On Friday 16 February 2007 20:42, Andy Speagle wrote:
> Hans,
>
> Thanks for the reply.... is the faulty DMA engine in the Conexant
> chips not something that can be coded around... at all? I'm not
> relishing rebuilding this system to get rid of my software RAID
> components.. ugh
Sorry, no. It seems to be a hardware issue. In fact, the 'offset'
messages you see are an attempt by the driver to work around it, but it
only works if the DMA errors only happen occasionally. When they happen
frequently the card will eventually hang.
Hans
>
> Thanks.
>
> Andy
>
> <snip>
>
> > The cx23415/6 Conexant chips don't have a very good DMA engine and
> > certain situations can cause it to do strange things and ultimate
> > hang. Using a CPU frequency changer application is one know
> > problem. RAID is another. Whether or not it will affect you seems
> > to depend on magic, although VIA chipsets seem to be more prone to
> > it than others.
> >
> > Interestingly, the PVR500 is immune to these problems, almost
> > certainly because it has a PCI bridge, which seems to isolate the
> > conexant chip from any unpleasantness. (Immune provided you use the
> > 0.10.0rc1 driver or up, older drivers can still hang the PVR500).
> >
> > I'm not sure whether interrupts are really a problem here. But I
> > would try to give the PVR350 an interrupt of its own (or at least,
> > not sharing any of the ide interrupts). To be honest, I think it is
> > the RAID array that is the culprit. Switching to using the
> > harddisks 'as is' would probably solve it.
_______________________________________________
ivtv-users mailing list
[email protected]
http://ivtvdriver.org/mailman/listinfo/ivtv-users
