[linux-2.6-block.git] / arch / i386 / mm / discontig.c

/*
 * Written by: Patricia Gaughen <gone@us.ibm.com>, IBM Corporation
 * August 2002: added remote node KVA remap - Martin J. Bligh 
 *
 * Copyright (C) 2002, IBM Corp.
 *
 * All rights reserved.          
 *
 * This program is free software; you can redistribute it and/or modify
 * it under the terms of the GNU General Public License as published by
 * the Free Software Foundation; either version 2 of the License, or
 * (at your option) any later version.
 *
 * This program is distributed in the hope that it will be useful, but
 * WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE, GOOD TITLE or
 * NON INFRINGEMENT.  See the GNU General Public License for more
 * details.
 *
 * You should have received a copy of the GNU General Public License
 * along with this program; if not, write to the Free Software
 * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
 */

#include <linux/config.h>
#include <linux/mm.h>
#include <linux/bootmem.h>
#include <linux/mmzone.h>
#include <linux/highmem.h>
#include <linux/initrd.h>
#include <linux/nodemask.h>
#include <asm/e820.h>
#include <asm/setup.h>
#include <asm/mmzone.h>
#include <bios_ebda.h>

struct pglist_data *node_data[MAX_NUMNODES];
bootmem_data_t node0_bdata;

/*
 * numa interface - we expect the numa architecture specfic code to have
 *                  populated the following initialisation.
 *
 * 1) node_online_map  - the map of all nodes configured (online) in the system
 * 2) physnode_map     - the mapping between a pfn and owning node
 * 3) node_start_pfn   - the starting page frame number for a node
 * 3) node_end_pfn     - the ending page fram number for a node
 */

/*
 * physnode_map keeps track of the physical memory layout of a generic
 * numa node on a 256Mb break (each element of the array will
 * represent 256Mb of memory and will be marked by the node id.  so,
 * if the first gig is on node 0, and the second gig is on node 1
 * physnode_map will contain:
 *
 *     physnode_map[0-3] = 0;
 *     physnode_map[4-7] = 1;
 *     physnode_map[8- ] = -1;
 */
s8 physnode_map[MAX_ELEMENTS] = { [0 ... (MAX_ELEMENTS - 1)] = -1};

void memory_present(int nid, unsigned long start, unsigned long end)
{
	unsigned long pfn;

	printk(KERN_INFO "Node: %d, start_pfn: %ld, end_pfn: %ld\n",
			nid, start, end);
	printk(KERN_DEBUG "  Setting physnode_map array to node %d for pfns:\n", nid);
	printk(KERN_DEBUG "  ");
	for (pfn = start; pfn < end; pfn += PAGES_PER_ELEMENT) {
		physnode_map[pfn / PAGES_PER_ELEMENT] = nid;
		printk("%ld ", pfn);
	}
	printk("\n");
}

unsigned long node_memmap_size_bytes(int nid, unsigned long start_pfn,
					      unsigned long end_pfn)
{
	unsigned long nr_pages = end_pfn - start_pfn;

	if (!nr_pages)
		return 0;

	return (nr_pages + 1) * sizeof(struct page);
}

unsigned long node_start_pfn[MAX_NUMNODES];
unsigned long node_end_pfn[MAX_NUMNODES];

extern unsigned long find_max_low_pfn(void);
extern void find_max_pfn(void);
extern void one_highpage_init(struct page *, int, int);

extern struct e820map e820;
extern unsigned long init_pg_tables_end;
extern unsigned long highend_pfn, highstart_pfn;
extern unsigned long max_low_pfn;
extern unsigned long totalram_pages;
extern unsigned long totalhigh_pages;

#define LARGE_PAGE_BYTES (PTRS_PER_PTE * PAGE_SIZE)

unsigned long node_remap_start_pfn[MAX_NUMNODES];
unsigned long node_remap_size[MAX_NUMNODES];
unsigned long node_remap_offset[MAX_NUMNODES];
void *node_remap_start_vaddr[MAX_NUMNODES];
void set_pmd_pfn(unsigned long vaddr, unsigned long pfn, pgprot_t flags);

void *node_remap_end_vaddr[MAX_NUMNODES];
void *node_remap_alloc_vaddr[MAX_NUMNODES];

/*
 * FLAT - support for basic PC memory model with discontig enabled, essentially
 *        a single node with all available processors in it with a flat
 *        memory map.
 */
int __init get_memcfg_numa_flat(void)
{
	printk("NUMA - single node, flat memory mode\n");

	/* Run the memory configuration and find the top of memory. */
	find_max_pfn();
	node_start_pfn[0] = 0;
	node_end_pfn[0] = max_pfn;
	memory_present(0, 0, max_pfn);

        /* Indicate there is one node available. */
	nodes_clear(node_online_map);
	node_set_online(0);
	return 1;
}

/*
 * Find the highest page frame number we have available for the node
 */
static void __init find_max_pfn_node(int nid)
{
	if (node_end_pfn[nid] > max_pfn)
		node_end_pfn[nid] = max_pfn;
	/*
	 * if a user has given mem=XXXX, then we need to make sure 
	 * that the node _starts_ before that, too, not just ends
	 */
	if (node_start_pfn[nid] > max_pfn)
		node_start_pfn[nid] = max_pfn;
	if (node_start_pfn[nid] > node_end_pfn[nid])
		BUG();
}

/* Find the owning node for a pfn. */
int early_pfn_to_nid(unsigned long pfn)
{
	int nid;

	for_each_node(nid) {
		if (node_end_pfn[nid] == 0)
			break;
		if (node_start_pfn[nid] <= pfn && node_end_pfn[nid] >= pfn)
			return nid;
	}

	return 0;
}

/* 
 * Allocate memory for the pg_data_t for this node via a crude pre-bootmem
 * method.  For node zero take this from the bottom of memory, for
 * subsequent nodes place them at node_remap_start_vaddr which contains
 * node local data in physically node local memory.  See setup_memory()
 * for details.
 */
static void __init allocate_pgdat(int nid)
{
	if (nid && node_has_online_mem(nid))
		NODE_DATA(nid) = (pg_data_t *)node_remap_start_vaddr[nid];
	else {
		NODE_DATA(nid) = (pg_data_t *)(__va(min_low_pfn << PAGE_SHIFT));
		min_low_pfn += PFN_UP(sizeof(pg_data_t));
	}
}

void *alloc_remap(int nid, unsigned long size)
{
	void *allocation = node_remap_alloc_vaddr[nid];

	size = ALIGN(size, L1_CACHE_BYTES);

	if (!allocation || (allocation + size) >= node_remap_end_vaddr[nid])
		return 0;

	node_remap_alloc_vaddr[nid] += size;
	memset(allocation, 0, size);

	return allocation;
}

void __init remap_numa_kva(void)
{
	void *vaddr;
	unsigned long pfn;
	int node;

	for_each_online_node(node) {
		for (pfn=0; pfn < node_remap_size[node]; pfn += PTRS_PER_PTE) {
			vaddr = node_remap_start_vaddr[node]+(pfn<<PAGE_SHIFT);
			set_pmd_pfn((ulong) vaddr, 
				node_remap_start_pfn[node] + pfn, 
				PAGE_KERNEL_LARGE);
		}
	}
}

static unsigned long calculate_numa_remap_pages(void)
{
	int nid;
	unsigned long size, reserve_pages = 0;
	unsigned long pfn;

	for_each_online_node(nid) {
		/*
		 * The acpi/srat node info can show hot-add memroy zones
		 * where memory could be added but not currently present.
		 */
		if (node_start_pfn[nid] > max_pfn)
			continue;
		if (node_end_pfn[nid] > max_pfn)
			node_end_pfn[nid] = max_pfn;

		/* ensure the remap includes space for the pgdat. */
		size = node_remap_size[nid] + sizeof(pg_data_t);

		/* convert size to large (pmd size) pages, rounding up */
		size = (size + LARGE_PAGE_BYTES - 1) / LARGE_PAGE_BYTES;
		/* now the roundup is correct, convert to PAGE_SIZE pages */
		size = size * PTRS_PER_PTE;

		/*
		 * Validate the region we are allocating only contains valid
		 * pages.
		 */
		for (pfn = node_end_pfn[nid] - size;
		     pfn < node_end_pfn[nid]; pfn++)
			if (!page_is_ram(pfn))
				break;

		if (pfn != node_end_pfn[nid])
			size = 0;

		printk("Reserving %ld pages of KVA for lmem_map of node %d\n",
				size, nid);
		node_remap_size[nid] = size;
		node_remap_offset[nid] = reserve_pages;
		reserve_pages += size;
		printk("Shrinking node %d from %ld pages to %ld pages\n",
			nid, node_end_pfn[nid], node_end_pfn[nid] - size);
		node_end_pfn[nid] -= size;
		node_remap_start_pfn[nid] = node_end_pfn[nid];
	}
	printk("Reserving total of %ld pages for numa KVA remap\n",
			reserve_pages);
	return reserve_pages;
}

extern void setup_bootmem_allocator(void);
unsigned long __init setup_memory(void)
{
	int nid;
	unsigned long system_start_pfn, system_max_low_pfn;
	unsigned long reserve_pages;

	/*
	 * When mapping a NUMA machine we allocate the node_mem_map arrays
	 * from node local memory.  They are then mapped directly into KVA
	 * between zone normal and vmalloc space.  Calculate the size of
	 * this space and use it to adjust the boundry between ZONE_NORMAL
	 * and ZONE_HIGHMEM.
	 */
	find_max_pfn();
	get_memcfg_numa();

	reserve_pages = calculate_numa_remap_pages();

	/* partially used pages are not usable - thus round upwards */
	system_start_pfn = min_low_pfn = PFN_UP(init_pg_tables_end);

	system_max_low_pfn = max_low_pfn = find_max_low_pfn() - reserve_pages;
	printk("reserve_pages = %ld find_max_low_pfn() ~ %ld\n",
			reserve_pages, max_low_pfn + reserve_pages);
	printk("max_pfn = %ld\n", max_pfn);
#ifdef CONFIG_HIGHMEM
	highstart_pfn = highend_pfn = max_pfn;
	if (max_pfn > system_max_low_pfn)
		highstart_pfn = system_max_low_pfn;
	printk(KERN_NOTICE "%ldMB HIGHMEM available.\n",
	       pages_to_mb(highend_pfn - highstart_pfn));
#endif
	printk(KERN_NOTICE "%ldMB LOWMEM available.\n",
			pages_to_mb(system_max_low_pfn));
	printk("min_low_pfn = %ld, max_low_pfn = %ld, highstart_pfn = %ld\n", 
			min_low_pfn, max_low_pfn, highstart_pfn);

	printk("Low memory ends at vaddr %08lx\n",
			(ulong) pfn_to_kaddr(max_low_pfn));
	for_each_online_node(nid) {
		node_remap_start_vaddr[nid] = pfn_to_kaddr(
				highstart_pfn + node_remap_offset[nid]);
		/* Init the node remap allocator */
		node_remap_end_vaddr[nid] = node_remap_start_vaddr[nid] +
			(node_remap_size[nid] * PAGE_SIZE);
		node_remap_alloc_vaddr[nid] = node_remap_start_vaddr[nid] +
			ALIGN(sizeof(pg_data_t), PAGE_SIZE);

		allocate_pgdat(nid);
		printk ("node %d will remap to vaddr %08lx - %08lx\n", nid,
			(ulong) node_remap_start_vaddr[nid],
			(ulong) pfn_to_kaddr(highstart_pfn
			   + node_remap_offset[nid] + node_remap_size[nid]));
	}
	printk("High memory starts at vaddr %08lx\n",
			(ulong) pfn_to_kaddr(highstart_pfn));
	vmalloc_earlyreserve = reserve_pages * PAGE_SIZE;
	for_each_online_node(nid)
		find_max_pfn_node(nid);

	memset(NODE_DATA(0), 0, sizeof(struct pglist_data));
	NODE_DATA(0)->bdata = &node0_bdata;
	setup_bootmem_allocator();
	return max_low_pfn;
}

void __init zone_sizes_init(void)
{
	int nid;

	/*
	 * Insert nodes into pgdat_list backward so they appear in order.
	 * Clobber node 0's links and NULL out pgdat_list before starting.
	 */
	pgdat_list = NULL;
	for (nid = MAX_NUMNODES - 1; nid >= 0; nid--) {
		if (!node_online(nid))
			continue;
		NODE_DATA(nid)->pgdat_next = pgdat_list;
		pgdat_list = NODE_DATA(nid);
	}

	for_each_online_node(nid) {
		unsigned long zones_size[MAX_NR_ZONES] = {0, 0, 0};
		unsigned long *zholes_size;
		unsigned int max_dma;

		unsigned long low = max_low_pfn;
		unsigned long start = node_start_pfn[nid];
		unsigned long high = node_end_pfn[nid];

		max_dma = virt_to_phys((char *)MAX_DMA_ADDRESS) >> PAGE_SHIFT;

		if (node_has_online_mem(nid)){
			if (start > low) {
#ifdef CONFIG_HIGHMEM
				BUG_ON(start > high);
				zones_size[ZONE_HIGHMEM] = high - start;
#endif
			} else {
				if (low < max_dma)
					zones_size[ZONE_DMA] = low;
				else {
					BUG_ON(max_dma > low);
					BUG_ON(low > high);
					zones_size[ZONE_DMA] = max_dma;
					zones_size[ZONE_NORMAL] = low - max_dma;
#ifdef CONFIG_HIGHMEM
					zones_size[ZONE_HIGHMEM] = high - low;
#endif
				}
			}
		}

		zholes_size = get_zholes_size(nid);

		free_area_init_node(nid, NODE_DATA(nid), zones_size, start,
				zholes_size);
	}
	return;
}

void __init set_highmem_pages_init(int bad_ppro) 
{
#ifdef CONFIG_HIGHMEM
	struct zone *zone;

	for_each_zone(zone) {
		unsigned long node_pfn, node_high_size, zone_start_pfn;
		struct page * zone_mem_map;
		
		if (!is_highmem(zone))
			continue;

		printk("Initializing %s for node %d\n", zone->name,
			zone->zone_pgdat->node_id);

		node_high_size = zone->spanned_pages;
		zone_mem_map = zone->zone_mem_map;
		zone_start_pfn = zone->zone_start_pfn;

		for (node_pfn = 0; node_pfn < node_high_size; node_pfn++) {
			one_highpage_init((struct page *)(zone_mem_map + node_pfn),
					  zone_start_pfn + node_pfn, bad_ppro);
		}
	}
	totalram_pages += totalhigh_pages;
#endif
}
Commit	Line	Data
1da177e4 LT	1	/*
	2	* Written by: Patricia Gaughen <gone@us.ibm.com>, IBM Corporation
	3	* August 2002: added remote node KVA remap - Martin J. Bligh
	4	*
	5	* Copyright (C) 2002, IBM Corp.
	6	*
	7	* All rights reserved.
	8	*
	9	* This program is free software; you can redistribute it and/or modify
	10	* it under the terms of the GNU General Public License as published by
	11	* the Free Software Foundation; either version 2 of the License, or
	12	* (at your option) any later version.
	13	*
	14	* This program is distributed in the hope that it will be useful, but
	15	* WITHOUT ANY WARRANTY; without even the implied warranty of
	16	* MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE, GOOD TITLE or
	17	* NON INFRINGEMENT. See the GNU General Public License for more
	18	* details.
	19	*
	20	* You should have received a copy of the GNU General Public License
	21	* along with this program; if not, write to the Free Software
	22	* Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
	23	*/
	24
	25	#include <linux/config.h>
	26	#include <linux/mm.h>
	27	#include <linux/bootmem.h>
	28	#include <linux/mmzone.h>
	29	#include <linux/highmem.h>
	30	#include <linux/initrd.h>
	31	#include <linux/nodemask.h>
	32	#include <asm/e820.h>
	33	#include <asm/setup.h>
	34	#include <asm/mmzone.h>
	35	#include <bios_ebda.h>
	36
	37	struct pglist_data *node_data[MAX_NUMNODES];
	38	bootmem_data_t node0_bdata;
	39
	40	/*
	41	* numa interface - we expect the numa architecture specfic code to have
	42	* populated the following initialisation.
	43	*
	44	* 1) node_online_map - the map of all nodes configured (online) in the system
	45	* 2) physnode_map - the mapping between a pfn and owning node
	46	* 3) node_start_pfn - the starting page frame number for a node
	47	* 3) node_end_pfn - the ending page fram number for a node
	48	*/
	49
	50	/*
	51	* physnode_map keeps track of the physical memory layout of a generic
	52	* numa node on a 256Mb break (each element of the array will
	53	* represent 256Mb of memory and will be marked by the node id. so,
	54	* if the first gig is on node 0, and the second gig is on node 1
	55	* physnode_map will contain:
	56	*
	57	* physnode_map[0-3] = 0;
	58	* physnode_map[4-7] = 1;
	59	* physnode_map[8- ] = -1;
	60	*/
	61	s8 physnode_map[MAX_ELEMENTS] = { [0 ... (MAX_ELEMENTS - 1)] = -1};
	62
	63	void memory_present(int nid, unsigned long start, unsigned long end)
	64	{
65	unsigned long pfn;
66
67	printk(KERN_INFO "Node: %d, start_pfn: %ld, end_pfn: %ld\n",
68	nid, start, end);
69	printk(KERN_DEBUG " Setting physnode_map array to node %d for pfns:\n", nid);
70	printk(KERN_DEBUG " ");
71	for (pfn = start; pfn < end; pfn += PAGES_PER_ELEMENT) {
72	physnode_map[pfn / PAGES_PER_ELEMENT] = nid;
73	printk("%ld ", pfn);
74	}
75	printk("\n");
76	}
77
78	unsigned long node_memmap_size_bytes(int nid, unsigned long start_pfn,
79	unsigned long end_pfn)
80	{
81	unsigned long nr_pages = end_pfn - start_pfn;
82
83	if (!nr_pages)
84	return 0;
85
86	return (nr_pages + 1) * sizeof(struct page);
87	}
88
89	unsigned long node_start_pfn[MAX_NUMNODES];
90	unsigned long node_end_pfn[MAX_NUMNODES];
91
92	extern unsigned long find_max_low_pfn(void);
93	extern void find_max_pfn(void);
94	extern void one_highpage_init(struct page *, int, int);
95
96	extern struct e820map e820;
97	extern unsigned long init_pg_tables_end;
98	extern unsigned long highend_pfn, highstart_pfn;
99	extern unsigned long max_low_pfn;
100	extern unsigned long totalram_pages;
101	extern unsigned long totalhigh_pages;
102
103	#define LARGE_PAGE_BYTES (PTRS_PER_PTE * PAGE_SIZE)
104
105	unsigned long node_remap_start_pfn[MAX_NUMNODES];
106	unsigned long node_remap_size[MAX_NUMNODES];
107	unsigned long node_remap_offset[MAX_NUMNODES];
108	void *node_remap_start_vaddr[MAX_NUMNODES];
109	void set_pmd_pfn(unsigned long vaddr, unsigned long pfn, pgprot_t flags);
110
6f167ec7 DH	111	void *node_remap_end_vaddr[MAX_NUMNODES];
	112	void *node_remap_alloc_vaddr[MAX_NUMNODES];
	113
1da177e4 LT	114	/*
	115	* FLAT - support for basic PC memory model with discontig enabled, essentially
	116	* a single node with all available processors in it with a flat
	117	* memory map.
	118	*/
	119	int __init get_memcfg_numa_flat(void)
	120	{
	121	printk("NUMA - single node, flat memory mode\n");
	122
	123	/* Run the memory configuration and find the top of memory. */
	124	find_max_pfn();
	125	node_start_pfn[0] = 0;
	126	node_end_pfn[0] = max_pfn;
	127	memory_present(0, 0, max_pfn);
	128
	129	/* Indicate there is one node available. */
	130	nodes_clear(node_online_map);
	131	node_set_online(0);
	132	return 1;
	133	}
	134
	135	/*
	136	* Find the highest page frame number we have available for the node
	137	*/
	138	static void __init find_max_pfn_node(int nid)
	139	{
	140	if (node_end_pfn[nid] > max_pfn)
	141	node_end_pfn[nid] = max_pfn;
	142	/*
	143	* if a user has given mem=XXXX, then we need to make sure
	144	* that the node _starts_ before that, too, not just ends
	145	*/
	146	if (node_start_pfn[nid] > max_pfn)
	147	node_start_pfn[nid] = max_pfn;
	148	if (node_start_pfn[nid] > node_end_pfn[nid])
	149	BUG();
	150	}
	151
c2ebaa42 DH	152	/* Find the owning node for a pfn. */
	153	int early_pfn_to_nid(unsigned long pfn)
	154	{
	155	int nid;
	156
	157	for_each_node(nid) {
	158	if (node_end_pfn[nid] == 0)
	159	break;
	160	if (node_start_pfn[nid] <= pfn && node_end_pfn[nid] >= pfn)
	161	return nid;
	162	}
	163
	164	return 0;
	165	}
	166
1da177e4 LT	167	/*
	168	* Allocate memory for the pg_data_t for this node via a crude pre-bootmem
	169	* method. For node zero take this from the bottom of memory, for
	170	* subsequent nodes place them at node_remap_start_vaddr which contains
	171	* node local data in physically node local memory. See setup_memory()
	172	* for details.
	173	*/
	174	static void __init allocate_pgdat(int nid)
	175	{
	176	if (nid && node_has_online_mem(nid))
	177	NODE_DATA(nid) = (pg_data_t *)node_remap_start_vaddr[nid];
	178	else {
	179	NODE_DATA(nid) = (pg_data_t *)(__va(min_low_pfn << PAGE_SHIFT));
	180	min_low_pfn += PFN_UP(sizeof(pg_data_t));
	181	}
	182	}
	183
6f167ec7 DH	184	void *alloc_remap(int nid, unsigned long size)
	185	{
	186	void *allocation = node_remap_alloc_vaddr[nid];
	187
	188	size = ALIGN(size, L1_CACHE_BYTES);
	189
	190	if (!allocation \|\| (allocation + size) >= node_remap_end_vaddr[nid])
	191	return 0;
	192
	193	node_remap_alloc_vaddr[nid] += size;
	194	memset(allocation, 0, size);
	195
	196	return allocation;
	197	}
	198
1da177e4 LT	199	void __init remap_numa_kva(void)
	200	{
	201	void *vaddr;
	202	unsigned long pfn;
	203	int node;
	204
	205	for_each_online_node(node) {
1da177e4 LT	206	for (pfn=0; pfn < node_remap_size[node]; pfn += PTRS_PER_PTE) {
	207	vaddr = node_remap_start_vaddr[node]+(pfn<<PAGE_SHIFT);
	208	set_pmd_pfn((ulong) vaddr,
	209	node_remap_start_pfn[node] + pfn,
	210	PAGE_KERNEL_LARGE);
	211	}
	212	}
	213	}
	214
	215	static unsigned long calculate_numa_remap_pages(void)
	216	{
	217	int nid;
	218	unsigned long size, reserve_pages = 0;
5b505b90	219	unsigned long pfn;
1da177e4 LT	220
1da177e4 LT	221	for_each_online_node(nid) {
1da177e4 LT	222	/*
	223	* The acpi/srat node info can show hot-add memroy zones
	224	* where memory could be added but not currently present.
	225	*/
	226	if (node_start_pfn[nid] > max_pfn)
	227	continue;
	228	if (node_end_pfn[nid] > max_pfn)
	229	node_end_pfn[nid] = max_pfn;
	230
	231	/* ensure the remap includes space for the pgdat. */
	232	size = node_remap_size[nid] + sizeof(pg_data_t);
	233
	234	/* convert size to large (pmd size) pages, rounding up */
	235	size = (size + LARGE_PAGE_BYTES - 1) / LARGE_PAGE_BYTES;
	236	/* now the roundup is correct, convert to PAGE_SIZE pages */
	237	size = size * PTRS_PER_PTE;
5b505b90 DH	238
	239	/*
	240	* Validate the region we are allocating only contains valid
	241	* pages.
	242	*/
	243	for (pfn = node_end_pfn[nid] - size;
	244	pfn < node_end_pfn[nid]; pfn++)
	245	if (!page_is_ram(pfn))
	246	break;
	247
	248	if (pfn != node_end_pfn[nid])
	249	size = 0;
	250
1da177e4 LT	251	printk("Reserving %ld pages of KVA for lmem_map of node %d\n",
	252	size, nid);
	253	node_remap_size[nid] = size;
1da177e4	254	node_remap_offset[nid] = reserve_pages;
6f167ec7	255	reserve_pages += size;
1da177e4 LT	256	printk("Shrinking node %d from %ld pages to %ld pages\n",
	257	nid, node_end_pfn[nid], node_end_pfn[nid] - size);
	258	node_end_pfn[nid] -= size;
	259	node_remap_start_pfn[nid] = node_end_pfn[nid];
	260	}
	261	printk("Reserving total of %ld pages for numa KVA remap\n",
	262	reserve_pages);
	263	return reserve_pages;
	264	}
	265
	266	extern void setup_bootmem_allocator(void);
	267	unsigned long __init setup_memory(void)
	268	{
	269	int nid;
	270	unsigned long system_start_pfn, system_max_low_pfn;
	271	unsigned long reserve_pages;
	272
	273	/*
	274	* When mapping a NUMA machine we allocate the node_mem_map arrays
	275	* from node local memory. They are then mapped directly into KVA
	276	* between zone normal and vmalloc space. Calculate the size of
	277	* this space and use it to adjust the boundry between ZONE_NORMAL
	278	* and ZONE_HIGHMEM.
	279	*/
	280	find_max_pfn();
	281	get_memcfg_numa();
	282
	283	reserve_pages = calculate_numa_remap_pages();
	284
	285	/* partially used pages are not usable - thus round upwards */
	286	system_start_pfn = min_low_pfn = PFN_UP(init_pg_tables_end);
	287
	288	system_max_low_pfn = max_low_pfn = find_max_low_pfn() - reserve_pages;
	289	printk("reserve_pages = %ld find_max_low_pfn() ~ %ld\n",
	290	reserve_pages, max_low_pfn + reserve_pages);
	291	printk("max_pfn = %ld\n", max_pfn);
	292	#ifdef CONFIG_HIGHMEM
	293	highstart_pfn = highend_pfn = max_pfn;
	294	if (max_pfn > system_max_low_pfn)
	295	highstart_pfn = system_max_low_pfn;
	296	printk(KERN_NOTICE "%ldMB HIGHMEM available.\n",
	297	pages_to_mb(highend_pfn - highstart_pfn));
	298	#endif
	299	printk(KERN_NOTICE "%ldMB LOWMEM available.\n",
	300	pages_to_mb(system_max_low_pfn));
	301	printk("min_low_pfn = %ld, max_low_pfn = %ld, highstart_pfn = %ld\n",
	302	min_low_pfn, max_low_pfn, highstart_pfn);
	303
	304	printk("Low memory ends at vaddr %08lx\n",
	305	(ulong) pfn_to_kaddr(max_low_pfn));
	306	for_each_online_node(nid) {
	307	node_remap_start_vaddr[nid] = pfn_to_kaddr(
6f167ec7 DH	308	highstart_pfn + node_remap_offset[nid]);
	309	/* Init the node remap allocator */
	310	node_remap_end_vaddr[nid] = node_remap_start_vaddr[nid] +
	311	(node_remap_size[nid] * PAGE_SIZE);
	312	node_remap_alloc_vaddr[nid] = node_remap_start_vaddr[nid] +
	313	ALIGN(sizeof(pg_data_t), PAGE_SIZE);
	314
1da177e4 LT	315	allocate_pgdat(nid);
	316	printk ("node %d will remap to vaddr %08lx - %08lx\n", nid,
	317	(ulong) node_remap_start_vaddr[nid],
6f167ec7 DH	318	(ulong) pfn_to_kaddr(highstart_pfn
6f167ec7 DH	319	+ node_remap_offset[nid] + node_remap_size[nid]));
1da177e4 LT	320	}
	321	printk("High memory starts at vaddr %08lx\n",
	322	(ulong) pfn_to_kaddr(highstart_pfn));
	323	vmalloc_earlyreserve = reserve_pages * PAGE_SIZE;
	324	for_each_online_node(nid)
	325	find_max_pfn_node(nid);
	326
	327	memset(NODE_DATA(0), 0, sizeof(struct pglist_data));
	328	NODE_DATA(0)->bdata = &node0_bdata;
	329	setup_bootmem_allocator();
	330	return max_low_pfn;
	331	}
	332
	333	void __init zone_sizes_init(void)
	334	{
	335	int nid;
	336
	337	/*
	338	* Insert nodes into pgdat_list backward so they appear in order.
	339	* Clobber node 0's links and NULL out pgdat_list before starting.
	340	*/
	341	pgdat_list = NULL;
	342	for (nid = MAX_NUMNODES - 1; nid >= 0; nid--) {
	343	if (!node_online(nid))
	344	continue;
	345	NODE_DATA(nid)->pgdat_next = pgdat_list;
	346	pgdat_list = NODE_DATA(nid);
	347	}
	348
	349	for_each_online_node(nid) {
	350	unsigned long zones_size[MAX_NR_ZONES] = {0, 0, 0};
	351	unsigned long *zholes_size;
	352	unsigned int max_dma;
	353
	354	unsigned long low = max_low_pfn;
	355	unsigned long start = node_start_pfn[nid];
	356	unsigned long high = node_end_pfn[nid];
	357
	358	max_dma = virt_to_phys((char *)MAX_DMA_ADDRESS) >> PAGE_SHIFT;
	359
	360	if (node_has_online_mem(nid)){
	361	if (start > low) {
	362	#ifdef CONFIG_HIGHMEM
	363	BUG_ON(start > high);
	364	zones_size[ZONE_HIGHMEM] = high - start;
	365	#endif
	366	} else {
	367	if (low < max_dma)
	368	zones_size[ZONE_DMA] = low;
	369	else {
	370	BUG_ON(max_dma > low);
	371	BUG_ON(low > high);
	372	zones_size[ZONE_DMA] = max_dma;
	373	zones_size[ZONE_NORMAL] = low - max_dma;
	374	#ifdef CONFIG_HIGHMEM
	375	zones_size[ZONE_HIGHMEM] = high - low;
	376	#endif
	377	}
	378	}
	379	}
	380
	381	zholes_size = get_zholes_size(nid);
6f167ec7 DH	382
	383	free_area_init_node(nid, NODE_DATA(nid), zones_size, start,
	384	zholes_size);
1da177e4 LT	385	}
	386	return;
	387	}
	388
	389	void __init set_highmem_pages_init(int bad_ppro)
	390	{
	391	#ifdef CONFIG_HIGHMEM
	392	struct zone *zone;
	393
	394	for_each_zone(zone) {
	395	unsigned long node_pfn, node_high_size, zone_start_pfn;
	396	struct page * zone_mem_map;
	397
	398	if (!is_highmem(zone))
	399	continue;
	400
	401	printk("Initializing %s for node %d\n", zone->name,
	402	zone->zone_pgdat->node_id);
	403
	404	node_high_size = zone->spanned_pages;
	405	zone_mem_map = zone->zone_mem_map;
	406	zone_start_pfn = zone->zone_start_pfn;
	407
	408	for (node_pfn = 0; node_pfn < node_high_size; node_pfn++) {
	409	one_highpage_init((struct page *)(zone_mem_map + node_pfn),
	410	zone_start_pfn + node_pfn, bad_ppro);
	411	}
	412	}
	413	totalram_pages += totalhigh_pages;
	414	#endif
	415	}