Messages in this thread | | | Date | Wed, 24 Sep 2014 10:36:38 +0200 | From | Peter Zijlstra <> | Subject | Re: [PATCH 09/11] net: Clean up sk_wait_event() vs might_sleep() |
| |
On Wed, Sep 24, 2014 at 10:18:54AM +0200, Peter Zijlstra wrote: > WARNING: CPU: 1 PID: 1744 at kernel/sched/core.c:7104 __might_sleep+0x58/0x90() > do not call blocking ops when !TASK_RUNNING; state=1 set at [<ffffffff81070e10>] prepare_to_wait+0x50 /0xa0 > > [<ffffffff8105bc38>] __might_sleep+0x58/0x90 > [<ffffffff8148c671>] lock_sock_nested+0x31/0xb0 > [<ffffffff81498aaa>] sk_stream_wait_memory+0x18a/0x2d0 > > Which is a false positive because sk_wait_event() will already have > TASK_RUNNING at that point if it would've gone through > schedule_timeout(). > > So annotate with fixup_sleep(); which goes away on !DEBUG builds. > > Cc: netdev@vger.kernel.org > Cc: David Miller <davem@davemloft.net> > Reported-by: Ilya Dryomov <ilya.dryomov@inktank.com> > Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Dave, FYI this patch relies on a previous one that introduces the fixup_sleep()
lkml.kernel.org/r/20140924082242.186408915@infradead.org
If you're ok with things, I'll merge this one along with the rest of things.
> --- > include/net/sock.h | 1 + > 1 file changed, 1 insertion(+) > > --- a/include/net/sock.h > +++ b/include/net/sock.h > @@ -902,6 +902,7 @@ static inline void sock_rps_reset_rxhash > if (!__rc) { \ > *(__timeo) = schedule_timeout(*(__timeo)); \ > } \ > + fixup_sleep(); \ > lock_sock(__sk); \ > __rc = __condition; \ > __rc; \ > >
| |