[PATCH] Light weight event counters
The remaining counters in page_state after the zoned VM counter patches
have been applied are all just for show in /proc/vmstat. They have no
essential function for the VM.
We use a simple increment of per cpu variables. In order to avoid the most
severe races we disable preempt. Preempt does not prevent the race between
an increment and an interrupt handler incrementing the same statistics
counter. However, that race is exceedingly rare, we may only loose one
increment or so and there is no requirement (at least not in kernel) that
the vm event counters have to be accurate.
In the non preempt case this results in a simple increment for each
counter. For many architectures this will be reduced by the compiler to a
single instruction. This single instruction is atomic for i386 and x86_64.
And therefore even the rare race condition in an interrupt is avoided for
both architectures in most cases.
The patchset also adds an off switch for embedded systems that allows a
building of linux kernels without these counters.
The implementation of these counters is through inline code that hopefully
results in only a single instruction increment instruction being emitted
(i386, x86_64) or in the increment being hidden though instruction
concurrency (EPIC architectures such as ia64 can get that done).
Benefits:
- VM event counter operations usually reduce to a single inline instruction
on i386 and x86_64.
- No interrupt disable, only preempt disable for the preempt case.
Preempt disable can also be avoided by moving the counter into a spinlock.
- Handling is similar to zoned VM counters.
- Simple and easily extendable.
- Can be omitted to reduce memory use for embedded use.
References:
RFC http://marc.theaimsgroup.com/?l=linux-kernel&m=113512330605497&w=2
RFC http://marc.theaimsgroup.com/?l=linux-kernel&m=114988082814934&w=2
local_t http://marc.theaimsgroup.com/?l=linux-kernel&m=114991748606690&w=2
V2 http://marc.theaimsgroup.com/?t=115014808400007&r=1&w=2
V3 http://marc.theaimsgroup.com/?l=linux-kernel&m=115024767022346&w=2
V4 http://marc.theaimsgroup.com/?l=linux-kernel&m=115047968808926&w=2
Signed-off-by: Christoph Lameter <clameter@sgi.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
diff --git a/arch/s390/appldata/appldata_base.c b/arch/s390/appldata/appldata_base.c
index 61bc446..2476ca7 100644
--- a/arch/s390/appldata/appldata_base.c
+++ b/arch/s390/appldata/appldata_base.c
@@ -766,7 +766,6 @@
#endif /* MODULE */
EXPORT_SYMBOL_GPL(si_swapinfo);
EXPORT_SYMBOL_GPL(nr_threads);
-EXPORT_SYMBOL_GPL(get_full_page_state);
EXPORT_SYMBOL_GPL(nr_running);
EXPORT_SYMBOL_GPL(nr_iowait);
//EXPORT_SYMBOL_GPL(nr_context_switches);
diff --git a/arch/s390/appldata/appldata_mem.c b/arch/s390/appldata/appldata_mem.c
index 180ba79..4811e2d 100644
--- a/arch/s390/appldata/appldata_mem.c
+++ b/arch/s390/appldata/appldata_mem.c
@@ -107,21 +107,21 @@
* serialized through the appldata_ops_lock and can use static
*/
static struct sysinfo val;
- static struct page_state ps;
+ unsigned long ev[NR_VM_EVENT_ITEMS];
struct appldata_mem_data *mem_data;
mem_data = data;
mem_data->sync_count_1++;
- get_full_page_state(&ps);
- mem_data->pgpgin = ps.pgpgin >> 1;
- mem_data->pgpgout = ps.pgpgout >> 1;
- mem_data->pswpin = ps.pswpin;
- mem_data->pswpout = ps.pswpout;
- mem_data->pgalloc = ps.pgalloc_high + ps.pgalloc_normal +
- ps.pgalloc_dma;
- mem_data->pgfault = ps.pgfault;
- mem_data->pgmajfault = ps.pgmajfault;
+ all_vm_events(ev);
+ mem_data->pgpgin = ev[PGPGIN] >> 1;
+ mem_data->pgpgout = ev[PGPGOUT] >> 1;
+ mem_data->pswpin = ev[PSWPIN];
+ mem_data->pswpout = ev[PSWPOUT];
+ mem_data->pgalloc = ev[PGALLOC_HIGH] + ev[PGALLOC_NORMAL] +
+ ev[PGALLOC_DMA];
+ mem_data->pgfault = ev[PGFAULT];
+ mem_data->pgmajfault = ev[PGMAJFAULT];
si_meminfo(&val);
mem_data->sharedram = val.sharedram;
diff --git a/block/ll_rw_blk.c b/block/ll_rw_blk.c
index eee03a3..fb83547 100644
--- a/block/ll_rw_blk.c
+++ b/block/ll_rw_blk.c
@@ -3117,9 +3117,9 @@
BIO_BUG_ON(!bio->bi_io_vec);
bio->bi_rw |= rw;
if (rw & WRITE)
- mod_page_state(pgpgout, count);
+ count_vm_events(PGPGOUT, count);
else
- mod_page_state(pgpgin, count);
+ count_vm_events(PGPGIN, count);
if (unlikely(block_dump)) {
char b[BDEVNAME_SIZE];
diff --git a/drivers/parisc/led.c b/drivers/parisc/led.c
index 298f2dd..d7024c7 100644
--- a/drivers/parisc/led.c
+++ b/drivers/parisc/led.c
@@ -411,16 +411,17 @@
static __inline__ int led_get_diskio_activity(void)
{
static unsigned long last_pgpgin, last_pgpgout;
- struct page_state pgstat;
+ unsigned long events[NR_VM_EVENT_ITEMS];
int changed;
- get_full_page_state(&pgstat); /* get no of sectors in & out */
+ all_vm_events(events);
/* Just use a very simple calculation here. Do not care about overflow,
since we only want to know if there was activity or not. */
- changed = (pgstat.pgpgin != last_pgpgin) || (pgstat.pgpgout != last_pgpgout);
- last_pgpgin = pgstat.pgpgin;
- last_pgpgout = pgstat.pgpgout;
+ changed = (events[PGPGIN] != last_pgpgin) ||
+ (events[PGPGOUT] != last_pgpgout);
+ last_pgpgin = events[PGPGIN];
+ last_pgpgout = events[PGPGOUT];
return (changed ? LED_DISK_IO : 0);
}
diff --git a/fs/inode.c b/fs/inode.c
index f42961e..14a6c41 100644
--- a/fs/inode.c
+++ b/fs/inode.c
@@ -452,15 +452,14 @@
nr_pruned++;
}
inodes_stat.nr_unused -= nr_pruned;
+ if (current_is_kswapd())
+ __count_vm_events(KSWAPD_INODESTEAL, reap);
+ else
+ __count_vm_events(PGINODESTEAL, reap);
spin_unlock(&inode_lock);
dispose_list(&freeable);
mutex_unlock(&iprune_mutex);
-
- if (current_is_kswapd())
- mod_page_state(kswapd_inodesteal, reap);
- else
- mod_page_state(pginodesteal, reap);
}
/*
diff --git a/fs/ncpfs/mmap.c b/fs/ncpfs/mmap.c
index 52d60c3..e7d5a30 100644
--- a/fs/ncpfs/mmap.c
+++ b/fs/ncpfs/mmap.c
@@ -93,7 +93,7 @@
*/
if (type)
*type = VM_FAULT_MAJOR;
- inc_page_state(pgmajfault);
+ count_vm_event(PGMAJFAULT);
return page;
}
diff --git a/include/linux/vmstat.h b/include/linux/vmstat.h
index 16173b6..3e0daf5 100644
--- a/include/linux/vmstat.h
+++ b/include/linux/vmstat.h
@@ -7,115 +7,77 @@
#include <linux/mmzone.h>
#include <asm/atomic.h>
+#ifdef CONFIG_VM_EVENT_COUNTERS
/*
- * Global page accounting. One instance per CPU. Only unsigned longs are
- * allowed.
+ * Light weight per cpu counter implementation.
*
- * - Fields can be modified with xxx_page_state and xxx_page_state_zone at
- * any time safely (which protects the instance from modification by
- * interrupt.
- * - The __xxx_page_state variants can be used safely when interrupts are
- * disabled.
- * - The __xxx_page_state variants can be used if the field is only
- * modified from process context and protected from preemption, or only
- * modified from interrupt context. In this case, the field should be
- * commented here.
+ * Counters should only be incremented and no critical kernel component
+ * should rely on the counter values.
+ *
+ * Counters are handled completely inline. On many platforms the code
+ * generated will simply be the increment of a global address.
*/
-struct page_state {
- unsigned long pgpgin; /* Disk reads */
- unsigned long pgpgout; /* Disk writes */
- unsigned long pswpin; /* swap reads */
- unsigned long pswpout; /* swap writes */
- unsigned long pgalloc_high; /* page allocations */
- unsigned long pgalloc_normal;
- unsigned long pgalloc_dma32;
- unsigned long pgalloc_dma;
+#define FOR_ALL_ZONES(x) x##_DMA, x##_DMA32, x##_NORMAL, x##_HIGH
- unsigned long pgfree; /* page freeings */
- unsigned long pgactivate; /* pages moved inactive->active */
- unsigned long pgdeactivate; /* pages moved active->inactive */
-
- unsigned long pgfault; /* faults (major+minor) */
- unsigned long pgmajfault; /* faults (major only) */
-
- unsigned long pgrefill_high; /* inspected in refill_inactive_zone */
- unsigned long pgrefill_normal;
- unsigned long pgrefill_dma32;
- unsigned long pgrefill_dma;
-
- unsigned long pgsteal_high; /* total highmem pages reclaimed */
- unsigned long pgsteal_normal;
- unsigned long pgsteal_dma32;
- unsigned long pgsteal_dma;
-
- unsigned long pgscan_kswapd_high;/* total highmem pages scanned */
- unsigned long pgscan_kswapd_normal;
- unsigned long pgscan_kswapd_dma32;
- unsigned long pgscan_kswapd_dma;
-
- unsigned long pgscan_direct_high;/* total highmem pages scanned */
- unsigned long pgscan_direct_normal;
- unsigned long pgscan_direct_dma32;
- unsigned long pgscan_direct_dma;
-
- unsigned long pginodesteal; /* pages reclaimed via inode freeing */
- unsigned long slabs_scanned; /* slab objects scanned */
- unsigned long kswapd_steal; /* pages reclaimed by kswapd */
- unsigned long kswapd_inodesteal;/* reclaimed via kswapd inode freeing */
- unsigned long pageoutrun; /* kswapd's calls to page reclaim */
- unsigned long allocstall; /* direct reclaim calls */
-
- unsigned long pgrotated; /* pages rotated to tail of the LRU */
+enum vm_event_item { PGPGIN, PGPGOUT, PSWPIN, PSWPOUT,
+ FOR_ALL_ZONES(PGALLOC),
+ PGFREE, PGACTIVATE, PGDEACTIVATE,
+ PGFAULT, PGMAJFAULT,
+ FOR_ALL_ZONES(PGREFILL),
+ FOR_ALL_ZONES(PGSTEAL),
+ FOR_ALL_ZONES(PGSCAN_KSWAPD),
+ FOR_ALL_ZONES(PGSCAN_DIRECT),
+ PGINODESTEAL, SLABS_SCANNED, KSWAPD_STEAL, KSWAPD_INODESTEAL,
+ PAGEOUTRUN, ALLOCSTALL, PGROTATED,
+ NR_VM_EVENT_ITEMS
};
-extern void get_full_page_state(struct page_state *ret);
-extern void mod_page_state_offset(unsigned long offset, unsigned long delta);
-extern void __mod_page_state_offset(unsigned long offset, unsigned long delta);
+struct vm_event_state {
+ unsigned long event[NR_VM_EVENT_ITEMS];
+};
-#define mod_page_state(member, delta) \
- mod_page_state_offset(offsetof(struct page_state, member), (delta))
+DECLARE_PER_CPU(struct vm_event_state, vm_event_states);
-#define __mod_page_state(member, delta) \
- __mod_page_state_offset(offsetof(struct page_state, member), (delta))
+static inline void __count_vm_event(enum vm_event_item item)
+{
+ __get_cpu_var(vm_event_states.event[item])++;
+}
-#define inc_page_state(member) mod_page_state(member, 1UL)
-#define dec_page_state(member) mod_page_state(member, 0UL - 1)
-#define add_page_state(member,delta) mod_page_state(member, (delta))
-#define sub_page_state(member,delta) mod_page_state(member, 0UL - (delta))
+static inline void count_vm_event(enum vm_event_item item)
+{
+ get_cpu_var(vm_event_states.event[item])++;
+ put_cpu();
+}
-#define __inc_page_state(member) __mod_page_state(member, 1UL)
-#define __dec_page_state(member) __mod_page_state(member, 0UL - 1)
-#define __add_page_state(member,delta) __mod_page_state(member, (delta))
-#define __sub_page_state(member,delta) __mod_page_state(member, 0UL - (delta))
+static inline void __count_vm_events(enum vm_event_item item, long delta)
+{
+ __get_cpu_var(vm_event_states.event[item]) += delta;
+}
-#define page_state(member) (*__page_state(offsetof(struct page_state, member)))
+static inline void count_vm_events(enum vm_event_item item, long delta)
+{
+ get_cpu_var(vm_event_states.event[item])++;
+ put_cpu();
+}
-#define state_zone_offset(zone, member) \
-({ \
- unsigned offset; \
- if (is_highmem(zone)) \
- offset = offsetof(struct page_state, member##_high); \
- else if (is_normal(zone)) \
- offset = offsetof(struct page_state, member##_normal); \
- else if (is_dma32(zone)) \
- offset = offsetof(struct page_state, member##_dma32); \
- else \
- offset = offsetof(struct page_state, member##_dma); \
- offset; \
-})
+extern void all_vm_events(unsigned long *);
+extern void vm_events_fold_cpu(int cpu);
-#define __mod_page_state_zone(zone, member, delta) \
- do { \
- __mod_page_state_offset(state_zone_offset(zone, member), (delta)); \
- } while (0)
+#else
-#define mod_page_state_zone(zone, member, delta) \
- do { \
- mod_page_state_offset(state_zone_offset(zone, member), (delta)); \
- } while (0)
+/* Disable counters */
+#define get_cpu_vm_events(e) 0L
+#define count_vm_event(e) do { } while (0)
+#define count_vm_events(e,d) do { } while (0)
+#define __count_vm_event(e) do { } while (0)
+#define __count_vm_events(e,d) do { } while (0)
+#define vm_events_fold_cpu(x) do { } while (0)
-DECLARE_PER_CPU(struct page_state, page_states);
+#endif /* CONFIG_VM_EVENT_COUNTERS */
+
+#define __count_zone_vm_events(item, zone, delta) \
+ __count_vm_events(item##_DMA + zone_idx(zone), delta)
/*
* Zone based page accounting with per cpu differentials.
diff --git a/init/Kconfig b/init/Kconfig
index f70f2fd..f515948 100644
--- a/init/Kconfig
+++ b/init/Kconfig
@@ -379,6 +379,15 @@
SLOB is more space efficient but does not scale well and is
more susceptible to fragmentation.
+config VM_EVENT_COUNTERS
+ default y
+ bool "Enable VM event counters for /proc/vmstat" if EMBEDDED
+ help
+ VM event counters are only needed to for event counts to be
+ shown. They have no function for the kernel itself. This
+ option allows the disabling of the VM event counters.
+ /proc/vmstat will only show page counts.
+
endmenu # General setup
config TINY_SHMEM
diff --git a/mm/filemap.c b/mm/filemap.c
index 87d62c4..796a547 100644
--- a/mm/filemap.c
+++ b/mm/filemap.c
@@ -1416,7 +1416,7 @@
*/
if (!did_readaround) {
majmin = VM_FAULT_MAJOR;
- inc_page_state(pgmajfault);
+ count_vm_event(PGMAJFAULT);
}
did_readaround = 1;
ra_pages = max_sane_readahead(file->f_ra.ra_pages);
@@ -1487,7 +1487,7 @@
page_not_uptodate:
if (!did_readaround) {
majmin = VM_FAULT_MAJOR;
- inc_page_state(pgmajfault);
+ count_vm_event(PGMAJFAULT);
}
lock_page(page);
diff --git a/mm/memory.c b/mm/memory.c
index 1a78791..7e2a4b1 100644
--- a/mm/memory.c
+++ b/mm/memory.c
@@ -1951,7 +1951,7 @@
/* Had to read the page from swap area: Major fault */
ret = VM_FAULT_MAJOR;
- inc_page_state(pgmajfault);
+ count_vm_event(PGMAJFAULT);
grab_swap_token();
}
@@ -2324,7 +2324,7 @@
__set_current_state(TASK_RUNNING);
- inc_page_state(pgfault);
+ count_vm_event(PGFAULT);
if (unlikely(is_vm_hugetlb_page(vma)))
return hugetlb_fault(mm, vma, address, write_access);
diff --git a/mm/page_alloc.c b/mm/page_alloc.c
index d616712..30b0b97 100644
--- a/mm/page_alloc.c
+++ b/mm/page_alloc.c
@@ -456,7 +456,7 @@
kernel_map_pages(page, 1 << order, 0);
local_irq_save(flags);
- __mod_page_state(pgfree, 1 << order);
+ __count_vm_events(PGFREE, 1 << order);
free_one_page(page_zone(page), page, order);
local_irq_restore(flags);
}
@@ -729,7 +729,7 @@
pcp = &zone_pcp(zone, get_cpu())->pcp[cold];
local_irq_save(flags);
- __inc_page_state(pgfree);
+ __count_vm_event(PGFREE);
list_add(&page->lru, &pcp->list);
pcp->count++;
if (pcp->count >= pcp->high) {
@@ -805,7 +805,7 @@
goto failed;
}
- __mod_page_state_zone(zone, pgalloc, 1 << order);
+ __count_zone_vm_events(PGALLOC, zone, 1 << order);
zone_statistics(zonelist, zone);
local_irq_restore(flags);
put_cpu();
@@ -2101,24 +2101,11 @@
unsigned long action, void *hcpu)
{
int cpu = (unsigned long)hcpu;
- unsigned long *src, *dest;
if (action == CPU_DEAD) {
- int i;
-
local_irq_disable();
__drain_pages(cpu);
-
- /* Add dead cpu's page_states to our own. */
- dest = (unsigned long *)&__get_cpu_var(page_states);
- src = (unsigned long *)&per_cpu(page_states, cpu);
-
- for (i = 0; i < sizeof(struct page_state)/sizeof(unsigned long);
- i++) {
- dest[i] += src[i];
- src[i] = 0;
- }
-
+ vm_events_fold_cpu(cpu);
local_irq_enable();
refresh_cpu_vm_stats(cpu);
}
diff --git a/mm/page_io.c b/mm/page_io.c
index bb2b0d5..8802994 100644
--- a/mm/page_io.c
+++ b/mm/page_io.c
@@ -101,7 +101,7 @@
}
if (wbc->sync_mode == WB_SYNC_ALL)
rw |= (1 << BIO_RW_SYNC);
- inc_page_state(pswpout);
+ count_vm_event(PSWPOUT);
set_page_writeback(page);
unlock_page(page);
submit_bio(rw, bio);
@@ -123,7 +123,7 @@
ret = -ENOMEM;
goto out;
}
- inc_page_state(pswpin);
+ count_vm_event(PSWPIN);
submit_bio(READ, bio);
out:
return ret;
diff --git a/mm/shmem.c b/mm/shmem.c
index b14ff81..a9c09e0 100644
--- a/mm/shmem.c
+++ b/mm/shmem.c
@@ -1045,12 +1045,12 @@
swappage = lookup_swap_cache(swap);
if (!swappage) {
shmem_swp_unmap(entry);
- spin_unlock(&info->lock);
/* here we actually do the io */
if (type && *type == VM_FAULT_MINOR) {
- inc_page_state(pgmajfault);
+ __count_vm_event(PGMAJFAULT);
*type = VM_FAULT_MAJOR;
}
+ spin_unlock(&info->lock);
swappage = shmem_swapin(info, swap, idx);
if (!swappage) {
spin_lock(&info->lock);
diff --git a/mm/swap.c b/mm/swap.c
index 990868a..8fd095c 100644
--- a/mm/swap.c
+++ b/mm/swap.c
@@ -87,7 +87,7 @@
spin_lock_irqsave(&zone->lru_lock, flags);
if (PageLRU(page) && !PageActive(page)) {
list_move_tail(&page->lru, &zone->inactive_list);
- inc_page_state(pgrotated);
+ __count_vm_event(PGROTATED);
}
if (!test_clear_page_writeback(page))
BUG();
@@ -107,7 +107,7 @@
del_page_from_inactive_list(zone, page);
SetPageActive(page);
add_page_to_active_list(zone, page);
- inc_page_state(pgactivate);
+ __count_vm_event(PGACTIVATE);
}
spin_unlock_irq(&zone->lru_lock);
}
diff --git a/mm/vmscan.c b/mm/vmscan.c
index d694243..ff2ebe9 100644
--- a/mm/vmscan.c
+++ b/mm/vmscan.c
@@ -215,7 +215,7 @@
break;
if (shrink_ret < nr_before)
ret += nr_before - shrink_ret;
- mod_page_state(slabs_scanned, this_scan);
+ count_vm_events(SLABS_SCANNED, this_scan);
total_scan -= this_scan;
cond_resched();
@@ -569,7 +569,7 @@
list_splice(&ret_pages, page_list);
if (pagevec_count(&freed_pvec))
__pagevec_release_nonlru(&freed_pvec);
- mod_page_state(pgactivate, pgactivate);
+ count_vm_events(PGACTIVATE, pgactivate);
return nr_reclaimed;
}
@@ -659,11 +659,11 @@
nr_reclaimed += nr_freed;
local_irq_disable();
if (current_is_kswapd()) {
- __mod_page_state_zone(zone, pgscan_kswapd, nr_scan);
- __mod_page_state(kswapd_steal, nr_freed);
+ __count_zone_vm_events(PGSCAN_KSWAPD, zone, nr_scan);
+ __count_vm_events(KSWAPD_STEAL, nr_freed);
} else
- __mod_page_state_zone(zone, pgscan_direct, nr_scan);
- __mod_page_state_zone(zone, pgsteal, nr_freed);
+ __count_zone_vm_events(PGSCAN_DIRECT, zone, nr_scan);
+ __count_vm_events(PGACTIVATE, nr_freed);
if (nr_taken == 0)
goto done;
@@ -841,11 +841,10 @@
}
}
zone->nr_active += pgmoved;
- spin_unlock(&zone->lru_lock);
- __mod_page_state_zone(zone, pgrefill, pgscanned);
- __mod_page_state(pgdeactivate, pgdeactivate);
- local_irq_enable();
+ __count_zone_vm_events(PGREFILL, zone, pgscanned);
+ __count_vm_events(PGDEACTIVATE, pgdeactivate);
+ spin_unlock_irq(&zone->lru_lock);
pagevec_release(&pvec);
}
@@ -977,7 +976,7 @@
.swappiness = vm_swappiness,
};
- inc_page_state(allocstall);
+ count_vm_event(ALLOCSTALL);
for (i = 0; zones[i] != NULL; i++) {
struct zone *zone = zones[i];
@@ -1074,7 +1073,7 @@
total_scanned = 0;
nr_reclaimed = 0;
sc.may_writepage = !laptop_mode;
- inc_page_state(pageoutrun);
+ count_vm_event(PAGEOUTRUN);
for (i = 0; i < pgdat->nr_zones; i++) {
struct zone *zone = pgdat->node_zones + i;
diff --git a/mm/vmstat.c b/mm/vmstat.c
index ee7f896..73b83d6 100644
--- a/mm/vmstat.c
+++ b/mm/vmstat.c
@@ -13,66 +13,6 @@
#include <linux/mm.h>
#include <linux/module.h>
-/*
- * Accumulate the page_state information across all CPUs.
- * The result is unavoidably approximate - it can change
- * during and after execution of this function.
- */
-DEFINE_PER_CPU(struct page_state, page_states) = {0};
-
-static void __get_page_state(struct page_state *ret, int nr, cpumask_t *cpumask)
-{
- unsigned cpu;
-
- memset(ret, 0, nr * sizeof(unsigned long));
- cpus_and(*cpumask, *cpumask, cpu_online_map);
-
- for_each_cpu_mask(cpu, *cpumask) {
- unsigned long *in;
- unsigned long *out;
- unsigned off;
- unsigned next_cpu;
-
- in = (unsigned long *)&per_cpu(page_states, cpu);
-
- next_cpu = next_cpu(cpu, *cpumask);
- if (likely(next_cpu < NR_CPUS))
- prefetch(&per_cpu(page_states, next_cpu));
-
- out = (unsigned long *)ret;
- for (off = 0; off < nr; off++)
- *out++ += *in++;
- }
-}
-
-void get_full_page_state(struct page_state *ret)
-{
- cpumask_t mask = CPU_MASK_ALL;
-
- __get_page_state(ret, sizeof(*ret) / sizeof(unsigned long), &mask);
-}
-
-void __mod_page_state_offset(unsigned long offset, unsigned long delta)
-{
- void *ptr;
-
- ptr = &__get_cpu_var(page_states);
- *(unsigned long *)(ptr + offset) += delta;
-}
-EXPORT_SYMBOL(__mod_page_state_offset);
-
-void mod_page_state_offset(unsigned long offset, unsigned long delta)
-{
- unsigned long flags;
- void *ptr;
-
- local_irq_save(flags);
- ptr = &__get_cpu_var(page_states);
- *(unsigned long *)(ptr + offset) += delta;
- local_irq_restore(flags);
-}
-EXPORT_SYMBOL(mod_page_state_offset);
-
void __get_zone_counts(unsigned long *active, unsigned long *inactive,
unsigned long *free, struct pglist_data *pgdat)
{
@@ -106,6 +46,63 @@
}
}
+#ifdef CONFIG_VM_EVENT_COUNTERS
+DEFINE_PER_CPU(struct vm_event_state, vm_event_states) = {{0}};
+EXPORT_PER_CPU_SYMBOL(vm_event_states);
+
+static void sum_vm_events(unsigned long *ret, cpumask_t *cpumask)
+{
+ int cpu = 0;
+ int i;
+
+ memset(ret, 0, NR_VM_EVENT_ITEMS * sizeof(unsigned long));
+
+ cpu = first_cpu(*cpumask);
+ while (cpu < NR_CPUS) {
+ struct vm_event_state *this = &per_cpu(vm_event_states, cpu);
+
+ cpu = next_cpu(cpu, *cpumask);
+
+ if (cpu < NR_CPUS)
+ prefetch(&per_cpu(vm_event_states, cpu));
+
+
+ for (i = 0; i < NR_VM_EVENT_ITEMS; i++)
+ ret[i] += this->event[i];
+ }
+}
+
+/*
+ * Accumulate the vm event counters across all CPUs.
+ * The result is unavoidably approximate - it can change
+ * during and after execution of this function.
+*/
+void all_vm_events(unsigned long *ret)
+{
+ sum_vm_events(ret, &cpu_online_map);
+}
+
+#ifdef CONFIG_HOTPLUG
+/*
+ * Fold the foreign cpu events into our own.
+ *
+ * This is adding to the events on one processor
+ * but keeps the global counts constant.
+ */
+void vm_events_fold_cpu(int cpu)
+{
+ struct vm_event_state *fold_state = &per_cpu(vm_event_states, cpu);
+ int i;
+
+ for (i = 0; i < NR_VM_EVENT_ITEMS; i++) {
+ count_vm_events(i, fold_state->event[i]);
+ fold_state->event[i] = 0;
+ }
+}
+#endif /* CONFIG_HOTPLUG */
+
+#endif /* CONFIG_VM_EVENT_COUNTERS */
+
/*
* Manage combined zone based / global counters
*
@@ -405,16 +402,16 @@
"numa_other",
#endif
- /* Event counters */
+#ifdef CONFIG_VM_EVENT_COUNTERS
"pgpgin",
"pgpgout",
"pswpin",
"pswpout",
- "pgalloc_high",
- "pgalloc_normal",
- "pgalloc_dma32",
"pgalloc_dma",
+ "pgalloc_dma32",
+ "pgalloc_normal",
+ "pgalloc_high",
"pgfree",
"pgactivate",
@@ -423,25 +420,25 @@
"pgfault",
"pgmajfault",
- "pgrefill_high",
- "pgrefill_normal",
- "pgrefill_dma32",
"pgrefill_dma",
+ "pgrefill_dma32",
+ "pgrefill_normal",
+ "pgrefill_high",
- "pgsteal_high",
- "pgsteal_normal",
- "pgsteal_dma32",
"pgsteal_dma",
+ "pgsteal_dma32",
+ "pgsteal_normal",
+ "pgsteal_high",
- "pgscan_kswapd_high",
- "pgscan_kswapd_normal",
- "pgscan_kswapd_dma32",
"pgscan_kswapd_dma",
+ "pgscan_kswapd_dma32",
+ "pgscan_kswapd_normal",
+ "pgscan_kswapd_high",
- "pgscan_direct_high",
- "pgscan_direct_normal",
- "pgscan_direct_dma32",
"pgscan_direct_dma",
+ "pgscan_direct_dma32",
+ "pgscan_direct_normal",
+ "pgscan_direct_high",
"pginodesteal",
"slabs_scanned",
@@ -451,6 +448,7 @@
"allocstall",
"pgrotated",
+#endif
};
/*
@@ -553,23 +551,32 @@
static void *vmstat_start(struct seq_file *m, loff_t *pos)
{
unsigned long *v;
- struct page_state *ps;
+#ifdef CONFIG_VM_EVENT_COUNTERS
+ unsigned long *e;
+#endif
int i;
if (*pos >= ARRAY_SIZE(vmstat_text))
return NULL;
+#ifdef CONFIG_VM_EVENT_COUNTERS
v = kmalloc(NR_VM_ZONE_STAT_ITEMS * sizeof(unsigned long)
- + sizeof(*ps), GFP_KERNEL);
+ + sizeof(struct vm_event_state), GFP_KERNEL);
+#else
+ v = kmalloc(NR_VM_ZONE_STAT_ITEMS * sizeof(unsigned long),
+ GFP_KERNEL);
+#endif
m->private = v;
if (!v)
return ERR_PTR(-ENOMEM);
for (i = 0; i < NR_VM_ZONE_STAT_ITEMS; i++)
v[i] = global_page_state(i);
- ps = (struct page_state *)(v + NR_VM_ZONE_STAT_ITEMS);
- get_full_page_state(ps);
- ps->pgpgin /= 2; /* sectors -> kbytes */
- ps->pgpgout /= 2;
+#ifdef CONFIG_VM_EVENT_COUNTERS
+ e = v + NR_VM_ZONE_STAT_ITEMS;
+ all_vm_events(e);
+ e[PGPGIN] /= 2; /* sectors -> kbytes */
+ e[PGPGOUT] /= 2;
+#endif
return v + *pos;
}