| What | Removed | Added |
|---|---|---|
| Assignee | xorg-driver-ati@lists.x.org | dri-devel@lists.freedesktop.org |
| QA Contact | xorg-team@lists.x.org | |
| Version | 6.9.0 | unspecified |
| Component | Driver/Radeon | DRM/Radeon |
| Product | xorg | DRI |
Does disabling AGP help? Load the radeon module with agpmode=-1
Indeed that solve the symptoms. I discover this was also documented here: https://wiki.debian.org/PowerPC/FAQ#How_do_I_get_graphics_working.3F [...] image=/boot/vmlinux label=Linux read-only initrd=/boot/initrd.img append="radeon.agpmode=-1 modprobe.blacklist=ams" [...] I am not sure what the 'ams' thingy is doing, but I am using it over here. The funny thing is that dragging window and moving them around makes the pointer and the window out-of-sync. Still usable though after a while. If AGP is simply not supported on PowerPC, it would be nice to have a clearer message (or at least a warning). Thanks
(In reply to Mathieu Malaterre from comment #3) > If AGP is simply not supported on PowerPC, it would be nice to have a > clearer message (or at least a warning). It is supported, it's just unstable on many PowerMacs. A patch disabling AGP by default on PowerMacs or even PPC in general probably wouldn't be rejected. :)
Created attachment 123411 [details] [review] default to PCI for PowerPC
(In reply to Michel Dänzer from comment #4) > (In reply to Mathieu Malaterre from comment #3) > > If AGP is simply not supported on PowerPC, it would be nice to have a > > clearer message (or at least a warning). > > It is supported, it's just unstable on many PowerMacs. > > A patch disabling AGP by default on PowerMacs or even PPC in general > probably wouldn't be rejected. :) Something like the patch: defaulttopci_powerpc.patch ? Should I submit it first on mesa-dev ?
FWIW on nouveau we just default agpmode=0 in a #ifdef __powerpc__. Perhaps the same thing can work for radeon.
(In reply to Ilia Mirkin from comment #7) > FWIW on nouveau we just default agpmode=0 in a #ifdef __powerpc__. Perhaps > the same thing can work for radeon. Well radeon_agpmode is already set to 0 already[*]. My patch force to PCI(-1) when user did not specify any value. [*]https://github.com/torvalds/linux/blob/master/drivers/gpu/drm/radeon/radeon_drv.c#L173
(In reply to Mathieu Malaterre from comment #8) > (In reply to Ilia Mirkin from comment #7) > > FWIW on nouveau we just default agpmode=0 in a #ifdef __powerpc__. Perhaps > > the same thing can work for radeon. > > Well radeon_agpmode is already set to 0 already[*]. My patch force to > PCI(-1) when user did not specify any value. > > [*]https://github.com/torvalds/linux/blob/master/drivers/gpu/drm/radeon/ > radeon_drv.c#L173 Oh, the logic is flipped from nouveau... on nouveau, -1 == autodetect and 0 == no agp. So in your case, just set it to -1 instead of 0 for powerpc. i.e. #ifdef __powerpc__ int radeon_agpmode = -1; #else int radeon_agpmode = 0; #endif or something similar.
(In reply to Michel Dänzer from comment #4) > (In reply to Mathieu Malaterre from comment #3) > > If AGP is simply not supported on PowerPC, it would be nice to have a > > clearer message (or at least a warning). > > It is supported, it's just unstable on many PowerMacs. > > A patch disabling AGP by default on PowerMacs or even PPC in general > probably wouldn't be rejected. :) Hi, Michel I don't understand. I know about the coherency issues on some UniNorth bridges (DMA writes through the GART going directly to RAM, IIRC), but how did OS X cope with them? Did it also disable AGP transfers? If not, what can be done to fix this bug for real, instead of working around it? (I have a very vague memory that these hangs didn't happen before KMS, but I may be completely wrong.) Also, on a somewhat related note, there are still issues with the Radeon (R600, not SI) DRM on big endian, as for https://bugs.freedesktop.org/show_bug.cgi?id=95015. Thanks, Rui
Also been hit by this bug on 2 of my 3 machines: PowerMac 7,3 (A1047) w. Radeon 9600 Pro and PowerBook 5,6 (A1106). My PowerBook 5,8 (A1138) magically works with AGP! Which is interesting because both the 5,6 and the 5,8 use a Mobility Radeon 9700.
Patch has been submitted to mesa-dev: https://lists.freedesktop.org/archives/mesa-dev/2016-May/115453.html<= /a> and Reviewed-By Christian K=C3=B6nig
Patch has also been submitted to dri-devel: https://lists.freedesktop.org/archives/dri-devel/2016-May/107479.htm= l
It seems the patches to force Radeon cards into PCI-mode (or f= orce it via radeon.agpmode=3D-1) are no definitive solution. My G4 and G5 Macs hold out longer running X but sooner or later a hard lock is unavoidable. Tested thi= s on a recent software stack: 4.8.1-kernel, xorg-server-1.18.4, xf86-video-ati-7= .7.1 and mesa-12.0.3. See my comment at Bug 94877.
Played around a bit with xorg and kernel options on my G5 and =
worked out that I
can drive the Radeon 9650 at AGP x8 if I disable DRI completely via Option
"NoAccel" "true" in /etc/xorg.conf.
I had no GPU-lockups since then and xorg is much more responsive despite
lacking DRI hardware acceleration (radeon.agpmode=3D-1 with DRI enabled).
linux-4.8.3-gentoo # dmesg | grep -i agp [ 0.000000] Found U3-AGP PCI host bridge. Firmware bus number: 240->= 255 [ 79.271669] Linux agpgart interface v0.103 [ 79.286888] agpgart-uninorth 0000:f0:0b.0: Apple U3H chipset [ 79.300467] agpgart-uninorth 0000:f0:0b.0: configuring for size idx: 64 [ 79.308235] agpgart-uninorth 0000:f0:0b.0: AGP aperture is 256M @ 0x0 [ 82.883029] agpgart-uninorth 0000:f0:0b.0: putting AGP V3 device into 8x mode [ 82.883040] radeon 0000:f0:10.0: putting AGP V3 device into 8x mode
(In reply to erhard_f from comment #15) > Played around a bit with xorg and kernel options= on my G5 and worked out > that I can drive the Radeon 9650 at AGP x8 if I disable DRI completely= via > Option "NoAccel" "true" in /etc/xorg.conf. >=20 > I had no GPU-lockups since then Sure, because you're not actually using the problematic AGP functionality. = :) > and xorg is much more responsive despite lacking= DRI hardware acceleration > (radeon.agpmode=3D-1 with DRI enabled). FWIW, the value of the radeon.agpmode parameter shouldn't matter if you dis= able DRI (which BTW means no GPU hardware acceleration can be used at all).
> FWIW, the value of the radeon.agpmo=
de parameter shouldn't matter if you disable
> DRI (which BTW means no GPU hardware acceleration can be used at all).=
Yep. And I was astonished to see the desktop running faster actually. ;)
*** Bug 95177 has been marked as a dupl= icate of this bug. ***
| What | Removed | Added |
|---|---|---|
| CC | sur3@gmx.de |
Created attachment 142691 [details]<=
/span>
dmesg
I think I have the same error with my Radeon HD 4670 on x64_64 arch with li=
nux,
I attached my dmesg with the error in the end.
| What | Removed | Added |
|---|---|---|
| Hardware | PowerPC | All |
nothing seems to help, I tried agpmode=3D-1, disabling glamor = that didnt help and Option "NoAccel" "true" wont let me login on the grafic= al user interface at all.
| What | Removed | Added |
|---|---|---|
| Resolution | --- | MOVED |
| Status | NEW | RESOLVED |
-- GitLab Migration Automatic Message -- This bug has been migrated to freedesktop.org's GitLab instance and has been closed from further activity. You can subscribe and participate further through the new bug through this = link to our GitLab instance: https://gitlab.freedesktop.org/drm/amd/issues/711.