| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Message-Id: <1410142342-8185-4-git-send-email-tj@kernel.org>
Date: Mon, 8 Sep 2014 11:12:22 +0900
From: Tejun Heo <tj@...nel.org>
To: cl@...ux-foundation.org, kmo@...erainc.com
Cc: linux-kernel@...r.kernel.org, Tejun Heo <tj@...nel.org>,
Johannes Weiner <hannes@...xchg.org>
Subject: [PATCH 3/3] percpu-refcount: make percpu_ref based on longs instead of ints
percpu_ref is currently based on ints and the number of refs it can
cover is (1 << 31). This makes it impossible to use a percpu_ref to
count memory objects or pages on 64bit machines as it may overflow.
This forces those users to somehow aggregate the references before
contributing to the percpu_ref which is often cumbersome and sometimes
challenging to get the same level of performance as using the
percpu_ref directly.
While using ints for the percpu counters makes them pack tighter on
64bit machines, the possible gain from using ints instead of longs is
extremely small compared to the overall gain from per-cpu operation.
This patch makes percpu_ref based on longs so that it can be used to
directly count memory objects or pages.
Signed-off-by: Tejun Heo <tj@...nel.org>
Cc: Kent Overstreet <kmo@...erainc.com>
Cc: Johannes Weiner <hannes@...xchg.org>
---
include/linux/percpu-refcount.h | 24 ++++++++++++------------
lib/percpu-refcount.c | 33 +++++++++++++++++----------------
2 files changed, 29 insertions(+), 28 deletions(-)
diff --git a/include/linux/percpu-refcount.h b/include/linux/percpu-refcount.h
index 97a7d2a..4c55ad4 100644
--- a/include/linux/percpu-refcount.h
+++ b/include/linux/percpu-refcount.h
@@ -55,7 +55,7 @@ struct percpu_ref;
typedef void (percpu_ref_func_t)(struct percpu_ref *);
struct percpu_ref {
- atomic_t count;
+ atomic_long_t count;
/*
* The low bit of the pointer indicates whether the ref is in percpu
* mode; if set, then get/put will manipulate the atomic_t.
@@ -98,7 +98,7 @@ static inline void percpu_ref_kill(struct percpu_ref *ref)
* branches as it can't assume that @ref->pcpu_count is not NULL.
*/
static inline bool __pcpu_ref_alive(struct percpu_ref *ref,
- unsigned __percpu **pcpu_countp)
+ unsigned long __percpu **pcpu_countp)
{
unsigned long pcpu_ptr = ACCESS_ONCE(ref->pcpu_count_ptr);
@@ -108,7 +108,7 @@ static inline bool __pcpu_ref_alive(struct percpu_ref *ref,
if (unlikely(pcpu_ptr & PCPU_REF_DEAD))
return false;
- *pcpu_countp = (unsigned __percpu *)pcpu_ptr;
+ *pcpu_countp = (unsigned long __percpu *)pcpu_ptr;
return true;
}
@@ -120,14 +120,14 @@ static inline bool __pcpu_ref_alive(struct percpu_ref *ref,
*/
static inline void percpu_ref_get(struct percpu_ref *ref)
{
- unsigned __percpu *pcpu_count;
+ unsigned long __percpu *pcpu_count;
rcu_read_lock_sched();
if (__pcpu_ref_alive(ref, &pcpu_count))
this_cpu_inc(*pcpu_count);
else
- atomic_inc(&ref->count);
+ atomic_long_inc(&ref->count);
rcu_read_unlock_sched();
}
@@ -143,7 +143,7 @@ static inline void percpu_ref_get(struct percpu_ref *ref)
*/
static inline bool percpu_ref_tryget(struct percpu_ref *ref)
{
- unsigned __percpu *pcpu_count;
+ unsigned long __percpu *pcpu_count;
int ret = false;
rcu_read_lock_sched();
@@ -152,7 +152,7 @@ static inline bool percpu_ref_tryget(struct percpu_ref *ref)
this_cpu_inc(*pcpu_count);
ret = true;
} else {
- ret = atomic_inc_not_zero(&ref->count);
+ ret = atomic_long_inc_not_zero(&ref->count);
}
rcu_read_unlock_sched();
@@ -176,7 +176,7 @@ static inline bool percpu_ref_tryget(struct percpu_ref *ref)
*/
static inline bool percpu_ref_tryget_live(struct percpu_ref *ref)
{
- unsigned __percpu *pcpu_count;
+ unsigned long __percpu *pcpu_count;
int ret = false;
rcu_read_lock_sched();
@@ -200,13 +200,13 @@ static inline bool percpu_ref_tryget_live(struct percpu_ref *ref)
*/
static inline void percpu_ref_put(struct percpu_ref *ref)
{
- unsigned __percpu *pcpu_count;
+ unsigned long __percpu *pcpu_count;
rcu_read_lock_sched();
if (__pcpu_ref_alive(ref, &pcpu_count))
this_cpu_dec(*pcpu_count);
- else if (unlikely(atomic_dec_and_test(&ref->count)))
+ else if (unlikely(atomic_long_dec_and_test(&ref->count)))
ref->release(ref);
rcu_read_unlock_sched();
@@ -220,11 +220,11 @@ static inline void percpu_ref_put(struct percpu_ref *ref)
*/
static inline bool percpu_ref_is_zero(struct percpu_ref *ref)
{
- unsigned __percpu *pcpu_count;
+ unsigned long __percpu *pcpu_count;
if (__pcpu_ref_alive(ref, &pcpu_count))
return false;
- return !atomic_read(&ref->count);
+ return !atomic_long_read(&ref->count);
}
#endif
diff --git a/lib/percpu-refcount.c b/lib/percpu-refcount.c
index a6768f6..2c42855 100644
--- a/lib/percpu-refcount.c
+++ b/lib/percpu-refcount.c
@@ -29,11 +29,11 @@
* can't hit 0 before we've added up all the percpu refs.
*/
-#define PCPU_COUNT_BIAS (1U << 31)
+#define PCPU_COUNT_BIAS (1LU << (BITS_PER_LONG - 1))
-static unsigned __percpu *pcpu_count_ptr(struct percpu_ref *ref)
+static unsigned long __percpu *pcpu_count_ptr(struct percpu_ref *ref)
{
- return (unsigned __percpu *)(ref->pcpu_count_ptr & ~PCPU_REF_DEAD);
+ return (unsigned long __percpu *)(ref->pcpu_count_ptr & ~PCPU_REF_DEAD);
}
/**
@@ -51,9 +51,9 @@ static unsigned __percpu *pcpu_count_ptr(struct percpu_ref *ref)
int percpu_ref_init(struct percpu_ref *ref, percpu_ref_func_t *release,
gfp_t gfp)
{
- atomic_set(&ref->count, 1 + PCPU_COUNT_BIAS);
+ atomic_long_set(&ref->count, 1 + PCPU_COUNT_BIAS);
- ref->pcpu_count_ptr = (unsigned long)alloc_percpu_gfp(unsigned, gfp);
+ ref->pcpu_count_ptr = (unsigned long)alloc_percpu_gfp(unsigned long, gfp);
if (!ref->pcpu_count_ptr)
return -ENOMEM;
@@ -75,13 +75,13 @@ EXPORT_SYMBOL_GPL(percpu_ref_init);
*/
void percpu_ref_reinit(struct percpu_ref *ref)
{
- unsigned __percpu *pcpu_count = pcpu_count_ptr(ref);
+ unsigned long __percpu *pcpu_count = pcpu_count_ptr(ref);
int cpu;
BUG_ON(!pcpu_count);
WARN_ON(!percpu_ref_is_zero(ref));
- atomic_set(&ref->count, 1 + PCPU_COUNT_BIAS);
+ atomic_long_set(&ref->count, 1 + PCPU_COUNT_BIAS);
/*
* Restore per-cpu operation. smp_store_release() is paired with
@@ -108,7 +108,7 @@ EXPORT_SYMBOL_GPL(percpu_ref_reinit);
void percpu_ref_set_killed(struct percpu_ref *ref)
{
ref->pcpu_count_ptr |= PCPU_REF_DEAD;
- atomic_set(&ref->count, 0);
+ atomic_long_set(&ref->count, 0);
}
/**
@@ -123,7 +123,7 @@ void percpu_ref_set_killed(struct percpu_ref *ref)
*/
void percpu_ref_exit(struct percpu_ref *ref)
{
- unsigned __percpu *pcpu_count = pcpu_count_ptr(ref);
+ unsigned long __percpu *pcpu_count = pcpu_count_ptr(ref);
if (pcpu_count) {
free_percpu(pcpu_count);
@@ -135,14 +135,15 @@ EXPORT_SYMBOL_GPL(percpu_ref_exit);
static void percpu_ref_kill_rcu(struct rcu_head *rcu)
{
struct percpu_ref *ref = container_of(rcu, struct percpu_ref, rcu);
- unsigned __percpu *pcpu_count = pcpu_count_ptr(ref);
- unsigned count = 0;
+ unsigned long __percpu *pcpu_count = pcpu_count_ptr(ref);
+ unsigned long count = 0;
int cpu;
for_each_possible_cpu(cpu)
count += *per_cpu_ptr(pcpu_count, cpu);
- pr_debug("global %i pcpu %i", atomic_read(&ref->count), (int) count);
+ pr_debug("global %ld pcpu %ld",
+ atomic_long_read(&ref->count), (long)count);
/*
* It's crucial that we sum the percpu counters _before_ adding the sum
@@ -157,11 +158,11 @@ static void percpu_ref_kill_rcu(struct rcu_head *rcu)
* time is equivalent and saves us atomic operations:
*/
- atomic_add((int) count - PCPU_COUNT_BIAS, &ref->count);
+ atomic_long_add((long)count - PCPU_COUNT_BIAS, &ref->count);
- WARN_ONCE(atomic_read(&ref->count) <= 0,
- "percpu ref (%pf) <= 0 (%i) after killed",
- ref->release, atomic_read(&ref->count));
+ WARN_ONCE(atomic_long_read(&ref->count) <= 0,
+ "percpu ref (%pf) <= 0 (%ld) after killed",
+ ref->release, atomic_long_read(&ref->count));
/* @ref is viewed as dead on all CPUs, send out kill confirmation */
if (ref->confirm_kill)
--
1.9.3
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
Powered by blists - more mailing lists