[mirror_ubuntu-bionic-kernel.git] / mm / page_vma_mapped.c

#include <linux/mm.h>
#include <linux/rmap.h>
#include <linux/hugetlb.h>
#include <linux/swap.h>
#include <linux/swapops.h>

#include "internal.h"

static inline bool check_pmd(struct page_vma_mapped_walk *pvmw)
{
	pmd_t pmde;
	/*
	 * Make sure we don't re-load pmd between present and !trans_huge check.
	 * We need a consistent view.
	 */
	pmde = READ_ONCE(*pvmw->pmd);
	return pmd_present(pmde) && !pmd_trans_huge(pmde);
}

static inline bool not_found(struct page_vma_mapped_walk *pvmw)
{
	page_vma_mapped_walk_done(pvmw);
	return false;
}

static bool map_pte(struct page_vma_mapped_walk *pvmw)
{
	pvmw->pte = pte_offset_map(pvmw->pmd, pvmw->address);
	if (!(pvmw->flags & PVMW_SYNC)) {
		if (pvmw->flags & PVMW_MIGRATION) {
			if (!is_swap_pte(*pvmw->pte))
				return false;
		} else {
			if (!pte_present(*pvmw->pte))
				return false;
		}
	}
	pvmw->ptl = pte_lockptr(pvmw->vma->vm_mm, pvmw->pmd);
	spin_lock(pvmw->ptl);
	return true;
}

static bool check_pte(struct page_vma_mapped_walk *pvmw)
{
	if (pvmw->flags & PVMW_MIGRATION) {
#ifdef CONFIG_MIGRATION
		swp_entry_t entry;
		if (!is_swap_pte(*pvmw->pte))
			return false;
		entry = pte_to_swp_entry(*pvmw->pte);

		if (!is_migration_entry(entry))
			return false;
		if (migration_entry_to_page(entry) - pvmw->page >=
				hpage_nr_pages(pvmw->page)) {
			return false;
		}
		if (migration_entry_to_page(entry) < pvmw->page)
			return false;
#else
		WARN_ON_ONCE(1);
#endif
	} else {
		if (is_swap_pte(*pvmw->pte)) {
			swp_entry_t entry;

			entry = pte_to_swp_entry(*pvmw->pte);
			if (is_device_private_entry(entry) &&
			    device_private_entry_to_page(entry) == pvmw->page)
				return true;
		}

		if (!pte_present(*pvmw->pte))
			return false;

		/* THP can be referenced by any subpage */
		if (pte_page(*pvmw->pte) - pvmw->page >=
				hpage_nr_pages(pvmw->page)) {
			return false;
		}
		if (pte_page(*pvmw->pte) < pvmw->page)
			return false;
	}

	return true;
}

/**
 * page_vma_mapped_walk - check if @pvmw->page is mapped in @pvmw->vma at
 * @pvmw->address
 * @pvmw: pointer to struct page_vma_mapped_walk. page, vma, address and flags
 * must be set. pmd, pte and ptl must be NULL.
 *
 * Returns true if the page is mapped in the vma. @pvmw->pmd and @pvmw->pte point
 * to relevant page table entries. @pvmw->ptl is locked. @pvmw->address is
 * adjusted if needed (for PTE-mapped THPs).
 *
 * If @pvmw->pmd is set but @pvmw->pte is not, you have found PMD-mapped page
 * (usually THP). For PTE-mapped THP, you should run page_vma_mapped_walk() in
 * a loop to find all PTEs that map the THP.
 *
 * For HugeTLB pages, @pvmw->pte is set to the relevant page table entry
 * regardless of which page table level the page is mapped at. @pvmw->pmd is
 * NULL.
 *
 * Retruns false if there are no more page table entries for the page in
 * the vma. @pvmw->ptl is unlocked and @pvmw->pte is unmapped.
 *
 * If you need to stop the walk before page_vma_mapped_walk() returned false,
 * use page_vma_mapped_walk_done(). It will do the housekeeping.
 */
bool page_vma_mapped_walk(struct page_vma_mapped_walk *pvmw)
{
	struct mm_struct *mm = pvmw->vma->vm_mm;
	struct page *page = pvmw->page;
	pgd_t *pgd;
	p4d_t *p4d;
	pud_t *pud;

	/* The only possible pmd mapping has been handled on last iteration */
	if (pvmw->pmd && !pvmw->pte)
		return not_found(pvmw);

	if (pvmw->pte)
		goto next_pte;

	if (unlikely(PageHuge(pvmw->page))) {
		/* when pud is not present, pte will be NULL */
		pvmw->pte = huge_pte_offset(mm, pvmw->address,
					    PAGE_SIZE << compound_order(page));
		if (!pvmw->pte)
			return false;

		pvmw->ptl = huge_pte_lockptr(page_hstate(page), mm, pvmw->pte);
		spin_lock(pvmw->ptl);
		if (!check_pte(pvmw))
			return not_found(pvmw);
		return true;
	}
restart:
	pgd = pgd_offset(mm, pvmw->address);
	if (!pgd_present(*pgd))
		return false;
	p4d = p4d_offset(pgd, pvmw->address);
	if (!p4d_present(*p4d))
		return false;
	pud = pud_offset(p4d, pvmw->address);
	if (!pud_present(*pud))
		return false;
	pvmw->pmd = pmd_offset(pud, pvmw->address);
	if (pmd_trans_huge(*pvmw->pmd) || is_pmd_migration_entry(*pvmw->pmd)) {
		pvmw->ptl = pmd_lock(mm, pvmw->pmd);
		if (likely(pmd_trans_huge(*pvmw->pmd))) {
			if (pvmw->flags & PVMW_MIGRATION)
				return not_found(pvmw);
			if (pmd_page(*pvmw->pmd) != page)
				return not_found(pvmw);
			return true;
		} else if (!pmd_present(*pvmw->pmd)) {
			if (thp_migration_supported()) {
				if (!(pvmw->flags & PVMW_MIGRATION))
					return not_found(pvmw);
				if (is_migration_entry(pmd_to_swp_entry(*pvmw->pmd))) {
					swp_entry_t entry = pmd_to_swp_entry(*pvmw->pmd);

					if (migration_entry_to_page(entry) != page)
						return not_found(pvmw);
					return true;
				}
			} else
				WARN_ONCE(1, "Non present huge pmd without pmd migration enabled!");
			return not_found(pvmw);
		} else {
			/* THP pmd was split under us: handle on pte level */
			spin_unlock(pvmw->ptl);
			pvmw->ptl = NULL;
		}
	} else {
		if (!check_pmd(pvmw))
			return false;
	}
	if (!map_pte(pvmw))
		goto next_pte;
	while (1) {
		if (check_pte(pvmw))
			return true;
next_pte:
		/* Seek to next pte only makes sense for THP */
		if (!PageTransHuge(pvmw->page) || PageHuge(pvmw->page))
			return not_found(pvmw);
		do {
			pvmw->address += PAGE_SIZE;
			if (pvmw->address >= pvmw->vma->vm_end ||
			    pvmw->address >=
					__vma_address(pvmw->page, pvmw->vma) +
					hpage_nr_pages(pvmw->page) * PAGE_SIZE)
				return not_found(pvmw);
			/* Did we cross page table boundary? */
			if (pvmw->address % PMD_SIZE == 0) {
				pte_unmap(pvmw->pte);
				if (pvmw->ptl) {
					spin_unlock(pvmw->ptl);
					pvmw->ptl = NULL;
				}
				goto restart;
			} else {
				pvmw->pte++;
			}
		} while (pte_none(*pvmw->pte));

		if (!pvmw->ptl) {
			pvmw->ptl = pte_lockptr(mm, pvmw->pmd);
			spin_lock(pvmw->ptl);
		}
	}
}

/**
 * page_mapped_in_vma - check whether a page is really mapped in a VMA
 * @page: the page to test
 * @vma: the VMA to test
 *
 * Returns 1 if the page is mapped into the page tables of the VMA, 0
 * if the page is not mapped into the page tables of this VMA.  Only
 * valid for normal file or anonymous VMAs.
 */
int page_mapped_in_vma(struct page *page, struct vm_area_struct *vma)
{
	struct page_vma_mapped_walk pvmw = {
		.page = page,
		.vma = vma,
		.flags = PVMW_SYNC,
	};
	unsigned long start, end;

	start = __vma_address(page, vma);
	end = start + PAGE_SIZE * (hpage_nr_pages(page) - 1);

	if (unlikely(end < vma->vm_start || start >= vma->vm_end))
		return 0;
	pvmw.address = max(start, vma->vm_start);
	if (!page_vma_mapped_walk(&pvmw))
		return 0;
	page_vma_mapped_walk_done(&pvmw);
	return 1;
}
Commit	Line	Data
ace71a19 KS	1	#include <linux/mm.h>
	2	#include <linux/rmap.h>
	3	#include <linux/hugetlb.h>
	4	#include <linux/swap.h>
	5	#include <linux/swapops.h>
	6
	7	#include "internal.h"
	8
	9	static inline bool check_pmd(struct page_vma_mapped_walk *pvmw)
	10	{
	11	pmd_t pmde;
	12	/*
	13	* Make sure we don't re-load pmd between present and !trans_huge check.
	14	* We need a consistent view.
	15	*/
	16	pmde = READ_ONCE(*pvmw->pmd);
	17	return pmd_present(pmde) && !pmd_trans_huge(pmde);
	18	}
	19
	20	static inline bool not_found(struct page_vma_mapped_walk *pvmw)
	21	{
	22	page_vma_mapped_walk_done(pvmw);
	23	return false;
	24	}
	25
	26	static bool map_pte(struct page_vma_mapped_walk *pvmw)
	27	{
	28	pvmw->pte = pte_offset_map(pvmw->pmd, pvmw->address);
	29	if (!(pvmw->flags & PVMW_SYNC)) {
	30	if (pvmw->flags & PVMW_MIGRATION) {
	31	if (!is_swap_pte(*pvmw->pte))
	32	return false;
	33	} else {
	34	if (!pte_present(*pvmw->pte))
	35	return false;
	36	}
	37	}
	38	pvmw->ptl = pte_lockptr(pvmw->vma->vm_mm, pvmw->pmd);
	39	spin_lock(pvmw->ptl);
	40	return true;
	41	}
	42
	43	static bool check_pte(struct page_vma_mapped_walk *pvmw)
	44	{
	45	if (pvmw->flags & PVMW_MIGRATION) {
	46	#ifdef CONFIG_MIGRATION
	47	swp_entry_t entry;
	48	if (!is_swap_pte(*pvmw->pte))
	49	return false;
	50	entry = pte_to_swp_entry(*pvmw->pte);
a5430dda	51
ace71a19 KS	52	if (!is_migration_entry(entry))
	53	return false;
	54	if (migration_entry_to_page(entry) - pvmw->page >=
	55	hpage_nr_pages(pvmw->page)) {
	56	return false;
	57	}
	58	if (migration_entry_to_page(entry) < pvmw->page)
	59	return false;
	60	#else
	61	WARN_ON_ONCE(1);
	62	#endif
	63	} else {
a5430dda JG	64	if (is_swap_pte(*pvmw->pte)) {
	65	swp_entry_t entry;
	66
	67	entry = pte_to_swp_entry(*pvmw->pte);
	68	if (is_device_private_entry(entry) &&
	69	device_private_entry_to_page(entry) == pvmw->page)
	70	return true;
	71	}
	72
ace71a19 KS	73	if (!pte_present(*pvmw->pte))
	74	return false;
	75
	76	/* THP can be referenced by any subpage */
	77	if (pte_page(*pvmw->pte) - pvmw->page >=
	78	hpage_nr_pages(pvmw->page)) {
	79	return false;
	80	}
	81	if (pte_page(*pvmw->pte) < pvmw->page)
	82	return false;
	83	}
	84
	85	return true;
	86	}
	87
	88	/**
	89	* page_vma_mapped_walk - check if @pvmw->page is mapped in @pvmw->vma at
	90	* @pvmw->address
	91	* @pvmw: pointer to struct page_vma_mapped_walk. page, vma, address and flags
	92	* must be set. pmd, pte and ptl must be NULL.
	93	*
	94	* Returns true if the page is mapped in the vma. @pvmw->pmd and @pvmw->pte point
	95	* to relevant page table entries. @pvmw->ptl is locked. @pvmw->address is
	96	* adjusted if needed (for PTE-mapped THPs).
	97	*
	98	* If @pvmw->pmd is set but @pvmw->pte is not, you have found PMD-mapped page
	99	* (usually THP). For PTE-mapped THP, you should run page_vma_mapped_walk() in
	100	* a loop to find all PTEs that map the THP.
	101	*
	102	* For HugeTLB pages, @pvmw->pte is set to the relevant page table entry
	103	* regardless of which page table level the page is mapped at. @pvmw->pmd is
	104	* NULL.
	105	*
	106	* Retruns false if there are no more page table entries for the page in
	107	* the vma. @pvmw->ptl is unlocked and @pvmw->pte is unmapped.
	108	*
	109	* If you need to stop the walk before page_vma_mapped_walk() returned false,
	110	* use page_vma_mapped_walk_done(). It will do the housekeeping.
	111	*/
	112	bool page_vma_mapped_walk(struct page_vma_mapped_walk *pvmw)
	113	{
	114	struct mm_struct *mm = pvmw->vma->vm_mm;
	115	struct page *page = pvmw->page;
	116	pgd_t *pgd;
c2febafc	117	p4d_t *p4d;
ace71a19 KS	118	pud_t *pud;
	119
	120	/* The only possible pmd mapping has been handled on last iteration */
	121	if (pvmw->pmd && !pvmw->pte)
	122	return not_found(pvmw);
	123
d75450ff	124	if (pvmw->pte)
ace71a19	125	goto next_pte;
ace71a19 KS	126
	127	if (unlikely(PageHuge(pvmw->page))) {
	128	/* when pud is not present, pte will be NULL */
7868a208 PA	129	pvmw->pte = huge_pte_offset(mm, pvmw->address,
7868a208 PA	130	PAGE_SIZE << compound_order(page));
ace71a19 KS	131	if (!pvmw->pte)
	132	return false;
	133
	134	pvmw->ptl = huge_pte_lockptr(page_hstate(page), mm, pvmw->pte);
	135	spin_lock(pvmw->ptl);
	136	if (!check_pte(pvmw))
	137	return not_found(pvmw);
	138	return true;
	139	}
	140	restart:
	141	pgd = pgd_offset(mm, pvmw->address);
	142	if (!pgd_present(*pgd))
	143	return false;
c2febafc KS	144	p4d = p4d_offset(pgd, pvmw->address);
	145	if (!p4d_present(*p4d))
	146	return false;
	147	pud = pud_offset(p4d, pvmw->address);
ace71a19 KS	148	if (!pud_present(*pud))
	149	return false;
	150	pvmw->pmd = pmd_offset(pud, pvmw->address);
616b8371	151	if (pmd_trans_huge(pvmw->pmd) \|\| is_pmd_migration_entry(pvmw->pmd)) {
ace71a19	152	pvmw->ptl = pmd_lock(mm, pvmw->pmd);
ace71a19 KS	153	if (likely(pmd_trans_huge(*pvmw->pmd))) {
	154	if (pvmw->flags & PVMW_MIGRATION)
	155	return not_found(pvmw);
	156	if (pmd_page(*pvmw->pmd) != page)
	157	return not_found(pvmw);
	158	return true;
616b8371 ZY	159	} else if (!pmd_present(*pvmw->pmd)) {
	160	if (thp_migration_supported()) {
	161	if (!(pvmw->flags & PVMW_MIGRATION))
	162	return not_found(pvmw);
	163	if (is_migration_entry(pmd_to_swp_entry(*pvmw->pmd))) {
	164	swp_entry_t entry = pmd_to_swp_entry(*pvmw->pmd);
	165
	166	if (migration_entry_to_page(entry) != page)
	167	return not_found(pvmw);
	168	return true;
	169	}
	170	} else
	171	WARN_ONCE(1, "Non present huge pmd without pmd migration enabled!");
	172	return not_found(pvmw);
ace71a19 KS	173	} else {
	174	/* THP pmd was split under us: handle on pte level */
	175	spin_unlock(pvmw->ptl);
	176	pvmw->ptl = NULL;
	177	}
	178	} else {
	179	if (!check_pmd(pvmw))
	180	return false;
	181	}
	182	if (!map_pte(pvmw))
	183	goto next_pte;
	184	while (1) {
	185	if (check_pte(pvmw))
	186	return true;
d75450ff HD	187	next_pte:
	188	/* Seek to next pte only makes sense for THP */
	189	if (!PageTransHuge(pvmw->page) \|\| PageHuge(pvmw->page))
	190	return not_found(pvmw);
	191	do {
ace71a19	192	pvmw->address += PAGE_SIZE;
d75450ff HD	193	if (pvmw->address >= pvmw->vma->vm_end \|\|
d75450ff HD	194	pvmw->address >=
ace71a19 KS	195	__vma_address(pvmw->page, pvmw->vma) +
	196	hpage_nr_pages(pvmw->page) * PAGE_SIZE)
	197	return not_found(pvmw);
	198	/* Did we cross page table boundary? */
	199	if (pvmw->address % PMD_SIZE == 0) {
	200	pte_unmap(pvmw->pte);
	201	if (pvmw->ptl) {
	202	spin_unlock(pvmw->ptl);
	203	pvmw->ptl = NULL;
	204	}
	205	goto restart;
	206	} else {
	207	pvmw->pte++;
	208	}
	209	} while (pte_none(*pvmw->pte));
	210
	211	if (!pvmw->ptl) {
	212	pvmw->ptl = pte_lockptr(mm, pvmw->pmd);
	213	spin_lock(pvmw->ptl);
	214	}
	215	}
	216	}
6a328a62 KS	217
	218	/**
	219	* page_mapped_in_vma - check whether a page is really mapped in a VMA
	220	* @page: the page to test
	221	* @vma: the VMA to test
	222	*
	223	* Returns 1 if the page is mapped into the page tables of the VMA, 0
	224	* if the page is not mapped into the page tables of this VMA. Only
	225	* valid for normal file or anonymous VMAs.
	226	*/
	227	int page_mapped_in_vma(struct page page, struct vm_area_struct vma)
	228	{
	229	struct page_vma_mapped_walk pvmw = {
	230	.page = page,
	231	.vma = vma,
	232	.flags = PVMW_SYNC,
	233	};
	234	unsigned long start, end;
	235
	236	start = __vma_address(page, vma);
	237	end = start + PAGE_SIZE * (hpage_nr_pages(page) - 1);
	238
	239	if (unlikely(end < vma->vm_start \|\| start >= vma->vm_end))
	240	return 0;
	241	pvmw.address = max(start, vma->vm_start);
	242	if (!page_vma_mapped_walk(&pvmw))
	243	return 0;
	244	page_vma_mapped_walk_done(&pvmw);
	245	return 1;
	246	}