->rep_remote_cma.initiator_depth != + ep->rep_remote_cma.responder_resources)) { + if (ep->rep_remote_cma.responder_resources == 0) + ep->rep_remote_cma.responder_resources = 1; + ep->rep_remote_cma.initiator_depth = + ep->rep_remote_cma.responder_resources; + goto retry; + } + rc = ep->rep_connected; + } else { + dprintk("RPC: %s: connected\n", __func__); + } + +out: + if (rc) + ep->rep_connected = rc; + return rc; +} + +/* + * rpcrdma_ep_disconnect + * + * This is separate from destroy to facilitate the ability + * to reconnect without recreating the endpoint. + * + * This call is not reentrant, and must not be made in parallel + * on the same endpoint. + */ +int +rpcrdma_ep_disconnect(struct rpcrdma_ep *ep, struct rpcrdma_ia *ia) +{ + int rc; + + rpcrdma_clean_cq(ep->rep_attr.recv_cq); + rpcrdma_clean_cq(ep->rep_attr.send_cq); + rc = rdma_disconnect(ia->ri_id); + if (!rc) { + /* returns without wait if not connected */ + wait_event_interruptible(ep->rep_connect_wait, + ep->rep_connected != 1); + dprintk("RPC: %s: after wait, %sconnected\n", __func__, + (ep->rep_connected == 1) ? "still " : "dis"); + } else { + dprintk("RPC: %s: rdma_disconnect %i\n", __func__, rc); + ep->rep_connected = rc; + } + return rc; +} + +/* + * Initialize buffer memory + */ +int +rpcrdma_buffer_create(struct rpcrdma_buffer *buf, struct rpcrdma_ep *ep, + struct rpcrdma_ia *ia, struct rpcrdma_create_data_internal *cdata) +{ + char *p; + size_t len, rlen, wlen; + int i, rc; + struct rpcrdma_mw *r; + + buf->rb_max_requests = cdata->max_requests; + spin_lock_init(&buf->rb_lock); + atomic_set(&buf->rb_credits, 1); + + /* Need to allocate: + * 1. arrays for send and recv pointers + * 2. arrays of struct rpcrdma_req to fill in pointers + * 3. array of struct rpcrdma_rep for replies + * 4. padding, if any + * 5. mw's, fmr's or frmr's, if any + * Send/recv buffers in req/rep need to be registered + */ + + len = buf->rb_max_requests * + (sizeof(struct rpcrdma_req *) + sizeof(struct rpcrdma_rep *)); + len += cdata->padding; + switch (ia->ri_memreg_strategy) { + case RPCRDMA_FRMR: + len += buf->rb_max_requests * RPCRDMA_MAX_SEGS * + sizeof(struct rpcrdma_mw); + break; + case RPCRDMA_MTHCAFMR: + /* TBD we are perhaps overallocating here */ + len += (buf->rb_max_requests + 1) * RPCRDMA_MAX_SEGS * + sizeof(struct rpcrdma_mw); + break; + default: + break; + } + + /* allocate 1, 4 and 5 in one shot */ + p = kzalloc(len, GFP_KERNEL); + if (p == NULL) { + dprintk("RPC: %s: req_t/rep_t/pad kzalloc(%zd) failed\n", + __func__, len); + rc = -ENOMEM; + goto out; + } + buf->rb_pool = p; /* for freeing it later */ + + buf->rb_send_bufs = (struct rpcrdma_req **) p; + p = (char *) &buf->rb_send_bufs[buf->rb_max_requests]; + buf->rb_recv_bufs = (struct rpcrdma_rep **) p; + p = (char *) &buf->rb_recv_bufs[buf->rb_max_requests]; + + /* + * Register the zeroed pad buffer, if any. + */ + if (cdata->padding) { + rc = rpcrdma_register_internal(ia, p, cdata->padding, + &ep->rep_pad_mr, &ep->rep_pad); + if (rc) + goto out; + } + p += cdata->padding; + + INIT_LIST_HEAD(&buf->rb_mws); + r = (struct rpcrdma_mw *)p; + switch (ia->ri_memreg_strategy) { + case RPCRDMA_FRMR: + for (i = buf->rb_max_requests * RPCRDMA_MAX_SEGS; i; i--) { + r->r.frmr.fr_mr = ib_alloc_fast_reg_mr(ia->ri_pd, + ia->ri_max_frmr_depth); + if (IS_ERR(r->r.frmr.fr_mr)) { + rc = PTR_ERR(r->r.frmr.fr_mr); + dprintk("RPC: %s: ib_alloc_fast_reg_mr" + " failed %i\n", __func__, rc); + goto out; + } + r->r.frmr.fr_pgl = ib_alloc_fast_reg_page_list( + ia->ri_id->device, + ia->ri_max_frmr_depth); + if (IS_ERR(r->r.frmr.fr_pgl)) { + rc = PTR_ERR(r->r.frmr.fr_pgl); + dprintk("RPC: %s: " + "ib_alloc_fast_reg_page_list " + "failed %i\n", __func__, rc); + + ib_dereg_mr(r->r.frmr.fr_mr); + goto out; + } + list_add(&r->mw_list, &buf->rb_mws); + ++r; + } + break; + case RPCRDMA_MTHCAFMR: + /* TBD we are perhaps overallocating here */ + for (i = (buf->rb_max_requests+1) * RPCRDMA_MAX_SEGS; i; i--) { + static struct ib_fmr_attr fa = + { RPCRDMA_MAX_DATA_SEGS, 1, PAGE_SHIFT }; + r->r.fmr = ib_alloc_fmr(ia->ri_pd, + IB_ACCESS_REMOTE_WRITE | IB_ACCESS_REMOTE_READ, + &fa); + if (IS_ERR(r->r.fmr)) { + rc = PTR_ERR(r->r.fmr); + dprintk("RPC: %s: ib_alloc_fmr" + " failed %i\n", __func__, rc); + goto out; + } + list_add(&r->mw_list, &buf->rb_mws); + ++r; + } + break; + default: + break; + } + + /* + * Allocate/init the request/reply buffers. Doing this + * using kmalloc for now -- one for each buf. + */ + wlen = 1 << fls(cdata->inline_wsize + sizeof(struct rpcrdma_req)); + rlen = 1 << fls(cdata->inline_rsize + sizeof(struct rpcrdma_rep)); + dprintk("RPC: %s: wlen = %zu, rlen = %zu\n", + __func__, wlen, rlen); + + for (i = 0; i < buf->rb_max_requests; i++) { + struct rpcrdma_req *req; + struct rpcrdma_rep *rep; + + req = kmalloc(wlen, GFP_KERNEL); + if (req == NULL) { + dprintk("RPC: %s: request buffer %d alloc" + " failed\n", __func__, i); + rc = -ENOMEM; + goto out; + } + memset(req, 0, sizeof(struct rpcrdma_req)); + buf->rb_send_bufs[i] = req; + buf->rb_send_bufs[i]->rl_buffer = buf; + + rc = rpcrdma_register_internal(ia, req->rl_base, + wlen - offsetof(struct rpcrdma_req, rl_base), + &buf->rb_send_bufs[i]->rl_handle, + &buf->rb_send_bufs[i]->rl_iov); + if (rc) + goto out; + + buf->rb_send_bufs[i]->rl_size = wlen - + sizeof(struct rpcrdma_req); + + rep = kmalloc(rlen, GFP_KERNEL); + if (rep == NULL) { + dprintk("RPC: %s: reply buffer %d alloc failed\n", + __func__, i); + rc = -ENOMEM; + goto out; + } + memset(rep, 0, sizeof(struct rpcrdma_rep)); + buf->rb_recv_bufs[i] = rep; + buf->rb_recv_bufs[i]->rr_buffer = buf; + + rc = rpcrdma_register_internal(ia, rep->rr_base, + rlen - offsetof(struct rpcrdma_rep, rr_base), + &buf->rb_recv_bufs[i]->rr_handle, + &buf->rb_recv_bufs[i]->rr_iov); + if (rc) + goto out; + + } + dprintk("RPC: %s: max_requests %d\n", + __func__, buf->rb_max_requests); + /* done */ + return 0; +out: + rpcrdma_buffer_destroy(buf); + return rc; +} + +/* + * Unregister and destroy buffer memory. Need to deal with + * partial initialization, so it's callable from failed create. + * Must be called before destroying endpoint, as registrations + * reference it. + */ +void +rpcrdma_buffer_destroy(struct rpcrdma_buffer *buf) +{ + int rc, i; + struct rpcrdma_ia *ia = rdmab_to_ia(buf); + struct rpcrdma_mw *r; + + /* clean up in reverse order from create + * 1. recv mr memory (mr free, then kfree) + * 2. send mr memory (mr free, then kfree) + * 3. padding (if any) [moved to rpcrdma_ep_destroy] + * 4. arrays + */ + dprintk("RPC: %s: entering\n", __func__); + + for (i = 0; i < buf->rb_max_requests; i++) { + if (buf->rb_recv_bufs && buf->rb_recv_bufs[i]) { + rpcrdma_deregister_internal(ia, + buf->rb_recv_bufs[i]->rr_handle, + &buf->rb_recv_bufs[i]->rr_iov); + kfree(buf->rb_recv_bufs[i]); + } + if (buf->rb_send_bufs && buf->rb_send_bufs[i]) { + rpcrdma_deregister_internal(ia, + buf->rb_send_bufs[i]->rl_handle, + &buf->rb_send_bufs[i]->rl_iov); + kfree(buf->rb_send_bufs[i]); + } + } + + while (!list_empty(&buf->rb_mws)) { + r = list_entry(buf->rb_mws.next, + struct rpcrdma_mw, mw_list); + list_del(&r->mw_list); + switch (ia->ri_memreg_strategy) { + case RPCRDMA_FRMR: + rc = ib_dereg_mr(r->r.frmr.fr_mr); + if (rc) + dprintk("RPC: %s:" + " ib_dereg_mr" + " failed %i\n", + __func__, rc); + ib_free_fast_reg_page_list(r->r.frmr.fr_pgl); + break; + case RPCRDMA_MTHCAFMR: + rc = ib_dealloc_fmr(r->r.fmr); + if (rc) + dprintk("RPC: %s:" + " ib_dealloc_fmr" + " failed %i\n", + __func__, rc); + break; + default: + break; + } + } + + kfree(buf->rb_pool); +} + +/* + * Get a set of request/reply buffers. + * + * Reply buffer (if needed) is attached to send buffer upon return. + * Rule: + * rb_send_index and rb_recv_index MUST always be pointing to the + * *next* available buffer (non-NULL). They are incremented after + * removing buffers, and decremented *before* returning them. + */ +struct rpcrdma_req * +rpcrdma_buffer_get(struct rpcrdma_buffer *buffers) +{ + struct rpcrdma_req *req; + unsigned long flags; + int i; + struct rpcrdma_mw *r; + + spin_lock_irqsave(&buffers->rb_lock, flags); + if (buffers->rb_send_index == buffers->rb_max_requests) { + spin_unlock_irqrestore(&buffers->rb_lock, flags); + dprintk("RPC: %s: out of request buffers\n", __func__); + return ((struct rpcrdma_req *)NULL); + } + + req = buffers->rb_send_bufs[buffers->rb_send_index]; + if (buffers->rb_send_index < buffers->rb_recv_index) { + dprintk("RPC: %s: %d extra receives outstanding (ok)\n", + __func__, + buffers->rb_recv_index - buffers->rb_send_index); + req->rl_reply = NULL; + } else { + req->rl_reply = buffers->rb_recv_bufs[buffers->rb_recv_index]; + buffers->rb_recv_bufs[buffers->rb_recv_index++] = NULL; + } + buffers->rb_send_bufs[buffers->rb_send_index++] = NULL; + if (!list_empty(&buffers->rb_mws)) { + i = RPCRDMA_MAX_SEGS - 1; + do { + r = list_entry(buffers->rb_mws.next, + struct rpcrdma_mw, mw_list); + list_del(&r->mw_list); + req->rl_segments[i].mr_chunk.rl_mw = r; + } while (--i >= 0); + } + spin_unlock_irqrestore(&buffers->rb_lock, flags); + return req; +} + +/* + * Put request/reply buffers back into pool. + * Pre-decrement counter/array index. + */ +void +rpcrdma_buffer_put(struct rpcrdma_req *req) +{ + struct rpcrdma_buffer *buffers = req->rl_buffer; + struct rpcrdma_ia *ia = rdmab_to_ia(buffers); + int i; + unsigned long flags; + + spin_lock_irqsave(&buffers->rb_lock, flags); + buffers->rb_send_bufs[--buffers->rb_send_index] = req; + req->rl_niovs = 0; + if (req->rl_reply) { + buffers->rb_recv_bufs[--buffers->rb_recv_index] = req->rl_reply; + req->rl_reply->rr_func = NULL; + req->rl_reply = NULL; + } + switch (ia->ri_memreg_strategy) { + case RPCRDMA_FRMR: + case RPCRDMA_MTHCAFMR: + /* + * Cycle mw's back in reverse order, and "spin" them. + * This delays and scrambles reuse as much as possible. + */ + i = 1; + do { + struct rpcrdma_mw **mw; + mw = &req->rl_segments[i].mr_chunk.rl_mw; + list_add_tail(&(*mw)->mw_list, &buffers->rb_mws); + *mw = NULL; + } while (++i < RPCRDMA_MAX_SEGS); + list_add_tail(&req->rl_segments[0].mr_chunk.rl_mw->mw_list, + &buffers->rb_mws); + req->rl_segments[0].mr_chunk.rl_mw = NULL; + break; + default: + break; + } + spin_unlock_irqrestore(&buffers->rb_lock, flags); +} + +/* + * Recover reply buffers from pool. + * This happens when recovering from error conditions. + * Post-increment counter/array index. + */ +void +rpcrdma_recv_buffer_get(struct rpcrdma_req *req) +{ + struct rpcrdma_buffer *buffers = req->rl_buffer; + unsigned long flags; + + if (req->rl_iov.length == 0) /* special case xprt_rdma_allocate() */ + buffers = ((struct rpcrdma_req *) buffers)->rl_buffer; + spin_lock_irqsave(&buffers->rb_lock, flags); + if (buffers->rb_recv_index < buffers->rb_max_requests) { + req->rl_reply = buffers->rb_recv_bufs[buffers->rb_recv_index]; + buffers->rb_recv_bufs[buffers->rb_recv_index++] = NULL; + } + spin_unlock_irqrestore(&buffers->rb_lock, flags); +} + +/* + * Put reply buffers back into pool when not attached to + * request. This happens in error conditions. + */ +void +rpcrdma_recv_buffer_put(struct rpcrdma_rep *rep) +{ + struct rpcrdma_buffer *buffers = rep->rr_buffer; + unsigned long flags; + + rep->rr_func = NULL; + spin_lock_irqsave(&buffers->rb_lock, flags); + buffers->rb_recv_bufs[--buffers->rb_recv_index] = rep; + spin_unlock_irqrestore(&buffers->rb_lock, flags); +} + +/* + * Wrappers for internal-use kmalloc memory registration, used by buffer code. + */ + +int +rpcrdma_register_internal(struct rpcrdma_ia *ia, void *va, int len, + struct ib_mr **mrp, struct ib_sge *iov) +{ + struct ib_phys_buf ipb; + struct ib_mr *mr; + int rc; + + /* + * All memory passed here was kmalloc'ed, therefore phys-contiguous. + */ + iov->addr = ib_dma_map_single(ia->ri_id->device, + va, len, DMA_BIDIRECTIONAL); + iov->length = len; + + if (ia->ri_have_dma_lkey) { + *mrp = NULL; + iov->lkey = ia->ri_dma_lkey; + return 0; + } else if (ia->ri_bind_mem != NULL) { + *mrp = NULL; + iov->lkey = ia->ri_bind_mem->lkey; + return 0; + } + + ipb.addr = iov->addr; + ipb.size = iov->length; + mr = ib_reg_phys_mr(ia->ri_pd, &ipb, 1, + IB_ACCESS_LOCAL_WRITE, &iov->addr); + + dprintk("RPC: %s: phys convert: 0x%llx " + "registered 0x%llx length %d\n", + __func__, (unsigned long long)ipb.addr, + (unsigned long long)iov->addr, len); + + if (IS_ERR(mr)) { + *mrp = NULL; + rc = PTR_ERR(mr); + dprintk("RPC: %s: failed with %i\n", __func__, rc); + } else { + *mrp = mr; + iov->lkey = mr->lkey; + rc = 0; + } + + return rc; +} + +int +rpcrdma_deregister_internal(struct rpcrdma_ia *ia, + struct ib_mr *mr, struct ib_sge *iov) +{ + int rc; + + ib_dma_unmap_single(ia->ri_id->device, + iov->addr, iov->length, DMA_BIDIRECTIONAL); + + if (NULL == mr) + return 0; + + rc = ib_dereg_mr(mr); + if (rc) + dprintk("RPC: %s: ib_dereg_mr failed %i\n", __func__, rc); + return rc; +} + +/* + * Wrappers for chunk registration, shared by read/write chunk code. + */ + +static void +rpcrdma_map_one(struct rpcrdma_ia *ia, struct rpcrdma_mr_seg *seg, int writing) +{ + seg->mr_dir = writing ? DMA_FROM_DEVICE : DMA_TO_DEVICE; + seg->mr_dmalen = seg->mr_len; + if (seg->mr_page) + seg->mr_dma = ib_dma_map_page(ia->ri_id->device, + seg->mr_page, offset_in_page(seg->mr_offset), + seg->mr_dmalen, seg->mr_dir); + else + seg->mr_dma = ib_dma_map_single(ia->ri_id->device, + seg->mr_offset, + seg->mr_dmalen, seg->mr_dir); + if (ib_dma_mapping_error(ia->ri_id->device, seg->mr_dma)) { + dprintk("RPC: %s: mr_dma %llx mr_offset %p mr_dma_len %zu\n", + __func__, + (unsigned long long)seg->mr_dma, + seg->mr_offset, seg->mr_dmalen); + } +} + +static void +rpcrdma_unmap_one(struct rpcrdma_ia *ia, struct rpcrdma_mr_seg *seg) +{ + if (seg->mr_page) + ib_dma_unmap_page(ia->ri_id->device, + seg->mr_dma, seg->mr_dmalen, seg->mr_dir); + else + ib_dma_unmap_single(ia->ri_id->device, + seg->mr_dma, seg->mr_dmalen, seg->mr_dir); +} + +static int +rpcrdma_register_frmr_external(struct rpcrdma_mr_seg *seg, + int *nsegs, int writing, struct rpcrdma_ia *ia, + struct rpcrdma_xprt *r_xprt) +{ + struct rpcrdma_mr_seg *seg1 = seg; + struct ib_send_wr invalidate_wr, frmr_wr, *bad_wr, *post_wr; + + u8 key; + int len, pageoff; + int i, rc; + int seg_len; + u64 pa; + int page_no; + + pageoff = offset_in_page(seg1->mr_offset); + seg1->mr_offset -= pageoff; /* start of page */ + seg1->mr_len += pageoff; + len = -pageoff; + if (*nsegs > ia->ri_max_frmr_depth) + *nsegs = ia->ri_max_frmr_depth; + for (page_no = i = 0; i < *nsegs;) { + rpcrdma_map_one(ia, seg, writing); + pa = seg->mr_dma; + for (seg_len = seg->mr_len; seg_len > 0; seg_len -= PAGE_SIZE) { + seg1->mr_chunk.rl_mw->r.frmr.fr_pgl-> + page_list[page_no++] = pa; + pa += PAGE_SIZE; + } + len += seg->mr_len; + ++seg; + ++i; + /* Check for holes */ + if ((i < *nsegs && offset_in_page(seg->mr_offset)) || + offset_in_page((seg-1)->mr_offset + (seg-1)->mr_len)) + break; + } + dprintk("RPC: %s: Using frmr %p to map %d segments\n", + __func__, seg1->mr_chunk.rl_mw, i); + + if (unlikely(seg1->mr_chunk.rl_mw->r.frmr.state == FRMR_IS_VALID)) { + dprintk("RPC: %s: frmr %x left valid, posting invalidate.\n", + __func__, + seg1->mr_chunk.rl_mw->r.frmr.fr_mr->rkey); + /* Invalidate before using. */ + memset(&invalidate_wr, 0, sizeof invalidate_wr); + invalidate_wr.wr_id = (unsigned long)(void *)seg1->mr_chunk.rl_mw; + invalidate_wr.next = &frmr_wr; + invalidate_wr.opcode = IB_WR_LOCAL_INV; + invalidate_wr.send_flags = IB_SEND_SIGNALED; + invalidate_wr.ex.invalidate_rkey = + seg1->mr_chunk.rl_mw->r.frmr.fr_mr->rkey; + DECR_CQCOUNT(&r_xprt->rx_ep); + post_wr = &invalidate_wr; + } else + post_wr = &frmr_wr; + + /* Prepare FRMR WR */ + memset(&frmr_wr, 0, sizeof frmr_wr); + frmr_wr.wr_id = (unsigned long)(void *)seg1->mr_chunk.rl_mw; + frmr_wr.opcode = IB_WR_FAST_REG_MR; + frmr_wr.send_flags = IB_SEND_SIGNALED; + frmr_wr.wr.fast_reg.iova_start = seg1->mr_dma; + frmr_wr.wr.fast_reg.page_list = seg1->mr_chunk.rl_mw->r.frmr.fr_pgl; + frmr_wr.wr.fast_reg.page_list_len = page_no; + frmr_wr.wr.fast_reg.page_shift = PAGE_SHIFT; + frmr_wr.wr.fast_reg.length = page_no << PAGE_SHIFT; + if (frmr_wr.wr.fast_reg.length < len) { + while (seg1->mr_nsegs--) + rpcrdma_unmap_one(ia, seg++); + return -EIO; + } + + /* Bump the key */ + key = (u8)(seg1->mr_chunk.rl_mw->r.frmr.fr_mr->rkey & 0x000000FF); + ib_update_fast_reg_key(seg1->mr_chunk.rl_mw->r.frmr.fr_mr, ++key); + + frmr_wr.wr.fast_reg.access_flags = (writing ? + IB_ACCESS_REMOTE_WRITE | IB_ACCESS_LOCAL_WRITE : + IB_ACCESS_REMOTE_READ); + frmr_wr.wr.fast_reg.rkey = seg1->mr_chunk.rl_mw->r.frmr.fr_mr->rkey; + DECR_CQCOUNT(&r_xprt->rx_ep); + + rc = ib_post_send(ia->ri_id->qp, post_wr, &bad_wr); + + if (rc) { + dprintk("RPC: %s: failed ib_post_send for register," + " status %i\n", __func__, rc); + while (i--) + rpcrdma_unmap_one(ia, --seg); + } else { + seg1->mr_rkey = seg1->mr_chunk.rl_mw->r.frmr.fr_mr->rkey; + seg1->mr_base = seg1->mr_dma + pageoff; + seg1->mr_nsegs = i; + seg1->mr_len = len; + } + *nsegs = i; + return rc; +} + +static int +rpcrdma_deregister_frmr_external(struct rpcrdma_mr_seg *seg, + struct rpcrdma_ia *ia, struct rpcrdma_xprt *r_xprt) +{ + struct rpcrdma_mr_seg *seg1 = seg; + struct ib_send_wr invalidate_wr, *bad_wr; + int rc; + + while (seg1->mr_nsegs--) + rpcrdma_unmap_one(ia, seg++); + + memset(&invalidate_wr, 0, sizeof invalidate_wr); + invalidate_wr.wr_id = (unsigned long)(void *)seg1->mr_chunk.rl_mw; + invalidate_wr.opcode = IB_WR_LOCAL_INV; + invalidate_wr.send_flags = IB_SEND_SIGNALED; + invalidate_wr.ex.invalidate_rkey = seg1->mr_chunk.rl_mw->r.frmr.fr_mr->rkey; + DECR_CQCOUNT(&r_xprt->rx_ep); + + rc = ib_post_send(ia->ri_id->qp, &invalidate_wr, &bad_wr); + if (rc) + dprintk("RPC: %s: failed ib_post_send for invalidate," + " status %i\n", __func__, rc); + return rc; +} + +static int +rpcrdma_register_fmr_external(struct rpcrdma_mr_seg *seg, + int *nsegs, int writing, struct rpcrdma_ia *ia) +{ + struct rpcrdma_mr_seg *seg1 = seg; + u64 physaddrs[RPCRDMA_MAX_DATA_SEGS]; + int len, pageoff, i, rc; + + pageoff = offset_in_page(seg1->mr_offset); + seg1->mr_offset -= pageoff; /* start of page */ + seg1->mr_len += pageoff; + len = -pageoff; + if (*nsegs > RPCRDMA_MAX_DATA_SEGS) + *nsegs = RPCRDMA_MAX_DATA_SEGS; + for (i = 0; i < *nsegs;) { + rpcrdma_map_one(ia, seg, writing); + physaddrs[i] = seg->mr_dma; + len += seg->mr_len; + ++seg; + ++i; + /* Check for holes */ + if ((i < *nsegs && offset_in_page(seg->mr_offset)) || + offset_in_page((seg-1)->mr_offset + (seg-1)->mr_len)) + break; + } + rc = ib_map_phys_fmr(seg1->mr_chunk.rl_mw->r.fmr, + physaddrs, i, seg1->mr_dma); + if (rc) { + dprintk("RPC: %s: failed ib_map_phys_fmr " + "%u@0x%llx+%i (%d)... status %i\n", __func__, + len, (unsigned long long)seg1->mr_dma, + pageoff, i, rc); + while (i--) + rpcrdma_unmap_one(ia, --seg); + } else { + seg1->mr_rkey = seg1->mr_chunk.rl_mw->r.fmr->rkey; + seg1->mr_base = seg1->mr_dma + pageoff; + seg1->mr_nsegs = i; + seg1->mr_len = len; + } + *nsegs = i; + return rc; +} + +static int +rpcrdma_deregister_fmr_external(struct rpcrdma_mr_seg *seg, + struct rpcrdma_ia *ia) +{ + struct rpcrdma_mr_seg *seg1 = seg; + LIST_HEAD(l); + int rc; + + list_add(&seg1->mr_chunk.rl_mw->r.fmr->list, &l); + rc = ib_unmap_fmr(&l); + while (seg1->mr_nsegs--) + rpcrdma_unmap_one(ia, seg++); + if (rc) + dprintk("RPC: %s: failed ib_unmap_fmr," + " status %i\n", __func__, rc); + return rc; +} + +int +rpcrdma_register_external(struct rpcrdma_mr_seg *seg, + int nsegs, int writing, struct rpcrdma_xprt *r_xprt) +{ + struct rpcrdma_ia *ia = &r_xprt->rx_ia; + int rc = 0; + + switch (ia->ri_memreg_strategy) { + +#if RPCRDMA_PERSISTENT_REGISTRATION + case RPCRDMA_ALLPHYSICAL: + rpcrdma_map_one(ia, seg, writing); + seg->mr_rkey = ia->ri_bind_mem->rkey; + seg->mr_base = seg->mr_dma; + seg->mr_nsegs = 1; + nsegs = 1; + break; +#endif + + /* Registration using frmr registration */ + case RPCRDMA_FRMR: + rc = rpcrdma_register_frmr_external(seg, &nsegs, writing, ia, r_xprt); + break; + + /* Registration using fmr memory registration */ + case RPCRDMA_MTHCAFMR: + rc = rpcrdma_register_fmr_external(seg, &nsegs, writing, ia); + break; + + default: + return -1; + } + if (rc) + return -1; + + return nsegs; +} + +int +rpcrdma_deregister_external(struct rpcrdma_mr_seg *seg, + struct rpcrdma_xprt *r_xprt) +{ + struct rpcrdma_ia *ia = &r_xprt->rx_ia; + int nsegs = seg->mr_nsegs, rc; + + switch (ia->ri_memreg_strategy) { + +#if RPCRDMA_PERSISTENT_REGISTRATION + case RPCRDMA_ALLPHYSICAL: + rpcrdma_unmap_one(ia, seg); + break; +#endif + + case RPCRDMA_FRMR: + rc = rpcrdma_deregister_frmr_external(seg, ia, r_xprt); + break; + + case RPCRDMA_MTHCAFMR: + rc = rpcrdma_deregister_fmr_external(seg, ia); + break; + + default: + break; + } + return nsegs; +} + +/* + * Prepost any receive buffer, then post send. + * + * Receive buffer is donated to hardware, reclaimed upon recv completion. + */ +int +rpcrdma_ep_post(struct rpcrdma_ia *ia, + struct rpcrdma_ep *ep, + struct rpcrdma_req *req) +{ + struct ib_send_wr send_wr, *send_wr_fail; + struct rpcrdma_rep *rep = req->rl_reply; + int rc; + + if (rep) { + rc = rpcrdma_ep_post_recv(ia, ep, rep); + if (rc) + goto out; + req->rl_reply = NULL; + } + + send_wr.next = NULL; + send_wr.wr_id = 0ULL; /* no send cookie */ + send_wr.sg_list = req->rl_send_iov; + send_wr.num_sge = req->rl_niovs; + send_wr.opcode = IB_WR_SEND; + if (send_wr.num_sge == 4) /* no need to sync any pad (constant) */ + ib_dma_sync_single_for_device(ia->ri_id->device, + req->rl_send_iov[3].addr, req->rl_send_iov[3].length, + DMA_TO_DEVICE); + ib_dma_sync_single_for_device(ia->ri_id->device, + req->rl_send_iov[1].addr, req->rl_send_iov[1].length, + DMA_TO_DEVICE); + ib_dma_sync_single_for_device(ia->ri_id->device, + req->rl_send_iov[0].addr, req->rl_send_iov[0].length, + DMA_TO_DEVICE); + + if (DECR_CQCOUNT(ep) > 0) + send_wr.send_flags = 0; + else { /* Provider must take a send completion every now and then */ + INIT_CQCOUNT(ep); + send_wr.send_flags = IB_SEND_SIGNALED; + } + + rc = ib_post_send(ia->ri_id->qp, &send_wr, &send_wr_fail); + if (rc) + dprintk("RPC: %s: ib_post_send returned %i\n", __func__, + rc); +out: + return rc; +} + +/* + * (Re)post a receive buffer. + */ +int +rpcrdma_ep_post_recv(struct rpcrdma_ia *ia, + struct rpcrdma_ep *ep, + struct rpcrdma_rep *rep) +{ + struct ib_recv_wr recv_wr, *recv_wr_fail; + int rc; + + recv_wr.next = NULL; + recv_wr.wr_id = (u64) (unsigned long) rep; + recv_wr.sg_list = &rep->rr_iov; + recv_wr.num_sge = 1; + + ib_dma_sync_single_for_cpu(ia->ri_id->device, + rep->rr_iov.addr, rep->rr_iov.length, DMA_BIDIRECTIONAL); + + rc = ib_post_recv(ia->ri_id->qp, &recv_wr, &recv_wr_fail); + + if (rc) + dprintk("RPC: %s: ib_post_recv returned %i\n", __func__, + rc); + return rc; +} diff --git a/net/sunrpc/xprtrdma/xprt_rdma.h b/net/sunrpc/xprtrdma/xprt_rdma.h new file mode 100644 index 00000000000..89e7cd47970 --- /dev/null +++ b/net/sunrpc/xprtrdma/xprt_rdma.h @@ -0,0 +1,358 @@ +/* + * Copyright (c) 2003-2007 Network Appliance, Inc. All rights reserved. + * + * This software is available to you under a choice of one of two + * licenses. You may choose to be licensed under the terms of the GNU + * General Public License (GPL) Version 2, available from the file + * COPYING in the main directory of this source tree, or the BSD-type + * license below: + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * + * Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * + * Redistributions in binary form must reproduce the above + * copyright notice, this list of conditions and the following + * disclaimer in the documentation and/or other materials provided + * with the distribution. + * + * Neither the name of the Network Appliance, Inc. nor the names of + * its contributors may be used to endorse or promote products + * derived from this software without specific prior written + * permission. + * + * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS + * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT + * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR + * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT + * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, + * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT + * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE + * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#ifndef _LINUX_SUNRPC_XPRT_RDMA_H +#define _LINUX_SUNRPC_XPRT_RDMA_H + +#include <linux/wait.h> /* wait_queue_head_t, etc */ +#include <linux/spinlock.h> /* spinlock_t, etc */ +#include <linux/atomic.h> /* atomic_t, etc */ +#include <linux/workqueue.h> /* struct work_struct */ + +#include <rdma/rdma_cm.h> /* RDMA connection api */ +#include <rdma/ib_verbs.h> /* RDMA verbs api */ + +#include <linux/sunrpc/clnt.h> /* rpc_xprt */ +#include <linux/sunrpc/rpc_rdma.h> /* RPC/RDMA protocol */ +#include <linux/sunrpc/xprtrdma.h> /* xprt parameters */ + +#define RDMA_RESOLVE_TIMEOUT (5000) /* 5 seconds */ +#define RDMA_CONNECT_RETRY_MAX (2) /* retries if no listener backlog */ + +/* + * Interface Adapter -- one per transport instance + */ +struct rpcrdma_ia { + struct rdma_cm_id *ri_id; + struct ib_pd *ri_pd; + struct ib_mr *ri_bind_mem; + u32 ri_dma_lkey; + int ri_have_dma_lkey; + struct completion ri_done; + int ri_async_rc; + enum rpcrdma_memreg ri_memreg_strategy; + unsigned int ri_max_frmr_depth; +}; + +/* + * RDMA Endpoint -- one per transport instance + */ + +#define RPCRDMA_WC_BUDGET (128) +#define RPCRDMA_POLLSIZE (16) + +struct rpcrdma_ep { + atomic_t rep_cqcount; + int rep_cqinit; + int rep_connected; + struct rpcrdma_ia *rep_ia; + struct ib_qp_init_attr rep_attr; + wait_queue_head_t rep_connect_wait; + struct ib_sge rep_pad; /* holds zeroed pad */ + struct ib_mr *rep_pad_mr; /* holds zeroed pad */ + void (*rep_func)(struct rpcrdma_ep *); + struct rpc_xprt *rep_xprt; /* for rep_func */ + struct rdma_conn_param rep_remote_cma; + struct sockaddr_storage rep_remote_addr; + struct delayed_work rep_connect_worker; + struct ib_wc rep_send_wcs[RPCRDMA_POLLSIZE]; + struct ib_wc rep_recv_wcs[RPCRDMA_POLLSIZE]; +}; + +#define INIT_CQCOUNT(ep) atomic_set(&(ep)->rep_cqcount, (ep)->rep_cqinit) +#define DECR_CQCOUNT(ep) atomic_sub_return(1, &(ep)->rep_cqcount) + +/* + * struct rpcrdma_rep -- this structure encapsulates state required to recv + * and complete a reply, asychronously. It needs several pieces of + * state: + * o recv buffer (posted to provider) + * o ib_sge (also donated to provider) + * o status of reply (length, success or not) + * o bookkeeping state to get run by tasklet (list, etc) + * + * These are allocated during initialization, per-transport instance; + * however, the tasklet execution list itself is global, as it should + * always be pretty short. + * + * N of these are associated with a transport instance, and stored in + * struct rpcrdma_buffer. N is the max number of outstanding requests. + */ + +/* temporary static scatter/gather max */ +#define RPCRDMA_MAX_DATA_SEGS (64) /* max scatter/gather */ +#define RPCRDMA_MAX_SEGS (RPCRDMA_MAX_DATA_SEGS + 2) /* head+tail = 2 */ +#define MAX_RPCRDMAHDR (\ + /* max supported RPC/RDMA header */ \ + sizeof(struct rpcrdma_msg) + (2 * sizeof(u32)) + \ + (sizeof(struct rpcrdma_read_chunk) * RPCRDMA_MAX_SEGS) + sizeof(u32)) + +struct rpcrdma_buffer; + +struct rpcrdma_rep { + unsigned int rr_len; /* actual received reply length */ + struct rpcrdma_buffer *rr_buffer; /* home base for this structure */ + struct rpc_xprt *rr_xprt; /* needed for request/reply matching */ + void (*rr_func)(struct rpcrdma_rep *);/* called by tasklet in softint */ + struct list_head rr_list; /* tasklet list */ + struct ib_sge rr_iov; /* for posting */ + struct ib_mr *rr_handle; /* handle for mem in rr_iov */ + char rr_base[MAX_RPCRDMAHDR]; /* minimal inline receive buffer */ +}; + +/* + * struct rpcrdma_req -- structure central to the request/reply sequence. + * + * N of these are associated with a transport instance, and stored in + * struct rpcrdma_buffer. N is the max number of outstanding requests. + * + * It includes pre-registered buffer memory for send AND recv. + * The recv buffer, however, is not owned by this structure, and + * is "donated" to the hardware when a recv is posted. When a + * reply is handled, the recv buffer used is given back to the + * struct rpcrdma_req associated with the request. + * + * In addition to the basic memory, this structure includes an array + * of iovs for send operations. The reason is that the iovs passed to + * ib_post_{send,recv} must not be modified until the work request + * completes. + * + * NOTES: + * o RPCRDMA_MAX_SEGS is the max number of addressible chunk elements we + * marshal. The number needed varies depending on the iov lists that + * are passed to us, the memory registration mode we are in, and if + * physical addressing is used, the layout. + */ + +struct rpcrdma_mr_seg { /* chunk descriptors */ + union { /* chunk memory handles */ + struct ib_mr *rl_mr; /* if registered directly */ + struct rpcrdma_mw { /* if registered from region */ + union { + struct ib_fmr *fmr; + struct { + struct ib_fast_reg_page_list *fr_pgl; + struct ib_mr *fr_mr; + enum { FRMR_IS_INVALID, FRMR_IS_VALID } state; + } frmr; + } r; + struct list_head mw_list; + } *rl_mw; + } mr_chunk; + u64 mr_base; /* registration result */ + u32 mr_rkey; /* registration result */ + u32 mr_len; /* length of chunk or segment */ + int mr_nsegs; /* number of segments in chunk or 0 */ + enum dma_data_direction mr_dir; /* segment mapping direction */ + dma_addr_t mr_dma; /* segment mapping address */ + size_t mr_dmalen; /* segment mapping length */ + struct page *mr_page; /* owning page, if any */ + char *mr_offset; /* kva if no page, else offset */ +}; + +struct rpcrdma_req { + size_t rl_size; /* actual length of buffer */ + unsigned int rl_niovs; /* 0, 2 or 4 */ + unsigned int rl_nchunks; /* non-zero if chunks */ + unsigned int rl_connect_cookie; /* retry detection */ + struct rpcrdma_buffer *rl_buffer; /* home base for this structure */ + struct rpcrdma_rep *rl_reply;/* holder for reply buffer */ + struct rpcrdma_mr_seg rl_segments[RPCRDMA_MAX_SEGS];/* chunk segments */ + struct ib_sge rl_send_iov[4]; /* for active requests */ + struct ib_sge rl_iov; /* for posting */ + struct ib_mr *rl_handle; /* handle for mem in rl_iov */ + char rl_base[MAX_RPCRDMAHDR]; /* start of actual buffer */ + __u32 rl_xdr_buf[0]; /* start of returned rpc rq_buffer */ +}; +#define rpcr_to_rdmar(r) \ + container_of((r)->rq_buffer, struct rpcrdma_req, rl_xdr_buf[0]) + +/* + * struct rpcrdma_buffer -- holds list/queue of pre-registered memory for + * inline requests/replies, and client/server credits. + * + * One of these is associated with a transport instance + */ +struct rpcrdma_buffer { + spinlock_t rb_lock; /* protects indexes */ + atomic_t rb_credits; /* most recent server credits */ + int rb_max_requests;/* client max requests */ + struct list_head rb_mws; /* optional memory windows/fmrs/frmrs */ + int rb_send_index; + struct rpcrdma_req **rb_send_bufs; + int rb_recv_index; + struct rpcrdma_rep **rb_recv_bufs; + char *rb_pool; +}; +#define rdmab_to_ia(b) (&container_of((b), struct rpcrdma_xprt, rx_buf)->rx_ia) + +/* + * Internal structure for transport instance creation. This + * exists primarily for modularity. + * + * This data should be set with mount options + */ +struct rpcrdma_create_data_internal { + struct sockaddr_storage addr; /* RDMA server address */ + unsigned int max_requests; /* max requests (slots) in flight */ + unsigned int rsize; /* mount rsize - max read hdr+data */ + unsigned int wsize; /* mount wsize - max write hdr+data */ + unsigned int inline_rsize; /* max non-rdma read data payload */ + unsigned int inline_wsize; /* max non-rdma write data payload */ + unsigned int padding; /* non-rdma write header padding */ +}; + +#define RPCRDMA_INLINE_READ_THRESHOLD(rq) \ + (rpcx_to_rdmad(rq->rq_xprt).inline_rsize) + +#define RPCRDMA_INLINE_WRITE_THRESHOLD(rq)\ + (rpcx_to_rdmad(rq->rq_xprt).inline_wsize) + +#define RPCRDMA_INLINE_PAD_VALUE(rq)\ + rpcx_to_rdmad(rq->rq_xprt).padding + +/* + * Statistics for RPCRDMA + */ +struct rpcrdma_stats { + unsigned long read_chunk_count; + unsigned long write_chunk_count; + unsigned long reply_chunk_count; + + unsigned long long total_rdma_request; + unsigned long long total_rdma_reply; + + unsigned long long pullup_copy_count; + unsigned long long fixup_copy_count; + unsigned long hardway_register_count; + unsigned long failed_marshal_count; + unsigned long bad_reply_count; +}; + +/* + * RPCRDMA transport -- encapsulates the structures above for + * integration with RPC. + * + * The contained structures are embedded, not pointers, + * for convenience. This structure need not be visible externally. + * + * It is allocated and initialized during mount, and released + * during unmount. + */ +struct rpcrdma_xprt { + struct rpc_xprt xprt; + struct rpcrdma_ia rx_ia; + struct rpcrdma_ep rx_ep; + struct rpcrdma_buffer rx_buf; + struct rpcrdma_create_data_internal rx_data; + struct delayed_work rdma_connect; + struct rpcrdma_stats rx_stats; +}; + +#define rpcx_to_rdmax(x) container_of(x, struct rpcrdma_xprt, xprt) +#define rpcx_to_rdmad(x) (rpcx_to_rdmax(x)->rx_data) + +/* Setting this to 0 ensures interoperability with early servers. + * Setting this to 1 enhances certain unaligned read/write performance. + * Default is 0, see sysctl entry and rpc_rdma.c rpcrdma_convert_iovs() */ +extern int xprt_rdma_pad_optimize; + +/* + * Interface Adapter calls - xprtrdma/verbs.c + */ +int rpcrdma_ia_open(struct rpcrdma_xprt *, struct sockaddr *, int); +void rpcrdma_ia_close(struct rpcrdma_ia *); + +/* + * Endpoint calls - xprtrdma/verbs.c + */ +int rpcrdma_ep_create(struct rpcrdma_ep *, struct rpcrdma_ia *, + struct rpcrdma_create_data_internal *); +void rpcrdma_ep_destroy(struct rpcrdma_ep *, struct rpcrdma_ia *); +int rpcrdma_ep_connect(struct rpcrdma_ep *, struct rpcrdma_ia *); +int rpcrdma_ep_disconnect(struct rpcrdma_ep *, struct rpcrdma_ia *); + +int rpcrdma_ep_post(struct rpcrdma_ia *, struct rpcrdma_ep *, + struct rpcrdma_req *); +int rpcrdma_ep_post_recv(struct rpcrdma_ia *, struct rpcrdma_ep *, + struct rpcrdma_rep *); + +/* + * Buffer calls - xprtrdma/verbs.c + */ +int rpcrdma_buffer_create(struct rpcrdma_buffer *, struct rpcrdma_ep *, + struct rpcrdma_ia *, + struct rpcrdma_create_data_internal *); +void rpcrdma_buffer_destroy(struct rpcrdma_buffer *); + +struct rpcrdma_req *rpcrdma_buffer_get(struct rpcrdma_buffer *); +void rpcrdma_buffer_put(struct rpcrdma_req *); +void rpcrdma_recv_buffer_get(struct rpcrdma_req *); +void rpcrdma_recv_buffer_put(struct rpcrdma_rep *); + +int rpcrdma_register_internal(struct rpcrdma_ia *, void *, int, + struct ib_mr **, struct ib_sge *); +int rpcrdma_deregister_internal(struct rpcrdma_ia *, + struct ib_mr *, struct ib_sge *); + +int rpcrdma_register_external(struct rpcrdma_mr_seg *, + int, int, struct rpcrdma_xprt *); +int rpcrdma_deregister_external(struct rpcrdma_mr_seg *, + struct rpcrdma_xprt *); + +/* + * RPC/RDMA connection management calls - xprtrdma/rpc_rdma.c + */ +void rpcrdma_connect_worker(struct work_struct *); +void rpcrdma_conn_func(struct rpcrdma_ep *); +void rpcrdma_reply_handler(struct rpcrdma_rep *); + +/* + * RPC/RDMA protocol calls - xprtrdma/rpc_rdma.c + */ +int rpcrdma_marshal_req(struct rpc_rqst *); + +/* Temporary NFS request map cache. Created in svc_rdma.c */ +extern struct kmem_cache *svc_rdma_map_cachep; +/* WR context cache. Created in svc_rdma.c */ +extern struct kmem_cache *svc_rdma_ctxt_cachep; +/* Workqueue created in svc_rdma.c */ +extern struct workqueue_struct *svc_rdma_wq; + +#endif /* _LINUX_SUNRPC_XPRT_RDMA_H */ diff --git a/net/sunrpc/xprtsock.c b/net/sunrpc/xprtsock.c index 0a51fd46a84..be8bbd5d65e 100644 --- a/net/sunrpc/xprtsock.c +++ b/net/sunrpc/xprtsock.c @@ -3,8 +3,8 @@ * * Client-side transport implementation for sockets. * - * TCP callback races fixes (C) 1998 Red Hat Software <alan@redhat.com> - * TCP send fixes (C) 1998 Red Hat Software <alan@redhat.com> + * TCP callback races fixes (C) 1998 Red Hat + * TCP send fixes (C) 1998 Red Hat * TCP NFS related read + write fixes * (C) 1999 Dave Airlie, University of Limerick, Ireland <airlied@linux.ie> * @@ -13,56 +13,148 @@ * (C) 1999 Trond Myklebust <trond.myklebust@fys.uio.no> * * IP socket transport implementation, (C) 2005 Chuck Lever <cel@netapp.com> + * + * IPv6 support contributed by Gilles Quillard, Bull Open Source, 2005. + * <gilles.quillard@bull.net> */ #include <linux/types.h> +#include <linux/string.h> #include <linux/slab.h> +#include <linux/module.h> #include <linux/capability.h> -#include <linux/sched.h> #include <linux/pagemap.h> #include <linux/errno.h> #include <linux/socket.h> #include <linux/in.h> #include <linux/net.h> #include <linux/mm.h> +#include <linux/un.h> #include <linux/udp.h> #include <linux/tcp.h> #include <linux/sunrpc/clnt.h> +#include <linux/sunrpc/addr.h> +#include <linux/sunrpc/sched.h> +#include <linux/sunrpc/svcsock.h> +#include <linux/sunrpc/xprtsock.h> #include <linux/file.h> +#ifdef CONFIG_SUNRPC_BACKCHANNEL +#include <linux/sunrpc/bc_xprt.h> +#endif #include <net/sock.h> #include <net/checksum.h> #include <net/udp.h> #include <net/tcp.h> +#include <trace/events/sunrpc.h> + +#include "sunrpc.h" + +static void xs_close(struct rpc_xprt *xprt); + /* * xprtsock tunables */ -unsigned int xprt_udp_slot_table_entries = RPC_DEF_SLOT_TABLE; -unsigned int xprt_tcp_slot_table_entries = RPC_DEF_SLOT_TABLE; +static unsigned int xprt_udp_slot_table_entries = RPC_DEF_SLOT_TABLE; +static unsigned int xprt_tcp_slot_table_entries = RPC_MIN_SLOT_TABLE; +static unsigned int xprt_max_tcp_slot_table_entries = RPC_MAX_SLOT_TABLE; -unsigned int xprt_min_resvport = RPC_DEF_MIN_RESVPORT; -unsigned int xprt_max_resvport = RPC_DEF_MAX_RESVPORT; +static unsigned int xprt_min_resvport = RPC_DEF_MIN_RESVPORT; +static unsigned int xprt_max_resvport = RPC_DEF_MAX_RESVPORT; -/* - * How many times to try sending a request on a socket before waiting - * for the socket buffer to clear. - */ -#define XS_SENDMSG_RETRY (10U) +#define XS_TCP_LINGER_TO (15U * HZ) +static unsigned int xs_tcp_fin_timeout __read_mostly = XS_TCP_LINGER_TO; /* - * Time out for an RPC UDP socket connect. UDP socket connects are - * synchronous, but we set a timeout anyway in case of resource - * exhaustion on the local host. + * We can register our own files under /proc/sys/sunrpc by + * calling register_sysctl_table() again. The files in that + * directory become the union of all files registered there. + * + * We simply need to make sure that we don't collide with + * someone else's file names! */ -#define XS_UDP_CONN_TO (5U * HZ) + +#ifdef RPC_DEBUG + +static unsigned int min_slot_table_size = RPC_MIN_SLOT_TABLE; +static unsigned int max_slot_table_size = RPC_MAX_SLOT_TABLE; +static unsigned int max_tcp_slot_table_limit = RPC_MAX_SLOT_TABLE_LIMIT; +static unsigned int xprt_min_resvport_limit = RPC_MIN_RESVPORT; +static unsigned int xprt_max_resvport_limit = RPC_MAX_RESVPORT; + +static struct ctl_table_header *sunrpc_table_header; /* - * Wait duration for an RPC TCP connection to be established. Solaris - * NFS over TCP uses 60 seconds, for example, which is in line with how - * long a server takes to reboot. + * FIXME: changing the UDP slot table size should also resize the UDP + * socket buffers for existing UDP transports */ -#define XS_TCP_CONN_TO (60U * HZ) +static struct ctl_table xs_tunables_table[] = { + { + .procname = "udp_slot_table_entries", + .data = &xprt_udp_slot_table_entries, + .maxlen = sizeof(unsigned int), + .mode = 0644, + .proc_handler = proc_dointvec_minmax, + .extra1 = &min_slot_table_size, + .extra2 = &max_slot_table_size + }, + { + .procname = "tcp_slot_table_entries", + .data = &xprt_tcp_slot_table_entries, + .maxlen = sizeof(unsigned int), + .mode = 0644, + .proc_handler = proc_dointvec_minmax, + .extra1 = &min_slot_table_size, + .extra2 = &max_slot_table_size + }, + { + .procname = "tcp_max_slot_table_entries", + .data = &xprt_max_tcp_slot_table_entries, + .maxlen = sizeof(unsigned int), + .mode = 0644, + .proc_handler = proc_dointvec_minmax, + .extra1 = &min_slot_table_size, + .extra2 = &max_tcp_slot_table_limit + }, + { + .procname = "min_resvport", + .data = &xprt_min_resvport, + .maxlen = sizeof(unsigned int), + .mode = 0644, + .proc_handler = proc_dointvec_minmax, + .extra1 = &xprt_min_resvport_limit, + .extra2 = &xprt_max_resvport_limit + }, + { + .procname = "max_resvport", + .data = &xprt_max_resvport, + .maxlen = sizeof(unsigned int), + .mode = 0644, + .proc_handler = proc_dointvec_minmax, + .extra1 = &xprt_min_resvport_limit, + .extra2 = &xprt_max_resvport_limit + }, + { + .procname = "tcp_fin_timeout", + .data = &xs_tcp_fin_timeout, + .maxlen = sizeof(xs_tcp_fin_timeout), + .mode = 0644, + .proc_handler = proc_dointvec_jiffies, + }, + { }, +}; + +static struct ctl_table sunrpc_table[] = { + { + .procname = "sunrpc", + .mode = 0555, + .child = xs_tunables_table + }, + { }, +}; + +#endif /* * Wait duration for a reply from the RPC portmapper. @@ -105,7 +197,7 @@ static void xs_pktdump(char *msg, u32 *packet, unsigned int count) u8 *buf = (u8 *) packet; int j; - dprintk("RPC: %s\n", msg); + dprintk("RPC: %s\n", msg); for (j = 0; j < count && j < 128; j += 4) { if (!(j & 31)) { if (j) @@ -124,39 +216,223 @@ static inline void xs_pktdump(char *msg, u32 *packet, unsigned int count) } #endif +struct sock_xprt { + struct rpc_xprt xprt; + + /* + * Network layer + */ + struct socket * sock; + struct sock * inet; + + /* + * State of TCP reply receive + */ + __be32 tcp_fraghdr, + tcp_xid, + tcp_calldir; + + u32 tcp_offset, + tcp_reclen; + + unsigned long tcp_copied, + tcp_flags; + + /* + * Connection of transports + */ + struct delayed_work connect_worker; + struct sockaddr_storage srcaddr; + unsigned short srcport; + + /* + * UDP socket buffer size parameters + */ + size_t rcvsize, + sndsize; + + /* + * Saved socket callback addresses + */ + void (*old_data_ready)(struct sock *); + void (*old_state_change)(struct sock *); + void (*old_write_space)(struct sock *); + void (*old_error_report)(struct sock *); +}; + +/* + * TCP receive state flags + */ +#define TCP_RCV_LAST_FRAG (1UL << 0) +#define TCP_RCV_COPY_FRAGHDR (1UL << 1) +#define TCP_RCV_COPY_XID (1UL << 2) +#define TCP_RCV_COPY_DATA (1UL << 3) +#define TCP_RCV_READ_CALLDIR (1UL << 4) +#define TCP_RCV_COPY_CALLDIR (1UL << 5) + +/* + * TCP RPC flags + */ +#define TCP_RPC_REPLY (1UL << 6) + +static inline struct rpc_xprt *xprt_from_sock(struct sock *sk) +{ + return (struct rpc_xprt *) sk->sk_user_data; +} + +static inline struct sockaddr *xs_addr(struct rpc_xprt *xprt) +{ + return (struct sockaddr *) &xprt->addr; +} + +static inline struct sockaddr_un *xs_addr_un(struct rpc_xprt *xprt) +{ + return (struct sockaddr_un *) &xprt->addr; +} + +static inline struct sockaddr_in *xs_addr_in(struct rpc_xprt *xprt) +{ + return (struct sockaddr_in *) &xprt->addr; +} + +static inline struct sockaddr_in6 *xs_addr_in6(struct rpc_xprt *xprt) +{ + return (struct sockaddr_in6 *) &xprt->addr; +} + +static void xs_format_common_peer_addresses(struct rpc_xprt *xprt) +{ + struct sockaddr *sap = xs_addr(xprt); + struct sockaddr_in6 *sin6; + struct sockaddr_in *sin; + struct sockaddr_un *sun; + char buf[128]; + + switch (sap->sa_family) { + case AF_LOCAL: + sun = xs_addr_un(xprt); + strlcpy(buf, sun->sun_path, sizeof(buf)); + xprt->address_strings[RPC_DISPLAY_ADDR] = + kstrdup(buf, GFP_KERNEL); + break; + case AF_INET: + (void)rpc_ntop(sap, buf, sizeof(buf)); + xprt->address_strings[RPC_DISPLAY_ADDR] = + kstrdup(buf, GFP_KERNEL); + sin = xs_addr_in(xprt); + snprintf(buf, sizeof(buf), "%08x", ntohl(sin->sin_addr.s_addr)); + break; + case AF_INET6: + (void)rpc_ntop(sap, buf, sizeof(buf)); + xprt->address_strings[RPC_DISPLAY_ADDR] = + kstrdup(buf, GFP_KERNEL); + sin6 = xs_addr_in6(xprt); + snprintf(buf, sizeof(buf), "%pi6", &sin6->sin6_addr); + break; + default: + BUG(); + } + + xprt->address_strings[RPC_DISPLAY_HEX_ADDR] = kstrdup(buf, GFP_KERNEL); +} + +static void xs_format_common_peer_ports(struct rpc_xprt *xprt) +{ + struct sockaddr *sap = xs_addr(xprt); + char buf[128]; + + snprintf(buf, sizeof(buf), "%u", rpc_get_port(sap)); + xprt->address_strings[RPC_DISPLAY_PORT] = kstrdup(buf, GFP_KERNEL); + + snprintf(buf, sizeof(buf), "%4hx", rpc_get_port(sap)); + xprt->address_strings[RPC_DISPLAY_HEX_PORT] = kstrdup(buf, GFP_KERNEL); +} + +static void xs_format_peer_addresses(struct rpc_xprt *xprt, + const char *protocol, + const char *netid) +{ + xprt->address_strings[RPC_DISPLAY_PROTO] = protocol; + xprt->address_strings[RPC_DISPLAY_NETID] = netid; + xs_format_common_peer_addresses(xprt); + xs_format_common_peer_ports(xprt); +} + +static void xs_update_peer_port(struct rpc_xprt *xprt) +{ + kfree(xprt->address_strings[RPC_DISPLAY_HEX_PORT]); + kfree(xprt->address_strings[RPC_DISPLAY_PORT]); + + xs_format_common_peer_ports(xprt); +} + +static void xs_free_peer_addresses(struct rpc_xprt *xprt) +{ + unsigned int i; + + for (i = 0; i < RPC_DISPLAY_MAX; i++) + switch (i) { + case RPC_DISPLAY_PROTO: + case RPC_DISPLAY_NETID: + continue; + default: + kfree(xprt->address_strings[i]); + } +} + #define XS_SENDMSG_FLAGS (MSG_DONTWAIT | MSG_NOSIGNAL) -static inline int xs_send_head(struct socket *sock, struct sockaddr *addr, int addrlen, struct xdr_buf *xdr, unsigned int base, unsigned int len) +static int xs_send_kvec(struct socket *sock, struct sockaddr *addr, int addrlen, struct kvec *vec, unsigned int base, int more) { - struct kvec iov = { - .iov_base = xdr->head[0].iov_base + base, - .iov_len = len - base, - }; struct msghdr msg = { .msg_name = addr, .msg_namelen = addrlen, - .msg_flags = XS_SENDMSG_FLAGS, + .msg_flags = XS_SENDMSG_FLAGS | (more ? MSG_MORE : 0), + }; + struct kvec iov = { + .iov_base = vec->iov_base + base, + .iov_len = vec->iov_len - base, }; - if (xdr->len > len) - msg.msg_flags |= MSG_MORE; - - if (likely(iov.iov_len)) + if (iov.iov_len != 0) return kernel_sendmsg(sock, &msg, &iov, 1, iov.iov_len); return kernel_sendmsg(sock, &msg, NULL, 0, 0); } -static int xs_send_tail(struct socket *sock, struct xdr_buf *xdr, unsigned int base, unsigned int len) +static int xs_send_pagedata(struct socket *sock, struct xdr_buf *xdr, unsigned int base, int more, bool zerocopy) { - struct kvec iov = { - .iov_base = xdr->tail[0].iov_base + base, - .iov_len = len - base, - }; - struct msghdr msg = { - .msg_flags = XS_SENDMSG_FLAGS, - }; + ssize_t (*do_sendpage)(struct socket *sock, struct page *page, + int offset, size_t size, int flags); + struct page **ppage; + unsigned int remainder; + int err, sent = 0; + + remainder = xdr->page_len - base; + base += xdr->page_base; + ppage = xdr->pages + (base >> PAGE_SHIFT); + base &= ~PAGE_MASK; + do_sendpage = sock->ops->sendpage; + if (!zerocopy) + do_sendpage = sock_no_sendpage; + for(;;) { + unsigned int len = min_t(unsigned int, PAGE_SIZE - base, remainder); + int flags = XS_SENDMSG_FLAGS; - return kernel_sendmsg(sock, &msg, &iov, 1, iov.iov_len); + remainder -= len; + if (remainder != 0 || more) + flags |= MSG_MORE; + err = do_sendpage(sock, *ppage, base, len, flags); + if (remainder == 0 || err != len) + break; + sent += err; + ppage++; + base = 0; + } + if (sent == 0) + return err; + if (err > 0) + sent += err; + return sent; } /** @@ -166,83 +442,62 @@ static int xs_send_tail(struct socket *sock, struct xdr_buf *xdr, unsigned int b * @addrlen: UDP only -- length of destination address * @xdr: buffer containing this request * @base: starting position in the buffer + * @zerocopy: true if it is safe to use sendpage() * */ -static inline int xs_sendpages(struct socket *sock, struct sockaddr *addr, int addrlen, struct xdr_buf *xdr, unsigned int base) +static int xs_sendpages(struct socket *sock, struct sockaddr *addr, int addrlen, struct xdr_buf *xdr, unsigned int base, bool zerocopy) { - struct page **ppage = xdr->pages; - unsigned int len, pglen = xdr->page_len; - int err, ret = 0; - ssize_t (*sendpage)(struct socket *, struct page *, int, size_t, int); + unsigned int remainder = xdr->len - base; + int err, sent = 0; if (unlikely(!sock)) - return -ENOTCONN; + return -ENOTSOCK; clear_bit(SOCK_ASYNC_NOSPACE, &sock->flags); + if (base != 0) { + addr = NULL; + addrlen = 0; + } - len = xdr->head[0].iov_len; - if (base < len || (addr != NULL && base == 0)) { - err = xs_send_head(sock, addr, addrlen, xdr, base, len); - if (ret == 0) - ret = err; - else if (err > 0) - ret += err; - if (err != (len - base)) + if (base < xdr->head[0].iov_len || addr != NULL) { + unsigned int len = xdr->head[0].iov_len - base; + remainder -= len; + err = xs_send_kvec(sock, addr, addrlen, &xdr->head[0], base, remainder != 0); + if (remainder == 0 || err != len) goto out; + sent += err; base = 0; } else - base -= len; + base -= xdr->head[0].iov_len; - if (unlikely(pglen == 0)) - goto copy_tail; - if (unlikely(base >= pglen)) { - base -= pglen; - goto copy_tail; - } - if (base || xdr->page_base) { - pglen -= base; - base += xdr->page_base; - ppage += base >> PAGE_CACHE_SHIFT; - base &= ~PAGE_CACHE_MASK; - } - - sendpage = sock->ops->sendpage ? : sock_no_sendpage; - do { - int flags = XS_SENDMSG_FLAGS; - - len = PAGE_CACHE_SIZE; - if (base) - len -= base; - if (pglen < len) - len = pglen; - - if (pglen != len || xdr->tail[0].iov_len != 0) - flags |= MSG_MORE; - - /* Hmm... We might be dealing with highmem pages */ - if (PageHighMem(*ppage)) - sendpage = sock_no_sendpage; - err = sendpage(sock, *ppage, base, len, flags); - if (ret == 0) - ret = err; - else if (err > 0) - ret += err; - if (err != len) + if (base < xdr->page_len) { + unsigned int len = xdr->page_len - base; + remainder -= len; + err = xs_send_pagedata(sock, xdr, base, remainder != 0, zerocopy); + if (remainder == 0 || err != len) goto out; + sent += err; base = 0; - ppage++; - } while ((pglen -= len) != 0); -copy_tail: - len = xdr->tail[0].iov_len; - if (base < len) { - err = xs_send_tail(sock, xdr, base, len); - if (ret == 0) - ret = err; - else if (err > 0) - ret += err; - } + } else + base -= xdr->page_len; + + if (base >= xdr->tail[0].iov_len) + return sent; + err = xs_send_kvec(sock, NULL, 0, &xdr->tail[0], base, 0); out: - return ret; + if (sent == 0) + return err; + if (err > 0) + sent += err; + return sent; +} + +static void xs_nospace_callback(struct rpc_task *task) +{ + struct sock_xprt *transport = container_of(task->tk_rqstp->rq_xprt, struct sock_xprt, xprt); + + transport->inet->sk_write_pending--; + clear_bit(SOCK_ASYNC_NOSPACE, &transport->sock->flags); } /** @@ -250,29 +505,107 @@ out: * @task: task to put to sleep * */ -static void xs_nospace(struct rpc_task *task) +static int xs_nospace(struct rpc_task *task) { struct rpc_rqst *req = task->tk_rqstp; struct rpc_xprt *xprt = req->rq_xprt; + struct sock_xprt *transport = container_of(xprt, struct sock_xprt, xprt); + struct sock *sk = transport->inet; + int ret = -EAGAIN; - dprintk("RPC: %4d xmit incomplete (%u left of %u)\n", + dprintk("RPC: %5u xmit incomplete (%u left of %u)\n", task->tk_pid, req->rq_slen - req->rq_bytes_sent, req->rq_slen); - if (test_bit(SOCK_ASYNC_NOSPACE, &xprt->sock->flags)) { - /* Protect against races with write_space */ - spin_lock_bh(&xprt->transport_lock); + /* Protect against races with write_space */ + spin_lock_bh(&xprt->transport_lock); + + /* Don't race with disconnect */ + if (xprt_connected(xprt)) { + if (test_bit(SOCK_ASYNC_NOSPACE, &transport->sock->flags)) { + /* + * Notify TCP that we're limited by the application + * window size + */ + set_bit(SOCK_NOSPACE, &transport->sock->flags); + sk->sk_write_pending++; + /* ...and wait for more buffer space */ + xprt_wait_for_buffer_space(task, xs_nospace_callback); + } + } else { + clear_bit(SOCK_ASYNC_NOSPACE, &transport->sock->flags); + ret = -ENOTCONN; + } - /* Don't race with disconnect */ - if (!xprt_connected(xprt)) - task->tk_status = -ENOTCONN; - else if (test_bit(SOCK_NOSPACE, &xprt->sock->flags)) - xprt_wait_for_buffer_space(task); + spin_unlock_bh(&xprt->transport_lock); - spin_unlock_bh(&xprt->transport_lock); - } else - /* Keep holding the socket if it is blocked */ - rpc_delay(task, HZ>>4); + /* Race breaker in case memory is freed before above code is called */ + sk->sk_write_space(sk); + return ret; +} + +/* + * Construct a stream transport record marker in @buf. + */ +static inline void xs_encode_stream_record_marker(struct xdr_buf *buf) +{ + u32 reclen = buf->len - sizeof(rpc_fraghdr); + rpc_fraghdr *base = buf->head[0].iov_base; + *base = cpu_to_be32(RPC_LAST_STREAM_FRAGMENT | reclen); +} + +/** + * xs_local_send_request - write an RPC request to an AF_LOCAL socket + * @task: RPC task that manages the state of an RPC request + * + * Return values: + * 0: The request has been sent + * EAGAIN: The socket was blocked, please call again later to + * complete the request + * ENOTCONN: Caller needs to invoke connect logic then call again + * other: Some other error occured, the request was not sent + */ +static int xs_local_send_request(struct rpc_task *task) +{ + struct rpc_rqst *req = task->tk_rqstp; + struct rpc_xprt *xprt = req->rq_xprt; + struct sock_xprt *transport = + container_of(xprt, struct sock_xprt, xprt); + struct xdr_buf *xdr = &req->rq_snd_buf; + int status; + + xs_encode_stream_record_marker(&req->rq_snd_buf); + + xs_pktdump("packet data:", + req->rq_svec->iov_base, req->rq_svec->iov_len); + + status = xs_sendpages(transport->sock, NULL, 0, + xdr, req->rq_bytes_sent, true); + dprintk("RPC: %s(%u) = %d\n", + __func__, xdr->len - req->rq_bytes_sent, status); + if (likely(status >= 0)) { + req->rq_bytes_sent += status; + req->rq_xmit_bytes_sent += status; + if (likely(req->rq_bytes_sent >= req->rq_slen)) { + req->rq_bytes_sent = 0; + return 0; + } + status = -EAGAIN; + } + + switch (status) { + case -EAGAIN: + status = xs_nospace(task); + break; + default: + dprintk("RPC: sendmsg returned unrecognized error %d\n", + -status); + case -EPIPE: + xs_close(xprt); + status = -ENOTCONN; + } + + return status; } /** @@ -284,12 +617,13 @@ static void xs_nospace(struct rpc_task *task) * EAGAIN: The socket was blocked, please call again later to * complete the request * ENOTCONN: Caller needs to invoke connect logic then call again - * other: Some other error occured, the request was not sent + * other: Some other error occurred, the request was not sent */ static int xs_udp_send_request(struct rpc_task *task) { struct rpc_rqst *req = task->tk_rqstp; struct rpc_xprt *xprt = req->rq_xprt; + struct sock_xprt *transport = container_of(xprt, struct sock_xprt, xprt); struct xdr_buf *xdr = &req->rq_snd_buf; int status; @@ -297,44 +631,62 @@ static int xs_udp_send_request(struct rpc_task *task) req->rq_svec->iov_base, req->rq_svec->iov_len); - req->rq_xtime = jiffies; - status = xs_sendpages(xprt->sock, (struct sockaddr *) &xprt->addr, - sizeof(xprt->addr), xdr, req->rq_bytes_sent); + if (!xprt_bound(xprt)) + return -ENOTCONN; + status = xs_sendpages(transport->sock, + xs_addr(xprt), + xprt->addrlen, xdr, + req->rq_bytes_sent, true); - dprintk("RPC: xs_udp_send_request(%u) = %d\n", + dprintk("RPC: xs_udp_send_request(%u) = %d\n", xdr->len - req->rq_bytes_sent, status); - if (likely(status >= (int) req->rq_slen)) - return 0; - - /* Still some bytes left; set up for a retry later. */ - if (status > 0) + if (status >= 0) { + req->rq_xmit_bytes_sent += status; + if (status >= req->rq_slen) + return 0; + /* Still some bytes left; set up for a retry later. */ status = -EAGAIN; + } switch (status) { - case -ENETUNREACH: - case -EPIPE: - case -ECONNREFUSED: - /* When the server has died, an ICMP port unreachable message - * prompts ECONNREFUSED. */ + case -ENOTSOCK: + status = -ENOTCONN; + /* Should we call xs_close() here? */ break; case -EAGAIN: - xs_nospace(task); + status = xs_nospace(task); break; default: - dprintk("RPC: sendmsg returned unrecognized error %d\n", + dprintk("RPC: sendmsg returned unrecognized error %d\n", -status); - break; + case -ENETUNREACH: + case -EPIPE: + case -ECONNREFUSED: + /* When the server has died, an ICMP port unreachable message + * prompts ECONNREFUSED. */ + clear_bit(SOCK_ASYNC_NOSPACE, &transport->sock->flags); } return status; } -static inline void xs_encode_tcp_record_marker(struct xdr_buf *buf) +/** + * xs_tcp_shutdown - gracefully shut down a TCP socket + * @xprt: transport + * + * Initiates a graceful shutdown of the TCP socket by calling the + * equivalent of shutdown(SHUT_WR); + */ +static void xs_tcp_shutdown(struct rpc_xprt *xprt) { - u32 reclen = buf->len - sizeof(rpc_fraghdr); - rpc_fraghdr *base = buf->head[0].iov_base; - *base = htonl(RPC_LAST_STREAM_FRAGMENT | reclen); + struct sock_xprt *transport = container_of(xprt, struct sock_xprt, xprt); + struct socket *sock = transport->sock; + + if (sock != NULL) { + kernel_sock_shutdown(sock, SHUT_WR); + trace_rpc_socket_shutdown(xprt, sock); + } } /** @@ -346,7 +698,7 @@ static inline void xs_encode_tcp_record_marker(struct xdr_buf *buf) * EAGAIN: The socket was blocked, please call again later to * complete the request * ENOTCONN: Caller needs to invoke connect logic then call again - * other: Some other error occured, the request was not sent + * other: Some other error occurred, the request was not sent * * XXX: In the case of soft timeouts, should we eventually give up * if sendmsg is not able to make progress? @@ -355,24 +707,32 @@ static int xs_tcp_send_request(struct rpc_task *task) { struct rpc_rqst *req = task->tk_rqstp; struct rpc_xprt *xprt = req->rq_xprt; + struct sock_xprt *transport = container_of(xprt, struct sock_xprt, xprt); struct xdr_buf *xdr = &req->rq_snd_buf; - int status, retry = 0; + bool zerocopy = true; + int status; - xs_encode_tcp_record_marker(&req->rq_snd_buf); + xs_encode_stream_record_marker(&req->rq_snd_buf); xs_pktdump("packet data:", req->rq_svec->iov_base, req->rq_svec->iov_len); + /* Don't use zero copy if this is a resend. If the RPC call + * completes while the socket holds a reference to the pages, + * then we may end up resending corrupted data. + */ + if (task->tk_flags & RPC_TASK_SENT) + zerocopy = false; /* Continue transmitting the packet/record. We must be careful * to cope with writespace callbacks arriving _after_ we have * called sendmsg(). */ while (1) { - req->rq_xtime = jiffies; - status = xs_sendpages(xprt->sock, NULL, 0, xdr, - req->rq_bytes_sent); + status = xs_sendpages(transport->sock, + NULL, 0, xdr, req->rq_bytes_sent, + zerocopy); - dprintk("RPC: xs_tcp_send_request(%u) = %d\n", + dprintk("RPC: xs_tcp_send_request(%u) = %d\n", xdr->len - req->rq_bytes_sent, status); if (unlikely(status < 0)) @@ -381,88 +741,269 @@ static int xs_tcp_send_request(struct rpc_task *task) /* If we've sent the entire packet, immediately * reset the count of bytes sent. */ req->rq_bytes_sent += status; + req->rq_xmit_bytes_sent += status; if (likely(req->rq_bytes_sent >= req->rq_slen)) { req->rq_bytes_sent = 0; return 0; } + if (status != 0) + continue; status = -EAGAIN; - if (retry++ > XS_SENDMSG_RETRY) - break; + break; } switch (status) { + case -ENOTSOCK: + status = -ENOTCONN; + /* Should we call xs_close() here? */ + break; case -EAGAIN: - xs_nospace(task); + status = xs_nospace(task); break; - case -ECONNREFUSED: + default: + dprintk("RPC: sendmsg returned unrecognized error %d\n", + -status); case -ECONNRESET: + xs_tcp_shutdown(xprt); + case -ECONNREFUSED: case -ENOTCONN: case -EPIPE: - status = -ENOTCONN; - break; - default: - dprintk("RPC: sendmsg returned unrecognized error %d\n", - -status); - xprt_disconnect(xprt); - break; + clear_bit(SOCK_ASYNC_NOSPACE, &transport->sock->flags); } return status; } /** - * xs_close - close a socket + * xs_tcp_release_xprt - clean up after a tcp transmission * @xprt: transport + * @task: rpc task * - * This is used when all requests are complete; ie, no DRC state remains - * on the server we want to save. + * This cleans up if an error causes us to abort the transmission of a request. + * In this case, the socket may need to be reset in order to avoid confusing + * the server. */ -static void xs_close(struct rpc_xprt *xprt) +static void xs_tcp_release_xprt(struct rpc_xprt *xprt, struct rpc_task *task) +{ + struct rpc_rqst *req; + + if (task != xprt->snd_task) + return; + if (task == NULL) + goto out_release; + req = task->tk_rqstp; + if (req == NULL) + goto out_release; + if (req->rq_bytes_sent == 0) + goto out_release; + if (req->rq_bytes_sent == req->rq_snd_buf.len) + goto out_release; + set_bit(XPRT_CLOSE_WAIT, &xprt->state); +out_release: + xprt_release_xprt(xprt, task); +} + +static void xs_save_old_callbacks(struct sock_xprt *transport, struct sock *sk) +{ + transport->old_data_ready = sk->sk_data_ready; + transport->old_state_change = sk->sk_state_change; + transport->old_write_space = sk->sk_write_space; + transport->old_error_report = sk->sk_error_report; +} + +static void xs_restore_old_callbacks(struct sock_xprt *transport, struct sock *sk) +{ + sk->sk_data_ready = transport->old_data_ready; + sk->sk_state_change = transport->old_state_change; + sk->sk_write_space = transport->old_write_space; + sk->sk_error_report = transport->old_error_report; +} + +/** + * xs_error_report - callback to handle TCP socket state errors + * @sk: socket + * + * Note: we don't call sock_error() since there may be a rpc_task + * using the socket, and so we don't want to clear sk->sk_err. + */ +static void xs_error_report(struct sock *sk) { - struct socket *sock = xprt->sock; - struct sock *sk = xprt->inet; + struct rpc_xprt *xprt; + int err; + + read_lock_bh(&sk->sk_callback_lock); + if (!(xprt = xprt_from_sock(sk))) + goto out; - if (!sk) + err = -sk->sk_err; + if (err == 0) + goto out; + dprintk("RPC: xs_error_report client %p, error=%d...\n", + xprt, -err); + trace_rpc_socket_error(xprt, sk->sk_socket, err); + xprt_wake_pending_tasks(xprt, err); + out: + read_unlock_bh(&sk->sk_callback_lock); +} + +static void xs_reset_transport(struct sock_xprt *transport) +{ + struct socket *sock = transport->sock; + struct sock *sk = transport->inet; + + if (sk == NULL) return; - dprintk("RPC: xs_close xprt %p\n", xprt); + transport->srcport = 0; write_lock_bh(&sk->sk_callback_lock); - xprt->inet = NULL; - xprt->sock = NULL; + transport->inet = NULL; + transport->sock = NULL; sk->sk_user_data = NULL; - sk->sk_data_ready = xprt->old_data_ready; - sk->sk_state_change = xprt->old_state_change; - sk->sk_write_space = xprt->old_write_space; - write_unlock_bh(&sk->sk_callback_lock); - sk->sk_no_check = 0; + xs_restore_old_callbacks(transport, sk); + write_unlock_bh(&sk->sk_callback_lock); + trace_rpc_socket_close(&transport->xprt, sock); sock_release(sock); } /** + * xs_close - close a socket + * @xprt: transport + * + * This is used when all requests are complete; ie, no DRC state remains + * on the server we want to save. + * + * The caller _must_ be holding XPRT_LOCKED in order to avoid issues with + * xs_reset_transport() zeroing the socket from underneath a writer. + */ +static void xs_close(struct rpc_xprt *xprt) +{ + struct sock_xprt *transport = container_of(xprt, struct sock_xprt, xprt); + + dprintk("RPC: xs_close xprt %p\n", xprt); + + cancel_delayed_work_sync(&transport->connect_worker); + + xs_reset_transport(transport); + xprt->reestablish_timeout = 0; + + smp_mb__before_atomic(); + clear_bit(XPRT_CONNECTION_ABORT, &xprt->state); + clear_bit(XPRT_CLOSE_WAIT, &xprt->state); + clear_bit(XPRT_CLOSING, &xprt->state); + smp_mb__after_atomic(); + xprt_disconnect_done(xprt); +} + +static void xs_tcp_close(struct rpc_xprt *xprt) +{ + if (test_and_clear_bit(XPRT_CONNECTION_CLOSE, &xprt->state)) + xs_close(xprt); + else + xs_tcp_shutdown(xprt); +} + +static void xs_xprt_free(struct rpc_xprt *xprt) +{ + xs_free_peer_addresses(xprt); + xprt_free(xprt); +} + +/** * xs_destroy - prepare to shutdown a transport * @xprt: doomed transport * */ static void xs_destroy(struct rpc_xprt *xprt) { - dprintk("RPC: xs_destroy xprt %p\n", xprt); + dprintk("RPC: xs_destroy xprt %p\n", xprt); - cancel_delayed_work(&xprt->connect_worker); - flush_scheduled_work(); - - xprt_disconnect(xprt); xs_close(xprt); - kfree(xprt->slot); + xs_xprt_free(xprt); + module_put(THIS_MODULE); } -static inline struct rpc_xprt *xprt_from_sock(struct sock *sk) +static int xs_local_copy_to_xdr(struct xdr_buf *xdr, struct sk_buff *skb) { - return (struct rpc_xprt *) sk->sk_user_data; + struct xdr_skb_reader desc = { + .skb = skb, + .offset = sizeof(rpc_fraghdr), + .count = skb->len - sizeof(rpc_fraghdr), + }; + + if (xdr_partial_copy_from_skb(xdr, 0, &desc, xdr_skb_read_bits) < 0) + return -1; + if (desc.count) + return -1; + return 0; +} + +/** + * xs_local_data_ready - "data ready" callback for AF_LOCAL sockets + * @sk: socket with data to read + * @len: how much data to read + * + * Currently this assumes we can read the whole reply in a single gulp. + */ +static void xs_local_data_ready(struct sock *sk) +{ + struct rpc_task *task; + struct rpc_xprt *xprt; + struct rpc_rqst *rovr; + struct sk_buff *skb; + int err, repsize, copied; + u32 _xid; + __be32 *xp; + + read_lock_bh(&sk->sk_callback_lock); + dprintk("RPC: %s...\n", __func__); + xprt = xprt_from_sock(sk); + if (xprt == NULL) + goto out; + + skb = skb_recv_datagram(sk, 0, 1, &err); + if (skb == NULL) + goto out; + + repsize = skb->len - sizeof(rpc_fraghdr); + if (repsize < 4) { + dprintk("RPC: impossible RPC reply size %d\n", repsize); + goto dropit; + } + + /* Copy the XID from the skb... */ + xp = skb_header_pointer(skb, sizeof(rpc_fraghdr), sizeof(_xid), &_xid); + if (xp == NULL) + goto dropit; + + /* Look up and lock the request corresponding to the given XID */ + spin_lock(&xprt->transport_lock); + rovr = xprt_lookup_rqst(xprt, *xp); + if (!rovr) + goto out_unlock; + task = rovr->rq_task; + + copied = rovr->rq_private_buf.buflen; + if (copied > repsize) + copied = repsize; + + if (xs_local_copy_to_xdr(&rovr->rq_private_buf, skb)) { + dprintk("RPC: sk_buff copy failed\n"); + goto out_unlock; + } + + xprt_complete_rqst(task, copied); + + out_unlock: + spin_unlock(&xprt->transport_lock); + dropit: + skb_free_datagram(sk, skb); + out: + read_unlock_bh(&sk->sk_callback_lock); } /** @@ -471,29 +1012,27 @@ static inline struct rpc_xprt *xprt_from_sock(struct sock *sk) * @len: how much data to read * */ -static void xs_udp_data_ready(struct sock *sk, int len) +static void xs_udp_data_ready(struct sock *sk) { struct rpc_task *task; struct rpc_xprt *xprt; struct rpc_rqst *rovr; struct sk_buff *skb; int err, repsize, copied; - u32 _xid, *xp; + u32 _xid; + __be32 *xp; - read_lock(&sk->sk_callback_lock); - dprintk("RPC: xs_udp_data_ready...\n"); + read_lock_bh(&sk->sk_callback_lock); + dprintk("RPC: xs_udp_data_ready...\n"); if (!(xprt = xprt_from_sock(sk))) goto out; if ((skb = skb_recv_datagram(sk, 0, 1, &err)) == NULL) goto out; - if (xprt->shutdown) - goto dropit; - repsize = skb->len - sizeof(struct udphdr); if (repsize < 4) { - dprintk("RPC: impossible RPC reply size %d!\n", repsize); + dprintk("RPC: impossible RPC reply size %d!\n", repsize); goto dropit; } @@ -514,14 +1053,14 @@ static void xs_udp_data_ready(struct sock *sk, int len) copied = repsize; /* Suck it into the iovec, verify checksum if not done by hw. */ - if (csum_partial_copy_to_xdr(&rovr->rq_private_buf, skb)) + if (csum_partial_copy_to_xdr(&rovr->rq_private_buf, skb)) { + UDPX_INC_STATS_BH(sk, UDP_MIB_INERRORS); goto out_unlock; + } - /* Something worked... */ - dst_confirm(skb->dst); + UDPX_INC_STATS_BH(sk, UDP_MIB_INDATAGRAMS); - xprt_adjust_cwnd(task, copied); - xprt_update_rtt(task); + xprt_adjust_cwnd(xprt, task, copied); xprt_complete_rqst(task, copied); out_unlock: @@ -529,213 +1068,364 @@ static void xs_udp_data_ready(struct sock *sk, int len) dropit: skb_free_datagram(sk, skb); out: - read_unlock(&sk->sk_callback_lock); + read_unlock_bh(&sk->sk_callback_lock); } -static inline size_t xs_tcp_copy_data(skb_reader_t *desc, void *p, size_t len) +/* + * Helper function to force a TCP close if the server is sending + * junk and/or it has put us in CLOSE_WAIT + */ +static void xs_tcp_force_close(struct rpc_xprt *xprt) { - if (len > desc->count) - len = desc->count; - if (skb_copy_bits(desc->skb, desc->offset, p, len)) { - dprintk("RPC: failed to copy %zu bytes from skb. %zu bytes remain\n", - len, desc->count); - return 0; - } - desc->offset += len; - desc->count -= len; - dprintk("RPC: copied %zu bytes from skb. %zu bytes remain\n", - len, desc->count); - return len; + set_bit(XPRT_CONNECTION_CLOSE, &xprt->state); + xprt_force_disconnect(xprt); } -static inline void xs_tcp_read_fraghdr(struct rpc_xprt *xprt, skb_reader_t *desc) +static inline void xs_tcp_read_fraghdr(struct rpc_xprt *xprt, struct xdr_skb_reader *desc) { + struct sock_xprt *transport = container_of(xprt, struct sock_xprt, xprt); size_t len, used; char *p; - p = ((char *) &xprt->tcp_recm) + xprt->tcp_offset; - len = sizeof(xprt->tcp_recm) - xprt->tcp_offset; - used = xs_tcp_copy_data(desc, p, len); - xprt->tcp_offset += used; + p = ((char *) &transport->tcp_fraghdr) + transport->tcp_offset; + len = sizeof(transport->tcp_fraghdr) - transport->tcp_offset; + used = xdr_skb_read_bits(desc, p, len); + transport->tcp_offset += used; if (used != len) return; - xprt->tcp_reclen = ntohl(xprt->tcp_recm); - if (xprt->tcp_reclen & RPC_LAST_STREAM_FRAGMENT) - xprt->tcp_flags |= XPRT_LAST_FRAG; + transport->tcp_reclen = ntohl(transport->tcp_fraghdr); + if (transport->tcp_reclen & RPC_LAST_STREAM_FRAGMENT) + transport->tcp_flags |= TCP_RCV_LAST_FRAG; else - xprt->tcp_flags &= ~XPRT_LAST_FRAG; - xprt->tcp_reclen &= RPC_FRAGMENT_SIZE_MASK; + transport->tcp_flags &= ~TCP_RCV_LAST_FRAG; + transport->tcp_reclen &= RPC_FRAGMENT_SIZE_MASK; - xprt->tcp_flags &= ~XPRT_COPY_RECM; - xprt->tcp_offset = 0; + transport->tcp_flags &= ~TCP_RCV_COPY_FRAGHDR; + transport->tcp_offset = 0; /* Sanity check of the record length */ - if (unlikely(xprt->tcp_reclen < 4)) { - dprintk("RPC: invalid TCP record fragment length\n"); - xprt_disconnect(xprt); + if (unlikely(transport->tcp_reclen < 8)) { + dprintk("RPC: invalid TCP record fragment length\n"); + xs_tcp_force_close(xprt); return; } - dprintk("RPC: reading TCP record fragment of length %d\n", - xprt->tcp_reclen); + dprintk("RPC: reading TCP record fragment of length %d\n", + transport->tcp_reclen); } -static void xs_tcp_check_recm(struct rpc_xprt *xprt) +static void xs_tcp_check_fraghdr(struct sock_xprt *transport) { - dprintk("RPC: xprt = %p, tcp_copied = %lu, tcp_offset = %u, tcp_reclen = %u, tcp_flags = %lx\n", - xprt, xprt->tcp_copied, xprt->tcp_offset, xprt->tcp_reclen, xprt->tcp_flags); - if (xprt->tcp_offset == xprt->tcp_reclen) { - xprt->tcp_flags |= XPRT_COPY_RECM; - xprt->tcp_offset = 0; - if (xprt->tcp_flags & XPRT_LAST_FRAG) { - xprt->tcp_flags &= ~XPRT_COPY_DATA; - xprt->tcp_flags |= XPRT_COPY_XID; - xprt->tcp_copied = 0; + if (transport->tcp_offset == transport->tcp_reclen) { + transport->tcp_flags |= TCP_RCV_COPY_FRAGHDR; + transport->tcp_offset = 0; + if (transport->tcp_flags & TCP_RCV_LAST_FRAG) { + transport->tcp_flags &= ~TCP_RCV_COPY_DATA; + transport->tcp_flags |= TCP_RCV_COPY_XID; + transport->tcp_copied = 0; } } } -static inline void xs_tcp_read_xid(struct rpc_xprt *xprt, skb_reader_t *desc) +static inline void xs_tcp_read_xid(struct sock_xprt *transport, struct xdr_skb_reader *desc) { size_t len, used; char *p; - len = sizeof(xprt->tcp_xid) - xprt->tcp_offset; - dprintk("RPC: reading XID (%Zu bytes)\n", len); - p = ((char *) &xprt->tcp_xid) + xprt->tcp_offset; - used = xs_tcp_copy_data(desc, p, len); - xprt->tcp_offset += used; + len = sizeof(transport->tcp_xid) - transport->tcp_offset; + dprintk("RPC: reading XID (%Zu bytes)\n", len); + p = ((char *) &transport->tcp_xid) + transport->tcp_offset; + used = xdr_skb_read_bits(desc, p, len); + transport->tcp_offset += used; if (used != len) return; - xprt->tcp_flags &= ~XPRT_COPY_XID; - xprt->tcp_flags |= XPRT_COPY_DATA; - xprt->tcp_copied = 4; - dprintk("RPC: reading reply for XID %08x\n", - ntohl(xprt->tcp_xid)); - xs_tcp_check_recm(xprt); + transport->tcp_flags &= ~TCP_RCV_COPY_XID; + transport->tcp_flags |= TCP_RCV_READ_CALLDIR; + transport->tcp_copied = 4; + dprintk("RPC: reading %s XID %08x\n", + (transport->tcp_flags & TCP_RPC_REPLY) ? "reply for" + : "request with", + ntohl(transport->tcp_xid)); + xs_tcp_check_fraghdr(transport); } -static inline void xs_tcp_read_request(struct rpc_xprt *xprt, skb_reader_t *desc) +static inline void xs_tcp_read_calldir(struct sock_xprt *transport, + struct xdr_skb_reader *desc) { - struct rpc_rqst *req; + size_t len, used; + u32 offset; + char *p; + + /* + * We want transport->tcp_offset to be 8 at the end of this routine + * (4 bytes for the xid and 4 bytes for the call/reply flag). + * When this function is called for the first time, + * transport->tcp_offset is 4 (after having already read the xid). + */ + offset = transport->tcp_offset - sizeof(transport->tcp_xid); + len = sizeof(transport->tcp_calldir) - offset; + dprintk("RPC: reading CALL/REPLY flag (%Zu bytes)\n", len); + p = ((char *) &transport->tcp_calldir) + offset; + used = xdr_skb_read_bits(desc, p, len); + transport->tcp_offset += used; + if (used != len) + return; + transport->tcp_flags &= ~TCP_RCV_READ_CALLDIR; + /* + * We don't yet have the XDR buffer, so we will write the calldir + * out after we get the buffer from the 'struct rpc_rqst' + */ + switch (ntohl(transport->tcp_calldir)) { + case RPC_REPLY: + transport->tcp_flags |= TCP_RCV_COPY_CALLDIR; + transport->tcp_flags |= TCP_RCV_COPY_DATA; + transport->tcp_flags |= TCP_RPC_REPLY; + break; + case RPC_CALL: + transport->tcp_flags |= TCP_RCV_COPY_CALLDIR; + transport->tcp_flags |= TCP_RCV_COPY_DATA; + transport->tcp_flags &= ~TCP_RPC_REPLY; + break; + default: + dprintk("RPC: invalid request message type\n"); + xs_tcp_force_close(&transport->xprt); + } + xs_tcp_check_fraghdr(transport); +} + +static inline void xs_tcp_read_common(struct rpc_xprt *xprt, + struct xdr_skb_reader *desc, + struct rpc_rqst *req) +{ + struct sock_xprt *transport = + container_of(xprt, struct sock_xprt, xprt); struct xdr_buf *rcvbuf; size_t len; ssize_t r; - /* Find and lock the request corresponding to this xid */ - spin_lock(&xprt->transport_lock); - req = xprt_lookup_rqst(xprt, xprt->tcp_xid); - if (!req) { - xprt->tcp_flags &= ~XPRT_COPY_DATA; - dprintk("RPC: XID %08x request not found!\n", - ntohl(xprt->tcp_xid)); - spin_unlock(&xprt->transport_lock); - return; + rcvbuf = &req->rq_private_buf; + + if (transport->tcp_flags & TCP_RCV_COPY_CALLDIR) { + /* + * Save the RPC direction in the XDR buffer + */ + memcpy(rcvbuf->head[0].iov_base + transport->tcp_copied, + &transport->tcp_calldir, + sizeof(transport->tcp_calldir)); + transport->tcp_copied += sizeof(transport->tcp_calldir); + transport->tcp_flags &= ~TCP_RCV_COPY_CALLDIR; } - rcvbuf = &req->rq_private_buf; len = desc->count; - if (len > xprt->tcp_reclen - xprt->tcp_offset) { - skb_reader_t my_desc; + if (len > transport->tcp_reclen - transport->tcp_offset) { + struct xdr_skb_reader my_desc; - len = xprt->tcp_reclen - xprt->tcp_offset; + len = transport->tcp_reclen - transport->tcp_offset; memcpy(&my_desc, desc, sizeof(my_desc)); my_desc.count = len; - r = xdr_partial_copy_from_skb(rcvbuf, xprt->tcp_copied, - &my_desc, xs_tcp_copy_data); + r = xdr_partial_copy_from_skb(rcvbuf, transport->tcp_copied, + &my_desc, xdr_skb_read_bits); desc->count -= r; desc->offset += r; } else - r = xdr_partial_copy_from_skb(rcvbuf, xprt->tcp_copied, - desc, xs_tcp_copy_data); + r = xdr_partial_copy_from_skb(rcvbuf, transport->tcp_copied, + desc, xdr_skb_read_bits); if (r > 0) { - xprt->tcp_copied += r; - xprt->tcp_offset += r; + transport->tcp_copied += r; + transport->tcp_offset += r; } if (r != len) { /* Error when copying to the receive buffer, * usually because we weren't able to allocate * additional buffer pages. All we can do now - * is turn off XPRT_COPY_DATA, so the request + * is turn off TCP_RCV_COPY_DATA, so the request * will not receive any additional updates, * and time out. * Any remaining data from this record will * be discarded. */ - xprt->tcp_flags &= ~XPRT_COPY_DATA; - dprintk("RPC: XID %08x truncated request\n", - ntohl(xprt->tcp_xid)); - dprintk("RPC: xprt = %p, tcp_copied = %lu, tcp_offset = %u, tcp_reclen = %u\n", - xprt, xprt->tcp_copied, xprt->tcp_offset, xprt->tcp_reclen); - goto out; + transport->tcp_flags &= ~TCP_RCV_COPY_DATA; + dprintk("RPC: XID %08x truncated request\n", + ntohl(transport->tcp_xid)); + dprintk("RPC: xprt = %p, tcp_copied = %lu, " + "tcp_offset = %u, tcp_reclen = %u\n", + xprt, transport->tcp_copied, + transport->tcp_offset, transport->tcp_reclen); + return; } - dprintk("RPC: XID %08x read %Zd bytes\n", - ntohl(xprt->tcp_xid), r); - dprintk("RPC: xprt = %p, tcp_copied = %lu, tcp_offset = %u, tcp_reclen = %u\n", - xprt, xprt->tcp_copied, xprt->tcp_offset, xprt->tcp_reclen); + dprintk("RPC: XID %08x read %Zd bytes\n", + ntohl(transport->tcp_xid), r); + dprintk("RPC: xprt = %p, tcp_copied = %lu, tcp_offset = %u, " + "tcp_reclen = %u\n", xprt, transport->tcp_copied, + transport->tcp_offset, transport->tcp_reclen); + + if (transport->tcp_copied == req->rq_private_buf.buflen) + transport->tcp_flags &= ~TCP_RCV_COPY_DATA; + else if (transport->tcp_offset == transport->tcp_reclen) { + if (transport->tcp_flags & TCP_RCV_LAST_FRAG) + transport->tcp_flags &= ~TCP_RCV_COPY_DATA; + } +} - if (xprt->tcp_copied == req->rq_private_buf.buflen) - xprt->tcp_flags &= ~XPRT_COPY_DATA; - else if (xprt->tcp_offset == xprt->tcp_reclen) { - if (xprt->tcp_flags & XPRT_LAST_FRAG) - xprt->tcp_flags &= ~XPRT_COPY_DATA; +/* + * Finds the request corresponding to the RPC xid and invokes the common + * tcp read code to read the data. + */ +static inline int xs_tcp_read_reply(struct rpc_xprt *xprt, + struct xdr_skb_reader *desc) +{ + struct sock_xprt *transport = + container_of(xprt, struct sock_xprt, xprt); + struct rpc_rqst *req; + + dprintk("RPC: read reply XID %08x\n", ntohl(transport->tcp_xid)); + + /* Find and lock the request corresponding to this xid */ + spin_lock(&xprt->transport_lock); + req = xprt_lookup_rqst(xprt, transport->tcp_xid); + if (!req) { + dprintk("RPC: XID %08x request not found!\n", + ntohl(transport->tcp_xid)); + spin_unlock(&xprt->transport_lock); + return -1; } -out: - if (!(xprt->tcp_flags & XPRT_COPY_DATA)) - xprt_complete_rqst(req->rq_task, xprt->tcp_copied); + xs_tcp_read_common(xprt, desc, req); + + if (!(transport->tcp_flags & TCP_RCV_COPY_DATA)) + xprt_complete_rqst(req->rq_task, transport->tcp_copied); + + spin_unlock(&xprt->transport_lock); + return 0; +} + +#if defined(CONFIG_SUNRPC_BACKCHANNEL) +/* + * Obtains an rpc_rqst previously allocated and invokes the common + * tcp read code to read the data. The result is placed in the callback + * queue. + * If we're unable to obtain the rpc_rqst we schedule the closing of the + * connection and return -1. + */ +static int xs_tcp_read_callback(struct rpc_xprt *xprt, + struct xdr_skb_reader *desc) +{ + struct sock_xprt *transport = + container_of(xprt, struct sock_xprt, xprt); + struct rpc_rqst *req; + + /* Look up and lock the request corresponding to the given XID */ + spin_lock(&xprt->transport_lock); + req = xprt_lookup_bc_request(xprt, transport->tcp_xid); + if (req == NULL) { + spin_unlock(&xprt->transport_lock); + printk(KERN_WARNING "Callback slot table overflowed\n"); + xprt_force_disconnect(xprt); + return -1; + } + + dprintk("RPC: read callback XID %08x\n", ntohl(req->rq_xid)); + xs_tcp_read_common(xprt, desc, req); + + if (!(transport->tcp_flags & TCP_RCV_COPY_DATA)) + xprt_complete_bc_request(req, transport->tcp_copied); spin_unlock(&xprt->transport_lock); - xs_tcp_check_recm(xprt); + + return 0; } -static inline void xs_tcp_read_discard(struct rpc_xprt *xprt, skb_reader_t *desc) +static inline int _xs_tcp_read_data(struct rpc_xprt *xprt, + struct xdr_skb_reader *desc) +{ + struct sock_xprt *transport = + container_of(xprt, struct sock_xprt, xprt); + + return (transport->tcp_flags & TCP_RPC_REPLY) ? + xs_tcp_read_reply(xprt, desc) : + xs_tcp_read_callback(xprt, desc); +} +#else +static inline int _xs_tcp_read_data(struct rpc_xprt *xprt, + struct xdr_skb_reader *desc) +{ + return xs_tcp_read_reply(xprt, desc); +} +#endif /* CONFIG_SUNRPC_BACKCHANNEL */ + +/* + * Read data off the transport. This can be either an RPC_CALL or an + * RPC_REPLY. Relay the processing to helper functions. + */ +static void xs_tcp_read_data(struct rpc_xprt *xprt, + struct xdr_skb_reader *desc) +{ + struct sock_xprt *transport = + container_of(xprt, struct sock_xprt, xprt); + + if (_xs_tcp_read_data(xprt, desc) == 0) + xs_tcp_check_fraghdr(transport); + else { + /* + * The transport_lock protects the request handling. + * There's no need to hold it to update the tcp_flags. + */ + transport->tcp_flags &= ~TCP_RCV_COPY_DATA; + } +} + +static inline void xs_tcp_read_discard(struct sock_xprt *transport, struct xdr_skb_reader *desc) { size_t len; - len = xprt->tcp_reclen - xprt->tcp_offset; + len = transport->tcp_reclen - transport->tcp_offset; if (len > desc->count) len = desc->count; desc->count -= len; desc->offset += len; - xprt->tcp_offset += len; - dprintk("RPC: discarded %Zu bytes\n", len); - xs_tcp_check_recm(xprt); + transport->tcp_offset += len; + dprintk("RPC: discarded %Zu bytes\n", len); + xs_tcp_check_fraghdr(transport); } static int xs_tcp_data_recv(read_descriptor_t *rd_desc, struct sk_buff *skb, unsigned int offset, size_t len) { struct rpc_xprt *xprt = rd_desc->arg.data; - skb_reader_t desc = { + struct sock_xprt *transport = container_of(xprt, struct sock_xprt, xprt); + struct xdr_skb_reader desc = { .skb = skb, .offset = offset, .count = len, - .csum = 0 }; - dprintk("RPC: xs_tcp_data_recv started\n"); + dprintk("RPC: xs_tcp_data_recv started\n"); do { /* Read in a new fragment marker if necessary */ /* Can we ever really expect to get completely empty fragments? */ - if (xprt->tcp_flags & XPRT_COPY_RECM) { + if (transport->tcp_flags & TCP_RCV_COPY_FRAGHDR) { xs_tcp_read_fraghdr(xprt, &desc); continue; } /* Read in the xid if necessary */ - if (xprt->tcp_flags & XPRT_COPY_XID) { - xs_tcp_read_xid(xprt, &desc); + if (transport->tcp_flags & TCP_RCV_COPY_XID) { + xs_tcp_read_xid(transport, &desc); + continue; + } + /* Read in the call/reply flag */ + if (transport->tcp_flags & TCP_RCV_READ_CALLDIR) { + xs_tcp_read_calldir(transport, &desc); continue; } /* Read in the request data */ - if (xprt->tcp_flags & XPRT_COPY_DATA) { - xs_tcp_read_request(xprt, &desc); + if (transport->tcp_flags & TCP_RCV_COPY_DATA) { + xs_tcp_read_data(xprt, &desc); continue; } /* Skip over any trailing bytes on short reads */ - xs_tcp_read_discard(xprt, &desc); + xs_tcp_read_discard(transport, &desc); } while (desc.count); - dprintk("RPC: xs_tcp_data_recv done\n"); + dprintk("RPC: xs_tcp_data_recv done\n"); return len - desc.count; } @@ -745,24 +1435,79 @@ static int xs_tcp_data_recv(read_descriptor_t *rd_desc, struct sk_buff *skb, uns * @bytes: how much data to read * */ -static void xs_tcp_data_ready(struct sock *sk, int bytes) +static void xs_tcp_data_ready(struct sock *sk) { struct rpc_xprt *xprt; read_descriptor_t rd_desc; + int read; - read_lock(&sk->sk_callback_lock); - dprintk("RPC: xs_tcp_data_ready...\n"); + dprintk("RPC: xs_tcp_data_ready...\n"); + + read_lock_bh(&sk->sk_callback_lock); if (!(xprt = xprt_from_sock(sk))) goto out; - if (xprt->shutdown) - goto out; + /* Any data means we had a useful conversation, so + * the we don't need to delay the next reconnect + */ + if (xprt->reestablish_timeout) + xprt->reestablish_timeout = 0; /* We use rd_desc to pass struct xprt to xs_tcp_data_recv */ rd_desc.arg.data = xprt; - rd_desc.count = 65536; - tcp_read_sock(sk, &rd_desc, xs_tcp_data_recv); + do { + rd_desc.count = 65536; + read = tcp_read_sock(sk, &rd_desc, xs_tcp_data_recv); + } while (read > 0); out: - read_unlock(&sk->sk_callback_lock); + read_unlock_bh(&sk->sk_callback_lock); +} + +/* + * Do the equivalent of linger/linger2 handling for dealing with + * broken servers that don't close the socket in a timely + * fashion + */ +static void xs_tcp_schedule_linger_timeout(struct rpc_xprt *xprt, + unsigned long timeout) +{ + struct sock_xprt *transport; + + if (xprt_test_and_set_connecting(xprt)) + return; + set_bit(XPRT_CONNECTION_ABORT, &xprt->state); + transport = container_of(xprt, struct sock_xprt, xprt); + queue_delayed_work(rpciod_workqueue, &transport->connect_worker, + timeout); +} + +static void xs_tcp_cancel_linger_timeout(struct rpc_xprt *xprt) +{ + struct sock_xprt *transport; + + transport = container_of(xprt, struct sock_xprt, xprt); + + if (!test_bit(XPRT_CONNECTION_ABORT, &xprt->state) || + !cancel_delayed_work(&transport->connect_worker)) + return; + clear_bit(XPRT_CONNECTION_ABORT, &xprt->state); + xprt_clear_connecting(xprt); +} + +static void xs_sock_reset_connection_flags(struct rpc_xprt *xprt) +{ + smp_mb__before_atomic(); + clear_bit(XPRT_CONNECTION_ABORT, &xprt->state); + clear_bit(XPRT_CONNECTION_CLOSE, &xprt->state); + clear_bit(XPRT_CLOSE_WAIT, &xprt->state); + clear_bit(XPRT_CLOSING, &xprt->state); + smp_mb__after_atomic(); +} + +static void xs_sock_mark_closed(struct rpc_xprt *xprt) +{ + xs_sock_reset_connection_flags(xprt); + /* Mark transport as closed and wake up all pending tasks */ + xprt_disconnect_done(xprt); } /** @@ -774,38 +1519,90 @@ static void xs_tcp_state_change(struct sock *sk) { struct rpc_xprt *xprt; - read_lock(&sk->sk_callback_lock); + read_lock_bh(&sk->sk_callback_lock); if (!(xprt = xprt_from_sock(sk))) goto out; - dprintk("RPC: xs_tcp_state_change client %p...\n", xprt); - dprintk("RPC: state %x conn %d dead %d zapped %d\n", - sk->sk_state, xprt_connected(xprt), - sock_flag(sk, SOCK_DEAD), - sock_flag(sk, SOCK_ZAPPED)); - + dprintk("RPC: xs_tcp_state_change client %p...\n", xprt); + dprintk("RPC: state %x conn %d dead %d zapped %d sk_shutdown %d\n", + sk->sk_state, xprt_connected(xprt), + sock_flag(sk, SOCK_DEAD), + sock_flag(sk, SOCK_ZAPPED), + sk->sk_shutdown); + + trace_rpc_socket_state_change(xprt, sk->sk_socket); switch (sk->sk_state) { case TCP_ESTABLISHED: - spin_lock_bh(&xprt->transport_lock); + spin_lock(&xprt->transport_lock); if (!xprt_test_and_set_connected(xprt)) { + struct sock_xprt *transport = container_of(xprt, + struct sock_xprt, xprt); + /* Reset TCP record info */ - xprt->tcp_offset = 0; - xprt->tcp_reclen = 0; - xprt->tcp_copied = 0; - xprt->tcp_flags = XPRT_COPY_RECM | XPRT_COPY_XID; - xprt->reestablish_timeout = XS_TCP_INIT_REEST_TO; - xprt_wake_pending_tasks(xprt, 0); + transport->tcp_offset = 0; + transport->tcp_reclen = 0; + transport->tcp_copied = 0; + transport->tcp_flags = + TCP_RCV_COPY_FRAGHDR | TCP_RCV_COPY_XID; + xprt->connect_cookie++; + + xprt_wake_pending_tasks(xprt, -EAGAIN); } - spin_unlock_bh(&xprt->transport_lock); + spin_unlock(&xprt->transport_lock); break; - case TCP_SYN_SENT: - case TCP_SYN_RECV: + case TCP_FIN_WAIT1: + /* The client initiated a shutdown of the socket */ + xprt->connect_cookie++; + xprt->reestablish_timeout = 0; + set_bit(XPRT_CLOSING, &xprt->state); + smp_mb__before_atomic(); + clear_bit(XPRT_CONNECTED, &xprt->state); + clear_bit(XPRT_CLOSE_WAIT, &xprt->state); + smp_mb__after_atomic(); + xs_tcp_schedule_linger_timeout(xprt, xs_tcp_fin_timeout); break; - default: - xprt_disconnect(xprt); + case TCP_CLOSE_WAIT: + /* The server initiated a shutdown of the socket */ + xprt->connect_cookie++; + clear_bit(XPRT_CONNECTED, &xprt->state); + xs_tcp_force_close(xprt); + case TCP_CLOSING: + /* + * If the server closed down the connection, make sure that + * we back off before reconnecting + */ + if (xprt->reestablish_timeout < XS_TCP_INIT_REEST_TO) + xprt->reestablish_timeout = XS_TCP_INIT_REEST_TO; + break; + case TCP_LAST_ACK: + set_bit(XPRT_CLOSING, &xprt->state); + xs_tcp_schedule_linger_timeout(xprt, xs_tcp_fin_timeout); + smp_mb__before_atomic(); + clear_bit(XPRT_CONNECTED, &xprt->state); + smp_mb__after_atomic(); break; + case TCP_CLOSE: + xs_tcp_cancel_linger_timeout(xprt); + xs_sock_mark_closed(xprt); } out: - read_unlock(&sk->sk_callback_lock); + read_unlock_bh(&sk->sk_callback_lock); +} + +static void xs_write_space(struct sock *sk) +{ + struct socket *sock; + struct rpc_xprt *xprt; + + if (unlikely(!(sock = sk->sk_socket))) + return; + clear_bit(SOCK_NOSPACE, &sock->flags); + + if (unlikely(!(xprt = xprt_from_sock(sk)))) + return; + if (test_and_clear_bit(SOCK_ASYNC_NOSPACE, &sock->flags) == 0) + return; + + xprt_write_space(xprt); } /** @@ -820,25 +1617,13 @@ static void xs_tcp_state_change(struct sock *sk) */ static void xs_udp_write_space(struct sock *sk) { - read_lock(&sk->sk_callback_lock); + read_lock_bh(&sk->sk_callback_lock); /* from net/core/sock.c:sock_def_write_space */ - if (sock_writeable(sk)) { - struct socket *sock; - struct rpc_xprt *xprt; + if (sock_writeable(sk)) + xs_write_space(sk); - if (unlikely(!(sock = sk->sk_socket))) - goto out; - if (unlikely(!(xprt = xprt_from_sock(sk)))) - goto out; - if (unlikely(!test_and_clear_bit(SOCK_NOSPACE, &sock->flags))) - goto out; - - xprt_write_space(xprt); - } - - out: - read_unlock(&sk->sk_callback_lock); + read_unlock_bh(&sk->sk_callback_lock); } /** @@ -853,38 +1638,27 @@ static void xs_udp_write_space(struct sock *sk) */ static void xs_tcp_write_space(struct sock *sk) { - read_lock(&sk->sk_callback_lock); + read_lock_bh(&sk->sk_callback_lock); /* from net/core/stream.c:sk_stream_write_space */ - if (sk_stream_wspace(sk) >= sk_stream_min_wspace(sk)) { - struct socket *sock; - struct rpc_xprt *xprt; - - if (unlikely(!(sock = sk->sk_socket))) - goto out; - if (unlikely(!(xprt = xprt_from_sock(sk)))) - goto out; - if (unlikely(!test_and_clear_bit(SOCK_NOSPACE, &sock->flags))) - goto out; + if (sk_stream_is_writeable(sk)) + xs_write_space(sk); - xprt_write_space(xprt); - } - - out: - read_unlock(&sk->sk_callback_lock); + read_unlock_bh(&sk->sk_callback_lock); } static void xs_udp_do_set_buffer_size(struct rpc_xprt *xprt) { - struct sock *sk = xprt->inet; + struct sock_xprt *transport = container_of(xprt, struct sock_xprt, xprt); + struct sock *sk = transport->inet; - if (xprt->rcvsize) { + if (transport->rcvsize) { sk->sk_userlocks |= SOCK_RCVBUF_LOCK; - sk->sk_rcvbuf = xprt->rcvsize * xprt->max_reqs * 2; + sk->sk_rcvbuf = transport->rcvsize * xprt->max_reqs * 2; } - if (xprt->sndsize) { + if (transport->sndsize) { sk->sk_userlocks |= SOCK_SNDBUF_LOCK; - sk->sk_sndbuf = xprt->sndsize * xprt->max_reqs * 2; + sk->sk_sndbuf = transport->sndsize * xprt->max_reqs * 2; sk->sk_write_space(sk); } } @@ -899,12 +1673,14 @@ static void xs_udp_do_set_buffer_size(struct rpc_xprt *xprt) */ static void xs_udp_set_buffer_size(struct rpc_xprt *xprt, size_t sndsize, size_t rcvsize) { - xprt->sndsize = 0; + struct sock_xprt *transport = container_of(xprt, struct sock_xprt, xprt); + + transport->sndsize = 0; if (sndsize) - xprt->sndsize = sndsize + 1024; - xprt->rcvsize = 0; + transport->sndsize = sndsize + 1024; + transport->rcvsize = 0; if (rcvsize) - xprt->rcvsize = rcvsize + 1024; + transport->rcvsize = rcvsize + 1024; xs_udp_do_set_buffer_size(xprt); } @@ -915,108 +1691,416 @@ static void xs_udp_set_buffer_size(struct rpc_xprt *xprt, size_t sndsize, size_t * * Adjust the congestion window after a retransmit timeout has occurred. */ -static void xs_udp_timer(struct rpc_task *task) +static void xs_udp_timer(struct rpc_xprt *xprt, struct rpc_task *task) { - xprt_adjust_cwnd(task, -ETIMEDOUT); + xprt_adjust_cwnd(xprt, task, -ETIMEDOUT); } -static int xs_bindresvport(struct rpc_xprt *xprt, struct socket *sock) +static unsigned short xs_get_random_port(void) { - struct sockaddr_in myaddr = { - .sin_family = AF_INET, - }; - int err; - unsigned short port = xprt->port; + unsigned short range = xprt_max_resvport - xprt_min_resvport; + unsigned short rand = (unsigned short) prandom_u32() % range; + return rand + xprt_min_resvport; +} + +/** + * xs_set_port - reset the port number in the remote endpoint address + * @xprt: generic transport + * @port: new port number + * + */ +static void xs_set_port(struct rpc_xprt *xprt, unsigned short port) +{ + dprintk("RPC: setting port for xprt %p to %u\n", xprt, port); + + rpc_set_port(xs_addr(xprt), port); + xs_update_peer_port(xprt); +} + +static unsigned short xs_get_srcport(struct sock_xprt *transport) +{ + unsigned short port = transport->srcport; + + if (port == 0 && transport->xprt.resvport) + port = xs_get_random_port(); + return port; +} +static unsigned short xs_next_srcport(struct sock_xprt *transport, unsigned short port) +{ + if (transport->srcport != 0) + transport->srcport = 0; + if (!transport->xprt.resvport) + return 0; + if (port <= xprt_min_resvport || port > xprt_max_resvport) + return xprt_max_resvport; + return --port; +} +static int xs_bind(struct sock_xprt *transport, struct socket *sock) +{ + struct sockaddr_storage myaddr; + int err, nloop = 0; + unsigned short port = xs_get_srcport(transport); + unsigned short last; + + memcpy(&myaddr, &transport->srcaddr, transport->xprt.addrlen); do { - myaddr.sin_port = htons(port); - err = sock->ops->bind(sock, (struct sockaddr *) &myaddr, - sizeof(myaddr)); + rpc_set_port((struct sockaddr *)&myaddr, port); + err = kernel_bind(sock, (struct sockaddr *)&myaddr, + transport->xprt.addrlen); + if (port == 0) + break; if (err == 0) { - xprt->port = port; - dprintk("RPC: xs_bindresvport bound to port %u\n", - port); - return 0; + transport->srcport = port; + break; } - if (port <= xprt_min_resvport) - port = xprt_max_resvport; - else - port--; - } while (err == -EADDRINUSE && port != xprt->port); - - dprintk("RPC: can't bind to reserved port (%d).\n", -err); + last = port; + port = xs_next_srcport(transport, port); + if (port > last) + nloop++; + } while (err == -EADDRINUSE && nloop != 2); + + if (myaddr.ss_family == AF_INET) + dprintk("RPC: %s %pI4:%u: %s (%d)\n", __func__, + &((struct sockaddr_in *)&myaddr)->sin_addr, + port, err ? "failed" : "ok", err); + else + dprintk("RPC: %s %pI6:%u: %s (%d)\n", __func__, + &((struct sockaddr_in6 *)&myaddr)->sin6_addr, + port, err ? "failed" : "ok", err); return err; } -/** - * xs_udp_connect_worker - set up a UDP socket - * @args: RPC transport to connect - * - * Invoked by a work queue tasklet. +/* + * We don't support autobind on AF_LOCAL sockets */ -static void xs_udp_connect_worker(void *args) +static void xs_local_rpcbind(struct rpc_task *task) { - struct rpc_xprt *xprt = (struct rpc_xprt *) args; - struct socket *sock = xprt->sock; - int err, status = -EIO; + rcu_read_lock(); + xprt_set_bound(rcu_dereference(task->tk_client->cl_xprt)); + rcu_read_unlock(); +} - if (xprt->shutdown || xprt->addr.sin_port == 0) - goto out; +static void xs_local_set_port(struct rpc_xprt *xprt, unsigned short port) +{ +} - dprintk("RPC: xs_udp_connect_worker for xprt %p\n", xprt); +#ifdef CONFIG_DEBUG_LOCK_ALLOC +static struct lock_class_key xs_key[2]; +static struct lock_class_key xs_slock_key[2]; - /* Start by resetting any existing state */ - xs_close(xprt); +static inline void xs_reclassify_socketu(struct socket *sock) +{ + struct sock *sk = sock->sk; + + sock_lock_init_class_and_name(sk, "slock-AF_LOCAL-RPC", + &xs_slock_key[1], "sk_lock-AF_LOCAL-RPC", &xs_key[1]); +} - if ((err = sock_create_kern(PF_INET, SOCK_DGRAM, IPPROTO_UDP, &sock)) < 0) { - dprintk("RPC: can't create UDP transport socket (%d).\n", -err); +static inline void xs_reclassify_socket4(struct socket *sock) +{ + struct sock *sk = sock->sk; + + sock_lock_init_class_and_name(sk, "slock-AF_INET-RPC", + &xs_slock_key[0], "sk_lock-AF_INET-RPC", &xs_key[0]); +} + +static inline void xs_reclassify_socket6(struct socket *sock) +{ + struct sock *sk = sock->sk; + + sock_lock_init_class_and_name(sk, "slock-AF_INET6-RPC", + &xs_slock_key[1], "sk_lock-AF_INET6-RPC", &xs_key[1]); +} + +static inline void xs_reclassify_socket(int family, struct socket *sock) +{ + WARN_ON_ONCE(sock_owned_by_user(sock->sk)); + if (sock_owned_by_user(sock->sk)) + return; + + switch (family) { + case AF_LOCAL: + xs_reclassify_socketu(sock); + break; + case AF_INET: + xs_reclassify_socket4(sock); + break; + case AF_INET6: + xs_reclassify_socket6(sock); + break; + } +} +#else +static inline void xs_reclassify_socketu(struct socket *sock) +{ +} + +static inline void xs_reclassify_socket4(struct socket *sock) +{ +} + +static inline void xs_reclassify_socket6(struct socket *sock) +{ +} + +static inline void xs_reclassify_socket(int family, struct socket *sock) +{ +} +#endif + +static void xs_dummy_setup_socket(struct work_struct *work) +{ +} + +static struct socket *xs_create_sock(struct rpc_xprt *xprt, + struct sock_xprt *transport, int family, int type, int protocol) +{ + struct socket *sock; + int err; + + err = __sock_create(xprt->xprt_net, family, type, protocol, &sock, 1); + if (err < 0) { + dprintk("RPC: can't create %d transport socket (%d).\n", + protocol, -err); goto out; } + xs_reclassify_socket(family, sock); - if (xprt->resvport && xs_bindresvport(xprt, sock) < 0) { + err = xs_bind(transport, sock); + if (err) { sock_release(sock); goto out; } - if (!xprt->inet) { + return sock; +out: + return ERR_PTR(err); +} + +static int xs_local_finish_connecting(struct rpc_xprt *xprt, + struct socket *sock) +{ + struct sock_xprt *transport = container_of(xprt, struct sock_xprt, + xprt); + + if (!transport->inet) { struct sock *sk = sock->sk; write_lock_bh(&sk->sk_callback_lock); + xs_save_old_callbacks(transport, sk); + + sk->sk_user_data = xprt; + sk->sk_data_ready = xs_local_data_ready; + sk->sk_write_space = xs_udp_write_space; + sk->sk_error_report = xs_error_report; + sk->sk_allocation = GFP_ATOMIC; + + xprt_clear_connected(xprt); + + /* Reset to new socket */ + transport->sock = sock; + transport->inet = sk; + + write_unlock_bh(&sk->sk_callback_lock); + } + + /* Tell the socket layer to start connecting... */ + xprt->stat.connect_count++; + xprt->stat.connect_start = jiffies; + return kernel_connect(sock, xs_addr(xprt), xprt->addrlen, 0); +} + +/** + * xs_local_setup_socket - create AF_LOCAL socket, connect to a local endpoint + * @xprt: RPC transport to connect + * @transport: socket transport to connect + * @create_sock: function to create a socket of the correct type + */ +static int xs_local_setup_socket(struct sock_xprt *transport) +{ + struct rpc_xprt *xprt = &transport->xprt; + struct socket *sock; + int status = -EIO; + + current->flags |= PF_FSTRANS; + + clear_bit(XPRT_CONNECTION_ABORT, &xprt->state); + status = __sock_create(xprt->xprt_net, AF_LOCAL, + SOCK_STREAM, 0, &sock, 1); + if (status < 0) { + dprintk("RPC: can't create AF_LOCAL " + "transport socket (%d).\n", -status); + goto out; + } + xs_reclassify_socketu(sock); + + dprintk("RPC: worker connecting xprt %p via AF_LOCAL to %s\n", + xprt, xprt->address_strings[RPC_DISPLAY_ADDR]); + + status = xs_local_finish_connecting(xprt, sock); + trace_rpc_socket_connect(xprt, sock, status); + switch (status) { + case 0: + dprintk("RPC: xprt %p connected to %s\n", + xprt, xprt->address_strings[RPC_DISPLAY_ADDR]); + xprt_set_connected(xprt); + break; + case -ENOENT: + dprintk("RPC: xprt %p: socket %s does not exist\n", + xprt, xprt->address_strings[RPC_DISPLAY_ADDR]); + break; + case -ECONNREFUSED: + dprintk("RPC: xprt %p: connection refused for %s\n", + xprt, xprt->address_strings[RPC_DISPLAY_ADDR]); + break; + default: + printk(KERN_ERR "%s: unhandled error (%d) connecting to %s\n", + __func__, -status, + xprt->address_strings[RPC_DISPLAY_ADDR]); + } + +out: + xprt_clear_connecting(xprt); + xprt_wake_pending_tasks(xprt, status); + current->flags &= ~PF_FSTRANS; + return status; +} + +static void xs_local_connect(struct rpc_xprt *xprt, struct rpc_task *task) +{ + struct sock_xprt *transport = container_of(xprt, struct sock_xprt, xprt); + int ret; + + if (RPC_IS_ASYNC(task)) { + /* + * We want the AF_LOCAL connect to be resolved in the + * filesystem namespace of the process making the rpc + * call. Thus we connect synchronously. + * + * If we want to support asynchronous AF_LOCAL calls, + * we'll need to figure out how to pass a namespace to + * connect. + */ + rpc_exit(task, -ENOTCONN); + return; + } + ret = xs_local_setup_socket(transport); + if (ret && !RPC_IS_SOFTCONN(task)) + msleep_interruptible(15000); +} + +#ifdef CONFIG_SUNRPC_SWAP +static void xs_set_memalloc(struct rpc_xprt *xprt) +{ + struct sock_xprt *transport = container_of(xprt, struct sock_xprt, + xprt); + + if (xprt->swapper) + sk_set_memalloc(transport->inet); +} + +/** + * xs_swapper - Tag this transport as being used for swap. + * @xprt: transport to tag + * @enable: enable/disable + * + */ +int xs_swapper(struct rpc_xprt *xprt, int enable) +{ + struct sock_xprt *transport = container_of(xprt, struct sock_xprt, + xprt); + int err = 0; + + if (enable) { + xprt->swapper++; + xs_set_memalloc(xprt); + } else if (xprt->swapper) { + xprt->swapper--; + sk_clear_memalloc(transport->inet); + } + + return err; +} +EXPORT_SYMBOL_GPL(xs_swapper); +#else +static void xs_set_memalloc(struct rpc_xprt *xprt) +{ +} +#endif + +static void xs_udp_finish_connecting(struct rpc_xprt *xprt, struct socket *sock) +{ + struct sock_xprt *transport = container_of(xprt, struct sock_xprt, xprt); + + if (!transport->inet) { + struct sock *sk = sock->sk; + + write_lock_bh(&sk->sk_callback_lock); + + xs_save_old_callbacks(transport, sk); + sk->sk_user_data = xprt; - xprt->old_data_ready = sk->sk_data_ready; - xprt->old_state_change = sk->sk_state_change; - xprt->old_write_space = sk->sk_write_space; sk->sk_data_ready = xs_udp_data_ready; sk->sk_write_space = xs_udp_write_space; - sk->sk_no_check = UDP_CSUM_NORCV; + sk->sk_allocation = GFP_ATOMIC; xprt_set_connected(xprt); /* Reset to new socket */ - xprt->sock = sock; - xprt->inet = sk; + transport->sock = sock; + transport->inet = sk; + + xs_set_memalloc(xprt); write_unlock_bh(&sk->sk_callback_lock); } xs_udp_do_set_buffer_size(xprt); +} + +static void xs_udp_setup_socket(struct work_struct *work) +{ + struct sock_xprt *transport = + container_of(work, struct sock_xprt, connect_worker.work); + struct rpc_xprt *xprt = &transport->xprt; + struct socket *sock = transport->sock; + int status = -EIO; + + current->flags |= PF_FSTRANS; + + /* Start by resetting any existing state */ + xs_reset_transport(transport); + sock = xs_create_sock(xprt, transport, + xs_addr(xprt)->sa_family, SOCK_DGRAM, IPPROTO_UDP); + if (IS_ERR(sock)) + goto out; + + dprintk("RPC: worker connecting xprt %p via %s to " + "%s (port %s)\n", xprt, + xprt->address_strings[RPC_DISPLAY_PROTO], + xprt->address_strings[RPC_DISPLAY_ADDR], + xprt->address_strings[RPC_DISPLAY_PORT]); + + xs_udp_finish_connecting(xprt, sock); + trace_rpc_socket_connect(xprt, sock, 0); status = 0; out: - xprt_wake_pending_tasks(xprt, status); xprt_clear_connecting(xprt); + xprt_wake_pending_tasks(xprt, status); + current->flags &= ~PF_FSTRANS; } /* * We need to preserve the port number so the reply cache on the server can * find our cached RPC replies when we get around to reconnecting. */ -static void xs_tcp_reuse_connection(struct rpc_xprt *xprt) +static void xs_abort_connection(struct sock_xprt *transport) { int result; - struct socket *sock = xprt->sock; struct sockaddr any; - dprintk("RPC: disconnecting xprt %p to reuse port\n", xprt); + dprintk("RPC: disconnecting xprt %p to reuse port\n", transport); /* * Disconnect the transport socket by doing a connect operation @@ -1024,56 +2108,71 @@ static void xs_tcp_reuse_connection(struct rpc_xprt *xprt) */ memset(&any, 0, sizeof(any)); any.sa_family = AF_UNSPEC; - result = sock->ops->connect(sock, &any, sizeof(any), 0); - if (result) - dprintk("RPC: AF_UNSPEC connect return code %d\n", - result); + result = kernel_connect(transport->sock, &any, sizeof(any), 0); + trace_rpc_socket_reset_connection(&transport->xprt, + transport->sock, result); + if (!result) + xs_sock_reset_connection_flags(&transport->xprt); + dprintk("RPC: AF_UNSPEC connect return code %d\n", result); } -/** - * xs_tcp_connect_worker - connect a TCP socket to a remote endpoint - * @args: RPC transport to connect - * - * Invoked by a work queue tasklet. - */ -static void xs_tcp_connect_worker(void *args) +static void xs_tcp_reuse_connection(struct sock_xprt *transport) { - struct rpc_xprt *xprt = (struct rpc_xprt *)args; - struct socket *sock = xprt->sock; - int err, status = -EIO; + unsigned int state = transport->inet->sk_state; - if (xprt->shutdown || xprt->addr.sin_port == 0) - goto out; - - dprintk("RPC: xs_tcp_connect_worker for xprt %p\n", xprt); - - if (!xprt->sock) { - /* start from scratch */ - if ((err = sock_create_kern(PF_INET, SOCK_STREAM, IPPROTO_TCP, &sock)) < 0) { - dprintk("RPC: can't create TCP transport socket (%d).\n", -err); - goto out; - } + if (state == TCP_CLOSE && transport->sock->state == SS_UNCONNECTED) { + /* we don't need to abort the connection if the socket + * hasn't undergone a shutdown + */ + if (transport->inet->sk_shutdown == 0) + return; + dprintk("RPC: %s: TCP_CLOSEd and sk_shutdown set to %d\n", + __func__, transport->inet->sk_shutdown); + } + if ((1 << state) & (TCPF_ESTABLISHED|TCPF_SYN_SENT)) { + /* we don't need to abort the connection if the socket + * hasn't undergone a shutdown + */ + if (transport->inet->sk_shutdown == 0) + return; + dprintk("RPC: %s: ESTABLISHED/SYN_SENT " + "sk_shutdown set to %d\n", + __func__, transport->inet->sk_shutdown); + } + xs_abort_connection(transport); +} - if (xprt->resvport && xs_bindresvport(xprt, sock) < 0) { - sock_release(sock); - goto out; - } - } else - /* "close" the socket, preserving the local port */ - xs_tcp_reuse_connection(xprt); +static int xs_tcp_finish_connecting(struct rpc_xprt *xprt, struct socket *sock) +{ + struct sock_xprt *transport = container_of(xprt, struct sock_xprt, xprt); + int ret = -ENOTCONN; - if (!xprt->inet) { + if (!transport->inet) { struct sock *sk = sock->sk; + unsigned int keepidle = xprt->timeout->to_initval / HZ; + unsigned int keepcnt = xprt->timeout->to_retries + 1; + unsigned int opt_on = 1; + + /* TCP Keepalive options */ + kernel_setsockopt(sock, SOL_SOCKET, SO_KEEPALIVE, + (char *)&opt_on, sizeof(opt_on)); + kernel_setsockopt(sock, SOL_TCP, TCP_KEEPIDLE, + (char *)&keepidle, sizeof(keepidle)); + kernel_setsockopt(sock, SOL_TCP, TCP_KEEPINTVL, + (char *)&keepidle, sizeof(keepidle)); + kernel_setsockopt(sock, SOL_TCP, TCP_KEEPCNT, + (char *)&keepcnt, sizeof(keepcnt)); write_lock_bh(&sk->sk_callback_lock); + xs_save_old_callbacks(transport, sk); + sk->sk_user_data = xprt; - xprt->old_data_ready = sk->sk_data_ready; - xprt->old_state_change = sk->sk_state_change; - xprt->old_write_space = sk->sk_write_space; sk->sk_data_ready = xs_tcp_data_ready; sk->sk_state_change = xs_tcp_state_change; sk->sk_write_space = xs_tcp_write_space; + sk->sk_error_report = xs_error_report; + sk->sk_allocation = GFP_ATOMIC; /* socket options */ sk->sk_userlocks |= SOCK_BINDPORT_LOCK; @@ -1084,40 +2183,118 @@ static void xs_tcp_connect_worker(void *args) xprt_clear_connected(xprt); /* Reset to new socket */ - xprt->sock = sock; - xprt->inet = sk; + transport->sock = sock; + transport->inet = sk; write_unlock_bh(&sk->sk_callback_lock); } + if (!xprt_bound(xprt)) + goto out; + + xs_set_memalloc(xprt); + /* Tell the socket layer to start connecting... */ - status = sock->ops->connect(sock, (struct sockaddr *) &xprt->addr, - sizeof(xprt->addr), O_NONBLOCK); - dprintk("RPC: %p connect status %d connected %d sock state %d\n", - xprt, -status, xprt_connected(xprt), sock->sk->sk_state); - if (status < 0) { - switch (status) { - case -EINPROGRESS: - case -EALREADY: - goto out_clear; - case -ECONNREFUSED: - case -ECONNRESET: - /* retry with existing socket, after a delay */ - break; - default: - /* get rid of existing socket, and retry */ - xs_close(xprt); - break; + xprt->stat.connect_count++; + xprt->stat.connect_start = jiffies; + ret = kernel_connect(sock, xs_addr(xprt), xprt->addrlen, O_NONBLOCK); + switch (ret) { + case 0: + case -EINPROGRESS: + /* SYN_SENT! */ + if (xprt->reestablish_timeout < XS_TCP_INIT_REEST_TO) + xprt->reestablish_timeout = XS_TCP_INIT_REEST_TO; + } +out: + return ret; +} + +/** + * xs_tcp_setup_socket - create a TCP socket and connect to a remote endpoint + * @xprt: RPC transport to connect + * @transport: socket transport to connect + * @create_sock: function to create a socket of the correct type + * + * Invoked by a work queue tasklet. + */ +static void xs_tcp_setup_socket(struct work_struct *work) +{ + struct sock_xprt *transport = + container_of(work, struct sock_xprt, connect_worker.work); + struct socket *sock = transport->sock; + struct rpc_xprt *xprt = &transport->xprt; + int status = -EIO; + + current->flags |= PF_FSTRANS; + + if (!sock) { + clear_bit(XPRT_CONNECTION_ABORT, &xprt->state); + sock = xs_create_sock(xprt, transport, + xs_addr(xprt)->sa_family, SOCK_STREAM, IPPROTO_TCP); + if (IS_ERR(sock)) { + status = PTR_ERR(sock); + goto out; } + } else { + int abort_and_exit; + + abort_and_exit = test_and_clear_bit(XPRT_CONNECTION_ABORT, + &xprt->state); + /* "close" the socket, preserving the local port */ + xs_tcp_reuse_connection(transport); + + if (abort_and_exit) + goto out_eagain; + } + + dprintk("RPC: worker connecting xprt %p via %s to " + "%s (port %s)\n", xprt, + xprt->address_strings[RPC_DISPLAY_PROTO], + xprt->address_strings[RPC_DISPLAY_ADDR], + xprt->address_strings[RPC_DISPLAY_PORT]); + + status = xs_tcp_finish_connecting(xprt, sock); + trace_rpc_socket_connect(xprt, sock, status); + dprintk("RPC: %p connect status %d connected %d sock state %d\n", + xprt, -status, xprt_connected(xprt), + sock->sk->sk_state); + switch (status) { + default: + printk("%s: connect returned unhandled error %d\n", + __func__, status); + case -EADDRNOTAVAIL: + /* We're probably in TIME_WAIT. Get rid of existing socket, + * and retry + */ + xs_tcp_force_close(xprt); + break; + case 0: + case -EINPROGRESS: + case -EALREADY: + xprt_clear_connecting(xprt); + current->flags &= ~PF_FSTRANS; + return; + case -EINVAL: + /* Happens, for instance, if the user specified a link + * local IPv6 address without a scope-id. + */ + case -ECONNREFUSED: + case -ECONNRESET: + case -ENETUNREACH: + /* retry with existing socket, after a delay */ + goto out; } +out_eagain: + status = -EAGAIN; out: - xprt_wake_pending_tasks(xprt, status); -out_clear: xprt_clear_connecting(xprt); + xprt_wake_pending_tasks(xprt, status); + current->flags &= ~PF_FSTRANS; } /** * xs_connect - connect a socket to a remote endpoint + * @xprt: pointer to transport structure * @task: address of RPC task that manages state of connect request * * TCP: If the remote end dropped the connection, delay reconnecting. @@ -1129,133 +2306,820 @@ out_clear: * If a UDP socket connect fails, the delay behavior here prevents * retry floods (hard mounts). */ -static void xs_connect(struct rpc_task *task) +static void xs_connect(struct rpc_xprt *xprt, struct rpc_task *task) { - struct rpc_xprt *xprt = task->tk_xprt; + struct sock_xprt *transport = container_of(xprt, struct sock_xprt, xprt); - if (xprt_test_and_set_connecting(xprt)) - return; - - if (xprt->sock != NULL) { - dprintk("RPC: xs_connect delayed xprt %p for %lu seconds\n", + if (transport->sock != NULL && !RPC_IS_SOFTCONN(task)) { + dprintk("RPC: xs_connect delayed xprt %p for %lu " + "seconds\n", xprt, xprt->reestablish_timeout / HZ); - schedule_delayed_work(&xprt->connect_worker, - xprt->reestablish_timeout); + queue_delayed_work(rpciod_workqueue, + &transport->connect_worker, + xprt->reestablish_timeout); xprt->reestablish_timeout <<= 1; + if (xprt->reestablish_timeout < XS_TCP_INIT_REEST_TO) + xprt->reestablish_timeout = XS_TCP_INIT_REEST_TO; if (xprt->reestablish_timeout > XS_TCP_MAX_REEST_TO) xprt->reestablish_timeout = XS_TCP_MAX_REEST_TO; } else { - dprintk("RPC: xs_connect scheduled xprt %p\n", xprt); - schedule_work(&xprt->connect_worker); + dprintk("RPC: xs_connect scheduled xprt %p\n", xprt); + queue_delayed_work(rpciod_workqueue, + &transport->connect_worker, 0); + } +} + +/** + * xs_local_print_stats - display AF_LOCAL socket-specifc stats + * @xprt: rpc_xprt struct containing statistics + * @seq: output file + * + */ +static void xs_local_print_stats(struct rpc_xprt *xprt, struct seq_file *seq) +{ + long idle_time = 0; + + if (xprt_connected(xprt)) + idle_time = (long)(jiffies - xprt->last_used) / HZ; + + seq_printf(seq, "\txprt:\tlocal %lu %lu %lu %ld %lu %lu %lu " + "%llu %llu %lu %llu %llu\n", + xprt->stat.bind_count, + xprt->stat.connect_count, + xprt->stat.connect_time, + idle_time, + xprt->stat.sends, + xprt->stat.recvs, + xprt->stat.bad_xids, + xprt->stat.req_u, + xprt->stat.bklog_u, + xprt->stat.max_slots, + xprt->stat.sending_u, + xprt->stat.pending_u); +} - /* flush_scheduled_work can sleep... */ - if (!RPC_IS_ASYNC(task)) - flush_scheduled_work(); +/** + * xs_udp_print_stats - display UDP socket-specifc stats + * @xprt: rpc_xprt struct containing statistics + * @seq: output file + * + */ +static void xs_udp_print_stats(struct rpc_xprt *xprt, struct seq_file *seq) +{ + struct sock_xprt *transport = container_of(xprt, struct sock_xprt, xprt); + + seq_printf(seq, "\txprt:\tudp %u %lu %lu %lu %lu %llu %llu " + "%lu %llu %llu\n", + transport->srcport, + xprt->stat.bind_count, + xprt->stat.sends, + xprt->stat.recvs, + xprt->stat.bad_xids, + xprt->stat.req_u, + xprt->stat.bklog_u, + xprt->stat.max_slots, + xprt->stat.sending_u, + xprt->stat.pending_u); +} + +/** + * xs_tcp_print_stats - display TCP socket-specifc stats + * @xprt: rpc_xprt struct containing statistics + * @seq: output file + * + */ +static void xs_tcp_print_stats(struct rpc_xprt *xprt, struct seq_file *seq) +{ + struct sock_xprt *transport = container_of(xprt, struct sock_xprt, xprt); + long idle_time = 0; + + if (xprt_connected(xprt)) + idle_time = (long)(jiffies - xprt->last_used) / HZ; + + seq_printf(seq, "\txprt:\ttcp %u %lu %lu %lu %ld %lu %lu %lu " + "%llu %llu %lu %llu %llu\n", + transport->srcport, + xprt->stat.bind_count, + xprt->stat.connect_count, + xprt->stat.connect_time, + idle_time, + xprt->stat.sends, + xprt->stat.recvs, + xprt->stat.bad_xids, + xprt->stat.req_u, + xprt->stat.bklog_u, + xprt->stat.max_slots, + xprt->stat.sending_u, + xprt->stat.pending_u); +} + +/* + * Allocate a bunch of pages for a scratch buffer for the rpc code. The reason + * we allocate pages instead doing a kmalloc like rpc_malloc is because we want + * to use the server side send routines. + */ +static void *bc_malloc(struct rpc_task *task, size_t size) +{ + struct page *page; + struct rpc_buffer *buf; + + WARN_ON_ONCE(size > PAGE_SIZE - sizeof(struct rpc_buffer)); + if (size > PAGE_SIZE - sizeof(struct rpc_buffer)) + return NULL; + + page = alloc_page(GFP_KERNEL); + if (!page) + return NULL; + + buf = page_address(page); + buf->len = PAGE_SIZE; + + return buf->data; +} + +/* + * Free the space allocated in the bc_alloc routine + */ +static void bc_free(void *buffer) +{ + struct rpc_buffer *buf; + + if (!buffer) + return; + + buf = container_of(buffer, struct rpc_buffer, data); + free_page((unsigned long)buf); +} + +/* + * Use the svc_sock to send the callback. Must be called with svsk->sk_mutex + * held. Borrows heavily from svc_tcp_sendto and xs_tcp_send_request. + */ +static int bc_sendto(struct rpc_rqst *req) +{ + int len; + struct xdr_buf *xbufp = &req->rq_snd_buf; + struct rpc_xprt *xprt = req->rq_xprt; + struct sock_xprt *transport = + container_of(xprt, struct sock_xprt, xprt); + struct socket *sock = transport->sock; + unsigned long headoff; + unsigned long tailoff; + + xs_encode_stream_record_marker(xbufp); + + tailoff = (unsigned long)xbufp->tail[0].iov_base & ~PAGE_MASK; + headoff = (unsigned long)xbufp->head[0].iov_base & ~PAGE_MASK; + len = svc_send_common(sock, xbufp, + virt_to_page(xbufp->head[0].iov_base), headoff, + xbufp->tail[0].iov_base, tailoff); + + if (len != xbufp->len) { + printk(KERN_NOTICE "Error sending entire callback!\n"); + len = -EAGAIN; + } + + return len; +} + +/* + * The send routine. Borrows from svc_send + */ +static int bc_send_request(struct rpc_task *task) +{ + struct rpc_rqst *req = task->tk_rqstp; + struct svc_xprt *xprt; + u32 len; + + dprintk("sending request with xid: %08x\n", ntohl(req->rq_xid)); + /* + * Get the server socket associated with this callback xprt + */ + xprt = req->rq_xprt->bc_xprt; + + /* + * Grab the mutex to serialize data as the connection is shared + * with the fore channel + */ + if (!mutex_trylock(&xprt->xpt_mutex)) { + rpc_sleep_on(&xprt->xpt_bc_pending, task, NULL); + if (!mutex_trylock(&xprt->xpt_mutex)) + return -EAGAIN; + rpc_wake_up_queued_task(&xprt->xpt_bc_pending, task); } + if (test_bit(XPT_DEAD, &xprt->xpt_flags)) + len = -ENOTCONN; + else + len = bc_sendto(req); + mutex_unlock(&xprt->xpt_mutex); + + if (len > 0) + len = 0; + + return len; +} + +/* + * The close routine. Since this is client initiated, we do nothing + */ + +static void bc_close(struct rpc_xprt *xprt) +{ +} + +/* + * The xprt destroy routine. Again, because this connection is client + * initiated, we do nothing + */ + +static void bc_destroy(struct rpc_xprt *xprt) +{ + dprintk("RPC: bc_destroy xprt %p\n", xprt); + + xs_xprt_free(xprt); + module_put(THIS_MODULE); } +static struct rpc_xprt_ops xs_local_ops = { + .reserve_xprt = xprt_reserve_xprt, + .release_xprt = xs_tcp_release_xprt, + .alloc_slot = xprt_alloc_slot, + .rpcbind = xs_local_rpcbind, + .set_port = xs_local_set_port, + .connect = xs_local_connect, + .buf_alloc = rpc_malloc, + .buf_free = rpc_free, + .send_request = xs_local_send_request, + .set_retrans_timeout = xprt_set_retrans_timeout_def, + .close = xs_close, + .destroy = xs_destroy, + .print_stats = xs_local_print_stats, +}; + static struct rpc_xprt_ops xs_udp_ops = { .set_buffer_size = xs_udp_set_buffer_size, .reserve_xprt = xprt_reserve_xprt_cong, .release_xprt = xprt_release_xprt_cong, + .alloc_slot = xprt_alloc_slot, + .rpcbind = rpcb_getport_async, + .set_port = xs_set_port, .connect = xs_connect, + .buf_alloc = rpc_malloc, + .buf_free = rpc_free, .send_request = xs_udp_send_request, .set_retrans_timeout = xprt_set_retrans_timeout_rtt, .timer = xs_udp_timer, .release_request = xprt_release_rqst_cong, .close = xs_close, .destroy = xs_destroy, + .print_stats = xs_udp_print_stats, }; static struct rpc_xprt_ops xs_tcp_ops = { .reserve_xprt = xprt_reserve_xprt, - .release_xprt = xprt_release_xprt, + .release_xprt = xs_tcp_release_xprt, + .alloc_slot = xprt_lock_and_alloc_slot, + .rpcbind = rpcb_getport_async, + .set_port = xs_set_port, .connect = xs_connect, + .buf_alloc = rpc_malloc, + .buf_free = rpc_free, .send_request = xs_tcp_send_request, .set_retrans_timeout = xprt_set_retrans_timeout_def, - .close = xs_close, + .close = xs_tcp_close, .destroy = xs_destroy, + .print_stats = xs_tcp_print_stats, +}; + +/* + * The rpc_xprt_ops for the server backchannel + */ + +static struct rpc_xprt_ops bc_tcp_ops = { + .reserve_xprt = xprt_reserve_xprt, + .release_xprt = xprt_release_xprt, + .alloc_slot = xprt_alloc_slot, + .buf_alloc = bc_malloc, + .buf_free = bc_free, + .send_request = bc_send_request, + .set_retrans_timeout = xprt_set_retrans_timeout_def, + .close = bc_close, + .destroy = bc_destroy, + .print_stats = xs_tcp_print_stats, +}; + +static int xs_init_anyaddr(const int family, struct sockaddr *sap) +{ + static const struct sockaddr_in sin = { + .sin_family = AF_INET, + .sin_addr.s_addr = htonl(INADDR_ANY), + }; + static const struct sockaddr_in6 sin6 = { + .sin6_family = AF_INET6, + .sin6_addr = IN6ADDR_ANY_INIT, + }; + + switch (family) { + case AF_LOCAL: + break; + case AF_INET: + memcpy(sap, &sin, sizeof(sin)); + break; + case AF_INET6: + memcpy(sap, &sin6, sizeof(sin6)); + break; + default: + dprintk("RPC: %s: Bad address family\n", __func__); + return -EAFNOSUPPORT; + } + return 0; +} + +static struct rpc_xprt *xs_setup_xprt(struct xprt_create *args, + unsigned int slot_table_size, + unsigned int max_slot_table_size) +{ + struct rpc_xprt *xprt; + struct sock_xprt *new; + + if (args->addrlen > sizeof(xprt->addr)) { + dprintk("RPC: xs_setup_xprt: address too large\n"); + return ERR_PTR(-EBADF); + } + + xprt = xprt_alloc(args->net, sizeof(*new), slot_table_size, + max_slot_table_size); + if (xprt == NULL) { + dprintk("RPC: xs_setup_xprt: couldn't allocate " + "rpc_xprt\n"); + return ERR_PTR(-ENOMEM); + } + + new = container_of(xprt, struct sock_xprt, xprt); + memcpy(&xprt->addr, args->dstaddr, args->addrlen); + xprt->addrlen = args->addrlen; + if (args->srcaddr) + memcpy(&new->srcaddr, args->srcaddr, args->addrlen); + else { + int err; + err = xs_init_anyaddr(args->dstaddr->sa_family, + (struct sockaddr *)&new->srcaddr); + if (err != 0) { + xprt_free(xprt); + return ERR_PTR(err); + } + } + + return xprt; +} + +static const struct rpc_timeout xs_local_default_timeout = { + .to_initval = 10 * HZ, + .to_maxval = 10 * HZ, + .to_retries = 2, }; /** - * xs_setup_udp - Set up transport to use a UDP socket - * @xprt: transport to set up - * @to: timeout parameters + * xs_setup_local - Set up transport to use an AF_LOCAL socket + * @args: rpc transport creation arguments * + * AF_LOCAL is a "tpi_cots_ord" transport, just like TCP */ -int xs_setup_udp(struct rpc_xprt *xprt, struct rpc_timeout *to) +static struct rpc_xprt *xs_setup_local(struct xprt_create *args) { - size_t slot_table_size; + struct sockaddr_un *sun = (struct sockaddr_un *)args->dstaddr; + struct sock_xprt *transport; + struct rpc_xprt *xprt; + struct rpc_xprt *ret; + + xprt = xs_setup_xprt(args, xprt_tcp_slot_table_entries, + xprt_max_tcp_slot_table_entries); + if (IS_ERR(xprt)) + return xprt; + transport = container_of(xprt, struct sock_xprt, xprt); + + xprt->prot = 0; + xprt->tsh_size = sizeof(rpc_fraghdr) / sizeof(u32); + xprt->max_payload = RPC_MAX_FRAGMENT_SIZE; + + xprt->bind_timeout = XS_BIND_TO; + xprt->reestablish_timeout = XS_TCP_INIT_REEST_TO; + xprt->idle_timeout = XS_IDLE_DISC_TO; + + xprt->ops = &xs_local_ops; + xprt->timeout = &xs_local_default_timeout; + + INIT_DELAYED_WORK(&transport->connect_worker, + xs_dummy_setup_socket); + + switch (sun->sun_family) { + case AF_LOCAL: + if (sun->sun_path[0] != '/') { + dprintk("RPC: bad AF_LOCAL address: %s\n", + sun->sun_path); + ret = ERR_PTR(-EINVAL); + goto out_err; + } + xprt_set_bound(xprt); + xs_format_peer_addresses(xprt, "local", RPCBIND_NETID_LOCAL); + ret = ERR_PTR(xs_local_setup_socket(transport)); + if (ret) + goto out_err; + break; + default: + ret = ERR_PTR(-EAFNOSUPPORT); + goto out_err; + } + + dprintk("RPC: set up xprt to %s via AF_LOCAL\n", + xprt->address_strings[RPC_DISPLAY_ADDR]); - dprintk("RPC: setting up udp-ipv4 transport...\n"); + if (try_module_get(THIS_MODULE)) + return xprt; + ret = ERR_PTR(-EINVAL); +out_err: + xs_xprt_free(xprt); + return ret; +} - xprt->max_reqs = xprt_udp_slot_table_entries; - slot_table_size = xprt->max_reqs * sizeof(xprt->slot[0]); - xprt->slot = kmalloc(slot_table_size, GFP_KERNEL); - if (xprt->slot == NULL) - return -ENOMEM; - memset(xprt->slot, 0, slot_table_size); +static const struct rpc_timeout xs_udp_default_timeout = { + .to_initval = 5 * HZ, + .to_maxval = 30 * HZ, + .to_increment = 5 * HZ, + .to_retries = 5, +}; + +/** + * xs_setup_udp - Set up transport to use a UDP socket + * @args: rpc transport creation arguments + * + */ +static struct rpc_xprt *xs_setup_udp(struct xprt_create *args) +{ + struct sockaddr *addr = args->dstaddr; + struct rpc_xprt *xprt; + struct sock_xprt *transport; + struct rpc_xprt *ret; + + xprt = xs_setup_xprt(args, xprt_udp_slot_table_entries, + xprt_udp_slot_table_entries); + if (IS_ERR(xprt)) + return xprt; + transport = container_of(xprt, struct sock_xprt, xprt); xprt->prot = IPPROTO_UDP; - xprt->port = xprt_max_resvport; xprt->tsh_size = 0; - xprt->resvport = capable(CAP_NET_BIND_SERVICE) ? 1 : 0; /* XXX: header size can vary due to auth type, IPv6, etc. */ xprt->max_payload = (1U << 16) - (MAX_HEADER << 3); - INIT_WORK(&xprt->connect_worker, xs_udp_connect_worker, xprt); xprt->bind_timeout = XS_BIND_TO; - xprt->connect_timeout = XS_UDP_CONN_TO; xprt->reestablish_timeout = XS_UDP_REEST_TO; xprt->idle_timeout = XS_IDLE_DISC_TO; xprt->ops = &xs_udp_ops; - if (to) - xprt->timeout = *to; - else - xprt_set_timeout(&xprt->timeout, 5, 5 * HZ); + xprt->timeout = &xs_udp_default_timeout; - return 0; + switch (addr->sa_family) { + case AF_INET: + if (((struct sockaddr_in *)addr)->sin_port != htons(0)) + xprt_set_bound(xprt); + + INIT_DELAYED_WORK(&transport->connect_worker, + xs_udp_setup_socket); + xs_format_peer_addresses(xprt, "udp", RPCBIND_NETID_UDP); + break; + case AF_INET6: + if (((struct sockaddr_in6 *)addr)->sin6_port != htons(0)) + xprt_set_bound(xprt); + + INIT_DELAYED_WORK(&transport->connect_worker, + xs_udp_setup_socket); + xs_format_peer_addresses(xprt, "udp", RPCBIND_NETID_UDP6); + break; + default: + ret = ERR_PTR(-EAFNOSUPPORT); + goto out_err; + } + + if (xprt_bound(xprt)) + dprintk("RPC: set up xprt to %s (port %s) via %s\n", + xprt->address_strings[RPC_DISPLAY_ADDR], + xprt->address_strings[RPC_DISPLAY_PORT], + xprt->address_strings[RPC_DISPLAY_PROTO]); + else + dprintk("RPC: set up xprt to %s (autobind) via %s\n", + xprt->address_strings[RPC_DISPLAY_ADDR], + xprt->address_strings[RPC_DISPLAY_PROTO]); + + if (try_module_get(THIS_MODULE)) + return xprt; + ret = ERR_PTR(-EINVAL); +out_err: + xs_xprt_free(xprt); + return ret; } +static const struct rpc_timeout xs_tcp_default_timeout = { + .to_initval = 60 * HZ, + .to_maxval = 60 * HZ, + .to_retries = 2, +}; + /** * xs_setup_tcp - Set up transport to use a TCP socket - * @xprt: transport to set up - * @to: timeout parameters + * @args: rpc transport creation arguments * */ -int xs_setup_tcp(struct rpc_xprt *xprt, struct rpc_timeout *to) +static struct rpc_xprt *xs_setup_tcp(struct xprt_create *args) { - size_t slot_table_size; + struct sockaddr *addr = args->dstaddr; + struct rpc_xprt *xprt; + struct sock_xprt *transport; + struct rpc_xprt *ret; + unsigned int max_slot_table_size = xprt_max_tcp_slot_table_entries; - dprintk("RPC: setting up tcp-ipv4 transport...\n"); + if (args->flags & XPRT_CREATE_INFINITE_SLOTS) + max_slot_table_size = RPC_MAX_SLOT_TABLE_LIMIT; - xprt->max_reqs = xprt_tcp_slot_table_entries; - slot_table_size = xprt->max_reqs * sizeof(xprt->slot[0]); - xprt->slot = kmalloc(slot_table_size, GFP_KERNEL); - if (xprt->slot == NULL) - return -ENOMEM; - memset(xprt->slot, 0, slot_table_size); + xprt = xs_setup_xprt(args, xprt_tcp_slot_table_entries, + max_slot_table_size); + if (IS_ERR(xprt)) + return xprt; + transport = container_of(xprt, struct sock_xprt, xprt); xprt->prot = IPPROTO_TCP; - xprt->port = xprt_max_resvport; xprt->tsh_size = sizeof(rpc_fraghdr) / sizeof(u32); - xprt->resvport = capable(CAP_NET_BIND_SERVICE) ? 1 : 0; xprt->max_payload = RPC_MAX_FRAGMENT_SIZE; - INIT_WORK(&xprt->connect_worker, xs_tcp_connect_worker, xprt); xprt->bind_timeout = XS_BIND_TO; - xprt->connect_timeout = XS_TCP_CONN_TO; xprt->reestablish_timeout = XS_TCP_INIT_REEST_TO; xprt->idle_timeout = XS_IDLE_DISC_TO; xprt->ops = &xs_tcp_ops; + xprt->timeout = &xs_tcp_default_timeout; + + switch (addr->sa_family) { + case AF_INET: + if (((struct sockaddr_in *)addr)->sin_port != htons(0)) + xprt_set_bound(xprt); + + INIT_DELAYED_WORK(&transport->connect_worker, + xs_tcp_setup_socket); + xs_format_peer_addresses(xprt, "tcp", RPCBIND_NETID_TCP); + break; + case AF_INET6: + if (((struct sockaddr_in6 *)addr)->sin6_port != htons(0)) + xprt_set_bound(xprt); + + INIT_DELAYED_WORK(&transport->connect_worker, + xs_tcp_setup_socket); + xs_format_peer_addresses(xprt, "tcp", RPCBIND_NETID_TCP6); + break; + default: + ret = ERR_PTR(-EAFNOSUPPORT); + goto out_err; + } - if (to) - xprt->timeout = *to; + if (xprt_bound(xprt)) + dprintk("RPC: set up xprt to %s (port %s) via %s\n", + xprt->address_strings[RPC_DISPLAY_ADDR], + xprt->address_strings[RPC_DISPLAY_PORT], + xprt->address_strings[RPC_DISPLAY_PROTO]); else - xprt_set_timeout(&xprt->timeout, 2, 60 * HZ); + dprintk("RPC: set up xprt to %s (autobind) via %s\n", + xprt->address_strings[RPC_DISPLAY_ADDR], + xprt->address_strings[RPC_DISPLAY_PROTO]); + + if (try_module_get(THIS_MODULE)) + return xprt; + ret = ERR_PTR(-EINVAL); +out_err: + xs_xprt_free(xprt); + return ret; +} + +/** + * xs_setup_bc_tcp - Set up transport to use a TCP backchannel socket + * @args: rpc transport creation arguments + * + */ +static struct rpc_xprt *xs_setup_bc_tcp(struct xprt_create *args) +{ + struct sockaddr *addr = args->dstaddr; + struct rpc_xprt *xprt; + struct sock_xprt *transport; + struct svc_sock *bc_sock; + struct rpc_xprt *ret; + + xprt = xs_setup_xprt(args, xprt_tcp_slot_table_entries, + xprt_tcp_slot_table_entries); + if (IS_ERR(xprt)) + return xprt; + transport = container_of(xprt, struct sock_xprt, xprt); + + xprt->prot = IPPROTO_TCP; + xprt->tsh_size = sizeof(rpc_fraghdr) / sizeof(u32); + xprt->max_payload = RPC_MAX_FRAGMENT_SIZE; + xprt->timeout = &xs_tcp_default_timeout; + + /* backchannel */ + xprt_set_bound(xprt); + xprt->bind_timeout = 0; + xprt->reestablish_timeout = 0; + xprt->idle_timeout = 0; + + xprt->ops = &bc_tcp_ops; + + switch (addr->sa_family) { + case AF_INET: + xs_format_peer_addresses(xprt, "tcp", + RPCBIND_NETID_TCP); + break; + case AF_INET6: + xs_format_peer_addresses(xprt, "tcp", + RPCBIND_NETID_TCP6); + break; + default: + ret = ERR_PTR(-EAFNOSUPPORT); + goto out_err; + } + + dprintk("RPC: set up xprt to %s (port %s) via %s\n", + xprt->address_strings[RPC_DISPLAY_ADDR], + xprt->address_strings[RPC_DISPLAY_PORT], + xprt->address_strings[RPC_DISPLAY_PROTO]); + + /* + * Once we've associated a backchannel xprt with a connection, + * we want to keep it around as long as the connection lasts, + * in case we need to start using it for a backchannel again; + * this reference won't be dropped until bc_xprt is destroyed. + */ + xprt_get(xprt); + args->bc_xprt->xpt_bc_xprt = xprt; + xprt->bc_xprt = args->bc_xprt; + bc_sock = container_of(args->bc_xprt, struct svc_sock, sk_xprt); + transport->sock = bc_sock->sk_sock; + transport->inet = bc_sock->sk_sk; + /* + * Since we don't want connections for the backchannel, we set + * the xprt status to connected + */ + xprt_set_connected(xprt); + + if (try_module_get(THIS_MODULE)) + return xprt; + + args->bc_xprt->xpt_bc_xprt = NULL; + xprt_put(xprt); + ret = ERR_PTR(-EINVAL); +out_err: + xs_xprt_free(xprt); + return ret; +} + +static struct xprt_class xs_local_transport = { + .list = LIST_HEAD_INIT(xs_local_transport.list), + .name = "named UNIX socket", + .owner = THIS_MODULE, + .ident = XPRT_TRANSPORT_LOCAL, + .setup = xs_setup_local, +}; + +static struct xprt_class xs_udp_transport = { + .list = LIST_HEAD_INIT(xs_udp_transport.list), + .name = "udp", + .owner = THIS_MODULE, + .ident = XPRT_TRANSPORT_UDP, + .setup = xs_setup_udp, +}; + +static struct xprt_class xs_tcp_transport = { + .list = LIST_HEAD_INIT(xs_tcp_transport.list), + .name = "tcp", + .owner = THIS_MODULE, + .ident = XPRT_TRANSPORT_TCP, + .setup = xs_setup_tcp, +}; + +static struct xprt_class xs_bc_tcp_transport = { + .list = LIST_HEAD_INIT(xs_bc_tcp_transport.list), + .name = "tcp NFSv4.1 backchannel", + .owner = THIS_MODULE, + .ident = XPRT_TRANSPORT_BC_TCP, + .setup = xs_setup_bc_tcp, +}; + +/** + * init_socket_xprt - set up xprtsock's sysctls, register with RPC client + * + */ +int init_socket_xprt(void) +{ +#ifdef RPC_DEBUG + if (!sunrpc_table_header) + sunrpc_table_header = register_sysctl_table(sunrpc_table); +#endif + + xprt_register_transport(&xs_local_transport); + xprt_register_transport(&xs_udp_transport); + xprt_register_transport(&xs_tcp_transport); + xprt_register_transport(&xs_bc_tcp_transport); + + return 0; +} + +/** + * cleanup_socket_xprt - remove xprtsock's sysctls, unregister + * + */ +void cleanup_socket_xprt(void) +{ +#ifdef RPC_DEBUG + if (sunrpc_table_header) { + unregister_sysctl_table(sunrpc_table_header); + sunrpc_table_header = NULL; + } +#endif + + xprt_unregister_transport(&xs_local_transport); + xprt_unregister_transport(&xs_udp_transport); + xprt_unregister_transport(&xs_tcp_transport); + xprt_unregister_transport(&xs_bc_tcp_transport); +} + +static int param_set_uint_minmax(const char *val, + const struct kernel_param *kp, + unsigned int min, unsigned int max) +{ + unsigned long num; + int ret; + + if (!val) + return -EINVAL; + ret = strict_strtoul(val, 0, &num); + if (ret == -EINVAL || num < min || num > max) + return -EINVAL; + *((unsigned int *)kp->arg) = num; return 0; } + +static int param_set_portnr(const char *val, const struct kernel_param *kp) +{ + return param_set_uint_minmax(val, kp, + RPC_MIN_RESVPORT, + RPC_MAX_RESVPORT); +} + +static struct kernel_param_ops param_ops_portnr = { + .set = param_set_portnr, + .get = param_get_uint, +}; + +#define param_check_portnr(name, p) \ + __param_check(name, p, unsigned int); + +module_param_named(min_resvport, xprt_min_resvport, portnr, 0644); +module_param_named(max_resvport, xprt_max_resvport, portnr, 0644); + +static int param_set_slot_table_size(const char *val, + const struct kernel_param *kp) +{ + return param_set_uint_minmax(val, kp, + RPC_MIN_SLOT_TABLE, + RPC_MAX_SLOT_TABLE); +} + +static struct kernel_param_ops param_ops_slot_table_size = { + .set = param_set_slot_table_size, + .get = param_get_uint, +}; + +#define param_check_slot_table_size(name, p) \ + __param_check(name, p, unsigned int); + +static int param_set_max_slot_table_size(const char *val, + const struct kernel_param *kp) +{ + return param_set_uint_minmax(val, kp, + RPC_MIN_SLOT_TABLE, + RPC_MAX_SLOT_TABLE_LIMIT); +} + +static struct kernel_param_ops param_ops_max_slot_table_size = { + .set = param_set_max_slot_table_size, + .get = param_get_uint, +}; + +#define param_check_max_slot_table_size(name, p) \ + __param_check(name, p, unsigned int); + +module_param_named(tcp_slot_table_entries, xprt_tcp_slot_table_entries, + slot_table_size, 0644); +module_param_named(tcp_max_slot_table_entries, xprt_max_tcp_slot_table_entries, + max_slot_table_size, 0644); +module_param_named(udp_slot_table_entries, xprt_udp_slot_table_entries, + slot_table_size, 0644); + |