[linux-block.git] / arch / riscv / mm / hugetlbpage.c

// SPDX-License-Identifier: GPL-2.0
#include <linux/hugetlb.h>
#include <linux/err.h>

#ifdef CONFIG_RISCV_ISA_SVNAPOT
pte_t huge_ptep_get(pte_t *ptep)
{
	unsigned long pte_num;
	int i;
	pte_t orig_pte = ptep_get(ptep);

	if (!pte_present(orig_pte) || !pte_napot(orig_pte))
		return orig_pte;

	pte_num = napot_pte_num(napot_cont_order(orig_pte));

	for (i = 0; i < pte_num; i++, ptep++) {
		pte_t pte = ptep_get(ptep);

		if (pte_dirty(pte))
			orig_pte = pte_mkdirty(orig_pte);

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

	return orig_pte;
}

pte_t *huge_pte_alloc(struct mm_struct *mm,
		      struct vm_area_struct *vma,
		      unsigned long addr,
		      unsigned long sz)
{
	unsigned long order;
	pte_t *pte = NULL;
	pgd_t *pgd;
	p4d_t *p4d;
	pud_t *pud;
	pmd_t *pmd;

	pgd = pgd_offset(mm, addr);
	p4d = p4d_alloc(mm, pgd, addr);
	if (!p4d)
		return NULL;

	pud = pud_alloc(mm, p4d, addr);
	if (!pud)
		return NULL;

	if (sz == PUD_SIZE) {
		pte = (pte_t *)pud;
		goto out;
	}

	if (sz == PMD_SIZE) {
		if (want_pmd_share(vma, addr) && pud_none(pudp_get(pud)))
			pte = huge_pmd_share(mm, vma, addr, pud);
		else
			pte = (pte_t *)pmd_alloc(mm, pud, addr);
		goto out;
	}

	pmd = pmd_alloc(mm, pud, addr);
	if (!pmd)
		return NULL;

	for_each_napot_order(order) {
		if (napot_cont_size(order) == sz) {
			pte = pte_alloc_huge(mm, pmd, addr & napot_cont_mask(order));
			break;
		}
	}

out:
	if (pte) {
		pte_t pteval = ptep_get_lockless(pte);

		WARN_ON_ONCE(pte_present(pteval) && !pte_huge(pteval));
	}
	return pte;
}

pte_t *huge_pte_offset(struct mm_struct *mm,
		       unsigned long addr,
		       unsigned long sz)
{
	unsigned long order;
	pte_t *pte = NULL;
	pgd_t *pgd;
	p4d_t *p4d;
	pud_t *pud;
	pmd_t *pmd;

	pgd = pgd_offset(mm, addr);
	if (!pgd_present(pgdp_get(pgd)))
		return NULL;

	p4d = p4d_offset(pgd, addr);
	if (!p4d_present(p4dp_get(p4d)))
		return NULL;

	pud = pud_offset(p4d, addr);
	if (sz == PUD_SIZE)
		/* must be pud huge, non-present or none */
		return (pte_t *)pud;

	if (!pud_present(pudp_get(pud)))
		return NULL;

	pmd = pmd_offset(pud, addr);
	if (sz == PMD_SIZE)
		/* must be pmd huge, non-present or none */
		return (pte_t *)pmd;

	if (!pmd_present(pmdp_get(pmd)))
		return NULL;

	for_each_napot_order(order) {
		if (napot_cont_size(order) == sz) {
			pte = pte_offset_huge(pmd, addr & napot_cont_mask(order));
			break;
		}
	}
	return pte;
}

static pte_t get_clear_contig(struct mm_struct *mm,
			      unsigned long addr,
			      pte_t *ptep,
			      unsigned long pte_num)
{
	pte_t orig_pte = ptep_get(ptep);
	unsigned long i;

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

		if (pte_dirty(pte))
			orig_pte = pte_mkdirty(orig_pte);

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

	return orig_pte;
}

static pte_t get_clear_contig_flush(struct mm_struct *mm,
				    unsigned long addr,
				    pte_t *ptep,
				    unsigned long pte_num)
{
	pte_t orig_pte = get_clear_contig(mm, addr, ptep, pte_num);
	struct vm_area_struct vma = TLB_FLUSH_VMA(mm, 0);
	bool valid = !pte_none(orig_pte);

	if (valid)
		flush_tlb_range(&vma, addr, addr + (PAGE_SIZE * pte_num));

	return orig_pte;
}

pte_t arch_make_huge_pte(pte_t entry, unsigned int shift, vm_flags_t flags)
{
	unsigned long order;

	for_each_napot_order(order) {
		if (shift == napot_cont_shift(order)) {
			entry = pte_mknapot(entry, order);
			break;
		}
	}
	if (order == NAPOT_ORDER_MAX)
		entry = pte_mkhuge(entry);

	return entry;
}

void set_huge_pte_at(struct mm_struct *mm,
		     unsigned long addr,
		     pte_t *ptep,
		     pte_t pte,
		     unsigned long sz)
{
	unsigned long hugepage_shift;
	int i, pte_num;

	if (sz >= PGDIR_SIZE)
		hugepage_shift = PGDIR_SHIFT;
	else if (sz >= P4D_SIZE)
		hugepage_shift = P4D_SHIFT;
	else if (sz >= PUD_SIZE)
		hugepage_shift = PUD_SHIFT;
	else if (sz >= PMD_SIZE)
		hugepage_shift = PMD_SHIFT;
	else
		hugepage_shift = PAGE_SHIFT;

	pte_num = sz >> hugepage_shift;
	for (i = 0; i < pte_num; i++, ptep++, addr += (1 << hugepage_shift))
		set_pte_at(mm, addr, ptep, pte);
}

int huge_ptep_set_access_flags(struct vm_area_struct *vma,
			       unsigned long addr,
			       pte_t *ptep,
			       pte_t pte,
			       int dirty)
{
	struct mm_struct *mm = vma->vm_mm;
	unsigned long order;
	pte_t orig_pte;
	int i, pte_num;

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

	order = napot_cont_order(pte);
	pte_num = napot_pte_num(order);
	ptep = huge_pte_offset(mm, addr, napot_cont_size(order));
	orig_pte = get_clear_contig_flush(mm, addr, ptep, pte_num);

	if (pte_dirty(orig_pte))
		pte = pte_mkdirty(pte);

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

	for (i = 0; i < pte_num; i++, addr += PAGE_SIZE, ptep++)
		set_pte_at(mm, addr, ptep, pte);

	return true;
}

pte_t huge_ptep_get_and_clear(struct mm_struct *mm,
			      unsigned long addr,
			      pte_t *ptep)
{
	pte_t orig_pte = ptep_get(ptep);
	int pte_num;

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

	pte_num = napot_pte_num(napot_cont_order(orig_pte));

	return get_clear_contig(mm, addr, ptep, pte_num);
}

void huge_ptep_set_wrprotect(struct mm_struct *mm,
			     unsigned long addr,
			     pte_t *ptep)
{
	pte_t pte = ptep_get(ptep);
	unsigned long order;
	pte_t orig_pte;
	int i, pte_num;

	if (!pte_napot(pte)) {
		ptep_set_wrprotect(mm, addr, ptep);
		return;
	}

	order = napot_cont_order(pte);
	pte_num = napot_pte_num(order);
	ptep = huge_pte_offset(mm, addr, napot_cont_size(order));
	orig_pte = get_clear_contig_flush(mm, addr, ptep, pte_num);

	orig_pte = pte_wrprotect(orig_pte);

	for (i = 0; i < pte_num; i++, addr += PAGE_SIZE, ptep++)
		set_pte_at(mm, addr, ptep, orig_pte);
}

pte_t huge_ptep_clear_flush(struct vm_area_struct *vma,
			    unsigned long addr,
			    pte_t *ptep)
{
	pte_t pte = ptep_get(ptep);
	int pte_num;

	if (!pte_napot(pte))
		return ptep_clear_flush(vma, addr, ptep);

	pte_num = napot_pte_num(napot_cont_order(pte));

	return get_clear_contig_flush(vma->vm_mm, addr, ptep, pte_num);
}

void huge_pte_clear(struct mm_struct *mm,
		    unsigned long addr,
		    pte_t *ptep,
		    unsigned long sz)
{
	pte_t pte = ptep_get(ptep);
	int i, pte_num;

	if (!pte_napot(pte)) {
		pte_clear(mm, addr, ptep);
		return;
	}

	pte_num = napot_pte_num(napot_cont_order(pte));
	for (i = 0; i < pte_num; i++, addr += PAGE_SIZE, ptep++)
		pte_clear(mm, addr, ptep);
}

static __init bool is_napot_size(unsigned long size)
{
	unsigned long order;

	if (!has_svnapot())
		return false;

	for_each_napot_order(order) {
		if (size == napot_cont_size(order))
			return true;
	}
	return false;
}

static __init int napot_hugetlbpages_init(void)
{
	if (has_svnapot()) {
		unsigned long order;

		for_each_napot_order(order)
			hugetlb_add_hstate(order);
	}
	return 0;
}
arch_initcall(napot_hugetlbpages_init);

#else

static __init bool is_napot_size(unsigned long size)
{
	return false;
}

#endif /*CONFIG_RISCV_ISA_SVNAPOT*/

int pud_huge(pud_t pud)
{
	return pud_leaf(pud);
}

int pmd_huge(pmd_t pmd)
{
	return pmd_leaf(pmd);
}

bool __init arch_hugetlb_valid_size(unsigned long size)
{
	if (size == HPAGE_SIZE)
		return true;
	else if (IS_ENABLED(CONFIG_64BIT) && size == PUD_SIZE)
		return true;
	else if (is_napot_size(size))
		return true;
	else
		return false;
}

#ifdef CONFIG_CONTIG_ALLOC
static __init int gigantic_pages_init(void)
{
	/* With CONTIG_ALLOC, we can allocate gigantic pages at runtime */
	if (IS_ENABLED(CONFIG_64BIT))
		hugetlb_add_hstate(PUD_SHIFT - PAGE_SHIFT);
	return 0;
}
arch_initcall(gigantic_pages_init);
#endif
Commit	Line	Data
9e953cda AG	1	// SPDX-License-Identifier: GPL-2.0
	2	#include <linux/hugetlb.h>
	3	#include <linux/err.h>
	4
82a1a1f3	5	#ifdef CONFIG_RISCV_ISA_SVNAPOT
6966d798 AG	6	pte_t huge_ptep_get(pte_t *ptep)
	7	{
	8	unsigned long pte_num;
	9	int i;
	10	pte_t orig_pte = ptep_get(ptep);
	11
	12	if (!pte_present(orig_pte) \|\| !pte_napot(orig_pte))
	13	return orig_pte;
	14
	15	pte_num = napot_pte_num(napot_cont_order(orig_pte));
	16
	17	for (i = 0; i < pte_num; i++, ptep++) {
	18	pte_t pte = ptep_get(ptep);
	19
	20	if (pte_dirty(pte))
	21	orig_pte = pte_mkdirty(orig_pte);
	22
	23	if (pte_young(pte))
	24	orig_pte = pte_mkyoung(orig_pte);
	25	}
	26
	27	return orig_pte;
	28	}
	29
82a1a1f3 QP	30	pte_t huge_pte_alloc(struct mm_struct mm,
	31	struct vm_area_struct *vma,
	32	unsigned long addr,
	33	unsigned long sz)
	34	{
	35	unsigned long order;
	36	pte_t *pte = NULL;
	37	pgd_t *pgd;
	38	p4d_t *p4d;
	39	pud_t *pud;
	40	pmd_t *pmd;
	41
	42	pgd = pgd_offset(mm, addr);
	43	p4d = p4d_alloc(mm, pgd, addr);
	44	if (!p4d)
	45	return NULL;
	46
	47	pud = pud_alloc(mm, p4d, addr);
	48	if (!pud)
	49	return NULL;
	50
	51	if (sz == PUD_SIZE) {
	52	pte = (pte_t *)pud;
	53	goto out;
	54	}
	55
	56	if (sz == PMD_SIZE) {
edf95564	57	if (want_pmd_share(vma, addr) && pud_none(pudp_get(pud)))
82a1a1f3 QP	58	pte = huge_pmd_share(mm, vma, addr, pud);
	59	else
	60	pte = (pte_t *)pmd_alloc(mm, pud, addr);
	61	goto out;
	62	}
	63
	64	pmd = pmd_alloc(mm, pud, addr);
	65	if (!pmd)
	66	return NULL;
	67
	68	for_each_napot_order(order) {
	69	if (napot_cont_size(order) == sz) {
893f667f	70	pte = pte_alloc_huge(mm, pmd, addr & napot_cont_mask(order));
82a1a1f3 QP	71	break;
	72	}
	73	}
	74
	75	out:
62ba41d2 JH	76	if (pte) {
	77	pte_t pteval = ptep_get_lockless(pte);
	78
	79	WARN_ON_ONCE(pte_present(pteval) && !pte_huge(pteval));
	80	}
82a1a1f3 QP	81	return pte;
	82	}
	83
	84	pte_t huge_pte_offset(struct mm_struct mm,
	85	unsigned long addr,
	86	unsigned long sz)
	87	{
	88	unsigned long order;
	89	pte_t *pte = NULL;
	90	pgd_t *pgd;
	91	p4d_t *p4d;
	92	pud_t *pud;
	93	pmd_t *pmd;
	94
	95	pgd = pgd_offset(mm, addr);
edf95564	96	if (!pgd_present(pgdp_get(pgd)))
82a1a1f3 QP	97	return NULL;
	98
	99	p4d = p4d_offset(pgd, addr);
edf95564	100	if (!p4d_present(p4dp_get(p4d)))
82a1a1f3 QP	101	return NULL;
	102
	103	pud = pud_offset(p4d, addr);
	104	if (sz == PUD_SIZE)
	105	/* must be pud huge, non-present or none */
	106	return (pte_t *)pud;
	107
edf95564	108	if (!pud_present(pudp_get(pud)))
82a1a1f3 QP	109	return NULL;
	110
	111	pmd = pmd_offset(pud, addr);
	112	if (sz == PMD_SIZE)
	113	/* must be pmd huge, non-present or none */
	114	return (pte_t *)pmd;
	115
edf95564	116	if (!pmd_present(pmdp_get(pmd)))
82a1a1f3 QP	117	return NULL;
	118
	119	for_each_napot_order(order) {
	120	if (napot_cont_size(order) == sz) {
893f667f	121	pte = pte_offset_huge(pmd, addr & napot_cont_mask(order));
82a1a1f3 QP	122	break;
	123	}
	124	}
	125	return pte;
	126	}
	127
	128	static pte_t get_clear_contig(struct mm_struct *mm,
	129	unsigned long addr,
	130	pte_t *ptep,
	131	unsigned long pte_num)
	132	{
	133	pte_t orig_pte = ptep_get(ptep);
	134	unsigned long i;
	135
	136	for (i = 0; i < pte_num; i++, addr += PAGE_SIZE, ptep++) {
	137	pte_t pte = ptep_get_and_clear(mm, addr, ptep);
	138
	139	if (pte_dirty(pte))
	140	orig_pte = pte_mkdirty(orig_pte);
	141
	142	if (pte_young(pte))
	143	orig_pte = pte_mkyoung(orig_pte);
	144	}
	145
	146	return orig_pte;
	147	}
	148
	149	static pte_t get_clear_contig_flush(struct mm_struct *mm,
	150	unsigned long addr,
	151	pte_t *ptep,
	152	unsigned long pte_num)
	153	{
	154	pte_t orig_pte = get_clear_contig(mm, addr, ptep, pte_num);
	155	struct vm_area_struct vma = TLB_FLUSH_VMA(mm, 0);
	156	bool valid = !pte_none(orig_pte);
	157
	158	if (valid)
	159	flush_tlb_range(&vma, addr, addr + (PAGE_SIZE * pte_num));
	160
	161	return orig_pte;
	162	}
	163
	164	pte_t arch_make_huge_pte(pte_t entry, unsigned int shift, vm_flags_t flags)
	165	{
	166	unsigned long order;
	167
	168	for_each_napot_order(order) {
	169	if (shift == napot_cont_shift(order)) {
	170	entry = pte_mknapot(entry, order);
	171	break;
	172	}
	173	}
	174	if (order == NAPOT_ORDER_MAX)
	175	entry = pte_mkhuge(entry);
	176
	177	return entry;
	178	}
	179
	180	void set_huge_pte_at(struct mm_struct *mm,
	181	unsigned long addr,
	182	pte_t *ptep,
935d4f0c RR	183	pte_t pte,
935d4f0c RR	184	unsigned long sz)
82a1a1f3	185	{
1de195dd	186	unsigned long hugepage_shift;
82a1a1f3 QP	187	int i, pte_num;
82a1a1f3 QP	188
1de195dd AG	189	if (sz >= PGDIR_SIZE)
	190	hugepage_shift = PGDIR_SHIFT;
	191	else if (sz >= P4D_SIZE)
	192	hugepage_shift = P4D_SHIFT;
	193	else if (sz >= PUD_SIZE)
	194	hugepage_shift = PUD_SHIFT;
	195	else if (sz >= PMD_SIZE)
	196	hugepage_shift = PMD_SHIFT;
	197	else
	198	hugepage_shift = PAGE_SHIFT;
	199
	200	pte_num = sz >> hugepage_shift;
	201	for (i = 0; i < pte_num; i++, ptep++, addr += (1 << hugepage_shift))
82a1a1f3 QP	202	set_pte_at(mm, addr, ptep, pte);
	203	}
	204
	205	int huge_ptep_set_access_flags(struct vm_area_struct *vma,
	206	unsigned long addr,
	207	pte_t *ptep,
	208	pte_t pte,
	209	int dirty)
	210	{
	211	struct mm_struct *mm = vma->vm_mm;
	212	unsigned long order;
	213	pte_t orig_pte;
	214	int i, pte_num;
	215
	216	if (!pte_napot(pte))
	217	return ptep_set_access_flags(vma, addr, ptep, pte, dirty);
	218
	219	order = napot_cont_order(pte);
	220	pte_num = napot_pte_num(order);
	221	ptep = huge_pte_offset(mm, addr, napot_cont_size(order));
	222	orig_pte = get_clear_contig_flush(mm, addr, ptep, pte_num);
	223
	224	if (pte_dirty(orig_pte))
	225	pte = pte_mkdirty(pte);
	226
	227	if (pte_young(orig_pte))
	228	pte = pte_mkyoung(pte);
	229
	230	for (i = 0; i < pte_num; i++, addr += PAGE_SIZE, ptep++)
	231	set_pte_at(mm, addr, ptep, pte);
	232
	233	return true;
	234	}
	235
	236	pte_t huge_ptep_get_and_clear(struct mm_struct *mm,
	237	unsigned long addr,
	238	pte_t *ptep)
	239	{
	240	pte_t orig_pte = ptep_get(ptep);
	241	int pte_num;
	242
	243	if (!pte_napot(orig_pte))
	244	return ptep_get_and_clear(mm, addr, ptep);
	245
	246	pte_num = napot_pte_num(napot_cont_order(orig_pte));
	247
	248	return get_clear_contig(mm, addr, ptep, pte_num);
	249	}
	250
	251	void huge_ptep_set_wrprotect(struct mm_struct *mm,
	252	unsigned long addr,
	253	pte_t *ptep)
	254	{
	255	pte_t pte = ptep_get(ptep);
	256	unsigned long order;
835e5ac3	257	pte_t orig_pte;
82a1a1f3 QP	258	int i, pte_num;
	259
	260	if (!pte_napot(pte)) {
	261	ptep_set_wrprotect(mm, addr, ptep);
	262	return;
	263	}
	264
	265	order = napot_cont_order(pte);
	266	pte_num = napot_pte_num(order);
	267	ptep = huge_pte_offset(mm, addr, napot_cont_size(order));
835e5ac3 AG	268	orig_pte = get_clear_contig_flush(mm, addr, ptep, pte_num);
	269
	270	orig_pte = pte_wrprotect(orig_pte);
82a1a1f3 QP	271
82a1a1f3 QP	272	for (i = 0; i < pte_num; i++, addr += PAGE_SIZE, ptep++)
835e5ac3	273	set_pte_at(mm, addr, ptep, orig_pte);
82a1a1f3 QP	274	}
	275
	276	pte_t huge_ptep_clear_flush(struct vm_area_struct *vma,
	277	unsigned long addr,
	278	pte_t *ptep)
	279	{
	280	pte_t pte = ptep_get(ptep);
	281	int pte_num;
	282
	283	if (!pte_napot(pte))
	284	return ptep_clear_flush(vma, addr, ptep);
	285
	286	pte_num = napot_pte_num(napot_cont_order(pte));
	287
	288	return get_clear_contig_flush(vma->vm_mm, addr, ptep, pte_num);
	289	}
	290
	291	void huge_pte_clear(struct mm_struct *mm,
	292	unsigned long addr,
	293	pte_t *ptep,
	294	unsigned long sz)
	295	{
edf95564	296	pte_t pte = ptep_get(ptep);
82a1a1f3 QP	297	int i, pte_num;
	298
	299	if (!pte_napot(pte)) {
	300	pte_clear(mm, addr, ptep);
	301	return;
	302	}
	303
	304	pte_num = napot_pte_num(napot_cont_order(pte));
	305	for (i = 0; i < pte_num; i++, addr += PAGE_SIZE, ptep++)
	306	pte_clear(mm, addr, ptep);
	307	}
	308
	309	static __init bool is_napot_size(unsigned long size)
	310	{
	311	unsigned long order;
	312
	313	if (!has_svnapot())
	314	return false;
	315
	316	for_each_napot_order(order) {
	317	if (size == napot_cont_size(order))
	318	return true;
	319	}
	320	return false;
	321	}
	322
	323	static __init int napot_hugetlbpages_init(void)
	324	{
	325	if (has_svnapot()) {
	326	unsigned long order;
	327
	328	for_each_napot_order(order)
	329	hugetlb_add_hstate(order);
	330	}
	331	return 0;
	332	}
	333	arch_initcall(napot_hugetlbpages_init);
	334
	335	#else
	336
	337	static __init bool is_napot_size(unsigned long size)
	338	{
	339	return false;
	340	}
	341
	342	#endif /CONFIG_RISCV_ISA_SVNAPOT/
	343
9e953cda AG	344	int pud_huge(pud_t pud)
9e953cda AG	345	{
3133287b	346	return pud_leaf(pud);
9e953cda AG	347	}
	348
	349	int pmd_huge(pmd_t pmd)
	350	{
3133287b	351	return pmd_leaf(pmd);
9e953cda AG	352	}
9e953cda AG	353
ae94da89 MK	354	bool __init arch_hugetlb_valid_size(unsigned long size)
	355	{
	356	if (size == HPAGE_SIZE)
	357	return true;
	358	else if (IS_ENABLED(CONFIG_64BIT) && size == PUD_SIZE)
	359	return true;
82a1a1f3 QP	360	else if (is_napot_size(size))
82a1a1f3 QP	361	return true;
ae94da89 MK	362	else
	363	return false;
	364	}
	365
9e953cda AG	366	#ifdef CONFIG_CONTIG_ALLOC
	367	static __init int gigantic_pages_init(void)
	368	{
	369	/* With CONTIG_ALLOC, we can allocate gigantic pages at runtime */
38237830	370	if (IS_ENABLED(CONFIG_64BIT))
9e953cda AG	371	hugetlb_add_hstate(PUD_SHIFT - PAGE_SHIFT);
	372	return 0;
	373	}
	374	arch_initcall(gigantic_pages_init);
	375	#endif