[ivtv-devel] 0.7.0 : "1000 ms time out waiting forfirmware" problem becomes worse

Hans Verkuil hverkuil at xs4all.nl
Wed Aug 23 07:43:47 CEST 2006


On Tuesday 22 August 2006 08:46, Tim Berger wrote:
> On 8/21/06, Ricardo Lugo <punka at cfl.rr.com> wrote:
> > Keith,
> > I have had some success minimizing these kinds of problems on my
> > SMP machine by tweaking my machine's IRQs. If any devices share
> > your ivtv card's IRQ, remove / shuffle the cards until you have a
> > good config where the card has its own IRQ.
>
> Ricardo, was this on 64bit as well?  Because it's looking to me like
> some 64bit bug was introduced after 0.6.1.  If people on 64bit  are
> only having trouble after this version but with 32bit are ok, then
> can this still be IRQ related?  On my machine, ivtv0 and ivtv1 have
> their own, non-shared interrupts.
>
> I never looked at the source before this evening, but it looks like
> ivtv_pause_encoder() was completely ripped out and replaced with some
> new api call ivtv_vapi().  Maybe that has somethig to do with it?  I
> really have no idea.  Something new has caused the firmware to go out
> to lunch though (or appear to be out to lunch), and only on dual core
> 64bit as far as I can tell.
>
> Maybe I'll just punt and reinstall with FC5 32bit.  I wonder how many
> people out there have had ivtv success with an athlon X2, and 64bit
> disto.  Please let me know off-list. Thx.

Odd, on my 64-bit production server I'm running ivtv-0.6.3 with kernel 
2.6.16.16 without problems. When I've finished working on the DMA error 
bug I'll see if I can reproduce this on my athlon X2.

Regards,

	Hans



More information about the ivtv-devel mailing list