1 # SOME DESCRIPTIVE TITLE
2 # Copyright (C) YEAR Free Software Foundation, Inc.
3 # This file is distributed under the same license as the PACKAGE package.
4 # FIRST AUTHOR <EMAIL@ADDRESS>, YEAR.
9 "Project-Id-Version: PACKAGE VERSION\n"
10 "POT-Creation-Date: 2012-04-23 07:53+0900\n"
11 "PO-Revision-Date: YEAR-MO-DA HO:MI+ZONE\n"
12 "Last-Translator: FULL NAME <EMAIL@ADDRESS>\n"
13 "Language-Team: LANGUAGE <LL@li.org>\n"
16 "Content-Type: text/plain; charset=UTF-8\n"
17 "Content-Transfer-Encoding: 8bit\n"
20 #: build/C/man2/alloc_hugepages.2:25
22 msgid "ALLOC_HUGEPAGES"
26 #: build/C/man2/alloc_hugepages.2:25
32 #: build/C/man2/alloc_hugepages.2:25 build/C/man2/cacheflush.2:24 build/C/man2/fallocate.2:8 build/C/man2/madvise.2:35 build/C/man3/mallopt.3:24 build/C/man2/mlock.2:27 build/C/man2/mmap.2:40 build/C/man2/mmap2.2:29 build/C/man2/mprotect.2:32 build/C/man2/mremap.2:31 build/C/man2/msync.2:25 build/C/man2/posix_fadvise.2:27 build/C/man2/readahead.2:28 build/C/man2/remap_file_pages.2:26 build/C/man3/shm_open.3:26 build/C/man7/shm_overview.7:27 build/C/man2/shmctl.2:43 build/C/man2/shmget.2:36 build/C/man2/shmop.2:39 build/C/man2/subpage_prot.2:28 build/C/man2/sync_file_range.2:30
38 #: build/C/man2/alloc_hugepages.2:25 build/C/man3/alloca.3:41 build/C/man2/cacheflush.2:24 build/C/man2/fallocate.2:8 build/C/man2/madvise.2:35 build/C/man3/malloc_hook.3:6 build/C/man3/mallopt.3:24 build/C/man2/mlock.2:27 build/C/man2/mmap.2:40 build/C/man2/mmap2.2:29 build/C/man2/mprotect.2:32 build/C/man2/mremap.2:31 build/C/man2/msync.2:25 build/C/man3/mtrace.3:23 build/C/man2/posix_fadvise.2:27 build/C/man3/posix_fallocate.3:23 build/C/man3/posix_memalign.3:28 build/C/man2/readahead.2:28 build/C/man2/remap_file_pages.2:26 build/C/man3/shm_open.3:26 build/C/man7/shm_overview.7:27 build/C/man2/shmctl.2:43 build/C/man2/shmget.2:36 build/C/man2/shmop.2:39 build/C/man2/subpage_prot.2:28 build/C/man2/sync_file_range.2:30
40 msgid "Linux Programmer's Manual"
44 #: build/C/man2/alloc_hugepages.2:26 build/C/man3/alloca.3:42 build/C/man2/cacheflush.2:25 build/C/man2/fallocate.2:9 build/C/man2/madvise.2:36 build/C/man3/malloc_hook.3:7 build/C/man3/mallopt.3:25 build/C/man2/mlock.2:28 build/C/man2/mmap.2:41 build/C/man2/mmap2.2:30 build/C/man2/mprotect.2:33 build/C/man2/mremap.2:32 build/C/man2/msync.2:26 build/C/man3/mtrace.3:24 build/C/man2/posix_fadvise.2:28 build/C/man3/posix_fallocate.3:24 build/C/man3/posix_memalign.3:29 build/C/man2/readahead.2:29 build/C/man2/remap_file_pages.2:27 build/C/man3/shm_open.3:27 build/C/man7/shm_overview.7:28 build/C/man2/shmctl.2:44 build/C/man2/shmget.2:37 build/C/man2/shmop.2:40 build/C/man2/subpage_prot.2:29 build/C/man2/sync_file_range.2:31
50 #: build/C/man2/alloc_hugepages.2:28
51 msgid "alloc_hugepages, free_hugepages - allocate or free huge pages"
55 #: build/C/man2/alloc_hugepages.2:28 build/C/man3/alloca.3:44 build/C/man2/cacheflush.2:27 build/C/man2/fallocate.2:11 build/C/man2/madvise.2:38 build/C/man3/malloc_hook.3:11 build/C/man3/mallopt.3:27 build/C/man2/mlock.2:30 build/C/man2/mmap.2:43 build/C/man2/mmap2.2:32 build/C/man2/mprotect.2:35 build/C/man2/mremap.2:34 build/C/man2/msync.2:28 build/C/man3/mtrace.3:26 build/C/man2/posix_fadvise.2:30 build/C/man3/posix_fallocate.3:26 build/C/man3/posix_memalign.3:31 build/C/man2/readahead.2:31 build/C/man2/remap_file_pages.2:29 build/C/man3/shm_open.3:29 build/C/man2/shmctl.2:46 build/C/man2/shmget.2:39 build/C/man2/shmop.2:42 build/C/man2/subpage_prot.2:31 build/C/man2/sync_file_range.2:33
60 #. asmlinkage unsigned long sys_alloc_hugepages(int key, unsigned long addr,
61 #. unsigned long len, int prot, int flag);
63 #: build/C/man2/alloc_hugepages.2:34
66 "B<void *alloc_hugepages(int >I<key>B<, void *>I<addr>B<, size_t "
68 "B< int >I<prot>B<, int >I<flag>B<);>\n"
71 #. asmlinkage int sys_free_hugepages(unsigned long addr);
73 #: build/C/man2/alloc_hugepages.2:37
75 msgid "B<int free_hugepages(void *>I<addr>B<);>\n"
79 #: build/C/man2/alloc_hugepages.2:38 build/C/man3/alloca.3:48 build/C/man2/cacheflush.2:33 build/C/man2/fallocate.2:19 build/C/man2/madvise.2:50 build/C/man3/malloc_hook.3:29 build/C/man3/mallopt.3:31 build/C/man2/mlock.2:40 build/C/man2/mmap.2:52 build/C/man2/mmap2.2:39 build/C/man2/mprotect.2:41 build/C/man2/mremap.2:43 build/C/man2/msync.2:32 build/C/man3/mtrace.3:32 build/C/man2/posix_fadvise.2:49 build/C/man3/posix_fallocate.3:44 build/C/man3/posix_memalign.3:83 build/C/man2/readahead.2:38 build/C/man2/remap_file_pages.2:37 build/C/man3/shm_open.3:41 build/C/man7/shm_overview.7:30 build/C/man2/shmctl.2:54 build/C/man2/shmget.2:47 build/C/man2/shmop.2:51 build/C/man2/subpage_prot.2:36 build/C/man2/sync_file_range.2:41
85 #: build/C/man2/alloc_hugepages.2:49
87 "The system calls B<alloc_hugepages>() and B<free_hugepages>() were "
88 "introduced in Linux 2.5.36 and removed again in 2.5.54. They existed only "
89 "on i386 and ia64 (when built with B<CONFIG_HUGETLB_PAGE>). In Linux 2.4.20 "
90 "the syscall numbers exist, but the calls fail with the error B<ENOSYS>."
94 #: build/C/man2/alloc_hugepages.2:57
96 "On i386 the memory management hardware knows about ordinary pages (4 KiB) "
97 "and huge pages (2 or 4 MiB). Similarly ia64 knows about huge pages of "
98 "several sizes. These system calls serve to map huge pages into the "
99 "process's memory or to free them again. Huge pages are locked into memory, "
100 "and are not swapped."
104 #: build/C/man2/alloc_hugepages.2:66
106 "The I<key> argument is an identifier. When zero the pages are private, and "
107 "not inherited by children. When positive the pages are shared with other "
108 "applications using the same I<key>, and inherited by child processes."
112 #: build/C/man2/alloc_hugepages.2:81
114 "The I<addr> argument of B<free_hugepages>() tells which page is being "
115 "freed: it was the return value of a call to B<alloc_hugepages>(). (The "
116 "memory is first actually freed when all users have released it.) The "
117 "I<addr> argument of B<alloc_hugepages>() is a hint, that the kernel may or "
118 "may not follow. Addresses must be properly aligned."
122 #: build/C/man2/alloc_hugepages.2:86
124 "The I<len> argument is the length of the required segment. It must be a "
125 "multiple of the huge page size."
129 #: build/C/man2/alloc_hugepages.2:94
131 "The I<prot> argument specifies the memory protection of the segment. It is "
132 "one of B<PROT_READ>, B<PROT_WRITE>, B<PROT_EXEC>."
136 #: build/C/man2/alloc_hugepages.2:109
138 "The I<flag> argument is ignored, unless I<key> is positive. In that case, "
139 "if I<flag> is B<IPC_CREAT>, then a new huge page segment is created when "
140 "none with the given key existed. If this flag is not set, then B<ENOENT> is "
141 "returned when no segment with the given key exists."
145 #: build/C/man2/alloc_hugepages.2:109 build/C/man3/alloca.3:58 build/C/man2/cacheflush.2:52 build/C/man2/fallocate.2:109 build/C/man2/madvise.2:250 build/C/man3/mallopt.3:377 build/C/man2/mlock.2:117 build/C/man2/mmap.2:373 build/C/man2/mmap2.2:50 build/C/man2/mprotect.2:72 build/C/man2/mremap.2:128 build/C/man2/msync.2:68 build/C/man2/posix_fadvise.2:86 build/C/man3/posix_fallocate.3:64 build/C/man3/posix_memalign.3:141 build/C/man2/readahead.2:65 build/C/man2/remap_file_pages.2:122 build/C/man3/shm_open.3:169 build/C/man2/shmctl.2:272 build/C/man2/shmget.2:178 build/C/man2/shmop.2:169 build/C/man2/subpage_prot.2:61 build/C/man2/sync_file_range.2:141
151 #: build/C/man2/alloc_hugepages.2:118
153 "On success, B<alloc_hugepages>() returns the allocated virtual address, and "
154 "B<free_hugepages>() returns zero. On error, -1 is returned, and I<errno> "
155 "is set appropriately."
159 #: build/C/man2/alloc_hugepages.2:118 build/C/man2/cacheflush.2:58 build/C/man2/fallocate.2:112 build/C/man2/madvise.2:257 build/C/man3/mallopt.3:382 build/C/man2/mlock.2:123 build/C/man2/mmap.2:390 build/C/man2/mmap2.2:57 build/C/man2/mprotect.2:79 build/C/man2/mremap.2:136 build/C/man2/msync.2:73 build/C/man2/posix_fadvise.2:89 build/C/man3/posix_fallocate.3:70 build/C/man3/posix_memalign.3:155 build/C/man2/readahead.2:71 build/C/man2/remap_file_pages.2:129 build/C/man3/shm_open.3:178 build/C/man2/shmctl.2:294 build/C/man2/shmget.2:182 build/C/man2/shmop.2:183 build/C/man2/subpage_prot.2:66 build/C/man2/sync_file_range.2:147
165 #: build/C/man2/alloc_hugepages.2:119 build/C/man2/fallocate.2:151
171 #: build/C/man2/alloc_hugepages.2:122
172 msgid "The system call is not supported on this kernel."
176 #: build/C/man2/alloc_hugepages.2:122
182 #: build/C/man2/alloc_hugepages.2:126
184 "I</proc/sys/vm/nr_hugepages> Number of configured hugetlb pages. This can "
185 "be read and written."
189 #: build/C/man2/alloc_hugepages.2:130
191 "I</proc/meminfo> Gives info on the number of configured hugetlb pages and on "
192 "their size in the three variables HugePages_Total, HugePages_Free, "
197 #: build/C/man2/alloc_hugepages.2:130 build/C/man3/alloca.3:63 build/C/man2/cacheflush.2:74 build/C/man2/fallocate.2:186 build/C/man2/madvise.2:307 build/C/man3/malloc_hook.3:76 build/C/man3/mallopt.3:390 build/C/man2/mlock.2:187 build/C/man2/mmap.2:483 build/C/man2/mmap2.2:73 build/C/man2/mprotect.2:107 build/C/man2/mremap.2:186 build/C/man2/msync.2:96 build/C/man3/mtrace.3:76 build/C/man2/posix_fadvise.2:111 build/C/man3/posix_fallocate.3:101 build/C/man3/posix_memalign.3:180 build/C/man2/readahead.2:87 build/C/man2/remap_file_pages.2:151 build/C/man3/shm_open.3:249 build/C/man7/shm_overview.7:101 build/C/man2/shmctl.2:366 build/C/man2/shmget.2:229 build/C/man2/shmop.2:226 build/C/man2/subpage_prot.2:91 build/C/man2/sync_file_range.2:180
199 msgid "CONFORMING TO"
203 #: build/C/man2/alloc_hugepages.2:133
205 "These calls are specific to Linux on Intel processors, and should not be "
206 "used in programs intended to be portable."
210 #: build/C/man2/alloc_hugepages.2:133 build/C/man3/alloca.3:71 build/C/man2/madvise.2:328 build/C/man3/malloc_hook.3:78 build/C/man2/mlock.2:214 build/C/man2/mmap.2:501 build/C/man2/mmap2.2:75 build/C/man2/mprotect.2:116 build/C/man2/mremap.2:192 build/C/man3/mtrace.3:78 build/C/man2/posix_fadvise.2:120 build/C/man3/posix_memalign.3:219 build/C/man3/shm_open.3:255 build/C/man7/shm_overview.7:103 build/C/man2/shmctl.2:371 build/C/man2/shmget.2:235 build/C/man2/shmop.2:245 build/C/man2/subpage_prot.2:93 build/C/man2/sync_file_range.2:183
216 #: build/C/man2/alloc_hugepages.2:141
218 "These system calls are gone; they existed only in Linux 2.5.36 through to "
219 "2.5.54. Now the hugetlbfs file system can be used instead. Memory backed "
220 "by huge pages (if the CPU supports them) is obtained by using B<mmap>(2) to "
221 "map files in this virtual file system."
225 #: build/C/man2/alloc_hugepages.2:145
227 "The maximal number of huge pages can be specified using the B<hugepages=> "
232 #: build/C/man2/alloc_hugepages.2:150 build/C/man3/alloca.3:153 build/C/man2/cacheflush.2:88 build/C/man2/fallocate.2:193 build/C/man2/madvise.2:360 build/C/man3/malloc_hook.3:138 build/C/man3/mallopt.3:578 build/C/man2/mlock.2:337 build/C/man2/mmap.2:667 build/C/man2/mmap2.2:94 build/C/man2/mprotect.2:223 build/C/man2/mremap.2:214 build/C/man2/msync.2:122 build/C/man3/mtrace.3:169 build/C/man2/posix_fadvise.2:188 build/C/man3/posix_fallocate.3:130 build/C/man3/posix_memalign.3:275 build/C/man2/readahead.2:98 build/C/man2/remap_file_pages.2:162 build/C/man3/shm_open.3:280 build/C/man7/shm_overview.7:127 build/C/man2/shmctl.2:410 build/C/man2/shmget.2:298 build/C/man2/shmop.2:290 build/C/man2/subpage_prot.2:126 build/C/man2/sync_file_range.2:222
238 #: build/C/man2/alloc_hugepages.2:157 build/C/man3/alloca.3:160 build/C/man2/cacheflush.2:95 build/C/man2/fallocate.2:200 build/C/man2/madvise.2:367 build/C/man3/malloc_hook.3:145 build/C/man3/mallopt.3:585 build/C/man2/mlock.2:344 build/C/man2/mmap.2:674 build/C/man2/mmap2.2:101 build/C/man2/mprotect.2:230 build/C/man2/mremap.2:221 build/C/man2/msync.2:129 build/C/man3/mtrace.3:176 build/C/man2/posix_fadvise.2:195 build/C/man3/posix_fallocate.3:137 build/C/man3/posix_memalign.3:282 build/C/man2/readahead.2:105 build/C/man2/remap_file_pages.2:169 build/C/man3/shm_open.3:287 build/C/man7/shm_overview.7:134 build/C/man2/shmctl.2:417 build/C/man2/shmget.2:305 build/C/man2/shmop.2:297 build/C/man2/subpage_prot.2:133 build/C/man2/sync_file_range.2:229
240 "This page is part of release 3.38 of the Linux I<man-pages> project. A "
241 "description of the project, and information about reporting bugs, can be "
242 "found at http://www.kernel.org/doc/man-pages/."
246 #: build/C/man3/alloca.3:41
252 #: build/C/man3/alloca.3:41
258 #: build/C/man3/alloca.3:41 build/C/man3/malloc_hook.3:6 build/C/man3/mtrace.3:23 build/C/man3/posix_fallocate.3:23 build/C/man3/posix_memalign.3:28
264 #: build/C/man3/alloca.3:44
265 msgid "alloca - allocate memory that is automatically freed"
269 #: build/C/man3/alloca.3:46
270 msgid "B<#include E<lt>alloca.hE<gt>>"
274 #: build/C/man3/alloca.3:48
275 msgid "B<void *alloca(size_t >I<size>B<);>"
279 #: build/C/man3/alloca.3:58
281 "The B<alloca>() function allocates I<size> bytes of space in the stack "
282 "frame of the caller. This temporary space is automatically freed when the "
283 "function that called B<alloca>() returns to its caller."
287 #: build/C/man3/alloca.3:63
289 "The B<alloca>() function returns a pointer to the beginning of the "
290 "allocated space. If the allocation causes stack overflow, program behavior "
295 #: build/C/man3/alloca.3:65
296 msgid "This function is not in POSIX.1-2001."
300 #: build/C/man3/alloca.3:71
302 "There is evidence that the B<alloca>() function appeared in 32V, PWB, "
303 "PWB.2, 3BSD, and 4BSD. There is a man page for it in 4.3BSD. Linux uses "
308 #: build/C/man3/alloca.3:86
310 "The B<alloca>() function is machine- and compiler-dependent. For certain "
311 "applications, its use can improve efficiency compared to the use of "
312 "B<malloc>(3) plus B<free>(3). In certain cases, it can also simplify "
313 "memory deallocation in applications that use B<longjmp>(3) or "
314 "B<siglongjmp>(3). Otherwise, its use is discouraged."
318 #: build/C/man3/alloca.3:95
320 "Because the space allocated by B<alloca>() is allocated within the stack "
321 "frame, that space is automatically freed if the function return is jumped "
322 "over by a call to B<longjmp>(3) or B<siglongjmp>(3)."
326 #: build/C/man3/alloca.3:100
327 msgid "Do not attempt to B<free>(3) space allocated by B<alloca>()!"
331 #: build/C/man3/alloca.3:100
333 msgid "Notes on the GNU Version"
337 #: build/C/man3/alloca.3:122
339 "Normally, B<gcc>(1) translates calls to B<alloca>() with inlined code. "
340 "This is not done when either the I<-ansi>, I<-std=c89>, I<-std=c99>, or the "
341 "I<-fno-builtin> option is given (and the header I<E<lt>alloca.hE<gt>> is not "
342 "included). But beware! By default the glibc version of "
343 "I<E<lt>stdlib.hE<gt>> includes I<E<lt>alloca.hE<gt>> and that contains the "
348 #: build/C/man3/alloca.3:125
350 msgid " #define alloca(size) __builtin_alloca (size)\n"
354 #: build/C/man3/alloca.3:128
355 msgid "with messy consequences if one has a private version of this function."
359 #: build/C/man3/alloca.3:132
361 "The fact that the code is inlined means that it is impossible to take the "
362 "address of this function, or to change its behavior by linking with a "
367 #: build/C/man3/alloca.3:136
369 "The inlined code often consists of a single instruction adjusting the stack "
370 "pointer, and does not check for stack overflow. Thus, there is no NULL "
375 #: build/C/man3/alloca.3:136 build/C/man2/cacheflush.2:81 build/C/man3/mallopt.3:404 build/C/man2/mlock.2:305 build/C/man2/mmap.2:540 build/C/man3/mtrace.3:100 build/C/man2/posix_fadvise.2:177 build/C/man2/shmget.2:284
381 #: build/C/man3/alloca.3:141
383 "There is no error indication if the stack frame cannot be extended. "
384 "(However, after a failed allocation, the program is likely to receive a "
385 "B<SIGSEGV> signal if it attempts to access the unallocated space.)"
389 #: build/C/man3/alloca.3:149
391 "On many systems B<alloca>() cannot be used inside the list of arguments of "
392 "a function call, because the stack space reserved by B<alloca>() would "
393 "appear on the stack in the middle of the space for the function arguments."
397 #: build/C/man3/alloca.3:149 build/C/man2/fallocate.2:189 build/C/man2/madvise.2:353 build/C/man3/malloc_hook.3:133 build/C/man3/mallopt.3:568 build/C/man2/mlock.2:330 build/C/man2/mmap.2:649 build/C/man2/mmap2.2:88 build/C/man2/mprotect.2:220 build/C/man2/mremap.2:200 build/C/man2/msync.2:118 build/C/man3/mtrace.3:165 build/C/man2/posix_fadvise.2:182 build/C/man3/posix_fallocate.3:126 build/C/man3/posix_memalign.3:270 build/C/man2/readahead.2:92 build/C/man2/remap_file_pages.2:155 build/C/man3/shm_open.3:269 build/C/man7/shm_overview.7:114 build/C/man2/shmctl.2:403 build/C/man2/shmget.2:290 build/C/man2/shmop.2:282 build/C/man2/subpage_prot.2:120 build/C/man2/sync_file_range.2:217
403 #: build/C/man3/alloca.3:153
404 msgid "B<brk>(2), B<longjmp>(3), B<malloc>(3)"
408 #: build/C/man2/cacheflush.2:24
414 #: build/C/man2/cacheflush.2:24
420 #: build/C/man2/cacheflush.2:27
421 msgid "cacheflush - flush contents of instruction and/or data cache"
425 #: build/C/man2/cacheflush.2:30
427 msgid "B<#include E<lt>asm/cachectl.hE<gt>>\n"
431 #: build/C/man2/cacheflush.2:32
433 msgid "B<int cacheflush(char *>I<addr>B<, int >I<nbytes>B<, int >I<cache>B<);>\n"
437 #: build/C/man2/cacheflush.2:42
439 "B<cacheflush>() flushes the contents of the indicated cache(s) for the user "
440 "addresses in the range I<addr> to I<(addr+nbytes-1)>. I<cache> may be one "
445 #: build/C/man2/cacheflush.2:42
451 #: build/C/man2/cacheflush.2:45
452 msgid "Flush the instruction cache."
456 #: build/C/man2/cacheflush.2:45
462 #: build/C/man2/cacheflush.2:48
463 msgid "Write back to memory and invalidate the affected valid cache lines."
467 #: build/C/man2/cacheflush.2:48
473 #: build/C/man2/cacheflush.2:52
474 msgid "Same as B<(ICACHE|DCACHE)>."
478 #: build/C/man2/cacheflush.2:58
480 "B<cacheflush>() returns 0 on success or -1 on error. If errors are "
481 "detected, I<errno> will indicate the error."
485 #: build/C/man2/cacheflush.2:59 build/C/man2/mmap2.2:58 build/C/man2/mremap.2:143 build/C/man2/shmctl.2:303 build/C/man2/subpage_prot.2:67
491 #: build/C/man2/cacheflush.2:66
493 "Some or all of the address range I<addr> to I<(addr+nbytes-1)> is not "
498 #: build/C/man2/cacheflush.2:66 build/C/man2/fallocate.2:124 build/C/man2/madvise.2:264 build/C/man2/mlock.2:158 build/C/man2/mlock.2:165 build/C/man2/mlock.2:177 build/C/man2/mmap.2:421 build/C/man2/mmap.2:429 build/C/man2/mmap.2:434 build/C/man2/mmap2.2:61 build/C/man2/mprotect.2:89 build/C/man2/mremap.2:152 build/C/man2/msync.2:80 build/C/man2/posix_fadvise.2:93 build/C/man3/posix_fallocate.3:79 build/C/man3/posix_memalign.3:156 build/C/man2/readahead.2:76 build/C/man2/remap_file_pages.2:130 build/C/man2/remap_file_pages.2:137 build/C/man3/shm_open.3:211 build/C/man2/shmctl.2:317 build/C/man2/shmget.2:196 build/C/man2/shmop.2:195 build/C/man2/shmop.2:218 build/C/man2/subpage_prot.2:72 build/C/man2/sync_file_range.2:152
504 #: build/C/man2/cacheflush.2:74
505 msgid "I<cache> is not one of B<ICACHE>, B<DCACHE>, or B<BCACHE>."
508 #. FIXME This system call was only on MIPS back in 1.2 days, but
509 #. by now it is on a number of other architectures (but not i386).
510 #. Investigate the details and update this page.
511 #. Irix 6.5 appears to have a cacheflush() syscall -- mtk
513 #: build/C/man2/cacheflush.2:81
515 "This Linux-specific system call is only available on MIPS-based systems. It "
516 "should not be used in programs intended to be portable."
520 #: build/C/man2/cacheflush.2:88
522 "The current implementation ignores the I<addr> and I<nbytes> arguments. "
523 "Therefore, the whole cache is always flushed."
527 #: build/C/man2/fallocate.2:8
533 #: build/C/man2/fallocate.2:8
539 #: build/C/man2/fallocate.2:11
540 msgid "fallocate - manipulate file space"
544 #: build/C/man2/fallocate.2:15 build/C/man2/readahead.2:35
547 "B<#define _GNU_SOURCE> /* See feature_test_macros(7) */\n"
548 "B<#include E<lt>fcntl.hE<gt>>\n"
552 #: build/C/man2/fallocate.2:18
555 "B<int fallocate(int >I<fd>B<, int >I<mode>B<, off_t >I<offset>B<, off_t "
560 #: build/C/man2/fallocate.2:24
562 "This is a nonportable, Linux-specific system call. For the portable, "
563 "POSIX.1-specified method of ensuring that space is allocated for a file, see "
564 "B<posix_fallocate>(3)."
568 #: build/C/man2/fallocate.2:34
570 "B<fallocate>() allows the caller to directly manipulate the allocated disk "
571 "space for the file referred to by I<fd> for the byte range starting at "
572 "I<offset> and continuing for I<len> bytes."
576 #: build/C/man2/fallocate.2:39
578 "The I<mode> argument determines the operation to be performed on the given "
579 "range. Details of the supported operations are given in the subsections "
584 #: build/C/man2/fallocate.2:39
586 msgid "Allocating disk space"
590 #: build/C/man2/fallocate.2:58
592 "The default operation (i.e., I<mode> is zero) of B<fallocate>() allocates "
593 "and initializes to zero the disk space within the range specified by "
594 "I<offset> and I<len>. The file size (as reported by B<stat>(2)) will be "
595 "changed if I<offset>+I<len> is greater than the file size. This default "
596 "behavior closely resembles the behavior of the B<posix_fallocate>(3) "
597 "library function, and is intended as a method of optimally implementing that "
602 #: build/C/man2/fallocate.2:64
604 "After a successful call, subsequent writes into the range specified by "
605 "I<offset> and I<len> are guaranteed not to fail because of lack of disk "
610 #: build/C/man2/fallocate.2:75
612 "If the B<FALLOC_FL_KEEP_SIZE> flag is specified in I<mode>, the behavior of "
613 "the call is similar, but the file size will not be changed even if "
614 "I<offset>+I<len> is greater than the file size. Preallocating zeroed blocks "
615 "beyond the end of the file in this manner is useful for optimizing append "
620 #: build/C/man2/fallocate.2:79
622 "Because allocation is done in block size chunks, B<fallocate>() may "
623 "allocate a larger range of disk space than was specified."
627 #: build/C/man2/fallocate.2:79
629 msgid "Deallocating file space"
633 #: build/C/man2/fallocate.2:94
635 "Specifying the B<FALLOC_FL_PUNCH_HOLE> flag (available since Linux 2.6.38) "
636 "in I<mode> deallocates space (i.e., creates a hole) in the byte range "
637 "starting at I<offset> and continuing for I<len> bytes. Within the specified "
638 "range, partial file system blocks are zeroed, and whole file system blocks "
639 "are removed from the file. After a successful call, subsequent reads from "
640 "this range will return zeroes."
644 #: build/C/man2/fallocate.2:105
646 "The B<FALLOC_FL_PUNCH_HOLE> flag must be ORed with B<FALLOC_FL_KEEP_SIZE> in "
647 "I<mode>; in other words, even when punching off the end of the file, the "
648 "file size (as reported by B<stat>(2)) does not change."
652 #: build/C/man2/fallocate.2:109
654 "Not all file systems support B<FALLOC_FL_PUNCH_HOLE>; if a file system "
655 "doesn't support the operation, an error is returned."
659 #: build/C/man2/fallocate.2:112
660 msgid "B<fallocate>() returns zero on success, and -1 on failure."
664 #: build/C/man2/fallocate.2:113 build/C/man2/madvise.2:261 build/C/man2/mmap.2:415 build/C/man2/posix_fadvise.2:90 build/C/man3/posix_fallocate.3:71 build/C/man2/readahead.2:72 build/C/man2/sync_file_range.2:148
670 #: build/C/man2/fallocate.2:117 build/C/man3/posix_fallocate.3:75
671 msgid "I<fd> is not a valid file descriptor, or is not opened for writing."
675 #: build/C/man2/fallocate.2:117 build/C/man3/posix_fallocate.3:75
681 #: build/C/man2/fallocate.2:121
682 msgid "I<offset>+I<len> exceeds the maximum file size."
686 #: build/C/man2/fallocate.2:121
692 #: build/C/man2/fallocate.2:124
693 msgid "A signal was caught during execution."
697 #: build/C/man2/fallocate.2:136 build/C/man3/posix_fallocate.3:85
698 msgid "I<offset> was less than 0, or I<len> was less than or equal to 0."
702 #: build/C/man2/fallocate.2:136 build/C/man2/madvise.2:292 build/C/man2/sync_file_range.2:160
708 #: build/C/man2/fallocate.2:139
709 msgid "An I/O error occurred while reading from or writing to a file system."
713 #: build/C/man2/fallocate.2:139 build/C/man2/mmap.2:450 build/C/man3/posix_fallocate.3:85
719 #: build/C/man2/fallocate.2:146
721 "I<fd> does not refer to a regular file or a directory. (If I<fd> is a pipe "
722 "or FIFO, a different error results.)"
726 #: build/C/man2/fallocate.2:146 build/C/man3/posix_fallocate.3:89 build/C/man2/shmget.2:214 build/C/man2/sync_file_range.2:166
732 #: build/C/man2/fallocate.2:151 build/C/man3/posix_fallocate.3:94
734 "There is not enough space left on the device containing the file referred to "
739 #: build/C/man2/fallocate.2:156
741 "The file system containing the file referred to by I<fd> does not support "
746 #: build/C/man2/fallocate.2:156
748 msgid "B<EOPNOTSUPP>"
752 #: build/C/man2/fallocate.2:162
754 "The I<mode> is not supported by the file system containing the file referred "
759 #: build/C/man2/fallocate.2:162 build/C/man2/mlock.2:144 build/C/man2/mlock.2:183 build/C/man2/mmap.2:458 build/C/man2/shmctl.2:342 build/C/man2/shmget.2:222
765 #: build/C/man2/fallocate.2:178
767 "The file referred to by I<fd> is marked immutable (see B<chattr>(1)). Or: "
768 "I<mode> specifies B<FALLOC_FL_PUNCH_HOLE> and the file referred to by I<fd> "
769 "is marked append-only (see B<chattr>(1))."
773 #: build/C/man2/fallocate.2:178 build/C/man2/posix_fadvise.2:96 build/C/man3/posix_fallocate.3:94 build/C/man2/sync_file_range.2:169
779 #: build/C/man2/fallocate.2:182
780 msgid "I<fd> refers to a pipe or FIFO."
784 #: build/C/man2/fallocate.2:182 build/C/man2/mmap2.2:70 build/C/man2/posix_fadvise.2:103 build/C/man3/posix_fallocate.3:98 build/C/man3/posix_memalign.3:165 build/C/man2/readahead.2:82 build/C/man2/remap_file_pages.2:146 build/C/man3/shm_open.3:247 build/C/man2/subpage_prot.2:85 build/C/man2/sync_file_range.2:177
790 #: build/C/man2/fallocate.2:186
792 "B<fallocate>() is available on Linux since kernel 2.6.23. Support is "
793 "provided by glibc since version 2.10."
797 #: build/C/man2/fallocate.2:189
798 msgid "B<fallocate>() is Linux-specific."
802 #: build/C/man2/fallocate.2:193
803 msgid "B<ftruncate>(2), B<posix_fadvise>(3), B<posix_fallocate>(3)"
807 #: build/C/man2/madvise.2:35
813 #: build/C/man2/madvise.2:35
819 #: build/C/man2/madvise.2:38
820 msgid "madvise - give advice about use of memory"
824 #: build/C/man2/madvise.2:40 build/C/man2/msync.2:30 build/C/man3/shm_open.3:31
825 msgid "B<#include E<lt>sys/mman.hE<gt>>"
829 #: build/C/man2/madvise.2:42
830 msgid "B<int madvise(void *>I<addr>B<, size_t >I<length>B<, int >I<advice>B<);>"
834 #: build/C/man2/madvise.2:46 build/C/man2/posix_fadvise.2:42 build/C/man3/posix_fallocate.3:37 build/C/man3/posix_memalign.3:48
835 msgid "Feature Test Macro Requirements for glibc (see B<feature_test_macros>(7)):"
839 #: build/C/man2/madvise.2:50
840 msgid "B<madvise>(): _BSD_SOURCE"
844 #: build/C/man2/madvise.2:68
846 "The B<madvise>() system call advises the kernel about how to handle paging "
847 "input/output in the address range beginning at address I<addr> and with size "
848 "I<length> bytes. It allows an application to tell the kernel how it expects "
849 "to use some mapped or shared memory areas, so that the kernel can choose "
850 "appropriate read-ahead and caching techniques. This call does not influence "
851 "the semantics of the application (except in the case of B<MADV_DONTNEED>), "
852 "but may influence its performance. The kernel is free to ignore the advice."
856 #: build/C/man2/madvise.2:72
857 msgid "The advice is indicated in the I<advice> argument which can be"
861 #: build/C/man2/madvise.2:72
863 msgid "B<MADV_NORMAL>"
867 #: build/C/man2/madvise.2:76
868 msgid "No special treatment. This is the default."
872 #: build/C/man2/madvise.2:76
874 msgid "B<MADV_RANDOM>"
878 #: build/C/man2/madvise.2:80
880 "Expect page references in random order. (Hence, read ahead may be less "
881 "useful than normally.)"
885 #: build/C/man2/madvise.2:80
887 msgid "B<MADV_SEQUENTIAL>"
891 #: build/C/man2/madvise.2:85
893 "Expect page references in sequential order. (Hence, pages in the given "
894 "range can be aggressively read ahead, and may be freed soon after they are "
899 #: build/C/man2/madvise.2:85
901 msgid "B<MADV_WILLNEED>"
905 #: build/C/man2/madvise.2:89
907 "Expect access in the near future. (Hence, it might be a good idea to read "
912 #: build/C/man2/madvise.2:89
914 msgid "B<MADV_DONTNEED>"
918 #: build/C/man2/madvise.2:100
920 "Do not expect access in the near future. (For the time being, the "
921 "application is finished with the given range, so the kernel can free "
922 "resources associated with it.) Subsequent accesses of pages in this range "
923 "will succeed, but will result either in reloading of the memory contents "
924 "from the underlying mapped file (see B<mmap>(2)) or zero-fill-on-demand "
925 "pages for mappings without an underlying file."
929 #: build/C/man2/madvise.2:100
931 msgid "B<MADV_REMOVE> (Since Linux 2.6.16)"
935 #. Databases want to use this feature to drop a section of their
936 #. bufferpool (shared memory segments) - without writing back to
937 #. disk/swap space. This feature is also useful for supporting
938 #. hot-plug memory on UML.
940 #: build/C/man2/madvise.2:113
942 "Free up a given range of pages and its associated backing store. Currently, "
943 "only shmfs/tmpfs supports this; other file systems return with the error "
948 #: build/C/man2/madvise.2:113
950 msgid "B<MADV_DONTFORK> (Since Linux 2.6.16)"
953 #. See http://lwn.net/Articles/171941/
954 #. [PATCH] madvise MADV_DONTFORK/MADV_DOFORK
955 #. Currently, copy-on-write may change the physical address of
956 #. a page even if the user requested that the page is pinned in
957 #. memory (either by mlock or by get_user_pages). This happens
958 #. if the process forks meanwhile, and the parent writes to that
959 #. page. As a result, the page is orphaned: in case of
960 #. get_user_pages, the application will never see any data hardware
961 #. DMA's into this page after the COW. In case of mlock'd memory,
962 #. the parent is not getting the realtime/security benefits of mlock.
964 #. In particular, this affects the Infiniband modules which do DMA from
965 #. and into user pages all the time.
967 #. This patch adds madvise options to control whether memory range is
968 #. inherited across fork. Useful e.g. for when hardware is doing DMA
969 #. from/into these pages. Could also be useful to an application
970 #. wanting to speed up its forks by cutting large areas out of
973 #. SEE ALSO: http://lwn.net/Articles/171941/
974 #. "Tweaks to madvise() and posix_fadvise()", 14 Feb 2006
976 #: build/C/man2/madvise.2:144
978 "Do not make the pages in this range available to the child after a "
979 "B<fork>(2). This is useful to prevent copy-on-write semantics from changing "
980 "the physical location of a page(s) if the parent writes to it after a "
981 "B<fork>(2). (Such page relocations cause problems for hardware that DMAs "
986 #: build/C/man2/madvise.2:144
988 msgid "B<MADV_DOFORK> (Since Linux 2.6.16)"
992 #: build/C/man2/madvise.2:150
994 "Undo the effect of B<MADV_DONTFORK>, restoring the default behavior, whereby "
995 "a mapping is inherited across B<fork>(2)."
999 #: build/C/man2/madvise.2:150
1001 msgid "B<MADV_HWPOISON> (Since Linux 2.6.32)"
1005 #: build/C/man2/madvise.2:162
1007 "Poison a page and handle it like a hardware memory corruption. This "
1008 "operation is only available for privileged (B<CAP_SYS_ADMIN>) processes. "
1009 "This operation may result in the calling process receiving a B<SIGBUS> and "
1010 "the page being unmapped. This feature is intended for testing of memory "
1011 "error-handling code; it is only available if the kernel was configured with "
1012 "B<CONFIG_MEMORY_FAILURE>."
1016 #: build/C/man2/madvise.2:162
1018 msgid "B<MADV_SOFT_OFFLINE> (Since Linux 2.6.33)"
1022 #: build/C/man2/madvise.2:180
1024 "Soft offline the pages in the range specified by I<addr> and I<length>. The "
1025 "memory of each page in the specified range is preserved (i.e., when next "
1026 "accessed, the same content will be visible, but in a new physical page "
1027 "frame), and the original page is offlined (i.e., no longer used, and taken "
1028 "out of normal memory management). The effect of the B<MADV_SOFT_OFFLINE> "
1029 "operation is invisible to (i.e., does not change the semantics of) the "
1030 "calling process. This feature is intended for testing of memory "
1031 "error-handling code; it is only available if the kernel was configured with "
1032 "B<CONFIG_MEMORY_FAILURE>."
1036 #: build/C/man2/madvise.2:180
1038 msgid "B<MADV_MERGEABLE> (since Linux 2.6.32)"
1042 #: build/C/man2/madvise.2:205
1044 "Enable Kernel Samepage Merging (KSM) for the pages in the range specified by "
1045 "I<addr> and I<length>. The kernel regularly scans those areas of user "
1046 "memory that have been marked as mergeable, looking for pages with identical "
1047 "content. These are replaced by a single write-protected page (which is "
1048 "automatically copied if a process later wants to update the content of the "
1049 "page). KSM only merges private anonymous pages (see B<mmap>(2)). The KSM "
1050 "feature is intended for applications that generate many instances of the "
1051 "same data (e.g., virtualization systems such as KVM). It can consume a lot "
1052 "of processing power; use with care. See the kernel source file "
1053 "I<Documentation/vm/ksm.txt> for more details. The B<MADV_MERGEABLE> and "
1054 "B<MADV_UNMERGEABLE> operations are only available if the kernel was "
1055 "configured with B<CONFIG_KSM>."
1059 #: build/C/man2/madvise.2:205
1061 msgid "B<MADV_UNMERGEABLE> (since Linux 2.6.32)"
1065 #: build/C/man2/madvise.2:214
1067 "Undo the effect of an earlier B<MADV_MERGEABLE> operation on the specified "
1068 "address range; KSM unmerges whatever pages it had merged in the address "
1069 "range specified by I<addr> and I<length>."
1073 #: build/C/man2/madvise.2:214
1075 msgid "B<MADV_HUGEPAGE> (since Linux 2.6.38)"
1078 #. http://lwn.net/Articles/358904/
1079 #. https://lwn.net/Articles/423584/
1081 #: build/C/man2/madvise.2:243
1083 "Enables Transparent Huge Pages (THP) for pages in the range specified by "
1084 "I<addr> and I<length>. Currently, Transparent Huge Pages only work with "
1085 "private anonymous pages (see B<mmap>(2)). The kernel will regularly scan "
1086 "the areas marked as huge page candidates to replace them with huge pages. "
1087 "The kernel will also allocate huge pages directly when the region is "
1088 "naturally aligned to the huge page size (see B<posix_memalign>(2)). This "
1089 "feature is primarily aimed at applications that use large mappings of data "
1090 "and access large regions of that memory at a time (e.g. virtualization "
1091 "systems such as QEMU). It can very easily waste memory (e.g. a 2MB mapping "
1092 "that only ever accesses 1 byte will result in 2MB of wired memory instead of "
1093 "one 4KB page). See the kernel source file I<Documentation/vm/transhuge.txt> "
1094 "for more details. The B<MADV_HUGEPAGE> and B<MADV_NOHUGEPAGE> operations "
1095 "are only available if the kernel was configured with "
1096 "B<CONFIG_TRANSPARENT_HUGEPAGE>."
1100 #: build/C/man2/madvise.2:243
1102 msgid "B<MADV_NOHUGEPAGE> (since Linux 2.6.38)"
1106 #: build/C/man2/madvise.2:250
1108 "Ensures that memory in the address range specified by I<addr> and I<length> "
1109 "will not be collapsed into huge pages."
1113 #: build/C/man2/madvise.2:257
1115 "On success B<madvise>() returns zero. On error, it returns -1 and I<errno> "
1116 "is set appropriately."
1120 #: build/C/man2/madvise.2:258 build/C/man2/mlock.2:155 build/C/man2/mmap.2:411 build/C/man2/mremap.2:137
1126 #: build/C/man2/madvise.2:261
1127 msgid "A kernel resource was temporarily unavailable."
1131 #: build/C/man2/madvise.2:264
1132 msgid "The map exists, but the area maps something that isn't a file."
1136 #: build/C/man2/madvise.2:267
1137 msgid "This error can occur for the following reasons:"
1141 #: build/C/man2/madvise.2:268 build/C/man2/madvise.2:274 build/C/man2/madvise.2:277 build/C/man2/madvise.2:280 build/C/man2/madvise.2:283 build/C/man3/mallopt.3:233 build/C/man3/mallopt.3:239
1149 #: build/C/man2/madvise.2:274
1150 msgid "The value I<len> is negative."
1154 #: build/C/man2/madvise.2:277
1155 msgid "I<addr> is not page-aligned."
1159 #: build/C/man2/madvise.2:280
1160 msgid "I<advice> is not a valid value"
1164 #: build/C/man2/madvise.2:283
1166 "The application is attempting to release locked or shared pages (with "
1167 "B<MADV_DONTNEED>)."
1171 #: build/C/man2/madvise.2:291
1173 "B<MADV_MERGEABLE> or B<MADV_UNMERGEABLE> was specified in I<advice>, but the "
1174 "kernel was not configured with B<CONFIG_KSM>."
1178 #: build/C/man2/madvise.2:298
1180 "(for B<MADV_WILLNEED>) Paging in this area would exceed the process's "
1181 "maximum resident set size."
1185 #: build/C/man2/madvise.2:298 build/C/man2/madvise.2:303 build/C/man2/mlock.2:124 build/C/man2/mlock.2:132 build/C/man2/mlock.2:170 build/C/man2/mmap.2:454 build/C/man2/mprotect.2:94 build/C/man2/mprotect.2:97 build/C/man2/mremap.2:180 build/C/man2/msync.2:93 build/C/man3/posix_memalign.3:162 build/C/man2/shmctl.2:326 build/C/man2/shmget.2:211 build/C/man2/shmop.2:209 build/C/man2/subpage_prot.2:82 build/C/man2/sync_file_range.2:163
1191 #: build/C/man2/madvise.2:303
1192 msgid "(for B<MADV_WILLNEED>) Not enough memory: paging in failed."
1196 #: build/C/man2/madvise.2:307
1198 "Addresses in the specified range are not currently mapped, or are outside "
1199 "the address space of the process."
1202 #. FIXME . Write a posix_fadvise(3) page.
1204 #: build/C/man2/madvise.2:319
1206 "POSIX.1b. POSIX.1-2001 describes B<posix_madvise>(3) with constants "
1207 "B<POSIX_MADV_NORMAL>, etc., with a behavior close to that described here. "
1208 "There is a similar B<posix_fadvise>(2) for file access."
1212 #: build/C/man2/madvise.2:328
1214 "B<MADV_REMOVE>, B<MADV_DONTFORK>, B<MADV_DOFORK>, B<MADV_HWPOISON>, "
1215 "B<MADV_MERGEABLE>, and B<MADV_UNMERGEABLE> are Linux-specific."
1219 #: build/C/man2/madvise.2:329 build/C/man2/mlock.2:268 build/C/man2/shmget.2:278
1225 #: build/C/man2/madvise.2:337
1227 "The current Linux implementation (2.4.0) views this system call more as a "
1228 "command than as advice and hence may return an error when it cannot do what "
1229 "it usually would do in response to this advice. (See the ERRORS description "
1230 "above.) This is nonstandard behavior."
1236 #. function first appeared in 4.4BSD.
1238 #: build/C/man2/madvise.2:353
1240 "The Linux implementation requires that the address I<addr> be page-aligned, "
1241 "and allows I<length> to be zero. If there are some parts of the specified "
1242 "address range that are not mapped, the Linux version of B<madvise>() "
1243 "ignores them and applies the call to the rest (but returns B<ENOMEM> from "
1244 "the system call, as it should)."
1248 #: build/C/man2/madvise.2:360
1250 "B<getrlimit>(2), B<mincore>(2), B<mmap>(2), B<mprotect>(2), B<msync>(2), "
1255 #: build/C/man3/malloc_hook.3:6
1261 #: build/C/man3/malloc_hook.3:6
1267 #: build/C/man3/malloc_hook.3:11
1269 "__malloc_hook, __malloc_initialize_hook, __memalign_hook, __free_hook, "
1270 "__realloc_hook, __after_morecore_hook - malloc debugging variables"
1274 #: build/C/man3/malloc_hook.3:14 build/C/man3/posix_memalign.3:40
1276 msgid "B<#include E<lt>malloc.hE<gt>>\n"
1280 #: build/C/man3/malloc_hook.3:16
1282 msgid "B<void *(*__malloc_hook)(size_t >I<size>B<, const void *>I<caller>B<);>\n"
1286 #: build/C/man3/malloc_hook.3:19
1289 "B<void *(*__realloc_hook)(void *>I<ptr>B<, size_t >I<size>B<, const void "
1290 "*>I<caller>B<);>\n"
1294 #: build/C/man3/malloc_hook.3:22
1297 "B<void *(*__memalign_hook)(size_t >I<alignment>B<, size_t >I<size>B<,>\n"
1298 "B< const void *>I<caller>B<);>\n"
1302 #: build/C/man3/malloc_hook.3:24
1304 msgid "B<void (*__free_hook)(void *>I<ptr>B<, const void *>I<caller>B<);>\n"
1308 #: build/C/man3/malloc_hook.3:26
1310 msgid "B<void (*__malloc_initialize_hook)(void);>\n"
1314 #: build/C/man3/malloc_hook.3:28
1316 msgid "B<void (*__after_morecore_hook)(void);>\n"
1320 #: build/C/man3/malloc_hook.3:39
1322 "The GNU C library lets you modify the behavior of B<malloc>(3), "
1323 "B<realloc>(3), and B<free>(3) by specifying appropriate hook functions. "
1324 "You can use these hooks to help you debug programs that use dynamic memory "
1325 "allocation, for example."
1329 #: build/C/man3/malloc_hook.3:46
1331 "The variable B<__malloc_initialize_hook> points at a function that is called "
1332 "once when the malloc implementation is initialized. This is a weak "
1333 "variable, so it can be overridden in the application with a definition like "
1338 #: build/C/man3/malloc_hook.3:49
1340 msgid " void (*__malloc_initialize_hook)(void) = my_init_hook;\n"
1344 #: build/C/man3/malloc_hook.3:54
1345 msgid "Now the function I<my_init_hook>() can do the initialization of all hooks."
1349 #: build/C/man3/malloc_hook.3:70
1351 "The four functions pointed to by B<__malloc_hook>, B<__realloc_hook>, "
1352 "B<__memalign_hook>, B<__free_hook> have a prototype like the functions "
1353 "B<malloc>(3), B<realloc>(3), B<memalign>(3), B<free>(3), respectively, "
1354 "except that they have a final argument I<caller> that gives the address of "
1355 "the caller of B<malloc>(3), etc."
1359 #: build/C/man3/malloc_hook.3:76
1361 "The variable B<__after_morecore_hook> points at a function that is called "
1362 "each time after B<sbrk>(2) was asked for more memory."
1366 #: build/C/man3/malloc_hook.3:78 build/C/man3/mtrace.3:78
1367 msgid "These functions are GNU extensions."
1370 #. https://bugzilla.redhat.com/show_bug.cgi?id=450187
1371 #. http://sourceware.org/bugzilla/show_bug.cgi?id=9957
1373 #: build/C/man3/malloc_hook.3:85
1375 "The use of these hook functions is not safe in multithreaded programs, and "
1376 "they are now deprecated. Programmers should instead preempt calls to the "
1377 "relevant functions by defining and exporting functions such as \"malloc\" "
1382 #: build/C/man3/malloc_hook.3:85 build/C/man3/mallopt.3:464 build/C/man2/mmap.2:567 build/C/man2/mprotect.2:142 build/C/man3/mtrace.3:106
1388 #: build/C/man3/malloc_hook.3:87
1389 msgid "Here is a short example of how to use these variables."
1393 #: build/C/man3/malloc_hook.3:91
1396 "#include E<lt>stdio.hE<gt>\n"
1397 "#include E<lt>malloc.hE<gt>\n"
1401 #: build/C/man3/malloc_hook.3:95
1404 "/* Prototypes for our hooks. */\n"
1405 "static void my_init_hook(void);\n"
1406 "static void *my_malloc_hook(size_t, const void *);\n"
1410 #: build/C/man3/malloc_hook.3:98
1413 "/* Variables to save original hooks. */\n"
1414 "static void *(*old_malloc_hook)(size_t, const void *);\n"
1418 #: build/C/man3/malloc_hook.3:101
1421 "/* Override initializing hook from the C library. */\n"
1422 "void (*__malloc_initialize_hook) (void) = my_init_hook;\n"
1426 #: build/C/man3/malloc_hook.3:108
1430 "my_init_hook(void)\n"
1432 " old_malloc_hook = __malloc_hook;\n"
1433 " __malloc_hook = my_malloc_hook;\n"
1438 #: build/C/man3/malloc_hook.3:113
1442 "my_malloc_hook(size_t size, const void *caller)\n"
1448 #: build/C/man3/malloc_hook.3:116
1451 " /* Restore all old hooks */\n"
1452 " __malloc_hook = old_malloc_hook;\n"
1456 #: build/C/man3/malloc_hook.3:119
1459 " /* Call recursively */\n"
1460 " result = malloc(size);\n"
1464 #: build/C/man3/malloc_hook.3:122
1467 " /* Save underlying hooks */\n"
1468 " old_malloc_hook = __malloc_hook;\n"
1472 #: build/C/man3/malloc_hook.3:126
1475 " /* printf() might call malloc(), so protect it too. */\n"
1476 " printf(\"malloc(%u) called from %p returns %p\\en\",\n"
1477 " (unsigned int) size, caller, result);\n"
1481 #: build/C/man3/malloc_hook.3:129
1484 " /* Restore our own hooks */\n"
1485 " __malloc_hook = my_malloc_hook;\n"
1489 #: build/C/man3/malloc_hook.3:132
1497 #: build/C/man3/malloc_hook.3:138
1498 msgid "B<mallinfo>(3), B<malloc>(3), B<mcheck>(3), B<mtrace>(3)"
1502 #: build/C/man3/mallopt.3:24
1508 #: build/C/man3/mallopt.3:24 build/C/man3/mtrace.3:23
1514 #: build/C/man3/mallopt.3:27
1515 msgid "mallopt - set memory allocation parameters"
1519 #: build/C/man3/mallopt.3:29
1520 msgid "B<#include E<lt>malloc.hE<gt>>"
1524 #: build/C/man3/mallopt.3:31
1525 msgid "B<int mallopt(int >I<param>B<, int >I<value>B<);>"
1529 #: build/C/man3/mallopt.3:42
1531 "The B<mallopt>() function adjusts parameters that control the behavior of "
1532 "the memory-allocation functions (see B<malloc>(3)). The I<param> argument "
1533 "specifies the parameter to be modified, and I<value> specifies the new value "
1534 "for that parameter."
1538 #: build/C/man3/mallopt.3:45
1539 msgid "The following values can be specified for I<param>:"
1543 #: build/C/man3/mallopt.3:45
1545 msgid "B<M_CHECK_ACTION>"
1549 #: build/C/man3/mallopt.3:51
1551 "Setting this parameter controls how glibc responds when various kinds of "
1552 "programming errors are detected (e.g., freeing the same pointer twice). The "
1553 "3 least significant bits (2, 1, and 0) of the value assigned to this "
1554 "parameter determine the glibc behavior, as follows:"
1558 #: build/C/man3/mallopt.3:52
1564 #: build/C/man3/mallopt.3:62
1566 "If this bit is set, then print a one-line message on I<stderr> that provides "
1567 "details about the error. The message starts with the string \"***\\ glibc "
1568 "detected\\ ***\", followed by the program name, the name of the "
1569 "memory-allocation function in which the error was detected, a brief "
1570 "description of the error, and the memory address where the error was "
1575 #: build/C/man3/mallopt.3:62
1581 #: build/C/man3/mallopt.3:77
1583 "If this bit is set, then, after printing any error message specified by bit "
1584 "0, the program is terminated by calling B<abort>(3). In glibc versions "
1585 "since 2.4, if bit 0 is also set, then, between printing the error message "
1586 "and aborting, the program also prints a stack trace in the manner of "
1587 "B<backtrace>(3), and prints the process's memory mapping in the style of "
1588 "I</proc/[pid]/maps> (see B<proc>(5))."
1592 #: build/C/man3/mallopt.3:77
1594 msgid "Bit 2 (since glibc 2.4)"
1598 #: build/C/man3/mallopt.3:84
1600 "This bit has an effect only if bit 0 is also set. If this bit is set, then "
1601 "the one-line message describing the error is simplified to contain just the "
1602 "name of the function where the error was detected and the brief description "
1607 #: build/C/man3/mallopt.3:89
1608 msgid "The remaining bits in I<value> are ignored."
1612 #: build/C/man3/mallopt.3:93
1614 "Combining the above details, the following numeric values are meaningful for "
1615 "B<M_CHECK_ACTION>:"
1619 #: build/C/man3/mallopt.3:94
1625 #: build/C/man3/mallopt.3:96
1626 msgid "Ignore error conditions; continue execution (with undefined results)."
1630 #: build/C/man3/mallopt.3:96
1636 #: build/C/man3/mallopt.3:98
1637 msgid "Print a detailed error message and continue execution."
1641 #: build/C/man3/mallopt.3:98
1647 #: build/C/man3/mallopt.3:100
1648 msgid "Abort the program."
1652 #: build/C/man3/mallopt.3:100
1658 #: build/C/man3/mallopt.3:103
1660 "Print detailed error message, stack trace, and memory mappings, and abort "
1665 #: build/C/man3/mallopt.3:103
1671 #: build/C/man3/mallopt.3:105
1672 msgid "Print a simple error message and continue execution."
1676 #: build/C/man3/mallopt.3:105
1682 #: build/C/man3/mallopt.3:108
1684 "Print simple error message, stack trace, and memory mappings, and abort the "
1689 #: build/C/man3/mallopt.3:114
1691 "Since glibc 2.3.4, the default value for the B<M_CHECK_ACTION> parameter is "
1692 "3. In glibc version 2.3.3 and earlier, the default value is 1."
1696 #: build/C/man3/mallopt.3:119
1698 "Using a nonzero B<M_CHECK_ACTION> value can be useful because otherwise a "
1699 "crash may happen much later, and the true cause of the problem is then very "
1700 "hard to track down."
1704 #: build/C/man3/mallopt.3:119
1706 msgid "B<M_MMAP_MAX>"
1709 #. The following text adapted from comments in the glibc source:
1711 #: build/C/man3/mallopt.3:129
1713 "This parameter specifies the maximum number of allocation requests that may "
1714 "be simultaneously serviced using B<mmap>(2). This parameter exists because "
1715 "some systems have a limited number of internal tables for use by B<mmap>(2), "
1716 "and using more than a few of them may degrade performance."
1720 #: build/C/man3/mallopt.3:136
1722 "The default value is 65,536, a value which has no special significance and "
1723 "which servers only as a safeguard. Setting this parameter to 0 disables the "
1724 "use of B<mmap>(2) for servicing large allocation requests."
1728 #: build/C/man3/mallopt.3:136
1730 msgid "B<M_MMAP_THRESHOLD>"
1734 #: build/C/man3/mallopt.3:145
1736 "For allocations greater than or equal to the limit specified (in bytes) by "
1737 "B<M_MMAP_THRESHOLD> that can't be satisfied from the free list, the "
1738 "memory-allocation functions employ B<mmap>(2) instead of increasing the "
1739 "program break using B<sbrk>(2)."
1743 #: build/C/man3/mallopt.3:165
1745 "Allocating memory using B<mmap>(2) has the significant advantage that the "
1746 "allocated memory blocks can always be independently released back to the "
1747 "system. (By contrast, the heap can be trimmed only if memory is freed at "
1748 "the top end.) On the other hand, there are some disadvantages to the use of "
1749 "B<mmap>(2): deallocated space is not placed on the free list for reuse by "
1750 "later allocations; memory may be wasted because B<mmap>(2) allocations must "
1751 "be page-aligned; and the kernel must perform the expensive task of zeroing "
1752 "out memory allocated via B<mmap>(2). Balancing these factors leads to a "
1753 "default setting of 128*1024 for the B<M_MMAP_THRESHOLD> parameter."
1757 #: build/C/man3/mallopt.3:172
1759 "The lower limit for this parameter is 0. The upper limit is "
1760 "B<DEFAULT_MMAP_THRESHOLD_MAX>: 512*1024 on 32-bit systems or "
1761 "I<4*1024*1024*sizeof(long)> on 64-bit systems."
1765 #: build/C/man3/mallopt.3:190
1767 "I<Note:> Nowadays, glibc uses a dynamic mmap threshold by default. The "
1768 "initial value of the threshold is 128*1024, but when blocks larger than the "
1769 "current threshold and less than or equal to B<DEFAULT_MMAP_THRESHOLD_MAX> "
1770 "are freed, the threshold is adjusted upwards to the size of the freed "
1771 "block. When dynamic mmap thresholding is in effect, the threshold for "
1772 "trimming the heap is also dynamically adjusted to be twice the dynamic mmap "
1773 "threshold. Dynamic adjustment of the mmap threshold is disabled if any of "
1774 "the B<M_TRIM_THRESHOLD>, B<M_TOP_PAD>, B<M_MMAP_THRESHOLD>, or B<M_MMAP_MAX> "
1775 "parameters is set."
1779 #: build/C/man3/mallopt.3:190
1781 msgid "B<M_MXFAST> (since glibc 2.3)"
1784 #. The following text adapted from comments in the glibc sources:
1786 #: build/C/man3/mallopt.3:210
1788 "Set the upper limit for memory allocation requests that are satisfied using "
1789 "\"fastbins\". (The measurement unit for this parameter is bytes.) Fastbins "
1790 "are storage areas that hold deallocated blocks of memory of the same size "
1791 "without merging adjacent free blocks. Subsequent reallocation of blocks of "
1792 "the same size can be handled very quickly by allocating from the fastbin, "
1793 "although memory fragmentation and the overall memory footprint of the "
1794 "program can increase. The default value for this parameter is "
1795 "I<64*sizeof(size_t)/4> (i.e., 64 on 32-bit architectures). The range for "
1796 "this parameter is 0 to I<80*sizeof(size_t)/4>. Setting B<M_MXFAST> to 0 "
1797 "disables the use of fastbins."
1801 #: build/C/man3/mallopt.3:210
1803 msgid "B<M_PERTURB> (since glibc 2.4)"
1807 #: build/C/man3/mallopt.3:225
1809 "If this parameter is set to a nonzero value, then bytes of allocated memory "
1810 "(other than allocations via B<calloc>(3)) are initialized to the complement "
1811 "of the value in the least significant byte of I<value>, and when allocated "
1812 "memory is released using B<free>(3), the freed bytes are ANded with the "
1813 "value in the least significant byte of I<value>. This can be useful for "
1814 "detecting errors where programs incorrectly rely on allocated memory being "
1815 "initialized to zero, or reuse values in memory that has already been freed."
1819 #: build/C/man3/mallopt.3:225
1821 msgid "B<M_TOP_PAD>"
1825 #: build/C/man3/mallopt.3:232
1827 "This parameter defines the amount of padding to employ when calling "
1828 "B<sbrk>(2) to modify the program break. (The measurement unit for this "
1829 "parameter is bytes.) This parameter has an effect in the following "
1834 #: build/C/man3/mallopt.3:239
1836 "When the program break is increased, then B<M_TOP_PAD> bytes are added to "
1837 "the B<sbrk>(2) request."
1841 #: build/C/man3/mallopt.3:245
1843 "When the heap is trimmed as a consequence of calling B<free>(3) (see the "
1844 "discussion of B<M_TRIM_THRESHOLD>) this much free space is preserved at the "
1849 #: build/C/man3/mallopt.3:249
1851 "In either case, the amount of padding is always rounded to a system page "
1856 #: build/C/man3/mallopt.3:256
1858 "Modifying B<M_TOP_PAD> is a trade-off between increasing the number of "
1859 "system calls (when the parameter is set low) and wasting unused memory at "
1860 "the top of the heap (when the parameter is set high)."
1863 #. DEFAULT_TOP_PAD in glibc source
1865 #: build/C/man3/mallopt.3:259
1866 msgid "The default value for this parameter is 128*1024."
1870 #: build/C/man3/mallopt.3:259
1872 msgid "B<M_TRIM_THRESHOLD>"
1876 #: build/C/man3/mallopt.3:275
1878 "When the amount of contiguous free memory at the top of the heap grows "
1879 "sufficiently large, B<free>(3) employs B<sbrk>(2) to release this memory "
1880 "back to the system. (This can be useful in programs that continue to "
1881 "execute for a long period after freeing a significant amount of memory.) "
1882 "The B<M_TRIM_THRESHOLD> parameter specifies the minimum size (in bytes) that "
1883 "this block of memory must reach before B<sbrk>(2) is used to trim the heap."
1887 #: build/C/man3/mallopt.3:280
1889 "The default value for this parameter is 128*1024. Setting "
1890 "B<M_TRIM_THRESHOLD> to -1 disables trimming completely."
1893 #. FIXME Do the arena parameters need to be documented?
1895 #. .BR M_ARENA_TEST " (since glibc 2.10)"
1897 #. .BR M_ARENA_MAX " (since glibc 2.10)"
1899 #. Environment variables
1900 #. MALLOC_ARENA_MAX_
1901 #. MALLOC_ARENA_TEST_
1903 #. http://udrepper.livejournal.com/20948.html describes some details
1904 #. of the MALLOC_ARENA_* environment variables.
1906 #. These macros aren't enabled in production releases until 2.15?
1907 #. (see glibc malloc/Makefile)
1909 #: build/C/man3/mallopt.3:303
1911 "Modifying B<M_TRIM_THRESHOLD> is a trade-off between increasing the number "
1912 "of system calls (when the parameter is set low) and wasting unused memory "
1913 "at the top of the heap (when the parameter is set high)."
1917 #: build/C/man3/mallopt.3:303
1919 msgid "Environment Variables"
1923 #: build/C/man3/mallopt.3:318
1925 "A number of environment variables can be defined to modify some of the same "
1926 "parameters as are controlled by B<mallopt>(). Using these variables has the "
1927 "advantage that the source code of the program need not be changed. To be "
1928 "effective, these variables must be defined before the first call to a "
1929 "memory-allocation function. (If the same parameters are adjusted via "
1930 "B<mallopt>() then the B<mallopt>() settings take precedence.) For "
1931 "security reasons, these variables are ignored in set-user-ID and "
1932 "set-group-ID programs."
1936 #: build/C/man3/mallopt.3:321
1938 "The environment variables are as follows (note the trailing underscore at "
1939 "the end of the name of each variable):"
1943 #: build/C/man3/mallopt.3:321
1945 msgid "B<MALLOC_CHECK_>"
1948 #. On glibc 2.12/x86, a simple malloc()+free() loop is about 70% slower
1949 #. when MALLOC_CHECK_ was set.
1951 #: build/C/man3/mallopt.3:338
1953 "This environment variable controls the same parameter as B<mallopt>() "
1954 "B<M_CHECK_ACTION>. If this variable is set to a nonzero value, then a "
1955 "special implementation of the memory-allocation functions is used. (This is "
1956 "accomplished using the B<malloc_hook>(3) feature.) This implementation "
1957 "performs additional error checking, but is slower than the standard set of "
1958 "memory-allocation functions. (This implementation does not detect all "
1959 "possible errors; memory leaks can still occur.)"
1963 #: build/C/man3/mallopt.3:343
1965 "The value assigned to this environment variable should be a single digit, "
1966 "whose meaning is as described for B<M_CHECK_ACTION>. Any characters beyond "
1967 "the initial digit are ignored."
1971 #: build/C/man3/mallopt.3:352
1973 "For security reasons, the effect of B<MALLOC_CHECK_> is disabled by default "
1974 "for set-user-ID and set-group-ID programs. However, if the file "
1975 "I</etc/suid-debug> exists (the content of the file is irrelevant), then "
1976 "B<MALLOC_CHECK_> also has an effect for set-user-ID and set-group-ID "
1981 #: build/C/man3/mallopt.3:352
1983 msgid "B<MALLOC_MMAP_MAX_>"
1987 #: build/C/man3/mallopt.3:357
1988 msgid "Controls the same parameter as B<mallopt>() B<M_MMAP_MAX>."
1992 #: build/C/man3/mallopt.3:357
1994 msgid "B<MALLOC_MMAP_THRESHOLD_>"
1998 #: build/C/man3/mallopt.3:362
1999 msgid "Controls the same parameter as B<mallopt>() B<M_MMAP_THRESHOLD>."
2003 #: build/C/man3/mallopt.3:362
2005 msgid "B<MALLOC_PERTURB_>"
2009 #: build/C/man3/mallopt.3:367
2010 msgid "Controls the same parameter as B<mallopt>() B<M_PERTURB>."
2014 #: build/C/man3/mallopt.3:367
2016 msgid "B<MALLOC_TRIM_THRESHOLD_>"
2020 #: build/C/man3/mallopt.3:372
2021 msgid "Controls the same parameter as B<mallopt>() B<M_TRIM_THRESHOLD>."
2025 #: build/C/man3/mallopt.3:372
2027 msgid "B<MALLOC_TOP_PAD_>"
2031 #: build/C/man3/mallopt.3:377
2032 msgid "Controls the same parameter as B<mallopt>() B<M_TOP_PAD>."
2036 #: build/C/man3/mallopt.3:382
2037 msgid "On success, B<mallopt>() returns 1. On error, it returns 0."
2041 #. Available already in glibc 2.0, possibly earlier
2043 #: build/C/man3/mallopt.3:390
2044 msgid "On error, I<errno> is I<not> set."
2049 #: build/C/man3/mallopt.3:404
2051 "This function is not specified by POSIX or the C standards. A similar "
2052 "function exists on many System V derivatives, but the range of values for "
2053 "I<param> varies across systems. The SVID defined options B<M_MXFAST>, "
2054 "B<M_NLBLKS>, B<M_GRAIN>, and B<M_KEEP>, but only the first of these is "
2055 "implemented in glibc."
2059 #: build/C/man3/mallopt.3:408
2060 msgid "Specifying an invalid value for I<param> does not generate an error."
2063 #. FIXME This looks buggy:
2064 #. setting the M_MXFAST limit rounds up: (s + SIZE_SZ) & ~MALLOC_ALIGN_MASK)
2065 #. malloc requests are rounded up:
2066 #. (req) + SIZE_SZ + MALLOC_ALIGN_MASK) & ~MALLOC_ALIGN_MASK
2067 #. http://sources.redhat.com/bugzilla/show_bug.cgi?id=12129
2069 #: build/C/man3/mallopt.3:416
2071 "A calculation error within the glibc implementation means that a call of the "
2076 #: build/C/man3/mallopt.3:419
2078 msgid " mallopt(M_MXFAST, n)\n"
2081 #. Bins are multiples of 2 * sizeof(size_t) + sizeof(size_t)
2083 #: build/C/man3/mallopt.3:431
2085 "does not result in fastbins being employed for all allocations of size up to "
2086 "I<n>. To ensure desired results, I<n> should be rounded up to the next "
2087 "multiple greater than or equal to I<(2k+1)*sizeof(size_t)>, where I<k> is an "
2091 #. FIXME MALLOC_MMAP_THRESHOLD_ and MALLOC_MMAP_MAX_
2092 #. do have an effect for set-user-ID programs (but not
2093 #. set-group-ID programs).
2094 #. http://sources.redhat.com/bugzilla/show_bug.cgi?id=12155
2096 #: build/C/man3/mallopt.3:443
2098 "The B<MALLOC_MMAP_THRESHOLD_> and B<MALLOC_MMAP_MAX_> variables are I<not> "
2099 "ignored in set-group-ID programs."
2102 #. FIXME http://sources.redhat.com/bugzilla/show_bug.cgi?id=12140
2104 #: build/C/man3/mallopt.3:464
2106 "If B<mallopt>() is used to set B<M_PERTURB>, then, as expected, the bytes "
2107 "of allocated memory are initialized to the complement of the byte in "
2108 "I<value>, and when that memory is freed, the bytes of the region are "
2109 "initialized to the byte specified in I<value>. However, there is an "
2110 "off-by-I<sizeof(size_t)> error in the implementation: instead of "
2111 "initializing precisely the block of memory being freed by the call "
2112 "I<free(p)>, the block starting at I<p+sizeof(size_t)> is initialized."
2116 #: build/C/man3/mallopt.3:473
2118 "The program below demonstrates the use of B<M_CHECK_ACTION>. If the program "
2119 "is supplied with an (integer) command-line argument, then that argument is "
2120 "used to set the B<M_CHECK_ACTION> parameter. The program then allocates a "
2121 "block of memory, and frees it twice (an error)."
2125 #: build/C/man3/mallopt.3:477
2127 "The following shell session shows what happens when we run this program "
2128 "under glibc, with the default value for B<M_CHECK_ACTION>:"
2132 #: build/C/man3/mallopt.3:496
2136 "*** glibc detected *** ./a.out: double free or corruption (top): 0x09d30008 "
2138 "======= Backtrace: =========\n"
2139 "/lib/libc.so.6(+0x6c501)[0x523501]\n"
2140 "/lib/libc.so.6(+0x6dd70)[0x524d70]\n"
2141 "/lib/libc.so.6(cfree+0x6d)[0x527e5d]\n"
2142 "\\&./a.out[0x80485db]\n"
2143 "/lib/libc.so.6(__libc_start_main+0xe7)[0x4cdce7]\n"
2144 "\\&./a.out[0x8048471]\n"
2145 "======= Memory map: ========\n"
2146 "001e4000-001fe000 r-xp 00000000 08:06 1083555 /lib/libgcc_s.so.1\n"
2147 "001fe000-001ff000 r--p 00019000 08:06 1083555 /lib/libgcc_s.so.1\n"
2148 "[some lines omitted]\n"
2149 "b7814000-b7817000 rw-p 00000000 00:00 0\n"
2150 "bff53000-bff74000 rw-p 00000000 00:00 0 [stack]\n"
2151 "Aborted (core dumped)\n"
2155 #: build/C/man3/mallopt.3:501
2157 "The following runs show the results when employing other values for "
2158 "B<M_CHECK_ACTION:>"
2162 #: build/C/man3/mallopt.3:514
2165 "$ B<./a.out 1> # Diagnose error and continue\n"
2166 "main(): returned from first free() call\n"
2167 "*** glibc detected *** ./a.out: double free or corruption (top): 0x09cbe008 "
2169 "main(): returned from second free() call\n"
2170 "$ B<./a.out 2> # Abort without error message\n"
2171 "main(): returned from first free() call\n"
2172 "Aborted (core dumped)\n"
2173 "$ B<./a.out 0> # Ignore error and continue\n"
2174 "main(): returned from first free() call\n"
2175 "main(): returned from second free() call\n"
2179 #: build/C/man3/mallopt.3:520
2181 "The next run shows how to set the same parameter using the B<MALLOC_CHECK_> "
2182 "environment variable:"
2186 #: build/C/man3/mallopt.3:527
2189 "$ B<MALLOC_CHECK_=1 ./a.out>\n"
2190 "main(): returned from first free() call\n"
2191 "*** glibc detected *** ./a.out: free(): invalid pointer: 0x092c2008 ***\n"
2192 "main(): returned from second free() call\n"
2196 #: build/C/man3/mallopt.3:529 build/C/man2/mprotect.2:159
2198 msgid "Program source"
2202 #: build/C/man3/mallopt.3:535
2205 "#include E<lt>malloc.hE<gt>\n"
2206 "#include E<lt>stdio.hE<gt>\n"
2207 "#include E<lt>stdlib.hE<gt>\n"
2211 #: build/C/man3/mallopt.3:540
2215 "main(int argc, char *argv[])\n"
2221 #: build/C/man3/mallopt.3:547
2224 " if (argc E<gt> 1) {\n"
2225 " if (argc != 2) {\n"
2226 " fprintf(stderr, \"%s E<lt>M_CHECK_ACTION-valueE<gt>\\en\", "
2228 " exit(EXIT_FAILURE);\n"
2234 #: build/C/man3/mallopt.3:552
2237 " if (mallopt(M_CHECK_ACTION, atoi(argv[1])) != 1) {\n"
2238 " fprintf(stderr, \"mallopt() failed\");\n"
2239 " exit(EXIT_FAILURE);\n"
2244 #: build/C/man3/mallopt.3:558
2247 " p = malloc(1000);\n"
2248 " if (p == NULL) {\n"
2249 " fprintf(stderr, \"malloc() failed\");\n"
2250 " exit(EXIT_FAILURE);\n"
2255 #: build/C/man3/mallopt.3:561
2259 " printf(\"main(): returned from first free() call\\en\");\n"
2263 #: build/C/man3/mallopt.3:564
2267 " printf(\"main(): returned from second free() call\\en\");\n"
2271 #: build/C/man3/mallopt.3:567 build/C/man2/mmap.2:648
2274 " exit(EXIT_SUCCESS);\n"
2279 #: build/C/man3/mallopt.3:578
2281 "B<mmap>(2) B<sbrk>(2), B<mallinfo>(3), B<malloc>(3), B<malloc_hook>(3), "
2282 "B<mtrace>(3), B<posix_memalign>(3)"
2286 #: build/C/man2/mlock.2:27
2292 #: build/C/man2/mlock.2:27
2298 #: build/C/man2/mlock.2:30
2299 msgid "mlock, munlock, mlockall, munlockall - lock and unlock memory"
2303 #: build/C/man2/mlock.2:33 build/C/man2/mmap.2:46 build/C/man2/mmap2.2:35 build/C/man2/mprotect.2:38 build/C/man2/mremap.2:39
2305 msgid "B<#include E<lt>sys/mman.hE<gt>>\n"
2309 #: build/C/man2/mlock.2:36
2312 "B<int mlock(const void *>I<addr>B<, size_t >I<len>B<);>\n"
2313 "B<int munlock(const void *>I<addr>B<, size_t >I<len>B<);>\n"
2317 #: build/C/man2/mlock.2:39
2320 "B<int mlockall(int >I<flags>B<);>\n"
2321 "B<int munlockall(void);>\n"
2325 #: build/C/man2/mlock.2:55
2327 "B<mlock>() and B<mlockall>() respectively lock part or all of the calling "
2328 "process's virtual address space into RAM, preventing that memory from being "
2329 "paged to the swap area. B<munlock>() and B<munlockall>() perform the "
2330 "converse operation, respectively unlocking part or all of the calling "
2331 "process's virtual address space, so that pages in the specified virtual "
2332 "address range may once more to be swapped out if required by the kernel "
2333 "memory manager. Memory locking and unlocking are performed in units of "
2338 #: build/C/man2/mlock.2:55
2340 msgid "mlock() and munlock()"
2344 #: build/C/man2/mlock.2:65
2346 "B<mlock>() locks pages in the address range starting at I<addr> and "
2347 "continuing for I<len> bytes. All pages that contain a part of the specified "
2348 "address range are guaranteed to be resident in RAM when the call returns "
2349 "successfully; the pages are guaranteed to stay in RAM until later unlocked."
2353 #: build/C/man2/mlock.2:74
2355 "B<munlock>() unlocks pages in the address range starting at I<addr> and "
2356 "continuing for I<len> bytes. After this call, all pages that contain a part "
2357 "of the specified memory range can be moved to external swap space again by "
2362 #: build/C/man2/mlock.2:74
2364 msgid "mlockall() and munlockall()"
2368 #: build/C/man2/mlock.2:84
2370 "B<mlockall>() locks all pages mapped into the address space of the calling "
2371 "process. This includes the pages of the code, data and stack segment, as "
2372 "well as shared libraries, user space kernel data, shared memory, and "
2373 "memory-mapped files. All mapped pages are guaranteed to be resident in RAM "
2374 "when the call returns successfully; the pages are guaranteed to stay in RAM "
2375 "until later unlocked."
2379 #: build/C/man2/mlock.2:89
2381 "The I<flags> argument is constructed as the bitwise OR of one or more of the "
2382 "following constants:"
2386 #: build/C/man2/mlock.2:89
2388 msgid "B<MCL_CURRENT>"
2392 #: build/C/man2/mlock.2:93
2394 "Lock all pages which are currently mapped into the address space of the "
2399 #: build/C/man2/mlock.2:93
2401 msgid "B<MCL_FUTURE>"
2405 #: build/C/man2/mlock.2:100
2407 "Lock all pages which will become mapped into the address space of the "
2408 "process in the future. These could be for instance new pages required by a "
2409 "growing heap and stack as well as new memory mapped files or shared memory "
2414 #: build/C/man2/mlock.2:113
2416 "If B<MCL_FUTURE> has been specified, then a later system call (e.g., "
2417 "B<mmap>(2), B<sbrk>(2), B<malloc>(3)), may fail if it would cause the number "
2418 "of locked bytes to exceed the permitted maximum (see below). In the same "
2419 "circumstances, stack growth may likewise fail: the kernel will deny stack "
2420 "expansion and deliver a B<SIGSEGV> signal to the process."
2424 #: build/C/man2/mlock.2:117
2426 "B<munlockall>() unlocks all pages mapped into the address space of the "
2431 #: build/C/man2/mlock.2:123
2433 "On success these system calls return 0. On error, -1 is returned, I<errno> "
2434 "is set appropriately, and no changes are made to any locks in the address "
2435 "space of the process."
2439 #: build/C/man2/mlock.2:132
2441 "(Linux 2.6.9 and later) the caller had a nonzero B<RLIMIT_MEMLOCK> soft "
2442 "resource limit, but tried to lock more memory than the limit permitted. "
2443 "This limit is not enforced if the process is privileged (B<CAP_IPC_LOCK>)."
2446 #. In the case of mlock(), this check is somewhat buggy: it doesn't
2447 #. take into account whether the to-be-locked range overlaps with
2448 #. already locked pages. Thus, suppose we allocate
2449 #. (num_physpages / 4 + 1) of memory, and lock those pages once using
2450 #. mlock(), and then lock the *same* page range a second time.
2451 #. In the case, the second mlock() call will fail, since the check
2452 #. calculates that the process is trying to lock (num_physpages / 2 + 2)
2453 #. pages, which of course is not true. (MTK, Nov 04, kernel 2.4.28)
2455 #: build/C/man2/mlock.2:144
2457 "(Linux 2.4 and earlier) the calling process tried to lock more than half of "
2461 #. SVr4 documents an additional EAGAIN error code.
2463 #: build/C/man2/mlock.2:150
2465 "The caller is not privileged, but needs privilege (B<CAP_IPC_LOCK>) to "
2466 "perform the requested operation."
2470 #: build/C/man2/mlock.2:155
2471 msgid "For B<mlock>() and B<munlock>():"
2475 #: build/C/man2/mlock.2:158
2476 msgid "Some or all of the specified address range could not be locked."
2480 #: build/C/man2/mlock.2:165
2482 "The result of the addition I<start>+I<len> was less than I<start> (e.g., the "
2483 "addition may have resulted in an overflow)."
2487 #: build/C/man2/mlock.2:170
2488 msgid "(Not on Linux) I<addr> was not a multiple of the page size."
2492 #: build/C/man2/mlock.2:174
2494 "Some of the specified address range does not correspond to mapped pages in "
2495 "the address space of the process."
2499 #: build/C/man2/mlock.2:177
2500 msgid "For B<mlockall>():"
2504 #: build/C/man2/mlock.2:180
2505 msgid "Unknown I<flags> were specified."
2509 #: build/C/man2/mlock.2:183
2510 msgid "For B<munlockall>():"
2514 #: build/C/man2/mlock.2:187
2515 msgid "(Linux 2.6.8 and earlier) The caller was not privileged (B<CAP_IPC_LOCK>)."
2519 #: build/C/man2/mlock.2:189
2520 msgid "POSIX.1-2001, SVr4."
2524 #: build/C/man2/mlock.2:189 build/C/man2/mmap.2:487 build/C/man2/msync.2:105
2526 msgid "AVAILABILITY"
2530 #: build/C/man2/mlock.2:201
2532 "On POSIX systems on which B<mlock>() and B<munlock>() are available, "
2533 "B<_POSIX_MEMLOCK_RANGE> is defined in I<E<lt>unistd.hE<gt>> and the number "
2534 "of bytes in a page can be determined from the constant B<PAGESIZE> (if "
2535 "defined) in I<E<lt>limits.hE<gt>> or by calling I<sysconf(_SC_PAGESIZE)>."
2538 #. POSIX.1-2001: It shall be defined to -1 or 0 or 200112L.
2539 #. -1: unavailable, 0: ask using sysconf().
2540 #. glibc defines it to 1.
2542 #: build/C/man2/mlock.2:214
2544 "On POSIX systems on which B<mlockall>() and B<munlockall>() are available, "
2545 "B<_POSIX_MEMLOCK> is defined in I<E<lt>unistd.hE<gt>> to a value greater "
2546 "than 0. (See also B<sysconf>(3).)"
2550 #: build/C/man2/mlock.2:232
2552 "Memory locking has two main applications: real-time algorithms and "
2553 "high-security data processing. Real-time applications require deterministic "
2554 "timing, and, like scheduling, paging is one major cause of unexpected "
2555 "program execution delays. Real-time applications will usually also switch "
2556 "to a real-time scheduler with B<sched_setscheduler>(2). Cryptographic "
2557 "security software often handles critical bytes like passwords or secret keys "
2558 "as data structures. As a result of paging, these secrets could be "
2559 "transferred onto a persistent swap store medium, where they might be "
2560 "accessible to the enemy long after the security software has erased the "
2561 "secrets in RAM and terminated. (But be aware that the suspend mode on "
2562 "laptops and some desktop computers will save a copy of the system's RAM to "
2563 "disk, regardless of memory locks.)"
2567 #: build/C/man2/mlock.2:245
2569 "Real-time processes that are using B<mlockall>() to prevent delays on page "
2570 "faults should reserve enough locked stack pages before entering the "
2571 "time-critical section, so that no page fault can be caused by function "
2572 "calls. This can be achieved by calling a function that allocates a "
2573 "sufficiently large automatic variable (an array) and writes to the memory "
2574 "occupied by this array in order to touch these stack pages. This way, "
2575 "enough pages will be mapped for the stack and can be locked into RAM. The "
2576 "dummy writes ensure that not even copy-on-write page faults can occur in the "
2581 #: build/C/man2/mlock.2:251
2583 "Memory locks are not inherited by a child created via B<fork>(2) and are "
2584 "automatically removed (unlocked) during an B<execve>(2) or when the process "
2589 #: build/C/man2/mlock.2:255
2591 "The memory lock on an address range is automatically removed if the address "
2592 "range is unmapped via B<munmap>(2)."
2596 #: build/C/man2/mlock.2:268
2598 "Memory locks do not stack, that is, pages which have been locked several "
2599 "times by calls to B<mlock>() or B<mlockall>() will be unlocked by a single "
2600 "call to B<munlock>() for the corresponding range or by B<munlockall>(). "
2601 "Pages which are mapped to several locations or by several processes stay "
2602 "locked into RAM as long as they are locked at least at one location or by at "
2603 "least one process."
2607 #: build/C/man2/mlock.2:279
2609 "Under Linux, B<mlock>() and B<munlock>() automatically round I<addr> down "
2610 "to the nearest page boundary. However, POSIX.1-2001 allows an "
2611 "implementation to require that I<addr> is page aligned, so portable "
2612 "applications should ensure this."
2616 #: build/C/man2/mlock.2:292
2618 "The I<VmLck> field of the Linux-specific I</proc/PID/status> file shows how "
2619 "many kilobytes of memory the process with ID I<PID> has locked using "
2620 "B<mlock>(), B<mlockall>(), and B<mmap>(2) B<MAP_LOCKED>."
2624 #: build/C/man2/mlock.2:292
2626 msgid "Limits and permissions"
2630 #: build/C/man2/mlock.2:299
2632 "In Linux 2.6.8 and earlier, a process must be privileged (B<CAP_IPC_LOCK>) "
2633 "in order to lock memory and the B<RLIMIT_MEMLOCK> soft resource limit "
2634 "defines a limit on how much memory the process may lock."
2638 #: build/C/man2/mlock.2:305
2640 "Since Linux 2.6.9, no limits are placed on the amount of memory that a "
2641 "privileged process can lock and the B<RLIMIT_MEMLOCK> soft resource limit "
2642 "instead defines a limit on how much memory an unprivileged process may lock."
2646 #: build/C/man2/mlock.2:313
2648 "In the 2.4 series Linux kernels up to and including 2.4.17, a bug caused the "
2649 "B<mlockall>() B<MCL_FUTURE> flag to be inherited across a B<fork>(2). This "
2650 "was rectified in kernel 2.4.18."
2653 #. See the following LKML thread:
2654 #. http://marc.theaimsgroup.com/?l=linux-kernel&m=113801392825023&w=2
2655 #. "Rationale for RLIMIT_MEMLOCK"
2658 #: build/C/man2/mlock.2:330
2660 "Since kernel 2.6.9, if a privileged process calls I<mlockall(MCL_FUTURE)> "
2661 "and later drops privileges (loses the B<CAP_IPC_LOCK> capability by, for "
2662 "example, setting its effective UID to a nonzero value), then subsequent "
2663 "memory allocations (e.g., B<mmap>(2), B<brk>(2)) will fail if the "
2664 "B<RLIMIT_MEMLOCK> resource limit is encountered."
2668 #: build/C/man2/mlock.2:337
2670 "B<mmap>(2), B<setrlimit>(2), B<shmctl>(2), B<sysconf>(3), B<proc>(5), "
2671 "B<capabilities>(7)"
2675 #: build/C/man2/mmap.2:40
2681 #: build/C/man2/mmap.2:40
2687 #: build/C/man2/mmap.2:43
2688 msgid "mmap, munmap - map or unmap files or devices into memory"
2692 #: build/C/man2/mmap.2:51
2695 "B<void *mmap(void *>I<addr>B<, size_t >I<length>B<, int >I<prot>B<, int "
2697 "B< int >I<fd>B<, off_t >I<offset>B<);>\n"
2698 "B<int munmap(void *>I<addr>B<, size_t >I<length>B<);>\n"
2702 #: build/C/man2/mmap.2:61
2704 "B<mmap>() creates a new mapping in the virtual address space of the calling "
2705 "process. The starting address for the new mapping is specified in I<addr>. "
2706 "The I<length> argument specifies the length of the mapping."
2709 #. Before Linux 2.6.24, the address was rounded up to the next page
2710 #. boundary; since 2.6.24, it is rounded down!
2712 #: build/C/man2/mmap.2:75
2714 "If I<addr> is NULL, then the kernel chooses the address at which to create "
2715 "the mapping; this is the most portable method of creating a new mapping. If "
2716 "I<addr> is not NULL, then the kernel takes it as a hint about where to place "
2717 "the mapping; on Linux, the mapping will be created at a nearby page "
2718 "boundary. The address of the new mapping is returned as the result of the "
2723 #: build/C/man2/mmap.2:87
2725 "The contents of a file mapping (as opposed to an anonymous mapping; see "
2726 "B<MAP_ANONYMOUS> below), are initialized using I<length> bytes starting at "
2727 "offset I<offset> in the file (or other object) referred to by the file "
2728 "descriptor I<fd>. I<offset> must be a multiple of the page size as returned "
2729 "by I<sysconf(_SC_PAGE_SIZE)>."
2733 #: build/C/man2/mmap.2:95
2735 "The I<prot> argument describes the desired memory protection of the mapping "
2736 "(and must not conflict with the open mode of the file). It is either "
2737 "B<PROT_NONE> or the bitwise OR of one or more of the following flags:"
2741 #: build/C/man2/mmap.2:95 build/C/man2/mprotect.2:67
2743 msgid "B<PROT_EXEC>"
2747 #: build/C/man2/mmap.2:98
2748 msgid "Pages may be executed."
2752 #: build/C/man2/mmap.2:98 build/C/man2/mprotect.2:61
2754 msgid "B<PROT_READ>"
2758 #: build/C/man2/mmap.2:101
2759 msgid "Pages may be read."
2763 #: build/C/man2/mmap.2:101 build/C/man2/mprotect.2:64
2765 msgid "B<PROT_WRITE>"
2769 #: build/C/man2/mmap.2:104
2770 msgid "Pages may be written."
2774 #: build/C/man2/mmap.2:104 build/C/man2/mprotect.2:58
2776 msgid "B<PROT_NONE>"
2780 #: build/C/man2/mmap.2:107
2781 msgid "Pages may not be accessed."
2785 #: build/C/man2/mmap.2:116
2787 "The I<flags> argument determines whether updates to the mapping are visible "
2788 "to other processes mapping the same region, and whether updates are carried "
2789 "through to the underlying file. This behavior is determined by including "
2790 "exactly one of the following values in I<flags>:"
2794 #: build/C/man2/mmap.2:116
2796 msgid "B<MAP_SHARED>"
2800 #: build/C/man2/mmap.2:126
2802 "Share this mapping. Updates to the mapping are visible to other processes "
2803 "that map this file, and are carried through to the underlying file. The "
2804 "file may not actually be updated until B<msync>(2) or B<munmap>() is "
2809 #: build/C/man2/mmap.2:126
2811 msgid "B<MAP_PRIVATE>"
2815 #: build/C/man2/mmap.2:135
2817 "Create a private copy-on-write mapping. Updates to the mapping are not "
2818 "visible to other processes mapping the same file, and are not carried "
2819 "through to the underlying file. It is unspecified whether changes made to "
2820 "the file after the B<mmap>() call are visible in the mapped region."
2824 #: build/C/man2/mmap.2:137
2825 msgid "Both of these flags are described in POSIX.1-2001."
2829 #: build/C/man2/mmap.2:140
2830 msgid "In addition, zero or more of the following values can be ORed in I<flags>:"
2834 #: build/C/man2/mmap.2:140
2836 msgid "B<MAP_32BIT> (since Linux 2.4.20, 2.6)"
2839 #. See http://lwn.net/Articles/294642 "Tangled up in threads", 19 Aug 08
2841 #: build/C/man2/mmap.2:156
2843 "Put the mapping into the first 2 Gigabytes of the process address space. "
2844 "This flag is only supported on x86-64, for 64-bit programs. It was added to "
2845 "allow thread stacks to be allocated somewhere in the first 2GB of memory, so "
2846 "as to improve context-switch performance on some early 64-bit processors. "
2847 "Modern x86-64 processors no longer have this performance problem, so use of "
2848 "this flag is not required on those systems. The B<MAP_32BIT> flag is "
2849 "ignored when B<MAP_FIXED> is set."
2853 #: build/C/man2/mmap.2:156
2859 #: build/C/man2/mmap.2:161
2860 msgid "Synonym for B<MAP_ANONYMOUS>. Deprecated."
2864 #: build/C/man2/mmap.2:161
2866 msgid "B<MAP_ANONYMOUS>"
2870 #: build/C/man2/mmap.2:183
2872 "The mapping is not backed by any file; its contents are initialized to "
2873 "zero. The I<fd> and I<offset> arguments are ignored; however, some "
2874 "implementations require I<fd> to be -1 if B<MAP_ANONYMOUS> (or B<MAP_ANON>) "
2875 "is specified, and portable applications should ensure this. The use of "
2876 "B<MAP_ANONYMOUS> in conjunction with B<MAP_SHARED> is only supported on "
2877 "Linux since kernel 2.4."
2881 #: build/C/man2/mmap.2:183
2883 msgid "B<MAP_DENYWRITE>"
2886 #. Introduced in 1.1.36, removed in 1.3.24.
2888 #: build/C/man2/mmap.2:191
2890 "This flag is ignored. (Long ago, it signaled that attempts to write to the "
2891 "underlying file should fail with B<ETXTBUSY>. But this was a source of "
2892 "denial-of-service attacks.)"
2896 #: build/C/man2/mmap.2:191
2898 msgid "B<MAP_EXECUTABLE>"
2901 #. Introduced in 1.1.38, removed in 1.3.24. Flag tested in proc_follow_link.
2902 #. (Long ago, it signaled that the underlying file is an executable.
2903 #. However, that information was not really used anywhere.)
2904 #. Linus talked about DOS related to MAP_EXECUTABLE, but he was thinking of
2907 #: build/C/man2/mmap.2:199
2908 msgid "This flag is ignored."
2912 #: build/C/man2/mmap.2:199
2917 #. On some systems, this was required as the opposite of
2918 #. MAP_ANONYMOUS -- mtk, 1 May 2007
2920 #: build/C/man2/mmap.2:205
2921 msgid "Compatibility flag. Ignored."
2925 #: build/C/man2/mmap.2:205
2927 msgid "B<MAP_FIXED>"
2931 #: build/C/man2/mmap.2:223
2933 "Don't interpret I<addr> as a hint: place the mapping at exactly that "
2934 "address. I<addr> must be a multiple of the page size. If the memory region "
2935 "specified by I<addr> and I<len> overlaps pages of any existing mapping(s), "
2936 "then the overlapped part of the existing mapping(s) will be discarded. If "
2937 "the specified address cannot be used, B<mmap>() will fail. Because "
2938 "requiring a fixed address for a mapping is less portable, the use of this "
2939 "option is discouraged."
2943 #: build/C/man2/mmap.2:223
2945 msgid "B<MAP_GROWSDOWN>"
2949 #: build/C/man2/mmap.2:228
2951 "Used for stacks. Indicates to the kernel virtual memory system that the "
2952 "mapping should extend downward in memory."
2956 #: build/C/man2/mmap.2:228
2958 msgid "B<MAP_HUGETLB> (since Linux 2.6.32)"
2962 #: build/C/man2/mmap.2:234
2964 "Allocate the mapping using \"huge pages.\" See the kernel source file "
2965 "I<Documentation/vm/hugetlbpage.txt> for further information."
2969 #: build/C/man2/mmap.2:234
2971 msgid "B<MAP_LOCKED> (since Linux 2.5.37)"
2974 #. If set, the mapped pages will not be swapped out.
2976 #: build/C/man2/mmap.2:240
2978 "Lock the pages of the mapped region into memory in the manner of "
2979 "B<mlock>(2). This flag is ignored in older kernels."
2983 #: build/C/man2/mmap.2:240
2985 msgid "B<MAP_NONBLOCK> (since Linux 2.5.46)"
2989 #: build/C/man2/mmap.2:255
2991 "Only meaningful in conjunction with B<MAP_POPULATE>. Don't perform "
2992 "read-ahead: only create page tables entries for pages that are already "
2993 "present in RAM. Since Linux 2.6.23, this flag causes B<MAP_POPULATE> to do "
2994 "nothing. One day the combination of B<MAP_POPULATE> and B<MAP_NONBLOCK> may "
2999 #: build/C/man2/mmap.2:255
3001 msgid "B<MAP_NORESERVE>"
3005 #: build/C/man2/mmap.2:270
3007 "Do not reserve swap space for this mapping. When swap space is reserved, "
3008 "one has the guarantee that it is possible to modify the mapping. When swap "
3009 "space is not reserved one might get B<SIGSEGV> upon a write if no physical "
3010 "memory is available. See also the discussion of the file "
3011 "I</proc/sys/vm/overcommit_memory> in B<proc>(5). In kernels before 2.6, "
3012 "this flag only had effect for private writable mappings."
3016 #: build/C/man2/mmap.2:270
3018 msgid "B<MAP_POPULATE> (since Linux 2.5.46)"
3022 #: build/C/man2/mmap.2:277
3024 "Populate (prefault) page tables for a mapping. For a file mapping, this "
3025 "causes read-ahead on the file. Later accesses to the mapping will not be "
3026 "blocked by page faults. B<MAP_POPULATE> is only supported for private "
3027 "mappings since Linux 2.6.23."
3031 #: build/C/man2/mmap.2:277
3033 msgid "B<MAP_STACK> (since Linux 2.6.27)"
3036 #. See http://lwn.net/Articles/294642 "Tangled up in threads", 19 Aug 08
3037 #. commit cd98a04a59e2f94fa64d5bf1e26498d27427d5e7
3038 #. http://thread.gmane.org/gmane.linux.kernel/720412
3039 #. "pthread_create() slow for many threads; also time to revisit 64b
3040 #. context switch optimization?"
3042 #: build/C/man2/mmap.2:290
3044 "Allocate the mapping at an address suitable for a process or thread stack. "
3045 "This flag is currently a no-op, but is used in the glibc threading "
3046 "implementation so that if some architectures require special treatment for "
3047 "stack allocations, support can later be transparently implemented for glibc."
3051 #: build/C/man2/mmap.2:290
3053 msgid "B<MAP_UNINITIALIZED> (since Linux 2.6.33)"
3057 #: build/C/man2/mmap.2:300
3059 "Don't clear anonymous pages. This flag is intended to improve performance "
3060 "on embedded devices. This flag is only honored if the kernel was configured "
3061 "with the B<CONFIG_MMAP_ALLOW_UNINITIALIZED> option. Because of the security "
3062 "implications, that option is normally enabled only on embedded devices "
3063 "(i.e., devices where one has complete control of the contents of user "
3068 #: build/C/man2/mmap.2:308
3070 "Of the above flags, only B<MAP_FIXED> is specified in POSIX.1-2001. "
3071 "However, most systems also support B<MAP_ANONYMOUS> (or its synonym "
3076 #: build/C/man2/mmap.2:315
3078 "Some systems document the additional flags B<MAP_AUTOGROW>, "
3079 "B<MAP_AUTORESRV>, B<MAP_COPY>, and B<MAP_LOCAL>."
3083 #: build/C/man2/mmap.2:321
3085 "Memory mapped by B<mmap>() is preserved across B<fork>(2), with the same "
3090 #: build/C/man2/mmap.2:329
3092 "A file is mapped in multiples of the page size. For a file that is not a "
3093 "multiple of the page size, the remaining memory is zeroed when mapped, and "
3094 "writes to that region are not written out to the file. The effect of "
3095 "changing the size of the underlying file of a mapping on the pages that "
3096 "correspond to added or removed regions of the file is unspecified."
3100 #: build/C/man2/mmap.2:329
3106 #: build/C/man2/mmap.2:339
3108 "The B<munmap>() system call deletes the mappings for the specified address "
3109 "range, and causes further references to addresses within the range to "
3110 "generate invalid memory references. The region is also automatically "
3111 "unmapped when the process is terminated. On the other hand, closing the "
3112 "file descriptor does not unmap the region."
3116 #: build/C/man2/mmap.2:349
3118 "The address I<addr> must be a multiple of the page size. All pages "
3119 "containing a part of the indicated range are unmapped, and subsequent "
3120 "references to these pages will generate B<SIGSEGV>. It is not an error if "
3121 "the indicated range does not contain any mapped pages."
3125 #: build/C/man2/mmap.2:349
3127 msgid "Timestamps changes for file-backed mappings"
3131 #: build/C/man2/mmap.2:356
3133 "For file-backed mappings, the I<st_atime> field for the mapped file may be "
3134 "updated at any time between the B<mmap>() and the corresponding unmapping; "
3135 "the first reference to a mapped page will update the field if it has not "
3140 #: build/C/man2/mmap.2:373
3142 "The I<st_ctime> and I<st_mtime> field for a file mapped with B<PROT_WRITE> "
3143 "and B<MAP_SHARED> will be updated after a write to the mapped region, and "
3144 "before a subsequent B<msync>(2) with the B<MS_SYNC> or B<MS_ASYNC> flag, if "
3149 #: build/C/man2/mmap.2:390
3151 "On success, B<mmap>() returns a pointer to the mapped area. On error, the "
3152 "value B<MAP_FAILED> (that is, I<(void\\ *)\\ -1>) is returned, and I<errno> "
3153 "is set appropriately. On success, B<munmap>() returns 0, on failure -1, "
3154 "and I<errno> is set (probably to B<EINVAL>)."
3158 #: build/C/man2/mmap.2:391 build/C/man2/mprotect.2:80 build/C/man3/shm_open.3:185 build/C/man3/shm_open.3:190 build/C/man2/shmctl.2:295 build/C/man2/shmget.2:186 build/C/man2/shmop.2:189
3164 #: build/C/man2/mmap.2:411
3166 "A file descriptor refers to a non-regular file. Or B<MAP_PRIVATE> was "
3167 "requested, but I<fd> is not open for reading. Or B<MAP_SHARED> was "
3168 "requested and B<PROT_WRITE> is set, but I<fd> is not open in read/write "
3169 "(B<O_RDWR>) mode. Or B<PROT_WRITE> is set, but the file is append-only."
3173 #: build/C/man2/mmap.2:415
3175 "The file has been locked, or too much memory has been locked (see "
3180 #: build/C/man2/mmap.2:421
3181 msgid "I<fd> is not a valid file descriptor (and B<MAP_ANONYMOUS> was not set)."
3185 #: build/C/man2/mmap.2:429
3187 "We don't like I<addr>, I<length>, or I<offset> (e.g., they are too large, or "
3188 "not aligned on a page boundary)."
3192 #: build/C/man2/mmap.2:434
3193 msgid "(since Linux 2.6.12) I<length> was 0."
3197 #: build/C/man2/mmap.2:442
3199 "I<flags> contained neither B<MAP_PRIVATE> or B<MAP_SHARED>, or contained "
3200 "both of these values."
3204 #: build/C/man2/mmap.2:442 build/C/man3/shm_open.3:227 build/C/man2/shmget.2:202
3209 #. [2.6.7] shmem_zero_setup()-->shmem_file_setup()-->get_empty_filp()
3211 #: build/C/man2/mmap.2:450 build/C/man2/shmget.2:206
3212 msgid "The system limit on the total number of open files has been reached."
3216 #: build/C/man2/mmap.2:454
3218 "The underlying file system of the specified file does not support memory "
3223 #: build/C/man2/mmap.2:458
3225 "No memory is available, or the process's maximum number of mappings would "
3226 "have been exceeded."
3229 #. (Since 2.4.25 / 2.6.0.)
3231 #: build/C/man2/mmap.2:467
3233 "The I<prot> argument asks for B<PROT_EXEC> but the mapped area belongs to a "
3234 "file on a file system that was mounted no-exec."
3238 #: build/C/man2/mmap.2:467
3244 #: build/C/man2/mmap.2:473
3246 "B<MAP_DENYWRITE> was set but the object specified by I<fd> is open for "
3251 #: build/C/man2/mmap.2:475
3252 msgid "Use of a mapped region can result in these signals:"
3256 #: build/C/man2/mmap.2:475
3262 #: build/C/man2/mmap.2:478
3263 msgid "Attempted write into a region mapped as read-only."
3267 #: build/C/man2/mmap.2:478
3273 #: build/C/man2/mmap.2:483
3275 "Attempted access to a portion of the buffer that does not correspond to the "
3276 "file (for example, beyond the end of the file, including the case where "
3277 "another process has truncated the file)."
3280 #. SVr4 documents additional error codes ENXIO and ENODEV.
3281 #. SUSv2 documents additional error codes EMFILE and EOVERFLOW.
3283 #: build/C/man2/mmap.2:487
3284 msgid "SVr4, 4.4BSD, POSIX.1-2001."
3287 #. POSIX.1-2001: It shall be defined to -1 or 0 or 200112L.
3288 #. -1: unavailable, 0: ask using sysconf().
3289 #. glibc defines it to 1.
3291 #: build/C/man2/mmap.2:501
3293 "On POSIX systems on which B<mmap>(), B<msync>(2) and B<munmap>() are "
3294 "available, B<_POSIX_MAPPED_FILES> is defined in I<E<lt>unistd.hE<gt>> to a "
3295 "value greater than 0. (See also B<sysconf>(3).)"
3298 #. Since around glibc 2.1/2.2, depending on the platform.
3300 #: build/C/man2/mmap.2:512
3302 "Since kernel 2.4, this system call has been superseded by B<mmap2>(2). "
3303 "Nowadays, the glibc B<mmap>() wrapper function invokes B<mmap2>(2) with a "
3304 "suitably adjusted value for I<offset>."
3308 #: build/C/man2/mmap.2:525
3310 "On some hardware architectures (e.g., i386), B<PROT_WRITE> implies "
3311 "B<PROT_READ>. It is architecture dependent whether B<PROT_READ> implies "
3312 "B<PROT_EXEC> or not. Portable programs should always set B<PROT_EXEC> if "
3313 "they intend to execute code in the new mapping."
3317 #: build/C/man2/mmap.2:540
3319 "The portable way to create a mapping is to specify I<addr> as 0 (NULL), and "
3320 "omit B<MAP_FIXED> from I<flags>. In this case, the system chooses the "
3321 "address for the mapping; the address is chosen so as not to conflict with "
3322 "any existing mapping, and will not be 0. If the B<MAP_FIXED> flag is "
3323 "specified, and I<addr> is 0 (NULL), then the mapped address will be 0 "
3328 #: build/C/man2/mmap.2:545
3330 "On Linux there are no guarantees like those suggested above under "
3331 "B<MAP_NORESERVE>. By default, any process can be killed at any moment when "
3332 "the system runs out of memory."
3336 #: build/C/man2/mmap.2:552
3338 "In kernels before 2.6.7, the B<MAP_POPULATE> flag only has effect if I<prot> "
3339 "is specified as B<PROT_NONE>."
3343 #: build/C/man2/mmap.2:567
3345 "SUSv3 specifies that B<mmap>() should fail if I<length> is 0. However, in "
3346 "kernels before 2.6.12, B<mmap>() succeeded in this case: no mapping was "
3347 "created and the call returned I<addr>. Since kernel 2.6.12, B<mmap>() "
3348 "fails with the error B<EINVAL> for this case."
3352 #: build/C/man2/mmap.2:579
3354 "The following program prints part of the file specified in its first "
3355 "command-line argument to standard output. The range of bytes to be printed "
3356 "is specified via offset and length values in the second and third "
3357 "command-line arguments. The program creates a memory mapping of the "
3358 "required pages of the file and then uses B<write>(2) to output the desired "
3363 #: build/C/man2/mmap.2:587
3366 "#include E<lt>sys/mman.hE<gt>\n"
3367 "#include E<lt>sys/stat.hE<gt>\n"
3368 "#include E<lt>fcntl.hE<gt>\n"
3369 "#include E<lt>stdio.hE<gt>\n"
3370 "#include E<lt>stdlib.hE<gt>\n"
3371 "#include E<lt>unistd.hE<gt>\n"
3375 #: build/C/man2/mmap.2:590 build/C/man2/mprotect.2:172
3378 "#define handle_error(msg) \\e\n"
3379 " do { perror(msg); exit(EXIT_FAILURE); } while (0)\n"
3383 #: build/C/man2/mmap.2:600
3387 "main(int argc, char *argv[])\n"
3391 " struct stat sb;\n"
3392 " off_t offset, pa_offset;\n"
3398 #: build/C/man2/mmap.2:605
3401 " if (argc E<lt> 3 || argc E<gt> 4) {\n"
3402 " fprintf(stderr, \"%s file offset [length]\\en\", argv[0]);\n"
3403 " exit(EXIT_FAILURE);\n"
3408 #: build/C/man2/mmap.2:609
3411 " fd = open(argv[1], O_RDONLY);\n"
3413 " handle_error(\"open\");\n"
3417 #: build/C/man2/mmap.2:612
3420 " if (fstat(fd, &sb) == -1) /* To obtain file size */\n"
3421 " handle_error(\"fstat\");\n"
3425 #: build/C/man2/mmap.2:616
3428 " offset = atoi(argv[2]);\n"
3429 " pa_offset = offset & ~(sysconf(_SC_PAGE_SIZE) - 1);\n"
3430 " /* offset for mmap() must be page aligned */\n"
3434 #: build/C/man2/mmap.2:621
3437 " if (offset E<gt>= sb.st_size) {\n"
3438 " fprintf(stderr, \"offset is past end of file\\en\");\n"
3439 " exit(EXIT_FAILURE);\n"
3444 #: build/C/man2/mmap.2:627
3447 " if (argc == 4) {\n"
3448 " length = atoi(argv[3]);\n"
3449 " if (offset + length E<gt> sb.st_size)\n"
3450 " length = sb.st_size - offset;\n"
3451 " /* Can\\(aqt display bytes past end of file */\n"
3455 #: build/C/man2/mmap.2:631
3458 " } else { /* No length arg ==E<gt> display to end of file */\n"
3459 " length = sb.st_size - offset;\n"
3464 #: build/C/man2/mmap.2:636
3467 " addr = mmap(NULL, length + offset - pa_offset, PROT_READ,\n"
3468 " MAP_PRIVATE, fd, pa_offset);\n"
3469 " if (addr == MAP_FAILED)\n"
3470 " handle_error(\"mmap\");\n"
3474 #: build/C/man2/mmap.2:641
3477 " s = write(STDOUT_FILENO, addr + offset - pa_offset, length);\n"
3478 " if (s != length) {\n"
3480 " handle_error(\"write\");\n"
3484 #: build/C/man2/mmap.2:645
3487 " fprintf(stderr, \"partial write\");\n"
3488 " exit(EXIT_FAILURE);\n"
3493 #: build/C/man2/mmap.2:662
3495 "B<getpagesize>(2), B<mincore>(2), B<mlock>(2), B<mmap2>(2), B<mprotect>(2), "
3496 "B<mremap>(2), B<msync>(2), B<remap_file_pages>(2), B<setrlimit>(2), "
3497 "B<shmat>(2), B<shm_open>(3), B<shm_overview>(7)"
3501 #: build/C/man2/mmap.2:667 build/C/man2/msync.2:122
3502 msgid "B.O. Gallmeister, POSIX.4, O'Reilly, pp. 128-129 and 389-391."
3506 #: build/C/man2/mmap2.2:29
3512 #: build/C/man2/mmap2.2:29 build/C/man2/msync.2:25 build/C/man2/remap_file_pages.2:26
3518 #: build/C/man2/mmap2.2:32
3519 msgid "mmap2 - map files or devices into memory"
3523 #: build/C/man2/mmap2.2:38
3526 "B<void *mmap2(void *>I<addr>B<, size_t >I<length>B<, int >I<prot>B<,>\n"
3527 "B< int >I<flags>B<, int >I<fd>B<, off_t >I<pgoffset>B<);>\n"
3531 #: build/C/man2/mmap2.2:50
3533 "The B<mmap2>() system call operates in exactly the same way as B<mmap>(2), "
3534 "except that the final argument specifies the offset into the file in "
3535 "4096-byte units (instead of bytes, as is done by B<mmap>(2)). This enables "
3536 "applications that use a 32-bit I<off_t> to map large files (up to 2^44 "
3541 #: build/C/man2/mmap2.2:57
3543 "On success, B<mmap2>() returns a pointer to the mapped area. On error -1 "
3544 "is returned and I<errno> is set appropriately."
3548 #: build/C/man2/mmap2.2:61
3549 msgid "Problem with getting the data from userspace."
3553 #: build/C/man2/mmap2.2:66
3555 "(Various platforms where the page size is not 4096 bytes.) I<offset * 4096> "
3556 "is not a multiple of the system page size."
3560 #: build/C/man2/mmap2.2:70
3561 msgid "B<mmap2>() can return any of the same errors as B<mmap>(2)."
3565 #: build/C/man2/mmap2.2:73
3566 msgid "B<mmap2>() is available since Linux 2.3.31."
3570 #: build/C/man2/mmap2.2:75 build/C/man2/subpage_prot.2:93
3571 msgid "This system call is Linux-specific."
3575 #: build/C/man2/mmap2.2:81
3577 "Nowadays, the glibc B<mmap>() wrapper function invokes this system call "
3578 "rather than the B<mmap>(2) system call."
3581 #. ia64 can have page sizes ranging from 4kB to 64kB.
3582 #. On cris, it looks like the unit might also be the page size,
3583 #. which is 8192 bytes. -- mtk, June 2007
3585 #: build/C/man2/mmap2.2:88
3587 "On ia64, the unit for I<offset> is actually the system page size, rather "
3592 #: build/C/man2/mmap2.2:94
3593 msgid "B<getpagesize>(2), B<mmap>(2), B<mremap>(2), B<msync>(2), B<shm_open>(3)"
3597 #: build/C/man2/mprotect.2:32
3603 #: build/C/man2/mprotect.2:32
3609 #: build/C/man2/mprotect.2:35
3610 msgid "mprotect - set protection on a region of memory"
3614 #: build/C/man2/mprotect.2:40
3616 msgid "B<int mprotect(const void *>I<addr>B<, size_t >I<len>B<, int >I<prot>B<);>\n"
3620 #: build/C/man2/mprotect.2:48
3622 "B<mprotect>() changes protection for the calling process's memory page(s) "
3623 "containing any part of the address range in the interval [I<addr>,\\ "
3624 "I<addr>+I<len>-1]. I<addr> must be aligned to a page boundary."
3628 #: build/C/man2/mprotect.2:53
3630 "If the calling process tries to access memory in a manner that violates the "
3631 "protection, then the kernel generates a B<SIGSEGV> signal for the process."
3635 #: build/C/man2/mprotect.2:58
3637 "I<prot> is either B<PROT_NONE> or a bitwise-or of the other values in the "
3642 #: build/C/man2/mprotect.2:61
3643 msgid "The memory cannot be accessed at all."
3647 #: build/C/man2/mprotect.2:64
3648 msgid "The memory can be read."
3652 #: build/C/man2/mprotect.2:67
3653 msgid "The memory can be modified."
3657 #. Document PROT_GROWSUP and PROT_GROWSDOWN
3659 #: build/C/man2/mprotect.2:72
3660 msgid "The memory can be executed."
3664 #: build/C/man2/mprotect.2:79
3666 "On success, B<mprotect>() returns zero. On error, -1 is returned, and "
3667 "I<errno> is set appropriately."
3671 #: build/C/man2/mprotect.2:89
3673 "The memory cannot be given the specified access. This can happen, for "
3674 "example, if you B<mmap>(2) a file to which you have read-only access, then "
3675 "ask B<mprotect>() to mark it B<PROT_WRITE>."
3678 #. Or: both PROT_GROWSUP and PROT_GROWSDOWN were specified in 'prot'.
3680 #: build/C/man2/mprotect.2:94
3681 msgid "I<addr> is not a valid pointer, or not a multiple of the system page size."
3685 #: build/C/man2/mprotect.2:97
3686 msgid "Internal kernel structures could not be allocated."
3690 #: build/C/man2/mprotect.2:107
3692 "Addresses in the range [I<addr>, I<addr>+I<len>-1] are invalid for the "
3693 "address space of the process, or specify one or more pages that are not "
3694 "mapped. (Before kernel 2.4.19, the error B<EFAULT> was incorrectly produced "
3698 #. SVr4 defines an additional error
3699 #. code EAGAIN. The SVr4 error conditions don't map neatly onto Linux's.
3701 #: build/C/man2/mprotect.2:116
3703 "SVr4, POSIX.1-2001. POSIX says that the behavior of B<mprotect>() is "
3704 "unspecified if it is applied to a region of memory that was not obtained via "
3709 #: build/C/man2/mprotect.2:123
3711 "On Linux it is always permissible to call B<mprotect>() on any address in a "
3712 "process's address space (except for the kernel vsyscall area). In "
3713 "particular it can be used to change existing code mappings to be writable."
3717 #: build/C/man2/mprotect.2:133
3719 "Whether B<PROT_EXEC> has any effect different from B<PROT_READ> is "
3720 "architecture- and kernel version-dependent. On some hardware architectures "
3721 "(e.g., i386), B<PROT_WRITE> implies B<PROT_READ>."
3725 #: build/C/man2/mprotect.2:142
3727 "POSIX.1-2001 says that an implementation may permit access other than that "
3728 "specified in I<prot>, but at a minimum can only allow write access if "
3729 "B<PROT_WRITE> has been set, and must not allow any access if B<PROT_NONE> "
3734 #: build/C/man2/mprotect.2:148
3736 "The program below allocates four pages of memory, makes the third of these "
3737 "pages read-only, and then executes a loop that walks upward through the "
3738 "allocated region modifying bytes."
3742 #: build/C/man2/mprotect.2:151
3743 msgid "An example of what we might see when running the program is the following:"
3747 #: build/C/man2/mprotect.2:157
3751 "Start of region: 0x804c000\n"
3752 "Got SIGSEGV at address: 0x804e000\n"
3756 #: build/C/man2/mprotect.2:169
3759 "#include E<lt>unistd.hE<gt>\n"
3760 "#include E<lt>signal.hE<gt>\n"
3761 "#include E<lt>stdio.hE<gt>\n"
3762 "#include E<lt>malloc.hE<gt>\n"
3763 "#include E<lt>stdlib.hE<gt>\n"
3764 "#include E<lt>errno.hE<gt>\n"
3765 "#include E<lt>sys/mman.hE<gt>\n"
3769 #: build/C/man2/mprotect.2:174
3771 msgid "char *buffer;\n"
3775 #: build/C/man2/mprotect.2:182
3779 "handler(int sig, siginfo_t *si, void *unused)\n"
3781 " printf(\"Got SIGSEGV at address: 0x%lx\\en\",\n"
3782 " (long) si-E<gt>si_addr);\n"
3783 " exit(EXIT_FAILURE);\n"
3788 #: build/C/man2/mprotect.2:189
3792 "main(int argc, char *argv[])\n"
3796 " struct sigaction sa;\n"
3800 #: build/C/man2/mprotect.2:195
3803 " sa.sa_flags = SA_SIGINFO;\n"
3804 " sigemptyset(&sa.sa_mask);\n"
3805 " sa.sa_sigaction = handler;\n"
3806 " if (sigaction(SIGSEGV, &sa, NULL) == -1)\n"
3807 " handle_error(\"sigaction\");\n"
3811 #: build/C/man2/mprotect.2:199
3814 " pagesize = sysconf(_SC_PAGE_SIZE);\n"
3815 " if (pagesize == -1)\n"
3816 " handle_error(\"sysconf\");\n"
3820 #: build/C/man2/mprotect.2:202
3823 " /* Allocate a buffer aligned on a page boundary;\n"
3824 " initial protection is PROT_READ | PROT_WRITE */\n"
3828 #: build/C/man2/mprotect.2:206
3831 " buffer = memalign(pagesize, 4 * pagesize);\n"
3832 " if (buffer == NULL)\n"
3833 " handle_error(\"memalign\");\n"
3837 #: build/C/man2/mprotect.2:208
3839 msgid " printf(\"Start of region: 0x%lx\\en\", (long) buffer);\n"
3843 #: build/C/man2/mprotect.2:212
3846 " if (mprotect(buffer + pagesize * 2, pagesize,\n"
3847 " PROT_READ) == -1)\n"
3848 " handle_error(\"mprotect\");\n"
3852 #: build/C/man2/mprotect.2:215
3855 " for (p = buffer ; ; )\n"
3856 " *(p++) = \\(aqa\\(aq;\n"
3860 #: build/C/man2/mprotect.2:219
3863 " printf(\"Loop completed\\en\"); /* Should never happen */\n"
3864 " exit(EXIT_SUCCESS);\n"
3869 #: build/C/man2/mprotect.2:223
3870 msgid "B<mmap>(2), B<sysconf>(3)"
3874 #: build/C/man2/mremap.2:31
3880 #: build/C/man2/mremap.2:31
3886 #: build/C/man2/mremap.2:34
3887 msgid "mremap - remap a virtual memory address"
3891 #: build/C/man2/mremap.2:37
3893 msgid "B<#define _GNU_SOURCE> /* See feature_test_macros(7) */\n"
3897 #: build/C/man2/mremap.2:42
3900 "B<void *mremap(void *>I<old_address>B<, size_t >I<old_size>B<,>\n"
3901 "B< size_t >I<new_size>B<, int >I<flags>B<, ... /* void "
3902 "*>I<new_address>B< */);>\n"
3906 #: build/C/man2/mremap.2:48
3908 "B<mremap>() expands (or shrinks) an existing memory mapping, potentially "
3909 "moving it at the same time (controlled by the I<flags> argument and the "
3910 "available virtual address space)."
3914 #: build/C/man2/mremap.2:62
3916 "I<old_address> is the old address of the virtual memory block that you want "
3917 "to expand (or shrink). Note that I<old_address> has to be page aligned. "
3918 "I<old_size> is the old size of the virtual memory block. I<new_size> is the "
3919 "requested size of the virtual memory block after the resize. An optional "
3920 "fifth argument, I<new_address>, may be provided; see the description of "
3921 "B<MREMAP_FIXED> below."
3925 #: build/C/man2/mremap.2:74
3927 "In Linux the memory is divided into pages. A user process has (one or) "
3928 "several linear virtual memory segments. Each virtual memory segment has one "
3929 "or more mappings to real memory pages (in the page table). Each virtual "
3930 "memory segment has its own protection (access rights), which may cause a "
3931 "segmentation violation if the memory is accessed incorrectly (e.g., writing "
3932 "to a read-only segment). Accessing virtual memory outside of the segments "
3933 "will also cause a segmentation violation."
3937 #: build/C/man2/mremap.2:82
3939 "B<mremap>() uses the Linux page table scheme. B<mremap>() changes the "
3940 "mapping between virtual addresses and memory pages. This can be used to "
3941 "implement a very efficient B<realloc>(3)."
3945 #: build/C/man2/mremap.2:84
3946 msgid "The I<flags> bit-mask argument may be 0, or include the following flag:"
3950 #: build/C/man2/mremap.2:84
3952 msgid "B<MREMAP_MAYMOVE>"
3956 #: build/C/man2/mremap.2:96
3958 "By default, if there is not sufficient space to expand a mapping at its "
3959 "current location, then B<mremap>() fails. If this flag is specified, then "
3960 "the kernel is permitted to relocate the mapping to a new virtual address, if "
3961 "necessary. If the mapping is relocated, then absolute pointers into the old "
3962 "mapping location become invalid (offsets relative to the starting address of "
3963 "the mapping should be employed)."
3967 #: build/C/man2/mremap.2:96
3969 msgid "B<MREMAP_FIXED> (since Linux 2.3.31)"
3973 #: build/C/man2/mremap.2:118
3975 "This flag serves a similar purpose to the B<MAP_FIXED> flag of B<mmap>(2). "
3976 "If this flag is specified, then B<mremap>() accepts a fifth argument, "
3977 "I<void *new_address>, which specifies a page-aligned address to which the "
3978 "mapping must be moved. Any previous mapping at the address range specified "
3979 "by I<new_address> and I<new_size> is unmapped. If B<MREMAP_FIXED> is "
3980 "specified, then B<MREMAP_MAYMOVE> must also be specified."
3984 #: build/C/man2/mremap.2:128
3986 "If the memory segment specified by I<old_address> and I<old_size> is locked "
3987 "(using B<mlock>(2) or similar), then this lock is maintained when the "
3988 "segment is resized and/or relocated. As a consequence, the amount of memory "
3989 "locked by the process may change."
3993 #: build/C/man2/mremap.2:136
3995 "On success B<mremap>() returns a pointer to the new virtual memory area. "
3996 "On error, the value B<MAP_FAILED> (that is, I<(void\\ *)\\ -1>) is returned, "
3997 "and I<errno> is set appropriately."
4001 #: build/C/man2/mremap.2:143
4003 "The caller tried to expand a memory segment that is locked, but this was not "
4004 "possible without exceeding the B<RLIMIT_MEMLOCK> resource limit."
4008 #: build/C/man2/mremap.2:152
4010 "\"Segmentation fault.\" Some address in the range I<old_address> to "
4011 "I<old_address>+I<old_size> is an invalid virtual memory address for this "
4012 "process. You can also get B<EFAULT> even if there exist mappings that cover "
4013 "the whole address space requested, but those mappings are of different "
4018 #: build/C/man2/mremap.2:180
4020 "An invalid argument was given. Possible causes are: I<old_address> was not "
4021 "page aligned; a value other than B<MREMAP_MAYMOVE> or B<MREMAP_FIXED> was "
4022 "specified in I<flags>; I<new_size> was zero; I<new_size> or I<new_address> "
4023 "was invalid; or the new address range specified by I<new_address> and "
4024 "I<new_size> overlapped the old address range specified by I<old_address> and "
4025 "I<old_size>; or B<MREMAP_FIXED> was specified without also specifying "
4026 "B<MREMAP_MAYMOVE>."
4030 #: build/C/man2/mremap.2:186
4032 "The memory area cannot be expanded at the current virtual address, and the "
4033 "B<MREMAP_MAYMOVE> flag is not set in I<flags>. Or, there is not enough "
4034 "(virtual) memory available."
4037 #. 4.2BSD had a (never actually implemented)
4039 #. call with completely different semantics.
4041 #: build/C/man2/mremap.2:192
4043 "This call is Linux-specific, and should not be used in programs intended to "
4048 #: build/C/man2/mremap.2:200
4050 "Prior to version 2.4, glibc did not expose the definition of "
4051 "B<MREMAP_FIXED>, and the prototype for B<mremap>() did not allow for the "
4052 "I<new_address> argument."
4056 #: build/C/man2/mremap.2:209
4058 "B<brk>(2), B<getpagesize>(2), B<getrlimit>(2), B<mlock>(2), B<mmap>(2), "
4059 "B<sbrk>(2), B<malloc>(3), B<realloc>(3)"
4063 #: build/C/man2/mremap.2:214
4065 "Your favorite OS text book for more information on paged memory. (I<Modern "
4066 "Operating Systems> by Andrew S. Tanenbaum, I<Inside Linux> by Randolf "
4067 "Bentson, I<The Design of the UNIX Operating System> by Maurice J. Bach.)"
4071 #: build/C/man2/msync.2:25
4077 #: build/C/man2/msync.2:28
4078 msgid "msync - synchronize a file with a memory map"
4082 #: build/C/man2/msync.2:32
4083 msgid "B<int msync(void *>I<addr>B<, size_t >I<length>B<, int >I<flags>B<);>"
4087 #: build/C/man2/msync.2:48
4089 "B<msync>() flushes changes made to the in-core copy of a file that was "
4090 "mapped into memory using B<mmap>(2) back to disk. Without use of this call "
4091 "there is no guarantee that changes are written back before B<munmap>(2) is "
4092 "called. To be more precise, the part of the file that corresponds to the "
4093 "memory area starting at I<addr> and having length I<length> is updated."
4097 #: build/C/man2/msync.2:68
4099 "The I<flags> argument may have the bits B<MS_ASYNC>, B<MS_SYNC>, and "
4100 "B<MS_INVALIDATE> set, but not both B<MS_ASYNC> and B<MS_SYNC>. B<MS_ASYNC> "
4101 "specifies that an update be scheduled, but the call returns immediately. "
4102 "B<MS_SYNC> asks for an update and waits for it to complete. "
4103 "B<MS_INVALIDATE> asks to invalidate other mappings of the same file (so that "
4104 "they can be updated with the fresh values just written)."
4108 #: build/C/man2/msync.2:73
4110 "On success, zero is returned. On error, -1 is returned, and I<errno> is set "
4115 #: build/C/man2/msync.2:74
4121 #: build/C/man2/msync.2:80
4123 "B<MS_INVALIDATE> was specified in I<flags>, and a memory lock exists for the "
4124 "specified address range."
4128 #: build/C/man2/msync.2:93
4130 "I<addr> is not a multiple of PAGESIZE; or any bit other than B<MS_ASYNC> | "
4131 "B<MS_INVALIDATE> | B<MS_SYNC> is set in I<flags>; or both B<MS_SYNC> and "
4132 "B<MS_ASYNC> are set in I<flags>."
4136 #: build/C/man2/msync.2:96
4137 msgid "The indicated memory (or part of it) was not mapped."
4141 #: build/C/man2/msync.2:98 build/C/man3/posix_fallocate.3:103 build/C/man3/shm_open.3:251 build/C/man7/shm_overview.7:103
4142 msgid "POSIX.1-2001."
4146 #: build/C/man2/msync.2:105
4148 "This call was introduced in Linux 1.3.21, and then used B<EFAULT> instead of "
4149 "B<ENOMEM>. In Linux 2.4.19 this was changed to the POSIX value B<ENOMEM>."
4152 #. POSIX.1-2001: It shall be defined to -1 or 0 or 200112L.
4153 #. -1: unavailable, 0: ask using sysconf().
4154 #. glibc defines them to 1.
4156 #: build/C/man2/msync.2:118
4158 "On POSIX systems on which B<msync>() is available, both "
4159 "B<_POSIX_MAPPED_FILES> and B<_POSIX_SYNCHRONIZED_IO> are defined in "
4160 "I<E<lt>unistd.hE<gt>> to a value greater than 0. (See also B<sysconf>(3).)"
4164 #: build/C/man2/msync.2:120 build/C/man7/shm_overview.7:46
4170 #: build/C/man3/mtrace.3:23
4176 #: build/C/man3/mtrace.3:26
4177 msgid "mtrace, muntrace - malloc tracing"
4181 #: build/C/man3/mtrace.3:28
4182 msgid "B<#include E<lt>mcheck.hE<gt>>"
4186 #: build/C/man3/mtrace.3:30
4187 msgid "B<void mtrace(void);>"
4191 #: build/C/man3/mtrace.3:32
4192 msgid "B<void muntrace(void);>"
4196 #: build/C/man3/mtrace.3:44
4198 "The B<mtrace>() function installs hook functions for the memory-allocation "
4199 "functions (B<malloc>(3), B<realloc>(3) B<memalign>(3), B<free>(3)). These "
4200 "hook functions record tracing information about memory allocation and "
4201 "deallocation. The tracing information can be used to discover memory leaks "
4202 "and attempts to free nonallocated memory in a program."
4206 #: build/C/man3/mtrace.3:55
4208 "The B<muntrace>() function disables the hook functions installed by "
4209 "B<mtrace>(), so that tracing information is no longer recorded for the "
4210 "memory-allocation functions. If no hook functions were successfully "
4211 "installed by B<mtrace>(), B<muntrace>() does nothing."
4215 #: build/C/man3/mtrace.3:63
4217 "When B<mtrace>(3) is called, it checks the value of the environment "
4218 "variable B<MALLOC_TRACE>, which should contain the pathname of a file in "
4219 "which the tracing information is to be recorded. If the pathname is "
4220 "successfully opened, it is truncated to zero length."
4224 #: build/C/man3/mtrace.3:76
4226 "If B<MALLOC_TRACE> is not set, or the pathname it specifies is invalid or "
4227 "not writable, then no hook functions are installed, and B<mtrace>() has no "
4228 "effect. In set-user-ID and set-group-ID programs, B<MALLOC_TRACE> is "
4229 "ignored, and B<mtrace>() has no effect."
4233 #: build/C/man3/mtrace.3:84
4235 "In normal usage, B<mtrace>() is called once at the start of execution of a "
4236 "program, and B<muntrace>() is never called."
4240 #: build/C/man3/mtrace.3:94
4242 "The tracing output produced after a call to B<mtrace>() is textual, but not "
4243 "designed to be human readable. The GNU C library provides a Perl script, "
4244 "B<mtrace>(1), that interprets the trace log and produces human-readable "
4245 "output. For best results, the traced program should be compiled with "
4246 "debugging enabled, so that line-number information is recorded in the "
4251 #: build/C/man3/mtrace.3:100
4253 "The tracing performed by B<mtrace>() incurs a performance penalty (if "
4254 "B<MALLOC_TRACE> points to a valid, writable pathname)."
4258 #: build/C/man3/mtrace.3:106
4260 "The line-number information produced by B<mtrace>(1) is not always precise: "
4261 "the line number references may refer to the previous or following "
4262 "(non-blank) line of the source code."
4266 #: build/C/man3/mtrace.3:113
4268 "The shell session below demonstrates the use of the B<mtrace>() function "
4269 "and the B<mtrace>(1) command in a program that has memory leaks at two "
4270 "different locations. The demonstration uses the following program:"
4274 #: build/C/man3/mtrace.3:120
4277 "$ B<cat t_mtrace.c>\n"
4278 "#include E<lt>mcheck.hE<gt>\n"
4279 "#include E<lt>stdlib.hE<gt>\n"
4280 "#include E<lt>stdio.hE<gt>\n"
4284 #: build/C/man3/mtrace.3:125
4288 "main(int argc, char *argv[])\n"
4294 #: build/C/man3/mtrace.3:127
4296 msgid " mtrace();\n"
4300 #: build/C/man3/mtrace.3:130
4303 " for (j = 0; j E<lt> 2; j++)\n"
4304 " malloc(100); /* Never freed--a memory leak */\n"
4308 #: build/C/man3/mtrace.3:134
4311 " calloc(16, 16); /* Never freed--a memory leak */\n"
4312 " exit(EXIT_SUCCESS);\n"
4317 #: build/C/man3/mtrace.3:140
4319 "When we run the program as follows, we see that B<mtrace>() diagnosed "
4320 "memory leaks at two different locations in the program:"
4324 #: build/C/man3/mtrace.3:153
4327 "$ B<cc -g t_mtrace.c -o t_mtrace>\n"
4328 "$ B<export MALLOC_TRACE=/tmp/t>\n"
4330 "$ B<mtrace ./t_mtrace $MALLOC_TRACE>\n"
4331 "Memory not freed:\n"
4332 "-----------------\n"
4333 " Address Size Caller\n"
4334 "0x084c9378 0x64 at /home/cecilia/t_mtrace.c:12\n"
4335 "0x084c93e0 0x64 at /home/cecilia/t_mtrace.c:12\n"
4336 "0x084c9448 0x100 at /home/cecilia/t_mtrace.c:16\n"
4340 #: build/C/man3/mtrace.3:165
4342 "The first two messages about unfreed memory correspond to the two "
4343 "B<malloc>(3) calls inside the I<for> loop. The final message corresponds "
4344 "to the call to B<calloc>(3) (which in turn calls B<malloc>(3))."
4348 #: build/C/man3/mtrace.3:169
4349 msgid "B<mtrace>(1), B<malloc>(3), B<malloc_hook>(3)"
4353 #: build/C/man2/posix_fadvise.2:27
4355 msgid "POSIX_FADVISE"
4359 #: build/C/man2/posix_fadvise.2:27 build/C/man2/sync_file_range.2:30
4365 #: build/C/man2/posix_fadvise.2:30
4366 msgid "posix_fadvise - predeclare an access pattern for file data"
4370 #: build/C/man2/posix_fadvise.2:33 build/C/man3/posix_fallocate.3:29
4372 msgid "B<#include E<lt>fcntl.hE<gt>>\n"
4376 #: build/C/man2/posix_fadvise.2:36
4379 "B<int posix_fadvise(int >I<fd>B<, off_t >I<offset>B<, off_t >I<len>B<, int "
4384 #: build/C/man2/posix_fadvise.2:45
4385 msgid "B<posix_fadvise>():"
4389 #: build/C/man2/posix_fadvise.2:47 build/C/man3/posix_fallocate.3:42
4390 msgid "_XOPEN_SOURCE\\ E<gt>=\\ 600 || _POSIX_C_SOURCE\\ E<gt>=\\ 200112L"
4394 #: build/C/man2/posix_fadvise.2:55
4396 "Programs can use B<posix_fadvise>() to announce an intention to access file "
4397 "data in a specific pattern in the future, thus allowing the kernel to "
4398 "perform appropriate optimizations."
4402 #: build/C/man2/posix_fadvise.2:62
4404 "The I<advice> applies to a (not necessarily existent) region starting at "
4405 "I<offset> and extending for I<len> bytes (or until the end of the file if "
4406 "I<len> is 0) within the file referred to by I<fd>. The I<advice> is not "
4407 "binding; it merely constitutes an expectation on behalf of the application."
4411 #: build/C/man2/posix_fadvise.2:64
4412 msgid "Permissible values for I<advice> include:"
4416 #: build/C/man2/posix_fadvise.2:64
4418 msgid "B<POSIX_FADV_NORMAL>"
4422 #: build/C/man2/posix_fadvise.2:70
4424 "Indicates that the application has no advice to give about its access "
4425 "pattern for the specified data. If no advice is given for an open file, "
4426 "this is the default assumption."
4430 #: build/C/man2/posix_fadvise.2:70
4432 msgid "B<POSIX_FADV_SEQUENTIAL>"
4436 #: build/C/man2/posix_fadvise.2:74
4438 "The application expects to access the specified data sequentially (with "
4439 "lower offsets read before higher ones)."
4443 #: build/C/man2/posix_fadvise.2:74
4445 msgid "B<POSIX_FADV_RANDOM>"
4449 #: build/C/man2/posix_fadvise.2:77
4450 msgid "The specified data will be accessed in random order."
4454 #: build/C/man2/posix_fadvise.2:77
4456 msgid "B<POSIX_FADV_NOREUSE>"
4460 #: build/C/man2/posix_fadvise.2:80
4461 msgid "The specified data will be accessed only once."
4465 #: build/C/man2/posix_fadvise.2:80
4467 msgid "B<POSIX_FADV_WILLNEED>"
4471 #: build/C/man2/posix_fadvise.2:83
4472 msgid "The specified data will be accessed in the near future."
4476 #: build/C/man2/posix_fadvise.2:83
4478 msgid "B<POSIX_FADV_DONTNEED>"
4482 #: build/C/man2/posix_fadvise.2:86
4483 msgid "The specified data will not be accessed in the near future."
4487 #: build/C/man2/posix_fadvise.2:89
4488 msgid "On success, zero is returned. On error, an error number is returned."
4492 #: build/C/man2/posix_fadvise.2:93
4493 msgid "The I<fd> argument was not a valid file descriptor."
4497 #: build/C/man2/posix_fadvise.2:96
4498 msgid "An invalid value was specified for I<advice>."
4502 #: build/C/man2/posix_fadvise.2:103
4504 "The specified file descriptor refers to a pipe or FIFO. (Linux actually "
4505 "returns B<EINVAL> in this case.)"
4508 #. of fadvise64_64()
4510 #: build/C/man2/posix_fadvise.2:111
4512 "Kernel support first appeared in Linux 2.5.60; the underlying system call is "
4513 "called B<fadvise64>(). Library support has been provided since glibc "
4514 "version 2.2, via the wrapper function B<posix_fadvise>()."
4518 #: build/C/man2/posix_fadvise.2:120
4520 "POSIX.1-2001. Note that the type of the I<len> argument was changed from "
4521 "I<size_t> to I<off_t> in POSIX.1-2003 TC1."
4525 #: build/C/man2/posix_fadvise.2:126
4527 "Under Linux, B<POSIX_FADV_NORMAL> sets the readahead window to the default "
4528 "size for the backing device; B<POSIX_FADV_SEQUENTIAL> doubles this size, and "
4529 "B<POSIX_FADV_RANDOM> disables file readahead entirely. These changes affect "
4530 "the entire file, not just the specified region (but other open file handles "
4531 "to the same file are unaffected)."
4535 #: build/C/man2/posix_fadvise.2:133
4537 "B<POSIX_FADV_WILLNEED> initiates a nonblocking read of the specified region "
4538 "into the page cache. The amount of data read may be decreased by the kernel "
4539 "depending on virtual memory load. (A few megabytes will usually be fully "
4540 "satisfied, and more is rarely useful.)"
4544 #: build/C/man2/posix_fadvise.2:137
4546 "In kernels before 2.6.18, B<POSIX_FADV_NOREUSE> had the same semantics as "
4547 "B<POSIX_FADV_WILLNEED>. This was probably a bug; since kernel 2.6.18, this "
4552 #: build/C/man2/posix_fadvise.2:145
4554 "B<POSIX_FADV_DONTNEED> attempts to free cached pages associated with the "
4555 "specified region. This is useful, for example, while streaming large "
4556 "files. A program may periodically request the kernel to free cached data "
4557 "that has already been used, so that more useful cached pages are not "
4558 "discarded instead."
4562 #: build/C/man2/posix_fadvise.2:153
4564 "Pages that have not yet been written out will be unaffected, so if the "
4565 "application wishes to guarantee that pages will be released, it should call "
4566 "B<fsync>(2) or B<fdatasync>(2) first."
4570 #: build/C/man2/posix_fadvise.2:153
4572 msgid "arm_fadvise()"
4576 #: build/C/man2/posix_fadvise.2:165
4578 "The ARM architecture needs 64-bit arguments to be aligned in a suitable pair "
4579 "of registers. On this architecture, the call signature of "
4580 "B<posix_fadvise>() is flawed, since it forces a register to be wasted as "
4581 "padding between the I<fd> and I<len> arguments. Therefore, since Linux "
4582 "2.6.14, ARM defines a different system call that orders the arguments "
4587 #: build/C/man2/posix_fadvise.2:170
4590 "B<long arm_fadvise64_64(int >I<fd>B<, int >I<advice>B<,>\n"
4591 "B< loff_t >I<offset>B<, loff_t >I<len>B<);>\n"
4594 #. No ARM support in glibc.
4596 #: build/C/man2/posix_fadvise.2:177
4598 "The behavior of this system call is otherwise exactly the same as "
4599 "B<posix_fadvise>(). No library support is provided for this system call in "
4604 #: build/C/man2/posix_fadvise.2:182
4606 "In kernels before 2.6.6, if I<len> was specified as 0, then this was "
4607 "interpreted literally as \"zero bytes\", rather than as meaning \"all bytes "
4608 "through to the end of the file\"."
4611 #. FIXME . Write a posix_fadvise(3) page.
4613 #: build/C/man2/posix_fadvise.2:188
4615 "B<readahead>(2), B<sync_file_range>(2), B<posix_fallocate>(3), "
4616 "B<posix_madvise>(3)"
4620 #: build/C/man3/posix_fallocate.3:23
4622 msgid "POSIX_FALLOCATE"
4626 #: build/C/man3/posix_fallocate.3:23
4632 #: build/C/man3/posix_fallocate.3:26
4633 msgid "posix_fallocate - allocate file space"
4637 #: build/C/man3/posix_fallocate.3:31
4639 msgid "B<int posix_fallocate(int >I<fd>B<, off_t >I<offset>B<, off_t >I<len>B<);>\n"
4643 #: build/C/man3/posix_fallocate.3:40
4644 msgid "B<posix_fallocate>():"
4648 #: build/C/man3/posix_fallocate.3:59
4650 "The function B<posix_fallocate>() ensures that disk space is allocated for "
4651 "the file referred to by the descriptor I<fd> for the bytes in the range "
4652 "starting at I<offset> and continuing for I<len> bytes. After a successful "
4653 "call to B<posix_fallocate>(), subsequent writes to bytes in the specified "
4654 "range are guaranteed not to fail because of lack of disk space."
4658 #: build/C/man3/posix_fallocate.3:64
4660 "If the size of the file is less than I<offset>+I<len>, then the file is "
4661 "increased to this size; otherwise the file size is left unchanged."
4665 #: build/C/man3/posix_fallocate.3:70
4667 "B<posix_fallocate>() returns zero on success, or an error number on "
4668 "failure. Note that I<errno> is not set."
4672 #: build/C/man3/posix_fallocate.3:79
4673 msgid "I<offset+len> exceeds the maximum file size."
4677 #: build/C/man3/posix_fallocate.3:89
4678 msgid "I<fd> does not refer to a regular file."
4682 #: build/C/man3/posix_fallocate.3:98
4683 msgid "I<fd> refers to a pipe."
4687 #: build/C/man3/posix_fallocate.3:101
4688 msgid "B<posix_fallocate>() is available since glibc 2.1.94."
4692 #: build/C/man3/posix_fallocate.3:126
4694 "POSIX.1-2008 says that an implementation I<shall> give the B<EINVAL> error "
4695 "if I<len> was 0, or I<offset> was less than 0. POSIX.1-2001 says that an "
4696 "implementation I<shall> give the B<EINVAL> error if I<len> is less than 0, "
4697 "or I<offset> was less than 0, and I<may> give the error if I<len> equals "
4702 #: build/C/man3/posix_fallocate.3:130
4703 msgid "B<fallocate>(2), B<lseek>(2), B<posix_fadvise>(2)"
4707 #: build/C/man3/posix_memalign.3:28
4709 msgid "POSIX_MEMALIGN"
4713 #: build/C/man3/posix_memalign.3:28
4719 #: build/C/man3/posix_memalign.3:31
4721 "posix_memalign, aligned_alloc, memalign, valloc, pvalloc - Allocate aligned "
4726 #: build/C/man3/posix_memalign.3:34
4728 msgid "B<#include E<lt>stdlib.hE<gt>>\n"
4732 #: build/C/man3/posix_memalign.3:38
4735 "B<int posix_memalign(void **>I<memptr>B<, size_t >I<alignment>B<, size_t "
4737 "B<void *aligned_alloc(size_t >I<alignment>B<, size_t >I<size>B<);>\n"
4738 "B<void *valloc(size_t >I<size>B<);>\n"
4742 #: build/C/man3/posix_memalign.3:43
4745 "B<void *memalign(size_t >I<alignment>B<, size_t >I<size>B<);>\n"
4746 "B<void *pvalloc(size_t >I<size>B<);>\n"
4750 #: build/C/man3/posix_memalign.3:53
4752 "B<posix_memalign>(): _POSIX_C_SOURCE\\ E<gt>=\\ 200112L || _XOPEN_SOURCE\\ "
4757 #: build/C/man3/posix_memalign.3:56
4758 msgid "B<aligned_alloc>(): _ISOC11_SOURCE"
4762 #: build/C/man3/posix_memalign.3:58
4763 msgid "B<valloc>():"
4767 #: build/C/man3/posix_memalign.3:61
4769 msgid "Since glibc 2.12:"
4773 #: build/C/man3/posix_memalign.3:68
4777 " (_XOPEN_SOURCE\\ E<gt>=\\ 500 ||\n"
4778 " _XOPEN_SOURCE\\ &&\\ _XOPEN_SOURCE_EXTENDED) &&\n"
4779 " !(_POSIX_C_SOURCE\\ E<gt>=\\ 200112L || _XOPEN_SOURCE\\ E<gt>=\\ 600)\n"
4783 #: build/C/man3/posix_memalign.3:70
4785 msgid "Before glibc 2.12:"
4789 #: build/C/man3/posix_memalign.3:74
4791 "_BSD_SOURCE || _XOPEN_SOURCE\\ E<gt>=\\ 500 || _XOPEN_SOURCE\\ &&\\ "
4792 "_XOPEN_SOURCE_EXTENDED"
4796 #: build/C/man3/posix_memalign.3:81
4798 "(The (nonstandard) header file I<E<lt>malloc.hE<gt>> also exposes the "
4799 "declaration of B<valloc>(); no feature test macros are required.)"
4804 #: build/C/man3/posix_memalign.3:102
4806 "The function B<posix_memalign>() allocates I<size> bytes and places the "
4807 "address of the allocated memory in I<*memptr>. The address of the allocated "
4808 "memory will be a multiple of I<alignment>, which must be a power of two and "
4809 "a multiple of I<sizeof(void *)>. If I<size> is 0, then B<posix_memalign>() "
4810 "returns either NULL, or a unique pointer value that can later be "
4811 "successfully passed to B<free>(3)."
4814 #. The behavior of memalign() for size==0 is as for posix_memalign()
4815 #. but no standards govern this.
4817 #: build/C/man3/posix_memalign.3:113
4819 "The obsolete function B<memalign>() allocates I<size> bytes and returns a "
4820 "pointer to the allocated memory. The memory address will be a multiple of "
4821 "I<alignment>, which must be a power of two."
4825 #: build/C/man3/posix_memalign.3:122
4827 "The function B<aligned_alloc>() is the same as B<memalign>(), except for "
4828 "the added restriction that I<size> should be a multiple of I<alignment>."
4832 #: build/C/man3/posix_memalign.3:132
4834 "The obsolete function B<valloc>() allocates I<size> bytes and returns a "
4835 "pointer to the allocated memory. The memory address will be a multiple of "
4836 "the page size. It is equivalent to I<memalign(sysconf(_SC_PAGESIZE),size)>."
4840 #: build/C/man3/posix_memalign.3:139
4842 "The obsolete function B<pvalloc>() is similar to B<valloc>(), but rounds "
4843 "the size of the allocation up to the next multiple of the system page size."
4847 #: build/C/man3/posix_memalign.3:141
4848 msgid "For all of these functions, the memory is not zeroed."
4852 #: build/C/man3/posix_memalign.3:148
4854 "B<aligned_alloc>(), B<memalign>(), B<valloc>(), and B<pvalloc>() return a "
4855 "pointer to the allocated memory, or NULL if the request fails."
4859 #: build/C/man3/posix_memalign.3:155
4861 "B<posix_memalign>() returns zero on success, or one of the error values "
4862 "listed in the next section on failure. Note that I<errno> is not set."
4866 #: build/C/man3/posix_memalign.3:162
4868 "The I<alignment> argument was not a power of two, or was not a multiple of "
4869 "I<sizeof(void *)>."
4873 #: build/C/man3/posix_memalign.3:165
4874 msgid "There was insufficient memory to fulfill the allocation request."
4878 #: build/C/man3/posix_memalign.3:172
4880 "The functions B<memalign>(), B<valloc>(), and B<pvalloc>() have been "
4881 "available in all Linux libc libraries."
4885 #: build/C/man3/posix_memalign.3:176
4886 msgid "The function B<aligned_alloc>() was added to glibc in version 2.16."
4890 #: build/C/man3/posix_memalign.3:180
4891 msgid "The function B<posix_memalign>() is available since glibc 2.1.91."
4895 #: build/C/man3/posix_memalign.3:187
4897 "The function B<valloc>() appeared in 3.0BSD. It is documented as being "
4898 "obsolete in 4.3BSD, and as legacy in SUSv2. It does not appear in "
4903 #: build/C/man3/posix_memalign.3:191
4904 msgid "The function B<pvalloc>() is a GNU extension."
4908 #: build/C/man3/posix_memalign.3:195
4909 msgid "The function B<memalign>() appears in SunOS 4.1.3 but not in 4.4BSD."
4913 #: build/C/man3/posix_memalign.3:199
4914 msgid "The function B<posix_memalign>() comes from POSIX.1d."
4918 #: build/C/man3/posix_memalign.3:204
4919 msgid "The function I<aligned_alloc ()> is specified in the C11 standard."
4923 #: build/C/man3/posix_memalign.3:204
4929 #: build/C/man3/posix_memalign.3:208
4931 "Everybody agrees that B<posix_memalign>() is declared in "
4932 "I<E<lt>stdlib.hE<gt>>."
4936 #: build/C/man3/posix_memalign.3:212
4938 "On some systems B<memalign>() is declared in I<E<lt>stdlib.hE<gt>> instead "
4939 "of I<E<lt>malloc.hE<gt>>."
4943 #: build/C/man3/posix_memalign.3:219
4945 "According to SUSv2, B<valloc>() is declared in I<E<lt>stdlib.hE<gt>>. "
4946 "Libc4,5 and glibc declare it in I<E<lt>malloc.hE<gt>>, and also in "
4947 "I<E<lt>stdlib.hE<gt>> if suitable feature test macros are defined (see "
4952 #: build/C/man3/posix_memalign.3:228
4954 "On many systems there are alignment restrictions, for example, on buffers "
4955 "used for direct block device I/O. POSIX specifies the "
4956 "I<pathconf(path,_PC_REC_XFER_ALIGN)> call that tells what alignment is "
4957 "needed. Now one can use B<posix_memalign>() to satisfy this requirement."
4961 #: build/C/man3/posix_memalign.3:237
4963 "B<posix_memalign>() verifies that I<alignment> matches the requirements "
4964 "detailed above. B<memalign>() may not check that the I<alignment> argument "
4968 #. Other systems allow passing the result of
4975 #: build/C/man3/posix_memalign.3:265
4977 "POSIX requires that memory obtained from B<posix_memalign>() can be freed "
4978 "using B<free>(3). Some systems provide no way to reclaim memory allocated "
4979 "with B<memalign>() or B<valloc>() (because one can only pass to B<free>(3) "
4980 "a pointer gotten from B<malloc>(3), while, for example, B<memalign>() would "
4981 "call B<malloc>(3) and then align the obtained value). The glibc "
4982 "implementation allows memory obtained from any of these these functions to "
4983 "be reclaimed with B<free>(3)."
4987 #: build/C/man3/posix_memalign.3:270
4989 "The glibc B<malloc>(3) always returns 8-byte aligned memory addresses, so "
4990 "these functions are only needed if you require larger alignment values."
4994 #: build/C/man3/posix_memalign.3:275
4995 msgid "B<brk>(2), B<getpagesize>(2), B<free>(3), B<malloc>(3)"
4999 #: build/C/man2/readahead.2:28
5005 #: build/C/man2/readahead.2:28 build/C/man7/shm_overview.7:27
5011 #: build/C/man2/readahead.2:31
5012 msgid "readahead - perform file readahead into page cache"
5016 #: build/C/man2/readahead.2:37
5019 "B<ssize_t readahead(int >I<fd>B<, off64_t >I<offset>B<, size_t "
5024 #: build/C/man2/readahead.2:65
5026 "B<readahead>() populates the page cache with data from a file so that "
5027 "subsequent reads from that file will not block on disk I/O. The I<fd> "
5028 "argument is a file descriptor identifying the file which is to be read. The "
5029 "I<offset> argument specifies the starting point from which data is to be "
5030 "read and I<count> specifies the number of bytes to be read. I/O is "
5031 "performed in whole pages, so that I<offset> is effectively rounded down to a "
5032 "page boundary and bytes are read up to the next page boundary greater than "
5033 "or equal to I<(offset+count)>. B<readahead>() does not read beyond the end "
5034 "of the file. B<readahead>() blocks until the specified data has been "
5035 "read. The current file offset of the open file referred to by I<fd> is left "
5040 #: build/C/man2/readahead.2:71
5042 "On success, B<readahead>() returns 0; on failure, -1 is returned, with "
5043 "I<errno> set to indicate the cause of the error."
5047 #: build/C/man2/readahead.2:76
5048 msgid "I<fd> is not a valid file descriptor or is not open for reading."
5052 #: build/C/man2/readahead.2:82
5053 msgid "I<fd> does not refer to a file type to which B<readahead>() can be applied."
5057 #: build/C/man2/readahead.2:87
5059 "The B<readahead>() system call appeared in Linux 2.4.13; glibc support has "
5060 "been provided since version 2.3."
5064 #: build/C/man2/readahead.2:92
5066 "The B<readahead>() system call is Linux-specific, and its use should be "
5067 "avoided in portable applications."
5071 #: build/C/man2/readahead.2:98
5072 msgid "B<lseek>(2), B<madvise>(2), B<mmap>(2), B<posix_fadvise>(2), B<read>(2)"
5076 #: build/C/man2/remap_file_pages.2:26
5078 msgid "REMAP_FILE_PAGES"
5082 #: build/C/man2/remap_file_pages.2:29
5083 msgid "remap_file_pages - create a nonlinear file mapping"
5087 #: build/C/man2/remap_file_pages.2:33
5090 "B<#define _GNU_SOURCE> /* See feature_test_macros(7) */\n"
5091 "B<#include E<lt>sys/mman.hE<gt>>\n"
5095 #: build/C/man2/remap_file_pages.2:36
5098 "B<int remap_file_pages(void *>I<addr>B<, size_t >I<size>B<, int "
5100 "B< ssize_t >I<pgoff>B<, int >I<flags>B<);>\n"
5104 #: build/C/man2/remap_file_pages.2:49
5106 "The B<remap_file_pages>() system call is used to create a nonlinear "
5107 "mapping, that is, a mapping in which the pages of the file are mapped into a "
5108 "nonsequential order in memory. The advantage of using B<remap_file_pages>() "
5109 "over using repeated calls to B<mmap>(2) is that the former approach does "
5110 "not require the kernel to create additional VMA (Virtual Memory Area) data "
5115 #: build/C/man2/remap_file_pages.2:51
5116 msgid "To create a nonlinear mapping we perform the following steps:"
5120 #: build/C/man2/remap_file_pages.2:51
5126 #: build/C/man2/remap_file_pages.2:59
5128 "Use B<mmap>(2) to create a mapping (which is initially linear). This "
5129 "mapping must be created with the B<MAP_SHARED> flag."
5133 #: build/C/man2/remap_file_pages.2:59
5139 #: build/C/man2/remap_file_pages.2:67
5141 "Use one or more calls to B<remap_file_pages>() to rearrange the "
5142 "correspondence between the pages of the mapping and the pages of the file. "
5143 "It is possible to map the same page of a file into multiple locations within "
5144 "the mapped region."
5148 #: build/C/man2/remap_file_pages.2:78
5150 "The I<pgoff> and I<size> arguments specify the region of the file that is to "
5151 "be relocated within the mapping: I<pgoff> is a file offset in units of the "
5152 "system page size; I<size> is the length of the region in bytes."
5156 #: build/C/man2/remap_file_pages.2:96
5158 "The I<addr> argument serves two purposes. First, it identifies the mapping "
5159 "whose pages we want to rearrange. Thus, I<addr> must be an address that "
5160 "falls within a region previously mapped by a call to B<mmap>(2). Second, "
5161 "I<addr> specifies the address at which the file pages identified by I<pgoff> "
5162 "and I<size> will be placed."
5165 #. This rounding is weird, and not consistent with the treatment of
5166 #. the analogous arguments for munmap()/mprotect() and for mlock().
5169 #: build/C/man2/remap_file_pages.2:110
5171 "The values specified in I<addr> and I<size> should be multiples of the "
5172 "system page size. If they are not, then the kernel rounds I<both> values "
5173 "I<down> to the nearest multiple of the page size."
5177 #: build/C/man2/remap_file_pages.2:114
5178 msgid "The I<prot> argument must be specified as 0."
5182 #: build/C/man2/remap_file_pages.2:122
5184 "The I<flags> argument has the same meaning as for B<mmap>(2), but all flags "
5185 "other than B<MAP_NONBLOCK> are ignored."
5189 #: build/C/man2/remap_file_pages.2:129
5191 "On success, B<remap_file_pages>() returns 0. On error, -1 is returned, and "
5192 "I<errno> is set appropriately."
5196 #: build/C/man2/remap_file_pages.2:137
5198 "I<addr> does not refer to a valid mapping created with the B<MAP_SHARED> "
5202 #. And possibly others from vma->vm_ops->populate()
5204 #: build/C/man2/remap_file_pages.2:146
5205 msgid "I<addr>, I<size>, I<prot>, or I<pgoff> is invalid."
5209 #: build/C/man2/remap_file_pages.2:151
5211 "The B<remap_file_pages>() system call appeared in Linux 2.5.46; glibc "
5212 "support was added in version 2.3.3."
5216 #: build/C/man2/remap_file_pages.2:155
5217 msgid "The B<remap_file_pages>() system call is Linux-specific."
5221 #: build/C/man2/remap_file_pages.2:162
5223 "B<getpagesize>(2), B<mmap>(2), B<mmap2>(2), B<mprotect>(2), B<mremap>(2), "
5228 #: build/C/man3/shm_open.3:26
5234 #: build/C/man3/shm_open.3:26
5240 #: build/C/man3/shm_open.3:29
5241 msgid "shm_open, shm_unlink - Create/open or unlink POSIX shared memory objects"
5245 #: build/C/man3/shm_open.3:33
5246 msgid "B<#include E<lt>sys/stat.hE<gt>> /* For mode constants */"
5250 #: build/C/man3/shm_open.3:35
5251 msgid "B<#include E<lt>fcntl.hE<gt>> /* For O_* constants */"
5255 #: build/C/man3/shm_open.3:37
5256 msgid "B<int shm_open(const char *>I<name>B<, int >I<oflag>B<, mode_t >I<mode>B<);>"
5260 #: build/C/man3/shm_open.3:39
5261 msgid "B<int shm_unlink(const char *>I<name>B<);>"
5265 #: build/C/man3/shm_open.3:41
5266 msgid "Link with I<-lrt>."
5270 #: build/C/man3/shm_open.3:53
5272 "B<shm_open>() creates and opens a new, or opens an existing, POSIX shared "
5273 "memory object. A POSIX shared memory object is in effect a handle which can "
5274 "be used by unrelated processes to B<mmap>(2) the same region of shared "
5275 "memory. The B<shm_unlink>() function performs the converse operation, "
5276 "removing an object previously created by B<shm_open>()."
5279 #. glibc allows the initial slash to be omitted, and makes
5280 #. multiple initial slashes equivalent to a single slash.
5281 #. This differs from the implementation of POSIX message queues.
5282 #. glibc allows subdirectory components in the name, in which
5283 #. case the subdirectory must exist under /dev/shm, and allow the
5284 #. required permissions if a user wants to create a shared memory
5285 #. object in that subdirectory.
5287 #: build/C/man3/shm_open.3:74
5289 "The operation of B<shm_open>() is analogous to that of B<open>(2). I<name> "
5290 "specifies the shared memory object to be created or opened. For portable "
5291 "use, a shared memory object should be identified by a name of the form "
5292 "I</somename>; that is, a null-terminated string of up to B<NAME_MAX> (i.e., "
5293 "255) characters consisting of an initial slash, followed by one or more "
5294 "characters, none of which are slashes."
5298 #: build/C/man3/shm_open.3:81
5300 "I<oflag> is a bit mask created by ORing together exactly one of B<O_RDONLY> "
5301 "or B<O_RDWR> and any of the other flags listed here:"
5305 #: build/C/man3/shm_open.3:81
5311 #: build/C/man3/shm_open.3:87
5313 "Open the object for read access. A shared memory object opened in this way "
5314 "can only be B<mmap>(2)ed for read (B<PROT_READ>) access."
5318 #: build/C/man3/shm_open.3:87
5324 #: build/C/man3/shm_open.3:90
5325 msgid "Open the object for read-write access."
5329 #: build/C/man3/shm_open.3:90
5334 #. In truth it is actually the file system IDs on Linux, but these
5335 #. are nearly always the same as the effective IDs. (MTK, Jul 05)
5337 #: build/C/man3/shm_open.3:110
5339 "Create the shared memory object if it does not exist. The user and group "
5340 "ownership of the object are taken from the corresponding effective IDs of "
5341 "the calling process, and the object's permission bits are set according to "
5342 "the low-order 9 bits of I<mode>, except that those bits set in the process "
5343 "file mode creation mask (see B<umask>(2)) are cleared for the new object. "
5344 "A set of macro constants which can be used to define I<mode> is listed in "
5345 "B<open>(2). (Symbolic definitions of these constants can be obtained by "
5346 "including I<E<lt>sys/stat.hE<gt>>.)"
5350 #: build/C/man3/shm_open.3:116
5352 "A new shared memory object initially has zero length\\(emthe size of the "
5353 "object can be set using B<ftruncate>(2). The newly allocated bytes of a "
5354 "shared memory object are automatically initialized to 0."
5358 #: build/C/man3/shm_open.3:116
5364 #: build/C/man3/shm_open.3:125
5366 "If B<O_CREAT> was also specified, and a shared memory object with the given "
5367 "I<name> already exists, return an error. The check for the existence of the "
5368 "object, and its creation if it does not exist, are performed atomically."
5372 #: build/C/man3/shm_open.3:125
5378 #: build/C/man3/shm_open.3:128
5379 msgid "If the shared memory object already exists, truncate it to zero bytes."
5383 #: build/C/man3/shm_open.3:131
5385 "Definitions of these flag values can be obtained by including "
5386 "I<E<lt>fcntl.hE<gt>>."
5390 #: build/C/man3/shm_open.3:142
5392 "On successful completion B<shm_open>() returns a new file descriptor "
5393 "referring to the shared memory object. This file descriptor is guaranteed "
5394 "to be the lowest-numbered file descriptor not previously opened within the "
5395 "process. The B<FD_CLOEXEC> flag (see B<fcntl>(2)) is set for the file "
5400 #: build/C/man3/shm_open.3:151
5402 "The file descriptor is normally used in subsequent calls to B<ftruncate>(2) "
5403 "(for a newly created object) and B<mmap>(2). After a call to B<mmap>(2) "
5404 "the file descriptor may be closed without affecting the memory mapping."
5408 #: build/C/man3/shm_open.3:169
5410 "The operation of B<shm_unlink>() is analogous to B<unlink>(2): it removes a "
5411 "shared memory object name, and, once all processes have unmapped the object, "
5412 "de-allocates and destroys the contents of the associated memory region. "
5413 "After a successful B<shm_unlink>(), attempts to B<shm_open>() an object "
5414 "with the same I<name> will fail (unless B<O_CREAT> was specified, in which "
5415 "case a new, distinct object is created)."
5419 #: build/C/man3/shm_open.3:178
5421 "On success, B<shm_open>() returns a nonnegative file descriptor. On "
5422 "failure, B<shm_open>() returns -1. B<shm_unlink>() returns 0 on success, "
5427 #: build/C/man3/shm_open.3:185
5429 "On failure, I<errno> is set to indicate the cause of the error. Values "
5430 "which may appear in I<errno> include the following:"
5434 #: build/C/man3/shm_open.3:190
5435 msgid "Permission to B<shm_unlink>() the shared memory object was denied."
5439 #: build/C/man3/shm_open.3:200
5441 "Permission was denied to B<shm_open>() I<name> in the specified I<mode>, or "
5442 "B<O_TRUNC> was specified and the caller does not have write permission on "
5447 #: build/C/man3/shm_open.3:200 build/C/man2/shmget.2:192
5453 #: build/C/man3/shm_open.3:211
5455 "Both B<O_CREAT> and B<O_EXCL> were specified to B<shm_open>() and the "
5456 "shared memory object specified by I<name> already exists."
5460 #: build/C/man3/shm_open.3:218
5461 msgid "The I<name> argument to B<shm_open>() was invalid."
5465 #: build/C/man3/shm_open.3:218
5471 #: build/C/man3/shm_open.3:221
5472 msgid "The process already has the maximum number of files open."
5476 #: build/C/man3/shm_open.3:221
5478 msgid "B<ENAMETOOLONG>"
5482 #: build/C/man3/shm_open.3:227
5483 msgid "The length of I<name> exceeds B<PATH_MAX>."
5487 #: build/C/man3/shm_open.3:231
5488 msgid "The limit on the total number of files open on the system has been reached."
5492 #: build/C/man3/shm_open.3:231 build/C/man3/shm_open.3:240 build/C/man2/shmget.2:206
5498 #: build/C/man3/shm_open.3:240
5500 "An attempt was made to B<shm_open>() a I<name> that did not exist, and "
5501 "B<O_CREAT> was not specified."
5505 #: build/C/man3/shm_open.3:247
5506 msgid "An attempt was to made to B<shm_unlink>() a I<name> that does not exist."
5510 #: build/C/man3/shm_open.3:249
5511 msgid "These functions are provided in glibc 2.2 and later."
5515 #: build/C/man3/shm_open.3:255
5517 "POSIX.1-2001 says that the group ownership of a newly created shared memory "
5518 "object is set to either the calling process's effective group ID or \"a "
5519 "system default group ID\"."
5523 #: build/C/man3/shm_open.3:264
5525 "POSIX leaves the behavior of the combination of B<O_RDONLY> and B<O_TRUNC> "
5526 "unspecified. On Linux, this will successfully truncate an existing shared "
5527 "memory object\\(emthis may not be so on other UNIX systems."
5531 #: build/C/man3/shm_open.3:269
5533 "The POSIX shared memory object implementation on Linux 2.4 makes use of a "
5534 "dedicated file system, which is normally mounted under I</dev/shm>."
5538 #: build/C/man3/shm_open.3:280
5540 "B<close>(2), B<fchmod>(2), B<fchown>(2), B<fcntl>(2), B<fstat>(2), "
5541 "B<ftruncate>(2), B<mmap>(2), B<open>(2), B<umask>(2), B<shm_overview>(7)"
5545 #: build/C/man7/shm_overview.7:27
5547 msgid "SHM_OVERVIEW"
5551 #: build/C/man7/shm_overview.7:30
5552 msgid "shm_overview - Overview of POSIX shared memory"
5556 #: build/C/man7/shm_overview.7:33
5558 "The POSIX shared memory API allows processes to communicate information by "
5559 "sharing a region of memory."
5563 #: build/C/man7/shm_overview.7:35
5564 msgid "The interfaces employed in the API are:"
5568 #: build/C/man7/shm_overview.7:35
5570 msgid "B<shm_open>(3)"
5574 #: build/C/man7/shm_overview.7:42
5576 "Create and open a new object, or open an existing object. This is analogous "
5577 "to B<open>(2). The call returns a file descriptor for use by the other "
5578 "interfaces listed below."
5582 #: build/C/man7/shm_overview.7:42
5584 msgid "B<ftruncate>(2)"
5588 #: build/C/man7/shm_overview.7:46
5590 "Set the size of the shared memory object. (A newly created shared memory "
5591 "object has a length of zero.)"
5595 #: build/C/man7/shm_overview.7:50
5597 "Map the shared memory object into the virtual address space of the calling "
5602 #: build/C/man7/shm_overview.7:50
5604 msgid "B<munmap>(2)"
5608 #: build/C/man7/shm_overview.7:54
5610 "Unmap the shared memory object from the virtual address space of the calling "
5615 #: build/C/man7/shm_overview.7:54
5617 msgid "B<shm_unlink>(3)"
5621 #: build/C/man7/shm_overview.7:57
5622 msgid "Remove a shared memory object name."
5626 #: build/C/man7/shm_overview.7:57
5632 #: build/C/man7/shm_overview.7:62
5634 "Close the file descriptor allocated by B<shm_open>(3) when it is no longer "
5639 #: build/C/man7/shm_overview.7:62
5645 #: build/C/man7/shm_overview.7:76
5647 "Obtain a I<stat> structure that describes the shared memory object. Among "
5648 "the information returned by this call are the object's size (I<st_size>), "
5649 "permissions (I<st_mode>), owner (I<st_uid>), and group (I<st_gid>)."
5653 #: build/C/man7/shm_overview.7:76
5655 msgid "B<fchown>(2)"
5659 #: build/C/man7/shm_overview.7:79
5660 msgid "To change the ownership of a shared memory object."
5664 #: build/C/man7/shm_overview.7:79
5666 msgid "B<fchmod>(2)"
5670 #: build/C/man7/shm_overview.7:82
5671 msgid "To change the permissions of a shared memory object."
5675 #: build/C/man7/shm_overview.7:82
5681 #: build/C/man7/shm_overview.7:84
5682 msgid "POSIX shared memory is supported since Linux 2.4 and glibc 2.2."
5686 #: build/C/man7/shm_overview.7:84
5692 #: build/C/man7/shm_overview.7:89
5694 "POSIX shared memory objects have kernel persistence: a shared memory object "
5695 "will exist until the system is shut down, or until all processes have "
5696 "unmapped the object and it has been deleted with B<shm_unlink>(3)"
5700 #: build/C/man7/shm_overview.7:89
5706 #: build/C/man7/shm_overview.7:94
5708 "Programs using the POSIX shared memory API must be compiled with I<cc -lrt> "
5709 "to link against the real-time library, I<librt>."
5713 #: build/C/man7/shm_overview.7:94
5715 msgid "Accessing shared memory objects via the file system"
5719 #: build/C/man7/shm_overview.7:101
5721 "On Linux, shared memory objects are created in a (I<tmpfs>) virtual file "
5722 "system, normally mounted under I</dev/shm>. Since kernel 2.6.19, Linux "
5723 "supports the use of access control lists (ACLs) to control the permissions "
5724 "of objects in the virtual file system."
5728 #: build/C/man7/shm_overview.7:106
5730 "Typically, processes must synchronize their access to a shared memory "
5731 "object, using, for example, POSIX semaphores."
5735 #: build/C/man7/shm_overview.7:114
5737 "System V shared memory (B<shmget>(2), B<shmop>(2), etc.) is an older shared "
5738 "memory API. POSIX shared memory provides a simpler, and better designed "
5739 "interface; on the other hand POSIX shared memory is somewhat less widely "
5740 "available (especially on older systems) than System V shared memory."
5744 #: build/C/man7/shm_overview.7:127
5746 "B<fchmod>(2), B<fchown>(2), B<fstat>(2), B<ftruncate>(2), B<mmap>(2), "
5747 "B<mprotect>(2), B<munmap>(2), B<shmget>(2), B<shmop>(2), B<shm_open>(3), "
5748 "B<shm_unlink>(3), B<sem_overview>(7)"
5752 #: build/C/man2/shmctl.2:43
5758 #: build/C/man2/shmctl.2:43
5764 #: build/C/man2/shmctl.2:46
5765 msgid "shmctl - shared memory control"
5769 #: build/C/man2/shmctl.2:49 build/C/man2/shmget.2:42
5770 msgid "B<#include E<lt>sys/ipc.hE<gt>>"
5774 #: build/C/man2/shmctl.2:51 build/C/man2/shmget.2:44
5775 msgid "B<#include E<lt>sys/shm.hE<gt>>"
5779 #: build/C/man2/shmctl.2:53
5780 msgid "B<int shmctl(int >I<shmid>B<, int >I<cmd>B<, struct shmid_ds *>I<buf>B<);>"
5784 #: build/C/man2/shmctl.2:60
5786 "B<shmctl>() performs the control operation specified by I<cmd> on the "
5787 "shared memory segment whose identifier is given in I<shmid>."
5791 #: build/C/man2/shmctl.2:65
5793 "The I<buf> argument is a pointer to a I<shmid_ds> structure, defined in "
5794 "I<E<lt>sys/shm.hE<gt>> as follows:"
5798 #: build/C/man2/shmctl.2:79
5801 "struct shmid_ds {\n"
5802 " struct ipc_perm shm_perm; /* Ownership and permissions */\n"
5803 " size_t shm_segsz; /* Size of segment (bytes) */\n"
5804 " time_t shm_atime; /* Last attach time */\n"
5805 " time_t shm_dtime; /* Last detach time */\n"
5806 " time_t shm_ctime; /* Last change time */\n"
5807 " pid_t shm_cpid; /* PID of creator */\n"
5808 " pid_t shm_lpid; /* PID of last shmat(2)/shmdt(2) */\n"
5809 " shmatt_t shm_nattch; /* No. of current attaches */\n"
5815 #: build/C/man2/shmctl.2:87
5817 "The I<ipc_perm> structure is defined in I<E<lt>sys/ipc.hE<gt>> as follows "
5818 "(the highlighted fields are settable using B<IPC_SET>):"
5822 #: build/C/man2/shmctl.2:100
5825 "struct ipc_perm {\n"
5826 " key_t __key; /* Key supplied to shmget(2) */\n"
5827 " uid_t B<uid>; /* Effective UID of owner */\n"
5828 " gid_t B<gid>; /* Effective GID of owner */\n"
5829 " uid_t cuid; /* Effective UID of creator */\n"
5830 " gid_t cgid; /* Effective GID of creator */\n"
5831 " unsigned short B<mode>; /* B<Permissions> + SHM_DEST and\n"
5832 " SHM_LOCKED flags */\n"
5833 " unsigned short __seq; /* Sequence number */\n"
5838 #: build/C/man2/shmctl.2:106
5839 msgid "Valid values for I<cmd> are:"
5843 #: build/C/man2/shmctl.2:107
5849 #: build/C/man2/shmctl.2:116
5851 "Copy information from the kernel data structure associated with I<shmid> "
5852 "into the I<shmid_ds> structure pointed to by I<buf>. The caller must have "
5853 "read permission on the shared memory segment."
5857 #: build/C/man2/shmctl.2:116
5863 #: build/C/man2/shmctl.2:134
5865 "Write the values of some members of the I<shmid_ds> structure pointed to by "
5866 "I<buf> to the kernel data structure associated with this shared memory "
5867 "segment, updating also its I<shm_ctime> member. The following fields can be "
5868 "changed: I<shm_perm.uid>, I<shm_perm.gid>, and (the least significant 9 bits "
5869 "of) I<shm_perm.mode>. The effective UID of the calling process must match "
5870 "the owner (I<shm_perm.uid>) or creator (I<shm_perm.cuid>) of the shared "
5871 "memory segment, or the caller must be privileged."
5875 #: build/C/man2/shmctl.2:134
5881 #: build/C/man2/shmctl.2:151
5883 "Mark the segment to be destroyed. The segment will only actually be "
5884 "destroyed after the last process detaches it (i.e., when the I<shm_nattch> "
5885 "member of the associated structure I<shmid_ds> is zero). The caller must be "
5886 "the owner or creator, or be privileged. If a segment has been marked for "
5887 "destruction, then the (nonstandard) B<SHM_DEST> flag of the "
5888 "I<shm_perm.mode> field in the associated data structure retrieved by "
5889 "B<IPC_STAT> will be set."
5893 #: build/C/man2/shmctl.2:154
5895 "The caller I<must> ensure that a segment is eventually destroyed; otherwise "
5896 "its pages that were faulted in will remain in memory or swap."
5900 #: build/C/man2/shmctl.2:154
5902 msgid "B<IPC_INFO> (Linux-specific)"
5906 #: build/C/man2/shmctl.2:167
5908 "Returns information about system-wide shared memory limits and parameters in "
5909 "the structure pointed to by I<buf>. This structure is of type I<shminfo> "
5910 "(thus, a cast is required), defined in I<E<lt>sys/shm.hE<gt>> if the "
5911 "B<_GNU_SOURCE> feature test macro is defined:"
5915 #: build/C/man2/shmctl.2:181
5918 "struct shminfo {\n"
5919 " unsigned long shmmax; /* Maximum segment size */\n"
5920 " unsigned long shmmin; /* Minimum segment size;\n"
5922 " unsigned long shmmni; /* Maximum number of segments */\n"
5923 " unsigned long shmseg; /* Maximum number of segments\n"
5924 " that a process can attach;\n"
5925 " unused within kernel */\n"
5926 " unsigned long shmall; /* Maximum number of pages of\n"
5927 " shared memory, system-wide */\n"
5932 #: build/C/man2/shmctl.2:194
5934 "The I<shmmni>, I<shmmax>, and I<shmall> settings can be changed via I</proc> "
5935 "files of the same name; see B<proc>(5) for details."
5939 #: build/C/man2/shmctl.2:194
5941 msgid "B<SHM_INFO> (Linux-specific)"
5945 #: build/C/man2/shmctl.2:205
5947 "Returns a I<shm_info> structure whose fields contain information about "
5948 "system resources consumed by shared memory. This structure is defined in "
5949 "I<E<lt>sys/shm.hE<gt>> if the B<_GNU_SOURCE> feature test macro is defined:"
5953 #: build/C/man2/shmctl.2:222
5956 "struct shm_info {\n"
5957 " int used_ids; /* # of currently existing\n"
5959 " unsigned long shm_tot; /* Total number of shared\n"
5960 " memory pages */\n"
5961 " unsigned long shm_rss; /* # of resident shared\n"
5962 " memory pages */\n"
5963 " unsigned long shm_swp; /* # of swapped shared\n"
5964 " memory pages */\n"
5965 " unsigned long swap_attempts;\n"
5966 " /* Unused since Linux 2.4 */\n"
5967 " unsigned long swap_successes;\n"
5968 " /* Unused since Linux 2.4 */\n"
5973 #: build/C/man2/shmctl.2:224
5975 msgid "B<SHM_STAT> (Linux-specific)"
5979 #: build/C/man2/shmctl.2:235
5981 "Returns a I<shmid_ds> structure as for B<IPC_STAT>. However, the I<shmid> "
5982 "argument is not a segment identifier, but instead an index into the kernel's "
5983 "internal array that maintains information about all shared memory segments "
5988 #: build/C/man2/shmctl.2:238
5990 "The caller can prevent or allow swapping of a shared memory segment with the "
5991 "following I<cmd> values:"
5995 #: build/C/man2/shmctl.2:239
5997 msgid "B<SHM_LOCK> (Linux-specific)"
6001 #: build/C/man2/shmctl.2:251
6003 "Prevent swapping of the shared memory segment. The caller must fault in any "
6004 "pages that are required to be present after locking is enabled. If a "
6005 "segment has been locked, then the (nonstandard) B<SHM_LOCKED> flag of the "
6006 "I<shm_perm.mode> field in the associated data structure retrieved by "
6007 "B<IPC_STAT> will be set."
6011 #: build/C/man2/shmctl.2:251
6013 msgid "B<SHM_UNLOCK> (Linux-specific)"
6017 #: build/C/man2/shmctl.2:254
6018 msgid "Unlock the segment, allowing it to be swapped out."
6021 #. There was some weirdness in 2.6.9: SHM_LOCK and SHM_UNLOCK could
6022 #. be applied to a segment, regardless of ownership of the segment.
6023 #. This was a botch-up in the move to RLIMIT_MEMLOCK, and was fixed
6024 #. in 2.6.10. MTK, May 2005
6026 #: build/C/man2/shmctl.2:272
6028 "In kernels before 2.6.10, only a privileged process could employ B<SHM_LOCK> "
6029 "and B<SHM_UNLOCK>. Since kernel 2.6.10, an unprivileged process can employ "
6030 "these operations if its effective UID matches the owner or creator UID of "
6031 "the segment, and (for B<SHM_LOCK>) the amount of memory to be locked falls "
6032 "within the B<RLIMIT_MEMLOCK> resource limit (see B<setrlimit>(2))."
6036 #: build/C/man2/shmctl.2:290
6038 "A successful B<IPC_INFO> or B<SHM_INFO> operation returns the index of the "
6039 "highest used entry in the kernel's internal array recording information "
6040 "about all shared memory segments. (This information can be used with "
6041 "repeated B<SHM_STAT> operations to obtain information about all shared "
6042 "memory segments on the system.) A successful B<SHM_STAT> operation returns "
6043 "the identifier of the shared memory segment whose index was given in "
6044 "I<shmid>. Other operations return 0 on success."
6048 #: build/C/man2/shmctl.2:294
6049 msgid "On error, -1 is returned, and I<errno> is set appropriately."
6053 #: build/C/man2/shmctl.2:303
6055 "B<IPC_STAT> or B<SHM_STAT> is requested and I<shm_perm.mode> does not allow "
6056 "read access for I<shmid>, and the calling process does not have the "
6057 "B<CAP_IPC_OWNER> capability."
6061 #: build/C/man2/shmctl.2:314
6063 "The argument I<cmd> has value B<IPC_SET> or B<IPC_STAT> but the address "
6064 "pointed to by I<buf> isn't accessible."
6068 #: build/C/man2/shmctl.2:314
6074 #: build/C/man2/shmctl.2:317
6075 msgid "I<shmid> points to a removed identifier."
6079 #: build/C/man2/shmctl.2:326
6081 "I<shmid> is not a valid identifier, or I<cmd> is not a valid command. Or: "
6082 "for a B<SHM_STAT> operation, the index value specified in I<shmid> referred "
6083 "to an array slot that is currently unused."
6087 #: build/C/man2/shmctl.2:337
6089 "(In kernels since 2.6.9), B<SHM_LOCK> was specified and the size of the "
6090 "to-be-locked segment would mean that the total bytes in locked shared memory "
6091 "segments would exceed the limit for the real user ID of the calling "
6092 "process. This limit is defined by the B<RLIMIT_MEMLOCK> soft resource limit "
6093 "(see B<setrlimit>(2))."
6097 #: build/C/man2/shmctl.2:337
6099 msgid "B<EOVERFLOW>"
6103 #: build/C/man2/shmctl.2:342
6105 "B<IPC_STAT> is attempted, and the GID or UID value is too large to be stored "
6106 "in the structure pointed to by I<buf>."
6110 #: build/C/man2/shmctl.2:354
6112 "B<IPC_SET> or B<IPC_RMID> is attempted, and the effective user ID of the "
6113 "calling process is not that of the creator (found in I<shm_perm.cuid>), or "
6114 "the owner (found in I<shm_perm.uid>), and the process was not privileged "
6115 "(Linux: did not have the B<CAP_SYS_ADMIN> capability)."
6119 #: build/C/man2/shmctl.2:366
6121 "Or (in kernels before 2.6.9), B<SHM_LOCK> or B<SHM_UNLOCK> was specified, "
6122 "but the process was not privileged (Linux: did not have the B<CAP_IPC_LOCK> "
6123 "capability). (Since Linux 2.6.9, this error can also occur if the "
6124 "B<RLIMIT_MEMLOCK> is 0 and the caller is not privileged.)"
6127 #. SVr4 documents an additional error condition EMFILE.
6129 #: build/C/man2/shmctl.2:371 build/C/man2/shmget.2:232 build/C/man2/shmop.2:229
6130 msgid "SVr4, POSIX.1-2001."
6134 #: build/C/man2/shmctl.2:382
6136 "The B<IPC_INFO>, B<SHM_STAT> and B<SHM_INFO> operations are used by the "
6137 "B<ipcs>(1) program to provide information on allocated resources. In the "
6138 "future these may modified or moved to a /proc file system interface."
6142 #: build/C/man2/shmctl.2:390
6144 "Linux permits a process to attach (B<shmat>(2)) a shared memory segment "
6145 "that has already been marked for deletion using I<shmctl(IPC_RMID)>. This "
6146 "feature is not available on other UNIX implementations; portable "
6147 "applications should avoid relying on it."
6151 #: build/C/man2/shmctl.2:403
6153 "Various fields in a I<struct shmid_ds> were typed as I<short> under Linux "
6154 "2.2 and have become I<long> under Linux 2.4. To take advantage of this, a "
6155 "recompilation under glibc-2.1.91 or later should suffice. (The kernel "
6156 "distinguishes old and new calls by an B<IPC_64> flag in I<cmd>.)"
6160 #: build/C/man2/shmctl.2:410
6162 "B<mlock>(2), B<setrlimit>(2), B<shmget>(2), B<shmop>(2), B<capabilities>(7), "
6167 #: build/C/man2/shmget.2:36
6173 #: build/C/man2/shmget.2:36
6179 #: build/C/man2/shmget.2:39
6180 msgid "shmget - allocates a shared memory segment"
6184 #: build/C/man2/shmget.2:46
6185 msgid "B<int shmget(key_t >I<key>B<, size_t >I<size>B<, int >I<shmflg>B<);>"
6189 #: build/C/man2/shmget.2:70
6191 "B<shmget>() returns the identifier of the shared memory segment associated "
6192 "with the value of the argument I<key>. A new shared memory segment, with "
6193 "size equal to the value of I<size> rounded up to a multiple of B<PAGE_SIZE>, "
6194 "is created if I<key> has the value B<IPC_PRIVATE> or I<key> isn't "
6195 "B<IPC_PRIVATE>, no shared memory segment corresponding to I<key> exists, and "
6196 "B<IPC_CREAT> is specified in I<shmflg>."
6200 #: build/C/man2/shmget.2:89
6202 "If I<shmflg> specifies both B<IPC_CREAT> and B<IPC_EXCL> and a shared memory "
6203 "segment already exists for I<key>, then B<shmget>() fails with I<errno> set "
6204 "to B<EEXIST>. (This is analogous to the effect of the combination B<O_CREAT "
6205 "| O_EXCL> for B<open>(2).)"
6209 #: build/C/man2/shmget.2:93
6210 msgid "The value I<shmflg> is composed of:"
6214 #: build/C/man2/shmget.2:93
6216 msgid "B<IPC_CREAT>"
6220 #: build/C/man2/shmget.2:100
6222 "to create a new segment. If this flag is not used, then B<shmget>() will "
6223 "find the segment associated with I<key> and check to see if the user has "
6224 "permission to access the segment."
6228 #: build/C/man2/shmget.2:100
6234 #: build/C/man2/shmget.2:103
6235 msgid "used with B<IPC_CREAT> to ensure failure if the segment already exists."
6239 #: build/C/man2/shmget.2:103
6241 msgid "I<mode_flags>"
6245 #: build/C/man2/shmget.2:113
6247 "(least significant 9 bits) specifying the permissions granted to the owner, "
6248 "group, and world. These bits have the same format, and the same meaning, as "
6249 "the I<mode> argument of B<open>(2). Presently, the execute permissions are "
6250 "not used by the system."
6254 #: build/C/man2/shmget.2:113
6256 msgid "B<SHM_HUGETLB> (since Linux 2.6)"
6260 #: build/C/man2/shmget.2:119
6262 "Allocate the segment using \"huge pages.\" See the kernel source file "
6263 "I<Documentation/vm/hugetlbpage.txt> for further information."
6267 #: build/C/man2/shmget.2:119
6269 msgid "B<SHM_NORESERVE> (since Linux 2.6.15)"
6272 #. As at 2.6.17-rc2, this flag has no effect if SHM_HUGETLB was also
6275 #: build/C/man2/shmget.2:138
6277 "This flag serves the same purpose as the B<mmap>(2) B<MAP_NORESERVE> flag. "
6278 "Do not reserve swap space for this segment. When swap space is reserved, "
6279 "one has the guarantee that it is possible to modify the segment. When swap "
6280 "space is not reserved one might get B<SIGSEGV> upon a write if no physical "
6281 "memory is available. See also the discussion of the file "
6282 "I</proc/sys/vm/overcommit_memory> in B<proc>(5)."
6286 #: build/C/man2/shmget.2:146
6288 "When a new shared memory segment is created, its contents are initialized to "
6289 "zero values, and its associated data structure, I<shmid_ds> (see "
6290 "B<shmctl>(2)), is initialized as follows:"
6294 #: build/C/man2/shmget.2:151
6296 "I<shm_perm.cuid> and I<shm_perm.uid> are set to the effective user ID of the "
6301 #: build/C/man2/shmget.2:156
6303 "I<shm_perm.cgid> and I<shm_perm.gid> are set to the effective group ID of "
6304 "the calling process."
6308 #: build/C/man2/shmget.2:161
6310 "The least significant 9 bits of I<shm_perm.mode> are set to the least "
6311 "significant 9 bit of I<shmflg>."
6315 #: build/C/man2/shmget.2:165
6316 msgid "I<shm_segsz> is set to the value of I<size>."
6320 #: build/C/man2/shmget.2:172
6321 msgid "I<shm_lpid>, I<shm_nattch>, I<shm_atime> and I<shm_dtime> are set to 0."
6325 #: build/C/man2/shmget.2:175
6326 msgid "I<shm_ctime> is set to the current time."
6330 #: build/C/man2/shmget.2:178
6332 "If the shared memory segment already exists, the permissions are verified, "
6333 "and a check is made to see if it is marked for destruction."
6337 #: build/C/man2/shmget.2:182
6338 msgid "A valid segment identifier, I<shmid>, is returned on success, -1 on error."
6342 #: build/C/man2/shmget.2:186
6343 msgid "On failure, I<errno> is set to one of the following:"
6347 #: build/C/man2/shmget.2:192
6349 "The user does not have permission to access the shared memory segment, and "
6350 "does not have the B<CAP_IPC_OWNER> capability."
6354 #: build/C/man2/shmget.2:196
6355 msgid "B<IPC_CREAT | IPC_EXCL> was specified and the segment exists."
6359 #: build/C/man2/shmget.2:202
6361 "A new segment was to be created and I<size> E<lt> B<SHMMIN> or I<size> E<gt> "
6362 "B<SHMMAX>, or no new segment was to be created, a segment with given key "
6363 "existed, but I<size> is greater than the size of that segment."
6367 #: build/C/man2/shmget.2:211
6368 msgid "No segment exists for the given I<key>, and B<IPC_CREAT> was not specified."
6372 #: build/C/man2/shmget.2:214
6373 msgid "No memory could be allocated for segment overhead."
6377 #: build/C/man2/shmget.2:222
6379 "All possible shared memory IDs have been taken (B<SHMMNI>), or allocating a "
6380 "segment of the requested I<size> would cause the system to exceed the "
6381 "system-wide limit on shared memory (B<SHMALL>)."
6385 #: build/C/man2/shmget.2:229
6387 "The B<SHM_HUGETLB> flag was specified, but the caller was not privileged "
6388 "(did not have the B<CAP_IPC_LOCK> capability)."
6392 #: build/C/man2/shmget.2:235
6393 msgid "B<SHM_HUGETLB> is a nonportable Linux extension."
6397 #: build/C/man2/shmget.2:245
6399 "B<IPC_PRIVATE> isn't a flag field but a I<key_t> type. If this special "
6400 "value is used for I<key>, the system call ignores everything but the least "
6401 "significant 9 bits of I<shmflg> and creates a new shared memory segment (on "
6406 #: build/C/man2/shmget.2:249
6408 "The following limits on shared memory segment resources affect the "
6413 #: build/C/man2/shmget.2:249
6419 #: build/C/man2/shmget.2:254
6421 "System wide maximum of shared memory pages (on Linux, this limit can be read "
6422 "and modified via I</proc/sys/kernel/shmall>)."
6426 #: build/C/man2/shmget.2:254
6432 #: build/C/man2/shmget.2:259
6434 "Maximum size in bytes for a shared memory segment: policy dependent (on "
6435 "Linux, this limit can be read and modified via I</proc/sys/kernel/shmmax>)."
6439 #: build/C/man2/shmget.2:259
6445 #: build/C/man2/shmget.2:265
6447 "Minimum size in bytes for a shared memory segment: implementation dependent "
6448 "(currently 1 byte, though B<PAGE_SIZE> is the effective minimum size)."
6452 #: build/C/man2/shmget.2:265
6457 #. Kernels between 2.4.x and 2.6.8 had an off-by-one error that meant
6458 #. that we could create one more segment than SHMMNI -- MTK
6459 #. This /proc file is not available in Linux 2.2 and earlier -- MTK
6461 #: build/C/man2/shmget.2:274
6463 "System wide maximum number of shared memory segments: implementation "
6464 "dependent (currently 4096, was 128 before Linux 2.3.99; on Linux, this limit "
6465 "can be read and modified via I</proc/sys/kernel/shmmni>)."
6469 #: build/C/man2/shmget.2:278
6471 "The implementation has no specific limits for the per-process maximum number "
6472 "of shared memory segments (B<SHMSEG>)."
6476 #: build/C/man2/shmget.2:284
6478 "Until version 2.3.30 Linux would return B<EIDRM> for a B<shmget>() on a "
6479 "shared memory segment scheduled for deletion."
6483 #: build/C/man2/shmget.2:290
6485 "The name choice B<IPC_PRIVATE> was perhaps unfortunate, B<IPC_NEW> would "
6486 "more clearly show its function."
6490 #: build/C/man2/shmget.2:298
6492 "B<shmat>(2), B<shmctl>(2), B<shmdt>(2), B<ftok>(3), B<capabilities>(7), "
6493 "B<shm_overview>(7), B<svipc>(7)"
6497 #: build/C/man2/shmop.2:39
6503 #: build/C/man2/shmop.2:39
6509 #: build/C/man2/shmop.2:42
6510 msgid "shmat, shmdt - shared memory operations"
6514 #: build/C/man2/shmop.2:46
6517 "B<#include E<lt>sys/types.hE<gt>>\n"
6518 "B<#include E<lt>sys/shm.hE<gt>>\n"
6522 #: build/C/man2/shmop.2:48
6525 "B<void *shmat(int >I<shmid>B<, const void *>I<shmaddr>B<, int "
6530 #: build/C/man2/shmop.2:50
6532 msgid "B<int shmdt(const void *>I<shmaddr>B<);>\n"
6536 #: build/C/man2/shmop.2:59
6538 "B<shmat>() attaches the shared memory segment identified by I<shmid> to the "
6539 "address space of the calling process. The attaching address is specified by "
6540 "I<shmaddr> with one of the following criteria:"
6544 #: build/C/man2/shmop.2:65
6546 "If I<shmaddr> is NULL, the system chooses a suitable (unused) address at "
6547 "which to attach the segment."
6551 #: build/C/man2/shmop.2:80
6553 "If I<shmaddr> isn't NULL and B<SHM_RND> is specified in I<shmflg>, the "
6554 "attach occurs at the address equal to I<shmaddr> rounded down to the nearest "
6555 "multiple of B<SHMLBA>. Otherwise I<shmaddr> must be a page-aligned address "
6556 "at which the attach occurs."
6560 #: build/C/man2/shmop.2:90
6562 "If B<SHM_RDONLY> is specified in I<shmflg>, the segment is attached for "
6563 "reading and the process must have read permission for the segment. "
6564 "Otherwise the segment is attached for read and write and the process must "
6565 "have read and write permission for the segment. There is no notion of a "
6566 "write-only shared memory segment."
6570 #: build/C/man2/shmop.2:105
6572 "The (Linux-specific) B<SHM_REMAP> flag may be specified in I<shmflg> to "
6573 "indicate that the mapping of the segment should replace any existing mapping "
6574 "in the range starting at I<shmaddr> and continuing for the size of the "
6575 "segment. (Normally an B<EINVAL> error would result if a mapping already "
6576 "exists in this address range.) In this case, I<shmaddr> must not be NULL."
6580 #: build/C/man2/shmop.2:112
6582 "The B<brk>(2) value of the calling process is not altered by the attach. "
6583 "The segment will automatically be detached at process exit. The same "
6584 "segment may be attached as a read and as a read-write one, and more than "
6585 "once, in the process's address space."
6589 #: build/C/man2/shmop.2:120
6591 "A successful B<shmat>() call updates the members of the I<shmid_ds> "
6592 "structure (see B<shmctl>(2)) associated with the shared memory segment as "
6597 #: build/C/man2/shmop.2:123
6598 msgid "I<shm_atime> is set to the current time."
6602 #: build/C/man2/shmop.2:126 build/C/man2/shmop.2:152
6603 msgid "I<shm_lpid> is set to the process-ID of the calling process."
6607 #: build/C/man2/shmop.2:129
6608 msgid "I<shm_nattch> is incremented by one."
6612 #: build/C/man2/shmop.2:140
6614 "B<shmdt>() detaches the shared memory segment located at the address "
6615 "specified by I<shmaddr> from the address space of the calling process. The "
6616 "to-be-detached segment must be currently attached with I<shmaddr> equal to "
6617 "the value returned by the attaching B<shmat>() call."
6621 #: build/C/man2/shmop.2:146
6623 "On a successful B<shmdt>() call the system updates the members of the "
6624 "I<shmid_ds> structure associated with the shared memory segment as follows:"
6628 #: build/C/man2/shmop.2:149
6629 msgid "I<shm_dtime> is set to the current time."
6633 #: build/C/man2/shmop.2:157
6635 "I<shm_nattch> is decremented by one. If it becomes 0 and the segment is "
6636 "marked for deletion, the segment is deleted."
6640 #: build/C/man2/shmop.2:161
6641 msgid "After a B<fork>(2) the child inherits the attached shared memory segments."
6645 #: build/C/man2/shmop.2:165
6647 "After an B<execve>(2) all attached shared memory segments are detached from "
6652 #: build/C/man2/shmop.2:169
6654 "Upon B<_exit>(2) all attached shared memory segments are detached from the "
6659 #: build/C/man2/shmop.2:177
6661 "On success B<shmat>() returns the address of the attached shared memory "
6662 "segment; on error I<(void\\ *)\\ -1> is returned, and I<errno> is set to "
6663 "indicate the cause of the error."
6667 #: build/C/man2/shmop.2:183
6669 "On success B<shmdt>() returns 0; on error -1 is returned, and I<errno> is "
6670 "set to indicate the cause of the error."
6674 #: build/C/man2/shmop.2:189
6675 msgid "When B<shmat>() fails, I<errno> is set to one of the following:"
6679 #: build/C/man2/shmop.2:195
6681 "The calling process does not have the required permissions for the requested "
6682 "attach type, and does not have the B<CAP_IPC_OWNER> capability."
6686 #: build/C/man2/shmop.2:209
6688 "Invalid I<shmid> value, unaligned (i.e., not page-aligned and B<SHM_RND> was "
6689 "not specified) or invalid I<shmaddr> value, or can't attach segment at "
6690 "I<shmaddr>, or B<SHM_REMAP> was specified and I<shmaddr> was NULL."
6694 #: build/C/man2/shmop.2:212
6695 msgid "Could not allocate memory for the descriptor or for the page tables."
6699 #: build/C/man2/shmop.2:218
6700 msgid "When B<shmdt>() fails, I<errno> is set as follows:"
6703 #. The following since 2.6.17-rc1:
6705 #: build/C/man2/shmop.2:226
6707 "There is no shared memory segment attached at I<shmaddr>; or, I<shmaddr> is "
6708 "not aligned on a page boundary."
6712 #: build/C/man2/shmop.2:245
6714 "In SVID 3 (or perhaps earlier) the type of the I<shmaddr> argument was "
6715 "changed from I<char *> into I<const void *>, and the returned type of "
6716 "B<shmat>() from I<char *> into I<void *>. (Linux libc4 and libc5 have the "
6717 "I<char *> prototypes; glibc2 has I<void *>.)"
6721 #: build/C/man2/shmop.2:257
6723 "Using B<shmat>() with I<shmaddr> equal to NULL is the preferred, portable "
6724 "way of attaching a shared memory segment. Be aware that the shared memory "
6725 "segment attached in this way may be attached at different addresses in "
6726 "different processes. Therefore, any pointers maintained within the shared "
6727 "memory must be made relative (typically to the starting address of the "
6728 "segment), rather than absolute."
6732 #: build/C/man2/shmop.2:262
6734 "On Linux, it is possible to attach a shared memory segment even if it is "
6735 "already marked to be deleted. However, POSIX.1-2001 does not specify this "
6736 "behavior and many other implementations do not support it."
6740 #: build/C/man2/shmop.2:265
6741 msgid "The following system parameter affects B<shmat>():"
6744 #. FIXME A good explanation of the rationale for the existence
6745 #. of SHMLBA would be useful here
6747 #: build/C/man2/shmop.2:265
6752 #. FIXME That last sentence isn't true for all Linux
6753 #. architectures (i.e., SHMLBA != PAGE_SIZE for some architectures)
6756 #: build/C/man2/shmop.2:278
6758 "Segment low boundary address multiple. Must be page aligned. For the "
6759 "current implementation the B<SHMLBA> value is B<PAGE_SIZE>."
6763 #: build/C/man2/shmop.2:282
6765 "The implementation places no intrinsic limit on the per-process maximum "
6766 "number of shared memory segments (B<SHMSEG>)."
6770 #: build/C/man2/shmop.2:290
6772 "B<brk>(2), B<mmap>(2), B<shmctl>(2), B<shmget>(2), B<capabilities>(7), "
6773 "B<shm_overview>(7), B<svipc>(7)"
6777 #: build/C/man2/subpage_prot.2:28
6779 msgid "SUBPAGE_PROT"
6783 #: build/C/man2/subpage_prot.2:28
6789 #: build/C/man2/subpage_prot.2:31
6790 msgid "subpage_prot - define a subpage protection for an address range"
6794 #: build/C/man2/subpage_prot.2:35
6797 "B<long subpage_prot(unsigned long >I<addr>B<, unsigned long >I<len>B<,>\n"
6798 "B< uint32_t *>I<map>B<);>\n"
6802 #: build/C/man2/subpage_prot.2:42
6804 "The PowerPC-specific B<subpage_prot>() system call provides the facility to "
6805 "control the access permissions on individual 4kB subpages on systems "
6806 "configured with a page size of 64kB."
6810 #: build/C/man2/subpage_prot.2:49
6812 "The protection map is applied to the memory pages in the region starting at "
6813 "I<addr> and continuing for I<len> bytes. Both of these arguments must be "
6814 "aligned to a 64-kB boundary."
6818 #: build/C/man2/subpage_prot.2:61
6820 "The protection map is specified in the buffer pointed to by I<map>. The map "
6821 "has 2 bits per 4kB subpage; thus each 32-bit word specifies the protections "
6822 "of 16 4kB subpages inside a 64kB page (so, the number of 32-bit words "
6823 "pointed to by I<map> should equate to the number of 64-kB pages specified by "
6824 "I<len>). Each 2-bit field in the protection map is either 0 to allow any "
6825 "access, 1 to prevent writes, or 2 or 3 to prevent all accesses."
6829 #: build/C/man2/subpage_prot.2:66
6831 "On success, B<subpage_prot>() returns 0. Otherwise, one of the error codes "
6832 "specified below is returned."
6836 #: build/C/man2/subpage_prot.2:72
6837 msgid "The buffer referred to by I<map> is not accessible."
6841 #: build/C/man2/subpage_prot.2:82
6843 "The I<addr> or I<len> arguments are incorrect. Both of these arguments must "
6844 "be aligned to a multiple of the system page size, and they must not refer to "
6845 "a region outside of the address space of the process or to a region that "
6846 "consists of huge pages."
6850 #: build/C/man2/subpage_prot.2:85 build/C/man2/sync_file_range.2:166
6851 msgid "Out of memory."
6855 #: build/C/man2/subpage_prot.2:91
6857 "This system call is provided on the PowerPC architecture since Linux "
6858 "2.6.25. The system call is provided only if the kernel is configured with "
6859 "B<CONFIG_PPC_64K_PAGES>. No library support is provided."
6863 #: build/C/man2/subpage_prot.2:98
6865 "Normal page protections (at the 64-kB page level) also apply; the subpage "
6866 "protection mechanism is an additional constraint, so putting 0 in a 2-bit "
6867 "field won't allow writes to a page that is otherwise write-protected."
6871 #: build/C/man2/subpage_prot.2:98
6876 #. In the initial implementation, it was the case that:
6877 #. In fact the whole process is switched to use 4k hardware pages when the
6878 #. subpage_prot system call is used, but this could be improved in future
6879 #. to switch only the affected segments.
6880 #. But Paul Mackerass says (Oct 2010): I'm pretty sure we now only switch
6881 #. the affected segment, not the whole process.
6883 #: build/C/man2/subpage_prot.2:120
6885 "This system call is provided to assist writing emulators that operate using "
6886 "64-kB pages on PowerPC systems. When emulating systems such as x86, which "
6887 "uses a smaller page size, the emulator can no longer use the "
6888 "memory-management unit (MMU) and normal system calls for controlling page "
6889 "protections. (The emulator could emulate the MMU by checking and possibly "
6890 "remapping the address for each memory access in software, but that is slow.) "
6891 "The idea is that the emulator supplies an array of protection masks to apply "
6892 "to a specified range of virtual addresses. These masks are applied at the "
6893 "level where hardware page-table entries (PTEs) are inserted into the "
6894 "hardware page table based on the Linux PTEs, so the Linux PTEs are not "
6895 "affected. Implicit in this is that the regions of the address space that "
6896 "are protected are switched to use 4-kB hardware pages rather than 64-kB "
6897 "hardware pages (on machines with hardware 64-kB page support)."
6901 #: build/C/man2/subpage_prot.2:123
6902 msgid "B<mprotect>(2), B<syscall>(2);"
6906 #: build/C/man2/subpage_prot.2:126
6907 msgid "the kernel source file I<Documentation/vm/hugetlbpage.txt>."
6911 #: build/C/man2/sync_file_range.2:30
6913 msgid "SYNC_FILE_RANGE"
6917 #: build/C/man2/sync_file_range.2:33
6918 msgid "sync_file_range - sync a file segment with disk"
6922 #: build/C/man2/sync_file_range.2:37
6925 "B<#define _GNU_SOURCE> /* See feature_test_macros(7) */\n"
6926 "B<#include E<lt>fcntl.hE<gt>>\n"
6930 #: build/C/man2/sync_file_range.2:40
6933 "B<int sync_file_range(int >I<fd>B<, off64_t >I<offset>B<, off64_t "
6935 "B< unsigned int >I<flags>B<);>\n"
6939 #: build/C/man2/sync_file_range.2:47
6941 "B<sync_file_range>() permits fine control when synchronizing the open file "
6942 "referred to by the file descriptor I<fd> with disk."
6946 #: build/C/man2/sync_file_range.2:61
6948 "I<offset> is the starting byte of the file range to be synchronized. "
6949 "I<nbytes> specifies the length of the range to be synchronized, in bytes; if "
6950 "I<nbytes> is zero, then all bytes from I<offset> through to the end of file "
6951 "are synchronized. Synchronization is in units of the system page size: "
6952 "I<offset> is rounded down to a page boundary; I<(offset+nbytes-1)> is "
6953 "rounded up to a page boundary."
6957 #: build/C/man2/sync_file_range.2:65
6958 msgid "The I<flags> bit-mask argument can include any of the following values:"
6962 #: build/C/man2/sync_file_range.2:65
6964 msgid "B<SYNC_FILE_RANGE_WAIT_BEFORE>"
6968 #: build/C/man2/sync_file_range.2:70
6970 "Wait upon write-out of all pages in the specified range that have already "
6971 "been submitted to the device driver for write-out before performing any "
6976 #: build/C/man2/sync_file_range.2:70 build/C/man2/sync_file_range.2:120
6978 msgid "B<SYNC_FILE_RANGE_WRITE>"
6982 #: build/C/man2/sync_file_range.2:76
6984 "Initiate write-out of all dirty pages in the specified range which are not "
6985 "presently submitted write-out. Note that even this may block if you attempt "
6986 "to write more than request queue size."
6990 #: build/C/man2/sync_file_range.2:76
6992 msgid "B<SYNC_FILE_RANGE_WAIT_AFTER>"
6996 #: build/C/man2/sync_file_range.2:80
6997 msgid "Wait upon write-out of all pages in the range after performing any write."
7001 #: build/C/man2/sync_file_range.2:84
7002 msgid "Specifying I<flags> as 0 is permitted, as a no-op."
7006 #: build/C/man2/sync_file_range.2:84
7012 #: build/C/man2/sync_file_range.2:100
7014 "This system call is extremely dangerous and should not be used in portable "
7015 "programs. None of these operations writes out the file's metadata. "
7016 "Therefore, unless the application is strictly performing overwrites of "
7017 "already-instantiated disk blocks, there are no guarantees that the data will "
7018 "be available after a crash. There is no user interface to know if a write "
7019 "is purely an overwrite. On file systems using copy-on-write semantics "
7020 "(e.g., I<btrfs>) an overwrite of existing allocated blocks is impossible. "
7021 "When writing into preallocated space, many file systems also require calls "
7022 "into the block allocator, which this system call does not sync out to disk. "
7023 "This system call does not flush disk write caches and thus does not provide "
7024 "any data integrity on systems with volatile disk write caches."
7028 #: build/C/man2/sync_file_range.2:100
7030 msgid "Some details"
7034 #: build/C/man2/sync_file_range.2:108
7036 "B<SYNC_FILE_RANGE_WAIT_BEFORE> and B<SYNC_FILE_RANGE_WAIT_AFTER> will detect "
7037 "any I/O errors or B<ENOSPC> conditions and will return these to the caller."
7041 #: build/C/man2/sync_file_range.2:112
7042 msgid "Useful combinations of the I<flags> bits are:"
7046 #: build/C/man2/sync_file_range.2:112
7048 msgid "B<SYNC_FILE_RANGE_WAIT_BEFORE | SYNC_FILE_RANGE_WRITE>"
7052 #: build/C/man2/sync_file_range.2:120
7054 "Ensures that all pages in the specified range which were dirty when "
7055 "B<sync_file_range>() was called are placed under write-out. This is a "
7056 "start-write-for-data-integrity operation."
7060 #: build/C/man2/sync_file_range.2:127
7062 "Start write-out of all dirty pages in the specified range which are not "
7063 "presently under write-out. This is an asynchronous flush-to-disk "
7064 "operation. This is not suitable for data integrity operations."
7068 #: build/C/man2/sync_file_range.2:127
7070 msgid "B<SYNC_FILE_RANGE_WAIT_BEFORE> (or B<SYNC_FILE_RANGE_WAIT_AFTER>)"
7074 #: build/C/man2/sync_file_range.2:134
7076 "Wait for completion of write-out of all pages in the specified range. This "
7077 "can be used after an earlier B<SYNC_FILE_RANGE_WAIT_BEFORE | "
7078 "SYNC_FILE_RANGE_WRITE> operation to wait for completion of that operation, "
7079 "and obtain its result."
7083 #: build/C/man2/sync_file_range.2:134
7086 "B<SYNC_FILE_RANGE_WAIT_BEFORE | SYNC_FILE_RANGE_WRITE | "
7087 "SYNC_FILE_RANGE_WAIT_AFTER>"
7091 #: build/C/man2/sync_file_range.2:141
7093 "This is a write-for-data-integrity operation that will ensure that all pages "
7094 "in the specified range which were dirty when B<sync_file_range>() was "
7095 "called are committed to disk."
7099 #: build/C/man2/sync_file_range.2:147
7101 "On success, B<sync_file_range>() returns 0; on failure -1 is returned and "
7102 "I<errno> is set to indicate the error."
7106 #: build/C/man2/sync_file_range.2:152
7107 msgid "I<fd> is not a valid file descriptor."
7111 #: build/C/man2/sync_file_range.2:160
7112 msgid "I<flags> specifies an invalid bit; or I<offset> or I<nbytes> is invalid."
7116 #: build/C/man2/sync_file_range.2:163
7121 #: build/C/man2/sync_file_range.2:169
7122 msgid "Out of disk space."
7125 #. FIXME . (bug?) Actually, how can 'fd' refer to a symbolic link (S_ISLNK)?
7126 #. (In userspace at least) it isn't possible to obtain a file descriptor
7127 #. for a symbolic link.
7129 #: build/C/man2/sync_file_range.2:177
7131 "I<fd> refers to something other than a regular file, a block device, a "
7132 "directory, or a symbolic link."
7136 #: build/C/man2/sync_file_range.2:180
7137 msgid "B<sync_file_range>() appeared on Linux in kernel 2.6.17."
7141 #: build/C/man2/sync_file_range.2:183
7143 "This system call is Linux-specific, and should be avoided in portable "
7147 #. See kernel commit edd5cd4a9424f22b0fa08bef5e299d41befd5622
7149 #: build/C/man2/sync_file_range.2:196
7151 "Some architectures (e.g., PowerPC, ARM) need 64-bit arguments to be aligned "
7152 "in a suitable pair of registers. On such architectures, the call signature "
7153 "of B<sync_file_range>() is flawed, since it forces a register to be wasted "
7154 "as padding between the I<fd> and I<offset> arguments. Therefore, these "
7155 "architectures define a different system call that orders the arguments "
7160 #: build/C/man2/sync_file_range.2:201
7163 "B<int sync_file_range2(int >I<fd>B<, unsigned int >I<flags>B<,>\n"
7164 "B< off64_t >I<offset>B<, off64_t >I<nbytes>B<);>\n"
7168 #: build/C/man2/sync_file_range.2:206
7170 "The behavior of this system call is otherwise exactly the same as "
7171 "B<sync_file_range>()."
7175 #: build/C/man2/sync_file_range.2:217
7177 "A system call with this signature first appeared on the ARM architecture in "
7178 "Linux 2.6.20, with the name B<arm_sync_file_range>(). It was renamed in "
7179 "Linux 2.6.22, when the analogous system call was added for PowerPC. On "
7180 "architectures where glibc support is provided, glibc transparently wraps "
7181 "B<sync_file_range2>() under the name B<sync_file_range>()."
7185 #: build/C/man2/sync_file_range.2:222
7186 msgid "B<fdatasync>(2), B<fsync>(2), B<msync>(2), B<sync>(2)"