R600/r700 kms + 3d dri1/dri2

[suokko]

There is now first public testing version available for r600 KMS. It also includes 3D support. But code is very experimental so you might end up having problems. All problems should be reported to https://bugs.freedekstop.org.

[Fran]

Doesn't work very well here (rv670, hd3850). As I said in the IRC:

- If I enable modesetting by default at boot, it doesn't find the firmware (possibly my fault, because it *is* in /lib/firmware/radeon).

- If I don't enable modesetting by default, build radeon as a module and load it with modeset=1, I get this

[drm] Initialized drm 1.1.0 20060810
[drm] radeon defaulting to userspace modesetting.
pci 0000:01:00.0: PCI INT A -> GSI 16 (level, low) -> IRQ 16
pci 0000:01:00.0: setting latency timer to 64
[drm] Initialized radeon 1.31.0 20080528 for 0000:01:00.0 on minor 0
[drm] Module unloaded
[drm] Initialized drm 1.1.0 20060810
[drm] radeon kernel modesetting enabled.
radeon 0000:01:00.0: setting latency timer to 64
[drm] radeon: Initializing kernel modesetting.
[drm] register mmio base: 0xE1000000
[drm] register mmio size: 65536
ATOM BIOS: 113
[drm] Clocks initialized !
[TTM] Zone  kernel: Available graphics memory: 2029444 kiB.
[drm] radeon: 256M of VRAM memory ready
[drm] radeon: 512M of GTT memory ready.
[drm] Loading RV670 CP Microcode
platform radeon_cp.0: firmware: requesting radeon/RV670_pfp.bin
platform radeon_cp.0: firmware: requesting radeon/RV670_me.bin
[drm] GART: num cpu pages 131072, num gpu pages 131072
[drm] ring test succeeded in 1 usecs
[drm] radeon: ib pool ready.
[drm:r600_ib_test] *ERROR* radeon: fence wait failed (-16).
[drm:r600_init] *ERROR* radeon: failled testing IB (-16).
[drm:radeon_driver_load_kms] *ERROR* Failed to initialize radeon, disabling IOCTL
[drm] radeon: finishing device.
[drm] radeon: fence finalized
[TTM] Zone  kernel: Used memory at exit: 1028 kiB.
[drm] radeon: ttm finalized
radeon: probe of 0000:01:00.0 failed with error -16

- If I forget about modesetting (modprobe radeon modeset=0), start X and launch glxgears, the screen suddenly goes black and the fan goes to 100%. That made me jump off my chair (the fan in a 3850 at 100% sounds like a fraking vacuum cleaner).

So for me it will be agd5f's drm for now.

[nanonyme]

I'd assumed the firmware belongs on initrd in KMS.

[Fran]

I'd assumed the firmware belongs on initrd in KMS.

Hum, maybe you're right. I thought that if ext4 support was built-in, the kernel should be able to locate the firmware in the root partition.

I hate initrd

Anyway, that's the least of my worries. The other two points are more annoying.

[tormod]

Hum, maybe you're right. I thought that if ext4 support was built-in, the kernel should be able to locate the firmware in the root partition.

The kernel does not look around in your partitions. It emits an event so that udev can call its firmware script to locate and load the firmware. See /lib/udev/rules.d/50-firmware.rules and /lib/udev/firmware.sh (on Debian and Ubuntu at least).

[lucky_]

Well, I guess it'gonna be hard to prevent us from wrecking our configs with those news.
I think I will let some kamikaze take the pain, before I dive into it.
As I can see it, it will be merges for 2.6.32, but how long will it stay in staging ?

[pvtcupcakes]

I think I'll wait until tomorrow and try out the Radeon test day for Fedora.

Hopefully I have time, I'll be spending most of my day at work and school. Will there be a midnight release of the test live cd? So I can have it ready to go when I get home.

[Fran]

Ok, I got it working. I just needed this patch to solve the second issue.

The third issue just went away.

So right know I have KMS working perfectly in my r600, yay! \o/

[kernelOfTruth]

great !

Fran, sorry for the über-n00bish question:

anything special how I need to handle the firmware ?

just copy it over from the kernel-sources firmware/radeon directory to /lib/firmware/radeon ?

thanks

[Fran]

A make install should copy it for you. And modprobe radeon should load it.

What I still haven't managed is to get it loaded when radeon is built-in. The boot process simply stops. If, as nanonyme says, an initrd is necessary, I'll pass for now. I've had bad experiences with initrds.