From: Michel Lespinasse on
Add down_read_unfair() API.

This is similar to down_read() with the following changes:
- when the rwsem is read owned with queued writers, down_read_unfair() callers
are allowed to acquire the rwsem for read without queueing;
- when the rwsem is write owned, down_read_unfair() callers get queued in
front of threads trying to acquire the rwsem by other means.

Signed-off-by: Michel Lespinasse <walken(a)google.com>
---
include/linux/rwsem-spinlock.h | 10 +++++++++-
include/linux/rwsem.h | 10 ++++++++++
kernel/rwsem.c | 17 +++++++++++++++++
lib/rwsem-spinlock.c | 10 +++++++---
4 files changed, 43 insertions(+), 4 deletions(-)

diff --git a/include/linux/rwsem-spinlock.h b/include/linux/rwsem-spinlock.h
index bdfcc25..dc849d9 100644
--- a/include/linux/rwsem-spinlock.h
+++ b/include/linux/rwsem-spinlock.h
@@ -60,7 +60,15 @@ do { \
__init_rwsem((sem), #sem, &__key); \
} while (0)

-extern void __down_read(struct rw_semaphore *sem);
+#define __HAVE_DOWN_READ_UNFAIR
+
+static inline void __down_read(struct rw_semaphore *sem)
+ { __down_read_internal(sem, 0); }
+
+static inline void __sched __down_read_unfair(struct rw_semaphore *sem)
+ { __down_read_internal(sem, 1); }
+
+extern void __down_read_internal(struct rw_semaphore *sem, int unfair);
extern int __down_read_trylock(struct rw_semaphore *sem);
extern void __down_write(struct rw_semaphore *sem);
extern void __down_write_nested(struct rw_semaphore *sem, int subclass);
diff --git a/include/linux/rwsem.h b/include/linux/rwsem.h
index efd348f..0d3310b 100644
--- a/include/linux/rwsem.h
+++ b/include/linux/rwsem.h
@@ -28,6 +28,16 @@ struct rw_semaphore;
extern void down_read(struct rw_semaphore *sem);

/*
+ * lock for reading - skip waiting threads
+ */
+#ifdef __HAVE_DOWN_READ_UNFAIR
+extern void down_read_unfair(struct rw_semaphore *sem);
+#else
+static inline void down_read_unfair(struct rw_semaphore *sem)
+ { down_read(sem); }
+#endif
+
+/*
* trylock for reading -- returns 1 if successful, 0 if contention
*/
extern int down_read_trylock(struct rw_semaphore *sem);
diff --git a/kernel/rwsem.c b/kernel/rwsem.c
index cae050b..d7b424b 100644
--- a/kernel/rwsem.c
+++ b/kernel/rwsem.c
@@ -26,6 +26,23 @@ void __sched down_read(struct rw_semaphore *sem)

EXPORT_SYMBOL(down_read);

+#ifdef __HAVE_DOWN_READ_UNFAIR
+
+/*
+ * lock for reading - skip waiting threads
+ */
+void __sched down_read_unfair(struct rw_semaphore *sem)
+{
+ might_sleep();
+ rwsem_acquire_read(&sem->dep_map, 0, 0, _RET_IP_);
+
+ LOCK_CONTENDED(sem, __down_read_trylock, __down_read_unfair);
+}
+
+EXPORT_SYMBOL(down_read_unfair);
+
+#endif
+
/*
* trylock for reading -- returns 1 if successful, 0 if contention
*/
diff --git a/lib/rwsem-spinlock.c b/lib/rwsem-spinlock.c
index ffc9fc7..b2fd5fb 100644
--- a/lib/rwsem-spinlock.c
+++ b/lib/rwsem-spinlock.c
@@ -139,7 +139,7 @@ __rwsem_wake_one_writer(struct rw_semaphore *sem)
/*
* get a read lock on the semaphore
*/
-void __sched __down_read(struct rw_semaphore *sem)
+void __sched __down_read_internal(struct rw_semaphore *sem, int unfair)
{
struct rwsem_waiter waiter;
struct task_struct *tsk;
@@ -147,7 +147,7 @@ void __sched __down_read(struct rw_semaphore *sem)

spin_lock_irqsave(&sem->wait_lock, flags);

- if (sem->activity >= 0 && list_empty(&sem->wait_list)) {
+ if (sem->activity >= 0 && (unfair || list_empty(&sem->wait_list))) {
/* granted */
sem->activity++;
spin_unlock_irqrestore(&sem->wait_lock, flags);
@@ -162,7 +162,11 @@ void __sched __down_read(struct rw_semaphore *sem)
waiter.flags = RWSEM_WAITING_FOR_READ;
get_task_struct(tsk);

- list_add_tail(&waiter.list, &sem->wait_list);
+ if (unfair) {
+ list_add(&waiter.list, &sem->wait_list);
+ } else {
+ list_add_tail(&waiter.list, &sem->wait_list);
+ }

/* we don't need to touch the semaphore struct anymore */
spin_unlock_irqrestore(&sem->wait_lock, flags);
--
1.7.0.1

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo(a)vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/