Freeze or Oops on recent kernels

Freeze or Oops on recent kernels

[yvahk-xreary]

I am getting either a a kernel Oops or freeze (without any console output)
on recent kernels.  I have tested on 2.6.32.26 PAE, 3.1.9 PAE, and 3.4.9 PAE
all with similar results.

The hardware involved comprises mainly:

	12 GB ram
	Intel DX58S02 motherboard
	Intel Xeon E5220
	2 x onboard ethernet
	Intel PRO/1000 PT Dual Port ethernet
	Hauppauge HVR1700 DVB-T PCIe card
	Technisat SkyStar2 DVB-T PCI card

The oops or freeze occurs when both DVB cards are recording
simultaneously. With either card installed on their own there
is never any problem.

I should also add that the exact same kernel and cards on
a Gigabyte GA-P31-S3G motherboard + Intel Pentium 4 the problem
NEVER occurs. So there may be a DX58S02/timing/interrupt issue.

When there is an oops it is like this (hand-transcribed
from 2.6.32.26 PAE kernel):

c0789444	panic + 3e/e9
		oops_end + 97/a6
		no_context + 13b/145
		__bad_area_nosemaphore + ec/f4
		? do_page_fault + 0/29f
		bad_area_nosemaphore + 12/15
		do_page_fault + 139/29f
		? do_page_fault + 139/29f
c078b54b	error_code + 73/78
f82084fb	? cx23885_video_irq + d8/1dc
f820b04d	cx23885_irq + 3df/3fe
c04834ac	handle_irq_event + 57/fd
		handle_fasteoi_irq + 6f/a2
		handle_irq + 40/4d
		do_IRQ + 46/9a
		common_interrupt + 30/38
c045b7a9	? prepare_to_wait + 14c
f7fef5a5	? videobug_waitm + 90/133
c045b601	? autoremove_waker_function + 0/34
f805e5bc	videobuf_dvb_thread + 73/135
f805e4f9	? videobuf_dvb_thread + 0/135
c045b3c9	kthread + 64/69
		? kthread + 0/69
		kernel_thread_helper + 7/10

Although I am a very experienced programmer I have next to zero
kernel expertise except for minor patching of a few drivers.

My guess from the stack trace is that there might be an issue
with page fault recursion (if that is at all possible).

Anyhow I don't want to waste too much of my time or anybody elses
on this - although with the problem occuring with 3.4.9 kernel which
has significant interrupt handling changes it probably is something
that somebody might want to know about. If anybody can spot a clue
as to where I should be looking and how I should go about isolating
the problem (if only kernel core dumped!) please let me know and
I will possibly try and assist. I need some guidance.

Regards
John W.

Re: Freeze or Oops on recent kernels

[Dan Carpenter]

I'm not going to file a bug for this on bugzilla because it's
ancient and not a new bug.  But I can forward it to linux-media
and the get_maintainer.pl people for cx23885_video.

HINT: We only care about the very most recent kernel.  If you can
take a photo of the stack trace, then file a bug report and attach
the .jpg.

regards,
dan carpenter

On Fri, Sep 07, 2012 at 09:24:13PM +1000, yvahk-xreary@zacglen.net wrote:
> 
> I am getting either a a kernel Oops or freeze (without any console output)
> on recent kernels.  I have tested on 2.6.32.26 PAE, 3.1.9 PAE, and 3.4.9 PAE
> all with similar results.
> 
> The hardware involved comprises mainly:
> 
> 	12 GB ram
> 	Intel DX58S02 motherboard
> 	Intel Xeon E5220
> 	2 x onboard ethernet
> 	Intel PRO/1000 PT Dual Port ethernet
> 	Hauppauge HVR1700 DVB-T PCIe card
> 	Technisat SkyStar2 DVB-T PCI card
> 
> The oops or freeze occurs when both DVB cards are recording
> simultaneously. With either card installed on their own there
> is never any problem.
> 
> I should also add that the exact same kernel and cards on
> a Gigabyte GA-P31-S3G motherboard + Intel Pentium 4 the problem
> NEVER occurs. So there may be a DX58S02/timing/interrupt issue.
> 
> When there is an oops it is like this (hand-transcribed
> from 2.6.32.26 PAE kernel):
> 
> c0789444	panic + 3e/e9
> 		oops_end + 97/a6
> 		no_context + 13b/145
> 		__bad_area_nosemaphore + ec/f4
> 		? do_page_fault + 0/29f
> 		bad_area_nosemaphore + 12/15
> 		do_page_fault + 139/29f
> 		? do_page_fault + 139/29f
> c078b54b	error_code + 73/78
> f82084fb	? cx23885_video_irq + d8/1dc
> f820b04d	cx23885_irq + 3df/3fe
> c04834ac	handle_irq_event + 57/fd
> 		handle_fasteoi_irq + 6f/a2
> 		handle_irq + 40/4d
> 		do_IRQ + 46/9a
> 		common_interrupt + 30/38
> c045b7a9	? prepare_to_wait + 14c
> f7fef5a5	? videobug_waitm + 90/133
> c045b601	? autoremove_waker_function + 0/34
> f805e5bc	videobuf_dvb_thread + 73/135
> f805e4f9	? videobuf_dvb_thread + 0/135
> c045b3c9	kthread + 64/69
> 		? kthread + 0/69
> 		kernel_thread_helper + 7/10
> 
> Although I am a very experienced programmer I have next to zero
> kernel expertise except for minor patching of a few drivers.
> 
> My guess from the stack trace is that there might be an issue
> with page fault recursion (if that is at all possible).
> 
> Anyhow I don't want to waste too much of my time or anybody elses
> on this - although with the problem occuring with 3.4.9 kernel which
> has significant interrupt handling changes it probably is something
> that somebody might want to know about. If anybody can spot a clue
> as to where I should be looking and how I should go about isolating
> the problem (if only kernel core dumped!) please let me know and
> I will possibly try and assist. I need some guidance.
> 
> Regards
> John W.
> 
> --
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/

Re: Freeze or Oops on recent kernels

[yvahk-xreary]

>
>HINT: We only care about the very most recent kernel.  If you can
>take a photo of the stack trace, then file a bug report and attach
>the .jpg.
>

After a bit of experimentation my guess is that is is all
about bad Intel DX58SO2 motherboard.

And my guess is that it has something to do with memory
mapping during an interrupt. Don't know much about memory
mapping but I suspect that something quite fundamental
goes wrong with this motherboard only.

Exact same type of oops also happens with a PCIe RS232 serial card.

Trouble is that one tends to select hardware for stability so that
means slightly older hardware and hopefully fewer bugs.
Unfortunately latest hardware may not be supported by older
more stable kernels, and it also wont be well supported by newer
kernels.

Worse still, everything lasts less than a year with manufacturers getting
away with vandalistic change of sockets or connectors or chipsets just to
achieve more sales, and new bugs. This constant cycle of pointless
change doesn't seem to be heading anywhere in particular.

John W

Re: Freeze or Oops on recent kernels

[Dan Carpenter]

On Mon, Sep 24, 2012 at 04:52:32PM +1000, yvahk-xreary@zacglen.net wrote:
> >
> >HINT: We only care about the very most recent kernel.  If you can
> >take a photo of the stack trace, then file a bug report and attach
> >the .jpg.
> >
> 
> After a bit of experimentation my guess is that is is all
> about bad Intel DX58SO2 motherboard.
> 
> And my guess is that it has something to do with memory
> mapping during an interrupt. Don't know much about memory
> mapping but I suspect that something quite fundamental
> goes wrong with this motherboard only.
> 
> Exact same type of oops also happens with a PCIe RS232 serial card.
> 

There is not enough information here for us to help you.  Sorry.

regards,
dan carpenter