[wdmaudiodev] Re: WLK tests behave VERY strange

  • From: Eugene Muzychenko <eugene@xxxxxxxxxxxxxx>
  • To: Tsai Tzung-Dar <wdmaudiodev@xxxxxxxxxxxxx>
  • Date: Sat, 29 Oct 2011 14:36:19 +0700

Hello Tsai,

> The test in MiniportWaveCyclic driver is pass

MSVAD is WaveCyclic-based too but it does not pass 1.10.

Could you please explain how a classic PortCls miniport driver could ever pass
this test? In both WaveCyclic and WavePci, a driver reports only its
internal stream position, it cannot know how many WRITE_STREAM
requests are queued to PortCls. KSPROPERTY_AUDIO_POSITION and
KSPROPERTY_AUDIO_POSITIONEX requests are processed only by PortCls.

I investigated how wavetest works in 1.10 case. It queues a
WRITE_STREAM request, starts the stream, waits a lot, queues another
WRITE_STREAM of the same size, waits a lot, cancels the second request
then expects stream position to reach *doubled* size (a sum of both
WRITE_STREAM requests). How a miniport driver can affect position
requests that PortCls handles without intercepting them?

> End Case: ID 16.26:1.10 : PASS : Standard Streaming\Rendering Pin
> Tests\Streaming buffer from different threads : [Sun Oct 23 00:21:44 2011]

What version of wavetest do you use? My wavetest from WLK 1.6.8367 has 
FileVersion 6.1.7601.17125.

Did you try to test MSVAD Simple with this wavetest?



WDMAUDIODEV addresses:
Post message: mailto:wdmaudiodev@xxxxxxxxxxxxx
Subscribe:    mailto:wdmaudiodev-request@xxxxxxxxxxxxx?subject=subscribe
Unsubscribe:  mailto:wdmaudiodev-request@xxxxxxxxxxxxx?subject=unsubscribe
Moderator:    mailto:wdmaudiodev-moderators@xxxxxxxxxxxxx


Other related posts: