[thirdparty/linux.git] / arch / arm64 / mm / hugetlbpage.c

// SPDX-License-Identifier: GPL-2.0-only
/*
 * arch/arm64/mm/hugetlbpage.c
 *
 * Copyright (C) 2013 Linaro Ltd.
 *
 * Based on arch/x86/mm/hugetlbpage.c.
 */

#include <linux/init.h>
#include <linux/fs.h>
#include <linux/mm.h>
#include <linux/hugetlb.h>
#include <linux/pagemap.h>
#include <linux/err.h>
#include <linux/sysctl.h>
#include <asm/mman.h>
#include <asm/tlb.h>
#include <asm/tlbflush.h>
#include <asm/pgalloc.h>

#ifdef CONFIG_ARCH_ENABLE_HUGEPAGE_MIGRATION
bool arch_hugetlb_migration_supported(struct hstate *h)
{
	size_t pagesize = huge_page_size(h);

	switch (pagesize) {
#ifdef CONFIG_ARM64_4K_PAGES
	case PUD_SIZE:
#endif
	case PMD_SIZE:
	case CONT_PMD_SIZE:
	case CONT_PTE_SIZE:
		return true;
	}
	pr_warn("%s: unrecognized huge page size 0x%lx\n",
			__func__, pagesize);
	return false;
}
#endif

int pmd_huge(pmd_t pmd)
{
	return pmd_val(pmd) && !(pmd_val(pmd) & PMD_TABLE_BIT);
}

int pud_huge(pud_t pud)
{
#ifndef __PAGETABLE_PMD_FOLDED
	return pud_val(pud) && !(pud_val(pud) & PUD_TABLE_BIT);
#else
	return 0;
#endif
}

/*
 * Select all bits except the pfn
 */
static inline pgprot_t pte_pgprot(pte_t pte)
{
	unsigned long pfn = pte_pfn(pte);

	return __pgprot(pte_val(pfn_pte(pfn, __pgprot(0))) ^ pte_val(pte));
}

static int find_num_contig(struct mm_struct *mm, unsigned long addr,
			   pte_t *ptep, size_t *pgsize)
{
	pgd_t *pgdp = pgd_offset(mm, addr);
	pud_t *pudp;
	pmd_t *pmdp;

	*pgsize = PAGE_SIZE;
	pudp = pud_offset(pgdp, addr);
	pmdp = pmd_offset(pudp, addr);
	if ((pte_t *)pmdp == ptep) {
		*pgsize = PMD_SIZE;
		return CONT_PMDS;
	}
	return CONT_PTES;
}

static inline int num_contig_ptes(unsigned long size, size_t *pgsize)
{
	int contig_ptes = 0;

	*pgsize = size;

	switch (size) {
#ifdef CONFIG_ARM64_4K_PAGES
	case PUD_SIZE:
#endif
	case PMD_SIZE:
		contig_ptes = 1;
		break;
	case CONT_PMD_SIZE:
		*pgsize = PMD_SIZE;
		contig_ptes = CONT_PMDS;
		break;
	case CONT_PTE_SIZE:
		*pgsize = PAGE_SIZE;
		contig_ptes = CONT_PTES;
		break;
	}

	return contig_ptes;
}

/*
 * Changing some bits of contiguous entries requires us to follow a
 * Break-Before-Make approach, breaking the whole contiguous set
 * before we can change any entries. See ARM DDI 0487A.k_iss10775,
 * "Misprogramming of the Contiguous bit", page D4-1762.
 *
 * This helper performs the break step.
 */
static pte_t get_clear_flush(struct mm_struct *mm,
			     unsigned long addr,
			     pte_t *ptep,
			     unsigned long pgsize,
			     unsigned long ncontig)
{
	pte_t orig_pte = huge_ptep_get(ptep);
	bool valid = pte_valid(orig_pte);
	unsigned long i, saddr = addr;

	for (i = 0; i < ncontig; i++, addr += pgsize, ptep++) {
		pte_t pte = ptep_get_and_clear(mm, addr, ptep);

		/*
		 * If HW_AFDBM is enabled, then the HW could turn on
		 * the dirty or accessed bit for any page in the set,
		 * so check them all.
		 */
		if (pte_dirty(pte))
			orig_pte = pte_mkdirty(orig_pte);

		if (pte_young(pte))
			orig_pte = pte_mkyoung(orig_pte);
	}

	if (valid) {
		struct vm_area_struct vma = TLB_FLUSH_VMA(mm, 0);
		flush_tlb_range(&vma, saddr, addr);
	}
	return orig_pte;
}

/*
 * Changing some bits of contiguous entries requires us to follow a
 * Break-Before-Make approach, breaking the whole contiguous set
 * before we can change any entries. See ARM DDI 0487A.k_iss10775,
 * "Misprogramming of the Contiguous bit", page D4-1762.
 *
 * This helper performs the break step for use cases where the
 * original pte is not needed.
 */
static void clear_flush(struct mm_struct *mm,
			     unsigned long addr,
			     pte_t *ptep,
			     unsigned long pgsize,
			     unsigned long ncontig)
{
	struct vm_area_struct vma = TLB_FLUSH_VMA(mm, 0);
	unsigned long i, saddr = addr;

	for (i = 0; i < ncontig; i++, addr += pgsize, ptep++)
		pte_clear(mm, addr, ptep);

	flush_tlb_range(&vma, saddr, addr);
}

void set_huge_pte_at(struct mm_struct *mm, unsigned long addr,
			    pte_t *ptep, pte_t pte)
{
	size_t pgsize;
	int i;
	int ncontig;
	unsigned long pfn, dpfn;
	pgprot_t hugeprot;

	/*
	 * Code needs to be expanded to handle huge swap and migration
	 * entries. Needed for HUGETLB and MEMORY_FAILURE.
	 */
	WARN_ON(!pte_present(pte));

	if (!pte_cont(pte)) {
		set_pte_at(mm, addr, ptep, pte);
		return;
	}

	ncontig = find_num_contig(mm, addr, ptep, &pgsize);
	pfn = pte_pfn(pte);
	dpfn = pgsize >> PAGE_SHIFT;
	hugeprot = pte_pgprot(pte);

	clear_flush(mm, addr, ptep, pgsize, ncontig);

	for (i = 0; i < ncontig; i++, ptep++, addr += pgsize, pfn += dpfn)
		set_pte_at(mm, addr, ptep, pfn_pte(pfn, hugeprot));
}

void set_huge_swap_pte_at(struct mm_struct *mm, unsigned long addr,
			  pte_t *ptep, pte_t pte, unsigned long sz)
{
	int i, ncontig;
	size_t pgsize;

	ncontig = num_contig_ptes(sz, &pgsize);

	for (i = 0; i < ncontig; i++, ptep++)
		set_pte(ptep, pte);
}

pte_t *huge_pte_alloc(struct mm_struct *mm,
		      unsigned long addr, unsigned long sz)
{
	pgd_t *pgdp;
	pud_t *pudp;
	pmd_t *pmdp;
	pte_t *ptep = NULL;

	pgdp = pgd_offset(mm, addr);
	pudp = pud_alloc(mm, pgdp, addr);
	if (!pudp)
		return NULL;

	if (sz == PUD_SIZE) {
		ptep = (pte_t *)pudp;
	} else if (sz == (CONT_PTE_SIZE)) {
		pmdp = pmd_alloc(mm, pudp, addr);
		if (!pmdp)
			return NULL;

		WARN_ON(addr & (sz - 1));
		/*
		 * Note that if this code were ever ported to the
		 * 32-bit arm platform then it will cause trouble in
		 * the case where CONFIG_HIGHPTE is set, since there
		 * will be no pte_unmap() to correspond with this
		 * pte_alloc_map().
		 */
		ptep = pte_alloc_map(mm, pmdp, addr);
	} else if (sz == PMD_SIZE) {
		if (IS_ENABLED(CONFIG_ARCH_WANT_HUGE_PMD_SHARE) &&
		    pud_none(READ_ONCE(*pudp)))
			ptep = huge_pmd_share(mm, addr, pudp);
		else
			ptep = (pte_t *)pmd_alloc(mm, pudp, addr);
	} else if (sz == (CONT_PMD_SIZE)) {
		pmdp = pmd_alloc(mm, pudp, addr);
		WARN_ON(addr & (sz - 1));
		return (pte_t *)pmdp;
	}

	return ptep;
}

pte_t *huge_pte_offset(struct mm_struct *mm,
		       unsigned long addr, unsigned long sz)
{
	pgd_t *pgdp;
	pud_t *pudp, pud;
	pmd_t *pmdp, pmd;

	pgdp = pgd_offset(mm, addr);
	if (!pgd_present(READ_ONCE(*pgdp)))
		return NULL;

	pudp = pud_offset(pgdp, addr);
	pud = READ_ONCE(*pudp);
	if (sz != PUD_SIZE && pud_none(pud))
		return NULL;
	/* hugepage or swap? */
	if (pud_huge(pud) || !pud_present(pud))
		return (pte_t *)pudp;
	/* table; check the next level */

	if (sz == CONT_PMD_SIZE)
		addr &= CONT_PMD_MASK;

	pmdp = pmd_offset(pudp, addr);
	pmd = READ_ONCE(*pmdp);
	if (!(sz == PMD_SIZE || sz == CONT_PMD_SIZE) &&
	    pmd_none(pmd))
		return NULL;
	if (pmd_huge(pmd) || !pmd_present(pmd))
		return (pte_t *)pmdp;

	if (sz == CONT_PTE_SIZE)
		return pte_offset_kernel(pmdp, (addr & CONT_PTE_MASK));

	return NULL;
}

pte_t arch_make_huge_pte(pte_t entry, struct vm_area_struct *vma,
			 struct page *page, int writable)
{
	size_t pagesize = huge_page_size(hstate_vma(vma));

	if (pagesize == CONT_PTE_SIZE) {
		entry = pte_mkcont(entry);
	} else if (pagesize == CONT_PMD_SIZE) {
		entry = pmd_pte(pmd_mkcont(pte_pmd(entry)));
	} else if (pagesize != PUD_SIZE && pagesize != PMD_SIZE) {
		pr_warn("%s: unrecognized huge page size 0x%lx\n",
			__func__, pagesize);
	}
	return entry;
}

void huge_pte_clear(struct mm_struct *mm, unsigned long addr,
		    pte_t *ptep, unsigned long sz)
{
	int i, ncontig;
	size_t pgsize;

	ncontig = num_contig_ptes(sz, &pgsize);

	for (i = 0; i < ncontig; i++, addr += pgsize, ptep++)
		pte_clear(mm, addr, ptep);
}

pte_t huge_ptep_get_and_clear(struct mm_struct *mm,
			      unsigned long addr, pte_t *ptep)
{
	int ncontig;
	size_t pgsize;
	pte_t orig_pte = huge_ptep_get(ptep);

	if (!pte_cont(orig_pte))
		return ptep_get_and_clear(mm, addr, ptep);

	ncontig = find_num_contig(mm, addr, ptep, &pgsize);

	return get_clear_flush(mm, addr, ptep, pgsize, ncontig);
}

/*
 * huge_ptep_set_access_flags will update access flags (dirty, accesssed)
 * and write permission.
 *
 * For a contiguous huge pte range we need to check whether or not write
 * permission has to change only on the first pte in the set. Then for
 * all the contiguous ptes we need to check whether or not there is a
 * discrepancy between dirty or young.
 */
static int __cont_access_flags_changed(pte_t *ptep, pte_t pte, int ncontig)
{
	int i;

	if (pte_write(pte) != pte_write(huge_ptep_get(ptep)))
		return 1;

	for (i = 0; i < ncontig; i++) {
		pte_t orig_pte = huge_ptep_get(ptep + i);

		if (pte_dirty(pte) != pte_dirty(orig_pte))
			return 1;

		if (pte_young(pte) != pte_young(orig_pte))
			return 1;
	}

	return 0;
}

int huge_ptep_set_access_flags(struct vm_area_struct *vma,
			       unsigned long addr, pte_t *ptep,
			       pte_t pte, int dirty)
{
	int ncontig, i;
	size_t pgsize = 0;
	unsigned long pfn = pte_pfn(pte), dpfn;
	pgprot_t hugeprot;
	pte_t orig_pte;

	if (!pte_cont(pte))
		return ptep_set_access_flags(vma, addr, ptep, pte, dirty);

	ncontig = find_num_contig(vma->vm_mm, addr, ptep, &pgsize);
	dpfn = pgsize >> PAGE_SHIFT;

	if (!__cont_access_flags_changed(ptep, pte, ncontig))
		return 0;

	orig_pte = get_clear_flush(vma->vm_mm, addr, ptep, pgsize, ncontig);

	/* Make sure we don't lose the dirty or young state */
	if (pte_dirty(orig_pte))
		pte = pte_mkdirty(pte);

	if (pte_young(orig_pte))
		pte = pte_mkyoung(pte);

	hugeprot = pte_pgprot(pte);
	for (i = 0; i < ncontig; i++, ptep++, addr += pgsize, pfn += dpfn)
		set_pte_at(vma->vm_mm, addr, ptep, pfn_pte(pfn, hugeprot));

	return 1;
}

void huge_ptep_set_wrprotect(struct mm_struct *mm,
			     unsigned long addr, pte_t *ptep)
{
	unsigned long pfn, dpfn;
	pgprot_t hugeprot;
	int ncontig, i;
	size_t pgsize;
	pte_t pte;

	if (!pte_cont(READ_ONCE(*ptep))) {
		ptep_set_wrprotect(mm, addr, ptep);
		return;
	}

	ncontig = find_num_contig(mm, addr, ptep, &pgsize);
	dpfn = pgsize >> PAGE_SHIFT;

	pte = get_clear_flush(mm, addr, ptep, pgsize, ncontig);
	pte = pte_wrprotect(pte);

	hugeprot = pte_pgprot(pte);
	pfn = pte_pfn(pte);

	for (i = 0; i < ncontig; i++, ptep++, addr += pgsize, pfn += dpfn)
		set_pte_at(mm, addr, ptep, pfn_pte(pfn, hugeprot));
}

void huge_ptep_clear_flush(struct vm_area_struct *vma,
			   unsigned long addr, pte_t *ptep)
{
	size_t pgsize;
	int ncontig;

	if (!pte_cont(READ_ONCE(*ptep))) {
		ptep_clear_flush(vma, addr, ptep);
		return;
	}

	ncontig = find_num_contig(vma->vm_mm, addr, ptep, &pgsize);
	clear_flush(vma->vm_mm, addr, ptep, pgsize, ncontig);
}

static void __init add_huge_page_size(unsigned long size)
{
	if (size_to_hstate(size))
		return;

	hugetlb_add_hstate(ilog2(size) - PAGE_SHIFT);
}

static int __init hugetlbpage_init(void)
{
#ifdef CONFIG_ARM64_4K_PAGES
	add_huge_page_size(PUD_SIZE);
#endif
	add_huge_page_size(CONT_PMD_SIZE);
	add_huge_page_size(PMD_SIZE);
	add_huge_page_size(CONT_PTE_SIZE);

	return 0;
}
arch_initcall(hugetlbpage_init);

static __init int setup_hugepagesz(char *opt)
{
	unsigned long ps = memparse(opt, &opt);

	switch (ps) {
#ifdef CONFIG_ARM64_4K_PAGES
	case PUD_SIZE:
#endif
	case CONT_PMD_SIZE:
	case PMD_SIZE:
	case CONT_PTE_SIZE:
		add_huge_page_size(ps);
		return 1;
	}

	hugetlb_bad_size();
	pr_err("hugepagesz: Unsupported page size %lu K\n", ps >> 10);
	return 0;
}
__setup("hugepagesz=", setup_hugepagesz);
Commit	Line	Data
1802d0be	1	// SPDX-License-Identifier: GPL-2.0-only
084bd298 SC	2	/*
	3	* arch/arm64/mm/hugetlbpage.c
	4	*
	5	* Copyright (C) 2013 Linaro Ltd.
	6	*
	7	* Based on arch/x86/mm/hugetlbpage.c.
084bd298 SC	8	*/
	9
	10	#include <linux/init.h>
	11	#include <linux/fs.h>
	12	#include <linux/mm.h>
	13	#include <linux/hugetlb.h>
	14	#include <linux/pagemap.h>
	15	#include <linux/err.h>
	16	#include <linux/sysctl.h>
	17	#include <asm/mman.h>
	18	#include <asm/tlb.h>
	19	#include <asm/tlbflush.h>
	20	#include <asm/pgalloc.h>
	21
5480280d AK	22	#ifdef CONFIG_ARCH_ENABLE_HUGEPAGE_MIGRATION
	23	bool arch_hugetlb_migration_supported(struct hstate *h)
	24	{
	25	size_t pagesize = huge_page_size(h);
	26
	27	switch (pagesize) {
	28	#ifdef CONFIG_ARM64_4K_PAGES
	29	case PUD_SIZE:
	30	#endif
	31	case PMD_SIZE:
	32	case CONT_PMD_SIZE:
	33	case CONT_PTE_SIZE:
	34	return true;
	35	}
	36	pr_warn("%s: unrecognized huge page size 0x%lx\n",
	37	__func__, pagesize);
	38	return false;
	39	}
	40	#endif
	41
084bd298 SC	42	int pmd_huge(pmd_t pmd)
084bd298 SC	43	{
fd28f5d4	44	return pmd_val(pmd) && !(pmd_val(pmd) & PMD_TABLE_BIT);
084bd298 SC	45	}
	46
	47	int pud_huge(pud_t pud)
	48	{
4797ec2d	49	#ifndef __PAGETABLE_PMD_FOLDED
fd28f5d4	50	return pud_val(pud) && !(pud_val(pud) & PUD_TABLE_BIT);
4797ec2d MS	51	#else
	52	return 0;
	53	#endif
084bd298 SC	54	}
084bd298 SC	55
b5b0be86 SC	56	/*
	57	* Select all bits except the pfn
	58	*/
	59	static inline pgprot_t pte_pgprot(pte_t pte)
	60	{
	61	unsigned long pfn = pte_pfn(pte);
	62
	63	return __pgprot(pte_val(pfn_pte(pfn, __pgprot(0))) ^ pte_val(pte));
	64	}
	65
66b3923a	66	static int find_num_contig(struct mm_struct *mm, unsigned long addr,
bb9dd3df	67	pte_t ptep, size_t pgsize)
66b3923a	68	{
20a004e7 WD	69	pgd_t *pgdp = pgd_offset(mm, addr);
	70	pud_t *pudp;
	71	pmd_t *pmdp;
66b3923a DW	72
66b3923a DW	73	*pgsize = PAGE_SIZE;
20a004e7 WD	74	pudp = pud_offset(pgdp, addr);
	75	pmdp = pmd_offset(pudp, addr);
	76	if ((pte_t *)pmdp == ptep) {
66b3923a DW	77	*pgsize = PMD_SIZE;
	78	return CONT_PMDS;
	79	}
	80	return CONT_PTES;
	81	}
	82
c3e4ed5c PA	83	static inline int num_contig_ptes(unsigned long size, size_t *pgsize)
	84	{
	85	int contig_ptes = 0;
	86
	87	*pgsize = size;
	88
	89	switch (size) {
	90	#ifdef CONFIG_ARM64_4K_PAGES
	91	case PUD_SIZE:
	92	#endif
	93	case PMD_SIZE:
	94	contig_ptes = 1;
	95	break;
	96	case CONT_PMD_SIZE:
	97	*pgsize = PMD_SIZE;
	98	contig_ptes = CONT_PMDS;
	99	break;
	100	case CONT_PTE_SIZE:
	101	*pgsize = PAGE_SIZE;
	102	contig_ptes = CONT_PTES;
	103	break;
	104	}
	105
	106	return contig_ptes;
	107	}
	108
d8bdcff2 SC	109	/*
	110	* Changing some bits of contiguous entries requires us to follow a
	111	* Break-Before-Make approach, breaking the whole contiguous set
	112	* before we can change any entries. See ARM DDI 0487A.k_iss10775,
	113	* "Misprogramming of the Contiguous bit", page D4-1762.
	114	*
	115	* This helper performs the break step.
	116	*/
	117	static pte_t get_clear_flush(struct mm_struct *mm,
	118	unsigned long addr,
	119	pte_t *ptep,
	120	unsigned long pgsize,
	121	unsigned long ncontig)
	122	{
d8bdcff2 SC	123	pte_t orig_pte = huge_ptep_get(ptep);
	124	bool valid = pte_valid(orig_pte);
	125	unsigned long i, saddr = addr;
	126
	127	for (i = 0; i < ncontig; i++, addr += pgsize, ptep++) {
	128	pte_t pte = ptep_get_and_clear(mm, addr, ptep);
	129
	130	/*
	131	* If HW_AFDBM is enabled, then the HW could turn on
469ed9d8 SC	132	* the dirty or accessed bit for any page in the set,
469ed9d8 SC	133	* so check them all.
d8bdcff2 SC	134	*/
	135	if (pte_dirty(pte))
	136	orig_pte = pte_mkdirty(orig_pte);
469ed9d8 SC	137
	138	if (pte_young(pte))
	139	orig_pte = pte_mkyoung(orig_pte);
d8bdcff2 SC	140	}
d8bdcff2 SC	141
8b11ec1b LT	142	if (valid) {
8b11ec1b LT	143	struct vm_area_struct vma = TLB_FLUSH_VMA(mm, 0);
d8bdcff2	144	flush_tlb_range(&vma, saddr, addr);
8b11ec1b	145	}
d8bdcff2 SC	146	return orig_pte;
	147	}
	148
	149	/*
	150	* Changing some bits of contiguous entries requires us to follow a
	151	* Break-Before-Make approach, breaking the whole contiguous set
	152	* before we can change any entries. See ARM DDI 0487A.k_iss10775,
	153	* "Misprogramming of the Contiguous bit", page D4-1762.
	154	*
	155	* This helper performs the break step for use cases where the
	156	* original pte is not needed.
	157	*/
	158	static void clear_flush(struct mm_struct *mm,
	159	unsigned long addr,
	160	pte_t *ptep,
	161	unsigned long pgsize,
	162	unsigned long ncontig)
	163	{
8b11ec1b	164	struct vm_area_struct vma = TLB_FLUSH_VMA(mm, 0);
d8bdcff2 SC	165	unsigned long i, saddr = addr;
	166
	167	for (i = 0; i < ncontig; i++, addr += pgsize, ptep++)
	168	pte_clear(mm, addr, ptep);
	169
	170	flush_tlb_range(&vma, saddr, addr);
	171	}
	172
66b3923a DW	173	void set_huge_pte_at(struct mm_struct *mm, unsigned long addr,
	174	pte_t *ptep, pte_t pte)
	175	{
	176	size_t pgsize;
	177	int i;
bb9dd3df	178	int ncontig;
29a7287d	179	unsigned long pfn, dpfn;
66b3923a DW	180	pgprot_t hugeprot;
66b3923a DW	181
d3ea7952 SC	182	/*
	183	* Code needs to be expanded to handle huge swap and migration
	184	* entries. Needed for HUGETLB and MEMORY_FAILURE.
	185	*/
	186	WARN_ON(!pte_present(pte));
	187
bb9dd3df	188	if (!pte_cont(pte)) {
66b3923a DW	189	set_pte_at(mm, addr, ptep, pte);
	190	return;
	191	}
	192
bb9dd3df	193	ncontig = find_num_contig(mm, addr, ptep, &pgsize);
66b3923a	194	pfn = pte_pfn(pte);
29a7287d	195	dpfn = pgsize >> PAGE_SHIFT;
b5b0be86	196	hugeprot = pte_pgprot(pte);
29a7287d	197
d8bdcff2 SC	198	clear_flush(mm, addr, ptep, pgsize, ncontig);
d8bdcff2 SC	199
20a004e7	200	for (i = 0; i < ncontig; i++, ptep++, addr += pgsize, pfn += dpfn)
66b3923a	201	set_pte_at(mm, addr, ptep, pfn_pte(pfn, hugeprot));
66b3923a DW	202	}
66b3923a DW	203
a8d623ee PA	204	void set_huge_swap_pte_at(struct mm_struct *mm, unsigned long addr,
	205	pte_t *ptep, pte_t pte, unsigned long sz)
	206	{
	207	int i, ncontig;
	208	size_t pgsize;
	209
	210	ncontig = num_contig_ptes(sz, &pgsize);
	211
	212	for (i = 0; i < ncontig; i++, ptep++)
	213	set_pte(ptep, pte);
	214	}
	215
66b3923a DW	216	pte_t huge_pte_alloc(struct mm_struct mm,
	217	unsigned long addr, unsigned long sz)
	218	{
20a004e7 WD	219	pgd_t *pgdp;
	220	pud_t *pudp;
	221	pmd_t *pmdp;
	222	pte_t *ptep = NULL;
	223
	224	pgdp = pgd_offset(mm, addr);
	225	pudp = pud_alloc(mm, pgdp, addr);
	226	if (!pudp)
66b3923a DW	227	return NULL;
	228
	229	if (sz == PUD_SIZE) {
20a004e7	230	ptep = (pte_t *)pudp;
441a6278	231	} else if (sz == (CONT_PTE_SIZE)) {
20a004e7	232	pmdp = pmd_alloc(mm, pudp, addr);
027d0c71 MR	233	if (!pmdp)
027d0c71 MR	234	return NULL;
66b3923a DW	235
	236	WARN_ON(addr & (sz - 1));
	237	/*
	238	* Note that if this code were ever ported to the
	239	* 32-bit arm platform then it will cause trouble in
	240	* the case where CONFIG_HIGHPTE is set, since there
	241	* will be no pte_unmap() to correspond with this
	242	* pte_alloc_map().
	243	*/
20a004e7	244	ptep = pte_alloc_map(mm, pmdp, addr);
66b3923a DW	245	} else if (sz == PMD_SIZE) {
66b3923a DW	246	if (IS_ENABLED(CONFIG_ARCH_WANT_HUGE_PMD_SHARE) &&
20a004e7 WD	247	pud_none(READ_ONCE(*pudp)))
20a004e7 WD	248	ptep = huge_pmd_share(mm, addr, pudp);
66b3923a	249	else
20a004e7	250	ptep = (pte_t *)pmd_alloc(mm, pudp, addr);
441a6278	251	} else if (sz == (CONT_PMD_SIZE)) {
20a004e7	252	pmdp = pmd_alloc(mm, pudp, addr);
66b3923a	253	WARN_ON(addr & (sz - 1));
20a004e7	254	return (pte_t *)pmdp;
66b3923a DW	255	}
66b3923a DW	256
20a004e7	257	return ptep;
66b3923a DW	258	}
66b3923a DW	259
7868a208 PA	260	pte_t huge_pte_offset(struct mm_struct mm,
7868a208 PA	261	unsigned long addr, unsigned long sz)
66b3923a	262	{
20a004e7 WD	263	pgd_t *pgdp;
	264	pud_t *pudp, pud;
	265	pmd_t *pmdp, pmd;
66b3923a	266
20a004e7 WD	267	pgdp = pgd_offset(mm, addr);
20a004e7 WD	268	if (!pgd_present(READ_ONCE(*pgdp)))
66b3923a	269	return NULL;
f02ab08a	270
20a004e7 WD	271	pudp = pud_offset(pgdp, addr);
	272	pud = READ_ONCE(*pudp);
	273	if (sz != PUD_SIZE && pud_none(pud))
66b3923a	274	return NULL;
30f3ac00	275	/* hugepage or swap? */
20a004e7 WD	276	if (pud_huge(pud) \|\| !pud_present(pud))
20a004e7 WD	277	return (pte_t *)pudp;
f02ab08a PA	278	/* table; check the next level */
f02ab08a PA	279
30f3ac00 PA	280	if (sz == CONT_PMD_SIZE)
	281	addr &= CONT_PMD_MASK;
	282
20a004e7 WD	283	pmdp = pmd_offset(pudp, addr);
20a004e7 WD	284	pmd = READ_ONCE(*pmdp);
30f3ac00	285	if (!(sz == PMD_SIZE \|\| sz == CONT_PMD_SIZE) &&
20a004e7	286	pmd_none(pmd))
66b3923a	287	return NULL;
20a004e7 WD	288	if (pmd_huge(pmd) \|\| !pmd_present(pmd))
20a004e7 WD	289	return (pte_t *)pmdp;
f02ab08a	290
20a004e7 WD	291	if (sz == CONT_PTE_SIZE)
20a004e7 WD	292	return pte_offset_kernel(pmdp, (addr & CONT_PTE_MASK));
30f3ac00	293
66b3923a DW	294	return NULL;
	295	}
	296
	297	pte_t arch_make_huge_pte(pte_t entry, struct vm_area_struct *vma,
	298	struct page *page, int writable)
	299	{
	300	size_t pagesize = huge_page_size(hstate_vma(vma));
	301
	302	if (pagesize == CONT_PTE_SIZE) {
	303	entry = pte_mkcont(entry);
	304	} else if (pagesize == CONT_PMD_SIZE) {
	305	entry = pmd_pte(pmd_mkcont(pte_pmd(entry)));
	306	} else if (pagesize != PUD_SIZE && pagesize != PMD_SIZE) {
	307	pr_warn("%s: unrecognized huge page size 0x%lx\n",
	308	__func__, pagesize);
	309	}
	310	return entry;
	311	}
	312
c3e4ed5c PA	313	void huge_pte_clear(struct mm_struct *mm, unsigned long addr,
	314	pte_t *ptep, unsigned long sz)
	315	{
	316	int i, ncontig;
	317	size_t pgsize;
	318
	319	ncontig = num_contig_ptes(sz, &pgsize);
	320
	321	for (i = 0; i < ncontig; i++, addr += pgsize, ptep++)
	322	pte_clear(mm, addr, ptep);
	323	}
	324
66b3923a DW	325	pte_t huge_ptep_get_and_clear(struct mm_struct *mm,
	326	unsigned long addr, pte_t *ptep)
	327	{
d8bdcff2	328	int ncontig;
29a7287d SC	329	size_t pgsize;
	330	pte_t orig_pte = huge_ptep_get(ptep);
	331
	332	if (!pte_cont(orig_pte))
66b3923a	333	return ptep_get_and_clear(mm, addr, ptep);
29a7287d SC	334
29a7287d SC	335	ncontig = find_num_contig(mm, addr, ptep, &pgsize);
29a7287d	336
d8bdcff2	337	return get_clear_flush(mm, addr, ptep, pgsize, ncontig);
66b3923a DW	338	}
66b3923a DW	339
031e6e6b SC	340	/*
	341	* huge_ptep_set_access_flags will update access flags (dirty, accesssed)
	342	* and write permission.
	343	*
	344	* For a contiguous huge pte range we need to check whether or not write
	345	* permission has to change only on the first pte in the set. Then for
	346	* all the contiguous ptes we need to check whether or not there is a
	347	* discrepancy between dirty or young.
	348	*/
	349	static int __cont_access_flags_changed(pte_t *ptep, pte_t pte, int ncontig)
	350	{
	351	int i;
	352
	353	if (pte_write(pte) != pte_write(huge_ptep_get(ptep)))
	354	return 1;
	355
	356	for (i = 0; i < ncontig; i++) {
	357	pte_t orig_pte = huge_ptep_get(ptep + i);
	358
	359	if (pte_dirty(pte) != pte_dirty(orig_pte))
	360	return 1;
	361
	362	if (pte_young(pte) != pte_young(orig_pte))
	363	return 1;
	364	}
	365
	366	return 0;
	367	}
	368
66b3923a DW	369	int huge_ptep_set_access_flags(struct vm_area_struct *vma,
	370	unsigned long addr, pte_t *ptep,
	371	pte_t pte, int dirty)
	372	{
031e6e6b	373	int ncontig, i;
29a7287d SC	374	size_t pgsize = 0;
	375	unsigned long pfn = pte_pfn(pte), dpfn;
	376	pgprot_t hugeprot;
d8bdcff2	377	pte_t orig_pte;
29a7287d SC	378
29a7287d SC	379	if (!pte_cont(pte))
66b3923a	380	return ptep_set_access_flags(vma, addr, ptep, pte, dirty);
29a7287d SC	381
	382	ncontig = find_num_contig(vma->vm_mm, addr, ptep, &pgsize);
	383	dpfn = pgsize >> PAGE_SHIFT;
29a7287d	384
031e6e6b SC	385	if (!__cont_access_flags_changed(ptep, pte, ncontig))
	386	return 0;
	387
d8bdcff2	388	orig_pte = get_clear_flush(vma->vm_mm, addr, ptep, pgsize, ncontig);
d8bdcff2	389
469ed9d8	390	/* Make sure we don't lose the dirty or young state */
d8bdcff2 SC	391	if (pte_dirty(orig_pte))
	392	pte = pte_mkdirty(pte);
	393
469ed9d8 SC	394	if (pte_young(orig_pte))
	395	pte = pte_mkyoung(pte);
	396
d8bdcff2 SC	397	hugeprot = pte_pgprot(pte);
	398	for (i = 0; i < ncontig; i++, ptep++, addr += pgsize, pfn += dpfn)
	399	set_pte_at(vma->vm_mm, addr, ptep, pfn_pte(pfn, hugeprot));
29a7287d	400
031e6e6b	401	return 1;
66b3923a DW	402	}
	403
	404	void huge_ptep_set_wrprotect(struct mm_struct *mm,
	405	unsigned long addr, pte_t *ptep)
	406	{
d8bdcff2 SC	407	unsigned long pfn, dpfn;
d8bdcff2 SC	408	pgprot_t hugeprot;
29a7287d SC	409	int ncontig, i;
29a7287d SC	410	size_t pgsize;
d8bdcff2	411	pte_t pte;
66b3923a	412
20a004e7	413	if (!pte_cont(READ_ONCE(*ptep))) {
66b3923a	414	ptep_set_wrprotect(mm, addr, ptep);
29a7287d	415	return;
66b3923a	416	}
29a7287d SC	417
29a7287d SC	418	ncontig = find_num_contig(mm, addr, ptep, &pgsize);
d8bdcff2 SC	419	dpfn = pgsize >> PAGE_SHIFT;
	420
	421	pte = get_clear_flush(mm, addr, ptep, pgsize, ncontig);
	422	pte = pte_wrprotect(pte);
	423
	424	hugeprot = pte_pgprot(pte);
	425	pfn = pte_pfn(pte);
	426
	427	for (i = 0; i < ncontig; i++, ptep++, addr += pgsize, pfn += dpfn)
	428	set_pte_at(mm, addr, ptep, pfn_pte(pfn, hugeprot));
66b3923a DW	429	}
	430
	431	void huge_ptep_clear_flush(struct vm_area_struct *vma,
	432	unsigned long addr, pte_t *ptep)
	433	{
29a7287d	434	size_t pgsize;
d8bdcff2	435	int ncontig;
29a7287d	436
20a004e7	437	if (!pte_cont(READ_ONCE(*ptep))) {
66b3923a	438	ptep_clear_flush(vma, addr, ptep);
29a7287d	439	return;
66b3923a	440	}
29a7287d SC	441
29a7287d SC	442	ncontig = find_num_contig(vma->vm_mm, addr, ptep, &pgsize);
d8bdcff2	443	clear_flush(vma->vm_mm, addr, ptep, pgsize, ncontig);
66b3923a DW	444	}
66b3923a DW	445
a21b0b78 AP	446	static void __init add_huge_page_size(unsigned long size)
	447	{
	448	if (size_to_hstate(size))
	449	return;
	450
	451	hugetlb_add_hstate(ilog2(size) - PAGE_SHIFT);
	452	}
	453
	454	static int __init hugetlbpage_init(void)
	455	{
	456	#ifdef CONFIG_ARM64_4K_PAGES
	457	add_huge_page_size(PUD_SIZE);
	458	#endif
441a6278	459	add_huge_page_size(CONT_PMD_SIZE);
a21b0b78	460	add_huge_page_size(PMD_SIZE);
441a6278	461	add_huge_page_size(CONT_PTE_SIZE);
a21b0b78 AP	462
	463	return 0;
	464	}
	465	arch_initcall(hugetlbpage_init);
	466
084bd298 SC	467	static __init int setup_hugepagesz(char *opt)
	468	{
	469	unsigned long ps = memparse(opt, &opt);
66b3923a	470
828f193d SC	471	switch (ps) {
	472	#ifdef CONFIG_ARM64_4K_PAGES
	473	case PUD_SIZE:
	474	#endif
441a6278	475	case CONT_PMD_SIZE:
828f193d	476	case PMD_SIZE:
441a6278	477	case CONT_PTE_SIZE:
a21b0b78	478	add_huge_page_size(ps);
828f193d	479	return 1;
084bd298	480	}
828f193d SC	481
	482	hugetlb_bad_size();
	483	pr_err("hugepagesz: Unsupported page size %lu K\n", ps >> 10);
	484	return 0;
084bd298 SC	485	}
084bd298 SC	486	__setup("hugepagesz=", setup_hugepagesz);