On Tue, Jun 02, 2009 at 12:18:54AM -0400, Mike Frysinger wrote:
snd_pcm_hw_constraint_minmax(substream->runtime,SNDRV_PCM_HW_PARAM_RATE,master_runtime->rate,master_runtime->rate);snd_pcm_hw_constraint_minmax(substream->runtime,SNDRV_PCM_HW_PARAM_SAMPLE_BITS,master_runtime->sample_bits,master_runtime->sample_bits);
if (master_runtime->rate != 0)snd_pcm_hw_constraint_minmax(substream->runtime,SNDRV_PCM_HW_PARAM_RATE,master_runtime->rate,master_runtime->rate);
if (master_runtime->sample_bits != 0)snd_pcm_hw_constraint_minmax(substream->runtime,SNDRV_PCM_HW_PARAM_SAMPLE_BITS,master_runtime->sample_bits,master_runtime->sample_bits);
This change is entirely unrelated to your changelog and should have been submitted as a separate patch. It's preventing the use of the configuration from an unconfigured master which isn't what the patch is about). It's also going over the 80 column limit.
For the rate constraint you should switch to using the symmetric_rates flag in the DAI which implements the equivalent constraints with no per-driver code.
However, given that this fix has missed two kernel releases due to the very slow rate of responses to review comments I've applied it so users get the benefit of the bug fix. Things would go more smoothly if you could get new versions of patches submitted faster after the original review.