[linux-block.git] / mm / page_vma_mapped.c

// SPDX-License-Identifier: GPL-2.0
#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 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 {
			/*
			 * We get here when we are trying to unmap a private
			 * device page from the process address space. Such
			 * page is not CPU accessible and thus is mapped as
			 * a special swap entry, nonetheless it still does
			 * count as a valid regular mapping for the page (and
			 * is accounted as such in page maps count).
			 *
			 * So handle this special case as if it was a normal
			 * page mapping ie lock CPU page table and returns
			 * true.
			 *
			 * For more details on device private memory see HMM
			 * (include/linux/hmm.h or mm/hmm.c).
			 */
			if (is_swap_pte(*pvmw->pte)) {
				swp_entry_t entry;

				/* Handle un-addressable ZONE_DEVICE memory */
				entry = pte_to_swp_entry(*pvmw->pte);
				if (!is_device_private_entry(entry))
					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 inline bool pfn_is_match(struct page *page, unsigned long pfn)
{
	unsigned long page_pfn = page_to_pfn(page);

	/* normal page and hugetlbfs page */
	if (!PageTransCompound(page) || PageHuge(page))
		return page_pfn == pfn;

	/* THP can be referenced by any subpage */
	return pfn >= page_pfn && pfn - page_pfn < thp_nr_pages(page);
}

/**
 * check_pte - check if @pvmw->page is mapped at the @pvmw->pte
 * @pvmw: page_vma_mapped_walk struct, includes a pair pte and page for checking
 *
 * page_vma_mapped_walk() found a place where @pvmw->page is *potentially*
 * mapped. check_pte() has to validate this.
 *
 * pvmw->pte may point to empty PTE, swap PTE or PTE pointing to
 * arbitrary page.
 *
 * If PVMW_MIGRATION flag is set, returns true if @pvmw->pte contains migration
 * entry that points to @pvmw->page or any subpage in case of THP.
 *
 * If PVMW_MIGRATION flag is not set, returns true if pvmw->pte points to
 * pvmw->page or any subpage in case of THP.
 *
 * Otherwise, return false.
 *
 */
static bool check_pte(struct page_vma_mapped_walk *pvmw)
{
	unsigned long pfn;

	if (pvmw->flags & PVMW_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;

		pfn = migration_entry_to_pfn(entry);
	} else if (is_swap_pte(*pvmw->pte)) {
		swp_entry_t entry;

		/* Handle un-addressable ZONE_DEVICE memory */
		entry = pte_to_swp_entry(*pvmw->pte);
		if (!is_device_private_entry(entry))
			return false;

		pfn = device_private_entry_to_pfn(entry);
	} else {
		if (!pte_present(*pvmw->pte))
			return false;

		pfn = pte_pfn(*pvmw->pte);
	}

	return pfn_is_match(pvmw->page, pfn);
}

/**
 * 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.
 *
 * Returns 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;
	pmd_t pmde;

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

	if (unlikely(PageHuge(page))) {
		/* The only possible mapping was handled on last iteration */
		if (pvmw->pte)
			return not_found(pvmw);

		/* when pud is not present, pte will be NULL */
		pvmw->pte = huge_pte_offset(mm, pvmw->address, page_size(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;
	}

	if (pvmw->pte)
		goto next_pte;
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);
	/*
	 * Make sure the pmd value isn't cached in a register by the
	 * compiler and used as a stale value after we've observed a
	 * subsequent update.
	 */
	pmde = READ_ONCE(*pvmw->pmd);
	if (pmd_trans_huge(pmde) || is_pmd_migration_entry(pmde)) {
		pvmw->ptl = pmd_lock(mm, pvmw->pmd);
		pmde = *pvmw->pmd;
		if (likely(pmd_trans_huge(pmde))) {
			if (pvmw->flags & PVMW_MIGRATION)
				return not_found(pvmw);
			if (pmd_page(pmde) != page)
				return not_found(pvmw);
			return true;
		}
		if (!pmd_present(pmde)) {
			swp_entry_t entry;

			if (!thp_migration_supported() ||
			    !(pvmw->flags & PVMW_MIGRATION))
				return not_found(pvmw);
			entry = pmd_to_swp_entry(pmde);
			if (!is_migration_entry(entry) ||
			    migration_entry_to_page(entry) != page)
				return not_found(pvmw);
			return true;
		}
		/* THP pmd was split under us: handle on pte level */
		spin_unlock(pvmw->ptl);
		pvmw->ptl = NULL;
	} else if (!pmd_present(pmde)) {
		/*
		 * If PVMW_SYNC, take and drop THP pmd lock so that we
		 * cannot return prematurely, while zap_huge_pmd() has
		 * cleared *pmd but not decremented compound_mapcount().
		 */
		if ((pvmw->flags & PVMW_SYNC) && PageTransCompound(page)) {
			spinlock_t *ptl = pmd_lock(mm, pvmw->pmd);

			spin_unlock(ptl);
		}
		return false;
	}
	if (!map_pte(pvmw))
		goto next_pte;
	while (1) {
		unsigned long end;

		if (check_pte(pvmw))
			return true;
next_pte:
		/* Seek to next pte only makes sense for THP */
		if (!PageTransHuge(page))
			return not_found(pvmw);
		end = vma_address_end(page, pvmw->vma);
		do {
			pvmw->address += PAGE_SIZE;
			if (pvmw->address >= end)
				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,
	};

	pvmw.address = vma_address(page, vma);
	if (pvmw.address == -EFAULT)
		return 0;
	if (!page_vma_mapped_walk(&pvmw))
		return 0;
	page_vma_mapped_walk_done(&pvmw);
	return 1;
}
Commit	Line	Data
b2441318	1	// SPDX-License-Identifier: GPL-2.0
ace71a19 KS	2	#include <linux/mm.h>
	3	#include <linux/rmap.h>
	4	#include <linux/hugetlb.h>
	5	#include <linux/swap.h>
	6	#include <linux/swapops.h>
	7
	8	#include "internal.h"
	9
ace71a19 KS	10	static inline bool not_found(struct page_vma_mapped_walk *pvmw)
	11	{
	12	page_vma_mapped_walk_done(pvmw);
	13	return false;
	14	}
	15
	16	static bool map_pte(struct page_vma_mapped_walk *pvmw)
	17	{
	18	pvmw->pte = pte_offset_map(pvmw->pmd, pvmw->address);
	19	if (!(pvmw->flags & PVMW_SYNC)) {
	20	if (pvmw->flags & PVMW_MIGRATION) {
	21	if (!is_swap_pte(*pvmw->pte))
	22	return false;
	23	} else {
aab8d052 RC	24	/*
	25	* We get here when we are trying to unmap a private
	26	* device page from the process address space. Such
	27	* page is not CPU accessible and thus is mapped as
	28	* a special swap entry, nonetheless it still does
	29	* count as a valid regular mapping for the page (and
	30	* is accounted as such in page maps count).
	31	*
	32	* So handle this special case as if it was a normal
	33	* page mapping ie lock CPU page table and returns
	34	* true.
	35	*
	36	* For more details on device private memory see HMM
	37	* (include/linux/hmm.h or mm/hmm.c).
	38	*/
	39	if (is_swap_pte(*pvmw->pte)) {
	40	swp_entry_t entry;
	41
	42	/* Handle un-addressable ZONE_DEVICE memory */
	43	entry = pte_to_swp_entry(*pvmw->pte);
	44	if (!is_device_private_entry(entry))
	45	return false;
	46	} else if (!pte_present(*pvmw->pte))
ace71a19 KS	47	return false;
	48	}
	49	}
	50	pvmw->ptl = pte_lockptr(pvmw->vma->vm_mm, pvmw->pmd);
	51	spin_lock(pvmw->ptl);
	52	return true;
	53	}
	54
5b8d6e37	55	static inline bool pfn_is_match(struct page *page, unsigned long pfn)
7222708e	56	{
5b8d6e37 LX	57	unsigned long page_pfn = page_to_pfn(page);
	58
	59	/* normal page and hugetlbfs page */
	60	if (!PageTransCompound(page) \|\| PageHuge(page))
	61	return page_pfn == pfn;
7222708e KS	62
7222708e KS	63	/* THP can be referenced by any subpage */
6c357848	64	return pfn >= page_pfn && pfn - page_pfn < thp_nr_pages(page);
7222708e KS	65	}
7222708e KS	66
0d665e7b KS	67	/**
0d665e7b KS	68	* check_pte - check if @pvmw->page is mapped at the @pvmw->pte
777f303c	69	* @pvmw: page_vma_mapped_walk struct, includes a pair pte and page for checking
0d665e7b KS	70	*
	71	* page_vma_mapped_walk() found a place where @pvmw->page is potentially
	72	* mapped. check_pte() has to validate this.
	73	*
777f303c AS	74	* pvmw->pte may point to empty PTE, swap PTE or PTE pointing to
777f303c AS	75	* arbitrary page.
0d665e7b KS	76	*
	77	* If PVMW_MIGRATION flag is set, returns true if @pvmw->pte contains migration
	78	* entry that points to @pvmw->page or any subpage in case of THP.
	79	*
777f303c AS	80	* If PVMW_MIGRATION flag is not set, returns true if pvmw->pte points to
777f303c AS	81	* pvmw->page or any subpage in case of THP.
0d665e7b KS	82	*
	83	* Otherwise, return false.
	84	*
	85	*/
ace71a19 KS	86	static bool check_pte(struct page_vma_mapped_walk *pvmw)
ace71a19 KS	87	{
0d665e7b KS	88	unsigned long pfn;
0d665e7b KS	89
ace71a19	90	if (pvmw->flags & PVMW_MIGRATION) {
ace71a19 KS	91	swp_entry_t entry;
	92	if (!is_swap_pte(*pvmw->pte))
	93	return false;
	94	entry = pte_to_swp_entry(*pvmw->pte);
a5430dda	95
ace71a19 KS	96	if (!is_migration_entry(entry))
ace71a19 KS	97	return false;
a5430dda	98
0d665e7b KS	99	pfn = migration_entry_to_pfn(entry);
	100	} else if (is_swap_pte(*pvmw->pte)) {
	101	swp_entry_t entry;
a5430dda	102
0d665e7b KS	103	/* Handle un-addressable ZONE_DEVICE memory */
	104	entry = pte_to_swp_entry(*pvmw->pte);
	105	if (!is_device_private_entry(entry))
ace71a19 KS	106	return false;
ace71a19 KS	107
0d665e7b KS	108	pfn = device_private_entry_to_pfn(entry);
	109	} else {
	110	if (!pte_present(*pvmw->pte))
ace71a19	111	return false;
0d665e7b KS	112
0d665e7b KS	113	pfn = pte_pfn(*pvmw->pte);
ace71a19 KS	114	}
ace71a19 KS	115
5b8d6e37	116	return pfn_is_match(pvmw->page, pfn);
ace71a19 KS	117	}
	118
	119	/**
	120	* page_vma_mapped_walk - check if @pvmw->page is mapped in @pvmw->vma at
	121	* @pvmw->address
	122	* @pvmw: pointer to struct page_vma_mapped_walk. page, vma, address and flags
	123	* must be set. pmd, pte and ptl must be NULL.
	124	*
	125	* Returns true if the page is mapped in the vma. @pvmw->pmd and @pvmw->pte point
	126	* to relevant page table entries. @pvmw->ptl is locked. @pvmw->address is
	127	* adjusted if needed (for PTE-mapped THPs).
	128	*
	129	* If @pvmw->pmd is set but @pvmw->pte is not, you have found PMD-mapped page
	130	* (usually THP). For PTE-mapped THP, you should run page_vma_mapped_walk() in
	131	* a loop to find all PTEs that map the THP.
	132	*
	133	* For HugeTLB pages, @pvmw->pte is set to the relevant page table entry
	134	* regardless of which page table level the page is mapped at. @pvmw->pmd is
	135	* NULL.
	136	*
baf2f90b	137	* Returns false if there are no more page table entries for the page in
ace71a19 KS	138	* the vma. @pvmw->ptl is unlocked and @pvmw->pte is unmapped.
	139	*
	140	* If you need to stop the walk before page_vma_mapped_walk() returned false,
	141	* use page_vma_mapped_walk_done(). It will do the housekeeping.
	142	*/
	143	bool page_vma_mapped_walk(struct page_vma_mapped_walk *pvmw)
	144	{
	145	struct mm_struct *mm = pvmw->vma->vm_mm;
	146	struct page *page = pvmw->page;
	147	pgd_t *pgd;
c2febafc	148	p4d_t *p4d;
ace71a19	149	pud_t *pud;
a7b10095	150	pmd_t pmde;
ace71a19 KS	151
	152	/* The only possible pmd mapping has been handled on last iteration */
	153	if (pvmw->pmd && !pvmw->pte)
	154	return not_found(pvmw);
	155
f003c03b	156	if (unlikely(PageHuge(page))) {
6d0fd598 HD	157	/* The only possible mapping was handled on last iteration */
	158	if (pvmw->pte)
	159	return not_found(pvmw);
	160
ace71a19	161	/* when pud is not present, pte will be NULL */
a50b854e	162	pvmw->pte = huge_pte_offset(mm, pvmw->address, page_size(page));
ace71a19 KS	163	if (!pvmw->pte)
	164	return false;
	165
	166	pvmw->ptl = huge_pte_lockptr(page_hstate(page), mm, pvmw->pte);
	167	spin_lock(pvmw->ptl);
	168	if (!check_pte(pvmw))
	169	return not_found(pvmw);
	170	return true;
	171	}
6d0fd598 HD	172
	173	if (pvmw->pte)
	174	goto next_pte;
ace71a19 KS	175	restart:
	176	pgd = pgd_offset(mm, pvmw->address);
	177	if (!pgd_present(*pgd))
	178	return false;
c2febafc KS	179	p4d = p4d_offset(pgd, pvmw->address);
	180	if (!p4d_present(*p4d))
	181	return false;
	182	pud = pud_offset(p4d, pvmw->address);
ace71a19 KS	183	if (!pud_present(*pud))
	184	return false;
	185	pvmw->pmd = pmd_offset(pud, pvmw->address);
a7b10095 WD	186	/*
	187	* Make sure the pmd value isn't cached in a register by the
	188	* compiler and used as a stale value after we've observed a
	189	* subsequent update.
	190	*/
	191	pmde = READ_ONCE(*pvmw->pmd);
	192	if (pmd_trans_huge(pmde) \|\| is_pmd_migration_entry(pmde)) {
ace71a19	193	pvmw->ptl = pmd_lock(mm, pvmw->pmd);
3306d311 HD	194	pmde = *pvmw->pmd;
3306d311 HD	195	if (likely(pmd_trans_huge(pmde))) {
ace71a19 KS	196	if (pvmw->flags & PVMW_MIGRATION)
ace71a19 KS	197	return not_found(pvmw);
3306d311	198	if (pmd_page(pmde) != page)
ace71a19 KS	199	return not_found(pvmw);
ace71a19 KS	200	return true;
ace71a19	201	}
e2e1d407 HD	202	if (!pmd_present(pmde)) {
	203	swp_entry_t entry;
	204
	205	if (!thp_migration_supported() \|\|
	206	!(pvmw->flags & PVMW_MIGRATION))
	207	return not_found(pvmw);
	208	entry = pmd_to_swp_entry(pmde);
	209	if (!is_migration_entry(entry) \|\|
	210	migration_entry_to_page(entry) != page)
	211	return not_found(pvmw);
	212	return true;
	213	}
	214	/* THP pmd was split under us: handle on pte level */
	215	spin_unlock(pvmw->ptl);
	216	pvmw->ptl = NULL;
a7b10095	217	} else if (!pmd_present(pmde)) {
732ed558 HD	218	/*
	219	* If PVMW_SYNC, take and drop THP pmd lock so that we
	220	* cannot return prematurely, while zap_huge_pmd() has
	221	* cleared *pmd but not decremented compound_mapcount().
	222	*/
f003c03b	223	if ((pvmw->flags & PVMW_SYNC) && PageTransCompound(page)) {
732ed558 HD	224	spinlock_t *ptl = pmd_lock(mm, pvmw->pmd);
	225
	226	spin_unlock(ptl);
	227	}
a7b10095	228	return false;
ace71a19 KS	229	}
	230	if (!map_pte(pvmw))
	231	goto next_pte;
	232	while (1) {
494334e4 HD	233	unsigned long end;
494334e4 HD	234
ace71a19 KS	235	if (check_pte(pvmw))
ace71a19 KS	236	return true;
d75450ff HD	237	next_pte:
d75450ff HD	238	/* Seek to next pte only makes sense for THP */
6d0fd598	239	if (!PageTransHuge(page))
d75450ff	240	return not_found(pvmw);
f003c03b	241	end = vma_address_end(page, pvmw->vma);
d75450ff	242	do {
ace71a19	243	pvmw->address += PAGE_SIZE;
494334e4	244	if (pvmw->address >= end)
ace71a19 KS	245	return not_found(pvmw);
	246	/* Did we cross page table boundary? */
	247	if (pvmw->address % PMD_SIZE == 0) {
	248	pte_unmap(pvmw->pte);
	249	if (pvmw->ptl) {
	250	spin_unlock(pvmw->ptl);
	251	pvmw->ptl = NULL;
	252	}
	253	goto restart;
	254	} else {
	255	pvmw->pte++;
	256	}
	257	} while (pte_none(*pvmw->pte));
	258
	259	if (!pvmw->ptl) {
	260	pvmw->ptl = pte_lockptr(mm, pvmw->pmd);
	261	spin_lock(pvmw->ptl);
	262	}
	263	}
	264	}
6a328a62 KS	265
	266	/**
	267	* page_mapped_in_vma - check whether a page is really mapped in a VMA
	268	* @page: the page to test
	269	* @vma: the VMA to test
	270	*
	271	* Returns 1 if the page is mapped into the page tables of the VMA, 0
	272	* if the page is not mapped into the page tables of this VMA. Only
	273	* valid for normal file or anonymous VMAs.
	274	*/
	275	int page_mapped_in_vma(struct page page, struct vm_area_struct vma)
	276	{
	277	struct page_vma_mapped_walk pvmw = {
	278	.page = page,
	279	.vma = vma,
	280	.flags = PVMW_SYNC,
	281	};
6a328a62	282
494334e4 HD	283	pvmw.address = vma_address(page, vma);
494334e4 HD	284	if (pvmw.address == -EFAULT)
6a328a62	285	return 0;
6a328a62 KS	286	if (!page_vma_mapped_walk(&pvmw))
	287	return 0;
	288	page_vma_mapped_walk_done(&pvmw);
	289	return 1;
	290	}