public inbox for [email protected]
 help / color / mirror / Atom feed
From: Pavel Begunkov <[email protected]>
To: Jens Axboe <[email protected]>, [email protected]
Subject: [PATCH 2/3] io_uring: fail io-wq submission from a task_work
Date: Thu, 18 Feb 2021 22:32:52 +0000	[thread overview]
Message-ID: <ae6848eec1847ff3811f13363f15308f033e7d41.1613687339.git.asml.silence@gmail.com> (raw)
In-Reply-To: <[email protected]>

In case of failure io_wq_submit_work() needs to post an CQE and so
potentially take uring_lock. The safest way to deal with it is to do
that from under task_work where we can safely take the lock.

Also, as io_iopoll_check() holds the lock tight and releases it
reluctantly, it will play nicer in the furuter with notifying an
iopolling task about new such pending failed requests.

Signed-off-by: Pavel Begunkov <[email protected]>
---
 fs/io_uring.c | 48 ++++++++++++++++++------------------------------
 1 file changed, 18 insertions(+), 30 deletions(-)

diff --git a/fs/io_uring.c b/fs/io_uring.c
index 8dab07f42b34..1cb5e40d9822 100644
--- a/fs/io_uring.c
+++ b/fs/io_uring.c
@@ -2338,7 +2338,7 @@ static void io_req_task_cancel(struct callback_head *cb)
 	struct io_ring_ctx *ctx = req->ctx;
 
 	mutex_lock(&ctx->uring_lock);
-	__io_req_task_cancel(req, -ECANCELED);
+	__io_req_task_cancel(req, req->result);
 	mutex_unlock(&ctx->uring_lock);
 	percpu_ref_put(&ctx->refs);
 }
@@ -2371,11 +2371,22 @@ static void io_req_task_queue(struct io_kiocb *req)
 	req->task_work.func = io_req_task_submit;
 	ret = io_req_task_work_add(req);
 	if (unlikely(ret)) {
+		ret = -ECANCELED;
 		percpu_ref_get(&req->ctx->refs);
 		io_req_task_work_add_fallback(req, io_req_task_cancel);
 	}
 }
 
+static void io_req_task_queue_fail(struct io_kiocb *req, int ret)
+{
+	percpu_ref_get(&req->ctx->refs);
+	req->result = ret;
+	req->task_work.func = io_req_task_cancel;
+
+	if (unlikely(io_req_task_work_add(req)))
+		io_req_task_work_add_fallback(req, io_req_task_cancel);
+}
+
 static inline void io_queue_next(struct io_kiocb *req)
 {
 	struct io_kiocb *nxt = io_req_find_next(req);
@@ -6428,13 +6439,8 @@ static void io_wq_submit_work(struct io_wq_work *work)
 	if (timeout)
 		io_queue_linked_timeout(timeout);
 
-	if (work->flags & IO_WQ_WORK_CANCEL) {
-		/* io-wq is going to take down one */
-		refcount_inc(&req->refs);
-		percpu_ref_get(&req->ctx->refs);
-		io_req_task_work_add_fallback(req, io_req_task_cancel);
-		return;
-	}
+	if (work->flags & IO_WQ_WORK_CANCEL)
+		ret = -ECANCELED;
 
 	if (!ret) {
 		do {
@@ -6450,29 +6456,11 @@ static void io_wq_submit_work(struct io_wq_work *work)
 		} while (1);
 	}
 
+	/* avoid locking problems by failing it from a clean context */
 	if (ret) {
-		struct io_ring_ctx *lock_ctx = NULL;
-
-		if (req->ctx->flags & IORING_SETUP_IOPOLL)
-			lock_ctx = req->ctx;
-
-		/*
-		 * io_iopoll_complete() does not hold completion_lock to
-		 * complete polled io, so here for polled io, we can not call
-		 * io_req_complete() directly, otherwise there maybe concurrent
-		 * access to cqring, defer_list, etc, which is not safe. Given
-		 * that io_iopoll_complete() is always called under uring_lock,
-		 * so here for polled io, we also get uring_lock to complete
-		 * it.
-		 */
-		if (lock_ctx)
-			mutex_lock(&lock_ctx->uring_lock);
-
-		req_set_fail_links(req);
-		io_req_complete(req, ret);
-
-		if (lock_ctx)
-			mutex_unlock(&lock_ctx->uring_lock);
+		/* io-wq is going to take one down */
+		refcount_inc(&req->refs);
+		io_req_task_queue_fail(req, ret);
 	}
 }
 
-- 
2.24.0


  parent reply	other threads:[~2021-02-18 22:37 UTC|newest]

Thread overview: 8+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-02-18 22:32 [PATCH 0/3] fix nested uring_lock Pavel Begunkov
2021-02-18 22:32 ` [PATCH 1/3] io_uring: don't take uring_lock during iowq cancel Pavel Begunkov
2021-02-18 22:32 ` Pavel Begunkov [this message]
2021-02-21  1:01   ` [PATCH 2/3] io_uring: fail io-wq submission from a task_work Pavel Begunkov
2021-02-21  2:03     ` Jens Axboe
2021-02-18 22:32 ` [PATCH 3/3] io_uring: avoid taking ctx refs for task-cancel Pavel Begunkov
2021-02-18 23:14   ` Jens Axboe
2021-02-18 23:16 ` [PATCH 0/3] fix nested uring_lock 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=ae6848eec1847ff3811f13363f15308f033e7d41.1613687339.git.asml.silence@gmail.com \
    [email protected] \
    [email protected] \
    [email protected] \
    /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