summaryrefslogtreecommitdiffstats
path: root/arch/arm/mm/fault-armv.c
blob: 82df01a72f4a6181b2c7ee0dadfe08ba0e7b1b2d (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
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
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
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
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
/*
 *  linux/arch/arm/mm/fault-armv.c
 *
 *  Copyright (C) 1995  Linus Torvalds
 *  Modifications for ARM processor (c) 1995-2002 Russell King
 *
 * This program is free software; you can redistribute it and/or modify
 * it under the terms of the GNU General Public License version 2 as
 * published by the Free Software Foundation.
 */
#include <linux/module.h>
#include <linux/sched.h>
#include <linux/kernel.h>
#include <linux/mm.h>
#include <linux/bitops.h>
#include <linux/vmalloc.h>
#include <linux/init.h>
#include <linux/pagemap.h>

#include <asm/bugs.h>
#include <asm/cacheflush.h>
#include <asm/cachetype.h>
#include <asm/pgtable.h>
#include <asm/tlbflush.h>

#include "mm.h"

static unsigned long shared_pte_mask = L_PTE_MT_BUFFERABLE;

/*
 * We take the easy way out of this problem - we make the
 * PTE uncacheable.  However, we leave the write buffer on.
 *
 * Note that the pte lock held when calling update_mmu_cache must also
 * guard the pte (somewhere else in the same mm) that we modify here.
 * Therefore those configurations which might call adjust_pte (those
 * without CONFIG_CPU_CACHE_VIPT) cannot support split page_table_lock.
 */
static int do_adjust_pte(struct vm_area_struct *vma, unsigned long address,
	unsigned long pfn, pte_t *ptep)
{
	pte_t entry = *ptep;
	int ret;

	/*
	 * If this page is present, it's actually being shared.
	 */
	ret = pte_present(entry);

	/*
	 * If this page isn't present, or is already setup to
	 * fault (ie, is old), we can safely ignore any issues.
	 */
	if (ret && (pte_val(entry) & L_PTE_MT_MASK) != shared_pte_mask) {
		flush_cache_page(vma, address, pfn);
		outer_flush_range((pfn << PAGE_SHIFT),
				  (pfn << PAGE_SHIFT) + PAGE_SIZE);
		pte_val(entry) &= ~L_PTE_MT_MASK;
		pte_val(entry) |= shared_pte_mask;
		set_pte_at(vma->vm_mm, address, ptep, entry);
		flush_tlb_page(vma, address);
	}

	return ret;
}

static int adjust_pte(struct vm_area_struct *vma, unsigned long address,
	unsigned long pfn)
{
	spinlock_t *ptl;
	pgd_t *pgd;
	pmd_t *pmd;
	pte_t *pte;
	int ret;

	pgd = pgd_offset(vma->vm_mm, address);
	if (pgd_none_or_clear_bad(pgd))
		return 0;

	pmd = pmd_offset(pgd, address);
	if (pmd_none_or_clear_bad(pmd))
		return 0;

	/*
	 * This is called while another page table is mapped, so we
	 * must use the nested version.  This also means we need to
	 * open-code the spin-locking.
	 */
	ptl = pte_lockptr(vma->vm_mm, pmd);
	pte = pte_offset_map_nested(pmd, address);
	spin_lock(ptl);

	ret = do_adjust_pte(vma, address, pfn, pte);

	spin_unlock(ptl);
	pte_unmap_nested(pte);

	return ret;
}

static void
make_coherent(struct address_space *mapping, struct vm_area_struct *vma,
	unsigned long addr, pte_t *ptep, unsigned long pfn)
{
	struct mm_struct *mm = vma->vm_mm;
	struct vm_area_struct *mpnt;
	struct prio_tree_iter iter;
	unsigned long offset;
	pgoff_t pgoff;
	int aliases = 0;

	pgoff = vma->vm_pgoff + ((addr - vma->vm_start) >> PAGE_SHIFT);

	/*
	 * If we have any shared mappings that are in the same mm
	 * space, then we need to handle them specially to maintain
	 * cache coherency.
	 */
	flush_dcache_mmap_lock(mapping);
	vma_prio_tree_foreach(mpnt, &iter, &mapping->i_mmap, pgoff, pgoff) {
		/*
		 * If this VMA is not in our MM, we can ignore it.
		 * Note that we intentionally mask out the VMA
		 * that we are fixing up.
		 */
		if (mpnt->vm_mm != mm || mpnt == vma)
			continue;
		if (!(mpnt->vm_flags & VM_MAYSHARE))
			continue;
		offset = (pgoff - mpnt->vm_pgoff) << PAGE_SHIFT;
		aliases += adjust_pte(mpnt, mpnt->vm_start + offset, pfn);
	}
	flush_dcache_mmap_unlock(mapping);
	if (aliases)
		do_adjust_pte(vma, addr, pfn, ptep);
}

/*
 * Take care of architecture specific things when placing a new PTE into
 * a page table, or changing an existing PTE.  Basically, there are two
 * things that we need to take care of:
 *
 *  1. If PG_dcache_dirty is set for the page, we need to ensure
 *     that any cache entries for the kernels virtual memory
 *     range are written back to the page.
 *  2. If we have multiple shared mappings of the same space in
 *     an object, we need to deal with the cache aliasing issues.
 *
 * Note that the pte lock will be held.
 */
void update_mmu_cache(struct vm_area_struct *vma, unsigned long addr,
	pte_t *ptep)
{
	unsigned long pfn = pte_pfn(*ptep);
	struct address_space *mapping;
	struct page *page;

	if (!pfn_valid(pfn))
		return;

	/*
	 * The zero page is never written to, so never has any dirty
	 * cache lines, and therefore never needs to be flushed.
	 */
	page = pfn_to_page(pfn);
	if (page == ZERO_PAGE(0))
		return;

	mapping = page_mapping(page);
#ifndef CONFIG_SMP
	if (test_and_clear_bit(PG_dcache_dirty, &page->flags))
		__flush_dcache_page(mapping, page);
#endif
	if (mapping) {
		if (cache_is_vivt())
			make_coherent(mapping, vma, addr, ptep, pfn);
		else if (vma->vm_flags & VM_EXEC)
			__flush_icache_all();
	}
}

/*
 * Check whether the write buffer has physical address aliasing
 * issues.  If it has, we need to avoid them for the case where
 * we have several shared mappings of the same object in user
 * space.
 */
static int __init check_writebuffer(unsigned long *p1, unsigned long *p2)
{
	register unsigned long zero = 0, one = 1, val;

	local_irq_disable();
	mb();
	*p1 = one;
	mb();
	*p2 = zero;
	mb();
	val = *p1;
	mb();
	local_irq_enable();
	return val != zero;
}

void __init check_writebuffer_bugs(void)
{
	struct page *page;
	const char *reason;
	unsigned long v = 1;

	printk(KERN_INFO "CPU: Testing write buffer coherency: ");

	page = alloc_page(GFP_KERNEL);
	if (page) {
		unsigned long *p1, *p2;
		pgprot_t prot = __pgprot_modify(PAGE_KERNEL,
					L_PTE_MT_MASK, L_PTE_MT_BUFFERABLE);

		p1 = vmap(&page, 1, VM_IOREMAP, prot);
		p2 = vmap(&page, 1, VM_IOREMAP, prot);

		if (p1 && p2) {
			v = check_writebuffer(p1, p2);
			reason = "enabling work-around";
		} else {
			reason = "unable to map memory\n";
		}

		vunmap(p1);
		vunmap(p2);
		put_page(page);
	} else {
		reason = "unable to grab page\n";
	}

	if (v) {
		printk("failed, %s\n", reason);
		shared_pte_mask = L_PTE_MT_UNCACHED;
	} else {
		printk("ok\n");
	}
}