Pulling in fix for bug #140713 for Gutsy release

[Steve Langasek]

On Fri, Oct 12, 2007 at 11:09:55AM -0500, Anthony Liguori wrote:
> Matt Zimmerman wrote:
> > It's up to the release managers whether this is worthwhile, though with
> > gfxboot being such a critical and hardware-dependent component, making a
> > change like this seems awfully risky at the 11th hour.  It simply can't be
> > properly tested at this point.

> Here's a larger chunk of the code:

<snip>

> It's not are to see from inspection that the new code is correct.

Unfortunately it is for me; I don't speak x86 asm well enough for the
correctness to be obvious to me, and as Matt points out there's really no
way to get this adequately tested with the change landing after the RC.

If another member of the release team were able to vouch for the
self-evident correctness of the fix I would be content for them to allow it
in, but otherwise I'm afraid my own ignorance here prevents me from
accepting the fix this late, sorry.