Re: [PATCH v2] exec: make de_thread() freezable

[Michal Hocko]

On Wed 14-11-18 19:18:42, Chanho Min wrote:
> > > > It's been some time since I have looked into this code so bear with
> me.
> > > > One thing is not really clear to me. Why does it help to exclude this
> > > > particular task from the freezer
> > >
> > > we don't exclude it,
> > >
> > > > when it is not sleeping in the freezer.
> > >
> > > Yes, it is not sleeping in __refrigerator(), but it does
> > >
> > > 	schedule();
> > > 	freezer_count();
> > >
> > > so it will enter __refrigerator() right after wakeup. If it won't be
> > woken
> > > up we do not care, we can consider it "frozen".
> > 
> > Right, but this is just silencing the freezing code to exclude this
> > task, right?
> > 
> > > > I can see how other threads need to be zapped and TASK_WAKEKILL
> > doesn't
> > > > do that but shouldn't we fix that instead?
> > >
> > > Not sure I understand, but unlikely we can (or want) to make
> > __refrigerator()
> > > killable.
> > 
> > Why would that be a problem. If the kill is fatal then why to keep the
> > killed task in the fridge?
> > 
> 
> Is it  different between 'the killed task is frozen' and '__refrigerator()
> is killable'?
> From a general '__refrigerator()' implementation point of view I know that
> it should not be killable.

Is that because there are many paths that do not terminate right after
the task get out of the fridge? Like the signal path?

> > > Otherwise, how can we fix that?
> > 
> > We can mark all threads PF_NOFREEZE and wake them up. This would require
> > some more changes of course but wouldn't that be a more appropriate
> > solution? Do we want to block exec for ever just because some threads
> > are in the fridge?
> > 
> 
> IMHO, It seems to be difficult and buggy to control with PF_NOFREEZE.
> Because,
> The sub-thread can freeze and receive SIG_KILL before the marking of
> PF_NOFREEZE
> and it should be freezable in other cases.

But we do control the ordering in this path no?

> I don't understand why it isn't appropriate for exec to block. The
> exec can freeze. When tasks are thawed, the killed sub-thread will die
> and wake de_thread(). The exec will continue to work from resume.

Because this is fragile. I haven't checked the full set of resources the
task holds when in this path but I can imagine we can introduce lock
dependency on freezing really easily.
-- 
Michal Hocko
SUSE Labs