[igt-dev] [PATCH i-g-t 7/9] lib: Make sure we leak no child processes

[Daniel Vetter]

There's a lot more ways to leak children than igt_fork, some even
handrolled. So check for that. Also have a nice littel testcase for
that too.

v2: Don't hang if there's a leaked child process (Chris). Has the
added benefit that my library unit test also gets faster!

v3: Rebase.

Cc: Chris Wilson <chris at chris-wilson.co.uk>
Reviewed-by: Chris Wilson <chris at chris-wilson.co.uk>
Signed-off-by: Daniel Vetter <daniel.vetter at intel.com>
---
 lib/igt_core.c       |  4 ++++
 lib/tests/igt_fork.c | 18 ++++++++++++++++++
 2 files changed, 22 insertions(+)

diff --git a/lib/igt_core.c b/lib/igt_core.c
index 607b34cd0a25..425b96ff15b4 100644
--- a/lib/igt_core.c
+++ b/lib/igt_core.c
@@ -1492,6 +1492,8 @@ void __igt_fail_assert(const char *domain, const char *file, const int line,
  */
 void igt_exit(void)
 {
+	int tmp;
+
 	igt_exit_called = true;
 
 	if (igt_key_file)
@@ -1527,6 +1529,8 @@ void igt_exit(void)
 		kill(test_children[c], SIGKILL);
 	assert(!num_test_children);
 
+	assert(waitpid(-1, &tmp, WNOHANG) == -1 && errno == ECHILD);
+
 	if (!test_with_subtests) {
 		struct timespec now;
 		const char *result;
diff --git a/lib/tests/igt_fork.c b/lib/tests/igt_fork.c
index 1c10a8014e7c..fa5bb7701c09 100644
--- a/lib/tests/igt_fork.c
+++ b/lib/tests/igt_fork.c
@@ -68,6 +68,20 @@ static void igt_fork_leak(void)
 	}
 }
 
+static void plain_fork_leak(void)
+{
+	int pid;
+
+	switch (pid = fork()) {
+	case -1:
+		internal_assert(0);
+	case 0:
+		sleep(1);
+	default:
+		exit(0);
+	}
+}
+
 static void igt_fork_timeout_leak(void)
 {
 	igt_fork(i, 1) {
@@ -119,4 +133,8 @@ int main(int argc, char **argv)
 	/* check that igt_waitchildren_timeout cleans up*/
 	ret = do_fork(igt_fork_timeout_leak);
 	internal_assert(WEXITSTATUS(ret) == SIGKILL + 128);
+
+	/* check that any other process leaks are caught*/
+	ret = do_fork(plain_fork_leak);
+	internal_assert(WTERMSIG(ret) == SIGABRT);
 }
-- 
2.20.1