[net-next-2.6.git] / arch / ia64 / mm / hugetlbpage.c

/*
 * IA-64 Huge TLB Page Support for Kernel.
 *
 * Copyright (C) 2002-2004 Rohit Seth <rohit.seth@intel.com>
 * Copyright (C) 2003-2004 Ken Chen <kenneth.w.chen@intel.com>
 *
 * Sep, 2003: add numa support
 * Feb, 2004: dynamic hugetlb page size via boot parameter
 */

#include <linux/init.h>
#include <linux/fs.h>
#include <linux/mm.h>
#include <linux/hugetlb.h>
#include <linux/pagemap.h>
#include <linux/module.h>
#include <linux/sysctl.h>
#include <linux/log2.h>
#include <asm/mman.h>
#include <asm/pgalloc.h>
#include <asm/tlb.h>
#include <asm/tlbflush.h>

unsigned int hpage_shift = HPAGE_SHIFT_DEFAULT;
EXPORT_SYMBOL(hpage_shift);

pte_t *
huge_pte_alloc(struct mm_struct *mm, unsigned long addr, unsigned long sz)
{
	unsigned long taddr = htlbpage_to_page(addr);
	pgd_t *pgd;
	pud_t *pud;
	pmd_t *pmd;
	pte_t *pte = NULL;

	pgd = pgd_offset(mm, taddr);
	pud = pud_alloc(mm, pgd, taddr);
	if (pud) {
		pmd = pmd_alloc(mm, pud, taddr);
		if (pmd)
			pte = pte_alloc_map(mm, pmd, taddr);
	}
	return pte;
}

pte_t *
huge_pte_offset (struct mm_struct *mm, unsigned long addr)
{
	unsigned long taddr = htlbpage_to_page(addr);
	pgd_t *pgd;
	pud_t *pud;
	pmd_t *pmd;
	pte_t *pte = NULL;

	pgd = pgd_offset(mm, taddr);
	if (pgd_present(*pgd)) {
		pud = pud_offset(pgd, taddr);
		if (pud_present(*pud)) {
			pmd = pmd_offset(pud, taddr);
			if (pmd_present(*pmd))
				pte = pte_offset_map(pmd, taddr);
		}
	}

	return pte;
}

int huge_pmd_unshare(struct mm_struct *mm, unsigned long *addr, pte_t *ptep)
{
	return 0;
}

#define mk_pte_huge(entry) { pte_val(entry) |= _PAGE_P; }

/*
 * Don't actually need to do any preparation, but need to make sure
 * the address is in the right region.
 */
int prepare_hugepage_range(struct file *file,
			unsigned long addr, unsigned long len)
{
	if (len & ~HPAGE_MASK)
		return -EINVAL;
	if (addr & ~HPAGE_MASK)
		return -EINVAL;
	if (REGION_NUMBER(addr) != RGN_HPAGE)
		return -EINVAL;

	return 0;
}

struct page *follow_huge_addr(struct mm_struct *mm, unsigned long addr, int write)
{
	struct page *page;
	pte_t *ptep;

	if (REGION_NUMBER(addr) != RGN_HPAGE)
		return ERR_PTR(-EINVAL);

	ptep = huge_pte_offset(mm, addr);
	if (!ptep || pte_none(*ptep))
		return NULL;
	page = pte_page(*ptep);
	page += ((addr & ~HPAGE_MASK) >> PAGE_SHIFT);
	return page;
}
int pmd_huge(pmd_t pmd)
{
	return 0;
}

int pud_huge(pud_t pud)
{
	return 0;
}

struct page *
follow_huge_pmd(struct mm_struct *mm, unsigned long address, pmd_t *pmd, int write)
{
	return NULL;
}

void hugetlb_free_pgd_range(struct mmu_gather *tlb,
			unsigned long addr, unsigned long end,
			unsigned long floor, unsigned long ceiling)
{
	/*
	 * This is called to free hugetlb page tables.
	 *
	 * The offset of these addresses from the base of the hugetlb
	 * region must be scaled down by HPAGE_SIZE/PAGE_SIZE so that
	 * the standard free_pgd_range will free the right page tables.
	 *
	 * If floor and ceiling are also in the hugetlb region, they
	 * must likewise be scaled down; but if outside, left unchanged.
	 */

	addr = htlbpage_to_page(addr);
	end  = htlbpage_to_page(end);
	if (REGION_NUMBER(floor) == RGN_HPAGE)
		floor = htlbpage_to_page(floor);
	if (REGION_NUMBER(ceiling) == RGN_HPAGE)
		ceiling = htlbpage_to_page(ceiling);

	free_pgd_range(tlb, addr, end, floor, ceiling);
}

unsigned long hugetlb_get_unmapped_area(struct file *file, unsigned long addr, unsigned long len,
		unsigned long pgoff, unsigned long flags)
{
	struct vm_area_struct *vmm;

	if (len > RGN_MAP_LIMIT)
		return -ENOMEM;
	if (len & ~HPAGE_MASK)
		return -EINVAL;

	/* Handle MAP_FIXED */
	if (flags & MAP_FIXED) {
		if (prepare_hugepage_range(file, addr, len))
			return -EINVAL;
		return addr;
	}

	/* This code assumes that RGN_HPAGE != 0. */
	if ((REGION_NUMBER(addr) != RGN_HPAGE) || (addr & (HPAGE_SIZE - 1)))
		addr = HPAGE_REGION_BASE;
	else
		addr = ALIGN(addr, HPAGE_SIZE);
	for (vmm = find_vma(current->mm, addr); ; vmm = vmm->vm_next) {
		/* At this point:  (!vmm || addr < vmm->vm_end). */
		if (REGION_OFFSET(addr) + len > RGN_MAP_LIMIT)
			return -ENOMEM;
		if (!vmm || (addr + len) <= vmm->vm_start)
			return addr;
		addr = ALIGN(vmm->vm_end, HPAGE_SIZE);
	}
}

static int __init hugetlb_setup_sz(char *str)
{
	u64 tr_pages;
	unsigned long long size;

	if (ia64_pal_vm_page_size(&tr_pages, NULL) != 0)
		/*
		 * shouldn't happen, but just in case.
		 */
		tr_pages = 0x15557000UL;

	size = memparse(str, &str);
	if (*str || !is_power_of_2(size) || !(tr_pages & size) ||
		size <= PAGE_SIZE ||
		size >= (1UL << PAGE_SHIFT << MAX_ORDER)) {
		printk(KERN_WARNING "Invalid huge page size specified\n");
		return 1;
	}

	hpage_shift = __ffs(size);
	/*
	 * boot cpu already executed ia64_mmu_init, and has HPAGE_SHIFT_DEFAULT
	 * override here with new page shift.
	 */
	ia64_set_rr(HPAGE_REGION_BASE, hpage_shift << 2);
	return 0;
}
early_param("hugepagesz", hugetlb_setup_sz);
Commit	Line	Data
1da177e4 LT	1	/*
	2	* IA-64 Huge TLB Page Support for Kernel.
	3	*
	4	* Copyright (C) 2002-2004 Rohit Seth <rohit.seth@intel.com>
	5	* Copyright (C) 2003-2004 Ken Chen <kenneth.w.chen@intel.com>
	6	*
	7	* Sep, 2003: add numa support
	8	* Feb, 2004: dynamic hugetlb page size via boot parameter
	9	*/
	10
1da177e4 LT	11	#include <linux/init.h>
	12	#include <linux/fs.h>
	13	#include <linux/mm.h>
	14	#include <linux/hugetlb.h>
	15	#include <linux/pagemap.h>
34d8a380	16	#include <linux/module.h>
1da177e4	17	#include <linux/sysctl.h>
9be26f4c	18	#include <linux/log2.h>
1da177e4 LT	19	#include <asm/mman.h>
	20	#include <asm/pgalloc.h>
	21	#include <asm/tlb.h>
	22	#include <asm/tlbflush.h>
	23
34d8a380 JS	24	unsigned int hpage_shift = HPAGE_SHIFT_DEFAULT;
34d8a380 JS	25	EXPORT_SYMBOL(hpage_shift);
1da177e4	26
63551ae0	27	pte_t *
a5516438	28	huge_pte_alloc(struct mm_struct *mm, unsigned long addr, unsigned long sz)
1da177e4 LT	29	{
	30	unsigned long taddr = htlbpage_to_page(addr);
	31	pgd_t *pgd;
	32	pud_t *pud;
	33	pmd_t *pmd;
	34	pte_t *pte = NULL;
	35
	36	pgd = pgd_offset(mm, taddr);
	37	pud = pud_alloc(mm, pgd, taddr);
	38	if (pud) {
	39	pmd = pmd_alloc(mm, pud, taddr);
	40	if (pmd)
	41	pte = pte_alloc_map(mm, pmd, taddr);
	42	}
	43	return pte;
	44	}
	45
63551ae0	46	pte_t *
1da177e4 LT	47	huge_pte_offset (struct mm_struct *mm, unsigned long addr)
	48	{
	49	unsigned long taddr = htlbpage_to_page(addr);
	50	pgd_t *pgd;
	51	pud_t *pud;
	52	pmd_t *pmd;
	53	pte_t *pte = NULL;
	54
	55	pgd = pgd_offset(mm, taddr);
	56	if (pgd_present(*pgd)) {
	57	pud = pud_offset(pgd, taddr);
	58	if (pud_present(*pud)) {
	59	pmd = pmd_offset(pud, taddr);
	60	if (pmd_present(*pmd))
	61	pte = pte_offset_map(pmd, taddr);
	62	}
	63	}
	64
	65	return pte;
	66	}
	67
39dde65c KC	68	int huge_pmd_unshare(struct mm_struct mm, unsigned long addr, pte_t *ptep)
	69	{
	70	return 0;
	71	}
	72
1da177e4 LT	73	#define mk_pte_huge(entry) { pte_val(entry) \|= _PAGE_P; }
1da177e4 LT	74
1da177e4	75	/*
42b88bef DG	76	* Don't actually need to do any preparation, but need to make sure
42b88bef DG	77	* the address is in the right region.
1da177e4	78	*/
a5516438 AK	79	int prepare_hugepage_range(struct file *file,
a5516438 AK	80	unsigned long addr, unsigned long len)
1da177e4 LT	81	{
	82	if (len & ~HPAGE_MASK)
	83	return -EINVAL;
	84	if (addr & ~HPAGE_MASK)
	85	return -EINVAL;
0a41e250	86	if (REGION_NUMBER(addr) != RGN_HPAGE)
1da177e4 LT	87	return -EINVAL;
	88
	89	return 0;
	90	}
	91
1da177e4 LT	92	struct page follow_huge_addr(struct mm_struct mm, unsigned long addr, int write)
	93	{
	94	struct page *page;
	95	pte_t *ptep;
	96
0a41e250	97	if (REGION_NUMBER(addr) != RGN_HPAGE)
1da177e4 LT	98	return ERR_PTR(-EINVAL);
	99
	100	ptep = huge_pte_offset(mm, addr);
	101	if (!ptep \|\| pte_none(*ptep))
	102	return NULL;
	103	page = pte_page(*ptep);
	104	page += ((addr & ~HPAGE_MASK) >> PAGE_SHIFT);
	105	return page;
	106	}
	107	int pmd_huge(pmd_t pmd)
	108	{
	109	return 0;
	110	}
ceb86879 AK	111
	112	int pud_huge(pud_t pud)
	113	{
	114	return 0;
	115	}
	116
1da177e4 LT	117	struct page *
	118	follow_huge_pmd(struct mm_struct mm, unsigned long address, pmd_t pmd, int write)
	119	{
	120	return NULL;
	121	}
	122
42b77728	123	void hugetlb_free_pgd_range(struct mmu_gather *tlb,
3bf5ee95 HD	124	unsigned long addr, unsigned long end,
3bf5ee95 HD	125	unsigned long floor, unsigned long ceiling)
1da177e4	126	{
3bf5ee95	127	/*
2332c9ae	128	* This is called to free hugetlb page tables.
3bf5ee95 HD	129	*
	130	* The offset of these addresses from the base of the hugetlb
	131	* region must be scaled down by HPAGE_SIZE/PAGE_SIZE so that
	132	* the standard free_pgd_range will free the right page tables.
	133	*
	134	* If floor and ceiling are also in the hugetlb region, they
	135	* must likewise be scaled down; but if outside, left unchanged.
	136	*/
	137
	138	addr = htlbpage_to_page(addr);
	139	end = htlbpage_to_page(end);
2332c9ae	140	if (REGION_NUMBER(floor) == RGN_HPAGE)
3bf5ee95	141	floor = htlbpage_to_page(floor);
2332c9ae	142	if (REGION_NUMBER(ceiling) == RGN_HPAGE)
3bf5ee95 HD	143	ceiling = htlbpage_to_page(ceiling);
	144
	145	free_pgd_range(tlb, addr, end, floor, ceiling);
1da177e4 LT	146	}
1da177e4 LT	147
1da177e4 LT	148	unsigned long hugetlb_get_unmapped_area(struct file *file, unsigned long addr, unsigned long len,
	149	unsigned long pgoff, unsigned long flags)
	150	{
	151	struct vm_area_struct *vmm;
	152
	153	if (len > RGN_MAP_LIMIT)
	154	return -ENOMEM;
	155	if (len & ~HPAGE_MASK)
	156	return -EINVAL;
afa37394 BH	157
	158	/* Handle MAP_FIXED */
	159	if (flags & MAP_FIXED) {
a5516438	160	if (prepare_hugepage_range(file, addr, len))
afa37394 BH	161	return -EINVAL;
	162	return addr;
	163	}
	164
0a41e250 PC	165	/* This code assumes that RGN_HPAGE != 0. */
0a41e250 PC	166	if ((REGION_NUMBER(addr) != RGN_HPAGE) \|\| (addr & (HPAGE_SIZE - 1)))
1da177e4 LT	167	addr = HPAGE_REGION_BASE;
	168	else
	169	addr = ALIGN(addr, HPAGE_SIZE);
	170	for (vmm = find_vma(current->mm, addr); ; vmm = vmm->vm_next) {
	171	/* At this point: (!vmm \|\| addr < vmm->vm_end). */
	172	if (REGION_OFFSET(addr) + len > RGN_MAP_LIMIT)
	173	return -ENOMEM;
	174	if (!vmm \|\| (addr + len) <= vmm->vm_start)
	175	return addr;
	176	addr = ALIGN(vmm->vm_end, HPAGE_SIZE);
	177	}
	178	}
	179
	180	static int __init hugetlb_setup_sz(char *str)
	181	{
	182	u64 tr_pages;
	183	unsigned long long size;
	184
	185	if (ia64_pal_vm_page_size(&tr_pages, NULL) != 0)
	186	/*
	187	* shouldn't happen, but just in case.
	188	*/
	189	tr_pages = 0x15557000UL;
	190
	191	size = memparse(str, &str);
9be26f4c	192	if (*str \|\| !is_power_of_2(size) \|\| !(tr_pages & size) \|\|
1da177e4 LT	193	size <= PAGE_SIZE \|\|
	194	size >= (1UL << PAGE_SHIFT << MAX_ORDER)) {
	195	printk(KERN_WARNING "Invalid huge page size specified\n");
	196	return 1;
	197	}
	198
	199	hpage_shift = __ffs(size);
	200	/*
	201	* boot cpu already executed ia64_mmu_init, and has HPAGE_SHIFT_DEFAULT
	202	* override here with new page shift.
	203	*/
	204	ia64_set_rr(HPAGE_REGION_BASE, hpage_shift << 2);
d9c23400	205	return 0;
1da177e4	206	}
d9c23400	207	early_param("hugepagesz", hugetlb_setup_sz);