locking/rwsem: Disable preemption while trying for rwsem lock

Gokul krishna Krishnakumar · Peter Zijlstra · commit 48dfb5d2560d · 2022-09-15T16:14:02.000+02:00
Make the region inside the rwsem_write_trylock non preemptible. We observe RT task is hogging CPU when trying to acquire rwsem lock which was acquired by a kworker task but before the rwsem owner was set. Here is the scenario: 1. CFS task (affined to a particular CPU) takes rwsem lock. 2. CFS task gets preempted by a RT task before setting owner. 3. RT task (FIFO) is trying to acquire the lock, but spinning until RT throttling happens for the lock as the lock was taken by CFS task. This patch attempts to fix the above issue by disabling preemption until owner is set for the lock. While at it also fix the issues at the places where rwsem_{set,clear}_owner() are called. This also adds lockdep annotation of preemption disable in rwsem_{set,clear}_owner() on Peter Z. suggestion. Signed-off-by: Gokul krishna Krishnakumar <quic_gokukris@quicinc.com> Signed-off-by: Mukesh Ojha <quic_mojha@quicinc.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Reviewed-by: Waiman Long <longman@redhat.com> Link: https://lore.kernel.org/r/1662661467-24203-1-git-send-email-quic_mojha@quicinc.com
diff --git a/kernel/locking/rwsem.c b/kernel/locking/rwsem.c
@@ -133,14 +133,19 @@
  * the owner value concurrently without lock. Read from owner, however,
  * may not need READ_ONCE() as long as the pointer value is only used
  * for comparison and isn't being dereferenced.
+ *
+ * Both rwsem_{set,clear}_owner() functions should be in the same
+ * preempt disable section as the atomic op that changes sem->count.
  */
 static inline void rwsem_set_owner(struct rw_semaphore *sem)
 {
+	lockdep_assert_preemption_disabled();
 	atomic_long_set(&sem->owner, (long)current);
 }
 
 static inline void rwsem_clear_owner(struct rw_semaphore *sem)
 {
+	lockdep_assert_preemption_disabled();
 	atomic_long_set(&sem->owner, 0);
 }
 
@@ -251,13 +256,16 @@ static inline bool rwsem_read_trylock(struct rw_semaphore *sem, long *cntp)
 static inline bool rwsem_write_trylock(struct rw_semaphore *sem)
 {
 	long tmp = RWSEM_UNLOCKED_VALUE;
+	bool ret = false;
 
+	preempt_disable();
 	if (atomic_long_try_cmpxchg_acquire(&sem->count, &tmp, RWSEM_WRITER_LOCKED)) {
 		rwsem_set_owner(sem);
-		return true;
+		ret = true;
 	}
 
-	return false;
+	preempt_enable();
+	return ret;
 }
 
 /*
@@ -1352,8 +1360,10 @@ static inline void __up_write(struct rw_semaphore *sem)
 	DEBUG_RWSEMS_WARN_ON((rwsem_owner(sem) != current) &&
 			    !rwsem_test_oflags(sem, RWSEM_NONSPINNABLE), sem);
 
+	preempt_disable();
 	rwsem_clear_owner(sem);
 	tmp = atomic_long_fetch_add_release(-RWSEM_WRITER_LOCKED, &sem->count);
+	preempt_enable();
 	if (unlikely(tmp & RWSEM_FLAG_WAITERS))
 		rwsem_wake(sem);
 }

Original file line number	Diff line number	Diff line change
`@@ -133,14 +133,19 @@`
`133`	`133`	`* the owner value concurrently without lock. Read from owner, however,`
`134`	`134`	`* may not need READ_ONCE() as long as the pointer value is only used`
`135`	`135`	`* for comparison and isn't being dereferenced.`
	`136`	`+ *`
	`137`	`+ * Both rwsem_{set,clear}_owner() functions should be in the same`
	`138`	`+ * preempt disable section as the atomic op that changes sem->count.`
`136`	`139`	`*/`
`137`	`140`	`static inline void rwsem_set_owner(struct rw_semaphore *sem)`
`138`	`141`	`{`
	`142`	`+ lockdep_assert_preemption_disabled();`
`139`	`143`	`atomic_long_set(&sem->owner, (long)current);`
`140`	`144`	`}`
`141`	`145`
`142`	`146`	`static inline void rwsem_clear_owner(struct rw_semaphore *sem)`
`143`	`147`	`{`
	`148`	`+ lockdep_assert_preemption_disabled();`
`144`	`149`	`atomic_long_set(&sem->owner, 0);`
`145`	`150`	`}`
`146`	`151`
`@@ -251,13 +256,16 @@ static inline bool rwsem_read_trylock(struct rw_semaphore sem, long cntp)`
`251`	`256`	`static inline bool rwsem_write_trylock(struct rw_semaphore *sem)`
`252`	`257`	`{`
`253`	`258`	`long tmp = RWSEM_UNLOCKED_VALUE;`
	`259`	`+ bool ret = false;`
`254`	`260`
	`261`	`+ preempt_disable();`
`255`	`262`	`if (atomic_long_try_cmpxchg_acquire(&sem->count, &tmp, RWSEM_WRITER_LOCKED)) {`
`256`	`263`	`rwsem_set_owner(sem);`
`257`		`- return true;`
	`264`	`+ ret = true;`
`258`	`265`	`}`
`259`	`266`
`260`		`- return false;`
	`267`	`+ preempt_enable();`
	`268`	`+ return ret;`
`261`	`269`	`}`
`262`	`270`
`263`	`271`	`/*`
`@@ -1352,8 +1360,10 @@ static inline void __up_write(struct rw_semaphore *sem)`
`1352`	`1360`	`DEBUG_RWSEMS_WARN_ON((rwsem_owner(sem) != current) &&`
`1353`	`1361`	`!rwsem_test_oflags(sem, RWSEM_NONSPINNABLE), sem);`
`1354`	`1362`
	`1363`	`+ preempt_disable();`
`1355`	`1364`	`rwsem_clear_owner(sem);`
`1356`	`1365`	`tmp = atomic_long_fetch_add_release(-RWSEM_WRITER_LOCKED, &sem->count);`
	`1366`	`+ preempt_enable();`
`1357`	`1367`	`if (unlikely(tmp & RWSEM_FLAG_WAITERS))`
`1358`	`1368`	`rwsem_wake(sem);`
`1359`	`1369`	`}`