Re: Linux 3.4-rc4

From: Nick Bowler <nbowler@elliptictech.com>
To: Linus Torvalds <torvalds@linux-foundation.org>
Cc: David Airlie <airlied@linux.ie>, Ben Skeggs <bskeggs@redhat.com>,
	Martin Peres <martin.peres@labri.fr>,
	dri-devel@lists.freedesktop.org,
	Linux Kernel Mailing List <linux-kernel@vger.kernel.org>
Subject: Re: Linux 3.4-rc4
Date: Sun, 22 Apr 2012 20:05:54 -0400	[thread overview]
Message-ID: <20120423000554.GA893@elliptictech.com> (raw)
In-Reply-To: <20120422164023.GA32342@elliptictech.com>

On 2012-04-22 12:40 -0400, Nick Bowler wrote:
> On 2012-04-21 21:51 -0700, Linus Torvalds wrote:
> > Nick, I realize you had trouble with a bisection already, but it might
> > really be worth trying again. Do a
> > 
> >     git bisect visualize
> > 
> > and try to pick a good commit (avoding the problems you hit) when you
> > hit a problem, and then do
> > 
> >    git reset --hard <that-point>
> > 
> > to force bisection to try another place. That way you can sometimes
> > avoid the problem spots, and continue the bisection.
> 
> Unfortunately, I think the whole swath of commits bisect wants to test
> are broken (as in, they panic before I get to see whether or not the VGA
> is working), because the commit from which most of the drm trees were
> based appears to be broken.  Nevertheless, I've included the new bisect
> log (four new commits marked skip as opposed to last time).  I've also
> included the boot log from a crashing kernel, in case someone recognizes
> how I can avoid this during bisection.  Note that this crash is *not* a
> regression that exists in current mainline -- bisecting this issue was
> the first time I had ever seen it.

Following up on the above, the commit which introduces the panics during
boot is this one:

commit 8e7e70522d760c4ccd4cd370ebfa0ba69e006c6e
Author: Jerome Glisse <jglisse@redhat.com>
Date:   Wed Nov 9 17:15:26 2011 -0500

    drm/ttm: isolate dma data from ttm_tt V4

    Move dma data to a superset ttm_dma_tt structure which herit
    from ttm_tt. This allow driver that don't use dma functionalities
    to not have to waste memory for it.

    V2 Rebase on top of no memory account changes (where/when is my
       delorean when i need it ?)
    V3 Make sure page list is initialized empty
    V4 typo/syntax fixes

    Signed-off-by: Jerome Glisse <jglisse@redhat.com>
    Reviewed-by: Thomas Hellstrom <thellstrom@vmware.com>

and the previous commit (3230cfc34fca: "drm/nouveau: enable the ttm dma
pool when swiotlb is active V3") works properly.

Sometime this week I suppose I'll try to track down the commit which
fixed the crashes...

Cheers,
-- 
Nick Bowler, Elliptic Technologies (http://www.elliptictech.com/)