1 files changed, 399 insertions, 108 deletions

diff --git a/fs/nfs/callback.c b/fs/nfs/callback.c
index fcd97406a77..073b4cf67ed 100644
--- a/fs/nfs/callback.c
+++ b/fs/nfs/callback.c
@@ -6,166 +6,458 @@
  * NFSv4 callback handling
  */
 
-#include <linux/config.h>
 #include <linux/completion.h>
 #include <linux/ip.h>
 #include <linux/module.h>
-#include <linux/smp_lock.h>
 #include <linux/sunrpc/svc.h>
 #include <linux/sunrpc/svcsock.h>
 #include <linux/nfs_fs.h>
+#include <linux/errno.h>
+#include <linux/mutex.h>
+#include <linux/freezer.h>
+#include <linux/kthread.h>
+#include <linux/sunrpc/svcauth_gss.h>
+#include <linux/sunrpc/bc_xprt.h>
 
 #include <net/inet_sock.h>
 
 #include "nfs4_fs.h"
 #include "callback.h"
+#include "internal.h"
+#include "netns.h"
 
 #define NFSDBG_FACILITY NFSDBG_CALLBACK
 
 struct nfs_callback_data {
 	unsigned int users;
 	struct svc_serv *serv;
-	pid_t pid;
-	struct completion started;
-	struct completion stopped;
+	struct svc_rqst *rqst;
+	struct task_struct *task;
 };
 
-static struct nfs_callback_data nfs_callback_info;
-static DECLARE_MUTEX(nfs_callback_sema);
+static struct nfs_callback_data nfs_callback_info[NFS4_MAX_MINOR_VERSION + 1];
+static DEFINE_MUTEX(nfs_callback_mutex);
 static struct svc_program nfs4_callback_program;
 
-unsigned int nfs_callback_set_tcpport;
-unsigned short nfs_callback_tcpport;
+static int nfs4_callback_up_net(struct svc_serv *serv, struct net *net)
+{
+	int ret;
+	struct nfs_net *nn = net_generic(net, nfs_net_id);
+
+	ret = svc_create_xprt(serv, "tcp", net, PF_INET,
+				nfs_callback_set_tcpport, SVC_SOCK_ANONYMOUS);
+	if (ret <= 0)
+		goto out_err;
+	nn->nfs_callback_tcpport = ret;
+	dprintk("NFS: Callback listener port = %u (af %u, net %p)\n",
+			nn->nfs_callback_tcpport, PF_INET, net);
+
+	ret = svc_create_xprt(serv, "tcp", net, PF_INET6,
+				nfs_callback_set_tcpport, SVC_SOCK_ANONYMOUS);
+	if (ret > 0) {
+		nn->nfs_callback_tcpport6 = ret;
+		dprintk("NFS: Callback listener port = %u (af %u, net %p)\n",
+				nn->nfs_callback_tcpport6, PF_INET6, net);
+	} else if (ret != -EAFNOSUPPORT)
+		goto out_err;
+	return 0;
+
+out_err:
+	return (ret) ? ret : -ENOMEM;
+}
 
 /*
- * This is the callback kernel thread.
+ * This is the NFSv4 callback kernel thread.
  */
-static void nfs_callback_svc(struct svc_rqst *rqstp)
+static int
+nfs4_callback_svc(void *vrqstp)
 {
-	struct svc_serv *serv = rqstp->rq_server;
 	int err;
+	struct svc_rqst *rqstp = vrqstp;
 
-	__module_get(THIS_MODULE);
-	lock_kernel();
-
-	nfs_callback_info.pid = current->pid;
-	daemonize("nfsv4-svc");
-	/* Process request with signals blocked, but allow SIGKILL.  */
-	allow_signal(SIGKILL);
+	set_freezable();
 
-	complete(&nfs_callback_info.started);
-
-	while (nfs_callback_info.users != 0 || !signalled()) {
+	while (!kthread_should_stop()) {
 		/*
 		 * Listen for a request on the socket
 		 */
-		err = svc_recv(serv, rqstp, MAX_SCHEDULE_TIMEOUT);
+		err = svc_recv(rqstp, MAX_SCHEDULE_TIMEOUT);
 		if (err == -EAGAIN || err == -EINTR)
 			continue;
-		if (err < 0) {
-			printk(KERN_WARNING
-					"%s: terminating on error %d\n",
-					__FUNCTION__, -err);
-			break;
+		svc_process(rqstp);
+	}
+	return 0;
+}
+
+/*
+ * Prepare to bring up the NFSv4 callback service
+ */
+static struct svc_rqst *
+nfs4_callback_up(struct svc_serv *serv)
+{
+	return svc_prepare_thread(serv, &serv->sv_pools[0], NUMA_NO_NODE);
+}
+
+#if defined(CONFIG_NFS_V4_1)
+static int nfs41_callback_up_net(struct svc_serv *serv, struct net *net)
+{
+	/*
+	 * Create an svc_sock for the back channel service that shares the
+	 * fore channel connection.
+	 * Returns the input port (0) and sets the svc_serv bc_xprt on success
+	 */
+	return svc_create_xprt(serv, "tcp-bc", net, PF_INET, 0,
+			      SVC_SOCK_ANONYMOUS);
+}
+
+/*
+ * The callback service for NFSv4.1 callbacks
+ */
+static int
+nfs41_callback_svc(void *vrqstp)
+{
+	struct svc_rqst *rqstp = vrqstp;
+	struct svc_serv *serv = rqstp->rq_server;
+	struct rpc_rqst *req;
+	int error;
+	DEFINE_WAIT(wq);
+
+	set_freezable();
+
+	while (!kthread_should_stop()) {
+		if (try_to_freeze())
+			continue;
+
+		prepare_to_wait(&serv->sv_cb_waitq, &wq, TASK_INTERRUPTIBLE);
+		spin_lock_bh(&serv->sv_cb_lock);
+		if (!list_empty(&serv->sv_cb_list)) {
+			req = list_first_entry(&serv->sv_cb_list,
+					struct rpc_rqst, rq_bc_list);
+			list_del(&req->rq_bc_list);
+			spin_unlock_bh(&serv->sv_cb_lock);
+			dprintk("Invoking bc_svc_process()\n");
+			error = bc_svc_process(serv, req, rqstp);
+			dprintk("bc_svc_process() returned w/ error code= %d\n",
+				error);
+		} else {
+			spin_unlock_bh(&serv->sv_cb_lock);
+			schedule();
 		}
-		dprintk("%s: request from %u.%u.%u.%u\n", __FUNCTION__,
-				NIPQUAD(rqstp->rq_addr.sin_addr.s_addr));
-		svc_process(serv, rqstp);
+		finish_wait(&serv->sv_cb_waitq, &wq);
+	}
+	return 0;
+}
+
+/*
+ * Bring up the NFSv4.1 callback service
+ */
+static struct svc_rqst *
+nfs41_callback_up(struct svc_serv *serv)
+{
+	struct svc_rqst *rqstp;
+
+	INIT_LIST_HEAD(&serv->sv_cb_list);
+	spin_lock_init(&serv->sv_cb_lock);
+	init_waitqueue_head(&serv->sv_cb_waitq);
+	rqstp = svc_prepare_thread(serv, &serv->sv_pools[0], NUMA_NO_NODE);
+	if (IS_ERR(rqstp)) {
+		svc_xprt_put(serv->sv_bc_xprt);
+		serv->sv_bc_xprt = NULL;
+	}
+	dprintk("--> %s return %d\n", __func__, PTR_ERR_OR_ZERO(rqstp));
+	return rqstp;
+}
+
+static void nfs_minorversion_callback_svc_setup(struct svc_serv *serv,
+		struct svc_rqst **rqstpp, int (**callback_svc)(void *vrqstp))
+{
+	*rqstpp = nfs41_callback_up(serv);
+	*callback_svc = nfs41_callback_svc;
+}
+
+static inline void nfs_callback_bc_serv(u32 minorversion, struct rpc_xprt *xprt,
+		struct svc_serv *serv)
+{
+	if (minorversion)
+		/*
+		 * Save the svc_serv in the transport so that it can
+		 * be referenced when the session backchannel is initialized
+		 */
+		xprt->bc_serv = serv;
+}
+#else
+static int nfs41_callback_up_net(struct svc_serv *serv, struct net *net)
+{
+	return 0;
+}
+
+static void nfs_minorversion_callback_svc_setup(struct svc_serv *serv,
+		struct svc_rqst **rqstpp, int (**callback_svc)(void *vrqstp))
+{
+	*rqstpp = ERR_PTR(-ENOTSUPP);
+	*callback_svc = ERR_PTR(-ENOTSUPP);
+}
+
+static inline void nfs_callback_bc_serv(u32 minorversion, struct rpc_xprt *xprt,
+		struct svc_serv *serv)
+{
+}
+#endif /* CONFIG_NFS_V4_1 */
+
+static int nfs_callback_start_svc(int minorversion, struct rpc_xprt *xprt,
+				  struct svc_serv *serv)
+{
+	struct svc_rqst *rqstp;
+	int (*callback_svc)(void *vrqstp);
+	struct nfs_callback_data *cb_info = &nfs_callback_info[minorversion];
+	int ret;
+
+	nfs_callback_bc_serv(minorversion, xprt, serv);
+
+	if (cb_info->task)
+		return 0;
+
+	switch (minorversion) {
+	case 0:
+		/* v4.0 callback setup */
+		rqstp = nfs4_callback_up(serv);
+		callback_svc = nfs4_callback_svc;
+		break;
+	default:
+		nfs_minorversion_callback_svc_setup(serv,
+				&rqstp, &callback_svc);
+	}
+
+	if (IS_ERR(rqstp))
+		return PTR_ERR(rqstp);
+
+	svc_sock_update_bufs(serv);
+
+	cb_info->serv = serv;
+	cb_info->rqst = rqstp;
+	cb_info->task = kthread_run(callback_svc, cb_info->rqst,
+				    "nfsv4.%u-svc", minorversion);
+	if (IS_ERR(cb_info->task)) {
+		ret = PTR_ERR(cb_info->task);
+		svc_exit_thread(cb_info->rqst);
+		cb_info->rqst = NULL;
+		cb_info->task = NULL;
+		return ret;
+	}
+	dprintk("nfs_callback_up: service started\n");
+	return 0;
+}
+
+static void nfs_callback_down_net(u32 minorversion, struct svc_serv *serv, struct net *net)
+{
+	struct nfs_net *nn = net_generic(net, nfs_net_id);
+
+	if (--nn->cb_users[minorversion])
+		return;
+
+	dprintk("NFS: destroy per-net callback data; net=%p\n", net);
+	svc_shutdown_net(serv, net);
+}
+
+static int nfs_callback_up_net(int minorversion, struct svc_serv *serv, struct net *net)
+{
+	struct nfs_net *nn = net_generic(net, nfs_net_id);
+	int ret;
+
+	if (nn->cb_users[minorversion]++)
+		return 0;
+
+	dprintk("NFS: create per-net callback data; net=%p\n", net);
+
+	ret = svc_bind(serv, net);
+	if (ret < 0) {
+		printk(KERN_WARNING "NFS: bind callback service failed\n");
+		goto err_bind;
+	}
+
+	switch (minorversion) {
+		case 0:
+			ret = nfs4_callback_up_net(serv, net);
+			break;
+		case 1:
+		case 2:
+			ret = nfs41_callback_up_net(serv, net);
+			break;
+		default:
+			printk(KERN_ERR "NFS: unknown callback version: %d\n",
+					minorversion);
+			ret = -EINVAL;
+			break;
+	}
+
+	if (ret < 0) {
+		printk(KERN_ERR "NFS: callback service start failed\n");
+		goto err_socks;
+	}
+	return 0;
+
+err_socks:
+	svc_rpcb_cleanup(serv, net);
+err_bind:
+	dprintk("NFS: Couldn't create callback socket: err = %d; "
+			"net = %p\n", ret, net);
+	return ret;
+}
+
+static struct svc_serv *nfs_callback_create_svc(int minorversion)
+{
+	struct nfs_callback_data *cb_info = &nfs_callback_info[minorversion];
+	struct svc_serv *serv;
+
+	/*
+	 * Check whether we're already up and running.
+	 */
+	if (cb_info->task) {
+		/*
+		 * Note: increase service usage, because later in case of error
+		 * svc_destroy() will be called.
+		 */
+		svc_get(cb_info->serv);
+		return cb_info->serv;
 	}
 
-	nfs_callback_info.pid = 0;
-	complete(&nfs_callback_info.stopped);
-	unlock_kernel();
-	module_put_and_exit(0);
+	/*
+	 * Sanity check: if there's no task,
+	 * we should be the first user ...
+	 */
+	if (cb_info->users)
+		printk(KERN_WARNING "nfs_callback_create_svc: no kthread, %d users??\n",
+			cb_info->users);
+
+	serv = svc_create(&nfs4_callback_program, NFS4_CALLBACK_BUFSIZE, NULL);
+	if (!serv) {
+		printk(KERN_ERR "nfs_callback_create_svc: create service failed\n");
+		return ERR_PTR(-ENOMEM);
+	}
+	/* As there is only one thread we need to over-ride the
+	 * default maximum of 80 connections
+	 */
+	serv->sv_maxconn = 1024;
+	dprintk("nfs_callback_create_svc: service created\n");
+	return serv;
 }
 
 /*
- * Bring up the server process if it is not already up.
+ * Bring up the callback thread if it is not already up.
  */
-int nfs_callback_up(void)
+int nfs_callback_up(u32 minorversion, struct rpc_xprt *xprt)
 {
 	struct svc_serv *serv;
-	struct svc_sock *svsk;
-	int ret = 0;
-
-	lock_kernel();
-	down(&nfs_callback_sema);
-	if (nfs_callback_info.users++ || nfs_callback_info.pid != 0)
-		goto out;
-	init_completion(&nfs_callback_info.started);
-	init_completion(&nfs_callback_info.stopped);
-	serv = svc_create(&nfs4_callback_program, NFS4_CALLBACK_BUFSIZE);
-	ret = -ENOMEM;
-	if (!serv)
-		goto out_err;
-	/* FIXME: We don't want to register this socket with the portmapper */
-	ret = svc_makesock(serv, IPPROTO_TCP, nfs_callback_set_tcpport);
+	struct nfs_callback_data *cb_info = &nfs_callback_info[minorversion];
+	int ret;
+	struct net *net = xprt->xprt_net;
+
+	mutex_lock(&nfs_callback_mutex);
+
+	serv = nfs_callback_create_svc(minorversion);
+	if (IS_ERR(serv)) {
+		ret = PTR_ERR(serv);
+		goto err_create;
+	}
+
+	ret = nfs_callback_up_net(minorversion, serv, net);
 	if (ret < 0)
-		goto out_destroy;
-	if (!list_empty(&serv->sv_permsocks)) {
-		svsk = list_entry(serv->sv_permsocks.next,
-				struct svc_sock, sk_list);
-		nfs_callback_tcpport = ntohs(inet_sk(svsk->sk_sk)->sport);
-		dprintk ("Callback port = 0x%x\n", nfs_callback_tcpport);
-	} else
-		BUG();
-	ret = svc_create_thread(nfs_callback_svc, serv);
+		goto err_net;
+
+	ret = nfs_callback_start_svc(minorversion, xprt, serv);
 	if (ret < 0)
-		goto out_destroy;
-	nfs_callback_info.serv = serv;
-	wait_for_completion(&nfs_callback_info.started);
-out:
-	up(&nfs_callback_sema);
-	unlock_kernel();
-	return ret;
-out_destroy:
+		goto err_start;
+
+	cb_info->users++;
+	/*
+	 * svc_create creates the svc_serv with sv_nrthreads == 1, and then
+	 * svc_prepare_thread increments that. So we need to call svc_destroy
+	 * on both success and failure so that the refcount is 1 when the
+	 * thread exits.
+	 */
+err_net:
 	svc_destroy(serv);
-out_err:
-	nfs_callback_info.users--;
-	goto out;
+err_create:
+	mutex_unlock(&nfs_callback_mutex);
+	return ret;
+
+err_start:
+	nfs_callback_down_net(minorversion, serv, net);
+	dprintk("NFS: Couldn't create server thread; err = %d\n", ret);
+	goto err_net;
 }
 
 /*
- * Kill the server process if it is not already up.
+ * Kill the callback thread if it's no longer being used.
  */
-int nfs_callback_down(void)
-{
-	int ret = 0;
-
-	lock_kernel();
-	down(&nfs_callback_sema);
-	if (--nfs_callback_info.users || nfs_callback_info.pid == 0)
-		goto out;
-	kill_proc(nfs_callback_info.pid, SIGKILL, 1);
-	wait_for_completion(&nfs_callback_info.stopped);
-out:
-	up(&nfs_callback_sema);
-	unlock_kernel();
-	return ret;
+void nfs_callback_down(int minorversion, struct net *net)
+{
+	struct nfs_callback_data *cb_info = &nfs_callback_info[minorversion];
+
+	mutex_lock(&nfs_callback_mutex);
+	nfs_callback_down_net(minorversion, cb_info->serv, net);
+	cb_info->users--;
+	if (cb_info->users == 0 && cb_info->task != NULL) {
+		kthread_stop(cb_info->task);
+		dprintk("nfs_callback_down: service stopped\n");
+		svc_exit_thread(cb_info->rqst);
+		dprintk("nfs_callback_down: service destroyed\n");
+		cb_info->serv = NULL;
+		cb_info->rqst = NULL;
+		cb_info->task = NULL;
+	}
+	mutex_unlock(&nfs_callback_mutex);
 }
 
-static int nfs_callback_authenticate(struct svc_rqst *rqstp)
+/* Boolean check of RPC_AUTH_GSS principal */
+int
+check_gss_callback_principal(struct nfs_client *clp, struct svc_rqst *rqstp)
 {
-	struct in_addr *addr = &rqstp->rq_addr.sin_addr;
-	struct nfs4_client *clp;
+	char *p = rqstp->rq_cred.cr_principal;
+
+	if (rqstp->rq_authop->flavour != RPC_AUTH_GSS)
+		return 1;
+
+	/* No RPC_AUTH_GSS on NFSv4.1 back channel yet */
+	if (clp->cl_minorversion != 0)
+		return 0;
+	/*
+	 * It might just be a normal user principal, in which case
+	 * userspace won't bother to tell us the name at all.
+	 */
+	if (p == NULL)
+		return 0;
+
+	/* Expect a GSS_C_NT_HOSTBASED_NAME like "nfs@serverhostname" */
 
-	/* Don't talk to strangers */
-	clp = nfs4_find_client(addr);
-	if (clp == NULL)
-		return SVC_DROP;
-	dprintk("%s: %u.%u.%u.%u NFSv4 callback!\n", __FUNCTION__, NIPQUAD(addr));
-	nfs4_put_client(clp);
+	if (memcmp(p, "nfs@", 4) != 0)
+		return 0;
+	p += 4;
+	if (strcmp(p, clp->cl_hostname) != 0)
+		return 0;
+	return 1;
+}
+
+/*
+ * pg_authenticate method for nfsv4 callback threads.
+ *
+ * The authflavor has been negotiated, so an incorrect flavor is a server
+ * bug. Drop packets with incorrect authflavor.
+ *
+ * All other checking done after NFS decoding where the nfs_client can be
+ * found in nfs4_callback_compound
+ */
+static int nfs_callback_authenticate(struct svc_rqst *rqstp)
+{
 	switch (rqstp->rq_authop->flavour) {
-		case RPC_AUTH_NULL:
-			if (rqstp->rq_proc != CB_NULL)
-				return SVC_DENIED;
-			break;
-		case RPC_AUTH_UNIX:
-			break;
-		case RPC_AUTH_GSS:
-			/* FIXME: RPCSEC_GSS handling? */
-		default:
-			return SVC_DENIED;
+	case RPC_AUTH_NULL:
+		if (rqstp->rq_proc != CB_NULL)
+			return SVC_DROP;
+		break;
+	case RPC_AUTH_GSS:
+		/* No RPC_AUTH_GSS support yet in NFSv4.1 */
+		 if (svc_is_backchannel(rqstp))
+			return SVC_DROP;
 	}
 	return SVC_OK;
 }
@@ -173,10 +465,9 @@ static int nfs_callback_authenticate(struct svc_rqst *rqstp)
 /*
  * Define NFS4 callback program
  */
-extern struct svc_version nfs4_callback_version1;
-
 static struct svc_version *nfs4_callback_version[] = {
 	[1] = &nfs4_callback_version1,
+	[4] = &nfs4_callback_version4,
 };
 
 static struct svc_stat nfs4_callback_stats;