[SOLVED] Post-Mortem Debug Using Recorded Trace Data

schmalr · Dec 13th 2018, 11:52pm

I am trying to figure out how to record trace data from a unit under test and then play that data back for post-mortem debugging. In section 3.11.11 of the Ozone documentation (2.60 Revision 3), it shows a field at the bottom of the Trace Settings dialog for setting the "Trace File". However, in my Ozone (V2.60i), that field does not exist. I have hunted for hours trying to find an equivalent setting in Ozone or SEGGER Embedded Studio, with no success. Then, if I did succeed in generating a trace file, I've seen nothing anywhere in the programs or documentation to suggest how I would set the debugger up to analyze that data file. Please give me some guidance on this.

Russ

SEGGER - Nino · Dec 14th 2018, 5:48pm

Hello Russ,

Thank you for your inquiry.
Not sure if we understand your setup correctly.
So you are tracing via pins and are using Ozone. Correct?
Which target device are you tracing?
You now want to dump the instruction trace data into a local file and load it after ending the debug session in Ozone for post analysis?
If you answered that question with yes then we can inform you about the great news that exactly that feature will be part of our next big J-Trace PRO feature extension planned for the beginning of next year.
Then you will be able to dump an infinite(as long you have disk space left) amount of trace data locally and analyze it at any later point in time.

If you are looking to export code coverage and profiling reports then this is alreay possible.
Simply right click inside the Code Coverage window and press export.

Best regards,
Nino

schmalr · Dec 17th 2018, 4:25pm

For now, I am tracing with a J-Trace Pro (Cortex M) and the Cortex M Reference Board included with it. I am simply exploring what we will be able to do (and how) in the future when we have our custom ASIC (which is under development).

We will (eventually) be performing automated overnight testing. I would like to be able to record trace data on the fly, triggering a halt of the trace when an anomaly is detected. We would then perform post-mortem analysis in the debugger on the anomaly using the recorded trace.

Russ

SEGGER - Nino · Dec 18th 2018, 9:41am

Hello Russ,

schmalr wrote:

I am simply exploring what we will be able to do (and how) in the future when we have our custom ASIC (which is under development).

Depending on how your devices implements the trace pins you will most likely need to enable and configure them in some way.
This can be done in a JLinkScript: wiki.segger.com/How_to_configu…t_files_to_enable_tracing

schmalr wrote:

We will (eventually) be performing automated overnight testing. I would like to be able to record trace data on the fly, triggering a halt of the trace when an anomaly is detected. We would then perform post-mortem analysis in the debugger on the anomaly using the recorded trace.

Automated testing is already possible with Ozone: wiki.segger.com/SEGGER_Ozone#Automated_test_example
If you can find a trigger condition that Ozone can catch (e.g. BP on Hardfault handler) your setup should be possible even now as Ozone will currently display up to 10 million of the last executed instruction in the instruction trace window.
For a longer time period the logging to a file for later analysis will be available in future as well.

Best regards,
Nino

schmalr · Dec 18th 2018, 4:32pm

Thanks for the info. I look forward to the future update.

Russ

SEGGER - Nino · Dec 18th 2018, 5:16pm

Hello Russ,

To stay up to date with our releases we suggest subscribing to our update notification list:
segger.com/notification/subscribe.php?prodid=178,7
This thread will be closed now.

Best regards,
Nino

[SOLVED] Post-Mortem Debug Using Recorded Trace Data

[SOLVED] Post-Mortem Debug Using Recorded Trace Data

schmalr wrote:

schmalr wrote:

Share

Tags