Re: [Qemu-devel] [PATCH 12/25] audio: basic support for multi backend audio

["Kővágó Zoltán" <dirty.ice.hu@gmail.com>]

Hi

2015-08-20 20:01 keltezéssel, Marc-André Lureau írta:
[snip]
>> diff --git a/qemu-options.hx b/qemu-options.hx
>> index efd57e6..60a3563 100644
>> --- a/qemu-options.hx
>> +++ b/qemu-options.hx
>> @@ -1538,6 +1538,11 @@ everybody else.  'ignore' completely ignores the shared flag and
>>   allows everybody connect unconditionally.  Doesn't conform to the rfb
>>   spec but is traditional QEMU behavior.
>>
>> +@item audiodev=@var{audiodev}
>> +
>> +Use the specified @var{audiodev} when the VNC client requests audio
>> +transmission.
>
> That doesn't make much sense to me, vnc (or spice) should be connected
> to the guest device, not to the audiodev. Same for for the wavcapture.

Hmm.  Maybe it would make more sense, but it's not how the audio system 
works.  The situation cyrrently (before my patch) is the following: 
VNC/spice/wavcapture attaches a capture to the global AudioState and 
starts recording/transmitting the mixed together audio of all connected 
guest sound cards.

Now with more audiodevs, you have to specify which one you wan't to 
capture from.  If there's multiple sound cards attached to a single 
audiodev (AudioState), then we will transmit the mix of them.  (If 
there's only one attached though, then we will only transmit that...)

To support capture from a single guest device would require some 
rewriting of the audio capture logic, in a backward compatible way 
(which would mean support both capturing form a single card and 
capturing a mix of all cards.  And in the latter case we still need some 
way to specify audiodev...)

Zoltan