2015-08-04 16:35:40

by Oleg Nesterov

Subject: [PATCH] net: pktgen: don't abuse current->state in pktgen_thread_worker()

Commit 1fbe4b46caca "net: pktgen: kill the Wait for kthread_stop
code in pktgen_thread_worker()" removed (in particular) the final
__set_current_state(TASK_RUNNING) and I didn't notice the previous
set_current_state(TASK_INTERRUPTIBLE). This triggers the warning
in __might_sleep() after return.

Afaics, we can simply remove both set_current_state()'s, and we
could do this a long ago right after ef87979c273a2 "pktgen: better
scheduler friendliness" which changed pktgen_thread_worker() to
use wait_event_interruptible_timeout().

Reported-by: Huang Ying <[email protected]>
Signed-off-by: Oleg Nesterov <[email protected]>
---
net/core/pktgen.c | 3 ---
1 files changed, 0 insertions(+), 3 deletions(-)

diff --git a/net/core/pktgen.c b/net/core/pktgen.c
index 8e0181a..6717684 100644
--- a/net/core/pktgen.c
+++ b/net/core/pktgen.c
@@ -3432,8 +3432,6 @@ static int pktgen_thread_worker(void *arg)

set_freezable();

- __set_current_state(TASK_RUNNING);
-
while (!kthread_should_stop()) {
pkt_dev = next_to_run(t);

@@ -3478,7 +3476,6 @@ static int pktgen_thread_worker(void *arg)

try_to_freeze();
}
- set_current_state(TASK_INTERRUPTIBLE);

pr_debug("%s stopping all device\n", t->tsk->comm);
pktgen_stop(t);
--
1.5.5.1

2015-08-04 16:41:00

by Marcelo Ricardo Leitner

Subject: Re: [PATCH] net: pktgen: don't abuse current->state in pktgen_thread_worker()

On Tue, Aug 04, 2015 at 06:33:34PM +0200, Oleg Nesterov wrote:
> Commit 1fbe4b46caca "net: pktgen: kill the Wait for kthread_stop
> code in pktgen_thread_worker()" removed (in particular) the final
> __set_current_state(TASK_RUNNING) and I didn't notice the previous
> set_current_state(TASK_INTERRUPTIBLE). This triggers the warning
> in __might_sleep() after return.
>
> Afaics, we can simply remove both set_current_state()'s, and we
> could do this a long ago right after ef87979c273a2 "pktgen: better
> scheduler friendliness" which changed pktgen_thread_worker() to
> use wait_event_interruptible_timeout().
>
> Reported-by: Huang Ying <[email protected]>
> Signed-off-by: Oleg Nesterov <[email protected]>

Interesting that it didn't happen in my tests yet per description it
should have. Huang, did you do anything special to trigger this?

2015-08-04 17:10:35

by Oleg Nesterov

Subject: Re: [PATCH] net: pktgen: don't abuse current->state in pktgen_thread_worker()

On 08/04, Marcelo Ricardo Leitner wrote:
>
> On Tue, Aug 04, 2015 at 06:33:34PM +0200, Oleg Nesterov wrote:
> > Commit 1fbe4b46caca "net: pktgen: kill the Wait for kthread_stop
> > code in pktgen_thread_worker()" removed (in particular) the final
> > __set_current_state(TASK_RUNNING) and I didn't notice the previous
> > set_current_state(TASK_INTERRUPTIBLE). This triggers the warning
> > in __might_sleep() after return.
> >
> > Afaics, we can simply remove both set_current_state()'s, and we
> > could do this a long ago right after ef87979c273a2 "pktgen: better
> > scheduler friendliness" which changed pktgen_thread_worker() to
> > use wait_event_interruptible_timeout().
> >
> > Reported-by: Huang Ying <[email protected]>
> > Signed-off-by: Oleg Nesterov <[email protected]>
>
> Interesting that it didn't happen in my tests yet per description it
> should have. Huang, did you do anything special to trigger this?

Do you have CONFIG_DEBUG_ATOMIC_SLEEP=y ?

Oleg.

2015-08-04 17:30:16

by Marcelo Ricardo Leitner

Subject: Re: [PATCH] net: pktgen: don't abuse current->state in pktgen_thread_worker()

On Tue, Aug 04, 2015 at 07:08:27PM +0200, Oleg Nesterov wrote:
> On 08/04, Marcelo Ricardo Leitner wrote:
> >
> > On Tue, Aug 04, 2015 at 06:33:34PM +0200, Oleg Nesterov wrote:
> > > Commit 1fbe4b46caca "net: pktgen: kill the Wait for kthread_stop
> > > code in pktgen_thread_worker()" removed (in particular) the final
> > > __set_current_state(TASK_RUNNING) and I didn't notice the previous
> > > set_current_state(TASK_INTERRUPTIBLE). This triggers the warning
> > > in __might_sleep() after return.
> > >
> > > Afaics, we can simply remove both set_current_state()'s, and we
> > > could do this a long ago right after ef87979c273a2 "pktgen: better
> > > scheduler friendliness" which changed pktgen_thread_worker() to
> > > use wait_event_interruptible_timeout().
> > >
> > > Reported-by: Huang Ying <[email protected]>
> > > Signed-off-by: Oleg Nesterov <[email protected]>
> >
> > Interesting that it didn't happen in my tests yet per description it
> > should have. Huang, did you do anything special to trigger this?
>
> Do you have CONFIG_DEBUG_ATOMIC_SLEEP=y ?

Ahh yes, probably not. Thanks.

Marcelo

2015-08-07 06:53:01

by David Miller

Subject: Re: [PATCH] net: pktgen: don't abuse current->state in pktgen_thread_worker()

From: Oleg Nesterov <[email protected]>
Date: Tue, 4 Aug 2015 18:33:34 +0200

> Commit 1fbe4b46caca "net: pktgen: kill the Wait for kthread_stop
> code in pktgen_thread_worker()" removed (in particular) the final
> __set_current_state(TASK_RUNNING) and I didn't notice the previous
> set_current_state(TASK_INTERRUPTIBLE). This triggers the warning
> in __might_sleep() after return.
>
> Afaics, we can simply remove both set_current_state()'s, and we
> could do this a long ago right after ef87979c273a2 "pktgen: better
> scheduler friendliness" which changed pktgen_thread_worker() to
> use wait_event_interruptible_timeout().
>
> Reported-by: Huang Ying <[email protected]>
> Signed-off-by: Oleg Nesterov <[email protected]>

Applied, thank you.