5 Commits

Author	SHA1	Message	Date
ad	f87898b1c5	Signal masks are thread private, and there is no guarantee of the order in ... which the two signal handlers in this test will execute, at least not beyond SIGUSR1 being caught first, and SIGUSR2 being caught after that. One way that it can fail erroneously is: 2 kill(getpid(), SIGUSR1) 2 Receive signal SIGUSR1 2 kill(getpid(), SIGUSR2) 1 Receive signal SIGUSR2 1 Exit handler 2 Exit handler So if the handlers appear not to have run in sequence, at least verify that they were run by different threads.	2006-12-26 20:29:03 +00:00
drochner	4b58be0197	rely a bit less on undocumented behaviour: we don't know which thread ... the signal is delivered to, and we can't be sure it is delivered synchonously (without that sleep(), the test fails also on Linux)	2005-08-29 18:49:14 +00:00
he	dc60e3b6f4	sig_atomic_t isn't necessarily compatible with the %d printf format, ... so cast to int before printing. The value appears to be either 1 or 2, so no information should be lost this way.	2005-03-05 14:34:29 +00:00
nathanw	5394d17172	Add a second flag to indicate whether handler2 ran at all.	2005-02-26 20:12:58 +00:00
nathanw	17b913020a	Test for proper sigmask handling in a signal handler with threads running.	2005-02-26 20:08:14 +00:00