[PATCH weston] Revert "compositor-drm: disable hardware cursors"

[Giulio Camuffo]

2015-09-09 9:58 GMT+03:00 Pekka Paalanen <ppaalanen at gmail.com>:
> On Tue, 8 Sep 2015 22:52:51 -0700
> Bryce Harrington <bryce at osg.samsung.com> wrote:
>
>> On Mon, Sep 07, 2015 at 04:42:16PM +0300, Pekka Paalanen wrote:
>> > On Mon, 7 Sep 2015 15:36:35 +0200
>> > Giulio Camuffo <giuliocamuffo at gmail.com> wrote:
>> >
>> > > 2015-09-07 16:27 GMT+03:00 Pekka Paalanen <ppaalanen at gmail.com>:
>> > > > From: Pekka Paalanen <pekka.paalanen at collabora.co.uk>
>> > > >
>> > > > This reverts commit 6858383d51b12632481370fdc7d886a1e6bb4ebd.
>> > > >
>> > > > The main reason for the original patch was a regression on the Intel
>> > > > drivers causing compositor framerate to drop to half during cursor
>> > > > updates or motion. The Intel problem was fixed in
>> > > > https://github.com/torvalds/linux/commit/2e7f43c41c042d6fed4d67aceeaae32d8f102e98
>> > > > The fix is included in Linux 4.1-rc6 and 4.0.5.
>> > >
>> > > Yeah, it seems to be fixed indeed.
>> > >
>> > > Reviewed-by: Giulio Camuffo <giuliocamuffo at gmail.com>
>> >
>> > Bryce, would it be ok to get this in for -rc?
>>
>> Since this re-enables lower level functionality, this feels like that
>> type of change that potentially could reveal unexpected side effects
>> (particularly if no one's run with this change on the newer kernels
>> yet.)
>>
>> However, there are still a couple weeks ahead of us, so if there are
>> side effects there's likely a good chance of discovering them still
>> pre-release.
>>
>> So... if we can get at least one Tested-by, I'd be ok with landing this
>> for -rc, and just be attentive to any problems that crop up.
>
> I tested this while checking if cursor motion without a cursor plane
> causes extra texture uploads on gl-renderer (it does not, but
> weston_surface_damage() *is* still not right), but since it's my patch,
> another vote would be nice.
>
> I'm on a 3.15 kernel atm. so I haven't seen the original problem to
> begin with. So yeah, let's get a Tested-by.

I tried it on 4.1.6, it worked fine.

>
>
> Thanks,
> pq