[linux-dvb] How to access /dev/videoK from /dev/dvb/adapterN?

Manu Abraham abraham.manu at gmail.com
Sat Jun 17 22:09:54 CEST 2006


Klaus Schmidinger wrote:
> Manu Abraham wrote:
>> Ralph Metzler wrote:
>>> Klaus Schmidinger writes:
>>>  > An application that uses the LinuxDVB driver can access the
>>>  > DVB devices via /dev/dvb/adapterN/* (N = 0, 1, ...). However,
>>>  > when it wants to, for instance, capture the current image on
>>>  > a device (like VDR with its GRAB command), it needs to access
>>>  > /dev/videoK, where 'K' is a number that is not directly related
>>>  > to the 'N' mentioned above for the DVB device.
>>>  >  > So VDR has to do some guessing on which /dev/videoK device might
>>>  > belong to the /dev/dvb/adapterN device. This is getting particularly
>>>  > problematic if there are also other /dev/video devices that have
>>>  > nothing to do with any DVB devices.
>>>  >  > Is there a defined way in which an application can query a DVB
>>>  > device for its related /dev/video device?
>>>
>>> Currently not.
>>> Since upcoming cards with decoder and V4L interface for those will
>>> need it too, should we add a call for this? In the DVB video device?
>>> We might also need something like this for audio.
>>>   
>>
>> Like what we discussed earlier, what about making the nodes children 
>> to the adapter device, this will shoot at least 2 or 3 birds in one 
>> shot. ie a /dev/dvb/adapterX device, all are just child nodes of the 
>> adapter device.
>>
>> This would solve
>>
>> (1) solve the issue with the demux for multistandard devices
>> (2) This current situation
>> (3) Calls that might belong to more than one device/belonging to the 
>> card etc
>
> I wouldn't mind _where_ the video device is that I need to use
> for grabbing the image, so that would be fine with me.
>
> However, the name /dev/dvb/adapterN/videoM is already taken - or
> could that same node be used for image grabbing?

Since the adapter device "knows" what devices are present , we can 
probably handle subsets within the adapter device. It could say, for eg. 
have it's own capability field. One of the advantages is that this will 
make it easier to handle other devices as well.

Last day, i was looking at a device from the professional Broadcast 
industry, well eventually i think, we will need more abstraction out 
from the adapter device, similarly to the frontend device. These devices 
vary in functionality a lot from hardware to hardware and a generic way 
to handle them is necessary as well.

I asked for a second opinion from Ralph as well on the same aspect. 
Maybe he has some further ideas/thoughts on the same.

Well, in any case we need an adapter device for handling multi-standard 
devices, for different protocols. We did in fact discuss on the adapter 
device a while back as to handle the devices that which is not a part of 
the frontend/demux/ca, possible through the adapter device. So in any 
way, i think the adapter device superset, would be a road to the future.


Manu




More information about the linux-dvb mailing list