1 #ifndef _X86_64_PGTABLE_H
2 #define _X86_64_PGTABLE_H
5 * This file contains the functions and defines necessary to modify and use
6 * the x86-64 page table tree.
8 #include <asm/processor.h>
9 #include <asm/fixmap.h>
10 #include <asm/bitops.h>
11 #include <linux/threads.h>
14 extern pud_t level3_kernel_pgt
[512];
15 extern pud_t level3_physmem_pgt
[512];
16 extern pud_t level3_ident_pgt
[512];
17 extern pmd_t level2_kernel_pgt
[512];
18 extern pgd_t init_level4_pgt
[];
19 extern pgd_t boot_level4_pgt
[];
20 extern unsigned long __supported_pte_mask
;
22 #define swapper_pg_dir init_level4_pgt
24 extern void paging_init(void);
25 extern void clear_kernel_mapping(unsigned long addr
, unsigned long size
);
28 * ZERO_PAGE is a global shared page that is always zero: used
29 * for zero-mapped memory areas etc..
31 extern unsigned long empty_zero_page
[PAGE_SIZE
/sizeof(unsigned long)];
32 #define ZERO_PAGE(vaddr) (virt_to_page(empty_zero_page))
35 * PGDIR_SHIFT determines what a top-level page table entry can map
37 #define PGDIR_SHIFT 39
38 #define PTRS_PER_PGD 512
44 #define PTRS_PER_PUD 512
47 * PMD_SHIFT determines the size of the area a middle-level
51 #define PTRS_PER_PMD 512
54 * entries per page directory level
56 #define PTRS_PER_PTE 512
58 #define pte_ERROR(e) \
59 printk("%s:%d: bad pte %p(%016lx).\n", __FILE__, __LINE__, &(e), pte_val(e))
60 #define pmd_ERROR(e) \
61 printk("%s:%d: bad pmd %p(%016lx).\n", __FILE__, __LINE__, &(e), pmd_val(e))
62 #define pud_ERROR(e) \
63 printk("%s:%d: bad pud %p(%016lx).\n", __FILE__, __LINE__, &(e), pud_val(e))
64 #define pgd_ERROR(e) \
65 printk("%s:%d: bad pgd %p(%016lx).\n", __FILE__, __LINE__, &(e), pgd_val(e))
67 #define pgd_none(x) (!pgd_val(x))
68 #define pud_none(x) (!pud_val(x))
70 static inline void set_pte(pte_t
*dst
, pte_t val
)
72 pte_val(*dst
) = pte_val(val
);
74 #define set_pte_at(mm,addr,ptep,pteval) set_pte(ptep,pteval)
76 static inline void set_pmd(pmd_t
*dst
, pmd_t val
)
78 pmd_val(*dst
) = pmd_val(val
);
81 static inline void set_pud(pud_t
*dst
, pud_t val
)
83 pud_val(*dst
) = pud_val(val
);
86 static inline void pud_clear (pud_t
*pud
)
88 set_pud(pud
, __pud(0));
91 static inline void set_pgd(pgd_t
*dst
, pgd_t val
)
93 pgd_val(*dst
) = pgd_val(val
);
96 static inline void pgd_clear (pgd_t
* pgd
)
98 set_pgd(pgd
, __pgd(0));
101 #define ptep_get_and_clear(mm,addr,xp) __pte(xchg(&(xp)->pte, 0))
105 static inline pte_t
ptep_get_and_clear_full(struct mm_struct
*mm
, unsigned long addr
, pte_t
*ptep
, int full
)
112 pte
= ptep_get_and_clear(mm
, addr
, ptep
);
117 #define pte_same(a, b) ((a).pte == (b).pte)
119 #define pte_pgprot(a) (__pgprot((a).pte & ~PHYSICAL_PAGE_MASK))
121 #define PMD_SIZE (1UL << PMD_SHIFT)
122 #define PMD_MASK (~(PMD_SIZE-1))
123 #define PUD_SIZE (1UL << PUD_SHIFT)
124 #define PUD_MASK (~(PUD_SIZE-1))
125 #define PGDIR_SIZE (1UL << PGDIR_SHIFT)
126 #define PGDIR_MASK (~(PGDIR_SIZE-1))
128 #define USER_PTRS_PER_PGD ((TASK_SIZE-1)/PGDIR_SIZE+1)
129 #define FIRST_USER_ADDRESS 0
132 #define MAXMEM 0x3fffffffffffUL
133 #define VMALLOC_START 0xffffc20000000000UL
134 #define VMALLOC_END 0xffffe1ffffffffffUL
135 #define MODULES_VADDR 0xffffffff88000000UL
136 #define MODULES_END 0xfffffffffff00000UL
137 #define MODULES_LEN (MODULES_END - MODULES_VADDR)
139 #define _PAGE_BIT_PRESENT 0
140 #define _PAGE_BIT_RW 1
141 #define _PAGE_BIT_USER 2
142 #define _PAGE_BIT_PWT 3
143 #define _PAGE_BIT_PCD 4
144 #define _PAGE_BIT_ACCESSED 5
145 #define _PAGE_BIT_DIRTY 6
146 #define _PAGE_BIT_PSE 7 /* 4 MB (or 2MB) page */
147 #define _PAGE_BIT_GLOBAL 8 /* Global TLB entry PPro+ */
148 #define _PAGE_BIT_NX 63 /* No execute: only valid after cpuid check */
150 #define _PAGE_PRESENT 0x001
151 #define _PAGE_RW 0x002
152 #define _PAGE_USER 0x004
153 #define _PAGE_PWT 0x008
154 #define _PAGE_PCD 0x010
155 #define _PAGE_ACCESSED 0x020
156 #define _PAGE_DIRTY 0x040
157 #define _PAGE_PSE 0x080 /* 2MB page */
158 #define _PAGE_FILE 0x040 /* nonlinear file mapping, saved PTE; unset:swap */
159 #define _PAGE_GLOBAL 0x100 /* Global TLB entry */
161 #define _PAGE_PROTNONE 0x080 /* If not present */
162 #define _PAGE_NX (1UL<<_PAGE_BIT_NX)
164 #define _PAGE_TABLE (_PAGE_PRESENT | _PAGE_RW | _PAGE_USER | _PAGE_ACCESSED | _PAGE_DIRTY)
165 #define _KERNPG_TABLE (_PAGE_PRESENT | _PAGE_RW | _PAGE_ACCESSED | _PAGE_DIRTY)
167 #define _PAGE_CHG_MASK (PTE_MASK | _PAGE_ACCESSED | _PAGE_DIRTY)
169 #define PAGE_NONE __pgprot(_PAGE_PROTNONE | _PAGE_ACCESSED)
170 #define PAGE_SHARED __pgprot(_PAGE_PRESENT | _PAGE_RW | _PAGE_USER | _PAGE_ACCESSED | _PAGE_NX)
171 #define PAGE_SHARED_EXEC __pgprot(_PAGE_PRESENT | _PAGE_RW | _PAGE_USER | _PAGE_ACCESSED)
172 #define PAGE_COPY_NOEXEC __pgprot(_PAGE_PRESENT | _PAGE_USER | _PAGE_ACCESSED | _PAGE_NX)
173 #define PAGE_COPY PAGE_COPY_NOEXEC
174 #define PAGE_COPY_EXEC __pgprot(_PAGE_PRESENT | _PAGE_USER | _PAGE_ACCESSED)
175 #define PAGE_READONLY __pgprot(_PAGE_PRESENT | _PAGE_USER | _PAGE_ACCESSED | _PAGE_NX)
176 #define PAGE_READONLY_EXEC __pgprot(_PAGE_PRESENT | _PAGE_USER | _PAGE_ACCESSED)
177 #define __PAGE_KERNEL \
178 (_PAGE_PRESENT | _PAGE_RW | _PAGE_DIRTY | _PAGE_ACCESSED | _PAGE_NX)
179 #define __PAGE_KERNEL_EXEC \
180 (_PAGE_PRESENT | _PAGE_RW | _PAGE_DIRTY | _PAGE_ACCESSED)
181 #define __PAGE_KERNEL_NOCACHE \
182 (_PAGE_PRESENT | _PAGE_RW | _PAGE_DIRTY | _PAGE_PCD | _PAGE_ACCESSED | _PAGE_NX)
183 #define __PAGE_KERNEL_RO \
184 (_PAGE_PRESENT | _PAGE_DIRTY | _PAGE_ACCESSED | _PAGE_NX)
185 #define __PAGE_KERNEL_VSYSCALL \
186 (_PAGE_PRESENT | _PAGE_USER | _PAGE_ACCESSED)
187 #define __PAGE_KERNEL_VSYSCALL_NOCACHE \
188 (_PAGE_PRESENT | _PAGE_USER | _PAGE_ACCESSED | _PAGE_PCD)
189 #define __PAGE_KERNEL_LARGE \
190 (__PAGE_KERNEL | _PAGE_PSE)
191 #define __PAGE_KERNEL_LARGE_EXEC \
192 (__PAGE_KERNEL_EXEC | _PAGE_PSE)
194 #define MAKE_GLOBAL(x) __pgprot((x) | _PAGE_GLOBAL)
196 #define PAGE_KERNEL MAKE_GLOBAL(__PAGE_KERNEL)
197 #define PAGE_KERNEL_EXEC MAKE_GLOBAL(__PAGE_KERNEL_EXEC)
198 #define PAGE_KERNEL_RO MAKE_GLOBAL(__PAGE_KERNEL_RO)
199 #define PAGE_KERNEL_NOCACHE MAKE_GLOBAL(__PAGE_KERNEL_NOCACHE)
200 #define PAGE_KERNEL_VSYSCALL32 __pgprot(__PAGE_KERNEL_VSYSCALL)
201 #define PAGE_KERNEL_VSYSCALL MAKE_GLOBAL(__PAGE_KERNEL_VSYSCALL)
202 #define PAGE_KERNEL_LARGE MAKE_GLOBAL(__PAGE_KERNEL_LARGE)
203 #define PAGE_KERNEL_VSYSCALL_NOCACHE MAKE_GLOBAL(__PAGE_KERNEL_VSYSCALL_NOCACHE)
206 #define __P000 PAGE_NONE
207 #define __P001 PAGE_READONLY
208 #define __P010 PAGE_COPY
209 #define __P011 PAGE_COPY
210 #define __P100 PAGE_READONLY_EXEC
211 #define __P101 PAGE_READONLY_EXEC
212 #define __P110 PAGE_COPY_EXEC
213 #define __P111 PAGE_COPY_EXEC
215 #define __S000 PAGE_NONE
216 #define __S001 PAGE_READONLY
217 #define __S010 PAGE_SHARED
218 #define __S011 PAGE_SHARED
219 #define __S100 PAGE_READONLY_EXEC
220 #define __S101 PAGE_READONLY_EXEC
221 #define __S110 PAGE_SHARED_EXEC
222 #define __S111 PAGE_SHARED_EXEC
224 static inline unsigned long pgd_bad(pgd_t pgd
)
226 unsigned long val
= pgd_val(pgd
);
228 val
&= ~(_PAGE_USER
| _PAGE_DIRTY
);
229 return val
& ~(_PAGE_PRESENT
| _PAGE_RW
| _PAGE_ACCESSED
);
232 static inline unsigned long pud_bad(pud_t pud
)
234 unsigned long val
= pud_val(pud
);
236 val
&= ~(_PAGE_USER
| _PAGE_DIRTY
);
237 return val
& ~(_PAGE_PRESENT
| _PAGE_RW
| _PAGE_ACCESSED
);
240 #define pte_none(x) (!pte_val(x))
241 #define pte_present(x) (pte_val(x) & (_PAGE_PRESENT | _PAGE_PROTNONE))
242 #define pte_clear(mm,addr,xp) do { set_pte_at(mm, addr, xp, __pte(0)); } while (0)
244 #define pages_to_mb(x) ((x) >> (20-PAGE_SHIFT)) /* FIXME: is this
246 #define pte_page(x) pfn_to_page(pte_pfn(x))
247 #define pte_pfn(x) ((pte_val(x) & __PHYSICAL_MASK) >> PAGE_SHIFT)
249 static inline pte_t
pfn_pte(unsigned long page_nr
, pgprot_t pgprot
)
252 pte_val(pte
) = (page_nr
<< PAGE_SHIFT
);
253 pte_val(pte
) |= pgprot_val(pgprot
);
254 pte_val(pte
) &= __supported_pte_mask
;
259 * The following only work if pte_present() is true.
260 * Undefined behaviour if not..
262 #define __LARGE_PTE (_PAGE_PSE|_PAGE_PRESENT)
263 static inline int pte_user(pte_t pte
) { return pte_val(pte
) & _PAGE_USER
; }
264 static inline int pte_read(pte_t pte
) { return pte_val(pte
) & _PAGE_USER
; }
265 static inline int pte_exec(pte_t pte
) { return !(pte_val(pte
) & _PAGE_NX
); }
266 static inline int pte_dirty(pte_t pte
) { return pte_val(pte
) & _PAGE_DIRTY
; }
267 static inline int pte_young(pte_t pte
) { return pte_val(pte
) & _PAGE_ACCESSED
; }
268 static inline int pte_write(pte_t pte
) { return pte_val(pte
) & _PAGE_RW
; }
269 static inline int pte_file(pte_t pte
) { return pte_val(pte
) & _PAGE_FILE
; }
270 static inline int pte_huge(pte_t pte
) { return pte_val(pte
) & _PAGE_PSE
; }
272 static inline pte_t
pte_rdprotect(pte_t pte
) { set_pte(&pte
, __pte(pte_val(pte
) & ~_PAGE_USER
)); return pte
; }
273 static inline pte_t
pte_exprotect(pte_t pte
) { set_pte(&pte
, __pte(pte_val(pte
) & ~_PAGE_USER
)); return pte
; }
274 static inline pte_t
pte_mkclean(pte_t pte
) { set_pte(&pte
, __pte(pte_val(pte
) & ~_PAGE_DIRTY
)); return pte
; }
275 static inline pte_t
pte_mkold(pte_t pte
) { set_pte(&pte
, __pte(pte_val(pte
) & ~_PAGE_ACCESSED
)); return pte
; }
276 static inline pte_t
pte_wrprotect(pte_t pte
) { set_pte(&pte
, __pte(pte_val(pte
) & ~_PAGE_RW
)); return pte
; }
277 static inline pte_t
pte_mkread(pte_t pte
) { set_pte(&pte
, __pte(pte_val(pte
) | _PAGE_USER
)); return pte
; }
278 static inline pte_t
pte_mkexec(pte_t pte
) { set_pte(&pte
, __pte(pte_val(pte
) & ~_PAGE_NX
)); return pte
; }
279 static inline pte_t
pte_mkdirty(pte_t pte
) { set_pte(&pte
, __pte(pte_val(pte
) | _PAGE_DIRTY
)); return pte
; }
280 static inline pte_t
pte_mkyoung(pte_t pte
) { set_pte(&pte
, __pte(pte_val(pte
) | _PAGE_ACCESSED
)); return pte
; }
281 static inline pte_t
pte_mkwrite(pte_t pte
) { set_pte(&pte
, __pte(pte_val(pte
) | _PAGE_RW
)); return pte
; }
282 static inline pte_t
pte_mkhuge(pte_t pte
) { set_pte(&pte
, __pte(pte_val(pte
) | _PAGE_PSE
)); return pte
; }
283 static inline pte_t
pte_clrhuge(pte_t pte
) { set_pte(&pte
, __pte(pte_val(pte
) & ~_PAGE_PSE
)); return pte
; }
285 struct vm_area_struct
;
287 static inline int ptep_test_and_clear_dirty(struct vm_area_struct
*vma
, unsigned long addr
, pte_t
*ptep
)
289 if (!pte_dirty(*ptep
))
291 return test_and_clear_bit(_PAGE_BIT_DIRTY
, &ptep
->pte
);
294 static inline int ptep_test_and_clear_young(struct vm_area_struct
*vma
, unsigned long addr
, pte_t
*ptep
)
296 if (!pte_young(*ptep
))
298 return test_and_clear_bit(_PAGE_BIT_ACCESSED
, &ptep
->pte
);
301 static inline void ptep_set_wrprotect(struct mm_struct
*mm
, unsigned long addr
, pte_t
*ptep
)
303 clear_bit(_PAGE_BIT_RW
, &ptep
->pte
);
307 * Macro to mark a page protection value as "uncacheable".
309 #define pgprot_noncached(prot) (__pgprot(pgprot_val(prot) | _PAGE_PCD | _PAGE_PWT))
311 static inline int pmd_large(pmd_t pte
) {
312 return (pmd_val(pte
) & __LARGE_PTE
) == __LARGE_PTE
;
317 * Conversion functions: convert a page and protection to a page entry,
318 * and a page entry and page directory to the page they refer to.
324 #define pgd_page_vaddr(pgd) ((unsigned long) __va((unsigned long)pgd_val(pgd) & PTE_MASK))
325 #define pgd_page(pgd) (pfn_to_page(pgd_val(pgd) >> PAGE_SHIFT))
326 #define pgd_index(address) (((address) >> PGDIR_SHIFT) & (PTRS_PER_PGD-1))
327 #define pgd_offset(mm, addr) ((mm)->pgd + pgd_index(addr))
328 #define pgd_offset_k(address) (init_level4_pgt + pgd_index(address))
329 #define pgd_present(pgd) (pgd_val(pgd) & _PAGE_PRESENT)
330 #define mk_kernel_pgd(address) ((pgd_t){ (address) | _KERNPG_TABLE })
332 /* PUD - Level3 access */
333 /* to find an entry in a page-table-directory. */
334 #define pud_page_vaddr(pud) ((unsigned long) __va(pud_val(pud) & PHYSICAL_PAGE_MASK))
335 #define pud_page(pud) (pfn_to_page(pud_val(pud) >> PAGE_SHIFT))
336 #define pud_index(address) (((address) >> PUD_SHIFT) & (PTRS_PER_PUD-1))
337 #define pud_offset(pgd, address) ((pud_t *) pgd_page_vaddr(*(pgd)) + pud_index(address))
338 #define pud_present(pud) (pud_val(pud) & _PAGE_PRESENT)
340 /* PMD - Level 2 access */
341 #define pmd_page_vaddr(pmd) ((unsigned long) __va(pmd_val(pmd) & PTE_MASK))
342 #define pmd_page(pmd) (pfn_to_page(pmd_val(pmd) >> PAGE_SHIFT))
344 #define pmd_index(address) (((address) >> PMD_SHIFT) & (PTRS_PER_PMD-1))
345 #define pmd_offset(dir, address) ((pmd_t *) pud_page_vaddr(*(dir)) + \
347 #define pmd_none(x) (!pmd_val(x))
348 #define pmd_present(x) (pmd_val(x) & _PAGE_PRESENT)
349 #define pmd_clear(xp) do { set_pmd(xp, __pmd(0)); } while (0)
350 #define pmd_bad(x) ((pmd_val(x) & (~PTE_MASK & ~_PAGE_USER)) != _KERNPG_TABLE )
351 #define pfn_pmd(nr,prot) (__pmd(((nr) << PAGE_SHIFT) | pgprot_val(prot)))
352 #define pmd_pfn(x) ((pmd_val(x) & __PHYSICAL_MASK) >> PAGE_SHIFT)
354 #define pte_to_pgoff(pte) ((pte_val(pte) & PHYSICAL_PAGE_MASK) >> PAGE_SHIFT)
355 #define pgoff_to_pte(off) ((pte_t) { ((off) << PAGE_SHIFT) | _PAGE_FILE })
356 #define PTE_FILE_MAX_BITS __PHYSICAL_MASK_SHIFT
358 /* PTE - Level 1 access. */
360 /* page, protection -> pte */
361 #define mk_pte(page, pgprot) pfn_pte(page_to_pfn(page), (pgprot))
362 #define mk_pte_huge(entry) (pte_val(entry) |= _PAGE_PRESENT | _PAGE_PSE)
364 /* physical address -> PTE */
365 static inline pte_t
mk_pte_phys(unsigned long physpage
, pgprot_t pgprot
)
368 pte_val(pte
) = physpage
| pgprot_val(pgprot
);
372 /* Change flags of a PTE */
373 static inline pte_t
pte_modify(pte_t pte
, pgprot_t newprot
)
375 pte_val(pte
) &= _PAGE_CHG_MASK
;
376 pte_val(pte
) |= pgprot_val(newprot
);
377 pte_val(pte
) &= __supported_pte_mask
;
381 #define pte_index(address) \
382 (((address) >> PAGE_SHIFT) & (PTRS_PER_PTE - 1))
383 #define pte_offset_kernel(dir, address) ((pte_t *) pmd_page_vaddr(*(dir)) + \
386 /* x86-64 always has all page tables mapped. */
387 #define pte_offset_map(dir,address) pte_offset_kernel(dir,address)
388 #define pte_offset_map_nested(dir,address) pte_offset_kernel(dir,address)
389 #define pte_unmap(pte) /* NOP */
390 #define pte_unmap_nested(pte) /* NOP */
392 #define update_mmu_cache(vma,address,pte) do { } while (0)
394 /* We only update the dirty/accessed state if we set
395 * the dirty bit by hand in the kernel, since the hardware
396 * will do the accessed bit for us, and we don't want to
397 * race with other CPU's that might be updating the dirty
398 * bit at the same time. */
399 #define __HAVE_ARCH_PTEP_SET_ACCESS_FLAGS
400 #define ptep_set_access_flags(__vma, __address, __ptep, __entry, __dirty) \
403 set_pte(__ptep, __entry); \
404 flush_tlb_page(__vma, __address); \
408 /* Encode and de-code a swap entry */
409 #define __swp_type(x) (((x).val >> 1) & 0x3f)
410 #define __swp_offset(x) ((x).val >> 8)
411 #define __swp_entry(type, offset) ((swp_entry_t) { ((type) << 1) | ((offset) << 8) })
412 #define __pte_to_swp_entry(pte) ((swp_entry_t) { pte_val(pte) })
413 #define __swp_entry_to_pte(x) ((pte_t) { (x).val })
415 extern spinlock_t pgd_lock
;
416 extern struct page
*pgd_list
;
417 void vmalloc_sync_all(void);
419 #endif /* !__ASSEMBLY__ */
421 extern int kern_addr_valid(unsigned long addr
);
423 #define io_remap_pfn_range(vma, vaddr, pfn, size, prot) \
424 remap_pfn_range(vma, vaddr, pfn, size, prot)
426 #define MK_IOSPACE_PFN(space, pfn) (pfn)
427 #define GET_IOSPACE(pfn) 0
428 #define GET_PFN(pfn) (pfn)
430 #define HAVE_ARCH_UNMAPPED_AREA
432 #define pgtable_cache_init() do { } while (0)
433 #define check_pgt_cache() do { } while (0)
435 #define PAGE_AGP PAGE_KERNEL_NOCACHE
436 #define HAVE_PAGE_AGP 1
438 /* fs/proc/kcore.c */
439 #define kc_vaddr_to_offset(v) ((v) & __VIRTUAL_MASK)
440 #define kc_offset_to_vaddr(o) \
441 (((o) & (1UL << (__VIRTUAL_MASK_SHIFT-1))) ? ((o) | (~__VIRTUAL_MASK)) : (o))
443 #define __HAVE_ARCH_PTEP_TEST_AND_CLEAR_YOUNG
444 #define __HAVE_ARCH_PTEP_TEST_AND_CLEAR_DIRTY
445 #define __HAVE_ARCH_PTEP_GET_AND_CLEAR
446 #define __HAVE_ARCH_PTEP_GET_AND_CLEAR_FULL
447 #define __HAVE_ARCH_PTEP_SET_WRPROTECT
448 #define __HAVE_ARCH_PTE_SAME
449 #include <asm-generic/pgtable.h>
451 #endif /* _X86_64_PGTABLE_H */