[linux-dvb] Re: Kernel Oops with saa7134 (v4l / analogue) + dvb-ttusb-budget

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

On Monday 04 Oct 2004 13:28, Richard Taylor wrote:
> > > ---- begin dmseg output ----
> > >
> > > saa7130/34: v4l2 driver version 0.2.11 loaded
> > > saa7134[0]: found at 0000:00:0e.0, rev: 1, irq: 17, latency: 32, mmio:
> > > 0xe4000000 saa7134[0]: subsystem: 1043:0210, board: KNC One TV-Station
> > > RDS / Typhoon TV Tuner RDS [card=7,insmod option] saa7134[0]: board
> > > init: gpio is 10700
> > > tuner: chip found at addr 0xc0 i2c-bus saa7134[0]
> > > tuner: type set to 10 (Alps TSBE1) by saa7134[0]
> > > tda9887: chip found @ 0x86
> >
> > It looks like the tda9887 device is erroneously clamining the same i2c
> > device as the ALPS TDMB7 device. I've seen this as well with the cx88
> > DVB-T cards. If you can avoid loading one of those modules, that would be
> > the easiest way to fix it.
> >
> > Otherwise.. umm, you'll either have to hack something in yourself to stop
> > the tda9887 from jumping on the bus, or wait for the new code refactoring
> > + kernel i2c fixes.
>
> I was very hopeful for a moment :)
>
> I removed the tda module from my kernel (deleted from /lib/.../media/video)
>
> ran depmod -a, and tried again.
>
> I still have the oops, but no reference to tda just before it...
>
> I'm beginning to wonder if dvb-core is seeing an saa7134 registering, and
> thinking it should add it as a dvb frontend?


AAaargh yeah, that'll be it! The tda9887 has the same i2c address as the alps 
tdmb7. 

You need to add a check at the start of attach_adapter() in alps_tdmb7.c to 
check the adapter type... something like:

if (!(adapter->class & I2C_CLASS_TV_DIGITAL)) return -ENODEV;

so it doesn't try to attach to non-DVB i2c adapters.

The refactoring and the i2c patches will sort all this out properly.