[mirror_ubuntu-focal-kernel.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_in_hpage(struct page *hpage, unsigned long pfn)
{
	unsigned long hpage_pfn = page_to_pfn(hpage);

	/* THP can be referenced by any subpage */
	return pfn >= hpage_pfn && pfn - hpage_pfn < hpage_nr_pages(hpage);
}

/**
 * check_pte - check if @pvmw->page is mapped at the @pvmw->pte
 *
 * 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_in_hpage(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.
 *
 * 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;
	unsigned long end;
	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;
	}

	/*
	 * Seek to next pte only makes sense for THP.
	 * But more important than that optimization, is to filter out
	 * any PageKsm page: whose page->index misleads vma_address()
	 * and vma_address_end() to disaster.
	 */
	end = PageTransCompound(page) ?
		vma_address_end(page, pvmw->vma) :
		pvmw->address + PAGE_SIZE;
	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;
this_pte:
		if (check_pte(pvmw))
			return true;
next_pte:
		do {
			pvmw->address += PAGE_SIZE;
			if (pvmw->address >= end)
				return not_found(pvmw);
			/* Did we cross page table boundary? */
			if ((pvmw->address & (PMD_SIZE - PAGE_SIZE)) == 0) {
				if (pvmw->ptl) {
					spin_unlock(pvmw->ptl);
					pvmw->ptl = NULL;
				}
				pte_unmap(pvmw->pte);
				pvmw->pte = NULL;
				goto restart;
			}
			pvmw->pte++;
		} while (pte_none(*pvmw->pte));

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

/**
 * 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
7222708e KS	55	static inline bool pfn_in_hpage(struct page *hpage, unsigned long pfn)
	56	{
	57	unsigned long hpage_pfn = page_to_pfn(hpage);
	58
	59	/* THP can be referenced by any subpage */
	60	return pfn >= hpage_pfn && pfn - hpage_pfn < hpage_nr_pages(hpage);
	61	}
	62
0d665e7b KS	63	/**
	64	* check_pte - check if @pvmw->page is mapped at the @pvmw->pte
	65	*
	66	* page_vma_mapped_walk() found a place where @pvmw->page is potentially
	67	* mapped. check_pte() has to validate this.
	68	*
	69	* @pvmw->pte may point to empty PTE, swap PTE or PTE pointing to arbitrary
	70	* page.
	71	*
	72	* If PVMW_MIGRATION flag is set, returns true if @pvmw->pte contains migration
	73	* entry that points to @pvmw->page or any subpage in case of THP.
	74	*
	75	* If PVMW_MIGRATION flag is not set, returns true if @pvmw->pte points to
	76	* @pvmw->page or any subpage in case of THP.
	77	*
	78	* Otherwise, return false.
	79	*
	80	*/
ace71a19 KS	81	static bool check_pte(struct page_vma_mapped_walk *pvmw)
ace71a19 KS	82	{
0d665e7b KS	83	unsigned long pfn;
0d665e7b KS	84
ace71a19	85	if (pvmw->flags & PVMW_MIGRATION) {
ace71a19 KS	86	swp_entry_t entry;
	87	if (!is_swap_pte(*pvmw->pte))
	88	return false;
	89	entry = pte_to_swp_entry(*pvmw->pte);
a5430dda	90
ace71a19 KS	91	if (!is_migration_entry(entry))
ace71a19 KS	92	return false;
a5430dda	93
0d665e7b KS	94	pfn = migration_entry_to_pfn(entry);
	95	} else if (is_swap_pte(*pvmw->pte)) {
	96	swp_entry_t entry;
a5430dda	97
0d665e7b KS	98	/* Handle un-addressable ZONE_DEVICE memory */
	99	entry = pte_to_swp_entry(*pvmw->pte);
	100	if (!is_device_private_entry(entry))
ace71a19 KS	101	return false;
ace71a19 KS	102
0d665e7b KS	103	pfn = device_private_entry_to_pfn(entry);
	104	} else {
	105	if (!pte_present(*pvmw->pte))
ace71a19	106	return false;
0d665e7b KS	107
0d665e7b KS	108	pfn = pte_pfn(*pvmw->pte);
ace71a19 KS	109	}
ace71a19 KS	110
7222708e	111	return pfn_in_hpage(pvmw->page, pfn);
ace71a19 KS	112	}
	113
	114	/**
	115	* page_vma_mapped_walk - check if @pvmw->page is mapped in @pvmw->vma at
	116	* @pvmw->address
	117	* @pvmw: pointer to struct page_vma_mapped_walk. page, vma, address and flags
	118	* must be set. pmd, pte and ptl must be NULL.
	119	*
	120	* Returns true if the page is mapped in the vma. @pvmw->pmd and @pvmw->pte point
	121	* to relevant page table entries. @pvmw->ptl is locked. @pvmw->address is
	122	* adjusted if needed (for PTE-mapped THPs).
	123	*
	124	* If @pvmw->pmd is set but @pvmw->pte is not, you have found PMD-mapped page
	125	* (usually THP). For PTE-mapped THP, you should run page_vma_mapped_walk() in
	126	* a loop to find all PTEs that map the THP.
	127	*
	128	* For HugeTLB pages, @pvmw->pte is set to the relevant page table entry
	129	* regardless of which page table level the page is mapped at. @pvmw->pmd is
	130	* NULL.
	131	*
	132	* Retruns false if there are no more page table entries for the page in
	133	* the vma. @pvmw->ptl is unlocked and @pvmw->pte is unmapped.
	134	*
	135	* If you need to stop the walk before page_vma_mapped_walk() returned false,
	136	* use page_vma_mapped_walk_done(). It will do the housekeeping.
	137	*/
	138	bool page_vma_mapped_walk(struct page_vma_mapped_walk *pvmw)
	139	{
	140	struct mm_struct *mm = pvmw->vma->vm_mm;
	141	struct page *page = pvmw->page;
81c31e3f	142	unsigned long end;
ace71a19	143	pgd_t *pgd;
c2febafc	144	p4d_t *p4d;
ace71a19	145	pud_t *pud;
a7b10095	146	pmd_t pmde;
ace71a19 KS	147
	148	/* The only possible pmd mapping has been handled on last iteration */
	149	if (pvmw->pmd && !pvmw->pte)
	150	return not_found(pvmw);
	151
95ae8818	152	if (unlikely(PageHuge(page))) {
bdac6048 HD	153	/* The only possible mapping was handled on last iteration */
	154	if (pvmw->pte)
	155	return not_found(pvmw);
	156
ace71a19	157	/* when pud is not present, pte will be NULL */
a50b854e	158	pvmw->pte = huge_pte_offset(mm, pvmw->address, page_size(page));
ace71a19 KS	159	if (!pvmw->pte)
	160	return false;
	161
	162	pvmw->ptl = huge_pte_lockptr(page_hstate(page), mm, pvmw->pte);
	163	spin_lock(pvmw->ptl);
	164	if (!check_pte(pvmw))
	165	return not_found(pvmw);
	166	return true;
	167	}
bdac6048	168
f3592d7e HD	169	/*
	170	* Seek to next pte only makes sense for THP.
	171	* But more important than that optimization, is to filter out
	172	* any PageKsm page: whose page->index misleads vma_address()
	173	* and vma_address_end() to disaster.
	174	*/
	175	end = PageTransCompound(page) ?
	176	vma_address_end(page, pvmw->vma) :
	177	pvmw->address + PAGE_SIZE;
bdac6048 HD	178	if (pvmw->pte)
bdac6048 HD	179	goto next_pte;
ace71a19	180	restart:
74430a62 HD	181	{
	182	pgd = pgd_offset(mm, pvmw->address);
	183	if (!pgd_present(*pgd))
	184	return false;
	185	p4d = p4d_offset(pgd, pvmw->address);
	186	if (!p4d_present(*p4d))
	187	return false;
	188	pud = pud_offset(p4d, pvmw->address);
	189	if (!pud_present(*pud))
	190	return false;
b8ead7b3	191
74430a62	192	pvmw->pmd = pmd_offset(pud, pvmw->address);
10f78390	193	/*
74430a62 HD	194	* Make sure the pmd value isn't cached in a register by the
	195	* compiler and used as a stale value after we've observed a
	196	* subsequent update.
10f78390	197	*/
74430a62 HD	198	pmde = READ_ONCE(*pvmw->pmd);
	199
	200	if (pmd_trans_huge(pmde) \|\| is_pmd_migration_entry(pmde)) {
	201	pvmw->ptl = pmd_lock(mm, pvmw->pmd);
	202	pmde = *pvmw->pmd;
	203	if (likely(pmd_trans_huge(pmde))) {
	204	if (pvmw->flags & PVMW_MIGRATION)
	205	return not_found(pvmw);
	206	if (pmd_page(pmde) != page)
	207	return not_found(pvmw);
	208	return true;
	209	}
	210	if (!pmd_present(pmde)) {
	211	swp_entry_t entry;
	212
	213	if (!thp_migration_supported() \|\|
	214	!(pvmw->flags & PVMW_MIGRATION))
	215	return not_found(pvmw);
	216	entry = pmd_to_swp_entry(pmde);
	217	if (!is_migration_entry(entry) \|\|
	218	migration_entry_to_page(entry) != page)
	219	return not_found(pvmw);
	220	return true;
	221	}
	222	/* THP pmd was split under us: handle on pte level */
	223	spin_unlock(pvmw->ptl);
	224	pvmw->ptl = NULL;
	225	} else if (!pmd_present(pmde)) {
	226	/*
	227	* If PVMW_SYNC, take and drop THP pmd lock so that we
	228	* cannot return prematurely, while zap_huge_pmd() has
	229	* cleared *pmd but not decremented compound_mapcount().
	230	*/
	231	if ((pvmw->flags & PVMW_SYNC) &&
	232	PageTransCompound(page)) {
	233	spinlock_t *ptl = pmd_lock(mm, pvmw->pmd);
10f78390	234
74430a62 HD	235	spin_unlock(ptl);
	236	}
	237	return false;
10f78390	238	}
74430a62 HD	239	if (!map_pte(pvmw))
74430a62 HD	240	goto next_pte;
81c31e3f	241	this_pte:
ace71a19 KS	242	if (check_pte(pvmw))
ace71a19 KS	243	return true;
d75450ff	244	next_pte:
d75450ff	245	do {
ace71a19	246	pvmw->address += PAGE_SIZE;
5bbd1b12	247	if (pvmw->address >= end)
ace71a19 KS	248	return not_found(pvmw);
ace71a19 KS	249	/* Did we cross page table boundary? */
c99dcd7f	250	if ((pvmw->address & (PMD_SIZE - PAGE_SIZE)) == 0) {
ace71a19 KS	251	if (pvmw->ptl) {
	252	spin_unlock(pvmw->ptl);
	253	pvmw->ptl = NULL;
	254	}
c99dcd7f HD	255	pte_unmap(pvmw->pte);
c99dcd7f HD	256	pvmw->pte = NULL;
ace71a19	257	goto restart;
ace71a19	258	}
c99dcd7f	259	pvmw->pte++;
ace71a19 KS	260	} while (pte_none(*pvmw->pte));
	261
	262	if (!pvmw->ptl) {
	263	pvmw->ptl = pte_lockptr(mm, pvmw->pmd);
	264	spin_lock(pvmw->ptl);
	265	}
81c31e3f	266	goto this_pte;
ace71a19 KS	267	}
ace71a19 KS	268	}
6a328a62 KS	269
	270	/**
	271	* page_mapped_in_vma - check whether a page is really mapped in a VMA
	272	* @page: the page to test
	273	* @vma: the VMA to test
	274	*
	275	* Returns 1 if the page is mapped into the page tables of the VMA, 0
	276	* if the page is not mapped into the page tables of this VMA. Only
	277	* valid for normal file or anonymous VMAs.
	278	*/
	279	int page_mapped_in_vma(struct page page, struct vm_area_struct vma)
	280	{
	281	struct page_vma_mapped_walk pvmw = {
	282	.page = page,
	283	.vma = vma,
	284	.flags = PVMW_SYNC,
	285	};
6a328a62	286
5bbd1b12 HD	287	pvmw.address = vma_address(page, vma);
5bbd1b12 HD	288	if (pvmw.address == -EFAULT)
6a328a62	289	return 0;
6a328a62 KS	290	if (!page_vma_mapped_walk(&pvmw))
	291	return 0;
	292	page_vma_mapped_walk_done(&pvmw);
	293	return 1;
	294	}