[prev in list] [next in list] [prev in thread] [next in thread] 

List:       linux-smp
Subject:    Re: help for sw suspend needed
From:       MOLNAR Ingo <mingo () chiara ! csoma ! elte ! hu>
Date:       1999-01-28 15:52:09
[Download RAW message or body]


On Thu, 28 Jan 1999, Gabor Kuti wrote:

> Launch as many kernel threads as many processors I have. With every thread
> catch a processor and wait for a common flag (what means we have resumed).
> With processor #0 I do the suspend thing..

this looks like the cleanest solution. Dont make kernel threads
unrunnable, it might be that you need certain threads for the suspension
to function properly. This solution also has the benefit that there are no
extra branches anywhere in any 'normal' kernel code. (thus no overhead)
Those 'extra' threads are all sw-suspend related. 

> One more question: should resuming run on the same processor as
> suspending? (which are one code flow (process)) from the kernel
> (scheduling) view..

i would be surprised, but anyway it's i think wise to do it on the 'boot'
CPU (which is (smp_processor() == 0)). Also, make sure all other CPUs are
able to handle interrupts, even if they are 'stopped' on that lock.

-- mingo

-
Linux SMP list: FIRST see FAQ at http://www.irisa.fr/prive/mentre/smp-faq/
To Unsubscribe: send "unsubscribe linux-smp" to majordomo@vger.rutgers.edu

[prev in list] [next in list] [prev in thread] [next in thread]