io_uring: avoid io-wq -EAGAIN looping for !IOPOLL
authorPavel Begunkov <asml.silence@gmail.com>
Fri, 13 May 2022 10:24:56 +0000 (11:24 +0100)
committerJens Axboe <axboe@kernel.dk>
Fri, 13 May 2022 12:50:42 +0000 (06:50 -0600)
If an opcode handler semi-reliably returns -EAGAIN, io_wq_submit_work()
might continue busily hammer the same handler over and over again, which
is not ideal. The -EAGAIN handling in question was put there only for
IOPOLL, so restrict it to IOPOLL mode only where there is no other
recourse than to retry as we cannot wait.

Fixes: def596e9557c9 ("io_uring: support for IO polling")
Signed-off-by: Pavel Begunkov <asml.silence@gmail.com>
Link: https://lore.kernel.org/r/f168b4f24181942f3614dd8ff648221736f572e6.1652433740.git.asml.silence@gmail.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>
fs/io_uring.c

index 3621898..f5f3832 100644 (file)
@@ -7573,6 +7573,8 @@ fail:
                 * wait for request slots on the block side.
                 */
                if (!needs_poll) {
+                       if (!(req->ctx->flags & IORING_SETUP_IOPOLL))
+                               break;
                        cond_resched();
                        continue;
                }