lkml.org 
[lkml]   [2004]   [Nov]   [24]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: Suspend 2 merge: 22/51: Suspend2 lowlevel code.
On Wed, 24 Nov 2004, Nigel Cunningham wrote:

> + * SMP support:
> + * All SMP processors enter this routine during suspend. The one through
> + * which the suspend is initiated (which, for simplicity, is always CPU 0)
> + * sends the others here using an IPI during do_suspend2_suspend_1. They
> + * remain here until after the atomic copy of the kernel is made, to ensure
> + * that they don't mess with memory in the meantime (even just idling will
> + * do that). Once the atomic copy is made, they are free to carry on idling.
> + * Note that we must let them go, because if we're using compression, the
> + * vfree calls in the compressors will result in IPIs being called and hanging
> + * because the CPUs are still here.
> + *
> + * At resume time, we do a similar thing. CPU 0 sends the others in here using
> + * an IPI. It then copies the original kernel back, restores its own processor
> + * context and flushes local tlbs before freeing the others to do the same.
> + * They can then go back to idling while CPU 0 reloads pageset 2, cleans up
> + * and unfreezes the processes.
> + *
> + * (Remember that freezing and thawing processes also uses IPIs, as may
> + * decompressing the data. Again, therefore, we cannot leave the other processors
> + * in here).
> + *
> + * At the moment, we do nothing about APICs, even though the code is there.

Ok,
Do you see anything missing (from an implementation point of view)
for the following?

Suspend:
1) suspend all cpus, save cpu0
2) proceed with state saving on cpu0 only
3) begin suspend

Resume:
1) begin resume
2) offline all currently online cpus
3) proceed with state restoring
4) online all previously online cpus

A lot of the subsystems which have work split across cpus will now have
work migrated across to cpu0, in that regard, which have you made swsusp
savvy? It looks like the timer changes might need looking at any others?

Thanks,
Zwane
-
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/

\
 
 \ /
  Last update: 2005-03-22 14:08    [W:0.576 / U:0.080 seconds]
©2003-2018 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site