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

/*
 * Virtual Memory Map support
 *
 * (C) 2007 sgi. Christoph Lameter.
 *
 * Virtual memory maps allow VM primitives pfn_to_page, page_to_pfn,
 * virt_to_page, page_address() to be implemented as a base offset
 * calculation without memory access.
 *
 * However, virtual mappings need a page table and TLBs. Many Linux
 * architectures already map their physical space using 1-1 mappings
 * via TLBs. For those arches the virtual memmory map is essentially
 * for free if we use the same page size as the 1-1 mappings. In that
 * case the overhead consists of a few additional pages that are
 * allocated to create a view of memory for vmemmap.
 *
 * The architecture is expected to provide a vmemmap_populate() function
 * to instantiate the mapping.
 */
#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 <linux/sched.h>
#include <asm/dma.h>
#include <asm/pgalloc.h>
#include <asm/pgtable.h>

/*
 * Allocate a block of memory to be used to back the virtual memory map
 * or to back the page tables that are used to create the mapping.
 * Uses the main allocators if they are available, else bootmem.
 */

static void * __init_refok __earlyonly_bootmem_alloc(int node,
				unsigned long size,
				unsigned long align,
				unsigned long goal)
{
	return __alloc_bootmem_node_high(NODE_DATA(node), size, align, goal);
}

static void *vmemmap_buf;
static void *vmemmap_buf_end;

void * __meminit vmemmap_alloc_block(unsigned long size, int node)
{
	/* If the main allocator is up use that, fallback to bootmem. */
	if (slab_is_available()) {
		struct page *page;

		if (node_state(node, N_HIGH_MEMORY))
			page = alloc_pages_node(node,
				GFP_KERNEL | __GFP_ZERO, get_order(size));
		else
			page = alloc_pages(GFP_KERNEL | __GFP_ZERO,
				get_order(size));
		if (page)
			return page_address(page);
		return NULL;
	} else
		return __earlyonly_bootmem_alloc(node, size, size,
				__pa(MAX_DMA_ADDRESS));
}

/* need to make sure size is all the same during early stage */
void * __meminit vmemmap_alloc_block_buf(unsigned long size, int node)
{
	void *ptr;

	if (!vmemmap_buf)
		return vmemmap_alloc_block(size, node);

	/* take the from buf */
	ptr = (void *)ALIGN((unsigned long)vmemmap_buf, size);
	if (ptr + size > vmemmap_buf_end)
		return vmemmap_alloc_block(size, node);

	vmemmap_buf = ptr + size;

	return ptr;
}

void __meminit vmemmap_verify(pte_t *pte, int node,
				unsigned long start, unsigned long end)
{
	unsigned long pfn = pte_pfn(*pte);
	int actual_node = early_pfn_to_nid(pfn);

	if (node_distance(actual_node, node) > LOCAL_DISTANCE)
		printk(KERN_WARNING "[%lx-%lx] potential offnode "
			"page_structs\n", start, end - 1);
}

pte_t * __meminit vmemmap_pte_populate(pmd_t *pmd, unsigned long addr, int node)
{
	pte_t *pte = pte_offset_kernel(pmd, addr);
	if (pte_none(*pte)) {
		pte_t entry;
		void *p = vmemmap_alloc_block_buf(PAGE_SIZE, node);
		if (!p)
			return NULL;
		entry = pfn_pte(__pa(p) >> PAGE_SHIFT, PAGE_KERNEL);
		set_pte_at(&init_mm, addr, pte, entry);
	}
	return pte;
}

pmd_t * __meminit vmemmap_pmd_populate(pud_t *pud, unsigned long addr, int node)
{
	pmd_t *pmd = pmd_offset(pud, addr);
	if (pmd_none(*pmd)) {
		void *p = vmemmap_alloc_block(PAGE_SIZE, node);
		if (!p)
			return NULL;
		pmd_populate_kernel(&init_mm, pmd, p);
	}
	return pmd;
}

pud_t * __meminit vmemmap_pud_populate(pgd_t *pgd, unsigned long addr, int node)
{
	pud_t *pud = pud_offset(pgd, addr);
	if (pud_none(*pud)) {
		void *p = vmemmap_alloc_block(PAGE_SIZE, node);
		if (!p)
			return NULL;
		pud_populate(&init_mm, pud, p);
	}
	return pud;
}

pgd_t * __meminit vmemmap_pgd_populate(unsigned long addr, int node)
{
	pgd_t *pgd = pgd_offset_k(addr);
	if (pgd_none(*pgd)) {
		void *p = vmemmap_alloc_block(PAGE_SIZE, node);
		if (!p)
			return NULL;
		pgd_populate(&init_mm, pgd, p);
	}
	return pgd;
}

int __meminit vmemmap_populate_basepages(struct page *start_page,
						unsigned long size, int node)
{
	unsigned long addr = (unsigned long)start_page;
	unsigned long end = (unsigned long)(start_page + size);
	pgd_t *pgd;
	pud_t *pud;
	pmd_t *pmd;
	pte_t *pte;

	for (; addr < end; addr += PAGE_SIZE) {
		pgd = vmemmap_pgd_populate(addr, node);
		if (!pgd)
			return -ENOMEM;
		pud = vmemmap_pud_populate(pgd, addr, node);
		if (!pud)
			return -ENOMEM;
		pmd = vmemmap_pmd_populate(pud, addr, node);
		if (!pmd)
			return -ENOMEM;
		pte = vmemmap_pte_populate(pmd, addr, node);
		if (!pte)
			return -ENOMEM;
		vmemmap_verify(pte, node, addr, addr + PAGE_SIZE);
	}

	return 0;
}

struct page * __meminit sparse_mem_map_populate(unsigned long pnum, int nid)
{
	struct page *map = pfn_to_page(pnum * PAGES_PER_SECTION);
	int error = vmemmap_populate(map, PAGES_PER_SECTION, nid);
	if (error)
		return NULL;

	return map;
}

void __init sparse_mem_maps_populate_node(struct page **map_map,
					  unsigned long pnum_begin,
					  unsigned long pnum_end,
					  unsigned long map_count, int nodeid)
{
	unsigned long pnum;
	unsigned long size = sizeof(struct page) * PAGES_PER_SECTION;
	void *vmemmap_buf_start;

	size = ALIGN(size, PMD_SIZE);
	vmemmap_buf_start = __earlyonly_bootmem_alloc(nodeid, size * map_count,
			 PMD_SIZE, __pa(MAX_DMA_ADDRESS));

	if (vmemmap_buf_start) {
		vmemmap_buf = vmemmap_buf_start;
		vmemmap_buf_end = vmemmap_buf_start + size * map_count;
	}

	for (pnum = pnum_begin; pnum < pnum_end; pnum++) {
		struct mem_section *ms;

		if (!present_section_nr(pnum))
			continue;

		map_map[pnum] = sparse_mem_map_populate(pnum, nodeid);
		if (map_map[pnum])
			continue;
		ms = __nr_to_section(pnum);
		printk(KERN_ERR "%s: sparsemem memory map backing failed "
			"some memory will not be available.\n", __func__);
		ms->section_mem_map = 0;
	}

	if (vmemmap_buf_start) {
		/* need to free left buf */
#ifdef CONFIG_NO_BOOTMEM
		free_early(__pa(vmemmap_buf_start), __pa(vmemmap_buf_end));
		if (vmemmap_buf_start < vmemmap_buf) {
			char name[15];

			snprintf(name, sizeof(name), "MEMMAP %d", nodeid);
			reserve_early_without_check(__pa(vmemmap_buf_start),
						    __pa(vmemmap_buf), name);
		}
#else
		free_bootmem(__pa(vmemmap_buf), vmemmap_buf_end - vmemmap_buf);
#endif
		vmemmap_buf = NULL;
		vmemmap_buf_end = NULL;
	}
}
Commit	Line	Data
8f6aac41 CL	1	/*
	2	* Virtual Memory Map support
	3	*
cde53535	4	* (C) 2007 sgi. Christoph Lameter.
8f6aac41 CL	5	*
	6	* Virtual memory maps allow VM primitives pfn_to_page, page_to_pfn,
	7	* virt_to_page, page_address() to be implemented as a base offset
	8	* calculation without memory access.
	9	*
	10	* However, virtual mappings need a page table and TLBs. Many Linux
	11	* architectures already map their physical space using 1-1 mappings
	12	* via TLBs. For those arches the virtual memmory map is essentially
	13	* for free if we use the same page size as the 1-1 mappings. In that
	14	* case the overhead consists of a few additional pages that are
	15	* allocated to create a view of memory for vmemmap.
	16	*
29c71111 AW	17	* The architecture is expected to provide a vmemmap_populate() function
29c71111 AW	18	* to instantiate the mapping.
8f6aac41 CL	19	*/
	20	#include <linux/mm.h>
	21	#include <linux/mmzone.h>
	22	#include <linux/bootmem.h>
	23	#include <linux/highmem.h>
	24	#include <linux/module.h>
	25	#include <linux/spinlock.h>
	26	#include <linux/vmalloc.h>
8bca44bb	27	#include <linux/sched.h>
8f6aac41 CL	28	#include <asm/dma.h>
	29	#include <asm/pgalloc.h>
	30	#include <asm/pgtable.h>
	31
	32	/*
	33	* Allocate a block of memory to be used to back the virtual memory map
	34	* or to back the page tables that are used to create the mapping.
	35	* Uses the main allocators if they are available, else bootmem.
	36	*/
e0dc3a53 KH	37
	38	static void * __init_refok __earlyonly_bootmem_alloc(int node,
	39	unsigned long size,
	40	unsigned long align,
	41	unsigned long goal)
	42	{
08677214	43	return __alloc_bootmem_node_high(NODE_DATA(node), size, align, goal);
e0dc3a53 KH	44	}
e0dc3a53 KH	45
9bdac914 YL	46	static void *vmemmap_buf;
9bdac914 YL	47	static void *vmemmap_buf_end;
e0dc3a53	48
8f6aac41 CL	49	void * __meminit vmemmap_alloc_block(unsigned long size, int node)
	50	{
	51	/* If the main allocator is up use that, fallback to bootmem. */
	52	if (slab_is_available()) {
f52407ce SL	53	struct page *page;
	54
	55	if (node_state(node, N_HIGH_MEMORY))
	56	page = alloc_pages_node(node,
8f6aac41	57	GFP_KERNEL \| __GFP_ZERO, get_order(size));
f52407ce SL	58	else
	59	page = alloc_pages(GFP_KERNEL \| __GFP_ZERO,
	60	get_order(size));
8f6aac41 CL	61	if (page)
	62	return page_address(page);
	63	return NULL;
	64	} else
e0dc3a53	65	return __earlyonly_bootmem_alloc(node, size, size,
8f6aac41 CL	66	__pa(MAX_DMA_ADDRESS));
	67	}
	68
9bdac914 YL	69	/* need to make sure size is all the same during early stage */
	70	void * __meminit vmemmap_alloc_block_buf(unsigned long size, int node)
	71	{
	72	void *ptr;
	73
	74	if (!vmemmap_buf)
	75	return vmemmap_alloc_block(size, node);
	76
	77	/* take the from buf */
	78	ptr = (void *)ALIGN((unsigned long)vmemmap_buf, size);
	79	if (ptr + size > vmemmap_buf_end)
	80	return vmemmap_alloc_block(size, node);
	81
	82	vmemmap_buf = ptr + size;
	83
	84	return ptr;
	85	}
	86
8f6aac41 CL	87	void __meminit vmemmap_verify(pte_t *pte, int node,
	88	unsigned long start, unsigned long end)
	89	{
	90	unsigned long pfn = pte_pfn(*pte);
	91	int actual_node = early_pfn_to_nid(pfn);
	92
b41ad14c	93	if (node_distance(actual_node, node) > LOCAL_DISTANCE)
8f6aac41 CL	94	printk(KERN_WARNING "[%lx-%lx] potential offnode "
	95	"page_structs\n", start, end - 1);
	96	}
	97
29c71111	98	pte_t * __meminit vmemmap_pte_populate(pmd_t *pmd, unsigned long addr, int node)
8f6aac41	99	{
29c71111 AW	100	pte_t *pte = pte_offset_kernel(pmd, addr);
	101	if (pte_none(*pte)) {
	102	pte_t entry;
9bdac914	103	void *p = vmemmap_alloc_block_buf(PAGE_SIZE, node);
29c71111	104	if (!p)
9dce07f1	105	return NULL;
29c71111 AW	106	entry = pfn_pte(__pa(p) >> PAGE_SHIFT, PAGE_KERNEL);
	107	set_pte_at(&init_mm, addr, pte, entry);
	108	}
	109	return pte;
8f6aac41 CL	110	}
8f6aac41 CL	111
29c71111	112	pmd_t * __meminit vmemmap_pmd_populate(pud_t *pud, unsigned long addr, int node)
8f6aac41	113	{
29c71111 AW	114	pmd_t *pmd = pmd_offset(pud, addr);
	115	if (pmd_none(*pmd)) {
	116	void *p = vmemmap_alloc_block(PAGE_SIZE, node);
	117	if (!p)
9dce07f1	118	return NULL;
29c71111	119	pmd_populate_kernel(&init_mm, pmd, p);
8f6aac41	120	}
29c71111	121	return pmd;
8f6aac41	122	}
8f6aac41	123
29c71111	124	pud_t * __meminit vmemmap_pud_populate(pgd_t *pgd, unsigned long addr, int node)
8f6aac41	125	{
29c71111 AW	126	pud_t *pud = pud_offset(pgd, addr);
	127	if (pud_none(*pud)) {
	128	void *p = vmemmap_alloc_block(PAGE_SIZE, node);
	129	if (!p)
9dce07f1	130	return NULL;
29c71111 AW	131	pud_populate(&init_mm, pud, p);
	132	}
	133	return pud;
	134	}
8f6aac41	135
29c71111 AW	136	pgd_t * __meminit vmemmap_pgd_populate(unsigned long addr, int node)
	137	{
	138	pgd_t *pgd = pgd_offset_k(addr);
	139	if (pgd_none(*pgd)) {
	140	void *p = vmemmap_alloc_block(PAGE_SIZE, node);
	141	if (!p)
9dce07f1	142	return NULL;
29c71111	143	pgd_populate(&init_mm, pgd, p);
8f6aac41	144	}
29c71111	145	return pgd;
8f6aac41 CL	146	}
8f6aac41 CL	147
29c71111 AW	148	int __meminit vmemmap_populate_basepages(struct page *start_page,
29c71111 AW	149	unsigned long size, int node)
8f6aac41	150	{
8f6aac41	151	unsigned long addr = (unsigned long)start_page;
29c71111 AW	152	unsigned long end = (unsigned long)(start_page + size);
	153	pgd_t *pgd;
	154	pud_t *pud;
	155	pmd_t *pmd;
	156	pte_t *pte;
8f6aac41	157
29c71111 AW	158	for (; addr < end; addr += PAGE_SIZE) {
	159	pgd = vmemmap_pgd_populate(addr, node);
	160	if (!pgd)
	161	return -ENOMEM;
	162	pud = vmemmap_pud_populate(pgd, addr, node);
	163	if (!pud)
	164	return -ENOMEM;
	165	pmd = vmemmap_pmd_populate(pud, addr, node);
	166	if (!pmd)
	167	return -ENOMEM;
	168	pte = vmemmap_pte_populate(pmd, addr, node);
	169	if (!pte)
	170	return -ENOMEM;
	171	vmemmap_verify(pte, node, addr, addr + PAGE_SIZE);
8f6aac41	172	}
29c71111 AW	173
29c71111 AW	174	return 0;
8f6aac41	175	}
8f6aac41	176
98f3cfc1	177	struct page * __meminit sparse_mem_map_populate(unsigned long pnum, int nid)
8f6aac41 CL	178	{
	179	struct page map = pfn_to_page(pnum PAGES_PER_SECTION);
	180	int error = vmemmap_populate(map, PAGES_PER_SECTION, nid);
	181	if (error)
	182	return NULL;
	183
	184	return map;
	185	}
9bdac914 YL	186
	187	void __init sparse_mem_maps_populate_node(struct page **map_map,
	188	unsigned long pnum_begin,
	189	unsigned long pnum_end,
	190	unsigned long map_count, int nodeid)
	191	{
	192	unsigned long pnum;
	193	unsigned long size = sizeof(struct page) * PAGES_PER_SECTION;
	194	void *vmemmap_buf_start;
	195
	196	size = ALIGN(size, PMD_SIZE);
	197	vmemmap_buf_start = __earlyonly_bootmem_alloc(nodeid, size * map_count,
	198	PMD_SIZE, __pa(MAX_DMA_ADDRESS));
	199
	200	if (vmemmap_buf_start) {
	201	vmemmap_buf = vmemmap_buf_start;
	202	vmemmap_buf_end = vmemmap_buf_start + size * map_count;
	203	}
	204
	205	for (pnum = pnum_begin; pnum < pnum_end; pnum++) {
	206	struct mem_section *ms;
	207
	208	if (!present_section_nr(pnum))
	209	continue;
	210
	211	map_map[pnum] = sparse_mem_map_populate(pnum, nodeid);
	212	if (map_map[pnum])
	213	continue;
	214	ms = __nr_to_section(pnum);
	215	printk(KERN_ERR "%s: sparsemem memory map backing failed "
	216	"some memory will not be available.\n", __func__);
	217	ms->section_mem_map = 0;
	218	}
	219
	220	if (vmemmap_buf_start) {
	221	/* need to free left buf */
	222	#ifdef CONFIG_NO_BOOTMEM
	223	free_early(__pa(vmemmap_buf_start), __pa(vmemmap_buf_end));
	224	if (vmemmap_buf_start < vmemmap_buf) {
	225	char name[15];
	226
	227	snprintf(name, sizeof(name), "MEMMAP %d", nodeid);
	228	reserve_early_without_check(__pa(vmemmap_buf_start),
	229	__pa(vmemmap_buf), name);
	230	}
	231	#else
	232	free_bootmem(__pa(vmemmap_buf), vmemmap_buf_end - vmemmap_buf);
	233	#endif
	234	vmemmap_buf = NULL;
	235	vmemmap_buf_end = NULL;
	236	}
	237	}