Prev: A tiny linux + SQL
Next: Recommendations for Linux "WE"-replacement?
From: Ohmster on 26 Mar 2010 02:03
Hey guys, I got a problem that is just breaking my heart and I cannot
stand it! I need your help, bad...

I have a nice system setup with Fedora 12, older machine, PIII
Coppermine, 1.5Gb SDRAM, Nvidia GeForce 6500 card, and I had compiz-
fusion working so good with this system, rotating cube, animated skydome,
the cool atomic screensaver, and it was such a treat to use. Well, I
usually install all of the updates when they appear in my gnome session
and this was no different, everything seemed fine but then when my
screensaver came up, the screen was locked up. Nothing would get it free,
not even Ctl-Alt-Backspace. I ssh'ed in from another machine, killed the
x server and could see my monitor go all to hell, but I could not get a
terminal back on it again. I had to reboot.

This happens every time now so while in gnome, I tried to rotate the cube
in compiz. It started to rotate and was very slow and "chuggish". It got
stuck but after a couple of minutes, I could see the CPU fire applet
making flames again, as I wiped the mouse over the screen, it "wiped
off" the frozen image and showed my desktop again. I tried to run
glxgears and they ran just fine, not slow at all. But anything compiz or
3D screensaver, and it is game over. It also takes longer for the desktop
to come back now after running startx. (I like runlevel 3).

I reinstalled kmod-nvidia-PAE.i686 1:190.53-3.fc12 and it did not do a
thing to help, as soon as I try to rotate the compiz cube, the cpu flame
applet freezes and then I get a partially turned cube frozen on the
screen for many minutes, like 3-5, and then the desktop comes back.

I have been trying to look through the messages file for any errors,
cannot find the X11 log file because I don't know where to look, but what
gives here and can I ever get my sleek, smooth running compiz-fusion
desktop back again? If only I could just "undo" whatever the hell those
50 or so updates were I would be fine.

Help!

--
~Ohmster | ohmster59 /a/t/ gmail dot com
Put "messageforohmster" in message body
(That is Message Body, not Subject!)
to pass my spam filter.
From: Ohmster on 26 Mar 2010 02:12
Ohmster <root(a)dev.nul.invalid> wrote in
news:Xns9D4714C47A250MyBigKitty(a)216.196.97.131:

> I have been trying to look through the messages file for any errors,
> cannot find the X11 log file because I don't know where to look, but
> what gives here and can I ever get my sleek, smooth running
> compiz-fusion desktop back again? If only I could just "undo" whatever
> the hell those 50 or so updates were I would be fine.
>
> Help!

Here is a little more info, I grepped messages for nvidia and the older
messages shows 4x mode for the AGP card but now it seems to be forced
into 0x mode? What does that mean? Oh please help!

Older messages:

[ohmster(a)ohmster log]$ sudo grep nvidia messages-20100228
Feb 21 21:35:09 ohmster yum: Installed: kmod-nvidia-2.6.31.12-
174.2.22.fc12.i686-190.53-1.fc12.5.i686
Feb 21 21:35:18 ohmster yum: Installed: kmod-nvidia-2.6.31.12-
174.2.22.fc12.i686.PAE-190.53-1.fc12.5.i686
Feb 21 21:35:19 ohmster yum: Updated: kmod-nvidia-190.53-1.fc12.5.i686
Feb 21 21:35:22 ohmster yum: Updated: kmod-nvidia-PAE-190.53-
1.fc12.5.i686
Feb 21 21:35:22 ohmster yum: Erased: kmod-nvidia-2.6.31.9-
174.fc12.i686.PAE
Feb 21 21:35:54 ohmster yum: Erased: kmod-nvidia-2.6.31.9-174.fc12.i686
Feb 21 21:40:06 ohmster kernel: nvidia: module license 'NVIDIA' taints
kernel.
Feb 21 21:40:06 ohmster kernel: nvidia 0000:01:00.0: PCI INT A -> Link
[LNKA] -> GSI 11 (level, low) -> IRQ 11
Feb 21 21:43:15 ohmster kernel: nvidia 0000:01:00.0: putting AGP V2
device into 4x mode
Feb 27 20:11:38 ohmster kernel: nvidia: module license 'NVIDIA' taints
kernel.
Feb 27 20:11:38 ohmster kernel: nvidia 0000:01:00.0: PCI INT A -> Link
[LNKA] -> GSI 11 (level, low) -> IRQ 11
Feb 27 22:10:14 ohmster kernel: nvidia 0000:01:00.0: putting AGP V2
device into 4x mode
[ohmster(a)ohmster log]$

Recent messages:

[ohmster(a)ohmster log]$ sudo grep nvidia messages
Mar 22 01:25:50 ohmster kernel: nvidia 0000:01:00.0: putting AGP V2
device into 4x mode
Mar 23 11:01:04 ohmster yum: Updated: nvidia-settings-1.0-3.4.fc12.i686
Mar 23 11:01:28 ohmster yum: Updated: nvidia-xconfig-1.0-2.fc12.i686
Mar 23 11:35:33 ohmster yum: Updated: 1:xorg-x11-drv-nvidia-libs-190.53-
4.fc12.i686
Mar 23 11:35:54 ohmster yum: Installed: 1:kmod-nvidia-2.6.32.9-
70.fc12.i686.PAE-190.53-3.fc12.i686
Mar 23 11:35:59 ohmster yum: Updated: 1:xorg-x11-drv-nvidia-190.53-
4.fc12.i686
Mar 23 11:36:18 ohmster yum: Installed: 1:kmod-nvidia-2.6.32.9-
70.fc12.i686-190.53-3.fc12.i686
Mar 23 11:36:27 ohmster yum: Updated: 1:kmod-nvidia-190.53-3.fc12.i686
Mar 23 11:36:28 ohmster yum: Updated: 1:kmod-nvidia-PAE-190.53-
3.fc12.i686
Mar 23 11:40:06 ohmster yum: Erased: kmod-nvidia-2.6.31.12-
174.2.3.fc12.i686.PAE
Mar 23 11:45:59 ohmster yum: Erased: kmod-nvidia-2.6.31.12-
174.2.3.fc12.i686
Mar 25 00:28:54 ohmster abrtd: Package 'xorg-x11-drv-nvidia' isn't signed
with proper key
Mar 25 00:35:07 ohmster kernel: nvidia: module license 'NVIDIA' taints
kernel.
Mar 25 00:35:07 ohmster kernel: nvidia 0000:01:00.0: PCI INT A -> Link
[LNKA] -> GSI 11 (level, low)
Mar 25 00:38:15 ohmster kernel: nvidia 0000:01:00.0: putting AGP V2
device into 4x mode
Mar 25 02:28:02 ohmster kernel: nvidia: module license 'NVIDIA' taints
kernel.
Mar 25 02:28:02 ohmster kernel: nvidia 0000:01:00.0: PCI INT A -> Link
[LNKA] -> GSI 11 (level, low)
Mar 25 03:44:36 ohmster kernel: nvidia 0000:01:00.0: putting AGP V2
device into 4x mode
Mar 25 04:38:14 ohmster kernel: nvidia 0000:01:00.0: putting AGP V2
device into 0x mode
Mar 25 13:36:01 ohmster kernel: nvidia: module license 'NVIDIA' taints
kernel.
Mar 25 13:36:01 ohmster kernel: nvidia 0000:01:00.0: PCI INT A -> Link
[LNKA] -> GSI 11 (level, low)
Mar 26 01:27:49 ohmster kernel: nvidia 0000:01:00.0: putting AGP V2
device into 4x mode
Mar 26 01:52:16 ohmster yum: Installed: 1:kmod-nvidia-PAE-190.53-
3.fc12.i686
Mar 26 01:58:14 ohmster kernel: nvidia 0000:01:00.0: putting AGP V2
device into 0x mode
[ohmster(a)ohmster log]$

--
~Ohmster | ohmster59 /a/t/ gmail dot com
Put "messageforohmster" in message body
(That is Message Body, not Subject!)
to pass my spam filter.
From: J.O. Aho on 26 Mar 2010 02:56
Ohmster wrote:
> Ohmster <root(a)dev.nul.invalid> wrote in
> news:Xns9D4714C47A250MyBigKitty(a)216.196.97.131:
>
>> I have been trying to look through the messages file for any errors,
>> cannot find the X11 log file because I don't know where to look, but
>> what gives here and can I ever get my sleek, smooth running
>> compiz-fusion desktop back again? If only I could just "undo" whatever
>> the hell those 50 or so updates were I would be fine.
>>

Should say I don't have much to say, Xorg.0.log could have been good to link
to, as the out puts of /proc/driver/nvidia/registry,
/proc/driver/nvidia/warnings/* (not README), /proc/driver/nvidia/agp/card.

I assume you did kill all X processes and rmmod the nvidia driver ant then
started X again.

> Here is a little more info, I grepped messages for nvidia and the older
> messages shows 4x mode for the AGP card but now it seems to be forced
> into 0x mode? What does that mean? Oh please help!

As far as I know there is no 0x

Should be 1x, 2x, 4x.


> Older messages:
>
> [ohmster(a)ohmster log]$ sudo grep nvidia messages-20100228
> Feb 21 21:35:09 ohmster yum: Installed: kmod-nvidia-2.6.31.12-
> 174.2.22.fc12.i686-190.53-1.fc12.5.i686
> Feb 21 21:35:18 ohmster yum: Installed: kmod-nvidia-2.6.31.12-
> 174.2.22.fc12.i686.PAE-190.53-1.fc12.5.i686
> Feb 21 21:35:19 ohmster yum: Updated: kmod-nvidia-190.53-1.fc12.5.i686
> Feb 21 21:35:22 ohmster yum: Updated: kmod-nvidia-PAE-190.53-
> 1.fc12.5.i686

As you have less than 4G ram, don't use the PAE kernel, it will hurt you more
than you gain (your gain is nothing).


--

//Aho
From: J G Miller on 26 Mar 2010 08:50
On Fri, 26 Mar 2010 01:03:20 -0500, Ohmster wrote:

> But anything compiz or 3D screensaver, and it is game over.

There appears to be some sort of bug in the GL screensavers,
or underlying GL libraries, and this is distribution independent.

On a Mint Linux system which I am running with an ATI Radeon
video card, if the GL screensaver started up, eventually there
would be a serious memory fault and my root / file system
just disappeared.

At first I was having a panic attack that the disk was failing,
because on rebooting and fscking everything seemed fine, then
the machine would become unusable again. After a while I realized
that if I was using the machine for extended periods of time there
was no problem, but if I left the machine on its own and then
came back, then the problem arose. This immediately made me
think of the screen saver kicking in and a web search revealed
others having similar problems.

So have a look at your /var/log/kernel.log or equivalent and
see if you are getting nasty errors in there when the GL screensaver
or compiz GL special effects start running.

If so, disable the GL screensavers (regular old fashioned ones
not using GL do not cause a problem) and turn off compiz special
effects until they get a fix.

Also, have you posted this bug on the Fedora tracking bug system?
From: Ohmster on 26 Mar 2010 10:38
Steve Ackman <steve(a)SNIP-THIS.twoloonscoffee.com> wrote in
news:slrnhqonjh.c3a.steve(a)sorceror.wizard.dyndns.org:

[..]
> I know you're super busy, way too busy to read
> entire man pages, (actually, I'm surprised you have
> time to even read this group, but <shrug> even if you
> don't have time to read the whole man page, jump down
> to the bottom, and there's *usually* a FILES section
> to tell you where related files will be found.
>
> Let's say you don't know what the <command> actually
> is. You at least know some context, right? Here's
> another man command you should keep on your cheatsheet
> right next to your computer.
>
> $ man -k <keyword>
[..]

Steve,

A lot of this stuff I do know, but when I need it, that is when it
escapes me. What you have written is a super neat, refresher that is what
I call "reference material" so I will make a reference card out of it and
keep multiple copies on both machines. Thank you.

The Linux machine is a dedicated Fedora 12 server. I have a Windows 7
machine on the same LAN.

I am soooo bummed out man because compiz-fusion worked so good, sleek as
hell, not a glitch or a problem with it, and now the moment you do
anything with 3D desktop, it freezes up for many minutes. Whether you
will get your terminal back ever again without rebooting is anyones
guess. I can ssh into the machine with putty from Windows to see what is
going on. Top reveals that when this happens, compiz and xorg are at the
top of the CPU list, eating 63% or more CPU time and I have NO local
control of the system. Only remote login via ssh works after that.

This is NOT once in a while, but always now. If I boot to run level 3 and
stay out of X, I am good. If I use startx, I eventually get in and can
run a browser and even glx gears turn fast and give high frame rates
(Over 1,000, maybe up to 1,900), yet anytime you call compiz to go to
work, it is game over. Rotate the cube and it is game over. It will
rotate just a few degrees and the screen freezes. Wait a couple minutes
and you can refresh the screen (I tried wiping with the mouse as a
squeegee and it worked) and log out of your x session to the much more
stable CLI run level 3.

Thanks Steve!
--
~Ohmster
 |  Next  |  Last
Pages: 1 2 3
Prev: A tiny linux + SQL
Next: Recommendations for Linux "WE"-replacement?