Error installing ATI Catalyst Driver

moisespedro · 01-24-2014, 09:33 PM

After reading everything and thinking about it I think I will stick with the default driver

JackHair · 01-25-2014, 01:30 AM

Quote:

Originally Posted by mostlyharmless

Probably nothing, but, with kernel 3.12.1, version 13.11v9.95 black screens and hangs the keyboard. Fortunately one can still ssh into the system. There appears to be a dmar error, related to IOMMU/vt-d. I don't know whether that's a driver fault, or a flaky bios though. The radeon drivers appear to work fine however.

Okay that's odd behavior. I haven't used 3.12.1 so I don't know if it could be a kernel flaw. I've only used 3.12.5 and up. My system is AMD/ATI all the way might be the reason I have no problem with either Radeon or fglrx (although Radeon is slower). I never understood the fglrx "bashing" for me it's just as good as when I ran nVidia. Guess I'm lucky

TobiSGD · 01-25-2014, 02:08 AM

Quote:

Originally Posted by Stuferus

hd6850 here.. i tryed the beta driver with the default generic kernel from 14.1 x64.. for example: minecraft looks unplayable with it only artefacts.. so i reverted to the default radeon driver.. works well.. minecraft (for example) has 30fps, thats more than fine for me.

i dont think i want to compile 3.12 kernel never did in slackware so im a bit feard something could go wrong. 3.10.17 is ok.. an unofficial 3.10.27 package would be nice i must admit.

You really should use a later kernel, I recommend to go for 3.13. The 3.10 kernel does not have dynamic power management, so that your card all the time runs in the highest power mode, unnecessarily heating up the videocard.

Stuferus · 01-25-2014, 12:24 PM

no thanks.. 3.13 brings the new firewall stuff which i not really want right now.. also the d-bus in kernel stuff from "the evil redhat guy" is on its way if im not mistaken. so.. 3.12 if i try it - and im not to positive if i really want to.

the radeon driver is fine.. it works.. im not fixing unless its really brocken (no x)!

but thx.

ReaperX7 · 01-26-2014, 07:17 PM

The kernel dbus implementation is supposed to provide a lower latency daemon protocol. I'd hardly consider that on the same level of systemd, but the implementation hasn't been finalized, added, or introduced, and the Linux Foundation controls the kernel, not GNOME.

Kernel 3.13 is recommended for Radeon video cards using the Southern Islands or later chipsets.

If AMD/ATI and other users recommend Proprietary driver 13.12, you should use that driver, not the BETAs, and use the appropriate kernel, which is recommended 3.13.x.

TobiSGD · 01-26-2014, 10:55 PM

Quote:

Originally Posted by Stuferus

no thanks.. 3.13 brings the new firewall stuff which i not really want right now.. also the d-bus in kernel stuff from "the evil redhat guy" is on its way if im not mistaken. so.. 3.12 if i try it - and im not to positive if i really want to.

the radeon driver is fine.. it works.. im not fixing unless its really brocken (no x)!

but thx.

Actually, the 3.10 kernel is broken. It has a bug in the ondemand CPU governor that was fixed in kernel 3.12. This fix can speed up some applications about 90%. But of course it is up to you, if you don't care for performance and heat production/power usage just keep the 3.10 kernel.

santaslilslacker · 01-27-2014, 02:51 AM

Hi

Just to share my pain with proprietary AMD so called driver... I had that misfortune to acquire an HP laptop with AMD chipset on it. Default graphics driver was ok, but it wasn't fit for 3D I'm occasionally using. So I've started experimenting with their proprietary POS since early October. All in all I've tried around 10 different driver blobs ranging from post-13.4 up to 13.12, including current beta. It was rather agonizing and painfully frustrating

Thing is that with each driver there was something wrong. One didn't have proper 3D support, other was retarded with power management or brightness management was messed up. There were problems with sleep resuming, hibernation that were solved with one nice & long power button press.
Finally, 2 or 3 weeks ago I've installed 3.12.6 kernel and latest beta from January (amd-catalyst-13.11-betav9.95-linux-x86.x86_64). For some unknown reason that combination worked for me. Honestly I don't know what changed (and don't care any more) but with this combo - everything works as it's supposed to: 3D, power events and even brightness; so there is no more 100% brightness super nova for me. At the moment it's rather power efficient

Click image for larger version

Name: uptime.png
Views: 22
Size: 90.2 KB
ID: 14625

as you can see from the screenshot. So 3 hours on battery including compiling, browsing, wifi, 2 web servers with mysql dumping and importing with some music and gtalk.
Thing is that I was stupid enough to get that hardware in the first place, so from now on it's "HP & AMD never again". That approach I would like to recommend to anyone. Sell that crap soon as possible and get something with nvidia/intel inside. It's way less painful, blob drivers are a breeze to install and they work oob most of the time. I'm not talking about bumblebee combo but about regular, single vga cards in laptops or desktops. Personally I didn't have any problems with them.
So my final advice would be that if you have all those problems above try 3.12.6 with beta above. It just might work(TM)
Cheers,

S.

enorbet · 01-27-2014, 05:15 AM

Quote:

Originally Posted by Stuferus

no thanks.. 3.13 brings the new firewall stuff which i not really want right now.. also the d-bus in kernel stuff from "the evil redhat guy" is on its way if im not mistaken. so.. 3.12 if i try it - and im not to positive if i really want to.

the radeon driver is fine.. it works.. im not fixing unless its really brocken (no x)!

but thx.

If the radeon driver is sufficient, and it has gotten quite good for some chips) then so be it, but I wouldn't worry much about kernel implementation of systemd just yet. Linus seems to hate those guys at least as much I, and many others, do, and he is far less polite about it.... in fact, rather outspoken and vitriolic. For some information and fun, google "Linus on systemd" or "Linus on udev", etc. DRAMA!

Stuferus · 01-27-2014, 06:54 PM

Quote:

Originally Posted by TobiSGD

Actually, the 3.10 kernel is broken. It has a bug in the ondemand CPU governor that was fixed in kernel 3.12. This fix can speed up some applications about 90%. But of course it is up to you, if you don't care for performance and heat production/power usage just keep the 3.10 kernel.

than the question which must be asked is.. why did 14.1 ship with it if it is "broken"?

my lm_sensors says:
cpu 30°c in idle on desktop, gpu 60°c idleing on desktop..

*desktop=xfce

thats the same values i have in windows vista 32bit (which sits on an other harddrive) with ati catalyst 13.4 drivers (13.12 dont work so well for me there too). So im not worryed, at least i did read the card can be 100°C before it dies of heat.

i have enabled the dynpm and it looks like its working as the card gets to 64°C (max!) inside yt or a game (minecraft, flightgear etc).

i really like the slackware way: "not broke, no fixing"

- it doesnt feel broke for me..

TobiSGD · 01-27-2014, 08:35 PM

Quote:

Originally Posted by Stuferus

than the question which must be asked is.. why did 14.1 ship with it if it is "broken"?

You will always have bugs in the kernel and at the time 14.1 was released this bug was not known. Also, 3.10 is a LTS version, so it suits Slackware best.

Quote:

i have enabled the dynpm and it looks like its working

dynpm is not even implemented in 3.10, so it simply can't work.

Quote:

as the card gets to 64°C (max!) inside yt or a game (minecraft, flightgear etc).

A temperature difference of only 4°C between idle and full load indicates that power management is not working and the card is either running under full clockspeed/high voltages all the time or under lowest clockspeed/voltages all the time. Neither situation is what I would call optimal.

Anyways, it is of course up to you what you decide to use, it is just that your decision to stick with the od kernel IMHO does not make much sense when there are much better options available.

moisespedro · 03-07-2014, 07:30 PM

So I finally decided to upgrade to Kernel 3.13.6. Everything went fine but I still can't install the driver.

Code:

pedro@slack:~$ cat /usr/share/ati/fglrx-install.log 
AMD kernel module generator version 2.1
doing Makefile based build for kernel 2.6.x and higher
rm -rf *.c *.h *.o *.ko *.a .??* *.symvers
make -C /lib/modules/3.13.6/build SUBDIRS=/home/pedro/Downloads/fglrx-install.EOiIfh/common/lib/modules/fglrx/build_mod/2.6.x modules
make[1]: Entrando no diretório `/usr/src/linux-3.13.6'
  CC [M]  /home/pedro/Downloads/fglrx-install.EOiIfh/common/lib/modules/fglrx/build_mod/2.6.x/firegl_public.o
  CC [M]  /home/pedro/Downloads/fglrx-install.EOiIfh/common/lib/modules/fglrx/build_mod/2.6.x/kcl_acpi.o
/home/pedro/Downloads/fglrx-install.EOiIfh/common/lib/modules/fglrx/build_mod/2.6.x/kcl_acpi.c: In function ‘KCL_ACPI_GetHandles’:
/home/pedro/Downloads/fglrx-install.EOiIfh/common/lib/modules/fglrx/build_mod/2.6.x/kcl_acpi.c:796:59: error: ‘struct acpi_dev_node’ has no member named ‘handle’
         pInfo->video_handle = pInfo->pcidev->dev.acpi_node.handle;
                                                           ^
/home/pedro/Downloads/fglrx-install.EOiIfh/common/lib/modules/fglrx/build_mod/2.6.x/kcl_acpi.c: In function ‘KCL_ACPI_ParseTable’:
/home/pedro/Downloads/fglrx-install.EOiIfh/common/lib/modules/fglrx/build_mod/2.6.x/kcl_acpi.c:999:5: warning: passing argument 1 of ‘(acpi_status (*)(u32,  void *, void *))handler’ makes integer from pointer without a cast [enabled by default]
     ((acpi_table_handler)handler)(hdr);
     ^
/home/pedro/Downloads/fglrx-install.EOiIfh/common/lib/modules/fglrx/build_mod/2.6.x/kcl_acpi.c:999:5: note: expected ‘u32’ but argument is of type ‘struct acpi_table_header *’
/home/pedro/Downloads/fglrx-install.EOiIfh/common/lib/modules/fglrx/build_mod/2.6.x/kcl_acpi.c:999:5: error: too few arguments to function ‘(acpi_status (*)(u32,  void *, void *))handler’
make[2]: ** [/home/pedro/Downloads/fglrx-install.EOiIfh/common/lib/modules/fglrx/build_mod/2.6.x/kcl_acpi.o] Erro 1
make[1]: ** [_module_/home/pedro/Downloads/fglrx-install.EOiIfh/common/lib/modules/fglrx/build_mod/2.6.x] Erro 2
make[1]: Saindo do diretório `/usr/src/linux-3.13.6'
make: ** [kmod_build] Erro 2
build failed with return value 2
ERROR: I didn't make module
[Error] Generate Package - error generating package : Slackware/Slackware

EDIT: The beta driver worked but a solution would still be nice

moisespedro · 03-07-2014, 07:49 PM

I am testing it with fgl_glxgears and I am getting one weird message
[code] pedro@slack:~$ fgl_glxgears
Using GLX_SGIX_pbuffer
libGL error: failed to load driver: swrast
libGL error: Try again with LIBGL_DEBUG=verbose for more details.
8727 frames in 5.0 seconds = 1745.400 FPS
10525 frames in 5.0 seconds = 2105.000 FPS
10498 frames in 5.0 seconds = 2099.600 FPS
^C [code]

Code:

pedro@slack:~$ LIBGL_DEBUG=verbose fgl_glxgears
Using GLX_SGIX_pbuffer
libGL: screen 0 does not appear to be DRI2 capable
libGL: OpenDriver: trying /usr/lib64/xorg/modules/dri/swrast_dri.so
libGL error: failed to load driver: swrast
8314 frames in 5.0 seconds = 1662.800 FPS
^C

It seems harmless

EDIT: fixed it by running "root@slack:~# /usr/lib/fglrx/switchlibGL amd"