linux/fs/ext4
Chunguang Xu 8c80fb312d ext4: fix a possible ABBA deadlock due to busy PA
We found on older kernel (3.10) that in the scenario of insufficient
disk space, system may trigger an ABBA deadlock problem, it seems that
this problem still exists in latest kernel, try to fix it here. The
main process triggered by this problem is that task A occupies the PA
and waits for the jbd2 transaction finish, the jbd2 transaction waits
for the completion of task B's IO (plug_list), but task B waits for
the release of PA by task A to finish discard, which indirectly forms
an ABBA deadlock. The related calltrace is as follows:

    Task A
    vfs_write
    ext4_mb_new_blocks()
    ext4_mb_mark_diskspace_used()       JBD2
    jbd2_journal_get_write_access()  -> jbd2_journal_commit_transaction()
  ->schedule()                          filemap_fdatawait()
 |                                              |
 | Task B                                       |
 | do_unlinkat()                                |
 | ext4_evict_inode()                           |
 | jbd2_journal_begin_ordered_truncate()        |
 | filemap_fdatawrite_range()                   |
 | ext4_mb_new_blocks()                         |
  -ext4_mb_discard_group_preallocations() <-----

Here, try to cancel ext4_mb_discard_group_preallocations() internal
retry due to PA busy, and do a limited number of retries inside
ext4_mb_discard_preallocations(), which can circumvent the above
problems, but also has some advantages:

1. Since the PA is in a busy state, if other groups have free PAs,
   keeping the current PA may help to reduce fragmentation.
2. Continue to traverse forward instead of waiting for the current
   group PA to be released. In most scenarios, the PA discard time
   can be reduced.

However, in the case of smaller free space, if only a few groups have
space, then due to multiple traversals of the group, it may increase
CPU overhead. But in contrast, I feel that the overall benefit is
better than the cost.

Signed-off-by: Chunguang Xu <brookxu@tencent.com>
Reported-by: kernel test robot <lkp@intel.com>
Reviewed-by: Jan Kara <jack@suse.cz>
Link: https://lore.kernel.org/r/1637630277-23496-1-git-send-email-brookxu.cn@gmail.com
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Cc: stable@kernel.org
2022-01-10 13:25:55 -05:00
..
.kunitconfig ext4: add .kunitconfig fragment to enable ext4-specific tests 2021-02-11 23:16:30 -05:00
acl.c ext4: use ext4_journal_start/stop for fast commit transactions 2021-12-23 18:13:25 -05:00
acl.h vfs: add rcu argument to ->get_acl() callback 2021-08-18 22:08:24 +02:00
balloc.c ext4: flush background discard kwork when retry allocation 2021-08-30 23:35:53 -04:00
bitmap.c
block_validity.c ext4: standardize error message in ext4_protect_reserved_inode() 2020-12-17 13:30:55 -05:00
dir.c ext4: fix potential infinite loop in ext4_dx_readdir() 2021-10-01 00:05:09 -04:00
ext4_extents.h ext4: fix sparse warnings 2021-08-30 23:36:50 -04:00
ext4_jbd2.c ext4: Support for checksumming from journal triggers 2021-08-30 23:36:50 -04:00
ext4_jbd2.h ext4: Support for checksumming from journal triggers 2021-08-30 23:36:50 -04:00
ext4.h ext4: destroy ext4_fc_dentry_cachep kmemcache on module removal 2022-01-10 13:25:54 -05:00
extents_status.c ext4: correct the cache_nr in tracepoint ext4_es_shrink_exit 2021-06-22 21:34:17 -04:00
extents_status.h ext4: fix extent_status trace points 2020-01-25 02:03:03 -05:00
extents.c ext4: fix fast commit may miss tracking range for FALLOC_FL_ZERO_RANGE 2022-01-10 13:25:44 -05:00
fast_commit.c ext4: destroy ext4_fc_dentry_cachep kmemcache on module removal 2022-01-10 13:25:54 -05:00
fast_commit.h ext4: simplify updating of fast commit stats 2021-12-23 18:13:25 -05:00
file.c ext4: use ext4_journal_start/stop for fast commit transactions 2021-12-23 18:13:25 -05:00
fsmap.c treewide: Change list_sort to use const pointers 2021-04-08 16:04:22 -07:00
fsmap.h ext4: fsmap: fix the block/inode bitmap comment 2021-06-24 09:48:29 -04:00
fsync.c block: use an on-stack bio in blkdev_issue_flush 2021-01-27 09:51:48 -07:00
hash.c ext4: handle casefolding with encryption 2021-04-05 22:04:20 -04:00
ialloc.c ext4: Support for checksumming from journal triggers 2021-08-30 23:36:50 -04:00
indirect.c ext4: Support for checksumming from journal triggers 2021-08-30 23:36:50 -04:00
inline.c ext4: remove extent cache entries when truncating inline data 2021-09-09 10:52:05 -04:00
inode-test.c fs: ext4: Modify inode-test.c to use KUnit parameterized testing feature 2020-12-02 16:07:25 -07:00
inode.c ext4: fast commit may miss tracking unwritten range during ftruncate 2022-01-10 13:25:54 -05:00
ioctl.c ext4: drop ineligible txn start stop APIs 2021-12-23 18:13:25 -05:00
Kconfig ext: EXT4_KUNIT_TESTS should depend on EXT4_FS instead of selecting it 2021-02-11 23:12:59 -05:00
Makefile ext4: Move orphan inode handling into a separate file 2021-08-30 23:36:51 -04:00
mballoc.c ext4: fix a possible ABBA deadlock due to busy PA 2022-01-10 13:25:55 -05:00
mballoc.h ext4: fix various seppling typos 2021-04-09 23:14:59 -04:00
migrate.c ext4: fix various seppling typos 2021-04-09 23:14:59 -04:00
mmp.c ext4: fix potential uninitialized access to retval in kmmpd 2021-07-23 07:31:29 -04:00
move_extent.c ext4: use common helpers in all places reading metadata buffers 2020-10-18 10:37:14 -04:00
namei.c ext4: fix boolreturn.cocci warnings in fs/ext4/name.c 2021-11-04 10:33:25 -04:00
orphan.c ext4: Improve scalability of ext4 orphan file handling 2021-08-30 23:36:51 -04:00
page-io.c ext4: convert from atomic_t to refcount_t on ext4_io_end->count 2021-11-04 10:33:24 -04:00
readpage.c block: Add bio_max_segs 2021-02-26 15:49:51 -07:00
resize.c ext4: Support for checksumming from journal triggers 2021-08-30 23:36:50 -04:00
super.c ext4: make sure to reset inode lockdep class when quota enabling fails 2022-01-10 13:25:55 -05:00
symlink.c ext4: report correct st_size for encrypted symlinks 2021-07-25 20:01:06 -07:00
sysfs.c ext4: replace snprintf in show functions with sysfs_emit 2022-01-10 13:25:55 -05:00
truncate.h ext4: Convert to use mapping->invalidate_lock 2021-07-13 14:29:00 +02:00
verity.c New features for ext4 this cycle include support for encrypted 2021-04-30 15:35:30 -07:00
xattr_hurd.c acl: handle idmapped mounts 2021-01-24 14:27:17 +01:00
xattr_security.c acl: handle idmapped mounts 2021-01-24 14:27:17 +01:00
xattr_trusted.c acl: handle idmapped mounts 2021-01-24 14:27:17 +01:00
xattr_user.c acl: handle idmapped mounts 2021-01-24 14:27:17 +01:00
xattr.c ext4: Support for checksumming from journal triggers 2021-08-30 23:36:50 -04:00
xattr.h ext4: remove duplicate definition of ext4_xattr_ibody_inline_set() 2021-06-24 10:09:39 -04:00