[PATCH] block: advance iov_iter on bio_add_hw_page failure

[PATCH] block: advance iov_iter on bio_add_hw_page failure

[Naohiro Aota]

When the bio's size reaches max_append_sectors, bio_add_hw_page returns
0 then __bio_iov_append_get_pages returns -EINVAL. This is an expected
result of building a small enough bio not to be split in the IO path.
However, iov_iter is not advanced in this case, causing the same pages
are filled for the bio again and again.

Fix the case by properly advancing the iov_iter for already processed
pages.

Fixes: 0512a75b98f8 ("block: Introduce REQ_OP_ZONE_APPEND")
Cc: stable@vger.kernel.org # 5.8+
Signed-off-by: Naohiro Aota <naohiro.aota@wdc.com>
---
 block/bio.c | 11 +++++++----
 1 file changed, 7 insertions(+), 4 deletions(-)

diff --git a/block/bio.c b/block/bio.c
index e0d41ccc4e90..2dfe40be4d6b 100644
--- a/block/bio.c
+++ b/block/bio.c
@@ -1082,6 +1082,7 @@ static int __bio_iov_append_get_pages(struct bio *bio, struct iov_iter *iter)
 	ssize_t size, left;
 	unsigned len, i;
 	size_t offset;
+	int ret = 0;
 
 	if (WARN_ON_ONCE(!max_append_sectors))
 		return 0;
@@ -1104,15 +1105,17 @@ static int __bio_iov_append_get_pages(struct bio *bio, struct iov_iter *iter)
 
 		len = min_t(size_t, PAGE_SIZE - offset, left);
 		if (bio_add_hw_page(q, bio, page, len, offset,
-				max_append_sectors, &same_page) != len)
-			return -EINVAL;
+				max_append_sectors, &same_page) != len) {
+			ret = -EINVAL;
+			break;
+		}
 		if (same_page)
 			put_page(page);
 		offset = 0;
 	}
 
-	iov_iter_advance(iter, size);
-	return 0;
+	iov_iter_advance(iter, size - left);
+	return ret;
 }
 
 /**
-- 
2.27.0

Re: [PATCH] block: advance iov_iter on bio_add_hw_page failure

[Johannes Thumshirn]

Thanks,
Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>

Re: [PATCH] block: advance iov_iter on bio_add_hw_page failure

[Jens Axboe]

On 10/28/20 1:25 AM, Naohiro Aota wrote:
> When the bio's size reaches max_append_sectors, bio_add_hw_page returns
> 0 then __bio_iov_append_get_pages returns -EINVAL. This is an expected
> result of building a small enough bio not to be split in the IO path.
> However, iov_iter is not advanced in this case, causing the same pages
> are filled for the bio again and again.
> 
> Fix the case by properly advancing the iov_iter for already processed
> pages.

Applied, thanks.

-- 
Jens Axboe