Amd CCC...nexuiz... and X full CPU...but glxgears fix the problem?!

[karlpiuemeno]

Hi folks.... I finally managed to get (a little) work fglrx in my fedora...I had conflicts between runlevels..

but now I have another problem..

after boot, if I run the amd/ati control panel or a game....nothing happen... with top I see that X is at 84 % CPU and GLXINFO HI cpu usage too...

for unlocking the problem, I just run GLXGEARS and all works....

what happen? How to fix this issue?

I'm using the fglrx 8.44 by livna... same problem with 8.45 (livna) and official installer...

kernel: 2.6.23.14-107

# Xorg configuration created by livna-config-display

Section "ServerLayout"
Identifier "Default Layout"
Screen 0 "Screen0" 0 0
InputDevice "Keyboard0" "CoreKeyboard"
EndSection

Section "Files"
ModulePath "/usr/lib/xorg/modules"
EndSection

Section "ServerFlags"
Option "AIGLX" "on"
EndSection

Section "InputDevice"
Identifier "Keyboard0"
Driver "kbd"
Option "XkbModel" "pc105"
Option "XkbLayout" "it"
EndSection

Section "Device"
Identifier "Videocard0"
Driver "fglrx"
# Option "AGPMode" "8"
Option "RenderAccel" "on"
Option "XAANoOffscreenPixmaps"
Option "ColorTiling" "on"
Option "OpenGLOverlay" "off"
Option "VideoOverlay" "on"
Option "AGPFastWrite" "on"
Option "EnablePageFlip" "true"


Option "UseFastTLS" "0"
Option "BlockSignalsOnLock" "on"
Option "ForceGenericCPU" "off"
Option "KernelModuleParm" "locked-userpages=0"

EndSection

Section "Screen"
Identifier "Screen0"
Device "Videocard0"
DefaultDepth 24
SubSection "Display"
Viewport 0 0
Depth 24
EndSubSection
EndSection

Section "Extensions"
Option "Composite" "Enable"
EndSection

[Kano]

That error is in every driver since 8.41.7 and Xserver 1.4. If you use an older driver (with -ignoreABI override) then it works without that problem.

[karlpiuemeno]

That error is in every driver since 8.41.7 and Xserver 1.4. If you use an older driver (with -ignoreABI override) then it works without that problem.

please explain better....

where to put that string???

[Kano]

Depends on the DM you use. Do you use KDM, GDM or whatever?

[karlpiuemeno]

Depends on the DM you use. Do you use KDM, GDM or whatever?


gdm :) I use fedora like mom did it :D

[Kano]

For Debian you would use this - I will not install Fedora:

echo [server-Standard] >> /etc/gdm/gdm.conf
echo command=/usr/bin/X -audit 0 -ignoreABI >> /etc/gdm/gdm.conf

Also when you want to use 8.40.4 you will need this:

http://mlau.at/files/fglrx-842-2623.patch

[karlpiuemeno]

tnx for your reply... but doesn't work for me... :(


tried also ignoreabi in other ways (tnx google :rolleyes: ) but nothing to do :(

[rbmorse]

Same here (I see problem, Kano's fix doesn't work...unless I'm screwing it up which is possible). Ubuntu 8.04 (alpha-2).

I'd test it on my production machine running 7.10 except it has radeonhd, which I trust more than fglrx.

[karlpiuemeno]

so.....the only solution is running glxgears at every xorg start? WOW :D great fglrx driver -_-" :D

[Kano]

That fix is for using 8.40.4 or older. My fglrx script has it included. As long as your xorg.conf has already a driver fglrx statement you can execute it for testing with -v 8.39.4 or -v 8.40.4 as options. It does not yet work with those pure default xorg.conf files.

http://kanotix.com/files/install-fglrx-debian.sh

[rbmorse]

so.....the only solution is running glxgears at every xorg start? WOW :D great fglrx driver -_-" :D

Or, just don't run the control center (amdcccle).

If you do run the control center you have to kill it manually OR run glxgears after you close the CCC window.

Other than that, it works pretty well for me.

[karlpiuemeno]

Or, just don't run the control center (amdcccle).

If you do run the control center you have to kill it manually OR run glxgears after you close the CCC window.

Other than that, it works pretty well for me.

no.. xorg goes crazy also if I run only nexuiz...
for example..


I turn on my pc... ok desktop.... I open NEXUIZ... nothing happen.... I open terminal, glxgears... puff, nexuiz screen pops up!

same thing with amdccc.......

[rbmorse]

Ok...then don't do that <g>.

Have you filed a bug report? Sometimes it helps.

[wojtek5001]

It is bug #1003 http://ati.cchtml.com/show_bug.cgi?id=1003
I have also submited simple workaround there. It works for me.


I've also made some deeper investigation - compare strace output (strace glxinfo) at first time and second time. Here are results (buggy output is submitted in bugzilla):
Line 528 bad:
ioctl(5, 0x4008642a, 0x7fffb2c30cb0) = 0
not called in normal run

Line 576 good:
ioctl(5, 0x4008642a, 0x7fff22eeac50) = 0
not called in badly run

loop seems to begin on line 754, where in
normal run couple of:
ioctl(5, 0xc03064a6, 0x7fff22eecca0) is called,

and fews calls later output messages are printed.

In case of bug, loop begins:

select(4, [3], [3], NULL, NULL) = 1 (out [3])
writev(3, [{"+\1\1\0", 4}], 1) = 4
select(4, [3], [], NULL, NULL) = 1 (in [3])
read(3, "\1\2&\0\0\0\0\0\f\0`\1\0\0\0\0D\f\0\0\0\0\0\0\340\37>\245"..., 4096) = 32
read(3, 0x605eb4, 4096) = -1 EAGAIN (Resource temporarily unavailable)
select(4, [3], [3], NULL, NULL) = 1 (out [3])
writev(3, [{"\3\1\2\0\2\0\200\1\16\0\2\0\2\0\200\1", 16}], 1) = 16
select(4, [3], [], NULL, NULL) = 1 (in [3])
read(3, "\1\0\'\0\3\0\0\0*\0\0\0\1\0\0\1\377\377\377\377\0\ 0\0\0"..., 4096) = 76
read(3, 0x605eb4, 4096) = -1 EAGAIN (Resource temporarily unavailable)
select(4, [3], [3], NULL, NULL) = 1 (out [3])
writev(3, [{"\223\1\3\0\0\0\0\0\2\0\200\1", 12}], 1) = 12
select(4, [3], [], NULL, NULL) = 1 (in [3])
read(3, "\1\334)\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0d\0d\0\0\ 0\0\0"..., 4096) = 32
read(3, 0x605eb4, 4096) = -1 EAGAIN (Resource temporarily unavailable)
ioctl(5, 0x4008642a, 0x7fffb2c32ac0) = 0

(that strange ioclt 0x4008642a again ;)

Can someone explain that ioctl. I can't find it in any kernel sources, so it must be some proprietary AMD/ATI driver call (or I overlooked something?).

I probably shuld write that to bugzilla, but no one seems to care ...