December 13, 2017
A test program I added to core/thread.d as a part of a pull request fails with only the 32-bit Darwin (OSX) tests:

  https://github.com/dlang/druntime/pull/1989

Here is the bug:

  https://issues.dlang.org/show_bug.cgi?id=18063

I was able to reproduce the issue on a MacBook Pro. Unfortunately, gdb backtrace is not helpful as it looks like stack corruption, far from the source of the issue. malloc says

  "malloc: *** error for object 0x10000000: pointer being freed was not allocated"


https://auto-tester.puremagic.com/show-run.ghtml?projectid=1&runid=2940199&dataid=20802787&isPull=true

Is OSX too-different a Posix platform that my test program is wrong? What are the differences that fail just the 32-bit test?

Is my bug fix for core/thread.d is incorrect or incomplete? If incomplete, are the failures unrelated to my changes? But then, why just the 32-bit builds?

Thank you,
Ali