Re: [PATCH 1/3] sched: fix TASK_WAKEKILL vs SIGKILL race

[Oleg Nesterov <oleg@tv-sign.ru>]

On 06/09, Roland McGrath wrote:
>
   ^^^^^
Oops, something is wrong! I received this message today, 06/28.

> > Note this "__TASK_STOPPED | __TASK_TRACED" check in signal_pending_state().
> > This is needed to preserve the current behaviour (ptrace_notify). I hope
> > this check will be removed soon, but this (afaics good) change needs the
> > separate discussion.
> 
> Agreed.  I think it might actually already be safe to drop it, but we can
> get to that after this settles.

Great! I'll re-send the patch which drops it in a minute.

Could you also look at other patches?

	[PATCH 2/3] ptrace: never sleep in TASK_TRACED if SIGKILL'ed
	http://marc.info/?l=linux-kernel&m=121362601402886

	[PATCH 3/3] ptrace: kill may_ptrace_stop()
	http://marc.info/?l=linux-kernel&m=121362601702897

I think ptrace_stop() really needs changes. But I don't know what is
the supposed behaviour of PT_TRACE_EXIT. Should we sleep even if the
task was killed? Should we sleep if the thread was killed because another
one does exit_group() or exec() ?

We can use "SIGNAL_GROUP_EXIT && (group_exit_code & 0x7f)" instead
of signal_group_exit() to be sure that task was killed by the fatal
signal.

Oleg.