Skip to contentSkip to conversation

io_uring:IORING_SETUP_SQPOLL don't need to enter io_cqring_wait

PATCH In review [email protected] 3 participants 4 messages Apr 7, 2020 – Apr 8, 2020 io_uring

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 io_uring:IORING_SETUP_SQPOLL don't need to enter io_cqring_wait

From: Liming Wu <[email protected]>

When SETUP_IOPOLL and SETUP_SQPOLL are both enabled, app don't need to enter io_cqring_wait too. If I misunderstand, please give me some advise.

Signed-off-by Liming Wu <[email protected]>

---
 io_uring.c | 11 ++++++-----
 1 file changed, 6 insertions(+), 5 deletions(-)
io_uring.c +6 -5
index b12d33b..36e884f 100644
@@ -7418,11 +7418,12 @@ SYSCALL_DEFINE6(io_uring_enter, unsigned int, fd, u32, to_submit,
 		 * polling again, they can rely on io_sq_thread to do polling
 		 * work, which can reduce cpu usage and uring_lock contention.
 		 */
-		if (ctx->flags & IORING_SETUP_IOPOLL &&
-		    !(ctx->flags & IORING_SETUP_SQPOLL)) {
-			ret = io_iopoll_check(ctx, &nr_events, min_complete);
-		} else {
-			ret = io_cqring_wait(ctx, min_complete, sig, sigsz);
+		if (!(ctx->flags & IORING_SETUP_SQPOLL)) {
+		    if (ctx->flags & IORING_SETUP_IOPOLL) {
+		    	ret = io_iopoll_check(ctx, &nr_events, min_complete);
+		    } else {
+		    	ret = io_cqring_wait(ctx, min_complete, sig, sigsz);
+		    }
 		}
 	}
 
Signature · 1 line
1.8.3.1
maintainer
#1 lore raw
Quoted 5 lines · On 4/7/20 1:44 AM, [email protected] wrote:

From: Liming Wu <[email protected]>

When SETUP_IOPOLL and SETUP_SQPOLL are both enabled, app don't need to enter io_cqring_wait too. If I misunderstand, please give me some advise.

The logic should be as follows:

flags method
---------------------------------------------
0 io_cqring_wait()
IOPOLL io_iopoll_check()
IOPOLL | SQPOLL io_cqring_wait()
SQPOLL io_cqring_wait()

The reasoning being that we do want to enter cqring_wait() for SQPOLL, as the application may want to wait for completions. Even with IOPOLL set. As far as I can tell, the current code is correct, as long as we know SQPOLL will always poll for events for us.

So I'm curious why you think your patch is needed? Leaving it below and CC'ing Xiaoguang, who made the most recent change, so he can comment.

Quoted 28 lines

Signed-off-by Liming Wu <[email protected]>

---
 io_uring.c | 11 ++++++-----
 1 file changed, 6 insertions(+), 5 deletions(-)
io_uring.c +6 -5
index b12d33b..36e884f 100644
@@ -7418,11 +7418,12 @@ SYSCALL_DEFINE6(io_uring_enter, unsigned int, fd, u32, to_submit,
 		 * polling again, they can rely on io_sq_thread to do polling
 		 * work, which can reduce cpu usage and uring_lock contention.
 		 */
-		if (ctx->flags & IORING_SETUP_IOPOLL &&
-		    !(ctx->flags & IORING_SETUP_SQPOLL)) {
-			ret = io_iopoll_check(ctx, &nr_events, min_complete);
-		} else {
-			ret = io_cqring_wait(ctx, min_complete, sig, sigsz);
+		if (!(ctx->flags & IORING_SETUP_SQPOLL)) {
+		    if (ctx->flags & IORING_SETUP_IOPOLL) {
+		    	ret = io_iopoll_check(ctx, &nr_events, min_complete);
+		    } else {
+		    	ret = io_cqring_wait(ctx, min_complete, sig, sigsz);
+		    }
 		}
 	}
 
Signature · 1 line
Jens Axboe
#2 lore raw

hi,

Quoted 16 lines · On 4/7/20 1:44 AM, [email protected] wrote:
Quoted 5 lines

From: Liming Wu <[email protected]>

When SETUP_IOPOLL and SETUP_SQPOLL are both enabled, app don't need to enter io_cqring_wait too. If I misunderstand, please give me some advise.

The logic should be as follows:

flags method
---------------------------------------------
0 io_cqring_wait()
IOPOLL io_iopoll_check()
IOPOLL | SQPOLL io_cqring_wait()
SQPOLL io_cqring_wait()

Quoted 4 lines

The reasoning being that we do want to enter cqring_wait() for SQPOLL, as the application may want to wait for completions. Even with IOPOLL set. As far as I can tell, the current code is correct, as long as we know SQPOLL will always poll for events for us.

Yes, agree.

Regards,
Xiaoguang Wang

Quoted 34 lines

So I'm curious why you think your patch is needed? Leaving it below and CC'ing Xiaoguang, who made the most recent change, so he can comment.

Quoted 28 lines

Signed-off-by Liming Wu <[email protected]>

---
  io_uring.c | 11 ++++++-----
  1 file changed, 6 insertions(+), 5 deletions(-)
io_uring.c +6 -5
index b12d33b..36e884f 100644
@@ -7418,11 +7418,12 @@ SYSCALL_DEFINE6(io_uring_enter, unsigned int, fd, u32, to_submit,
  		 * polling again, they can rely on io_sq_thread to do polling
  		 * work, which can reduce cpu usage and uring_lock contention.
  		 */
-		if (ctx->flags & IORING_SETUP_IOPOLL &&
-		    !(ctx->flags & IORING_SETUP_SQPOLL)) {
-			ret = io_iopoll_check(ctx, &nr_events, min_complete);
-		} else {
-			ret = io_cqring_wait(ctx, min_complete, sig, sigsz);
+		if (!(ctx->flags & IORING_SETUP_SQPOLL)) {
+		    if (ctx->flags & IORING_SETUP_IOPOLL) {
+		    	ret = io_iopoll_check(ctx, &nr_events, min_complete);
+		    } else {
+		    	ret = io_cqring_wait(ctx, min_complete, sig, sigsz);
+		    }
  		}
  	}
 
#3 lore raw
In reply to [email protected]

hi,

Quoted 21 lines

From: Liming Wu <[email protected]>

When SETUP_IOPOLL and SETUP_SQPOLL are both enabled, app don't need to enter io_cqring_wait too. If I misunderstand, please give me some advise.

Signed-off-by Liming Wu <[email protected]>

---
  io_uring.c | 11 ++++++-----
  1 file changed, 6 insertions(+), 5 deletions(-)
io_uring.c +0 -2
index b12d33b..36e884f 100644
@@ -7418,11 +7418,12 @@ SYSCALL_DEFINE6(io_uring_enter, unsigned int, fd, u32, to_submit,
  		 * polling again, they can rely on io_sq_thread to do polling
  		 * work, which can reduce cpu usage and uring_lock contention.
  		 */
-		if (ctx->flags & IORING_SETUP_IOPOLL &&
-		    !(ctx->flags & IORING_SETUP_SQPOLL)) {

io_uring.c SYSCALL_DEFINE6(io_uring_enter, unsigned int, fd, u32, to_submit, · line 7427

Quoted 10 lines
io_uring.c +6 -3
-			ret = io_iopoll_check(ctx, &nr_events, min_complete);
-		} else {
-			ret = io_cqring_wait(ctx, min_complete, sig, sigsz);
+		if (!(ctx->flags & IORING_SETUP_SQPOLL)) {
+		    if (ctx->flags & IORING_SETUP_IOPOLL) {
+		    	ret = io_iopoll_check(ctx, &nr_events, min_complete);
+		    } else {
+		    	ret = io_cqring_wait(ctx, min_complete, sig, sigsz);
+		    }
  		}Indeed I had checked your patch yesterday, and also did not understand why you have

above modifications. For your codes, if IORING_SETUP_SQPOLL is enabed, we'll do nothing for IORING_ENTER_GETEVENTS, I think it's not correct.

In patch's commit message, I think you should explain more why you make such changes, what's the benefit? You can have a look at my previous patch:
io_uring: io_uring_enter(2) don't poll while SETUP_IOPOLL|SETUP_SQPOLL enabled

Finally, seems like that your patch did't pass checkpatch.pl, you should checkpatch before sending patches:)

Regards,
Xiaoguang Wang

Quoted 3 lines
io_uring.c +0 -0
  	}