2013-06-01 00:26:45 +02:00
|
|
|
#define pr_fmt(fmt) "%s: " fmt "\n", __func__
|
|
|
|
|
|
|
|
#include <linux/kernel.h>
|
2014-09-24 19:31:49 +02:00
|
|
|
#include <linux/sched.h>
|
|
|
|
#include <linux/wait.h>
|
2013-06-01 00:26:45 +02:00
|
|
|
#include <linux/percpu-refcount.h>
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Initially, a percpu refcount is just a set of percpu counters. Initially, we
|
|
|
|
* don't try to detect the ref hitting 0 - which means that get/put can just
|
|
|
|
* increment or decrement the local counter. Note that the counter on a
|
|
|
|
* particular cpu can (and will) wrap - this is fine, when we go to shutdown the
|
|
|
|
* percpu counters will all sum to the correct value
|
|
|
|
*
|
|
|
|
* (More precisely: because moduler arithmatic is commutative the sum of all the
|
2014-09-24 19:31:48 +02:00
|
|
|
* percpu_count vars will be equal to what it would have been if all the gets
|
|
|
|
* and puts were done to a single integer, even if some of the percpu integers
|
2013-06-01 00:26:45 +02:00
|
|
|
* overflow or underflow).
|
|
|
|
*
|
|
|
|
* The real trick to implementing percpu refcounts is shutdown. We can't detect
|
|
|
|
* the ref hitting 0 on every put - this would require global synchronization
|
|
|
|
* and defeat the whole purpose of using percpu refs.
|
|
|
|
*
|
|
|
|
* What we do is require the user to keep track of the initial refcount; we know
|
|
|
|
* the ref can't hit 0 before the user drops the initial ref, so as long as we
|
|
|
|
* convert to non percpu mode before the initial ref is dropped everything
|
|
|
|
* works.
|
|
|
|
*
|
|
|
|
* Converting to non percpu mode is done with some RCUish stuff in
|
2014-09-20 07:27:25 +02:00
|
|
|
* percpu_ref_kill. Additionally, we need a bias value so that the
|
|
|
|
* atomic_long_t can't hit 0 before we've added up all the percpu refs.
|
2013-06-01 00:26:45 +02:00
|
|
|
*/
|
|
|
|
|
2014-09-24 19:31:48 +02:00
|
|
|
#define PERCPU_COUNT_BIAS (1LU << (BITS_PER_LONG - 1))
|
2013-06-01 00:26:45 +02:00
|
|
|
|
2014-09-24 19:31:49 +02:00
|
|
|
static DECLARE_WAIT_QUEUE_HEAD(percpu_ref_switch_waitq);
|
|
|
|
|
2014-09-24 19:31:48 +02:00
|
|
|
static unsigned long __percpu *percpu_count_ptr(struct percpu_ref *ref)
|
2014-06-28 14:10:13 +02:00
|
|
|
{
|
2014-09-24 19:31:48 +02:00
|
|
|
return (unsigned long __percpu *)
|
2014-09-24 19:31:49 +02:00
|
|
|
(ref->percpu_count_ptr & ~__PERCPU_REF_ATOMIC_DEAD);
|
2014-06-28 14:10:13 +02:00
|
|
|
}
|
|
|
|
|
2013-06-01 00:26:45 +02:00
|
|
|
/**
|
|
|
|
* percpu_ref_init - initialize a percpu refcount
|
2013-06-13 05:43:06 +02:00
|
|
|
* @ref: percpu_ref to initialize
|
|
|
|
* @release: function which will be called when refcount hits 0
|
2014-09-08 02:51:30 +02:00
|
|
|
* @gfp: allocation mask to use
|
2013-06-01 00:26:45 +02:00
|
|
|
*
|
|
|
|
* Initializes the refcount in single atomic counter mode with a refcount of 1;
|
2014-09-20 07:27:25 +02:00
|
|
|
* analagous to atomic_long_set(ref, 1).
|
2013-06-01 00:26:45 +02:00
|
|
|
*
|
|
|
|
* Note that @release must not sleep - it may potentially be called from RCU
|
|
|
|
* callback context by percpu_ref_kill().
|
|
|
|
*/
|
2014-09-08 02:51:30 +02:00
|
|
|
int percpu_ref_init(struct percpu_ref *ref, percpu_ref_func_t *release,
|
|
|
|
gfp_t gfp)
|
2013-06-01 00:26:45 +02:00
|
|
|
{
|
2014-09-24 19:31:49 +02:00
|
|
|
size_t align = max_t(size_t, 1 << __PERCPU_REF_FLAG_BITS,
|
|
|
|
__alignof__(unsigned long));
|
|
|
|
|
2014-09-24 19:31:48 +02:00
|
|
|
atomic_long_set(&ref->count, 1 + PERCPU_COUNT_BIAS);
|
2013-06-01 00:26:45 +02:00
|
|
|
|
2014-09-24 19:31:49 +02:00
|
|
|
ref->percpu_count_ptr = (unsigned long)
|
|
|
|
__alloc_percpu_gfp(sizeof(unsigned long), align, gfp);
|
2014-09-24 19:31:48 +02:00
|
|
|
if (!ref->percpu_count_ptr)
|
2013-06-01 00:26:45 +02:00
|
|
|
return -ENOMEM;
|
|
|
|
|
|
|
|
ref->release = release;
|
|
|
|
return 0;
|
|
|
|
}
|
2013-10-16 22:47:01 +02:00
|
|
|
EXPORT_SYMBOL_GPL(percpu_ref_init);
|
2013-06-01 00:26:45 +02:00
|
|
|
|
2013-06-13 05:52:35 +02:00
|
|
|
/**
|
2014-06-28 14:10:14 +02:00
|
|
|
* percpu_ref_exit - undo percpu_ref_init()
|
|
|
|
* @ref: percpu_ref to exit
|
2013-06-13 05:52:35 +02:00
|
|
|
*
|
2014-06-28 14:10:14 +02:00
|
|
|
* This function exits @ref. The caller is responsible for ensuring that
|
|
|
|
* @ref is no longer in active use. The usual places to invoke this
|
|
|
|
* function from are the @ref->release() callback or in init failure path
|
|
|
|
* where percpu_ref_init() succeeded but other parts of the initialization
|
|
|
|
* of the embedding object failed.
|
2013-06-13 05:52:35 +02:00
|
|
|
*/
|
2014-06-28 14:10:14 +02:00
|
|
|
void percpu_ref_exit(struct percpu_ref *ref)
|
2013-06-13 05:52:35 +02:00
|
|
|
{
|
2014-09-24 19:31:48 +02:00
|
|
|
unsigned long __percpu *percpu_count = percpu_count_ptr(ref);
|
2013-06-13 05:52:35 +02:00
|
|
|
|
2014-09-24 19:31:48 +02:00
|
|
|
if (percpu_count) {
|
|
|
|
free_percpu(percpu_count);
|
2014-09-24 19:31:49 +02:00
|
|
|
ref->percpu_count_ptr = __PERCPU_REF_ATOMIC_DEAD;
|
2013-06-13 05:52:35 +02:00
|
|
|
}
|
|
|
|
}
|
2014-06-28 14:10:14 +02:00
|
|
|
EXPORT_SYMBOL_GPL(percpu_ref_exit);
|
2013-06-13 05:52:35 +02:00
|
|
|
|
2014-09-24 19:31:49 +02:00
|
|
|
static void percpu_ref_call_confirm_rcu(struct rcu_head *rcu)
|
|
|
|
{
|
|
|
|
struct percpu_ref *ref = container_of(rcu, struct percpu_ref, rcu);
|
|
|
|
|
|
|
|
ref->confirm_switch(ref);
|
|
|
|
ref->confirm_switch = NULL;
|
|
|
|
wake_up_all(&percpu_ref_switch_waitq);
|
|
|
|
|
|
|
|
/* drop ref from percpu_ref_switch_to_atomic() */
|
|
|
|
percpu_ref_put(ref);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void percpu_ref_switch_to_atomic_rcu(struct rcu_head *rcu)
|
2013-06-01 00:26:45 +02:00
|
|
|
{
|
|
|
|
struct percpu_ref *ref = container_of(rcu, struct percpu_ref, rcu);
|
2014-09-24 19:31:48 +02:00
|
|
|
unsigned long __percpu *percpu_count = percpu_count_ptr(ref);
|
2014-09-20 07:27:25 +02:00
|
|
|
unsigned long count = 0;
|
2013-06-01 00:26:45 +02:00
|
|
|
int cpu;
|
|
|
|
|
|
|
|
for_each_possible_cpu(cpu)
|
2014-09-24 19:31:48 +02:00
|
|
|
count += *per_cpu_ptr(percpu_count, cpu);
|
2013-06-01 00:26:45 +02:00
|
|
|
|
2014-09-24 19:31:48 +02:00
|
|
|
pr_debug("global %ld percpu %ld",
|
2014-09-20 07:27:25 +02:00
|
|
|
atomic_long_read(&ref->count), (long)count);
|
2013-06-01 00:26:45 +02:00
|
|
|
|
|
|
|
/*
|
|
|
|
* It's crucial that we sum the percpu counters _before_ adding the sum
|
|
|
|
* to &ref->count; since gets could be happening on one cpu while puts
|
|
|
|
* happen on another, adding a single cpu's count could cause
|
|
|
|
* @ref->count to hit 0 before we've got a consistent value - but the
|
|
|
|
* sum of all the counts will be consistent and correct.
|
|
|
|
*
|
|
|
|
* Subtracting the bias value then has to happen _after_ adding count to
|
|
|
|
* &ref->count; we need the bias value to prevent &ref->count from
|
|
|
|
* reaching 0 before we add the percpu counts. But doing it at the same
|
|
|
|
* time is equivalent and saves us atomic operations:
|
|
|
|
*/
|
2014-09-24 19:31:48 +02:00
|
|
|
atomic_long_add((long)count - PERCPU_COUNT_BIAS, &ref->count);
|
2013-06-01 00:26:45 +02:00
|
|
|
|
2014-09-20 07:27:25 +02:00
|
|
|
WARN_ONCE(atomic_long_read(&ref->count) <= 0,
|
2014-09-24 19:31:49 +02:00
|
|
|
"percpu ref (%pf) <= 0 (%ld) after switching to atomic",
|
2014-09-20 07:27:25 +02:00
|
|
|
ref->release, atomic_long_read(&ref->count));
|
2014-01-06 22:13:26 +01:00
|
|
|
|
2014-09-24 19:31:49 +02:00
|
|
|
/* @ref is viewed as dead on all CPUs, send out switch confirmation */
|
|
|
|
percpu_ref_call_confirm_rcu(rcu);
|
|
|
|
}
|
2013-06-14 04:23:53 +02:00
|
|
|
|
2014-09-24 19:31:49 +02:00
|
|
|
static void percpu_ref_noop_confirm_switch(struct percpu_ref *ref)
|
|
|
|
{
|
|
|
|
}
|
|
|
|
|
|
|
|
static void __percpu_ref_switch_to_atomic(struct percpu_ref *ref,
|
|
|
|
percpu_ref_func_t *confirm_switch)
|
|
|
|
{
|
|
|
|
if (!(ref->percpu_count_ptr & __PERCPU_REF_ATOMIC)) {
|
|
|
|
/* switching from percpu to atomic */
|
|
|
|
ref->percpu_count_ptr |= __PERCPU_REF_ATOMIC;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Non-NULL ->confirm_switch is used to indicate that
|
|
|
|
* switching is in progress. Use noop one if unspecified.
|
|
|
|
*/
|
|
|
|
WARN_ON_ONCE(ref->confirm_switch);
|
|
|
|
ref->confirm_switch =
|
|
|
|
confirm_switch ?: percpu_ref_noop_confirm_switch;
|
|
|
|
|
|
|
|
percpu_ref_get(ref); /* put after confirmation */
|
|
|
|
call_rcu_sched(&ref->rcu, percpu_ref_switch_to_atomic_rcu);
|
|
|
|
} else if (confirm_switch) {
|
|
|
|
/*
|
|
|
|
* Somebody already set ATOMIC. Switching may still be in
|
|
|
|
* progress. @confirm_switch must be invoked after the
|
|
|
|
* switching is complete and a full sched RCU grace period
|
|
|
|
* has passed. Wait synchronously for the previous
|
|
|
|
* switching and schedule @confirm_switch invocation.
|
|
|
|
*/
|
|
|
|
wait_event(percpu_ref_switch_waitq, !ref->confirm_switch);
|
|
|
|
ref->confirm_switch = confirm_switch;
|
|
|
|
|
|
|
|
percpu_ref_get(ref); /* put after confirmation */
|
|
|
|
call_rcu_sched(&ref->rcu, percpu_ref_call_confirm_rcu);
|
|
|
|
}
|
2013-06-01 00:26:45 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
2014-09-24 19:31:49 +02:00
|
|
|
* percpu_ref_switch_to_atomic - switch a percpu_ref to atomic mode
|
|
|
|
* @ref: percpu_ref to switch to atomic mode
|
|
|
|
* @confirm_switch: optional confirmation callback
|
2013-06-01 00:26:45 +02:00
|
|
|
*
|
2014-09-24 19:31:49 +02:00
|
|
|
* There's no reason to use this function for the usual reference counting.
|
|
|
|
* Use percpu_ref_kill[_and_confirm]().
|
|
|
|
*
|
|
|
|
* Schedule switching of @ref to atomic mode. All its percpu counts will
|
|
|
|
* be collected to the main atomic counter. On completion, when all CPUs
|
|
|
|
* are guaraneed to be in atomic mode, @confirm_switch, which may not
|
|
|
|
* block, is invoked. This function may be invoked concurrently with all
|
|
|
|
* the get/put operations and can safely be mixed with kill and reinit
|
|
|
|
* operations.
|
2013-06-01 00:26:45 +02:00
|
|
|
*
|
2014-09-24 19:31:49 +02:00
|
|
|
* This function normally doesn't block and can be called from any context
|
|
|
|
* but it may block if @confirm_kill is specified and @ref is already in
|
|
|
|
* the process of switching to atomic mode. In such cases, @confirm_switch
|
|
|
|
* will be invoked after the switching is complete.
|
|
|
|
*
|
|
|
|
* Due to the way percpu_ref is implemented, @confirm_switch will be called
|
|
|
|
* after at least one full sched RCU grace period has passed but this is an
|
|
|
|
* implementation detail and must not be depended upon.
|
2013-06-01 00:26:45 +02:00
|
|
|
*/
|
2014-09-24 19:31:49 +02:00
|
|
|
void percpu_ref_switch_to_atomic(struct percpu_ref *ref,
|
|
|
|
percpu_ref_func_t *confirm_switch)
|
2013-06-01 00:26:45 +02:00
|
|
|
{
|
2014-09-24 19:31:49 +02:00
|
|
|
__percpu_ref_switch_to_atomic(ref, confirm_switch);
|
2013-06-01 00:26:45 +02:00
|
|
|
}
|
2014-09-24 19:31:48 +02:00
|
|
|
|
2014-09-24 19:31:49 +02:00
|
|
|
static void __percpu_ref_switch_to_percpu(struct percpu_ref *ref)
|
2014-09-24 19:31:48 +02:00
|
|
|
{
|
2014-09-24 19:31:48 +02:00
|
|
|
unsigned long __percpu *percpu_count = percpu_count_ptr(ref);
|
2014-09-24 19:31:48 +02:00
|
|
|
int cpu;
|
|
|
|
|
2014-09-24 19:31:48 +02:00
|
|
|
BUG_ON(!percpu_count);
|
2014-09-24 19:31:48 +02:00
|
|
|
|
2014-09-24 19:31:49 +02:00
|
|
|
if (!(ref->percpu_count_ptr & __PERCPU_REF_ATOMIC))
|
|
|
|
return;
|
|
|
|
|
|
|
|
wait_event(percpu_ref_switch_waitq, !ref->confirm_switch);
|
|
|
|
|
|
|
|
atomic_long_add(PERCPU_COUNT_BIAS, &ref->count);
|
2014-09-24 19:31:48 +02:00
|
|
|
|
|
|
|
/*
|
|
|
|
* Restore per-cpu operation. smp_store_release() is paired with
|
2014-09-24 19:31:48 +02:00
|
|
|
* smp_read_barrier_depends() in __ref_is_percpu() and guarantees
|
|
|
|
* that the zeroing is visible to all percpu accesses which can see
|
2014-09-24 19:31:49 +02:00
|
|
|
* the following __PERCPU_REF_ATOMIC clearing.
|
2014-09-24 19:31:48 +02:00
|
|
|
*/
|
|
|
|
for_each_possible_cpu(cpu)
|
2014-09-24 19:31:48 +02:00
|
|
|
*per_cpu_ptr(percpu_count, cpu) = 0;
|
2014-09-24 19:31:48 +02:00
|
|
|
|
2014-09-24 19:31:48 +02:00
|
|
|
smp_store_release(&ref->percpu_count_ptr,
|
2014-09-24 19:31:49 +02:00
|
|
|
ref->percpu_count_ptr & ~__PERCPU_REF_ATOMIC);
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* percpu_ref_switch_to_percpu - switch a percpu_ref to percpu mode
|
|
|
|
* @ref: percpu_ref to switch to percpu mode
|
|
|
|
*
|
|
|
|
* There's no reason to use this function for the usual reference counting.
|
|
|
|
* To re-use an expired ref, use percpu_ref_reinit().
|
|
|
|
*
|
|
|
|
* Switch @ref to percpu mode. This function may be invoked concurrently
|
|
|
|
* with all the get/put operations and can safely be mixed with kill and
|
|
|
|
* reinit operations.
|
|
|
|
*
|
|
|
|
* This function normally doesn't block and can be called from any context
|
|
|
|
* but it may block if @ref is in the process of switching to atomic mode
|
|
|
|
* by percpu_ref_switch_atomic().
|
|
|
|
*/
|
|
|
|
void percpu_ref_switch_to_percpu(struct percpu_ref *ref)
|
|
|
|
{
|
|
|
|
/* a dying or dead ref can't be switched to percpu mode w/o reinit */
|
|
|
|
if (!(ref->percpu_count_ptr & __PERCPU_REF_DEAD))
|
|
|
|
__percpu_ref_switch_to_percpu(ref);
|
2014-09-24 19:31:48 +02:00
|
|
|
}
|
2014-09-24 19:31:49 +02:00
|
|
|
|
|
|
|
/**
|
|
|
|
* percpu_ref_kill_and_confirm - drop the initial ref and schedule confirmation
|
|
|
|
* @ref: percpu_ref to kill
|
|
|
|
* @confirm_kill: optional confirmation callback
|
|
|
|
*
|
|
|
|
* Equivalent to percpu_ref_kill() but also schedules kill confirmation if
|
|
|
|
* @confirm_kill is not NULL. @confirm_kill, which may not block, will be
|
|
|
|
* called after @ref is seen as dead from all CPUs at which point all
|
|
|
|
* further invocations of percpu_ref_tryget_live() will fail. See
|
|
|
|
* percpu_ref_tryget_live() for details.
|
|
|
|
*
|
|
|
|
* This function normally doesn't block and can be called from any context
|
2014-09-24 19:31:49 +02:00
|
|
|
* but it may block if @confirm_kill is specified and @ref is in the
|
|
|
|
* process of switching to atomic mode by percpu_ref_switch_atomic().
|
2014-09-24 19:31:49 +02:00
|
|
|
*
|
|
|
|
* Due to the way percpu_ref is implemented, @confirm_switch will be called
|
|
|
|
* after at least one full sched RCU grace period has passed but this is an
|
|
|
|
* implementation detail and must not be depended upon.
|
|
|
|
*/
|
|
|
|
void percpu_ref_kill_and_confirm(struct percpu_ref *ref,
|
|
|
|
percpu_ref_func_t *confirm_kill)
|
|
|
|
{
|
|
|
|
WARN_ONCE(ref->percpu_count_ptr & __PERCPU_REF_DEAD,
|
|
|
|
"%s called more than once on %pf!", __func__, ref->release);
|
|
|
|
|
|
|
|
ref->percpu_count_ptr |= __PERCPU_REF_DEAD;
|
|
|
|
__percpu_ref_switch_to_atomic(ref, confirm_kill);
|
|
|
|
percpu_ref_put(ref);
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL_GPL(percpu_ref_kill_and_confirm);
|
2014-09-24 19:31:49 +02:00
|
|
|
|
|
|
|
/**
|
|
|
|
* percpu_ref_reinit - re-initialize a percpu refcount
|
|
|
|
* @ref: perpcu_ref to re-initialize
|
|
|
|
*
|
|
|
|
* Re-initialize @ref so that it's in the same state as when it finished
|
|
|
|
* percpu_ref_init(). @ref must have been initialized successfully and
|
|
|
|
* reached 0 but not exited.
|
|
|
|
*
|
|
|
|
* Note that percpu_ref_tryget[_live]() are safe to perform on @ref while
|
|
|
|
* this function is in progress.
|
|
|
|
*/
|
|
|
|
void percpu_ref_reinit(struct percpu_ref *ref)
|
|
|
|
{
|
|
|
|
WARN_ON_ONCE(!percpu_ref_is_zero(ref));
|
|
|
|
|
|
|
|
ref->percpu_count_ptr &= ~__PERCPU_REF_DEAD;
|
|
|
|
percpu_ref_get(ref);
|
|
|
|
__percpu_ref_switch_to_percpu(ref);
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL_GPL(percpu_ref_reinit);
|