Re: [RFC PATCH 0/3] Refcount Based Cpu Hotplug V3

[Ingo Molnar <mingo@elte.hu>]

* Gautham R Shenoy <ego@in.ibm.com> wrote:

> Hi, 
> This is Try #3 for the Refcount + Waitqueue based implementation
> for cpu-hotplug locking. The earlier versions can be found at
> 
> http://lkml.org/lkml/2007/10/24/36
> and
> http://lkml.org/lkml/2007/10/16/118.
> 
> This version drops the patch 4 from the earlier series, which was 
> basically removing the CPU_DEAD and CPU_UP_CANCELLED event handling in 
> the workqueue subsystem. This was based on Oleg Nestertov's comments 
> since the problem which the patch was attempting to solve was more 
> complicated and needs to be looked into seperately.
> 
> The patch series is against v2.6.24-rc2 and has survived 
> overnight kern_bench + cpu-offline-online tests on 
> - 8 way x86, 
> - 8 way x86_64,
> - 128 way ppc64 boxes.
> 
> The patch series has gone through many iterations and looks to have 
> matured in the process.
> 
> Andrew, could you please consider it for the next -mm release if there 
> are no objections ?

really nice stuff!

Acked-by: Ingo Molnar <mingo@elte.hu>

since this is in essence a regression fix, how about a v2.6.24 merge? 
CPU online/offline never worked really reliably and it's used by 
suspend/resume as wel.

	Ingo