geforce 6600 random system lock

[auouymous]

Have you tried appending "irqpoll" to the kernel?

[Hannibal]

auouymous, it doesn't help. Switching to stronger power suply was also unsucessful. I have only one new Xid, but it occurs near X start not during hang.

Oct 11 18:34:13 hannibal kernel: NVRM: Xid (0001:00): 6, PE0000 0300 00000000 00000030 ffffffff 00010001

[a7v]

If you've tried everything else you could try lowering/raising AGP drive strength slightly in your BIOS settings. It has helped me with both a Fx 5200 and a Gf 6600 on two different motherboards.

[Hannibal]

I haven't such BIOS option

[Hannibal]

I don't is this important, but Xorg CPU usage on Ti 4200 is 2,3 times lower than under 6600...

[Hannibal]

I try to invastigate what is a problem, but I was unsucessful. But maybe my tries help someone...

I've add some printk to handle_IRQ_event function in kernel/irq/handle.c file.
It prits data something like that:

Oct 16 19:06:30 hannibal kernel: DATA: dff23448 2 dff234ec 1 2 2
Oct 16 19:06:30 hannibal kernel: DATA: 31687465 0 0 0 0 c0441048
Oct 16 19:06:30 hannibal kernel: DATA: e104c000 e19bb640 1 1 1 f210de
Oct 16 19:06:30 hannibal kernel:     SUBDATA1 d9b48000
Oct 16 19:06:30 hannibal kernel:     SUBDATA2 e104c000
Oct 16 19:06:30 hannibal kernel: IRQ: 16 - Device: e19bb640 Handler:  nv_kern_isr+0x0/0x72 [nvidia]
Oct 16 19:06:30 hannibal kernel: DATA: dff23448 2 dff234ec 1 2 2
Oct 16 19:06:30 hannibal kernel: DATA: 31687465 0 0 0 0 c0441048
Oct 16 19:06:30 hannibal kernel: DATA: e104c000 e19bb640 1 1 1 f210de
Oct 16 19:06:30 hannibal kernel:     SUBDATA1 d9b48000
Oct 16 19:06:30 hannibal kernel:     SUBDATA2 e104c000
Oct 16 19:06:30 hannibal kernel: IRQ: 16 - Device: e19bb640 Handler:  nv_kern_isr+0x0/0x72 [nvidia]
Oct 16 19:06:30 hannibal kernel: DATA: dff23448 2 dff234ec 1 2 2
Oct 16 19:06:30 hannibal kernel: DATA: 31687465 0 0 0 0 c0441048
Oct 16 19:06:30 hannibal kernel: DATA: e104c000 e19bb640 1 1 1 f210de
Oct 16 19:06:30 hannibal kernel:     SUBDATA1 d9b48000
Oct 16 19:06:30 hannibal kernel:     SUBDATA2 e104c000
Oct 16 19:06:30 hannibal kernel: IRQ: 16 - Device: e19bb640 Handler:  nv_kern_isr+0x0/0x72 [nvidia]

DATA means action->handle and first 5 dwords from action->dev_id
SUBDATAX id data from NVIDIA board (2 first dwords used as pointers )
Last line tells which handler takes interrupt.

From time to time appear this log:

Oct 16 19:06:30 hannibal kernel: DATA:  2 dff234ec 1 2 2
Oct 16 19:06:30 hannibal kernel: DATA: 31687465 0 0 0 0 c0441048
Oct 16 19:06:30 hannibal kernel: DATA: e104c000 e19bb640 1 1 1 f210de
Oct 16 19:06:30 hannibal kernel:     SUBDATA1 d9b48000
Oct 16 19:06:30 hannibal kernel:     SUBDATA2 e104c000
Oct 16 19:06:30 hannibal kernel: IRQ: 16 - Device: 00000000 Handler: NULL

I think this is not detected properly nvidia interrupt. Almost for sure this aren't USB (0xdff23448) and Realtek (0x31687465) ints. Because the not run at that moment (and I have lockups even when I disable usb in BIOS and remove realtek).

For sure this NULL apperar when I swich from X to console. Second time this appear is just before lockup. Then apperar losts of this NULLS and also somethin like that:

Oct 16 19:06:34 hannibal kernel: DATA: e104c000 e31687465 0 0 0 0 c0441048
Oct 16 19:06:34 hannibal kernel: DATA: e104c000 31687465 0 0 0 0 c0441048
Oct 16 19:06:34 hannibal kernel: DATA: e104c000 e131687465 0 0 0 0 c0441048
Oct 16 19:06:34 hannibal kernel: DATA: e104c000 e19b31687465 0 0 0 0 c0441048
Oct 16 19:06:34 hannibal kernel: DATA: e104c000 e131687465 0 0 0 0 c0441048
Oct 16 19:06:34 hannibal kernel: DATA: e104c000 e1931687465 0 0 0 0 c0441048
Oct 16 19:06:34 hannibal last message repeated 2 times
Oct 16 19:06:34 hannibal kernel: DA31687465 0 0 0 0 c0441048
Oct 16 19:06:34 hannibal kernel: DATA: e104c000 31687465 0 0 0 0 c0441048
Oct 16 19:06:34 hannibal kernel: DATA: e104c000 e31687465 0 0 0 0 c0441048
Oct 16 19:06:34 hannibal kernel: DATA: e104c000 e1931687465 0 0 0 0 c0441048
Oct 16 19:06:34 hannibal kernel: DATA: e104c000 e131687465 0 0 0 0 c0441048
Oct 16 19:06:34 hannibal kernel: DATA: e1031687465 0 0 0 0 31687465 0 0 0 0 c0441048
Oct 16 19:06:34 hannibal kernel: DATA: e104c000 31687465 0 0 0 0 c0441048
Oct 16 19:06:34 hannibal kernel: DATA: e104c000 e31687465 0 0 0 0 c0441048
Oct 16 19:06:34 hannibal kernel: DATA: e104c000 e19b31687465 0 0 0 0 c0441048
Oct 16 19:06:34 hannibal kernel: DATA: e104c000 e131687465 0 0 0 0 c0441048
Oct 16 19:06:34 hannibal kernel: DATA: e104c000 e19b31687465 0 0 0 0 c0441048
Oct 16 19:06:34 hannibal kernel: DATA: e104c000 e19bb31687465 0 0 0 0 c0441048
Oct 16 19:06:34 hannibal kernel: DATA: e104c031687465 0 0 0 0 c0441048
Oct 16 19:06:34 hannibal kernel: DATA: e104c00031687465 0 0 0 0 c0441048
Oct 16 19:06:34 hannibal kernel: DATA: e104c000 e31687465 0 0 0 0 c0441048
Oct 16 19:06:34 hannibal last message repeated 2 times
Oct 16 19:06:34 hannibal kernel: DATA: e104c000 31687465 0 0 0 0 c0441048

It looks like so many interrupts are commning, that printk doesn't have enough time to print whole message. This ints are probably those NULLS because 9900 unahandled ints is requied to disable IRQ, but not all are listed explictly in log.

Code that detect nv interrupt will probably help, but unfortunatlyin this is in binary part...

I hope it was understandable and helpful ;-)

[gentel]

Hi!

I think I have the same problem like you!
My problem occured after a world update on my gentoo box. (testing version with kde 3.5.5)
I also have a geforce 6600 gt and I have testet the newest beta and stable driver.
In the console my computer doesn't freeze.
When I boot W*****S my machine works very well.

I think I found an another thread about this problem:
http://forums.gentoo.org/viewtopic-t-507935.html

I don't know what I should test now!

By the way: I have succesfully used the beta driver with beryl, so then I have disabled it, but the error still occur.

Best wishes

[Hannibal]

9629 and 2.6.19-rc4 freezes still occurs on 6600...
It seems that konqeror (with this forum open! ) cause that freeze occurs very early.

[netllama]

Hannibal,
Is this problem only present in 2.6.19-rc kernels, or does it also occur in stable, released kernels?

Does this only reproduce with Composite turned on?
Have you verified that you're using the latest BIOS for the motherboard?

Thanks,
Lonni

[Hannibal]

I've test it on 8776 and 2.6.18 with RenderAccel On/Off , AllowGLXWithComposite On/Off. But I forget about 'Option "Composite" "Enable"' :P . I will test this also, but probably this nothing change. And yes, I have most recent BIOS for my mainboard.

As I wrote before, there is massive number of unhandled IRQs for sure belonging to graphics card. But I no idea this is cause or effect of error.

[netllama]

If there are 'unhandled' IRQs that is likely a BIOS problem.

[Hannibal]

Originally Posted by netllama

If there are 'unhandled' IRQs that is likely a BIOS problem.

This isn't good information, and as i suspect turning off composite didn't help...