Issues with Merging coverage with RTEMS Tester.

Cillian O'Donnell cpodonnell8 at gmail.com
Thu Mar 15 10:01:54 UTC 2018


Alright so last time we looked at this the major issues were:

Covoars reliance on external tools that would cause portability issues.
Those were nm (which I got rid of but the fix is still only on my github,
I'll submit the patch when everything's rebased with current rtems-tools
master and I can test it again)

addr2line piped to dos2unix to find the source lines matching the objdump
assembly lines. Chris might there be some rtemstoolkit sorcery to get rid
of this dependency.

There was a number of temp files created in the process that needed to be
cleaned up afterwards. I got rid of most of these with
rld::process::tempfile, however there was still the trace files generated
by couverture when each test is run, that are then used later on by covoar.
There being handled at the moment with a try finally statement in python
which doesn't work for SIGTERM. These are the files ending in .cov

These were the major blockers last time.

At the time it would've been nice to update the config files from .mc to
.INI and it now has become a necessity and probably the first thing that
needs to be done as everything else is using INI now. I'd say this is the
only thing that needs to be fixed to rebase to the current master.

If you want to have a look at the python stuff it would be:

https://github.com/cillianodonnell/rtems-tools/blob/coverage-merge/tester/rt/coverage.py

and the changes to rtems-test

https://github.com/cillianodonnell/rtems-tools/blob/coverage-merge/tester/rt/test.py
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.rtems.org/pipermail/devel/attachments/20180315/6271dc3c/attachment-0001.html>


More information about the devel mailing list