[Quantal][SRU] Updated i915 driver for HSW support

[Anca Emanuel]

Publish to an ppa ? or something else for people to test ?

On Mon, Dec 3, 2012 at 7:24 PM, Leann Ogasawara
<leann.ogasawara at canonical.com> wrote:
> On 12/03/2012 05:14 AM, Anca Emanuel wrote:
>> VGA fixes here:
>> http://permalink.gmane.org/gmane.comp.freedesktop.xorg.drivers.intel/16580
>
> Thanks for the heads up.  I've applied the 5 noted patches from that
> thread to the updated i915 driver I've proposed and tested.  I can
> confirm the suspend/resume issue I was seeing when testing over  VGA is
> resolved with this additional set of patches.  I've gone ahead and
> pushed them to my branch:
>
> git://kernel.ubuntu.com/ogasawara/ubuntu-quantal.git hsw-backport-3.6
>
> Thanks,
> Leann
>
>>
>> On Sat, Dec 1, 2012 at 9:13 AM, Leann Ogasawara
>> <leann.ogasawara at canonical.com> wrote:
>>> BugLink: http://bugs.launchpad.net/bugs/1085245
>>>
>>> == Quantal SRU Justification ==
>>>
>>> Beginning with the 12.04.2 point release, we are providing the 12.10
>>> kernel as an enablement vehicle.  Enablement for Haswell graphics has
>>> been a frequent request and highly sought after for the 12.04.2 point
>>> release coupled with the 12.10 kernel and X stack.  Full enablement
>>> support for Haswell will unfortunately not be available by default until
>>> the upstream v3.8 kernel.  Given our release time frames, waiting for
>>> the v3.8 kernel to land in an official Ubuntu release and propagate back
>>> into 12.04 will greatly delay our ability to support Haswell in the near
>>> future.  As an alternative solution, I'm am proposing we provide an
>>> updated i915 driver in Quantal with Haswell support.
>>>
>>> == Fix ==
>>> We can provide an updated snapshot of the i915 driver which supports
>>> Haswell.  This updated snapshot is also making it's way through the
>>> drm-intel tree and targeting to land in the upstream v3.8 kernel.  This
>>> updated version of the i915 driver being proposed here for the Quantal
>>> kernel will only support Haswell pci id's which should greatly reduce
>>> the risk of regression to non-Haswell hardware.  The following set of
>>> patches introduces this updated i915 driver as an Ubuntu driver (ie
>>> ubuntu-quantal/ubuntu/i915) and will be renamed to "i915_hsw".
>>>
>>> The updated i915 snapshot was taken from
>>> "git://people.freedesktop.org/~danvet/drm backport-3.6" which was
>>> specifically provided by Intel to help with backporting efforts.  It is
>>> a v3.6 based tree with primarily Haswell enablement patches applied on
>>> top.  It should be less volatile than a drm-intel-nightly branch which
>>> contains the same Haswell support.  The set of patches I'm proposing for
>>> the Quantal kernel can be reviewed at:
>>>
>>>   git://kernel.ubuntu.com/ogasawara/ubuntu-quantal.git hsw-backport-3.6
>>>
>>> I would note I did have to include some patches which touch outside the
>>> ubuntu/i915 space.  However, I hope most are straightforward enough to
>>> see they pose no risk of regression (ie they only introduce new
>>> #define's or functions which are only used by this newer driver).  I
>>> would also note I did need to also cherry-pick some drm/fb-helper
>>> patches which were listed in the backport-3.6 branch from Intel in order
>>> to resolve some VGA issues I was seeing while testing.  I hope those are
>>> also straightforward when reviewing.
>>>
>>> == Testing ==
>>> I've tested this on a Beta Shark Bay Desktop SDP replicating the same
>>> package combination we'll see with the 12.04.2 point release, ie Precise
>>> + Quantal X stack + Quantal test kernel.  I see significant improvements
>>> compared to a stock Quantal kernel.  With a stock Quantal kernel, the
>>> system is unusable and the monitor powers off 3secs into booting
>>> anyways.  With the proposed set of patches I'm able to successfully boot
>>> and connect a monitor over HDMI and VGA.  I also did some general smoke
>>> testing of suspend/resume, audio, ethernet, wifi, reboot/shutdown, 3D,
>>> and monitor hotplugging all of which passed with the exception of
>>> suspend/resume failures over VGA which I'm told is a known issue that is
>>> being addressed.  I have also tested this on all my non-Haswell hardware
>>> and can confirm I see no regressions.
>>>
>>> Given this set of patches should be well contained and really only
>>> affect Haswell hardware, I propose we try and pull it in for the
>>> upcoming Quantal kernel SRU cycle.  It will allow us to ensure we see no
>>> regressions on non-Haswell hardware but also allow us to start more
>>> widespread testing against the Haswell hardware we do have access to.
>>>
>>> Thanks,
>>> Leann
>>>
>>> --
>>> kernel-team mailing list
>>> kernel-team at lists.ubuntu.com
>>> https://lists.ubuntu.com/mailman/listinfo/kernel-team
>