Re: netbsd-7 on intel GPU (improvements and regressions)

[Robert Swindells <rjs%fdy2.co.uk@localhost>]

Manuel Bouyer wrote:
>On Fri, Jan 16, 2015 at 01:10:59PM +0100, Manuel Bouyer wrote:
>> On Thu, Jan 15, 2015 at 11:38:05AM +1100, matthew green wrote:
>> > 
>> > there seems to be an i386 GL problem, it was introduced after
>> > the mesa 10 upgrade, i'm fairly sure, but i have not looked
>> > closely at it yet.  it occurs on radeon as well.
>> 
>> At last I found why it happens on i386 and not amd64; the attached
>> patch makes it work on i386 for me.
>> On amd64 version, an assembly version of the functions in
>> dist/src/mapi/entry.c is used only when GLX_USE_TLS is defined.
>> On i386, there are assembly versions for both GLX_USE_TLS and !GLX_USE_TLS
>> cases. I guess the assembly version is wrong, but I've not looked closely.
>> The attached patch makes i386 use the C version as amd64 does, and makes it
>> work.
>
>But, I'm still seeing display corruption (epecially in fonts) with
>stellarium and acceleration enabled on i386 (on a 82965PM chipset,
>which xorg identifies as a 965GM). I also see display problems on amd64
>running on a HD Graphics 4600, with opencpm.

What kind of fonts ?

The server will use completely different code paths for TrueType fonts
than for .pcf ones. I see less corruption with non-scaled fonts.

I have a theory that we need some cache flush calls in intel_ringbuffer.c.

I had started looking at this but my main test machine also has problems
with how i2c detects the screen.

Robert Swindells