From: Janice M. Girouard <jgirouar@redhat.com>
Subject: Re: [RHEL 5 PPC PATCH] REGRESSION: enter xmon (.early_pfn_to_nid+0x0/0x50)duri  (fwd)
Date: Tue, 9 Jan 2007 20:52:17 -0500 (Eastern Standard Time)
Bugzilla: 220065
Message-Id: <Pine.WNT.4.64.0701092047560.252@IBM-3MTQI3AXJFW>
Changelog: ppc64: initialization of hotplug memory fixes


RHBZ#: 220065
https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=220065

Description:
------------
The following patch fixes 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.

Without this fix, the kernel will oops when memmap_init_zone calls
early_pfn_in_nid, which is marked __init.  Customers will not be able to
add memory to a partion.  This is a serious regression from previous RHEL5
builds.

RHEL Version Found:
-------------------
Kernel 2.6.18-1.2839.el

Upstream Status:
----------------
Community dialogue and acceptance can be viewed:
http://lkml.org/lkml/2006/12/19/208

Test Status:
------------
Tested by John Rose of IBM.  This is based & builds without error on
kernel-2.6.18-1.2839.el5.src.rpm.

Proposed Patch:
----------------
Please review and ACK for RHEL 5.0


diff -ru -X /home/dave/work/linux/exclude --exclude-from=/home/dave/work/linux/exclude rh-clean/arch/ia64/mm/init.c rh/arch/ia64/mm/init.c
--- rh-clean/arch/ia64/mm/init.c	2006-12-20 21:52:31.000000000 -0800
+++ rh/arch/ia64/mm/init.c	2007-01-09 11:17:22.000000000 -0800
@@ -550,7 +550,8 @@
 
 	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;
 }
 
@@ -559,7 +560,7 @@
 	     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;
diff -ru -X /home/dave/work/linux/exclude --exclude-from=/home/dave/work/linux/exclude rh-clean/include/linux/mm.h rh/include/linux/mm.h
--- rh-clean/include/linux/mm.h	2006-12-20 21:52:38.000000000 -0800
+++ rh/include/linux/mm.h	2007-01-09 11:18:00.000000000 -0800
@@ -916,7 +916,8 @@
 extern void free_area_init_node(int nid, pg_data_t *pgdat,
 	unsigned long * zones_size, unsigned long zone_start_pfn, 
 	unsigned long *zholes_size);
-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);
diff -ru -X /home/dave/work/linux/exclude --exclude-from=/home/dave/work/linux/exclude rh-clean/include/linux/mmzone.h rh/include/linux/mmzone.h
--- rh-clean/include/linux/mmzone.h	2006-12-20 21:52:28.000000000 -0800
+++ rh/include/linux/mmzone.h	2007-01-09 11:17:22.000000000 -0800
@@ -346,9 +346,13 @@
 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);
Only in rh: linux-2.6.18-1.2910.el5
diff -ru -X /home/dave/work/linux/exclude --exclude-from=/home/dave/work/linux/exclude rh-clean/mm/memory_hotplug.c rh/mm/memory_hotplug.c
--- rh-clean/mm/memory_hotplug.c	2006-09-19 20:42:06.000000000 -0700
+++ rh/mm/memory_hotplug.c	2007-01-09 11:18:55.000000000 -0800
@@ -36,11 +36,13 @@
 	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);
 	zonetable_add(zone, nid, zone_type, phys_start_pfn, nr_pages);
 	return 0;
 }
diff -ru -X /home/dave/work/linux/exclude --exclude-from=/home/dave/work/linux/exclude rh-clean/mm/page_alloc.c rh/mm/page_alloc.c
--- rh-clean/mm/page_alloc.c	2006-12-20 21:52:33.000000000 -0800
+++ rh/mm/page_alloc.c	2007-01-09 11:19:20.000000000 -0800
@@ -1666,17 +1666,24 @@
  * 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);
@@ -1717,7 +1724,7 @@
 
 #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)
@@ -1960,7 +1967,8 @@
 
 __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;
@@ -2038,7 +2046,8 @@
 			continue;
 
 		zonetable_add(zone, nid, j, zone_start_pfn, size);
-		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;
 	}