On Wed, 2006-16-08 at 13:21 -0700, Kristo Kriechbaum wrote:
> Sander Sweers wrote:
> > On Wed, 2006-08-16 at 12:13 +0200, Simen Thoresen wrote:
> > [..]
> >> The core functionality of this powermanagement feature is compiled into 
> >> the kernel (or as a module). On my kernel (2.6.16.6), the feature is 
> >> built as
> >> CONFIG_X86_POWERNOW_K8=m
> >> CONFIG_X86_POWERNOW_K8_ACPI=y
> >>
> >> ...for my Athlon64 cpu, in i386 mode - I assume it will be similar on 
> >> your system.
> >>
> >> On my system (no errors for a week or so), I am not only not running the 
> >> userspace manager, but I am also /not/ loading the module.
> > 
> > You can control the CPUFREQ governors with sysfs. Try modprobing the
> > cpufreq_ondemand and echo ondemand to the scaling governor. "echo
> > ondemand > /sys/devices/system/cpu/cpu0/cpufreq/scaling_governor. Echo
> > performance to make the cpu stay at maximum performance.
> > 
> > Powernow will not only lower the cpu's frequentie but also the voltage.
> > And as the memory controller in on the cpu with AMD64 this might cause
> > the dma errors.
> > 
> > Not sure about the above but if all the people with dma errors have an
> > AMD64 this could be the issue.
> 
> Interesting.  I have an AMD64, and I did (do?) have the DMA problems.
> I've been running with the PIO mode patch for some time now, but in the
> next few days I hope to do some testing with your above suggestion.  I
> did have cpuspeed turned on.  My machine is running an FC4 kernel.
> 
> Kristo

I have been watching this thread with great interest.  Here are the
things that I noticed.

- My problems only started with the 2.6.17 kernel and ivtv 0.7.0
- I read somewhere that this is an IRQ migration problem on SMP kernels
so I turned off the irqbalance service.  Stability increased since
failures when from 1 - 5 minutes to 5 - 10 days.
- When I switch to a UP kernel, stability increases again.
- Someone mentioned a memory timing issue.  I have two sets of dual
channel memory.  That might also be the cause.

What this all tells me is there is an IRQ timing issue in ivtv.  I have
no way to prove this since I do not dive into code and am an average end
user using RPMS from Axel.

Does this help?  Do others have similar experiences?

My machine is an AMD64 x2 in x86_64 mode.

Preet


_______________________________________________
ivtv-users mailing list
[email protected]
http://ivtvdriver.org/mailman/listinfo/ivtv-users

Reply via email to