On Wed, Jun 22, 2011 at 8:36 PM, Mark Brown broonie@opensource.wolfsonmicro.com wrote:
On Wed, Jun 22, 2011 at 08:21:06PM +0200, Daniel Mack wrote:
I thought so too, but I'm facing an Ooops on 3.0.0-rc4, while the same platform code was running fine on 2.6.39. And while searching for potential reasons for this regression, this commit was quite obvious. But maybe that's indeed fixing the wrong side. What happens is that pxa2xx_pcm_preallocate_dma_buffer() is called for SNDRV_PCM_STREAM_CAPTURE, but pcm->streams[1].substream is NULL.
Is this the same issue that Timur was having where he was relying on the channel count of one of the DAI drivers to tell him if the substream was allocated instead of looking to see if the substream was allocated which is the more direct and reliable thing to do? Looking at the code it appears that's the case for PXA.
And so the bug was always there, it was just hidden by another bug. Thanks for the pointer :)
Daniel