[BUG][Zephyr] FW not suspended after dmic test

[kv2019i]

Describe the bug
check-runtime-pm-status-15 fails on SOF Zephyr build on UP WHL board
See Intel internal daily plan #9293

To Reproduce
TPLG=/lib/firmware/intel/sof-tplg/sof-hda-generic-4ch.tplg MODEL=WHL_UPEXT_HDA_ZEPHYR ~/sof-test/test-case/check-runtime-pm-status.sh -l 15

Reproduction Rate

Expected behavior
Test should pass (system should go to runtime suspend)

Impact
Increased power consumption.

Environment

1. Branch name and commit hash of the 2 repositories: sof (firmware/topology) and linux (kernel driver).
   • Kernel: 55744127
   • SOF Commit: 9c002f3
   • Zephyr Commit: 768c5d42c0bd
2. Name of the topology file
   • Topology: sof-hda-generic-4ch.tplg
3. Name of the platform(s) on which the bug is observed.
   • Platform: {PLATFORM}

Screenshots or console output

[kv2019i]

Can you @marc-hb take a look (you have the board)? Tagging P1 as this is blocking CI.

[XiaoyunWu6666]

This may not be a single issue . @kv2019i @marc-hb

Below are some findings observed from inner CI daily testresults and our issue identification records :
Since UP WHL board was added to CI (late Nov ,2021) , it had the issue of"ipc timed out when TRIG_STOP with multiple-pipeline test" .
check-runtime-pm-status-15 is the case that goes after this , and it got affected and always failed at the first round (so it appears to fail to suspend after dmic arecord)

[XiaoyunWu6666]

I add a 'reset' after multiple-pipeline-all-50 testcase , then the runtime-pm-status doesn't get affected
please check inner planresult 9342

Also check inner daily 9399

[miRoox]

This failure is caused by the previous failure about multiple pipelines, which is already reported in #4680. As @XiaoyunWu6666 said, if we reboot system after multiple-pipeline-all-50, the subsequent tests won't fail. Therefore, this is not a real bug but just DSP state had been corrupted at this point.