]> git.proxmox.com Git - mirror_ubuntu-disco-kernel.git/commitdiff
memory_hotplug: clear zone when removing the memory
authorYasuaki Ishimatsu <isimatu.yasuaki@jp.fujitsu.com>
Sat, 23 Feb 2013 00:33:12 +0000 (16:33 -0800)
committerLinus Torvalds <torvalds@linux-foundation.org>
Sun, 24 Feb 2013 01:50:12 +0000 (17:50 -0800)
When memory is added, we update zone's and pgdat's start_pfn and
spanned_pages in __add_zone().  So we should revert them when the memory
is removed.

The patch adds a new function __remove_zone() to do this.

Signed-off-by: Yasuaki Ishimatsu <isimatu.yasuaki@jp.fujitsu.com>
Signed-off-by: Wen Congyang <wency@cn.fujitsu.com>
Signed-off-by: Tang Chen <tangchen@cn.fujitsu.com>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Jiang Liu <jiang.liu@huawei.com>
Cc: Jianguo Wu <wujianguo@huawei.com>
Cc: Kamezawa Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Lai Jiangshan <laijs@cn.fujitsu.com>
Cc: Wu Jianguo <wujianguo@huawei.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
mm/memory_hotplug.c

index 46c58be2fdc4cdc84ea1f17c788b5a438d74b65d..3f792375f3267f53fafdecf5b2c0426f9d181a3d 100644 (file)
@@ -432,8 +432,211 @@ static int __meminit __add_section(int nid, struct zone *zone,
        return register_new_memory(nid, __pfn_to_section(phys_start_pfn));
 }
 
+/* find the smallest valid pfn in the range [start_pfn, end_pfn) */
+static int find_smallest_section_pfn(int nid, struct zone *zone,
+                                    unsigned long start_pfn,
+                                    unsigned long end_pfn)
+{
+       struct mem_section *ms;
+
+       for (; start_pfn < end_pfn; start_pfn += PAGES_PER_SECTION) {
+               ms = __pfn_to_section(start_pfn);
+
+               if (unlikely(!valid_section(ms)))
+                       continue;
+
+               if (unlikely(pfn_to_nid(start_pfn) != nid))
+                       continue;
+
+               if (zone && zone != page_zone(pfn_to_page(start_pfn)))
+                       continue;
+
+               return start_pfn;
+       }
+
+       return 0;
+}
+
+/* find the biggest valid pfn in the range [start_pfn, end_pfn). */
+static int find_biggest_section_pfn(int nid, struct zone *zone,
+                                   unsigned long start_pfn,
+                                   unsigned long end_pfn)
+{
+       struct mem_section *ms;
+       unsigned long pfn;
+
+       /* pfn is the end pfn of a memory section. */
+       pfn = end_pfn - 1;
+       for (; pfn >= start_pfn; pfn -= PAGES_PER_SECTION) {
+               ms = __pfn_to_section(pfn);
+
+               if (unlikely(!valid_section(ms)))
+                       continue;
+
+               if (unlikely(pfn_to_nid(pfn) != nid))
+                       continue;
+
+               if (zone && zone != page_zone(pfn_to_page(pfn)))
+                       continue;
+
+               return pfn;
+       }
+
+       return 0;
+}
+
+static void shrink_zone_span(struct zone *zone, unsigned long start_pfn,
+                            unsigned long end_pfn)
+{
+       unsigned long zone_start_pfn =  zone->zone_start_pfn;
+       unsigned long zone_end_pfn = zone->zone_start_pfn + zone->spanned_pages;
+       unsigned long pfn;
+       struct mem_section *ms;
+       int nid = zone_to_nid(zone);
+
+       zone_span_writelock(zone);
+       if (zone_start_pfn == start_pfn) {
+               /*
+                * If the section is smallest section in the zone, it need
+                * shrink zone->zone_start_pfn and zone->zone_spanned_pages.
+                * In this case, we find second smallest valid mem_section
+                * for shrinking zone.
+                */
+               pfn = find_smallest_section_pfn(nid, zone, end_pfn,
+                                               zone_end_pfn);
+               if (pfn) {
+                       zone->zone_start_pfn = pfn;
+                       zone->spanned_pages = zone_end_pfn - pfn;
+               }
+       } else if (zone_end_pfn == end_pfn) {
+               /*
+                * If the section is biggest section in the zone, it need
+                * shrink zone->spanned_pages.
+                * In this case, we find second biggest valid mem_section for
+                * shrinking zone.
+                */
+               pfn = find_biggest_section_pfn(nid, zone, zone_start_pfn,
+                                              start_pfn);
+               if (pfn)
+                       zone->spanned_pages = pfn - zone_start_pfn + 1;
+       }
+
+       /*
+        * The section is not biggest or smallest mem_section in the zone, it
+        * only creates a hole in the zone. So in this case, we need not
+        * change the zone. But perhaps, the zone has only hole data. Thus
+        * it check the zone has only hole or not.
+        */
+       pfn = zone_start_pfn;
+       for (; pfn < zone_end_pfn; pfn += PAGES_PER_SECTION) {
+               ms = __pfn_to_section(pfn);
+
+               if (unlikely(!valid_section(ms)))
+                       continue;
+
+               if (page_zone(pfn_to_page(pfn)) != zone)
+                       continue;
+
+                /* If the section is current section, it continues the loop */
+               if (start_pfn == pfn)
+                       continue;
+
+               /* If we find valid section, we have nothing to do */
+               zone_span_writeunlock(zone);
+               return;
+       }
+
+       /* The zone has no valid section */
+       zone->zone_start_pfn = 0;
+       zone->spanned_pages = 0;
+       zone_span_writeunlock(zone);
+}
+
+static void shrink_pgdat_span(struct pglist_data *pgdat,
+                             unsigned long start_pfn, unsigned long end_pfn)
+{
+       unsigned long pgdat_start_pfn =  pgdat->node_start_pfn;
+       unsigned long pgdat_end_pfn =
+               pgdat->node_start_pfn + pgdat->node_spanned_pages;
+       unsigned long pfn;
+       struct mem_section *ms;
+       int nid = pgdat->node_id;
+
+       if (pgdat_start_pfn == start_pfn) {
+               /*
+                * If the section is smallest section in the pgdat, it need
+                * shrink pgdat->node_start_pfn and pgdat->node_spanned_pages.
+                * In this case, we find second smallest valid mem_section
+                * for shrinking zone.
+                */
+               pfn = find_smallest_section_pfn(nid, NULL, end_pfn,
+                                               pgdat_end_pfn);
+               if (pfn) {
+                       pgdat->node_start_pfn = pfn;
+                       pgdat->node_spanned_pages = pgdat_end_pfn - pfn;
+               }
+       } else if (pgdat_end_pfn == end_pfn) {
+               /*
+                * If the section is biggest section in the pgdat, it need
+                * shrink pgdat->node_spanned_pages.
+                * In this case, we find second biggest valid mem_section for
+                * shrinking zone.
+                */
+               pfn = find_biggest_section_pfn(nid, NULL, pgdat_start_pfn,
+                                              start_pfn);
+               if (pfn)
+                       pgdat->node_spanned_pages = pfn - pgdat_start_pfn + 1;
+       }
+
+       /*
+        * If the section is not biggest or smallest mem_section in the pgdat,
+        * it only creates a hole in the pgdat. So in this case, we need not
+        * change the pgdat.
+        * But perhaps, the pgdat has only hole data. Thus it check the pgdat
+        * has only hole or not.
+        */
+       pfn = pgdat_start_pfn;
+       for (; pfn < pgdat_end_pfn; pfn += PAGES_PER_SECTION) {
+               ms = __pfn_to_section(pfn);
+
+               if (unlikely(!valid_section(ms)))
+                       continue;
+
+               if (pfn_to_nid(pfn) != nid)
+                       continue;
+
+                /* If the section is current section, it continues the loop */
+               if (start_pfn == pfn)
+                       continue;
+
+               /* If we find valid section, we have nothing to do */
+               return;
+       }
+
+       /* The pgdat has no valid section */
+       pgdat->node_start_pfn = 0;
+       pgdat->node_spanned_pages = 0;
+}
+
+static void __remove_zone(struct zone *zone, unsigned long start_pfn)
+{
+       struct pglist_data *pgdat = zone->zone_pgdat;
+       int nr_pages = PAGES_PER_SECTION;
+       int zone_type;
+       unsigned long flags;
+
+       zone_type = zone - pgdat->node_zones;
+
+       pgdat_resize_lock(zone->zone_pgdat, &flags);
+       shrink_zone_span(zone, start_pfn, start_pfn + nr_pages);
+       shrink_pgdat_span(pgdat, start_pfn, start_pfn + nr_pages);
+       pgdat_resize_unlock(zone->zone_pgdat, &flags);
+}
+
 static int __remove_section(struct zone *zone, struct mem_section *ms)
 {
+       unsigned long start_pfn;
+       int scn_nr;
        int ret = -EINVAL;
 
        if (!valid_section(ms))
@@ -443,6 +646,10 @@ static int __remove_section(struct zone *zone, struct mem_section *ms)
        if (ret)
                return ret;
 
+       scn_nr = __section_nr(ms);
+       start_pfn = section_nr_to_pfn(scn_nr);
+       __remove_zone(zone, start_pfn);
+
        sparse_remove_one_section(zone, ms);
        return 0;
 }