1 files changed, 134 insertions, 60 deletions

diff --git a/kernel/locking/percpu-rwsem.c b/kernel/locking/percpu-rwsem.c
index 364d38a0c444..a008a1ba21a7 100644
--- a/kernel/locking/percpu-rwsem.c
+++ b/kernel/locking/percpu-rwsem.c
@@ -1,27 +1,29 @@
 // SPDX-License-Identifier: GPL-2.0-only
 #include <linux/atomic.h>
-#include <linux/rwsem.h>
 #include <linux/percpu.h>
+#include <linux/wait.h>
 #include <linux/lockdep.h>
 #include <linux/percpu-rwsem.h>
 #include <linux/rcupdate.h>
 #include <linux/sched.h>
+#include <linux/sched/task.h>
 #include <linux/errno.h>
 
-#include "rwsem.h"
-
 int __percpu_init_rwsem(struct percpu_rw_semaphore *sem,
-			const char *name, struct lock_class_key *rwsem_key)
+			const char *name, struct lock_class_key *key)
 {
 	sem->read_count = alloc_percpu(int);
 	if (unlikely(!sem->read_count))
 		return -ENOMEM;
 
-	/* ->rw_sem represents the whole percpu_rw_semaphore for lockdep */
 	rcu_sync_init(&sem->rss);
-	__init_rwsem(&sem->rw_sem, name, rwsem_key);
 	rcuwait_init(&sem->writer);
-	sem->readers_block = 0;
+	init_waitqueue_head(&sem->waiters);
+	atomic_set(&sem->block, 0);
+#ifdef CONFIG_DEBUG_LOCK_ALLOC
+	debug_check_no_locks_freed((void *)sem, sizeof(*sem));
+	lockdep_init_map(&sem->dep_map, name, key, 0);
+#endif
 	return 0;
 }
 EXPORT_SYMBOL_GPL(__percpu_init_rwsem);
@@ -41,73 +43,139 @@ void percpu_free_rwsem(struct percpu_rw_semaphore *sem)
 }
 EXPORT_SYMBOL_GPL(percpu_free_rwsem);
 
-int __percpu_down_read(struct percpu_rw_semaphore *sem, int try)
+static bool __percpu_down_read_trylock(struct percpu_rw_semaphore *sem)
 {
+	__this_cpu_inc(*sem->read_count);
+
 	/*
 	 * Due to having preemption disabled the decrement happens on
 	 * the same CPU as the increment, avoiding the
 	 * increment-on-one-CPU-and-decrement-on-another problem.
 	 *
-	 * If the reader misses the writer's assignment of readers_block, then
-	 * the writer is guaranteed to see the reader's increment.
+	 * If the reader misses the writer's assignment of sem->block, then the
+	 * writer is guaranteed to see the reader's increment.
 	 *
 	 * Conversely, any readers that increment their sem->read_count after
-	 * the writer looks are guaranteed to see the readers_block value,
-	 * which in turn means that they are guaranteed to immediately
-	 * decrement their sem->read_count, so that it doesn't matter that the
-	 * writer missed them.
+	 * the writer looks are guaranteed to see the sem->block value, which
+	 * in turn means that they are guaranteed to immediately decrement
+	 * their sem->read_count, so that it doesn't matter that the writer
+	 * missed them.
 	 */
 
 	smp_mb(); /* A matches D */
 
 	/*
-	 * If !readers_block the critical section starts here, matched by the
+	 * If !sem->block the critical section starts here, matched by the
 	 * release in percpu_up_write().
 	 */
-	if (likely(!smp_load_acquire(&sem->readers_block)))
+	if (likely(!atomic_read_acquire(&sem->block)))
+		return true;
+
+	__this_cpu_dec(*sem->read_count);
+
+	/* Prod writer to re-evaluate readers_active_check() */
+	rcuwait_wake_up(&sem->writer);
+
+	return false;
+}
+
+static inline bool __percpu_down_write_trylock(struct percpu_rw_semaphore *sem)
+{
+	if (atomic_read(&sem->block))
+		return false;
+
+	return atomic_xchg(&sem->block, 1) == 0;
+}
+
+static bool __percpu_rwsem_trylock(struct percpu_rw_semaphore *sem, bool reader)
+{
+	if (reader) {
+		bool ret;
+
+		preempt_disable();
+		ret = __percpu_down_read_trylock(sem);
+		preempt_enable();
+
+		return ret;
+	}
+	return __percpu_down_write_trylock(sem);
+}
+
+/*
+ * The return value of wait_queue_entry::func means:
+ *
+ *  <0 - error, wakeup is terminated and the error is returned
+ *   0 - no wakeup, a next waiter is tried
+ *  >0 - woken, if EXCLUSIVE, counted towards @nr_exclusive.
+ *
+ * We use EXCLUSIVE for both readers and writers to preserve FIFO order,
+ * and play games with the return value to allow waking multiple readers.
+ *
+ * Specifically, we wake readers until we've woken a single writer, or until a
+ * trylock fails.
+ */
+static int percpu_rwsem_wake_function(struct wait_queue_entry *wq_entry,
+				      unsigned int mode, int wake_flags,
+				      void *key)
+{
+	struct task_struct *p = get_task_struct(wq_entry->private);
+	bool reader = wq_entry->flags & WQ_FLAG_CUSTOM;
+	struct percpu_rw_semaphore *sem = key;
+
+	/* concurrent against percpu_down_write(), can get stolen */
+	if (!__percpu_rwsem_trylock(sem, reader))
 		return 1;
 
-	/*
-	 * Per the above comment; we still have preemption disabled and
-	 * will thus decrement on the same CPU as we incremented.
-	 */
-	__percpu_up_read(sem);
+	list_del_init(&wq_entry->entry);
+	smp_store_release(&wq_entry->private, NULL);
 
-	if (try)
-		return 0;
+	wake_up_process(p);
+	put_task_struct(p);
 
-	/*
-	 * We either call schedule() in the wait, or we'll fall through
-	 * and reschedule on the preempt_enable() in percpu_down_read().
-	 */
-	preempt_enable_no_resched();
+	return !reader; /* wake (readers until) 1 writer */
+}
+
+static void percpu_rwsem_wait(struct percpu_rw_semaphore *sem, bool reader)
+{
+	DEFINE_WAIT_FUNC(wq_entry, percpu_rwsem_wake_function);
+	bool wait;
 
+	spin_lock_irq(&sem->waiters.lock);
 	/*
-	 * Avoid lockdep for the down/up_read() we already have them.
+	 * Serialize against the wakeup in percpu_up_write(), if we fail
+	 * the trylock, the wakeup must see us on the list.
 	 */
-	__down_read(&sem->rw_sem);
-	this_cpu_inc(*sem->read_count);
-	__up_read(&sem->rw_sem);
+	wait = !__percpu_rwsem_trylock(sem, reader);
+	if (wait) {
+		wq_entry.flags |= WQ_FLAG_EXCLUSIVE | reader * WQ_FLAG_CUSTOM;
+		__add_wait_queue_entry_tail(&sem->waiters, &wq_entry);
+	}
+	spin_unlock_irq(&sem->waiters.lock);
 
-	preempt_disable();
-	return 1;
+	while (wait) {
+		set_current_state(TASK_UNINTERRUPTIBLE);
+		if (!smp_load_acquire(&wq_entry.private))
+			break;
+		schedule();
+	}
+	__set_current_state(TASK_RUNNING);
 }
-EXPORT_SYMBOL_GPL(__percpu_down_read);
 
-void __percpu_up_read(struct percpu_rw_semaphore *sem)
+bool __percpu_down_read(struct percpu_rw_semaphore *sem, bool try)
 {
-	smp_mb(); /* B matches C */
-	/*
-	 * In other words, if they see our decrement (presumably to aggregate
-	 * zero, as that is the only time it matters) they will also see our
-	 * critical section.
-	 */
-	__this_cpu_dec(*sem->read_count);
+	if (__percpu_down_read_trylock(sem))
+		return true;
 
-	/* Prod writer to recheck readers_active */
-	rcuwait_wake_up(&sem->writer);
+	if (try)
+		return false;
+
+	preempt_enable();
+	percpu_rwsem_wait(sem, /* .reader = */ true);
+	preempt_disable();
+
+	return true;
 }
-EXPORT_SYMBOL_GPL(__percpu_up_read);
+EXPORT_SYMBOL_GPL(__percpu_down_read);
 
 #define per_cpu_sum(var)						\
 ({									\
@@ -124,6 +192,8 @@ EXPORT_SYMBOL_GPL(__percpu_up_read);
  * zero.  If this sum is zero, then it is stable due to the fact that if any
  * newly arriving readers increment a given counter, they will immediately
  * decrement that same counter.
+ *
+ * Assumes sem->block is set.
  */
 static bool readers_active_check(struct percpu_rw_semaphore *sem)
 {
@@ -142,32 +212,36 @@ static bool readers_active_check(struct percpu_rw_semaphore *sem)
 
 void percpu_down_write(struct percpu_rw_semaphore *sem)
 {
+	might_sleep();
+	rwsem_acquire(&sem->dep_map, 0, 0, _RET_IP_);
+
 	/* Notify readers to take the slow path. */
 	rcu_sync_enter(&sem->rss);
 
-	down_write(&sem->rw_sem);
-
 	/*
-	 * Notify new readers to block; up until now, and thus throughout the
-	 * longish rcu_sync_enter() above, new readers could still come in.
+	 * Try set sem->block; this provides writer-writer exclusion.
+	 * Having sem->block set makes new readers block.
 	 */
-	WRITE_ONCE(sem->readers_block, 1);
+	if (!__percpu_down_write_trylock(sem))
+		percpu_rwsem_wait(sem, /* .reader = */ false);
 
-	smp_mb(); /* D matches A */
+	/* smp_mb() implied by __percpu_down_write_trylock() on success -- D matches A */
 
 	/*
-	 * If they don't see our writer of readers_block, then we are
-	 * guaranteed to see their sem->read_count increment, and therefore
-	 * will wait for them.
+	 * If they don't see our store of sem->block, then we are guaranteed to
+	 * see their sem->read_count increment, and therefore will wait for
+	 * them.
 	 */
 
-	/* Wait for all now active readers to complete. */
-	rcuwait_wait_event(&sem->writer, readers_active_check(sem));
+	/* Wait for all active readers to complete. */
+	rcuwait_wait_event(&sem->writer, readers_active_check(sem), TASK_UNINTERRUPTIBLE);
 }
 EXPORT_SYMBOL_GPL(percpu_down_write);
 
 void percpu_up_write(struct percpu_rw_semaphore *sem)
 {
+	rwsem_release(&sem->dep_map, _RET_IP_);
+
 	/*
 	 * Signal the writer is done, no fast path yet.
 	 *
@@ -178,12 +252,12 @@ void percpu_up_write(struct percpu_rw_semaphore *sem)
 	 * Therefore we force it through the slow path which guarantees an
 	 * acquire and thereby guarantees the critical section's consistency.
 	 */
-	smp_store_release(&sem->readers_block, 0);
+	atomic_set_release(&sem->block, 0);
 
 	/*
-	 * Release the write lock, this will allow readers back in the game.
+	 * Prod any pending reader/writer to make progress.
 	 */
-	up_write(&sem->rw_sem);
+	__wake_up(&sem->waiters, TASK_NORMAL, 1, sem);
 
 	/*
 	 * Once this completes (at least one RCU-sched grace period hence) the