[net-next-2.6.git] / mm / sparse.c

/*
 * sparse memory mappings.
 */
#include <linux/mm.h>
#include <linux/mmzone.h>
#include <linux/bootmem.h>
#include <linux/highmem.h>
#include <linux/module.h>
#include <linux/spinlock.h>
#include <linux/vmalloc.h>
#include "internal.h"
#include <asm/dma.h>
#include <asm/pgalloc.h>
#include <asm/pgtable.h>

/*
 * Permanent SPARSEMEM data:
 *
 * 1) mem_section	- memory sections, mem_map's for valid memory
 */
#ifdef CONFIG_SPARSEMEM_EXTREME
struct mem_section *mem_section[NR_SECTION_ROOTS]
	____cacheline_internodealigned_in_smp;
#else
struct mem_section mem_section[NR_SECTION_ROOTS][SECTIONS_PER_ROOT]
	____cacheline_internodealigned_in_smp;
#endif
EXPORT_SYMBOL(mem_section);

#ifdef NODE_NOT_IN_PAGE_FLAGS
/*
 * If we did not store the node number in the page then we have to
 * do a lookup in the section_to_node_table in order to find which
 * node the page belongs to.
 */
#if MAX_NUMNODES <= 256
static u8 section_to_node_table[NR_MEM_SECTIONS] __cacheline_aligned;
#else
static u16 section_to_node_table[NR_MEM_SECTIONS] __cacheline_aligned;
#endif

int page_to_nid(struct page *page)
{
	return section_to_node_table[page_to_section(page)];
}
EXPORT_SYMBOL(page_to_nid);

static void set_section_nid(unsigned long section_nr, int nid)
{
	section_to_node_table[section_nr] = nid;
}
#else /* !NODE_NOT_IN_PAGE_FLAGS */
static inline void set_section_nid(unsigned long section_nr, int nid)
{
}
#endif

#ifdef CONFIG_SPARSEMEM_EXTREME
static struct mem_section noinline __init_refok *sparse_index_alloc(int nid)
{
	struct mem_section *section = NULL;
	unsigned long array_size = SECTIONS_PER_ROOT *
				   sizeof(struct mem_section);

	if (slab_is_available())
		section = kmalloc_node(array_size, GFP_KERNEL, nid);
	else
		section = alloc_bootmem_node(NODE_DATA(nid), array_size);

	if (section)
		memset(section, 0, array_size);

	return section;
}

static int __meminit sparse_index_init(unsigned long section_nr, int nid)
{
	static DEFINE_SPINLOCK(index_init_lock);
	unsigned long root = SECTION_NR_TO_ROOT(section_nr);
	struct mem_section *section;
	int ret = 0;

	if (mem_section[root])
		return -EEXIST;

	section = sparse_index_alloc(nid);
	if (!section)
		return -ENOMEM;
	/*
	 * This lock keeps two different sections from
	 * reallocating for the same index
	 */
	spin_lock(&index_init_lock);

	if (mem_section[root]) {
		ret = -EEXIST;
		goto out;
	}

	mem_section[root] = section;
out:
	spin_unlock(&index_init_lock);
	return ret;
}
#else /* !SPARSEMEM_EXTREME */
static inline int sparse_index_init(unsigned long section_nr, int nid)
{
	return 0;
}
#endif

/*
 * Although written for the SPARSEMEM_EXTREME case, this happens
 * to also work for the flat array case because
 * NR_SECTION_ROOTS==NR_MEM_SECTIONS.
 */
int __section_nr(struct mem_section* ms)
{
	unsigned long root_nr;
	struct mem_section* root;

	for (root_nr = 0; root_nr < NR_SECTION_ROOTS; root_nr++) {
		root = __nr_to_section(root_nr * SECTIONS_PER_ROOT);
		if (!root)
			continue;

		if ((ms >= root) && (ms < (root + SECTIONS_PER_ROOT)))
		     break;
	}

	return (root_nr * SECTIONS_PER_ROOT) + (ms - root);
}

/*
 * During early boot, before section_mem_map is used for an actual
 * mem_map, we use section_mem_map to store the section's NUMA
 * node.  This keeps us from having to use another data structure.  The
 * node information is cleared just before we store the real mem_map.
 */
static inline unsigned long sparse_encode_early_nid(int nid)
{
	return (nid << SECTION_NID_SHIFT);
}

static inline int sparse_early_nid(struct mem_section *section)
{
	return (section->section_mem_map >> SECTION_NID_SHIFT);
}

/* Record a memory area against a node. */
void __init memory_present(int nid, unsigned long start, unsigned long end)
{
	unsigned long max_arch_pfn = 1UL << (MAX_PHYSMEM_BITS-PAGE_SHIFT);
	unsigned long pfn;

	/*
	 * Sanity checks - do not allow an architecture to pass
	 * in larger pfns than the maximum scope of sparsemem:
	 */
	if (start >= max_arch_pfn)
		return;
	if (end >= max_arch_pfn)
		end = max_arch_pfn;

	start &= PAGE_SECTION_MASK;
	for (pfn = start; pfn < end; pfn += PAGES_PER_SECTION) {
		unsigned long section = pfn_to_section_nr(pfn);
		struct mem_section *ms;

		sparse_index_init(section, nid);
		set_section_nid(section, nid);

		ms = __nr_to_section(section);
		if (!ms->section_mem_map)
			ms->section_mem_map = sparse_encode_early_nid(nid) |
							SECTION_MARKED_PRESENT;
	}
}

/*
 * Only used by the i386 NUMA architecures, but relatively
 * generic code.
 */
unsigned long __init node_memmap_size_bytes(int nid, unsigned long start_pfn,
						     unsigned long end_pfn)
{
	unsigned long pfn;
	unsigned long nr_pages = 0;

	for (pfn = start_pfn; pfn < end_pfn; pfn += PAGES_PER_SECTION) {
		if (nid != early_pfn_to_nid(pfn))
			continue;

		if (pfn_present(pfn))
			nr_pages += PAGES_PER_SECTION;
	}

	return nr_pages * sizeof(struct page);
}

/*
 * Subtle, we encode the real pfn into the mem_map such that
 * the identity pfn - section_mem_map will return the actual
 * physical page frame number.
 */
static unsigned long sparse_encode_mem_map(struct page *mem_map, unsigned long pnum)
{
	return (unsigned long)(mem_map - (section_nr_to_pfn(pnum)));
}

/*
 * Decode mem_map from the coded memmap
 */
struct page *sparse_decode_mem_map(unsigned long coded_mem_map, unsigned long pnum)
{
	/* mask off the extra low bits of information */
	coded_mem_map &= SECTION_MAP_MASK;
	return ((struct page *)coded_mem_map) + section_nr_to_pfn(pnum);
}

static int __meminit sparse_init_one_section(struct mem_section *ms,
		unsigned long pnum, struct page *mem_map,
		unsigned long *pageblock_bitmap)
{
	if (!present_section(ms))
		return -EINVAL;

	ms->section_mem_map &= ~SECTION_MAP_MASK;
	ms->section_mem_map |= sparse_encode_mem_map(mem_map, pnum) |
							SECTION_HAS_MEM_MAP;
 	ms->pageblock_flags = pageblock_bitmap;

	return 1;
}

unsigned long usemap_size(void)
{
	unsigned long size_bytes;
	size_bytes = roundup(SECTION_BLOCKFLAGS_BITS, 8) / 8;
	size_bytes = roundup(size_bytes, sizeof(unsigned long));
	return size_bytes;
}

#ifdef CONFIG_MEMORY_HOTPLUG
static unsigned long *__kmalloc_section_usemap(void)
{
	return kmalloc(usemap_size(), GFP_KERNEL);
}
#endif /* CONFIG_MEMORY_HOTPLUG */

static unsigned long *__init sparse_early_usemap_alloc(unsigned long pnum)
{
	unsigned long *usemap;
	struct mem_section *ms = __nr_to_section(pnum);
	int nid = sparse_early_nid(ms);

	usemap = alloc_bootmem_node(NODE_DATA(nid), usemap_size());
	if (usemap)
		return usemap;

	/* Stupid: suppress gcc warning for SPARSEMEM && !NUMA */
	nid = 0;

	printk(KERN_WARNING "%s: allocation failed\n", __func__);
	return NULL;
}

#ifndef CONFIG_SPARSEMEM_VMEMMAP
struct page __init *sparse_mem_map_populate(unsigned long pnum, int nid)
{
	struct page *map;

	map = alloc_remap(nid, sizeof(struct page) * PAGES_PER_SECTION);
	if (map)
		return map;

	map = alloc_bootmem_pages_node(NODE_DATA(nid),
		       PAGE_ALIGN(sizeof(struct page) * PAGES_PER_SECTION));
	return map;
}
#endif /* !CONFIG_SPARSEMEM_VMEMMAP */

struct page __init *sparse_early_mem_map_alloc(unsigned long pnum)
{
	struct page *map;
	struct mem_section *ms = __nr_to_section(pnum);
	int nid = sparse_early_nid(ms);

	map = sparse_mem_map_populate(pnum, nid);
	if (map)
		return map;

	printk(KERN_ERR "%s: sparsemem memory map backing failed "
			"some memory will not be available.\n", __func__);
	ms->section_mem_map = 0;
	return NULL;
}

void __attribute__((weak)) __meminit vmemmap_populate_print_last(void)
{
}
/*
 * Allocate the accumulated non-linear sections, allocate a mem_map
 * for each and record the physical to section mapping.
 */
void __init sparse_init(void)
{
	unsigned long pnum;
	struct page *map;
	unsigned long *usemap;
	unsigned long **usemap_map;
	int size;

	/*
	 * map is using big page (aka 2M in x86 64 bit)
	 * usemap is less one page (aka 24 bytes)
	 * so alloc 2M (with 2M align) and 24 bytes in turn will
	 * make next 2M slip to one more 2M later.
	 * then in big system, the memory will have a lot of holes...
	 * here try to allocate 2M pages continously.
	 *
	 * powerpc need to call sparse_init_one_section right after each
	 * sparse_early_mem_map_alloc, so allocate usemap_map at first.
	 */
	size = sizeof(unsigned long *) * NR_MEM_SECTIONS;
	usemap_map = alloc_bootmem(size);
	if (!usemap_map)
		panic("can not allocate usemap_map\n");

	for (pnum = 0; pnum < NR_MEM_SECTIONS; pnum++) {
		if (!present_section_nr(pnum))
			continue;
		usemap_map[pnum] = sparse_early_usemap_alloc(pnum);
	}

	for (pnum = 0; pnum < NR_MEM_SECTIONS; pnum++) {
		if (!present_section_nr(pnum))
			continue;

		usemap = usemap_map[pnum];
		if (!usemap)
			continue;

		map = sparse_early_mem_map_alloc(pnum);
		if (!map)
			continue;

		sparse_init_one_section(__nr_to_section(pnum), pnum, map,
								usemap);
	}

	vmemmap_populate_print_last();

	free_bootmem(__pa(usemap_map), size);
}

#ifdef CONFIG_MEMORY_HOTPLUG
#ifdef CONFIG_SPARSEMEM_VMEMMAP
static inline struct page *kmalloc_section_memmap(unsigned long pnum, int nid,
						 unsigned long nr_pages)
{
	/* This will make the necessary allocations eventually. */
	return sparse_mem_map_populate(pnum, nid);
}
static void __kfree_section_memmap(struct page *memmap, unsigned long nr_pages)
{
	return; /* XXX: Not implemented yet */
}
static void free_map_bootmem(struct page *page, unsigned long nr_pages)
{
}
#else
static struct page *__kmalloc_section_memmap(unsigned long nr_pages)
{
	struct page *page, *ret;
	unsigned long memmap_size = sizeof(struct page) * nr_pages;

	page = alloc_pages(GFP_KERNEL|__GFP_NOWARN, get_order(memmap_size));
	if (page)
		goto got_map_page;

	ret = vmalloc(memmap_size);
	if (ret)
		goto got_map_ptr;

	return NULL;
got_map_page:
	ret = (struct page *)pfn_to_kaddr(page_to_pfn(page));
got_map_ptr:
	memset(ret, 0, memmap_size);

	return ret;
}

static inline struct page *kmalloc_section_memmap(unsigned long pnum, int nid,
						  unsigned long nr_pages)
{
	return __kmalloc_section_memmap(nr_pages);
}

static void __kfree_section_memmap(struct page *memmap, unsigned long nr_pages)
{
	if (is_vmalloc_addr(memmap))
		vfree(memmap);
	else
		free_pages((unsigned long)memmap,
			   get_order(sizeof(struct page) * nr_pages));
}

static void free_map_bootmem(struct page *page, unsigned long nr_pages)
{
	unsigned long maps_section_nr, removing_section_nr, i;
	int magic;

	for (i = 0; i < nr_pages; i++, page++) {
		magic = atomic_read(&page->_mapcount);

		BUG_ON(magic == NODE_INFO);

		maps_section_nr = pfn_to_section_nr(page_to_pfn(page));
		removing_section_nr = page->private;

		/*
		 * When this function is called, the removing section is
		 * logical offlined state. This means all pages are isolated
		 * from page allocator. If removing section's memmap is placed
		 * on the same section, it must not be freed.
		 * If it is freed, page allocator may allocate it which will
		 * be removed physically soon.
		 */
		if (maps_section_nr != removing_section_nr)
			put_page_bootmem(page);
	}
}
#endif /* CONFIG_SPARSEMEM_VMEMMAP */

static void free_section_usemap(struct page *memmap, unsigned long *usemap)
{
	struct page *usemap_page;
	unsigned long nr_pages;

	if (!usemap)
		return;

	usemap_page = virt_to_page(usemap);
	/*
	 * Check to see if allocation came from hot-plug-add
	 */
	if (PageSlab(usemap_page)) {
		kfree(usemap);
		if (memmap)
			__kfree_section_memmap(memmap, PAGES_PER_SECTION);
		return;
	}

	/*
	 * The usemap came from bootmem. This is packed with other usemaps
	 * on the section which has pgdat at boot time. Just keep it as is now.
	 */

	if (memmap) {
		struct page *memmap_page;
		memmap_page = virt_to_page(memmap);

		nr_pages = PAGE_ALIGN(PAGES_PER_SECTION * sizeof(struct page))
			>> PAGE_SHIFT;

		free_map_bootmem(memmap_page, nr_pages);
	}
}

/*
 * returns the number of sections whose mem_maps were properly
 * set.  If this is <=0, then that means that the passed-in
 * map was not consumed and must be freed.
 */
int sparse_add_one_section(struct zone *zone, unsigned long start_pfn,
			   int nr_pages)
{
	unsigned long section_nr = pfn_to_section_nr(start_pfn);
	struct pglist_data *pgdat = zone->zone_pgdat;
	struct mem_section *ms;
	struct page *memmap;
	unsigned long *usemap;
	unsigned long flags;
	int ret;

	/*
	 * no locking for this, because it does its own
	 * plus, it does a kmalloc
	 */
	ret = sparse_index_init(section_nr, pgdat->node_id);
	if (ret < 0 && ret != -EEXIST)
		return ret;
	memmap = kmalloc_section_memmap(section_nr, pgdat->node_id, nr_pages);
	if (!memmap)
		return -ENOMEM;
	usemap = __kmalloc_section_usemap();
	if (!usemap) {
		__kfree_section_memmap(memmap, nr_pages);
		return -ENOMEM;
	}

	pgdat_resize_lock(pgdat, &flags);

	ms = __pfn_to_section(start_pfn);
	if (ms->section_mem_map & SECTION_MARKED_PRESENT) {
		ret = -EEXIST;
		goto out;
	}

	ms->section_mem_map |= SECTION_MARKED_PRESENT;

	ret = sparse_init_one_section(ms, section_nr, memmap, usemap);

out:
	pgdat_resize_unlock(pgdat, &flags);
	if (ret <= 0) {
		kfree(usemap);
		__kfree_section_memmap(memmap, nr_pages);
	}
	return ret;
}

void sparse_remove_one_section(struct zone *zone, struct mem_section *ms)
{
	struct page *memmap = NULL;
	unsigned long *usemap = NULL;

	if (ms->section_mem_map) {
		usemap = ms->pageblock_flags;
		memmap = sparse_decode_mem_map(ms->section_mem_map,
						__section_nr(ms));
		ms->section_mem_map = 0;
		ms->pageblock_flags = NULL;
	}

	free_section_usemap(memmap, usemap);
}
#endif
Commit	Line	Data
d41dee36 AW	1	/*
	2	* sparse memory mappings.
	3	*/
d41dee36 AW	4	#include <linux/mm.h>
	5	#include <linux/mmzone.h>
	6	#include <linux/bootmem.h>
0b0acbec	7	#include <linux/highmem.h>
d41dee36	8	#include <linux/module.h>
28ae55c9	9	#include <linux/spinlock.h>
0b0acbec	10	#include <linux/vmalloc.h>
0c0a4a51	11	#include "internal.h"
d41dee36	12	#include <asm/dma.h>
8f6aac41 CL	13	#include <asm/pgalloc.h>
8f6aac41 CL	14	#include <asm/pgtable.h>
d41dee36 AW	15
	16	/*
	17	* Permanent SPARSEMEM data:
	18	*
	19	* 1) mem_section - memory sections, mem_map's for valid memory
	20	*/
3e347261	21	#ifdef CONFIG_SPARSEMEM_EXTREME
802f192e	22	struct mem_section *mem_section[NR_SECTION_ROOTS]
22fc6ecc	23	____cacheline_internodealigned_in_smp;
3e347261 BP	24	#else
3e347261 BP	25	struct mem_section mem_section[NR_SECTION_ROOTS][SECTIONS_PER_ROOT]
22fc6ecc	26	____cacheline_internodealigned_in_smp;
3e347261 BP	27	#endif
	28	EXPORT_SYMBOL(mem_section);
	29
89689ae7 CL	30	#ifdef NODE_NOT_IN_PAGE_FLAGS
	31	/*
	32	* If we did not store the node number in the page then we have to
	33	* do a lookup in the section_to_node_table in order to find which
	34	* node the page belongs to.
	35	*/
	36	#if MAX_NUMNODES <= 256
	37	static u8 section_to_node_table[NR_MEM_SECTIONS] __cacheline_aligned;
	38	#else
	39	static u16 section_to_node_table[NR_MEM_SECTIONS] __cacheline_aligned;
	40	#endif
	41
25ba77c1	42	int page_to_nid(struct page *page)
89689ae7 CL	43	{
	44	return section_to_node_table[page_to_section(page)];
	45	}
	46	EXPORT_SYMBOL(page_to_nid);
85770ffe AW	47
	48	static void set_section_nid(unsigned long section_nr, int nid)
	49	{
	50	section_to_node_table[section_nr] = nid;
	51	}
	52	#else /* !NODE_NOT_IN_PAGE_FLAGS */
	53	static inline void set_section_nid(unsigned long section_nr, int nid)
	54	{
	55	}
89689ae7 CL	56	#endif
89689ae7 CL	57
3e347261	58	#ifdef CONFIG_SPARSEMEM_EXTREME
577a32f6	59	static struct mem_section noinline __init_refok *sparse_index_alloc(int nid)
28ae55c9 DH	60	{
	61	struct mem_section *section = NULL;
	62	unsigned long array_size = SECTIONS_PER_ROOT *
	63	sizeof(struct mem_section);
	64
39d24e64	65	if (slab_is_available())
46a66eec MK	66	section = kmalloc_node(array_size, GFP_KERNEL, nid);
	67	else
	68	section = alloc_bootmem_node(NODE_DATA(nid), array_size);
28ae55c9 DH	69
	70	if (section)
	71	memset(section, 0, array_size);
	72
	73	return section;
3e347261	74	}
802f192e	75
a3142c8e	76	static int __meminit sparse_index_init(unsigned long section_nr, int nid)
802f192e	77	{
34af946a	78	static DEFINE_SPINLOCK(index_init_lock);
28ae55c9 DH	79	unsigned long root = SECTION_NR_TO_ROOT(section_nr);
	80	struct mem_section *section;
	81	int ret = 0;
802f192e BP	82
802f192e BP	83	if (mem_section[root])
28ae55c9	84	return -EEXIST;
3e347261	85
28ae55c9	86	section = sparse_index_alloc(nid);
af0cd5a7 WC	87	if (!section)
af0cd5a7 WC	88	return -ENOMEM;
28ae55c9 DH	89	/*
	90	* This lock keeps two different sections from
	91	* reallocating for the same index
	92	*/
	93	spin_lock(&index_init_lock);
3e347261	94
28ae55c9 DH	95	if (mem_section[root]) {
	96	ret = -EEXIST;
	97	goto out;
	98	}
	99
	100	mem_section[root] = section;
	101	out:
	102	spin_unlock(&index_init_lock);
	103	return ret;
	104	}
	105	#else /* !SPARSEMEM_EXTREME */
	106	static inline int sparse_index_init(unsigned long section_nr, int nid)
	107	{
	108	return 0;
802f192e	109	}
28ae55c9 DH	110	#endif
28ae55c9 DH	111
4ca644d9 DH	112	/*
4ca644d9 DH	113	* Although written for the SPARSEMEM_EXTREME case, this happens
cd881a6b	114	* to also work for the flat array case because
4ca644d9 DH	115	* NR_SECTION_ROOTS==NR_MEM_SECTIONS.
	116	*/
	117	int __section_nr(struct mem_section* ms)
	118	{
	119	unsigned long root_nr;
	120	struct mem_section* root;
	121
12783b00 MK	122	for (root_nr = 0; root_nr < NR_SECTION_ROOTS; root_nr++) {
12783b00 MK	123	root = __nr_to_section(root_nr * SECTIONS_PER_ROOT);
4ca644d9 DH	124	if (!root)
	125	continue;
	126
	127	if ((ms >= root) && (ms < (root + SECTIONS_PER_ROOT)))
	128	break;
	129	}
	130
	131	return (root_nr * SECTIONS_PER_ROOT) + (ms - root);
	132	}
	133
30c253e6 AW	134	/*
	135	* During early boot, before section_mem_map is used for an actual
	136	* mem_map, we use section_mem_map to store the section's NUMA
	137	* node. This keeps us from having to use another data structure. The
	138	* node information is cleared just before we store the real mem_map.
	139	*/
	140	static inline unsigned long sparse_encode_early_nid(int nid)
	141	{
	142	return (nid << SECTION_NID_SHIFT);
	143	}
	144
	145	static inline int sparse_early_nid(struct mem_section *section)
	146	{
	147	return (section->section_mem_map >> SECTION_NID_SHIFT);
	148	}
	149
d41dee36	150	/* Record a memory area against a node. */
a3142c8e	151	void __init memory_present(int nid, unsigned long start, unsigned long end)
d41dee36	152	{
bead9a3a	153	unsigned long max_arch_pfn = 1UL << (MAX_PHYSMEM_BITS-PAGE_SHIFT);
d41dee36 AW	154	unsigned long pfn;
d41dee36 AW	155
bead9a3a IM	156	/*
	157	* Sanity checks - do not allow an architecture to pass
	158	* in larger pfns than the maximum scope of sparsemem:
	159	*/
	160	if (start >= max_arch_pfn)
	161	return;
	162	if (end >= max_arch_pfn)
	163	end = max_arch_pfn;
	164
d41dee36 AW	165	start &= PAGE_SECTION_MASK;
	166	for (pfn = start; pfn < end; pfn += PAGES_PER_SECTION) {
	167	unsigned long section = pfn_to_section_nr(pfn);
802f192e BP	168	struct mem_section *ms;
	169
	170	sparse_index_init(section, nid);
85770ffe	171	set_section_nid(section, nid);
802f192e BP	172
	173	ms = __nr_to_section(section);
	174	if (!ms->section_mem_map)
30c253e6 AW	175	ms->section_mem_map = sparse_encode_early_nid(nid) \|
30c253e6 AW	176	SECTION_MARKED_PRESENT;
d41dee36 AW	177	}
	178	}
	179
	180	/*
	181	* Only used by the i386 NUMA architecures, but relatively
	182	* generic code.
	183	*/
	184	unsigned long __init node_memmap_size_bytes(int nid, unsigned long start_pfn,
	185	unsigned long end_pfn)
	186	{
	187	unsigned long pfn;
	188	unsigned long nr_pages = 0;
	189
	190	for (pfn = start_pfn; pfn < end_pfn; pfn += PAGES_PER_SECTION) {
	191	if (nid != early_pfn_to_nid(pfn))
	192	continue;
	193
540557b9	194	if (pfn_present(pfn))
d41dee36 AW	195	nr_pages += PAGES_PER_SECTION;
	196	}
	197
	198	return nr_pages * sizeof(struct page);
	199	}
	200
29751f69 AW	201	/*
	202	* Subtle, we encode the real pfn into the mem_map such that
	203	* the identity pfn - section_mem_map will return the actual
	204	* physical page frame number.
	205	*/
	206	static unsigned long sparse_encode_mem_map(struct page *mem_map, unsigned long pnum)
	207	{
	208	return (unsigned long)(mem_map - (section_nr_to_pfn(pnum)));
	209	}
	210
	211	/*
ea01ea93	212	* Decode mem_map from the coded memmap
29751f69	213	*/
29751f69 AW	214	struct page *sparse_decode_mem_map(unsigned long coded_mem_map, unsigned long pnum)
29751f69 AW	215	{
ea01ea93 BP	216	/* mask off the extra low bits of information */
ea01ea93 BP	217	coded_mem_map &= SECTION_MAP_MASK;
29751f69 AW	218	return ((struct page *)coded_mem_map) + section_nr_to_pfn(pnum);
	219	}
	220
a3142c8e	221	static int __meminit sparse_init_one_section(struct mem_section *ms,
5c0e3066 MG	222	unsigned long pnum, struct page *mem_map,
5c0e3066 MG	223	unsigned long *pageblock_bitmap)
29751f69	224	{
540557b9	225	if (!present_section(ms))
29751f69 AW	226	return -EINVAL;
29751f69 AW	227
30c253e6	228	ms->section_mem_map &= ~SECTION_MAP_MASK;
540557b9 AW	229	ms->section_mem_map \|= sparse_encode_mem_map(mem_map, pnum) \|
540557b9 AW	230	SECTION_HAS_MEM_MAP;
5c0e3066	231	ms->pageblock_flags = pageblock_bitmap;
29751f69 AW	232
	233	return 1;
	234	}
	235
04753278	236	unsigned long usemap_size(void)
5c0e3066 MG	237	{
	238	unsigned long size_bytes;
	239	size_bytes = roundup(SECTION_BLOCKFLAGS_BITS, 8) / 8;
	240	size_bytes = roundup(size_bytes, sizeof(unsigned long));
	241	return size_bytes;
	242	}
	243
	244	#ifdef CONFIG_MEMORY_HOTPLUG
	245	static unsigned long *__kmalloc_section_usemap(void)
	246	{
	247	return kmalloc(usemap_size(), GFP_KERNEL);
	248	}
	249	#endif /* CONFIG_MEMORY_HOTPLUG */
	250
a322f8ab	251	static unsigned long *__init sparse_early_usemap_alloc(unsigned long pnum)
5c0e3066	252	{
51674644	253	unsigned long *usemap;
5c0e3066 MG	254	struct mem_section *ms = __nr_to_section(pnum);
	255	int nid = sparse_early_nid(ms);
	256
51674644	257	usemap = alloc_bootmem_node(NODE_DATA(nid), usemap_size());
5c0e3066 MG	258	if (usemap)
	259	return usemap;
	260
	261	/* Stupid: suppress gcc warning for SPARSEMEM && !NUMA */
	262	nid = 0;
	263
d40cee24	264	printk(KERN_WARNING "%s: allocation failed\n", __func__);
5c0e3066 MG	265	return NULL;
	266	}
	267
8f6aac41	268	#ifndef CONFIG_SPARSEMEM_VMEMMAP
98f3cfc1	269	struct page __init *sparse_mem_map_populate(unsigned long pnum, int nid)
29751f69 AW	270	{
29751f69 AW	271	struct page *map;
29751f69 AW	272
	273	map = alloc_remap(nid, sizeof(struct page) * PAGES_PER_SECTION);
	274	if (map)
	275	return map;
	276
9d99217a YG	277	map = alloc_bootmem_pages_node(NODE_DATA(nid),
9d99217a YG	278	PAGE_ALIGN(sizeof(struct page) * PAGES_PER_SECTION));
8f6aac41 CL	279	return map;
	280	}
	281	#endif /* !CONFIG_SPARSEMEM_VMEMMAP */
	282
	283	struct page __init *sparse_early_mem_map_alloc(unsigned long pnum)
	284	{
	285	struct page *map;
	286	struct mem_section *ms = __nr_to_section(pnum);
	287	int nid = sparse_early_nid(ms);
	288
98f3cfc1	289	map = sparse_mem_map_populate(pnum, nid);
29751f69 AW	290	if (map)
	291	return map;
	292
8f6aac41	293	printk(KERN_ERR "%s: sparsemem memory map backing failed "
d40cee24	294	"some memory will not be available.\n", __func__);
802f192e	295	ms->section_mem_map = 0;
29751f69 AW	296	return NULL;
	297	}
	298
c2b91e2e YL	299	void __attribute__((weak)) __meminit vmemmap_populate_print_last(void)
	300	{
	301	}
193faea9 SR	302	/*
	303	* Allocate the accumulated non-linear sections, allocate a mem_map
	304	* for each and record the physical to section mapping.
	305	*/
	306	void __init sparse_init(void)
	307	{
	308	unsigned long pnum;
	309	struct page *map;
5c0e3066	310	unsigned long *usemap;
e123dd3f YL	311	unsigned long **usemap_map;
	312	int size;
	313
	314	/*
	315	* map is using big page (aka 2M in x86 64 bit)
	316	* usemap is less one page (aka 24 bytes)
	317	* so alloc 2M (with 2M align) and 24 bytes in turn will
	318	* make next 2M slip to one more 2M later.
	319	* then in big system, the memory will have a lot of holes...
	320	* here try to allocate 2M pages continously.
	321	*
	322	* powerpc need to call sparse_init_one_section right after each
	323	* sparse_early_mem_map_alloc, so allocate usemap_map at first.
	324	*/
	325	size = sizeof(unsigned long ) NR_MEM_SECTIONS;
	326	usemap_map = alloc_bootmem(size);
	327	if (!usemap_map)
	328	panic("can not allocate usemap_map\n");
193faea9 SR	329
193faea9 SR	330	for (pnum = 0; pnum < NR_MEM_SECTIONS; pnum++) {
540557b9	331	if (!present_section_nr(pnum))
193faea9	332	continue;
e123dd3f YL	333	usemap_map[pnum] = sparse_early_usemap_alloc(pnum);
e123dd3f YL	334	}
193faea9	335
e123dd3f YL	336	for (pnum = 0; pnum < NR_MEM_SECTIONS; pnum++) {
e123dd3f YL	337	if (!present_section_nr(pnum))
193faea9	338	continue;
5c0e3066	339
e123dd3f	340	usemap = usemap_map[pnum];
5c0e3066 MG	341	if (!usemap)
	342	continue;
	343
e123dd3f YL	344	map = sparse_early_mem_map_alloc(pnum);
	345	if (!map)
	346	continue;
	347
5c0e3066 MG	348	sparse_init_one_section(__nr_to_section(pnum), pnum, map,
5c0e3066 MG	349	usemap);
193faea9	350	}
e123dd3f	351
c2b91e2e YL	352	vmemmap_populate_print_last();
c2b91e2e YL	353
e123dd3f	354	free_bootmem(__pa(usemap_map), size);
193faea9 SR	355	}
	356
	357	#ifdef CONFIG_MEMORY_HOTPLUG
98f3cfc1 YG	358	#ifdef CONFIG_SPARSEMEM_VMEMMAP
	359	static inline struct page *kmalloc_section_memmap(unsigned long pnum, int nid,
	360	unsigned long nr_pages)
	361	{
	362	/* This will make the necessary allocations eventually. */
	363	return sparse_mem_map_populate(pnum, nid);
	364	}
	365	static void __kfree_section_memmap(struct page *memmap, unsigned long nr_pages)
	366	{
	367	return; /* XXX: Not implemented yet */
	368	}
0c0a4a51 YG	369	static void free_map_bootmem(struct page *page, unsigned long nr_pages)
	370	{
	371	}
98f3cfc1	372	#else
0b0acbec DH	373	static struct page *__kmalloc_section_memmap(unsigned long nr_pages)
	374	{
	375	struct page page, ret;
	376	unsigned long memmap_size = sizeof(struct page) * nr_pages;
	377
f2d0aa5b	378	page = alloc_pages(GFP_KERNEL\|__GFP_NOWARN, get_order(memmap_size));
0b0acbec DH	379	if (page)
	380	goto got_map_page;
	381
	382	ret = vmalloc(memmap_size);
	383	if (ret)
	384	goto got_map_ptr;
	385
	386	return NULL;
	387	got_map_page:
	388	ret = (struct page *)pfn_to_kaddr(page_to_pfn(page));
	389	got_map_ptr:
	390	memset(ret, 0, memmap_size);
	391
	392	return ret;
	393	}
	394
98f3cfc1 YG	395	static inline struct page *kmalloc_section_memmap(unsigned long pnum, int nid,
	396	unsigned long nr_pages)
	397	{
	398	return __kmalloc_section_memmap(nr_pages);
	399	}
	400
0b0acbec DH	401	static void __kfree_section_memmap(struct page *memmap, unsigned long nr_pages)
0b0acbec DH	402	{
9e2779fa	403	if (is_vmalloc_addr(memmap))
0b0acbec DH	404	vfree(memmap);
	405	else
	406	free_pages((unsigned long)memmap,
	407	get_order(sizeof(struct page) * nr_pages));
	408	}
0c0a4a51 YG	409
	410	static void free_map_bootmem(struct page *page, unsigned long nr_pages)
	411	{
	412	unsigned long maps_section_nr, removing_section_nr, i;
	413	int magic;
	414
	415	for (i = 0; i < nr_pages; i++, page++) {
	416	magic = atomic_read(&page->_mapcount);
	417
	418	BUG_ON(magic == NODE_INFO);
	419
	420	maps_section_nr = pfn_to_section_nr(page_to_pfn(page));
	421	removing_section_nr = page->private;
	422
	423	/*
	424	* When this function is called, the removing section is
	425	* logical offlined state. This means all pages are isolated
	426	* from page allocator. If removing section's memmap is placed
	427	* on the same section, it must not be freed.
	428	* If it is freed, page allocator may allocate it which will
	429	* be removed physically soon.
	430	*/
	431	if (maps_section_nr != removing_section_nr)
	432	put_page_bootmem(page);
	433	}
	434	}
98f3cfc1	435	#endif /* CONFIG_SPARSEMEM_VMEMMAP */
0b0acbec	436
ea01ea93 BP	437	static void free_section_usemap(struct page memmap, unsigned long usemap)
ea01ea93 BP	438	{
0c0a4a51 YG	439	struct page *usemap_page;
	440	unsigned long nr_pages;
	441
ea01ea93 BP	442	if (!usemap)
	443	return;
	444
0c0a4a51	445	usemap_page = virt_to_page(usemap);
ea01ea93 BP	446	/*
	447	* Check to see if allocation came from hot-plug-add
	448	*/
0c0a4a51	449	if (PageSlab(usemap_page)) {
ea01ea93 BP	450	kfree(usemap);
	451	if (memmap)
	452	__kfree_section_memmap(memmap, PAGES_PER_SECTION);
	453	return;
	454	}
	455
	456	/*
0c0a4a51 YG	457	* The usemap came from bootmem. This is packed with other usemaps
0c0a4a51 YG	458	* on the section which has pgdat at boot time. Just keep it as is now.
ea01ea93	459	*/
0c0a4a51 YG	460
	461	if (memmap) {
	462	struct page *memmap_page;
	463	memmap_page = virt_to_page(memmap);
	464
	465	nr_pages = PAGE_ALIGN(PAGES_PER_SECTION * sizeof(struct page))
	466	>> PAGE_SHIFT;
	467
	468	free_map_bootmem(memmap_page, nr_pages);
	469	}
ea01ea93 BP	470	}
ea01ea93 BP	471
29751f69 AW	472	/*
	473	* returns the number of sections whose mem_maps were properly
	474	* set. If this is <=0, then that means that the passed-in
	475	* map was not consumed and must be freed.
	476	*/
0b0acbec DH	477	int sparse_add_one_section(struct zone *zone, unsigned long start_pfn,
0b0acbec DH	478	int nr_pages)
29751f69	479	{
0b0acbec DH	480	unsigned long section_nr = pfn_to_section_nr(start_pfn);
	481	struct pglist_data *pgdat = zone->zone_pgdat;
	482	struct mem_section *ms;
	483	struct page *memmap;
5c0e3066	484	unsigned long *usemap;
0b0acbec DH	485	unsigned long flags;
0b0acbec DH	486	int ret;
29751f69	487
0b0acbec DH	488	/*
	489	* no locking for this, because it does its own
	490	* plus, it does a kmalloc
	491	*/
bbd06825 WC	492	ret = sparse_index_init(section_nr, pgdat->node_id);
	493	if (ret < 0 && ret != -EEXIST)
	494	return ret;
98f3cfc1	495	memmap = kmalloc_section_memmap(section_nr, pgdat->node_id, nr_pages);
bbd06825 WC	496	if (!memmap)
bbd06825 WC	497	return -ENOMEM;
5c0e3066	498	usemap = __kmalloc_section_usemap();
bbd06825 WC	499	if (!usemap) {
	500	__kfree_section_memmap(memmap, nr_pages);
	501	return -ENOMEM;
	502	}
0b0acbec DH	503
0b0acbec DH	504	pgdat_resize_lock(pgdat, &flags);
29751f69	505
0b0acbec DH	506	ms = __pfn_to_section(start_pfn);
	507	if (ms->section_mem_map & SECTION_MARKED_PRESENT) {
	508	ret = -EEXIST;
	509	goto out;
	510	}
5c0e3066	511
29751f69 AW	512	ms->section_mem_map \|= SECTION_MARKED_PRESENT;
29751f69 AW	513
5c0e3066	514	ret = sparse_init_one_section(ms, section_nr, memmap, usemap);
0b0acbec	515
0b0acbec DH	516	out:
0b0acbec DH	517	pgdat_resize_unlock(pgdat, &flags);
bbd06825 WC	518	if (ret <= 0) {
bbd06825 WC	519	kfree(usemap);
46a66eec	520	__kfree_section_memmap(memmap, nr_pages);
bbd06825	521	}
0b0acbec	522	return ret;
29751f69	523	}
ea01ea93 BP	524
	525	void sparse_remove_one_section(struct zone zone, struct mem_section ms)
	526	{
	527	struct page *memmap = NULL;
	528	unsigned long *usemap = NULL;
	529
	530	if (ms->section_mem_map) {
	531	usemap = ms->pageblock_flags;
	532	memmap = sparse_decode_mem_map(ms->section_mem_map,
	533	__section_nr(ms));
	534	ms->section_mem_map = 0;
	535	ms->pageblock_flags = NULL;
	536	}
	537
	538	free_section_usemap(memmap, usemap);
	539	}
a3142c8e	540	#endif