Comment # 33 on bug 921577 from Michel Normand

Created attachment 653799 [details]
doxygen_1810_vm61_cpu_4_threads4_gdb_bt_data.log

* The attached log file keep trace of the random SIGSEGV of doxygen
  while call in loop as detailed below.
* as already reported before such failure reported only
  when guest configured like: 4 threads/core (as per lscpu in log)
* gdb reported regs confirm reason of SIGSEGV (invalid r9 and r10 values)
  but those values seems strange versus related disassembly code.
* I would need help from there.
===
=== extract doxygen_1810_vm61_cpu_4_threads4_gdb_bt_data.log
===
[michel@vm61:~/work]
$osc co openSUSE:Factory:PowerPC doxygen-doc
$cd openSUSE:Factory:PowerPC/doxygen-doc
$vi doxygen-doc.spec <= to remove doxygen-dot-one-thread.patch (bypass
bnc#921577)
$osc build --clean
...
$target=/home/michel/buildtmp/build-root/standard-ppc64le
$for xx in dev sys proc; do sudo mount -o bind /$xx $target/$xx;done
[michel@vm61:~/work/openSUSE:Factory:PowerPC/doxygen-doc]
$osc chroot
running: sudo chroot /home/michel/buildtmp/build-root/standard-ppc64le su -
abuild
abuild@vm61:~> set -o vi
abuild@vm61:~> cd rpmbuild/BUILD/doxygen-1.8.10/
abuild@vm61:~/rpmbuild/BUILD/doxygen-1.8.10> ulimit -c unlimited
abuild@vm61:~/rpmbuild/BUILD/doxygen-1.8.10> idx=1; while test 1; do echo "===
trial $idx"; rm -rf doxygen.tag doxygen_docs; ./build/bin/doxygen
>/tmp/doxyen.log 2>&1 || break; ((idx++)); done
...
Segmentation fault (core dumped)
===