Re: Automated report: NetBSD-current/i386 test failure

[Andreas Gustafsson <gson%gson.org@localhost>]

The NetBSD Test Fixture wrote:
> This is an automatically generated notice of new failures of the
> NetBSD test suite.
> 
> The newly failing test cases are:
> 
>     usr.bin/c++/t_asan_poison:poison
>     usr.bin/c++/t_asan_poison:poison_pic
>     usr.bin/cc/t_asan_poison:poison
>     usr.bin/cc/t_asan_poison:poison_pic

At first glance this looked like a duplicate report, but I now think
what actually happened was that these four test cases were fixed and
then quickly broken again, and the second report incorrectly identified
the commits leading to the first failure rather than the second,
because when a test passes just a few times and then fails again,
it's hard to tell if it's because it was fixed and re-broken or just
unreliable.

The correct set of commits for the second failure is:

  2018.12.28.19.54.36 thorpej src/tests/rump/kernspace/kernspace.h 1.8
  2018.12.28.19.54.36 thorpej src/tests/rump/kernspace/threadpool.c 1.4
  2018.12.28.19.54.36 thorpej src/tests/rump/rumpkern/t_threadpool.c 1.2
  2018.12.28.20.12.35 christos src/lib/csu/common/crt0-common.c 1.23
  2018.12.28.20.43.25 christos src/external/gpl3/binutils/dist/ld/configure.tgt 1.31

-- 
Andreas Gustafsson, gson%gson.org@localhost