Trying to make headway into finding the Xid crashes source...

[deadhamster]

nukem: Just use the NvAGP "0" option in your xorg.conf and don't compile agpgart support in your kernel.

-DH

[LubosD]

@cpbotha: BTW: I had to comment line 52...

/tmp/vtkFlRenderWindowInteractor-nvcrash/examples/Cone3/Cone3.cxx: In function `void quit_cb(Fl_Widget*, void*)':
/tmp/vtkFlRenderWindowInteractor-nvcrash/examples/Cone3/Cone3.cxx:52: error: 'class vtkRenderWindow' has no member named 'Finalize'

Beautiful:

May 11 15:03:27 localhost kernel: NVRM: Xid: 23,  L1 -> L0
May 11 15:03:27 localhost kernel: NVRM: Xid: 6, PE0002 0184 beef0201 00310568 00000000 02440208
May 11 15:03:28 localhost kernel: NVRM: Xid: 23,  L1 -> L0
May 11 15:03:28 localhost kernel: NVRM: Xid: 6, PE0002 0184 beef0201 00000030 00000000 02440208
May 11 15:03:28 localhost kernel: NVRM: Xid: 23,  L0 -> L0
May 11 15:03:28 localhost kernel: NVRM: Xid: 6, PE0002 0184 beef0201 00000000 00000000 00000000
May 11 15:03:28 localhost kernel: NVRM: Xid: 6, PE0002 0184 beef0201 00000000 00000000 00000000
May 11 15:03:28 localhost kernel: NVRM: Xid: 6, PE0002 0104 00000000 00000000 00000000 00000000
May 11 15:03:28 localhost kernel: NVRM: Xid: 6, PE0002 0104 00000000 00000000 00000000 00000000
May 11 15:03:28 localhost kernel: NVRM: Xid: 23,  L1 -> L0
May 11 15:03:28 localhost kernel: NVRM: Xid: 6, PE0002 0184 beef0201 00310568 fff8f8f8 02440208
May 11 15:03:28 localhost kernel: NVRM: Xid: 23,  L1 -> L0
May 11 15:03:28 localhost kernel: NVRM: Xid: 6, PE0002 0184 beef0201 00310568 ffe7e8e7 02440208
May 11 15:03:29 localhost kernel: NVRM: Xid: 23,  L1 -> L0
May 11 15:03:29 localhost kernel: NVRM: Xid: 6, PE0002 0184 beef0201 00310568 ffa93f3f 02440208
May 11 15:03:29 localhost kernel: NVRM: Xid: 23,  L1 -> L0

[LubosD]

What about creating some online petition for adequate drivers for Linux?

[superm1]

Originally Posted by Soul-Crusher

Interesting... I started using APIC as someone else suggested earlier in this thread. That didn't help, but I got a message in my kernel log:
** PCI interrupts are no longer routed automatically. If this
** causes a device to stop working, it is probably because the
** driver failed to call pci_enable_device(). As a temporary
** workaround, the "pci=routeirq" argument restores the old
** behavior. If this argument makes the device work again,
** please email the output of "lspci" to bjorn.helgaas@hp.com
** so I can fix the driver

So I did just that and added pci=routeirq to my kernel line in grub on startup. Now glxgears seems to run just fine at normal speed without wrenching anything.

Edit: It seems that wasn't a permanent fix, but I would suggest trying it out anyway.

Well - gave this a shot with no love. X is still taking up 99 percent CPU before it even gives a chance to any other process, and erroring the log with Xid 13 errors. This seems to be occuring for me for any nvidia driver revision however, so maybe I'm having a worse problem.h

[Saulzar]

Hehe - "beef0201" I wonder if that is intentional...

[xenofon]

Hello,
I have had both the "live pointer" and the complete freeze occur to me. Glxgears freezes the system too. Also, I get artifacts in racer just before things freeze. UT2004 gives no artifacts, but system freezes eventually.

System specs bellow:

MSI NX6600GT AGP
2x Athlon MP 2200+
MSI k7D Master
1,5 GB DDR 2100 ECC reg
adaptec 39160 running seagate 36,5 GB 15krpm disk
Creative labs SB Audigy
Realtek based Ethernet
USB2 card
Hauppauge PVR3

Currently running: Mandriva 2005 LE, custom 2.6.11.8 kernel with ivtv and nvidia 7174 modules. Alsa 1.0.8 drivers for the Audigy.

-----------------------------------------------------------------------------
/proc/interupts:

CPU0 CPU1
0: 3390078 809 IO-APIC-edge timer
1: 5014 7 IO-APIC-edge i8042
7: 0 0 IO-APIC-edge parport0
8: 0 1 IO-APIC-edge rtc
9: 0 0 IO-APIC-level acpi
12: 4 89 IO-APIC-edge i8042
14: 4066 2 IO-APIC-edge ide0
15: 59058 1 IO-APIC-edge ide1
16: 70787 0 IO-APIC-level ohci_hcd, eth0
17: 254069 0 IO-APIC-level aic7xxx, ohci_hcd, nvidia, EMU10K1
18: 387396 9266 IO-APIC-level aic7xxx, ehci_hcd, ivtv0, ohci1394
NMI: 0 0
LOC: 3390781 3390966
ERR: 0
MIS: 0
--------------------------------------------------------------------------
/proc/driver/nvidia/agp/status

Status: Enabled
Driver: NVIDIA
AGP Rate: 4x
Fast Writes: Disabled
SBA: Disabled
---------------------------------------------------------------------------

Passing the noapic and pci=routeirq parameters do not seem to solve the problem.

A typical Xid error appearing in the logs is as follows:

May 12 00:23:46 quicksilver kernel: NVRM: Xid: 6, PE0000 0f60 00ffffff 0000fc142000ffff 00ffffff


It should be noted that the same system equipped with an Albatron 5900XTv runs rock solid.


I hope this information helps in gathering some statistics on the problem.

[cpbotha]

@LubosD: Hi there!

Originally Posted by LubosD

BTW: I had to comment line 52...

Aaaah, that's right... it's a relatively recent addition to VTK. You could have replaced it with WindowRemap(), but it's not so important.

Originally Posted by LubosD

Beautiful:

May 11 15:03:27 localhost kernel: NVRM: Xid: 23,  L1 -> L0
May 11 15:03:27 localhost kernel: NVRM: Xid: 6, PE0002 0184 beef0201 00310568 00000000 02440208

Does this mean you could reproduce it? I.e. did you get the temporary X-freeze also with WAIT() calls in your X log?

Thank you for giving this a shot!

[LubosD]

I haven't looked in to the X.org log file so I don't know about WAIT()s. But it crashed...
The crash was rather permanent than temporary

[xenofon]

Originally Posted by xenofon

Hello,
I have had both the "live pointer" and the complete freeze occur to me. Glxgears freezes the system too. Also, I get artifacts in racer just before things freeze. UT2004 gives no artifacts, but system freezes eventually.

System specs bellow:

MSI NX6600GT AGP
2x Athlon MP 2200+
MSI k7D Master
1,5 GB DDR 2100 ECC reg
adaptec 39160 running seagate 36,5 GB 15krpm disk
Creative labs SB Audigy
Realtek based Ethernet
USB2 card
Hauppauge PVR3

Currently running: Mandriva 2005 LE, custom 2.6.11.8 kernel with ivtv and nvidia 7174 modules. Alsa 1.0.8 drivers for the Audigy.

-----------------------------------------------------------------------------
/proc/interupts:

CPU0 CPU1
0: 3390078 809 IO-APIC-edge timer
1: 5014 7 IO-APIC-edge i8042
7: 0 0 IO-APIC-edge parport0
8: 0 1 IO-APIC-edge rtc
9: 0 0 IO-APIC-level acpi
12: 4 89 IO-APIC-edge i8042
14: 4066 2 IO-APIC-edge ide0
15: 59058 1 IO-APIC-edge ide1
16: 70787 0 IO-APIC-level ohci_hcd, eth0
17: 254069 0 IO-APIC-level aic7xxx, ohci_hcd, nvidia, EMU10K1
18: 387396 9266 IO-APIC-level aic7xxx, ehci_hcd, ivtv0, ohci1394
NMI: 0 0
LOC: 3390781 3390966
ERR: 0
MIS: 0
--------------------------------------------------------------------------
/proc/driver/nvidia/agp/status

Status: Enabled
Driver: NVIDIA
AGP Rate: 4x
Fast Writes: Disabled
SBA: Disabled
---------------------------------------------------------------------------

Passing the noapic and pci=routeirq parameters do not seem to solve the problem.

A typical Xid error appearing in the logs is as follows:

May 12 00:23:46 quicksilver kernel: NVRM: Xid: 6, PE0000 0f60 00ffffff 0000fc142000ffff 00ffffff


It should be noted that the same system equipped with an Albatron 5900XTv runs rock solid.


I hope this information helps in gathering some statistics on the problem.

After reading a couple more threads about the 6600GT and in an effort to silence my system a little bit, while ensuring good cooling for the graphics card, I installed the Zalman VF700-Cu cooler on my card.

Also, just to make sure, I connected the card to the PSU on its own cable (no other devices are connected on this cable) and I played around with the PCI cards until the 6600GT was not sharing an interupt with the AUDIGY.

---------------------------------------- /proc/interrupts-----------------------------------------

CPU0 CPU1
0: 3251164 567 IO-APIC-edge timer
1: 7521 2 IO-APIC-edge i8042
7: 0 0 IO-APIC-edge parport0
8: 0 1 IO-APIC-edge rtc
9: 0 0 IO-APIC-level acpi
12: 91 3 IO-APIC-edge i8042
14: 5575 2 IO-APIC-edge ide0
15: 56610 1 IO-APIC-edge ide1
16: 263130 0 IO-APIC-level ivtv0, eth0
17: 276721 0 IO-APIC-level aic7xxx, ohci_hcd, nvidia
18: 19993 0 IO-APIC-level aic7xxx, ohci_hcd, EMU10K1
19: 36 1 IO-APIC-level ehci_hcd, ohci1394
NMI: 0 0
LOC: 3251551 3252066
ERR: 0
MIS: 0
--------------------------------------------------------------------------------------------------------




Now, glxgears has been running for 30min + without a crash . Also, no more artifacts in racer. So far no crashes at all, but I do get some new Xid errors in my logs:
May 12 05:01:27 quicksilver kernel: NVRM: Xid: 16, Head 00000000 Count 000734e9
May 12 15:07:22 quicksilver kernel: NVRM: Xid: 23, L1 -> L0
May 12 15:07:22 quicksilver kernel: NVRM: Xid: 6, PE0000 07e8 00ffffff 0000fb1003f8ffff 00ffffff
May 12 15:07:22 quicksilver kernel: NVRM: Xid: 23, L0 -> L0
May 12 15:07:22 quicksilver kernel: NVRM: Xid: 1, Channel 00000002 Method 00000000 Data c044dbc8
May 12 15:07:22 quicksilver kernel: NVRM: Xid: 23, L1 -> L0
May 12 15:07:22 quicksilver kernel: NVRM: Xid: 6, PE0002 1f08 00000000 0000000000002ab3 00000000
May 12 15:07:22 quicksilver kernel: NVRM: Xid: 23, L1 -> L0
May 12 15:07:22 quicksilver kernel: NVRM: Xid: 6, PE0002 1f08 00000000 00210004


This might suggest that my problem was part-cooling related, although MSI had a pretty hefty cooler on the card to start with and my case is full of fans and blowholes.

But why the new error messages? I will keep you posted if any new crashes occur.[

[LubosD]

I have 6800 Ultra, so the card must share the power supply cables with other devices. My PSU is 500W so I don't think it suffers. It's fan is huge and noisy and temperatures are quite good (cca 69˚C at max - under load).

Nvidia shares it's IRQ with three other devices including Audigy (EMU10K1)...but I can't find any way how to change IRQs. I can only change IRQ assignment in BIOS: Automatic/Reserved. When is IRQ reserved it is not used by any devices so it is useless.
Does anyone have an idea how to force some IRQ to some driver (when that IRQ is marked as "Reserved" in BIOS)?

[xenofon]

I have no way to change the irq's in my bios, but I do know which pci slots on my mobo share irqs -- this information is usually in the manual. Then, by rearranging the pci cards I can avoid the graphics card sharing an interupt with the audigy and this is what I was talking about in my previous post.
Nonetheless, this still does not resolve the Xid error issue discussed in this thread. However, I have had no crashes since the rearrangements mentioned in the other post.

[deadhamster]

I don't know if anyone has tried this, I know I haven't. But has anyone tried not installing the glx portion of the driver package at all? It's easy enough on gentoo. That would at least help narrow things down some. Need I try this tonight?

-DH