io-uring.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Ming Lei <ming.lei@redhat.com>
To: Jens Axboe <axboe@kernel.dk>,
	io-uring@vger.kernel.org, linux-block@vger.kernel.org
Cc: Ming Lei <ming.lei@redhat.com>
Subject: [PATCH for-6.7/io_uring 7/7] ublk: simplify aborting request
Date: Mon,  9 Oct 2023 17:33:22 +0800	[thread overview]
Message-ID: <20231009093324.957829-8-ming.lei@redhat.com> (raw)
In-Reply-To: <20231009093324.957829-1-ming.lei@redhat.com>

Now ublk_abort_queue() is run exclusively with ublk_queue_rq() and the
ubq_daemon task, so simplify aborting request:

- set UBLK_IO_FLAG_ABORTED in ublk_abort_queue() just for aborting
this request

- abort request in ublk_queue_rq() if ubq->canceling is set

Signed-off-by: Ming Lei <ming.lei@redhat.com>
---
 drivers/block/ublk_drv.c | 41 +++++++++++-----------------------------
 1 file changed, 11 insertions(+), 30 deletions(-)

diff --git a/drivers/block/ublk_drv.c b/drivers/block/ublk_drv.c
index 75ed7b87a844..20237b8f318a 100644
--- a/drivers/block/ublk_drv.c
+++ b/drivers/block/ublk_drv.c
@@ -1083,13 +1083,10 @@ static void __ublk_fail_req(struct ublk_queue *ubq, struct ublk_io *io,
 {
 	WARN_ON_ONCE(io->flags & UBLK_IO_FLAG_ACTIVE);
 
-	if (!(io->flags & UBLK_IO_FLAG_ABORTED)) {
-		io->flags |= UBLK_IO_FLAG_ABORTED;
-		if (ublk_queue_can_use_recovery_reissue(ubq))
-			blk_mq_requeue_request(req, false);
-		else
-			ublk_put_req_ref(ubq, req);
-	}
+	if (ublk_queue_can_use_recovery_reissue(ubq))
+		blk_mq_requeue_request(req, false);
+	else
+		ublk_put_req_ref(ubq, req);
 }
 
 static void ubq_complete_io_cmd(struct ublk_io *io, int res,
@@ -1230,27 +1227,10 @@ static void ublk_rq_task_work_cb(struct io_uring_cmd *cmd, unsigned issue_flags)
 static void ublk_queue_cmd(struct ublk_queue *ubq, struct request *rq)
 {
 	struct ublk_rq_data *data = blk_mq_rq_to_pdu(rq);
-	struct ublk_io *io;
 
-	if (!llist_add(&data->node, &ubq->io_cmds))
-		return;
+	if (llist_add(&data->node, &ubq->io_cmds)) {
+		struct ublk_io *io = &ubq->ios[rq->tag];
 
-	io = &ubq->ios[rq->tag];
-	/*
-	 * If the check pass, we know that this is a re-issued request aborted
-	 * previously in cancel fn because the ubq_daemon(cmd's task) is
-	 * PF_EXITING. We cannot call io_uring_cmd_complete_in_task() anymore
-	 * because this ioucmd's io_uring context may be freed now if no inflight
-	 * ioucmd exists. Otherwise we may cause null-deref in ctx->fallback_work.
-	 *
-	 * Note: cancel fn sets UBLK_IO_FLAG_ABORTED and ends this request(releasing
-	 * the tag). Then the request is re-started(allocating the tag) and we are here.
-	 * Since releasing/allocating a tag implies smp_mb(), finding UBLK_IO_FLAG_ABORTED
-	 * guarantees that here is a re-issued request aborted previously.
-	 */
-	if (unlikely(io->flags & UBLK_IO_FLAG_ABORTED)) {
-		ublk_abort_io_cmds(ubq);
-	} else {
 		io_uring_cmd_complete_in_task(io->cmd, ublk_rq_task_work_cb);
 	}
 }
@@ -1320,13 +1300,12 @@ static blk_status_t ublk_queue_rq(struct blk_mq_hw_ctx *hctx,
 	if (ublk_queue_can_use_recovery(ubq) && unlikely(ubq->force_abort))
 		return BLK_STS_IOERR;
 
-	blk_mq_start_request(bd->rq);
-
 	if (unlikely(ubq->canceling)) {
 		__ublk_abort_rq(ubq, rq);
 		return BLK_STS_OK;
 	}
 
+	blk_mq_start_request(bd->rq);
 	ublk_queue_cmd(ubq, rq);
 
 	return BLK_STS_OK;
@@ -1449,8 +1428,10 @@ static void ublk_abort_queue(struct ublk_device *ub, struct ublk_queue *ubq)
 			 * will do it
 			 */
 			rq = blk_mq_tag_to_rq(ub->tag_set.tags[ubq->q_id], i);
-			if (rq)
+			if (rq && blk_mq_request_started(rq)) {
+				io->flags |= UBLK_IO_FLAG_ABORTED;
 				__ublk_fail_req(ubq, io, rq);
+			}
 		}
 	}
 }
@@ -1534,7 +1515,7 @@ static void ublk_uring_cmd_cancel_fn(struct io_uring_cmd *cmd,
 
 	io = &ubq->ios[pdu->tag];
 	WARN_ON_ONCE(io->cmd != cmd);
-	ublk_cancel_cmd(ubq, &ubq->ios[pdu->tag], issue_flags);
+	ublk_cancel_cmd(ubq, io, issue_flags);
 
 	if (need_schedule) {
 		if (ublk_can_use_recovery(ub))
-- 
2.41.0


  parent reply	other threads:[~2023-10-09  9:36 UTC|newest]

Thread overview: 13+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2023-10-09  9:33 [PATCH for-6.7/io_uring 0/7] ublk: simplify abort with cancelable uring_cmd Ming Lei
2023-10-09  9:33 ` [PATCH for-6.7/io_uring 1/7] ublk: don't get ublk device reference in ublk_abort_queue() Ming Lei
2023-10-09  9:33 ` [PATCH for-6.7/io_uring 2/7] ublk: make sure io cmd handled in submitter task context Ming Lei
2023-10-09  9:33 ` [PATCH for-6.7/io_uring 3/7] ublk: move ublk_cancel_dev() out of ub->mutex Ming Lei
2023-10-09  9:33 ` [PATCH for-6.7/io_uring 4/7] ublk: rename mm_lock as lock Ming Lei
2023-10-09  9:33 ` [PATCH for-6.7/io_uring 5/7] ublk: quiesce request queue when aborting queue Ming Lei
2023-10-09  9:33 ` [PATCH for-6.7/io_uring 6/7] ublk: replace monitor with cancelable uring_cmd Ming Lei
2023-10-09  9:33 ` Ming Lei [this message]
2023-10-16 23:47 ` [PATCH for-6.7/io_uring 0/7] ublk: simplify abort " Ming Lei
2023-10-17  1:14   ` Jens Axboe
2023-10-17  1:13 ` Jens Axboe
2023-10-17  5:00   ` Ming Lei
2023-10-17 14:24     ` Jens Axboe

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20231009093324.957829-8-ming.lei@redhat.com \
    --to=ming.lei@redhat.com \
    --cc=axboe@kernel.dk \
    --cc=io-uring@vger.kernel.org \
    --cc=linux-block@vger.kernel.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).