Commit | Line | Data |
---|---|---|
16d69265 | 1 | #include <linux/mm.h> |
30992c97 MM |
2 | #include <linux/slab.h> |
3 | #include <linux/string.h> | |
b95f1b31 | 4 | #include <linux/export.h> |
96840aa0 | 5 | #include <linux/err.h> |
3b8f14b4 | 6 | #include <linux/sched.h> |
eb36c587 | 7 | #include <linux/security.h> |
96840aa0 | 8 | #include <asm/uaccess.h> |
30992c97 | 9 | |
6038def0 NK |
10 | #include "internal.h" |
11 | ||
a8d154b0 | 12 | #define CREATE_TRACE_POINTS |
ad8d75ff | 13 | #include <trace/events/kmem.h> |
a8d154b0 | 14 | |
30992c97 | 15 | /** |
30992c97 | 16 | * kstrdup - allocate space for and copy an existing string |
30992c97 MM |
17 | * @s: the string to duplicate |
18 | * @gfp: the GFP mask used in the kmalloc() call when allocating memory | |
19 | */ | |
20 | char *kstrdup(const char *s, gfp_t gfp) | |
21 | { | |
22 | size_t len; | |
23 | char *buf; | |
24 | ||
25 | if (!s) | |
26 | return NULL; | |
27 | ||
28 | len = strlen(s) + 1; | |
1d2c8eea | 29 | buf = kmalloc_track_caller(len, gfp); |
30992c97 MM |
30 | if (buf) |
31 | memcpy(buf, s, len); | |
32 | return buf; | |
33 | } | |
34 | EXPORT_SYMBOL(kstrdup); | |
96840aa0 | 35 | |
1e66df3e JF |
36 | /** |
37 | * kstrndup - allocate space for and copy an existing string | |
38 | * @s: the string to duplicate | |
39 | * @max: read at most @max chars from @s | |
40 | * @gfp: the GFP mask used in the kmalloc() call when allocating memory | |
41 | */ | |
42 | char *kstrndup(const char *s, size_t max, gfp_t gfp) | |
43 | { | |
44 | size_t len; | |
45 | char *buf; | |
46 | ||
47 | if (!s) | |
48 | return NULL; | |
49 | ||
50 | len = strnlen(s, max); | |
51 | buf = kmalloc_track_caller(len+1, gfp); | |
52 | if (buf) { | |
53 | memcpy(buf, s, len); | |
54 | buf[len] = '\0'; | |
55 | } | |
56 | return buf; | |
57 | } | |
58 | EXPORT_SYMBOL(kstrndup); | |
59 | ||
1a2f67b4 AD |
60 | /** |
61 | * kmemdup - duplicate region of memory | |
62 | * | |
63 | * @src: memory region to duplicate | |
64 | * @len: memory region length | |
65 | * @gfp: GFP mask to use | |
66 | */ | |
67 | void *kmemdup(const void *src, size_t len, gfp_t gfp) | |
68 | { | |
69 | void *p; | |
70 | ||
1d2c8eea | 71 | p = kmalloc_track_caller(len, gfp); |
1a2f67b4 AD |
72 | if (p) |
73 | memcpy(p, src, len); | |
74 | return p; | |
75 | } | |
76 | EXPORT_SYMBOL(kmemdup); | |
77 | ||
610a77e0 LZ |
78 | /** |
79 | * memdup_user - duplicate memory region from user space | |
80 | * | |
81 | * @src: source address in user space | |
82 | * @len: number of bytes to copy | |
83 | * | |
84 | * Returns an ERR_PTR() on failure. | |
85 | */ | |
86 | void *memdup_user(const void __user *src, size_t len) | |
87 | { | |
88 | void *p; | |
89 | ||
90 | /* | |
91 | * Always use GFP_KERNEL, since copy_from_user() can sleep and | |
92 | * cause pagefault, which makes it pointless to use GFP_NOFS | |
93 | * or GFP_ATOMIC. | |
94 | */ | |
95 | p = kmalloc_track_caller(len, GFP_KERNEL); | |
96 | if (!p) | |
97 | return ERR_PTR(-ENOMEM); | |
98 | ||
99 | if (copy_from_user(p, src, len)) { | |
100 | kfree(p); | |
101 | return ERR_PTR(-EFAULT); | |
102 | } | |
103 | ||
104 | return p; | |
105 | } | |
106 | EXPORT_SYMBOL(memdup_user); | |
107 | ||
ef2ad80c | 108 | /** |
93bc4e89 | 109 | * __krealloc - like krealloc() but don't free @p. |
ef2ad80c CL |
110 | * @p: object to reallocate memory for. |
111 | * @new_size: how many bytes of memory are required. | |
112 | * @flags: the type of memory to allocate. | |
113 | * | |
93bc4e89 PE |
114 | * This function is like krealloc() except it never frees the originally |
115 | * allocated buffer. Use this if you don't want to free the buffer immediately | |
116 | * like, for example, with RCU. | |
ef2ad80c | 117 | */ |
93bc4e89 | 118 | void *__krealloc(const void *p, size_t new_size, gfp_t flags) |
ef2ad80c CL |
119 | { |
120 | void *ret; | |
ef8b4520 | 121 | size_t ks = 0; |
ef2ad80c | 122 | |
93bc4e89 | 123 | if (unlikely(!new_size)) |
6cb8f913 | 124 | return ZERO_SIZE_PTR; |
ef2ad80c | 125 | |
ef8b4520 CL |
126 | if (p) |
127 | ks = ksize(p); | |
128 | ||
ef2ad80c CL |
129 | if (ks >= new_size) |
130 | return (void *)p; | |
131 | ||
132 | ret = kmalloc_track_caller(new_size, flags); | |
93bc4e89 | 133 | if (ret && p) |
be21f0ab | 134 | memcpy(ret, p, ks); |
93bc4e89 PE |
135 | |
136 | return ret; | |
137 | } | |
138 | EXPORT_SYMBOL(__krealloc); | |
139 | ||
140 | /** | |
141 | * krealloc - reallocate memory. The contents will remain unchanged. | |
142 | * @p: object to reallocate memory for. | |
143 | * @new_size: how many bytes of memory are required. | |
144 | * @flags: the type of memory to allocate. | |
145 | * | |
146 | * The contents of the object pointed to are preserved up to the | |
147 | * lesser of the new and old sizes. If @p is %NULL, krealloc() | |
148 | * behaves exactly like kmalloc(). If @size is 0 and @p is not a | |
149 | * %NULL pointer, the object pointed to is freed. | |
150 | */ | |
151 | void *krealloc(const void *p, size_t new_size, gfp_t flags) | |
152 | { | |
153 | void *ret; | |
154 | ||
155 | if (unlikely(!new_size)) { | |
ef2ad80c | 156 | kfree(p); |
93bc4e89 | 157 | return ZERO_SIZE_PTR; |
ef2ad80c | 158 | } |
93bc4e89 PE |
159 | |
160 | ret = __krealloc(p, new_size, flags); | |
161 | if (ret && p != ret) | |
162 | kfree(p); | |
163 | ||
ef2ad80c CL |
164 | return ret; |
165 | } | |
166 | EXPORT_SYMBOL(krealloc); | |
167 | ||
3ef0e5ba JW |
168 | /** |
169 | * kzfree - like kfree but zero memory | |
170 | * @p: object to free memory of | |
171 | * | |
172 | * The memory of the object @p points to is zeroed before freed. | |
173 | * If @p is %NULL, kzfree() does nothing. | |
a234bdc9 PE |
174 | * |
175 | * Note: this function zeroes the whole allocated buffer which can be a good | |
176 | * deal bigger than the requested buffer size passed to kmalloc(). So be | |
177 | * careful when using this function in performance sensitive code. | |
3ef0e5ba JW |
178 | */ |
179 | void kzfree(const void *p) | |
180 | { | |
181 | size_t ks; | |
182 | void *mem = (void *)p; | |
183 | ||
184 | if (unlikely(ZERO_OR_NULL_PTR(mem))) | |
185 | return; | |
186 | ks = ksize(mem); | |
187 | memset(mem, 0, ks); | |
188 | kfree(mem); | |
189 | } | |
190 | EXPORT_SYMBOL(kzfree); | |
191 | ||
96840aa0 DA |
192 | /* |
193 | * strndup_user - duplicate an existing string from user space | |
96840aa0 DA |
194 | * @s: The string to duplicate |
195 | * @n: Maximum number of bytes to copy, including the trailing NUL. | |
196 | */ | |
197 | char *strndup_user(const char __user *s, long n) | |
198 | { | |
199 | char *p; | |
200 | long length; | |
201 | ||
202 | length = strnlen_user(s, n); | |
203 | ||
204 | if (!length) | |
205 | return ERR_PTR(-EFAULT); | |
206 | ||
207 | if (length > n) | |
208 | return ERR_PTR(-EINVAL); | |
209 | ||
90d74045 | 210 | p = memdup_user(s, length); |
96840aa0 | 211 | |
90d74045 JL |
212 | if (IS_ERR(p)) |
213 | return p; | |
96840aa0 DA |
214 | |
215 | p[length - 1] = '\0'; | |
216 | ||
217 | return p; | |
218 | } | |
219 | EXPORT_SYMBOL(strndup_user); | |
16d69265 | 220 | |
6038def0 NK |
221 | void __vma_link_list(struct mm_struct *mm, struct vm_area_struct *vma, |
222 | struct vm_area_struct *prev, struct rb_node *rb_parent) | |
223 | { | |
224 | struct vm_area_struct *next; | |
225 | ||
226 | vma->vm_prev = prev; | |
227 | if (prev) { | |
228 | next = prev->vm_next; | |
229 | prev->vm_next = vma; | |
230 | } else { | |
231 | mm->mmap = vma; | |
232 | if (rb_parent) | |
233 | next = rb_entry(rb_parent, | |
234 | struct vm_area_struct, vm_rb); | |
235 | else | |
236 | next = NULL; | |
237 | } | |
238 | vma->vm_next = next; | |
239 | if (next) | |
240 | next->vm_prev = vma; | |
241 | } | |
242 | ||
b7643757 SP |
243 | /* Check if the vma is being used as a stack by this task */ |
244 | static int vm_is_stack_for_task(struct task_struct *t, | |
245 | struct vm_area_struct *vma) | |
246 | { | |
247 | return (vma->vm_start <= KSTK_ESP(t) && vma->vm_end >= KSTK_ESP(t)); | |
248 | } | |
249 | ||
250 | /* | |
251 | * Check if the vma is being used as a stack. | |
252 | * If is_group is non-zero, check in the entire thread group or else | |
253 | * just check in the current task. Returns the pid of the task that | |
254 | * the vma is stack for. | |
255 | */ | |
256 | pid_t vm_is_stack(struct task_struct *task, | |
257 | struct vm_area_struct *vma, int in_group) | |
258 | { | |
259 | pid_t ret = 0; | |
260 | ||
261 | if (vm_is_stack_for_task(task, vma)) | |
262 | return task->pid; | |
263 | ||
264 | if (in_group) { | |
265 | struct task_struct *t; | |
266 | rcu_read_lock(); | |
267 | if (!pid_alive(task)) | |
268 | goto done; | |
269 | ||
270 | t = task; | |
271 | do { | |
272 | if (vm_is_stack_for_task(t, vma)) { | |
273 | ret = t->pid; | |
274 | goto done; | |
275 | } | |
276 | } while_each_thread(task, t); | |
277 | done: | |
278 | rcu_read_unlock(); | |
279 | } | |
280 | ||
281 | return ret; | |
282 | } | |
283 | ||
efc1a3b1 | 284 | #if defined(CONFIG_MMU) && !defined(HAVE_ARCH_PICK_MMAP_LAYOUT) |
16d69265 AM |
285 | void arch_pick_mmap_layout(struct mm_struct *mm) |
286 | { | |
287 | mm->mmap_base = TASK_UNMAPPED_BASE; | |
288 | mm->get_unmapped_area = arch_get_unmapped_area; | |
289 | mm->unmap_area = arch_unmap_area; | |
290 | } | |
291 | #endif | |
912985dc | 292 | |
45888a0c XG |
293 | /* |
294 | * Like get_user_pages_fast() except its IRQ-safe in that it won't fall | |
295 | * back to the regular GUP. | |
25985edc | 296 | * If the architecture not support this function, simply return with no |
45888a0c XG |
297 | * page pinned |
298 | */ | |
299 | int __attribute__((weak)) __get_user_pages_fast(unsigned long start, | |
300 | int nr_pages, int write, struct page **pages) | |
301 | { | |
302 | return 0; | |
303 | } | |
304 | EXPORT_SYMBOL_GPL(__get_user_pages_fast); | |
305 | ||
9de100d0 AG |
306 | /** |
307 | * get_user_pages_fast() - pin user pages in memory | |
308 | * @start: starting user address | |
309 | * @nr_pages: number of pages from start to pin | |
310 | * @write: whether pages will be written to | |
311 | * @pages: array that receives pointers to the pages pinned. | |
312 | * Should be at least nr_pages long. | |
313 | * | |
9de100d0 AG |
314 | * Returns number of pages pinned. This may be fewer than the number |
315 | * requested. If nr_pages is 0 or negative, returns 0. If no pages | |
316 | * were pinned, returns -errno. | |
d2bf6be8 NP |
317 | * |
318 | * get_user_pages_fast provides equivalent functionality to get_user_pages, | |
319 | * operating on current and current->mm, with force=0 and vma=NULL. However | |
320 | * unlike get_user_pages, it must be called without mmap_sem held. | |
321 | * | |
322 | * get_user_pages_fast may take mmap_sem and page table locks, so no | |
323 | * assumptions can be made about lack of locking. get_user_pages_fast is to be | |
324 | * implemented in a way that is advantageous (vs get_user_pages()) when the | |
325 | * user memory area is already faulted in and present in ptes. However if the | |
326 | * pages have to be faulted in, it may turn out to be slightly slower so | |
327 | * callers need to carefully consider what to use. On many architectures, | |
328 | * get_user_pages_fast simply falls back to get_user_pages. | |
9de100d0 | 329 | */ |
912985dc RR |
330 | int __attribute__((weak)) get_user_pages_fast(unsigned long start, |
331 | int nr_pages, int write, struct page **pages) | |
332 | { | |
333 | struct mm_struct *mm = current->mm; | |
334 | int ret; | |
335 | ||
336 | down_read(&mm->mmap_sem); | |
337 | ret = get_user_pages(current, mm, start, nr_pages, | |
338 | write, 0, pages, NULL); | |
339 | up_read(&mm->mmap_sem); | |
340 | ||
341 | return ret; | |
342 | } | |
343 | EXPORT_SYMBOL_GPL(get_user_pages_fast); | |
ca2b84cb | 344 | |
eb36c587 AV |
345 | unsigned long vm_mmap_pgoff(struct file *file, unsigned long addr, |
346 | unsigned long len, unsigned long prot, | |
347 | unsigned long flag, unsigned long pgoff) | |
348 | { | |
349 | unsigned long ret; | |
350 | struct mm_struct *mm = current->mm; | |
351 | ||
352 | ret = security_mmap_file(file, prot, flag); | |
353 | if (!ret) { | |
354 | down_write(&mm->mmap_sem); | |
355 | ret = do_mmap_pgoff(file, addr, len, prot, flag, pgoff); | |
356 | up_write(&mm->mmap_sem); | |
357 | } | |
358 | return ret; | |
359 | } | |
360 | ||
361 | unsigned long vm_mmap(struct file *file, unsigned long addr, | |
362 | unsigned long len, unsigned long prot, | |
363 | unsigned long flag, unsigned long offset) | |
364 | { | |
365 | if (unlikely(offset + PAGE_ALIGN(len) < offset)) | |
366 | return -EINVAL; | |
367 | if (unlikely(offset & ~PAGE_MASK)) | |
368 | return -EINVAL; | |
369 | ||
370 | return vm_mmap_pgoff(file, addr, len, prot, flag, offset >> PAGE_SHIFT); | |
371 | } | |
372 | EXPORT_SYMBOL(vm_mmap); | |
373 | ||
ca2b84cb | 374 | /* Tracepoints definitions. */ |
ca2b84cb EGM |
375 | EXPORT_TRACEPOINT_SYMBOL(kmalloc); |
376 | EXPORT_TRACEPOINT_SYMBOL(kmem_cache_alloc); | |
377 | EXPORT_TRACEPOINT_SYMBOL(kmalloc_node); | |
378 | EXPORT_TRACEPOINT_SYMBOL(kmem_cache_alloc_node); | |
379 | EXPORT_TRACEPOINT_SYMBOL(kfree); | |
380 | EXPORT_TRACEPOINT_SYMBOL(kmem_cache_free); |