Make error handling in parallel pg_upgrade less bogus.

tglsfdc · tglsfdc · commit 16fda4b853cd · 2018-12-16T14:51:47.000-05:00
reap_child() basically ignored the possibility of either an error in
waitpid() itself or a child process failure on signal.  We don't really
need to do more than report and crash hard, but proceeding as though
nothing is wrong is definitely Not Acceptable.  The error report for
nonzero child exit status was pretty off-point, as well.

Noted while fooling around with child-process failure detection
logic elsewhere.  It's been like this a long time, so back-patch to
all supported branches.
diff --git a/src/bin/pg_upgrade/parallel.c b/src/bin/pg_upgrade/parallel.c
@@ -290,7 +290,7 @@ reap_child(bool wait_for_child)
 {
 #ifndef WIN32
 	int			work_status;
-	int			ret;
+	pid_t		child;
 #else
 	int			thread_num;
 	DWORD		res;
@@ -300,14 +300,13 @@ reap_child(bool wait_for_child)
 		return false;
 
 #ifndef WIN32
-	ret = waitpid(-1, &work_status, wait_for_child ? 0 : WNOHANG);
-
-	/* no children or, for WNOHANG, no dead children */
-	if (ret <= 0 || !WIFEXITED(work_status))
-		return false;
-
-	if (WEXITSTATUS(work_status) != 0)
-		pg_fatal("child worker exited abnormally: %s\n", strerror(errno));
+	child = waitpid(-1, &work_status, wait_for_child ? 0 : WNOHANG);
+	if (child == (pid_t) -1)
+		pg_fatal("waitpid() failed: %s\n", strerror(errno));
+	if (child == 0)
+		return false;			/* no children, or no dead children */
+	if (work_status != 0)
+		pg_fatal("child process exited abnormally: status %d\n", work_status);
 #else
 	/* wait for one to finish */
 	thread_num = WaitForMultipleObjects(parallel_jobs, thread_handles,
