[mirror_ubuntu-focal-kernel.git] / include / asm-generic / tlb.h

/* include/asm-generic/tlb.h
 *
 *	Generic TLB shootdown code
 *
 * Copyright 2001 Red Hat, Inc.
 * Based on code from mm/memory.c Copyright Linus Torvalds and others.
 *
 * Copyright 2011 Red Hat, Inc., Peter Zijlstra <pzijlstr@redhat.com>
 *
 * This program is free software; you can redistribute it and/or
 * modify it under the terms of the GNU General Public License
 * as published by the Free Software Foundation; either version
 * 2 of the License, or (at your option) any later version.
 */
#ifndef _ASM_GENERIC__TLB_H
#define _ASM_GENERIC__TLB_H

#include <linux/swap.h>
#include <asm/pgalloc.h>
#include <asm/tlbflush.h>

#ifdef CONFIG_HAVE_RCU_TABLE_FREE
/*
 * Semi RCU freeing of the page directories.
 *
 * This is needed by some architectures to implement software pagetable walkers.
 *
 * gup_fast() and other software pagetable walkers do a lockless page-table
 * walk and therefore needs some synchronization with the freeing of the page
 * directories. The chosen means to accomplish that is by disabling IRQs over
 * the walk.
 *
 * Architectures that use IPIs to flush TLBs will then automagically DTRT,
 * since we unlink the page, flush TLBs, free the page. Since the disabling of
 * IRQs delays the completion of the TLB flush we can never observe an already
 * freed page.
 *
 * Architectures that do not have this (PPC) need to delay the freeing by some
 * other means, this is that means.
 *
 * What we do is batch the freed directory pages (tables) and RCU free them.
 * We use the sched RCU variant, as that guarantees that IRQ/preempt disabling
 * holds off grace periods.
 *
 * However, in order to batch these pages we need to allocate storage, this
 * allocation is deep inside the MM code and can thus easily fail on memory
 * pressure. To guarantee progress we fall back to single table freeing, see
 * the implementation of tlb_remove_table_one().
 *
 */
struct mmu_table_batch {
	struct rcu_head		rcu;
	unsigned int		nr;
	void			*tables[0];
};

#define MAX_TABLE_BATCH		\
	((PAGE_SIZE - sizeof(struct mmu_table_batch)) / sizeof(void *))

extern void tlb_table_flush(struct mmu_gather *tlb);
extern void tlb_remove_table(struct mmu_gather *tlb, void *table);

#endif

/*
 * If we can't allocate a page to make a big batch of page pointers
 * to work on, then just handle a few from the on-stack structure.
 */
#define MMU_GATHER_BUNDLE	8

struct mmu_gather_batch {
	struct mmu_gather_batch	*next;
	unsigned int		nr;
	unsigned int		max;
	struct page		*pages[0];
};

#define MAX_GATHER_BATCH	\
	((PAGE_SIZE - sizeof(struct mmu_gather_batch)) / sizeof(void *))

/* struct mmu_gather is an opaque type used by the mm code for passing around
 * any data needed by arch specific code for tlb_remove_page.
 */
struct mmu_gather {
	struct mm_struct	*mm;
#ifdef CONFIG_HAVE_RCU_TABLE_FREE
	struct mmu_table_batch	*batch;
#endif
	unsigned int		need_flush : 1,	/* Did free PTEs */
				fast_mode  : 1; /* No batching   */

	unsigned int		fullmm;

	struct mmu_gather_batch *active;
	struct mmu_gather_batch	local;
	struct page		*__pages[MMU_GATHER_BUNDLE];
};

/*
 * For UP we don't need to worry about TLB flush
 * and page free order so much..
 */
#ifdef CONFIG_SMP
  #define tlb_fast_mode(tlb) (tlb->fast_mode)
#else
  #define tlb_fast_mode(tlb) 1
#endif

static inline int tlb_next_batch(struct mmu_gather *tlb)
{
	struct mmu_gather_batch *batch;

	batch = tlb->active;
	if (batch->next) {
		tlb->active = batch->next;
		return 1;
	}

	batch = (void *)__get_free_pages(GFP_NOWAIT | __GFP_NOWARN, 0);
	if (!batch)
		return 0;

	batch->next = NULL;
	batch->nr   = 0;
	batch->max  = MAX_GATHER_BATCH;

	tlb->active->next = batch;
	tlb->active = batch;

	return 1;
}

/* tlb_gather_mmu
 *	Called to initialize an (on-stack) mmu_gather structure for page-table
 *	tear-down from @mm. The @fullmm argument is used when @mm is without
 *	users and we're going to destroy the full address space (exit/execve).
 */
static inline void
tlb_gather_mmu(struct mmu_gather *tlb, struct mm_struct *mm, bool fullmm)
{
	tlb->mm = mm;

	tlb->fullmm     = fullmm;
	tlb->need_flush = 0;
	tlb->fast_mode  = (num_possible_cpus() == 1);
	tlb->local.next = NULL;
	tlb->local.nr   = 0;
	tlb->local.max  = ARRAY_SIZE(tlb->__pages);
	tlb->active     = &tlb->local;

#ifdef CONFIG_HAVE_RCU_TABLE_FREE
	tlb->batch = NULL;
#endif
}

static inline void
tlb_flush_mmu(struct mmu_gather *tlb)
{
	struct mmu_gather_batch *batch;

	if (!tlb->need_flush)
		return;
	tlb->need_flush = 0;
	tlb_flush(tlb);
#ifdef CONFIG_HAVE_RCU_TABLE_FREE
	tlb_table_flush(tlb);
#endif

	if (tlb_fast_mode(tlb))
		return;

	for (batch = &tlb->local; batch; batch = batch->next) {
		free_pages_and_swap_cache(batch->pages, batch->nr);
		batch->nr = 0;
	}
	tlb->active = &tlb->local;
}

/* tlb_finish_mmu
 *	Called at the end of the shootdown operation to free up any resources
 *	that were required.
 */
static inline void
tlb_finish_mmu(struct mmu_gather *tlb, unsigned long start, unsigned long end)
{
	struct mmu_gather_batch *batch, *next;

	tlb_flush_mmu(tlb);

	/* keep the page table cache within bounds */
	check_pgt_cache();

	for (batch = tlb->local.next; batch; batch = next) {
		next = batch->next;
		free_pages((unsigned long)batch, 0);
	}
	tlb->local.next = NULL;
}

/* __tlb_remove_page
 *	Must perform the equivalent to __free_pte(pte_get_and_clear(ptep)), while
 *	handling the additional races in SMP caused by other CPUs caching valid
 *	mappings in their TLBs. Returns the number of free page slots left.
 *	When out of page slots we must call tlb_flush_mmu().
 */
static inline int __tlb_remove_page(struct mmu_gather *tlb, struct page *page)
{
	struct mmu_gather_batch *batch;

	tlb->need_flush = 1;

	if (tlb_fast_mode(tlb)) {
		free_page_and_swap_cache(page);
		return 1; /* avoid calling tlb_flush_mmu() */
	}

	batch = tlb->active;
	batch->pages[batch->nr++] = page;
	VM_BUG_ON(batch->nr > batch->max);
	if (batch->nr == batch->max) {
		if (!tlb_next_batch(tlb))
			return 0;
	}

	return batch->max - batch->nr;
}

/* tlb_remove_page
 *	Similar to __tlb_remove_page but will call tlb_flush_mmu() itself when
 *	required.
 */
static inline void tlb_remove_page(struct mmu_gather *tlb, struct page *page)
{
	if (!__tlb_remove_page(tlb, page))
		tlb_flush_mmu(tlb);
}

/**
 * tlb_remove_tlb_entry - remember a pte unmapping for later tlb invalidation.
 *
 * Record the fact that pte's were really umapped in ->need_flush, so we can
 * later optimise away the tlb invalidate.   This helps when userspace is
 * unmapping already-unmapped pages, which happens quite a lot.
 */
#define tlb_remove_tlb_entry(tlb, ptep, address)		\
	do {							\
		tlb->need_flush = 1;				\
		__tlb_remove_tlb_entry(tlb, ptep, address);	\
	} while (0)

#define pte_free_tlb(tlb, ptep, address)			\
	do {							\
		tlb->need_flush = 1;				\
		__pte_free_tlb(tlb, ptep, address);		\
	} while (0)

#ifndef __ARCH_HAS_4LEVEL_HACK
#define pud_free_tlb(tlb, pudp, address)			\
	do {							\
		tlb->need_flush = 1;				\
		__pud_free_tlb(tlb, pudp, address);		\
	} while (0)
#endif

#define pmd_free_tlb(tlb, pmdp, address)			\
	do {							\
		tlb->need_flush = 1;				\
		__pmd_free_tlb(tlb, pmdp, address);		\
	} while (0)

#define tlb_migrate_finish(mm) do {} while (0)

#endif /* _ASM_GENERIC__TLB_H */
Commit	Line	Data
f30c2269	1	/* include/asm-generic/tlb.h
1da177e4 LT	2	*
	3	* Generic TLB shootdown code
	4	*
	5	* Copyright 2001 Red Hat, Inc.
	6	* Based on code from mm/memory.c Copyright Linus Torvalds and others.
	7	*
d16dfc55 PZ	8	* Copyright 2011 Red Hat, Inc., Peter Zijlstra <pzijlstr@redhat.com>
d16dfc55 PZ	9	*
1da177e4 LT	10	* This program is free software; you can redistribute it and/or
	11	* modify it under the terms of the GNU General Public License
	12	* as published by the Free Software Foundation; either version
	13	* 2 of the License, or (at your option) any later version.
	14	*/
	15	#ifndef _ASM_GENERIC__TLB_H
	16	#define _ASM_GENERIC__TLB_H
	17
1da177e4	18	#include <linux/swap.h>
62152d0e	19	#include <asm/pgalloc.h>
1da177e4 LT	20	#include <asm/tlbflush.h>
1da177e4 LT	21
26723911 PZ	22	#ifdef CONFIG_HAVE_RCU_TABLE_FREE
	23	/*
	24	* Semi RCU freeing of the page directories.
	25	*
	26	* This is needed by some architectures to implement software pagetable walkers.
	27	*
	28	* gup_fast() and other software pagetable walkers do a lockless page-table
	29	* walk and therefore needs some synchronization with the freeing of the page
	30	* directories. The chosen means to accomplish that is by disabling IRQs over
	31	* the walk.
	32	*
	33	* Architectures that use IPIs to flush TLBs will then automagically DTRT,
	34	* since we unlink the page, flush TLBs, free the page. Since the disabling of
	35	* IRQs delays the completion of the TLB flush we can never observe an already
	36	* freed page.
	37	*
	38	* Architectures that do not have this (PPC) need to delay the freeing by some
	39	* other means, this is that means.
	40	*
	41	* What we do is batch the freed directory pages (tables) and RCU free them.
	42	* We use the sched RCU variant, as that guarantees that IRQ/preempt disabling
	43	* holds off grace periods.
	44	*
	45	* However, in order to batch these pages we need to allocate storage, this
	46	* allocation is deep inside the MM code and can thus easily fail on memory
	47	* pressure. To guarantee progress we fall back to single table freeing, see
	48	* the implementation of tlb_remove_table_one().
	49	*
	50	*/
	51	struct mmu_table_batch {
	52	struct rcu_head rcu;
	53	unsigned int nr;
	54	void *tables[0];
	55	};
	56
	57	#define MAX_TABLE_BATCH \
	58	((PAGE_SIZE - sizeof(struct mmu_table_batch)) / sizeof(void *))
	59
	60	extern void tlb_table_flush(struct mmu_gather *tlb);
	61	extern void tlb_remove_table(struct mmu_gather tlb, void table);
	62
	63	#endif
	64
d16dfc55 PZ	65	/*
	66	* If we can't allocate a page to make a big batch of page pointers
	67	* to work on, then just handle a few from the on-stack structure.
	68	*/
	69	#define MMU_GATHER_BUNDLE 8
	70
e303297e PZ	71	struct mmu_gather_batch {
	72	struct mmu_gather_batch *next;
	73	unsigned int nr;
	74	unsigned int max;
	75	struct page *pages[0];
	76	};
	77
	78	#define MAX_GATHER_BATCH \
	79	((PAGE_SIZE - sizeof(struct mmu_gather_batch)) / sizeof(void *))
	80
1da177e4	81	/* struct mmu_gather is an opaque type used by the mm code for passing around
15a23ffa	82	* any data needed by arch specific code for tlb_remove_page.
1da177e4 LT	83	*/
	84	struct mmu_gather {
	85	struct mm_struct *mm;
26723911 PZ	86	#ifdef CONFIG_HAVE_RCU_TABLE_FREE
	87	struct mmu_table_batch *batch;
	88	#endif
e303297e PZ	89	unsigned int need_flush : 1, /* Did free PTEs */
	90	fast_mode : 1; /* No batching */
	91
	92	unsigned int fullmm;
	93
	94	struct mmu_gather_batch *active;
	95	struct mmu_gather_batch local;
	96	struct page *__pages[MMU_GATHER_BUNDLE];
1da177e4 LT	97	};
1da177e4 LT	98
e303297e PZ	99	/*
	100	* For UP we don't need to worry about TLB flush
	101	* and page free order so much..
	102	*/
	103	#ifdef CONFIG_SMP
	104	#define tlb_fast_mode(tlb) (tlb->fast_mode)
	105	#else
	106	#define tlb_fast_mode(tlb) 1
	107	#endif
	108
	109	static inline int tlb_next_batch(struct mmu_gather *tlb)
d16dfc55	110	{
e303297e	111	struct mmu_gather_batch *batch;
d16dfc55	112
e303297e PZ	113	batch = tlb->active;
	114	if (batch->next) {
	115	tlb->active = batch->next;
	116	return 1;
d16dfc55	117	}
e303297e PZ	118
	119	batch = (void *)__get_free_pages(GFP_NOWAIT \| __GFP_NOWARN, 0);
	120	if (!batch)
	121	return 0;
	122
	123	batch->next = NULL;
	124	batch->nr = 0;
	125	batch->max = MAX_GATHER_BATCH;
	126
	127	tlb->active->next = batch;
	128	tlb->active = batch;
	129
	130	return 1;
d16dfc55	131	}
1da177e4 LT	132
1da177e4 LT	133	/* tlb_gather_mmu
d16dfc55 PZ	134	* Called to initialize an (on-stack) mmu_gather structure for page-table
	135	* tear-down from @mm. The @fullmm argument is used when @mm is without
	136	* users and we're going to destroy the full address space (exit/execve).
1da177e4	137	*/
d16dfc55 PZ	138	static inline void
d16dfc55 PZ	139	tlb_gather_mmu(struct mmu_gather tlb, struct mm_struct mm, bool fullmm)
1da177e4	140	{
1da177e4 LT	141	tlb->mm = mm;
1da177e4 LT	142
e303297e PZ	143	tlb->fullmm = fullmm;
	144	tlb->need_flush = 0;
	145	tlb->fast_mode = (num_possible_cpus() == 1);
	146	tlb->local.next = NULL;
	147	tlb->local.nr = 0;
	148	tlb->local.max = ARRAY_SIZE(tlb->__pages);
	149	tlb->active = &tlb->local;
1da177e4	150
26723911 PZ	151	#ifdef CONFIG_HAVE_RCU_TABLE_FREE
26723911 PZ	152	tlb->batch = NULL;
d16dfc55	153	#endif
1da177e4 LT	154	}
	155
	156	static inline void
d16dfc55	157	tlb_flush_mmu(struct mmu_gather *tlb)
1da177e4	158	{
e303297e PZ	159	struct mmu_gather_batch *batch;
e303297e PZ	160
1da177e4 LT	161	if (!tlb->need_flush)
	162	return;
	163	tlb->need_flush = 0;
	164	tlb_flush(tlb);
26723911 PZ	165	#ifdef CONFIG_HAVE_RCU_TABLE_FREE
	166	tlb_table_flush(tlb);
	167	#endif
e303297e PZ	168
	169	if (tlb_fast_mode(tlb))
	170	return;
	171
	172	for (batch = &tlb->local; batch; batch = batch->next) {
	173	free_pages_and_swap_cache(batch->pages, batch->nr);
	174	batch->nr = 0;
1da177e4	175	}
e303297e	176	tlb->active = &tlb->local;
1da177e4 LT	177	}
	178
	179	/* tlb_finish_mmu
	180	* Called at the end of the shootdown operation to free up any resources
15a23ffa	181	* that were required.
1da177e4 LT	182	*/
	183	static inline void
	184	tlb_finish_mmu(struct mmu_gather *tlb, unsigned long start, unsigned long end)
	185	{
e303297e PZ	186	struct mmu_gather_batch batch, next;
e303297e PZ	187
d16dfc55	188	tlb_flush_mmu(tlb);
1da177e4 LT	189
	190	/* keep the page table cache within bounds */
	191	check_pgt_cache();
15a23ffa	192
e303297e PZ	193	for (batch = tlb->local.next; batch; batch = next) {
	194	next = batch->next;
	195	free_pages((unsigned long)batch, 0);
	196	}
	197	tlb->local.next = NULL;
1da177e4 LT	198	}
1da177e4 LT	199
d16dfc55	200	/* __tlb_remove_page
1da177e4 LT	201	* Must perform the equivalent to __free_pte(pte_get_and_clear(ptep)), while
1da177e4 LT	202	* handling the additional races in SMP caused by other CPUs caching valid
d16dfc55 PZ	203	* mappings in their TLBs. Returns the number of free page slots left.
d16dfc55 PZ	204	* When out of page slots we must call tlb_flush_mmu().
1da177e4	205	*/
d16dfc55	206	static inline int __tlb_remove_page(struct mmu_gather tlb, struct page page)
1da177e4	207	{
e303297e PZ	208	struct mmu_gather_batch *batch;
e303297e PZ	209
1da177e4	210	tlb->need_flush = 1;
e303297e	211
1da177e4 LT	212	if (tlb_fast_mode(tlb)) {
1da177e4 LT	213	free_page_and_swap_cache(page);
d16dfc55	214	return 1; /* avoid calling tlb_flush_mmu() */
1da177e4	215	}
d16dfc55	216
e303297e PZ	217	batch = tlb->active;
	218	batch->pages[batch->nr++] = page;
	219	VM_BUG_ON(batch->nr > batch->max);
	220	if (batch->nr == batch->max) {
	221	if (!tlb_next_batch(tlb))
	222	return 0;
	223	}
	224
	225	return batch->max - batch->nr;
d16dfc55 PZ	226	}
	227
	228	/* tlb_remove_page
	229	* Similar to __tlb_remove_page but will call tlb_flush_mmu() itself when
	230	* required.
	231	*/
	232	static inline void tlb_remove_page(struct mmu_gather tlb, struct page page)
	233	{
	234	if (!__tlb_remove_page(tlb, page))
	235	tlb_flush_mmu(tlb);
1da177e4 LT	236	}
	237
	238	/**
	239	* tlb_remove_tlb_entry - remember a pte unmapping for later tlb invalidation.
	240	*
	241	* Record the fact that pte's were really umapped in ->need_flush, so we can
	242	* later optimise away the tlb invalidate. This helps when userspace is
	243	* unmapping already-unmapped pages, which happens quite a lot.
	244	*/
	245	#define tlb_remove_tlb_entry(tlb, ptep, address) \
	246	do { \
	247	tlb->need_flush = 1; \
	248	__tlb_remove_tlb_entry(tlb, ptep, address); \
	249	} while (0)
	250
9e1b32ca	251	#define pte_free_tlb(tlb, ptep, address) \
1da177e4 LT	252	do { \
1da177e4 LT	253	tlb->need_flush = 1; \
9e1b32ca	254	__pte_free_tlb(tlb, ptep, address); \
1da177e4 LT	255	} while (0)
	256
	257	#ifndef __ARCH_HAS_4LEVEL_HACK
9e1b32ca	258	#define pud_free_tlb(tlb, pudp, address) \
1da177e4 LT	259	do { \
1da177e4 LT	260	tlb->need_flush = 1; \
9e1b32ca	261	__pud_free_tlb(tlb, pudp, address); \
1da177e4 LT	262	} while (0)
	263	#endif
	264
9e1b32ca	265	#define pmd_free_tlb(tlb, pmdp, address) \
1da177e4 LT	266	do { \
1da177e4 LT	267	tlb->need_flush = 1; \
9e1b32ca	268	__pmd_free_tlb(tlb, pmdp, address); \
1da177e4 LT	269	} while (0)
	270
	271	#define tlb_migrate_finish(mm) do {} while (0)
	272
	273	#endif /* _ASM_GENERIC__TLB_H */