lkml.org 
[lkml]   [2009]   [Apr]   [14]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: [PATCH 0/3] v2 Make hierarchical RCU less IPI-happy and add more tracing

* Paul E. McKenney <paulmck@linux.vnet.ibm.com> wrote:

> This three-patch set fixes a hierarchical-RCU performance bug
> located by Anton Blanchard. This bug affects certain high-end
> floating-point computation-heavy workloads with closely spaced
> barrier-synchronized iterations, where an interruption of any one
> CPU's processing during a given iteration slows the system as a
> whole. This bug manifests itself as excessive resched IPIs, and
> is strictly a performance regression. It does not affect
> correctness.
>
> The first patch provides the fix for the problem, the second patch
> adds the tracing, and the third patch adds the documentation for
> the tracing.
>
> Located-by: Anton Blanchard <anton@au1.ibm.com>
> Tested-by: Anton Blanchard <anton@au1.ibm.com>
> Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>

i've applied the first patch to tip:core/urgent (for .30) and the
second and third one to tip:core/rcu (for .31).

Thanks Paul,

Ingo


\
 
 \ /
  Last update: 2009-04-14 11:37    [from the cache]
©2003-2011 Jasper Spaans