2013-06-01 06:26:45 +08:00
|
|
|
#define pr_fmt(fmt) "%s: " fmt "\n", __func__
|
|
|
|
|
|
|
|
#include <linux/kernel.h>
|
|
|
|
#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
|
|
|
|
* pcpu_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
|
|
|
|
* 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 13:27:25 +08:00
|
|
|
* 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 06:26:45 +08:00
|
|
|
*/
|
|
|
|
|
2014-09-20 13:27:25 +08:00
|
|
|
#define PCPU_COUNT_BIAS (1LU << (BITS_PER_LONG - 1))
|
2013-06-01 06:26:45 +08:00
|
|
|
|
2014-09-20 13:27:25 +08:00
|
|
|
static unsigned long __percpu *pcpu_count_ptr(struct percpu_ref *ref)
|
2014-06-28 20:10:13 +08:00
|
|
|
{
|
2014-09-20 13:27:25 +08:00
|
|
|
return (unsigned long __percpu *)(ref->pcpu_count_ptr & ~PCPU_REF_DEAD);
|
2014-06-28 20:10:13 +08:00
|
|
|
}
|
|
|
|
|
2013-06-01 06:26:45 +08:00
|
|
|
/**
|
|
|
|
* percpu_ref_init - initialize a percpu refcount
|
2013-06-13 11:43:06 +08:00
|
|
|
* @ref: percpu_ref to initialize
|
|
|
|
* @release: function which will be called when refcount hits 0
|
2014-09-08 08:51:30 +08:00
|
|
|
* @gfp: allocation mask to use
|
2013-06-01 06:26:45 +08:00
|
|
|
*
|
|
|
|
* Initializes the refcount in single atomic counter mode with a refcount of 1;
|
2014-09-20 13:27:25 +08:00
|
|
|
* analagous to atomic_long_set(ref, 1).
|
2013-06-01 06:26:45 +08:00
|
|
|
*
|
|
|
|
* Note that @release must not sleep - it may potentially be called from RCU
|
|
|
|
* callback context by percpu_ref_kill().
|
|
|
|
*/
|
2014-09-08 08:51:30 +08:00
|
|
|
int percpu_ref_init(struct percpu_ref *ref, percpu_ref_func_t *release,
|
|
|
|
gfp_t gfp)
|
2013-06-01 06:26:45 +08:00
|
|
|
{
|
2014-09-20 13:27:25 +08:00
|
|
|
atomic_long_set(&ref->count, 1 + PCPU_COUNT_BIAS);
|
2013-06-01 06:26:45 +08:00
|
|
|
|
2014-09-20 13:27:25 +08:00
|
|
|
ref->pcpu_count_ptr = (unsigned long)alloc_percpu_gfp(unsigned long, gfp);
|
2014-06-28 20:10:13 +08:00
|
|
|
if (!ref->pcpu_count_ptr)
|
2013-06-01 06:26:45 +08:00
|
|
|
return -ENOMEM;
|
|
|
|
|
|
|
|
ref->release = release;
|
|
|
|
return 0;
|
|
|
|
}
|
2013-10-17 04:47:01 +08:00
|
|
|
EXPORT_SYMBOL_GPL(percpu_ref_init);
|
2013-06-01 06:26:45 +08:00
|
|
|
|
2014-06-28 20:10:14 +08:00
|
|
|
/**
|
|
|
|
* 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, killed
|
|
|
|
* 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)
|
|
|
|
{
|
2014-09-20 13:27:25 +08:00
|
|
|
unsigned long __percpu *pcpu_count = pcpu_count_ptr(ref);
|
2014-06-28 20:10:14 +08:00
|
|
|
int cpu;
|
|
|
|
|
|
|
|
BUG_ON(!pcpu_count);
|
|
|
|
WARN_ON(!percpu_ref_is_zero(ref));
|
|
|
|
|
2014-09-20 13:27:25 +08:00
|
|
|
atomic_long_set(&ref->count, 1 + PCPU_COUNT_BIAS);
|
2014-06-28 20:10:14 +08:00
|
|
|
|
|
|
|
/*
|
|
|
|
* Restore per-cpu operation. smp_store_release() is paired with
|
|
|
|
* smp_read_barrier_depends() in __pcpu_ref_alive() and guarantees
|
|
|
|
* that the zeroing is visible to all percpu accesses which can see
|
|
|
|
* the following PCPU_REF_DEAD clearing.
|
|
|
|
*/
|
|
|
|
for_each_possible_cpu(cpu)
|
|
|
|
*per_cpu_ptr(pcpu_count, cpu) = 0;
|
|
|
|
|
|
|
|
smp_store_release(&ref->pcpu_count_ptr,
|
|
|
|
ref->pcpu_count_ptr & ~PCPU_REF_DEAD);
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL_GPL(percpu_ref_reinit);
|
|
|
|
|
2013-06-13 11:52:35 +08:00
|
|
|
/**
|
2014-06-28 20:10:14 +08:00
|
|
|
* percpu_ref_exit - undo percpu_ref_init()
|
|
|
|
* @ref: percpu_ref to exit
|
2013-06-13 11:52:35 +08:00
|
|
|
*
|
2014-06-28 20:10:14 +08:00
|
|
|
* 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 11:52:35 +08:00
|
|
|
*/
|
2014-06-28 20:10:14 +08:00
|
|
|
void percpu_ref_exit(struct percpu_ref *ref)
|
2013-06-13 11:52:35 +08:00
|
|
|
{
|
2014-09-20 13:27:25 +08:00
|
|
|
unsigned long __percpu *pcpu_count = pcpu_count_ptr(ref);
|
2013-06-13 11:52:35 +08:00
|
|
|
|
|
|
|
if (pcpu_count) {
|
2014-06-28 20:10:13 +08:00
|
|
|
free_percpu(pcpu_count);
|
2014-06-28 20:10:14 +08:00
|
|
|
ref->pcpu_count_ptr = PCPU_REF_DEAD;
|
2013-06-13 11:52:35 +08:00
|
|
|
}
|
|
|
|
}
|
2014-06-28 20:10:14 +08:00
|
|
|
EXPORT_SYMBOL_GPL(percpu_ref_exit);
|
2013-06-13 11:52:35 +08:00
|
|
|
|
2013-06-01 06:26:45 +08:00
|
|
|
static void percpu_ref_kill_rcu(struct rcu_head *rcu)
|
|
|
|
{
|
|
|
|
struct percpu_ref *ref = container_of(rcu, struct percpu_ref, rcu);
|
2014-09-20 13:27:25 +08:00
|
|
|
unsigned long __percpu *pcpu_count = pcpu_count_ptr(ref);
|
|
|
|
unsigned long count = 0;
|
2013-06-01 06:26:45 +08:00
|
|
|
int cpu;
|
|
|
|
|
|
|
|
for_each_possible_cpu(cpu)
|
|
|
|
count += *per_cpu_ptr(pcpu_count, cpu);
|
|
|
|
|
2014-09-20 13:27:25 +08:00
|
|
|
pr_debug("global %ld pcpu %ld",
|
|
|
|
atomic_long_read(&ref->count), (long)count);
|
2013-06-01 06:26:45 +08:00
|
|
|
|
|
|
|
/*
|
|
|
|
* 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:
|
|
|
|
*/
|
|
|
|
|
2014-09-20 13:27:25 +08:00
|
|
|
atomic_long_add((long)count - PCPU_COUNT_BIAS, &ref->count);
|
2013-06-01 06:26:45 +08:00
|
|
|
|
2014-09-20 13:27:25 +08:00
|
|
|
WARN_ONCE(atomic_long_read(&ref->count) <= 0,
|
|
|
|
"percpu ref (%pf) <= 0 (%ld) after killed",
|
|
|
|
ref->release, atomic_long_read(&ref->count));
|
2014-01-07 05:13:26 +08:00
|
|
|
|
2013-06-14 10:23:53 +08:00
|
|
|
/* @ref is viewed as dead on all CPUs, send out kill confirmation */
|
|
|
|
if (ref->confirm_kill)
|
|
|
|
ref->confirm_kill(ref);
|
|
|
|
|
2013-06-01 06:26:45 +08:00
|
|
|
/*
|
|
|
|
* Now we're in single atomic_t mode with a consistent refcount, so it's
|
|
|
|
* safe to drop our initial ref:
|
|
|
|
*/
|
|
|
|
percpu_ref_put(ref);
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
2013-06-14 10:23:53 +08:00
|
|
|
* percpu_ref_kill_and_confirm - drop the initial ref and schedule confirmation
|
2013-06-13 11:43:06 +08:00
|
|
|
* @ref: percpu_ref to kill
|
2013-06-14 10:23:53 +08:00
|
|
|
* @confirm_kill: optional confirmation callback
|
2013-06-01 06:26:45 +08:00
|
|
|
*
|
2013-06-14 10:23:53 +08:00
|
|
|
* 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 - all further
|
|
|
|
* invocations of percpu_ref_tryget() will fail. See percpu_ref_tryget()
|
|
|
|
* for more details.
|
2013-06-01 06:26:45 +08:00
|
|
|
*
|
2013-06-14 10:23:53 +08:00
|
|
|
* Due to the way percpu_ref is implemented, @confirm_kill will be called
|
|
|
|
* after at least one full RCU grace period has passed but this is an
|
|
|
|
* implementation detail and callers must not depend on it.
|
2013-06-01 06:26:45 +08:00
|
|
|
*/
|
2013-06-14 10:23:53 +08:00
|
|
|
void percpu_ref_kill_and_confirm(struct percpu_ref *ref,
|
|
|
|
percpu_ref_func_t *confirm_kill)
|
2013-06-01 06:26:45 +08:00
|
|
|
{
|
2014-06-28 20:10:13 +08:00
|
|
|
WARN_ONCE(ref->pcpu_count_ptr & PCPU_REF_DEAD,
|
2014-09-20 13:27:24 +08:00
|
|
|
"percpu_ref_kill() called more than once on %pf!",
|
|
|
|
ref->release);
|
2013-06-01 06:26:45 +08:00
|
|
|
|
2014-06-28 20:10:13 +08:00
|
|
|
ref->pcpu_count_ptr |= PCPU_REF_DEAD;
|
2013-06-14 10:23:53 +08:00
|
|
|
ref->confirm_kill = confirm_kill;
|
2013-06-01 06:26:45 +08:00
|
|
|
|
2013-06-17 07:12:26 +08:00
|
|
|
call_rcu_sched(&ref->rcu, percpu_ref_kill_rcu);
|
2013-06-01 06:26:45 +08:00
|
|
|
}
|
2013-10-17 04:47:01 +08:00
|
|
|
EXPORT_SYMBOL_GPL(percpu_ref_kill_and_confirm);
|