[Xorg-driver-geode] RandR code merged into main tree

[Martin-Éric Racine]

On Tue, Sep 16, 2008 at 6:30 PM, Jordan Crouse <jordan.crouse at amd.com> wrote:
> On 16/09/08 17:38 +0300, Martin-Éric Racine wrote:
>> On Sat, Sep 13, 2008 at 2:27 AM, Jordan Crouse <jordan.crouse at amd.com> wrote:
>> > On 11/09/08 23:17 +0300, Martin-Éric Racine wrote:
>> >> On Thu, Sep 11, 2008 at 2:36 AM, Jordan Crouse <jordan.crouse at amd.com> wrote:
>> >> > We are only 5 days away from the expected release date, and since
>> >> > I haven't heard any problems with the RandR 1.2 changes, I have
>> >> > merged them into the main tree.  The code is confirmed working for
>> >> > X 1.5 vanilla, Hardy and Lenny.  I don't have any other operating
>> >> > systems handy. Please test.
>> >>
>> >> On Ubuntu/Hardy (X 1.4.1):
>> >>
>> >> Impeccable operation using 8.04.1 on LTSP, in which an updated GEODE
>> >> is installed, followed by rebuilding the LTSP squashfs image. fbcon
>> >> without fb. Stock kernel.
>> >>
>> >> On Debian/Lenny (X 1.4.2):
>> >>
>> >> Switching back and forth between a vcons and X produces a situation
>> >> where the default X-shaped mouse pointer glyph remains in the middle
>> >> of the screen, while the real pointer configured by GTK2 preferences
>> >> moves with the mouse. fbcon with lxfb. Custom kernel based on stock
>> >> Debian kernel, but with support for non-Geode peripherals removed.
>> >>
>> >> On Ubuntu/Hardy (X 1.5):
>> >>
>> >> Square box around the mouse pointer glyph when X initially launches.
>> >> fbcon wihout fb. Stock kernel.
>> >
>> > That sounds like alpha cursor problems to me.  can you try the attached
>> > patch and see if it makes a difference?  If so, then I'll clean it up
>> > and add it to the driver.
>>
>> On both Debian/Lenny and Ubuntu/Intrepid, it apparently offsets the
>> perceived location of the mouse pointer (glyph location versus pointer
>> physical location) by one vertical GNOME menu row and it makes the X
>> server crash at random after just a few minutes. However, the garbage
>> around the pointer is indeed gone in both cases.
>
> Okay - good to know.  Crash logs would be great if you get a chance.
>
> I'll look at this next week.  Needless to say, the release will be delayed
> by about a week or so, unless you want to release what we have now for
> Hardy and hope to roll it quickly enough for Intrepid.

Hardy only accepts critical bugfixes and/or support for new hardware.
Final OLPC support and DDC fixes for CS5535 barely qualified.

Intrepid is already in feature freeze and my best guess is that
support for Randr 1.2 won't be given much priority.  Ditto for Lenny.

In other words, RandR 1.2 can wait until we get it right.

-- 
Martin-Éric Racine
http://q-funk.iki.fi