> On Tue, Aug 28, 2007 at 06:24:55PM +0200, Hans Verkuil wrote:
>> The good news is that the message is harmless, the bad news is that it
>> shouldn't happen.
>
> On second thought, it might not be so harmless. See below(*).
>
>> If you do not load the nxt2004 driver but only ivtv, does the message
>> still appear?
>
> The message does not appear when I don't load the nxt200x (nor
> saa7134*) modules.
>
> (*)I first tried to do this by unloading all v4l drivers and then
> reloading just ivtv and its dependencies. When I tried to reload
> ivtv, I had a complete system lock up. That reminded me of the 3 hard
> lock ups I had this weekend. I had attributed those to very heavy
> stress testing I was doing to help track down some MythTV bugs.
Looks like another i2c probing f*ck-up to me. Try this:
rename saa7115.ko to saa7115.ko.bak so that this module cannot be loaded.
Then try to load nxt200x and ivtv the usual way. You will of course get
errors about the saa7115 not being present, but do you still get the tuner
errors as well?
The core problem is that the way i2c devices are discovered is
fundamentally wrong. Starting with 2.6.22 the i2c subsystem was cleaned up
so that it became possible to fix these problems, but there was
insufficient time to use it in v4l2. I'm working on it to get it fixed for
ivtv in kernel 2.6.24.
There are some module options for i2c modules to force them to ignore
probing of certain addresses and/or adapters, but it is very
user-unfriendly and in fact hard to figure out which i2c module is the
culprit.
Regards,
Hans
_______________________________________________
ivtv-devel mailing list
[email protected]
http://ivtvdriver.org/mailman/listinfo/ivtv-devel
