Skip to contentSkip to conversation

block: free copied pages when blk_rq_map_kern() fails

PATCH Applied Jackie Liu 3 participants 3 messages Jul 15 – Jul 16 Block layer

lore
Share & tools
lore

Jackie Liu

author
#0 lore raw

PATCH 1/1 block: free copied pages when blk_rq_map_kern() fails

From: Jackie Liu <[email protected]>

bio_copy_kern() allocates pages that are normally freed by the bio completion callback. If blk_rq_append_bio() rejects the bio, however, blk_rq_map_kern() only drops the bio reference. Since bio_put() does not free pages referenced by the bio vectors, those pages leak.

This can happen when the bio exceeds the queue segment constraints or when a later mapping cannot be merged into a request built by earlier calls. Track whether the buffer was copied and free those pages before dropping the rejected bio.

Fixes 3a5a39276d2a ("block: allow blk_rq_map_kern to append to requests")
Assisted-by Codex:gpt-5.6-sol
---
 block/blk-map.c | 9 +++++++--
 1 file changed, 7 insertions(+), 2 deletions(-)
block/blk-map.c +7 -2
index 768549f19f97..d1d6bbe0ecf1 100644
@@ -653,6 +653,7 @@ int blk_rq_map_kern(struct request *rq, void *kbuf, unsigned int len,
 		gfp_t gfp_mask)
 {
 	unsigned long addr = (unsigned long) kbuf;
+	bool do_copy;
 	struct bio *bio;
 	int ret;
 
@@ -661,7 +662,8 @@ int blk_rq_map_kern(struct request *rq, void *kbuf, unsigned int len,
 	if (!len || !kbuf)
 		return -EINVAL;
 
-	if (!blk_rq_aligned(rq->q, addr, len) || object_is_on_stack(kbuf))
+	do_copy = !blk_rq_aligned(rq->q, addr, len) || object_is_on_stack(kbuf);
+	if (do_copy)
 		bio = bio_copy_kern(rq, kbuf, len, gfp_mask);
 	else
 		bio = bio_map_kern(rq, kbuf, len, gfp_mask);
@@ -670,8 +672,11 @@ int blk_rq_map_kern(struct request *rq, void *kbuf, unsigned int len,
 		return PTR_ERR(bio);
 
 	ret = blk_rq_append_bio(rq, bio);
-	if (unlikely(ret))
+	if (unlikely(ret)) {
+		if (do_copy)
+			bio_free_pages(bio);
 		blk_mq_map_bio_put(bio);
+	}
 	return ret;
 }
 EXPORT_SYMBOL(blk_rq_map_kern);
Signature · 1 line
2.54.0
maintainer Applied
#2 lore raw
In reply to Jackie Liu
Quoted 11 lines · Jackie Liu

bio_copy_kern() allocates pages that are normally freed by the bio completion callback. If blk_rq_append_bio() rejects the bio, however, blk_rq_map_kern() only drops the bio reference. Since bio_put() does not free pages referenced by the bio vectors, those pages leak.

This can happen when the bio exceeds the queue segment constraints or when a later mapping cannot be merged into a request built by earlier calls. Track whether the buffer was copied and free those pages before dropping the rejected bio.

[...]

Applied, thanks!

[1/1] block: free copied pages when blk_rq_map_kern() fails

      commit: bd2df8dc72201f626d66ca97f2afc8b7f2740713

Best regards,

Signature · 1 line
Jens Axboe