Re: [PATCH 3.12 25/91] x86/setup: Extend low identity map to cover whole kernel range

[Luis Henriques]

 [ Adding Greg and Kamal ]

On Wed, Jan 06, 2016 at 01:31:55PM +0000, Matt Fleming wrote:
> On Wed, 06 Jan, at 11:24:55AM, Luis Henriques wrote:
> > On Wed, Jan 06, 2016 at 11:00:31AM +0000, Matt Fleming wrote:
> > > On Wed, 06 Jan, at 11:47:20AM, Paolo Bonzini wrote:
> > > > 
> > > > Without testing the problematic scenario explicitly (32-bit UEFI
> > > > kernel), I think this patch and 26/91 should not be backported to
> > > > kernels that do not have 23a0d4e8fa6d.
> > > 
> > > I tend to agree.
> > 
> > I can see these 2 commits in kernels as old as 3.10 (which definitely do
> > not include 23a0d4e8fa6d).  Does this mean these should be reverted from
> > stable kernels that already include these patches?  Or would you rather
> > recommend to backport 23a0d4e8fa6d?
> 
> That depends on your appetite for risk ;-)
>

Heh, I guess stable kernels aren't really about appetite for risk :-)

> 23a0d4e8fa6d does fix a legitimate bug, albeit one that no one seems
> to have ever hit. Personally, I'd go for backporting 23a0d4e8fa6d.

This commit doesn't seem to be too bad to backport.  I'm attaching 2
backports:

 - one is for the 3.16 stable kernel,
 - the other can be applied to 3.10, 3.12 and 3.13

(For the other kernels, I believe 23a0d4e8fa6d will be a clean
cherry-pick.)

Cheers,
--
Luís