include/asm-x86/pgtable.h

   1 #ifndef _ASM_X86_PGTABLE_H
   2 #define _ASM_X86_PGTABLE_H
   3
   4 #define USER_PTRS_PER_PGD       ((TASK_SIZE-1)/PGDIR_SIZE+1)
   5 #define FIRST_USER_ADDRESS      0
   6
   7 #define _PAGE_BIT_PRESENT       0
   8 #define _PAGE_BIT_RW            1
   9 #define _PAGE_BIT_USER          2
  10 #define _PAGE_BIT_PWT           3
  11 #define _PAGE_BIT_PCD           4
  12 #define _PAGE_BIT_ACCESSED      5
  13 #define _PAGE_BIT_DIRTY         6
  14 #define _PAGE_BIT_FILE          6
  15 #define _PAGE_BIT_PSE           7       /* 4 MB (or 2MB) page */
  16 #define _PAGE_BIT_PAT           7       /* on 4KB pages */
  17 #define _PAGE_BIT_GLOBAL        8       /* Global TLB entry PPro+ */
  18 #define _PAGE_BIT_UNUSED1       9       /* available for programmer */
  19 #define _PAGE_BIT_UNUSED2       10
  20 #define _PAGE_BIT_UNUSED3       11
  21 #define _PAGE_BIT_PAT_LARGE     12      /* On 2MB or 1GB pages */
  22 #define _PAGE_BIT_NX           63       /* No execute: only valid after cpuid check */
  23
  24 /*
  25  * Note: we use _AC(1, L) instead of _AC(1, UL) so that we get a
  26  * sign-extended value on 32-bit with all 1's in the upper word,
  27  * which preserves the upper pte values on 64-bit ptes:
  28  */
  29 #define _PAGE_PRESENT   (_AC(1, L)<<_PAGE_BIT_PRESENT)
  30 #define _PAGE_RW        (_AC(1, L)<<_PAGE_BIT_RW)
  31 #define _PAGE_USER      (_AC(1, L)<<_PAGE_BIT_USER)
  32 #define _PAGE_PWT       (_AC(1, L)<<_PAGE_BIT_PWT)
  33 #define _PAGE_PCD       (_AC(1, L)<<_PAGE_BIT_PCD)
  34 #define _PAGE_ACCESSED  (_AC(1, L)<<_PAGE_BIT_ACCESSED)
  35 #define _PAGE_DIRTY     (_AC(1, L)<<_PAGE_BIT_DIRTY)
  36 #define _PAGE_PSE       (_AC(1, L)<<_PAGE_BIT_PSE)      /* 2MB page */
  37 #define _PAGE_GLOBAL    (_AC(1, L)<<_PAGE_BIT_GLOBAL)   /* Global TLB entry */
  38 #define _PAGE_UNUSED1   (_AC(1, L)<<_PAGE_BIT_UNUSED1)
  39 #define _PAGE_UNUSED2   (_AC(1, L)<<_PAGE_BIT_UNUSED2)
  40 #define _PAGE_UNUSED3   (_AC(1, L)<<_PAGE_BIT_UNUSED3)
  41 #define _PAGE_PAT       (_AC(1, L)<<_PAGE_BIT_PAT)
  42 #define _PAGE_PAT_LARGE (_AC(1, L)<<_PAGE_BIT_PAT_LARGE)
  43
  44 #if defined(CONFIG_X86_64) || defined(CONFIG_X86_PAE)
  45 #define _PAGE_NX        (_AC(1, ULL) << _PAGE_BIT_NX)
  46 #else
  47 #define _PAGE_NX        0
  48 #endif
  49
  50 /* If _PAGE_PRESENT is clear, we use these: */
  51 #define _PAGE_FILE      _PAGE_DIRTY     /* nonlinear file mapping, saved PTE; unset:swap */
  52 #define _PAGE_PROTNONE  _PAGE_PSE       /* if the user mapped it with PROT_NONE;
  53                                            pte_present gives true */
  54
  55 #define _PAGE_TABLE     (_PAGE_PRESENT | _PAGE_RW | _PAGE_USER | _PAGE_ACCESSED | _PAGE_DIRTY)
  56 #define _KERNPG_TABLE   (_PAGE_PRESENT | _PAGE_RW | _PAGE_ACCESSED | _PAGE_DIRTY)
  57
  58 #define _PAGE_CHG_MASK  (PTE_MASK | _PAGE_ACCESSED | _PAGE_DIRTY)
  59
  60 #define PAGE_NONE       __pgprot(_PAGE_PROTNONE | _PAGE_ACCESSED)
  61 #define PAGE_SHARED     __pgprot(_PAGE_PRESENT | _PAGE_RW | _PAGE_USER | _PAGE_ACCESSED | _PAGE_NX)
  62
  63 #define PAGE_SHARED_EXEC        __pgprot(_PAGE_PRESENT | _PAGE_RW | _PAGE_USER | _PAGE_ACCESSED)
  64 #define PAGE_COPY_NOEXEC        __pgprot(_PAGE_PRESENT | _PAGE_USER | _PAGE_ACCESSED | _PAGE_NX)
  65 #define PAGE_COPY_EXEC          __pgprot(_PAGE_PRESENT | _PAGE_USER | _PAGE_ACCESSED)
  66 #define PAGE_COPY               PAGE_COPY_NOEXEC
  67 #define PAGE_READONLY           __pgprot(_PAGE_PRESENT | _PAGE_USER | _PAGE_ACCESSED | _PAGE_NX)
  68 #define PAGE_READONLY_EXEC      __pgprot(_PAGE_PRESENT | _PAGE_USER | _PAGE_ACCESSED)
  69
  70 #ifdef CONFIG_X86_32
  71 #define _PAGE_KERNEL_EXEC \
  72         (_PAGE_PRESENT | _PAGE_RW | _PAGE_DIRTY | _PAGE_ACCESSED)
  73 #define _PAGE_KERNEL (_PAGE_KERNEL_EXEC | _PAGE_NX)
  74
  75 #ifndef __ASSEMBLY__
  76 extern pteval_t __PAGE_KERNEL, __PAGE_KERNEL_EXEC;
  77 #endif  /* __ASSEMBLY__ */
  78 #else
  79 #define __PAGE_KERNEL_EXEC                                              \
  80         (_PAGE_PRESENT | _PAGE_RW | _PAGE_DIRTY | _PAGE_ACCESSED)
  81 #define __PAGE_KERNEL           (__PAGE_KERNEL_EXEC | _PAGE_NX)
  82 #endif
  83
  84 #define __PAGE_KERNEL_RO                (__PAGE_KERNEL & ~_PAGE_RW)
  85 #define __PAGE_KERNEL_RX                (__PAGE_KERNEL_EXEC & ~_PAGE_RW)
  86 #define __PAGE_KERNEL_EXEC_NOCACHE      (__PAGE_KERNEL_EXEC | _PAGE_PCD | _PAGE_PWT)
  87 #define __PAGE_KERNEL_NOCACHE           (__PAGE_KERNEL | _PAGE_PCD | _PAGE_PWT)
  88 #define __PAGE_KERNEL_UC_MINUS          (__PAGE_KERNEL | _PAGE_PCD)
  89 #define __PAGE_KERNEL_VSYSCALL          (__PAGE_KERNEL_RX | _PAGE_USER)
  90 #define __PAGE_KERNEL_VSYSCALL_NOCACHE  (__PAGE_KERNEL_VSYSCALL | _PAGE_PCD | _PAGE_PWT)
  91 #define __PAGE_KERNEL_LARGE             (__PAGE_KERNEL | _PAGE_PSE)
  92 #define __PAGE_KERNEL_LARGE_EXEC        (__PAGE_KERNEL_EXEC | _PAGE_PSE)
  93
  94 #ifdef CONFIG_X86_32
  95 # define MAKE_GLOBAL(x)                 __pgprot((x))
  96 #else
  97 # define MAKE_GLOBAL(x)                 __pgprot((x) | _PAGE_GLOBAL)
  98 #endif
  99
 100 #define PAGE_KERNEL                     MAKE_GLOBAL(__PAGE_KERNEL)
 101 #define PAGE_KERNEL_RO                  MAKE_GLOBAL(__PAGE_KERNEL_RO)
 102 #define PAGE_KERNEL_EXEC                MAKE_GLOBAL(__PAGE_KERNEL_EXEC)
 103 #define PAGE_KERNEL_RX                  MAKE_GLOBAL(__PAGE_KERNEL_RX)
 104 #define PAGE_KERNEL_NOCACHE             MAKE_GLOBAL(__PAGE_KERNEL_NOCACHE)
 105 #define PAGE_KERNEL_UC_MINUS            MAKE_GLOBAL(__PAGE_KERNEL_UC_MINUS)
 106 #define PAGE_KERNEL_EXEC_NOCACHE        MAKE_GLOBAL(__PAGE_KERNEL_EXEC_NOCACHE)
 107 #define PAGE_KERNEL_LARGE               MAKE_GLOBAL(__PAGE_KERNEL_LARGE)
 108 #define PAGE_KERNEL_LARGE_EXEC          MAKE_GLOBAL(__PAGE_KERNEL_LARGE_EXEC)
 109 #define PAGE_KERNEL_VSYSCALL            MAKE_GLOBAL(__PAGE_KERNEL_VSYSCALL)
 110 #define PAGE_KERNEL_VSYSCALL_NOCACHE    MAKE_GLOBAL(__PAGE_KERNEL_VSYSCALL_NOCACHE)
 111
 112 /*         xwr */
 113 #define __P000  PAGE_NONE
 114 #define __P001  PAGE_READONLY
 115 #define __P010  PAGE_COPY
 116 #define __P011  PAGE_COPY
 117 #define __P100  PAGE_READONLY_EXEC
 118 #define __P101  PAGE_READONLY_EXEC
 119 #define __P110  PAGE_COPY_EXEC
 120 #define __P111  PAGE_COPY_EXEC
 121
 122 #define __S000  PAGE_NONE
 123 #define __S001  PAGE_READONLY
 124 #define __S010  PAGE_SHARED
 125 #define __S011  PAGE_SHARED
 126 #define __S100  PAGE_READONLY_EXEC
 127 #define __S101  PAGE_READONLY_EXEC
 128 #define __S110  PAGE_SHARED_EXEC
 129 #define __S111  PAGE_SHARED_EXEC
 130
 131 #ifndef __ASSEMBLY__
 132
 133 /*
 134  * ZERO_PAGE is a global shared page that is always zero: used
 135  * for zero-mapped memory areas etc..
 136  */
 137 extern unsigned long empty_zero_page[PAGE_SIZE/sizeof(unsigned long)];
 138 #define ZERO_PAGE(vaddr) (virt_to_page(empty_zero_page))
 139
 140 extern spinlock_t pgd_lock;
 141 extern struct list_head pgd_list;
 142
 143 /*
 144  * The following only work if pte_present() is true.
 145  * Undefined behaviour if not..
 146  */
 147 static inline int pte_dirty(pte_t pte)          { return pte_val(pte) & _PAGE_DIRTY; }
 148 static inline int pte_young(pte_t pte)          { return pte_val(pte) & _PAGE_ACCESSED; }
 149 static inline int pte_write(pte_t pte)          { return pte_val(pte) & _PAGE_RW; }
 150 static inline int pte_file(pte_t pte)           { return pte_val(pte) & _PAGE_FILE; }
 151 static inline int pte_huge(pte_t pte)           { return pte_val(pte) & _PAGE_PSE; }
 152 static inline int pte_global(pte_t pte)         { return pte_val(pte) & _PAGE_GLOBAL; }
 153 static inline int pte_exec(pte_t pte)           { return !(pte_val(pte) & _PAGE_NX); }
 154
 155 static inline int pmd_large(pmd_t pte) {
 156         return (pmd_val(pte) & (_PAGE_PSE|_PAGE_PRESENT)) ==
 157                 (_PAGE_PSE|_PAGE_PRESENT);
 158 }
 159
 160 static inline pte_t pte_mkclean(pte_t pte)      { return __pte(pte_val(pte) & ~(pteval_t)_PAGE_DIRTY); }
 161 static inline pte_t pte_mkold(pte_t pte)        { return __pte(pte_val(pte) & ~(pteval_t)_PAGE_ACCESSED); }
 162 static inline pte_t pte_wrprotect(pte_t pte)    { return __pte(pte_val(pte) & ~(pteval_t)_PAGE_RW); }
 163 static inline pte_t pte_mkexec(pte_t pte)       { return __pte(pte_val(pte) & ~(pteval_t)_PAGE_NX); }
 164 static inline pte_t pte_mkdirty(pte_t pte)      { return __pte(pte_val(pte) | _PAGE_DIRTY); }
 165 static inline pte_t pte_mkyoung(pte_t pte)      { return __pte(pte_val(pte) | _PAGE_ACCESSED); }
 166 static inline pte_t pte_mkwrite(pte_t pte)      { return __pte(pte_val(pte) | _PAGE_RW); }
 167 static inline pte_t pte_mkhuge(pte_t pte)       { return __pte(pte_val(pte) | _PAGE_PSE); }
 168 static inline pte_t pte_clrhuge(pte_t pte)      { return __pte(pte_val(pte) & ~(pteval_t)_PAGE_PSE); }
 169 static inline pte_t pte_mkglobal(pte_t pte)     { return __pte(pte_val(pte) | _PAGE_GLOBAL); }
 170 static inline pte_t pte_clrglobal(pte_t pte)    { return __pte(pte_val(pte) & ~(pteval_t)_PAGE_GLOBAL); }
 171
 172 extern pteval_t __supported_pte_mask;
 173
 174 static inline pte_t pfn_pte(unsigned long page_nr, pgprot_t pgprot)
 175 {
 176         return __pte((((phys_addr_t)page_nr << PAGE_SHIFT) |
 177                       pgprot_val(pgprot)) & __supported_pte_mask);
 178 }
 179
 180 static inline pmd_t pfn_pmd(unsigned long page_nr, pgprot_t pgprot)
 181 {
 182         return __pmd((((phys_addr_t)page_nr << PAGE_SHIFT) |
 183                       pgprot_val(pgprot)) & __supported_pte_mask);
 184 }
 185
 186 static inline pte_t pte_modify(pte_t pte, pgprot_t newprot)
 187 {
 188         pteval_t val = pte_val(pte);
 189
 190         /*
 191          * Chop off the NX bit (if present), and add the NX portion of
 192          * the newprot (if present):
 193          */
 194         val &= _PAGE_CHG_MASK & ~_PAGE_NX;
 195         val |= pgprot_val(newprot) & __supported_pte_mask;
 196
 197         return __pte(val);
 198 }
 199
 200 #define pte_pgprot(x) __pgprot(pte_val(x) & (0xfff | _PAGE_NX))
 201
 202 #define canon_pgprot(p) __pgprot(pgprot_val(p) & __supported_pte_mask)
 203
 204 #ifdef CONFIG_PARAVIRT
 205 #include <asm/paravirt.h>
 206 #else  /* !CONFIG_PARAVIRT */
 207 #define set_pte(ptep, pte)              native_set_pte(ptep, pte)
 208 #define set_pte_at(mm, addr, ptep, pte) native_set_pte_at(mm, addr, ptep, pte)
 209
 210 #define set_pte_present(mm, addr, ptep, pte)                            \
 211         native_set_pte_present(mm, addr, ptep, pte)
 212 #define set_pte_atomic(ptep, pte)                                       \
 213         native_set_pte_atomic(ptep, pte)
 214
 215 #define set_pmd(pmdp, pmd)              native_set_pmd(pmdp, pmd)
 216
 217 #ifndef __PAGETABLE_PUD_FOLDED
 218 #define set_pgd(pgdp, pgd)              native_set_pgd(pgdp, pgd)
 219 #define pgd_clear(pgd)                  native_pgd_clear(pgd)
 220 #endif
 221
 222 #ifndef set_pud
 223 # define set_pud(pudp, pud)             native_set_pud(pudp, pud)
 224 #endif
 225
 226 #ifndef __PAGETABLE_PMD_FOLDED
 227 #define pud_clear(pud)                  native_pud_clear(pud)
 228 #endif
 229
 230 #define pte_clear(mm, addr, ptep)       native_pte_clear(mm, addr, ptep)
 231 #define pmd_clear(pmd)                  native_pmd_clear(pmd)
 232
 233 #define pte_update(mm, addr, ptep)              do { } while (0)
 234 #define pte_update_defer(mm, addr, ptep)        do { } while (0)
 235 #endif  /* CONFIG_PARAVIRT */
 236
 237 #endif  /* __ASSEMBLY__ */
 238
 239 #ifdef CONFIG_X86_32
 240 # include "pgtable_32.h"
 241 #else
 242 # include "pgtable_64.h"
 243 #endif
 244
 245 #ifndef __ASSEMBLY__
 246
 247 enum {
 248         PG_LEVEL_NONE,
 249         PG_LEVEL_4K,
 250         PG_LEVEL_2M,
 251         PG_LEVEL_1G,
 252 };
 253
 254 /*
 255  * Helper function that returns the kernel pagetable entry controlling
 256  * the virtual address 'address'. NULL means no pagetable entry present.
 257  * NOTE: the return type is pte_t but if the pmd is PSE then we return it
 258  * as a pte too.
 259  */
 260 extern pte_t *lookup_address(unsigned long address, unsigned int *level);
 261
 262 /* local pte updates need not use xchg for locking */
 263 static inline pte_t native_local_ptep_get_and_clear(pte_t *ptep)
 264 {
 265         pte_t res = *ptep;
 266
 267         /* Pure native function needs no input for mm, addr */
 268         native_pte_clear(NULL, 0, ptep);
 269         return res;
 270 }
 271
 272 static inline void native_set_pte_at(struct mm_struct *mm, unsigned long addr,
 273                                      pte_t *ptep , pte_t pte)
 274 {
 275         native_set_pte(ptep, pte);
 276 }
 277
 278 #ifndef CONFIG_PARAVIRT
 279 /*
 280  * Rules for using pte_update - it must be called after any PTE update which
 281  * has not been done using the set_pte / clear_pte interfaces.  It is used by
 282  * shadow mode hypervisors to resynchronize the shadow page tables.  Kernel PTE
 283  * updates should either be sets, clears, or set_pte_atomic for P->P
 284  * transitions, which means this hook should only be called for user PTEs.
 285  * This hook implies a P->P protection or access change has taken place, which
 286  * requires a subsequent TLB flush.  The notification can optionally be delayed
 287  * until the TLB flush event by using the pte_update_defer form of the
 288  * interface, but care must be taken to assure that the flush happens while
 289  * still holding the same page table lock so that the shadow and primary pages
 290  * do not become out of sync on SMP.
 291  */
 292 #define pte_update(mm, addr, ptep)              do { } while (0)
 293 #define pte_update_defer(mm, addr, ptep)        do { } while (0)
 294 #endif
 295
 296 /*
 297  * We only update the dirty/accessed state if we set
 298  * the dirty bit by hand in the kernel, since the hardware
 299  * will do the accessed bit for us, and we don't want to
 300  * race with other CPU's that might be updating the dirty
 301  * bit at the same time.
 302  */
 303 #define  __HAVE_ARCH_PTEP_SET_ACCESS_FLAGS
 304 #define ptep_set_access_flags(vma, address, ptep, entry, dirty)         \
 305 ({                                                                      \
 306         int __changed = !pte_same(*(ptep), entry);                      \
 307         if (__changed && dirty) {                                       \
 308                 *ptep = entry;                                          \
 309                 pte_update_defer((vma)->vm_mm, (address), (ptep));      \
 310                 flush_tlb_page(vma, address);                           \
 311         }                                                               \
 312         __changed;                                                      \
 313 })
 314
 315 #define __HAVE_ARCH_PTEP_TEST_AND_CLEAR_YOUNG
 316 #define ptep_test_and_clear_young(vma, addr, ptep) ({                   \
 317         int __ret = 0;                                                  \
 318         if (pte_young(*(ptep)))                                         \
 319                 __ret = test_and_clear_bit(_PAGE_BIT_ACCESSED,          \
 320                                            &(ptep)->pte);               \
 321         if (__ret)                                                      \
 322                 pte_update((vma)->vm_mm, addr, ptep);                   \
 323         __ret;                                                          \
 324 })
 325
 326 #define __HAVE_ARCH_PTEP_CLEAR_YOUNG_FLUSH
 327 #define ptep_clear_flush_young(vma, address, ptep)                      \
 328 ({                                                                      \
 329         int __young;                                                    \
 330         __young = ptep_test_and_clear_young((vma), (address), (ptep));  \
 331         if (__young)                                                    \
 332                 flush_tlb_page(vma, address);                           \
 333         __young;                                                        \
 334 })
 335
 336 #define __HAVE_ARCH_PTEP_GET_AND_CLEAR
 337 static inline pte_t ptep_get_and_clear(struct mm_struct *mm, unsigned long addr, pte_t *ptep)
 338 {
 339         pte_t pte = native_ptep_get_and_clear(ptep);
 340         pte_update(mm, addr, ptep);
 341         return pte;
 342 }
 343
 344 #define __HAVE_ARCH_PTEP_GET_AND_CLEAR_FULL
 345 static inline pte_t ptep_get_and_clear_full(struct mm_struct *mm, unsigned long addr, pte_t *ptep, int full)
 346 {
 347         pte_t pte;
 348         if (full) {
 349                 /*
 350                  * Full address destruction in progress; paravirt does not
 351                  * care about updates and native needs no locking
 352                  */
 353                 pte = native_local_ptep_get_and_clear(ptep);
 354         } else {
 355                 pte = ptep_get_and_clear(mm, addr, ptep);
 356         }
 357         return pte;
 358 }
 359
 360 #define __HAVE_ARCH_PTEP_SET_WRPROTECT
 361 static inline void ptep_set_wrprotect(struct mm_struct *mm, unsigned long addr, pte_t *ptep)
 362 {
 363         clear_bit(_PAGE_BIT_RW, (unsigned long *)&ptep->pte);
 364         pte_update(mm, addr, ptep);
 365 }
 366
 367 #include <asm-generic/pgtable.h>
 368 #endif  /* __ASSEMBLY__ */
 369
 370 #endif  /* _ASM_X86_PGTABLE_H */