]>
git.proxmox.com Git - mirror_ubuntu-zesty-kernel.git/blob - mm/util.c
2 #include <linux/slab.h>
3 #include <linux/string.h>
4 #include <linux/compiler.h>
5 #include <linux/export.h>
7 #include <linux/sched.h>
8 #include <linux/security.h>
9 #include <linux/swap.h>
10 #include <linux/swapops.h>
11 #include <linux/mman.h>
12 #include <linux/hugetlb.h>
14 #include <asm/uaccess.h>
18 #define CREATE_TRACE_POINTS
19 #include <trace/events/kmem.h>
22 * kstrdup - allocate space for and copy an existing string
23 * @s: the string to duplicate
24 * @gfp: the GFP mask used in the kmalloc() call when allocating memory
26 char *kstrdup(const char *s
, gfp_t gfp
)
35 buf
= kmalloc_track_caller(len
, gfp
);
40 EXPORT_SYMBOL(kstrdup
);
43 * kstrndup - allocate space for and copy an existing string
44 * @s: the string to duplicate
45 * @max: read at most @max chars from @s
46 * @gfp: the GFP mask used in the kmalloc() call when allocating memory
48 char *kstrndup(const char *s
, size_t max
, gfp_t gfp
)
56 len
= strnlen(s
, max
);
57 buf
= kmalloc_track_caller(len
+1, gfp
);
64 EXPORT_SYMBOL(kstrndup
);
67 * kmemdup - duplicate region of memory
69 * @src: memory region to duplicate
70 * @len: memory region length
71 * @gfp: GFP mask to use
73 void *kmemdup(const void *src
, size_t len
, gfp_t gfp
)
77 p
= kmalloc_track_caller(len
, gfp
);
82 EXPORT_SYMBOL(kmemdup
);
85 * memdup_user - duplicate memory region from user space
87 * @src: source address in user space
88 * @len: number of bytes to copy
90 * Returns an ERR_PTR() on failure.
92 void *memdup_user(const void __user
*src
, size_t len
)
97 * Always use GFP_KERNEL, since copy_from_user() can sleep and
98 * cause pagefault, which makes it pointless to use GFP_NOFS
101 p
= kmalloc_track_caller(len
, GFP_KERNEL
);
103 return ERR_PTR(-ENOMEM
);
105 if (copy_from_user(p
, src
, len
)) {
107 return ERR_PTR(-EFAULT
);
112 EXPORT_SYMBOL(memdup_user
);
114 static __always_inline
void *__do_krealloc(const void *p
, size_t new_size
,
126 ret
= kmalloc_track_caller(new_size
, flags
);
134 * __krealloc - like krealloc() but don't free @p.
135 * @p: object to reallocate memory for.
136 * @new_size: how many bytes of memory are required.
137 * @flags: the type of memory to allocate.
139 * This function is like krealloc() except it never frees the originally
140 * allocated buffer. Use this if you don't want to free the buffer immediately
141 * like, for example, with RCU.
143 void *__krealloc(const void *p
, size_t new_size
, gfp_t flags
)
145 if (unlikely(!new_size
))
146 return ZERO_SIZE_PTR
;
148 return __do_krealloc(p
, new_size
, flags
);
151 EXPORT_SYMBOL(__krealloc
);
154 * krealloc - reallocate memory. The contents will remain unchanged.
155 * @p: object to reallocate memory for.
156 * @new_size: how many bytes of memory are required.
157 * @flags: the type of memory to allocate.
159 * The contents of the object pointed to are preserved up to the
160 * lesser of the new and old sizes. If @p is %NULL, krealloc()
161 * behaves exactly like kmalloc(). If @new_size is 0 and @p is not a
162 * %NULL pointer, the object pointed to is freed.
164 void *krealloc(const void *p
, size_t new_size
, gfp_t flags
)
168 if (unlikely(!new_size
)) {
170 return ZERO_SIZE_PTR
;
173 ret
= __do_krealloc(p
, new_size
, flags
);
179 EXPORT_SYMBOL(krealloc
);
182 * kzfree - like kfree but zero memory
183 * @p: object to free memory of
185 * The memory of the object @p points to is zeroed before freed.
186 * If @p is %NULL, kzfree() does nothing.
188 * Note: this function zeroes the whole allocated buffer which can be a good
189 * deal bigger than the requested buffer size passed to kmalloc(). So be
190 * careful when using this function in performance sensitive code.
192 void kzfree(const void *p
)
195 void *mem
= (void *)p
;
197 if (unlikely(ZERO_OR_NULL_PTR(mem
)))
203 EXPORT_SYMBOL(kzfree
);
206 * strndup_user - duplicate an existing string from user space
207 * @s: The string to duplicate
208 * @n: Maximum number of bytes to copy, including the trailing NUL.
210 char *strndup_user(const char __user
*s
, long n
)
215 length
= strnlen_user(s
, n
);
218 return ERR_PTR(-EFAULT
);
221 return ERR_PTR(-EINVAL
);
223 p
= memdup_user(s
, length
);
228 p
[length
- 1] = '\0';
232 EXPORT_SYMBOL(strndup_user
);
234 void __vma_link_list(struct mm_struct
*mm
, struct vm_area_struct
*vma
,
235 struct vm_area_struct
*prev
, struct rb_node
*rb_parent
)
237 struct vm_area_struct
*next
;
241 next
= prev
->vm_next
;
246 next
= rb_entry(rb_parent
,
247 struct vm_area_struct
, vm_rb
);
256 /* Check if the vma is being used as a stack by this task */
257 static int vm_is_stack_for_task(struct task_struct
*t
,
258 struct vm_area_struct
*vma
)
260 return (vma
->vm_start
<= KSTK_ESP(t
) && vma
->vm_end
>= KSTK_ESP(t
));
264 * Check if the vma is being used as a stack.
265 * If is_group is non-zero, check in the entire thread group or else
266 * just check in the current task. Returns the pid of the task that
267 * the vma is stack for.
269 pid_t
vm_is_stack(struct task_struct
*task
,
270 struct vm_area_struct
*vma
, int in_group
)
274 if (vm_is_stack_for_task(task
, vma
))
278 struct task_struct
*t
;
280 if (!pid_alive(task
))
285 if (vm_is_stack_for_task(t
, vma
)) {
289 } while_each_thread(task
, t
);
297 #if defined(CONFIG_MMU) && !defined(HAVE_ARCH_PICK_MMAP_LAYOUT)
298 void arch_pick_mmap_layout(struct mm_struct
*mm
)
300 mm
->mmap_base
= TASK_UNMAPPED_BASE
;
301 mm
->get_unmapped_area
= arch_get_unmapped_area
;
306 * Like get_user_pages_fast() except its IRQ-safe in that it won't fall
307 * back to the regular GUP.
308 * If the architecture not support this function, simply return with no
311 int __weak
__get_user_pages_fast(unsigned long start
,
312 int nr_pages
, int write
, struct page
**pages
)
316 EXPORT_SYMBOL_GPL(__get_user_pages_fast
);
319 * get_user_pages_fast() - pin user pages in memory
320 * @start: starting user address
321 * @nr_pages: number of pages from start to pin
322 * @write: whether pages will be written to
323 * @pages: array that receives pointers to the pages pinned.
324 * Should be at least nr_pages long.
326 * Returns number of pages pinned. This may be fewer than the number
327 * requested. If nr_pages is 0 or negative, returns 0. If no pages
328 * were pinned, returns -errno.
330 * get_user_pages_fast provides equivalent functionality to get_user_pages,
331 * operating on current and current->mm, with force=0 and vma=NULL. However
332 * unlike get_user_pages, it must be called without mmap_sem held.
334 * get_user_pages_fast may take mmap_sem and page table locks, so no
335 * assumptions can be made about lack of locking. get_user_pages_fast is to be
336 * implemented in a way that is advantageous (vs get_user_pages()) when the
337 * user memory area is already faulted in and present in ptes. However if the
338 * pages have to be faulted in, it may turn out to be slightly slower so
339 * callers need to carefully consider what to use. On many architectures,
340 * get_user_pages_fast simply falls back to get_user_pages.
342 int __weak
get_user_pages_fast(unsigned long start
,
343 int nr_pages
, int write
, struct page
**pages
)
345 struct mm_struct
*mm
= current
->mm
;
348 down_read(&mm
->mmap_sem
);
349 ret
= get_user_pages(current
, mm
, start
, nr_pages
,
350 write
, 0, pages
, NULL
);
351 up_read(&mm
->mmap_sem
);
355 EXPORT_SYMBOL_GPL(get_user_pages_fast
);
357 unsigned long vm_mmap_pgoff(struct file
*file
, unsigned long addr
,
358 unsigned long len
, unsigned long prot
,
359 unsigned long flag
, unsigned long pgoff
)
362 struct mm_struct
*mm
= current
->mm
;
363 unsigned long populate
;
365 ret
= security_mmap_file(file
, prot
, flag
);
367 down_write(&mm
->mmap_sem
);
368 ret
= do_mmap_pgoff(file
, addr
, len
, prot
, flag
, pgoff
,
370 up_write(&mm
->mmap_sem
);
372 mm_populate(ret
, populate
);
377 unsigned long vm_mmap(struct file
*file
, unsigned long addr
,
378 unsigned long len
, unsigned long prot
,
379 unsigned long flag
, unsigned long offset
)
381 if (unlikely(offset
+ PAGE_ALIGN(len
) < offset
))
383 if (unlikely(offset
& ~PAGE_MASK
))
386 return vm_mmap_pgoff(file
, addr
, len
, prot
, flag
, offset
>> PAGE_SHIFT
);
388 EXPORT_SYMBOL(vm_mmap
);
390 struct address_space
*page_mapping(struct page
*page
)
392 struct address_space
*mapping
= page
->mapping
;
394 /* This happens if someone calls flush_dcache_page on slab page */
395 if (unlikely(PageSlab(page
)))
398 if (unlikely(PageSwapCache(page
))) {
401 entry
.val
= page_private(page
);
402 mapping
= swap_address_space(entry
);
403 } else if ((unsigned long)mapping
& PAGE_MAPPING_ANON
)
408 int overcommit_ratio_handler(struct ctl_table
*table
, int write
,
409 void __user
*buffer
, size_t *lenp
,
414 ret
= proc_dointvec(table
, write
, buffer
, lenp
, ppos
);
415 if (ret
== 0 && write
)
416 sysctl_overcommit_kbytes
= 0;
420 int overcommit_kbytes_handler(struct ctl_table
*table
, int write
,
421 void __user
*buffer
, size_t *lenp
,
426 ret
= proc_doulongvec_minmax(table
, write
, buffer
, lenp
, ppos
);
427 if (ret
== 0 && write
)
428 sysctl_overcommit_ratio
= 0;
433 * Committed memory limit enforced when OVERCOMMIT_NEVER policy is used
435 unsigned long vm_commit_limit(void)
437 unsigned long allowed
;
439 if (sysctl_overcommit_kbytes
)
440 allowed
= sysctl_overcommit_kbytes
>> (PAGE_SHIFT
- 10);
442 allowed
= ((totalram_pages
- hugetlb_total_pages())
443 * sysctl_overcommit_ratio
/ 100);
444 allowed
+= total_swap_pages
;
450 /* Tracepoints definitions. */
451 EXPORT_TRACEPOINT_SYMBOL(kmalloc
);
452 EXPORT_TRACEPOINT_SYMBOL(kmem_cache_alloc
);
453 EXPORT_TRACEPOINT_SYMBOL(kmalloc_node
);
454 EXPORT_TRACEPOINT_SYMBOL(kmem_cache_alloc_node
);
455 EXPORT_TRACEPOINT_SYMBOL(kfree
);
456 EXPORT_TRACEPOINT_SYMBOL(kmem_cache_free
);