blob: 6cb80b4973d2db0db22021d043d71b49b136991d [file] [log] [blame]
Russell Kingd111e8f2006-09-27 15:27:33 +01001/*
2 * linux/arch/arm/mm/mmu.c
3 *
4 * Copyright (C) 1995-2005 Russell King
5 *
6 * This program is free software; you can redistribute it and/or modify
7 * it under the terms of the GNU General Public License version 2 as
8 * published by the Free Software Foundation.
9 */
Russell Kingae8f1542006-09-27 15:38:34 +010010#include <linux/module.h>
Russell Kingd111e8f2006-09-27 15:27:33 +010011#include <linux/kernel.h>
12#include <linux/errno.h>
13#include <linux/init.h>
14#include <linux/bootmem.h>
15#include <linux/mman.h>
16#include <linux/nodemask.h>
17
18#include <asm/mach-types.h>
19#include <asm/setup.h>
20#include <asm/sizes.h>
21#include <asm/tlb.h>
22
23#include <asm/mach/arch.h>
24#include <asm/mach/map.h>
25
26#include "mm.h"
27
28DEFINE_PER_CPU(struct mmu_gather, mmu_gathers);
29
Russell King6ae5a6e2006-09-30 10:50:05 +010030extern void _stext, _etext, __data_start, _end;
Russell Kingd111e8f2006-09-27 15:27:33 +010031extern pgd_t swapper_pg_dir[PTRS_PER_PGD];
32
33/*
34 * empty_zero_page is a special page that is used for
35 * zero-initialized data and COW.
36 */
37struct page *empty_zero_page;
38
39/*
40 * The pmd table for the upper-most set of pages.
41 */
42pmd_t *top_pmd;
43
Russell Kingae8f1542006-09-27 15:38:34 +010044#define CPOLICY_UNCACHED 0
45#define CPOLICY_BUFFERED 1
46#define CPOLICY_WRITETHROUGH 2
47#define CPOLICY_WRITEBACK 3
48#define CPOLICY_WRITEALLOC 4
49
50static unsigned int cachepolicy __initdata = CPOLICY_WRITEBACK;
51static unsigned int ecc_mask __initdata = 0;
Imre_Deak44b18692007-02-11 13:45:13 +010052pgprot_t pgprot_user;
Russell Kingae8f1542006-09-27 15:38:34 +010053pgprot_t pgprot_kernel;
54
Imre_Deak44b18692007-02-11 13:45:13 +010055EXPORT_SYMBOL(pgprot_user);
Russell Kingae8f1542006-09-27 15:38:34 +010056EXPORT_SYMBOL(pgprot_kernel);
57
58struct cachepolicy {
59 const char policy[16];
60 unsigned int cr_mask;
61 unsigned int pmd;
62 unsigned int pte;
63};
64
65static struct cachepolicy cache_policies[] __initdata = {
66 {
67 .policy = "uncached",
68 .cr_mask = CR_W|CR_C,
69 .pmd = PMD_SECT_UNCACHED,
70 .pte = 0,
71 }, {
72 .policy = "buffered",
73 .cr_mask = CR_C,
74 .pmd = PMD_SECT_BUFFERED,
75 .pte = PTE_BUFFERABLE,
76 }, {
77 .policy = "writethrough",
78 .cr_mask = 0,
79 .pmd = PMD_SECT_WT,
80 .pte = PTE_CACHEABLE,
81 }, {
82 .policy = "writeback",
83 .cr_mask = 0,
84 .pmd = PMD_SECT_WB,
85 .pte = PTE_BUFFERABLE|PTE_CACHEABLE,
86 }, {
87 .policy = "writealloc",
88 .cr_mask = 0,
89 .pmd = PMD_SECT_WBWA,
90 .pte = PTE_BUFFERABLE|PTE_CACHEABLE,
91 }
92};
93
94/*
95 * These are useful for identifing cache coherency
96 * problems by allowing the cache or the cache and
97 * writebuffer to be turned off. (Note: the write
98 * buffer should not be on and the cache off).
99 */
100static void __init early_cachepolicy(char **p)
101{
102 int i;
103
104 for (i = 0; i < ARRAY_SIZE(cache_policies); i++) {
105 int len = strlen(cache_policies[i].policy);
106
107 if (memcmp(*p, cache_policies[i].policy, len) == 0) {
108 cachepolicy = i;
109 cr_alignment &= ~cache_policies[i].cr_mask;
110 cr_no_alignment &= ~cache_policies[i].cr_mask;
111 *p += len;
112 break;
113 }
114 }
115 if (i == ARRAY_SIZE(cache_policies))
116 printk(KERN_ERR "ERROR: unknown or unsupported cache policy\n");
117 flush_cache_all();
118 set_cr(cr_alignment);
119}
120__early_param("cachepolicy=", early_cachepolicy);
121
122static void __init early_nocache(char **__unused)
123{
124 char *p = "buffered";
125 printk(KERN_WARNING "nocache is deprecated; use cachepolicy=%s\n", p);
126 early_cachepolicy(&p);
127}
128__early_param("nocache", early_nocache);
129
130static void __init early_nowrite(char **__unused)
131{
132 char *p = "uncached";
133 printk(KERN_WARNING "nowb is deprecated; use cachepolicy=%s\n", p);
134 early_cachepolicy(&p);
135}
136__early_param("nowb", early_nowrite);
137
138static void __init early_ecc(char **p)
139{
140 if (memcmp(*p, "on", 2) == 0) {
141 ecc_mask = PMD_PROTECTION;
142 *p += 2;
143 } else if (memcmp(*p, "off", 3) == 0) {
144 ecc_mask = 0;
145 *p += 3;
146 }
147}
148__early_param("ecc=", early_ecc);
149
150static int __init noalign_setup(char *__unused)
151{
152 cr_alignment &= ~CR_A;
153 cr_no_alignment &= ~CR_A;
154 set_cr(cr_alignment);
155 return 1;
156}
157__setup("noalign", noalign_setup);
158
Russell King255d1f82006-12-18 00:12:47 +0000159#ifndef CONFIG_SMP
160void adjust_cr(unsigned long mask, unsigned long set)
161{
162 unsigned long flags;
163
164 mask &= ~CR_A;
165
166 set &= mask;
167
168 local_irq_save(flags);
169
170 cr_no_alignment = (cr_no_alignment & ~mask) | set;
171 cr_alignment = (cr_alignment & ~mask) | set;
172
173 set_cr((get_cr() & ~mask) | set);
174
175 local_irq_restore(flags);
176}
177#endif
178
Russell Kingb29e9f52007-04-21 10:47:29 +0100179static struct mem_type mem_types[] = {
Russell Kingae8f1542006-09-27 15:38:34 +0100180 [MT_DEVICE] = {
181 .prot_pte = L_PTE_PRESENT | L_PTE_YOUNG | L_PTE_DIRTY |
182 L_PTE_WRITE,
183 .prot_l1 = PMD_TYPE_TABLE,
184 .prot_sect = PMD_TYPE_SECT | PMD_BIT4 | PMD_SECT_UNCACHED |
185 PMD_SECT_AP_WRITE,
186 .domain = DOMAIN_IO,
187 },
188 [MT_CACHECLEAN] = {
189 .prot_sect = PMD_TYPE_SECT | PMD_BIT4,
190 .domain = DOMAIN_KERNEL,
191 },
192 [MT_MINICLEAN] = {
193 .prot_sect = PMD_TYPE_SECT | PMD_BIT4 | PMD_SECT_MINICACHE,
194 .domain = DOMAIN_KERNEL,
195 },
196 [MT_LOW_VECTORS] = {
197 .prot_pte = L_PTE_PRESENT | L_PTE_YOUNG | L_PTE_DIRTY |
198 L_PTE_EXEC,
199 .prot_l1 = PMD_TYPE_TABLE,
200 .domain = DOMAIN_USER,
201 },
202 [MT_HIGH_VECTORS] = {
203 .prot_pte = L_PTE_PRESENT | L_PTE_YOUNG | L_PTE_DIRTY |
204 L_PTE_USER | L_PTE_EXEC,
205 .prot_l1 = PMD_TYPE_TABLE,
206 .domain = DOMAIN_USER,
207 },
208 [MT_MEMORY] = {
209 .prot_sect = PMD_TYPE_SECT | PMD_BIT4 | PMD_SECT_AP_WRITE,
210 .domain = DOMAIN_KERNEL,
211 },
212 [MT_ROM] = {
213 .prot_sect = PMD_TYPE_SECT | PMD_BIT4,
214 .domain = DOMAIN_KERNEL,
215 },
216 [MT_IXP2000_DEVICE] = { /* IXP2400 requires XCB=101 for on-chip I/O */
217 .prot_pte = L_PTE_PRESENT | L_PTE_YOUNG | L_PTE_DIRTY |
218 L_PTE_WRITE,
219 .prot_l1 = PMD_TYPE_TABLE,
220 .prot_sect = PMD_TYPE_SECT | PMD_BIT4 | PMD_SECT_UNCACHED |
221 PMD_SECT_AP_WRITE | PMD_SECT_BUFFERABLE |
222 PMD_SECT_TEX(1),
223 .domain = DOMAIN_IO,
224 },
225 [MT_NONSHARED_DEVICE] = {
226 .prot_l1 = PMD_TYPE_TABLE,
227 .prot_sect = PMD_TYPE_SECT | PMD_BIT4 | PMD_SECT_NONSHARED_DEV |
228 PMD_SECT_AP_WRITE,
229 .domain = DOMAIN_IO,
230 }
231};
232
Russell Kingb29e9f52007-04-21 10:47:29 +0100233const struct mem_type *get_mem_type(unsigned int type)
234{
235 return type < ARRAY_SIZE(mem_types) ? &mem_types[type] : NULL;
236}
237
Russell Kingae8f1542006-09-27 15:38:34 +0100238/*
239 * Adjust the PMD section entries according to the CPU in use.
240 */
241static void __init build_mem_type_table(void)
242{
243 struct cachepolicy *cp;
244 unsigned int cr = get_cr();
245 unsigned int user_pgprot, kern_pgprot;
246 int cpu_arch = cpu_architecture();
247 int i;
248
249#if defined(CONFIG_CPU_DCACHE_DISABLE)
250 if (cachepolicy > CPOLICY_BUFFERED)
251 cachepolicy = CPOLICY_BUFFERED;
252#elif defined(CONFIG_CPU_DCACHE_WRITETHROUGH)
253 if (cachepolicy > CPOLICY_WRITETHROUGH)
254 cachepolicy = CPOLICY_WRITETHROUGH;
255#endif
256 if (cpu_arch < CPU_ARCH_ARMv5) {
257 if (cachepolicy >= CPOLICY_WRITEALLOC)
258 cachepolicy = CPOLICY_WRITEBACK;
259 ecc_mask = 0;
260 }
261
262 /*
263 * Xscale must not have PMD bit 4 set for section mappings.
264 */
265 if (cpu_is_xscale())
266 for (i = 0; i < ARRAY_SIZE(mem_types); i++)
267 mem_types[i].prot_sect &= ~PMD_BIT4;
268
269 /*
270 * ARMv5 and lower, excluding Xscale, bit 4 must be set for
271 * page tables.
272 */
273 if (cpu_arch < CPU_ARCH_ARMv6 && !cpu_is_xscale())
274 for (i = 0; i < ARRAY_SIZE(mem_types); i++)
275 if (mem_types[i].prot_l1)
276 mem_types[i].prot_l1 |= PMD_BIT4;
277
278 cp = &cache_policies[cachepolicy];
279 kern_pgprot = user_pgprot = cp->pte;
280
281 /*
282 * Enable CPU-specific coherency if supported.
283 * (Only available on XSC3 at the moment.)
284 */
285 if (arch_is_coherent()) {
286 if (cpu_is_xsc3()) {
287 mem_types[MT_MEMORY].prot_sect |= PMD_SECT_S;
Lennert Buytenhek0e5fdca2006-12-02 00:03:47 +0100288 mem_types[MT_MEMORY].prot_pte |= L_PTE_SHARED;
Russell Kingae8f1542006-09-27 15:38:34 +0100289 }
290 }
291
292 /*
293 * ARMv6 and above have extended page tables.
294 */
295 if (cpu_arch >= CPU_ARCH_ARMv6 && (cr & CR_XP)) {
296 /*
297 * bit 4 becomes XN which we must clear for the
298 * kernel memory mapping.
299 */
300 mem_types[MT_MEMORY].prot_sect &= ~PMD_SECT_XN;
301 mem_types[MT_ROM].prot_sect &= ~PMD_SECT_XN;
302
303 /*
304 * Mark cache clean areas and XIP ROM read only
305 * from SVC mode and no access from userspace.
306 */
307 mem_types[MT_ROM].prot_sect |= PMD_SECT_APX|PMD_SECT_AP_WRITE;
308 mem_types[MT_MINICLEAN].prot_sect |= PMD_SECT_APX|PMD_SECT_AP_WRITE;
309 mem_types[MT_CACHECLEAN].prot_sect |= PMD_SECT_APX|PMD_SECT_AP_WRITE;
310
311 /*
312 * Mark the device area as "shared device"
313 */
314 mem_types[MT_DEVICE].prot_pte |= L_PTE_BUFFERABLE;
315 mem_types[MT_DEVICE].prot_sect |= PMD_SECT_BUFFERED;
316
Russell Kingae8f1542006-09-27 15:38:34 +0100317#ifdef CONFIG_SMP
318 /*
319 * Mark memory with the "shared" attribute for SMP systems
320 */
321 user_pgprot |= L_PTE_SHARED;
322 kern_pgprot |= L_PTE_SHARED;
323 mem_types[MT_MEMORY].prot_sect |= PMD_SECT_S;
324#endif
325 }
326
327 for (i = 0; i < 16; i++) {
328 unsigned long v = pgprot_val(protection_map[i]);
329 v = (v & ~(L_PTE_BUFFERABLE|L_PTE_CACHEABLE)) | user_pgprot;
330 protection_map[i] = __pgprot(v);
331 }
332
333 mem_types[MT_LOW_VECTORS].prot_pte |= kern_pgprot;
334 mem_types[MT_HIGH_VECTORS].prot_pte |= kern_pgprot;
335
336 if (cpu_arch >= CPU_ARCH_ARMv5) {
337#ifndef CONFIG_SMP
338 /*
339 * Only use write-through for non-SMP systems
340 */
341 mem_types[MT_LOW_VECTORS].prot_pte &= ~L_PTE_BUFFERABLE;
342 mem_types[MT_HIGH_VECTORS].prot_pte &= ~L_PTE_BUFFERABLE;
343#endif
344 } else {
345 mem_types[MT_MINICLEAN].prot_sect &= ~PMD_SECT_TEX(1);
346 }
347
Imre_Deak44b18692007-02-11 13:45:13 +0100348 pgprot_user = __pgprot(L_PTE_PRESENT | L_PTE_YOUNG | user_pgprot);
Russell Kingae8f1542006-09-27 15:38:34 +0100349 pgprot_kernel = __pgprot(L_PTE_PRESENT | L_PTE_YOUNG |
350 L_PTE_DIRTY | L_PTE_WRITE |
351 L_PTE_EXEC | kern_pgprot);
352
353 mem_types[MT_LOW_VECTORS].prot_l1 |= ecc_mask;
354 mem_types[MT_HIGH_VECTORS].prot_l1 |= ecc_mask;
355 mem_types[MT_MEMORY].prot_sect |= ecc_mask | cp->pmd;
356 mem_types[MT_ROM].prot_sect |= cp->pmd;
357
358 switch (cp->pmd) {
359 case PMD_SECT_WT:
360 mem_types[MT_CACHECLEAN].prot_sect |= PMD_SECT_WT;
361 break;
362 case PMD_SECT_WB:
363 case PMD_SECT_WBWA:
364 mem_types[MT_CACHECLEAN].prot_sect |= PMD_SECT_WB;
365 break;
366 }
367 printk("Memory policy: ECC %sabled, Data cache %s\n",
368 ecc_mask ? "en" : "dis", cp->policy);
Russell King2497f0a2007-04-21 09:59:44 +0100369
370 for (i = 0; i < ARRAY_SIZE(mem_types); i++) {
371 struct mem_type *t = &mem_types[i];
372 if (t->prot_l1)
373 t->prot_l1 |= PMD_DOMAIN(t->domain);
374 if (t->prot_sect)
375 t->prot_sect |= PMD_DOMAIN(t->domain);
376 }
Russell Kingae8f1542006-09-27 15:38:34 +0100377}
378
379#define vectors_base() (vectors_high() ? 0xffff0000 : 0)
380
Russell King24e6c692007-04-21 10:21:28 +0100381static void __init alloc_init_pte(pmd_t *pmd, unsigned long addr,
382 unsigned long end, unsigned long pfn,
383 const struct mem_type *type)
Russell Kingae8f1542006-09-27 15:38:34 +0100384{
Russell King24e6c692007-04-21 10:21:28 +0100385 pte_t *pte;
Russell Kingae8f1542006-09-27 15:38:34 +0100386
Russell King24e6c692007-04-21 10:21:28 +0100387 if (pmd_none(*pmd)) {
388 pte = alloc_bootmem_low_pages(2 * PTRS_PER_PTE * sizeof(pte_t));
389 __pmd_populate(pmd, __pa(pte) | type->prot_l1);
390 }
Russell Kingae8f1542006-09-27 15:38:34 +0100391
Russell King24e6c692007-04-21 10:21:28 +0100392 pte = pte_offset_kernel(pmd, addr);
393 do {
394 set_pte_ext(pte, pfn_pte(pfn, __pgprot(type->prot_pte)), 0);
395 pfn++;
396 } while (pte++, addr += PAGE_SIZE, addr != end);
Russell Kingae8f1542006-09-27 15:38:34 +0100397}
398
Russell King24e6c692007-04-21 10:21:28 +0100399static void __init alloc_init_section(pgd_t *pgd, unsigned long addr,
400 unsigned long end, unsigned long phys,
401 const struct mem_type *type)
Russell Kingae8f1542006-09-27 15:38:34 +0100402{
Russell King24e6c692007-04-21 10:21:28 +0100403 pmd_t *pmd = pmd_offset(pgd, addr);
Russell Kingae8f1542006-09-27 15:38:34 +0100404
Russell King24e6c692007-04-21 10:21:28 +0100405 /*
406 * Try a section mapping - end, addr and phys must all be aligned
407 * to a section boundary. Note that PMDs refer to the individual
408 * L1 entries, whereas PGDs refer to a group of L1 entries making
409 * up one logical pointer to an L2 table.
410 */
411 if (((addr | end | phys) & ~SECTION_MASK) == 0) {
412 pmd_t *p = pmd;
Russell Kingae8f1542006-09-27 15:38:34 +0100413
Russell King24e6c692007-04-21 10:21:28 +0100414 if (addr & SECTION_SIZE)
415 pmd++;
416
417 do {
418 *pmd = __pmd(phys | type->prot_sect);
419 phys += SECTION_SIZE;
420 } while (pmd++, addr += SECTION_SIZE, addr != end);
421
422 flush_pmd_entry(p);
423 } else {
424 /*
425 * No need to loop; pte's aren't interested in the
426 * individual L1 entries.
427 */
428 alloc_init_pte(pmd, addr, end, __phys_to_pfn(phys), type);
Russell Kingae8f1542006-09-27 15:38:34 +0100429 }
Russell Kingae8f1542006-09-27 15:38:34 +0100430}
431
Russell King4a56c1e2007-04-21 10:16:48 +0100432static void __init create_36bit_mapping(struct map_desc *md,
433 const struct mem_type *type)
434{
435 unsigned long phys, addr, length, end;
436 pgd_t *pgd;
437
438 addr = md->virtual;
439 phys = (unsigned long)__pfn_to_phys(md->pfn);
440 length = PAGE_ALIGN(md->length);
441
442 if (!(cpu_architecture() >= CPU_ARCH_ARMv6 || cpu_is_xsc3())) {
443 printk(KERN_ERR "MM: CPU does not support supersection "
444 "mapping for 0x%08llx at 0x%08lx\n",
445 __pfn_to_phys((u64)md->pfn), addr);
446 return;
447 }
448
449 /* N.B. ARMv6 supersections are only defined to work with domain 0.
450 * Since domain assignments can in fact be arbitrary, the
451 * 'domain == 0' check below is required to insure that ARMv6
452 * supersections are only allocated for domain 0 regardless
453 * of the actual domain assignments in use.
454 */
455 if (type->domain) {
456 printk(KERN_ERR "MM: invalid domain in supersection "
457 "mapping for 0x%08llx at 0x%08lx\n",
458 __pfn_to_phys((u64)md->pfn), addr);
459 return;
460 }
461
462 if ((addr | length | __pfn_to_phys(md->pfn)) & ~SUPERSECTION_MASK) {
463 printk(KERN_ERR "MM: cannot create mapping for "
464 "0x%08llx at 0x%08lx invalid alignment\n",
465 __pfn_to_phys((u64)md->pfn), addr);
466 return;
467 }
468
469 /*
470 * Shift bits [35:32] of address into bits [23:20] of PMD
471 * (See ARMv6 spec).
472 */
473 phys |= (((md->pfn >> (32 - PAGE_SHIFT)) & 0xF) << 20);
474
475 pgd = pgd_offset_k(addr);
476 end = addr + length;
477 do {
478 pmd_t *pmd = pmd_offset(pgd, addr);
479 int i;
480
481 for (i = 0; i < 16; i++)
482 *pmd++ = __pmd(phys | type->prot_sect | PMD_SECT_SUPER);
483
484 addr += SUPERSECTION_SIZE;
485 phys += SUPERSECTION_SIZE;
486 pgd += SUPERSECTION_SIZE >> PGDIR_SHIFT;
487 } while (addr != end);
488}
489
Russell Kingae8f1542006-09-27 15:38:34 +0100490/*
491 * Create the page directory entries and any necessary
492 * page tables for the mapping specified by `md'. We
493 * are able to cope here with varying sizes and address
494 * offsets, and we take full advantage of sections and
495 * supersections.
496 */
497void __init create_mapping(struct map_desc *md)
498{
Russell King24e6c692007-04-21 10:21:28 +0100499 unsigned long phys, addr, length, end;
Russell Kingd5c98172007-04-21 10:05:32 +0100500 const struct mem_type *type;
Russell King24e6c692007-04-21 10:21:28 +0100501 pgd_t *pgd;
Russell Kingae8f1542006-09-27 15:38:34 +0100502
503 if (md->virtual != vectors_base() && md->virtual < TASK_SIZE) {
504 printk(KERN_WARNING "BUG: not creating mapping for "
505 "0x%08llx at 0x%08lx in user region\n",
506 __pfn_to_phys((u64)md->pfn), md->virtual);
507 return;
508 }
509
510 if ((md->type == MT_DEVICE || md->type == MT_ROM) &&
511 md->virtual >= PAGE_OFFSET && md->virtual < VMALLOC_END) {
512 printk(KERN_WARNING "BUG: mapping for 0x%08llx at 0x%08lx "
513 "overlaps vmalloc space\n",
514 __pfn_to_phys((u64)md->pfn), md->virtual);
515 }
516
Russell Kingd5c98172007-04-21 10:05:32 +0100517 type = &mem_types[md->type];
Russell Kingae8f1542006-09-27 15:38:34 +0100518
519 /*
520 * Catch 36-bit addresses
521 */
Russell King4a56c1e2007-04-21 10:16:48 +0100522 if (md->pfn >= 0x100000) {
523 create_36bit_mapping(md, type);
524 return;
Russell Kingae8f1542006-09-27 15:38:34 +0100525 }
526
Russell King24e6c692007-04-21 10:21:28 +0100527 addr = md->virtual;
528 phys = (unsigned long)__pfn_to_phys(md->pfn);
529 length = PAGE_ALIGN(md->length);
Russell Kingae8f1542006-09-27 15:38:34 +0100530
Russell King24e6c692007-04-21 10:21:28 +0100531 if (type->prot_l1 == 0 && ((addr | phys | length) & ~SECTION_MASK)) {
Russell Kingae8f1542006-09-27 15:38:34 +0100532 printk(KERN_WARNING "BUG: map for 0x%08lx at 0x%08lx can not "
533 "be mapped using pages, ignoring.\n",
Russell King24e6c692007-04-21 10:21:28 +0100534 __pfn_to_phys(md->pfn), addr);
Russell Kingae8f1542006-09-27 15:38:34 +0100535 return;
536 }
537
Russell King24e6c692007-04-21 10:21:28 +0100538 pgd = pgd_offset_k(addr);
539 end = addr + length;
540 do {
541 unsigned long next = pgd_addr_end(addr, end);
Russell Kingae8f1542006-09-27 15:38:34 +0100542
Russell King24e6c692007-04-21 10:21:28 +0100543 alloc_init_section(pgd, addr, next, phys, type);
Russell Kingae8f1542006-09-27 15:38:34 +0100544
Russell King24e6c692007-04-21 10:21:28 +0100545 phys += next - addr;
546 addr = next;
547 } while (pgd++, addr != end);
Russell Kingae8f1542006-09-27 15:38:34 +0100548}
549
550/*
551 * Create the architecture specific mappings
552 */
553void __init iotable_init(struct map_desc *io_desc, int nr)
554{
555 int i;
556
557 for (i = 0; i < nr; i++)
558 create_mapping(io_desc + i);
559}
560
Russell Kingd111e8f2006-09-27 15:27:33 +0100561static inline void prepare_page_table(struct meminfo *mi)
562{
563 unsigned long addr;
564
565 /*
566 * Clear out all the mappings below the kernel image.
567 */
568 for (addr = 0; addr < MODULE_START; addr += PGDIR_SIZE)
569 pmd_clear(pmd_off_k(addr));
570
571#ifdef CONFIG_XIP_KERNEL
572 /* The XIP kernel is mapped in the module area -- skip over it */
573 addr = ((unsigned long)&_etext + PGDIR_SIZE - 1) & PGDIR_MASK;
574#endif
575 for ( ; addr < PAGE_OFFSET; addr += PGDIR_SIZE)
576 pmd_clear(pmd_off_k(addr));
577
578 /*
579 * Clear out all the kernel space mappings, except for the first
580 * memory bank, up to the end of the vmalloc region.
581 */
582 for (addr = __phys_to_virt(mi->bank[0].start + mi->bank[0].size);
583 addr < VMALLOC_END; addr += PGDIR_SIZE)
584 pmd_clear(pmd_off_k(addr));
585}
586
587/*
588 * Reserve the various regions of node 0
589 */
590void __init reserve_node_zero(pg_data_t *pgdat)
591{
592 unsigned long res_size = 0;
593
594 /*
595 * Register the kernel text and data with bootmem.
596 * Note that this can only be in node 0.
597 */
598#ifdef CONFIG_XIP_KERNEL
599 reserve_bootmem_node(pgdat, __pa(&__data_start), &_end - &__data_start);
600#else
601 reserve_bootmem_node(pgdat, __pa(&_stext), &_end - &_stext);
602#endif
603
604 /*
605 * Reserve the page tables. These are already in use,
606 * and can only be in node 0.
607 */
608 reserve_bootmem_node(pgdat, __pa(swapper_pg_dir),
609 PTRS_PER_PGD * sizeof(pgd_t));
610
611 /*
612 * Hmm... This should go elsewhere, but we really really need to
613 * stop things allocating the low memory; ideally we need a better
614 * implementation of GFP_DMA which does not assume that DMA-able
615 * memory starts at zero.
616 */
617 if (machine_is_integrator() || machine_is_cintegrator())
618 res_size = __pa(swapper_pg_dir) - PHYS_OFFSET;
619
620 /*
621 * These should likewise go elsewhere. They pre-reserve the
622 * screen memory region at the start of main system memory.
623 */
624 if (machine_is_edb7211())
625 res_size = 0x00020000;
626 if (machine_is_p720t())
627 res_size = 0x00014000;
628
Ben Dooksbbf6f282006-12-07 20:47:58 +0100629 /* H1940 and RX3715 need to reserve this for suspend */
630
631 if (machine_is_h1940() || machine_is_rx3715()) {
Ben Dooks90733412006-12-06 01:50:24 +0100632 reserve_bootmem_node(pgdat, 0x30003000, 0x1000);
633 reserve_bootmem_node(pgdat, 0x30081000, 0x1000);
634 }
635
Russell Kingd111e8f2006-09-27 15:27:33 +0100636#ifdef CONFIG_SA1111
637 /*
638 * Because of the SA1111 DMA bug, we want to preserve our
639 * precious DMA-able memory...
640 */
641 res_size = __pa(swapper_pg_dir) - PHYS_OFFSET;
642#endif
643 if (res_size)
644 reserve_bootmem_node(pgdat, PHYS_OFFSET, res_size);
645}
646
647/*
648 * Set up device the mappings. Since we clear out the page tables for all
649 * mappings above VMALLOC_END, we will remove any debug device mappings.
650 * This means you have to be careful how you debug this function, or any
651 * called function. This means you can't use any function or debugging
652 * method which may touch any device, otherwise the kernel _will_ crash.
653 */
654static void __init devicemaps_init(struct machine_desc *mdesc)
655{
656 struct map_desc map;
657 unsigned long addr;
658 void *vectors;
659
660 /*
661 * Allocate the vector page early.
662 */
663 vectors = alloc_bootmem_low_pages(PAGE_SIZE);
664 BUG_ON(!vectors);
665
666 for (addr = VMALLOC_END; addr; addr += PGDIR_SIZE)
667 pmd_clear(pmd_off_k(addr));
668
669 /*
670 * Map the kernel if it is XIP.
671 * It is always first in the modulearea.
672 */
673#ifdef CONFIG_XIP_KERNEL
674 map.pfn = __phys_to_pfn(CONFIG_XIP_PHYS_ADDR & SECTION_MASK);
675 map.virtual = MODULE_START;
676 map.length = ((unsigned long)&_etext - map.virtual + ~SECTION_MASK) & SECTION_MASK;
677 map.type = MT_ROM;
678 create_mapping(&map);
679#endif
680
681 /*
682 * Map the cache flushing regions.
683 */
684#ifdef FLUSH_BASE
685 map.pfn = __phys_to_pfn(FLUSH_BASE_PHYS);
686 map.virtual = FLUSH_BASE;
687 map.length = SZ_1M;
688 map.type = MT_CACHECLEAN;
689 create_mapping(&map);
690#endif
691#ifdef FLUSH_BASE_MINICACHE
692 map.pfn = __phys_to_pfn(FLUSH_BASE_PHYS + SZ_1M);
693 map.virtual = FLUSH_BASE_MINICACHE;
694 map.length = SZ_1M;
695 map.type = MT_MINICLEAN;
696 create_mapping(&map);
697#endif
698
699 /*
700 * Create a mapping for the machine vectors at the high-vectors
701 * location (0xffff0000). If we aren't using high-vectors, also
702 * create a mapping at the low-vectors virtual address.
703 */
704 map.pfn = __phys_to_pfn(virt_to_phys(vectors));
705 map.virtual = 0xffff0000;
706 map.length = PAGE_SIZE;
707 map.type = MT_HIGH_VECTORS;
708 create_mapping(&map);
709
710 if (!vectors_high()) {
711 map.virtual = 0;
712 map.type = MT_LOW_VECTORS;
713 create_mapping(&map);
714 }
715
716 /*
717 * Ask the machine support to map in the statically mapped devices.
718 */
719 if (mdesc->map_io)
720 mdesc->map_io();
721
722 /*
723 * Finally flush the caches and tlb to ensure that we're in a
724 * consistent state wrt the writebuffer. This also ensures that
725 * any write-allocated cache lines in the vector page are written
726 * back. After this point, we can start to touch devices again.
727 */
728 local_flush_tlb_all();
729 flush_cache_all();
730}
731
732/*
733 * paging_init() sets up the page tables, initialises the zone memory
734 * maps, and sets up the zero page, bad page and bad page tables.
735 */
736void __init paging_init(struct meminfo *mi, struct machine_desc *mdesc)
737{
738 void *zero_page;
739
740 build_mem_type_table();
741 prepare_page_table(mi);
742 bootmem_init(mi);
743 devicemaps_init(mdesc);
744
745 top_pmd = pmd_off_k(0xffff0000);
746
747 /*
748 * allocate the zero page. Note that we count on this going ok.
749 */
750 zero_page = alloc_bootmem_low_pages(PAGE_SIZE);
751 memzero(zero_page, PAGE_SIZE);
752 empty_zero_page = virt_to_page(zero_page);
753 flush_dcache_page(empty_zero_page);
754}
Russell Kingae8f1542006-09-27 15:38:34 +0100755
756/*
757 * In order to soft-boot, we need to insert a 1:1 mapping in place of
758 * the user-mode pages. This will then ensure that we have predictable
759 * results when turning the mmu off
760 */
761void setup_mm_for_reboot(char mode)
762{
763 unsigned long base_pmdval;
764 pgd_t *pgd;
765 int i;
766
767 if (current->mm && current->mm->pgd)
768 pgd = current->mm->pgd;
769 else
770 pgd = init_mm.pgd;
771
772 base_pmdval = PMD_SECT_AP_WRITE | PMD_SECT_AP_READ | PMD_TYPE_SECT;
773 if (cpu_architecture() <= CPU_ARCH_ARMv5TEJ && !cpu_is_xscale())
774 base_pmdval |= PMD_BIT4;
775
776 for (i = 0; i < FIRST_USER_PGD_NR + USER_PTRS_PER_PGD; i++, pgd++) {
777 unsigned long pmdval = (i << PGDIR_SHIFT) | base_pmdval;
778 pmd_t *pmd;
779
780 pmd = pmd_off(pgd, i << PGDIR_SHIFT);
781 pmd[0] = __pmd(pmdval);
782 pmd[1] = __pmd(pmdval + (1 << (PGDIR_SHIFT - 1)));
783 flush_pmd_entry(pmd);
784 }
785}