From a0dff78dab0ff8d78bd5c9e33c105cf1292f2282 Mon Sep 17 00:00:00 2001 From: Sage Weil Date: Thu, 22 Jul 2010 13:47:21 -0700 Subject: ceph: avoid dcache readdir for snapdir We should always go to the MDS for readdir on the hidden snapdir. The set of snapshots can change at any time; the client can't trust its cache for that. Signed-off-by: Sage Weil --- fs/ceph/dir.c | 1 + 1 file changed, 1 insertion(+) diff --git a/fs/ceph/dir.c b/fs/ceph/dir.c index f85719310db..ea36ba9960d 100644 --- a/fs/ceph/dir.c +++ b/fs/ceph/dir.c @@ -266,6 +266,7 @@ static int ceph_readdir(struct file *filp, void *dirent, filldir_t filldir) spin_lock(&inode->i_lock); if ((filp->f_pos == 2 || fi->dentry) && !ceph_test_opt(client, NOASYNCREADDIR) && + ceph_snap(inode) != CEPH_SNAPDIR && (ci->i_ceph_flags & CEPH_I_COMPLETE) && __ceph_caps_issued_mask(ci, CEPH_CAP_FILE_SHARED, 1)) { err = __dcache_readdir(filp, dirent, filldir); -- cgit v1.2.3 From 252af5214682191e34e57204e1a31924fb82c207 Mon Sep 17 00:00:00 2001 From: Sage Weil Date: Thu, 22 Jul 2010 13:49:08 -0700 Subject: ceph: fix d_release dop for snapdir, snapped dentries We need to set the d_release dop for snapdir and snapped dentries so that the ceph_dentry_info struct gets released. We also use the dcache to cache readdir results when possible, which only works if we know when dentries are dropped from the cache. Since we don't use the dcache for readdir in the hidden snapdir, avoid that case in ceph_dentry_release. Signed-off-by: Sage Weil --- fs/ceph/dir.c | 12 +++++++++--- 1 file changed, 9 insertions(+), 3 deletions(-) diff --git a/fs/ceph/dir.c b/fs/ceph/dir.c index ea36ba9960d..f94ed3c7f6a 100644 --- a/fs/ceph/dir.c +++ b/fs/ceph/dir.c @@ -1014,18 +1014,22 @@ out_touch: /* * When a dentry is released, clear the dir I_COMPLETE if it was part - * of the current dir gen. + * of the current dir gen or if this is in the snapshot namespace. */ static void ceph_dentry_release(struct dentry *dentry) { struct ceph_dentry_info *di = ceph_dentry(dentry); struct inode *parent_inode = dentry->d_parent->d_inode; + u64 snapid = ceph_snap(parent_inode); - if (parent_inode) { + dout("dentry_release %p parent %p\n", dentry, parent_inode); + + if (parent_inode && snapid != CEPH_SNAPDIR) { struct ceph_inode_info *ci = ceph_inode(parent_inode); spin_lock(&parent_inode->i_lock); - if (ci->i_shared_gen == di->lease_shared_gen) { + if (ci->i_shared_gen == di->lease_shared_gen || + snapid <= CEPH_MAXSNAP) { dout(" clearing %p complete (d_release)\n", parent_inode); ci->i_ceph_flags &= ~CEPH_I_COMPLETE; @@ -1242,7 +1246,9 @@ struct dentry_operations ceph_dentry_ops = { struct dentry_operations ceph_snapdir_dentry_ops = { .d_revalidate = ceph_snapdir_d_revalidate, + .d_release = ceph_dentry_release, }; struct dentry_operations ceph_snap_dentry_ops = { + .d_release = ceph_dentry_release, }; -- cgit v1.2.3 From bc4fdca85734d12cd2c7a25c52323ef6e6e5adef Mon Sep 17 00:00:00 2001 From: Sage Weil Date: Tue, 20 Jul 2010 16:19:56 -0700 Subject: ceph: fix pg_mapping leak on pg_temp updates Free the ceph_pg_mapping structs when they are removed from the pg_temp rbtree. Also fix a leak in the __insert_pg_mapping() error path. Signed-off-by: Sage Weil --- fs/ceph/osdmap.c | 26 +++++++++++++++----------- 1 file changed, 15 insertions(+), 11 deletions(-) diff --git a/fs/ceph/osdmap.c b/fs/ceph/osdmap.c index 277f8b33957..416d46adbf8 100644 --- a/fs/ceph/osdmap.c +++ b/fs/ceph/osdmap.c @@ -831,12 +831,13 @@ struct ceph_osdmap *osdmap_apply_incremental(void **p, void *end, /* remove any? */ while (rbp && pgid_cmp(rb_entry(rbp, struct ceph_pg_mapping, node)->pgid, pgid) <= 0) { - struct rb_node *cur = rbp; + struct ceph_pg_mapping *cur = + rb_entry(rbp, struct ceph_pg_mapping, node); + rbp = rb_next(rbp); - dout(" removed pg_temp %llx\n", - *(u64 *)&rb_entry(cur, struct ceph_pg_mapping, - node)->pgid); - rb_erase(cur, &map->pg_temp); + dout(" removed pg_temp %llx\n", *(u64 *)&cur->pgid); + rb_erase(&cur->node, &map->pg_temp); + kfree(cur); } if (pglen) { @@ -852,19 +853,22 @@ struct ceph_osdmap *osdmap_apply_incremental(void **p, void *end, for (j = 0; j < pglen; j++) pg->osds[j] = ceph_decode_32(p); err = __insert_pg_mapping(pg, &map->pg_temp); - if (err) + if (err) { + kfree(pg); goto bad; + } dout(" added pg_temp %llx len %d\n", *(u64 *)&pgid, pglen); } } while (rbp) { - struct rb_node *cur = rbp; + struct ceph_pg_mapping *cur = + rb_entry(rbp, struct ceph_pg_mapping, node); + rbp = rb_next(rbp); - dout(" removed pg_temp %llx\n", - *(u64 *)&rb_entry(cur, struct ceph_pg_mapping, - node)->pgid); - rb_erase(cur, &map->pg_temp); + dout(" removed pg_temp %llx\n", *(u64 *)&cur->pgid); + rb_erase(&cur->node, &map->pg_temp); + kfree(cur); } /* ignore the rest */ -- cgit v1.2.3 From 8c696737aa61316a252c4514d09dd163f1464d33 Mon Sep 17 00:00:00 2001 From: Sage Weil Date: Thu, 22 Jul 2010 14:11:56 -0700 Subject: ceph: fix leak of dentry in ceph_init_dentry() error path If we fail to allocate a ceph_dentry_info, don't leak the dn reference. Signed-off-by: Sage Weil --- fs/ceph/inode.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) diff --git a/fs/ceph/inode.c b/fs/ceph/inode.c index 8f9b9fe8ef9..3582e79f46e 100644 --- a/fs/ceph/inode.c +++ b/fs/ceph/inode.c @@ -1199,8 +1199,10 @@ retry_lookup: goto out; } err = ceph_init_dentry(dn); - if (err < 0) + if (err < 0) { + dput(dn); goto out; + } } else if (dn->d_inode && (ceph_ino(dn->d_inode) != vino.ino || ceph_snap(dn->d_inode) != vino.snap)) { -- cgit v1.2.3 From 1dadcce358a4c4078e1ea0bc4365c3f67b8e373e Mon Sep 17 00:00:00 2001 From: Sage Weil Date: Fri, 23 Jul 2010 13:54:21 -0700 Subject: ceph: fix dentry lease release When we embed a dentry lease release notification in a request, invalidate our lease so we don't think we still have it. Otherwise we can get all sorts of incorrect client behavior when multiple clients are interacting with the same part of the namespace. Signed-off-by: Sage Weil --- fs/ceph/caps.c | 1 + 1 file changed, 1 insertion(+) diff --git a/fs/ceph/caps.c b/fs/ceph/caps.c index 74144d6389f..6afc1affb50 100644 --- a/fs/ceph/caps.c +++ b/fs/ceph/caps.c @@ -2984,6 +2984,7 @@ int ceph_encode_dentry_release(void **p, struct dentry *dentry, memcpy(*p, dentry->d_name.name, dentry->d_name.len); *p += dentry->d_name.len; rel->dname_seq = cpu_to_le32(di->lease_seq); + __ceph_mdsc_drop_dentry_lease(dentry); } spin_unlock(&dentry->d_lock); return ret; -- cgit v1.2.3 From 25848b3ec681c7018e3746dd850c1e8ed0a3dd6b Mon Sep 17 00:00:00 2001 From: "Robert P. J. Day" Date: Sat, 24 Jul 2010 06:41:18 -0400 Subject: ceph: Correct obvious typo of Kconfig variable "CRYPTO_AES" Signed-off-by: Robert P. J. Day Signed-off-by: Sage Weil --- fs/ceph/Kconfig | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/fs/ceph/Kconfig b/fs/ceph/Kconfig index 04b8280582a..bc87b9c1d27 100644 --- a/fs/ceph/Kconfig +++ b/fs/ceph/Kconfig @@ -2,7 +2,7 @@ config CEPH_FS tristate "Ceph distributed file system (EXPERIMENTAL)" depends on INET && EXPERIMENTAL select LIBCRC32C - select CONFIG_CRYPTO_AES + select CRYPTO_AES help Choose Y or M here to include support for mounting the experimental Ceph distributed file system. Ceph is an extremely -- cgit v1.2.3 From 03066f23452ff088ad8e2c8acdf4443043f35b51 Mon Sep 17 00:00:00 2001 From: Yehuda Sadeh Date: Tue, 27 Jul 2010 13:11:08 -0700 Subject: ceph: use complete_all and wake_up_all This fixes an issue triggered by running concurrent syncs. One of the syncs would go through while the other would just hang indefinitely. In any case, we never actually want to wake a single waiter, so the *_all functions should be used. Signed-off-by: Yehuda Sadeh Signed-off-by: Sage Weil --- fs/ceph/caps.c | 14 +++++++------- fs/ceph/file.c | 2 +- fs/ceph/inode.c | 2 +- fs/ceph/mds_client.c | 10 +++++----- fs/ceph/mon_client.c | 6 +++--- fs/ceph/osd_client.c | 6 +++--- 6 files changed, 20 insertions(+), 20 deletions(-) diff --git a/fs/ceph/caps.c b/fs/ceph/caps.c index 6afc1affb50..b81be9a5648 100644 --- a/fs/ceph/caps.c +++ b/fs/ceph/caps.c @@ -627,7 +627,7 @@ retry: if (fmode >= 0) __ceph_get_fmode(ci, fmode); spin_unlock(&inode->i_lock); - wake_up(&ci->i_cap_wq); + wake_up_all(&ci->i_cap_wq); return 0; } @@ -1181,7 +1181,7 @@ static int __send_cap(struct ceph_mds_client *mdsc, struct ceph_cap *cap, } if (wake) - wake_up(&ci->i_cap_wq); + wake_up_all(&ci->i_cap_wq); return delayed; } @@ -2153,7 +2153,7 @@ void ceph_put_cap_refs(struct ceph_inode_info *ci, int had) else if (flushsnaps) ceph_flush_snaps(ci); if (wake) - wake_up(&ci->i_cap_wq); + wake_up_all(&ci->i_cap_wq); if (put) iput(inode); } @@ -2229,7 +2229,7 @@ void ceph_put_wrbuffer_cap_refs(struct ceph_inode_info *ci, int nr, iput(inode); } else if (complete_capsnap) { ceph_flush_snaps(ci); - wake_up(&ci->i_cap_wq); + wake_up_all(&ci->i_cap_wq); } if (drop_capsnap) iput(inode); @@ -2405,7 +2405,7 @@ static void handle_cap_grant(struct inode *inode, struct ceph_mds_caps *grant, if (queue_invalidate) ceph_queue_invalidate(inode); if (wake) - wake_up(&ci->i_cap_wq); + wake_up_all(&ci->i_cap_wq); if (check_caps == 1) ceph_check_caps(ci, CHECK_CAPS_NODELAY|CHECK_CAPS_AUTHONLY, @@ -2460,7 +2460,7 @@ static void handle_cap_flush_ack(struct inode *inode, u64 flush_tid, struct ceph_inode_info, i_flushing_item)->vfs_inode); mdsc->num_cap_flushing--; - wake_up(&mdsc->cap_flushing_wq); + wake_up_all(&mdsc->cap_flushing_wq); dout(" inode %p now !flushing\n", inode); if (ci->i_dirty_caps == 0) { @@ -2472,7 +2472,7 @@ static void handle_cap_flush_ack(struct inode *inode, u64 flush_tid, } } spin_unlock(&mdsc->cap_dirty_lock); - wake_up(&ci->i_cap_wq); + wake_up_all(&ci->i_cap_wq); out: spin_unlock(&inode->i_lock); diff --git a/fs/ceph/file.c b/fs/ceph/file.c index 6251a1574b9..7c08698fad3 100644 --- a/fs/ceph/file.c +++ b/fs/ceph/file.c @@ -265,7 +265,7 @@ int ceph_release(struct inode *inode, struct file *file) kmem_cache_free(ceph_file_cachep, cf); /* wake up anyone waiting for caps on this inode */ - wake_up(&ci->i_cap_wq); + wake_up_all(&ci->i_cap_wq); return 0; } diff --git a/fs/ceph/inode.c b/fs/ceph/inode.c index 3582e79f46e..389f9dbd994 100644 --- a/fs/ceph/inode.c +++ b/fs/ceph/inode.c @@ -1501,7 +1501,7 @@ retry: if (wrbuffer_refs == 0) ceph_check_caps(ci, CHECK_CAPS_AUTHONLY, NULL); if (wake) - wake_up(&ci->i_cap_wq); + wake_up_all(&ci->i_cap_wq); } diff --git a/fs/ceph/mds_client.c b/fs/ceph/mds_client.c index 416c08d315d..dd440bd438a 100644 --- a/fs/ceph/mds_client.c +++ b/fs/ceph/mds_client.c @@ -868,7 +868,7 @@ static int wake_up_session_cb(struct inode *inode, struct ceph_cap *cap, { struct ceph_inode_info *ci = ceph_inode(inode); - wake_up(&ci->i_cap_wq); + wake_up_all(&ci->i_cap_wq); if (arg) { spin_lock(&inode->i_lock); ci->i_wanted_max_size = 0; @@ -1564,7 +1564,7 @@ static void complete_request(struct ceph_mds_client *mdsc, if (req->r_callback) req->r_callback(mdsc, req); else - complete(&req->r_completion); + complete_all(&req->r_completion); } /* @@ -1932,7 +1932,7 @@ static void handle_reply(struct ceph_mds_session *session, struct ceph_msg *msg) if (head->safe) { req->r_got_safe = true; __unregister_request(mdsc, req); - complete(&req->r_safe_completion); + complete_all(&req->r_safe_completion); if (req->r_got_unsafe) { /* @@ -1947,7 +1947,7 @@ static void handle_reply(struct ceph_mds_session *session, struct ceph_msg *msg) /* last unsafe request during umount? */ if (mdsc->stopping && !__get_oldest_req(mdsc)) - complete(&mdsc->safe_umount_waiters); + complete_all(&mdsc->safe_umount_waiters); mutex_unlock(&mdsc->mutex); goto out; } @@ -2126,7 +2126,7 @@ static void handle_session(struct ceph_mds_session *session, pr_info("mds%d reconnect denied\n", session->s_mds); remove_session_caps(session); wake = 1; /* for good measure */ - complete(&mdsc->session_close_waiters); + complete_all(&mdsc->session_close_waiters); kick_requests(mdsc, mds); break; diff --git a/fs/ceph/mon_client.c b/fs/ceph/mon_client.c index cc115eafae1..54fe01c5070 100644 --- a/fs/ceph/mon_client.c +++ b/fs/ceph/mon_client.c @@ -345,7 +345,7 @@ static void ceph_monc_handle_map(struct ceph_mon_client *monc, out: mutex_unlock(&monc->mutex); - wake_up(&client->auth_wq); + wake_up_all(&client->auth_wq); } /* @@ -462,7 +462,7 @@ static void handle_statfs_reply(struct ceph_mon_client *monc, } mutex_unlock(&monc->mutex); if (req) { - complete(&req->completion); + complete_all(&req->completion); put_generic_request(req); } return; @@ -718,7 +718,7 @@ static void handle_auth_reply(struct ceph_mon_client *monc, monc->m_auth->front_max); if (ret < 0) { monc->client->auth_err = ret; - wake_up(&monc->client->auth_wq); + wake_up_all(&monc->client->auth_wq); } else if (ret > 0) { __send_prepared_auth_request(monc, ret); } else if (!was_auth && monc->auth->ops->is_authenticated(monc->auth)) { diff --git a/fs/ceph/osd_client.c b/fs/ceph/osd_client.c index 92b7251a53f..e3852234789 100644 --- a/fs/ceph/osd_client.c +++ b/fs/ceph/osd_client.c @@ -862,12 +862,12 @@ static void handle_reply(struct ceph_osd_client *osdc, struct ceph_msg *msg, if (req->r_callback) req->r_callback(req, msg); else - complete(&req->r_completion); + complete_all(&req->r_completion); if (flags & CEPH_OSD_FLAG_ONDISK) { if (req->r_safe_callback) req->r_safe_callback(req, msg); - complete(&req->r_safe_completion); /* fsync waiter */ + complete_all(&req->r_safe_completion); /* fsync waiter */ } done: @@ -1083,7 +1083,7 @@ done: if (newmap) kick_requests(osdc, NULL); up_read(&osdc->map_sem); - wake_up(&osdc->client->auth_wq); + wake_up_all(&osdc->client->auth_wq); return; bad: -- cgit v1.2.3