From: Muchun Song <songmuchun@xxxxxxxxxxxxx>
Subject: mm: memcontrol: convert NR_SHMEM_PMDMAPPED account to pages
Currently we use struct per_cpu_nodestat to cache the vmstat counters,
which leads to inaccurate statistics especially THP vmstat counters. In
the systems with hundreds of processors it can be GBs of memory. For
example, for a 96 CPUs system, the threshold is the maximum number of 125.
And the per cpu counters can cache 23.4375 GB in total.
The THP page is already a form of batched addition (it will add 512 worth
of memory in one go) so skipping the batching seems like sensible.
Although every THP stats update overflows the per-cpu counter, resorting
to atomic global updates. But it can make the statistics more accuracy
for the THP vmstat counters.
So we convert the NR_SHMEM_PMDMAPPED account to pages. This patch is
consistent with 8f182270dfec ("mm/swap.c: flush lru pvecs on compound page
arrival"). Doing this also can make the unit of vmstat counters more
unified. Finally, the unit of the vmstat counters are pages, kB and
bytes. The B/KB suffix can tell us that the unit is bytes or kB. The
rest which is without suffix are pages.
Link: https://lkml.kernel.org/r/20201228164110.2838-6-songmuchun@xxxxxxxxxxxxx
Signed-off-by: Muchun Song <songmuchun@xxxxxxxxxxxxx>
Cc: Alexey Dobriyan <adobriyan@xxxxxxxxx>
Cc: Feng Tang <feng.tang@xxxxxxxxx>
Cc: Greg Kroah-Hartman <gregkh@xxxxxxxxxxxxxxxxxxx>
Cc: Hugh Dickins <hughd@xxxxxxxxxx>
Cc: Johannes Weiner <hannes@xxxxxxxxxxx>
Cc: Joonsoo Kim <iamjoonsoo.kim@xxxxxxx>
Cc: Michal Hocko <mhocko@xxxxxxxx>
Cc: NeilBrown <neilb@xxxxxxx>
Cc: Pankaj Gupta <pankaj.gupta@xxxxxxxxxxxxxxx>
Cc: Rafael. J. Wysocki <rafael@xxxxxxxxxx>
Cc: Randy Dunlap <rdunlap@xxxxxxxxxxxxx>
Cc: Roman Gushchin <guro@xxxxxx>
Cc: Sami Tolvanen <samitolvanen@xxxxxxxxxx>
Cc: Shakeel Butt <shakeelb@xxxxxxxxxx>
Cc: Vladimir Davydov <vdavydov.dev@xxxxxxxxx>
Signed-off-by: Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx>
---
drivers/base/node.c | 3 +--
fs/proc/meminfo.c | 2 +-
include/linux/mmzone.h | 3 ++-
mm/page_alloc.c | 3 +--
mm/rmap.c | 14 ++++++++++----
5 files changed, 15 insertions(+), 10 deletions(-)
--- a/drivers/base/node.c~mm-memcontrol-convert-nr_shmem_pmdmapped-account-to-pages
+++ a/drivers/base/node.c
@@ -463,8 +463,7 @@ static ssize_t node_read_meminfo(struct
,
nid, K(node_page_state(pgdat, NR_ANON_THPS)),
nid, K(node_page_state(pgdat, NR_SHMEM_THPS)),
- nid, K(node_page_state(pgdat, NR_SHMEM_PMDMAPPED) *
- HPAGE_PMD_NR),
+ nid, K(node_page_state(pgdat, NR_SHMEM_PMDMAPPED)),
nid, K(node_page_state(pgdat, NR_FILE_THPS)),
nid, K(node_page_state(pgdat, NR_FILE_PMDMAPPED) *
HPAGE_PMD_NR)
--- a/fs/proc/meminfo.c~mm-memcontrol-convert-nr_shmem_pmdmapped-account-to-pages
+++ a/fs/proc/meminfo.c
@@ -133,7 +133,7 @@ static int meminfo_proc_show(struct seq_
show_val_kb(m, "ShmemHugePages: ",
global_node_page_state(NR_SHMEM_THPS));
show_val_kb(m, "ShmemPmdMapped: ",
- global_node_page_state(NR_SHMEM_PMDMAPPED) * HPAGE_PMD_NR);
+ global_node_page_state(NR_SHMEM_PMDMAPPED));
show_val_kb(m, "FileHugePages: ",
global_node_page_state(NR_FILE_THPS));
show_val_kb(m, "FilePmdMapped: ",
--- a/include/linux/mmzone.h~mm-memcontrol-convert-nr_shmem_pmdmapped-account-to-pages
+++ a/include/linux/mmzone.h
@@ -221,7 +221,8 @@ static __always_inline bool vmstat_item_
return item == NR_ANON_THPS ||
item == NR_FILE_THPS ||
- item == NR_SHMEM_THPS;
+ item == NR_SHMEM_THPS ||
+ item == NR_SHMEM_PMDMAPPED;
}
/*
--- a/mm/page_alloc.c~mm-memcontrol-convert-nr_shmem_pmdmapped-account-to-pages
+++ a/mm/page_alloc.c
@@ -5585,8 +5585,7 @@ void show_free_areas(unsigned int filter
K(node_page_state(pgdat, NR_SHMEM)),
#ifdef CONFIG_TRANSPARENT_HUGEPAGE
K(node_page_state(pgdat, NR_SHMEM_THPS)),
- K(node_page_state(pgdat, NR_SHMEM_PMDMAPPED)
- * HPAGE_PMD_NR),
+ K(node_page_state(pgdat, NR_SHMEM_PMDMAPPED)),
K(node_page_state(pgdat, NR_ANON_THPS)),
#endif
K(node_page_state(pgdat, NR_WRITEBACK_TEMP)),
--- a/mm/rmap.c~mm-memcontrol-convert-nr_shmem_pmdmapped-account-to-pages
+++ a/mm/rmap.c
@@ -1211,14 +1211,17 @@ void page_add_file_rmap(struct page *pag
VM_BUG_ON_PAGE(compound && !PageTransHuge(page), page);
lock_page_memcg(page);
if (compound && PageTransHuge(page)) {
- for (i = 0, nr = 0; i < thp_nr_pages(page); i++) {
+ int nr_pages = thp_nr_pages(page);
+
+ for (i = 0, nr = 0; i < nr_pages; i++) {
if (atomic_inc_and_test(&page[i]._mapcount))
nr++;
}
if (!atomic_inc_and_test(compound_mapcount_ptr(page)))
goto out;
if (PageSwapBacked(page))
- __inc_node_page_state(page, NR_SHMEM_PMDMAPPED);
+ __mod_lruvec_page_state(page, NR_SHMEM_PMDMAPPED,
+ nr_pages);
else
__inc_node_page_state(page, NR_FILE_PMDMAPPED);
} else {
@@ -1252,14 +1255,17 @@ static void page_remove_file_rmap(struct
/* page still mapped by someone else? */
if (compound && PageTransHuge(page)) {
- for (i = 0, nr = 0; i < thp_nr_pages(page); i++) {
+ int nr_pages = thp_nr_pages(page);
+
+ for (i = 0, nr = 0; i < nr_pages; i++) {
if (atomic_add_negative(-1, &page[i]._mapcount))
nr++;
}
if (!atomic_add_negative(-1, compound_mapcount_ptr(page)))
return;
if (PageSwapBacked(page))
- __dec_node_page_state(page, NR_SHMEM_PMDMAPPED);
+ __mod_lruvec_page_state(page, NR_SHMEM_PMDMAPPED,
+ -nr_pages);
else
__dec_node_page_state(page, NR_FILE_PMDMAPPED);
} else {
_
