[linux-2.6-block.git] / mm / sparse-vmemmap.c

// SPDX-License-Identifier: GPL-2.0
/*
 * 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_raw(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()) {
		gfp_t gfp_mask = GFP_KERNEL|__GFP_RETRY_MAYFAIL|__GFP_NOWARN;
		int order = get_order(size);
		static bool warned;
		struct page *page;

		page = alloc_pages_node(node, gfp_mask, order);
		if (page)
			return page_address(page);

		if (!warned) {
			warn_alloc(gfp_mask & ~__GFP_NOWARN, NULL,
				   "vmemmap alloc failure: order:%u", order);
			warned = true;
		}
		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;
}

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;
}

/**
 * altmap_alloc_block_buf - allocate pages from the device page map
 * @altmap:	device page map
 * @size:	size (in bytes) of the allocation
 *
 * Allocations are aligned to the size of the request.
 */
void * __meminit altmap_alloc_block_buf(unsigned long size,
		struct vmem_altmap *altmap)
{
	unsigned long pfn, nr_pfns, nr_align;

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

	pfn = vmem_altmap_next_pfn(altmap);
	nr_pfns = size >> PAGE_SHIFT;
	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 NULL;

	altmap->alloc += nr_pfns;
	altmap->align += nr_align;
	pfn += nr_align;

	pr_debug("%s: pfn: %#lx alloc: %ld align: %ld nr: %#lx\n",
			__func__, pfn, altmap->alloc, altmap->align, nr_pfns);
	return __va(__pfn_to_phys(pfn));
}

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 = 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;
}

static void * __meminit vmemmap_alloc_block_zero(unsigned long size, int node)
{
	void *p = vmemmap_alloc_block(size, node);

	if (!p)
		return NULL;
	memset(p, 0, size);

	return p;
}

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_zero(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_zero(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_zero(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_zero(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,
		struct vmem_altmap *altmap)
{
	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, altmap))
		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, NULL);
		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
b2441318	1	// SPDX-License-Identifier: GPL-2.0
8f6aac41 CL	2	/*
	3	* Virtual Memory Map support
	4	*
cde53535	5	* (C) 2007 sgi. Christoph Lameter.
8f6aac41 CL	6	*
	7	* Virtual memory maps allow VM primitives pfn_to_page, page_to_pfn,
	8	* virt_to_page, page_address() to be implemented as a base offset
	9	* calculation without memory access.
	10	*
	11	* However, virtual mappings need a page table and TLBs. Many Linux
	12	* architectures already map their physical space using 1-1 mappings
b595076a	13	* via TLBs. For those arches the virtual memory map is essentially
8f6aac41 CL	14	* for free if we use the same page size as the 1-1 mappings. In that
	15	* case the overhead consists of a few additional pages that are
	16	* allocated to create a view of memory for vmemmap.
	17	*
29c71111 AW	18	* The architecture is expected to provide a vmemmap_populate() function
29c71111 AW	19	* to instantiate the mapping.
8f6aac41 CL	20	*/
	21	#include <linux/mm.h>
	22	#include <linux/mmzone.h>
	23	#include <linux/bootmem.h>
4b94ffdc	24	#include <linux/memremap.h>
8f6aac41	25	#include <linux/highmem.h>
5a0e3ad6	26	#include <linux/slab.h>
8f6aac41 CL	27	#include <linux/spinlock.h>
8f6aac41 CL	28	#include <linux/vmalloc.h>
8bca44bb	29	#include <linux/sched.h>
8f6aac41 CL	30	#include <asm/dma.h>
	31	#include <asm/pgalloc.h>
	32	#include <asm/pgtable.h>
	33
	34	/*
	35	* Allocate a block of memory to be used to back the virtual memory map
	36	* or to back the page tables that are used to create the mapping.
	37	* Uses the main allocators if they are available, else bootmem.
	38	*/
e0dc3a53	39
bd721ea7	40	static void * __ref __earlyonly_bootmem_alloc(int node,
e0dc3a53 KH	41	unsigned long size,
	42	unsigned long align,
	43	unsigned long goal)
	44	{
f7f99100	45	return memblock_virt_alloc_try_nid_raw(size, align, goal,
bb016b84	46	BOOTMEM_ALLOC_ACCESSIBLE, node);
e0dc3a53 KH	47	}
e0dc3a53 KH	48
9bdac914 YL	49	static void *vmemmap_buf;
9bdac914 YL	50	static void *vmemmap_buf_end;
e0dc3a53	51
8f6aac41 CL	52	void * __meminit vmemmap_alloc_block(unsigned long size, int node)
	53	{
	54	/* If the main allocator is up use that, fallback to bootmem. */
	55	if (slab_is_available()) {
fcdaf842 MH	56	gfp_t gfp_mask = GFP_KERNEL\|__GFP_RETRY_MAYFAIL\|__GFP_NOWARN;
	57	int order = get_order(size);
	58	static bool warned;
f52407ce SL	59	struct page *page;
f52407ce SL	60
fcdaf842	61	page = alloc_pages_node(node, gfp_mask, order);
8f6aac41 CL	62	if (page)
8f6aac41 CL	63	return page_address(page);
fcdaf842 MH	64
	65	if (!warned) {
	66	warn_alloc(gfp_mask & ~__GFP_NOWARN, NULL,
	67	"vmemmap alloc failure: order:%u", order);
	68	warned = true;
	69	}
8f6aac41 CL	70	return NULL;
8f6aac41 CL	71	} else
e0dc3a53	72	return __earlyonly_bootmem_alloc(node, size, size,
8f6aac41 CL	73	__pa(MAX_DMA_ADDRESS));
	74	}
	75
9bdac914	76	/* need to make sure size is all the same during early stage */
a8fc357b	77	void * __meminit vmemmap_alloc_block_buf(unsigned long size, int node)
9bdac914 YL	78	{
	79	void *ptr;
	80
	81	if (!vmemmap_buf)
	82	return vmemmap_alloc_block(size, node);
	83
	84	/* take the from buf */
	85	ptr = (void *)ALIGN((unsigned long)vmemmap_buf, size);
	86	if (ptr + size > vmemmap_buf_end)
	87	return vmemmap_alloc_block(size, node);
	88
	89	vmemmap_buf = ptr + size;
	90
	91	return ptr;
	92	}
	93
4b94ffdc DW	94	static unsigned long __meminit vmem_altmap_next_pfn(struct vmem_altmap *altmap)
	95	{
	96	return altmap->base_pfn + altmap->reserve + altmap->alloc
	97	+ altmap->align;
	98	}
	99
	100	static unsigned long __meminit vmem_altmap_nr_free(struct vmem_altmap *altmap)
	101	{
	102	unsigned long allocated = altmap->alloc + altmap->align;
	103
	104	if (altmap->free > allocated)
	105	return altmap->free - allocated;
	106	return 0;
	107	}
	108
	109	/**
eb804533 CH	110	* altmap_alloc_block_buf - allocate pages from the device page map
	111	* @altmap: device page map
	112	* @size: size (in bytes) of the allocation
4b94ffdc	113	*
eb804533	114	* Allocations are aligned to the size of the request.
4b94ffdc	115	*/
a8fc357b	116	void * __meminit altmap_alloc_block_buf(unsigned long size,
4b94ffdc DW	117	struct vmem_altmap *altmap)
4b94ffdc DW	118	{
eb804533	119	unsigned long pfn, nr_pfns, nr_align;
4b94ffdc DW	120
	121	if (size & ~PAGE_MASK) {
	122	pr_warn_once("%s: allocations must be multiple of PAGE_SIZE (%ld)\n",
	123	__func__, size);
	124	return NULL;
	125	}
	126
eb804533	127	pfn = vmem_altmap_next_pfn(altmap);
4b94ffdc	128	nr_pfns = size >> PAGE_SHIFT;
eb804533 CH	129	nr_align = 1UL << find_first_bit(&nr_pfns, BITS_PER_LONG);
	130	nr_align = ALIGN(pfn, nr_align) - pfn;
	131	if (nr_pfns + nr_align > vmem_altmap_nr_free(altmap))
	132	return NULL;
	133
	134	altmap->alloc += nr_pfns;
	135	altmap->align += nr_align;
	136	pfn += nr_align;
	137
4b94ffdc DW	138	pr_debug("%s: pfn: %#lx alloc: %ld align: %ld nr: %#lx\n",
4b94ffdc DW	139	__func__, pfn, altmap->alloc, altmap->align, nr_pfns);
eb804533	140	return __va(__pfn_to_phys(pfn));
4b94ffdc DW	141	}
4b94ffdc DW	142
8f6aac41 CL	143	void __meminit vmemmap_verify(pte_t *pte, int node,
	144	unsigned long start, unsigned long end)
	145	{
	146	unsigned long pfn = pte_pfn(*pte);
	147	int actual_node = early_pfn_to_nid(pfn);
	148
b41ad14c	149	if (node_distance(actual_node, node) > LOCAL_DISTANCE)
1170532b JP	150	pr_warn("[%lx-%lx] potential offnode page_structs\n",
1170532b JP	151	start, end - 1);
8f6aac41 CL	152	}
8f6aac41 CL	153
29c71111	154	pte_t * __meminit vmemmap_pte_populate(pmd_t *pmd, unsigned long addr, int node)
8f6aac41	155	{
29c71111 AW	156	pte_t *pte = pte_offset_kernel(pmd, addr);
	157	if (pte_none(*pte)) {
	158	pte_t entry;
a8fc357b	159	void *p = vmemmap_alloc_block_buf(PAGE_SIZE, node);
29c71111	160	if (!p)
9dce07f1	161	return NULL;
29c71111 AW	162	entry = pfn_pte(__pa(p) >> PAGE_SHIFT, PAGE_KERNEL);
	163	set_pte_at(&init_mm, addr, pte, entry);
	164	}
	165	return pte;
8f6aac41 CL	166	}
8f6aac41 CL	167
f7f99100 PT	168	static void * __meminit vmemmap_alloc_block_zero(unsigned long size, int node)
	169	{
	170	void *p = vmemmap_alloc_block(size, node);
	171
	172	if (!p)
	173	return NULL;
	174	memset(p, 0, size);
	175
	176	return p;
	177	}
	178
29c71111	179	pmd_t * __meminit vmemmap_pmd_populate(pud_t *pud, unsigned long addr, int node)
8f6aac41	180	{
29c71111 AW	181	pmd_t *pmd = pmd_offset(pud, addr);
29c71111 AW	182	if (pmd_none(*pmd)) {
f7f99100	183	void *p = vmemmap_alloc_block_zero(PAGE_SIZE, node);
29c71111	184	if (!p)
9dce07f1	185	return NULL;
29c71111	186	pmd_populate_kernel(&init_mm, pmd, p);
8f6aac41	187	}
29c71111	188	return pmd;
8f6aac41	189	}
8f6aac41	190
c2febafc	191	pud_t * __meminit vmemmap_pud_populate(p4d_t *p4d, unsigned long addr, int node)
8f6aac41	192	{
c2febafc	193	pud_t *pud = pud_offset(p4d, addr);
29c71111	194	if (pud_none(*pud)) {
f7f99100	195	void *p = vmemmap_alloc_block_zero(PAGE_SIZE, node);
29c71111	196	if (!p)
9dce07f1	197	return NULL;
29c71111 AW	198	pud_populate(&init_mm, pud, p);
	199	}
	200	return pud;
	201	}
8f6aac41	202
c2febafc KS	203	p4d_t * __meminit vmemmap_p4d_populate(pgd_t *pgd, unsigned long addr, int node)
	204	{
	205	p4d_t *p4d = p4d_offset(pgd, addr);
	206	if (p4d_none(*p4d)) {
f7f99100	207	void *p = vmemmap_alloc_block_zero(PAGE_SIZE, node);
c2febafc KS	208	if (!p)
	209	return NULL;
	210	p4d_populate(&init_mm, p4d, p);
	211	}
	212	return p4d;
	213	}
	214
29c71111 AW	215	pgd_t * __meminit vmemmap_pgd_populate(unsigned long addr, int node)
	216	{
	217	pgd_t *pgd = pgd_offset_k(addr);
	218	if (pgd_none(*pgd)) {
f7f99100	219	void *p = vmemmap_alloc_block_zero(PAGE_SIZE, node);
29c71111	220	if (!p)
9dce07f1	221	return NULL;
29c71111	222	pgd_populate(&init_mm, pgd, p);
8f6aac41	223	}
29c71111	224	return pgd;
8f6aac41 CL	225	}
8f6aac41 CL	226
0aad818b JW	227	int __meminit vmemmap_populate_basepages(unsigned long start,
0aad818b JW	228	unsigned long end, int node)
8f6aac41	229	{
0aad818b	230	unsigned long addr = start;
29c71111	231	pgd_t *pgd;
c2febafc	232	p4d_t *p4d;
29c71111 AW	233	pud_t *pud;
	234	pmd_t *pmd;
	235	pte_t *pte;
8f6aac41	236
29c71111 AW	237	for (; addr < end; addr += PAGE_SIZE) {
	238	pgd = vmemmap_pgd_populate(addr, node);
	239	if (!pgd)
	240	return -ENOMEM;
c2febafc KS	241	p4d = vmemmap_p4d_populate(pgd, addr, node);
	242	if (!p4d)
	243	return -ENOMEM;
	244	pud = vmemmap_pud_populate(p4d, addr, node);
29c71111 AW	245	if (!pud)
	246	return -ENOMEM;
	247	pmd = vmemmap_pmd_populate(pud, addr, node);
	248	if (!pmd)
	249	return -ENOMEM;
	250	pte = vmemmap_pte_populate(pmd, addr, node);
	251	if (!pte)
	252	return -ENOMEM;
	253	vmemmap_verify(pte, node, addr, addr + PAGE_SIZE);
8f6aac41	254	}
29c71111 AW	255
29c71111 AW	256	return 0;
8f6aac41	257	}
8f6aac41	258
7b73d978 CH	259	struct page * __meminit sparse_mem_map_populate(unsigned long pnum, int nid,
7b73d978 CH	260	struct vmem_altmap *altmap)
8f6aac41	261	{
0aad818b JW	262	unsigned long start;
	263	unsigned long end;
	264	struct page *map;
	265
	266	map = pfn_to_page(pnum * PAGES_PER_SECTION);
	267	start = (unsigned long)map;
	268	end = (unsigned long)(map + PAGES_PER_SECTION);
	269
7b73d978	270	if (vmemmap_populate(start, end, nid, altmap))
8f6aac41 CL	271	return NULL;
	272
	273	return map;
	274	}
9bdac914 YL	275
	276	void __init sparse_mem_maps_populate_node(struct page **map_map,
	277	unsigned long pnum_begin,
	278	unsigned long pnum_end,
	279	unsigned long map_count, int nodeid)
	280	{
	281	unsigned long pnum;
	282	unsigned long size = sizeof(struct page) * PAGES_PER_SECTION;
	283	void *vmemmap_buf_start;
	284
	285	size = ALIGN(size, PMD_SIZE);
	286	vmemmap_buf_start = __earlyonly_bootmem_alloc(nodeid, size * map_count,
	287	PMD_SIZE, __pa(MAX_DMA_ADDRESS));
	288
	289	if (vmemmap_buf_start) {
	290	vmemmap_buf = vmemmap_buf_start;
	291	vmemmap_buf_end = vmemmap_buf_start + size * map_count;
	292	}
	293
	294	for (pnum = pnum_begin; pnum < pnum_end; pnum++) {
	295	struct mem_section *ms;
	296
	297	if (!present_section_nr(pnum))
	298	continue;
	299
7b73d978	300	map_map[pnum] = sparse_mem_map_populate(pnum, nodeid, NULL);
9bdac914 YL	301	if (map_map[pnum])
	302	continue;
	303	ms = __nr_to_section(pnum);
1170532b	304	pr_err("%s: sparsemem memory map backing failed some memory will not be available\n",
756a025f	305	__func__);
9bdac914 YL	306	ms->section_mem_map = 0;
	307	}
	308
	309	if (vmemmap_buf_start) {
	310	/* need to free left buf */
bb016b84 SS	311	memblock_free_early(__pa(vmemmap_buf),
bb016b84 SS	312	vmemmap_buf_end - vmemmap_buf);
9bdac914 YL	313	vmemmap_buf = NULL;
	314	vmemmap_buf_end = NULL;
	315	}
	316	}