On 22:15, Robert Millan wrote: > On 03/12/2013 14:24, Steven Chamberlain wrote: > > This is probably why Konstantin's sysctl checks are a good idea, and > > the radeonkms module should try to implement them. Otherwise, there's > > no proper fallback., > > Well, this was simpler than I thought. > > Please can you test?
Yes, that works, thanks! libdrm2 can detect when KMS is really available, without the "return 0" hack. But I still notice something odd: With upstream BETA3 or BETA4 kernel and libdrm2 hack: * Running `kldload radeonkms` enables KMS * Xorg is accelerated - simple glxgears test shows 230-240fps With Debian svn258623-1 and your patch implementing the sysctl, firmware modules available, and whether libdrm2 hack is used or not: * `kldload radeonkms` fails to automatically load the drm2.ko prerequisite - I have to manually `kldload drm2` first, whereas upstream kernels just automatically loaded it from e.g. /lib/modules/10.0-BETA4/ * libdrm2 detects KMS as available, Xorg uses the radeon driver * but graphics are *slow*, Xorg uses 100% CPU, keyboard and mouse are not very responsive and the same glxgears test reports 19fps I'm hoping the last point is something to do with the debugging options enabled in the Debian kernel builds for experimental. I'll try again soon with the debug options turned off, but I expect to find something else is causing this. Regards, -- Steven Chamberlain ste...@pyro.eu.org -- To UNSUBSCRIBE, email to debian-bsd-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/20131204013907.ga1...@squeeze.pyro.eu.org