The commit 2d146aa3aa84 ("mm: memcontrol: switch to rstat") but skipped
the conversion of the lruvec stats as such stats are read in the
performance critical code paths and flushing stats may have impacted the
performances of the applications. This patch converts the lruvec stats
to rstat and later patch adds the periodic flushing of the stats and
thus remove the need to synchronously flushing the stats in the
performance critical code paths.
The rstat conversion comes with the price i.e. memory cost. Effectively
this patch reverts the savings done by the commit f3344adf38bd ("mm:
memcontrol: optimize per-lruvec stats counter memory usage"). However
this cost is justified due to negative impact of the inaccurate lruvec
stats on many heuristics. One such case is reported in [1].
The memory reclaim code is filled with plethora of heuristics and many
of those heuristics reads the lruvec stats. So, inaccurate stats can
make such heuristics ineffective. [1] reports the impact of inaccurate
lruvec stats on the "cache trim mode" heuristic. Inaccurate lruvec stats
can impact the deactivation and aging anon heuristics as well.
[1] https://lore.kernel.org/linux-mm/[email protected]/
Signed-off-by: Shakeel Butt <[email protected]>
---
include/linux/memcontrol.h | 42 +++++++------
mm/memcontrol.c | 118 +++++++++++++------------------------
2 files changed, 60 insertions(+), 100 deletions(-)
diff --git a/include/linux/memcontrol.h b/include/linux/memcontrol.h
index 3cc18c2176e7..81d65d32ec2a 100644
--- a/include/linux/memcontrol.h
+++ b/include/linux/memcontrol.h
@@ -105,14 +105,6 @@ struct mem_cgroup_reclaim_iter {
unsigned int generation;
};
-struct lruvec_stat {
- long count[NR_VM_NODE_STAT_ITEMS];
-};
-
-struct batched_lruvec_stat {
- s32 count[NR_VM_NODE_STAT_ITEMS];
-};
-
/*
* Bitmap and deferred work of shrinker::id corresponding to memcg-aware
* shrinkers, which have elements charged to this memcg.
@@ -123,24 +115,30 @@ struct shrinker_info {
unsigned long *map;
};
+struct lruvec_stats_percpu {
+ /* Local (CPU and cgroup) state */
+ long state[NR_VM_NODE_STAT_ITEMS];
+
+ /* Delta calculation for lockless upward propagation */
+ long state_prev[NR_VM_NODE_STAT_ITEMS];
+};
+
+struct lruvec_stats {
+ /* Aggregated (CPU and subtree) state */
+ long state[NR_VM_NODE_STAT_ITEMS];
+
+ /* Pending child counts during tree propagation */
+ long state_pending[NR_VM_NODE_STAT_ITEMS];
+};
+
/*
* per-node information in memory controller.
*/
struct mem_cgroup_per_node {
struct lruvec lruvec;
- /*
- * 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 batched_lruvec_stat __percpu *lruvec_stat_cpu;
- atomic_long_t lruvec_stat[NR_VM_NODE_STAT_ITEMS];
+ struct lruvec_stats_percpu __percpu *lruvec_stats_percpu;
+ struct lruvec_stats lruvec_stats;
unsigned long lru_zone_size[MAX_NR_ZONES][NR_LRU_LISTS];
@@ -965,7 +963,7 @@ static inline unsigned long lruvec_page_state(struct lruvec *lruvec,
return node_page_state(lruvec_pgdat(lruvec), idx);
pn = container_of(lruvec, struct mem_cgroup_per_node, lruvec);
- x = atomic_long_read(&pn->lruvec_stat[idx]);
+ x = READ_ONCE(pn->lruvec_stats.state[idx]);
#ifdef CONFIG_SMP
if (x < 0)
x = 0;
@@ -985,7 +983,7 @@ static inline unsigned long lruvec_page_state_local(struct lruvec *lruvec,
pn = container_of(lruvec, struct mem_cgroup_per_node, lruvec);
for_each_possible_cpu(cpu)
- x += per_cpu(pn->lruvec_stat_local->count[idx], cpu);
+ x += per_cpu(pn->lruvec_stats_percpu->state[idx], cpu);
#ifdef CONFIG_SMP
if (x < 0)
x = 0;
diff --git a/mm/memcontrol.c b/mm/memcontrol.c
index b9a6db6a7d4f..d48f727bec05 100644
--- a/mm/memcontrol.c
+++ b/mm/memcontrol.c
@@ -665,46 +665,20 @@ static unsigned long memcg_page_state_local(struct mem_cgroup *memcg, int idx)
return x;
}
-static struct mem_cgroup_per_node *
-parent_nodeinfo(struct mem_cgroup_per_node *pn, int nid)
-{
- struct mem_cgroup *parent;
-
- parent = parent_mem_cgroup(pn->memcg);
- if (!parent)
- return NULL;
- return parent->nodeinfo[nid];
-}
-
void __mod_memcg_lruvec_state(struct lruvec *lruvec, enum node_stat_item idx,
int val)
{
struct mem_cgroup_per_node *pn;
struct mem_cgroup *memcg;
- long x, threshold = MEMCG_CHARGE_BATCH;
pn = container_of(lruvec, struct mem_cgroup_per_node, lruvec);
memcg = pn->memcg;
- /* Update memcg */
- __mod_memcg_state(memcg, idx, val);
-
/* Update lruvec */
- __this_cpu_add(pn->lruvec_stat_local->count[idx], val);
+ __this_cpu_add(pn->lruvec_stats_percpu->state[idx], val);
- if (vmstat_item_in_bytes(idx))
- threshold <<= PAGE_SHIFT;
-
- x = val + __this_cpu_read(pn->lruvec_stat_cpu->count[idx]);
- if (unlikely(abs(x) > threshold)) {
- pg_data_t *pgdat = lruvec_pgdat(lruvec);
- struct mem_cgroup_per_node *pi;
-
- for (pi = pn; pi; pi = parent_nodeinfo(pi, pgdat->node_id))
- atomic_long_add(x, &pi->lruvec_stat[idx]);
- x = 0;
- }
- __this_cpu_write(pn->lruvec_stat_cpu->count[idx], x);
+ /* Update memcg */
+ __mod_memcg_state(memcg, idx, val);
}
/**
@@ -2271,40 +2245,13 @@ static void drain_all_stock(struct mem_cgroup *root_memcg)
mutex_unlock(&percpu_charge_mutex);
}
-static void memcg_flush_lruvec_page_state(struct mem_cgroup *memcg, int cpu)
-{
- int nid;
-
- for_each_node(nid) {
- struct mem_cgroup_per_node *pn = memcg->nodeinfo[nid];
- unsigned long stat[NR_VM_NODE_STAT_ITEMS];
- struct batched_lruvec_stat *lstatc;
- int i;
-
- lstatc = per_cpu_ptr(pn->lruvec_stat_cpu, cpu);
- for (i = 0; i < NR_VM_NODE_STAT_ITEMS; i++) {
- stat[i] = lstatc->count[i];
- lstatc->count[i] = 0;
- }
-
- do {
- for (i = 0; i < NR_VM_NODE_STAT_ITEMS; i++)
- atomic_long_add(stat[i], &pn->lruvec_stat[i]);
- } while ((pn = parent_nodeinfo(pn, nid)));
- }
-}
-
static int memcg_hotplug_cpu_dead(unsigned int cpu)
{
struct memcg_stock_pcp *stock;
- struct mem_cgroup *memcg;
stock = &per_cpu(memcg_stock, cpu);
drain_stock(stock);
- for_each_mem_cgroup(memcg)
- memcg_flush_lruvec_page_state(memcg, cpu);
-
return 0;
}
@@ -5108,17 +5055,9 @@ static int alloc_mem_cgroup_per_node_info(struct mem_cgroup *memcg, int node)
if (!pn)
return 1;
- pn->lruvec_stat_local = alloc_percpu_gfp(struct lruvec_stat,
- GFP_KERNEL_ACCOUNT);
- if (!pn->lruvec_stat_local) {
- kfree(pn);
- return 1;
- }
-
- 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);
+ pn->lruvec_stats_percpu = alloc_percpu_gfp(struct lruvec_stats_percpu,
+ GFP_KERNEL_ACCOUNT);
+ if (!pn->lruvec_stats_percpu) {
kfree(pn);
return 1;
}
@@ -5139,8 +5078,7 @@ static void free_mem_cgroup_per_node_info(struct mem_cgroup *memcg, int node)
if (!pn)
return;
- free_percpu(pn->lruvec_stat_cpu);
- free_percpu(pn->lruvec_stat_local);
+ free_percpu(pn->lruvec_stats_percpu);
kfree(pn);
}
@@ -5156,15 +5094,7 @@ static void __mem_cgroup_free(struct mem_cgroup *memcg)
static void mem_cgroup_free(struct mem_cgroup *memcg)
{
- int cpu;
-
memcg_wb_domain_exit(memcg);
- /*
- * Flush percpu lruvec stats to guarantee the value
- * correctness on parent's and all ancestor levels.
- */
- for_each_online_cpu(cpu)
- memcg_flush_lruvec_page_state(memcg, cpu);
__mem_cgroup_free(memcg);
}
@@ -5397,7 +5327,7 @@ static void mem_cgroup_css_rstat_flush(struct cgroup_subsys_state *css, int cpu)
struct mem_cgroup *parent = parent_mem_cgroup(memcg);
struct memcg_vmstats_percpu *statc;
long delta, v;
- int i;
+ int i, nid;
statc = per_cpu_ptr(memcg->vmstats_percpu, cpu);
@@ -5445,6 +5375,36 @@ static void mem_cgroup_css_rstat_flush(struct cgroup_subsys_state *css, int cpu)
if (parent)
parent->vmstats.events_pending[i] += delta;
}
+
+ for_each_node_state(nid, N_MEMORY) {
+ struct mem_cgroup_per_node *pn = memcg->nodeinfo[nid];
+ struct mem_cgroup_per_node *ppn = NULL;
+ struct lruvec_stats_percpu *lstatc;
+
+ if (parent)
+ ppn = parent->nodeinfo[nid];
+
+ lstatc = per_cpu_ptr(pn->lruvec_stats_percpu, cpu);
+
+ for (i = 0; i < NR_VM_NODE_STAT_ITEMS; i++) {
+ delta = pn->lruvec_stats.state_pending[i];
+ if (delta)
+ pn->lruvec_stats.state_pending[i] = 0;
+
+ v = READ_ONCE(lstatc->state[i]);
+ if (v != lstatc->state_prev[i]) {
+ delta += v - lstatc->state_prev[i];
+ lstatc->state_prev[i] = v;
+ }
+
+ if (!delta)
+ continue;
+
+ pn->lruvec_stats.state[i] += delta;
+ if (ppn)
+ ppn->lruvec_stats.state_pending[i] += delta;
+ }
+ }
}
#ifdef CONFIG_MMU
@@ -6378,6 +6338,8 @@ static int memory_numa_stat_show(struct seq_file *m, void *v)
int i;
struct mem_cgroup *memcg = mem_cgroup_from_seq(m);
+ cgroup_rstat_flush(memcg->css.cgroup);
+
for (i = 0; i < ARRAY_SIZE(memory_stats); i++) {
int nid;
--
2.32.0.rc1.229.g3e70b5a671-goog
At the moment memcg stats are read in four contexts:
1. memcg stat user interfaces
2. dirty throttling
3. page fault
4. memory reclaim
Currently the kernel flushes the stats for first two cases. Flushing the
stats for remaining two casese may have performance impact. Always
flushing the memcg stats on the page fault code path may negatively
impacts the performance of the applications. In addition flushing in the
memory reclaim code path, though treated as slowpath, can become the
source of contention for the global lock taken for stat flushing because
when system or memcg is under memory pressure, many tasks may enter the
reclaim path.
Instead of synchronously flushing the stats, this patch adds support of
asynchronous periodic flushing of the memcg stats. For now the flushing
period is hardcoded to 2*HZ but that can be changed later through maybe
sysctl if need arise.
This patch does add the explicit flushing in the kswapd thread as the
number of kswapd threads which corresponds to the number of nodes on
realistic machines are usually low.
Signed-off-by: Shakeel Butt <[email protected]>
---
include/linux/memcontrol.h | 10 ++++++++++
mm/memcontrol.c | 14 ++++++++++++++
mm/vmscan.c | 6 ++++++
3 files changed, 30 insertions(+)
diff --git a/include/linux/memcontrol.h b/include/linux/memcontrol.h
index 81d65d32ec2a..222c00e76ef9 100644
--- a/include/linux/memcontrol.h
+++ b/include/linux/memcontrol.h
@@ -991,6 +991,12 @@ static inline unsigned long lruvec_page_state_local(struct lruvec *lruvec,
return x;
}
+static inline void mem_cgroup_flush_stats(void)
+{
+ if (!mem_cgroup_disabled())
+ cgroup_rstat_flush(root_mem_cgroup->css.cgroup);
+}
+
void __mod_memcg_lruvec_state(struct lruvec *lruvec, enum node_stat_item idx,
int val);
void __mod_lruvec_kmem_state(void *p, enum node_stat_item idx, int val);
@@ -1394,6 +1400,10 @@ static inline unsigned long lruvec_page_state_local(struct lruvec *lruvec,
return node_page_state(lruvec_pgdat(lruvec), idx);
}
+static inline void mem_cgroup_flush_stats(void)
+{
+}
+
static inline void __mod_memcg_lruvec_state(struct lruvec *lruvec,
enum node_stat_item idx, int val)
{
diff --git a/mm/memcontrol.c b/mm/memcontrol.c
index d48f727bec05..6c8578faa8b4 100644
--- a/mm/memcontrol.c
+++ b/mm/memcontrol.c
@@ -96,6 +96,10 @@ bool cgroup_memory_noswap __read_mostly;
static DECLARE_WAIT_QUEUE_HEAD(memcg_cgwb_frn_waitq);
#endif
+/* Periodically flush memcg and lruvec stats. */
+static void flush_memcg_stats(struct work_struct *w);
+static DECLARE_DEFERRABLE_WORK(stats_flush, flush_memcg_stats);
+
/* Whether legacy memory+swap accounting is active */
static bool do_memsw_account(void)
{
@@ -5230,6 +5234,10 @@ static int mem_cgroup_css_online(struct cgroup_subsys_state *css)
/* Online state pins memcg ID, memcg ID pins CSS */
refcount_set(&memcg->id.ref, 1);
css_get(css);
+
+ if (unlikely(mem_cgroup_is_root(memcg)))
+ schedule_delayed_work(&stats_flush, round_jiffies(2UL*HZ));
+
return 0;
}
@@ -5321,6 +5329,12 @@ static void mem_cgroup_css_reset(struct cgroup_subsys_state *css)
memcg_wb_domain_size_changed(memcg);
}
+static void flush_memcg_stats(struct work_struct *w)
+{
+ cgroup_rstat_flush(root_mem_cgroup->css.cgroup);
+ schedule_delayed_work(&stats_flush, round_jiffies(2UL*HZ));
+}
+
static void mem_cgroup_css_rstat_flush(struct cgroup_subsys_state *css, int cpu)
{
struct mem_cgroup *memcg = mem_cgroup_from_css(css);
diff --git a/mm/vmscan.c b/mm/vmscan.c
index 60a19fd6ea3f..16546a5be922 100644
--- a/mm/vmscan.c
+++ b/mm/vmscan.c
@@ -3872,6 +3872,12 @@ static int balance_pgdat(pg_data_t *pgdat, int order, int highest_zoneidx)
sc.may_writepage = !laptop_mode && !nr_boost_reclaim;
sc.may_swap = !nr_boost_reclaim;
+ /*
+ * Flush the memory cgroup stats, so that we read accurate
+ * per-memcg lruvec stats for heuristics later.
+ */
+ mem_cgroup_flush_stats();
+
/*
* Do some background aging of the anon list, to give
* pages a chance to be referenced before reclaiming. All
--
2.32.0.rc1.229.g3e70b5a671-goog
On Fri, Jun 04, 2021 at 02:18:16PM +0800, Hillf Danton wrote:
> On Thu, 3 Jun 2021 18:56:40 -0700 Shakeel Butt wrote:
> >
> > +static void flush_memcg_stats(struct work_struct *w)
> > +{
> > + cgroup_rstat_flush(root_mem_cgroup->css.cgroup);
> > + schedule_delayed_work(&stats_flush, round_jiffies(2UL*HZ));
> > +}
>
> Given flush may block, the unbound wq is what you need.
>
> queue_delayed_work(system_unbound_wq, &stats_flush, 2 * HZ);
Default per-cpu workqueue can block just fine. I don't see a strong reason
why this would need to be unbound.
--
tejun
On Sat, Jun 05, 2021 at 09:54:21AM +0800, Hillf Danton wrote:
> The cond_resched() in cgroup_rstat_flush_locked() matches its appearence in
> your post [1]. So does unbound IMHO.
Ah yeah, this either needs CPU_INTENSIVE or UNBOUND, prolly the latter is
better.
> And the short stuff [2] looks to me like it is incorrect to queue a work
> acquiring mutex lock on to the system_wq. IOW the unbound wq is the right
> thing for any work that might sleep.
This part doesn't make sense. Blocking from per-cpu workqueue is completely
fine. What's not fine is consuming a lot of CPU cycles.
Thanks.
--
tejun
On Fri, Jun 4, 2021 at 7:33 PM Tejun Heo <[email protected]> wrote:
>
> On Sat, Jun 05, 2021 at 09:54:21AM +0800, Hillf Danton wrote:
> > The cond_resched() in cgroup_rstat_flush_locked() matches its appearence in
> > your post [1]. So does unbound IMHO.
>
> Ah yeah, this either needs CPU_INTENSIVE or UNBOUND, prolly the latter is
> better.
>
I will change the next version to the system_unbound_wq.