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_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_NX           63       /* No execute: only valid after cpuid check */
  21
  22 #define _PAGE_PRESENT   (_AC(1, L)<<_PAGE_BIT_PRESENT)
  23 #define _PAGE_RW        (_AC(1, L)<<_PAGE_BIT_RW)
  24 #define _PAGE_USER      (_AC(1, L)<<_PAGE_BIT_USER)
  25 #define _PAGE_PWT       (_AC(1, L)<<_PAGE_BIT_PWT)
  26 #define _PAGE_PCD       (_AC(1, L)<<_PAGE_BIT_PCD)
  27 #define _PAGE_ACCESSED  (_AC(1, L)<<_PAGE_BIT_ACCESSED)
  28 #define _PAGE_DIRTY     (_AC(1, L)<<_PAGE_BIT_DIRTY)
  29 #define _PAGE_PSE       (_AC(1, L)<<_PAGE_BIT_PSE)      /* 2MB page */
  30 #define _PAGE_GLOBAL    (_AC(1, L)<<_PAGE_BIT_GLOBAL)   /* Global TLB entry */
  31 #define _PAGE_UNUSED1   (_AC(1, L)<<_PAGE_BIT_UNUSED1)
  32 #define _PAGE_UNUSED2   (_AC(1, L)<<_PAGE_BIT_UNUSED2)
  33 #define _PAGE_UNUSED3   (_AC(1, L)<<_PAGE_BIT_UNUSED3)
  34
  35 #if defined(CONFIG_X86_64) || defined(CONFIG_X86_PAE)
  36 #define _PAGE_NX        (_AC(1, ULL) << _PAGE_BIT_NX)
  37 #else
  38 #define _PAGE_NX        0
  39 #endif
  40
  41 /* If _PAGE_PRESENT is clear, we use these: */
  42 #define _PAGE_FILE      _PAGE_DIRTY     /* nonlinear file mapping, saved PTE; unset:swap */
  43 #define _PAGE_PROTNONE  _PAGE_PSE       /* if the user mapped it with PROT_NONE;
  44                                            pte_present gives true */
  45
  46 #define _PAGE_TABLE     (_PAGE_PRESENT | _PAGE_RW | _PAGE_USER | _PAGE_ACCESSED | _PAGE_DIRTY)
  47 #define _KERNPG_TABLE   (_PAGE_PRESENT | _PAGE_RW | _PAGE_ACCESSED | _PAGE_DIRTY)
  48
  49 #define _PAGE_CHG_MASK  (PTE_MASK | _PAGE_ACCESSED | _PAGE_DIRTY)
  50
  51 #define PAGE_NONE       __pgprot(_PAGE_PROTNONE | _PAGE_ACCESSED)
  52 #define PAGE_SHARED     __pgprot(_PAGE_PRESENT | _PAGE_RW | _PAGE_USER | _PAGE_ACCESSED | _PAGE_NX)
  53
  54 #define PAGE_SHARED_EXEC        __pgprot(_PAGE_PRESENT | _PAGE_RW | _PAGE_USER | _PAGE_ACCESSED)
  55 #define PAGE_COPY_NOEXEC        __pgprot(_PAGE_PRESENT | _PAGE_USER | _PAGE_ACCESSED | _PAGE_NX)
  56 #define PAGE_COPY_EXEC          __pgprot(_PAGE_PRESENT | _PAGE_USER | _PAGE_ACCESSED)
  57 #define PAGE_COPY               PAGE_COPY_NOEXEC
  58 #define PAGE_READONLY           __pgprot(_PAGE_PRESENT | _PAGE_USER | _PAGE_ACCESSED | _PAGE_NX)
  59 #define PAGE_READONLY_EXEC      __pgprot(_PAGE_PRESENT | _PAGE_USER | _PAGE_ACCESSED)
  60
  61 #ifdef CONFIG_X86_32
  62 #define _PAGE_KERNEL_EXEC \
  63         (_PAGE_PRESENT | _PAGE_RW | _PAGE_DIRTY | _PAGE_ACCESSED)
  64 #define _PAGE_KERNEL (_PAGE_KERNEL_EXEC | _PAGE_NX)
  65
  66 #ifndef __ASSEMBLY__
  67 extern unsigned long long __PAGE_KERNEL, __PAGE_KERNEL_EXEC;
  68 #endif  /* __ASSEMBLY__ */
  69 #else
  70 #define __PAGE_KERNEL_EXEC                                              \
  71         (_PAGE_PRESENT | _PAGE_RW | _PAGE_DIRTY | _PAGE_ACCESSED)
  72 #define __PAGE_KERNEL           (__PAGE_KERNEL_EXEC | _PAGE_NX)
  73 #endif
  74
  75 #define __PAGE_KERNEL_RO                (__PAGE_KERNEL & ~_PAGE_RW)
  76 #define __PAGE_KERNEL_RX                (__PAGE_KERNEL_EXEC & ~_PAGE_RW)
  77 #define __PAGE_KERNEL_NOCACHE           (__PAGE_KERNEL | _PAGE_PCD | _PAGE_PWT)
  78 #define __PAGE_KERNEL_VSYSCALL          (__PAGE_KERNEL_RX | _PAGE_USER)
  79 #define __PAGE_KERNEL_VSYSCALL_NOCACHE  (__PAGE_KERNEL_VSYSCALL | _PAGE_PCD | _PAGE_PWT)
  80 #define __PAGE_KERNEL_LARGE             (__PAGE_KERNEL | _PAGE_PSE)
  81 #define __PAGE_KERNEL_LARGE_EXEC        (__PAGE_KERNEL_EXEC | _PAGE_PSE)
  82
  83 #ifdef CONFIG_X86_32
  84 # define MAKE_GLOBAL(x)                 __pgprot((x))
  85 #else
  86 # define MAKE_GLOBAL(x)                 __pgprot((x) | _PAGE_GLOBAL)
  87 #endif
  88
  89 #define PAGE_KERNEL                     MAKE_GLOBAL(__PAGE_KERNEL)
  90 #define PAGE_KERNEL_RO                  MAKE_GLOBAL(__PAGE_KERNEL_RO)
  91 #define PAGE_KERNEL_EXEC                MAKE_GLOBAL(__PAGE_KERNEL_EXEC)
  92 #define PAGE_KERNEL_RX                  MAKE_GLOBAL(__PAGE_KERNEL_RX)
  93 #define PAGE_KERNEL_NOCACHE             MAKE_GLOBAL(__PAGE_KERNEL_NOCACHE)
  94 #define PAGE_KERNEL_LARGE               MAKE_GLOBAL(__PAGE_KERNEL_LARGE)
  95 #define PAGE_KERNEL_LARGE_EXEC          MAKE_GLOBAL(__PAGE_KERNEL_LARGE_EXEC)
  96 #define PAGE_KERNEL_VSYSCALL            MAKE_GLOBAL(__PAGE_KERNEL_VSYSCALL)
  97 #define PAGE_KERNEL_VSYSCALL_NOCACHE    MAKE_GLOBAL(__PAGE_KERNEL_VSYSCALL_NOCACHE)
  98
  99 /*         xwr */
 100 #define __P000  PAGE_NONE
 101 #define __P001  PAGE_READONLY
 102 #define __P010  PAGE_COPY
 103 #define __P011  PAGE_COPY
 104 #define __P100  PAGE_READONLY_EXEC
 105 #define __P101  PAGE_READONLY_EXEC
 106 #define __P110  PAGE_COPY_EXEC
 107 #define __P111  PAGE_COPY_EXEC
 108
 109 #define __S000  PAGE_NONE
 110 #define __S001  PAGE_READONLY
 111 #define __S010  PAGE_SHARED
 112 #define __S011  PAGE_SHARED
 113 #define __S100  PAGE_READONLY_EXEC
 114 #define __S101  PAGE_READONLY_EXEC
 115 #define __S110  PAGE_SHARED_EXEC
 116 #define __S111  PAGE_SHARED_EXEC
 117
 118 #ifndef __ASSEMBLY__
 119
 120 /*
 121  * ZERO_PAGE is a global shared page that is always zero: used
 122  * for zero-mapped memory areas etc..
 123  */
 124 extern unsigned long empty_zero_page[PAGE_SIZE/sizeof(unsigned long)];
 125 #define ZERO_PAGE(vaddr) (virt_to_page(empty_zero_page))
 126
 127
 128 /*
 129  * The following only work if pte_present() is true.
 130  * Undefined behaviour if not..
 131  */
 132 static inline int pte_dirty(pte_t pte)          { return pte_val(pte) & _PAGE_DIRTY; }
 133 static inline int pte_young(pte_t pte)          { return pte_val(pte) & _PAGE_ACCESSED; }
 134 static inline int pte_write(pte_t pte)          { return pte_val(pte) & _PAGE_RW; }
 135 static inline int pte_file(pte_t pte)           { return pte_val(pte) & _PAGE_FILE; }
 136 static inline int pte_huge(pte_t pte)           { return pte_val(pte) & _PAGE_PSE; }
 137
 138 static inline int pmd_large(pmd_t pte) {
 139         return (pmd_val(pte) & (_PAGE_PSE|_PAGE_PRESENT)) ==
 140                 (_PAGE_PSE|_PAGE_PRESENT);
 141 }
 142
 143 static inline pte_t pte_mkclean(pte_t pte)      { return __pte(pte_val(pte) & ~_PAGE_DIRTY); }
 144 static inline pte_t pte_mkold(pte_t pte)        { return __pte(pte_val(pte) & ~_PAGE_ACCESSED); }
 145 static inline pte_t pte_wrprotect(pte_t pte)    { return __pte(pte_val(pte) & ~_PAGE_RW); }
 146 static inline pte_t pte_mkexec(pte_t pte)       { return __pte(pte_val(pte) & ~_PAGE_NX); }
 147 static inline pte_t pte_mkdirty(pte_t pte)      { return __pte(pte_val(pte) | _PAGE_DIRTY); }
 148 static inline pte_t pte_mkyoung(pte_t pte)      { return __pte(pte_val(pte) | _PAGE_ACCESSED); }
 149 static inline pte_t pte_mkwrite(pte_t pte)      { return __pte(pte_val(pte) | _PAGE_RW); }
 150 static inline pte_t pte_mkhuge(pte_t pte)       { return __pte(pte_val(pte) | _PAGE_PSE); }
 151 static inline pte_t pte_clrhuge(pte_t pte)      { return __pte(pte_val(pte) & ~_PAGE_PSE); }
 152
 153 extern pteval_t __supported_pte_mask;
 154
 155 static inline pte_t pfn_pte(unsigned long page_nr, pgprot_t pgprot)
 156 {
 157         return __pte((((phys_addr_t)page_nr << PAGE_SHIFT) |
 158                       pgprot_val(pgprot)) & __supported_pte_mask);
 159 }
 160
 161 static inline pmd_t pfn_pmd(unsigned long page_nr, pgprot_t pgprot)
 162 {
 163         return __pmd((((phys_addr_t)page_nr << PAGE_SHIFT) |
 164                       pgprot_val(pgprot)) & __supported_pte_mask);
 165 }
 166
 167 static inline pte_t pte_modify(pte_t pte, pgprot_t newprot)
 168 {
 169         pteval_t val = pte_val(pte);
 170
 171         /*
 172          * Chop off the NX bit (if present), and add the NX portion of
 173          * the newprot (if present):
 174          */
 175         val &= _PAGE_CHG_MASK & ~_PAGE_NX;
 176         val |= pgprot_val(newprot) & __supported_pte_mask;
 177
 178         return __pte(val);
 179 }
 180
 181 #endif  /* __ASSEMBLY__ */
 182
 183 #ifdef CONFIG_X86_32
 184 # include "pgtable_32.h"
 185 #else
 186 # include "pgtable_64.h"
 187 #endif
 188
 189 #ifndef __ASSEMBLY__
 190
 191 #ifndef CONFIG_PARAVIRT
 192 /*
 193  * Rules for using pte_update - it must be called after any PTE update which
 194  * has not been done using the set_pte / clear_pte interfaces.  It is used by
 195  * shadow mode hypervisors to resynchronize the shadow page tables.  Kernel PTE
 196  * updates should either be sets, clears, or set_pte_atomic for P->P
 197  * transitions, which means this hook should only be called for user PTEs.
 198  * This hook implies a P->P protection or access change has taken place, which
 199  * requires a subsequent TLB flush.  The notification can optionally be delayed
 200  * until the TLB flush event by using the pte_update_defer form of the
 201  * interface, but care must be taken to assure that the flush happens while
 202  * still holding the same page table lock so that the shadow and primary pages
 203  * do not become out of sync on SMP.
 204  */
 205 #define pte_update(mm, addr, ptep)              do { } while (0)
 206 #define pte_update_defer(mm, addr, ptep)        do { } while (0)
 207 #endif
 208
 209 /* local pte updates need not use xchg for locking */
 210 static inline pte_t native_local_ptep_get_and_clear(pte_t *ptep)
 211 {
 212         pte_t res = *ptep;
 213
 214         /* Pure native function needs no input for mm, addr */
 215         native_pte_clear(NULL, 0, ptep);
 216         return res;
 217 }
 218
 219 /*
 220  * We only update the dirty/accessed state if we set
 221  * the dirty bit by hand in the kernel, since the hardware
 222  * will do the accessed bit for us, and we don't want to
 223  * race with other CPU's that might be updating the dirty
 224  * bit at the same time.
 225  */
 226 #define  __HAVE_ARCH_PTEP_SET_ACCESS_FLAGS
 227 #define ptep_set_access_flags(vma, address, ptep, entry, dirty)         \
 228 ({                                                                      \
 229         int __changed = !pte_same(*(ptep), entry);                      \
 230         if (__changed && dirty) {                                       \
 231                 *ptep = entry;                                          \
 232                 pte_update_defer((vma)->vm_mm, (address), (ptep));      \
 233                 flush_tlb_page(vma, address);                           \
 234         }                                                               \
 235         __changed;                                                      \
 236 })
 237
 238 #define __HAVE_ARCH_PTEP_TEST_AND_CLEAR_YOUNG
 239 #define ptep_test_and_clear_young(vma, addr, ptep) ({                   \
 240         int __ret = 0;                                                  \
 241         if (pte_young(*(ptep)))                                         \
 242                 __ret = test_and_clear_bit(_PAGE_BIT_ACCESSED,          \
 243                                            &(ptep)->pte);               \
 244         if (__ret)                                                      \
 245                 pte_update((vma)->vm_mm, addr, ptep);                   \
 246         __ret;                                                          \
 247 })
 248
 249 #define __HAVE_ARCH_PTEP_CLEAR_YOUNG_FLUSH
 250 #define ptep_clear_flush_young(vma, address, ptep)                      \
 251 ({                                                                      \
 252         int __young;                                                    \
 253         __young = ptep_test_and_clear_young((vma), (address), (ptep));  \
 254         if (__young)                                                    \
 255                 flush_tlb_page(vma, address);                           \
 256         __young;                                                        \
 257 })
 258
 259 #define __HAVE_ARCH_PTEP_GET_AND_CLEAR
 260 static inline pte_t ptep_get_and_clear(struct mm_struct *mm, unsigned long addr, pte_t *ptep)
 261 {
 262         pte_t pte = native_ptep_get_and_clear(ptep);
 263         pte_update(mm, addr, ptep);
 264         return pte;
 265 }
 266
 267 #define __HAVE_ARCH_PTEP_GET_AND_CLEAR_FULL
 268 static inline pte_t ptep_get_and_clear_full(struct mm_struct *mm, unsigned long addr, pte_t *ptep, int full)
 269 {
 270         pte_t pte;
 271         if (full) {
 272                 /*
 273                  * Full address destruction in progress; paravirt does not
 274                  * care about updates and native needs no locking
 275                  */
 276                 pte = native_local_ptep_get_and_clear(ptep);
 277         } else {
 278                 pte = ptep_get_and_clear(mm, addr, ptep);
 279         }
 280         return pte;
 281 }
 282
 283 #define __HAVE_ARCH_PTEP_SET_WRPROTECT
 284 static inline void ptep_set_wrprotect(struct mm_struct *mm, unsigned long addr, pte_t *ptep)
 285 {
 286         clear_bit(_PAGE_BIT_RW, (unsigned long *)&ptep->pte);
 287         pte_update(mm, addr, ptep);
 288 }
 289
 290 #ifndef CONFIG_PARAVIRT
 291 #define pte_clear(mm, addr, ptep)               native_pte_clear(mm, addr, ptep)
 292 #endif  /* !CONFIG_PARAVIRT */
 293
 294 #include <asm-generic/pgtable.h>
 295 #endif  /* __ASSEMBLY__ */
 296
 297 #endif  /* _ASM_X86_PGTABLE_H */