[PATCH] Remove XAA
alexdeucher at gmail.com
Mon Jan 23 06:26:12 PST 2012
On Mon, Jan 23, 2012 at 2:24 AM, Dave Airlie <airlied at gmail.com> wrote:
> On Sun, Jan 22, 2012 at 10:47 PM, Matt Turner <mattst88 at gmail.com> wrote:
>> On Sun, Jan 22, 2012 at 4:01 PM, Daniel Stone <daniel at fooishbar.org> wrote:
>>> On 22 January 2012 14:33, Mark Kettenis <mark.kettenis at xs4all.nl> wrote:
>>>> No way! XAA works for accelerated scrolling and makes a significant
>>>> difference on slow CPUs. It often works better for EXA. If there is
>>>> something that should be removed it's that since very few drivers actually
>>>> have working EXA support.
>>> 'very few drivers' ... which cover about 99% of our userbase if you
>>> just look at the big three. But anyway, even if you don't think any
>>> metric which gives equal weighting to xf86-video-intel as to
>>> xf86-video-imstt is absolutely pointless, there are 12 drivers that we
>>> ship ported to EXA.
>>>> Deliberately breaking drivers without giving people a chance to fix them
>>>> isn't fair. Because of the current development model of Xorg people may
>>>> not notice that stuff is broken for more than six months.
>>> If you read the mail you're replying to, you'll note that I
>>> specifically advocated holding off the merge so the drivers could be
>>> fixed. Although if no-one even notices for six months or more, I
>>> think that tells you quite a lot about the state of the driver, its
>>> maintenance, and its usage.
>> I like killing code, but I don't understand how killing the XAA code
>> helps us (other than killing code).
> Oh it helps me a lot, when you are trying to rework the X server ABI
> to be hotplug capable, the less useless ancient untestable pieces of
> crap you have to port makes the chance of users with modern hardware
> getting a modern feature everyone expects us to have had 10 years ago
>> Clearly they're not very important, but some hardware can't do EXA.
>> I'm not going to claim there are defensible reasons for using these
>> cards -- except perhaps in the cases of hardware like newport and the
>> SGI O2.
> But what do you run on these systems? maybe we need a mini-EXA that
> only accelerates upload to the front buffer, like shadowfb but is also
> DRI1 compatible. All XAA is doing for you is fast X core protocol
> operations on the front buffer, its not accelerating anything the
> current DE or apps use.
Maybe, but is it worth the effort? I mean we are talking about 10+
year old hardware here. We already don't have enough people to
support all the new stuff well. Is it really that onerous to run an
older xserver if you really want to keep using XAA?
More information about the xorg-devel