[PATCH] Fix sparsemem on Cell

Fix an oops experienced on the Cell architecture when init-time functions, early_*(), are called at runtime. It alters the call paths to make sure that the callers explicitly say whether the call is being made on behalf of a hotplug even, or happening at boot-time. It has been compile tested on ppc64, ia64, s390, i386 and x86_64. Acked-by: Arnd Bergmann <arndb@de.ibm.com> Signed-off-by: Dave Hansen <haveblue@us.ibm.com> Cc: Yasunori Goto <y-goto@jp.fujitsu.com> Acked-by: Andy Whitcroft <apw@shadowen.org> Cc: Christoph Lameter <clameter@engr.sgi.com> Cc: Martin Schwidefsky <schwidefsky@de.ibm.com> Acked-by: Heiko Carstens <heiko.carstens@de.ibm.com> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Paul Mackerras <paulus@samba.org> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] Fix sparsemem on Cell
Fix an oops experienced on the Cell architecture when init-time functions, early_*(), are called at runtime. It alters the call paths to make sure that the callers explicitly say whether the call is being made on behalf of a hotplug even, or happening at boot-time. It has been compile tested on ppc64, ia64, s390, i386 and x86_64. Acked-by: Arnd Bergmann <arndb@de.ibm.com> Signed-off-by: Dave Hansen <haveblue@us.ibm.com> Cc: Yasunori Goto <y-goto@jp.fujitsu.com> Acked-by: Andy Whitcroft <apw@shadowen.org> Cc: Christoph Lameter <clameter@engr.sgi.com> Cc: Martin Schwidefsky <schwidefsky@de.ibm.com> Acked-by: Heiko Carstens <heiko.carstens@de.ibm.com> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Paul Mackerras <paulus@samba.org> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Dave Hansen · Linus Torvalds · Eric Lee · Eric Lee · Eric Lee · Eric Lee
1 parent 47a4d5be7c
Showing 6 changed files with 34 additions and 16 deletions Side-by-side Diff
arch/ia64/mm/init.c
arch/s390/mm/vmem.c
include/linux/mm.h
include/linux/mmzone.h
mm/memory_hotplug.c
mm/page_alloc.c
@@ -543,7 +543,8 @@
  
 	if (map_start < map_end)
 		memmap_init_zone((unsigned long)(map_end - map_start),
-				 args->nid, args->zone, page_to_pfn(map_start));
+				 args->nid, args->zone, page_to_pfn(map_start),
+				 MEMMAP_EARLY);
 	return 0;
 }
  
@@ -552,7 +553,7 @@
 	     unsigned long start_pfn)
 {
 	if (!vmem_map)
-		memmap_init_zone(size, nid, zone, start_pfn);
+		memmap_init_zone(size, nid, zone, start_pfn, MEMMAP_EARLY);
 	else {
 		struct page *start;
 		struct memmap_init_callback_data args;
@@ -61,7 +61,8 @@
  
 		if (map_start < map_end)
 			memmap_init_zone((unsigned long)(map_end - map_start),
-					 nid, zone, page_to_pfn(map_start));
+					 nid, zone, page_to_pfn(map_start),
+					 MEMMAP_EARLY);
 	}
 }
  
@@ -978,7 +978,8 @@
 #endif /* CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID */
 #endif /* CONFIG_ARCH_POPULATES_NODE_MAP */
 extern void set_dma_reserve(unsigned long new_dma_reserve);
-extern void memmap_init_zone(unsigned long, int, unsigned long, unsigned long);
+extern void memmap_init_zone(unsigned long, int, unsigned long,
+				unsigned long, enum memmap_context);
 extern void setup_per_zone_pages_min(void);
 extern void mem_init(void);
 extern void show_mem(void);
@@ -450,9 +450,13 @@
 void wakeup_kswapd(struct zone *zone, int order);
 int zone_watermark_ok(struct zone *z, int order, unsigned long mark,
 		int classzone_idx, int alloc_flags);
-
+enum memmap_context {
+	MEMMAP_EARLY,
+	MEMMAP_HOTPLUG,
+};
 extern int init_currently_empty_zone(struct zone *zone, unsigned long start_pfn,
-				     unsigned long size);
+				     unsigned long size,
+				     enum memmap_context context);
  
 #ifdef CONFIG_HAVE_MEMORY_PRESENT
 void memory_present(int nid, unsigned long start, unsigned long end);
@@ -67,11 +67,13 @@
 	zone_type = zone - pgdat->node_zones;
 	if (!populated_zone(zone)) {
 		int ret = 0;
-		ret = init_currently_empty_zone(zone, phys_start_pfn, nr_pages);
+		ret = init_currently_empty_zone(zone, phys_start_pfn,
+						nr_pages, MEMMAP_HOTPLUG);
 		if (ret < 0)
 			return ret;
 	}
-	memmap_init_zone(nr_pages, nid, zone_type, phys_start_pfn);
+	memmap_init_zone(nr_pages, nid, zone_type,
+			 phys_start_pfn, MEMMAP_HOTPLUG);
 	return 0;
 }
  
@@ -1956,17 +1956,24 @@
  * done. Non-atomic initialization, single-pass.
  */
 void __meminit memmap_init_zone(unsigned long size, int nid, unsigned long zone,
-		unsigned long start_pfn)
+		unsigned long start_pfn, enum memmap_context context)
 {
 	struct page *page;
 	unsigned long end_pfn = start_pfn + size;
 	unsigned long pfn;
  
 	for (pfn = start_pfn; pfn < end_pfn; pfn++) {
-		if (!early_pfn_valid(pfn))
-			continue;
-		if (!early_pfn_in_nid(pfn, nid))
-			continue;
+		/*
+		 * There can be holes in boot-time mem_map[]s
+		 * handed to this function.  They do not
+		 * exist on hotplugged memory.
+		 */
+		if (context == MEMMAP_EARLY) {
+			if (!early_pfn_valid(pfn))
+				continue;
+			if (!early_pfn_in_nid(pfn, nid))
+				continue;
+		}
 		page = pfn_to_page(pfn);
 		set_page_links(page, zone, nid, pfn);
 		init_page_count(page);
@@ -1993,7 +2000,7 @@
  
 #ifndef __HAVE_ARCH_MEMMAP_INIT
 #define memmap_init(size, nid, zone, start_pfn) \
-	memmap_init_zone((size), (nid), (zone), (start_pfn))
+	memmap_init_zone((size), (nid), (zone), (start_pfn), MEMMAP_EARLY)
 #endif
  
 static int __cpuinit zone_batchsize(struct zone *zone)
@@ -2239,7 +2246,8 @@
  
 __meminit int init_currently_empty_zone(struct zone *zone,
 					unsigned long zone_start_pfn,
-					unsigned long size)
+					unsigned long size,
+					enum memmap_context context)
 {
 	struct pglist_data *pgdat = zone->zone_pgdat;
 	int ret;
@@ -2683,7 +2691,8 @@
 		if (!size)
 			continue;
  
-		ret = init_currently_empty_zone(zone, zone_start_pfn, size);
+		ret = init_currently_empty_zone(zone, zone_start_pfn,
+						size, MEMMAP_EARLY);
 		BUG_ON(ret);
 		zone_start_pfn += size;
 	}
...	...	@@ -543,7 +543,8 @@
543	543
544	544	if (map_start < map_end)
545	545	memmap_init_zone((unsigned long)(map_end - map_start),
546		- args->nid, args->zone, page_to_pfn(map_start));
	546	+ args->nid, args->zone, page_to_pfn(map_start),
	547	+ MEMMAP_EARLY);
547	548	return 0;
548	549	}
549	550
...	...	@@ -552,7 +553,7 @@
552	553	unsigned long start_pfn)
553	554	{
554	555	if (!vmem_map)
555		- memmap_init_zone(size, nid, zone, start_pfn);
	556	+ memmap_init_zone(size, nid, zone, start_pfn, MEMMAP_EARLY);
556	557	else {
557	558	struct page *start;
558	559	struct memmap_init_callback_data args;
...	...	@@ -61,7 +61,8 @@
61	61
62	62	if (map_start < map_end)
63	63	memmap_init_zone((unsigned long)(map_end - map_start),
64		- nid, zone, page_to_pfn(map_start));
	64	+ nid, zone, page_to_pfn(map_start),
	65	+ MEMMAP_EARLY);
65	66	}
66	67	}
67	68
...	...	@@ -978,7 +978,8 @@
978	978	#endif /* CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID */
979	979	#endif /* CONFIG_ARCH_POPULATES_NODE_MAP */
980	980	extern void set_dma_reserve(unsigned long new_dma_reserve);
981		-extern void memmap_init_zone(unsigned long, int, unsigned long, unsigned long);
	981	+extern void memmap_init_zone(unsigned long, int, unsigned long,
	982	+ unsigned long, enum memmap_context);
982	983	extern void setup_per_zone_pages_min(void);
983	984	extern void mem_init(void);
984	985	extern void show_mem(void);
...	...	@@ -450,9 +450,13 @@
450	450	void wakeup_kswapd(struct zone *zone, int order);
451	451	int zone_watermark_ok(struct zone *z, int order, unsigned long mark,
452	452	int classzone_idx, int alloc_flags);
453		-
	453	+enum memmap_context {
	454	+ MEMMAP_EARLY,
	455	+ MEMMAP_HOTPLUG,
	456	+};
454	457	extern int init_currently_empty_zone(struct zone *zone, unsigned long start_pfn,
455		- unsigned long size);
	458	+ unsigned long size,
	459	+ enum memmap_context context);
456	460
457	461	#ifdef CONFIG_HAVE_MEMORY_PRESENT
458	462	void memory_present(int nid, unsigned long start, unsigned long end);
...	...	@@ -67,11 +67,13 @@
67	67	zone_type = zone - pgdat->node_zones;
68	68	if (!populated_zone(zone)) {
69	69	int ret = 0;
70		- ret = init_currently_empty_zone(zone, phys_start_pfn, nr_pages);
	70	+ ret = init_currently_empty_zone(zone, phys_start_pfn,
	71	+ nr_pages, MEMMAP_HOTPLUG);
71	72	if (ret < 0)
72	73	return ret;
73	74	}
74		- memmap_init_zone(nr_pages, nid, zone_type, phys_start_pfn);
	75	+ memmap_init_zone(nr_pages, nid, zone_type,
	76	+ phys_start_pfn, MEMMAP_HOTPLUG);
75	77	return 0;
76	78	}
77	79
...	...	@@ -1956,17 +1956,24 @@
1956	1956	* done. Non-atomic initialization, single-pass.
1957	1957	*/
1958	1958	void __meminit memmap_init_zone(unsigned long size, int nid, unsigned long zone,
1959		- unsigned long start_pfn)
	1959	+ unsigned long start_pfn, enum memmap_context context)
1960	1960	{
1961	1961	struct page *page;
1962	1962	unsigned long end_pfn = start_pfn + size;
1963	1963	unsigned long pfn;
1964	1964
1965	1965	for (pfn = start_pfn; pfn < end_pfn; pfn++) {
1966		- if (!early_pfn_valid(pfn))
1967		- continue;
1968		- if (!early_pfn_in_nid(pfn, nid))
1969		- continue;
	1966	+ /*
	1967	+ * There can be holes in boot-time mem_map[]s
	1968	+ * handed to this function. They do not
	1969	+ * exist on hotplugged memory.
	1970	+ */
	1971	+ if (context == MEMMAP_EARLY) {
	1972	+ if (!early_pfn_valid(pfn))
	1973	+ continue;
	1974	+ if (!early_pfn_in_nid(pfn, nid))
	1975	+ continue;
	1976	+ }
1970	1977	page = pfn_to_page(pfn);
1971	1978	set_page_links(page, zone, nid, pfn);
1972	1979	init_page_count(page);
...	...	@@ -1993,7 +2000,7 @@
1993	2000
1994	2001	#ifndef __HAVE_ARCH_MEMMAP_INIT
1995	2002	#define memmap_init(size, nid, zone, start_pfn) \
1996		- memmap_init_zone((size), (nid), (zone), (start_pfn))
	2003	+ memmap_init_zone((size), (nid), (zone), (start_pfn), MEMMAP_EARLY)
1997	2004	#endif
1998	2005
1999	2006	static int __cpuinit zone_batchsize(struct zone *zone)
...	...	@@ -2239,7 +2246,8 @@
2239	2246
2240	2247	__meminit int init_currently_empty_zone(struct zone *zone,
2241	2248	unsigned long zone_start_pfn,
2242		- unsigned long size)
	2249	+ unsigned long size,
	2250	+ enum memmap_context context)
2243	2251	{
2244	2252	struct pglist_data *pgdat = zone->zone_pgdat;
2245	2253	int ret;
...	...	@@ -2683,7 +2691,8 @@
2683	2691	if (!size)
2684	2692	continue;
2685	2693
2686		- ret = init_currently_empty_zone(zone, zone_start_pfn, size);
	2694	+ ret = init_currently_empty_zone(zone, zone_start_pfn,
	2695	+ size, MEMMAP_EARLY);
2687	2696	BUG_ON(ret);
2688	2697	zone_start_pfn += size;
2689	2698	}