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 unsigned long __supported_pte_mask;
21 #define swapper_pg_dir init_level4_pgt
23 extern int nonx_setup(char *str);
24 extern void paging_init(void);
25 extern void clear_kernel_mapping(unsigned long addr, unsigned long size);
27 extern unsigned long pgkern_mask;
30 * ZERO_PAGE is a global shared page that is always zero: used
31 * for zero-mapped memory areas etc..
33 extern unsigned long empty_zero_page[PAGE_SIZE/sizeof(unsigned long)];
34 #define ZERO_PAGE(vaddr) (virt_to_page(empty_zero_page))
37 * PGDIR_SHIFT determines what a top-level page table entry can map
39 #define PGDIR_SHIFT 39
40 #define PTRS_PER_PGD 512
46 #define PTRS_PER_PUD 512
49 * PMD_SHIFT determines the size of the area a middle-level
53 #define PTRS_PER_PMD 512
56 * entries per page directory level
58 #define PTRS_PER_PTE 512
60 #define pte_ERROR(e) \
61 printk("%s:%d: bad pte %p(%016lx).\n", __FILE__, __LINE__, &(e), pte_val(e))
62 #define pmd_ERROR(e) \
63 printk("%s:%d: bad pmd %p(%016lx).\n", __FILE__, __LINE__, &(e), pmd_val(e))
64 #define pud_ERROR(e) \
65 printk("%s:%d: bad pud %p(%016lx).\n", __FILE__, __LINE__, &(e), pud_val(e))
66 #define pgd_ERROR(e) \
67 printk("%s:%d: bad pgd %p(%016lx).\n", __FILE__, __LINE__, &(e), pgd_val(e))
69 #define pgd_none(x) (!pgd_val(x))
70 #define pud_none(x) (!pud_val(x))
72 static inline void set_pte(pte_t *dst, pte_t val)
74 pte_val(*dst) = pte_val(val);
76 #define set_pte_at(mm,addr,ptep,pteval) set_pte(ptep,pteval)
78 static inline void set_pmd(pmd_t *dst, pmd_t val)
80 pmd_val(*dst) = pmd_val(val);
83 static inline void set_pud(pud_t *dst, pud_t val)
85 pud_val(*dst) = pud_val(val);
88 extern inline void pud_clear (pud_t *pud)
90 set_pud(pud, __pud(0));
93 static inline void set_pgd(pgd_t *dst, pgd_t val)
95 pgd_val(*dst) = pgd_val(val);
98 extern inline void pgd_clear (pgd_t * pgd)
100 set_pgd(pgd, __pgd(0));
103 #define pud_page(pud) \
104 ((unsigned long) __va(pud_val(pud) & PHYSICAL_PAGE_MASK))
106 #define ptep_get_and_clear(mm,addr,xp) __pte(xchg(&(xp)->pte, 0))
107 #define pte_same(a, b) ((a).pte == (b).pte)
109 #define PMD_SIZE (1UL << PMD_SHIFT)
110 #define PMD_MASK (~(PMD_SIZE-1))
111 #define PUD_SIZE (1UL << PUD_SHIFT)
112 #define PUD_MASK (~(PUD_SIZE-1))
113 #define PGDIR_SIZE (1UL << PGDIR_SHIFT)
114 #define PGDIR_MASK (~(PGDIR_SIZE-1))
116 #define USER_PTRS_PER_PGD (TASK_SIZE/PGDIR_SIZE)
117 #define FIRST_USER_ADDRESS 0
120 #define MAXMEM 0x3fffffffffffUL
121 #define VMALLOC_START 0xffffc20000000000UL
122 #define VMALLOC_END 0xffffe1ffffffffffUL
123 #define MODULES_VADDR 0xffffffff88000000UL
124 #define MODULES_END 0xfffffffffff00000UL
125 #define MODULES_LEN (MODULES_END - MODULES_VADDR)
127 #define _PAGE_BIT_PRESENT 0
128 #define _PAGE_BIT_RW 1
129 #define _PAGE_BIT_USER 2
130 #define _PAGE_BIT_PWT 3
131 #define _PAGE_BIT_PCD 4
132 #define _PAGE_BIT_ACCESSED 5
133 #define _PAGE_BIT_DIRTY 6
134 #define _PAGE_BIT_PSE 7 /* 4 MB (or 2MB) page */
135 #define _PAGE_BIT_GLOBAL 8 /* Global TLB entry PPro+ */
136 #define _PAGE_BIT_NX 63 /* No execute: only valid after cpuid check */
138 #define _PAGE_PRESENT 0x001
139 #define _PAGE_RW 0x002
140 #define _PAGE_USER 0x004
141 #define _PAGE_PWT 0x008
142 #define _PAGE_PCD 0x010
143 #define _PAGE_ACCESSED 0x020
144 #define _PAGE_DIRTY 0x040
145 #define _PAGE_PSE 0x080 /* 2MB page */
146 #define _PAGE_FILE 0x040 /* set:pagecache, unset:swap */
147 #define _PAGE_GLOBAL 0x100 /* Global TLB entry */
149 #define _PAGE_PROTNONE 0x080 /* If not present */
150 #define _PAGE_NX (1UL<<_PAGE_BIT_NX)
152 #define _PAGE_TABLE (_PAGE_PRESENT | _PAGE_RW | _PAGE_USER | _PAGE_ACCESSED | _PAGE_DIRTY)
153 #define _KERNPG_TABLE (_PAGE_PRESENT | _PAGE_RW | _PAGE_ACCESSED | _PAGE_DIRTY)
155 #define _PAGE_CHG_MASK (PTE_MASK | _PAGE_ACCESSED | _PAGE_DIRTY)
157 #define PAGE_NONE __pgprot(_PAGE_PROTNONE | _PAGE_ACCESSED)
158 #define PAGE_SHARED __pgprot(_PAGE_PRESENT | _PAGE_RW | _PAGE_USER | _PAGE_ACCESSED | _PAGE_NX)
159 #define PAGE_SHARED_EXEC __pgprot(_PAGE_PRESENT | _PAGE_RW | _PAGE_USER | _PAGE_ACCESSED)
160 #define PAGE_COPY_NOEXEC __pgprot(_PAGE_PRESENT | _PAGE_USER | _PAGE_ACCESSED | _PAGE_NX)
161 #define PAGE_COPY PAGE_COPY_NOEXEC
162 #define PAGE_COPY_EXEC __pgprot(_PAGE_PRESENT | _PAGE_USER | _PAGE_ACCESSED)
163 #define PAGE_READONLY __pgprot(_PAGE_PRESENT | _PAGE_USER | _PAGE_ACCESSED | _PAGE_NX)
164 #define PAGE_READONLY_EXEC __pgprot(_PAGE_PRESENT | _PAGE_USER | _PAGE_ACCESSED)
165 #define __PAGE_KERNEL \
166 (_PAGE_PRESENT | _PAGE_RW | _PAGE_DIRTY | _PAGE_ACCESSED | _PAGE_NX)
167 #define __PAGE_KERNEL_EXEC \
168 (_PAGE_PRESENT | _PAGE_RW | _PAGE_DIRTY | _PAGE_ACCESSED)
169 #define __PAGE_KERNEL_NOCACHE \
170 (_PAGE_PRESENT | _PAGE_RW | _PAGE_DIRTY | _PAGE_PCD | _PAGE_ACCESSED | _PAGE_NX)
171 #define __PAGE_KERNEL_RO \
172 (_PAGE_PRESENT | _PAGE_DIRTY | _PAGE_ACCESSED | _PAGE_NX)
173 #define __PAGE_KERNEL_VSYSCALL \
174 (_PAGE_PRESENT | _PAGE_USER | _PAGE_ACCESSED)
175 #define __PAGE_KERNEL_VSYSCALL_NOCACHE \
176 (_PAGE_PRESENT | _PAGE_USER | _PAGE_ACCESSED | _PAGE_PCD)
177 #define __PAGE_KERNEL_LARGE \
178 (__PAGE_KERNEL | _PAGE_PSE)
180 #define MAKE_GLOBAL(x) __pgprot((x) | _PAGE_GLOBAL)
182 #define PAGE_KERNEL MAKE_GLOBAL(__PAGE_KERNEL)
183 #define PAGE_KERNEL_EXEC MAKE_GLOBAL(__PAGE_KERNEL_EXEC)
184 #define PAGE_KERNEL_RO MAKE_GLOBAL(__PAGE_KERNEL_RO)
185 #define PAGE_KERNEL_NOCACHE MAKE_GLOBAL(__PAGE_KERNEL_NOCACHE)
186 #define PAGE_KERNEL_VSYSCALL32 __pgprot(__PAGE_KERNEL_VSYSCALL)
187 #define PAGE_KERNEL_VSYSCALL MAKE_GLOBAL(__PAGE_KERNEL_VSYSCALL)
188 #define PAGE_KERNEL_LARGE MAKE_GLOBAL(__PAGE_KERNEL_LARGE)
189 #define PAGE_KERNEL_VSYSCALL_NOCACHE MAKE_GLOBAL(__PAGE_KERNEL_VSYSCALL_NOCACHE)
192 #define __P000 PAGE_NONE
193 #define __P001 PAGE_READONLY
194 #define __P010 PAGE_COPY
195 #define __P011 PAGE_COPY
196 #define __P100 PAGE_READONLY_EXEC
197 #define __P101 PAGE_READONLY_EXEC
198 #define __P110 PAGE_COPY_EXEC
199 #define __P111 PAGE_COPY_EXEC
201 #define __S000 PAGE_NONE
202 #define __S001 PAGE_READONLY
203 #define __S010 PAGE_SHARED
204 #define __S011 PAGE_SHARED
205 #define __S100 PAGE_READONLY_EXEC
206 #define __S101 PAGE_READONLY_EXEC
207 #define __S110 PAGE_SHARED_EXEC
208 #define __S111 PAGE_SHARED_EXEC
210 static inline unsigned long pgd_bad(pgd_t pgd)
212 unsigned long val = pgd_val(pgd);
214 val &= ~(_PAGE_USER | _PAGE_DIRTY);
215 return val & ~(_PAGE_PRESENT | _PAGE_RW | _PAGE_ACCESSED);
218 static inline unsigned long pud_bad(pud_t pud)
220 unsigned long val = pud_val(pud);
222 val &= ~(_PAGE_USER | _PAGE_DIRTY);
223 return val & ~(_PAGE_PRESENT | _PAGE_RW | _PAGE_ACCESSED);
226 #define pte_none(x) (!pte_val(x))
227 #define pte_present(x) (pte_val(x) & (_PAGE_PRESENT | _PAGE_PROTNONE))
228 #define pte_clear(mm,addr,xp) do { set_pte_at(mm, addr, xp, __pte(0)); } while (0)
230 #define pages_to_mb(x) ((x) >> (20-PAGE_SHIFT)) /* FIXME: is this
232 #define pte_page(x) pfn_to_page(pte_pfn(x))
233 #define pte_pfn(x) ((pte_val(x) >> PAGE_SHIFT) & __PHYSICAL_MASK)
235 static inline pte_t pfn_pte(unsigned long page_nr, pgprot_t pgprot)
238 pte_val(pte) = (page_nr << PAGE_SHIFT);
239 pte_val(pte) |= pgprot_val(pgprot);
240 pte_val(pte) &= __supported_pte_mask;
245 * The following only work if pte_present() is true.
246 * Undefined behaviour if not..
248 static inline int pte_user(pte_t pte) { return pte_val(pte) & _PAGE_USER; }
249 extern inline int pte_read(pte_t pte) { return pte_val(pte) & _PAGE_USER; }
250 extern inline int pte_exec(pte_t pte) { return pte_val(pte) & _PAGE_USER; }
251 extern inline int pte_dirty(pte_t pte) { return pte_val(pte) & _PAGE_DIRTY; }
252 extern inline int pte_young(pte_t pte) { return pte_val(pte) & _PAGE_ACCESSED; }
253 extern inline int pte_write(pte_t pte) { return pte_val(pte) & _PAGE_RW; }
254 static inline int pte_file(pte_t pte) { return pte_val(pte) & _PAGE_FILE; }
256 extern inline pte_t pte_rdprotect(pte_t pte) { set_pte(&pte, __pte(pte_val(pte) & ~_PAGE_USER)); return pte; }
257 extern inline pte_t pte_exprotect(pte_t pte) { set_pte(&pte, __pte(pte_val(pte) & ~_PAGE_USER)); return pte; }
258 extern inline pte_t pte_mkclean(pte_t pte) { set_pte(&pte, __pte(pte_val(pte) & ~_PAGE_DIRTY)); return pte; }
259 extern inline pte_t pte_mkold(pte_t pte) { set_pte(&pte, __pte(pte_val(pte) & ~_PAGE_ACCESSED)); return pte; }
260 extern inline pte_t pte_wrprotect(pte_t pte) { set_pte(&pte, __pte(pte_val(pte) & ~_PAGE_RW)); return pte; }
261 extern inline pte_t pte_mkread(pte_t pte) { set_pte(&pte, __pte(pte_val(pte) | _PAGE_USER)); return pte; }
262 extern inline pte_t pte_mkexec(pte_t pte) { set_pte(&pte, __pte(pte_val(pte) | _PAGE_USER)); return pte; }
263 extern inline pte_t pte_mkdirty(pte_t pte) { set_pte(&pte, __pte(pte_val(pte) | _PAGE_DIRTY)); return pte; }
264 extern inline pte_t pte_mkyoung(pte_t pte) { set_pte(&pte, __pte(pte_val(pte) | _PAGE_ACCESSED)); return pte; }
265 extern inline pte_t pte_mkwrite(pte_t pte) { set_pte(&pte, __pte(pte_val(pte) | _PAGE_RW)); return pte; }
267 struct vm_area_struct;
269 static inline int ptep_test_and_clear_dirty(struct vm_area_struct *vma, unsigned long addr, pte_t *ptep)
271 if (!pte_dirty(*ptep))
273 return test_and_clear_bit(_PAGE_BIT_DIRTY, ptep);
276 static inline int ptep_test_and_clear_young(struct vm_area_struct *vma, unsigned long addr, pte_t *ptep)
278 if (!pte_young(*ptep))
280 return test_and_clear_bit(_PAGE_BIT_ACCESSED, ptep);
283 static inline void ptep_set_wrprotect(struct mm_struct *mm, unsigned long addr, pte_t *ptep)
285 clear_bit(_PAGE_BIT_RW, ptep);
289 * Macro to mark a page protection value as "uncacheable".
291 #define pgprot_noncached(prot) (__pgprot(pgprot_val(prot) | _PAGE_PCD | _PAGE_PWT))
293 #define __LARGE_PTE (_PAGE_PSE|_PAGE_PRESENT)
294 static inline int pmd_large(pmd_t pte) {
295 return (pmd_val(pte) & __LARGE_PTE) == __LARGE_PTE;
300 * Conversion functions: convert a page and protection to a page entry,
301 * and a page entry and page directory to the page they refer to.
304 #define page_pte(page) page_pte_prot(page, __pgprot(0))
309 #define pgd_page(pgd) ((unsigned long) __va((unsigned long)pgd_val(pgd) & PTE_MASK))
310 #define pgd_index(address) (((address) >> PGDIR_SHIFT) & (PTRS_PER_PGD-1))
311 #define pgd_offset(mm, addr) ((mm)->pgd + pgd_index(addr))
312 #define pgd_offset_k(address) (init_level4_pgt + pgd_index(address))
313 #define pgd_present(pgd) (pgd_val(pgd) & _PAGE_PRESENT)
314 #define mk_kernel_pgd(address) ((pgd_t){ (address) | _KERNPG_TABLE })
316 /* PUD - Level3 access */
317 /* to find an entry in a page-table-directory. */
318 #define pud_index(address) (((address) >> PUD_SHIFT) & (PTRS_PER_PUD-1))
319 #define pud_offset(pgd, address) ((pud_t *) pgd_page(*(pgd)) + pud_index(address))
320 #define pud_offset_k(pgd, addr) pud_offset(pgd, addr)
321 #define pud_present(pud) (pud_val(pud) & _PAGE_PRESENT)
323 static inline pud_t *__pud_offset_k(pud_t *pud, unsigned long address)
325 return pud + pud_index(address);
328 /* PMD - Level 2 access */
329 #define pmd_page_kernel(pmd) ((unsigned long) __va(pmd_val(pmd) & PTE_MASK))
330 #define pmd_page(pmd) (pfn_to_page(pmd_val(pmd) >> PAGE_SHIFT))
332 #define pmd_index(address) (((address) >> PMD_SHIFT) & (PTRS_PER_PMD-1))
333 #define pmd_offset(dir, address) ((pmd_t *) pud_page(*(dir)) + \
335 #define pmd_none(x) (!pmd_val(x))
336 #define pmd_present(x) (pmd_val(x) & _PAGE_PRESENT)
337 #define pmd_clear(xp) do { set_pmd(xp, __pmd(0)); } while (0)
338 #define pmd_bad(x) ((pmd_val(x) & (~PTE_MASK & ~_PAGE_USER)) != _KERNPG_TABLE )
339 #define pfn_pmd(nr,prot) (__pmd(((nr) << PAGE_SHIFT) | pgprot_val(prot)))
340 #define pmd_pfn(x) ((pmd_val(x) >> PAGE_SHIFT) & __PHYSICAL_MASK)
342 #define pte_to_pgoff(pte) ((pte_val(pte) & PHYSICAL_PAGE_MASK) >> PAGE_SHIFT)
343 #define pgoff_to_pte(off) ((pte_t) { ((off) << PAGE_SHIFT) | _PAGE_FILE })
344 #define PTE_FILE_MAX_BITS __PHYSICAL_MASK_SHIFT
346 /* PTE - Level 1 access. */
348 /* page, protection -> pte */
349 #define mk_pte(page, pgprot) pfn_pte(page_to_pfn(page), (pgprot))
350 #define mk_pte_huge(entry) (pte_val(entry) |= _PAGE_PRESENT | _PAGE_PSE)
352 /* physical address -> PTE */
353 static inline pte_t mk_pte_phys(unsigned long physpage, pgprot_t pgprot)
356 pte_val(pte) = physpage | pgprot_val(pgprot);
360 /* Change flags of a PTE */
361 extern inline pte_t pte_modify(pte_t pte, pgprot_t newprot)
363 pte_val(pte) &= _PAGE_CHG_MASK;
364 pte_val(pte) |= pgprot_val(newprot);
365 pte_val(pte) &= __supported_pte_mask;
369 #define pte_index(address) \
370 ((address >> PAGE_SHIFT) & (PTRS_PER_PTE - 1))
371 #define pte_offset_kernel(dir, address) ((pte_t *) pmd_page_kernel(*(dir)) + \
374 /* x86-64 always has all page tables mapped. */
375 #define pte_offset_map(dir,address) pte_offset_kernel(dir,address)
376 #define pte_offset_map_nested(dir,address) pte_offset_kernel(dir,address)
377 #define pte_unmap(pte) /* NOP */
378 #define pte_unmap_nested(pte) /* NOP */
380 #define update_mmu_cache(vma,address,pte) do { } while (0)
382 /* We only update the dirty/accessed state if we set
383 * the dirty bit by hand in the kernel, since the hardware
384 * will do the accessed bit for us, and we don't want to
385 * race with other CPU's that might be updating the dirty
386 * bit at the same time. */
387 #define __HAVE_ARCH_PTEP_SET_ACCESS_FLAGS
388 #define ptep_set_access_flags(__vma, __address, __ptep, __entry, __dirty) \
391 set_pte(__ptep, __entry); \
392 flush_tlb_page(__vma, __address); \
396 /* Encode and de-code a swap entry */
397 #define __swp_type(x) (((x).val >> 1) & 0x3f)
398 #define __swp_offset(x) ((x).val >> 8)
399 #define __swp_entry(type, offset) ((swp_entry_t) { ((type) << 1) | ((offset) << 8) })
400 #define __pte_to_swp_entry(pte) ((swp_entry_t) { pte_val(pte) })
401 #define __swp_entry_to_pte(x) ((pte_t) { (x).val })
403 #endif /* !__ASSEMBLY__ */
405 extern int kern_addr_valid(unsigned long addr);
407 #define io_remap_page_range(vma, vaddr, paddr, size, prot) \
408 remap_pfn_range(vma, vaddr, (paddr) >> PAGE_SHIFT, size, prot)
410 #define io_remap_pfn_range(vma, vaddr, pfn, size, prot) \
411 remap_pfn_range(vma, vaddr, pfn, size, prot)
413 #define MK_IOSPACE_PFN(space, pfn) (pfn)
414 #define GET_IOSPACE(pfn) 0
415 #define GET_PFN(pfn) (pfn)
417 #define HAVE_ARCH_UNMAPPED_AREA
419 #define pgtable_cache_init() do { } while (0)
420 #define check_pgt_cache() do { } while (0)
422 #define PAGE_AGP PAGE_KERNEL_NOCACHE
423 #define HAVE_PAGE_AGP 1
425 /* fs/proc/kcore.c */
426 #define kc_vaddr_to_offset(v) ((v) & __VIRTUAL_MASK)
427 #define kc_offset_to_vaddr(o) \
428 (((o) & (1UL << (__VIRTUAL_MASK_SHIFT-1))) ? ((o) | (~__VIRTUAL_MASK)) : (o))
430 #define __HAVE_ARCH_PTEP_TEST_AND_CLEAR_YOUNG
431 #define __HAVE_ARCH_PTEP_TEST_AND_CLEAR_DIRTY
432 #define __HAVE_ARCH_PTEP_GET_AND_CLEAR
433 #define __HAVE_ARCH_PTEP_SET_WRPROTECT
434 #define __HAVE_ARCH_PTE_SAME
435 #include <asm-generic/pgtable.h>
437 #endif /* _X86_64_PGTABLE_H */