[linux.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 memory 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/memremap.h>
#include <linux/highmem.h>
#include <linux/slab.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 * __ref __earlyonly_bootmem_alloc(int node,
				unsigned long size,
				unsigned long align,
				unsigned long goal)
{
	return memblock_virt_alloc_try_nid(size, align, goal,
					    BOOTMEM_ALLOC_ACCESSIBLE, node);
}

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 | __GFP_REPEAT,
				get_order(size));
		else
			page = alloc_pages(
				GFP_KERNEL | __GFP_ZERO | __GFP_REPEAT,
				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 */
static void * __meminit 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;
}

static unsigned long __meminit vmem_altmap_next_pfn(struct vmem_altmap *altmap)
{
	return altmap->base_pfn + altmap->reserve + altmap->alloc
		+ altmap->align;
}

static unsigned long __meminit vmem_altmap_nr_free(struct vmem_altmap *altmap)
{
	unsigned long allocated = altmap->alloc + altmap->align;

	if (altmap->free > allocated)
		return altmap->free - allocated;
	return 0;
}

/**
 * vmem_altmap_alloc - allocate pages from the vmem_altmap reservation
 * @altmap - reserved page pool for the allocation
 * @nr_pfns - size (in pages) of the allocation
 *
 * Allocations are aligned to the size of the request
 */
static unsigned long __meminit vmem_altmap_alloc(struct vmem_altmap *altmap,
		unsigned long nr_pfns)
{
	unsigned long pfn = vmem_altmap_next_pfn(altmap);
	unsigned long nr_align;

	nr_align = 1UL << find_first_bit(&nr_pfns, BITS_PER_LONG);
	nr_align = ALIGN(pfn, nr_align) - pfn;

	if (nr_pfns + nr_align > vmem_altmap_nr_free(altmap))
		return ULONG_MAX;
	altmap->alloc += nr_pfns;
	altmap->align += nr_align;
	return pfn + nr_align;
}

static void * __meminit altmap_alloc_block_buf(unsigned long size,
		struct vmem_altmap *altmap)
{
	unsigned long pfn, nr_pfns;
	void *ptr;

	if (size & ~PAGE_MASK) {
		pr_warn_once("%s: allocations must be multiple of PAGE_SIZE (%ld)\n",
				__func__, size);
		return NULL;
	}

	nr_pfns = size >> PAGE_SHIFT;
	pfn = vmem_altmap_alloc(altmap, nr_pfns);
	if (pfn < ULONG_MAX)
		ptr = __va(__pfn_to_phys(pfn));
	else
		ptr = NULL;
	pr_debug("%s: pfn: %#lx alloc: %ld align: %ld nr: %#lx\n",
			__func__, pfn, altmap->alloc, altmap->align, nr_pfns);

	return ptr;
}

/* need to make sure size is all the same during early stage */
void * __meminit __vmemmap_alloc_block_buf(unsigned long size, int node,
		struct vmem_altmap *altmap)
{
	if (altmap)
		return altmap_alloc_block_buf(size, altmap);
	return alloc_block_buf(size, node);
}

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)
		pr_warn("[%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 = 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(p4d_t *p4d, unsigned long addr, int node)
{
	pud_t *pud = pud_offset(p4d, 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;
}

p4d_t * __meminit vmemmap_p4d_populate(pgd_t *pgd, unsigned long addr, int node)
{
	p4d_t *p4d = p4d_offset(pgd, addr);
	if (p4d_none(*p4d)) {
		void *p = vmemmap_alloc_block(PAGE_SIZE, node);
		if (!p)
			return NULL;
		p4d_populate(&init_mm, p4d, p);
	}
	return p4d;
}

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(unsigned long start,
					 unsigned long end, int node)
{
	unsigned long addr = start;
	pgd_t *pgd;
	p4d_t *p4d;
	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;
		p4d = vmemmap_p4d_populate(pgd, addr, node);
		if (!p4d)
			return -ENOMEM;
		pud = vmemmap_pud_populate(p4d, 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)
{
	unsigned long start;
	unsigned long end;
	struct page *map;

	map = pfn_to_page(pnum * PAGES_PER_SECTION);
	start = (unsigned long)map;
	end = (unsigned long)(map + PAGES_PER_SECTION);

	if (vmemmap_populate(start, end, nid))
		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);
		pr_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 */
		memblock_free_early(__pa(vmemmap_buf),
				    vmemmap_buf_end - vmemmap_buf);
		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
b595076a	12	* via TLBs. For those arches the virtual memory map is essentially
8f6aac41 CL	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>
4b94ffdc	23	#include <linux/memremap.h>
8f6aac41	24	#include <linux/highmem.h>
5a0e3ad6	25	#include <linux/slab.h>
8f6aac41 CL	26	#include <linux/spinlock.h>
8f6aac41 CL	27	#include <linux/vmalloc.h>
8bca44bb	28	#include <linux/sched.h>
8f6aac41 CL	29	#include <asm/dma.h>
	30	#include <asm/pgalloc.h>
	31	#include <asm/pgtable.h>
	32
	33	/*
	34	* Allocate a block of memory to be used to back the virtual memory map
	35	* or to back the page tables that are used to create the mapping.
	36	* Uses the main allocators if they are available, else bootmem.
	37	*/
e0dc3a53	38
bd721ea7	39	static void * __ref __earlyonly_bootmem_alloc(int node,
e0dc3a53 KH	40	unsigned long size,
	41	unsigned long align,
	42	unsigned long goal)
	43	{
bb016b84 SS	44	return memblock_virt_alloc_try_nid(size, align, goal,
bb016b84 SS	45	BOOTMEM_ALLOC_ACCESSIBLE, node);
e0dc3a53 KH	46	}
e0dc3a53 KH	47
9bdac914 YL	48	static void *vmemmap_buf;
9bdac914 YL	49	static void *vmemmap_buf_end;
e0dc3a53	50
8f6aac41 CL	51	void * __meminit vmemmap_alloc_block(unsigned long size, int node)
	52	{
	53	/* If the main allocator is up use that, fallback to bootmem. */
	54	if (slab_is_available()) {
f52407ce SL	55	struct page *page;
	56
	57	if (node_state(node, N_HIGH_MEMORY))
055e4fd9 BH	58	page = alloc_pages_node(
	59	node, GFP_KERNEL \| __GFP_ZERO \| __GFP_REPEAT,
	60	get_order(size));
f52407ce	61	else
055e4fd9 BH	62	page = alloc_pages(
055e4fd9 BH	63	GFP_KERNEL \| __GFP_ZERO \| __GFP_REPEAT,
f52407ce	64	get_order(size));
8f6aac41 CL	65	if (page)
	66	return page_address(page);
	67	return NULL;
	68	} else
e0dc3a53	69	return __earlyonly_bootmem_alloc(node, size, size,
8f6aac41 CL	70	__pa(MAX_DMA_ADDRESS));
	71	}
	72
9bdac914	73	/* need to make sure size is all the same during early stage */
4b94ffdc	74	static void * __meminit alloc_block_buf(unsigned long size, int node)
9bdac914 YL	75	{
	76	void *ptr;
	77
	78	if (!vmemmap_buf)
	79	return vmemmap_alloc_block(size, node);
	80
	81	/* take the from buf */
	82	ptr = (void *)ALIGN((unsigned long)vmemmap_buf, size);
	83	if (ptr + size > vmemmap_buf_end)
	84	return vmemmap_alloc_block(size, node);
	85
	86	vmemmap_buf = ptr + size;
	87
	88	return ptr;
	89	}
	90
4b94ffdc DW	91	static unsigned long __meminit vmem_altmap_next_pfn(struct vmem_altmap *altmap)
	92	{
	93	return altmap->base_pfn + altmap->reserve + altmap->alloc
	94	+ altmap->align;
	95	}
	96
	97	static unsigned long __meminit vmem_altmap_nr_free(struct vmem_altmap *altmap)
	98	{
	99	unsigned long allocated = altmap->alloc + altmap->align;
	100
	101	if (altmap->free > allocated)
	102	return altmap->free - allocated;
	103	return 0;
	104	}
	105
	106	/**
	107	* vmem_altmap_alloc - allocate pages from the vmem_altmap reservation
	108	* @altmap - reserved page pool for the allocation
	109	* @nr_pfns - size (in pages) of the allocation
	110	*
	111	* Allocations are aligned to the size of the request
	112	*/
	113	static unsigned long __meminit vmem_altmap_alloc(struct vmem_altmap *altmap,
	114	unsigned long nr_pfns)
	115	{
	116	unsigned long pfn = vmem_altmap_next_pfn(altmap);
	117	unsigned long nr_align;
	118
	119	nr_align = 1UL << find_first_bit(&nr_pfns, BITS_PER_LONG);
	120	nr_align = ALIGN(pfn, nr_align) - pfn;
	121
	122	if (nr_pfns + nr_align > vmem_altmap_nr_free(altmap))
	123	return ULONG_MAX;
	124	altmap->alloc += nr_pfns;
	125	altmap->align += nr_align;
	126	return pfn + nr_align;
	127	}
	128
	129	static void * __meminit altmap_alloc_block_buf(unsigned long size,
	130	struct vmem_altmap *altmap)
	131	{
	132	unsigned long pfn, nr_pfns;
	133	void *ptr;
	134
	135	if (size & ~PAGE_MASK) {
	136	pr_warn_once("%s: allocations must be multiple of PAGE_SIZE (%ld)\n",
	137	__func__, size);
	138	return NULL;
	139	}
	140
	141	nr_pfns = size >> PAGE_SHIFT;
	142	pfn = vmem_altmap_alloc(altmap, nr_pfns);
	143	if (pfn < ULONG_MAX)
	144	ptr = __va(__pfn_to_phys(pfn));
	145	else
	146	ptr = NULL;
	147	pr_debug("%s: pfn: %#lx alloc: %ld align: %ld nr: %#lx\n",
	148	__func__, pfn, altmap->alloc, altmap->align, nr_pfns);
	149
	150	return ptr;
	151	}
	152
	153	/* need to make sure size is all the same during early stage */
	154	void * __meminit __vmemmap_alloc_block_buf(unsigned long size, int node,
155	struct vmem_altmap *altmap)
156	{
157	if (altmap)
158	return altmap_alloc_block_buf(size, altmap);
159	return alloc_block_buf(size, node);
160	}
161
8f6aac41 CL	162	void __meminit vmemmap_verify(pte_t *pte, int node,
	163	unsigned long start, unsigned long end)
	164	{
	165	unsigned long pfn = pte_pfn(*pte);
	166	int actual_node = early_pfn_to_nid(pfn);
	167
b41ad14c	168	if (node_distance(actual_node, node) > LOCAL_DISTANCE)
1170532b JP	169	pr_warn("[%lx-%lx] potential offnode page_structs\n",
1170532b JP	170	start, end - 1);
8f6aac41 CL	171	}
8f6aac41 CL	172
29c71111	173	pte_t * __meminit vmemmap_pte_populate(pmd_t *pmd, unsigned long addr, int node)
8f6aac41	174	{
29c71111 AW	175	pte_t *pte = pte_offset_kernel(pmd, addr);
	176	if (pte_none(*pte)) {
	177	pte_t entry;
4b94ffdc	178	void *p = alloc_block_buf(PAGE_SIZE, node);
29c71111	179	if (!p)
9dce07f1	180	return NULL;
29c71111 AW	181	entry = pfn_pte(__pa(p) >> PAGE_SHIFT, PAGE_KERNEL);
	182	set_pte_at(&init_mm, addr, pte, entry);
	183	}
	184	return pte;
8f6aac41 CL	185	}
8f6aac41 CL	186
29c71111	187	pmd_t * __meminit vmemmap_pmd_populate(pud_t *pud, unsigned long addr, int node)
8f6aac41	188	{
29c71111 AW	189	pmd_t *pmd = pmd_offset(pud, addr);
	190	if (pmd_none(*pmd)) {
	191	void *p = vmemmap_alloc_block(PAGE_SIZE, node);
	192	if (!p)
9dce07f1	193	return NULL;
29c71111	194	pmd_populate_kernel(&init_mm, pmd, p);
8f6aac41	195	}
29c71111	196	return pmd;
8f6aac41	197	}
8f6aac41	198
c2febafc	199	pud_t * __meminit vmemmap_pud_populate(p4d_t *p4d, unsigned long addr, int node)
8f6aac41	200	{
c2febafc	201	pud_t *pud = pud_offset(p4d, addr);
29c71111 AW	202	if (pud_none(*pud)) {
	203	void *p = vmemmap_alloc_block(PAGE_SIZE, node);
	204	if (!p)
9dce07f1	205	return NULL;
29c71111 AW	206	pud_populate(&init_mm, pud, p);
	207	}
	208	return pud;
	209	}
8f6aac41	210
c2febafc KS	211	p4d_t * __meminit vmemmap_p4d_populate(pgd_t *pgd, unsigned long addr, int node)
	212	{
	213	p4d_t *p4d = p4d_offset(pgd, addr);
	214	if (p4d_none(*p4d)) {
	215	void *p = vmemmap_alloc_block(PAGE_SIZE, node);
	216	if (!p)
	217	return NULL;
	218	p4d_populate(&init_mm, p4d, p);
	219	}
	220	return p4d;
	221	}
	222
29c71111 AW	223	pgd_t * __meminit vmemmap_pgd_populate(unsigned long addr, int node)
	224	{
	225	pgd_t *pgd = pgd_offset_k(addr);
	226	if (pgd_none(*pgd)) {
	227	void *p = vmemmap_alloc_block(PAGE_SIZE, node);
	228	if (!p)
9dce07f1	229	return NULL;
29c71111	230	pgd_populate(&init_mm, pgd, p);
8f6aac41	231	}
29c71111	232	return pgd;
8f6aac41 CL	233	}
8f6aac41 CL	234
0aad818b JW	235	int __meminit vmemmap_populate_basepages(unsigned long start,
0aad818b JW	236	unsigned long end, int node)
8f6aac41	237	{
0aad818b	238	unsigned long addr = start;
29c71111	239	pgd_t *pgd;
c2febafc	240	p4d_t *p4d;
29c71111 AW	241	pud_t *pud;
	242	pmd_t *pmd;
	243	pte_t *pte;
8f6aac41	244
29c71111 AW	245	for (; addr < end; addr += PAGE_SIZE) {
	246	pgd = vmemmap_pgd_populate(addr, node);
	247	if (!pgd)
	248	return -ENOMEM;
c2febafc KS	249	p4d = vmemmap_p4d_populate(pgd, addr, node);
	250	if (!p4d)
	251	return -ENOMEM;
	252	pud = vmemmap_pud_populate(p4d, addr, node);
29c71111 AW	253	if (!pud)
	254	return -ENOMEM;
	255	pmd = vmemmap_pmd_populate(pud, addr, node);
	256	if (!pmd)
	257	return -ENOMEM;
	258	pte = vmemmap_pte_populate(pmd, addr, node);
	259	if (!pte)
	260	return -ENOMEM;
	261	vmemmap_verify(pte, node, addr, addr + PAGE_SIZE);
8f6aac41	262	}
29c71111 AW	263
29c71111 AW	264	return 0;
8f6aac41	265	}
8f6aac41	266
98f3cfc1	267	struct page * __meminit sparse_mem_map_populate(unsigned long pnum, int nid)
8f6aac41	268	{
0aad818b JW	269	unsigned long start;
	270	unsigned long end;
	271	struct page *map;
	272
	273	map = pfn_to_page(pnum * PAGES_PER_SECTION);
	274	start = (unsigned long)map;
	275	end = (unsigned long)(map + PAGES_PER_SECTION);
	276
	277	if (vmemmap_populate(start, end, nid))
8f6aac41 CL	278	return NULL;
	279
	280	return map;
	281	}
9bdac914 YL	282
	283	void __init sparse_mem_maps_populate_node(struct page **map_map,
	284	unsigned long pnum_begin,
	285	unsigned long pnum_end,
	286	unsigned long map_count, int nodeid)
	287	{
	288	unsigned long pnum;
	289	unsigned long size = sizeof(struct page) * PAGES_PER_SECTION;
	290	void *vmemmap_buf_start;
	291
	292	size = ALIGN(size, PMD_SIZE);
	293	vmemmap_buf_start = __earlyonly_bootmem_alloc(nodeid, size * map_count,
	294	PMD_SIZE, __pa(MAX_DMA_ADDRESS));
	295
	296	if (vmemmap_buf_start) {
	297	vmemmap_buf = vmemmap_buf_start;
	298	vmemmap_buf_end = vmemmap_buf_start + size * map_count;
	299	}
	300
	301	for (pnum = pnum_begin; pnum < pnum_end; pnum++) {
	302	struct mem_section *ms;
	303
	304	if (!present_section_nr(pnum))
	305	continue;
	306
	307	map_map[pnum] = sparse_mem_map_populate(pnum, nodeid);
	308	if (map_map[pnum])
	309	continue;
	310	ms = __nr_to_section(pnum);
1170532b	311	pr_err("%s: sparsemem memory map backing failed some memory will not be available\n",
756a025f	312	__func__);
9bdac914 YL	313	ms->section_mem_map = 0;
	314	}
	315
	316	if (vmemmap_buf_start) {
	317	/* need to free left buf */
bb016b84 SS	318	memblock_free_early(__pa(vmemmap_buf),
bb016b84 SS	319	vmemmap_buf_end - vmemmap_buf);
9bdac914 YL	320	vmemmap_buf = NULL;
	321	vmemmap_buf_end = NULL;
	322	}
	323	}