On 2023-10-06 10:44 AM, Takashi Iwai wrote:
On Tue, 26 Sep 2023 10:06:19 +0200, Cezary Rojewski wrote:
The patchset targets two intertwined topics:
The driver shall poll SDxFIFOS to ensure a valid value is set by the controller after programming SDxFMT. Due to amount of users and limited-number of configuration available in our CI when compared to overall possibilities on the market, the check is non-blocking.
Second topic relates to stream setup procedure. The procedure differs between HDAudio controller device revisions. Right now those differences are handled directly by a platform driver. Existing top-level 'if (pci->device == APL)' could be replaced by a abstraction in lower parts of the code instead.
With that done, the two users are updated accordingly. In avs-driver case, this updates the flow to the recommended one.
Changes in v3:
- fixed issues pointed out by scripts/kernel-doc
Changes in v2:
- fixed ->host_setup assignment in patch 02/04
Cezary Rojewski (4): ALSA: hda: Poll SDxFIFOS after programming SDxFMT ALSA: hda: Introduce HOST stream setup mechanism ASoC: Intel: avs: Use helper to setup HOST stream ASoC: Intel: Skylake: Use helper to setup HOST stream
Sorry for the late reaction, as I've been (still) off since the last week.
Now applied now to for-next branch.
Hello Takashi,
Now I'm the one sorry for the late replay - I've found two new things when fixing the debug message (reported by me in patch 1/4). - null-ptr-deref when assigning hdac_stream (when type=COUPLED) - azx_dev->fifo_size is initialized incorrectly
It's good to debug things, you never know what you may find!
May I send v4 as a collective update? It would address the two above and the false-positive debug message that appears when code-loading AudioDSP firmware.
Kind regards, Czarek