[linux-dvb] Re: Frontend experimental patch v1

[a.share@t-online.de (Andreas Share)]

> Hi, this patch does the following:
>
> FE_RESET. This is removed.
>
> FE_CAN_CLEAN_SETUP. This is removed.
>
> FE_POLL. This is added as a more meaningful replacement for FE_RESET.
> Currently it is only used by the ves1x93 frontend. FE_POLL will be called
> repeatedly after the frontend parameters have been set until it returns
> nonzero.
>
> FE_GET_TUNE_SETTINGS. This passes a structure as follows:
> struct dvb_frontend_tune_settings {
>         int min_delay_ms;
>         int step_size;
>         int max_drift;
>         struct dvb_frontend_parameters parameters;
> };
> The parameters field is initialised with the frontend parameters about to
be
> set. The frontend should fill in min_delay_ms, step_size, and max_drift
with
> appropriate values. If not implemented, return an error, and suitable
> defaults will be used.
>
> FE_NEEDS_BENDING. Indicates the frontend has problems with side-by-side
tuners
> being tuned to the same frequency.
>
> FE_NEEDS_POLL. Indicates that the FE_POLL ioctl should be called after
tuning.
>
> Module parameter dvb_force_auto_inversion. Workaround for buggy client
apps;
> if set to 1, forces INVERSION_AUTO always. The apps should really be fixed
to
> always use INVERSION_AUTO to start with, and remember the setting
determined
> by the frontend, but in the meantime...... Default is off.
>
> Module parameter dvb_override_frequency_bending. Has three possibilities:
> 0: default. Frequency bending will occur iff a frontend has
FE_NEEDS_BENDING
> set.
> 1: Frequency bending will never occur.
> 2: Frequency bending will always occur.
> The default is 0.
>
> I've updated all the frontends with the new flags. Note that any frontend
> previously marked with FE_CAN_RECOVER now also returns a max_drift value
of
> 0. I don't know if this is correct, but it means the behaviour should be
the
> same as how it was with the previous code. I don't have any of these
> frontends, but if someone who does wants to finetune these settings,
please
> do so!
>
> Also, any frontends which previously implemented FE_RESET, but didn't
really
> need it have been changed so they execute the code as part of
> FE_SET_FRONTEND. Basically, any frontend except the ves1x93.
>
> None of the frontends have had FE_NEEDS_BENDING set, as I do not know
which
> require it. Certainly the tda1004x and the su1278 frontends do not. I've
> tried machines with 5 cards in them all tuned to the same frequency with
no
> problems.


Hi,

the ves1x93 driver do FE_RESET (aka ves1x93_clr_bit) also at the end of
ves1x93_set_symbolrate (after tuning), so i think we should remove this
also. Or need this tuner clr_bit twice?

Greetings

Andreas Share



-- 
Info:
To unsubscribe send a mail to ecartis@linuxtv.org with "unsubscribe linux-dvb" as subject.