Fwd: [RFC 2/2] io_uring: mark REQ_NOWAIT for a non-mq queue as unspported

Fwd: [RFC 2/2] io_uring: mark REQ_NOWAIT for a non-mq queue as unspported

[Jens Axboe]

-------- Forwarded Message --------
Subject: [RFC 2/2] io_uring: mark REQ_NOWAIT for a non-mq queue as unspported
Date: Tue, 19 May 2020 14:52:50 -0700
From: Bijan Mottahedeh <bijan.mottahedeh@oracle.com>
To: axboe@kernel.dk
CC: io-uring@vger.kernel.org

Mark a REQ_NOWAIT request for a non-mq queue as unspported instead of
retryable since otherwise the io_uring layer will keep resubmitting
the request.

Signed-off-by: Bijan Mottahedeh <bijan.mottahedeh@oracle.com>
---
 block/blk-core.c | 10 +++-------
 1 file changed, 3 insertions(+), 7 deletions(-)

diff --git a/block/blk-core.c b/block/blk-core.c
index 5847993..3807140 100644
--- a/block/blk-core.c
+++ b/block/blk-core.c
@@ -962,14 +962,10 @@ static inline blk_status_t blk_check_zone_append(struct request_queue *q,
 	}
 
 	/*
-	 * Non-mq queues do not honor REQ_NOWAIT, so complete a bio
-	 * with BLK_STS_AGAIN status in order to catch -EAGAIN and
-	 * to give a chance to the caller to repeat request gracefully.
+	 * Non-mq queues do not honor REQ_NOWAIT, return -EOPNOTSUPP.
 	 */
-	if ((bio->bi_opf & REQ_NOWAIT) && !queue_is_mq(q)) {
-		status = BLK_STS_AGAIN;
-		goto end_io;
-	}
+	if ((bio->bi_opf & REQ_NOWAIT) && !queue_is_mq(q))
+		goto not_supported;
 
 	if (should_fail_bio(bio))
 		goto end_io;
-- 
1.8.3.1

Re: Fwd: [RFC 2/2] io_uring: mark REQ_NOWAIT for a non-mq queue as unspported

[Christoph Hellwig]

On Tue, May 19, 2020 at 04:16:45PM -0600, Jens Axboe wrote:
> 
> 
> 
> -------- Forwarded Message --------
> Subject: [RFC 2/2] io_uring: mark REQ_NOWAIT for a non-mq queue as unspported
> Date: Tue, 19 May 2020 14:52:50 -0700
> From: Bijan Mottahedeh <bijan.mottahedeh@oracle.com>
> To: axboe@kernel.dk
> CC: io-uring@vger.kernel.org
> 
> Mark a REQ_NOWAIT request for a non-mq queue as unspported instead of
> retryable since otherwise the io_uring layer will keep resubmitting
> the request.

Someone needs to audit if this translates into the right errors message
for the syscalls, as they eventually need to return -EAGAIN
to userspace.