1 #ifndef ASM_X86__PGTABLE_H
2 #define ASM_X86__PGTABLE_H
4 #define FIRST_USER_ADDRESS 0
6 #define _PAGE_BIT_PRESENT 0 /* is present */
7 #define _PAGE_BIT_RW 1 /* writeable */
8 #define _PAGE_BIT_USER 2 /* userspace addressable */
9 #define _PAGE_BIT_PWT 3 /* page write through */
10 #define _PAGE_BIT_PCD 4 /* page cache disabled */
11 #define _PAGE_BIT_ACCESSED 5 /* was accessed (raised by CPU) */
12 #define _PAGE_BIT_DIRTY 6 /* was written to (raised by CPU) */
13 #define _PAGE_BIT_FILE 6
14 #define _PAGE_BIT_PSE 7 /* 4 MB (or 2MB) page */
15 #define _PAGE_BIT_PAT 7 /* on 4KB pages */
16 #define _PAGE_BIT_GLOBAL 8 /* Global TLB entry PPro+ */
17 #define _PAGE_BIT_UNUSED1 9 /* available for programmer */
18 #define _PAGE_BIT_UNUSED2 10
19 #define _PAGE_BIT_UNUSED3 11
20 #define _PAGE_BIT_PAT_LARGE 12 /* On 2MB or 1GB pages */
21 #define _PAGE_BIT_SPECIAL _PAGE_BIT_UNUSED1
22 #define _PAGE_BIT_CPA_TEST _PAGE_BIT_UNUSED1
23 #define _PAGE_BIT_NX 63 /* No execute: only valid after cpuid check */
25 #define _PAGE_PRESENT (_AT(pteval_t, 1) << _PAGE_BIT_PRESENT)
26 #define _PAGE_RW (_AT(pteval_t, 1) << _PAGE_BIT_RW)
27 #define _PAGE_USER (_AT(pteval_t, 1) << _PAGE_BIT_USER)
28 #define _PAGE_PWT (_AT(pteval_t, 1) << _PAGE_BIT_PWT)
29 #define _PAGE_PCD (_AT(pteval_t, 1) << _PAGE_BIT_PCD)
30 #define _PAGE_ACCESSED (_AT(pteval_t, 1) << _PAGE_BIT_ACCESSED)
31 #define _PAGE_DIRTY (_AT(pteval_t, 1) << _PAGE_BIT_DIRTY)
32 #define _PAGE_PSE (_AT(pteval_t, 1) << _PAGE_BIT_PSE)
33 #define _PAGE_GLOBAL (_AT(pteval_t, 1) << _PAGE_BIT_GLOBAL)
34 #define _PAGE_UNUSED1 (_AT(pteval_t, 1) << _PAGE_BIT_UNUSED1)
35 #define _PAGE_UNUSED2 (_AT(pteval_t, 1) << _PAGE_BIT_UNUSED2)
36 #define _PAGE_UNUSED3 (_AT(pteval_t, 1) << _PAGE_BIT_UNUSED3)
37 #define _PAGE_PAT (_AT(pteval_t, 1) << _PAGE_BIT_PAT)
38 #define _PAGE_PAT_LARGE (_AT(pteval_t, 1) << _PAGE_BIT_PAT_LARGE)
39 #define _PAGE_SPECIAL (_AT(pteval_t, 1) << _PAGE_BIT_SPECIAL)
40 #define _PAGE_CPA_TEST (_AT(pteval_t, 1) << _PAGE_BIT_CPA_TEST)
41 #define __HAVE_ARCH_PTE_SPECIAL
43 #if defined(CONFIG_X86_64) || defined(CONFIG_X86_PAE)
44 #define _PAGE_NX (_AT(pteval_t, 1) << _PAGE_BIT_NX)
46 #define _PAGE_NX (_AT(pteval_t, 0))
49 /* If _PAGE_PRESENT is clear, we use these: */
50 #define _PAGE_FILE _PAGE_DIRTY /* nonlinear file mapping,
51 * saved PTE; unset:swap */
52 #define _PAGE_PROTNONE _PAGE_PSE /* if the user mapped it with PROT_NONE;
53 pte_present gives true */
55 #define _PAGE_TABLE (_PAGE_PRESENT | _PAGE_RW | _PAGE_USER | \
56 _PAGE_ACCESSED | _PAGE_DIRTY)
57 #define _KERNPG_TABLE (_PAGE_PRESENT | _PAGE_RW | _PAGE_ACCESSED | \
60 /* Set of bits not changed in pte_modify */
61 #define _PAGE_CHG_MASK (PTE_PFN_MASK | _PAGE_PCD | _PAGE_PWT | \
62 _PAGE_SPECIAL | _PAGE_ACCESSED | _PAGE_DIRTY)
64 #define _PAGE_CACHE_MASK (_PAGE_PCD | _PAGE_PWT)
65 #define _PAGE_CACHE_WB (0)
66 #define _PAGE_CACHE_WC (_PAGE_PWT)
67 #define _PAGE_CACHE_UC_MINUS (_PAGE_PCD)
68 #define _PAGE_CACHE_UC (_PAGE_PCD | _PAGE_PWT)
70 #define PAGE_NONE __pgprot(_PAGE_PROTNONE | _PAGE_ACCESSED)
71 #define PAGE_SHARED __pgprot(_PAGE_PRESENT | _PAGE_RW | _PAGE_USER | \
72 _PAGE_ACCESSED | _PAGE_NX)
74 #define PAGE_SHARED_EXEC __pgprot(_PAGE_PRESENT | _PAGE_RW | \
75 _PAGE_USER | _PAGE_ACCESSED)
76 #define PAGE_COPY_NOEXEC __pgprot(_PAGE_PRESENT | _PAGE_USER | \
77 _PAGE_ACCESSED | _PAGE_NX)
78 #define PAGE_COPY_EXEC __pgprot(_PAGE_PRESENT | _PAGE_USER | \
80 #define PAGE_COPY PAGE_COPY_NOEXEC
81 #define PAGE_READONLY __pgprot(_PAGE_PRESENT | _PAGE_USER | \
82 _PAGE_ACCESSED | _PAGE_NX)
83 #define PAGE_READONLY_EXEC __pgprot(_PAGE_PRESENT | _PAGE_USER | \
86 #define __PAGE_KERNEL_EXEC \
87 (_PAGE_PRESENT | _PAGE_RW | _PAGE_DIRTY | _PAGE_ACCESSED | _PAGE_GLOBAL)
88 #define __PAGE_KERNEL (__PAGE_KERNEL_EXEC | _PAGE_NX)
90 #define __PAGE_KERNEL_RO (__PAGE_KERNEL & ~_PAGE_RW)
91 #define __PAGE_KERNEL_RX (__PAGE_KERNEL_EXEC & ~_PAGE_RW)
92 #define __PAGE_KERNEL_EXEC_NOCACHE (__PAGE_KERNEL_EXEC | _PAGE_PCD | _PAGE_PWT)
93 #define __PAGE_KERNEL_WC (__PAGE_KERNEL | _PAGE_CACHE_WC)
94 #define __PAGE_KERNEL_NOCACHE (__PAGE_KERNEL | _PAGE_PCD | _PAGE_PWT)
95 #define __PAGE_KERNEL_UC_MINUS (__PAGE_KERNEL | _PAGE_PCD)
96 #define __PAGE_KERNEL_VSYSCALL (__PAGE_KERNEL_RX | _PAGE_USER)
97 #define __PAGE_KERNEL_VSYSCALL_NOCACHE (__PAGE_KERNEL_VSYSCALL | _PAGE_PCD | _PAGE_PWT)
98 #define __PAGE_KERNEL_LARGE (__PAGE_KERNEL | _PAGE_PSE)
99 #define __PAGE_KERNEL_LARGE_NOCACHE (__PAGE_KERNEL | _PAGE_CACHE_UC | _PAGE_PSE)
100 #define __PAGE_KERNEL_LARGE_EXEC (__PAGE_KERNEL_EXEC | _PAGE_PSE)
102 #define PAGE_KERNEL __pgprot(__PAGE_KERNEL)
103 #define PAGE_KERNEL_RO __pgprot(__PAGE_KERNEL_RO)
104 #define PAGE_KERNEL_EXEC __pgprot(__PAGE_KERNEL_EXEC)
105 #define PAGE_KERNEL_RX __pgprot(__PAGE_KERNEL_RX)
106 #define PAGE_KERNEL_WC __pgprot(__PAGE_KERNEL_WC)
107 #define PAGE_KERNEL_NOCACHE __pgprot(__PAGE_KERNEL_NOCACHE)
108 #define PAGE_KERNEL_UC_MINUS __pgprot(__PAGE_KERNEL_UC_MINUS)
109 #define PAGE_KERNEL_EXEC_NOCACHE __pgprot(__PAGE_KERNEL_EXEC_NOCACHE)
110 #define PAGE_KERNEL_LARGE __pgprot(__PAGE_KERNEL_LARGE)
111 #define PAGE_KERNEL_LARGE_NOCACHE __pgprot(__PAGE_KERNEL_LARGE_NOCACHE)
112 #define PAGE_KERNEL_LARGE_EXEC __pgprot(__PAGE_KERNEL_LARGE_EXEC)
113 #define PAGE_KERNEL_VSYSCALL __pgprot(__PAGE_KERNEL_VSYSCALL)
114 #define PAGE_KERNEL_VSYSCALL_NOCACHE __pgprot(__PAGE_KERNEL_VSYSCALL_NOCACHE)
117 #define __P000 PAGE_NONE
118 #define __P001 PAGE_READONLY
119 #define __P010 PAGE_COPY
120 #define __P011 PAGE_COPY
121 #define __P100 PAGE_READONLY_EXEC
122 #define __P101 PAGE_READONLY_EXEC
123 #define __P110 PAGE_COPY_EXEC
124 #define __P111 PAGE_COPY_EXEC
126 #define __S000 PAGE_NONE
127 #define __S001 PAGE_READONLY
128 #define __S010 PAGE_SHARED
129 #define __S011 PAGE_SHARED
130 #define __S100 PAGE_READONLY_EXEC
131 #define __S101 PAGE_READONLY_EXEC
132 #define __S110 PAGE_SHARED_EXEC
133 #define __S111 PAGE_SHARED_EXEC
136 * early identity mapping pte attrib macros.
139 #define __PAGE_KERNEL_IDENT_LARGE_EXEC __PAGE_KERNEL_LARGE_EXEC
141 #define PTE_IDENT_ATTR 0x003 /* PRESENT+RW */
142 #define PDE_IDENT_ATTR 0x063 /* PRESENT+RW+DIRTY+ACCESSED */
143 #define PGD_IDENT_ATTR 0x001 /* PRESENT (no other attributes) */
149 * ZERO_PAGE is a global shared page that is always zero: used
150 * for zero-mapped memory areas etc..
152 extern unsigned long empty_zero_page
[PAGE_SIZE
/ sizeof(unsigned long)];
153 #define ZERO_PAGE(vaddr) (virt_to_page(empty_zero_page))
155 extern spinlock_t pgd_lock
;
156 extern struct list_head pgd_list
;
159 * The following only work if pte_present() is true.
160 * Undefined behaviour if not..
162 static inline int pte_dirty(pte_t pte
)
164 return pte_flags(pte
) & _PAGE_DIRTY
;
167 static inline int pte_young(pte_t pte
)
169 return pte_flags(pte
) & _PAGE_ACCESSED
;
172 static inline int pte_write(pte_t pte
)
174 return pte_flags(pte
) & _PAGE_RW
;
177 static inline int pte_file(pte_t pte
)
179 return pte_flags(pte
) & _PAGE_FILE
;
182 static inline int pte_huge(pte_t pte
)
184 return pte_flags(pte
) & _PAGE_PSE
;
187 static inline int pte_global(pte_t pte
)
189 return pte_flags(pte
) & _PAGE_GLOBAL
;
192 static inline int pte_exec(pte_t pte
)
194 return !(pte_flags(pte
) & _PAGE_NX
);
197 static inline int pte_special(pte_t pte
)
199 return pte_val(pte
) & _PAGE_SPECIAL
;
202 static inline unsigned long pte_pfn(pte_t pte
)
204 return (pte_val(pte
) & PTE_PFN_MASK
) >> PAGE_SHIFT
;
207 #define pte_page(pte) pfn_to_page(pte_pfn(pte))
209 static inline int pmd_large(pmd_t pte
)
211 return (pmd_val(pte
) & (_PAGE_PSE
| _PAGE_PRESENT
)) ==
212 (_PAGE_PSE
| _PAGE_PRESENT
);
215 static inline pte_t
pte_mkclean(pte_t pte
)
217 return __pte(pte_val(pte
) & ~_PAGE_DIRTY
);
220 static inline pte_t
pte_mkold(pte_t pte
)
222 return __pte(pte_val(pte
) & ~_PAGE_ACCESSED
);
225 static inline pte_t
pte_wrprotect(pte_t pte
)
227 return __pte(pte_val(pte
) & ~_PAGE_RW
);
230 static inline pte_t
pte_mkexec(pte_t pte
)
232 return __pte(pte_val(pte
) & ~_PAGE_NX
);
235 static inline pte_t
pte_mkdirty(pte_t pte
)
237 return __pte(pte_val(pte
) | _PAGE_DIRTY
);
240 static inline pte_t
pte_mkyoung(pte_t pte
)
242 return __pte(pte_val(pte
) | _PAGE_ACCESSED
);
245 static inline pte_t
pte_mkwrite(pte_t pte
)
247 return __pte(pte_val(pte
) | _PAGE_RW
);
250 static inline pte_t
pte_mkhuge(pte_t pte
)
252 return __pte(pte_val(pte
) | _PAGE_PSE
);
255 static inline pte_t
pte_clrhuge(pte_t pte
)
257 return __pte(pte_val(pte
) & ~_PAGE_PSE
);
260 static inline pte_t
pte_mkglobal(pte_t pte
)
262 return __pte(pte_val(pte
) | _PAGE_GLOBAL
);
265 static inline pte_t
pte_clrglobal(pte_t pte
)
267 return __pte(pte_val(pte
) & ~_PAGE_GLOBAL
);
270 static inline pte_t
pte_mkspecial(pte_t pte
)
272 return __pte(pte_val(pte
) | _PAGE_SPECIAL
);
275 extern pteval_t __supported_pte_mask
;
277 static inline pte_t
pfn_pte(unsigned long page_nr
, pgprot_t pgprot
)
279 return __pte((((phys_addr_t
)page_nr
<< PAGE_SHIFT
) |
280 pgprot_val(pgprot
)) & __supported_pte_mask
);
283 static inline pmd_t
pfn_pmd(unsigned long page_nr
, pgprot_t pgprot
)
285 return __pmd((((phys_addr_t
)page_nr
<< PAGE_SHIFT
) |
286 pgprot_val(pgprot
)) & __supported_pte_mask
);
289 static inline pte_t
pte_modify(pte_t pte
, pgprot_t newprot
)
291 pteval_t val
= pte_val(pte
);
294 * Chop off the NX bit (if present), and add the NX portion of
295 * the newprot (if present):
297 val
&= _PAGE_CHG_MASK
;
298 val
|= pgprot_val(newprot
) & (~_PAGE_CHG_MASK
) & __supported_pte_mask
;
303 /* mprotect needs to preserve PAT bits when updating vm_page_prot */
304 #define pgprot_modify pgprot_modify
305 static inline pgprot_t
pgprot_modify(pgprot_t oldprot
, pgprot_t newprot
)
307 pgprotval_t preservebits
= pgprot_val(oldprot
) & _PAGE_CHG_MASK
;
308 pgprotval_t addbits
= pgprot_val(newprot
);
309 return __pgprot(preservebits
| addbits
);
312 #define pte_pgprot(x) __pgprot(pte_flags(x) & PTE_FLAGS_MASK)
314 #define canon_pgprot(p) __pgprot(pgprot_val(p) & __supported_pte_mask)
317 #define __HAVE_PHYS_MEM_ACCESS_PROT
319 pgprot_t
phys_mem_access_prot(struct file
*file
, unsigned long pfn
,
320 unsigned long size
, pgprot_t vma_prot
);
321 int phys_mem_access_prot_allowed(struct file
*file
, unsigned long pfn
,
322 unsigned long size
, pgprot_t
*vma_prot
);
325 /* Install a pte for a particular vaddr in kernel space. */
326 void set_pte_vaddr(unsigned long vaddr
, pte_t pte
);
329 extern void native_pagetable_setup_start(pgd_t
*base
);
330 extern void native_pagetable_setup_done(pgd_t
*base
);
332 static inline void native_pagetable_setup_start(pgd_t
*base
) {}
333 static inline void native_pagetable_setup_done(pgd_t
*base
) {}
336 extern int arch_report_meminfo(char *page
);
338 #ifdef CONFIG_PARAVIRT
339 #include <asm/paravirt.h>
340 #else /* !CONFIG_PARAVIRT */
341 #define set_pte(ptep, pte) native_set_pte(ptep, pte)
342 #define set_pte_at(mm, addr, ptep, pte) native_set_pte_at(mm, addr, ptep, pte)
344 #define set_pte_present(mm, addr, ptep, pte) \
345 native_set_pte_present(mm, addr, ptep, pte)
346 #define set_pte_atomic(ptep, pte) \
347 native_set_pte_atomic(ptep, pte)
349 #define set_pmd(pmdp, pmd) native_set_pmd(pmdp, pmd)
351 #ifndef __PAGETABLE_PUD_FOLDED
352 #define set_pgd(pgdp, pgd) native_set_pgd(pgdp, pgd)
353 #define pgd_clear(pgd) native_pgd_clear(pgd)
357 # define set_pud(pudp, pud) native_set_pud(pudp, pud)
360 #ifndef __PAGETABLE_PMD_FOLDED
361 #define pud_clear(pud) native_pud_clear(pud)
364 #define pte_clear(mm, addr, ptep) native_pte_clear(mm, addr, ptep)
365 #define pmd_clear(pmd) native_pmd_clear(pmd)
367 #define pte_update(mm, addr, ptep) do { } while (0)
368 #define pte_update_defer(mm, addr, ptep) do { } while (0)
370 static inline void __init
paravirt_pagetable_setup_start(pgd_t
*base
)
372 native_pagetable_setup_start(base
);
375 static inline void __init
paravirt_pagetable_setup_done(pgd_t
*base
)
377 native_pagetable_setup_done(base
);
379 #endif /* CONFIG_PARAVIRT */
381 #endif /* __ASSEMBLY__ */
384 # include "pgtable_32.h"
386 # include "pgtable_64.h"
390 * the pgd page can be thought of an array like this: pgd_t[PTRS_PER_PGD]
392 * this macro returns the index of the entry in the pgd page which would
393 * control the given virtual address
395 #define pgd_index(address) (((address) >> PGDIR_SHIFT) & (PTRS_PER_PGD - 1))
398 * pgd_offset() returns a (pgd_t *)
399 * pgd_index() is used get the offset into the pgd page's array of pgd_t's;
401 #define pgd_offset(mm, address) ((mm)->pgd + pgd_index((address)))
403 * a shortcut which implies the use of the kernel's pgd, instead
406 #define pgd_offset_k(address) pgd_offset(&init_mm, (address))
409 #define KERNEL_PGD_BOUNDARY pgd_index(PAGE_OFFSET)
410 #define KERNEL_PGD_PTRS (PTRS_PER_PGD - KERNEL_PGD_BOUNDARY)
422 #ifdef CONFIG_PROC_FS
423 extern void update_page_count(int level
, unsigned long pages
);
425 static inline void update_page_count(int level
, unsigned long pages
) { }
429 * Helper function that returns the kernel pagetable entry controlling
430 * the virtual address 'address'. NULL means no pagetable entry present.
431 * NOTE: the return type is pte_t but if the pmd is PSE then we return it
434 extern pte_t
*lookup_address(unsigned long address
, unsigned int *level
);
436 /* local pte updates need not use xchg for locking */
437 static inline pte_t
native_local_ptep_get_and_clear(pte_t
*ptep
)
441 /* Pure native function needs no input for mm, addr */
442 native_pte_clear(NULL
, 0, ptep
);
446 static inline void native_set_pte_at(struct mm_struct
*mm
, unsigned long addr
,
447 pte_t
*ptep
, pte_t pte
)
449 native_set_pte(ptep
, pte
);
452 #ifndef CONFIG_PARAVIRT
454 * Rules for using pte_update - it must be called after any PTE update which
455 * has not been done using the set_pte / clear_pte interfaces. It is used by
456 * shadow mode hypervisors to resynchronize the shadow page tables. Kernel PTE
457 * updates should either be sets, clears, or set_pte_atomic for P->P
458 * transitions, which means this hook should only be called for user PTEs.
459 * This hook implies a P->P protection or access change has taken place, which
460 * requires a subsequent TLB flush. The notification can optionally be delayed
461 * until the TLB flush event by using the pte_update_defer form of the
462 * interface, but care must be taken to assure that the flush happens while
463 * still holding the same page table lock so that the shadow and primary pages
464 * do not become out of sync on SMP.
466 #define pte_update(mm, addr, ptep) do { } while (0)
467 #define pte_update_defer(mm, addr, ptep) do { } while (0)
471 * We only update the dirty/accessed state if we set
472 * the dirty bit by hand in the kernel, since the hardware
473 * will do the accessed bit for us, and we don't want to
474 * race with other CPU's that might be updating the dirty
475 * bit at the same time.
477 struct vm_area_struct
;
479 #define __HAVE_ARCH_PTEP_SET_ACCESS_FLAGS
480 extern int ptep_set_access_flags(struct vm_area_struct
*vma
,
481 unsigned long address
, pte_t
*ptep
,
482 pte_t entry
, int dirty
);
484 #define __HAVE_ARCH_PTEP_TEST_AND_CLEAR_YOUNG
485 extern int ptep_test_and_clear_young(struct vm_area_struct
*vma
,
486 unsigned long addr
, pte_t
*ptep
);
488 #define __HAVE_ARCH_PTEP_CLEAR_YOUNG_FLUSH
489 extern int ptep_clear_flush_young(struct vm_area_struct
*vma
,
490 unsigned long address
, pte_t
*ptep
);
492 #define __HAVE_ARCH_PTEP_GET_AND_CLEAR
493 static inline pte_t
ptep_get_and_clear(struct mm_struct
*mm
, unsigned long addr
,
496 pte_t pte
= native_ptep_get_and_clear(ptep
);
497 pte_update(mm
, addr
, ptep
);
501 #define __HAVE_ARCH_PTEP_GET_AND_CLEAR_FULL
502 static inline pte_t
ptep_get_and_clear_full(struct mm_struct
*mm
,
503 unsigned long addr
, pte_t
*ptep
,
509 * Full address destruction in progress; paravirt does not
510 * care about updates and native needs no locking
512 pte
= native_local_ptep_get_and_clear(ptep
);
514 pte
= ptep_get_and_clear(mm
, addr
, ptep
);
519 #define __HAVE_ARCH_PTEP_SET_WRPROTECT
520 static inline void ptep_set_wrprotect(struct mm_struct
*mm
,
521 unsigned long addr
, pte_t
*ptep
)
523 clear_bit(_PAGE_BIT_RW
, (unsigned long *)&ptep
->pte
);
524 pte_update(mm
, addr
, ptep
);
528 * clone_pgd_range(pgd_t *dst, pgd_t *src, int count);
530 * dst - pointer to pgd range anwhere on a pgd page
532 * count - the number of pgds to copy.
534 * dst and src can be on the same page, but the range must not overlap,
535 * and must not cross a page boundary.
537 static inline void clone_pgd_range(pgd_t
*dst
, pgd_t
*src
, int count
)
539 memcpy(dst
, src
, count
* sizeof(pgd_t
));
543 #include <asm-generic/pgtable.h>
544 #endif /* __ASSEMBLY__ */
546 #endif /* ASM_X86__PGTABLE_H */