BackPup/Linux
1
0 Fork 0
mirror of https://github.com/torvalds/linux.git synced 2026-10-07 22:37:45 -04:00
Code Issues Packages Projects Releases Wiki Activity

mm: memcontrol: optimize per-lruvec stats counter memory usage

Browse Source 
The vmstat threshold is 32 (MEMCG_CHARGE_BATCH), Actually the threshold
can be as big as MEMCG_CHARGE_BATCH * PAGE_SIZE.  It still fits into s32.
So introduce struct batched_lruvec_stat to optimize memory usage.

The size of struct lruvec_stat is 304 bytes on 64 bit systems.  As it is a
per-cpu structure.  So with this patch, we can save 304 / 2 * ncpu bytes
per-memcg per-node where ncpu is the number of the possible CPU.  If there
are c memory cgroup (include dying cgroup) and n NUMA node in the system.
Finally, we can save (152 * ncpu * c * n) bytes.

[[email protected]: fix typo in comment]

Link: https://lkml.kernel.org/r/[email protected]
Signed-off-by: Muchun Song <[email protected]>
Reviewed-by: Shakeel Butt <[email protected]>
Cc: Johannes Weiner <[email protected]>
Cc: Michal Hocko <[email protected]>
Cc: Vladimir Davydov <[email protected]>
Cc: Shakeel Butt <[email protected]>
Cc: Roman Gushchin <[email protected]>
Cc: Stephen Rothwell <[email protected]>
Cc: Chris Down <[email protected]>
Cc: Yafang Shao <[email protected]>
Cc: Wei Yang <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
This commit is contained in:
Muchun Song authored and Linus Torvalds committed 2021-02-24 13:38:29 -08:00
1 parent 2e9bd48315
commit f3344adf38
2 files changed
+21 -3
Download Patch File Download Diff File Expand all files Collapse all files

No files matched your search

include/linux/memcontrol.h
+12 -2
View File
@@ -92,6 +92,10 @@ struct lruvec_stat {
long count[NR_VM_NODE_STAT_ITEMS];
};
struct batched_lruvec_stat {
s32 count[NR_VM_NODE_STAT_ITEMS];
};
/*
* Bitmap of shrinker::id corresponding to memcg-aware shrinkers,
* which have elements charged to this memcg.
@@ -107,11 +111,17 @@ struct memcg_shrinker_map {
struct mem_cgroup_per_node {
struct lruvec lruvec;
/* Legacy local VM stats */
/*
* Legacy local VM stats. This should be struct lruvec_stat and
* cannot be optimized to struct batched_lruvec_stat. Because
* the threshold of the lruvec_stat_cpu can be as big as
* MEMCG_CHARGE_BATCH * PAGE_SIZE. It can fit into s32. But this
* filed has no upper limit.
*/
struct lruvec_stat __percpu *lruvec_stat_local;
/* Subtree VM stats (batched updates) */
struct lruvec_stat __percpu *lruvec_stat_cpu;
struct batched_lruvec_stat __percpu *lruvec_stat_cpu;
atomic_long_t lruvec_stat[NR_VM_NODE_STAT_ITEMS];
unsigned long lru_zone_size[MAX_NR_ZONES][NR_LRU_LISTS];
mm/memcontrol.c
+9 -1
View File
@@ -5208,7 +5208,7 @@ static int alloc_mem_cgroup_per_node_info(struct mem_cgroup *memcg, int node)
return 1;
}
pn->lruvec_stat_cpu = alloc_percpu_gfp(struct lruvec_stat,
pn->lruvec_stat_cpu = alloc_percpu_gfp(struct batched_lruvec_stat,
GFP_KERNEL_ACCOUNT);
if (!pn->lruvec_stat_cpu) {
free_percpu(pn->lruvec_stat_local);
@@ -7093,6 +7093,14 @@ static int __init mem_cgroup_init(void)
{
int cpu, node;
/*
* Currently s32 type (can refer to struct batched_lruvec_stat) is
* used for per-memcg-per-cpu caching of per-node statistics. In order
* to work fine, we should make sure that the overfill threshold can't
* exceed S32_MAX / PAGE_SIZE.
*/
BUILD_BUG_ON(MEMCG_CHARGE_BATCH > S32_MAX / PAGE_SIZE);
cpuhp_setup_state_nocalls(CPUHP_MM_MEMCQ_DEAD, "mm/memctrl:dead", NULL,
memcg_hotplug_cpu_dead);