Re: [PATCH] fix de_thread() vs do_coredump() deadlock

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

On 04/11, Roland McGrath wrote:
>
> > So, de_thread() sets SIGNAL_GROUP_EXEC and sends SIGKILL to other thereads.
> > 
> > Sub-thread receives the signal, and calls get_signal_to_deliver->do_group_exit.
> > do_group_exit() calls zap_other_threads(SIGNAL_GROUP_EXIT) because there is no
> > SIGNAL_GROUP_EXIT set. zap_other_threads() notices SIGNAL_GROUP_EXEC, wakes up
> > execer, and changes ->signal->flags to SIGNAL_GROUP_EXIT.
> > 
> > de_thread() re-locks sighand, sees !SIGNAL_GROUP_EXEC and goes to 'dying:'.
> 
> That is what I intend.  The exec'ing thread backs out and processes its SIGKILL.
> It sounds like you are calling this scenario a problem, but I don't know why.

Oh, probably I missed something. But as I can see it, MT exec can never succeed!

Once again. Process starts exec, it has no pending signals. Execer thread sets
SIGNAL_GROUP_EXEC, sends SIGKILL to other threads, and waits them to die.
The first thread which dequeues SIGKILL will change SIGNAL_GROUP_EXEC to
SIGNAL_GROUP_EXIT, thus aborting exec.

Sorry for persistance if I really misunderstand this patch.

Oleg.