From 082afec92d1052305af1195f591602f4d0f44277 Mon Sep 17 00:00:00 2001 From: Sage Weil Date: Sun, 22 Aug 2010 15:16:41 -0700 Subject: ceph: fix xattr cap writeback We should include the xattr metadata blob in the cap update message any time we are flushing dirty state, NOT just when we are also dropping the cap. This fixes async xattr writeback. Also, clean up the code slightly to avoid duplicating the bit test. Signed-off-by: Sage Weil --- fs/ceph/caps.c | 10 +++++----- 1 file changed, 5 insertions(+), 5 deletions(-) (limited to 'fs/ceph/caps.c') diff --git a/fs/ceph/caps.c b/fs/ceph/caps.c index 7bf182b0397..0ac2703f3bd 100644 --- a/fs/ceph/caps.c +++ b/fs/ceph/caps.c @@ -1082,6 +1082,7 @@ static int __send_cap(struct ceph_mds_client *mdsc, struct ceph_cap *cap, gid_t gid; struct ceph_mds_session *session; u64 xattr_version = 0; + struct ceph_buffer *xattr_blob = NULL; int delayed = 0; u64 flush_tid = 0; int i; @@ -1160,9 +1161,10 @@ static int __send_cap(struct ceph_mds_client *mdsc, struct ceph_cap *cap, gid = inode->i_gid; mode = inode->i_mode; - if (dropping & CEPH_CAP_XATTR_EXCL) { + if (flushing & CEPH_CAP_XATTR_EXCL) { __ceph_build_xattrs_blob(ci); - xattr_version = ci->i_xattrs.version + 1; + xattr_blob = ci->i_xattrs.blob; + xattr_version = ci->i_xattrs.version; } spin_unlock(&inode->i_lock); @@ -1170,9 +1172,7 @@ static int __send_cap(struct ceph_mds_client *mdsc, struct ceph_cap *cap, ret = send_cap_msg(session, ceph_vino(inode).ino, cap_id, op, keep, want, flushing, seq, flush_tid, issue_seq, mseq, size, max_size, &mtime, &atime, time_warp_seq, - uid, gid, mode, - xattr_version, - (flushing & CEPH_CAP_XATTR_EXCL) ? ci->i_xattrs.blob : NULL, + uid, gid, mode, xattr_version, xattr_blob, follows); if (ret < 0) { dout("error sending cap msg, must requeue %p\n", inode); -- cgit v1.2.3-70-g09d2 From 4a625be47243e0e07dedd0a1a6b94c66c2ab93ba Mon Sep 17 00:00:00 2001 From: Sage Weil Date: Sun, 22 Aug 2010 15:03:56 -0700 Subject: ceph: include dirty xattrs state in snapped caps When we snapshot dirty metadata that needs to be written back to the MDS, include dirty xattr metadata. Make the capsnap reference the encoded xattr blob so that it will be written back in the FLUSHSNAP op. Also fix the capsnap creation guard to include dirty auth or file bits, not just tests specific to dirty file data or file writes in progress (this fixes auth metadata writeback). Signed-off-by: Sage Weil --- fs/ceph/caps.c | 2 +- fs/ceph/snap.c | 23 ++++++++++++++++------- fs/ceph/super.h | 8 +++++--- fs/ceph/xattr.c | 1 + 4 files changed, 23 insertions(+), 11 deletions(-) (limited to 'fs/ceph/caps.c') diff --git a/fs/ceph/caps.c b/fs/ceph/caps.c index 0ac2703f3bd..ba5bbf318fe 100644 --- a/fs/ceph/caps.c +++ b/fs/ceph/caps.c @@ -1282,7 +1282,7 @@ retry: &capsnap->mtime, &capsnap->atime, capsnap->time_warp_seq, capsnap->uid, capsnap->gid, capsnap->mode, - 0, NULL, + capsnap->xattr_version, capsnap->xattr_blob, capsnap->follows); next_follows = capsnap->follows + 1; diff --git a/fs/ceph/snap.c b/fs/ceph/snap.c index c0b26b6badb..2cb190c2bd9 100644 --- a/fs/ceph/snap.c +++ b/fs/ceph/snap.c @@ -435,7 +435,7 @@ void ceph_queue_cap_snap(struct ceph_inode_info *ci) { struct inode *inode = &ci->vfs_inode; struct ceph_cap_snap *capsnap; - int used; + int used, dirty; capsnap = kzalloc(sizeof(*capsnap), GFP_NOFS); if (!capsnap) { @@ -445,6 +445,7 @@ void ceph_queue_cap_snap(struct ceph_inode_info *ci) spin_lock(&inode->i_lock); used = __ceph_caps_used(ci); + dirty = __ceph_caps_dirty(ci); if (__ceph_have_pending_cap_snap(ci)) { /* there is no point in queuing multiple "pending" cap_snaps, as no new writes are allowed to start when pending, so any @@ -452,11 +453,13 @@ void ceph_queue_cap_snap(struct ceph_inode_info *ci) cap_snap. lucky us. */ dout("queue_cap_snap %p already pending\n", inode); kfree(capsnap); - } else if (ci->i_wrbuffer_ref_head || (used & CEPH_CAP_FILE_WR)) { + } else if (ci->i_wrbuffer_ref_head || (used & CEPH_CAP_FILE_WR) || + (dirty & (CEPH_CAP_AUTH_EXCL|CEPH_CAP_XATTR_EXCL| + CEPH_CAP_FILE_EXCL|CEPH_CAP_FILE_WR))) { struct ceph_snap_context *snapc = ci->i_head_snapc; igrab(inode); - + atomic_set(&capsnap->nref, 1); capsnap->ci = ci; INIT_LIST_HEAD(&capsnap->ci_item); @@ -464,15 +467,21 @@ void ceph_queue_cap_snap(struct ceph_inode_info *ci) capsnap->follows = snapc->seq - 1; capsnap->issued = __ceph_caps_issued(ci, NULL); - capsnap->dirty = __ceph_caps_dirty(ci); + capsnap->dirty = dirty; capsnap->mode = inode->i_mode; capsnap->uid = inode->i_uid; capsnap->gid = inode->i_gid; - /* fixme? */ - capsnap->xattr_blob = NULL; - capsnap->xattr_len = 0; + if (dirty & CEPH_CAP_XATTR_EXCL) { + __ceph_build_xattrs_blob(ci); + capsnap->xattr_blob = + ceph_buffer_get(ci->i_xattrs.blob); + capsnap->xattr_version = ci->i_xattrs.version; + } else { + capsnap->xattr_blob = NULL; + capsnap->xattr_version = 0; + } /* dirty page count moved from _head to this cap_snap; all subsequent writes page dirties occur _after_ this diff --git a/fs/ceph/super.h b/fs/ceph/super.h index 2482d696f0d..b33929d8f28 100644 --- a/fs/ceph/super.h +++ b/fs/ceph/super.h @@ -216,8 +216,7 @@ struct ceph_cap_snap { uid_t uid; gid_t gid; - void *xattr_blob; - int xattr_len; + struct ceph_buffer *xattr_blob; u64 xattr_version; u64 size; @@ -229,8 +228,11 @@ struct ceph_cap_snap { static inline void ceph_put_cap_snap(struct ceph_cap_snap *capsnap) { - if (atomic_dec_and_test(&capsnap->nref)) + if (atomic_dec_and_test(&capsnap->nref)) { + if (capsnap->xattr_blob) + ceph_buffer_put(capsnap->xattr_blob); kfree(capsnap); + } } /* diff --git a/fs/ceph/xattr.c b/fs/ceph/xattr.c index 097a2654c00..9578af610b7 100644 --- a/fs/ceph/xattr.c +++ b/fs/ceph/xattr.c @@ -485,6 +485,7 @@ void __ceph_build_xattrs_blob(struct ceph_inode_info *ci) ci->i_xattrs.blob = ci->i_xattrs.prealloc_blob; ci->i_xattrs.prealloc_blob = NULL; ci->i_xattrs.dirty = false; + ci->i_xattrs.version++; } } -- cgit v1.2.3-70-g09d2 From 7d8cb26d7dcb911f110b7762bd5941e8f009d6c3 Mon Sep 17 00:00:00 2001 From: Sage Weil Date: Tue, 24 Aug 2010 08:44:16 -0700 Subject: ceph: maintain i_head_snapc when any caps are dirty, not just for data We used to use i_head_snapc to keep track of which snapc the current epoch of dirty data was dirtied under. It is used by queue_cap_snap to set up the cap_snap. However, since we queue cap snaps for any dirty caps, not just for dirty file data, we need to keep a valid i_head_snapc anytime we have dirty|flushing caps. This fixes a NULL pointer deref in queue_cap_snap when writing back dirty caps without data (e.g., snaptest-authwb.sh). Signed-off-by: Sage Weil --- fs/ceph/addr.c | 4 ++-- fs/ceph/caps.c | 20 +++++++++++++++++--- fs/ceph/snap.c | 6 +++++- fs/ceph/super.h | 3 ++- 4 files changed, 26 insertions(+), 7 deletions(-) (limited to 'fs/ceph/caps.c') diff --git a/fs/ceph/addr.c b/fs/ceph/addr.c index 420d46974ec..4cfce1ee31f 100644 --- a/fs/ceph/addr.c +++ b/fs/ceph/addr.c @@ -87,7 +87,7 @@ static int ceph_set_page_dirty(struct page *page) /* dirty the head */ spin_lock(&inode->i_lock); - if (ci->i_wrbuffer_ref_head == 0) + if (ci->i_head_snapc == NULL) ci->i_head_snapc = ceph_get_snap_context(snapc); ++ci->i_wrbuffer_ref_head; if (ci->i_wrbuffer_ref == 0) @@ -346,7 +346,7 @@ static struct ceph_snap_context *get_oldest_context(struct inode *inode, break; } } - if (!snapc && ci->i_head_snapc) { + if (!snapc && ci->i_wrbuffer_ref_head) { snapc = ceph_get_snap_context(ci->i_head_snapc); dout(" head snapc %p has %d dirty pages\n", snapc, ci->i_wrbuffer_ref_head); diff --git a/fs/ceph/caps.c b/fs/ceph/caps.c index ba5bbf318fe..a2069b6680a 100644 --- a/fs/ceph/caps.c +++ b/fs/ceph/caps.c @@ -1143,6 +1143,10 @@ static int __send_cap(struct ceph_mds_client *mdsc, struct ceph_cap *cap, for (i = 0; i < CEPH_CAP_BITS; i++) if (flushing & (1 << i)) ci->i_cap_flush_tid[i] = flush_tid; + + follows = ci->i_head_snapc->seq; + } else { + follows = 0; } keep = cap->implemented; @@ -1156,7 +1160,6 @@ static int __send_cap(struct ceph_mds_client *mdsc, struct ceph_cap *cap, mtime = inode->i_mtime; atime = inode->i_atime; time_warp_seq = ci->i_time_warp_seq; - follows = ci->i_snap_realm->cached_context->seq; uid = inode->i_uid; gid = inode->i_gid; mode = inode->i_mode; @@ -1332,7 +1335,11 @@ void __ceph_mark_dirty_caps(struct ceph_inode_info *ci, int mask) ceph_cap_string(was | mask)); ci->i_dirty_caps |= mask; if (was == 0) { - dout(" inode %p now dirty\n", &ci->vfs_inode); + if (!ci->i_head_snapc) + ci->i_head_snapc = ceph_get_snap_context( + ci->i_snap_realm->cached_context); + dout(" inode %p now dirty snapc %p\n", &ci->vfs_inode, + ci->i_head_snapc); BUG_ON(!list_empty(&ci->i_dirty_item)); spin_lock(&mdsc->cap_dirty_lock); list_add(&ci->i_dirty_item, &mdsc->cap_dirty); @@ -2190,7 +2197,9 @@ void ceph_put_wrbuffer_cap_refs(struct ceph_inode_info *ci, int nr, if (ci->i_head_snapc == snapc) { ci->i_wrbuffer_ref_head -= nr; - if (!ci->i_wrbuffer_ref_head) { + if (ci->i_wrbuffer_ref_head == 0 && + ci->i_dirty_caps == 0 && ci->i_flushing_caps == 0) { + BUG_ON(!ci->i_head_snapc); ceph_put_snap_context(ci->i_head_snapc); ci->i_head_snapc = NULL; } @@ -2483,6 +2492,11 @@ static void handle_cap_flush_ack(struct inode *inode, u64 flush_tid, dout(" inode %p now clean\n", inode); BUG_ON(!list_empty(&ci->i_dirty_item)); drop = 1; + if (ci->i_wrbuffer_ref_head == 0) { + BUG_ON(!ci->i_head_snapc); + ceph_put_snap_context(ci->i_head_snapc); + ci->i_head_snapc = NULL; + } } else { BUG_ON(list_empty(&ci->i_dirty_item)); } diff --git a/fs/ceph/snap.c b/fs/ceph/snap.c index 6bdbf3ae708..4868b9dcac5 100644 --- a/fs/ceph/snap.c +++ b/fs/ceph/snap.c @@ -458,6 +458,8 @@ void ceph_queue_cap_snap(struct ceph_inode_info *ci) CEPH_CAP_FILE_EXCL|CEPH_CAP_FILE_WR))) { struct ceph_snap_context *snapc = ci->i_head_snapc; + dout("queue_cap_snap %p cap_snap %p queuing under %p\n", inode, + capsnap, snapc); igrab(inode); atomic_set(&capsnap->nref, 1); @@ -489,7 +491,9 @@ void ceph_queue_cap_snap(struct ceph_inode_info *ci) capsnap->dirty_pages = ci->i_wrbuffer_ref_head; ci->i_wrbuffer_ref_head = 0; capsnap->context = snapc; - ci->i_head_snapc = NULL; + ci->i_head_snapc = + ceph_get_snap_context(ci->i_snap_realm->cached_context); + dout(" new snapc is %p\n", ci->i_head_snapc); list_add_tail(&capsnap->ci_item, &ci->i_cap_snaps); if (used & CEPH_CAP_FILE_WR) { diff --git a/fs/ceph/super.h b/fs/ceph/super.h index b33929d8f28..c33897ae572 100644 --- a/fs/ceph/super.h +++ b/fs/ceph/super.h @@ -344,7 +344,8 @@ struct ceph_inode_info { unsigned i_cap_exporting_issued; struct ceph_cap_reservation i_cap_migration_resv; struct list_head i_cap_snaps; /* snapped state pending flush to mds */ - struct ceph_snap_context *i_head_snapc; /* set if wr_buffer_head > 0 */ + struct ceph_snap_context *i_head_snapc; /* set if wr_buffer_head > 0 or + dirty|flushing caps */ unsigned i_snap_caps; /* cap bits for snapped files */ int i_nr_by_mode[CEPH_FILE_MODE_NUM]; /* open file counts */ -- cgit v1.2.3-70-g09d2 From cfc0bf6640dfd0f43bf8bfec5a475284809baa4d Mon Sep 17 00:00:00 2001 From: Sage Weil Date: Tue, 14 Sep 2010 15:50:59 -0700 Subject: ceph: stop sending FLUSHSNAPs when we hit a dirty capsnap Stop sending FLUSHSNAP messages when we hit a capsnap that has dirty_pages or is still writing. We'll send the newer capsnaps only after the older ones complete. Signed-off-by: Sage Weil --- fs/ceph/caps.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) (limited to 'fs/ceph/caps.c') diff --git a/fs/ceph/caps.c b/fs/ceph/caps.c index a2069b6680a..9fbe9019155 100644 --- a/fs/ceph/caps.c +++ b/fs/ceph/caps.c @@ -1227,7 +1227,7 @@ retry: * pages to be written out. */ if (capsnap->dirty_pages || capsnap->writing) - continue; + break; /* * if cap writeback already occurred, we should have dropped @@ -1276,8 +1276,8 @@ retry: &session->s_cap_snaps_flushing); spin_unlock(&inode->i_lock); - dout("flush_snaps %p cap_snap %p follows %lld size %llu\n", - inode, capsnap, next_follows, capsnap->size); + dout("flush_snaps %p cap_snap %p follows %lld tid %llu\n", + inode, capsnap, capsnap->follows, capsnap->flush_tid); send_cap_msg(session, ceph_vino(inode).ino, 0, CEPH_CAP_OP_FLUSHSNAP, capsnap->issued, 0, capsnap->dirty, 0, capsnap->flush_tid, 0, mseq, -- cgit v1.2.3-70-g09d2 From e835124c2be289515b918f2688ced4249e2de566 Mon Sep 17 00:00:00 2001 From: Sage Weil Date: Fri, 17 Sep 2010 08:03:08 -0700 Subject: ceph: only send one flushsnap per cap_snap per mds session Sending multiple flushsnap messages is problematic because we ignore the response if the tid doesn't match, and the server may only respond to each one once. It's also a waste. So, skip cap_snaps that are already on the flushing list, unless the caller tells us to resend (because we are reconnecting). Signed-off-by: Sage Weil --- fs/ceph/caps.c | 19 +++++++++++++++---- fs/ceph/snap.c | 2 +- fs/ceph/super.h | 3 ++- 3 files changed, 18 insertions(+), 6 deletions(-) (limited to 'fs/ceph/caps.c') diff --git a/fs/ceph/caps.c b/fs/ceph/caps.c index 9fbe9019155..b01c316a814 100644 --- a/fs/ceph/caps.c +++ b/fs/ceph/caps.c @@ -1195,10 +1195,14 @@ static int __send_cap(struct ceph_mds_client *mdsc, struct ceph_cap *cap, * asynchronously back to the MDS once sync writes complete and dirty * data is written out. * + * Unless @again is true, skip cap_snaps that were already sent to + * the MDS (i.e., during this session). + * * Called under i_lock. Takes s_mutex as needed. */ void __ceph_flush_snaps(struct ceph_inode_info *ci, - struct ceph_mds_session **psession) + struct ceph_mds_session **psession, + int again) __releases(ci->vfs_inode->i_lock) __acquires(ci->vfs_inode->i_lock) { @@ -1240,6 +1244,13 @@ retry: dout("no auth cap (migrating?), doing nothing\n"); goto out; } + + /* only flush each capsnap once */ + if (!again && !list_empty(&capsnap->flushing_item)) { + dout("already flushed %p, skipping\n", capsnap); + continue; + } + mds = ci->i_auth_cap->session->s_mds; mseq = ci->i_auth_cap->mseq; @@ -1314,7 +1325,7 @@ static void ceph_flush_snaps(struct ceph_inode_info *ci) struct inode *inode = &ci->vfs_inode; spin_lock(&inode->i_lock); - __ceph_flush_snaps(ci, NULL); + __ceph_flush_snaps(ci, NULL, 0); spin_unlock(&inode->i_lock); } @@ -1477,7 +1488,7 @@ void ceph_check_caps(struct ceph_inode_info *ci, int flags, /* flush snaps first time around only */ if (!list_empty(&ci->i_cap_snaps)) - __ceph_flush_snaps(ci, &session); + __ceph_flush_snaps(ci, &session, 0); goto retry_locked; retry: spin_lock(&inode->i_lock); @@ -1894,7 +1905,7 @@ static void kick_flushing_capsnaps(struct ceph_mds_client *mdsc, if (cap && cap->session == session) { dout("kick_flushing_caps %p cap %p capsnap %p\n", inode, cap, capsnap); - __ceph_flush_snaps(ci, &session); + __ceph_flush_snaps(ci, &session, 1); } else { pr_err("%p auth cap %p not mds%d ???\n", inode, cap, session->s_mds); diff --git a/fs/ceph/snap.c b/fs/ceph/snap.c index 9e6eef14b7d..190b6c4a6f2 100644 --- a/fs/ceph/snap.c +++ b/fs/ceph/snap.c @@ -717,7 +717,7 @@ static void flush_snaps(struct ceph_mds_client *mdsc) igrab(inode); spin_unlock(&mdsc->snap_flush_lock); spin_lock(&inode->i_lock); - __ceph_flush_snaps(ci, &session); + __ceph_flush_snaps(ci, &session, 0); spin_unlock(&inode->i_lock); iput(inode); spin_lock(&mdsc->snap_flush_lock); diff --git a/fs/ceph/super.h b/fs/ceph/super.h index c80bfbe27b0..b87638e84c4 100644 --- a/fs/ceph/super.h +++ b/fs/ceph/super.h @@ -828,7 +828,8 @@ extern void ceph_put_cap_refs(struct ceph_inode_info *ci, int had); extern void ceph_put_wrbuffer_cap_refs(struct ceph_inode_info *ci, int nr, struct ceph_snap_context *snapc); extern void __ceph_flush_snaps(struct ceph_inode_info *ci, - struct ceph_mds_session **psession); + struct ceph_mds_session **psession, + int again); extern void ceph_check_caps(struct ceph_inode_info *ci, int flags, struct ceph_mds_session *session); extern void ceph_check_delayed_caps(struct ceph_mds_client *mdsc); -- cgit v1.2.3-70-g09d2 From a43fb73101eaf6db0b33d22c152b338ab8b3edbb Mon Sep 17 00:00:00 2001 From: Sage Weil Date: Fri, 17 Sep 2010 09:54:08 -0700 Subject: ceph: check mapping to determine if FILE_CACHE cap is used See if the i_data mapping has any pages to determine if the FILE_CACHE capability is currently in use, instead of assuming it is any time the rdcache_gen value is set (i.e., issued -> used). This allows the MDS RECALL_STATE process work for inodes that have cached pages. Signed-off-by: Sage Weil --- fs/ceph/caps.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'fs/ceph/caps.c') diff --git a/fs/ceph/caps.c b/fs/ceph/caps.c index b01c316a814..73c153092f7 100644 --- a/fs/ceph/caps.c +++ b/fs/ceph/caps.c @@ -814,7 +814,7 @@ int __ceph_caps_used(struct ceph_inode_info *ci) used |= CEPH_CAP_PIN; if (ci->i_rd_ref) used |= CEPH_CAP_FILE_RD; - if (ci->i_rdcache_ref || ci->i_rdcache_gen) + if (ci->i_rdcache_ref || ci->vfs_inode.i_data.nrpages) used |= CEPH_CAP_FILE_CACHE; if (ci->i_wr_ref) used |= CEPH_CAP_FILE_WR; -- cgit v1.2.3-70-g09d2 From 21b559de56695d36b3f0819b7e2454737db254f8 Mon Sep 17 00:00:00 2001 From: Greg Farnum Date: Wed, 6 Oct 2010 15:46:30 -0700 Subject: ceph: send cap release message early on failed revoke. If an MDS tries to revoke caps that we don't have, we want to send releases early since they probably contain the caps message the MDS is looking for. Previously, we only sent the messages if we didn't have the inode either. But in a multi-mds system we can retain the inode after dropping all caps for a single MDS. Signed-off-by: Greg Farnum Signed-off-by: Sage Weil --- fs/ceph/caps.c | 23 +++++++++++++---------- 1 file changed, 13 insertions(+), 10 deletions(-) (limited to 'fs/ceph/caps.c') diff --git a/fs/ceph/caps.c b/fs/ceph/caps.c index 73c153092f7..97de325a49f 100644 --- a/fs/ceph/caps.c +++ b/fs/ceph/caps.c @@ -2774,15 +2774,7 @@ void ceph_handle_caps(struct ceph_mds_session *session, if (op == CEPH_CAP_OP_IMPORT) __queue_cap_release(session, vino.ino, cap_id, mseq, seq); - - /* - * send any full release message to try to move things - * along for the mds (who clearly thinks we still have this - * cap). - */ - ceph_add_cap_releases(mdsc, session); - ceph_send_cap_releases(mdsc, session); - goto done; + goto flush_cap_releases; } /* these will work even if we don't have a cap yet */ @@ -2810,7 +2802,7 @@ void ceph_handle_caps(struct ceph_mds_session *session, dout(" no cap on %p ino %llx.%llx from mds%d\n", inode, ceph_ino(inode), ceph_snap(inode), mds); spin_unlock(&inode->i_lock); - goto done; + goto flush_cap_releases; } /* note that each of these drops i_lock for us */ @@ -2834,6 +2826,17 @@ void ceph_handle_caps(struct ceph_mds_session *session, ceph_cap_op_name(op)); } + goto done; + +flush_cap_releases: + /* + * send any full release message to try to move things + * along for the mds (who clearly thinks we still have this + * cap). + */ + ceph_add_cap_releases(mdsc, session); + ceph_send_cap_releases(mdsc, session); + done: mutex_unlock(&session->s_mutex); done_unlocked: -- cgit v1.2.3-70-g09d2 From d91f2438d881514e4a923fd786dbd94b764a9440 Mon Sep 17 00:00:00 2001 From: Sage Weil Date: Wed, 22 Sep 2010 11:16:00 -0700 Subject: ceph: update issue_seq on cap grant We need to update the issue_seq on any grant operation, be it via an MDS reply or a separate grant message. The update in the grant path was missing. This broke cap release for inodes in which the MDS sent an explicit grant message that was not soon after followed by a successful MDS reply on the same inode. Also fix the signedness on seq locals. Signed-off-by: Sage Weil --- fs/ceph/caps.c | 8 +++++--- 1 file changed, 5 insertions(+), 3 deletions(-) (limited to 'fs/ceph/caps.c') diff --git a/fs/ceph/caps.c b/fs/ceph/caps.c index 97de325a49f..5e9da996a15 100644 --- a/fs/ceph/caps.c +++ b/fs/ceph/caps.c @@ -2283,7 +2283,8 @@ static void handle_cap_grant(struct inode *inode, struct ceph_mds_caps *grant, { struct ceph_inode_info *ci = ceph_inode(inode); int mds = session->s_mds; - int seq = le32_to_cpu(grant->seq); + unsigned seq = le32_to_cpu(grant->seq); + unsigned issue_seq = le32_to_cpu(grant->issue_seq); int newcaps = le32_to_cpu(grant->caps); int issued, implemented, used, wanted, dirty; u64 size = le64_to_cpu(grant->size); @@ -2295,8 +2296,8 @@ static void handle_cap_grant(struct inode *inode, struct ceph_mds_caps *grant, int revoked_rdcache = 0; int queue_invalidate = 0; - dout("handle_cap_grant inode %p cap %p mds%d seq %d %s\n", - inode, cap, mds, seq, ceph_cap_string(newcaps)); + dout("handle_cap_grant inode %p cap %p mds%d seq %u/%u %s\n", + inode, cap, mds, seq, issue_seq, ceph_cap_string(newcaps)); dout(" size %llu max_size %llu, i_size %llu\n", size, max_size, inode->i_size); @@ -2392,6 +2393,7 @@ static void handle_cap_grant(struct inode *inode, struct ceph_mds_caps *grant, } cap->seq = seq; + cap->issue_seq = issue_seq; /* file layout may have changed */ ci->i_layout = grant->layout; -- cgit v1.2.3-70-g09d2