#2 unhandled signal error when life expires

[Danny Faught]

Subtest 2 of the "simple" test fails intermittently
(maybe about half the time). Subtest 1 also hits a rare
failure, perhaps from the same cause. Here's a command
line to reproduce it:
stress_driver -life 3s -log foo -- /bin/true

When the failure occurs, stress_driver returns 1 and
prints something like this in the log:

[06/27/103 23:21:30] starting pid 22688, user=dfaught,
timeout=0s (1 tests running):
/home/dfaught/sd/t/sdtrue
[06/27/103 23:21:30] life expired, attempting to kill tests
[06/27/103 23:21:50] pid 22688 died from signal 9 (0
tests running)
[06/27/103 23:21:50] FAIL - error statistics for 46
total tests:
[06/27/103 23:21:50] 1 died by an unhandled signal
[06/27/103 23:21:50] 0 exited with a non-zero status
[06/27/103 23:21:50] 0 never died after a KILL
signal
[06/27/103 23:21:50] 0 internal errors

Looks like it might be a race condition between
exit_event and end_event. A log with debuggin enabled
is attached.

[Danny Faught]

log file for failed run

[Danny Faught]

Logged In: YES
user_id=576825

Subtest 1 of the lifetime test also hits this problem.