FS#7513 - X freezes
Attached to Project:
Arch Linux
Opened by Urs Wolfer (uwolfer) - Monday, 25 June 2007, 14:47 GMT
Last edited by Alexander Baldeck (kth5) - Friday, 09 November 2007, 14:49 GMT
Opened by Urs Wolfer (uwolfer) - Monday, 25 June 2007, 14:47 GMT
Last edited by Alexander Baldeck (kth5) - Friday, 09 November 2007, 14:49 GMT
|
Details
Some users have the problem that the X server freezes from
time to time.
Please see this form thread for more information: http://bbs.archlinux.org/viewtopic.php?id=34369 This problem happens with nvidia driver for me. I think I have also had this problem once wiht my macbook, which has AFAIK i810 drives. |
This task depends upon
Closed by Alexander Baldeck (kth5)
Friday, 09 November 2007, 14:49 GMT
Reason for closing: Works for me
Additional comments about closing: Now that X.org 7.3 has been moved to [extra], more precise input on issues in needed again. Please open more specific bugs.
Friday, 09 November 2007, 14:49 GMT
Reason for closing: Works for me
Additional comments about closing: Now that X.org 7.3 has been moved to [extra], more precise input on issues in needed again. Please open more specific bugs.
AMD64 Athlon
ASUS K8V-MX with chipsets VIA K8M800 and VIA VT8237
2x512 DDR 400 MHz RAM
ATA 80GB SAMSUNG
17'' SAMSUNG SyncMaster 793s monitor
i use VIA drivers for X
32-bit arch linux
Intel Core 2 Duo T7200
256MB ATI MOBILITY RADEON X1400 (fglrx)
2x512 667MHz DDR2 SDRAM
recently i installed xorg 1.3.0 and i don't remember if i had a lockup since then. i'll let you know if it freezes again.
The thread on the forum talks about 2 problems:
#1) when running "X" EVERYTHING just freezes. C-M-F2(3,4 rtc) does not work, my computer gets locked up. the only option is to use "restart" button on Cabinet.
#2) exactly same as #1 but this time mouse moved, i mean mouse pointer does not freeze. but again, the solution is same as the #1.
Exactly problem #2 occured to me yesterday twice at random moments.
Exactly problem #1 occured yesterday two times out of three when trying to start wine so this one seems to be reproducible.
The things above happen while I'm working on my computer.
But for a few weeks now I have a similar problem when kdm asks me to login. I cannot write my password because there is no response to keyboard input. The mouse moves fine. I can also click the buttons in kdm(this is different from #2 above since there clicking the buttons has no effect) but if I choose restart from kdm options it freezes.This is quite rare (say once every 20 - 30 reboots).
I'm not sure if the 3 above are related since there are subtle differences but I reported them hoping that it would help the devs figuring out what's the problem.
Maybe a bug in nvidia proprietary drivers?
I also had hard freezes begin to appear when I upgraded to nvidia 100.14.11. The machine completely freezes with no log message. I can't even ssh into the box. I rolled back to 100.14.09 and they appear to have stopped.
I will also say that I am experiencing the other X freezing that people are complaining about. The mouse still moves, but does appear to interact with any elements on the screen. This problem seems to present itself after the computer has been idle for some time. I have a dual-core proc and I can ssh in to the box and look at the CPU stats in htop while this is going on, the two cores sitch back and forth on high cpu usage. Core0 will show the x process at 90+% cpu for a couple of seconds, then core1 will take on the load, then core0, etc. If you wait long enough, it seems that the system actually is responding to mouse clicks, just very slowly. One time I left the machine alone for about fifteen minutes and it appeared to have stopped, although I have not been able to recreate this.
I have tried to look for a pattern based on which apps were left open, whether beryl was running or not, etc. and I can't seem to find a common denominator other than it usually is occurring when I sit down at the machine after being away for some time.
AMD Athlon64 X2 5600+
Video Card = NVIDIA GeForce 7300 GS
chipset = NVIDIA nforce5 (MCP55)
AMD Athlon64 X2 5600+
Video Card = NVIDIA GeForce 7300 GS
chipset = NVIDIA nforce5 (MCP55)
It has happened 2x in the past two days, both times after coming back from being away. When I sat down, the high cpu usage was going on and had been for awhile based on my cpu usage meter. I am using the xorg-server from testing btw, in fact, my whole box runs testing and is up to date.
The programs running when this occurred were different each time.
yesterday
KDE, Beryl running but not active, kwalett, superkaramba and kmix were active in the tray. open apps were Opera, Amarok, Skype, Kmail and vmware had a windowsXP session running.
Today
KDE, Beryl running but not active, kwalett, superkaramba and kmix were active in the tray. open apps were Opera, Firefox and Konqueror.
I am going to try now without kdm loaded, to see if that has anything to do with it, and perhaps I should try rolling back the xorg-server. I wish I could trigger this bug at will so it would be easier to test various combinations.
Also, my logs show nothing, and I tried to attach gdb to the session and get a backtrace, but it is useless because of the stripped binaries.
http://bugs.archlinux.org/task/7230
It has happened 2x in the past two days, both times after coming back from being away. When I sat down, the high cpu usage was going on and had been for awhile based on my cpu usage meter. I am using the xorg-server from testing btw, in fact, my whole box runs testing and is up to date.
The programs running when this occurred were different each time.
yesterday
KDE, Beryl running but not active, kwalett, superkaramba and kmix were active in the tray. open apps were Opera, Amarok, Skype, Kmail and vmware had a windowsXP session running.
Today
KDE, Beryl running but not active, kwalett, superkaramba and kmix were active in the tray. open apps were Opera, Firefox and Konqueror.
I am going to try now without kdm loaded, to see if that has anything to do with it, and perhaps I should try rolling back the xorg-server. I wish I could trigger this bug at will so it would be easier to test various combinations.
Also, my logs show nothing, and I tried to attach gdb to the session and get a backtrace, but it is useless because of the stripped binaries.
I'll keep testing and report back.
Actually the kernek spits warnings when booting about not being able to claim some regions of memory, on my lappy while on the other stationary machine not.
* Ach32/Arch64
* kernel version
* xorg-server version
* full card name
* turbo cache / dedicated mem only
* special configs (like xinerama dualhead, twinvew etc)
No beryl please. That is to be considered unstable on all setups per se.
My Setup that locks up every now and then but works with xf86-video-nv:
* Arch64
* kernel: 2.6.22-rc5
* xorg-server 1.3.0.0 & 1.2.0
* nVidia Corporation GeForce Go 7400
* turbo cache (128M/512M)
* xinerama dualhead & single screen
My specs are:
* Arch32
* Kernel 2.6.21.5-ARCH
* xorg-server 1.3
* nVidia GeForce 7300 GS
* Single Display
The specs for my video card say it has turbo cache. There is 256MB on board and it says it can use up to 512MB. This is the link to my card http://www.gigabyte-usa.com/Products/VGA/Products_Spec.aspx?ClassValue=VGA&ProductID=2376&ProductName=GV-NX73G256D-RH
*kernel26 2.6.21.5-1 and kernel26suspend2 2.6.21.1-2
*xorg-server 1.3.0.0-1
*ati radeon mobility x1400 (fglrx 8.37.6-1)
*single display
most recently i had a lockup (where you can still move the mouse) at loading planeshift. i'll try to reproduce it again.
So anyway, when I re-enabled the power management in KDE, I have experenced ths lockup three times. I'll keep testing but i wanted to post this to see if it got any diagniostic juices flowing with anybody else watching this bug.
BTW, I was using 100.14.09 because 100.14.11 was very unstable for me.
> All of them share code, but they're maintained by 3 different maintainers, each with their own
> view on what hardware to support in what way.
hey JGC, thanks a lot buddy. i never knew that. xf86-video-unichrome cannot be used because my chipset(VT8237 on K8M800) is not supported: http://unichrome.sourceforge.net/
OpenChrome has my chipset but if i install it what "name" will i put in the "Driver" section of "/etc/X11/xorg.conf" ???
> in your configuration file. I think unichrome uses the same.
> That's why they're drop-in replacements for eachother.
ok, now i am using "openchrome" driver and from last 2 days no single X lockup or freeze :-), only "audacious" got a single but bad freeze and everything else is fine. i still want to keep a check for 1 full week so be ready for my next comment :0)
thanks a lot JGC
> only "audacious" got a single but bad freeze and everything else is fine.
> i still want to keep a check for 1 full week so be ready for my next comment :0)
OK... still same lockups with same intensity and same in number :-( ... this shows that there is no problem with the driver. this hardware is total-crap. please do not buy this one.
need to buy new graphics card :(
> this shows that there is no problem with the driver. this hardware is total-crap.
> please do not buy this one.
> need to buy new graphics card :(
JGC, isn't there the possibility that there is bug in present Xorg ?
Problem solved.
I removed the nvidia driver and compiled my own (0-9755). Now it's been working fine and not a single crash. So the problem seems to be a driver issue related to nvidia.
I've been using Gnome & Beryl extensively for 2 days with not a single crash or hard lock (it used to lock very often).
Now I'm noticing that every now and then windows such as: firefox, gnome-terminal, or Nautilus turn white and display no text or icons. If I minimize and maximize then everything is fine. Maybe this is the original issue that was crashing the Nvidia driver?? There's something wrong somewhere regardless, but at least no more hard locks. I hate having to press the reboot button, if you know what I mean. (After all instability is why I went from Win98 to Linux way back when in the year 2000, and it's one of the reasons why I've stayed with Linux. So this issue is very important.)
Also, i have rebuilt my nvidia module and it still happens.
https://bugs.launchpad.net/ubuntu/+source/xorg/+bug/51991
https://bugs.launchpad.net/ubuntu/+source/kdeutils/+bug/109507
How do I explain that other people with non-Nvidia cards are experiencing this? Easy, read my second comment. Now the windows turn white and all content disappears. This can be solved by minimizing and maximizing the affected window. So, obviously there is something else that's wrong, but at least the system doesn't hang (for me at least). It appears that this issue is related to the graphics driver and something else that is causing the crash.
> How do I explain that other people with non-Nvidia cards are experiencing this? Easy, read my
> second comment. Now the windows turn white and all content disappears. This can be solved by
> minimizing and maximizing the affected window.
well, your solution is for people using Desktops. you can not mimnimize/maximize a window in a Tiled Window Manager like stumpwm or ratpoison or wmii.
???
It's not really a solution, but it just happens to be what I'm doing and I'm only doing this because I don't know how to fix it. =S
Essentially what I'm saying is this:
1. The problem affects the driver but may not be directly caused by the video driver alone. Proof, it stopped on my computer when I compiled my own driver module. I'm using a little bit of an older driver so it may be that newer drivers may have made the issue worse?? I'm using 0-9755.
2. Even though the screen no longer freezes, windows now go all white and their contents disappears. This indicates that the problem still exists, it just no longer causes a hard crash on my particular system.
By the way, I don't think that it's related to superkaramba. Not only do I not use it now, but I did use it on Ubuntu and it did not crash on me.
[sda@black video]$ uptime
10:02:20 up 1 day, 9:06, 0 users, load average: 0.36, 0.38, 0.24
top - 10:02:39 up 1 day, 9:07, 0 users, load average: 0.26, 0.36, 0.23
Tasks: 72 total, 3 running, 68 sleeping, 1 stopped, 0 zombie
Cpu(s): 19.1%us, 7.6%sy, 0.0%ni, 71.9%id, 0.0%wa, 0.3%hi, 1.0%si, 0.0%st
Mem: 1035416k total, 1017000k used, 18416k free, 34508k buffers
Swap: 2104472k total, 0k used, 2104472k free, 576576k cached
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
5676 root 15 0 153m 66m 6008 S 19.1 6.5 31:27.53 X
xrestop - Display: localhost:0
Monitoring 13 clients. XErrors: 0
Pixmaps: 61312K total, Other: 47K total, All: 61360K total
res-base Wins GCs Fnts Pxms Misc Pxm mem Other Total PID Identifier
1000000 12 25 1 16 89 39616K 3K 39620K 5717 urxvt
0600000 220 5 3 314 652 8363K 23K 8386K ? Enlightenment
0400000 0 2 0 2 8 5658K 240B 5659K ? <unknown>
0800000 27 38 1 215 64 3358K 4K 3362K 5707 gkrellm
1200000 36 26 2 37 10 2824K 3K 2828K ? XMMS - 414. Various - (65:51)
0a00000 108 1 0 107 4 870K 2K 873K ? E-Exec
1600000 186 42 1 142 100 571K 8K 580K 2577
FS#7513: X freezes - SeaMonkey0e00000 19 2 0 10 3 27K 576B 27K ? E-ScreenShoot
0c00000 13 1 0 5 4 20K 432B 21K ? Emix
1a00000 1 1 0 0 0 0B 48B 48B ? xrestop
1400000 1 1 0 0 0 0B 48B 48B ? <unknown>
1800000 1 0 0 0 0 0B 24B 24B ? <unknown>
0200000 0 1 0 0 0 0B 24B 24B ? <unknown>
no freezes, no issues ever. I do had some "fleas" with "audacious" - xmms rocks. nVidia drivers from nvidia.com - libglx.so.100.14.09
no frills, no hassle... composite extension is loaded...
may be removal of libdri.so from /usr/lib/xorg/modules/extensions/ (and /etc/X11/xorg.conf) could help?
Anyway, at least we've got a proven xorg-6.9.* as a last resort... But it seems to me that the matter of all the thrills could be in running apps instead of drivers/xorg itself...
>... But it seems to me that the matter of all the thrills could be in running apps instead of drivers/xorg itself...
Well, I have Arch installed on the same computer twice. One native and one in Vmware on WinXP. The Vmware version hasn't had any problems since I loaded it up a few weeks ago. The only real difference in the programs, between the two systems, is Beryl... and OpenOffice come to think of it. So I suppose we should ask ourselves are we all running Beryl?? Perhaps there is an incompatability between beryl and several graphics drivers?
> So I suppose we should ask ourselves are we all running Beryl?? Perhaps there is an incompatability
> between beryl and several graphics drivers?
NO, i am not running Beryl. i never ran Beryl. i used to use GNOME, then Xfce, then Windowmaker, then wmii and today stumpwm. BTW, i had the same lockups on Gentoo 2007.0 and i do not even have Nvidia card, i have VIA.
i must mention that replacing "xf86-video-via" with "openchrome" driver have reduced the number X lock-ups to 20-30%. this solution was suggested by "Jan de Groot (JGC)" here. may JGC live long :-)
On my Laptop (specs below) running Beryl (and tried compiz), nVidia 7400 Go, Gnome Firefox latest released nVidia drivers (100.14.11), freezes occur randomly. It seems I can make the frequency increase by rapidly scrolling down USING THE MOUSE WHEEL in Firefox WHILE a page is still loading alot of data/images. There's no guarantee that this will cause a freeze, but it does 50% of the time. the other time that a freeze isn't caused, the screen starts flickering on the bottom quarter of the screen pretty bad -- like it's having a hard time redrawing it. I'm using the Avant window navigator down there, so maybe that's a clue.
Oh, and whenever the problem arises, I get the following messages (or similar) in /var/log/everything.log:
Jul 23 08:24:26 plunkerbunkey NVRM: Xid (0001:00): 6, PE0000 0fc0 00dedfdf 000083c8 00000000 00fbf8f1
Jul 23 08:24:26 plunkerbunkey NVRM: Xid (0001:00): 30, L2 -> L1
Sometimes the "L2 -> L1" will say "L1 -> L0" or something. sometimes the "30" is a "28" or something. When the system "locks up" the mouse moves but nothing else works except a hard power-off. I've not let the machine sit for hours to test and see if it comes back. When it doesn't lock up but just "flickers", I just restart X and all is well.
I've tried using the Epiphany and Opera browsers and it doesn't fix the issue.
Laptop Spec:
HP DV8000 (actually it's a dv8230us -- specs on HP's site)
Centrino Duo (T2300)
2GB RAM
nVidia 7400 Go with 256MB RAM
Dual 80GB drives
Wireless
CD/DVD, etc.
Spec:
* Ach32
* kernel 2.6.22.1
* xorg-server 1.2.0-5
* nvidia geforce 7600GT 256MB
Anyway, I tried the "idle=poll" boot parameter. I booted up and TRIED to crash things by scrolling too fast, etc. I couldn't get the machine to lock up. But I only used it for about 30 minutes at that sitting. The lockup is so random that just 30 minutes hasn't proven anything. I'll keep trying that though and see if it helps.
But something is definitely going on. (Oh, I switched to compiz instead of beryl and still got a lockup -- though that was BEFORE the idle=poll.)
All I know for sure is this:
a) It's been 4 days of heavy use and no more lockups. Solved by changing my video driver.
b) Screens going white and contents disappearing is related to Beryl, and more specifically the experimental add ons such as snow and tabs. I was able to reproduce this on Ubuntu by enabling these extra add-ons as well.
Jul 26 23:11:46 plunkerbunkey NVRM: Xid (0001:00): 6, PE0000 0000 beef4901 000018bc 00000000 00000000
Jul 26 23:11:46 plunkerbunkey NVRM: Xid (0001:00): 30, L2 -> L1
To fix this I just restart X and the flicker clears. I don't KNOW that the idle=poll is "saving" me from what WOULD have been a lockup, but the "idle=poll" is not STOPPING the nVidia error being logged. I don;t know what this error means, though so I really can't check anything else right now.
My system specs are listed above:
DV8000 Laptop, Centrino Duo, nVidia 7400 Go 256MB RAM, etc. etc.
I hesitate to change to the "open" drivers because of my need for acceleration for GL apps (Blender, a game or two, etc.).
I was hunting around nVidia's forums for more options and came across this:
http://www.nvnews.net/vbulletin/showthread.php?t=95090
That is a link to a message where someone has some code that they say locks up their machine. I compiled that code and ran it on a machine that has never locked up and on my laptop, which does lock up.
On the machine that never locks up -- the code ran fine and fast -- just spit out some lines and went fine. Nothing in the logs, no lockups, etc.
On my laptop -- which DOES lockup randomly -- I ran it and it STALLS at the line reading "glXMakeContextCurrent". What makes this even weirder is here is the "everything.log" dump of what happens when I run that program:
Jul 27 19:46:42 plunkerbunkey NVRM: Xid (0001:00): 13, 0003 beef3097 00004497 0000021c 00000000 00000002
Jul 27 19:46:46 plunkerbunkey NVRM: Xid (0001:00): 30, L0 -> L0
Jul 27 19:46:46 plunkerbunkey NVRM: Xid (0001:00): 9, Channel 00000020 Instance 00000000 Intr 00100000
Jul 27 19:46:46 plunkerbunkey ipw3945: Microcode SW error detected. Restarting.
Jul 27 19:46:46 plunkerbunkey ipw3945: request scan called when driver not ready.
Jul 27 19:46:46 plunkerbunkey NVRM: Xid (0001:00): 13, 0000 01016100 0000008a 0000057c 00ede9e3 00004000
Jul 27 19:46:46 plunkerbunkey NVRM: Xid (0001:00): 9, Channel 00000020 Instance 00000000 Intr 00100000
Jul 27 19:46:47 plunkerbunkey ipw3945: Error sending ADD_STA: time out after 500ms.
Jul 27 19:46:48 plunkerbunkey ipw3945: Can't stop Rx DMA.
Jul 27 19:46:48 plunkerbunkey ipw3945: Detected geography ABG (11 802.11bg channels, 13 802.11a channels)
Notice the 4-second delay in the logs when this started? That's the "stall" I mentioned above. The laptop screen flickered too! The blue, wireless "network" status light on my laptop (a DV8000) flickers when I run the program and then goes back to normal. Maybe this is related to why most of m lockups seem to occur when browsing/scrolling on the internet. To make it even "weirder", after I ran the program 4-5 times it eventually QUIT STALLING and no more messages got put in the logs! Does this give anyone any clues?
Also, following the lead in the ubuntu forums linked to above, I removed superkaramba, and three days have gone by with no problems. I suppose superkaramba could be using the offending commands, so it is still a driver issue, although this does not explain why people have reported this with hardware other than nvidia.
I was scrolling down a large page, reading as I went. As I scrolled to the bottom (using Firefox and the mouse wheel), the PC locked up but the mouse still moved. I waited and after about a minute, it "unlocked". When it did so the bottom quarter of the screen would "flicker" about once every second or two. All this phenomena is mentioned through my earlier posts.
What I did this time, though is to run the sample code I mention in my most previous to this one. I ran it and it would stall a few seconds. Each time the wireless connection would reset (I'm using the ipw3945 drivers). After running it 4 times, it stopped stalling a few seconds, stopped resetting the wireless driver/card AND the screen flickering stopped.
Does this shed any information on the subject for anyone?
I've got nvidia card & iwlwifi network wireless...
I'm fed up using a customized kernel (pacman -Syu break module for kernel: not same version)
oggb4mp3 - Has installing the 9755 driver helped on your system?
So maybe my bug is in superkaramba after all, or maybe one of the themes I am using. I'll keep on it.
Bad: I Googled this issue and this is actually a pretty old problem that you can see lots of even way back in 2005. It seems to be either related to the kernel or Xorg, because it started happening when Xorg started being introduced in Linux. It seems that under certain HARDWARE and software conditions Linux just crashes.... =( So sad.
My GeForce 6200 was fine with the 9755 driver, but I decided to do a small upgrade to a GeForce 8500 which didn't work under that driver. This resulted in me having to use the newest drivers and splat, die, crash. I reinstalled Ubuntu and experienced the same thing. In frustration I looked up the problem in the Nvidia forums which brings me to the good news.
Good: 1. I'm back on Arch. 2. Good news for the people with NVidia cards and VIA chipsets. There are some incompatibilities there and there is a patch which works for me. The patch is EXPERIMENTAL, but is provided by NVIDIA. And again, for me it seems to fix the problem. I've been running Compiz-Fusion without any crashes for a few hours. I'll keep you posted.
Here's a link to the NVidia forums. Scroll down until you see the NVidia employee linking to the patch. There are some INSTRUCTIONS there. The Nvidia patch is here:
http://www.nvnews.net/vbulletin/showthread.php?t=87918&page=7&highlight=freezes
Hopefully it helps a few of you out.
Oh hell, I'll also copy and paste for the lazy people.... =) lol The patch is also attached.
Quote:
I found that the workaround referred to earlier is incomplete in 100.14.xx, supporting reports of a regression since 1.0-96xx; could those of you seeing problems with 100.14.11 try the attached (experimental) patch? It should improve stability with NV44 and G72 based graphics cards on certain ATi chipsets.
To apply the patch, use:
# sh /path/to/NVIDIA-Linux-x86-100.14.11-pkg0.run --apply-patch /path/to/NVIDIA_kernel-100.14.11-336672.diff.txt
# sh NVIDIA-Linux-x86-100.14.11-pkg0-custom.run
>> I tried the patch, but unfortunately it didn't work. It froze rather quickly actually.....
Remember that this fix is only for people with motherboards that have ATI chipsets AND who also nave certain NVIDIA graphics card. I'm sorry that this didn't work for you though.
This problem has been around for 2 years, so I don't think that any 1 patch will help everyone unless it's a proper Xorg or Kernel patch. On the other hand this has helped some a few people, as you'll see if you read the forum.
For those that meet the above criteria try out the patch. It's been about a day for me and I'm solid, although I've noticed a small performance hit with Compiz-fusion.
More info:
http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=434900
http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=435100
http://www.nvidia.com/object/linux_display_archive.html
Apparently worked for this 8600 GTS (running compiz-fusion). Please report back if it worked for you.
6600GT
Turned on a screensaver (through the kde control panel)
updated xorg to 7.3 (testing repo)
updated nvidia driver to x.19 (testing repo)
I'll be gone for a week, so I can't really test it for longer until I get back, but I will post if it crashes on me....
-Fixed stability problems with some GeForce 8 GPUs
So it may be the solution to -some- of the stability problems reported here
* Ach32
* 2.6.22.9
* 1.4-3 (and 1.4-1)
* Intel i945GM
* xfce4, no composite, no beryl, no compiz
* happens only when I'm at the computer, never when the computer is sitting by itself
* Arch32
* 2.6.22.9-1
* xorg-server 1.2.0-5
* GeForce 7800GTX, nvidia 100.14.11-4
* xfce4, composite on, no beryl or compiz
* Happened only while scrolling in Firefox.
When I close a graphical session cleanly, with the Xfce4 logout, my console text is light blue, and all the other colors seem to be "shifted" in the same direction. If I log in and then out again the same happens, but light blue becomes dark blue. After the third time everything is black, but still works. If I don't exit cleanly, and just kill X with Ctrl-Alt-Backspace instead, colors are fine. Killing X also has the power to "heal" the console colors and turn them white again (seems like a bad RPG o.O ).
This last problem, anyway, seems like an unrelated bug that appeared in the same release.
However, I believe I have worked around it, the solution being turning DPMS off.
Just add the following to your .xinitrc or similar:
exec xset -dpms
I've already kept the machine running for 24 hours, during the night - no freeze when I wake up. I will try leaving it on again today to see what happens.
Hope this helps.
However, X started freezing for me when opening mplayer (or one of the mplayer frontends) fairly frequently...
What a shame.
In my case the instability of Nvidia 8600GTS (and other Geforce 8 GPUs) was solved by upgrading to nvidia and nvidia-utils 100.14.19 - 100.14.x before this was unstable.
Also I cannot switch to plain text console during freeze.
I use nvidia with Testing disabled.
I'm using xf86-video-intel, 945GM card (MacBook Core 2 Duo).
Never happened with old intel driver (xf86-video-i810).