Re: 2.6.22-rc5 regression

[Paul Mundt <lethal@linux-sh.org>]

On Mon, Jun 18, 2007 at 02:10:49AM +0200, Carlo Wood wrote:
> On Mon, Jun 18, 2007 at 08:18:58AM +0900, Paul Mundt wrote:
> > The obvious question of course is whether actually reverting this
> > changeset fixes your problem? I would be very surprised if it did.
> 
> Having a git commit Id, like d09c6b809432668371b5de9102f4f9aa6a7c79cc,
> how can I create a .diff file from that?
> 
$ git show d09c6b809432668371b5de9102f4f9aa6a7c79cc | patch -p1 -R
... build and test ...
$ git show d09c6b809432668371b5de9102f4f9aa6a7c79cc | patch -p1
... back to original (or git checkout-index -f mm/page_alloc.c) ...

if you just want to test that single change.

> I think git bisect made an error... as you can see for yourself
> with the list of tested git Id's that I gave - it didn't test
> both sides of that patch imho. If you tell me how to extract a patch
> then I'll manually unapply it to some failing kernel revisions and
> see if that makes a difference.
> 
Assuming that reverting that patch makes no difference, can you provide
the output of 'git bisect log'?