Merge branch 'master' of master.kernel.org:/pub/scm/linux/kernel/git/davem/net-2.6

Conflicts:
	arch/microblaze/include/asm/socket.h

1 files changed, 18 insertions, 25 deletions

diff --git a/drivers/cpufreq/cpufreq_conservative.c b/drivers/cpufreq/cpufreq_conservative.c
index 7fc58af748b..bdea7e2f94b 100644
--- a/drivers/cpufreq/cpufreq_conservative.c
+++ b/drivers/cpufreq/cpufreq_conservative.c
@@ -64,21 +64,20 @@ struct cpu_dbs_info_s {
 	unsigned int requested_freq;
 	int cpu;
 	unsigned int enable:1;
+	/*
+	 * percpu mutex that serializes governor limit change with
+	 * do_dbs_timer invocation. We do not want do_dbs_timer to run
+	 * when user is changing the governor or limits.
+	 */
+	struct mutex timer_mutex;
 };
 static DEFINE_PER_CPU(struct cpu_dbs_info_s, cpu_dbs_info);
 
 static unsigned int dbs_enable;	/* number of CPUs using this policy */
 
 /*
- * DEADLOCK ALERT! There is a ordering requirement between cpu_hotplug
- * lock and dbs_mutex. cpu_hotplug lock should always be held before
- * dbs_mutex. If any function that can potentially take cpu_hotplug lock
- * (like __cpufreq_driver_target()) is being called with dbs_mutex taken, then
- * cpu_hotplug lock should be taken before that. Note that cpu_hotplug lock
- * is recursive for the same process. -Venki
- * DEADLOCK ALERT! (2) : do_dbs_timer() must not take the dbs_mutex, because it
- * would deadlock with cancel_delayed_work_sync(), which is needed for proper
- * raceless workqueue teardown.
+ * dbs_mutex protects data in dbs_tuners_ins from concurrent changes on
+ * different CPUs. It protects dbs_enable in governor start/stop.
  */
 static DEFINE_MUTEX(dbs_mutex);
 
@@ -488,18 +487,12 @@ static void do_dbs_timer(struct work_struct *work)
 
 	delay -= jiffies % delay;
 
-	if (lock_policy_rwsem_write(cpu) < 0)
-		return;
-
-	if (!dbs_info->enable) {
-		unlock_policy_rwsem_write(cpu);
-		return;
-	}
+	mutex_lock(&dbs_info->timer_mutex);
 
 	dbs_check_cpu(dbs_info);
 
 	queue_delayed_work_on(cpu, kconservative_wq, &dbs_info->work, delay);
-	unlock_policy_rwsem_write(cpu);
+	mutex_unlock(&dbs_info->timer_mutex);
 }
 
 static inline void dbs_timer_init(struct cpu_dbs_info_s *dbs_info)
@@ -535,9 +528,6 @@ static int cpufreq_governor_dbs(struct cpufreq_policy *policy,
 		if ((!cpu_online(cpu)) || (!policy->cur))
 			return -EINVAL;
 
-		if (this_dbs_info->enable) /* Already enabled */
-			break;
-
 		mutex_lock(&dbs_mutex);
 
 		rc = sysfs_create_group(&policy->kobj, &dbs_attr_group);
@@ -561,6 +551,7 @@ static int cpufreq_governor_dbs(struct cpufreq_policy *policy,
 		this_dbs_info->down_skip = 0;
 		this_dbs_info->requested_freq = policy->cur;
 
+		mutex_init(&this_dbs_info->timer_mutex);
 		dbs_enable++;
 		/*
 		 * Start the timerschedule work, when this governor
@@ -590,17 +581,19 @@ static int cpufreq_governor_dbs(struct cpufreq_policy *policy,
 					&dbs_cpufreq_notifier_block,
 					CPUFREQ_TRANSITION_NOTIFIER);
 		}
-		dbs_timer_init(this_dbs_info);
-
 		mutex_unlock(&dbs_mutex);
 
+		dbs_timer_init(this_dbs_info);
+
 		break;
 
 	case CPUFREQ_GOV_STOP:
-		mutex_lock(&dbs_mutex);
 		dbs_timer_exit(this_dbs_info);
+
+		mutex_lock(&dbs_mutex);
 		sysfs_remove_group(&policy->kobj, &dbs_attr_group);
 		dbs_enable--;
+		mutex_destroy(&this_dbs_info->timer_mutex);
 
 		/*
 		 * Stop the timerschedule work, when this governor
@@ -616,7 +609,7 @@ static int cpufreq_governor_dbs(struct cpufreq_policy *policy,
 		break;
 
 	case CPUFREQ_GOV_LIMITS:
-		mutex_lock(&dbs_mutex);
+		mutex_lock(&this_dbs_info->timer_mutex);
 		if (policy->max < this_dbs_info->cur_policy->cur)
 			__cpufreq_driver_target(
 					this_dbs_info->cur_policy,
@@ -625,7 +618,7 @@ static int cpufreq_governor_dbs(struct cpufreq_policy *policy,
 			__cpufreq_driver_target(
 					this_dbs_info->cur_policy,
 					policy->min, CPUFREQ_RELATION_L);
-		mutex_unlock(&dbs_mutex);
+		mutex_unlock(&this_dbs_info->timer_mutex);
 
 		break;
 	}