kernel.git/fs/gfs2, branch linux-6.5.y

gfs2: don't withdraw if init_threads() got interrupted

2023-11-28T17:15:13Z

commit 0cdc6f44e9fdc2d20d720145bf99a39f611f6d61 upstream. In gfs2_fill_super(), when mounting a gfs2 filesystem is interrupted, kthread_create() can return -EINTR. When that happens, we roll back what has already been done and abort the mount. Since commit 62dd0f98a0e5 ("gfs2: Flag a withdraw if init_threads() fails), we are calling gfs2_withdraw_delayed() in gfs2_fill_super(); first via gfs2_make_fs_rw(), then directly. But gfs2_withdraw_delayed() only marks the filesystem as withdrawing and relies on a caller further up the stack to do the actual withdraw, which doesn't exist in the gfs2_fill_super() case. Because the filesystem is marked as withdrawing / withdrawn, function gfs2_lm_unmount() doesn't release the dlm lockspace, so when we try to mount that filesystem again, we get: gfs2: fsid=gohan:gohan0: Trying to join cluster "lock_dlm", "gohan:gohan0" gfs2: fsid=gohan:gohan0: dlm_new_lockspace error -17 Since commit b77b4a4815a9 ("gfs2: Rework freeze / thaw logic"), the deadlock this gfs2_withdraw_delayed() call was supposed to work around cannot occur anymore because freeze_go_callback() won't take the sb->s_umount semaphore unconditionally anymore, so we can get rid of the gfs2_withdraw_delayed() in gfs2_fill_super() entirely. Reported-by: Alexander Aring Signed-off-by: Andreas Gruenbacher Cc: stable@vger.kernel.org # v6.5+ Signed-off-by: Greg Kroah-Hartman

gfs2: Silence "suspicious RCU usage in gfs2_permission" warning

2023-11-28T17:14:53Z

[ Upstream commit 074d7306a4fe22fcac0b53f699f92757ab1cee99 ] Commit 0abd1557e21c added rcu_dereference() for dereferencing ip->i_gl in gfs2_permission. This now causes lockdep to complain when gfs2_permission is called in non-RCU context: WARNING: suspicious RCU usage in gfs2_permission Switch to rcu_dereference_check() and check for the MAY_NOT_BLOCK flag to shut up lockdep when we know that dereferencing ip->i_gl is safe. Fixes: 0abd1557e21c ("gfs2: fix an oops in gfs2_permission") Reported-by: syzbot+3e5130844b0c0e2b4948@syzkaller.appspotmail.com Signed-off-by: Andreas Gruenbacher Signed-off-by: Sasha Levin

gfs2: fix an oops in gfs2_permission

2023-11-28T17:14:51Z

[ Upstream commit 0abd1557e21c617bd13fc18f7725fc6363c05913 ] In RCU mode, we might race with gfs2_evict_inode(), which zeroes ->i_gl. Freeing of the object it points to is RCU-delayed, so if we manage to fetch the pointer before it's been replaced with NULL, we are fine. Check if we'd fetched NULL and treat that as "bail out and tell the caller to get out of RCU mode". Signed-off-by: Al Viro Signed-off-by: Andreas Gruenbacher Signed-off-by: Sasha Levin

gfs2: ignore negated quota changes

2023-11-28T17:14:51Z

[ Upstream commit 4c6a08125f2249531ec01783a5f4317d7342add5 ] When lots of quota changes are made, there may be cases in which an inode's quota information is increased and then decreased, such as when blocks are added to a file, then deleted from it. If the timing is right, function do_qc can add pending quota changes to a transaction, then later, another call to do_qc can negate those changes, resulting in a net gain of 0. The quota_change information is recorded in the qc buffer (and qd element of the inode as well). The buffer is added to the transaction by the first call to do_qc, but a subsequent call changes the value from non-zero back to zero. At that point it's too late to remove the buffer_head from the transaction. Later, when the quota sync code is called, the zero-change qd element is discovered and flagged as an assert warning. If the fs is mounted with errors=panic, the kernel will panic. This is usually seen when files are truncated and the quota changes are negated by punch_hole/truncate which uses gfs2_quota_hold and gfs2_quota_unhold rather than block allocations that use gfs2_quota_lock and gfs2_quota_unlock which automatically do quota sync. This patch solves the problem by adding a check to qd_check_sync such that net-zero quota changes already added to the transaction are no longer deemed necessary to be synced, and skipped. In this case references are taken for the qd and the slot from do_qc so those need to be put. The normal sequence of events for a normal non-zero quota change is as follows: gfs2_quota_change do_qc qd_hold slot_hold Later, when the changes are to be synced: gfs2_quota_sync qd_fish qd_check_sync gets qd ref via lockref_get_not_dead do_sync do_qc(QC_SYNC) qd_put lockref_put_or_lock qd_unlock qd_put lockref_put_or_lock In the net-zero change case, we add a check to qd_check_sync so it puts the qd and slot references acquired in gfs2_quota_change and skip the unneeded sync. Signed-off-by: Bob Peterson Signed-off-by: Andreas Gruenbacher Signed-off-by: Sasha Levin

gfs2: fix glock shrinker ref issues

2023-10-06T11:16:17Z

[ Upstream commit 62862485a4c3a52029fc30f4bdde9af04afdafc9 ] Before this patch, function gfs2_scan_glock_lru would only try to free glocks that had a reference count of 0. But if the reference count ever got to 0, the glock should have already been freed. Shrinker function gfs2_dispose_glock_lru checks whether glocks on the LRU are demote_ok, and if so, tries to demote them. But that's only possible if the reference count is at least 1. This patch changes gfs2_scan_glock_lru so it will try to demote and/or dispose of glocks that have a reference count of 1 and which are either demotable, or are already unlocked. Signed-off-by: Bob Peterson Signed-off-by: Andreas Gruenbacher Signed-off-by: Sasha Levin

gfs2: Fix another freeze/thaw hang

2023-10-06T11:15:45Z

[ Upstream commit 52954b750958dcab9e44935f0c32643279091c85 ] On a thawed filesystem, the freeze glock is held in shared mode. In order to initiate a cluster-wide freeze, the node initiating the freeze drops the freeze glock and grabs it in exclusive mode. The other nodes recognize this as contention on the freeze glock; function freeze_go_callback is invoked. This indicates to them that they must freeze the filesystem locally, drop the freeze glock, and then re-acquire it in shared mode before being able to unfreeze the filesystem locally. While a node is trying to re-acquire the freeze glock in shared mode, additional contention can occur. In that case, the node must behave in the same way as above. Unfortunately, freeze_go_callback() contains a check that causes it to bail out when the freeze glock isn't held in shared mode. Fix that to allow the glock to be unlocked or held in shared mode. In addition, update a reference to trylock_super() which has been renamed to super_trylock_shared() in the meantime. Fixes: b77b4a4815a9 ("gfs2: Rework freeze / thaw logic") Signed-off-by: Andreas Gruenbacher Signed-off-by: Sasha Levin

gfs2: low-memory forced flush fixes

2023-09-19T10:30:13Z

[ Upstream commit b74cd55aa9a9d0aca760028a51343ec79812e410 ] First, function gfs2_ail_flush_reqd checks the SDF_FORCE_AIL_FLUSH flag to determine if an AIL flush should be forced in low-memory situations. However, it also immediately clears the flag, and when called repeatedly as in function gfs2_logd, the flag will be lost. Fix that by pulling the SDF_FORCE_AIL_FLUSH flag check out of gfs2_ail_flush_reqd. Second, function gfs2_writepages sets the SDF_FORCE_AIL_FLUSH flag whether or not enough pages were written. If enough pages could be written, flushing the AIL is unnecessary, though. Third, gfs2_writepages doesn't wake up logd after setting the SDF_FORCE_AIL_FLUSH flag, so it can take a long time for logd to react. It would be preferable to wake up logd, but that hurts the performance of some workloads and we don't quite understand why so far, so don't wake up logd so far. Fixes: b066a4eebd4f ("gfs2: forcibly flush ail to relieve memory pressure") Signed-off-by: Andreas Gruenbacher Signed-off-by: Sasha Levin

gfs2: Switch to wait_event in gfs2_logd

2023-09-19T10:30:12Z

[ Upstream commit 6df373b09b1dcf2f7d579f515f653f89a896d417 ] In gfs2_logd(), switch from an open-coded wait loop to wait_event_interruptible_timeout(). Signed-off-by: Andreas Gruenbacher Stable-dep-of: b74cd55aa9a9 ("gfs2: low-memory forced flush fixes") Signed-off-by: Sasha Levin

gfs2: Don't use filemap_splice_read

2023-08-07T16:42:04Z

Starting with patch 2cb1e08985, gfs2 started using the new function filemap_splice_read rather than the old (and subsequently deleted) function generic_file_splice_read. filemap_splice_read works by taking references to a number of folios in the page cache and splicing those folios into a pipe. The folios are then read from the pipe and the folio references are dropped. This can take an arbitrary amount of time. We cannot allow that in gfs2 because those folio references will pin the inode glock to the node and prevent it from being demoted, which can lead to cluster-wide deadlocks. Instead, use copy_splice_read. (In addition, the old generic_file_splice_read called into ->read_iter, which called gfs2_file_read_iter, which took the inode glock during the operation. The new filemap_splice_read interface does not take the inode glock anymore. This is fixable, but it still wouldn't prevent cluster-wide deadlocks.) Fixes: 2cb1e08985e3 ("splice: Use filemap_splice_read() instead of generic_file_splice_read()") Signed-off-by: Bob Peterson Signed-off-by: Andreas Gruenbacher

gfs2: Fix freeze consistency check in gfs2_trans_add_meta

2023-08-07T16:40:51Z

Function gfs2_trans_add_meta() checks for the SDF_FROZEN flag to make sure that no buffers are added to a transaction while the filesystem is frozen. With the recent freeze/thaw rework, the SDF_FROZEN flag is cleared after thaw_super() is called, which is sufficient for serializing freeze/thaw. However, other filesystem operations started after thaw_super() may now be calling gfs2_trans_add_meta() before the SDF_FROZEN flag is cleared, which will trigger the SDF_FROZEN check in gfs2_trans_add_meta(). Fix that by checking the s_writers.frozen state instead. In addition, make sure not to call gfs2_assert_withdraw() with the sd_log_lock spin lock held. Check for a withdrawn filesystem before checking for a frozen filesystem, and don't pin/add buffers to the current transaction in case of a failure in either case. Signed-off-by: Andreas Gruenbacher Signed-off-by: Bob Peterson