Some stress tests create both the signal helper and a lot of competing
processes. In these tests, the parent is just waiting upon the children,
and the intention is not to keep waking up the waiting parent, but to
keep interrupting the children (as we hope to trigger races in our
kernel code). kill(-pid) sends the signal to all members of the process
group, not just the target pid.
We also switch from using SIGUSR1 to SIGCONT to paper over a race
condition when forking children that saw the default signal action being
run (and thus killing the child).
Signed-off-by: Chris Wilson <chris@xxxxxxxxxxxxxxxxxx>
---
lib/igt_aux.c | 10 ++++++----
1 file changed, 6 insertions(+), 4 deletions(-)
diff --git a/lib/igt_aux.c b/lib/igt_aux.c
index 4d08d68..f6b5792 100644
--- a/lib/igt_aux.c
+++ b/lib/igt_aux.c
@@ -79,7 +79,7 @@ static void __attribute__((noreturn)) signal_helper_process(pid_t pid)
/* Interrupt the parent process at 500Hz, just to be annoying */
while (1) {
usleep(1000 * 1000 / 500);
- if (kill(pid, SIGUSR1)) /* Parent has died, so must we. */
+ if (kill(pid, SIGCONT)) /* Parent has died, so must we. */
exit(0);
}
}
@@ -93,7 +93,7 @@ static void sig_handler(int i)
* igt_fork_signal_helper:
*
* Fork a child process using #igt_fork_helper to interrupt the parent process
- * with a SIGUSR1 signal at regular quick intervals. The corresponding dummy
+ * with a SIGCONT signal at regular quick intervals. The corresponding dummy
* signal handler is installed in the parent process.
*
* This is useful to exercise ioctl error paths, at least where those can be
@@ -108,10 +108,12 @@ void igt_fork_signal_helper(void)
if (igt_only_list_subtests())
return;
- signal(SIGUSR1, sig_handler);
+ signal(SIGCONT, sig_handler);
+ setpgrp(); /* define a new process group for ourselves */
igt_fork_helper(&signal_helper) {
- signal_helper_process(getppid());
+ signal(SIGCONT, SIG_IGN);
+ signal_helper_process(-getppid());
}
}
--
2.7.0.rc3
_______________________________________________
Intel-gfx mailing list
Intel-gfx@xxxxxxxxxxxxxxxxxxxxx
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
