[linux-2.6-block.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>

#ifndef CONFIG_KMAP_LOCAL
#ifdef CONFIG_HIGHMEM
DEFINE_PER_CPU(int, __kmap_atomic_idx);
#endif
#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

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
298fa1ad	34	#ifndef CONFIG_KMAP_LOCAL
157e118b	35	#ifdef CONFIG_HIGHMEM
a8e23a29 PZ	36	DEFINE_PER_CPU(int, __kmap_atomic_idx);
a8e23a29 PZ	37	#endif
298fa1ad	38	#endif
a8e23a29	39
1da177e4 LT	40	/*
	41	* Virtual_count is not a pure "count".
	42	* 0 means that it is not mapped, and has not been mapped
	43	* since a TLB flush - it is usable.
	44	* 1 means that there are no users, but it has been mapped
	45	* since the last TLB flush - so we can't use it.
	46	* n means that there are (n-1) current users of it.
	47	*/
	48	#ifdef CONFIG_HIGHMEM
260b2367	49
15de36a4 MF	50	/*
	51	* Architecture with aliasing data cache may define the following family of
	52	* helper functions in its asm/highmem.h to control cache color of virtual
	53	* addresses where physical memory pages are mapped by kmap.
	54	*/
	55	#ifndef get_pkmap_color
	56
	57	/*
	58	* Determine color of virtual address where the page should be mapped.
	59	*/
	60	static inline unsigned int get_pkmap_color(struct page *page)
	61	{
	62	return 0;
	63	}
	64	#define get_pkmap_color get_pkmap_color
	65
	66	/*
	67	* Get next index for mapping inside PKMAP region for page with given color.
	68	*/
	69	static inline unsigned int get_next_pkmap_nr(unsigned int color)
	70	{
	71	static unsigned int last_pkmap_nr;
	72
	73	last_pkmap_nr = (last_pkmap_nr + 1) & LAST_PKMAP_MASK;
	74	return last_pkmap_nr;
	75	}
	76
	77	/*
	78	* Determine if page index inside PKMAP region (pkmap_nr) of given color
	79	* has wrapped around PKMAP region end. When this happens an attempt to
	80	* flush all unused PKMAP slots is made.
	81	*/
	82	static inline int no_more_pkmaps(unsigned int pkmap_nr, unsigned int color)
	83	{
	84	return pkmap_nr == 0;
	85	}
	86
	87	/*
	88	* Get the number of PKMAP entries of the given color. If no free slot is
	89	* found after checking that many entries, kmap will sleep waiting for
	90	* someone to call kunmap and free PKMAP slot.
	91	*/
	92	static inline int get_pkmap_entries_count(unsigned int color)
	93	{
	94	return LAST_PKMAP;
	95	}
	96
	97	/*
	98	* Get head of a wait queue for PKMAP entries of the given color.
	99	* Wait queues for different mapping colors should be independent to avoid
	100	* unnecessary wakeups caused by freeing of slots of other colors.
	101	*/
	102	static inline wait_queue_head_t *get_pkmap_wait_queue_head(unsigned int color)
	103	{
	104	static DECLARE_WAIT_QUEUE_HEAD(pkmap_map_wait);
	105
	106	return &pkmap_map_wait;
	107	}
	108	#endif
	109
ca79b0c2 AK	110	atomic_long_t _totalhigh_pages __read_mostly;
ca79b0c2 AK	111	EXPORT_SYMBOL(_totalhigh_pages);
3e4d3af5	112
c1f60a5a CL	113	unsigned int nr_free_highpages (void)
c1f60a5a CL	114	{
33499bfe	115	struct zone *zone;
c1f60a5a CL	116	unsigned int pages = 0;
c1f60a5a CL	117
33499bfe JK	118	for_each_populated_zone(zone) {
	119	if (is_highmem(zone))
	120	pages += zone_page_state(zone, NR_FREE_PAGES);
2a1e274a	121	}
c1f60a5a CL	122
	123	return pages;
	124	}
	125
1da177e4	126	static int pkmap_count[LAST_PKMAP];
1da177e4 LT	127	static __cacheline_aligned_in_smp DEFINE_SPINLOCK(kmap_lock);
	128
	129	pte_t * pkmap_page_table;
	130
3297e760 NP	131	/*
	132	* Most architectures have no use for kmap_high_get(), so let's abstract
	133	* the disabling of IRQ out of the locking in that case to save on a
	134	* potential useless overhead.
	135	*/
	136	#ifdef ARCH_NEEDS_KMAP_HIGH_GET
	137	#define lock_kmap() spin_lock_irq(&kmap_lock)
	138	#define unlock_kmap() spin_unlock_irq(&kmap_lock)
	139	#define lock_kmap_any(flags) spin_lock_irqsave(&kmap_lock, flags)
	140	#define unlock_kmap_any(flags) spin_unlock_irqrestore(&kmap_lock, flags)
	141	#else
	142	#define lock_kmap() spin_lock(&kmap_lock)
	143	#define unlock_kmap() spin_unlock(&kmap_lock)
	144	#define lock_kmap_any(flags) \
	145	do { spin_lock(&kmap_lock); (void)(flags); } while (0)
	146	#define unlock_kmap_any(flags) \
	147	do { spin_unlock(&kmap_lock); (void)(flags); } while (0)
	148	#endif
	149
5a178119 MG	150	struct page kmap_to_page(void vaddr)
	151	{
	152	unsigned long addr = (unsigned long)vaddr;
	153
498c2280	154	if (addr >= PKMAP_ADDR(0) && addr < PKMAP_ADDR(LAST_PKMAP)) {
4de22c05	155	int i = PKMAP_NR(addr);
5a178119 MG	156	return pte_page(pkmap_page_table[i]);
	157	}
	158
	159	return virt_to_page(addr);
	160	}
f0263d2d	161	EXPORT_SYMBOL(kmap_to_page);
5a178119	162
1da177e4 LT	163	static void flush_all_zero_pkmaps(void)
	164	{
	165	int i;
5843d9a4	166	int need_flush = 0;
1da177e4 LT	167
	168	flush_cache_kmaps();
	169
	170	for (i = 0; i < LAST_PKMAP; i++) {
	171	struct page *page;
	172
	173	/*
	174	* zero means we don't have anything to do,
	175	* >1 means that it is still in use. Only
	176	* a count of 1 means that it is free but
	177	* needs to be unmapped
	178	*/
	179	if (pkmap_count[i] != 1)
	180	continue;
	181	pkmap_count[i] = 0;
	182
	183	/* sanity check */
75babcac	184	BUG_ON(pte_none(pkmap_page_table[i]));
1da177e4 LT	185
	186	/*
	187	* Don't need an atomic fetch-and-clear op here;
	188	* no-one has the page mapped, and cannot get at
	189	* its virtual address (and hence PTE) without first
	190	* getting the kmap_lock (which is held here).
	191	* So no dangers, even with speculative execution.
	192	*/
	193	page = pte_page(pkmap_page_table[i]);
eb2db439	194	pte_clear(&init_mm, PKMAP_ADDR(i), &pkmap_page_table[i]);
1da177e4 LT	195
1da177e4 LT	196	set_page_address(page, NULL);
5843d9a4	197	need_flush = 1;
1da177e4	198	}
5843d9a4 NP	199	if (need_flush)
5843d9a4 NP	200	flush_tlb_kernel_range(PKMAP_ADDR(0), PKMAP_ADDR(LAST_PKMAP));
1da177e4 LT	201	}
1da177e4 LT	202
77f6078a RD	203	/**
	204	* kmap_flush_unused - flush all unused kmap mappings in order to remove stray mappings
	205	*/
ce6234b5 JF	206	void kmap_flush_unused(void)
ce6234b5 JF	207	{
3297e760	208	lock_kmap();
ce6234b5	209	flush_all_zero_pkmaps();
3297e760	210	unlock_kmap();
ce6234b5 JF	211	}
ce6234b5 JF	212
1da177e4 LT	213	static inline unsigned long map_new_virtual(struct page *page)
	214	{
	215	unsigned long vaddr;
	216	int count;
15de36a4 MF	217	unsigned int last_pkmap_nr;
15de36a4 MF	218	unsigned int color = get_pkmap_color(page);
1da177e4 LT	219
1da177e4 LT	220	start:
15de36a4	221	count = get_pkmap_entries_count(color);
1da177e4 LT	222	/* Find an empty entry */
1da177e4 LT	223	for (;;) {
15de36a4 MF	224	last_pkmap_nr = get_next_pkmap_nr(color);
15de36a4 MF	225	if (no_more_pkmaps(last_pkmap_nr, color)) {
1da177e4	226	flush_all_zero_pkmaps();
15de36a4	227	count = get_pkmap_entries_count(color);
1da177e4 LT	228	}
	229	if (!pkmap_count[last_pkmap_nr])
	230	break; /* Found a usable entry */
	231	if (--count)
	232	continue;
	233
	234	/*
	235	* Sleep for somebody else to unmap their entries
	236	*/
	237	{
	238	DECLARE_WAITQUEUE(wait, current);
15de36a4 MF	239	wait_queue_head_t *pkmap_map_wait =
15de36a4 MF	240	get_pkmap_wait_queue_head(color);
1da177e4 LT	241
1da177e4 LT	242	__set_current_state(TASK_UNINTERRUPTIBLE);
15de36a4	243	add_wait_queue(pkmap_map_wait, &wait);
3297e760	244	unlock_kmap();
1da177e4	245	schedule();
15de36a4	246	remove_wait_queue(pkmap_map_wait, &wait);
3297e760	247	lock_kmap();
1da177e4 LT	248
	249	/* Somebody else might have mapped it while we slept */
	250	if (page_address(page))
	251	return (unsigned long)page_address(page);
	252
	253	/* Re-start */
	254	goto start;
	255	}
	256	}
	257	vaddr = PKMAP_ADDR(last_pkmap_nr);
	258	set_pte_at(&init_mm, vaddr,
	259	&(pkmap_page_table[last_pkmap_nr]), mk_pte(page, kmap_prot));
	260
	261	pkmap_count[last_pkmap_nr] = 1;
	262	set_page_address(page, (void *)vaddr);
	263
	264	return vaddr;
	265	}
	266
77f6078a RD	267	/**
	268	* kmap_high - map a highmem page into memory
	269	* @page: &struct page to map
	270	*
	271	* Returns the page's virtual memory address.
	272	*
	273	* We cannot call this from interrupts, as it may block.
	274	*/
920c7a5d	275	void kmap_high(struct page page)
1da177e4 LT	276	{
	277	unsigned long vaddr;
	278
	279	/*
	280	* For highmem pages, we can't trust "virtual" until
	281	* after we have the lock.
1da177e4	282	*/
3297e760	283	lock_kmap();
1da177e4 LT	284	vaddr = (unsigned long)page_address(page);
	285	if (!vaddr)
	286	vaddr = map_new_virtual(page);
	287	pkmap_count[PKMAP_NR(vaddr)]++;
75babcac	288	BUG_ON(pkmap_count[PKMAP_NR(vaddr)] < 2);
3297e760	289	unlock_kmap();
1da177e4 LT	290	return (void*) vaddr;
	291	}
	292
	293	EXPORT_SYMBOL(kmap_high);
	294
3297e760 NP	295	#ifdef ARCH_NEEDS_KMAP_HIGH_GET
	296	/**
	297	* kmap_high_get - pin a highmem page into memory
	298	* @page: &struct page to pin
	299	*
	300	* Returns the page's current virtual memory address, or NULL if no mapping
5e39df56	301	* exists. If and only if a non null address is returned then a
3297e760 NP	302	* matching call to kunmap_high() is necessary.
	303	*
	304	* This can be called from any context.
	305	*/
	306	void kmap_high_get(struct page page)
	307	{
	308	unsigned long vaddr, flags;
	309
	310	lock_kmap_any(flags);
	311	vaddr = (unsigned long)page_address(page);
	312	if (vaddr) {
	313	BUG_ON(pkmap_count[PKMAP_NR(vaddr)] < 1);
	314	pkmap_count[PKMAP_NR(vaddr)]++;
	315	}
	316	unlock_kmap_any(flags);
	317	return (void*) vaddr;
	318	}
	319	#endif
	320
77f6078a	321	/**
4e9dc5df	322	* kunmap_high - unmap a highmem page into memory
77f6078a	323	* @page: &struct page to unmap
3297e760 NP	324	*
	325	* If ARCH_NEEDS_KMAP_HIGH_GET is not defined then this may be called
	326	* only from user context.
77f6078a	327	*/
920c7a5d	328	void kunmap_high(struct page *page)
1da177e4 LT	329	{
	330	unsigned long vaddr;
	331	unsigned long nr;
3297e760	332	unsigned long flags;
1da177e4	333	int need_wakeup;
15de36a4 MF	334	unsigned int color = get_pkmap_color(page);
15de36a4 MF	335	wait_queue_head_t *pkmap_map_wait;
1da177e4	336
3297e760	337	lock_kmap_any(flags);
1da177e4	338	vaddr = (unsigned long)page_address(page);
75babcac	339	BUG_ON(!vaddr);
1da177e4 LT	340	nr = PKMAP_NR(vaddr);
	341
	342	/*
	343	* A count must never go down to zero
	344	* without a TLB flush!
	345	*/
	346	need_wakeup = 0;
	347	switch (--pkmap_count[nr]) {
	348	case 0:
	349	BUG();
	350	case 1:
	351	/*
	352	* Avoid an unnecessary wake_up() function call.
	353	* The common case is pkmap_count[] == 1, but
	354	* no waiters.
	355	* The tasks queued in the wait-queue are guarded
	356	* by both the lock in the wait-queue-head and by
	357	* the kmap_lock. As the kmap_lock is held here,
	358	* no need for the wait-queue-head's lock. Simply
	359	* test if the queue is empty.
	360	*/
15de36a4 MF	361	pkmap_map_wait = get_pkmap_wait_queue_head(color);
15de36a4 MF	362	need_wakeup = waitqueue_active(pkmap_map_wait);
1da177e4	363	}
3297e760	364	unlock_kmap_any(flags);
1da177e4 LT	365
	366	/* do wake-up, if needed, race-free outside of the spin lock */
	367	if (need_wakeup)
15de36a4	368	wake_up(pkmap_map_wait);
1da177e4	369	}
1da177e4	370	EXPORT_SYMBOL(kunmap_high);
298fa1ad TG	371	#endif /* CONFIG_HIGHMEM */
	372
	373	#ifdef CONFIG_KMAP_LOCAL
	374
	375	#include <asm/kmap_size.h>
	376
	377	static DEFINE_PER_CPU(int, __kmap_local_idx);
	378
389755c2 TG	379	/*
	380	* With DEBUG_HIGHMEM the stack depth is doubled and every second
	381	* slot is unused which acts as a guard page
	382	*/
	383	#ifdef CONFIG_DEBUG_HIGHMEM
	384	# define KM_INCR 2
	385	#else
	386	# define KM_INCR 1
	387	#endif
	388
298fa1ad TG	389	static inline int kmap_local_idx_push(void)
298fa1ad TG	390	{
389755c2	391	int idx = __this_cpu_add_return(__kmap_local_idx, KM_INCR) - 1;
298fa1ad TG	392
	393	WARN_ON_ONCE(in_irq() && !irqs_disabled());
	394	BUG_ON(idx >= KM_MAX_IDX);
	395	return idx;
	396	}
	397
	398	static inline int kmap_local_idx(void)
	399	{
	400	return __this_cpu_read(__kmap_local_idx) - 1;
	401	}
	402
	403	static inline void kmap_local_idx_pop(void)
	404	{
389755c2	405	int idx = __this_cpu_sub_return(__kmap_local_idx, KM_INCR);
298fa1ad TG	406
	407	BUG_ON(idx < 0);
	408	}
	409
	410	#ifndef arch_kmap_local_post_map
	411	# define arch_kmap_local_post_map(vaddr, pteval) do { } while (0)
	412	#endif
	413	#ifndef arch_kmap_local_pre_unmap
	414	# define arch_kmap_local_pre_unmap(vaddr) do { } while (0)
	415	#endif
	416
	417	#ifndef arch_kmap_local_post_unmap
	418	# define arch_kmap_local_post_unmap(vaddr) do { } while (0)
	419	#endif
	420
	421	#ifndef arch_kmap_local_map_idx
	422	#define arch_kmap_local_map_idx(idx, pfn) kmap_local_calc_idx(idx)
	423	#endif
	424
	425	#ifndef arch_kmap_local_unmap_idx
	426	#define arch_kmap_local_unmap_idx(idx, vaddr) kmap_local_calc_idx(idx)
	427	#endif
	428
	429	#ifndef arch_kmap_local_high_get
	430	static inline void arch_kmap_local_high_get(struct page page)
	431	{
	432	return NULL;
	433	}
	434	#endif
	435
	436	/* Unmap a local mapping which was obtained by kmap_high_get() */
	437	static inline void kmap_high_unmap_local(unsigned long vaddr)
	438	{
	439	#ifdef ARCH_NEEDS_KMAP_HIGH_GET
	440	if (vaddr >= PKMAP_ADDR(0) && vaddr < PKMAP_ADDR(LAST_PKMAP))
	441	kunmap_high(pte_page(pkmap_page_table[PKMAP_NR(vaddr)]));
	442	#endif
	443	}
	444
	445	static inline int kmap_local_calc_idx(int idx)
	446	{
	447	return idx + KM_MAX_IDX * smp_processor_id();
	448	}
	449
	450	static pte_t *__kmap_pte;
	451
	452	static pte_t *kmap_get_pte(void)
	453	{
	454	if (!__kmap_pte)
	455	__kmap_pte = virt_to_kpte(__fix_to_virt(FIX_KMAP_BEGIN));
	456	return __kmap_pte;
	457	}
	458
	459	void *__kmap_local_pfn_prot(unsigned long pfn, pgprot_t prot)
	460	{
	461	pte_t pteval, *kmap_pte = kmap_get_pte();
	462	unsigned long vaddr;
	463	int idx;
	464
	465	preempt_disable();
	466	idx = arch_kmap_local_map_idx(kmap_local_idx_push(), pfn);
	467	vaddr = __fix_to_virt(FIX_KMAP_BEGIN + idx);
	468	BUG_ON(!pte_none(*(kmap_pte - idx)));
	469	pteval = pfn_pte(pfn, prot);
470	set_pte_at(&init_mm, vaddr, kmap_pte - idx, pteval);
471	arch_kmap_local_post_map(vaddr, pteval);
472	preempt_enable();
473
474	return (void *)vaddr;
475	}
476	EXPORT_SYMBOL_GPL(__kmap_local_pfn_prot);
477
478	void __kmap_local_page_prot(struct page page, pgprot_t prot)
479	{
480	void *kmap;
481
482	if (!PageHighMem(page))
483	return page_address(page);
484
485	/* Try kmap_high_get() if architecture has it enabled */
486	kmap = arch_kmap_local_high_get(page);
487	if (kmap)
488	return kmap;
489
490	return __kmap_local_pfn_prot(page_to_pfn(page), prot);
491	}
492	EXPORT_SYMBOL(__kmap_local_page_prot);
493
494	void kunmap_local_indexed(void *vaddr)
495	{
496	unsigned long addr = (unsigned long) vaddr & PAGE_MASK;
497	pte_t *kmap_pte = kmap_get_pte();
498	int idx;
499
500	if (addr < __fix_to_virt(FIX_KMAP_END) \|\|
501	addr > __fix_to_virt(FIX_KMAP_BEGIN)) {
502	WARN_ON_ONCE(addr < PAGE_OFFSET);
503
504	/* Handle mappings which were obtained by kmap_high_get() */
505	kmap_high_unmap_local(addr);
506	return;
507	}
508
509	preempt_disable();
510	idx = arch_kmap_local_unmap_idx(kmap_local_idx(), addr);
511	WARN_ON_ONCE(addr != __fix_to_virt(FIX_KMAP_BEGIN + idx));
512
513	arch_kmap_local_pre_unmap(addr);
514	pte_clear(&init_mm, addr, kmap_pte - idx);
515	arch_kmap_local_post_unmap(addr);
516	kmap_local_idx_pop();
517	preempt_enable();
518	}
519	EXPORT_SYMBOL(kunmap_local_indexed);
520	#endif
1da177e4	521
1da177e4 LT	522	#if defined(HASHED_PAGE_VIRTUAL)
	523
	524	#define PA_HASH_ORDER 7
	525
	526	/*
	527	* Describes one page->virtual association
	528	*/
	529	struct page_address_map {
	530	struct page *page;
	531	void *virtual;
	532	struct list_head list;
	533	};
	534
a354e2c8	535	static struct page_address_map page_address_maps[LAST_PKMAP];
1da177e4 LT	536
	537	/*
	538	* Hash table bucket
	539	*/
	540	static struct page_address_slot {
	541	struct list_head lh; /* List of page_address_maps */
	542	spinlock_t lock; /* Protect this bucket's list */
	543	} ____cacheline_aligned_in_smp page_address_htable[1<<PA_HASH_ORDER];
	544
f9918794	545	static struct page_address_slot page_slot(const struct page page)
1da177e4 LT	546	{
	547	return &page_address_htable[hash_ptr(page, PA_HASH_ORDER)];
	548	}
	549
77f6078a RD	550	/**
	551	* page_address - get the mapped virtual address of a page
	552	* @page: &struct page to get the virtual address of
	553	*
	554	* Returns the page's virtual address.
	555	*/
f9918794	556	void page_address(const struct page page)
1da177e4 LT	557	{
	558	unsigned long flags;
	559	void *ret;
	560	struct page_address_slot *pas;
	561
	562	if (!PageHighMem(page))
	563	return lowmem_page_address(page);
	564
	565	pas = page_slot(page);
	566	ret = NULL;
	567	spin_lock_irqsave(&pas->lock, flags);
	568	if (!list_empty(&pas->lh)) {
	569	struct page_address_map *pam;
	570
	571	list_for_each_entry(pam, &pas->lh, list) {
	572	if (pam->page == page) {
	573	ret = pam->virtual;
	574	goto done;
	575	}
	576	}
	577	}
	578	done:
	579	spin_unlock_irqrestore(&pas->lock, flags);
	580	return ret;
	581	}
	582
	583	EXPORT_SYMBOL(page_address);
	584
77f6078a RD	585	/**
	586	* set_page_address - set a page's virtual address
	587	* @page: &struct page to set
	588	* @virtual: virtual address to use
	589	*/
1da177e4 LT	590	void set_page_address(struct page page, void virtual)
	591	{
	592	unsigned long flags;
	593	struct page_address_slot *pas;
	594	struct page_address_map *pam;
	595
	596	BUG_ON(!PageHighMem(page));
	597
	598	pas = page_slot(page);
	599	if (virtual) { /* Add */
a354e2c8	600	pam = &page_address_maps[PKMAP_NR((unsigned long)virtual)];
1da177e4 LT	601	pam->page = page;
	602	pam->virtual = virtual;
	603
	604	spin_lock_irqsave(&pas->lock, flags);
	605	list_add_tail(&pam->list, &pas->lh);
	606	spin_unlock_irqrestore(&pas->lock, flags);
	607	} else { /* Remove */
	608	spin_lock_irqsave(&pas->lock, flags);
	609	list_for_each_entry(pam, &pas->lh, list) {
	610	if (pam->page == page) {
	611	list_del(&pam->list);
	612	spin_unlock_irqrestore(&pas->lock, flags);
1da177e4 LT	613	goto done;
	614	}
	615	}
	616	spin_unlock_irqrestore(&pas->lock, flags);
	617	}
	618	done:
	619	return;
	620	}
	621
1da177e4 LT	622	void __init page_address_init(void)
	623	{
	624	int i;
	625
1da177e4 LT	626	for (i = 0; i < ARRAY_SIZE(page_address_htable); i++) {
	627	INIT_LIST_HEAD(&page_address_htable[i].lh);
	628	spin_lock_init(&page_address_htable[i].lock);
	629	}
1da177e4 LT	630	}
1da177e4 LT	631
955cc774	632	#endif /* defined(HASHED_PAGE_VIRTUAL) */