Re: [PATCH] Revert "exec: make de_thread() freezable (was: Re: Linux 4.20-rc4)

[Michal Hocko]

On Mon 03-12-18 09:06:18, Linus Torvalds wrote:
> On Mon, Dec 3, 2018 at 6:17 AM Michal Hocko <mhocko@xxxxxxxxxx> wrote:
> >
> > This argument just doesn't make any sense. Rare bugs are maybe even more
> > annoying because you do not expect them to happen.
> 
> Absolutely.
> 
> And valid lockdep complaints are a real issue too.

No questions about that. But in this case there is no real deadlock so
the splat is disputable. And we have a "do not use" freezable_schedule_unsafe
which shouldn't tigger the splat yet it would make the suspend issue go
away AFIU. Not a great fix and we should target for something better
long term but good enough as a poor's man stop gap fix.

> So I don't think there's any question that this should be reverted,
> the only question is whether I take the revert directly or get it from
> the PM tree.
> 
> It does sound like the de_thread() behavior needs more work. Maybe,
> for example, we need to make it clear that zapped threads are *not*
> frozen, and instead the freezer will wait for them to exit?

I was actually proposing something like that during the review
discussion. Basically set PF_NOFREEZE all the threads and wake them up.
But that is not so simple because they might be at any path when they
sleep (this is the biggest design flaw of the freezer IMHO).

-- 
Michal Hocko
SUSE Labs