1 | #ifndef _I386_PGTABLE_H |
---|
2 | #define _I386_PGTABLE_H |
---|
3 | |
---|
4 | #include <asm/hypervisor.h> |
---|
5 | |
---|
6 | /* |
---|
7 | * The Linux memory management assumes a three-level page table setup. On |
---|
8 | * the i386, we use that, but "fold" the mid level into the top-level page |
---|
9 | * table, so that we physically have the same two-level page table as the |
---|
10 | * i386 mmu expects. |
---|
11 | * |
---|
12 | * This file contains the functions and defines necessary to modify and use |
---|
13 | * the i386 page table tree. |
---|
14 | */ |
---|
15 | #ifndef __ASSEMBLY__ |
---|
16 | #include <asm/processor.h> |
---|
17 | #include <asm/fixmap.h> |
---|
18 | #include <linux/threads.h> |
---|
19 | |
---|
20 | #ifndef _I386_BITOPS_H |
---|
21 | #include <asm/bitops.h> |
---|
22 | #endif |
---|
23 | |
---|
24 | #include <linux/slab.h> |
---|
25 | #include <linux/list.h> |
---|
26 | #include <linux/spinlock.h> |
---|
27 | |
---|
28 | /* Is this pagetable pinned? */ |
---|
29 | #define PG_pinned PG_arch_1 |
---|
30 | |
---|
31 | struct mm_struct; |
---|
32 | struct vm_area_struct; |
---|
33 | |
---|
34 | /* |
---|
35 | * ZERO_PAGE is a global shared page that is always zero: used |
---|
36 | * for zero-mapped memory areas etc.. |
---|
37 | */ |
---|
38 | #define ZERO_PAGE(vaddr) (virt_to_page(empty_zero_page)) |
---|
39 | extern unsigned long empty_zero_page[1024]; |
---|
40 | extern pgd_t *swapper_pg_dir; |
---|
41 | extern kmem_cache_t *pgd_cache; |
---|
42 | extern kmem_cache_t *pmd_cache; |
---|
43 | extern spinlock_t pgd_lock; |
---|
44 | extern struct page *pgd_list; |
---|
45 | |
---|
46 | void pmd_ctor(void *, kmem_cache_t *, unsigned long); |
---|
47 | void pgd_ctor(void *, kmem_cache_t *, unsigned long); |
---|
48 | void pgd_dtor(void *, kmem_cache_t *, unsigned long); |
---|
49 | void pgtable_cache_init(void); |
---|
50 | void paging_init(void); |
---|
51 | |
---|
52 | /* |
---|
53 | * The Linux x86 paging architecture is 'compile-time dual-mode', it |
---|
54 | * implements both the traditional 2-level x86 page tables and the |
---|
55 | * newer 3-level PAE-mode page tables. |
---|
56 | */ |
---|
57 | #ifdef CONFIG_X86_PAE |
---|
58 | # include <asm/pgtable-3level-defs.h> |
---|
59 | # define PMD_SIZE (1UL << PMD_SHIFT) |
---|
60 | # define PMD_MASK (~(PMD_SIZE-1)) |
---|
61 | #else |
---|
62 | # include <asm/pgtable-2level-defs.h> |
---|
63 | #endif |
---|
64 | |
---|
65 | #define PGDIR_SIZE (1UL << PGDIR_SHIFT) |
---|
66 | #define PGDIR_MASK (~(PGDIR_SIZE-1)) |
---|
67 | |
---|
68 | #define USER_PTRS_PER_PGD (TASK_SIZE/PGDIR_SIZE) |
---|
69 | #define FIRST_USER_ADDRESS 0 |
---|
70 | |
---|
71 | #define USER_PGD_PTRS (PAGE_OFFSET >> PGDIR_SHIFT) |
---|
72 | #define KERNEL_PGD_PTRS (PTRS_PER_PGD-USER_PGD_PTRS) |
---|
73 | |
---|
74 | #define TWOLEVEL_PGDIR_SHIFT 22 |
---|
75 | #define BOOT_USER_PGD_PTRS (__PAGE_OFFSET >> TWOLEVEL_PGDIR_SHIFT) |
---|
76 | #define BOOT_KERNEL_PGD_PTRS (1024-BOOT_USER_PGD_PTRS) |
---|
77 | |
---|
78 | /* Just any arbitrary offset to the start of the vmalloc VM area: the |
---|
79 | * current 8MB value just means that there will be a 8MB "hole" after the |
---|
80 | * physical memory until the kernel virtual memory starts. That means that |
---|
81 | * any out-of-bounds memory accesses will hopefully be caught. |
---|
82 | * The vmalloc() routines leaves a hole of 4kB between each vmalloced |
---|
83 | * area for the same reason. ;) |
---|
84 | */ |
---|
85 | #define VMALLOC_OFFSET (8*1024*1024) |
---|
86 | #define VMALLOC_START (((unsigned long) high_memory + vmalloc_earlyreserve + \ |
---|
87 | 2*VMALLOC_OFFSET-1) & ~(VMALLOC_OFFSET-1)) |
---|
88 | #ifdef CONFIG_HIGHMEM |
---|
89 | # define VMALLOC_END (PKMAP_BASE-2*PAGE_SIZE) |
---|
90 | #else |
---|
91 | # define VMALLOC_END (FIXADDR_START-2*PAGE_SIZE) |
---|
92 | #endif |
---|
93 | |
---|
94 | /* |
---|
95 | * _PAGE_PSE set in the page directory entry just means that |
---|
96 | * the page directory entry points directly to a 4MB-aligned block of |
---|
97 | * memory. |
---|
98 | */ |
---|
99 | #define _PAGE_BIT_PRESENT 0 |
---|
100 | #define _PAGE_BIT_RW 1 |
---|
101 | #define _PAGE_BIT_USER 2 |
---|
102 | #define _PAGE_BIT_PWT 3 |
---|
103 | #define _PAGE_BIT_PCD 4 |
---|
104 | #define _PAGE_BIT_ACCESSED 5 |
---|
105 | #define _PAGE_BIT_DIRTY 6 |
---|
106 | #define _PAGE_BIT_PSE 7 /* 4 MB (or 2MB) page, Pentium+, if present.. */ |
---|
107 | #define _PAGE_BIT_GLOBAL 8 /* Global TLB entry PPro+ */ |
---|
108 | #define _PAGE_BIT_UNUSED1 9 /* available for programmer */ |
---|
109 | #define _PAGE_BIT_UNUSED2 10 |
---|
110 | #define _PAGE_BIT_UNUSED3 11 |
---|
111 | #define _PAGE_BIT_NX 63 |
---|
112 | |
---|
113 | #define _PAGE_PRESENT 0x001 |
---|
114 | #define _PAGE_RW 0x002 |
---|
115 | #define _PAGE_USER 0x004 |
---|
116 | #define _PAGE_PWT 0x008 |
---|
117 | #define _PAGE_PCD 0x010 |
---|
118 | #define _PAGE_ACCESSED 0x020 |
---|
119 | #define _PAGE_DIRTY 0x040 |
---|
120 | #define _PAGE_PSE 0x080 /* 4 MB (or 2MB) page, Pentium+, if present.. */ |
---|
121 | #define _PAGE_GLOBAL 0x100 /* Global TLB entry PPro+ */ |
---|
122 | #define _PAGE_UNUSED1 0x200 /* available for programmer */ |
---|
123 | #define _PAGE_UNUSED2 0x400 |
---|
124 | #define _PAGE_UNUSED3 0x800 |
---|
125 | |
---|
126 | /* If _PAGE_PRESENT is clear, we use these: */ |
---|
127 | #define _PAGE_FILE 0x040 /* nonlinear file mapping, saved PTE; unset:swap */ |
---|
128 | #define _PAGE_PROTNONE 0x080 /* if the user mapped it with PROT_NONE; |
---|
129 | pte_present gives true */ |
---|
130 | #ifdef CONFIG_X86_PAE |
---|
131 | #define _PAGE_NX (1ULL<<_PAGE_BIT_NX) |
---|
132 | #else |
---|
133 | #define _PAGE_NX 0 |
---|
134 | #endif |
---|
135 | |
---|
136 | #define _PAGE_TABLE (_PAGE_PRESENT | _PAGE_RW | _PAGE_USER | _PAGE_ACCESSED | _PAGE_DIRTY) |
---|
137 | #define _KERNPG_TABLE (_PAGE_PRESENT | _PAGE_RW | _PAGE_ACCESSED | _PAGE_DIRTY) |
---|
138 | #define _PAGE_CHG_MASK (PTE_MASK | _PAGE_ACCESSED | _PAGE_DIRTY) |
---|
139 | |
---|
140 | #define PAGE_NONE \ |
---|
141 | __pgprot(_PAGE_PROTNONE | _PAGE_ACCESSED) |
---|
142 | #define PAGE_SHARED \ |
---|
143 | __pgprot(_PAGE_PRESENT | _PAGE_RW | _PAGE_USER | _PAGE_ACCESSED) |
---|
144 | |
---|
145 | #define PAGE_SHARED_EXEC \ |
---|
146 | __pgprot(_PAGE_PRESENT | _PAGE_RW | _PAGE_USER | _PAGE_ACCESSED) |
---|
147 | #define PAGE_COPY_NOEXEC \ |
---|
148 | __pgprot(_PAGE_PRESENT | _PAGE_USER | _PAGE_ACCESSED | _PAGE_NX) |
---|
149 | #define PAGE_COPY_EXEC \ |
---|
150 | __pgprot(_PAGE_PRESENT | _PAGE_USER | _PAGE_ACCESSED) |
---|
151 | #define PAGE_COPY \ |
---|
152 | PAGE_COPY_NOEXEC |
---|
153 | #define PAGE_READONLY \ |
---|
154 | __pgprot(_PAGE_PRESENT | _PAGE_USER | _PAGE_ACCESSED | _PAGE_NX) |
---|
155 | #define PAGE_READONLY_EXEC \ |
---|
156 | __pgprot(_PAGE_PRESENT | _PAGE_USER | _PAGE_ACCESSED) |
---|
157 | |
---|
158 | #define _PAGE_KERNEL \ |
---|
159 | (_PAGE_PRESENT | _PAGE_RW | _PAGE_DIRTY | _PAGE_ACCESSED | _PAGE_NX) |
---|
160 | #define _PAGE_KERNEL_EXEC \ |
---|
161 | (_PAGE_PRESENT | _PAGE_RW | _PAGE_DIRTY | _PAGE_ACCESSED) |
---|
162 | |
---|
163 | extern unsigned long long __PAGE_KERNEL, __PAGE_KERNEL_EXEC; |
---|
164 | #define __PAGE_KERNEL_RO (__PAGE_KERNEL & ~_PAGE_RW) |
---|
165 | #define __PAGE_KERNEL_NOCACHE (__PAGE_KERNEL | _PAGE_PCD) |
---|
166 | #define __PAGE_KERNEL_LARGE (__PAGE_KERNEL | _PAGE_PSE) |
---|
167 | #define __PAGE_KERNEL_LARGE_EXEC (__PAGE_KERNEL_EXEC | _PAGE_PSE) |
---|
168 | |
---|
169 | #define PAGE_KERNEL __pgprot(__PAGE_KERNEL) |
---|
170 | #define PAGE_KERNEL_RO __pgprot(__PAGE_KERNEL_RO) |
---|
171 | #define PAGE_KERNEL_EXEC __pgprot(__PAGE_KERNEL_EXEC) |
---|
172 | #define PAGE_KERNEL_NOCACHE __pgprot(__PAGE_KERNEL_NOCACHE) |
---|
173 | #define PAGE_KERNEL_LARGE __pgprot(__PAGE_KERNEL_LARGE) |
---|
174 | #define PAGE_KERNEL_LARGE_EXEC __pgprot(__PAGE_KERNEL_LARGE_EXEC) |
---|
175 | |
---|
176 | /* |
---|
177 | * The i386 can't do page protection for execute, and considers that |
---|
178 | * the same are read. Also, write permissions imply read permissions. |
---|
179 | * This is the closest we can get.. |
---|
180 | */ |
---|
181 | #define __P000 PAGE_NONE |
---|
182 | #define __P001 PAGE_READONLY |
---|
183 | #define __P010 PAGE_COPY |
---|
184 | #define __P011 PAGE_COPY |
---|
185 | #define __P100 PAGE_READONLY_EXEC |
---|
186 | #define __P101 PAGE_READONLY_EXEC |
---|
187 | #define __P110 PAGE_COPY_EXEC |
---|
188 | #define __P111 PAGE_COPY_EXEC |
---|
189 | |
---|
190 | #define __S000 PAGE_NONE |
---|
191 | #define __S001 PAGE_READONLY |
---|
192 | #define __S010 PAGE_SHARED |
---|
193 | #define __S011 PAGE_SHARED |
---|
194 | #define __S100 PAGE_READONLY_EXEC |
---|
195 | #define __S101 PAGE_READONLY_EXEC |
---|
196 | #define __S110 PAGE_SHARED_EXEC |
---|
197 | #define __S111 PAGE_SHARED_EXEC |
---|
198 | |
---|
199 | /* |
---|
200 | * Define this if things work differently on an i386 and an i486: |
---|
201 | * it will (on an i486) warn about kernel memory accesses that are |
---|
202 | * done without a 'access_ok(VERIFY_WRITE,..)' |
---|
203 | */ |
---|
204 | #undef TEST_ACCESS_OK |
---|
205 | |
---|
206 | /* The boot page tables (all created as a single array) */ |
---|
207 | extern unsigned long pg0[]; |
---|
208 | |
---|
209 | #define pte_present(x) ((x).pte_low & (_PAGE_PRESENT | _PAGE_PROTNONE)) |
---|
210 | |
---|
211 | /* To avoid harmful races, pmd_none(x) should check only the lower when PAE */ |
---|
212 | #define pmd_none(x) (!(unsigned long)pmd_val(x)) |
---|
213 | #if CONFIG_XEN_COMPAT <= 0x030002 |
---|
214 | /* pmd_present doesn't just test the _PAGE_PRESENT bit since wr.p.t. |
---|
215 | can temporarily clear it. */ |
---|
216 | #define pmd_present(x) (pmd_val(x)) |
---|
217 | #else |
---|
218 | #define pmd_present(x) (pmd_val(x) & _PAGE_PRESENT) |
---|
219 | #endif |
---|
220 | #define pmd_bad(x) ((pmd_val(x) & (~PAGE_MASK & ~_PAGE_USER & ~_PAGE_PRESENT)) != (_KERNPG_TABLE & ~_PAGE_PRESENT)) |
---|
221 | |
---|
222 | |
---|
223 | #define pages_to_mb(x) ((x) >> (20-PAGE_SHIFT)) |
---|
224 | |
---|
225 | /* |
---|
226 | * The following only work if pte_present() is true. |
---|
227 | * Undefined behaviour if not.. |
---|
228 | */ |
---|
229 | static inline int pte_user(pte_t pte) { return (pte).pte_low & _PAGE_USER; } |
---|
230 | static inline int pte_read(pte_t pte) { return (pte).pte_low & _PAGE_USER; } |
---|
231 | static inline int pte_dirty(pte_t pte) { return (pte).pte_low & _PAGE_DIRTY; } |
---|
232 | static inline int pte_young(pte_t pte) { return (pte).pte_low & _PAGE_ACCESSED; } |
---|
233 | static inline int pte_write(pte_t pte) { return (pte).pte_low & _PAGE_RW; } |
---|
234 | static inline int pte_huge(pte_t pte) { return (pte).pte_low & _PAGE_PSE; } |
---|
235 | |
---|
236 | /* |
---|
237 | * The following only works if pte_present() is not true. |
---|
238 | */ |
---|
239 | static inline int pte_file(pte_t pte) { return (pte).pte_low & _PAGE_FILE; } |
---|
240 | |
---|
241 | static inline pte_t pte_rdprotect(pte_t pte) { (pte).pte_low &= ~_PAGE_USER; return pte; } |
---|
242 | static inline pte_t pte_exprotect(pte_t pte) { (pte).pte_low &= ~_PAGE_USER; return pte; } |
---|
243 | static inline pte_t pte_mkclean(pte_t pte) { (pte).pte_low &= ~_PAGE_DIRTY; return pte; } |
---|
244 | static inline pte_t pte_mkold(pte_t pte) { (pte).pte_low &= ~_PAGE_ACCESSED; return pte; } |
---|
245 | static inline pte_t pte_wrprotect(pte_t pte) { (pte).pte_low &= ~_PAGE_RW; return pte; } |
---|
246 | static inline pte_t pte_mkread(pte_t pte) { (pte).pte_low |= _PAGE_USER; return pte; } |
---|
247 | static inline pte_t pte_mkexec(pte_t pte) { (pte).pte_low |= _PAGE_USER; return pte; } |
---|
248 | static inline pte_t pte_mkdirty(pte_t pte) { (pte).pte_low |= _PAGE_DIRTY; return pte; } |
---|
249 | static inline pte_t pte_mkyoung(pte_t pte) { (pte).pte_low |= _PAGE_ACCESSED; return pte; } |
---|
250 | static inline pte_t pte_mkwrite(pte_t pte) { (pte).pte_low |= _PAGE_RW; return pte; } |
---|
251 | static inline pte_t pte_mkhuge(pte_t pte) { (pte).pte_low |= _PAGE_PSE; return pte; } |
---|
252 | |
---|
253 | #ifdef CONFIG_X86_PAE |
---|
254 | # include <asm/pgtable-3level.h> |
---|
255 | #else |
---|
256 | # include <asm/pgtable-2level.h> |
---|
257 | #endif |
---|
258 | |
---|
259 | #define ptep_test_and_clear_dirty(vma, addr, ptep) \ |
---|
260 | ({ \ |
---|
261 | pte_t __pte = *(ptep); \ |
---|
262 | int __ret = pte_dirty(__pte); \ |
---|
263 | if (__ret) { \ |
---|
264 | __pte = pte_mkclean(__pte); \ |
---|
265 | if ((vma)->vm_mm != current->mm || \ |
---|
266 | HYPERVISOR_update_va_mapping(addr, __pte, 0)) \ |
---|
267 | (ptep)->pte_low = __pte.pte_low; \ |
---|
268 | } \ |
---|
269 | __ret; \ |
---|
270 | }) |
---|
271 | |
---|
272 | #define ptep_test_and_clear_young(vma, addr, ptep) \ |
---|
273 | ({ \ |
---|
274 | pte_t __pte = *(ptep); \ |
---|
275 | int __ret = pte_young(__pte); \ |
---|
276 | if (__ret) \ |
---|
277 | __pte = pte_mkold(__pte); \ |
---|
278 | if ((vma)->vm_mm != current->mm || \ |
---|
279 | HYPERVISOR_update_va_mapping(addr, __pte, 0)) \ |
---|
280 | (ptep)->pte_low = __pte.pte_low; \ |
---|
281 | __ret; \ |
---|
282 | }) |
---|
283 | |
---|
284 | #define ptep_get_and_clear_full(mm, addr, ptep, full) \ |
---|
285 | ((full) ? ({ \ |
---|
286 | pte_t __res = *(ptep); \ |
---|
287 | if (test_bit(PG_pinned, &virt_to_page((mm)->pgd)->flags)) \ |
---|
288 | xen_l1_entry_update(ptep, __pte(0)); \ |
---|
289 | else \ |
---|
290 | *(ptep) = __pte(0); \ |
---|
291 | __res; \ |
---|
292 | }) : \ |
---|
293 | ptep_get_and_clear(mm, addr, ptep)) |
---|
294 | |
---|
295 | static inline void ptep_set_wrprotect(struct mm_struct *mm, unsigned long addr, pte_t *ptep) |
---|
296 | { |
---|
297 | pte_t pte = *ptep; |
---|
298 | if (pte_write(pte)) |
---|
299 | set_pte_at(mm, addr, ptep, pte_wrprotect(pte)); |
---|
300 | } |
---|
301 | |
---|
302 | /* |
---|
303 | * clone_pgd_range(pgd_t *dst, pgd_t *src, int count); |
---|
304 | * |
---|
305 | * dst - pointer to pgd range anwhere on a pgd page |
---|
306 | * src - "" |
---|
307 | * count - the number of pgds to copy. |
---|
308 | * |
---|
309 | * dst and src can be on the same page, but the range must not overlap, |
---|
310 | * and must not cross a page boundary. |
---|
311 | */ |
---|
312 | static inline void clone_pgd_range(pgd_t *dst, pgd_t *src, int count) |
---|
313 | { |
---|
314 | memcpy(dst, src, count * sizeof(pgd_t)); |
---|
315 | } |
---|
316 | |
---|
317 | /* |
---|
318 | * Macro to mark a page protection value as "uncacheable". On processors which do not support |
---|
319 | * it, this is a no-op. |
---|
320 | */ |
---|
321 | #define pgprot_noncached(prot) ((boot_cpu_data.x86 > 3) \ |
---|
322 | ? (__pgprot(pgprot_val(prot) | _PAGE_PCD | _PAGE_PWT)) : (prot)) |
---|
323 | |
---|
324 | /* |
---|
325 | * Conversion functions: convert a page and protection to a page entry, |
---|
326 | * and a page entry and page directory to the page they refer to. |
---|
327 | */ |
---|
328 | |
---|
329 | #define mk_pte(page, pgprot) pfn_pte(page_to_pfn(page), (pgprot)) |
---|
330 | |
---|
331 | static inline pte_t pte_modify(pte_t pte, pgprot_t newprot) |
---|
332 | { |
---|
333 | /* |
---|
334 | * Since this might change the present bit (which controls whether |
---|
335 | * a pte_t object has undergone p2m translation), we must use |
---|
336 | * pte_val() on the input pte and __pte() for the return value. |
---|
337 | */ |
---|
338 | paddr_t pteval = pte_val(pte); |
---|
339 | |
---|
340 | pteval &= _PAGE_CHG_MASK; |
---|
341 | pteval |= pgprot_val(newprot); |
---|
342 | #ifdef CONFIG_X86_PAE |
---|
343 | pteval &= __supported_pte_mask; |
---|
344 | #endif |
---|
345 | return __pte(pteval); |
---|
346 | } |
---|
347 | |
---|
348 | #define pmd_large(pmd) \ |
---|
349 | ((pmd_val(pmd) & (_PAGE_PSE|_PAGE_PRESENT)) == (_PAGE_PSE|_PAGE_PRESENT)) |
---|
350 | |
---|
351 | /* |
---|
352 | * the pgd page can be thought of an array like this: pgd_t[PTRS_PER_PGD] |
---|
353 | * |
---|
354 | * this macro returns the index of the entry in the pgd page which would |
---|
355 | * control the given virtual address |
---|
356 | */ |
---|
357 | #define pgd_index(address) (((address) >> PGDIR_SHIFT) & (PTRS_PER_PGD-1)) |
---|
358 | #define pgd_index_k(addr) pgd_index(addr) |
---|
359 | |
---|
360 | /* |
---|
361 | * pgd_offset() returns a (pgd_t *) |
---|
362 | * pgd_index() is used get the offset into the pgd page's array of pgd_t's; |
---|
363 | */ |
---|
364 | #define pgd_offset(mm, address) ((mm)->pgd+pgd_index(address)) |
---|
365 | |
---|
366 | /* |
---|
367 | * a shortcut which implies the use of the kernel's pgd, instead |
---|
368 | * of a process's |
---|
369 | */ |
---|
370 | #define pgd_offset_k(address) pgd_offset(&init_mm, address) |
---|
371 | |
---|
372 | /* |
---|
373 | * the pmd page can be thought of an array like this: pmd_t[PTRS_PER_PMD] |
---|
374 | * |
---|
375 | * this macro returns the index of the entry in the pmd page which would |
---|
376 | * control the given virtual address |
---|
377 | */ |
---|
378 | #define pmd_index(address) \ |
---|
379 | (((address) >> PMD_SHIFT) & (PTRS_PER_PMD-1)) |
---|
380 | |
---|
381 | /* |
---|
382 | * the pte page can be thought of an array like this: pte_t[PTRS_PER_PTE] |
---|
383 | * |
---|
384 | * this macro returns the index of the entry in the pte page which would |
---|
385 | * control the given virtual address |
---|
386 | */ |
---|
387 | #define pte_index(address) \ |
---|
388 | (((address) >> PAGE_SHIFT) & (PTRS_PER_PTE - 1)) |
---|
389 | #define pte_offset_kernel(dir, address) \ |
---|
390 | ((pte_t *) pmd_page_kernel(*(dir)) + pte_index(address)) |
---|
391 | |
---|
392 | #define pmd_page(pmd) (pfn_to_page(pmd_val(pmd) >> PAGE_SHIFT)) |
---|
393 | |
---|
394 | #define pmd_page_kernel(pmd) \ |
---|
395 | ((unsigned long) __va(pmd_val(pmd) & PAGE_MASK)) |
---|
396 | |
---|
397 | /* |
---|
398 | * Helper function that returns the kernel pagetable entry controlling |
---|
399 | * the virtual address 'address'. NULL means no pagetable entry present. |
---|
400 | * NOTE: the return type is pte_t but if the pmd is PSE then we return it |
---|
401 | * as a pte too. |
---|
402 | */ |
---|
403 | extern pte_t *lookup_address(unsigned long address); |
---|
404 | |
---|
405 | /* |
---|
406 | * Make a given kernel text page executable/non-executable. |
---|
407 | * Returns the previous executability setting of that page (which |
---|
408 | * is used to restore the previous state). Used by the SMP bootup code. |
---|
409 | * NOTE: this is an __init function for security reasons. |
---|
410 | */ |
---|
411 | #ifdef CONFIG_X86_PAE |
---|
412 | extern int set_kernel_exec(unsigned long vaddr, int enable); |
---|
413 | #else |
---|
414 | static inline int set_kernel_exec(unsigned long vaddr, int enable) { return 0;} |
---|
415 | #endif |
---|
416 | |
---|
417 | extern void noexec_setup(const char *str); |
---|
418 | |
---|
419 | #if defined(CONFIG_HIGHPTE) |
---|
420 | #define pte_offset_map(dir, address) \ |
---|
421 | ((pte_t *)kmap_atomic_pte(pmd_page(*(dir)),KM_PTE0) + \ |
---|
422 | pte_index(address)) |
---|
423 | #define pte_offset_map_nested(dir, address) \ |
---|
424 | ((pte_t *)kmap_atomic_pte(pmd_page(*(dir)),KM_PTE1) + \ |
---|
425 | pte_index(address)) |
---|
426 | #define pte_unmap(pte) kunmap_atomic(pte, KM_PTE0) |
---|
427 | #define pte_unmap_nested(pte) kunmap_atomic(pte, KM_PTE1) |
---|
428 | #else |
---|
429 | #define pte_offset_map(dir, address) \ |
---|
430 | ((pte_t *)page_address(pmd_page(*(dir))) + pte_index(address)) |
---|
431 | #define pte_offset_map_nested(dir, address) pte_offset_map(dir, address) |
---|
432 | #define pte_unmap(pte) do { } while (0) |
---|
433 | #define pte_unmap_nested(pte) do { } while (0) |
---|
434 | #endif |
---|
435 | |
---|
436 | #define __HAVE_ARCH_PTEP_ESTABLISH |
---|
437 | #define ptep_establish(vma, address, ptep, pteval) \ |
---|
438 | do { \ |
---|
439 | if ( likely((vma)->vm_mm == current->mm) ) { \ |
---|
440 | BUG_ON(HYPERVISOR_update_va_mapping(address, \ |
---|
441 | pteval, \ |
---|
442 | (unsigned long)(vma)->vm_mm->cpu_vm_mask.bits| \ |
---|
443 | UVMF_INVLPG|UVMF_MULTI)); \ |
---|
444 | } else { \ |
---|
445 | xen_l1_entry_update(ptep, pteval); \ |
---|
446 | flush_tlb_page(vma, address); \ |
---|
447 | } \ |
---|
448 | } while (0) |
---|
449 | |
---|
450 | /* |
---|
451 | * The i386 doesn't have any external MMU info: the kernel page |
---|
452 | * tables contain all the necessary information. |
---|
453 | * |
---|
454 | * Also, we only update the dirty/accessed state if we set |
---|
455 | * the dirty bit by hand in the kernel, since the hardware |
---|
456 | * will do the accessed bit for us, and we don't want to |
---|
457 | * race with other CPU's that might be updating the dirty |
---|
458 | * bit at the same time. |
---|
459 | */ |
---|
460 | #define update_mmu_cache(vma,address,pte) do { } while (0) |
---|
461 | #define __HAVE_ARCH_PTEP_SET_ACCESS_FLAGS |
---|
462 | #define ptep_set_access_flags(vma, address, ptep, entry, dirty) \ |
---|
463 | do { \ |
---|
464 | if (dirty) \ |
---|
465 | ptep_establish(vma, address, ptep, entry); \ |
---|
466 | } while (0) |
---|
467 | |
---|
468 | #include <xen/features.h> |
---|
469 | void make_lowmem_page_readonly(void *va, unsigned int feature); |
---|
470 | void make_lowmem_page_writable(void *va, unsigned int feature); |
---|
471 | void make_page_readonly(void *va, unsigned int feature); |
---|
472 | void make_page_writable(void *va, unsigned int feature); |
---|
473 | void make_pages_readonly(void *va, unsigned int nr, unsigned int feature); |
---|
474 | void make_pages_writable(void *va, unsigned int nr, unsigned int feature); |
---|
475 | |
---|
476 | #define virt_to_ptep(__va) \ |
---|
477 | ({ \ |
---|
478 | pgd_t *__pgd = pgd_offset_k((unsigned long)(__va)); \ |
---|
479 | pud_t *__pud = pud_offset(__pgd, (unsigned long)(__va)); \ |
---|
480 | pmd_t *__pmd = pmd_offset(__pud, (unsigned long)(__va)); \ |
---|
481 | pte_offset_kernel(__pmd, (unsigned long)(__va)); \ |
---|
482 | }) |
---|
483 | |
---|
484 | #define arbitrary_virt_to_machine(__va) \ |
---|
485 | ({ \ |
---|
486 | maddr_t m = (maddr_t)pte_mfn(*virt_to_ptep(__va)) << PAGE_SHIFT;\ |
---|
487 | m | ((unsigned long)(__va) & (PAGE_SIZE-1)); \ |
---|
488 | }) |
---|
489 | |
---|
490 | #endif /* !__ASSEMBLY__ */ |
---|
491 | |
---|
492 | #ifdef CONFIG_FLATMEM |
---|
493 | #define kern_addr_valid(addr) (1) |
---|
494 | #endif /* CONFIG_FLATMEM */ |
---|
495 | |
---|
496 | int direct_remap_pfn_range(struct vm_area_struct *vma, |
---|
497 | unsigned long address, |
---|
498 | unsigned long mfn, |
---|
499 | unsigned long size, |
---|
500 | pgprot_t prot, |
---|
501 | domid_t domid); |
---|
502 | int direct_kernel_remap_pfn_range(unsigned long address, |
---|
503 | unsigned long mfn, |
---|
504 | unsigned long size, |
---|
505 | pgprot_t prot, |
---|
506 | domid_t domid); |
---|
507 | int create_lookup_pte_addr(struct mm_struct *mm, |
---|
508 | unsigned long address, |
---|
509 | uint64_t *ptep); |
---|
510 | int touch_pte_range(struct mm_struct *mm, |
---|
511 | unsigned long address, |
---|
512 | unsigned long size); |
---|
513 | |
---|
514 | #define io_remap_pfn_range(vma,from,pfn,size,prot) \ |
---|
515 | direct_remap_pfn_range(vma,from,pfn,size,prot,DOMID_IO) |
---|
516 | |
---|
517 | #define MK_IOSPACE_PFN(space, pfn) (pfn) |
---|
518 | #define GET_IOSPACE(pfn) 0 |
---|
519 | #define GET_PFN(pfn) (pfn) |
---|
520 | |
---|
521 | #define __HAVE_ARCH_PTEP_TEST_AND_CLEAR_YOUNG |
---|
522 | #define __HAVE_ARCH_PTEP_TEST_AND_CLEAR_DIRTY |
---|
523 | #define __HAVE_ARCH_PTEP_GET_AND_CLEAR |
---|
524 | #define __HAVE_ARCH_PTEP_GET_AND_CLEAR_FULL |
---|
525 | #define __HAVE_ARCH_PTEP_CLEAR_FLUSH |
---|
526 | #define __HAVE_ARCH_PTEP_SET_WRPROTECT |
---|
527 | #define __HAVE_ARCH_PTE_SAME |
---|
528 | #include <asm-generic/pgtable.h> |
---|
529 | |
---|
530 | #endif /* _I386_PGTABLE_H */ |
---|