verbose tracing doesn't refresh #526
Copy link
Copy link
Closed
Labels
GLKApplies to Gemini Lake platformApplies to Gemini Lake platformP2Critical bugs or normal featuresCritical bugs or normal featuresbugSomething isn't working as expectedSomething isn't working as expectedverifiedQA/Bug reporter verifies solution worked.QA/Bug reporter verifies solution worked.
Description
sof-logger -t doesn't refresh under one condition, when runtime is suspended, run sof-logger first and then do audio playback.
How to reproduce:
- enable verbose in trace.h
- make sure runtime is suspended
cat /sys/bus/pci/devices/0000:00:0e.0/power/runtime_status - open the trace log on one terminal
sof-logger -l sof-apl.ldc -t - run audio in another terminal
speaker-test -Dhw:0,1 -c 2 -t wav - check the trace log output, no refresh.
But if we run audio first then open trace log, no such issue, trace log will continue to update.
There is no obvious log in kernel dmesg. logger output shows as below:
CORE LEVEL COMP_ID TIMESTAMP DELTA FILE_NAME CONTENT
0 2 WAIT 12693428.697917 12693429.000000 ../../src/include/sof/wait.h:76 WFX
0 2 PIPE 12693432.447917 3.750000 schedule.c:368 sch
0 2 WAIT 12693436.666667 4.218750 ../../src/include/sof/wait.h:73 WFE
0 2 WAIT 12694428.697917 992.031250 ../../src/include/sof/wait.h:76 WFX
0 2 PIPE 12694432.604167 3.906250 schedule.c:368 sch
0 2 WAIT 12694436.875000 4.270833 ../../src/include/sof/wait.h:73 WFE
0 2 WAIT 12695428.697917 991.822937 ../../src/include/sof/wait.h:76 WFX
0 2 PIPE 12695432.447917 3.750000 schedule.c:368 sch
0 2 WAIT 12695436.822917 4.375000 ../../src/include/sof/wait.h:73 WFE
0 2 WAIT 12696428.697917 991.875000 ../../src/include/sof/wait.h:76 WFX
Reactions are currently unavailable
Metadata
Metadata
Labels
GLKApplies to Gemini Lake platformApplies to Gemini Lake platformP2Critical bugs or normal featuresCritical bugs or normal featuresbugSomething isn't working as expectedSomething isn't working as expectedverifiedQA/Bug reporter verifies solution worked.QA/Bug reporter verifies solution worked.
Activity
lgirdwood commentedon Nov 2, 2018
@keyonjie could be related to your recent fix for debufFS PM runtime ?
lgirdwood commentedon Nov 2, 2018
@keqiaozhang please use markdown when pasting logs. See above changes I've made.
keqiaozhang commentedon Nov 2, 2018
@lgirdwood thanks, Learned.
xiulipan commentedon Nov 2, 2018
@lgirdwood
I think NO. thesofproject/linux#237 is not merged yet.
And another reason is dma-trace used another file_ops as other debugfs.
https://github.com/thesofproject/linux/blob/a5ddc3d78b47b4d3e77ac371b6f016f3a6a9278c/sound/soc/sof/trace.c#L117
.read = sof_dfsentry_trace_read,https://github.com/thesofproject/linux/blob/a5ddc3d78b47b4d3e77ac371b6f016f3a6a9278c/sound/soc/sof/debug.c#L92
.read = sof_dfsentry_read,So maybe we should goto
sof_dfsentry_trace_readfunction to have a look.mengdonglin commentedon Nov 8, 2018
@keyonjie Is any of your recent PR will fix this issue?
keyonjie commentedon Nov 8, 2018
@mengdonglin yes, PR thesofproject/linux#248 should fix this.
keyonjie commentedon Nov 20, 2018
thesofproject/linux#287 merged to sof-dev, @ranj063 please backport it to cros branch then we can verify and close this issue on GLK.
keqiaozhang commentedon Nov 27, 2018
Confirmed that this issue is fixed.
Tested with:
kernel-cros-sof-v4.14-rebase: f8f388ef
sof-master: 443233
tplg-master: ad3f3e4