git: 9front

Info  •  Files  •  Log  •  Branches

kernel: never do sched() in unlock() when not in "Running" state.

Some callers of unlock change the process state
(such as qlock()) which means they are committed
to calling sched() anyway, so no ned to call it
internally.

--- a/sys/src/9/port/taslock.c

+++ b/sys/src/9/port/taslock.c

@@ -204,13 +204,13 @@

 	coherence();

 	l->key = 0;

-	if(up && --up->nlocks == 0 && up->delaysched && islo()){

-		/*

-		 * Call sched if the need arose while locks were held

-		 * But, don't do it from interrupt routines, hence the islo() test

-		 */

+	/*

+	 * Call sched if the need arose while locks were held

+	 * But, don't do it from interrupt routines, hence the islo() test

+	 */

+	if(up && --up->nlocks == 0)

+	if(up->state == Running && up->delaysched && islo())

 		sched();

-	}

 }

 uintptr ilockpcs[0x100] = { [0xff] = 1 };

-- 

⑨