summaryrefslogtreecommitdiffstats
diff options
context:
space:
mode:
-rw-r--r--include/linux/gfp.h4
-rw-r--r--mm/page_alloc.c17
-rw-r--r--mm/slab.c65
3 files changed, 72 insertions, 14 deletions
diff --git a/include/linux/gfp.h b/include/linux/gfp.h
index 20f9148..7851e6b 100644
--- a/include/linux/gfp.h
+++ b/include/linux/gfp.h
@@ -157,9 +157,9 @@ extern void FASTCALL(free_cold_page(struct page *page));
void page_alloc_init(void);
#ifdef CONFIG_NUMA
-void drain_remote_pages(void);
+void drain_node_pages(int node);
#else
-static inline void drain_remote_pages(void) { };
+static inline void drain_node_pages(int node) { };
#endif
#endif /* __LINUX_GFP_H */
diff --git a/mm/page_alloc.c b/mm/page_alloc.c
index 791690d..234bd48 100644
--- a/mm/page_alloc.c
+++ b/mm/page_alloc.c
@@ -590,21 +590,20 @@ static int rmqueue_bulk(struct zone *zone, unsigned int order,
}
#ifdef CONFIG_NUMA
-/* Called from the slab reaper to drain remote pagesets */
-void drain_remote_pages(void)
+/*
+ * Called from the slab reaper to drain pagesets on a particular node that
+ * belong to the currently executing processor.
+ */
+void drain_node_pages(int nodeid)
{
- struct zone *zone;
- int i;
+ int i, z;
unsigned long flags;
local_irq_save(flags);
- for_each_zone(zone) {
+ for (z = 0; z < MAX_NR_ZONES; z++) {
+ struct zone *zone = NODE_DATA(nodeid)->node_zones + z;
struct per_cpu_pageset *pset;
- /* Do not drain local pagesets */
- if (zone->zone_pgdat->node_id == numa_node_id())
- continue;
-
pset = zone_pcp(zone, smp_processor_id());
for (i = 0; i < ARRAY_SIZE(pset->pcp); i++) {
struct per_cpu_pages *pcp;
diff --git a/mm/slab.c b/mm/slab.c
index 61800b8..d0bd7f0 100644
--- a/mm/slab.c
+++ b/mm/slab.c
@@ -789,6 +789,47 @@ static void __slab_error(const char *function, struct kmem_cache *cachep, char *
dump_stack();
}
+#ifdef CONFIG_NUMA
+/*
+ * Special reaping functions for NUMA systems called from cache_reap().
+ * These take care of doing round robin flushing of alien caches (containing
+ * objects freed on different nodes from which they were allocated) and the
+ * flushing of remote pcps by calling drain_node_pages.
+ */
+static DEFINE_PER_CPU(unsigned long, reap_node);
+
+static void init_reap_node(int cpu)
+{
+ int node;
+
+ node = next_node(cpu_to_node(cpu), node_online_map);
+ if (node == MAX_NUMNODES)
+ node = 0;
+
+ __get_cpu_var(reap_node) = node;
+}
+
+static void next_reap_node(void)
+{
+ int node = __get_cpu_var(reap_node);
+
+ /*
+ * Also drain per cpu pages on remote zones
+ */
+ if (node != numa_node_id())
+ drain_node_pages(node);
+
+ node = next_node(node, node_online_map);
+ if (unlikely(node >= MAX_NUMNODES))
+ node = first_node(node_online_map);
+ __get_cpu_var(reap_node) = node;
+}
+
+#else
+#define init_reap_node(cpu) do { } while (0)
+#define next_reap_node(void) do { } while (0)
+#endif
+
/*
* Initiate the reap timer running on the target CPU. We run at around 1 to 2Hz
* via the workqueue/eventd.
@@ -806,6 +847,7 @@ static void __devinit start_cpu_timer(int cpu)
* at that time.
*/
if (keventd_up() && reap_work->func == NULL) {
+ init_reap_node(cpu);
INIT_WORK(reap_work, cache_reap, NULL);
schedule_delayed_work_on(cpu, reap_work, HZ + 3 * cpu);
}
@@ -884,6 +926,23 @@ static void __drain_alien_cache(struct kmem_cache *cachep,
}
}
+/*
+ * Called from cache_reap() to regularly drain alien caches round robin.
+ */
+static void reap_alien(struct kmem_cache *cachep, struct kmem_list3 *l3)
+{
+ int node = __get_cpu_var(reap_node);
+
+ if (l3->alien) {
+ struct array_cache *ac = l3->alien[node];
+ if (ac && ac->avail) {
+ spin_lock_irq(&ac->lock);
+ __drain_alien_cache(cachep, ac, node);
+ spin_unlock_irq(&ac->lock);
+ }
+ }
+}
+
static void drain_alien_cache(struct kmem_cache *cachep, struct array_cache **alien)
{
int i = 0;
@@ -902,6 +961,7 @@ static void drain_alien_cache(struct kmem_cache *cachep, struct array_cache **al
#else
#define drain_alien_cache(cachep, alien) do { } while (0)
+#define reap_alien(cachep, l3) do { } while (0)
static inline struct array_cache **alloc_alien_cache(int node, int limit)
{
@@ -3497,8 +3557,7 @@ static void cache_reap(void *unused)
check_irq_on();
l3 = searchp->nodelists[numa_node_id()];
- if (l3->alien)
- drain_alien_cache(searchp, l3->alien);
+ reap_alien(searchp, l3);
spin_lock_irq(&l3->list_lock);
drain_array_locked(searchp, cpu_cache_get(searchp), 0,
@@ -3548,7 +3607,7 @@ static void cache_reap(void *unused)
}
check_irq_on();
mutex_unlock(&cache_chain_mutex);
- drain_remote_pages();
+ next_reap_node();
/* Setup the next iteration */
schedule_delayed_work(&__get_cpu_var(reap_work), REAPTIMEOUT_CPUC);
}
OpenPOWER on IntegriCloud