Re: [PATCH 4/5] rcu: join rcu_ctrlblk and rcu_state

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

On Mon, Jan 09, 2006 at 04:43:17PM -0800, Linus Torvalds wrote:
> 
> On Mon, 9 Jan 2006, Paul E. McKenney wrote:
> > 
> > This patch looks sane to me.  It passes a short one-hour rcutorture
> > on ppc64 and x86, firing up some overnight runs as well.
> > 
> > Dipankar, Manfred, any other concerns?  Cacheline alignment?  (Seems
> > to me this code is far enough from the fastpath that this should not
> > be a problem, but thought I should ask.)
> 
> I'd ask you and Oleg to re-synchronize, and perhaps Oleg to re-send the 
> (part of?) the series that has no debate. I'm unsure, for example, whether 
> #2 was just to be dropped.

I believe that the original #2 is to be dropped, but that the patch Oleg
submitted in:

	http://marc.theaimsgroup.com/?l=linux-kernel&m=113681388600342&w=2

may be needed.  I have added Vatsa to the CC to get his take on this.

However, this patch should be independent from #4, so it should be OK
to apply an updated #4 first while we are working out what to do about #2.

> I already applied #1, and it looks like there's agreement on #3 and #4, 
> but basically, just to make sure, can Oleg please re-send to make sure I 
> got it right?
> 
> Getting a screwed-up RCU thing is going to be too painful to debug, so I'd 
> rather get it right the first time it hits my tree..

Been there more times than I care to admit, and I most definitely agree!!!

							Thanx, Paul