Re: [PATCH 0/1] exit: kill signal_struct->quick_threads
From: Eric W. Biederman
Date: Sat Jun 15 2024 - 10:54:26 EST
Oleg Nesterov <oleg@xxxxxxxxxx> writes:
> So...
>
> Eric, do you agree with this patch or not?
I really don't.
I think skipping some work if SIGNAL_GROUP_EXIT is already
set is not necessarily wrong.
I think we need the quick_threads count, and related cleanups.
I was hoping to be able to post a patchset with this reply
to explain things, but it looks like that is still a couple
of days off.
Where I was going, and where I think we should go with quick_threads is
an exit path that is much easier to reason about and maintain. But the
decrement needs to move into the same code that sets SIGNAL_GROUP_EXIT.
Which makes it something very different from signal->live.
Eric