[PATCH v2 2/4] percpu_stats: Enable 64-bit counts in 32-bit architectures
From: Waiman Long
Date: Fri Apr 08 2016 - 12:17:50 EST
The unsigned long type in 32-bit architectures is only 32-bit. This may
not be enough from some statistics counts that may well go over 2^32
over time. This patch optionally enables the use of 64-bit counts in
32-bit architecture, though it does add a bit of performance overhead
if enabled.
This patch adds a flags argument to the percpu_stats_init() function:
int percpu_stats_init(struct percpu_stats *pcs, int num, int flags)
Currently, the following 2 flags are supported:
1) PCPU_STAT_64BIT - enable 64-bit counts
2) PCPU_STAT_INTSAFE - make the 64-bit count update interrupt safe
The second flag isn't active if the first flag is not set.
Signed-off-by: Waiman Long <Waiman.Long@xxxxxxx>
---
include/linux/percpu_stats.h | 28 +++++++++++-
lib/percpu_stats.c | 96 +++++++++++++++++++++++++++++++++++-------
2 files changed, 105 insertions(+), 19 deletions(-)
diff --git a/include/linux/percpu_stats.h b/include/linux/percpu_stats.h
index ed6e8ac..641f211 100644
--- a/include/linux/percpu_stats.h
+++ b/include/linux/percpu_stats.h
@@ -6,15 +6,34 @@
*/
#include <linux/percpu.h>
#include <linux/types.h>
+#include <linux/u64_stats_sync.h>
+
+/*
+ * Supported flags for percpu_stats_init()
+ */
+#define PCPU_STAT_64BIT 1 /* Use 64-bit statistics count */
+#define PCPU_STAT_INTSAFE 2 /* Make percpu_add interrupt safe */
struct percpu_stats {
- unsigned long __percpu *stats;
+ union {
+ unsigned long __percpu *stats;
+ uint64_t __percpu *stats64;
+ };
+ struct u64_stats_sync sync;
int nstats; /* Number of statistics counts in stats array */
+ int flags;
};
extern void percpu_stats_destroy(struct percpu_stats *pcs);
-extern int percpu_stats_init(struct percpu_stats *pcs, int num);
+extern int percpu_stats_init(struct percpu_stats *pcs, int num, int flags);
extern uint64_t percpu_stats_sum(struct percpu_stats *pcs, int stat);
+extern void __percpu_stats_add(struct percpu_stats *pcs, int stat, int cnt);
+
+#ifdef CONFIG_64BIT
+#define PERCPU_STATS_FLAGS(pcs) false
+#else
+#define PERCPU_STATS_FLAGS(pcs) ((pcs)->flags)
+#endif
/**
* percpu_stats_add - Add the given value to a statistics count
@@ -26,7 +45,10 @@ static inline void
percpu_stats_add(struct percpu_stats *pcs, int stat, int cnt)
{
BUG_ON((unsigned int)stat >= pcs->nstats);
- this_cpu_add(pcs->stats[stat], cnt);
+ if (unlikely(PERCPU_STATS_FLAGS(pcs)))
+ __percpu_stats_add(pcs, stat, cnt);
+ else
+ this_cpu_add(pcs->stats[stat], cnt);
}
static inline void percpu_stats_inc(struct percpu_stats *pcs, int stat)
diff --git a/lib/percpu_stats.c b/lib/percpu_stats.c
index bc9f26d..2ec739e 100644
--- a/lib/percpu_stats.c
+++ b/lib/percpu_stats.c
@@ -5,29 +5,47 @@
#include <linux/percpu_stats.h>
#include <linux/bug.h>
+#ifdef CONFIG_64BIT
+/*
+ * Ignore PCPU_STAT_64BIT & PCPU_STAT_INTSAFE flags for 64-bit architectures
+ * as 64-bit count is the default.
+ */
+#define IS_STATS64(pcs) false
+#define GET_FLAGS(f) ((f) & ~(PCPU_STAT_64BIT | PCPU_STAT_INTSAFE))
+#else
+#define IS_STATS64(pcs) ((pcs)->flags & PCPU_STAT_64BIT)
+#define GET_FLAGS(f) (f)
+#endif
+
/**
* percpu_stats_init - allocate memory for the percpu statistics counts
- * @pcs: Pointer to percpu_stats structure
- * @num: Number of statistics counts to be used
+ * @pcs : Pointer to percpu_stats structure
+ * @num : Number of statistics counts to be used
+ * @flags: Optional feature bits
* Return: 0 if successful, -ENOMEM if memory allocation fails.
*/
-int percpu_stats_init(struct percpu_stats *pcs, int num)
+int percpu_stats_init(struct percpu_stats *pcs, int num, int flags)
{
- int cpu;
+ int cpu, size;
+ pcs->flags = GET_FLAGS(flags);
pcs->nstats = num;
- pcs->stats = __alloc_percpu(sizeof(unsigned long) * num,
- __alignof__(unsigned long));
- if (!pcs->stats)
- return -ENOMEM;
+ if (IS_STATS64(pcs)) {
+ size = sizeof(uint64_t) * num;
+ pcs->stats64 = __alloc_percpu(size, __alignof__(uint64_t));
+ if (!pcs->stats64)
+ return -ENOMEM;
+ u64_stats_init(&pcs->sync);
+ } else {
+ size = sizeof(unsigned long) * num;
+ pcs->stats = __alloc_percpu(size, __alignof__(unsigned long));
+ if (!pcs->stats)
+ return -ENOMEM;
+ }
- for_each_possible_cpu(cpu) {
- unsigned long *pstats = per_cpu_ptr(pcs->stats, cpu);
- int stat;
+ for_each_possible_cpu(cpu)
+ memset(per_cpu_ptr(pcs->stats, cpu), 0, size);
- for (stat = 0; stat < pcs->nstats; stat++, pstats++)
- *pstats = 0;
- }
return 0;
}
EXPORT_SYMBOL(percpu_stats_init);
@@ -57,8 +75,54 @@ uint64_t percpu_stats_sum(struct percpu_stats *pcs, int stat)
BUG_ON((unsigned int)stat >= pcs->nstats);
- for_each_possible_cpu(cpu)
- sum += per_cpu(pcs->stats[stat], cpu);
+ if (IS_STATS64(pcs)) {
+ for_each_possible_cpu(cpu) {
+ uint64_t val;
+ unsigned int seq;
+
+ do {
+ seq = u64_stats_fetch_begin(&pcs->sync);
+ val = per_cpu(pcs->stats64[stat], cpu);
+ } while (u64_stats_fetch_retry(&pcs->sync, seq));
+ sum += val;
+ }
+ } else {
+ for_each_possible_cpu(cpu)
+ sum += per_cpu(pcs->stats[stat], cpu);
+ }
return sum;
}
EXPORT_SYMBOL(percpu_stats_sum);
+
+/**
+ * __percpu_stats_add - add given count to percpu value
+ * @pcs : Pointer to percpu_stats structure
+ * @stat: The statistics count that needs to be updated
+ * @cnt: The value to be added to the statistics count
+ */
+void __percpu_stats_add(struct percpu_stats *pcs, int stat, int cnt)
+{
+ /*
+ * u64_stats_update_begin/u64_stats_update_end alone are not safe
+ * against recursive add on the same CPU caused by interrupt.
+ * So we need to set the PCPU_STAT_INTSAFE flag if this is required.
+ */
+ if (IS_STATS64(pcs)) {
+ uint64_t *pstats64;
+ unsigned long flags;
+
+ pstats64 = get_cpu_ptr(pcs->stats64);
+ if (pcs->flags & PCPU_STAT_INTSAFE)
+ local_irq_save(flags);
+
+ u64_stats_update_begin(&pcs->sync);
+ pstats64[stat] += cnt;
+ u64_stats_update_end(&pcs->sync);
+
+ if (pcs->flags & PCPU_STAT_INTSAFE)
+ local_irq_restore(flags);
+
+ put_cpu_ptr(pcs->stats64);
+ }
+}
+EXPORT_SYMBOL(__percpu_stats_add);
--
1.7.1