Bug #992
closedPacket Context end time before start time
0%
Description
In a 32 bit system, setting up rolling traces in UST on lttng 2.7 makes end times of packet contexts appear BEFORE the start time. This is an issue since then seeking within a trace file does not work anymore.
I have attached a sample trace to show the problem.
Babeltrace reads it fine and does not even warn about the problem.
Tracecompass fails with an error message "A trace has failed validation
Reading error: org.eclipse.tracecompass.ctf.core.CTFException: Packet begin timestamp is after end timestamp
Packet begin timestamp is after end timestamp"
Feel free to efficiosify the trace and put it in the test-spec if you want.
Files
Updated by Bernd Hufmann about 8 years ago
See below for information about the tracing session configuration that caused such a trace. The application is running on a 3.10.55 Kernel with PREEMPT_RT and LTTng 2.7.0.
Session configuration:
lttng create --snapshot mysession
lttng enable-channel mychannel -u -s mysession --switch-timer 200000
lttng enable-event -s mysession -u ust* -c mychannel
lttng start mysession
...
lttng stop mysession
lttng snapshot record -s mysession /tmp
Note: The problem doesn't happen if the switch-timer is set to 0
Updated by Jonathan Rajotte Julien about 4 years ago
- Status changed from New to Invalid
State of babeltrace moved a lot since.
Closing this ticket as invalid. Reopen it if it stills apply to Babeltrace 2.