The patch titled
Subject: mm: fix vm-scalability regression in cgroup-aware workingset code
has been added to the -mm tree. Its filename is
mm-fix-vm-scalability-regression-in-cgroup-aware-workingset-code.patch
This patch should soon appear at
http://ozlabs.org/~akpm/mmots/broken-out/mm-fix-vm-scalability-regression-in-cgroup-aware-workingset-code.patch
and later at
http://ozlabs.org/~akpm/mmotm/broken-out/mm-fix-vm-scalability-regression-in-cgroup-aware-workingset-code.patch
Before you just go and hit "reply", please:
a) Consider who else should be cc'ed
b) Prefer to cc a suitable mailing list as well
c) Ideally: find the original patch on the mailing list and do a
reply-to-all to that, adding suitable additional cc's
*** Remember to use Documentation/SubmitChecklist when testing your code ***
The -mm tree is included into linux-next and is updated
there every 3-4 working days
------------------------------------------------------
From: Johannes Weiner <hannes@xxxxxxxxxxx>
Subject: mm: fix vm-scalability regression in cgroup-aware workingset code
23047a96d7cf ("mm: workingset: per-cgroup cache thrash detection") added a
page->mem_cgroup lookup to the cache eviction, refault, and activation
paths, as well as locking to the activation path, and the vm-scalability
tests showed a regression of -23%. While the test in question is an
artificial worst-case scenario that doesn't occur in real workloads -
reading two sparse files in parallel at full CPU speed just to hammer the
LRU paths - there is still some optimizations that can be done in those
paths.
Inline the lookup functions to eliminate calls. Also, page->mem_cgroup
doesn't need to be stabilized when counting an activation; we merely need
to hold the RCU lock to prevent the memcg from being freed.
This cuts down on overhead quite a bit:
23047a96d7cfcfca 063f6715e77a7be5770d6081fe
---------------- --------------------------
%stddev %change %stddev
| 21621405 ± 0% +11.3% 24069657 ± 2% vm-scalability.throughput
Link: http://lkml.kernel.org/r/20160622182019.24064-1-hannes@xxxxxxxxxxx
Reported-by: Ye Xiaolong <xiaolong.ye@xxxxxxxxx>
Signed-off-by: Johannes Weiner <hannes@xxxxxxxxxxx>
Cc: Michal Hocko <mhocko@xxxxxxx>
Cc: Vladimir Davydov <vdavydov@xxxxxxxxxxxxx>
Signed-off-by: Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx>
---
include/linux/memcontrol.h | 39 ++++++++++++++++++++++++++++++++++-
include/linux/mm.h | 8 +++++++
mm/memcontrol.c | 39 -----------------------------------
mm/workingset.c | 10 +++++---
4 files changed, 52 insertions(+), 44 deletions(-)
diff -puN include/linux/memcontrol.h~mm-fix-vm-scalability-regression-in-cgroup-aware-workingset-code include/linux/memcontrol.h
--- a/include/linux/memcontrol.h~mm-fix-vm-scalability-regression-in-cgroup-aware-workingset-code
+++ a/include/linux/memcontrol.h
@@ -310,7 +310,44 @@ void mem_cgroup_uncharge_list(struct lis
void mem_cgroup_migrate(struct page *oldpage, struct page *newpage);
-struct lruvec *mem_cgroup_lruvec(struct pglist_data *, struct mem_cgroup *);
+static inline struct mem_cgroup_per_node *
+mem_cgroup_nodeinfo(struct mem_cgroup *memcg, int nid)
+{
+ return memcg->nodeinfo[nid];
+}
+
+/**
+ * mem_cgroup_lruvec - get the lru list vector for a memcg node
+ * @node: node of the wanted lruvec
+ * @memcg: memcg of the wanted lruvec
+ *
+ * Returns the lru list vector holding pages for a given @node and @memcg.
+ * This can be the node lruvec, if the memory controller is disabled.
+ */
+static inline struct lruvec *mem_cgroup_lruvec(struct pglist_data *pgdat,
+ struct mem_cgroup *memcg)
+{
+ struct mem_cgroup_per_node *mz;
+ struct lruvec *lruvec;
+
+ if (mem_cgroup_disabled()) {
+ lruvec = node_lruvec(pgdat);
+ goto out;
+ }
+
+ mz = mem_cgroup_nodeinfo(memcg, pgdat->node_id);
+ lruvec = &mz->lruvec;
+out:
+ /*
+ * Since a node can be onlined after the mem_cgroup was created,
+ * we have to be prepared to initialize lruvec->zone here;
+ * and if offlined then reonlined, we need to reinitialize it.
+ */
+ if (unlikely(lruvec->pgdat != pgdat))
+ lruvec->pgdat = pgdat;
+ return lruvec;
+}
+
struct lruvec *mem_cgroup_page_lruvec(struct page *, struct pglist_data *);
bool task_in_mem_cgroup(struct task_struct *task, struct mem_cgroup *memcg);
diff -puN include/linux/mm.h~mm-fix-vm-scalability-regression-in-cgroup-aware-workingset-code include/linux/mm.h
--- a/include/linux/mm.h~mm-fix-vm-scalability-regression-in-cgroup-aware-workingset-code
+++ a/include/linux/mm.h
@@ -978,11 +978,19 @@ static inline struct mem_cgroup *page_me
{
return page->mem_cgroup;
}
+static inline struct mem_cgroup *page_memcg_rcu(struct page *page)
+{
+ return READ_ONCE(page->mem_cgroup);
+}
#else
static inline struct mem_cgroup *page_memcg(struct page *page)
{
return NULL;
}
+static inline struct mem_cgroup *page_memcg_rcu(struct page *page)
+{
+ return NULL;
+}
#endif
/*
diff -puN mm/memcontrol.c~mm-fix-vm-scalability-regression-in-cgroup-aware-workingset-code mm/memcontrol.c
--- a/mm/memcontrol.c~mm-fix-vm-scalability-regression-in-cgroup-aware-workingset-code
+++ a/mm/memcontrol.c
@@ -319,12 +319,6 @@ EXPORT_SYMBOL(memcg_kmem_enabled_key);
#endif /* !CONFIG_SLOB */
-static struct mem_cgroup_per_node *
-mem_cgroup_nodeinfo(struct mem_cgroup *memcg, int nid)
-{
- return memcg->nodeinfo[nid];
-}
-
/**
* mem_cgroup_css_from_page - css of the memcg associated with a page
* @page: page of interest
@@ -927,39 +921,6 @@ static void invalidate_reclaim_iterators
iter = mem_cgroup_iter(NULL, iter, NULL))
/**
- * mem_cgroup_lruvec - get the lru list vector for a node or a memcg zone
- * @node: node of the wanted lruvec
- * @memcg: memcg of the wanted lruvec
- *
- * Returns the lru list vector holding pages for a given @node or a given
- * @memcg and @zone. This can be the node lruvec, if the memory controller
- * is disabled.
- */
-struct lruvec *mem_cgroup_lruvec(struct pglist_data *pgdat,
- struct mem_cgroup *memcg)
-{
- struct mem_cgroup_per_node *mz;
- struct lruvec *lruvec;
-
- if (mem_cgroup_disabled()) {
- lruvec = node_lruvec(pgdat);
- goto out;
- }
-
- mz = mem_cgroup_nodeinfo(memcg, pgdat->node_id);
- lruvec = &mz->lruvec;
-out:
- /*
- * Since a node can be onlined after the mem_cgroup was created,
- * we have to be prepared to initialize lruvec->zone here;
- * and if offlined then reonlined, we need to reinitialize it.
- */
- if (unlikely(lruvec->pgdat != pgdat))
- lruvec->pgdat = pgdat;
- return lruvec;
-}
-
-/**
* mem_cgroup_page_lruvec - return lruvec for isolating/putting an LRU page
* @page: the page
* @zone: zone of the page
diff -puN mm/workingset.c~mm-fix-vm-scalability-regression-in-cgroup-aware-workingset-code mm/workingset.c
--- a/mm/workingset.c~mm-fix-vm-scalability-regression-in-cgroup-aware-workingset-code
+++ a/mm/workingset.c
@@ -302,9 +302,10 @@ bool workingset_refault(void *shadow)
*/
void workingset_activation(struct page *page)
{
+ struct mem_cgroup *memcg;
struct lruvec *lruvec;
- lock_page_memcg(page);
+ rcu_read_lock();
/*
* Filter non-memcg pages here, e.g. unmap can call
* mark_page_accessed() on VDSO pages.
@@ -312,12 +313,13 @@ void workingset_activation(struct page *
* XXX: See workingset_refault() - this should return
* root_mem_cgroup even for !CONFIG_MEMCG.
*/
- if (!mem_cgroup_disabled() && !page_memcg(page))
+ memcg = page_memcg_rcu(page);
+ if (!mem_cgroup_disabled() && !memcg)
goto out;
- lruvec = mem_cgroup_lruvec(page_pgdat(page), page_memcg(page));
+ lruvec = mem_cgroup_lruvec(page_pgdat(page), memcg);
atomic_long_inc(&lruvec->inactive_age);
out:
- unlock_page_memcg(page);
+ rcu_read_unlock();
}
/*
_
Patches currently in -mm which might be from hannes@xxxxxxxxxxx are
cgroup-fix-idr-leak-for-the-first-cgroup-root.patch
cgroup-remove-unnecessary-0-check-from-css_from_id.patch
mm-memcontrol-fix-cgroup-creation-failure-after-many-small-jobs.patch
mm-memcontrol-fix-cgroup-creation-failure-after-many-small-jobs-fix.patch
mm-fix-vm-scalability-regression-in-cgroup-aware-workingset-code.patch
--
To unsubscribe from this list: send the line "unsubscribe mm-commits" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
