Hi Song,

what's your command to get this log? it seems to have better information
than mine.

I just do old-school tracing:

# enable all available events
echo 1 > /sys/kernel/debug/tracing/events/enable
# start tracing
echo 1 > /sys/kernel/debug/tracing/tracing_on

Then I run cyclictest with -b and --tracemark to stop tracing when a
latency hits and write a mark to the file.
The trace can then be obtained by:
cat /sys/kernel/debug/tracing/trace

HTH
Christoph