On Fri, 6 Aug 2010 14:34:22 +0100 Owain Ainsworth
<zerooa@googlemail.com> wrote:
quoted text
>
> Other than intagp, pretty much every agp driver is almost identical,
> with only small details differing. Unlike intagp, the ptes are not in
> a BAR, but in dma memory. so this diff just repeats every register
> setting that we do on attach at resume (including aperture size ones
> incase this was reset by PSS, calling that twice on attach is the
> only functional change for attach time).
> 
> sthen@ has tested this on an x31 (or was it a t31... an old thinkpad
> with a radeon and a 855 agp chipset), and it helped out. For one thing
> it fixes the ``no X started, suspend, resume, startx, crash'' problem
> for him.  All other agp bridges still need testing though.
> 
> This diff is NOT in snapshots and will not make release unless it
> gets a decent amount of testing, so please test it and let me know
> what your results are.
> 
> Cheers,
> -0-

Most of the testing done will probably be done on laptops or other
systems with built-in graphics. Today I'm starting testing on an
IBM/Lenovo desktop with built-in intel with a optional AGP slot. I'll
try to cover all the other cards I have.

	jcr

--
The OpenBSD Journal - http://www.undeadly.org