[wdmaudiodev] Re: Need help in Passing a test case from KS-Position test
- From: Gaurav Khuntale <gauravkhuntale@xxxxxxxxx>
- To: wdmaudiodev@xxxxxxxxxxxxx
- Date: Thu, 22 May 2014 14:25:09 +0530
Hi Matthew,
If i use below code in GetPosition, then complete KS-Position test is
getting passed,
LARGE_INTEGER currentTime = { 0 };
currentTime = KeQueryPerformanceCounter( NULL );
ULONGLONG timeElapsedInUs = currentTime.QuadPart -
m_dmaStartTime.QuadPart;
timeElapsedInUs = timeElapsedInUs * 1000000;
timeElapsedInUs = ( timeElapsedInUs + m_ullElapsedTimeCarryForward
) / m_startFrequency.QuadPart ;
m_ullElapsedTimeCarryForward = ( timeElapsedInUs +
m_ullElapsedTimeCarryForward ) % m_startFrequency.QuadPart ;
ULONG bytesDisplaceMent = ( m_ulDmaMovementRate *
(ULONG)timeElapsedInUs + m_ulByteDisplacementCarryForward ) / 1000000 ;
m_ulByteDisplacementCarryForward = ( m_ulDmaMovementRate *
(ULONG)timeElapsedInUs + m_ulByteDisplacementCarryForward) % 1000000;
m_ulDmaPosition = ( m_ulDmaPosition + bytesDisplaceMent )%
m_ulDmaBufferSize;
*Position = m_ulDmaPosition;
m_dmaStartTime = currentTime;
Thanks.
On Thu, May 22, 2014 at 6:02 AM, Matthew van Eerde <
Matthew.van.Eerde@xxxxxxxxxxxxx> wrote:
> > I’m working on a better version which attempts to maintain sample
> granularity but eliminates the slop.
>
> OK, I think I have something. Just want to run it few some tests before I
> share.
>
> Unfortunately the WaveCyclic model does not allow miniports to report a
> correlated time + QPC, so extrapolation is forced. MSVad doesn't care much,
> but hardware-based drivers are affected.
> ------------------------------
> From: Gaurav Khuntale <gauravkhuntale@xxxxxxxxx>
> Sent: 5/5/2014 7:01 AM
>
> To: wdmaudiodev@xxxxxxxxxxxxx
> Subject: [wdmaudiodev] Re: Need help in Passing a test case from
> KS-Position test
>
> Hi Matthew,
>
> I tried below code in GetPosition() and driver passed the test case.
>
> ULONGLONG currentTime = KeQueryInterruptTime();
>
> ULONG timeElapsedInNS = (ULONG) (currentTime - m_ullDmaTimeStamp);
> // Diff in nanoseconds
>
> ULONGLONG bytesDisplaceMent = (ULONGLONG)( (ULONGLONG)m_ulDmaMovementRate
> * (ULONGLONG)timeElapsedInNS ) / 10000000;
>
> m_ulDmaPosition = (m_ulDmaPosition + (ULONG)bytesDisplaceMent) %
> m_ulDmaBufferSize;
>
> *Position = m_ulDmaPosition;
>
> m_ullDmaTimeStamp = currentTime;
>
> Is this a proper way to update position from GetPosition?
>
> Thanks,
> Gaurav
>
>
> On Sun, May 4, 2014 at 10:19 AM, Gaurav Khuntale <gauravkhuntale@xxxxxxxxx
> > wrote:
>
>> Hmmm..i don't have a way to get a position from the layer below me
>> because i am throwing the data over the network after processing it. I
>> don't have any hardware for which my driver is working.
>>
>> Can you please let me know when your better version with sample
>> granularity is ready?
>>
>> Thanks.
>>
>>
>> On Sat, May 3, 2014 at 6:09 PM, Matthew van Eerde <
>> Matthew.van.Eerde@xxxxxxxxxxxxx> wrote:
>>
>>> Do you have any way of getting a real position from the layer below
>>> you? That would be much better than relying on KeQueryInterruptTime(). Or
>>> are you throwing the data on the floor like MSVad does?
>>>
>>>
>>>
>>> Don’t use my code; it’s just a proof-of-concept to demonstrate that you
>>> can get much better than 0.29 ms jitter using sample accuracy instead of
>>> millisecond accuracy.
>>>
>>> In particular, I really don’t like this line:
>>>
>>>
>>>
>>> ULONG ulByteDisplacement = (ULONG)((m_ulDmaMovementRate * ullElapsed +
>>> ullHnsInHalfASec) / ullHnsPerSec);
>>>
>>>
>>>
>>> This line introduces slop into the byte position which accumulates over
>>> time. I think that’s why my skew actually regressed (but I’m not sure.)
>>>
>>>
>>>
>>> I’m working on a better version which attempts to maintain sample
>>> granularity but eliminates the slop.
>>>
>>>
>>>
>>>
>>>
>>
>>
>
Other related posts:
