[linux-2.6-block.git] / mm / pagewalk.c

#include <linux/mm.h>
#include <linux/highmem.h>
#include <linux/sched.h>
#include <linux/hugetlb.h>

static int walk_pte_range(pmd_t *pmd, unsigned long addr, unsigned long end,
			  struct mm_walk *walk)
{
	pte_t *pte;
	int err = 0;

	pte = pte_offset_map(pmd, addr);
	for (;;) {
		err = walk->pte_entry(pte, addr, addr + PAGE_SIZE, walk);
		if (err)
		       break;
		addr += PAGE_SIZE;
		if (addr == end)
			break;
		pte++;
	}

	pte_unmap(pte);
	return err;
}

static int walk_pmd_range(pud_t *pud, unsigned long addr, unsigned long end,
			  struct mm_walk *walk)
{
	pmd_t *pmd;
	unsigned long next;
	int err = 0;

	pmd = pmd_offset(pud, addr);
	do {
		next = pmd_addr_end(addr, end);
		split_huge_page_pmd(walk->mm, pmd);
		if (pmd_none_or_clear_bad(pmd)) {
			if (walk->pte_hole)
				err = walk->pte_hole(addr, next, walk);
			if (err)
				break;
			continue;
		}
		if (walk->pmd_entry)
			err = walk->pmd_entry(pmd, addr, next, walk);
		if (!err && walk->pte_entry)
			err = walk_pte_range(pmd, addr, next, walk);
		if (err)
			break;
	} while (pmd++, addr = next, addr != end);

	return err;
}

static int walk_pud_range(pgd_t *pgd, unsigned long addr, unsigned long end,
			  struct mm_walk *walk)
{
	pud_t *pud;
	unsigned long next;
	int err = 0;

	pud = pud_offset(pgd, addr);
	do {
		next = pud_addr_end(addr, end);
		if (pud_none_or_clear_bad(pud)) {
			if (walk->pte_hole)
				err = walk->pte_hole(addr, next, walk);
			if (err)
				break;
			continue;
		}
		if (walk->pud_entry)
			err = walk->pud_entry(pud, addr, next, walk);
		if (!err && (walk->pmd_entry || walk->pte_entry))
			err = walk_pmd_range(pud, addr, next, walk);
		if (err)
			break;
	} while (pud++, addr = next, addr != end);

	return err;
}

#ifdef CONFIG_HUGETLB_PAGE
static unsigned long hugetlb_entry_end(struct hstate *h, unsigned long addr,
				       unsigned long end)
{
	unsigned long boundary = (addr & huge_page_mask(h)) + huge_page_size(h);
	return boundary < end ? boundary : end;
}

static int walk_hugetlb_range(struct vm_area_struct *vma,
			      unsigned long addr, unsigned long end,
			      struct mm_walk *walk)
{
	struct hstate *h = hstate_vma(vma);
	unsigned long next;
	unsigned long hmask = huge_page_mask(h);
	pte_t *pte;
	int err = 0;

	do {
		next = hugetlb_entry_end(h, addr, end);
		pte = huge_pte_offset(walk->mm, addr & hmask);
		if (pte && walk->hugetlb_entry)
			err = walk->hugetlb_entry(pte, hmask, addr, next, walk);
		if (err)
			return err;
	} while (addr = next, addr != end);

	return 0;
}
#endif

/**
 * walk_page_range - walk a memory map's page tables with a callback
 * @mm: memory map to walk
 * @addr: starting address
 * @end: ending address
 * @walk: set of callbacks to invoke for each level of the tree
 *
 * Recursively walk the page table for the memory area in a VMA,
 * calling supplied callbacks. Callbacks are called in-order (first
 * PGD, first PUD, first PMD, first PTE, second PTE... second PMD,
 * etc.). If lower-level callbacks are omitted, walking depth is reduced.
 *
 * Each callback receives an entry pointer and the start and end of the
 * associated range, and a copy of the original mm_walk for access to
 * the ->private or ->mm fields.
 *
 * No locks are taken, but the bottom level iterator will map PTE
 * directories from highmem if necessary.
 *
 * If any callback returns a non-zero value, the walk is aborted and
 * the return value is propagated back to the caller. Otherwise 0 is returned.
 */
int walk_page_range(unsigned long addr, unsigned long end,
		    struct mm_walk *walk)
{
	pgd_t *pgd;
	unsigned long next;
	int err = 0;

	if (addr >= end)
		return err;

	if (!walk->mm)
		return -EINVAL;

	pgd = pgd_offset(walk->mm, addr);
	do {
		struct vm_area_struct *uninitialized_var(vma);

		next = pgd_addr_end(addr, end);

#ifdef CONFIG_HUGETLB_PAGE
		/*
		 * handle hugetlb vma individually because pagetable walk for
		 * the hugetlb page is dependent on the architecture and
		 * we can't handled it in the same manner as non-huge pages.
		 */
		vma = find_vma(walk->mm, addr);
		if (vma && is_vm_hugetlb_page(vma)) {
			if (vma->vm_end < next)
				next = vma->vm_end;
			/*
			 * Hugepage is very tightly coupled with vma, so
			 * walk through hugetlb entries within a given vma.
			 */
			err = walk_hugetlb_range(vma, addr, next, walk);
			if (err)
				break;
			pgd = pgd_offset(walk->mm, next);
			continue;
		}
#endif
		if (pgd_none_or_clear_bad(pgd)) {
			if (walk->pte_hole)
				err = walk->pte_hole(addr, next, walk);
			if (err)
				break;
			pgd++;
			continue;
		}
		if (walk->pgd_entry)
			err = walk->pgd_entry(pgd, addr, next, walk);
		if (!err &&
		    (walk->pud_entry || walk->pmd_entry || walk->pte_entry))
			err = walk_pud_range(pgd, addr, next, walk);
		if (err)
			break;
		pgd++;
	} while (addr = next, addr != end);

	return err;
}
Commit	Line	Data
e6473092 MM	1	#include <linux/mm.h>
	2	#include <linux/highmem.h>
	3	#include <linux/sched.h>
d33b9f45	4	#include <linux/hugetlb.h>
e6473092 MM	5
e6473092 MM	6	static int walk_pte_range(pmd_t *pmd, unsigned long addr, unsigned long end,
2165009b	7	struct mm_walk *walk)
e6473092 MM	8	{
	9	pte_t *pte;
	10	int err = 0;
	11
	12	pte = pte_offset_map(pmd, addr);
556637cd	13	for (;;) {
2165009b	14	err = walk->pte_entry(pte, addr, addr + PAGE_SIZE, walk);
e6473092 MM	15	if (err)
e6473092 MM	16	break;
556637cd JW	17	addr += PAGE_SIZE;
	18	if (addr == end)
	19	break;
	20	pte++;
	21	}
e6473092 MM	22
	23	pte_unmap(pte);
	24	return err;
	25	}
	26
	27	static int walk_pmd_range(pud_t *pud, unsigned long addr, unsigned long end,
2165009b	28	struct mm_walk *walk)
e6473092 MM	29	{
	30	pmd_t *pmd;
	31	unsigned long next;
	32	int err = 0;
	33
	34	pmd = pmd_offset(pud, addr);
	35	do {
	36	next = pmd_addr_end(addr, end);
bae9c19b	37	split_huge_page_pmd(walk->mm, pmd);
e6473092 MM	38	if (pmd_none_or_clear_bad(pmd)) {
e6473092 MM	39	if (walk->pte_hole)
2165009b	40	err = walk->pte_hole(addr, next, walk);
e6473092 MM	41	if (err)
	42	break;
	43	continue;
	44	}
	45	if (walk->pmd_entry)
2165009b	46	err = walk->pmd_entry(pmd, addr, next, walk);
e6473092	47	if (!err && walk->pte_entry)
2165009b	48	err = walk_pte_range(pmd, addr, next, walk);
e6473092 MM	49	if (err)
	50	break;
	51	} while (pmd++, addr = next, addr != end);
	52
	53	return err;
	54	}
	55
	56	static int walk_pud_range(pgd_t *pgd, unsigned long addr, unsigned long end,
2165009b	57	struct mm_walk *walk)
e6473092 MM	58	{
	59	pud_t *pud;
	60	unsigned long next;
	61	int err = 0;
	62
	63	pud = pud_offset(pgd, addr);
	64	do {
	65	next = pud_addr_end(addr, end);
	66	if (pud_none_or_clear_bad(pud)) {
	67	if (walk->pte_hole)
2165009b	68	err = walk->pte_hole(addr, next, walk);
e6473092 MM	69	if (err)
	70	break;
	71	continue;
	72	}
	73	if (walk->pud_entry)
2165009b	74	err = walk->pud_entry(pud, addr, next, walk);
e6473092	75	if (!err && (walk->pmd_entry \|\| walk->pte_entry))
2165009b	76	err = walk_pmd_range(pud, addr, next, walk);
e6473092 MM	77	if (err)
	78	break;
	79	} while (pud++, addr = next, addr != end);
	80
	81	return err;
	82	}
	83
116354d1 NH	84	#ifdef CONFIG_HUGETLB_PAGE
	85	static unsigned long hugetlb_entry_end(struct hstate *h, unsigned long addr,
	86	unsigned long end)
	87	{
	88	unsigned long boundary = (addr & huge_page_mask(h)) + huge_page_size(h);
	89	return boundary < end ? boundary : end;
	90	}
	91
	92	static int walk_hugetlb_range(struct vm_area_struct *vma,
	93	unsigned long addr, unsigned long end,
	94	struct mm_walk *walk)
	95	{
	96	struct hstate *h = hstate_vma(vma);
	97	unsigned long next;
	98	unsigned long hmask = huge_page_mask(h);
	99	pte_t *pte;
	100	int err = 0;
	101
	102	do {
	103	next = hugetlb_entry_end(h, addr, end);
	104	pte = huge_pte_offset(walk->mm, addr & hmask);
	105	if (pte && walk->hugetlb_entry)
	106	err = walk->hugetlb_entry(pte, hmask, addr, next, walk);
	107	if (err)
	108	return err;
	109	} while (addr = next, addr != end);
	110
	111	return 0;
	112	}
	113	#endif
	114
e6473092 MM	115	/**
e6473092 MM	116	* walk_page_range - walk a memory map's page tables with a callback
7682486b RD	117	* @mm: memory map to walk
	118	* @addr: starting address
	119	* @end: ending address
	120	* @walk: set of callbacks to invoke for each level of the tree
e6473092 MM	121	*
	122	* Recursively walk the page table for the memory area in a VMA,
	123	* calling supplied callbacks. Callbacks are called in-order (first
	124	* PGD, first PUD, first PMD, first PTE, second PTE... second PMD,
	125	* etc.). If lower-level callbacks are omitted, walking depth is reduced.
	126	*
2165009b DH	127	* Each callback receives an entry pointer and the start and end of the
	128	* associated range, and a copy of the original mm_walk for access to
	129	* the ->private or ->mm fields.
e6473092 MM	130	*
	131	* No locks are taken, but the bottom level iterator will map PTE
	132	* directories from highmem if necessary.
	133	*
	134	* If any callback returns a non-zero value, the walk is aborted and
	135	* the return value is propagated back to the caller. Otherwise 0 is returned.
	136	*/
2165009b DH	137	int walk_page_range(unsigned long addr, unsigned long end,
2165009b DH	138	struct mm_walk *walk)
e6473092 MM	139	{
	140	pgd_t *pgd;
	141	unsigned long next;
	142	int err = 0;
	143
	144	if (addr >= end)
	145	return err;
	146
2165009b DH	147	if (!walk->mm)
	148	return -EINVAL;
	149
	150	pgd = pgd_offset(walk->mm, addr);
e6473092	151	do {
5f0af70a DS	152	struct vm_area_struct *uninitialized_var(vma);
5f0af70a DS	153
e6473092	154	next = pgd_addr_end(addr, end);
d33b9f45	155
5f0af70a	156	#ifdef CONFIG_HUGETLB_PAGE
5dc37642 NH	157	/*
	158	* handle hugetlb vma individually because pagetable walk for
	159	* the hugetlb page is dependent on the architecture and
	160	* we can't handled it in the same manner as non-huge pages.
	161	*/
d33b9f45 NH	162	vma = find_vma(walk->mm, addr);
	163	if (vma && is_vm_hugetlb_page(vma)) {
	164	if (vma->vm_end < next)
	165	next = vma->vm_end;
116354d1 NH	166	/*
	167	* Hugepage is very tightly coupled with vma, so
	168	* walk through hugetlb entries within a given vma.
	169	*/
	170	err = walk_hugetlb_range(vma, addr, next, walk);
5dc37642 NH	171	if (err)
5dc37642 NH	172	break;
116354d1	173	pgd = pgd_offset(walk->mm, next);
d33b9f45 NH	174	continue;
d33b9f45 NH	175	}
5dc37642	176	#endif
e6473092 MM	177	if (pgd_none_or_clear_bad(pgd)) {
e6473092 MM	178	if (walk->pte_hole)
2165009b	179	err = walk->pte_hole(addr, next, walk);
e6473092 MM	180	if (err)
e6473092 MM	181	break;
d33b9f45	182	pgd++;
e6473092 MM	183	continue;
	184	}
	185	if (walk->pgd_entry)
2165009b	186	err = walk->pgd_entry(pgd, addr, next, walk);
e6473092 MM	187	if (!err &&
e6473092 MM	188	(walk->pud_entry \|\| walk->pmd_entry \|\| walk->pte_entry))
2165009b	189	err = walk_pud_range(pgd, addr, next, walk);
e6473092 MM	190	if (err)
e6473092 MM	191	break;
d33b9f45 NH	192	pgd++;
d33b9f45 NH	193	} while (addr = next, addr != end);
e6473092 MM	194
	195	return err;
	196	}