Re: [PATCH 2.6.19 5/5] fs: freeze_bdev with semaphore not mutex

[Pavel Machek <pavel@ucw.cz>]

Hi!

> > > > > This is from a work queue, so in fact from a process context, but from
> > > > > a process that is running with PF_NOFREEZE.
> > > > 
> > > > Why not simply &~ PF_NOFREEZE on that particular process? Filesystems
> > > > are free to use threads/work queues/whatever, but refrigerator should
> > > > mean "no writes to filesystem" for them...
> > > 
> > > But how we differentiate worker_threads used by filesystems from the
> > > other ones?
> > 
> > I'd expect filesystems to do &~ PF_NOFREEZE by hand.
> > 
> > > BTW, I think that worker_threads run with PF_NOFREEZE for a reason,
> > > but what exactly is it?
> > 
> > I do not think we had particulary good reasons...
> 
> Well, it looks like quite a lot of drivers depend on them, including libata.
> 
> I think we can add a flag to __create_workqueue() that will indicate if
> this one is to be running with PF_NOFREEZE and a corresponding macro like
> create_freezable_workqueue() to be used wherever we want the worker thread
> to freeze (in which case it should be calling try_to_freeze() somewhere).
> Then, we can teach filesystems to use this macro instead of
> create_workqueue().

Works for me.

> Having done that we'd be able to drop the freezing of bdevs patch and forget
> about the dm-related complexity.

yes.

> [Still I wonder if the sys_sync() in freeze_processes() is actually safe if
> there's a suspended dm device somewhere in the stack, because in the other
> case the freezing of bdevs would be no more dangerous than the thing
> that we're already doing.]

? Not sure if I quite understand, but if dm breaks sync... something
is teribly wrong with dm. And we do simple sys_sync()... so I do not
think we have a problem.

									Pavel
-- 
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html