Skip to content

ext4: Various fixes of ext4 handling of fs errors

Jan Kara linux-ext4 1 version Applied

Status from this message: Thanks, applied.

Get series with b4
b4 am [email protected]

Versions newest first

  1. v1 Latest12 patches 21 repliesApplied

Patches in v1 Applied

#Subject and filesSizeRepliesApplied
1/12ext4: Don't remount read-only with errors=continue on rebootfs/ext4/super.c+6 -82 replies
2/12ext4: Remove redundant sb checksum recomputationfs/ext4/super.c+0 -23 replies
3/12ext4: Standardize error message in ext4_protect_reserved_inode()fs/ext4/block_validity.c+4 -62 replies
4/12ext4: Make ext4_abort() use __ext4_error()fs/ext4/ext4.h fs/ext4/ext4_jbd2.c fs/ext4/inode.c +1 more+37 -822 replies
5/12ext4: Move functions in super.cfs/ext4/super.c+98 -982 replies
6/12ext4: Simplify ext4 error translationfs/ext4/super.c+40 -552 replies
7/12ext4: Defer saving error info from atomic contextfs/ext4/ext4.h fs/ext4/super.c+99 -373 replies
8/12ext4: Combine ext4_handle_error() and save_error_info()fs/ext4/super.c+15 -163 replies
9/12ext4: Drop sync argument of ext4_commit_super()fs/ext4/super.c+22 -251 reply
10/12ext4: Protect superblock modifications with a buffer lockfs/ext4/ext4_jbd2.c fs/ext4/file.c fs/ext4/inode.c +4 more+28 -20 replies
11/12ext4: Save error info to sb through journal if availablefs/ext4/super.c+72 -250 replies
12/12ext4: Use sbi instead of EXT4_SB(sb) in ext4_update_super()fs/ext4/super.c+10 -110 replies

Reviews trailers per patch

7 of 12 patches have a Reviewed-by or Acked-by.

PatchAndreas DilgerHarshad Shirwadkar
1/12R-b
2/12R-b
3/12R-b
4/12R-b
5/12R-b
6/12R-b
7/12
8/12
9/12R-b
10/12
11/12
12/12

Trailers given in replies are filled; those carried in the patches from earlier versions are outlined with dashes.

Read the v1 thread

Cover letter [PATCH 00/12] ext4: Various fixes of ext4 handling of fs errors

Hello,

this patches addresses problems in handling of filesystem errors in ext4. When we hit metadata error, we want to store information about the error in the superblock. Currently we do it through direct superblock modification which can lead to lost information, checksum failures, or DIF/DIX failures. Fix various races in the error handling so that the superblock update is reliable.

The patches have passed xfstests for me in various configurations and some targetted manual testing of the error handling.

Honza