Commit f47ad457 authored by Tejun Heo's avatar Tejun Heo

percpu_ref: decouple switching to percpu mode and reinit

percpu_ref has treated the dropping of the base reference and
switching to atomic mode as an integral operation; however, there's
nothing inherent tying the two together.

The use cases for percpu_ref have been expanding continuously.  While
the current init/kill/reinit/exit model can cover a lot, the coupling
of kill/reinit with atomic/percpu mode switching is turning out to be
too restrictive for use cases where many percpu_refs are created and
destroyed back-to-back with only some of them reaching extended
operation.  The coupling also makes implementing always-atomic debug
mode difficult.

This patch separates out percpu mode switching into
percpu_ref_switch_to_percpu() and reimplements percpu_ref_reinit() on
top of it.

* DEAD still requires ATOMIC.  A dead ref can't be switched to percpu
  mode w/o going through reinit.

v2: __percpu_ref_switch_to_percpu() was missing static.  Fixed.
    Reported by Fengguang aka kbuild test robot.
Signed-off-by: default avatarTejun Heo <tj@kernel.org>
Reviewed-by: default avatarKent Overstreet <kmo@daterainc.com>
Cc: Jens Axboe <axboe@kernel.dk>
Cc: Christoph Hellwig <hch@infradead.org>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: kbuild test robot <fengguang.wu@intel.com>
parent 490c79a6
...@@ -80,9 +80,10 @@ int __must_check percpu_ref_init(struct percpu_ref *ref, ...@@ -80,9 +80,10 @@ int __must_check percpu_ref_init(struct percpu_ref *ref,
void percpu_ref_exit(struct percpu_ref *ref); void percpu_ref_exit(struct percpu_ref *ref);
void percpu_ref_switch_to_atomic(struct percpu_ref *ref, void percpu_ref_switch_to_atomic(struct percpu_ref *ref,
percpu_ref_func_t *confirm_switch); percpu_ref_func_t *confirm_switch);
void percpu_ref_reinit(struct percpu_ref *ref); void percpu_ref_switch_to_percpu(struct percpu_ref *ref);
void percpu_ref_kill_and_confirm(struct percpu_ref *ref, void percpu_ref_kill_and_confirm(struct percpu_ref *ref,
percpu_ref_func_t *confirm_kill); percpu_ref_func_t *confirm_kill);
void percpu_ref_reinit(struct percpu_ref *ref);
/** /**
* percpu_ref_kill - drop the initial ref * percpu_ref_kill - drop the initial ref
......
...@@ -206,40 +206,54 @@ void percpu_ref_switch_to_atomic(struct percpu_ref *ref, ...@@ -206,40 +206,54 @@ void percpu_ref_switch_to_atomic(struct percpu_ref *ref,
__percpu_ref_switch_to_atomic(ref, confirm_switch); __percpu_ref_switch_to_atomic(ref, confirm_switch);
} }
/** static void __percpu_ref_switch_to_percpu(struct percpu_ref *ref)
* 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)
{ {
unsigned long __percpu *percpu_count = percpu_count_ptr(ref); unsigned long __percpu *percpu_count = percpu_count_ptr(ref);
int cpu; int cpu;
BUG_ON(!percpu_count); BUG_ON(!percpu_count);
WARN_ON_ONCE(!percpu_ref_is_zero(ref));
atomic_long_set(&ref->count, 1 + PERCPU_COUNT_BIAS); 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);
/* /*
* Restore per-cpu operation. smp_store_release() is paired with * Restore per-cpu operation. smp_store_release() is paired with
* smp_read_barrier_depends() in __ref_is_percpu() and guarantees * smp_read_barrier_depends() in __ref_is_percpu() and guarantees
* that the zeroing is visible to all percpu accesses which can see * that the zeroing is visible to all percpu accesses which can see
* the following __PERCPU_REF_ATOMIC_DEAD clearing. * the following __PERCPU_REF_ATOMIC clearing.
*/ */
for_each_possible_cpu(cpu) for_each_possible_cpu(cpu)
*per_cpu_ptr(percpu_count, cpu) = 0; *per_cpu_ptr(percpu_count, cpu) = 0;
smp_store_release(&ref->percpu_count_ptr, smp_store_release(&ref->percpu_count_ptr,
ref->percpu_count_ptr & ~__PERCPU_REF_ATOMIC_DEAD); 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);
} }
EXPORT_SYMBOL_GPL(percpu_ref_reinit);
/** /**
* percpu_ref_kill_and_confirm - drop the initial ref and schedule confirmation * percpu_ref_kill_and_confirm - drop the initial ref and schedule confirmation
...@@ -253,8 +267,8 @@ EXPORT_SYMBOL_GPL(percpu_ref_reinit); ...@@ -253,8 +267,8 @@ EXPORT_SYMBOL_GPL(percpu_ref_reinit);
* percpu_ref_tryget_live() for details. * percpu_ref_tryget_live() for details.
* *
* This function normally doesn't block and can be called from any context * 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 * but it may block if @confirm_kill is specified and @ref is in the
* the process of switching to atomic mode by percpu_ref_switch_atomic(). * process of switching to atomic mode by percpu_ref_switch_atomic().
* *
* Due to the way percpu_ref is implemented, @confirm_switch will be called * 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 * after at least one full sched RCU grace period has passed but this is an
...@@ -271,3 +285,24 @@ void percpu_ref_kill_and_confirm(struct percpu_ref *ref, ...@@ -271,3 +285,24 @@ void percpu_ref_kill_and_confirm(struct percpu_ref *ref,
percpu_ref_put(ref); percpu_ref_put(ref);
} }
EXPORT_SYMBOL_GPL(percpu_ref_kill_and_confirm); EXPORT_SYMBOL_GPL(percpu_ref_kill_and_confirm);
/**
* 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);
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment