On 4/11/06, Michael Krufky <mkrufky@xxxxxxx> wrote: > Markus Rechberger wrote: > > >are you serious that this is a nice way to detect the dvb demodulator? > >you bind the mt352 module to devices that might only contain a zl10353 > >which isn't very nice. > >This will end up in a mess like the tuner-core module. > > > > > Markus- > > For once we agree with each other, however: > I think we'd agree way oftener if you would just read back what's requested... > #1) This is an experimental patch, and I purposefully did not include my > sign-off. > there are general issues with the zl10353 module I'm on it with pascoe we need to merge our changes (register writes) it doesn't work that general as it is. My version for sure can't be taken as it is since it would break other dvb-t drivers..my one is only supposed to work with the em2880-dvb module but we need to work out an acceptable way for that module to get all devices covered. > #2) You forgot to mention that I put needless braces around the check > for (card->fe = NULL) -- if the patch works, I will remove those before > applying. > don't really care about this one since we have other problems with it.. > #3) This is exactly the same way that Chris Pascoe did it in cx88-dvb > (see linux/drivers/media/video/cx88/cx88-dvb.c , lines 557 - 573) -- I > told him that I don't like it, however, neither your argument nor mine > holds any water -- Chris did the right thing. This is also how it is > done in b2c2-flexcop, and many of the other dvb drivers, and this is the > acceptable probing method for dvb frontends. > > The case is this: Boards out there based on the mt352 frontend have > been re-released with the zl10353 instead of the mt352, without changing > the PCI subsystem ID or eeprom contents. > in general what would we do if the mt352 or zl10353 gets replaced with a foo123 add another attach probe function? ... no we should define a more general way for that. I don't even like to call attach_mt352 within the driver... there's a new way of attaching external tuner modules in my v4l-dvb repository - that's one part I wanted Mauro to check - though I had no time to change what he pointed out yet but generally it will stay that way. tuner-core maintains a global list of external tuners, the xc3028-tuner module registers itself and extends that global list so that the tuner-core module can work with it. It could be made completly dynamically but Mauro stated out he still wants an entry in tuner-types.h and so I did for now. Think something like that would also be nice in such cases.. though of course I'm open for better ways to handle that. > I refer you to linux/drivers/media/dvb/b2c2/flexcop-fe-tuner.c , lines > 436 - 508 > very ugly yes... > NOW you know, and knowing is half the battle. > > -G.I. Joe > -- Markus Rechberger _______________________________________________ linux-dvb@xxxxxxxxxxx http://www.linuxtv.org/cgi-bin/mailman/listinfo/linux-dvb
