[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/init.h>
#include <linux/pfn.h>
#include <linux/bootmem.h>
#include <linux/module.h>

#include <asm/bug.h>
#include <asm/io.h>
#include <asm/processor.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_UNUSED_SYMBOL(max_pfn);  /*  June 2006  */

static LIST_HEAD(bdata_list);
#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;
}

/*
 * link bdata in order
 */
static void __init link_bootmem(bootmem_data_t *bdata)
{
	bootmem_data_t *ent;

	if (list_empty(&bdata_list)) {
		list_add(&bdata->list, &bdata_list);
		return;
	}
	/* insert in order */
	list_for_each_entry(ent, &bdata_list, list) {
		if (bdata->node_boot_start < ent->node_boot_start) {
			list_add_tail(&bdata->list, &ent->list);
			return;
		}
	}
	list_add_tail(&bdata->list, &bdata_list);
}

/*
 * Given an initialised bdata, it returns the size of the boot bitmap
 */
static unsigned long __init get_mapsize(bootmem_data_t *bdata)
{
	unsigned long mapsize;
	unsigned long start = PFN_DOWN(bdata->node_boot_start);
	unsigned long end = bdata->node_low_pfn;

	mapsize = ((end - start) + 7) / 8;
	return ALIGN(mapsize, sizeof(long));
}

/*
 * 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;

	bdata->node_bootmem_map = phys_to_virt(PFN_PHYS(mapstart));
	bdata->node_boot_start = PFN_PHYS(start);
	bdata->node_low_pfn = end;
	link_bootmem(bdata);

	/*
	 * Initially all pages are reserved - setup_arch() has to
	 * register free RAM areas explicitly.
	 */
	mapsize = get_mapsize(bdata);
	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 sidx, eidx;
	unsigned long i;

	/*
	 * round up, partially reserved pages are considered
	 * fully reserved.
	 */
	BUG_ON(!size);
	BUG_ON(PFN_DOWN(addr) >= bdata->node_low_pfn);
	BUG_ON(PFN_UP(addr + size) > bdata->node_low_pfn);

	sidx = PFN_DOWN(addr - bdata->node_boot_start);
	eidx = PFN_UP(addr + size - bdata->node_boot_start);

	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 sidx, eidx;
	unsigned long i;

	/*
	 * round down end of usable mem, partially free pages are
	 * considered reserved.
	 */
	BUG_ON(!size);
	BUG_ON(PFN_DOWN(addr + size) > bdata->node_low_pfn);

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

	/*
	 * Round up the beginning of the address.
	 */
	sidx = PFN_UP(addr) - PFN_DOWN(bdata->node_boot_start);
	eidx = PFN_DOWN(addr + size - bdata->node_boot_start);

	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.
 */
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;
	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;

	end_pfn = bdata->node_low_pfn;
	limit = PFN_DOWN(limit);
	if (limit && end_pfn > limit)
		end_pfn = limit;

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

	/*
	 * We try to allocate bootmem pages above 'goal'
	 * first, then we try to allocate lower pages.
	 */
	if (goal && goal >= bdata->node_boot_start && PFN_DOWN(goal) < 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 = PFN_DOWN(ALIGN(preferred, align)) + 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 (i >= eidx)
			break;
		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 = PFN_PHYS(start);
	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 = PFN_DOWN(bdata->node_boot_start);
	idx = bdata->node_low_pfn - pfn;
	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 order;

			page = pfn_to_page(pfn);
			count += BITS_PER_LONG;
			order = ffs(BITS_PER_LONG) - 1;
			__free_pages_bootmem(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++;
					__free_pages_bootmem(page, 0);
				}
			}
		} 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;
	idx = (get_mapsize(bdata) + PAGE_SIZE-1) >> PAGE_SHIFT;
	for (i = 0; i < idx; i++, page++) {
		__free_pages_bootmem(page, 0);
		count++;
	}
	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_nopanic(unsigned long size, unsigned long align,
				      unsigned long goal)
{
	bootmem_data_t *bdata;
	void *ptr;

	list_for_each_entry(bdata, &bdata_list, list) {
		ptr = __alloc_bootmem_core(bdata, size, align, goal, 0);
		if (ptr)
			return ptr;
	}
	return NULL;
}

void * __init __alloc_bootmem(unsigned long size, unsigned long align,
			      unsigned long goal)
{
	void *mem = __alloc_bootmem_nopanic(size,align,goal);

	if (mem)
		return mem;
	/*
	 * 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(pg_data_t *pgdat, unsigned long size,
				   unsigned long align, unsigned long goal)
{
	void *ptr;

	ptr = __alloc_bootmem_core(pgdat->bdata, size, align, goal, 0);
	if (ptr)
		return ptr;

	return __alloc_bootmem(size, align, goal);
}

#ifndef ARCH_LOW_ADDRESS_LIMIT
#define ARCH_LOW_ADDRESS_LIMIT	0xffffffffUL
#endif

void * __init __alloc_bootmem_low(unsigned long size, unsigned long align,
				  unsigned long goal)
{
	bootmem_data_t *bdata;
	void *ptr;

	list_for_each_entry(bdata, &bdata_list, list) {
		ptr = __alloc_bootmem_core(bdata, size, align, goal,
						ARCH_LOW_ADDRESS_LIMIT);
		if (ptr)
			return ptr;
	}

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

void * __init __alloc_bootmem_low_node(pg_data_t *pgdat, unsigned long size,
				       unsigned long align, unsigned long goal)
{
	return __alloc_bootmem_core(pgdat->bdata, size, align, goal,
				    ARCH_LOW_ADDRESS_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	*/
1da177e4	11	#include <linux/init.h>
bbc7b92e	12	#include <linux/pfn.h>
1da177e4	13	#include <linux/bootmem.h>
1da177e4	14	#include <linux/module.h>
e786e86a FBH	15
e786e86a FBH	16	#include <asm/bug.h>
1da177e4	17	#include <asm/io.h>
dfd54cbc	18	#include <asm/processor.h>
e786e86a	19
1da177e4 LT	20	#include "internal.h"
	21
	22	/*
	23	* Access to this subsystem has to be serialized externally. (this is
	24	* true for the boot process anyway)
	25	*/
	26	unsigned long max_low_pfn;
	27	unsigned long min_low_pfn;
	28	unsigned long max_pfn;
	29
6d46cc6b	30	EXPORT_UNUSED_SYMBOL(max_pfn); /* June 2006 */
1da177e4	31
679bc9fb	32	static LIST_HEAD(bdata_list);
92aa63a5 VG	33	#ifdef CONFIG_CRASH_DUMP
	34	/*
	35	* If we have booted due to a crash, max_pfn will be a very low value. We need
	36	* to know the amount of memory that the previous kernel used.
	37	*/
	38	unsigned long saved_max_pfn;
	39	#endif
	40
1da177e4	41	/* return the number of _pages_ that will be allocated for the boot bitmap */
f71bf0ca	42	unsigned long __init bootmem_bootmap_pages(unsigned long pages)
1da177e4 LT	43	{
	44	unsigned long mapsize;
	45
	46	mapsize = (pages+7)/8;
	47	mapsize = (mapsize + ~PAGE_MASK) & PAGE_MASK;
	48	mapsize >>= PAGE_SHIFT;
	49
	50	return mapsize;
	51	}
f71bf0ca	52
679bc9fb KH	53	/*
	54	* link bdata in order
	55	*/
69d49e68	56	static void __init link_bootmem(bootmem_data_t *bdata)
679bc9fb KH	57	{
679bc9fb KH	58	bootmem_data_t *ent;
f71bf0ca	59
679bc9fb KH	60	if (list_empty(&bdata_list)) {
	61	list_add(&bdata->list, &bdata_list);
	62	return;
	63	}
	64	/* insert in order */
	65	list_for_each_entry(ent, &bdata_list, list) {
	66	if (bdata->node_boot_start < ent->node_boot_start) {
	67	list_add_tail(&bdata->list, &ent->list);
	68	return;
	69	}
	70	}
	71	list_add_tail(&bdata->list, &bdata_list);
679bc9fb KH	72	}
679bc9fb KH	73
bbc7b92e FBH	74	/*
	75	* Given an initialised bdata, it returns the size of the boot bitmap
	76	*/
	77	static unsigned long __init get_mapsize(bootmem_data_t *bdata)
	78	{
	79	unsigned long mapsize;
	80	unsigned long start = PFN_DOWN(bdata->node_boot_start);
	81	unsigned long end = bdata->node_low_pfn;
	82
	83	mapsize = ((end - start) + 7) / 8;
	84	return ALIGN(mapsize, sizeof(long));
	85	}
1da177e4 LT	86
	87	/*
	88	* Called once to set up the allocator itself.
	89	*/
f71bf0ca	90	static unsigned long __init init_bootmem_core(pg_data_t *pgdat,
1da177e4 LT	91	unsigned long mapstart, unsigned long start, unsigned long end)
	92	{
	93	bootmem_data_t *bdata = pgdat->bdata;
bbc7b92e	94	unsigned long mapsize;
1da177e4	95
bbc7b92e FBH	96	bdata->node_bootmem_map = phys_to_virt(PFN_PHYS(mapstart));
bbc7b92e FBH	97	bdata->node_boot_start = PFN_PHYS(start);
1da177e4	98	bdata->node_low_pfn = end;
679bc9fb	99	link_bootmem(bdata);
1da177e4 LT	100
	101	/*
	102	* Initially all pages are reserved - setup_arch() has to
	103	* register free RAM areas explicitly.
	104	*/
bbc7b92e	105	mapsize = get_mapsize(bdata);
1da177e4 LT	106	memset(bdata->node_bootmem_map, 0xff, mapsize);
	107
	108	return mapsize;
	109	}
	110
	111	/*
	112	* Marks a particular physical memory range as unallocatable. Usable RAM
	113	* might be used for boot-time allocations - or it might get added
	114	* to the free page pool later on.
	115	*/
bb0923a6 FBH	116	static void __init reserve_bootmem_core(bootmem_data_t *bdata, unsigned long addr,
bb0923a6 FBH	117	unsigned long size)
1da177e4	118	{
bbc7b92e	119	unsigned long sidx, eidx;
1da177e4	120	unsigned long i;
bbc7b92e	121
1da177e4 LT	122	/*
	123	* round up, partially reserved pages are considered
	124	* fully reserved.
	125	*/
1da177e4	126	BUG_ON(!size);
bbc7b92e FBH	127	BUG_ON(PFN_DOWN(addr) >= bdata->node_low_pfn);
	128	BUG_ON(PFN_UP(addr + size) > bdata->node_low_pfn);
	129
	130	sidx = PFN_DOWN(addr - bdata->node_boot_start);
	131	eidx = PFN_UP(addr + size - bdata->node_boot_start);
1da177e4 LT	132
	133	for (i = sidx; i < eidx; i++)
	134	if (test_and_set_bit(i, bdata->node_bootmem_map)) {
	135	#ifdef CONFIG_DEBUG_BOOTMEM
	136	printk("hm, page %08lx reserved twice.\n", i*PAGE_SIZE);
	137	#endif
	138	}
	139	}
	140
bb0923a6 FBH	141	static void __init free_bootmem_core(bootmem_data_t *bdata, unsigned long addr,
bb0923a6 FBH	142	unsigned long size)
1da177e4	143	{
bbc7b92e	144	unsigned long sidx, eidx;
1da177e4	145	unsigned long i;
bbc7b92e	146
1da177e4 LT	147	/*
	148	* round down end of usable mem, partially free pages are
	149	* considered reserved.
	150	*/
1da177e4	151	BUG_ON(!size);
bbc7b92e	152	BUG_ON(PFN_DOWN(addr + size) > bdata->node_low_pfn);
1da177e4 LT	153
	154	if (addr < bdata->last_success)
	155	bdata->last_success = addr;
	156
	157	/*
	158	* Round up the beginning of the address.
	159	*/
bbc7b92e FBH	160	sidx = PFN_UP(addr) - PFN_DOWN(bdata->node_boot_start);
bbc7b92e FBH	161	eidx = PFN_DOWN(addr + size - bdata->node_boot_start);
1da177e4 LT	162
	163	for (i = sidx; i < eidx; i++) {
	164	if (unlikely(!test_and_clear_bit(i, bdata->node_bootmem_map)))
	165	BUG();
	166	}
	167	}
	168
	169	/*
	170	* We 'merge' subsequent allocations to save space. We might 'lose'
	171	* some fraction of a page if allocations cannot be satisfied due to
	172	* size constraints on boxes where there is physical RAM space
	173	* fragmentation - in these cases (mostly large memory boxes) this
	174	* is not a problem.
	175	*
	176	* On low memory boxes we get it right in 100% of the cases.
	177	*
	178	* alignment has to be a power of 2 value.
	179	*
	180	* NOTE: This function is _not_ reentrant.
	181	*/
267b4801	182	void * __init
1da177e4	183	__alloc_bootmem_core(struct bootmem_data *bdata, unsigned long size,
281dd25c	184	unsigned long align, unsigned long goal, unsigned long limit)
1da177e4 LT	185	{
1da177e4 LT	186	unsigned long offset, remaining_size, areasize, preferred;
bbc7b92e	187	unsigned long i, start = 0, incr, eidx, end_pfn;
1da177e4 LT	188	void *ret;
1da177e4 LT	189
f71bf0ca	190	if (!size) {
1da177e4 LT	191	printk("__alloc_bootmem_core(): zero-sized request\n");
	192	BUG();
	193	}
	194	BUG_ON(align & (align-1));
	195
281dd25c YG	196	if (limit && bdata->node_boot_start >= limit)
	197	return NULL;
	198
bbc7b92e FBH	199	end_pfn = bdata->node_low_pfn;
bbc7b92e FBH	200	limit = PFN_DOWN(limit);
281dd25c YG	201	if (limit && end_pfn > limit)
	202	end_pfn = limit;
	203
bbc7b92e	204	eidx = end_pfn - PFN_DOWN(bdata->node_boot_start);
1da177e4	205	offset = 0;
bbc7b92e FBH	206	if (align && (bdata->node_boot_start & (align - 1UL)) != 0)
	207	offset = align - (bdata->node_boot_start & (align - 1UL));
	208	offset = PFN_DOWN(offset);
1da177e4 LT	209
	210	/*
	211	* We try to allocate bootmem pages above 'goal'
	212	* first, then we try to allocate lower pages.
	213	*/
bbc7b92e	214	if (goal && goal >= bdata->node_boot_start && PFN_DOWN(goal) < end_pfn) {
1da177e4 LT	215	preferred = goal - bdata->node_boot_start;
	216
	217	if (bdata->last_success >= preferred)
281dd25c YG	218	if (!limit \|\| (limit && limit > bdata->last_success))
281dd25c YG	219	preferred = bdata->last_success;
1da177e4 LT	220	} else
	221	preferred = 0;
	222
bbc7b92e FBH	223	preferred = PFN_DOWN(ALIGN(preferred, align)) + offset;
bbc7b92e FBH	224	areasize = (size + PAGE_SIZE-1) / PAGE_SIZE;
1da177e4 LT	225	incr = align >> PAGE_SHIFT ? : 1;
	226
	227	restart_scan:
	228	for (i = preferred; i < eidx; i += incr) {
	229	unsigned long j;
	230	i = find_next_zero_bit(bdata->node_bootmem_map, eidx, i);
	231	i = ALIGN(i, incr);
66d43e98 HM	232	if (i >= eidx)
66d43e98 HM	233	break;
1da177e4 LT	234	if (test_bit(i, bdata->node_bootmem_map))
	235	continue;
	236	for (j = i + 1; j < i + areasize; ++j) {
	237	if (j >= eidx)
	238	goto fail_block;
f71bf0ca	239	if (test_bit(j, bdata->node_bootmem_map))
1da177e4 LT	240	goto fail_block;
	241	}
	242	start = i;
	243	goto found;
	244	fail_block:
	245	i = ALIGN(j, incr);
	246	}
	247
	248	if (preferred > offset) {
	249	preferred = offset;
	250	goto restart_scan;
	251	}
	252	return NULL;
	253
	254	found:
bbc7b92e	255	bdata->last_success = PFN_PHYS(start);
1da177e4 LT	256	BUG_ON(start >= eidx);
	257
	258	/*
	259	* Is the next page of the previous allocation-end the start
	260	* of this allocation's buffer? If yes then we can 'merge'
	261	* the previous partial page with this allocation.
	262	*/
	263	if (align < PAGE_SIZE &&
	264	bdata->last_offset && bdata->last_pos+1 == start) {
8c0e33c1	265	offset = ALIGN(bdata->last_offset, align);
1da177e4	266	BUG_ON(offset > PAGE_SIZE);
f71bf0ca	267	remaining_size = PAGE_SIZE - offset;
1da177e4 LT	268	if (size < remaining_size) {
	269	areasize = 0;
	270	/* last_pos unchanged */
f71bf0ca FBH	271	bdata->last_offset = offset + size;
	272	ret = phys_to_virt(bdata->last_pos * PAGE_SIZE +
	273	offset +
	274	bdata->node_boot_start);
1da177e4 LT	275	} else {
1da177e4 LT	276	remaining_size = size - remaining_size;
f71bf0ca FBH	277	areasize = (remaining_size + PAGE_SIZE-1) / PAGE_SIZE;
	278	ret = phys_to_virt(bdata->last_pos * PAGE_SIZE +
	279	offset +
	280	bdata->node_boot_start);
	281	bdata->last_pos = start + areasize - 1;
1da177e4 LT	282	bdata->last_offset = remaining_size;
	283	}
	284	bdata->last_offset &= ~PAGE_MASK;
	285	} else {
	286	bdata->last_pos = start + areasize - 1;
	287	bdata->last_offset = size & ~PAGE_MASK;
	288	ret = phys_to_virt(start * PAGE_SIZE + bdata->node_boot_start);
	289	}
	290
	291	/*
	292	* Reserve the area now:
	293	*/
f71bf0ca	294	for (i = start; i < start + areasize; i++)
1da177e4 LT	295	if (unlikely(test_and_set_bit(i, bdata->node_bootmem_map)))
	296	BUG();
	297	memset(ret, 0, size);
	298	return ret;
	299	}
	300
	301	static unsigned long __init free_all_bootmem_core(pg_data_t *pgdat)
	302	{
	303	struct page *page;
d41dee36	304	unsigned long pfn;
1da177e4 LT	305	bootmem_data_t *bdata = pgdat->bdata;
	306	unsigned long i, count, total = 0;
	307	unsigned long idx;
	308	unsigned long *map;
	309	int gofast = 0;
	310
	311	BUG_ON(!bdata->node_bootmem_map);
	312
	313	count = 0;
	314	/* first extant page of the node */
bbc7b92e FBH	315	pfn = PFN_DOWN(bdata->node_boot_start);
bbc7b92e FBH	316	idx = bdata->node_low_pfn - pfn;
1da177e4 LT	317	map = bdata->node_bootmem_map;
	318	/* Check physaddr is O(LOG2(BITS_PER_LONG)) page aligned */
	319	if (bdata->node_boot_start == 0 \|\|
	320	ffs(bdata->node_boot_start) - PAGE_SHIFT > ffs(BITS_PER_LONG))
	321	gofast = 1;
	322	for (i = 0; i < idx; ) {
	323	unsigned long v = ~map[i / BITS_PER_LONG];
d41dee36	324
1da177e4	325	if (gofast && v == ~0UL) {
a226f6c8	326	int order;
1da177e4	327
d41dee36	328	page = pfn_to_page(pfn);
1da177e4	329	count += BITS_PER_LONG;
1da177e4	330	order = ffs(BITS_PER_LONG) - 1;
a226f6c8	331	__free_pages_bootmem(page, order);
1da177e4 LT	332	i += BITS_PER_LONG;
	333	page += BITS_PER_LONG;
	334	} else if (v) {
	335	unsigned long m;
d41dee36 AW	336
d41dee36 AW	337	page = pfn_to_page(pfn);
1da177e4 LT	338	for (m = 1; m && i < idx; m<<=1, page++, i++) {
	339	if (v & m) {
	340	count++;
a226f6c8	341	__free_pages_bootmem(page, 0);
1da177e4 LT	342	}
	343	}
	344	} else {
f71bf0ca	345	i += BITS_PER_LONG;
1da177e4	346	}
d41dee36	347	pfn += BITS_PER_LONG;
1da177e4 LT	348	}
	349	total += count;
	350
	351	/*
	352	* Now free the allocator bitmap itself, it's not
	353	* needed anymore:
	354	*/
	355	page = virt_to_page(bdata->node_bootmem_map);
	356	count = 0;
bbc7b92e FBH	357	idx = (get_mapsize(bdata) + PAGE_SIZE-1) >> PAGE_SHIFT;
bbc7b92e FBH	358	for (i = 0; i < idx; i++, page++) {
a226f6c8	359	__free_pages_bootmem(page, 0);
bbc7b92e	360	count++;
1da177e4 LT	361	}
	362	total += count;
	363	bdata->node_bootmem_map = NULL;
	364
	365	return total;
	366	}
	367
f71bf0ca	368	unsigned long __init init_bootmem_node(pg_data_t *pgdat, unsigned long freepfn,
bb0923a6	369	unsigned long startpfn, unsigned long endpfn)
1da177e4	370	{
f71bf0ca	371	return init_bootmem_core(pgdat, freepfn, startpfn, endpfn);
1da177e4 LT	372	}
1da177e4 LT	373
f71bf0ca FBH	374	void __init reserve_bootmem_node(pg_data_t *pgdat, unsigned long physaddr,
f71bf0ca FBH	375	unsigned long size)
1da177e4 LT	376	{
	377	reserve_bootmem_core(pgdat->bdata, physaddr, size);
	378	}
	379
f71bf0ca FBH	380	void __init free_bootmem_node(pg_data_t *pgdat, unsigned long physaddr,
f71bf0ca FBH	381	unsigned long size)
1da177e4 LT	382	{
	383	free_bootmem_core(pgdat->bdata, physaddr, size);
	384	}
	385
f71bf0ca	386	unsigned long __init free_all_bootmem_node(pg_data_t *pgdat)
1da177e4	387	{
f71bf0ca	388	return free_all_bootmem_core(pgdat);
1da177e4 LT	389	}
1da177e4 LT	390
f71bf0ca	391	unsigned long __init init_bootmem(unsigned long start, unsigned long pages)
1da177e4 LT	392	{
	393	max_low_pfn = pages;
	394	min_low_pfn = start;
f71bf0ca	395	return init_bootmem_core(NODE_DATA(0), start, 0, pages);
1da177e4 LT	396	}
	397
	398	#ifndef CONFIG_HAVE_ARCH_BOOTMEM_NODE
f71bf0ca	399	void __init reserve_bootmem(unsigned long addr, unsigned long size)
1da177e4 LT	400	{
	401	reserve_bootmem_core(NODE_DATA(0)->bdata, addr, size);
	402	}
	403	#endif /* !CONFIG_HAVE_ARCH_BOOTMEM_NODE */
	404
f71bf0ca	405	void __init free_bootmem(unsigned long addr, unsigned long size)
1da177e4 LT	406	{
	407	free_bootmem_core(NODE_DATA(0)->bdata, addr, size);
	408	}
	409
f71bf0ca	410	unsigned long __init free_all_bootmem(void)
1da177e4	411	{
f71bf0ca	412	return free_all_bootmem_core(NODE_DATA(0));
1da177e4 LT	413	}
1da177e4 LT	414
bb0923a6 FBH	415	void * __init __alloc_bootmem_nopanic(unsigned long size, unsigned long align,
bb0923a6 FBH	416	unsigned long goal)
1da177e4	417	{
679bc9fb	418	bootmem_data_t *bdata;
1da177e4 LT	419	void *ptr;
1da177e4 LT	420
f71bf0ca FBH	421	list_for_each_entry(bdata, &bdata_list, list) {
	422	ptr = __alloc_bootmem_core(bdata, size, align, goal, 0);
	423	if (ptr)
	424	return ptr;
	425	}
a8062231 AK	426	return NULL;
a8062231 AK	427	}
1da177e4	428
bb0923a6 FBH	429	void * __init __alloc_bootmem(unsigned long size, unsigned long align,
bb0923a6 FBH	430	unsigned long goal)
a8062231 AK	431	{
a8062231 AK	432	void *mem = __alloc_bootmem_nopanic(size,align,goal);
f71bf0ca	433
a8062231 AK	434	if (mem)
a8062231 AK	435	return mem;
1da177e4 LT	436	/*
	437	* Whoops, we cannot satisfy the allocation request.
	438	*/
	439	printk(KERN_ALERT "bootmem alloc of %lu bytes failed!\n", size);
	440	panic("Out of memory");
	441	return NULL;
	442	}
	443
281dd25c	444
bb0923a6 FBH	445	void * __init __alloc_bootmem_node(pg_data_t *pgdat, unsigned long size,
bb0923a6 FBH	446	unsigned long align, unsigned long goal)
1da177e4 LT	447	{
	448	void *ptr;
	449
008857c1	450	ptr = __alloc_bootmem_core(pgdat->bdata, size, align, goal, 0);
1da177e4	451	if (ptr)
f71bf0ca	452	return ptr;
1da177e4	453
008857c1	454	return __alloc_bootmem(size, align, goal);
1da177e4 LT	455	}
1da177e4 LT	456
dfd54cbc HC	457	#ifndef ARCH_LOW_ADDRESS_LIMIT
	458	#define ARCH_LOW_ADDRESS_LIMIT 0xffffffffUL
	459	#endif
008857c1	460
bb0923a6 FBH	461	void * __init __alloc_bootmem_low(unsigned long size, unsigned long align,
bb0923a6 FBH	462	unsigned long goal)
008857c1	463	{
679bc9fb	464	bootmem_data_t *bdata;
008857c1 RT	465	void *ptr;
008857c1 RT	466
f71bf0ca	467	list_for_each_entry(bdata, &bdata_list, list) {
dfd54cbc HC	468	ptr = __alloc_bootmem_core(bdata, size, align, goal,
dfd54cbc HC	469	ARCH_LOW_ADDRESS_LIMIT);
f71bf0ca FBH	470	if (ptr)
	471	return ptr;
	472	}
008857c1 RT	473
	474	/*
	475	* Whoops, we cannot satisfy the allocation request.
	476	*/
	477	printk(KERN_ALERT "low bootmem alloc of %lu bytes failed!\n", size);
	478	panic("Out of low memory");
	479	return NULL;
	480	}
	481
	482	void * __init __alloc_bootmem_low_node(pg_data_t *pgdat, unsigned long size,
	483	unsigned long align, unsigned long goal)
	484	{
dfd54cbc HC	485	return __alloc_bootmem_core(pgdat->bdata, size, align, goal,
dfd54cbc HC	486	ARCH_LOW_ADDRESS_LIMIT);
008857c1	487	}