Skip to contentSkip to conversation

locking/rwsem: Take read lock immediate if empty queue with no writer

PATCH Superseded Waiman Long 2 participants 3 messages Jul 10, 2018 – Jul 13, 2018 Locking primitives

v1 of 3 Superseded Latest is v3 →

lore
Share & tools
lore

Waiman Long

author reviewer
#0 lore raw

PATCH 1/1 locking/rwsem: Take read lock immediate if empty queue with no writer

It was found that a constant stream of readers might cause the count to go negative most of the time after an initial trigger by a writer even if no writer was present afterward. As a result, most of the readers would have to go through the slowpath reducing their performance.

To avoid that from happening, an additional check is added to detect the special case that the reader in the critical section is the only one in the wait queue and no writer is present. When that happens, it can just have the lock and return immediately without further action. Other incoming readers won't see a waiter is present and be forced into the slowpath.

After the list_empty() calls, the CPU should have the lock cacheline anyway, so an additional semaphore count check shouldn't have any performance impact.

---
 kernel/locking/rwsem-xadd.c | 16 +++++++++++++++-
 1 file changed, 15 insertions(+), 1 deletion(-)
kernel/locking/rwsem-xadd.c +15 -1
index 3064c50..ef8a5f3 100644
@@ -233,8 +233,22 @@ static void __rwsem_mark_wake(struct rw_semaphore *sem,
 	waiter.type = RWSEM_WAITING_FOR_READ;
 
 	raw_spin_lock_irq(&sem->wait_lock);
-	if (list_empty(&sem->wait_list))
+	if (list_empty(&sem->wait_list)) {
+		/*
+		 * In the unlikely event that the task is the only one in
+		 * the wait queue and a writer isn't present, it can have
+		 * the lock and return immediately without going through
+		 * the remaining slowpath code.
+		 *
+		 * Count won't be 0, but allowing it will probably generate
+		 * better code.
+		 */
+		if (unlikely(atomic_long_read(&sem->count) >= 0)) {
+			raw_spin_unlock_irq(&sem->wait_lock);
+			return sem;
+		}
 		adjustment += RWSEM_WAITING_BIAS;
+	}
 	list_add_tail(&waiter.list, &sem->wait_list);
 
 	/* we're now waiting on the lock, but no longer actively locking */
Signature · 1 line
1.8.3.1
#1 lore raw
Quoted 12 lines · Waiman Long

It was found that a constant stream of readers might cause the count to go negative most of the time after an initial trigger by a writer even if no writer was present afterward. As a result, most of the readers would have to go through the slowpath reducing their performance.

To avoid that from happening, an additional check is added to detect the special case that the reader in the critical section is the only one in the wait queue and no writer is present. When that happens, it can just have the lock and return immediately without further action. Other incoming readers won't see a waiter is present and be forced into the slowpath.

Quoted 8 lines

After the list_empty() calls, the CPU should have the lock cacheline anyway, so an additional semaphore count check shouldn't have any performance impact.

---
 kernel/locking/rwsem-xadd.c | 16 +++++++++++++++-
 1 file changed, 15 insertions(+), 1 deletion(-)

This looks ok to me, but it would be nice to include some performance figures in the commit log. Do you have any? Phrases such as "shouldn't have any performance impact" and "probably generate better code" don't fill me with good feelings ;)

Will

Quoted 31 lines
kernel/locking/rwsem-xadd.c +15 -1
index 3064c50..ef8a5f3 100644
@@ -233,8 +233,22 @@ static void __rwsem_mark_wake(struct rw_semaphore *sem,
 	waiter.type = RWSEM_WAITING_FOR_READ;
 
 	raw_spin_lock_irq(&sem->wait_lock);
-	if (list_empty(&sem->wait_list))
+	if (list_empty(&sem->wait_list)) {
+		/*
+		 * In the unlikely event that the task is the only one in
+		 * the wait queue and a writer isn't present, it can have
+		 * the lock and return immediately without going through
+		 * the remaining slowpath code.
+		 *
+		 * Count won't be 0, but allowing it will probably generate
+		 * better code.
+		 */
+		if (unlikely(atomic_long_read(&sem->count) >= 0)) {
+			raw_spin_unlock_irq(&sem->wait_lock);
+			return sem;
+		}
 		adjustment += RWSEM_WAITING_BIAS;
+	}
 	list_add_tail(&waiter.list, &sem->wait_list);
 
 	/* we're now waiting on the lock, but no longer actively locking */
Signature · 1 line
1.8.3.1
author reviewer
#2 lore raw
Quoted 21 lines · Will Deacon
Quoted 12 lines · Waiman Long

It was found that a constant stream of readers might cause the count to go negative most of the time after an initial trigger by a writer even if no writer was present afterward. As a result, most of the readers would have to go through the slowpath reducing their performance.

To avoid that from happening, an additional check is added to detect the special case that the reader in the critical section is the only one in the wait queue and no writer is present. When that happens, it can just have the lock and return immediately without further action. Other incoming readers won't see a waiter is present and be forced into the slowpath.

Quoted 8 lines

After the list_empty() calls, the CPU should have the lock cacheline anyway, so an additional semaphore count check shouldn't have any performance impact.

---
 kernel/locking/rwsem-xadd.c | 16 +++++++++++++++-
 1 file changed, 15 insertions(+), 1 deletion(-)
Quoted 6 lines

This looks ok to me, but it would be nice to include some performance figures in the commit log. Do you have any? Phrases such as "shouldn't have any performance impact" and "probably generate better code" don't fill me with good feelings ;)

Will

I just got the customer testing results back. I had posted a v2 patch with the customer data. I also remove wordings that may cause confusion.

Cheers,
Longman