aboutsummaryrefslogtreecommitdiffstats
diff options
context:
space:
mode:
-rw-r--r--Documentation/cgroups/memory.txt4
-rw-r--r--arch/ia64/mm/init.c1
-rw-r--r--arch/mips/fw/arc/misc.c1
-rw-r--r--drivers/rapidio/rio.c2
-rw-r--r--include/linux/mm.h4
-rw-r--r--include/linux/mmzone.h2
-rw-r--r--include/linux/rio.h2
-rw-r--r--mm/bootmem.c10
-rw-r--r--mm/highmem.c2
-rw-r--r--mm/memcontrol.c67
-rw-r--r--mm/memory.c10
-rw-r--r--mm/memory_hotplug.c7
-rw-r--r--mm/mmap.c2
-rw-r--r--mm/mmzone.c6
-rw-r--r--mm/nobootmem.c3
-rw-r--r--mm/page_alloc.c36
-rw-r--r--mm/shmem.c18
-rw-r--r--mm/swapfile.c4
-rw-r--r--mm/vmscan.c25
19 files changed, 86 insertions, 120 deletions
diff --git a/Documentation/cgroups/memory.txt b/Documentation/cgroups/memory.txt
index c07f7b4fb88d..71c4da413444 100644
--- a/Documentation/cgroups/memory.txt
+++ b/Documentation/cgroups/memory.txt
@@ -466,6 +466,10 @@ Note:
4665.3 swappiness 4665.3 swappiness
467 467
468Similar to /proc/sys/vm/swappiness, but affecting a hierarchy of groups only. 468Similar to /proc/sys/vm/swappiness, but affecting a hierarchy of groups only.
469Please note that unlike the global swappiness, memcg knob set to 0
470really prevents from any swapping even if there is a swap storage
471available. This might lead to memcg OOM killer if there are no file
472pages to reclaim.
469 473
470Following cgroups' swappiness can't be changed. 474Following cgroups' swappiness can't be changed.
471- root cgroup (uses /proc/sys/vm/swappiness). 475- root cgroup (uses /proc/sys/vm/swappiness).
diff --git a/arch/ia64/mm/init.c b/arch/ia64/mm/init.c
index acd5b68e8871..082e383c1b6f 100644
--- a/arch/ia64/mm/init.c
+++ b/arch/ia64/mm/init.c
@@ -637,7 +637,6 @@ mem_init (void)
637 637
638 high_memory = __va(max_low_pfn * PAGE_SIZE); 638 high_memory = __va(max_low_pfn * PAGE_SIZE);
639 639
640 reset_zone_present_pages();
641 for_each_online_pgdat(pgdat) 640 for_each_online_pgdat(pgdat)
642 if (pgdat->bdata->node_bootmem_map) 641 if (pgdat->bdata->node_bootmem_map)
643 totalram_pages += free_all_bootmem_node(pgdat); 642 totalram_pages += free_all_bootmem_node(pgdat);
diff --git a/arch/mips/fw/arc/misc.c b/arch/mips/fw/arc/misc.c
index 7cf80ca2c1d2..f9f5307434c2 100644
--- a/arch/mips/fw/arc/misc.c
+++ b/arch/mips/fw/arc/misc.c
@@ -11,6 +11,7 @@
11 */ 11 */
12#include <linux/init.h> 12#include <linux/init.h>
13#include <linux/kernel.h> 13#include <linux/kernel.h>
14#include <linux/irqflags.h>
14 15
15#include <asm/bcache.h> 16#include <asm/bcache.h>
16 17
diff --git a/drivers/rapidio/rio.c b/drivers/rapidio/rio.c
index c17ae22567e0..0c6fcb461faf 100644
--- a/drivers/rapidio/rio.c
+++ b/drivers/rapidio/rio.c
@@ -401,7 +401,7 @@ EXPORT_SYMBOL_GPL(rio_release_inb_pwrite);
401/** 401/**
402 * rio_map_inb_region -- Map inbound memory region. 402 * rio_map_inb_region -- Map inbound memory region.
403 * @mport: Master port. 403 * @mport: Master port.
404 * @lstart: physical address of memory region to be mapped 404 * @local: physical address of memory region to be mapped
405 * @rbase: RIO base address assigned to this window 405 * @rbase: RIO base address assigned to this window
406 * @size: Size of the memory region 406 * @size: Size of the memory region
407 * @rflags: Flags for mapping. 407 * @rflags: Flags for mapping.
diff --git a/include/linux/mm.h b/include/linux/mm.h
index fa0680402738..bcaab4e6fe91 100644
--- a/include/linux/mm.h
+++ b/include/linux/mm.h
@@ -1684,9 +1684,5 @@ static inline unsigned int debug_guardpage_minorder(void) { return 0; }
1684static inline bool page_is_guard(struct page *page) { return false; } 1684static inline bool page_is_guard(struct page *page) { return false; }
1685#endif /* CONFIG_DEBUG_PAGEALLOC */ 1685#endif /* CONFIG_DEBUG_PAGEALLOC */
1686 1686
1687extern void reset_zone_present_pages(void);
1688extern void fixup_zone_present_pages(int nid, unsigned long start_pfn,
1689 unsigned long end_pfn);
1690
1691#endif /* __KERNEL__ */ 1687#endif /* __KERNEL__ */
1692#endif /* _LINUX_MM_H */ 1688#endif /* _LINUX_MM_H */
diff --git a/include/linux/mmzone.h b/include/linux/mmzone.h
index 50aaca81f63d..a23923ba8263 100644
--- a/include/linux/mmzone.h
+++ b/include/linux/mmzone.h
@@ -752,7 +752,7 @@ extern int init_currently_empty_zone(struct zone *zone, unsigned long start_pfn,
752 unsigned long size, 752 unsigned long size,
753 enum memmap_context context); 753 enum memmap_context context);
754 754
755extern void lruvec_init(struct lruvec *lruvec, struct zone *zone); 755extern void lruvec_init(struct lruvec *lruvec);
756 756
757static inline struct zone *lruvec_zone(struct lruvec *lruvec) 757static inline struct zone *lruvec_zone(struct lruvec *lruvec)
758{ 758{
diff --git a/include/linux/rio.h b/include/linux/rio.h
index 4187da511006..a3e784278667 100644
--- a/include/linux/rio.h
+++ b/include/linux/rio.h
@@ -275,9 +275,11 @@ struct rio_id_table {
275 * struct rio_net - RIO network info 275 * struct rio_net - RIO network info
276 * @node: Node in global list of RIO networks 276 * @node: Node in global list of RIO networks
277 * @devices: List of devices in this network 277 * @devices: List of devices in this network
278 * @switches: List of switches in this netowrk
278 * @mports: List of master ports accessing this network 279 * @mports: List of master ports accessing this network
279 * @hport: Default port for accessing this network 280 * @hport: Default port for accessing this network
280 * @id: RIO network ID 281 * @id: RIO network ID
282 * @destid_table: destID allocation table
281 */ 283 */
282struct rio_net { 284struct rio_net {
283 struct list_head node; /* node in list of networks */ 285 struct list_head node; /* node in list of networks */
diff --git a/mm/bootmem.c b/mm/bootmem.c
index 434be4ae7a04..f468185b3b28 100644
--- a/mm/bootmem.c
+++ b/mm/bootmem.c
@@ -198,8 +198,6 @@ static unsigned long __init free_all_bootmem_core(bootmem_data_t *bdata)
198 int order = ilog2(BITS_PER_LONG); 198 int order = ilog2(BITS_PER_LONG);
199 199
200 __free_pages_bootmem(pfn_to_page(start), order); 200 __free_pages_bootmem(pfn_to_page(start), order);
201 fixup_zone_present_pages(page_to_nid(pfn_to_page(start)),
202 start, start + BITS_PER_LONG);
203 count += BITS_PER_LONG; 201 count += BITS_PER_LONG;
204 start += BITS_PER_LONG; 202 start += BITS_PER_LONG;
205 } else { 203 } else {
@@ -210,9 +208,6 @@ static unsigned long __init free_all_bootmem_core(bootmem_data_t *bdata)
210 if (vec & 1) { 208 if (vec & 1) {
211 page = pfn_to_page(start + off); 209 page = pfn_to_page(start + off);
212 __free_pages_bootmem(page, 0); 210 __free_pages_bootmem(page, 0);
213 fixup_zone_present_pages(
214 page_to_nid(page),
215 start + off, start + off + 1);
216 count++; 211 count++;
217 } 212 }
218 vec >>= 1; 213 vec >>= 1;
@@ -226,11 +221,8 @@ static unsigned long __init free_all_bootmem_core(bootmem_data_t *bdata)
226 pages = bdata->node_low_pfn - bdata->node_min_pfn; 221 pages = bdata->node_low_pfn - bdata->node_min_pfn;
227 pages = bootmem_bootmap_pages(pages); 222 pages = bootmem_bootmap_pages(pages);
228 count += pages; 223 count += pages;
229 while (pages--) { 224 while (pages--)
230 fixup_zone_present_pages(page_to_nid(page),
231 page_to_pfn(page), page_to_pfn(page) + 1);
232 __free_pages_bootmem(page++, 0); 225 __free_pages_bootmem(page++, 0);
233 }
234 226
235 bdebug("nid=%td released=%lx\n", bdata - bootmem_node_data, count); 227 bdebug("nid=%td released=%lx\n", bdata - bootmem_node_data, count);
236 228
diff --git a/mm/highmem.c b/mm/highmem.c
index d517cd16a6eb..2da13a5c50e2 100644
--- a/mm/highmem.c
+++ b/mm/highmem.c
@@ -98,7 +98,7 @@ struct page *kmap_to_page(void *vaddr)
98{ 98{
99 unsigned long addr = (unsigned long)vaddr; 99 unsigned long addr = (unsigned long)vaddr;
100 100
101 if (addr >= PKMAP_ADDR(0) && addr <= PKMAP_ADDR(LAST_PKMAP)) { 101 if (addr >= PKMAP_ADDR(0) && addr < PKMAP_ADDR(LAST_PKMAP)) {
102 int i = (addr - PKMAP_ADDR(0)) >> PAGE_SHIFT; 102 int i = (addr - PKMAP_ADDR(0)) >> PAGE_SHIFT;
103 return pte_page(pkmap_page_table[i]); 103 return pte_page(pkmap_page_table[i]);
104 } 104 }
diff --git a/mm/memcontrol.c b/mm/memcontrol.c
index 7acf43bf04a2..dd39ba000b31 100644
--- a/mm/memcontrol.c
+++ b/mm/memcontrol.c
@@ -1055,12 +1055,24 @@ struct lruvec *mem_cgroup_zone_lruvec(struct zone *zone,
1055 struct mem_cgroup *memcg) 1055 struct mem_cgroup *memcg)
1056{ 1056{
1057 struct mem_cgroup_per_zone *mz; 1057 struct mem_cgroup_per_zone *mz;
1058 struct lruvec *lruvec;
1058 1059
1059 if (mem_cgroup_disabled()) 1060 if (mem_cgroup_disabled()) {
1060 return &zone->lruvec; 1061 lruvec = &zone->lruvec;
1062 goto out;
1063 }
1061 1064
1062 mz = mem_cgroup_zoneinfo(memcg, zone_to_nid(zone), zone_idx(zone)); 1065 mz = mem_cgroup_zoneinfo(memcg, zone_to_nid(zone), zone_idx(zone));
1063 return &mz->lruvec; 1066 lruvec = &mz->lruvec;
1067out:
1068 /*
1069 * Since a node can be onlined after the mem_cgroup was created,
1070 * we have to be prepared to initialize lruvec->zone here;
1071 * and if offlined then reonlined, we need to reinitialize it.
1072 */
1073 if (unlikely(lruvec->zone != zone))
1074 lruvec->zone = zone;
1075 return lruvec;
1064} 1076}
1065 1077
1066/* 1078/*
@@ -1087,9 +1099,12 @@ struct lruvec *mem_cgroup_page_lruvec(struct page *page, struct zone *zone)
1087 struct mem_cgroup_per_zone *mz; 1099 struct mem_cgroup_per_zone *mz;
1088 struct mem_cgroup *memcg; 1100 struct mem_cgroup *memcg;
1089 struct page_cgroup *pc; 1101 struct page_cgroup *pc;
1102 struct lruvec *lruvec;
1090 1103
1091 if (mem_cgroup_disabled()) 1104 if (mem_cgroup_disabled()) {
1092 return &zone->lruvec; 1105 lruvec = &zone->lruvec;
1106 goto out;
1107 }
1093 1108
1094 pc = lookup_page_cgroup(page); 1109 pc = lookup_page_cgroup(page);
1095 memcg = pc->mem_cgroup; 1110 memcg = pc->mem_cgroup;
@@ -1107,7 +1122,16 @@ struct lruvec *mem_cgroup_page_lruvec(struct page *page, struct zone *zone)
1107 pc->mem_cgroup = memcg = root_mem_cgroup; 1122 pc->mem_cgroup = memcg = root_mem_cgroup;
1108 1123
1109 mz = page_cgroup_zoneinfo(memcg, page); 1124 mz = page_cgroup_zoneinfo(memcg, page);
1110 return &mz->lruvec; 1125 lruvec = &mz->lruvec;
1126out:
1127 /*
1128 * Since a node can be onlined after the mem_cgroup was created,
1129 * we have to be prepared to initialize lruvec->zone here;
1130 * and if offlined then reonlined, we need to reinitialize it.
1131 */
1132 if (unlikely(lruvec->zone != zone))
1133 lruvec->zone = zone;
1134 return lruvec;
1111} 1135}
1112 1136
1113/** 1137/**
@@ -1452,17 +1476,26 @@ static int mem_cgroup_count_children(struct mem_cgroup *memcg)
1452static u64 mem_cgroup_get_limit(struct mem_cgroup *memcg) 1476static u64 mem_cgroup_get_limit(struct mem_cgroup *memcg)
1453{ 1477{
1454 u64 limit; 1478 u64 limit;
1455 u64 memsw;
1456 1479
1457 limit = res_counter_read_u64(&memcg->res, RES_LIMIT); 1480 limit = res_counter_read_u64(&memcg->res, RES_LIMIT);
1458 limit += total_swap_pages << PAGE_SHIFT;
1459 1481
1460 memsw = res_counter_read_u64(&memcg->memsw, RES_LIMIT);
1461 /* 1482 /*
1462 * If memsw is finite and limits the amount of swap space available 1483 * Do not consider swap space if we cannot swap due to swappiness
1463 * to this memcg, return that limit.
1464 */ 1484 */
1465 return min(limit, memsw); 1485 if (mem_cgroup_swappiness(memcg)) {
1486 u64 memsw;
1487
1488 limit += total_swap_pages << PAGE_SHIFT;
1489 memsw = res_counter_read_u64(&memcg->memsw, RES_LIMIT);
1490
1491 /*
1492 * If memsw is finite and limits the amount of swap space
1493 * available to this memcg, return that limit.
1494 */
1495 limit = min(limit, memsw);
1496 }
1497
1498 return limit;
1466} 1499}
1467 1500
1468void mem_cgroup_out_of_memory(struct mem_cgroup *memcg, gfp_t gfp_mask, 1501void mem_cgroup_out_of_memory(struct mem_cgroup *memcg, gfp_t gfp_mask,
@@ -3688,17 +3721,17 @@ unsigned long mem_cgroup_soft_limit_reclaim(struct zone *zone, int order,
3688static bool mem_cgroup_force_empty_list(struct mem_cgroup *memcg, 3721static bool mem_cgroup_force_empty_list(struct mem_cgroup *memcg,
3689 int node, int zid, enum lru_list lru) 3722 int node, int zid, enum lru_list lru)
3690{ 3723{
3691 struct mem_cgroup_per_zone *mz; 3724 struct lruvec *lruvec;
3692 unsigned long flags, loop; 3725 unsigned long flags, loop;
3693 struct list_head *list; 3726 struct list_head *list;
3694 struct page *busy; 3727 struct page *busy;
3695 struct zone *zone; 3728 struct zone *zone;
3696 3729
3697 zone = &NODE_DATA(node)->node_zones[zid]; 3730 zone = &NODE_DATA(node)->node_zones[zid];
3698 mz = mem_cgroup_zoneinfo(memcg, node, zid); 3731 lruvec = mem_cgroup_zone_lruvec(zone, memcg);
3699 list = &mz->lruvec.lists[lru]; 3732 list = &lruvec->lists[lru];
3700 3733
3701 loop = mz->lru_size[lru]; 3734 loop = mem_cgroup_get_lru_size(lruvec, lru);
3702 /* give some margin against EBUSY etc...*/ 3735 /* give some margin against EBUSY etc...*/
3703 loop += 256; 3736 loop += 256;
3704 busy = NULL; 3737 busy = NULL;
@@ -4736,7 +4769,7 @@ static int alloc_mem_cgroup_per_zone_info(struct mem_cgroup *memcg, int node)
4736 4769
4737 for (zone = 0; zone < MAX_NR_ZONES; zone++) { 4770 for (zone = 0; zone < MAX_NR_ZONES; zone++) {
4738 mz = &pn->zoneinfo[zone]; 4771 mz = &pn->zoneinfo[zone];
4739 lruvec_init(&mz->lruvec, &NODE_DATA(node)->node_zones[zone]); 4772 lruvec_init(&mz->lruvec);
4740 mz->usage_in_excess = 0; 4773 mz->usage_in_excess = 0;
4741 mz->on_tree = false; 4774 mz->on_tree = false;
4742 mz->memcg = memcg; 4775 mz->memcg = memcg;
diff --git a/mm/memory.c b/mm/memory.c
index fb135ba4aba9..221fc9ffcab1 100644
--- a/mm/memory.c
+++ b/mm/memory.c
@@ -2527,9 +2527,8 @@ static int do_wp_page(struct mm_struct *mm, struct vm_area_struct *vma,
2527 int ret = 0; 2527 int ret = 0;
2528 int page_mkwrite = 0; 2528 int page_mkwrite = 0;
2529 struct page *dirty_page = NULL; 2529 struct page *dirty_page = NULL;
2530 unsigned long mmun_start; /* For mmu_notifiers */ 2530 unsigned long mmun_start = 0; /* For mmu_notifiers */
2531 unsigned long mmun_end; /* For mmu_notifiers */ 2531 unsigned long mmun_end = 0; /* For mmu_notifiers */
2532 bool mmun_called = false; /* For mmu_notifiers */
2533 2532
2534 old_page = vm_normal_page(vma, address, orig_pte); 2533 old_page = vm_normal_page(vma, address, orig_pte);
2535 if (!old_page) { 2534 if (!old_page) {
@@ -2708,8 +2707,7 @@ gotten:
2708 goto oom_free_new; 2707 goto oom_free_new;
2709 2708
2710 mmun_start = address & PAGE_MASK; 2709 mmun_start = address & PAGE_MASK;
2711 mmun_end = (address & PAGE_MASK) + PAGE_SIZE; 2710 mmun_end = mmun_start + PAGE_SIZE;
2712 mmun_called = true;
2713 mmu_notifier_invalidate_range_start(mm, mmun_start, mmun_end); 2711 mmu_notifier_invalidate_range_start(mm, mmun_start, mmun_end);
2714 2712
2715 /* 2713 /*
@@ -2778,7 +2776,7 @@ gotten:
2778 page_cache_release(new_page); 2776 page_cache_release(new_page);
2779unlock: 2777unlock:
2780 pte_unmap_unlock(page_table, ptl); 2778 pte_unmap_unlock(page_table, ptl);
2781 if (mmun_called) 2779 if (mmun_end > mmun_start)
2782 mmu_notifier_invalidate_range_end(mm, mmun_start, mmun_end); 2780 mmu_notifier_invalidate_range_end(mm, mmun_start, mmun_end);
2783 if (old_page) { 2781 if (old_page) {
2784 /* 2782 /*
diff --git a/mm/memory_hotplug.c b/mm/memory_hotplug.c
index 56b758ae57d2..e4eeacae2b91 100644
--- a/mm/memory_hotplug.c
+++ b/mm/memory_hotplug.c
@@ -106,7 +106,6 @@ static void get_page_bootmem(unsigned long info, struct page *page,
106void __ref put_page_bootmem(struct page *page) 106void __ref put_page_bootmem(struct page *page)
107{ 107{
108 unsigned long type; 108 unsigned long type;
109 struct zone *zone;
110 109
111 type = (unsigned long) page->lru.next; 110 type = (unsigned long) page->lru.next;
112 BUG_ON(type < MEMORY_HOTPLUG_MIN_BOOTMEM_TYPE || 111 BUG_ON(type < MEMORY_HOTPLUG_MIN_BOOTMEM_TYPE ||
@@ -117,12 +116,6 @@ void __ref put_page_bootmem(struct page *page)
117 set_page_private(page, 0); 116 set_page_private(page, 0);
118 INIT_LIST_HEAD(&page->lru); 117 INIT_LIST_HEAD(&page->lru);
119 __free_pages_bootmem(page, 0); 118 __free_pages_bootmem(page, 0);
120
121 zone = page_zone(page);
122 zone_span_writelock(zone);
123 zone->present_pages++;
124 zone_span_writeunlock(zone);
125 totalram_pages++;
126 } 119 }
127 120
128} 121}
diff --git a/mm/mmap.c b/mm/mmap.c
index 2d942353d681..9a796c41e7d9 100644
--- a/mm/mmap.c
+++ b/mm/mmap.c
@@ -334,8 +334,10 @@ void validate_mm(struct mm_struct *mm)
334 struct vm_area_struct *vma = mm->mmap; 334 struct vm_area_struct *vma = mm->mmap;
335 while (vma) { 335 while (vma) {
336 struct anon_vma_chain *avc; 336 struct anon_vma_chain *avc;
337 vma_lock_anon_vma(vma);
337 list_for_each_entry(avc, &vma->anon_vma_chain, same_vma) 338 list_for_each_entry(avc, &vma->anon_vma_chain, same_vma)
338 anon_vma_interval_tree_verify(avc); 339 anon_vma_interval_tree_verify(avc);
340 vma_unlock_anon_vma(vma);
339 vma = vma->vm_next; 341 vma = vma->vm_next;
340 i++; 342 i++;
341 } 343 }
diff --git a/mm/mmzone.c b/mm/mmzone.c
index 3cef80f6ac79..4596d81b89b1 100644
--- a/mm/mmzone.c
+++ b/mm/mmzone.c
@@ -87,7 +87,7 @@ int memmap_valid_within(unsigned long pfn,
87} 87}
88#endif /* CONFIG_ARCH_HAS_HOLES_MEMORYMODEL */ 88#endif /* CONFIG_ARCH_HAS_HOLES_MEMORYMODEL */
89 89
90void lruvec_init(struct lruvec *lruvec, struct zone *zone) 90void lruvec_init(struct lruvec *lruvec)
91{ 91{
92 enum lru_list lru; 92 enum lru_list lru;
93 93
@@ -95,8 +95,4 @@ void lruvec_init(struct lruvec *lruvec, struct zone *zone)
95 95
96 for_each_lru(lru) 96 for_each_lru(lru)
97 INIT_LIST_HEAD(&lruvec->lists[lru]); 97 INIT_LIST_HEAD(&lruvec->lists[lru]);
98
99#ifdef CONFIG_MEMCG
100 lruvec->zone = zone;
101#endif
102} 98}
diff --git a/mm/nobootmem.c b/mm/nobootmem.c
index 714d5d650470..bd82f6b31411 100644
--- a/mm/nobootmem.c
+++ b/mm/nobootmem.c
@@ -116,8 +116,6 @@ static unsigned long __init __free_memory_core(phys_addr_t start,
116 return 0; 116 return 0;
117 117
118 __free_pages_memory(start_pfn, end_pfn); 118 __free_pages_memory(start_pfn, end_pfn);
119 fixup_zone_present_pages(pfn_to_nid(start >> PAGE_SHIFT),
120 start_pfn, end_pfn);
121 119
122 return end_pfn - start_pfn; 120 return end_pfn - start_pfn;
123} 121}
@@ -128,7 +126,6 @@ unsigned long __init free_low_memory_core_early(int nodeid)
128 phys_addr_t start, end, size; 126 phys_addr_t start, end, size;
129 u64 i; 127 u64 i;
130 128
131 reset_zone_present_pages();
132 for_each_free_mem_range(i, MAX_NUMNODES, &start, &end, NULL) 129 for_each_free_mem_range(i, MAX_NUMNODES, &start, &end, NULL)
133 count += __free_memory_core(start, end); 130 count += __free_memory_core(start, end);
134 131
diff --git a/mm/page_alloc.c b/mm/page_alloc.c
index 5b74de6702e0..7bb35ac0964a 100644
--- a/mm/page_alloc.c
+++ b/mm/page_alloc.c
@@ -4505,7 +4505,7 @@ static void __paginginit free_area_init_core(struct pglist_data *pgdat,
4505 zone->zone_pgdat = pgdat; 4505 zone->zone_pgdat = pgdat;
4506 4506
4507 zone_pcp_init(zone); 4507 zone_pcp_init(zone);
4508 lruvec_init(&zone->lruvec, zone); 4508 lruvec_init(&zone->lruvec);
4509 if (!size) 4509 if (!size)
4510 continue; 4510 continue;
4511 4511
@@ -6098,37 +6098,3 @@ void dump_page(struct page *page)
6098 dump_page_flags(page->flags); 6098 dump_page_flags(page->flags);
6099 mem_cgroup_print_bad_page(page); 6099 mem_cgroup_print_bad_page(page);
6100} 6100}
6101
6102/* reset zone->present_pages */
6103void reset_zone_present_pages(void)
6104{
6105 struct zone *z;
6106 int i, nid;
6107
6108 for_each_node_state(nid, N_HIGH_MEMORY) {
6109 for (i = 0; i < MAX_NR_ZONES; i++) {
6110 z = NODE_DATA(nid)->node_zones + i;
6111 z->present_pages = 0;
6112 }
6113 }
6114}
6115
6116/* calculate zone's present pages in buddy system */
6117void fixup_zone_present_pages(int nid, unsigned long start_pfn,
6118 unsigned long end_pfn)
6119{
6120 struct zone *z;
6121 unsigned long zone_start_pfn, zone_end_pfn;
6122 int i;
6123
6124 for (i = 0; i < MAX_NR_ZONES; i++) {
6125 z = NODE_DATA(nid)->node_zones + i;
6126 zone_start_pfn = z->zone_start_pfn;
6127 zone_end_pfn = zone_start_pfn + z->spanned_pages;
6128
6129 /* if the two regions intersect */
6130 if (!(zone_start_pfn >= end_pfn || zone_end_pfn <= start_pfn))
6131 z->present_pages += min(end_pfn, zone_end_pfn) -
6132 max(start_pfn, zone_start_pfn);
6133 }
6134}
diff --git a/mm/shmem.c b/mm/shmem.c
index 67afba5117f2..89341b658bd0 100644
--- a/mm/shmem.c
+++ b/mm/shmem.c
@@ -643,7 +643,7 @@ static void shmem_evict_inode(struct inode *inode)
643 kfree(info->symlink); 643 kfree(info->symlink);
644 644
645 simple_xattrs_free(&info->xattrs); 645 simple_xattrs_free(&info->xattrs);
646 BUG_ON(inode->i_blocks); 646 WARN_ON(inode->i_blocks);
647 shmem_free_inode(inode->i_sb); 647 shmem_free_inode(inode->i_sb);
648 clear_inode(inode); 648 clear_inode(inode);
649} 649}
@@ -1145,8 +1145,20 @@ repeat:
1145 if (!error) { 1145 if (!error) {
1146 error = shmem_add_to_page_cache(page, mapping, index, 1146 error = shmem_add_to_page_cache(page, mapping, index,
1147 gfp, swp_to_radix_entry(swap)); 1147 gfp, swp_to_radix_entry(swap));
1148 /* We already confirmed swap, and make no allocation */ 1148 /*
1149 VM_BUG_ON(error); 1149 * We already confirmed swap under page lock, and make
1150 * no memory allocation here, so usually no possibility
1151 * of error; but free_swap_and_cache() only trylocks a
1152 * page, so it is just possible that the entry has been
1153 * truncated or holepunched since swap was confirmed.
1154 * shmem_undo_range() will have done some of the
1155 * unaccounting, now delete_from_swap_cache() will do
1156 * the rest (including mem_cgroup_uncharge_swapcache).
1157 * Reset swap.val? No, leave it so "failed" goes back to
1158 * "repeat": reading a hole and writing should succeed.
1159 */
1160 if (error)
1161 delete_from_swap_cache(page);
1150 } 1162 }
1151 if (error) 1163 if (error)
1152 goto failed; 1164 goto failed;
diff --git a/mm/swapfile.c b/mm/swapfile.c
index 71cd288b2001..f91a25547ffe 100644
--- a/mm/swapfile.c
+++ b/mm/swapfile.c
@@ -1494,9 +1494,8 @@ SYSCALL_DEFINE1(swapoff, const char __user *, specialfile)
1494 BUG_ON(!current->mm); 1494 BUG_ON(!current->mm);
1495 1495
1496 pathname = getname(specialfile); 1496 pathname = getname(specialfile);
1497 err = PTR_ERR(pathname);
1498 if (IS_ERR(pathname)) 1497 if (IS_ERR(pathname))
1499 goto out; 1498 return PTR_ERR(pathname);
1500 1499
1501 victim = file_open_name(pathname, O_RDWR|O_LARGEFILE, 0); 1500 victim = file_open_name(pathname, O_RDWR|O_LARGEFILE, 0);
1502 err = PTR_ERR(victim); 1501 err = PTR_ERR(victim);
@@ -1608,6 +1607,7 @@ SYSCALL_DEFINE1(swapoff, const char __user *, specialfile)
1608out_dput: 1607out_dput:
1609 filp_close(victim, NULL); 1608 filp_close(victim, NULL);
1610out: 1609out:
1610 putname(pathname);
1611 return err; 1611 return err;
1612} 1612}
1613 1613
diff --git a/mm/vmscan.c b/mm/vmscan.c
index 8b055e9379bc..48550c66f1f2 100644
--- a/mm/vmscan.c
+++ b/mm/vmscan.c
@@ -1760,28 +1760,6 @@ static bool in_reclaim_compaction(struct scan_control *sc)
1760 return false; 1760 return false;
1761} 1761}
1762 1762
1763#ifdef CONFIG_COMPACTION
1764/*
1765 * If compaction is deferred for sc->order then scale the number of pages
1766 * reclaimed based on the number of consecutive allocation failures
1767 */
1768static unsigned long scale_for_compaction(unsigned long pages_for_compaction,
1769 struct lruvec *lruvec, struct scan_control *sc)
1770{
1771 struct zone *zone = lruvec_zone(lruvec);
1772
1773 if (zone->compact_order_failed <= sc->order)
1774 pages_for_compaction <<= zone->compact_defer_shift;
1775 return pages_for_compaction;
1776}
1777#else
1778static unsigned long scale_for_compaction(unsigned long pages_for_compaction,
1779 struct lruvec *lruvec, struct scan_control *sc)
1780{
1781 return pages_for_compaction;
1782}
1783#endif
1784
1785/* 1763/*
1786 * Reclaim/compaction is used for high-order allocation requests. It reclaims 1764 * Reclaim/compaction is used for high-order allocation requests. It reclaims
1787 * order-0 pages before compacting the zone. should_continue_reclaim() returns 1765 * order-0 pages before compacting the zone. should_continue_reclaim() returns
@@ -1829,9 +1807,6 @@ static inline bool should_continue_reclaim(struct lruvec *lruvec,
1829 * inactive lists are large enough, continue reclaiming 1807 * inactive lists are large enough, continue reclaiming
1830 */ 1808 */
1831 pages_for_compaction = (2UL << sc->order); 1809 pages_for_compaction = (2UL << sc->order);
1832
1833 pages_for_compaction = scale_for_compaction(pages_for_compaction,
1834 lruvec, sc);
1835 inactive_lru_pages = get_lru_size(lruvec, LRU_INACTIVE_FILE); 1810 inactive_lru_pages = get_lru_size(lruvec, LRU_INACTIVE_FILE);
1836 if (nr_swap_pages > 0) 1811 if (nr_swap_pages > 0)
1837 inactive_lru_pages += get_lru_size(lruvec, LRU_INACTIVE_ANON); 1812 inactive_lru_pages += get_lru_size(lruvec, LRU_INACTIVE_ANON);