PDA

View Full Version : Slamd64 (64 Bit slackware) and ATI Radeon (catalyst) driver problem


ljones
03-22-2009, 01:06 PM
Hello all !

I'm trying to figure out how to install ATI's catalyst drivers into my install of slamd64 (that's a 64-bit version of slackware) 12.2 . Unforunatly I don't seem to be getting very far, so far I've had no success at doing this.

Here's the system I am currently using (specs);

* Memory - 4 GB Ram

* Motherboard - ASUS M3A78EM

* CPU - AMD Phenom II x4 940

* Graphics - Onboard ATI Radeon HD 3200

Linux kernel version = 2.6.27.7

Unfortunatly I can't get the driver to even install properly let alone run. If I try to install the latest 9.2 catalyst driver from here ( http://support.amd.com/us/gpudownload/Pages/index.aspx ) and install it from the command line (outside of X) I get this strangeness at the end of installation;


n_US/LC_ADDRESS
7f1e95a3d000-7f1e95a3000 r--p 0000000 08:01 40566
/usr/lib64/locale/en_US/LC_TELEPHONE
7f1e95a3000-7f1e95a3f000 r--p 000000
000 08:01 40570
5a3f000-7f1e95a40000 r---p 000000 08:01 40573 /usr/lib
64/locale/en_US/LC_IDENTIFICATION
7f1e95a40000-7f1e95a4200 rw-p 7f1e95a40000 00:00 /lib64/ld-2.7.so


And right at the bottom of the screen;


./at-installer.sh: line 39: 11447 Aborted ./setup.data/bin/${ARCH_DIR}/setup 2>/dev/null
Removing temporary directory: fglrx-install.hxzQfq3
root@darkstar:~#


BTW that's not a formatting error above in what I'm guessing are error messages, the text actually looks like that on-screen, all messed up and out of line. I get the same results from the 9.2, 9.1 and 8-12 driver.

If I try to start X after this, it dosen't work -- I either get a completely blank screen, no X cursor or maybe some red lines or dots on the screen. The system then locks up hard; no response from mouse or keyboard - the only way out is to power off or hit the reset button.

Even after trying 'aticonfig --initial' I get the same problem, in fact the system went to a blank screen then reset itself!

xorgsetup also dosen't seem to work properly after installing the ATI drivers, it just crashes with these messages;


ast
apm
neomagic
newport
fbdev
vesa
vga

Backtrace:
0: /usr/X11R6/bin/X(xf86Sighandler+0x6a) [0x48666a]
1: /lib64/libc.so.6 [0x7f764663df70]
2: /usr/lib64/xorg/modules/drivers//fglrx_drv.so(atiddxProbeMain+0x8fd) [0x7f764129b1bd]
3: /usr/X11R6/bin/X(DoConfigure+0x1fa) [0x47efca]
4: /usr/X11R6/bin/X(InitOutput+0x6a5) [0x4615f5]
5: /usr/X11R6/bin/X(main+0x29f) [0x43017f]
6: /li64/libc.so.6(__libc_start_main+0xf4) [0x7f764662a164]
7: /usr/X11R6/bin/X [0x42f699]

Fatal server error:
Caught signal 11. Server aborting

/usr/bin/xorgsetup: line 170 : 24548 Aborted /usr/X11R6/bin/X -configure


I've tried reinstalling slamd64 (so I'm working off a clean install) and redownloading the drivers (just in case I got a bad download) but no luck, same problem every time.

Has anyone had any luck with the 9.2 ATI drivers at all? After googling a *lot* I'm finding most people are having *lots* of trouble x.x .... (add one to the list now!).

Can anyone think what is wrong? Thanks :)

ljones

ljones
03-22-2009, 04:03 PM
I may have an answer! Or maybe - 99% of one!

After a lot of poking about on this forum, and I forget which thread I read this on (d'oh) -- the thread was discussing problems regarding 64 bit systems, southbridge chips and 4GB of memory. It appears I may have this problem.

I have in the bios of my motherboard (ASUS M3A78EM) enabled "Memory hole remapping" as someone on the thread I was reading suggested this, but it didn't seem to work. Still the system locked up, no X, no response to keyboard or mouse, having to hit reset!

However there is another option, "Internal Graphics". I changed primary video controller from "GFX0-GPP-IGFX-PCI" to "IGFX-GFX0-GPP-PCI", not sure if that will make much difference. But more importantly changing another option "Frame buffer location" does seem to make a difference. Set it to "Above 4G" and it's crash and lock-up time. But set it to "Below 4G" and it works (it is set to this right now).

Only odd thing now is if I drop back out of X, the system restarts. However X seems to be running -- I'm running the default install of XFCE 4 that Slamd64 uses, and the (admittidaly mimimal) 3D functions that can be enabled in 'window manager tweaks' and then 'compositor'.

Glxgears seems to be a bit odd though - leave it at its default window size and I get 1424 FPS, make it twice as big and it drops to 241 FPS! No idea on that one ............

EDIT: I also had to set the graphics cards' memory to either "auto" or "256M" (as it is on board). I don't have a 512M option. Setting it (in the bios) to 64M caused crashes again x.x .

Just hope I'm right! :-)

ljones

ljones
03-22-2009, 04:13 PM
:-( Damn! I thought I had it ....

Now glxinfo is telloing me "Direct rendering: No" x.x !

What could be wrong?

Tried running aticonfig --initial, but it didn't seem to work, still direct rendering = no :( !

?? Does anyone know why direct rendering won't work ?? :(

ljones

ljones
03-22-2009, 05:19 PM
Ok, I'm going to try to reinstall everything, start from scratch. I just looked in my /var/log/Xorg.0.log file and saw this:


fglrx(0): driver needs X.org 7.1.x.y with x.y >= 0.0
fglrx(0): could not detect X server version (query_status=1)
fglrx(0): atiddxDriScreenInit failed. GPS not been initalised.


Thinking something has got messed up somewhere .....

ljones

ljones
03-22-2009, 05:59 PM
:( !

Output of /var/log/Xorg.0.log


(==) fglrx(0): NoDRI = NO
(II) Loading sub module "fglrxdrm"
(II) LoadModule: "fglrxdrm"
(II) Reloading /usr/lib64/xorg/modules/linux//libfglrxdrm.so
(==) fglrx(0): Capabilities: 0x00000000
(==) fglrx(0): CapabilitiesEx: 0x00000000
(==) fglrx(0): cpuFlags: 0x4000001f
(==) fglrx(0): OpenGL ClientDriverName: "fglrx_dri.so"
(WW) fglrx(0): No DRM connection for driver fglrx.
(II) fglrx(0): [drm] DRM buffer queue setup: nbufs = 100 bufsize = 65536
(==) fglrx(0): UseFastTLS=0
(==) fglrx(0): BlockSignalsOnLock=1
(II) fglrx(0): Direct rendering disabled
(--) Depth 24 pixmap format is 32 bpp
(II) do I need RAC? No, I don't.
(II) resource ranges after preInit:
[0] 0 0 0xfbb00000 - 0xfbbfffff (0x100000) MX[B]
[1] 0 0 0xfbcf0000 - 0xfbcfffff (0x10000) MX[B]
[2] 0 0 0xd0000000 - 0xdfffffff (0x10000000) MX[B]
[3] -1 0 0x00100000 - 0x3fffffff (0x3ff00000) MX[B]E(B)
[4] -1 0 0x000f0000 - 0x000fffff (0x10000) MX[B]
[5] -1 0 0x000c0000 - 0x000effff (0x30000) MX[B]
[6] -1 0 0x00000000 - 0x0009ffff (0xa0000) MX[B]
[7] -1 0 0xfbff0000 - 0xfbffffff (0x10000) MX[B]
[8] -1 0 0xfaff0000 - 0xfaffffff (0x10000) MX[B]
[9] -1 0 0xfbeff000 - 0xfbefffff (0x1000) MX[B]
[10] -1 0 0xfbdfec00 - 0xfbdfec7f (0x80) MX[B]
[11] -1 0 0xfbdff000 - 0xfbdff07f (0x80) MX[B]
[12] -1 0 0xfbdff400 - 0xfbdff47f (0x80) MX[B]
[13] -1 0 0xfbdff800 - 0xfbdfffff (0x800) MX[B]
[14] -1 0 0xfbce8000 - 0xfbcebfff (0x4000) MX[B]
[15] -1 0 0xfbaf9000 - 0xfbaf9fff (0x1000) MX[B]
[16] -1 0 0xfbaf4000 - 0xfbaf7fff (0x4000) MX[B]
[17] -1 0 0xfbafa800 - 0xfbafa8ff (0x100) MX[B]
[18] -1 0 0xfbafb000 - 0xfbafbfff (0x1000) MX[B]
[19] -1 0 0xfbafc000 - 0xfbafcfff (0x1000) MX[B]
[20] -1 0 0xfbaff000 - 0xfbaff0ff (0x100) MX[B]
[21] -1 0 0xfbafd000 - 0xfbafdfff (0x1000) MX[B]
[22] -1 0 0xfbafe000 - 0xfbafefff (0x1000) MX[B]
[23] -1 0 0xfbaff800 - 0xfbaffbff (0x400) MX[B]
[24] -1 0 0xfbb00000 - 0xfbbfffff (0x100000) MX[B](B)
[25] -1 0 0xfbcf0000 - 0xfbcfffff (0x10000) MX[B](B)
[26] -1 0 0xd0000000 - 0xdfffffff (0x10000000) MX[B](B)
[27] 0 0 0x000a0000 - 0x000affff (0x10000) MS[B](OprU)
[28] 0 0 0x000b0000 - 0x000b7fff (0x8000) MS[B](OprU)
[29] 0 0 0x000b8000 - 0x000bffff (0x8000) MS[B](OprU)
[30] 0 0 0x0000d000 - 0x0000d0ff (0x100) IX[B]
[31] -1 0 0x0000ffff - 0x0000ffff (0x1) IX[B]
[32] -1 0 0x00000000 - 0x000000ff (0x100) IX[B]
[33] -1 0 0x0000e800 - 0x0000e8ff (0x100) IX[B]
[34] -1 0 0x0000ff00 - 0x0000ff0f (0x10) IX[B]
[35] -1 0 0x000001f0 - 0x000001f0 (0x1) IX[B]
[36] -1 0 0x000001f0 - 0x000001f7 (0x8) IX[B]
[37] -1 0 0x000001f0 - 0x000001f0 (0x1) IX[B]
[38] -1 0 0x000001f0 - 0x000001f7 (0x8) IX[B]
[39] -1 0 0x00008000 - 0x0000800f (0x10) IX[B]
[40] -1 0 0x00009000 - 0x00009003 (0x4) IX[B]
[41] -1 0 0x0000a000 - 0x0000a007 (0x8) IX[B]
[42] -1 0 0x0000b000 - 0x0000b003 (0x4) IX[B]
[43] -1 0 0x0000c000 - 0x0000c007 (0x8) IX[B]
[44] -1 0 0x0000d000 - 0x0000d0ff (0x100) IX[B](B)
[45] 0 0 0x000003b0 - 0x000003bb (0xc) IS[B](OprU)
[46] 0 0 0x000003c0 - 0x000003df (0x20) IS[B](OprU)
(II) fglrx(0): driver needs X.org 7.1.x.y with x.y >= 0.0
(WW) fglrx(0): could not detect X server version (query_status=-1)
(EE) fglrx(0): atiddxDriScreenInit failed, GPS not been initialized.
(WW) fglrx(0): ***********************************************
(WW) fglrx(0): * DRI initialization failed! *
(WW) fglrx(0): * (maybe driver kernel module missing or bad) *
(WW) fglrx(0): * 2D acceleraton available (MMIO) *
(WW) fglrx(0): * no 3D acceleration available *
(WW) fglrx(0): ********************************************* *
(II) fglrx(0): FBADPhys: 0x0 FBMappedSize: 0x10000000
(==) fglrx(0): Write-combining range (0xd0000000,0x10000000)
(II) fglrx(0): FBMM initialized for area (0,0)-(1728,8191)
(II) fglrx(0): FBMM auto alloc for area (0,0)-(1728,1680) (front color buffer - assumption)
(II) fglrx(0): Largest offscreen area available: 1728 x 6511
(NI) fglrx(0): ASIC Family ID/Name:75/RS780

(NI) fglrx(0): ASIC Rev ID/Name:19/Not RV6xx family Device

(NI) fglrx(0): ASIC Device ID/Name:0X9610/Not RV6xx family Device

(NI) fglrx(0): CPLIB is initialized.

(==) fglrx(0): Backing store disabled


:( Ok, I'm stumped - I've installed the drivers but I just cannot get 3d to work properly. XFCE's compositing effects work and are fast if the windows are small, but slow down if they are larger. Windows are *really* slow if I turn off XFCE's compisiting effects!

:confused: :confused:

Anyone know what is going on?!

ljones

bridgman
03-22-2009, 06:45 PM
The kernel driver isn't starting up so you have no acceleration. The driver is saying that it can't find the X server version, which I don't remember seeing before.

Someone else was trying to get slamd working here a couple of days ago, let's see if we can find those posts.

ljones
03-22-2009, 07:38 PM
The kernel driver isn't starting up so you have no acceleration. The driver is saying that it can't find the X server version, which I don't remember seeing before.

Someone else was trying to get slamd working here a couple of days ago, let's see if we can find those posts.

Maybe my X server is too old? Though it is the one that slamd64 comes with. Wonder where those other posts are, I've not come across 'em yet.

:)

ljones

bridgman
03-22-2009, 07:51 PM
It's wierd, I'm not finding them either. Even if the discussion was on IRC rather than Phoronix it would have shown up in the logs. I found *this* thread easily enough ;)

Ah well... can you pls pastebin your entire log as well as dmesg output and maybe we'll find a clue there.

ljones
03-23-2009, 01:45 PM
How do I pastebin? I tried to put my log and dmesg in seperate posts here, but they are too big x.x

I could always try zipping them and uuencoding them, would that work? :)

ljones

bridgman
03-23-2009, 03:35 PM
Go to www.pastebin.ca or any of the other locations -- I think there are pastebin sites around the world). That site will let you paste in a big block of text, then when you save will give you a link to the resulting page.

ljones
03-23-2009, 03:54 PM
Y'learn something new every day :) . I had never heard of a pastebin before!

Anyways, here is the output of dmesg and the X log file;

First of all, dmesg;
http://www.pastebin.ca/1369456

And then the X server's log;
http://www.pastebin.ca/1369457

:) hope this helps!

ljones

ArTourter
03-23-2009, 05:16 PM
From your dmesg dump:
fglrx: module license 'Proprietary. (C) 2002 - ATI Technologies, Starnberg, GERMANY' taints kernel.
fglrx: Unknown symbol pci_disable_msi
fglrx: Unknown symbol pci_enable_msi


This is the same error as on Slackware 32bit. The requirements for the AMD kernel module have changed in 9.1 and above, without any mention in the changelog: it now requires PCI_MSI enabled in the kernel and this is not the case in the standard Slackware kernel (or Slamd64). You will need to recompile your kernel with this turned on for the newer fglrx to work. Otherwise you have to go back to 8.12 or prior.

Slackware-current with 2.6.28.8 has now PCI_MSI enabled by default in the kernel and so does Slamd64-current. However I haven't tried fglrx in -current yet (do not want to upgrade to kde4 just yet).

Hope this is helpful

ljones
03-23-2009, 06:56 PM
From your dmesg dump:


This is the same error as on Slackware 32bit. The requirements for the AMD kernel module have changed in 9.1 and above, without any mention in the changelog: it now requires PCI_MSI enabled in the kernel and this is not the case in the standard Slackware kernel (or Slamd64). You will need to recompile your kernel with this turned on for the newer fglrx to work. Otherwise you have to go back to 8.12 or prior.

Slackware-current with 2.6.28.8 has now PSI_MSI enabled by default in the kernel and so does Slamd64-current. However I haven't tried fglrx in -current yet (do not want to upgrade to kde4 just yet).

Hope this is helpful

Yep looks like I'll have to compile a kernel. I checked the normal slackware current version (it is currently 2.6.28.8 as you say) however the slamd64 kernel current appears to be different (example here - ftp://ftp.gtlib.gatech.edu/pub/slamd64/slamd64-current/slackware/a/ ), they are using 2.6.28.7 . I tried installing this on the offchance that PSI_MSI would be enabled, but no luck, installing the newer kernel made no difference, no accelration.

Guess its time to download a kernel and compile it !

EDIT
Now things might have worked -- after reinstalling the drivers at the command line and typing aticonfig --initial things may be working (plus reboot). Glxinfo now says "yes" to direct rendering; glxgears gives me around 521 FPS, while fgl_glxgears comes in at just a little more, around 644 FPS. Though I have no idea if those numbers are bad, good or about normal!

As for KDE 4 I see its packages are in current, however I have no idea which packages to install! (I don't think *everything* KDE 4 in slackware needs is in the KDE/ directory).

Should anyone need it btw after installing the new kernel bits 'n' pieces (from here - ftp://ftp.gtlib.gatech.edu/pub/slamd64/slamd64-current/slackware/a/ ) and installing these files :


kernel-firmware-2.6.28.7-noarch_slamd64-1.tgz
kernel-generic-2.6.28.7-x86_64_slamd64-1.tgz
kernel-huge-2.6.28.7-x86_64_slamd64-1.tgz
kernel-modules-2.6.28.7-x86_64_slamd64-1.tgz
kernel-source-2.6.28.7-noarch_slamd64-1.tgz


I also adjusted my lilo.conf (and and ran lilo) to point at /boot/vmlinuz-huge-2.6.28.7

Here's the X server log after doing the above:
http://www.pastebin.ca/1369730

And the output of dmesg:
http://www.pastebin.ca/1369732

uname -r reports 2.6.28.7

If this is all working now I must type it all out as a sequence of steps so I can follow it through! :) And the system didn't restart if I exited X :) !

ljones

ArTourter
03-23-2009, 07:02 PM
Yep looks like I'll have to compile a kernel. I checked the normal slackware current version (it is currently 2.6.28.8 as you say) however the slamd64 kernel current appears to be different (example here - ftp://ftp.gtlib.gatech.edu/pub/slamd64/slamd64-current/slackware/a/ ), they are using 2.6.28.7 . I tried installing this on the offchance that PSI_MSI would be enabled, but no luck, installing the newer kernel made no difference, no accelration.

Guess its time to download a kernel and compile it !

Hmm, if the 2.6.28.7 in slamd64 does not work, then I am not quite sure what else needs to be done as from the config file (http://www.slackware.org.uk/slamd64/slamd64-current/kernels/huge.s/config)
CONFIG_PCI_MSI=y

Do you actually get the same error, or do you get something else?

ArTourter
03-23-2009, 07:16 PM
I have a reply to prior your edit somehow stuck in moderation (probably due to a url I included)

but if fgl_glxgears actually runs, that means that the fglrx driver is running which seems to be a huge improvement to your previous situation.

regarding the framerate, I would not know as it is not something I really care about myself.

Out of interest, did you recompile your kernel at all or this is still the stock slamd64-current kernel where you only recompile the AMD driver and initialised it?

BTW, to explain the version difference between slackware and slamd64, the changes appear first in slack and are then ported to slamd64. 2.6.28.8 should appear in slamd64-current at the next batch of changes unless .9 or newer is released since then and slackware has it in the tree.

Kano
03-23-2009, 07:42 PM
You really use lilo? I currently switched over to grub2 for next system and you do not even use grub? With something like update-grub to integrated kernels automatically... Wow, that is really a legacy system.

ArTourter
03-23-2009, 08:26 PM
@Kano
go and troll somewhere else, will ya?

Kano
03-23-2009, 08:39 PM
Well best install grub if possible lilo is really outdated. You have to reinstall it everytime, grub only needs config file updates.

ArTourter
03-23-2009, 08:59 PM
yes because simply typing lilo is really hard and complicated.

listen, you use what you want and let every body else use what works for them, ok? your arrogant attempt at evangelism is out of place.

as I said earlier, troll somewhere else.

ljones
03-24-2009, 02:40 PM
x.x

There still seems to be just one more problem!

If I run X and (let's say) fgl_glxgears as root, no problems.

But as a normal (non-root) user, X crashes on first run with a blank screen, I have to stop it and restart it then it works. But then fgl_glxgears fails as non-root;


libGL error: open DRM failed (Operation not permitted)
libGL error: reverting to (slow) indirect rendering
X Error of failed request : GLXBadDrawable
Major opcode of failed request: 144 (GLX)
Minor opcode of failed request: 5 (X_GLXMakeCurrent)
Serial number of failed request: 45
Current serial number in output stream: 45


BTW I used the slamd64-current kernel.

I don't get the error message as root, and direct rendering works - but only as root!

EDIT
Ok, I've half solved the one with the fgl_glxgears not working as root with the following:


Section "DRI"
Mode 0666
EndSection


Though X as non root still fails first time as non-root!

Looks like there's something wrong in dmesg; (look at lines 360 onwards);

dmesg --
http://pastebin.ca/1370762

And even when X was working, I noticed this:

(from /var/log/Xorg.0.log)

AIGLX error: dlopen of /usr/lib64/xorg/modules/dri/fglrx_dri.so failed (/usr/lib64/xorg/modules/dri/fglrc_dri.so : cannot open shared object file : no such file or directory)


But I managed to fix that missing file with ' ln -s /usr/X11R6/lib64/modules/dri/fglrx_dri.so /usr/lib64/xorg/modules/dri/fglrx_dri.so ! :)

X Still died on 1st run though as non root.....

ljones

Kano
03-24-2009, 02:44 PM
Well I only needed a dri hack for an older ati-driver, usually it is controlled by this xorg.conf section, which seems to be default in newer xservers:

Section "DRI"
Mode 0666
EndSection

ljones
03-24-2009, 03:25 PM
I managed to solve some of the other problems such as the missing file, fixed with ' ln -s /usr/X11R6/lib64/modules/dri/fglrx_dri.so /usr/lib64/xorg/modules/dri/fglrx_dri.so ' .

X starting still seems to be a bit of a problem -- it gets even more weird! I rebooted, and after logging in as non-root (from cold) and starting X went in ok. But leaving X as non-root and then attempting to go back in again (as non root) failed as in my previous post (no screen, errors etc). Prehaps this is a bug in the drivers?

Weird, not sure why that would happen!

ljones

mounty
04-09-2009, 10:00 AM
weird! I rebooted, and after logging in as non-root (from cold) and starting X went in ok. But leaving X as non-root and then attempting to go back in again (as non root) failed as in my previous post (no screen, errors etc). Prehaps this is a bug in the drivers?This fixes X for me; I can run fgl_glxgears as a normal user:Section "dri"
Group "video"
Mode 0666
EndSectionThen add myself to group video.