Michael Krufky wrote: > Oliver Endriss wrote: > > Michael Krufky wrote: > >> Trent Piepho wrote: > >>> On Mon, 6 Aug 2007, e9hack wrote: > >>>> Michael Krufky schrieb: > >>>>> Now I'm beginning to have doubts about Oliver's original patch: > >>>>> > >>>>> dvb_frontend: Range check of frequency and symbol rate > >>>>> http://linuxtv.org/hg/v4l-dvb/rev/8186a34dd0a6 > >>>>> > >>>>> Should we be checking fe->ops.tuner_ops.info.frequency_min|max , > >>>>> instead of > >>>>> fe->ops.info.frequency_min|max ??? > >>>>> > >>>> I didn't see the ranges from the tuner, because the dvb-c drivers don't > >>>> use the pll framework. They > >>>> have very simple tuning functions only. We should use both values. One > >>>> part may be more restrictive > >>>> than the other one. > >>> Most demodulators don't have frequency ranges. They just take whatever > >>> the > >>> tuner gives them at a fixed intermediate frequency. It's really the tuner > >>> that has the frequency range. > > > > Agreed. > > > >>> I think it would make more sense for the demodulator drivers to fill > >>> fe->ops.info.frequency_min|max using > >>> fe->ops.tuner_ops.info.frequency_min|max. > >>> A frontend driver that doesn't use a separate tuner driver (like DST) > >>> would > >>> set the fe->ops.info.frequency_min|max directly. > > > > Afaics the demod driver cannot fill fe->ops.info.frequency_min|max using > > fe->ops.tuner_ops.info.frequency_min|max, because the tuner driver will > > be attached _after_ the demod driver (see budget-av.c for example). > > > >> The way I see it, the demod driver that doesn't have a separate tuner > >> driver > >> should just go ahead and fill ops.tuner_ops.info.frequency_min|max , > >> because > >> otherwise those fields will be there anyway, just remaining empty. Those > >> structures are not pointers, and we should always be able to rely on their > >> presence. > >> > >> There is no need for BOTH ops.info.frequency_min|max AND > >> ops.tuner_ops.info.frequency_min|max > > > > The following drivers set ops.tuner_ops.info.frequency_min|max: > > dvb-pll, mt2060, qt1010, tda826x, tda827x, tua6100. > > > > All other drivers use ops.info.frequency_min|max. > > > > What about this: > > dvb_frontend.c shall use ops.tuner_ops.info.frequency_min|max if > > non-zero. Otherwise it would continue to use ops.info.frequency_min|max. > > That's fine with me... except I just don't see why we shouldn't just have the > demod drivers that have the integrated tuning functionality just fill > tuner_ops.info.frequency_max|min anyway. The point is, the structures will be > present regardless of whether any tuner_ops are actually filled. This is an > opportunity to remove the frequency_min|max from the demod ops.info structure, > as we all agree that it is inappropriate there anyhow. If we do this, then we > will have a standard across the board.
That's fine if we agree that we have to touch most frontend drivers. If we choose to do that, we should remove ops.info.frequency_min|max, i.e. remove 'struct dvb_frontend_info' from 'struct frontend_ops' and add something like 'struct dvb_demod_info' instead. We must not modify 'struct dvb_frontend_info' because it is an API data structure. Afaics 'frequency_stepsize' can be substituted by 'frequency_step', but what should we do with 'frequency_tolerance'? CU Oliver -- ---------------------------------------------------------------- VDR Remote Plugin 0.3.9: http://www.escape-edv.de/endriss/vdr/ ---------------------------------------------------------------- _______________________________________________ linux-dvb mailing list linux-dvb@linuxtv.org http://www.linuxtv.org/cgi-bin/mailman/listinfo/linux-dvb
