* [PATCH v1] net/dev.c : Remove redundant state settings after waking up
@ 2023-01-10 9:14 李哲
2023-01-10 9:29 ` Eric Dumazet
0 siblings, 1 reply; 8+ messages in thread
From: 李哲 @ 2023-01-10 9:14 UTC (permalink / raw)
To: davem, edumazet, kuba, pabeni, bigeasy, imagedong, kuniyu, petrm
Cc: netdev, linux-kernel, 李哲
the task status has been set to TASK_RUNNING in shcedule(),
no need to set again here
Signed-off-by: 李哲 <sensor1010@163.com>
---
net/core/dev.c | 1 -
1 file changed, 1 deletion(-)
diff --git a/net/core/dev.c b/net/core/dev.c
index b76fb37b381e..4bd2d4b954c9 100644
--- a/net/core/dev.c
+++ b/net/core/dev.c
@@ -6580,7 +6580,6 @@ static int napi_thread_wait(struct napi_struct *napi)
schedule();
/* woken being true indicates this thread owns this napi. */
woken = true;
- set_current_state(TASK_INTERRUPTIBLE);
}
__set_current_state(TASK_RUNNING);
--
2.17.1
^ permalink raw reply related [flat|nested] 8+ messages in thread
* Re: [PATCH v1] net/dev.c : Remove redundant state settings after waking up
2023-01-10 9:14 [PATCH v1] net/dev.c : Remove redundant state settings after waking up 李哲
@ 2023-01-10 9:29 ` Eric Dumazet
2023-01-11 0:30 ` Jakub Kicinski
0 siblings, 1 reply; 8+ messages in thread
From: Eric Dumazet @ 2023-01-10 9:29 UTC (permalink / raw)
To: 李哲, Wei Wang
Cc: davem, kuba, pabeni, bigeasy, imagedong, kuniyu, petrm, netdev,
linux-kernel
On Tue, Jan 10, 2023 at 10:15 AM 李哲 <sensor1010@163.com> wrote:
>
> the task status has been set to TASK_RUNNING in shcedule(),
> no need to set again here
Changelog is rather confusing, this does not match the patch, which
removes one set_current_state(TASK_INTERRUPTIBLE);
TASK_INTERRUPTIBLE != TASK_RUNNING
Patch itself looks okay (but has nothing to do with thread state after
schedule()),
you should have CC Wei Wang because she
authored commit cb038357937e net: fix race between napi kthread mode
and busy poll
>
> Signed-off-by: 李哲 <sensor1010@163.com>
> ---
> net/core/dev.c | 1 -
> 1 file changed, 1 deletion(-)
>
> diff --git a/net/core/dev.c b/net/core/dev.c
> index b76fb37b381e..4bd2d4b954c9 100644
> --- a/net/core/dev.c
> +++ b/net/core/dev.c
> @@ -6580,7 +6580,6 @@ static int napi_thread_wait(struct napi_struct *napi)
> schedule();
> /* woken being true indicates this thread owns this napi. */
> woken = true;
> - set_current_state(TASK_INTERRUPTIBLE);
> }
> __set_current_state(TASK_RUNNING);
>
> --
> 2.17.1
>
^ permalink raw reply [flat|nested] 8+ messages in thread
* Re: [PATCH v1] net/dev.c : Remove redundant state settings after waking up
2023-01-10 9:29 ` Eric Dumazet
@ 2023-01-11 0:30 ` Jakub Kicinski
2023-01-11 0:42 ` Wei Wang
0 siblings, 1 reply; 8+ messages in thread
From: Jakub Kicinski @ 2023-01-11 0:30 UTC (permalink / raw)
To: Eric Dumazet
Cc: 李哲,
Wei Wang, davem, pabeni, bigeasy, imagedong, kuniyu, petrm,
netdev, linux-kernel
On Tue, 10 Jan 2023 10:29:20 +0100 Eric Dumazet wrote:
> > the task status has been set to TASK_RUNNING in shcedule(),
> > no need to set again here
>
> Changelog is rather confusing, this does not match the patch, which
> removes one set_current_state(TASK_INTERRUPTIBLE);
>
> TASK_INTERRUPTIBLE != TASK_RUNNING
>
> Patch itself looks okay (but has nothing to do with thread state after
> schedule()),
> you should have CC Wei Wang because she
> authored commit cb038357937e net: fix race between napi kthread mode
> and busy poll
AFAIU this is the semi-idiomatic way of handling wait loops.
It's not schedule() that may set the task state to TASK_RUNNING,
it's whoever wakes the process and makes the "wait condition" true.
In this case - test_bit(NAPI_STATE_SCHED, &napi->state)
I vote to not futz with this logic.
^ permalink raw reply [flat|nested] 8+ messages in thread
* Re: [PATCH v1] net/dev.c : Remove redundant state settings after waking up
2023-01-11 0:30 ` Jakub Kicinski
@ 2023-01-11 0:42 ` Wei Wang
2023-01-11 7:32 ` Sebastian Andrzej Siewior
0 siblings, 1 reply; 8+ messages in thread
From: Wei Wang @ 2023-01-11 0:42 UTC (permalink / raw)
To: Jakub Kicinski
Cc: Eric Dumazet, 李哲,
davem, pabeni, bigeasy, imagedong, kuniyu, petrm, netdev,
linux-kernel
I was not able to see the entire changelog, but I don't think
> - set_current_state(TASK_INTERRUPTIBLE);
is redundant.
It makes sure that if the previous if statement:
if (test_bit(NAPI_STATE_SCHED_THREADED, &napi->state) || woken)
is false, this napi thread yields the CPU to other threads waiting to
be run by calling schedule().
And the napi thread gets into running state again after the next
wake_up_process() is called from ____napi_schedule().
On Tue, Jan 10, 2023 at 4:30 PM Jakub Kicinski <kuba@kernel.org> wrote:
>
> On Tue, 10 Jan 2023 10:29:20 +0100 Eric Dumazet wrote:
> > > the task status has been set to TASK_RUNNING in shcedule(),
> > > no need to set again here
> >
> > Changelog is rather confusing, this does not match the patch, which
> > removes one set_current_state(TASK_INTERRUPTIBLE);
> >
> > TASK_INTERRUPTIBLE != TASK_RUNNING
> >
> > Patch itself looks okay (but has nothing to do with thread state after
> > schedule()),
> > you should have CC Wei Wang because she
> > authored commit cb038357937e net: fix race between napi kthread mode
> > and busy poll
>
> AFAIU this is the semi-idiomatic way of handling wait loops.
> It's not schedule() that may set the task state to TASK_RUNNING,
> it's whoever wakes the process and makes the "wait condition" true.
> In this case - test_bit(NAPI_STATE_SCHED, &napi->state)
>
> I vote to not futz with this logic.
^ permalink raw reply [flat|nested] 8+ messages in thread
* Re: [PATCH v1] net/dev.c : Remove redundant state settings after waking up
2023-01-11 0:42 ` Wei Wang
@ 2023-01-11 7:32 ` Sebastian Andrzej Siewior
2023-01-11 18:20 ` Jakub Kicinski
0 siblings, 1 reply; 8+ messages in thread
From: Sebastian Andrzej Siewior @ 2023-01-11 7:32 UTC (permalink / raw)
To: Wei Wang
Cc: Jakub Kicinski, Eric Dumazet, 李哲,
davem, pabeni, imagedong, kuniyu, petrm, netdev, linux-kernel
On 2023-01-10 16:42:56 [-0800], Wei Wang wrote:
> I was not able to see the entire changelog, but I don't think
> > - set_current_state(TASK_INTERRUPTIBLE);
> is redundant.
>
> It makes sure that if the previous if statement:
> if (test_bit(NAPI_STATE_SCHED_THREADED, &napi->state) || woken)
> is false, this napi thread yields the CPU to other threads waiting to
> be run by calling schedule().
It made sense in the beginning but now the suggested patch is a clean
up. First the `woken' parameter was added in commit
cb038357937ee ("net: fix race between napi kthread mode and busy poll")
and then the `napi_disable_pending' check was removed in commit
27f0ad71699de ("net: fix hangup on napi_disable for threaded napi")
which renders the code to:
| while (!kthread_should_stop()) {
| if (test_bit(NAPI_STATE_SCHED_THREADED, &napi->state) || woken) {
| WARN_ON(!list_empty(&napi->poll_list));
| __set_current_state(TASK_RUNNING);
| return 0;
| }
|
| schedule();
| /* woken being true indicates this thread owns this napi. */
| woken = true;
| set_current_state(TASK_INTERRUPTIBLE);
| }
| __set_current_state(TASK_RUNNING);
so when you get out of schedule() woken is set and even if
NAPI_STATE_SCHED_THREADED is not set, the while() loop is left due to
`woken = true'. So changing state to TASK_INTERRUPTIBLE makes no sense
since it will be set back to TASK_RUNNING cycles later.
Sebastian
^ permalink raw reply [flat|nested] 8+ messages in thread
* Re: [PATCH v1] net/dev.c : Remove redundant state settings after waking up
2023-01-11 7:32 ` Sebastian Andrzej Siewior
@ 2023-01-11 18:20 ` Jakub Kicinski
2023-01-11 19:39 ` Wei Wang
0 siblings, 1 reply; 8+ messages in thread
From: Jakub Kicinski @ 2023-01-11 18:20 UTC (permalink / raw)
To: Sebastian Andrzej Siewior
Cc: Wei Wang, Eric Dumazet, 李哲,
davem, pabeni, imagedong, kuniyu, petrm, netdev, linux-kernel
On Wed, 11 Jan 2023 08:32:42 +0100 Sebastian Andrzej Siewior wrote:
> It made sense in the beginning but now the suggested patch is a clean
> up. First the `woken' parameter was added in commit
> cb038357937ee ("net: fix race between napi kthread mode and busy poll")
>
> and then the `napi_disable_pending' check was removed in commit
> 27f0ad71699de ("net: fix hangup on napi_disable for threaded napi")
>
> which renders the code to:
> | while (!kthread_should_stop()) {
> | if (test_bit(NAPI_STATE_SCHED_THREADED, &napi->state) || woken) {
> | WARN_ON(!list_empty(&napi->poll_list));
> | __set_current_state(TASK_RUNNING);
> | return 0;
> | }
> |
> | schedule();
> | /* woken being true indicates this thread owns this napi. */
> | woken = true;
> | set_current_state(TASK_INTERRUPTIBLE);
> | }
> | __set_current_state(TASK_RUNNING);
>
> so when you get out of schedule() woken is set and even if
> NAPI_STATE_SCHED_THREADED is not set, the while() loop is left due to
> `woken = true'. So changing state to TASK_INTERRUPTIBLE makes no sense
> since it will be set back to TASK_RUNNING cycles later.
Ah, fair point, forgot about the woken optimization.
^ permalink raw reply [flat|nested] 8+ messages in thread
* Re: [PATCH v1] net/dev.c : Remove redundant state settings after waking up
2023-01-11 18:20 ` Jakub Kicinski
@ 2023-01-11 19:39 ` Wei Wang
[not found] ` <706ea669.7a09.1869dce0851.Coremail.sensor1010@163.com>
0 siblings, 1 reply; 8+ messages in thread
From: Wei Wang @ 2023-01-11 19:39 UTC (permalink / raw)
To: Jakub Kicinski
Cc: Sebastian Andrzej Siewior, Eric Dumazet, 李哲,
davem, pabeni, imagedong, kuniyu, petrm, netdev, linux-kernel
On Wed, Jan 11, 2023 at 10:21 AM Jakub Kicinski <kuba@kernel.org> wrote:
>
> On Wed, 11 Jan 2023 08:32:42 +0100 Sebastian Andrzej Siewior wrote:
> > It made sense in the beginning but now the suggested patch is a clean
> > up. First the `woken' parameter was added in commit
> > cb038357937ee ("net: fix race between napi kthread mode and busy poll")
> >
> > and then the `napi_disable_pending' check was removed in commit
> > 27f0ad71699de ("net: fix hangup on napi_disable for threaded napi")
> >
> > which renders the code to:
> > | while (!kthread_should_stop()) {
> > | if (test_bit(NAPI_STATE_SCHED_THREADED, &napi->state) || woken) {
> > | WARN_ON(!list_empty(&napi->poll_list));
> > | __set_current_state(TASK_RUNNING);
> > | return 0;
> > | }
> > |
> > | schedule();
> > | /* woken being true indicates this thread owns this napi. */
> > | woken = true;
> > | set_current_state(TASK_INTERRUPTIBLE);
> > | }
> > | __set_current_state(TASK_RUNNING);
> >
> > so when you get out of schedule() woken is set and even if
> > NAPI_STATE_SCHED_THREADED is not set, the while() loop is left due to
> > `woken = true'. So changing state to TASK_INTERRUPTIBLE makes no sense
> > since it will be set back to TASK_RUNNING cycles later.
>
> Ah, fair point, forgot about the woken optimization.
Agree. I think it is OK to remove this, since woken is set, and this
function will set TASK_RUNNING and return 0. And the next time, it
will enter with TASK_INTERRUPTIBLE and woken = false.
^ permalink raw reply [flat|nested] 8+ messages in thread
* Re: [PATCH v1] net/dev.c : Remove redundant state settings after waking up
[not found] ` <706ea669.7a09.1869dce0851.Coremail.sensor1010@163.com>
@ 2023-03-01 17:07 ` Jakub Kicinski
0 siblings, 0 replies; 8+ messages in thread
From: Jakub Kicinski @ 2023-03-01 17:07 UTC (permalink / raw)
To: lizhe
Cc: Wei Wang, Sebastian Andrzej Siewior, Eric Dumazet, davem, pabeni,
imagedong, kuniyu, petrm, netdev, linux-kernel
On Wed, 1 Mar 2023 23:32:50 +0800 (CST) lizhe wrote:
> HI :
> if want to merge this patch into the main line, what should i do ?
Stop sending HTML emails please, the list only accepts plain text.
Read this:
https://www.kernel.org/doc/html/latest/process/maintainer-netdev.html
Improve the commit message and the subject line (look at other related
commits to find examples). Repost when appropriate (net-next is closed).
^ permalink raw reply [flat|nested] 8+ messages in thread
end of thread, other threads:[~2023-03-01 17:07 UTC | newest]
Thread overview: 8+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2023-01-10 9:14 [PATCH v1] net/dev.c : Remove redundant state settings after waking up 李哲
2023-01-10 9:29 ` Eric Dumazet
2023-01-11 0:30 ` Jakub Kicinski
2023-01-11 0:42 ` Wei Wang
2023-01-11 7:32 ` Sebastian Andrzej Siewior
2023-01-11 18:20 ` Jakub Kicinski
2023-01-11 19:39 ` Wei Wang
[not found] ` <706ea669.7a09.1869dce0851.Coremail.sensor1010@163.com>
2023-03-01 17:07 ` Jakub Kicinski
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.