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_NX 63 /* No execute: only valid after cpuid check */
23 #define _PAGE_PRESENT (_AT(pteval_t, 1) << _PAGE_BIT_PRESENT)
24 #define _PAGE_RW (_AT(pteval_t, 1) << _PAGE_BIT_RW)
25 #define _PAGE_USER (_AT(pteval_t, 1) << _PAGE_BIT_USER)
26 #define _PAGE_PWT (_AT(pteval_t, 1) << _PAGE_BIT_PWT)
27 #define _PAGE_PCD (_AT(pteval_t, 1) << _PAGE_BIT_PCD)
28 #define _PAGE_ACCESSED (_AT(pteval_t, 1) << _PAGE_BIT_ACCESSED)
29 #define _PAGE_DIRTY (_AT(pteval_t, 1) << _PAGE_BIT_DIRTY)
30 #define _PAGE_PSE (_AT(pteval_t, 1) << _PAGE_BIT_PSE)
31 #define _PAGE_GLOBAL (_AT(pteval_t, 1) << _PAGE_BIT_GLOBAL)
32 #define _PAGE_UNUSED1 (_AT(pteval_t, 1) << _PAGE_BIT_UNUSED1)
33 #define _PAGE_UNUSED2 (_AT(pteval_t, 1) << _PAGE_BIT_UNUSED2)
34 #define _PAGE_UNUSED3 (_AT(pteval_t, 1) << _PAGE_BIT_UNUSED3)
35 #define _PAGE_PAT (_AT(pteval_t, 1) << _PAGE_BIT_PAT)
36 #define _PAGE_PAT_LARGE (_AT(pteval_t, 1) << _PAGE_BIT_PAT_LARGE)
38 #if defined(CONFIG_X86_64) || defined(CONFIG_X86_PAE)
39 #define _PAGE_NX (_AT(pteval_t, 1) << _PAGE_BIT_NX)
41 #define _PAGE_NX (_AT(pteval_t, 0))
44 /* If _PAGE_PRESENT is clear, we use these: */
45 #define _PAGE_FILE _PAGE_DIRTY /* nonlinear file mapping,
46 * saved PTE; unset:swap */
47 #define _PAGE_PROTNONE _PAGE_PSE /* if the user mapped it with PROT_NONE;
48 pte_present gives true */
50 #define _PAGE_TABLE (_PAGE_PRESENT | _PAGE_RW | _PAGE_USER | \
51 _PAGE_ACCESSED | _PAGE_DIRTY)
52 #define _KERNPG_TABLE (_PAGE_PRESENT | _PAGE_RW | _PAGE_ACCESSED | \
55 /* Set of bits not changed in pte_modify */
56 #define _PAGE_CHG_MASK (PTE_MASK | _PAGE_PCD | _PAGE_PWT | \
57 _PAGE_ACCESSED | _PAGE_DIRTY)
59 #define _PAGE_CACHE_MASK (_PAGE_PCD | _PAGE_PWT)
60 #define _PAGE_CACHE_WB (0)
61 #define _PAGE_CACHE_WC (_PAGE_PWT)
62 #define _PAGE_CACHE_UC_MINUS (_PAGE_PCD)
63 #define _PAGE_CACHE_UC (_PAGE_PCD | _PAGE_PWT)
65 #define PAGE_NONE __pgprot(_PAGE_PROTNONE | _PAGE_ACCESSED)
66 #define PAGE_SHARED __pgprot(_PAGE_PRESENT | _PAGE_RW | _PAGE_USER | \
67 _PAGE_ACCESSED | _PAGE_NX)
69 #define PAGE_SHARED_EXEC __pgprot(_PAGE_PRESENT | _PAGE_RW | \
70 _PAGE_USER | _PAGE_ACCESSED)
71 #define PAGE_COPY_NOEXEC __pgprot(_PAGE_PRESENT | _PAGE_USER | \
72 _PAGE_ACCESSED | _PAGE_NX)
73 #define PAGE_COPY_EXEC __pgprot(_PAGE_PRESENT | _PAGE_USER | \
75 #define PAGE_COPY PAGE_COPY_NOEXEC
76 #define PAGE_READONLY __pgprot(_PAGE_PRESENT | _PAGE_USER | \
77 _PAGE_ACCESSED | _PAGE_NX)
78 #define PAGE_READONLY_EXEC __pgprot(_PAGE_PRESENT | _PAGE_USER | \
81 #define __PAGE_KERNEL_EXEC \
82 (_PAGE_PRESENT | _PAGE_RW | _PAGE_DIRTY | _PAGE_ACCESSED | _PAGE_GLOBAL)
83 #define __PAGE_KERNEL (__PAGE_KERNEL_EXEC | _PAGE_NX)
85 #define __PAGE_KERNEL_RO (__PAGE_KERNEL & ~_PAGE_RW)
86 #define __PAGE_KERNEL_RX (__PAGE_KERNEL_EXEC & ~_PAGE_RW)
87 #define __PAGE_KERNEL_EXEC_NOCACHE (__PAGE_KERNEL_EXEC | _PAGE_PCD | _PAGE_PWT)
88 #define __PAGE_KERNEL_WC (__PAGE_KERNEL | _PAGE_CACHE_WC)
89 #define __PAGE_KERNEL_NOCACHE (__PAGE_KERNEL | _PAGE_PCD | _PAGE_PWT)
90 #define __PAGE_KERNEL_UC_MINUS (__PAGE_KERNEL | _PAGE_PCD)
91 #define __PAGE_KERNEL_VSYSCALL (__PAGE_KERNEL_RX | _PAGE_USER)
92 #define __PAGE_KERNEL_VSYSCALL_NOCACHE (__PAGE_KERNEL_VSYSCALL | _PAGE_PCD | _PAGE_PWT)
93 #define __PAGE_KERNEL_LARGE (__PAGE_KERNEL | _PAGE_PSE)
94 #define __PAGE_KERNEL_LARGE_EXEC (__PAGE_KERNEL_EXEC | _PAGE_PSE)
96 #define PAGE_KERNEL __pgprot(__PAGE_KERNEL)
97 #define PAGE_KERNEL_RO __pgprot(__PAGE_KERNEL_RO)
98 #define PAGE_KERNEL_EXEC __pgprot(__PAGE_KERNEL_EXEC)
99 #define PAGE_KERNEL_RX __pgprot(__PAGE_KERNEL_RX)
100 #define PAGE_KERNEL_WC __pgprot(__PAGE_KERNEL_WC)
101 #define PAGE_KERNEL_NOCACHE __pgprot(__PAGE_KERNEL_NOCACHE)
102 #define PAGE_KERNEL_UC_MINUS __pgprot(__PAGE_KERNEL_UC_MINUS)
103 #define PAGE_KERNEL_EXEC_NOCACHE __pgprot(__PAGE_KERNEL_EXEC_NOCACHE)
104 #define PAGE_KERNEL_LARGE __pgprot(__PAGE_KERNEL_LARGE)
105 #define PAGE_KERNEL_LARGE_EXEC __pgprot(__PAGE_KERNEL_LARGE_EXEC)
106 #define PAGE_KERNEL_VSYSCALL __pgprot(__PAGE_KERNEL_VSYSCALL)
107 #define PAGE_KERNEL_VSYSCALL_NOCACHE __pgprot(__PAGE_KERNEL_VSYSCALL_NOCACHE)
110 #define __P000 PAGE_NONE
111 #define __P001 PAGE_READONLY
112 #define __P010 PAGE_COPY
113 #define __P011 PAGE_COPY
114 #define __P100 PAGE_READONLY_EXEC
115 #define __P101 PAGE_READONLY_EXEC
116 #define __P110 PAGE_COPY_EXEC
117 #define __P111 PAGE_COPY_EXEC
119 #define __S000 PAGE_NONE
120 #define __S001 PAGE_READONLY
121 #define __S010 PAGE_SHARED
122 #define __S011 PAGE_SHARED
123 #define __S100 PAGE_READONLY_EXEC
124 #define __S101 PAGE_READONLY_EXEC
125 #define __S110 PAGE_SHARED_EXEC
126 #define __S111 PAGE_SHARED_EXEC
131 * ZERO_PAGE is a global shared page that is always zero: used
132 * for zero-mapped memory areas etc..
134 extern unsigned long empty_zero_page[PAGE_SIZE / sizeof(unsigned long)];
135 #define ZERO_PAGE(vaddr) (virt_to_page(empty_zero_page))
137 extern spinlock_t pgd_lock;
138 extern struct list_head pgd_list;
141 * The following only work if pte_present() is true.
142 * Undefined behaviour if not..
144 static inline int pte_dirty(pte_t pte)
146 return pte_flags(pte) & _PAGE_DIRTY;
149 static inline int pte_young(pte_t pte)
151 return pte_flags(pte) & _PAGE_ACCESSED;
154 static inline int pte_write(pte_t pte)
156 return pte_flags(pte) & _PAGE_RW;
159 static inline int pte_file(pte_t pte)
161 return pte_flags(pte) & _PAGE_FILE;
164 static inline int pte_huge(pte_t pte)
166 return pte_flags(pte) & _PAGE_PSE;
169 static inline int pte_global(pte_t pte)
171 return pte_flags(pte) & _PAGE_GLOBAL;
174 static inline int pte_exec(pte_t pte)
176 return !(pte_flags(pte) & _PAGE_NX);
179 static inline int pte_special(pte_t pte)
184 static inline int pmd_large(pmd_t pte)
186 return (pmd_val(pte) & (_PAGE_PSE | _PAGE_PRESENT)) ==
187 (_PAGE_PSE | _PAGE_PRESENT);
190 static inline pte_t pte_mkclean(pte_t pte)
192 return __pte(pte_val(pte) & ~_PAGE_DIRTY);
195 static inline pte_t pte_mkold(pte_t pte)
197 return __pte(pte_val(pte) & ~_PAGE_ACCESSED);
200 static inline pte_t pte_wrprotect(pte_t pte)
202 return __pte(pte_val(pte) & ~_PAGE_RW);
205 static inline pte_t pte_mkexec(pte_t pte)
207 return __pte(pte_val(pte) & ~_PAGE_NX);
210 static inline pte_t pte_mkdirty(pte_t pte)
212 return __pte(pte_val(pte) | _PAGE_DIRTY);
215 static inline pte_t pte_mkyoung(pte_t pte)
217 return __pte(pte_val(pte) | _PAGE_ACCESSED);
220 static inline pte_t pte_mkwrite(pte_t pte)
222 return __pte(pte_val(pte) | _PAGE_RW);
225 static inline pte_t pte_mkhuge(pte_t pte)
227 return __pte(pte_val(pte) | _PAGE_PSE);
230 static inline pte_t pte_clrhuge(pte_t pte)
232 return __pte(pte_val(pte) & ~_PAGE_PSE);
235 static inline pte_t pte_mkglobal(pte_t pte)
237 return __pte(pte_val(pte) | _PAGE_GLOBAL);
240 static inline pte_t pte_clrglobal(pte_t pte)
242 return __pte(pte_val(pte) & ~_PAGE_GLOBAL);
245 static inline pte_t pte_mkspecial(pte_t pte)
250 extern pteval_t __supported_pte_mask;
252 static inline pte_t pfn_pte(unsigned long page_nr, pgprot_t pgprot)
254 return __pte((((phys_addr_t)page_nr << PAGE_SHIFT) |
255 pgprot_val(pgprot)) & __supported_pte_mask);
258 static inline pmd_t pfn_pmd(unsigned long page_nr, pgprot_t pgprot)
260 return __pmd((((phys_addr_t)page_nr << PAGE_SHIFT) |
261 pgprot_val(pgprot)) & __supported_pte_mask);
264 static inline pte_t pte_modify(pte_t pte, pgprot_t newprot)
266 pteval_t val = pte_val(pte);
269 * Chop off the NX bit (if present), and add the NX portion of
270 * the newprot (if present):
272 val &= _PAGE_CHG_MASK;
273 val |= pgprot_val(newprot) & (~_PAGE_CHG_MASK) & __supported_pte_mask;
278 /* mprotect needs to preserve PAT bits when updating vm_page_prot */
279 #define pgprot_modify pgprot_modify
280 static inline pgprot_t pgprot_modify(pgprot_t oldprot, pgprot_t newprot)
282 pgprotval_t preservebits = pgprot_val(oldprot) & _PAGE_CHG_MASK;
283 pgprotval_t addbits = pgprot_val(newprot);
284 return __pgprot(preservebits | addbits);
287 #define pte_pgprot(x) __pgprot(pte_flags(x) & ~PTE_MASK)
289 #define canon_pgprot(p) __pgprot(pgprot_val(p) & __supported_pte_mask)
292 #define __HAVE_PHYS_MEM_ACCESS_PROT
294 pgprot_t phys_mem_access_prot(struct file *file, unsigned long pfn,
295 unsigned long size, pgprot_t vma_prot);
296 int phys_mem_access_prot_allowed(struct file *file, unsigned long pfn,
297 unsigned long size, pgprot_t *vma_prot);
300 /* Install a pte for a particular vaddr in kernel space. */
301 void set_pte_vaddr(unsigned long vaddr, pte_t pte);
303 #ifdef CONFIG_PARAVIRT
304 #include <asm/paravirt.h>
305 #else /* !CONFIG_PARAVIRT */
306 #define set_pte(ptep, pte) native_set_pte(ptep, pte)
307 #define set_pte_at(mm, addr, ptep, pte) native_set_pte_at(mm, addr, ptep, pte)
309 #define set_pte_present(mm, addr, ptep, pte) \
310 native_set_pte_present(mm, addr, ptep, pte)
311 #define set_pte_atomic(ptep, pte) \
312 native_set_pte_atomic(ptep, pte)
314 #define set_pmd(pmdp, pmd) native_set_pmd(pmdp, pmd)
316 #ifndef __PAGETABLE_PUD_FOLDED
317 #define set_pgd(pgdp, pgd) native_set_pgd(pgdp, pgd)
318 #define pgd_clear(pgd) native_pgd_clear(pgd)
322 # define set_pud(pudp, pud) native_set_pud(pudp, pud)
325 #ifndef __PAGETABLE_PMD_FOLDED
326 #define pud_clear(pud) native_pud_clear(pud)
329 #define pte_clear(mm, addr, ptep) native_pte_clear(mm, addr, ptep)
330 #define pmd_clear(pmd) native_pmd_clear(pmd)
332 #define pte_update(mm, addr, ptep) do { } while (0)
333 #define pte_update_defer(mm, addr, ptep) do { } while (0)
334 #endif /* CONFIG_PARAVIRT */
336 #endif /* __ASSEMBLY__ */
339 # include "pgtable_32.h"
341 # include "pgtable_64.h"
345 * the pgd page can be thought of an array like this: pgd_t[PTRS_PER_PGD]
347 * this macro returns the index of the entry in the pgd page which would
348 * control the given virtual address
350 #define pgd_index(address) (((address) >> PGDIR_SHIFT) & (PTRS_PER_PGD - 1))
353 * pgd_offset() returns a (pgd_t *)
354 * pgd_index() is used get the offset into the pgd page's array of pgd_t's;
356 #define pgd_offset(mm, address) ((mm)->pgd + pgd_index((address)))
358 * a shortcut which implies the use of the kernel's pgd, instead
361 #define pgd_offset_k(address) pgd_offset(&init_mm, (address))
364 #define KERNEL_PGD_BOUNDARY pgd_index(PAGE_OFFSET)
365 #define KERNEL_PGD_PTRS (PTRS_PER_PGD - KERNEL_PGD_BOUNDARY)
377 #ifdef CONFIG_PROC_FS
378 extern void update_page_count(int level, unsigned long pages);
380 static inline void update_page_count(int level, unsigned long pages) { }
384 * Helper function that returns the kernel pagetable entry controlling
385 * the virtual address 'address'. NULL means no pagetable entry present.
386 * NOTE: the return type is pte_t but if the pmd is PSE then we return it
389 extern pte_t *lookup_address(unsigned long address, unsigned int *level);
391 /* local pte updates need not use xchg for locking */
392 static inline pte_t native_local_ptep_get_and_clear(pte_t *ptep)
396 /* Pure native function needs no input for mm, addr */
397 native_pte_clear(NULL, 0, ptep);
401 static inline void native_set_pte_at(struct mm_struct *mm, unsigned long addr,
402 pte_t *ptep , pte_t pte)
404 native_set_pte(ptep, pte);
407 #ifndef CONFIG_PARAVIRT
409 * Rules for using pte_update - it must be called after any PTE update which
410 * has not been done using the set_pte / clear_pte interfaces. It is used by
411 * shadow mode hypervisors to resynchronize the shadow page tables. Kernel PTE
412 * updates should either be sets, clears, or set_pte_atomic for P->P
413 * transitions, which means this hook should only be called for user PTEs.
414 * This hook implies a P->P protection or access change has taken place, which
415 * requires a subsequent TLB flush. The notification can optionally be delayed
416 * until the TLB flush event by using the pte_update_defer form of the
417 * interface, but care must be taken to assure that the flush happens while
418 * still holding the same page table lock so that the shadow and primary pages
419 * do not become out of sync on SMP.
421 #define pte_update(mm, addr, ptep) do { } while (0)
422 #define pte_update_defer(mm, addr, ptep) do { } while (0)
426 * We only update the dirty/accessed state if we set
427 * the dirty bit by hand in the kernel, since the hardware
428 * will do the accessed bit for us, and we don't want to
429 * race with other CPU's that might be updating the dirty
430 * bit at the same time.
432 struct vm_area_struct;
434 #define __HAVE_ARCH_PTEP_SET_ACCESS_FLAGS
435 extern int ptep_set_access_flags(struct vm_area_struct *vma,
436 unsigned long address, pte_t *ptep,
437 pte_t entry, int dirty);
439 #define __HAVE_ARCH_PTEP_TEST_AND_CLEAR_YOUNG
440 extern int ptep_test_and_clear_young(struct vm_area_struct *vma,
441 unsigned long addr, pte_t *ptep);
443 #define __HAVE_ARCH_PTEP_CLEAR_YOUNG_FLUSH
444 extern int ptep_clear_flush_young(struct vm_area_struct *vma,
445 unsigned long address, pte_t *ptep);
447 #define __HAVE_ARCH_PTEP_GET_AND_CLEAR
448 static inline pte_t ptep_get_and_clear(struct mm_struct *mm, unsigned long addr,
451 pte_t pte = native_ptep_get_and_clear(ptep);
452 pte_update(mm, addr, ptep);
456 #define __HAVE_ARCH_PTEP_GET_AND_CLEAR_FULL
457 static inline pte_t ptep_get_and_clear_full(struct mm_struct *mm,
458 unsigned long addr, pte_t *ptep,
464 * Full address destruction in progress; paravirt does not
465 * care about updates and native needs no locking
467 pte = native_local_ptep_get_and_clear(ptep);
469 pte = ptep_get_and_clear(mm, addr, ptep);
474 #define __HAVE_ARCH_PTEP_SET_WRPROTECT
475 static inline void ptep_set_wrprotect(struct mm_struct *mm,
476 unsigned long addr, pte_t *ptep)
478 clear_bit(_PAGE_BIT_RW, (unsigned long *)&ptep->pte);
479 pte_update(mm, addr, ptep);
483 * clone_pgd_range(pgd_t *dst, pgd_t *src, int count);
485 * dst - pointer to pgd range anwhere on a pgd page
487 * count - the number of pgds to copy.
489 * dst and src can be on the same page, but the range must not overlap,
490 * and must not cross a page boundary.
492 static inline void clone_pgd_range(pgd_t *dst, pgd_t *src, int count)
494 memcpy(dst, src, count * sizeof(pgd_t));
498 #include <asm-generic/pgtable.h>
499 #endif /* __ASSEMBLY__ */
501 #endif /* _ASM_X86_PGTABLE_H */