Here is the second version of the patch. I have used alloc_pages_exact
instead of the complex double array approach.
I still fallback to kmalloc/vmalloc because hotplug can happen quite
some time after boot and we can end up not having enough continuous
pages at that time.
I am also thinking whether it would make sense to introduce
alloc_pages_exact_node function which would allocate pages from the
given node.
Any thoughts?
---
>From e8909bbd1d759de274a6ed7812530e576ad8bc44 Mon Sep 17 00:00:00 2001
From: Michal Hocko <mhocko@xxxxxxx>
Date: Thu, 24 Feb 2011 11:25:44 +0100
Subject: [PATCH] page_cgroup: Reduce allocation overhead for page_cgroup array for CONFIG_SPARSEMEM
Currently we are allocating a single page_cgroup array per memory
section (stored in mem_section->base) when CONFIG_SPARSEMEM is selected.
This is correct but memory inefficient solution because the allocated
memory (unless we fall back to vmalloc) is not kmalloc friendly:
- 32b - 16384 entries (20B per entry) fit into 327680B so the
524288B slab cache is used
- 32b with PAE - 131072 entries with 2621440B fit into 4194304B
- 64b - 32768 entries (40B per entry) fit into 2097152 cache
This is ~37% wasted space per memory section and it sumps up for the
whole memory. On a x86_64 machine it is something like 6MB per 1GB of
RAM.
We can reduce the internal fragmentation either by imeplementing 2
dimensional array and allocate kmalloc aligned sizes for each entry (as
suggested in https://lkml.org/lkml/2011/2/23/232) or we can get rid of
kmalloc altogether and allocate directly from the buddy allocator (use
alloc_pages_exact) as suggested by Dave Hansen.
The later solution is much simpler and the internal fragmentation is
comparable (~1 page per section).
We still need a fallback to kmalloc/vmalloc because we have no
guarantees that we will have a continuous memory of that size (order-10)
later on the hotplug events.
Signed-off-by: Michal Hocko <mhocko@xxxxxxx>
---
mm/page_cgroup.c | 62 ++++++++++++++++++++++++++++++++++--------------------
1 files changed, 39 insertions(+), 23 deletions(-)
diff --git a/mm/page_cgroup.c b/mm/page_cgroup.c
index 5bffada..eaae7de 100644
--- a/mm/page_cgroup.c
+++ b/mm/page_cgroup.c
@@ -105,7 +105,41 @@ struct page_cgroup *lookup_page_cgroup(struct page *page)
return section->page_cgroup + pfn;
}
-/* __alloc_bootmem...() is protected by !slab_available() */
+static void *__init_refok alloc_mcg_table(size_t size, int nid)
+{
+ void *addr = NULL;
+ if((addr = alloc_pages_exact(size, GFP_KERNEL | __GFP_NOWARN)))
+ return addr;
+
+ if (node_state(nid, N_HIGH_MEMORY)) {
+ addr = kmalloc_node(size, GFP_KERNEL | __GFP_NOWARN, nid);
+ if (!addr)
+ addr = vmalloc_node(size, nid);
+ } else {
+ addr = kmalloc(size, GFP_KERNEL | __GFP_NOWARN);
+ if (!addr)
+ addr = vmalloc(size);
+ }
+
+ return addr;
+}
+
+static void *free__mcg_table(void *addr)
+{
+ if (is_vmalloc_addr(addr)) {
+ vfree(addr);
+ } else {
+ struct page *page = virt_to_page(addr);
+ if (!PageReserved(page)) { /* Is bootmem ? */
+ if (!PageSlab(page)) {
+ size_t table_size = sizeof(struct page_cgroup) * PAGES_PER_SECTION;
+ free_pages_exact(addr, table_size);
+ } else
+ kfree(addr);
+ }
+ }
+}
+
static int __init_refok init_section_page_cgroup(unsigned long pfn)
{
struct mem_section *section = __pfn_to_section(pfn);
@@ -114,19 +148,9 @@ static int __init_refok init_section_page_cgroup(unsigned long pfn)
int nid, index;
if (!section->page_cgroup) {
- nid = page_to_nid(pfn_to_page(pfn));
table_size = sizeof(struct page_cgroup) * PAGES_PER_SECTION;
- VM_BUG_ON(!slab_is_available());
- if (node_state(nid, N_HIGH_MEMORY)) {
- base = kmalloc_node(table_size,
- GFP_KERNEL | __GFP_NOWARN, nid);
- if (!base)
- base = vmalloc_node(table_size, nid);
- } else {
- base = kmalloc(table_size, GFP_KERNEL | __GFP_NOWARN);
- if (!base)
- base = vmalloc(table_size);
- }
+ nid = page_to_nid(pfn_to_page(pfn));
+ base = alloc_mcg_table(table_size, nid);
/*
* The value stored in section->page_cgroup is (base - pfn)
* and it does not point to the memory block allocated above,
@@ -170,16 +194,8 @@ void __free_page_cgroup(unsigned long pfn)
if (!ms || !ms->page_cgroup)
return;
base = ms->page_cgroup + pfn;
- if (is_vmalloc_addr(base)) {
- vfree(base);
- ms->page_cgroup = NULL;
- } else {
- struct page *page = virt_to_page(base);
- if (!PageReserved(page)) { /* Is bootmem ? */
- kfree(base);
- ms->page_cgroup = NULL;
- }
- }
+ free__mcg_table(base);
+ ms->page_cgroup = NULL;
}
int __meminit online_page_cgroup(unsigned long start_pfn,
--
1.7.2.3
--
Michal Hocko
SUSE Labs
SUSE LINUX s.r.o.
Lihovarska 1060/12
190 00 Praha 9
Czech Republic
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@xxxxxxxxxx For more info on Linux MM,
see: http://www.linux-mm.org/ .
Fight unfair telecom internet charges in Canada: sign http://stopthemeter.ca/
Don't email: <a href=mailto:"dont@xxxxxxxxx";> email@xxxxxxxxx </a>