Donate to e Foundation | Murena handsets with /e/OS | Own a part of Murena! Learn more

Commit b393e8b3 authored Oct 09, 2017 by Paul E. McKenney

percpu: READ_ONCE() now implies smp_read_barrier_depends()



Because READ_ONCE() now implies smp_read_barrier_depends(), this commit
removes the now-redundant smp_read_barrier_depends() following the
READ_ONCE() in __ref_is_percpu().

Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Acked-by: Tejun Heo <tj@kernel.org>
Cc: Christoph Lameter <cl@linux.com>

parent 7088efa9

include/linux/percpu-refcount.h

+3 −3

Original line number	Original line	Diff line number	Diff line
	@@ -139,12 +139,12 @@ static inline bool __ref_is_percpu(struct percpu_ref *ref,
	* when using it as a pointer, __PERCPU_REF_ATOMIC may be set in		* when using it as a pointer, __PERCPU_REF_ATOMIC may be set in
	* between contaminating the pointer value, meaning that		* between contaminating the pointer value, meaning that
	* READ_ONCE() is required when fetching it.		* READ_ONCE() is required when fetching it.
			*
			* The smp_read_barrier_depends() implied by READ_ONCE() pairs
			* with smp_store_release() in __percpu_ref_switch_to_percpu().
	*/		*/
	percpu_ptr = READ_ONCE(ref->percpu_count_ptr);		percpu_ptr = READ_ONCE(ref->percpu_count_ptr);

	/* paired with smp_store_release() in __percpu_ref_switch_to_percpu() */
	smp_read_barrier_depends();

	/*		/*
	* Theoretically, the following could test just ATOMIC; however,		* Theoretically, the following could test just ATOMIC; however,
	* then we'd have to mask off DEAD separately as DEAD may be		* then we'd have to mask off DEAD separately as DEAD may be

lib/percpu-refcount.c

+4 −4

Original line number	Original line	Diff line number	Diff line
	@@ -197,10 +197,10 @@ static void __percpu_ref_switch_to_percpu(struct percpu_ref *ref)
	atomic_long_add(PERCPU_COUNT_BIAS, &ref->count);		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
	* smp_read_barrier_depends() in __ref_is_percpu() and guarantees		* with READ_ONCE() in __ref_is_percpu() and guarantees that the
	* that the zeroing is visible to all percpu accesses which can see		* zeroing is visible to all percpu accesses which can see the
	* the following __PERCPU_REF_ATOMIC clearing.		* 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;