Originally posted by Maarte
View Post
Announcement
Collapse
No announcement yet.
Linux 2.6.32 To Get R600 KMS Along With 3D
Collapse
X
-
Originally posted by LiquidAcid View PostUUID?
I'm just booting the kernel built from airlied's drm-next tree and once the login is available I modprobe radeon with the corresponding parameters. Is this the wrong approach?
The stability problems of yesterday seem to be fixed now with todays updates to kernel and xf86-video-ati.
FWIW when compiling libdrm I had to (unless it was something else I messed up) do
./autogen.sh --enable-radeon-experimental-api
Without that KMS worked, but I couldn't get modeset=0 or an old kernel + agd5f modules to work - The driver would hang on startx.
Comment
-
Originally posted by Maarte View PostWill Fedora 12 (Constantine) be 2.6.31 or 2.6.32? There seem to be conflicting opinions across both official sites and this forum. Thankyou for any clarification offered.
Comment
-
Originally posted by deanjo View PostIt's way to late for inclusion in OS 11.2. They already made an exception putting in 2.6.31. Beside's you can always install the Factory KOTD.
Comment
-
Originally posted by bridgman View PostI imagine the driver code needs to know about multiple GPUs even if it is only using one of them. I guess there are four stages of "support" :
nothing at all - driver gets confused and crashes when >1 GPU
awareness - driver can only work with one GPU at a time but knows enough to ignore the other ones
multi-GPU - driver can support multiple GPUs with different screens on each GPU
multi-GPU rendering - like Crossfire, driver can divide work across multiple GPU engines and combine the results to a single screen
Comment
-
Originally posted by pvtcupcakes View PostI disabled KMS, and I'm able to get into X now.
When I had KMS enabled, I got the nice high res framebuffer, but as soon as X started I got a black screen and no signal on my monitor.
Comment
-
Test with fedora_test_day_live_cd
Hello Everyone!
Yesterday I tested the fedora test day live cd with 3 systems. It was a disaster!
On my ASUS 1005HA Netbook (Intel 945) the 3d acceleration was disabled. On the next system with a Nvidia card (GeForce 6700 XL) KMS and xrandr works. XV seems to be work but wasn't checked. And finally on my system with HD 3870 the most things work! I started with radeon.modeset=1 and after some reboots (which depends on my crypted disks) i got a desktop with working compiz! The test with xv passed. But suddenly the system hang while i test the quick user switch. I turned power off because even the reset button was dead. Afterwards the radeon fan runs at full speed for ~ 20 seconds. This was very uncommon. The system started windows and debian normal. So i tried the live cd again. The system hang on the login and my monitor failed to show a picture. At the same time the radeon fan start with full speed. I checked with radeon.modeset=0 but the system hang on the login screen.
Comment
-
Originally posted by zappbrannigan View PostI turned power off because even the reset button was dead. Afterwards the radeon fan runs at full speed for ~ 20 seconds. This was very uncommon. The system started windows and debian normal. So i tried the live cd again. The system hang on the login and my monitor failed to show a picture. At the same time the radeon fan start with full speed. I checked with radeon.modeset=0 but the system hang on the login screen.
Comment
-
Doesn't work too well here. hd 3470 mobility on toshiba laptop (RV620).
gentoo, xorg-server 1.6.4-rc1, latest drm-next kernel, xf86-video-ati, mesa and libdrm from git (all compiled 10 mins ago). It loads the firmware only when radeon is compiled as module, then gives a very nice high-res console; starting X works /but/
1) sometimes X crashes when launching even 'simple' apps, and the whole laptop hangs
2) if it manages to survive for a while, on dmesg I get a bunch of
Code:[drm:radeon_ring_write] *ERROR* radeon: writting more dword to ring than expected !
When switching back works, I get in Xorg.0.log some messages like
Code:space check failed in flush
Code:Unpin not necessary for ffff8800bd793a00 !
Comment
Comment