[GitHub/mt8127/android_kernel_alcatel_ttab.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 can_reserve_bootmem_core(bootmem_data_t *bdata,
			unsigned long addr, unsigned long size, int flags)
{
	unsigned long sidx, eidx;
	unsigned long i;

	BUG_ON(!size);

	/* out of range, don't hold other */
	if (addr + size < bdata->node_boot_start ||
		PFN_DOWN(addr) > bdata->node_low_pfn)
		return 0;

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

	eidx = PFN_UP(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 (test_bit(i, bdata->node_bootmem_map)) {
			if (flags & BOOTMEM_EXCLUSIVE)
				return -EBUSY;
		}
	}

	return 0;

}

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

	BUG_ON(!size);

	/* out of range */
	if (addr + size < bdata->node_boot_start ||
		PFN_DOWN(addr) > bdata->node_low_pfn)
		return;

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

	eidx = PFN_UP(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 (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;

	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 areasize, preferred;
	unsigned long i, start = 0, incr, eidx, end_pfn;
	void *ret;
	unsigned long node_boot_start;
	void *node_bootmem_map;

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

	/* on nodes without memory - bootmem_map is NULL */
	if (!bdata->node_bootmem_map)
		return NULL;

	/* bdata->node_boot_start is supposed to be (12+6)bits alignment on x86_64 ? */
	node_boot_start = bdata->node_boot_start;
	node_bootmem_map = bdata->node_bootmem_map;
	if (align) {
		node_boot_start = ALIGN(bdata->node_boot_start, align);
		if (node_boot_start > bdata->node_boot_start)
			node_bootmem_map = (unsigned long *)bdata->node_bootmem_map +
			    PFN_DOWN(node_boot_start - bdata->node_boot_start)/BITS_PER_LONG;
	}

	if (limit && 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(node_boot_start);

	/*
	 * We try to allocate bootmem pages above 'goal'
	 * first, then we try to allocate lower pages.
	 */
	preferred = 0;
	if (goal && PFN_DOWN(goal) < end_pfn) {
		if (goal > node_boot_start)
			preferred = goal - node_boot_start;

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

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

restart_scan:
	for (i = preferred; i < eidx;) {
		unsigned long j;

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

	if (preferred > 0) {
		preferred = 0;
		goto restart_scan;
	}
	return NULL;

found:
	bdata->last_success = PFN_PHYS(start) + node_boot_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) {
		unsigned long offset, remaining_size;
		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 + 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 + 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 + node_boot_start);
	}

	/*
	 * Reserve the area now:
	 */
	for (i = start; i < start + areasize; i++)
		if (unlikely(test_and_set_bit(i, 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)
{
	int ret;

	ret = can_reserve_bootmem_core(pgdat->bdata, physaddr, size, flags);
	if (ret < 0)
		return;
	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)
{
	register_page_bootmem_info_node(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)
{
	bootmem_data_t *bdata;
	int ret;

	list_for_each_entry(bdata, &bdata_list, list) {
		ret = can_reserve_bootmem_core(bdata, addr, size, flags);
		if (ret < 0)
			return ret;
	}
	list_for_each_entry(bdata, &bdata_list, list)
		reserve_bootmem_core(bdata, addr, size, flags);

	return 0;
}
#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);
}

#ifdef CONFIG_SPARSEMEM
void * __init alloc_bootmem_section(unsigned long size,
				    unsigned long section_nr)
{
	void *ptr;
	unsigned long limit, goal, start_nr, end_nr, pfn;
	struct pglist_data *pgdat;

	pfn = section_nr_to_pfn(section_nr);
	goal = PFN_PHYS(pfn);
	limit = PFN_PHYS(section_nr_to_pfn(section_nr + 1)) - 1;
	pgdat = NODE_DATA(early_pfn_to_nid(pfn));
	ptr = __alloc_bootmem_core(pgdat->bdata, size, SMP_CACHE_BYTES, goal,
				   limit);

	if (!ptr)
		return NULL;

	start_nr = pfn_to_section_nr(PFN_DOWN(__pa(ptr)));
	end_nr = pfn_to_section_nr(PFN_DOWN(__pa(ptr) + size));
	if (start_nr != section_nr || end_nr != section_nr) {
		printk(KERN_WARNING "alloc_bootmem failed on section %ld.\n",
		       section_nr);
		free_bootmem_core(pgdat->bdata, __pa(ptr), size);
		ptr = NULL;
	}

	return ptr;
}
#endif

#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	*/
a5645a61	114	static int __init can_reserve_bootmem_core(bootmem_data_t *bdata,
72a7fe39	115	unsigned long addr, unsigned long size, int flags)
1da177e4	116	{
bbc7b92e	117	unsigned long sidx, eidx;
1da177e4	118	unsigned long i;
a5645a61 YL	119
	120	BUG_ON(!size);
	121
	122	/* out of range, don't hold other */
	123	if (addr + size < bdata->node_boot_start \|\|
	124	PFN_DOWN(addr) > bdata->node_low_pfn)
	125	return 0;
bbc7b92e	126
1da177e4	127	/*
a5645a61	128	* Round up to index to the range.
1da177e4	129	*/
a5645a61 YL	130	if (addr > bdata->node_boot_start)
	131	sidx= PFN_DOWN(addr - bdata->node_boot_start);
	132	else
	133	sidx = 0;
	134
	135	eidx = PFN_UP(addr + size - bdata->node_boot_start);
	136	if (eidx > bdata->node_low_pfn - PFN_DOWN(bdata->node_boot_start))
	137	eidx = bdata->node_low_pfn - PFN_DOWN(bdata->node_boot_start);
	138
	139	for (i = sidx; i < eidx; i++) {
	140	if (test_bit(i, bdata->node_bootmem_map)) {
	141	if (flags & BOOTMEM_EXCLUSIVE)
	142	return -EBUSY;
	143	}
	144	}
	145
	146	return 0;
	147
	148	}
	149
	150	static void __init reserve_bootmem_core(bootmem_data_t *bdata,
	151	unsigned long addr, unsigned long size, int flags)
	152	{
	153	unsigned long sidx, eidx;
	154	unsigned long i;
	155
1da177e4	156	BUG_ON(!size);
bbc7b92e	157
a5645a61 YL	158	/* out of range */
	159	if (addr + size < bdata->node_boot_start \|\|
	160	PFN_DOWN(addr) > bdata->node_low_pfn)
	161	return;
	162
	163	/*
	164	* Round up to index to the range.
	165	*/
	166	if (addr > bdata->node_boot_start)
	167	sidx= PFN_DOWN(addr - bdata->node_boot_start);
	168	else
	169	sidx = 0;
	170
bbc7b92e	171	eidx = PFN_UP(addr + size - bdata->node_boot_start);
a5645a61 YL	172	if (eidx > bdata->node_low_pfn - PFN_DOWN(bdata->node_boot_start))
a5645a61 YL	173	eidx = bdata->node_low_pfn - PFN_DOWN(bdata->node_boot_start);
1da177e4	174
a5645a61	175	for (i = sidx; i < eidx; i++) {
1da177e4 LT	176	if (test_and_set_bit(i, bdata->node_bootmem_map)) {
	177	#ifdef CONFIG_DEBUG_BOOTMEM
	178	printk("hm, page %08lx reserved twice.\n", i*PAGE_SIZE);
	179	#endif
	180	}
a5645a61	181	}
1da177e4 LT	182	}
1da177e4 LT	183
bb0923a6 FBH	184	static void __init free_bootmem_core(bootmem_data_t *bdata, unsigned long addr,
bb0923a6 FBH	185	unsigned long size)
1da177e4	186	{
bbc7b92e	187	unsigned long sidx, eidx;
1da177e4	188	unsigned long i;
bbc7b92e	189
5a982cbc YL	190	BUG_ON(!size);
	191
	192	/* out range */
	193	if (addr + size < bdata->node_boot_start \|\|
	194	PFN_DOWN(addr) > bdata->node_low_pfn)
	195	return;
1da177e4 LT	196	/*
	197	* round down end of usable mem, partially free pages are
	198	* considered reserved.
	199	*/
1da177e4	200
5a982cbc	201	if (addr >= bdata->node_boot_start && addr < bdata->last_success)
1da177e4 LT	202	bdata->last_success = addr;
	203
	204	/*
5a982cbc	205	* Round up to index to the range.
1da177e4	206	*/
5a982cbc YL	207	if (PFN_UP(addr) > PFN_DOWN(bdata->node_boot_start))
	208	sidx = PFN_UP(addr) - PFN_DOWN(bdata->node_boot_start);
	209	else
	210	sidx = 0;
	211
bbc7b92e	212	eidx = PFN_DOWN(addr + size - bdata->node_boot_start);
5a982cbc YL	213	if (eidx > bdata->node_low_pfn - PFN_DOWN(bdata->node_boot_start))
5a982cbc YL	214	eidx = bdata->node_low_pfn - PFN_DOWN(bdata->node_boot_start);
1da177e4 LT	215
	216	for (i = sidx; i < eidx; i++) {
	217	if (unlikely(!test_and_clear_bit(i, bdata->node_bootmem_map)))
	218	BUG();
	219	}
	220	}
	221
	222	/*
	223	* We 'merge' subsequent allocations to save space. We might 'lose'
	224	* some fraction of a page if allocations cannot be satisfied due to
	225	* size constraints on boxes where there is physical RAM space
	226	* fragmentation - in these cases (mostly large memory boxes) this
	227	* is not a problem.
	228	*
	229	* On low memory boxes we get it right in 100% of the cases.
	230	*
	231	* alignment has to be a power of 2 value.
	232	*
	233	* NOTE: This function is _not_ reentrant.
	234	*/
267b4801	235	void * __init
1da177e4	236	__alloc_bootmem_core(struct bootmem_data *bdata, unsigned long size,
281dd25c	237	unsigned long align, unsigned long goal, unsigned long limit)
1da177e4	238	{
9a2dc04c	239	unsigned long areasize, preferred;
bbc7b92e	240	unsigned long i, start = 0, incr, eidx, end_pfn;
1da177e4	241	void *ret;
9a2dc04c YL	242	unsigned long node_boot_start;
9a2dc04c YL	243	void *node_bootmem_map;
1da177e4	244
f71bf0ca	245	if (!size) {
1da177e4 LT	246	printk("__alloc_bootmem_core(): zero-sized request\n");
	247	BUG();
	248	}
	249	BUG_ON(align & (align-1));
	250
7c309a64 CK	251	/* on nodes without memory - bootmem_map is NULL */
	252	if (!bdata->node_bootmem_map)
	253	return NULL;
	254
9a2dc04c YL	255	/* bdata->node_boot_start is supposed to be (12+6)bits alignment on x86_64 ? */
	256	node_boot_start = bdata->node_boot_start;
	257	node_bootmem_map = bdata->node_bootmem_map;
	258	if (align) {
	259	node_boot_start = ALIGN(bdata->node_boot_start, align);
	260	if (node_boot_start > bdata->node_boot_start)
	261	node_bootmem_map = (unsigned long *)bdata->node_bootmem_map +
	262	PFN_DOWN(node_boot_start - bdata->node_boot_start)/BITS_PER_LONG;
	263	}
	264
	265	if (limit && node_boot_start >= limit)
	266	return NULL;
	267
bbc7b92e FBH	268	end_pfn = bdata->node_low_pfn;
bbc7b92e FBH	269	limit = PFN_DOWN(limit);
281dd25c YG	270	if (limit && end_pfn > limit)
	271	end_pfn = limit;
	272
9a2dc04c	273	eidx = end_pfn - PFN_DOWN(node_boot_start);
1da177e4 LT	274
	275	/*
	276	* We try to allocate bootmem pages above 'goal'
	277	* first, then we try to allocate lower pages.
	278	*/
ad09315c YL	279	preferred = 0;
ad09315c YL	280	if (goal && PFN_DOWN(goal) < end_pfn) {
9a2dc04c YL	281	if (goal > node_boot_start)
9a2dc04c YL	282	preferred = goal - node_boot_start;
1da177e4	283
9a2dc04c YL	284	if (bdata->last_success > node_boot_start &&
9a2dc04c YL	285	bdata->last_success - node_boot_start >= preferred)
281dd25c	286	if (!limit \|\| (limit && limit > bdata->last_success))
9a2dc04c	287	preferred = bdata->last_success - node_boot_start;
ad09315c	288	}
1da177e4	289
9a2dc04c	290	preferred = PFN_DOWN(ALIGN(preferred, align));
bbc7b92e	291	areasize = (size + PAGE_SIZE-1) / PAGE_SIZE;
1da177e4 LT	292	incr = align >> PAGE_SHIFT ? : 1;
	293
	294	restart_scan:
ad09315c	295	for (i = preferred; i < eidx;) {
1da177e4	296	unsigned long j;
ad09315c	297
9a2dc04c	298	i = find_next_zero_bit(node_bootmem_map, eidx, i);
1da177e4	299	i = ALIGN(i, incr);
66d43e98 HM	300	if (i >= eidx)
66d43e98 HM	301	break;
9a2dc04c	302	if (test_bit(i, node_bootmem_map)) {
ad09315c	303	i += incr;
1da177e4	304	continue;
ad09315c	305	}
1da177e4 LT	306	for (j = i + 1; j < i + areasize; ++j) {
	307	if (j >= eidx)
	308	goto fail_block;
9a2dc04c	309	if (test_bit(j, node_bootmem_map))
1da177e4 LT	310	goto fail_block;
	311	}
	312	start = i;
	313	goto found;
	314	fail_block:
	315	i = ALIGN(j, incr);
ad09315c YL	316	if (i == j)
ad09315c YL	317	i += incr;
1da177e4 LT	318	}
1da177e4 LT	319
9a2dc04c YL	320	if (preferred > 0) {
9a2dc04c YL	321	preferred = 0;
1da177e4 LT	322	goto restart_scan;
	323	}
	324	return NULL;
	325
	326	found:
9a2dc04c	327	bdata->last_success = PFN_PHYS(start) + node_boot_start;
1da177e4 LT	328	BUG_ON(start >= eidx);
	329
	330	/*
	331	* Is the next page of the previous allocation-end the start
	332	* of this allocation's buffer? If yes then we can 'merge'
	333	* the previous partial page with this allocation.
	334	*/
	335	if (align < PAGE_SIZE &&
	336	bdata->last_offset && bdata->last_pos+1 == start) {
9a2dc04c	337	unsigned long offset, remaining_size;
8c0e33c1	338	offset = ALIGN(bdata->last_offset, align);
1da177e4	339	BUG_ON(offset > PAGE_SIZE);
f71bf0ca	340	remaining_size = PAGE_SIZE - offset;
1da177e4 LT	341	if (size < remaining_size) {
	342	areasize = 0;
	343	/* last_pos unchanged */
f71bf0ca FBH	344	bdata->last_offset = offset + size;
f71bf0ca FBH	345	ret = phys_to_virt(bdata->last_pos * PAGE_SIZE +
9a2dc04c	346	offset + node_boot_start);
1da177e4 LT	347	} else {
1da177e4 LT	348	remaining_size = size - remaining_size;
f71bf0ca FBH	349	areasize = (remaining_size + PAGE_SIZE-1) / PAGE_SIZE;
f71bf0ca FBH	350	ret = phys_to_virt(bdata->last_pos * PAGE_SIZE +
9a2dc04c	351	offset + node_boot_start);
f71bf0ca	352	bdata->last_pos = start + areasize - 1;
1da177e4 LT	353	bdata->last_offset = remaining_size;
	354	}
	355	bdata->last_offset &= ~PAGE_MASK;
	356	} else {
	357	bdata->last_pos = start + areasize - 1;
	358	bdata->last_offset = size & ~PAGE_MASK;
9a2dc04c	359	ret = phys_to_virt(start * PAGE_SIZE + node_boot_start);
1da177e4 LT	360	}
	361
	362	/*
	363	* Reserve the area now:
	364	*/
f71bf0ca	365	for (i = start; i < start + areasize; i++)
9a2dc04c	366	if (unlikely(test_and_set_bit(i, node_bootmem_map)))
1da177e4 LT	367	BUG();
	368	memset(ret, 0, size);
	369	return ret;
	370	}
	371
	372	static unsigned long __init free_all_bootmem_core(pg_data_t *pgdat)
	373	{
	374	struct page *page;
d41dee36	375	unsigned long pfn;
1da177e4 LT	376	bootmem_data_t *bdata = pgdat->bdata;
	377	unsigned long i, count, total = 0;
	378	unsigned long idx;
	379	unsigned long *map;
	380	int gofast = 0;
	381
	382	BUG_ON(!bdata->node_bootmem_map);
	383
	384	count = 0;
	385	/* first extant page of the node */
bbc7b92e FBH	386	pfn = PFN_DOWN(bdata->node_boot_start);
bbc7b92e FBH	387	idx = bdata->node_low_pfn - pfn;
1da177e4 LT	388	map = bdata->node_bootmem_map;
	389	/* Check physaddr is O(LOG2(BITS_PER_LONG)) page aligned */
	390	if (bdata->node_boot_start == 0 \|\|
	391	ffs(bdata->node_boot_start) - PAGE_SHIFT > ffs(BITS_PER_LONG))
	392	gofast = 1;
	393	for (i = 0; i < idx; ) {
	394	unsigned long v = ~map[i / BITS_PER_LONG];
d41dee36	395
1da177e4	396	if (gofast && v == ~0UL) {
a226f6c8	397	int order;
1da177e4	398
d41dee36	399	page = pfn_to_page(pfn);
1da177e4	400	count += BITS_PER_LONG;
1da177e4	401	order = ffs(BITS_PER_LONG) - 1;
a226f6c8	402	__free_pages_bootmem(page, order);
1da177e4 LT	403	i += BITS_PER_LONG;
	404	page += BITS_PER_LONG;
	405	} else if (v) {
	406	unsigned long m;
d41dee36 AW	407
d41dee36 AW	408	page = pfn_to_page(pfn);
1da177e4 LT	409	for (m = 1; m && i < idx; m<<=1, page++, i++) {
	410	if (v & m) {
	411	count++;
a226f6c8	412	__free_pages_bootmem(page, 0);
1da177e4 LT	413	}
	414	}
	415	} else {
f71bf0ca	416	i += BITS_PER_LONG;
1da177e4	417	}
d41dee36	418	pfn += BITS_PER_LONG;
1da177e4 LT	419	}
	420	total += count;
	421
	422	/*
	423	* Now free the allocator bitmap itself, it's not
	424	* needed anymore:
	425	*/
	426	page = virt_to_page(bdata->node_bootmem_map);
	427	count = 0;
bbc7b92e FBH	428	idx = (get_mapsize(bdata) + PAGE_SIZE-1) >> PAGE_SHIFT;
bbc7b92e FBH	429	for (i = 0; i < idx; i++, page++) {
a226f6c8	430	__free_pages_bootmem(page, 0);
bbc7b92e	431	count++;
1da177e4 LT	432	}
	433	total += count;
	434	bdata->node_bootmem_map = NULL;
	435
	436	return total;
	437	}
	438
f71bf0ca	439	unsigned long __init init_bootmem_node(pg_data_t *pgdat, unsigned long freepfn,
bb0923a6	440	unsigned long startpfn, unsigned long endpfn)
1da177e4	441	{
f71bf0ca	442	return init_bootmem_core(pgdat, freepfn, startpfn, endpfn);
1da177e4 LT	443	}
1da177e4 LT	444
f71bf0ca	445	void __init reserve_bootmem_node(pg_data_t *pgdat, unsigned long physaddr,
72a7fe39	446	unsigned long size, int flags)
1da177e4	447	{
a5645a61 YL	448	int ret;
	449
	450	ret = can_reserve_bootmem_core(pgdat->bdata, physaddr, size, flags);
	451	if (ret < 0)
	452	return;
72a7fe39	453	reserve_bootmem_core(pgdat->bdata, physaddr, size, flags);
1da177e4 LT	454	}
1da177e4 LT	455
f71bf0ca FBH	456	void __init free_bootmem_node(pg_data_t *pgdat, unsigned long physaddr,
f71bf0ca FBH	457	unsigned long size)
1da177e4 LT	458	{
	459	free_bootmem_core(pgdat->bdata, physaddr, size);
	460	}
	461
f71bf0ca	462	unsigned long __init free_all_bootmem_node(pg_data_t *pgdat)
1da177e4	463	{
04753278	464	register_page_bootmem_info_node(pgdat);
f71bf0ca	465	return free_all_bootmem_core(pgdat);
1da177e4 LT	466	}
1da177e4 LT	467
f71bf0ca	468	unsigned long __init init_bootmem(unsigned long start, unsigned long pages)
1da177e4 LT	469	{
	470	max_low_pfn = pages;
	471	min_low_pfn = start;
f71bf0ca	472	return init_bootmem_core(NODE_DATA(0), start, 0, pages);
1da177e4 LT	473	}
	474
	475	#ifndef CONFIG_HAVE_ARCH_BOOTMEM_NODE
72a7fe39 BW	476	int __init reserve_bootmem(unsigned long addr, unsigned long size,
72a7fe39 BW	477	int flags)
1da177e4	478	{
a5645a61 YL	479	bootmem_data_t *bdata;
	480	int ret;
	481
	482	list_for_each_entry(bdata, &bdata_list, list) {
	483	ret = can_reserve_bootmem_core(bdata, addr, size, flags);
	484	if (ret < 0)
	485	return ret;
	486	}
	487	list_for_each_entry(bdata, &bdata_list, list)
	488	reserve_bootmem_core(bdata, addr, size, flags);
	489
	490	return 0;
1da177e4 LT	491	}
	492	#endif /* !CONFIG_HAVE_ARCH_BOOTMEM_NODE */
	493
f71bf0ca	494	void __init free_bootmem(unsigned long addr, unsigned long size)
1da177e4	495	{
5a982cbc YL	496	bootmem_data_t *bdata;
	497	list_for_each_entry(bdata, &bdata_list, list)
	498	free_bootmem_core(bdata, addr, size);
1da177e4 LT	499	}
1da177e4 LT	500
f71bf0ca	501	unsigned long __init free_all_bootmem(void)
1da177e4	502	{
f71bf0ca	503	return free_all_bootmem_core(NODE_DATA(0));
1da177e4 LT	504	}
1da177e4 LT	505
bb0923a6 FBH	506	void * __init __alloc_bootmem_nopanic(unsigned long size, unsigned long align,
bb0923a6 FBH	507	unsigned long goal)
1da177e4	508	{
679bc9fb	509	bootmem_data_t *bdata;
1da177e4 LT	510	void *ptr;
1da177e4 LT	511
f71bf0ca FBH	512	list_for_each_entry(bdata, &bdata_list, list) {
	513	ptr = __alloc_bootmem_core(bdata, size, align, goal, 0);
	514	if (ptr)
	515	return ptr;
	516	}
a8062231 AK	517	return NULL;
a8062231 AK	518	}
1da177e4	519
bb0923a6 FBH	520	void * __init __alloc_bootmem(unsigned long size, unsigned long align,
bb0923a6 FBH	521	unsigned long goal)
a8062231 AK	522	{
a8062231 AK	523	void *mem = __alloc_bootmem_nopanic(size,align,goal);
f71bf0ca	524
a8062231 AK	525	if (mem)
a8062231 AK	526	return mem;
1da177e4 LT	527	/*
	528	* Whoops, we cannot satisfy the allocation request.
	529	*/
	530	printk(KERN_ALERT "bootmem alloc of %lu bytes failed!\n", size);
	531	panic("Out of memory");
	532	return NULL;
	533	}
	534
281dd25c	535
bb0923a6 FBH	536	void * __init __alloc_bootmem_node(pg_data_t *pgdat, unsigned long size,
bb0923a6 FBH	537	unsigned long align, unsigned long goal)
1da177e4 LT	538	{
	539	void *ptr;
	540
008857c1	541	ptr = __alloc_bootmem_core(pgdat->bdata, size, align, goal, 0);
1da177e4	542	if (ptr)
f71bf0ca	543	return ptr;
1da177e4	544
008857c1	545	return __alloc_bootmem(size, align, goal);
1da177e4 LT	546	}
1da177e4 LT	547
e70260aa YG	548	#ifdef CONFIG_SPARSEMEM
	549	void * __init alloc_bootmem_section(unsigned long size,
	550	unsigned long section_nr)
	551	{
	552	void *ptr;
	553	unsigned long limit, goal, start_nr, end_nr, pfn;
	554	struct pglist_data *pgdat;
	555
	556	pfn = section_nr_to_pfn(section_nr);
	557	goal = PFN_PHYS(pfn);
	558	limit = PFN_PHYS(section_nr_to_pfn(section_nr + 1)) - 1;
	559	pgdat = NODE_DATA(early_pfn_to_nid(pfn));
	560	ptr = __alloc_bootmem_core(pgdat->bdata, size, SMP_CACHE_BYTES, goal,
	561	limit);
	562
	563	if (!ptr)
	564	return NULL;
	565
	566	start_nr = pfn_to_section_nr(PFN_DOWN(__pa(ptr)));
	567	end_nr = pfn_to_section_nr(PFN_DOWN(__pa(ptr) + size));
	568	if (start_nr != section_nr \|\| end_nr != section_nr) {
	569	printk(KERN_WARNING "alloc_bootmem failed on section %ld.\n",
	570	section_nr);
	571	free_bootmem_core(pgdat->bdata, __pa(ptr), size);
	572	ptr = NULL;
	573	}
	574
	575	return ptr;
	576	}
	577	#endif
	578
dfd54cbc HC	579	#ifndef ARCH_LOW_ADDRESS_LIMIT
	580	#define ARCH_LOW_ADDRESS_LIMIT 0xffffffffUL
	581	#endif
008857c1	582
bb0923a6 FBH	583	void * __init __alloc_bootmem_low(unsigned long size, unsigned long align,
bb0923a6 FBH	584	unsigned long goal)
008857c1	585	{
679bc9fb	586	bootmem_data_t *bdata;
008857c1 RT	587	void *ptr;
008857c1 RT	588
f71bf0ca	589	list_for_each_entry(bdata, &bdata_list, list) {
dfd54cbc HC	590	ptr = __alloc_bootmem_core(bdata, size, align, goal,
dfd54cbc HC	591	ARCH_LOW_ADDRESS_LIMIT);
f71bf0ca FBH	592	if (ptr)
	593	return ptr;
	594	}
008857c1 RT	595
	596	/*
	597	* Whoops, we cannot satisfy the allocation request.
	598	*/
	599	printk(KERN_ALERT "low bootmem alloc of %lu bytes failed!\n", size);
	600	panic("Out of low memory");
	601	return NULL;
	602	}
	603
	604	void * __init __alloc_bootmem_low_node(pg_data_t *pgdat, unsigned long size,
	605	unsigned long align, unsigned long goal)
	606	{
dfd54cbc HC	607	return __alloc_bootmem_core(pgdat->bdata, size, align, goal,
dfd54cbc HC	608	ARCH_LOW_ADDRESS_LIMIT);
008857c1	609	}