[linux-dvb] Re: Problems with WinTV NOVA-CI

[Andrew de Quincey <adq_dvb@lidskialf.net>]

On Monday 13 Dec 2004 14:49, Simeon Walker wrote:
> On Mon, 2004-12-13 at 15:31 +0100, Johannes Stezenbach wrote:
> > Stefan Taferner wrote:
> > > On Monday 13 December 2004 14:54, Andrew de Quincey wrote:
> > > [...]
> > >
> > > > Werner said he'd compared the saa7146 PCI memory between working and
> > > > non-working and there wasn't really an obvious difference. On the
> > > > Nova-S
> >
> > You could check for yourself: run lspci -vvxxx after cold and
> > warm boot from windoze. As for interpretation of the values one
> > has to consult the pci spec. But it's also possible that irq
> > routing is screwed up which won't show up in lspci I think.
> >
> > > > This sounds _extremely_ like the issue many people who use VDR with
> > > > an FF and a budget card have - after a bit the budget card just stops
> > > > delivering data, but it can still tune OK. (the FF card works fine
> > > > though). A reboot (although not to windows) is required to fix it.
> > >
> > > I assume that one cannot reset the DVB card without rebooting the
> > > computer?
> >
> > saa7146_core.c:saa7146_init_one()
> > ...
> >         /* we don't do a master reset here anymore, it screws up
> >            some boards that don't have an i2c-eeprom for configuration
> >            values */
> > /*
> >         saa7146_write(dev, MC1, MASK_31);
> > */
>
> I have an old Nova-T (not CI) and on every kernel/driver install I
> uncomment this line. The card will then work if the modules are loaded,
> unloaded and loaded again after a cold boot. On subsequent warm boots
> this is not required.
>
> I have had many different kernels from 2.4.x to 2.6.10rc2, the card has
> been in two completely different machines (an MSI Hermes 650 and now a
> VIA M10K based box) but none of this changes the behaviour.

Interesting.. so that card DOES need reset - but its not being set up 
right/waiting long enough post-reset so you need to unload+reload the drivers 
to fix it. Does that sum it up?


On a related issue:
> >         /* we don't do a master reset here anymore, it screws up
> >            some boards that don't have an i2c-eeprom for configuration
> >            values */

Can someone explain what exactly goes wrong with these boards?