ext4: Make sure BH_New bit is cleared in ->write_end handler
authorJan Kara <jack@suse.cz>
Wed, 9 Jul 2025 08:48:32 +0000 (10:48 +0200)
committerTheodore Ts'o <tytso@mit.edu>
Fri, 18 Jul 2025 03:25:21 +0000 (23:25 -0400)
commit91b8ca8b26729b729dda8a4eddb9aceaea706f37
tree7d69629902d36182c8440a8648cab1b71a063b2f
parentc678bdc998754589cea2e6afab9401d7d8312ac4
ext4: Make sure BH_New bit is cleared in ->write_end handler

Currently we clear BH_New bit in case of error and also in the standard
ext4_write_end() handler (in block_commit_write()). However
ext4_journalled_write_end() misses this clearing and thus we are leaving
stale BH_New bits behind. Generally ext4_block_write_begin() clears
these bits before any harm can be done but in case blocksize < pagesize
and we hit some error when processing a page with these stale bits,
we'll try to zero buffers with these stale BH_New bits and jbd2 will
complain (as buffers were not prepared for writing in this transaction).
Fix the problem by clearing BH_New bits in ext4_journalled_write_end()
and WARN if ext4_block_write_begin() sees stale BH_New bits.

Reported-by: Baolin Liu <liubaolin12138@163.com>
Reported-by: Zhi Long <longzhi@sangfor.com.cn>
Fixes: 3910b513fcdf ("ext4: persist the new uptodate buffers in ext4_journalled_zero_new_buffers")
Signed-off-by: Jan Kara <jack@suse.cz>
Link: https://patch.msgid.link/20250709084831.23876-2-jack@suse.cz
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
fs/ext4/inline.c
fs/ext4/inode.c