[linux-2.6-block.git] / include / linux / mmu_notifier.h

/* SPDX-License-Identifier: GPL-2.0 */
#ifndef _LINUX_MMU_NOTIFIER_H
#define _LINUX_MMU_NOTIFIER_H

#include <linux/types.h>
#include <linux/list.h>
#include <linux/spinlock.h>
#include <linux/mm_types.h>
#include <linux/srcu.h>

struct mmu_notifier;
struct mmu_notifier_ops;

/* mmu_notifier_ops flags */
#define MMU_INVALIDATE_DOES_NOT_BLOCK	(0x01)

#ifdef CONFIG_MMU_NOTIFIER

/*
 * The mmu notifier_mm structure is allocated and installed in
 * mm->mmu_notifier_mm inside the mm_take_all_locks() protected
 * critical section and it's released only when mm_count reaches zero
 * in mmdrop().
 */
struct mmu_notifier_mm {
	/* all mmu notifiers registerd in this mm are queued in this list */
	struct hlist_head list;
	/* to serialize the list modifications and hlist_unhashed */
	spinlock_t lock;
};

struct mmu_notifier_ops {
	/*
	 * Flags to specify behavior of callbacks for this MMU notifier.
	 * Used to determine which context an operation may be called.
	 *
	 * MMU_INVALIDATE_DOES_NOT_BLOCK: invalidate_range_* callbacks do not
	 *	block
	 */
	int flags;

	/*
	 * Called either by mmu_notifier_unregister or when the mm is
	 * being destroyed by exit_mmap, always before all pages are
	 * freed. This can run concurrently with other mmu notifier
	 * methods (the ones invoked outside the mm context) and it
	 * should tear down all secondary mmu mappings and freeze the
	 * secondary mmu. If this method isn't implemented you've to
	 * be sure that nothing could possibly write to the pages
	 * through the secondary mmu by the time the last thread with
	 * tsk->mm == mm exits.
	 *
	 * As side note: the pages freed after ->release returns could
	 * be immediately reallocated by the gart at an alias physical
	 * address with a different cache model, so if ->release isn't
	 * implemented because all _software_ driven memory accesses
	 * through the secondary mmu are terminated by the time the
	 * last thread of this mm quits, you've also to be sure that
	 * speculative _hardware_ operations can't allocate dirty
	 * cachelines in the cpu that could not be snooped and made
	 * coherent with the other read and write operations happening
	 * through the gart alias address, so leading to memory
	 * corruption.
	 */
	void (*release)(struct mmu_notifier *mn,
			struct mm_struct *mm);

	/*
	 * clear_flush_young is called after the VM is
	 * test-and-clearing the young/accessed bitflag in the
	 * pte. This way the VM will provide proper aging to the
	 * accesses to the page through the secondary MMUs and not
	 * only to the ones through the Linux pte.
	 * Start-end is necessary in case the secondary MMU is mapping the page
	 * at a smaller granularity than the primary MMU.
	 */
	int (*clear_flush_young)(struct mmu_notifier *mn,
				 struct mm_struct *mm,
				 unsigned long start,
				 unsigned long end);

	/*
	 * clear_young is a lightweight version of clear_flush_young. Like the
	 * latter, it is supposed to test-and-clear the young/accessed bitflag
	 * in the secondary pte, but it may omit flushing the secondary tlb.
	 */
	int (*clear_young)(struct mmu_notifier *mn,
			   struct mm_struct *mm,
			   unsigned long start,
			   unsigned long end);

	/*
	 * test_young is called to check the young/accessed bitflag in
	 * the secondary pte. This is used to know if the page is
	 * frequently used without actually clearing the flag or tearing
	 * down the secondary mapping on the page.
	 */
	int (*test_young)(struct mmu_notifier *mn,
			  struct mm_struct *mm,
			  unsigned long address);

	/*
	 * change_pte is called in cases that pte mapping to page is changed:
	 * for example, when ksm remaps pte to point to a new shared page.
	 */
	void (*change_pte)(struct mmu_notifier *mn,
			   struct mm_struct *mm,
			   unsigned long address,
			   pte_t pte);

	/*
	 * invalidate_range_start() and invalidate_range_end() must be
	 * paired and are called only when the mmap_sem and/or the
	 * locks protecting the reverse maps are held. If the subsystem
	 * can't guarantee that no additional references are taken to
	 * the pages in the range, it has to implement the
	 * invalidate_range() notifier to remove any references taken
	 * after invalidate_range_start().
	 *
	 * Invalidation of multiple concurrent ranges may be
	 * optionally permitted by the driver. Either way the
	 * establishment of sptes is forbidden in the range passed to
	 * invalidate_range_begin/end for the whole duration of the
	 * invalidate_range_begin/end critical section.
	 *
	 * invalidate_range_start() is called when all pages in the
	 * range are still mapped and have at least a refcount of one.
	 *
	 * invalidate_range_end() is called when all pages in the
	 * range have been unmapped and the pages have been freed by
	 * the VM.
	 *
	 * The VM will remove the page table entries and potentially
	 * the page between invalidate_range_start() and
	 * invalidate_range_end(). If the page must not be freed
	 * because of pending I/O or other circumstances then the
	 * invalidate_range_start() callback (or the initial mapping
	 * by the driver) must make sure that the refcount is kept
	 * elevated.
	 *
	 * If the driver increases the refcount when the pages are
	 * initially mapped into an address space then either
	 * invalidate_range_start() or invalidate_range_end() may
	 * decrease the refcount. If the refcount is decreased on
	 * invalidate_range_start() then the VM can free pages as page
	 * table entries are removed.  If the refcount is only
	 * droppped on invalidate_range_end() then the driver itself
	 * will drop the last refcount but it must take care to flush
	 * any secondary tlb before doing the final free on the
	 * page. Pages will no longer be referenced by the linux
	 * address space but may still be referenced by sptes until
	 * the last refcount is dropped.
	 *
	 * If blockable argument is set to false then the callback cannot
	 * sleep and has to return with -EAGAIN. 0 should be returned
	 * otherwise.
	 *
	 */
	int (*invalidate_range_start)(struct mmu_notifier *mn,
				       struct mm_struct *mm,
				       unsigned long start, unsigned long end,
				       bool blockable);
	void (*invalidate_range_end)(struct mmu_notifier *mn,
				     struct mm_struct *mm,
				     unsigned long start, unsigned long end);

	/*
	 * invalidate_range() is either called between
	 * invalidate_range_start() and invalidate_range_end() when the
	 * VM has to free pages that where unmapped, but before the
	 * pages are actually freed, or outside of _start()/_end() when
	 * a (remote) TLB is necessary.
	 *
	 * If invalidate_range() is used to manage a non-CPU TLB with
	 * shared page-tables, it not necessary to implement the
	 * invalidate_range_start()/end() notifiers, as
	 * invalidate_range() alread catches the points in time when an
	 * external TLB range needs to be flushed. For more in depth
	 * discussion on this see Documentation/vm/mmu_notifier.rst
	 *
	 * Note that this function might be called with just a sub-range
	 * of what was passed to invalidate_range_start()/end(), if
	 * called between those functions.
	 *
	 * If this callback cannot block, and invalidate_range_{start,end}
	 * cannot block, mmu_notifier_ops.flags should have
	 * MMU_INVALIDATE_DOES_NOT_BLOCK set.
	 */
	void (*invalidate_range)(struct mmu_notifier *mn, struct mm_struct *mm,
				 unsigned long start, unsigned long end);
};

/*
 * The notifier chains are protected by mmap_sem and/or the reverse map
 * semaphores. Notifier chains are only changed when all reverse maps and
 * the mmap_sem locks are taken.
 *
 * Therefore notifier chains can only be traversed when either
 *
 * 1. mmap_sem is held.
 * 2. One of the reverse map locks is held (i_mmap_rwsem or anon_vma->rwsem).
 * 3. No other concurrent thread can access the list (release)
 */
struct mmu_notifier {
	struct hlist_node hlist;
	const struct mmu_notifier_ops *ops;
};

static inline int mm_has_notifiers(struct mm_struct *mm)
{
	return unlikely(mm->mmu_notifier_mm);
}

extern int mmu_notifier_register(struct mmu_notifier *mn,
				 struct mm_struct *mm);
extern int __mmu_notifier_register(struct mmu_notifier *mn,
				   struct mm_struct *mm);
extern void mmu_notifier_unregister(struct mmu_notifier *mn,
				    struct mm_struct *mm);
extern void mmu_notifier_unregister_no_release(struct mmu_notifier *mn,
					       struct mm_struct *mm);
extern void __mmu_notifier_mm_destroy(struct mm_struct *mm);
extern void __mmu_notifier_release(struct mm_struct *mm);
extern int __mmu_notifier_clear_flush_young(struct mm_struct *mm,
					  unsigned long start,
					  unsigned long end);
extern int __mmu_notifier_clear_young(struct mm_struct *mm,
				      unsigned long start,
				      unsigned long end);
extern int __mmu_notifier_test_young(struct mm_struct *mm,
				     unsigned long address);
extern void __mmu_notifier_change_pte(struct mm_struct *mm,
				      unsigned long address, pte_t pte);
extern int __mmu_notifier_invalidate_range_start(struct mm_struct *mm,
				  unsigned long start, unsigned long end,
				  bool blockable);
extern void __mmu_notifier_invalidate_range_end(struct mm_struct *mm,
				  unsigned long start, unsigned long end,
				  bool only_end);
extern void __mmu_notifier_invalidate_range(struct mm_struct *mm,
				  unsigned long start, unsigned long end);
extern bool mm_has_blockable_invalidate_notifiers(struct mm_struct *mm);

static inline void mmu_notifier_release(struct mm_struct *mm)
{
	if (mm_has_notifiers(mm))
		__mmu_notifier_release(mm);
}

static inline int mmu_notifier_clear_flush_young(struct mm_struct *mm,
					  unsigned long start,
					  unsigned long end)
{
	if (mm_has_notifiers(mm))
		return __mmu_notifier_clear_flush_young(mm, start, end);
	return 0;
}

static inline int mmu_notifier_clear_young(struct mm_struct *mm,
					   unsigned long start,
					   unsigned long end)
{
	if (mm_has_notifiers(mm))
		return __mmu_notifier_clear_young(mm, start, end);
	return 0;
}

static inline int mmu_notifier_test_young(struct mm_struct *mm,
					  unsigned long address)
{
	if (mm_has_notifiers(mm))
		return __mmu_notifier_test_young(mm, address);
	return 0;
}

static inline void mmu_notifier_change_pte(struct mm_struct *mm,
					   unsigned long address, pte_t pte)
{
	if (mm_has_notifiers(mm))
		__mmu_notifier_change_pte(mm, address, pte);
}

static inline void mmu_notifier_invalidate_range_start(struct mm_struct *mm,
				  unsigned long start, unsigned long end)
{
	if (mm_has_notifiers(mm))
		__mmu_notifier_invalidate_range_start(mm, start, end, true);
}

static inline int mmu_notifier_invalidate_range_start_nonblock(struct mm_struct *mm,
				  unsigned long start, unsigned long end)
{
	if (mm_has_notifiers(mm))
		return __mmu_notifier_invalidate_range_start(mm, start, end, false);
	return 0;
}

static inline void mmu_notifier_invalidate_range_end(struct mm_struct *mm,
				  unsigned long start, unsigned long end)
{
	if (mm_has_notifiers(mm))
		__mmu_notifier_invalidate_range_end(mm, start, end, false);
}

static inline void mmu_notifier_invalidate_range_only_end(struct mm_struct *mm,
				  unsigned long start, unsigned long end)
{
	if (mm_has_notifiers(mm))
		__mmu_notifier_invalidate_range_end(mm, start, end, true);
}

static inline void mmu_notifier_invalidate_range(struct mm_struct *mm,
				  unsigned long start, unsigned long end)
{
	if (mm_has_notifiers(mm))
		__mmu_notifier_invalidate_range(mm, start, end);
}

static inline void mmu_notifier_mm_init(struct mm_struct *mm)
{
	mm->mmu_notifier_mm = NULL;
}

static inline void mmu_notifier_mm_destroy(struct mm_struct *mm)
{
	if (mm_has_notifiers(mm))
		__mmu_notifier_mm_destroy(mm);
}

#define ptep_clear_flush_young_notify(__vma, __address, __ptep)		\
({									\
	int __young;							\
	struct vm_area_struct *___vma = __vma;				\
	unsigned long ___address = __address;				\
	__young = ptep_clear_flush_young(___vma, ___address, __ptep);	\
	__young |= mmu_notifier_clear_flush_young(___vma->vm_mm,	\
						  ___address,		\
						  ___address +		\
							PAGE_SIZE);	\
	__young;							\
})

#define pmdp_clear_flush_young_notify(__vma, __address, __pmdp)		\
({									\
	int __young;							\
	struct vm_area_struct *___vma = __vma;				\
	unsigned long ___address = __address;				\
	__young = pmdp_clear_flush_young(___vma, ___address, __pmdp);	\
	__young |= mmu_notifier_clear_flush_young(___vma->vm_mm,	\
						  ___address,		\
						  ___address +		\
							PMD_SIZE);	\
	__young;							\
})

#define ptep_clear_young_notify(__vma, __address, __ptep)		\
({									\
	int __young;							\
	struct vm_area_struct *___vma = __vma;				\
	unsigned long ___address = __address;				\
	__young = ptep_test_and_clear_young(___vma, ___address, __ptep);\
	__young |= mmu_notifier_clear_young(___vma->vm_mm, ___address,	\
					    ___address + PAGE_SIZE);	\
	__young;							\
})

#define pmdp_clear_young_notify(__vma, __address, __pmdp)		\
({									\
	int __young;							\
	struct vm_area_struct *___vma = __vma;				\
	unsigned long ___address = __address;				\
	__young = pmdp_test_and_clear_young(___vma, ___address, __pmdp);\
	__young |= mmu_notifier_clear_young(___vma->vm_mm, ___address,	\
					    ___address + PMD_SIZE);	\
	__young;							\
})

#define	ptep_clear_flush_notify(__vma, __address, __ptep)		\
({									\
	unsigned long ___addr = __address & PAGE_MASK;			\
	struct mm_struct *___mm = (__vma)->vm_mm;			\
	pte_t ___pte;							\
									\
	___pte = ptep_clear_flush(__vma, __address, __ptep);		\
	mmu_notifier_invalidate_range(___mm, ___addr,			\
					___addr + PAGE_SIZE);		\
									\
	___pte;								\
})

#define pmdp_huge_clear_flush_notify(__vma, __haddr, __pmd)		\
({									\
	unsigned long ___haddr = __haddr & HPAGE_PMD_MASK;		\
	struct mm_struct *___mm = (__vma)->vm_mm;			\
	pmd_t ___pmd;							\
									\
	___pmd = pmdp_huge_clear_flush(__vma, __haddr, __pmd);		\
	mmu_notifier_invalidate_range(___mm, ___haddr,			\
				      ___haddr + HPAGE_PMD_SIZE);	\
									\
	___pmd;								\
})

#define pudp_huge_clear_flush_notify(__vma, __haddr, __pud)		\
({									\
	unsigned long ___haddr = __haddr & HPAGE_PUD_MASK;		\
	struct mm_struct *___mm = (__vma)->vm_mm;			\
	pud_t ___pud;							\
									\
	___pud = pudp_huge_clear_flush(__vma, __haddr, __pud);		\
	mmu_notifier_invalidate_range(___mm, ___haddr,			\
				      ___haddr + HPAGE_PUD_SIZE);	\
									\
	___pud;								\
})

/*
 * set_pte_at_notify() sets the pte _after_ running the notifier.
 * This is safe to start by updating the secondary MMUs, because the primary MMU
 * pte invalidate must have already happened with a ptep_clear_flush() before
 * set_pte_at_notify() has been invoked.  Updating the secondary MMUs first is
 * required when we change both the protection of the mapping from read-only to
 * read-write and the pfn (like during copy on write page faults). Otherwise the
 * old page would remain mapped readonly in the secondary MMUs after the new
 * page is already writable by some CPU through the primary MMU.
 */
#define set_pte_at_notify(__mm, __address, __ptep, __pte)		\
({									\
	struct mm_struct *___mm = __mm;					\
	unsigned long ___address = __address;				\
	pte_t ___pte = __pte;						\
									\
	mmu_notifier_change_pte(___mm, ___address, ___pte);		\
	set_pte_at(___mm, ___address, __ptep, ___pte);			\
})

extern void mmu_notifier_call_srcu(struct rcu_head *rcu,
				   void (*func)(struct rcu_head *rcu));
extern void mmu_notifier_synchronize(void);

#else /* CONFIG_MMU_NOTIFIER */

static inline int mm_has_notifiers(struct mm_struct *mm)
{
	return 0;
}

static inline void mmu_notifier_release(struct mm_struct *mm)
{
}

static inline int mmu_notifier_clear_flush_young(struct mm_struct *mm,
					  unsigned long start,
					  unsigned long end)
{
	return 0;
}

static inline int mmu_notifier_test_young(struct mm_struct *mm,
					  unsigned long address)
{
	return 0;
}

static inline void mmu_notifier_change_pte(struct mm_struct *mm,
					   unsigned long address, pte_t pte)
{
}

static inline void mmu_notifier_invalidate_range_start(struct mm_struct *mm,
				  unsigned long start, unsigned long end)
{
}

static inline int mmu_notifier_invalidate_range_start_nonblock(struct mm_struct *mm,
				  unsigned long start, unsigned long end)
{
	return 0;
}

static inline void mmu_notifier_invalidate_range_end(struct mm_struct *mm,
				  unsigned long start, unsigned long end)
{
}

static inline void mmu_notifier_invalidate_range_only_end(struct mm_struct *mm,
				  unsigned long start, unsigned long end)
{
}

static inline void mmu_notifier_invalidate_range(struct mm_struct *mm,
				  unsigned long start, unsigned long end)
{
}

static inline bool mm_has_blockable_invalidate_notifiers(struct mm_struct *mm)
{
	return false;
}

static inline void mmu_notifier_mm_init(struct mm_struct *mm)
{
}

static inline void mmu_notifier_mm_destroy(struct mm_struct *mm)
{
}

#define ptep_clear_flush_young_notify ptep_clear_flush_young
#define pmdp_clear_flush_young_notify pmdp_clear_flush_young
#define ptep_clear_young_notify ptep_test_and_clear_young
#define pmdp_clear_young_notify pmdp_test_and_clear_young
#define	ptep_clear_flush_notify ptep_clear_flush
#define pmdp_huge_clear_flush_notify pmdp_huge_clear_flush
#define pudp_huge_clear_flush_notify pudp_huge_clear_flush
#define set_pte_at_notify set_pte_at

#endif /* CONFIG_MMU_NOTIFIER */

#endif /* _LINUX_MMU_NOTIFIER_H */
Commit	Line	Data
b2441318	1	/* SPDX-License-Identifier: GPL-2.0 */
cddb8a5c AA	2	#ifndef _LINUX_MMU_NOTIFIER_H
	3	#define _LINUX_MMU_NOTIFIER_H
	4
5ff7091f	5	#include <linux/types.h>
cddb8a5c AA	6	#include <linux/list.h>
	7	#include <linux/spinlock.h>
	8	#include <linux/mm_types.h>
21a92735	9	#include <linux/srcu.h>
cddb8a5c AA	10
	11	struct mmu_notifier;
	12	struct mmu_notifier_ops;
	13
5ff7091f DR	14	/* mmu_notifier_ops flags */
	15	#define MMU_INVALIDATE_DOES_NOT_BLOCK (0x01)
	16
cddb8a5c AA	17	#ifdef CONFIG_MMU_NOTIFIER
	18
	19	/*
	20	* The mmu notifier_mm structure is allocated and installed in
	21	* mm->mmu_notifier_mm inside the mm_take_all_locks() protected
	22	* critical section and it's released only when mm_count reaches zero
	23	* in mmdrop().
	24	*/
	25	struct mmu_notifier_mm {
	26	/* all mmu notifiers registerd in this mm are queued in this list */
	27	struct hlist_head list;
	28	/* to serialize the list modifications and hlist_unhashed */
	29	spinlock_t lock;
	30	};
	31
	32	struct mmu_notifier_ops {
5ff7091f DR	33	/*
	34	* Flags to specify behavior of callbacks for this MMU notifier.
	35	* Used to determine which context an operation may be called.
	36	*
	37	* MMU_INVALIDATE_DOES_NOT_BLOCK: invalidate_range_* callbacks do not
	38	* block
	39	*/
	40	int flags;
	41
cddb8a5c AA	42	/*
	43	* Called either by mmu_notifier_unregister or when the mm is
	44	* being destroyed by exit_mmap, always before all pages are
	45	* freed. This can run concurrently with other mmu notifier
	46	* methods (the ones invoked outside the mm context) and it
	47	* should tear down all secondary mmu mappings and freeze the
	48	* secondary mmu. If this method isn't implemented you've to
	49	* be sure that nothing could possibly write to the pages
	50	* through the secondary mmu by the time the last thread with
	51	* tsk->mm == mm exits.
	52	*
	53	* As side note: the pages freed after ->release returns could
	54	* be immediately reallocated by the gart at an alias physical
	55	* address with a different cache model, so if ->release isn't
	56	* implemented because all _software_ driven memory accesses
	57	* through the secondary mmu are terminated by the time the
	58	* last thread of this mm quits, you've also to be sure that
	59	* speculative _hardware_ operations can't allocate dirty
	60	* cachelines in the cpu that could not be snooped and made
	61	* coherent with the other read and write operations happening
	62	* through the gart alias address, so leading to memory
	63	* corruption.
	64	*/
	65	void (release)(struct mmu_notifier mn,
	66	struct mm_struct *mm);
	67
	68	/*
	69	* clear_flush_young is called after the VM is
	70	* test-and-clearing the young/accessed bitflag in the
	71	* pte. This way the VM will provide proper aging to the
	72	* accesses to the page through the secondary MMUs and not
	73	* only to the ones through the Linux pte.
57128468 ALC	74	* Start-end is necessary in case the secondary MMU is mapping the page
57128468 ALC	75	* at a smaller granularity than the primary MMU.
cddb8a5c AA	76	*/
	77	int (clear_flush_young)(struct mmu_notifier mn,
	78	struct mm_struct *mm,
57128468 ALC	79	unsigned long start,
57128468 ALC	80	unsigned long end);
cddb8a5c	81
1d7715c6 VD	82	/*
	83	* clear_young is a lightweight version of clear_flush_young. Like the
	84	* latter, it is supposed to test-and-clear the young/accessed bitflag
	85	* in the secondary pte, but it may omit flushing the secondary tlb.
	86	*/
	87	int (clear_young)(struct mmu_notifier mn,
	88	struct mm_struct *mm,
	89	unsigned long start,
	90	unsigned long end);
	91
8ee53820 AA	92	/*
	93	* test_young is called to check the young/accessed bitflag in
	94	* the secondary pte. This is used to know if the page is
	95	* frequently used without actually clearing the flag or tearing
	96	* down the secondary mapping on the page.
	97	*/
	98	int (test_young)(struct mmu_notifier mn,
	99	struct mm_struct *mm,
	100	unsigned long address);
	101
828502d3 IE	102	/*
	103	* change_pte is called in cases that pte mapping to page is changed:
	104	* for example, when ksm remaps pte to point to a new shared page.
	105	*/
	106	void (change_pte)(struct mmu_notifier mn,
	107	struct mm_struct *mm,
	108	unsigned long address,
	109	pte_t pte);
	110
cddb8a5c AA	111	/*
	112	* invalidate_range_start() and invalidate_range_end() must be
	113	* paired and are called only when the mmap_sem and/or the
0f0a327f JR	114	* locks protecting the reverse maps are held. If the subsystem
	115	* can't guarantee that no additional references are taken to
	116	* the pages in the range, it has to implement the
	117	* invalidate_range() notifier to remove any references taken
	118	* after invalidate_range_start().
cddb8a5c AA	119	*
	120	* Invalidation of multiple concurrent ranges may be
	121	* optionally permitted by the driver. Either way the
	122	* establishment of sptes is forbidden in the range passed to
	123	* invalidate_range_begin/end for the whole duration of the
	124	* invalidate_range_begin/end critical section.
	125	*
	126	* invalidate_range_start() is called when all pages in the
	127	* range are still mapped and have at least a refcount of one.
	128	*
	129	* invalidate_range_end() is called when all pages in the
	130	* range have been unmapped and the pages have been freed by
	131	* the VM.
	132	*
	133	* The VM will remove the page table entries and potentially
	134	* the page between invalidate_range_start() and
	135	* invalidate_range_end(). If the page must not be freed
	136	* because of pending I/O or other circumstances then the
	137	* invalidate_range_start() callback (or the initial mapping
	138	* by the driver) must make sure that the refcount is kept
	139	* elevated.
	140	*
	141	* If the driver increases the refcount when the pages are
	142	* initially mapped into an address space then either
	143	* invalidate_range_start() or invalidate_range_end() may
	144	* decrease the refcount. If the refcount is decreased on
	145	* invalidate_range_start() then the VM can free pages as page
	146	* table entries are removed. If the refcount is only
	147	* droppped on invalidate_range_end() then the driver itself
	148	* will drop the last refcount but it must take care to flush
	149	* any secondary tlb before doing the final free on the
	150	* page. Pages will no longer be referenced by the linux
	151	* address space but may still be referenced by sptes until
	152	* the last refcount is dropped.
5ff7091f	153	*
93065ac7 MH	154	* If blockable argument is set to false then the callback cannot
	155	* sleep and has to return with -EAGAIN. 0 should be returned
	156	* otherwise.
	157	*
cddb8a5c	158	*/
93065ac7	159	int (invalidate_range_start)(struct mmu_notifier mn,
cddb8a5c	160	struct mm_struct *mm,
93065ac7 MH	161	unsigned long start, unsigned long end,
93065ac7 MH	162	bool blockable);
cddb8a5c AA	163	void (invalidate_range_end)(struct mmu_notifier mn,
	164	struct mm_struct *mm,
	165	unsigned long start, unsigned long end);
0f0a327f JR	166
	167	/*
	168	* invalidate_range() is either called between
	169	* invalidate_range_start() and invalidate_range_end() when the
	170	* VM has to free pages that where unmapped, but before the
	171	* pages are actually freed, or outside of _start()/_end() when
	172	* a (remote) TLB is necessary.
	173	*
	174	* If invalidate_range() is used to manage a non-CPU TLB with
	175	* shared page-tables, it not necessary to implement the
	176	* invalidate_range_start()/end() notifiers, as
	177	* invalidate_range() alread catches the points in time when an
0f10851e	178	* external TLB range needs to be flushed. For more in depth
ad56b738	179	* discussion on this see Documentation/vm/mmu_notifier.rst
0f0a327f	180	*
0f0a327f JR	181	* Note that this function might be called with just a sub-range
	182	* of what was passed to invalidate_range_start()/end(), if
	183	* called between those functions.
5ff7091f DR	184	*
	185	* If this callback cannot block, and invalidate_range_{start,end}
	186	* cannot block, mmu_notifier_ops.flags should have
	187	* MMU_INVALIDATE_DOES_NOT_BLOCK set.
0f0a327f JR	188	*/
	189	void (invalidate_range)(struct mmu_notifier mn, struct mm_struct *mm,
	190	unsigned long start, unsigned long end);
cddb8a5c AA	191	};
	192
	193	/*
	194	* The notifier chains are protected by mmap_sem and/or the reverse map
	195	* semaphores. Notifier chains are only changed when all reverse maps and
	196	* the mmap_sem locks are taken.
	197	*
	198	* Therefore notifier chains can only be traversed when either
	199	*
	200	* 1. mmap_sem is held.
c8c06efa	201	* 2. One of the reverse map locks is held (i_mmap_rwsem or anon_vma->rwsem).
cddb8a5c AA	202	* 3. No other concurrent thread can access the list (release)
	203	*/
	204	struct mmu_notifier {
	205	struct hlist_node hlist;
	206	const struct mmu_notifier_ops *ops;
	207	};
	208
	209	static inline int mm_has_notifiers(struct mm_struct *mm)
	210	{
	211	return unlikely(mm->mmu_notifier_mm);
	212	}
	213
	214	extern int mmu_notifier_register(struct mmu_notifier *mn,
	215	struct mm_struct *mm);
	216	extern int __mmu_notifier_register(struct mmu_notifier *mn,
	217	struct mm_struct *mm);
	218	extern void mmu_notifier_unregister(struct mmu_notifier *mn,
	219	struct mm_struct *mm);
b972216e PZ	220	extern void mmu_notifier_unregister_no_release(struct mmu_notifier *mn,
b972216e PZ	221	struct mm_struct *mm);
cddb8a5c AA	222	extern void __mmu_notifier_mm_destroy(struct mm_struct *mm);
	223	extern void __mmu_notifier_release(struct mm_struct *mm);
	224	extern int __mmu_notifier_clear_flush_young(struct mm_struct *mm,
57128468 ALC	225	unsigned long start,
57128468 ALC	226	unsigned long end);
1d7715c6 VD	227	extern int __mmu_notifier_clear_young(struct mm_struct *mm,
	228	unsigned long start,
	229	unsigned long end);
8ee53820 AA	230	extern int __mmu_notifier_test_young(struct mm_struct *mm,
8ee53820 AA	231	unsigned long address);
828502d3 IE	232	extern void __mmu_notifier_change_pte(struct mm_struct *mm,
828502d3 IE	233	unsigned long address, pte_t pte);
93065ac7 MH	234	extern int __mmu_notifier_invalidate_range_start(struct mm_struct *mm,
	235	unsigned long start, unsigned long end,
	236	bool blockable);
cddb8a5c	237	extern void __mmu_notifier_invalidate_range_end(struct mm_struct *mm,
4645b9fe JG	238	unsigned long start, unsigned long end,
4645b9fe JG	239	bool only_end);
0f0a327f JR	240	extern void __mmu_notifier_invalidate_range(struct mm_struct *mm,
0f0a327f JR	241	unsigned long start, unsigned long end);
5ff7091f	242	extern bool mm_has_blockable_invalidate_notifiers(struct mm_struct *mm);
cddb8a5c AA	243
	244	static inline void mmu_notifier_release(struct mm_struct *mm)
	245	{
	246	if (mm_has_notifiers(mm))
	247	__mmu_notifier_release(mm);
	248	}
	249
	250	static inline int mmu_notifier_clear_flush_young(struct mm_struct *mm,
57128468 ALC	251	unsigned long start,
57128468 ALC	252	unsigned long end)
cddb8a5c AA	253	{
cddb8a5c AA	254	if (mm_has_notifiers(mm))
57128468	255	return __mmu_notifier_clear_flush_young(mm, start, end);
cddb8a5c AA	256	return 0;
	257	}
	258
1d7715c6 VD	259	static inline int mmu_notifier_clear_young(struct mm_struct *mm,
	260	unsigned long start,
	261	unsigned long end)
	262	{
	263	if (mm_has_notifiers(mm))
	264	return __mmu_notifier_clear_young(mm, start, end);
	265	return 0;
	266	}
	267
8ee53820 AA	268	static inline int mmu_notifier_test_young(struct mm_struct *mm,
	269	unsigned long address)
	270	{
	271	if (mm_has_notifiers(mm))
	272	return __mmu_notifier_test_young(mm, address);
	273	return 0;
	274	}
	275
828502d3 IE	276	static inline void mmu_notifier_change_pte(struct mm_struct *mm,
	277	unsigned long address, pte_t pte)
	278	{
	279	if (mm_has_notifiers(mm))
	280	__mmu_notifier_change_pte(mm, address, pte);
	281	}
	282
cddb8a5c AA	283	static inline void mmu_notifier_invalidate_range_start(struct mm_struct *mm,
	284	unsigned long start, unsigned long end)
	285	{
	286	if (mm_has_notifiers(mm))
93065ac7 MH	287	__mmu_notifier_invalidate_range_start(mm, start, end, true);
	288	}
	289
	290	static inline int mmu_notifier_invalidate_range_start_nonblock(struct mm_struct *mm,
	291	unsigned long start, unsigned long end)
	292	{
	293	if (mm_has_notifiers(mm))
	294	return __mmu_notifier_invalidate_range_start(mm, start, end, false);
	295	return 0;
cddb8a5c AA	296	}
	297
	298	static inline void mmu_notifier_invalidate_range_end(struct mm_struct *mm,
	299	unsigned long start, unsigned long end)
	300	{
	301	if (mm_has_notifiers(mm))
4645b9fe JG	302	__mmu_notifier_invalidate_range_end(mm, start, end, false);
	303	}
	304
	305	static inline void mmu_notifier_invalidate_range_only_end(struct mm_struct *mm,
	306	unsigned long start, unsigned long end)
	307	{
	308	if (mm_has_notifiers(mm))
	309	__mmu_notifier_invalidate_range_end(mm, start, end, true);
cddb8a5c AA	310	}
cddb8a5c AA	311
1897bdc4 JR	312	static inline void mmu_notifier_invalidate_range(struct mm_struct *mm,
	313	unsigned long start, unsigned long end)
	314	{
0f0a327f JR	315	if (mm_has_notifiers(mm))
0f0a327f JR	316	__mmu_notifier_invalidate_range(mm, start, end);
1897bdc4 JR	317	}
1897bdc4 JR	318
cddb8a5c AA	319	static inline void mmu_notifier_mm_init(struct mm_struct *mm)
	320	{
	321	mm->mmu_notifier_mm = NULL;
	322	}
	323
	324	static inline void mmu_notifier_mm_destroy(struct mm_struct *mm)
	325	{
	326	if (mm_has_notifiers(mm))
	327	__mmu_notifier_mm_destroy(mm);
	328	}
	329
cddb8a5c AA	330	#define ptep_clear_flush_young_notify(__vma, __address, __ptep) \
	331	({ \
	332	int __young; \
	333	struct vm_area_struct *___vma = __vma; \
	334	unsigned long ___address = __address; \
	335	__young = ptep_clear_flush_young(___vma, ___address, __ptep); \
	336	__young \|= mmu_notifier_clear_flush_young(___vma->vm_mm, \
57128468 ALC	337	___address, \
	338	___address + \
	339	PAGE_SIZE); \
cddb8a5c AA	340	__young; \
	341	})
	342
91a4ee26 AA	343	#define pmdp_clear_flush_young_notify(__vma, __address, __pmdp) \
	344	({ \
	345	int __young; \
	346	struct vm_area_struct *___vma = __vma; \
	347	unsigned long ___address = __address; \
	348	__young = pmdp_clear_flush_young(___vma, ___address, __pmdp); \
	349	__young \|= mmu_notifier_clear_flush_young(___vma->vm_mm, \
57128468 ALC	350	___address, \
	351	___address + \
	352	PMD_SIZE); \
91a4ee26 AA	353	__young; \
	354	})
	355
1d7715c6 VD	356	#define ptep_clear_young_notify(__vma, __address, __ptep) \
	357	({ \
	358	int __young; \
	359	struct vm_area_struct *___vma = __vma; \
	360	unsigned long ___address = __address; \
	361	__young = ptep_test_and_clear_young(___vma, ___address, __ptep);\
	362	__young \|= mmu_notifier_clear_young(___vma->vm_mm, ___address, \
	363	___address + PAGE_SIZE); \
	364	__young; \
	365	})
	366
	367	#define pmdp_clear_young_notify(__vma, __address, __pmdp) \
	368	({ \
	369	int __young; \
	370	struct vm_area_struct *___vma = __vma; \
	371	unsigned long ___address = __address; \
	372	__young = pmdp_test_and_clear_young(___vma, ___address, __pmdp);\
	373	__young \|= mmu_notifier_clear_young(___vma->vm_mm, ___address, \
	374	___address + PMD_SIZE); \
	375	__young; \
	376	})
	377
34ee645e JR	378	#define ptep_clear_flush_notify(__vma, __address, __ptep) \
	379	({ \
	380	unsigned long ___addr = __address & PAGE_MASK; \
	381	struct mm_struct *___mm = (__vma)->vm_mm; \
	382	pte_t ___pte; \
	383	\
	384	___pte = ptep_clear_flush(__vma, __address, __ptep); \
	385	mmu_notifier_invalidate_range(___mm, ___addr, \
	386	___addr + PAGE_SIZE); \
	387	\
	388	___pte; \
	389	})
	390
8809aa2d	391	#define pmdp_huge_clear_flush_notify(__vma, __haddr, __pmd) \
34ee645e JR	392	({ \
	393	unsigned long ___haddr = __haddr & HPAGE_PMD_MASK; \
	394	struct mm_struct *___mm = (__vma)->vm_mm; \
	395	pmd_t ___pmd; \
	396	\
8809aa2d	397	___pmd = pmdp_huge_clear_flush(__vma, __haddr, __pmd); \
34ee645e JR	398	mmu_notifier_invalidate_range(___mm, ___haddr, \
	399	___haddr + HPAGE_PMD_SIZE); \
	400	\
	401	___pmd; \
	402	})
	403
a00cc7d9 MW	404	#define pudp_huge_clear_flush_notify(__vma, __haddr, __pud) \
	405	({ \
	406	unsigned long ___haddr = __haddr & HPAGE_PUD_MASK; \
	407	struct mm_struct *___mm = (__vma)->vm_mm; \
	408	pud_t ___pud; \
	409	\
	410	___pud = pudp_huge_clear_flush(__vma, __haddr, __pud); \
	411	mmu_notifier_invalidate_range(___mm, ___haddr, \
	412	___haddr + HPAGE_PUD_SIZE); \
	413	\
	414	___pud; \
	415	})
	416
48af0d7c XG	417	/*
	418	* set_pte_at_notify() sets the pte _after_ running the notifier.
	419	* This is safe to start by updating the secondary MMUs, because the primary MMU
	420	* pte invalidate must have already happened with a ptep_clear_flush() before
	421	* set_pte_at_notify() has been invoked. Updating the secondary MMUs first is
	422	* required when we change both the protection of the mapping from read-only to
	423	* read-write and the pfn (like during copy on write page faults). Otherwise the
	424	* old page would remain mapped readonly in the secondary MMUs after the new
	425	* page is already writable by some CPU through the primary MMU.
	426	*/
828502d3 IE	427	#define set_pte_at_notify(__mm, __address, __ptep, __pte) \
	428	({ \
	429	struct mm_struct *___mm = __mm; \
	430	unsigned long ___address = __address; \
	431	pte_t ___pte = __pte; \
	432	\
828502d3	433	mmu_notifier_change_pte(___mm, ___address, ___pte); \
48af0d7c	434	set_pte_at(___mm, ___address, __ptep, ___pte); \
828502d3 IE	435	})
828502d3 IE	436
b972216e PZ	437	extern void mmu_notifier_call_srcu(struct rcu_head *rcu,
	438	void (func)(struct rcu_head rcu));
	439	extern void mmu_notifier_synchronize(void);
	440
cddb8a5c AA	441	#else /* CONFIG_MMU_NOTIFIER */
cddb8a5c AA	442
4d4bbd85 MH	443	static inline int mm_has_notifiers(struct mm_struct *mm)
	444	{
	445	return 0;
	446	}
	447
cddb8a5c AA	448	static inline void mmu_notifier_release(struct mm_struct *mm)
	449	{
	450	}
	451
	452	static inline int mmu_notifier_clear_flush_young(struct mm_struct *mm,
57128468 ALC	453	unsigned long start,
57128468 ALC	454	unsigned long end)
8ee53820 AA	455	{
	456	return 0;
	457	}
	458
	459	static inline int mmu_notifier_test_young(struct mm_struct *mm,
	460	unsigned long address)
cddb8a5c AA	461	{
	462	return 0;
	463	}
	464
828502d3 IE	465	static inline void mmu_notifier_change_pte(struct mm_struct *mm,
	466	unsigned long address, pte_t pte)
	467	{
	468	}
	469
cddb8a5c AA	470	static inline void mmu_notifier_invalidate_range_start(struct mm_struct *mm,
	471	unsigned long start, unsigned long end)
	472	{
	473	}
	474
93065ac7 MH	475	static inline int mmu_notifier_invalidate_range_start_nonblock(struct mm_struct *mm,
	476	unsigned long start, unsigned long end)
	477	{
	478	return 0;
	479	}
	480
cddb8a5c AA	481	static inline void mmu_notifier_invalidate_range_end(struct mm_struct *mm,
	482	unsigned long start, unsigned long end)
	483	{
	484	}
	485
4645b9fe JG	486	static inline void mmu_notifier_invalidate_range_only_end(struct mm_struct *mm,
	487	unsigned long start, unsigned long end)
	488	{
	489	}
	490
1897bdc4 JR	491	static inline void mmu_notifier_invalidate_range(struct mm_struct *mm,
	492	unsigned long start, unsigned long end)
	493	{
	494	}
	495
5ff7091f DR	496	static inline bool mm_has_blockable_invalidate_notifiers(struct mm_struct *mm)
	497	{
	498	return false;
	499	}
	500
cddb8a5c AA	501	static inline void mmu_notifier_mm_init(struct mm_struct *mm)
	502	{
	503	}
	504
	505	static inline void mmu_notifier_mm_destroy(struct mm_struct *mm)
	506	{
	507	}
	508
	509	#define ptep_clear_flush_young_notify ptep_clear_flush_young
91a4ee26	510	#define pmdp_clear_flush_young_notify pmdp_clear_flush_young
33c3fc71 VD	511	#define ptep_clear_young_notify ptep_test_and_clear_young
33c3fc71 VD	512	#define pmdp_clear_young_notify pmdp_test_and_clear_young
34ee645e	513	#define ptep_clear_flush_notify ptep_clear_flush
8809aa2d	514	#define pmdp_huge_clear_flush_notify pmdp_huge_clear_flush
a00cc7d9	515	#define pudp_huge_clear_flush_notify pudp_huge_clear_flush
828502d3	516	#define set_pte_at_notify set_pte_at
cddb8a5c AA	517
	518	#endif /* CONFIG_MMU_NOTIFIER */
	519
	520	#endif /* _LINUX_MMU_NOTIFIER_H */