]> git.proxmox.com Git - mirror_ubuntu-kernels.git/commit
ext4: add error checking to ext4_ext_replay_set_iblocks()
authorTheodore Ts'o <tytso@mit.edu>
Thu, 2 Sep 2021 15:36:01 +0000 (11:36 -0400)
committerTheodore Ts'o <tytso@mit.edu>
Thu, 2 Sep 2021 15:36:01 +0000 (11:36 -0400)
commit1fd95c05d8f742abfe906620780aee4dbe1a2db0
tree47923c8ffa2cdbb3ca996b0c470688b947b0d1c9
parentbaaae979b112642a41b71c71c599d875c067d257
ext4: add error checking to ext4_ext_replay_set_iblocks()

If the call to ext4_map_blocks() fails due to an corrupted file
system, ext4_ext_replay_set_iblocks() can get stuck in an infinite
loop.  This could be reproduced by running generic/526 with a file
system that has inline_data and fast_commit enabled.  The system will
repeatedly log to the console:

EXT4-fs warning (device dm-3): ext4_block_to_path:105: block 1074800922 > max in inode 131076

and the stack that it gets stuck in is:

   ext4_block_to_path+0xe3/0x130
   ext4_ind_map_blocks+0x93/0x690
   ext4_map_blocks+0x100/0x660
   skip_hole+0x47/0x70
   ext4_ext_replay_set_iblocks+0x223/0x440
   ext4_fc_replay_inode+0x29e/0x3b0
   ext4_fc_replay+0x278/0x550
   do_one_pass+0x646/0xc10
   jbd2_journal_recover+0x14a/0x270
   jbd2_journal_load+0xc4/0x150
   ext4_load_journal+0x1f3/0x490
   ext4_fill_super+0x22d4/0x2c00

With this patch, generic/526 still fails, but system is no longer
locking up in a tight loop.  It's likely the root casue is that
fast_commit replay is corrupting file systems with inline_data, and we
probably need to add better error handling in the fast commit replay
code path beyond what is done here, which essentially just breaks the
infinite loop without reporting the to the higher levels of the code.

Fixes: 8016E29F4362 ("ext4: fast commit recovery path")
Cc: stable@kernel.org
Cc: Harshad Shirwadkar <harshadshirwadkar@gmail.com>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
fs/ext4/extents.c