[deliverable/linux.git] / mm / mincore.c

/*
 *	linux/mm/mincore.c
 *
 * Copyright (C) 1994-2006  Linus Torvalds
 */

/*
 * The mincore() system call.
 */
#include <linux/pagemap.h>
#include <linux/gfp.h>
#include <linux/mm.h>
#include <linux/mman.h>
#include <linux/syscalls.h>
#include <linux/swap.h>
#include <linux/swapops.h>
#include <linux/hugetlb.h>

#include <asm/uaccess.h>
#include <asm/pgtable.h>

static void mincore_hugetlb_page_range(struct vm_area_struct *vma,
				unsigned long addr, unsigned long end,
				unsigned char *vec)
{
#ifdef CONFIG_HUGETLB_PAGE
	struct hstate *h;

	h = hstate_vma(vma);
	while (1) {
		unsigned char present;
		pte_t *ptep;
		/*
		 * Huge pages are always in RAM for now, but
		 * theoretically it needs to be checked.
		 */
		ptep = huge_pte_offset(current->mm,
				       addr & huge_page_mask(h));
		present = ptep && !huge_pte_none(huge_ptep_get(ptep));
		while (1) {
			*vec = present;
			vec++;
			addr += PAGE_SIZE;
			if (addr == end)
				return;
			/* check hugepage border */
			if (!(addr & ~huge_page_mask(h)))
				break;
		}
	}
#else
	BUG();
#endif
}

/*
 * Later we can get more picky about what "in core" means precisely.
 * For now, simply check to see if the page is in the page cache,
 * and is up to date; i.e. that no page-in operation would be required
 * at this time if an application were to map and access this page.
 */
static unsigned char mincore_page(struct address_space *mapping, pgoff_t pgoff)
{
	unsigned char present = 0;
	struct page *page;

	/*
	 * When tmpfs swaps out a page from a file, any process mapping that
	 * file will not get a swp_entry_t in its pte, but rather it is like
	 * any other file mapping (ie. marked !present and faulted in with
	 * tmpfs's .fault). So swapped out tmpfs mappings are tested here.
	 */
#ifdef CONFIG_SWAP
	if (shmem_mapping(mapping)) {
		page = find_get_entry(mapping, pgoff);
		/*
		 * shmem/tmpfs may return swap: account for swapcache
		 * page too.
		 */
		if (radix_tree_exceptional_entry(page)) {
			swp_entry_t swp = radix_to_swp_entry(page);
			page = find_get_page(swap_address_space(swp), swp.val);
		}
	} else
		page = find_get_page(mapping, pgoff);
#else
	page = find_get_page(mapping, pgoff);
#endif
	if (page) {
		present = PageUptodate(page);
		page_cache_release(page);
	}

	return present;
}

static void mincore_unmapped_range(struct vm_area_struct *vma,
				unsigned long addr, unsigned long end,
				unsigned char *vec)
{
	unsigned long nr = (end - addr) >> PAGE_SHIFT;
	int i;

	if (vma->vm_file) {
		pgoff_t pgoff;

		pgoff = linear_page_index(vma, addr);
		for (i = 0; i < nr; i++, pgoff++)
			vec[i] = mincore_page(vma->vm_file->f_mapping, pgoff);
	} else {
		for (i = 0; i < nr; i++)
			vec[i] = 0;
	}
}

static void mincore_pte_range(struct vm_area_struct *vma, pmd_t *pmd,
			unsigned long addr, unsigned long end,
			unsigned char *vec)
{
	unsigned long next;
	spinlock_t *ptl;
	pte_t *ptep;

	ptep = pte_offset_map_lock(vma->vm_mm, pmd, addr, &ptl);
	do {
		pte_t pte = *ptep;
		pgoff_t pgoff;

		next = addr + PAGE_SIZE;
		if (pte_none(pte))
			mincore_unmapped_range(vma, addr, next, vec);
		else if (pte_present(pte))
			*vec = 1;
		else if (pte_file(pte)) {
			pgoff = pte_to_pgoff(pte);
			*vec = mincore_page(vma->vm_file->f_mapping, pgoff);
		} else { /* pte is a swap entry */
			swp_entry_t entry = pte_to_swp_entry(pte);

			if (non_swap_entry(entry)) {
				/*
				 * migration or hwpoison entries are always
				 * uptodate
				 */
				*vec = 1;
			} else {
#ifdef CONFIG_SWAP
				pgoff = entry.val;
				*vec = mincore_page(swap_address_space(entry),
					pgoff);
#else
				WARN_ON(1);
				*vec = 1;
#endif
			}
		}
		vec++;
	} while (ptep++, addr = next, addr != end);
	pte_unmap_unlock(ptep - 1, ptl);
}

static void mincore_pmd_range(struct vm_area_struct *vma, pud_t *pud,
			unsigned long addr, unsigned long end,
			unsigned char *vec)
{
	unsigned long next;
	pmd_t *pmd;

	pmd = pmd_offset(pud, addr);
	do {
		next = pmd_addr_end(addr, end);
		if (pmd_trans_huge(*pmd)) {
			if (mincore_huge_pmd(vma, pmd, addr, next, vec)) {
				vec += (next - addr) >> PAGE_SHIFT;
				continue;
			}
			/* fall through */
		}
		if (pmd_none_or_trans_huge_or_clear_bad(pmd))
			mincore_unmapped_range(vma, addr, next, vec);
		else
			mincore_pte_range(vma, pmd, addr, next, vec);
		vec += (next - addr) >> PAGE_SHIFT;
	} while (pmd++, addr = next, addr != end);
}

static void mincore_pud_range(struct vm_area_struct *vma, pgd_t *pgd,
			unsigned long addr, unsigned long end,
			unsigned char *vec)
{
	unsigned long next;
	pud_t *pud;

	pud = pud_offset(pgd, addr);
	do {
		next = pud_addr_end(addr, end);
		if (pud_none_or_clear_bad(pud))
			mincore_unmapped_range(vma, addr, next, vec);
		else
			mincore_pmd_range(vma, pud, addr, next, vec);
		vec += (next - addr) >> PAGE_SHIFT;
	} while (pud++, addr = next, addr != end);
}

static void mincore_page_range(struct vm_area_struct *vma,
			unsigned long addr, unsigned long end,
			unsigned char *vec)
{
	unsigned long next;
	pgd_t *pgd;

	pgd = pgd_offset(vma->vm_mm, addr);
	do {
		next = pgd_addr_end(addr, end);
		if (pgd_none_or_clear_bad(pgd))
			mincore_unmapped_range(vma, addr, next, vec);
		else
			mincore_pud_range(vma, pgd, addr, next, vec);
		vec += (next - addr) >> PAGE_SHIFT;
	} while (pgd++, addr = next, addr != end);
}

/*
 * Do a chunk of "sys_mincore()". We've already checked
 * all the arguments, we hold the mmap semaphore: we should
 * just return the amount of info we're asked for.
 */
static long do_mincore(unsigned long addr, unsigned long pages, unsigned char *vec)
{
	struct vm_area_struct *vma;
	unsigned long end;

	vma = find_vma(current->mm, addr);
	if (!vma || addr < vma->vm_start)
		return -ENOMEM;

	end = min(vma->vm_end, addr + (pages << PAGE_SHIFT));

	if (is_vm_hugetlb_page(vma))
		mincore_hugetlb_page_range(vma, addr, end, vec);
	else
		mincore_page_range(vma, addr, end, vec);

	return (end - addr) >> PAGE_SHIFT;
}

/*
 * The mincore(2) system call.
 *
 * mincore() returns the memory residency status of the pages in the
 * current process's address space specified by [addr, addr + len).
 * The status is returned in a vector of bytes.  The least significant
 * bit of each byte is 1 if the referenced page is in memory, otherwise
 * it is zero.
 *
 * Because the status of a page can change after mincore() checks it
 * but before it returns to the application, the returned vector may
 * contain stale information.  Only locked pages are guaranteed to
 * remain in memory.
 *
 * return values:
 *  zero    - success
 *  -EFAULT - vec points to an illegal address
 *  -EINVAL - addr is not a multiple of PAGE_CACHE_SIZE
 *  -ENOMEM - Addresses in the range [addr, addr + len] are
 *		invalid for the address space of this process, or
 *		specify one or more pages which are not currently
 *		mapped
 *  -EAGAIN - A kernel resource was temporarily unavailable.
 */
SYSCALL_DEFINE3(mincore, unsigned long, start, size_t, len,
		unsigned char __user *, vec)
{
	long retval;
	unsigned long pages;
	unsigned char *tmp;

	/* Check the start address: needs to be page-aligned.. */
 	if (start & ~PAGE_CACHE_MASK)
		return -EINVAL;

	/* ..and we need to be passed a valid user-space range */
	if (!access_ok(VERIFY_READ, (void __user *) start, len))
		return -ENOMEM;

	/* This also avoids any overflows on PAGE_CACHE_ALIGN */
	pages = len >> PAGE_SHIFT;
	pages += (len & ~PAGE_MASK) != 0;

	if (!access_ok(VERIFY_WRITE, vec, pages))
		return -EFAULT;

	tmp = (void *) __get_free_page(GFP_USER);
	if (!tmp)
		return -EAGAIN;

	retval = 0;
	while (pages) {
		/*
		 * Do at most PAGE_SIZE entries per iteration, due to
		 * the temporary buffer size.
		 */
		down_read(&current->mm->mmap_sem);
		retval = do_mincore(start, min(pages, PAGE_SIZE), tmp);
		up_read(&current->mm->mmap_sem);

		if (retval <= 0)
			break;
		if (copy_to_user(vec, tmp, retval)) {
			retval = -EFAULT;
			break;
		}
		pages -= retval;
		vec += retval;
		start += retval << PAGE_SHIFT;
		retval = 0;
	}
	free_page((unsigned long) tmp);
	return retval;
}
Commit	Line	Data
1da177e4 LT	1	/*
	2	* linux/mm/mincore.c
	3	*
2f77d107	4	* Copyright (C) 1994-2006 Linus Torvalds
1da177e4 LT	5	*/
	6
	7	/*
	8	* The mincore() system call.
	9	*/
1da177e4	10	#include <linux/pagemap.h>
5a0e3ad6	11	#include <linux/gfp.h>
1da177e4 LT	12	#include <linux/mm.h>
	13	#include <linux/mman.h>
	14	#include <linux/syscalls.h>
42da9cbd NP	15	#include <linux/swap.h>
42da9cbd NP	16	#include <linux/swapops.h>
4f16fc10	17	#include <linux/hugetlb.h>
1da177e4 LT	18
	19	#include <asm/uaccess.h>
	20	#include <asm/pgtable.h>
	21
f4884010	22	static void mincore_hugetlb_page_range(struct vm_area_struct *vma,
25ef0e50	23	unsigned long addr, unsigned long end,
f4884010 JW	24	unsigned char *vec)
	25	{
	26	#ifdef CONFIG_HUGETLB_PAGE
	27	struct hstate *h;
f4884010	28
f4884010 JW	29	h = hstate_vma(vma);
	30	while (1) {
	31	unsigned char present;
	32	pte_t *ptep;
	33	/*
	34	* Huge pages are always in RAM for now, but
	35	* theoretically it needs to be checked.
	36	*/
	37	ptep = huge_pte_offset(current->mm,
	38	addr & huge_page_mask(h));
	39	present = ptep && !huge_pte_none(huge_ptep_get(ptep));
	40	while (1) {
25ef0e50 JW	41	*vec = present;
25ef0e50 JW	42	vec++;
f4884010	43	addr += PAGE_SIZE;
25ef0e50	44	if (addr == end)
f4884010 JW	45	return;
	46	/* check hugepage border */
	47	if (!(addr & ~huge_page_mask(h)))
	48	break;
	49	}
	50	}
	51	#else
	52	BUG();
	53	#endif
	54	}
	55
1da177e4 LT	56	/*
	57	* Later we can get more picky about what "in core" means precisely.
	58	* For now, simply check to see if the page is in the page cache,
	59	* and is up to date; i.e. that no page-in operation would be required
	60	* at this time if an application were to map and access this page.
	61	*/
42da9cbd	62	static unsigned char mincore_page(struct address_space *mapping, pgoff_t pgoff)
1da177e4 LT	63	{
1da177e4 LT	64	unsigned char present = 0;
42da9cbd	65	struct page *page;
1da177e4	66
42da9cbd NP	67	/*
	68	* When tmpfs swaps out a page from a file, any process mapping that
	69	* file will not get a swp_entry_t in its pte, but rather it is like
	70	* any other file mapping (ie. marked !present and faulted in with
3c18ddd1	71	* tmpfs's .fault). So swapped out tmpfs mappings are tested here.
42da9cbd	72	*/
31475dd6	73	#ifdef CONFIG_SWAP
0cd6144a JW	74	if (shmem_mapping(mapping)) {
	75	page = find_get_entry(mapping, pgoff);
	76	/*
	77	* shmem/tmpfs may return swap: account for swapcache
	78	* page too.
	79	*/
	80	if (radix_tree_exceptional_entry(page)) {
	81	swp_entry_t swp = radix_to_swp_entry(page);
	82	page = find_get_page(swap_address_space(swp), swp.val);
	83	}
	84	} else
	85	page = find_get_page(mapping, pgoff);
	86	#else
	87	page = find_get_page(mapping, pgoff);
31475dd6	88	#endif
1da177e4 LT	89	if (page) {
	90	present = PageUptodate(page);
	91	page_cache_release(page);
	92	}
	93
	94	return present;
	95	}
	96
f4884010	97	static void mincore_unmapped_range(struct vm_area_struct *vma,
25ef0e50	98	unsigned long addr, unsigned long end,
f4884010 JW	99	unsigned char *vec)
f4884010 JW	100	{
25ef0e50	101	unsigned long nr = (end - addr) >> PAGE_SHIFT;
f4884010 JW	102	int i;
	103
	104	if (vma->vm_file) {
	105	pgoff_t pgoff;
	106
	107	pgoff = linear_page_index(vma, addr);
	108	for (i = 0; i < nr; i++, pgoff++)
	109	vec[i] = mincore_page(vma->vm_file->f_mapping, pgoff);
	110	} else {
	111	for (i = 0; i < nr; i++)
	112	vec[i] = 0;
	113	}
	114	}
	115
	116	static void mincore_pte_range(struct vm_area_struct vma, pmd_t pmd,
25ef0e50	117	unsigned long addr, unsigned long end,
f4884010 JW	118	unsigned char *vec)
f4884010 JW	119	{
25ef0e50	120	unsigned long next;
f4884010 JW	121	spinlock_t *ptl;
f4884010 JW	122	pte_t *ptep;
f4884010 JW	123
f4884010 JW	124	ptep = pte_offset_map_lock(vma->vm_mm, pmd, addr, &ptl);
25ef0e50	125	do {
f4884010 JW	126	pte_t pte = *ptep;
	127	pgoff_t pgoff;
	128
25ef0e50	129	next = addr + PAGE_SIZE;
f4884010	130	if (pte_none(pte))
25ef0e50	131	mincore_unmapped_range(vma, addr, next, vec);
f4884010	132	else if (pte_present(pte))
25ef0e50	133	*vec = 1;
f4884010 JW	134	else if (pte_file(pte)) {
f4884010 JW	135	pgoff = pte_to_pgoff(pte);
25ef0e50	136	*vec = mincore_page(vma->vm_file->f_mapping, pgoff);
f4884010 JW	137	} else { /* pte is a swap entry */
	138	swp_entry_t entry = pte_to_swp_entry(pte);
	139
c313dc5d WY	140	if (non_swap_entry(entry)) {
	141	/*
	142	* migration or hwpoison entries are always
	143	* uptodate
	144	*/
25ef0e50	145	*vec = 1;
f4884010 JW	146	} else {
	147	#ifdef CONFIG_SWAP
	148	pgoff = entry.val;
33806f06 SL	149	*vec = mincore_page(swap_address_space(entry),
33806f06 SL	150	pgoff);
f4884010 JW	151	#else
f4884010 JW	152	WARN_ON(1);
25ef0e50	153	*vec = 1;
f4884010 JW	154	#endif
	155	}
	156	}
25ef0e50 JW	157	vec++;
25ef0e50 JW	158	} while (ptep++, addr = next, addr != end);
f4884010 JW	159	pte_unmap_unlock(ptep - 1, ptl);
	160	}
	161
e48293fd JW	162	static void mincore_pmd_range(struct vm_area_struct vma, pud_t pud,
	163	unsigned long addr, unsigned long end,
	164	unsigned char *vec)
	165	{
	166	unsigned long next;
	167	pmd_t *pmd;
	168
	169	pmd = pmd_offset(pud, addr);
	170	do {
	171	next = pmd_addr_end(addr, end);
0ca1634d JW	172	if (pmd_trans_huge(*pmd)) {
	173	if (mincore_huge_pmd(vma, pmd, addr, next, vec)) {
	174	vec += (next - addr) >> PAGE_SHIFT;
	175	continue;
	176	}
	177	/* fall through */
	178	}
1a5a9906	179	if (pmd_none_or_trans_huge_or_clear_bad(pmd))
e48293fd JW	180	mincore_unmapped_range(vma, addr, next, vec);
	181	else
	182	mincore_pte_range(vma, pmd, addr, next, vec);
	183	vec += (next - addr) >> PAGE_SHIFT;
	184	} while (pmd++, addr = next, addr != end);
	185	}
	186
	187	static void mincore_pud_range(struct vm_area_struct vma, pgd_t pgd,
	188	unsigned long addr, unsigned long end,
	189	unsigned char *vec)
	190	{
	191	unsigned long next;
	192	pud_t *pud;
	193
	194	pud = pud_offset(pgd, addr);
	195	do {
	196	next = pud_addr_end(addr, end);
	197	if (pud_none_or_clear_bad(pud))
	198	mincore_unmapped_range(vma, addr, next, vec);
	199	else
	200	mincore_pmd_range(vma, pud, addr, next, vec);
	201	vec += (next - addr) >> PAGE_SHIFT;
	202	} while (pud++, addr = next, addr != end);
	203	}
	204
	205	static void mincore_page_range(struct vm_area_struct *vma,
	206	unsigned long addr, unsigned long end,
	207	unsigned char *vec)
	208	{
	209	unsigned long next;
	210	pgd_t *pgd;
	211
	212	pgd = pgd_offset(vma->vm_mm, addr);
	213	do {
	214	next = pgd_addr_end(addr, end);
	215	if (pgd_none_or_clear_bad(pgd))
	216	mincore_unmapped_range(vma, addr, next, vec);
	217	else
	218	mincore_pud_range(vma, pgd, addr, next, vec);
	219	vec += (next - addr) >> PAGE_SHIFT;
	220	} while (pgd++, addr = next, addr != end);
	221	}
	222
2f77d107 LT	223	/*
	224	* Do a chunk of "sys_mincore()". We've already checked
	225	* all the arguments, we hold the mmap semaphore: we should
	226	* just return the amount of info we're asked for.
	227	*/
6a60f1b3	228	static long do_mincore(unsigned long addr, unsigned long pages, unsigned char *vec)
1da177e4	229	{
6a60f1b3	230	struct vm_area_struct *vma;
25ef0e50	231	unsigned long end;
1da177e4	232
6a60f1b3	233	vma = find_vma(current->mm, addr);
4fb23e43 LT	234	if (!vma \|\| addr < vma->vm_start)
4fb23e43 LT	235	return -ENOMEM;
1da177e4	236
25ef0e50	237	end = min(vma->vm_end, addr + (pages << PAGE_SHIFT));
6a60f1b3	238
e48293fd JW	239	if (is_vm_hugetlb_page(vma))
	240	mincore_hugetlb_page_range(vma, addr, end, vec);
	241	else
	242	mincore_page_range(vma, addr, end, vec);
42da9cbd	243
25ef0e50	244	return (end - addr) >> PAGE_SHIFT;
1da177e4 LT	245	}
	246
	247	/*
	248	* The mincore(2) system call.
	249	*
	250	* mincore() returns the memory residency status of the pages in the
	251	* current process's address space specified by [addr, addr + len).
	252	* The status is returned in a vector of bytes. The least significant
	253	* bit of each byte is 1 if the referenced page is in memory, otherwise
	254	* it is zero.
	255	*
	256	* Because the status of a page can change after mincore() checks it
	257	* but before it returns to the application, the returned vector may
	258	* contain stale information. Only locked pages are guaranteed to
	259	* remain in memory.
	260	*
	261	* return values:
	262	* zero - success
	263	* -EFAULT - vec points to an illegal address
	264	* -EINVAL - addr is not a multiple of PAGE_CACHE_SIZE
	265	* -ENOMEM - Addresses in the range [addr, addr + len] are
	266	* invalid for the address space of this process, or
	267	* specify one or more pages which are not currently
	268	* mapped
	269	* -EAGAIN - A kernel resource was temporarily unavailable.
	270	*/
3480b257 HC	271	SYSCALL_DEFINE3(mincore, unsigned long, start, size_t, len,
3480b257 HC	272	unsigned char __user *, vec)
1da177e4	273	{
2f77d107 LT	274	long retval;
	275	unsigned long pages;
	276	unsigned char *tmp;
1da177e4	277
2f77d107 LT	278	/* Check the start address: needs to be page-aligned.. */
	279	if (start & ~PAGE_CACHE_MASK)
	280	return -EINVAL;
1da177e4	281
2f77d107 LT	282	/* ..and we need to be passed a valid user-space range */
	283	if (!access_ok(VERIFY_READ, (void __user *) start, len))
	284	return -ENOMEM;
1da177e4	285
2f77d107 LT	286	/* This also avoids any overflows on PAGE_CACHE_ALIGN */
	287	pages = len >> PAGE_SHIFT;
	288	pages += (len & ~PAGE_MASK) != 0;
1da177e4	289
2f77d107 LT	290	if (!access_ok(VERIFY_WRITE, vec, pages))
2f77d107 LT	291	return -EFAULT;
1da177e4	292
2f77d107 LT	293	tmp = (void *) __get_free_page(GFP_USER);
2f77d107 LT	294	if (!tmp)
4fb23e43	295	return -EAGAIN;
2f77d107 LT	296
	297	retval = 0;
	298	while (pages) {
	299	/*
	300	* Do at most PAGE_SIZE entries per iteration, due to
	301	* the temporary buffer size.
	302	*/
	303	down_read(&current->mm->mmap_sem);
6a60f1b3	304	retval = do_mincore(start, min(pages, PAGE_SIZE), tmp);
2f77d107 LT	305	up_read(&current->mm->mmap_sem);
	306
	307	if (retval <= 0)
	308	break;
	309	if (copy_to_user(vec, tmp, retval)) {
	310	retval = -EFAULT;
	311	break;
1da177e4	312	}
2f77d107 LT	313	pages -= retval;
	314	vec += retval;
	315	start += retval << PAGE_SHIFT;
	316	retval = 0;
1da177e4	317	}
2f77d107 LT	318	free_page((unsigned long) tmp);
2f77d107 LT	319	return retval;
1da177e4	320	}