Branch data Line data Source code
1 : : /* SPDX-License-Identifier: GPL-2.0 */
2 : : #ifndef _LINUX_MM_H
3 : : #define _LINUX_MM_H
4 : :
5 : : #include <linux/errno.h>
6 : :
7 : : #ifdef __KERNEL__
8 : :
9 : : #include <linux/mmdebug.h>
10 : : #include <linux/gfp.h>
11 : : #include <linux/bug.h>
12 : : #include <linux/list.h>
13 : : #include <linux/mmzone.h>
14 : : #include <linux/rbtree.h>
15 : : #include <linux/atomic.h>
16 : : #include <linux/debug_locks.h>
17 : : #include <linux/mm_types.h>
18 : : #include <linux/range.h>
19 : : #include <linux/pfn.h>
20 : : #include <linux/percpu-refcount.h>
21 : : #include <linux/bit_spinlock.h>
22 : : #include <linux/shrinker.h>
23 : : #include <linux/resource.h>
24 : : #include <linux/page_ext.h>
25 : : #include <linux/err.h>
26 : : #include <linux/page_ref.h>
27 : : #include <linux/memremap.h>
28 : : #include <linux/overflow.h>
29 : : #include <linux/sizes.h>
30 : :
31 : : struct mempolicy;
32 : : struct anon_vma;
33 : : struct anon_vma_chain;
34 : : struct file_ra_state;
35 : : struct user_struct;
36 : : struct writeback_control;
37 : : struct bdi_writeback;
38 : :
39 : : void init_mm_internals(void);
40 : :
41 : : #ifndef CONFIG_NEED_MULTIPLE_NODES /* Don't use mapnrs, do it properly */
42 : : extern unsigned long max_mapnr;
43 : :
44 : : static inline void set_max_mapnr(unsigned long limit)
45 : : {
46 : : max_mapnr = limit;
47 : : }
48 : : #else
49 : : static inline void set_max_mapnr(unsigned long limit) { }
50 : : #endif
51 : :
52 : : extern atomic_long_t _totalram_pages;
53 : 578328 : static inline unsigned long totalram_pages(void)
54 : : {
55 [ + - ]: 578328 : return (unsigned long)atomic_long_read(&_totalram_pages);
56 : : }
57 : :
58 : 0 : static inline void totalram_pages_inc(void)
59 : : {
60 : 0 : atomic_long_inc(&_totalram_pages);
61 : : }
62 : :
63 : : static inline void totalram_pages_dec(void)
64 : : {
65 : : atomic_long_dec(&_totalram_pages);
66 : : }
67 : :
68 : 18382 : static inline void totalram_pages_add(long count)
69 : : {
70 : 18382 : atomic_long_add(count, &_totalram_pages);
71 : 18382 : }
72 : :
73 : : extern void * high_memory;
74 : : extern int page_cluster;
75 : :
76 : : #ifdef CONFIG_SYSCTL
77 : : extern int sysctl_legacy_va_layout;
78 : : #else
79 : : #define sysctl_legacy_va_layout 0
80 : : #endif
81 : :
82 : : #ifdef CONFIG_HAVE_ARCH_MMAP_RND_BITS
83 : : extern const int mmap_rnd_bits_min;
84 : : extern const int mmap_rnd_bits_max;
85 : : extern int mmap_rnd_bits __read_mostly;
86 : : #endif
87 : : #ifdef CONFIG_HAVE_ARCH_MMAP_RND_COMPAT_BITS
88 : : extern const int mmap_rnd_compat_bits_min;
89 : : extern const int mmap_rnd_compat_bits_max;
90 : : extern int mmap_rnd_compat_bits __read_mostly;
91 : : #endif
92 : :
93 : : #include <asm/page.h>
94 : : #include <asm/pgtable.h>
95 : : #include <asm/processor.h>
96 : :
97 : : /*
98 : : * Architectures that support memory tagging (assigning tags to memory regions,
99 : : * embedding these tags into addresses that point to these memory regions, and
100 : : * checking that the memory and the pointer tags match on memory accesses)
101 : : * redefine this macro to strip tags from pointers.
102 : : * It's defined as noop for arcitectures that don't support memory tagging.
103 : : */
104 : : #ifndef untagged_addr
105 : : #define untagged_addr(addr) (addr)
106 : : #endif
107 : :
108 : : #ifndef __pa_symbol
109 : : #define __pa_symbol(x) __pa(RELOC_HIDE((unsigned long)(x), 0))
110 : : #endif
111 : :
112 : : #ifndef page_to_virt
113 : : #define page_to_virt(x) __va(PFN_PHYS(page_to_pfn(x)))
114 : : #endif
115 : :
116 : : #ifndef lm_alias
117 : : #define lm_alias(x) __va(__pa_symbol(x))
118 : : #endif
119 : :
120 : : /*
121 : : * To prevent common memory management code establishing
122 : : * a zero page mapping on a read fault.
123 : : * This macro should be defined within <asm/pgtable.h>.
124 : : * s390 does this to prevent multiplexing of hardware bits
125 : : * related to the physical page in case of virtualization.
126 : : */
127 : : #ifndef mm_forbids_zeropage
128 : : #define mm_forbids_zeropage(X) (0)
129 : : #endif
130 : :
131 : : /*
132 : : * On some architectures it is expensive to call memset() for small sizes.
133 : : * If an architecture decides to implement their own version of
134 : : * mm_zero_struct_page they should wrap the defines below in a #ifndef and
135 : : * define their own version of this macro in <asm/pgtable.h>
136 : : */
137 : : #if BITS_PER_LONG == 64
138 : : /* This function must be updated when the size of struct page grows above 80
139 : : * or reduces below 56. The idea that compiler optimizes out switch()
140 : : * statement, and only leaves move/store instructions. Also the compiler can
141 : : * combine write statments if they are both assignments and can be reordered,
142 : : * this can result in several of the writes here being dropped.
143 : : */
144 : : #define mm_zero_struct_page(pp) __mm_zero_struct_page(pp)
145 : : static inline void __mm_zero_struct_page(struct page *page)
146 : : {
147 : : unsigned long *_pp = (void *)page;
148 : :
149 : : /* Check that struct page is either 56, 64, 72, or 80 bytes */
150 : : BUILD_BUG_ON(sizeof(struct page) & 7);
151 : : BUILD_BUG_ON(sizeof(struct page) < 56);
152 : : BUILD_BUG_ON(sizeof(struct page) > 80);
153 : :
154 : : switch (sizeof(struct page)) {
155 : : case 80:
156 : : _pp[9] = 0; /* fallthrough */
157 : : case 72:
158 : : _pp[8] = 0; /* fallthrough */
159 : : case 64:
160 : : _pp[7] = 0; /* fallthrough */
161 : : case 56:
162 : : _pp[6] = 0;
163 : : _pp[5] = 0;
164 : : _pp[4] = 0;
165 : : _pp[3] = 0;
166 : : _pp[2] = 0;
167 : : _pp[1] = 0;
168 : : _pp[0] = 0;
169 : : }
170 : : }
171 : : #else
172 : : #define mm_zero_struct_page(pp) ((void)memset((pp), 0, sizeof(struct page)))
173 : : #endif
174 : :
175 : : /*
176 : : * Default maximum number of active map areas, this limits the number of vmas
177 : : * per mm struct. Users can overwrite this number by sysctl but there is a
178 : : * problem.
179 : : *
180 : : * When a program's coredump is generated as ELF format, a section is created
181 : : * per a vma. In ELF, the number of sections is represented in unsigned short.
182 : : * This means the number of sections should be smaller than 65535 at coredump.
183 : : * Because the kernel adds some informative sections to a image of program at
184 : : * generating coredump, we need some margin. The number of extra sections is
185 : : * 1-3 now and depends on arch. We use "5" as safe margin, here.
186 : : *
187 : : * ELF extended numbering allows more than 65535 sections, so 16-bit bound is
188 : : * not a hard limit any more. Although some userspace tools can be surprised by
189 : : * that.
190 : : */
191 : : #define MAPCOUNT_ELF_CORE_MARGIN (5)
192 : : #define DEFAULT_MAX_MAP_COUNT (USHRT_MAX - MAPCOUNT_ELF_CORE_MARGIN)
193 : :
194 : : extern int sysctl_max_map_count;
195 : :
196 : : extern unsigned long sysctl_user_reserve_kbytes;
197 : : extern unsigned long sysctl_admin_reserve_kbytes;
198 : :
199 : : extern int sysctl_overcommit_memory;
200 : : extern int sysctl_overcommit_ratio;
201 : : extern unsigned long sysctl_overcommit_kbytes;
202 : :
203 : : extern int overcommit_ratio_handler(struct ctl_table *, int, void __user *,
204 : : size_t *, loff_t *);
205 : : extern int overcommit_kbytes_handler(struct ctl_table *, int, void __user *,
206 : : size_t *, loff_t *);
207 : :
208 : : #define nth_page(page,n) pfn_to_page(page_to_pfn((page)) + (n))
209 : :
210 : : /* to align the pointer to the (next) page boundary */
211 : : #define PAGE_ALIGN(addr) ALIGN(addr, PAGE_SIZE)
212 : :
213 : : /* test whether an address (unsigned long or pointer) is aligned to PAGE_SIZE */
214 : : #define PAGE_ALIGNED(addr) IS_ALIGNED((unsigned long)(addr), PAGE_SIZE)
215 : :
216 : : #define lru_to_page(head) (list_entry((head)->prev, struct page, lru))
217 : :
218 : : /*
219 : : * Linux kernel virtual memory manager primitives.
220 : : * The idea being to have a "virtual" mm in the same way
221 : : * we have a virtual fs - giving a cleaner interface to the
222 : : * mm details, and allowing different kinds of memory mappings
223 : : * (from shared memory to executable loading to arbitrary
224 : : * mmap() functions).
225 : : */
226 : :
227 : : struct vm_area_struct *vm_area_alloc(struct mm_struct *);
228 : : struct vm_area_struct *vm_area_dup(struct vm_area_struct *);
229 : : void vm_area_free(struct vm_area_struct *);
230 : :
231 : : #ifndef CONFIG_MMU
232 : : extern struct rb_root nommu_region_tree;
233 : : extern struct rw_semaphore nommu_region_sem;
234 : :
235 : : extern unsigned int kobjsize(const void *objp);
236 : : #endif
237 : :
238 : : /*
239 : : * vm_flags in vm_area_struct, see mm_types.h.
240 : : * When changing, update also include/trace/events/mmflags.h
241 : : */
242 : : #define VM_NONE 0x00000000
243 : :
244 : : #define VM_READ 0x00000001 /* currently active flags */
245 : : #define VM_WRITE 0x00000002
246 : : #define VM_EXEC 0x00000004
247 : : #define VM_SHARED 0x00000008
248 : :
249 : : /* mprotect() hardcodes VM_MAYREAD >> 4 == VM_READ, and so for r/w/x bits. */
250 : : #define VM_MAYREAD 0x00000010 /* limits for mprotect() etc */
251 : : #define VM_MAYWRITE 0x00000020
252 : : #define VM_MAYEXEC 0x00000040
253 : : #define VM_MAYSHARE 0x00000080
254 : :
255 : : #define VM_GROWSDOWN 0x00000100 /* general info on the segment */
256 : : #define VM_UFFD_MISSING 0x00000200 /* missing pages tracking */
257 : : #define VM_PFNMAP 0x00000400 /* Page-ranges managed without "struct page", just pure PFN */
258 : : #define VM_DENYWRITE 0x00000800 /* ETXTBSY on write attempts.. */
259 : : #define VM_UFFD_WP 0x00001000 /* wrprotect pages tracking */
260 : :
261 : : #define VM_LOCKED 0x00002000
262 : : #define VM_IO 0x00004000 /* Memory mapped I/O or similar */
263 : :
264 : : /* Used by sys_madvise() */
265 : : #define VM_SEQ_READ 0x00008000 /* App will access data sequentially */
266 : : #define VM_RAND_READ 0x00010000 /* App will not benefit from clustered reads */
267 : :
268 : : #define VM_DONTCOPY 0x00020000 /* Do not copy this vma on fork */
269 : : #define VM_DONTEXPAND 0x00040000 /* Cannot expand with mremap() */
270 : : #define VM_LOCKONFAULT 0x00080000 /* Lock the pages covered when they are faulted in */
271 : : #define VM_ACCOUNT 0x00100000 /* Is a VM accounted object */
272 : : #define VM_NORESERVE 0x00200000 /* should the VM suppress accounting */
273 : : #define VM_HUGETLB 0x00400000 /* Huge TLB Page VM */
274 : : #define VM_SYNC 0x00800000 /* Synchronous page faults */
275 : : #define VM_ARCH_1 0x01000000 /* Architecture-specific flag */
276 : : #define VM_WIPEONFORK 0x02000000 /* Wipe VMA contents in child. */
277 : : #define VM_DONTDUMP 0x04000000 /* Do not include in the core dump */
278 : :
279 : : #ifdef CONFIG_MEM_SOFT_DIRTY
280 : : # define VM_SOFTDIRTY 0x08000000 /* Not soft dirty clean area */
281 : : #else
282 : : # define VM_SOFTDIRTY 0
283 : : #endif
284 : :
285 : : #define VM_MIXEDMAP 0x10000000 /* Can contain "struct page" and pure PFN pages */
286 : : #define VM_HUGEPAGE 0x20000000 /* MADV_HUGEPAGE marked this vma */
287 : : #define VM_NOHUGEPAGE 0x40000000 /* MADV_NOHUGEPAGE marked this vma */
288 : : #define VM_MERGEABLE 0x80000000 /* KSM may merge identical pages */
289 : :
290 : : #ifdef CONFIG_ARCH_USES_HIGH_VMA_FLAGS
291 : : #define VM_HIGH_ARCH_BIT_0 32 /* bit only usable on 64-bit architectures */
292 : : #define VM_HIGH_ARCH_BIT_1 33 /* bit only usable on 64-bit architectures */
293 : : #define VM_HIGH_ARCH_BIT_2 34 /* bit only usable on 64-bit architectures */
294 : : #define VM_HIGH_ARCH_BIT_3 35 /* bit only usable on 64-bit architectures */
295 : : #define VM_HIGH_ARCH_BIT_4 36 /* bit only usable on 64-bit architectures */
296 : : #define VM_HIGH_ARCH_0 BIT(VM_HIGH_ARCH_BIT_0)
297 : : #define VM_HIGH_ARCH_1 BIT(VM_HIGH_ARCH_BIT_1)
298 : : #define VM_HIGH_ARCH_2 BIT(VM_HIGH_ARCH_BIT_2)
299 : : #define VM_HIGH_ARCH_3 BIT(VM_HIGH_ARCH_BIT_3)
300 : : #define VM_HIGH_ARCH_4 BIT(VM_HIGH_ARCH_BIT_4)
301 : : #endif /* CONFIG_ARCH_USES_HIGH_VMA_FLAGS */
302 : :
303 : : #ifdef CONFIG_ARCH_HAS_PKEYS
304 : : # define VM_PKEY_SHIFT VM_HIGH_ARCH_BIT_0
305 : : # define VM_PKEY_BIT0 VM_HIGH_ARCH_0 /* A protection key is a 4-bit value */
306 : : # define VM_PKEY_BIT1 VM_HIGH_ARCH_1 /* on x86 and 5-bit value on ppc64 */
307 : : # define VM_PKEY_BIT2 VM_HIGH_ARCH_2
308 : : # define VM_PKEY_BIT3 VM_HIGH_ARCH_3
309 : : #ifdef CONFIG_PPC
310 : : # define VM_PKEY_BIT4 VM_HIGH_ARCH_4
311 : : #else
312 : : # define VM_PKEY_BIT4 0
313 : : #endif
314 : : #endif /* CONFIG_ARCH_HAS_PKEYS */
315 : :
316 : : #if defined(CONFIG_X86)
317 : : # define VM_PAT VM_ARCH_1 /* PAT reserves whole VMA at once (x86) */
318 : : #elif defined(CONFIG_PPC)
319 : : # define VM_SAO VM_ARCH_1 /* Strong Access Ordering (powerpc) */
320 : : #elif defined(CONFIG_PARISC)
321 : : # define VM_GROWSUP VM_ARCH_1
322 : : #elif defined(CONFIG_IA64)
323 : : # define VM_GROWSUP VM_ARCH_1
324 : : #elif defined(CONFIG_SPARC64)
325 : : # define VM_SPARC_ADI VM_ARCH_1 /* Uses ADI tag for access control */
326 : : # define VM_ARCH_CLEAR VM_SPARC_ADI
327 : : #elif !defined(CONFIG_MMU)
328 : : # define VM_MAPPED_COPY VM_ARCH_1 /* T if mapped copy of data (nommu mmap) */
329 : : #endif
330 : :
331 : : #if defined(CONFIG_X86_INTEL_MPX)
332 : : /* MPX specific bounds table or bounds directory */
333 : : # define VM_MPX VM_HIGH_ARCH_4
334 : : #else
335 : : # define VM_MPX VM_NONE
336 : : #endif
337 : :
338 : : #ifndef VM_GROWSUP
339 : : # define VM_GROWSUP VM_NONE
340 : : #endif
341 : :
342 : : /* Bits set in the VMA until the stack is in its final location */
343 : : #define VM_STACK_INCOMPLETE_SETUP (VM_RAND_READ | VM_SEQ_READ)
344 : :
345 : : #ifndef VM_STACK_DEFAULT_FLAGS /* arch can override this */
346 : : #define VM_STACK_DEFAULT_FLAGS VM_DATA_DEFAULT_FLAGS
347 : : #endif
348 : :
349 : : #ifdef CONFIG_STACK_GROWSUP
350 : : #define VM_STACK VM_GROWSUP
351 : : #else
352 : : #define VM_STACK VM_GROWSDOWN
353 : : #endif
354 : :
355 : : #define VM_STACK_FLAGS (VM_STACK | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT)
356 : :
357 : : /*
358 : : * Special vmas that are non-mergable, non-mlock()able.
359 : : * Note: mm/huge_memory.c VM_NO_THP depends on this definition.
360 : : */
361 : : #define VM_SPECIAL (VM_IO | VM_DONTEXPAND | VM_PFNMAP | VM_MIXEDMAP)
362 : :
363 : : /* This mask defines which mm->def_flags a process can inherit its parent */
364 : : #define VM_INIT_DEF_MASK VM_NOHUGEPAGE
365 : :
366 : : /* This mask is used to clear all the VMA flags used by mlock */
367 : : #define VM_LOCKED_CLEAR_MASK (~(VM_LOCKED | VM_LOCKONFAULT))
368 : :
369 : : /* Arch-specific flags to clear when updating VM flags on protection change */
370 : : #ifndef VM_ARCH_CLEAR
371 : : # define VM_ARCH_CLEAR VM_NONE
372 : : #endif
373 : : #define VM_FLAGS_CLEAR (ARCH_VM_PKEY_FLAGS | VM_ARCH_CLEAR)
374 : :
375 : : /*
376 : : * mapping from the currently active vm_flags protection bits (the
377 : : * low four bits) to a page protection mask..
378 : : */
379 : : extern pgprot_t protection_map[16];
380 : :
381 : : #define FAULT_FLAG_WRITE 0x01 /* Fault was a write access */
382 : : #define FAULT_FLAG_MKWRITE 0x02 /* Fault was mkwrite of existing pte */
383 : : #define FAULT_FLAG_ALLOW_RETRY 0x04 /* Retry fault if blocking */
384 : : #define FAULT_FLAG_RETRY_NOWAIT 0x08 /* Don't drop mmap_sem and wait when retrying */
385 : : #define FAULT_FLAG_KILLABLE 0x10 /* The fault task is in SIGKILL killable region */
386 : : #define FAULT_FLAG_TRIED 0x20 /* Second try */
387 : : #define FAULT_FLAG_USER 0x40 /* The fault originated in userspace */
388 : : #define FAULT_FLAG_REMOTE 0x80 /* faulting for non current tsk/mm */
389 : : #define FAULT_FLAG_INSTRUCTION 0x100 /* The fault was during an instruction fetch */
390 : :
391 : : #define FAULT_FLAG_TRACE \
392 : : { FAULT_FLAG_WRITE, "WRITE" }, \
393 : : { FAULT_FLAG_MKWRITE, "MKWRITE" }, \
394 : : { FAULT_FLAG_ALLOW_RETRY, "ALLOW_RETRY" }, \
395 : : { FAULT_FLAG_RETRY_NOWAIT, "RETRY_NOWAIT" }, \
396 : : { FAULT_FLAG_KILLABLE, "KILLABLE" }, \
397 : : { FAULT_FLAG_TRIED, "TRIED" }, \
398 : : { FAULT_FLAG_USER, "USER" }, \
399 : : { FAULT_FLAG_REMOTE, "REMOTE" }, \
400 : : { FAULT_FLAG_INSTRUCTION, "INSTRUCTION" }
401 : :
402 : : /*
403 : : * vm_fault is filled by the the pagefault handler and passed to the vma's
404 : : * ->fault function. The vma's ->fault is responsible for returning a bitmask
405 : : * of VM_FAULT_xxx flags that give details about how the fault was handled.
406 : : *
407 : : * MM layer fills up gfp_mask for page allocations but fault handler might
408 : : * alter it if its implementation requires a different allocation context.
409 : : *
410 : : * pgoff should be used in favour of virtual_address, if possible.
411 : : */
412 : : struct vm_fault {
413 : : struct vm_area_struct *vma; /* Target VMA */
414 : : unsigned int flags; /* FAULT_FLAG_xxx flags */
415 : : gfp_t gfp_mask; /* gfp mask to be used for allocations */
416 : : pgoff_t pgoff; /* Logical page offset based on vma */
417 : : unsigned long address; /* Faulting virtual address */
418 : : pmd_t *pmd; /* Pointer to pmd entry matching
419 : : * the 'address' */
420 : : pud_t *pud; /* Pointer to pud entry matching
421 : : * the 'address'
422 : : */
423 : : pte_t orig_pte; /* Value of PTE at the time of fault */
424 : :
425 : : struct page *cow_page; /* Page handler may use for COW fault */
426 : : struct mem_cgroup *memcg; /* Cgroup cow_page belongs to */
427 : : struct page *page; /* ->fault handlers should return a
428 : : * page here, unless VM_FAULT_NOPAGE
429 : : * is set (which is also implied by
430 : : * VM_FAULT_ERROR).
431 : : */
432 : : /* These three entries are valid only while holding ptl lock */
433 : : pte_t *pte; /* Pointer to pte entry matching
434 : : * the 'address'. NULL if the page
435 : : * table hasn't been allocated.
436 : : */
437 : : spinlock_t *ptl; /* Page table lock.
438 : : * Protects pte page table if 'pte'
439 : : * is not NULL, otherwise pmd.
440 : : */
441 : : pgtable_t prealloc_pte; /* Pre-allocated pte page table.
442 : : * vm_ops->map_pages() calls
443 : : * alloc_set_pte() from atomic context.
444 : : * do_fault_around() pre-allocates
445 : : * page table to avoid allocation from
446 : : * atomic context.
447 : : */
448 : : };
449 : :
450 : : /* page entry size for vm->huge_fault() */
451 : : enum page_entry_size {
452 : : PE_SIZE_PTE = 0,
453 : : PE_SIZE_PMD,
454 : : PE_SIZE_PUD,
455 : : };
456 : :
457 : : /*
458 : : * These are the virtual MM functions - opening of an area, closing and
459 : : * unmapping it (needed to keep files on disk up-to-date etc), pointer
460 : : * to the functions called when a no-page or a wp-page exception occurs.
461 : : */
462 : : struct vm_operations_struct {
463 : : void (*open)(struct vm_area_struct * area);
464 : : void (*close)(struct vm_area_struct * area);
465 : : int (*split)(struct vm_area_struct * area, unsigned long addr);
466 : : int (*mremap)(struct vm_area_struct * area);
467 : : vm_fault_t (*fault)(struct vm_fault *vmf);
468 : : vm_fault_t (*huge_fault)(struct vm_fault *vmf,
469 : : enum page_entry_size pe_size);
470 : : void (*map_pages)(struct vm_fault *vmf,
471 : : pgoff_t start_pgoff, pgoff_t end_pgoff);
472 : : unsigned long (*pagesize)(struct vm_area_struct * area);
473 : :
474 : : /* notification that a previously read-only page is about to become
475 : : * writable, if an error is returned it will cause a SIGBUS */
476 : : vm_fault_t (*page_mkwrite)(struct vm_fault *vmf);
477 : :
478 : : /* same as page_mkwrite when using VM_PFNMAP|VM_MIXEDMAP */
479 : : vm_fault_t (*pfn_mkwrite)(struct vm_fault *vmf);
480 : :
481 : : /* called by access_process_vm when get_user_pages() fails, typically
482 : : * for use by special VMAs that can switch between memory and hardware
483 : : */
484 : : int (*access)(struct vm_area_struct *vma, unsigned long addr,
485 : : void *buf, int len, int write);
486 : :
487 : : /* Called by the /proc/PID/maps code to ask the vma whether it
488 : : * has a special name. Returning non-NULL will also cause this
489 : : * vma to be dumped unconditionally. */
490 : : const char *(*name)(struct vm_area_struct *vma);
491 : :
492 : : #ifdef CONFIG_NUMA
493 : : /*
494 : : * set_policy() op must add a reference to any non-NULL @new mempolicy
495 : : * to hold the policy upon return. Caller should pass NULL @new to
496 : : * remove a policy and fall back to surrounding context--i.e. do not
497 : : * install a MPOL_DEFAULT policy, nor the task or system default
498 : : * mempolicy.
499 : : */
500 : : int (*set_policy)(struct vm_area_struct *vma, struct mempolicy *new);
501 : :
502 : : /*
503 : : * get_policy() op must add reference [mpol_get()] to any policy at
504 : : * (vma,addr) marked as MPOL_SHARED. The shared policy infrastructure
505 : : * in mm/mempolicy.c will do this automatically.
506 : : * get_policy() must NOT add a ref if the policy at (vma,addr) is not
507 : : * marked as MPOL_SHARED. vma policies are protected by the mmap_sem.
508 : : * If no [shared/vma] mempolicy exists at the addr, get_policy() op
509 : : * must return NULL--i.e., do not "fallback" to task or system default
510 : : * policy.
511 : : */
512 : : struct mempolicy *(*get_policy)(struct vm_area_struct *vma,
513 : : unsigned long addr);
514 : : #endif
515 : : /*
516 : : * Called by vm_normal_page() for special PTEs to find the
517 : : * page for @addr. This is useful if the default behavior
518 : : * (using pte_page()) would not find the correct page.
519 : : */
520 : : struct page *(*find_special_page)(struct vm_area_struct *vma,
521 : : unsigned long addr);
522 : : };
523 : :
524 : 245348 : static inline void vma_init(struct vm_area_struct *vma, struct mm_struct *mm)
525 : : {
526 : 245348 : static const struct vm_operations_struct dummy_vm_ops = {};
527 : :
528 : 245348 : memset(vma, 0, sizeof(*vma));
529 : 245348 : vma->vm_mm = mm;
530 : 245348 : vma->vm_ops = &dummy_vm_ops;
531 : 245348 : INIT_LIST_HEAD(&vma->anon_vma_chain);
532 : 228253 : }
533 : :
534 : 66481 : static inline void vma_set_anonymous(struct vm_area_struct *vma)
535 : : {
536 [ + - ]: 36722 : vma->vm_ops = NULL;
537 : 29759 : }
538 : :
539 : 1013844 : static inline bool vma_is_anonymous(struct vm_area_struct *vma)
540 : : {
541 [ + + + + ]: 1013844 : return !vma->vm_ops;
542 : : }
543 : :
544 : : #ifdef CONFIG_SHMEM
545 : : /*
546 : : * The vma_is_shmem is not inline because it is used only by slow
547 : : * paths in userfault.
548 : : */
549 : : bool vma_is_shmem(struct vm_area_struct *vma);
550 : : #else
551 : : static inline bool vma_is_shmem(struct vm_area_struct *vma) { return false; }
552 : : #endif
553 : :
554 : : int vma_is_stack_for_current(struct vm_area_struct *vma);
555 : :
556 : : /* flush_tlb_range() takes a vma, not a mm, and can care about flags */
557 : : #define TLB_FLUSH_VMA(mm,flags) { .vm_mm = (mm), .vm_flags = (flags) }
558 : :
559 : : struct mmu_gather;
560 : : struct inode;
561 : :
562 : : /*
563 : : * FIXME: take this include out, include page-flags.h in
564 : : * files which need it (119 of them)
565 : : */
566 : : #include <linux/page-flags.h>
567 : : #include <linux/huge_mm.h>
568 : :
569 : : /*
570 : : * Methods to modify the page usage count.
571 : : *
572 : : * What counts for a page usage:
573 : : * - cache mapping (page->mapping)
574 : : * - private data (page->private)
575 : : * - page mapped in a task's page tables, each mapping
576 : : * is counted separately
577 : : *
578 : : * Also, many kernel routines increase the page count before a critical
579 : : * routine so they can be sure the page doesn't go away from under them.
580 : : */
581 : :
582 : : /*
583 : : * Drop a ref, return true if the refcount fell to zero (the page has no users)
584 : : */
585 : 10605543 : static inline int put_page_testzero(struct page *page)
586 : : {
587 : 10605543 : VM_BUG_ON_PAGE(page_ref_count(page) == 0, page);
588 : 10605543 : return page_ref_dec_and_test(page);
589 : : }
590 : :
591 : : /*
592 : : * Try to grab a ref unless the page has a refcount of zero, return false if
593 : : * that is the case.
594 : : * This can be called when MMU is off so it must not access
595 : : * any of the virtual mappings.
596 : : */
597 : 0 : static inline int get_page_unless_zero(struct page *page)
598 : : {
599 : 0 : return page_ref_add_unless(page, 1, 0);
600 : : }
601 : :
602 : : extern int page_is_ram(unsigned long pfn);
603 : :
604 : : enum {
605 : : REGION_INTERSECTS,
606 : : REGION_DISJOINT,
607 : : REGION_MIXED,
608 : : };
609 : :
610 : : int region_intersects(resource_size_t offset, size_t size, unsigned long flags,
611 : : unsigned long desc);
612 : :
613 : : /* Support for virtually mapped pages */
614 : : struct page *vmalloc_to_page(const void *addr);
615 : : unsigned long vmalloc_to_pfn(const void *addr);
616 : :
617 : : /*
618 : : * Determine if an address is within the vmalloc range
619 : : *
620 : : * On nommu, vmalloc/vfree wrap through kmalloc/kfree directly, so there
621 : : * is no special casing required.
622 : : */
623 : :
624 : : #ifndef is_ioremap_addr
625 : : #define is_ioremap_addr(x) is_vmalloc_addr(x)
626 : : #endif
627 : :
628 : : #ifdef CONFIG_MMU
629 : : extern bool is_vmalloc_addr(const void *x);
630 : : extern int is_vmalloc_or_module_addr(const void *x);
631 : : #else
632 : : static inline bool is_vmalloc_addr(const void *x)
633 : : {
634 : : return false;
635 : : }
636 : : static inline int is_vmalloc_or_module_addr(const void *x)
637 : : {
638 : : return 0;
639 : : }
640 : : #endif
641 : :
642 : : extern void *kvmalloc_node(size_t size, gfp_t flags, int node);
643 : 25851 : static inline void *kvmalloc(size_t size, gfp_t flags)
644 : : {
645 : 25526 : return kvmalloc_node(size, flags, NUMA_NO_NODE);
646 : : }
647 : 0 : static inline void *kvzalloc_node(size_t size, gfp_t flags, int node)
648 : : {
649 : 0 : return kvmalloc_node(size, flags | __GFP_ZERO, node);
650 : : }
651 : 1159 : static inline void *kvzalloc(size_t size, gfp_t flags)
652 : : {
653 : 1159 : return kvmalloc(size, flags | __GFP_ZERO);
654 : : }
655 : :
656 : 417 : static inline void *kvmalloc_array(size_t n, size_t size, gfp_t flags)
657 : : {
658 : 417 : size_t bytes;
659 : :
660 [ + - - - : 417 : if (unlikely(check_mul_overflow(n, size, &bytes)))
# # ]
661 : : return NULL;
662 : :
663 : 417 : return kvmalloc(bytes, flags);
664 : : }
665 : :
666 : 104 : static inline void *kvcalloc(size_t n, size_t size, gfp_t flags)
667 : : {
668 : 104 : return kvmalloc_array(n, size, flags | __GFP_ZERO);
669 : : }
670 : :
671 : : extern void kvfree(const void *addr);
672 : :
673 : 0 : static inline int compound_mapcount(struct page *page)
674 : : {
675 : 0 : VM_BUG_ON_PAGE(!PageCompound(page), page);
676 [ # # ]: 0 : page = compound_head(page);
677 : 0 : return atomic_read(compound_mapcount_ptr(page)) + 1;
678 : : }
679 : :
680 : : /*
681 : : * The atomic page->_mapcount, starts from -1: so that transitions
682 : : * both from it and to it can be tracked, using atomic_inc_and_test
683 : : * and atomic_add_negative(-1).
684 : : */
685 : 0 : static inline void page_mapcount_reset(struct page *page)
686 : : {
687 : 0 : atomic_set(&(page)->_mapcount, -1);
688 : : }
689 : :
690 : : int __page_mapcount(struct page *page);
691 : :
692 : 8111459 : static inline int page_mapcount(struct page *page)
693 : : {
694 : 8111459 : VM_BUG_ON_PAGE(PageSlab(page), page);
695 : :
696 [ - + ]: 16222904 : if (unlikely(PageCompound(page)))
697 : 0 : return __page_mapcount(page);
698 : 8111459 : return atomic_read(&page->_mapcount) + 1;
699 : : }
700 : :
701 : : #ifdef CONFIG_TRANSPARENT_HUGEPAGE
702 : : int total_mapcount(struct page *page);
703 : : int page_trans_huge_mapcount(struct page *page, int *total_mapcount);
704 : : #else
705 : 0 : static inline int total_mapcount(struct page *page)
706 : : {
707 : 0 : return page_mapcount(page);
708 : : }
709 : 261462 : static inline int page_trans_huge_mapcount(struct page *page,
710 : : int *total_mapcount)
711 : : {
712 : 261462 : int mapcount = page_mapcount(page);
713 [ + - ]: 261462 : if (total_mapcount)
714 : 261462 : *total_mapcount = mapcount;
715 [ - + ]: 261462 : return mapcount;
716 : : }
717 : : #endif
718 : :
719 : 38395286 : static inline struct page *virt_to_head_page(const void *x)
720 : : {
721 [ + - + - : 38395286 : struct page *page = virt_to_page(x);
+ - + - +
- + - - -
# # # # #
# ]
722 : :
723 [ + + + + : 38395286 : return compound_head(page);
+ + + + +
+ + + - -
# # # # #
# ]
724 : : }
725 : :
726 : : void __put_page(struct page *page);
727 : :
728 : : void put_pages_list(struct list_head *pages);
729 : :
730 : : void split_page(struct page *page, unsigned int order);
731 : :
732 : : /*
733 : : * Compound pages have a destructor function. Provide a
734 : : * prototype for that function and accessor functions.
735 : : * These are _only_ valid on the head of a compound page.
736 : : */
737 : : typedef void compound_page_dtor(struct page *);
738 : :
739 : : /* Keep the enum in sync with compound_page_dtors array in mm/page_alloc.c */
740 : : enum compound_dtor_id {
741 : : NULL_COMPOUND_DTOR,
742 : : COMPOUND_PAGE_DTOR,
743 : : #ifdef CONFIG_HUGETLB_PAGE
744 : : HUGETLB_PAGE_DTOR,
745 : : #endif
746 : : #ifdef CONFIG_TRANSPARENT_HUGEPAGE
747 : : TRANSHUGE_PAGE_DTOR,
748 : : #endif
749 : : NR_COMPOUND_DTORS,
750 : : };
751 : : extern compound_page_dtor * const compound_page_dtors[];
752 : :
753 : 136035 : static inline void set_compound_page_dtor(struct page *page,
754 : : enum compound_dtor_id compound_dtor)
755 : : {
756 : 136035 : VM_BUG_ON_PAGE(compound_dtor >= NR_COMPOUND_DTORS, page);
757 : 136035 : page[1].compound_dtor = compound_dtor;
758 : : }
759 : :
760 : 0 : static inline compound_page_dtor *get_compound_page_dtor(struct page *page)
761 : : {
762 : 0 : VM_BUG_ON_PAGE(page[1].compound_dtor >= NR_COMPOUND_DTORS, page);
763 : 0 : return compound_page_dtors[page[1].compound_dtor];
764 : : }
765 : :
766 : 9796385 : static inline unsigned int compound_order(struct page *page)
767 : : {
768 [ + + + + : 9728005 : if (!PageHead(page))
+ + + + -
+ - + -
- ]
769 : : return 0;
770 [ # # # # ]: 1260272 : return page[1].compound_order;
771 : : }
772 : :
773 : 136035 : static inline void set_compound_order(struct page *page, unsigned int order)
774 : : {
775 : 136035 : page[1].compound_order = order;
776 : : }
777 : :
778 : : /* Returns the number of pages in this potentially compound page. */
779 : 9083044 : static inline unsigned long compound_nr(struct page *page)
780 : : {
781 [ + + ]: 10547786 : return 1UL << compound_order(page);
782 : : }
783 : :
784 : : /* Returns the number of bytes in this potentially compound page. */
785 : 370424 : static inline unsigned long page_size(struct page *page)
786 : : {
787 [ - - ]: 397438 : return PAGE_SIZE << compound_order(page);
788 : : }
789 : :
790 : : /* Returns the number of bits needed for the number of bytes in a page */
791 : : static inline unsigned int page_shift(struct page *page)
792 : : {
793 : : return PAGE_SHIFT + compound_order(page);
794 : : }
795 : :
796 : : void free_compound_page(struct page *page);
797 : :
798 : : #ifdef CONFIG_MMU
799 : : /*
800 : : * Do pte_mkwrite, but only if the vma says VM_WRITE. We do this when
801 : : * servicing faults for write access. In the normal case, do always want
802 : : * pte_mkwrite. But get_user_pages can cause write faults for mappings
803 : : * that do not have writing enabled, when used by access_process_vm.
804 : : */
805 : 478815 : static inline pte_t maybe_mkwrite(pte_t pte, struct vm_area_struct *vma)
806 : : {
807 [ + - - - : 478815 : if (likely(vma->vm_flags & VM_WRITE))
+ - + - -
- - - ]
808 : 478815 : pte = pte_mkwrite(pte);
809 : 386665 : return pte;
810 : : }
811 : :
812 : : vm_fault_t alloc_set_pte(struct vm_fault *vmf, struct mem_cgroup *memcg,
813 : : struct page *page);
814 : : vm_fault_t finish_fault(struct vm_fault *vmf);
815 : : vm_fault_t finish_mkwrite_fault(struct vm_fault *vmf);
816 : : #endif
817 : :
818 : : /*
819 : : * Multiple processes may "see" the same page. E.g. for untouched
820 : : * mappings of /dev/null, all processes see the same page full of
821 : : * zeroes, and text pages of executables and shared libraries have
822 : : * only one copy in memory, at most, normally.
823 : : *
824 : : * For the non-reserved pages, page_count(page) denotes a reference count.
825 : : * page_count() == 0 means the page is free. page->lru is then used for
826 : : * freelist management in the buddy allocator.
827 : : * page_count() > 0 means the page has been allocated.
828 : : *
829 : : * Pages are allocated by the slab allocator in order to provide memory
830 : : * to kmalloc and kmem_cache_alloc. In this case, the management of the
831 : : * page, and the fields in 'struct page' are the responsibility of mm/slab.c
832 : : * unless a particular usage is carefully commented. (the responsibility of
833 : : * freeing the kmalloc memory is the caller's, of course).
834 : : *
835 : : * A page may be used by anyone else who does a __get_free_page().
836 : : * In this case, page_count still tracks the references, and should only
837 : : * be used through the normal accessor functions. The top bits of page->flags
838 : : * and page->virtual store page management information, but all other fields
839 : : * are unused and could be used privately, carefully. The management of this
840 : : * page is the responsibility of the one who allocated it, and those who have
841 : : * subsequently been given references to it.
842 : : *
843 : : * The other pages (we may call them "pagecache pages") are completely
844 : : * managed by the Linux memory manager: I/O, buffers, swapping etc.
845 : : * The following discussion applies only to them.
846 : : *
847 : : * A pagecache page contains an opaque `private' member, which belongs to the
848 : : * page's address_space. Usually, this is the address of a circular list of
849 : : * the page's disk buffers. PG_private must be set to tell the VM to call
850 : : * into the filesystem to release these pages.
851 : : *
852 : : * A page may belong to an inode's memory mapping. In this case, page->mapping
853 : : * is the pointer to the inode, and page->index is the file offset of the page,
854 : : * in units of PAGE_SIZE.
855 : : *
856 : : * If pagecache pages are not associated with an inode, they are said to be
857 : : * anonymous pages. These may become associated with the swapcache, and in that
858 : : * case PG_swapcache is set, and page->private is an offset into the swapcache.
859 : : *
860 : : * In either case (swapcache or inode backed), the pagecache itself holds one
861 : : * reference to the page. Setting PG_private should also increment the
862 : : * refcount. The each user mapping also has a reference to the page.
863 : : *
864 : : * The pagecache pages are stored in a per-mapping radix tree, which is
865 : : * rooted at mapping->i_pages, and indexed by offset.
866 : : * Where 2.4 and early 2.6 kernels kept dirty/clean pages in per-address_space
867 : : * lists, we instead now tag pages as dirty/writeback in the radix tree.
868 : : *
869 : : * All pagecache pages may be subject to I/O:
870 : : * - inode pages may need to be read from disk,
871 : : * - inode pages which have been modified and are MAP_SHARED may need
872 : : * to be written back to the inode on disk,
873 : : * - anonymous pages (including MAP_PRIVATE file mappings) which have been
874 : : * modified may need to be swapped out to swap space and (later) to be read
875 : : * back into memory.
876 : : */
877 : :
878 : : /*
879 : : * The zone field is never updated after free_area_init_core()
880 : : * sets it, so none of the operations on it need to be atomic.
881 : : */
882 : :
883 : : /* Page flags: | [SECTION] | [NODE] | ZONE | [LAST_CPUPID] | ... | FLAGS | */
884 : : #define SECTIONS_PGOFF ((sizeof(unsigned long)*8) - SECTIONS_WIDTH)
885 : : #define NODES_PGOFF (SECTIONS_PGOFF - NODES_WIDTH)
886 : : #define ZONES_PGOFF (NODES_PGOFF - ZONES_WIDTH)
887 : : #define LAST_CPUPID_PGOFF (ZONES_PGOFF - LAST_CPUPID_WIDTH)
888 : : #define KASAN_TAG_PGOFF (LAST_CPUPID_PGOFF - KASAN_TAG_WIDTH)
889 : :
890 : : /*
891 : : * Define the bit shifts to access each section. For non-existent
892 : : * sections we define the shift as 0; that plus a 0 mask ensures
893 : : * the compiler will optimise away reference to them.
894 : : */
895 : : #define SECTIONS_PGSHIFT (SECTIONS_PGOFF * (SECTIONS_WIDTH != 0))
896 : : #define NODES_PGSHIFT (NODES_PGOFF * (NODES_WIDTH != 0))
897 : : #define ZONES_PGSHIFT (ZONES_PGOFF * (ZONES_WIDTH != 0))
898 : : #define LAST_CPUPID_PGSHIFT (LAST_CPUPID_PGOFF * (LAST_CPUPID_WIDTH != 0))
899 : : #define KASAN_TAG_PGSHIFT (KASAN_TAG_PGOFF * (KASAN_TAG_WIDTH != 0))
900 : :
901 : : /* NODE:ZONE or SECTION:ZONE is used to ID a zone for the buddy allocator */
902 : : #ifdef NODE_NOT_IN_PAGE_FLAGS
903 : : #define ZONEID_SHIFT (SECTIONS_SHIFT + ZONES_SHIFT)
904 : : #define ZONEID_PGOFF ((SECTIONS_PGOFF < ZONES_PGOFF)? \
905 : : SECTIONS_PGOFF : ZONES_PGOFF)
906 : : #else
907 : : #define ZONEID_SHIFT (NODES_SHIFT + ZONES_SHIFT)
908 : : #define ZONEID_PGOFF ((NODES_PGOFF < ZONES_PGOFF)? \
909 : : NODES_PGOFF : ZONES_PGOFF)
910 : : #endif
911 : :
912 : : #define ZONEID_PGSHIFT (ZONEID_PGOFF * (ZONEID_SHIFT != 0))
913 : :
914 : : #define ZONES_MASK ((1UL << ZONES_WIDTH) - 1)
915 : : #define NODES_MASK ((1UL << NODES_WIDTH) - 1)
916 : : #define SECTIONS_MASK ((1UL << SECTIONS_WIDTH) - 1)
917 : : #define LAST_CPUPID_MASK ((1UL << LAST_CPUPID_SHIFT) - 1)
918 : : #define KASAN_TAG_MASK ((1UL << KASAN_TAG_WIDTH) - 1)
919 : : #define ZONEID_MASK ((1UL << ZONEID_SHIFT) - 1)
920 : :
921 : 1938252 : static inline enum zone_type page_zonenum(const struct page *page)
922 : : {
923 [ - + ]: 1763968 : return (page->flags >> ZONES_PGSHIFT) & ZONES_MASK;
924 : : }
925 : :
926 : : #ifdef CONFIG_ZONE_DEVICE
927 : : static inline bool is_zone_device_page(const struct page *page)
928 : : {
929 : : return page_zonenum(page) == ZONE_DEVICE;
930 : : }
931 : : extern void memmap_init_zone_device(struct zone *, unsigned long,
932 : : unsigned long, struct dev_pagemap *);
933 : : #else
934 : 8540278 : static inline bool is_zone_device_page(const struct page *page)
935 : : {
936 [ - + ]: 8540278 : return false;
937 : : }
938 : : #endif
939 : :
940 : : #ifdef CONFIG_DEV_PAGEMAP_OPS
941 : : void free_devmap_managed_page(struct page *page);
942 : : DECLARE_STATIC_KEY_FALSE(devmap_managed_key);
943 : :
944 : : static inline bool page_is_devmap_managed(struct page *page)
945 : : {
946 : : if (!static_branch_unlikely(&devmap_managed_key))
947 : : return false;
948 : : if (!is_zone_device_page(page))
949 : : return false;
950 : : switch (page->pgmap->type) {
951 : : case MEMORY_DEVICE_PRIVATE:
952 : : case MEMORY_DEVICE_FS_DAX:
953 : : return true;
954 : : default:
955 : : break;
956 : : }
957 : : return false;
958 : : }
959 : :
960 : : void put_devmap_managed_page(struct page *page);
961 : :
962 : : #else /* CONFIG_DEV_PAGEMAP_OPS */
963 : 1530141 : static inline bool page_is_devmap_managed(struct page *page)
964 : : {
965 : 1530141 : return false;
966 : : }
967 : :
968 : : static inline void put_devmap_managed_page(struct page *page)
969 : : {
970 : : }
971 : : #endif /* CONFIG_DEV_PAGEMAP_OPS */
972 : :
973 : 0 : static inline bool is_device_private_page(const struct page *page)
974 : : {
975 [ # # # # ]: 0 : return IS_ENABLED(CONFIG_DEV_PAGEMAP_OPS) &&
976 : : IS_ENABLED(CONFIG_DEVICE_PRIVATE) &&
977 : : is_zone_device_page(page) &&
978 : : page->pgmap->type == MEMORY_DEVICE_PRIVATE;
979 : : }
980 : :
981 : : static inline bool is_pci_p2pdma_page(const struct page *page)
982 : : {
983 : : return IS_ENABLED(CONFIG_DEV_PAGEMAP_OPS) &&
984 : : IS_ENABLED(CONFIG_PCI_P2PDMA) &&
985 : : is_zone_device_page(page) &&
986 : : page->pgmap->type == MEMORY_DEVICE_PCI_P2PDMA;
987 : : }
988 : :
989 : : /* 127: arbitrary random number, small enough to assemble well */
990 : : #define page_ref_zero_or_close_to_overflow(page) \
991 : : ((unsigned int) page_ref_count(page) + 127u <= 127u)
992 : :
993 : 1862615 : static inline void get_page(struct page *page)
994 : : {
995 [ - + - + : 1862615 : page = compound_head(page);
- + - + -
- - - # #
# # # # #
# # # #
# ]
996 : : /*
997 : : * Getting a normal page or the head of a compound page
998 : : * requires to already have an elevated page->_refcount.
999 : : */
1000 : 1862615 : VM_BUG_ON_PAGE(page_ref_zero_or_close_to_overflow(page), page);
1001 : 1862615 : page_ref_inc(page);
1002 : 65 : }
1003 : :
1004 : 33023 : static inline __must_check bool try_get_page(struct page *page)
1005 : : {
1006 [ - + ]: 33023 : page = compound_head(page);
1007 [ - + + - ]: 33023 : if (WARN_ON_ONCE(page_ref_count(page) <= 0))
1008 : : return false;
1009 : 33023 : page_ref_inc(page);
1010 : 33023 : return true;
1011 : : }
1012 : :
1013 : 1530141 : static inline void put_page(struct page *page)
1014 : : {
1015 [ - + ]: 1530141 : page = compound_head(page);
1016 : :
1017 : : /*
1018 : : * For devmap managed pages we need to catch refcount transition from
1019 : : * 2 to 1, when refcount reach one it means the page is free and we
1020 : : * need to inform the device driver through callback. See
1021 : : * include/linux/memremap.h and HMM for details.
1022 : : */
1023 : 1530141 : if (page_is_devmap_managed(page)) {
1024 : : put_devmap_managed_page(page);
1025 : : return;
1026 : : }
1027 : :
1028 [ + + ]: 1530141 : if (put_page_testzero(page))
1029 : 1458 : __put_page(page);
1030 : : }
1031 : :
1032 : : /**
1033 : : * unpin_user_page() - release a gup-pinned page
1034 : : * @page: pointer to page to be released
1035 : : *
1036 : : * Pages that were pinned via pin_user_pages*() must be released via either
1037 : : * unpin_user_page(), or one of the unpin_user_pages*() routines. This is so
1038 : : * that eventually such pages can be separately tracked and uniquely handled. In
1039 : : * particular, interactions with RDMA and filesystems need special handling.
1040 : : *
1041 : : * unpin_user_page() and put_page() are not interchangeable, despite this early
1042 : : * implementation that makes them look the same. unpin_user_page() calls must
1043 : : * be perfectly matched up with pin*() calls.
1044 : : */
1045 : 0 : static inline void unpin_user_page(struct page *page)
1046 : : {
1047 : 0 : put_page(page);
1048 : : }
1049 : :
1050 : : void unpin_user_pages_dirty_lock(struct page **pages, unsigned long npages,
1051 : : bool make_dirty);
1052 : :
1053 : : void unpin_user_pages(struct page **pages, unsigned long npages);
1054 : :
1055 : : #if defined(CONFIG_SPARSEMEM) && !defined(CONFIG_SPARSEMEM_VMEMMAP)
1056 : : #define SECTION_IN_PAGE_FLAGS
1057 : : #endif
1058 : :
1059 : : /*
1060 : : * The identification function is mainly used by the buddy allocator for
1061 : : * determining if two pages could be buddies. We are not really identifying
1062 : : * the zone since we could be using the section number id if we do not have
1063 : : * node id available in page flags.
1064 : : * We only guarantee that it will return the same value for two combinable
1065 : : * pages in a zone.
1066 : : */
1067 : 311439 : static inline int page_zone_id(struct page *page)
1068 : : {
1069 [ - + + - : 311439 : return (page->flags >> ZONEID_PGSHIFT) & ZONEID_MASK;
+ - ]
1070 : : }
1071 : :
1072 : : #ifdef NODE_NOT_IN_PAGE_FLAGS
1073 : : extern int page_to_nid(const struct page *page);
1074 : : #else
1075 : 5088977 : static inline int page_to_nid(const struct page *page)
1076 : : {
1077 : 5088977 : struct page *p = (struct page *)page;
1078 : :
1079 [ + - - - : 4394087 : return (PF_POISONED_CHECK(p)->flags >> NODES_PGSHIFT) & NODES_MASK;
- - - + -
+ - + - +
- + + + ]
1080 : : }
1081 : : #endif
1082 : :
1083 : : #ifdef CONFIG_NUMA_BALANCING
1084 : : static inline int cpu_pid_to_cpupid(int cpu, int pid)
1085 : : {
1086 : : return ((cpu & LAST__CPU_MASK) << LAST__PID_SHIFT) | (pid & LAST__PID_MASK);
1087 : : }
1088 : :
1089 : : static inline int cpupid_to_pid(int cpupid)
1090 : : {
1091 : : return cpupid & LAST__PID_MASK;
1092 : : }
1093 : :
1094 : : static inline int cpupid_to_cpu(int cpupid)
1095 : : {
1096 : : return (cpupid >> LAST__PID_SHIFT) & LAST__CPU_MASK;
1097 : : }
1098 : :
1099 : : static inline int cpupid_to_nid(int cpupid)
1100 : : {
1101 : : return cpu_to_node(cpupid_to_cpu(cpupid));
1102 : : }
1103 : :
1104 : : static inline bool cpupid_pid_unset(int cpupid)
1105 : : {
1106 : : return cpupid_to_pid(cpupid) == (-1 & LAST__PID_MASK);
1107 : : }
1108 : :
1109 : : static inline bool cpupid_cpu_unset(int cpupid)
1110 : : {
1111 : : return cpupid_to_cpu(cpupid) == (-1 & LAST__CPU_MASK);
1112 : : }
1113 : :
1114 : : static inline bool __cpupid_match_pid(pid_t task_pid, int cpupid)
1115 : : {
1116 : : return (task_pid & LAST__PID_MASK) == cpupid_to_pid(cpupid);
1117 : : }
1118 : :
1119 : : #define cpupid_match_pid(task, cpupid) __cpupid_match_pid(task->pid, cpupid)
1120 : : #ifdef LAST_CPUPID_NOT_IN_PAGE_FLAGS
1121 : : static inline int page_cpupid_xchg_last(struct page *page, int cpupid)
1122 : : {
1123 : : return xchg(&page->_last_cpupid, cpupid & LAST_CPUPID_MASK);
1124 : : }
1125 : :
1126 : : static inline int page_cpupid_last(struct page *page)
1127 : : {
1128 : : return page->_last_cpupid;
1129 : : }
1130 : : static inline void page_cpupid_reset_last(struct page *page)
1131 : : {
1132 : : page->_last_cpupid = -1 & LAST_CPUPID_MASK;
1133 : : }
1134 : : #else
1135 : : static inline int page_cpupid_last(struct page *page)
1136 : : {
1137 : : return (page->flags >> LAST_CPUPID_PGSHIFT) & LAST_CPUPID_MASK;
1138 : : }
1139 : :
1140 : : extern int page_cpupid_xchg_last(struct page *page, int cpupid);
1141 : :
1142 : : static inline void page_cpupid_reset_last(struct page *page)
1143 : : {
1144 : : page->flags |= LAST_CPUPID_MASK << LAST_CPUPID_PGSHIFT;
1145 : : }
1146 : : #endif /* LAST_CPUPID_NOT_IN_PAGE_FLAGS */
1147 : : #else /* !CONFIG_NUMA_BALANCING */
1148 : 0 : static inline int page_cpupid_xchg_last(struct page *page, int cpupid)
1149 : : {
1150 [ # # ]: 0 : return page_to_nid(page); /* XXX */
1151 : : }
1152 : :
1153 : : static inline int page_cpupid_last(struct page *page)
1154 : : {
1155 : : return page_to_nid(page); /* XXX */
1156 : : }
1157 : :
1158 : : static inline int cpupid_to_nid(int cpupid)
1159 : : {
1160 : : return -1;
1161 : : }
1162 : :
1163 : : static inline int cpupid_to_pid(int cpupid)
1164 : : {
1165 : : return -1;
1166 : : }
1167 : :
1168 : : static inline int cpupid_to_cpu(int cpupid)
1169 : : {
1170 : : return -1;
1171 : : }
1172 : :
1173 : : static inline int cpu_pid_to_cpupid(int nid, int pid)
1174 : : {
1175 : : return -1;
1176 : : }
1177 : :
1178 : : static inline bool cpupid_pid_unset(int cpupid)
1179 : : {
1180 : : return 1;
1181 : : }
1182 : :
1183 : : static inline void page_cpupid_reset_last(struct page *page)
1184 : : {
1185 : : }
1186 : :
1187 : : static inline bool cpupid_match_pid(struct task_struct *task, int cpupid)
1188 : : {
1189 : : return false;
1190 : : }
1191 : : #endif /* CONFIG_NUMA_BALANCING */
1192 : :
1193 : : #ifdef CONFIG_KASAN_SW_TAGS
1194 : : static inline u8 page_kasan_tag(const struct page *page)
1195 : : {
1196 : : return (page->flags >> KASAN_TAG_PGSHIFT) & KASAN_TAG_MASK;
1197 : : }
1198 : :
1199 : : static inline void page_kasan_tag_set(struct page *page, u8 tag)
1200 : : {
1201 : : page->flags &= ~(KASAN_TAG_MASK << KASAN_TAG_PGSHIFT);
1202 : : page->flags |= (tag & KASAN_TAG_MASK) << KASAN_TAG_PGSHIFT;
1203 : : }
1204 : :
1205 : : static inline void page_kasan_tag_reset(struct page *page)
1206 : : {
1207 : : page_kasan_tag_set(page, 0xff);
1208 : : }
1209 : : #else
1210 : : static inline u8 page_kasan_tag(const struct page *page)
1211 : : {
1212 : : return 0xff;
1213 : : }
1214 : :
1215 : : static inline void page_kasan_tag_set(struct page *page, u8 tag) { }
1216 : 1096879 : static inline void page_kasan_tag_reset(struct page *page) { }
1217 : : #endif
1218 : :
1219 : 577696 : static inline struct zone *page_zone(const struct page *page)
1220 : : {
1221 [ - - - - : 577696 : return &NODE_DATA(page_to_nid(page))->node_zones[page_zonenum(page)];
+ - ]
1222 : : }
1223 : :
1224 : 3710774 : static inline pg_data_t *page_pgdat(const struct page *page)
1225 : : {
1226 [ + + + + ]: 3693679 : return NODE_DATA(page_to_nid(page));
1227 : : }
1228 : :
1229 : : #ifdef SECTION_IN_PAGE_FLAGS
1230 : : static inline void set_page_section(struct page *page, unsigned long section)
1231 : : {
1232 : : page->flags &= ~(SECTIONS_MASK << SECTIONS_PGSHIFT);
1233 : : page->flags |= (section & SECTIONS_MASK) << SECTIONS_PGSHIFT;
1234 : : }
1235 : :
1236 : : static inline unsigned long page_to_section(const struct page *page)
1237 : : {
1238 : : return (page->flags >> SECTIONS_PGSHIFT) & SECTIONS_MASK;
1239 : : }
1240 : : #endif
1241 : :
1242 : : static inline void set_page_zone(struct page *page, enum zone_type zone)
1243 : : {
1244 : : page->flags &= ~(ZONES_MASK << ZONES_PGSHIFT);
1245 : : page->flags |= (zone & ZONES_MASK) << ZONES_PGSHIFT;
1246 : : }
1247 : :
1248 : : static inline void set_page_node(struct page *page, unsigned long node)
1249 : : {
1250 : : page->flags &= ~(NODES_MASK << NODES_PGSHIFT);
1251 : : page->flags |= (node & NODES_MASK) << NODES_PGSHIFT;
1252 : : }
1253 : :
1254 : : static inline void set_page_links(struct page *page, enum zone_type zone,
1255 : : unsigned long node, unsigned long pfn)
1256 : : {
1257 : : set_page_zone(page, zone);
1258 : : set_page_node(page, node);
1259 : : #ifdef SECTION_IN_PAGE_FLAGS
1260 : : set_page_section(page, pfn_to_section_nr(pfn));
1261 : : #endif
1262 : : }
1263 : :
1264 : : #ifdef CONFIG_MEMCG
1265 : : static inline struct mem_cgroup *page_memcg(struct page *page)
1266 : : {
1267 : : return page->mem_cgroup;
1268 : : }
1269 : : static inline struct mem_cgroup *page_memcg_rcu(struct page *page)
1270 : : {
1271 : : WARN_ON_ONCE(!rcu_read_lock_held());
1272 : : return READ_ONCE(page->mem_cgroup);
1273 : : }
1274 : : #else
1275 : 0 : static inline struct mem_cgroup *page_memcg(struct page *page)
1276 : : {
1277 : 0 : return NULL;
1278 : : }
1279 : 60147 : static inline struct mem_cgroup *page_memcg_rcu(struct page *page)
1280 : : {
1281 : 60147 : WARN_ON_ONCE(!rcu_read_lock_held());
1282 : 60147 : return NULL;
1283 : : }
1284 : : #endif
1285 : :
1286 : : /*
1287 : : * Some inline functions in vmstat.h depend on page_zone()
1288 : : */
1289 : : #include <linux/vmstat.h>
1290 : :
1291 : 15205715 : static __always_inline void *lowmem_page_address(const struct page *page)
1292 : : {
1293 [ + + + + : 15203460 : return page_to_virt(page);
+ + # # #
# # # # #
# # ]
1294 : : }
1295 : :
1296 : : #if defined(CONFIG_HIGHMEM) && !defined(WANT_PAGE_VIRTUAL)
1297 : : #define HASHED_PAGE_VIRTUAL
1298 : : #endif
1299 : :
1300 : : #if defined(WANT_PAGE_VIRTUAL)
1301 : : static inline void *page_address(const struct page *page)
1302 : : {
1303 : : return page->virtual;
1304 : : }
1305 : : static inline void set_page_address(struct page *page, void *address)
1306 : : {
1307 : : page->virtual = address;
1308 : : }
1309 : : #define page_address_init() do { } while(0)
1310 : : #endif
1311 : :
1312 : : #if defined(HASHED_PAGE_VIRTUAL)
1313 : : void *page_address(const struct page *page);
1314 : : void set_page_address(struct page *page, void *virtual);
1315 : : void page_address_init(void);
1316 : : #endif
1317 : :
1318 : : #if !defined(HASHED_PAGE_VIRTUAL) && !defined(WANT_PAGE_VIRTUAL)
1319 : : #define page_address(page) lowmem_page_address(page)
1320 : : #define set_page_address(page, address) do { } while(0)
1321 : : #define page_address_init() do { } while(0)
1322 : : #endif
1323 : :
1324 : : extern void *page_rmapping(struct page *page);
1325 : : extern struct anon_vma *page_anon_vma(struct page *page);
1326 : : extern struct address_space *page_mapping(struct page *page);
1327 : :
1328 : : extern struct address_space *__page_file_mapping(struct page *);
1329 : :
1330 : : static inline
1331 : 0 : struct address_space *page_file_mapping(struct page *page)
1332 : : {
1333 [ # # # # ]: 0 : if (unlikely(PageSwapCache(page)))
1334 : 0 : return __page_file_mapping(page);
1335 : :
1336 : 0 : return page->mapping;
1337 : : }
1338 : :
1339 : : extern pgoff_t __page_file_index(struct page *page);
1340 : :
1341 : : /*
1342 : : * Return the pagecache index of the passed page. Regular pagecache pages
1343 : : * use ->index whereas swapcache pages use swp_offset(->private)
1344 : : */
1345 : 8219 : static inline pgoff_t page_index(struct page *page)
1346 : : {
1347 [ - + - + ]: 16438 : if (unlikely(PageSwapCache(page)))
1348 : 0 : return __page_file_index(page);
1349 : 8219 : return page->index;
1350 : : }
1351 : :
1352 : : bool page_mapped(struct page *page);
1353 : : struct address_space *page_mapping(struct page *page);
1354 : : struct address_space *page_mapping_file(struct page *page);
1355 : :
1356 : : /*
1357 : : * Return true only if the page has been allocated with
1358 : : * ALLOC_NO_WATERMARKS and the low watermark was not
1359 : : * met implying that the system is under some pressure.
1360 : : */
1361 : 343324 : static inline bool page_is_pfmemalloc(struct page *page)
1362 : : {
1363 : : /*
1364 : : * Page index cannot be this large so this must be
1365 : : * a pfmemalloc page.
1366 : : */
1367 [ - + # # : 343324 : return page->index == -1UL;
# # # # #
# # # # #
# # # # ]
1368 : : }
1369 : :
1370 : : /*
1371 : : * Only to be called by the page allocator on a freshly allocated
1372 : : * page.
1373 : : */
1374 : 0 : static inline void set_page_pfmemalloc(struct page *page)
1375 : : {
1376 : 0 : page->index = -1UL;
1377 : 0 : }
1378 : :
1379 : 1632038 : static inline void clear_page_pfmemalloc(struct page *page)
1380 : : {
1381 : 1632038 : page->index = 0;
1382 : 1632038 : }
1383 : :
1384 : : /*
1385 : : * Can be called by the pagefault handler when it gets a VM_FAULT_OOM.
1386 : : */
1387 : : extern void pagefault_out_of_memory(void);
1388 : :
1389 : : #define offset_in_page(p) ((unsigned long)(p) & ~PAGE_MASK)
1390 : :
1391 : : /*
1392 : : * Flags passed to show_mem() and show_free_areas() to suppress output in
1393 : : * various contexts.
1394 : : */
1395 : : #define SHOW_MEM_FILTER_NODES (0x0001u) /* disallowed nodes */
1396 : :
1397 : : extern void show_free_areas(unsigned int flags, nodemask_t *nodemask);
1398 : :
1399 : : #ifdef CONFIG_MMU
1400 : : extern bool can_do_mlock(void);
1401 : : #else
1402 : : static inline bool can_do_mlock(void) { return false; }
1403 : : #endif
1404 : : extern int user_shm_lock(size_t, struct user_struct *);
1405 : : extern void user_shm_unlock(size_t, struct user_struct *);
1406 : :
1407 : : /*
1408 : : * Parameter block passed down to zap_pte_range in exceptional cases.
1409 : : */
1410 : : struct zap_details {
1411 : : struct address_space *check_mapping; /* Check page->mapping if set */
1412 : : pgoff_t first_index; /* Lowest page->index to unmap */
1413 : : pgoff_t last_index; /* Highest page->index to unmap */
1414 : : };
1415 : :
1416 : : struct page *vm_normal_page(struct vm_area_struct *vma, unsigned long addr,
1417 : : pte_t pte);
1418 : : struct page *vm_normal_page_pmd(struct vm_area_struct *vma, unsigned long addr,
1419 : : pmd_t pmd);
1420 : :
1421 : : void zap_vma_ptes(struct vm_area_struct *vma, unsigned long address,
1422 : : unsigned long size);
1423 : : void zap_page_range(struct vm_area_struct *vma, unsigned long address,
1424 : : unsigned long size);
1425 : : void unmap_vmas(struct mmu_gather *tlb, struct vm_area_struct *start_vma,
1426 : : unsigned long start, unsigned long end);
1427 : :
1428 : : struct mmu_notifier_range;
1429 : :
1430 : : void free_pgd_range(struct mmu_gather *tlb, unsigned long addr,
1431 : : unsigned long end, unsigned long floor, unsigned long ceiling);
1432 : : int copy_page_range(struct mm_struct *dst, struct mm_struct *src,
1433 : : struct vm_area_struct *vma);
1434 : : int follow_pte_pmd(struct mm_struct *mm, unsigned long address,
1435 : : struct mmu_notifier_range *range,
1436 : : pte_t **ptepp, pmd_t **pmdpp, spinlock_t **ptlp);
1437 : : int follow_pfn(struct vm_area_struct *vma, unsigned long address,
1438 : : unsigned long *pfn);
1439 : : int follow_phys(struct vm_area_struct *vma, unsigned long address,
1440 : : unsigned int flags, unsigned long *prot, resource_size_t *phys);
1441 : : int generic_access_phys(struct vm_area_struct *vma, unsigned long addr,
1442 : : void *buf, int len, int write);
1443 : :
1444 : : extern void truncate_pagecache(struct inode *inode, loff_t new);
1445 : : extern void truncate_setsize(struct inode *inode, loff_t newsize);
1446 : : void pagecache_isize_extended(struct inode *inode, loff_t from, loff_t to);
1447 : : void truncate_pagecache_range(struct inode *inode, loff_t offset, loff_t end);
1448 : : int truncate_inode_page(struct address_space *mapping, struct page *page);
1449 : : int generic_error_remove_page(struct address_space *mapping, struct page *page);
1450 : : int invalidate_inode_page(struct page *page);
1451 : :
1452 : : #ifdef CONFIG_MMU
1453 : : extern vm_fault_t handle_mm_fault(struct vm_area_struct *vma,
1454 : : unsigned long address, unsigned int flags);
1455 : : extern int fixup_user_fault(struct task_struct *tsk, struct mm_struct *mm,
1456 : : unsigned long address, unsigned int fault_flags,
1457 : : bool *unlocked);
1458 : : void unmap_mapping_pages(struct address_space *mapping,
1459 : : pgoff_t start, pgoff_t nr, bool even_cows);
1460 : : void unmap_mapping_range(struct address_space *mapping,
1461 : : loff_t const holebegin, loff_t const holelen, int even_cows);
1462 : : #else
1463 : : static inline vm_fault_t handle_mm_fault(struct vm_area_struct *vma,
1464 : : unsigned long address, unsigned int flags)
1465 : : {
1466 : : /* should never happen if there's no MMU */
1467 : : BUG();
1468 : : return VM_FAULT_SIGBUS;
1469 : : }
1470 : : static inline int fixup_user_fault(struct task_struct *tsk,
1471 : : struct mm_struct *mm, unsigned long address,
1472 : : unsigned int fault_flags, bool *unlocked)
1473 : : {
1474 : : /* should never happen if there's no MMU */
1475 : : BUG();
1476 : : return -EFAULT;
1477 : : }
1478 : : static inline void unmap_mapping_pages(struct address_space *mapping,
1479 : : pgoff_t start, pgoff_t nr, bool even_cows) { }
1480 : : static inline void unmap_mapping_range(struct address_space *mapping,
1481 : : loff_t const holebegin, loff_t const holelen, int even_cows) { }
1482 : : #endif
1483 : :
1484 : : static inline void unmap_shared_mapping_range(struct address_space *mapping,
1485 : : loff_t const holebegin, loff_t const holelen)
1486 : : {
1487 : : unmap_mapping_range(mapping, holebegin, holelen, 0);
1488 : : }
1489 : :
1490 : : extern int access_process_vm(struct task_struct *tsk, unsigned long addr,
1491 : : void *buf, int len, unsigned int gup_flags);
1492 : : extern int access_remote_vm(struct mm_struct *mm, unsigned long addr,
1493 : : void *buf, int len, unsigned int gup_flags);
1494 : : extern int __access_remote_vm(struct task_struct *tsk, struct mm_struct *mm,
1495 : : unsigned long addr, void *buf, int len, unsigned int gup_flags);
1496 : :
1497 : : long get_user_pages_remote(struct task_struct *tsk, struct mm_struct *mm,
1498 : : unsigned long start, unsigned long nr_pages,
1499 : : unsigned int gup_flags, struct page **pages,
1500 : : struct vm_area_struct **vmas, int *locked);
1501 : : long pin_user_pages_remote(struct task_struct *tsk, struct mm_struct *mm,
1502 : : unsigned long start, unsigned long nr_pages,
1503 : : unsigned int gup_flags, struct page **pages,
1504 : : struct vm_area_struct **vmas, int *locked);
1505 : : long get_user_pages(unsigned long start, unsigned long nr_pages,
1506 : : unsigned int gup_flags, struct page **pages,
1507 : : struct vm_area_struct **vmas);
1508 : : long pin_user_pages(unsigned long start, unsigned long nr_pages,
1509 : : unsigned int gup_flags, struct page **pages,
1510 : : struct vm_area_struct **vmas);
1511 : : long get_user_pages_locked(unsigned long start, unsigned long nr_pages,
1512 : : unsigned int gup_flags, struct page **pages, int *locked);
1513 : : long get_user_pages_unlocked(unsigned long start, unsigned long nr_pages,
1514 : : struct page **pages, unsigned int gup_flags);
1515 : :
1516 : : int get_user_pages_fast(unsigned long start, int nr_pages,
1517 : : unsigned int gup_flags, struct page **pages);
1518 : : int pin_user_pages_fast(unsigned long start, int nr_pages,
1519 : : unsigned int gup_flags, struct page **pages);
1520 : :
1521 : : int account_locked_vm(struct mm_struct *mm, unsigned long pages, bool inc);
1522 : : int __account_locked_vm(struct mm_struct *mm, unsigned long pages, bool inc,
1523 : : struct task_struct *task, bool bypass_rlim);
1524 : :
1525 : : /* Container for pinned pfns / pages */
1526 : : struct frame_vector {
1527 : : unsigned int nr_allocated; /* Number of frames we have space for */
1528 : : unsigned int nr_frames; /* Number of frames stored in ptrs array */
1529 : : bool got_ref; /* Did we pin pages by getting page ref? */
1530 : : bool is_pfns; /* Does array contain pages or pfns? */
1531 : : void *ptrs[0]; /* Array of pinned pfns / pages. Use
1532 : : * pfns_vector_pages() or pfns_vector_pfns()
1533 : : * for access */
1534 : : };
1535 : :
1536 : : struct frame_vector *frame_vector_create(unsigned int nr_frames);
1537 : : void frame_vector_destroy(struct frame_vector *vec);
1538 : : int get_vaddr_frames(unsigned long start, unsigned int nr_pfns,
1539 : : unsigned int gup_flags, struct frame_vector *vec);
1540 : : void put_vaddr_frames(struct frame_vector *vec);
1541 : : int frame_vector_to_pages(struct frame_vector *vec);
1542 : : void frame_vector_to_pfns(struct frame_vector *vec);
1543 : :
1544 : : static inline unsigned int frame_vector_count(struct frame_vector *vec)
1545 : : {
1546 : : return vec->nr_frames;
1547 : : }
1548 : :
1549 : : static inline struct page **frame_vector_pages(struct frame_vector *vec)
1550 : : {
1551 : : if (vec->is_pfns) {
1552 : : int err = frame_vector_to_pages(vec);
1553 : :
1554 : : if (err)
1555 : : return ERR_PTR(err);
1556 : : }
1557 : : return (struct page **)(vec->ptrs);
1558 : : }
1559 : :
1560 : : static inline unsigned long *frame_vector_pfns(struct frame_vector *vec)
1561 : : {
1562 : : if (!vec->is_pfns)
1563 : : frame_vector_to_pfns(vec);
1564 : : return (unsigned long *)(vec->ptrs);
1565 : : }
1566 : :
1567 : : struct kvec;
1568 : : int get_kernel_pages(const struct kvec *iov, int nr_pages, int write,
1569 : : struct page **pages);
1570 : : int get_kernel_page(unsigned long start, int write, struct page **pages);
1571 : : struct page *get_dump_page(unsigned long addr);
1572 : :
1573 : : extern int try_to_release_page(struct page * page, gfp_t gfp_mask);
1574 : : extern void do_invalidatepage(struct page *page, unsigned int offset,
1575 : : unsigned int length);
1576 : :
1577 : : void __set_page_dirty(struct page *, struct address_space *, int warn);
1578 : : int __set_page_dirty_nobuffers(struct page *page);
1579 : : int __set_page_dirty_no_writeback(struct page *page);
1580 : : int redirty_page_for_writepage(struct writeback_control *wbc,
1581 : : struct page *page);
1582 : : void account_page_dirtied(struct page *page, struct address_space *mapping);
1583 : : void account_page_cleaned(struct page *page, struct address_space *mapping,
1584 : : struct bdi_writeback *wb);
1585 : : int set_page_dirty(struct page *page);
1586 : : int set_page_dirty_lock(struct page *page);
1587 : : void __cancel_dirty_page(struct page *page);
1588 : 7410 : static inline void cancel_dirty_page(struct page *page)
1589 : : {
1590 : : /* Avoid atomic ops, locking, etc. when not actually needed. */
1591 [ + + - + ]: 14820 : if (PageDirty(page))
1592 : 442 : __cancel_dirty_page(page);
1593 : 7410 : }
1594 : : int clear_page_dirty_for_io(struct page *page);
1595 : :
1596 : : int get_cmdline(struct task_struct *task, char *buffer, int buflen);
1597 : :
1598 : : extern unsigned long move_page_tables(struct vm_area_struct *vma,
1599 : : unsigned long old_addr, struct vm_area_struct *new_vma,
1600 : : unsigned long new_addr, unsigned long len,
1601 : : bool need_rmap_locks);
1602 : : extern unsigned long change_protection(struct vm_area_struct *vma, unsigned long start,
1603 : : unsigned long end, pgprot_t newprot,
1604 : : int dirty_accountable, int prot_numa);
1605 : : extern int mprotect_fixup(struct vm_area_struct *vma,
1606 : : struct vm_area_struct **pprev, unsigned long start,
1607 : : unsigned long end, unsigned long newflags);
1608 : :
1609 : : /*
1610 : : * doesn't attempt to fault and will return short.
1611 : : */
1612 : : int __get_user_pages_fast(unsigned long start, int nr_pages, int write,
1613 : : struct page **pages);
1614 : : /*
1615 : : * per-process(per-mm_struct) statistics.
1616 : : */
1617 : 594917 : static inline unsigned long get_mm_counter(struct mm_struct *mm, int member)
1618 : : {
1619 : 594056 : long val = atomic_long_read(&mm->rss_stat.count[member]);
1620 : :
1621 : : #ifdef SPLIT_RSS_COUNTING
1622 : : /*
1623 : : * counter is updated in asynchronous manner and may go to minus.
1624 : : * But it's never be expected number for users.
1625 : : */
1626 : 594917 : if (val < 0)
1627 : : val = 0;
1628 : : #endif
1629 : 198210 : return (unsigned long)val;
1630 : : }
1631 : :
1632 : : void mm_trace_rss_stat(struct mm_struct *mm, int member, long count);
1633 : :
1634 : 1165371 : static inline void add_mm_counter(struct mm_struct *mm, int member, long value)
1635 : : {
1636 : 1165371 : long count = atomic_long_add_return(value, &mm->rss_stat.count[member]);
1637 : :
1638 : 1165371 : mm_trace_rss_stat(mm, member, count);
1639 : 1011025 : }
1640 : :
1641 : 0 : static inline void inc_mm_counter(struct mm_struct *mm, int member)
1642 : : {
1643 : 0 : long count = atomic_long_inc_return(&mm->rss_stat.count[member]);
1644 : :
1645 : 0 : mm_trace_rss_stat(mm, member, count);
1646 : 0 : }
1647 : :
1648 : 0 : static inline void dec_mm_counter(struct mm_struct *mm, int member)
1649 : : {
1650 : 0 : long count = atomic_long_dec_return(&mm->rss_stat.count[member]);
1651 : :
1652 : 0 : mm_trace_rss_stat(mm, member, count);
1653 : 0 : }
1654 : :
1655 : : /* Optimized variant when page is already known not to be PageAnon */
1656 : 14008390 : static inline int mm_counter_file(struct page *page)
1657 : : {
1658 [ + + ]: 28016800 : if (PageSwapBacked(page))
1659 : 1760 : return MM_SHMEMPAGES;
1660 : : return MM_FILEPAGES;
1661 : : }
1662 : :
1663 : 8602088 : static inline int mm_counter(struct page *page)
1664 : : {
1665 [ + + ]: 8602088 : if (PageAnon(page))
1666 : : return MM_ANONPAGES;
1667 : 6766845 : return mm_counter_file(page);
1668 : : }
1669 : :
1670 : 197923 : static inline unsigned long get_mm_rss(struct mm_struct *mm)
1671 : : {
1672 : 395846 : return get_mm_counter(mm, MM_FILEPAGES) +
1673 : 197923 : get_mm_counter(mm, MM_ANONPAGES) +
1674 : : get_mm_counter(mm, MM_SHMEMPAGES);
1675 : : }
1676 : :
1677 : 16224 : static inline unsigned long get_mm_hiwater_rss(struct mm_struct *mm)
1678 : : {
1679 : 32448 : return max(mm->hiwater_rss, get_mm_rss(mm));
1680 : : }
1681 : :
1682 : 0 : static inline unsigned long get_mm_hiwater_vm(struct mm_struct *mm)
1683 : : {
1684 : 0 : return max(mm->hiwater_vm, mm->total_vm);
1685 : : }
1686 : :
1687 : 158034 : static inline void update_hiwater_rss(struct mm_struct *mm)
1688 : : {
1689 : 158034 : unsigned long _rss = get_mm_rss(mm);
1690 : :
1691 [ + + # # ]: 158034 : if ((mm)->hiwater_rss < _rss)
1692 : 71586 : (mm)->hiwater_rss = _rss;
1693 : : }
1694 : :
1695 : 158034 : static inline void update_hiwater_vm(struct mm_struct *mm)
1696 : : {
1697 [ + + ]: 158034 : if (mm->hiwater_vm < mm->total_vm)
1698 : 65573 : mm->hiwater_vm = mm->total_vm;
1699 : : }
1700 : :
1701 : 0 : static inline void reset_mm_hiwater_rss(struct mm_struct *mm)
1702 : : {
1703 : 0 : mm->hiwater_rss = get_mm_rss(mm);
1704 : : }
1705 : :
1706 : 16224 : static inline void setmax_mm_hiwater_rss(unsigned long *maxrss,
1707 : : struct mm_struct *mm)
1708 : : {
1709 : 16224 : unsigned long hiwater_rss = get_mm_hiwater_rss(mm);
1710 : :
1711 [ + + ]: 16224 : if (*maxrss < hiwater_rss)
1712 : 15980 : *maxrss = hiwater_rss;
1713 : : }
1714 : :
1715 : : #if defined(SPLIT_RSS_COUNTING)
1716 : : void sync_mm_rss(struct mm_struct *mm);
1717 : : #else
1718 : : static inline void sync_mm_rss(struct mm_struct *mm)
1719 : : {
1720 : : }
1721 : : #endif
1722 : :
1723 : : #ifndef CONFIG_ARCH_HAS_PTE_DEVMAP
1724 : : static inline int pte_devmap(pte_t pte)
1725 : : {
1726 : : return 0;
1727 : : }
1728 : : #endif
1729 : :
1730 : : int vma_wants_writenotify(struct vm_area_struct *vma, pgprot_t vm_page_prot);
1731 : :
1732 : : extern pte_t *__get_locked_pte(struct mm_struct *mm, unsigned long addr,
1733 : : spinlock_t **ptl);
1734 : 130284 : static inline pte_t *get_locked_pte(struct mm_struct *mm, unsigned long addr,
1735 : : spinlock_t **ptl)
1736 : : {
1737 : 130284 : pte_t *ptep;
1738 : 130284 : __cond_lock(*ptl, ptep = __get_locked_pte(mm, addr, ptl));
1739 [ + + ]: 130284 : return ptep;
1740 : : }
1741 : :
1742 : : #ifdef __PAGETABLE_P4D_FOLDED
1743 : : static inline int __p4d_alloc(struct mm_struct *mm, pgd_t *pgd,
1744 : : unsigned long address)
1745 : : {
1746 : : return 0;
1747 : : }
1748 : : #else
1749 : : int __p4d_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address);
1750 : : #endif
1751 : :
1752 : : #if defined(__PAGETABLE_PUD_FOLDED) || !defined(CONFIG_MMU)
1753 : : static inline int __pud_alloc(struct mm_struct *mm, p4d_t *p4d,
1754 : : unsigned long address)
1755 : : {
1756 : : return 0;
1757 : : }
1758 : : static inline void mm_inc_nr_puds(struct mm_struct *mm) {}
1759 : : static inline void mm_dec_nr_puds(struct mm_struct *mm) {}
1760 : :
1761 : : #else
1762 : : int __pud_alloc(struct mm_struct *mm, p4d_t *p4d, unsigned long address);
1763 : :
1764 : 54755 : static inline void mm_inc_nr_puds(struct mm_struct *mm)
1765 : : {
1766 : 54755 : if (mm_pud_folded(mm))
1767 : : return;
1768 : 54755 : atomic_long_add(PTRS_PER_PUD * sizeof(pud_t), &mm->pgtables_bytes);
1769 : : }
1770 : :
1771 : 53952 : static inline void mm_dec_nr_puds(struct mm_struct *mm)
1772 : : {
1773 : 53952 : if (mm_pud_folded(mm))
1774 : : return;
1775 : 53952 : atomic_long_sub(PTRS_PER_PUD * sizeof(pud_t), &mm->pgtables_bytes);
1776 : : }
1777 : : #endif
1778 : :
1779 : : #if defined(__PAGETABLE_PMD_FOLDED) || !defined(CONFIG_MMU)
1780 : : static inline int __pmd_alloc(struct mm_struct *mm, pud_t *pud,
1781 : : unsigned long address)
1782 : : {
1783 : : return 0;
1784 : : }
1785 : :
1786 : : static inline void mm_inc_nr_pmds(struct mm_struct *mm) {}
1787 : : static inline void mm_dec_nr_pmds(struct mm_struct *mm) {}
1788 : :
1789 : : #else
1790 : : int __pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address);
1791 : :
1792 : 74540 : static inline void mm_inc_nr_pmds(struct mm_struct *mm)
1793 : : {
1794 : 74540 : if (mm_pmd_folded(mm))
1795 : : return;
1796 : 74540 : atomic_long_add(PTRS_PER_PMD * sizeof(pmd_t), &mm->pgtables_bytes);
1797 : : }
1798 : :
1799 : 73594 : static inline void mm_dec_nr_pmds(struct mm_struct *mm)
1800 : : {
1801 : 73594 : if (mm_pmd_folded(mm))
1802 : : return;
1803 : 73594 : atomic_long_sub(PTRS_PER_PMD * sizeof(pmd_t), &mm->pgtables_bytes);
1804 : : }
1805 : : #endif
1806 : :
1807 : : #ifdef CONFIG_MMU
1808 : : static inline void mm_pgtables_bytes_init(struct mm_struct *mm)
1809 : : {
1810 : : atomic_long_set(&mm->pgtables_bytes, 0);
1811 : : }
1812 : :
1813 : 9283 : static inline unsigned long mm_pgtables_bytes(const struct mm_struct *mm)
1814 : : {
1815 : 9283 : return atomic_long_read(&mm->pgtables_bytes);
1816 : : }
1817 : :
1818 : 157775 : static inline void mm_inc_nr_ptes(struct mm_struct *mm)
1819 : : {
1820 : 157775 : atomic_long_add(PTRS_PER_PTE * sizeof(pte_t), &mm->pgtables_bytes);
1821 : 157775 : }
1822 : :
1823 : 154465 : static inline void mm_dec_nr_ptes(struct mm_struct *mm)
1824 : : {
1825 : 154465 : atomic_long_sub(PTRS_PER_PTE * sizeof(pte_t), &mm->pgtables_bytes);
1826 : 154465 : }
1827 : : #else
1828 : :
1829 : : static inline void mm_pgtables_bytes_init(struct mm_struct *mm) {}
1830 : : static inline unsigned long mm_pgtables_bytes(const struct mm_struct *mm)
1831 : : {
1832 : : return 0;
1833 : : }
1834 : :
1835 : : static inline void mm_inc_nr_ptes(struct mm_struct *mm) {}
1836 : : static inline void mm_dec_nr_ptes(struct mm_struct *mm) {}
1837 : : #endif
1838 : :
1839 : : int __pte_alloc(struct mm_struct *mm, pmd_t *pmd);
1840 : : int __pte_alloc_kernel(pmd_t *pmd);
1841 : :
1842 : : #if defined(CONFIG_MMU)
1843 : :
1844 : : /*
1845 : : * The following ifdef needed to get the 5level-fixup.h header to work.
1846 : : * Remove it when 5level-fixup.h has been removed.
1847 : : */
1848 : : #ifndef __ARCH_HAS_5LEVEL_HACK
1849 : 1918253 : static inline p4d_t *p4d_alloc(struct mm_struct *mm, pgd_t *pgd,
1850 : : unsigned long address)
1851 : : {
1852 [ - - ]: 1918253 : return (unlikely(pgd_none(*pgd)) && __p4d_alloc(mm, pgd, address)) ?
1853 [ - + ]: 1918253 : NULL : p4d_offset(pgd, address);
1854 : : }
1855 : :
1856 : 1918227 : static inline pud_t *pud_alloc(struct mm_struct *mm, p4d_t *p4d,
1857 : : unsigned long address)
1858 : : {
1859 [ + + + - ]: 1918227 : return (unlikely(p4d_none(*p4d)) && __pud_alloc(mm, p4d, address)) ?
1860 [ + + ]: 1918227 : NULL : pud_offset(p4d, address);
1861 : : }
1862 : : #endif /* !__ARCH_HAS_5LEVEL_HACK */
1863 : :
1864 : 1918214 : static inline pmd_t *pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address)
1865 : : {
1866 [ + + + - ]: 1918214 : return (unlikely(pud_none(*pud)) && __pmd_alloc(mm, pud, address))?
1867 [ + + ]: 1918214 : NULL: pmd_offset(pud, address);
1868 : : }
1869 : : #endif /* CONFIG_MMU */
1870 : :
1871 : : #if USE_SPLIT_PTE_PTLOCKS
1872 : : #if ALLOC_SPLIT_PTLOCKS
1873 : : void __init ptlock_cache_init(void);
1874 : : extern bool ptlock_alloc(struct page *page);
1875 : : extern void ptlock_free(struct page *page);
1876 : :
1877 : : static inline spinlock_t *ptlock_ptr(struct page *page)
1878 : : {
1879 : : return page->ptl;
1880 : : }
1881 : : #else /* ALLOC_SPLIT_PTLOCKS */
1882 : 13 : static inline void ptlock_cache_init(void)
1883 : : {
1884 : 13 : }
1885 : :
1886 : 0 : static inline bool ptlock_alloc(struct page *page)
1887 : : {
1888 : 0 : return true;
1889 : : }
1890 : :
1891 : 1388174 : static inline void ptlock_free(struct page *page)
1892 : : {
1893 : 1388174 : }
1894 : :
1895 : 3639557 : static inline spinlock_t *ptlock_ptr(struct page *page)
1896 : : {
1897 : 3639557 : return &page->ptl;
1898 : : }
1899 : : #endif /* ALLOC_SPLIT_PTLOCKS */
1900 : :
1901 : 3639557 : static inline spinlock_t *pte_lockptr(struct mm_struct *mm, pmd_t *pmd)
1902 : : {
1903 [ + - + - : 7279114 : return ptlock_ptr(pmd_page(*pmd));
+ + + - +
- + - - -
- - - - -
- + - - -
- - - - -
- + - + -
+ - + - ]
1904 : : }
1905 : :
1906 : 0 : static inline bool ptlock_init(struct page *page)
1907 : : {
1908 : : /*
1909 : : * prep_new_page() initialize page->private (and therefore page->ptl)
1910 : : * with 0. Make sure nobody took it in use in between.
1911 : : *
1912 : : * It can happen if arch try to use slab for page table allocation:
1913 : : * slab code uses page->slab_cache, which share storage with page->ptl.
1914 : : */
1915 : 0 : VM_BUG_ON_PAGE(*(unsigned long *)&page->ptl, page);
1916 : 0 : if (!ptlock_alloc(page))
1917 : : return false;
1918 : 0 : spin_lock_init(ptlock_ptr(page));
1919 : 0 : return true;
1920 : : }
1921 : :
1922 : : #else /* !USE_SPLIT_PTE_PTLOCKS */
1923 : : /*
1924 : : * We use mm->page_table_lock to guard all pagetable pages of the mm.
1925 : : */
1926 : : static inline spinlock_t *pte_lockptr(struct mm_struct *mm, pmd_t *pmd)
1927 : : {
1928 : : return &mm->page_table_lock;
1929 : : }
1930 : : static inline void ptlock_cache_init(void) {}
1931 : : static inline bool ptlock_init(struct page *page) { return true; }
1932 : : static inline void ptlock_free(struct page *page) {}
1933 : : #endif /* USE_SPLIT_PTE_PTLOCKS */
1934 : :
1935 : 13 : static inline void pgtable_init(void)
1936 : : {
1937 : 13 : ptlock_cache_init();
1938 : 13 : pgtable_cache_init();
1939 : : }
1940 : :
1941 : : static inline bool pgtable_pte_page_ctor(struct page *page)
1942 : : {
1943 : : if (!ptlock_init(page))
1944 : : return false;
1945 : : __SetPageTable(page);
1946 : : inc_zone_page_state(page, NR_PAGETABLE);
1947 : : return true;
1948 : : }
1949 : :
1950 : 907674 : static inline void pgtable_pte_page_dtor(struct page *page)
1951 : : {
1952 : 907674 : ptlock_free(page);
1953 : 907674 : __ClearPageTable(page);
1954 : 907674 : dec_zone_page_state(page, NR_PAGETABLE);
1955 : : }
1956 : :
1957 : : #define pte_offset_map_lock(mm, pmd, address, ptlp) \
1958 : : ({ \
1959 : : spinlock_t *__ptl = pte_lockptr(mm, pmd); \
1960 : : pte_t *__pte = pte_offset_map(pmd, address); \
1961 : : *(ptlp) = __ptl; \
1962 : : spin_lock(__ptl); \
1963 : : __pte; \
1964 : : })
1965 : :
1966 : : #define pte_unmap_unlock(pte, ptl) do { \
1967 : : spin_unlock(ptl); \
1968 : : pte_unmap(pte); \
1969 : : } while (0)
1970 : :
1971 : : #define pte_alloc(mm, pmd) (unlikely(pmd_none(*(pmd))) && __pte_alloc(mm, pmd))
1972 : :
1973 : : #define pte_alloc_map(mm, pmd, address) \
1974 : : (pte_alloc(mm, pmd) ? NULL : pte_offset_map(pmd, address))
1975 : :
1976 : : #define pte_alloc_map_lock(mm, pmd, address, ptlp) \
1977 : : (pte_alloc(mm, pmd) ? \
1978 : : NULL : pte_offset_map_lock(mm, pmd, address, ptlp))
1979 : :
1980 : : #define pte_alloc_kernel(pmd, address) \
1981 : : ((unlikely(pmd_none(*(pmd))) && __pte_alloc_kernel(pmd))? \
1982 : : NULL: pte_offset_kernel(pmd, address))
1983 : :
1984 : : #if USE_SPLIT_PMD_PTLOCKS
1985 : :
1986 : 0 : static struct page *pmd_to_page(pmd_t *pmd)
1987 : : {
1988 : 0 : unsigned long mask = ~(PTRS_PER_PMD * sizeof(pmd_t) - 1);
1989 [ # # ]: 0 : return virt_to_page((void *)((unsigned long) pmd & mask));
1990 : : }
1991 : :
1992 : 0 : static inline spinlock_t *pmd_lockptr(struct mm_struct *mm, pmd_t *pmd)
1993 : : {
1994 [ # # # # : 0 : return ptlock_ptr(pmd_to_page(pmd));
# # # # ]
1995 : : }
1996 : :
1997 : 0 : static inline bool pgtable_pmd_page_ctor(struct page *page)
1998 : : {
1999 : : #ifdef CONFIG_TRANSPARENT_HUGEPAGE
2000 : : page->pmd_huge_pte = NULL;
2001 : : #endif
2002 : 0 : return ptlock_init(page);
2003 : : }
2004 : :
2005 : 480500 : static inline void pgtable_pmd_page_dtor(struct page *page)
2006 : : {
2007 : : #ifdef CONFIG_TRANSPARENT_HUGEPAGE
2008 : : VM_BUG_ON_PAGE(page->pmd_huge_pte, page);
2009 : : #endif
2010 : 480500 : ptlock_free(page);
2011 : : }
2012 : :
2013 : : #define pmd_huge_pte(mm, pmd) (pmd_to_page(pmd)->pmd_huge_pte)
2014 : :
2015 : : #else
2016 : :
2017 : : static inline spinlock_t *pmd_lockptr(struct mm_struct *mm, pmd_t *pmd)
2018 : : {
2019 : : return &mm->page_table_lock;
2020 : : }
2021 : :
2022 : : static inline bool pgtable_pmd_page_ctor(struct page *page) { return true; }
2023 : : static inline void pgtable_pmd_page_dtor(struct page *page) {}
2024 : :
2025 : : #define pmd_huge_pte(mm, pmd) ((mm)->pmd_huge_pte)
2026 : :
2027 : : #endif
2028 : :
2029 : : static inline spinlock_t *pmd_lock(struct mm_struct *mm, pmd_t *pmd)
2030 : : {
2031 : : spinlock_t *ptl = pmd_lockptr(mm, pmd);
2032 : : spin_lock(ptl);
2033 : : return ptl;
2034 : : }
2035 : :
2036 : : /*
2037 : : * No scalability reason to split PUD locks yet, but follow the same pattern
2038 : : * as the PMD locks to make it easier if we decide to. The VM should not be
2039 : : * considered ready to switch to split PUD locks yet; there may be places
2040 : : * which need to be converted from page_table_lock.
2041 : : */
2042 : 74540 : static inline spinlock_t *pud_lockptr(struct mm_struct *mm, pud_t *pud)
2043 : : {
2044 : 74540 : return &mm->page_table_lock;
2045 : : }
2046 : :
2047 : 74540 : static inline spinlock_t *pud_lock(struct mm_struct *mm, pud_t *pud)
2048 : : {
2049 : 74540 : spinlock_t *ptl = pud_lockptr(mm, pud);
2050 : :
2051 : 74540 : spin_lock(ptl);
2052 [ + - ]: 74540 : return ptl;
2053 : : }
2054 : :
2055 : : extern void __init pagecache_init(void);
2056 : : extern void free_area_init(unsigned long * zones_size);
2057 : : extern void __init free_area_init_node(int nid, unsigned long * zones_size,
2058 : : unsigned long zone_start_pfn, unsigned long *zholes_size);
2059 : : extern void free_initmem(void);
2060 : :
2061 : : /*
2062 : : * Free reserved pages within range [PAGE_ALIGN(start), end & PAGE_MASK)
2063 : : * into the buddy system. The freed pages will be poisoned with pattern
2064 : : * "poison" if it's within range [0, UCHAR_MAX].
2065 : : * Return pages freed into the buddy system.
2066 : : */
2067 : : extern unsigned long free_reserved_area(void *start, void *end,
2068 : : int poison, const char *s);
2069 : :
2070 : : #ifdef CONFIG_HIGHMEM
2071 : : /*
2072 : : * Free a highmem page into the buddy system, adjusting totalhigh_pages
2073 : : * and totalram_pages.
2074 : : */
2075 : : extern void free_highmem_page(struct page *page);
2076 : : #endif
2077 : :
2078 : : extern void adjust_managed_page_count(struct page *page, long count);
2079 : : extern void mem_init_print_info(const char *str);
2080 : :
2081 : : extern void reserve_bootmem_region(phys_addr_t start, phys_addr_t end);
2082 : :
2083 : : /* Free the reserved page into the buddy system, so it gets managed. */
2084 : 18369 : static inline void __free_reserved_page(struct page *page)
2085 : : {
2086 : 18369 : ClearPageReserved(page);
2087 : 18369 : init_page_count(page);
2088 : 18369 : __free_page(page);
2089 : 18369 : }
2090 : :
2091 : 18369 : static inline void free_reserved_page(struct page *page)
2092 : : {
2093 : 18369 : __free_reserved_page(page);
2094 : 18369 : adjust_managed_page_count(page, 1);
2095 : : }
2096 : :
2097 : : static inline void mark_page_reserved(struct page *page)
2098 : : {
2099 : : SetPageReserved(page);
2100 : : adjust_managed_page_count(page, -1);
2101 : : }
2102 : :
2103 : : /*
2104 : : * Default method to free all the __init memory into the buddy system.
2105 : : * The freed pages will be poisoned with pattern "poison" if it's within
2106 : : * range [0, UCHAR_MAX].
2107 : : * Return pages freed into the buddy system.
2108 : : */
2109 : 0 : static inline unsigned long free_initmem_default(int poison)
2110 : : {
2111 : 0 : extern char __init_begin[], __init_end[];
2112 : :
2113 : 0 : return free_reserved_area(&__init_begin, &__init_end,
2114 : : poison, "unused kernel");
2115 : : }
2116 : :
2117 : 13 : static inline unsigned long get_num_physpages(void)
2118 : : {
2119 : 13 : int nid;
2120 : 13 : unsigned long phys_pages = 0;
2121 : :
2122 [ + + ]: 39 : for_each_online_node(nid)
2123 : 13 : phys_pages += node_present_pages(nid);
2124 : :
2125 : 13 : return phys_pages;
2126 : : }
2127 : :
2128 : : #ifdef CONFIG_HAVE_MEMBLOCK_NODE_MAP
2129 : : /*
2130 : : * With CONFIG_HAVE_MEMBLOCK_NODE_MAP set, an architecture may initialise its
2131 : : * zones, allocate the backing mem_map and account for memory holes in a more
2132 : : * architecture independent manner. This is a substitute for creating the
2133 : : * zone_sizes[] and zholes_size[] arrays and passing them to
2134 : : * free_area_init_node()
2135 : : *
2136 : : * An architecture is expected to register range of page frames backed by
2137 : : * physical memory with memblock_add[_node]() before calling
2138 : : * free_area_init_nodes() passing in the PFN each zone ends at. At a basic
2139 : : * usage, an architecture is expected to do something like
2140 : : *
2141 : : * unsigned long max_zone_pfns[MAX_NR_ZONES] = {max_dma, max_normal_pfn,
2142 : : * max_highmem_pfn};
2143 : : * for_each_valid_physical_page_range()
2144 : : * memblock_add_node(base, size, nid)
2145 : : * free_area_init_nodes(max_zone_pfns);
2146 : : *
2147 : : * free_bootmem_with_active_regions() calls free_bootmem_node() for each
2148 : : * registered physical page range. Similarly
2149 : : * sparse_memory_present_with_active_regions() calls memory_present() for
2150 : : * each range when SPARSEMEM is enabled.
2151 : : *
2152 : : * See mm/page_alloc.c for more information on each function exposed by
2153 : : * CONFIG_HAVE_MEMBLOCK_NODE_MAP.
2154 : : */
2155 : : extern void free_area_init_nodes(unsigned long *max_zone_pfn);
2156 : : unsigned long node_map_pfn_alignment(void);
2157 : : unsigned long __absent_pages_in_range(int nid, unsigned long start_pfn,
2158 : : unsigned long end_pfn);
2159 : : extern unsigned long absent_pages_in_range(unsigned long start_pfn,
2160 : : unsigned long end_pfn);
2161 : : extern void get_pfn_range_for_nid(unsigned int nid,
2162 : : unsigned long *start_pfn, unsigned long *end_pfn);
2163 : : extern unsigned long find_min_pfn_with_active_regions(void);
2164 : : extern void free_bootmem_with_active_regions(int nid,
2165 : : unsigned long max_low_pfn);
2166 : : extern void sparse_memory_present_with_active_regions(int nid);
2167 : :
2168 : : #endif /* CONFIG_HAVE_MEMBLOCK_NODE_MAP */
2169 : :
2170 : : #if !defined(CONFIG_HAVE_MEMBLOCK_NODE_MAP) && \
2171 : : !defined(CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID)
2172 : : static inline int __early_pfn_to_nid(unsigned long pfn,
2173 : : struct mminit_pfnnid_cache *state)
2174 : : {
2175 : : return 0;
2176 : : }
2177 : : #else
2178 : : /* please see mm/page_alloc.c */
2179 : : extern int __meminit early_pfn_to_nid(unsigned long pfn);
2180 : : /* there is a per-arch backend function. */
2181 : : extern int __meminit __early_pfn_to_nid(unsigned long pfn,
2182 : : struct mminit_pfnnid_cache *state);
2183 : : #endif
2184 : :
2185 : : extern void set_dma_reserve(unsigned long new_dma_reserve);
2186 : : extern void memmap_init_zone(unsigned long, int, unsigned long, unsigned long,
2187 : : enum memmap_context, struct vmem_altmap *);
2188 : : extern void setup_per_zone_wmarks(void);
2189 : : extern int __meminit init_per_zone_wmark_min(void);
2190 : : extern void mem_init(void);
2191 : : extern void __init mmap_init(void);
2192 : : extern void show_mem(unsigned int flags, nodemask_t *nodemask);
2193 : : extern long si_mem_available(void);
2194 : : extern void si_meminfo(struct sysinfo * val);
2195 : : extern void si_meminfo_node(struct sysinfo *val, int nid);
2196 : : #ifdef __HAVE_ARCH_RESERVED_KERNEL_PAGES
2197 : : extern unsigned long arch_reserved_kernel_pages(void);
2198 : : #endif
2199 : :
2200 : : extern __printf(3, 4)
2201 : : void warn_alloc(gfp_t gfp_mask, nodemask_t *nodemask, const char *fmt, ...);
2202 : :
2203 : : extern void setup_per_cpu_pageset(void);
2204 : :
2205 : : /* page_alloc.c */
2206 : : extern int min_free_kbytes;
2207 : : extern int watermark_boost_factor;
2208 : : extern int watermark_scale_factor;
2209 : :
2210 : : /* nommu.c */
2211 : : extern atomic_long_t mmap_pages_allocated;
2212 : : extern int nommu_shrink_inode_mappings(struct inode *, size_t, size_t);
2213 : :
2214 : : /* interval_tree.c */
2215 : : void vma_interval_tree_insert(struct vm_area_struct *node,
2216 : : struct rb_root_cached *root);
2217 : : void vma_interval_tree_insert_after(struct vm_area_struct *node,
2218 : : struct vm_area_struct *prev,
2219 : : struct rb_root_cached *root);
2220 : : void vma_interval_tree_remove(struct vm_area_struct *node,
2221 : : struct rb_root_cached *root);
2222 : : struct vm_area_struct *vma_interval_tree_iter_first(struct rb_root_cached *root,
2223 : : unsigned long start, unsigned long last);
2224 : : struct vm_area_struct *vma_interval_tree_iter_next(struct vm_area_struct *node,
2225 : : unsigned long start, unsigned long last);
2226 : :
2227 : : #define vma_interval_tree_foreach(vma, root, start, last) \
2228 : : for (vma = vma_interval_tree_iter_first(root, start, last); \
2229 : : vma; vma = vma_interval_tree_iter_next(vma, start, last))
2230 : :
2231 : : void anon_vma_interval_tree_insert(struct anon_vma_chain *node,
2232 : : struct rb_root_cached *root);
2233 : : void anon_vma_interval_tree_remove(struct anon_vma_chain *node,
2234 : : struct rb_root_cached *root);
2235 : : struct anon_vma_chain *
2236 : : anon_vma_interval_tree_iter_first(struct rb_root_cached *root,
2237 : : unsigned long start, unsigned long last);
2238 : : struct anon_vma_chain *anon_vma_interval_tree_iter_next(
2239 : : struct anon_vma_chain *node, unsigned long start, unsigned long last);
2240 : : #ifdef CONFIG_DEBUG_VM_RB
2241 : : void anon_vma_interval_tree_verify(struct anon_vma_chain *node);
2242 : : #endif
2243 : :
2244 : : #define anon_vma_interval_tree_foreach(avc, root, start, last) \
2245 : : for (avc = anon_vma_interval_tree_iter_first(root, start, last); \
2246 : : avc; avc = anon_vma_interval_tree_iter_next(avc, start, last))
2247 : :
2248 : : /* mmap.c */
2249 : : extern int __vm_enough_memory(struct mm_struct *mm, long pages, int cap_sys_admin);
2250 : : extern int __vma_adjust(struct vm_area_struct *vma, unsigned long start,
2251 : : unsigned long end, pgoff_t pgoff, struct vm_area_struct *insert,
2252 : : struct vm_area_struct *expand);
2253 : 261980 : static inline int vma_adjust(struct vm_area_struct *vma, unsigned long start,
2254 : : unsigned long end, pgoff_t pgoff, struct vm_area_struct *insert)
2255 : : {
2256 : 261980 : return __vma_adjust(vma, start, end, pgoff, insert, NULL);
2257 : : }
2258 : : extern struct vm_area_struct *vma_merge(struct mm_struct *,
2259 : : struct vm_area_struct *prev, unsigned long addr, unsigned long end,
2260 : : unsigned long vm_flags, struct anon_vma *, struct file *, pgoff_t,
2261 : : struct mempolicy *, struct vm_userfaultfd_ctx);
2262 : : extern struct anon_vma *find_mergeable_anon_vma(struct vm_area_struct *);
2263 : : extern int __split_vma(struct mm_struct *, struct vm_area_struct *,
2264 : : unsigned long addr, int new_below);
2265 : : extern int split_vma(struct mm_struct *, struct vm_area_struct *,
2266 : : unsigned long addr, int new_below);
2267 : : extern int insert_vm_struct(struct mm_struct *, struct vm_area_struct *);
2268 : : extern void __vma_link_rb(struct mm_struct *, struct vm_area_struct *,
2269 : : struct rb_node **, struct rb_node *);
2270 : : extern void unlink_file_vma(struct vm_area_struct *);
2271 : : extern struct vm_area_struct *copy_vma(struct vm_area_struct **,
2272 : : unsigned long addr, unsigned long len, pgoff_t pgoff,
2273 : : bool *need_rmap_locks);
2274 : : extern void exit_mmap(struct mm_struct *);
2275 : :
2276 : 12381 : static inline int check_data_rlimit(unsigned long rlim,
2277 : : unsigned long new,
2278 : : unsigned long start,
2279 : : unsigned long end_data,
2280 : : unsigned long start_data)
2281 : : {
2282 [ - + ]: 12381 : if (rlim < RLIM_INFINITY) {
2283 [ # # ]: 0 : if (((new - start) + (end_data - start_data)) > rlim)
2284 : 0 : return -ENOSPC;
2285 : : }
2286 : :
2287 : : return 0;
2288 : : }
2289 : :
2290 : : extern int mm_take_all_locks(struct mm_struct *mm);
2291 : : extern void mm_drop_all_locks(struct mm_struct *mm);
2292 : :
2293 : : extern void set_mm_exe_file(struct mm_struct *mm, struct file *new_exe_file);
2294 : : extern struct file *get_mm_exe_file(struct mm_struct *mm);
2295 : : extern struct file *get_task_exe_file(struct task_struct *task);
2296 : :
2297 : : extern bool may_expand_vm(struct mm_struct *, vm_flags_t, unsigned long npages);
2298 : : extern void vm_stat_account(struct mm_struct *, vm_flags_t, long npages);
2299 : :
2300 : : extern bool vma_is_special_mapping(const struct vm_area_struct *vma,
2301 : : const struct vm_special_mapping *sm);
2302 : : extern struct vm_area_struct *_install_special_mapping(struct mm_struct *mm,
2303 : : unsigned long addr, unsigned long len,
2304 : : unsigned long flags,
2305 : : const struct vm_special_mapping *spec);
2306 : : /* This is an obsolete alternative to _install_special_mapping. */
2307 : : extern int install_special_mapping(struct mm_struct *mm,
2308 : : unsigned long addr, unsigned long len,
2309 : : unsigned long flags, struct page **pages);
2310 : :
2311 : : unsigned long randomize_stack_top(unsigned long stack_top);
2312 : :
2313 : : extern unsigned long get_unmapped_area(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
2314 : :
2315 : : extern unsigned long mmap_region(struct file *file, unsigned long addr,
2316 : : unsigned long len, vm_flags_t vm_flags, unsigned long pgoff,
2317 : : struct list_head *uf);
2318 : : extern unsigned long do_mmap(struct file *file, unsigned long addr,
2319 : : unsigned long len, unsigned long prot, unsigned long flags,
2320 : : vm_flags_t vm_flags, unsigned long pgoff, unsigned long *populate,
2321 : : struct list_head *uf);
2322 : : extern int __do_munmap(struct mm_struct *, unsigned long, size_t,
2323 : : struct list_head *uf, bool downgrade);
2324 : : extern int do_munmap(struct mm_struct *, unsigned long, size_t,
2325 : : struct list_head *uf);
2326 : : extern int do_madvise(unsigned long start, size_t len_in, int behavior);
2327 : :
2328 : : static inline unsigned long
2329 : 351677 : do_mmap_pgoff(struct file *file, unsigned long addr,
2330 : : unsigned long len, unsigned long prot, unsigned long flags,
2331 : : unsigned long pgoff, unsigned long *populate,
2332 : : struct list_head *uf)
2333 : : {
2334 : 351677 : return do_mmap(file, addr, len, prot, flags, 0, pgoff, populate, uf);
2335 : : }
2336 : :
2337 : : #ifdef CONFIG_MMU
2338 : : extern int __mm_populate(unsigned long addr, unsigned long len,
2339 : : int ignore_errors);
2340 : 0 : static inline void mm_populate(unsigned long addr, unsigned long len)
2341 : : {
2342 : : /* Ignore errors */
2343 : 0 : (void) __mm_populate(addr, len, 1);
2344 : 0 : }
2345 : : #else
2346 : : static inline void mm_populate(unsigned long addr, unsigned long len) {}
2347 : : #endif
2348 : :
2349 : : /* These take the mm semaphore themselves */
2350 : : extern int __must_check vm_brk(unsigned long, unsigned long);
2351 : : extern int __must_check vm_brk_flags(unsigned long, unsigned long, unsigned long);
2352 : : extern int vm_munmap(unsigned long, size_t);
2353 : : extern unsigned long __must_check vm_mmap(struct file *, unsigned long,
2354 : : unsigned long, unsigned long,
2355 : : unsigned long, unsigned long);
2356 : :
2357 : : struct vm_unmapped_area_info {
2358 : : #define VM_UNMAPPED_AREA_TOPDOWN 1
2359 : : unsigned long flags;
2360 : : unsigned long length;
2361 : : unsigned long low_limit;
2362 : : unsigned long high_limit;
2363 : : unsigned long align_mask;
2364 : : unsigned long align_offset;
2365 : : };
2366 : :
2367 : : extern unsigned long unmapped_area(struct vm_unmapped_area_info *info);
2368 : : extern unsigned long unmapped_area_topdown(struct vm_unmapped_area_info *info);
2369 : :
2370 : : /*
2371 : : * Search for an unmapped address range.
2372 : : *
2373 : : * We are looking for a range that:
2374 : : * - does not intersect with any VMA;
2375 : : * - is contained within the [low_limit, high_limit) interval;
2376 : : * - is at least the desired size.
2377 : : * - satisfies (begin_addr & align_mask) == (align_offset & align_mask)
2378 : : */
2379 : : static inline unsigned long
2380 : 767431 : vm_unmapped_area(struct vm_unmapped_area_info *info)
2381 : : {
2382 [ + - ]: 767431 : if (info->flags & VM_UNMAPPED_AREA_TOPDOWN)
2383 : 767431 : return unmapped_area_topdown(info);
2384 : : else
2385 : 0 : return unmapped_area(info);
2386 : : }
2387 : :
2388 : : /* truncate.c */
2389 : : extern void truncate_inode_pages(struct address_space *, loff_t);
2390 : : extern void truncate_inode_pages_range(struct address_space *,
2391 : : loff_t lstart, loff_t lend);
2392 : : extern void truncate_inode_pages_final(struct address_space *);
2393 : :
2394 : : /* generic vm_area_ops exported for stackable file systems */
2395 : : extern vm_fault_t filemap_fault(struct vm_fault *vmf);
2396 : : extern void filemap_map_pages(struct vm_fault *vmf,
2397 : : pgoff_t start_pgoff, pgoff_t end_pgoff);
2398 : : extern vm_fault_t filemap_page_mkwrite(struct vm_fault *vmf);
2399 : :
2400 : : /* mm/page-writeback.c */
2401 : : int __must_check write_one_page(struct page *page);
2402 : : void task_dirty_inc(struct task_struct *tsk);
2403 : :
2404 : : /* readahead.c */
2405 : : #define VM_READAHEAD_PAGES (SZ_128K / PAGE_SIZE)
2406 : :
2407 : : int force_page_cache_readahead(struct address_space *mapping, struct file *filp,
2408 : : pgoff_t offset, unsigned long nr_to_read);
2409 : :
2410 : : void page_cache_sync_readahead(struct address_space *mapping,
2411 : : struct file_ra_state *ra,
2412 : : struct file *filp,
2413 : : pgoff_t offset,
2414 : : unsigned long size);
2415 : :
2416 : : void page_cache_async_readahead(struct address_space *mapping,
2417 : : struct file_ra_state *ra,
2418 : : struct file *filp,
2419 : : struct page *pg,
2420 : : pgoff_t offset,
2421 : : unsigned long size);
2422 : :
2423 : : extern unsigned long stack_guard_gap;
2424 : : /* Generic expand stack which grows the stack according to GROWS{UP,DOWN} */
2425 : : extern int expand_stack(struct vm_area_struct *vma, unsigned long address);
2426 : :
2427 : : /* CONFIG_STACK_GROWSUP still needs to to grow downwards at some places */
2428 : : extern int expand_downwards(struct vm_area_struct *vma,
2429 : : unsigned long address);
2430 : : #if VM_GROWSUP
2431 : : extern int expand_upwards(struct vm_area_struct *vma, unsigned long address);
2432 : : #else
2433 : : #define expand_upwards(vma, address) (0)
2434 : : #endif
2435 : :
2436 : : /* Look up the first VMA which satisfies addr < vm_end, NULL if none. */
2437 : : extern struct vm_area_struct * find_vma(struct mm_struct * mm, unsigned long addr);
2438 : : extern struct vm_area_struct * find_vma_prev(struct mm_struct * mm, unsigned long addr,
2439 : : struct vm_area_struct **pprev);
2440 : :
2441 : : /* Look up the first VMA which intersects the interval start_addr..end_addr-1,
2442 : : NULL if none. Assume start_addr < end_addr. */
2443 : 0 : static inline struct vm_area_struct * find_vma_intersection(struct mm_struct * mm, unsigned long start_addr, unsigned long end_addr)
2444 : : {
2445 : 0 : struct vm_area_struct * vma = find_vma(mm,start_addr);
2446 : :
2447 [ # # # # ]: 0 : if (vma && end_addr <= vma->vm_start)
2448 : : vma = NULL;
2449 [ # # ]: 0 : return vma;
2450 : : }
2451 : :
2452 : 6523395 : static inline unsigned long vm_start_gap(struct vm_area_struct *vma)
2453 : : {
2454 : 6523395 : unsigned long vm_start = vma->vm_start;
2455 : :
2456 [ + + - - : 6523395 : if (vma->vm_flags & VM_GROWSDOWN) {
- - + + -
+ ]
2457 : 446428 : vm_start -= stack_guard_gap;
2458 [ - + - - : 446428 : if (vm_start > vma->vm_start)
- - - + -
- ]
2459 : 0 : vm_start = 0;
2460 : : }
2461 [ + + - - : 6523395 : return vm_start;
+ + - + ]
2462 : : }
2463 : :
2464 : 7677275 : static inline unsigned long vm_end_gap(struct vm_area_struct *vma)
2465 : : {
2466 : 7677275 : unsigned long vm_end = vma->vm_end;
2467 : :
2468 : 7677275 : if (vma->vm_flags & VM_GROWSUP) {
2469 : : vm_end += stack_guard_gap;
2470 : : if (vm_end < vma->vm_end)
2471 : : vm_end = -PAGE_SIZE;
2472 : : }
2473 [ - - + + ]: 7677275 : return vm_end;
2474 : : }
2475 : :
2476 : 6973827 : static inline unsigned long vma_pages(struct vm_area_struct *vma)
2477 : : {
2478 [ + + + - : 6973827 : return (vma->vm_end - vma->vm_start) >> PAGE_SHIFT;
+ + + + -
- - - + +
+ + ]
2479 : : }
2480 : :
2481 : : /* Look up the first VMA which exactly match the interval vm_start ... vm_end */
2482 : 0 : static inline struct vm_area_struct *find_exact_vma(struct mm_struct *mm,
2483 : : unsigned long vm_start, unsigned long vm_end)
2484 : : {
2485 : 0 : struct vm_area_struct *vma = find_vma(mm, vm_start);
2486 : :
2487 [ # # # # : 0 : if (vma && (vma->vm_start != vm_start || vma->vm_end != vm_end))
# # # # #
# # # # #
# # # # ]
2488 : 0 : vma = NULL;
2489 : :
2490 [ # # # # ]: 0 : return vma;
2491 : : }
2492 : :
2493 : 0 : static inline bool range_in_vma(struct vm_area_struct *vma,
2494 : : unsigned long start, unsigned long end)
2495 : : {
2496 [ # # # # : 0 : return (vma && vma->vm_start <= start && end <= vma->vm_end);
# # # # #
# # # # #
# # ]
2497 : : }
2498 : :
2499 : : #ifdef CONFIG_MMU
2500 : : pgprot_t vm_get_page_prot(unsigned long vm_flags);
2501 : : void vma_set_page_prot(struct vm_area_struct *vma);
2502 : : #else
2503 : : static inline pgprot_t vm_get_page_prot(unsigned long vm_flags)
2504 : : {
2505 : : return __pgprot(0);
2506 : : }
2507 : : static inline void vma_set_page_prot(struct vm_area_struct *vma)
2508 : : {
2509 : : vma->vm_page_prot = vm_get_page_prot(vma->vm_flags);
2510 : : }
2511 : : #endif
2512 : :
2513 : : #ifdef CONFIG_NUMA_BALANCING
2514 : : unsigned long change_prot_numa(struct vm_area_struct *vma,
2515 : : unsigned long start, unsigned long end);
2516 : : #endif
2517 : :
2518 : : struct vm_area_struct *find_extend_vma(struct mm_struct *, unsigned long addr);
2519 : : int remap_pfn_range(struct vm_area_struct *, unsigned long addr,
2520 : : unsigned long pfn, unsigned long size, pgprot_t);
2521 : : int vm_insert_page(struct vm_area_struct *, unsigned long addr, struct page *);
2522 : : int vm_map_pages(struct vm_area_struct *vma, struct page **pages,
2523 : : unsigned long num);
2524 : : int vm_map_pages_zero(struct vm_area_struct *vma, struct page **pages,
2525 : : unsigned long num);
2526 : : vm_fault_t vmf_insert_pfn(struct vm_area_struct *vma, unsigned long addr,
2527 : : unsigned long pfn);
2528 : : vm_fault_t vmf_insert_pfn_prot(struct vm_area_struct *vma, unsigned long addr,
2529 : : unsigned long pfn, pgprot_t pgprot);
2530 : : vm_fault_t vmf_insert_mixed(struct vm_area_struct *vma, unsigned long addr,
2531 : : pfn_t pfn);
2532 : : vm_fault_t vmf_insert_mixed_prot(struct vm_area_struct *vma, unsigned long addr,
2533 : : pfn_t pfn, pgprot_t pgprot);
2534 : : vm_fault_t vmf_insert_mixed_mkwrite(struct vm_area_struct *vma,
2535 : : unsigned long addr, pfn_t pfn);
2536 : : int vm_iomap_memory(struct vm_area_struct *vma, phys_addr_t start, unsigned long len);
2537 : :
2538 : 0 : static inline vm_fault_t vmf_insert_page(struct vm_area_struct *vma,
2539 : : unsigned long addr, struct page *page)
2540 : : {
2541 : 0 : int err = vm_insert_page(vma, addr, page);
2542 : :
2543 [ # # ]: 0 : if (err == -ENOMEM)
2544 : : return VM_FAULT_OOM;
2545 [ # # ]: 0 : if (err < 0 && err != -EBUSY)
2546 : 0 : return VM_FAULT_SIGBUS;
2547 : :
2548 : : return VM_FAULT_NOPAGE;
2549 : : }
2550 : :
2551 : 0 : static inline vm_fault_t vmf_error(int err)
2552 : : {
2553 [ # # # # ]: 0 : if (err == -ENOMEM)
2554 : 0 : return VM_FAULT_OOM;
2555 : : return VM_FAULT_SIGBUS;
2556 : : }
2557 : :
2558 : : struct page *follow_page(struct vm_area_struct *vma, unsigned long address,
2559 : : unsigned int foll_flags);
2560 : :
2561 : : #define FOLL_WRITE 0x01 /* check pte is writable */
2562 : : #define FOLL_TOUCH 0x02 /* mark page accessed */
2563 : : #define FOLL_GET 0x04 /* do get_page on page */
2564 : : #define FOLL_DUMP 0x08 /* give error on hole if it would be zero */
2565 : : #define FOLL_FORCE 0x10 /* get_user_pages read/write w/o permission */
2566 : : #define FOLL_NOWAIT 0x20 /* if a disk transfer is needed, start the IO
2567 : : * and return without waiting upon it */
2568 : : #define FOLL_POPULATE 0x40 /* fault in page */
2569 : : #define FOLL_SPLIT 0x80 /* don't return transhuge pages, split them */
2570 : : #define FOLL_HWPOISON 0x100 /* check page is hwpoisoned */
2571 : : #define FOLL_NUMA 0x200 /* force NUMA hinting page fault */
2572 : : #define FOLL_MIGRATION 0x400 /* wait for page to replace migration entry */
2573 : : #define FOLL_TRIED 0x800 /* a retry, previous pass started an IO */
2574 : : #define FOLL_MLOCK 0x1000 /* lock present pages */
2575 : : #define FOLL_REMOTE 0x2000 /* we are working on non-current tsk/mm */
2576 : : #define FOLL_COW 0x4000 /* internal GUP flag */
2577 : : #define FOLL_ANON 0x8000 /* don't do file mappings */
2578 : : #define FOLL_LONGTERM 0x10000 /* mapping lifetime is indefinite: see below */
2579 : : #define FOLL_SPLIT_PMD 0x20000 /* split huge pmd before returning */
2580 : : #define FOLL_PIN 0x40000 /* pages must be released via unpin_user_page */
2581 : :
2582 : : /*
2583 : : * FOLL_PIN and FOLL_LONGTERM may be used in various combinations with each
2584 : : * other. Here is what they mean, and how to use them:
2585 : : *
2586 : : * FOLL_LONGTERM indicates that the page will be held for an indefinite time
2587 : : * period _often_ under userspace control. This is in contrast to
2588 : : * iov_iter_get_pages(), whose usages are transient.
2589 : : *
2590 : : * FIXME: For pages which are part of a filesystem, mappings are subject to the
2591 : : * lifetime enforced by the filesystem and we need guarantees that longterm
2592 : : * users like RDMA and V4L2 only establish mappings which coordinate usage with
2593 : : * the filesystem. Ideas for this coordination include revoking the longterm
2594 : : * pin, delaying writeback, bounce buffer page writeback, etc. As FS DAX was
2595 : : * added after the problem with filesystems was found FS DAX VMAs are
2596 : : * specifically failed. Filesystem pages are still subject to bugs and use of
2597 : : * FOLL_LONGTERM should be avoided on those pages.
2598 : : *
2599 : : * FIXME: Also NOTE that FOLL_LONGTERM is not supported in every GUP call.
2600 : : * Currently only get_user_pages() and get_user_pages_fast() support this flag
2601 : : * and calls to get_user_pages_[un]locked are specifically not allowed. This
2602 : : * is due to an incompatibility with the FS DAX check and
2603 : : * FAULT_FLAG_ALLOW_RETRY.
2604 : : *
2605 : : * In the CMA case: long term pins in a CMA region would unnecessarily fragment
2606 : : * that region. And so, CMA attempts to migrate the page before pinning, when
2607 : : * FOLL_LONGTERM is specified.
2608 : : *
2609 : : * FOLL_PIN indicates that a special kind of tracking (not just page->_refcount,
2610 : : * but an additional pin counting system) will be invoked. This is intended for
2611 : : * anything that gets a page reference and then touches page data (for example,
2612 : : * Direct IO). This lets the filesystem know that some non-file-system entity is
2613 : : * potentially changing the pages' data. In contrast to FOLL_GET (whose pages
2614 : : * are released via put_page()), FOLL_PIN pages must be released, ultimately, by
2615 : : * a call to unpin_user_page().
2616 : : *
2617 : : * FOLL_PIN is similar to FOLL_GET: both of these pin pages. They use different
2618 : : * and separate refcounting mechanisms, however, and that means that each has
2619 : : * its own acquire and release mechanisms:
2620 : : *
2621 : : * FOLL_GET: get_user_pages*() to acquire, and put_page() to release.
2622 : : *
2623 : : * FOLL_PIN: pin_user_pages*() to acquire, and unpin_user_pages to release.
2624 : : *
2625 : : * FOLL_PIN and FOLL_GET are mutually exclusive for a given function call.
2626 : : * (The underlying pages may experience both FOLL_GET-based and FOLL_PIN-based
2627 : : * calls applied to them, and that's perfectly OK. This is a constraint on the
2628 : : * callers, not on the pages.)
2629 : : *
2630 : : * FOLL_PIN should be set internally by the pin_user_pages*() APIs, never
2631 : : * directly by the caller. That's in order to help avoid mismatches when
2632 : : * releasing pages: get_user_pages*() pages must be released via put_page(),
2633 : : * while pin_user_pages*() pages must be released via unpin_user_page().
2634 : : *
2635 : : * Please see Documentation/vm/pin_user_pages.rst for more information.
2636 : : */
2637 : :
2638 : 0 : static inline int vm_fault_to_errno(vm_fault_t vm_fault, int foll_flags)
2639 : : {
2640 [ # # # # ]: 0 : if (vm_fault & VM_FAULT_OOM)
2641 : : return -ENOMEM;
2642 [ # # # # ]: 0 : if (vm_fault & (VM_FAULT_HWPOISON | VM_FAULT_HWPOISON_LARGE))
2643 [ # # ]: 0 : return (foll_flags & FOLL_HWPOISON) ? -EHWPOISON : -EFAULT;
2644 [ # # # # ]: 0 : if (vm_fault & (VM_FAULT_SIGBUS | VM_FAULT_SIGSEGV))
2645 : 0 : return -EFAULT;
2646 : : return 0;
2647 : : }
2648 : :
2649 : : typedef int (*pte_fn_t)(pte_t *pte, unsigned long addr, void *data);
2650 : : extern int apply_to_page_range(struct mm_struct *mm, unsigned long address,
2651 : : unsigned long size, pte_fn_t fn, void *data);
2652 : : extern int apply_to_existing_page_range(struct mm_struct *mm,
2653 : : unsigned long address, unsigned long size,
2654 : : pte_fn_t fn, void *data);
2655 : :
2656 : : #ifdef CONFIG_PAGE_POISONING
2657 : : extern bool page_poisoning_enabled(void);
2658 : : extern void kernel_poison_pages(struct page *page, int numpages, int enable);
2659 : : #else
2660 : : static inline bool page_poisoning_enabled(void) { return false; }
2661 : 1632038 : static inline void kernel_poison_pages(struct page *page, int numpages,
2662 [ + - ]: 1632038 : int enable) { }
2663 : : #endif
2664 : :
2665 : : #ifdef CONFIG_INIT_ON_ALLOC_DEFAULT_ON
2666 : : DECLARE_STATIC_KEY_TRUE(init_on_alloc);
2667 : : #else
2668 : : DECLARE_STATIC_KEY_FALSE(init_on_alloc);
2669 : : #endif
2670 : 1637108 : static inline bool want_init_on_alloc(gfp_t flags)
2671 : : {
2672 [ + - + - ]: 3274216 : if (static_branch_unlikely(&init_on_alloc) &&
2673 : : !page_poisoning_enabled())
2674 : : return true;
2675 [ + + ]: 1637108 : return flags & __GFP_ZERO;
2676 : : }
2677 : :
2678 : : #ifdef CONFIG_INIT_ON_FREE_DEFAULT_ON
2679 : : DECLARE_STATIC_KEY_TRUE(init_on_free);
2680 : : #else
2681 : : DECLARE_STATIC_KEY_FALSE(init_on_free);
2682 : : #endif
2683 : 2931207 : static inline bool want_init_on_free(void)
2684 : : {
2685 [ + - + - : 7641046 : return static_branch_unlikely(&init_on_free) &&
+ - + - +
- - + + -
+ - ]
2686 : : !page_poisoning_enabled();
2687 : : }
2688 : :
2689 : : #ifdef CONFIG_DEBUG_PAGEALLOC
2690 : : extern void init_debug_pagealloc(void);
2691 : : #else
2692 : 13 : static inline void init_debug_pagealloc(void) {}
2693 : : #endif
2694 : : extern bool _debug_pagealloc_enabled_early;
2695 : : DECLARE_STATIC_KEY_FALSE(_debug_pagealloc_enabled);
2696 : :
2697 : 174875 : static inline bool debug_pagealloc_enabled(void)
2698 : : {
2699 : 174875 : return IS_ENABLED(CONFIG_DEBUG_PAGEALLOC) &&
2700 : : _debug_pagealloc_enabled_early;
2701 : : }
2702 : :
2703 : : /*
2704 : : * For use in fast paths after init_debug_pagealloc() has run, or when a
2705 : : * false negative result is not harmful when called too early.
2706 : : */
2707 : 14867934 : static inline bool debug_pagealloc_enabled_static(void)
2708 : : {
2709 : 14867934 : if (!IS_ENABLED(CONFIG_DEBUG_PAGEALLOC))
2710 : 5531979 : return false;
2711 : :
2712 : : return static_branch_unlikely(&_debug_pagealloc_enabled);
2713 : : }
2714 : :
2715 : : #if defined(CONFIG_DEBUG_PAGEALLOC) || defined(CONFIG_ARCH_HAS_SET_DIRECT_MAP)
2716 : : extern void __kernel_map_pages(struct page *page, int numpages, int enable);
2717 : :
2718 : : /*
2719 : : * When called in DEBUG_PAGEALLOC context, the call should most likely be
2720 : : * guarded by debug_pagealloc_enabled() or debug_pagealloc_enabled_static()
2721 : : */
2722 : : static inline void
2723 : 0 : kernel_map_pages(struct page *page, int numpages, int enable)
2724 : : {
2725 : 0 : __kernel_map_pages(page, numpages, enable);
2726 : 0 : }
2727 : : #ifdef CONFIG_HIBERNATION
2728 : : extern bool kernel_page_present(struct page *page);
2729 : : #endif /* CONFIG_HIBERNATION */
2730 : : #else /* CONFIG_DEBUG_PAGEALLOC || CONFIG_ARCH_HAS_SET_DIRECT_MAP */
2731 : : static inline void
2732 : : kernel_map_pages(struct page *page, int numpages, int enable) {}
2733 : : #ifdef CONFIG_HIBERNATION
2734 : : static inline bool kernel_page_present(struct page *page) { return true; }
2735 : : #endif /* CONFIG_HIBERNATION */
2736 : : #endif /* CONFIG_DEBUG_PAGEALLOC || CONFIG_ARCH_HAS_SET_DIRECT_MAP */
2737 : :
2738 : : #ifdef __HAVE_ARCH_GATE_AREA
2739 : : extern struct vm_area_struct *get_gate_vma(struct mm_struct *mm);
2740 : : extern int in_gate_area_no_mm(unsigned long addr);
2741 : : extern int in_gate_area(struct mm_struct *mm, unsigned long addr);
2742 : : #else
2743 : : static inline struct vm_area_struct *get_gate_vma(struct mm_struct *mm)
2744 : : {
2745 : : return NULL;
2746 : : }
2747 : : static inline int in_gate_area_no_mm(unsigned long addr) { return 0; }
2748 : : static inline int in_gate_area(struct mm_struct *mm, unsigned long addr)
2749 : : {
2750 : : return 0;
2751 : : }
2752 : : #endif /* __HAVE_ARCH_GATE_AREA */
2753 : :
2754 : : extern bool process_shares_mm(struct task_struct *p, struct mm_struct *mm);
2755 : :
2756 : : #ifdef CONFIG_SYSCTL
2757 : : extern int sysctl_drop_caches;
2758 : : int drop_caches_sysctl_handler(struct ctl_table *, int,
2759 : : void __user *, size_t *, loff_t *);
2760 : : #endif
2761 : :
2762 : : void drop_slab(void);
2763 : : void drop_slab_node(int nid);
2764 : :
2765 : : #ifndef CONFIG_MMU
2766 : : #define randomize_va_space 0
2767 : : #else
2768 : : extern int randomize_va_space;
2769 : : #endif
2770 : :
2771 : : const char * arch_vma_name(struct vm_area_struct *vma);
2772 : : #ifdef CONFIG_MMU
2773 : : void print_vma_addr(char *prefix, unsigned long rip);
2774 : : #else
2775 : : static inline void print_vma_addr(char *prefix, unsigned long rip)
2776 : : {
2777 : : }
2778 : : #endif
2779 : :
2780 : : void *sparse_buffer_alloc(unsigned long size);
2781 : : struct page * __populate_section_memmap(unsigned long pfn,
2782 : : unsigned long nr_pages, int nid, struct vmem_altmap *altmap);
2783 : : pgd_t *vmemmap_pgd_populate(unsigned long addr, int node);
2784 : : p4d_t *vmemmap_p4d_populate(pgd_t *pgd, unsigned long addr, int node);
2785 : : pud_t *vmemmap_pud_populate(p4d_t *p4d, unsigned long addr, int node);
2786 : : pmd_t *vmemmap_pmd_populate(pud_t *pud, unsigned long addr, int node);
2787 : : pte_t *vmemmap_pte_populate(pmd_t *pmd, unsigned long addr, int node);
2788 : : void *vmemmap_alloc_block(unsigned long size, int node);
2789 : : struct vmem_altmap;
2790 : : void *vmemmap_alloc_block_buf(unsigned long size, int node);
2791 : : void *altmap_alloc_block_buf(unsigned long size, struct vmem_altmap *altmap);
2792 : : void vmemmap_verify(pte_t *, int, unsigned long, unsigned long);
2793 : : int vmemmap_populate_basepages(unsigned long start, unsigned long end,
2794 : : int node);
2795 : : int vmemmap_populate(unsigned long start, unsigned long end, int node,
2796 : : struct vmem_altmap *altmap);
2797 : : void vmemmap_populate_print_last(void);
2798 : : #ifdef CONFIG_MEMORY_HOTPLUG
2799 : : void vmemmap_free(unsigned long start, unsigned long end,
2800 : : struct vmem_altmap *altmap);
2801 : : #endif
2802 : : void register_page_bootmem_memmap(unsigned long section_nr, struct page *map,
2803 : : unsigned long nr_pages);
2804 : :
2805 : : enum mf_flags {
2806 : : MF_COUNT_INCREASED = 1 << 0,
2807 : : MF_ACTION_REQUIRED = 1 << 1,
2808 : : MF_MUST_KILL = 1 << 2,
2809 : : MF_SOFT_OFFLINE = 1 << 3,
2810 : : };
2811 : : extern int memory_failure(unsigned long pfn, int flags);
2812 : : extern void memory_failure_queue(unsigned long pfn, int flags);
2813 : : extern int unpoison_memory(unsigned long pfn);
2814 : : extern int get_hwpoison_page(struct page *page);
2815 : : #define put_hwpoison_page(page) put_page(page)
2816 : : extern int sysctl_memory_failure_early_kill;
2817 : : extern int sysctl_memory_failure_recovery;
2818 : : extern void shake_page(struct page *p, int access);
2819 : : extern atomic_long_t num_poisoned_pages __read_mostly;
2820 : : extern int soft_offline_page(unsigned long pfn, int flags);
2821 : :
2822 : :
2823 : : /*
2824 : : * Error handlers for various types of pages.
2825 : : */
2826 : : enum mf_result {
2827 : : MF_IGNORED, /* Error: cannot be handled */
2828 : : MF_FAILED, /* Error: handling failed */
2829 : : MF_DELAYED, /* Will be handled later */
2830 : : MF_RECOVERED, /* Successfully recovered */
2831 : : };
2832 : :
2833 : : enum mf_action_page_type {
2834 : : MF_MSG_KERNEL,
2835 : : MF_MSG_KERNEL_HIGH_ORDER,
2836 : : MF_MSG_SLAB,
2837 : : MF_MSG_DIFFERENT_COMPOUND,
2838 : : MF_MSG_POISONED_HUGE,
2839 : : MF_MSG_HUGE,
2840 : : MF_MSG_FREE_HUGE,
2841 : : MF_MSG_NON_PMD_HUGE,
2842 : : MF_MSG_UNMAP_FAILED,
2843 : : MF_MSG_DIRTY_SWAPCACHE,
2844 : : MF_MSG_CLEAN_SWAPCACHE,
2845 : : MF_MSG_DIRTY_MLOCKED_LRU,
2846 : : MF_MSG_CLEAN_MLOCKED_LRU,
2847 : : MF_MSG_DIRTY_UNEVICTABLE_LRU,
2848 : : MF_MSG_CLEAN_UNEVICTABLE_LRU,
2849 : : MF_MSG_DIRTY_LRU,
2850 : : MF_MSG_CLEAN_LRU,
2851 : : MF_MSG_TRUNCATED_LRU,
2852 : : MF_MSG_BUDDY,
2853 : : MF_MSG_BUDDY_2ND,
2854 : : MF_MSG_DAX,
2855 : : MF_MSG_UNKNOWN,
2856 : : };
2857 : :
2858 : : #if defined(CONFIG_TRANSPARENT_HUGEPAGE) || defined(CONFIG_HUGETLBFS)
2859 : : extern void clear_huge_page(struct page *page,
2860 : : unsigned long addr_hint,
2861 : : unsigned int pages_per_huge_page);
2862 : : extern void copy_user_huge_page(struct page *dst, struct page *src,
2863 : : unsigned long addr_hint,
2864 : : struct vm_area_struct *vma,
2865 : : unsigned int pages_per_huge_page);
2866 : : extern long copy_huge_page_from_user(struct page *dst_page,
2867 : : const void __user *usr_src,
2868 : : unsigned int pages_per_huge_page,
2869 : : bool allow_pagefault);
2870 : : #endif /* CONFIG_TRANSPARENT_HUGEPAGE || CONFIG_HUGETLBFS */
2871 : :
2872 : : #ifdef CONFIG_DEBUG_PAGEALLOC
2873 : : extern unsigned int _debug_guardpage_minorder;
2874 : : DECLARE_STATIC_KEY_FALSE(_debug_guardpage_enabled);
2875 : :
2876 : : static inline unsigned int debug_guardpage_minorder(void)
2877 : : {
2878 : : return _debug_guardpage_minorder;
2879 : : }
2880 : :
2881 : : static inline bool debug_guardpage_enabled(void)
2882 : : {
2883 : : return static_branch_unlikely(&_debug_guardpage_enabled);
2884 : : }
2885 : :
2886 : : static inline bool page_is_guard(struct page *page)
2887 : : {
2888 : : if (!debug_guardpage_enabled())
2889 : : return false;
2890 : :
2891 : : return PageGuard(page);
2892 : : }
2893 : : #else
2894 : 13 : static inline unsigned int debug_guardpage_minorder(void) { return 0; }
2895 : : static inline bool debug_guardpage_enabled(void) { return false; }
2896 [ + + + + ]: 1207442 : static inline bool page_is_guard(struct page *page) { return false; }
2897 : : #endif /* CONFIG_DEBUG_PAGEALLOC */
2898 : :
2899 : : #if MAX_NUMNODES > 1
2900 : : void __init setup_nr_node_ids(void);
2901 : : #else
2902 : : static inline void setup_nr_node_ids(void) {}
2903 : : #endif
2904 : :
2905 : : extern int memcmp_pages(struct page *page1, struct page *page2);
2906 : :
2907 : 0 : static inline int pages_identical(struct page *page1, struct page *page2)
2908 : : {
2909 : 0 : return !memcmp_pages(page1, page2);
2910 : : }
2911 : :
2912 : : #ifdef CONFIG_MAPPING_DIRTY_HELPERS
2913 : : unsigned long clean_record_shared_mapping_range(struct address_space *mapping,
2914 : : pgoff_t first_index, pgoff_t nr,
2915 : : pgoff_t bitmap_pgoff,
2916 : : unsigned long *bitmap,
2917 : : pgoff_t *start,
2918 : : pgoff_t *end);
2919 : :
2920 : : unsigned long wp_shared_mapping_range(struct address_space *mapping,
2921 : : pgoff_t first_index, pgoff_t nr);
2922 : : #endif
2923 : :
2924 : : #endif /* __KERNEL__ */
2925 : : #endif /* _LINUX_MM_H */
|
