* Christoph Lameter <christoph@lameter.com>
*/
-#include <linux/config.h>
#include <linux/mm.h>
#include <linux/module.h>
-
-void __get_zone_counts(unsigned long *active, unsigned long *inactive,
- unsigned long *free, struct pglist_data *pgdat)
-{
- struct zone *zones = pgdat->node_zones;
- int i;
-
- *active = 0;
- *inactive = 0;
- *free = 0;
- for (i = 0; i < MAX_NR_ZONES; i++) {
- *active += zones[i].nr_active;
- *inactive += zones[i].nr_inactive;
- *free += zones[i].free_pages;
- }
-}
-
-void get_zone_counts(unsigned long *active,
- unsigned long *inactive, unsigned long *free)
-{
- struct pglist_data *pgdat;
-
- *active = 0;
- *inactive = 0;
- *free = 0;
- for_each_online_pgdat(pgdat) {
- unsigned long l, m, n;
- __get_zone_counts(&l, &m, &n, pgdat);
- *active += l;
- *inactive += m;
- *free += n;
- }
-}
+#include <linux/cpu.h>
+#include <linux/sched.h>
#ifdef CONFIG_VM_EVENT_COUNTERS
DEFINE_PER_CPU(struct vm_event_state, vm_event_states) = {{0}};
{
sum_vm_events(ret, &cpu_online_map);
}
+EXPORT_SYMBOL_GPL(all_vm_events);
#ifdef CONFIG_HOTPLUG
/*
#ifdef CONFIG_SMP
-#define STAT_THRESHOLD 32
+static int calculate_threshold(struct zone *zone)
+{
+ int threshold;
+ int mem; /* memory in 128 MB units */
+
+ /*
+ * The threshold scales with the number of processors and the amount
+ * of memory per zone. More memory means that we can defer updates for
+ * longer, more processors could lead to more contention.
+ * fls() is used to have a cheap way of logarithmic scaling.
+ *
+ * Some sample thresholds:
+ *
+ * Threshold Processors (fls) Zonesize fls(mem+1)
+ * ------------------------------------------------------------------
+ * 8 1 1 0.9-1 GB 4
+ * 16 2 2 0.9-1 GB 4
+ * 20 2 2 1-2 GB 5
+ * 24 2 2 2-4 GB 6
+ * 28 2 2 4-8 GB 7
+ * 32 2 2 8-16 GB 8
+ * 4 2 2 <128M 1
+ * 30 4 3 2-4 GB 5
+ * 48 4 3 8-16 GB 8
+ * 32 8 4 1-2 GB 4
+ * 32 8 4 0.9-1GB 4
+ * 10 16 5 <128M 1
+ * 40 16 5 900M 4
+ * 70 64 7 2-4 GB 5
+ * 84 64 7 4-8 GB 6
+ * 108 512 9 4-8 GB 6
+ * 125 1024 10 8-16 GB 8
+ * 125 1024 10 16-32 GB 9
+ */
+
+ mem = zone->present_pages >> (27 - PAGE_SHIFT);
+
+ threshold = 2 * fls(num_online_cpus()) * (1 + fls(mem));
+
+ /*
+ * Maximum threshold is 125
+ */
+ threshold = min(125, threshold);
+
+ return threshold;
+}
/*
- * Determine pointer to currently valid differential byte given a zone and
- * the item number.
- *
- * Preemption must be off
+ * Refresh the thresholds for each zone.
*/
-static inline s8 *diff_pointer(struct zone *zone, enum zone_stat_item item)
+static void refresh_zone_stat_thresholds(void)
{
- return &zone_pcp(zone, smp_processor_id())->vm_stat_diff[item];
+ struct zone *zone;
+ int cpu;
+ int threshold;
+
+ for_each_zone(zone) {
+
+ if (!zone->present_pages)
+ continue;
+
+ threshold = calculate_threshold(zone);
+
+ for_each_online_cpu(cpu)
+ zone_pcp(zone, cpu)->stat_threshold = threshold;
+ }
}
/*
void __mod_zone_page_state(struct zone *zone, enum zone_stat_item item,
int delta)
{
- s8 *p;
+ struct per_cpu_pageset *pcp = zone_pcp(zone, smp_processor_id());
+ s8 *p = pcp->vm_stat_diff + item;
long x;
- p = diff_pointer(zone, item);
x = delta + *p;
- if (unlikely(x > STAT_THRESHOLD || x < -STAT_THRESHOLD)) {
+ if (unlikely(x > pcp->stat_threshold || x < -pcp->stat_threshold)) {
zone_page_state_add(x, zone, item);
x = 0;
}
-
*p = x;
}
EXPORT_SYMBOL(__mod_zone_page_state);
* No overflow check is necessary and therefore the differential can be
* incremented or decremented in place which may allow the compilers to
* generate better code.
- *
* The increment or decrement is known and therefore one boundary check can
* be omitted.
*
+ * NOTE: These functions are very performance sensitive. Change only
+ * with care.
+ *
* Some processors have inc/dec instructions that are atomic vs an interrupt.
* However, the code must first determine the differential location in a zone
* based on the processor number and then inc/dec the counter. There is no
* in between and therefore the atomicity vs. interrupt cannot be exploited
* in a useful way here.
*/
-static void __inc_zone_state(struct zone *zone, enum zone_stat_item item)
+void __inc_zone_state(struct zone *zone, enum zone_stat_item item)
{
- s8 *p = diff_pointer(zone, item);
+ struct per_cpu_pageset *pcp = zone_pcp(zone, smp_processor_id());
+ s8 *p = pcp->vm_stat_diff + item;
(*p)++;
- if (unlikely(*p > STAT_THRESHOLD)) {
- zone_page_state_add(*p, zone, item);
- *p = 0;
+ if (unlikely(*p > pcp->stat_threshold)) {
+ int overstep = pcp->stat_threshold / 2;
+
+ zone_page_state_add(*p + overstep, zone, item);
+ *p = -overstep;
}
}
}
EXPORT_SYMBOL(__inc_zone_page_state);
-void __dec_zone_page_state(struct page *page, enum zone_stat_item item)
+void __dec_zone_state(struct zone *zone, enum zone_stat_item item)
{
- struct zone *zone = page_zone(page);
- s8 *p = diff_pointer(zone, item);
+ struct per_cpu_pageset *pcp = zone_pcp(zone, smp_processor_id());
+ s8 *p = pcp->vm_stat_diff + item;
(*p)--;
- if (unlikely(*p < -STAT_THRESHOLD)) {
- zone_page_state_add(*p, zone, item);
- *p = 0;
+ if (unlikely(*p < - pcp->stat_threshold)) {
+ int overstep = pcp->stat_threshold / 2;
+
+ zone_page_state_add(*p - overstep, zone, item);
+ *p = overstep;
}
}
+
+void __dec_zone_page_state(struct page *page, enum zone_stat_item item)
+{
+ __dec_zone_state(page_zone(page), item);
+}
EXPORT_SYMBOL(__dec_zone_page_state);
void inc_zone_state(struct zone *zone, enum zone_stat_item item)
void dec_zone_page_state(struct page *page, enum zone_stat_item item)
{
unsigned long flags;
- struct zone *zone;
- s8 *p;
- zone = page_zone(page);
local_irq_save(flags);
- p = diff_pointer(zone, item);
-
- (*p)--;
-
- if (unlikely(*p < -STAT_THRESHOLD)) {
- zone_page_state_add(*p, zone, item);
- *p = 0;
- }
+ __dec_zone_page_state(page, item);
local_irq_restore(flags);
}
EXPORT_SYMBOL(dec_zone_page_state);
/*
* Update the zone counters for one cpu.
+ *
+ * Note that refresh_cpu_vm_stats strives to only access
+ * node local memory. The per cpu pagesets on remote zones are placed
+ * in the memory local to the processor using that pageset. So the
+ * loop over all zones will access a series of cachelines local to
+ * the processor.
+ *
+ * The call to zone_page_state_add updates the cachelines with the
+ * statistics in the remote zone struct as well as the global cachelines
+ * with the global counters. These could cause remote node cache line
+ * bouncing and will have to be only done when necessary.
*/
void refresh_cpu_vm_stats(int cpu)
{
unsigned long flags;
for_each_zone(zone) {
- struct per_cpu_pageset *pcp;
+ struct per_cpu_pageset *p;
+
+ if (!populated_zone(zone))
+ continue;
- pcp = zone_pcp(zone, cpu);
+ p = zone_pcp(zone, cpu);
for (i = 0; i < NR_VM_ZONE_STAT_ITEMS; i++)
- if (pcp->vm_stat_diff[i]) {
+ if (p->vm_stat_diff[i]) {
local_irq_save(flags);
- zone_page_state_add(pcp->vm_stat_diff[i],
+ zone_page_state_add(p->vm_stat_diff[i],
zone, i);
- pcp->vm_stat_diff[i] = 0;
+ p->vm_stat_diff[i] = 0;
+#ifdef CONFIG_NUMA
+ /* 3 seconds idle till flush */
+ p->expire = 3;
+#endif
local_irq_restore(flags);
}
+#ifdef CONFIG_NUMA
+ /*
+ * Deal with draining the remote pageset of this
+ * processor
+ *
+ * Check if there are pages remaining in this pageset
+ * if not then there is nothing to expire.
+ */
+ if (!p->expire || (!p->pcp[0].count && !p->pcp[1].count))
+ continue;
+
+ /*
+ * We never drain zones local to this processor.
+ */
+ if (zone_to_nid(zone) == numa_node_id()) {
+ p->expire = 0;
+ continue;
+ }
+
+ p->expire--;
+ if (p->expire)
+ continue;
+
+ if (p->pcp[0].count)
+ drain_zone_pages(zone, p->pcp + 0);
+
+ if (p->pcp[1].count)
+ drain_zone_pages(zone, p->pcp + 1);
+#endif
}
}
__inc_zone_state(z, NUMA_MISS);
__inc_zone_state(zonelist->zones[0], NUMA_FOREIGN);
}
- if (z->zone_pgdat == NODE_DATA(numa_node_id()))
+ if (z->node == numa_node_id())
__inc_zone_state(z, NUMA_LOCAL);
else
__inc_zone_state(z, NUMA_OTHER);
return 0;
}
-struct seq_operations fragmentation_op = {
+const struct seq_operations fragmentation_op = {
.start = frag_start,
.next = frag_next,
.stop = frag_stop,
.show = frag_show,
};
-static char *vmstat_text[] = {
+#ifdef CONFIG_ZONE_DMA
+#define TEXT_FOR_DMA(xx) xx "_dma",
+#else
+#define TEXT_FOR_DMA(xx)
+#endif
+
+#ifdef CONFIG_ZONE_DMA32
+#define TEXT_FOR_DMA32(xx) xx "_dma32",
+#else
+#define TEXT_FOR_DMA32(xx)
+#endif
+
+#ifdef CONFIG_HIGHMEM
+#define TEXT_FOR_HIGHMEM(xx) xx "_high",
+#else
+#define TEXT_FOR_HIGHMEM(xx)
+#endif
+
+#define TEXTS_FOR_ZONES(xx) TEXT_FOR_DMA(xx) TEXT_FOR_DMA32(xx) xx "_normal", \
+ TEXT_FOR_HIGHMEM(xx) xx "_movable",
+
+static const char * const vmstat_text[] = {
/* Zoned VM counters */
+ "nr_free_pages",
+ "nr_inactive",
+ "nr_active",
"nr_anon_pages",
"nr_mapped",
"nr_file_pages",
- "nr_slab",
- "nr_page_table_pages",
"nr_dirty",
"nr_writeback",
+ "nr_slab_reclaimable",
+ "nr_slab_unreclaimable",
+ "nr_page_table_pages",
"nr_unstable",
"nr_bounce",
+ "nr_vmscan_write",
#ifdef CONFIG_NUMA
"numa_hit",
"pswpin",
"pswpout",
- "pgalloc_dma",
- "pgalloc_dma32",
- "pgalloc_normal",
- "pgalloc_high",
+ TEXTS_FOR_ZONES("pgalloc")
"pgfree",
"pgactivate",
"pgfault",
"pgmajfault",
- "pgrefill_dma",
- "pgrefill_dma32",
- "pgrefill_normal",
- "pgrefill_high",
-
- "pgsteal_dma",
- "pgsteal_dma32",
- "pgsteal_normal",
- "pgsteal_high",
-
- "pgscan_kswapd_dma",
- "pgscan_kswapd_dma32",
- "pgscan_kswapd_normal",
- "pgscan_kswapd_high",
-
- "pgscan_direct_dma",
- "pgscan_direct_dma32",
- "pgscan_direct_normal",
- "pgscan_direct_high",
+ TEXTS_FOR_ZONES("pgrefill")
+ TEXTS_FOR_ZONES("pgsteal")
+ TEXTS_FOR_ZONES("pgscan_kswapd")
+ TEXTS_FOR_ZONES("pgscan_direct")
"pginodesteal",
"slabs_scanned",
"\n min %lu"
"\n low %lu"
"\n high %lu"
- "\n active %lu"
- "\n inactive %lu"
"\n scanned %lu (a: %lu i: %lu)"
"\n spanned %lu"
"\n present %lu",
- zone->free_pages,
+ zone_page_state(zone, NR_FREE_PAGES),
zone->pages_min,
zone->pages_low,
zone->pages_high,
- zone->nr_active,
- zone->nr_inactive,
zone->pages_scanned,
zone->nr_scan_active, zone->nr_scan_inactive,
zone->spanned_pages,
pageset = zone_pcp(zone, i);
for (j = 0; j < ARRAY_SIZE(pageset->pcp); j++) {
- if (pageset->pcp[j].count)
- break;
- }
- if (j == ARRAY_SIZE(pageset->pcp))
- continue;
- for (j = 0; j < ARRAY_SIZE(pageset->pcp); j++) {
seq_printf(m,
"\n cpu: %i pcp: %i"
"\n count: %i"
pageset->pcp[j].high,
pageset->pcp[j].batch);
}
+#ifdef CONFIG_SMP
+ seq_printf(m, "\n vm stats threshold: %d",
+ pageset->stat_threshold);
+#endif
}
seq_printf(m,
"\n all_unreclaimable: %u"
"\n prev_priority: %i"
- "\n temp_priority: %i"
"\n start_pfn: %lu",
zone->all_unreclaimable,
zone->prev_priority,
- zone->temp_priority,
zone->zone_start_pfn);
spin_unlock_irqrestore(&zone->lock, flags);
seq_putc(m, '\n');
return 0;
}
-struct seq_operations zoneinfo_op = {
+const struct seq_operations zoneinfo_op = {
.start = frag_start, /* iterate over all zones. The same as in
* fragmentation. */
.next = frag_next,
m->private = NULL;
}
-struct seq_operations vmstat_op = {
+const struct seq_operations vmstat_op = {
.start = vmstat_start,
.next = vmstat_next,
.stop = vmstat_stop,
#endif /* CONFIG_PROC_FS */
+#ifdef CONFIG_SMP
+static DEFINE_PER_CPU(struct delayed_work, vmstat_work);
+int sysctl_stat_interval __read_mostly = HZ;
+
+static void vmstat_update(struct work_struct *w)
+{
+ refresh_cpu_vm_stats(smp_processor_id());
+ schedule_delayed_work(&__get_cpu_var(vmstat_work),
+ sysctl_stat_interval);
+}
+
+static void __devinit start_cpu_timer(int cpu)
+{
+ struct delayed_work *vmstat_work = &per_cpu(vmstat_work, cpu);
+
+ INIT_DELAYED_WORK_DEFERRABLE(vmstat_work, vmstat_update);
+ schedule_delayed_work_on(cpu, vmstat_work, HZ + cpu);
+}
+
+/*
+ * Use the cpu notifier to insure that the thresholds are recalculated
+ * when necessary.
+ */
+static int __cpuinit vmstat_cpuup_callback(struct notifier_block *nfb,
+ unsigned long action,
+ void *hcpu)
+{
+ long cpu = (long)hcpu;
+
+ switch (action) {
+ case CPU_ONLINE:
+ case CPU_ONLINE_FROZEN:
+ start_cpu_timer(cpu);
+ break;
+ case CPU_DOWN_PREPARE:
+ case CPU_DOWN_PREPARE_FROZEN:
+ cancel_rearming_delayed_work(&per_cpu(vmstat_work, cpu));
+ per_cpu(vmstat_work, cpu).work.func = NULL;
+ break;
+ case CPU_DOWN_FAILED:
+ case CPU_DOWN_FAILED_FROZEN:
+ start_cpu_timer(cpu);
+ break;
+ case CPU_DEAD:
+ case CPU_DEAD_FROZEN:
+ refresh_zone_stat_thresholds();
+ break;
+ default:
+ break;
+ }
+ return NOTIFY_OK;
+}
+
+static struct notifier_block __cpuinitdata vmstat_notifier =
+ { &vmstat_cpuup_callback, NULL, 0 };
+
+int __init setup_vmstat(void)
+{
+ int cpu;
+
+ refresh_zone_stat_thresholds();
+ register_cpu_notifier(&vmstat_notifier);
+
+ for_each_online_cpu(cpu)
+ start_cpu_timer(cpu);
+ return 0;
+}
+module_init(setup_vmstat)
+#endif
git://ftp.safe.ca / safe / jmp / linux-2.6 / blobdiff