On Sun, 18 Apr 2021 16:16:56 +0200, Geraldo Nascimento wrote:
Em Dom, 18 de abr de 2021 09:45, Olivia Mackintosh livvy@base.nu escreveu:
> As a blind shot: might the stall be avoided by passing the recently > introduced chip->playback_first flag? The revised patch is below. This appears to fix the issue, thank you! I am curious as to why this is not required for Geraldo's DDJ-SR2.Heh, sometimes I think Takashi Iwai is psychic ;-)
Heh, it was our luck that we've worked on the similar problem very recently.
Just kidding, he's "just" very talented and very experienced. Anyway, the DDJ-SR2 is originally a Serato controller. It still requires two MIDI SysEx messages to unmute the RCA inputs - engaging "Serato Mode" in marketing-speak.
My point is that Pioneer may be a little more compliant / a little less deviant from the UAC2 standard when designing gear for use with third-party software.
I think many devices require the implicit feedback seem following this pattern: for such devices, the full-duplex streams are mandatory, i.e. both streams have to be started always. And the implicit feedback is triggered at the later point for a proper sync.
My initial thought was that certain devices that restrict capture unless regular SysEx/Control URBs are sent may mean that Playback first support is required (e.g. 750, 850, 450). Please correct me if this intuition is incorrect.
As mentioned previously, it might be the difference of the PCM parameters to be deployed. I guess Geraldo didn't test speaker-test program but maybe only the standard sound backends like JACK, etc.
In anyway, I'm going to format and submit the proper patch for merging.
thanks,
Takashi