The stack traces (as with every other bsps) have some toolchain specific library and function name variations. In this case, the library parent directories are not exactly the same as for qemuppc or intel-x86-64.
----
dumping the difference line that filed when comparing apitest.bas with apitest.log:
root@128:/usr/lib64/heapscope/ptest# cat apitest.filter*
Match not found at line: 7 token: (/lib64/libc-2.20.so)
editing the apitest.bas at line 7
[ADDR] generic_start_main (/lib64/libc-2.20.so)
editing the apitest.log output file to see what was reported:
80b5b95fec generic_start_main.isra.0 (/lib64/libc-2.20.so)
which matches except for the '.isra.0' suffix after generic_start_main.
either run-ptest or run-hstest must be modified on a per arch variant (ppc64) to account for this expected value. The workbench 4 had no issues with capturing the data.
These are regression test scripts, and do not reflect the core functionality. The impact to the user is that the user will see function names towards the top of callstacks, of little interest for this problem usecase, that would be slightly different than the function names of these uninteresting functions for another bsp.
However, we want the regression tests to adjust for this and PASS not fail.
