I can't have install both xorg-server-1.5.0 and ati-drivers, what do I have to do to have 3d acceleration on ati card? Reproducible: Always
For now you can just add >=x11-base/xorg-server-1.4.99 to /etc/portage/package.mask
Also, I believe you are supposed to post the output of emerge --info with any bug report
I'm not dumb, Ive masked it, but I think it's a quite big problem that quite essential package conflicts with popular drivers. I think that setting 'xorg' USE flag triggered this conflict though I don't know what unsetting this flag might do.
'emerge --info' wouldn't be helpful here. As we're talking about ati-drivers, the only way this can be fixed is wait till ati releases new version, one meant for xorg-server 1.5 (I think it was the same when 1.4 was out). And I don't think this rates as a blocker.
Anyway (just to be annoying), xf86-video-ati still works (as good (or bad) as it did). That's just the way it is with binary-only drivers.
But...maybe someone could create a patch to make the compatible?
*** Bug 236874 has been marked as a duplicate of this bug. ***
No, it is not feasible to create a patch to make binary drivers work with a different xorg version. If you use binary drivers, expect to wait a little while until the vendor releases a version that has been tested against the new xorg. Both Nvidia and ATI binary drivers should be available by the time a new xorg goes stable, but not necessarily when a new xorg is marked as "testing". On the other hand, it sounds like Rafal has had good luck on running old ATI binary driver against the new xorg. If further testing confirms that the new xorg happened not to break old ati drivers, the xorg maintainer may want to remove the blocker early this time (instead of waiting for binary ATI drivers that officially support xorg-server-1.5)
I've used some patches to compile ati-drivers (fglrx) against unsupported kernel versions, I thought it would be possible to create ones for xorg.
Well, no, you misunderstood me. I was just being annoying. I simply meant: the open driver still works as good/bad as it did. And about those patches: you were patching source of a kernel module, the ati-driver xorg driver is binary-only, so upstream has to recompile and release it. Till then, you'll just have to wait.
So the only thing that's compiling in ati-drivers is kernel module:/?
Has anyone tried ati-drivers with xorg-server-1.5.0 by modifying the ati-drivers ebuild to allow it? Does it run alright?
It also is blocked when you use a mask of >=x11-drivers/nvidia-drivers-97.00 as suggested in http://www.gentoo.org/doc/en/nvidia-guide.xml because of an old video card. So people with old hardware are not able to upgrade to xorg-server-1.5.0 when they want to use the nvidia drivers :-(
(In reply to comment #12) > Has anyone tried ati-drivers with xorg-server-1.5.0 by modifying the > ati-drivers ebuild to allow it? Does it run alright? > yeah I've tested that. But as expected: it does not work >(II) Loading /usr/lib64/xorg/modules/drivers//fglrx_drv.so >dlopen: /usr/lib64/xorg/modules/drivers//fglrx_drv.so: undefined symbol: miZeroLineScreenIndex >(EE) Failed to load /usr/lib64/xorg/modules/drivers//fglrx_drv.so >(II) UnloadModule: "fglrx" >(EE) Failed to load module "fglrx" (loader failed, 7) >(EE) No drivers available.
(In reply to comment #14) > (In reply to comment #12) > > Has anyone tried ati-drivers with xorg-server-1.5.0 by modifying the > > ati-drivers ebuild to allow it? Does it run alright? > > > > yeah I've tested that. But as expected: it does not work You have successfully figured out the obvious reason there is an ati-drivers block in xorg-server-1.5.0 :) What is the problem here? The blocker is more than likely there to save from people using ati-drivers from unexpected breakages after upgrading xorg-server, leading them to frustration and having to downgrade a long xorg-server compile after the fact if they choose to stick with the proprietary drivers instead of using xf86-video-ati. So is portage showing you the blockers and giving you pain during a normal emerge -uDva world or similar run, or what is the problem here? The blocker is probably meant to keep you from upgrading xorg-server in a systematic world upgrade while it would break an installed driver - so you have to choose if to upgrade xorg-server after uninstalling ati-drivers or wait until ati-drivers works with xorg-server-1.5, at which point the blocker would be removed or adjusted to cover only the versions lower than the compatible version.
Created attachment 169084 [details, diff] ati-drivers 8.543 ebuild While the latest officially released fglrx driver (8.542) does not work with xorg-server-1.5, the ubuntu folk has gotten their hands on a prerelease (8.543) that does. That prerelease isn't available from amd, but it was rather trivial to modify the ebuild to use the .orig.tar.gz from archive.ubuntu.com instead. The attached patch does just that. It "works for me", though with a few quirks: The wine problems are still there, and amdcccle segfaults. But the rest seams to work, including direct rendering and dual head setup. Your mileage may vary...
Created attachment 169086 [details, diff] Patch to xorg-server-1.5.2 to remove the blocker For portage not to complain, you also need to apply this small patch to xorg-server to remove the blocker.
I would also like to comment that perhaps the xorg-server-1.5 ebuild should complain if you try to emerge it with the fglrx VIDEO_CARDS option set. I had a testing install and emerged xorg-x11 expecting fglrx VIDEO_CARDS to pull in the driver. If it isn't supported, it either shouldn't be listed as a valid VIDEO_CARD or complain if you have the option set.
Created attachment 170101 [details, diff] ati-drivers 8.543 ebuild modified for kernel 2.6.27 fixes src_url to ubuntu repository corrected stop compiling for kernel 2.6.27
on AMD64 the package fails to build due to a bug in firegl_public.c according to http://ubuntuforums.org/showthread.php?p=5943228 As I understand it the fix is to replace return compat_alloc_user_space(size); on line 1973 with return __ke_compat_alloc_user_space(size); let me know if you need emerge --info
sorry, I was wrong about the fix for amd64, I still get implicit declaration errors at the same point with the change.
sorry about the triple post, the problem I was having is that the version modified for 2.6.27 doesn't work for amd64, it builds fine under 2.6.25. The build for 2.6.27 fails in linux-mod_src_compile: /var/tmp/portage/x11-drivers/ati-drivers-8.542/work/lib/modules/fglrx/build_mod/2.6.x/firegl_public.c: In function 'KCL_AMD64_COMPAT32_AllocUserSpace': /var/tmp/portage/x11-drivers/ati-drivers-8.542/work/lib/modules/fglrx/build_mod/2.6.x/firegl_public.c:1973: error: implicit declaration of function 'compat_alloc_user_space' /var/tmp/portage/x11-drivers/ati-drivers-8.542/work/lib/modules/fglrx/build_mod/2.6.x/firegl_public.c:1973: warning: return makes pointer from integer without a cast /var/tmp/portage/x11-drivers/ati-drivers-8.542/work/lib/modules/fglrx/build_mod/2.6.x/firegl_public.c: In function '__ke_register_ioctl32_conversion': /var/tmp/portage/x11-drivers/ati-drivers-8.542/work/lib/modules/fglrx/build_mod/2.6.x/firegl_public.c:2911: error: implicit declaration of function 'register_ioctl32_conversion' /var/tmp/portage/x11-drivers/ati-drivers-8.542/work/lib/modules/fglrx/build_mod/2.6.x/firegl_public.c: In function '__ke_unregister_ioctl32_conversion': /var/tmp/portage/x11-drivers/ati-drivers-8.542/work/lib/modules/fglrx/build_mod/2.6.x/firegl_public.c:2920: error: implicit declaration of function 'unregister_ioctl32_conversion' /var/tmp/portage/x11-drivers/ati-drivers-8.542/work/lib/modules/fglrx/build_mod/2.6.x/firegl_public.c: In function '__ke_vm_phys_addr_str': /var/tmp/portage/x11-drivers/ati-drivers-8.542/work/lib/modules/fglrx/build_mod/2.6.x/firegl_public.c:3475: warning: return makes pointer from integer without a cast /var/tmp/portage/x11-drivers/ati-drivers-8.542/work/lib/modules/fglrx/build_mod/2.6.x/firegl_public.c:3476: warning: return makes pointer from integer without a cast /var/tmp/portage/x11-drivers/ati-drivers-8.542/work/lib/modules/fglrx/build_mod/2.6.x/firegl_public.c:3477: warning: return makes pointer from integer without a cast /var/tmp/portage/x11-drivers/ati-drivers-8.542/work/lib/modules/fglrx/build_mod/2.6.x/firegl_public.c:3479: warning: return makes pointer from integer without a cast /var/tmp/portage/x11-drivers/ati-drivers-8.542/work/lib/modules/fglrx/build_mod/2.6.x/firegl_public.c: At top level: /var/tmp/portage/x11-drivers/ati-drivers-8.542/work/lib/modules/fglrx/build_mod/2.6.x/firegl_public.c:5727: warning: initialization from incompatible pointer type /var/tmp/portage/x11-drivers/ati-drivers-8.542/work/lib/modules/fglrx/build_mod/2.6.x/firegl_public.c:5753: warning: initialization from incompatible pointer type
(In reply to comment #22) > sorry about the triple post, the problem I was having is that the version > modified for 2.6.27 doesn't work for amd64, it builds fine under 2.6.25. The > build for 2.6.27 fails in linux-mod_src_compile: > /var/tmp/portage/x11-drivers/ati-drivers-8.542/ ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ try the ati-drivers-8.543
(In reply to comment #16) > [..snip..] > > That prerelease isn't available from amd, but it was rather trivial to modify > the ebuild to use the .orig.tar.gz from archive.ubuntu.com instead. The > attached patch does just that. > FYI, the .orig.tar.gz is nowhere to be found on: SRC_URI="http://archive.ubuntu.com/ so, I changed the URL to start with: http://mirror.ne.gov/ubuntu/pool/multiverse/f/fglrx-installer/
(In reply to comment #24) > > FYI, the .orig.tar.gz is nowhere to be found on: > SRC_URI="http://archive.ubuntu.com/ > so, I changed the URL to start with: > http://mirror.ne.gov/ubuntu/pool/multiverse/f/fglrx-installer/ > never mind. Attachment 170101 [details, diff] has the correct URL.
Whoever posted that for Ubuntu violated their NDA with ATI (now AMD). It's from a private beta. We won't be adding it.
need version bump ^^ https://a248.e.akamai.net/f/674/9206/0/www2.ati.com/drivers/linux/catalyst_811_linux.pdf > Xorg 7.4 (Xserver 1.5) support ISSUES RESOLVED
*** Bug 247017 has been marked as a duplicate of this bug. ***
x11-drivers/ati-drivers-8.552-r1 is in portage (masked). I installed it and it works with xorg-server-1.5.2 (portage) but also with 1.5.3 and git (x11 overlay). I don't know if the block should be removed, since the drivers are still masked in portage.
This is still an issue. Why has this bug been open for over 6 months with no resolution and xorg-server 1.5.3 marked stable? If this is a problem with the binary drivers not being up to date then maybe there should be a block in place or at the very least some documentation in the upgrade guide on how to resolve the issue. Can someone please escalate this.
Jeff, I think you should comment on this. I have absolutely no idea how the latest ati-driver stuff works on 1.5.3. Thanks
(In reply to comment #31) > Jeff, I think you should comment on this. I have absolutely no idea how the > latest ati-driver stuff works on 1.5.3. Catalyst 9.1, 9.2, 9.3 and 9.4 fully support X 1.5.x. The version of the drivers this bug is about is ancient.
Alright then, no need to keep open. Thanks for the insight.
yes, this is no longer a problem with recent drivers.
It seems that the only problem is that xorg-server is marked stable while all the drivers that work with it are marked unstable. (Unless I didn't try one of the drivers). It seems like people without unstable keywords set are going to get broken, so this needs to be documented or newer ati-drivers should be marked stable.