[patch V3 29/32] cpu/hotplug: Convert hotplug locking to percpu rwsem

From: Thomas Gleixner
Date: Wed May 24 2017 - 04:32:49 EST


There are no more (known) nested calls to get_online_cpus() and all
observed lock ordering problems have been addressed.

Replace the magic nested 'rwsem' hackery with a percpu-rwsem.

Signed-off-by: Thomas Gleixner <tglx@xxxxxxxxxxxxx>
Cc: Peter Zijlstra <peterz@xxxxxxxxxxxxx>
Cc: Sebastian Siewior <bigeasy@xxxxxxxxxxxxx>
Cc: Steven Rostedt <rostedt@xxxxxxxxxxx>

---
include/linux/cpu.h | 2
kernel/cpu.c | 107 ++++++----------------------------------------------
2 files changed, 14 insertions(+), 95 deletions(-)

Index: b/include/linux/cpu.h
===================================================================
--- a/include/linux/cpu.h
+++ b/include/linux/cpu.h
@@ -103,7 +103,7 @@ extern void cpus_write_lock(void);
extern void cpus_write_unlock(void);
extern void cpus_read_lock(void);
extern void cpus_read_unlock(void);
-static inline void lockdep_assert_cpus_held(void) { }
+extern void lockdep_assert_cpus_held(void);
extern void cpu_hotplug_disable(void);
extern void cpu_hotplug_enable(void);
void clear_tasks_mm_cpumask(int cpu);
Index: b/kernel/cpu.c
===================================================================
--- a/kernel/cpu.c
+++ b/kernel/cpu.c
@@ -27,6 +27,7 @@
#include <linux/smpboot.h>
#include <linux/relay.h>
#include <linux/slab.h>
+#include <linux/percpu-rwsem.h>

#include <trace/events/power.h>
#define CREATE_TRACE_POINTS
@@ -196,121 +197,41 @@ void cpu_maps_update_done(void)
mutex_unlock(&cpu_add_remove_lock);
}

-/* If set, cpu_up and cpu_down will return -EBUSY and do nothing.
+/*
+ * If set, cpu_up and cpu_down will return -EBUSY and do nothing.
* Should always be manipulated under cpu_add_remove_lock
*/
static int cpu_hotplug_disabled;

#ifdef CONFIG_HOTPLUG_CPU

-static struct {
- struct task_struct *active_writer;
- /* wait queue to wake up the active_writer */
- wait_queue_head_t wq;
- /* verifies that no writer will get active while readers are active */
- struct mutex lock;
- /*
- * Also blocks the new readers during
- * an ongoing cpu hotplug operation.
- */
- atomic_t refcount;
-
-#ifdef CONFIG_DEBUG_LOCK_ALLOC
- struct lockdep_map dep_map;
-#endif
-} cpu_hotplug = {
- .active_writer = NULL,
- .wq = __WAIT_QUEUE_HEAD_INITIALIZER(cpu_hotplug.wq),
- .lock = __MUTEX_INITIALIZER(cpu_hotplug.lock),
-#ifdef CONFIG_DEBUG_LOCK_ALLOC
- .dep_map = STATIC_LOCKDEP_MAP_INIT("cpu_hotplug.dep_map", &cpu_hotplug.dep_map),
-#endif
-};
-
-/* Lockdep annotations for get/put_online_cpus() and cpu_hotplug_begin/end() */
-#define cpuhp_lock_acquire_read() lock_map_acquire_read(&cpu_hotplug.dep_map)
-#define cpuhp_lock_acquire_tryread() \
- lock_map_acquire_tryread(&cpu_hotplug.dep_map)
-#define cpuhp_lock_acquire() lock_map_acquire(&cpu_hotplug.dep_map)
-#define cpuhp_lock_release() lock_map_release(&cpu_hotplug.dep_map)
-
+DEFINE_STATIC_PERCPU_RWSEM(cpu_hotplug_lock);

void cpus_read_lock(void)
{
- might_sleep();
- if (cpu_hotplug.active_writer == current)
- return;
- cpuhp_lock_acquire_read();
- mutex_lock(&cpu_hotplug.lock);
- atomic_inc(&cpu_hotplug.refcount);
- mutex_unlock(&cpu_hotplug.lock);
+ percpu_down_read(&cpu_hotplug_lock);
}
EXPORT_SYMBOL_GPL(cpus_read_lock);

void cpus_read_unlock(void)
{
- int refcount;
-
- if (cpu_hotplug.active_writer == current)
- return;
-
- refcount = atomic_dec_return(&cpu_hotplug.refcount);
- if (WARN_ON(refcount < 0)) /* try to fix things up */
- atomic_inc(&cpu_hotplug.refcount);
-
- if (refcount <= 0 && waitqueue_active(&cpu_hotplug.wq))
- wake_up(&cpu_hotplug.wq);
-
- cpuhp_lock_release();
-
+ percpu_up_read(&cpu_hotplug_lock);
}
EXPORT_SYMBOL_GPL(cpus_read_unlock);

-/*
- * This ensures that the hotplug operation can begin only when the
- * refcount goes to zero.
- *
- * Note that during a cpu-hotplug operation, the new readers, if any,
- * will be blocked by the cpu_hotplug.lock
- *
- * Since cpu_hotplug_begin() is always called after invoking
- * cpu_maps_update_begin(), we can be sure that only one writer is active.
- *
- * Note that theoretically, there is a possibility of a livelock:
- * - Refcount goes to zero, last reader wakes up the sleeping
- * writer.
- * - Last reader unlocks the cpu_hotplug.lock.
- * - A new reader arrives at this moment, bumps up the refcount.
- * - The writer acquires the cpu_hotplug.lock finds the refcount
- * non zero and goes to sleep again.
- *
- * However, this is very difficult to achieve in practice since
- * get_online_cpus() not an api which is called all that often.
- *
- */
void cpus_write_lock(void)
{
- DEFINE_WAIT(wait);
-
- cpu_hotplug.active_writer = current;
- cpuhp_lock_acquire();
-
- for (;;) {
- mutex_lock(&cpu_hotplug.lock);
- prepare_to_wait(&cpu_hotplug.wq, &wait, TASK_UNINTERRUPTIBLE);
- if (likely(!atomic_read(&cpu_hotplug.refcount)))
- break;
- mutex_unlock(&cpu_hotplug.lock);
- schedule();
- }
- finish_wait(&cpu_hotplug.wq, &wait);
+ percpu_down_write(&cpu_hotplug_lock);
}

void cpus_write_unlock(void)
{
- cpu_hotplug.active_writer = NULL;
- mutex_unlock(&cpu_hotplug.lock);
- cpuhp_lock_release();
+ percpu_up_write(&cpu_hotplug_lock);
+}
+
+void lockdep_assert_cpus_held(void)
+{
+ percpu_rwsem_assert_held(&cpu_hotplug_lock);
}

/*
@@ -344,8 +265,6 @@ void cpu_hotplug_enable(void)
EXPORT_SYMBOL_GPL(cpu_hotplug_enable);
#endif /* CONFIG_HOTPLUG_CPU */

-/* Notifier wrappers for transitioning to state machine */
-
static int bringup_wait_for_ap(unsigned int cpu)
{
struct cpuhp_cpu_state *st = per_cpu_ptr(&cpuhp_state, cpu);