Zero unexpected ATF test failures

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

Hi all,

I'm happy to report that today my amd64 bare metal testbed completed
an ATF test run with zero unexpected failures, produding a line of
all green in the monthly report:

  http://www.gson.org/netbsd/bugs/build/amd64-baremetal/commits-2019.01.html#2019.01.15.03.43.15

That's a first for this test setup, and pretty rare overall - the
last ones I remember seeing were on Martin's sparc64 testbed in
May 2018, and on the TNF i386 and amd64 testbeds in 2012.

Thanks to all those who have helped make this happen by fixing bugs,
and let's try to keep the positive trend going - it would be nice if
the 9.0 release could reliably show zero unexpected test failures on
at least one architecture.

We're still not quite there with the "reliably" part, because there
are still several tests that fail randomly.  Here's a list of some
test cases exhibiting frequent random failures on real amd64 hardware:

  fs/vfs/t_rmdirrace:lfs_race
  lib/libc/sys/t_posix_fadvise:posix_fadvise_reg
  fs/ffs/t_quotalimit:slimit_le_1_user
  modules/t_builtin:disable

i386 currently has some 8 test cases failing on real hardware,
fixes for these would also be appreciated:

  http://www.gson.org/netbsd/bugs/build/i386-baremetal/2019/2019.01.15.12.55.03/test.html#failed-tcs-summary

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