1 /*- 2 * Copyright (c) 2010 Isilon Systems, Inc. 3 * Copyright (c) 2010 iX Systems, Inc. 4 * Copyright (c) 2010 Panasas, Inc. 5 * Copyright (c) 2013-2017 Mellanox Technologies, Ltd. 6 * Copyright (c) 2015 François Tigeot 7 * Copyright (c) 2015 Matthew Dillon <dillon@backplane.com> 8 * All rights reserved. 9 * 10 * Redistribution and use in source and binary forms, with or without 11 * modification, are permitted provided that the following conditions 12 * are met: 13 * 1. Redistributions of source code must retain the above copyright 14 * notice unmodified, this list of conditions, and the following 15 * disclaimer. 16 * 2. Redistributions in binary form must reproduce the above copyright 17 * notice, this list of conditions and the following disclaimer in the 18 * documentation and/or other materials provided with the distribution. 19 * 20 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR 21 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES 22 * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. 23 * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, 24 * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT 25 * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, 26 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY 27 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT 28 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF 29 * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. 30 */ 31 #ifndef _LINUXKPI_LINUX_MM_H_ 32 #define _LINUXKPI_LINUX_MM_H_ 33 34 #include <linux/spinlock.h> 35 #include <linux/gfp.h> 36 #include <linux/kernel.h> 37 #include <linux/mm_types.h> 38 #include <linux/mmzone.h> 39 #include <linux/pfn.h> 40 #include <linux/list.h> 41 #include <linux/mmap_lock.h> 42 #include <linux/overflow.h> 43 #include <linux/shrinker.h> 44 #include <linux/page.h> 45 46 #include <asm/pgtable.h> 47 48 #define PAGE_ALIGN(x) ALIGN(x, PAGE_SIZE) 49 50 /* 51 * Make sure our LinuxKPI defined virtual memory flags don't conflict 52 * with the ones defined by FreeBSD: 53 */ 54 CTASSERT((VM_PROT_ALL & -(1 << 8)) == 0); 55 56 #define VM_READ VM_PROT_READ 57 #define VM_WRITE VM_PROT_WRITE 58 #define VM_EXEC VM_PROT_EXECUTE 59 60 #define VM_ACCESS_FLAGS (VM_READ | VM_WRITE | VM_EXEC) 61 62 #define VM_PFNINTERNAL (1 << 8) /* FreeBSD private flag to vm_insert_pfn() */ 63 #define VM_MIXEDMAP (1 << 9) 64 #define VM_NORESERVE (1 << 10) 65 #define VM_PFNMAP (1 << 11) 66 #define VM_IO (1 << 12) 67 #define VM_MAYWRITE (1 << 13) 68 #define VM_DONTCOPY (1 << 14) 69 #define VM_DONTEXPAND (1 << 15) 70 #define VM_DONTDUMP (1 << 16) 71 #define VM_SHARED (1 << 17) 72 73 #define VMA_MAX_PREFAULT_RECORD 1 74 75 #define FOLL_WRITE (1 << 0) 76 #define FOLL_FORCE (1 << 1) 77 78 #define VM_FAULT_OOM (1 << 0) 79 #define VM_FAULT_SIGBUS (1 << 1) 80 #define VM_FAULT_MAJOR (1 << 2) 81 #define VM_FAULT_WRITE (1 << 3) 82 #define VM_FAULT_HWPOISON (1 << 4) 83 #define VM_FAULT_HWPOISON_LARGE (1 << 5) 84 #define VM_FAULT_SIGSEGV (1 << 6) 85 #define VM_FAULT_NOPAGE (1 << 7) 86 #define VM_FAULT_LOCKED (1 << 8) 87 #define VM_FAULT_RETRY (1 << 9) 88 #define VM_FAULT_FALLBACK (1 << 10) 89 90 #define VM_FAULT_ERROR (VM_FAULT_OOM | VM_FAULT_SIGBUS | VM_FAULT_SIGSEGV | \ 91 VM_FAULT_HWPOISON |VM_FAULT_HWPOISON_LARGE | VM_FAULT_FALLBACK) 92 93 #define FAULT_FLAG_WRITE (1 << 0) 94 #define FAULT_FLAG_MKWRITE (1 << 1) 95 #define FAULT_FLAG_ALLOW_RETRY (1 << 2) 96 #define FAULT_FLAG_RETRY_NOWAIT (1 << 3) 97 #define FAULT_FLAG_KILLABLE (1 << 4) 98 #define FAULT_FLAG_TRIED (1 << 5) 99 #define FAULT_FLAG_USER (1 << 6) 100 #define FAULT_FLAG_REMOTE (1 << 7) 101 #define FAULT_FLAG_INSTRUCTION (1 << 8) 102 103 #define fault_flag_allow_retry_first(flags) \ 104 (((flags) & (FAULT_FLAG_ALLOW_RETRY | FAULT_FLAG_TRIED)) == FAULT_FLAG_ALLOW_RETRY) 105 106 typedef int (*pte_fn_t)(linux_pte_t *, unsigned long addr, void *data); 107 108 struct vm_area_struct { 109 vm_offset_t vm_start; 110 vm_offset_t vm_end; 111 vm_offset_t vm_pgoff; 112 pgprot_t vm_page_prot; 113 unsigned long vm_flags; 114 struct mm_struct *vm_mm; 115 void *vm_private_data; 116 const struct vm_operations_struct *vm_ops; 117 struct linux_file *vm_file; 118 119 /* internal operation */ 120 vm_paddr_t vm_pfn; /* PFN for memory map */ 121 vm_size_t vm_len; /* length for memory map */ 122 vm_pindex_t vm_pfn_first; 123 int vm_pfn_count; 124 int *vm_pfn_pcount; 125 vm_object_t vm_obj; 126 vm_map_t vm_cached_map; 127 TAILQ_ENTRY(vm_area_struct) vm_entry; 128 }; 129 130 struct vm_fault { 131 unsigned int flags; 132 pgoff_t pgoff; 133 union { 134 /* user-space address */ 135 void *virtual_address; /* < 4.11 */ 136 unsigned long address; /* >= 4.11 */ 137 }; 138 struct page *page; 139 struct vm_area_struct *vma; 140 }; 141 142 struct vm_operations_struct { 143 void (*open) (struct vm_area_struct *); 144 void (*close) (struct vm_area_struct *); 145 int (*fault) (struct vm_fault *); 146 int (*access) (struct vm_area_struct *, unsigned long, void *, int, int); 147 }; 148 149 struct sysinfo { 150 uint64_t totalram; /* Total usable main memory size */ 151 uint64_t freeram; /* Available memory size */ 152 uint64_t totalhigh; /* Total high memory size */ 153 uint64_t freehigh; /* Available high memory size */ 154 uint32_t mem_unit; /* Memory unit size in bytes */ 155 }; 156 157 static inline struct page * 158 virt_to_head_page(const void *p) 159 { 160 161 return (virt_to_page(p)); 162 } 163 164 static inline struct folio * 165 virt_to_folio(const void *p) 166 { 167 struct page *page = virt_to_page(p); 168 169 return (page_folio(page)); 170 } 171 172 /* 173 * Compute log2 of the power of two rounded up count of pages 174 * needed for size bytes. 175 */ 176 static inline int 177 get_order(unsigned long size) 178 { 179 int order; 180 181 size = (size - 1) >> PAGE_SHIFT; 182 order = 0; 183 while (size) { 184 order++; 185 size >>= 1; 186 } 187 return (order); 188 } 189 190 /* 191 * Resolve a page into a virtual address: 192 * 193 * NOTE: This function only works for pages allocated by the kernel. 194 */ 195 void *linux_page_address(const struct page *); 196 #define page_address(page) linux_page_address(page) 197 198 static inline void * 199 lowmem_page_address(struct page *page) 200 { 201 return (page_address(page)); 202 } 203 204 /* 205 * This only works via memory map operations. 206 */ 207 static inline int 208 io_remap_pfn_range(struct vm_area_struct *vma, 209 unsigned long addr, unsigned long pfn, unsigned long size, 210 vm_memattr_t prot) 211 { 212 vma->vm_page_prot = prot; 213 vma->vm_pfn = pfn; 214 vma->vm_len = size; 215 216 return (0); 217 } 218 219 vm_fault_t 220 lkpi_vmf_insert_pfn_prot_locked(struct vm_area_struct *vma, unsigned long addr, 221 unsigned long pfn, pgprot_t prot); 222 223 static inline vm_fault_t 224 vmf_insert_pfn_prot(struct vm_area_struct *vma, unsigned long addr, 225 unsigned long pfn, pgprot_t prot) 226 { 227 vm_fault_t ret; 228 229 VM_OBJECT_WLOCK(vma->vm_obj); 230 ret = lkpi_vmf_insert_pfn_prot_locked(vma, addr, pfn, prot); 231 VM_OBJECT_WUNLOCK(vma->vm_obj); 232 233 return (ret); 234 } 235 #define vmf_insert_pfn_prot(...) \ 236 _Static_assert(false, \ 237 "This function is always called in a loop. Consider using the locked version") 238 239 static inline int 240 apply_to_page_range(struct mm_struct *mm, unsigned long address, 241 unsigned long size, pte_fn_t fn, void *data) 242 { 243 return (-ENOTSUP); 244 } 245 246 int zap_vma_ptes(struct vm_area_struct *vma, unsigned long address, 247 unsigned long size); 248 249 int lkpi_remap_pfn_range(struct vm_area_struct *vma, 250 unsigned long start_addr, unsigned long start_pfn, unsigned long size, 251 pgprot_t prot); 252 253 static inline int 254 remap_pfn_range(struct vm_area_struct *vma, unsigned long addr, 255 unsigned long pfn, unsigned long size, pgprot_t prot) 256 { 257 return (lkpi_remap_pfn_range(vma, addr, pfn, size, prot)); 258 } 259 260 static inline unsigned long 261 vma_pages(struct vm_area_struct *vma) 262 { 263 return ((vma->vm_end - vma->vm_start) >> PAGE_SHIFT); 264 } 265 266 #define offset_in_page(off) ((unsigned long)(off) & (PAGE_SIZE - 1)) 267 268 static inline void 269 set_page_dirty(struct page *page) 270 { 271 vm_page_dirty(page); 272 } 273 274 static inline void 275 mark_page_accessed(struct page *page) 276 { 277 vm_page_reference(page); 278 } 279 280 static inline void 281 get_page(struct page *page) 282 { 283 vm_page_wire(page); 284 } 285 286 static inline void 287 put_page(struct page *page) 288 { 289 /* `__free_page()` takes care of the refcounting (unwire). */ 290 __free_page(page); 291 } 292 293 static inline void 294 folio_get(struct folio *folio) 295 { 296 get_page(&folio->page); 297 } 298 299 static inline void 300 folio_put(struct folio *folio) 301 { 302 put_page(&folio->page); 303 } 304 305 /* 306 * Linux uses the following "transparent" union so that `release_pages()` 307 * accepts both a list of `struct page` or a list of `struct folio`. This 308 * relies on the fact that a `struct folio` can be cast to a `struct page`. 309 */ 310 typedef union { 311 struct page **pages; 312 struct folio **folios; 313 } release_pages_arg __attribute__ ((__transparent_union__)); 314 315 void linux_release_pages(release_pages_arg arg, int nr); 316 #define release_pages(arg, nr) linux_release_pages((arg), (nr)) 317 318 extern long 319 lkpi_get_user_pages(unsigned long start, unsigned long nr_pages, 320 unsigned int gup_flags, struct page **); 321 #if defined(LINUXKPI_VERSION) && LINUXKPI_VERSION >= 60500 322 #define get_user_pages(start, nr_pages, gup_flags, pages) \ 323 lkpi_get_user_pages(start, nr_pages, gup_flags, pages) 324 #else 325 #define get_user_pages(start, nr_pages, gup_flags, pages, vmas) \ 326 lkpi_get_user_pages(start, nr_pages, gup_flags, pages) 327 #endif 328 329 #if defined(LINUXKPI_VERSION) && LINUXKPI_VERSION >= 60500 330 static inline long 331 pin_user_pages(unsigned long start, unsigned long nr_pages, 332 unsigned int gup_flags, struct page **pages) 333 { 334 return (get_user_pages(start, nr_pages, gup_flags, pages)); 335 } 336 #else 337 static inline long 338 pin_user_pages(unsigned long start, unsigned long nr_pages, 339 unsigned int gup_flags, struct page **pages, 340 struct vm_area_struct **vmas) 341 { 342 return (get_user_pages(start, nr_pages, gup_flags, pages, vmas)); 343 } 344 #endif 345 346 extern int 347 __get_user_pages_fast(unsigned long start, int nr_pages, int write, 348 struct page **); 349 350 static inline int 351 pin_user_pages_fast(unsigned long start, int nr_pages, 352 unsigned int gup_flags, struct page **pages) 353 { 354 return __get_user_pages_fast( 355 start, nr_pages, !!(gup_flags & FOLL_WRITE), pages); 356 } 357 358 extern long 359 get_user_pages_remote(struct task_struct *, struct mm_struct *, 360 unsigned long start, unsigned long nr_pages, 361 unsigned int gup_flags, struct page **, 362 struct vm_area_struct **); 363 364 static inline long 365 pin_user_pages_remote(struct task_struct *task, struct mm_struct *mm, 366 unsigned long start, unsigned long nr_pages, 367 unsigned int gup_flags, struct page **pages, 368 struct vm_area_struct **vmas) 369 { 370 return get_user_pages_remote( 371 task, mm, start, nr_pages, gup_flags, pages, vmas); 372 } 373 374 #define unpin_user_page(page) put_page(page) 375 #define unpin_user_pages(pages, npages) release_pages(pages, npages) 376 377 #define copy_highpage(to, from) pmap_copy_page(from, to) 378 379 static inline pgprot_t 380 vm_get_page_prot(unsigned long vm_flags) 381 { 382 return (vm_flags & VM_PROT_ALL); 383 } 384 385 static inline void 386 vm_flags_set(struct vm_area_struct *vma, unsigned long flags) 387 { 388 vma->vm_flags |= flags; 389 } 390 391 static inline void 392 vm_flags_clear(struct vm_area_struct *vma, unsigned long flags) 393 { 394 vma->vm_flags &= ~flags; 395 } 396 397 static inline struct page * 398 vmalloc_to_page(const void *addr) 399 { 400 vm_paddr_t paddr; 401 402 paddr = pmap_kextract((vm_offset_t)addr); 403 return (PHYS_TO_VM_PAGE(paddr)); 404 } 405 406 static inline int 407 trylock_page(struct page *page) 408 { 409 return (vm_page_tryxbusy(page)); 410 } 411 412 static inline void 413 unlock_page(struct page *page) 414 { 415 416 vm_page_xunbusy(page); 417 } 418 419 extern int is_vmalloc_addr(const void *addr); 420 void si_meminfo(struct sysinfo *si); 421 422 static inline unsigned long 423 totalram_pages(void) 424 { 425 return ((unsigned long)physmem); 426 } 427 428 #define unmap_mapping_range(...) lkpi_unmap_mapping_range(__VA_ARGS__) 429 void lkpi_unmap_mapping_range(void *obj, loff_t const holebegin __unused, 430 loff_t const holelen, int even_cows __unused); 431 432 #define PAGE_ALIGNED(p) __is_aligned(p, PAGE_SIZE) 433 434 void vma_set_file(struct vm_area_struct *vma, struct linux_file *file); 435 436 static inline void 437 might_alloc(gfp_t gfp_mask __unused) 438 { 439 } 440 441 #define is_cow_mapping(flags) (false) 442 443 static inline bool 444 want_init_on_free(void) 445 { 446 return (false); 447 } 448 449 static inline unsigned long 450 folio_pfn(struct folio *folio) 451 { 452 return (page_to_pfn(&folio->page)); 453 } 454 455 static inline long 456 folio_nr_pages(struct folio *folio) 457 { 458 return (1); 459 } 460 461 static inline size_t 462 folio_size(struct folio *folio) 463 { 464 return (PAGE_SIZE); 465 } 466 467 static inline void 468 folio_mark_dirty(struct folio *folio) 469 { 470 set_page_dirty(&folio->page); 471 } 472 473 static inline void * 474 folio_address(const struct folio *folio) 475 { 476 return (page_address(&folio->page)); 477 } 478 479 #endif /* _LINUXKPI_LINUX_MM_H_ */ 480