GNOME Bugzilla – Bug 793353
make check failed at random timing with Segfaults or BUS error
Last modified: 2018-11-03 10:48:33 UTC
On armhf, make check in orc fails at random timing. (occasionally pass all tests) [Environment] - HW: Raspberry Pi3 - OS: Linux raspberrypi 4.9.79-v7+ #1091 SMP Tue Feb 6 13:18:45 GMT 2018 armv7l GNU/Linux (Raspbian Stretch) - Compiler: Both gcc6.3.0(default) and gcc4.9.3-14 are tried; the results were same. - Builder: meson and autoconf are tried; the results were same. - Code Tag: orc-0.4.26 to orc-0.4.28 are tried; the results were same. I have located the code of segfaults/bus error while running benchmorc. The errors always happen at the following code point. However, what test item fails depends on individual executions and looks like at random. orcexecutor.c: ---------------------------------------- void orc_executor_run (OrcExecutor *ex) { void (*func) (OrcExecutor *); if (ex->program) { func = ex->program->code_exec; } else { OrcCode *code = (OrcCode *)ex->arrays[ORC_VAR_A2]; func = code->exec; } if (func) { func (ex); ****** <--- BAD THINGS WILL HAPPEN HERE ***** /* ORC_ERROR("counters %d %d %d", ex->counter1, ex->counter2, ex->counter3); */ } else { orc_executor_emulate (ex); } } ---------------------------------------- Due to the failure mode, does the generated code have memory alignment issues? I didn't observe the same issue on Raspbian Jessie. So, the kernel and glibc behaviors might be related IMHO.
-- GitLab Migration Automatic Message -- This bug has been migrated to freedesktop.org's GitLab instance and has been closed from further activity. You can subscribe and participate further through the new bug through this link to our GitLab instance: https://gitlab.freedesktop.org/gstreamer/orc/issues/14.