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
242
243
244
245
246
247
248
249
250
251
|
/*
* Microblaze support for cache consistent memory.
* Copyright (C) 2010 Michal Simek <monstr@monstr.eu>
* Copyright (C) 2010 PetaLogix
* Copyright (C) 2005 John Williams <jwilliams@itee.uq.edu.au>
*
* Based on PowerPC version derived from arch/arm/mm/consistent.c
* Copyright (C) 2001 Dan Malek (dmalek@jlc.net)
* Copyright (C) 2000 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/export.h>
#include <linux/signal.h>
#include <linux/sched.h>
#include <linux/kernel.h>
#include <linux/errno.h>
#include <linux/string.h>
#include <linux/types.h>
#include <linux/ptrace.h>
#include <linux/mman.h>
#include <linux/mm.h>
#include <linux/swap.h>
#include <linux/stddef.h>
#include <linux/vmalloc.h>
#include <linux/init.h>
#include <linux/delay.h>
#include <linux/bootmem.h>
#include <linux/highmem.h>
#include <linux/pci.h>
#include <linux/interrupt.h>
#include <linux/gfp.h>
#include <asm/pgalloc.h>
#include <linux/io.h>
#include <linux/hardirq.h>
#include <linux/mmu_context.h>
#include <asm/mmu.h>
#include <linux/uaccess.h>
#include <asm/pgtable.h>
#include <asm/cpuinfo.h>
#include <asm/tlbflush.h>
#ifndef CONFIG_MMU
/* I have to use dcache values because I can't relate on ram size */
# define UNCACHED_SHADOW_MASK (cpuinfo.dcache_high - cpuinfo.dcache_base + 1)
#endif
/*
* Consistent memory allocators. Used for DMA devices that want to
* share uncached memory with the processor core.
* My crufty no-MMU approach is simple. In the HW platform we can optionally
* mirror the DDR up above the processor cacheable region. So, memory accessed
* in this mirror region will not be cached. It's alloced from the same
* pool as normal memory, but the handle we return is shifted up into the
* uncached region. This will no doubt cause big problems if memory allocated
* here is not also freed properly. -- JW
*/
void *consistent_alloc(gfp_t gfp, size_t size, dma_addr_t *dma_handle)
{
unsigned long order, vaddr;
void *ret;
unsigned int i, err = 0;
struct page *page, *end;
#ifdef CONFIG_MMU
phys_addr_t pa;
struct vm_struct *area;
unsigned long va;
#endif
if (in_interrupt())
BUG();
/* Only allocate page size areas. */
size = PAGE_ALIGN(size);
order = get_order(size);
vaddr = __get_free_pages(gfp, order);
if (!vaddr)
return NULL;
/*
* we need to ensure that there are no cachelines in use,
* or worse dirty in this area.
*/
flush_dcache_range(virt_to_phys((void *)vaddr),
virt_to_phys((void *)vaddr) + size);
#ifndef CONFIG_MMU
ret = (void *)vaddr;
/*
* Here's the magic! Note if the uncached shadow is not implemented,
* it's up to the calling code to also test that condition and make
* other arranegments, such as manually flushing the cache and so on.
*/
# ifdef CONFIG_XILINX_UNCACHED_SHADOW
ret = (void *)((unsigned) ret | UNCACHED_SHADOW_MASK);
# endif
if ((unsigned int)ret > cpuinfo.dcache_base &&
(unsigned int)ret < cpuinfo.dcache_high)
pr_warn("ERROR: Your cache coherent area is CACHED!!!\n");
/* dma_handle is same as physical (shadowed) address */
*dma_handle = (dma_addr_t)ret;
#else
/* Allocate some common virtual space to map the new pages. */
area = get_vm_area(size, VM_ALLOC);
if (!area) {
free_pages(vaddr, order);
return NULL;
}
va = (unsigned long) area->addr;
ret = (void *)va;
/* This gives us the real physical address of the first page. */
*dma_handle = pa = virt_to_bus((void *)vaddr);
#endif
/*
* free wasted pages. We skip the first page since we know
* that it will have count = 1 and won't require freeing.
* We also mark the pages in use as reserved so that
* remap_page_range works.
*/
page = virt_to_page(vaddr);
end = page + (1 << order);
split_page(page, order);
for (i = 0; i < size && err == 0; i += PAGE_SIZE) {
#ifdef CONFIG_MMU
/* MS: This is the whole magic - use cache inhibit pages */
err = map_page(va + i, pa + i, _PAGE_KERNEL | _PAGE_NO_CACHE);
#endif
SetPageReserved(page);
page++;
}
/* Free the otherwise unused pages. */
while (page < end) {
__free_page(page);
page++;
}
if (err) {
free_pages(vaddr, order);
return NULL;
}
return ret;
}
EXPORT_SYMBOL(consistent_alloc);
/*
* free page(s) as defined by the above mapping.
*/
void consistent_free(size_t size, void *vaddr)
{
struct page *page;
if (in_interrupt())
BUG();
size = PAGE_ALIGN(size);
#ifndef CONFIG_MMU
/* Clear SHADOW_MASK bit in address, and free as per usual */
# ifdef CONFIG_XILINX_UNCACHED_SHADOW
vaddr = (void *)((unsigned)vaddr & ~UNCACHED_SHADOW_MASK);
# endif
page = virt_to_page(vaddr);
do {
__free_reserved_page(page);
page++;
} while (size -= PAGE_SIZE);
#else
do {
pte_t *ptep;
unsigned long pfn;
ptep = pte_offset_kernel(pmd_offset(pgd_offset_k(
(unsigned int)vaddr),
(unsigned int)vaddr),
(unsigned int)vaddr);
if (!pte_none(*ptep) && pte_present(*ptep)) {
pfn = pte_pfn(*ptep);
pte_clear(&init_mm, (unsigned int)vaddr, ptep);
if (pfn_valid(pfn)) {
page = pfn_to_page(pfn);
__free_reserved_page(page);
}
}
vaddr += PAGE_SIZE;
} while (size -= PAGE_SIZE);
/* flush tlb */
flush_tlb_all();
#endif
}
EXPORT_SYMBOL(consistent_free);
/*
* make an area consistent.
*/
void consistent_sync(void *vaddr, size_t size, int direction)
{
unsigned long start;
unsigned long end;
start = (unsigned long)vaddr;
/* Convert start address back down to unshadowed memory region */
#ifdef CONFIG_XILINX_UNCACHED_SHADOW
start &= ~UNCACHED_SHADOW_MASK;
#endif
end = start + size;
switch (direction) {
case PCI_DMA_NONE:
BUG();
case PCI_DMA_FROMDEVICE: /* invalidate only */
invalidate_dcache_range(start, end);
break;
case PCI_DMA_TODEVICE: /* writeback only */
flush_dcache_range(start, end);
break;
case PCI_DMA_BIDIRECTIONAL: /* writeback and invalidate */
flush_dcache_range(start, end);
break;
}
}
EXPORT_SYMBOL(consistent_sync);
/*
* consistent_sync_page makes memory consistent. identical
* to consistent_sync, but takes a struct page instead of a
* virtual address
*/
void consistent_sync_page(struct page *page, unsigned long offset,
size_t size, int direction)
{
unsigned long start = (unsigned long)page_address(page) + offset;
consistent_sync((void *)start, size, direction);
}
EXPORT_SYMBOL(consistent_sync_page);
|
