Re: [dvbtools-devel] [linux-dvb] Re: dvbtune-0.5 inverts diseqc pol and band bits

[Zoilo G. <zo...@xs...>]

Nico Sabbi wrote:
> Nico Sabbi wrote:
>> Zoilo Gomez wrote:
>>
>>>>> The Diseqc-specs at 
>>>>> http://www.eutelsat.com/satellites/pdf/Diseqc/Reference%20docs/bus_spec.pdf 
>>>>> (table on page 13) confirm that dvbtune is in fact sending the 
>>>>> wrong command.
>>>>>
>>>>> With a patched dvbtune (pol-bit and band-bit both inverted) it 
>>>>> works OK, and is also according to the Eutelsat specs.
>>>>>
>>>
>>> I have attached a patch for you.
>>>
>>> The problem is only in dvbtune (version 0.5); you may not have been 
>>> using dvbtune (dvbtools); dvbscan and szap (linuxtv-dvb-apps) do not 
>>> suffer from this problem.
>>>
>>> Z.
>>>
>>>
>>> ------------------------------------------------------------------------ 
>>>
>>>
>>> diff -Naur dvbtune-0.5.orig/tune.c dvbtune-0.5/tune.c
>>> --- dvbtune-0.5.orig/tune.c    2004-02-06 15:00:36.000000000 +0100
>>> +++ dvbtune-0.5/tune.c    2007-04-30 21:22:53.000000000 +0200
>>> @@ -203,7 +203,7 @@
>>>      * bits are: option, position, polarizaion, band
>>>      */
>>>     cmd.cmd.msg[3] =
>>> -       0xf0 | (((sat_no * 4) & 0x0f) | (hi_lo ? 1 : 0) | (pol ? 0 : 
>>> 2));
>>> +       0xf0 | (((sat_no * 4) & 0x0f) | (hi_lo ? 0 : 1) | (pol ? 2 : 
>>> 0));
>>>  
>>>     diseqc_send_msg(secfd, pol,
>>>             &cmd, hi_lo,
>>
>>
>
>
> I disagree with your changes:
> according to the specs that you linked the fields are:
> *switch-number
> *switch option
> *satellite position
> *polarization
> *frequency_is_in_hi_band
> *bytecode
>
>
>                         1 A A V Lo F0
>                         2 A A V Hi F1
>                         3 A A H Lo F2
>                         4 A A H Hi F3
>                         5 A B V Lo F4
>                         6 A B V Hi F5
>                         7 A B H Lo F6
>                         8 A B H Hi F7
>                         9 B A V Lo F8
>                        10 B A V Hi F9
>                        11 B A H Lo FA
>                        12 B A H Hi FB
>                        13 B B V Lo FC
>                        14 B B V Hi FD
>                        15 B B H Lo FE
>                        16 B B H Hi FF
>
> assuming that in the code "pol" stands for "pol_is_vertical" and that 
> "hi_lo" stands for "hi_band" ( frequency above SLOF) and counting bits 
> from 8..1
> - the last bit must be set to 1 when hi_band is 1
> - the forelast bit (0x2) must be set when the polarization is 
> horizontal (this polv is 0).
>
> Please, explain why you believe that the current code is incorrect

Not so much a matter of believe, but of results.

My switch was not working, and selecting the wrong input. Hence I looked 
into the source, put some printfs and found that the DiSEqC command 
being sent was incorrect, i.e. different from the table above. So I 
changed the code, then it was working correctly.

Perhaps it would be better to change the code in the calling function; 
your opinion that (hi_lo==1) means high band seems logical, but in that 
case the value being passed to the function is wrong.

(pol==1) meaning "vertical" seems somewhat more ambiguous to me, perhaps 
a better name would be "vpol" (in which case however the calling 
function is sending the wrong value, inverted) or hpol (in which case 
it's OK).

Z.