include/asm-x86/pgtable_64.h

   1 #ifndef _X86_64_PGTABLE_H
   2 #define _X86_64_PGTABLE_H
   3
   4 #include <linux/const.h>
   5 #ifndef __ASSEMBLY__
   6
   7 /*
   8  * This file contains the functions and defines necessary to modify and use
   9  * the x86-64 page table tree.
  10  */
  11 #include <asm/processor.h>
  12 #include <linux/bitops.h>
  13 #include <linux/threads.h>
  14 #include <asm/pda.h>
  15
  16 extern pud_t level3_kernel_pgt[512];
  17 extern pud_t level3_ident_pgt[512];
  18 extern pmd_t level2_kernel_pgt[512];
  19 extern pgd_t init_level4_pgt[];
  20 extern unsigned long __supported_pte_mask;
  21
  22 #define swapper_pg_dir init_level4_pgt
  23
  24 extern void paging_init(void);
  25 extern void clear_kernel_mapping(unsigned long addr, unsigned long size);
  26
  27 /*
  28  * ZERO_PAGE is a global shared page that is always zero: used
  29  * for zero-mapped memory areas etc..
  30  */
  31 extern unsigned long empty_zero_page[PAGE_SIZE/sizeof(unsigned long)];
  32 #define ZERO_PAGE(vaddr) (virt_to_page(empty_zero_page))
  33
  34 #endif /* !__ASSEMBLY__ */
  35
  36 /*
  37  * PGDIR_SHIFT determines what a top-level page table entry can map
  38  */
  39 #define PGDIR_SHIFT     39
  40 #define PTRS_PER_PGD    512
  41
  42 /*
  43  * 3rd level page
  44  */
  45 #define PUD_SHIFT       30
  46 #define PTRS_PER_PUD    512
  47
  48 /*
  49  * PMD_SHIFT determines the size of the area a middle-level
  50  * page table can map
  51  */
  52 #define PMD_SHIFT       21
  53 #define PTRS_PER_PMD    512
  54
  55 /*
  56  * entries per page directory level
  57  */
  58 #define PTRS_PER_PTE    512
  59
  60 #ifndef __ASSEMBLY__
  61
  62 #define pte_ERROR(e) \
  63         printk("%s:%d: bad pte %p(%016lx).\n", __FILE__, __LINE__, &(e), pte_val(e))
  64 #define pmd_ERROR(e) \
  65         printk("%s:%d: bad pmd %p(%016lx).\n", __FILE__, __LINE__, &(e), pmd_val(e))
  66 #define pud_ERROR(e) \
  67         printk("%s:%d: bad pud %p(%016lx).\n", __FILE__, __LINE__, &(e), pud_val(e))
  68 #define pgd_ERROR(e) \
  69         printk("%s:%d: bad pgd %p(%016lx).\n", __FILE__, __LINE__, &(e), pgd_val(e))
  70
  71 #define pgd_none(x)     (!pgd_val(x))
  72 #define pud_none(x)     (!pud_val(x))
  73
  74 static inline void set_pte(pte_t *dst, pte_t val)
  75 {
  76         pte_val(*dst) = pte_val(val);
  77 }
  78 #define set_pte_at(mm,addr,ptep,pteval) set_pte(ptep,pteval)
  79
  80 static inline void set_pmd(pmd_t *dst, pmd_t val)
  81 {
  82         *dst = val;
  83 }
  84
  85 static inline void set_pud(pud_t *dst, pud_t val)
  86 {
  87         *dst = val;
  88 }
  89
  90 static inline void pud_clear (pud_t *pud)
  91 {
  92         set_pud(pud, __pud(0));
  93 }
  94
  95 static inline void set_pgd(pgd_t *dst, pgd_t val)
  96 {
  97         *dst = val;
  98 }
  99
 100 static inline void pgd_clear (pgd_t * pgd)
 101 {
 102         set_pgd(pgd, __pgd(0));
 103 }
 104
 105 #define ptep_get_and_clear(mm,addr,xp)  __pte(xchg(&(xp)->pte, 0))
 106
 107 struct mm_struct;
 108
 109 static inline pte_t ptep_get_and_clear_full(struct mm_struct *mm, unsigned long addr, pte_t *ptep, int full)
 110 {
 111         pte_t pte;
 112         if (full) {
 113                 pte = *ptep;
 114                 *ptep = __pte(0);
 115         } else {
 116                 pte = ptep_get_and_clear(mm, addr, ptep);
 117         }
 118         return pte;
 119 }
 120
 121 #define pte_same(a, b)          ((a).pte == (b).pte)
 122
 123 #define pte_pgprot(a)   (__pgprot((a).pte & ~PHYSICAL_PAGE_MASK))
 124
 125 #endif /* !__ASSEMBLY__ */
 126
 127 #define PMD_SIZE        (_AC(1,UL) << PMD_SHIFT)
 128 #define PMD_MASK        (~(PMD_SIZE-1))
 129 #define PUD_SIZE        (_AC(1,UL) << PUD_SHIFT)
 130 #define PUD_MASK        (~(PUD_SIZE-1))
 131 #define PGDIR_SIZE      (_AC(1,UL) << PGDIR_SHIFT)
 132 #define PGDIR_MASK      (~(PGDIR_SIZE-1))
 133
 134
 135 #define MAXMEM           _AC(0x3fffffffffff, UL)
 136 #define VMALLOC_START    _AC(0xffffc20000000000, UL)
 137 #define VMALLOC_END      _AC(0xffffe1ffffffffff, UL)
 138 #define VMEMMAP_START    _AC(0xffffe20000000000, UL)
 139 #define MODULES_VADDR    _AC(0xffffffff88000000, UL)
 140 #define MODULES_END      _AC(0xfffffffffff00000, UL)
 141 #define MODULES_LEN   (MODULES_END - MODULES_VADDR)
 142
 143 #ifndef __ASSEMBLY__
 144
 145 static inline unsigned long pgd_bad(pgd_t pgd)
 146 {
 147         return pgd_val(pgd) & ~(PTE_MASK | _KERNPG_TABLE | _PAGE_USER);
 148 }
 149
 150 static inline unsigned long pud_bad(pud_t pud)
 151 {
 152         return pud_val(pud) & ~(PTE_MASK | _KERNPG_TABLE | _PAGE_USER);
 153 }
 154
 155 static inline unsigned long pmd_bad(pmd_t pmd)
 156 {
 157         return pmd_val(pmd) & ~(PTE_MASK | _KERNPG_TABLE | _PAGE_USER);
 158 }
 159
 160 #define pte_none(x)     (!pte_val(x))
 161 #define pte_present(x)  (pte_val(x) & (_PAGE_PRESENT | _PAGE_PROTNONE))
 162 #define pte_clear(mm,addr,xp)   do { set_pte_at(mm, addr, xp, __pte(0)); } while (0)
 163
 164 #define pages_to_mb(x) ((x) >> (20-PAGE_SHIFT)) /* FIXME: is this right? */
 165 #define pte_page(x)     pfn_to_page(pte_pfn(x))
 166 #define pte_pfn(x)  ((pte_val(x) & __PHYSICAL_MASK) >> PAGE_SHIFT)
 167
 168 static inline pte_t pfn_pte(unsigned long page_nr, pgprot_t pgprot)
 169 {
 170         pte_t pte;
 171         pte_val(pte) = (page_nr << PAGE_SHIFT);
 172         pte_val(pte) |= pgprot_val(pgprot);
 173         pte_val(pte) &= __supported_pte_mask;
 174         return pte;
 175 }
 176
 177 /*
 178  * The following only work if pte_present() is true.
 179  * Undefined behaviour if not..
 180  */
 181 #define __LARGE_PTE (_PAGE_PSE|_PAGE_PRESENT)
 182
 183 static inline int pte_dirty(pte_t pte)          { return pte_val(pte) & _PAGE_DIRTY; }
 184 static inline int pte_young(pte_t pte)          { return pte_val(pte) & _PAGE_ACCESSED; }
 185 static inline int pte_write(pte_t pte)          { return pte_val(pte) & _PAGE_RW; }
 186 static inline int pte_file(pte_t pte)           { return pte_val(pte) & _PAGE_FILE; }
 187 static inline int pte_huge(pte_t pte)           { return pte_val(pte) & _PAGE_PSE; }
 188
 189 static inline int pmd_large(pmd_t pte) {
 190         return (pmd_val(pte) & (_PAGE_PSE|_PAGE_PRESENT)) ==
 191                 (_PAGE_PSE|_PAGE_PRESENT);
 192 }
 193
 194 static inline pte_t pte_mkclean(pte_t pte)      { return __pte(pte_val(pte) & ~_PAGE_DIRTY); }
 195 static inline pte_t pte_mkold(pte_t pte)        { return __pte(pte_val(pte) & ~_PAGE_ACCESSED); }
 196 static inline pte_t pte_wrprotect(pte_t pte)    { return __pte(pte_val(pte) & ~_PAGE_RW); }
 197 static inline pte_t pte_mkexec(pte_t pte)       { return __pte(pte_val(pte) & ~_PAGE_NX); }
 198 static inline pte_t pte_mkdirty(pte_t pte)      { return __pte(pte_val(pte) | _PAGE_DIRTY); }
 199 static inline pte_t pte_mkyoung(pte_t pte)      { return __pte(pte_val(pte) | _PAGE_ACCESSED); }
 200 static inline pte_t pte_mkwrite(pte_t pte)      { return __pte(pte_val(pte) | _PAGE_RW); }
 201 static inline pte_t pte_mkhuge(pte_t pte)       { return __pte(pte_val(pte) | _PAGE_PSE); }
 202 static inline pte_t pte_clrhuge(pte_t pte)      { return __pte(pte_val(pte) & ~_PAGE_PSE); }
 203
 204 struct vm_area_struct;
 205
 206 static inline int ptep_test_and_clear_young(struct vm_area_struct *vma, unsigned long addr, pte_t *ptep)
 207 {
 208         if (!pte_young(*ptep))
 209                 return 0;
 210         return test_and_clear_bit(_PAGE_BIT_ACCESSED, &ptep->pte);
 211 }
 212
 213 static inline void ptep_set_wrprotect(struct mm_struct *mm, unsigned long addr, pte_t *ptep)
 214 {
 215         clear_bit(_PAGE_BIT_RW, &ptep->pte);
 216 }
 217
 218 /*
 219  * Macro to mark a page protection value as "uncacheable".
 220  */
 221 #define pgprot_noncached(prot)  (__pgprot(pgprot_val(prot) | _PAGE_PCD | _PAGE_PWT))
 222
 223 static inline int pmd_large(pmd_t pte) {
 224         return (pmd_val(pte) & __LARGE_PTE) == __LARGE_PTE;
 225 }
 226
 227
 228 /*
 229  * Conversion functions: convert a page and protection to a page entry,
 230  * and a page entry and page directory to the page they refer to.
 231  */
 232
 233 /*
 234  * Level 4 access.
 235  */
 236 #define pgd_page_vaddr(pgd) ((unsigned long) __va((unsigned long)pgd_val(pgd) & PTE_MASK))
 237 #define pgd_page(pgd)           (pfn_to_page(pgd_val(pgd) >> PAGE_SHIFT))
 238 #define pgd_index(address) (((address) >> PGDIR_SHIFT) & (PTRS_PER_PGD-1))
 239 #define pgd_offset(mm, addr) ((mm)->pgd + pgd_index(addr))
 240 #define pgd_offset_k(address) (init_level4_pgt + pgd_index(address))
 241 #define pgd_present(pgd) (pgd_val(pgd) & _PAGE_PRESENT)
 242 #define mk_kernel_pgd(address) ((pgd_t){ (address) | _KERNPG_TABLE })
 243
 244 /* PUD - Level3 access */
 245 /* to find an entry in a page-table-directory. */
 246 #define pud_page_vaddr(pud) ((unsigned long) __va(pud_val(pud) & PHYSICAL_PAGE_MASK))
 247 #define pud_page(pud)           (pfn_to_page(pud_val(pud) >> PAGE_SHIFT))
 248 #define pud_index(address) (((address) >> PUD_SHIFT) & (PTRS_PER_PUD-1))
 249 #define pud_offset(pgd, address) ((pud_t *) pgd_page_vaddr(*(pgd)) + pud_index(address))
 250 #define pud_present(pud) (pud_val(pud) & _PAGE_PRESENT)
 251
 252 /* PMD  - Level 2 access */
 253 #define pmd_page_vaddr(pmd) ((unsigned long) __va(pmd_val(pmd) & PTE_MASK))
 254 #define pmd_page(pmd)           (pfn_to_page(pmd_val(pmd) >> PAGE_SHIFT))
 255
 256 #define pmd_index(address) (((address) >> PMD_SHIFT) & (PTRS_PER_PMD-1))
 257 #define pmd_offset(dir, address) ((pmd_t *) pud_page_vaddr(*(dir)) + \
 258                         pmd_index(address))
 259 #define pmd_none(x)     (!pmd_val(x))
 260 #define pmd_present(x)  (pmd_val(x) & _PAGE_PRESENT)
 261 #define pmd_clear(xp)   do { set_pmd(xp, __pmd(0)); } while (0)
 262 #define pfn_pmd(nr,prot) (__pmd(((nr) << PAGE_SHIFT) | pgprot_val(prot)))
 263 #define pmd_pfn(x)  ((pmd_val(x) & __PHYSICAL_MASK) >> PAGE_SHIFT)
 264
 265 #define pte_to_pgoff(pte) ((pte_val(pte) & PHYSICAL_PAGE_MASK) >> PAGE_SHIFT)
 266 #define pgoff_to_pte(off) ((pte_t) { ((off) << PAGE_SHIFT) | _PAGE_FILE })
 267 #define PTE_FILE_MAX_BITS __PHYSICAL_MASK_SHIFT
 268
 269 /* PTE - Level 1 access. */
 270
 271 /* page, protection -> pte */
 272 #define mk_pte(page, pgprot)    pfn_pte(page_to_pfn(page), (pgprot))
 273
 274 /* Change flags of a PTE */
 275 static inline pte_t pte_modify(pte_t pte, pgprot_t newprot)
 276 {
 277         pte_val(pte) &= _PAGE_CHG_MASK;
 278         pte_val(pte) |= pgprot_val(newprot);
 279         pte_val(pte) &= __supported_pte_mask;
 280        return pte;
 281 }
 282
 283 #define pte_index(address) \
 284                 (((address) >> PAGE_SHIFT) & (PTRS_PER_PTE - 1))
 285 #define pte_offset_kernel(dir, address) ((pte_t *) pmd_page_vaddr(*(dir)) + \
 286                         pte_index(address))
 287
 288 /* x86-64 always has all page tables mapped. */
 289 #define pte_offset_map(dir,address) pte_offset_kernel(dir,address)
 290 #define pte_offset_map_nested(dir,address) pte_offset_kernel(dir,address)
 291 #define pte_unmap(pte) /* NOP */
 292 #define pte_unmap_nested(pte) /* NOP */
 293
 294 #define update_mmu_cache(vma,address,pte) do { } while (0)
 295
 296 /* We only update the dirty/accessed state if we set
 297  * the dirty bit by hand in the kernel, since the hardware
 298  * will do the accessed bit for us, and we don't want to
 299  * race with other CPU's that might be updating the dirty
 300  * bit at the same time. */
 301 #define  __HAVE_ARCH_PTEP_SET_ACCESS_FLAGS
 302 #define ptep_set_access_flags(__vma, __address, __ptep, __entry, __dirty) \
 303 ({                                                                        \
 304         int __changed = !pte_same(*(__ptep), __entry);                    \
 305         if (__changed && __dirty) {                                       \
 306                 set_pte(__ptep, __entry);                                 \
 307                 flush_tlb_page(__vma, __address);                         \
 308         }                                                                 \
 309         __changed;                                                        \
 310 })
 311
 312 /* Encode and de-code a swap entry */
 313 #define __swp_type(x)                   (((x).val >> 1) & 0x3f)
 314 #define __swp_offset(x)                 ((x).val >> 8)
 315 #define __swp_entry(type, offset)       ((swp_entry_t) { ((type) << 1) | ((offset) << 8) })
 316 #define __pte_to_swp_entry(pte)         ((swp_entry_t) { pte_val(pte) })
 317 #define __swp_entry_to_pte(x)           ((pte_t) { (x).val })
 318
 319 extern spinlock_t pgd_lock;
 320 extern struct list_head pgd_list;
 321
 322 extern int kern_addr_valid(unsigned long addr);
 323
 324 pte_t *lookup_address(unsigned long addr);
 325
 326 #define io_remap_pfn_range(vma, vaddr, pfn, size, prot)         \
 327                 remap_pfn_range(vma, vaddr, pfn, size, prot)
 328
 329 #define HAVE_ARCH_UNMAPPED_AREA
 330 #define HAVE_ARCH_UNMAPPED_AREA_TOPDOWN
 331
 332 #define pgtable_cache_init()   do { } while (0)
 333 #define check_pgt_cache()      do { } while (0)
 334
 335 #define PAGE_AGP    PAGE_KERNEL_NOCACHE
 336 #define HAVE_PAGE_AGP 1
 337
 338 /* fs/proc/kcore.c */
 339 #define kc_vaddr_to_offset(v) ((v) & __VIRTUAL_MASK)
 340 #define kc_offset_to_vaddr(o) \
 341    (((o) & (1UL << (__VIRTUAL_MASK_SHIFT-1))) ? ((o) | (~__VIRTUAL_MASK)) : (o))
 342
 343 #define __HAVE_ARCH_PTEP_TEST_AND_CLEAR_YOUNG
 344 #define __HAVE_ARCH_PTEP_GET_AND_CLEAR
 345 #define __HAVE_ARCH_PTEP_GET_AND_CLEAR_FULL
 346 #define __HAVE_ARCH_PTEP_SET_WRPROTECT
 347 #define __HAVE_ARCH_PTE_SAME
 348 #include <asm-generic/pgtable.h>
 349 #endif /* !__ASSEMBLY__ */
 350
 351 #endif /* _X86_64_PGTABLE_H */