[CPUFREQ] remove rwsem lock from CPUFREQ_GOV_STOP call (second call site)
Mathieu Desnoyers [Mon, 8 Jun 2009 17:17:31 +0000 (13:17 -0400)]
remove rwsem lock from CPUFREQ_GOV_STOP call (second call site)

commit 42a06f2166f2f6f7bf04f32b4e823eacdceafdc9

Missed a call site for CPUFREQ_GOV_STOP to remove the rwlock taken around the
teardown. To make a long story short, the rwlock write-lock causes a circular
dependency with cancel_delayed_work_sync(), because the timer handler takes the
read lock.

Note that all callers to __cpufreq_set_policy are taking the rwsem. All sysfs
callers (writers) hold the write rwsem at the earliest sysfs calling stage.

However, the rwlock write-lock is not needed upon governor stop.

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@polymtl.ca>
Acked-by: Venkatesh Pallipadi <venkatesh.pallipadi@intel.com>
CC: rjw@sisk.pl
CC: mingo@elte.hu
CC: Shaohua Li <shaohua.li@intel.com>
CC: Pekka Enberg <penberg@cs.helsinki.fi>
CC: Dave Young <hidave.darkstar@gmail.com>
CC: "Rafael J. Wysocki" <rjw@sisk.pl>
CC: Rusty Russell <rusty@rustcorp.com.au>
CC: trenn@suse.de
CC: sven.wegener@stealer.net
CC: cpufreq@vger.kernel.org
Signed-off-by: Dave Jones <davej@redhat.com>


index 4da2844..3938c78 100644 (file)
@@ -61,6 +61,8 @@ static DEFINE_SPINLOCK(cpufreq_driver_lock);
  *   are concerned with are online after they get the lock.
  * - Governor routines that can be called in cpufreq hotplug path should not
  *   take this sem as top level hotplug notifier handler takes this.
+ * - Lock should not be held across
+ *     __cpufreq_governor(data, CPUFREQ_GOV_STOP);
 static DEFINE_PER_CPU(int, policy_cpu);
 static DEFINE_PER_CPU(struct rw_semaphore, cpu_policy_rwsem);
@@ -1707,8 +1709,17 @@ static int __cpufreq_set_policy(struct cpufreq_policy *data,
                        dprintk("governor switch\n");
                        /* end old governor */
-                       if (data->governor)
+                       if (data->governor) {
+                               /*
+                                * Need to release the rwsem around governor
+                                * stop due to lock dependency between
+                                * cancel_delayed_work_sync and the read lock
+                                * taken in the delayed work handler.
+                                */
+                               unlock_policy_rwsem_write(data->cpu);
                                __cpufreq_governor(data, CPUFREQ_GOV_STOP);
+                               lock_policy_rwsem_write(data->cpu);
+                       }
                        /* start new governor */
                        data->governor = policy->governor;