From: Peter Zijlstra <a.p.zijlstra@chello.nl>
Date: Thu, 10 Sep 2009 16:18:47 +0000 (+0200)
Subject: sched: for_each_domain() vs RCU
X-Git-Tag: v2.6.32-rc1~652^2~26
X-Git-Url: https://openfabrics.org/gitweb/?a=commitdiff_plain;h=83f54960c11a14942ab00b54c51e91906b9d8235;p=~shefty%2Frdma-dev.git

sched: for_each_domain() vs RCU

for_each_domain() uses RCU to serialize the sched_domains, except
it doesn't actually use rcu_read_lock() and instead relies on
disabling preemption -> FAIL.

XXX: audit other sched_domain code.

Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
LKML-Reference: <new-submission>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
---

diff --git a/kernel/sched_fair.c b/kernel/sched_fair.c
index eaa00014b49..43dc6d1d9e8 100644
--- a/kernel/sched_fair.c
+++ b/kernel/sched_fair.c
@@ -1331,6 +1331,7 @@ static int select_task_rq_fair(struct task_struct *p, int flag, int sync)
 		new_cpu = prev_cpu;
 	}
 
+	rcu_read_lock();
 	for_each_domain(cpu, tmp) {
 		/*
 		 * If power savings logic is enabled for a domain, see if we
@@ -1369,8 +1370,10 @@ static int select_task_rq_fair(struct task_struct *p, int flag, int sync)
 		if (want_affine && (tmp->flags & SD_WAKE_AFFINE) &&
 		    cpumask_test_cpu(prev_cpu, sched_domain_span(tmp))) {
 
-			if (wake_affine(tmp, p, sync))
-				return cpu;
+			if (wake_affine(tmp, p, sync)) {
+				new_cpu = cpu;
+				goto out;
+			}
 
 			want_affine = 0;
 		}
@@ -1416,6 +1419,8 @@ static int select_task_rq_fair(struct task_struct *p, int flag, int sync)
 		/* while loop will break here if sd == NULL */
 	}
 
+out:
+	rcu_read_unlock();
 	return new_cpu;
 }
 #endif /* CONFIG_SMP */