Re: [PATCH 1/2] block: Make __submit_bio_noacct() preserve the bio submission order

[Bart Van Assche <bvanassche@xxxxxxx>]

On 6/8/25 3:47 PM, Damien Le Moal wrote:
> So yes, we need a fix. Can you work on one ?

The patch below seems to be sufficient but I'm not sure whether this
approach is acceptable:

Subject: [PATCH] block: Preserve the LBA order when splitting a bio

Preserve the bio order if bio_split() is called on the prefix returned
by an earlier bio_split() call. This can happen with fscrypt and the
inline encryption fallback code if max_sectors is less than the maximum
bio size supported by the inline encryption fallback code (1 MiB for 4 KiB
pages) or when using zoned storage and the distance from the start of the
bio to the next zone boundary is less than 1 MiB.

Fixes: 488f6682c832 ("block: blk-crypto-fallback for Inline Encryption")
Signed-off-by: Bart Van Assche <bvanassche@xxxxxxx>
---
 block/bio.c               |  8 ++++++++
 block/blk-core.c          | 12 ++++++++----
 include/linux/blk_types.h |  5 +++++
 3 files changed, 21 insertions(+), 4 deletions(-)

diff --git a/block/bio.c b/block/bio.c
index 3c0a558c90f5..440ed443545c 100644
--- a/block/bio.c
+++ b/block/bio.c
@@ -1689,6 +1689,14 @@ struct bio *bio_split(struct bio *bio, int sectors,

 	bio_advance(bio, split->bi_iter.bi_size);

+	/*
+	 * If bio_split() is called on a prefix from an earlier bio_split()
+	 * call, adding it at the head of current->bio_list[0] preserves the
+	 * LBA order. This is essential when writing data to a zoned block
+	 * device and when using REQ_OP_WRITE instead of REQ_OP_ZONE_APPEND.
+	 */
+	bio_set_flag(bio, BIO_ADD_AT_HEAD);
+
 	if (bio_flagged(bio, BIO_TRACE_COMPLETION))
 		bio_set_flag(split, BIO_TRACE_COMPLETION);

diff --git a/block/blk-core.c b/block/blk-core.c
index b862c66018f2..570a14a7bcd4 100644
--- a/block/blk-core.c
+++ b/block/blk-core.c
@@ -745,12 +745,16 @@ void submit_bio_noacct_nocheck(struct bio *bio)
 	 * to collect a list of requests submited by a ->submit_bio method while
 	 * it is active, and then process them after it returned.
 	 */
-	if (current->bio_list)
-		bio_list_add(&current->bio_list[0], bio);
-	else if (!bdev_test_flag(bio->bi_bdev, BD_HAS_SUBMIT_BIO))
+	if (current->bio_list) {
+		if (bio_flagged(bio, BIO_ADD_AT_HEAD))
+			bio_list_add_head(&current->bio_list[0], bio);
+		else
+			bio_list_add(&current->bio_list[0], bio);
+	} else if (!bdev_test_flag(bio->bi_bdev, BD_HAS_SUBMIT_BIO)) {
 		__submit_bio_noacct_mq(bio);
-	else
+	} else {
 		__submit_bio_noacct(bio);
+	}
 }

 static blk_status_t blk_validate_atomic_write_op_size(struct 
request_queue *q,
diff --git a/include/linux/blk_types.h b/include/linux/blk_types.h
index 3d1577f07c1c..0e2d3fd8d40a 100644
--- a/include/linux/blk_types.h
+++ b/include/linux/blk_types.h
@@ -308,6 +308,11 @@ enum {
 	BIO_REMAPPED,
 	BIO_ZONE_WRITE_PLUGGING, /* bio handled through zone write plugging */
 	BIO_EMULATES_ZONE_APPEND, /* bio emulates a zone append operation */
+	/*
+	 * make submit_bio_noacct_nocheck() add this bio at the head of
+	 * current->bio_list[0].
+	 */
+	BIO_ADD_AT_HEAD,
 	BIO_FLAG_LAST
 };