Re: suspend2 merge (was Re: [Suspend2-devel] Re: CFS and suspend2: hang in atomic copy)

[Pavel Machek <pavel@ucw.cz>]

Hi!

> > Not the same... but they are still related. "freeze" (for atomic
> > snapshot) is actually subset of "suspend"... freeze needs DMAs off and
> > saved state, and you need DMAs off and saved state for "suspend".
> 
> THEY HAVE ABSOLUTELY NOTHING IN COMMON!
> 
> Nobody in their right mind thinks that "disable DMA" and "suspend" are 
> similar operations. 
> 
> > So it is actually correct to do "suspend" when you want "freeze"; it
> > is just slow. That's why they only differ in parameter these days.
> 
> It is *not* correct to "suspend" when you want "freeze".

Example?

> I don't understand how you can even *claim* something like that.
> 
> Here's a trivial example:
>  - SCSI disk
> 
> Tell me, what does "suspend" do, and what does "freeze" (snapshot) do?

Suspend syncs caches/spins down. Freeze does not do anything.

That's okay, I keep claiming "freeze" is subset of "suspend". Can you
name device where that is not true?

Remember we do

suspend(PMSG_FREEZE)
atomic snapshot
resume()
write snapshot.

So if we do spin the scsi disk down, nothing really bad happens, we'll
just spin it up. (So scsi disk is not example I want. Spining down
scsi disk on freeze is slow and stupid, but it is not incorrect).

Yes, If I'd knew what I know now, drivers would have
suspend/freeze/thaw/resume methods. We probably still can do that
change. Unfortunately, it needs driver authors to understand 4 hooks
(not 2) and do the right thing.
								Pavel
-- 
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/