Fix off-by-one error in iov_iter_advance()
[linux-2.6.git] / mm / memory_hotplug.c
index 71da5c9..89fee2d 100644 (file)
@@ -4,7 +4,6 @@
  *  Copyright (C)
  */
 
-#include <linux/config.h>
 #include <linux/stddef.h>
 #include <linux/mm.h>
 #include <linux/swap.h>
@@ -14,6 +13,7 @@
 #include <linux/compiler.h>
 #include <linux/module.h>
 #include <linux/pagevec.h>
+#include <linux/writeback.h>
 #include <linux/slab.h>
 #include <linux/sysctl.h>
 #include <linux/cpu.h>
 #include <linux/memory_hotplug.h>
 #include <linux/highmem.h>
 #include <linux/vmalloc.h>
+#include <linux/ioport.h>
+#include <linux/cpuset.h>
+#include <linux/delay.h>
+#include <linux/migrate.h>
+#include <linux/page-isolation.h>
 
 #include <asm/tlbflush.h>
 
-extern void zonetable_add(struct zone *zone, int nid, int zid, unsigned long pfn,
-                         unsigned long size);
+#include "internal.h"
+
+/* add this memory to iomem resource */
+static struct resource *register_memory_resource(u64 start, u64 size)
+{
+       struct resource *res;
+       res = kzalloc(sizeof(struct resource), GFP_KERNEL);
+       BUG_ON(!res);
+
+       res->name = "System RAM";
+       res->start = start;
+       res->end = start + size - 1;
+       res->flags = IORESOURCE_MEM | IORESOURCE_BUSY;
+       if (request_resource(&iomem_resource, res) < 0) {
+               printk("System RAM resource %llx - %llx cannot be added\n",
+               (unsigned long long)res->start, (unsigned long long)res->end);
+               kfree(res);
+               res = NULL;
+       }
+       return res;
+}
+
+static void release_memory_resource(struct resource *res)
+{
+       if (!res)
+               return;
+       release_resource(res);
+       kfree(res);
+       return;
+}
+
+#ifdef CONFIG_MEMORY_HOTPLUG_SPARSE
+#ifndef CONFIG_SPARSEMEM_VMEMMAP
+static void get_page_bootmem(unsigned long info,  struct page *page, int type)
+{
+       atomic_set(&page->_mapcount, type);
+       SetPagePrivate(page);
+       set_page_private(page, info);
+       atomic_inc(&page->_count);
+}
+
+void put_page_bootmem(struct page *page)
+{
+       int type;
+
+       type = atomic_read(&page->_mapcount);
+       BUG_ON(type >= -1);
+
+       if (atomic_dec_return(&page->_count) == 1) {
+               ClearPagePrivate(page);
+               set_page_private(page, 0);
+               reset_page_mapcount(page);
+               __free_pages_bootmem(page, 0);
+       }
+
+}
+
+static void register_page_bootmem_info_section(unsigned long start_pfn)
+{
+       unsigned long *usemap, mapsize, section_nr, i;
+       struct mem_section *ms;
+       struct page *page, *memmap;
+
+       if (!pfn_valid(start_pfn))
+               return;
+
+       section_nr = pfn_to_section_nr(start_pfn);
+       ms = __nr_to_section(section_nr);
+
+       /* Get section's memmap address */
+       memmap = sparse_decode_mem_map(ms->section_mem_map, section_nr);
+
+       /*
+        * Get page for the memmap's phys address
+        * XXX: need more consideration for sparse_vmemmap...
+        */
+       page = virt_to_page(memmap);
+       mapsize = sizeof(struct page) * PAGES_PER_SECTION;
+       mapsize = PAGE_ALIGN(mapsize) >> PAGE_SHIFT;
+
+       /* remember memmap's page */
+       for (i = 0; i < mapsize; i++, page++)
+               get_page_bootmem(section_nr, page, SECTION_INFO);
+
+       usemap = __nr_to_section(section_nr)->pageblock_flags;
+       page = virt_to_page(usemap);
+
+       mapsize = PAGE_ALIGN(usemap_size()) >> PAGE_SHIFT;
+
+       for (i = 0; i < mapsize; i++, page++)
+               get_page_bootmem(section_nr, page, MIX_SECTION_INFO);
+
+}
+
+void register_page_bootmem_info_node(struct pglist_data *pgdat)
+{
+       unsigned long i, pfn, end_pfn, nr_pages;
+       int node = pgdat->node_id;
+       struct page *page;
+       struct zone *zone;
+
+       nr_pages = PAGE_ALIGN(sizeof(struct pglist_data)) >> PAGE_SHIFT;
+       page = virt_to_page(pgdat);
+
+       for (i = 0; i < nr_pages; i++, page++)
+               get_page_bootmem(node, page, NODE_INFO);
+
+       zone = &pgdat->node_zones[0];
+       for (; zone < pgdat->node_zones + MAX_NR_ZONES - 1; zone++) {
+               if (zone->wait_table) {
+                       nr_pages = zone->wait_table_hash_nr_entries
+                               * sizeof(wait_queue_head_t);
+                       nr_pages = PAGE_ALIGN(nr_pages) >> PAGE_SHIFT;
+                       page = virt_to_page(zone->wait_table);
+
+                       for (i = 0; i < nr_pages; i++, page++)
+                               get_page_bootmem(node, page, NODE_INFO);
+               }
+       }
+
+       pfn = pgdat->node_start_pfn;
+       end_pfn = pfn + pgdat->node_spanned_pages;
+
+       /* register_section info */
+       for (; pfn < end_pfn; pfn += PAGES_PER_SECTION)
+               register_page_bootmem_info_section(pfn);
+
+}
+#endif /* !CONFIG_SPARSEMEM_VMEMMAP */
+
+static void grow_zone_span(struct zone *zone, unsigned long start_pfn,
+                          unsigned long end_pfn)
+{
+       unsigned long old_zone_end_pfn;
+
+       zone_span_writelock(zone);
+
+       old_zone_end_pfn = zone->zone_start_pfn + zone->spanned_pages;
+       if (start_pfn < zone->zone_start_pfn)
+               zone->zone_start_pfn = start_pfn;
+
+       zone->spanned_pages = max(old_zone_end_pfn, end_pfn) -
+                               zone->zone_start_pfn;
+
+       zone_span_writeunlock(zone);
+}
+
+static void grow_pgdat_span(struct pglist_data *pgdat, unsigned long start_pfn,
+                           unsigned long end_pfn)
+{
+       unsigned long old_pgdat_end_pfn =
+               pgdat->node_start_pfn + pgdat->node_spanned_pages;
+
+       if (start_pfn < pgdat->node_start_pfn)
+               pgdat->node_start_pfn = start_pfn;
+
+       pgdat->node_spanned_pages = max(old_pgdat_end_pfn, end_pfn) -
+                                       pgdat->node_start_pfn;
+}
+
 static int __add_zone(struct zone *zone, unsigned long phys_start_pfn)
 {
        struct pglist_data *pgdat = zone->zone_pgdat;
        int nr_pages = PAGES_PER_SECTION;
        int nid = pgdat->node_id;
        int zone_type;
+       unsigned long flags;
 
        zone_type = zone - pgdat->node_zones;
-       if (!populated_zone(zone)) {
-               int ret = 0;
-               ret = init_currently_empty_zone(zone, phys_start_pfn, nr_pages);
-               if (ret < 0)
+       if (!zone->wait_table) {
+               int ret;
+
+               ret = init_currently_empty_zone(zone, phys_start_pfn,
+                                               nr_pages, MEMMAP_HOTPLUG);
+               if (ret)
                        return ret;
        }
-       memmap_init_zone(nr_pages, nid, zone_type, phys_start_pfn);
-       zonetable_add(zone, nid, zone_type, phys_start_pfn, nr_pages);
+       pgdat_resize_lock(zone->zone_pgdat, &flags);
+       grow_zone_span(zone, phys_start_pfn, phys_start_pfn + nr_pages);
+       grow_pgdat_span(zone->zone_pgdat, phys_start_pfn,
+                       phys_start_pfn + nr_pages);
+       pgdat_resize_unlock(zone->zone_pgdat, &flags);
+       memmap_init_zone(nr_pages, nid, zone_type,
+                        phys_start_pfn, MEMMAP_HOTPLUG);
        return 0;
 }
 
-extern int sparse_add_one_section(struct zone *zone, unsigned long start_pfn,
-                                 int nr_pages);
 static int __add_section(struct zone *zone, unsigned long phys_start_pfn)
 {
        int nr_pages = PAGES_PER_SECTION;
        int ret;
 
+       if (pfn_valid(phys_start_pfn))
+               return -EEXIST;
+
        ret = sparse_add_one_section(zone, phys_start_pfn, nr_pages);
 
        if (ret < 0)
@@ -65,6 +237,36 @@ static int __add_section(struct zone *zone, unsigned long phys_start_pfn)
        return register_new_memory(__pfn_to_section(phys_start_pfn));
 }
 
+#ifdef CONFIG_SPARSEMEM_VMEMMAP
+static int __remove_section(struct zone *zone, struct mem_section *ms)
+{
+       /*
+        * XXX: Freeing memmap with vmemmap is not implement yet.
+        *      This should be removed later.
+        */
+       return -EBUSY;
+}
+#else
+static int __remove_section(struct zone *zone, struct mem_section *ms)
+{
+       unsigned long flags;
+       struct pglist_data *pgdat = zone->zone_pgdat;
+       int ret = -EINVAL;
+
+       if (!valid_section(ms))
+               return ret;
+
+       ret = unregister_memory_section(ms);
+       if (ret)
+               return ret;
+
+       pgdat_resize_lock(pgdat, &flags);
+       sparse_remove_one_section(zone, ms);
+       pgdat_resize_unlock(pgdat, &flags);
+       return 0;
+}
+#endif
+
 /*
  * Reasonably generic function for adding memory.  It is
  * expected that archs that support memory hotplug will
@@ -76,78 +278,581 @@ int __add_pages(struct zone *zone, unsigned long phys_start_pfn,
 {
        unsigned long i;
        int err = 0;
+       int start_sec, end_sec;
+       /* during initialize mem_map, align hot-added range to section */
+       start_sec = pfn_to_section_nr(phys_start_pfn);
+       end_sec = pfn_to_section_nr(phys_start_pfn + nr_pages - 1);
 
-       for (i = 0; i < nr_pages; i += PAGES_PER_SECTION) {
-               err = __add_section(zone, phys_start_pfn + i);
+       for (i = start_sec; i <= end_sec; i++) {
+               err = __add_section(zone, i << PFN_SECTION_SHIFT);
 
-               /* We want to keep adding the rest of the
-                * sections if the first ones already exist
+               /*
+                * EEXIST is finally dealt with by ioresource collision
+                * check. see add_memory() => register_memory_resource()
+                * Warning will be printed if there is collision.
                 */
                if (err && (err != -EEXIST))
                        break;
+               err = 0;
        }
 
        return err;
 }
 EXPORT_SYMBOL_GPL(__add_pages);
 
-static void grow_zone_span(struct zone *zone,
-               unsigned long start_pfn, unsigned long end_pfn)
+/**
+ * __remove_pages() - remove sections of pages from a zone
+ * @zone: zone from which pages need to be removed
+ * @phys_start_pfn: starting pageframe (must be aligned to start of a section)
+ * @nr_pages: number of pages to remove (must be multiple of section size)
+ *
+ * Generic helper function to remove section mappings and sysfs entries
+ * for the section of the memory we are removing. Caller needs to make
+ * sure that pages are marked reserved and zones are adjust properly by
+ * calling offline_pages().
+ */
+int __remove_pages(struct zone *zone, unsigned long phys_start_pfn,
+                unsigned long nr_pages)
 {
-       unsigned long old_zone_end_pfn;
-
-       zone_span_writelock(zone);
+       unsigned long i, ret = 0;
+       int sections_to_remove;
 
-       old_zone_end_pfn = zone->zone_start_pfn + zone->spanned_pages;
-       if (start_pfn < zone->zone_start_pfn)
-               zone->zone_start_pfn = start_pfn;
+       /*
+        * We can only remove entire sections
+        */
+       BUG_ON(phys_start_pfn & ~PAGE_SECTION_MASK);
+       BUG_ON(nr_pages % PAGES_PER_SECTION);
 
-       zone->spanned_pages = max(old_zone_end_pfn, end_pfn) -
-                               zone->zone_start_pfn;
+       release_mem_region(phys_start_pfn << PAGE_SHIFT, nr_pages * PAGE_SIZE);
 
-       zone_span_writeunlock(zone);
+       sections_to_remove = nr_pages / PAGES_PER_SECTION;
+       for (i = 0; i < sections_to_remove; i++) {
+               unsigned long pfn = phys_start_pfn + i*PAGES_PER_SECTION;
+               ret = __remove_section(zone, __pfn_to_section(pfn));
+               if (ret)
+                       break;
+       }
+       return ret;
 }
+EXPORT_SYMBOL_GPL(__remove_pages);
 
-static void grow_pgdat_span(struct pglist_data *pgdat,
-               unsigned long start_pfn, unsigned long end_pfn)
+void online_page(struct page *page)
 {
-       unsigned long old_pgdat_end_pfn =
-               pgdat->node_start_pfn + pgdat->node_spanned_pages;
+       totalram_pages++;
+       num_physpages++;
 
-       if (start_pfn < pgdat->node_start_pfn)
-               pgdat->node_start_pfn = start_pfn;
+#ifdef CONFIG_HIGHMEM
+       if (PageHighMem(page))
+               totalhigh_pages++;
+#endif
 
-       pgdat->node_spanned_pages = max(old_pgdat_end_pfn, end_pfn) -
-                                       pgdat->node_start_pfn;
+#ifdef CONFIG_FLATMEM
+       max_mapnr = max(page_to_pfn(page), max_mapnr);
+#endif
+
+       ClearPageReserved(page);
+       init_page_count(page);
+       __free_page(page);
 }
 
-int online_pages(unsigned long pfn, unsigned long nr_pages)
+static int online_pages_range(unsigned long start_pfn, unsigned long nr_pages,
+                       void *arg)
 {
        unsigned long i;
-       unsigned long flags;
+       unsigned long onlined_pages = *(unsigned long *)arg;
+       struct page *page;
+       if (PageReserved(pfn_to_page(start_pfn)))
+               for (i = 0; i < nr_pages; i++) {
+                       page = pfn_to_page(start_pfn + i);
+                       online_page(page);
+                       onlined_pages++;
+               }
+       *(unsigned long *)arg = onlined_pages;
+       return 0;
+}
+
+
+int online_pages(unsigned long pfn, unsigned long nr_pages)
+{
        unsigned long onlined_pages = 0;
        struct zone *zone;
+       int need_zonelists_rebuild = 0;
+       int nid;
+       int ret;
+       struct memory_notify arg;
+
+       arg.start_pfn = pfn;
+       arg.nr_pages = nr_pages;
+       arg.status_change_nid = -1;
 
+       nid = page_to_nid(pfn_to_page(pfn));
+       if (node_present_pages(nid) == 0)
+               arg.status_change_nid = nid;
+
+       ret = memory_notify(MEM_GOING_ONLINE, &arg);
+       ret = notifier_to_errno(ret);
+       if (ret) {
+               memory_notify(MEM_CANCEL_ONLINE, &arg);
+               return ret;
+       }
        /*
         * This doesn't need a lock to do pfn_to_page().
         * The section can't be removed here because of the
-        * memory_block->state_sem.
+        * memory_block->state_mutex.
         */
        zone = page_zone(pfn_to_page(pfn));
-       pgdat_resize_lock(zone->zone_pgdat, &flags);
-       grow_zone_span(zone, pfn, pfn + nr_pages);
-       grow_pgdat_span(zone->zone_pgdat, pfn, pfn + nr_pages);
-       pgdat_resize_unlock(zone->zone_pgdat, &flags);
+       /*
+        * If this zone is not populated, then it is not in zonelist.
+        * This means the page allocator ignores this zone.
+        * So, zonelist must be updated after online.
+        */
+       if (!populated_zone(zone))
+               need_zonelists_rebuild = 1;
 
-       for (i = 0; i < nr_pages; i++) {
-               struct page *page = pfn_to_page(pfn + i);
-               online_page(page);
-               onlined_pages++;
+       ret = walk_memory_resource(pfn, nr_pages, &onlined_pages,
+               online_pages_range);
+       if (ret) {
+               printk(KERN_DEBUG "online_pages %lx at %lx failed\n",
+                       nr_pages, pfn);
+               memory_notify(MEM_CANCEL_ONLINE, &arg);
+               return ret;
        }
+
        zone->present_pages += onlined_pages;
        zone->zone_pgdat->node_present_pages += onlined_pages;
 
        setup_per_zone_pages_min();
+       if (onlined_pages) {
+               kswapd_run(zone_to_nid(zone));
+               node_set_state(zone_to_nid(zone), N_HIGH_MEMORY);
+       }
+
+       if (need_zonelists_rebuild)
+               build_all_zonelists();
+       else
+               vm_total_pages = nr_free_pagecache_pages();
+
+       writeback_set_ratelimit();
+
+       if (onlined_pages)
+               memory_notify(MEM_ONLINE, &arg);
+
+       return 0;
+}
+#endif /* CONFIG_MEMORY_HOTPLUG_SPARSE */
+
+static pg_data_t *hotadd_new_pgdat(int nid, u64 start)
+{
+       struct pglist_data *pgdat;
+       unsigned long zones_size[MAX_NR_ZONES] = {0};
+       unsigned long zholes_size[MAX_NR_ZONES] = {0};
+       unsigned long start_pfn = start >> PAGE_SHIFT;
+
+       pgdat = arch_alloc_nodedata(nid);
+       if (!pgdat)
+               return NULL;
+
+       arch_refresh_nodedata(nid, pgdat);
+
+       /* we can use NODE_DATA(nid) from here */
+
+       /* init node's zones as empty zones, we don't have any present pages.*/
+       free_area_init_node(nid, zones_size, start_pfn, zholes_size);
+
+       return pgdat;
+}
+
+static void rollback_node_hotadd(int nid, pg_data_t *pgdat)
+{
+       arch_refresh_nodedata(nid, NULL);
+       arch_free_nodedata(pgdat);
+       return;
+}
+
+
+int add_memory(int nid, u64 start, u64 size)
+{
+       pg_data_t *pgdat = NULL;
+       int new_pgdat = 0;
+       struct resource *res;
+       int ret;
+
+       res = register_memory_resource(start, size);
+       if (!res)
+               return -EEXIST;
+
+       if (!node_online(nid)) {
+               pgdat = hotadd_new_pgdat(nid, start);
+               if (!pgdat)
+                       return -ENOMEM;
+               new_pgdat = 1;
+       }
+
+       /* call arch's memory hotadd */
+       ret = arch_add_memory(nid, start, size);
+
+       if (ret < 0)
+               goto error;
+
+       /* we online node here. we can't roll back from here. */
+       node_set_online(nid);
+
+       cpuset_track_online_nodes();
+
+       if (new_pgdat) {
+               ret = register_one_node(nid);
+               /*
+                * If sysfs file of new node can't create, cpu on the node
+                * can't be hot-added. There is no rollback way now.
+                * So, check by BUG_ON() to catch it reluctantly..
+                */
+               BUG_ON(ret);
+       }
+
+       return ret;
+error:
+       /* rollback pgdat allocation and others */
+       if (new_pgdat)
+               rollback_node_hotadd(nid, pgdat);
+       if (res)
+               release_memory_resource(res);
+
+       return ret;
+}
+EXPORT_SYMBOL_GPL(add_memory);
+
+#ifdef CONFIG_MEMORY_HOTREMOVE
+/*
+ * A free page on the buddy free lists (not the per-cpu lists) has PageBuddy
+ * set and the size of the free page is given by page_order(). Using this,
+ * the function determines if the pageblock contains only free pages.
+ * Due to buddy contraints, a free page at least the size of a pageblock will
+ * be located at the start of the pageblock
+ */
+static inline int pageblock_free(struct page *page)
+{
+       return PageBuddy(page) && page_order(page) >= pageblock_order;
+}
+
+/* Return the start of the next active pageblock after a given page */
+static struct page *next_active_pageblock(struct page *page)
+{
+       int pageblocks_stride;
+
+       /* Ensure the starting page is pageblock-aligned */
+       BUG_ON(page_to_pfn(page) & (pageblock_nr_pages - 1));
+
+       /* Move forward by at least 1 * pageblock_nr_pages */
+       pageblocks_stride = 1;
+
+       /* If the entire pageblock is free, move to the end of free page */
+       if (pageblock_free(page))
+               pageblocks_stride += page_order(page) - pageblock_order;
+
+       return page + (pageblocks_stride * pageblock_nr_pages);
+}
+
+/* Checks if this range of memory is likely to be hot-removable. */
+int is_mem_section_removable(unsigned long start_pfn, unsigned long nr_pages)
+{
+       int type;
+       struct page *page = pfn_to_page(start_pfn);
+       struct page *end_page = page + nr_pages;
+
+       /* Check the starting page of each pageblock within the range */
+       for (; page < end_page; page = next_active_pageblock(page)) {
+               type = get_pageblock_migratetype(page);
+
+               /*
+                * A pageblock containing MOVABLE or free pages is considered
+                * removable
+                */
+               if (type != MIGRATE_MOVABLE && !pageblock_free(page))
+                       return 0;
+
+               /*
+                * A pageblock starting with a PageReserved page is not
+                * considered removable.
+                */
+               if (PageReserved(page))
+                       return 0;
+       }
+
+       /* All pageblocks in the memory block are likely to be hot-removable */
+       return 1;
+}
+
+/*
+ * Confirm all pages in a range [start, end) is belongs to the same zone.
+ */
+static int test_pages_in_a_zone(unsigned long start_pfn, unsigned long end_pfn)
+{
+       unsigned long pfn;
+       struct zone *zone = NULL;
+       struct page *page;
+       int i;
+       for (pfn = start_pfn;
+            pfn < end_pfn;
+            pfn += MAX_ORDER_NR_PAGES) {
+               i = 0;
+               /* This is just a CONFIG_HOLES_IN_ZONE check.*/
+               while ((i < MAX_ORDER_NR_PAGES) && !pfn_valid_within(pfn + i))
+                       i++;
+               if (i == MAX_ORDER_NR_PAGES)
+                       continue;
+               page = pfn_to_page(pfn + i);
+               if (zone && page_zone(page) != zone)
+                       return 0;
+               zone = page_zone(page);
+       }
+       return 1;
+}
+
+/*
+ * Scanning pfn is much easier than scanning lru list.
+ * Scan pfn from start to end and Find LRU page.
+ */
+int scan_lru_pages(unsigned long start, unsigned long end)
+{
+       unsigned long pfn;
+       struct page *page;
+       for (pfn = start; pfn < end; pfn++) {
+               if (pfn_valid(pfn)) {
+                       page = pfn_to_page(pfn);
+                       if (PageLRU(page))
+                               return pfn;
+               }
+       }
+       return 0;
+}
+
+static struct page *
+hotremove_migrate_alloc(struct page *page,
+                       unsigned long private,
+                       int **x)
+{
+       /* This should be improoooooved!! */
+       return alloc_page(GFP_HIGHUSER_PAGECACHE);
+}
+
+
+#define NR_OFFLINE_AT_ONCE_PAGES       (256)
+static int
+do_migrate_range(unsigned long start_pfn, unsigned long end_pfn)
+{
+       unsigned long pfn;
+       struct page *page;
+       int move_pages = NR_OFFLINE_AT_ONCE_PAGES;
+       int not_managed = 0;
+       int ret = 0;
+       LIST_HEAD(source);
+
+       for (pfn = start_pfn; pfn < end_pfn && move_pages > 0; pfn++) {
+               if (!pfn_valid(pfn))
+                       continue;
+               page = pfn_to_page(pfn);
+               if (!page_count(page))
+                       continue;
+               /*
+                * We can skip free pages. And we can only deal with pages on
+                * LRU.
+                */
+               ret = isolate_lru_page(page, &source);
+               if (!ret) { /* Success */
+                       move_pages--;
+               } else {
+                       /* Becasue we don't have big zone->lock. we should
+                          check this again here. */
+                       if (page_count(page))
+                               not_managed++;
+#ifdef CONFIG_DEBUG_VM
+                       printk(KERN_INFO "removing from LRU failed"
+                                        " %lx/%d/%lx\n",
+                               pfn, page_count(page), page->flags);
+#endif
+               }
+       }
+       ret = -EBUSY;
+       if (not_managed) {
+               if (!list_empty(&source))
+                       putback_lru_pages(&source);
+               goto out;
+       }
+       ret = 0;
+       if (list_empty(&source))
+               goto out;
+       /* this function returns # of failed pages */
+       ret = migrate_pages(&source, hotremove_migrate_alloc, 0);
+
+out:
+       return ret;
+}
+
+/*
+ * remove from free_area[] and mark all as Reserved.
+ */
+static int
+offline_isolated_pages_cb(unsigned long start, unsigned long nr_pages,
+                       void *data)
+{
+       __offline_isolated_pages(start, start + nr_pages);
+       return 0;
+}
+
+static void
+offline_isolated_pages(unsigned long start_pfn, unsigned long end_pfn)
+{
+       walk_memory_resource(start_pfn, end_pfn - start_pfn, NULL,
+                               offline_isolated_pages_cb);
+}
+
+/*
+ * Check all pages in range, recoreded as memory resource, are isolated.
+ */
+static int
+check_pages_isolated_cb(unsigned long start_pfn, unsigned long nr_pages,
+                       void *data)
+{
+       int ret;
+       long offlined = *(long *)data;
+       ret = test_pages_isolated(start_pfn, start_pfn + nr_pages);
+       offlined = nr_pages;
+       if (!ret)
+               *(long *)data += offlined;
+       return ret;
+}
+
+static long
+check_pages_isolated(unsigned long start_pfn, unsigned long end_pfn)
+{
+       long offlined = 0;
+       int ret;
+
+       ret = walk_memory_resource(start_pfn, end_pfn - start_pfn, &offlined,
+                       check_pages_isolated_cb);
+       if (ret < 0)
+               offlined = (long)ret;
+       return offlined;
+}
+
+int offline_pages(unsigned long start_pfn,
+                 unsigned long end_pfn, unsigned long timeout)
+{
+       unsigned long pfn, nr_pages, expire;
+       long offlined_pages;
+       int ret, drain, retry_max, node;
+       struct zone *zone;
+       struct memory_notify arg;
 
+       BUG_ON(start_pfn >= end_pfn);
+       /* at least, alignment against pageblock is necessary */
+       if (!IS_ALIGNED(start_pfn, pageblock_nr_pages))
+               return -EINVAL;
+       if (!IS_ALIGNED(end_pfn, pageblock_nr_pages))
+               return -EINVAL;
+       /* This makes hotplug much easier...and readable.
+          we assume this for now. .*/
+       if (!test_pages_in_a_zone(start_pfn, end_pfn))
+               return -EINVAL;
+
+       zone = page_zone(pfn_to_page(start_pfn));
+       node = zone_to_nid(zone);
+       nr_pages = end_pfn - start_pfn;
+
+       /* set above range as isolated */
+       ret = start_isolate_page_range(start_pfn, end_pfn);
+       if (ret)
+               return ret;
+
+       arg.start_pfn = start_pfn;
+       arg.nr_pages = nr_pages;
+       arg.status_change_nid = -1;
+       if (nr_pages >= node_present_pages(node))
+               arg.status_change_nid = node;
+
+       ret = memory_notify(MEM_GOING_OFFLINE, &arg);
+       ret = notifier_to_errno(ret);
+       if (ret)
+               goto failed_removal;
+
+       pfn = start_pfn;
+       expire = jiffies + timeout;
+       drain = 0;
+       retry_max = 5;
+repeat:
+       /* start memory hot removal */
+       ret = -EAGAIN;
+       if (time_after(jiffies, expire))
+               goto failed_removal;
+       ret = -EINTR;
+       if (signal_pending(current))
+               goto failed_removal;
+       ret = 0;
+       if (drain) {
+               lru_add_drain_all();
+               flush_scheduled_work();
+               cond_resched();
+               drain_all_pages();
+       }
+
+       pfn = scan_lru_pages(start_pfn, end_pfn);
+       if (pfn) { /* We have page on LRU */
+               ret = do_migrate_range(pfn, end_pfn);
+               if (!ret) {
+                       drain = 1;
+                       goto repeat;
+               } else {
+                       if (ret < 0)
+                               if (--retry_max == 0)
+                                       goto failed_removal;
+                       yield();
+                       drain = 1;
+                       goto repeat;
+               }
+       }
+       /* drain all zone's lru pagevec, this is asyncronous... */
+       lru_add_drain_all();
+       flush_scheduled_work();
+       yield();
+       /* drain pcp pages , this is synchrouns. */
+       drain_all_pages();
+       /* check again */
+       offlined_pages = check_pages_isolated(start_pfn, end_pfn);
+       if (offlined_pages < 0) {
+               ret = -EBUSY;
+               goto failed_removal;
+       }
+       printk(KERN_INFO "Offlined Pages %ld\n", offlined_pages);
+       /* Ok, all of our target is islaoted.
+          We cannot do rollback at this point. */
+       offline_isolated_pages(start_pfn, end_pfn);
+       /* reset pagetype flags and makes migrate type to be MOVABLE */
+       undo_isolate_page_range(start_pfn, end_pfn);
+       /* removal success */
+       zone->present_pages -= offlined_pages;
+       zone->zone_pgdat->node_present_pages -= offlined_pages;
+       totalram_pages -= offlined_pages;
+       num_physpages -= offlined_pages;
+
+       vm_total_pages = nr_free_pagecache_pages();
+       writeback_set_ratelimit();
+
+       memory_notify(MEM_OFFLINE, &arg);
        return 0;
+
+failed_removal:
+       printk(KERN_INFO "memory offlining %lx to %lx failed\n",
+               start_pfn, end_pfn);
+       memory_notify(MEM_CANCEL_OFFLINE, &arg);
+       /* pushback to free area */
+       undo_isolate_page_range(start_pfn, end_pfn);
+
+       return ret;
+}
+#else
+int remove_memory(u64 start, u64 size)
+{
+       return -EINVAL;
 }
+EXPORT_SYMBOL_GPL(remove_memory);
+#endif /* CONFIG_MEMORY_HOTREMOVE */