memory-hotplug: update memory block's state and notify userspace

remove_memory() will be called when hot removing a memory device. But even if offlining memory, we cannot notice it. So the patch updates the memory block's state and sends notification to userspace. Additionally, the memory device may contain more than one memory block. If the memory block has been offlined, __offline_pages() will fail. So we should try to offline one memory block at a time. Thus remove_memory() also check each memory block's state. So there is no need to check the memory block's state before calling remove_memory(). Signed-off-by: Wen Congyang <wency@cn.fujitsu.com> Signed-off-by: Yasuaki Ishimatsu <isimatu.yasuaki@jp.fujitsu.com> Cc: David Rientjes <rientjes@google.com> Cc: Jiang Liu <liuj97@gmail.com> Cc: Len Brown <len.brown@intel.com> Cc: Christoph Lameter <cl@linux.com> Cc: Minchan Kim <minchan.kim@gmail.com> Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

memory-hotplug: update memory block's state and notify userspace
remove_memory() will be called when hot removing a memory device. But even if offlining memory, we cannot notice it. So the patch updates the memory block's state and sends notification to userspace. Additionally, the memory device may contain more than one memory block. If the memory block has been offlined, __offline_pages() will fail. So we should try to offline one memory block at a time. Thus remove_memory() also check each memory block's state. So there is no need to check the memory block's state before calling remove_memory(). Signed-off-by: Wen Congyang <wency@cn.fujitsu.com> Signed-off-by: Yasuaki Ishimatsu <isimatu.yasuaki@jp.fujitsu.com> Cc: David Rientjes <rientjes@google.com> Cc: Jiang Liu <liuj97@gmail.com> Cc: Len Brown <len.brown@intel.com> Cc: Christoph Lameter <cl@linux.com> Cc: Minchan Kim <minchan.kim@gmail.com> Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Wen Congyang · Linus Torvalds
1 parent a16cee10c7
Showing 3 changed files with 61 additions and 5 deletions Side-by-side Diff
drivers/base/memory.c
include/linux/memory_hotplug.h
mm/memory_hotplug.c
@@ -275,13 +275,11 @@
 	return ret;
 }
  
-static int memory_block_change_state(struct memory_block *mem,
+static int __memory_block_change_state(struct memory_block *mem,
 		unsigned long to_state, unsigned long from_state_req)
 {
 	int ret = 0;
  
-	mutex_lock(&mem->state_mutex);
-
 	if (mem->state != from_state_req) {
 		ret = -EINVAL;
 		goto out;
  
@@ -309,10 +307,20 @@
 		break;
 	}
 out:
-	mutex_unlock(&mem->state_mutex);
 	return ret;
 }
  
+static int memory_block_change_state(struct memory_block *mem,
+		unsigned long to_state, unsigned long from_state_req)
+{
+	int ret;
+
+	mutex_lock(&mem->state_mutex);
+	ret = __memory_block_change_state(mem, to_state, from_state_req);
+	mutex_unlock(&mem->state_mutex);
+
+	return ret;
+}
 static ssize_t
 store_mem_state(struct device *dev,
 		struct device_attribute *attr, const char *buf, size_t count)
@@ -650,6 +658,21 @@
 		return -EINVAL;
  
 	return remove_memory_block(0, section, 0);
+}
+
+/*
+ * offline one memory block. If the memory block has been offlined, do nothing.
+ */
+int offline_memory_block(struct memory_block *mem)
+{
+	int ret = 0;
+
+	mutex_lock(&mem->state_mutex);
+	if (mem->state != MEM_OFFLINE)
+		ret = __memory_block_change_state(mem, MEM_OFFLINE, MEM_ONLINE);
+	mutex_unlock(&mem->state_mutex);
+
+	return ret;
 }
  
 /*
@@ -10,6 +10,7 @@
 struct zone;
 struct pglist_data;
 struct mem_section;
+struct memory_block;
  
 #ifdef CONFIG_MEMORY_HOTPLUG
  
@@ -234,6 +235,7 @@
 extern int add_memory(int nid, u64 start, u64 size);
 extern int arch_add_memory(int nid, u64 start, u64 size);
 extern int offline_pages(unsigned long start_pfn, unsigned long nr_pages);
+extern int offline_memory_block(struct memory_block *mem);
 extern int remove_memory(u64 start, u64 size);
 extern int sparse_add_one_section(struct zone *zone, unsigned long start_pfn,
 								int nr_pages);
@@ -1014,11 +1014,42 @@
  
 int remove_memory(u64 start, u64 size)
 {
+	struct memory_block *mem = NULL;
+	struct mem_section *section;
 	unsigned long start_pfn, end_pfn;
+	unsigned long pfn, section_nr;
+	int ret;
  
 	start_pfn = PFN_DOWN(start);
 	end_pfn = start_pfn + PFN_DOWN(size);
-	return __offline_pages(start_pfn, end_pfn, 120 * HZ);
+
+	for (pfn = start_pfn; pfn < end_pfn; pfn += PAGES_PER_SECTION) {
+		section_nr = pfn_to_section_nr(pfn);
+		if (!present_section_nr(section_nr))
+			continue;
+
+		section = __nr_to_section(section_nr);
+		/* same memblock? */
+		if (mem)
+			if ((section_nr >= mem->start_section_nr) &&
+			    (section_nr <= mem->end_section_nr))
+				continue;
+
+		mem = find_memory_block_hinted(section, mem);
+		if (!mem)
+			continue;
+
+		ret = offline_memory_block(mem);
+		if (ret) {
+			kobject_put(&mem->dev.kobj);
+			return ret;
+		}
+	}
+
+	if (mem)
+		kobject_put(&mem->dev.kobj);
+
+	return 0;
 }
 #else
 int offline_pages(unsigned long start_pfn, unsigned long nr_pages)
...	...	@@ -275,13 +275,11 @@
275	275	return ret;
276	276	}
277	277
278		-static int memory_block_change_state(struct memory_block *mem,
	278	+static int __memory_block_change_state(struct memory_block *mem,
279	279	unsigned long to_state, unsigned long from_state_req)
280	280	{
281	281	int ret = 0;
282	282
283		- mutex_lock(&mem->state_mutex);
284		-
285	283	if (mem->state != from_state_req) {
286	284	ret = -EINVAL;
287	285	goto out;
288	286
...	...	@@ -309,10 +307,20 @@
309	307	break;
310	308	}
311	309	out:
312		- mutex_unlock(&mem->state_mutex);
313	310	return ret;
314	311	}
315	312
	313	+static int memory_block_change_state(struct memory_block *mem,
	314	+ unsigned long to_state, unsigned long from_state_req)
	315	+{
	316	+ int ret;
	317	+
	318	+ mutex_lock(&mem->state_mutex);
	319	+ ret = __memory_block_change_state(mem, to_state, from_state_req);
	320	+ mutex_unlock(&mem->state_mutex);
	321	+
	322	+ return ret;
	323	+}
316	324	static ssize_t
317	325	store_mem_state(struct device *dev,
318	326	struct device_attribute attr, const char buf, size_t count)
...	...	@@ -650,6 +658,21 @@
650	658	return -EINVAL;
651	659
652	660	return remove_memory_block(0, section, 0);
	661	+}
	662	+
	663	+/*
	664	+ * offline one memory block. If the memory block has been offlined, do nothing.
	665	+ */
	666	+int offline_memory_block(struct memory_block *mem)
	667	+{
	668	+ int ret = 0;
	669	+
	670	+ mutex_lock(&mem->state_mutex);
	671	+ if (mem->state != MEM_OFFLINE)
	672	+ ret = __memory_block_change_state(mem, MEM_OFFLINE, MEM_ONLINE);
	673	+ mutex_unlock(&mem->state_mutex);
	674	+
	675	+ return ret;
653	676	}
654	677
655	678	/*
...	...	@@ -10,6 +10,7 @@
10	10	struct zone;
11	11	struct pglist_data;
12	12	struct mem_section;
	13	+struct memory_block;
13	14
14	15	#ifdef CONFIG_MEMORY_HOTPLUG
15	16
...	...	@@ -234,6 +235,7 @@
234	235	extern int add_memory(int nid, u64 start, u64 size);
235	236	extern int arch_add_memory(int nid, u64 start, u64 size);
236	237	extern int offline_pages(unsigned long start_pfn, unsigned long nr_pages);
	238	+extern int offline_memory_block(struct memory_block *mem);
237	239	extern int remove_memory(u64 start, u64 size);
238	240	extern int sparse_add_one_section(struct zone *zone, unsigned long start_pfn,
239	241	int nr_pages);
...	...	@@ -1014,11 +1014,42 @@
1014	1014
1015	1015	int remove_memory(u64 start, u64 size)
1016	1016	{
	1017	+ struct memory_block *mem = NULL;
	1018	+ struct mem_section *section;
1017	1019	unsigned long start_pfn, end_pfn;
	1020	+ unsigned long pfn, section_nr;
	1021	+ int ret;
1018	1022
1019	1023	start_pfn = PFN_DOWN(start);
1020	1024	end_pfn = start_pfn + PFN_DOWN(size);
1021		- return __offline_pages(start_pfn, end_pfn, 120 * HZ);
	1025	+
	1026	+ for (pfn = start_pfn; pfn < end_pfn; pfn += PAGES_PER_SECTION) {
	1027	+ section_nr = pfn_to_section_nr(pfn);
	1028	+ if (!present_section_nr(section_nr))
	1029	+ continue;
	1030	+
	1031	+ section = __nr_to_section(section_nr);
	1032	+ /* same memblock? */
	1033	+ if (mem)
	1034	+ if ((section_nr >= mem->start_section_nr) &&
	1035	+ (section_nr <= mem->end_section_nr))
	1036	+ continue;
	1037	+
	1038	+ mem = find_memory_block_hinted(section, mem);
	1039	+ if (!mem)
	1040	+ continue;
	1041	+
	1042	+ ret = offline_memory_block(mem);
	1043	+ if (ret) {
	1044	+ kobject_put(&mem->dev.kobj);
	1045	+ return ret;
	1046	+ }
	1047	+ }
	1048	+
	1049	+ if (mem)
	1050	+ kobject_put(&mem->dev.kobj);
	1051	+
	1052	+ return 0;
1022	1053	}
1023	1054	#else
1024	1055	int offline_pages(unsigned long start_pfn, unsigned long nr_pages)