[linux.git] / mm / bootmem.c

/*
 *  linux/mm/bootmem.c
 *
 *  Copyright (C) 1999 Ingo Molnar
 *  Discontiguous memory support, Kanoj Sarcar, SGI, Nov 1999
 *
 *  simple boot-time physical memory area allocator and
 *  free memory collector. It's used to deal with reserved
 *  system memory and memory holes as well.
 */

#include <linux/mm.h>
#include <linux/kernel_stat.h>
#include <linux/swap.h>
#include <linux/interrupt.h>
#include <linux/init.h>
#include <linux/bootmem.h>
#include <linux/mmzone.h>
#include <linux/module.h>
#include <asm/dma.h>
#include <asm/io.h>
#include "internal.h"

/*
 * Access to this subsystem has to be serialized externally. (this is
 * true for the boot process anyway)
 */
unsigned long max_low_pfn;
unsigned long min_low_pfn;
unsigned long max_pfn;

EXPORT_SYMBOL(max_pfn);		/* This is exported so
				 * dma_get_required_mask(), which uses
				 * it, can be an inline function */

#ifdef CONFIG_CRASH_DUMP
/*
 * If we have booted due to a crash, max_pfn will be a very low value. We need
 * to know the amount of memory that the previous kernel used.
 */
unsigned long saved_max_pfn;
#endif

/* return the number of _pages_ that will be allocated for the boot bitmap */
unsigned long __init bootmem_bootmap_pages (unsigned long pages)
{
	unsigned long mapsize;

	mapsize = (pages+7)/8;
	mapsize = (mapsize + ~PAGE_MASK) & PAGE_MASK;
	mapsize >>= PAGE_SHIFT;

	return mapsize;
}

/*
 * Called once to set up the allocator itself.
 */
static unsigned long __init init_bootmem_core (pg_data_t *pgdat,
	unsigned long mapstart, unsigned long start, unsigned long end)
{
	bootmem_data_t *bdata = pgdat->bdata;
	unsigned long mapsize = ((end - start)+7)/8;

	pgdat->pgdat_next = pgdat_list;
	pgdat_list = pgdat;

	mapsize = ALIGN(mapsize, sizeof(long));
	bdata->node_bootmem_map = phys_to_virt(mapstart << PAGE_SHIFT);
	bdata->node_boot_start = (start << PAGE_SHIFT);
	bdata->node_low_pfn = end;

	/*
	 * Initially all pages are reserved - setup_arch() has to
	 * register free RAM areas explicitly.
	 */
	memset(bdata->node_bootmem_map, 0xff, mapsize);

	return mapsize;
}

/*
 * Marks a particular physical memory range as unallocatable. Usable RAM
 * might be used for boot-time allocations - or it might get added
 * to the free page pool later on.
 */
static void __init reserve_bootmem_core(bootmem_data_t *bdata, unsigned long addr, unsigned long size)
{
	unsigned long i;
	/*
	 * round up, partially reserved pages are considered
	 * fully reserved.
	 */
	unsigned long sidx = (addr - bdata->node_boot_start)/PAGE_SIZE;
	unsigned long eidx = (addr + size - bdata->node_boot_start + 
							PAGE_SIZE-1)/PAGE_SIZE;
	unsigned long end = (addr + size + PAGE_SIZE-1)/PAGE_SIZE;

	BUG_ON(!size);
	BUG_ON(sidx >= eidx);
	BUG_ON((addr >> PAGE_SHIFT) >= bdata->node_low_pfn);
	BUG_ON(end > bdata->node_low_pfn);

	for (i = sidx; i < eidx; i++)
		if (test_and_set_bit(i, bdata->node_bootmem_map)) {
#ifdef CONFIG_DEBUG_BOOTMEM
			printk("hm, page %08lx reserved twice.\n", i*PAGE_SIZE);
#endif
		}
}

static void __init free_bootmem_core(bootmem_data_t *bdata, unsigned long addr, unsigned long size)
{
	unsigned long i;
	unsigned long start;
	/*
	 * round down end of usable mem, partially free pages are
	 * considered reserved.
	 */
	unsigned long sidx;
	unsigned long eidx = (addr + size - bdata->node_boot_start)/PAGE_SIZE;
	unsigned long end = (addr + size)/PAGE_SIZE;

	BUG_ON(!size);
	BUG_ON(end > bdata->node_low_pfn);

	if (addr < bdata->last_success)
		bdata->last_success = addr;

	/*
	 * Round up the beginning of the address.
	 */
	start = (addr + PAGE_SIZE-1) / PAGE_SIZE;
	sidx = start - (bdata->node_boot_start/PAGE_SIZE);

	for (i = sidx; i < eidx; i++) {
		if (unlikely(!test_and_clear_bit(i, bdata->node_bootmem_map)))
			BUG();
	}
}

/*
 * We 'merge' subsequent allocations to save space. We might 'lose'
 * some fraction of a page if allocations cannot be satisfied due to
 * size constraints on boxes where there is physical RAM space
 * fragmentation - in these cases (mostly large memory boxes) this
 * is not a problem.
 *
 * On low memory boxes we get it right in 100% of the cases.
 *
 * alignment has to be a power of 2 value.
 *
 * NOTE:  This function is _not_ reentrant.
 */
static void * __init
__alloc_bootmem_core(struct bootmem_data *bdata, unsigned long size,
	      unsigned long align, unsigned long goal, unsigned long limit)
{
	unsigned long offset, remaining_size, areasize, preferred;
	unsigned long i, start = 0, incr, eidx, end_pfn = bdata->node_low_pfn;
	void *ret;

	if(!size) {
		printk("__alloc_bootmem_core(): zero-sized request\n");
		BUG();
	}
	BUG_ON(align & (align-1));

	if (limit && bdata->node_boot_start >= limit)
		return NULL;

        limit >>=PAGE_SHIFT;
	if (limit && end_pfn > limit)
		end_pfn = limit;

	eidx = end_pfn - (bdata->node_boot_start >> PAGE_SHIFT);
	offset = 0;
	if (align &&
	    (bdata->node_boot_start & (align - 1UL)) != 0)
		offset = (align - (bdata->node_boot_start & (align - 1UL)));
	offset >>= PAGE_SHIFT;

	/*
	 * We try to allocate bootmem pages above 'goal'
	 * first, then we try to allocate lower pages.
	 */
	if (goal && (goal >= bdata->node_boot_start) && 
	    ((goal >> PAGE_SHIFT) < end_pfn)) {
		preferred = goal - bdata->node_boot_start;

		if (bdata->last_success >= preferred)
			if (!limit || (limit && limit > bdata->last_success))
				preferred = bdata->last_success;
	} else
		preferred = 0;

	preferred = ALIGN(preferred, align) >> PAGE_SHIFT;
	preferred += offset;
	areasize = (size+PAGE_SIZE-1)/PAGE_SIZE;
	incr = align >> PAGE_SHIFT ? : 1;

restart_scan:
	for (i = preferred; i < eidx; i += incr) {
		unsigned long j;
		i = find_next_zero_bit(bdata->node_bootmem_map, eidx, i);
		i = ALIGN(i, incr);
		if (test_bit(i, bdata->node_bootmem_map))
			continue;
		for (j = i + 1; j < i + areasize; ++j) {
			if (j >= eidx)
				goto fail_block;
			if (test_bit (j, bdata->node_bootmem_map))
				goto fail_block;
		}
		start = i;
		goto found;
	fail_block:
		i = ALIGN(j, incr);
	}

	if (preferred > offset) {
		preferred = offset;
		goto restart_scan;
	}
	return NULL;

found:
	bdata->last_success = start << PAGE_SHIFT;
	BUG_ON(start >= eidx);

	/*
	 * Is the next page of the previous allocation-end the start
	 * of this allocation's buffer? If yes then we can 'merge'
	 * the previous partial page with this allocation.
	 */
	if (align < PAGE_SIZE &&
	    bdata->last_offset && bdata->last_pos+1 == start) {
		offset = ALIGN(bdata->last_offset, align);
		BUG_ON(offset > PAGE_SIZE);
		remaining_size = PAGE_SIZE-offset;
		if (size < remaining_size) {
			areasize = 0;
			/* last_pos unchanged */
			bdata->last_offset = offset+size;
			ret = phys_to_virt(bdata->last_pos*PAGE_SIZE + offset +
						bdata->node_boot_start);
		} else {
			remaining_size = size - remaining_size;
			areasize = (remaining_size+PAGE_SIZE-1)/PAGE_SIZE;
			ret = phys_to_virt(bdata->last_pos*PAGE_SIZE + offset +
						bdata->node_boot_start);
			bdata->last_pos = start+areasize-1;
			bdata->last_offset = remaining_size;
		}
		bdata->last_offset &= ~PAGE_MASK;
	} else {
		bdata->last_pos = start + areasize - 1;
		bdata->last_offset = size & ~PAGE_MASK;
		ret = phys_to_virt(start * PAGE_SIZE + bdata->node_boot_start);
	}

	/*
	 * Reserve the area now:
	 */
	for (i = start; i < start+areasize; i++)
		if (unlikely(test_and_set_bit(i, bdata->node_bootmem_map)))
			BUG();
	memset(ret, 0, size);
	return ret;
}

static unsigned long __init free_all_bootmem_core(pg_data_t *pgdat)
{
	struct page *page;
	unsigned long pfn;
	bootmem_data_t *bdata = pgdat->bdata;
	unsigned long i, count, total = 0;
	unsigned long idx;
	unsigned long *map; 
	int gofast = 0;

	BUG_ON(!bdata->node_bootmem_map);

	count = 0;
	/* first extant page of the node */
	pfn = bdata->node_boot_start >> PAGE_SHIFT;
	idx = bdata->node_low_pfn - (bdata->node_boot_start >> PAGE_SHIFT);
	map = bdata->node_bootmem_map;
	/* Check physaddr is O(LOG2(BITS_PER_LONG)) page aligned */
	if (bdata->node_boot_start == 0 ||
	    ffs(bdata->node_boot_start) - PAGE_SHIFT > ffs(BITS_PER_LONG))
		gofast = 1;
	for (i = 0; i < idx; ) {
		unsigned long v = ~map[i / BITS_PER_LONG];

		if (gofast && v == ~0UL) {
			int j, order;

			page = pfn_to_page(pfn);
			count += BITS_PER_LONG;
			__ClearPageReserved(page);
			order = ffs(BITS_PER_LONG) - 1;
			set_page_refs(page, order);
			for (j = 1; j < BITS_PER_LONG; j++) {
				if (j + 16 < BITS_PER_LONG)
					prefetchw(page + j + 16);
				__ClearPageReserved(page + j);
			}
			__free_pages(page, order);
			i += BITS_PER_LONG;
			page += BITS_PER_LONG;
		} else if (v) {
			unsigned long m;

			page = pfn_to_page(pfn);
			for (m = 1; m && i < idx; m<<=1, page++, i++) {
				if (v & m) {
					count++;
					__ClearPageReserved(page);
					set_page_refs(page, 0);
					__free_page(page);
				}
			}
		} else {
			i+=BITS_PER_LONG;
		}
		pfn += BITS_PER_LONG;
	}
	total += count;

	/*
	 * Now free the allocator bitmap itself, it's not
	 * needed anymore:
	 */
	page = virt_to_page(bdata->node_bootmem_map);
	count = 0;
	for (i = 0; i < ((bdata->node_low_pfn-(bdata->node_boot_start >> PAGE_SHIFT))/8 + PAGE_SIZE-1)/PAGE_SIZE; i++,page++) {
		count++;
		__ClearPageReserved(page);
		set_page_count(page, 1);
		__free_page(page);
	}
	total += count;
	bdata->node_bootmem_map = NULL;

	return total;
}

unsigned long __init init_bootmem_node (pg_data_t *pgdat, unsigned long freepfn, unsigned long startpfn, unsigned long endpfn)
{
	return(init_bootmem_core(pgdat, freepfn, startpfn, endpfn));
}

void __init reserve_bootmem_node (pg_data_t *pgdat, unsigned long physaddr, unsigned long size)
{
	reserve_bootmem_core(pgdat->bdata, physaddr, size);
}

void __init free_bootmem_node (pg_data_t *pgdat, unsigned long physaddr, unsigned long size)
{
	free_bootmem_core(pgdat->bdata, physaddr, size);
}

unsigned long __init free_all_bootmem_node (pg_data_t *pgdat)
{
	return(free_all_bootmem_core(pgdat));
}

unsigned long __init init_bootmem (unsigned long start, unsigned long pages)
{
	max_low_pfn = pages;
	min_low_pfn = start;
	return(init_bootmem_core(NODE_DATA(0), start, 0, pages));
}

#ifndef CONFIG_HAVE_ARCH_BOOTMEM_NODE
void __init reserve_bootmem (unsigned long addr, unsigned long size)
{
	reserve_bootmem_core(NODE_DATA(0)->bdata, addr, size);
}
#endif /* !CONFIG_HAVE_ARCH_BOOTMEM_NODE */

void __init free_bootmem (unsigned long addr, unsigned long size)
{
	free_bootmem_core(NODE_DATA(0)->bdata, addr, size);
}

unsigned long __init free_all_bootmem (void)
{
	return(free_all_bootmem_core(NODE_DATA(0)));
}

void * __init __alloc_bootmem_limit (unsigned long size, unsigned long align, unsigned long goal,
				unsigned long limit)
{
	pg_data_t *pgdat = pgdat_list;
	void *ptr;

	for_each_pgdat(pgdat)
		if ((ptr = __alloc_bootmem_core(pgdat->bdata, size,
						 align, goal, limit)))
			return(ptr);

	/*
	 * Whoops, we cannot satisfy the allocation request.
	 */
	printk(KERN_ALERT "bootmem alloc of %lu bytes failed!\n", size);
	panic("Out of memory");
	return NULL;
}


void * __init __alloc_bootmem_node_limit (pg_data_t *pgdat, unsigned long size, unsigned long align,
				     unsigned long goal, unsigned long limit)
{
	void *ptr;

	ptr = __alloc_bootmem_core(pgdat->bdata, size, align, goal, limit);
	if (ptr)
		return (ptr);

	return __alloc_bootmem_limit(size, align, goal, limit);
}
Commit	Line	Data
1da177e4 LT	1	/*
	2	* linux/mm/bootmem.c
	3	*
	4	* Copyright (C) 1999 Ingo Molnar
	5	* Discontiguous memory support, Kanoj Sarcar, SGI, Nov 1999
	6	*
	7	* simple boot-time physical memory area allocator and
	8	* free memory collector. It's used to deal with reserved
	9	* system memory and memory holes as well.
	10	*/
	11
	12	#include <linux/mm.h>
	13	#include <linux/kernel_stat.h>
	14	#include <linux/swap.h>
	15	#include <linux/interrupt.h>
	16	#include <linux/init.h>
	17	#include <linux/bootmem.h>
	18	#include <linux/mmzone.h>
	19	#include <linux/module.h>
	20	#include <asm/dma.h>
	21	#include <asm/io.h>
	22	#include "internal.h"
	23
	24	/*
	25	* Access to this subsystem has to be serialized externally. (this is
	26	* true for the boot process anyway)
	27	*/
	28	unsigned long max_low_pfn;
	29	unsigned long min_low_pfn;
	30	unsigned long max_pfn;
	31
	32	EXPORT_SYMBOL(max_pfn); /* This is exported so
	33	* dma_get_required_mask(), which uses
	34	* it, can be an inline function */
	35
92aa63a5 VG	36	#ifdef CONFIG_CRASH_DUMP
	37	/*
	38	* If we have booted due to a crash, max_pfn will be a very low value. We need
	39	* to know the amount of memory that the previous kernel used.
	40	*/
	41	unsigned long saved_max_pfn;
	42	#endif
	43
1da177e4 LT	44	/* return the number of _pages_ that will be allocated for the boot bitmap */
	45	unsigned long __init bootmem_bootmap_pages (unsigned long pages)
	46	{
	47	unsigned long mapsize;
	48
	49	mapsize = (pages+7)/8;
	50	mapsize = (mapsize + ~PAGE_MASK) & PAGE_MASK;
	51	mapsize >>= PAGE_SHIFT;
	52
	53	return mapsize;
	54	}
	55
	56	/*
	57	* Called once to set up the allocator itself.
	58	*/
	59	static unsigned long __init init_bootmem_core (pg_data_t *pgdat,
	60	unsigned long mapstart, unsigned long start, unsigned long end)
	61	{
	62	bootmem_data_t *bdata = pgdat->bdata;
	63	unsigned long mapsize = ((end - start)+7)/8;
	64
6e3254c4 LT	65	pgdat->pgdat_next = pgdat_list;
6e3254c4 LT	66	pgdat_list = pgdat;
1da177e4	67
8c0e33c1	68	mapsize = ALIGN(mapsize, sizeof(long));
1da177e4 LT	69	bdata->node_bootmem_map = phys_to_virt(mapstart << PAGE_SHIFT);
	70	bdata->node_boot_start = (start << PAGE_SHIFT);
	71	bdata->node_low_pfn = end;
	72
	73	/*
	74	* Initially all pages are reserved - setup_arch() has to
	75	* register free RAM areas explicitly.
	76	*/
	77	memset(bdata->node_bootmem_map, 0xff, mapsize);
	78
	79	return mapsize;
	80	}
	81
	82	/*
	83	* Marks a particular physical memory range as unallocatable. Usable RAM
	84	* might be used for boot-time allocations - or it might get added
	85	* to the free page pool later on.
	86	*/
	87	static void __init reserve_bootmem_core(bootmem_data_t *bdata, unsigned long addr, unsigned long size)
	88	{
	89	unsigned long i;
	90	/*
	91	* round up, partially reserved pages are considered
	92	* fully reserved.
	93	*/
	94	unsigned long sidx = (addr - bdata->node_boot_start)/PAGE_SIZE;
	95	unsigned long eidx = (addr + size - bdata->node_boot_start +
	96	PAGE_SIZE-1)/PAGE_SIZE;
	97	unsigned long end = (addr + size + PAGE_SIZE-1)/PAGE_SIZE;
	98
	99	BUG_ON(!size);
	100	BUG_ON(sidx >= eidx);
	101	BUG_ON((addr >> PAGE_SHIFT) >= bdata->node_low_pfn);
	102	BUG_ON(end > bdata->node_low_pfn);
	103
	104	for (i = sidx; i < eidx; i++)
	105	if (test_and_set_bit(i, bdata->node_bootmem_map)) {
	106	#ifdef CONFIG_DEBUG_BOOTMEM
	107	printk("hm, page %08lx reserved twice.\n", i*PAGE_SIZE);
	108	#endif
	109	}
	110	}
	111
	112	static void __init free_bootmem_core(bootmem_data_t *bdata, unsigned long addr, unsigned long size)
	113	{
	114	unsigned long i;
	115	unsigned long start;
	116	/*
	117	* round down end of usable mem, partially free pages are
	118	* considered reserved.
	119	*/
	120	unsigned long sidx;
	121	unsigned long eidx = (addr + size - bdata->node_boot_start)/PAGE_SIZE;
	122	unsigned long end = (addr + size)/PAGE_SIZE;
	123
	124	BUG_ON(!size);
	125	BUG_ON(end > bdata->node_low_pfn);
	126
	127	if (addr < bdata->last_success)
	128	bdata->last_success = addr;
	129
	130	/*
	131	* Round up the beginning of the address.
	132	*/
133	start = (addr + PAGE_SIZE-1) / PAGE_SIZE;
134	sidx = start - (bdata->node_boot_start/PAGE_SIZE);
135
136	for (i = sidx; i < eidx; i++) {
137	if (unlikely(!test_and_clear_bit(i, bdata->node_bootmem_map)))
138	BUG();
139	}
140	}
141
142	/*
143	* We 'merge' subsequent allocations to save space. We might 'lose'
144	* some fraction of a page if allocations cannot be satisfied due to
145	* size constraints on boxes where there is physical RAM space
146	* fragmentation - in these cases (mostly large memory boxes) this
147	* is not a problem.
148	*
149	* On low memory boxes we get it right in 100% of the cases.
150	*
151	* alignment has to be a power of 2 value.
152	*
153	* NOTE: This function is _not_ reentrant.
154	*/
155	static void * __init
156	__alloc_bootmem_core(struct bootmem_data *bdata, unsigned long size,
281dd25c	157	unsigned long align, unsigned long goal, unsigned long limit)
1da177e4 LT	158	{
1da177e4 LT	159	unsigned long offset, remaining_size, areasize, preferred;
281dd25c	160	unsigned long i, start = 0, incr, eidx, end_pfn = bdata->node_low_pfn;
1da177e4 LT	161	void *ret;
	162
	163	if(!size) {
	164	printk("__alloc_bootmem_core(): zero-sized request\n");
	165	BUG();
	166	}
	167	BUG_ON(align & (align-1));
	168
281dd25c YG	169	if (limit && bdata->node_boot_start >= limit)
	170	return NULL;
	171
	172	limit >>=PAGE_SHIFT;
	173	if (limit && end_pfn > limit)
	174	end_pfn = limit;
	175
	176	eidx = end_pfn - (bdata->node_boot_start >> PAGE_SHIFT);
1da177e4 LT	177	offset = 0;
	178	if (align &&
	179	(bdata->node_boot_start & (align - 1UL)) != 0)
	180	offset = (align - (bdata->node_boot_start & (align - 1UL)));
	181	offset >>= PAGE_SHIFT;
	182
	183	/*
	184	* We try to allocate bootmem pages above 'goal'
	185	* first, then we try to allocate lower pages.
	186	*/
	187	if (goal && (goal >= bdata->node_boot_start) &&
281dd25c	188	((goal >> PAGE_SHIFT) < end_pfn)) {
1da177e4 LT	189	preferred = goal - bdata->node_boot_start;
	190
	191	if (bdata->last_success >= preferred)
281dd25c YG	192	if (!limit \|\| (limit && limit > bdata->last_success))
281dd25c YG	193	preferred = bdata->last_success;
1da177e4 LT	194	} else
	195	preferred = 0;
	196
8c0e33c1	197	preferred = ALIGN(preferred, align) >> PAGE_SHIFT;
1da177e4 LT	198	preferred += offset;
	199	areasize = (size+PAGE_SIZE-1)/PAGE_SIZE;
	200	incr = align >> PAGE_SHIFT ? : 1;
	201
	202	restart_scan:
	203	for (i = preferred; i < eidx; i += incr) {
	204	unsigned long j;
	205	i = find_next_zero_bit(bdata->node_bootmem_map, eidx, i);
	206	i = ALIGN(i, incr);
	207	if (test_bit(i, bdata->node_bootmem_map))
	208	continue;
	209	for (j = i + 1; j < i + areasize; ++j) {
	210	if (j >= eidx)
	211	goto fail_block;
	212	if (test_bit (j, bdata->node_bootmem_map))
	213	goto fail_block;
	214	}
	215	start = i;
	216	goto found;
	217	fail_block:
	218	i = ALIGN(j, incr);
	219	}
	220
	221	if (preferred > offset) {
	222	preferred = offset;
	223	goto restart_scan;
	224	}
	225	return NULL;
	226
	227	found:
	228	bdata->last_success = start << PAGE_SHIFT;
	229	BUG_ON(start >= eidx);
	230
	231	/*
	232	* Is the next page of the previous allocation-end the start
	233	* of this allocation's buffer? If yes then we can 'merge'
	234	* the previous partial page with this allocation.
	235	*/
	236	if (align < PAGE_SIZE &&
	237	bdata->last_offset && bdata->last_pos+1 == start) {
8c0e33c1	238	offset = ALIGN(bdata->last_offset, align);
1da177e4 LT	239	BUG_ON(offset > PAGE_SIZE);
	240	remaining_size = PAGE_SIZE-offset;
	241	if (size < remaining_size) {
	242	areasize = 0;
	243	/* last_pos unchanged */
	244	bdata->last_offset = offset+size;
	245	ret = phys_to_virt(bdata->last_pos*PAGE_SIZE + offset +
	246	bdata->node_boot_start);
	247	} else {
	248	remaining_size = size - remaining_size;
	249	areasize = (remaining_size+PAGE_SIZE-1)/PAGE_SIZE;
	250	ret = phys_to_virt(bdata->last_pos*PAGE_SIZE + offset +
	251	bdata->node_boot_start);
	252	bdata->last_pos = start+areasize-1;
	253	bdata->last_offset = remaining_size;
	254	}
	255	bdata->last_offset &= ~PAGE_MASK;
	256	} else {
	257	bdata->last_pos = start + areasize - 1;
	258	bdata->last_offset = size & ~PAGE_MASK;
	259	ret = phys_to_virt(start * PAGE_SIZE + bdata->node_boot_start);
	260	}
	261
	262	/*
	263	* Reserve the area now:
	264	*/
	265	for (i = start; i < start+areasize; i++)
	266	if (unlikely(test_and_set_bit(i, bdata->node_bootmem_map)))
	267	BUG();
	268	memset(ret, 0, size);
	269	return ret;
	270	}
	271
	272	static unsigned long __init free_all_bootmem_core(pg_data_t *pgdat)
	273	{
	274	struct page *page;
d41dee36	275	unsigned long pfn;
1da177e4 LT	276	bootmem_data_t *bdata = pgdat->bdata;
	277	unsigned long i, count, total = 0;
	278	unsigned long idx;
	279	unsigned long *map;
	280	int gofast = 0;
	281
	282	BUG_ON(!bdata->node_bootmem_map);
	283
	284	count = 0;
	285	/* first extant page of the node */
d41dee36	286	pfn = bdata->node_boot_start >> PAGE_SHIFT;
1da177e4 LT	287	idx = bdata->node_low_pfn - (bdata->node_boot_start >> PAGE_SHIFT);
	288	map = bdata->node_bootmem_map;
	289	/* Check physaddr is O(LOG2(BITS_PER_LONG)) page aligned */
	290	if (bdata->node_boot_start == 0 \|\|
	291	ffs(bdata->node_boot_start) - PAGE_SHIFT > ffs(BITS_PER_LONG))
	292	gofast = 1;
	293	for (i = 0; i < idx; ) {
	294	unsigned long v = ~map[i / BITS_PER_LONG];
d41dee36	295
1da177e4 LT	296	if (gofast && v == ~0UL) {
	297	int j, order;
	298
d41dee36	299	page = pfn_to_page(pfn);
1da177e4 LT	300	count += BITS_PER_LONG;
	301	__ClearPageReserved(page);
	302	order = ffs(BITS_PER_LONG) - 1;
	303	set_page_refs(page, order);
	304	for (j = 1; j < BITS_PER_LONG; j++) {
	305	if (j + 16 < BITS_PER_LONG)
	306	prefetchw(page + j + 16);
	307	__ClearPageReserved(page + j);
	308	}
	309	__free_pages(page, order);
	310	i += BITS_PER_LONG;
	311	page += BITS_PER_LONG;
	312	} else if (v) {
	313	unsigned long m;
d41dee36 AW	314
d41dee36 AW	315	page = pfn_to_page(pfn);
1da177e4 LT	316	for (m = 1; m && i < idx; m<<=1, page++, i++) {
	317	if (v & m) {
	318	count++;
	319	__ClearPageReserved(page);
	320	set_page_refs(page, 0);
	321	__free_page(page);
	322	}
	323	}
	324	} else {
	325	i+=BITS_PER_LONG;
1da177e4	326	}
d41dee36	327	pfn += BITS_PER_LONG;
1da177e4 LT	328	}
	329	total += count;
	330
	331	/*
	332	* Now free the allocator bitmap itself, it's not
	333	* needed anymore:
	334	*/
	335	page = virt_to_page(bdata->node_bootmem_map);
	336	count = 0;
	337	for (i = 0; i < ((bdata->node_low_pfn-(bdata->node_boot_start >> PAGE_SHIFT))/8 + PAGE_SIZE-1)/PAGE_SIZE; i++,page++) {
	338	count++;
	339	__ClearPageReserved(page);
	340	set_page_count(page, 1);
	341	__free_page(page);
	342	}
	343	total += count;
	344	bdata->node_bootmem_map = NULL;
	345
	346	return total;
	347	}
	348
	349	unsigned long __init init_bootmem_node (pg_data_t *pgdat, unsigned long freepfn, unsigned long startpfn, unsigned long endpfn)
	350	{
	351	return(init_bootmem_core(pgdat, freepfn, startpfn, endpfn));
	352	}
	353
	354	void __init reserve_bootmem_node (pg_data_t *pgdat, unsigned long physaddr, unsigned long size)
	355	{
	356	reserve_bootmem_core(pgdat->bdata, physaddr, size);
	357	}
	358
	359	void __init free_bootmem_node (pg_data_t *pgdat, unsigned long physaddr, unsigned long size)
	360	{
	361	free_bootmem_core(pgdat->bdata, physaddr, size);
	362	}
	363
	364	unsigned long __init free_all_bootmem_node (pg_data_t *pgdat)
	365	{
	366	return(free_all_bootmem_core(pgdat));
	367	}
	368
	369	unsigned long __init init_bootmem (unsigned long start, unsigned long pages)
	370	{
	371	max_low_pfn = pages;
	372	min_low_pfn = start;
	373	return(init_bootmem_core(NODE_DATA(0), start, 0, pages));
	374	}
	375
	376	#ifndef CONFIG_HAVE_ARCH_BOOTMEM_NODE
	377	void __init reserve_bootmem (unsigned long addr, unsigned long size)
	378	{
	379	reserve_bootmem_core(NODE_DATA(0)->bdata, addr, size);
	380	}
	381	#endif /* !CONFIG_HAVE_ARCH_BOOTMEM_NODE */
	382
	383	void __init free_bootmem (unsigned long addr, unsigned long size)
	384	{
	385	free_bootmem_core(NODE_DATA(0)->bdata, addr, size);
	386	}
	387
	388	unsigned long __init free_all_bootmem (void)
	389	{
	390	return(free_all_bootmem_core(NODE_DATA(0)));
	391	}
392
281dd25c YG	393	void * __init __alloc_bootmem_limit (unsigned long size, unsigned long align, unsigned long goal,
281dd25c YG	394	unsigned long limit)
1da177e4 LT	395	{
	396	pg_data_t *pgdat = pgdat_list;
	397	void *ptr;
	398
	399	for_each_pgdat(pgdat)
	400	if ((ptr = __alloc_bootmem_core(pgdat->bdata, size,
281dd25c	401	align, goal, limit)))
1da177e4 LT	402	return(ptr);
	403
	404	/*
	405	* Whoops, we cannot satisfy the allocation request.
	406	*/
	407	printk(KERN_ALERT "bootmem alloc of %lu bytes failed!\n", size);
	408	panic("Out of memory");
	409	return NULL;
	410	}
	411
281dd25c YG	412
	413	void * __init __alloc_bootmem_node_limit (pg_data_t *pgdat, unsigned long size, unsigned long align,
	414	unsigned long goal, unsigned long limit)
1da177e4 LT	415	{
	416	void *ptr;
	417
281dd25c	418	ptr = __alloc_bootmem_core(pgdat->bdata, size, align, goal, limit);
1da177e4 LT	419	if (ptr)
	420	return (ptr);
	421
281dd25c	422	return __alloc_bootmem_limit(size, align, goal, limit);
1da177e4 LT	423	}
1da177e4 LT	424