CPU : Davinci DM6467
OS : Linux 2.6.38
On our current product we are trying to read 8, 32bit streams/channels on the McASP.
The streams are interleaved and hence after reading a bloack of data using Sound_read call we separate the streams in software.
What we observe is that the amount of data received is more than the calculated data size that needs to come for the configured
sample rate.
This causes problems when we try to compare the received data with a reference data for fingerprint matching.
What could be the possible cause for this since we have observed that there is no clock skew (using oscilloscope) and
also that the davinci-pcm isr runs in correct time as is calculated for the sample rate configured.