On Thu, 12 Dec 2019 18:23:08 +0100, Takashi Iwai wrote:
On Thu, 12 Dec 2019 18:17:14 +0100, Kai Vehmanen wrote:
Hey,
On Tue, 10 Dec 2019, Takashi Iwai wrote:
This patch attempts to improve the situation by introducing the standard waitqueue in the RIRB waiter side instead of polling. The
this seems like a good change. It's a bit unfortunate this logic is not yet shared and SOF uses the variant in snd_hdac_bus_get_response() -- that would need similar treatment. I can make a matching change there and see how this fairs in SOF CI. But basic tests with snd-hda-intel on a few systems didn't reveal anything odd and code seems correct.
Yes, please, the adaption to SOF is really appreciated. This is a 5.6 material and I applied only the changes I could actually test as a start.
And now I realized that adapting this for the common code is trivial, and it's a good chance to unify the code there. I'll submit a test patch set soon later. Stay tuned.
Takashi