[mirror_ubuntu-jammy-kernel.git] / mm / highmem.c

// SPDX-License-Identifier: GPL-2.0
/*
 * High memory handling common code and variables.
 *
 * (C) 1999 Andrea Arcangeli, SuSE GmbH, andrea@suse.de
 *          Gerhard Wichert, Siemens AG, Gerhard.Wichert@pdb.siemens.de
 *
 *
 * 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 <mingo@redhat.com>
 */

#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>
#include <linux/vmalloc.h>

/*
 * 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

atomic_long_t _totalhigh_pages __read_mostly;
EXPORT_SYMBOL(_totalhigh_pages);

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 /* CONFIG_HIGHMEM */

#ifdef CONFIG_KMAP_LOCAL

#include <asm/kmap_size.h>

static DEFINE_PER_CPU(int, __kmap_local_idx);

/*
 * With DEBUG_HIGHMEM the stack depth is doubled and every second
 * slot is unused which acts as a guard page
 */
#ifdef CONFIG_DEBUG_HIGHMEM
# define KM_INCR	2
#else
# define KM_INCR	1
#endif

static inline int kmap_local_idx_push(void)
{
	int idx = __this_cpu_add_return(__kmap_local_idx, KM_INCR) - 1;

	WARN_ON_ONCE(in_irq() && !irqs_disabled());
	BUG_ON(idx >= KM_MAX_IDX);
	return idx;
}

static inline int kmap_local_idx(void)
{
	return __this_cpu_read(__kmap_local_idx) - 1;
}

static inline void kmap_local_idx_pop(void)
{
	int idx = __this_cpu_sub_return(__kmap_local_idx, KM_INCR);

	BUG_ON(idx < 0);
}

#ifndef arch_kmap_local_post_map
# define arch_kmap_local_post_map(vaddr, pteval)	do { } while (0)
#endif

#ifndef arch_kmap_local_pre_unmap
# define arch_kmap_local_pre_unmap(vaddr)		do { } while (0)
#endif

#ifndef arch_kmap_local_post_unmap
# define arch_kmap_local_post_unmap(vaddr)		do { } while (0)
#endif

#ifndef arch_kmap_local_map_idx
#define arch_kmap_local_map_idx(idx, pfn)	kmap_local_calc_idx(idx)
#endif

#ifndef arch_kmap_local_unmap_idx
#define arch_kmap_local_unmap_idx(idx, vaddr)	kmap_local_calc_idx(idx)
#endif

#ifndef arch_kmap_local_high_get
static inline void *arch_kmap_local_high_get(struct page *page)
{
	return NULL;
}
#endif

/* Unmap a local mapping which was obtained by kmap_high_get() */
static inline void kmap_high_unmap_local(unsigned long vaddr)
{
#ifdef ARCH_NEEDS_KMAP_HIGH_GET
	if (vaddr >= PKMAP_ADDR(0) && vaddr < PKMAP_ADDR(LAST_PKMAP))
		kunmap_high(pte_page(pkmap_page_table[PKMAP_NR(vaddr)]));
#endif
}

static inline int kmap_local_calc_idx(int idx)
{
	return idx + KM_MAX_IDX * smp_processor_id();
}

static pte_t *__kmap_pte;

static pte_t *kmap_get_pte(void)
{
	if (!__kmap_pte)
		__kmap_pte = virt_to_kpte(__fix_to_virt(FIX_KMAP_BEGIN));
	return __kmap_pte;
}

void *__kmap_local_pfn_prot(unsigned long pfn, pgprot_t prot)
{
	pte_t pteval, *kmap_pte = kmap_get_pte();
	unsigned long vaddr;
	int idx;

	preempt_disable();
	idx = arch_kmap_local_map_idx(kmap_local_idx_push(), pfn);
	vaddr = __fix_to_virt(FIX_KMAP_BEGIN + idx);
	BUG_ON(!pte_none(*(kmap_pte - idx)));
	pteval = pfn_pte(pfn, prot);
	set_pte_at(&init_mm, vaddr, kmap_pte - idx, pteval);
	arch_kmap_local_post_map(vaddr, pteval);
	preempt_enable();

	return (void *)vaddr;
}
EXPORT_SYMBOL_GPL(__kmap_local_pfn_prot);

void *__kmap_local_page_prot(struct page *page, pgprot_t prot)
{
	void *kmap;

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

	/* Try kmap_high_get() if architecture has it enabled */
	kmap = arch_kmap_local_high_get(page);
	if (kmap)
		return kmap;

	return __kmap_local_pfn_prot(page_to_pfn(page), prot);
}
EXPORT_SYMBOL(__kmap_local_page_prot);

void kunmap_local_indexed(void *vaddr)
{
	unsigned long addr = (unsigned long) vaddr & PAGE_MASK;
	pte_t *kmap_pte = kmap_get_pte();
	int idx;

	if (addr < __fix_to_virt(FIX_KMAP_END) ||
	    addr > __fix_to_virt(FIX_KMAP_BEGIN)) {
		WARN_ON_ONCE(addr < PAGE_OFFSET);

		/* Handle mappings which were obtained by kmap_high_get() */
		kmap_high_unmap_local(addr);
		return;
	}

	preempt_disable();
	idx = arch_kmap_local_unmap_idx(kmap_local_idx(), addr);
	WARN_ON_ONCE(addr != __fix_to_virt(FIX_KMAP_BEGIN + idx));

	arch_kmap_local_pre_unmap(addr);
	pte_clear(&init_mm, addr, kmap_pte - idx);
	arch_kmap_local_post_unmap(addr);
	kmap_local_idx_pop();
	preempt_enable();
}
EXPORT_SYMBOL(kunmap_local_indexed);
#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(HASHED_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, andrea@suse.de
	6	* Gerhard Wichert, Siemens AG, Gerhard.Wichert@pdb.siemens.de
	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 <mingo@redhat.com>
	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	31	#include <asm/tlbflush.h>
186525bd	32	#include <linux/vmalloc.h>
a8e23a29	33
1da177e4 LT	34	/*
	35	* Virtual_count is not a pure "count".
	36	* 0 means that it is not mapped, and has not been mapped
	37	* since a TLB flush - it is usable.
	38	* 1 means that there are no users, but it has been mapped
	39	* since the last TLB flush - so we can't use it.
	40	* n means that there are (n-1) current users of it.
	41	*/
	42	#ifdef CONFIG_HIGHMEM
260b2367	43
15de36a4 MF	44	/*
	45	* Architecture with aliasing data cache may define the following family of
	46	* helper functions in its asm/highmem.h to control cache color of virtual
	47	* addresses where physical memory pages are mapped by kmap.
	48	*/
	49	#ifndef get_pkmap_color
	50
	51	/*
	52	* Determine color of virtual address where the page should be mapped.
	53	*/
	54	static inline unsigned int get_pkmap_color(struct page *page)
	55	{
	56	return 0;
	57	}
	58	#define get_pkmap_color get_pkmap_color
	59
	60	/*
	61	* Get next index for mapping inside PKMAP region for page with given color.
	62	*/
	63	static inline unsigned int get_next_pkmap_nr(unsigned int color)
	64	{
	65	static unsigned int last_pkmap_nr;
	66
	67	last_pkmap_nr = (last_pkmap_nr + 1) & LAST_PKMAP_MASK;
	68	return last_pkmap_nr;
	69	}
	70
	71	/*
	72	* Determine if page index inside PKMAP region (pkmap_nr) of given color
	73	* has wrapped around PKMAP region end. When this happens an attempt to
	74	* flush all unused PKMAP slots is made.
	75	*/
	76	static inline int no_more_pkmaps(unsigned int pkmap_nr, unsigned int color)
	77	{
	78	return pkmap_nr == 0;
	79	}
	80
	81	/*
	82	* Get the number of PKMAP entries of the given color. If no free slot is
	83	* found after checking that many entries, kmap will sleep waiting for
	84	* someone to call kunmap and free PKMAP slot.
	85	*/
	86	static inline int get_pkmap_entries_count(unsigned int color)
	87	{
	88	return LAST_PKMAP;
	89	}
	90
	91	/*
	92	* Get head of a wait queue for PKMAP entries of the given color.
	93	* Wait queues for different mapping colors should be independent to avoid
	94	* unnecessary wakeups caused by freeing of slots of other colors.
	95	*/
	96	static inline wait_queue_head_t *get_pkmap_wait_queue_head(unsigned int color)
	97	{
	98	static DECLARE_WAIT_QUEUE_HEAD(pkmap_map_wait);
	99
	100	return &pkmap_map_wait;
	101	}
	102	#endif
	103
ca79b0c2 AK	104	atomic_long_t _totalhigh_pages __read_mostly;
ca79b0c2 AK	105	EXPORT_SYMBOL(_totalhigh_pages);
3e4d3af5	106
c1f60a5a CL	107	unsigned int nr_free_highpages (void)
c1f60a5a CL	108	{
33499bfe	109	struct zone *zone;
c1f60a5a CL	110	unsigned int pages = 0;
c1f60a5a CL	111
33499bfe JK	112	for_each_populated_zone(zone) {
	113	if (is_highmem(zone))
	114	pages += zone_page_state(zone, NR_FREE_PAGES);
2a1e274a	115	}
c1f60a5a CL	116
	117	return pages;
	118	}
	119
1da177e4	120	static int pkmap_count[LAST_PKMAP];
1da177e4 LT	121	static __cacheline_aligned_in_smp DEFINE_SPINLOCK(kmap_lock);
	122
	123	pte_t * pkmap_page_table;
	124
3297e760 NP	125	/*
	126	* Most architectures have no use for kmap_high_get(), so let's abstract
	127	* the disabling of IRQ out of the locking in that case to save on a
	128	* potential useless overhead.
	129	*/
	130	#ifdef ARCH_NEEDS_KMAP_HIGH_GET
	131	#define lock_kmap() spin_lock_irq(&kmap_lock)
	132	#define unlock_kmap() spin_unlock_irq(&kmap_lock)
	133	#define lock_kmap_any(flags) spin_lock_irqsave(&kmap_lock, flags)
	134	#define unlock_kmap_any(flags) spin_unlock_irqrestore(&kmap_lock, flags)
	135	#else
	136	#define lock_kmap() spin_lock(&kmap_lock)
	137	#define unlock_kmap() spin_unlock(&kmap_lock)
	138	#define lock_kmap_any(flags) \
	139	do { spin_lock(&kmap_lock); (void)(flags); } while (0)
	140	#define unlock_kmap_any(flags) \
	141	do { spin_unlock(&kmap_lock); (void)(flags); } while (0)
	142	#endif
	143
5a178119 MG	144	struct page kmap_to_page(void vaddr)
	145	{
	146	unsigned long addr = (unsigned long)vaddr;
	147
498c2280	148	if (addr >= PKMAP_ADDR(0) && addr < PKMAP_ADDR(LAST_PKMAP)) {
4de22c05	149	int i = PKMAP_NR(addr);
5a178119 MG	150	return pte_page(pkmap_page_table[i]);
	151	}
	152
	153	return virt_to_page(addr);
	154	}
f0263d2d	155	EXPORT_SYMBOL(kmap_to_page);
5a178119	156
1da177e4 LT	157	static void flush_all_zero_pkmaps(void)
	158	{
	159	int i;
5843d9a4	160	int need_flush = 0;
1da177e4 LT	161
	162	flush_cache_kmaps();
	163
	164	for (i = 0; i < LAST_PKMAP; i++) {
	165	struct page *page;
	166
	167	/*
	168	* zero means we don't have anything to do,
	169	* >1 means that it is still in use. Only
	170	* a count of 1 means that it is free but
	171	* needs to be unmapped
	172	*/
	173	if (pkmap_count[i] != 1)
	174	continue;
	175	pkmap_count[i] = 0;
	176
	177	/* sanity check */
75babcac	178	BUG_ON(pte_none(pkmap_page_table[i]));
1da177e4 LT	179
	180	/*
	181	* Don't need an atomic fetch-and-clear op here;
	182	* no-one has the page mapped, and cannot get at
	183	* its virtual address (and hence PTE) without first
	184	* getting the kmap_lock (which is held here).
	185	* So no dangers, even with speculative execution.
	186	*/
	187	page = pte_page(pkmap_page_table[i]);
eb2db439	188	pte_clear(&init_mm, PKMAP_ADDR(i), &pkmap_page_table[i]);
1da177e4 LT	189
1da177e4 LT	190	set_page_address(page, NULL);
5843d9a4	191	need_flush = 1;
1da177e4	192	}
5843d9a4 NP	193	if (need_flush)
5843d9a4 NP	194	flush_tlb_kernel_range(PKMAP_ADDR(0), PKMAP_ADDR(LAST_PKMAP));
1da177e4 LT	195	}
1da177e4 LT	196
77f6078a RD	197	/**
	198	* kmap_flush_unused - flush all unused kmap mappings in order to remove stray mappings
	199	*/
ce6234b5 JF	200	void kmap_flush_unused(void)
ce6234b5 JF	201	{
3297e760	202	lock_kmap();
ce6234b5	203	flush_all_zero_pkmaps();
3297e760	204	unlock_kmap();
ce6234b5 JF	205	}
ce6234b5 JF	206
1da177e4 LT	207	static inline unsigned long map_new_virtual(struct page *page)
	208	{
	209	unsigned long vaddr;
	210	int count;
15de36a4 MF	211	unsigned int last_pkmap_nr;
15de36a4 MF	212	unsigned int color = get_pkmap_color(page);
1da177e4 LT	213
1da177e4 LT	214	start:
15de36a4	215	count = get_pkmap_entries_count(color);
1da177e4 LT	216	/* Find an empty entry */
1da177e4 LT	217	for (;;) {
15de36a4 MF	218	last_pkmap_nr = get_next_pkmap_nr(color);
15de36a4 MF	219	if (no_more_pkmaps(last_pkmap_nr, color)) {
1da177e4	220	flush_all_zero_pkmaps();
15de36a4	221	count = get_pkmap_entries_count(color);
1da177e4 LT	222	}
	223	if (!pkmap_count[last_pkmap_nr])
	224	break; /* Found a usable entry */
	225	if (--count)
	226	continue;
	227
	228	/*
	229	* Sleep for somebody else to unmap their entries
	230	*/
	231	{
	232	DECLARE_WAITQUEUE(wait, current);
15de36a4 MF	233	wait_queue_head_t *pkmap_map_wait =
15de36a4 MF	234	get_pkmap_wait_queue_head(color);
1da177e4 LT	235
1da177e4 LT	236	__set_current_state(TASK_UNINTERRUPTIBLE);
15de36a4	237	add_wait_queue(pkmap_map_wait, &wait);
3297e760	238	unlock_kmap();
1da177e4	239	schedule();
15de36a4	240	remove_wait_queue(pkmap_map_wait, &wait);
3297e760	241	lock_kmap();
1da177e4 LT	242
	243	/* Somebody else might have mapped it while we slept */
	244	if (page_address(page))
	245	return (unsigned long)page_address(page);
	246
	247	/* Re-start */
	248	goto start;
	249	}
	250	}
	251	vaddr = PKMAP_ADDR(last_pkmap_nr);
	252	set_pte_at(&init_mm, vaddr,
	253	&(pkmap_page_table[last_pkmap_nr]), mk_pte(page, kmap_prot));
	254
	255	pkmap_count[last_pkmap_nr] = 1;
	256	set_page_address(page, (void *)vaddr);
	257
	258	return vaddr;
	259	}
	260
77f6078a RD	261	/**
	262	* kmap_high - map a highmem page into memory
	263	* @page: &struct page to map
	264	*
	265	* Returns the page's virtual memory address.
	266	*
	267	* We cannot call this from interrupts, as it may block.
	268	*/
920c7a5d	269	void kmap_high(struct page page)
1da177e4 LT	270	{
	271	unsigned long vaddr;
	272
	273	/*
	274	* For highmem pages, we can't trust "virtual" until
	275	* after we have the lock.
1da177e4	276	*/
3297e760	277	lock_kmap();
1da177e4 LT	278	vaddr = (unsigned long)page_address(page);
	279	if (!vaddr)
	280	vaddr = map_new_virtual(page);
	281	pkmap_count[PKMAP_NR(vaddr)]++;
75babcac	282	BUG_ON(pkmap_count[PKMAP_NR(vaddr)] < 2);
3297e760	283	unlock_kmap();
1da177e4 LT	284	return (void*) vaddr;
	285	}
	286
	287	EXPORT_SYMBOL(kmap_high);
	288
3297e760 NP	289	#ifdef ARCH_NEEDS_KMAP_HIGH_GET
	290	/**
	291	* kmap_high_get - pin a highmem page into memory
	292	* @page: &struct page to pin
	293	*
	294	* Returns the page's current virtual memory address, or NULL if no mapping
5e39df56	295	* exists. If and only if a non null address is returned then a
3297e760 NP	296	* matching call to kunmap_high() is necessary.
	297	*
	298	* This can be called from any context.
	299	*/
	300	void kmap_high_get(struct page page)
	301	{
	302	unsigned long vaddr, flags;
	303
	304	lock_kmap_any(flags);
	305	vaddr = (unsigned long)page_address(page);
	306	if (vaddr) {
	307	BUG_ON(pkmap_count[PKMAP_NR(vaddr)] < 1);
	308	pkmap_count[PKMAP_NR(vaddr)]++;
	309	}
	310	unlock_kmap_any(flags);
	311	return (void*) vaddr;
	312	}
	313	#endif
	314
77f6078a	315	/**
4e9dc5df	316	* kunmap_high - unmap a highmem page into memory
77f6078a	317	* @page: &struct page to unmap
3297e760 NP	318	*
	319	* If ARCH_NEEDS_KMAP_HIGH_GET is not defined then this may be called
	320	* only from user context.
77f6078a	321	*/
920c7a5d	322	void kunmap_high(struct page *page)
1da177e4 LT	323	{
	324	unsigned long vaddr;
	325	unsigned long nr;
3297e760	326	unsigned long flags;
1da177e4	327	int need_wakeup;
15de36a4 MF	328	unsigned int color = get_pkmap_color(page);
15de36a4 MF	329	wait_queue_head_t *pkmap_map_wait;
1da177e4	330
3297e760	331	lock_kmap_any(flags);
1da177e4	332	vaddr = (unsigned long)page_address(page);
75babcac	333	BUG_ON(!vaddr);
1da177e4 LT	334	nr = PKMAP_NR(vaddr);
	335
	336	/*
	337	* A count must never go down to zero
	338	* without a TLB flush!
	339	*/
	340	need_wakeup = 0;
	341	switch (--pkmap_count[nr]) {
	342	case 0:
	343	BUG();
	344	case 1:
	345	/*
	346	* Avoid an unnecessary wake_up() function call.
	347	* The common case is pkmap_count[] == 1, but
	348	* no waiters.
	349	* The tasks queued in the wait-queue are guarded
	350	* by both the lock in the wait-queue-head and by
	351	* the kmap_lock. As the kmap_lock is held here,
	352	* no need for the wait-queue-head's lock. Simply
	353	* test if the queue is empty.
	354	*/
15de36a4 MF	355	pkmap_map_wait = get_pkmap_wait_queue_head(color);
15de36a4 MF	356	need_wakeup = waitqueue_active(pkmap_map_wait);
1da177e4	357	}
3297e760	358	unlock_kmap_any(flags);
1da177e4 LT	359
	360	/* do wake-up, if needed, race-free outside of the spin lock */
	361	if (need_wakeup)
15de36a4	362	wake_up(pkmap_map_wait);
1da177e4	363	}
1da177e4	364	EXPORT_SYMBOL(kunmap_high);
298fa1ad TG	365	#endif /* CONFIG_HIGHMEM */
	366
	367	#ifdef CONFIG_KMAP_LOCAL
	368
	369	#include <asm/kmap_size.h>
	370
	371	static DEFINE_PER_CPU(int, __kmap_local_idx);
	372
389755c2 TG	373	/*
	374	* With DEBUG_HIGHMEM the stack depth is doubled and every second
	375	* slot is unused which acts as a guard page
	376	*/
	377	#ifdef CONFIG_DEBUG_HIGHMEM
	378	# define KM_INCR 2
	379	#else
	380	# define KM_INCR 1
	381	#endif
	382
298fa1ad TG	383	static inline int kmap_local_idx_push(void)
298fa1ad TG	384	{
389755c2	385	int idx = __this_cpu_add_return(__kmap_local_idx, KM_INCR) - 1;
298fa1ad TG	386
	387	WARN_ON_ONCE(in_irq() && !irqs_disabled());
	388	BUG_ON(idx >= KM_MAX_IDX);
	389	return idx;
	390	}
	391
	392	static inline int kmap_local_idx(void)
	393	{
	394	return __this_cpu_read(__kmap_local_idx) - 1;
	395	}
	396
	397	static inline void kmap_local_idx_pop(void)
	398	{
389755c2	399	int idx = __this_cpu_sub_return(__kmap_local_idx, KM_INCR);
298fa1ad TG	400
	401	BUG_ON(idx < 0);
	402	}
	403
	404	#ifndef arch_kmap_local_post_map
	405	# define arch_kmap_local_post_map(vaddr, pteval) do { } while (0)
	406	#endif
3c1016b5	407
298fa1ad TG	408	#ifndef arch_kmap_local_pre_unmap
	409	# define arch_kmap_local_pre_unmap(vaddr) do { } while (0)
	410	#endif
	411
	412	#ifndef arch_kmap_local_post_unmap
	413	# define arch_kmap_local_post_unmap(vaddr) do { } while (0)
	414	#endif
	415
	416	#ifndef arch_kmap_local_map_idx
	417	#define arch_kmap_local_map_idx(idx, pfn) kmap_local_calc_idx(idx)
	418	#endif
	419
	420	#ifndef arch_kmap_local_unmap_idx
	421	#define arch_kmap_local_unmap_idx(idx, vaddr) kmap_local_calc_idx(idx)
	422	#endif
	423
	424	#ifndef arch_kmap_local_high_get
	425	static inline void arch_kmap_local_high_get(struct page page)
	426	{
	427	return NULL;
	428	}
	429	#endif
	430
	431	/* Unmap a local mapping which was obtained by kmap_high_get() */
	432	static inline void kmap_high_unmap_local(unsigned long vaddr)
	433	{
	434	#ifdef ARCH_NEEDS_KMAP_HIGH_GET
	435	if (vaddr >= PKMAP_ADDR(0) && vaddr < PKMAP_ADDR(LAST_PKMAP))
	436	kunmap_high(pte_page(pkmap_page_table[PKMAP_NR(vaddr)]));
	437	#endif
	438	}
	439
	440	static inline int kmap_local_calc_idx(int idx)
	441	{
	442	return idx + KM_MAX_IDX * smp_processor_id();
	443	}
	444
	445	static pte_t *__kmap_pte;
	446
	447	static pte_t *kmap_get_pte(void)
	448	{
	449	if (!__kmap_pte)
	450	__kmap_pte = virt_to_kpte(__fix_to_virt(FIX_KMAP_BEGIN));
	451	return __kmap_pte;
	452	}
	453
	454	void *__kmap_local_pfn_prot(unsigned long pfn, pgprot_t prot)
	455	{
	456	pte_t pteval, *kmap_pte = kmap_get_pte();
	457	unsigned long vaddr;
	458	int idx;
	459
	460	preempt_disable();
	461	idx = arch_kmap_local_map_idx(kmap_local_idx_push(), pfn);
	462	vaddr = __fix_to_virt(FIX_KMAP_BEGIN + idx);
	463	BUG_ON(!pte_none(*(kmap_pte - idx)));
	464	pteval = pfn_pte(pfn, prot);
	465	set_pte_at(&init_mm, vaddr, kmap_pte - idx, pteval);
	466	arch_kmap_local_post_map(vaddr, pteval);
	467	preempt_enable();
	468
	469	return (void *)vaddr;
	470	}
	471	EXPORT_SYMBOL_GPL(__kmap_local_pfn_prot);
472
473	void __kmap_local_page_prot(struct page page, pgprot_t prot)
474	{
475	void *kmap;
476
477	if (!PageHighMem(page))
478	return page_address(page);
479
480	/* Try kmap_high_get() if architecture has it enabled */
481	kmap = arch_kmap_local_high_get(page);
482	if (kmap)
483	return kmap;
484
485	return __kmap_local_pfn_prot(page_to_pfn(page), prot);
486	}
487	EXPORT_SYMBOL(__kmap_local_page_prot);
488
489	void kunmap_local_indexed(void *vaddr)
490	{
491	unsigned long addr = (unsigned long) vaddr & PAGE_MASK;
492	pte_t *kmap_pte = kmap_get_pte();
493	int idx;
494
495	if (addr < __fix_to_virt(FIX_KMAP_END) \|\|
496	addr > __fix_to_virt(FIX_KMAP_BEGIN)) {
497	WARN_ON_ONCE(addr < PAGE_OFFSET);
498
499	/* Handle mappings which were obtained by kmap_high_get() */
500	kmap_high_unmap_local(addr);
501	return;
502	}
503
504	preempt_disable();
505	idx = arch_kmap_local_unmap_idx(kmap_local_idx(), addr);
506	WARN_ON_ONCE(addr != __fix_to_virt(FIX_KMAP_BEGIN + idx));
507
508	arch_kmap_local_pre_unmap(addr);
509	pte_clear(&init_mm, addr, kmap_pte - idx);
510	arch_kmap_local_post_unmap(addr);
511	kmap_local_idx_pop();
512	preempt_enable();
513	}
514	EXPORT_SYMBOL(kunmap_local_indexed);
515	#endif
1da177e4	516
1da177e4 LT	517	#if defined(HASHED_PAGE_VIRTUAL)
	518
	519	#define PA_HASH_ORDER 7
	520
	521	/*
	522	* Describes one page->virtual association
	523	*/
	524	struct page_address_map {
	525	struct page *page;
	526	void *virtual;
	527	struct list_head list;
	528	};
	529
a354e2c8	530	static struct page_address_map page_address_maps[LAST_PKMAP];
1da177e4 LT	531
	532	/*
	533	* Hash table bucket
	534	*/
	535	static struct page_address_slot {
	536	struct list_head lh; /* List of page_address_maps */
	537	spinlock_t lock; /* Protect this bucket's list */
	538	} ____cacheline_aligned_in_smp page_address_htable[1<<PA_HASH_ORDER];
	539
f9918794	540	static struct page_address_slot page_slot(const struct page page)
1da177e4 LT	541	{
	542	return &page_address_htable[hash_ptr(page, PA_HASH_ORDER)];
	543	}
	544
77f6078a RD	545	/**
	546	* page_address - get the mapped virtual address of a page
	547	* @page: &struct page to get the virtual address of
	548	*
	549	* Returns the page's virtual address.
	550	*/
f9918794	551	void page_address(const struct page page)
1da177e4 LT	552	{
	553	unsigned long flags;
	554	void *ret;
	555	struct page_address_slot *pas;
	556
	557	if (!PageHighMem(page))
	558	return lowmem_page_address(page);
	559
	560	pas = page_slot(page);
	561	ret = NULL;
	562	spin_lock_irqsave(&pas->lock, flags);
	563	if (!list_empty(&pas->lh)) {
	564	struct page_address_map *pam;
	565
	566	list_for_each_entry(pam, &pas->lh, list) {
	567	if (pam->page == page) {
	568	ret = pam->virtual;
	569	goto done;
	570	}
	571	}
	572	}
	573	done:
	574	spin_unlock_irqrestore(&pas->lock, flags);
	575	return ret;
	576	}
	577
	578	EXPORT_SYMBOL(page_address);
	579
77f6078a RD	580	/**
	581	* set_page_address - set a page's virtual address
	582	* @page: &struct page to set
	583	* @virtual: virtual address to use
	584	*/
1da177e4 LT	585	void set_page_address(struct page page, void virtual)
	586	{
	587	unsigned long flags;
	588	struct page_address_slot *pas;
	589	struct page_address_map *pam;
	590
	591	BUG_ON(!PageHighMem(page));
	592
	593	pas = page_slot(page);
	594	if (virtual) { /* Add */
a354e2c8	595	pam = &page_address_maps[PKMAP_NR((unsigned long)virtual)];
1da177e4 LT	596	pam->page = page;
	597	pam->virtual = virtual;
	598
	599	spin_lock_irqsave(&pas->lock, flags);
	600	list_add_tail(&pam->list, &pas->lh);
	601	spin_unlock_irqrestore(&pas->lock, flags);
	602	} else { /* Remove */
	603	spin_lock_irqsave(&pas->lock, flags);
	604	list_for_each_entry(pam, &pas->lh, list) {
	605	if (pam->page == page) {
	606	list_del(&pam->list);
	607	spin_unlock_irqrestore(&pas->lock, flags);
1da177e4 LT	608	goto done;
	609	}
	610	}
	611	spin_unlock_irqrestore(&pas->lock, flags);
	612	}
	613	done:
	614	return;
	615	}
	616
1da177e4 LT	617	void __init page_address_init(void)
	618	{
	619	int i;
	620
1da177e4 LT	621	for (i = 0; i < ARRAY_SIZE(page_address_htable); i++) {
	622	INIT_LIST_HEAD(&page_address_htable[i].lh);
	623	spin_lock_init(&page_address_htable[i].lock);
	624	}
1da177e4 LT	625	}
1da177e4 LT	626
955cc774	627	#endif /* defined(HASHED_PAGE_VIRTUAL) */