[Bug][HDA][WHL] Can't stay in runtime PM suspend status when headset connected.

[ClarexZhou]

Bug Description
Can't stay in runtime PM suspend status when headset connected.
Keeps boot firmware when DSP power state: D3.

To Reproduce

1. Connect Headset
2. Boot up system
3. Wait till runtim PM suspended

Reproduction Rate
5/5

Impact
Firmware will be auto triggered again.

Kernel: sof-dev commit: 77b49b3
FW:[2ef369d57][ https://github.com/thesofproject/sof-bin/tree/stable-v1.4.2/lib/firmware/intel/sof/v1.4.2]
tplg: Same with FW branch [sof-hda-generic-4ch.tplg ]
Platform:[WHL laptop with onboard codec ALC285 in HDA mode]
UCM2: https://github.com/thesofproject/alsa-ucm-conf, commit 4722f5b3

dmesg_0327_1837.txt
amixercontents_0327_1837.txt

[ClarexZhou]

The root cause is the sof-dev branch merged this upstream commit: 76f7dec (ALSA: hda/realtek - Add Headset Button supported for ThinkPad X1), with this commit, if we set the codec to auto-suspend, the codec driver will trigger a fake key event during calling rt_suspend(), so the driver will rt_resume() and rt_suspend() periodically.
This issue also exists in the legacy hda driver, and please set CONFIG_SND_HDA_POWER_SAVE_DEFAULT to 1 otherwise rt_pm is not neabled in the hda codec driver.

So it is not my patches introduced this regression, my patches exposed a hidden problem. I just sent a patch for fixing this problem: https://mailman.alsa-project.org/pipermail/alsa-devel/2020-March/165395.html

from #1837 (comment)

[ClarexZhou]

close as a codec issue and the fix has already been applied in upstream https://mailman.alsa-project.org/pipermail/alsa-devel/2020-March/165400.html