* [PATCH 5.19] io_uring: fix req->apoll_events
@ 2022-06-21 12:25 Pavel Begunkov
2022-06-21 12:28 ` Pavel Begunkov
` (2 more replies)
0 siblings, 3 replies; 4+ messages in thread
From: Pavel Begunkov @ 2022-06-21 12:25 UTC (permalink / raw)
To: io-uring; +Cc: Jens Axboe, asml.silence
apoll_events should be set once in the beginning of poll arming just as
poll->events and not change after. However, currently io_uring resets it
on each __io_poll_execute() for no clear reason. There is also a place
in __io_arm_poll_handler() where we add EPOLLONESHOT to downgrade a
multishot, but forget to do the same thing with ->apoll_events, which is
buggy.
Fixes: 81459350d581e ("io_uring: cache req->apoll->events in req->cflags")
Signed-off-by: Pavel Begunkov <[email protected]>
---
fs/io_uring.c | 12 ++++++++----
1 file changed, 8 insertions(+), 4 deletions(-)
diff --git a/fs/io_uring.c b/fs/io_uring.c
index 87c65a358678..ebda9a565fc0 100644
--- a/fs/io_uring.c
+++ b/fs/io_uring.c
@@ -6954,7 +6954,8 @@ static void io_apoll_task_func(struct io_kiocb *req, bool *locked)
io_req_complete_failed(req, ret);
}
-static void __io_poll_execute(struct io_kiocb *req, int mask, __poll_t events)
+static void __io_poll_execute(struct io_kiocb *req, int mask,
+ __poll_t __maybe_unused events)
{
req->cqe.res = mask;
/*
@@ -6963,7 +6964,6 @@ static void __io_poll_execute(struct io_kiocb *req, int mask, __poll_t events)
* CPU. We want to avoid pulling in req->apoll->events for that
* case.
*/
- req->apoll_events = events;
if (req->opcode == IORING_OP_POLL_ADD)
req->io_task_work.func = io_poll_task_func;
else
@@ -7114,6 +7114,8 @@ static int __io_arm_poll_handler(struct io_kiocb *req,
io_init_poll_iocb(poll, mask, io_poll_wake);
poll->file = req->file;
+ req->apoll_events = poll->events;
+
ipt->pt._key = mask;
ipt->req = req;
ipt->error = 0;
@@ -7144,8 +7146,10 @@ static int __io_arm_poll_handler(struct io_kiocb *req,
if (mask) {
/* can't multishot if failed, just queue the event we've got */
- if (unlikely(ipt->error || !ipt->nr_entries))
+ if (unlikely(ipt->error || !ipt->nr_entries)) {
poll->events |= EPOLLONESHOT;
+ req->apoll_events |= EPOLLONESHOT;
+ }
__io_poll_execute(req, mask, poll->events);
return 0;
}
@@ -7392,7 +7396,7 @@ static int io_poll_add_prep(struct io_kiocb *req, const struct io_uring_sqe *sqe
return -EINVAL;
io_req_set_refcount(req);
- req->apoll_events = poll->events = io_poll_parse_events(sqe, flags);
+ poll->events = io_poll_parse_events(sqe, flags);
return 0;
}
--
2.36.1
^ permalink raw reply related [flat|nested] 4+ messages in thread
* Re: [PATCH 5.19] io_uring: fix req->apoll_events
2022-06-21 12:25 [PATCH 5.19] io_uring: fix req->apoll_events Pavel Begunkov
@ 2022-06-21 12:28 ` Pavel Begunkov
2022-06-21 13:28 ` Hao Xu
2022-06-21 13:49 ` Jens Axboe
2 siblings, 0 replies; 4+ messages in thread
From: Pavel Begunkov @ 2022-06-21 12:28 UTC (permalink / raw)
To: io-uring; +Cc: Jens Axboe
On 6/21/22 13:25, Pavel Begunkov wrote:
> apoll_events should be set once in the beginning of poll arming just as
> poll->events and not change after. However, currently io_uring resets it
> on each __io_poll_execute() for no clear reason. There is also a place
> in __io_arm_poll_handler() where we add EPOLLONESHOT to downgrade a
> multishot, but forget to do the same thing with ->apoll_events, which is
> buggy.
>
> Fixes: 81459350d581e ("io_uring: cache req->apoll->events in req->cflags")
> Signed-off-by: Pavel Begunkov <[email protected]>
> ---
> fs/io_uring.c | 12 ++++++++----
> 1 file changed, 8 insertions(+), 4 deletions(-)
>
> diff --git a/fs/io_uring.c b/fs/io_uring.c
> index 87c65a358678..ebda9a565fc0 100644
> --- a/fs/io_uring.c
> +++ b/fs/io_uring.c
> @@ -6954,7 +6954,8 @@ static void io_apoll_task_func(struct io_kiocb *req, bool *locked)
> io_req_complete_failed(req, ret);
> }
>
> -static void __io_poll_execute(struct io_kiocb *req, int mask, __poll_t events)
> +static void __io_poll_execute(struct io_kiocb *req, int mask,
> + __poll_t __maybe_unused events)
Killing @events would add extra rebase conflicts, so left it here and will
send a clean up for 5.20.
> {
> req->cqe.res = mask;
> /*
> @@ -6963,7 +6964,6 @@ static void __io_poll_execute(struct io_kiocb *req, int mask, __poll_t events)
> * CPU. We want to avoid pulling in req->apoll->events for that
> * case.
> */
> - req->apoll_events = events;
> if (req->opcode == IORING_OP_POLL_ADD)
> req->io_task_work.func = io_poll_task_func;
> else
> @@ -7114,6 +7114,8 @@ static int __io_arm_poll_handler(struct io_kiocb *req,
> io_init_poll_iocb(poll, mask, io_poll_wake);
> poll->file = req->file;
>
> + req->apoll_events = poll->events;
> +
> ipt->pt._key = mask;
> ipt->req = req;
> ipt->error = 0;
> @@ -7144,8 +7146,10 @@ static int __io_arm_poll_handler(struct io_kiocb *req,
>
> if (mask) {
> /* can't multishot if failed, just queue the event we've got */
> - if (unlikely(ipt->error || !ipt->nr_entries))
> + if (unlikely(ipt->error || !ipt->nr_entries)) {
> poll->events |= EPOLLONESHOT;
> + req->apoll_events |= EPOLLONESHOT;
> + }
> __io_poll_execute(req, mask, poll->events);
> return 0;
> }
> @@ -7392,7 +7396,7 @@ static int io_poll_add_prep(struct io_kiocb *req, const struct io_uring_sqe *sqe
> return -EINVAL;
>
> io_req_set_refcount(req);
> - req->apoll_events = poll->events = io_poll_parse_events(sqe, flags);
> + poll->events = io_poll_parse_events(sqe, flags);
> return 0;
> }
>
--
Pavel Begunkov
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: [PATCH 5.19] io_uring: fix req->apoll_events
2022-06-21 12:25 [PATCH 5.19] io_uring: fix req->apoll_events Pavel Begunkov
2022-06-21 12:28 ` Pavel Begunkov
@ 2022-06-21 13:28 ` Hao Xu
2022-06-21 13:49 ` Jens Axboe
2 siblings, 0 replies; 4+ messages in thread
From: Hao Xu @ 2022-06-21 13:28 UTC (permalink / raw)
To: Pavel Begunkov, io-uring; +Cc: Jens Axboe
On 6/21/22 20:25, Pavel Begunkov wrote:
> apoll_events should be set once in the beginning of poll arming just as
> poll->events and not change after. However, currently io_uring resets it
> on each __io_poll_execute() for no clear reason. There is also a place
> in __io_arm_poll_handler() where we add EPOLLONESHOT to downgrade a
> multishot, but forget to do the same thing with ->apoll_events, which is
> buggy.
>
> Fixes: 81459350d581e ("io_uring: cache req->apoll->events in req->cflags")
> Signed-off-by: Pavel Begunkov <[email protected]>
> ---
> fs/io_uring.c | 12 ++++++++----
> 1 file changed, 8 insertions(+), 4 deletions(-)
>
> diff --git a/fs/io_uring.c b/fs/io_uring.c
> index 87c65a358678..ebda9a565fc0 100644
> --- a/fs/io_uring.c
> +++ b/fs/io_uring.c
> @@ -6954,7 +6954,8 @@ static void io_apoll_task_func(struct io_kiocb *req, bool *locked)
> io_req_complete_failed(req, ret);
> }
>
> -static void __io_poll_execute(struct io_kiocb *req, int mask, __poll_t events)
> +static void __io_poll_execute(struct io_kiocb *req, int mask,
> + __poll_t __maybe_unused events)
> {
> req->cqe.res = mask;
> /*
> @@ -6963,7 +6964,6 @@ static void __io_poll_execute(struct io_kiocb *req, int mask, __poll_t events)
> * CPU. We want to avoid pulling in req->apoll->events for that
> * case.
> */
> - req->apoll_events = events;
> if (req->opcode == IORING_OP_POLL_ADD)
> req->io_task_work.func = io_poll_task_func;
> else
> @@ -7114,6 +7114,8 @@ static int __io_arm_poll_handler(struct io_kiocb *req,
> io_init_poll_iocb(poll, mask, io_poll_wake);
> poll->file = req->file;
>
> + req->apoll_events = poll->events;
> +
> ipt->pt._key = mask;
> ipt->req = req;
> ipt->error = 0;
> @@ -7144,8 +7146,10 @@ static int __io_arm_poll_handler(struct io_kiocb *req,
>
> if (mask) {
> /* can't multishot if failed, just queue the event we've got */
> - if (unlikely(ipt->error || !ipt->nr_entries))
> + if (unlikely(ipt->error || !ipt->nr_entries)) {
> poll->events |= EPOLLONESHOT;
> + req->apoll_events |= EPOLLONESHOT;
> + }
> __io_poll_execute(req, mask, poll->events);
> return 0;
> }
> @@ -7392,7 +7396,7 @@ static int io_poll_add_prep(struct io_kiocb *req, const struct io_uring_sqe *sqe
> return -EINVAL;
>
> io_req_set_refcount(req);
> - req->apoll_events = poll->events = io_poll_parse_events(sqe, flags);
> + poll->events = io_poll_parse_events(sqe, flags);
> return 0;
> }
>
Make sense,
Reviewed-by: Hao Xu <[email protected]>
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: [PATCH 5.19] io_uring: fix req->apoll_events
2022-06-21 12:25 [PATCH 5.19] io_uring: fix req->apoll_events Pavel Begunkov
2022-06-21 12:28 ` Pavel Begunkov
2022-06-21 13:28 ` Hao Xu
@ 2022-06-21 13:49 ` Jens Axboe
2 siblings, 0 replies; 4+ messages in thread
From: Jens Axboe @ 2022-06-21 13:49 UTC (permalink / raw)
To: io-uring, asml.silence
On Tue, 21 Jun 2022 13:25:06 +0100, Pavel Begunkov wrote:
> apoll_events should be set once in the beginning of poll arming just as
> poll->events and not change after. However, currently io_uring resets it
> on each __io_poll_execute() for no clear reason. There is also a place
> in __io_arm_poll_handler() where we add EPOLLONESHOT to downgrade a
> multishot, but forget to do the same thing with ->apoll_events, which is
> buggy.
>
> [...]
Applied, thanks!
[1/1] io_uring: fix req->apoll_events
commit: aacf2f9f382c91df73f33317e28a4c34c8038986
Best regards,
--
Jens Axboe
^ permalink raw reply [flat|nested] 4+ messages in thread
end of thread, other threads:[~2022-06-21 13:49 UTC | newest]
Thread overview: 4+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2022-06-21 12:25 [PATCH 5.19] io_uring: fix req->apoll_events Pavel Begunkov
2022-06-21 12:28 ` Pavel Begunkov
2022-06-21 13:28 ` Hao Xu
2022-06-21 13:49 ` Jens Axboe
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox