From: Daniel Drake <d.drake@mmm.com>

To avoid race conditions, sometimes runscript exits early with a "foo is not
under our control" message. Currently we interpret that as failure in the
start_done handler. This change refines the failure handling to purely failure
conditions.

I confirmed that this only triggers once for affected services.

Index: splashutils-gentoo-1.0.3/splash.c
===================================================================
--- splashutils-gentoo-1.0.3.orig/splash.c
+++ splashutils-gentoo-1.0.3/splash.c
@@ -753,8 +753,8 @@ do_start:
 		if (splash_check_daemon(&pid_daemon, false))
 			return -1;
 
-		if (rc_service_state(name, rc_service_started) ||
-		    rc_service_state(name, rc_service_inactive)) {
+		if (!rc_service_state(name, rc_service_failed) &&
+		    !rc_service_state(name, rc_service_stopped)) {
 			bool gpm = false;
 
 			if (!strcmp(name, "gpm")) {