Re: [RFC][PATCH] rcu classic: new algorithm for callbacks-processing

["Paul E. McKenney" <paulmck@linux.vnet.ibm.com>]

On Sun, Jul 06, 2008 at 05:23:47PM +0800, Lai Jiangshan wrote:
> Paul E. McKenney wrote:
> 
> > 
> > You tried this without your changes and it passed, correct?
> > 
> > Never forget to test the base kernel.  ;-)
> > 
> 
> Thanks, but I make some mistake here.
> 
> I tested the base kernel, no bug occurred.
> So I try to find out bugs in my changes in 2 weeks.
> 
> I added debug codes and these codes showed all processing
> are right and no callback broke its grace period
> but the bug still occurred.
> 
> When I was puzzled after 2 weeks, I looked back the base
> kernel that I had tested, and I found that I forget to set
> CONFIG_NO_HZ=y.
> 
> Then I set CONFIG_NO_HZ=y in the base kernel, and the bug
> occurred. It's exactly as the bug when my patch applied.
> 
> I have reported this bug here:
> http://lkml.org/lkml/2008/7/3/24

Hello, Jiangshan,

Does Gautham's patch (http://lkml.org/lkml/2008/6/27/9) help?

							Thanx, Paul