Re: Radeon, KMS, OpenGL and black screens
On Wed, 28 Dec 2011, Camaleón wrote:
> On Thu, 22 Dec 2011 23:51:43 +0100, Andre Majorel wrote:
> > As far as I can tell, enabling KMS causes the screen to go black shortly
> > after "ib test succeeded in 0 usecs" and disabling it prevents hardware
> > 3D acceleration.
> >
> > Any ideas ? Yes, drm, drm_kms_helper and radeon are loaded before startx
> > and no, it's not new hardware (RV370).
> >
> > Thanks in advance.
>
> Yes: report it at Debian BTS.
>
> KMS has to work flawlessly.
The RV370 is probably very close to the M22, so my KMS experience with the
Mobile Radeon X300 probably applies to the RV370 (non-mobile X300 ?) as
well.
I am typing this in a M22 (Mobile Radeon X300) in a ThinkPad T43, and it is
working fine in KMS mode, as long as you consider toy power management in a
laptop GPU to be worth the "working fine" label. Fortunately, this box
works fine with the X300 underclocked and undervolted, and therefore one can
use the "low" power_profile if one doesn't mind the SERR NMIs and resulting
kernel warnings the radeon driver causes when switching PCIe link speed.
Getting KMS to work on this Radeon X300 required a newer kernel, though:
2.6.32 KMS was hopeless on a lot of hardware, and even after so many
stable/longterm releases, it might still not be entirely fixed.
2.6.35 works fine, but that particular "longterm" kernel is best avoided (I
am not going to explain why. You can either trust me on this, or not).
3.0.x works really well, but some squeeze userspace is broken by the uts
version change... Bugs have been reported, but at least the laptop-mode one
has not been acted upon yet. You can hack the kernel to report itself as
2.6.4x if you want to avoid this, or use backported packages from testing or
unstable.
--
"One disk to rule them all, One disk to find them. One disk to bring
them all and in the darkness grind them. In the Land of Redmond
where the shadows lie." -- The Silicon Valley Tarot
Henrique Holschuh
Reply to: