pgtable-3level.h 9.3 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281
  1. /*
  2. * arch/arm/include/asm/pgtable-3level.h
  3. *
  4. * Copyright (C) 2011 ARM Ltd.
  5. * Author: Catalin Marinas <catalin.marinas@arm.com>
  6. *
  7. * This program is free software; you can redistribute it and/or modify
  8. * it under the terms of the GNU General Public License version 2 as
  9. * published by the Free Software Foundation.
  10. *
  11. * This program is distributed in the hope that it will be useful,
  12. * but WITHOUT ANY WARRANTY; without even the implied warranty of
  13. * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  14. * GNU General Public License for more details.
  15. *
  16. * You should have received a copy of the GNU General Public License
  17. * along with this program; if not, write to the Free Software
  18. * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
  19. */
  20. #ifndef _ASM_PGTABLE_3LEVEL_H
  21. #define _ASM_PGTABLE_3LEVEL_H
  22. /*
  23. * With LPAE, there are 3 levels of page tables. Each level has 512 entries of
  24. * 8 bytes each, occupying a 4K page. The first level table covers a range of
  25. * 512GB, each entry representing 1GB. Since we are limited to 4GB input
  26. * address range, only 4 entries in the PGD are used.
  27. *
  28. * There are enough spare bits in a page table entry for the kernel specific
  29. * state.
  30. */
  31. #define PTRS_PER_PTE 512
  32. #define PTRS_PER_PMD 512
  33. #define PTRS_PER_PGD 4
  34. #define PTE_HWTABLE_PTRS (0)
  35. #define PTE_HWTABLE_OFF (0)
  36. #define PTE_HWTABLE_SIZE (PTRS_PER_PTE * sizeof(u64))
  37. /*
  38. * PGDIR_SHIFT determines the size a top-level page table entry can map.
  39. */
  40. #define PGDIR_SHIFT 30
  41. /*
  42. * PMD_SHIFT determines the size a middle-level page table entry can map.
  43. */
  44. #define PMD_SHIFT 21
  45. #define PMD_SIZE (1UL << PMD_SHIFT)
  46. #define PMD_MASK (~((1 << PMD_SHIFT) - 1))
  47. #define PGDIR_SIZE (1UL << PGDIR_SHIFT)
  48. #define PGDIR_MASK (~((1 << PGDIR_SHIFT) - 1))
  49. /*
  50. * section address mask and size definitions.
  51. */
  52. #define SECTION_SHIFT 21
  53. #define SECTION_SIZE (1UL << SECTION_SHIFT)
  54. #define SECTION_MASK (~((1 << SECTION_SHIFT) - 1))
  55. #define USER_PTRS_PER_PGD (PAGE_OFFSET / PGDIR_SIZE)
  56. /*
  57. * Hugetlb definitions.
  58. */
  59. #define HPAGE_SHIFT PMD_SHIFT
  60. #define HPAGE_SIZE (_AC(1, UL) << HPAGE_SHIFT)
  61. #define HPAGE_MASK (~(HPAGE_SIZE - 1))
  62. #define HUGETLB_PAGE_ORDER (HPAGE_SHIFT - PAGE_SHIFT)
  63. /*
  64. * "Linux" PTE definitions for LPAE.
  65. *
  66. * These bits overlap with the hardware bits but the naming is preserved for
  67. * consistency with the classic page table format.
  68. */
  69. #define L_PTE_VALID (_AT(pteval_t, 1) << 0) /* Valid */
  70. #define L_PTE_PRESENT (_AT(pteval_t, 3) << 0) /* Present */
  71. #define L_PTE_FILE (_AT(pteval_t, 1) << 2) /* only when !PRESENT */
  72. #define L_PTE_USER (_AT(pteval_t, 1) << 6) /* AP[1] */
  73. #define L_PTE_SHARED (_AT(pteval_t, 3) << 8) /* SH[1:0], inner shareable */
  74. #define L_PTE_YOUNG (_AT(pteval_t, 1) << 10) /* AF */
  75. #define L_PTE_XN (_AT(pteval_t, 1) << 54) /* XN */
  76. #define L_PTE_DIRTY (_AT(pteval_t, 1) << 55)
  77. #define L_PTE_SPECIAL (_AT(pteval_t, 1) << 56)
  78. #define L_PTE_NONE (_AT(pteval_t, 1) << 57) /* PROT_NONE */
  79. #define L_PTE_RDONLY (_AT(pteval_t, 1) << 58) /* READ ONLY */
  80. #define L_PMD_SECT_VALID (_AT(pmdval_t, 1) << 0)
  81. #define L_PMD_SECT_DIRTY (_AT(pmdval_t, 1) << 55)
  82. #define L_PMD_SECT_SPLITTING (_AT(pmdval_t, 1) << 56)
  83. #define L_PMD_SECT_NONE (_AT(pmdval_t, 1) << 57)
  84. #define L_PMD_SECT_RDONLY (_AT(pteval_t, 1) << 58)
  85. /*
  86. * To be used in assembly code with the upper page attributes.
  87. */
  88. #define L_PTE_XN_HIGH (1 << (54 - 32))
  89. #define L_PTE_DIRTY_HIGH (1 << (55 - 32))
  90. /*
  91. * AttrIndx[2:0] encoding (mapping attributes defined in the MAIR* registers).
  92. */
  93. #define L_PTE_MT_UNCACHED (_AT(pteval_t, 0) << 2) /* strongly ordered */
  94. #define L_PTE_MT_BUFFERABLE (_AT(pteval_t, 1) << 2) /* normal non-cacheable */
  95. #define L_PTE_MT_WRITETHROUGH (_AT(pteval_t, 2) << 2) /* normal inner write-through */
  96. #define L_PTE_MT_WRITEBACK (_AT(pteval_t, 3) << 2) /* normal inner write-back */
  97. #define L_PTE_MT_WRITEALLOC (_AT(pteval_t, 7) << 2) /* normal inner write-alloc */
  98. #define L_PTE_MT_DEV_SHARED (_AT(pteval_t, 4) << 2) /* device */
  99. #define L_PTE_MT_DEV_NONSHARED (_AT(pteval_t, 4) << 2) /* device */
  100. #define L_PTE_MT_DEV_WC (_AT(pteval_t, 1) << 2) /* normal non-cacheable */
  101. #define L_PTE_MT_DEV_CACHED (_AT(pteval_t, 3) << 2) /* normal inner write-back */
  102. #define L_PTE_MT_MASK (_AT(pteval_t, 7) << 2)
  103. /*
  104. * Software PGD flags.
  105. */
  106. #define L_PGD_SWAPPER (_AT(pgdval_t, 1) << 55) /* swapper_pg_dir entry */
  107. /*
  108. * 2nd stage PTE definitions for LPAE.
  109. */
  110. #define L_PTE_S2_MT_UNCACHED (_AT(pteval_t, 0x0) << 2) /* strongly ordered */
  111. #define L_PTE_S2_MT_WRITETHROUGH (_AT(pteval_t, 0xa) << 2) /* normal inner write-through */
  112. #define L_PTE_S2_MT_WRITEBACK (_AT(pteval_t, 0xf) << 2) /* normal inner write-back */
  113. #define L_PTE_S2_MT_DEV_SHARED (_AT(pteval_t, 0x1) << 2) /* device */
  114. #define L_PTE_S2_MT_MASK (_AT(pteval_t, 0xf) << 2)
  115. #define L_PTE_S2_RDONLY (_AT(pteval_t, 1) << 6) /* HAP[1] */
  116. #define L_PTE_S2_RDWR (_AT(pteval_t, 3) << 6) /* HAP[2:1] */
  117. #define L_PMD_S2_RDWR (_AT(pmdval_t, 3) << 6) /* HAP[2:1] */
  118. /*
  119. * Hyp-mode PL2 PTE definitions for LPAE.
  120. */
  121. #define L_PTE_HYP L_PTE_USER
  122. #ifndef __ASSEMBLY__
  123. #define pud_none(pud) (!pud_val(pud))
  124. #define pud_bad(pud) (!(pud_val(pud) & 2))
  125. #define pud_present(pud) (pud_val(pud))
  126. #define pmd_table(pmd) ((pmd_val(pmd) & PMD_TYPE_MASK) == \
  127. PMD_TYPE_TABLE)
  128. #define pmd_sect(pmd) ((pmd_val(pmd) & PMD_TYPE_MASK) == \
  129. PMD_TYPE_SECT)
  130. #define pmd_large(pmd) pmd_sect(pmd)
  131. #define pud_clear(pudp) \
  132. do { \
  133. *pudp = __pud(0); \
  134. clean_pmd_entry(pudp); \
  135. } while (0)
  136. #define set_pud(pudp, pud) \
  137. do { \
  138. *pudp = pud; \
  139. flush_pmd_entry(pudp); \
  140. } while (0)
  141. static inline pmd_t *pud_page_vaddr(pud_t pud)
  142. {
  143. return __va(pud_val(pud) & PHYS_MASK & (s32)PAGE_MASK);
  144. }
  145. /* Find an entry in the second-level page table.. */
  146. #define pmd_index(addr) (((addr) >> PMD_SHIFT) & (PTRS_PER_PMD - 1))
  147. static inline pmd_t *pmd_offset(pud_t *pud, unsigned long addr)
  148. {
  149. return (pmd_t *)pud_page_vaddr(*pud) + pmd_index(addr);
  150. }
  151. #define pmd_bad(pmd) (!(pmd_val(pmd) & 2))
  152. #define copy_pmd(pmdpd,pmdps) \
  153. do { \
  154. *pmdpd = *pmdps; \
  155. flush_pmd_entry(pmdpd); \
  156. } while (0)
  157. #define pmd_clear(pmdp) \
  158. do { \
  159. *pmdp = __pmd(0); \
  160. clean_pmd_entry(pmdp); \
  161. } while (0)
  162. /*
  163. * For 3 levels of paging the PTE_EXT_NG bit will be set for user address ptes
  164. * that are written to a page table but not for ptes created with mk_pte.
  165. *
  166. * In hugetlb_no_page, a new huge pte (new_pte) is generated and passed to
  167. * hugetlb_cow, where it is compared with an entry in a page table.
  168. * This comparison test fails erroneously leading ultimately to a memory leak.
  169. *
  170. * To correct this behaviour, we mask off PTE_EXT_NG for any pte that is
  171. * present before running the comparison.
  172. */
  173. #define __HAVE_ARCH_PTE_SAME
  174. #define pte_same(pte_a,pte_b) ((pte_present(pte_a) ? pte_val(pte_a) & ~PTE_EXT_NG \
  175. : pte_val(pte_a)) \
  176. == (pte_present(pte_b) ? pte_val(pte_b) & ~PTE_EXT_NG \
  177. : pte_val(pte_b)))
  178. #define set_pte_ext(ptep,pte,ext) cpu_set_pte_ext(ptep,__pte(pte_val(pte)|(ext)))
  179. #define pte_huge(pte) (pte_val(pte) && !(pte_val(pte) & PTE_TABLE_BIT))
  180. #define pte_mkhuge(pte) (__pte(pte_val(pte) & ~PTE_TABLE_BIT))
  181. #define pmd_isset(pmd, val) ((u32)(val) == (val) ? pmd_val(pmd) & (val) \
  182. : !!(pmd_val(pmd) & (val)))
  183. #define pmd_isclear(pmd, val) (!(pmd_val(pmd) & (val)))
  184. #define pmd_young(pmd) (pmd_isset((pmd), PMD_SECT_AF))
  185. #define __HAVE_ARCH_PMD_WRITE
  186. #define pmd_write(pmd) (pmd_isclear((pmd), L_PMD_SECT_RDONLY))
  187. #define pmd_dirty(pmd) (pmd_isset((pmd), L_PMD_SECT_DIRTY))
  188. #define pmd_hugewillfault(pmd) (!pmd_young(pmd) || !pmd_write(pmd))
  189. #define pmd_thp_or_huge(pmd) (pmd_huge(pmd) || pmd_trans_huge(pmd))
  190. #ifdef CONFIG_TRANSPARENT_HUGEPAGE
  191. #define pmd_trans_huge(pmd) (pmd_val(pmd) && !pmd_table(pmd))
  192. #define pmd_trans_splitting(pmd) (pmd_isset((pmd), L_PMD_SECT_SPLITTING))
  193. #endif
  194. #define PMD_BIT_FUNC(fn,op) \
  195. static inline pmd_t pmd_##fn(pmd_t pmd) { pmd_val(pmd) op; return pmd; }
  196. PMD_BIT_FUNC(wrprotect, |= L_PMD_SECT_RDONLY);
  197. PMD_BIT_FUNC(mkold, &= ~PMD_SECT_AF);
  198. PMD_BIT_FUNC(mksplitting, |= L_PMD_SECT_SPLITTING);
  199. PMD_BIT_FUNC(mkwrite, &= ~L_PMD_SECT_RDONLY);
  200. PMD_BIT_FUNC(mkdirty, |= L_PMD_SECT_DIRTY);
  201. PMD_BIT_FUNC(mkyoung, |= PMD_SECT_AF);
  202. #define pmd_mkhuge(pmd) (__pmd(pmd_val(pmd) & ~PMD_TABLE_BIT))
  203. #define pmd_pfn(pmd) (((pmd_val(pmd) & PMD_MASK) & PHYS_MASK) >> PAGE_SHIFT)
  204. #define pfn_pmd(pfn,prot) (__pmd(((phys_addr_t)(pfn) << PAGE_SHIFT) | pgprot_val(prot)))
  205. #define mk_pmd(page,prot) pfn_pmd(page_to_pfn(page),prot)
  206. /* represent a notpresent pmd by zero, this is used by pmdp_invalidate */
  207. #define pmd_mknotpresent(pmd) (__pmd(0))
  208. static inline pmd_t pmd_modify(pmd_t pmd, pgprot_t newprot)
  209. {
  210. const pmdval_t mask = PMD_SECT_USER | PMD_SECT_XN | L_PMD_SECT_RDONLY |
  211. L_PMD_SECT_VALID | L_PMD_SECT_NONE;
  212. pmd_val(pmd) = (pmd_val(pmd) & ~mask) | (pgprot_val(newprot) & mask);
  213. return pmd;
  214. }
  215. static inline void set_pmd_at(struct mm_struct *mm, unsigned long addr,
  216. pmd_t *pmdp, pmd_t pmd)
  217. {
  218. BUG_ON(addr >= TASK_SIZE);
  219. /* create a faulting entry if PROT_NONE protected */
  220. if (pmd_val(pmd) & L_PMD_SECT_NONE)
  221. pmd_val(pmd) &= ~L_PMD_SECT_VALID;
  222. if (pmd_write(pmd) && pmd_dirty(pmd))
  223. pmd_val(pmd) &= ~PMD_SECT_AP2;
  224. else
  225. pmd_val(pmd) |= PMD_SECT_AP2;
  226. *pmdp = __pmd(pmd_val(pmd) | PMD_SECT_nG);
  227. flush_pmd_entry(pmdp);
  228. }
  229. static inline int has_transparent_hugepage(void)
  230. {
  231. return 1;
  232. }
  233. #endif /* __ASSEMBLY__ */
  234. #endif /* _ASM_PGTABLE_3LEVEL_H */