At Wed, 4 Jul 2007 15:37:22 -0500, Matt Mullins wrote:
On a whim, I decided to extract the sound/pci/hda from vanilla 2.6.20 kernel sources and compile it under 2.6.22-rc5 (yes, I know, I probably shouldn't have done that). I had to add a typedef to patch_sigmatel.c to get it to compile. When I loaded the modules, I did *not* get the "switching to single_cmd" error, but I haven't gotten sound yet, either, probably because I'm using pieces of two different versions of ALSA. However, this suggests to me that perhaps there is a problem with the hda_intel driver, and not necessarily an ACPI problem.
That's possible. It might be that an access to an unsupported widget may trigger the unexpected behavior of the codec chip and the communication is disturbed.
Maybe it's worth to trace which verb call triggers the communication breakage. (BTW, for testing, you should concentrate only on audio codec, so skip modem codec via probe_mask option.)
Takashi