Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > linux.kernel > #1405970 > unrolled thread

[PATCH RESEND 0/8] More stuff to charge to kmemcg

Back to article view | Back to linux.kernel

Contents

  [PATCH RESEND 0/8] More stuff to charge to kmemcg Vladimir Davydov <vdavydov@virtuozzo.com> - 2016-05-24 11:10 +0200
    [PATCH RESEND 4/8] mm: charge/uncharge kmemcg from generic page allocator paths Vladimir Davydov <vdavydov@virtuozzo.com> - 2016-05-24 11:10 +0200
    [PATCH RESEND 5/8] mm: memcontrol: teach uncharge_list to deal with kmem pages Vladimir Davydov <vdavydov@virtuozzo.com> - 2016-05-24 11:10 +0200
    [PATCH RESEND 6/8] arch: x86: charge page tables to kmemcg Vladimir Davydov <vdavydov@virtuozzo.com> - 2016-05-24 11:30 +0200
    [PATCH RESEND 1/8] mm: remove pointless struct in struct page definition Vladimir Davydov <vdavydov@virtuozzo.com> - 2016-05-24 11:30 +0200
    [PATCH RESEND 8/8] af_unix: charge buffers to kmemcg Vladimir Davydov <vdavydov@virtuozzo.com> - 2016-05-24 12:30 +0200
      Re: [PATCH RESEND 8/8] af_unix: charge buffers to kmemcg Eric Dumazet <eric.dumazet@gmail.com> - 2016-05-24 15:10 +0200
        Re: [PATCH RESEND 8/8] af_unix: charge buffers to kmemcg Vladimir Davydov <vdavydov@virtuozzo.com> - 2016-05-24 18:40 +0200

#1405970 — [PATCH RESEND 0/8] More stuff to charge to kmemcg

[resending with all relevant lists in Cc]

Hi,

This patch implements per kmemcg accounting of page tables (x86-only),
pipe buffers, and unix socket buffers.

Basically, this is v2 of my earlier attempt [1], addressing comments by
Andrew, namely: lack of comments to non-standard _mapcount usage, extra
overhead even when kmemcg is unused, wrong handling of stolen pipe
buffer pages.

Patches 1-3 are just cleanups that are not supposed to introduce any
functional changes. Patches 4 and 5 move charge/uncharge to generic page
allocator paths for the sake of accounting pipe and unix socket buffers.
Patches 5-7 make x86 page tables, pipe buffers, and unix socket buffers
accountable.

[1] http://lkml.kernel.org/r/%3Ccover.1443262808.git.vdavydov@parallels.com%3E

Thanks,

Vladimir Davydov (8):
  mm: remove pointless struct in struct page definition
  mm: clean up non-standard page->_mapcount users
  mm: memcontrol: cleanup kmem charge functions
  mm: charge/uncharge kmemcg from generic page allocator paths
  mm: memcontrol: teach uncharge_list to deal with kmem pages
  arch: x86: charge page tables to kmemcg
  pipe: account to kmemcg
  af_unix: charge buffers to kmemcg

 arch/x86/include/asm/pgalloc.h |  12 ++++-
 arch/x86/mm/pgtable.c          |  11 ++--
 fs/pipe.c                      |  32 ++++++++---
 include/linux/gfp.h            |  10 +---
 include/linux/memcontrol.h     | 103 +++---------------------------------
 include/linux/mm_types.h       |  73 ++++++++++++-------------
 include/linux/page-flags.h     |  78 +++++++++++++--------------
 kernel/fork.c                  |   6 +--
 mm/memcontrol.c                | 117 ++++++++++++++++++++++++++++-------------
 mm/page_alloc.c                |  63 +++++-----------------
 mm/slab.h                      |  16 ++++--
 mm/slab_common.c               |   2 +-
 mm/slub.c                      |   6 +--
 mm/vmalloc.c                   |   6 +--
 net/unix/af_unix.c             |   1 +
 scripts/tags.sh                |   3 ++
 16 files changed, 245 insertions(+), 294 deletions(-)

-- 
2.1.4

[toc] | [next] | [standalone]

#1405971 — [PATCH RESEND 4/8] mm: charge/uncharge kmemcg from generic page allocator paths

Currently, to charge a non-slab allocation to kmemcg one has to use
alloc_kmem_pages helper with __GFP_ACCOUNT flag. A page allocated with
this helper should finally be freed using free_kmem_pages, otherwise it
won't be uncharged.

This API suits its current users fine, but it turns out to be impossible
to use along with page reference counting, i.e. when an allocation is
supposed to be freed with put_page, as it is the case with pipe or unix
socket buffers.

To overcome this limitation, this patch moves charging/uncharging to
generic page allocator paths, i.e. to __alloc_pages_nodemask and
free_pages_prepare, and zaps alloc/free_kmem_pages helpers. This way,
one can use any of the available page allocation functions to get the
allocated page charged to kmemcg - it's enough to pass __GFP_ACCOUNT,
just like in case of kmalloc and friends. A charged page will be
automatically uncharged on free.

To make it possible, we need to mark pages charged to kmemcg somehow. To
avoid introducing a new page flag, we make use of page->_mapcount for
marking such pages. Since pages charged to kmemcg are not supposed to be
mapped to userspace, it should work just fine. There are other (ab)users
of page->_mapcount - buddy and balloon pages - but we don't conflict
with them.

In case kmemcg is compiled out or not used at runtime, this patch
introduces no overhead to generic page allocator paths. If kmemcg is
used, it will be plus one gfp flags check on alloc and plus one
page->_mapcount check on free, which shouldn't hurt performance, because
the data accessed are hot.

Signed-off-by: Vladimir Davydov <vdavydov@virtuozzo.com>
---
 include/linux/gfp.h        | 10 +------
 include/linux/page-flags.h |  7 +++++
 kernel/fork.c              |  6 ++---
 mm/page_alloc.c            | 66 +++++++++-------------------------------------
 mm/slab_common.c           |  2 +-
 mm/slub.c                  |  6 ++---
 mm/vmalloc.c               |  6 ++---
 7 files changed, 31 insertions(+), 72 deletions(-)

diff --git a/include/linux/gfp.h b/include/linux/gfp.h
index 570383a41853..c29e9d347bc6 100644
--- a/include/linux/gfp.h
+++ b/include/linux/gfp.h
@@ -78,8 +78,7 @@ struct vm_area_struct;
  * __GFP_THISNODE forces the allocation to be satisified from the requested
  *   node with no fallbacks or placement policy enforcements.
  *
- * __GFP_ACCOUNT causes the allocation to be accounted to kmemcg (only relevant
- *   to kmem allocations).
+ * __GFP_ACCOUNT causes the allocation to be accounted to kmemcg.
  */
 #define __GFP_RECLAIMABLE ((__force gfp_t)___GFP_RECLAIMABLE)
 #define __GFP_WRITE	((__force gfp_t)___GFP_WRITE)
@@ -486,10 +485,6 @@ extern struct page *alloc_pages_vma(gfp_t gfp_mask, int order,
 #define alloc_page_vma_node(gfp_mask, vma, addr, node)		\
 	alloc_pages_vma(gfp_mask, 0, vma, addr, node, false)
 
-extern struct page *alloc_kmem_pages(gfp_t gfp_mask, unsigned int order);
-extern struct page *alloc_kmem_pages_node(int nid, gfp_t gfp_mask,
-					  unsigned int order);
-
 extern unsigned long __get_free_pages(gfp_t gfp_mask, unsigned int order);
 extern unsigned long get_zeroed_page(gfp_t gfp_mask);
 
@@ -513,9 +508,6 @@ extern void *__alloc_page_frag(struct page_frag_cache *nc,
 			       unsigned int fragsz, gfp_t gfp_mask);
 extern void __free_page_frag(void *addr);
 
-extern void __free_kmem_pages(struct page *page, unsigned int order);
-extern void free_kmem_pages(unsigned long addr, unsigned int order);
-
 #define __free_page(page) __free_pages((page), 0)
 #define free_page(addr) free_pages((addr), 0)
 
diff --git a/include/linux/page-flags.h b/include/linux/page-flags.h
index 9940ade6a25e..b51e75a47e82 100644
--- a/include/linux/page-flags.h
+++ b/include/linux/page-flags.h
@@ -630,6 +630,13 @@ PAGE_MAPCOUNT_OPS(Buddy, BUDDY)
 #define PAGE_BALLOON_MAPCOUNT_VALUE		(-256)
 PAGE_MAPCOUNT_OPS(Balloon, BALLOON)
 
+/*
+ * If kmemcg is enabled, the buddy allocator will set PageKmemcg() on
+ * pages allocated with __GFP_ACCOUNT. It gets cleared on page free.
+ */
+#define PAGE_KMEMCG_MAPCOUNT_VALUE		(-512)
+PAGE_MAPCOUNT_OPS(Kmemcg, KMEMCG)
+
 extern bool is_free_buddy_page(struct page *page);
 
 /*
diff --git a/kernel/fork.c b/kernel/fork.c
index 66cc2e0e137e..3f3c30f80786 100644
--- a/kernel/fork.c
+++ b/kernel/fork.c
@@ -162,8 +162,8 @@ void __weak arch_release_thread_info(struct thread_info *ti)
 static struct thread_info *alloc_thread_info_node(struct task_struct *tsk,
 						  int node)
 {
-	struct page *page = alloc_kmem_pages_node(node, THREADINFO_GFP,
-						  THREAD_SIZE_ORDER);
+	struct page *page = alloc_pages_node(node, THREADINFO_GFP,
+					     THREAD_SIZE_ORDER);
 
 	if (page)
 		memcg_kmem_update_page_stat(page, MEMCG_KERNEL_STACK,
@@ -178,7 +178,7 @@ static inline void free_thread_info(struct thread_info *ti)
 
 	memcg_kmem_update_page_stat(page, MEMCG_KERNEL_STACK,
 				    -(1 << THREAD_SIZE_ORDER));
-	__free_kmem_pages(page, THREAD_SIZE_ORDER);
+	__free_pages(page, THREAD_SIZE_ORDER);
 }
 # else
 static struct kmem_cache *thread_info_cache;
diff --git a/mm/page_alloc.c b/mm/page_alloc.c
index f948f4e02e89..c9ee98f1d3cc 100644
--- a/mm/page_alloc.c
+++ b/mm/page_alloc.c
@@ -63,6 +63,7 @@
 #include <linux/sched/rt.h>
 #include <linux/page_owner.h>
 #include <linux/kthread.h>
+#include <linux/memcontrol.h>
 
 #include <asm/sections.h>
 #include <asm/tlbflush.h>
@@ -1017,6 +1018,10 @@ static __always_inline bool free_pages_prepare(struct page *page,
 	}
 	if (PageAnonHead(page))
 		page->mapping = NULL;
+	if (memcg_kmem_enabled() && PageKmemcg(page)) {
+		memcg_kmem_uncharge(page, order);
+		__ClearPageKmemcg(page);
+	}
 	if (check_free)
 		bad += free_pages_check(page);
 	if (bad)
@@ -3833,6 +3838,14 @@ no_zone:
 	}
 
 out:
+	if (memcg_kmem_enabled() && (gfp_mask & __GFP_ACCOUNT) && page) {
+		if (unlikely(memcg_kmem_charge(page, gfp_mask, order))) {
+			__free_pages(page, order);
+			page = NULL;
+		} else
+			__SetPageKmemcg(page);
+	}
+
 	if (kmemcheck_enabled && page)
 		kmemcheck_pagealloc_alloc(page, order, gfp_mask);
 
@@ -3988,59 +4001,6 @@ void __free_page_frag(void *addr)
 }
 EXPORT_SYMBOL(__free_page_frag);
 
-/*
- * alloc_kmem_pages charges newly allocated pages to the kmem resource counter
- * of the current memory cgroup if __GFP_ACCOUNT is set, other than that it is
- * equivalent to alloc_pages.
- *
- * It should be used when the caller would like to use kmalloc, but since the
- * allocation is large, it has to fall back to the page allocator.
- */
-struct page *alloc_kmem_pages(gfp_t gfp_mask, unsigned int order)
-{
-	struct page *page;
-
-	page = alloc_pages(gfp_mask, order);
-	if (memcg_kmem_enabled() && (gfp_mask & __GFP_ACCOUNT) &&
-	    page && memcg_kmem_charge(page, gfp_mask, order) != 0) {
-		__free_pages(page, order);
-		page = NULL;
-	}
-	return page;
-}
-
-struct page *alloc_kmem_pages_node(int nid, gfp_t gfp_mask, unsigned int order)
-{
-	struct page *page;
-
-	page = alloc_pages_node(nid, gfp_mask, order);
-	if (memcg_kmem_enabled() && (gfp_mask & __GFP_ACCOUNT) &&
-	    page && memcg_kmem_charge(page, gfp_mask, order) != 0) {
-		__free_pages(page, order);
-		page = NULL;
-	}
-	return page;
-}
-
-/*
- * __free_kmem_pages and free_kmem_pages will free pages allocated with
- * alloc_kmem_pages.
- */
-void __free_kmem_pages(struct page *page, unsigned int order)
-{
-	if (memcg_kmem_enabled())
-		memcg_kmem_uncharge(page, order);
-	__free_pages(page, order);
-}
-
-void free_kmem_pages(unsigned long addr, unsigned int order)
-{
-	if (addr != 0) {
-		VM_BUG_ON(!virt_addr_valid((void *)addr));
-		__free_kmem_pages(virt_to_page((void *)addr), order);
-	}
-}
-
 static void *make_alloc_exact(unsigned long addr, unsigned int order,
 		size_t size)
 {
diff --git a/mm/slab_common.c b/mm/slab_common.c
index a65dad7fdcd1..3bf8f63a6abe 100644
--- a/mm/slab_common.c
+++ b/mm/slab_common.c
@@ -1012,7 +1012,7 @@ void *kmalloc_order(size_t size, gfp_t flags, unsigned int order)
 	struct page *page;
 
 	flags |= __GFP_COMP;
-	page = alloc_kmem_pages(flags, order);
+	page = alloc_pages(flags, order);
 	ret = page ? page_address(page) : NULL;
 	kmemleak_alloc(ret, size, 1, flags);
 	kasan_kmalloc_large(ret, size, flags);
diff --git a/mm/slub.c b/mm/slub.c
index 538c8584ba8d..ac7b5a55e4c9 100644
--- a/mm/slub.c
+++ b/mm/slub.c
@@ -2867,7 +2867,7 @@ int build_detached_freelist(struct kmem_cache *s, size_t size,
 		if (unlikely(!PageSlab(page))) {
 			BUG_ON(!PageCompound(page));
 			kfree_hook(object);
-			__free_kmem_pages(page, compound_order(page));
+			__free_pages(page, compound_order(page));
 			p[size] = NULL; /* mark object processed */
 			return size;
 		}
@@ -3575,7 +3575,7 @@ static void *kmalloc_large_node(size_t size, gfp_t flags, int node)
 	void *ptr = NULL;
 
 	flags |= __GFP_COMP | __GFP_NOTRACK;
-	page = alloc_kmem_pages_node(node, flags, get_order(size));
+	page = alloc_pages_node(node, flags, get_order(size));
 	if (page)
 		ptr = page_address(page);
 
@@ -3656,7 +3656,7 @@ void kfree(const void *x)
 	if (unlikely(!PageSlab(page))) {
 		BUG_ON(!PageCompound(page));
 		kfree_hook(x);
-		__free_kmem_pages(page, compound_order(page));
+		__free_pages(page, compound_order(page));
 		return;
 	}
 	slab_free(page->slab_cache, page, object, NULL, 1, _RET_IP_);
diff --git a/mm/vmalloc.c b/mm/vmalloc.c
index cf7ad1a53be0..a806236ecd62 100644
--- a/mm/vmalloc.c
+++ b/mm/vmalloc.c
@@ -1501,7 +1501,7 @@ static void __vunmap(const void *addr, int deallocate_pages)
 			struct page *page = area->pages[i];
 
 			BUG_ON(!page);
-			__free_kmem_pages(page, 0);
+			__free_pages(page, 0);
 		}
 
 		kvfree(area->pages);
@@ -1628,9 +1628,9 @@ static void *__vmalloc_area_node(struct vm_struct *area, gfp_t gfp_mask,
 		struct page *page;
 
 		if (node == NUMA_NO_NODE)
-			page = alloc_kmem_pages(alloc_mask, order);
+			page = alloc_pages(alloc_mask, order);
 		else
-			page = alloc_kmem_pages_node(node, alloc_mask, order);
+			page = alloc_pages_node(node, alloc_mask, order);
 
 		if (unlikely(!page)) {
 			/* Successfully allocated i pages, free them in __vunmap() */
-- 
2.1.4

[toc] | [prev] | [next] | [standalone]

#1405972 — [PATCH RESEND 5/8] mm: memcontrol: teach uncharge_list to deal with kmem pages

Page table pages are batched-freed in release_pages on most
architectures. If we want to charge them to kmemcg (this is what is done
later in this series), we need to teach mem_cgroup_uncharge_list to
handle kmem pages.

Signed-off-by: Vladimir Davydov <vdavydov@virtuozzo.com>
---
 mm/memcontrol.c | 42 ++++++++++++++++++++++++------------------
 1 file changed, 24 insertions(+), 18 deletions(-)

diff --git a/mm/memcontrol.c b/mm/memcontrol.c
index 482b4a0c97e4..89a421ee4713 100644
--- a/mm/memcontrol.c
+++ b/mm/memcontrol.c
@@ -5432,15 +5432,18 @@ void mem_cgroup_cancel_charge(struct page *page, struct mem_cgroup *memcg,
 
 static void uncharge_batch(struct mem_cgroup *memcg, unsigned long pgpgout,
 			   unsigned long nr_anon, unsigned long nr_file,
-			   unsigned long nr_huge, struct page *dummy_page)
+			   unsigned long nr_huge, unsigned long nr_kmem,
+			   struct page *dummy_page)
 {
-	unsigned long nr_pages = nr_anon + nr_file;
+	unsigned long nr_pages = nr_anon + nr_file + nr_kmem;
 	unsigned long flags;
 
 	if (!mem_cgroup_is_root(memcg)) {
 		page_counter_uncharge(&memcg->memory, nr_pages);
 		if (do_memsw_account())
 			page_counter_uncharge(&memcg->memsw, nr_pages);
+		if (!cgroup_subsys_on_dfl(memory_cgrp_subsys) && nr_kmem)
+			page_counter_uncharge(&memcg->kmem, nr_kmem);
 		memcg_oom_recover(memcg);
 	}
 
@@ -5463,6 +5466,7 @@ static void uncharge_list(struct list_head *page_list)
 	unsigned long nr_anon = 0;
 	unsigned long nr_file = 0;
 	unsigned long nr_huge = 0;
+	unsigned long nr_kmem = 0;
 	unsigned long pgpgout = 0;
 	struct list_head *next;
 	struct page *page;
@@ -5473,8 +5477,6 @@ static void uncharge_list(struct list_head *page_list)
 	 */
 	next = page_list->next;
 	do {
-		unsigned int nr_pages = 1;
-
 		page = list_entry(next, struct page, lru);
 		next = page->lru.next;
 
@@ -5493,31 +5495,35 @@ static void uncharge_list(struct list_head *page_list)
 		if (memcg != page->mem_cgroup) {
 			if (memcg) {
 				uncharge_batch(memcg, pgpgout, nr_anon, nr_file,
-					       nr_huge, page);
-				pgpgout = nr_anon = nr_file = nr_huge = 0;
+					       nr_huge, nr_kmem, page);
+				pgpgout = nr_anon = nr_file =
+					nr_huge = nr_kmem = 0;
 			}
 			memcg = page->mem_cgroup;
 		}
 
-		if (PageTransHuge(page)) {
-			nr_pages <<= compound_order(page);
-			VM_BUG_ON_PAGE(!PageTransHuge(page), page);
-			nr_huge += nr_pages;
-		}
+		if (!PageKmemcg(page)) {
+			unsigned int nr_pages = 1;
 
-		if (PageAnon(page))
-			nr_anon += nr_pages;
-		else
-			nr_file += nr_pages;
+			if (PageTransHuge(page)) {
+				nr_pages <<= compound_order(page);
+				VM_BUG_ON_PAGE(!PageTransHuge(page), page);
+				nr_huge += nr_pages;
+			}
+			if (PageAnon(page))
+				nr_anon += nr_pages;
+			else
+				nr_file += nr_pages;
+			pgpgout++;
+		} else
+			nr_kmem += 1 << compound_order(page);
 
 		page->mem_cgroup = NULL;
-
-		pgpgout++;
 	} while (next != page_list);
 
 	if (memcg)
 		uncharge_batch(memcg, pgpgout, nr_anon, nr_file,
-			       nr_huge, page);
+			       nr_huge, nr_kmem, page);
 }
 
 /**
-- 
2.1.4

[toc] | [prev] | [next] | [standalone]

#1405988 — [PATCH RESEND 6/8] arch: x86: charge page tables to kmemcg

Page tables can bite a relatively big chunk off system memory and their
allocations are easy to trigger from userspace, so they should be
accounted to kmemcg.

This patch marks page table allocations as __GFP_ACCOUNT for x86. Note
we must not charge allocations of kernel page tables, because they can
be shared among processes from different cgroups so accounting them to a
particular one can pin other cgroups for indefinitely long. So we clear
__GFP_ACCOUNT flag if a page table is allocated for the kernel.

Signed-off-by: Vladimir Davydov <vdavydov@virtuozzo.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: "H. Peter Anvin" <hpa@zytor.com>
---
 arch/x86/include/asm/pgalloc.h | 12 ++++++++++--
 arch/x86/mm/pgtable.c          | 11 ++++++++---
 2 files changed, 18 insertions(+), 5 deletions(-)

diff --git a/arch/x86/include/asm/pgalloc.h b/arch/x86/include/asm/pgalloc.h
index bf7f8b55b0f9..2f531633cb16 100644
--- a/arch/x86/include/asm/pgalloc.h
+++ b/arch/x86/include/asm/pgalloc.h
@@ -81,7 +81,11 @@ static inline void pmd_populate(struct mm_struct *mm, pmd_t *pmd,
 static inline pmd_t *pmd_alloc_one(struct mm_struct *mm, unsigned long addr)
 {
 	struct page *page;
-	page = alloc_pages(GFP_KERNEL | __GFP_REPEAT | __GFP_ZERO, 0);
+	gfp_t gfp = GFP_KERNEL_ACCOUNT | __GFP_REPEAT | __GFP_ZERO;
+
+	if (mm == &init_mm)
+		gfp &= ~__GFP_ACCOUNT;
+	page = alloc_pages(gfp, 0);
 	if (!page)
 		return NULL;
 	if (!pgtable_pmd_page_ctor(page)) {
@@ -125,7 +129,11 @@ static inline void pgd_populate(struct mm_struct *mm, pgd_t *pgd, pud_t *pud)
 
 static inline pud_t *pud_alloc_one(struct mm_struct *mm, unsigned long addr)
 {
-	return (pud_t *)get_zeroed_page(GFP_KERNEL|__GFP_REPEAT);
+	gfp_t gfp = GFP_KERNEL_ACCOUNT | __GFP_REPEAT;
+
+	if (mm == &init_mm)
+		gfp &= ~__GFP_ACCOUNT;
+	return (pud_t *)get_zeroed_page(gfp);
 }
 
 static inline void pud_free(struct mm_struct *mm, pud_t *pud)
diff --git a/arch/x86/mm/pgtable.c b/arch/x86/mm/pgtable.c
index 4eb287e25043..421ac6b74d11 100644
--- a/arch/x86/mm/pgtable.c
+++ b/arch/x86/mm/pgtable.c
@@ -6,7 +6,8 @@
 #include <asm/fixmap.h>
 #include <asm/mtrr.h>
 
-#define PGALLOC_GFP GFP_KERNEL | __GFP_NOTRACK | __GFP_REPEAT | __GFP_ZERO
+#define PGALLOC_GFP (GFP_KERNEL_ACCOUNT | __GFP_NOTRACK | __GFP_REPEAT | \
+		     __GFP_ZERO)
 
 #ifdef CONFIG_HIGHPTE
 #define PGALLOC_USER_GFP __GFP_HIGHMEM
@@ -18,7 +19,7 @@ gfp_t __userpte_alloc_gfp = PGALLOC_GFP | PGALLOC_USER_GFP;
 
 pte_t *pte_alloc_one_kernel(struct mm_struct *mm, unsigned long address)
 {
-	return (pte_t *)__get_free_page(PGALLOC_GFP);
+	return (pte_t *)__get_free_page(PGALLOC_GFP & ~__GFP_ACCOUNT);
 }
 
 pgtable_t pte_alloc_one(struct mm_struct *mm, unsigned long address)
@@ -207,9 +208,13 @@ static int preallocate_pmds(struct mm_struct *mm, pmd_t *pmds[])
 {
 	int i;
 	bool failed = false;
+	gfp_t gfp = PGALLOC_GFP;
+
+	if (mm == &init_mm)
+		gfp &= ~__GFP_ACCOUNT;
 
 	for(i = 0; i < PREALLOCATED_PMDS; i++) {
-		pmd_t *pmd = (pmd_t *)__get_free_page(PGALLOC_GFP);
+		pmd_t *pmd = (pmd_t *)__get_free_page(gfp);
 		if (!pmd)
 			failed = true;
 		if (pmd && !pgtable_pmd_page_ctor(virt_to_page(pmd))) {
-- 
2.1.4

[toc] | [prev] | [next] | [standalone]

#1405990 — [PATCH RESEND 1/8] mm: remove pointless struct in struct page definition

... to reduce indentation level thus leaving more space for comments.

Signed-off-by: Vladimir Davydov <vdavydov@virtuozzo.com>
---
 include/linux/mm_types.h | 68 +++++++++++++++++++++++-------------------------
 1 file changed, 32 insertions(+), 36 deletions(-)

diff --git a/include/linux/mm_types.h b/include/linux/mm_types.h
index d553855503e6..3cc5977a9cab 100644
--- a/include/linux/mm_types.h
+++ b/include/linux/mm_types.h
@@ -60,51 +60,47 @@ struct page {
 	};
 
 	/* Second double word */
-	struct {
-		union {
-			pgoff_t index;		/* Our offset within mapping. */
-			void *freelist;		/* sl[aou]b first free object */
-			/* page_deferred_list().prev	-- second tail page */
-		};
+	union {
+		pgoff_t index;		/* Our offset within mapping. */
+		void *freelist;		/* sl[aou]b first free object */
+		/* page_deferred_list().prev	-- second tail page */
+	};
 
-		union {
+	union {
 #if defined(CONFIG_HAVE_CMPXCHG_DOUBLE) && \
 	defined(CONFIG_HAVE_ALIGNED_STRUCT_PAGE)
-			/* Used for cmpxchg_double in slub */
-			unsigned long counters;
+		/* Used for cmpxchg_double in slub */
+		unsigned long counters;
 #else
-			/*
-			 * Keep _refcount separate from slub cmpxchg_double
-			 * data.  As the rest of the double word is protected by
-			 * slab_lock but _refcount is not.
-			 */
-			unsigned counters;
+		/*
+		 * Keep _refcount separate from slub cmpxchg_double data.
+		 * As the rest of the double word is protected by slab_lock
+		 * but _refcount is not.
+		 */
+		unsigned counters;
 #endif
+		struct {
 
-			struct {
-
-				union {
-					/*
-					 * Count of ptes mapped in mms, to show
-					 * when page is mapped & limit reverse
-					 * map searches.
-					 */
-					atomic_t _mapcount;
-
-					struct { /* SLUB */
-						unsigned inuse:16;
-						unsigned objects:15;
-						unsigned frozen:1;
-					};
-					int units;	/* SLOB */
-				};
+			union {
 				/*
-				 * Usage count, *USE WRAPPER FUNCTION*
-				 * when manual accounting. See page_ref.h
+				 * Count of ptes mapped in mms, to show when
+				 * page is mapped & limit reverse map searches.
 				 */
-				atomic_t _refcount;
+				atomic_t _mapcount;
+
+				unsigned int active;		/* SLAB */
+				struct {			/* SLUB */
+					unsigned inuse:16;
+					unsigned objects:15;
+					unsigned frozen:1;
+				};
+				int units;			/* SLOB */
 			};
-			unsigned int active;	/* SLAB */
+			/*
+			 * Usage count, *USE WRAPPER FUNCTION* when manual
+			 * accounting. See page_ref.h
+			 */
+			atomic_t _refcount;
 		};
 	};
 
-- 
2.1.4

[toc] | [prev] | [next] | [standalone]

#1406047 — [PATCH RESEND 8/8] af_unix: charge buffers to kmemcg

Unix sockets can consume a significant amount of system memory, hence
they should be accounted to kmemcg.

Since unix socket buffers are always allocated from process context,
all we need to do to charge them to kmemcg is set __GFP_ACCOUNT in
sock->sk_allocation mask.

Signed-off-by: Vladimir Davydov <vdavydov@virtuozzo.com>
Cc: "David S. Miller" <davem@davemloft.net>
---
 net/unix/af_unix.c | 1 +
 1 file changed, 1 insertion(+)

diff --git a/net/unix/af_unix.c b/net/unix/af_unix.c
index 80aa6a3e6817..022bdd3ab7d9 100644
--- a/net/unix/af_unix.c
+++ b/net/unix/af_unix.c
@@ -769,6 +769,7 @@ static struct sock *unix_create1(struct net *net, struct socket *sock, int kern)
 	lockdep_set_class(&sk->sk_receive_queue.lock,
 				&af_unix_sk_receive_queue_lock_key);
 
+	sk->sk_allocation	= GFP_KERNEL_ACCOUNT;
 	sk->sk_write_space	= unix_write_space;
 	sk->sk_max_ack_backlog	= net->unx.sysctl_max_dgram_qlen;
 	sk->sk_destruct		= unix_sock_destructor;
-- 
2.1.4

[toc] | [prev] | [next] | [standalone]

#1406144 — Re: [PATCH RESEND 8/8] af_unix: charge buffers to kmemcg

On Tue, 2016-05-24 at 11:49 +0300, Vladimir Davydov wrote:
> Unix sockets can consume a significant amount of system memory, hence
> they should be accounted to kmemcg.
> 
> Since unix socket buffers are always allocated from process context,
> all we need to do to charge them to kmemcg is set __GFP_ACCOUNT in
> sock->sk_allocation mask.

I have two questions : 

1) What happens when a buffer, allocated from socket <A> lands in a
different socket <B>, maybe owned by another user/process.

Who owns it now, in term of kmemcg accounting ?

2) Has performance impact been evaluated ?

Thanks.

[toc] | [prev] | [next] | [standalone]

#1406273 — Re: [PATCH RESEND 8/8] af_unix: charge buffers to kmemcg

On Tue, May 24, 2016 at 06:02:06AM -0700, Eric Dumazet wrote:
> On Tue, 2016-05-24 at 11:49 +0300, Vladimir Davydov wrote:
> > Unix sockets can consume a significant amount of system memory, hence
> > they should be accounted to kmemcg.
> > 
> > Since unix socket buffers are always allocated from process context,
> > all we need to do to charge them to kmemcg is set __GFP_ACCOUNT in
> > sock->sk_allocation mask.
> 
> I have two questions : 
> 
> 1) What happens when a buffer, allocated from socket <A> lands in a
> different socket <B>, maybe owned by another user/process.
> 
> Who owns it now, in term of kmemcg accounting ?

We never move memcg charges. E.g. if two processes from different
cgroups are sharing a memory region, each page will be charged to the
process which touched it first. Or if two processes are working with the
same directory tree, inodes and dentries will be charged to the first
user. The same is fair for unix socket buffers - they will be charged to
the sender.

> 
> 2) Has performance impact been evaluated ?

I ran netperf STREAM_STREAM with default options in a kmemcg on
a 4 core x 2 HT box. The results are below:

 # clients            bandwidth (10^6bits/sec)
                    base              patched
         1      67643 +-  725      64874 +-  353    - 4.0 %
         4     193585 +- 2516     186715 +- 1460    - 3.5 %
         8     194820 +-  377     187443 +- 1229    - 3.7 %

So the accounting doesn't come for free - it takes ~4% of performance.
I believe we could optimize it by using per cpu batching not only on
charge, but also on uncharge in memcg core, but that's beyond the scope
of this patch set - I'll take a look at this later.

Anyway, if performance impact is found to be unacceptable, it is always
possible to disable kmem accounting at boot time (cgroup.memory=nokmem)
or not use memory cgroups at runtime at all (thanks to jump labels
there'll be no overhead even if they are compiled in).

Thanks,
Vladimir

[toc] | [prev] | [standalone]

Back to top | Article view | linux.kernel

csiph-web