On Sunday 20 August 2006 23:16, Simen Thoresen wrote:
> Hans Verkuil wrote:
> > The only thing I know of that causes artefacts etc. is de DMA
> > error. Once that occurs the only 'fix' at this moment is to reboot.
>
> I too noticed that the errors would continue past the 'damaged'
> segment, and also from one recording to the next. Can't we reset the
> chip (reloading the driver) or similar?
>
> I have not experimented with this myself yet.
Reloading won't work, only rebooting. I've tested with all
three 'recommended' firmwares, and all exhibit the same behavior.
Somewhere a wrong pointer is stored in the cx23415 memory. It isn't the
driver, that's completely reloaded without any effect.
>
> Thank you for running the 'special' frequency policy manager - it is
> about the same that I intended to run .-)
>
> Is there any benefit in me reproducing this to confirm that the
> changing CPU speed in mid-DMA can cause the errors?
No. Now that I finally can reproduce this I can proceed trying to fix
it.
Hans
>
> If not, I can return to running with PIO, and reenable the cpuspeed
> function.
>
> Yours,
> -S
>
> > Hans
> >
> > On Sunday 20 August 2006 22:39, Michael Zanetti wrote:
> >> Hello,
> >>
> >> I'm back home now. I just tried disabling CPU Frequency scaling
> >> again. The machine is running now for about 3 hours. I haven't
> >> seen any DMA error yet but I can see already artefacts and
> >> corrupted mpg video. Usually this happens just before the DMA
> >> errors come up. Do you have this behavior on your setups? Could
> >> someone with a working setup please attach or upload a kernel
> >> config file? I hope I can find out some more settings that cause
> >> this problem.
> >>
> >> Michael
> >>
> >> Am 20.08.2006 um 21:39 schrieb John Drescher:
> >>>> I've finally managed to reproduce the DMA error reliably and
> >>>> quickly using the following shell script on my dual AMD:
> >>>>
> >>>> #!/bin/sh
> >>>> cd /sys/devices/system/cpu
> >>>> echo userspace >cpu0/cpufreq/scaling_governor
> >>>> echo userspace >cpu1/cpufreq/scaling_governor
> >>>> while true; do \
> >>>> echo 2200000 >cpu0/cpufreq/scaling_setspeed; \
> >>>> echo 2200000 >cpu1/cpufreq/scaling_setspeed; \
> >>>> sleep 0.3; \
> >>>> echo 1000000 >cpu0/cpufreq/scaling_setspeed; \
> >>>> echo 1000000 >cpu1/cpufreq/scaling_setspeed; \
> >>>> sleep 0.3; \
> >>>> done
> >>>>
> >>>> For some reason powernowd triggers the bug only very rarely, but
> >>>> this shell script is very effective.
> >>>
> >>> Thanks for the script. I got "dmabug.sh: line 9:
> >>> cpu0/cpufreq/scaling_setspeed: Permission denied" errors when I
> >>> tried that as root so I guess I'll have to dig a little deaper
> >>> into my kernel config to see why this is happening.
> >>>
> >>> John
> >>>
> >>> _______________________________________________
> >>> ivtv-users mailing list
> >>> [email protected]
> >>> http://ivtvdriver.org/mailman/listinfo/ivtv-users
> >>
> >> _______________________________________________
> >> ivtv-users mailing list
> >> [email protected]
> >> http://ivtvdriver.org/mailman/listinfo/ivtv-users
> >
> > _______________________________________________
> > ivtv-users mailing list
> > [email protected]
> > http://ivtvdriver.org/mailman/listinfo/ivtv-users
_______________________________________________
ivtv-users mailing list
[email protected]
http://ivtvdriver.org/mailman/listinfo/ivtv-users
