[linux-2.6-block.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/bootmem.h>
#include <linux/module.h>

#include <asm/bug.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_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);
	return;
}


/*
 * 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;

	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;
	link_bootmem(bdata);

	/*
	 * 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.
 */
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 (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 = 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 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;
	for (i = 0; i < ((bdata->node_low_pfn-(bdata->node_boot_start >> PAGE_SHIFT))/8 + PAGE_SIZE-1)/PAGE_SIZE; i++,page++) {
		count++;
		__free_pages_bootmem(page, 0);
	}
	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)
		if ((ptr = __alloc_bootmem_core(bdata, size, align, goal, 0)))
			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);
}

#define LOW32LIMIT 0xffffffff

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)
		if ((ptr = __alloc_bootmem_core(bdata, size,
						 align, goal, LOW32LIMIT)))
			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, LOW32LIMIT);
}
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 LT	11	#include <linux/init.h>
1da177e4 LT	12	#include <linux/bootmem.h>
1da177e4	13	#include <linux/module.h>
e786e86a FBH	14
e786e86a FBH	15	#include <asm/bug.h>
1da177e4	16	#include <asm/io.h>
e786e86a	17
1da177e4 LT	18	#include "internal.h"
	19
	20	/*
	21	* Access to this subsystem has to be serialized externally. (this is
	22	* true for the boot process anyway)
	23	*/
	24	unsigned long max_low_pfn;
	25	unsigned long min_low_pfn;
	26	unsigned long max_pfn;
	27
6d46cc6b	28	EXPORT_UNUSED_SYMBOL(max_pfn); /* June 2006 */
1da177e4	29
679bc9fb	30	static LIST_HEAD(bdata_list);
92aa63a5 VG	31	#ifdef CONFIG_CRASH_DUMP
	32	/*
	33	* If we have booted due to a crash, max_pfn will be a very low value. We need
	34	* to know the amount of memory that the previous kernel used.
	35	*/
	36	unsigned long saved_max_pfn;
	37	#endif
	38
1da177e4 LT	39	/* return the number of _pages_ that will be allocated for the boot bitmap */
	40	unsigned long __init bootmem_bootmap_pages (unsigned long pages)
	41	{
	42	unsigned long mapsize;
	43
	44	mapsize = (pages+7)/8;
	45	mapsize = (mapsize + ~PAGE_MASK) & PAGE_MASK;
	46	mapsize >>= PAGE_SHIFT;
	47
	48	return mapsize;
	49	}
679bc9fb KH	50	/*
	51	* link bdata in order
	52	*/
69d49e68	53	static void __init link_bootmem(bootmem_data_t *bdata)
679bc9fb KH	54	{
	55	bootmem_data_t *ent;
	56	if (list_empty(&bdata_list)) {
	57	list_add(&bdata->list, &bdata_list);
	58	return;
	59	}
	60	/* insert in order */
	61	list_for_each_entry(ent, &bdata_list, list) {
	62	if (bdata->node_boot_start < ent->node_boot_start) {
	63	list_add_tail(&bdata->list, &ent->list);
	64	return;
	65	}
	66	}
	67	list_add_tail(&bdata->list, &bdata_list);
	68	return;
	69	}
	70
1da177e4 LT	71
	72	/*
	73	* Called once to set up the allocator itself.
	74	*/
	75	static unsigned long __init init_bootmem_core (pg_data_t *pgdat,
	76	unsigned long mapstart, unsigned long start, unsigned long end)
	77	{
	78	bootmem_data_t *bdata = pgdat->bdata;
	79	unsigned long mapsize = ((end - start)+7)/8;
	80
8c0e33c1	81	mapsize = ALIGN(mapsize, sizeof(long));
1da177e4 LT	82	bdata->node_bootmem_map = phys_to_virt(mapstart << PAGE_SHIFT);
	83	bdata->node_boot_start = (start << PAGE_SHIFT);
	84	bdata->node_low_pfn = end;
679bc9fb	85	link_bootmem(bdata);
1da177e4 LT	86
	87	/*
	88	* Initially all pages are reserved - setup_arch() has to
	89	* register free RAM areas explicitly.
	90	*/
	91	memset(bdata->node_bootmem_map, 0xff, mapsize);
	92
	93	return mapsize;
	94	}
	95
	96	/*
	97	* Marks a particular physical memory range as unallocatable. Usable RAM
	98	* might be used for boot-time allocations - or it might get added
	99	* to the free page pool later on.
	100	*/
bb0923a6 FBH	101	static void __init reserve_bootmem_core(bootmem_data_t *bdata, unsigned long addr,
bb0923a6 FBH	102	unsigned long size)
1da177e4 LT	103	{
	104	unsigned long i;
	105	/*
	106	* round up, partially reserved pages are considered
	107	* fully reserved.
	108	*/
	109	unsigned long sidx = (addr - bdata->node_boot_start)/PAGE_SIZE;
	110	unsigned long eidx = (addr + size - bdata->node_boot_start +
	111	PAGE_SIZE-1)/PAGE_SIZE;
	112	unsigned long end = (addr + size + PAGE_SIZE-1)/PAGE_SIZE;
	113
	114	BUG_ON(!size);
	115	BUG_ON(sidx >= eidx);
	116	BUG_ON((addr >> PAGE_SHIFT) >= bdata->node_low_pfn);
	117	BUG_ON(end > bdata->node_low_pfn);
	118
	119	for (i = sidx; i < eidx; i++)
	120	if (test_and_set_bit(i, bdata->node_bootmem_map)) {
	121	#ifdef CONFIG_DEBUG_BOOTMEM
	122	printk("hm, page %08lx reserved twice.\n", i*PAGE_SIZE);
	123	#endif
	124	}
	125	}
	126
bb0923a6 FBH	127	static void __init free_bootmem_core(bootmem_data_t *bdata, unsigned long addr,
bb0923a6 FBH	128	unsigned long size)
1da177e4 LT	129	{
	130	unsigned long i;
	131	unsigned long start;
	132	/*
	133	* round down end of usable mem, partially free pages are
	134	* considered reserved.
	135	*/
	136	unsigned long sidx;
	137	unsigned long eidx = (addr + size - bdata->node_boot_start)/PAGE_SIZE;
	138	unsigned long end = (addr + size)/PAGE_SIZE;
	139
	140	BUG_ON(!size);
	141	BUG_ON(end > bdata->node_low_pfn);
	142
	143	if (addr < bdata->last_success)
	144	bdata->last_success = addr;
	145
	146	/*
	147	* Round up the beginning of the address.
	148	*/
	149	start = (addr + PAGE_SIZE-1) / PAGE_SIZE;
	150	sidx = start - (bdata->node_boot_start/PAGE_SIZE);
	151
	152	for (i = sidx; i < eidx; i++) {
	153	if (unlikely(!test_and_clear_bit(i, bdata->node_bootmem_map)))
	154	BUG();
	155	}
	156	}
	157
	158	/*
	159	* We 'merge' subsequent allocations to save space. We might 'lose'
	160	* some fraction of a page if allocations cannot be satisfied due to
	161	* size constraints on boxes where there is physical RAM space
	162	* fragmentation - in these cases (mostly large memory boxes) this
	163	* is not a problem.
	164	*
	165	* On low memory boxes we get it right in 100% of the cases.
	166	*
	167	* alignment has to be a power of 2 value.
	168	*
	169	* NOTE: This function is _not_ reentrant.
	170	*/
267b4801	171	void * __init
1da177e4	172	__alloc_bootmem_core(struct bootmem_data *bdata, unsigned long size,
281dd25c	173	unsigned long align, unsigned long goal, unsigned long limit)
1da177e4 LT	174	{
1da177e4 LT	175	unsigned long offset, remaining_size, areasize, preferred;
281dd25c	176	unsigned long i, start = 0, incr, eidx, end_pfn = bdata->node_low_pfn;
1da177e4 LT	177	void *ret;
	178
	179	if(!size) {
	180	printk("__alloc_bootmem_core(): zero-sized request\n");
	181	BUG();
	182	}
	183	BUG_ON(align & (align-1));
	184
281dd25c YG	185	if (limit && bdata->node_boot_start >= limit)
	186	return NULL;
	187
	188	limit >>=PAGE_SHIFT;
	189	if (limit && end_pfn > limit)
	190	end_pfn = limit;
	191
	192	eidx = end_pfn - (bdata->node_boot_start >> PAGE_SHIFT);
1da177e4 LT	193	offset = 0;
	194	if (align &&
	195	(bdata->node_boot_start & (align - 1UL)) != 0)
	196	offset = (align - (bdata->node_boot_start & (align - 1UL)));
	197	offset >>= PAGE_SHIFT;
	198
	199	/*
	200	* We try to allocate bootmem pages above 'goal'
	201	* first, then we try to allocate lower pages.
	202	*/
	203	if (goal && (goal >= bdata->node_boot_start) &&
281dd25c	204	((goal >> PAGE_SHIFT) < end_pfn)) {
1da177e4 LT	205	preferred = goal - bdata->node_boot_start;
	206
	207	if (bdata->last_success >= preferred)
281dd25c YG	208	if (!limit \|\| (limit && limit > bdata->last_success))
281dd25c YG	209	preferred = bdata->last_success;
1da177e4 LT	210	} else
	211	preferred = 0;
	212
8c0e33c1	213	preferred = ALIGN(preferred, align) >> PAGE_SHIFT;
1da177e4 LT	214	preferred += offset;
	215	areasize = (size+PAGE_SIZE-1)/PAGE_SIZE;
	216	incr = align >> PAGE_SHIFT ? : 1;
	217
	218	restart_scan:
	219	for (i = preferred; i < eidx; i += incr) {
	220	unsigned long j;
	221	i = find_next_zero_bit(bdata->node_bootmem_map, eidx, i);
	222	i = ALIGN(i, incr);
66d43e98 HM	223	if (i >= eidx)
66d43e98 HM	224	break;
1da177e4 LT	225	if (test_bit(i, bdata->node_bootmem_map))
	226	continue;
	227	for (j = i + 1; j < i + areasize; ++j) {
	228	if (j >= eidx)
	229	goto fail_block;
	230	if (test_bit (j, bdata->node_bootmem_map))
	231	goto fail_block;
	232	}
	233	start = i;
	234	goto found;
	235	fail_block:
	236	i = ALIGN(j, incr);
	237	}
	238
	239	if (preferred > offset) {
	240	preferred = offset;
	241	goto restart_scan;
	242	}
	243	return NULL;
	244
	245	found:
	246	bdata->last_success = start << PAGE_SHIFT;
	247	BUG_ON(start >= eidx);
	248
	249	/*
	250	* Is the next page of the previous allocation-end the start
	251	* of this allocation's buffer? If yes then we can 'merge'
	252	* the previous partial page with this allocation.
	253	*/
	254	if (align < PAGE_SIZE &&
	255	bdata->last_offset && bdata->last_pos+1 == start) {
8c0e33c1	256	offset = ALIGN(bdata->last_offset, align);
1da177e4 LT	257	BUG_ON(offset > PAGE_SIZE);
	258	remaining_size = PAGE_SIZE-offset;
	259	if (size < remaining_size) {
	260	areasize = 0;
	261	/* last_pos unchanged */
	262	bdata->last_offset = offset+size;
	263	ret = phys_to_virt(bdata->last_pos*PAGE_SIZE + offset +
	264	bdata->node_boot_start);
	265	} else {
	266	remaining_size = size - remaining_size;
	267	areasize = (remaining_size+PAGE_SIZE-1)/PAGE_SIZE;
	268	ret = phys_to_virt(bdata->last_pos*PAGE_SIZE + offset +
	269	bdata->node_boot_start);
	270	bdata->last_pos = start+areasize-1;
	271	bdata->last_offset = remaining_size;
	272	}
	273	bdata->last_offset &= ~PAGE_MASK;
	274	} else {
	275	bdata->last_pos = start + areasize - 1;
	276	bdata->last_offset = size & ~PAGE_MASK;
	277	ret = phys_to_virt(start * PAGE_SIZE + bdata->node_boot_start);
	278	}
	279
	280	/*
	281	* Reserve the area now:
	282	*/
	283	for (i = start; i < start+areasize; i++)
	284	if (unlikely(test_and_set_bit(i, bdata->node_bootmem_map)))
	285	BUG();
	286	memset(ret, 0, size);
	287	return ret;
	288	}
	289
	290	static unsigned long __init free_all_bootmem_core(pg_data_t *pgdat)
	291	{
	292	struct page *page;
d41dee36	293	unsigned long pfn;
1da177e4 LT	294	bootmem_data_t *bdata = pgdat->bdata;
	295	unsigned long i, count, total = 0;
	296	unsigned long idx;
	297	unsigned long *map;
	298	int gofast = 0;
	299
	300	BUG_ON(!bdata->node_bootmem_map);
	301
	302	count = 0;
	303	/* first extant page of the node */
d41dee36	304	pfn = bdata->node_boot_start >> PAGE_SHIFT;
1da177e4 LT	305	idx = bdata->node_low_pfn - (bdata->node_boot_start >> PAGE_SHIFT);
	306	map = bdata->node_bootmem_map;
	307	/* Check physaddr is O(LOG2(BITS_PER_LONG)) page aligned */
	308	if (bdata->node_boot_start == 0 \|\|
	309	ffs(bdata->node_boot_start) - PAGE_SHIFT > ffs(BITS_PER_LONG))
	310	gofast = 1;
	311	for (i = 0; i < idx; ) {
	312	unsigned long v = ~map[i / BITS_PER_LONG];
d41dee36	313
1da177e4	314	if (gofast && v == ~0UL) {
a226f6c8	315	int order;
1da177e4	316
d41dee36	317	page = pfn_to_page(pfn);
1da177e4	318	count += BITS_PER_LONG;
1da177e4	319	order = ffs(BITS_PER_LONG) - 1;
a226f6c8	320	__free_pages_bootmem(page, order);
1da177e4 LT	321	i += BITS_PER_LONG;
	322	page += BITS_PER_LONG;
	323	} else if (v) {
	324	unsigned long m;
d41dee36 AW	325
d41dee36 AW	326	page = pfn_to_page(pfn);
1da177e4 LT	327	for (m = 1; m && i < idx; m<<=1, page++, i++) {
	328	if (v & m) {
	329	count++;
a226f6c8	330	__free_pages_bootmem(page, 0);
1da177e4 LT	331	}
	332	}
	333	} else {
	334	i+=BITS_PER_LONG;
1da177e4	335	}
d41dee36	336	pfn += BITS_PER_LONG;
1da177e4 LT	337	}
	338	total += count;
	339
	340	/*
	341	* Now free the allocator bitmap itself, it's not
	342	* needed anymore:
	343	*/
	344	page = virt_to_page(bdata->node_bootmem_map);
	345	count = 0;
	346	for (i = 0; i < ((bdata->node_low_pfn-(bdata->node_boot_start >> PAGE_SHIFT))/8 + PAGE_SIZE-1)/PAGE_SIZE; i++,page++) {
	347	count++;
a226f6c8	348	__free_pages_bootmem(page, 0);
1da177e4 LT	349	}
	350	total += count;
	351	bdata->node_bootmem_map = NULL;
	352
	353	return total;
	354	}
	355
bb0923a6 FBH	356	unsigned long __init init_bootmem_node (pg_data_t *pgdat, unsigned long freepfn,
bb0923a6 FBH	357	unsigned long startpfn, unsigned long endpfn)
1da177e4 LT	358	{
	359	return(init_bootmem_core(pgdat, freepfn, startpfn, endpfn));
	360	}
	361
bb0923a6 FBH	362	void __init reserve_bootmem_node (pg_data_t *pgdat, unsigned long physaddr,
bb0923a6 FBH	363	unsigned long size)
1da177e4 LT	364	{
	365	reserve_bootmem_core(pgdat->bdata, physaddr, size);
	366	}
	367
bb0923a6 FBH	368	void __init free_bootmem_node (pg_data_t *pgdat, unsigned long physaddr,
bb0923a6 FBH	369	unsigned long size)
1da177e4 LT	370	{
	371	free_bootmem_core(pgdat->bdata, physaddr, size);
	372	}
	373
	374	unsigned long __init free_all_bootmem_node (pg_data_t *pgdat)
	375	{
	376	return(free_all_bootmem_core(pgdat));
	377	}
	378
	379	unsigned long __init init_bootmem (unsigned long start, unsigned long pages)
	380	{
	381	max_low_pfn = pages;
	382	min_low_pfn = start;
	383	return(init_bootmem_core(NODE_DATA(0), start, 0, pages));
	384	}
	385
	386	#ifndef CONFIG_HAVE_ARCH_BOOTMEM_NODE
	387	void __init reserve_bootmem (unsigned long addr, unsigned long size)
	388	{
	389	reserve_bootmem_core(NODE_DATA(0)->bdata, addr, size);
	390	}
	391	#endif /* !CONFIG_HAVE_ARCH_BOOTMEM_NODE */
	392
	393	void __init free_bootmem (unsigned long addr, unsigned long size)
	394	{
	395	free_bootmem_core(NODE_DATA(0)->bdata, addr, size);
	396	}
	397
	398	unsigned long __init free_all_bootmem (void)
	399	{
	400	return(free_all_bootmem_core(NODE_DATA(0)));
	401	}
	402
bb0923a6 FBH	403	void * __init __alloc_bootmem_nopanic(unsigned long size, unsigned long align,
bb0923a6 FBH	404	unsigned long goal)
1da177e4	405	{
679bc9fb	406	bootmem_data_t *bdata;
1da177e4 LT	407	void *ptr;
1da177e4 LT	408
679bc9fb KH	409	list_for_each_entry(bdata, &bdata_list, list)
679bc9fb KH	410	if ((ptr = __alloc_bootmem_core(bdata, size, align, goal, 0)))
1da177e4	411	return(ptr);
a8062231 AK	412	return NULL;
a8062231 AK	413	}
1da177e4	414
bb0923a6 FBH	415	void * __init __alloc_bootmem(unsigned long size, unsigned long align,
bb0923a6 FBH	416	unsigned long goal)
a8062231 AK	417	{
	418	void *mem = __alloc_bootmem_nopanic(size,align,goal);
	419	if (mem)
	420	return mem;
1da177e4 LT	421	/*
	422	* Whoops, we cannot satisfy the allocation request.
	423	*/
	424	printk(KERN_ALERT "bootmem alloc of %lu bytes failed!\n", size);
	425	panic("Out of memory");
	426	return NULL;
	427	}
	428
281dd25c	429
bb0923a6 FBH	430	void * __init __alloc_bootmem_node(pg_data_t *pgdat, unsigned long size,
bb0923a6 FBH	431	unsigned long align, unsigned long goal)
1da177e4 LT	432	{
	433	void *ptr;
	434
008857c1	435	ptr = __alloc_bootmem_core(pgdat->bdata, size, align, goal, 0);
1da177e4 LT	436	if (ptr)
	437	return (ptr);
	438
008857c1	439	return __alloc_bootmem(size, align, goal);
1da177e4 LT	440	}
1da177e4 LT	441
008857c1 RT	442	#define LOW32LIMIT 0xffffffff
008857c1 RT	443
bb0923a6 FBH	444	void * __init __alloc_bootmem_low(unsigned long size, unsigned long align,
bb0923a6 FBH	445	unsigned long goal)
008857c1	446	{
679bc9fb	447	bootmem_data_t *bdata;
008857c1 RT	448	void *ptr;
008857c1 RT	449
679bc9fb KH	450	list_for_each_entry(bdata, &bdata_list, list)
679bc9fb KH	451	if ((ptr = __alloc_bootmem_core(bdata, size,
008857c1 RT	452	align, goal, LOW32LIMIT)))
	453	return(ptr);
	454
	455	/*
	456	* Whoops, we cannot satisfy the allocation request.
	457	*/
	458	printk(KERN_ALERT "low bootmem alloc of %lu bytes failed!\n", size);
	459	panic("Out of low memory");
	460	return NULL;
	461	}
	462
	463	void * __init __alloc_bootmem_low_node(pg_data_t *pgdat, unsigned long size,
	464	unsigned long align, unsigned long goal)
	465	{
	466	return __alloc_bootmem_core(pgdat->bdata, size, align, goal, LOW32LIMIT);
	467	}