[RFC] [Maverick] e1000e driver update to support 82579 LOM's

[Tim Gardner]

On 08/31/2010 08:20 PM, Leann Ogasawara wrote:
> Hi All,
>
> It's been requested that we update the e1000e driver in Maverick to
> enable support for 82579 LOM's [1].  The patches to enable support
> already exist upstream and are included as of 2.6.36-rc1.  I was advised
> however to *not* just cherry-pick the two patches [2] which I originally
> felt were necessary.  The reason being is that testing and validation
> from the original patch authors is done against the tip of the stack
> from which the patches are applied.  Thus by only cherry-picking a few
> patches in between, it invalidates any of their testing and they can't
> guarantee the resulting driver we then ship in Maverick is "approved".
> This could then have unintended consequences where we've inadvertently
> introduced issues which would otherwise not have existed. It was advised
> to instead backport ~20 patches between v2.6.35 and v2.6.36-rc1 for the
> e1000e driver.
>
> I've taken the liberty to backport these 20 patches to our Maverick
> kernel [3].  There was only 1 minor conflict that needed fixing up, the
> rest were clean cherry-picks.  I've subsequently built a test kernel and
> verified this indeed enables support for an 82579 LOM.  Additionally,
> I've tested this kernel on an older system which also leverages the
> e1000e driver and confirmed there were no regressions that I was able to
> see.
>
> I'd like to get the team's feedback if they feel applying these 20
> patches to the main distro Maverick linux kernel tree seems reasonable
> or if I should instead pursue enabling this LOM through
> linux-backports-modules.
>
> Thanks in advance,
> Leann
>
> [1] https://bugs.launchpad.net/ubuntu/+bug/601044
> [2] d3738bb8203acf8552c3ec8b3447133fc0938ddd e1000e: initial support for 82579 LOMs
>      96f2bd13bfb6df5beec7fe55405ad94b528b8b4c e1000e: correct MAC-PHY interconnect register offset for 82579
> [3] git://kernel.ubuntu.com/ogasawara/ubuntu-maverick.git lp601044-all
>
>

I'm thinking the best route for this driver update is via LBM once 
2.6.36 is final. We could certainly enable the 82579 LOM with the 2 
patches you've listed above, but a full scale backport ought to go into LBM.

rtg
-- 
Tim Gardner tim.gardner at canonical.com