Bug 2354

Summary: RFE: Catalyst 12.6 for F17
Product: Fedora Reporter: Patrick <rpmfusion>
Component: catalyst-kmodAssignee: Stewart Adam <s.adam>
Status: RESOLVED FIXED    
Severity: normal CC: harm, kwizart, leigh123linux, maxim.yegorushkin
Priority: P5    
Version: 17   
Hardware: All   
OS: GNU/Linux   
namespace:
Attachments: catalyst-kmod 12.4 -> 12.6 patch
xorg-x11-drv-catalyst 12.4 -> 12.6 patch
Catalyst Control Center Information page

Description Patrick 2012-06-01 18:07:37 CEST
Would it perhaps be possible to create Catalyst 12.6 packages for F17 x86_64?

This release is available at:
http://www2.ati.com/drivers/hotfix/catalyst_12.6_hotfixes/amd-driver-installer-8.98-x86.x86_64.zip

It's compatible with Linux 3.3 and X.Org Server 1.12 so maybe it's not too challenging to create them for F17. Unofficial changelog here:
http://www.phoronix.com/scan.php?page=news_item&px=MTExMTU

Thank you for your consideration!
Comment 1 Nicolas Chauvet 2012-06-01 18:16:22 CEST
Can you create a patch against the previous src.rpm ?
Comment 2 Patrick 2012-06-01 18:58:37 CEST
Created attachment 905 [details]
catalyst-kmod 12.4 -> 12.6 patch
Comment 3 Patrick 2012-06-01 18:59:15 CEST
Created attachment 906 [details]
xorg-x11-drv-catalyst 12.4 -> 12.6 patch
Comment 4 Patrick 2012-06-01 19:00:17 CEST
Both catalyst-kmod and xorg-x11-drv-catalyst require little work to compile (but I did not test if the rpms actualy work, nor did I build them with mock):

In catalyst-kmod.spec:

1) update version to 12.6
2) update release
3) disable rename_debug.patch

In xorg-x11-drv-catalyst.spec

1) update version to 12.6
2) update release
3) in %files add: %{_sysconfdir}/ati/atiapfxx.blb

Patches uploaded for both spec files.
Comment 5 leigh scott 2012-06-12 14:41:11 CEST
(In reply to comment #0)
> Would it perhaps be possible to create Catalyst 12.6 packages for F17 x86_64?

> It's compatible with Linux 3.3 and X.Org Server 1.12 so maybe it's not too

12.6 isn't compatible with the F17 3.4.x kernel.
Comment 6 Patrick 2012-06-12 15:07:27 CEST
That must be a runtime issue then because both they built fine on F17 x86_64 with only minor modifications in the spec file.

I also recall a report on the Fedora users mailing list where someone mentioned that 12.6 worked fine on F17.

Have you tried building the 12.4 SRPMs updated to 12.6 via the two spec file patches and tested if they work?
Comment 7 leigh scott 2012-06-12 15:25:50 CEST
(In reply to comment #6)
> That must be a runtime issue then because both they built fine on F17 x86_64
> with only minor modifications in the spec file.

It's a build issue

$ make
make -C /lib/modules/3.4.2-1.fc16.x86_64/build SUBDIRS=/home/leigh/rpmbuild/BUILD/fglrx/common/lib/modules/fglrx/build_mod/2.6.x modules
make[1]: Entering directory `/usr/src/kernels/3.4.2-1.fc16.x86_64'
  CC [M]  /home/leigh/rpmbuild/BUILD/fglrx/common/lib/modules/fglrx/build_mod/2.6.x/firegl_public.o
/home/leigh/rpmbuild/BUILD/fglrx/common/lib/modules/fglrx/build_mod/2.6.x/firegl_public.c: In function ‘kasInitExecutionLevels’:
/home/leigh/rpmbuild/BUILD/fglrx/common/lib/modules/fglrx/build_mod/2.6.x/firegl_public.c:4159:5: error: ‘cpu_possible_map’ undeclared (first use in this function)
/home/leigh/rpmbuild/BUILD/fglrx/common/lib/modules/fglrx/build_mod/2.6.x/firegl_public.c:4159:5: note: each undeclared identifier is reported only once for each function it appears in
/home/leigh/rpmbuild/BUILD/fglrx/common/lib/modules/fglrx/build_mod/2.6.x/firegl_public.c:4159:5: warning: left-hand operand of comma expression has no effect [-Wunused-value]
make[2]: *** [/home/leigh/rpmbuild/BUILD/fglrx/common/lib/modules/fglrx/build_mod/2.6.x/firegl_public.o] Error 1
make[1]: *** [_module_/home/leigh/rpmbuild/BUILD/fglrx/common/lib/modules/fglrx/build_mod/2.6.x] Error 2
make[1]: Leaving directory `/usr/src/kernels/3.4.2-1.fc16.x86_64'
make: *** [kmod_build] Error 2



> Have you tried building the 12.4 SRPMs updated to 12.6 via the two spec file
> patches and tested if they work?


No, I haven't got a hd5xxx + card
Comment 8 Patrick 2012-06-12 16:14:46 CEST
You are building on F16 and not F17 like I did. The test I did was on F17. Maybe that makes a difference, I don't know. All I know is that the SRPMs updated to F17 built for me.

Maybe one of the packagers/developers in the ATI/AMD Phoronix forum have a patch that enables compatibility with that 3.4 kernel.
Comment 9 Nicolas Chauvet 2012-06-12 16:28:32 CEST
(In reply to comment #8)
> You are building on F16 and not F17 like I did. The test I did was on F17.
The kmod will not build because F17 was recently updated to kernel 3.4.0.
Anyhow, leigh123linux lacks the needed hardware and express his intention to drop maintainance of a main most updated branch of the driver.

@Patrick, please have a look on current bugzilla entry for this package and try to propose a fix if you intend to take over maintenance of the catalyst driver.

Please also reminds that currently catalyst 12.6 is pre-release software, so there is no rush to brick end-users systems.
Comment 10 Patrick 2012-06-12 16:44:23 CEST
Sorry for the confusion. I totally missed that F16/F17 got the kernel updated to 3.4.

@Nicolas: I do not have the knowledge to take over maintenance of this package as I don't know C. No rush needed on my part.
Comment 11 leigh scott 2012-06-12 17:12:23 CEST
(In reply to comment #9)
> Anyhow, leigh123linux lacks the needed hardware and express his intention to
> drop maintainance of a main most updated branch of the driver.

> Please also reminds that currently catalyst 12.6 is pre-release software, so
> there is no rush to brick end-users systems.

Users are already destroying their systems with the .run file, it maybe better to package the pre-release and leave it in testing to the final release.
I have seen reports that it works ok, at least if it fails the users will find it easier to remove.

If Patrick can provide spec files that work (tested) with the 3.4 kernel I would be willing to build it if Stewart is unavailable.
Comment 12 Harm 2012-06-24 13:33:41 CEST
(In reply to comment #11)
> Users are already destroying their systems with the .run file, it maybe better
> to package the pre-release and leave it in testing to the final release.
> I have seen reports that it works ok, at least if it fails the users will find
> it easier to remove.

Is there any news on when rpm files will be available (albeit in -testing)?

Building the kmod using the attached files worked out ok on kernel-3.4.0. I do remember having to tar my own set of http://downloads.diffingo.com/rpmfusion/kmod-data/catalyst-kmod-data-%{version}.tar.bz2 using files from the .run file. In the end, I got several working rpms, a watermark but no graphics acceleration. So something clearly wasn't done right.
Comment 13 leigh scott 2012-06-24 14:32:10 CEST
(In reply to comment #12)
> (In reply to comment #11)
> > Users are already destroying their systems with the .run file, it maybe better
> > to package the pre-release and leave it in testing to the final release.
> > I have seen reports that it works ok, at least if it fails the users will find
> > it easier to remove.
> 
> Is there any news on when rpm files will be available (albeit in -testing)?

I'm not going to do any work on packaging it unless someone else provides a working patch for the catalyst-kmod package as it doesn't build with the 3.4 kernel.


> Building the kmod using the attached files worked out ok on kernel-3.4.0. I do
> remember having to tar my own set of
> http://downloads.diffingo.com/rpmfusion/kmod-data/catalyst-kmod-data-%{version}.tar.bz2
> using files from the .run file. In the end, I got several working rpms, a
> watermark but no graphics acceleration. So something clearly wasn't done right.

What did you expect?, it needs patching!!
Comment 14 Maxim Yegorushkin 2012-06-26 23:50:36 CEST
I built Catalyst 12.6 beta after applying the patch from here http://henryhermawan.blogspot.co.uk/2012/06/amd-catalyst-patch-for-kernel-34x.html and currently run it with no issues.

After building I got the "internal testing only" watermark when running the driver and had to drop in control and signature files into /etc/ati to make the logo disappear. I don't remember where I got the files from but I still have them. Would it be okay to include the files into the patch?

Which prerequisites would I need to download to play around with patching the spec files and rpmbuilding it?
Comment 15 leigh scott 2012-06-27 04:44:59 CEST
(In reply to comment #14)

> After building I got the "internal testing only" watermark when running the
> driver and had to drop in control and signature files into /etc/ati to make the
> logo disappear. I don't remember where I got the files from but I still have
> them. Would it be okay to include the files into the patch?


12.6 should be in the testing repo soon, as for the watermark, I refuse to remove it as it identifies the driver as a beta.
Comment 16 leigh scott 2012-06-27 09:15:12 CEST
$ yum --releasever 17 --enablerepo rpmfusion-nonfree-updates-testing list *\mod-catalyst xorg-x11-drv-catalys\*
Loaded plugins: langpacks, presto, refresh-packagekit, remove-with-leaves, show-leaves
Available Packages
akmod-catalyst.x86_64                                        12.6-0.1.fc17                           rpmfusion-nonfree-updates-testing
kmod-catalyst.x86_64                                         12.6-0.1.fc17                           rpmfusion-nonfree-updates-testing
xorg-x11-drv-catalyst.x86_64                                 12.6-0.1.fc17                           rpmfusion-nonfree-updates-testing
xorg-x11-drv-catalyst-devel.i686                             12.6-0.1.fc17                           rpmfusion-nonfree-updates-testing
xorg-x11-drv-catalyst-devel.x86_64                           12.6-0.1.fc17                           rpmfusion-nonfree-updates-testing
xorg-x11-drv-catalyst-libs.x86_64                            12.6-0.1.fc17                           rpmfusion-nonfree-updates-testing
Comment 17 Maxim Yegorushkin 2012-07-03 22:51:50 CEST
Nicely done. I've installed xorg-x11-drv-catalyst.12.6-1.fc17 driver from rpmfusion-nonfree-updates and it works as expected.

amdcccle application still shows "Catalyst Version 12.4" though. Is it expected?

I attach a screenshot.
Comment 18 Maxim Yegorushkin 2012-07-03 22:52:40 CEST
Created attachment 923 [details]
Catalyst Control Center Information page