ASoC: SOF: HDA: initialize period_bytes at prepare for Code Loader and Trace

[keyonjie]

This is aimed to fix#308
Just compile passed, not tested yet.

For Code Loader and DMA Trace, we don't have hw_params() stage, but do
the related settings at prepare(), we should set period_bytes of them to
be 0 to make BDL setting up works, here add this explicit initialization
for both of them.

Without this initialization, the first_boot works fine, but Code Loader
or DMA Trace may failed after suspend/resume, as the stream might be used
by other streams with different period_bytes setting before suspend.

Signed-off-by: Keyon Jie yang.jie@linux.intel.com

[ranj063]

@keyonjie instead of doing this individually for each case, does it make sense to add this in hda_dsp_stream_put() instead to cover any future issues?

[lgirdwood]

@keyonjie@ranj063 yes, it does make sense to do it in one place but I've merged this now since it's a blocker for HDA. @keyonjie or @ranj063 incremental update if you have time tonight or tomorrow.

[keyonjie]

@keyonjie instead of doing this individually for each case, does it make sense to add this in hda_dsp_stream_put() instead to cover any future issues?

Yes, I thought about that, we need not only reset period_bytes, we need add resetting to all hw_params related at put() next step.

[ranj063]

@keyonjie so will you be doing that next or do you want me to look into it?

[keyonjie]

@ranj063 I am thinking that should we do that in get() or put(), or stream_hw_params().

Doing something in only put() but no opposite in get() looks somewhat odd.

[keyonjie]

@ranj063 maybe better to keep current solution, initialize period_bytes/frags/bufsize actually is must for PCM stream, so the same for CL and DMA trace, we should treat it as a kind of initialization, not a kind of reset.