From 30c71233a1d4b5c23ad6652847285bf6b57086e2 Mon Sep 17 00:00:00 2001 From: Jeff Layton Date: Thu, 15 Dec 2016 08:37:56 -0500 Subject: ceph: clean up unsafe d_parent access in __choose_mds __choose_mds exists to pick an MDS to use when issuing a call. Doing that typically involves picking an inode and using the authoritative MDS for it. In most cases, that's pretty straightforward, as we are using an inode to which we hold a reference (usually represented by r_dentry or r_inode in the request). In the case of a snapshotted directory however, we need to fetch the non-snapped parent, which involves walking back up the parents in the tree. The dentries in the snapshot dir are effectively frozen but the overall parent is _not_, and could vanish if a concurrent rename were to occur. Clean this code up and take special care to ensure the validity of the entries we're working with. First, try to use the inode in r_locked_dir if one exists. If not and all we have is r_dentry, then we have to walk back up the tree. Use the rcu_read_lock for this so we can ensure that any d_parent we find won't go away, and take extra care to deal with the possibility that the dentries could go negative. Change get_nonsnap_parent to return an inode, and take a reference to that inode before returning (if any). Change all of the other places where we set "inode" in __choose_mds to also take a reference, and then call iput on that inode before exiting the function. Link: http://tracker.ceph.com/issues/18148 Signed-off-by: Jeff Layton Reviewed-by: Yan, Zheng Signed-off-by: Ilya Dryomov --- fs/ceph/mds_client.c | 64 ++++++++++++++++++++++++++++++++++------------------ 1 file changed, 42 insertions(+), 22 deletions(-) (limited to 'fs/ceph') diff --git a/fs/ceph/mds_client.c b/fs/ceph/mds_client.c index c9d2e553a6c4..377ac34ddbb3 100644 --- a/fs/ceph/mds_client.c +++ b/fs/ceph/mds_client.c @@ -667,6 +667,27 @@ static void __unregister_request(struct ceph_mds_client *mdsc, ceph_mdsc_put_request(req); } +/* + * Walk back up the dentry tree until we hit a dentry representing a + * non-snapshot inode. We do this using the rcu_read_lock (which must be held + * when calling this) to ensure that the objects won't disappear while we're + * working with them. Once we hit a candidate dentry, we attempt to take a + * reference to it, and return that as the result. + */ +static struct inode *get_nonsnap_parent(struct dentry *dentry) { struct inode + *inode = NULL; + + while (dentry && !IS_ROOT(dentry)) { + inode = d_inode_rcu(dentry); + if (!inode || ceph_snap(inode) == CEPH_NOSNAP) + break; + dentry = dentry->d_parent; + } + if (inode) + inode = igrab(inode); + return inode; +} + /* * Choose mds to send request to next. If there is a hint set in the * request (e.g., due to a prior forward hint from the mds), use that. @@ -675,19 +696,6 @@ static void __unregister_request(struct ceph_mds_client *mdsc, * * Called under mdsc->mutex. */ -static struct dentry *get_nonsnap_parent(struct dentry *dentry) -{ - /* - * we don't need to worry about protecting the d_parent access - * here because we never renaming inside the snapped namespace - * except to resplice to another snapdir, and either the old or new - * result is a valid result. - */ - while (!IS_ROOT(dentry) && ceph_snap(d_inode(dentry)) != CEPH_NOSNAP) - dentry = dentry->d_parent; - return dentry; -} - static int __choose_mds(struct ceph_mds_client *mdsc, struct ceph_mds_request *req) { @@ -717,30 +725,39 @@ static int __choose_mds(struct ceph_mds_client *mdsc, inode = NULL; if (req->r_inode) { inode = req->r_inode; + ihold(inode); } else if (req->r_dentry) { /* ignore race with rename; old or new d_parent is okay */ - struct dentry *parent = req->r_dentry->d_parent; - struct inode *dir = d_inode(parent); + struct dentry *parent; + struct inode *dir; + + rcu_read_lock(); + parent = req->r_dentry->d_parent; + dir = req->r_locked_dir ? : d_inode_rcu(parent); - if (dir->i_sb != mdsc->fsc->sb) { - /* not this fs! */ + if (!dir || dir->i_sb != mdsc->fsc->sb) { + /* not this fs or parent went negative */ inode = d_inode(req->r_dentry); + if (inode) + ihold(inode); } else if (ceph_snap(dir) != CEPH_NOSNAP) { /* direct snapped/virtual snapdir requests * based on parent dir inode */ - struct dentry *dn = get_nonsnap_parent(parent); - inode = d_inode(dn); + inode = get_nonsnap_parent(parent); dout("__choose_mds using nonsnap parent %p\n", inode); } else { /* dentry target */ inode = d_inode(req->r_dentry); if (!inode || mode == USE_AUTH_MDS) { /* dir + name */ - inode = dir; + inode = igrab(dir); hash = ceph_dentry_hash(dir, req->r_dentry); is_hash = true; + } else { + ihold(inode); } } + rcu_read_unlock(); } dout("__choose_mds %p is_hash=%d (%d) mode %d\n", inode, (int)is_hash, @@ -769,7 +786,7 @@ static int __choose_mds(struct ceph_mds_client *mdsc, (int)r, frag.ndist); if (ceph_mdsmap_get_state(mdsc->mdsmap, mds) >= CEPH_MDS_STATE_ACTIVE) - return mds; + goto out; } /* since this file/dir wasn't known to be @@ -784,7 +801,7 @@ static int __choose_mds(struct ceph_mds_client *mdsc, inode, ceph_vinop(inode), frag.frag, mds); if (ceph_mdsmap_get_state(mdsc->mdsmap, mds) >= CEPH_MDS_STATE_ACTIVE) - return mds; + goto out; } } } @@ -797,6 +814,7 @@ static int __choose_mds(struct ceph_mds_client *mdsc, cap = rb_entry(rb_first(&ci->i_caps), struct ceph_cap, ci_node); if (!cap) { spin_unlock(&ci->i_ceph_lock); + iput(inode); goto random; } mds = cap->session->s_mds; @@ -804,6 +822,8 @@ static int __choose_mds(struct ceph_mds_client *mdsc, inode, ceph_vinop(inode), mds, cap == ci->i_auth_cap ? "auth " : "", cap); spin_unlock(&ci->i_ceph_lock); +out: + iput(inode); return mds; random: -- cgit v1.2.3-59-g8ed1b From c6b0b656ca24ede6657abb4a2cd910fa9c1879ba Mon Sep 17 00:00:00 2001 From: Jeff Layton Date: Thu, 15 Dec 2016 08:37:57 -0500 Subject: ceph: clean up unsafe d_parent accesses in build_dentry_path While we hold a reference to the dentry when build_dentry_path is called, we could end up racing with a rename that changes d_parent. Handle that situation correctly, by using the rcu_read_lock to ensure that the parent dentry and inode stick around long enough to safely check ceph_snap and ceph_ino. Link: http://tracker.ceph.com/issues/18148 Signed-off-by: Jeff Layton Reviewed-by: Yan, Zheng Signed-off-by: Ilya Dryomov --- fs/ceph/mds_client.c | 9 +++++++-- 1 file changed, 7 insertions(+), 2 deletions(-) (limited to 'fs/ceph') diff --git a/fs/ceph/mds_client.c b/fs/ceph/mds_client.c index 377ac34ddbb3..7c2eb28c24f6 100644 --- a/fs/ceph/mds_client.c +++ b/fs/ceph/mds_client.c @@ -1800,13 +1800,18 @@ static int build_dentry_path(struct dentry *dentry, int *pfreepath) { char *path; + struct inode *dir; - if (ceph_snap(d_inode(dentry->d_parent)) == CEPH_NOSNAP) { - *pino = ceph_ino(d_inode(dentry->d_parent)); + rcu_read_lock(); + dir = d_inode_rcu(dentry->d_parent); + if (dir && ceph_snap(dir) == CEPH_NOSNAP) { + *pino = ceph_ino(dir); + rcu_read_unlock(); *ppath = dentry->d_name.name; *ppathlen = dentry->d_name.len; return 0; } + rcu_read_unlock(); path = ceph_mdsc_build_path(dentry, ppathlen, pino, 1); if (IS_ERR(path)) return PTR_ERR(path); -- cgit v1.2.3-59-g8ed1b From fd36a71762f3b0fcb9741ed24021afabec7e0c45 Mon Sep 17 00:00:00 2001 From: Jeff Layton Date: Thu, 15 Dec 2016 08:37:58 -0500 Subject: ceph: pass parent dir ino info to build_dentry_path In the event that we have a parent inode reference in the request, we can use that instead of mucking about in the dcache. Pass any parent inode info we have down to build_dentry_path so it can make use of it. Link: http://tracker.ceph.com/issues/18148 Signed-off-by: Jeff Layton Reviewed-by: Yan, Zheng Signed-off-by: Ilya Dryomov --- fs/ceph/mds_client.c | 16 +++++++++------- 1 file changed, 9 insertions(+), 7 deletions(-) (limited to 'fs/ceph') diff --git a/fs/ceph/mds_client.c b/fs/ceph/mds_client.c index 7c2eb28c24f6..c41c6aab3232 100644 --- a/fs/ceph/mds_client.c +++ b/fs/ceph/mds_client.c @@ -1795,15 +1795,15 @@ retry: return path; } -static int build_dentry_path(struct dentry *dentry, +static int build_dentry_path(struct dentry *dentry, struct inode *dir, const char **ppath, int *ppathlen, u64 *pino, int *pfreepath) { char *path; - struct inode *dir; rcu_read_lock(); - dir = d_inode_rcu(dentry->d_parent); + if (!dir) + dir = d_inode_rcu(dentry->d_parent); if (dir && ceph_snap(dir) == CEPH_NOSNAP) { *pino = ceph_ino(dir); rcu_read_unlock(); @@ -1847,8 +1847,8 @@ static int build_inode_path(struct inode *inode, * an explicit ino+path. */ static int set_request_path_attr(struct inode *rinode, struct dentry *rdentry, - const char *rpath, u64 rino, - const char **ppath, int *pathlen, + struct inode *rdiri, const char *rpath, + u64 rino, const char **ppath, int *pathlen, u64 *ino, int *freepath) { int r = 0; @@ -1858,7 +1858,8 @@ static int set_request_path_attr(struct inode *rinode, struct dentry *rdentry, dout(" inode %p %llx.%llx\n", rinode, ceph_ino(rinode), ceph_snap(rinode)); } else if (rdentry) { - r = build_dentry_path(rdentry, ppath, pathlen, ino, freepath); + r = build_dentry_path(rdentry, rdiri, ppath, pathlen, ino, + freepath); dout(" dentry %p %llx/%.*s\n", rdentry, *ino, *pathlen, *ppath); } else if (rpath || rino) { @@ -1891,7 +1892,7 @@ static struct ceph_msg *create_request_message(struct ceph_mds_client *mdsc, int ret; ret = set_request_path_attr(req->r_inode, req->r_dentry, - req->r_path1, req->r_ino1.ino, + req->r_locked_dir, req->r_path1, req->r_ino1.ino, &path1, &pathlen1, &ino1, &freepath1); if (ret < 0) { msg = ERR_PTR(ret); @@ -1899,6 +1900,7 @@ static struct ceph_msg *create_request_message(struct ceph_mds_client *mdsc, } ret = set_request_path_attr(NULL, req->r_old_dentry, + req->r_old_dentry_dir, req->r_path2, req->r_ino2.ino, &path2, &pathlen2, &ino2, &freepath2); if (ret < 0) { -- cgit v1.2.3-59-g8ed1b From adf0d68701c7f3e50f21308c76f41e60956a6832 Mon Sep 17 00:00:00 2001 From: Jeff Layton Date: Thu, 15 Dec 2016 08:37:58 -0500 Subject: ceph: fix unsafe dcache access in ceph_encode_dentry_release Accessing d_parent requires some sort of locking or it could vanish out from under us. Since we take the d_lock anyway, use that to fetch d_parent and take a reference to it, and then use that reference to call ceph_encode_inode_release. Link: http://tracker.ceph.com/issues/18148 Signed-off-by: Jeff Layton Reviewed-by: Yan, Zheng Signed-off-by: Ilya Dryomov --- fs/ceph/caps.c | 7 +++++-- 1 file changed, 5 insertions(+), 2 deletions(-) (limited to 'fs/ceph') diff --git a/fs/ceph/caps.c b/fs/ceph/caps.c index 94fd76d04683..d1b4c543cab1 100644 --- a/fs/ceph/caps.c +++ b/fs/ceph/caps.c @@ -3926,7 +3926,7 @@ int ceph_encode_inode_release(void **p, struct inode *inode, int ceph_encode_dentry_release(void **p, struct dentry *dentry, int mds, int drop, int unless) { - struct inode *dir = d_inode(dentry->d_parent); + struct dentry *parent; struct ceph_mds_request_release *rel = *p; struct ceph_dentry_info *di = ceph_dentry(dentry); int force = 0; @@ -3941,9 +3941,12 @@ int ceph_encode_dentry_release(void **p, struct dentry *dentry, spin_lock(&dentry->d_lock); if (di->lease_session && di->lease_session->s_mds == mds) force = 1; + parent = dget(dentry->d_parent); spin_unlock(&dentry->d_lock); - ret = ceph_encode_inode_release(p, dir, mds, drop, unless, force); + ret = ceph_encode_inode_release(p, d_inode(parent), mds, drop, + unless, force); + dput(parent); spin_lock(&dentry->d_lock); if (ret && di->lease_session && di->lease_session->s_mds == mds) { -- cgit v1.2.3-59-g8ed1b From ca6c8ae0f7930dad7e10664e3b5bc657dd75be60 Mon Sep 17 00:00:00 2001 From: Jeff Layton Date: Thu, 15 Dec 2016 08:37:59 -0500 Subject: ceph: pass parent inode info to ceph_encode_dentry_release if we have it If we have a parent inode reference already, then we don't need to go back up the directory tree to find one. Link: http://tracker.ceph.com/issues/18148 Signed-off-by: Jeff Layton Reviewed-by: Yan, Zheng Signed-off-by: Ilya Dryomov --- fs/ceph/caps.c | 11 +++++++---- fs/ceph/mds_client.c | 7 +++++-- fs/ceph/super.h | 1 + 3 files changed, 13 insertions(+), 6 deletions(-) (limited to 'fs/ceph') diff --git a/fs/ceph/caps.c b/fs/ceph/caps.c index d1b4c543cab1..4951ab96ffc8 100644 --- a/fs/ceph/caps.c +++ b/fs/ceph/caps.c @@ -3924,9 +3924,10 @@ int ceph_encode_inode_release(void **p, struct inode *inode, } int ceph_encode_dentry_release(void **p, struct dentry *dentry, + struct inode *dir, int mds, int drop, int unless) { - struct dentry *parent; + struct dentry *parent = NULL; struct ceph_mds_request_release *rel = *p; struct ceph_dentry_info *di = ceph_dentry(dentry); int force = 0; @@ -3941,11 +3942,13 @@ int ceph_encode_dentry_release(void **p, struct dentry *dentry, spin_lock(&dentry->d_lock); if (di->lease_session && di->lease_session->s_mds == mds) force = 1; - parent = dget(dentry->d_parent); + if (!dir) { + parent = dget(dentry->d_parent); + dir = d_inode(parent); + } spin_unlock(&dentry->d_lock); - ret = ceph_encode_inode_release(p, d_inode(parent), mds, drop, - unless, force); + ret = ceph_encode_inode_release(p, dir, mds, drop, unless, force); dput(parent); spin_lock(&dentry->d_lock); diff --git a/fs/ceph/mds_client.c b/fs/ceph/mds_client.c index c41c6aab3232..1b78457c992b 100644 --- a/fs/ceph/mds_client.c +++ b/fs/ceph/mds_client.c @@ -1954,10 +1954,13 @@ static struct ceph_msg *create_request_message(struct ceph_mds_client *mdsc, mds, req->r_inode_drop, req->r_inode_unless, 0); if (req->r_dentry_drop) releases += ceph_encode_dentry_release(&p, req->r_dentry, - mds, req->r_dentry_drop, req->r_dentry_unless); + req->r_locked_dir, mds, req->r_dentry_drop, + req->r_dentry_unless); if (req->r_old_dentry_drop) releases += ceph_encode_dentry_release(&p, req->r_old_dentry, - mds, req->r_old_dentry_drop, req->r_old_dentry_unless); + req->r_old_dentry_dir, mds, + req->r_old_dentry_drop, + req->r_old_dentry_unless); if (req->r_old_inode_drop) releases += ceph_encode_inode_release(&p, d_inode(req->r_old_dentry), diff --git a/fs/ceph/super.h b/fs/ceph/super.h index 3373b61faefd..a80a915ca247 100644 --- a/fs/ceph/super.h +++ b/fs/ceph/super.h @@ -904,6 +904,7 @@ extern void ceph_flush_dirty_caps(struct ceph_mds_client *mdsc); extern int ceph_encode_inode_release(void **p, struct inode *inode, int mds, int drop, int unless, int force); extern int ceph_encode_dentry_release(void **p, struct dentry *dn, + struct inode *dir, int mds, int drop, int unless); extern int ceph_get_caps(struct ceph_inode_info *ci, int need, int want, -- cgit v1.2.3-59-g8ed1b From 52953d55917e45fdf62d5e6cb37c61c32dc8d373 Mon Sep 17 00:00:00 2001 From: Seraphime Kirkovski Date: Mon, 26 Dec 2016 10:26:34 +0100 Subject: ceph: cleanup ACCESS_ONCE -> READ_ONCE This removes the uses of ACCESS_ONCE in favor of READ_ONCE Signed-off-by: Seraphime Kirkovski Signed-off-by: Yan, Zheng --- fs/ceph/addr.c | 4 ++-- fs/ceph/caps.c | 2 +- fs/ceph/dir.c | 2 +- fs/ceph/inode.c | 2 +- fs/ceph/mds_client.c | 10 +++++----- 5 files changed, 10 insertions(+), 10 deletions(-) (limited to 'fs/ceph') diff --git a/fs/ceph/addr.c b/fs/ceph/addr.c index e4b066cd912a..39852567d66e 100644 --- a/fs/ceph/addr.c +++ b/fs/ceph/addr.c @@ -771,7 +771,7 @@ static int ceph_writepages_start(struct address_space *mapping, wbc->sync_mode == WB_SYNC_NONE ? "NONE" : (wbc->sync_mode == WB_SYNC_ALL ? "ALL" : "HOLD")); - if (ACCESS_ONCE(fsc->mount_state) == CEPH_MOUNT_SHUTDOWN) { + if (READ_ONCE(fsc->mount_state) == CEPH_MOUNT_SHUTDOWN) { if (ci->i_wrbuffer_ref > 0) { pr_warn_ratelimited( "writepage_start %p %lld forced umount\n", @@ -1194,7 +1194,7 @@ static int ceph_update_writeable_page(struct file *file, int r; struct ceph_snap_context *snapc, *oldest; - if (ACCESS_ONCE(fsc->mount_state) == CEPH_MOUNT_SHUTDOWN) { + if (READ_ONCE(fsc->mount_state) == CEPH_MOUNT_SHUTDOWN) { dout(" page %p forced umount\n", page); unlock_page(page); return -EIO; diff --git a/fs/ceph/caps.c b/fs/ceph/caps.c index 4951ab96ffc8..e3a2a3f32568 100644 --- a/fs/ceph/caps.c +++ b/fs/ceph/caps.c @@ -2477,7 +2477,7 @@ again: if (ci->i_ceph_flags & CEPH_I_CAP_DROPPED) { int mds_wanted; - if (ACCESS_ONCE(mdsc->fsc->mount_state) == + if (READ_ONCE(mdsc->fsc->mount_state) == CEPH_MOUNT_SHUTDOWN) { dout("get_cap_refs %p forced umount\n", inode); *err = -EIO; diff --git a/fs/ceph/dir.c b/fs/ceph/dir.c index 8ab1fdf0bd49..d4385563b70a 100644 --- a/fs/ceph/dir.c +++ b/fs/ceph/dir.c @@ -1194,7 +1194,7 @@ static int ceph_d_revalidate(struct dentry *dentry, unsigned int flags) struct inode *dir; if (flags & LOOKUP_RCU) { - parent = ACCESS_ONCE(dentry->d_parent); + parent = READ_ONCE(dentry->d_parent); dir = d_inode_rcu(parent); if (!dir) return -ECHILD; diff --git a/fs/ceph/inode.c b/fs/ceph/inode.c index 5e659d054b40..4926265f4223 100644 --- a/fs/ceph/inode.c +++ b/fs/ceph/inode.c @@ -1720,7 +1720,7 @@ static void ceph_invalidate_work(struct work_struct *work) mutex_lock(&ci->i_truncate_mutex); - if (ACCESS_ONCE(fsc->mount_state) == CEPH_MOUNT_SHUTDOWN) { + if (READ_ONCE(fsc->mount_state) == CEPH_MOUNT_SHUTDOWN) { pr_warn_ratelimited("invalidate_pages %p %lld forced umount\n", inode, ceph_ino(inode)); mapping_set_error(inode->i_mapping, -EIO); diff --git a/fs/ceph/mds_client.c b/fs/ceph/mds_client.c index 1b78457c992b..176512960b14 100644 --- a/fs/ceph/mds_client.c +++ b/fs/ceph/mds_client.c @@ -1166,7 +1166,7 @@ static int remove_session_caps_cb(struct inode *inode, struct ceph_cap *cap, ci->i_ceph_flags |= CEPH_I_CAP_DROPPED; if (ci->i_wrbuffer_ref > 0 && - ACCESS_ONCE(fsc->mount_state) == CEPH_MOUNT_SHUTDOWN) + READ_ONCE(fsc->mount_state) == CEPH_MOUNT_SHUTDOWN) invalidate = true; while (!list_empty(&ci->i_cap_flush_list)) { @@ -2126,12 +2126,12 @@ static int __do_request(struct ceph_mds_client *mdsc, err = -EIO; goto finish; } - if (ACCESS_ONCE(mdsc->fsc->mount_state) == CEPH_MOUNT_SHUTDOWN) { + if (READ_ONCE(mdsc->fsc->mount_state) == CEPH_MOUNT_SHUTDOWN) { dout("do_request forced umount\n"); err = -EIO; goto finish; } - if (ACCESS_ONCE(mdsc->fsc->mount_state) == CEPH_MOUNT_MOUNTING) { + if (READ_ONCE(mdsc->fsc->mount_state) == CEPH_MOUNT_MOUNTING) { if (mdsc->mdsmap_err) { err = mdsc->mdsmap_err; dout("do_request mdsmap err %d\n", err); @@ -3586,7 +3586,7 @@ void ceph_mdsc_sync(struct ceph_mds_client *mdsc) { u64 want_tid, want_flush; - if (ACCESS_ONCE(mdsc->fsc->mount_state) == CEPH_MOUNT_SHUTDOWN) + if (READ_ONCE(mdsc->fsc->mount_state) == CEPH_MOUNT_SHUTDOWN) return; dout("sync\n"); @@ -3617,7 +3617,7 @@ void ceph_mdsc_sync(struct ceph_mds_client *mdsc) */ static bool done_closing_sessions(struct ceph_mds_client *mdsc, int skipped) { - if (ACCESS_ONCE(mdsc->fsc->mount_state) == CEPH_MOUNT_SHUTDOWN) + if (READ_ONCE(mdsc->fsc->mount_state) == CEPH_MOUNT_SHUTDOWN) return true; return atomic_read(&mdsc->num_sessions) <= skipped; } -- cgit v1.2.3-59-g8ed1b From 0fbc5360bf433ffeccbb94ef806d637049899741 Mon Sep 17 00:00:00 2001 From: Colin Ian King Date: Thu, 29 Dec 2016 20:19:32 +0000 Subject: ceph: fix spelling mistake: "enabing" -> "enabling" trivial fix to spelling mistake in debug message Signed-off-by: Colin Ian King Signed-off-by: Yan, Zheng --- fs/ceph/cache.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'fs/ceph') diff --git a/fs/ceph/cache.c b/fs/ceph/cache.c index 5bc5d37b1217..4e7421caf380 100644 --- a/fs/ceph/cache.c +++ b/fs/ceph/cache.c @@ -234,7 +234,7 @@ void ceph_fscache_file_set_cookie(struct inode *inode, struct file *filp) fscache_enable_cookie(ci->fscache, ceph_fscache_can_enable, inode); if (fscache_cookie_enabled(ci->fscache)) { - dout("fscache_file_set_cookie %p %p enabing cache\n", + dout("fscache_file_set_cookie %p %p enabling cache\n", inode, filp); } } -- cgit v1.2.3-59-g8ed1b From 7c94ba27903c3c46c5829e96efb7c2a8ab7510e9 Mon Sep 17 00:00:00 2001 From: Andreas Gerstmayr Date: Tue, 10 Jan 2017 14:17:56 +0100 Subject: ceph: set io_pages bdi hint This patch sets the io_pages bdi hint based on the rsize mount option. Without this patch large buffered reads (request size > max readahead) are processed sequentially in chunks of the readahead size (i.e. read requests are sent out up to the readahead size, then the do_generic_file_read() function waits until the first page is received). With this patch read requests are sent out at once up to the size specified in the rsize mount option (default: 64 MB). Signed-off-by: Andreas Gerstmayr Acked-by: Jeff Layton Signed-off-by: Yan, Zheng --- Documentation/filesystems/ceph.txt | 5 ++--- fs/ceph/super.c | 8 ++++++++ fs/ceph/super.h | 4 ++-- 3 files changed, 12 insertions(+), 5 deletions(-) (limited to 'fs/ceph') diff --git a/Documentation/filesystems/ceph.txt b/Documentation/filesystems/ceph.txt index f5306ee40ea9..0b302a11718a 100644 --- a/Documentation/filesystems/ceph.txt +++ b/Documentation/filesystems/ceph.txt @@ -98,11 +98,10 @@ Mount Options size. rsize=X - Specify the maximum read size in bytes. By default there is no - maximum. + Specify the maximum read size in bytes. Default: 64 MB. rasize=X - Specify the maximum readahead. + Specify the maximum readahead. Default: 8 MB. mount_timeout=X Specify the timeout value for mount (in seconds), in the case diff --git a/fs/ceph/super.c b/fs/ceph/super.c index 6bd20d707bfd..a0a0b6d02f89 100644 --- a/fs/ceph/super.c +++ b/fs/ceph/super.c @@ -952,6 +952,14 @@ static int ceph_register_bdi(struct super_block *sb, fsc->backing_dev_info.ra_pages = VM_MAX_READAHEAD * 1024 / PAGE_SIZE; + if (fsc->mount_options->rsize > fsc->mount_options->rasize && + fsc->mount_options->rsize >= PAGE_SIZE) + fsc->backing_dev_info.io_pages = + (fsc->mount_options->rsize + PAGE_SIZE - 1) + >> PAGE_SHIFT; + else if (fsc->mount_options->rsize == 0) + fsc->backing_dev_info.io_pages = ULONG_MAX; + err = bdi_register(&fsc->backing_dev_info, NULL, "ceph-%ld", atomic_long_inc_return(&bdi_seq)); if (!err) diff --git a/fs/ceph/super.h b/fs/ceph/super.h index a80a915ca247..f3f9215abf8e 100644 --- a/fs/ceph/super.h +++ b/fs/ceph/super.h @@ -45,8 +45,8 @@ #define ceph_test_mount_opt(fsc, opt) \ (!!((fsc)->mount_options->flags & CEPH_MOUNT_OPT_##opt)) -#define CEPH_RSIZE_DEFAULT 0 /* max read size */ -#define CEPH_RASIZE_DEFAULT (8192*1024) /* readahead */ +#define CEPH_RSIZE_DEFAULT (64*1024*1024) /* max read size */ +#define CEPH_RASIZE_DEFAULT (8192*1024) /* max readahead */ #define CEPH_MAX_READDIR_DEFAULT 1024 #define CEPH_MAX_READDIR_BYTES_DEFAULT (512*1024) #define CEPH_SNAPDIRNAME_DEFAULT ".snap" -- cgit v1.2.3-59-g8ed1b From eb65b919b914e65485fd8134912066f4fafc4f1e Mon Sep 17 00:00:00 2001 From: "Yan, Zheng" Date: Thu, 12 Jan 2017 17:18:00 +0800 Subject: ceph: avoid updating mds_wanted too frequently user space may open/close single file frequently. It's not good to send a clientcaps message to mds for each open/close syscall. Signed-off-by: Yan, Zheng --- fs/ceph/caps.c | 12 +++++++++--- 1 file changed, 9 insertions(+), 3 deletions(-) (limited to 'fs/ceph') diff --git a/fs/ceph/caps.c b/fs/ceph/caps.c index e3a2a3f32568..d941c48e8bff 100644 --- a/fs/ceph/caps.c +++ b/fs/ceph/caps.c @@ -1184,6 +1184,13 @@ static int __send_cap(struct ceph_mds_client *mdsc, struct ceph_cap *cap, delayed = 1; } ci->i_ceph_flags &= ~(CEPH_I_NODELAY | CEPH_I_FLUSH); + if (want & ~cap->mds_wanted) { + /* user space may open/close single file frequently. + * This avoids droping mds_wanted immediately after + * requesting new mds_wanted. + */ + __cap_set_timeouts(mdsc, ci); + } cap->issued &= retain; /* drop bits we don't want */ if (cap->implemented & ~cap->issued) { @@ -2485,15 +2492,14 @@ again: goto out_unlock; } mds_wanted = __ceph_caps_mds_wanted(ci); - if ((mds_wanted & need) != need) { + if (need & ~(mds_wanted & need)) { dout("get_cap_refs %p caps were dropped" " (session killed?)\n", inode); *err = -ESTALE; ret = 1; goto out_unlock; } - if ((mds_wanted & file_wanted) == - (file_wanted & (CEPH_CAP_FILE_RD|CEPH_CAP_FILE_WR))) + if (!(file_wanted & ~mds_wanted)) ci->i_ceph_flags &= ~CEPH_I_CAP_DROPPED; } -- cgit v1.2.3-59-g8ed1b From 24c149ad6914d349d8b64749f20f3f8ea5031fe0 Mon Sep 17 00:00:00 2001 From: Jeff Layton Date: Thu, 12 Jan 2017 14:42:40 -0500 Subject: ceph: fix bogus endianness change in ceph_ioctl_set_layout sparse says: fs/ceph/ioctl.c:100:28: warning: cast to restricted __le64 preferred_osd is a __s64 so we don't need to do any conversion. Also, just remove the cast in ceph_ioctl_get_layout as it's not needed. Signed-off-by: Jeff Layton Reviewed-by: Sage Weil Signed-off-by: Ilya Dryomov --- fs/ceph/ioctl.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'fs/ceph') diff --git a/fs/ceph/ioctl.c b/fs/ceph/ioctl.c index 7d752d53353a..4c9c72f26eb9 100644 --- a/fs/ceph/ioctl.c +++ b/fs/ceph/ioctl.c @@ -25,7 +25,7 @@ static long ceph_ioctl_get_layout(struct file *file, void __user *arg) l.stripe_count = ci->i_layout.stripe_count; l.object_size = ci->i_layout.object_size; l.data_pool = ci->i_layout.pool_id; - l.preferred_osd = (s32)-1; + l.preferred_osd = -1; if (copy_to_user(arg, &l, sizeof(l))) return -EFAULT; } @@ -97,7 +97,7 @@ static long ceph_ioctl_set_layout(struct file *file, void __user *arg) nl.data_pool = ci->i_layout.pool_id; /* this is obsolete, and always -1 */ - nl.preferred_osd = le64_to_cpu(-1); + nl.preferred_osd = -1; err = __validate_layout(mdsc, &nl); if (err) -- cgit v1.2.3-59-g8ed1b From d641df819db8b80198fd85d9de91137e8a823b07 Mon Sep 17 00:00:00 2001 From: "Yan, Zheng" Date: Thu, 19 Jan 2017 11:21:29 +0800 Subject: ceph: update readpages osd request according to size of pages add_to_page_cache_lru() can fails, so the actual pages to read can be smaller than the initial size of osd request. We need to update osd request size in that case. Signed-off-by: Yan, Zheng Reviewed-by: Jeff Layton --- fs/ceph/addr.c | 1 + net/ceph/osd_client.c | 3 ++- 2 files changed, 3 insertions(+), 1 deletion(-) (limited to 'fs/ceph') diff --git a/fs/ceph/addr.c b/fs/ceph/addr.c index 39852567d66e..4547bbf80e4f 100644 --- a/fs/ceph/addr.c +++ b/fs/ceph/addr.c @@ -391,6 +391,7 @@ static int start_read(struct inode *inode, struct list_head *page_list, int max) nr_pages = i; if (nr_pages > 0) { len = nr_pages << PAGE_SHIFT; + osd_req_op_extent_update(req, 0, len); break; } goto out_pages; diff --git a/net/ceph/osd_client.c b/net/ceph/osd_client.c index 842f049abb86..3a2417bb6ff0 100644 --- a/net/ceph/osd_client.c +++ b/net/ceph/osd_client.c @@ -672,7 +672,8 @@ void osd_req_op_extent_update(struct ceph_osd_request *osd_req, BUG_ON(length > previous); op->extent.length = length; - op->indata_len -= previous - length; + if (op->op == CEPH_OSD_OP_WRITE || op->op == CEPH_OSD_OP_WRITEFULL) + op->indata_len -= previous - length; } EXPORT_SYMBOL(osd_req_op_extent_update); -- cgit v1.2.3-59-g8ed1b From 00f06cba53f53f3f7be8ac4f9ba2c2f6a94bca6f Mon Sep 17 00:00:00 2001 From: "Yan, Zheng" Date: Tue, 24 Jan 2017 10:02:32 +0800 Subject: ceph: make sure flushing inode in proper session's cap_flushing list when flushing inode's auth cap changes, we need to move it into the new auth cap session's cap_flushing list Signed-off-by: Yan, Zheng --- fs/ceph/caps.c | 10 ++++++++++ 1 file changed, 10 insertions(+) (limited to 'fs/ceph') diff --git a/fs/ceph/caps.c b/fs/ceph/caps.c index d941c48e8bff..ed8c7addce91 100644 --- a/fs/ceph/caps.c +++ b/fs/ceph/caps.c @@ -3410,6 +3410,7 @@ retry: tcap->implemented |= issued; if (cap == ci->i_auth_cap) ci->i_auth_cap = tcap; + if (!list_empty(&ci->i_cap_flush_list) && ci->i_auth_cap == tcap) { spin_lock(&mdsc->cap_dirty_lock); @@ -3423,9 +3424,18 @@ retry: } else if (tsession) { /* add placeholder for the export tagert */ int flag = (cap == ci->i_auth_cap) ? CEPH_CAP_FLAG_AUTH : 0; + tcap = new_cap; ceph_add_cap(inode, tsession, t_cap_id, -1, issued, 0, t_seq - 1, t_mseq, (u64)-1, flag, &new_cap); + if (!list_empty(&ci->i_cap_flush_list) && + ci->i_auth_cap == tcap) { + spin_lock(&mdsc->cap_dirty_lock); + list_move_tail(&ci->i_flushing_item, + &tcap->session->s_cap_flushing); + spin_unlock(&mdsc->cap_dirty_lock); + } + __ceph_remove_cap(cap, false); goto out_unlock; } -- cgit v1.2.3-59-g8ed1b From c1944fedd8c492ce1c1a99ca9064dcc7bafa80e9 Mon Sep 17 00:00:00 2001 From: "Yan, Zheng" Date: Sun, 29 Jan 2017 22:15:47 +0800 Subject: ceph: avoid calling ceph_renew_caps() infinitely __ceph_caps_mds_wanted() ignores caps from stale session. So the return value of __ceph_caps_mds_wanted() can keep the same across ceph_renew_caps(). This causes try_get_cap_refs() to keep calling ceph_renew_caps(). The fix is ignore the session valid check for the try_get_cap_refs() case. If session is stale, just let the caps requester sleep. Signed-off-by: Yan, Zheng --- fs/ceph/caps.c | 6 +++--- fs/ceph/file.c | 2 +- fs/ceph/super.h | 2 +- 3 files changed, 5 insertions(+), 5 deletions(-) (limited to 'fs/ceph') diff --git a/fs/ceph/caps.c b/fs/ceph/caps.c index ed8c7addce91..3c2dfd72e5b2 100644 --- a/fs/ceph/caps.c +++ b/fs/ceph/caps.c @@ -867,7 +867,7 @@ int __ceph_caps_file_wanted(struct ceph_inode_info *ci) /* * Return caps we have registered with the MDS(s) as 'wanted'. */ -int __ceph_caps_mds_wanted(struct ceph_inode_info *ci) +int __ceph_caps_mds_wanted(struct ceph_inode_info *ci, bool check) { struct ceph_cap *cap; struct rb_node *p; @@ -875,7 +875,7 @@ int __ceph_caps_mds_wanted(struct ceph_inode_info *ci) for (p = rb_first(&ci->i_caps); p; p = rb_next(p)) { cap = rb_entry(p, struct ceph_cap, ci_node); - if (!__cap_is_valid(cap)) + if (check && !__cap_is_valid(cap)) continue; if (cap == ci->i_auth_cap) mds_wanted |= cap->mds_wanted; @@ -2491,7 +2491,7 @@ again: ret = 1; goto out_unlock; } - mds_wanted = __ceph_caps_mds_wanted(ci); + mds_wanted = __ceph_caps_mds_wanted(ci, false); if (need & ~(mds_wanted & need)) { dout("get_cap_refs %p caps were dropped" " (session killed?)\n", inode); diff --git a/fs/ceph/file.c b/fs/ceph/file.c index 045d30d26624..beb24f8bb917 100644 --- a/fs/ceph/file.c +++ b/fs/ceph/file.c @@ -283,7 +283,7 @@ int ceph_open(struct inode *inode, struct file *file) spin_lock(&ci->i_ceph_lock); if (__ceph_is_any_real_caps(ci) && (((fmode & CEPH_FILE_MODE_WR) == 0) || ci->i_auth_cap)) { - int mds_wanted = __ceph_caps_mds_wanted(ci); + int mds_wanted = __ceph_caps_mds_wanted(ci, true); int issued = __ceph_caps_issued(ci, NULL); dout("open %p fmode %d want %s issued %s using existing\n", diff --git a/fs/ceph/super.h b/fs/ceph/super.h index f3f9215abf8e..6477264bfc7e 100644 --- a/fs/ceph/super.h +++ b/fs/ceph/super.h @@ -602,7 +602,7 @@ static inline int __ceph_caps_wanted(struct ceph_inode_info *ci) } /* what the mds thinks we want */ -extern int __ceph_caps_mds_wanted(struct ceph_inode_info *ci); +extern int __ceph_caps_mds_wanted(struct ceph_inode_info *ci, bool check); extern void ceph_caps_init(struct ceph_mds_client *mdsc); extern void ceph_caps_finalize(struct ceph_mds_client *mdsc); -- cgit v1.2.3-59-g8ed1b From 6fffaef954c432f6ddb42ec1064087e8ed704ca8 Mon Sep 17 00:00:00 2001 From: Jeff Layton Date: Tue, 31 Jan 2017 10:55:38 -0500 Subject: ceph: remove "Debugging hook" from ceph_fill_trace Keeping around commented out code is just asking for it to bitrot and makes viewing the code under cscope more confusing. If we really need this, then we can revert this patch and put it under a Kconfig option. Signed-off-by: Jeff Layton Reviewed-by: Yan, Zheng Signed-off-by: Ilya Dryomov --- fs/ceph/inode.c | 34 ---------------------------------- 1 file changed, 34 deletions(-) (limited to 'fs/ceph') diff --git a/fs/ceph/inode.c b/fs/ceph/inode.c index 4926265f4223..38a54ebf647d 100644 --- a/fs/ceph/inode.c +++ b/fs/ceph/inode.c @@ -1120,40 +1120,6 @@ int ceph_fill_trace(struct super_block *sb, struct ceph_mds_request *req, dout("fill_trace %p is_dentry %d is_target %d\n", req, rinfo->head->is_dentry, rinfo->head->is_target); -#if 0 - /* - * Debugging hook: - * - * If we resend completed ops to a recovering mds, we get no - * trace. Since that is very rare, pretend this is the case - * to ensure the 'no trace' handlers in the callers behave. - * - * Fill in inodes unconditionally to avoid breaking cap - * invariants. - */ - if (rinfo->head->op & CEPH_MDS_OP_WRITE) { - pr_info("fill_trace faking empty trace on %lld %s\n", - req->r_tid, ceph_mds_op_name(rinfo->head->op)); - if (rinfo->head->is_dentry) { - rinfo->head->is_dentry = 0; - err = fill_inode(req->r_locked_dir, - &rinfo->diri, rinfo->dirfrag, - session, req->r_request_started, -1); - } - if (rinfo->head->is_target) { - rinfo->head->is_target = 0; - ininfo = rinfo->targeti.in; - vino.ino = le64_to_cpu(ininfo->ino); - vino.snap = le64_to_cpu(ininfo->snapid); - in = ceph_get_inode(sb, vino); - err = fill_inode(in, &rinfo->targeti, NULL, - session, req->r_request_started, - req->r_fmode); - iput(in); - } - } -#endif - if (!rinfo->head->is_target && !rinfo->head->is_dentry) { dout("fill_trace reply is empty!\n"); if (rinfo->head->result == 0 && req->r_locked_dir) -- cgit v1.2.3-59-g8ed1b From f5a03b080450784e671998921feb62fd3846c953 Mon Sep 17 00:00:00 2001 From: Jeff Layton Date: Tue, 31 Jan 2017 11:06:13 -0500 Subject: ceph: drop session argument to ceph_fill_trace Just get it from r_session since that's what's always passed in. Signed-off-by: Jeff Layton Reviewed-by: Yan, Zheng Signed-off-by: Ilya Dryomov --- fs/ceph/inode.c | 4 ++-- fs/ceph/mds_client.c | 2 +- fs/ceph/super.h | 3 +-- 3 files changed, 4 insertions(+), 5 deletions(-) (limited to 'fs/ceph') diff --git a/fs/ceph/inode.c b/fs/ceph/inode.c index 38a54ebf647d..b18462c64cdd 100644 --- a/fs/ceph/inode.c +++ b/fs/ceph/inode.c @@ -1108,9 +1108,9 @@ out: * * Called with snap_rwsem (read). */ -int ceph_fill_trace(struct super_block *sb, struct ceph_mds_request *req, - struct ceph_mds_session *session) +int ceph_fill_trace(struct super_block *sb, struct ceph_mds_request *req) { + struct ceph_mds_session *session = req->r_session; struct ceph_mds_reply_info_parsed *rinfo = &req->r_reply_info; struct inode *in = NULL; struct ceph_vino vino; diff --git a/fs/ceph/mds_client.c b/fs/ceph/mds_client.c index 176512960b14..4a68067a76b5 100644 --- a/fs/ceph/mds_client.c +++ b/fs/ceph/mds_client.c @@ -2516,7 +2516,7 @@ static void handle_reply(struct ceph_mds_session *session, struct ceph_msg *msg) /* insert trace into our cache */ mutex_lock(&req->r_fill_mutex); current->journal_info = req; - err = ceph_fill_trace(mdsc->fsc->sb, req, req->r_session); + err = ceph_fill_trace(mdsc->fsc->sb, req); if (err == 0) { if (result == 0 && (req->r_op == CEPH_MDS_OP_READDIR || req->r_op == CEPH_MDS_OP_LSSNAP)) diff --git a/fs/ceph/super.h b/fs/ceph/super.h index 6477264bfc7e..950170136be9 100644 --- a/fs/ceph/super.h +++ b/fs/ceph/super.h @@ -764,8 +764,7 @@ extern void ceph_fill_file_time(struct inode *inode, int issued, u64 time_warp_seq, struct timespec *ctime, struct timespec *mtime, struct timespec *atime); extern int ceph_fill_trace(struct super_block *sb, - struct ceph_mds_request *req, - struct ceph_mds_session *session); + struct ceph_mds_request *req); extern int ceph_readdir_prepopulate(struct ceph_mds_request *req, struct ceph_mds_session *session); -- cgit v1.2.3-59-g8ed1b From bc2de10dc4da5036ada3381775bd966f0c21c603 Mon Sep 17 00:00:00 2001 From: Jeff Layton Date: Wed, 1 Feb 2017 13:49:09 -0500 Subject: ceph: convert bools in ceph_mds_request to a new r_req_flags field Currently, we have a bunch of bool flags in struct ceph_mds_request. We need more flags though, but each bool takes (at least) a byte. Those add up over time. Merge all of the existing bools in this struct into a single unsigned long, and use the set/test/clear_bit macros to manipulate them. These are atomic operations, but that is required here to prevent load/modify/store races. The existing flags are protected by different locks, so we can't rely on them for that purpose. Signed-off-by: Jeff Layton Reviewed-by: Yan, Zheng Signed-off-by: Ilya Dryomov --- fs/ceph/debugfs.c | 2 +- fs/ceph/dir.c | 4 ++-- fs/ceph/inode.c | 18 +++++++++--------- fs/ceph/mds_client.c | 47 +++++++++++++++++++++++++---------------------- fs/ceph/mds_client.h | 12 ++++++++---- 5 files changed, 45 insertions(+), 38 deletions(-) (limited to 'fs/ceph') diff --git a/fs/ceph/debugfs.c b/fs/ceph/debugfs.c index 39ff678e567f..f2ae393e2c31 100644 --- a/fs/ceph/debugfs.c +++ b/fs/ceph/debugfs.c @@ -70,7 +70,7 @@ static int mdsc_show(struct seq_file *s, void *p) seq_printf(s, "%s", ceph_mds_op_name(req->r_op)); - if (req->r_got_unsafe) + if (test_bit(CEPH_MDS_R_GOT_UNSAFE, &req->r_req_flags)) seq_puts(s, "\t(unsafe)"); else seq_puts(s, "\t"); diff --git a/fs/ceph/dir.c b/fs/ceph/dir.c index d4385563b70a..5fb86d71220e 100644 --- a/fs/ceph/dir.c +++ b/fs/ceph/dir.c @@ -371,7 +371,7 @@ more: /* hints to request -> mds selection code */ req->r_direct_mode = USE_AUTH_MDS; req->r_direct_hash = ceph_frag_value(frag); - req->r_direct_is_hash = true; + __set_bit(CEPH_MDS_R_DIRECT_IS_HASH, &req->r_req_flags); if (fi->last_name) { req->r_path2 = kstrdup(fi->last_name, GFP_KERNEL); if (!req->r_path2) { @@ -417,7 +417,7 @@ more: fi->frag = frag; fi->last_readdir = req; - if (req->r_did_prepopulate) { + if (test_bit(CEPH_MDS_R_DID_PREPOPULATE, &req->r_req_flags)) { fi->readdir_cache_idx = req->r_readdir_cache_idx; if (fi->readdir_cache_idx < 0) { /* preclude from marking dir ordered */ diff --git a/fs/ceph/inode.c b/fs/ceph/inode.c index b18462c64cdd..ebfb156aba89 100644 --- a/fs/ceph/inode.c +++ b/fs/ceph/inode.c @@ -1199,8 +1199,8 @@ retry_lookup: err = fill_inode(in, req->r_locked_page, &rinfo->targeti, NULL, session, req->r_request_started, - (!req->r_aborted && rinfo->head->result == 0) ? - req->r_fmode : -1, + (!test_bit(CEPH_MDS_R_ABORTED, &req->r_req_flags) && + rinfo->head->result == 0) ? req->r_fmode : -1, &req->r_caps_reservation); if (err < 0) { pr_err("fill_inode badness %p %llx.%llx\n", @@ -1213,8 +1213,8 @@ retry_lookup: * ignore null lease/binding on snapdir ENOENT, or else we * will have trouble splicing in the virtual snapdir later */ - if (rinfo->head->is_dentry && !req->r_aborted && - req->r_locked_dir && + if (rinfo->head->is_dentry && req->r_locked_dir && + !test_bit(CEPH_MDS_R_ABORTED, &req->r_req_flags) && (rinfo->head->is_target || strncmp(req->r_dentry->d_name.name, fsc->mount_options->snapdir_name, req->r_dentry->d_name.len))) { @@ -1317,9 +1317,9 @@ retry_lookup: update_dentry_lease(dn, rinfo->dlease, session, req->r_request_started); dout(" final dn %p\n", dn); - } else if (!req->r_aborted && - (req->r_op == CEPH_MDS_OP_LOOKUPSNAP || - req->r_op == CEPH_MDS_OP_MKSNAP)) { + } else if ((req->r_op == CEPH_MDS_OP_LOOKUPSNAP || + req->r_op == CEPH_MDS_OP_MKSNAP) && + !test_bit(CEPH_MDS_R_ABORTED, &req->r_req_flags)) { struct dentry *dn = req->r_dentry; struct inode *dir = req->r_locked_dir; @@ -1444,7 +1444,7 @@ int ceph_readdir_prepopulate(struct ceph_mds_request *req, u32 fpos_offset; struct ceph_readdir_cache_control cache_ctl = {}; - if (req->r_aborted) + if (test_bit(CEPH_MDS_R_ABORTED, &req->r_req_flags)) return readdir_prepopulate_inodes_only(req, session); if (rinfo->hash_order && req->r_path2) { @@ -1598,7 +1598,7 @@ next_item: } out: if (err == 0 && skipped == 0) { - req->r_did_prepopulate = true; + set_bit(CEPH_MDS_R_DID_PREPOPULATE, &req->r_req_flags); req->r_readdir_cache_idx = cache_ctl.index; } ceph_readdir_cache_release(&cache_ctl); diff --git a/fs/ceph/mds_client.c b/fs/ceph/mds_client.c index 4a68067a76b5..ccf75a3260e8 100644 --- a/fs/ceph/mds_client.c +++ b/fs/ceph/mds_client.c @@ -644,13 +644,15 @@ static void __unregister_request(struct ceph_mds_client *mdsc, erase_request(&mdsc->request_tree, req); - if (req->r_unsafe_dir && req->r_got_unsafe) { + if (req->r_unsafe_dir && + test_bit(CEPH_MDS_R_GOT_UNSAFE, &req->r_req_flags)) { struct ceph_inode_info *ci = ceph_inode(req->r_unsafe_dir); spin_lock(&ci->i_unsafe_lock); list_del_init(&req->r_unsafe_dir_item); spin_unlock(&ci->i_unsafe_lock); } - if (req->r_target_inode && req->r_got_unsafe) { + if (req->r_target_inode && + test_bit(CEPH_MDS_R_GOT_UNSAFE, &req->r_req_flags)) { struct ceph_inode_info *ci = ceph_inode(req->r_target_inode); spin_lock(&ci->i_unsafe_lock); list_del_init(&req->r_unsafe_target_item); @@ -705,7 +707,7 @@ static int __choose_mds(struct ceph_mds_client *mdsc, int mode = req->r_direct_mode; int mds = -1; u32 hash = req->r_direct_hash; - bool is_hash = req->r_direct_is_hash; + bool is_hash = test_bit(CEPH_MDS_R_DIRECT_IS_HASH, &req->r_req_flags); /* * is there a specific mds we should try? ignore hint if we have @@ -2042,7 +2044,7 @@ static int __prepare_send_request(struct ceph_mds_client *mdsc, dout("prepare_send_request %p tid %lld %s (attempt %d)\n", req, req->r_tid, ceph_mds_op_name(req->r_op), req->r_attempts); - if (req->r_got_unsafe) { + if (test_bit(CEPH_MDS_R_GOT_UNSAFE, &req->r_req_flags)) { void *p; /* * Replay. Do not regenerate message (and rebuild @@ -2091,7 +2093,7 @@ static int __prepare_send_request(struct ceph_mds_client *mdsc, rhead = msg->front.iov_base; rhead->oldest_client_tid = cpu_to_le64(__get_oldest_tid(mdsc)); - if (req->r_got_unsafe) + if (test_bit(CEPH_MDS_R_GOT_UNSAFE, &req->r_req_flags)) flags |= CEPH_MDS_FLAG_REPLAY; if (req->r_locked_dir) flags |= CEPH_MDS_FLAG_WANT_DENTRY; @@ -2114,8 +2116,8 @@ static int __do_request(struct ceph_mds_client *mdsc, int mds = -1; int err = 0; - if (req->r_err || req->r_got_result) { - if (req->r_aborted) + if (req->r_err || test_bit(CEPH_MDS_R_GOT_RESULT, &req->r_req_flags)) { + if (test_bit(CEPH_MDS_R_ABORTED, &req->r_req_flags)) __unregister_request(mdsc, req); goto out; } @@ -2245,7 +2247,7 @@ static void kick_requests(struct ceph_mds_client *mdsc, int mds) while (p) { req = rb_entry(p, struct ceph_mds_request, r_node); p = rb_next(p); - if (req->r_got_unsafe) + if (test_bit(CEPH_MDS_R_GOT_UNSAFE, &req->r_req_flags)) continue; if (req->r_attempts > 0) continue; /* only new requests */ @@ -2319,7 +2321,7 @@ int ceph_mdsc_do_request(struct ceph_mds_client *mdsc, mutex_lock(&mdsc->mutex); /* only abort if we didn't race with a real reply */ - if (req->r_got_result) { + if (test_bit(CEPH_MDS_R_GOT_RESULT, &req->r_req_flags)) { err = le32_to_cpu(req->r_reply_info.head->result); } else if (err < 0) { dout("aborted request %lld with %d\n", req->r_tid, err); @@ -2331,7 +2333,7 @@ int ceph_mdsc_do_request(struct ceph_mds_client *mdsc, */ mutex_lock(&req->r_fill_mutex); req->r_err = err; - req->r_aborted = true; + set_bit(CEPH_MDS_R_ABORTED, &req->r_req_flags); mutex_unlock(&req->r_fill_mutex); if (req->r_locked_dir && @@ -2409,14 +2411,14 @@ static void handle_reply(struct ceph_mds_session *session, struct ceph_msg *msg) } /* dup? */ - if ((req->r_got_unsafe && !head->safe) || - (req->r_got_safe && head->safe)) { + if ((test_bit(CEPH_MDS_R_GOT_UNSAFE, &req->r_req_flags) && !head->safe) || + (test_bit(CEPH_MDS_R_GOT_SAFE, &req->r_req_flags) && head->safe)) { pr_warn("got a dup %s reply on %llu from mds%d\n", head->safe ? "safe" : "unsafe", tid, mds); mutex_unlock(&mdsc->mutex); goto out; } - if (req->r_got_safe) { + if (test_bit(CEPH_MDS_R_GOT_SAFE, &req->r_req_flags)) { pr_warn("got unsafe after safe on %llu from mds%d\n", tid, mds); mutex_unlock(&mdsc->mutex); @@ -2455,10 +2457,10 @@ static void handle_reply(struct ceph_mds_session *session, struct ceph_msg *msg) if (head->safe) { - req->r_got_safe = true; + set_bit(CEPH_MDS_R_GOT_SAFE, &req->r_req_flags); __unregister_request(mdsc, req); - if (req->r_got_unsafe) { + if (test_bit(CEPH_MDS_R_GOT_UNSAFE, &req->r_req_flags)) { /* * We already handled the unsafe response, now do the * cleanup. No need to examine the response; the MDS @@ -2476,7 +2478,7 @@ static void handle_reply(struct ceph_mds_session *session, struct ceph_msg *msg) goto out; } } else { - req->r_got_unsafe = true; + set_bit(CEPH_MDS_R_GOT_UNSAFE, &req->r_req_flags); list_add_tail(&req->r_unsafe_item, &req->r_session->s_unsafe); if (req->r_unsafe_dir) { struct ceph_inode_info *ci = @@ -2530,7 +2532,8 @@ static void handle_reply(struct ceph_mds_session *session, struct ceph_msg *msg) if (realm) ceph_put_snap_realm(mdsc, realm); - if (err == 0 && req->r_got_unsafe && req->r_target_inode) { + if (err == 0 && req->r_target_inode && + test_bit(CEPH_MDS_R_GOT_UNSAFE, &req->r_req_flags)) { struct ceph_inode_info *ci = ceph_inode(req->r_target_inode); spin_lock(&ci->i_unsafe_lock); list_add_tail(&req->r_unsafe_target_item, &ci->i_unsafe_iops); @@ -2538,12 +2541,12 @@ static void handle_reply(struct ceph_mds_session *session, struct ceph_msg *msg) } out_err: mutex_lock(&mdsc->mutex); - if (!req->r_aborted) { + if (!test_bit(CEPH_MDS_R_ABORTED, &req->r_req_flags)) { if (err) { req->r_err = err; } else { req->r_reply = ceph_msg_get(msg); - req->r_got_result = true; + set_bit(CEPH_MDS_R_GOT_RESULT, &req->r_req_flags); } } else { dout("reply arrived after request %lld was aborted\n", tid); @@ -2587,7 +2590,7 @@ static void handle_forward(struct ceph_mds_client *mdsc, goto out; /* dup reply? */ } - if (req->r_aborted) { + if (test_bit(CEPH_MDS_R_ABORTED, &req->r_req_flags)) { dout("forward tid %llu aborted, unregistering\n", tid); __unregister_request(mdsc, req); } else if (fwd_seq <= req->r_num_fwd) { @@ -2597,7 +2600,7 @@ static void handle_forward(struct ceph_mds_client *mdsc, /* resend. forward race not possible; mds would drop */ dout("forward tid %llu to mds%d (we resend)\n", tid, next_mds); BUG_ON(req->r_err); - BUG_ON(req->r_got_result); + BUG_ON(test_bit(CEPH_MDS_R_GOT_RESULT, &req->r_req_flags)); req->r_attempts = 0; req->r_num_fwd = fwd_seq; req->r_resend_mds = next_mds; @@ -2762,7 +2765,7 @@ static void replay_unsafe_requests(struct ceph_mds_client *mdsc, while (p) { req = rb_entry(p, struct ceph_mds_request, r_node); p = rb_next(p); - if (req->r_got_unsafe) + if (test_bit(CEPH_MDS_R_GOT_UNSAFE, &req->r_req_flags)) continue; if (req->r_attempts == 0) continue; /* only old requests */ diff --git a/fs/ceph/mds_client.h b/fs/ceph/mds_client.h index 3c6f77b7bb02..409b0e3c3b7a 100644 --- a/fs/ceph/mds_client.h +++ b/fs/ceph/mds_client.h @@ -205,6 +205,14 @@ struct ceph_mds_request { struct inode *r_locked_dir; /* dir (if any) i_mutex locked by vfs */ struct inode *r_target_inode; /* resulting inode */ +#define CEPH_MDS_R_DIRECT_IS_HASH (1) /* r_direct_hash is valid */ +#define CEPH_MDS_R_ABORTED (2) /* call was aborted */ +#define CEPH_MDS_R_GOT_UNSAFE (3) /* got an unsafe reply */ +#define CEPH_MDS_R_GOT_SAFE (4) /* got a safe reply */ +#define CEPH_MDS_R_GOT_RESULT (5) /* got a result */ +#define CEPH_MDS_R_DID_PREPOPULATE (6) /* prepopulated readdir */ + unsigned long r_req_flags; + struct mutex r_fill_mutex; union ceph_mds_request_args r_args; @@ -216,7 +224,6 @@ struct ceph_mds_request { /* for choosing which mds to send this request to */ int r_direct_mode; u32 r_direct_hash; /* choose dir frag based on this dentry hash */ - bool r_direct_is_hash; /* true if r_direct_hash is valid */ /* data payload is used for xattr ops */ struct ceph_pagelist *r_pagelist; @@ -234,7 +241,6 @@ struct ceph_mds_request { struct ceph_mds_reply_info_parsed r_reply_info; struct page *r_locked_page; int r_err; - bool r_aborted; unsigned long r_timeout; /* optional. jiffies, 0 is "wait forever" */ unsigned long r_started; /* start time to measure timeout against */ @@ -262,9 +268,7 @@ struct ceph_mds_request { ceph_mds_request_callback_t r_callback; ceph_mds_request_wait_callback_t r_wait_for_completion; struct list_head r_unsafe_item; /* per-session unsafe list item */ - bool r_got_unsafe, r_got_safe, r_got_result; - bool r_did_prepopulate; long long r_dir_release_cnt; long long r_dir_ordered_cnt; int r_readdir_cache_idx; -- cgit v1.2.3-59-g8ed1b From 3dd69aabcef3d835446a9a1e11d2eab0e6e35e95 Mon Sep 17 00:00:00 2001 From: Jeff Layton Date: Tue, 31 Jan 2017 10:28:26 -0500 Subject: ceph: add a new flag to indicate whether parent is locked struct ceph_mds_request has an r_locked_dir pointer, which is set to indicate the parent inode and that its i_rwsem is locked. In some critical places, we need to be able to indicate the parent inode to the request handling code, even when its i_rwsem may not be locked. Most of the code that operates on r_locked_dir doesn't require that the i_rwsem be locked. We only really need it to handle manipulation of the dcache. The rest (filling of the inode, updating dentry leases, etc.) already has its own locking. Add a new r_req_flags bit that indicates whether the parent is locked when doing the request, and rename the pointer to "r_parent". For now, all the places that set r_parent also set this flag, but that will change in a later patch. Signed-off-by: Jeff Layton Reviewed-by: Yan, Zheng Signed-off-by: Ilya Dryomov --- fs/ceph/dir.c | 21 ++++++++++++++------- fs/ceph/export.c | 3 ++- fs/ceph/file.c | 3 ++- fs/ceph/inode.c | 13 +++++++------ fs/ceph/mds_client.c | 24 ++++++++++++------------ fs/ceph/mds_client.h | 3 ++- 6 files changed, 39 insertions(+), 28 deletions(-) (limited to 'fs/ceph') diff --git a/fs/ceph/dir.c b/fs/ceph/dir.c index 5fb86d71220e..c8562b8a8c7e 100644 --- a/fs/ceph/dir.c +++ b/fs/ceph/dir.c @@ -752,7 +752,8 @@ static struct dentry *ceph_lookup(struct inode *dir, struct dentry *dentry, mask |= CEPH_CAP_XATTR_SHARED; req->r_args.getattr.mask = cpu_to_le32(mask); - req->r_locked_dir = dir; + req->r_parent = dir; + set_bit(CEPH_MDS_R_PARENT_LOCKED, &req->r_req_flags); err = ceph_mdsc_do_request(mdsc, NULL, req); err = ceph_handle_snapdir(req, dentry, err); dentry = ceph_finish_lookup(req, dentry, err); @@ -813,7 +814,8 @@ static int ceph_mknod(struct inode *dir, struct dentry *dentry, } req->r_dentry = dget(dentry); req->r_num_caps = 2; - req->r_locked_dir = dir; + req->r_parent = dir; + set_bit(CEPH_MDS_R_PARENT_LOCKED, &req->r_req_flags); req->r_args.mknod.mode = cpu_to_le32(mode); req->r_args.mknod.rdev = cpu_to_le32(rdev); req->r_dentry_drop = CEPH_CAP_FILE_SHARED; @@ -864,7 +866,8 @@ static int ceph_symlink(struct inode *dir, struct dentry *dentry, ceph_mdsc_put_request(req); goto out; } - req->r_locked_dir = dir; + req->r_parent = dir; + set_bit(CEPH_MDS_R_PARENT_LOCKED, &req->r_req_flags); req->r_dentry = dget(dentry); req->r_num_caps = 2; req->r_dentry_drop = CEPH_CAP_FILE_SHARED; @@ -913,7 +916,8 @@ static int ceph_mkdir(struct inode *dir, struct dentry *dentry, umode_t mode) req->r_dentry = dget(dentry); req->r_num_caps = 2; - req->r_locked_dir = dir; + req->r_parent = dir; + set_bit(CEPH_MDS_R_PARENT_LOCKED, &req->r_req_flags); req->r_args.mkdir.mode = cpu_to_le32(mode); req->r_dentry_drop = CEPH_CAP_FILE_SHARED; req->r_dentry_unless = CEPH_CAP_FILE_EXCL; @@ -957,7 +961,8 @@ static int ceph_link(struct dentry *old_dentry, struct inode *dir, req->r_dentry = dget(dentry); req->r_num_caps = 2; req->r_old_dentry = dget(old_dentry); - req->r_locked_dir = dir; + req->r_parent = dir; + set_bit(CEPH_MDS_R_PARENT_LOCKED, &req->r_req_flags); req->r_dentry_drop = CEPH_CAP_FILE_SHARED; req->r_dentry_unless = CEPH_CAP_FILE_EXCL; /* release LINK_SHARED on source inode (mds will lock it) */ @@ -1023,7 +1028,8 @@ static int ceph_unlink(struct inode *dir, struct dentry *dentry) } req->r_dentry = dget(dentry); req->r_num_caps = 2; - req->r_locked_dir = dir; + req->r_parent = dir; + set_bit(CEPH_MDS_R_PARENT_LOCKED, &req->r_req_flags); req->r_dentry_drop = CEPH_CAP_FILE_SHARED; req->r_dentry_unless = CEPH_CAP_FILE_EXCL; req->r_inode_drop = drop_caps_for_unlink(inode); @@ -1066,7 +1072,8 @@ static int ceph_rename(struct inode *old_dir, struct dentry *old_dentry, req->r_num_caps = 2; req->r_old_dentry = dget(old_dentry); req->r_old_dentry_dir = old_dir; - req->r_locked_dir = new_dir; + req->r_parent = new_dir; + set_bit(CEPH_MDS_R_PARENT_LOCKED, &req->r_req_flags); req->r_old_dentry_drop = CEPH_CAP_FILE_SHARED; req->r_old_dentry_unless = CEPH_CAP_FILE_EXCL; req->r_dentry_drop = CEPH_CAP_FILE_SHARED; diff --git a/fs/ceph/export.c b/fs/ceph/export.c index 180bbef760f2..e8f11fa565c5 100644 --- a/fs/ceph/export.c +++ b/fs/ceph/export.c @@ -207,7 +207,8 @@ static int ceph_get_name(struct dentry *parent, char *name, req->r_inode = d_inode(child); ihold(d_inode(child)); req->r_ino2 = ceph_vino(d_inode(parent)); - req->r_locked_dir = d_inode(parent); + req->r_parent = d_inode(parent); + set_bit(CEPH_MDS_R_PARENT_LOCKED, &req->r_req_flags); req->r_num_caps = 2; err = ceph_mdsc_do_request(mdsc, NULL, req); diff --git a/fs/ceph/file.c b/fs/ceph/file.c index beb24f8bb917..baad3b688ada 100644 --- a/fs/ceph/file.c +++ b/fs/ceph/file.c @@ -379,7 +379,8 @@ int ceph_atomic_open(struct inode *dir, struct dentry *dentry, mask |= CEPH_CAP_XATTR_SHARED; req->r_args.open.mask = cpu_to_le32(mask); - req->r_locked_dir = dir; /* caller holds dir->i_mutex */ + req->r_parent = dir; + set_bit(CEPH_MDS_R_PARENT_LOCKED, &req->r_req_flags); err = ceph_mdsc_do_request(mdsc, (flags & (O_CREAT|O_TRUNC)) ? dir : NULL, req); diff --git a/fs/ceph/inode.c b/fs/ceph/inode.c index ebfb156aba89..35a8c453bea6 100644 --- a/fs/ceph/inode.c +++ b/fs/ceph/inode.c @@ -1122,13 +1122,13 @@ int ceph_fill_trace(struct super_block *sb, struct ceph_mds_request *req) if (!rinfo->head->is_target && !rinfo->head->is_dentry) { dout("fill_trace reply is empty!\n"); - if (rinfo->head->result == 0 && req->r_locked_dir) + if (rinfo->head->result == 0 && req->r_parent) ceph_invalidate_dir_request(req); return 0; } if (rinfo->head->is_dentry) { - struct inode *dir = req->r_locked_dir; + struct inode *dir = req->r_parent; if (dir) { err = fill_inode(dir, NULL, @@ -1213,8 +1213,9 @@ retry_lookup: * ignore null lease/binding on snapdir ENOENT, or else we * will have trouble splicing in the virtual snapdir later */ - if (rinfo->head->is_dentry && req->r_locked_dir && - !test_bit(CEPH_MDS_R_ABORTED, &req->r_req_flags) && + if (rinfo->head->is_dentry && + !test_bit(CEPH_MDS_R_ABORTED, &req->r_req_flags) && + test_bit(CEPH_MDS_R_PARENT_LOCKED, &req->r_req_flags) && (rinfo->head->is_target || strncmp(req->r_dentry->d_name.name, fsc->mount_options->snapdir_name, req->r_dentry->d_name.len))) { @@ -1223,7 +1224,7 @@ retry_lookup: * mknod symlink mkdir : null -> new inode * unlink : linked -> null */ - struct inode *dir = req->r_locked_dir; + struct inode *dir = req->r_parent; struct dentry *dn = req->r_dentry; bool have_dir_cap, have_lease; @@ -1321,7 +1322,7 @@ retry_lookup: req->r_op == CEPH_MDS_OP_MKSNAP) && !test_bit(CEPH_MDS_R_ABORTED, &req->r_req_flags)) { struct dentry *dn = req->r_dentry; - struct inode *dir = req->r_locked_dir; + struct inode *dir = req->r_parent; /* fill out a snapdir LOOKUPSNAP dentry */ BUG_ON(!dn); diff --git a/fs/ceph/mds_client.c b/fs/ceph/mds_client.c index ccf75a3260e8..52521f339745 100644 --- a/fs/ceph/mds_client.c +++ b/fs/ceph/mds_client.c @@ -547,8 +547,8 @@ void ceph_mdsc_release_request(struct kref *kref) ceph_put_cap_refs(ceph_inode(req->r_inode), CEPH_CAP_PIN); iput(req->r_inode); } - if (req->r_locked_dir) - ceph_put_cap_refs(ceph_inode(req->r_locked_dir), CEPH_CAP_PIN); + if (req->r_parent) + ceph_put_cap_refs(ceph_inode(req->r_parent), CEPH_CAP_PIN); iput(req->r_target_inode); if (req->r_dentry) dput(req->r_dentry); @@ -735,7 +735,7 @@ static int __choose_mds(struct ceph_mds_client *mdsc, rcu_read_lock(); parent = req->r_dentry->d_parent; - dir = req->r_locked_dir ? : d_inode_rcu(parent); + dir = req->r_parent ? : d_inode_rcu(parent); if (!dir || dir->i_sb != mdsc->fsc->sb) { /* not this fs or parent went negative */ @@ -1894,7 +1894,7 @@ static struct ceph_msg *create_request_message(struct ceph_mds_client *mdsc, int ret; ret = set_request_path_attr(req->r_inode, req->r_dentry, - req->r_locked_dir, req->r_path1, req->r_ino1.ino, + req->r_parent, req->r_path1, req->r_ino1.ino, &path1, &pathlen1, &ino1, &freepath1); if (ret < 0) { msg = ERR_PTR(ret); @@ -1956,7 +1956,7 @@ static struct ceph_msg *create_request_message(struct ceph_mds_client *mdsc, mds, req->r_inode_drop, req->r_inode_unless, 0); if (req->r_dentry_drop) releases += ceph_encode_dentry_release(&p, req->r_dentry, - req->r_locked_dir, mds, req->r_dentry_drop, + req->r_parent, mds, req->r_dentry_drop, req->r_dentry_unless); if (req->r_old_dentry_drop) releases += ceph_encode_dentry_release(&p, req->r_old_dentry, @@ -2095,14 +2095,14 @@ static int __prepare_send_request(struct ceph_mds_client *mdsc, rhead->oldest_client_tid = cpu_to_le64(__get_oldest_tid(mdsc)); if (test_bit(CEPH_MDS_R_GOT_UNSAFE, &req->r_req_flags)) flags |= CEPH_MDS_FLAG_REPLAY; - if (req->r_locked_dir) + if (req->r_parent) flags |= CEPH_MDS_FLAG_WANT_DENTRY; rhead->flags = cpu_to_le32(flags); rhead->num_fwd = req->r_num_fwd; rhead->num_retry = req->r_attempts - 1; rhead->ino = 0; - dout(" r_locked_dir = %p\n", req->r_locked_dir); + dout(" r_parent = %p\n", req->r_parent); return 0; } @@ -2282,11 +2282,11 @@ int ceph_mdsc_do_request(struct ceph_mds_client *mdsc, dout("do_request on %p\n", req); - /* take CAP_PIN refs for r_inode, r_locked_dir, r_old_dentry */ + /* take CAP_PIN refs for r_inode, r_parent, r_old_dentry */ if (req->r_inode) ceph_get_cap_refs(ceph_inode(req->r_inode), CEPH_CAP_PIN); - if (req->r_locked_dir) - ceph_get_cap_refs(ceph_inode(req->r_locked_dir), CEPH_CAP_PIN); + if (req->r_parent) + ceph_get_cap_refs(ceph_inode(req->r_parent), CEPH_CAP_PIN); if (req->r_old_dentry_dir) ceph_get_cap_refs(ceph_inode(req->r_old_dentry_dir), CEPH_CAP_PIN); @@ -2336,7 +2336,7 @@ int ceph_mdsc_do_request(struct ceph_mds_client *mdsc, set_bit(CEPH_MDS_R_ABORTED, &req->r_req_flags); mutex_unlock(&req->r_fill_mutex); - if (req->r_locked_dir && + if (req->r_parent && (req->r_op & CEPH_MDS_OP_WRITE)) ceph_invalidate_dir_request(req); } else { @@ -2355,7 +2355,7 @@ out: */ void ceph_invalidate_dir_request(struct ceph_mds_request *req) { - struct inode *inode = req->r_locked_dir; + struct inode *inode = req->r_parent; dout("invalidate_dir_request %p (complete, lease(s))\n", inode); diff --git a/fs/ceph/mds_client.h b/fs/ceph/mds_client.h index 409b0e3c3b7a..ac0475a2daa7 100644 --- a/fs/ceph/mds_client.h +++ b/fs/ceph/mds_client.h @@ -202,7 +202,7 @@ struct ceph_mds_request { char *r_path1, *r_path2; struct ceph_vino r_ino1, r_ino2; - struct inode *r_locked_dir; /* dir (if any) i_mutex locked by vfs */ + struct inode *r_parent; /* parent dir inode */ struct inode *r_target_inode; /* resulting inode */ #define CEPH_MDS_R_DIRECT_IS_HASH (1) /* r_direct_hash is valid */ @@ -211,6 +211,7 @@ struct ceph_mds_request { #define CEPH_MDS_R_GOT_SAFE (4) /* got a safe reply */ #define CEPH_MDS_R_GOT_RESULT (5) /* got a result */ #define CEPH_MDS_R_DID_PREPOPULATE (6) /* prepopulated readdir */ +#define CEPH_MDS_R_PARENT_LOCKED (7) /* is r_parent->i_rwsem wlocked? */ unsigned long r_req_flags; struct mutex r_fill_mutex; -- cgit v1.2.3-59-g8ed1b From 80d025ffede88969f6adf7266fbdedfd5641148a Mon Sep 17 00:00:00 2001 From: Jeff Layton Date: Thu, 26 Jan 2017 16:14:18 -0500 Subject: ceph: don't update_dentry_lease unless we actually got one This if block updates the dentry lease even in the case where the MDS didn't grant one. Signed-off-by: Jeff Layton Reviewed-by: Yan, Zheng Signed-off-by: Ilya Dryomov --- fs/ceph/inode.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'fs/ceph') diff --git a/fs/ceph/inode.c b/fs/ceph/inode.c index 35a8c453bea6..542e3c6b0f1b 100644 --- a/fs/ceph/inode.c +++ b/fs/ceph/inode.c @@ -1286,8 +1286,8 @@ retry_lookup: ceph_dir_clear_ordered(dir); dout("d_delete %p\n", dn); d_delete(dn); - } else { - if (have_lease && d_unhashed(dn)) + } else if (have_lease) { + if (d_unhashed(dn)) d_add(dn, NULL); update_dentry_lease(dn, rinfo->dlease, session, -- cgit v1.2.3-59-g8ed1b From f5d55f03973f79a008bf8635452849f8ef589c23 Mon Sep 17 00:00:00 2001 From: Jeff Layton Date: Fri, 27 Jan 2017 09:13:57 -0500 Subject: ceph: vet the target and parent inodes before updating dentry lease In a later patch, we're going to need to allow ceph_fill_trace to update the dentry's lease when the parent is not locked. This is potentially racy though -- by the time we get around to processing the trace, the parent may have already changed. Change update_dentry_lease to take a ceph_vino pointer and use that to ensure that the dentry's parent still matches it before updating the lease. Signed-off-by: Jeff Layton Reviewed-by: Yan, Zheng Signed-off-by: Ilya Dryomov --- fs/ceph/inode.c | 72 ++++++++++++++++++++++++++++++++++++++------------------- 1 file changed, 48 insertions(+), 24 deletions(-) (limited to 'fs/ceph') diff --git a/fs/ceph/inode.c b/fs/ceph/inode.c index 542e3c6b0f1b..f5111df561e4 100644 --- a/fs/ceph/inode.c +++ b/fs/ceph/inode.c @@ -1016,7 +1016,9 @@ out: static void update_dentry_lease(struct dentry *dentry, struct ceph_mds_reply_lease *lease, struct ceph_mds_session *session, - unsigned long from_time) + unsigned long from_time, + struct ceph_vino *tgt_vino, + struct ceph_vino *dir_vino) { struct ceph_dentry_info *di = ceph_dentry(dentry); long unsigned duration = le32_to_cpu(lease->duration_ms); @@ -1024,13 +1026,27 @@ static void update_dentry_lease(struct dentry *dentry, long unsigned half_ttl = from_time + (duration * HZ / 2) / 1000; struct inode *dir; + /* + * Make sure dentry's inode matches tgt_vino. NULL tgt_vino means that + * we expect a negative dentry. + */ + if (!tgt_vino && d_really_is_positive(dentry)) + return; + + if (tgt_vino && (d_really_is_negative(dentry) || + !ceph_ino_compare(d_inode(dentry), tgt_vino))) + return; + spin_lock(&dentry->d_lock); dout("update_dentry_lease %p duration %lu ms ttl %lu\n", dentry, duration, ttl); - /* make lease_rdcache_gen match directory */ dir = d_inode(dentry->d_parent); + /* make sure parent matches dir_vino */ + if (!ceph_ino_compare(dir, dir_vino)) + goto out_unlock; + /* only track leases on regular dentries */ if (ceph_snap(dir) != CEPH_NOSNAP) goto out_unlock; @@ -1113,7 +1129,7 @@ int ceph_fill_trace(struct super_block *sb, struct ceph_mds_request *req) struct ceph_mds_session *session = req->r_session; struct ceph_mds_reply_info_parsed *rinfo = &req->r_reply_info; struct inode *in = NULL; - struct ceph_vino vino; + struct ceph_vino tvino, dvino; struct ceph_fs_client *fsc = ceph_sb_to_client(sb); int err = 0; @@ -1154,8 +1170,8 @@ int ceph_fill_trace(struct super_block *sb, struct ceph_mds_request *req) dname.name = rinfo->dname; dname.len = rinfo->dname_len; dname.hash = full_name_hash(parent, dname.name, dname.len); - vino.ino = le64_to_cpu(rinfo->targeti.in->ino); - vino.snap = le64_to_cpu(rinfo->targeti.in->snapid); + tvino.ino = le64_to_cpu(rinfo->targeti.in->ino); + tvino.snap = le64_to_cpu(rinfo->targeti.in->snapid); retry_lookup: dn = d_lookup(parent, &dname); dout("d_lookup on parent=%p name=%.*s got %p\n", @@ -1172,8 +1188,8 @@ retry_lookup: } err = 0; } else if (d_really_is_positive(dn) && - (ceph_ino(d_inode(dn)) != vino.ino || - ceph_snap(d_inode(dn)) != vino.snap)) { + (ceph_ino(d_inode(dn)) != tvino.ino || + ceph_snap(d_inode(dn)) != tvino.snap)) { dout(" dn %p points to wrong inode %p\n", dn, d_inode(dn)); d_delete(dn); @@ -1187,10 +1203,10 @@ retry_lookup: } if (rinfo->head->is_target) { - vino.ino = le64_to_cpu(rinfo->targeti.in->ino); - vino.snap = le64_to_cpu(rinfo->targeti.in->snapid); + tvino.ino = le64_to_cpu(rinfo->targeti.in->ino); + tvino.snap = le64_to_cpu(rinfo->targeti.in->snapid); - in = ceph_get_inode(sb, vino); + in = ceph_get_inode(sb, tvino); if (IS_ERR(in)) { err = PTR_ERR(in); goto done; @@ -1231,10 +1247,12 @@ retry_lookup: BUG_ON(!dn); BUG_ON(!dir); BUG_ON(d_inode(dn->d_parent) != dir); - BUG_ON(ceph_ino(dir) != - le64_to_cpu(rinfo->diri.in->ino)); - BUG_ON(ceph_snap(dir) != - le64_to_cpu(rinfo->diri.in->snapid)); + + dvino.ino = le64_to_cpu(rinfo->diri.in->ino); + dvino.snap = le64_to_cpu(rinfo->diri.in->snapid); + + BUG_ON(ceph_ino(dir) != dvino.ino); + BUG_ON(ceph_snap(dir) != dvino.snap); /* do we have a lease on the whole dir? */ have_dir_cap = @@ -1291,7 +1309,8 @@ retry_lookup: d_add(dn, NULL); update_dentry_lease(dn, rinfo->dlease, session, - req->r_request_started); + req->r_request_started, + NULL, &dvino); } goto done; } @@ -1314,9 +1333,13 @@ retry_lookup: have_lease = false; } - if (have_lease) + if (have_lease) { + tvino.ino = le64_to_cpu(rinfo->targeti.in->ino); + tvino.snap = le64_to_cpu(rinfo->targeti.in->snapid); update_dentry_lease(dn, rinfo->dlease, session, - req->r_request_started); + req->r_request_started, + &tvino, &dvino); + } dout(" final dn %p\n", dn); } else if ((req->r_op == CEPH_MDS_OP_LOOKUPSNAP || req->r_op == CEPH_MDS_OP_MKSNAP) && @@ -1490,14 +1513,14 @@ int ceph_readdir_prepopulate(struct ceph_mds_request *req, /* FIXME: release caps/leases if error occurs */ for (i = 0; i < rinfo->dir_nr; i++) { struct ceph_mds_reply_dir_entry *rde = rinfo->dir_entries + i; - struct ceph_vino vino; + struct ceph_vino tvino, dvino; dname.name = rde->name; dname.len = rde->name_len; dname.hash = full_name_hash(parent, dname.name, dname.len); - vino.ino = le64_to_cpu(rde->inode.in->ino); - vino.snap = le64_to_cpu(rde->inode.in->snapid); + tvino.ino = le64_to_cpu(rde->inode.in->ino); + tvino.snap = le64_to_cpu(rde->inode.in->snapid); if (rinfo->hash_order) { u32 hash = ceph_str_hash(ci->i_dir_layout.dl_dir_hash, @@ -1526,8 +1549,8 @@ retry_lookup: goto out; } } else if (d_really_is_positive(dn) && - (ceph_ino(d_inode(dn)) != vino.ino || - ceph_snap(d_inode(dn)) != vino.snap)) { + (ceph_ino(d_inode(dn)) != tvino.ino || + ceph_snap(d_inode(dn)) != tvino.snap)) { dout(" dn %p points to wrong inode %p\n", dn, d_inode(dn)); d_delete(dn); @@ -1539,7 +1562,7 @@ retry_lookup: if (d_really_is_positive(dn)) { in = d_inode(dn); } else { - in = ceph_get_inode(parent->d_sb, vino); + in = ceph_get_inode(parent->d_sb, tvino); if (IS_ERR(in)) { dout("new_inode badness\n"); d_drop(dn); @@ -1584,8 +1607,9 @@ retry_lookup: ceph_dentry(dn)->offset = rde->offset; + dvino = ceph_vino(d_inode(parent)); update_dentry_lease(dn, rde->lease, req->r_session, - req->r_request_started); + req->r_request_started, &tvino, &dvino); if (err == 0 && skipped == 0 && cache_ctl.index >= 0) { ret = fill_readdir_cache(d_inode(parent), dn, -- cgit v1.2.3-59-g8ed1b From cdde7c43513de7d8baec3ad7f6a01a177e00e968 Mon Sep 17 00:00:00 2001 From: Jeff Layton Date: Fri, 27 Jan 2017 13:07:10 -0500 Subject: ceph: call update_dentry_lease even when r_locked dir is not set We don't really require that the parent be locked in order to update the lease on a dentry. Lease info is protected by the d_lock. In the event that the parent is not locked in ceph_fill_trace, and we have both parent and target info, go ahead and update the dentry lease. Signed-off-by: Jeff Layton Reviewed-by: Yan, Zheng Signed-off-by: Ilya Dryomov --- fs/ceph/inode.c | 20 ++++++++++++++++++++ 1 file changed, 20 insertions(+) (limited to 'fs/ceph') diff --git a/fs/ceph/inode.c b/fs/ceph/inode.c index f5111df561e4..68f46132b157 100644 --- a/fs/ceph/inode.c +++ b/fs/ceph/inode.c @@ -1360,6 +1360,26 @@ retry_lookup: goto done; } req->r_dentry = dn; /* may have spliced */ + } else if (rinfo->head->is_dentry) { + struct ceph_vino *ptvino = NULL; + + if ((le32_to_cpu(rinfo->diri.in->cap.caps) & CEPH_CAP_FILE_SHARED) || + le32_to_cpu(rinfo->dlease->duration_ms)) { + dvino.ino = le64_to_cpu(rinfo->diri.in->ino); + dvino.snap = le64_to_cpu(rinfo->diri.in->snapid); + + if (rinfo->head->is_target) { + tvino.ino = le64_to_cpu(rinfo->targeti.in->ino); + tvino.snap = le64_to_cpu(rinfo->targeti.in->snapid); + ptvino = &tvino; + } + + update_dentry_lease(req->r_dentry, rinfo->dlease, + session, req->r_request_started, ptvino, + &dvino); + } else { + dout("%s: no dentry lease or dir cap\n", __func__); + } } done: dout("fill_trace done err=%d\n", err); -- cgit v1.2.3-59-g8ed1b From 5eb9f6040f3854c5b92c60fd211dc0235ef25ca8 Mon Sep 17 00:00:00 2001 From: Jeff Layton Date: Mon, 30 Jan 2017 09:47:25 -0500 Subject: ceph: do a LOOKUP in d_revalidate instead of GETATTR In commit c3f4688a08f (ceph: don't set req->r_locked_dir in ceph_d_revalidate), we changed the code to do a GETATTR instead of a LOOKUP as the parent info isn't strictly necessary to revalidate the dentry. What we missed there though is that in order to update the lease on the dentry after revalidating it, we _do_ need parent info. Change ceph_d_revalidate back to doing a LOOKUP instead of a GETATTR so that we can get the parent info in order to update the lease from ceph_fill_trace. Note that we set req->r_parent here, but we cannot set the CEPH_MDS_R_PARENT_LOCKED flag as we can't guarantee that it is. Signed-off-by: Jeff Layton Reviewed-by: Yan, Zheng Signed-off-by: Ilya Dryomov --- fs/ceph/dir.c | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) (limited to 'fs/ceph') diff --git a/fs/ceph/dir.c b/fs/ceph/dir.c index c8562b8a8c7e..3e9ad501addf 100644 --- a/fs/ceph/dir.c +++ b/fs/ceph/dir.c @@ -1244,11 +1244,12 @@ static int ceph_d_revalidate(struct dentry *dentry, unsigned int flags) return -ECHILD; op = ceph_snap(dir) == CEPH_SNAPDIR ? - CEPH_MDS_OP_LOOKUPSNAP : CEPH_MDS_OP_GETATTR; + CEPH_MDS_OP_LOOKUPSNAP : CEPH_MDS_OP_LOOKUP; req = ceph_mdsc_create_request(mdsc, op, USE_ANY_MDS); if (!IS_ERR(req)) { req->r_dentry = dget(dentry); - req->r_num_caps = op == CEPH_MDS_OP_GETATTR ? 1 : 2; + req->r_num_caps = 2; + req->r_parent = dir; mask = CEPH_STAT_CAP_INODE | CEPH_CAP_AUTH_SHARED; if (ceph_security_xattr_wanted(dir)) -- cgit v1.2.3-59-g8ed1b From df963ea8a082d31521a120e8e31a29ad8a1dc215 Mon Sep 17 00:00:00 2001 From: Jeff Layton Date: Tue, 14 Feb 2017 10:09:40 -0500 Subject: ceph: remove req from unsafe list when unregistering it There's no reason a request should ever be on a s_unsafe list but not in the request tree. Cc: stable@vger.kernel.org Link: http://tracker.ceph.com/issues/18474 Signed-off-by: Jeff Layton Reviewed-by: Yan, Zheng Signed-off-by: Ilya Dryomov --- fs/ceph/mds_client.c | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) (limited to 'fs/ceph') diff --git a/fs/ceph/mds_client.c b/fs/ceph/mds_client.c index 52521f339745..fdbc3544e41c 100644 --- a/fs/ceph/mds_client.c +++ b/fs/ceph/mds_client.c @@ -628,6 +628,9 @@ static void __unregister_request(struct ceph_mds_client *mdsc, { dout("__unregister_request %p tid %lld\n", req, req->r_tid); + /* Never leave an unregistered request on an unsafe list! */ + list_del_init(&req->r_unsafe_item); + if (req->r_tid == mdsc->oldest_tid) { struct rb_node *p = rb_next(&req->r_node); mdsc->oldest_tid = 0; @@ -1058,7 +1061,6 @@ static void cleanup_session_requests(struct ceph_mds_client *mdsc, while (!list_empty(&session->s_unsafe)) { req = list_first_entry(&session->s_unsafe, struct ceph_mds_request, r_unsafe_item); - list_del_init(&req->r_unsafe_item); pr_warn_ratelimited(" dropping unsafe request %llu\n", req->r_tid); __unregister_request(mdsc, req); @@ -2469,7 +2471,6 @@ static void handle_reply(struct ceph_mds_session *session, struct ceph_msg *msg) * useful we could do with a revised return value. */ dout("got safe reply %llu, mds%d\n", tid, mds); - list_del_init(&req->r_unsafe_item); /* last unsafe request during umount? */ if (mdsc->stopping && !__get_oldest_req(mdsc)) -- cgit v1.2.3-59-g8ed1b From f107548039807eb890e65ce5cd29d6ac52562f09 Mon Sep 17 00:00:00 2001 From: Dan Carpenter Date: Thu, 23 Feb 2017 13:39:59 +0300 Subject: ceph: tidy some white space in get_nonsnap_parent() The white space here seems slightly messed up. Signed-off-by: Dan Carpenter Signed-off-by: Ilya Dryomov --- fs/ceph/mds_client.c | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) (limited to 'fs/ceph') diff --git a/fs/ceph/mds_client.c b/fs/ceph/mds_client.c index fdbc3544e41c..c681762d76e6 100644 --- a/fs/ceph/mds_client.c +++ b/fs/ceph/mds_client.c @@ -679,8 +679,9 @@ static void __unregister_request(struct ceph_mds_client *mdsc, * working with them. Once we hit a candidate dentry, we attempt to take a * reference to it, and return that as the result. */ -static struct inode *get_nonsnap_parent(struct dentry *dentry) { struct inode - *inode = NULL; +static struct inode *get_nonsnap_parent(struct dentry *dentry) +{ + struct inode *inode = NULL; while (dentry && !IS_ROOT(dentry)) { inode = d_inode_rcu(dentry); -- cgit v1.2.3-59-g8ed1b From 55f2a04588c5881d90e22631b17a84cd25d17cc4 Mon Sep 17 00:00:00 2001 From: Ilya Dryomov Date: Mon, 13 Feb 2017 14:44:19 +0100 Subject: ceph: remove special ack vs commit behavior - ask for a commit reply instead of an ack reply in __ceph_pool_perm_get() - don't ask for both ack and commit replies in ceph_sync_write() - since just only one reply is requested now, i_unsafe_writes list will always be empty -- kill ceph_sync_write_wait() and go back to a standard ->evict_inode() Signed-off-by: Ilya Dryomov Reviewed-by: Jeff Layton Reviewed-by: Sage Weil --- fs/ceph/addr.c | 2 +- fs/ceph/caps.c | 2 -- fs/ceph/file.c | 88 +-------------------------------------------------------- fs/ceph/inode.c | 9 ------ fs/ceph/super.c | 1 - fs/ceph/super.h | 4 +-- 6 files changed, 3 insertions(+), 103 deletions(-) (limited to 'fs/ceph') diff --git a/fs/ceph/addr.c b/fs/ceph/addr.c index 4547bbf80e4f..3f0474c55f05 100644 --- a/fs/ceph/addr.c +++ b/fs/ceph/addr.c @@ -1872,7 +1872,7 @@ static int __ceph_pool_perm_get(struct ceph_inode_info *ci, goto out_unlock; } - wr_req->r_flags = CEPH_OSD_FLAG_WRITE | CEPH_OSD_FLAG_ACK; + wr_req->r_flags = CEPH_OSD_FLAG_WRITE | CEPH_OSD_FLAG_ONDISK; osd_req_op_init(wr_req, 0, CEPH_OSD_OP_CREATE, CEPH_OSD_OP_FLAG_EXCL); ceph_oloc_copy(&wr_req->r_base_oloc, &rd_req->r_base_oloc); ceph_oid_copy(&wr_req->r_base_oid, &rd_req->r_base_oid); diff --git a/fs/ceph/caps.c b/fs/ceph/caps.c index 3c2dfd72e5b2..cd966f276a8d 100644 --- a/fs/ceph/caps.c +++ b/fs/ceph/caps.c @@ -2091,8 +2091,6 @@ int ceph_fsync(struct file *file, loff_t start, loff_t end, int datasync) dout("fsync %p%s\n", inode, datasync ? " datasync" : ""); - ceph_sync_write_wait(inode); - ret = filemap_write_and_wait_range(inode->i_mapping, start, end); if (ret < 0) goto out; diff --git a/fs/ceph/file.c b/fs/ceph/file.c index baad3b688ada..023cb7fd9b5f 100644 --- a/fs/ceph/file.c +++ b/fs/ceph/file.c @@ -795,89 +795,6 @@ out: kfree(aio_work); } -/* - * Write commit request unsafe callback, called to tell us when a - * request is unsafe (that is, in flight--has been handed to the - * messenger to send to its target osd). It is called again when - * we've received a response message indicating the request is - * "safe" (its CEPH_OSD_FLAG_ONDISK flag is set), or when a request - * is completed early (and unsuccessfully) due to a timeout or - * interrupt. - * - * This is used if we requested both an ACK and ONDISK commit reply - * from the OSD. - */ -static void ceph_sync_write_unsafe(struct ceph_osd_request *req, bool unsafe) -{ - struct ceph_inode_info *ci = ceph_inode(req->r_inode); - - dout("%s %p tid %llu %ssafe\n", __func__, req, req->r_tid, - unsafe ? "un" : ""); - if (unsafe) { - ceph_get_cap_refs(ci, CEPH_CAP_FILE_WR); - spin_lock(&ci->i_unsafe_lock); - list_add_tail(&req->r_unsafe_item, - &ci->i_unsafe_writes); - spin_unlock(&ci->i_unsafe_lock); - - complete_all(&req->r_completion); - } else { - spin_lock(&ci->i_unsafe_lock); - list_del_init(&req->r_unsafe_item); - spin_unlock(&ci->i_unsafe_lock); - ceph_put_cap_refs(ci, CEPH_CAP_FILE_WR); - } -} - -/* - * Wait on any unsafe replies for the given inode. First wait on the - * newest request, and make that the upper bound. Then, if there are - * more requests, keep waiting on the oldest as long as it is still older - * than the original request. - */ -void ceph_sync_write_wait(struct inode *inode) -{ - struct ceph_inode_info *ci = ceph_inode(inode); - struct list_head *head = &ci->i_unsafe_writes; - struct ceph_osd_request *req; - u64 last_tid; - - if (!S_ISREG(inode->i_mode)) - return; - - spin_lock(&ci->i_unsafe_lock); - if (list_empty(head)) - goto out; - - /* set upper bound as _last_ entry in chain */ - - req = list_last_entry(head, struct ceph_osd_request, - r_unsafe_item); - last_tid = req->r_tid; - - do { - ceph_osdc_get_request(req); - spin_unlock(&ci->i_unsafe_lock); - - dout("sync_write_wait on tid %llu (until %llu)\n", - req->r_tid, last_tid); - wait_for_completion(&req->r_done_completion); - ceph_osdc_put_request(req); - - spin_lock(&ci->i_unsafe_lock); - /* - * from here on look at first entry in chain, since we - * only want to wait for anything older than last_tid - */ - if (list_empty(head)) - break; - req = list_first_entry(head, struct ceph_osd_request, - r_unsafe_item); - } while (req->r_tid < last_tid); -out: - spin_unlock(&ci->i_unsafe_lock); -} - static ssize_t ceph_direct_read_write(struct kiocb *iocb, struct iov_iter *iter, struct ceph_snap_context *snapc, @@ -1119,8 +1036,7 @@ ceph_sync_write(struct kiocb *iocb, struct iov_iter *from, loff_t pos, flags = CEPH_OSD_FLAG_ORDERSNAP | CEPH_OSD_FLAG_ONDISK | - CEPH_OSD_FLAG_WRITE | - CEPH_OSD_FLAG_ACK; + CEPH_OSD_FLAG_WRITE; while ((len = iov_iter_count(from)) > 0) { size_t left; @@ -1166,8 +1082,6 @@ ceph_sync_write(struct kiocb *iocb, struct iov_iter *from, loff_t pos, goto out; } - /* get a second commit callback */ - req->r_unsafe_callback = ceph_sync_write_unsafe; req->r_inode = inode; osd_req_op_extent_osd_data_pages(req, 0, pages, len, 0, diff --git a/fs/ceph/inode.c b/fs/ceph/inode.c index 68f46132b157..fd8f771f99b7 100644 --- a/fs/ceph/inode.c +++ b/fs/ceph/inode.c @@ -499,7 +499,6 @@ struct inode *ceph_alloc_inode(struct super_block *sb) ci->i_rdcache_gen = 0; ci->i_rdcache_revoking = 0; - INIT_LIST_HEAD(&ci->i_unsafe_writes); INIT_LIST_HEAD(&ci->i_unsafe_dirops); INIT_LIST_HEAD(&ci->i_unsafe_iops); spin_lock_init(&ci->i_unsafe_lock); @@ -583,14 +582,6 @@ int ceph_drop_inode(struct inode *inode) return 1; } -void ceph_evict_inode(struct inode *inode) -{ - /* wait unsafe sync writes */ - ceph_sync_write_wait(inode); - truncate_inode_pages_final(&inode->i_data); - clear_inode(inode); -} - static inline blkcnt_t calc_inode_blocks(u64 size) { return (size + (1<<9) - 1) >> 9; diff --git a/fs/ceph/super.c b/fs/ceph/super.c index a0a0b6d02f89..0ec8d0114e57 100644 --- a/fs/ceph/super.c +++ b/fs/ceph/super.c @@ -757,7 +757,6 @@ static const struct super_operations ceph_super_ops = { .destroy_inode = ceph_destroy_inode, .write_inode = ceph_write_inode, .drop_inode = ceph_drop_inode, - .evict_inode = ceph_evict_inode, .sync_fs = ceph_sync_fs, .put_super = ceph_put_super, .show_options = ceph_show_options, diff --git a/fs/ceph/super.h b/fs/ceph/super.h index 950170136be9..e9410bcf4113 100644 --- a/fs/ceph/super.h +++ b/fs/ceph/super.h @@ -343,7 +343,6 @@ struct ceph_inode_info { u32 i_rdcache_gen; /* incremented each time we get FILE_CACHE. */ u32 i_rdcache_revoking; /* RDCACHE gen to async invalidate, if any */ - struct list_head i_unsafe_writes; /* uncommitted sync writes */ struct list_head i_unsafe_dirops; /* uncommitted mds dir ops */ struct list_head i_unsafe_iops; /* uncommitted mds inode ops */ spinlock_t i_unsafe_lock; @@ -753,7 +752,6 @@ extern const struct inode_operations ceph_file_iops; extern struct inode *ceph_alloc_inode(struct super_block *sb); extern void ceph_destroy_inode(struct inode *inode); extern int ceph_drop_inode(struct inode *inode); -extern void ceph_evict_inode(struct inode *inode); extern struct inode *ceph_get_inode(struct super_block *sb, struct ceph_vino vino); @@ -933,7 +931,7 @@ extern int ceph_atomic_open(struct inode *dir, struct dentry *dentry, extern int ceph_release(struct inode *inode, struct file *filp); extern void ceph_fill_inline_data(struct inode *inode, struct page *locked_page, char *data, size_t len); -extern void ceph_sync_write_wait(struct inode *inode); + /* dir.c */ extern const struct file_operations ceph_dir_fops; extern const struct file_operations ceph_snapdir_fops; -- cgit v1.2.3-59-g8ed1b From 54ea0046b6fe36ec18e82d282a29a18da6cdea0f Mon Sep 17 00:00:00 2001 From: Ilya Dryomov Date: Sat, 11 Feb 2017 18:48:41 +0100 Subject: libceph, rbd, ceph: WRITE | ONDISK -> WRITE CEPH_OSD_FLAG_ONDISK is set in account_request(). Signed-off-by: Ilya Dryomov Reviewed-by: Jeff Layton Reviewed-by: Sage Weil --- drivers/block/rbd.c | 6 ++---- fs/ceph/addr.c | 14 +++++--------- fs/ceph/file.c | 15 ++++----------- net/ceph/cls_lock_client.c | 12 ++++++------ net/ceph/osd_client.c | 9 ++++----- 5 files changed, 21 insertions(+), 35 deletions(-) (limited to 'fs/ceph') diff --git a/drivers/block/rbd.c b/drivers/block/rbd.c index ffb8d3afc6b4..695ef552aaf0 100644 --- a/drivers/block/rbd.c +++ b/drivers/block/rbd.c @@ -1981,8 +1981,7 @@ static struct ceph_osd_request *rbd_osd_req_create( return __rbd_osd_req_create(rbd_dev, snapc, num_ops, (op_type == OBJ_OP_WRITE || op_type == OBJ_OP_DISCARD) ? - CEPH_OSD_FLAG_WRITE | CEPH_OSD_FLAG_ONDISK : CEPH_OSD_FLAG_READ, - obj_request); + CEPH_OSD_FLAG_WRITE : CEPH_OSD_FLAG_READ, obj_request); } /* @@ -2008,8 +2007,7 @@ rbd_osd_req_create_copyup(struct rbd_obj_request *obj_request) return __rbd_osd_req_create(img_request->rbd_dev, img_request->snapc, num_osd_ops, - CEPH_OSD_FLAG_WRITE | CEPH_OSD_FLAG_ONDISK, - obj_request); + CEPH_OSD_FLAG_WRITE, obj_request); } static void rbd_osd_req_destroy(struct ceph_osd_request *osd_req) diff --git a/fs/ceph/addr.c b/fs/ceph/addr.c index 3f0474c55f05..6ecb920602ed 100644 --- a/fs/ceph/addr.c +++ b/fs/ceph/addr.c @@ -1018,8 +1018,7 @@ new_request: &ci->i_layout, vino, offset, &len, 0, num_ops, CEPH_OSD_OP_WRITE, - CEPH_OSD_FLAG_WRITE | - CEPH_OSD_FLAG_ONDISK, + CEPH_OSD_FLAG_WRITE, snapc, truncate_seq, truncate_size, false); if (IS_ERR(req)) { @@ -1029,8 +1028,7 @@ new_request: min(num_ops, CEPH_OSD_SLAB_OPS), CEPH_OSD_OP_WRITE, - CEPH_OSD_FLAG_WRITE | - CEPH_OSD_FLAG_ONDISK, + CEPH_OSD_FLAG_WRITE, snapc, truncate_seq, truncate_size, true); BUG_ON(IS_ERR(req)); @@ -1680,8 +1678,7 @@ int ceph_uninline_data(struct file *filp, struct page *locked_page) req = ceph_osdc_new_request(&fsc->client->osdc, &ci->i_layout, ceph_vino(inode), 0, &len, 0, 1, - CEPH_OSD_OP_CREATE, - CEPH_OSD_FLAG_ONDISK | CEPH_OSD_FLAG_WRITE, + CEPH_OSD_OP_CREATE, CEPH_OSD_FLAG_WRITE, NULL, 0, 0, false); if (IS_ERR(req)) { err = PTR_ERR(req); @@ -1698,8 +1695,7 @@ int ceph_uninline_data(struct file *filp, struct page *locked_page) req = ceph_osdc_new_request(&fsc->client->osdc, &ci->i_layout, ceph_vino(inode), 0, &len, 1, 3, - CEPH_OSD_OP_WRITE, - CEPH_OSD_FLAG_ONDISK | CEPH_OSD_FLAG_WRITE, + CEPH_OSD_OP_WRITE, CEPH_OSD_FLAG_WRITE, NULL, ci->i_truncate_seq, ci->i_truncate_size, false); if (IS_ERR(req)) { @@ -1872,7 +1868,7 @@ static int __ceph_pool_perm_get(struct ceph_inode_info *ci, goto out_unlock; } - wr_req->r_flags = CEPH_OSD_FLAG_WRITE | CEPH_OSD_FLAG_ONDISK; + wr_req->r_flags = CEPH_OSD_FLAG_WRITE; osd_req_op_init(wr_req, 0, CEPH_OSD_OP_CREATE, CEPH_OSD_OP_FLAG_EXCL); ceph_oloc_copy(&wr_req->r_base_oloc, &rd_req->r_base_oloc); ceph_oid_copy(&wr_req->r_base_oid, &rd_req->r_base_oid); diff --git a/fs/ceph/file.c b/fs/ceph/file.c index 023cb7fd9b5f..26cc95421cca 100644 --- a/fs/ceph/file.c +++ b/fs/ceph/file.c @@ -759,9 +759,7 @@ static void ceph_aio_retry_work(struct work_struct *work) goto out; } - req->r_flags = CEPH_OSD_FLAG_ORDERSNAP | - CEPH_OSD_FLAG_ONDISK | - CEPH_OSD_FLAG_WRITE; + req->r_flags = CEPH_OSD_FLAG_ORDERSNAP | CEPH_OSD_FLAG_WRITE; ceph_oloc_copy(&req->r_base_oloc, &orig_req->r_base_oloc); ceph_oid_copy(&req->r_base_oid, &orig_req->r_base_oid); @@ -833,9 +831,7 @@ ceph_direct_read_write(struct kiocb *iocb, struct iov_iter *iter, if (ret2 < 0) dout("invalidate_inode_pages2_range returned %d\n", ret2); - flags = CEPH_OSD_FLAG_ORDERSNAP | - CEPH_OSD_FLAG_ONDISK | - CEPH_OSD_FLAG_WRITE; + flags = CEPH_OSD_FLAG_ORDERSNAP | CEPH_OSD_FLAG_WRITE; } else { flags = CEPH_OSD_FLAG_READ; } @@ -1034,9 +1030,7 @@ ceph_sync_write(struct kiocb *iocb, struct iov_iter *from, loff_t pos, if (ret < 0) dout("invalidate_inode_pages2_range returned %d\n", ret); - flags = CEPH_OSD_FLAG_ORDERSNAP | - CEPH_OSD_FLAG_ONDISK | - CEPH_OSD_FLAG_WRITE; + flags = CEPH_OSD_FLAG_ORDERSNAP | CEPH_OSD_FLAG_WRITE; while ((len = iov_iter_count(from)) > 0) { size_t left; @@ -1531,8 +1525,7 @@ static int ceph_zero_partial_object(struct inode *inode, ceph_vino(inode), offset, length, 0, 1, op, - CEPH_OSD_FLAG_WRITE | - CEPH_OSD_FLAG_ONDISK, + CEPH_OSD_FLAG_WRITE, NULL, 0, 0, false); if (IS_ERR(req)) { ret = PTR_ERR(req); diff --git a/net/ceph/cls_lock_client.c b/net/ceph/cls_lock_client.c index f13a1ea87459..b9233b990399 100644 --- a/net/ceph/cls_lock_client.c +++ b/net/ceph/cls_lock_client.c @@ -69,8 +69,8 @@ int ceph_cls_lock(struct ceph_osd_client *osdc, dout("%s lock_name %s type %d cookie %s tag %s desc %s flags 0x%x\n", __func__, lock_name, type, cookie, tag, desc, flags); ret = ceph_osdc_call(osdc, oid, oloc, "lock", "lock", - CEPH_OSD_FLAG_WRITE | CEPH_OSD_FLAG_ONDISK, - lock_op_page, lock_op_buf_size, NULL, NULL); + CEPH_OSD_FLAG_WRITE, lock_op_page, + lock_op_buf_size, NULL, NULL); dout("%s: status %d\n", __func__, ret); __free_page(lock_op_page); @@ -117,8 +117,8 @@ int ceph_cls_unlock(struct ceph_osd_client *osdc, dout("%s lock_name %s cookie %s\n", __func__, lock_name, cookie); ret = ceph_osdc_call(osdc, oid, oloc, "lock", "unlock", - CEPH_OSD_FLAG_WRITE | CEPH_OSD_FLAG_ONDISK, - unlock_op_page, unlock_op_buf_size, NULL, NULL); + CEPH_OSD_FLAG_WRITE, unlock_op_page, + unlock_op_buf_size, NULL, NULL); dout("%s: status %d\n", __func__, ret); __free_page(unlock_op_page); @@ -170,8 +170,8 @@ int ceph_cls_break_lock(struct ceph_osd_client *osdc, dout("%s lock_name %s cookie %s locker %s%llu\n", __func__, lock_name, cookie, ENTITY_NAME(*locker)); ret = ceph_osdc_call(osdc, oid, oloc, "lock", "break_lock", - CEPH_OSD_FLAG_WRITE | CEPH_OSD_FLAG_ONDISK, - break_op_page, break_op_buf_size, NULL, NULL); + CEPH_OSD_FLAG_WRITE, break_op_page, + break_op_buf_size, NULL, NULL); dout("%s: status %d\n", __func__, ret); __free_page(break_op_page); diff --git a/net/ceph/osd_client.c b/net/ceph/osd_client.c index e1c6c2b4a295..5c0938ddddf6 100644 --- a/net/ceph/osd_client.c +++ b/net/ceph/osd_client.c @@ -1704,7 +1704,7 @@ promote: static void account_request(struct ceph_osd_request *req) { - WARN_ON(req->r_flags & CEPH_OSD_FLAG_ACK); + WARN_ON(req->r_flags & (CEPH_OSD_FLAG_ACK | CEPH_OSD_FLAG_ONDISK)); WARN_ON(!(req->r_flags & (CEPH_OSD_FLAG_READ | CEPH_OSD_FLAG_WRITE))); req->r_flags |= CEPH_OSD_FLAG_ONDISK; @@ -3539,7 +3539,7 @@ ceph_osdc_watch(struct ceph_osd_client *osdc, ceph_oid_copy(&lreq->t.base_oid, oid); ceph_oloc_copy(&lreq->t.base_oloc, oloc); - lreq->t.flags = CEPH_OSD_FLAG_WRITE | CEPH_OSD_FLAG_ONDISK; + lreq->t.flags = CEPH_OSD_FLAG_WRITE; lreq->mtime = CURRENT_TIME; lreq->reg_req = alloc_linger_request(lreq); @@ -3597,7 +3597,7 @@ int ceph_osdc_unwatch(struct ceph_osd_client *osdc, ceph_oid_copy(&req->r_base_oid, &lreq->t.base_oid); ceph_oloc_copy(&req->r_base_oloc, &lreq->t.base_oloc); - req->r_flags = CEPH_OSD_FLAG_WRITE | CEPH_OSD_FLAG_ONDISK; + req->r_flags = CEPH_OSD_FLAG_WRITE; req->r_mtime = CURRENT_TIME; osd_req_op_watch_init(req, 0, lreq->linger_id, CEPH_OSD_WATCH_OP_UNWATCH); @@ -4163,8 +4163,7 @@ int ceph_osdc_writepages(struct ceph_osd_client *osdc, struct ceph_vino vino, int page_align = off & ~PAGE_MASK; req = ceph_osdc_new_request(osdc, layout, vino, off, &len, 0, 1, - CEPH_OSD_OP_WRITE, - CEPH_OSD_FLAG_ONDISK | CEPH_OSD_FLAG_WRITE, + CEPH_OSD_OP_WRITE, CEPH_OSD_FLAG_WRITE, snapc, truncate_seq, truncate_size, true); if (IS_ERR(req)) -- cgit v1.2.3-59-g8ed1b