jbd2 submits buffer_heads it allocates directly for journal descriptor
and commit-record blocks, which have no folio backing at all - as the
comment on buffer_set_crypto_ctx() just above already notes. Unlike
that check, and unlike the bh->b_folio check further down in the same
function, __bh_submit() calls folio_test_dropbehind(bh->b_folio)
unconditionally, so committing a transaction on one of these bh's
null-derefs while reading the folio's flags:

  BUG: kernel NULL pointer dereference, address: 0000000000000000
  RIP: __bh_submit+0x90/0x140
  Call Trace:
   jbd2_journal_commit_transaction
   kjournald2
   kthread

Guard the call the same way.

Fixes: a2c924c240e7 ("buffer: set BIO_COMPLETE_IN_TASK for dropbehind writeback")
Signed-off-by: Tanawat Jukmongkol <tanawat.jukmon@gmail.com>
---
 fs/buffer.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/fs/buffer.c b/fs/buffer.c
index 427d8a817..4aea0bb08 100644
--- a/fs/buffer.c
+++ b/fs/buffer.c
@@ -1106,7 +1106,7 @@ static void __bh_submit(struct buffer_head *bh, blk_opf_t opf,

 	bio = bio_alloc(bh->b_bdev, 1, opf, GFP_NOIO);

-	if (folio_test_dropbehind(bh->b_folio) && op_is_write(opf))
+	if (bh->b_folio && folio_test_dropbehind(bh->b_folio) && op_is_write(opf))
 		bio_set_flag(bio, BIO_COMPLETE_IN_TASK);

 	if (IS_ENABLED(CONFIG_FS_ENCRYPTION))

Please disregard this patch — it duplicates a fix already posted and
under review:

  https://lore.kernel.org/lkml/20260902013357.2815214-1-joseph.qi@linux.alibaba.com/

Sorry for the noise.

Tanawat Jukmongkol