[PATCH 22/23] io_uring: don't wait on CQ exclusively
From: Pavel Begunkov <asml.silence@gmail.com>
Date: 2021-05-19 14:16:14
Also in:
bpf, io-uring, lkml
Subsystem:
filesystems (vfs and infrastructure), the rest · Maintainers:
Alexander Viro, Christian Brauner, Linus Torvalds
It doesn't make much sense for several tasks to wait on a single CQ, it would be racy and involve rather strange code flow with synchronisation, so we don't really care optimising this case. Don't do exclusive CQ waiting and wake up everyone in the queue, it will be handy for implementing waiting non-default CQs. Signed-off-by: Pavel Begunkov <asml.silence@gmail.com> --- fs/io_uring.c | 5 ++--- 1 file changed, 2 insertions(+), 3 deletions(-)
diff --git a/fs/io_uring.c b/fs/io_uring.c
index c4682146afa4..805c10be7ea4 100644
--- a/fs/io_uring.c
+++ b/fs/io_uring.c@@ -7085,7 +7085,7 @@ static int io_wake_function(struct wait_queue_entry *curr, unsigned int mode, */ if (io_should_wake(iowq) || test_bit(0, &iowq->ctx->cq_check_overflow)) return autoremove_wake_function(curr, mode, wake_flags, key); - return -1; + return 0; } static int io_run_task_work_sig(void)
@@ -7176,8 +7176,7 @@ static int io_cqring_wait(struct io_ring_ctx *ctx, int min_events, ret = -EBUSY; break; } - prepare_to_wait_exclusive(&ctx->wait, &iowq.wq, - TASK_INTERRUPTIBLE); + prepare_to_wait(&ctx->wait, &iowq.wq, TASK_INTERRUPTIBLE); ret = io_cqring_wait_schedule(ctx, &iowq, &timeout); finish_wait(&ctx->wait, &iowq.wq); cond_resched();
--
2.31.1