[GitHub/mt8127/android_kernel_alcatel_ttab.git] / arch / i386 / mm / pgtable.c

/*
 *  linux/arch/i386/mm/pgtable.c
 */

#include <linux/sched.h>
#include <linux/kernel.h>
#include <linux/errno.h>
#include <linux/mm.h>
#include <linux/swap.h>
#include <linux/smp.h>
#include <linux/highmem.h>
#include <linux/slab.h>
#include <linux/pagemap.h>
#include <linux/spinlock.h>
#include <linux/module.h>
#include <linux/quicklist.h>

#include <asm/system.h>
#include <asm/pgtable.h>
#include <asm/pgalloc.h>
#include <asm/fixmap.h>
#include <asm/e820.h>
#include <asm/tlb.h>
#include <asm/tlbflush.h>

void show_mem(void)
{
	int total = 0, reserved = 0;
	int shared = 0, cached = 0;
	int highmem = 0;
	struct page *page;
	pg_data_t *pgdat;
	unsigned long i;
	unsigned long flags;

	printk(KERN_INFO "Mem-info:\n");
	show_free_areas();
	printk(KERN_INFO "Free swap:       %6ldkB\n", nr_swap_pages<<(PAGE_SHIFT-10));
	for_each_online_pgdat(pgdat) {
		pgdat_resize_lock(pgdat, &flags);
		for (i = 0; i < pgdat->node_spanned_pages; ++i) {
			page = pgdat_page_nr(pgdat, i);
			total++;
			if (PageHighMem(page))
				highmem++;
			if (PageReserved(page))
				reserved++;
			else if (PageSwapCache(page))
				cached++;
			else if (page_count(page))
				shared += page_count(page) - 1;
		}
		pgdat_resize_unlock(pgdat, &flags);
	}
	printk(KERN_INFO "%d pages of RAM\n", total);
	printk(KERN_INFO "%d pages of HIGHMEM\n", highmem);
	printk(KERN_INFO "%d reserved pages\n", reserved);
	printk(KERN_INFO "%d pages shared\n", shared);
	printk(KERN_INFO "%d pages swap cached\n", cached);

	printk(KERN_INFO "%lu pages dirty\n", global_page_state(NR_FILE_DIRTY));
	printk(KERN_INFO "%lu pages writeback\n",
					global_page_state(NR_WRITEBACK));
	printk(KERN_INFO "%lu pages mapped\n", global_page_state(NR_FILE_MAPPED));
	printk(KERN_INFO "%lu pages slab\n",
		global_page_state(NR_SLAB_RECLAIMABLE) +
		global_page_state(NR_SLAB_UNRECLAIMABLE));
	printk(KERN_INFO "%lu pages pagetables\n",
					global_page_state(NR_PAGETABLE));
}

/*
 * Associate a virtual page frame with a given physical page frame 
 * and protection flags for that frame.
 */ 
static void set_pte_pfn(unsigned long vaddr, unsigned long pfn, pgprot_t flags)
{
	pgd_t *pgd;
	pud_t *pud;
	pmd_t *pmd;
	pte_t *pte;

	pgd = swapper_pg_dir + pgd_index(vaddr);
	if (pgd_none(*pgd)) {
		BUG();
		return;
	}
	pud = pud_offset(pgd, vaddr);
	if (pud_none(*pud)) {
		BUG();
		return;
	}
	pmd = pmd_offset(pud, vaddr);
	if (pmd_none(*pmd)) {
		BUG();
		return;
	}
	pte = pte_offset_kernel(pmd, vaddr);
	if (pgprot_val(flags))
		/* <pfn,flags> stored as-is, to permit clearing entries */
		set_pte(pte, pfn_pte(pfn, flags));
	else
		pte_clear(&init_mm, vaddr, pte);

	/*
	 * It's enough to flush this one mapping.
	 * (PGE mappings get flushed as well)
	 */
	__flush_tlb_one(vaddr);
}

/*
 * Associate a large virtual page frame with a given physical page frame 
 * and protection flags for that frame. pfn is for the base of the page,
 * vaddr is what the page gets mapped to - both must be properly aligned. 
 * The pmd must already be instantiated. Assumes PAE mode.
 */ 
void set_pmd_pfn(unsigned long vaddr, unsigned long pfn, pgprot_t flags)
{
	pgd_t *pgd;
	pud_t *pud;
	pmd_t *pmd;

	if (vaddr & (PMD_SIZE-1)) {		/* vaddr is misaligned */
		printk(KERN_WARNING "set_pmd_pfn: vaddr misaligned\n");
		return; /* BUG(); */
	}
	if (pfn & (PTRS_PER_PTE-1)) {		/* pfn is misaligned */
		printk(KERN_WARNING "set_pmd_pfn: pfn misaligned\n");
		return; /* BUG(); */
	}
	pgd = swapper_pg_dir + pgd_index(vaddr);
	if (pgd_none(*pgd)) {
		printk(KERN_WARNING "set_pmd_pfn: pgd_none\n");
		return; /* BUG(); */
	}
	pud = pud_offset(pgd, vaddr);
	pmd = pmd_offset(pud, vaddr);
	set_pmd(pmd, pfn_pmd(pfn, flags));
	/*
	 * It's enough to flush this one mapping.
	 * (PGE mappings get flushed as well)
	 */
	__flush_tlb_one(vaddr);
}

static int fixmaps;
unsigned long __FIXADDR_TOP = 0xfffff000;
EXPORT_SYMBOL(__FIXADDR_TOP);

void __set_fixmap (enum fixed_addresses idx, unsigned long phys, pgprot_t flags)
{
	unsigned long address = __fix_to_virt(idx);

	if (idx >= __end_of_fixed_addresses) {
		BUG();
		return;
	}
	set_pte_pfn(address, phys >> PAGE_SHIFT, flags);
	fixmaps++;
}

/**
 * reserve_top_address - reserves a hole in the top of kernel address space
 * @reserve - size of hole to reserve
 *
 * Can be used to relocate the fixmap area and poke a hole in the top
 * of kernel address space to make room for a hypervisor.
 */
void reserve_top_address(unsigned long reserve)
{
	BUG_ON(fixmaps > 0);
	printk(KERN_INFO "Reserving virtual address space above 0x%08x\n",
	       (int)-reserve);
	__FIXADDR_TOP = -reserve - PAGE_SIZE;
	__VMALLOC_RESERVE += reserve;
}

pte_t *pte_alloc_one_kernel(struct mm_struct *mm, unsigned long address)
{
	return (pte_t *)__get_free_page(GFP_KERNEL|__GFP_REPEAT|__GFP_ZERO);
}

struct page *pte_alloc_one(struct mm_struct *mm, unsigned long address)
{
	struct page *pte;

#ifdef CONFIG_HIGHPTE
	pte = alloc_pages(GFP_KERNEL|__GFP_HIGHMEM|__GFP_REPEAT|__GFP_ZERO, 0);
#else
	pte = alloc_pages(GFP_KERNEL|__GFP_REPEAT|__GFP_ZERO, 0);
#endif
	return pte;
}

void pmd_ctor(void *pmd, struct kmem_cache *cache, unsigned long flags)
{
	memset(pmd, 0, PTRS_PER_PMD*sizeof(pmd_t));
}

/*
 * List of all pgd's needed for non-PAE so it can invalidate entries
 * in both cached and uncached pgd's; not needed for PAE since the
 * kernel pmd is shared. If PAE were not to share the pmd a similar
 * tactic would be needed. This is essentially codepath-based locking
 * against pageattr.c; it is the unique case in which a valid change
 * of kernel pagetables can't be lazily synchronized by vmalloc faults.
 * vmalloc faults work because attached pagetables are never freed.
 * -- wli
 */
DEFINE_SPINLOCK(pgd_lock);
struct page *pgd_list;

static inline void pgd_list_add(pgd_t *pgd)
{
	struct page *page = virt_to_page(pgd);
	page->index = (unsigned long)pgd_list;
	if (pgd_list)
		set_page_private(pgd_list, (unsigned long)&page->index);
	pgd_list = page;
	set_page_private(page, (unsigned long)&pgd_list);
}

static inline void pgd_list_del(pgd_t *pgd)
{
	struct page *next, **pprev, *page = virt_to_page(pgd);
	next = (struct page *)page->index;
	pprev = (struct page **)page_private(page);
	*pprev = next;
	if (next)
		set_page_private(next, (unsigned long)pprev);
}


#if (PTRS_PER_PMD == 1)
/* Non-PAE pgd constructor */
void pgd_ctor(void *pgd)
{
	unsigned long flags;

	/* !PAE, no pagetable sharing */
	memset(pgd, 0, USER_PTRS_PER_PGD*sizeof(pgd_t));

	spin_lock_irqsave(&pgd_lock, flags);

	/* must happen under lock */
	clone_pgd_range((pgd_t *)pgd + USER_PTRS_PER_PGD,
			swapper_pg_dir + USER_PTRS_PER_PGD,
			KERNEL_PGD_PTRS);
	paravirt_alloc_pd_clone(__pa(pgd) >> PAGE_SHIFT,
				__pa(swapper_pg_dir) >> PAGE_SHIFT,
				USER_PTRS_PER_PGD,
				KERNEL_PGD_PTRS);
	pgd_list_add(pgd);
	spin_unlock_irqrestore(&pgd_lock, flags);
}
#else  /* PTRS_PER_PMD > 1 */
/* PAE pgd constructor */
void pgd_ctor(void *pgd)
{
	/* PAE, kernel PMD may be shared */

	if (SHARED_KERNEL_PMD) {
		clone_pgd_range((pgd_t *)pgd + USER_PTRS_PER_PGD,
				swapper_pg_dir + USER_PTRS_PER_PGD,
				KERNEL_PGD_PTRS);
	} else {
		unsigned long flags;

		memset(pgd, 0, USER_PTRS_PER_PGD*sizeof(pgd_t));
		spin_lock_irqsave(&pgd_lock, flags);
		pgd_list_add(pgd);
		spin_unlock_irqrestore(&pgd_lock, flags);
	}
}
#endif	/* PTRS_PER_PMD */

void pgd_dtor(void *pgd)
{
	unsigned long flags; /* can be called from interrupt context */

	if (SHARED_KERNEL_PMD)
		return;

	paravirt_release_pd(__pa(pgd) >> PAGE_SHIFT);
	spin_lock_irqsave(&pgd_lock, flags);
	pgd_list_del(pgd);
	spin_unlock_irqrestore(&pgd_lock, flags);
}

#define UNSHARED_PTRS_PER_PGD				\
	(SHARED_KERNEL_PMD ? USER_PTRS_PER_PGD : PTRS_PER_PGD)

/* If we allocate a pmd for part of the kernel address space, then
   make sure its initialized with the appropriate kernel mappings.
   Otherwise use a cached zeroed pmd.  */
static pmd_t *pmd_cache_alloc(int idx)
{
	pmd_t *pmd;

	if (idx >= USER_PTRS_PER_PGD) {
		pmd = (pmd_t *)__get_free_page(GFP_KERNEL);

		if (pmd)
			memcpy(pmd,
			       (void *)pgd_page_vaddr(swapper_pg_dir[idx]),
			       sizeof(pmd_t) * PTRS_PER_PMD);
	} else
		pmd = kmem_cache_alloc(pmd_cache, GFP_KERNEL);

	return pmd;
}

static void pmd_cache_free(pmd_t *pmd, int idx)
{
	if (idx >= USER_PTRS_PER_PGD)
		free_page((unsigned long)pmd);
	else
		kmem_cache_free(pmd_cache, pmd);
}

pgd_t *pgd_alloc(struct mm_struct *mm)
{
	int i;
	pgd_t *pgd = quicklist_alloc(0, GFP_KERNEL, pgd_ctor);

	if (PTRS_PER_PMD == 1 || !pgd)
		return pgd;

 	for (i = 0; i < UNSHARED_PTRS_PER_PGD; ++i) {
		pmd_t *pmd = pmd_cache_alloc(i);

		if (!pmd)
			goto out_oom;

		paravirt_alloc_pd(__pa(pmd) >> PAGE_SHIFT);
		set_pgd(&pgd[i], __pgd(1 + __pa(pmd)));
	}
	return pgd;

out_oom:
	for (i--; i >= 0; i--) {
		pgd_t pgdent = pgd[i];
		void* pmd = (void *)__va(pgd_val(pgdent)-1);
		paravirt_release_pd(__pa(pmd) >> PAGE_SHIFT);
		pmd_cache_free(pmd, i);
	}
	quicklist_free(0, pgd_dtor, pgd);
	return NULL;
}

void pgd_free(pgd_t *pgd)
{
	int i;

	/* in the PAE case user pgd entries are overwritten before usage */
	if (PTRS_PER_PMD > 1)
		for (i = 0; i < UNSHARED_PTRS_PER_PGD; ++i) {
			pgd_t pgdent = pgd[i];
			void* pmd = (void *)__va(pgd_val(pgdent)-1);
			paravirt_release_pd(__pa(pmd) >> PAGE_SHIFT);
			pmd_cache_free(pmd, i);
		}
	/* in the non-PAE case, free_pgtables() clears user pgd entries */
	quicklist_free(0, pgd_dtor, pgd);
}

void check_pgt_cache(void)
{
	quicklist_trim(0, pgd_dtor, 25, 16);
}
Commit	Line	Data
1da177e4 LT	1	/*
	2	* linux/arch/i386/mm/pgtable.c
	3	*/
	4
1da177e4 LT	5	#include <linux/sched.h>
	6	#include <linux/kernel.h>
	7	#include <linux/errno.h>
	8	#include <linux/mm.h>
	9	#include <linux/swap.h>
	10	#include <linux/smp.h>
	11	#include <linux/highmem.h>
	12	#include <linux/slab.h>
	13	#include <linux/pagemap.h>
	14	#include <linux/spinlock.h>
052e7994	15	#include <linux/module.h>
f1d1a842	16	#include <linux/quicklist.h>
1da177e4 LT	17
	18	#include <asm/system.h>
	19	#include <asm/pgtable.h>
	20	#include <asm/pgalloc.h>
	21	#include <asm/fixmap.h>
	22	#include <asm/e820.h>
	23	#include <asm/tlb.h>
	24	#include <asm/tlbflush.h>
	25
	26	void show_mem(void)
	27	{
	28	int total = 0, reserved = 0;
	29	int shared = 0, cached = 0;
	30	int highmem = 0;
	31	struct page *page;
	32	pg_data_t *pgdat;
	33	unsigned long i;
208d54e5	34	unsigned long flags;
1da177e4	35
f90e7185	36	printk(KERN_INFO "Mem-info:\n");
1da177e4	37	show_free_areas();
f90e7185	38	printk(KERN_INFO "Free swap: %6ldkB\n", nr_swap_pages<<(PAGE_SHIFT-10));
ec936fc5	39	for_each_online_pgdat(pgdat) {
208d54e5	40	pgdat_resize_lock(pgdat, &flags);
1da177e4	41	for (i = 0; i < pgdat->node_spanned_pages; ++i) {
408fde81	42	page = pgdat_page_nr(pgdat, i);
1da177e4 LT	43	total++;
	44	if (PageHighMem(page))
	45	highmem++;
	46	if (PageReserved(page))
	47	reserved++;
	48	else if (PageSwapCache(page))
	49	cached++;
	50	else if (page_count(page))
	51	shared += page_count(page) - 1;
	52	}
208d54e5	53	pgdat_resize_unlock(pgdat, &flags);
1da177e4	54	}
f90e7185 CL	55	printk(KERN_INFO "%d pages of RAM\n", total);
	56	printk(KERN_INFO "%d pages of HIGHMEM\n", highmem);
	57	printk(KERN_INFO "%d reserved pages\n", reserved);
	58	printk(KERN_INFO "%d pages shared\n", shared);
	59	printk(KERN_INFO "%d pages swap cached\n", cached);
6f4e1e50	60
b1e7a8fd	61	printk(KERN_INFO "%lu pages dirty\n", global_page_state(NR_FILE_DIRTY));
ce866b34 CL	62	printk(KERN_INFO "%lu pages writeback\n",
ce866b34 CL	63	global_page_state(NR_WRITEBACK));
65ba55f5	64	printk(KERN_INFO "%lu pages mapped\n", global_page_state(NR_FILE_MAPPED));
972d1a7b CL	65	printk(KERN_INFO "%lu pages slab\n",
	66	global_page_state(NR_SLAB_RECLAIMABLE) +
	67	global_page_state(NR_SLAB_UNRECLAIMABLE));
df849a15 CL	68	printk(KERN_INFO "%lu pages pagetables\n",
df849a15 CL	69	global_page_state(NR_PAGETABLE));
1da177e4 LT	70	}
	71
	72	/*
	73	* Associate a virtual page frame with a given physical page frame
	74	* and protection flags for that frame.
	75	*/
	76	static void set_pte_pfn(unsigned long vaddr, unsigned long pfn, pgprot_t flags)
	77	{
	78	pgd_t *pgd;
	79	pud_t *pud;
	80	pmd_t *pmd;
	81	pte_t *pte;
	82
	83	pgd = swapper_pg_dir + pgd_index(vaddr);
	84	if (pgd_none(*pgd)) {
	85	BUG();
	86	return;
	87	}
	88	pud = pud_offset(pgd, vaddr);
	89	if (pud_none(*pud)) {
	90	BUG();
	91	return;
	92	}
	93	pmd = pmd_offset(pud, vaddr);
	94	if (pmd_none(*pmd)) {
	95	BUG();
	96	return;
	97	}
	98	pte = pte_offset_kernel(pmd, vaddr);
b0bfece4 JB	99	if (pgprot_val(flags))
	100	/* <pfn,flags> stored as-is, to permit clearing entries */
	101	set_pte(pte, pfn_pte(pfn, flags));
	102	else
	103	pte_clear(&init_mm, vaddr, pte);
1da177e4 LT	104
	105	/*
	106	* It's enough to flush this one mapping.
	107	* (PGE mappings get flushed as well)
	108	*/
	109	__flush_tlb_one(vaddr);
	110	}
	111
	112	/*
	113	* Associate a large virtual page frame with a given physical page frame
	114	* and protection flags for that frame. pfn is for the base of the page,
	115	* vaddr is what the page gets mapped to - both must be properly aligned.
	116	* The pmd must already be instantiated. Assumes PAE mode.
	117	*/
	118	void set_pmd_pfn(unsigned long vaddr, unsigned long pfn, pgprot_t flags)
	119	{
	120	pgd_t *pgd;
	121	pud_t *pud;
	122	pmd_t *pmd;
	123
	124	if (vaddr & (PMD_SIZE-1)) { /* vaddr is misaligned */
f90e7185	125	printk(KERN_WARNING "set_pmd_pfn: vaddr misaligned\n");
1da177e4 LT	126	return; /* BUG(); */
	127	}
	128	if (pfn & (PTRS_PER_PTE-1)) { /* pfn is misaligned */
f90e7185	129	printk(KERN_WARNING "set_pmd_pfn: pfn misaligned\n");
1da177e4 LT	130	return; /* BUG(); */
	131	}
	132	pgd = swapper_pg_dir + pgd_index(vaddr);
	133	if (pgd_none(*pgd)) {
f90e7185	134	printk(KERN_WARNING "set_pmd_pfn: pgd_none\n");
1da177e4 LT	135	return; /* BUG(); */
	136	}
	137	pud = pud_offset(pgd, vaddr);
	138	pmd = pmd_offset(pud, vaddr);
	139	set_pmd(pmd, pfn_pmd(pfn, flags));
	140	/*
	141	* It's enough to flush this one mapping.
	142	* (PGE mappings get flushed as well)
	143	*/
	144	__flush_tlb_one(vaddr);
	145	}
	146
052e7994	147	static int fixmaps;
052e7994 JF	148	unsigned long __FIXADDR_TOP = 0xfffff000;
052e7994 JF	149	EXPORT_SYMBOL(__FIXADDR_TOP);
052e7994	150
1da177e4 LT	151	void __set_fixmap (enum fixed_addresses idx, unsigned long phys, pgprot_t flags)
	152	{
	153	unsigned long address = __fix_to_virt(idx);
	154
	155	if (idx >= __end_of_fixed_addresses) {
	156	BUG();
	157	return;
	158	}
	159	set_pte_pfn(address, phys >> PAGE_SHIFT, flags);
052e7994 JF	160	fixmaps++;
	161	}
	162
	163	/**
	164	* reserve_top_address - reserves a hole in the top of kernel address space
	165	* @reserve - size of hole to reserve
	166	*
	167	* Can be used to relocate the fixmap area and poke a hole in the top
	168	* of kernel address space to make room for a hypervisor.
	169	*/
	170	void reserve_top_address(unsigned long reserve)
	171	{
	172	BUG_ON(fixmaps > 0);
7ce0bcfd ZA	173	printk(KERN_INFO "Reserving virtual address space above 0x%08x\n",
7ce0bcfd ZA	174	(int)-reserve);
052e7994 JF	175	__FIXADDR_TOP = -reserve - PAGE_SIZE;
052e7994 JF	176	__VMALLOC_RESERVE += reserve;
1da177e4 LT	177	}
	178
	179	pte_t pte_alloc_one_kernel(struct mm_struct mm, unsigned long address)
	180	{
	181	return (pte_t *)__get_free_page(GFP_KERNEL\|__GFP_REPEAT\|__GFP_ZERO);
	182	}
	183
	184	struct page pte_alloc_one(struct mm_struct mm, unsigned long address)
	185	{
	186	struct page *pte;
	187
	188	#ifdef CONFIG_HIGHPTE
	189	pte = alloc_pages(GFP_KERNEL\|__GFP_HIGHMEM\|__GFP_REPEAT\|__GFP_ZERO, 0);
	190	#else
	191	pte = alloc_pages(GFP_KERNEL\|__GFP_REPEAT\|__GFP_ZERO, 0);
	192	#endif
	193	return pte;
	194	}
	195
e18b890b	196	void pmd_ctor(void pmd, struct kmem_cache cache, unsigned long flags)
1da177e4 LT	197	{
	198	memset(pmd, 0, PTRS_PER_PMD*sizeof(pmd_t));
	199	}
	200
	201	/*
	202	* List of all pgd's needed for non-PAE so it can invalidate entries
	203	* in both cached and uncached pgd's; not needed for PAE since the
	204	* kernel pmd is shared. If PAE were not to share the pmd a similar
	205	* tactic would be needed. This is essentially codepath-based locking
	206	* against pageattr.c; it is the unique case in which a valid change
	207	* of kernel pagetables can't be lazily synchronized by vmalloc faults.
	208	* vmalloc faults work because attached pagetables are never freed.
1da177e4 LT	209	* -- wli
	210	*/
	211	DEFINE_SPINLOCK(pgd_lock);
	212	struct page *pgd_list;
	213
	214	static inline void pgd_list_add(pgd_t *pgd)
	215	{
	216	struct page *page = virt_to_page(pgd);
	217	page->index = (unsigned long)pgd_list;
	218	if (pgd_list)
4c21e2f2	219	set_page_private(pgd_list, (unsigned long)&page->index);
1da177e4	220	pgd_list = page;
4c21e2f2	221	set_page_private(page, (unsigned long)&pgd_list);
1da177e4 LT	222	}
	223
	224	static inline void pgd_list_del(pgd_t *pgd)
	225	{
	226	struct page next, pprev, page = virt_to_page(pgd);
	227	next = (struct page *)page->index;
4c21e2f2	228	pprev = (struct page **)page_private(page);
1da177e4 LT	229	*pprev = next;
1da177e4 LT	230	if (next)
4c21e2f2	231	set_page_private(next, (unsigned long)pprev);
1da177e4 LT	232	}
1da177e4 LT	233
f1d1a842 CL	234
f1d1a842 CL	235
5311ab62 JF	236	#if (PTRS_PER_PMD == 1)
5311ab62 JF	237	/* Non-PAE pgd constructor */
f1d1a842	238	void pgd_ctor(void *pgd)
1da177e4 LT	239	{
	240	unsigned long flags;
	241
5311ab62 JF	242	/* !PAE, no pagetable sharing */
	243	memset(pgd, 0, USER_PTRS_PER_PGD*sizeof(pgd_t));
	244
	245	spin_lock_irqsave(&pgd_lock, flags);
1da177e4	246
5311ab62	247	/* must happen under lock */
d7271b14	248	clone_pgd_range((pgd_t *)pgd + USER_PTRS_PER_PGD,
1da177e4	249	swapper_pg_dir + USER_PTRS_PER_PGD,
d7271b14	250	KERNEL_PGD_PTRS);
c119ecce	251	paravirt_alloc_pd_clone(__pa(pgd) >> PAGE_SHIFT,
5311ab62 JF	252	__pa(swapper_pg_dir) >> PAGE_SHIFT,
	253	USER_PTRS_PER_PGD,
	254	KERNEL_PGD_PTRS);
1da177e4 LT	255	pgd_list_add(pgd);
1da177e4 LT	256	spin_unlock_irqrestore(&pgd_lock, flags);
1da177e4	257	}
5311ab62 JF	258	#else /* PTRS_PER_PMD > 1 */
5311ab62 JF	259	/* PAE pgd constructor */
f1d1a842	260	void pgd_ctor(void *pgd)
5311ab62 JF	261	{
	262	/* PAE, kernel PMD may be shared */
	263
	264	if (SHARED_KERNEL_PMD) {
	265	clone_pgd_range((pgd_t *)pgd + USER_PTRS_PER_PGD,
	266	swapper_pg_dir + USER_PTRS_PER_PGD,
	267	KERNEL_PGD_PTRS);
	268	} else {
	269	unsigned long flags;
	270
	271	memset(pgd, 0, USER_PTRS_PER_PGD*sizeof(pgd_t));
	272	spin_lock_irqsave(&pgd_lock, flags);
	273	pgd_list_add(pgd);
	274	spin_unlock_irqrestore(&pgd_lock, flags);
	275	}
	276	}
	277	#endif /* PTRS_PER_PMD */
1da177e4	278
f1d1a842	279	void pgd_dtor(void *pgd)
1da177e4 LT	280	{
	281	unsigned long flags; /* can be called from interrupt context */
	282
f1d1a842 CL	283	if (SHARED_KERNEL_PMD)
f1d1a842 CL	284	return;
5311ab62	285
c119ecce	286	paravirt_release_pd(__pa(pgd) >> PAGE_SHIFT);
1da177e4 LT	287	spin_lock_irqsave(&pgd_lock, flags);
	288	pgd_list_del(pgd);
	289	spin_unlock_irqrestore(&pgd_lock, flags);
	290	}
	291
5311ab62 JF	292	#define UNSHARED_PTRS_PER_PGD \
	293	(SHARED_KERNEL_PMD ? USER_PTRS_PER_PGD : PTRS_PER_PGD)
	294
	295	/* If we allocate a pmd for part of the kernel address space, then
	296	make sure its initialized with the appropriate kernel mappings.
	297	Otherwise use a cached zeroed pmd. */
	298	static pmd_t *pmd_cache_alloc(int idx)
	299	{
	300	pmd_t *pmd;
	301
	302	if (idx >= USER_PTRS_PER_PGD) {
	303	pmd = (pmd_t *)__get_free_page(GFP_KERNEL);
	304
	305	if (pmd)
	306	memcpy(pmd,
	307	(void *)pgd_page_vaddr(swapper_pg_dir[idx]),
	308	sizeof(pmd_t) * PTRS_PER_PMD);
	309	} else
	310	pmd = kmem_cache_alloc(pmd_cache, GFP_KERNEL);
	311
	312	return pmd;
	313	}
	314
	315	static void pmd_cache_free(pmd_t *pmd, int idx)
	316	{
	317	if (idx >= USER_PTRS_PER_PGD)
	318	free_page((unsigned long)pmd);
	319	else
	320	kmem_cache_free(pmd_cache, pmd);
	321	}
	322
1da177e4 LT	323	pgd_t pgd_alloc(struct mm_struct mm)
	324	{
	325	int i;
f1d1a842	326	pgd_t *pgd = quicklist_alloc(0, GFP_KERNEL, pgd_ctor);
1da177e4 LT	327
	328	if (PTRS_PER_PMD == 1 \|\| !pgd)
	329	return pgd;
	330
5311ab62 JF	331	for (i = 0; i < UNSHARED_PTRS_PER_PGD; ++i) {
	332	pmd_t *pmd = pmd_cache_alloc(i);
	333
1da177e4 LT	334	if (!pmd)
1da177e4 LT	335	goto out_oom;
5311ab62	336
c119ecce	337	paravirt_alloc_pd(__pa(pmd) >> PAGE_SHIFT);
1da177e4 LT	338	set_pgd(&pgd[i], __pgd(1 + __pa(pmd)));
	339	}
	340	return pgd;
	341
	342	out_oom:
c119ecce ZA	343	for (i--; i >= 0; i--) {
	344	pgd_t pgdent = pgd[i];
	345	void* pmd = (void *)__va(pgd_val(pgdent)-1);
	346	paravirt_release_pd(__pa(pmd) >> PAGE_SHIFT);
5311ab62	347	pmd_cache_free(pmd, i);
c119ecce	348	}
f1d1a842	349	quicklist_free(0, pgd_dtor, pgd);
1da177e4 LT	350	return NULL;
	351	}
	352
	353	void pgd_free(pgd_t *pgd)
	354	{
	355	int i;
	356
	357	/* in the PAE case user pgd entries are overwritten before usage */
	358	if (PTRS_PER_PMD > 1)
5311ab62	359	for (i = 0; i < UNSHARED_PTRS_PER_PGD; ++i) {
c119ecce ZA	360	pgd_t pgdent = pgd[i];
	361	void* pmd = (void *)__va(pgd_val(pgdent)-1);
	362	paravirt_release_pd(__pa(pmd) >> PAGE_SHIFT);
5311ab62	363	pmd_cache_free(pmd, i);
c119ecce	364	}
e0da382c	365	/* in the non-PAE case, free_pgtables() clears user pgd entries */
f1d1a842	366	quicklist_free(0, pgd_dtor, pgd);
1da177e4	367	}
f1d1a842 CL	368
	369	void check_pgt_cache(void)
	370	{
	371	quicklist_trim(0, pgd_dtor, 25, 16);
	372	}
	373