]>
Commit | Line | Data |
---|---|---|
28f540f4 | 1 | @node Low-Level I/O, File System Interface, I/O on Streams, Top |
7a68c94a | 2 | @c %MENU% Low-level, less portable I/O |
28f540f4 RM |
3 | @chapter Low-Level Input/Output |
4 | ||
5 | This chapter describes functions for performing low-level input/output | |
6 | operations on file descriptors. These functions include the primitives | |
7 | for the higher-level I/O functions described in @ref{I/O on Streams}, as | |
8 | well as functions for performing low-level control operations for which | |
9 | there are no equivalents on streams. | |
10 | ||
11 | Stream-level I/O is more flexible and usually more convenient; | |
12 | therefore, programmers generally use the descriptor-level functions only | |
13 | when necessary. These are some of the usual reasons: | |
14 | ||
15 | @itemize @bullet | |
16 | @item | |
17 | For reading binary files in large chunks. | |
18 | ||
19 | @item | |
20 | For reading an entire file into core before parsing it. | |
21 | ||
22 | @item | |
23 | To perform operations other than data transfer, which can only be done | |
24 | with a descriptor. (You can use @code{fileno} to get the descriptor | |
25 | corresponding to a stream.) | |
26 | ||
27 | @item | |
28 | To pass descriptors to a child process. (The child can create its own | |
29 | stream to use a descriptor that it inherits, but cannot inherit a stream | |
30 | directly.) | |
31 | @end itemize | |
32 | ||
33 | @menu | |
34 | * Opening and Closing Files:: How to open and close file | |
2c6fe0bd | 35 | descriptors. |
28f540f4 RM |
36 | * I/O Primitives:: Reading and writing data. |
37 | * File Position Primitive:: Setting a descriptor's file | |
2c6fe0bd | 38 | position. |
28f540f4 RM |
39 | * Descriptors and Streams:: Converting descriptor to stream |
40 | or vice-versa. | |
41 | * Stream/Descriptor Precautions:: Precautions needed if you use both | |
42 | descriptors and streams. | |
49c091e5 | 43 | * Scatter-Gather:: Fast I/O to discontinuous buffers. |
bad7a0c8 | 44 | * Copying File Data:: Copying data between files. |
07435eb4 | 45 | * Memory-mapped I/O:: Using files like memory. |
28f540f4 RM |
46 | * Waiting for I/O:: How to check for input or output |
47 | on multiple file descriptors. | |
dfd2257a | 48 | * Synchronizing I/O:: Making sure all I/O actions completed. |
b07d03e0 | 49 | * Asynchronous I/O:: Perform I/O in parallel. |
28f540f4 RM |
50 | * Control Operations:: Various other operations on file |
51 | descriptors. | |
52 | * Duplicating Descriptors:: Fcntl commands for duplicating | |
53 | file descriptors. | |
54 | * Descriptor Flags:: Fcntl commands for manipulating | |
55 | flags associated with file | |
2c6fe0bd | 56 | descriptors. |
28f540f4 RM |
57 | * File Status Flags:: Fcntl commands for manipulating |
58 | flags associated with open files. | |
59 | * File Locks:: Fcntl commands for implementing | |
60 | file locking. | |
0961f7e1 JL |
61 | * Open File Description Locks:: Fcntl commands for implementing |
62 | open file description locking. | |
63 | * Open File Description Locks Example:: An example of open file description lock | |
64 | usage | |
28f540f4 RM |
65 | * Interrupt Input:: Getting an asynchronous signal when |
66 | input arrives. | |
07435eb4 | 67 | * IOCTLs:: Generic I/O Control operations. |
28f540f4 RM |
68 | @end menu |
69 | ||
70 | ||
71 | @node Opening and Closing Files | |
72 | @section Opening and Closing Files | |
73 | ||
74 | @cindex opening a file descriptor | |
75 | @cindex closing a file descriptor | |
76 | This section describes the primitives for opening and closing files | |
77 | using file descriptors. The @code{open} and @code{creat} functions are | |
78 | declared in the header file @file{fcntl.h}, while @code{close} is | |
79 | declared in @file{unistd.h}. | |
80 | @pindex unistd.h | |
81 | @pindex fcntl.h | |
82 | ||
28f540f4 | 83 | @deftypefun int open (const char *@var{filename}, int @var{flags}[, mode_t @var{mode}]) |
d08a7e4c | 84 | @standards{POSIX.1, fcntl.h} |
2cc3615c | 85 | @safety{@prelim{}@mtsafe{}@assafe{}@acsafe{@acsfd{}}} |
624254b1 SC |
86 | The @code{open} function creates and returns a new file descriptor for |
87 | the file named by @var{filename}. Initially, the file position | |
28f540f4 | 88 | indicator for the file is at the beginning of the file. The argument |
624254b1 SC |
89 | @var{mode} (@pxref{Permission Bits}) is used only when a file is |
90 | created, but it doesn't hurt to supply the argument in any case. | |
28f540f4 RM |
91 | |
92 | The @var{flags} argument controls how the file is to be opened. This is | |
93 | a bit mask; you create the value by the bitwise OR of the appropriate | |
94 | parameters (using the @samp{|} operator in C). | |
95 | @xref{File Status Flags}, for the parameters available. | |
96 | ||
97 | The normal return value from @code{open} is a non-negative integer file | |
07435eb4 | 98 | descriptor. In the case of an error, a value of @math{-1} is returned |
28f540f4 RM |
99 | instead. In addition to the usual file name errors (@pxref{File |
100 | Name Errors}), the following @code{errno} error conditions are defined | |
101 | for this function: | |
102 | ||
103 | @table @code | |
104 | @item EACCES | |
19e4c7dd | 105 | The file exists but is not readable/writable as requested by the @var{flags} |
9739d2d5 | 106 | argument, or the file does not exist and the directory is unwritable so |
28f540f4 RM |
107 | it cannot be created. |
108 | ||
109 | @item EEXIST | |
110 | Both @code{O_CREAT} and @code{O_EXCL} are set, and the named file already | |
111 | exists. | |
112 | ||
113 | @item EINTR | |
114 | The @code{open} operation was interrupted by a signal. | |
115 | @xref{Interrupted Primitives}. | |
116 | ||
117 | @item EISDIR | |
118 | The @var{flags} argument specified write access, and the file is a directory. | |
119 | ||
120 | @item EMFILE | |
121 | The process has too many files open. | |
122 | The maximum number of file descriptors is controlled by the | |
123 | @code{RLIMIT_NOFILE} resource limit; @pxref{Limits on Resources}. | |
124 | ||
125 | @item ENFILE | |
126 | The entire system, or perhaps the file system which contains the | |
127 | directory, cannot support any additional open files at the moment. | |
a7a93d50 | 128 | (This problem cannot happen on @gnuhurdsystems{}.) |
28f540f4 RM |
129 | |
130 | @item ENOENT | |
131 | The named file does not exist, and @code{O_CREAT} is not specified. | |
132 | ||
133 | @item ENOSPC | |
134 | The directory or file system that would contain the new file cannot be | |
135 | extended, because there is no disk space left. | |
136 | ||
137 | @item ENXIO | |
138 | @code{O_NONBLOCK} and @code{O_WRONLY} are both set in the @var{flags} | |
139 | argument, the file named by @var{filename} is a FIFO (@pxref{Pipes and | |
140 | FIFOs}), and no process has the file open for reading. | |
141 | ||
142 | @item EROFS | |
143 | The file resides on a read-only file system and any of @w{@code{O_WRONLY}}, | |
144 | @code{O_RDWR}, and @code{O_TRUNC} are set in the @var{flags} argument, | |
145 | or @code{O_CREAT} is set and the file does not already exist. | |
146 | @end table | |
147 | ||
148 | @c !!! umask | |
149 | ||
04b9968b | 150 | If on a 32 bit machine the sources are translated with |
b07d03e0 UD |
151 | @code{_FILE_OFFSET_BITS == 64} the function @code{open} returns a file |
152 | descriptor opened in the large file mode which enables the file handling | |
9ceeb279 OB |
153 | functions to use files up to @twoexp{63} bytes in size and offset from |
154 | @minus{}@twoexp{63} to @twoexp{63}. This happens transparently for the user | |
9739d2d5 | 155 | since all of the low-level file handling functions are equally replaced. |
b07d03e0 | 156 | |
04b9968b | 157 | This function is a cancellation point in multi-threaded programs. This |
dfd2257a UD |
158 | is a problem if the thread allocates some resources (like memory, file |
159 | descriptors, semaphores or whatever) at the time @code{open} is | |
19e4c7dd | 160 | called. If the thread gets canceled these resources stay allocated |
dfd2257a | 161 | until the program ends. To avoid this calls to @code{open} should be |
04b9968b | 162 | protected using cancellation handlers. |
dfd2257a UD |
163 | @c ref pthread_cleanup_push / pthread_cleanup_pop |
164 | ||
28f540f4 RM |
165 | The @code{open} function is the underlying primitive for the @code{fopen} |
166 | and @code{freopen} functions, that create streams. | |
167 | @end deftypefun | |
168 | ||
b07d03e0 | 169 | @deftypefun int open64 (const char *@var{filename}, int @var{flags}[, mode_t @var{mode}]) |
d08a7e4c | 170 | @standards{Unix98, fcntl.h} |
2cc3615c | 171 | @safety{@prelim{}@mtsafe{}@assafe{}@acsafe{@acsfd{}}} |
b07d03e0 UD |
172 | This function is similar to @code{open}. It returns a file descriptor |
173 | which can be used to access the file named by @var{filename}. The only | |
04b9968b | 174 | difference is that on 32 bit systems the file is opened in the |
b07d03e0 UD |
175 | large file mode. I.e., file length and file offsets can exceed 31 bits. |
176 | ||
b07d03e0 UD |
177 | When the sources are translated with @code{_FILE_OFFSET_BITS == 64} this |
178 | function is actually available under the name @code{open}. I.e., the | |
179 | new, extended API using 64 bit file sizes and offsets transparently | |
180 | replaces the old API. | |
181 | @end deftypefun | |
182 | ||
28f540f4 | 183 | @deftypefn {Obsolete function} int creat (const char *@var{filename}, mode_t @var{mode}) |
d08a7e4c | 184 | @standards{POSIX.1, fcntl.h} |
2cc3615c | 185 | @safety{@prelim{}@mtsafe{}@assafe{}@acsafe{@acsfd{}}} |
28f540f4 RM |
186 | This function is obsolete. The call: |
187 | ||
188 | @smallexample | |
189 | creat (@var{filename}, @var{mode}) | |
190 | @end smallexample | |
191 | ||
192 | @noindent | |
193 | is equivalent to: | |
194 | ||
195 | @smallexample | |
196 | open (@var{filename}, O_WRONLY | O_CREAT | O_TRUNC, @var{mode}) | |
197 | @end smallexample | |
b07d03e0 | 198 | |
04b9968b | 199 | If on a 32 bit machine the sources are translated with |
b07d03e0 UD |
200 | @code{_FILE_OFFSET_BITS == 64} the function @code{creat} returns a file |
201 | descriptor opened in the large file mode which enables the file handling | |
9ceeb279 OB |
202 | functions to use files up to @twoexp{63} in size and offset from |
203 | @minus{}@twoexp{63} to @twoexp{63}. This happens transparently for the user | |
9739d2d5 | 204 | since all of the low-level file handling functions are equally replaced. |
b07d03e0 UD |
205 | @end deftypefn |
206 | ||
b07d03e0 | 207 | @deftypefn {Obsolete function} int creat64 (const char *@var{filename}, mode_t @var{mode}) |
d08a7e4c | 208 | @standards{Unix98, fcntl.h} |
2cc3615c | 209 | @safety{@prelim{}@mtsafe{}@assafe{}@acsafe{@acsfd{}}} |
b07d03e0 UD |
210 | This function is similar to @code{creat}. It returns a file descriptor |
211 | which can be used to access the file named by @var{filename}. The only | |
9739d2d5 | 212 | difference is that on 32 bit systems the file is opened in the |
b07d03e0 UD |
213 | large file mode. I.e., file length and file offsets can exceed 31 bits. |
214 | ||
215 | To use this file descriptor one must not use the normal operations but | |
216 | instead the counterparts named @code{*64}, e.g., @code{read64}. | |
217 | ||
218 | When the sources are translated with @code{_FILE_OFFSET_BITS == 64} this | |
219 | function is actually available under the name @code{open}. I.e., the | |
220 | new, extended API using 64 bit file sizes and offsets transparently | |
221 | replaces the old API. | |
28f540f4 RM |
222 | @end deftypefn |
223 | ||
28f540f4 | 224 | @deftypefun int close (int @var{filedes}) |
d08a7e4c | 225 | @standards{POSIX.1, unistd.h} |
2cc3615c | 226 | @safety{@prelim{}@mtsafe{}@assafe{}@acsafe{@acsfd{}}} |
28f540f4 RM |
227 | The function @code{close} closes the file descriptor @var{filedes}. |
228 | Closing a file has the following consequences: | |
229 | ||
230 | @itemize @bullet | |
2c6fe0bd | 231 | @item |
28f540f4 RM |
232 | The file descriptor is deallocated. |
233 | ||
234 | @item | |
235 | Any record locks owned by the process on the file are unlocked. | |
236 | ||
237 | @item | |
238 | When all file descriptors associated with a pipe or FIFO have been closed, | |
239 | any unread data is discarded. | |
240 | @end itemize | |
241 | ||
04b9968b | 242 | This function is a cancellation point in multi-threaded programs. This |
dfd2257a UD |
243 | is a problem if the thread allocates some resources (like memory, file |
244 | descriptors, semaphores or whatever) at the time @code{close} is | |
19e4c7dd | 245 | called. If the thread gets canceled these resources stay allocated |
04b9968b UD |
246 | until the program ends. To avoid this, calls to @code{close} should be |
247 | protected using cancellation handlers. | |
dfd2257a UD |
248 | @c ref pthread_cleanup_push / pthread_cleanup_pop |
249 | ||
07435eb4 | 250 | The normal return value from @code{close} is @math{0}; a value of @math{-1} |
28f540f4 RM |
251 | is returned in case of failure. The following @code{errno} error |
252 | conditions are defined for this function: | |
253 | ||
254 | @table @code | |
255 | @item EBADF | |
256 | The @var{filedes} argument is not a valid file descriptor. | |
257 | ||
258 | @item EINTR | |
259 | The @code{close} call was interrupted by a signal. | |
260 | @xref{Interrupted Primitives}. | |
261 | Here is an example of how to handle @code{EINTR} properly: | |
262 | ||
263 | @smallexample | |
264 | TEMP_FAILURE_RETRY (close (desc)); | |
265 | @end smallexample | |
266 | ||
267 | @item ENOSPC | |
268 | @itemx EIO | |
269 | @itemx EDQUOT | |
2c6fe0bd | 270 | When the file is accessed by NFS, these errors from @code{write} can sometimes |
28f540f4 RM |
271 | not be detected until @code{close}. @xref{I/O Primitives}, for details |
272 | on their meaning. | |
273 | @end table | |
b07d03e0 UD |
274 | |
275 | Please note that there is @emph{no} separate @code{close64} function. | |
276 | This is not necessary since this function does not determine nor depend | |
fed8f7f7 | 277 | on the mode of the file. The kernel which performs the @code{close} |
04b9968b | 278 | operation knows which mode the descriptor is used for and can handle |
b07d03e0 | 279 | this situation. |
28f540f4 RM |
280 | @end deftypefun |
281 | ||
282 | To close a stream, call @code{fclose} (@pxref{Closing Streams}) instead | |
283 | of trying to close its underlying file descriptor with @code{close}. | |
284 | This flushes any buffered output and updates the stream object to | |
285 | indicate that it is closed. | |
286 | ||
287 | @node I/O Primitives | |
288 | @section Input and Output Primitives | |
289 | ||
290 | This section describes the functions for performing primitive input and | |
291 | output operations on file descriptors: @code{read}, @code{write}, and | |
292 | @code{lseek}. These functions are declared in the header file | |
293 | @file{unistd.h}. | |
294 | @pindex unistd.h | |
295 | ||
28f540f4 | 296 | @deftp {Data Type} ssize_t |
d08a7e4c | 297 | @standards{POSIX.1, unistd.h} |
28f540f4 RM |
298 | This data type is used to represent the sizes of blocks that can be |
299 | read or written in a single operation. It is similar to @code{size_t}, | |
300 | but must be a signed type. | |
301 | @end deftp | |
302 | ||
303 | @cindex reading from a file descriptor | |
28f540f4 | 304 | @deftypefun ssize_t read (int @var{filedes}, void *@var{buffer}, size_t @var{size}) |
d08a7e4c | 305 | @standards{POSIX.1, unistd.h} |
2cc3615c | 306 | @safety{@prelim{}@mtsafe{}@assafe{}@acsafe{}} |
28f540f4 RM |
307 | The @code{read} function reads up to @var{size} bytes from the file |
308 | with descriptor @var{filedes}, storing the results in the @var{buffer}. | |
04b9968b UD |
309 | (This is not necessarily a character string, and no terminating null |
310 | character is added.) | |
28f540f4 RM |
311 | |
312 | @cindex end-of-file, on a file descriptor | |
313 | The return value is the number of bytes actually read. This might be | |
314 | less than @var{size}; for example, if there aren't that many bytes left | |
315 | in the file or if there aren't that many bytes immediately available. | |
316 | The exact behavior depends on what kind of file it is. Note that | |
317 | reading less than @var{size} bytes is not an error. | |
318 | ||
319 | A value of zero indicates end-of-file (except if the value of the | |
320 | @var{size} argument is also zero). This is not considered an error. | |
321 | If you keep calling @code{read} while at end-of-file, it will keep | |
322 | returning zero and doing nothing else. | |
323 | ||
324 | If @code{read} returns at least one character, there is no way you can | |
325 | tell whether end-of-file was reached. But if you did reach the end, the | |
326 | next read will return zero. | |
327 | ||
07435eb4 | 328 | In case of an error, @code{read} returns @math{-1}. The following |
28f540f4 RM |
329 | @code{errno} error conditions are defined for this function: |
330 | ||
331 | @table @code | |
332 | @item EAGAIN | |
333 | Normally, when no input is immediately available, @code{read} waits for | |
334 | some input. But if the @code{O_NONBLOCK} flag is set for the file | |
335 | (@pxref{File Status Flags}), @code{read} returns immediately without | |
336 | reading any data, and reports this error. | |
337 | ||
338 | @strong{Compatibility Note:} Most versions of BSD Unix use a different | |
1f77f049 | 339 | error code for this: @code{EWOULDBLOCK}. In @theglibc{}, |
28f540f4 RM |
340 | @code{EWOULDBLOCK} is an alias for @code{EAGAIN}, so it doesn't matter |
341 | which name you use. | |
342 | ||
343 | On some systems, reading a large amount of data from a character special | |
344 | file can also fail with @code{EAGAIN} if the kernel cannot find enough | |
345 | physical memory to lock down the user's pages. This is limited to | |
346 | devices that transfer with direct memory access into the user's memory, | |
347 | which means it does not include terminals, since they always use | |
a7a93d50 JM |
348 | separate buffers inside the kernel. This problem never happens on |
349 | @gnuhurdsystems{}. | |
28f540f4 RM |
350 | |
351 | Any condition that could result in @code{EAGAIN} can instead result in a | |
352 | successful @code{read} which returns fewer bytes than requested. | |
353 | Calling @code{read} again immediately would result in @code{EAGAIN}. | |
354 | ||
355 | @item EBADF | |
356 | The @var{filedes} argument is not a valid file descriptor, | |
357 | or is not open for reading. | |
358 | ||
359 | @item EINTR | |
360 | @code{read} was interrupted by a signal while it was waiting for input. | |
9739d2d5 | 361 | @xref{Interrupted Primitives}. A signal will not necessarily cause |
28f540f4 RM |
362 | @code{read} to return @code{EINTR}; it may instead result in a |
363 | successful @code{read} which returns fewer bytes than requested. | |
364 | ||
365 | @item EIO | |
366 | For many devices, and for disk files, this error code indicates | |
367 | a hardware error. | |
368 | ||
369 | @code{EIO} also occurs when a background process tries to read from the | |
370 | controlling terminal, and the normal action of stopping the process by | |
371 | sending it a @code{SIGTTIN} signal isn't working. This might happen if | |
04b9968b | 372 | the signal is being blocked or ignored, or because the process group is |
28f540f4 RM |
373 | orphaned. @xref{Job Control}, for more information about job control, |
374 | and @ref{Signal Handling}, for information about signals. | |
7e583a52 RM |
375 | |
376 | @item EINVAL | |
377 | In some systems, when reading from a character or block device, position | |
378 | and size offsets must be aligned to a particular block size. This error | |
379 | indicates that the offsets were not properly aligned. | |
28f540f4 RM |
380 | @end table |
381 | ||
b07d03e0 UD |
382 | Please note that there is no function named @code{read64}. This is not |
383 | necessary since this function does not directly modify or handle the | |
384 | possibly wide file offset. Since the kernel handles this state | |
04b9968b | 385 | internally, the @code{read} function can be used for all cases. |
b07d03e0 | 386 | |
04b9968b | 387 | This function is a cancellation point in multi-threaded programs. This |
dfd2257a UD |
388 | is a problem if the thread allocates some resources (like memory, file |
389 | descriptors, semaphores or whatever) at the time @code{read} is | |
19e4c7dd | 390 | called. If the thread gets canceled these resources stay allocated |
04b9968b UD |
391 | until the program ends. To avoid this, calls to @code{read} should be |
392 | protected using cancellation handlers. | |
dfd2257a UD |
393 | @c ref pthread_cleanup_push / pthread_cleanup_pop |
394 | ||
28f540f4 RM |
395 | The @code{read} function is the underlying primitive for all of the |
396 | functions that read from streams, such as @code{fgetc}. | |
397 | @end deftypefun | |
398 | ||
a5a0310d | 399 | @deftypefun ssize_t pread (int @var{filedes}, void *@var{buffer}, size_t @var{size}, off_t @var{offset}) |
d08a7e4c | 400 | @standards{Unix98, unistd.h} |
2cc3615c AO |
401 | @safety{@prelim{}@mtsafe{}@assafe{}@acsafe{}} |
402 | @c This is usually a safe syscall. The sysdeps/posix fallback emulation | |
403 | @c is not MT-Safe because it uses lseek, read and lseek back, but is it | |
404 | @c used anywhere? | |
a5a0310d | 405 | The @code{pread} function is similar to the @code{read} function. The |
04b9968b UD |
406 | first three arguments are identical, and the return values and error |
407 | codes also correspond. | |
a5a0310d UD |
408 | |
409 | The difference is the fourth argument and its handling. The data block | |
410 | is not read from the current position of the file descriptor | |
411 | @code{filedes}. Instead the data is read from the file starting at | |
412 | position @var{offset}. The position of the file descriptor itself is | |
04b9968b | 413 | not affected by the operation. The value is the same as before the call. |
a5a0310d | 414 | |
b07d03e0 UD |
415 | When the source file is compiled with @code{_FILE_OFFSET_BITS == 64} the |
416 | @code{pread} function is in fact @code{pread64} and the type | |
04b9968b | 417 | @code{off_t} has 64 bits, which makes it possible to handle files up to |
9ceeb279 | 418 | @twoexp{63} bytes in length. |
b07d03e0 | 419 | |
a5a0310d UD |
420 | The return value of @code{pread} describes the number of bytes read. |
421 | In the error case it returns @math{-1} like @code{read} does and the | |
04b9968b UD |
422 | error codes are also the same, with these additions: |
423 | ||
a5a0310d UD |
424 | @table @code |
425 | @item EINVAL | |
426 | The value given for @var{offset} is negative and therefore illegal. | |
427 | ||
428 | @item ESPIPE | |
9739d2d5 | 429 | The file descriptor @var{filedes} is associated with a pipe or a FIFO and |
a5a0310d UD |
430 | this device does not allow positioning of the file pointer. |
431 | @end table | |
432 | ||
433 | The function is an extension defined in the Unix Single Specification | |
434 | version 2. | |
435 | @end deftypefun | |
436 | ||
b07d03e0 | 437 | @deftypefun ssize_t pread64 (int @var{filedes}, void *@var{buffer}, size_t @var{size}, off64_t @var{offset}) |
d08a7e4c | 438 | @standards{Unix98, unistd.h} |
2cc3615c AO |
439 | @safety{@prelim{}@mtsafe{}@assafe{}@acsafe{}} |
440 | @c This is usually a safe syscall. The sysdeps/posix fallback emulation | |
441 | @c is not MT-Safe because it uses lseek64, read and lseek64 back, but is | |
442 | @c it used anywhere? | |
b07d03e0 UD |
443 | This function is similar to the @code{pread} function. The difference |
444 | is that the @var{offset} parameter is of type @code{off64_t} instead of | |
04b9968b | 445 | @code{off_t} which makes it possible on 32 bit machines to address |
9ceeb279 | 446 | files larger than @twoexp{31} bytes and up to @twoexp{63} bytes. The |
b07d03e0 UD |
447 | file descriptor @code{filedes} must be opened using @code{open64} since |
448 | otherwise the large offsets possible with @code{off64_t} will lead to | |
449 | errors with a descriptor in small file mode. | |
450 | ||
c756c71c | 451 | When the source file is compiled with @code{_FILE_OFFSET_BITS == 64} on a |
04b9968b UD |
452 | 32 bit machine this function is actually available under the name |
453 | @code{pread} and so transparently replaces the 32 bit interface. | |
b07d03e0 UD |
454 | @end deftypefun |
455 | ||
28f540f4 | 456 | @cindex writing to a file descriptor |
28f540f4 | 457 | @deftypefun ssize_t write (int @var{filedes}, const void *@var{buffer}, size_t @var{size}) |
d08a7e4c | 458 | @standards{POSIX.1, unistd.h} |
2cc3615c | 459 | @safety{@prelim{}@mtsafe{}@assafe{}@acsafe{}} |
0c6891a0 CD |
460 | @c Some say write is thread-unsafe on Linux without O_APPEND. In the VFS layer |
461 | @c the vfs_write() does no locking around the acquisition of a file offset and | |
462 | @c therefore multiple threads / kernel tasks may race and get the same offset | |
463 | @c resulting in data loss. | |
464 | @c | |
465 | @c See: | |
466 | @c http://thread.gmane.org/gmane.linux.kernel/397980 | |
467 | @c http://lwn.net/Articles/180387/ | |
468 | @c | |
469 | @c The counter argument is that POSIX only says that the write starts at the | |
470 | @c file position and that the file position is updated *before* the function | |
471 | @c returns. What that really means is that any expectation of atomic writes is | |
472 | @c strictly an invention of the interpretation of the reader. Data loss could | |
473 | @c happen if two threads start the write at the same time. Only writes that | |
474 | @c come after the return of another write are guaranteed to follow the other | |
475 | @c write. | |
476 | @c | |
477 | @c The other side of the coin is that POSIX goes on further to say in | |
478 | @c "2.9.7 Thread Interactions with Regular File Operations" that threads | |
479 | @c should never see interleaving sets of file operations, but it is insane | |
480 | @c to do anything like that because it kills performance, so you don't get | |
481 | @c those guarantees in Linux. | |
482 | @c | |
483 | @c So we mark it thread safe, it doesn't blow up, but you might loose | |
484 | @c data, and we don't strictly meet the POSIX requirements. | |
a2887bdb CD |
485 | @c |
486 | @c The fix for file offsets racing was merged in 3.14, the commits were: | |
487 | @c 9c225f2655e36a470c4f58dbbc99244c5fc7f2d4, and | |
488 | @c d7a15f8d0777955986a2ab00ab181795cab14b01. Therefore after Linux 3.14 you | |
489 | @c should get mostly MT-safe writes. | |
28f540f4 RM |
490 | The @code{write} function writes up to @var{size} bytes from |
491 | @var{buffer} to the file with descriptor @var{filedes}. The data in | |
492 | @var{buffer} is not necessarily a character string and a null character is | |
493 | output like any other character. | |
494 | ||
495 | The return value is the number of bytes actually written. This may be | |
496 | @var{size}, but can always be smaller. Your program should always call | |
497 | @code{write} in a loop, iterating until all the data is written. | |
498 | ||
499 | Once @code{write} returns, the data is enqueued to be written and can be | |
500 | read back right away, but it is not necessarily written out to permanent | |
501 | storage immediately. You can use @code{fsync} when you need to be sure | |
502 | your data has been permanently stored before continuing. (It is more | |
503 | efficient for the system to batch up consecutive writes and do them all | |
504 | at once when convenient. Normally they will always be written to disk | |
a5a0310d UD |
505 | within a minute or less.) Modern systems provide another function |
506 | @code{fdatasync} which guarantees integrity only for the file data and | |
507 | is therefore faster. | |
508 | @c !!! xref fsync, fdatasync | |
2c6fe0bd | 509 | You can use the @code{O_FSYNC} open mode to make @code{write} always |
28f540f4 RM |
510 | store the data to disk before returning; @pxref{Operating Modes}. |
511 | ||
07435eb4 | 512 | In the case of an error, @code{write} returns @math{-1}. The following |
28f540f4 RM |
513 | @code{errno} error conditions are defined for this function: |
514 | ||
515 | @table @code | |
516 | @item EAGAIN | |
517 | Normally, @code{write} blocks until the write operation is complete. | |
518 | But if the @code{O_NONBLOCK} flag is set for the file (@pxref{Control | |
04b9968b | 519 | Operations}), it returns immediately without writing any data and |
28f540f4 RM |
520 | reports this error. An example of a situation that might cause the |
521 | process to block on output is writing to a terminal device that supports | |
522 | flow control, where output has been suspended by receipt of a STOP | |
523 | character. | |
524 | ||
525 | @strong{Compatibility Note:} Most versions of BSD Unix use a different | |
1f77f049 | 526 | error code for this: @code{EWOULDBLOCK}. In @theglibc{}, |
28f540f4 RM |
527 | @code{EWOULDBLOCK} is an alias for @code{EAGAIN}, so it doesn't matter |
528 | which name you use. | |
529 | ||
530 | On some systems, writing a large amount of data from a character special | |
531 | file can also fail with @code{EAGAIN} if the kernel cannot find enough | |
532 | physical memory to lock down the user's pages. This is limited to | |
533 | devices that transfer with direct memory access into the user's memory, | |
534 | which means it does not include terminals, since they always use | |
a7a93d50 JM |
535 | separate buffers inside the kernel. This problem does not arise on |
536 | @gnuhurdsystems{}. | |
28f540f4 RM |
537 | |
538 | @item EBADF | |
539 | The @var{filedes} argument is not a valid file descriptor, | |
540 | or is not open for writing. | |
541 | ||
542 | @item EFBIG | |
543 | The size of the file would become larger than the implementation can support. | |
544 | ||
545 | @item EINTR | |
546 | The @code{write} operation was interrupted by a signal while it was | |
04b9968b | 547 | blocked waiting for completion. A signal will not necessarily cause |
28f540f4 RM |
548 | @code{write} to return @code{EINTR}; it may instead result in a |
549 | successful @code{write} which writes fewer bytes than requested. | |
550 | @xref{Interrupted Primitives}. | |
551 | ||
552 | @item EIO | |
553 | For many devices, and for disk files, this error code indicates | |
554 | a hardware error. | |
555 | ||
556 | @item ENOSPC | |
557 | The device containing the file is full. | |
558 | ||
559 | @item EPIPE | |
560 | This error is returned when you try to write to a pipe or FIFO that | |
561 | isn't open for reading by any process. When this happens, a @code{SIGPIPE} | |
562 | signal is also sent to the process; see @ref{Signal Handling}. | |
7e583a52 RM |
563 | |
564 | @item EINVAL | |
565 | In some systems, when writing to a character or block device, position | |
566 | and size offsets must be aligned to a particular block size. This error | |
567 | indicates that the offsets were not properly aligned. | |
28f540f4 RM |
568 | @end table |
569 | ||
570 | Unless you have arranged to prevent @code{EINTR} failures, you should | |
571 | check @code{errno} after each failing call to @code{write}, and if the | |
572 | error was @code{EINTR}, you should simply repeat the call. | |
573 | @xref{Interrupted Primitives}. The easy way to do this is with the | |
574 | macro @code{TEMP_FAILURE_RETRY}, as follows: | |
575 | ||
576 | @smallexample | |
577 | nbytes = TEMP_FAILURE_RETRY (write (desc, buffer, count)); | |
578 | @end smallexample | |
579 | ||
b07d03e0 UD |
580 | Please note that there is no function named @code{write64}. This is not |
581 | necessary since this function does not directly modify or handle the | |
582 | possibly wide file offset. Since the kernel handles this state | |
583 | internally the @code{write} function can be used for all cases. | |
584 | ||
04b9968b | 585 | This function is a cancellation point in multi-threaded programs. This |
dfd2257a UD |
586 | is a problem if the thread allocates some resources (like memory, file |
587 | descriptors, semaphores or whatever) at the time @code{write} is | |
19e4c7dd | 588 | called. If the thread gets canceled these resources stay allocated |
04b9968b UD |
589 | until the program ends. To avoid this, calls to @code{write} should be |
590 | protected using cancellation handlers. | |
dfd2257a UD |
591 | @c ref pthread_cleanup_push / pthread_cleanup_pop |
592 | ||
28f540f4 RM |
593 | The @code{write} function is the underlying primitive for all of the |
594 | functions that write to streams, such as @code{fputc}. | |
595 | @end deftypefun | |
596 | ||
a5a0310d | 597 | @deftypefun ssize_t pwrite (int @var{filedes}, const void *@var{buffer}, size_t @var{size}, off_t @var{offset}) |
d08a7e4c | 598 | @standards{Unix98, unistd.h} |
2cc3615c AO |
599 | @safety{@prelim{}@mtsafe{}@assafe{}@acsafe{}} |
600 | @c This is usually a safe syscall. The sysdeps/posix fallback emulation | |
601 | @c is not MT-Safe because it uses lseek, write and lseek back, but is it | |
602 | @c used anywhere? | |
a5a0310d | 603 | The @code{pwrite} function is similar to the @code{write} function. The |
04b9968b UD |
604 | first three arguments are identical, and the return values and error codes |
605 | also correspond. | |
a5a0310d UD |
606 | |
607 | The difference is the fourth argument and its handling. The data block | |
608 | is not written to the current position of the file descriptor | |
609 | @code{filedes}. Instead the data is written to the file starting at | |
610 | position @var{offset}. The position of the file descriptor itself is | |
04b9968b | 611 | not affected by the operation. The value is the same as before the call. |
a5a0310d | 612 | |
717da4b3 AZ |
613 | However, on Linux, if a file is opened with @code{O_APPEND}, @code{pwrite} |
614 | appends data to the end of the file, regardless of the value of | |
615 | @code{offset}. | |
616 | ||
b07d03e0 UD |
617 | When the source file is compiled with @code{_FILE_OFFSET_BITS == 64} the |
618 | @code{pwrite} function is in fact @code{pwrite64} and the type | |
04b9968b | 619 | @code{off_t} has 64 bits, which makes it possible to handle files up to |
9ceeb279 | 620 | @twoexp{63} bytes in length. |
b07d03e0 | 621 | |
a5a0310d UD |
622 | The return value of @code{pwrite} describes the number of written bytes. |
623 | In the error case it returns @math{-1} like @code{write} does and the | |
04b9968b UD |
624 | error codes are also the same, with these additions: |
625 | ||
a5a0310d UD |
626 | @table @code |
627 | @item EINVAL | |
628 | The value given for @var{offset} is negative and therefore illegal. | |
629 | ||
630 | @item ESPIPE | |
04b9968b | 631 | The file descriptor @var{filedes} is associated with a pipe or a FIFO and |
a5a0310d UD |
632 | this device does not allow positioning of the file pointer. |
633 | @end table | |
634 | ||
635 | The function is an extension defined in the Unix Single Specification | |
636 | version 2. | |
637 | @end deftypefun | |
638 | ||
b07d03e0 | 639 | @deftypefun ssize_t pwrite64 (int @var{filedes}, const void *@var{buffer}, size_t @var{size}, off64_t @var{offset}) |
d08a7e4c | 640 | @standards{Unix98, unistd.h} |
2cc3615c AO |
641 | @safety{@prelim{}@mtsafe{}@assafe{}@acsafe{}} |
642 | @c This is usually a safe syscall. The sysdeps/posix fallback emulation | |
643 | @c is not MT-Safe because it uses lseek64, write and lseek64 back, but | |
644 | @c is it used anywhere? | |
b07d03e0 UD |
645 | This function is similar to the @code{pwrite} function. The difference |
646 | is that the @var{offset} parameter is of type @code{off64_t} instead of | |
04b9968b | 647 | @code{off_t} which makes it possible on 32 bit machines to address |
9ceeb279 | 648 | files larger than @twoexp{31} bytes and up to @twoexp{63} bytes. The |
b07d03e0 UD |
649 | file descriptor @code{filedes} must be opened using @code{open64} since |
650 | otherwise the large offsets possible with @code{off64_t} will lead to | |
651 | errors with a descriptor in small file mode. | |
652 | ||
c756c71c | 653 | When the source file is compiled using @code{_FILE_OFFSET_BITS == 64} on a |
04b9968b UD |
654 | 32 bit machine this function is actually available under the name |
655 | @code{pwrite} and so transparently replaces the 32 bit interface. | |
b07d03e0 | 656 | @end deftypefun |
717da4b3 | 657 | |
28f540f4 RM |
658 | @node File Position Primitive |
659 | @section Setting the File Position of a Descriptor | |
660 | ||
661 | Just as you can set the file position of a stream with @code{fseek}, you | |
662 | can set the file position of a descriptor with @code{lseek}. This | |
663 | specifies the position in the file for the next @code{read} or | |
664 | @code{write} operation. @xref{File Positioning}, for more information | |
665 | on the file position and what it means. | |
666 | ||
667 | To read the current file position value from a descriptor, use | |
668 | @code{lseek (@var{desc}, 0, SEEK_CUR)}. | |
669 | ||
670 | @cindex file positioning on a file descriptor | |
671 | @cindex positioning a file descriptor | |
672 | @cindex seeking on a file descriptor | |
28f540f4 | 673 | @deftypefun off_t lseek (int @var{filedes}, off_t @var{offset}, int @var{whence}) |
d08a7e4c | 674 | @standards{POSIX.1, unistd.h} |
2cc3615c | 675 | @safety{@prelim{}@mtsafe{}@assafe{}@acsafe{}} |
28f540f4 RM |
676 | The @code{lseek} function is used to change the file position of the |
677 | file with descriptor @var{filedes}. | |
678 | ||
679 | The @var{whence} argument specifies how the @var{offset} should be | |
04b9968b UD |
680 | interpreted, in the same way as for the @code{fseek} function, and it must |
681 | be one of the symbolic constants @code{SEEK_SET}, @code{SEEK_CUR}, or | |
28f540f4 RM |
682 | @code{SEEK_END}. |
683 | ||
2fe82ca6 | 684 | @vtable @code |
28f540f4 | 685 | @item SEEK_SET |
4dad7bab | 686 | Specifies that @var{offset} is a count of characters from the beginning |
28f540f4 RM |
687 | of the file. |
688 | ||
689 | @item SEEK_CUR | |
4dad7bab | 690 | Specifies that @var{offset} is a count of characters from the current |
28f540f4 RM |
691 | file position. This count may be positive or negative. |
692 | ||
693 | @item SEEK_END | |
4dad7bab | 694 | Specifies that @var{offset} is a count of characters from the end of |
28f540f4 RM |
695 | the file. A negative count specifies a position within the current |
696 | extent of the file; a positive count specifies a position past the | |
2c6fe0bd | 697 | current end. If you set the position past the current end, and |
28f540f4 | 698 | actually write data, you will extend the file with zeros up to that |
336dfb2d | 699 | position. |
2fe82ca6 | 700 | @end vtable |
28f540f4 RM |
701 | |
702 | The return value from @code{lseek} is normally the resulting file | |
703 | position, measured in bytes from the beginning of the file. | |
704 | You can use this feature together with @code{SEEK_CUR} to read the | |
705 | current file position. | |
706 | ||
707 | If you want to append to the file, setting the file position to the | |
708 | current end of file with @code{SEEK_END} is not sufficient. Another | |
709 | process may write more data after you seek but before you write, | |
710 | extending the file so the position you write onto clobbers their data. | |
711 | Instead, use the @code{O_APPEND} operating mode; @pxref{Operating Modes}. | |
712 | ||
713 | You can set the file position past the current end of the file. This | |
714 | does not by itself make the file longer; @code{lseek} never changes the | |
715 | file. But subsequent output at that position will extend the file. | |
716 | Characters between the previous end of file and the new position are | |
717 | filled with zeros. Extending the file in this way can create a | |
718 | ``hole'': the blocks of zeros are not actually allocated on disk, so the | |
78759725 | 719 | file takes up less space than it appears to; it is then called a |
28f540f4 RM |
720 | ``sparse file''. |
721 | @cindex sparse files | |
722 | @cindex holes in files | |
723 | ||
724 | If the file position cannot be changed, or the operation is in some way | |
07435eb4 | 725 | invalid, @code{lseek} returns a value of @math{-1}. The following |
28f540f4 RM |
726 | @code{errno} error conditions are defined for this function: |
727 | ||
728 | @table @code | |
729 | @item EBADF | |
730 | The @var{filedes} is not a valid file descriptor. | |
731 | ||
732 | @item EINVAL | |
733 | The @var{whence} argument value is not valid, or the resulting | |
734 | file offset is not valid. A file offset is invalid. | |
735 | ||
736 | @item ESPIPE | |
737 | The @var{filedes} corresponds to an object that cannot be positioned, | |
738 | such as a pipe, FIFO or terminal device. (POSIX.1 specifies this error | |
a7a93d50 | 739 | only for pipes and FIFOs, but on @gnusystems{}, you always get |
28f540f4 RM |
740 | @code{ESPIPE} if the object is not seekable.) |
741 | @end table | |
742 | ||
b07d03e0 UD |
743 | When the source file is compiled with @code{_FILE_OFFSET_BITS == 64} the |
744 | @code{lseek} function is in fact @code{lseek64} and the type | |
745 | @code{off_t} has 64 bits which makes it possible to handle files up to | |
9ceeb279 | 746 | @twoexp{63} bytes in length. |
b07d03e0 | 747 | |
04b9968b | 748 | This function is a cancellation point in multi-threaded programs. This |
dfd2257a UD |
749 | is a problem if the thread allocates some resources (like memory, file |
750 | descriptors, semaphores or whatever) at the time @code{lseek} is | |
19e4c7dd | 751 | called. If the thread gets canceled these resources stay allocated |
dfd2257a | 752 | until the program ends. To avoid this calls to @code{lseek} should be |
04b9968b | 753 | protected using cancellation handlers. |
dfd2257a UD |
754 | @c ref pthread_cleanup_push / pthread_cleanup_pop |
755 | ||
28f540f4 | 756 | The @code{lseek} function is the underlying primitive for the |
dfd2257a UD |
757 | @code{fseek}, @code{fseeko}, @code{ftell}, @code{ftello} and |
758 | @code{rewind} functions, which operate on streams instead of file | |
759 | descriptors. | |
28f540f4 RM |
760 | @end deftypefun |
761 | ||
b07d03e0 | 762 | @deftypefun off64_t lseek64 (int @var{filedes}, off64_t @var{offset}, int @var{whence}) |
d08a7e4c | 763 | @standards{Unix98, unistd.h} |
2cc3615c | 764 | @safety{@prelim{}@mtsafe{}@assafe{}@acsafe{}} |
b07d03e0 UD |
765 | This function is similar to the @code{lseek} function. The difference |
766 | is that the @var{offset} parameter is of type @code{off64_t} instead of | |
04b9968b | 767 | @code{off_t} which makes it possible on 32 bit machines to address |
9ceeb279 | 768 | files larger than @twoexp{31} bytes and up to @twoexp{63} bytes. The |
b07d03e0 UD |
769 | file descriptor @code{filedes} must be opened using @code{open64} since |
770 | otherwise the large offsets possible with @code{off64_t} will lead to | |
771 | errors with a descriptor in small file mode. | |
772 | ||
c756c71c | 773 | When the source file is compiled with @code{_FILE_OFFSET_BITS == 64} on a |
b07d03e0 | 774 | 32 bits machine this function is actually available under the name |
04b9968b | 775 | @code{lseek} and so transparently replaces the 32 bit interface. |
b07d03e0 UD |
776 | @end deftypefun |
777 | ||
28f540f4 | 778 | You can have multiple descriptors for the same file if you open the file |
2c6fe0bd | 779 | more than once, or if you duplicate a descriptor with @code{dup}. |
28f540f4 RM |
780 | Descriptors that come from separate calls to @code{open} have independent |
781 | file positions; using @code{lseek} on one descriptor has no effect on the | |
2c6fe0bd | 782 | other. For example, |
28f540f4 RM |
783 | |
784 | @smallexample | |
785 | @group | |
786 | @{ | |
787 | int d1, d2; | |
788 | char buf[4]; | |
789 | d1 = open ("foo", O_RDONLY); | |
790 | d2 = open ("foo", O_RDONLY); | |
791 | lseek (d1, 1024, SEEK_SET); | |
792 | read (d2, buf, 4); | |
793 | @} | |
794 | @end group | |
795 | @end smallexample | |
796 | ||
797 | @noindent | |
798 | will read the first four characters of the file @file{foo}. (The | |
799 | error-checking code necessary for a real program has been omitted here | |
800 | for brevity.) | |
801 | ||
802 | By contrast, descriptors made by duplication share a common file | |
803 | position with the original descriptor that was duplicated. Anything | |
804 | which alters the file position of one of the duplicates, including | |
805 | reading or writing data, affects all of them alike. Thus, for example, | |
806 | ||
807 | @smallexample | |
808 | @{ | |
809 | int d1, d2, d3; | |
810 | char buf1[4], buf2[4]; | |
811 | d1 = open ("foo", O_RDONLY); | |
812 | d2 = dup (d1); | |
813 | d3 = dup (d2); | |
814 | lseek (d3, 1024, SEEK_SET); | |
815 | read (d1, buf1, 4); | |
816 | read (d2, buf2, 4); | |
817 | @} | |
818 | @end smallexample | |
819 | ||
820 | @noindent | |
821 | will read four characters starting with the 1024'th character of | |
822 | @file{foo}, and then four more characters starting with the 1028'th | |
823 | character. | |
824 | ||
28f540f4 | 825 | @deftp {Data Type} off_t |
d08a7e4c | 826 | @standards{POSIX.1, sys/types.h} |
07e12bb3 JM |
827 | This is a signed integer type used to represent file sizes. In |
828 | @theglibc{}, this type is no narrower than @code{int}. | |
a3a4a74e UD |
829 | |
830 | If the source is compiled with @code{_FILE_OFFSET_BITS == 64} this type | |
831 | is transparently replaced by @code{off64_t}. | |
28f540f4 RM |
832 | @end deftp |
833 | ||
b07d03e0 | 834 | @deftp {Data Type} off64_t |
d08a7e4c | 835 | @standards{Unix98, sys/types.h} |
b07d03e0 | 836 | This type is used similar to @code{off_t}. The difference is that even |
04b9968b | 837 | on 32 bit machines, where the @code{off_t} type would have 32 bits, |
b07d03e0 | 838 | @code{off64_t} has 64 bits and so is able to address files up to |
9ceeb279 | 839 | @twoexp{63} bytes in length. |
a3a4a74e UD |
840 | |
841 | When compiling with @code{_FILE_OFFSET_BITS == 64} this type is | |
842 | available under the name @code{off_t}. | |
b07d03e0 UD |
843 | @end deftp |
844 | ||
28f540f4 RM |
845 | These aliases for the @samp{SEEK_@dots{}} constants exist for the sake |
846 | of compatibility with older BSD systems. They are defined in two | |
847 | different header files: @file{fcntl.h} and @file{sys/file.h}. | |
848 | ||
2fe82ca6 | 849 | @vtable @code |
28f540f4 RM |
850 | @item L_SET |
851 | An alias for @code{SEEK_SET}. | |
852 | ||
853 | @item L_INCR | |
854 | An alias for @code{SEEK_CUR}. | |
855 | ||
856 | @item L_XTND | |
857 | An alias for @code{SEEK_END}. | |
2fe82ca6 | 858 | @end vtable |
28f540f4 RM |
859 | |
860 | @node Descriptors and Streams | |
861 | @section Descriptors and Streams | |
862 | @cindex streams, and file descriptors | |
863 | @cindex converting file descriptor to stream | |
864 | @cindex extracting file descriptor from stream | |
865 | ||
866 | Given an open file descriptor, you can create a stream for it with the | |
867 | @code{fdopen} function. You can get the underlying file descriptor for | |
868 | an existing stream with the @code{fileno} function. These functions are | |
869 | declared in the header file @file{stdio.h}. | |
870 | @pindex stdio.h | |
871 | ||
28f540f4 | 872 | @deftypefun {FILE *} fdopen (int @var{filedes}, const char *@var{opentype}) |
d08a7e4c | 873 | @standards{POSIX.1, stdio.h} |
2cc3615c | 874 | @safety{@prelim{}@mtsafe{}@asunsafe{@ascuheap{} @asulock{}}@acunsafe{@acsmem{} @aculock{}}} |
28f540f4 RM |
875 | The @code{fdopen} function returns a new stream for the file descriptor |
876 | @var{filedes}. | |
877 | ||
878 | The @var{opentype} argument is interpreted in the same way as for the | |
879 | @code{fopen} function (@pxref{Opening Streams}), except that | |
a7a93d50 | 880 | the @samp{b} option is not permitted; this is because @gnusystems{} make no |
28f540f4 | 881 | distinction between text and binary files. Also, @code{"w"} and |
04b9968b | 882 | @code{"w+"} do not cause truncation of the file; these have an effect only |
28f540f4 RM |
883 | when opening a file, and in this case the file has already been opened. |
884 | You must make sure that the @var{opentype} argument matches the actual | |
885 | mode of the open file descriptor. | |
886 | ||
887 | The return value is the new stream. If the stream cannot be created | |
888 | (for example, if the modes for the file indicated by the file descriptor | |
889 | do not permit the access specified by the @var{opentype} argument), a | |
890 | null pointer is returned instead. | |
891 | ||
892 | In some other systems, @code{fdopen} may fail to detect that the modes | |
9739d2d5 | 893 | for file descriptors do not permit the access specified by |
1f77f049 | 894 | @code{opentype}. @Theglibc{} always checks for this. |
28f540f4 RM |
895 | @end deftypefun |
896 | ||
897 | For an example showing the use of the @code{fdopen} function, | |
898 | see @ref{Creating a Pipe}. | |
899 | ||
28f540f4 | 900 | @deftypefun int fileno (FILE *@var{stream}) |
d08a7e4c | 901 | @standards{POSIX.1, stdio.h} |
2cc3615c | 902 | @safety{@prelim{}@mtsafe{}@assafe{}@acsafe{}} |
28f540f4 RM |
903 | This function returns the file descriptor associated with the stream |
904 | @var{stream}. If an error is detected (for example, if the @var{stream} | |
905 | is not valid) or if @var{stream} does not do I/O to a file, | |
07435eb4 | 906 | @code{fileno} returns @math{-1}. |
28f540f4 RM |
907 | @end deftypefun |
908 | ||
7b4161bb | 909 | @deftypefun int fileno_unlocked (FILE *@var{stream}) |
d08a7e4c | 910 | @standards{GNU, stdio.h} |
2cc3615c | 911 | @safety{@prelim{}@mtsafe{}@assafe{}@acsafe{}} |
7b4161bb UD |
912 | The @code{fileno_unlocked} function is equivalent to the @code{fileno} |
913 | function except that it does not implicitly lock the stream if the state | |
914 | is @code{FSETLOCKING_INTERNAL}. | |
915 | ||
916 | This function is a GNU extension. | |
917 | @end deftypefun | |
918 | ||
28f540f4 RM |
919 | @cindex standard file descriptors |
920 | @cindex file descriptors, standard | |
921 | There are also symbolic constants defined in @file{unistd.h} for the | |
922 | file descriptors belonging to the standard streams @code{stdin}, | |
923 | @code{stdout}, and @code{stderr}; see @ref{Standard Streams}. | |
924 | @pindex unistd.h | |
925 | ||
2fe82ca6 | 926 | @vtable @code |
28f540f4 | 927 | @item STDIN_FILENO |
d08a7e4c | 928 | @standards{POSIX.1, unistd.h} |
28f540f4 RM |
929 | This macro has value @code{0}, which is the file descriptor for |
930 | standard input. | |
931 | @cindex standard input file descriptor | |
932 | ||
28f540f4 | 933 | @item STDOUT_FILENO |
d08a7e4c | 934 | @standards{POSIX.1, unistd.h} |
28f540f4 RM |
935 | This macro has value @code{1}, which is the file descriptor for |
936 | standard output. | |
937 | @cindex standard output file descriptor | |
938 | ||
28f540f4 | 939 | @item STDERR_FILENO |
d08a7e4c | 940 | @standards{POSIX.1, unistd.h} |
28f540f4 RM |
941 | This macro has value @code{2}, which is the file descriptor for |
942 | standard error output. | |
2fe82ca6 | 943 | @end vtable |
28f540f4 RM |
944 | @cindex standard error file descriptor |
945 | ||
946 | @node Stream/Descriptor Precautions | |
947 | @section Dangers of Mixing Streams and Descriptors | |
948 | @cindex channels | |
949 | @cindex streams and descriptors | |
950 | @cindex descriptors and streams | |
951 | @cindex mixing descriptors and streams | |
952 | ||
953 | You can have multiple file descriptors and streams (let's call both | |
954 | streams and descriptors ``channels'' for short) connected to the same | |
955 | file, but you must take care to avoid confusion between channels. There | |
956 | are two cases to consider: @dfn{linked} channels that share a single | |
957 | file position value, and @dfn{independent} channels that have their own | |
958 | file positions. | |
959 | ||
960 | It's best to use just one channel in your program for actual data | |
961 | transfer to any given file, except when all the access is for input. | |
962 | For example, if you open a pipe (something you can only do at the file | |
963 | descriptor level), either do all I/O with the descriptor, or construct a | |
964 | stream from the descriptor with @code{fdopen} and then do all I/O with | |
965 | the stream. | |
966 | ||
967 | @menu | |
968 | * Linked Channels:: Dealing with channels sharing a file position. | |
969 | * Independent Channels:: Dealing with separately opened, unlinked channels. | |
2c6fe0bd | 970 | * Cleaning Streams:: Cleaning a stream makes it safe to use |
28f540f4 RM |
971 | another channel. |
972 | @end menu | |
973 | ||
974 | @node Linked Channels | |
975 | @subsection Linked Channels | |
976 | @cindex linked channels | |
977 | ||
978 | Channels that come from a single opening share the same file position; | |
979 | we call them @dfn{linked} channels. Linked channels result when you | |
980 | make a stream from a descriptor using @code{fdopen}, when you get a | |
981 | descriptor from a stream with @code{fileno}, when you copy a descriptor | |
982 | with @code{dup} or @code{dup2}, and when descriptors are inherited | |
983 | during @code{fork}. For files that don't support random access, such as | |
984 | terminals and pipes, @emph{all} channels are effectively linked. On | |
985 | random-access files, all append-type output streams are effectively | |
986 | linked to each other. | |
987 | ||
988 | @cindex cleaning up a stream | |
0295d266 UD |
989 | If you have been using a stream for I/O (or have just opened the stream), |
990 | and you want to do I/O using | |
28f540f4 RM |
991 | another channel (either a stream or a descriptor) that is linked to it, |
992 | you must first @dfn{clean up} the stream that you have been using. | |
993 | @xref{Cleaning Streams}. | |
994 | ||
995 | Terminating a process, or executing a new program in the process, | |
996 | destroys all the streams in the process. If descriptors linked to these | |
997 | streams persist in other processes, their file positions become | |
998 | undefined as a result. To prevent this, you must clean up the streams | |
999 | before destroying them. | |
1000 | ||
1001 | @node Independent Channels | |
1002 | @subsection Independent Channels | |
1003 | @cindex independent channels | |
1004 | ||
1005 | When you open channels (streams or descriptors) separately on a seekable | |
1006 | file, each channel has its own file position. These are called | |
1007 | @dfn{independent channels}. | |
1008 | ||
1009 | The system handles each channel independently. Most of the time, this | |
1010 | is quite predictable and natural (especially for input): each channel | |
1011 | can read or write sequentially at its own place in the file. However, | |
1012 | if some of the channels are streams, you must take these precautions: | |
1013 | ||
1014 | @itemize @bullet | |
1015 | @item | |
1016 | You should clean an output stream after use, before doing anything else | |
1017 | that might read or write from the same part of the file. | |
1018 | ||
1019 | @item | |
1020 | You should clean an input stream before reading data that may have been | |
1021 | modified using an independent channel. Otherwise, you might read | |
1022 | obsolete data that had been in the stream's buffer. | |
1023 | @end itemize | |
1024 | ||
1025 | If you do output to one channel at the end of the file, this will | |
1026 | certainly leave the other independent channels positioned somewhere | |
1027 | before the new end. You cannot reliably set their file positions to the | |
1028 | new end of file before writing, because the file can always be extended | |
1029 | by another process between when you set the file position and when you | |
1030 | write the data. Instead, use an append-type descriptor or stream; they | |
1031 | always output at the current end of the file. In order to make the | |
1032 | end-of-file position accurate, you must clean the output channel you | |
1033 | were using, if it is a stream. | |
1034 | ||
1035 | It's impossible for two channels to have separate file pointers for a | |
1036 | file that doesn't support random access. Thus, channels for reading or | |
1037 | writing such files are always linked, never independent. Append-type | |
1038 | channels are also always linked. For these channels, follow the rules | |
1039 | for linked channels; see @ref{Linked Channels}. | |
1040 | ||
1041 | @node Cleaning Streams | |
1042 | @subsection Cleaning Streams | |
1043 | ||
6664049b | 1044 | You can use @code{fflush} to clean a stream in most |
28f540f4 RM |
1045 | cases. |
1046 | ||
6664049b | 1047 | You can skip the @code{fflush} if you know the stream |
28f540f4 RM |
1048 | is already clean. A stream is clean whenever its buffer is empty. For |
1049 | example, an unbuffered stream is always clean. An input stream that is | |
1050 | at end-of-file is clean. A line-buffered stream is clean when the last | |
0295d266 UD |
1051 | character output was a newline. However, a just-opened input stream |
1052 | might not be clean, as its input buffer might not be empty. | |
28f540f4 RM |
1053 | |
1054 | There is one case in which cleaning a stream is impossible on most | |
1055 | systems. This is when the stream is doing input from a file that is not | |
1056 | random-access. Such streams typically read ahead, and when the file is | |
1057 | not random access, there is no way to give back the excess data already | |
1058 | read. When an input stream reads from a random-access file, | |
1059 | @code{fflush} does clean the stream, but leaves the file pointer at an | |
1060 | unpredictable place; you must set the file pointer before doing any | |
6664049b | 1061 | further I/O. |
28f540f4 RM |
1062 | |
1063 | Closing an output-only stream also does @code{fflush}, so this is a | |
6664049b | 1064 | valid way of cleaning an output stream. |
28f540f4 RM |
1065 | |
1066 | You need not clean a stream before using its descriptor for control | |
1067 | operations such as setting terminal modes; these operations don't affect | |
1068 | the file position and are not affected by it. You can use any | |
1069 | descriptor for these operations, and all channels are affected | |
1070 | simultaneously. However, text already ``output'' to a stream but still | |
1071 | buffered by the stream will be subject to the new terminal modes when | |
1072 | subsequently flushed. To make sure ``past'' output is covered by the | |
1073 | terminal settings that were in effect at the time, flush the output | |
1074 | streams for that terminal before setting the modes. @xref{Terminal | |
1075 | Modes}. | |
1076 | ||
07435eb4 UD |
1077 | @node Scatter-Gather |
1078 | @section Fast Scatter-Gather I/O | |
1079 | @cindex scatter-gather | |
1080 | ||
1081 | Some applications may need to read or write data to multiple buffers, | |
04b9968b | 1082 | which are separated in memory. Although this can be done easily enough |
19e4c7dd | 1083 | with multiple calls to @code{read} and @code{write}, it is inefficient |
07435eb4 UD |
1084 | because there is overhead associated with each kernel call. |
1085 | ||
1086 | Instead, many platforms provide special high-speed primitives to perform | |
1f77f049 JM |
1087 | these @dfn{scatter-gather} operations in a single kernel call. @Theglibc{} |
1088 | will provide an emulation on any system that lacks these | |
07435eb4 UD |
1089 | primitives, so they are not a portability threat. They are defined in |
1090 | @code{sys/uio.h}. | |
1091 | ||
1092 | These functions are controlled with arrays of @code{iovec} structures, | |
1093 | which describe the location and size of each buffer. | |
1094 | ||
1095 | @deftp {Data Type} {struct iovec} | |
d08a7e4c | 1096 | @standards{BSD, sys/uio.h} |
07435eb4 | 1097 | |
cf822e3c | 1098 | The @code{iovec} structure describes a buffer. It contains two fields: |
07435eb4 UD |
1099 | |
1100 | @table @code | |
1101 | ||
1102 | @item void *iov_base | |
1103 | Contains the address of a buffer. | |
1104 | ||
1105 | @item size_t iov_len | |
1106 | Contains the length of the buffer. | |
1107 | ||
1108 | @end table | |
1109 | @end deftp | |
1110 | ||
1111 | @deftypefun ssize_t readv (int @var{filedes}, const struct iovec *@var{vector}, int @var{count}) | |
d08a7e4c | 1112 | @standards{BSD, sys/uio.h} |
2cc3615c AO |
1113 | @safety{@prelim{}@mtsafe{}@asunsafe{@ascuheap{}}@acunsafe{@acsmem{}}} |
1114 | @c The fallback sysdeps/posix implementation, used even on GNU/Linux | |
1115 | @c with old kernels that lack a full readv/writev implementation, may | |
1116 | @c malloc the buffer into which data is read, if the total read size is | |
1117 | @c too large for alloca. | |
07435eb4 UD |
1118 | |
1119 | The @code{readv} function reads data from @var{filedes} and scatters it | |
1120 | into the buffers described in @var{vector}, which is taken to be | |
1121 | @var{count} structures long. As each buffer is filled, data is sent to the | |
1122 | next. | |
1123 | ||
1124 | Note that @code{readv} is not guaranteed to fill all the buffers. | |
1125 | It may stop at any point, for the same reasons @code{read} would. | |
1126 | ||
1127 | The return value is a count of bytes (@emph{not} buffers) read, @math{0} | |
1128 | indicating end-of-file, or @math{-1} indicating an error. The possible | |
1129 | errors are the same as in @code{read}. | |
1130 | ||
1131 | @end deftypefun | |
1132 | ||
1133 | @deftypefun ssize_t writev (int @var{filedes}, const struct iovec *@var{vector}, int @var{count}) | |
d08a7e4c | 1134 | @standards{BSD, sys/uio.h} |
2cc3615c AO |
1135 | @safety{@prelim{}@mtsafe{}@asunsafe{@ascuheap{}}@acunsafe{@acsmem{}}} |
1136 | @c The fallback sysdeps/posix implementation, used even on GNU/Linux | |
1137 | @c with old kernels that lack a full readv/writev implementation, may | |
1138 | @c malloc the buffer from which data is written, if the total write size | |
1139 | @c is too large for alloca. | |
07435eb4 UD |
1140 | |
1141 | The @code{writev} function gathers data from the buffers described in | |
1142 | @var{vector}, which is taken to be @var{count} structures long, and writes | |
1143 | them to @code{filedes}. As each buffer is written, it moves on to the | |
1144 | next. | |
1145 | ||
1146 | Like @code{readv}, @code{writev} may stop midstream under the same | |
1147 | conditions @code{write} would. | |
1148 | ||
1149 | The return value is a count of bytes written, or @math{-1} indicating an | |
1150 | error. The possible errors are the same as in @code{write}. | |
1151 | ||
1152 | @end deftypefun | |
1153 | ||
f6e965ee FW |
1154 | @deftypefun ssize_t preadv (int @var{fd}, const struct iovec *@var{iov}, int @var{iovcnt}, off_t @var{offset}) |
1155 | @standards{BSD, sys/uio.h} | |
1156 | @safety{@prelim{}@mtsafe{}@assafe{}@acsafe{}} | |
1157 | @c This is a syscall for Linux 3.2 for all architectures but microblaze | |
1158 | @c (which was added on 3.15). The sysdeps/posix fallback emulation | |
1159 | @c is also MT-Safe since it calls pread, and it is now a syscall on all | |
1160 | @c targets. | |
1161 | ||
1162 | This function is similar to the @code{readv} function, with the difference | |
1163 | it adds an extra @var{offset} parameter of type @code{off_t} similar to | |
1164 | @code{pread}. The data is written to the file starting at position | |
1165 | @var{offset}. The position of the file descriptor itself is not affected | |
1166 | by the operation. The value is the same as before the call. | |
1167 | ||
1168 | When the source file is compiled with @code{_FILE_OFFSET_BITS == 64} the | |
1169 | @code{preadv} function is in fact @code{preadv64} and the type | |
1170 | @code{off_t} has 64 bits, which makes it possible to handle files up to | |
1171 | @twoexp{63} bytes in length. | |
1172 | ||
1173 | The return value is a count of bytes (@emph{not} buffers) read, @math{0} | |
1174 | indicating end-of-file, or @math{-1} indicating an error. The possible | |
1175 | errors are the same as in @code{readv} and @code{pread}. | |
1176 | @end deftypefun | |
1177 | ||
1178 | @deftypefun ssize_t preadv64 (int @var{fd}, const struct iovec *@var{iov}, int @var{iovcnt}, off64_t @var{offset}) | |
1179 | @standards{BSD, unistd.h} | |
1180 | @safety{@prelim{}@mtsafe{}@assafe{}@acsafe{}} | |
1181 | @c This is a syscall for Linux 3.2 for all architectures but microblaze | |
1182 | @c (which was added on 3.15). The sysdeps/posix fallback emulation | |
1183 | @c is also MT-Safe since it calls pread64, and it is now a syscall on all | |
1184 | @c targets. | |
1185 | ||
1186 | This function is similar to the @code{preadv} function with the difference | |
1187 | is that the @var{offset} parameter is of type @code{off64_t} instead of | |
1188 | @code{off_t}. It makes it possible on 32 bit machines to address | |
1189 | files larger than @twoexp{31} bytes and up to @twoexp{63} bytes. The | |
1190 | file descriptor @code{filedes} must be opened using @code{open64} since | |
1191 | otherwise the large offsets possible with @code{off64_t} will lead to | |
1192 | errors with a descriptor in small file mode. | |
1193 | ||
1194 | When the source file is compiled using @code{_FILE_OFFSET_BITS == 64} on a | |
1195 | 32 bit machine this function is actually available under the name | |
1196 | @code{preadv} and so transparently replaces the 32 bit interface. | |
1197 | @end deftypefun | |
1198 | ||
1199 | @deftypefun ssize_t pwritev (int @var{fd}, const struct iovec *@var{iov}, int @var{iovcnt}, off_t @var{offset}) | |
1200 | @standards{BSD, sys/uio.h} | |
1201 | @safety{@prelim{}@mtsafe{}@assafe{}@acsafe{}} | |
1202 | @c This is a syscall for Linux 3.2 for all architectures but microblaze | |
1203 | @c (which was added on 3.15). The sysdeps/posix fallback emulation | |
1204 | @c is also MT-Safe since it calls pwrite, and it is now a syscall on all | |
1205 | @c targets. | |
1206 | ||
1207 | This function is similar to the @code{writev} function, with the difference | |
1208 | it adds an extra @var{offset} parameter of type @code{off_t} similar to | |
1209 | @code{pwrite}. The data is written to the file starting at position | |
1210 | @var{offset}. The position of the file descriptor itself is not affected | |
1211 | by the operation. The value is the same as before the call. | |
1212 | ||
1213 | However, on Linux, if a file is opened with @code{O_APPEND}, @code{pwrite} | |
1214 | appends data to the end of the file, regardless of the value of | |
1215 | @code{offset}. | |
1216 | ||
1217 | When the source file is compiled with @code{_FILE_OFFSET_BITS == 64} the | |
1218 | @code{pwritev} function is in fact @code{pwritev64} and the type | |
1219 | @code{off_t} has 64 bits, which makes it possible to handle files up to | |
1220 | @twoexp{63} bytes in length. | |
1221 | ||
1222 | The return value is a count of bytes (@emph{not} buffers) written, @math{0} | |
1223 | indicating end-of-file, or @math{-1} indicating an error. The possible | |
1224 | errors are the same as in @code{writev} and @code{pwrite}. | |
1225 | @end deftypefun | |
1226 | ||
1227 | @deftypefun ssize_t pwritev64 (int @var{fd}, const struct iovec *@var{iov}, int @var{iovcnt}, off64_t @var{offset}) | |
1228 | @standards{BSD, unistd.h} | |
1229 | @safety{@prelim{}@mtsafe{}@assafe{}@acsafe{}} | |
1230 | @c This is a syscall for Linux 3.2 for all architectures but microblaze | |
1231 | @c (which was added on 3.15). The sysdeps/posix fallback emulation | |
1232 | @c is also MT-Safe since it calls pwrite64, and it is now a syscall on all | |
1233 | @c targets. | |
1234 | ||
1235 | This function is similar to the @code{pwritev} function with the difference | |
1236 | is that the @var{offset} parameter is of type @code{off64_t} instead of | |
1237 | @code{off_t}. It makes it possible on 32 bit machines to address | |
1238 | files larger than @twoexp{31} bytes and up to @twoexp{63} bytes. The | |
1239 | file descriptor @code{filedes} must be opened using @code{open64} since | |
1240 | otherwise the large offsets possible with @code{off64_t} will lead to | |
1241 | errors with a descriptor in small file mode. | |
1242 | ||
1243 | When the source file is compiled using @code{_FILE_OFFSET_BITS == 64} on a | |
1244 | 32 bit machine this function is actually available under the name | |
1245 | @code{pwritev} and so transparently replaces the 32 bit interface. | |
1246 | @end deftypefun | |
1247 | ||
1248 | @deftypefun ssize_t preadv2 (int @var{fd}, const struct iovec *@var{iov}, int @var{iovcnt}, off_t @var{offset}, int @var{flags}) | |
1249 | @standards{GNU, sys/uio.h} | |
1250 | @safety{@prelim{}@mtsafe{}@assafe{}@acsafe{}} | |
1251 | @c This is a syscall for Linux v4.6. The sysdeps/posix fallback emulation | |
1252 | @c is also MT-Safe since it calls preadv. | |
1253 | ||
d4b4a00a FW |
1254 | This function is similar to the @code{preadv} function, with the |
1255 | difference it adds an extra @var{flags} parameter of type @code{int}. | |
1256 | Additionally, if @var{offset} is @math{-1}, the current file position | |
1257 | is used and updated (like the @code{readv} function). | |
1258 | ||
1259 | The supported @var{flags} are dependent of the underlying system. For | |
1260 | Linux it supports: | |
f6e965ee FW |
1261 | |
1262 | @vtable @code | |
1263 | @item RWF_HIPRI | |
1264 | High priority request. This adds a flag that tells the file system that | |
1265 | this is a high priority request for which it is worth to poll the hardware. | |
1266 | The flag is purely advisory and can be ignored if not supported. The | |
1267 | @var{fd} must be opened using @code{O_DIRECT}. | |
1268 | ||
1269 | @item RWF_DSYNC | |
1270 | Per-IO synchronization as if the file was opened with @code{O_DSYNC} flag. | |
1271 | ||
1272 | @item RWF_SYNC | |
1273 | Per-IO synchronization as if the file was opened with @code{O_SYNC} flag. | |
1274 | ||
1275 | @item RWF_NOWAIT | |
1276 | Use nonblocking mode for this operation; that is, this call to @code{preadv2} | |
1277 | will fail and set @code{errno} to @code{EAGAIN} if the operation would block. | |
1278 | @end vtable | |
1279 | ||
1280 | When the source file is compiled with @code{_FILE_OFFSET_BITS == 64} the | |
1281 | @code{preadv2} function is in fact @code{preadv64v2} and the type | |
1282 | @code{off_t} has 64 bits, which makes it possible to handle files up to | |
1283 | @twoexp{63} bytes in length. | |
1284 | ||
1285 | The return value is a count of bytes (@emph{not} buffers) read, @math{0} | |
1286 | indicating end-of-file, or @math{-1} indicating an error. The possible | |
1287 | errors are the same as in @code{preadv} with the addition of: | |
1288 | ||
1289 | @table @code | |
1290 | ||
1291 | @item EOPNOTSUPP | |
1292 | ||
1293 | @c The default sysdeps/posix code will return it for any flags value | |
1294 | @c different than 0. | |
1295 | An unsupported @var{flags} was used. | |
1296 | ||
1297 | @end table | |
1298 | ||
1299 | @end deftypefun | |
1300 | ||
1301 | @deftypefun ssize_t preadv64v2 (int @var{fd}, const struct iovec *@var{iov}, int @var{iovcnt}, off64_t @var{offset}, int @var{flags}) | |
1302 | @standards{GNU, unistd.h} | |
1303 | @safety{@prelim{}@mtsafe{}@assafe{}@acsafe{}} | |
1304 | @c This is a syscall for Linux v4.6. The sysdeps/posix fallback emulation | |
1305 | @c is also MT-Safe since it calls preadv. | |
07435eb4 | 1306 | |
f6e965ee FW |
1307 | This function is similar to the @code{preadv2} function with the difference |
1308 | is that the @var{offset} parameter is of type @code{off64_t} instead of | |
1309 | @code{off_t}. It makes it possible on 32 bit machines to address | |
1310 | files larger than @twoexp{31} bytes and up to @twoexp{63} bytes. The | |
1311 | file descriptor @code{filedes} must be opened using @code{open64} since | |
1312 | otherwise the large offsets possible with @code{off64_t} will lead to | |
1313 | errors with a descriptor in small file mode. | |
1314 | ||
1315 | When the source file is compiled using @code{_FILE_OFFSET_BITS == 64} on a | |
1316 | 32 bit machine this function is actually available under the name | |
1317 | @code{preadv2} and so transparently replaces the 32 bit interface. | |
1318 | @end deftypefun | |
1319 | ||
1320 | ||
1321 | @deftypefun ssize_t pwritev2 (int @var{fd}, const struct iovec *@var{iov}, int @var{iovcnt}, off_t @var{offset}, int @var{flags}) | |
1322 | @standards{GNU, sys/uio.h} | |
1323 | @safety{@prelim{}@mtsafe{}@assafe{}@acsafe{}} | |
1324 | @c This is a syscall for Linux v4.6. The sysdeps/posix fallback emulation | |
1325 | @c is also MT-Safe since it calls pwritev. | |
1326 | ||
d4b4a00a FW |
1327 | This function is similar to the @code{pwritev} function, with the |
1328 | difference it adds an extra @var{flags} parameter of type @code{int}. | |
1329 | Additionally, if @var{offset} is @math{-1}, the current file position | |
1330 | should is used and updated (like the @code{writev} function). | |
1331 | ||
1332 | The supported @var{flags} are dependent of the underlying system. For | |
1333 | Linux, the supported flags are the same as those for @code{preadv2}. | |
f6e965ee FW |
1334 | |
1335 | When the source file is compiled with @code{_FILE_OFFSET_BITS == 64} the | |
1336 | @code{pwritev2} function is in fact @code{pwritev64v2} and the type | |
1337 | @code{off_t} has 64 bits, which makes it possible to handle files up to | |
1338 | @twoexp{63} bytes in length. | |
1339 | ||
1340 | The return value is a count of bytes (@emph{not} buffers) write, @math{0} | |
1341 | indicating end-of-file, or @math{-1} indicating an error. The possible | |
1342 | errors are the same as in @code{preadv2}. | |
1343 | @end deftypefun | |
1344 | ||
1345 | @deftypefun ssize_t pwritev64v2 (int @var{fd}, const struct iovec *@var{iov}, int @var{iovcnt}, off64_t @var{offset}, int @var{flags}) | |
1346 | @standards{GNU, unistd.h} | |
1347 | @safety{@prelim{}@mtsafe{}@assafe{}@acsafe{}} | |
1348 | @c This is a syscall for Linux v4.6. The sysdeps/posix fallback emulation | |
1349 | @c is also MT-Safe since it calls pwritev. | |
1350 | ||
1351 | This function is similar to the @code{pwritev2} function with the difference | |
1352 | is that the @var{offset} parameter is of type @code{off64_t} instead of | |
1353 | @code{off_t}. It makes it possible on 32 bit machines to address | |
1354 | files larger than @twoexp{31} bytes and up to @twoexp{63} bytes. The | |
1355 | file descriptor @code{filedes} must be opened using @code{open64} since | |
1356 | otherwise the large offsets possible with @code{off64_t} will lead to | |
1357 | errors with a descriptor in small file mode. | |
1358 | ||
1359 | When the source file is compiled using @code{_FILE_OFFSET_BITS == 64} on a | |
1360 | 32 bit machine this function is actually available under the name | |
1361 | @code{pwritev2} and so transparently replaces the 32 bit interface. | |
1362 | @end deftypefun | |
07435eb4 | 1363 | |
bad7a0c8 FW |
1364 | @node Copying File Data |
1365 | @section Copying data between two files | |
1366 | @cindex copying files | |
1367 | @cindex file copy | |
1368 | ||
1369 | A special function is provided to copy data between two files on the | |
1370 | same file system. The system can optimize such copy operations. This | |
1371 | is particularly important on network file systems, where the data would | |
1372 | otherwise have to be transferred twice over the network. | |
1373 | ||
1374 | Note that this function only copies file data, but not metadata such as | |
1375 | file permissions or extended attributes. | |
1376 | ||
1377 | @deftypefun ssize_t copy_file_range (int @var{inputfd}, off64_t *@var{inputpos}, int @var{outputfd}, off64_t *@var{outputpos}, ssize_t @var{length}, unsigned int @var{flags}) | |
1378 | @standards{GNU, unistd.h} | |
1379 | @safety{@prelim{}@mtsafe{}@assafe{}@acsafe{}} | |
1380 | ||
1381 | This function copies up to @var{length} bytes from the file descriptor | |
1382 | @var{inputfd} to the file descriptor @var{outputfd}. | |
1383 | ||
1384 | The function can operate on both the current file position (like | |
1385 | @code{read} and @code{write}) and an explicit offset (like @code{pread} | |
1386 | and @code{pwrite}). If the @var{inputpos} pointer is null, the file | |
1387 | position of @var{inputfd} is used as the starting point of the copy | |
1388 | operation, and the file position is advanced during it. If | |
1389 | @var{inputpos} is not null, then @code{*@var{inputpos}} is used as the | |
1390 | starting point of the copy operation, and @code{*@var{inputpos}} is | |
1391 | incremented by the number of copied bytes, but the file position remains | |
1392 | unchanged. Similar rules apply to @var{outputfd} and @var{outputpos} | |
1393 | for the output file position. | |
1394 | ||
1395 | The @var{flags} argument is currently reserved and must be zero. | |
1396 | ||
1397 | The @code{copy_file_range} function returns the number of bytes copied. | |
1398 | This can be less than the specified @var{length} in case the input file | |
1399 | contains fewer remaining bytes than @var{length}, or if a read or write | |
1400 | failure occurs. The return value is zero if the end of the input file | |
1401 | is encountered immediately. | |
1402 | ||
1403 | If no bytes can be copied, to report an error, @code{copy_file_range} | |
1404 | returns the value @math{-1} and sets @code{errno}. The following | |
1405 | @code{errno} error conditions are specific to this function: | |
1406 | ||
1407 | @table @code | |
1408 | @item EISDIR | |
1409 | At least one of the descriptors @var{inputfd} or @var{outputfd} refers | |
1410 | to a directory. | |
1411 | ||
1412 | @item EINVAL | |
1413 | At least one of the descriptors @var{inputfd} or @var{outputfd} refers | |
1414 | to a non-regular, non-directory file (such as a socket or a FIFO). | |
1415 | ||
1416 | The input or output positions before are after the copy operations are | |
1417 | outside of an implementation-defined limit. | |
1418 | ||
1419 | The @var{flags} argument is not zero. | |
1420 | ||
1421 | @item EFBIG | |
1422 | The new file size would exceed the process file size limit. | |
1423 | @xref{Limits on Resources}. | |
1424 | ||
1425 | The input or output positions before are after the copy operations are | |
1426 | outside of an implementation-defined limit. This can happen if the file | |
1427 | was not opened with large file support (LFS) on 32-bit machines, and the | |
1428 | copy operation would create a file which is larger than what | |
1429 | @code{off_t} could represent. | |
1430 | ||
1431 | @item EBADF | |
1432 | The argument @var{inputfd} is not a valid file descriptor open for | |
1433 | reading. | |
1434 | ||
1435 | The argument @var{outputfd} is not a valid file descriptor open for | |
1436 | writing, or @var{outputfd} has been opened with @code{O_APPEND}. | |
1437 | ||
1438 | @item EXDEV | |
1439 | The input and output files reside on different file systems. | |
1440 | @end table | |
1441 | ||
1442 | In addition, @code{copy_file_range} can fail with the error codes | |
1443 | which are used by @code{read}, @code{pread}, @code{write}, and | |
1444 | @code{pwrite}. | |
1445 | ||
1446 | The @code{copy_file_range} function is a cancellation point. In case of | |
1447 | cancellation, the input location (the file position or the value at | |
1448 | @code{*@var{inputpos}}) is indeterminate. | |
1449 | @end deftypefun | |
1450 | ||
07435eb4 UD |
1451 | @node Memory-mapped I/O |
1452 | @section Memory-mapped I/O | |
1453 | ||
1454 | On modern operating systems, it is possible to @dfn{mmap} (pronounced | |
1455 | ``em-map'') a file to a region of memory. When this is done, the file can | |
1456 | be accessed just like an array in the program. | |
1457 | ||
19e4c7dd | 1458 | This is more efficient than @code{read} or @code{write}, as only the regions |
04b9968b | 1459 | of the file that a program actually accesses are loaded. Accesses to |
07435eb4 UD |
1460 | not-yet-loaded parts of the mmapped region are handled in the same way as |
1461 | swapped out pages. | |
1462 | ||
b642f101 UD |
1463 | Since mmapped pages can be stored back to their file when physical |
1464 | memory is low, it is possible to mmap files orders of magnitude larger | |
1465 | than both the physical memory @emph{and} swap space. The only limit is | |
1466 | address space. The theoretical limit is 4GB on a 32-bit machine - | |
1467 | however, the actual limit will be smaller since some areas will be | |
1468 | reserved for other purposes. If the LFS interface is used the file size | |
1469 | on 32-bit systems is not limited to 2GB (offsets are signed which | |
1470 | reduces the addressable area of 4GB by half); the full 64-bit are | |
1471 | available. | |
07435eb4 UD |
1472 | |
1473 | Memory mapping only works on entire pages of memory. Thus, addresses | |
1474 | for mapping must be page-aligned, and length values will be rounded up. | |
a465b89e | 1475 | To determine the default size of a page the machine uses one should use: |
07435eb4 | 1476 | |
b642f101 | 1477 | @vindex _SC_PAGESIZE |
07435eb4 UD |
1478 | @smallexample |
1479 | size_t page_size = (size_t) sysconf (_SC_PAGESIZE); | |
1480 | @end smallexample | |
1481 | ||
a465b89e FW |
1482 | On some systems, mappings can use larger page sizes |
1483 | for certain files, and applications can request larger page sizes for | |
1484 | anonymous mappings as well (see the @code{MAP_HUGETLB} flag below). | |
1485 | ||
1486 | The following functions are declared in @file{sys/mman.h}: | |
07435eb4 | 1487 | |
cc6e48bc | 1488 | @deftypefun {void *} mmap (void *@var{address}, size_t @var{length}, int @var{protect}, int @var{flags}, int @var{filedes}, off_t @var{offset}) |
d08a7e4c | 1489 | @standards{POSIX, sys/mman.h} |
2cc3615c | 1490 | @safety{@prelim{}@mtsafe{}@assafe{}@acsafe{}} |
07435eb4 UD |
1491 | |
1492 | The @code{mmap} function creates a new mapping, connected to bytes | |
b73147d0 | 1493 | (@var{offset}) to (@var{offset} + @var{length} - 1) in the file open on |
b61345a1 UD |
1494 | @var{filedes}. A new reference for the file specified by @var{filedes} |
1495 | is created, which is not removed by closing the file. | |
07435eb4 UD |
1496 | |
1497 | @var{address} gives a preferred starting address for the mapping. | |
cf822e3c OB |
1498 | @code{NULL} expresses no preference. Any previous mapping at that |
1499 | address is automatically removed. The address you give may still be | |
07435eb4 UD |
1500 | changed, unless you use the @code{MAP_FIXED} flag. |
1501 | ||
07435eb4 UD |
1502 | @var{protect} contains flags that control what kind of access is |
1503 | permitted. They include @code{PROT_READ}, @code{PROT_WRITE}, and | |
0f74bbf5 FW |
1504 | @code{PROT_EXEC}. The special flag @code{PROT_NONE} reserves a region |
1505 | of address space for future use. The @code{mprotect} function can be | |
1506 | used to change the protection flags. @xref{Memory Protection}. | |
07435eb4 UD |
1507 | |
1508 | @var{flags} contains flags that control the nature of the map. | |
1509 | One of @code{MAP_SHARED} or @code{MAP_PRIVATE} must be specified. | |
1510 | ||
1511 | They include: | |
1512 | ||
1513 | @vtable @code | |
1514 | @item MAP_PRIVATE | |
1515 | This specifies that writes to the region should never be written back | |
1516 | to the attached file. Instead, a copy is made for the process, and the | |
1517 | region will be swapped normally if memory runs low. No other process will | |
1518 | see the changes. | |
1519 | ||
1520 | Since private mappings effectively revert to ordinary memory | |
1521 | when written to, you must have enough virtual memory for a copy of | |
1522 | the entire mmapped region if you use this mode with @code{PROT_WRITE}. | |
1523 | ||
1524 | @item MAP_SHARED | |
1525 | This specifies that writes to the region will be written back to the | |
1526 | file. Changes made will be shared immediately with other processes | |
1527 | mmaping the same file. | |
1528 | ||
1529 | Note that actual writing may take place at any time. You need to use | |
1530 | @code{msync}, described below, if it is important that other processes | |
1531 | using conventional I/O get a consistent view of the file. | |
1532 | ||
1533 | @item MAP_FIXED | |
1534 | This forces the system to use the exact mapping address specified in | |
1535 | @var{address} and fail if it can't. | |
1536 | ||
1537 | @c One of these is official - the other is obviously an obsolete synonym | |
1538 | @c Which is which? | |
1539 | @item MAP_ANONYMOUS | |
1540 | @itemx MAP_ANON | |
1541 | This flag tells the system to create an anonymous mapping, not connected | |
9739d2d5 | 1542 | to a file. @var{filedes} and @var{offset} are ignored, and the region is |
07435eb4 UD |
1543 | initialized with zeros. |
1544 | ||
1545 | Anonymous maps are used as the basic primitive to extend the heap on some | |
1546 | systems. They are also useful to share data between multiple tasks | |
1547 | without creating a file. | |
1548 | ||
49c091e5 | 1549 | On some systems using private anonymous mmaps is more efficient than using |
1f77f049 | 1550 | @code{malloc} for large blocks. This is not an issue with @theglibc{}, |
07435eb4 UD |
1551 | as the included @code{malloc} automatically uses @code{mmap} where appropriate. |
1552 | ||
a465b89e FW |
1553 | @item MAP_HUGETLB |
1554 | @standards{Linux, sys/mman.h} | |
1555 | This requests that the system uses an alternative page size which is | |
1556 | larger than the default page size for the mapping. For some workloads, | |
1557 | increasing the page size for large mappings improves performance because | |
1558 | the system needs to handle far fewer pages. For other workloads which | |
1559 | require frequent transfer of pages between storage or different nodes, | |
1560 | the decreased page granularity may cause performance problems due to the | |
1561 | increased page size and larger transfers. | |
1562 | ||
1563 | In order to create the mapping, the system needs physically contiguous | |
1564 | memory of the size of the increased page size. As a result, | |
1565 | @code{MAP_HUGETLB} mappings are affected by memory fragmentation, and | |
1566 | their creation can fail even if plenty of memory is available in the | |
1567 | system. | |
1568 | ||
1569 | Not all file systems support mappings with an increased page size. | |
1570 | ||
1571 | The @code{MAP_HUGETLB} flag is specific to Linux. | |
1572 | ||
1573 | @c There is a mechanism to select different hugepage sizes; see | |
1574 | @c include/uapi/asm-generic/hugetlb_encode.h in the kernel sources. | |
1575 | ||
07435eb4 UD |
1576 | @c Linux has some other MAP_ options, which I have not discussed here. |
1577 | @c MAP_DENYWRITE, MAP_EXECUTABLE and MAP_GROWSDOWN don't seem applicable to | |
cf822e3c | 1578 | @c user programs (and I don't understand the last two). MAP_LOCKED does |
07435eb4 UD |
1579 | @c not appear to be implemented. |
1580 | ||
1581 | @end vtable | |
1582 | ||
52e6d801 FB |
1583 | @code{mmap} returns the address of the new mapping, or |
1584 | @code{MAP_FAILED} for an error. | |
07435eb4 UD |
1585 | |
1586 | Possible errors include: | |
1587 | ||
1588 | @table @code | |
1589 | ||
1590 | @item EINVAL | |
1591 | ||
a465b89e FW |
1592 | Either @var{address} was unusable (because it is not a multiple of the |
1593 | applicable page size), or inconsistent @var{flags} were given. | |
1594 | ||
1595 | If @code{MAP_HUGETLB} was specified, the file or system does not support | |
1596 | large page sizes. | |
07435eb4 UD |
1597 | |
1598 | @item EACCES | |
1599 | ||
1600 | @var{filedes} was not open for the type of access specified in @var{protect}. | |
1601 | ||
1602 | @item ENOMEM | |
1603 | ||
1604 | Either there is not enough memory for the operation, or the process is | |
1605 | out of address space. | |
1606 | ||
1607 | @item ENODEV | |
1608 | ||
1609 | This file is of a type that doesn't support mapping. | |
1610 | ||
1611 | @item ENOEXEC | |
1612 | ||
1613 | The file is on a filesystem that doesn't support mapping. | |
1614 | ||
1615 | @c On Linux, EAGAIN will appear if the file has a conflicting mandatory lock. | |
1616 | @c However mandatory locks are not discussed in this manual. | |
1617 | @c | |
1618 | @c Similarly, ETXTBSY will occur if the MAP_DENYWRITE flag (not documented | |
1619 | @c here) is used and the file is already open for writing. | |
1620 | ||
1621 | @end table | |
1622 | ||
1623 | @end deftypefun | |
1624 | ||
cc6e48bc | 1625 | @deftypefun {void *} mmap64 (void *@var{address}, size_t @var{length}, int @var{protect}, int @var{flags}, int @var{filedes}, off64_t @var{offset}) |
d08a7e4c | 1626 | @standards{LFS, sys/mman.h} |
2cc3615c AO |
1627 | @safety{@prelim{}@mtsafe{}@assafe{}@acsafe{}} |
1628 | @c The page_shift auto detection when MMAP2_PAGE_SHIFT is -1 (it never | |
1629 | @c is) would be thread-unsafe. | |
b642f101 UD |
1630 | The @code{mmap64} function is equivalent to the @code{mmap} function but |
1631 | the @var{offset} parameter is of type @code{off64_t}. On 32-bit systems | |
1632 | this allows the file associated with the @var{filedes} descriptor to be | |
1633 | larger than 2GB. @var{filedes} must be a descriptor returned from a | |
1634 | call to @code{open64} or @code{fopen64} and @code{freopen64} where the | |
1635 | descriptor is retrieved with @code{fileno}. | |
1636 | ||
1637 | When the sources are translated with @code{_FILE_OFFSET_BITS == 64} this | |
1638 | function is actually available under the name @code{mmap}. I.e., the | |
1639 | new, extended API using 64 bit file sizes and offsets transparently | |
1640 | replaces the old API. | |
1641 | @end deftypefun | |
1642 | ||
07435eb4 | 1643 | @deftypefun int munmap (void *@var{addr}, size_t @var{length}) |
d08a7e4c | 1644 | @standards{POSIX, sys/mman.h} |
2cc3615c | 1645 | @safety{@prelim{}@mtsafe{}@assafe{}@acsafe{}} |
07435eb4 UD |
1646 | |
1647 | @code{munmap} removes any memory maps from (@var{addr}) to (@var{addr} + | |
1648 | @var{length}). @var{length} should be the length of the mapping. | |
1649 | ||
04b9968b | 1650 | It is safe to unmap multiple mappings in one command, or include unmapped |
07435eb4 | 1651 | space in the range. It is also possible to unmap only part of an existing |
04b9968b | 1652 | mapping. However, only entire pages can be removed. If @var{length} is not |
07435eb4 UD |
1653 | an even number of pages, it will be rounded up. |
1654 | ||
1655 | It returns @math{0} for success and @math{-1} for an error. | |
1656 | ||
1657 | One error is possible: | |
1658 | ||
1659 | @table @code | |
1660 | ||
1661 | @item EINVAL | |
04b9968b | 1662 | The memory range given was outside the user mmap range or wasn't page |
07435eb4 UD |
1663 | aligned. |
1664 | ||
1665 | @end table | |
1666 | ||
1667 | @end deftypefun | |
1668 | ||
1669 | @deftypefun int msync (void *@var{address}, size_t @var{length}, int @var{flags}) | |
d08a7e4c | 1670 | @standards{POSIX, sys/mman.h} |
2cc3615c | 1671 | @safety{@prelim{}@mtsafe{}@assafe{}@acsafe{}} |
07435eb4 UD |
1672 | |
1673 | When using shared mappings, the kernel can write the file at any time | |
1674 | before the mapping is removed. To be certain data has actually been | |
49c091e5 UD |
1675 | written to the file and will be accessible to non-memory-mapped I/O, it |
1676 | is necessary to use this function. | |
07435eb4 UD |
1677 | |
1678 | It operates on the region @var{address} to (@var{address} + @var{length}). | |
1679 | It may be used on part of a mapping or multiple mappings, however the | |
1680 | region given should not contain any unmapped space. | |
1681 | ||
1682 | @var{flags} can contain some options: | |
1683 | ||
1684 | @vtable @code | |
1685 | ||
1686 | @item MS_SYNC | |
1687 | ||
1688 | This flag makes sure the data is actually written @emph{to disk}. | |
1689 | Normally @code{msync} only makes sure that accesses to a file with | |
1690 | conventional I/O reflect the recent changes. | |
1691 | ||
1692 | @item MS_ASYNC | |
1693 | ||
1694 | This tells @code{msync} to begin the synchronization, but not to wait for | |
1695 | it to complete. | |
1696 | ||
1697 | @c Linux also has MS_INVALIDATE, which I don't understand. | |
1698 | ||
1699 | @end vtable | |
1700 | ||
1701 | @code{msync} returns @math{0} for success and @math{-1} for | |
1702 | error. Errors include: | |
1703 | ||
1704 | @table @code | |
1705 | ||
1706 | @item EINVAL | |
1707 | An invalid region was given, or the @var{flags} were invalid. | |
1708 | ||
1709 | @item EFAULT | |
1710 | There is no existing mapping in at least part of the given region. | |
1711 | ||
1712 | @end table | |
1713 | ||
1714 | @end deftypefun | |
1715 | ||
1716 | @deftypefun {void *} mremap (void *@var{address}, size_t @var{length}, size_t @var{new_length}, int @var{flag}) | |
d08a7e4c | 1717 | @standards{GNU, sys/mman.h} |
2cc3615c | 1718 | @safety{@prelim{}@mtsafe{}@assafe{}@acsafe{}} |
07435eb4 UD |
1719 | |
1720 | This function can be used to change the size of an existing memory | |
1721 | area. @var{address} and @var{length} must cover a region entirely mapped | |
cf822e3c | 1722 | in the same @code{mmap} statement. A new mapping with the same |
04b9968b | 1723 | characteristics will be returned with the length @var{new_length}. |
07435eb4 | 1724 | |
cf822e3c | 1725 | One option is possible, @code{MREMAP_MAYMOVE}. If it is given in |
07435eb4 UD |
1726 | @var{flags}, the system may remove the existing mapping and create a new |
1727 | one of the desired length in another location. | |
1728 | ||
cf822e3c | 1729 | The address of the resulting mapping is returned, or @math{-1}. Possible |
07435eb4 UD |
1730 | error codes include: |
1731 | ||
07435eb4 UD |
1732 | @table @code |
1733 | ||
1734 | @item EFAULT | |
1735 | There is no existing mapping in at least part of the original region, or | |
1736 | the region covers two or more distinct mappings. | |
1737 | ||
1738 | @item EINVAL | |
1739 | The address given is misaligned or inappropriate. | |
1740 | ||
1741 | @item EAGAIN | |
1742 | The region has pages locked, and if extended it would exceed the | |
1743 | process's resource limit for locked pages. @xref{Limits on Resources}. | |
1744 | ||
1745 | @item ENOMEM | |
19e4c7dd | 1746 | The region is private writable, and insufficient virtual memory is |
07435eb4 UD |
1747 | available to extend it. Also, this error will occur if |
1748 | @code{MREMAP_MAYMOVE} is not given and the extension would collide with | |
1749 | another mapped region. | |
1750 | ||
1751 | @end table | |
1752 | @end deftypefun | |
1753 | ||
04b9968b UD |
1754 | This function is only available on a few systems. Except for performing |
1755 | optional optimizations one should not rely on this function. | |
1756 | ||
07435eb4 UD |
1757 | Not all file descriptors may be mapped. Sockets, pipes, and most devices |
1758 | only allow sequential access and do not fit into the mapping abstraction. | |
1759 | In addition, some regular files may not be mmapable, and older kernels may | |
1760 | not support mapping at all. Thus, programs using @code{mmap} should | |
1761 | have a fallback method to use should it fail. @xref{Mmap,,,standards,GNU | |
1762 | Coding Standards}. | |
1763 | ||
0bc93a2f | 1764 | @deftypefun int madvise (void *@var{addr}, size_t @var{length}, int @var{advice}) |
d08a7e4c | 1765 | @standards{POSIX, sys/mman.h} |
2cc3615c | 1766 | @safety{@prelim{}@mtsafe{}@assafe{}@acsafe{}} |
0bc93a2f AJ |
1767 | |
1768 | This function can be used to provide the system with @var{advice} about | |
1769 | the intended usage patterns of the memory region starting at @var{addr} | |
1770 | and extending @var{length} bytes. | |
1771 | ||
1772 | The valid BSD values for @var{advice} are: | |
1773 | ||
2fe82ca6 | 1774 | @vtable @code |
0bc93a2f AJ |
1775 | |
1776 | @item MADV_NORMAL | |
1777 | The region should receive no further special treatment. | |
1778 | ||
1779 | @item MADV_RANDOM | |
cf822e3c | 1780 | The region will be accessed via random page references. The kernel |
0bc93a2f AJ |
1781 | should page-in the minimal number of pages for each page fault. |
1782 | ||
1783 | @item MADV_SEQUENTIAL | |
cf822e3c | 1784 | The region will be accessed via sequential page references. This |
0bc93a2f AJ |
1785 | may cause the kernel to aggressively read-ahead, expecting further |
1786 | sequential references after any page fault within this region. | |
1787 | ||
1788 | @item MADV_WILLNEED | |
1789 | The region will be needed. The pages within this region may | |
1790 | be pre-faulted in by the kernel. | |
1791 | ||
1792 | @item MADV_DONTNEED | |
1793 | The region is no longer needed. The kernel may free these pages, | |
1794 | causing any changes to the pages to be lost, as well as swapped | |
1795 | out pages to be discarded. | |
1796 | ||
a465b89e FW |
1797 | @item MADV_HUGEPAGE |
1798 | @standards{Linux, sys/mman.h} | |
1799 | Indicate that it is beneficial to increase the page size for this | |
1800 | mapping. This can improve performance for larger mappings because the | |
1801 | system needs to handle far fewer pages. However, if parts of the | |
1802 | mapping are frequently transferred between storage or different nodes, | |
1803 | performance may suffer because individual transfers can become | |
1804 | substantially larger due to the increased page size. | |
1805 | ||
1806 | This flag is specific to Linux. | |
1807 | ||
1808 | @item MADV_NOHUGEPAGE | |
1809 | Undo the effect of a previous @code{MADV_HUGEPAGE} advice. This flag | |
1810 | is specific to Linux. | |
1811 | ||
2fe82ca6 | 1812 | @end vtable |
0bc93a2f AJ |
1813 | |
1814 | The POSIX names are slightly different, but with the same meanings: | |
1815 | ||
2fe82ca6 | 1816 | @vtable @code |
0bc93a2f AJ |
1817 | |
1818 | @item POSIX_MADV_NORMAL | |
1819 | This corresponds with BSD's @code{MADV_NORMAL}. | |
1820 | ||
1821 | @item POSIX_MADV_RANDOM | |
1822 | This corresponds with BSD's @code{MADV_RANDOM}. | |
1823 | ||
1824 | @item POSIX_MADV_SEQUENTIAL | |
1825 | This corresponds with BSD's @code{MADV_SEQUENTIAL}. | |
1826 | ||
1827 | @item POSIX_MADV_WILLNEED | |
1828 | This corresponds with BSD's @code{MADV_WILLNEED}. | |
1829 | ||
1830 | @item POSIX_MADV_DONTNEED | |
1831 | This corresponds with BSD's @code{MADV_DONTNEED}. | |
1832 | ||
2fe82ca6 | 1833 | @end vtable |
0bc93a2f | 1834 | |
bb4e6db2 | 1835 | @code{madvise} returns @math{0} for success and @math{-1} for |
0bc93a2f AJ |
1836 | error. Errors include: |
1837 | @table @code | |
1838 | ||
1839 | @item EINVAL | |
1840 | An invalid region was given, or the @var{advice} was invalid. | |
1841 | ||
1842 | @item EFAULT | |
1843 | There is no existing mapping in at least part of the given region. | |
1844 | ||
1845 | @end table | |
1846 | @end deftypefun | |
07435eb4 | 1847 | |
416e0145 | 1848 | @deftypefn Function int shm_open (const char *@var{name}, int @var{oflag}, mode_t @var{mode}) |
d08a7e4c | 1849 | @standards{POSIX, sys/mman.h} |
2cc3615c AO |
1850 | @safety{@prelim{}@mtsafe{@mtslocale{}}@asunsafe{@asuinit{} @ascuheap{} @asulock{}}@acunsafe{@aculock{} @acsmem{} @acsfd{}}} |
1851 | @c shm_open @mtslocale @asuinit @ascuheap @asulock @aculock @acsmem @acsfd | |
1852 | @c libc_once(where_is_shmfs) @mtslocale @asuinit @ascuheap @asulock @aculock @acsmem @acsfd | |
1853 | @c where_is_shmfs @mtslocale @ascuheap @asulock @aculock @acsmem @acsfd | |
1854 | @c statfs dup ok | |
1855 | @c setmntent dup @ascuheap @asulock @acsmem @acsfd @aculock | |
1856 | @c getmntent_r dup @mtslocale @ascuheap @aculock @acsmem [no @asucorrupt @acucorrupt; exclusive stream] | |
1857 | @c strcmp dup ok | |
1858 | @c strlen dup ok | |
1859 | @c malloc dup @ascuheap @acsmem | |
1860 | @c mempcpy dup ok | |
1861 | @c endmntent dup @ascuheap @asulock @aculock @acsmem @acsfd | |
1862 | @c strlen dup ok | |
1863 | @c strchr dup ok | |
1864 | @c mempcpy dup ok | |
1865 | @c open dup @acsfd | |
1866 | @c fcntl dup ok | |
1867 | @c close dup @acsfd | |
416e0145 OB |
1868 | |
1869 | This function returns a file descriptor that can be used to allocate shared | |
cf822e3c | 1870 | memory via mmap. Unrelated processes can use same @var{name} to create or |
416e0145 OB |
1871 | open existing shared memory objects. |
1872 | ||
1873 | A @var{name} argument specifies the shared memory object to be opened. | |
1874 | In @theglibc{} it must be a string smaller than @code{NAME_MAX} bytes starting | |
1875 | with an optional slash but containing no other slashes. | |
1876 | ||
1877 | The semantics of @var{oflag} and @var{mode} arguments is same as in @code{open}. | |
1878 | ||
1879 | @code{shm_open} returns the file descriptor on success or @math{-1} on error. | |
1880 | On failure @code{errno} is set. | |
1881 | @end deftypefn | |
1882 | ||
1883 | @deftypefn Function int shm_unlink (const char *@var{name}) | |
2cc3615c AO |
1884 | @safety{@prelim{}@mtsafe{@mtslocale{}}@asunsafe{@asuinit{} @ascuheap{} @asulock{}}@acunsafe{@aculock{} @acsmem{} @acsfd{}}} |
1885 | @c shm_unlink @mtslocale @asuinit @ascuheap @asulock @aculock @acsmem @acsfd | |
1886 | @c libc_once(where_is_shmfs) dup @mtslocale @asuinit @ascuheap @asulock @aculock @acsmem @acsfd | |
1887 | @c strlen dup ok | |
1888 | @c strchr dup ok | |
1889 | @c mempcpy dup ok | |
1890 | @c unlink dup ok | |
416e0145 | 1891 | |
9739d2d5 | 1892 | This function is the inverse of @code{shm_open} and removes the object with |
416e0145 OB |
1893 | the given @var{name} previously created by @code{shm_open}. |
1894 | ||
1895 | @code{shm_unlink} returns @math{0} on success or @math{-1} on error. | |
1896 | On failure @code{errno} is set. | |
1897 | @end deftypefn | |
1898 | ||
59d2cbb1 FW |
1899 | @deftypefun int memfd_create (const char *@var{name}, unsigned int @var{flags}) |
1900 | @standards{Linux, sys/mman.h} | |
1901 | @safety{@prelim{}@mtsafe{}@assafe{}@acsafe{@acsfd{}}} | |
1902 | The @code{memfd_create} function returns a file descriptor which can be | |
1903 | used to create memory mappings using the @code{mmap} function. It is | |
1904 | similar to the @code{shm_open} function in the sense that these mappings | |
1905 | are not backed by actual files. However, the descriptor returned by | |
1906 | @code{memfd_create} does not correspond to a named object; the | |
1907 | @var{name} argument is used for debugging purposes only (e.g., will | |
1908 | appear in @file{/proc}), and separate invocations of @code{memfd_create} | |
1909 | with the same @var{name} will not return descriptors for the same region | |
1910 | of memory. The descriptor can also be used to create alias mappings | |
1911 | within the same process. | |
1912 | ||
1913 | The descriptor initially refers to a zero-length file. Before mappings | |
1914 | can be created which are backed by memory, the file size needs to be | |
1915 | increased with the @code{ftruncate} function. @xref{File Size}. | |
1916 | ||
1917 | The @var{flags} argument can be a combination of the following flags: | |
1918 | ||
1919 | @vtable @code | |
1920 | @item MFD_CLOEXEC | |
1921 | @standards{Linux, sys/mman.h} | |
1922 | The descriptor is created with the @code{O_CLOEXEC} flag. | |
1923 | ||
1924 | @item MFD_ALLOW_SEALING | |
1925 | @standards{Linux, sys/mman.h} | |
1926 | The descriptor supports the addition of seals using the @code{fcntl} | |
1927 | function. | |
1928 | ||
1929 | @item MFD_HUGETLB | |
1930 | @standards{Linux, sys/mman.h} | |
1931 | This requests that mappings created using the returned file descriptor | |
1932 | use a larger page size. See @code{MAP_HUGETLB} above for details. | |
1933 | ||
1934 | This flag is incompatible with @code{MFD_ALLOW_SEALING}. | |
1935 | @end vtable | |
1936 | ||
1937 | @code{memfd_create} returns a file descriptor on success, and @math{-1} | |
1938 | on failure. | |
1939 | ||
1940 | The following @code{errno} error conditions are defined for this | |
1941 | function: | |
1942 | ||
1943 | @table @code | |
1944 | @item EINVAL | |
1945 | An invalid combination is specified in @var{flags}, or @var{name} is | |
1946 | too long. | |
1947 | ||
1948 | @item EFAULT | |
1949 | The @var{name} argument does not point to a string. | |
1950 | ||
1951 | @item EMFILE | |
1952 | The operation would exceed the file descriptor limit for this process. | |
1953 | ||
1954 | @item ENFILE | |
1955 | The operation would exceed the system-wide file descriptor limit. | |
1956 | ||
1957 | @item ENOMEM | |
1958 | There is not enough memory for the operation. | |
1959 | @end table | |
1960 | @end deftypefun | |
1961 | ||
28f540f4 RM |
1962 | @node Waiting for I/O |
1963 | @section Waiting for Input or Output | |
1964 | @cindex waiting for input or output | |
1965 | @cindex multiplexing input | |
1966 | @cindex input from multiple files | |
1967 | ||
1968 | Sometimes a program needs to accept input on multiple input channels | |
1969 | whenever input arrives. For example, some workstations may have devices | |
1970 | such as a digitizing tablet, function button box, or dial box that are | |
1971 | connected via normal asynchronous serial interfaces; good user interface | |
1972 | style requires responding immediately to input on any device. Another | |
1973 | example is a program that acts as a server to several other processes | |
1974 | via pipes or sockets. | |
1975 | ||
1976 | You cannot normally use @code{read} for this purpose, because this | |
1977 | blocks the program until input is available on one particular file | |
1978 | descriptor; input on other channels won't wake it up. You could set | |
1979 | nonblocking mode and poll each file descriptor in turn, but this is very | |
1980 | inefficient. | |
1981 | ||
1982 | A better solution is to use the @code{select} function. This blocks the | |
1983 | program until input or output is ready on a specified set of file | |
1984 | descriptors, or until a timer expires, whichever comes first. This | |
1985 | facility is declared in the header file @file{sys/types.h}. | |
1986 | @pindex sys/types.h | |
1987 | ||
1988 | In the case of a server socket (@pxref{Listening}), we say that | |
1989 | ``input'' is available when there are pending connections that could be | |
1990 | accepted (@pxref{Accepting Connections}). @code{accept} for server | |
1991 | sockets blocks and interacts with @code{select} just as @code{read} does | |
1992 | for normal input. | |
1993 | ||
1994 | @cindex file descriptor sets, for @code{select} | |
1995 | The file descriptor sets for the @code{select} function are specified | |
1996 | as @code{fd_set} objects. Here is the description of the data type | |
1997 | and some macros for manipulating these objects. | |
1998 | ||
28f540f4 | 1999 | @deftp {Data Type} fd_set |
d08a7e4c | 2000 | @standards{BSD, sys/types.h} |
28f540f4 RM |
2001 | The @code{fd_set} data type represents file descriptor sets for the |
2002 | @code{select} function. It is actually a bit array. | |
2003 | @end deftp | |
2004 | ||
28f540f4 | 2005 | @deftypevr Macro int FD_SETSIZE |
d08a7e4c | 2006 | @standards{BSD, sys/types.h} |
28f540f4 RM |
2007 | The value of this macro is the maximum number of file descriptors that a |
2008 | @code{fd_set} object can hold information about. On systems with a | |
2009 | fixed maximum number, @code{FD_SETSIZE} is at least that number. On | |
2010 | some systems, including GNU, there is no absolute limit on the number of | |
2011 | descriptors open, but this macro still has a constant value which | |
2012 | controls the number of bits in an @code{fd_set}; if you get a file | |
2013 | descriptor with a value as high as @code{FD_SETSIZE}, you cannot put | |
2014 | that descriptor into an @code{fd_set}. | |
2015 | @end deftypevr | |
2016 | ||
28f540f4 | 2017 | @deftypefn Macro void FD_ZERO (fd_set *@var{set}) |
d08a7e4c | 2018 | @standards{BSD, sys/types.h} |
2cc3615c | 2019 | @safety{@prelim{}@mtsafe{@mtsrace{:set}}@assafe{}@acsafe{}} |
28f540f4 RM |
2020 | This macro initializes the file descriptor set @var{set} to be the |
2021 | empty set. | |
2022 | @end deftypefn | |
2023 | ||
28f540f4 | 2024 | @deftypefn Macro void FD_SET (int @var{filedes}, fd_set *@var{set}) |
d08a7e4c | 2025 | @standards{BSD, sys/types.h} |
2cc3615c AO |
2026 | @safety{@prelim{}@mtsafe{@mtsrace{:set}}@assafe{}@acsafe{}} |
2027 | @c Setting a bit isn't necessarily atomic, so there's a potential race | |
2028 | @c here if set is not used exclusively. | |
28f540f4 | 2029 | This macro adds @var{filedes} to the file descriptor set @var{set}. |
d9997a45 UD |
2030 | |
2031 | The @var{filedes} parameter must not have side effects since it is | |
2032 | evaluated more than once. | |
28f540f4 RM |
2033 | @end deftypefn |
2034 | ||
28f540f4 | 2035 | @deftypefn Macro void FD_CLR (int @var{filedes}, fd_set *@var{set}) |
d08a7e4c | 2036 | @standards{BSD, sys/types.h} |
2cc3615c AO |
2037 | @safety{@prelim{}@mtsafe{@mtsrace{:set}}@assafe{}@acsafe{}} |
2038 | @c Setting a bit isn't necessarily atomic, so there's a potential race | |
2039 | @c here if set is not used exclusively. | |
28f540f4 | 2040 | This macro removes @var{filedes} from the file descriptor set @var{set}. |
d9997a45 UD |
2041 | |
2042 | The @var{filedes} parameter must not have side effects since it is | |
2043 | evaluated more than once. | |
28f540f4 RM |
2044 | @end deftypefn |
2045 | ||
d9997a45 | 2046 | @deftypefn Macro int FD_ISSET (int @var{filedes}, const fd_set *@var{set}) |
d08a7e4c | 2047 | @standards{BSD, sys/types.h} |
2cc3615c | 2048 | @safety{@prelim{}@mtsafe{@mtsrace{:set}}@assafe{}@acsafe{}} |
28f540f4 | 2049 | This macro returns a nonzero value (true) if @var{filedes} is a member |
3081378b | 2050 | of the file descriptor set @var{set}, and zero (false) otherwise. |
d9997a45 UD |
2051 | |
2052 | The @var{filedes} parameter must not have side effects since it is | |
2053 | evaluated more than once. | |
28f540f4 RM |
2054 | @end deftypefn |
2055 | ||
2056 | Next, here is the description of the @code{select} function itself. | |
2057 | ||
28f540f4 | 2058 | @deftypefun int select (int @var{nfds}, fd_set *@var{read-fds}, fd_set *@var{write-fds}, fd_set *@var{except-fds}, struct timeval *@var{timeout}) |
d08a7e4c | 2059 | @standards{BSD, sys/types.h} |
2cc3615c AO |
2060 | @safety{@prelim{}@mtsafe{@mtsrace{:read-fds} @mtsrace{:write-fds} @mtsrace{:except-fds}}@assafe{}@acsafe{}} |
2061 | @c The select syscall is preferred, but pselect6 may be used instead, | |
2062 | @c which requires converting timeout to a timespec and back. The | |
2063 | @c conversions are not atomic. | |
28f540f4 RM |
2064 | The @code{select} function blocks the calling process until there is |
2065 | activity on any of the specified sets of file descriptors, or until the | |
2066 | timeout period has expired. | |
2067 | ||
2068 | The file descriptors specified by the @var{read-fds} argument are | |
2069 | checked to see if they are ready for reading; the @var{write-fds} file | |
2070 | descriptors are checked to see if they are ready for writing; and the | |
2071 | @var{except-fds} file descriptors are checked for exceptional | |
2072 | conditions. You can pass a null pointer for any of these arguments if | |
2073 | you are not interested in checking for that kind of condition. | |
2074 | ||
76de2021 UD |
2075 | A file descriptor is considered ready for reading if a @code{read} |
2076 | call will not block. This usually includes the read offset being at | |
2077 | the end of the file or there is an error to report. A server socket | |
2078 | is considered ready for reading if there is a pending connection which | |
2079 | can be accepted with @code{accept}; @pxref{Accepting Connections}. A | |
2080 | client socket is ready for writing when its connection is fully | |
2081 | established; @pxref{Connecting}. | |
28f540f4 RM |
2082 | |
2083 | ``Exceptional conditions'' does not mean errors---errors are reported | |
2084 | immediately when an erroneous system call is executed, and do not | |
2085 | constitute a state of the descriptor. Rather, they include conditions | |
2086 | such as the presence of an urgent message on a socket. (@xref{Sockets}, | |
2087 | for information on urgent messages.) | |
2088 | ||
2089 | The @code{select} function checks only the first @var{nfds} file | |
2090 | descriptors. The usual thing is to pass @code{FD_SETSIZE} as the value | |
2091 | of this argument. | |
2092 | ||
2093 | The @var{timeout} specifies the maximum time to wait. If you pass a | |
2094 | null pointer for this argument, it means to block indefinitely until one | |
2095 | of the file descriptors is ready. Otherwise, you should provide the | |
2096 | time in @code{struct timeval} format; see @ref{High-Resolution | |
2097 | Calendar}. Specify zero as the time (a @code{struct timeval} containing | |
2098 | all zeros) if you want to find out which descriptors are ready without | |
2099 | waiting if none are ready. | |
2100 | ||
2101 | The normal return value from @code{select} is the total number of ready file | |
2102 | descriptors in all of the sets. Each of the argument sets is overwritten | |
2103 | with information about the descriptors that are ready for the corresponding | |
2104 | operation. Thus, to see if a particular descriptor @var{desc} has input, | |
2105 | use @code{FD_ISSET (@var{desc}, @var{read-fds})} after @code{select} returns. | |
2106 | ||
2107 | If @code{select} returns because the timeout period expires, it returns | |
2108 | a value of zero. | |
2109 | ||
2110 | Any signal will cause @code{select} to return immediately. So if your | |
2111 | program uses signals, you can't rely on @code{select} to keep waiting | |
2112 | for the full time specified. If you want to be sure of waiting for a | |
2113 | particular amount of time, you must check for @code{EINTR} and repeat | |
2114 | the @code{select} with a newly calculated timeout based on the current | |
2115 | time. See the example below. See also @ref{Interrupted Primitives}. | |
2116 | ||
2117 | If an error occurs, @code{select} returns @code{-1} and does not modify | |
2c6fe0bd | 2118 | the argument file descriptor sets. The following @code{errno} error |
28f540f4 RM |
2119 | conditions are defined for this function: |
2120 | ||
2121 | @table @code | |
2122 | @item EBADF | |
2123 | One of the file descriptor sets specified an invalid file descriptor. | |
2124 | ||
2125 | @item EINTR | |
2126 | The operation was interrupted by a signal. @xref{Interrupted Primitives}. | |
2127 | ||
2128 | @item EINVAL | |
2129 | The @var{timeout} argument is invalid; one of the components is negative | |
2130 | or too large. | |
2131 | @end table | |
2132 | @end deftypefun | |
2133 | ||
2134 | @strong{Portability Note:} The @code{select} function is a BSD Unix | |
2135 | feature. | |
2136 | ||
2137 | Here is an example showing how you can use @code{select} to establish a | |
2138 | timeout period for reading from a file descriptor. The @code{input_timeout} | |
2139 | function blocks the calling process until input is available on the | |
2140 | file descriptor, or until the timeout period expires. | |
2141 | ||
2142 | @smallexample | |
2143 | @include select.c.texi | |
2144 | @end smallexample | |
2145 | ||
2146 | There is another example showing the use of @code{select} to multiplex | |
2147 | input from multiple sockets in @ref{Server Example}. | |
2148 | ||
2149 | ||
dfd2257a UD |
2150 | @node Synchronizing I/O |
2151 | @section Synchronizing I/O operations | |
2152 | ||
2153 | @cindex synchronizing | |
19e4c7dd | 2154 | In most modern operating systems, the normal I/O operations are not |
dfd2257a | 2155 | executed synchronously. I.e., even if a @code{write} system call |
19e4c7dd | 2156 | returns, this does not mean the data is actually written to the media, |
dfd2257a UD |
2157 | e.g., the disk. |
2158 | ||
19e4c7dd | 2159 | In situations where synchronization points are necessary, you can use |
04b9968b | 2160 | special functions which ensure that all operations finish before |
dfd2257a UD |
2161 | they return. |
2162 | ||
8ded91fb | 2163 | @deftypefun void sync (void) |
d08a7e4c | 2164 | @standards{X/Open, unistd.h} |
2cc3615c | 2165 | @safety{@prelim{}@mtsafe{}@assafe{}@acsafe{}} |
dfd2257a | 2166 | A call to this function will not return as long as there is data which |
04b9968b | 2167 | has not been written to the device. All dirty buffers in the kernel will |
dfd2257a UD |
2168 | be written and so an overall consistent system can be achieved (if no |
2169 | other process in parallel writes data). | |
2170 | ||
2171 | A prototype for @code{sync} can be found in @file{unistd.h}. | |
dfd2257a UD |
2172 | @end deftypefun |
2173 | ||
04b9968b UD |
2174 | Programs more often want to ensure that data written to a given file is |
2175 | committed, rather than all data in the system. For this, @code{sync} is overkill. | |
2176 | ||
dfd2257a | 2177 | |
dfd2257a | 2178 | @deftypefun int fsync (int @var{fildes}) |
d08a7e4c | 2179 | @standards{POSIX, unistd.h} |
2cc3615c | 2180 | @safety{@prelim{}@mtsafe{}@assafe{}@acsafe{}} |
19e4c7dd AJ |
2181 | The @code{fsync} function can be used to make sure all data associated with |
2182 | the open file @var{fildes} is written to the device associated with the | |
dfd2257a UD |
2183 | descriptor. The function call does not return unless all actions have |
2184 | finished. | |
2185 | ||
2186 | A prototype for @code{fsync} can be found in @file{unistd.h}. | |
2187 | ||
04b9968b | 2188 | This function is a cancellation point in multi-threaded programs. This |
dfd2257a UD |
2189 | is a problem if the thread allocates some resources (like memory, file |
2190 | descriptors, semaphores or whatever) at the time @code{fsync} is | |
19e4c7dd | 2191 | called. If the thread gets canceled these resources stay allocated |
04b9968b UD |
2192 | until the program ends. To avoid this, calls to @code{fsync} should be |
2193 | protected using cancellation handlers. | |
dfd2257a UD |
2194 | @c ref pthread_cleanup_push / pthread_cleanup_pop |
2195 | ||
49c091e5 | 2196 | The return value of the function is zero if no error occurred. Otherwise |
dfd2257a UD |
2197 | it is @math{-1} and the global variable @var{errno} is set to the |
2198 | following values: | |
2199 | @table @code | |
2200 | @item EBADF | |
2201 | The descriptor @var{fildes} is not valid. | |
2202 | ||
2203 | @item EINVAL | |
2204 | No synchronization is possible since the system does not implement this. | |
2205 | @end table | |
2206 | @end deftypefun | |
2207 | ||
2208 | Sometimes it is not even necessary to write all data associated with a | |
2209 | file descriptor. E.g., in database files which do not change in size it | |
2210 | is enough to write all the file content data to the device. | |
19e4c7dd | 2211 | Meta-information, like the modification time etc., are not that important |
dfd2257a | 2212 | and leaving such information uncommitted does not prevent a successful |
9739d2d5 | 2213 | recovery of the file in case of a problem. |
dfd2257a | 2214 | |
dfd2257a | 2215 | @deftypefun int fdatasync (int @var{fildes}) |
d08a7e4c | 2216 | @standards{POSIX, unistd.h} |
2cc3615c | 2217 | @safety{@prelim{}@mtsafe{}@assafe{}@acsafe{}} |
04b9968b | 2218 | When a call to the @code{fdatasync} function returns, it is ensured |
dfd2257a | 2219 | that all of the file data is written to the device. For all pending I/O |
04b9968b | 2220 | operations, the parts guaranteeing data integrity finished. |
dfd2257a UD |
2221 | |
2222 | Not all systems implement the @code{fdatasync} operation. On systems | |
2223 | missing this functionality @code{fdatasync} is emulated by a call to | |
2224 | @code{fsync} since the performed actions are a superset of those | |
19e4c7dd | 2225 | required by @code{fdatasync}. |
dfd2257a UD |
2226 | |
2227 | The prototype for @code{fdatasync} is in @file{unistd.h}. | |
2228 | ||
49c091e5 | 2229 | The return value of the function is zero if no error occurred. Otherwise |
dfd2257a UD |
2230 | it is @math{-1} and the global variable @var{errno} is set to the |
2231 | following values: | |
2232 | @table @code | |
2233 | @item EBADF | |
2234 | The descriptor @var{fildes} is not valid. | |
2235 | ||
2236 | @item EINVAL | |
2237 | No synchronization is possible since the system does not implement this. | |
2238 | @end table | |
2239 | @end deftypefun | |
2240 | ||
2241 | ||
b07d03e0 UD |
2242 | @node Asynchronous I/O |
2243 | @section Perform I/O Operations in Parallel | |
2244 | ||
2245 | The POSIX.1b standard defines a new set of I/O operations which can | |
9739d2d5 | 2246 | significantly reduce the time an application spends waiting for I/O. The |
b07d03e0 | 2247 | new functions allow a program to initiate one or more I/O operations and |
04b9968b UD |
2248 | then immediately resume normal work while the I/O operations are |
2249 | executed in parallel. This functionality is available if the | |
a3a4a74e | 2250 | @file{unistd.h} file defines the symbol @code{_POSIX_ASYNCHRONOUS_IO}. |
b07d03e0 UD |
2251 | |
2252 | These functions are part of the library with realtime functions named | |
2253 | @file{librt}. They are not actually part of the @file{libc} binary. | |
2254 | The implementation of these functions can be done using support in the | |
c756c71c UD |
2255 | kernel (if available) or using an implementation based on threads at |
2256 | userlevel. In the latter case it might be necessary to link applications | |
fed8f7f7 | 2257 | with the thread library @file{libpthread} in addition to @file{librt}. |
b07d03e0 | 2258 | |
c756c71c | 2259 | All AIO operations operate on files which were opened previously. There |
04b9968b | 2260 | might be arbitrarily many operations running for one file. The |
b07d03e0 UD |
2261 | asynchronous I/O operations are controlled using a data structure named |
2262 | @code{struct aiocb} (@dfn{AIO control block}). It is defined in | |
2263 | @file{aio.h} as follows. | |
2264 | ||
b07d03e0 | 2265 | @deftp {Data Type} {struct aiocb} |
d08a7e4c | 2266 | @standards{POSIX.1b, aio.h} |
b07d03e0 UD |
2267 | The POSIX.1b standard mandates that the @code{struct aiocb} structure |
2268 | contains at least the members described in the following table. There | |
04b9968b | 2269 | might be more elements which are used by the implementation, but |
19e4c7dd | 2270 | depending upon these elements is not portable and is highly deprecated. |
b07d03e0 UD |
2271 | |
2272 | @table @code | |
2273 | @item int aio_fildes | |
19e4c7dd AJ |
2274 | This element specifies the file descriptor to be used for the |
2275 | operation. It must be a legal descriptor, otherwise the operation will | |
2276 | fail. | |
b07d03e0 UD |
2277 | |
2278 | The device on which the file is opened must allow the seek operation. | |
2279 | I.e., it is not possible to use any of the AIO operations on devices | |
2280 | like terminals where an @code{lseek} call would lead to an error. | |
2281 | ||
2282 | @item off_t aio_offset | |
19e4c7dd | 2283 | This element specifies the offset in the file at which the operation (input |
fed8f7f7 | 2284 | or output) is performed. Since the operations are carried out in arbitrary |
b07d03e0 UD |
2285 | order and more than one operation for one file descriptor can be |
2286 | started, one cannot expect a current read/write position of the file | |
2287 | descriptor. | |
2288 | ||
2289 | @item volatile void *aio_buf | |
2290 | This is a pointer to the buffer with the data to be written or the place | |
c756c71c | 2291 | where the read data is stored. |
b07d03e0 UD |
2292 | |
2293 | @item size_t aio_nbytes | |
2294 | This element specifies the length of the buffer pointed to by @code{aio_buf}. | |
2295 | ||
2296 | @item int aio_reqprio | |
c756c71c | 2297 | If the platform has defined @code{_POSIX_PRIORITIZED_IO} and |
19e4c7dd | 2298 | @code{_POSIX_PRIORITY_SCHEDULING}, the AIO requests are |
b07d03e0 UD |
2299 | processed based on the current scheduling priority. The |
2300 | @code{aio_reqprio} element can then be used to lower the priority of the | |
2301 | AIO operation. | |
2302 | ||
2303 | @item struct sigevent aio_sigevent | |
2304 | This element specifies how the calling process is notified once the | |
fed8f7f7 | 2305 | operation terminates. If the @code{sigev_notify} element is |
19e4c7dd AJ |
2306 | @code{SIGEV_NONE}, no notification is sent. If it is @code{SIGEV_SIGNAL}, |
2307 | the signal determined by @code{sigev_signo} is sent. Otherwise, | |
2308 | @code{sigev_notify} must be @code{SIGEV_THREAD}. In this case, a thread | |
c756c71c | 2309 | is created which starts executing the function pointed to by |
b07d03e0 UD |
2310 | @code{sigev_notify_function}. |
2311 | ||
2312 | @item int aio_lio_opcode | |
2313 | This element is only used by the @code{lio_listio} and | |
04b9968b UD |
2314 | @code{lio_listio64} functions. Since these functions allow an |
2315 | arbitrary number of operations to start at once, and each operation can be | |
2316 | input or output (or nothing), the information must be stored in the | |
b07d03e0 UD |
2317 | control block. The possible values are: |
2318 | ||
2319 | @vtable @code | |
2320 | @item LIO_READ | |
2321 | Start a read operation. Read from the file at position | |
2322 | @code{aio_offset} and store the next @code{aio_nbytes} bytes in the | |
2323 | buffer pointed to by @code{aio_buf}. | |
2324 | ||
2325 | @item LIO_WRITE | |
2326 | Start a write operation. Write @code{aio_nbytes} bytes starting at | |
2327 | @code{aio_buf} into the file starting at position @code{aio_offset}. | |
2328 | ||
2329 | @item LIO_NOP | |
2330 | Do nothing for this control block. This value is useful sometimes when | |
2331 | an array of @code{struct aiocb} values contains holes, i.e., some of the | |
fed8f7f7 | 2332 | values must not be handled although the whole array is presented to the |
b07d03e0 UD |
2333 | @code{lio_listio} function. |
2334 | @end vtable | |
2335 | @end table | |
a3a4a74e | 2336 | |
fed8f7f7 | 2337 | When the sources are compiled using @code{_FILE_OFFSET_BITS == 64} on a |
19e4c7dd | 2338 | 32 bit machine, this type is in fact @code{struct aiocb64}, since the LFS |
a3a4a74e UD |
2339 | interface transparently replaces the @code{struct aiocb} definition. |
2340 | @end deftp | |
2341 | ||
19e4c7dd | 2342 | For use with the AIO functions defined in the LFS, there is a similar type |
a3a4a74e | 2343 | defined which replaces the types of the appropriate members with larger |
04b9968b | 2344 | types but otherwise is equivalent to @code{struct aiocb}. Particularly, |
a3a4a74e UD |
2345 | all member names are the same. |
2346 | ||
a3a4a74e | 2347 | @deftp {Data Type} {struct aiocb64} |
d08a7e4c | 2348 | @standards{POSIX.1b, aio.h} |
a3a4a74e UD |
2349 | @table @code |
2350 | @item int aio_fildes | |
2351 | This element specifies the file descriptor which is used for the | |
2352 | operation. It must be a legal descriptor since otherwise the operation | |
2353 | fails for obvious reasons. | |
2354 | ||
2355 | The device on which the file is opened must allow the seek operation. | |
2356 | I.e., it is not possible to use any of the AIO operations on devices | |
2357 | like terminals where an @code{lseek} call would lead to an error. | |
2358 | ||
2359 | @item off64_t aio_offset | |
04b9968b | 2360 | This element specifies at which offset in the file the operation (input |
a3a4a74e UD |
2361 | or output) is performed. Since the operation are carried in arbitrary |
2362 | order and more than one operation for one file descriptor can be | |
2363 | started, one cannot expect a current read/write position of the file | |
2364 | descriptor. | |
2365 | ||
2366 | @item volatile void *aio_buf | |
2367 | This is a pointer to the buffer with the data to be written or the place | |
19e4c7dd | 2368 | where the read data is stored. |
a3a4a74e UD |
2369 | |
2370 | @item size_t aio_nbytes | |
2371 | This element specifies the length of the buffer pointed to by @code{aio_buf}. | |
2372 | ||
2373 | @item int aio_reqprio | |
2374 | If for the platform @code{_POSIX_PRIORITIZED_IO} and | |
04b9968b | 2375 | @code{_POSIX_PRIORITY_SCHEDULING} are defined the AIO requests are |
a3a4a74e UD |
2376 | processed based on the current scheduling priority. The |
2377 | @code{aio_reqprio} element can then be used to lower the priority of the | |
2378 | AIO operation. | |
2379 | ||
2380 | @item struct sigevent aio_sigevent | |
2381 | This element specifies how the calling process is notified once the | |
9739d2d5 | 2382 | operation terminates. If the @code{sigev_notify} element is |
19e4c7dd AJ |
2383 | @code{SIGEV_NONE} no notification is sent. If it is @code{SIGEV_SIGNAL}, |
2384 | the signal determined by @code{sigev_signo} is sent. Otherwise, | |
a3a4a74e | 2385 | @code{sigev_notify} must be @code{SIGEV_THREAD} in which case a thread |
9739d2d5 | 2386 | is created which starts executing the function pointed to by |
a3a4a74e UD |
2387 | @code{sigev_notify_function}. |
2388 | ||
2389 | @item int aio_lio_opcode | |
2390 | This element is only used by the @code{lio_listio} and | |
9739d2d5 | 2391 | @code{lio_listio64} functions. Since these functions allow an |
04b9968b UD |
2392 | arbitrary number of operations to start at once, and since each operation can be |
2393 | input or output (or nothing), the information must be stored in the | |
a3a4a74e UD |
2394 | control block. See the description of @code{struct aiocb} for a description |
2395 | of the possible values. | |
2396 | @end table | |
2397 | ||
2398 | When the sources are compiled using @code{_FILE_OFFSET_BITS == 64} on a | |
19e4c7dd AJ |
2399 | 32 bit machine, this type is available under the name @code{struct |
2400 | aiocb64}, since the LFS transparently replaces the old interface. | |
b07d03e0 UD |
2401 | @end deftp |
2402 | ||
2403 | @menu | |
a3a4a74e UD |
2404 | * Asynchronous Reads/Writes:: Asynchronous Read and Write Operations. |
2405 | * Status of AIO Operations:: Getting the Status of AIO Operations. | |
2406 | * Synchronizing AIO Operations:: Getting into a consistent state. | |
04b9968b | 2407 | * Cancel AIO Operations:: Cancellation of AIO Operations. |
a3a4a74e | 2408 | * Configuration of AIO:: How to optimize the AIO implementation. |
b07d03e0 UD |
2409 | @end menu |
2410 | ||
a3a4a74e UD |
2411 | @node Asynchronous Reads/Writes |
2412 | @subsection Asynchronous Read and Write Operations | |
b07d03e0 | 2413 | |
b07d03e0 | 2414 | @deftypefun int aio_read (struct aiocb *@var{aiocbp}) |
d08a7e4c | 2415 | @standards{POSIX.1b, aio.h} |
2cc3615c AO |
2416 | @safety{@prelim{}@mtsafe{}@asunsafe{@asulock{} @ascuheap{}}@acunsafe{@aculock{} @acsmem{}}} |
2417 | @c Calls aio_enqueue_request. | |
2418 | @c aio_enqueue_request @asulock @ascuheap @aculock @acsmem | |
2419 | @c pthread_self ok | |
2420 | @c pthread_getschedparam @asulock @aculock | |
2421 | @c lll_lock (pthread descriptor's lock) @asulock @aculock | |
2422 | @c sched_getparam ok | |
2423 | @c sched_getscheduler ok | |
2424 | @c lll_unlock @aculock | |
2425 | @c pthread_mutex_lock (aio_requests_mutex) @asulock @aculock | |
2426 | @c get_elem @ascuheap @acsmem [@asucorrupt @acucorrupt] | |
2427 | @c realloc @ascuheap @acsmem | |
2428 | @c calloc @ascuheap @acsmem | |
2429 | @c aio_create_helper_thread @asulock @ascuheap @aculock @acsmem | |
2430 | @c pthread_attr_init ok | |
2431 | @c pthread_attr_setdetachstate ok | |
2432 | @c pthread_get_minstack ok | |
2433 | @c pthread_attr_setstacksize ok | |
2434 | @c sigfillset ok | |
2435 | @c memset ok | |
2436 | @c sigdelset ok | |
2437 | @c SYSCALL rt_sigprocmask ok | |
2438 | @c pthread_create @asulock @ascuheap @aculock @acsmem | |
2439 | @c lll_lock (default_pthread_attr_lock) @asulock @aculock | |
2440 | @c alloca/malloc @ascuheap @acsmem | |
2441 | @c lll_unlock @aculock | |
2442 | @c allocate_stack @asulock @ascuheap @aculock @acsmem | |
2443 | @c getpagesize dup | |
2444 | @c lll_lock (default_pthread_attr_lock) @asulock @aculock | |
2445 | @c lll_unlock @aculock | |
2446 | @c _dl_allocate_tls @ascuheap @acsmem | |
2447 | @c _dl_allocate_tls_storage @ascuheap @acsmem | |
2448 | @c memalign @ascuheap @acsmem | |
2449 | @c memset ok | |
2450 | @c allocate_dtv dup | |
2451 | @c free @ascuheap @acsmem | |
2452 | @c allocate_dtv @ascuheap @acsmem | |
2453 | @c calloc @ascuheap @acsmem | |
2454 | @c INSTALL_DTV ok | |
2455 | @c list_add dup | |
2456 | @c get_cached_stack | |
2457 | @c lll_lock (stack_cache_lock) @asulock @aculock | |
2458 | @c list_for_each ok | |
2459 | @c list_entry dup | |
2460 | @c FREE_P dup | |
2461 | @c stack_list_del dup | |
2462 | @c stack_list_add dup | |
2463 | @c lll_unlock @aculock | |
2464 | @c _dl_allocate_tls_init ok | |
2465 | @c GET_DTV ok | |
2466 | @c mmap ok | |
2467 | @c atomic_increment_val ok | |
2468 | @c munmap ok | |
2469 | @c change_stack_perm ok | |
2470 | @c mprotect ok | |
2471 | @c mprotect ok | |
2472 | @c stack_list_del dup | |
2473 | @c _dl_deallocate_tls dup | |
2474 | @c munmap ok | |
2475 | @c THREAD_COPY_STACK_GUARD ok | |
2476 | @c THREAD_COPY_POINTER_GUARD ok | |
2477 | @c atomic_exchange_acq ok | |
2478 | @c lll_futex_wake ok | |
2479 | @c deallocate_stack @asulock @ascuheap @aculock @acsmem | |
2480 | @c lll_lock (state_cache_lock) @asulock @aculock | |
2481 | @c stack_list_del ok | |
2482 | @c atomic_write_barrier ok | |
2483 | @c list_del ok | |
2484 | @c atomic_write_barrier ok | |
2485 | @c queue_stack @ascuheap @acsmem | |
2486 | @c stack_list_add ok | |
2487 | @c atomic_write_barrier ok | |
2488 | @c list_add ok | |
2489 | @c atomic_write_barrier ok | |
2490 | @c free_stacks @ascuheap @acsmem | |
2491 | @c list_for_each_prev_safe ok | |
2492 | @c list_entry ok | |
2493 | @c FREE_P ok | |
2494 | @c stack_list_del dup | |
2495 | @c _dl_deallocate_tls dup | |
2496 | @c munmap ok | |
2497 | @c _dl_deallocate_tls @ascuheap @acsmem | |
2498 | @c free @ascuheap @acsmem | |
2499 | @c lll_unlock @aculock | |
2500 | @c create_thread @asulock @ascuheap @aculock @acsmem | |
2501 | @c td_eventword | |
2502 | @c td_eventmask | |
2503 | @c do_clone @asulock @ascuheap @aculock @acsmem | |
2504 | @c PREPARE_CREATE ok | |
2505 | @c lll_lock (pd->lock) @asulock @aculock | |
2506 | @c atomic_increment ok | |
2507 | @c clone ok | |
2508 | @c atomic_decrement ok | |
2509 | @c atomic_exchange_acq ok | |
2510 | @c lll_futex_wake ok | |
2511 | @c deallocate_stack dup | |
2512 | @c sched_setaffinity ok | |
2513 | @c tgkill ok | |
2514 | @c sched_setscheduler ok | |
2515 | @c atomic_compare_and_exchange_bool_acq ok | |
2516 | @c nptl_create_event ok | |
2517 | @c lll_unlock (pd->lock) @aculock | |
2518 | @c free @ascuheap @acsmem | |
2519 | @c pthread_attr_destroy ok (cpuset won't be set, so free isn't called) | |
2520 | @c add_request_to_runlist ok | |
2521 | @c pthread_cond_signal ok | |
2522 | @c aio_free_request ok | |
2523 | @c pthread_mutex_unlock @aculock | |
2524 | ||
2525 | @c (in the new thread, initiated with clone) | |
2526 | @c start_thread ok | |
2527 | @c HP_TIMING_NOW ok | |
2528 | @c ctype_init @mtslocale | |
2529 | @c atomic_exchange_acq ok | |
2530 | @c lll_futex_wake ok | |
2531 | @c sigemptyset ok | |
2532 | @c sigaddset ok | |
2533 | @c setjmp ok | |
2534 | @c CANCEL_ASYNC -> pthread_enable_asynccancel ok | |
2535 | @c do_cancel ok | |
2536 | @c pthread_unwind ok | |
2537 | @c Unwind_ForcedUnwind or longjmp ok [@ascuheap @acsmem?] | |
2538 | @c lll_lock @asulock @aculock | |
2539 | @c lll_unlock @asulock @aculock | |
2540 | @c CANCEL_RESET -> pthread_disable_asynccancel ok | |
2541 | @c lll_futex_wait ok | |
2542 | @c ->start_routine ok ----- | |
2543 | @c call_tls_dtors @asulock @ascuheap @aculock @acsmem | |
2544 | @c user-supplied dtor | |
2545 | @c rtld_lock_lock_recursive (dl_load_lock) @asulock @aculock | |
2546 | @c rtld_lock_unlock_recursive @aculock | |
2547 | @c free @ascuheap @acsmem | |
2548 | @c nptl_deallocate_tsd @ascuheap @acsmem | |
2549 | @c tsd user-supplied dtors ok | |
2550 | @c free @ascuheap @acsmem | |
2551 | @c libc_thread_freeres | |
2552 | @c libc_thread_subfreeres ok | |
2553 | @c atomic_decrement_and_test ok | |
2554 | @c td_eventword ok | |
2555 | @c td_eventmask ok | |
2556 | @c atomic_compare_exchange_bool_acq ok | |
2557 | @c nptl_death_event ok | |
2558 | @c lll_robust_dead ok | |
2559 | @c getpagesize ok | |
2560 | @c madvise ok | |
2561 | @c free_tcb @asulock @ascuheap @aculock @acsmem | |
2562 | @c free @ascuheap @acsmem | |
2563 | @c deallocate_stack @asulock @ascuheap @aculock @acsmem | |
2564 | @c lll_futex_wait ok | |
2565 | @c exit_thread_inline ok | |
2566 | @c syscall(exit) ok | |
2567 | ||
04b9968b UD |
2568 | This function initiates an asynchronous read operation. It |
2569 | immediately returns after the operation was enqueued or when an | |
fed8f7f7 | 2570 | error was encountered. |
b07d03e0 | 2571 | |
a3a4a74e | 2572 | The first @code{aiocbp->aio_nbytes} bytes of the file for which |
c756c71c UD |
2573 | @code{aiocbp->aio_fildes} is a descriptor are written to the buffer |
2574 | starting at @code{aiocbp->aio_buf}. Reading starts at the absolute | |
2575 | position @code{aiocbp->aio_offset} in the file. | |
b07d03e0 UD |
2576 | |
2577 | If prioritized I/O is supported by the platform the | |
2578 | @code{aiocbp->aio_reqprio} value is used to adjust the priority before | |
2579 | the request is actually enqueued. | |
2580 | ||
2581 | The calling process is notified about the termination of the read | |
2582 | request according to the @code{aiocbp->aio_sigevent} value. | |
2583 | ||
04b9968b | 2584 | When @code{aio_read} returns, the return value is zero if no error |
b07d03e0 | 2585 | occurred that can be found before the process is enqueued. If such an |
04b9968b UD |
2586 | early error is found, the function returns @math{-1} and sets |
2587 | @code{errno} to one of the following values: | |
b07d03e0 UD |
2588 | |
2589 | @table @code | |
2590 | @item EAGAIN | |
2591 | The request was not enqueued due to (temporarily) exceeded resource | |
2592 | limitations. | |
2593 | @item ENOSYS | |
2594 | The @code{aio_read} function is not implemented. | |
2595 | @item EBADF | |
2596 | The @code{aiocbp->aio_fildes} descriptor is not valid. This condition | |
04b9968b | 2597 | need not be recognized before enqueueing the request and so this error |
fed8f7f7 | 2598 | might also be signaled asynchronously. |
b07d03e0 UD |
2599 | @item EINVAL |
2600 | The @code{aiocbp->aio_offset} or @code{aiocbp->aio_reqpiro} value is | |
2601 | invalid. This condition need not be recognized before enqueueing the | |
49c091e5 | 2602 | request and so this error might also be signaled asynchronously. |
b07d03e0 UD |
2603 | @end table |
2604 | ||
04b9968b UD |
2605 | If @code{aio_read} returns zero, the current status of the request |
2606 | can be queried using @code{aio_error} and @code{aio_return} functions. | |
2607 | As long as the value returned by @code{aio_error} is @code{EINPROGRESS} | |
2608 | the operation has not yet completed. If @code{aio_error} returns zero, | |
78759725 UD |
2609 | the operation successfully terminated, otherwise the value is to be |
2610 | interpreted as an error code. If the function terminated, the result of | |
2611 | the operation can be obtained using a call to @code{aio_return}. The | |
2612 | returned value is the same as an equivalent call to @code{read} would | |
04b9968b | 2613 | have returned. Possible error codes returned by @code{aio_error} are: |
b07d03e0 UD |
2614 | |
2615 | @table @code | |
2616 | @item EBADF | |
2617 | The @code{aiocbp->aio_fildes} descriptor is not valid. | |
2618 | @item ECANCELED | |
19e4c7dd | 2619 | The operation was canceled before the operation was finished |
b07d03e0 UD |
2620 | (@pxref{Cancel AIO Operations}) |
2621 | @item EINVAL | |
2622 | The @code{aiocbp->aio_offset} value is invalid. | |
2623 | @end table | |
a3a4a74e UD |
2624 | |
2625 | When the sources are compiled with @code{_FILE_OFFSET_BITS == 64} this | |
2626 | function is in fact @code{aio_read64} since the LFS interface transparently | |
2627 | replaces the normal implementation. | |
b07d03e0 UD |
2628 | @end deftypefun |
2629 | ||
8ded91fb | 2630 | @deftypefun int aio_read64 (struct aiocb64 *@var{aiocbp}) |
d08a7e4c | 2631 | @standards{Unix98, aio.h} |
2cc3615c | 2632 | @safety{@prelim{}@mtsafe{}@asunsafe{@asulock{} @ascuheap{}}@acunsafe{@aculock{} @acsmem{}}} |
b07d03e0 | 2633 | This function is similar to the @code{aio_read} function. The only |
19e4c7dd AJ |
2634 | difference is that on @w{32 bit} machines, the file descriptor should |
2635 | be opened in the large file mode. Internally, @code{aio_read64} uses | |
a3a4a74e UD |
2636 | functionality equivalent to @code{lseek64} (@pxref{File Position |
2637 | Primitive}) to position the file descriptor correctly for the reading, | |
9739d2d5 | 2638 | as opposed to the @code{lseek} functionality used in @code{aio_read}. |
a3a4a74e | 2639 | |
19e4c7dd | 2640 | When the sources are compiled with @code{_FILE_OFFSET_BITS == 64}, this |
a3a4a74e | 2641 | function is available under the name @code{aio_read} and so transparently |
04b9968b | 2642 | replaces the interface for small files on 32 bit machines. |
b07d03e0 UD |
2643 | @end deftypefun |
2644 | ||
19e4c7dd | 2645 | To write data asynchronously to a file, there exists an equivalent pair |
a3a4a74e UD |
2646 | of functions with a very similar interface. |
2647 | ||
a3a4a74e | 2648 | @deftypefun int aio_write (struct aiocb *@var{aiocbp}) |
d08a7e4c | 2649 | @standards{POSIX.1b, aio.h} |
2cc3615c | 2650 | @safety{@prelim{}@mtsafe{}@asunsafe{@asulock{} @ascuheap{}}@acunsafe{@aculock{} @acsmem{}}} |
a3a4a74e UD |
2651 | This function initiates an asynchronous write operation. The function |
2652 | call immediately returns after the operation was enqueued or if before | |
fed8f7f7 | 2653 | this happens an error was encountered. |
a3a4a74e UD |
2654 | |
2655 | The first @code{aiocbp->aio_nbytes} bytes from the buffer starting at | |
2656 | @code{aiocbp->aio_buf} are written to the file for which | |
9dcc8f11 | 2657 | @code{aiocbp->aio_fildes} is a descriptor, starting at the absolute |
a3a4a74e UD |
2658 | position @code{aiocbp->aio_offset} in the file. |
2659 | ||
19e4c7dd | 2660 | If prioritized I/O is supported by the platform, the |
a3a4a74e UD |
2661 | @code{aiocbp->aio_reqprio} value is used to adjust the priority before |
2662 | the request is actually enqueued. | |
2663 | ||
2664 | The calling process is notified about the termination of the read | |
2665 | request according to the @code{aiocbp->aio_sigevent} value. | |
2666 | ||
19e4c7dd | 2667 | When @code{aio_write} returns, the return value is zero if no error |
a3a4a74e UD |
2668 | occurred that can be found before the process is enqueued. If such an |
2669 | early error is found the function returns @math{-1} and sets | |
2670 | @code{errno} to one of the following values. | |
2671 | ||
2672 | @table @code | |
2673 | @item EAGAIN | |
2674 | The request was not enqueued due to (temporarily) exceeded resource | |
2675 | limitations. | |
2676 | @item ENOSYS | |
2677 | The @code{aio_write} function is not implemented. | |
2678 | @item EBADF | |
2679 | The @code{aiocbp->aio_fildes} descriptor is not valid. This condition | |
19e4c7dd | 2680 | may not be recognized before enqueueing the request, and so this error |
fed8f7f7 | 2681 | might also be signaled asynchronously. |
a3a4a74e | 2682 | @item EINVAL |
19e4c7dd AJ |
2683 | The @code{aiocbp->aio_offset} or @code{aiocbp->aio_reqprio} value is |
2684 | invalid. This condition may not be recognized before enqueueing the | |
fed8f7f7 | 2685 | request and so this error might also be signaled asynchronously. |
a3a4a74e UD |
2686 | @end table |
2687 | ||
19e4c7dd | 2688 | In the case @code{aio_write} returns zero, the current status of the |
9739d2d5 | 2689 | request can be queried using the @code{aio_error} and @code{aio_return} |
c756c71c | 2690 | functions. As long as the value returned by @code{aio_error} is |
a3a4a74e | 2691 | @code{EINPROGRESS} the operation has not yet completed. If |
19e4c7dd | 2692 | @code{aio_error} returns zero, the operation successfully terminated, |
a3a4a74e | 2693 | otherwise the value is to be interpreted as an error code. If the |
9739d2d5 | 2694 | function terminated, the result of the operation can be obtained using a call |
a3a4a74e | 2695 | to @code{aio_return}. The returned value is the same as an equivalent |
19e4c7dd | 2696 | call to @code{read} would have returned. Possible error codes returned |
a3a4a74e UD |
2697 | by @code{aio_error} are: |
2698 | ||
2699 | @table @code | |
2700 | @item EBADF | |
2701 | The @code{aiocbp->aio_fildes} descriptor is not valid. | |
2702 | @item ECANCELED | |
19e4c7dd | 2703 | The operation was canceled before the operation was finished. |
a3a4a74e UD |
2704 | (@pxref{Cancel AIO Operations}) |
2705 | @item EINVAL | |
2706 | The @code{aiocbp->aio_offset} value is invalid. | |
2707 | @end table | |
2708 | ||
19e4c7dd | 2709 | When the sources are compiled with @code{_FILE_OFFSET_BITS == 64}, this |
a3a4a74e UD |
2710 | function is in fact @code{aio_write64} since the LFS interface transparently |
2711 | replaces the normal implementation. | |
2712 | @end deftypefun | |
2713 | ||
8ded91fb | 2714 | @deftypefun int aio_write64 (struct aiocb64 *@var{aiocbp}) |
d08a7e4c | 2715 | @standards{Unix98, aio.h} |
2cc3615c | 2716 | @safety{@prelim{}@mtsafe{}@asunsafe{@asulock{} @ascuheap{}}@acunsafe{@aculock{} @acsmem{}}} |
a3a4a74e | 2717 | This function is similar to the @code{aio_write} function. The only |
04b9968b | 2718 | difference is that on @w{32 bit} machines the file descriptor should |
a3a4a74e UD |
2719 | be opened in the large file mode. Internally @code{aio_write64} uses |
2720 | functionality equivalent to @code{lseek64} (@pxref{File Position | |
2721 | Primitive}) to position the file descriptor correctly for the writing, | |
9739d2d5 | 2722 | as opposed to the @code{lseek} functionality used in @code{aio_write}. |
a3a4a74e | 2723 | |
19e4c7dd | 2724 | When the sources are compiled with @code{_FILE_OFFSET_BITS == 64}, this |
a3a4a74e | 2725 | function is available under the name @code{aio_write} and so transparently |
04b9968b | 2726 | replaces the interface for small files on 32 bit machines. |
a3a4a74e UD |
2727 | @end deftypefun |
2728 | ||
19e4c7dd AJ |
2729 | Besides these functions with the more or less traditional interface, |
2730 | POSIX.1b also defines a function which can initiate more than one | |
2731 | operation at a time, and which can handle freely mixed read and write | |
2732 | operations. It is therefore similar to a combination of @code{readv} and | |
a3a4a74e UD |
2733 | @code{writev}. |
2734 | ||
a3a4a74e | 2735 | @deftypefun int lio_listio (int @var{mode}, struct aiocb *const @var{list}[], int @var{nent}, struct sigevent *@var{sig}) |
d08a7e4c | 2736 | @standards{POSIX.1b, aio.h} |
2cc3615c AO |
2737 | @safety{@prelim{}@mtsafe{}@asunsafe{@asulock{} @ascuheap{}}@acunsafe{@aculock{} @acsmem{}}} |
2738 | @c Call lio_listio_internal, that takes the aio_requests_mutex lock and | |
2739 | @c enqueues each request. Then, it waits for notification or prepares | |
2740 | @c for it before releasing the lock. Even though it performs memory | |
2741 | @c allocation and locking of its own, it doesn't add any classes of | |
2742 | @c safety issues that aren't already covered by aio_enqueue_request. | |
a3a4a74e UD |
2743 | The @code{lio_listio} function can be used to enqueue an arbitrary |
2744 | number of read and write requests at one time. The requests can all be | |
2745 | meant for the same file, all for different files or every solution in | |
2746 | between. | |
2747 | ||
2748 | @code{lio_listio} gets the @var{nent} requests from the array pointed to | |
19e4c7dd | 2749 | by @var{list}. The operation to be performed is determined by the |
a3a4a74e | 2750 | @code{aio_lio_opcode} member in each element of @var{list}. If this |
19e4c7dd | 2751 | field is @code{LIO_READ} a read operation is enqueued, similar to a call |
a3a4a74e UD |
2752 | of @code{aio_read} for this element of the array (except that the way |
2753 | the termination is signalled is different, as we will see below). If | |
19e4c7dd | 2754 | the @code{aio_lio_opcode} member is @code{LIO_WRITE} a write operation |
a3a4a74e UD |
2755 | is enqueued. Otherwise the @code{aio_lio_opcode} must be @code{LIO_NOP} |
2756 | in which case this element of @var{list} is simply ignored. This | |
2757 | ``operation'' is useful in situations where one has a fixed array of | |
2758 | @code{struct aiocb} elements from which only a few need to be handled at | |
2759 | a time. Another situation is where the @code{lio_listio} call was | |
19e4c7dd | 2760 | canceled before all requests are processed (@pxref{Cancel AIO |
a3a4a74e UD |
2761 | Operations}) and the remaining requests have to be reissued. |
2762 | ||
fed8f7f7 | 2763 | The other members of each element of the array pointed to by |
a3a4a74e UD |
2764 | @code{list} must have values suitable for the operation as described in |
2765 | the documentation for @code{aio_read} and @code{aio_write} above. | |
2766 | ||
2767 | The @var{mode} argument determines how @code{lio_listio} behaves after | |
2768 | having enqueued all the requests. If @var{mode} is @code{LIO_WAIT} it | |
2769 | waits until all requests terminated. Otherwise @var{mode} must be | |
fed8f7f7 | 2770 | @code{LIO_NOWAIT} and in this case the function returns immediately after |
a3a4a74e UD |
2771 | having enqueued all the requests. In this case the caller gets a |
2772 | notification of the termination of all requests according to the | |
2773 | @var{sig} parameter. If @var{sig} is @code{NULL} no notification is | |
9739d2d5 | 2774 | sent. Otherwise a signal is sent or a thread is started, just as |
a3a4a74e UD |
2775 | described in the description for @code{aio_read} or @code{aio_write}. |
2776 | ||
19e4c7dd | 2777 | If @var{mode} is @code{LIO_WAIT}, the return value of @code{lio_listio} |
a3a4a74e | 2778 | is @math{0} when all requests completed successfully. Otherwise the |
9739d2d5 | 2779 | function returns @math{-1} and @code{errno} is set accordingly. To find |
a3a4a74e UD |
2780 | out which request or requests failed one has to use the @code{aio_error} |
2781 | function on all the elements of the array @var{list}. | |
2782 | ||
19e4c7dd | 2783 | In case @var{mode} is @code{LIO_NOWAIT}, the function returns @math{0} if |
a3a4a74e UD |
2784 | all requests were enqueued correctly. The current state of the requests |
2785 | can be found using @code{aio_error} and @code{aio_return} as described | |
19e4c7dd | 2786 | above. If @code{lio_listio} returns @math{-1} in this mode, the |
a3a4a74e | 2787 | global variable @code{errno} is set accordingly. If a request did not |
19e4c7dd AJ |
2788 | yet terminate, a call to @code{aio_error} returns @code{EINPROGRESS}. If |
2789 | the value is different, the request is finished and the error value (or | |
a3a4a74e UD |
2790 | @math{0}) is returned and the result of the operation can be retrieved |
2791 | using @code{aio_return}. | |
2792 | ||
2793 | Possible values for @code{errno} are: | |
2794 | ||
2795 | @table @code | |
2796 | @item EAGAIN | |
19e4c7dd | 2797 | The resources necessary to queue all the requests are not available at |
a3a4a74e | 2798 | the moment. The error status for each element of @var{list} must be |
19e4c7dd | 2799 | checked to determine which request failed. |
a3a4a74e | 2800 | |
fed8f7f7 | 2801 | Another reason could be that the system wide limit of AIO requests is |
a7a93d50 | 2802 | exceeded. This cannot be the case for the implementation on @gnusystems{} |
a3a4a74e UD |
2803 | since no arbitrary limits exist. |
2804 | @item EINVAL | |
2805 | The @var{mode} parameter is invalid or @var{nent} is larger than | |
2806 | @code{AIO_LISTIO_MAX}. | |
2807 | @item EIO | |
2808 | One or more of the request's I/O operations failed. The error status of | |
19e4c7dd | 2809 | each request should be checked to determine which one failed. |
a3a4a74e UD |
2810 | @item ENOSYS |
2811 | The @code{lio_listio} function is not supported. | |
2812 | @end table | |
2813 | ||
2814 | If the @var{mode} parameter is @code{LIO_NOWAIT} and the caller cancels | |
19e4c7dd | 2815 | a request, the error status for this request returned by |
a3a4a74e UD |
2816 | @code{aio_error} is @code{ECANCELED}. |
2817 | ||
19e4c7dd | 2818 | When the sources are compiled with @code{_FILE_OFFSET_BITS == 64}, this |
a3a4a74e UD |
2819 | function is in fact @code{lio_listio64} since the LFS interface |
2820 | transparently replaces the normal implementation. | |
2821 | @end deftypefun | |
2822 | ||
8ded91fb | 2823 | @deftypefun int lio_listio64 (int @var{mode}, struct aiocb64 *const @var{list}[], int @var{nent}, struct sigevent *@var{sig}) |
d08a7e4c | 2824 | @standards{Unix98, aio.h} |
2cc3615c | 2825 | @safety{@prelim{}@mtsafe{}@asunsafe{@asulock{} @ascuheap{}}@acunsafe{@aculock{} @acsmem{}}} |
19e4c7dd AJ |
2826 | This function is similar to the @code{lio_listio} function. The only |
2827 | difference is that on @w{32 bit} machines, the file descriptor should | |
2828 | be opened in the large file mode. Internally, @code{lio_listio64} uses | |
a3a4a74e UD |
2829 | functionality equivalent to @code{lseek64} (@pxref{File Position |
2830 | Primitive}) to position the file descriptor correctly for the reading or | |
9739d2d5 | 2831 | writing, as opposed to the @code{lseek} functionality used in |
a3a4a74e UD |
2832 | @code{lio_listio}. |
2833 | ||
19e4c7dd | 2834 | When the sources are compiled with @code{_FILE_OFFSET_BITS == 64}, this |
a3a4a74e | 2835 | function is available under the name @code{lio_listio} and so |
04b9968b | 2836 | transparently replaces the interface for small files on 32 bit |
a3a4a74e UD |
2837 | machines. |
2838 | @end deftypefun | |
2839 | ||
2840 | @node Status of AIO Operations | |
2841 | @subsection Getting the Status of AIO Operations | |
2842 | ||
fed8f7f7 | 2843 | As already described in the documentation of the functions in the last |
04b9968b UD |
2844 | section, it must be possible to get information about the status of an I/O |
2845 | request. When the operation is performed truly asynchronously (as with | |
19e4c7dd AJ |
2846 | @code{aio_read} and @code{aio_write} and with @code{lio_listio} when the |
2847 | mode is @code{LIO_NOWAIT}), one sometimes needs to know whether a | |
2848 | specific request already terminated and if so, what the result was. | |
04b9968b | 2849 | The following two functions allow you to get this kind of information. |
a3a4a74e | 2850 | |
a3a4a74e | 2851 | @deftypefun int aio_error (const struct aiocb *@var{aiocbp}) |
d08a7e4c | 2852 | @standards{POSIX.1b, aio.h} |
2cc3615c | 2853 | @safety{@prelim{}@mtsafe{}@assafe{}@acsafe{}} |
a3a4a74e | 2854 | This function determines the error state of the request described by the |
fed8f7f7 | 2855 | @code{struct aiocb} variable pointed to by @var{aiocbp}. If the |
a3a4a74e UD |
2856 | request has not yet terminated the value returned is always |
2857 | @code{EINPROGRESS}. Once the request has terminated the value | |
2858 | @code{aio_error} returns is either @math{0} if the request completed | |
fed8f7f7 | 2859 | successfully or it returns the value which would be stored in the |
a3a4a74e UD |
2860 | @code{errno} variable if the request would have been done using |
2861 | @code{read}, @code{write}, or @code{fsync}. | |
2862 | ||
2863 | The function can return @code{ENOSYS} if it is not implemented. It | |
2864 | could also return @code{EINVAL} if the @var{aiocbp} parameter does not | |
2865 | refer to an asynchronous operation whose return status is not yet known. | |
2866 | ||
2867 | When the sources are compiled with @code{_FILE_OFFSET_BITS == 64} this | |
2868 | function is in fact @code{aio_error64} since the LFS interface | |
2869 | transparently replaces the normal implementation. | |
2870 | @end deftypefun | |
2871 | ||
a3a4a74e | 2872 | @deftypefun int aio_error64 (const struct aiocb64 *@var{aiocbp}) |
d08a7e4c | 2873 | @standards{Unix98, aio.h} |
2cc3615c | 2874 | @safety{@prelim{}@mtsafe{}@assafe{}@acsafe{}} |
a3a4a74e UD |
2875 | This function is similar to @code{aio_error} with the only difference |
2876 | that the argument is a reference to a variable of type @code{struct | |
2877 | aiocb64}. | |
2878 | ||
2879 | When the sources are compiled with @code{_FILE_OFFSET_BITS == 64} this | |
2880 | function is available under the name @code{aio_error} and so | |
04b9968b | 2881 | transparently replaces the interface for small files on 32 bit |
a3a4a74e UD |
2882 | machines. |
2883 | @end deftypefun | |
2884 | ||
8ded91fb | 2885 | @deftypefun ssize_t aio_return (struct aiocb *@var{aiocbp}) |
d08a7e4c | 2886 | @standards{POSIX.1b, aio.h} |
2cc3615c | 2887 | @safety{@prelim{}@mtsafe{}@assafe{}@acsafe{}} |
a3a4a74e UD |
2888 | This function can be used to retrieve the return status of the operation |
2889 | carried out by the request described in the variable pointed to by | |
2890 | @var{aiocbp}. As long as the error status of this request as returned | |
9739d2d5 | 2891 | by @code{aio_error} is @code{EINPROGRESS} the return value of this function is |
a3a4a74e UD |
2892 | undefined. |
2893 | ||
fed8f7f7 UD |
2894 | Once the request is finished this function can be used exactly once to |
2895 | retrieve the return value. Following calls might lead to undefined | |
19e4c7dd | 2896 | behavior. The return value itself is the value which would have been |
a3a4a74e UD |
2897 | returned by the @code{read}, @code{write}, or @code{fsync} call. |
2898 | ||
2899 | The function can return @code{ENOSYS} if it is not implemented. It | |
2900 | could also return @code{EINVAL} if the @var{aiocbp} parameter does not | |
2901 | refer to an asynchronous operation whose return status is not yet known. | |
2902 | ||
2903 | When the sources are compiled with @code{_FILE_OFFSET_BITS == 64} this | |
2904 | function is in fact @code{aio_return64} since the LFS interface | |
2905 | transparently replaces the normal implementation. | |
2906 | @end deftypefun | |
2907 | ||
8ded91fb | 2908 | @deftypefun ssize_t aio_return64 (struct aiocb64 *@var{aiocbp}) |
d08a7e4c | 2909 | @standards{Unix98, aio.h} |
2cc3615c | 2910 | @safety{@prelim{}@mtsafe{}@assafe{}@acsafe{}} |
a3a4a74e UD |
2911 | This function is similar to @code{aio_return} with the only difference |
2912 | that the argument is a reference to a variable of type @code{struct | |
2913 | aiocb64}. | |
2914 | ||
2915 | When the sources are compiled with @code{_FILE_OFFSET_BITS == 64} this | |
2916 | function is available under the name @code{aio_return} and so | |
04b9968b | 2917 | transparently replaces the interface for small files on 32 bit |
a3a4a74e UD |
2918 | machines. |
2919 | @end deftypefun | |
2920 | ||
2921 | @node Synchronizing AIO Operations | |
2922 | @subsection Getting into a Consistent State | |
2923 | ||
2924 | When dealing with asynchronous operations it is sometimes necessary to | |
fed8f7f7 | 2925 | get into a consistent state. This would mean for AIO that one wants to |
9739d2d5 | 2926 | know whether a certain request or a group of requests were processed. |
a3a4a74e | 2927 | This could be done by waiting for the notification sent by the system |
04b9968b | 2928 | after the operation terminated, but this sometimes would mean wasting |
a3a4a74e UD |
2929 | resources (mainly computation time). Instead POSIX.1b defines two |
2930 | functions which will help with most kinds of consistency. | |
2931 | ||
2932 | The @code{aio_fsync} and @code{aio_fsync64} functions are only available | |
19e4c7dd | 2933 | if the symbol @code{_POSIX_SYNCHRONIZED_IO} is defined in @file{unistd.h}. |
a3a4a74e UD |
2934 | |
2935 | @cindex synchronizing | |
a3a4a74e | 2936 | @deftypefun int aio_fsync (int @var{op}, struct aiocb *@var{aiocbp}) |
d08a7e4c | 2937 | @standards{POSIX.1b, aio.h} |
2cc3615c AO |
2938 | @safety{@prelim{}@mtsafe{}@asunsafe{@asulock{} @ascuheap{}}@acunsafe{@aculock{} @acsmem{}}} |
2939 | @c After fcntl to check that the FD is open, it calls | |
2940 | @c aio_enqueue_request. | |
9739d2d5 | 2941 | Calling this function forces all I/O operations queued at the |
fed8f7f7 | 2942 | time of the function call operating on the file descriptor |
a3a4a74e | 2943 | @code{aiocbp->aio_fildes} into the synchronized I/O completion state |
04b9968b | 2944 | (@pxref{Synchronizing I/O}). The @code{aio_fsync} function returns |
a3a4a74e UD |
2945 | immediately but the notification through the method described in |
2946 | @code{aiocbp->aio_sigevent} will happen only after all requests for this | |
04b9968b | 2947 | file descriptor have terminated and the file is synchronized. This also |
a3a4a74e | 2948 | means that requests for this very same file descriptor which are queued |
04b9968b | 2949 | after the synchronization request are not affected. |
a3a4a74e UD |
2950 | |
2951 | If @var{op} is @code{O_DSYNC} the synchronization happens as with a call | |
2952 | to @code{fdatasync}. Otherwise @var{op} should be @code{O_SYNC} and | |
fed8f7f7 | 2953 | the synchronization happens as with @code{fsync}. |
a3a4a74e | 2954 | |
19e4c7dd | 2955 | As long as the synchronization has not happened, a call to |
a3a4a74e | 2956 | @code{aio_error} with the reference to the object pointed to by |
fed8f7f7 UD |
2957 | @var{aiocbp} returns @code{EINPROGRESS}. Once the synchronization is |
2958 | done @code{aio_error} return @math{0} if the synchronization was not | |
a3a4a74e UD |
2959 | successful. Otherwise the value returned is the value to which the |
2960 | @code{fsync} or @code{fdatasync} function would have set the | |
2961 | @code{errno} variable. In this case nothing can be assumed about the | |
9739d2d5 | 2962 | consistency of the data written to this file descriptor. |
a3a4a74e UD |
2963 | |
2964 | The return value of this function is @math{0} if the request was | |
19e4c7dd | 2965 | successfully enqueued. Otherwise the return value is @math{-1} and |
a3a4a74e UD |
2966 | @code{errno} is set to one of the following values: |
2967 | ||
2968 | @table @code | |
2969 | @item EAGAIN | |
fed8f7f7 | 2970 | The request could not be enqueued due to temporary lack of resources. |
a3a4a74e | 2971 | @item EBADF |
47792506 | 2972 | The file descriptor @code{@var{aiocbp}->aio_fildes} is not valid. |
a3a4a74e UD |
2973 | @item EINVAL |
2974 | The implementation does not support I/O synchronization or the @var{op} | |
2975 | parameter is other than @code{O_DSYNC} and @code{O_SYNC}. | |
2976 | @item ENOSYS | |
2977 | This function is not implemented. | |
2978 | @end table | |
2979 | ||
2980 | When the sources are compiled with @code{_FILE_OFFSET_BITS == 64} this | |
37de3d55 | 2981 | function is in fact @code{aio_fsync64} since the LFS interface |
a3a4a74e UD |
2982 | transparently replaces the normal implementation. |
2983 | @end deftypefun | |
2984 | ||
a3a4a74e | 2985 | @deftypefun int aio_fsync64 (int @var{op}, struct aiocb64 *@var{aiocbp}) |
d08a7e4c | 2986 | @standards{Unix98, aio.h} |
2cc3615c | 2987 | @safety{@prelim{}@mtsafe{}@asunsafe{@asulock{} @ascuheap{}}@acunsafe{@aculock{} @acsmem{}}} |
a3a4a74e UD |
2988 | This function is similar to @code{aio_fsync} with the only difference |
2989 | that the argument is a reference to a variable of type @code{struct | |
2990 | aiocb64}. | |
2991 | ||
2992 | When the sources are compiled with @code{_FILE_OFFSET_BITS == 64} this | |
2993 | function is available under the name @code{aio_fsync} and so | |
04b9968b | 2994 | transparently replaces the interface for small files on 32 bit |
a3a4a74e UD |
2995 | machines. |
2996 | @end deftypefun | |
2997 | ||
fed8f7f7 | 2998 | Another method of synchronization is to wait until one or more requests of a |
a3a4a74e UD |
2999 | specific set terminated. This could be achieved by the @code{aio_*} |
3000 | functions to notify the initiating process about the termination but in | |
3001 | some situations this is not the ideal solution. In a program which | |
3002 | constantly updates clients somehow connected to the server it is not | |
3003 | always the best solution to go round robin since some connections might | |
9739d2d5 | 3004 | be slow. On the other hand letting the @code{aio_*} functions notify the |
a3a4a74e | 3005 | caller might also be not the best solution since whenever the process |
9739d2d5 | 3006 | works on preparing data for a client it makes no sense to be |
a3a4a74e UD |
3007 | interrupted by a notification since the new client will not be handled |
3008 | before the current client is served. For situations like this | |
3009 | @code{aio_suspend} should be used. | |
3010 | ||
a3a4a74e | 3011 | @deftypefun int aio_suspend (const struct aiocb *const @var{list}[], int @var{nent}, const struct timespec *@var{timeout}) |
d08a7e4c | 3012 | @standards{POSIX.1b, aio.h} |
2cc3615c AO |
3013 | @safety{@prelim{}@mtsafe{}@asunsafe{@asulock{}}@acunsafe{@aculock{}}} |
3014 | @c Take aio_requests_mutex, set up waitlist and requestlist, wait | |
3015 | @c for completion or timeout, and release the mutex. | |
19e4c7dd | 3016 | When calling this function, the calling thread is suspended until at |
a3a4a74e | 3017 | least one of the requests pointed to by the @var{nent} elements of the |
19e4c7dd AJ |
3018 | array @var{list} has completed. If any of the requests has already |
3019 | completed at the time @code{aio_suspend} is called, the function returns | |
3020 | immediately. Whether a request has terminated or not is determined by | |
a3a4a74e | 3021 | comparing the error status of the request with @code{EINPROGRESS}. If |
19e4c7dd | 3022 | an element of @var{list} is @code{NULL}, the entry is simply ignored. |
a3a4a74e | 3023 | |
19e4c7dd AJ |
3024 | If no request has finished, the calling process is suspended. If |
3025 | @var{timeout} is @code{NULL}, the process is not woken until a request | |
3026 | has finished. If @var{timeout} is not @code{NULL}, the process remains | |
3027 | suspended at least as long as specified in @var{timeout}. In this case, | |
a3a4a74e UD |
3028 | @code{aio_suspend} returns with an error. |
3029 | ||
fed8f7f7 | 3030 | The return value of the function is @math{0} if one or more requests |
a3a4a74e UD |
3031 | from the @var{list} have terminated. Otherwise the function returns |
3032 | @math{-1} and @code{errno} is set to one of the following values: | |
3033 | ||
3034 | @table @code | |
3035 | @item EAGAIN | |
3036 | None of the requests from the @var{list} completed in the time specified | |
3037 | by @var{timeout}. | |
3038 | @item EINTR | |
3039 | A signal interrupted the @code{aio_suspend} function. This signal might | |
3040 | also be sent by the AIO implementation while signalling the termination | |
3041 | of one of the requests. | |
3042 | @item ENOSYS | |
3043 | The @code{aio_suspend} function is not implemented. | |
3044 | @end table | |
3045 | ||
3046 | When the sources are compiled with @code{_FILE_OFFSET_BITS == 64} this | |
3047 | function is in fact @code{aio_suspend64} since the LFS interface | |
3048 | transparently replaces the normal implementation. | |
3049 | @end deftypefun | |
3050 | ||
a3a4a74e | 3051 | @deftypefun int aio_suspend64 (const struct aiocb64 *const @var{list}[], int @var{nent}, const struct timespec *@var{timeout}) |
d08a7e4c | 3052 | @standards{Unix98, aio.h} |
2cc3615c | 3053 | @safety{@prelim{}@mtsafe{}@asunsafe{@asulock{}}@acunsafe{@aculock{}}} |
a3a4a74e UD |
3054 | This function is similar to @code{aio_suspend} with the only difference |
3055 | that the argument is a reference to a variable of type @code{struct | |
3056 | aiocb64}. | |
3057 | ||
3058 | When the sources are compiled with @code{_FILE_OFFSET_BITS == 64} this | |
3059 | function is available under the name @code{aio_suspend} and so | |
04b9968b | 3060 | transparently replaces the interface for small files on 32 bit |
a3a4a74e UD |
3061 | machines. |
3062 | @end deftypefun | |
b07d03e0 UD |
3063 | |
3064 | @node Cancel AIO Operations | |
04b9968b | 3065 | @subsection Cancellation of AIO Operations |
b07d03e0 | 3066 | |
19e4c7dd | 3067 | When one or more requests are asynchronously processed, it might be |
a3a4a74e | 3068 | useful in some situations to cancel a selected operation, e.g., if it |
19e4c7dd AJ |
3069 | becomes obvious that the written data is no longer accurate and would |
3070 | have to be overwritten soon. As an example, assume an application, which | |
a3a4a74e UD |
3071 | writes data in files in a situation where new incoming data would have |
3072 | to be written in a file which will be updated by an enqueued request. | |
19e4c7dd AJ |
3073 | The POSIX AIO implementation provides such a function, but this function |
3074 | is not capable of forcing the cancellation of the request. It is up to the | |
a3a4a74e UD |
3075 | implementation to decide whether it is possible to cancel the operation |
3076 | or not. Therefore using this function is merely a hint. | |
3077 | ||
a3a4a74e | 3078 | @deftypefun int aio_cancel (int @var{fildes}, struct aiocb *@var{aiocbp}) |
d08a7e4c | 3079 | @standards{POSIX.1b, aio.h} |
2cc3615c AO |
3080 | @safety{@prelim{}@mtsafe{}@asunsafe{@asulock{} @ascuheap{}}@acunsafe{@aculock{} @acsmem{}}} |
3081 | @c After fcntl to check the fd is open, hold aio_requests_mutex, call | |
3082 | @c aio_find_req_fd, aio_remove_request, then aio_notify and | |
3083 | @c aio_free_request each request before releasing the lock. | |
3084 | @c aio_notify calls aio_notify_only and free, besides cond signal or | |
3085 | @c similar. aio_notify_only calls pthread_attr_init, | |
3086 | @c pthread_attr_setdetachstate, malloc, pthread_create, | |
3087 | @c notify_func_wrapper, aio_sigqueue, getpid, raise. | |
3088 | @c notify_func_wraper calls aio_start_notify_thread, free and then the | |
3089 | @c notifier function. | |
a3a4a74e | 3090 | The @code{aio_cancel} function can be used to cancel one or more |
19e4c7dd AJ |
3091 | outstanding requests. If the @var{aiocbp} parameter is @code{NULL}, the |
3092 | function tries to cancel all of the outstanding requests which would process | |
3093 | the file descriptor @var{fildes} (i.e., whose @code{aio_fildes} member | |
3094 | is @var{fildes}). If @var{aiocbp} is not @code{NULL}, @code{aio_cancel} | |
3095 | attempts to cancel the specific request pointed to by @var{aiocbp}. | |
a3a4a74e | 3096 | |
19e4c7dd | 3097 | For requests which were successfully canceled, the normal notification |
a3a4a74e UD |
3098 | about the termination of the request should take place. I.e., depending |
3099 | on the @code{struct sigevent} object which controls this, nothing | |
3100 | happens, a signal is sent or a thread is started. If the request cannot | |
19e4c7dd | 3101 | be canceled, it terminates the usual way after performing the operation. |
a3a4a74e | 3102 | |
19e4c7dd | 3103 | After a request is successfully canceled, a call to @code{aio_error} with |
a3a4a74e UD |
3104 | a reference to this request as the parameter will return |
3105 | @code{ECANCELED} and a call to @code{aio_return} will return @math{-1}. | |
19e4c7dd | 3106 | If the request wasn't canceled and is still running the error status is |
a3a4a74e UD |
3107 | still @code{EINPROGRESS}. |
3108 | ||
3109 | The return value of the function is @code{AIO_CANCELED} if there were | |
19e4c7dd AJ |
3110 | requests which haven't terminated and which were successfully canceled. |
3111 | If there is one or more requests left which couldn't be canceled, the | |
a3a4a74e | 3112 | return value is @code{AIO_NOTCANCELED}. In this case @code{aio_error} |
9739d2d5 | 3113 | must be used to find out which of the, perhaps multiple, requests (if |
19e4c7dd | 3114 | @var{aiocbp} is @code{NULL}) weren't successfully canceled. If all |
a3a4a74e UD |
3115 | requests already terminated at the time @code{aio_cancel} is called the |
3116 | return value is @code{AIO_ALLDONE}. | |
3117 | ||
3118 | If an error occurred during the execution of @code{aio_cancel} the | |
3119 | function returns @math{-1} and sets @code{errno} to one of the following | |
3120 | values. | |
3121 | ||
3122 | @table @code | |
3123 | @item EBADF | |
3124 | The file descriptor @var{fildes} is not valid. | |
3125 | @item ENOSYS | |
3126 | @code{aio_cancel} is not implemented. | |
3127 | @end table | |
3128 | ||
19e4c7dd | 3129 | When the sources are compiled with @code{_FILE_OFFSET_BITS == 64}, this |
a3a4a74e UD |
3130 | function is in fact @code{aio_cancel64} since the LFS interface |
3131 | transparently replaces the normal implementation. | |
3132 | @end deftypefun | |
3133 | ||
19e4c7dd | 3134 | @deftypefun int aio_cancel64 (int @var{fildes}, struct aiocb64 *@var{aiocbp}) |
d08a7e4c | 3135 | @standards{Unix98, aio.h} |
2cc3615c | 3136 | @safety{@prelim{}@mtsafe{}@asunsafe{@asulock{} @ascuheap{}}@acunsafe{@aculock{} @acsmem{}}} |
a3a4a74e UD |
3137 | This function is similar to @code{aio_cancel} with the only difference |
3138 | that the argument is a reference to a variable of type @code{struct | |
3139 | aiocb64}. | |
3140 | ||
19e4c7dd | 3141 | When the sources are compiled with @code{_FILE_OFFSET_BITS == 64}, this |
a3a4a74e | 3142 | function is available under the name @code{aio_cancel} and so |
04b9968b | 3143 | transparently replaces the interface for small files on 32 bit |
a3a4a74e UD |
3144 | machines. |
3145 | @end deftypefun | |
3146 | ||
3147 | @node Configuration of AIO | |
3148 | @subsection How to optimize the AIO implementation | |
3149 | ||
3150 | The POSIX standard does not specify how the AIO functions are | |
19e4c7dd | 3151 | implemented. They could be system calls, but it is also possible to |
a3a4a74e UD |
3152 | emulate them at userlevel. |
3153 | ||
9739d2d5 | 3154 | At the time of writing, the available implementation is a user-level |
19e4c7dd AJ |
3155 | implementation which uses threads for handling the enqueued requests. |
3156 | While this implementation requires making some decisions about | |
9739d2d5 | 3157 | limitations, hard limitations are something best avoided |
1f77f049 | 3158 | in @theglibc{}. Therefore, @theglibc{} provides a means |
19e4c7dd | 3159 | for tuning the AIO implementation according to the individual use. |
a3a4a74e | 3160 | |
a3a4a74e | 3161 | @deftp {Data Type} {struct aioinit} |
d08a7e4c | 3162 | @standards{GNU, aio.h} |
a3a4a74e UD |
3163 | This data type is used to pass the configuration or tunable parameters |
3164 | to the implementation. The program has to initialize the members of | |
3165 | this struct and pass it to the implementation using the @code{aio_init} | |
3166 | function. | |
3167 | ||
3168 | @table @code | |
3169 | @item int aio_threads | |
19e4c7dd | 3170 | This member specifies the maximal number of threads which may be used |
a3a4a74e UD |
3171 | at any one time. |
3172 | @item int aio_num | |
c756c71c | 3173 | This number provides an estimate on the maximal number of simultaneously |
a3a4a74e UD |
3174 | enqueued requests. |
3175 | @item int aio_locks | |
19e4c7dd | 3176 | Unused. |
a3a4a74e | 3177 | @item int aio_usedba |
19e4c7dd | 3178 | Unused. |
a3a4a74e | 3179 | @item int aio_debug |
19e4c7dd | 3180 | Unused. |
a3a4a74e | 3181 | @item int aio_numusers |
19e4c7dd | 3182 | Unused. |
a3a4a74e | 3183 | @item int aio_reserved[2] |
19e4c7dd | 3184 | Unused. |
a3a4a74e UD |
3185 | @end table |
3186 | @end deftp | |
3187 | ||
a3a4a74e | 3188 | @deftypefun void aio_init (const struct aioinit *@var{init}) |
d08a7e4c | 3189 | @standards{GNU, aio.h} |
2cc3615c AO |
3190 | @safety{@prelim{}@mtsafe{}@asunsafe{@asulock{}}@acunsafe{@aculock{}}} |
3191 | @c All changes to global objects are guarded by aio_requests_mutex. | |
a3a4a74e | 3192 | This function must be called before any other AIO function. Calling it |
19e4c7dd AJ |
3193 | is completely voluntary, as it is only meant to help the AIO |
3194 | implementation perform better. | |
a3a4a74e | 3195 | |
9739d2d5 | 3196 | Before calling @code{aio_init}, the members of a variable of |
a3a4a74e UD |
3197 | type @code{struct aioinit} must be initialized. Then a reference to |
3198 | this variable is passed as the parameter to @code{aio_init} which itself | |
3199 | may or may not pay attention to the hints. | |
3200 | ||
c756c71c | 3201 | The function has no return value and no error cases are defined. It is |
9739d2d5 | 3202 | an extension which follows a proposal from the SGI implementation in |
c756c71c | 3203 | @w{Irix 6}. It is not covered by POSIX.1b or Unix98. |
a3a4a74e | 3204 | @end deftypefun |
b07d03e0 | 3205 | |
28f540f4 RM |
3206 | @node Control Operations |
3207 | @section Control Operations on Files | |
3208 | ||
3209 | @cindex control operations on files | |
3210 | @cindex @code{fcntl} function | |
3211 | This section describes how you can perform various other operations on | |
3212 | file descriptors, such as inquiring about or setting flags describing | |
3213 | the status of the file descriptor, manipulating record locks, and the | |
3214 | like. All of these operations are performed by the function @code{fcntl}. | |
3215 | ||
3216 | The second argument to the @code{fcntl} function is a command that | |
3217 | specifies which operation to perform. The function and macros that name | |
3218 | various flags that are used with it are declared in the header file | |
3219 | @file{fcntl.h}. Many of these flags are also used by the @code{open} | |
3220 | function; see @ref{Opening and Closing Files}. | |
3221 | @pindex fcntl.h | |
3222 | ||
28f540f4 | 3223 | @deftypefun int fcntl (int @var{filedes}, int @var{command}, @dots{}) |
d08a7e4c | 3224 | @standards{POSIX.1, fcntl.h} |
2cc3615c | 3225 | @safety{@prelim{}@mtsafe{}@assafe{}@acsafe{}} |
28f540f4 RM |
3226 | The @code{fcntl} function performs the operation specified by |
3227 | @var{command} on the file descriptor @var{filedes}. Some commands | |
3228 | require additional arguments to be supplied. These additional arguments | |
3229 | and the return value and error conditions are given in the detailed | |
3230 | descriptions of the individual commands. | |
3231 | ||
3232 | Briefly, here is a list of what the various commands are. | |
3233 | ||
2fe82ca6 | 3234 | @vtable @code |
28f540f4 RM |
3235 | @item F_DUPFD |
3236 | Duplicate the file descriptor (return another file descriptor pointing | |
3237 | to the same open file). @xref{Duplicating Descriptors}. | |
3238 | ||
3239 | @item F_GETFD | |
3240 | Get flags associated with the file descriptor. @xref{Descriptor Flags}. | |
3241 | ||
3242 | @item F_SETFD | |
3243 | Set flags associated with the file descriptor. @xref{Descriptor Flags}. | |
3244 | ||
3245 | @item F_GETFL | |
3246 | Get flags associated with the open file. @xref{File Status Flags}. | |
3247 | ||
3248 | @item F_SETFL | |
3249 | Set flags associated with the open file. @xref{File Status Flags}. | |
3250 | ||
3251 | @item F_GETLK | |
0961f7e1 | 3252 | Test a file lock. @xref{File Locks}. |
28f540f4 RM |
3253 | |
3254 | @item F_SETLK | |
3255 | Set or clear a file lock. @xref{File Locks}. | |
3256 | ||
3257 | @item F_SETLKW | |
3258 | Like @code{F_SETLK}, but wait for completion. @xref{File Locks}. | |
3259 | ||
0961f7e1 JL |
3260 | @item F_OFD_GETLK |
3261 | Test an open file description lock. @xref{Open File Description Locks}. | |
3262 | Specific to Linux. | |
3263 | ||
3264 | @item F_OFD_SETLK | |
3265 | Set or clear an open file description lock. @xref{Open File Description Locks}. | |
3266 | Specific to Linux. | |
3267 | ||
3268 | @item F_OFD_SETLKW | |
3269 | Like @code{F_OFD_SETLK}, but block until lock is acquired. | |
3270 | @xref{Open File Description Locks}. Specific to Linux. | |
3271 | ||
28f540f4 RM |
3272 | @item F_GETOWN |
3273 | Get process or process group ID to receive @code{SIGIO} signals. | |
3274 | @xref{Interrupt Input}. | |
3275 | ||
3276 | @item F_SETOWN | |
3277 | Set process or process group ID to receive @code{SIGIO} signals. | |
3278 | @xref{Interrupt Input}. | |
2fe82ca6 | 3279 | @end vtable |
dfd2257a | 3280 | |
04b9968b | 3281 | This function is a cancellation point in multi-threaded programs. This |
dfd2257a UD |
3282 | is a problem if the thread allocates some resources (like memory, file |
3283 | descriptors, semaphores or whatever) at the time @code{fcntl} is | |
19e4c7dd | 3284 | called. If the thread gets canceled these resources stay allocated |
dfd2257a | 3285 | until the program ends. To avoid this calls to @code{fcntl} should be |
04b9968b | 3286 | protected using cancellation handlers. |
dfd2257a | 3287 | @c ref pthread_cleanup_push / pthread_cleanup_pop |
28f540f4 RM |
3288 | @end deftypefun |
3289 | ||
3290 | ||
3291 | @node Duplicating Descriptors | |
3292 | @section Duplicating Descriptors | |
3293 | ||
3294 | @cindex duplicating file descriptors | |
3295 | @cindex redirecting input and output | |
3296 | ||
3297 | You can @dfn{duplicate} a file descriptor, or allocate another file | |
3298 | descriptor that refers to the same open file as the original. Duplicate | |
3299 | descriptors share one file position and one set of file status flags | |
3300 | (@pxref{File Status Flags}), but each has its own set of file descriptor | |
3301 | flags (@pxref{Descriptor Flags}). | |
3302 | ||
3303 | The major use of duplicating a file descriptor is to implement | |
3304 | @dfn{redirection} of input or output: that is, to change the | |
3305 | file or pipe that a particular file descriptor corresponds to. | |
3306 | ||
3307 | You can perform this operation using the @code{fcntl} function with the | |
3308 | @code{F_DUPFD} command, but there are also convenient functions | |
3309 | @code{dup} and @code{dup2} for duplicating descriptors. | |
3310 | ||
3311 | @pindex unistd.h | |
3312 | @pindex fcntl.h | |
3313 | The @code{fcntl} function and flags are declared in @file{fcntl.h}, | |
3314 | while prototypes for @code{dup} and @code{dup2} are in the header file | |
3315 | @file{unistd.h}. | |
3316 | ||
28f540f4 | 3317 | @deftypefun int dup (int @var{old}) |
d08a7e4c | 3318 | @standards{POSIX.1, unistd.h} |
2cc3615c | 3319 | @safety{@prelim{}@mtsafe{}@assafe{}@acsafe{}} |
28f540f4 RM |
3320 | This function copies descriptor @var{old} to the first available |
3321 | descriptor number (the first number not currently open). It is | |
3322 | equivalent to @code{fcntl (@var{old}, F_DUPFD, 0)}. | |
3323 | @end deftypefun | |
3324 | ||
28f540f4 | 3325 | @deftypefun int dup2 (int @var{old}, int @var{new}) |
d08a7e4c | 3326 | @standards{POSIX.1, unistd.h} |
2cc3615c | 3327 | @safety{@prelim{}@mtsafe{}@assafe{}@acsafe{}} |
28f540f4 RM |
3328 | This function copies the descriptor @var{old} to descriptor number |
3329 | @var{new}. | |
3330 | ||
3331 | If @var{old} is an invalid descriptor, then @code{dup2} does nothing; it | |
3332 | does not close @var{new}. Otherwise, the new duplicate of @var{old} | |
3333 | replaces any previous meaning of descriptor @var{new}, as if @var{new} | |
3334 | were closed first. | |
3335 | ||
3336 | If @var{old} and @var{new} are different numbers, and @var{old} is a | |
3337 | valid descriptor number, then @code{dup2} is equivalent to: | |
3338 | ||
3339 | @smallexample | |
3340 | close (@var{new}); | |
3341 | fcntl (@var{old}, F_DUPFD, @var{new}) | |
3342 | @end smallexample | |
3343 | ||
3344 | However, @code{dup2} does this atomically; there is no instant in the | |
3345 | middle of calling @code{dup2} at which @var{new} is closed and not yet a | |
3346 | duplicate of @var{old}. | |
3347 | @end deftypefun | |
3348 | ||
28f540f4 | 3349 | @deftypevr Macro int F_DUPFD |
d08a7e4c | 3350 | @standards{POSIX.1, fcntl.h} |
28f540f4 RM |
3351 | This macro is used as the @var{command} argument to @code{fcntl}, to |
3352 | copy the file descriptor given as the first argument. | |
3353 | ||
3354 | The form of the call in this case is: | |
3355 | ||
3356 | @smallexample | |
3357 | fcntl (@var{old}, F_DUPFD, @var{next-filedes}) | |
3358 | @end smallexample | |
3359 | ||
3360 | The @var{next-filedes} argument is of type @code{int} and specifies that | |
3361 | the file descriptor returned should be the next available one greater | |
3362 | than or equal to this value. | |
3363 | ||
3364 | The return value from @code{fcntl} with this command is normally the value | |
07435eb4 | 3365 | of the new file descriptor. A return value of @math{-1} indicates an |
28f540f4 RM |
3366 | error. The following @code{errno} error conditions are defined for |
3367 | this command: | |
3368 | ||
3369 | @table @code | |
3370 | @item EBADF | |
3371 | The @var{old} argument is invalid. | |
3372 | ||
3373 | @item EINVAL | |
3374 | The @var{next-filedes} argument is invalid. | |
3375 | ||
3376 | @item EMFILE | |
3377 | There are no more file descriptors available---your program is already | |
3378 | using the maximum. In BSD and GNU, the maximum is controlled by a | |
3379 | resource limit that can be changed; @pxref{Limits on Resources}, for | |
3380 | more information about the @code{RLIMIT_NOFILE} limit. | |
3381 | @end table | |
3382 | ||
3383 | @code{ENFILE} is not a possible error code for @code{dup2} because | |
3384 | @code{dup2} does not create a new opening of a file; duplicate | |
3385 | descriptors do not count toward the limit which @code{ENFILE} | |
3386 | indicates. @code{EMFILE} is possible because it refers to the limit on | |
3387 | distinct descriptor numbers in use in one process. | |
3388 | @end deftypevr | |
3389 | ||
3390 | Here is an example showing how to use @code{dup2} to do redirection. | |
3391 | Typically, redirection of the standard streams (like @code{stdin}) is | |
3392 | done by a shell or shell-like program before calling one of the | |
3393 | @code{exec} functions (@pxref{Executing a File}) to execute a new | |
3394 | program in a child process. When the new program is executed, it | |
3395 | creates and initializes the standard streams to point to the | |
3396 | corresponding file descriptors, before its @code{main} function is | |
3397 | invoked. | |
3398 | ||
3399 | So, to redirect standard input to a file, the shell could do something | |
3400 | like: | |
3401 | ||
3402 | @smallexample | |
3403 | pid = fork (); | |
3404 | if (pid == 0) | |
3405 | @{ | |
3406 | char *filename; | |
3407 | char *program; | |
3408 | int file; | |
3409 | @dots{} | |
3410 | file = TEMP_FAILURE_RETRY (open (filename, O_RDONLY)); | |
3411 | dup2 (file, STDIN_FILENO); | |
3412 | TEMP_FAILURE_RETRY (close (file)); | |
3413 | execv (program, NULL); | |
3414 | @} | |
3415 | @end smallexample | |
3416 | ||
3417 | There is also a more detailed example showing how to implement redirection | |
3418 | in the context of a pipeline of processes in @ref{Launching Jobs}. | |
3419 | ||
3420 | ||
3421 | @node Descriptor Flags | |
3422 | @section File Descriptor Flags | |
3423 | @cindex file descriptor flags | |
3424 | ||
3425 | @dfn{File descriptor flags} are miscellaneous attributes of a file | |
3426 | descriptor. These flags are associated with particular file | |
3427 | descriptors, so that if you have created duplicate file descriptors | |
3428 | from a single opening of a file, each descriptor has its own set of flags. | |
3429 | ||
3430 | Currently there is just one file descriptor flag: @code{FD_CLOEXEC}, | |
3431 | which causes the descriptor to be closed if you use any of the | |
3432 | @code{exec@dots{}} functions (@pxref{Executing a File}). | |
3433 | ||
3434 | The symbols in this section are defined in the header file | |
3435 | @file{fcntl.h}. | |
3436 | @pindex fcntl.h | |
3437 | ||
28f540f4 | 3438 | @deftypevr Macro int F_GETFD |
d08a7e4c | 3439 | @standards{POSIX.1, fcntl.h} |
28f540f4 RM |
3440 | This macro is used as the @var{command} argument to @code{fcntl}, to |
3441 | specify that it should return the file descriptor flags associated | |
2c6fe0bd | 3442 | with the @var{filedes} argument. |
28f540f4 RM |
3443 | |
3444 | The normal return value from @code{fcntl} with this command is a | |
3445 | nonnegative number which can be interpreted as the bitwise OR of the | |
3446 | individual flags (except that currently there is only one flag to use). | |
3447 | ||
07435eb4 | 3448 | In case of an error, @code{fcntl} returns @math{-1}. The following |
28f540f4 RM |
3449 | @code{errno} error conditions are defined for this command: |
3450 | ||
3451 | @table @code | |
3452 | @item EBADF | |
3453 | The @var{filedes} argument is invalid. | |
3454 | @end table | |
3455 | @end deftypevr | |
3456 | ||
3457 | ||
28f540f4 | 3458 | @deftypevr Macro int F_SETFD |
d08a7e4c | 3459 | @standards{POSIX.1, fcntl.h} |
28f540f4 RM |
3460 | This macro is used as the @var{command} argument to @code{fcntl}, to |
3461 | specify that it should set the file descriptor flags associated with the | |
3462 | @var{filedes} argument. This requires a third @code{int} argument to | |
3463 | specify the new flags, so the form of the call is: | |
3464 | ||
3465 | @smallexample | |
3466 | fcntl (@var{filedes}, F_SETFD, @var{new-flags}) | |
3467 | @end smallexample | |
3468 | ||
3469 | The normal return value from @code{fcntl} with this command is an | |
07435eb4 | 3470 | unspecified value other than @math{-1}, which indicates an error. |
28f540f4 RM |
3471 | The flags and error conditions are the same as for the @code{F_GETFD} |
3472 | command. | |
3473 | @end deftypevr | |
3474 | ||
3475 | The following macro is defined for use as a file descriptor flag with | |
3476 | the @code{fcntl} function. The value is an integer constant usable | |
3477 | as a bit mask value. | |
3478 | ||
28f540f4 | 3479 | @deftypevr Macro int FD_CLOEXEC |
d08a7e4c | 3480 | @standards{POSIX.1, fcntl.h} |
28f540f4 RM |
3481 | @cindex close-on-exec (file descriptor flag) |
3482 | This flag specifies that the file descriptor should be closed when | |
3483 | an @code{exec} function is invoked; see @ref{Executing a File}. When | |
3484 | a file descriptor is allocated (as with @code{open} or @code{dup}), | |
3485 | this bit is initially cleared on the new file descriptor, meaning that | |
3486 | descriptor will survive into the new program after @code{exec}. | |
3487 | @end deftypevr | |
3488 | ||
3489 | If you want to modify the file descriptor flags, you should get the | |
3490 | current flags with @code{F_GETFD} and modify the value. Don't assume | |
3491 | that the flags listed here are the only ones that are implemented; your | |
3492 | program may be run years from now and more flags may exist then. For | |
3493 | example, here is a function to set or clear the flag @code{FD_CLOEXEC} | |
3494 | without altering any other flags: | |
3495 | ||
3496 | @smallexample | |
3497 | /* @r{Set the @code{FD_CLOEXEC} flag of @var{desc} if @var{value} is nonzero,} | |
3498 | @r{or clear the flag if @var{value} is 0.} | |
2c6fe0bd | 3499 | @r{Return 0 on success, or -1 on error with @code{errno} set.} */ |
28f540f4 RM |
3500 | |
3501 | int | |
3502 | set_cloexec_flag (int desc, int value) | |
3503 | @{ | |
3504 | int oldflags = fcntl (desc, F_GETFD, 0); | |
8e96ae1a | 3505 | /* @r{If reading the flags failed, return error indication now.} */ |
28f540f4 RM |
3506 | if (oldflags < 0) |
3507 | return oldflags; | |
3508 | /* @r{Set just the flag we want to set.} */ | |
3509 | if (value != 0) | |
3510 | oldflags |= FD_CLOEXEC; | |
3511 | else | |
3512 | oldflags &= ~FD_CLOEXEC; | |
3513 | /* @r{Store modified flag word in the descriptor.} */ | |
3514 | return fcntl (desc, F_SETFD, oldflags); | |
3515 | @} | |
3516 | @end smallexample | |
3517 | ||
3518 | @node File Status Flags | |
3519 | @section File Status Flags | |
3520 | @cindex file status flags | |
3521 | ||
3522 | @dfn{File status flags} are used to specify attributes of the opening of a | |
3523 | file. Unlike the file descriptor flags discussed in @ref{Descriptor | |
3524 | Flags}, the file status flags are shared by duplicated file descriptors | |
3525 | resulting from a single opening of the file. The file status flags are | |
3526 | specified with the @var{flags} argument to @code{open}; | |
3527 | @pxref{Opening and Closing Files}. | |
3528 | ||
3529 | File status flags fall into three categories, which are described in the | |
3530 | following sections. | |
3531 | ||
3532 | @itemize @bullet | |
3533 | @item | |
3534 | @ref{Access Modes}, specify what type of access is allowed to the | |
3535 | file: reading, writing, or both. They are set by @code{open} and are | |
3536 | returned by @code{fcntl}, but cannot be changed. | |
3537 | ||
3538 | @item | |
3539 | @ref{Open-time Flags}, control details of what @code{open} will do. | |
3540 | These flags are not preserved after the @code{open} call. | |
3541 | ||
3542 | @item | |
3543 | @ref{Operating Modes}, affect how operations such as @code{read} and | |
3544 | @code{write} are done. They are set by @code{open}, and can be fetched or | |
3545 | changed with @code{fcntl}. | |
3546 | @end itemize | |
3547 | ||
3548 | The symbols in this section are defined in the header file | |
3549 | @file{fcntl.h}. | |
3550 | @pindex fcntl.h | |
3551 | ||
3552 | @menu | |
3553 | * Access Modes:: Whether the descriptor can read or write. | |
3554 | * Open-time Flags:: Details of @code{open}. | |
3555 | * Operating Modes:: Special modes to control I/O operations. | |
3556 | * Getting File Status Flags:: Fetching and changing these flags. | |
3557 | @end menu | |
3558 | ||
3559 | @node Access Modes | |
3560 | @subsection File Access Modes | |
3561 | ||
3562 | The file access modes allow a file descriptor to be used for reading, | |
a7a93d50 | 3563 | writing, or both. (On @gnuhurdsystems{}, they can also allow none of these, |
28f540f4 RM |
3564 | and allow execution of the file as a program.) The access modes are chosen |
3565 | when the file is opened, and never change. | |
3566 | ||
28f540f4 | 3567 | @deftypevr Macro int O_RDONLY |
d08a7e4c | 3568 | @standards{POSIX.1, fcntl.h} |
28f540f4 RM |
3569 | Open the file for read access. |
3570 | @end deftypevr | |
3571 | ||
28f540f4 | 3572 | @deftypevr Macro int O_WRONLY |
d08a7e4c | 3573 | @standards{POSIX.1, fcntl.h} |
28f540f4 RM |
3574 | Open the file for write access. |
3575 | @end deftypevr | |
3576 | ||
28f540f4 | 3577 | @deftypevr Macro int O_RDWR |
d08a7e4c | 3578 | @standards{POSIX.1, fcntl.h} |
28f540f4 RM |
3579 | Open the file for both reading and writing. |
3580 | @end deftypevr | |
3581 | ||
a7a93d50 | 3582 | On @gnuhurdsystems{} (and not on other systems), @code{O_RDONLY} and |
28f540f4 RM |
3583 | @code{O_WRONLY} are independent bits that can be bitwise-ORed together, |
3584 | and it is valid for either bit to be set or clear. This means that | |
3585 | @code{O_RDWR} is the same as @code{O_RDONLY|O_WRONLY}. A file access | |
3586 | mode of zero is permissible; it allows no operations that do input or | |
3587 | output to the file, but does allow other operations such as | |
a7a93d50 | 3588 | @code{fchmod}. On @gnuhurdsystems{}, since ``read-only'' or ``write-only'' |
28f540f4 RM |
3589 | is a misnomer, @file{fcntl.h} defines additional names for the file |
3590 | access modes. These names are preferred when writing GNU-specific code. | |
3591 | But most programs will want to be portable to other POSIX.1 systems and | |
3592 | should use the POSIX.1 names above instead. | |
3593 | ||
28f540f4 | 3594 | @deftypevr Macro int O_READ |
d08a7e4c | 3595 | @standards{GNU, fcntl.h (optional)} |
c2835376 | 3596 | Open the file for reading. Same as @code{O_RDONLY}; only defined on GNU. |
28f540f4 RM |
3597 | @end deftypevr |
3598 | ||
28f540f4 | 3599 | @deftypevr Macro int O_WRITE |
d08a7e4c | 3600 | @standards{GNU, fcntl.h (optional)} |
c2835376 | 3601 | Open the file for writing. Same as @code{O_WRONLY}; only defined on GNU. |
28f540f4 RM |
3602 | @end deftypevr |
3603 | ||
28f540f4 | 3604 | @deftypevr Macro int O_EXEC |
d08a7e4c | 3605 | @standards{GNU, fcntl.h (optional)} |
28f540f4 RM |
3606 | Open the file for executing. Only defined on GNU. |
3607 | @end deftypevr | |
3608 | ||
3609 | To determine the file access mode with @code{fcntl}, you must extract | |
a7a93d50 JM |
3610 | the access mode bits from the retrieved file status flags. On |
3611 | @gnuhurdsystems{}, | |
3612 | you can just test the @code{O_READ} and @code{O_WRITE} bits in | |
28f540f4 RM |
3613 | the flags word. But in other POSIX.1 systems, reading and writing |
3614 | access modes are not stored as distinct bit flags. The portable way to | |
3615 | extract the file access mode bits is with @code{O_ACCMODE}. | |
3616 | ||
28f540f4 | 3617 | @deftypevr Macro int O_ACCMODE |
d08a7e4c | 3618 | @standards{POSIX.1, fcntl.h} |
28f540f4 RM |
3619 | This macro stands for a mask that can be bitwise-ANDed with the file |
3620 | status flag value to produce a value representing the file access mode. | |
3621 | The mode will be @code{O_RDONLY}, @code{O_WRONLY}, or @code{O_RDWR}. | |
a7a93d50 | 3622 | (On @gnuhurdsystems{} it could also be zero, and it never includes the |
28f540f4 RM |
3623 | @code{O_EXEC} bit.) |
3624 | @end deftypevr | |
3625 | ||
3626 | @node Open-time Flags | |
3627 | @subsection Open-time Flags | |
3628 | ||
3629 | The open-time flags specify options affecting how @code{open} will behave. | |
3630 | These options are not preserved once the file is open. The exception to | |
3631 | this is @code{O_NONBLOCK}, which is also an I/O operating mode and so it | |
3632 | @emph{is} saved. @xref{Opening and Closing Files}, for how to call | |
3633 | @code{open}. | |
3634 | ||
3635 | There are two sorts of options specified by open-time flags. | |
3636 | ||
3637 | @itemize @bullet | |
3638 | @item | |
3639 | @dfn{File name translation flags} affect how @code{open} looks up the | |
3640 | file name to locate the file, and whether the file can be created. | |
3641 | @cindex file name translation flags | |
3642 | @cindex flags, file name translation | |
3643 | ||
3644 | @item | |
3645 | @dfn{Open-time action flags} specify extra operations that @code{open} will | |
3646 | perform on the file once it is open. | |
3647 | @cindex open-time action flags | |
3648 | @cindex flags, open-time action | |
3649 | @end itemize | |
3650 | ||
3651 | Here are the file name translation flags. | |
3652 | ||
28f540f4 | 3653 | @deftypevr Macro int O_CREAT |
d08a7e4c | 3654 | @standards{POSIX.1, fcntl.h} |
28f540f4 RM |
3655 | If set, the file will be created if it doesn't already exist. |
3656 | @c !!! mode arg, umask | |
3657 | @cindex create on open (file status flag) | |
3658 | @end deftypevr | |
3659 | ||
28f540f4 | 3660 | @deftypevr Macro int O_EXCL |
d08a7e4c | 3661 | @standards{POSIX.1, fcntl.h} |
28f540f4 RM |
3662 | If both @code{O_CREAT} and @code{O_EXCL} are set, then @code{open} fails |
3663 | if the specified file already exists. This is guaranteed to never | |
3664 | clobber an existing file. | |
b9af29c0 FW |
3665 | |
3666 | The @code{O_EXCL} flag has a special meaning in combination with | |
3667 | @code{O_TMPFILE}; see below. | |
3668 | @end deftypevr | |
3669 | ||
3670 | @deftypevr Macro int O_TMPFILE | |
3671 | @standards{GNU, fcntl.h} | |
3672 | If this flag is specified, functions in the @code{open} family create an | |
3673 | unnamed temporary file. In this case, the pathname argument to the | |
3674 | @code{open} family of functions (@pxref{Opening and Closing Files}) is | |
3675 | interpreted as the directory in which the temporary file is created | |
3676 | (thus determining the file system which provides the storage for the | |
3677 | file). The @code{O_TMPFILE} flag must be combined with @code{O_WRONLY} | |
3678 | or @code{O_RDWR}, and the @var{mode} argument is required. | |
3679 | ||
3680 | The temporary file can later be given a name using @code{linkat}, | |
3681 | turning it into a regular file. This allows the atomic creation of a | |
3682 | file with the specific file attributes (mode and extended attributes) | |
3683 | and file contents. If, for security reasons, it is not desirable that a | |
3684 | name can be given to the file, the @code{O_EXCL} flag can be specified | |
3685 | along with @code{O_TMPFILE}. | |
3686 | ||
3687 | Not all kernels support this open flag. If this flag is unsupported, an | |
3688 | attempt to create an unnamed temporary file fails with an error of | |
3689 | @code{EINVAL}. If the underlying file system does not support the | |
3690 | @code{O_TMPFILE} flag, an @code{EOPNOTSUPP} error is the result. | |
3691 | ||
3692 | The @code{O_TMPFILE} flag is a GNU extension. | |
28f540f4 RM |
3693 | @end deftypevr |
3694 | ||
28f540f4 | 3695 | @deftypevr Macro int O_NONBLOCK |
d08a7e4c | 3696 | @standards{POSIX.1, fcntl.h} |
28f540f4 RM |
3697 | @cindex non-blocking open |
3698 | This prevents @code{open} from blocking for a ``long time'' to open the | |
3699 | file. This is only meaningful for some kinds of files, usually devices | |
3700 | such as serial ports; when it is not meaningful, it is harmless and | |
9739d2d5 | 3701 | ignored. Often, opening a port to a modem blocks until the modem reports |
28f540f4 RM |
3702 | carrier detection; if @code{O_NONBLOCK} is specified, @code{open} will |
3703 | return immediately without a carrier. | |
3704 | ||
3705 | Note that the @code{O_NONBLOCK} flag is overloaded as both an I/O operating | |
3706 | mode and a file name translation flag. This means that specifying | |
3707 | @code{O_NONBLOCK} in @code{open} also sets nonblocking I/O mode; | |
3708 | @pxref{Operating Modes}. To open the file without blocking but do normal | |
3709 | I/O that blocks, you must call @code{open} with @code{O_NONBLOCK} set and | |
3710 | then call @code{fcntl} to turn the bit off. | |
3711 | @end deftypevr | |
3712 | ||
28f540f4 | 3713 | @deftypevr Macro int O_NOCTTY |
d08a7e4c | 3714 | @standards{POSIX.1, fcntl.h} |
28f540f4 RM |
3715 | If the named file is a terminal device, don't make it the controlling |
3716 | terminal for the process. @xref{Job Control}, for information about | |
3717 | what it means to be the controlling terminal. | |
3718 | ||
a7a93d50 JM |
3719 | On @gnuhurdsystems{} and 4.4 BSD, opening a file never makes it the |
3720 | controlling terminal and @code{O_NOCTTY} is zero. However, @gnulinuxsystems{} | |
3721 | and some other systems use a nonzero value for @code{O_NOCTTY} and set the | |
28f540f4 RM |
3722 | controlling terminal when you open a file that is a terminal device; so |
3723 | to be portable, use @code{O_NOCTTY} when it is important to avoid this. | |
3724 | @cindex controlling terminal, setting | |
3725 | @end deftypevr | |
3726 | ||
a7a93d50 JM |
3727 | The following three file name translation flags exist only on |
3728 | @gnuhurdsystems{}. | |
28f540f4 | 3729 | |
28f540f4 | 3730 | @deftypevr Macro int O_IGNORE_CTTY |
d08a7e4c | 3731 | @standards{GNU, fcntl.h (optional)} |
28f540f4 RM |
3732 | Do not recognize the named file as the controlling terminal, even if it |
3733 | refers to the process's existing controlling terminal device. Operations | |
3734 | on the new file descriptor will never induce job control signals. | |
3735 | @xref{Job Control}. | |
3736 | @end deftypevr | |
3737 | ||
28f540f4 | 3738 | @deftypevr Macro int O_NOLINK |
d08a7e4c | 3739 | @standards{GNU, fcntl.h (optional)} |
28f540f4 RM |
3740 | If the named file is a symbolic link, open the link itself instead of |
3741 | the file it refers to. (@code{fstat} on the new file descriptor will | |
3742 | return the information returned by @code{lstat} on the link's name.) | |
3743 | @cindex symbolic link, opening | |
3744 | @end deftypevr | |
3745 | ||
28f540f4 | 3746 | @deftypevr Macro int O_NOTRANS |
d08a7e4c | 3747 | @standards{GNU, fcntl.h (optional)} |
28f540f4 RM |
3748 | If the named file is specially translated, do not invoke the translator. |
3749 | Open the bare file the translator itself sees. | |
3750 | @end deftypevr | |
3751 | ||
3752 | ||
3753 | The open-time action flags tell @code{open} to do additional operations | |
3754 | which are not really related to opening the file. The reason to do them | |
3755 | as part of @code{open} instead of in separate calls is that @code{open} | |
3756 | can do them @i{atomically}. | |
3757 | ||
28f540f4 | 3758 | @deftypevr Macro int O_TRUNC |
d08a7e4c | 3759 | @standards{POSIX.1, fcntl.h} |
28f540f4 RM |
3760 | Truncate the file to zero length. This option is only useful for |
3761 | regular files, not special files such as directories or FIFOs. POSIX.1 | |
3762 | requires that you open the file for writing to use @code{O_TRUNC}. In | |
3763 | BSD and GNU you must have permission to write the file to truncate it, | |
3764 | but you need not open for write access. | |
3765 | ||
3766 | This is the only open-time action flag specified by POSIX.1. There is | |
3767 | no good reason for truncation to be done by @code{open}, instead of by | |
3768 | calling @code{ftruncate} afterwards. The @code{O_TRUNC} flag existed in | |
3769 | Unix before @code{ftruncate} was invented, and is retained for backward | |
3770 | compatibility. | |
3771 | @end deftypevr | |
3772 | ||
27e309c1 UD |
3773 | The remaining operating modes are BSD extensions. They exist only |
3774 | on some systems. On other systems, these macros are not defined. | |
3775 | ||
28f540f4 | 3776 | @deftypevr Macro int O_SHLOCK |
d08a7e4c | 3777 | @standards{BSD, fcntl.h (optional)} |
28f540f4 RM |
3778 | Acquire a shared lock on the file, as with @code{flock}. |
3779 | @xref{File Locks}. | |
3780 | ||
3781 | If @code{O_CREAT} is specified, the locking is done atomically when | |
3782 | creating the file. You are guaranteed that no other process will get | |
3783 | the lock on the new file first. | |
3784 | @end deftypevr | |
3785 | ||
28f540f4 | 3786 | @deftypevr Macro int O_EXLOCK |
d08a7e4c | 3787 | @standards{BSD, fcntl.h (optional)} |
28f540f4 RM |
3788 | Acquire an exclusive lock on the file, as with @code{flock}. |
3789 | @xref{File Locks}. This is atomic like @code{O_SHLOCK}. | |
3790 | @end deftypevr | |
3791 | ||
3792 | @node Operating Modes | |
3793 | @subsection I/O Operating Modes | |
3794 | ||
3795 | The operating modes affect how input and output operations using a file | |
3796 | descriptor work. These flags are set by @code{open} and can be fetched | |
3797 | and changed with @code{fcntl}. | |
3798 | ||
28f540f4 | 3799 | @deftypevr Macro int O_APPEND |
d08a7e4c | 3800 | @standards{POSIX.1, fcntl.h} |
28f540f4 RM |
3801 | The bit that enables append mode for the file. If set, then all |
3802 | @code{write} operations write the data at the end of the file, extending | |
3803 | it, regardless of the current file position. This is the only reliable | |
3804 | way to append to a file. In append mode, you are guaranteed that the | |
3805 | data you write will always go to the current end of the file, regardless | |
3806 | of other processes writing to the file. Conversely, if you simply set | |
3807 | the file position to the end of file and write, then another process can | |
3808 | extend the file after you set the file position but before you write, | |
3809 | resulting in your data appearing someplace before the real end of file. | |
3810 | @end deftypevr | |
3811 | ||
2c6fe0bd | 3812 | @deftypevr Macro int O_NONBLOCK |
d08a7e4c | 3813 | @standards{POSIX.1, fcntl.h} |
28f540f4 RM |
3814 | The bit that enables nonblocking mode for the file. If this bit is set, |
3815 | @code{read} requests on the file can return immediately with a failure | |
3816 | status if there is no input immediately available, instead of blocking. | |
3817 | Likewise, @code{write} requests can also return immediately with a | |
3818 | failure status if the output can't be written immediately. | |
3819 | ||
3820 | Note that the @code{O_NONBLOCK} flag is overloaded as both an I/O | |
3821 | operating mode and a file name translation flag; @pxref{Open-time Flags}. | |
3822 | @end deftypevr | |
3823 | ||
28f540f4 | 3824 | @deftypevr Macro int O_NDELAY |
d08a7e4c | 3825 | @standards{BSD, fcntl.h} |
28f540f4 RM |
3826 | This is an obsolete name for @code{O_NONBLOCK}, provided for |
3827 | compatibility with BSD. It is not defined by the POSIX.1 standard. | |
3828 | @end deftypevr | |
3829 | ||
3830 | The remaining operating modes are BSD and GNU extensions. They exist only | |
3831 | on some systems. On other systems, these macros are not defined. | |
3832 | ||
28f540f4 | 3833 | @deftypevr Macro int O_ASYNC |
d08a7e4c | 3834 | @standards{BSD, fcntl.h} |
28f540f4 RM |
3835 | The bit that enables asynchronous input mode. If set, then @code{SIGIO} |
3836 | signals will be generated when input is available. @xref{Interrupt Input}. | |
3837 | ||
3838 | Asynchronous input mode is a BSD feature. | |
3839 | @end deftypevr | |
3840 | ||
28f540f4 | 3841 | @deftypevr Macro int O_FSYNC |
d08a7e4c | 3842 | @standards{BSD, fcntl.h} |
28f540f4 RM |
3843 | The bit that enables synchronous writing for the file. If set, each |
3844 | @code{write} call will make sure the data is reliably stored on disk before | |
3845 | returning. @c !!! xref fsync | |
3846 | ||
3847 | Synchronous writing is a BSD feature. | |
3848 | @end deftypevr | |
3849 | ||
28f540f4 | 3850 | @deftypevr Macro int O_SYNC |
d08a7e4c | 3851 | @standards{BSD, fcntl.h} |
28f540f4 RM |
3852 | This is another name for @code{O_FSYNC}. They have the same value. |
3853 | @end deftypevr | |
3854 | ||
28f540f4 | 3855 | @deftypevr Macro int O_NOATIME |
d08a7e4c | 3856 | @standards{GNU, fcntl.h} |
28f540f4 RM |
3857 | If this bit is set, @code{read} will not update the access time of the |
3858 | file. @xref{File Times}. This is used by programs that do backups, so | |
3859 | that backing a file up does not count as reading it. | |
3860 | Only the owner of the file or the superuser may use this bit. | |
3861 | ||
3862 | This is a GNU extension. | |
3863 | @end deftypevr | |
3864 | ||
3865 | @node Getting File Status Flags | |
3866 | @subsection Getting and Setting File Status Flags | |
3867 | ||
3868 | The @code{fcntl} function can fetch or change file status flags. | |
3869 | ||
28f540f4 | 3870 | @deftypevr Macro int F_GETFL |
d08a7e4c | 3871 | @standards{POSIX.1, fcntl.h} |
28f540f4 RM |
3872 | This macro is used as the @var{command} argument to @code{fcntl}, to |
3873 | read the file status flags for the open file with descriptor | |
3874 | @var{filedes}. | |
3875 | ||
3876 | The normal return value from @code{fcntl} with this command is a | |
3877 | nonnegative number which can be interpreted as the bitwise OR of the | |
3878 | individual flags. Since the file access modes are not single-bit values, | |
3879 | you can mask off other bits in the returned flags with @code{O_ACCMODE} | |
3880 | to compare them. | |
3881 | ||
07435eb4 | 3882 | In case of an error, @code{fcntl} returns @math{-1}. The following |
28f540f4 RM |
3883 | @code{errno} error conditions are defined for this command: |
3884 | ||
3885 | @table @code | |
3886 | @item EBADF | |
3887 | The @var{filedes} argument is invalid. | |
3888 | @end table | |
3889 | @end deftypevr | |
3890 | ||
28f540f4 | 3891 | @deftypevr Macro int F_SETFL |
d08a7e4c | 3892 | @standards{POSIX.1, fcntl.h} |
28f540f4 RM |
3893 | This macro is used as the @var{command} argument to @code{fcntl}, to set |
3894 | the file status flags for the open file corresponding to the | |
3895 | @var{filedes} argument. This command requires a third @code{int} | |
3896 | argument to specify the new flags, so the call looks like this: | |
3897 | ||
3898 | @smallexample | |
3899 | fcntl (@var{filedes}, F_SETFL, @var{new-flags}) | |
3900 | @end smallexample | |
3901 | ||
3902 | You can't change the access mode for the file in this way; that is, | |
3903 | whether the file descriptor was opened for reading or writing. | |
3904 | ||
3905 | The normal return value from @code{fcntl} with this command is an | |
07435eb4 | 3906 | unspecified value other than @math{-1}, which indicates an error. The |
28f540f4 RM |
3907 | error conditions are the same as for the @code{F_GETFL} command. |
3908 | @end deftypevr | |
3909 | ||
3910 | If you want to modify the file status flags, you should get the current | |
3911 | flags with @code{F_GETFL} and modify the value. Don't assume that the | |
3912 | flags listed here are the only ones that are implemented; your program | |
3913 | may be run years from now and more flags may exist then. For example, | |
3914 | here is a function to set or clear the flag @code{O_NONBLOCK} without | |
3915 | altering any other flags: | |
3916 | ||
3917 | @smallexample | |
3918 | @group | |
3919 | /* @r{Set the @code{O_NONBLOCK} flag of @var{desc} if @var{value} is nonzero,} | |
3920 | @r{or clear the flag if @var{value} is 0.} | |
2c6fe0bd | 3921 | @r{Return 0 on success, or -1 on error with @code{errno} set.} */ |
28f540f4 RM |
3922 | |
3923 | int | |
3924 | set_nonblock_flag (int desc, int value) | |
3925 | @{ | |
3926 | int oldflags = fcntl (desc, F_GETFL, 0); | |
3927 | /* @r{If reading the flags failed, return error indication now.} */ | |
3928 | if (oldflags == -1) | |
3929 | return -1; | |
3930 | /* @r{Set just the flag we want to set.} */ | |
3931 | if (value != 0) | |
3932 | oldflags |= O_NONBLOCK; | |
3933 | else | |
3934 | oldflags &= ~O_NONBLOCK; | |
3935 | /* @r{Store modified flag word in the descriptor.} */ | |
3936 | return fcntl (desc, F_SETFL, oldflags); | |
3937 | @} | |
3938 | @end group | |
3939 | @end smallexample | |
3940 | ||
3941 | @node File Locks | |
3942 | @section File Locks | |
3943 | ||
3944 | @cindex file locks | |
3945 | @cindex record locking | |
0961f7e1 JL |
3946 | This section describes record locks that are associated with the process. |
3947 | There is also a different type of record lock that is associated with the | |
3948 | open file description instead of the process. @xref{Open File Description Locks}. | |
3949 | ||
28f540f4 RM |
3950 | The remaining @code{fcntl} commands are used to support @dfn{record |
3951 | locking}, which permits multiple cooperating programs to prevent each | |
3952 | other from simultaneously accessing parts of a file in error-prone | |
3953 | ways. | |
3954 | ||
3955 | @cindex exclusive lock | |
3956 | @cindex write lock | |
3957 | An @dfn{exclusive} or @dfn{write} lock gives a process exclusive access | |
3958 | for writing to the specified part of the file. While a write lock is in | |
3959 | place, no other process can lock that part of the file. | |
3960 | ||
3961 | @cindex shared lock | |
3962 | @cindex read lock | |
3963 | A @dfn{shared} or @dfn{read} lock prohibits any other process from | |
3964 | requesting a write lock on the specified part of the file. However, | |
3965 | other processes can request read locks. | |
3966 | ||
3967 | The @code{read} and @code{write} functions do not actually check to see | |
3968 | whether there are any locks in place. If you want to implement a | |
3969 | locking protocol for a file shared by multiple processes, your application | |
3970 | must do explicit @code{fcntl} calls to request and clear locks at the | |
3971 | appropriate points. | |
3972 | ||
3973 | Locks are associated with processes. A process can only have one kind | |
3974 | of lock set for each byte of a given file. When any file descriptor for | |
3975 | that file is closed by the process, all of the locks that process holds | |
3976 | on that file are released, even if the locks were made using other | |
3977 | descriptors that remain open. Likewise, locks are released when a | |
3978 | process exits, and are not inherited by child processes created using | |
3979 | @code{fork} (@pxref{Creating a Process}). | |
3980 | ||
3981 | When making a lock, use a @code{struct flock} to specify what kind of | |
3982 | lock and where. This data type and the associated macros for the | |
3983 | @code{fcntl} function are declared in the header file @file{fcntl.h}. | |
3984 | @pindex fcntl.h | |
3985 | ||
28f540f4 | 3986 | @deftp {Data Type} {struct flock} |
d08a7e4c | 3987 | @standards{POSIX.1, fcntl.h} |
28f540f4 RM |
3988 | This structure is used with the @code{fcntl} function to describe a file |
3989 | lock. It has these members: | |
3990 | ||
3991 | @table @code | |
3992 | @item short int l_type | |
3993 | Specifies the type of the lock; one of @code{F_RDLCK}, @code{F_WRLCK}, or | |
3994 | @code{F_UNLCK}. | |
3995 | ||
3996 | @item short int l_whence | |
3997 | This corresponds to the @var{whence} argument to @code{fseek} or | |
3998 | @code{lseek}, and specifies what the offset is relative to. Its value | |
3999 | can be one of @code{SEEK_SET}, @code{SEEK_CUR}, or @code{SEEK_END}. | |
4000 | ||
4001 | @item off_t l_start | |
4002 | This specifies the offset of the start of the region to which the lock | |
9739d2d5 | 4003 | applies, and is given in bytes relative to the point specified by the |
28f540f4 RM |
4004 | @code{l_whence} member. |
4005 | ||
4006 | @item off_t l_len | |
4007 | This specifies the length of the region to be locked. A value of | |
4008 | @code{0} is treated specially; it means the region extends to the end of | |
4009 | the file. | |
4010 | ||
4011 | @item pid_t l_pid | |
4012 | This field is the process ID (@pxref{Process Creation Concepts}) of the | |
4013 | process holding the lock. It is filled in by calling @code{fcntl} with | |
0961f7e1 JL |
4014 | the @code{F_GETLK} command, but is ignored when making a lock. If the |
4015 | conflicting lock is an open file description lock | |
4016 | (@pxref{Open File Description Locks}), then this field will be set to | |
4017 | @math{-1}. | |
28f540f4 RM |
4018 | @end table |
4019 | @end deftp | |
4020 | ||
28f540f4 | 4021 | @deftypevr Macro int F_GETLK |
d08a7e4c | 4022 | @standards{POSIX.1, fcntl.h} |
28f540f4 RM |
4023 | This macro is used as the @var{command} argument to @code{fcntl}, to |
4024 | specify that it should get information about a lock. This command | |
4025 | requires a third argument of type @w{@code{struct flock *}} to be passed | |
4026 | to @code{fcntl}, so that the form of the call is: | |
4027 | ||
4028 | @smallexample | |
4029 | fcntl (@var{filedes}, F_GETLK, @var{lockp}) | |
4030 | @end smallexample | |
4031 | ||
4032 | If there is a lock already in place that would block the lock described | |
4033 | by the @var{lockp} argument, information about that lock overwrites | |
4034 | @code{*@var{lockp}}. Existing locks are not reported if they are | |
4035 | compatible with making a new lock as specified. Thus, you should | |
4036 | specify a lock type of @code{F_WRLCK} if you want to find out about both | |
4037 | read and write locks, or @code{F_RDLCK} if you want to find out about | |
4038 | write locks only. | |
4039 | ||
4040 | There might be more than one lock affecting the region specified by the | |
4041 | @var{lockp} argument, but @code{fcntl} only returns information about | |
4042 | one of them. The @code{l_whence} member of the @var{lockp} structure is | |
4043 | set to @code{SEEK_SET} and the @code{l_start} and @code{l_len} fields | |
4044 | set to identify the locked region. | |
4045 | ||
4046 | If no lock applies, the only change to the @var{lockp} structure is to | |
4047 | update the @code{l_type} to a value of @code{F_UNLCK}. | |
4048 | ||
4049 | The normal return value from @code{fcntl} with this command is an | |
07435eb4 | 4050 | unspecified value other than @math{-1}, which is reserved to indicate an |
28f540f4 RM |
4051 | error. The following @code{errno} error conditions are defined for |
4052 | this command: | |
4053 | ||
4054 | @table @code | |
4055 | @item EBADF | |
4056 | The @var{filedes} argument is invalid. | |
4057 | ||
4058 | @item EINVAL | |
4059 | Either the @var{lockp} argument doesn't specify valid lock information, | |
4060 | or the file associated with @var{filedes} doesn't support locks. | |
4061 | @end table | |
4062 | @end deftypevr | |
4063 | ||
28f540f4 | 4064 | @deftypevr Macro int F_SETLK |
d08a7e4c | 4065 | @standards{POSIX.1, fcntl.h} |
28f540f4 RM |
4066 | This macro is used as the @var{command} argument to @code{fcntl}, to |
4067 | specify that it should set or clear a lock. This command requires a | |
4068 | third argument of type @w{@code{struct flock *}} to be passed to | |
4069 | @code{fcntl}, so that the form of the call is: | |
4070 | ||
4071 | @smallexample | |
4072 | fcntl (@var{filedes}, F_SETLK, @var{lockp}) | |
4073 | @end smallexample | |
4074 | ||
4075 | If the process already has a lock on any part of the region, the old lock | |
4076 | on that part is replaced with the new lock. You can remove a lock | |
4077 | by specifying a lock type of @code{F_UNLCK}. | |
4078 | ||
4079 | If the lock cannot be set, @code{fcntl} returns immediately with a value | |
9739d2d5 RJ |
4080 | of @math{-1}. This function does not block while waiting for other processes |
4081 | to release locks. If @code{fcntl} succeeds, it returns a value other | |
07435eb4 | 4082 | than @math{-1}. |
28f540f4 RM |
4083 | |
4084 | The following @code{errno} error conditions are defined for this | |
4085 | function: | |
4086 | ||
4087 | @table @code | |
4088 | @item EAGAIN | |
4089 | @itemx EACCES | |
4090 | The lock cannot be set because it is blocked by an existing lock on the | |
4091 | file. Some systems use @code{EAGAIN} in this case, and other systems | |
4092 | use @code{EACCES}; your program should treat them alike, after | |
a7a93d50 | 4093 | @code{F_SETLK}. (@gnulinuxhurdsystems{} always use @code{EAGAIN}.) |
28f540f4 RM |
4094 | |
4095 | @item EBADF | |
4096 | Either: the @var{filedes} argument is invalid; you requested a read lock | |
4097 | but the @var{filedes} is not open for read access; or, you requested a | |
4098 | write lock but the @var{filedes} is not open for write access. | |
4099 | ||
4100 | @item EINVAL | |
4101 | Either the @var{lockp} argument doesn't specify valid lock information, | |
4102 | or the file associated with @var{filedes} doesn't support locks. | |
4103 | ||
4104 | @item ENOLCK | |
4105 | The system has run out of file lock resources; there are already too | |
4106 | many file locks in place. | |
4107 | ||
4108 | Well-designed file systems never report this error, because they have no | |
4109 | limitation on the number of locks. However, you must still take account | |
4110 | of the possibility of this error, as it could result from network access | |
4111 | to a file system on another machine. | |
4112 | @end table | |
4113 | @end deftypevr | |
4114 | ||
28f540f4 | 4115 | @deftypevr Macro int F_SETLKW |
d08a7e4c | 4116 | @standards{POSIX.1, fcntl.h} |
28f540f4 RM |
4117 | This macro is used as the @var{command} argument to @code{fcntl}, to |
4118 | specify that it should set or clear a lock. It is just like the | |
4119 | @code{F_SETLK} command, but causes the process to block (or wait) | |
4120 | until the request can be specified. | |
4121 | ||
4122 | This command requires a third argument of type @code{struct flock *}, as | |
4123 | for the @code{F_SETLK} command. | |
4124 | ||
4125 | The @code{fcntl} return values and errors are the same as for the | |
4126 | @code{F_SETLK} command, but these additional @code{errno} error conditions | |
4127 | are defined for this command: | |
4128 | ||
4129 | @table @code | |
4130 | @item EINTR | |
4131 | The function was interrupted by a signal while it was waiting. | |
4132 | @xref{Interrupted Primitives}. | |
4133 | ||
4134 | @item EDEADLK | |
4135 | The specified region is being locked by another process. But that | |
4136 | process is waiting to lock a region which the current process has | |
4137 | locked, so waiting for the lock would result in deadlock. The system | |
4138 | does not guarantee that it will detect all such conditions, but it lets | |
4139 | you know if it notices one. | |
4140 | @end table | |
4141 | @end deftypevr | |
4142 | ||
4143 | ||
4144 | The following macros are defined for use as values for the @code{l_type} | |
4145 | member of the @code{flock} structure. The values are integer constants. | |
4146 | ||
2fe82ca6 | 4147 | @vtable @code |
28f540f4 | 4148 | @item F_RDLCK |
d08a7e4c | 4149 | @standards{POSIX.1, fcntl.h} |
28f540f4 RM |
4150 | This macro is used to specify a read (or shared) lock. |
4151 | ||
28f540f4 | 4152 | @item F_WRLCK |
d08a7e4c | 4153 | @standards{POSIX.1, fcntl.h} |
28f540f4 RM |
4154 | This macro is used to specify a write (or exclusive) lock. |
4155 | ||
28f540f4 | 4156 | @item F_UNLCK |
d08a7e4c | 4157 | @standards{POSIX.1, fcntl.h} |
28f540f4 | 4158 | This macro is used to specify that the region is unlocked. |
2fe82ca6 | 4159 | @end vtable |
28f540f4 RM |
4160 | |
4161 | As an example of a situation where file locking is useful, consider a | |
4162 | program that can be run simultaneously by several different users, that | |
4163 | logs status information to a common file. One example of such a program | |
4164 | might be a game that uses a file to keep track of high scores. Another | |
4165 | example might be a program that records usage or accounting information | |
4166 | for billing purposes. | |
4167 | ||
4168 | Having multiple copies of the program simultaneously writing to the | |
4169 | file could cause the contents of the file to become mixed up. But | |
4170 | you can prevent this kind of problem by setting a write lock on the | |
2c6fe0bd | 4171 | file before actually writing to the file. |
28f540f4 RM |
4172 | |
4173 | If the program also needs to read the file and wants to make sure that | |
4174 | the contents of the file are in a consistent state, then it can also use | |
4175 | a read lock. While the read lock is set, no other process can lock | |
4176 | that part of the file for writing. | |
4177 | ||
4178 | @c ??? This section could use an example program. | |
4179 | ||
0961f7e1 | 4180 | Remember that file locks are only an @emph{advisory} protocol for |
28f540f4 RM |
4181 | controlling access to a file. There is still potential for access to |
4182 | the file by programs that don't use the lock protocol. | |
4183 | ||
0961f7e1 JL |
4184 | @node Open File Description Locks |
4185 | @section Open File Description Locks | |
4186 | ||
4187 | In contrast to process-associated record locks (@pxref{File Locks}), | |
4188 | open file description record locks are associated with an open file | |
4189 | description rather than a process. | |
4190 | ||
4191 | Using @code{fcntl} to apply an open file description lock on a region that | |
4192 | already has an existing open file description lock that was created via the | |
4193 | same file descriptor will never cause a lock conflict. | |
4194 | ||
4195 | Open file description locks are also inherited by child processes across | |
4196 | @code{fork}, or @code{clone} with @code{CLONE_FILES} set | |
4197 | (@pxref{Creating a Process}), along with the file descriptor. | |
4198 | ||
4199 | It is important to distinguish between the open file @emph{description} (an | |
4200 | instance of an open file, usually created by a call to @code{open}) and | |
4201 | an open file @emph{descriptor}, which is a numeric value that refers to the | |
4202 | open file description. The locks described here are associated with the | |
4203 | open file @emph{description} and not the open file @emph{descriptor}. | |
4204 | ||
4205 | Using @code{dup} (@pxref{Duplicating Descriptors}) to copy a file | |
4206 | descriptor does not give you a new open file description, but rather copies a | |
4207 | reference to an existing open file description and assigns it to a new | |
4208 | file descriptor. Thus, open file description locks set on a file | |
4209 | descriptor cloned by @code{dup} will never conflict with open file | |
4210 | description locks set on the original descriptor since they refer to the | |
4211 | same open file description. Depending on the range and type of lock | |
4212 | involved, the original lock may be modified by a @code{F_OFD_SETLK} or | |
4213 | @code{F_OFD_SETLKW} command in this situation however. | |
4214 | ||
4215 | Open file description locks always conflict with process-associated locks, | |
4216 | even if acquired by the same process or on the same open file | |
4217 | descriptor. | |
4218 | ||
4219 | Open file description locks use the same @code{struct flock} as | |
4220 | process-associated locks as an argument (@pxref{File Locks}) and the | |
4221 | macros for the @code{command} values are also declared in the header file | |
4222 | @file{fcntl.h}. To use them, the macro @code{_GNU_SOURCE} must be | |
4223 | defined prior to including any header file. | |
4224 | ||
4225 | In contrast to process-associated locks, any @code{struct flock} used as | |
4226 | an argument to open file description lock commands must have the @code{l_pid} | |
4227 | value set to @math{0}. Also, when returning information about an | |
4228 | open file description lock in a @code{F_GETLK} or @code{F_OFD_GETLK} request, | |
4229 | the @code{l_pid} field in @code{struct flock} will be set to @math{-1} | |
4230 | to indicate that the lock is not associated with a process. | |
4231 | ||
4232 | When the same @code{struct flock} is reused as an argument to a | |
4233 | @code{F_OFD_SETLK} or @code{F_OFD_SETLKW} request after being used for an | |
4234 | @code{F_OFD_GETLK} request, it is necessary to inspect and reset the | |
4235 | @code{l_pid} field to @math{0}. | |
4236 | ||
4237 | @pindex fcntl.h. | |
4238 | ||
4239 | @deftypevr Macro int F_OFD_GETLK | |
4240 | This macro is used as the @var{command} argument to @code{fcntl}, to | |
4241 | specify that it should get information about a lock. This command | |
4242 | requires a third argument of type @w{@code{struct flock *}} to be passed | |
4243 | to @code{fcntl}, so that the form of the call is: | |
4244 | ||
4245 | @smallexample | |
4246 | fcntl (@var{filedes}, F_OFD_GETLK, @var{lockp}) | |
4247 | @end smallexample | |
4248 | ||
4249 | If there is a lock already in place that would block the lock described | |
4250 | by the @var{lockp} argument, information about that lock is written to | |
4251 | @code{*@var{lockp}}. Existing locks are not reported if they are | |
4252 | compatible with making a new lock as specified. Thus, you should | |
4253 | specify a lock type of @code{F_WRLCK} if you want to find out about both | |
4254 | read and write locks, or @code{F_RDLCK} if you want to find out about | |
4255 | write locks only. | |
4256 | ||
4257 | There might be more than one lock affecting the region specified by the | |
4258 | @var{lockp} argument, but @code{fcntl} only returns information about | |
4259 | one of them. Which lock is returned in this situation is undefined. | |
4260 | ||
4261 | The @code{l_whence} member of the @var{lockp} structure are set to | |
4262 | @code{SEEK_SET} and the @code{l_start} and @code{l_len} fields are set | |
4263 | to identify the locked region. | |
4264 | ||
4265 | If no conflicting lock exists, the only change to the @var{lockp} structure | |
4266 | is to update the @code{l_type} field to the value @code{F_UNLCK}. | |
4267 | ||
4268 | The normal return value from @code{fcntl} with this command is either @math{0} | |
4269 | on success or @math{-1}, which indicates an error. The following @code{errno} | |
4270 | error conditions are defined for this command: | |
4271 | ||
4272 | @table @code | |
4273 | @item EBADF | |
4274 | The @var{filedes} argument is invalid. | |
4275 | ||
4276 | @item EINVAL | |
4277 | Either the @var{lockp} argument doesn't specify valid lock information, | |
4278 | the operating system kernel doesn't support open file description locks, or the file | |
4279 | associated with @var{filedes} doesn't support locks. | |
4280 | @end table | |
4281 | @end deftypevr | |
4282 | ||
0961f7e1 | 4283 | @deftypevr Macro int F_OFD_SETLK |
d08a7e4c | 4284 | @standards{POSIX.1, fcntl.h} |
0961f7e1 JL |
4285 | This macro is used as the @var{command} argument to @code{fcntl}, to |
4286 | specify that it should set or clear a lock. This command requires a | |
4287 | third argument of type @w{@code{struct flock *}} to be passed to | |
4288 | @code{fcntl}, so that the form of the call is: | |
4289 | ||
4290 | @smallexample | |
4291 | fcntl (@var{filedes}, F_OFD_SETLK, @var{lockp}) | |
4292 | @end smallexample | |
4293 | ||
4294 | If the open file already has a lock on any part of the | |
4295 | region, the old lock on that part is replaced with the new lock. You | |
4296 | can remove a lock by specifying a lock type of @code{F_UNLCK}. | |
4297 | ||
4298 | If the lock cannot be set, @code{fcntl} returns immediately with a value | |
4299 | of @math{-1}. This command does not wait for other tasks | |
4300 | to release locks. If @code{fcntl} succeeds, it returns @math{0}. | |
4301 | ||
4302 | The following @code{errno} error conditions are defined for this | |
4303 | command: | |
4304 | ||
4305 | @table @code | |
4306 | @item EAGAIN | |
4307 | The lock cannot be set because it is blocked by an existing lock on the | |
4308 | file. | |
4309 | ||
4310 | @item EBADF | |
4311 | Either: the @var{filedes} argument is invalid; you requested a read lock | |
4312 | but the @var{filedes} is not open for read access; or, you requested a | |
4313 | write lock but the @var{filedes} is not open for write access. | |
4314 | ||
4315 | @item EINVAL | |
4316 | Either the @var{lockp} argument doesn't specify valid lock information, | |
4317 | the operating system kernel doesn't support open file description locks, or the | |
4318 | file associated with @var{filedes} doesn't support locks. | |
4319 | ||
4320 | @item ENOLCK | |
4321 | The system has run out of file lock resources; there are already too | |
4322 | many file locks in place. | |
4323 | ||
4324 | Well-designed file systems never report this error, because they have no | |
4325 | limitation on the number of locks. However, you must still take account | |
4326 | of the possibility of this error, as it could result from network access | |
4327 | to a file system on another machine. | |
4328 | @end table | |
4329 | @end deftypevr | |
4330 | ||
0961f7e1 | 4331 | @deftypevr Macro int F_OFD_SETLKW |
d08a7e4c | 4332 | @standards{POSIX.1, fcntl.h} |
0961f7e1 JL |
4333 | This macro is used as the @var{command} argument to @code{fcntl}, to |
4334 | specify that it should set or clear a lock. It is just like the | |
4335 | @code{F_OFD_SETLK} command, but causes the process to wait until the request | |
4336 | can be completed. | |
4337 | ||
4338 | This command requires a third argument of type @code{struct flock *}, as | |
4339 | for the @code{F_OFD_SETLK} command. | |
4340 | ||
4341 | The @code{fcntl} return values and errors are the same as for the | |
4342 | @code{F_OFD_SETLK} command, but these additional @code{errno} error conditions | |
4343 | are defined for this command: | |
4344 | ||
4345 | @table @code | |
4346 | @item EINTR | |
4347 | The function was interrupted by a signal while it was waiting. | |
4348 | @xref{Interrupted Primitives}. | |
4349 | ||
4350 | @end table | |
4351 | @end deftypevr | |
4352 | ||
4353 | Open file description locks are useful in the same sorts of situations as | |
4354 | process-associated locks. They can also be used to synchronize file | |
4355 | access between threads within the same process by having each thread perform | |
4356 | its own @code{open} of the file, to obtain its own open file description. | |
4357 | ||
4358 | Because open file description locks are automatically freed only upon | |
4359 | closing the last file descriptor that refers to the open file | |
4360 | description, this locking mechanism avoids the possibility that locks | |
4361 | are inadvertently released due to a library routine opening and closing | |
4362 | a file without the application being aware. | |
4363 | ||
4364 | As with process-associated locks, open file description locks are advisory. | |
4365 | ||
4366 | @node Open File Description Locks Example | |
4367 | @section Open File Description Locks Example | |
4368 | ||
4369 | Here is an example of using open file description locks in a threaded | |
4370 | program. If this program used process-associated locks, then it would be | |
4371 | subject to data corruption because process-associated locks are shared | |
4372 | by the threads inside a process, and thus cannot be used by one thread | |
4373 | to lock out another thread in the same process. | |
4374 | ||
4375 | Proper error handling has been omitted in the following program for | |
4376 | brevity. | |
4377 | ||
4378 | @smallexample | |
4379 | @include ofdlocks.c.texi | |
4380 | @end smallexample | |
4381 | ||
4382 | This example creates three threads each of which loops five times, | |
4383 | appending to the file. Access to the file is serialized via open file | |
4384 | description locks. If we compile and run the above program, we'll end up | |
4385 | with /tmp/foo that has 15 lines in it. | |
4386 | ||
4387 | If we, however, were to replace the @code{F_OFD_SETLK} and | |
4388 | @code{F_OFD_SETLKW} commands with their process-associated lock | |
4389 | equivalents, the locking essentially becomes a noop since it is all done | |
4390 | within the context of the same process. That leads to data corruption | |
4391 | (typically manifested as missing lines) as some threads race in and | |
4392 | overwrite the data written by others. | |
4393 | ||
28f540f4 RM |
4394 | @node Interrupt Input |
4395 | @section Interrupt-Driven Input | |
4396 | ||
4397 | @cindex interrupt-driven input | |
4398 | If you set the @code{O_ASYNC} status flag on a file descriptor | |
4399 | (@pxref{File Status Flags}), a @code{SIGIO} signal is sent whenever | |
4400 | input or output becomes possible on that file descriptor. The process | |
4401 | or process group to receive the signal can be selected by using the | |
4402 | @code{F_SETOWN} command to the @code{fcntl} function. If the file | |
4403 | descriptor is a socket, this also selects the recipient of @code{SIGURG} | |
4404 | signals that are delivered when out-of-band data arrives on that socket; | |
4405 | see @ref{Out-of-Band Data}. (@code{SIGURG} is sent in any situation | |
4406 | where @code{select} would report the socket as having an ``exceptional | |
4407 | condition''. @xref{Waiting for I/O}.) | |
4408 | ||
4409 | If the file descriptor corresponds to a terminal device, then @code{SIGIO} | |
2c6fe0bd | 4410 | signals are sent to the foreground process group of the terminal. |
28f540f4 RM |
4411 | @xref{Job Control}. |
4412 | ||
4413 | @pindex fcntl.h | |
4414 | The symbols in this section are defined in the header file | |
4415 | @file{fcntl.h}. | |
4416 | ||
28f540f4 | 4417 | @deftypevr Macro int F_GETOWN |
d08a7e4c | 4418 | @standards{BSD, fcntl.h} |
28f540f4 RM |
4419 | This macro is used as the @var{command} argument to @code{fcntl}, to |
4420 | specify that it should get information about the process or process | |
4421 | group to which @code{SIGIO} signals are sent. (For a terminal, this is | |
4422 | actually the foreground process group ID, which you can get using | |
4423 | @code{tcgetpgrp}; see @ref{Terminal Access Functions}.) | |
4424 | ||
4425 | The return value is interpreted as a process ID; if negative, its | |
4426 | absolute value is the process group ID. | |
4427 | ||
4428 | The following @code{errno} error condition is defined for this command: | |
4429 | ||
4430 | @table @code | |
4431 | @item EBADF | |
4432 | The @var{filedes} argument is invalid. | |
4433 | @end table | |
4434 | @end deftypevr | |
4435 | ||
28f540f4 | 4436 | @deftypevr Macro int F_SETOWN |
d08a7e4c | 4437 | @standards{BSD, fcntl.h} |
28f540f4 RM |
4438 | This macro is used as the @var{command} argument to @code{fcntl}, to |
4439 | specify that it should set the process or process group to which | |
4440 | @code{SIGIO} signals are sent. This command requires a third argument | |
4441 | of type @code{pid_t} to be passed to @code{fcntl}, so that the form of | |
4442 | the call is: | |
4443 | ||
4444 | @smallexample | |
4445 | fcntl (@var{filedes}, F_SETOWN, @var{pid}) | |
4446 | @end smallexample | |
4447 | ||
4448 | The @var{pid} argument should be a process ID. You can also pass a | |
4449 | negative number whose absolute value is a process group ID. | |
4450 | ||
07435eb4 | 4451 | The return value from @code{fcntl} with this command is @math{-1} |
28f540f4 RM |
4452 | in case of error and some other value if successful. The following |
4453 | @code{errno} error conditions are defined for this command: | |
4454 | ||
4455 | @table @code | |
4456 | @item EBADF | |
4457 | The @var{filedes} argument is invalid. | |
4458 | ||
4459 | @item ESRCH | |
4460 | There is no process or process group corresponding to @var{pid}. | |
4461 | @end table | |
4462 | @end deftypevr | |
4463 | ||
4464 | @c ??? This section could use an example program. | |
07435eb4 UD |
4465 | |
4466 | @node IOCTLs | |
4467 | @section Generic I/O Control operations | |
4468 | @cindex generic i/o control operations | |
4469 | @cindex IOCTLs | |
4470 | ||
a7a93d50 | 4471 | @gnusystems{} can handle most input/output operations on many different |
07435eb4 UD |
4472 | devices and objects in terms of a few file primitives - @code{read}, |
4473 | @code{write} and @code{lseek}. However, most devices also have a few | |
cf822e3c | 4474 | peculiar operations which do not fit into this model. Such as: |
07435eb4 UD |
4475 | |
4476 | @itemize @bullet | |
4477 | ||
4478 | @item | |
4479 | Changing the character font used on a terminal. | |
4480 | ||
4481 | @item | |
4482 | Telling a magnetic tape system to rewind or fast forward. (Since they | |
4483 | cannot move in byte increments, @code{lseek} is inapplicable). | |
4484 | ||
4485 | @item | |
4486 | Ejecting a disk from a drive. | |
4487 | ||
4488 | @item | |
4489 | Playing an audio track from a CD-ROM drive. | |
4490 | ||
4491 | @item | |
4492 | Maintaining routing tables for a network. | |
4493 | ||
4494 | @end itemize | |
4495 | ||
4496 | Although some such objects such as sockets and terminals | |
4497 | @footnote{Actually, the terminal-specific functions are implemented with | |
4498 | IOCTLs on many platforms.} have special functions of their own, it would | |
4499 | not be practical to create functions for all these cases. | |
4500 | ||
4501 | Instead these minor operations, known as @dfn{IOCTL}s, are assigned code | |
4502 | numbers and multiplexed through the @code{ioctl} function, defined in | |
4503 | @code{sys/ioctl.h}. The code numbers themselves are defined in many | |
4504 | different headers. | |
4505 | ||
4506 | @deftypefun int ioctl (int @var{filedes}, int @var{command}, @dots{}) | |
d08a7e4c | 4507 | @standards{BSD, sys/ioctl.h} |
2cc3615c | 4508 | @safety{@prelim{}@mtsafe{}@assafe{}@acsafe{}} |
07435eb4 UD |
4509 | |
4510 | The @code{ioctl} function performs the generic I/O operation | |
4511 | @var{command} on @var{filedes}. | |
4512 | ||
4513 | A third argument is usually present, either a single number or a pointer | |
4514 | to a structure. The meaning of this argument, the returned value, and | |
4515 | any error codes depends upon the command used. Often @math{-1} is | |
4516 | returned for a failure. | |
4517 | ||
4518 | @end deftypefun | |
4519 | ||
4520 | On some systems, IOCTLs used by different devices share the same numbers. | |
4521 | Thus, although use of an inappropriate IOCTL @emph{usually} only produces | |
4522 | an error, you should not attempt to use device-specific IOCTLs on an | |
4523 | unknown device. | |
4524 | ||
4525 | Most IOCTLs are OS-specific and/or only used in special system utilities, | |
4526 | and are thus beyond the scope of this document. For an example of the use | |
8b7fb588 | 4527 | of an IOCTL, see @ref{Out-of-Band Data}. |
2cc3615c AO |
4528 | |
4529 | @c FIXME this is undocumented: | |
4530 | @c dup3 |