Total freeze with 7300LE using NVidia drivers under Ubuntu

[atonello]

Well, after a long time of useless trying, reading this forum and browsing ubuntu's ones, unable to find a solution, I've finally decided to post my problem, hoping to find a solution...

I can't use NVidia drivers in Ubuntu 6.10 32 bit (but it was also the same in 6.06, even tried 32 and 64 bit, since it's a problem I'm regularly having since I built my current system on last september) for more than some hours (sometimes more, sometimes just for few minutes) before it will inexorably freeze.
With freeze I mean that mouse and keyboard stop responding, system clock on the panel is equally freezed; if music was being played then it keeps looping last audio frame.
Ctrl+Alt+backspace, Ctrl+Alt+F1, Alt+SysReq+s/b - they're all useless since the keyboard stops responding, the only exit way is the reset button...

If I use "nv" in xorg.conf the system is rock solid.

Also tried various workarounds (such as noapic, nolapic, in kernel boot; disabling cpu frequency scaling; etc.), with no luck.
(and, though if I'm using now the latest 9746, I've been struggling with this problem for months and, obviously, with different versions on NVidia drivers)

Thank you

P.S. Here are my system specs:
M/B Asus A8R32mvp (Ati RD580) + Asus NVidia 7300LE 256 mb + A64 X2 4200+
P.P.S. Ram passed very long tests with memtest86+ and system is also stable under WinXp (other than using Ubuntu + "nv" in xorg.conf).


I'm attaching today's nvidia-bug-report.log

[atonello]

Hmmm... Shall I assume there's no answer to my problem?

[Mephisto1982]

i'm having the same issue, although it doesnt seem to be a total freeze, but just an xorg lockup. sometimes my mouse is still moving, sometimes it isnt, but that's just about the only thing i can still do. Sometimes X even gets killed and restarts...
It only seems to happen when running opengl apps (games), and some games crash X more than others. I've played UT2004 for hours straight while Warcraft 3 seems to hang X in about 30% of all games.
I've only built this system a few weeks ago, and the old one was totally stable with the same software installation. current specs:
Intel Core 2 Duo E6400
2GB ddr2 ram
GeForce 7600 GT 256MB PCI-E
Fedora Core 6
with different versions of the official driver.

Got these from logs:

Backtrace:
0: /usr/bin/Xorg(xf86SigHandler+0x81) [0x80e53b1]
1: [0x220420]
2: /usr/lib/xorg/modules/drivers/nvidia_drv.so(_nv001972X+0x27) [0x5f2b4b]

Fatal server error:
Caught signal 11. Server aborting

Jan 1 16:12:30 chii kernel: Code: Bad EIP value.
Jan 1 16:12:30 chii kernel: EIP: [<00000000>] 0x0 SS:ESP 0068:efc32fa4
Jan 1 16:12:30 chii kernel: <1>BUG: unable to handle kernel NULL pointer dereference at virtual address 00000000
Jan 1 16:12:30 chii kernel: printing eip:
Jan 1 16:12:30 chii kernel: 00000000
Jan 1 16:12:30 chii kernel: *pde = 7a8f6067
Jan 1 16:12:30 chii kernel: Oops: 0000 [#6]
Jan 1 16:12:30 chii kernel: SMP
Jan 1 16:12:30 chii kernel: last sysfs file: /block/sdb/sdb2/stat
Jan 1 16:12:30 chii kernel: Modules linked in: ipv6 dm_mirror dm_mod video sbs i2c_ec button battery asus_acpi ac parport_pc lp parport snd_ymfpci gameport
snd_ac97_codec ac97_bus snd_seq_dummy bt878 snd_seq_oss snd_seq_midi_event snd_seq tuner snd_pcm_oss snd_mixer_oss bttv snd_pcm video_buf ir_common compat_io
ctl32 i2c_algo_bit btcx_risc tveeprom videodev v4l2_common v4l1_compat snd_opl3_lib snd_timer snd_hwdep snd_page_alloc snd_mpu401_uart snd_rawmidi snd_seq_de
vice sky2 snd soundcore serio_raw nvidia(P) ata_piix usbhid hid i2c_i801<1>BUG: unable to handle kernel NULL pointer dereference pcspkr ff_memless iTCO_wdt i
TCO_vendor_support xpad i2c_core sr_mod cdrom sg ata_generic ahci pata_jmicron libata sd_mod scsi_mod ext3 jbd mbcache ehci_hcd ohci_hcd uhci_hcd
Jan 1 16:12:30 chii kernel: at virtual address 00000001
Jan 1 16:12:30 chii kernel: CPU: 0
Jan 1 16:12:30 chii kernel: EIP: 0060:[<00000000>] Tainted: P VLI
Jan 1 16:12:30 chii kernel: EFLAGS: 00013287 (2.6.19-1.2891.fc7 #1)
Jan 1 16:12:30 chii kernel: EIP is at 0x0
Jan 1 16:12:30 chii kernel: eax: ef05f55c ebx: fffffff7 ecx: 00000006 edx: ef05f55c
Jan 1 16:12:30 chii kernel: printing eip:
Jan 1 16:12:30 chii kernel: 00000001
Jan 1 16:12:30 chii kernel: *pde = 774e5067
Jan 1 16:12:30 chii kernel: esi: 00000001 edi: bff77f6c ebp: f3d12fb0 esp: f3d12fa4
Jan 1 16:12:30 chii kernel: ds: 007b es: 007b ss: 0068
Jan 1 16:12:30 chii kernel: Process Xorg (pid: 2501, ti=f3d12000 task=f3d50030 task.ti=f3d12000)
Jan 1 16:12:30 chii kernel: Stack: f3d12000 c0404098 00000006 f3d12000 c0404050 00000006 c0384641 bff77f6c
Jan 1 16:12:30 chii kernel: 00000000 00000001 bff77fa8 00000036 0000007b 0000007b 00000033 00000036
Jan 1 16:12:30 chii kernel: 007ad402 00000073 00003287 bff77f44 0000007b 00000000 00000000
Jan 1 16:12:30 chii kernel: Call Trace:
Jan 1 16:12:30 chii kernel: [<c04051c1>] show_trace_log_lvl+0x1a/0x2f
Jan 1 16:12:30 chii kernel: [<c0405271>] show_stack_log_lvl+0x9b/0xa3
Jan 1 16:12:30 chii kernel: [<c040540d>] show_registers+0x194/0x26a
Jan 1 16:12:30 chii kernel: [<c0405610>] die+0x12d/0x242
Jan 1 16:12:30 chii kernel: [<c05fda37>] do_page_fault+0x3ee/0x4ba
Jan 1 16:12:30 chii kernel: [<c05fc234>] error_code+0x7c/0x84
Jan 1 16:12:30 chii kernel: [<c0404050>] syscall_call+0x7/0xb
Jan 1 16:12:30 chii kernel: =======================

[edit:] attached an nvidia-bug-report.log

Also, some things I've tried:
SMbus patch
xenrt patch
disable composite
kernel options: pci=nommconf idle=poll
older xorg package (in case it was an xorg bug)
updating bios
latest driver (9746)

[edit2:] seems like the attachment failed. The forum didnt like the size of the file... I removed some irrelevant stuff from the NVRM messages from years ago (yes, this install is very old, and it has always been stable before i got my new hardware), which seemed to take up more than half of the file...

[atonello]

First of all, thank you Mephisto1982 for your response.

I don't know if our systems suffer for the same problem;
there's a thread (http://www.nvnews.net/vbulletin/showthread.php?t=64682) on this forum about some incompatibilities between 6200 cards (that incidentally have turbocache functionalities just like mine...) and ATI RS480 chipsets (more or less the same family as the RS580 hosted in my motherboard) that make me think... The fact is that people on that thread seem to have solved their problems with the 9746 drivers that still lockup my computer.

By the way, I've found a thread on Ubuntu forums about similar lockups on Core2 Duo machines, please see if it's useful (unfortunately, it didn't make the difference for me ): http://www.ubuntuforums.org/showthre...ghlight=freeze

btw, I'd like to hear some thought from NVidia guys, just to know if there's something I'm missing (maybe some tweak in xorg.conf)...

Thanks

[Mephisto1982]

Thanks for the suggestion. I read the thread and, although it wasn't about the nvidia driver but core 2 duo (my lockups/crashes seems to happen only when running opengl apps, so i don't think it's a general system instability issue) i now have some other options to try. I didn't see it in the thread anywhere, but all the talk about BIOS and APIC made me go through my bios options, and I'm trying with the virtualisation disabled now, since the nvidia driver doesn't support xen-kernels either, so it may have a problem with that. I can also turn off NX and see if it improves anything. If that doesn't help, I'll try with the noapic and nolapic kernel options.

[fldc]

I have the same problem with my ATI xpress 200 based ASUS P5RD1-VM and geforce 7300 GS, the 9746 driver did reduce those crashes, now i have a freeze, every couple a day of regular use instead. If using the composite extension and compiz a bit more often.

[atonello]

@ fldc
due to my stability problems, I never dared install things such as compiz/beryl or activate composite...

@ Mephisto1982
hope you'll find the clue; about kernel options, they deluded me only making lockups more sporadic...

P.S. I'm on "nv" for 2 days almost (no reboot, always up) and no freezes...

[atonello]

small update...

After a couple of days using "nv" in xorg.conf, I've tried once again "nvidia"; after approx 5-10 min, respectively, I've experienced other 2 crashes.
This time I had set a notebook on lan (through my adsl router) as client via ssh; the news is that after the crash there was no more host (i.e. the locked computer): the connection was lost. Checking my router setup page, there was only the ip of the client pc among its "connected peripherals".

latest debug.log

[netllama]

Are you able to setup a serial console to capture any kernel messages from the time of the crash?

[atonello]

well, if you might point me some (even generic - just a glimpse then I'll google ) indications about what do and what kernel messages you need (perhaps, are you referring to kern.log? And if so, couldn't I paste its content in a zipped log?)... Please consider I've shutdown and now I'm running with "nv"; anyway I can always revert back to "nvidia" an be ready to catch what needed.

Thanks, Antonello

[netllama]

I'm referring to kernel messages that would only appear over a serial console (or possibly a netdump server, but that is less likely to work).

[atonello]

yes, in the meanwhile I was gathering some information about...

what I'd like to know is whether I'm necessarily supposed to use a serial cable (crosswired): in that case I'll need more time (once I had one - very long ago), or I can act via lan (I'm aware that it's limited by the fall of connection once the events get triggered).

thanks