From 9208faf297dddfa97a86d7224b6bf94f2e346dd9 Mon Sep 17 00:00:00 2001 From: Yu Zhiguo Date: Mon, 6 Jul 2009 17:24:16 +0800 Subject: NFSv4: ACL in operations 'open' and 'create' should be used ACL in operations 'open' and 'create' is decoded but never be used. It should be set as the initial ACL for the object according to RFC3530. If error occurs when setting the ACL, just clear the ACL bit in the returned attr bitmap. Signed-off-by: Yu Zhiguo Signed-off-by: J. Bruce Fields --- fs/nfsd/nfs4proc.c | 47 ++++++++++++++++++++++++++++++++++++++++++----- 1 file changed, 42 insertions(+), 5 deletions(-) (limited to 'fs') diff --git a/fs/nfsd/nfs4proc.c b/fs/nfsd/nfs4proc.c index 7c8801769a3..d781658e808 100644 --- a/fs/nfsd/nfs4proc.c +++ b/fs/nfsd/nfs4proc.c @@ -123,6 +123,35 @@ nfsd4_check_open_attributes(struct svc_rqst *rqstp, return status; } +static int +is_create_with_attrs(struct nfsd4_open *open) +{ + return open->op_create == NFS4_OPEN_CREATE + && (open->op_createmode == NFS4_CREATE_UNCHECKED + || open->op_createmode == NFS4_CREATE_GUARDED + || open->op_createmode == NFS4_CREATE_EXCLUSIVE4_1); +} + +/* + * if error occurs when setting the acl, just clear the acl bit + * in the returned attr bitmap. + */ +static void +do_set_nfs4_acl(struct svc_rqst *rqstp, struct svc_fh *fhp, + struct nfs4_acl *acl, u32 *bmval) +{ + __be32 status; + + status = nfsd4_set_nfs4_acl(rqstp, fhp, acl); + if (status) + /* + * We should probably fail the whole open at this point, + * but we've already created the file, so it's too late; + * So this seems the least of evils: + */ + bmval[0] &= ~FATTR4_WORD0_ACL; +} + static inline void fh_dup2(struct svc_fh *dst, struct svc_fh *src) { @@ -206,6 +235,9 @@ do_open_lookup(struct svc_rqst *rqstp, struct svc_fh *current_fh, struct nfsd4_o if (status) goto out; + if (is_create_with_attrs(open) && open->op_acl != NULL) + do_set_nfs4_acl(rqstp, &resfh, open->op_acl, open->op_bmval); + set_change_info(&open->op_cinfo, current_fh); fh_dup2(current_fh, &resfh); @@ -536,12 +568,17 @@ nfsd4_create(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate, status = nfserr_badtype; } - if (!status) { - fh_unlock(&cstate->current_fh); - set_change_info(&create->cr_cinfo, &cstate->current_fh); - fh_dup2(&cstate->current_fh, &resfh); - } + if (status) + goto out; + if (create->cr_acl != NULL) + do_set_nfs4_acl(rqstp, &resfh, create->cr_acl, + create->cr_bmval); + + fh_unlock(&cstate->current_fh); + set_change_info(&create->cr_cinfo, &cstate->current_fh); + fh_dup2(&cstate->current_fh, &resfh); +out: fh_put(&resfh); return status; } -- cgit v1.2.3-18-g5258 From 4bd9b0f4afc76cf972578c702e1bc1b6f2d10ba5 Mon Sep 17 00:00:00 2001 From: Andy Adamson Date: Wed, 24 Jun 2009 15:37:45 -0400 Subject: nfsd41: use globals for DRC limits The version 4.1 DRC memory limit and tracking variables are server wide and session specific. Replace struct svc_serv fields with globals. Stop using the svc_serv sv_lock. Add a spinlock to serialize access to the DRC limit management variables which change on session creation and deletion (usage counter) or (future) administrative action to adjust the total DRC memory limit. Signed-off-by: Andy Adamson Signed-off-by: Benny Halevy --- fs/nfsd/nfs4state.c | 10 +++++----- fs/nfsd/nfssvc.c | 19 +++++++++++++++---- 2 files changed, 20 insertions(+), 9 deletions(-) (limited to 'fs') diff --git a/fs/nfsd/nfs4state.c b/fs/nfsd/nfs4state.c index 980a216a48c..2e6a44e3d2f 100644 --- a/fs/nfsd/nfs4state.c +++ b/fs/nfsd/nfs4state.c @@ -430,11 +430,11 @@ static int set_forechannel_maxreqs(struct nfsd4_channel_attrs *fchan) else if (fchan->maxreqs > NFSD_MAX_SLOTS_PER_SESSION) fchan->maxreqs = NFSD_MAX_SLOTS_PER_SESSION; - spin_lock(&nfsd_serv->sv_lock); - if (np + nfsd_serv->sv_drc_pages_used > nfsd_serv->sv_drc_max_pages) - np = nfsd_serv->sv_drc_max_pages - nfsd_serv->sv_drc_pages_used; - nfsd_serv->sv_drc_pages_used += np; - spin_unlock(&nfsd_serv->sv_lock); + spin_lock(&nfsd_drc_lock); + if (np + nfsd_drc_pages_used > nfsd_drc_max_pages) + np = nfsd_drc_max_pages - nfsd_drc_pages_used; + nfsd_drc_pages_used += np; + spin_unlock(&nfsd_drc_lock); if (np <= 0) { status = nfserr_resource; diff --git a/fs/nfsd/nfssvc.c b/fs/nfsd/nfssvc.c index d4c9884cd54..78d8fcd883f 100644 --- a/fs/nfsd/nfssvc.c +++ b/fs/nfsd/nfssvc.c @@ -67,6 +67,16 @@ struct timeval nfssvc_boot; DEFINE_MUTEX(nfsd_mutex); struct svc_serv *nfsd_serv; +/* + * nfsd_drc_lock protects nfsd_drc_max_pages and nfsd_drc_pages_used. + * nfsd_drc_max_pages limits the total amount of memory available for + * version 4.1 DRC caches. + * nfsd_drc_pages_used tracks the current version 4.1 DRC memory usage. + */ +spinlock_t nfsd_drc_lock; +unsigned int nfsd_drc_max_pages; +unsigned int nfsd_drc_pages_used; + #if defined(CONFIG_NFSD_V2_ACL) || defined(CONFIG_NFSD_V3_ACL) static struct svc_stat nfsd_acl_svcstats; static struct svc_version * nfsd_acl_version[] = { @@ -238,11 +248,12 @@ static void set_max_drc(void) { /* The percent of nr_free_buffer_pages used by the V4.1 server DRC */ #define NFSD_DRC_SIZE_SHIFT 7 - nfsd_serv->sv_drc_max_pages = nr_free_buffer_pages() + nfsd_drc_max_pages = nr_free_buffer_pages() >> NFSD_DRC_SIZE_SHIFT; - nfsd_serv->sv_drc_pages_used = 0; - dprintk("%s svc_drc_max_pages %u\n", __func__, - nfsd_serv->sv_drc_max_pages); + nfsd_drc_pages_used = 0; + spin_lock_init(&nfsd_drc_lock); + dprintk("%s nfsd_drc_max_pages %u\n", __func__, + nfsd_drc_max_pages); } int nfsd_create_serv(void) -- cgit v1.2.3-18-g5258 From 2522a776c1b9b5c93383d07717f895cc1a56a87a Mon Sep 17 00:00:00 2001 From: Eric Sesterhenn Date: Tue, 28 Jul 2009 14:32:08 +0200 Subject: Fix memory leak in write_pool_threads kmemleak produces the following warning unreferenced object 0xc9ec02a0 (size 8): comm "cat", pid 19048, jiffies 730243 backtrace: [] create_object+0x100/0x240 [] kmemleak_alloc+0x2b/0x60 [] __kmalloc+0x14b/0x270 [] write_pool_threads+0x87/0x1d0 [] nfsctl_transaction_write+0x58/0x70 [] nfsctl_transaction_read+0x4f/0x60 [] vfs_read+0x94/0x150 [] sys_read+0x3d/0x70 [] sysenter_do_call+0x12/0x32 [] 0xffffffff write_pool_threads() only frees nthreads on error paths, in the success case we leak it. Signed-off-by: Eric Sesterhenn Reviewed-by: Catalin Marinas Signed-off-by: J. Bruce Fields --- fs/nfsd/nfsctl.c | 1 + 1 file changed, 1 insertion(+) (limited to 'fs') diff --git a/fs/nfsd/nfsctl.c b/fs/nfsd/nfsctl.c index 1250fb978ac..48da164bb59 100644 --- a/fs/nfsd/nfsctl.c +++ b/fs/nfsd/nfsctl.c @@ -785,6 +785,7 @@ static ssize_t write_pool_threads(struct file *file, char *buf, size_t size) mesg += len; } + kfree(nthreads); mutex_unlock(&nfsd_mutex); return (mesg-buf); -- cgit v1.2.3-18-g5258 From 413d63d7106b914a4a004ac08698f10c618e4616 Mon Sep 17 00:00:00 2001 From: "J. Bruce Fields" Date: Tue, 28 Jul 2009 11:37:25 -0400 Subject: nfsd: minor write_pool_threads exit cleanup Signed-off-by: J. Bruce Fields --- fs/nfsd/nfsctl.c | 6 +----- 1 file changed, 1 insertion(+), 5 deletions(-) (limited to 'fs') diff --git a/fs/nfsd/nfsctl.c b/fs/nfsd/nfsctl.c index 48da164bb59..b51e7ae8b57 100644 --- a/fs/nfsd/nfsctl.c +++ b/fs/nfsd/nfsctl.c @@ -784,11 +784,7 @@ static ssize_t write_pool_threads(struct file *file, char *buf, size_t size) size -= len; mesg += len; } - - kfree(nthreads); - mutex_unlock(&nfsd_mutex); - return (mesg-buf); - + rv = mesg - buf; out_free: kfree(nthreads); mutex_unlock(&nfsd_mutex); -- cgit v1.2.3-18-g5258 From be98d1bbd1b872a10d64cdef0af10b9afcc48092 Mon Sep 17 00:00:00 2001 From: Andy Adamson Date: Mon, 27 Jul 2009 18:49:05 -0400 Subject: nfsd41: reclaim DRC memory on session free This fixes a leak which would eventually lock out new clients. Signed-off-by: Andy Adamson Signed-off-by: J. Bruce Fields --- fs/nfsd/nfs4state.c | 3 +++ 1 file changed, 3 insertions(+) (limited to 'fs') diff --git a/fs/nfsd/nfs4state.c b/fs/nfsd/nfs4state.c index 2e6a44e3d2f..69bd37e3fdc 100644 --- a/fs/nfsd/nfs4state.c +++ b/fs/nfsd/nfs4state.c @@ -585,6 +585,9 @@ free_session(struct kref *kref) struct nfsd4_cache_entry *e = &ses->se_slots[i].sl_cache_entry; nfsd4_release_respages(e->ce_respages, e->ce_resused); } + spin_lock(&nfsd_drc_lock); + nfsd_drc_pages_used -= ses->se_fchannel.maxreqs * NFSD_PAGES_PER_SLOT; + spin_unlock(&nfsd_drc_lock); kfree(ses); } -- cgit v1.2.3-18-g5258 From b101ebbc39f50f8af4657e517954ca874b13b364 Mon Sep 17 00:00:00 2001 From: Andy Adamson Date: Mon, 27 Jul 2009 18:40:09 -0400 Subject: nfsd41: minor set_forechannel_maxreqs cleanup Signed-off-by: Andy Adamson Signed-off-by: J. Bruce Fields --- fs/nfsd/nfs4state.c | 15 +++++++-------- 1 file changed, 7 insertions(+), 8 deletions(-) (limited to 'fs') diff --git a/fs/nfsd/nfs4state.c b/fs/nfsd/nfs4state.c index 69bd37e3fdc..70cba3fbfa6 100644 --- a/fs/nfsd/nfs4state.c +++ b/fs/nfsd/nfs4state.c @@ -423,26 +423,25 @@ gen_sessionid(struct nfsd4_session *ses) */ static int set_forechannel_maxreqs(struct nfsd4_channel_attrs *fchan) { - int status = 0, np = fchan->maxreqs * NFSD_PAGES_PER_SLOT; + int np; if (fchan->maxreqs < 1) return nfserr_inval; else if (fchan->maxreqs > NFSD_MAX_SLOTS_PER_SESSION) fchan->maxreqs = NFSD_MAX_SLOTS_PER_SESSION; + np = fchan->maxreqs * NFSD_PAGES_PER_SLOT; + spin_lock(&nfsd_drc_lock); if (np + nfsd_drc_pages_used > nfsd_drc_max_pages) np = nfsd_drc_max_pages - nfsd_drc_pages_used; nfsd_drc_pages_used += np; spin_unlock(&nfsd_drc_lock); - if (np <= 0) { - status = nfserr_resource; - fchan->maxreqs = 0; - } else - fchan->maxreqs = np / NFSD_PAGES_PER_SLOT; - - return status; + fchan->maxreqs = np / NFSD_PAGES_PER_SLOT; + if (fchan->maxreqs == 0) + return nfserr_resource; + return 0; } /* -- cgit v1.2.3-18-g5258 From 6a14dd1a4fe1bd00e02a96c97015cedfddda58ed Mon Sep 17 00:00:00 2001 From: Andy Adamson Date: Mon, 27 Jul 2009 19:06:45 -0400 Subject: nfsd41: reserve less memory for DRC Also remove a slightly misleading comment. Signed-off-by: Andy Adamson Signed-off-by: J. Bruce Fields --- fs/nfsd/nfssvc.c | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-) (limited to 'fs') diff --git a/fs/nfsd/nfssvc.c b/fs/nfsd/nfssvc.c index 78d8fcd883f..9be2a1932f8 100644 --- a/fs/nfsd/nfssvc.c +++ b/fs/nfsd/nfssvc.c @@ -246,8 +246,7 @@ void nfsd_reset_versions(void) */ static void set_max_drc(void) { - /* The percent of nr_free_buffer_pages used by the V4.1 server DRC */ - #define NFSD_DRC_SIZE_SHIFT 7 + #define NFSD_DRC_SIZE_SHIFT 10 nfsd_drc_max_pages = nr_free_buffer_pages() >> NFSD_DRC_SIZE_SHIFT; nfsd_drc_pages_used = 0; -- cgit v1.2.3-18-g5258 From 0c193054a4c1cf190d2f23e5e91bd14402e43912 Mon Sep 17 00:00:00 2001 From: Andy Adamson Date: Mon, 27 Jul 2009 19:09:19 -0400 Subject: nfsd41: hange from page to memory based drc limits NFSD_SLOT_CACHE_SIZE is the size of all encoded operation responses (excluding the sequence operation) that we want to cache. For now, keep NFSD_SLOT_CACHE_SIZE at PAGE_SIZE. It will be reduced when the DRC is changed from page based to memory based. Signed-off-by: Andy Adamson Signed-off-by: J. Bruce Fields --- fs/nfsd/nfs4state.c | 28 +++++++++++++--------------- fs/nfsd/nfssvc.c | 13 ++++++------- 2 files changed, 19 insertions(+), 22 deletions(-) (limited to 'fs') diff --git a/fs/nfsd/nfs4state.c b/fs/nfsd/nfs4state.c index 70cba3fbfa6..e2b11b1b515 100644 --- a/fs/nfsd/nfs4state.c +++ b/fs/nfsd/nfs4state.c @@ -414,31 +414,31 @@ gen_sessionid(struct nfsd4_session *ses) /* * Give the client the number of slots it requests bound by - * NFSD_MAX_SLOTS_PER_SESSION and by sv_drc_max_pages. + * NFSD_MAX_SLOTS_PER_SESSION and by nfsd_drc_max_mem. * - * If we run out of pages (sv_drc_pages_used == sv_drc_max_pages) we - * should (up to a point) re-negotiate active sessions and reduce their - * slot usage to make rooom for new connections. For now we just fail the - * create session. + * If we run out of reserved DRC memory we should (up to a point) re-negotiate + * active sessions and reduce their slot usage to make rooom for new + * connections. For now we just fail the create session. */ static int set_forechannel_maxreqs(struct nfsd4_channel_attrs *fchan) { - int np; + int mem; if (fchan->maxreqs < 1) return nfserr_inval; else if (fchan->maxreqs > NFSD_MAX_SLOTS_PER_SESSION) fchan->maxreqs = NFSD_MAX_SLOTS_PER_SESSION; - np = fchan->maxreqs * NFSD_PAGES_PER_SLOT; + mem = fchan->maxreqs * NFSD_SLOT_CACHE_SIZE; spin_lock(&nfsd_drc_lock); - if (np + nfsd_drc_pages_used > nfsd_drc_max_pages) - np = nfsd_drc_max_pages - nfsd_drc_pages_used; - nfsd_drc_pages_used += np; + if (mem + nfsd_drc_mem_used > nfsd_drc_max_mem) + mem = ((nfsd_drc_max_mem - nfsd_drc_mem_used) / + NFSD_SLOT_CACHE_SIZE) * NFSD_SLOT_CACHE_SIZE; + nfsd_drc_mem_used += mem; spin_unlock(&nfsd_drc_lock); - fchan->maxreqs = np / NFSD_PAGES_PER_SLOT; + fchan->maxreqs = mem / NFSD_SLOT_CACHE_SIZE; if (fchan->maxreqs == 0) return nfserr_resource; return 0; @@ -465,9 +465,7 @@ static int init_forechannel_attrs(struct svc_rqst *rqstp, fchan->maxresp_sz = maxcount; session_fchan->maxresp_sz = fchan->maxresp_sz; - /* Set the max response cached size our default which is - * a multiple of PAGE_SIZE and small */ - session_fchan->maxresp_cached = NFSD_PAGES_PER_SLOT * PAGE_SIZE; + session_fchan->maxresp_cached = NFSD_SLOT_CACHE_SIZE; fchan->maxresp_cached = session_fchan->maxresp_cached; /* Use the client's maxops if possible */ @@ -585,7 +583,7 @@ free_session(struct kref *kref) nfsd4_release_respages(e->ce_respages, e->ce_resused); } spin_lock(&nfsd_drc_lock); - nfsd_drc_pages_used -= ses->se_fchannel.maxreqs * NFSD_PAGES_PER_SLOT; + nfsd_drc_mem_used -= ses->se_fchannel.maxreqs * NFSD_SLOT_CACHE_SIZE; spin_unlock(&nfsd_drc_lock); kfree(ses); } diff --git a/fs/nfsd/nfssvc.c b/fs/nfsd/nfssvc.c index 9be2a1932f8..5a280a9cb54 100644 --- a/fs/nfsd/nfssvc.c +++ b/fs/nfsd/nfssvc.c @@ -74,8 +74,8 @@ struct svc_serv *nfsd_serv; * nfsd_drc_pages_used tracks the current version 4.1 DRC memory usage. */ spinlock_t nfsd_drc_lock; -unsigned int nfsd_drc_max_pages; -unsigned int nfsd_drc_pages_used; +unsigned int nfsd_drc_max_mem; +unsigned int nfsd_drc_mem_used; #if defined(CONFIG_NFSD_V2_ACL) || defined(CONFIG_NFSD_V3_ACL) static struct svc_stat nfsd_acl_svcstats; @@ -247,12 +247,11 @@ void nfsd_reset_versions(void) static void set_max_drc(void) { #define NFSD_DRC_SIZE_SHIFT 10 - nfsd_drc_max_pages = nr_free_buffer_pages() - >> NFSD_DRC_SIZE_SHIFT; - nfsd_drc_pages_used = 0; + nfsd_drc_max_mem = (nr_free_buffer_pages() + >> NFSD_DRC_SIZE_SHIFT) * PAGE_SIZE; + nfsd_drc_mem_used = 0; spin_lock_init(&nfsd_drc_lock); - dprintk("%s nfsd_drc_max_pages %u\n", __func__, - nfsd_drc_max_pages); + dprintk("%s nfsd_drc_max_mem %u \n", __func__, nfsd_drc_max_mem); } int nfsd_create_serv(void) -- cgit v1.2.3-18-g5258 From 5261dcf8eb3d098545a676030910cf2c05a00e6c Mon Sep 17 00:00:00 2001 From: Andy Adamson Date: Thu, 23 Jul 2009 19:02:14 -0400 Subject: nfsd41: remove redundant forechannel max requests check This check is done in set_forechannel_maxreqs. Signed-off-by: Andy Adamson Signed-off-by: J. Bruce Fields --- fs/nfsd/nfs4state.c | 4 ---- 1 file changed, 4 deletions(-) (limited to 'fs') diff --git a/fs/nfsd/nfs4state.c b/fs/nfsd/nfs4state.c index e2b11b1b515..0be417e3aab 100644 --- a/fs/nfsd/nfs4state.c +++ b/fs/nfsd/nfs4state.c @@ -473,10 +473,6 @@ static int init_forechannel_attrs(struct svc_rqst *rqstp, fchan->maxops = NFSD_MAX_OPS_PER_COMPOUND; session_fchan->maxops = fchan->maxops; - /* try to use the client requested number of slots */ - if (fchan->maxreqs > NFSD_MAX_SLOTS_PER_SESSION) - fchan->maxreqs = NFSD_MAX_SLOTS_PER_SESSION; - /* FIXME: Error means no more DRC pages so the server should * recover pages from existing sessions. For now fail session * creation. -- cgit v1.2.3-18-g5258 From 88e588d56a2f0226a34386b94a03fda97d2b8e67 Mon Sep 17 00:00:00 2001 From: Andy Adamson Date: Thu, 23 Jul 2009 19:02:15 -0400 Subject: nfsd41: change check_slot_seqid parameters For separation of session slot and clientid slot processing. Signed-off-by: Andy Adamson Signed-off-by: J. Bruce Fields --- fs/nfsd/nfs4state.c | 24 +++++++++++++----------- 1 file changed, 13 insertions(+), 11 deletions(-) (limited to 'fs') diff --git a/fs/nfsd/nfs4state.c b/fs/nfsd/nfs4state.c index 0be417e3aab..99df8e7a687 100644 --- a/fs/nfsd/nfs4state.c +++ b/fs/nfsd/nfs4state.c @@ -1309,26 +1309,26 @@ error: } static int -check_slot_seqid(u32 seqid, struct nfsd4_slot *slot) +check_slot_seqid(u32 seqid, u32 slot_seqid, int slot_inuse) { - dprintk("%s enter. seqid %d slot->sl_seqid %d\n", __func__, seqid, - slot->sl_seqid); + dprintk("%s enter. seqid %d slot_seqid %d\n", __func__, seqid, + slot_seqid); /* The slot is in use, and no response has been sent. */ - if (slot->sl_inuse) { - if (seqid == slot->sl_seqid) + if (slot_inuse) { + if (seqid == slot_seqid) return nfserr_jukebox; else return nfserr_seq_misordered; } /* Normal */ - if (likely(seqid == slot->sl_seqid + 1)) + if (likely(seqid == slot_seqid + 1)) return nfs_ok; /* Replay */ - if (seqid == slot->sl_seqid) + if (seqid == slot_seqid) return nfserr_replay_cache; /* Wraparound */ - if (seqid == 1 && (slot->sl_seqid + 1) == 0) + if (seqid == 1 && (slot_seqid + 1) == 0) return nfs_ok; /* Misordered replay or misordered new request */ return nfserr_seq_misordered; @@ -1351,7 +1351,8 @@ nfsd4_create_session(struct svc_rqst *rqstp, if (conf) { slot = &conf->cl_slot; - status = check_slot_seqid(cr_ses->seqid, slot); + status = check_slot_seqid(cr_ses->seqid, slot->sl_seqid, + slot->sl_inuse); if (status == nfserr_replay_cache) { dprintk("Got a create_session replay! seqid= %d\n", slot->sl_seqid); @@ -1376,7 +1377,8 @@ nfsd4_create_session(struct svc_rqst *rqstp, } slot = &unconf->cl_slot; - status = check_slot_seqid(cr_ses->seqid, slot); + status = check_slot_seqid(cr_ses->seqid, slot->sl_seqid, + slot->sl_inuse); if (status) { /* an unconfirmed replay returns misordered */ status = nfserr_seq_misordered; @@ -1477,7 +1479,7 @@ nfsd4_sequence(struct svc_rqst *rqstp, slot = &session->se_slots[seq->slotid]; dprintk("%s: slotid %d\n", __func__, seq->slotid); - status = check_slot_seqid(seq->seqid, slot); + status = check_slot_seqid(seq->seqid, slot->sl_seqid, slot->sl_inuse); if (status == nfserr_replay_cache) { cstate->slot = slot; cstate->session = session; -- cgit v1.2.3-18-g5258 From 49557cc74c7bdf6a984be227ead9a84b3a26f053 Mon Sep 17 00:00:00 2001 From: Andy Adamson Date: Thu, 23 Jul 2009 19:02:16 -0400 Subject: nfsd41: Use separate DRC for setclientid Instead of trying to share the generic 4.1 reply cache code for the CREATE_SESSION reply cache, it's simpler to handle CREATE_SESSION separately. The nfs41 single slot clientid DRC holds the results of create session processing. CREATE_SESSION can be preceeded by a SEQUENCE operation (an embedded CREATE_SESSION) and the create session single slot cache must be maintained. nfsd4_replay_cache_entry() and nfsd4_store_cache_entry() do not implement the replay of an embedded CREATE_SESSION. The clientid DRC slot does not need the inuse, cachethis or other fields that the multiple slot session cache uses. Replace the clientid DRC cache struct nfs4_slot cache with a new nfsd4_clid_slot cache. Save the xdr struct nfsd4_create_session into the cache at the end of processing, and on a replay, replace the struct for the replay request with the cached version all while under the state lock. nfsd4_proc_compound will handle both the solo and embedded CREATE_SESSION case via the normal use of encode_operation. Errors that do not change the create session cache: A create session NFS4ERR_STALE_CLIENTID error means that a client record (and associated create session slot) could not be found and therefore can't be changed. NFSERR_SEQ_MISORDERED errors do not change the slot cache. All other errors get cached. Remove the clientid DRC specific check in nfs4svc_encode_compoundres to put the session only if cstate.session is set which will now always be true. Signed-off-by: Andy Adamson Signed-off-by: J. Bruce Fields --- fs/nfsd/nfs4proc.c | 2 +- fs/nfsd/nfs4state.c | 64 +++++++++++++++++++++++++++++++---------------------- fs/nfsd/nfs4xdr.c | 3 +-- 3 files changed, 40 insertions(+), 29 deletions(-) (limited to 'fs') diff --git a/fs/nfsd/nfs4proc.c b/fs/nfsd/nfs4proc.c index d781658e808..d606c6a427d 100644 --- a/fs/nfsd/nfs4proc.c +++ b/fs/nfsd/nfs4proc.c @@ -1120,7 +1120,7 @@ nfsd4_proc_compound(struct svc_rqst *rqstp, BUG_ON(op->status == nfs_ok); encode_op: - /* Only from SEQUENCE or CREATE_SESSION */ + /* Only from SEQUENCE */ if (resp->cstate.status == nfserr_replay_cache) { dprintk("%s NFS4.1 replay from cache\n", __func__); if (nfsd4_not_cached(resp)) diff --git a/fs/nfsd/nfs4state.c b/fs/nfsd/nfs4state.c index 99df8e7a687..7729d092c8a 100644 --- a/fs/nfsd/nfs4state.c +++ b/fs/nfsd/nfs4state.c @@ -653,8 +653,6 @@ static inline void free_client(struct nfs4_client *clp) { shutdown_callback_client(clp); - nfsd4_release_respages(clp->cl_slot.sl_cache_entry.ce_respages, - clp->cl_slot.sl_cache_entry.ce_resused); if (clp->cl_cred.cr_group_info) put_group_info(clp->cl_cred.cr_group_info); kfree(clp->cl_principal); @@ -1293,12 +1291,11 @@ out_copy: exid->clientid.cl_boot = new->cl_clientid.cl_boot; exid->clientid.cl_id = new->cl_clientid.cl_id; - new->cl_slot.sl_seqid = 0; exid->seqid = 1; nfsd4_set_ex_flags(new, exid); dprintk("nfsd4_exchange_id seqid %d flags %x\n", - new->cl_slot.sl_seqid, new->cl_exchange_flags); + new->cl_cs_slot.sl_seqid, new->cl_exchange_flags); status = nfs_ok; out: @@ -1334,15 +1331,35 @@ check_slot_seqid(u32 seqid, u32 slot_seqid, int slot_inuse) return nfserr_seq_misordered; } +/* + * Cache the create session result into the create session single DRC + * slot cache by saving the xdr structure. sl_seqid has been set. + * Do this for solo or embedded create session operations. + */ +static void +nfsd4_cache_create_session(struct nfsd4_create_session *cr_ses, + struct nfsd4_clid_slot *slot, int nfserr) +{ + slot->sl_status = nfserr; + memcpy(&slot->sl_cr_ses, cr_ses, sizeof(*cr_ses)); +} + +static __be32 +nfsd4_replay_create_session(struct nfsd4_create_session *cr_ses, + struct nfsd4_clid_slot *slot) +{ + memcpy(cr_ses, &slot->sl_cr_ses, sizeof(*cr_ses)); + return slot->sl_status; +} + __be32 nfsd4_create_session(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate, struct nfsd4_create_session *cr_ses) { u32 ip_addr = svc_addr_in(rqstp)->sin_addr.s_addr; - struct nfsd4_compoundres *resp = rqstp->rq_resp; struct nfs4_client *conf, *unconf; - struct nfsd4_slot *slot = NULL; + struct nfsd4_clid_slot *cs_slot = NULL; int status = 0; nfs4_lock_state(); @@ -1350,25 +1367,22 @@ nfsd4_create_session(struct svc_rqst *rqstp, conf = find_confirmed_client(&cr_ses->clientid); if (conf) { - slot = &conf->cl_slot; - status = check_slot_seqid(cr_ses->seqid, slot->sl_seqid, - slot->sl_inuse); + cs_slot = &conf->cl_cs_slot; + status = check_slot_seqid(cr_ses->seqid, cs_slot->sl_seqid, 0); if (status == nfserr_replay_cache) { dprintk("Got a create_session replay! seqid= %d\n", - slot->sl_seqid); - cstate->slot = slot; - cstate->status = status; + cs_slot->sl_seqid); /* Return the cached reply status */ - status = nfsd4_replay_cache_entry(resp, NULL); + status = nfsd4_replay_create_session(cr_ses, cs_slot); goto out; - } else if (cr_ses->seqid != conf->cl_slot.sl_seqid + 1) { + } else if (cr_ses->seqid != cs_slot->sl_seqid + 1) { status = nfserr_seq_misordered; dprintk("Sequence misordered!\n"); dprintk("Expected seqid= %d but got seqid= %d\n", - slot->sl_seqid, cr_ses->seqid); + cs_slot->sl_seqid, cr_ses->seqid); goto out; } - conf->cl_slot.sl_seqid++; + cs_slot->sl_seqid++; } else if (unconf) { if (!same_creds(&unconf->cl_cred, &rqstp->rq_cred) || (ip_addr != unconf->cl_addr)) { @@ -1376,16 +1390,15 @@ nfsd4_create_session(struct svc_rqst *rqstp, goto out; } - slot = &unconf->cl_slot; - status = check_slot_seqid(cr_ses->seqid, slot->sl_seqid, - slot->sl_inuse); + cs_slot = &unconf->cl_cs_slot; + status = check_slot_seqid(cr_ses->seqid, cs_slot->sl_seqid, 0); if (status) { /* an unconfirmed replay returns misordered */ status = nfserr_seq_misordered; - goto out; + goto out_cache; } - slot->sl_seqid++; /* from 0 to 1 */ + cs_slot->sl_seqid++; /* from 0 to 1 */ move_to_confirmed(unconf); /* @@ -1406,12 +1419,11 @@ nfsd4_create_session(struct svc_rqst *rqstp, memcpy(cr_ses->sessionid.data, conf->cl_sessionid.data, NFS4_MAX_SESSIONID_LEN); - cr_ses->seqid = slot->sl_seqid; + cr_ses->seqid = cs_slot->sl_seqid; - slot->sl_inuse = true; - cstate->slot = slot; - /* Ensure a page is used for the cache */ - slot->sl_cache_entry.ce_cachethis = 1; +out_cache: + /* cache solo and embedded create sessions under the state lock */ + nfsd4_cache_create_session(cr_ses, cs_slot, status); out: nfs4_unlock_state(); dprintk("%s returns %d\n", __func__, ntohl(status)); diff --git a/fs/nfsd/nfs4xdr.c b/fs/nfsd/nfs4xdr.c index 2dcc7feaa6f..fdf632bf1cf 100644 --- a/fs/nfsd/nfs4xdr.c +++ b/fs/nfsd/nfs4xdr.c @@ -3313,8 +3313,7 @@ nfs4svc_encode_compoundres(struct svc_rqst *rqstp, __be32 *p, struct nfsd4_compo dprintk("%s: SET SLOT STATE TO AVAILABLE\n", __func__); resp->cstate.slot->sl_inuse = 0; } - if (resp->cstate.session) - nfsd4_put_session(resp->cstate.session); + nfsd4_put_session(resp->cstate.session); } return 1; } -- cgit v1.2.3-18-g5258 From c8647947f8c13ee2647505debae284ab1c859e65 Mon Sep 17 00:00:00 2001 From: Andy Adamson Date: Thu, 23 Jul 2009 19:02:17 -0400 Subject: nfsd41: rename nfsd4_enc_uncached_replay This function is only used for SEQUENCE replay. Signed-off-by: Andy Adamson Signed-off-by: J. Bruce Fields --- fs/nfsd/nfs4proc.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'fs') diff --git a/fs/nfsd/nfs4proc.c b/fs/nfsd/nfs4proc.c index d606c6a427d..23cd73869a2 100644 --- a/fs/nfsd/nfs4proc.c +++ b/fs/nfsd/nfs4proc.c @@ -989,7 +989,7 @@ static const char *nfsd4_op_name(unsigned opnum); * encode the uncache rep error on the next operation. */ static __be32 -nfsd4_enc_uncached_replay(struct nfsd4_compoundargs *args, +nfsd4_enc_sequence_replay(struct nfsd4_compoundargs *args, struct nfsd4_compoundres *resp) { struct nfsd4_op *op; @@ -1124,7 +1124,7 @@ encode_op: if (resp->cstate.status == nfserr_replay_cache) { dprintk("%s NFS4.1 replay from cache\n", __func__); if (nfsd4_not_cached(resp)) - status = nfsd4_enc_uncached_replay(args, resp); + status = nfsd4_enc_sequence_replay(args, resp); else status = op->status; goto out; -- cgit v1.2.3-18-g5258 From abfabf8cafa60e7876a7193fb344f739f690071d Mon Sep 17 00:00:00 2001 From: Andy Adamson Date: Thu, 23 Jul 2009 19:02:18 -0400 Subject: nfsd41: encode replay sequence from the slot values The sequence operation is not cached; always encode the sequence operation on a replay from the slot table and session values. This simplifies the sessions replay logic in nfsd4_proc_compound. If this is a replay of a compound that was specified not to be cached, return NFS4ERR_RETRY_UNCACHED_REP. Signed-off-by: Andy Adamson Signed-off-by: J. Bruce Fields --- fs/nfsd/nfs4proc.c | 33 +-------------------------------- fs/nfsd/nfs4state.c | 40 ++++++++++++++++++++++++++++++++++++---- 2 files changed, 37 insertions(+), 36 deletions(-) (limited to 'fs') diff --git a/fs/nfsd/nfs4proc.c b/fs/nfsd/nfs4proc.c index 23cd73869a2..6fde431df9e 100644 --- a/fs/nfsd/nfs4proc.c +++ b/fs/nfsd/nfs4proc.c @@ -983,34 +983,6 @@ static struct nfsd4_operation nfsd4_ops[]; static const char *nfsd4_op_name(unsigned opnum); -/* - * This is a replay of a compound for which no cache entry pages - * were used. Encode the sequence operation, and if cachethis is FALSE - * encode the uncache rep error on the next operation. - */ -static __be32 -nfsd4_enc_sequence_replay(struct nfsd4_compoundargs *args, - struct nfsd4_compoundres *resp) -{ - struct nfsd4_op *op; - - dprintk("--> %s resp->opcnt %d ce_cachethis %u \n", __func__, - resp->opcnt, resp->cstate.slot->sl_cache_entry.ce_cachethis); - - /* Encode the replayed sequence operation */ - BUG_ON(resp->opcnt != 1); - op = &args->ops[resp->opcnt - 1]; - nfsd4_encode_operation(resp, op); - - /*return nfserr_retry_uncached_rep in next operation. */ - if (resp->cstate.slot->sl_cache_entry.ce_cachethis == 0) { - op = &args->ops[resp->opcnt++]; - op->status = nfserr_retry_uncached_rep; - nfsd4_encode_operation(resp, op); - } - return op->status; -} - /* * Enforce NFSv4.1 COMPOUND ordering rules. * @@ -1123,10 +1095,7 @@ encode_op: /* Only from SEQUENCE */ if (resp->cstate.status == nfserr_replay_cache) { dprintk("%s NFS4.1 replay from cache\n", __func__); - if (nfsd4_not_cached(resp)) - status = nfsd4_enc_sequence_replay(args, resp); - else - status = op->status; + status = op->status; goto out; } if (op->status == nfserr_replay_me) { diff --git a/fs/nfsd/nfs4state.c b/fs/nfsd/nfs4state.c index 7729d092c8a..9295c4b56bc 100644 --- a/fs/nfsd/nfs4state.c +++ b/fs/nfsd/nfs4state.c @@ -1108,6 +1108,36 @@ nfsd41_copy_replay_data(struct nfsd4_compoundres *resp, return 1; } +/* + * Encode the replay sequence operation from the slot values. + * If cachethis is FALSE encode the uncached rep error on the next + * operation which sets resp->p and increments resp->opcnt for + * nfs4svc_encode_compoundres. + * + */ +static __be32 +nfsd4_enc_sequence_replay(struct nfsd4_compoundargs *args, + struct nfsd4_compoundres *resp) +{ + struct nfsd4_op *op; + struct nfsd4_slot *slot = resp->cstate.slot; + + dprintk("--> %s resp->opcnt %d cachethis %u \n", __func__, + resp->opcnt, resp->cstate.slot->sl_cache_entry.ce_cachethis); + + /* Encode the replayed sequence operation */ + op = &args->ops[resp->opcnt - 1]; + nfsd4_encode_operation(resp, op); + + /* Return nfserr_retry_uncached_rep in next operation. */ + if (args->opcnt > 1 && slot->sl_cache_entry.ce_cachethis == 0) { + op = &args->ops[resp->opcnt++]; + op->status = nfserr_retry_uncached_rep; + nfsd4_encode_operation(resp, op); + } + return op->status; +} + /* * Keep the first page of the replay. Copy the NFSv4.1 data from the first * cached page. Replace any futher replay pages from the cache. @@ -1131,10 +1161,12 @@ nfsd4_replay_cache_entry(struct nfsd4_compoundres *resp, * session inactivity timer fires and a solo sequence operation * is sent (lease renewal). */ - if (seq && nfsd4_not_cached(resp)) { - seq->maxslots = resp->cstate.session->se_fchannel.maxreqs; - return nfs_ok; - } + seq->maxslots = resp->cstate.session->se_fchannel.maxreqs; + + /* Either returns 0 or nfserr_retry_uncached */ + status = nfsd4_enc_sequence_replay(resp->rqstp->rq_argp, resp); + if (status == nfserr_retry_uncached_rep) + return status; if (!nfsd41_copy_replay_data(resp, entry)) { /* -- cgit v1.2.3-18-g5258 From 4516fc0454e7ffe2f369e80045b23c2b32155004 Mon Sep 17 00:00:00 2001 From: Jeff Layton Date: Fri, 14 Aug 2009 12:57:54 -0400 Subject: sunrpc: add routine for comparing addresses lockd needs these sort of routines, as does the NFSv4 callback code. Move lockd's routines into common code and rename them so that they can be used by others. Signed-off-by: Jeff Layton Acked-by: Chuck Lever Signed-off-by: J. Bruce Fields --- fs/lockd/clntlock.c | 2 +- fs/lockd/host.c | 4 ++-- fs/lockd/mon.c | 2 +- fs/lockd/svcsubs.c | 2 +- 4 files changed, 5 insertions(+), 5 deletions(-) (limited to 'fs') diff --git a/fs/lockd/clntlock.c b/fs/lockd/clntlock.c index 1f3b0fc0d35..fc9032dc886 100644 --- a/fs/lockd/clntlock.c +++ b/fs/lockd/clntlock.c @@ -166,7 +166,7 @@ __be32 nlmclnt_grant(const struct sockaddr *addr, const struct nlm_lock *lock) */ if (fl_blocked->fl_u.nfs_fl.owner->pid != lock->svid) continue; - if (!nlm_cmp_addr(nlm_addr(block->b_host), addr)) + if (!rpc_cmp_addr(nlm_addr(block->b_host), addr)) continue; if (nfs_compare_fh(NFS_FH(fl_blocked->fl_file->f_path.dentry->d_inode) ,fh) != 0) continue; diff --git a/fs/lockd/host.c b/fs/lockd/host.c index 7cb076ac6b4..4600c2037b8 100644 --- a/fs/lockd/host.c +++ b/fs/lockd/host.c @@ -111,7 +111,7 @@ static struct nlm_host *nlm_lookup_host(struct nlm_lookup_host_info *ni) */ chain = &nlm_hosts[nlm_hash_address(ni->sap)]; hlist_for_each_entry(host, pos, chain, h_hash) { - if (!nlm_cmp_addr(nlm_addr(host), ni->sap)) + if (!rpc_cmp_addr(nlm_addr(host), ni->sap)) continue; /* See if we have an NSM handle for this client */ @@ -125,7 +125,7 @@ static struct nlm_host *nlm_lookup_host(struct nlm_lookup_host_info *ni) if (host->h_server != ni->server) continue; if (ni->server && - !nlm_cmp_addr(nlm_srcaddr(host), ni->src_sap)) + !rpc_cmp_addr(nlm_srcaddr(host), ni->src_sap)) continue; /* Move to head of hash chain. */ diff --git a/fs/lockd/mon.c b/fs/lockd/mon.c index 30c933188dd..f956651d0f6 100644 --- a/fs/lockd/mon.c +++ b/fs/lockd/mon.c @@ -209,7 +209,7 @@ static struct nsm_handle *nsm_lookup_addr(const struct sockaddr *sap) struct nsm_handle *nsm; list_for_each_entry(nsm, &nsm_handles, sm_link) - if (nlm_cmp_addr(nsm_addr(nsm), sap)) + if (rpc_cmp_addr(nsm_addr(nsm), sap)) return nsm; return NULL; } diff --git a/fs/lockd/svcsubs.c b/fs/lockd/svcsubs.c index 9e4d6aab611..ad478da7ca6 100644 --- a/fs/lockd/svcsubs.c +++ b/fs/lockd/svcsubs.c @@ -417,7 +417,7 @@ EXPORT_SYMBOL_GPL(nlmsvc_unlock_all_by_sb); static int nlmsvc_match_ip(void *datap, struct nlm_host *host) { - return nlm_cmp_addr(nlm_srcaddr(host), datap); + return rpc_cmp_addr(nlm_srcaddr(host), datap); } /** -- cgit v1.2.3-18-g5258 From 363168b4ea8ec26aeb982ac6024a09f907ecd27e Mon Sep 17 00:00:00 2001 From: Jeff Layton Date: Fri, 14 Aug 2009 12:57:56 -0400 Subject: nfsd: make nfs4_client->cl_addr a struct sockaddr_storage It's currently a __be32, which isn't big enough to hold an IPv6 address. Signed-off-by: Jeff Layton Acked-by: Chuck Lever Signed-off-by: J. Bruce Fields --- fs/nfsd/nfs4state.c | 32 +++++++++++++++++++------------- 1 file changed, 19 insertions(+), 13 deletions(-) (limited to 'fs') diff --git a/fs/nfsd/nfs4state.c b/fs/nfsd/nfs4state.c index 9295c4b56bc..bfc14d879ea 100644 --- a/fs/nfsd/nfs4state.c +++ b/fs/nfsd/nfs4state.c @@ -55,6 +55,7 @@ #include #include #include +#include #define NFSDDBG_FACILITY NFSDDBG_PROC @@ -1220,13 +1221,15 @@ nfsd4_exchange_id(struct svc_rqst *rqstp, int status; unsigned int strhashval; char dname[HEXDIR_LEN]; + char addr_str[INET6_ADDRSTRLEN]; nfs4_verifier verf = exid->verifier; - u32 ip_addr = svc_addr_in(rqstp)->sin_addr.s_addr; + struct sockaddr *sa = svc_addr(rqstp); + rpc_ntop(sa, addr_str, sizeof(addr_str)); dprintk("%s rqstp=%p exid=%p clname.len=%u clname.data=%p " - " ip_addr=%u flags %x, spa_how %d\n", + "ip_addr=%s flags %x, spa_how %d\n", __func__, rqstp, exid, exid->clname.len, exid->clname.data, - ip_addr, exid->flags, exid->spa_how); + addr_str, exid->flags, exid->spa_how); if (!check_name(exid->clname) || (exid->flags & ~EXCHGID4_FLAG_MASK_A)) return nfserr_inval; @@ -1315,7 +1318,7 @@ out_new: copy_verf(new, &verf); copy_cred(&new->cl_cred, &rqstp->rq_cred); - new->cl_addr = ip_addr; + rpc_copy_addr((struct sockaddr *) &new->cl_addr, sa); gen_clid(new); gen_confirm(new); add_to_unconfirmed(new, strhashval); @@ -1389,7 +1392,7 @@ nfsd4_create_session(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate, struct nfsd4_create_session *cr_ses) { - u32 ip_addr = svc_addr_in(rqstp)->sin_addr.s_addr; + struct sockaddr *sa = svc_addr(rqstp); struct nfs4_client *conf, *unconf; struct nfsd4_clid_slot *cs_slot = NULL; int status = 0; @@ -1417,7 +1420,7 @@ nfsd4_create_session(struct svc_rqst *rqstp, cs_slot->sl_seqid++; } else if (unconf) { if (!same_creds(&unconf->cl_cred, &rqstp->rq_cred) || - (ip_addr != unconf->cl_addr)) { + !rpc_cmp_addr(sa, (struct sockaddr *) &unconf->cl_addr)) { status = nfserr_clid_inuse; goto out; } @@ -1564,7 +1567,7 @@ __be32 nfsd4_setclientid(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate, struct nfsd4_setclientid *setclid) { - struct sockaddr_in *sin = svc_addr_in(rqstp); + struct sockaddr *sa = svc_addr(rqstp); struct xdr_netobj clname = { .len = setclid->se_namelen, .data = setclid->se_name, @@ -1596,8 +1599,11 @@ nfsd4_setclientid(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate, /* RFC 3530 14.2.33 CASE 0: */ status = nfserr_clid_inuse; if (!same_creds(&conf->cl_cred, &rqstp->rq_cred)) { - dprintk("NFSD: setclientid: string in use by client" - " at %pI4\n", &conf->cl_addr); + char addr_str[INET6_ADDRSTRLEN]; + rpc_ntop((struct sockaddr *) &conf->cl_addr, addr_str, + sizeof(addr_str)); + dprintk("NFSD: setclientid: string in use by client " + "at %s\n", addr_str); goto out; } } @@ -1659,7 +1665,7 @@ nfsd4_setclientid(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate, gen_clid(new); } copy_verf(new, &clverifier); - new->cl_addr = sin->sin_addr.s_addr; + rpc_copy_addr((struct sockaddr *) &new->cl_addr, sa); new->cl_flavor = rqstp->rq_flavor; princ = svc_gss_principal(rqstp); if (princ) { @@ -1693,7 +1699,7 @@ nfsd4_setclientid_confirm(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate, struct nfsd4_setclientid_confirm *setclientid_confirm) { - struct sockaddr_in *sin = svc_addr_in(rqstp); + struct sockaddr *sa = svc_addr(rqstp); struct nfs4_client *conf, *unconf; nfs4_verifier confirm = setclientid_confirm->sc_confirm; clientid_t * clid = &setclientid_confirm->sc_clientid; @@ -1712,9 +1718,9 @@ nfsd4_setclientid_confirm(struct svc_rqst *rqstp, unconf = find_unconfirmed_client(clid); status = nfserr_clid_inuse; - if (conf && conf->cl_addr != sin->sin_addr.s_addr) + if (conf && !rpc_cmp_addr((struct sockaddr *) &conf->cl_addr, sa)) goto out; - if (unconf && unconf->cl_addr != sin->sin_addr.s_addr) + if (unconf && !rpc_cmp_addr((struct sockaddr *) &unconf->cl_addr, sa)) goto out; /* -- cgit v1.2.3-18-g5258 From aa9a4ec7707a5391cde556f3fa1b0eb4bca3bcf6 Mon Sep 17 00:00:00 2001 From: Jeff Layton Date: Fri, 14 Aug 2009 12:57:57 -0400 Subject: nfsd: convert nfs4_cb_conn struct to hold address in sockaddr_storage ...rather than as a separate address and port fields. This will be necessary for implementing callbacks over IPv6. Also, convert gen_callback to use the standard rpcuaddr2sockaddr routine rather than its own private one. Signed-off-by: Jeff Layton Acked-by: Chuck Lever Signed-off-by: J. Bruce Fields --- fs/nfsd/nfs4callback.c | 11 ++----- fs/nfsd/nfs4state.c | 81 ++++++-------------------------------------------- 2 files changed, 11 insertions(+), 81 deletions(-) (limited to 'fs') diff --git a/fs/nfsd/nfs4callback.c b/fs/nfsd/nfs4callback.c index 3fd23f7acec..81d1c5285dc 100644 --- a/fs/nfsd/nfs4callback.c +++ b/fs/nfsd/nfs4callback.c @@ -377,7 +377,6 @@ static int max_cb_time(void) int setup_callback_client(struct nfs4_client *clp) { - struct sockaddr_in addr; struct nfs4_cb_conn *cb = &clp->cl_cb_conn; struct rpc_timeout timeparms = { .to_initval = max_cb_time(), @@ -385,8 +384,8 @@ int setup_callback_client(struct nfs4_client *clp) }; struct rpc_create_args args = { .protocol = IPPROTO_TCP, - .address = (struct sockaddr *)&addr, - .addrsize = sizeof(addr), + .address = (struct sockaddr *) &cb->cb_addr, + .addrsize = cb->cb_addrlen, .timeout = &timeparms, .program = &cb_program, .prognumber = cb->cb_prog, @@ -400,12 +399,6 @@ int setup_callback_client(struct nfs4_client *clp) if (!clp->cl_principal && (clp->cl_flavor >= RPC_AUTH_GSS_KRB5)) return -EINVAL; - /* Initialize address */ - memset(&addr, 0, sizeof(addr)); - addr.sin_family = AF_INET; - addr.sin_port = htons(cb->cb_port); - addr.sin_addr.s_addr = htonl(cb->cb_addr); - /* Create RPC client */ client = rpc_create(&args); if (IS_ERR(client)) { diff --git a/fs/nfsd/nfs4state.c b/fs/nfsd/nfs4state.c index bfc14d879ea..96a742308ce 100644 --- a/fs/nfsd/nfs4state.c +++ b/fs/nfsd/nfs4state.c @@ -897,76 +897,6 @@ find_unconfirmed_client_by_str(const char *dname, unsigned int hashval, return NULL; } -/* a helper function for parse_callback */ -static int -parse_octet(unsigned int *lenp, char **addrp) -{ - unsigned int len = *lenp; - char *p = *addrp; - int n = -1; - char c; - - for (;;) { - if (!len) - break; - len--; - c = *p++; - if (c == '.') - break; - if ((c < '0') || (c > '9')) { - n = -1; - break; - } - if (n < 0) - n = 0; - n = (n * 10) + (c - '0'); - if (n > 255) { - n = -1; - break; - } - } - *lenp = len; - *addrp = p; - return n; -} - -/* parse and set the setclientid ipv4 callback address */ -static int -parse_ipv4(unsigned int addr_len, char *addr_val, unsigned int *cbaddrp, unsigned short *cbportp) -{ - int temp = 0; - u32 cbaddr = 0; - u16 cbport = 0; - u32 addrlen = addr_len; - char *addr = addr_val; - int i, shift; - - /* ipaddress */ - shift = 24; - for(i = 4; i > 0 ; i--) { - if ((temp = parse_octet(&addrlen, &addr)) < 0) { - return 0; - } - cbaddr |= (temp << shift); - if (shift > 0) - shift -= 8; - } - *cbaddrp = cbaddr; - - /* port */ - shift = 8; - for(i = 2; i > 0 ; i--) { - if ((temp = parse_octet(&addrlen, &addr)) < 0) { - return 0; - } - cbport |= (temp << shift); - if (shift > 0) - shift -= 8; - } - *cbportp = cbport; - return 1; -} - static void gen_callback(struct nfs4_client *clp, struct nfsd4_setclientid *se) { @@ -976,14 +906,21 @@ gen_callback(struct nfs4_client *clp, struct nfsd4_setclientid *se) if ((se->se_callback_netid_len != 3) || memcmp((char *)se->se_callback_netid_val, "tcp", 3)) goto out_err; - if ( !(parse_ipv4(se->se_callback_addr_len, se->se_callback_addr_val, - &cb->cb_addr, &cb->cb_port))) + cb->cb_addrlen = rpc_uaddr2sockaddr(se->se_callback_addr_val, + se->se_callback_addr_len, + (struct sockaddr *) &cb->cb_addr, + sizeof(cb->cb_addr)); + + if (!cb->cb_addrlen || cb->cb_addr.ss_family != AF_INET) goto out_err; + cb->cb_minorversion = 0; cb->cb_prog = se->se_callback_prog; cb->cb_ident = se->se_callback_ident; return; out_err: + cb->cb_addr.ss_family = AF_UNSPEC; + cb->cb_addrlen = 0; dprintk(KERN_INFO "NFSD: this client (clientid %08x/%08x) " "will not receive delegations\n", clp->cl_clientid.cl_boot, clp->cl_clientid.cl_id); -- cgit v1.2.3-18-g5258 From 7077ecbabd626cce1fcf5cc9766c83ec04d919f9 Mon Sep 17 00:00:00 2001 From: Jeff Layton Date: Fri, 14 Aug 2009 12:57:58 -0400 Subject: nfsd: add support for NFSv4 callbacks over IPv6 The framework to add this is all in place. Now, add the code to allow support for establishing a callback channel on an IPv6 socket. Signed-off-by: Jeff Layton Acked-by: Chuck Lever Signed-off-by: J. Bruce Fields --- fs/nfsd/nfs4state.c | 15 +++++++++++---- 1 file changed, 11 insertions(+), 4 deletions(-) (limited to 'fs') diff --git a/fs/nfsd/nfs4state.c b/fs/nfsd/nfs4state.c index 96a742308ce..9ec0ca1ef4e 100644 --- a/fs/nfsd/nfs4state.c +++ b/fs/nfsd/nfs4state.c @@ -901,9 +901,16 @@ static void gen_callback(struct nfs4_client *clp, struct nfsd4_setclientid *se) { struct nfs4_cb_conn *cb = &clp->cl_cb_conn; - - /* Currently, we only support tcp for the callback channel */ - if ((se->se_callback_netid_len != 3) || memcmp((char *)se->se_callback_netid_val, "tcp", 3)) + unsigned short expected_family; + + /* Currently, we only support tcp and tcp6 for the callback channel */ + if (se->se_callback_netid_len == 3 && + !memcmp(se->se_callback_netid_val, "tcp", 3)) + expected_family = AF_INET; + else if (se->se_callback_netid_len == 4 && + !memcmp(se->se_callback_netid_val, "tcp6", 4)) + expected_family = AF_INET6; + else goto out_err; cb->cb_addrlen = rpc_uaddr2sockaddr(se->se_callback_addr_val, @@ -911,7 +918,7 @@ gen_callback(struct nfs4_client *clp, struct nfsd4_setclientid *se) (struct sockaddr *) &cb->cb_addr, sizeof(cb->cb_addr)); - if (!cb->cb_addrlen || cb->cb_addr.ss_family != AF_INET) + if (!cb->cb_addrlen || cb->cb_addr.ss_family != expected_family) goto out_err; cb->cb_minorversion = 0; -- cgit v1.2.3-18-g5258 From fbf4665f41b02e757ab9d9198df65e319388e728 Mon Sep 17 00:00:00 2001 From: Jeff Layton Date: Fri, 14 Aug 2009 12:57:59 -0400 Subject: nfsd: populate sin6_scope_id on callback address with scopeid from rq_addr on SETCLIENTID call When a SETCLIENTID call comes in, one of the args given is the svc_rqst. This struct contains an rq_addr field which holds the address that sent the call. If this is an IPv6 address, then we can use the sin6_scope_id field in this address to populate the sin6_scope_id field in the callback address. AFAICT, the rq_addr.sin6_scope_id is non-zero if and only if the client mounted the server's link-local address. Signed-off-by: Jeff Layton Acked-by: Chuck Lever Signed-off-by: J. Bruce Fields --- fs/nfsd/nfs4state.c | 7 +++++-- 1 file changed, 5 insertions(+), 2 deletions(-) (limited to 'fs') diff --git a/fs/nfsd/nfs4state.c b/fs/nfsd/nfs4state.c index 9ec0ca1ef4e..d2a05248090 100644 --- a/fs/nfsd/nfs4state.c +++ b/fs/nfsd/nfs4state.c @@ -898,7 +898,7 @@ find_unconfirmed_client_by_str(const char *dname, unsigned int hashval, } static void -gen_callback(struct nfs4_client *clp, struct nfsd4_setclientid *se) +gen_callback(struct nfs4_client *clp, struct nfsd4_setclientid *se, u32 scopeid) { struct nfs4_cb_conn *cb = &clp->cl_cb_conn; unsigned short expected_family; @@ -921,6 +921,9 @@ gen_callback(struct nfs4_client *clp, struct nfsd4_setclientid *se) if (!cb->cb_addrlen || cb->cb_addr.ss_family != expected_family) goto out_err; + if (cb->cb_addr.ss_family == AF_INET6) + ((struct sockaddr_in6 *) &cb->cb_addr)->sin6_scope_id = scopeid; + cb->cb_minorversion = 0; cb->cb_prog = se->se_callback_prog; cb->cb_ident = se->se_callback_ident; @@ -1621,7 +1624,7 @@ nfsd4_setclientid(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate, } copy_cred(&new->cl_cred, &rqstp->rq_cred); gen_confirm(new); - gen_callback(new, setclid); + gen_callback(new, setclid, rpc_get_scope_id(sa)); add_to_unconfirmed(new, strhashval); setclid->se_clientid.cl_boot = new->cl_clientid.cl_boot; setclid->se_clientid.cl_id = new->cl_clientid.cl_id; -- cgit v1.2.3-18-g5258 From 55bb55dca0cecac2fb7b8c743db41361c011c8a8 Mon Sep 17 00:00:00 2001 From: Frank Filz Date: Fri, 14 Aug 2009 15:02:30 -0700 Subject: nfsd: Fix unnecessary deny bits in NFSv4 ACL The group deny entries end up denying tcy even though tcy was just allowed by the allow entry. This appears to be due to: ace->access_mask = mask_from_posix(deny, flags); instead of: ace->access_mask = deny_mask_from_posix(deny, flags); Denying a previously allowed bit has no effect, so this shouldn't affect behavior, but it's ugly. Signed-off-by: Frank Filz Signed-off-by: J. Bruce Fields --- fs/nfsd/nfs4acl.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/nfsd/nfs4acl.c b/fs/nfsd/nfs4acl.c index 54b8b4140c8..5320c2b6c7a 100644 --- a/fs/nfsd/nfs4acl.c +++ b/fs/nfsd/nfs4acl.c @@ -335,7 +335,7 @@ _posix_to_nfsv4_one(struct posix_acl *pacl, struct nfs4_acl *acl, if (deny) { ace->type = NFS4_ACE_ACCESS_DENIED_ACE_TYPE; ace->flag = eflag | NFS4_ACE_IDENTIFIER_GROUP; - ace->access_mask = mask_from_posix(deny, flags); + ace->access_mask = deny_mask_from_posix(deny, flags); ace->whotype = NFS4_ACL_WHO_NAMED; ace->who = pa->e_id; ace++; -- cgit v1.2.3-18-g5258 From ed2d8aed52212610d4cb79be3cbf535b04be38dc Mon Sep 17 00:00:00 2001 From: Ryusei Yamaguchi Date: Sun, 16 Aug 2009 00:54:41 +0900 Subject: knfsd: Replace lock_kernel with a mutex in nfsd pool stats. lock_kernel() in knfsd was replaced with a mutex. The later commit 03cf6c9f49a8fea953d38648d016e3f46e814991 ("knfsd: add file to export stats about nfsd pools") did not follow that change. This patch fixes the issue. Also move the get and put of nfsd_serv to the open and close methods (instead of start and stop methods) to allow atomic check and increment of reference count in the open method (where we can still return an error). Signed-off-by: Ryusei Yamaguchi Signed-off-by: Isaku Yamahata Signed-off-by: YOSHIFUJI Hideaki Cc: Greg Banks Signed-off-by: J. Bruce Fields --- fs/nfsd/nfsctl.c | 3 ++- fs/nfsd/nfssvc.c | 23 +++++++++++++++++++++-- 2 files changed, 23 insertions(+), 3 deletions(-) (limited to 'fs') diff --git a/fs/nfsd/nfsctl.c b/fs/nfsd/nfsctl.c index b764d7d898e..00388d2a3c9 100644 --- a/fs/nfsd/nfsctl.c +++ b/fs/nfsd/nfsctl.c @@ -174,12 +174,13 @@ static const struct file_operations exports_operations = { }; extern int nfsd_pool_stats_open(struct inode *inode, struct file *file); +extern int nfsd_pool_stats_release(struct inode *inode, struct file *file); static struct file_operations pool_stats_operations = { .open = nfsd_pool_stats_open, .read = seq_read, .llseek = seq_lseek, - .release = seq_release, + .release = nfsd_pool_stats_release, .owner = THIS_MODULE, }; diff --git a/fs/nfsd/nfssvc.c b/fs/nfsd/nfssvc.c index d68cd056b28..675d395c4ab 100644 --- a/fs/nfsd/nfssvc.c +++ b/fs/nfsd/nfssvc.c @@ -34,6 +34,7 @@ #include #include #include +#include #define NFSDDBG_FACILITY NFSDDBG_SVC @@ -614,7 +615,25 @@ nfsd_dispatch(struct svc_rqst *rqstp, __be32 *statp) int nfsd_pool_stats_open(struct inode *inode, struct file *file) { - if (nfsd_serv == NULL) + int ret; + mutex_lock(&nfsd_mutex); + if (nfsd_serv == NULL) { + mutex_unlock(&nfsd_mutex); return -ENODEV; - return svc_pool_stats_open(nfsd_serv, file); + } + /* bump up the psudo refcount while traversing */ + svc_get(nfsd_serv); + ret = svc_pool_stats_open(nfsd_serv, file); + mutex_unlock(&nfsd_mutex); + return ret; +} + +int nfsd_pool_stats_release(struct inode *inode, struct file *file) +{ + int ret = seq_release(inode, file); + mutex_lock(&nfsd_mutex); + /* this function really, really should have been called svc_put() */ + svc_destroy(nfsd_serv); + mutex_unlock(&nfsd_mutex); + return ret; } -- cgit v1.2.3-18-g5258 From aaf84eb95a9c610c6413cee4836764ea9194eed3 Mon Sep 17 00:00:00 2001 From: Benny Halevy Date: Thu, 20 Aug 2009 03:21:56 +0300 Subject: nfsd41: renew_client must be called under the state lock Until we work out the state locking so we can use a spin lock to protect the cl_lru, we need to take the state_lock to renew the client. Signed-off-by: Benny Halevy [nfsd41: Do not renew state on error] Signed-off-by: Ricardo Labiaga Signed-off-by: Benny Halevy [nfsd41: Simplify exit code] Signed-off-by: J. Bruce Fields --- fs/nfsd/nfs4state.c | 13 ++++++++----- 1 file changed, 8 insertions(+), 5 deletions(-) (limited to 'fs') diff --git a/fs/nfsd/nfs4state.c b/fs/nfsd/nfs4state.c index d2a05248090..5f634d24861 100644 --- a/fs/nfsd/nfs4state.c +++ b/fs/nfsd/nfs4state.c @@ -1481,7 +1481,7 @@ nfsd4_sequence(struct svc_rqst *rqstp, * for nfsd4_svc_encode_compoundres processing */ status = nfsd4_replay_cache_entry(resp, seq); cstate->status = nfserr_replay_cache; - goto replay_cache; + goto out; } if (status) goto out; @@ -1497,15 +1497,18 @@ nfsd4_sequence(struct svc_rqst *rqstp, cstate->slot = slot; cstate->session = session; -replay_cache: - /* Renew the clientid on success and on replay. - * Hold a session reference until done processing the compound: + /* Hold a session reference until done processing the compound: * nfsd4_put_session called only if the cstate slot is set. */ - renew_client(session->se_client); nfsd4_get_session(session); out: spin_unlock(&sessionid_lock); + /* Renew the clientid on success and on replay */ + if (cstate->session) { + nfs4_lock_state(); + renew_client(session->se_client); + nfs4_unlock_state(); + } dprintk("%s: return %d\n", __func__, ntohl(status)); return status; } -- cgit v1.2.3-18-g5258 From d8d0b85b11476ce59684ad2998e91a522df518a0 Mon Sep 17 00:00:00 2001 From: Frank Filz Date: Thu, 27 Aug 2009 17:35:41 -0400 Subject: nfsd4: remove ACE4_IDENTIFIER_GROUP flag from GROUP@ entry RFC 3530 says "ACE4_IDENTIFIER_GROUP flag MUST be ignored on entries with these special identifiers. When encoding entries with these special identifiers, the ACE4_IDENTIFIER_GROUP flag SHOULD be set to zero." It really shouldn't matter either way, but the point is that this flag is used to distinguish named users from named groups (since unix allows a group to have the same name as a user), so it doesn't really make sense to use it on a special identifier such as this.) Signed-off-by: J. Bruce Fields --- fs/nfsd/nfs4acl.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/nfsd/nfs4acl.c b/fs/nfsd/nfs4acl.c index 5320c2b6c7a..725d02f210e 100644 --- a/fs/nfsd/nfs4acl.c +++ b/fs/nfsd/nfs4acl.c @@ -321,7 +321,7 @@ _posix_to_nfsv4_one(struct posix_acl *pacl, struct nfs4_acl *acl, deny = ~pas.group & pas.other; if (deny) { ace->type = NFS4_ACE_ACCESS_DENIED_ACE_TYPE; - ace->flag = eflag | NFS4_ACE_IDENTIFIER_GROUP; + ace->flag = eflag; ace->access_mask = deny_mask_from_posix(deny, flags); ace->whotype = NFS4_ACL_WHO_GROUP; ace++; -- cgit v1.2.3-18-g5258 From 468de9e54a900559b55aa939a4daeaea1915e572 Mon Sep 17 00:00:00 2001 From: Andy Adamson Date: Thu, 27 Aug 2009 12:07:40 -0400 Subject: nfsd41: expand solo sequence check Compounds consisting of only a sequence operation don't need any additional caching beyond the sequence information we store in the slot entry. Fix nfsd4_is_solo_sequence to identify this case correctly. The additional check for a failed sequence in nfsd4_store_cache_entry() is redundant, since the nfsd4_is_solo_sequence call lower down catches this case. The final ce_cachethis set in nfsd4_sequence is also redundant. Signed-off-by: Andy Adamson Signed-off-by: J. Bruce Fields --- fs/nfsd/nfs4state.c | 9 --------- 1 file changed, 9 deletions(-) (limited to 'fs') diff --git a/fs/nfsd/nfs4state.c b/fs/nfsd/nfs4state.c index 5f634d24861..b44a2cfde6f 100644 --- a/fs/nfsd/nfs4state.c +++ b/fs/nfsd/nfs4state.c @@ -991,16 +991,10 @@ nfsd4_store_cache_entry(struct nfsd4_compoundres *resp) { struct nfsd4_cache_entry *entry = &resp->cstate.slot->sl_cache_entry; struct svc_rqst *rqstp = resp->rqstp; - struct nfsd4_compoundargs *args = rqstp->rq_argp; - struct nfsd4_op *op = &args->ops[resp->opcnt]; struct kvec *resv = &rqstp->rq_res.head[0]; dprintk("--> %s entry %p\n", __func__, entry); - /* Don't cache a failed OP_SEQUENCE. */ - if (resp->opcnt == 1 && op->opnum == OP_SEQUENCE && resp->cstate.status) - return; - nfsd4_release_respages(entry->ce_respages, entry->ce_resused); entry->ce_opcnt = resp->opcnt; entry->ce_status = resp->cstate.status; @@ -1490,9 +1484,6 @@ nfsd4_sequence(struct svc_rqst *rqstp, slot->sl_inuse = true; slot->sl_seqid = seq->seqid; slot->sl_cache_entry.ce_cachethis = seq->cachethis; - /* Always set the cache entry cachethis for solo sequence */ - if (nfsd4_is_solo_sequence(resp)) - slot->sl_cache_entry.ce_cachethis = 1; cstate->slot = slot; cstate->session = session; -- cgit v1.2.3-18-g5258 From a06b1261bdb580b35967d0e055d1ab131b332254 Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Mon, 31 Aug 2009 15:16:11 -0400 Subject: NFSD: Fix a bug in the NFSv4 'supported attrs' mandatory attribute The fact that the filesystem doesn't currently list any alternate locations does _not_ imply that the fs_locations attribute should be marked as "unsupported". Signed-off-by: Trond Myklebust Signed-off-by: J. Bruce Fields --- fs/nfsd/nfs4proc.c | 7 +------ fs/nfsd/nfs4xdr.c | 7 ------- 2 files changed, 1 insertion(+), 13 deletions(-) (limited to 'fs') diff --git a/fs/nfsd/nfs4proc.c b/fs/nfsd/nfs4proc.c index 6fde431df9e..bebc0c2e1b0 100644 --- a/fs/nfsd/nfs4proc.c +++ b/fs/nfsd/nfs4proc.c @@ -68,7 +68,6 @@ check_attr_support(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate, u32 *bmval, u32 *writable) { struct dentry *dentry = cstate->current_fh.fh_dentry; - struct svc_export *exp = cstate->current_fh.fh_export; /* * Check about attributes are supported by the NFSv4 server or not. @@ -80,17 +79,13 @@ check_attr_support(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate, return nfserr_attrnotsupp; /* - * Check FATTR4_WORD0_ACL & FATTR4_WORD0_FS_LOCATIONS can be supported + * Check FATTR4_WORD0_ACL can be supported * in current environment or not. */ if (bmval[0] & FATTR4_WORD0_ACL) { if (!IS_POSIXACL(dentry->d_inode)) return nfserr_attrnotsupp; } - if (bmval[0] & FATTR4_WORD0_FS_LOCATIONS) { - if (exp->ex_fslocs.locations == NULL) - return nfserr_attrnotsupp; - } /* * According to spec, read-only attributes return ERR_INVAL. diff --git a/fs/nfsd/nfs4xdr.c b/fs/nfsd/nfs4xdr.c index fdf632bf1cf..20c5e3db066 100644 --- a/fs/nfsd/nfs4xdr.c +++ b/fs/nfsd/nfs4xdr.c @@ -1793,11 +1793,6 @@ nfsd4_encode_fattr(struct svc_fh *fhp, struct svc_export *exp, goto out_nfserr; } } - if (bmval0 & FATTR4_WORD0_FS_LOCATIONS) { - if (exp->ex_fslocs.locations == NULL) { - bmval0 &= ~FATTR4_WORD0_FS_LOCATIONS; - } - } if ((buflen -= 16) < 0) goto out_resource; @@ -1825,8 +1820,6 @@ nfsd4_encode_fattr(struct svc_fh *fhp, struct svc_export *exp, goto out_resource; if (!aclsupport) word0 &= ~FATTR4_WORD0_ACL; - if (!exp->ex_fslocs.locations) - word0 &= ~FATTR4_WORD0_FS_LOCATIONS; if (!word2) { WRITE32(2); WRITE32(word0); -- cgit v1.2.3-18-g5258 From a649637c73a36174287a403cdda7607177d64523 Mon Sep 17 00:00:00 2001 From: Andy Adamson Date: Fri, 28 Aug 2009 08:45:01 -0400 Subject: nfsd41: bound forechannel drc size by memory usage By using the requested ca_maxresponsesize_cached * ca_maxresponses to bound a forechannel drc request size, clients can tailor a session to usage. For example, an I/O session (READ/WRITE only) can have a much smaller ca_maxresponsesize_cached (for only WRITE compound responses) and a lot larger ca_maxresponses to service a large in-flight data window. Signed-off-by: Andy Adamson Signed-off-by: J. Bruce Fields --- fs/nfsd/nfs4state.c | 66 ++++++++++++++++++++++++++++++++++++++--------------- 1 file changed, 48 insertions(+), 18 deletions(-) (limited to 'fs') diff --git a/fs/nfsd/nfs4state.c b/fs/nfsd/nfs4state.c index b44a2cfde6f..02b3ddd0bee 100644 --- a/fs/nfsd/nfs4state.c +++ b/fs/nfsd/nfs4state.c @@ -414,34 +414,64 @@ gen_sessionid(struct nfsd4_session *ses) } /* - * Give the client the number of slots it requests bound by - * NFSD_MAX_SLOTS_PER_SESSION and by nfsd_drc_max_mem. + * The protocol defines ca_maxresponssize_cached to include the size of + * the rpc header, but all we need to cache is the data starting after + * the end of the initial SEQUENCE operation--the rest we regenerate + * each time. Therefore we can advertise a ca_maxresponssize_cached + * value that is the number of bytes in our cache plus a few additional + * bytes. In order to stay on the safe side, and not promise more than + * we can cache, those additional bytes must be the minimum possible: 24 + * bytes of rpc header (xid through accept state, with AUTH_NULL + * verifier), 12 for the compound header (with zero-length tag), and 44 + * for the SEQUENCE op response: + */ +#define NFSD_MIN_HDR_SEQ_SZ (24 + 12 + 44) + +/* + * Give the client the number of ca_maxresponsesize_cached slots it + * requests, of size bounded by NFSD_SLOT_CACHE_SIZE, + * NFSD_MAX_MEM_PER_SESSION, and nfsd_drc_max_mem. Do not allow more + * than NFSD_MAX_SLOTS_PER_SESSION. * - * If we run out of reserved DRC memory we should (up to a point) re-negotiate - * active sessions and reduce their slot usage to make rooom for new - * connections. For now we just fail the create session. + * If we run out of reserved DRC memory we should (up to a point) + * re-negotiate active sessions and reduce their slot usage to make + * rooom for new connections. For now we just fail the create session. */ -static int set_forechannel_maxreqs(struct nfsd4_channel_attrs *fchan) +static int set_forechannel_drc_size(struct nfsd4_channel_attrs *fchan) { - int mem; + int mem, size = fchan->maxresp_cached; if (fchan->maxreqs < 1) return nfserr_inval; - else if (fchan->maxreqs > NFSD_MAX_SLOTS_PER_SESSION) - fchan->maxreqs = NFSD_MAX_SLOTS_PER_SESSION; - mem = fchan->maxreqs * NFSD_SLOT_CACHE_SIZE; + if (size < NFSD_MIN_HDR_SEQ_SZ) + size = NFSD_MIN_HDR_SEQ_SZ; + size -= NFSD_MIN_HDR_SEQ_SZ; + if (size > NFSD_SLOT_CACHE_SIZE) + size = NFSD_SLOT_CACHE_SIZE; + + /* bound the maxreqs by NFSD_MAX_MEM_PER_SESSION */ + mem = fchan->maxreqs * size; + if (mem > NFSD_MAX_MEM_PER_SESSION) { + fchan->maxreqs = NFSD_MAX_MEM_PER_SESSION / size; + if (fchan->maxreqs > NFSD_MAX_SLOTS_PER_SESSION) + fchan->maxreqs = NFSD_MAX_SLOTS_PER_SESSION; + mem = fchan->maxreqs * size; + } spin_lock(&nfsd_drc_lock); - if (mem + nfsd_drc_mem_used > nfsd_drc_max_mem) - mem = ((nfsd_drc_max_mem - nfsd_drc_mem_used) / - NFSD_SLOT_CACHE_SIZE) * NFSD_SLOT_CACHE_SIZE; + /* bound the total session drc memory ussage */ + if (mem + nfsd_drc_mem_used > nfsd_drc_max_mem) { + fchan->maxreqs = (nfsd_drc_max_mem - nfsd_drc_mem_used) / size; + mem = fchan->maxreqs * size; + } nfsd_drc_mem_used += mem; spin_unlock(&nfsd_drc_lock); - fchan->maxreqs = mem / NFSD_SLOT_CACHE_SIZE; if (fchan->maxreqs == 0) return nfserr_resource; + + fchan->maxresp_cached = size + NFSD_MIN_HDR_SEQ_SZ; return 0; } @@ -466,9 +496,6 @@ static int init_forechannel_attrs(struct svc_rqst *rqstp, fchan->maxresp_sz = maxcount; session_fchan->maxresp_sz = fchan->maxresp_sz; - session_fchan->maxresp_cached = NFSD_SLOT_CACHE_SIZE; - fchan->maxresp_cached = session_fchan->maxresp_cached; - /* Use the client's maxops if possible */ if (fchan->maxops > NFSD_MAX_OPS_PER_COMPOUND) fchan->maxops = NFSD_MAX_OPS_PER_COMPOUND; @@ -478,9 +505,12 @@ static int init_forech