New ATI drivers - black screen, system freeze and crash. Anyone in the same boat?

[zolookas]

Xorg with fglrx crashes with X1650. Screen goes
black. Keyboard becomes unresponsive. System
is thoroughly hung and requires hard reset to recover. Card works older fglrx included in ubuntu 7.10 and with vesa/"ati" driver in ubuntu 8.04 (although "ati" driver colour depth seems to be less than 24bit). I am using DVI connection. Is anyone else in the same boat and has some solutions? Because i didn't find any people who had found a workaround.

I've discussed this problem in mandriva:
http://forum.mandriva.com/viewtopic.php?p=430804

Other bug reports in unofficial ati bugtracker:
http://ati.cchtml.com/show_bug.cgi?id=1015
http://ati.cchtml.com/show_bug.cgi?id=910

[legume]

[QUOTE=zolookas;28021]Xorg with fglrx crashes with X1650. Screen goes
black. Keyboard becomes unresponsive. System
is thoroughly hung and requires hard reset to recover. Card works older fglrx included in ubuntu 7.10 and with vesa/"ati" driver in ubuntu 8.04 (although "ati" driver colour depth seems to be less than 24bit). I am using DVI connection. Is anyone else in the same boat and has some solutions? Because i didn't find any people who had found a workaround.

Try setting the GART/aperture size in your BIOS to 512.

Rather than a hard reset you can often use magic SysReq by pressing and keep holding down alt and SysRq then pressing and releasing s then u then b.

This will (s)ync your drives (u)n/remount them readonly and then re(b)oot.

[Kano]

I had similar issues when switching from ati->fglrx or when ModeLines have been in the xorg.conf (tested only X700SE PCI-E). fglrx drivers are not even able to init the hardware correctly.

[erdnussflipp]

Hey Guys,


Great that somebody experienced the same issues as i do.

In my case its absolutely the same, with all drivers i tried.


I'll hereby give you some detailed information what i did, and what my system specs are.

System specs:

MB: Asus Blitz Formula Special Edition
CPU: Intel C2D E6850
VC: Asus HD 2900 XT
MEM: 4 GB OCZ Special Ops DDR2-Ram - 800Mhz

VGA monitor (CRT)

tried os:
Ubuntu 7.10 (gutsy) x86_64
Ubuntu 8.04 (hardy) x86_64

drivers installed:
xorg-driver-fglrx - native gutsy / native hardy repository's
latest linux ati driver from ati.com

HowTo followed:
Gutsy:
http://wiki.cchtml.com/index.php/Ubuntu_Gutsy_Installation_Guide
- both methods

Hardy:
http://wiki.cchtml.com/index.php/Ubuntu_Hardy_Installation_Guide
- both methods

BOTH: different howto's from the Ubuntu forums.

every single HowTo had been proceed with a fresh installed Ubuntu system, neither 7.10 nor 8.04 worked.

here is a paste bin of one of my xorg.0.log's during the tries : http://pastebin.com/f1e13bff8

no errors/clues what so ever.

after installing various variants of HowTo's on any single boot the system just stays still on loading xorg/gnome. Sometimes with a gnome background, sometimes with a totally black screen.

i didn't try <Ctrl> + <Alt> SysRq
i didn't try setting bios features to other values

i will do so and post any outcome whatsoever

Hopefully someone gets a clue reading this.

Regards,

erdnussflipp

[oblivious_maximus]

I just bought a Sapphire 3650 last week. This week I'm still using my Asus 8600gt (which I've wanted rid of since the day I bought it, but it functions properly for the most part). I get similar lockups to the OP with the 3650 whenever I log out of KDE(3) or try to start a second, additional Xserver (with X :1 -layout layoutname). As the keyboard locks up along with the rest of X, the Magic SysReq key is no help at all :( .

I plugged the 3650 back in the other day to try this possible fix (http://www.phoronix.com/forums/showpost.php?p=26853&postcount=134) but it made no difference for me. Stopping atieventsd entirely is not an option I'm willing to consider because then the fan is constantly speeding up and slowing down in a loop and the higher speed is irritatingly loud.

I've only used the fglrx 8.3 driver so far, I tried the Debian packages in unstable(which are completely missing the /etc/ati/authatieventsd.sh file), and I've built Debian packages from the .run fglrx package on ATI's site (and the phorogit scripts because the .run package by itself wouldn't build the packages).

The rest of my system is comprised of:
Asus M3A AMD770 motherboard
X2 4400+ (Brisbane with a defective CPU temp sensor)
4GB ddr2-800
Promise SATA 300-TX2 controller w/1 PATA drive attached.
2 other PATA drives attached to the motherboard's PATA port. No SATA drives :(

I'm using Debian Lenny 64bit, KDE 3.5.8, and a Debian 2.6.24-4 kernel compiled from the linux-source package in sid. I installed a few X11 packages from experimental (libx11-xcb-dev and whatever else it pulled in as dependencies) a while back to compile Compiz and they're still installed. Otherwise my X11 should be whatever version is in testing now.

If I'm leaving out some info that someone might find useful please let me know.
edit:
X -version

X Window System Version 1.3.0
Release Date: 19 April 2007
X Protocol Version 11, Revision 0, Release 1.3
Build Operating System: Linux Debian (xorg-server 2:1.3.0.0.dfsg-12lenny2)
Current Operating System: Linux basestar 2.6.24-4-amd64 #1 SMP PREEMPT Wed Mar 5 17:01:51 EST 2008 x86_64
Build Date: 19 January 2008
--snipped bug reporting info--
Module Loader present

edit2: the OP might want to try checking for any BIOS updates their motherboard might need. I just checked for mine and it seems there is a newer version posted, with the changelog reading: "fixed the issue that the system freezing when using some certain VGA card." I haven't tried it yet but I will shortly. I know there's another person who posted about lockups like this (I think in the 8.3 driver thread) who updated their BIOS and fixed it, so my fingers are massively crossed here. will post results.

[hehe2]

Hi,

I got some freeze and crash too with the latest ati drivers ...

I've post about my issue there (http://www.phoronix.com/forums/showthread.php?t=8280) ...

I've not seen any new bios for my motherboard last time I've checked but I'll do it again, who knows ...

Still using my old computer (xp1800+ nvidia 7600GS agp ...) cause I've no pci express card except for my two new atis which aren't working :X

Best regards.

[c247]

690G ati x1200 IGP WAS using ati proprietary but had problems after upgrading from fedora 7 to 8 ( this is x86_64 btw ) I try to use yum for everything.

X was loop, lock, freeze, kernel panic, crash, bsod whatever you want to call it. I had to hard reboot when X started "looping" is how I would describe it. It was not only when playing games, but frequently or always when playing games (quake3 engine). The lock appeared randomly, but as I fiddled with X settings, it just got worse.

Uninstalled ati proprietary, and removed all previous "video drivers". Then rebooted into generic X just to be certain it isn't bad hardware. This is a new machine and has been great until the fedora upgrade.

So I tried radeonhd, looks really nice, but no 3D yet.

However, new kernel today

2.6.24.3-50.fc8

and new fglrx package from livna

kmod-fglrx-8.471-3.8.03.lvn8
kmod-fglrx-2.6.24.3-50.fc8-8.471-3.8.03.lvn8

Yes! so far so good with games. I have not tried AIGLX enable yet. Let me know if you have a similar experience. I wouldn't call this fixed yet, but I ran urbanterror(q3 engine) with its default settings and it looked and performed well with no sign of lock.

fglrxinfo

OpenGL vendor string: ATI Technologies Inc.
OpenGL renderer string: ATI Radeon X1200 Series
OpenGL version string: 2.1.7412 Release

Good luck

-- Update --

Its been hours now and no lock ups, no sign of any problems.

After all this experimenting while it was crashing I learned more about all the different drivers available for this card. I never tried the actual xorg radeon driver. Does anyone recommend it over the proprietary ( fglrx provided by livna ) ? To install it would I need to disable livna, so I get xorg-drv-radeon from fedora update channel?

-- Update --

Bad news. I just had a hard lock while playing urbanterror. It could be the framerate setting I have been setting low to try to improve smoothness. It is 80 by default, and I put it on 40 when I got the lock. ( will set back to 80 and try again ) Will also try setting the acceleration settings always on provided in amdcccl. In fc7 when I had no problems with games I think those were set always on? Could it be that urbanterror needs to do a release for 2.6.24 kernels?

brigman (in post #10) was right about my distro's authatieventsd.sh set incorrectly. :0.Xauth was in /var/gdm in fedora 8 not /var/lib/gdm Maybe having this set properly will help?

-- Update --

Tried all the xorg radeon stuff, no 3D yet. Went back to AMD proprietary download. No livna. Freshrpms instead. No locks while playing urbanterror, and very smooth. One important difference, no amarok running in the background this time. Not 100% sure yet, but I bet my hard locks are a combination of pulseaudio and the fglrx driver. I am using all the integrated devices on my mobo. The reason I think this is because I just tried a fresh vlc from freshrpms, and when I closed it, it caused the same sort of audio loop I hear when the machine hard locks. Had to kill -9 <pulseaudio pid> to get the loop to stop. No hard lock up yet. Very interesting.

[zappbrannigan]

MB: Asus Blitz Formula Special Edition
CPU: Intel C2D E6850
VC: Asus HD 2900 XT
MEM: 4 GB OCZ Special Ops DDR2-Ram - 800Mhz

...

after installing various variants of HowTo's on any single boot the system just stays still on loading xorg/gnome. Sometimes with a gnome background, sometimes with a totally black screen.

Have you read my (http://www.phoronix.com/forums/showthread.php?t=7507) thread ? My 4GB 64bit system hangs also with a black screen. I fixed my mtrr table. Please dump yours with cat /proc/mtrr
Or try to turn off memory remap in your bios.

[oblivious_maximus]

some good advice I should have followed (because I knew full well!): never use any Windows-based BIOS update utilities.

Anyway! having since realized I could just use a USB flash drive to install the BIOS from through "EZFlash", I flashed the new 0701 BIOS onto my brand new board. After checking that it would still boot my old setup, I threw in the HD3650, installed the fglrx 8.3 packages I built last week, edited /etc/ati/authatieventsd.sh to point to /var/run/xauth, and finally found absolutely no difference to the lockups on logout and when starting an additional Xserver.

zappbrannigan: I read the thread you linked but I don't see any info on what you did to your mtrr... Here's the output of cat /proc/mtrr on my system (but with my 8600gt installed, I don't know if that matters):
cat /proc/mtrr
reg00: base=0x00000000 ( 0MB), size=2048MB: write-back, count=1
reg01: base=0x80000000 (2048MB), size=1024MB: write-back, count=1
reg02: base=0xc0000000 (3072MB), size= 256MB: write-back, count=1I'm pretty sure I don't have any memory remap options in my BIOS, if there's something I can do with the mtrr please let me know.

[bridgman]

o_m, there's a fix/workaround/whatever which seems to handle the "locking up on exit" issue. The root problem seems to be that atieventsd looks for a certain file, which seems to be in different places on different distros, and fails "ungracefully" if it doesn't find the file. The fix involves editing a script to point to the correct location of the file in your distro. Not sure it works for everyone but so far it's running about 5 for 5.

have you tried this ? If not I should be able to dig it up again and maybe we can make it a sticky.

EDIT - here the last place I posted about it : http://www.phoronix.com/forums/showthread.php?p=27696#post27696

[BlueKoala]

Bridgman, that sounds like that is the exact issue I am having, although the lockups seem to be ""random"". Only when the videocard is being somewhat stressed on my system. I'll mess around with this and see what happens. Thanks a lot!

[zappbrannigan]

I get similar lockups to the OP with the 3650 whenever I log out of KDE(3) or try to start a second, additional Xserver (with X :1 -layout layoutname).
I'm sorry. I missed this two details. Your problem is different.


cat /proc/mtrr
reg00: base=0x00000000 ( 0MB), size=2048MB: write-back, count=1
reg01: base=0x80000000 (2048MB), size=1024MB: write-back, count=1
reg02: base=0xc0000000 (3072MB), size= 256MB: write-back, count=1I'm pretty sure I don't have any memory remap options in my BIOS, if there's something I can do with the mtrr please let me know.
Your system has 4GB ram and your bios do memory remapping by default. But your mtrr table handles only 3328MB. Please cat /proc/meminfo.

[oblivious_maximus]

bridgman: I did try editing authatieventsd.sh to point to /var/run/xauth right after updating my BIOS - but I must thank you anyway for mentioning that(thanks!), because I've just realized I never restarted atieventsd afterwards, only kdm. Presumably that would be necessary for the change to make any difference eh? There's always a little detail I've missed somewhere! I'll try this again shortly.

edit: I installed fglrx 8.3, stopped atieventsd, edited /etc/ati/authatieventsd.sh and started atieventsd again, all in single user mode before X started. Unfortunately it doesn't seem to have made any difference, it's still locking up on logout. I rebooted and tried again but it still locked up. /edit

Your system has 4GB ram ... But your mtrr table handles only 3328MBthat doesn't sound very good :(
here you go:
$cat /proc/meminfo
MemTotal: 4055724 kB
MemFree: 2719492 kB
Buffers: 1812 kB
Cached: 921208 kB
SwapCached: 0 kB
Active: 458556 kB
Inactive: 785172 kB
SwapTotal: 522072 kB
SwapFree: 522072 kB
Dirty: 272 kB
Writeback: 0 kB
AnonPages: 320856 kB
Mapped: 78920 kB
Slab: 38944 kB
SReclaimable: 22380 kB
SUnreclaim: 16564 kB
PageTables: 11396 kB
NFS_Unstable: 0 kB
Bounce: 0 kB
CommitLimit: 2549932 kB
Committed_AS: 626108 kB
VmallocTotal: 34359738367 kB
VmallocUsed: 59328 kB
VmallocChunk: 34359677947 kB
HugePages_Total: 0
HugePages_Free: 0
HugePages_Rsvd: 0
HugePages_Surp: 0
Hugepagesize: 2048 kB

[damdim]

I also had a system freeze with fglrx and after removing the line BusID "PCI:1:0:0" from xorg.conf I haven't had a freeze yet.

Update:
I have just had a freeze:(

[Angus]

My freeze happens whenever the server tries to start up. I don't have the ability to tweek that UDA/SIDEPORT crap, so I'm stuck. I hope ATI fixes this problem soon, because I'm tired of looking at this aspect ratio from outer space (1400x1050). I'm going to remember this the next time I shop for a video card. I've also not forgotten when the ATI Rage would freeze my system if I had the colour set to 32-bits.

[bridgman]

Angus, that sounds like a completely different problem. I looked back through the thread and didn't find any reference to UDA or sideport, can you fill us in ?

How is the aspect ratio tied into all this ?

If you are freezing at startup is there anything relevent in the logs ?

[Angus]

Angus, that sounds like a completely different problem. I looked back through the thread and didn't find any reference to UDA or sideport, can you fill us in ?

How is the aspect ratio tied into all this ?

If you are freezing at startup is there anything relevent in the logs ?

Really? Oh, perhaps I read about the UDA/SIDEPORT thing somewhere else. UDA is supposedly when the video card uses shared memory, and SIDEPORT is when it uses its on-board memory. I also read that if the card is configured for SIDEPORT only, the driver causes a freeze. Others have had the good fortune of BIOSes that allow them to select among those modes, but mine does not. Here's something to look at: http://wiki.cchtml.com/index.php/Troubleshooting#Computer_Freezes_while_using_fglrx _.28UMA_and_SIDEPORT.29

The reason I have an aspect ratio from hell is that the resolution that gives me the closest ratio to 16:10 offered by sax2 is 1400x1050. I hate that! I'd prefer a 1280x1024 monitor over this stretched widescreen, but the only 4:3 monitors we have around here are analogue only, and my card is DVI only.

I could try running it again and post those log messages you are looking for, if you like. I don't know where to look though.

[bridgman]

Really? Oh, perhaps I read about the UDA/SIDEPORT thing somewhere else. UDA is supposedly when the video card uses shared memory, and SIDEPORT is when it uses its on-board memory. I also read that if the card is configured for SIDEPORT only, the driver causes a freeze. Others have had the good fortune of BIOSes that allow them to select among those modes, but mine does not.

Thanks, Angus. Guessing that you are running an RS690 aka X1200/1250/1270 ?

The reason I have an aspect ratio from hell is that the resolution that gives me the closest ratio to 16:10 offered by sax2 is 1400x1050. I hate that! I'd prefer a 1280x1024 monitor over this stretched widescreen, but the only 4:3 monitors we have around here are analogue only, and my card is DVI only.

Ahh, you're probably falling back to the Vesa driver I guess ? That would explain the limited modes.

Just curious, what is the native resolution of your monitor ?

I could try running it again and post those log messages you are looking for, if you like. I don't know where to look though.

You mentioned Sax2 - are you running OpenSuSE ?

[Angus]

Thanks, Angus. Guessing that you are running an RS690 aka X1200/1250/1270 ?

No, it's an HD2600Pro.


Ahh, you're probably falling back to the Vesa driver I guess ? That would explain the limited modes.

Just curious, what is the native resolution of your monitor ?

1680x1050


You mentioned Sax2 - are you running OpenSuSE ?

Yes I am.

I don't suppose you think you know a solution.

[bridgman]

No, it's an HD2600Pro.

(native resolution) 1680x1050

(running OpenSuSE ?) Yes I am.

I don't suppose you think you know a solution.

OK, this helps. If you're running an HD2600 then UMA and Sideport are not an issue for you anyways -- they just affect the IGP parts (graphics integrated into the Northbridge part of the chipset on your motherboard).

Is your card AGP or PCIE (PCI Express) ?

Is it safe to assume there are no built-in graphics on your motherboard ?

Can someone jump in and point out where the relevant logs are for OpenSuSE ? I *think* they're in the same place on most Linux distros but don't want to mislead.

[Angus]

OK, this helps. If you're running an HD2600 then UMA and Sideport are not an issue for you anyways -- they just affect the IGP parts (graphics integrated into the Northbridge part of the chipset on your motherboard).

Is your card AGP or PCIE (PCI Express) ?


PCIE.

Is it safe to assume there are no built-in graphics on your motherboard ?


It'd be a very unsafe assumption. There is on-board graphics. From what I can tell from the BIOS, they don't matter if there is a video card installed.

[c247]

Not using AIGLX at this point, just wanted to get urbanterror working.

I'm feeling stupid for not trying this first, I was running all 3D options on in fc7, but it wasn't required to play games.

Set all 3D options on (always on) using amdcccle !

more here

http://www.phoronix.com/forums/showthread.php?p=28247#post28247

I played for about 30 minutes no problems. It was locking withing the first 10min. or less consistently.

If you are having any kind of hard locks or freezing regardless if its games, try this first.

Fedora 8 ATI x1250 IGP 128MB

Good Luck.

[Angus]

OK, this helps. If you're running an HD2600 then UMA and Sideport are not an issue for you anyways -- they just affect the IGP parts (graphics integrated into the Northbridge part of the chipset on your motherboard).

Is your card AGP or PCIE (PCI Express) ?

Is it safe to assume there are no built-in graphics on your motherboard ?

Can someone jump in and point out where the relevant logs are for OpenSuSE ? I *think* they're in the same place on most Linux distros but don't want to mislead.

In the absence of any clarification of the location of these logs, I decided to look into /var/logs/, and I found Xorg.0.log. Here's what the end of it looked like:

(II) fglrx(0): driver needs X.org 7.1.x.y with x.y >= 0.0
(II) fglrx(0): detected X.org 7.1.0.0
(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: 0xc0000000 FBMappedSize: 0x10000000
(WW) fglrx(0): Failed to set up 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)-(1728,1050) (front color buffer - assumption)
(II) fglrx(0): Largest offscreen area available: 1728 x 7141
(==) fglrx(0): Backing store disabled
(II) Loading extension FGLRXEXTENSION
(II) Loading extension ATITVOUT
(**) fglrx(0): DPMS enabled
(WW) fglrx(0): Textured Video not supported without DRI enabled.
(II) LoadModule: "glesx"
(II) Loading /usr/lib64/xorg/modules//glesx.so

Let me know if you want to see more of the file.

[GerbilSoft]

I've got a similar problem on my T60p [2623-DDU, FireGL V5200]. Since around v8.41, fglrx loads fine, but after five minutes or so, the screen goes black. The backlight is still on, but the display is blank. The system is still responsive, but nothing's visible. Switching to console and back doesn't help. Any ideas what might be causing this and/or how to fix it? (I'm currently running the latest radeon driver from git, and while it is stable, I'd like to be able to use 3D acceleration right now.)

[bridgman]

I think Dave Airlie might have the same laptop -- if so 3D can't be far away ;)

Seriously, please keep feeding us information re: anything else that might be happening around the time the screen goes blank. I don't remember seeing this specific problem before, but if anyone else has please sing out.

[GerbilSoft]

Seriously, please keep feeding us information re: anything else that might be happening around the time the screen goes blank. I don't remember seeing this specific problem before, but if anyone else has please sing out.

It seems to be completely random. Sometimes, it happens when the KDE splash screen's onscreen; other times, it happens while the system has been idle for some time. Then other times, it happens while I'm in the middle of actually doing something (which is REALLY irritating).

Though I did notice one thing; it seems as if the time is cumulative throughout X sessions. Suppose the blanking always happens at 10 minutes of usage. If I start X for 7 minutes, then exit X, the screen will blank 3 minutes into the next X session if I haven't rebooted. I'll have to do some actual timing with a stopwatch to check this.

EDIT: I ran a test. The system was idle for 30 minutes, and the screen didn't blank. So I guess it has something to do with GLX or some other library that causes things to screw up.

EDIT 2: It's definitely GLX. I ran glxgears for testing to make sure hardware acceleration was enabled. Less than a minute after closing glxgears, I got a white screen. The system was still responsive, but nothing showed up onscreen.

[bridgman]

Interesting. Is there any chance a screen saver is kicking in and failing somehow ?

[GerbilSoft]

xscreensaver is set to a time interval of 15 minutes. When I did the idle check, the screen saver did activate, and it functioned normally.

I did some more tests with GL apps, and almost immediately (up to a minute or so later) after running any GL app, the black/white/corrupted/etc screen appeared. I believe this didn't happen with 8.40 (the last fglrx I used), so it might have something to do with the new libGL introduced in 8.41.

[erdnussflipp]

Have you read
my (http://www.phoronix.com/forums/showthread.php?t=7507) thread ? My 4GB 64bit system hangs also with a black screen. I fixed my mtrr table. Please dump yours with cat /proc/mtrr
Or try to turn off memory remap in your bios.



here you go with mine:

reg00: base=0xd0000000 (3328MB), size= 256MB: uncachable, count=1
reg01: base=0xe0000000 (3584MB), size= 512MB: uncachable, count=1
reg02: base=0x00000000 ( 0MB), size=4096MB: write-back, count=1
reg03: base=0x100000000 (4096MB), size= 512MB: write-back, count=1
reg04: base=0x120000000 (4608MB), size= 256MB: write-back, count=1

[GerbilSoft]

It seems the black screen problem was a hardware issue. I had my ThinkPad's system board replaced due to a broken VGA port, and the black screen problem no longer happens. :)

fglrx is still rather slow at 2D, so I'm sticking with radeon for now. Hopefully radeon/radeonhd will gain 3D acceleration support for R5xx soon! :)

[oblivious_maximus]

I've just tried again with my HD3650 and only 2GB of RAM installed. I pointed authatieventsd.sh to /var/run/xauth/ and rebooted, but when I logout of KDE it still locks up so completely that even raising elephants with the magic sysreq key doesn't help. This was still with fglrx 8.3 btw.

[Talavis]

I've just tried again with my HD3650 and only 2GB of RAM installed. I pointed authatieventsd.sh to /var/run/xauth/ and rebooted, but when I logout of KDE it still locks up so completely that even raising elephants with the magic sysreq key doesn't help. This was still with fglrx 8.3 btw.
I wrote XDM_AUTH_MASK=/var/run/xauth/A$1* after a post on planetkde and has never had a problem since.

[Dandel]

I read most of this... I've been having the same issues myself on both 8.3 and 8.4, of course i lock terminal when i'm not at the computer. The lockups usually happen when the system is left running while logged in after well over 48 hours... should be worth nothing i am running on a Radeon X1900GT.

[oblivious_maximus]

I wrote XDM_AUTH_MASK=/var/run/xauth/A$1*that's exactly what I meant by "pointed authatieventsd.sh to /var/run/xauth", but thanks.

I'll be trying the 3650 again sometime soon, now that I'm using i386 Debian and not amd64. Reading the recent fglrx threads has me feeling less than enthused about getting the Radeon working though.

[jwhowarth]

I am seeing the same problem with Fedora 8 x86_64 on a Tyan Tomcat K8E-SLI with ATI Radeon X1650 Pro. The monitor will randomly go blank and the machine will freeze until it is rebooted (although it sometime reboots on its on). The messages log shows the following errors before this happens...

May 1 16:26:40 bromo kernel: BUG: soft lockup - CPU#1 stuck for 11s! [X:2977]
May 1 16:26:40 bromo kernel: CPU 1:
May 1 16:26:40 bromo kernel: Modules linked in: rfcomm l2cap bluetooth autofs4 fuse sunrpc nf_conntrack_netbios_ns nf_conntrack_ipv4 ipt_REJECT iptable_filter ip_tables nf_conntrack_ipv6 xt_state nf_conntrack xt_tcpudp ip6t_REJECT ip6table_filter ip6_tables x_tables dm_mirror dm_multipath dm_mod ipv6 floppy pcspkr e100 k8temp forcedeth hwmon tg3 serio_raw mii fglrx(P)(U) i2c_nforce2 i2c_core button sr_mod sg cdrom pata_amd sata_nv ata_generic pata_acpi libata sd_mod scsi_mod raid456 async_xor async_memcpy async_tx xor raid1 ext3 jbd mbcache uhci_hcd ohci_hcd ehci_hcd
May 1 16:26:40 bromo kernel: Pid: 2977, comm: X Tainted: P 2.6.24.5-85.fc8 #1
May 1 16:26:40 bromo kernel: RIP: 0010:[<ffffffff88198af0>] [<ffffffff88198af0>] :fglrx:_ZN4Asic16Is_WPTR_equ_RPTR19ConditionSucces sfulEv+0x0/0x60
May 1 16:26:40 bromo kernel: RSP: 0018:ffff81013d5d1c80 EFLAGS: 00200206
May 1 16:26:40 bromo kernel: RAX: ffffffff881f0670 RBX: 00000000007ffff9 RCX: 0000000000000000
May 1 16:26:40 bromo kernel: RDX: ffffc200013f14c0 RSI: ffffc200013f1020 RDI: ffff81013d5d1cf8
May 1 16:26:40 bromo kernel: RBP: 0000000000000043 R08: ffffffff881f0670 R09: 0000000000000043
May 1 16:26:40 bromo kernel: R10: ffff8100bf124ba0 R11: ffffffff810f9fec R12: ffff8100bf124ba0
May 1 16:26:40 bromo kernel: R13: ffffffff810f9fec R14: 00000000007ffff9 R15: 0000000000200282
May 1 16:26:40 bromo kernel: FS: 00002aaaaaad17b0(0000) GS:ffff81013fc01780(0000) knlGS:00000000f6f7aa50
May 1 16:26:40 bromo kernel: CS: 0010 DS: 0000 ES: 0000 CR0: 000000008005003b
May 1 16:26:40 bromo kernel: CR2: 0000003853aba380 CR3: 000000012a577000 CR4: 00000000000006a0
May 1 16:26:40 bromo kernel: DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
May 1 16:26:40 bromo kernel: DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400
May 1 16:26:40 bromo kernel:
May 1 16:26:40 bromo kernel: Call Trace:
May 1 16:26:40 bromo kernel: [<ffffffff88197e3f>] :fglrx:_ZN4Asic9WaitUntil15WaitForCompleteEv+0x1f/0xa0
May 1 16:26:40 bromo kernel: [<ffffffff8819ba72>] :fglrx:_ZN6AsicR616ASICIdleInternalEN4Asic15idle_W aitMethodE+0xa2/0x1e0
May 1 16:26:40 bromo kernel: [<ffffffff881806fb>] :fglrx:_ZN10QS_PRIVATE10submitListEP9CMMDriverP10_ QS_PARAM_+0xfb/0x460
May 1 16:26:40 bromo kernel: [<ffffffff88196945>] :fglrx:_ZN4Asic7PM4idleENS_15idle_WaitMethodE+0x55/0x90
May 1 16:26:40 bromo kernel: [<ffffffff88191526>] :fglrx:_ZN15QS_PRIVATE_CORE7PM4idleEN4Asic15idle_W aitMethodE+0x26/0x50
May 1 16:26:40 bromo kernel: [<ffffffff88180a8a>] :fglrx:_ZN10QS_PRIVATE11synchronizeEv+0x2a/0x30
May 1 16:26:40 bromo kernel: [<ffffffff881888e3>] :fglrx:_Z8uCWDDEQCmjjPvjS_+0x363/0xf80
May 1 16:26:40 bromo kernel: [<ffffffff88145f4a>] :fglrx:firegl_cmmqs_CWDDE_32+0x20a/0x310
May 1 16:26:40 bromo kernel: [<ffffffff88145037>] :fglrx:firegl_cmmqs_CWDDE32+0x67/0xf0
May 1 16:26:40 bromo kernel: [<ffffffff88144fd0>] :fglrx:firegl_cmmqs_CWDDE32+0x0/0xf0
May 1 16:26:40 bromo kernel: [<ffffffff88138d26>] :fglrx:firegl_ioctl+0x1b6/0x230
May 1 16:26:40 bromo kernel: [<ffffffff810abe61>] do_ioctl+0x55/0x6b
May 1 16:26:40 bromo kernel: [<ffffffff810ac0ba>] vfs_ioctl+0x243/0x25c
May 1 16:26:40 bromo kernel: [<ffffffff810ac124>] sys_ioctl+0x51/0x71
May 1 16:26:40 bromo kernel: [<ffffffff8100c005>] tracesys+0xd5/0xda
May 1 16:26:40 bromo kernel:
May 1 16:26:52 bromo kernel: BUG: soft lockup - CPU#1 stuck for 11s! [X:2977]
May 1 16:26:52 bromo kernel: CPU 1:

I've tried using noacpi, noapictimer and acpi_use_timer_override but none of those help. What are the other folks here seeing in there message logs from just before the blank screen/reboot?
Jack

[c247]

Sorry guys, I'm not seeing the black screen or the auto reboot, but I do have these problems persisting.

Hard Lockup when I play something like quake3 engine. ( probably anything with 3D accel ) The lockup happens in several situations, and I must hard reboot (using restart button on tower) system is non responsive. the cases are as follows:

I always use XFCE and gdm

1. I have been working for hours as my dev user in standard X session, and I have probably watched a ton of flash videos ( I mention this because I think it is related somehow ) Then I log out of that X session and login as my game user, again standard X XFCE session. After I start a game, like UrbanTerror, I will usually get a lockup within 10 minutes. After I restart and login as my game user, I have no problems playing the game for hours. There is something about that reboot that resets something, maybe onboard the card itself. I have tried rmmod fglrx and then modprobe to reload it, but that doesn't make the game stable like a fresh reboot.

2. I'm playing UrbanTerror and have just changed game system settings, like fps, compress textures, or vertical sync. The game will start to jitter and then the system will lockup. Again its definitely the card and the driver right?

3. I have seen your black screen that causes me to reboot, but that happens when I move between virtual consoles. I will go to console 1 and put the machine in init 3, but then if I go to virtual console 7 ( where X was spawning ) I get a black screen, and I'm not able to do anything but reboot (ctrl+alt+del)

I was hoping that this latest April driver would solve problems for us IGP guys, but I guess not. I'm using

Gigabyte ga-ma69G-s3h using the IGP ati x1250 and all other integrated devices.

I definitely had to remove pulseaudio, it was causing all kinds of weird problems, but the ATI driver is the culprit.

I'm waiting for radeonHD driver with 3D support, I bet it destroys ATI proprietary. Gotta have hope :D

Oh yeah, most importantly. If you are having a lot of problems with fedora 8 x86_64 I recommend this especially if you did an upgrade from fc7 to fc8

go to init 2 or 3 ( get out of X )

Then do a yum groupremove "X Window System" "KDE" whatever else you have related to X.

Then do a yum groupinstall "X Window System" "KDE" "XFCE" whatever.

Thanks to jwhowarth for setting me straight on fc7 packages that still apply to fc8. I had gotten rid of almost all fc7 packages, but he pointed out that a lot of them are still used.

I know this takes a long time and it seems like overkill, but trust me, it fixed a bunch of my problems. Granted my upgrade from fc7 to fc8 was eventful to say the least, but doing the above you are certain to reload all the latest libs and binaries.

Clay

[jwhowarth]

I hope you made a list of the fc7 rpms that you deleted. There are quite a few packages in Fedora 8 that still have the fc7 suffix. Critical things like grep for instance. You really should use 'yum list extras' to find those packages that aren't part of the current release of fedora.

[c247]

Wow you are right, a bunch of core fc7 packages made it back into my fc8 when I did the X related group installs. Crazy! I had gotten rid of all but a few fc7 packages and my system was stable. How Odd :D Where did I find an fc8 grep ? I'm totally confused now :D

Clay

[jwhowarth]

Look at the list of packages in the Fedora 8 distribution...

http://mirror.stanford.edu/fedora/linux/releases/8/Everything/x86_64/os/Packages/

Basically, the fedora developers didn't bump the fc version on those packages that didn't change versions or revisions for Fedora 8. You can generally assume that after upgrading Fedora and doing a yum update that all of the current packages will be installed. You then can use 'yum list extras' to identify those packages that are either depreciated out of the current fedora release or those that you installed yourself outside of the normal yum repos.

[Ren Höek]

In the absence of any clarification of the location of these logs, I decided to look into /var/logs/, and I found Xorg.0.log. Here's what the end of it looked like:

(II) fglrx(0): driver needs X.org 7.1.x.y with x.y >= 0.0
(II) fglrx(0): detected X.org 7.1.0.0
(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: 0xc0000000 FBMappedSize: 0x10000000
(WW) fglrx(0): Failed to set up 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)-(1728,1050) (front color buffer - assumption)
(II) fglrx(0): Largest offscreen area available: 1728 x 7141
(==) fglrx(0): Backing store disabled
(II) Loading extension FGLRXEXTENSION
(II) Loading extension ATITVOUT
(**) fglrx(0): DPMS enabled
(WW) fglrx(0): Textured Video not supported without DRI enabled.
(II) LoadModule: "glesx"
(II) Loading /usr/lib64/xorg/modules//glesx.so

Let me know if you want to see more of the file.

I had the same messages in my log...
If you try
$ modprobe -vf fglrx
and get something like
install /sbin/lrm-video fglrx
the solution is to modify the file "/etc/modprobe.d/lrm-video". Comment out the line with fglrx so it looks like
#install fglrx /sbin/lrm-video fglrx $CMDLINE_OPTS
and reboot.

I hope it works for you too
Ren