blockdev: turn a rw semaphore into a percpu rw semaphore
This avoids cache line bouncing when many processes lock the semaphore
for read.
New percpu lock implementation
The lock consists of an array of percpu unsigned integers, a boolean
variable and a mutex.
When we take the lock for read, we enter rcu read section, check for a
"locked" variable. If it is false, we increase a percpu counter on the
current cpu and exit the rcu section. If "locked" is true, we exit the
rcu section, take the mutex and drop it (this waits until a writer
finished) and retry.
Unlocking for read just decreases percpu variable. Note that we can
unlock on a difference cpu than where we locked, in this case the
counter underflows. The sum of all percpu counters represents the number
of processes that hold the lock for read.
When we need to lock for write, we take the mutex, set "locked" variable
to true and synchronize rcu. Since RCU has been synchronized, no
processes can create new read locks. We wait until the sum of percpu
counters is zero - when it is, there are no readers in the critical
section.
Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
2012-09-26 13:46:43 +08:00
|
|
|
#ifndef _LINUX_PERCPU_RWSEM_H
|
|
|
|
|
#define _LINUX_PERCPU_RWSEM_H
|
|
|
|
|
|
2012-12-18 08:01:36 +08:00
|
|
|
#include <linux/atomic.h>
|
2012-12-18 08:01:32 +08:00
|
|
|
#include <linux/rwsem.h>
|
blockdev: turn a rw semaphore into a percpu rw semaphore
This avoids cache line bouncing when many processes lock the semaphore
for read.
New percpu lock implementation
The lock consists of an array of percpu unsigned integers, a boolean
variable and a mutex.
When we take the lock for read, we enter rcu read section, check for a
"locked" variable. If it is false, we increase a percpu counter on the
current cpu and exit the rcu section. If "locked" is true, we exit the
rcu section, take the mutex and drop it (this waits until a writer
finished) and retry.
Unlocking for read just decreases percpu variable. Note that we can
unlock on a difference cpu than where we locked, in this case the
counter underflows. The sum of all percpu counters represents the number
of processes that hold the lock for read.
When we need to lock for write, we take the mutex, set "locked" variable
to true and synchronize rcu. Since RCU has been synchronized, no
processes can create new read locks. We wait until the sum of percpu
counters is zero - when it is, there are no readers in the critical
section.
Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
2012-09-26 13:46:43 +08:00
|
|
|
#include <linux/percpu.h>
|
2012-12-18 08:01:32 +08:00
|
|
|
#include <linux/wait.h>
|
2012-12-18 08:01:38 +08:00
|
|
|
#include <linux/lockdep.h>
|
blockdev: turn a rw semaphore into a percpu rw semaphore
This avoids cache line bouncing when many processes lock the semaphore
for read.
New percpu lock implementation
The lock consists of an array of percpu unsigned integers, a boolean
variable and a mutex.
When we take the lock for read, we enter rcu read section, check for a
"locked" variable. If it is false, we increase a percpu counter on the
current cpu and exit the rcu section. If "locked" is true, we exit the
rcu section, take the mutex and drop it (this waits until a writer
finished) and retry.
Unlocking for read just decreases percpu variable. Note that we can
unlock on a difference cpu than where we locked, in this case the
counter underflows. The sum of all percpu counters represents the number
of processes that hold the lock for read.
When we need to lock for write, we take the mutex, set "locked" variable
to true and synchronize rcu. Since RCU has been synchronized, no
processes can create new read locks. We wait until the sum of percpu
counters is zero - when it is, there are no readers in the critical
section.
Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
2012-09-26 13:46:43 +08:00
|
|
|
|
|
|
|
|
struct percpu_rw_semaphore {
|
2012-12-18 08:01:32 +08:00
|
|
|
unsigned int __percpu *fast_read_ctr;
|
2012-12-18 08:01:36 +08:00
|
|
|
atomic_t write_ctr;
|
2012-12-18 08:01:32 +08:00
|
|
|
struct rw_semaphore rw_sem;
|
|
|
|
|
atomic_t slow_read_ctr;
|
|
|
|
|
wait_queue_head_t write_waitq;
|
blockdev: turn a rw semaphore into a percpu rw semaphore
This avoids cache line bouncing when many processes lock the semaphore
for read.
New percpu lock implementation
The lock consists of an array of percpu unsigned integers, a boolean
variable and a mutex.
When we take the lock for read, we enter rcu read section, check for a
"locked" variable. If it is false, we increase a percpu counter on the
current cpu and exit the rcu section. If "locked" is true, we exit the
rcu section, take the mutex and drop it (this waits until a writer
finished) and retry.
Unlocking for read just decreases percpu variable. Note that we can
unlock on a difference cpu than where we locked, in this case the
counter underflows. The sum of all percpu counters represents the number
of processes that hold the lock for read.
When we need to lock for write, we take the mutex, set "locked" variable
to true and synchronize rcu. Since RCU has been synchronized, no
processes can create new read locks. We wait until the sum of percpu
counters is zero - when it is, there are no readers in the critical
section.
Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
2012-09-26 13:46:43 +08:00
|
|
|
};
|
|
|
|
|
|
2012-12-18 08:01:32 +08:00
|
|
|
extern void percpu_down_read(struct percpu_rw_semaphore *);
|
|
|
|
|
extern void percpu_up_read(struct percpu_rw_semaphore *);
|
2012-10-23 07:37:47 +08:00
|
|
|
|
2012-12-18 08:01:32 +08:00
|
|
|
extern void percpu_down_write(struct percpu_rw_semaphore *);
|
|
|
|
|
extern void percpu_up_write(struct percpu_rw_semaphore *);
|
blockdev: turn a rw semaphore into a percpu rw semaphore
This avoids cache line bouncing when many processes lock the semaphore
for read.
New percpu lock implementation
The lock consists of an array of percpu unsigned integers, a boolean
variable and a mutex.
When we take the lock for read, we enter rcu read section, check for a
"locked" variable. If it is false, we increase a percpu counter on the
current cpu and exit the rcu section. If "locked" is true, we exit the
rcu section, take the mutex and drop it (this waits until a writer
finished) and retry.
Unlocking for read just decreases percpu variable. Note that we can
unlock on a difference cpu than where we locked, in this case the
counter underflows. The sum of all percpu counters represents the number
of processes that hold the lock for read.
When we need to lock for write, we take the mutex, set "locked" variable
to true and synchronize rcu. Since RCU has been synchronized, no
processes can create new read locks. We wait until the sum of percpu
counters is zero - when it is, there are no readers in the critical
section.
Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
2012-09-26 13:46:43 +08:00
|
|
|
|
2012-12-18 08:01:38 +08:00
|
|
|
extern int __percpu_init_rwsem(struct percpu_rw_semaphore *,
|
|
|
|
|
const char *, struct lock_class_key *);
|
2012-12-18 08:01:32 +08:00
|
|
|
extern void percpu_free_rwsem(struct percpu_rw_semaphore *);
|
blockdev: turn a rw semaphore into a percpu rw semaphore
This avoids cache line bouncing when many processes lock the semaphore
for read.
New percpu lock implementation
The lock consists of an array of percpu unsigned integers, a boolean
variable and a mutex.
When we take the lock for read, we enter rcu read section, check for a
"locked" variable. If it is false, we increase a percpu counter on the
current cpu and exit the rcu section. If "locked" is true, we exit the
rcu section, take the mutex and drop it (this waits until a writer
finished) and retry.
Unlocking for read just decreases percpu variable. Note that we can
unlock on a difference cpu than where we locked, in this case the
counter underflows. The sum of all percpu counters represents the number
of processes that hold the lock for read.
When we need to lock for write, we take the mutex, set "locked" variable
to true and synchronize rcu. Since RCU has been synchronized, no
processes can create new read locks. We wait until the sum of percpu
counters is zero - when it is, there are no readers in the critical
section.
Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
2012-09-26 13:46:43 +08:00
|
|
|
|
2012-12-18 08:01:38 +08:00
|
|
|
#define percpu_init_rwsem(brw) \
|
|
|
|
|
({ \
|
|
|
|
|
static struct lock_class_key rwsem_key; \
|
|
|
|
|
__percpu_init_rwsem(brw, #brw, &rwsem_key); \
|
|
|
|
|
})
|
|
|
|
|
|
blockdev: turn a rw semaphore into a percpu rw semaphore
This avoids cache line bouncing when many processes lock the semaphore
for read.
New percpu lock implementation
The lock consists of an array of percpu unsigned integers, a boolean
variable and a mutex.
When we take the lock for read, we enter rcu read section, check for a
"locked" variable. If it is false, we increase a percpu counter on the
current cpu and exit the rcu section. If "locked" is true, we exit the
rcu section, take the mutex and drop it (this waits until a writer
finished) and retry.
Unlocking for read just decreases percpu variable. Note that we can
unlock on a difference cpu than where we locked, in this case the
counter underflows. The sum of all percpu counters represents the number
of processes that hold the lock for read.
When we need to lock for write, we take the mutex, set "locked" variable
to true and synchronize rcu. Since RCU has been synchronized, no
processes can create new read locks. We wait until the sum of percpu
counters is zero - when it is, there are no readers in the critical
section.
Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
2012-09-26 13:46:43 +08:00
|
|
|
#endif
|