Matthias Drochner wrote:
>> I managed to get drm working here on NetBSD 4.0_BETA2
>> using the i915 driver.
>>     
>
> Great work!
> I've adapted it to my -current system (i945) and it appears to work.
> The "glxgears" score is actually a bit less than with software
> rendering, but perhaps I'm doing something wrong. The amount
> of system time spent indicates that the kernel module does
> something...
>
>   
Did you turn off debugging?
I think it goes faster than software, if debugging is turned off.
>>    mknod /dev/dri/card0 c 34 0
>>     
>
> I've allocated the mi major number 188 for this which I intend
> to commit soon so that it is reserved.
> The libdrm and the xserver pkgs need to be adapted then.
>
>   
Thanks.
>> 3) I had to remove the workaround for an
>>    X server bug in agp_i810.c.
>>    Is this still true?
>>    (It does not seem to seem to affect my machine).
>>     
>
> It was me who introduced this. It was a BIOS bug, so while
> I don't have an affected board at hand I assume that the
> workaround is still necessary.
> We'll find a way to have them coexist.
>
>   
The last time I tried this it was messy. I will look again.
FreeBSD does not seem to have a workaround, I wonder
if they never encountered the problem?
> macallan@netbsd.org wrote:
>   
>> I'd still put them into sys/dev/drm - a while ago I was hacking on DRM 
>> support for Sun Creator3D graphics boards which are very non-PCI but 
>> that's really just cosmetics.
>>     
>
> I've already done that: pci card frontends in sys/dev/pci/drm, rest
> in sys/dev/drm. There is still some pci specific code in common
> files, but at least the general separation is done.
>
>   
Thanks.
> best regards
> Matthias
>
>
>   


-- 
Kind regards,

Yorick Hardy