[linux.git] / mm / highmem.c

// SPDX-License-Identifier: GPL-2.0
/*
 * High memory handling common code and variables.
 *
 * (C) 1999 Andrea Arcangeli, SuSE GmbH, [email protected]
 *          Gerhard Wichert, Siemens AG, [email protected]
 *
 *
 * Redesigned the x86 32-bit VM architecture to deal with
 * 64-bit physical space. With current x86 CPUs this
 * means up to 64 Gigabytes physical RAM.
 *
 * Rewrote high memory support to move the page cache into
 * high memory. Implemented permanent (schedulable) kmaps
 * based on Linus' idea.
 *
 * Copyright (C) 1999 Ingo Molnar <[email protected]>
 */

#include <linux/mm.h>
#include <linux/export.h>
#include <linux/swap.h>
#include <linux/bio.h>
#include <linux/pagemap.h>
#include <linux/mempool.h>
#include <linux/blkdev.h>
#include <linux/init.h>
#include <linux/hash.h>
#include <linux/highmem.h>
#include <linux/kgdb.h>
#include <asm/tlbflush.h>


#if defined(CONFIG_HIGHMEM) || defined(CONFIG_X86_32)
DEFINE_PER_CPU(int, __kmap_atomic_idx);
#endif

/*
 * Virtual_count is not a pure "count".
 *  0 means that it is not mapped, and has not been mapped
 *    since a TLB flush - it is usable.
 *  1 means that there are no users, but it has been mapped
 *    since the last TLB flush - so we can't use it.
 *  n means that there are (n-1) current users of it.
 */
#ifdef CONFIG_HIGHMEM

/*
 * Architecture with aliasing data cache may define the following family of
 * helper functions in its asm/highmem.h to control cache color of virtual
 * addresses where physical memory pages are mapped by kmap.
 */
#ifndef get_pkmap_color

/*
 * Determine color of virtual address where the page should be mapped.
 */
static inline unsigned int get_pkmap_color(struct page *page)
{
	return 0;
}
#define get_pkmap_color get_pkmap_color

/*
 * Get next index for mapping inside PKMAP region for page with given color.
 */
static inline unsigned int get_next_pkmap_nr(unsigned int color)
{
	static unsigned int last_pkmap_nr;

	last_pkmap_nr = (last_pkmap_nr + 1) & LAST_PKMAP_MASK;
	return last_pkmap_nr;
}

/*
 * Determine if page index inside PKMAP region (pkmap_nr) of given color
 * has wrapped around PKMAP region end. When this happens an attempt to
 * flush all unused PKMAP slots is made.
 */
static inline int no_more_pkmaps(unsigned int pkmap_nr, unsigned int color)
{
	return pkmap_nr == 0;
}

/*
 * Get the number of PKMAP entries of the given color. If no free slot is
 * found after checking that many entries, kmap will sleep waiting for
 * someone to call kunmap and free PKMAP slot.
 */
static inline int get_pkmap_entries_count(unsigned int color)
{
	return LAST_PKMAP;
}

/*
 * Get head of a wait queue for PKMAP entries of the given color.
 * Wait queues for different mapping colors should be independent to avoid
 * unnecessary wakeups caused by freeing of slots of other colors.
 */
static inline wait_queue_head_t *get_pkmap_wait_queue_head(unsigned int color)
{
	static DECLARE_WAIT_QUEUE_HEAD(pkmap_map_wait);

	return &pkmap_map_wait;
}
#endif

unsigned long totalhigh_pages __read_mostly;
EXPORT_SYMBOL(totalhigh_pages);


EXPORT_PER_CPU_SYMBOL(__kmap_atomic_idx);

unsigned int nr_free_highpages (void)
{
	struct zone *zone;
	unsigned int pages = 0;

	for_each_populated_zone(zone) {
		if (is_highmem(zone))
			pages += zone_page_state(zone, NR_FREE_PAGES);
	}

	return pages;
}

static int pkmap_count[LAST_PKMAP];
static  __cacheline_aligned_in_smp DEFINE_SPINLOCK(kmap_lock);

pte_t * pkmap_page_table;

/*
 * Most architectures have no use for kmap_high_get(), so let's abstract
 * the disabling of IRQ out of the locking in that case to save on a
 * potential useless overhead.
 */
#ifdef ARCH_NEEDS_KMAP_HIGH_GET
#define lock_kmap()             spin_lock_irq(&kmap_lock)
#define unlock_kmap()           spin_unlock_irq(&kmap_lock)
#define lock_kmap_any(flags)    spin_lock_irqsave(&kmap_lock, flags)
#define unlock_kmap_any(flags)  spin_unlock_irqrestore(&kmap_lock, flags)
#else
#define lock_kmap()             spin_lock(&kmap_lock)
#define unlock_kmap()           spin_unlock(&kmap_lock)
#define lock_kmap_any(flags)    \
		do { spin_lock(&kmap_lock); (void)(flags); } while (0)
#define unlock_kmap_any(flags)  \
		do { spin_unlock(&kmap_lock); (void)(flags); } while (0)
#endif

struct page *kmap_to_page(void *vaddr)
{
	unsigned long addr = (unsigned long)vaddr;

	if (addr >= PKMAP_ADDR(0) && addr < PKMAP_ADDR(LAST_PKMAP)) {
		int i = PKMAP_NR(addr);
		return pte_page(pkmap_page_table[i]);
	}

	return virt_to_page(addr);
}
EXPORT_SYMBOL(kmap_to_page);

static void flush_all_zero_pkmaps(void)
{
	int i;
	int need_flush = 0;

	flush_cache_kmaps();

	for (i = 0; i < LAST_PKMAP; i++) {
		struct page *page;

		/*
		 * zero means we don't have anything to do,
		 * >1 means that it is still in use. Only
		 * a count of 1 means that it is free but
		 * needs to be unmapped
		 */
		if (pkmap_count[i] != 1)
			continue;
		pkmap_count[i] = 0;

		/* sanity check */
		BUG_ON(pte_none(pkmap_page_table[i]));

		/*
		 * Don't need an atomic fetch-and-clear op here;
		 * no-one has the page mapped, and cannot get at
		 * its virtual address (and hence PTE) without first
		 * getting the kmap_lock (which is held here).
		 * So no dangers, even with speculative execution.
		 */
		page = pte_page(pkmap_page_table[i]);
		pte_clear(&init_mm, PKMAP_ADDR(i), &pkmap_page_table[i]);

		set_page_address(page, NULL);
		need_flush = 1;
	}
	if (need_flush)
		flush_tlb_kernel_range(PKMAP_ADDR(0), PKMAP_ADDR(LAST_PKMAP));
}

/**
 * kmap_flush_unused - flush all unused kmap mappings in order to remove stray mappings
 */
void kmap_flush_unused(void)
{
	lock_kmap();
	flush_all_zero_pkmaps();
	unlock_kmap();
}

static inline unsigned long map_new_virtual(struct page *page)
{
	unsigned long vaddr;
	int count;
	unsigned int last_pkmap_nr;
	unsigned int color = get_pkmap_color(page);

start:
	count = get_pkmap_entries_count(color);
	/* Find an empty entry */
	for (;;) {
		last_pkmap_nr = get_next_pkmap_nr(color);
		if (no_more_pkmaps(last_pkmap_nr, color)) {
			flush_all_zero_pkmaps();
			count = get_pkmap_entries_count(color);
		}
		if (!pkmap_count[last_pkmap_nr])
			break;	/* Found a usable entry */
		if (--count)
			continue;

		/*
		 * Sleep for somebody else to unmap their entries
		 */
		{
			DECLARE_WAITQUEUE(wait, current);
			wait_queue_head_t *pkmap_map_wait =
				get_pkmap_wait_queue_head(color);

			__set_current_state(TASK_UNINTERRUPTIBLE);
			add_wait_queue(pkmap_map_wait, &wait);
			unlock_kmap();
			schedule();
			remove_wait_queue(pkmap_map_wait, &wait);
			lock_kmap();

			/* Somebody else might have mapped it while we slept */
			if (page_address(page))
				return (unsigned long)page_address(page);

			/* Re-start */
			goto start;
		}
	}
	vaddr = PKMAP_ADDR(last_pkmap_nr);
	set_pte_at(&init_mm, vaddr,
		   &(pkmap_page_table[last_pkmap_nr]), mk_pte(page, kmap_prot));

	pkmap_count[last_pkmap_nr] = 1;
	set_page_address(page, (void *)vaddr);

	return vaddr;
}

/**
 * kmap_high - map a highmem page into memory
 * @page: &struct page to map
 *
 * Returns the page's virtual memory address.
 *
 * We cannot call this from interrupts, as it may block.
 */
void *kmap_high(struct page *page)
{
	unsigned long vaddr;

	/*
	 * For highmem pages, we can't trust "virtual" until
	 * after we have the lock.
	 */
	lock_kmap();
	vaddr = (unsigned long)page_address(page);
	if (!vaddr)
		vaddr = map_new_virtual(page);
	pkmap_count[PKMAP_NR(vaddr)]++;
	BUG_ON(pkmap_count[PKMAP_NR(vaddr)] < 2);
	unlock_kmap();
	return (void*) vaddr;
}

EXPORT_SYMBOL(kmap_high);

#ifdef ARCH_NEEDS_KMAP_HIGH_GET
/**
 * kmap_high_get - pin a highmem page into memory
 * @page: &struct page to pin
 *
 * Returns the page's current virtual memory address, or NULL if no mapping
 * exists.  If and only if a non null address is returned then a
 * matching call to kunmap_high() is necessary.
 *
 * This can be called from any context.
 */
void *kmap_high_get(struct page *page)
{
	unsigned long vaddr, flags;

	lock_kmap_any(flags);
	vaddr = (unsigned long)page_address(page);
	if (vaddr) {
		BUG_ON(pkmap_count[PKMAP_NR(vaddr)] < 1);
		pkmap_count[PKMAP_NR(vaddr)]++;
	}
	unlock_kmap_any(flags);
	return (void*) vaddr;
}
#endif

/**
 * kunmap_high - unmap a highmem page into memory
 * @page: &struct page to unmap
 *
 * If ARCH_NEEDS_KMAP_HIGH_GET is not defined then this may be called
 * only from user context.
 */
void kunmap_high(struct page *page)
{
	unsigned long vaddr;
	unsigned long nr;
	unsigned long flags;
	int need_wakeup;
	unsigned int color = get_pkmap_color(page);
	wait_queue_head_t *pkmap_map_wait;

	lock_kmap_any(flags);
	vaddr = (unsigned long)page_address(page);
	BUG_ON(!vaddr);
	nr = PKMAP_NR(vaddr);

	/*
	 * A count must never go down to zero
	 * without a TLB flush!
	 */
	need_wakeup = 0;
	switch (--pkmap_count[nr]) {
	case 0:
		BUG();
	case 1:
		/*
		 * Avoid an unnecessary wake_up() function call.
		 * The common case is pkmap_count[] == 1, but
		 * no waiters.
		 * The tasks queued in the wait-queue are guarded
		 * by both the lock in the wait-queue-head and by
		 * the kmap_lock.  As the kmap_lock is held here,
		 * no need for the wait-queue-head's lock.  Simply
		 * test if the queue is empty.
		 */
		pkmap_map_wait = get_pkmap_wait_queue_head(color);
		need_wakeup = waitqueue_active(pkmap_map_wait);
	}
	unlock_kmap_any(flags);

	/* do wake-up, if needed, race-free outside of the spin lock */
	if (need_wakeup)
		wake_up(pkmap_map_wait);
}

EXPORT_SYMBOL(kunmap_high);
#endif

#if defined(HASHED_PAGE_VIRTUAL)

#define PA_HASH_ORDER	7

/*
 * Describes one page->virtual association
 */
struct page_address_map {
	struct page *page;
	void *virtual;
	struct list_head list;
};

static struct page_address_map page_address_maps[LAST_PKMAP];

/*
 * Hash table bucket
 */
static struct page_address_slot {
	struct list_head lh;			/* List of page_address_maps */
	spinlock_t lock;			/* Protect this bucket's list */
} ____cacheline_aligned_in_smp page_address_htable[1<<PA_HASH_ORDER];

static struct page_address_slot *page_slot(const struct page *page)
{
	return &page_address_htable[hash_ptr(page, PA_HASH_ORDER)];
}

/**
 * page_address - get the mapped virtual address of a page
 * @page: &struct page to get the virtual address of
 *
 * Returns the page's virtual address.
 */
void *page_address(const struct page *page)
{
	unsigned long flags;
	void *ret;
	struct page_address_slot *pas;

	if (!PageHighMem(page))
		return lowmem_page_address(page);

	pas = page_slot(page);
	ret = NULL;
	spin_lock_irqsave(&pas->lock, flags);
	if (!list_empty(&pas->lh)) {
		struct page_address_map *pam;

		list_for_each_entry(pam, &pas->lh, list) {
			if (pam->page == page) {
				ret = pam->virtual;
				goto done;
			}
		}
	}
done:
	spin_unlock_irqrestore(&pas->lock, flags);
	return ret;
}

EXPORT_SYMBOL(page_address);

/**
 * set_page_address - set a page's virtual address
 * @page: &struct page to set
 * @virtual: virtual address to use
 */
void set_page_address(struct page *page, void *virtual)
{
	unsigned long flags;
	struct page_address_slot *pas;
	struct page_address_map *pam;

	BUG_ON(!PageHighMem(page));

	pas = page_slot(page);
	if (virtual) {		/* Add */
		pam = &page_address_maps[PKMAP_NR((unsigned long)virtual)];
		pam->page = page;
		pam->virtual = virtual;

		spin_lock_irqsave(&pas->lock, flags);
		list_add_tail(&pam->list, &pas->lh);
		spin_unlock_irqrestore(&pas->lock, flags);
	} else {		/* Remove */
		spin_lock_irqsave(&pas->lock, flags);
		list_for_each_entry(pam, &pas->lh, list) {
			if (pam->page == page) {
				list_del(&pam->list);
				spin_unlock_irqrestore(&pas->lock, flags);
				goto done;
			}
		}
		spin_unlock_irqrestore(&pas->lock, flags);
	}
done:
	return;
}

void __init page_address_init(void)
{
	int i;

	for (i = 0; i < ARRAY_SIZE(page_address_htable); i++) {
		INIT_LIST_HEAD(&page_address_htable[i].lh);
		spin_lock_init(&page_address_htable[i].lock);
	}
}

#endif	/* defined(CONFIG_HIGHMEM) && !defined(WANT_PAGE_VIRTUAL) */
Commit	Line	Data
b2441318	1	// SPDX-License-Identifier: GPL-2.0
1da177e4 LT	2	/*
	3	* High memory handling common code and variables.
	4	*
	5	* (C) 1999 Andrea Arcangeli, SuSE GmbH, [email protected]
	6	* Gerhard Wichert, Siemens AG, [email protected]
	7	*
	8	*
	9	* Redesigned the x86 32-bit VM architecture to deal with
	10	* 64-bit physical space. With current x86 CPUs this
	11	* means up to 64 Gigabytes physical RAM.
	12	*
	13	* Rewrote high memory support to move the page cache into
	14	* high memory. Implemented permanent (schedulable) kmaps
	15	* based on Linus' idea.
	16	*
	17	* Copyright (C) 1999 Ingo Molnar <[email protected]>
	18	*/
	19
	20	#include <linux/mm.h>
b95f1b31	21	#include <linux/export.h>
1da177e4 LT	22	#include <linux/swap.h>
	23	#include <linux/bio.h>
	24	#include <linux/pagemap.h>
	25	#include <linux/mempool.h>
	26	#include <linux/blkdev.h>
	27	#include <linux/init.h>
	28	#include <linux/hash.h>
	29	#include <linux/highmem.h>
eac79005	30	#include <linux/kgdb.h>
1da177e4 LT	31	#include <asm/tlbflush.h>
1da177e4 LT	32
a8e23a29 PZ	33
	34	#if defined(CONFIG_HIGHMEM) \|\| defined(CONFIG_X86_32)
	35	DEFINE_PER_CPU(int, __kmap_atomic_idx);
	36	#endif
	37
1da177e4 LT	38	/*
	39	* Virtual_count is not a pure "count".
	40	* 0 means that it is not mapped, and has not been mapped
	41	* since a TLB flush - it is usable.
	42	* 1 means that there are no users, but it has been mapped
	43	* since the last TLB flush - so we can't use it.
	44	* n means that there are (n-1) current users of it.
	45	*/
	46	#ifdef CONFIG_HIGHMEM
260b2367	47
15de36a4 MF	48	/*
	49	* Architecture with aliasing data cache may define the following family of
	50	* helper functions in its asm/highmem.h to control cache color of virtual
	51	* addresses where physical memory pages are mapped by kmap.
	52	*/
	53	#ifndef get_pkmap_color
	54
	55	/*
	56	* Determine color of virtual address where the page should be mapped.
	57	*/
	58	static inline unsigned int get_pkmap_color(struct page *page)
	59	{
	60	return 0;
	61	}
	62	#define get_pkmap_color get_pkmap_color
	63
	64	/*
	65	* Get next index for mapping inside PKMAP region for page with given color.
	66	*/
	67	static inline unsigned int get_next_pkmap_nr(unsigned int color)
	68	{
	69	static unsigned int last_pkmap_nr;
	70
	71	last_pkmap_nr = (last_pkmap_nr + 1) & LAST_PKMAP_MASK;
	72	return last_pkmap_nr;
	73	}
	74
	75	/*
	76	* Determine if page index inside PKMAP region (pkmap_nr) of given color
	77	* has wrapped around PKMAP region end. When this happens an attempt to
	78	* flush all unused PKMAP slots is made.
	79	*/
	80	static inline int no_more_pkmaps(unsigned int pkmap_nr, unsigned int color)
	81	{
	82	return pkmap_nr == 0;
	83	}
	84
	85	/*
	86	* Get the number of PKMAP entries of the given color. If no free slot is
	87	* found after checking that many entries, kmap will sleep waiting for
	88	* someone to call kunmap and free PKMAP slot.
	89	*/
	90	static inline int get_pkmap_entries_count(unsigned int color)
	91	{
	92	return LAST_PKMAP;
	93	}
	94
	95	/*
	96	* Get head of a wait queue for PKMAP entries of the given color.
	97	* Wait queues for different mapping colors should be independent to avoid
	98	* unnecessary wakeups caused by freeing of slots of other colors.
	99	*/
	100	static inline wait_queue_head_t *get_pkmap_wait_queue_head(unsigned int color)
	101	{
	102	static DECLARE_WAIT_QUEUE_HEAD(pkmap_map_wait);
	103
	104	return &pkmap_map_wait;
	105	}
	106	#endif
	107
c1f60a5a	108	unsigned long totalhigh_pages __read_mostly;
db7a94d6	109	EXPORT_SYMBOL(totalhigh_pages);
c1f60a5a	110
3e4d3af5	111
3e4d3af5 PZ	112	EXPORT_PER_CPU_SYMBOL(__kmap_atomic_idx);
3e4d3af5 PZ	113
c1f60a5a CL	114	unsigned int nr_free_highpages (void)
c1f60a5a CL	115	{
33499bfe	116	struct zone *zone;
c1f60a5a CL	117	unsigned int pages = 0;
c1f60a5a CL	118
33499bfe JK	119	for_each_populated_zone(zone) {
	120	if (is_highmem(zone))
	121	pages += zone_page_state(zone, NR_FREE_PAGES);
2a1e274a	122	}
c1f60a5a CL	123
	124	return pages;
	125	}
	126
1da177e4	127	static int pkmap_count[LAST_PKMAP];
1da177e4 LT	128	static __cacheline_aligned_in_smp DEFINE_SPINLOCK(kmap_lock);
	129
	130	pte_t * pkmap_page_table;
	131
3297e760 NP	132	/*
	133	* Most architectures have no use for kmap_high_get(), so let's abstract
	134	* the disabling of IRQ out of the locking in that case to save on a
	135	* potential useless overhead.
	136	*/
	137	#ifdef ARCH_NEEDS_KMAP_HIGH_GET
	138	#define lock_kmap() spin_lock_irq(&kmap_lock)
	139	#define unlock_kmap() spin_unlock_irq(&kmap_lock)
	140	#define lock_kmap_any(flags) spin_lock_irqsave(&kmap_lock, flags)
	141	#define unlock_kmap_any(flags) spin_unlock_irqrestore(&kmap_lock, flags)
	142	#else
	143	#define lock_kmap() spin_lock(&kmap_lock)
	144	#define unlock_kmap() spin_unlock(&kmap_lock)
	145	#define lock_kmap_any(flags) \
	146	do { spin_lock(&kmap_lock); (void)(flags); } while (0)
	147	#define unlock_kmap_any(flags) \
	148	do { spin_unlock(&kmap_lock); (void)(flags); } while (0)
	149	#endif
	150
5a178119 MG	151	struct page kmap_to_page(void vaddr)
	152	{
	153	unsigned long addr = (unsigned long)vaddr;
	154
498c2280	155	if (addr >= PKMAP_ADDR(0) && addr < PKMAP_ADDR(LAST_PKMAP)) {
4de22c05	156	int i = PKMAP_NR(addr);
5a178119 MG	157	return pte_page(pkmap_page_table[i]);
	158	}
	159
	160	return virt_to_page(addr);
	161	}
f0263d2d	162	EXPORT_SYMBOL(kmap_to_page);
5a178119	163
1da177e4 LT	164	static void flush_all_zero_pkmaps(void)
	165	{
	166	int i;
5843d9a4	167	int need_flush = 0;
1da177e4 LT	168
	169	flush_cache_kmaps();
	170
	171	for (i = 0; i < LAST_PKMAP; i++) {
	172	struct page *page;
	173
	174	/*
	175	* zero means we don't have anything to do,
	176	* >1 means that it is still in use. Only
	177	* a count of 1 means that it is free but
	178	* needs to be unmapped
	179	*/
	180	if (pkmap_count[i] != 1)
	181	continue;
	182	pkmap_count[i] = 0;
	183
	184	/* sanity check */
75babcac	185	BUG_ON(pte_none(pkmap_page_table[i]));
1da177e4 LT	186
	187	/*
	188	* Don't need an atomic fetch-and-clear op here;
	189	* no-one has the page mapped, and cannot get at
	190	* its virtual address (and hence PTE) without first
	191	* getting the kmap_lock (which is held here).
	192	* So no dangers, even with speculative execution.
	193	*/
	194	page = pte_page(pkmap_page_table[i]);
eb2db439	195	pte_clear(&init_mm, PKMAP_ADDR(i), &pkmap_page_table[i]);
1da177e4 LT	196
1da177e4 LT	197	set_page_address(page, NULL);
5843d9a4	198	need_flush = 1;
1da177e4	199	}
5843d9a4 NP	200	if (need_flush)
5843d9a4 NP	201	flush_tlb_kernel_range(PKMAP_ADDR(0), PKMAP_ADDR(LAST_PKMAP));
1da177e4 LT	202	}
1da177e4 LT	203
77f6078a RD	204	/**
	205	* kmap_flush_unused - flush all unused kmap mappings in order to remove stray mappings
	206	*/
ce6234b5 JF	207	void kmap_flush_unused(void)
ce6234b5 JF	208	{
3297e760	209	lock_kmap();
ce6234b5	210	flush_all_zero_pkmaps();
3297e760	211	unlock_kmap();
ce6234b5 JF	212	}
ce6234b5 JF	213
1da177e4 LT	214	static inline unsigned long map_new_virtual(struct page *page)
	215	{
	216	unsigned long vaddr;
	217	int count;
15de36a4 MF	218	unsigned int last_pkmap_nr;
15de36a4 MF	219	unsigned int color = get_pkmap_color(page);
1da177e4 LT	220
1da177e4 LT	221	start:
15de36a4	222	count = get_pkmap_entries_count(color);
1da177e4 LT	223	/* Find an empty entry */
1da177e4 LT	224	for (;;) {
15de36a4 MF	225	last_pkmap_nr = get_next_pkmap_nr(color);
15de36a4 MF	226	if (no_more_pkmaps(last_pkmap_nr, color)) {
1da177e4	227	flush_all_zero_pkmaps();
15de36a4	228	count = get_pkmap_entries_count(color);
1da177e4 LT	229	}
	230	if (!pkmap_count[last_pkmap_nr])
	231	break; /* Found a usable entry */
	232	if (--count)
	233	continue;
	234
	235	/*
	236	* Sleep for somebody else to unmap their entries
	237	*/
	238	{
	239	DECLARE_WAITQUEUE(wait, current);
15de36a4 MF	240	wait_queue_head_t *pkmap_map_wait =
15de36a4 MF	241	get_pkmap_wait_queue_head(color);
1da177e4 LT	242
1da177e4 LT	243	__set_current_state(TASK_UNINTERRUPTIBLE);
15de36a4	244	add_wait_queue(pkmap_map_wait, &wait);
3297e760	245	unlock_kmap();
1da177e4	246	schedule();
15de36a4	247	remove_wait_queue(pkmap_map_wait, &wait);
3297e760	248	lock_kmap();
1da177e4 LT	249
	250	/* Somebody else might have mapped it while we slept */
	251	if (page_address(page))
	252	return (unsigned long)page_address(page);
	253
	254	/* Re-start */
	255	goto start;
	256	}
	257	}
	258	vaddr = PKMAP_ADDR(last_pkmap_nr);
	259	set_pte_at(&init_mm, vaddr,
	260	&(pkmap_page_table[last_pkmap_nr]), mk_pte(page, kmap_prot));
	261
	262	pkmap_count[last_pkmap_nr] = 1;
	263	set_page_address(page, (void *)vaddr);
	264
	265	return vaddr;
	266	}
	267
77f6078a RD	268	/**
	269	* kmap_high - map a highmem page into memory
	270	* @page: &struct page to map
	271	*
	272	* Returns the page's virtual memory address.
	273	*
	274	* We cannot call this from interrupts, as it may block.
	275	*/
920c7a5d	276	void kmap_high(struct page page)
1da177e4 LT	277	{
	278	unsigned long vaddr;
	279
	280	/*
	281	* For highmem pages, we can't trust "virtual" until
	282	* after we have the lock.
1da177e4	283	*/
3297e760	284	lock_kmap();
1da177e4 LT	285	vaddr = (unsigned long)page_address(page);
	286	if (!vaddr)
	287	vaddr = map_new_virtual(page);
	288	pkmap_count[PKMAP_NR(vaddr)]++;
75babcac	289	BUG_ON(pkmap_count[PKMAP_NR(vaddr)] < 2);
3297e760	290	unlock_kmap();
1da177e4 LT	291	return (void*) vaddr;
	292	}
	293
	294	EXPORT_SYMBOL(kmap_high);
	295
3297e760 NP	296	#ifdef ARCH_NEEDS_KMAP_HIGH_GET
	297	/**
	298	* kmap_high_get - pin a highmem page into memory
	299	* @page: &struct page to pin
	300	*
	301	* Returns the page's current virtual memory address, or NULL if no mapping
5e39df56	302	* exists. If and only if a non null address is returned then a
3297e760 NP	303	* matching call to kunmap_high() is necessary.
	304	*
	305	* This can be called from any context.
	306	*/
	307	void kmap_high_get(struct page page)
	308	{
	309	unsigned long vaddr, flags;
	310
	311	lock_kmap_any(flags);
	312	vaddr = (unsigned long)page_address(page);
	313	if (vaddr) {
	314	BUG_ON(pkmap_count[PKMAP_NR(vaddr)] < 1);
	315	pkmap_count[PKMAP_NR(vaddr)]++;
	316	}
	317	unlock_kmap_any(flags);
	318	return (void*) vaddr;
	319	}
	320	#endif
	321
77f6078a	322	/**
4e9dc5df	323	* kunmap_high - unmap a highmem page into memory
77f6078a	324	* @page: &struct page to unmap
3297e760 NP	325	*
	326	* If ARCH_NEEDS_KMAP_HIGH_GET is not defined then this may be called
	327	* only from user context.
77f6078a	328	*/
920c7a5d	329	void kunmap_high(struct page *page)
1da177e4 LT	330	{
	331	unsigned long vaddr;
	332	unsigned long nr;
3297e760	333	unsigned long flags;
1da177e4	334	int need_wakeup;
15de36a4 MF	335	unsigned int color = get_pkmap_color(page);
15de36a4 MF	336	wait_queue_head_t *pkmap_map_wait;
1da177e4	337
3297e760	338	lock_kmap_any(flags);
1da177e4	339	vaddr = (unsigned long)page_address(page);
75babcac	340	BUG_ON(!vaddr);
1da177e4 LT	341	nr = PKMAP_NR(vaddr);
	342
	343	/*
	344	* A count must never go down to zero
	345	* without a TLB flush!
	346	*/
	347	need_wakeup = 0;
	348	switch (--pkmap_count[nr]) {
	349	case 0:
	350	BUG();
	351	case 1:
	352	/*
	353	* Avoid an unnecessary wake_up() function call.
	354	* The common case is pkmap_count[] == 1, but
	355	* no waiters.
	356	* The tasks queued in the wait-queue are guarded
	357	* by both the lock in the wait-queue-head and by
	358	* the kmap_lock. As the kmap_lock is held here,
	359	* no need for the wait-queue-head's lock. Simply
	360	* test if the queue is empty.
	361	*/
15de36a4 MF	362	pkmap_map_wait = get_pkmap_wait_queue_head(color);
15de36a4 MF	363	need_wakeup = waitqueue_active(pkmap_map_wait);
1da177e4	364	}
3297e760	365	unlock_kmap_any(flags);
1da177e4 LT	366
	367	/* do wake-up, if needed, race-free outside of the spin lock */
	368	if (need_wakeup)
15de36a4	369	wake_up(pkmap_map_wait);
1da177e4 LT	370	}
	371
	372	EXPORT_SYMBOL(kunmap_high);
1da177e4 LT	373	#endif
1da177e4 LT	374
1da177e4 LT	375	#if defined(HASHED_PAGE_VIRTUAL)
	376
	377	#define PA_HASH_ORDER 7
	378
	379	/*
	380	* Describes one page->virtual association
	381	*/
	382	struct page_address_map {
	383	struct page *page;
	384	void *virtual;
	385	struct list_head list;
	386	};
	387
a354e2c8	388	static struct page_address_map page_address_maps[LAST_PKMAP];
1da177e4 LT	389
	390	/*
	391	* Hash table bucket
	392	*/
	393	static struct page_address_slot {
	394	struct list_head lh; /* List of page_address_maps */
	395	spinlock_t lock; /* Protect this bucket's list */
	396	} ____cacheline_aligned_in_smp page_address_htable[1<<PA_HASH_ORDER];
	397
f9918794	398	static struct page_address_slot page_slot(const struct page page)
1da177e4 LT	399	{
	400	return &page_address_htable[hash_ptr(page, PA_HASH_ORDER)];
	401	}
	402
77f6078a RD	403	/**
	404	* page_address - get the mapped virtual address of a page
	405	* @page: &struct page to get the virtual address of
	406	*
	407	* Returns the page's virtual address.
	408	*/
f9918794	409	void page_address(const struct page page)
1da177e4 LT	410	{
	411	unsigned long flags;
	412	void *ret;
	413	struct page_address_slot *pas;
	414
	415	if (!PageHighMem(page))
	416	return lowmem_page_address(page);
	417
	418	pas = page_slot(page);
	419	ret = NULL;
	420	spin_lock_irqsave(&pas->lock, flags);
	421	if (!list_empty(&pas->lh)) {
	422	struct page_address_map *pam;
	423
	424	list_for_each_entry(pam, &pas->lh, list) {
	425	if (pam->page == page) {
	426	ret = pam->virtual;
	427	goto done;
	428	}
	429	}
	430	}
	431	done:
	432	spin_unlock_irqrestore(&pas->lock, flags);
	433	return ret;
	434	}
	435
	436	EXPORT_SYMBOL(page_address);
	437
77f6078a RD	438	/**
	439	* set_page_address - set a page's virtual address
	440	* @page: &struct page to set
	441	* @virtual: virtual address to use
	442	*/
1da177e4 LT	443	void set_page_address(struct page page, void virtual)
	444	{
	445	unsigned long flags;
	446	struct page_address_slot *pas;
	447	struct page_address_map *pam;
	448
	449	BUG_ON(!PageHighMem(page));
	450
	451	pas = page_slot(page);
	452	if (virtual) { /* Add */
a354e2c8	453	pam = &page_address_maps[PKMAP_NR((unsigned long)virtual)];
1da177e4 LT	454	pam->page = page;
	455	pam->virtual = virtual;
	456
	457	spin_lock_irqsave(&pas->lock, flags);
	458	list_add_tail(&pam->list, &pas->lh);
	459	spin_unlock_irqrestore(&pas->lock, flags);
	460	} else { /* Remove */
	461	spin_lock_irqsave(&pas->lock, flags);
	462	list_for_each_entry(pam, &pas->lh, list) {
	463	if (pam->page == page) {
	464	list_del(&pam->list);
	465	spin_unlock_irqrestore(&pas->lock, flags);
1da177e4 LT	466	goto done;
	467	}
	468	}
	469	spin_unlock_irqrestore(&pas->lock, flags);
	470	}
	471	done:
	472	return;
	473	}
	474
1da177e4 LT	475	void __init page_address_init(void)
	476	{
	477	int i;
	478
1da177e4 LT	479	for (i = 0; i < ARRAY_SIZE(page_address_htable); i++) {
	480	INIT_LIST_HEAD(&page_address_htable[i].lh);
	481	spin_lock_init(&page_address_htable[i].lock);
	482	}
1da177e4 LT	483	}
	484
	485	#endif /* defined(CONFIG_HIGHMEM) && !defined(WANT_PAGE_VIRTUAL) */