aboutsummaryrefslogtreecommitdiff
diff options
context:
space:
mode:
authorPaul E. McKenney <paulmck@linux.vnet.ibm.com>2012-06-19 11:58:27 -0700
committerPaul E. McKenney <paulmck@linux.vnet.ibm.com>2012-07-02 12:34:25 -0700
commit95f0c1de3e6ed4383cc4b5f52ce4ecfb21026b49 (patch)
tree8099a405c03d9caabbf8f1752346e0aa0e30ee28
parent1c17e4d4437d8045a596d9f06c1558dc09e2b372 (diff)
rcu: Disable preemption in rcu_blocking_is_gp()
It is time to optimize CONFIG_TREE_PREEMPT_RCU's synchronize_rcu() for uniprocessor optimization, which means that rcu_blocking_is_gp() can no longer rely on RCU read-side critical sections having disabled preemption. This commit therefore disables preemption across rcu_blocking_is_gp()'s scan of the cpu_online_mask. (Updated from previous version to fix embarrassing bug spotted by Wu Fengguang.) Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
-rw-r--r--kernel/rcutree.c24
1 files changed, 6 insertions, 18 deletions
diff --git a/kernel/rcutree.c b/kernel/rcutree.c
index 70c4da7d2a9..e000a623e63 100644
--- a/kernel/rcutree.c
+++ b/kernel/rcutree.c
@@ -1981,28 +1981,16 @@ EXPORT_SYMBOL_GPL(call_rcu_bh);
* occasionally incorrectly indicate that there are multiple CPUs online
* when there was in fact only one the whole time, as this just adds
* some overhead: RCU still operates correctly.
- *
- * Of course, sampling num_online_cpus() with preemption enabled can
- * give erroneous results if there are concurrent CPU-hotplug operations.
- * For example, given a demonic sequence of preemptions in num_online_cpus()
- * and CPU-hotplug operations, there could be two or more CPUs online at
- * all times, but num_online_cpus() might well return one (or even zero).
- *
- * However, all such demonic sequences require at least one CPU-offline
- * operation. Furthermore, rcu_blocking_is_gp() giving the wrong answer
- * is only a problem if there is an RCU read-side critical section executing
- * throughout. But RCU-sched and RCU-bh read-side critical sections
- * disable either preemption or bh, which prevents a CPU from going offline.
- * Therefore, the only way that rcu_blocking_is_gp() can incorrectly return
- * that there is only one CPU when in fact there was more than one throughout
- * is when there were no RCU readers in the system. If there are no
- * RCU readers, the grace period by definition can be of zero length,
- * regardless of the number of online CPUs.
*/
static inline int rcu_blocking_is_gp(void)
{
+ int ret;
+
might_sleep(); /* Check for RCU read-side critical section. */
- return num_online_cpus() <= 1;
+ preempt_disable();
+ ret = num_online_cpus() <= 1;
+ preempt_enable();
+ return ret;
}
/**