a regression fix caused by a refactoring in 5.11. A corrupted superblock
wouldn't be detected by checksum verification due to wrongly placed
initialization of the checksum length, thus making memcmp always work.

I've verified it manually and ran other test suites before sending this.
Please pull, thanks.

The following changes since commit 9ad6d91f056b99dbe59a262810cb342519ea8d39:

  btrfs: fix log replay failure due to race with space cache rebuild 
(2021-01-25 18:44:53 +0100)

are available in the Git repository at:

  git://git.kernel.org/pub/scm/linux/kernel/git/kdave/linux.git for-5.11-rc7-tag

for you to fetch changes up to 83c68bbcb6ac2dbbcaf12e2281a29a9f73b97d0f:

  btrfs: initialize fs_info::csum_size earlier in open_ctree (2021-02-12 
14:48:24 +0100)

Su Yue (1):
      btrfs: initialize fs_info::csum_size earlier in open_ctree

 fs/btrfs/disk-io.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

Reply via email to