tlb.h 7.56 KB
Newer Older
1
/* include/asm-generic/tlb.h
Linus Torvalds's avatar
Linus Torvalds committed
2 3 4 5 6 7
 *
 *	Generic TLB shootdown code
 *
 * Copyright 2001 Red Hat, Inc.
 * Based on code from mm/memory.c Copyright Linus Torvalds and others.
 *
8
 * Copyright 2011 Red Hat, Inc., Peter Zijlstra
Peter Zijlstra's avatar
Peter Zijlstra committed
9
 *
Linus Torvalds's avatar
Linus Torvalds committed
10 11 12 13 14 15 16 17 18
 * 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>
19
#include <asm/pgalloc.h>
Linus Torvalds's avatar
Linus Torvalds committed
20 21
#include <asm/tlbflush.h>

22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64
#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

Peter Zijlstra's avatar
Peter Zijlstra committed
65 66 67 68 69 70
/*
 * 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

71 72 73 74 75 76 77 78 79 80
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 *))

81 82 83 84 85 86 87 88
/*
 * Limit the maximum number of mmu_gather batches to reduce a risk of soft
 * lockups for non-preemptible kernels on huge machines when a lot of memory
 * is zapped during unmapping.
 * 10K pages freed at once should be safe even without a preemption point.
 */
#define MAX_GATHER_BATCH_COUNT	(10000UL/MAX_GATHER_BATCH)

Linus Torvalds's avatar
Linus Torvalds committed
89
/* struct mmu_gather is an opaque type used by the mm code for passing around
90
 * any data needed by arch specific code for tlb_remove_page.
Linus Torvalds's avatar
Linus Torvalds committed
91 92 93
 */
struct mmu_gather {
	struct mm_struct	*mm;
94 95 96
#ifdef CONFIG_HAVE_RCU_TABLE_FREE
	struct mmu_table_batch	*batch;
#endif
97 98
	unsigned long		start;
	unsigned long		end;
99 100
	/* we are in the middle of an operation to clear
	 * a full mm and can make some optimizations */
101
	unsigned int		fullmm : 1,
102 103 104
	/* we have performed an operation which
	 * requires a complete flush of the tlb */
				need_flush_all : 1;
105 106 107 108

	struct mmu_gather_batch *active;
	struct mmu_gather_batch	local;
	struct page		*__pages[MMU_GATHER_BUNDLE];
109
	unsigned int		batch_count;
110 111 112 113 114
	/*
	 * __tlb_adjust_range  will track the new addr here,
	 * that that we can adjust the range after the flush
	 */
	unsigned long addr;
115
	int page_size;
Linus Torvalds's avatar
Linus Torvalds committed
116 117
};

118
#define HAVE_GENERIC_MMU_GATHER
119

120
void tlb_gather_mmu(struct mmu_gather *tlb, struct mm_struct *mm, unsigned long start, unsigned long end);
121
void tlb_flush_mmu(struct mmu_gather *tlb);
122 123
void tlb_finish_mmu(struct mmu_gather *tlb, unsigned long start,
							unsigned long end);
124 125
extern bool __tlb_remove_page_size(struct mmu_gather *tlb, struct page *page,
				   int page_size);
Linus Torvalds's avatar
Linus Torvalds committed
126

127 128 129 130 131
static inline void __tlb_adjust_range(struct mmu_gather *tlb,
				      unsigned long address)
{
	tlb->start = min(tlb->start, address);
	tlb->end = max(tlb->end, address + PAGE_SIZE);
132 133 134 135 136 137
	/*
	 * Track the last address with which we adjusted the range. This
	 * will be used later to adjust again after a mmu_flush due to
	 * failed __tlb_remove_page
	 */
	tlb->addr = address;
138 139 140 141
}

static inline void __tlb_reset_range(struct mmu_gather *tlb)
{
142 143 144 145 146 147
	if (tlb->fullmm) {
		tlb->start = tlb->end = ~0;
	} else {
		tlb->start = TASK_SIZE;
		tlb->end = 0;
	}
148 149
}

150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165
static inline void tlb_remove_page_size(struct mmu_gather *tlb,
					struct page *page, int page_size)
{
	if (__tlb_remove_page_size(tlb, page, page_size)) {
		tlb_flush_mmu(tlb);
		tlb->page_size = page_size;
		__tlb_adjust_range(tlb, tlb->addr);
		__tlb_remove_page_size(tlb, page, page_size);
	}
}

static bool __tlb_remove_page(struct mmu_gather *tlb, struct page *page)
{
	return __tlb_remove_page_size(tlb, page, PAGE_SIZE);
}

166 167 168 169 170 171
/* 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)
{
172
	return tlb_remove_page_size(tlb, page, PAGE_SIZE);
173 174 175 176 177 178
}

static inline bool __tlb_remove_pte_page(struct mmu_gather *tlb, struct page *page)
{
	/* active->nr should be zero when we call this */
	VM_BUG_ON_PAGE(tlb->active->nr, page);
179
	tlb->page_size = PAGE_SIZE;
180 181 182 183
	__tlb_adjust_range(tlb, tlb->addr);
	return __tlb_remove_page(tlb, page);
}

184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208
/*
 * In the case of tlb vma handling, we can optimise these away in the
 * case where we're doing a full MM flush.  When we're doing a munmap,
 * the vmas are adjusted to only cover the region to be torn down.
 */
#ifndef tlb_start_vma
#define tlb_start_vma(tlb, vma) do { } while (0)
#endif

#define __tlb_end_vma(tlb, vma)					\
	do {							\
		if (!tlb->fullmm && tlb->end) {			\
			tlb_flush(tlb);				\
			__tlb_reset_range(tlb);			\
		}						\
	} while (0)

#ifndef tlb_end_vma
#define tlb_end_vma	__tlb_end_vma
#endif

#ifndef __tlb_remove_tlb_entry
#define __tlb_remove_tlb_entry(tlb, ptep, address) do { } while (0)
#endif

Linus Torvalds's avatar
Linus Torvalds committed
209 210 211
/**
 * tlb_remove_tlb_entry - remember a pte unmapping for later tlb invalidation.
 *
212 213 214
 * Record the fact that pte's were really unmapped by updating the range,
 * so we can later optimise away the tlb invalidate.   This helps when
 * userspace is unmapping already-unmapped pages, which happens quite a lot.
Linus Torvalds's avatar
Linus Torvalds committed
215 216 217
 */
#define tlb_remove_tlb_entry(tlb, ptep, address)		\
	do {							\
218
		__tlb_adjust_range(tlb, address);		\
Linus Torvalds's avatar
Linus Torvalds committed
219 220 221
		__tlb_remove_tlb_entry(tlb, ptep, address);	\
	} while (0)

222 223 224 225 226 227 228 229 230 231
/**
 * tlb_remove_pmd_tlb_entry - remember a pmd mapping for later tlb invalidation
 * This is a nop so far, because only x86 needs it.
 */
#ifndef __tlb_remove_pmd_tlb_entry
#define __tlb_remove_pmd_tlb_entry(tlb, pmdp, address) do {} while (0)
#endif

#define tlb_remove_pmd_tlb_entry(tlb, pmdp, address)		\
	do {							\
232
		__tlb_adjust_range(tlb, address);		\
233 234 235
		__tlb_remove_pmd_tlb_entry(tlb, pmdp, address);	\
	} while (0)

236
#define pte_free_tlb(tlb, ptep, address)			\
Linus Torvalds's avatar
Linus Torvalds committed
237
	do {							\
238
		__tlb_adjust_range(tlb, address);		\
239
		__pte_free_tlb(tlb, ptep, address);		\
Linus Torvalds's avatar
Linus Torvalds committed
240 241 242
	} while (0)

#ifndef __ARCH_HAS_4LEVEL_HACK
243
#define pud_free_tlb(tlb, pudp, address)			\
Linus Torvalds's avatar
Linus Torvalds committed
244
	do {							\
245
		__tlb_adjust_range(tlb, address);		\
246
		__pud_free_tlb(tlb, pudp, address);		\
Linus Torvalds's avatar
Linus Torvalds committed
247 248 249
	} while (0)
#endif

250
#define pmd_free_tlb(tlb, pmdp, address)			\
Linus Torvalds's avatar
Linus Torvalds committed
251
	do {							\
252
		__tlb_adjust_range(tlb, address);		\
253
		__pmd_free_tlb(tlb, pmdp, address);		\
Linus Torvalds's avatar
Linus Torvalds committed
254 255 256 257 258
	} while (0)

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

#endif /* _ASM_GENERIC__TLB_H */