Re: [PATCH] io_uring/sqpoll: Increase task_work submission batch size

[Jens Axboe <axboe@kernel.dk>]

On 4/3/25 1:56 PM, Gabriel Krisman Bertazi wrote:
> diff --git a/io_uring/sqpoll.c b/io_uring/sqpoll.c
> index d037cc68e9d3..e58e4d2b3bde 100644
> --- a/io_uring/sqpoll.c
> +++ b/io_uring/sqpoll.c
> @@ -20,7 +20,7 @@
>  #include "sqpoll.h"
>  
>  #define IORING_SQPOLL_CAP_ENTRIES_VALUE 8
> -#define IORING_TW_CAP_ENTRIES_VALUE	8
> +#define IORING_TW_CAP_ENTRIES_VALUE	1024

That's a huge bump! This should not be a submission side thing, it's
purely running the task work. For this test case, I'm assuming you don't
see any io-wq activity, and hence everything is done purely inline from
the SQPOLL thread? This confuses me a bit, as this should not be driving
the queue depth at all, as submissions would be done by
__io_sq_thread(). And that part only caps when there is more than a
single ctx in there, which your case would not have. IOW, it should
submit everything that's there and hence this change should not change
the submission/queueing side of things. It only really deals with
running the task_work that will post the completion.

Maybe we should just not submit more until we've depleted the tw list?

In any case, we can _probably_ make this 32 or something without
worrying too much about it, though I would like to fully understand why
it's slower. Maybe it's the getrusage() that we do for every loop? You
could try and disable that just to see if it makes a difference?

-- 
Jens Axboe