[PATCH 1/1] io_uring: further remove sqpoll limits on opcodes

[PATCH 1/1] io_uring: further remove sqpoll limits on opcodes

[Pavel Begunkov]

There are three types of requests that left disabled for sqpoll, namely
epoll ctx, statx, and resources update. Since SQPOLL task is now closely
mimics a userspace thread, remove the restrictions.

Signed-off-by: Pavel Begunkov <asml.silence@gmail.com>
---
 fs/io_uring.c | 6 ++----
 1 file changed, 2 insertions(+), 4 deletions(-)

diff --git a/fs/io_uring.c b/fs/io_uring.c
index 8f718d26f01c..981c0f86e054 100644
--- a/fs/io_uring.c
+++ b/fs/io_uring.c
@@ -4037,7 +4037,7 @@ static int io_epoll_ctl_prep(struct io_kiocb *req,
 #if defined(CONFIG_EPOLL)
 	if (sqe->ioprio || sqe->buf_index)
 		return -EINVAL;
-	if (unlikely(req->ctx->flags & (IORING_SETUP_IOPOLL | IORING_SETUP_SQPOLL)))
+	if (unlikely(req->ctx->flags & IORING_SETUP_IOPOLL))
 		return -EINVAL;
 
 	req->epoll.epfd = READ_ONCE(sqe->fd);
@@ -4152,7 +4152,7 @@ static int io_fadvise(struct io_kiocb *req, unsigned int issue_flags)
 
 static int io_statx_prep(struct io_kiocb *req, const struct io_uring_sqe *sqe)
 {
-	if (unlikely(req->ctx->flags & (IORING_SETUP_IOPOLL | IORING_SETUP_SQPOLL)))
+	if (unlikely(req->ctx->flags & IORING_SETUP_IOPOLL))
 		return -EINVAL;
 	if (sqe->ioprio || sqe->buf_index)
 		return -EINVAL;
@@ -5829,8 +5829,6 @@ static int io_async_cancel(struct io_kiocb *req, unsigned int issue_flags)
 static int io_rsrc_update_prep(struct io_kiocb *req,
 				const struct io_uring_sqe *sqe)
 {
-	if (unlikely(req->ctx->flags & IORING_SETUP_SQPOLL))
-		return -EINVAL;
 	if (unlikely(req->flags & (REQ_F_FIXED_FILE | REQ_F_BUFFER_SELECT)))
 		return -EINVAL;
 	if (sqe->ioprio || sqe->rw_flags)
-- 
2.31.1

Re: [PATCH 1/1] io_uring: further remove sqpoll limits on opcodes

[Jens Axboe]

On 5/14/21 5:05 AM, Pavel Begunkov wrote:
> There are three types of requests that left disabled for sqpoll, namely
> epoll ctx, statx, and resources update. Since SQPOLL task is now closely
> mimics a userspace thread, remove the restrictions.

This is nice, now SQPOLL is 100% like non sqpoll, which it should be with
the io threads. Applied, thanks.

-- 
Jens Axboe