[linux-dvb] Why do we prevent tda9887 autodetection on dvb-bt8xx boards?

[Mike Krufky]

Manu, Gerd, and anyone else that can offer some insight-

In troubleshooting the FusionHDTV5 Lite bug last week, Ricardo noticed 
the following in bttv-cards.c:

         if (0 == tda9887 && 0 == bttv_tvcards[btv->c.type].has_dvb &&
                   bttv_I2CRead(btv, I2C_TDA9887, "TDA9887") >=0)
                       tda9887 = 1;
               if (tda9887)
                         request_module("tda9887");

This I found to be very strange...  When I first programmed this board, 
tda9887 was automatically being detected.  Somehow, the code above wasnt 
working correctly, but it is now.

This is a mystery to me, but regardless.....

WHY do we prevent tda9887 autodetection on bttv boards with DVB 
capabilities?

I did some research... Gerd committed this change on /Wed Oct 6 13:45:14 
2004 UTC/ ... Unfortunately, his commit message was totally unrelated to 
the changes made:

- cx88: AverTV Studio 303 (M126) fixes.
http://linuxtv.org/cgi-bin/viewcvs.cgi/v4l-kernel/linux/drivers/media/video/bttv-cards.c?root=v4l&r1=1.27&r2=1.28


FusionHDTV5 Lite is a hybrid board, and it depends on tda9887 in order 
to deliver NTSC video.

Ricardo applied the following, and it fixed the FusionHDTV5 Lite bug:

http://linuxtv.org/cgi-bin/viewcvs.cgi/v4l-kernel/linux/drivers/media/video/bttv-cards.c?root=v4l&r1=1.97&r2=1.98

(Thanks again, RC)

However, I'd like to get to the root of this... Can we remove "0 == 
bttv_tvcards[btv->c.type].has_dvb &&" from the check?

Does tda9887 autodetection cause problems on other DVB boards???

I can only assume that Gerd did this back then, because there weren't 
many (if any) hybrid NTSC/ATSC boards around...  Maybe Gerd thought that 
it was pointless to attempt to probe tda9887, assuming that it would 
never be necessary?

Things are okay as-is, but I wonder if this should be changed.

Any ideas/feedback?

Regards,

Michael Krufky