Commit 60a5a19e7419ba0bc22ed01b3285e8940b42944c

Authored by Tang Chen
Committed by Linus Torvalds
1 parent 815121d2b5

memory-hotplug: remove sysfs file of node

Introduce a new function try_offline_node() to remove sysfs file of node
when all memory sections of this node are removed.  If some memory
sections of this node are not removed, this function does nothing.

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: Yasuaki Ishimatsu <isimatu.yasuaki@jp.fujitsu.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>

Showing 3 changed files with 63 additions and 5 deletions Side-by-side Diff

drivers/acpi/acpi_memhotplug.c
... ... @@ -280,9 +280,11 @@
280 280  
281 281 static int acpi_memory_remove_memory(struct acpi_memory_device *mem_device)
282 282 {
283   - int result = 0;
  283 + int result = 0, nid;
284 284 struct acpi_memory_info *info, *n;
285 285  
  286 + nid = acpi_get_node(mem_device->device->handle);
  287 +
286 288 list_for_each_entry_safe(info, n, &mem_device->res_list, list) {
287 289 if (info->failed)
288 290 /* The kernel does not use this memory block */
... ... @@ -295,7 +297,9 @@
295 297 */
296 298 return -EBUSY;
297 299  
298   - result = remove_memory(info->start_addr, info->length);
  300 + if (nid < 0)
  301 + nid = memory_add_physaddr_to_nid(info->start_addr);
  302 + result = remove_memory(nid, info->start_addr, info->length);
299 303 if (result)
300 304 return result;
301 305  
include/linux/memory_hotplug.h
... ... @@ -248,7 +248,7 @@
248 248 extern int offline_pages(unsigned long start_pfn, unsigned long nr_pages);
249 249 extern int offline_memory_block(struct memory_block *mem);
250 250 extern bool is_memblock_offlined(struct memory_block *mem);
251   -extern int remove_memory(u64 start, u64 size);
  251 +extern int remove_memory(int nid, u64 start, u64 size);
252 252 extern int sparse_add_one_section(struct zone *zone, unsigned long start_pfn,
253 253 int nr_pages);
254 254 extern void sparse_remove_one_section(struct zone *zone, struct mem_section *ms);
... ... @@ -29,6 +29,7 @@
29 29 #include <linux/suspend.h>
30 30 #include <linux/mm_inline.h>
31 31 #include <linux/firmware-map.h>
  32 +#include <linux/stop_machine.h>
32 33  
33 34 #include <asm/tlbflush.h>
34 35  
35 36  
... ... @@ -1679,8 +1680,59 @@
1679 1680 return ret;
1680 1681 }
1681 1682  
1682   -int __ref remove_memory(u64 start, u64 size)
  1683 +static int check_cpu_on_node(void *data)
1683 1684 {
  1685 + struct pglist_data *pgdat = data;
  1686 + int cpu;
  1687 +
  1688 + for_each_present_cpu(cpu) {
  1689 + if (cpu_to_node(cpu) == pgdat->node_id)
  1690 + /*
  1691 + * the cpu on this node isn't removed, and we can't
  1692 + * offline this node.
  1693 + */
  1694 + return -EBUSY;
  1695 + }
  1696 +
  1697 + return 0;
  1698 +}
  1699 +
  1700 +/* offline the node if all memory sections of this node are removed */
  1701 +static void try_offline_node(int nid)
  1702 +{
  1703 + unsigned long start_pfn = NODE_DATA(nid)->node_start_pfn;
  1704 + unsigned long end_pfn = start_pfn + NODE_DATA(nid)->node_spanned_pages;
  1705 + unsigned long pfn;
  1706 +
  1707 + for (pfn = start_pfn; pfn < end_pfn; pfn += PAGES_PER_SECTION) {
  1708 + unsigned long section_nr = pfn_to_section_nr(pfn);
  1709 +
  1710 + if (!present_section_nr(section_nr))
  1711 + continue;
  1712 +
  1713 + if (pfn_to_nid(pfn) != nid)
  1714 + continue;
  1715 +
  1716 + /*
  1717 + * some memory sections of this node are not removed, and we
  1718 + * can't offline node now.
  1719 + */
  1720 + return;
  1721 + }
  1722 +
  1723 + if (stop_machine(check_cpu_on_node, NODE_DATA(nid), NULL))
  1724 + return;
  1725 +
  1726 + /*
  1727 + * all memory/cpu of this node are removed, we can offline this
  1728 + * node now.
  1729 + */
  1730 + node_set_offline(nid);
  1731 + unregister_one_node(nid);
  1732 +}
  1733 +
  1734 +int __ref remove_memory(int nid, u64 start, u64 size)
  1735 +{
1684 1736 unsigned long start_pfn, end_pfn;
1685 1737 int ret = 0;
1686 1738 int retry = 1;
... ... @@ -1734,6 +1786,8 @@
1734 1786  
1735 1787 arch_remove_memory(start, size);
1736 1788  
  1789 + try_offline_node(nid);
  1790 +
1737 1791 unlock_memory_hotplug();
1738 1792  
1739 1793 return 0;
... ... @@ -1743,7 +1797,7 @@
1743 1797 {
1744 1798 return -EINVAL;
1745 1799 }
1746   -int remove_memory(u64 start, u64 size)
  1800 +int remove_memory(int nid, u64 start, u64 size)
1747 1801 {
1748 1802 return -EINVAL;
1749 1803 }