linux-stable/fs
Darrick J. Wong 30ac846da3 xfs: log recovery should replay deferred ops in order
[ Upstream commit 509955823c ]

As part of testing log recovery with dm_log_writes, Amir Goldstein
discovered an error in the deferred ops recovery that lead to corruption
of the filesystem metadata if a reflink+rmap filesystem happened to shut
down midway through a CoW remap:

"This is what happens [after failed log recovery]:

"Phase 1 - find and verify superblock...
"Phase 2 - using internal log
"        - zero log...
"        - scan filesystem freespace and inode maps...
"        - found root inode chunk
"Phase 3 - for each AG...
"        - scan (but don't clear) agi unlinked lists...
"        - process known inodes and perform inode discovery...
"        - agno = 0
"data fork in regular inode 134 claims CoW block 376
"correcting nextents for inode 134
"bad data fork in inode 134
"would have cleared inode 134"

Hou Tao dissected the log contents of exactly such a crash:

"According to the implementation of xfs_defer_finish(), these ops should
be completed in the following sequence:

"Have been done:
"(1) CUI: Oper (160)
"(2) BUI: Oper (161)
"(3) CUD: Oper (194), for CUI Oper (160)
"(4) RUI A: Oper (197), free rmap [0x155, 2, -9]

"Should be done:
"(5) BUD: for BUI Oper (161)
"(6) RUI B: add rmap [0x155, 2, 137]
"(7) RUD: for RUI A
"(8) RUD: for RUI B

"Actually be done by xlog_recover_process_intents()
"(5) BUD: for BUI Oper (161)
"(6) RUI B: add rmap [0x155, 2, 137]
"(7) RUD: for RUI B
"(8) RUD: for RUI A

"So the rmap entry [0x155, 2, -9] for COW should be freed firstly,
then a new rmap entry [0x155, 2, 137] will be added. However, as we can see
from the log record in post_mount.log (generated after umount) and the trace
print, the new rmap entry [0x155, 2, 137] are added firstly, then the rmap
entry [0x155, 2, -9] are freed."

When reconstructing the internal log state from the log items found on
disk, it's required that deferred ops replay in exactly the same order
that they would have had the filesystem not gone down.  However,
replaying unfinished deferred ops can create /more/ deferred ops.  These
new deferred ops are finished in the wrong order.  This causes fs
corruption and replay crashes, so let's create a single defer_ops to
handle the subsequent ops created during replay, then use one single
transaction at the end of log recovery to ensure that everything is
replayed in the same order as they're supposed to be.

Reported-by: Amir Goldstein <amir73il@gmail.com>
Analyzed-by: Hou Tao <houtao1@huawei.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Tested-by: Amir Goldstein <amir73il@gmail.com>
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Signed-off-by: Sasha Levin <alexander.levin@verizon.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2018-02-03 17:39:06 +01:00
..
9p fs/9p: Compare qid.path in v9fs_test_inode 2017-11-30 08:40:49 +00:00
adfs License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
affs License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
afs afs: Connect up the CB.ProbeUuid 2017-12-14 09:53:15 +01:00
autofs4 autofs: fix careless error in recent commit 2017-12-20 10:10:17 +01:00
befs License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
bfs License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
btrfs Btrfs: fix list_add corruption and soft lockups in fsync 2018-02-03 17:39:06 +01:00
cachefiles License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
ceph ceph: drop negative child dentries before try pruning inode's alias 2017-12-20 10:10:19 +01:00
cifs cifs: fix NULL deref in SMB2_read 2017-12-20 10:10:17 +01:00
coda coda: fix 'kernel memory exposure attempt' in fsync 2017-11-24 08:37:05 +01:00
configfs
cramfs License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
crypto fscrypt: lock mutex before checking for bounce page pool 2017-11-30 08:40:44 +00:00
debugfs
devpts pty: Repair TIOCGPTPEER 2017-08-24 13:23:03 -07:00
dlm License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
ecryptfs eCryptfs: use after free in ecryptfs_release_messaging() 2017-11-30 08:40:45 +00:00
efivarfs
efs License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
exofs Merge branch 'for-4.14/block' of git://git.kernel.dk/linux-block 2017-09-07 11:59:42 -07:00
exportfs
ext2 License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
ext4 ext4: fix crash when a directory's i_size is too small 2017-12-20 10:10:22 +01:00
f2fs f2fs: fix to clear FI_NO_PREALLOC 2017-12-14 09:53:14 +01:00
fat fs/fat/inode.c: fix sb_rdonly() change 2017-12-05 11:26:29 +01:00
freevxfs
fscache License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
fuse fuse: fix READDIRPLUS skipping an entry 2017-10-25 16:34:27 +02:00
gfs2 GFS2: Take inode off order_write list when setting jdata flag 2017-12-20 10:10:29 +01:00
hfs License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
hfsplus License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
hostfs License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
hpfs License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
hugetlbfs fs/hugetlbfs/inode.c: fix hwpoison reserve accounting 2017-11-03 07:39:19 -07:00
isofs isofs: fix timestamps beyond 2027 2017-11-30 08:40:45 +00:00
jbd2
jffs2 License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
jfs License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
kernfs Merge branch 'for-4.14/block' of git://git.kernel.dk/linux-block 2017-09-07 11:59:42 -07:00
lockd lockd: lost rollback of set_grace_period() in lockd_down_net() 2017-12-05 11:26:30 +01:00
minix License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
ncpfs License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
nfs nfs: don't wait on commit in nfs_commit_inode() if there were no commit requests 2017-12-20 10:10:20 +01:00
nfs_common
nfsd nfsd: auth: Fix gid sorting when rootsquash enabled 2018-01-31 14:03:50 +01:00
nilfs2 nilfs2: fix race condition that causes file system corruption 2017-11-30 08:40:44 +00:00
nls License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
notify fsnotify: fix pinning group in fsnotify_prepare_user_wait() 2017-11-30 08:40:47 +00:00
ntfs License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
ocfs2 ocfs2: should wait dio before inode lock in ocfs2_setattr() 2017-11-24 08:37:04 +01:00
omfs License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
openpromfs
orangefs orangefs: fix deadlock; do not write i_size in read_iter 2018-01-31 14:03:43 +01:00
overlayfs locking/barriers: Convert users of lockless_dereference() to READ_ONCE() 2017-12-25 14:26:21 +01:00
proc proc: fix coredump vs read /proc/*/stat race 2018-01-23 19:58:18 +01:00
pstore License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
qnx4 License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
qnx6 License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
quota License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
ramfs mm: make pagevec_lookup() update index 2017-09-06 17:27:26 -07:00
reiserfs reiserfs: remove unneeded i_version bump 2018-02-03 17:39:05 +01:00
romfs License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
squashfs License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
sysfs
sysv License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
tracefs
ubifs License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
udf udf: Avoid overflow when session starts at large offset 2017-12-20 10:10:36 +01:00
ufs License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
xfs xfs: log recovery should replay deferred ops in order 2018-02-03 17:39:06 +01:00
aio.c Merge branch 'work.read_write' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs 2017-09-14 19:29:55 -07:00
anon_inodes.c
attr.c License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
bad_inode.c License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
binfmt_aout.c fs: fix kernel_read prototype 2017-09-04 19:05:15 -04:00
binfmt_elf.c Merge branch 'work.set_fs' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs 2017-09-14 18:13:32 -07:00
binfmt_elf_fdpic.c Merge branch 'work.set_fs' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs 2017-09-14 18:13:32 -07:00
binfmt_em86.c
binfmt_flat.c License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
binfmt_misc.c fs/binfmt_misc.c: node could be NULL when evicting inode 2017-10-13 16:18:33 -07:00
binfmt_script.c exec: load_script: kill the onstack interp[BINPRM_BUF_SIZE] array 2017-10-03 17:54:25 -07:00
block_dev.c fs/mpage.c: fix mpage_writepage() for pages with buffers 2017-10-13 16:18:33 -07:00
buffer.c fs: guard_bio_eod() needs to consider partitions 2017-11-30 08:40:45 +00:00
char_dev.c License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
compat.c
compat_binfmt_elf.c
compat_ioctl.c License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
coredump.c License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
dax.c dax: fix PMD faults on zero-length files 2017-11-30 08:40:53 +00:00
dcache.c locking/barriers: Convert users of lockless_dereference() to READ_ONCE() 2017-12-25 14:26:21 +01:00
dcookies.c
direct-io.c Changes since last update: 2017-10-18 14:51:50 -04:00
drop_caches.c License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
eventfd.c
eventpoll.c fs/epoll: use faster rb_first_cached() 2017-09-08 18:26:49 -07:00
exec.c exec: Weaken dumpability for secureexec 2018-01-05 15:48:57 +01:00
fcntl.c fcntl: don't cap l_start and l_end values for F_GETLK64 in compat syscall 2017-12-17 15:07:59 +01:00
fhandle.c License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
file.c License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
file_table.c fput: Don't reinvent the wheel but use existing llist API 2017-08-28 00:50:23 -04:00
filesystems.c License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
fs-writeback.c
fs_pin.c License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
fs_struct.c
inode.c Merge branch 'overlayfs-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mszeredi/vfs 2017-09-13 09:11:44 -07:00
internal.h Merge branch 'overlayfs-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mszeredi/vfs 2017-09-13 09:11:44 -07:00
ioctl.c License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
iomap.c fs: invalidate page cache after end_io() in dio completion 2017-10-16 12:11:56 -07:00
Kconfig
Kconfig.binfmt
libfs.c
locks.c locks: restore a warn for leaked locks on close 2017-07-21 13:57:31 -04:00
Makefile License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
mbcache.c
mount.h License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
mpage.c License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
namei.c autofs: revert "autofs: fix AT_NO_AUTOMOUNT not being honored" 2017-12-05 11:26:29 +01:00
namespace.c VFS: Handle lazytime in do_mount() 2018-02-03 17:38:53 +01:00
no-block.c
nsfs.c License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
open.c ovl: don't allow writing ioctl on lower layer 2017-09-05 12:53:12 +02:00
pipe.c pipe: avoid round_pipe_size() nr_pages overflow on 32-bit 2018-01-23 19:58:14 +01:00
pnode.c
pnode.h
posix_acl.c
proc_namespace.c License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
read_write.c License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
readdir.c License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
select.c License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
seq_file.c License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
signalfd.c License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
splice.c fs: move kernel_write to fs/read_write.c 2017-09-04 19:05:15 -04:00
stack.c
stat.c License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
statfs.c License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
super.c License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
sync.c License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
timerfd.c License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
userfaultfd.c userfaultfd: clear the vma->vm_userfaultfd_ctx if UFFD_EVENT_FORK fails 2018-01-10 09:31:17 +01:00
utimes.c License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
xattr.c lsm: fix smack_inode_removexattr and xattr_getsecurity memleak 2017-10-04 18:03:15 +11:00