Re: Patch for lost wakeups

[Oleg Nesterov <oleg@redhat.com>]

On 08/08, Linus Torvalds wrote:
>
> > Name:    Xorg
> > State:    S (sleeping)
> > Tgid:    2597
> > Pid:    2597
> > PPid:    2595
> > TracerPid:    0
> > Uid:    0    0    0    0
> > Gid:    0    0    0    0
> > FDSize:    64
> > Groups:
> > VmPeak:       44640 kB
> > VmSize:       31232 kB
> > VmLck:           0 kB
> > VmHWM:       20560 kB
> > VmRSS:       20016 kB
> > VmData:        5728 kB
> > VmStk:         160 kB
> > VmExe:        1952 kB
> > VmLib:       11296 kB
> > VmPTE:         128 kB
> > VmSwap:           0 kB
> > Threads:    1
> > SigQ:    1/15809
> > SigPnd:    00000000000000000000000000000000
> > ShdPnd:    00000000000000000000000000200000
> > SigBlk:    00000000000000000000000000000000
> > SigIgn:    80000000000000000000000006001000
> > SigCgt:    000000000000000000000001e020eecf

OK, given that Threads == 1 this is wrong in any case.

Could you please reproduce with the trivial patch below ? (please also
should /proc/pid/stack).

If (with this patch) SigQ: should "1" at the end, then we have a scheduler
race or someone does set_tsk_thread_flag(non-current-task, TIF_SIGPENDING)
without locking/wakeup.

Otherwise we should identify the wrong clear_flag(TIF_SIGPENDING).

Oleg.


diff --git a/fs/proc/array.c b/fs/proc/array.c
index cbd0f1b..d4a8cbb 100644
--- a/fs/proc/array.c
+++ b/fs/proc/array.c
@@ -286,7 +286,7 @@ static inline void task_sig(struct seq_file *m, struct task_struct *p)
 	}
 
 	seq_printf(m, "Threads:\t%d\n", num_threads);
-	seq_printf(m, "SigQ:\t%lu/%lu\n", qsize, qlim);
+	seq_printf(m, "SigQ:\t%lu/%lu %d\n", qsize, qlim, !!signal_pending(p));
 
 	/* render them all */
 	render_sigset_t(m, "SigPnd:\t", &pending);