check-sof-logger.sh: fix no-log-file failures due to runtime Power Management by kv2019i · Pull Request #1109 · thesofproject/sof-test

kv2019i · 2023-10-04T11:47:59Z

The logger test case has a design limitation in that it assumes the SOF DSP to remain powered up between driver reload and start of the logger. This creates a possibility for a race with runtime PM (RPM).

If the DSP goes to runtime suspend during this window, it is not guaranteed that the SOF driver will provide any logs. Some logging mechanisms save the logs that were emitted before suspend (sof-logger/IPC3), but some not (cavstool/IPC3 and mtrace/IPC4).

To limit the window , add a simple ping of all PCM nodes just before logger process is started. This is a crude mechanism but something that works with all variants of SOF. End result is that DSP will be resumed to D0 just before the logger test starts.

The logger test case has a design limitation in that it assumes the SOF DSP to remain powered up between driver reload and start of the logger. This creates a possibility for a race with runtime PM (RPM). If the DSP goes to runtime suspend during this window, it is not guaranteed that the SOF driver will provide any logs. Some logging mechanisms save the logs that were emitted before suspend (sof-logger/IPC3), but some not (cavstool/IPC3 and mtrace/IPC4). To limit the window , add a simple ping of all PCM nodes just before logger process is started. This is a crude mechanism but something that works with all variants of SOF. End result is that DSP will be resumed to D0 just before the logger test starts. Signed-off-by: Kai Vehmanen <kai.vehmanen@linux.intel.com>

kv2019i · 2023-10-04T14:52:00Z

Only system-PM errors in https://sof-ci.01.org/softestpr/PR1109/build763/devicetest/index.html , https://sof-ci.01.org/softestpr/PR1109/build762/devicetest/index.html and https://sof-ci.01.org/softestpr/PR1109/build761/devicetest/index.html . And no cases of empty mtrace.

marc-hb · 2023-10-04T16:57:25Z

We don't run check-sof-logger in PR testing (like https://sof-ci.01.org/softestpr/PR1109/build761/devicetest/index.html), we never have - cause it was failing all the time. We only ever ran it in daily tests.

@fredoh9 could you add them to the PR test plan? Then we can SOFCI TEST this and merge.

fredoh9 · 2023-10-04T22:46:18Z

I added check-sof-logger.sh to mtl pr test.

And I manually ran check-sof-logger.sh with this PR on MTLP_RVP_HDA, it still failed. But I ran the test automatically thre result looks good.
planresultdetail/32679

fredoh9 · 2023-10-04T22:46:39Z

SOFCI TEST

fredoh9

I re-test again after re-deploying daily build. Now constantly PASS. I did not deploy daily build before, probably the build installed wasn't stable. There is no way I can track the build installed before.
Looks good!

fredoh9 · 2023-10-05T17:26:50Z

MTLP_RVP_HDA failed in PR test. Interestingly I tried on MTLP_RVP_HDA when I saw the failure with this PR. Apparently the result get better but not complete solution.

kv2019i · 2023-10-05T17:45:56Z

@fredoh9 I think this is a different problem:

Polling 'sudo test -e /proc/asound/card0/id' every 1s for 70s
Completed 'sudo test -e /proc/asound/card0/id' after 0s and 1 attempts
2023-10-04 23:00:14 UTC [REMOTE_INFO] Found /proc/asound/card0/id.
[...]
  File "/usr/local/bin/mtrace-reader.py", line 20, in 
    fd = os.open(MTRACE_FILE, os.O_RDONLY)
FileNotFoundError: [Errno 2] No such file or directory: '/sys/kernel/debug/sof/mtrace/core0'

I checked the DUT and there's the 60sec timeout due to missing i915 driver, and there's a non-SOF driver registered as card0, so the test case logic goes wrong.

We can try to make the test case logic more robust, but I think this is a different problem.

The check for DSP boot assumes SOF is always card0 in the system, and this leads to failure. Fix the issue by using the existing sof_firmware_boot_complete() check to wait for DSP boot to successfully complete. Signed-off-by: Kai Vehmanen <kai.vehmanen@linux.intel.com>

kv2019i · 2023-10-05T18:03:20Z

Ok, @fredoh9 , added a second patch to this PR to fix the issue that affects this specific DUT. The check should now pass even if some non-SOF ALSA driver is loaded as card0 (like is the case on this DUT).

fredoh9 · 2023-10-05T18:09:39Z

Thank you a lot! I will check again in this afternoon!

marc-hb · 2023-10-05T19:11:42Z

Interesting, do we have another race?
https://sof-ci.01.org/softestpr/PR1109/build768/devicetest/index.html

SKIP    	snd_sof_imx8m 	(not in tree)
MODPROBE	snd_usb_audio		 
2023-10-05 18:15:08 UTC [REMOTE_INFO] Polling , waiting for DSP boot...
Polling 'sof_firmware_boot_complete --since=@1696529699' every 1s for 3s
[  232.064597] kernel: snd_sof:snd_sof_run_firmware: sof-audio-pci-intel-tgl 0000:00:1f.3: firmware boot complete
Completed 'sof_firmware_boot_complete --since=@1696529699' after 0s and 1 attempts
2023-10-05 18:15:08 UTC [REMOTE_INFO] DSP booted successfully.
2023-10-05 18:15:08 UTC [REMOTE_INFO] Trying to get Zephyr logs from mtrace with background /usr/local/bin/mtrace-reader.py ...
2023-10-05 18:15:08 UTC [REMOTE_COMMAND] sudo  /usr/local/bin/mtrace-reader.py >  /home/ubuntu/sof-test/logs/check-sof-logger/2023-10-05-18:15:03-9615/logger.etrace.txt  2>  /home/ubuntu/sof-test/logs/check-sof-logger/2023-10-05-18:15:03-9615/logger.etrace_stderr.txt
Traceback (most recent call last):
  File "/usr/local/bin/mtrace-reader.py", line 20, in 
    fd = os.open(MTRACE_FILE, os.O_RDONLY)
FileNotFoundError: [Errno 2] No such file or directory: '/sys/kernel/debug/sof/mtrace/core0'
2023-10-05 18:15:10 UTC [REMOTE_ERROR] timeout /usr/local/bin/mtrace-reader.py returned unexpected: 1
2023-10-05 18:15:10 UTC [REMOTE_INFO] Starting func_exit_handler(1)
2023-10-05 18:15:10 UTC [REMOTE_ERROR] Starting func_exit_handler(), exit status=1, FUNCNAME stack:
2023-10-05 18:15:10 UTC [REMOTE_ERROR]  die()  @  /home/ubuntu/sof-test/case-lib/lib.sh
2023-10-05 18:15:10 UTC [REMOTE_ERROR]  run_loggers()  @  /home/ubuntu/sof-test/test-case/check-sof-logger.sh:166
2023-10-05 18:15:10 UTC [REMOTE_ERROR]  main()  @  /home/ubuntu/sof-test/test-case/check-sof-logger.sh:271
2023-10-05 18:15:10 UTC [REMOTE_ERROR]  main()  @  /home/ubuntu/sof-test/test-case/check-sof-logger.sh:364
2023-10-05 18:15:10 UTC [REMOTE_INFO] pkill -TERM -f mtrace-reader.py
2023-10-05 18:15:10 UTC [REMOTE_INFO] ktime=238 sof-test PID=3794: ending
2023-10-05 18:15:10 UTC [REMOTE_INFO] Test Result: FAIL!

fredoh9 · 2023-10-05T23:04:25Z

I tried on MTLP_RVP_HDA,

2023-10-05 22:56:14 UTC [INFO] Trying to get Zephyr logs from mtrace with background /usr/local/bin/mtrace-reader.py ...
2023-10-05 22:56:14 UTC [COMMAND] sudo  /usr/local/bin/mtrace-reader.py >  /home/ubuntu/sof-test/logs/check-sof-logger/2023-10-05-22:56:08-15319/logger.etrace.txt  2>  /home/ubuntu/sof-test/logs/check-sof-logger/2023-10-05-22:56:08-15319/logger.etrace_stderr.txt
Traceback (most recent call last):
  File "/usr/local/bin/mtrace-reader.py", line 20, in <module>
    fd = os.open(MTRACE_FILE, os.O_RDONLY)
FileNotFoundError: [Errno 2] No such file or directory: '/sys/kernel/debug/sof/mtrace/core0'

And trigger sof-logger test plan also, Similar errors found.
planresultdetail/32724

lib.sh:sof_firmware_boot_complete() ensures DSP has booted, but it does not guarantee SOF driver probe has completed. So it is still possible some of the interfaces SOF driver exposes to user-space are not available (including interfaces needed to run loggers). Tracing is initialized before ALSA card registration in SOF diver (see snd_sof_device_probe() in SOF driver), so by waiting for ALSA card to show up in procfs, we can guarantee tracing related interfaces have been initialized. Signed-off-by: Kai Vehmanen <kai.vehmanen@linux.intel.com>

marc-hb · 2023-10-06T14:28:37Z

check-sof-logger now all passed in https://sof-ci.01.org/softestpr/PR1109/build772/devicetest/index.html . By chance or design?

marc-hb · 2023-10-06T14:28:54Z

SOFCI TEST

fredoh9 · 2023-10-06T17:31:43Z

I re-ran sof-logger test, all passed!
planresultdetail/32767

marc-hb

Very clean code, easy to fix later if needed. Will hopefully help with current PRs, merging.

marc-hb · 2023-10-06T19:51:44Z

+
+wait_for_sof_alsa_card()
+{
+    if poll_wait_for 1 "$MAX_WAIT_FW_LOADING" sof_alsa_card_found ; then


In a follow-up PR please:

Suggested change

if poll_wait_for 1 "$MAX_WAIT_FW_LOADING" sof_alsa_card_found ; then

if poll_wait_for 1 "$1" sof_alsa_card_found ; then

marc-hb · 2023-10-09T14:30:41Z

mtrace still empty in 2 out of 6 devices in today's daily test run 32787

2023-10-08 12:29:48 UTC [REMOTE_ERROR] Log header not found in /home/ubuntu/sof-test/logs/check-sof-logger/2023-10-08-12:29:19-18463/logger.etrace.txt

The No such file or directory message is gone though!

kv2019i requested a review from a team as a code owner October 4, 2023 11:47

marc-hb previously approved these changes Oct 4, 2023

View reviewed changes

fredoh9 previously approved these changes Oct 4, 2023

View reviewed changes

kv2019i dismissed stale reviews from fredoh9 and marc-hb via 4bb1acb October 5, 2023 18:02

marc-hb previously approved these changes Oct 5, 2023

View reviewed changes

kv2019i dismissed marc-hb’s stale review via 1741c33 October 6, 2023 06:55

marc-hb mentioned this pull request Oct 6, 2023

west.yml: Bump up Zephyr revision thesofproject/sof#8295

Merged

fredoh9 approved these changes Oct 6, 2023

View reviewed changes

marc-hb approved these changes Oct 6, 2023

View reviewed changes

marc-hb merged commit e3d7664 into thesofproject:main Oct 6, 2023

marc-hb mentioned this pull request Apr 26, 2024

[BUG] Zephyr "mtrace" logs empty until the next reboot thesofproject/sof#9087

Closed

marc-hb changed the title ~~check-sof-logger.sh: fix no-log-file failures due to RPM~~ check-sof-logger.sh: fix no-log-file failures due to runtime Power Management Apr 26, 2024

	if poll_wait_for 1 "$MAX_WAIT_FW_LOADING" sof_alsa_card_found ; then
	if poll_wait_for 1 "$1" sof_alsa_card_found ; then

Conversation

kv2019i commented Oct 4, 2023

Uh oh!

kv2019i commented Oct 4, 2023

Uh oh!

marc-hb commented Oct 4, 2023

Uh oh!

fredoh9 commented Oct 4, 2023

Uh oh!

fredoh9 commented Oct 4, 2023

Uh oh!

fredoh9 left a comment

Choose a reason for hiding this comment

Uh oh!

fredoh9 commented Oct 5, 2023

Uh oh!

kv2019i commented Oct 5, 2023

Uh oh!

kv2019i commented Oct 5, 2023

Uh oh!

fredoh9 commented Oct 5, 2023

Uh oh!

marc-hb commented Oct 5, 2023

Uh oh!

fredoh9 commented Oct 5, 2023

Uh oh!

marc-hb commented Oct 6, 2023

Uh oh!

marc-hb commented Oct 6, 2023

Uh oh!

fredoh9 commented Oct 6, 2023

Uh oh!

marc-hb left a comment

Choose a reason for hiding this comment

Uh oh!

marc-hb Oct 6, 2023

Choose a reason for hiding this comment

Uh oh!

marc-hb commented Oct 9, 2023 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Reviewers

Uh oh!

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

marc-hb commented Oct 9, 2023 •

edited

Loading