[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;

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 int __init reserve_bootmem_core(bootmem_data_t *bdata,
			unsigned long addr, unsigned long size, int flags)
{
	unsigned long sidx, eidx;
	unsigned long i;
	int ret;

	/*
	 * 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);
	BUG_ON(addr < bdata->node_boot_start);

	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
			if (flags & BOOTMEM_EXCLUSIVE) {
				ret = -EBUSY;
				goto err;
			}
		}

	return 0;

err:
	/* unreserve memory we accidentally reserved */
	for (i--; i >= sidx; i--)
		clear_bit(i, bdata->node_bootmem_map);

	return ret;
}

static void __init free_bootmem_core(bootmem_data_t *bdata, unsigned long addr,
				     unsigned long size)
{
	unsigned long sidx, eidx;
	unsigned long i;

	BUG_ON(!size);

	/* out range */
	if (addr + size < bdata->node_boot_start ||
		PFN_DOWN(addr) > bdata->node_low_pfn)
		return;
	/*
	 * round down end of usable mem, partially free pages are
	 * considered reserved.
	 */

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

	/*
	 * Round up to index to the range.
	 */
	if (PFN_UP(addr) > PFN_DOWN(bdata->node_boot_start))
		sidx = PFN_UP(addr) - PFN_DOWN(bdata->node_boot_start);
	else
		sidx = 0;

	eidx = PFN_DOWN(addr + size - bdata->node_boot_start);
	if (eidx > bdata->node_low_pfn - PFN_DOWN(bdata->node_boot_start))
		eidx = bdata->node_low_pfn - PFN_DOWN(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;

	/* on nodes without memory - bootmem_map is NULL */
	if (!bdata->node_bootmem_map)
		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, int flags)
{
	reserve_bootmem_core(pgdat->bdata, physaddr, size, flags);
}

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

void __init free_bootmem(unsigned long addr, unsigned long size)
{
	bootmem_data_t *bdata;
	list_for_each_entry(bdata, &bdata_list, list)
		free_bootmem_core(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
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	39	/* return the number of _pages_ that will be allocated for the boot bitmap */
f71bf0ca	40	unsigned long __init bootmem_bootmap_pages(unsigned long pages)
1da177e4 LT	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	}
f71bf0ca	50
679bc9fb KH	51	/*
	52	* link bdata in order
	53	*/
69d49e68	54	static void __init link_bootmem(bootmem_data_t *bdata)
679bc9fb KH	55	{
679bc9fb KH	56	bootmem_data_t *ent;
f71bf0ca	57
679bc9fb KH	58	if (list_empty(&bdata_list)) {
	59	list_add(&bdata->list, &bdata_list);
	60	return;
	61	}
	62	/* insert in order */
	63	list_for_each_entry(ent, &bdata_list, list) {
	64	if (bdata->node_boot_start < ent->node_boot_start) {
	65	list_add_tail(&bdata->list, &ent->list);
	66	return;
	67	}
	68	}
	69	list_add_tail(&bdata->list, &bdata_list);
679bc9fb KH	70	}
679bc9fb KH	71
bbc7b92e FBH	72	/*
	73	* Given an initialised bdata, it returns the size of the boot bitmap
	74	*/
	75	static unsigned long __init get_mapsize(bootmem_data_t *bdata)
	76	{
	77	unsigned long mapsize;
	78	unsigned long start = PFN_DOWN(bdata->node_boot_start);
	79	unsigned long end = bdata->node_low_pfn;
	80
	81	mapsize = ((end - start) + 7) / 8;
	82	return ALIGN(mapsize, sizeof(long));
	83	}
1da177e4 LT	84
	85	/*
	86	* Called once to set up the allocator itself.
	87	*/
f71bf0ca	88	static unsigned long __init init_bootmem_core(pg_data_t *pgdat,
1da177e4 LT	89	unsigned long mapstart, unsigned long start, unsigned long end)
	90	{
	91	bootmem_data_t *bdata = pgdat->bdata;
bbc7b92e	92	unsigned long mapsize;
1da177e4	93
bbc7b92e FBH	94	bdata->node_bootmem_map = phys_to_virt(PFN_PHYS(mapstart));
bbc7b92e FBH	95	bdata->node_boot_start = PFN_PHYS(start);
1da177e4	96	bdata->node_low_pfn = end;
679bc9fb	97	link_bootmem(bdata);
1da177e4 LT	98
	99	/*
	100	* Initially all pages are reserved - setup_arch() has to
	101	* register free RAM areas explicitly.
	102	*/
bbc7b92e	103	mapsize = get_mapsize(bdata);
1da177e4 LT	104	memset(bdata->node_bootmem_map, 0xff, mapsize);
	105
	106	return mapsize;
	107	}
	108
	109	/*
	110	* Marks a particular physical memory range as unallocatable. Usable RAM
	111	* might be used for boot-time allocations - or it might get added
	112	* to the free page pool later on.
	113	*/
72a7fe39 BW	114	static int __init reserve_bootmem_core(bootmem_data_t *bdata,
72a7fe39 BW	115	unsigned long addr, unsigned long size, int flags)
1da177e4	116	{
bbc7b92e	117	unsigned long sidx, eidx;
1da177e4	118	unsigned long i;
72a7fe39	119	int ret;
bbc7b92e	120
1da177e4 LT	121	/*
	122	* round up, partially reserved pages are considered
	123	* fully reserved.
	124	*/
1da177e4	125	BUG_ON(!size);
bbc7b92e FBH	126	BUG_ON(PFN_DOWN(addr) >= bdata->node_low_pfn);
bbc7b92e FBH	127	BUG_ON(PFN_UP(addr + size) > bdata->node_low_pfn);
5a982cbc	128	BUG_ON(addr < bdata->node_boot_start);
bbc7b92e FBH	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
72a7fe39 BW	138	if (flags & BOOTMEM_EXCLUSIVE) {
	139	ret = -EBUSY;
	140	goto err;
	141	}
1da177e4	142	}
72a7fe39 BW	143
	144	return 0;
	145
	146	err:
	147	/* unreserve memory we accidentally reserved */
	148	for (i--; i >= sidx; i--)
	149	clear_bit(i, bdata->node_bootmem_map);
	150
	151	return ret;
1da177e4 LT	152	}
1da177e4 LT	153
bb0923a6 FBH	154	static void __init free_bootmem_core(bootmem_data_t *bdata, unsigned long addr,
bb0923a6 FBH	155	unsigned long size)
1da177e4	156	{
bbc7b92e	157	unsigned long sidx, eidx;
1da177e4	158	unsigned long i;
bbc7b92e	159
5a982cbc YL	160	BUG_ON(!size);
	161
	162	/* out range */
	163	if (addr + size < bdata->node_boot_start \|\|
	164	PFN_DOWN(addr) > bdata->node_low_pfn)
	165	return;
1da177e4 LT	166	/*
	167	* round down end of usable mem, partially free pages are
	168	* considered reserved.
	169	*/
1da177e4	170
5a982cbc	171	if (addr >= bdata->node_boot_start && addr < bdata->last_success)
1da177e4 LT	172	bdata->last_success = addr;
	173
	174	/*
5a982cbc	175	* Round up to index to the range.
1da177e4	176	*/
5a982cbc YL	177	if (PFN_UP(addr) > PFN_DOWN(bdata->node_boot_start))
	178	sidx = PFN_UP(addr) - PFN_DOWN(bdata->node_boot_start);
	179	else
	180	sidx = 0;
	181
bbc7b92e	182	eidx = PFN_DOWN(addr + size - bdata->node_boot_start);
5a982cbc YL	183	if (eidx > bdata->node_low_pfn - PFN_DOWN(bdata->node_boot_start))
5a982cbc YL	184	eidx = bdata->node_low_pfn - PFN_DOWN(bdata->node_boot_start);
1da177e4 LT	185
	186	for (i = sidx; i < eidx; i++) {
	187	if (unlikely(!test_and_clear_bit(i, bdata->node_bootmem_map)))
	188	BUG();
	189	}
	190	}
	191
	192	/*
	193	* We 'merge' subsequent allocations to save space. We might 'lose'
	194	* some fraction of a page if allocations cannot be satisfied due to
	195	* size constraints on boxes where there is physical RAM space
	196	* fragmentation - in these cases (mostly large memory boxes) this
	197	* is not a problem.
	198	*
	199	* On low memory boxes we get it right in 100% of the cases.
	200	*
	201	* alignment has to be a power of 2 value.
	202	*
	203	* NOTE: This function is _not_ reentrant.
	204	*/
267b4801	205	void * __init
1da177e4	206	__alloc_bootmem_core(struct bootmem_data *bdata, unsigned long size,
281dd25c	207	unsigned long align, unsigned long goal, unsigned long limit)
1da177e4 LT	208	{
1da177e4 LT	209	unsigned long offset, remaining_size, areasize, preferred;
bbc7b92e	210	unsigned long i, start = 0, incr, eidx, end_pfn;
1da177e4 LT	211	void *ret;
1da177e4 LT	212
f71bf0ca	213	if (!size) {
1da177e4 LT	214	printk("__alloc_bootmem_core(): zero-sized request\n");
	215	BUG();
	216	}
	217	BUG_ON(align & (align-1));
	218
281dd25c YG	219	if (limit && bdata->node_boot_start >= limit)
	220	return NULL;
	221
7c309a64 CK	222	/* on nodes without memory - bootmem_map is NULL */
	223	if (!bdata->node_bootmem_map)
	224	return NULL;
	225
bbc7b92e FBH	226	end_pfn = bdata->node_low_pfn;
bbc7b92e FBH	227	limit = PFN_DOWN(limit);
281dd25c YG	228	if (limit && end_pfn > limit)
	229	end_pfn = limit;
	230
bbc7b92e	231	eidx = end_pfn - PFN_DOWN(bdata->node_boot_start);
1da177e4	232	offset = 0;
bbc7b92e FBH	233	if (align && (bdata->node_boot_start & (align - 1UL)) != 0)
	234	offset = align - (bdata->node_boot_start & (align - 1UL));
	235	offset = PFN_DOWN(offset);
1da177e4 LT	236
	237	/*
	238	* We try to allocate bootmem pages above 'goal'
	239	* first, then we try to allocate lower pages.
	240	*/
bbc7b92e	241	if (goal && goal >= bdata->node_boot_start && PFN_DOWN(goal) < end_pfn) {
1da177e4 LT	242	preferred = goal - bdata->node_boot_start;
	243
	244	if (bdata->last_success >= preferred)
281dd25c YG	245	if (!limit \|\| (limit && limit > bdata->last_success))
281dd25c YG	246	preferred = bdata->last_success;
1da177e4 LT	247	} else
	248	preferred = 0;
	249
bbc7b92e FBH	250	preferred = PFN_DOWN(ALIGN(preferred, align)) + offset;
bbc7b92e FBH	251	areasize = (size + PAGE_SIZE-1) / PAGE_SIZE;
1da177e4 LT	252	incr = align >> PAGE_SHIFT ? : 1;
	253
	254	restart_scan:
	255	for (i = preferred; i < eidx; i += incr) {
	256	unsigned long j;
	257	i = find_next_zero_bit(bdata->node_bootmem_map, eidx, i);
	258	i = ALIGN(i, incr);
66d43e98 HM	259	if (i >= eidx)
66d43e98 HM	260	break;
1da177e4 LT	261	if (test_bit(i, bdata->node_bootmem_map))
	262	continue;
	263	for (j = i + 1; j < i + areasize; ++j) {
	264	if (j >= eidx)
	265	goto fail_block;
f71bf0ca	266	if (test_bit(j, bdata->node_bootmem_map))
1da177e4 LT	267	goto fail_block;
	268	}
	269	start = i;
	270	goto found;
	271	fail_block:
	272	i = ALIGN(j, incr);
	273	}
	274
	275	if (preferred > offset) {
	276	preferred = offset;
	277	goto restart_scan;
	278	}
	279	return NULL;
	280
	281	found:
bbc7b92e	282	bdata->last_success = PFN_PHYS(start);
1da177e4 LT	283	BUG_ON(start >= eidx);
	284
	285	/*
	286	* Is the next page of the previous allocation-end the start
	287	* of this allocation's buffer? If yes then we can 'merge'
	288	* the previous partial page with this allocation.
	289	*/
	290	if (align < PAGE_SIZE &&
	291	bdata->last_offset && bdata->last_pos+1 == start) {
8c0e33c1	292	offset = ALIGN(bdata->last_offset, align);
1da177e4	293	BUG_ON(offset > PAGE_SIZE);
f71bf0ca	294	remaining_size = PAGE_SIZE - offset;
1da177e4 LT	295	if (size < remaining_size) {
	296	areasize = 0;
	297	/* last_pos unchanged */
f71bf0ca FBH	298	bdata->last_offset = offset + size;
	299	ret = phys_to_virt(bdata->last_pos * PAGE_SIZE +
	300	offset +
	301	bdata->node_boot_start);
1da177e4 LT	302	} else {
1da177e4 LT	303	remaining_size = size - remaining_size;
f71bf0ca FBH	304	areasize = (remaining_size + PAGE_SIZE-1) / PAGE_SIZE;
	305	ret = phys_to_virt(bdata->last_pos * PAGE_SIZE +
	306	offset +
	307	bdata->node_boot_start);
	308	bdata->last_pos = start + areasize - 1;
1da177e4 LT	309	bdata->last_offset = remaining_size;
	310	}
	311	bdata->last_offset &= ~PAGE_MASK;
	312	} else {
	313	bdata->last_pos = start + areasize - 1;
	314	bdata->last_offset = size & ~PAGE_MASK;
	315	ret = phys_to_virt(start * PAGE_SIZE + bdata->node_boot_start);
	316	}
	317
	318	/*
	319	* Reserve the area now:
	320	*/
f71bf0ca	321	for (i = start; i < start + areasize; i++)
1da177e4 LT	322	if (unlikely(test_and_set_bit(i, bdata->node_bootmem_map)))
	323	BUG();
	324	memset(ret, 0, size);
	325	return ret;
	326	}
	327
	328	static unsigned long __init free_all_bootmem_core(pg_data_t *pgdat)
	329	{
	330	struct page *page;
d41dee36	331	unsigned long pfn;
1da177e4 LT	332	bootmem_data_t *bdata = pgdat->bdata;
	333	unsigned long i, count, total = 0;
	334	unsigned long idx;
	335	unsigned long *map;
	336	int gofast = 0;
	337
	338	BUG_ON(!bdata->node_bootmem_map);
	339
	340	count = 0;
	341	/* first extant page of the node */
bbc7b92e FBH	342	pfn = PFN_DOWN(bdata->node_boot_start);
bbc7b92e FBH	343	idx = bdata->node_low_pfn - pfn;
1da177e4 LT	344	map = bdata->node_bootmem_map;
	345	/* Check physaddr is O(LOG2(BITS_PER_LONG)) page aligned */
	346	if (bdata->node_boot_start == 0 \|\|
	347	ffs(bdata->node_boot_start) - PAGE_SHIFT > ffs(BITS_PER_LONG))
	348	gofast = 1;
	349	for (i = 0; i < idx; ) {
	350	unsigned long v = ~map[i / BITS_PER_LONG];
d41dee36	351
1da177e4	352	if (gofast && v == ~0UL) {
a226f6c8	353	int order;
1da177e4	354
d41dee36	355	page = pfn_to_page(pfn);
1da177e4	356	count += BITS_PER_LONG;
1da177e4	357	order = ffs(BITS_PER_LONG) - 1;
a226f6c8	358	__free_pages_bootmem(page, order);
1da177e4 LT	359	i += BITS_PER_LONG;
	360	page += BITS_PER_LONG;
	361	} else if (v) {
	362	unsigned long m;
d41dee36 AW	363
d41dee36 AW	364	page = pfn_to_page(pfn);
1da177e4 LT	365	for (m = 1; m && i < idx; m<<=1, page++, i++) {
	366	if (v & m) {
	367	count++;
a226f6c8	368	__free_pages_bootmem(page, 0);
1da177e4 LT	369	}
	370	}
	371	} else {
f71bf0ca	372	i += BITS_PER_LONG;
1da177e4	373	}
d41dee36	374	pfn += BITS_PER_LONG;
1da177e4 LT	375	}
	376	total += count;
	377
	378	/*
	379	* Now free the allocator bitmap itself, it's not
	380	* needed anymore:
	381	*/
	382	page = virt_to_page(bdata->node_bootmem_map);
	383	count = 0;
bbc7b92e FBH	384	idx = (get_mapsize(bdata) + PAGE_SIZE-1) >> PAGE_SHIFT;
bbc7b92e FBH	385	for (i = 0; i < idx; i++, page++) {
a226f6c8	386	__free_pages_bootmem(page, 0);
bbc7b92e	387	count++;
1da177e4 LT	388	}
	389	total += count;
	390	bdata->node_bootmem_map = NULL;
	391
	392	return total;
	393	}
	394
f71bf0ca	395	unsigned long __init init_bootmem_node(pg_data_t *pgdat, unsigned long freepfn,
bb0923a6	396	unsigned long startpfn, unsigned long endpfn)
1da177e4	397	{
f71bf0ca	398	return init_bootmem_core(pgdat, freepfn, startpfn, endpfn);
1da177e4 LT	399	}
1da177e4 LT	400
f71bf0ca	401	void __init reserve_bootmem_node(pg_data_t *pgdat, unsigned long physaddr,
72a7fe39	402	unsigned long size, int flags)
1da177e4	403	{
72a7fe39	404	reserve_bootmem_core(pgdat->bdata, physaddr, size, flags);
1da177e4 LT	405	}
1da177e4 LT	406
f71bf0ca FBH	407	void __init free_bootmem_node(pg_data_t *pgdat, unsigned long physaddr,
f71bf0ca FBH	408	unsigned long size)
1da177e4 LT	409	{
	410	free_bootmem_core(pgdat->bdata, physaddr, size);
	411	}
	412
f71bf0ca	413	unsigned long __init free_all_bootmem_node(pg_data_t *pgdat)
1da177e4	414	{
f71bf0ca	415	return free_all_bootmem_core(pgdat);
1da177e4 LT	416	}
1da177e4 LT	417
f71bf0ca	418	unsigned long __init init_bootmem(unsigned long start, unsigned long pages)
1da177e4 LT	419	{
	420	max_low_pfn = pages;
	421	min_low_pfn = start;
f71bf0ca	422	return init_bootmem_core(NODE_DATA(0), start, 0, pages);
1da177e4 LT	423	}
	424
	425	#ifndef CONFIG_HAVE_ARCH_BOOTMEM_NODE
72a7fe39 BW	426	int __init reserve_bootmem(unsigned long addr, unsigned long size,
72a7fe39 BW	427	int flags)
1da177e4	428	{
72a7fe39	429	return reserve_bootmem_core(NODE_DATA(0)->bdata, addr, size, flags);
1da177e4 LT	430	}
	431	#endif /* !CONFIG_HAVE_ARCH_BOOTMEM_NODE */
	432
f71bf0ca	433	void __init free_bootmem(unsigned long addr, unsigned long size)
1da177e4	434	{
5a982cbc YL	435	bootmem_data_t *bdata;
	436	list_for_each_entry(bdata, &bdata_list, list)
	437	free_bootmem_core(bdata, addr, size);
1da177e4 LT	438	}
1da177e4 LT	439
f71bf0ca	440	unsigned long __init free_all_bootmem(void)
1da177e4	441	{
f71bf0ca	442	return free_all_bootmem_core(NODE_DATA(0));
1da177e4 LT	443	}
1da177e4 LT	444
bb0923a6 FBH	445	void * __init __alloc_bootmem_nopanic(unsigned long size, unsigned long align,
bb0923a6 FBH	446	unsigned long goal)
1da177e4	447	{
679bc9fb	448	bootmem_data_t *bdata;
1da177e4 LT	449	void *ptr;
1da177e4 LT	450
f71bf0ca FBH	451	list_for_each_entry(bdata, &bdata_list, list) {
	452	ptr = __alloc_bootmem_core(bdata, size, align, goal, 0);
	453	if (ptr)
	454	return ptr;
	455	}
a8062231 AK	456	return NULL;
a8062231 AK	457	}
1da177e4	458
bb0923a6 FBH	459	void * __init __alloc_bootmem(unsigned long size, unsigned long align,
bb0923a6 FBH	460	unsigned long goal)
a8062231 AK	461	{
a8062231 AK	462	void *mem = __alloc_bootmem_nopanic(size,align,goal);
f71bf0ca	463
a8062231 AK	464	if (mem)
a8062231 AK	465	return mem;
1da177e4 LT	466	/*
	467	* Whoops, we cannot satisfy the allocation request.
	468	*/
	469	printk(KERN_ALERT "bootmem alloc of %lu bytes failed!\n", size);
	470	panic("Out of memory");
	471	return NULL;
	472	}
	473
281dd25c	474
bb0923a6 FBH	475	void * __init __alloc_bootmem_node(pg_data_t *pgdat, unsigned long size,
bb0923a6 FBH	476	unsigned long align, unsigned long goal)
1da177e4 LT	477	{
	478	void *ptr;
	479
008857c1	480	ptr = __alloc_bootmem_core(pgdat->bdata, size, align, goal, 0);
1da177e4	481	if (ptr)
f71bf0ca	482	return ptr;
1da177e4	483
008857c1	484	return __alloc_bootmem(size, align, goal);
1da177e4 LT	485	}
1da177e4 LT	486
dfd54cbc HC	487	#ifndef ARCH_LOW_ADDRESS_LIMIT
	488	#define ARCH_LOW_ADDRESS_LIMIT 0xffffffffUL
	489	#endif
008857c1	490
bb0923a6 FBH	491	void * __init __alloc_bootmem_low(unsigned long size, unsigned long align,
bb0923a6 FBH	492	unsigned long goal)
008857c1	493	{
679bc9fb	494	bootmem_data_t *bdata;
008857c1 RT	495	void *ptr;
008857c1 RT	496
f71bf0ca	497	list_for_each_entry(bdata, &bdata_list, list) {
dfd54cbc HC	498	ptr = __alloc_bootmem_core(bdata, size, align, goal,
dfd54cbc HC	499	ARCH_LOW_ADDRESS_LIMIT);
f71bf0ca FBH	500	if (ptr)
	501	return ptr;
	502	}
008857c1 RT	503
	504	/*
	505	* Whoops, we cannot satisfy the allocation request.
	506	*/
	507	printk(KERN_ALERT "low bootmem alloc of %lu bytes failed!\n", size);
	508	panic("Out of low memory");
	509	return NULL;
	510	}
	511
	512	void * __init __alloc_bootmem_low_node(pg_data_t *pgdat, unsigned long size,
	513	unsigned long align, unsigned long goal)
	514	{
dfd54cbc HC	515	return __alloc_bootmem_core(pgdat->bdata, size, align, goal,
dfd54cbc HC	516	ARCH_LOW_ADDRESS_LIMIT);
008857c1	517	}