HVR 4000 hybrid card still producing multiple frontends for single adapter

HVR 4000 hybrid card still producing multiple frontends for single adapter

[Hawes, Mark]

Hi,

I have a HVR 4000 hybrid card  which provides both DVB-S2 and DVB-T capabilities on the one adapter. Using the current media tree build updated with the contents of the linux media drivers tarball dated 22/01/2012 the drivers for this card are still generating two frontends on the adapter as below:

> Jan 23 12:16:44 Nutrigrain kernel: [    9.346240] DVB: registering adapter 1 frontend 0 (Conexant CX24116/CX24118)...
> Jan 23 12:16:44 Nutrigrain kernel: [    9.349110] DVB: registering adapter 1 frontend 1 (Conexant CX22702 DVB-T)...

I understand that this behaviour is now deprecated and that the correct behaviour should be to generate one front end with multiple capabilities. Can this please be corrected.

Thanks,

Mark Hawes.

Re: HVR 4000 hybrid card still producing multiple frontends for single adapter

[Antti Palosaari]

On 01/24/2012 06:41 AM, Hawes, Mark wrote:
> Hi,
>
> I have a HVR 4000 hybrid card  which provides both DVB-S2 and DVB-T capabilities on the one adapter. Using the current media tree build updated with the contents of the linux media drivers tarball dated 22/01/2012 the drivers for this card are still generating two frontends on the adapter as below:
>
>> Jan 23 12:16:44 Nutrigrain kernel: [    9.346240] DVB: registering adapter 1 frontend 0 (Conexant CX24116/CX24118)...
>> Jan 23 12:16:44 Nutrigrain kernel: [    9.349110] DVB: registering adapter 1 frontend 1 (Conexant CX22702 DVB-T)...
>
> I understand that this behaviour is now deprecated and that the correct behaviour should be to generate one front end with multiple capabilities. Can this please be corrected.

Same applies for many other devices too. For example some older Anysee 
E7 models have two chip and two frontends whilst new one have only one. 
Also TechnoTrend CT3650 and Hauppauge WinTV.

Maybe it those are implemented later as one frontend, it not clear for me.


Antti
-- 
http://palosaari.fi/

Re: HVR 4000 hybrid card still producing multiple frontends for single adapter

[Devin Heitmueller]

On Tue, Jan 24, 2012 at 6:48 AM, Antti Palosaari <crope@iki.fi> wrote:
> On 01/24/2012 06:41 AM, Hawes, Mark wrote:
>>
>> Hi,
>>
>> I have a HVR 4000 hybrid card  which provides both DVB-S2 and DVB-T
>> capabilities on the one adapter. Using the current media tree build updated
>> with the contents of the linux media drivers tarball dated 22/01/2012 the
>> drivers for this card are still generating two frontends on the adapter as
>> below:
>>
>>> Jan 23 12:16:44 Nutrigrain kernel: [    9.346240] DVB: registering
>>> adapter 1 frontend 0 (Conexant CX24116/CX24118)...
>>> Jan 23 12:16:44 Nutrigrain kernel: [    9.349110] DVB: registering
>>> adapter 1 frontend 1 (Conexant CX22702 DVB-T)...
>>
>>
>> I understand that this behaviour is now deprecated and that the correct
>> behaviour should be to generate one front end with multiple capabilities.
>> Can this please be corrected.
>
>
> Same applies for many other devices too. For example some older Anysee E7
> models have two chip and two frontends whilst new one have only one. Also
> TechnoTrend CT3650 and Hauppauge WinTV.
>
> Maybe it those are implemented later as one frontend, it not clear for me.

The merging of frontends is something that is only done if there are
multiple modulation types on the same demodulator chip.  As the
HVR-4000 has separate demods for DVB-T versus DVB-S2, they will always
be represented by two separate frontends (for the foreseeable future).

In other words, the recent work doesn't apply to this card (and others like it).

Devin

-- 
Devin J. Heitmueller - Kernel Labs
http://www.kernellabs.com

Re: HVR 4000 hybrid card still producing multiple frontends for single adapter

[Antti Palosaari]

On 01/24/2012 04:49 PM, Devin Heitmueller wrote:
> On Tue, Jan 24, 2012 at 6:48 AM, Antti Palosaari<crope@iki.fi>  wrote:
>> On 01/24/2012 06:41 AM, Hawes, Mark wrote:
>>>
>>> Hi,
>>>
>>> I have a HVR 4000 hybrid card  which provides both DVB-S2 and DVB-T
>>> capabilities on the one adapter. Using the current media tree build updated
>>> with the contents of the linux media drivers tarball dated 22/01/2012 the
>>> drivers for this card are still generating two frontends on the adapter as
>>> below:
>>>
>>>> Jan 23 12:16:44 Nutrigrain kernel: [    9.346240] DVB: registering
>>>> adapter 1 frontend 0 (Conexant CX24116/CX24118)...
>>>> Jan 23 12:16:44 Nutrigrain kernel: [    9.349110] DVB: registering
>>>> adapter 1 frontend 1 (Conexant CX22702 DVB-T)...
>>>
>>>
>>> I understand that this behaviour is now deprecated and that the correct
>>> behaviour should be to generate one front end with multiple capabilities.
>>> Can this please be corrected.
>>
>>
>> Same applies for many other devices too. For example some older Anysee E7
>> models have two chip and two frontends whilst new one have only one. Also
>> TechnoTrend CT3650 and Hauppauge WinTV.
>>
>> Maybe it those are implemented later as one frontend, it not clear for me.
>
> The merging of frontends is something that is only done if there are
> multiple modulation types on the same demodulator chip.  As the
> HVR-4000 has separate demods for DVB-T versus DVB-S2, they will always
> be represented by two separate frontends (for the foreseeable future).
>
> In other words, the recent work doesn't apply to this card (and others like it).

So what was the actual benefit then just introduce one way more to 
implement same thing. As I sometime understood from Manu's talk there 
will not be difference if my device is based of DVB-T + DVB-C demod 
combination or just single chip that does same. Now there is devices 
that have same characteristics but different interface.

regards
Antti
-- 
http://palosaari.fi/

Re: HVR 4000 hybrid card still producing multiple frontends for single adapter

[Devin Heitmueller]

On Tue, Jan 24, 2012 at 9:58 AM, Antti Palosaari <crope@iki.fi> wrote:
> So what was the actual benefit then just introduce one way more to implement
> same thing. As I sometime understood from Manu's talk there will not be
> difference if my device is based of DVB-T + DVB-C demod combination or just
> single chip that does same. Now there is devices that have same
> characteristics but different interface.

For one thing, you cannot use DVB-T and DVB-C at the same time if
they're on the same demod.  With many of the devices that have S/S2
and DVB-T, you can be using them both in parallel.  Having multiple
frontends actually makes sense since you don't want two applications
talking to the same frontend at the same time but operating on
different tuners/streams.

That said, there could be opportunities for consolidation if the
demods could not be used in parallel, but I believe that would require
a nontrivial restructuring of the core code and API.  In my opinion
the entry point for the kernel ABI should *never* have been the
demodulator but rather the bridge driver (where you can exercise
greater control over what can be used in parallel).

Devin

-- 
Devin J. Heitmueller - Kernel Labs
http://www.kernellabs.com

Re: HVR 4000 hybrid card still producing multiple frontends for single adapter

[Antti Palosaari]

On 01/24/2012 05:16 PM, Devin Heitmueller wrote:
> On Tue, Jan 24, 2012 at 9:58 AM, Antti Palosaari<crope@iki.fi>  wrote:
>> So what was the actual benefit then just introduce one way more to implement
>> same thing. As I sometime understood from Manu's talk there will not be
>> difference if my device is based of DVB-T + DVB-C demod combination or just
>> single chip that does same. Now there is devices that have same
>> characteristics but different interface.
>
> For one thing, you cannot use DVB-T and DVB-C at the same time if
> they're on the same demod.  With many of the devices that have S/S2
> and DVB-T, you can be using them both in parallel.  Having multiple
> frontends actually makes sense since you don't want two applications
> talking to the same frontend at the same time but operating on
> different tuners/streams.

For the demods that are not shared (like tuner shared) we register own 
frontend under own adapter. I don't see that is going to change. It have 
been ages as it is and I have not seen none have said it is needed to 
change.

> That said, there could be opportunities for consolidation if the
> demods could not be used in parallel, but I believe that would require
> a nontrivial restructuring of the core code and API.  In my opinion
> the entry point for the kernel ABI should *never* have been the
> demodulator but rather the bridge driver (where you can exercise
> greater control over what can be used in parallel).

Under the current situation I see it is better to select only one 
method. As it is now single frontend then it is just needed to make 
"virtual" frontend that combines multiple frontends as single and offers 
it through API.

And one thing I would like to mention, frontend is just logical entity 
that represent DigitalTV hardware. It is rather much mapped as hardware 
point of view to demod driver callbacks but it is not needed :)


regards
Antti

-- 
http://palosaari.fi/

Re: HVR 4000 hybrid card still producing multiple frontends for single adapter

[Manu Abraham]

On Tue, Jan 24, 2012 at 8:28 PM, Antti Palosaari <crope@iki.fi> wrote:
> On 01/24/2012 04:49 PM, Devin Heitmueller wrote:
>>
>> On Tue, Jan 24, 2012 at 6:48 AM, Antti Palosaari<crope@iki.fi>  wrote:
>>>
>>> On 01/24/2012 06:41 AM, Hawes, Mark wrote:
>>>>
>>>>
>>>> Hi,
>>>>
>>>> I have a HVR 4000 hybrid card  which provides both DVB-S2 and DVB-T
>>>> capabilities on the one adapter. Using the current media tree build
>>>> updated
>>>> with the contents of the linux media drivers tarball dated 22/01/2012
>>>> the
>>>> drivers for this card are still generating two frontends on the adapter
>>>> as
>>>> below:
>>>>
>>>>> Jan 23 12:16:44 Nutrigrain kernel: [    9.346240] DVB: registering
>>>>> adapter 1 frontend 0 (Conexant CX24116/CX24118)...
>>>>> Jan 23 12:16:44 Nutrigrain kernel: [    9.349110] DVB: registering
>>>>> adapter 1 frontend 1 (Conexant CX22702 DVB-T)...
>>>>
>>>>
>>>>
>>>> I understand that this behaviour is now deprecated and that the correct
>>>> behaviour should be to generate one front end with multiple
>>>> capabilities.
>>>> Can this please be corrected.
>>>
>>>
>>>
>>> Same applies for many other devices too. For example some older Anysee E7
>>> models have two chip and two frontends whilst new one have only one. Also
>>> TechnoTrend CT3650 and Hauppauge WinTV.
>>>
>>> Maybe it those are implemented later as one frontend, it not clear for
>>> me.
>>
>>
>> The merging of frontends is something that is only done if there are
>> multiple modulation types on the same demodulator chip.  As the
>> HVR-4000 has separate demods for DVB-T versus DVB-S2, they will always
>> be represented by two separate frontends (for the foreseeable future).
>>
>> In other words, the recent work doesn't apply to this card (and others
>> like it).
>
>
> So what was the actual benefit then just introduce one way more to implement
> same thing. As I sometime understood from Manu's talk there will not be
> difference if my device is based of DVB-T + DVB-C demod combination or just
> single chip that does same. Now there is devices that have same
> characteristics but different interface.

Yes, you are right. I had a very preliminary patch to handle this,
Will post it soon.

Regards,
Manu

Re: HVR 4000 hybrid card still producing multiple frontends for single adapter

[Lars Hanisch]

Hi,

Am 24.01.2012 16:16, schrieb Devin Heitmueller:
> On Tue, Jan 24, 2012 at 9:58 AM, Antti Palosaari<crope@iki.fi>  wrote:
>> So what was the actual benefit then just introduce one way more to implement
>> same thing. As I sometime understood from Manu's talk there will not be
>> difference if my device is based of DVB-T + DVB-C demod combination or just
>> single chip that does same. Now there is devices that have same
>> characteristics but different interface.
>
> For one thing, you cannot use DVB-T and DVB-C at the same time if
> they're on the same demod.  With many of the devices that have S/S2
> and DVB-T, you can be using them both in parallel.  Having multiple
> frontends actually makes sense since you don't want two applications
> talking to the same frontend at the same time but operating on
> different tuners/streams.

  The two frontends of the HVR 4000 can only be open mutually exclusive so I think the recent changes are for those 
devices, aren't they? Sure you can connect DVB-T and DVB-S at the same time to the HVR 4000, but you can't use it in 
parallel.

Lars.

>
> That said, there could be opportunities for consolidation if the
> demods could not be used in parallel, but I believe that would require
> a nontrivial restructuring of the core code and API.  In my opinion
> the entry point for the kernel ABI should *never* have been the
> demodulator but rather the bridge driver (where you can exercise
> greater control over what can be used in parallel).
>
> Devin
>