]>
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. |
07435eb4 | 44 | * Memory-mapped I/O:: Using files like memory. |
28f540f4 RM |
45 | * Waiting for I/O:: How to check for input or output |
46 | on multiple file descriptors. | |
dfd2257a | 47 | * Synchronizing I/O:: Making sure all I/O actions completed. |
b07d03e0 | 48 | * Asynchronous I/O:: Perform I/O in parallel. |
28f540f4 RM |
49 | * Control Operations:: Various other operations on file |
50 | descriptors. | |
51 | * Duplicating Descriptors:: Fcntl commands for duplicating | |
52 | file descriptors. | |
53 | * Descriptor Flags:: Fcntl commands for manipulating | |
54 | flags associated with file | |
2c6fe0bd | 55 | descriptors. |
28f540f4 RM |
56 | * File Status Flags:: Fcntl commands for manipulating |
57 | flags associated with open files. | |
58 | * File Locks:: Fcntl commands for implementing | |
59 | file locking. | |
60 | * Interrupt Input:: Getting an asynchronous signal when | |
61 | input arrives. | |
07435eb4 | 62 | * IOCTLs:: Generic I/O Control operations. |
28f540f4 RM |
63 | @end menu |
64 | ||
65 | ||
66 | @node Opening and Closing Files | |
67 | @section Opening and Closing Files | |
68 | ||
69 | @cindex opening a file descriptor | |
70 | @cindex closing a file descriptor | |
71 | This section describes the primitives for opening and closing files | |
72 | using file descriptors. The @code{open} and @code{creat} functions are | |
73 | declared in the header file @file{fcntl.h}, while @code{close} is | |
74 | declared in @file{unistd.h}. | |
75 | @pindex unistd.h | |
76 | @pindex fcntl.h | |
77 | ||
78 | @comment fcntl.h | |
79 | @comment POSIX.1 | |
80 | @deftypefun int open (const char *@var{filename}, int @var{flags}[, mode_t @var{mode}]) | |
81 | The @code{open} function creates and returns a new file descriptor | |
82 | for the file named by @var{filename}. Initially, the file position | |
83 | indicator for the file is at the beginning of the file. The argument | |
84 | @var{mode} is used only when a file is created, but it doesn't hurt | |
85 | to supply the argument in any case. | |
86 | ||
87 | The @var{flags} argument controls how the file is to be opened. This is | |
88 | a bit mask; you create the value by the bitwise OR of the appropriate | |
89 | parameters (using the @samp{|} operator in C). | |
90 | @xref{File Status Flags}, for the parameters available. | |
91 | ||
92 | The normal return value from @code{open} is a non-negative integer file | |
07435eb4 | 93 | descriptor. In the case of an error, a value of @math{-1} is returned |
28f540f4 RM |
94 | instead. In addition to the usual file name errors (@pxref{File |
95 | Name Errors}), the following @code{errno} error conditions are defined | |
96 | for this function: | |
97 | ||
98 | @table @code | |
99 | @item EACCES | |
19e4c7dd AJ |
100 | The file exists but is not readable/writable as requested by the @var{flags} |
101 | argument, the file does not exist and the directory is unwritable so | |
28f540f4 RM |
102 | it cannot be created. |
103 | ||
104 | @item EEXIST | |
105 | Both @code{O_CREAT} and @code{O_EXCL} are set, and the named file already | |
106 | exists. | |
107 | ||
108 | @item EINTR | |
109 | The @code{open} operation was interrupted by a signal. | |
110 | @xref{Interrupted Primitives}. | |
111 | ||
112 | @item EISDIR | |
113 | The @var{flags} argument specified write access, and the file is a directory. | |
114 | ||
115 | @item EMFILE | |
116 | The process has too many files open. | |
117 | The maximum number of file descriptors is controlled by the | |
118 | @code{RLIMIT_NOFILE} resource limit; @pxref{Limits on Resources}. | |
119 | ||
120 | @item ENFILE | |
121 | The entire system, or perhaps the file system which contains the | |
122 | directory, cannot support any additional open files at the moment. | |
123 | (This problem cannot happen on the GNU system.) | |
124 | ||
125 | @item ENOENT | |
126 | The named file does not exist, and @code{O_CREAT} is not specified. | |
127 | ||
128 | @item ENOSPC | |
129 | The directory or file system that would contain the new file cannot be | |
130 | extended, because there is no disk space left. | |
131 | ||
132 | @item ENXIO | |
133 | @code{O_NONBLOCK} and @code{O_WRONLY} are both set in the @var{flags} | |
134 | argument, the file named by @var{filename} is a FIFO (@pxref{Pipes and | |
135 | FIFOs}), and no process has the file open for reading. | |
136 | ||
137 | @item EROFS | |
138 | The file resides on a read-only file system and any of @w{@code{O_WRONLY}}, | |
139 | @code{O_RDWR}, and @code{O_TRUNC} are set in the @var{flags} argument, | |
140 | or @code{O_CREAT} is set and the file does not already exist. | |
141 | @end table | |
142 | ||
143 | @c !!! umask | |
144 | ||
04b9968b | 145 | If on a 32 bit machine the sources are translated with |
b07d03e0 UD |
146 | @code{_FILE_OFFSET_BITS == 64} the function @code{open} returns a file |
147 | descriptor opened in the large file mode which enables the file handling | |
fed8f7f7 | 148 | functions to use files up to @math{2^63} bytes in size and offset from |
b07d03e0 UD |
149 | @math{-2^63} to @math{2^63}. This happens transparently for the user |
150 | since all of the lowlevel file handling functions are equally replaced. | |
151 | ||
04b9968b | 152 | This function is a cancellation point in multi-threaded programs. This |
dfd2257a UD |
153 | is a problem if the thread allocates some resources (like memory, file |
154 | descriptors, semaphores or whatever) at the time @code{open} is | |
19e4c7dd | 155 | called. If the thread gets canceled these resources stay allocated |
dfd2257a | 156 | until the program ends. To avoid this calls to @code{open} should be |
04b9968b | 157 | protected using cancellation handlers. |
dfd2257a UD |
158 | @c ref pthread_cleanup_push / pthread_cleanup_pop |
159 | ||
28f540f4 RM |
160 | The @code{open} function is the underlying primitive for the @code{fopen} |
161 | and @code{freopen} functions, that create streams. | |
162 | @end deftypefun | |
163 | ||
b07d03e0 | 164 | @comment fcntl.h |
a3a4a74e | 165 | @comment Unix98 |
b07d03e0 UD |
166 | @deftypefun int open64 (const char *@var{filename}, int @var{flags}[, mode_t @var{mode}]) |
167 | This function is similar to @code{open}. It returns a file descriptor | |
168 | which can be used to access the file named by @var{filename}. The only | |
04b9968b | 169 | difference is that on 32 bit systems the file is opened in the |
b07d03e0 UD |
170 | large file mode. I.e., file length and file offsets can exceed 31 bits. |
171 | ||
b07d03e0 UD |
172 | When the sources are translated with @code{_FILE_OFFSET_BITS == 64} this |
173 | function is actually available under the name @code{open}. I.e., the | |
174 | new, extended API using 64 bit file sizes and offsets transparently | |
175 | replaces the old API. | |
176 | @end deftypefun | |
177 | ||
28f540f4 RM |
178 | @comment fcntl.h |
179 | @comment POSIX.1 | |
180 | @deftypefn {Obsolete function} int creat (const char *@var{filename}, mode_t @var{mode}) | |
181 | This function is obsolete. The call: | |
182 | ||
183 | @smallexample | |
184 | creat (@var{filename}, @var{mode}) | |
185 | @end smallexample | |
186 | ||
187 | @noindent | |
188 | is equivalent to: | |
189 | ||
190 | @smallexample | |
191 | open (@var{filename}, O_WRONLY | O_CREAT | O_TRUNC, @var{mode}) | |
192 | @end smallexample | |
b07d03e0 | 193 | |
04b9968b | 194 | If on a 32 bit machine the sources are translated with |
b07d03e0 UD |
195 | @code{_FILE_OFFSET_BITS == 64} the function @code{creat} returns a file |
196 | descriptor opened in the large file mode which enables the file handling | |
197 | functions to use files up to @math{2^63} in size and offset from | |
198 | @math{-2^63} to @math{2^63}. This happens transparently for the user | |
199 | since all of the lowlevel file handling functions are equally replaced. | |
200 | @end deftypefn | |
201 | ||
202 | @comment fcntl.h | |
a3a4a74e | 203 | @comment Unix98 |
b07d03e0 UD |
204 | @deftypefn {Obsolete function} int creat64 (const char *@var{filename}, mode_t @var{mode}) |
205 | This function is similar to @code{creat}. It returns a file descriptor | |
206 | which can be used to access the file named by @var{filename}. The only | |
04b9968b | 207 | the difference is that on 32 bit systems the file is opened in the |
b07d03e0 UD |
208 | large file mode. I.e., file length and file offsets can exceed 31 bits. |
209 | ||
210 | To use this file descriptor one must not use the normal operations but | |
211 | instead the counterparts named @code{*64}, e.g., @code{read64}. | |
212 | ||
213 | When the sources are translated with @code{_FILE_OFFSET_BITS == 64} this | |
214 | function is actually available under the name @code{open}. I.e., the | |
215 | new, extended API using 64 bit file sizes and offsets transparently | |
216 | replaces the old API. | |
28f540f4 RM |
217 | @end deftypefn |
218 | ||
219 | @comment unistd.h | |
220 | @comment POSIX.1 | |
221 | @deftypefun int close (int @var{filedes}) | |
222 | The function @code{close} closes the file descriptor @var{filedes}. | |
223 | Closing a file has the following consequences: | |
224 | ||
225 | @itemize @bullet | |
2c6fe0bd | 226 | @item |
28f540f4 RM |
227 | The file descriptor is deallocated. |
228 | ||
229 | @item | |
230 | Any record locks owned by the process on the file are unlocked. | |
231 | ||
232 | @item | |
233 | When all file descriptors associated with a pipe or FIFO have been closed, | |
234 | any unread data is discarded. | |
235 | @end itemize | |
236 | ||
04b9968b | 237 | This function is a cancellation point in multi-threaded programs. This |
dfd2257a UD |
238 | is a problem if the thread allocates some resources (like memory, file |
239 | descriptors, semaphores or whatever) at the time @code{close} is | |
19e4c7dd | 240 | called. If the thread gets canceled these resources stay allocated |
04b9968b UD |
241 | until the program ends. To avoid this, calls to @code{close} should be |
242 | protected using cancellation handlers. | |
dfd2257a UD |
243 | @c ref pthread_cleanup_push / pthread_cleanup_pop |
244 | ||
07435eb4 | 245 | The normal return value from @code{close} is @math{0}; a value of @math{-1} |
28f540f4 RM |
246 | is returned in case of failure. The following @code{errno} error |
247 | conditions are defined for this function: | |
248 | ||
249 | @table @code | |
250 | @item EBADF | |
251 | The @var{filedes} argument is not a valid file descriptor. | |
252 | ||
253 | @item EINTR | |
254 | The @code{close} call was interrupted by a signal. | |
255 | @xref{Interrupted Primitives}. | |
256 | Here is an example of how to handle @code{EINTR} properly: | |
257 | ||
258 | @smallexample | |
259 | TEMP_FAILURE_RETRY (close (desc)); | |
260 | @end smallexample | |
261 | ||
262 | @item ENOSPC | |
263 | @itemx EIO | |
264 | @itemx EDQUOT | |
2c6fe0bd | 265 | When the file is accessed by NFS, these errors from @code{write} can sometimes |
28f540f4 RM |
266 | not be detected until @code{close}. @xref{I/O Primitives}, for details |
267 | on their meaning. | |
268 | @end table | |
b07d03e0 UD |
269 | |
270 | Please note that there is @emph{no} separate @code{close64} function. | |
271 | This is not necessary since this function does not determine nor depend | |
fed8f7f7 | 272 | on the mode of the file. The kernel which performs the @code{close} |
04b9968b | 273 | operation knows which mode the descriptor is used for and can handle |
b07d03e0 | 274 | this situation. |
28f540f4 RM |
275 | @end deftypefun |
276 | ||
277 | To close a stream, call @code{fclose} (@pxref{Closing Streams}) instead | |
278 | of trying to close its underlying file descriptor with @code{close}. | |
279 | This flushes any buffered output and updates the stream object to | |
280 | indicate that it is closed. | |
281 | ||
282 | @node I/O Primitives | |
283 | @section Input and Output Primitives | |
284 | ||
285 | This section describes the functions for performing primitive input and | |
286 | output operations on file descriptors: @code{read}, @code{write}, and | |
287 | @code{lseek}. These functions are declared in the header file | |
288 | @file{unistd.h}. | |
289 | @pindex unistd.h | |
290 | ||
291 | @comment unistd.h | |
292 | @comment POSIX.1 | |
293 | @deftp {Data Type} ssize_t | |
294 | This data type is used to represent the sizes of blocks that can be | |
295 | read or written in a single operation. It is similar to @code{size_t}, | |
296 | but must be a signed type. | |
297 | @end deftp | |
298 | ||
299 | @cindex reading from a file descriptor | |
300 | @comment unistd.h | |
301 | @comment POSIX.1 | |
302 | @deftypefun ssize_t read (int @var{filedes}, void *@var{buffer}, size_t @var{size}) | |
303 | The @code{read} function reads up to @var{size} bytes from the file | |
304 | with descriptor @var{filedes}, storing the results in the @var{buffer}. | |
04b9968b UD |
305 | (This is not necessarily a character string, and no terminating null |
306 | character is added.) | |
28f540f4 RM |
307 | |
308 | @cindex end-of-file, on a file descriptor | |
309 | The return value is the number of bytes actually read. This might be | |
310 | less than @var{size}; for example, if there aren't that many bytes left | |
311 | in the file or if there aren't that many bytes immediately available. | |
312 | The exact behavior depends on what kind of file it is. Note that | |
313 | reading less than @var{size} bytes is not an error. | |
314 | ||
315 | A value of zero indicates end-of-file (except if the value of the | |
316 | @var{size} argument is also zero). This is not considered an error. | |
317 | If you keep calling @code{read} while at end-of-file, it will keep | |
318 | returning zero and doing nothing else. | |
319 | ||
320 | If @code{read} returns at least one character, there is no way you can | |
321 | tell whether end-of-file was reached. But if you did reach the end, the | |
322 | next read will return zero. | |
323 | ||
07435eb4 | 324 | In case of an error, @code{read} returns @math{-1}. The following |
28f540f4 RM |
325 | @code{errno} error conditions are defined for this function: |
326 | ||
327 | @table @code | |
328 | @item EAGAIN | |
329 | Normally, when no input is immediately available, @code{read} waits for | |
330 | some input. But if the @code{O_NONBLOCK} flag is set for the file | |
331 | (@pxref{File Status Flags}), @code{read} returns immediately without | |
332 | reading any data, and reports this error. | |
333 | ||
334 | @strong{Compatibility Note:} Most versions of BSD Unix use a different | |
335 | error code for this: @code{EWOULDBLOCK}. In the GNU library, | |
336 | @code{EWOULDBLOCK} is an alias for @code{EAGAIN}, so it doesn't matter | |
337 | which name you use. | |
338 | ||
339 | On some systems, reading a large amount of data from a character special | |
340 | file can also fail with @code{EAGAIN} if the kernel cannot find enough | |
341 | physical memory to lock down the user's pages. This is limited to | |
342 | devices that transfer with direct memory access into the user's memory, | |
343 | which means it does not include terminals, since they always use | |
344 | separate buffers inside the kernel. This problem never happens in the | |
345 | GNU system. | |
346 | ||
347 | Any condition that could result in @code{EAGAIN} can instead result in a | |
348 | successful @code{read} which returns fewer bytes than requested. | |
349 | Calling @code{read} again immediately would result in @code{EAGAIN}. | |
350 | ||
351 | @item EBADF | |
352 | The @var{filedes} argument is not a valid file descriptor, | |
353 | or is not open for reading. | |
354 | ||
355 | @item EINTR | |
356 | @code{read} was interrupted by a signal while it was waiting for input. | |
357 | @xref{Interrupted Primitives}. A signal will not necessary cause | |
358 | @code{read} to return @code{EINTR}; it may instead result in a | |
359 | successful @code{read} which returns fewer bytes than requested. | |
360 | ||
361 | @item EIO | |
362 | For many devices, and for disk files, this error code indicates | |
363 | a hardware error. | |
364 | ||
365 | @code{EIO} also occurs when a background process tries to read from the | |
366 | controlling terminal, and the normal action of stopping the process by | |
367 | sending it a @code{SIGTTIN} signal isn't working. This might happen if | |
04b9968b | 368 | the signal is being blocked or ignored, or because the process group is |
28f540f4 RM |
369 | orphaned. @xref{Job Control}, for more information about job control, |
370 | and @ref{Signal Handling}, for information about signals. | |
7e583a52 RM |
371 | |
372 | @item EINVAL | |
373 | In some systems, when reading from a character or block device, position | |
374 | and size offsets must be aligned to a particular block size. This error | |
375 | indicates that the offsets were not properly aligned. | |
28f540f4 RM |
376 | @end table |
377 | ||
b07d03e0 UD |
378 | Please note that there is no function named @code{read64}. This is not |
379 | necessary since this function does not directly modify or handle the | |
380 | possibly wide file offset. Since the kernel handles this state | |
04b9968b | 381 | internally, the @code{read} function can be used for all cases. |
b07d03e0 | 382 | |
04b9968b | 383 | This function is a cancellation point in multi-threaded programs. This |
dfd2257a UD |
384 | is a problem if the thread allocates some resources (like memory, file |
385 | descriptors, semaphores or whatever) at the time @code{read} is | |
19e4c7dd | 386 | called. If the thread gets canceled these resources stay allocated |
04b9968b UD |
387 | until the program ends. To avoid this, calls to @code{read} should be |
388 | protected using cancellation handlers. | |
dfd2257a UD |
389 | @c ref pthread_cleanup_push / pthread_cleanup_pop |
390 | ||
28f540f4 RM |
391 | The @code{read} function is the underlying primitive for all of the |
392 | functions that read from streams, such as @code{fgetc}. | |
393 | @end deftypefun | |
394 | ||
a5a0310d UD |
395 | @comment unistd.h |
396 | @comment Unix98 | |
397 | @deftypefun ssize_t pread (int @var{filedes}, void *@var{buffer}, size_t @var{size}, off_t @var{offset}) | |
398 | The @code{pread} function is similar to the @code{read} function. The | |
04b9968b UD |
399 | first three arguments are identical, and the return values and error |
400 | codes also correspond. | |
a5a0310d UD |
401 | |
402 | The difference is the fourth argument and its handling. The data block | |
403 | is not read from the current position of the file descriptor | |
404 | @code{filedes}. Instead the data is read from the file starting at | |
405 | position @var{offset}. The position of the file descriptor itself is | |
04b9968b | 406 | not affected by the operation. The value is the same as before the call. |
a5a0310d | 407 | |
b07d03e0 UD |
408 | When the source file is compiled with @code{_FILE_OFFSET_BITS == 64} the |
409 | @code{pread} function is in fact @code{pread64} and the type | |
04b9968b | 410 | @code{off_t} has 64 bits, which makes it possible to handle files up to |
c756c71c | 411 | @math{2^63} bytes in length. |
b07d03e0 | 412 | |
a5a0310d UD |
413 | The return value of @code{pread} describes the number of bytes read. |
414 | In the error case it returns @math{-1} like @code{read} does and the | |
04b9968b UD |
415 | error codes are also the same, with these additions: |
416 | ||
a5a0310d UD |
417 | @table @code |
418 | @item EINVAL | |
419 | The value given for @var{offset} is negative and therefore illegal. | |
420 | ||
421 | @item ESPIPE | |
422 | The file descriptor @var{filedes} is associate with a pipe or a FIFO and | |
423 | this device does not allow positioning of the file pointer. | |
424 | @end table | |
425 | ||
426 | The function is an extension defined in the Unix Single Specification | |
427 | version 2. | |
428 | @end deftypefun | |
429 | ||
b07d03e0 | 430 | @comment unistd.h |
a3a4a74e | 431 | @comment Unix98 |
b07d03e0 UD |
432 | @deftypefun ssize_t pread64 (int @var{filedes}, void *@var{buffer}, size_t @var{size}, off64_t @var{offset}) |
433 | This function is similar to the @code{pread} function. The difference | |
434 | is that the @var{offset} parameter is of type @code{off64_t} instead of | |
04b9968b | 435 | @code{off_t} which makes it possible on 32 bit machines to address |
c756c71c | 436 | files larger than @math{2^31} bytes and up to @math{2^63} bytes. The |
b07d03e0 UD |
437 | file descriptor @code{filedes} must be opened using @code{open64} since |
438 | otherwise the large offsets possible with @code{off64_t} will lead to | |
439 | errors with a descriptor in small file mode. | |
440 | ||
c756c71c | 441 | When the source file is compiled with @code{_FILE_OFFSET_BITS == 64} on a |
04b9968b UD |
442 | 32 bit machine this function is actually available under the name |
443 | @code{pread} and so transparently replaces the 32 bit interface. | |
b07d03e0 UD |
444 | @end deftypefun |
445 | ||
28f540f4 RM |
446 | @cindex writing to a file descriptor |
447 | @comment unistd.h | |
448 | @comment POSIX.1 | |
449 | @deftypefun ssize_t write (int @var{filedes}, const void *@var{buffer}, size_t @var{size}) | |
450 | The @code{write} function writes up to @var{size} bytes from | |
451 | @var{buffer} to the file with descriptor @var{filedes}. The data in | |
452 | @var{buffer} is not necessarily a character string and a null character is | |
453 | output like any other character. | |
454 | ||
455 | The return value is the number of bytes actually written. This may be | |
456 | @var{size}, but can always be smaller. Your program should always call | |
457 | @code{write} in a loop, iterating until all the data is written. | |
458 | ||
459 | Once @code{write} returns, the data is enqueued to be written and can be | |
460 | read back right away, but it is not necessarily written out to permanent | |
461 | storage immediately. You can use @code{fsync} when you need to be sure | |
462 | your data has been permanently stored before continuing. (It is more | |
463 | efficient for the system to batch up consecutive writes and do them all | |
464 | at once when convenient. Normally they will always be written to disk | |
a5a0310d UD |
465 | within a minute or less.) Modern systems provide another function |
466 | @code{fdatasync} which guarantees integrity only for the file data and | |
467 | is therefore faster. | |
468 | @c !!! xref fsync, fdatasync | |
2c6fe0bd | 469 | You can use the @code{O_FSYNC} open mode to make @code{write} always |
28f540f4 RM |
470 | store the data to disk before returning; @pxref{Operating Modes}. |
471 | ||
07435eb4 | 472 | In the case of an error, @code{write} returns @math{-1}. The following |
28f540f4 RM |
473 | @code{errno} error conditions are defined for this function: |
474 | ||
475 | @table @code | |
476 | @item EAGAIN | |
477 | Normally, @code{write} blocks until the write operation is complete. | |
478 | But if the @code{O_NONBLOCK} flag is set for the file (@pxref{Control | |
04b9968b | 479 | Operations}), it returns immediately without writing any data and |
28f540f4 RM |
480 | reports this error. An example of a situation that might cause the |
481 | process to block on output is writing to a terminal device that supports | |
482 | flow control, where output has been suspended by receipt of a STOP | |
483 | character. | |
484 | ||
485 | @strong{Compatibility Note:} Most versions of BSD Unix use a different | |
486 | error code for this: @code{EWOULDBLOCK}. In the GNU library, | |
487 | @code{EWOULDBLOCK} is an alias for @code{EAGAIN}, so it doesn't matter | |
488 | which name you use. | |
489 | ||
490 | On some systems, writing a large amount of data from a character special | |
491 | file can also fail with @code{EAGAIN} if the kernel cannot find enough | |
492 | physical memory to lock down the user's pages. This is limited to | |
493 | devices that transfer with direct memory access into the user's memory, | |
494 | which means it does not include terminals, since they always use | |
495 | separate buffers inside the kernel. This problem does not arise in the | |
496 | GNU system. | |
497 | ||
498 | @item EBADF | |
499 | The @var{filedes} argument is not a valid file descriptor, | |
500 | or is not open for writing. | |
501 | ||
502 | @item EFBIG | |
503 | The size of the file would become larger than the implementation can support. | |
504 | ||
505 | @item EINTR | |
506 | The @code{write} operation was interrupted by a signal while it was | |
04b9968b | 507 | blocked waiting for completion. A signal will not necessarily cause |
28f540f4 RM |
508 | @code{write} to return @code{EINTR}; it may instead result in a |
509 | successful @code{write} which writes fewer bytes than requested. | |
510 | @xref{Interrupted Primitives}. | |
511 | ||
512 | @item EIO | |
513 | For many devices, and for disk files, this error code indicates | |
514 | a hardware error. | |
515 | ||
516 | @item ENOSPC | |
517 | The device containing the file is full. | |
518 | ||
519 | @item EPIPE | |
520 | This error is returned when you try to write to a pipe or FIFO that | |
521 | isn't open for reading by any process. When this happens, a @code{SIGPIPE} | |
522 | signal is also sent to the process; see @ref{Signal Handling}. | |
7e583a52 RM |
523 | |
524 | @item EINVAL | |
525 | In some systems, when writing to a character or block device, position | |
526 | and size offsets must be aligned to a particular block size. This error | |
527 | indicates that the offsets were not properly aligned. | |
28f540f4 RM |
528 | @end table |
529 | ||
530 | Unless you have arranged to prevent @code{EINTR} failures, you should | |
531 | check @code{errno} after each failing call to @code{write}, and if the | |
532 | error was @code{EINTR}, you should simply repeat the call. | |
533 | @xref{Interrupted Primitives}. The easy way to do this is with the | |
534 | macro @code{TEMP_FAILURE_RETRY}, as follows: | |
535 | ||
536 | @smallexample | |
537 | nbytes = TEMP_FAILURE_RETRY (write (desc, buffer, count)); | |
538 | @end smallexample | |
539 | ||
b07d03e0 UD |
540 | Please note that there is no function named @code{write64}. This is not |
541 | necessary since this function does not directly modify or handle the | |
542 | possibly wide file offset. Since the kernel handles this state | |
543 | internally the @code{write} function can be used for all cases. | |
544 | ||
04b9968b | 545 | This function is a cancellation point in multi-threaded programs. This |
dfd2257a UD |
546 | is a problem if the thread allocates some resources (like memory, file |
547 | descriptors, semaphores or whatever) at the time @code{write} is | |
19e4c7dd | 548 | called. If the thread gets canceled these resources stay allocated |
04b9968b UD |
549 | until the program ends. To avoid this, calls to @code{write} should be |
550 | protected using cancellation handlers. | |
dfd2257a UD |
551 | @c ref pthread_cleanup_push / pthread_cleanup_pop |
552 | ||
28f540f4 RM |
553 | The @code{write} function is the underlying primitive for all of the |
554 | functions that write to streams, such as @code{fputc}. | |
555 | @end deftypefun | |
556 | ||
a5a0310d UD |
557 | @comment unistd.h |
558 | @comment Unix98 | |
559 | @deftypefun ssize_t pwrite (int @var{filedes}, const void *@var{buffer}, size_t @var{size}, off_t @var{offset}) | |
560 | The @code{pwrite} function is similar to the @code{write} function. The | |
04b9968b UD |
561 | first three arguments are identical, and the return values and error codes |
562 | also correspond. | |
a5a0310d UD |
563 | |
564 | The difference is the fourth argument and its handling. The data block | |
565 | is not written to the current position of the file descriptor | |
566 | @code{filedes}. Instead the data is written to the file starting at | |
567 | position @var{offset}. The position of the file descriptor itself is | |
04b9968b | 568 | not affected by the operation. The value is the same as before the call. |
a5a0310d | 569 | |
b07d03e0 UD |
570 | When the source file is compiled with @code{_FILE_OFFSET_BITS == 64} the |
571 | @code{pwrite} function is in fact @code{pwrite64} and the type | |
04b9968b | 572 | @code{off_t} has 64 bits, which makes it possible to handle files up to |
c756c71c | 573 | @math{2^63} bytes in length. |
b07d03e0 | 574 | |
a5a0310d UD |
575 | The return value of @code{pwrite} describes the number of written bytes. |
576 | In the error case it returns @math{-1} like @code{write} does and the | |
04b9968b UD |
577 | error codes are also the same, with these additions: |
578 | ||
a5a0310d UD |
579 | @table @code |
580 | @item EINVAL | |
581 | The value given for @var{offset} is negative and therefore illegal. | |
582 | ||
583 | @item ESPIPE | |
04b9968b | 584 | The file descriptor @var{filedes} is associated with a pipe or a FIFO and |
a5a0310d UD |
585 | this device does not allow positioning of the file pointer. |
586 | @end table | |
587 | ||
588 | The function is an extension defined in the Unix Single Specification | |
589 | version 2. | |
590 | @end deftypefun | |
591 | ||
b07d03e0 | 592 | @comment unistd.h |
a3a4a74e | 593 | @comment Unix98 |
b07d03e0 UD |
594 | @deftypefun ssize_t pwrite64 (int @var{filedes}, const void *@var{buffer}, size_t @var{size}, off64_t @var{offset}) |
595 | This function is similar to the @code{pwrite} function. The difference | |
596 | is that the @var{offset} parameter is of type @code{off64_t} instead of | |
04b9968b | 597 | @code{off_t} which makes it possible on 32 bit machines to address |
c756c71c | 598 | files larger than @math{2^31} bytes and up to @math{2^63} bytes. The |
b07d03e0 UD |
599 | file descriptor @code{filedes} must be opened using @code{open64} since |
600 | otherwise the large offsets possible with @code{off64_t} will lead to | |
601 | errors with a descriptor in small file mode. | |
602 | ||
c756c71c | 603 | When the source file is compiled using @code{_FILE_OFFSET_BITS == 64} on a |
04b9968b UD |
604 | 32 bit machine this function is actually available under the name |
605 | @code{pwrite} and so transparently replaces the 32 bit interface. | |
b07d03e0 UD |
606 | @end deftypefun |
607 | ||
a5a0310d | 608 | |
28f540f4 RM |
609 | @node File Position Primitive |
610 | @section Setting the File Position of a Descriptor | |
611 | ||
612 | Just as you can set the file position of a stream with @code{fseek}, you | |
613 | can set the file position of a descriptor with @code{lseek}. This | |
614 | specifies the position in the file for the next @code{read} or | |
615 | @code{write} operation. @xref{File Positioning}, for more information | |
616 | on the file position and what it means. | |
617 | ||
618 | To read the current file position value from a descriptor, use | |
619 | @code{lseek (@var{desc}, 0, SEEK_CUR)}. | |
620 | ||
621 | @cindex file positioning on a file descriptor | |
622 | @cindex positioning a file descriptor | |
623 | @cindex seeking on a file descriptor | |
624 | @comment unistd.h | |
625 | @comment POSIX.1 | |
626 | @deftypefun off_t lseek (int @var{filedes}, off_t @var{offset}, int @var{whence}) | |
627 | The @code{lseek} function is used to change the file position of the | |
628 | file with descriptor @var{filedes}. | |
629 | ||
630 | The @var{whence} argument specifies how the @var{offset} should be | |
04b9968b UD |
631 | interpreted, in the same way as for the @code{fseek} function, and it must |
632 | be one of the symbolic constants @code{SEEK_SET}, @code{SEEK_CUR}, or | |
28f540f4 RM |
633 | @code{SEEK_END}. |
634 | ||
635 | @table @code | |
636 | @item SEEK_SET | |
637 | Specifies that @var{whence} is a count of characters from the beginning | |
638 | of the file. | |
639 | ||
640 | @item SEEK_CUR | |
641 | Specifies that @var{whence} is a count of characters from the current | |
642 | file position. This count may be positive or negative. | |
643 | ||
644 | @item SEEK_END | |
645 | Specifies that @var{whence} is a count of characters from the end of | |
646 | the file. A negative count specifies a position within the current | |
647 | extent of the file; a positive count specifies a position past the | |
2c6fe0bd | 648 | current end. If you set the position past the current end, and |
28f540f4 | 649 | actually write data, you will extend the file with zeros up to that |
336dfb2d UD |
650 | position. |
651 | @end table | |
28f540f4 RM |
652 | |
653 | The return value from @code{lseek} is normally the resulting file | |
654 | position, measured in bytes from the beginning of the file. | |
655 | You can use this feature together with @code{SEEK_CUR} to read the | |
656 | current file position. | |
657 | ||
658 | If you want to append to the file, setting the file position to the | |
659 | current end of file with @code{SEEK_END} is not sufficient. Another | |
660 | process may write more data after you seek but before you write, | |
661 | extending the file so the position you write onto clobbers their data. | |
662 | Instead, use the @code{O_APPEND} operating mode; @pxref{Operating Modes}. | |
663 | ||
664 | You can set the file position past the current end of the file. This | |
665 | does not by itself make the file longer; @code{lseek} never changes the | |
666 | file. But subsequent output at that position will extend the file. | |
667 | Characters between the previous end of file and the new position are | |
668 | filled with zeros. Extending the file in this way can create a | |
669 | ``hole'': the blocks of zeros are not actually allocated on disk, so the | |
78759725 | 670 | file takes up less space than it appears to; it is then called a |
28f540f4 RM |
671 | ``sparse file''. |
672 | @cindex sparse files | |
673 | @cindex holes in files | |
674 | ||
675 | If the file position cannot be changed, or the operation is in some way | |
07435eb4 | 676 | invalid, @code{lseek} returns a value of @math{-1}. The following |
28f540f4 RM |
677 | @code{errno} error conditions are defined for this function: |
678 | ||
679 | @table @code | |
680 | @item EBADF | |
681 | The @var{filedes} is not a valid file descriptor. | |
682 | ||
683 | @item EINVAL | |
684 | The @var{whence} argument value is not valid, or the resulting | |
685 | file offset is not valid. A file offset is invalid. | |
686 | ||
687 | @item ESPIPE | |
688 | The @var{filedes} corresponds to an object that cannot be positioned, | |
689 | such as a pipe, FIFO or terminal device. (POSIX.1 specifies this error | |
690 | only for pipes and FIFOs, but in the GNU system, you always get | |
691 | @code{ESPIPE} if the object is not seekable.) | |
692 | @end table | |
693 | ||
b07d03e0 UD |
694 | When the source file is compiled with @code{_FILE_OFFSET_BITS == 64} the |
695 | @code{lseek} function is in fact @code{lseek64} and the type | |
696 | @code{off_t} has 64 bits which makes it possible to handle files up to | |
c756c71c | 697 | @math{2^63} bytes in length. |
b07d03e0 | 698 | |
04b9968b | 699 | This function is a cancellation point in multi-threaded programs. This |
dfd2257a UD |
700 | is a problem if the thread allocates some resources (like memory, file |
701 | descriptors, semaphores or whatever) at the time @code{lseek} is | |
19e4c7dd | 702 | called. If the thread gets canceled these resources stay allocated |
dfd2257a | 703 | until the program ends. To avoid this calls to @code{lseek} should be |
04b9968b | 704 | protected using cancellation handlers. |
dfd2257a UD |
705 | @c ref pthread_cleanup_push / pthread_cleanup_pop |
706 | ||
28f540f4 | 707 | The @code{lseek} function is the underlying primitive for the |
dfd2257a UD |
708 | @code{fseek}, @code{fseeko}, @code{ftell}, @code{ftello} and |
709 | @code{rewind} functions, which operate on streams instead of file | |
710 | descriptors. | |
28f540f4 RM |
711 | @end deftypefun |
712 | ||
b07d03e0 | 713 | @comment unistd.h |
a3a4a74e | 714 | @comment Unix98 |
b07d03e0 UD |
715 | @deftypefun off64_t lseek64 (int @var{filedes}, off64_t @var{offset}, int @var{whence}) |
716 | This function is similar to the @code{lseek} function. The difference | |
717 | is that the @var{offset} parameter is of type @code{off64_t} instead of | |
04b9968b | 718 | @code{off_t} which makes it possible on 32 bit machines to address |
c756c71c | 719 | files larger than @math{2^31} bytes and up to @math{2^63} bytes. The |
b07d03e0 UD |
720 | file descriptor @code{filedes} must be opened using @code{open64} since |
721 | otherwise the large offsets possible with @code{off64_t} will lead to | |
722 | errors with a descriptor in small file mode. | |
723 | ||
c756c71c | 724 | When the source file is compiled with @code{_FILE_OFFSET_BITS == 64} on a |
b07d03e0 | 725 | 32 bits machine this function is actually available under the name |
04b9968b | 726 | @code{lseek} and so transparently replaces the 32 bit interface. |
b07d03e0 UD |
727 | @end deftypefun |
728 | ||
28f540f4 | 729 | You can have multiple descriptors for the same file if you open the file |
2c6fe0bd | 730 | more than once, or if you duplicate a descriptor with @code{dup}. |
28f540f4 RM |
731 | Descriptors that come from separate calls to @code{open} have independent |
732 | file positions; using @code{lseek} on one descriptor has no effect on the | |
2c6fe0bd | 733 | other. For example, |
28f540f4 RM |
734 | |
735 | @smallexample | |
736 | @group | |
737 | @{ | |
738 | int d1, d2; | |
739 | char buf[4]; | |
740 | d1 = open ("foo", O_RDONLY); | |
741 | d2 = open ("foo", O_RDONLY); | |
742 | lseek (d1, 1024, SEEK_SET); | |
743 | read (d2, buf, 4); | |
744 | @} | |
745 | @end group | |
746 | @end smallexample | |
747 | ||
748 | @noindent | |
749 | will read the first four characters of the file @file{foo}. (The | |
750 | error-checking code necessary for a real program has been omitted here | |
751 | for brevity.) | |
752 | ||
753 | By contrast, descriptors made by duplication share a common file | |
754 | position with the original descriptor that was duplicated. Anything | |
755 | which alters the file position of one of the duplicates, including | |
756 | reading or writing data, affects all of them alike. Thus, for example, | |
757 | ||
758 | @smallexample | |
759 | @{ | |
760 | int d1, d2, d3; | |
761 | char buf1[4], buf2[4]; | |
762 | d1 = open ("foo", O_RDONLY); | |
763 | d2 = dup (d1); | |
764 | d3 = dup (d2); | |
765 | lseek (d3, 1024, SEEK_SET); | |
766 | read (d1, buf1, 4); | |
767 | read (d2, buf2, 4); | |
768 | @} | |
769 | @end smallexample | |
770 | ||
771 | @noindent | |
772 | will read four characters starting with the 1024'th character of | |
773 | @file{foo}, and then four more characters starting with the 1028'th | |
774 | character. | |
775 | ||
776 | @comment sys/types.h | |
777 | @comment POSIX.1 | |
778 | @deftp {Data Type} off_t | |
779 | This is an arithmetic data type used to represent file sizes. | |
780 | In the GNU system, this is equivalent to @code{fpos_t} or @code{long int}. | |
a3a4a74e UD |
781 | |
782 | If the source is compiled with @code{_FILE_OFFSET_BITS == 64} this type | |
783 | is transparently replaced by @code{off64_t}. | |
28f540f4 RM |
784 | @end deftp |
785 | ||
b07d03e0 | 786 | @comment sys/types.h |
a3a4a74e | 787 | @comment Unix98 |
b07d03e0 UD |
788 | @deftp {Data Type} off64_t |
789 | This type is used similar to @code{off_t}. The difference is that even | |
04b9968b | 790 | on 32 bit machines, where the @code{off_t} type would have 32 bits, |
b07d03e0 UD |
791 | @code{off64_t} has 64 bits and so is able to address files up to |
792 | @math{2^63} bytes in length. | |
a3a4a74e UD |
793 | |
794 | When compiling with @code{_FILE_OFFSET_BITS == 64} this type is | |
795 | available under the name @code{off_t}. | |
b07d03e0 UD |
796 | @end deftp |
797 | ||
28f540f4 RM |
798 | These aliases for the @samp{SEEK_@dots{}} constants exist for the sake |
799 | of compatibility with older BSD systems. They are defined in two | |
800 | different header files: @file{fcntl.h} and @file{sys/file.h}. | |
801 | ||
802 | @table @code | |
803 | @item L_SET | |
804 | An alias for @code{SEEK_SET}. | |
805 | ||
806 | @item L_INCR | |
807 | An alias for @code{SEEK_CUR}. | |
808 | ||
809 | @item L_XTND | |
810 | An alias for @code{SEEK_END}. | |
811 | @end table | |
812 | ||
813 | @node Descriptors and Streams | |
814 | @section Descriptors and Streams | |
815 | @cindex streams, and file descriptors | |
816 | @cindex converting file descriptor to stream | |
817 | @cindex extracting file descriptor from stream | |
818 | ||
819 | Given an open file descriptor, you can create a stream for it with the | |
820 | @code{fdopen} function. You can get the underlying file descriptor for | |
821 | an existing stream with the @code{fileno} function. These functions are | |
822 | declared in the header file @file{stdio.h}. | |
823 | @pindex stdio.h | |
824 | ||
825 | @comment stdio.h | |
826 | @comment POSIX.1 | |
827 | @deftypefun {FILE *} fdopen (int @var{filedes}, const char *@var{opentype}) | |
828 | The @code{fdopen} function returns a new stream for the file descriptor | |
829 | @var{filedes}. | |
830 | ||
831 | The @var{opentype} argument is interpreted in the same way as for the | |
832 | @code{fopen} function (@pxref{Opening Streams}), except that | |
833 | the @samp{b} option is not permitted; this is because GNU makes no | |
834 | distinction between text and binary files. Also, @code{"w"} and | |
04b9968b | 835 | @code{"w+"} do not cause truncation of the file; these have an effect only |
28f540f4 RM |
836 | when opening a file, and in this case the file has already been opened. |
837 | You must make sure that the @var{opentype} argument matches the actual | |
838 | mode of the open file descriptor. | |
839 | ||
840 | The return value is the new stream. If the stream cannot be created | |
841 | (for example, if the modes for the file indicated by the file descriptor | |
842 | do not permit the access specified by the @var{opentype} argument), a | |
843 | null pointer is returned instead. | |
844 | ||
845 | In some other systems, @code{fdopen} may fail to detect that the modes | |
846 | for file descriptor do not permit the access specified by | |
847 | @code{opentype}. The GNU C library always checks for this. | |
848 | @end deftypefun | |
849 | ||
850 | For an example showing the use of the @code{fdopen} function, | |
851 | see @ref{Creating a Pipe}. | |
852 | ||
853 | @comment stdio.h | |
854 | @comment POSIX.1 | |
855 | @deftypefun int fileno (FILE *@var{stream}) | |
856 | This function returns the file descriptor associated with the stream | |
857 | @var{stream}. If an error is detected (for example, if the @var{stream} | |
858 | is not valid) or if @var{stream} does not do I/O to a file, | |
07435eb4 | 859 | @code{fileno} returns @math{-1}. |
28f540f4 RM |
860 | @end deftypefun |
861 | ||
7b4161bb UD |
862 | @comment stdio.h |
863 | @comment GNU | |
864 | @deftypefun int fileno_unlocked (FILE *@var{stream}) | |
865 | The @code{fileno_unlocked} function is equivalent to the @code{fileno} | |
866 | function except that it does not implicitly lock the stream if the state | |
867 | is @code{FSETLOCKING_INTERNAL}. | |
868 | ||
869 | This function is a GNU extension. | |
870 | @end deftypefun | |
871 | ||
28f540f4 RM |
872 | @cindex standard file descriptors |
873 | @cindex file descriptors, standard | |
874 | There are also symbolic constants defined in @file{unistd.h} for the | |
875 | file descriptors belonging to the standard streams @code{stdin}, | |
876 | @code{stdout}, and @code{stderr}; see @ref{Standard Streams}. | |
877 | @pindex unistd.h | |
878 | ||
879 | @comment unistd.h | |
880 | @comment POSIX.1 | |
881 | @table @code | |
882 | @item STDIN_FILENO | |
883 | @vindex STDIN_FILENO | |
884 | This macro has value @code{0}, which is the file descriptor for | |
885 | standard input. | |
886 | @cindex standard input file descriptor | |
887 | ||
888 | @comment unistd.h | |
889 | @comment POSIX.1 | |
890 | @item STDOUT_FILENO | |
891 | @vindex STDOUT_FILENO | |
892 | This macro has value @code{1}, which is the file descriptor for | |
893 | standard output. | |
894 | @cindex standard output file descriptor | |
895 | ||
896 | @comment unistd.h | |
897 | @comment POSIX.1 | |
898 | @item STDERR_FILENO | |
899 | @vindex STDERR_FILENO | |
900 | This macro has value @code{2}, which is the file descriptor for | |
901 | standard error output. | |
902 | @end table | |
903 | @cindex standard error file descriptor | |
904 | ||
905 | @node Stream/Descriptor Precautions | |
906 | @section Dangers of Mixing Streams and Descriptors | |
907 | @cindex channels | |
908 | @cindex streams and descriptors | |
909 | @cindex descriptors and streams | |
910 | @cindex mixing descriptors and streams | |
911 | ||
912 | You can have multiple file descriptors and streams (let's call both | |
913 | streams and descriptors ``channels'' for short) connected to the same | |
914 | file, but you must take care to avoid confusion between channels. There | |
915 | are two cases to consider: @dfn{linked} channels that share a single | |
916 | file position value, and @dfn{independent} channels that have their own | |
917 | file positions. | |
918 | ||
919 | It's best to use just one channel in your program for actual data | |
920 | transfer to any given file, except when all the access is for input. | |
921 | For example, if you open a pipe (something you can only do at the file | |
922 | descriptor level), either do all I/O with the descriptor, or construct a | |
923 | stream from the descriptor with @code{fdopen} and then do all I/O with | |
924 | the stream. | |
925 | ||
926 | @menu | |
927 | * Linked Channels:: Dealing with channels sharing a file position. | |
928 | * Independent Channels:: Dealing with separately opened, unlinked channels. | |
2c6fe0bd | 929 | * Cleaning Streams:: Cleaning a stream makes it safe to use |
28f540f4 RM |
930 | another channel. |
931 | @end menu | |
932 | ||
933 | @node Linked Channels | |
934 | @subsection Linked Channels | |
935 | @cindex linked channels | |
936 | ||
937 | Channels that come from a single opening share the same file position; | |
938 | we call them @dfn{linked} channels. Linked channels result when you | |
939 | make a stream from a descriptor using @code{fdopen}, when you get a | |
940 | descriptor from a stream with @code{fileno}, when you copy a descriptor | |
941 | with @code{dup} or @code{dup2}, and when descriptors are inherited | |
942 | during @code{fork}. For files that don't support random access, such as | |
943 | terminals and pipes, @emph{all} channels are effectively linked. On | |
944 | random-access files, all append-type output streams are effectively | |
945 | linked to each other. | |
946 | ||
947 | @cindex cleaning up a stream | |
0295d266 UD |
948 | If you have been using a stream for I/O (or have just opened the stream), |
949 | and you want to do I/O using | |
28f540f4 RM |
950 | another channel (either a stream or a descriptor) that is linked to it, |
951 | you must first @dfn{clean up} the stream that you have been using. | |
952 | @xref{Cleaning Streams}. | |
953 | ||
954 | Terminating a process, or executing a new program in the process, | |
955 | destroys all the streams in the process. If descriptors linked to these | |
956 | streams persist in other processes, their file positions become | |
957 | undefined as a result. To prevent this, you must clean up the streams | |
958 | before destroying them. | |
959 | ||
960 | @node Independent Channels | |
961 | @subsection Independent Channels | |
962 | @cindex independent channels | |
963 | ||
964 | When you open channels (streams or descriptors) separately on a seekable | |
965 | file, each channel has its own file position. These are called | |
966 | @dfn{independent channels}. | |
967 | ||
968 | The system handles each channel independently. Most of the time, this | |
969 | is quite predictable and natural (especially for input): each channel | |
970 | can read or write sequentially at its own place in the file. However, | |
971 | if some of the channels are streams, you must take these precautions: | |
972 | ||
973 | @itemize @bullet | |
974 | @item | |
975 | You should clean an output stream after use, before doing anything else | |
976 | that might read or write from the same part of the file. | |
977 | ||
978 | @item | |
979 | You should clean an input stream before reading data that may have been | |
980 | modified using an independent channel. Otherwise, you might read | |
981 | obsolete data that had been in the stream's buffer. | |
982 | @end itemize | |
983 | ||
984 | If you do output to one channel at the end of the file, this will | |
985 | certainly leave the other independent channels positioned somewhere | |
986 | before the new end. You cannot reliably set their file positions to the | |
987 | new end of file before writing, because the file can always be extended | |
988 | by another process between when you set the file position and when you | |
989 | write the data. Instead, use an append-type descriptor or stream; they | |
990 | always output at the current end of the file. In order to make the | |
991 | end-of-file position accurate, you must clean the output channel you | |
992 | were using, if it is a stream. | |
993 | ||
994 | It's impossible for two channels to have separate file pointers for a | |
995 | file that doesn't support random access. Thus, channels for reading or | |
996 | writing such files are always linked, never independent. Append-type | |
997 | channels are also always linked. For these channels, follow the rules | |
998 | for linked channels; see @ref{Linked Channels}. | |
999 | ||
1000 | @node Cleaning Streams | |
1001 | @subsection Cleaning Streams | |
1002 | ||
1003 | On the GNU system, you can clean up any stream with @code{fclean}: | |
1004 | ||
1005 | @comment stdio.h | |
1006 | @comment GNU | |
1007 | @deftypefun int fclean (FILE *@var{stream}) | |
1008 | Clean up the stream @var{stream} so that its buffer is empty. If | |
1009 | @var{stream} is doing output, force it out. If @var{stream} is doing | |
1010 | input, give the data in the buffer back to the system, arranging to | |
1011 | reread it. | |
1012 | @end deftypefun | |
1013 | ||
1014 | On other systems, you can use @code{fflush} to clean a stream in most | |
1015 | cases. | |
1016 | ||
1017 | You can skip the @code{fclean} or @code{fflush} if you know the stream | |
1018 | is already clean. A stream is clean whenever its buffer is empty. For | |
1019 | example, an unbuffered stream is always clean. An input stream that is | |
1020 | at end-of-file is clean. A line-buffered stream is clean when the last | |
0295d266 UD |
1021 | character output was a newline. However, a just-opened input stream |
1022 | might not be clean, as its input buffer might not be empty. | |
28f540f4 RM |
1023 | |
1024 | There is one case in which cleaning a stream is impossible on most | |
1025 | systems. This is when the stream is doing input from a file that is not | |
1026 | random-access. Such streams typically read ahead, and when the file is | |
1027 | not random access, there is no way to give back the excess data already | |
1028 | read. When an input stream reads from a random-access file, | |
1029 | @code{fflush} does clean the stream, but leaves the file pointer at an | |
1030 | unpredictable place; you must set the file pointer before doing any | |
1031 | further I/O. On the GNU system, using @code{fclean} avoids both of | |
1032 | these problems. | |
1033 | ||
1034 | Closing an output-only stream also does @code{fflush}, so this is a | |
1035 | valid way of cleaning an output stream. On the GNU system, closing an | |
1036 | input stream does @code{fclean}. | |
1037 | ||
1038 | You need not clean a stream before using its descriptor for control | |
1039 | operations such as setting terminal modes; these operations don't affect | |
1040 | the file position and are not affected by it. You can use any | |
1041 | descriptor for these operations, and all channels are affected | |
1042 | simultaneously. However, text already ``output'' to a stream but still | |
1043 | buffered by the stream will be subject to the new terminal modes when | |
1044 | subsequently flushed. To make sure ``past'' output is covered by the | |
1045 | terminal settings that were in effect at the time, flush the output | |
1046 | streams for that terminal before setting the modes. @xref{Terminal | |
1047 | Modes}. | |
1048 | ||
07435eb4 UD |
1049 | @node Scatter-Gather |
1050 | @section Fast Scatter-Gather I/O | |
1051 | @cindex scatter-gather | |
1052 | ||
1053 | Some applications may need to read or write data to multiple buffers, | |
04b9968b | 1054 | which are separated in memory. Although this can be done easily enough |
19e4c7dd | 1055 | with multiple calls to @code{read} and @code{write}, it is inefficient |
07435eb4 UD |
1056 | because there is overhead associated with each kernel call. |
1057 | ||
1058 | Instead, many platforms provide special high-speed primitives to perform | |
1059 | these @dfn{scatter-gather} operations in a single kernel call. The GNU C | |
1060 | library will provide an emulation on any system that lacks these | |
1061 | primitives, so they are not a portability threat. They are defined in | |
1062 | @code{sys/uio.h}. | |
1063 | ||
1064 | These functions are controlled with arrays of @code{iovec} structures, | |
1065 | which describe the location and size of each buffer. | |
1066 | ||
4c450556 UD |
1067 | @comment sys/uio.h |
1068 | @comment BSD | |
07435eb4 UD |
1069 | @deftp {Data Type} {struct iovec} |
1070 | ||
1071 | The @code{iovec} structure describes a buffer. It contains two fields: | |
1072 | ||
1073 | @table @code | |
1074 | ||
1075 | @item void *iov_base | |
1076 | Contains the address of a buffer. | |
1077 | ||
1078 | @item size_t iov_len | |
1079 | Contains the length of the buffer. | |
1080 | ||
1081 | @end table | |
1082 | @end deftp | |
1083 | ||
4c450556 UD |
1084 | @comment sys/uio.h |
1085 | @comment BSD | |
07435eb4 UD |
1086 | @deftypefun ssize_t readv (int @var{filedes}, const struct iovec *@var{vector}, int @var{count}) |
1087 | ||
1088 | The @code{readv} function reads data from @var{filedes} and scatters it | |
1089 | into the buffers described in @var{vector}, which is taken to be | |
1090 | @var{count} structures long. As each buffer is filled, data is sent to the | |
1091 | next. | |
1092 | ||
1093 | Note that @code{readv} is not guaranteed to fill all the buffers. | |
1094 | It may stop at any point, for the same reasons @code{read} would. | |
1095 | ||
1096 | The return value is a count of bytes (@emph{not} buffers) read, @math{0} | |
1097 | indicating end-of-file, or @math{-1} indicating an error. The possible | |
1098 | errors are the same as in @code{read}. | |
1099 | ||
1100 | @end deftypefun | |
1101 | ||
4c450556 UD |
1102 | @comment sys/uio.h |
1103 | @comment BSD | |
07435eb4 UD |
1104 | @deftypefun ssize_t writev (int @var{filedes}, const struct iovec *@var{vector}, int @var{count}) |
1105 | ||
1106 | The @code{writev} function gathers data from the buffers described in | |
1107 | @var{vector}, which is taken to be @var{count} structures long, and writes | |
1108 | them to @code{filedes}. As each buffer is written, it moves on to the | |
1109 | next. | |
1110 | ||
1111 | Like @code{readv}, @code{writev} may stop midstream under the same | |
1112 | conditions @code{write} would. | |
1113 | ||
1114 | The return value is a count of bytes written, or @math{-1} indicating an | |
1115 | error. The possible errors are the same as in @code{write}. | |
1116 | ||
1117 | @end deftypefun | |
1118 | ||
1119 | @c Note - I haven't read this anywhere. I surmised it from my knowledge | |
1120 | @c of computer science. Thus, there could be subtleties I'm missing. | |
1121 | ||
1122 | Note that if the buffers are small (under about 1kB), high-level streams | |
1123 | may be easier to use than these functions. However, @code{readv} and | |
1124 | @code{writev} are more efficient when the individual buffers themselves | |
1125 | (as opposed to the total output), are large. In that case, a high-level | |
1126 | stream would not be able to cache the data effectively. | |
1127 | ||
1128 | @node Memory-mapped I/O | |
1129 | @section Memory-mapped I/O | |
1130 | ||
1131 | On modern operating systems, it is possible to @dfn{mmap} (pronounced | |
1132 | ``em-map'') a file to a region of memory. When this is done, the file can | |
1133 | be accessed just like an array in the program. | |
1134 | ||
19e4c7dd | 1135 | This is more efficient than @code{read} or @code{write}, as only the regions |
04b9968b | 1136 | of the file that a program actually accesses are loaded. Accesses to |
07435eb4 UD |
1137 | not-yet-loaded parts of the mmapped region are handled in the same way as |
1138 | swapped out pages. | |
1139 | ||
b642f101 UD |
1140 | Since mmapped pages can be stored back to their file when physical |
1141 | memory is low, it is possible to mmap files orders of magnitude larger | |
1142 | than both the physical memory @emph{and} swap space. The only limit is | |
1143 | address space. The theoretical limit is 4GB on a 32-bit machine - | |
1144 | however, the actual limit will be smaller since some areas will be | |
1145 | reserved for other purposes. If the LFS interface is used the file size | |
1146 | on 32-bit systems is not limited to 2GB (offsets are signed which | |
1147 | reduces the addressable area of 4GB by half); the full 64-bit are | |
1148 | available. | |
07435eb4 UD |
1149 | |
1150 | Memory mapping only works on entire pages of memory. Thus, addresses | |
1151 | for mapping must be page-aligned, and length values will be rounded up. | |
1152 | To determine the size of a page the machine uses one should use | |
1153 | ||
b642f101 | 1154 | @vindex _SC_PAGESIZE |
07435eb4 UD |
1155 | @smallexample |
1156 | size_t page_size = (size_t) sysconf (_SC_PAGESIZE); | |
1157 | @end smallexample | |
1158 | ||
b642f101 | 1159 | @noindent |
07435eb4 UD |
1160 | These functions are declared in @file{sys/mman.h}. |
1161 | ||
4c450556 UD |
1162 | @comment sys/mman.h |
1163 | @comment POSIX | |
07435eb4 UD |
1164 | @deftypefun {void *} mmap (void *@var{address}, size_t @var{length},int @var{protect}, int @var{flags}, int @var{filedes}, off_t @var{offset}) |
1165 | ||
1166 | The @code{mmap} function creates a new mapping, connected to bytes | |
b73147d0 | 1167 | (@var{offset}) to (@var{offset} + @var{length} - 1) in the file open on |
b61345a1 UD |
1168 | @var{filedes}. A new reference for the file specified by @var{filedes} |
1169 | is created, which is not removed by closing the file. | |
07435eb4 UD |
1170 | |
1171 | @var{address} gives a preferred starting address for the mapping. | |
1172 | @code{NULL} expresses no preference. Any previous mapping at that | |
1173 | address is automatically removed. The address you give may still be | |
1174 | changed, unless you use the @code{MAP_FIXED} flag. | |
1175 | ||
1176 | @vindex PROT_READ | |
1177 | @vindex PROT_WRITE | |
1178 | @vindex PROT_EXEC | |
1179 | @var{protect} contains flags that control what kind of access is | |
1180 | permitted. They include @code{PROT_READ}, @code{PROT_WRITE}, and | |
1181 | @code{PROT_EXEC}, which permit reading, writing, and execution, | |
1182 | respectively. Inappropriate access will cause a segfault (@pxref{Program | |
1183 | Error Signals}). | |
1184 | ||
1185 | Note that most hardware designs cannot support write permission without | |
1186 | read permission, and many do not distinguish read and execute permission. | |
49c091e5 | 1187 | Thus, you may receive wider permissions than you ask for, and mappings of |
07435eb4 UD |
1188 | write-only files may be denied even if you do not use @code{PROT_READ}. |
1189 | ||
1190 | @var{flags} contains flags that control the nature of the map. | |
1191 | One of @code{MAP_SHARED} or @code{MAP_PRIVATE} must be specified. | |
1192 | ||
1193 | They include: | |
1194 | ||
1195 | @vtable @code | |
1196 | @item MAP_PRIVATE | |
1197 | This specifies that writes to the region should never be written back | |
1198 | to the attached file. Instead, a copy is made for the process, and the | |
1199 | region will be swapped normally if memory runs low. No other process will | |
1200 | see the changes. | |
1201 | ||
1202 | Since private mappings effectively revert to ordinary memory | |
1203 | when written to, you must have enough virtual memory for a copy of | |
1204 | the entire mmapped region if you use this mode with @code{PROT_WRITE}. | |
1205 | ||
1206 | @item MAP_SHARED | |
1207 | This specifies that writes to the region will be written back to the | |
1208 | file. Changes made will be shared immediately with other processes | |
1209 | mmaping the same file. | |
1210 | ||
1211 | Note that actual writing may take place at any time. You need to use | |
1212 | @code{msync}, described below, if it is important that other processes | |
1213 | using conventional I/O get a consistent view of the file. | |
1214 | ||
1215 | @item MAP_FIXED | |
1216 | This forces the system to use the exact mapping address specified in | |
1217 | @var{address} and fail if it can't. | |
1218 | ||
1219 | @c One of these is official - the other is obviously an obsolete synonym | |
1220 | @c Which is which? | |
1221 | @item MAP_ANONYMOUS | |
1222 | @itemx MAP_ANON | |
1223 | This flag tells the system to create an anonymous mapping, not connected | |
1224 | to a file. @var{filedes} and @var{off} are ignored, and the region is | |
1225 | initialized with zeros. | |
1226 | ||
1227 | Anonymous maps are used as the basic primitive to extend the heap on some | |
1228 | systems. They are also useful to share data between multiple tasks | |
1229 | without creating a file. | |
1230 | ||
49c091e5 | 1231 | On some systems using private anonymous mmaps is more efficient than using |
07435eb4 UD |
1232 | @code{malloc} for large blocks. This is not an issue with the GNU C library, |
1233 | as the included @code{malloc} automatically uses @code{mmap} where appropriate. | |
1234 | ||
1235 | @c Linux has some other MAP_ options, which I have not discussed here. | |
1236 | @c MAP_DENYWRITE, MAP_EXECUTABLE and MAP_GROWSDOWN don't seem applicable to | |
1237 | @c user programs (and I don't understand the last two). MAP_LOCKED does | |
1238 | @c not appear to be implemented. | |
1239 | ||
1240 | @end vtable | |
1241 | ||
1242 | @code{mmap} returns the address of the new mapping, or @math{-1} for an | |
1243 | error. | |
1244 | ||
1245 | Possible errors include: | |
1246 | ||
1247 | @table @code | |
1248 | ||
1249 | @item EINVAL | |
1250 | ||
1251 | Either @var{address} was unusable, or inconsistent @var{flags} were | |
1252 | given. | |
1253 | ||
1254 | @item EACCES | |
1255 | ||
1256 | @var{filedes} was not open for the type of access specified in @var{protect}. | |
1257 | ||
1258 | @item ENOMEM | |
1259 | ||
1260 | Either there is not enough memory for the operation, or the process is | |
1261 | out of address space. | |
1262 | ||
1263 | @item ENODEV | |
1264 | ||
1265 | This file is of a type that doesn't support mapping. | |
1266 | ||
1267 | @item ENOEXEC | |
1268 | ||
1269 | The file is on a filesystem that doesn't support mapping. | |
1270 | ||
1271 | @c On Linux, EAGAIN will appear if the file has a conflicting mandatory lock. | |
1272 | @c However mandatory locks are not discussed in this manual. | |
1273 | @c | |
1274 | @c Similarly, ETXTBSY will occur if the MAP_DENYWRITE flag (not documented | |
1275 | @c here) is used and the file is already open for writing. | |
1276 | ||
1277 | @end table | |
1278 | ||
1279 | @end deftypefun | |
1280 | ||
4c450556 UD |
1281 | @comment sys/mman.h |
1282 | @comment LFS | |
b642f101 UD |
1283 | @deftypefun {void *} mmap64 (void *@var{address}, size_t @var{length},int @var{protect}, int @var{flags}, int @var{filedes}, off64_t @var{offset}) |
1284 | The @code{mmap64} function is equivalent to the @code{mmap} function but | |
1285 | the @var{offset} parameter is of type @code{off64_t}. On 32-bit systems | |
1286 | this allows the file associated with the @var{filedes} descriptor to be | |
1287 | larger than 2GB. @var{filedes} must be a descriptor returned from a | |
1288 | call to @code{open64} or @code{fopen64} and @code{freopen64} where the | |
1289 | descriptor is retrieved with @code{fileno}. | |
1290 | ||
1291 | When the sources are translated with @code{_FILE_OFFSET_BITS == 64} this | |
1292 | function is actually available under the name @code{mmap}. I.e., the | |
1293 | new, extended API using 64 bit file sizes and offsets transparently | |
1294 | replaces the old API. | |
1295 | @end deftypefun | |
1296 | ||
4c450556 UD |
1297 | @comment sys/mman.h |
1298 | @comment POSIX | |
07435eb4 UD |
1299 | @deftypefun int munmap (void *@var{addr}, size_t @var{length}) |
1300 | ||
1301 | @code{munmap} removes any memory maps from (@var{addr}) to (@var{addr} + | |
1302 | @var{length}). @var{length} should be the length of the mapping. | |
1303 | ||
04b9968b | 1304 | It is safe to unmap multiple mappings in one command, or include unmapped |
07435eb4 | 1305 | space in the range. It is also possible to unmap only part of an existing |
04b9968b | 1306 | mapping. However, only entire pages can be removed. If @var{length} is not |
07435eb4 UD |
1307 | an even number of pages, it will be rounded up. |
1308 | ||
1309 | It returns @math{0} for success and @math{-1} for an error. | |
1310 | ||
1311 | One error is possible: | |
1312 | ||
1313 | @table @code | |
1314 | ||
1315 | @item EINVAL | |
04b9968b | 1316 | The memory range given was outside the user mmap range or wasn't page |
07435eb4 UD |
1317 | aligned. |
1318 | ||
1319 | @end table | |
1320 | ||
1321 | @end deftypefun | |
1322 | ||
4c450556 UD |
1323 | @comment sys/mman.h |
1324 | @comment POSIX | |
07435eb4 UD |
1325 | @deftypefun int msync (void *@var{address}, size_t @var{length}, int @var{flags}) |
1326 | ||
1327 | When using shared mappings, the kernel can write the file at any time | |
1328 | before the mapping is removed. To be certain data has actually been | |
49c091e5 UD |
1329 | written to the file and will be accessible to non-memory-mapped I/O, it |
1330 | is necessary to use this function. | |
07435eb4 UD |
1331 | |
1332 | It operates on the region @var{address} to (@var{address} + @var{length}). | |
1333 | It may be used on part of a mapping or multiple mappings, however the | |
1334 | region given should not contain any unmapped space. | |
1335 | ||
1336 | @var{flags} can contain some options: | |
1337 | ||
1338 | @vtable @code | |
1339 | ||
1340 | @item MS_SYNC | |
1341 | ||
1342 | This flag makes sure the data is actually written @emph{to disk}. | |
1343 | Normally @code{msync} only makes sure that accesses to a file with | |
1344 | conventional I/O reflect the recent changes. | |
1345 | ||
1346 | @item MS_ASYNC | |
1347 | ||
1348 | This tells @code{msync} to begin the synchronization, but not to wait for | |
1349 | it to complete. | |
1350 | ||
1351 | @c Linux also has MS_INVALIDATE, which I don't understand. | |
1352 | ||
1353 | @end vtable | |
1354 | ||
1355 | @code{msync} returns @math{0} for success and @math{-1} for | |
1356 | error. Errors include: | |
1357 | ||
1358 | @table @code | |
1359 | ||
1360 | @item EINVAL | |
1361 | An invalid region was given, or the @var{flags} were invalid. | |
1362 | ||
1363 | @item EFAULT | |
1364 | There is no existing mapping in at least part of the given region. | |
1365 | ||
1366 | @end table | |
1367 | ||
1368 | @end deftypefun | |
1369 | ||
4c450556 UD |
1370 | @comment sys/mman.h |
1371 | @comment GNU | |
07435eb4 UD |
1372 | @deftypefun {void *} mremap (void *@var{address}, size_t @var{length}, size_t @var{new_length}, int @var{flag}) |
1373 | ||
1374 | This function can be used to change the size of an existing memory | |
1375 | area. @var{address} and @var{length} must cover a region entirely mapped | |
1376 | in the same @code{mmap} statement. A new mapping with the same | |
04b9968b | 1377 | characteristics will be returned with the length @var{new_length}. |
07435eb4 UD |
1378 | |
1379 | One option is possible, @code{MREMAP_MAYMOVE}. If it is given in | |
1380 | @var{flags}, the system may remove the existing mapping and create a new | |
1381 | one of the desired length in another location. | |
1382 | ||
1383 | The address of the resulting mapping is returned, or @math{-1}. Possible | |
1384 | error codes include: | |
1385 | ||
07435eb4 UD |
1386 | @table @code |
1387 | ||
1388 | @item EFAULT | |
1389 | There is no existing mapping in at least part of the original region, or | |
1390 | the region covers two or more distinct mappings. | |
1391 | ||
1392 | @item EINVAL | |
1393 | The address given is misaligned or inappropriate. | |
1394 | ||
1395 | @item EAGAIN | |
1396 | The region has pages locked, and if extended it would exceed the | |
1397 | process's resource limit for locked pages. @xref{Limits on Resources}. | |
1398 | ||
1399 | @item ENOMEM | |
19e4c7dd | 1400 | The region is private writable, and insufficient virtual memory is |
07435eb4 UD |
1401 | available to extend it. Also, this error will occur if |
1402 | @code{MREMAP_MAYMOVE} is not given and the extension would collide with | |
1403 | another mapped region. | |
1404 | ||
1405 | @end table | |
1406 | @end deftypefun | |
1407 | ||
04b9968b UD |
1408 | This function is only available on a few systems. Except for performing |
1409 | optional optimizations one should not rely on this function. | |
1410 | ||
07435eb4 UD |
1411 | Not all file descriptors may be mapped. Sockets, pipes, and most devices |
1412 | only allow sequential access and do not fit into the mapping abstraction. | |
1413 | In addition, some regular files may not be mmapable, and older kernels may | |
1414 | not support mapping at all. Thus, programs using @code{mmap} should | |
1415 | have a fallback method to use should it fail. @xref{Mmap,,,standards,GNU | |
1416 | Coding Standards}. | |
1417 | ||
0bc93a2f AJ |
1418 | @comment sys/mman.h |
1419 | @comment POSIX | |
1420 | @deftypefun int madvise (void *@var{addr}, size_t @var{length}, int @var{advice}) | |
1421 | ||
1422 | This function can be used to provide the system with @var{advice} about | |
1423 | the intended usage patterns of the memory region starting at @var{addr} | |
1424 | and extending @var{length} bytes. | |
1425 | ||
1426 | The valid BSD values for @var{advice} are: | |
1427 | ||
1428 | @table @code | |
1429 | ||
1430 | @item MADV_NORMAL | |
1431 | The region should receive no further special treatment. | |
1432 | ||
1433 | @item MADV_RANDOM | |
1434 | The region will be accessed via random page references. The kernel | |
1435 | should page-in the minimal number of pages for each page fault. | |
1436 | ||
1437 | @item MADV_SEQUENTIAL | |
1438 | The region will be accessed via sequential page references. This | |
1439 | may cause the kernel to aggressively read-ahead, expecting further | |
1440 | sequential references after any page fault within this region. | |
1441 | ||
1442 | @item MADV_WILLNEED | |
1443 | The region will be needed. The pages within this region may | |
1444 | be pre-faulted in by the kernel. | |
1445 | ||
1446 | @item MADV_DONTNEED | |
1447 | The region is no longer needed. The kernel may free these pages, | |
1448 | causing any changes to the pages to be lost, as well as swapped | |
1449 | out pages to be discarded. | |
1450 | ||
1451 | @end table | |
1452 | ||
1453 | The POSIX names are slightly different, but with the same meanings: | |
1454 | ||
1455 | @table @code | |
1456 | ||
1457 | @item POSIX_MADV_NORMAL | |
1458 | This corresponds with BSD's @code{MADV_NORMAL}. | |
1459 | ||
1460 | @item POSIX_MADV_RANDOM | |
1461 | This corresponds with BSD's @code{MADV_RANDOM}. | |
1462 | ||
1463 | @item POSIX_MADV_SEQUENTIAL | |
1464 | This corresponds with BSD's @code{MADV_SEQUENTIAL}. | |
1465 | ||
1466 | @item POSIX_MADV_WILLNEED | |
1467 | This corresponds with BSD's @code{MADV_WILLNEED}. | |
1468 | ||
1469 | @item POSIX_MADV_DONTNEED | |
1470 | This corresponds with BSD's @code{MADV_DONTNEED}. | |
1471 | ||
1472 | @end table | |
1473 | ||
1474 | @code{msync} returns @math{0} for success and @math{-1} for | |
1475 | error. Errors include: | |
1476 | @table @code | |
1477 | ||
1478 | @item EINVAL | |
1479 | An invalid region was given, or the @var{advice} was invalid. | |
1480 | ||
1481 | @item EFAULT | |
1482 | There is no existing mapping in at least part of the given region. | |
1483 | ||
1484 | @end table | |
1485 | @end deftypefun | |
07435eb4 | 1486 | |
28f540f4 RM |
1487 | @node Waiting for I/O |
1488 | @section Waiting for Input or Output | |
1489 | @cindex waiting for input or output | |
1490 | @cindex multiplexing input | |
1491 | @cindex input from multiple files | |
1492 | ||
1493 | Sometimes a program needs to accept input on multiple input channels | |
1494 | whenever input arrives. For example, some workstations may have devices | |
1495 | such as a digitizing tablet, function button box, or dial box that are | |
1496 | connected via normal asynchronous serial interfaces; good user interface | |
1497 | style requires responding immediately to input on any device. Another | |
1498 | example is a program that acts as a server to several other processes | |
1499 | via pipes or sockets. | |
1500 | ||
1501 | You cannot normally use @code{read} for this purpose, because this | |
1502 | blocks the program until input is available on one particular file | |
1503 | descriptor; input on other channels won't wake it up. You could set | |
1504 | nonblocking mode and poll each file descriptor in turn, but this is very | |
1505 | inefficient. | |
1506 | ||
1507 | A better solution is to use the @code{select} function. This blocks the | |
1508 | program until input or output is ready on a specified set of file | |
1509 | descriptors, or until a timer expires, whichever comes first. This | |
1510 | facility is declared in the header file @file{sys/types.h}. | |
1511 | @pindex sys/types.h | |
1512 | ||
1513 | In the case of a server socket (@pxref{Listening}), we say that | |
1514 | ``input'' is available when there are pending connections that could be | |
1515 | accepted (@pxref{Accepting Connections}). @code{accept} for server | |
1516 | sockets blocks and interacts with @code{select} just as @code{read} does | |
1517 | for normal input. | |
1518 | ||
1519 | @cindex file descriptor sets, for @code{select} | |
1520 | The file descriptor sets for the @code{select} function are specified | |
1521 | as @code{fd_set} objects. Here is the description of the data type | |
1522 | and some macros for manipulating these objects. | |
1523 | ||
1524 | @comment sys/types.h | |
1525 | @comment BSD | |
1526 | @deftp {Data Type} fd_set | |
1527 | The @code{fd_set} data type represents file descriptor sets for the | |
1528 | @code{select} function. It is actually a bit array. | |
1529 | @end deftp | |
1530 | ||
1531 | @comment sys/types.h | |
1532 | @comment BSD | |
1533 | @deftypevr Macro int FD_SETSIZE | |
1534 | The value of this macro is the maximum number of file descriptors that a | |
1535 | @code{fd_set} object can hold information about. On systems with a | |
1536 | fixed maximum number, @code{FD_SETSIZE} is at least that number. On | |
1537 | some systems, including GNU, there is no absolute limit on the number of | |
1538 | descriptors open, but this macro still has a constant value which | |
1539 | controls the number of bits in an @code{fd_set}; if you get a file | |
1540 | descriptor with a value as high as @code{FD_SETSIZE}, you cannot put | |
1541 | that descriptor into an @code{fd_set}. | |
1542 | @end deftypevr | |
1543 | ||
1544 | @comment sys/types.h | |
1545 | @comment BSD | |
1546 | @deftypefn Macro void FD_ZERO (fd_set *@var{set}) | |
1547 | This macro initializes the file descriptor set @var{set} to be the | |
1548 | empty set. | |
1549 | @end deftypefn | |
1550 | ||
1551 | @comment sys/types.h | |
1552 | @comment BSD | |
1553 | @deftypefn Macro void FD_SET (int @var{filedes}, fd_set *@var{set}) | |
1554 | This macro adds @var{filedes} to the file descriptor set @var{set}. | |
d9997a45 UD |
1555 | |
1556 | The @var{filedes} parameter must not have side effects since it is | |
1557 | evaluated more than once. | |
28f540f4 RM |
1558 | @end deftypefn |
1559 | ||
1560 | @comment sys/types.h | |
1561 | @comment BSD | |
1562 | @deftypefn Macro void FD_CLR (int @var{filedes}, fd_set *@var{set}) | |
1563 | This macro removes @var{filedes} from the file descriptor set @var{set}. | |
d9997a45 UD |
1564 | |
1565 | The @var{filedes} parameter must not have side effects since it is | |
1566 | evaluated more than once. | |
28f540f4 RM |
1567 | @end deftypefn |
1568 | ||
1569 | @comment sys/types.h | |
1570 | @comment BSD | |
d9997a45 | 1571 | @deftypefn Macro int FD_ISSET (int @var{filedes}, const fd_set *@var{set}) |
28f540f4 | 1572 | This macro returns a nonzero value (true) if @var{filedes} is a member |
3081378b | 1573 | of the file descriptor set @var{set}, and zero (false) otherwise. |
d9997a45 UD |
1574 | |
1575 | The @var{filedes} parameter must not have side effects since it is | |
1576 | evaluated more than once. | |
28f540f4 RM |
1577 | @end deftypefn |
1578 | ||
1579 | Next, here is the description of the @code{select} function itself. | |
1580 | ||
1581 | @comment sys/types.h | |
1582 | @comment BSD | |
1583 | @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}) | |
1584 | The @code{select} function blocks the calling process until there is | |
1585 | activity on any of the specified sets of file descriptors, or until the | |
1586 | timeout period has expired. | |
1587 | ||
1588 | The file descriptors specified by the @var{read-fds} argument are | |
1589 | checked to see if they are ready for reading; the @var{write-fds} file | |
1590 | descriptors are checked to see if they are ready for writing; and the | |
1591 | @var{except-fds} file descriptors are checked for exceptional | |
1592 | conditions. You can pass a null pointer for any of these arguments if | |
1593 | you are not interested in checking for that kind of condition. | |
1594 | ||
76de2021 UD |
1595 | A file descriptor is considered ready for reading if a @code{read} |
1596 | call will not block. This usually includes the read offset being at | |
1597 | the end of the file or there is an error to report. A server socket | |
1598 | is considered ready for reading if there is a pending connection which | |
1599 | can be accepted with @code{accept}; @pxref{Accepting Connections}. A | |
1600 | client socket is ready for writing when its connection is fully | |
1601 | established; @pxref{Connecting}. | |
28f540f4 RM |
1602 | |
1603 | ``Exceptional conditions'' does not mean errors---errors are reported | |
1604 | immediately when an erroneous system call is executed, and do not | |
1605 | constitute a state of the descriptor. Rather, they include conditions | |
1606 | such as the presence of an urgent message on a socket. (@xref{Sockets}, | |
1607 | for information on urgent messages.) | |
1608 | ||
1609 | The @code{select} function checks only the first @var{nfds} file | |
1610 | descriptors. The usual thing is to pass @code{FD_SETSIZE} as the value | |
1611 | of this argument. | |
1612 | ||
1613 | The @var{timeout} specifies the maximum time to wait. If you pass a | |
1614 | null pointer for this argument, it means to block indefinitely until one | |
1615 | of the file descriptors is ready. Otherwise, you should provide the | |
1616 | time in @code{struct timeval} format; see @ref{High-Resolution | |
1617 | Calendar}. Specify zero as the time (a @code{struct timeval} containing | |
1618 | all zeros) if you want to find out which descriptors are ready without | |
1619 | waiting if none are ready. | |
1620 | ||
1621 | The normal return value from @code{select} is the total number of ready file | |
1622 | descriptors in all of the sets. Each of the argument sets is overwritten | |
1623 | with information about the descriptors that are ready for the corresponding | |
1624 | operation. Thus, to see if a particular descriptor @var{desc} has input, | |
1625 | use @code{FD_ISSET (@var{desc}, @var{read-fds})} after @code{select} returns. | |
1626 | ||
1627 | If @code{select} returns because the timeout period expires, it returns | |
1628 | a value of zero. | |
1629 | ||
1630 | Any signal will cause @code{select} to return immediately. So if your | |
1631 | program uses signals, you can't rely on @code{select} to keep waiting | |
1632 | for the full time specified. If you want to be sure of waiting for a | |
1633 | particular amount of time, you must check for @code{EINTR} and repeat | |
1634 | the @code{select} with a newly calculated timeout based on the current | |
1635 | time. See the example below. See also @ref{Interrupted Primitives}. | |
1636 | ||
1637 | If an error occurs, @code{select} returns @code{-1} and does not modify | |
2c6fe0bd | 1638 | the argument file descriptor sets. The following @code{errno} error |
28f540f4 RM |
1639 | conditions are defined for this function: |
1640 | ||
1641 | @table @code | |
1642 | @item EBADF | |
1643 | One of the file descriptor sets specified an invalid file descriptor. | |
1644 | ||
1645 | @item EINTR | |
1646 | The operation was interrupted by a signal. @xref{Interrupted Primitives}. | |
1647 | ||
1648 | @item EINVAL | |
1649 | The @var{timeout} argument is invalid; one of the components is negative | |
1650 | or too large. | |
1651 | @end table | |
1652 | @end deftypefun | |
1653 | ||
1654 | @strong{Portability Note:} The @code{select} function is a BSD Unix | |
1655 | feature. | |
1656 | ||
1657 | Here is an example showing how you can use @code{select} to establish a | |
1658 | timeout period for reading from a file descriptor. The @code{input_timeout} | |
1659 | function blocks the calling process until input is available on the | |
1660 | file descriptor, or until the timeout period expires. | |
1661 | ||
1662 | @smallexample | |
1663 | @include select.c.texi | |
1664 | @end smallexample | |
1665 | ||
1666 | There is another example showing the use of @code{select} to multiplex | |
1667 | input from multiple sockets in @ref{Server Example}. | |
1668 | ||
1669 | ||
dfd2257a UD |
1670 | @node Synchronizing I/O |
1671 | @section Synchronizing I/O operations | |
1672 | ||
1673 | @cindex synchronizing | |
19e4c7dd | 1674 | In most modern operating systems, the normal I/O operations are not |
dfd2257a | 1675 | executed synchronously. I.e., even if a @code{write} system call |
19e4c7dd | 1676 | returns, this does not mean the data is actually written to the media, |
dfd2257a UD |
1677 | e.g., the disk. |
1678 | ||
19e4c7dd | 1679 | In situations where synchronization points are necessary, you can use |
04b9968b | 1680 | special functions which ensure that all operations finish before |
dfd2257a UD |
1681 | they return. |
1682 | ||
1683 | @comment unistd.h | |
1684 | @comment X/Open | |
1685 | @deftypefun int sync (void) | |
1686 | A call to this function will not return as long as there is data which | |
04b9968b | 1687 | has not been written to the device. All dirty buffers in the kernel will |
dfd2257a UD |
1688 | be written and so an overall consistent system can be achieved (if no |
1689 | other process in parallel writes data). | |
1690 | ||
1691 | A prototype for @code{sync} can be found in @file{unistd.h}. | |
1692 | ||
1693 | The return value is zero to indicate no error. | |
1694 | @end deftypefun | |
1695 | ||
04b9968b UD |
1696 | Programs more often want to ensure that data written to a given file is |
1697 | committed, rather than all data in the system. For this, @code{sync} is overkill. | |
1698 | ||
dfd2257a UD |
1699 | |
1700 | @comment unistd.h | |
1701 | @comment POSIX | |
1702 | @deftypefun int fsync (int @var{fildes}) | |
19e4c7dd AJ |
1703 | The @code{fsync} function can be used to make sure all data associated with |
1704 | the open file @var{fildes} is written to the device associated with the | |
dfd2257a UD |
1705 | descriptor. The function call does not return unless all actions have |
1706 | finished. | |
1707 | ||
1708 | A prototype for @code{fsync} can be found in @file{unistd.h}. | |
1709 | ||
04b9968b | 1710 | This function is a cancellation point in multi-threaded programs. This |
dfd2257a UD |
1711 | is a problem if the thread allocates some resources (like memory, file |
1712 | descriptors, semaphores or whatever) at the time @code{fsync} is | |
19e4c7dd | 1713 | called. If the thread gets canceled these resources stay allocated |
04b9968b UD |
1714 | until the program ends. To avoid this, calls to @code{fsync} should be |
1715 | protected using cancellation handlers. | |
dfd2257a UD |
1716 | @c ref pthread_cleanup_push / pthread_cleanup_pop |
1717 | ||
49c091e5 | 1718 | The return value of the function is zero if no error occurred. Otherwise |
dfd2257a UD |
1719 | it is @math{-1} and the global variable @var{errno} is set to the |
1720 | following values: | |
1721 | @table @code | |
1722 | @item EBADF | |
1723 | The descriptor @var{fildes} is not valid. | |
1724 | ||
1725 | @item EINVAL | |
1726 | No synchronization is possible since the system does not implement this. | |
1727 | @end table | |
1728 | @end deftypefun | |
1729 | ||
1730 | Sometimes it is not even necessary to write all data associated with a | |
1731 | file descriptor. E.g., in database files which do not change in size it | |
1732 | is enough to write all the file content data to the device. | |
19e4c7dd | 1733 | Meta-information, like the modification time etc., are not that important |
dfd2257a UD |
1734 | and leaving such information uncommitted does not prevent a successful |
1735 | recovering of the file in case of a problem. | |
1736 | ||
1737 | @comment unistd.h | |
1738 | @comment POSIX | |
1739 | @deftypefun int fdatasync (int @var{fildes}) | |
04b9968b | 1740 | When a call to the @code{fdatasync} function returns, it is ensured |
dfd2257a | 1741 | that all of the file data is written to the device. For all pending I/O |
04b9968b | 1742 | operations, the parts guaranteeing data integrity finished. |
dfd2257a UD |
1743 | |
1744 | Not all systems implement the @code{fdatasync} operation. On systems | |
1745 | missing this functionality @code{fdatasync} is emulated by a call to | |
1746 | @code{fsync} since the performed actions are a superset of those | |
19e4c7dd | 1747 | required by @code{fdatasync}. |
dfd2257a UD |
1748 | |
1749 | The prototype for @code{fdatasync} is in @file{unistd.h}. | |
1750 | ||
49c091e5 | 1751 | The return value of the function is zero if no error occurred. Otherwise |
dfd2257a UD |
1752 | it is @math{-1} and the global variable @var{errno} is set to the |
1753 | following values: | |
1754 | @table @code | |
1755 | @item EBADF | |
1756 | The descriptor @var{fildes} is not valid. | |
1757 | ||
1758 | @item EINVAL | |
1759 | No synchronization is possible since the system does not implement this. | |
1760 | @end table | |
1761 | @end deftypefun | |
1762 | ||
1763 | ||
b07d03e0 UD |
1764 | @node Asynchronous I/O |
1765 | @section Perform I/O Operations in Parallel | |
1766 | ||
1767 | The POSIX.1b standard defines a new set of I/O operations which can | |
04b9968b | 1768 | significantly reduce the time an application spends waiting at I/O. The |
b07d03e0 | 1769 | new functions allow a program to initiate one or more I/O operations and |
04b9968b UD |
1770 | then immediately resume normal work while the I/O operations are |
1771 | executed in parallel. This functionality is available if the | |
a3a4a74e | 1772 | @file{unistd.h} file defines the symbol @code{_POSIX_ASYNCHRONOUS_IO}. |
b07d03e0 UD |
1773 | |
1774 | These functions are part of the library with realtime functions named | |
1775 | @file{librt}. They are not actually part of the @file{libc} binary. | |
1776 | The implementation of these functions can be done using support in the | |
c756c71c UD |
1777 | kernel (if available) or using an implementation based on threads at |
1778 | userlevel. In the latter case it might be necessary to link applications | |
fed8f7f7 | 1779 | with the thread library @file{libpthread} in addition to @file{librt}. |
b07d03e0 | 1780 | |
c756c71c | 1781 | All AIO operations operate on files which were opened previously. There |
04b9968b | 1782 | might be arbitrarily many operations running for one file. The |
b07d03e0 UD |
1783 | asynchronous I/O operations are controlled using a data structure named |
1784 | @code{struct aiocb} (@dfn{AIO control block}). It is defined in | |
1785 | @file{aio.h} as follows. | |
1786 | ||
1787 | @comment aio.h | |
1788 | @comment POSIX.1b | |
1789 | @deftp {Data Type} {struct aiocb} | |
1790 | The POSIX.1b standard mandates that the @code{struct aiocb} structure | |
1791 | contains at least the members described in the following table. There | |
04b9968b | 1792 | might be more elements which are used by the implementation, but |
19e4c7dd | 1793 | depending upon these elements is not portable and is highly deprecated. |
b07d03e0 UD |
1794 | |
1795 | @table @code | |
1796 | @item int aio_fildes | |
19e4c7dd AJ |
1797 | This element specifies the file descriptor to be used for the |
1798 | operation. It must be a legal descriptor, otherwise the operation will | |
1799 | fail. | |
b07d03e0 UD |
1800 | |
1801 | The device on which the file is opened must allow the seek operation. | |
1802 | I.e., it is not possible to use any of the AIO operations on devices | |
1803 | like terminals where an @code{lseek} call would lead to an error. | |
1804 | ||
1805 | @item off_t aio_offset | |
19e4c7dd | 1806 | This element specifies the offset in the file at which the operation (input |
fed8f7f7 | 1807 | or output) is performed. Since the operations are carried out in arbitrary |
b07d03e0 UD |
1808 | order and more than one operation for one file descriptor can be |
1809 | started, one cannot expect a current read/write position of the file | |
1810 | descriptor. | |
1811 | ||
1812 | @item volatile void *aio_buf | |
1813 | This is a pointer to the buffer with the data to be written or the place | |
c756c71c | 1814 | where the read data is stored. |
b07d03e0 UD |
1815 | |
1816 | @item size_t aio_nbytes | |
1817 | This element specifies the length of the buffer pointed to by @code{aio_buf}. | |
1818 | ||
1819 | @item int aio_reqprio | |
c756c71c | 1820 | If the platform has defined @code{_POSIX_PRIORITIZED_IO} and |
19e4c7dd | 1821 | @code{_POSIX_PRIORITY_SCHEDULING}, the AIO requests are |
b07d03e0 UD |
1822 | processed based on the current scheduling priority. The |
1823 | @code{aio_reqprio} element can then be used to lower the priority of the | |
1824 | AIO operation. | |
1825 | ||
1826 | @item struct sigevent aio_sigevent | |
1827 | This element specifies how the calling process is notified once the | |
fed8f7f7 | 1828 | operation terminates. If the @code{sigev_notify} element is |
19e4c7dd AJ |
1829 | @code{SIGEV_NONE}, no notification is sent. If it is @code{SIGEV_SIGNAL}, |
1830 | the signal determined by @code{sigev_signo} is sent. Otherwise, | |
1831 | @code{sigev_notify} must be @code{SIGEV_THREAD}. In this case, a thread | |
c756c71c | 1832 | is created which starts executing the function pointed to by |
b07d03e0 UD |
1833 | @code{sigev_notify_function}. |
1834 | ||
1835 | @item int aio_lio_opcode | |
1836 | This element is only used by the @code{lio_listio} and | |
04b9968b UD |
1837 | @code{lio_listio64} functions. Since these functions allow an |
1838 | arbitrary number of operations to start at once, and each operation can be | |
1839 | input or output (or nothing), the information must be stored in the | |
b07d03e0 UD |
1840 | control block. The possible values are: |
1841 | ||
1842 | @vtable @code | |
1843 | @item LIO_READ | |
1844 | Start a read operation. Read from the file at position | |
1845 | @code{aio_offset} and store the next @code{aio_nbytes} bytes in the | |
1846 | buffer pointed to by @code{aio_buf}. | |
1847 | ||
1848 | @item LIO_WRITE | |
1849 | Start a write operation. Write @code{aio_nbytes} bytes starting at | |
1850 | @code{aio_buf} into the file starting at position @code{aio_offset}. | |
1851 | ||
1852 | @item LIO_NOP | |
1853 | Do nothing for this control block. This value is useful sometimes when | |
1854 | an array of @code{struct aiocb} values contains holes, i.e., some of the | |
fed8f7f7 | 1855 | values must not be handled although the whole array is presented to the |
b07d03e0 UD |
1856 | @code{lio_listio} function. |
1857 | @end vtable | |
1858 | @end table | |
a3a4a74e | 1859 | |
fed8f7f7 | 1860 | When the sources are compiled using @code{_FILE_OFFSET_BITS == 64} on a |
19e4c7dd | 1861 | 32 bit machine, this type is in fact @code{struct aiocb64}, since the LFS |
a3a4a74e UD |
1862 | interface transparently replaces the @code{struct aiocb} definition. |
1863 | @end deftp | |
1864 | ||
19e4c7dd | 1865 | For use with the AIO functions defined in the LFS, there is a similar type |
a3a4a74e | 1866 | defined which replaces the types of the appropriate members with larger |
04b9968b | 1867 | types but otherwise is equivalent to @code{struct aiocb}. Particularly, |
a3a4a74e UD |
1868 | all member names are the same. |
1869 | ||
1870 | @comment aio.h | |
1871 | @comment POSIX.1b | |
1872 | @deftp {Data Type} {struct aiocb64} | |
1873 | @table @code | |
1874 | @item int aio_fildes | |
1875 | This element specifies the file descriptor which is used for the | |
1876 | operation. It must be a legal descriptor since otherwise the operation | |
1877 | fails for obvious reasons. | |
1878 | ||
1879 | The device on which the file is opened must allow the seek operation. | |
1880 | I.e., it is not possible to use any of the AIO operations on devices | |
1881 | like terminals where an @code{lseek} call would lead to an error. | |
1882 | ||
1883 | @item off64_t aio_offset | |
04b9968b | 1884 | This element specifies at which offset in the file the operation (input |
a3a4a74e UD |
1885 | or output) is performed. Since the operation are carried in arbitrary |
1886 | order and more than one operation for one file descriptor can be | |
1887 | started, one cannot expect a current read/write position of the file | |
1888 | descriptor. | |
1889 | ||
1890 | @item volatile void *aio_buf | |
1891 | This is a pointer to the buffer with the data to be written or the place | |
19e4c7dd | 1892 | where the read data is stored. |
a3a4a74e UD |
1893 | |
1894 | @item size_t aio_nbytes | |
1895 | This element specifies the length of the buffer pointed to by @code{aio_buf}. | |
1896 | ||
1897 | @item int aio_reqprio | |
1898 | If for the platform @code{_POSIX_PRIORITIZED_IO} and | |
04b9968b | 1899 | @code{_POSIX_PRIORITY_SCHEDULING} are defined the AIO requests are |
a3a4a74e UD |
1900 | processed based on the current scheduling priority. The |
1901 | @code{aio_reqprio} element can then be used to lower the priority of the | |
1902 | AIO operation. | |
1903 | ||
1904 | @item struct sigevent aio_sigevent | |
1905 | This element specifies how the calling process is notified once the | |
19e4c7dd AJ |
1906 | operation terminates. If the @code{sigev_notify}, element is |
1907 | @code{SIGEV_NONE} no notification is sent. If it is @code{SIGEV_SIGNAL}, | |
1908 | the signal determined by @code{sigev_signo} is sent. Otherwise, | |
a3a4a74e | 1909 | @code{sigev_notify} must be @code{SIGEV_THREAD} in which case a thread |
04b9968b | 1910 | which starts executing the function pointed to by |
a3a4a74e UD |
1911 | @code{sigev_notify_function}. |
1912 | ||
1913 | @item int aio_lio_opcode | |
1914 | This element is only used by the @code{lio_listio} and | |
04b9968b UD |
1915 | @code{[lio_listio64} functions. Since these functions allow an |
1916 | arbitrary number of operations to start at once, and since each operation can be | |
1917 | input or output (or nothing), the information must be stored in the | |
a3a4a74e UD |
1918 | control block. See the description of @code{struct aiocb} for a description |
1919 | of the possible values. | |
1920 | @end table | |
1921 | ||
1922 | When the sources are compiled using @code{_FILE_OFFSET_BITS == 64} on a | |
19e4c7dd AJ |
1923 | 32 bit machine, this type is available under the name @code{struct |
1924 | aiocb64}, since the LFS transparently replaces the old interface. | |
b07d03e0 UD |
1925 | @end deftp |
1926 | ||
1927 | @menu | |
a3a4a74e UD |
1928 | * Asynchronous Reads/Writes:: Asynchronous Read and Write Operations. |
1929 | * Status of AIO Operations:: Getting the Status of AIO Operations. | |
1930 | * Synchronizing AIO Operations:: Getting into a consistent state. | |
04b9968b | 1931 | * Cancel AIO Operations:: Cancellation of AIO Operations. |
a3a4a74e | 1932 | * Configuration of AIO:: How to optimize the AIO implementation. |
b07d03e0 UD |
1933 | @end menu |
1934 | ||
a3a4a74e UD |
1935 | @node Asynchronous Reads/Writes |
1936 | @subsection Asynchronous Read and Write Operations | |
b07d03e0 UD |
1937 | |
1938 | @comment aio.h | |
1939 | @comment POSIX.1b | |
1940 | @deftypefun int aio_read (struct aiocb *@var{aiocbp}) | |
04b9968b UD |
1941 | This function initiates an asynchronous read operation. It |
1942 | immediately returns after the operation was enqueued or when an | |
fed8f7f7 | 1943 | error was encountered. |
b07d03e0 | 1944 | |
a3a4a74e | 1945 | The first @code{aiocbp->aio_nbytes} bytes of the file for which |
c756c71c UD |
1946 | @code{aiocbp->aio_fildes} is a descriptor are written to the buffer |
1947 | starting at @code{aiocbp->aio_buf}. Reading starts at the absolute | |
1948 | position @code{aiocbp->aio_offset} in the file. | |
b07d03e0 UD |
1949 | |
1950 | If prioritized I/O is supported by the platform the | |
1951 | @code{aiocbp->aio_reqprio} value is used to adjust the priority before | |
1952 | the request is actually enqueued. | |
1953 | ||
1954 | The calling process is notified about the termination of the read | |
1955 | request according to the @code{aiocbp->aio_sigevent} value. | |
1956 | ||
04b9968b | 1957 | When @code{aio_read} returns, the return value is zero if no error |
b07d03e0 | 1958 | occurred that can be found before the process is enqueued. If such an |
04b9968b UD |
1959 | early error is found, the function returns @math{-1} and sets |
1960 | @code{errno} to one of the following values: | |
b07d03e0 UD |
1961 | |
1962 | @table @code | |
1963 | @item EAGAIN | |
1964 | The request was not enqueued due to (temporarily) exceeded resource | |
1965 | limitations. | |
1966 | @item ENOSYS | |
1967 | The @code{aio_read} function is not implemented. | |
1968 | @item EBADF | |
1969 | The @code{aiocbp->aio_fildes} descriptor is not valid. This condition | |
04b9968b | 1970 | need not be recognized before enqueueing the request and so this error |
fed8f7f7 | 1971 | might also be signaled asynchronously. |
b07d03e0 UD |
1972 | @item EINVAL |
1973 | The @code{aiocbp->aio_offset} or @code{aiocbp->aio_reqpiro} value is | |
1974 | invalid. This condition need not be recognized before enqueueing the | |
49c091e5 | 1975 | request and so this error might also be signaled asynchronously. |
b07d03e0 UD |
1976 | @end table |
1977 | ||
04b9968b UD |
1978 | If @code{aio_read} returns zero, the current status of the request |
1979 | can be queried using @code{aio_error} and @code{aio_return} functions. | |
1980 | As long as the value returned by @code{aio_error} is @code{EINPROGRESS} | |
1981 | the operation has not yet completed. If @code{aio_error} returns zero, | |
78759725 UD |
1982 | the operation successfully terminated, otherwise the value is to be |
1983 | interpreted as an error code. If the function terminated, the result of | |
1984 | the operation can be obtained using a call to @code{aio_return}. The | |
1985 | returned value is the same as an equivalent call to @code{read} would | |
04b9968b | 1986 | have returned. Possible error codes returned by @code{aio_error} are: |
b07d03e0 UD |
1987 | |
1988 | @table @code | |
1989 | @item EBADF | |
1990 | The @code{aiocbp->aio_fildes} descriptor is not valid. | |
1991 | @item ECANCELED | |
19e4c7dd | 1992 | The operation was canceled before the operation was finished |
b07d03e0 UD |
1993 | (@pxref{Cancel AIO Operations}) |
1994 | @item EINVAL | |
1995 | The @code{aiocbp->aio_offset} value is invalid. | |
1996 | @end table | |
a3a4a74e UD |
1997 | |
1998 | When the sources are compiled with @code{_FILE_OFFSET_BITS == 64} this | |
1999 | function is in fact @code{aio_read64} since the LFS interface transparently | |
2000 | replaces the normal implementation. | |
b07d03e0 UD |
2001 | @end deftypefun |
2002 | ||
2003 | @comment aio.h | |
a3a4a74e | 2004 | @comment Unix98 |
b07d03e0 UD |
2005 | @deftypefun int aio_read64 (struct aiocb *@var{aiocbp}) |
2006 | This function is similar to the @code{aio_read} function. The only | |
19e4c7dd AJ |
2007 | difference is that on @w{32 bit} machines, the file descriptor should |
2008 | be opened in the large file mode. Internally, @code{aio_read64} uses | |
a3a4a74e UD |
2009 | functionality equivalent to @code{lseek64} (@pxref{File Position |
2010 | Primitive}) to position the file descriptor correctly for the reading, | |
fed8f7f7 | 2011 | as opposed to @code{lseek} functionality used in @code{aio_read}. |
a3a4a74e | 2012 | |
19e4c7dd | 2013 | When the sources are compiled with @code{_FILE_OFFSET_BITS == 64}, this |
a3a4a74e | 2014 | function is available under the name @code{aio_read} and so transparently |
04b9968b | 2015 | replaces the interface for small files on 32 bit machines. |
b07d03e0 UD |
2016 | @end deftypefun |
2017 | ||
19e4c7dd | 2018 | To write data asynchronously to a file, there exists an equivalent pair |
a3a4a74e UD |
2019 | of functions with a very similar interface. |
2020 | ||
2021 | @comment aio.h | |
2022 | @comment POSIX.1b | |
2023 | @deftypefun int aio_write (struct aiocb *@var{aiocbp}) | |
2024 | This function initiates an asynchronous write operation. The function | |
2025 | call immediately returns after the operation was enqueued or if before | |
fed8f7f7 | 2026 | this happens an error was encountered. |
a3a4a74e UD |
2027 | |
2028 | The first @code{aiocbp->aio_nbytes} bytes from the buffer starting at | |
2029 | @code{aiocbp->aio_buf} are written to the file for which | |
2030 | @code{aiocbp->aio_fildes} is an descriptor, starting at the absolute | |
2031 | position @code{aiocbp->aio_offset} in the file. | |
2032 | ||
19e4c7dd | 2033 | If prioritized I/O is supported by the platform, the |
a3a4a74e UD |
2034 | @code{aiocbp->aio_reqprio} value is used to adjust the priority before |
2035 | the request is actually enqueued. | |
2036 | ||
2037 | The calling process is notified about the termination of the read | |
2038 | request according to the @code{aiocbp->aio_sigevent} value. | |
2039 | ||
19e4c7dd | 2040 | When @code{aio_write} returns, the return value is zero if no error |
a3a4a74e UD |
2041 | occurred that can be found before the process is enqueued. If such an |
2042 | early error is found the function returns @math{-1} and sets | |
2043 | @code{errno} to one of the following values. | |
2044 | ||
2045 | @table @code | |
2046 | @item EAGAIN | |
2047 | The request was not enqueued due to (temporarily) exceeded resource | |
2048 | limitations. | |
2049 | @item ENOSYS | |
2050 | The @code{aio_write} function is not implemented. | |
2051 | @item EBADF | |
2052 | The @code{aiocbp->aio_fildes} descriptor is not valid. This condition | |
19e4c7dd | 2053 | may not be recognized before enqueueing the request, and so this error |
fed8f7f7 | 2054 | might also be signaled asynchronously. |
a3a4a74e | 2055 | @item EINVAL |
19e4c7dd AJ |
2056 | The @code{aiocbp->aio_offset} or @code{aiocbp->aio_reqprio} value is |
2057 | invalid. This condition may not be recognized before enqueueing the | |
fed8f7f7 | 2058 | request and so this error might also be signaled asynchronously. |
a3a4a74e UD |
2059 | @end table |
2060 | ||
19e4c7dd | 2061 | In the case @code{aio_write} returns zero, the current status of the |
a3a4a74e | 2062 | request can be queried using @code{aio_error} and @code{aio_return} |
c756c71c | 2063 | functions. As long as the value returned by @code{aio_error} is |
a3a4a74e | 2064 | @code{EINPROGRESS} the operation has not yet completed. If |
19e4c7dd | 2065 | @code{aio_error} returns zero, the operation successfully terminated, |
a3a4a74e | 2066 | otherwise the value is to be interpreted as an error code. If the |
19e4c7dd | 2067 | function terminated, the result of the operation can be get using a call |
a3a4a74e | 2068 | to @code{aio_return}. The returned value is the same as an equivalent |
19e4c7dd | 2069 | call to @code{read} would have returned. Possible error codes returned |
a3a4a74e UD |
2070 | by @code{aio_error} are: |
2071 | ||
2072 | @table @code | |
2073 | @item EBADF | |
2074 | The @code{aiocbp->aio_fildes} descriptor is not valid. | |
2075 | @item ECANCELED | |
19e4c7dd | 2076 | The operation was canceled before the operation was finished. |
a3a4a74e UD |
2077 | (@pxref{Cancel AIO Operations}) |
2078 | @item EINVAL | |
2079 | The @code{aiocbp->aio_offset} value is invalid. | |
2080 | @end table | |
2081 | ||
19e4c7dd | 2082 | When the sources are compiled with @code{_FILE_OFFSET_BITS == 64}, this |
a3a4a74e UD |
2083 | function is in fact @code{aio_write64} since the LFS interface transparently |
2084 | replaces the normal implementation. | |
2085 | @end deftypefun | |
2086 | ||
2087 | @comment aio.h | |
2088 | @comment Unix98 | |
2089 | @deftypefun int aio_write64 (struct aiocb *@var{aiocbp}) | |
2090 | This function is similar to the @code{aio_write} function. The only | |
04b9968b | 2091 | difference is that on @w{32 bit} machines the file descriptor should |
a3a4a74e UD |
2092 | be opened in the large file mode. Internally @code{aio_write64} uses |
2093 | functionality equivalent to @code{lseek64} (@pxref{File Position | |
2094 | Primitive}) to position the file descriptor correctly for the writing, | |
fed8f7f7 | 2095 | as opposed to @code{lseek} functionality used in @code{aio_write}. |
a3a4a74e | 2096 | |
19e4c7dd | 2097 | When the sources are compiled with @code{_FILE_OFFSET_BITS == 64}, this |
a3a4a74e | 2098 | function is available under the name @code{aio_write} and so transparently |
04b9968b | 2099 | replaces the interface for small files on 32 bit machines. |
a3a4a74e UD |
2100 | @end deftypefun |
2101 | ||
19e4c7dd AJ |
2102 | Besides these functions with the more or less traditional interface, |
2103 | POSIX.1b also defines a function which can initiate more than one | |
2104 | operation at a time, and which can handle freely mixed read and write | |
2105 | operations. It is therefore similar to a combination of @code{readv} and | |
a3a4a74e UD |
2106 | @code{writev}. |
2107 | ||
2108 | @comment aio.h | |
2109 | @comment POSIX.1b | |
2110 | @deftypefun int lio_listio (int @var{mode}, struct aiocb *const @var{list}[], int @var{nent}, struct sigevent *@var{sig}) | |
2111 | The @code{lio_listio} function can be used to enqueue an arbitrary | |
2112 | number of read and write requests at one time. The requests can all be | |
2113 | meant for the same file, all for different files or every solution in | |
2114 | between. | |
2115 | ||
2116 | @code{lio_listio} gets the @var{nent} requests from the array pointed to | |
19e4c7dd | 2117 | by @var{list}. The operation to be performed is determined by the |
a3a4a74e | 2118 | @code{aio_lio_opcode} member in each element of @var{list}. If this |
19e4c7dd | 2119 | field is @code{LIO_READ} a read operation is enqueued, similar to a call |
a3a4a74e UD |
2120 | of @code{aio_read} for this element of the array (except that the way |
2121 | the termination is signalled is different, as we will see below). If | |
19e4c7dd | 2122 | the @code{aio_lio_opcode} member is @code{LIO_WRITE} a write operation |
a3a4a74e UD |
2123 | is enqueued. Otherwise the @code{aio_lio_opcode} must be @code{LIO_NOP} |
2124 | in which case this element of @var{list} is simply ignored. This | |
2125 | ``operation'' is useful in situations where one has a fixed array of | |
2126 | @code{struct aiocb} elements from which only a few need to be handled at | |
2127 | a time. Another situation is where the @code{lio_listio} call was | |
19e4c7dd | 2128 | canceled before all requests are processed (@pxref{Cancel AIO |
a3a4a74e UD |
2129 | Operations}) and the remaining requests have to be reissued. |
2130 | ||
fed8f7f7 | 2131 | The other members of each element of the array pointed to by |
a3a4a74e UD |
2132 | @code{list} must have values suitable for the operation as described in |
2133 | the documentation for @code{aio_read} and @code{aio_write} above. | |
2134 | ||
2135 | The @var{mode} argument determines how @code{lio_listio} behaves after | |
2136 | having enqueued all the requests. If @var{mode} is @code{LIO_WAIT} it | |
2137 | waits until all requests terminated. Otherwise @var{mode} must be | |
fed8f7f7 | 2138 | @code{LIO_NOWAIT} and in this case the function returns immediately after |
a3a4a74e UD |
2139 | having enqueued all the requests. In this case the caller gets a |
2140 | notification of the termination of all requests according to the | |
2141 | @var{sig} parameter. If @var{sig} is @code{NULL} no notification is | |
2142 | send. Otherwise a signal is sent or a thread is started, just as | |
2143 | described in the description for @code{aio_read} or @code{aio_write}. | |
2144 | ||
19e4c7dd | 2145 | If @var{mode} is @code{LIO_WAIT}, the return value of @code{lio_listio} |
a3a4a74e UD |
2146 | is @math{0} when all requests completed successfully. Otherwise the |
2147 | function return @math{-1} and @code{errno} is set accordingly. To find | |
2148 | out which request or requests failed one has to use the @code{aio_error} | |
2149 | function on all the elements of the array @var{list}. | |
2150 | ||
19e4c7dd | 2151 | In case @var{mode} is @code{LIO_NOWAIT}, the function returns @math{0} if |
a3a4a74e UD |
2152 | all requests were enqueued correctly. The current state of the requests |
2153 | can be found using @code{aio_error} and @code{aio_return} as described | |
19e4c7dd | 2154 | above. If @code{lio_listio} returns @math{-1} in this mode, the |
a3a4a74e | 2155 | global variable @code{errno} is set accordingly. If a request did not |
19e4c7dd AJ |
2156 | yet terminate, a call to @code{aio_error} returns @code{EINPROGRESS}. If |
2157 | the value is different, the request is finished and the error value (or | |
a3a4a74e UD |
2158 | @math{0}) is returned and the result of the operation can be retrieved |
2159 | using @code{aio_return}. | |
2160 | ||
2161 | Possible values for @code{errno} are: | |
2162 | ||
2163 | @table @code | |
2164 | @item EAGAIN | |
19e4c7dd | 2165 | The resources necessary to queue all the requests are not available at |
a3a4a74e | 2166 | the moment. The error status for each element of @var{list} must be |
19e4c7dd | 2167 | checked to determine which request failed. |
a3a4a74e | 2168 | |
fed8f7f7 | 2169 | Another reason could be that the system wide limit of AIO requests is |
a3a4a74e UD |
2170 | exceeded. This cannot be the case for the implementation on GNU systems |
2171 | since no arbitrary limits exist. | |
2172 | @item EINVAL | |
2173 | The @var{mode} parameter is invalid or @var{nent} is larger than | |
2174 | @code{AIO_LISTIO_MAX}. | |
2175 | @item EIO | |
2176 | One or more of the request's I/O operations failed. The error status of | |
19e4c7dd | 2177 | each request should be checked to determine which one failed. |
a3a4a74e UD |
2178 | @item ENOSYS |
2179 | The @code{lio_listio} function is not supported. | |
2180 | @end table | |
2181 | ||
2182 | If the @var{mode} parameter is @code{LIO_NOWAIT} and the caller cancels | |
19e4c7dd | 2183 | a request, the error status for this request returned by |
a3a4a74e UD |
2184 | @code{aio_error} is @code{ECANCELED}. |
2185 | ||
19e4c7dd | 2186 | When the sources are compiled with @code{_FILE_OFFSET_BITS == 64}, this |
a3a4a74e UD |
2187 | function is in fact @code{lio_listio64} since the LFS interface |
2188 | transparently replaces the normal implementation. | |
2189 | @end deftypefun | |
2190 | ||
2191 | @comment aio.h | |
2192 | @comment Unix98 | |
2193 | @deftypefun int lio_listio64 (int @var{mode}, struct aiocb *const @var{list}, int @var{nent}, struct sigevent *@var{sig}) | |
19e4c7dd AJ |
2194 | This function is similar to the @code{lio_listio} function. The only |
2195 | difference is that on @w{32 bit} machines, the file descriptor should | |
2196 | be opened in the large file mode. Internally, @code{lio_listio64} uses | |
a3a4a74e UD |
2197 | functionality equivalent to @code{lseek64} (@pxref{File Position |
2198 | Primitive}) to position the file descriptor correctly for the reading or | |
fed8f7f7 | 2199 | writing, as opposed to @code{lseek} functionality used in |
a3a4a74e UD |
2200 | @code{lio_listio}. |
2201 | ||
19e4c7dd | 2202 | When the sources are compiled with @code{_FILE_OFFSET_BITS == 64}, this |
a3a4a74e | 2203 | function is available under the name @code{lio_listio} and so |
04b9968b | 2204 | transparently replaces the interface for small files on 32 bit |
a3a4a74e UD |
2205 | machines. |
2206 | @end deftypefun | |
2207 | ||
2208 | @node Status of AIO Operations | |
2209 | @subsection Getting the Status of AIO Operations | |
2210 | ||
fed8f7f7 | 2211 | As already described in the documentation of the functions in the last |
04b9968b UD |
2212 | section, it must be possible to get information about the status of an I/O |
2213 | request. When the operation is performed truly asynchronously (as with | |
19e4c7dd AJ |
2214 | @code{aio_read} and @code{aio_write} and with @code{lio_listio} when the |
2215 | mode is @code{LIO_NOWAIT}), one sometimes needs to know whether a | |
2216 | specific request already terminated and if so, what the result was. | |
04b9968b | 2217 | The following two functions allow you to get this kind of information. |
a3a4a74e UD |
2218 | |
2219 | @comment aio.h | |
2220 | @comment POSIX.1b | |
2221 | @deftypefun int aio_error (const struct aiocb *@var{aiocbp}) | |
2222 | This function determines the error state of the request described by the | |
fed8f7f7 | 2223 | @code{struct aiocb} variable pointed to by @var{aiocbp}. If the |
a3a4a74e UD |
2224 | request has not yet terminated the value returned is always |
2225 | @code{EINPROGRESS}. Once the request has terminated the value | |
2226 | @code{aio_error} returns is either @math{0} if the request completed | |
fed8f7f7 | 2227 | successfully or it returns the value which would be stored in the |
a3a4a74e UD |
2228 | @code{errno} variable if the request would have been done using |
2229 | @code{read}, @code{write}, or @code{fsync}. | |
2230 | ||
2231 | The function can return @code{ENOSYS} if it is not implemented. It | |
2232 | could also return @code{EINVAL} if the @var{aiocbp} parameter does not | |
2233 | refer to an asynchronous operation whose return status is not yet known. | |
2234 | ||
2235 | When the sources are compiled with @code{_FILE_OFFSET_BITS == 64} this | |
2236 | function is in fact @code{aio_error64} since the LFS interface | |
2237 | transparently replaces the normal implementation. | |
2238 | @end deftypefun | |
2239 | ||
2240 | @comment aio.h | |
2241 | @comment Unix98 | |
2242 | @deftypefun int aio_error64 (const struct aiocb64 *@var{aiocbp}) | |
2243 | This function is similar to @code{aio_error} with the only difference | |
2244 | that the argument is a reference to a variable of type @code{struct | |
2245 | aiocb64}. | |
2246 | ||
2247 | When the sources are compiled with @code{_FILE_OFFSET_BITS == 64} this | |
2248 | function is available under the name @code{aio_error} and so | |
04b9968b | 2249 | transparently replaces the interface for small files on 32 bit |
a3a4a74e UD |
2250 | machines. |
2251 | @end deftypefun | |
2252 | ||
2253 | @comment aio.h | |
2254 | @comment POSIX.1b | |
2255 | @deftypefun ssize_t aio_return (const struct aiocb *@var{aiocbp}) | |
2256 | This function can be used to retrieve the return status of the operation | |
2257 | carried out by the request described in the variable pointed to by | |
2258 | @var{aiocbp}. As long as the error status of this request as returned | |
2259 | by @code{aio_error} is @code{EINPROGRESS} the return of this function is | |
2260 | undefined. | |
2261 | ||
fed8f7f7 UD |
2262 | Once the request is finished this function can be used exactly once to |
2263 | retrieve the return value. Following calls might lead to undefined | |
19e4c7dd | 2264 | behavior. The return value itself is the value which would have been |
a3a4a74e UD |
2265 | returned by the @code{read}, @code{write}, or @code{fsync} call. |
2266 | ||
2267 | The function can return @code{ENOSYS} if it is not implemented. It | |
2268 | could also return @code{EINVAL} if the @var{aiocbp} parameter does not | |
2269 | refer to an asynchronous operation whose return status is not yet known. | |
2270 | ||
2271 | When the sources are compiled with @code{_FILE_OFFSET_BITS == 64} this | |
2272 | function is in fact @code{aio_return64} since the LFS interface | |
2273 | transparently replaces the normal implementation. | |
2274 | @end deftypefun | |
2275 | ||
2276 | @comment aio.h | |
2277 | @comment Unix98 | |
2278 | @deftypefun int aio_return64 (const struct aiocb64 *@var{aiocbp}) | |
2279 | This function is similar to @code{aio_return} with the only difference | |
2280 | that the argument is a reference to a variable of type @code{struct | |
2281 | aiocb64}. | |
2282 | ||
2283 | When the sources are compiled with @code{_FILE_OFFSET_BITS == 64} this | |
2284 | function is available under the name @code{aio_return} and so | |
04b9968b | 2285 | transparently replaces the interface for small files on 32 bit |
a3a4a74e UD |
2286 | machines. |
2287 | @end deftypefun | |
2288 | ||
2289 | @node Synchronizing AIO Operations | |
2290 | @subsection Getting into a Consistent State | |
2291 | ||
2292 | When dealing with asynchronous operations it is sometimes necessary to | |
fed8f7f7 | 2293 | get into a consistent state. This would mean for AIO that one wants to |
a3a4a74e UD |
2294 | know whether a certain request or a group of request were processed. |
2295 | This could be done by waiting for the notification sent by the system | |
04b9968b | 2296 | after the operation terminated, but this sometimes would mean wasting |
a3a4a74e UD |
2297 | resources (mainly computation time). Instead POSIX.1b defines two |
2298 | functions which will help with most kinds of consistency. | |
2299 | ||
2300 | The @code{aio_fsync} and @code{aio_fsync64} functions are only available | |
19e4c7dd | 2301 | if the symbol @code{_POSIX_SYNCHRONIZED_IO} is defined in @file{unistd.h}. |
a3a4a74e UD |
2302 | |
2303 | @cindex synchronizing | |
2304 | @comment aio.h | |
2305 | @comment POSIX.1b | |
2306 | @deftypefun int aio_fsync (int @var{op}, struct aiocb *@var{aiocbp}) | |
2307 | Calling this function forces all I/O operations operating queued at the | |
fed8f7f7 | 2308 | time of the function call operating on the file descriptor |
a3a4a74e | 2309 | @code{aiocbp->aio_fildes} into the synchronized I/O completion state |
04b9968b | 2310 | (@pxref{Synchronizing I/O}). The @code{aio_fsync} function returns |
a3a4a74e UD |
2311 | immediately but the notification through the method described in |
2312 | @code{aiocbp->aio_sigevent} will happen only after all requests for this | |
04b9968b | 2313 | file descriptor have terminated and the file is synchronized. This also |
a3a4a74e | 2314 | means that requests for this very same file descriptor which are queued |
04b9968b | 2315 | after the synchronization request are not affected. |
a3a4a74e UD |
2316 | |
2317 | If @var{op} is @code{O_DSYNC} the synchronization happens as with a call | |
2318 | to @code{fdatasync}. Otherwise @var{op} should be @code{O_SYNC} and | |
fed8f7f7 | 2319 | the synchronization happens as with @code{fsync}. |
a3a4a74e | 2320 | |
19e4c7dd | 2321 | As long as the synchronization has not happened, a call to |
a3a4a74e | 2322 | @code{aio_error} with the reference to the object pointed to by |
fed8f7f7 UD |
2323 | @var{aiocbp} returns @code{EINPROGRESS}. Once the synchronization is |
2324 | done @code{aio_error} return @math{0} if the synchronization was not | |
a3a4a74e UD |
2325 | successful. Otherwise the value returned is the value to which the |
2326 | @code{fsync} or @code{fdatasync} function would have set the | |
2327 | @code{errno} variable. In this case nothing can be assumed about the | |
2328 | consistency for the data written to this file descriptor. | |
2329 | ||
2330 | The return value of this function is @math{0} if the request was | |
19e4c7dd | 2331 | successfully enqueued. Otherwise the return value is @math{-1} and |
a3a4a74e UD |
2332 | @code{errno} is set to one of the following values: |
2333 | ||
2334 | @table @code | |
2335 | @item EAGAIN | |
fed8f7f7 | 2336 | The request could not be enqueued due to temporary lack of resources. |
a3a4a74e UD |
2337 | @item EBADF |
2338 | The file descriptor @code{aiocbp->aio_fildes} is not valid or not open | |
2339 | for writing. | |
2340 | @item EINVAL | |
2341 | The implementation does not support I/O synchronization or the @var{op} | |
2342 | parameter is other than @code{O_DSYNC} and @code{O_SYNC}. | |
2343 | @item ENOSYS | |
2344 | This function is not implemented. | |
2345 | @end table | |
2346 | ||
2347 | When the sources are compiled with @code{_FILE_OFFSET_BITS == 64} this | |
37de3d55 | 2348 | function is in fact @code{aio_fsync64} since the LFS interface |
a3a4a74e UD |
2349 | transparently replaces the normal implementation. |
2350 | @end deftypefun | |
2351 | ||
2352 | @comment aio.h | |
2353 | @comment Unix98 | |
2354 | @deftypefun int aio_fsync64 (int @var{op}, struct aiocb64 *@var{aiocbp}) | |
2355 | This function is similar to @code{aio_fsync} with the only difference | |
2356 | that the argument is a reference to a variable of type @code{struct | |
2357 | aiocb64}. | |
2358 | ||
2359 | When the sources are compiled with @code{_FILE_OFFSET_BITS == 64} this | |
2360 | function is available under the name @code{aio_fsync} and so | |
04b9968b | 2361 | transparently replaces the interface for small files on 32 bit |
a3a4a74e UD |
2362 | machines. |
2363 | @end deftypefun | |
2364 | ||
fed8f7f7 | 2365 | Another method of synchronization is to wait until one or more requests of a |
a3a4a74e UD |
2366 | specific set terminated. This could be achieved by the @code{aio_*} |
2367 | functions to notify the initiating process about the termination but in | |
2368 | some situations this is not the ideal solution. In a program which | |
2369 | constantly updates clients somehow connected to the server it is not | |
2370 | always the best solution to go round robin since some connections might | |
2371 | be slow. On the other hand letting the @code{aio_*} function notify the | |
2372 | caller might also be not the best solution since whenever the process | |
2373 | works on preparing data for on client it makes no sense to be | |
2374 | interrupted by a notification since the new client will not be handled | |
2375 | before the current client is served. For situations like this | |
2376 | @code{aio_suspend} should be used. | |
2377 | ||
2378 | @comment aio.h | |
2379 | @comment POSIX.1b | |
2380 | @deftypefun int aio_suspend (const struct aiocb *const @var{list}[], int @var{nent}, const struct timespec *@var{timeout}) | |
19e4c7dd | 2381 | When calling this function, the calling thread is suspended until at |
a3a4a74e | 2382 | least one of the requests pointed to by the @var{nent} elements of the |
19e4c7dd AJ |
2383 | array @var{list} has completed. If any of the requests has already |
2384 | completed at the time @code{aio_suspend} is called, the function returns | |
2385 | immediately. Whether a request has terminated or not is determined by | |
a3a4a74e | 2386 | comparing the error status of the request with @code{EINPROGRESS}. If |
19e4c7dd | 2387 | an element of @var{list} is @code{NULL}, the entry is simply ignored. |
a3a4a74e | 2388 | |
19e4c7dd AJ |
2389 | If no request has finished, the calling process is suspended. If |
2390 | @var{timeout} is @code{NULL}, the process is not woken until a request | |
2391 | has finished. If @var{timeout} is not @code{NULL}, the process remains | |
2392 | suspended at least as long as specified in @var{timeout}. In this case, | |
a3a4a74e UD |
2393 | @code{aio_suspend} returns with an error. |
2394 | ||
fed8f7f7 | 2395 | The return value of the function is @math{0} if one or more requests |
a3a4a74e UD |
2396 | from the @var{list} have terminated. Otherwise the function returns |
2397 | @math{-1} and @code{errno} is set to one of the following values: | |
2398 | ||
2399 | @table @code | |
2400 | @item EAGAIN | |
2401 | None of the requests from the @var{list} completed in the time specified | |
2402 | by @var{timeout}. | |
2403 | @item EINTR | |
2404 | A signal interrupted the @code{aio_suspend} function. This signal might | |
2405 | also be sent by the AIO implementation while signalling the termination | |
2406 | of one of the requests. | |
2407 | @item ENOSYS | |
2408 | The @code{aio_suspend} function is not implemented. | |
2409 | @end table | |
2410 | ||
2411 | When the sources are compiled with @code{_FILE_OFFSET_BITS == 64} this | |
2412 | function is in fact @code{aio_suspend64} since the LFS interface | |
2413 | transparently replaces the normal implementation. | |
2414 | @end deftypefun | |
2415 | ||
2416 | @comment aio.h | |
2417 | @comment Unix98 | |
2418 | @deftypefun int aio_suspend64 (const struct aiocb64 *const @var{list}[], int @var{nent}, const struct timespec *@var{timeout}) | |
2419 | This function is similar to @code{aio_suspend} with the only difference | |
2420 | that the argument is a reference to a variable of type @code{struct | |
2421 | aiocb64}. | |
2422 | ||
2423 | When the sources are compiled with @code{_FILE_OFFSET_BITS == 64} this | |
2424 | function is available under the name @code{aio_suspend} and so | |
04b9968b | 2425 | transparently replaces the interface for small files on 32 bit |
a3a4a74e UD |
2426 | machines. |
2427 | @end deftypefun | |
b07d03e0 UD |
2428 | |
2429 | @node Cancel AIO Operations | |
04b9968b | 2430 | @subsection Cancellation of AIO Operations |
b07d03e0 | 2431 | |
19e4c7dd | 2432 | When one or more requests are asynchronously processed, it might be |
a3a4a74e | 2433 | useful in some situations to cancel a selected operation, e.g., if it |
19e4c7dd AJ |
2434 | becomes obvious that the written data is no longer accurate and would |
2435 | have to be overwritten soon. As an example, assume an application, which | |
a3a4a74e UD |
2436 | writes data in files in a situation where new incoming data would have |
2437 | to be written in a file which will be updated by an enqueued request. | |
19e4c7dd AJ |
2438 | The POSIX AIO implementation provides such a function, but this function |
2439 | is not capable of forcing the cancellation of the request. It is up to the | |
a3a4a74e UD |
2440 | implementation to decide whether it is possible to cancel the operation |
2441 | or not. Therefore using this function is merely a hint. | |
2442 | ||
2443 | @comment aio.h | |
2444 | @comment POSIX.1b | |
2445 | @deftypefun int aio_cancel (int @var{fildes}, struct aiocb *@var{aiocbp}) | |
2446 | The @code{aio_cancel} function can be used to cancel one or more | |
19e4c7dd AJ |
2447 | outstanding requests. If the @var{aiocbp} parameter is @code{NULL}, the |
2448 | function tries to cancel all of the outstanding requests which would process | |
2449 | the file descriptor @var{fildes} (i.e., whose @code{aio_fildes} member | |
2450 | is @var{fildes}). If @var{aiocbp} is not @code{NULL}, @code{aio_cancel} | |
2451 | attempts to cancel the specific request pointed to by @var{aiocbp}. | |
a3a4a74e | 2452 | |
19e4c7dd | 2453 | For requests which were successfully canceled, the normal notification |
a3a4a74e UD |
2454 | about the termination of the request should take place. I.e., depending |
2455 | on the @code{struct sigevent} object which controls this, nothing | |
2456 | happens, a signal is sent or a thread is started. If the request cannot | |
19e4c7dd | 2457 | be canceled, it terminates the usual way after performing the operation. |
a3a4a74e | 2458 | |
19e4c7dd | 2459 | After a request is successfully canceled, a call to @code{aio_error} with |
a3a4a74e UD |
2460 | a reference to this request as the parameter will return |
2461 | @code{ECANCELED} and a call to @code{aio_return} will return @math{-1}. | |
19e4c7dd | 2462 | If the request wasn't canceled and is still running the error status is |
a3a4a74e UD |
2463 | still @code{EINPROGRESS}. |
2464 | ||
2465 | The return value of the function is @code{AIO_CANCELED} if there were | |
19e4c7dd AJ |
2466 | requests which haven't terminated and which were successfully canceled. |
2467 | If there is one or more requests left which couldn't be canceled, the | |
a3a4a74e | 2468 | return value is @code{AIO_NOTCANCELED}. In this case @code{aio_error} |
19e4c7dd AJ |
2469 | must be used to find out which of the, perhaps multiple, requests (in |
2470 | @var{aiocbp} is @code{NULL}) weren't successfully canceled. If all | |
a3a4a74e UD |
2471 | requests already terminated at the time @code{aio_cancel} is called the |
2472 | return value is @code{AIO_ALLDONE}. | |
2473 | ||
2474 | If an error occurred during the execution of @code{aio_cancel} the | |
2475 | function returns @math{-1} and sets @code{errno} to one of the following | |
2476 | values. | |
2477 | ||
2478 | @table @code | |
2479 | @item EBADF | |
2480 | The file descriptor @var{fildes} is not valid. | |
2481 | @item ENOSYS | |
2482 | @code{aio_cancel} is not implemented. | |
2483 | @end table | |
2484 | ||
19e4c7dd | 2485 | When the sources are compiled with @code{_FILE_OFFSET_BITS == 64}, this |
a3a4a74e UD |
2486 | function is in fact @code{aio_cancel64} since the LFS interface |
2487 | transparently replaces the normal implementation. | |
2488 | @end deftypefun | |
2489 | ||
2490 | @comment aio.h | |
2491 | @comment Unix98 | |
19e4c7dd | 2492 | @deftypefun int aio_cancel64 (int @var{fildes}, struct aiocb64 *@var{aiocbp}) |
a3a4a74e UD |
2493 | This function is similar to @code{aio_cancel} with the only difference |
2494 | that the argument is a reference to a variable of type @code{struct | |
2495 | aiocb64}. | |
2496 | ||
19e4c7dd | 2497 | When the sources are compiled with @code{_FILE_OFFSET_BITS == 64}, this |
a3a4a74e | 2498 | function is available under the name @code{aio_cancel} and so |
04b9968b | 2499 | transparently replaces the interface for small files on 32 bit |
a3a4a74e UD |
2500 | machines. |
2501 | @end deftypefun | |
2502 | ||
2503 | @node Configuration of AIO | |
2504 | @subsection How to optimize the AIO implementation | |
2505 | ||
2506 | The POSIX standard does not specify how the AIO functions are | |
19e4c7dd | 2507 | implemented. They could be system calls, but it is also possible to |
a3a4a74e UD |
2508 | emulate them at userlevel. |
2509 | ||
19e4c7dd AJ |
2510 | At the point of this writing, the available implementation is a userlevel |
2511 | implementation which uses threads for handling the enqueued requests. | |
2512 | While this implementation requires making some decisions about | |
2513 | limitations, hard limitations are something which is best avoided | |
2514 | in the GNU C library. Therefore, the GNU C library provides a means | |
2515 | for tuning the AIO implementation according to the individual use. | |
a3a4a74e UD |
2516 | |
2517 | @comment aio.h | |
2518 | @comment GNU | |
2519 | @deftp {Data Type} {struct aioinit} | |
2520 | This data type is used to pass the configuration or tunable parameters | |
2521 | to the implementation. The program has to initialize the members of | |
2522 | this struct and pass it to the implementation using the @code{aio_init} | |
2523 | function. | |
2524 | ||
2525 | @table @code | |
2526 | @item int aio_threads | |
19e4c7dd | 2527 | This member specifies the maximal number of threads which may be used |
a3a4a74e UD |
2528 | at any one time. |
2529 | @item int aio_num | |
c756c71c | 2530 | This number provides an estimate on the maximal number of simultaneously |
a3a4a74e UD |
2531 | enqueued requests. |
2532 | @item int aio_locks | |
19e4c7dd | 2533 | Unused. |
a3a4a74e | 2534 | @item int aio_usedba |
19e4c7dd | 2535 | Unused. |
a3a4a74e | 2536 | @item int aio_debug |
19e4c7dd | 2537 | Unused. |
a3a4a74e | 2538 | @item int aio_numusers |
19e4c7dd | 2539 | Unused. |
a3a4a74e | 2540 | @item int aio_reserved[2] |
19e4c7dd | 2541 | Unused. |
a3a4a74e UD |
2542 | @end table |
2543 | @end deftp | |
2544 | ||
2545 | @comment aio.h | |
2546 | @comment GNU | |
2547 | @deftypefun void aio_init (const struct aioinit *@var{init}) | |
2548 | This function must be called before any other AIO function. Calling it | |
19e4c7dd AJ |
2549 | is completely voluntary, as it is only meant to help the AIO |
2550 | implementation perform better. | |
a3a4a74e | 2551 | |
19e4c7dd | 2552 | Before calling the @code{aio_init}, function the members of a variable of |
a3a4a74e UD |
2553 | type @code{struct aioinit} must be initialized. Then a reference to |
2554 | this variable is passed as the parameter to @code{aio_init} which itself | |
2555 | may or may not pay attention to the hints. | |
2556 | ||
c756c71c UD |
2557 | The function has no return value and no error cases are defined. It is |
2558 | a extension which follows a proposal from the SGI implementation in | |
2559 | @w{Irix 6}. It is not covered by POSIX.1b or Unix98. | |
a3a4a74e | 2560 | @end deftypefun |
b07d03e0 | 2561 | |
28f540f4 RM |
2562 | @node Control Operations |
2563 | @section Control Operations on Files | |
2564 | ||
2565 | @cindex control operations on files | |
2566 | @cindex @code{fcntl} function | |
2567 | This section describes how you can perform various other operations on | |
2568 | file descriptors, such as inquiring about or setting flags describing | |
2569 | the status of the file descriptor, manipulating record locks, and the | |
2570 | like. All of these operations are performed by the function @code{fcntl}. | |
2571 | ||
2572 | The second argument to the @code{fcntl} function is a command that | |
2573 | specifies which operation to perform. The function and macros that name | |
2574 | various flags that are used with it are declared in the header file | |
2575 | @file{fcntl.h}. Many of these flags are also used by the @code{open} | |
2576 | function; see @ref{Opening and Closing Files}. | |
2577 | @pindex fcntl.h | |
2578 | ||
2579 | @comment fcntl.h | |
2580 | @comment POSIX.1 | |
2581 | @deftypefun int fcntl (int @var{filedes}, int @var{command}, @dots{}) | |
2582 | The @code{fcntl} function performs the operation specified by | |
2583 | @var{command} on the file descriptor @var{filedes}. Some commands | |
2584 | require additional arguments to be supplied. These additional arguments | |
2585 | and the return value and error conditions are given in the detailed | |
2586 | descriptions of the individual commands. | |
2587 | ||
2588 | Briefly, here is a list of what the various commands are. | |
2589 | ||
2590 | @table @code | |
2591 | @item F_DUPFD | |
2592 | Duplicate the file descriptor (return another file descriptor pointing | |
2593 | to the same open file). @xref{Duplicating Descriptors}. | |
2594 | ||
2595 | @item F_GETFD | |
2596 | Get flags associated with the file descriptor. @xref{Descriptor Flags}. | |
2597 | ||
2598 | @item F_SETFD | |
2599 | Set flags associated with the file descriptor. @xref{Descriptor Flags}. | |
2600 | ||
2601 | @item F_GETFL | |
2602 | Get flags associated with the open file. @xref{File Status Flags}. | |
2603 | ||
2604 | @item F_SETFL | |
2605 | Set flags associated with the open file. @xref{File Status Flags}. | |
2606 | ||
2607 | @item F_GETLK | |
2608 | Get a file lock. @xref{File Locks}. | |
2609 | ||
2610 | @item F_SETLK | |
2611 | Set or clear a file lock. @xref{File Locks}. | |
2612 | ||
2613 | @item F_SETLKW | |
2614 | Like @code{F_SETLK}, but wait for completion. @xref{File Locks}. | |
2615 | ||
2616 | @item F_GETOWN | |
2617 | Get process or process group ID to receive @code{SIGIO} signals. | |
2618 | @xref{Interrupt Input}. | |
2619 | ||
2620 | @item F_SETOWN | |
2621 | Set process or process group ID to receive @code{SIGIO} signals. | |
2622 | @xref{Interrupt Input}. | |
2623 | @end table | |
dfd2257a | 2624 | |
04b9968b | 2625 | This function is a cancellation point in multi-threaded programs. This |
dfd2257a UD |
2626 | is a problem if the thread allocates some resources (like memory, file |
2627 | descriptors, semaphores or whatever) at the time @code{fcntl} is | |
19e4c7dd | 2628 | called. If the thread gets canceled these resources stay allocated |
dfd2257a | 2629 | until the program ends. To avoid this calls to @code{fcntl} should be |
04b9968b | 2630 | protected using cancellation handlers. |
dfd2257a | 2631 | @c ref pthread_cleanup_push / pthread_cleanup_pop |
28f540f4 RM |
2632 | @end deftypefun |
2633 | ||
2634 | ||
2635 | @node Duplicating Descriptors | |
2636 | @section Duplicating Descriptors | |
2637 | ||
2638 | @cindex duplicating file descriptors | |
2639 | @cindex redirecting input and output | |
2640 | ||
2641 | You can @dfn{duplicate} a file descriptor, or allocate another file | |
2642 | descriptor that refers to the same open file as the original. Duplicate | |
2643 | descriptors share one file position and one set of file status flags | |
2644 | (@pxref{File Status Flags}), but each has its own set of file descriptor | |
2645 | flags (@pxref{Descriptor Flags}). | |
2646 | ||
2647 | The major use of duplicating a file descriptor is to implement | |
2648 | @dfn{redirection} of input or output: that is, to change the | |
2649 | file or pipe that a particular file descriptor corresponds to. | |
2650 | ||
2651 | You can perform this operation using the @code{fcntl} function with the | |
2652 | @code{F_DUPFD} command, but there are also convenient functions | |
2653 | @code{dup} and @code{dup2} for duplicating descriptors. | |
2654 | ||
2655 | @pindex unistd.h | |
2656 | @pindex fcntl.h | |
2657 | The @code{fcntl} function and flags are declared in @file{fcntl.h}, | |
2658 | while prototypes for @code{dup} and @code{dup2} are in the header file | |
2659 | @file{unistd.h}. | |
2660 | ||
2661 | @comment unistd.h | |
2662 | @comment POSIX.1 | |
2663 | @deftypefun int dup (int @var{old}) | |
2664 | This function copies descriptor @var{old} to the first available | |
2665 | descriptor number (the first number not currently open). It is | |
2666 | equivalent to @code{fcntl (@var{old}, F_DUPFD, 0)}. | |
2667 | @end deftypefun | |
2668 | ||
2669 | @comment unistd.h | |
2670 | @comment POSIX.1 | |
2671 | @deftypefun int dup2 (int @var{old}, int @var{new}) | |
2672 | This function copies the descriptor @var{old} to descriptor number | |
2673 | @var{new}. | |
2674 | ||
2675 | If @var{old} is an invalid descriptor, then @code{dup2} does nothing; it | |
2676 | does not close @var{new}. Otherwise, the new duplicate of @var{old} | |
2677 | replaces any previous meaning of descriptor @var{new}, as if @var{new} | |
2678 | were closed first. | |
2679 | ||
2680 | If @var{old} and @var{new} are different numbers, and @var{old} is a | |
2681 | valid descriptor number, then @code{dup2} is equivalent to: | |
2682 | ||
2683 | @smallexample | |
2684 | close (@var{new}); | |
2685 | fcntl (@var{old}, F_DUPFD, @var{new}) | |
2686 | @end smallexample | |
2687 | ||
2688 | However, @code{dup2} does this atomically; there is no instant in the | |
2689 | middle of calling @code{dup2} at which @var{new} is closed and not yet a | |
2690 | duplicate of @var{old}. | |
2691 | @end deftypefun | |
2692 | ||
2693 | @comment fcntl.h | |
2694 | @comment POSIX.1 | |
2695 | @deftypevr Macro int F_DUPFD | |
2696 | This macro is used as the @var{command} argument to @code{fcntl}, to | |
2697 | copy the file descriptor given as the first argument. | |
2698 | ||
2699 | The form of the call in this case is: | |
2700 | ||
2701 | @smallexample | |
2702 | fcntl (@var{old}, F_DUPFD, @var{next-filedes}) | |
2703 | @end smallexample | |
2704 | ||
2705 | The @var{next-filedes} argument is of type @code{int} and specifies that | |
2706 | the file descriptor returned should be the next available one greater | |
2707 | than or equal to this value. | |
2708 | ||
2709 | The return value from @code{fcntl} with this command is normally the value | |
07435eb4 | 2710 | of the new file descriptor. A return value of @math{-1} indicates an |
28f540f4 RM |
2711 | error. The following @code{errno} error conditions are defined for |
2712 | this command: | |
2713 | ||
2714 | @table @code | |
2715 | @item EBADF | |
2716 | The @var{old} argument is invalid. | |
2717 | ||
2718 | @item EINVAL | |
2719 | The @var{next-filedes} argument is invalid. | |
2720 | ||
2721 | @item EMFILE | |
2722 | There are no more file descriptors available---your program is already | |
2723 | using the maximum. In BSD and GNU, the maximum is controlled by a | |
2724 | resource limit that can be changed; @pxref{Limits on Resources}, for | |
2725 | more information about the @code{RLIMIT_NOFILE} limit. | |
2726 | @end table | |
2727 | ||
2728 | @code{ENFILE} is not a possible error code for @code{dup2} because | |
2729 | @code{dup2} does not create a new opening of a file; duplicate | |
2730 | descriptors do not count toward the limit which @code{ENFILE} | |
2731 | indicates. @code{EMFILE} is possible because it refers to the limit on | |
2732 | distinct descriptor numbers in use in one process. | |
2733 | @end deftypevr | |
2734 | ||
2735 | Here is an example showing how to use @code{dup2} to do redirection. | |
2736 | Typically, redirection of the standard streams (like @code{stdin}) is | |
2737 | done by a shell or shell-like program before calling one of the | |
2738 | @code{exec} functions (@pxref{Executing a File}) to execute a new | |
2739 | program in a child process. When the new program is executed, it | |
2740 | creates and initializes the standard streams to point to the | |
2741 | corresponding file descriptors, before its @code{main} function is | |
2742 | invoked. | |
2743 | ||
2744 | So, to redirect standard input to a file, the shell could do something | |
2745 | like: | |
2746 | ||
2747 | @smallexample | |
2748 | pid = fork (); | |
2749 | if (pid == 0) | |
2750 | @{ | |
2751 | char *filename; | |
2752 | char *program; | |
2753 | int file; | |
2754 | @dots{} | |
2755 | file = TEMP_FAILURE_RETRY (open (filename, O_RDONLY)); | |
2756 | dup2 (file, STDIN_FILENO); | |
2757 | TEMP_FAILURE_RETRY (close (file)); | |
2758 | execv (program, NULL); | |
2759 | @} | |
2760 | @end smallexample | |
2761 | ||
2762 | There is also a more detailed example showing how to implement redirection | |
2763 | in the context of a pipeline of processes in @ref{Launching Jobs}. | |
2764 | ||
2765 | ||
2766 | @node Descriptor Flags | |
2767 | @section File Descriptor Flags | |
2768 | @cindex file descriptor flags | |
2769 | ||
2770 | @dfn{File descriptor flags} are miscellaneous attributes of a file | |
2771 | descriptor. These flags are associated with particular file | |
2772 | descriptors, so that if you have created duplicate file descriptors | |
2773 | from a single opening of a file, each descriptor has its own set of flags. | |
2774 | ||
2775 | Currently there is just one file descriptor flag: @code{FD_CLOEXEC}, | |
2776 | which causes the descriptor to be closed if you use any of the | |
2777 | @code{exec@dots{}} functions (@pxref{Executing a File}). | |
2778 | ||
2779 | The symbols in this section are defined in the header file | |
2780 | @file{fcntl.h}. | |
2781 | @pindex fcntl.h | |
2782 | ||
2783 | @comment fcntl.h | |
2784 | @comment POSIX.1 | |
2785 | @deftypevr Macro int F_GETFD | |
2786 | This macro is used as the @var{command} argument to @code{fcntl}, to | |
2787 | specify that it should return the file descriptor flags associated | |
2c6fe0bd | 2788 | with the @var{filedes} argument. |
28f540f4 RM |
2789 | |
2790 | The normal return value from @code{fcntl} with this command is a | |
2791 | nonnegative number which can be interpreted as the bitwise OR of the | |
2792 | individual flags (except that currently there is only one flag to use). | |
2793 | ||
07435eb4 | 2794 | In case of an error, @code{fcntl} returns @math{-1}. The following |
28f540f4 RM |
2795 | @code{errno} error conditions are defined for this command: |
2796 | ||
2797 | @table @code | |
2798 | @item EBADF | |
2799 | The @var{filedes} argument is invalid. | |
2800 | @end table | |
2801 | @end deftypevr | |
2802 | ||
2803 | ||
2804 | @comment fcntl.h | |
2805 | @comment POSIX.1 | |
2806 | @deftypevr Macro int F_SETFD | |
2807 | This macro is used as the @var{command} argument to @code{fcntl}, to | |
2808 | specify that it should set the file descriptor flags associated with the | |
2809 | @var{filedes} argument. This requires a third @code{int} argument to | |
2810 | specify the new flags, so the form of the call is: | |
2811 | ||
2812 | @smallexample | |
2813 | fcntl (@var{filedes}, F_SETFD, @var{new-flags}) | |
2814 | @end smallexample | |
2815 | ||
2816 | The normal return value from @code{fcntl} with this command is an | |
07435eb4 | 2817 | unspecified value other than @math{-1}, which indicates an error. |
28f540f4 RM |
2818 | The flags and error conditions are the same as for the @code{F_GETFD} |
2819 | command. | |
2820 | @end deftypevr | |
2821 | ||
2822 | The following macro is defined for use as a file descriptor flag with | |
2823 | the @code{fcntl} function. The value is an integer constant usable | |
2824 | as a bit mask value. | |
2825 | ||
2826 | @comment fcntl.h | |
2827 | @comment POSIX.1 | |
2828 | @deftypevr Macro int FD_CLOEXEC | |
2829 | @cindex close-on-exec (file descriptor flag) | |
2830 | This flag specifies that the file descriptor should be closed when | |
2831 | an @code{exec} function is invoked; see @ref{Executing a File}. When | |
2832 | a file descriptor is allocated (as with @code{open} or @code{dup}), | |
2833 | this bit is initially cleared on the new file descriptor, meaning that | |
2834 | descriptor will survive into the new program after @code{exec}. | |
2835 | @end deftypevr | |
2836 | ||
2837 | If you want to modify the file descriptor flags, you should get the | |
2838 | current flags with @code{F_GETFD} and modify the value. Don't assume | |
2839 | that the flags listed here are the only ones that are implemented; your | |
2840 | program may be run years from now and more flags may exist then. For | |
2841 | example, here is a function to set or clear the flag @code{FD_CLOEXEC} | |
2842 | without altering any other flags: | |
2843 | ||
2844 | @smallexample | |
2845 | /* @r{Set the @code{FD_CLOEXEC} flag of @var{desc} if @var{value} is nonzero,} | |
2846 | @r{or clear the flag if @var{value} is 0.} | |
2c6fe0bd | 2847 | @r{Return 0 on success, or -1 on error with @code{errno} set.} */ |
28f540f4 RM |
2848 | |
2849 | int | |
2850 | set_cloexec_flag (int desc, int value) | |
2851 | @{ | |
2852 | int oldflags = fcntl (desc, F_GETFD, 0); | |
8e96ae1a | 2853 | /* @r{If reading the flags failed, return error indication now.} */ |
28f540f4 RM |
2854 | if (oldflags < 0) |
2855 | return oldflags; | |
2856 | /* @r{Set just the flag we want to set.} */ | |
2857 | if (value != 0) | |
2858 | oldflags |= FD_CLOEXEC; | |
2859 | else | |
2860 | oldflags &= ~FD_CLOEXEC; | |
2861 | /* @r{Store modified flag word in the descriptor.} */ | |
2862 | return fcntl (desc, F_SETFD, oldflags); | |
2863 | @} | |
2864 | @end smallexample | |
2865 | ||
2866 | @node File Status Flags | |
2867 | @section File Status Flags | |
2868 | @cindex file status flags | |
2869 | ||
2870 | @dfn{File status flags} are used to specify attributes of the opening of a | |
2871 | file. Unlike the file descriptor flags discussed in @ref{Descriptor | |
2872 | Flags}, the file status flags are shared by duplicated file descriptors | |
2873 | resulting from a single opening of the file. The file status flags are | |
2874 | specified with the @var{flags} argument to @code{open}; | |
2875 | @pxref{Opening and Closing Files}. | |
2876 | ||
2877 | File status flags fall into three categories, which are described in the | |
2878 | following sections. | |
2879 | ||
2880 | @itemize @bullet | |
2881 | @item | |
2882 | @ref{Access Modes}, specify what type of access is allowed to the | |
2883 | file: reading, writing, or both. They are set by @code{open} and are | |
2884 | returned by @code{fcntl}, but cannot be changed. | |
2885 | ||
2886 | @item | |
2887 | @ref{Open-time Flags}, control details of what @code{open} will do. | |
2888 | These flags are not preserved after the @code{open} call. | |
2889 | ||
2890 | @item | |
2891 | @ref{Operating Modes}, affect how operations such as @code{read} and | |
2892 | @code{write} are done. They are set by @code{open}, and can be fetched or | |
2893 | changed with @code{fcntl}. | |
2894 | @end itemize | |
2895 | ||
2896 | The symbols in this section are defined in the header file | |
2897 | @file{fcntl.h}. | |
2898 | @pindex fcntl.h | |
2899 | ||
2900 | @menu | |
2901 | * Access Modes:: Whether the descriptor can read or write. | |
2902 | * Open-time Flags:: Details of @code{open}. | |
2903 | * Operating Modes:: Special modes to control I/O operations. | |
2904 | * Getting File Status Flags:: Fetching and changing these flags. | |
2905 | @end menu | |
2906 | ||
2907 | @node Access Modes | |
2908 | @subsection File Access Modes | |
2909 | ||
2910 | The file access modes allow a file descriptor to be used for reading, | |
2911 | writing, or both. (In the GNU system, they can also allow none of these, | |
2912 | and allow execution of the file as a program.) The access modes are chosen | |
2913 | when the file is opened, and never change. | |
2914 | ||
2915 | @comment fcntl.h | |
2916 | @comment POSIX.1 | |
2917 | @deftypevr Macro int O_RDONLY | |
2918 | Open the file for read access. | |
2919 | @end deftypevr | |
2920 | ||
2921 | @comment fcntl.h | |
2922 | @comment POSIX.1 | |
2923 | @deftypevr Macro int O_WRONLY | |
2924 | Open the file for write access. | |
2925 | @end deftypevr | |
2926 | ||
2927 | @comment fcntl.h | |
2928 | @comment POSIX.1 | |
2929 | @deftypevr Macro int O_RDWR | |
2930 | Open the file for both reading and writing. | |
2931 | @end deftypevr | |
2932 | ||
2933 | In the GNU system (and not in other systems), @code{O_RDONLY} and | |
2934 | @code{O_WRONLY} are independent bits that can be bitwise-ORed together, | |
2935 | and it is valid for either bit to be set or clear. This means that | |
2936 | @code{O_RDWR} is the same as @code{O_RDONLY|O_WRONLY}. A file access | |
2937 | mode of zero is permissible; it allows no operations that do input or | |
2938 | output to the file, but does allow other operations such as | |
2939 | @code{fchmod}. On the GNU system, since ``read-only'' or ``write-only'' | |
2940 | is a misnomer, @file{fcntl.h} defines additional names for the file | |
2941 | access modes. These names are preferred when writing GNU-specific code. | |
2942 | But most programs will want to be portable to other POSIX.1 systems and | |
2943 | should use the POSIX.1 names above instead. | |
2944 | ||
2945 | @comment fcntl.h | |
2946 | @comment GNU | |
2947 | @deftypevr Macro int O_READ | |
c2835376 | 2948 | Open the file for reading. Same as @code{O_RDONLY}; only defined on GNU. |
28f540f4 RM |
2949 | @end deftypevr |
2950 | ||
2951 | @comment fcntl.h | |
2952 | @comment GNU | |
2953 | @deftypevr Macro int O_WRITE | |
c2835376 | 2954 | Open the file for writing. Same as @code{O_WRONLY}; only defined on GNU. |
28f540f4 RM |
2955 | @end deftypevr |
2956 | ||
2957 | @comment fcntl.h | |
2958 | @comment GNU | |
2959 | @deftypevr Macro int O_EXEC | |
2960 | Open the file for executing. Only defined on GNU. | |
2961 | @end deftypevr | |
2962 | ||
2963 | To determine the file access mode with @code{fcntl}, you must extract | |
2964 | the access mode bits from the retrieved file status flags. In the GNU | |
2965 | system, you can just test the @code{O_READ} and @code{O_WRITE} bits in | |
2966 | the flags word. But in other POSIX.1 systems, reading and writing | |
2967 | access modes are not stored as distinct bit flags. The portable way to | |
2968 | extract the file access mode bits is with @code{O_ACCMODE}. | |
2969 | ||
2970 | @comment fcntl.h | |
2971 | @comment POSIX.1 | |
2972 | @deftypevr Macro int O_ACCMODE | |
2973 | This macro stands for a mask that can be bitwise-ANDed with the file | |
2974 | status flag value to produce a value representing the file access mode. | |
2975 | The mode will be @code{O_RDONLY}, @code{O_WRONLY}, or @code{O_RDWR}. | |
2976 | (In the GNU system it could also be zero, and it never includes the | |
2977 | @code{O_EXEC} bit.) | |
2978 | @end deftypevr | |
2979 | ||
2980 | @node Open-time Flags | |
2981 | @subsection Open-time Flags | |
2982 | ||
2983 | The open-time flags specify options affecting how @code{open} will behave. | |
2984 | These options are not preserved once the file is open. The exception to | |
2985 | this is @code{O_NONBLOCK}, which is also an I/O operating mode and so it | |
2986 | @emph{is} saved. @xref{Opening and Closing Files}, for how to call | |
2987 | @code{open}. | |
2988 | ||
2989 | There are two sorts of options specified by open-time flags. | |
2990 | ||
2991 | @itemize @bullet | |
2992 | @item | |
2993 | @dfn{File name translation flags} affect how @code{open} looks up the | |
2994 | file name to locate the file, and whether the file can be created. | |
2995 | @cindex file name translation flags | |
2996 | @cindex flags, file name translation | |
2997 | ||
2998 | @item | |
2999 | @dfn{Open-time action flags} specify extra operations that @code{open} will | |
3000 | perform on the file once it is open. | |
3001 | @cindex open-time action flags | |
3002 | @cindex flags, open-time action | |
3003 | @end itemize | |
3004 | ||
3005 | Here are the file name translation flags. | |
3006 | ||
3007 | @comment fcntl.h | |
3008 | @comment POSIX.1 | |
3009 | @deftypevr Macro int O_CREAT | |
3010 | If set, the file will be created if it doesn't already exist. | |
3011 | @c !!! mode arg, umask | |
3012 | @cindex create on open (file status flag) | |
3013 | @end deftypevr | |
3014 | ||
3015 | @comment fcntl.h | |
3016 | @comment POSIX.1 | |
3017 | @deftypevr Macro int O_EXCL | |
3018 | If both @code{O_CREAT} and @code{O_EXCL} are set, then @code{open} fails | |
3019 | if the specified file already exists. This is guaranteed to never | |
3020 | clobber an existing file. | |
3021 | @end deftypevr | |
3022 | ||
3023 | @comment fcntl.h | |
3024 | @comment POSIX.1 | |
3025 | @deftypevr Macro int O_NONBLOCK | |
3026 | @cindex non-blocking open | |
3027 | This prevents @code{open} from blocking for a ``long time'' to open the | |
3028 | file. This is only meaningful for some kinds of files, usually devices | |
3029 | such as serial ports; when it is not meaningful, it is harmless and | |
3030 | ignored. Often opening a port to a modem blocks until the modem reports | |
3031 | carrier detection; if @code{O_NONBLOCK} is specified, @code{open} will | |
3032 | return immediately without a carrier. | |
3033 | ||
3034 | Note that the @code{O_NONBLOCK} flag is overloaded as both an I/O operating | |
3035 | mode and a file name translation flag. This means that specifying | |
3036 | @code{O_NONBLOCK} in @code{open} also sets nonblocking I/O mode; | |
3037 | @pxref{Operating Modes}. To open the file without blocking but do normal | |
3038 | I/O that blocks, you must call @code{open} with @code{O_NONBLOCK} set and | |
3039 | then call @code{fcntl} to turn the bit off. | |
3040 | @end deftypevr | |
3041 | ||
3042 | @comment fcntl.h | |
3043 | @comment POSIX.1 | |
3044 | @deftypevr Macro int O_NOCTTY | |
3045 | If the named file is a terminal device, don't make it the controlling | |
3046 | terminal for the process. @xref{Job Control}, for information about | |
3047 | what it means to be the controlling terminal. | |
3048 | ||
3049 | In the GNU system and 4.4 BSD, opening a file never makes it the | |
3050 | controlling terminal and @code{O_NOCTTY} is zero. However, other | |
3051 | systems may use a nonzero value for @code{O_NOCTTY} and set the | |
3052 | controlling terminal when you open a file that is a terminal device; so | |
3053 | to be portable, use @code{O_NOCTTY} when it is important to avoid this. | |
3054 | @cindex controlling terminal, setting | |
3055 | @end deftypevr | |
3056 | ||
3057 | The following three file name translation flags exist only in the GNU system. | |
3058 | ||
3059 | @comment fcntl.h | |
3060 | @comment GNU | |
3061 | @deftypevr Macro int O_IGNORE_CTTY | |
3062 | Do not recognize the named file as the controlling terminal, even if it | |
3063 | refers to the process's existing controlling terminal device. Operations | |
3064 | on the new file descriptor will never induce job control signals. | |
3065 | @xref{Job Control}. | |
3066 | @end deftypevr | |
3067 | ||
3068 | @comment fcntl.h | |
3069 | @comment GNU | |
3070 | @deftypevr Macro int O_NOLINK | |
3071 | If the named file is a symbolic link, open the link itself instead of | |
3072 | the file it refers to. (@code{fstat} on the new file descriptor will | |
3073 | return the information returned by @code{lstat} on the link's name.) | |
3074 | @cindex symbolic link, opening | |
3075 | @end deftypevr | |
3076 | ||
3077 | @comment fcntl.h | |
3078 | @comment GNU | |
3079 | @deftypevr Macro int O_NOTRANS | |
3080 | If the named file is specially translated, do not invoke the translator. | |
3081 | Open the bare file the translator itself sees. | |
3082 | @end deftypevr | |
3083 | ||
3084 | ||
3085 | The open-time action flags tell @code{open} to do additional operations | |
3086 | which are not really related to opening the file. The reason to do them | |
3087 | as part of @code{open} instead of in separate calls is that @code{open} | |
3088 | can do them @i{atomically}. | |
3089 | ||
3090 | @comment fcntl.h | |
3091 | @comment POSIX.1 | |
3092 | @deftypevr Macro int O_TRUNC | |
3093 | Truncate the file to zero length. This option is only useful for | |
3094 | regular files, not special files such as directories or FIFOs. POSIX.1 | |
3095 | requires that you open the file for writing to use @code{O_TRUNC}. In | |
3096 | BSD and GNU you must have permission to write the file to truncate it, | |
3097 | but you need not open for write access. | |
3098 | ||
3099 | This is the only open-time action flag specified by POSIX.1. There is | |
3100 | no good reason for truncation to be done by @code{open}, instead of by | |
3101 | calling @code{ftruncate} afterwards. The @code{O_TRUNC} flag existed in | |
3102 | Unix before @code{ftruncate} was invented, and is retained for backward | |
3103 | compatibility. | |
3104 | @end deftypevr | |
3105 | ||
27e309c1 UD |
3106 | The remaining operating modes are BSD extensions. They exist only |
3107 | on some systems. On other systems, these macros are not defined. | |
3108 | ||
28f540f4 RM |
3109 | @comment fcntl.h |
3110 | @comment BSD | |
3111 | @deftypevr Macro int O_SHLOCK | |
3112 | Acquire a shared lock on the file, as with @code{flock}. | |
3113 | @xref{File Locks}. | |
3114 | ||
3115 | If @code{O_CREAT} is specified, the locking is done atomically when | |
3116 | creating the file. You are guaranteed that no other process will get | |
3117 | the lock on the new file first. | |
3118 | @end deftypevr | |
3119 | ||
3120 | @comment fcntl.h | |
3121 | @comment BSD | |
3122 | @deftypevr Macro int O_EXLOCK | |
3123 | Acquire an exclusive lock on the file, as with @code{flock}. | |
3124 | @xref{File Locks}. This is atomic like @code{O_SHLOCK}. | |
3125 | @end deftypevr | |
3126 | ||
3127 | @node Operating Modes | |
3128 | @subsection I/O Operating Modes | |
3129 | ||
3130 | The operating modes affect how input and output operations using a file | |
3131 | descriptor work. These flags are set by @code{open} and can be fetched | |
3132 | and changed with @code{fcntl}. | |
3133 | ||
3134 | @comment fcntl.h | |
3135 | @comment POSIX.1 | |
3136 | @deftypevr Macro int O_APPEND | |
3137 | The bit that enables append mode for the file. If set, then all | |
3138 | @code{write} operations write the data at the end of the file, extending | |
3139 | it, regardless of the current file position. This is the only reliable | |
3140 | way to append to a file. In append mode, you are guaranteed that the | |
3141 | data you write will always go to the current end of the file, regardless | |
3142 | of other processes writing to the file. Conversely, if you simply set | |
3143 | the file position to the end of file and write, then another process can | |
3144 | extend the file after you set the file position but before you write, | |
3145 | resulting in your data appearing someplace before the real end of file. | |
3146 | @end deftypevr | |
3147 | ||
3148 | @comment fcntl.h | |
3149 | @comment POSIX.1 | |
2c6fe0bd | 3150 | @deftypevr Macro int O_NONBLOCK |
28f540f4 RM |
3151 | The bit that enables nonblocking mode for the file. If this bit is set, |
3152 | @code{read} requests on the file can return immediately with a failure | |
3153 | status if there is no input immediately available, instead of blocking. | |
3154 | Likewise, @code{write} requests can also return immediately with a | |
3155 | failure status if the output can't be written immediately. | |
3156 | ||
3157 | Note that the @code{O_NONBLOCK} flag is overloaded as both an I/O | |
3158 | operating mode and a file name translation flag; @pxref{Open-time Flags}. | |
3159 | @end deftypevr | |
3160 | ||
3161 | @comment fcntl.h | |
3162 | @comment BSD | |
3163 | @deftypevr Macro int O_NDELAY | |
3164 | This is an obsolete name for @code{O_NONBLOCK}, provided for | |
3165 | compatibility with BSD. It is not defined by the POSIX.1 standard. | |
3166 | @end deftypevr | |
3167 | ||
3168 | The remaining operating modes are BSD and GNU extensions. They exist only | |
3169 | on some systems. On other systems, these macros are not defined. | |
3170 | ||
3171 | @comment fcntl.h | |
3172 | @comment BSD | |
3173 | @deftypevr Macro int O_ASYNC | |
3174 | The bit that enables asynchronous input mode. If set, then @code{SIGIO} | |
3175 | signals will be generated when input is available. @xref{Interrupt Input}. | |
3176 | ||
3177 | Asynchronous input mode is a BSD feature. | |
3178 | @end deftypevr | |
3179 | ||
3180 | @comment fcntl.h | |
3181 | @comment BSD | |
3182 | @deftypevr Macro int O_FSYNC | |
3183 | The bit that enables synchronous writing for the file. If set, each | |
3184 | @code{write} call will make sure the data is reliably stored on disk before | |
3185 | returning. @c !!! xref fsync | |
3186 | ||
3187 | Synchronous writing is a BSD feature. | |
3188 | @end deftypevr | |
3189 | ||
3190 | @comment fcntl.h | |
3191 | @comment BSD | |
3192 | @deftypevr Macro int O_SYNC | |
3193 | This is another name for @code{O_FSYNC}. They have the same value. | |
3194 | @end deftypevr | |
3195 | ||
3196 | @comment fcntl.h | |
3197 | @comment GNU | |
3198 | @deftypevr Macro int O_NOATIME | |
3199 | If this bit is set, @code{read} will not update the access time of the | |
3200 | file. @xref{File Times}. This is used by programs that do backups, so | |
3201 | that backing a file up does not count as reading it. | |
3202 | Only the owner of the file or the superuser may use this bit. | |
3203 | ||
3204 | This is a GNU extension. | |
3205 | @end deftypevr | |
3206 | ||
3207 | @node Getting File Status Flags | |
3208 | @subsection Getting and Setting File Status Flags | |
3209 | ||
3210 | The @code{fcntl} function can fetch or change file status flags. | |
3211 | ||
3212 | @comment fcntl.h | |
3213 | @comment POSIX.1 | |
3214 | @deftypevr Macro int F_GETFL | |
3215 | This macro is used as the @var{command} argument to @code{fcntl}, to | |
3216 | read the file status flags for the open file with descriptor | |
3217 | @var{filedes}. | |
3218 | ||
3219 | The normal return value from @code{fcntl} with this command is a | |
3220 | nonnegative number which can be interpreted as the bitwise OR of the | |
3221 | individual flags. Since the file access modes are not single-bit values, | |
3222 | you can mask off other bits in the returned flags with @code{O_ACCMODE} | |
3223 | to compare them. | |
3224 | ||
07435eb4 | 3225 | In case of an error, @code{fcntl} returns @math{-1}. The following |
28f540f4 RM |
3226 | @code{errno} error conditions are defined for this command: |
3227 | ||
3228 | @table @code | |
3229 | @item EBADF | |
3230 | The @var{filedes} argument is invalid. | |
3231 | @end table | |
3232 | @end deftypevr | |
3233 | ||
3234 | @comment fcntl.h | |
3235 | @comment POSIX.1 | |
3236 | @deftypevr Macro int F_SETFL | |
3237 | This macro is used as the @var{command} argument to @code{fcntl}, to set | |
3238 | the file status flags for the open file corresponding to the | |
3239 | @var{filedes} argument. This command requires a third @code{int} | |
3240 | argument to specify the new flags, so the call looks like this: | |
3241 | ||
3242 | @smallexample | |
3243 | fcntl (@var{filedes}, F_SETFL, @var{new-flags}) | |
3244 | @end smallexample | |
3245 | ||
3246 | You can't change the access mode for the file in this way; that is, | |
3247 | whether the file descriptor was opened for reading or writing. | |
3248 | ||
3249 | The normal return value from @code{fcntl} with this command is an | |
07435eb4 | 3250 | unspecified value other than @math{-1}, which indicates an error. The |
28f540f4 RM |
3251 | error conditions are the same as for the @code{F_GETFL} command. |
3252 | @end deftypevr | |
3253 | ||
3254 | If you want to modify the file status flags, you should get the current | |
3255 | flags with @code{F_GETFL} and modify the value. Don't assume that the | |
3256 | flags listed here are the only ones that are implemented; your program | |
3257 | may be run years from now and more flags may exist then. For example, | |
3258 | here is a function to set or clear the flag @code{O_NONBLOCK} without | |
3259 | altering any other flags: | |
3260 | ||
3261 | @smallexample | |
3262 | @group | |
3263 | /* @r{Set the @code{O_NONBLOCK} flag of @var{desc} if @var{value} is nonzero,} | |
3264 | @r{or clear the flag if @var{value} is 0.} | |
2c6fe0bd | 3265 | @r{Return 0 on success, or -1 on error with @code{errno} set.} */ |
28f540f4 RM |
3266 | |
3267 | int | |
3268 | set_nonblock_flag (int desc, int value) | |
3269 | @{ | |
3270 | int oldflags = fcntl (desc, F_GETFL, 0); | |
3271 | /* @r{If reading the flags failed, return error indication now.} */ | |
3272 | if (oldflags == -1) | |
3273 | return -1; | |
3274 | /* @r{Set just the flag we want to set.} */ | |
3275 | if (value != 0) | |
3276 | oldflags |= O_NONBLOCK; | |
3277 | else | |
3278 | oldflags &= ~O_NONBLOCK; | |
3279 | /* @r{Store modified flag word in the descriptor.} */ | |
3280 | return fcntl (desc, F_SETFL, oldflags); | |
3281 | @} | |
3282 | @end group | |
3283 | @end smallexample | |
3284 | ||
3285 | @node File Locks | |
3286 | @section File Locks | |
3287 | ||
3288 | @cindex file locks | |
3289 | @cindex record locking | |
3290 | The remaining @code{fcntl} commands are used to support @dfn{record | |
3291 | locking}, which permits multiple cooperating programs to prevent each | |
3292 | other from simultaneously accessing parts of a file in error-prone | |
3293 | ways. | |
3294 | ||
3295 | @cindex exclusive lock | |
3296 | @cindex write lock | |
3297 | An @dfn{exclusive} or @dfn{write} lock gives a process exclusive access | |
3298 | for writing to the specified part of the file. While a write lock is in | |
3299 | place, no other process can lock that part of the file. | |
3300 | ||
3301 | @cindex shared lock | |
3302 | @cindex read lock | |
3303 | A @dfn{shared} or @dfn{read} lock prohibits any other process from | |
3304 | requesting a write lock on the specified part of the file. However, | |
3305 | other processes can request read locks. | |
3306 | ||
3307 | The @code{read} and @code{write} functions do not actually check to see | |
3308 | whether there are any locks in place. If you want to implement a | |
3309 | locking protocol for a file shared by multiple processes, your application | |
3310 | must do explicit @code{fcntl} calls to request and clear locks at the | |
3311 | appropriate points. | |
3312 | ||
3313 | Locks are associated with processes. A process can only have one kind | |
3314 | of lock set for each byte of a given file. When any file descriptor for | |
3315 | that file is closed by the process, all of the locks that process holds | |
3316 | on that file are released, even if the locks were made using other | |
3317 | descriptors that remain open. Likewise, locks are released when a | |
3318 | process exits, and are not inherited by child processes created using | |
3319 | @code{fork} (@pxref{Creating a Process}). | |
3320 | ||
3321 | When making a lock, use a @code{struct flock} to specify what kind of | |
3322 | lock and where. This data type and the associated macros for the | |
3323 | @code{fcntl} function are declared in the header file @file{fcntl.h}. | |
3324 | @pindex fcntl.h | |
3325 | ||
3326 | @comment fcntl.h | |
3327 | @comment POSIX.1 | |
3328 | @deftp {Data Type} {struct flock} | |
3329 | This structure is used with the @code{fcntl} function to describe a file | |
3330 | lock. It has these members: | |
3331 | ||
3332 | @table @code | |
3333 | @item short int l_type | |
3334 | Specifies the type of the lock; one of @code{F_RDLCK}, @code{F_WRLCK}, or | |
3335 | @code{F_UNLCK}. | |
3336 | ||
3337 | @item short int l_whence | |
3338 | This corresponds to the @var{whence} argument to @code{fseek} or | |
3339 | @code{lseek}, and specifies what the offset is relative to. Its value | |
3340 | can be one of @code{SEEK_SET}, @code{SEEK_CUR}, or @code{SEEK_END}. | |
3341 | ||
3342 | @item off_t l_start | |
3343 | This specifies the offset of the start of the region to which the lock | |
3344 | applies, and is given in bytes relative to the point specified by | |
3345 | @code{l_whence} member. | |
3346 | ||
3347 | @item off_t l_len | |
3348 | This specifies the length of the region to be locked. A value of | |
3349 | @code{0} is treated specially; it means the region extends to the end of | |
3350 | the file. | |
3351 | ||
3352 | @item pid_t l_pid | |
3353 | This field is the process ID (@pxref{Process Creation Concepts}) of the | |
3354 | process holding the lock. It is filled in by calling @code{fcntl} with | |
3355 | the @code{F_GETLK} command, but is ignored when making a lock. | |
3356 | @end table | |
3357 | @end deftp | |
3358 | ||
3359 | @comment fcntl.h | |
3360 | @comment POSIX.1 | |
3361 | @deftypevr Macro int F_GETLK | |
3362 | This macro is used as the @var{command} argument to @code{fcntl}, to | |
3363 | specify that it should get information about a lock. This command | |
3364 | requires a third argument of type @w{@code{struct flock *}} to be passed | |
3365 | to @code{fcntl}, so that the form of the call is: | |
3366 | ||
3367 | @smallexample | |
3368 | fcntl (@var{filedes}, F_GETLK, @var{lockp}) | |
3369 | @end smallexample | |
3370 | ||
3371 | If there is a lock already in place that would block the lock described | |
3372 | by the @var{lockp} argument, information about that lock overwrites | |
3373 | @code{*@var{lockp}}. Existing locks are not reported if they are | |
3374 | compatible with making a new lock as specified. Thus, you should | |
3375 | specify a lock type of @code{F_WRLCK} if you want to find out about both | |
3376 | read and write locks, or @code{F_RDLCK} if you want to find out about | |
3377 | write locks only. | |
3378 | ||
3379 | There might be more than one lock affecting the region specified by the | |
3380 | @var{lockp} argument, but @code{fcntl} only returns information about | |
3381 | one of them. The @code{l_whence} member of the @var{lockp} structure is | |
3382 | set to @code{SEEK_SET} and the @code{l_start} and @code{l_len} fields | |
3383 | set to identify the locked region. | |
3384 | ||
3385 | If no lock applies, the only change to the @var{lockp} structure is to | |
3386 | update the @code{l_type} to a value of @code{F_UNLCK}. | |
3387 | ||
3388 | The normal return value from @code{fcntl} with this command is an | |
07435eb4 | 3389 | unspecified value other than @math{-1}, which is reserved to indicate an |
28f540f4 RM |
3390 | error. The following @code{errno} error conditions are defined for |
3391 | this command: | |
3392 | ||
3393 | @table @code | |
3394 | @item EBADF | |
3395 | The @var{filedes} argument is invalid. | |
3396 | ||
3397 | @item EINVAL | |
3398 | Either the @var{lockp} argument doesn't specify valid lock information, | |
3399 | or the file associated with @var{filedes} doesn't support locks. | |
3400 | @end table | |
3401 | @end deftypevr | |
3402 | ||
3403 | @comment fcntl.h | |
3404 | @comment POSIX.1 | |
3405 | @deftypevr Macro int F_SETLK | |
3406 | This macro is used as the @var{command} argument to @code{fcntl}, to | |
3407 | specify that it should set or clear a lock. This command requires a | |
3408 | third argument of type @w{@code{struct flock *}} to be passed to | |
3409 | @code{fcntl}, so that the form of the call is: | |
3410 | ||
3411 | @smallexample | |
3412 | fcntl (@var{filedes}, F_SETLK, @var{lockp}) | |
3413 | @end smallexample | |
3414 | ||
3415 | If the process already has a lock on any part of the region, the old lock | |
3416 | on that part is replaced with the new lock. You can remove a lock | |
3417 | by specifying a lock type of @code{F_UNLCK}. | |
3418 | ||
3419 | If the lock cannot be set, @code{fcntl} returns immediately with a value | |
07435eb4 | 3420 | of @math{-1}. This function does not block waiting for other processes |
28f540f4 | 3421 | to release locks. If @code{fcntl} succeeds, it return a value other |
07435eb4 | 3422 | than @math{-1}. |
28f540f4 RM |
3423 | |
3424 | The following @code{errno} error conditions are defined for this | |
3425 | function: | |
3426 | ||
3427 | @table @code | |
3428 | @item EAGAIN | |
3429 | @itemx EACCES | |
3430 | The lock cannot be set because it is blocked by an existing lock on the | |
3431 | file. Some systems use @code{EAGAIN} in this case, and other systems | |
3432 | use @code{EACCES}; your program should treat them alike, after | |
3433 | @code{F_SETLK}. (The GNU system always uses @code{EAGAIN}.) | |
3434 | ||
3435 | @item EBADF | |
3436 | Either: the @var{filedes} argument is invalid; you requested a read lock | |
3437 | but the @var{filedes} is not open for read access; or, you requested a | |
3438 | write lock but the @var{filedes} is not open for write access. | |
3439 | ||
3440 | @item EINVAL | |
3441 | Either the @var{lockp} argument doesn't specify valid lock information, | |
3442 | or the file associated with @var{filedes} doesn't support locks. | |
3443 | ||
3444 | @item ENOLCK | |
3445 | The system has run out of file lock resources; there are already too | |
3446 | many file locks in place. | |
3447 | ||
3448 | Well-designed file systems never report this error, because they have no | |
3449 | limitation on the number of locks. However, you must still take account | |
3450 | of the possibility of this error, as it could result from network access | |
3451 | to a file system on another machine. | |
3452 | @end table | |
3453 | @end deftypevr | |
3454 | ||
3455 | @comment fcntl.h | |
3456 | @comment POSIX.1 | |
3457 | @deftypevr Macro int F_SETLKW | |
3458 | This macro is used as the @var{command} argument to @code{fcntl}, to | |
3459 | specify that it should set or clear a lock. It is just like the | |
3460 | @code{F_SETLK} command, but causes the process to block (or wait) | |
3461 | until the request can be specified. | |
3462 | ||
3463 | This command requires a third argument of type @code{struct flock *}, as | |
3464 | for the @code{F_SETLK} command. | |
3465 | ||
3466 | The @code{fcntl} return values and errors are the same as for the | |
3467 | @code{F_SETLK} command, but these additional @code{errno} error conditions | |
3468 | are defined for this command: | |
3469 | ||
3470 | @table @code | |
3471 | @item EINTR | |
3472 | The function was interrupted by a signal while it was waiting. | |
3473 | @xref{Interrupted Primitives}. | |
3474 | ||
3475 | @item EDEADLK | |
3476 | The specified region is being locked by another process. But that | |
3477 | process is waiting to lock a region which the current process has | |
3478 | locked, so waiting for the lock would result in deadlock. The system | |
3479 | does not guarantee that it will detect all such conditions, but it lets | |
3480 | you know if it notices one. | |
3481 | @end table | |
3482 | @end deftypevr | |
3483 | ||
3484 | ||
3485 | The following macros are defined for use as values for the @code{l_type} | |
3486 | member of the @code{flock} structure. The values are integer constants. | |
3487 | ||
3488 | @table @code | |
3489 | @comment fcntl.h | |
3490 | @comment POSIX.1 | |
3491 | @vindex F_RDLCK | |
3492 | @item F_RDLCK | |
3493 | This macro is used to specify a read (or shared) lock. | |
3494 | ||
3495 | @comment fcntl.h | |
3496 | @comment POSIX.1 | |
3497 | @vindex F_WRLCK | |
3498 | @item F_WRLCK | |
3499 | This macro is used to specify a write (or exclusive) lock. | |
3500 | ||
3501 | @comment fcntl.h | |
3502 | @comment POSIX.1 | |
3503 | @vindex F_UNLCK | |
3504 | @item F_UNLCK | |
3505 | This macro is used to specify that the region is unlocked. | |
3506 | @end table | |
3507 | ||
3508 | As an example of a situation where file locking is useful, consider a | |
3509 | program that can be run simultaneously by several different users, that | |
3510 | logs status information to a common file. One example of such a program | |
3511 | might be a game that uses a file to keep track of high scores. Another | |
3512 | example might be a program that records usage or accounting information | |
3513 | for billing purposes. | |
3514 | ||
3515 | Having multiple copies of the program simultaneously writing to the | |
3516 | file could cause the contents of the file to become mixed up. But | |
3517 | you can prevent this kind of problem by setting a write lock on the | |
2c6fe0bd | 3518 | file before actually writing to the file. |
28f540f4 RM |
3519 | |
3520 | If the program also needs to read the file and wants to make sure that | |
3521 | the contents of the file are in a consistent state, then it can also use | |
3522 | a read lock. While the read lock is set, no other process can lock | |
3523 | that part of the file for writing. | |
3524 | ||
3525 | @c ??? This section could use an example program. | |
3526 | ||
3527 | Remember that file locks are only a @emph{voluntary} protocol for | |
3528 | controlling access to a file. There is still potential for access to | |
3529 | the file by programs that don't use the lock protocol. | |
3530 | ||
3531 | @node Interrupt Input | |
3532 | @section Interrupt-Driven Input | |
3533 | ||
3534 | @cindex interrupt-driven input | |
3535 | If you set the @code{O_ASYNC} status flag on a file descriptor | |
3536 | (@pxref{File Status Flags}), a @code{SIGIO} signal is sent whenever | |
3537 | input or output becomes possible on that file descriptor. The process | |
3538 | or process group to receive the signal can be selected by using the | |
3539 | @code{F_SETOWN} command to the @code{fcntl} function. If the file | |
3540 | descriptor is a socket, this also selects the recipient of @code{SIGURG} | |
3541 | signals that are delivered when out-of-band data arrives on that socket; | |
3542 | see @ref{Out-of-Band Data}. (@code{SIGURG} is sent in any situation | |
3543 | where @code{select} would report the socket as having an ``exceptional | |
3544 | condition''. @xref{Waiting for I/O}.) | |
3545 | ||
3546 | If the file descriptor corresponds to a terminal device, then @code{SIGIO} | |
2c6fe0bd | 3547 | signals are sent to the foreground process group of the terminal. |
28f540f4 RM |
3548 | @xref{Job Control}. |
3549 | ||
3550 | @pindex fcntl.h | |
3551 | The symbols in this section are defined in the header file | |
3552 | @file{fcntl.h}. | |
3553 | ||
3554 | @comment fcntl.h | |
3555 | @comment BSD | |
3556 | @deftypevr Macro int F_GETOWN | |
3557 | This macro is used as the @var{command} argument to @code{fcntl}, to | |
3558 | specify that it should get information about the process or process | |
3559 | group to which @code{SIGIO} signals are sent. (For a terminal, this is | |
3560 | actually the foreground process group ID, which you can get using | |
3561 | @code{tcgetpgrp}; see @ref{Terminal Access Functions}.) | |
3562 | ||
3563 | The return value is interpreted as a process ID; if negative, its | |
3564 | absolute value is the process group ID. | |
3565 | ||
3566 | The following @code{errno} error condition is defined for this command: | |
3567 | ||
3568 | @table @code | |
3569 | @item EBADF | |
3570 | The @var{filedes} argument is invalid. | |
3571 | @end table | |
3572 | @end deftypevr | |
3573 | ||
3574 | @comment fcntl.h | |
3575 | @comment BSD | |
3576 | @deftypevr Macro int F_SETOWN | |
3577 | This macro is used as the @var{command} argument to @code{fcntl}, to | |
3578 | specify that it should set the process or process group to which | |
3579 | @code{SIGIO} signals are sent. This command requires a third argument | |
3580 | of type @code{pid_t} to be passed to @code{fcntl}, so that the form of | |
3581 | the call is: | |
3582 | ||
3583 | @smallexample | |
3584 | fcntl (@var{filedes}, F_SETOWN, @var{pid}) | |
3585 | @end smallexample | |
3586 | ||
3587 | The @var{pid} argument should be a process ID. You can also pass a | |
3588 | negative number whose absolute value is a process group ID. | |
3589 | ||
07435eb4 | 3590 | The return value from @code{fcntl} with this command is @math{-1} |
28f540f4 RM |
3591 | in case of error and some other value if successful. The following |
3592 | @code{errno} error conditions are defined for this command: | |
3593 | ||
3594 | @table @code | |
3595 | @item EBADF | |
3596 | The @var{filedes} argument is invalid. | |
3597 | ||
3598 | @item ESRCH | |
3599 | There is no process or process group corresponding to @var{pid}. | |
3600 | @end table | |
3601 | @end deftypevr | |
3602 | ||
3603 | @c ??? This section could use an example program. | |
07435eb4 UD |
3604 | |
3605 | @node IOCTLs | |
3606 | @section Generic I/O Control operations | |
3607 | @cindex generic i/o control operations | |
3608 | @cindex IOCTLs | |
3609 | ||
3610 | The GNU system can handle most input/output operations on many different | |
3611 | devices and objects in terms of a few file primitives - @code{read}, | |
3612 | @code{write} and @code{lseek}. However, most devices also have a few | |
3613 | peculiar operations which do not fit into this model. Such as: | |
3614 | ||
3615 | @itemize @bullet | |
3616 | ||
3617 | @item | |
3618 | Changing the character font used on a terminal. | |
3619 | ||
3620 | @item | |
3621 | Telling a magnetic tape system to rewind or fast forward. (Since they | |
3622 | cannot move in byte increments, @code{lseek} is inapplicable). | |
3623 | ||
3624 | @item | |
3625 | Ejecting a disk from a drive. | |
3626 | ||
3627 | @item | |
3628 | Playing an audio track from a CD-ROM drive. | |
3629 | ||
3630 | @item | |
3631 | Maintaining routing tables for a network. | |
3632 | ||
3633 | @end itemize | |
3634 | ||
3635 | Although some such objects such as sockets and terminals | |
3636 | @footnote{Actually, the terminal-specific functions are implemented with | |
3637 | IOCTLs on many platforms.} have special functions of their own, it would | |
3638 | not be practical to create functions for all these cases. | |
3639 | ||
3640 | Instead these minor operations, known as @dfn{IOCTL}s, are assigned code | |
3641 | numbers and multiplexed through the @code{ioctl} function, defined in | |
3642 | @code{sys/ioctl.h}. The code numbers themselves are defined in many | |
3643 | different headers. | |
3644 | ||
b2e3d177 UD |
3645 | @comment sys/ioctl.h |
3646 | @comment BSD | |
07435eb4 UD |
3647 | @deftypefun int ioctl (int @var{filedes}, int @var{command}, @dots{}) |
3648 | ||
3649 | The @code{ioctl} function performs the generic I/O operation | |
3650 | @var{command} on @var{filedes}. | |
3651 | ||
3652 | A third argument is usually present, either a single number or a pointer | |
3653 | to a structure. The meaning of this argument, the returned value, and | |
3654 | any error codes depends upon the command used. Often @math{-1} is | |
3655 | returned for a failure. | |
3656 | ||
3657 | @end deftypefun | |
3658 | ||
3659 | On some systems, IOCTLs used by different devices share the same numbers. | |
3660 | Thus, although use of an inappropriate IOCTL @emph{usually} only produces | |
3661 | an error, you should not attempt to use device-specific IOCTLs on an | |
3662 | unknown device. | |
3663 | ||
3664 | Most IOCTLs are OS-specific and/or only used in special system utilities, | |
3665 | and are thus beyond the scope of this document. For an example of the use | |
8b7fb588 | 3666 | of an IOCTL, see @ref{Out-of-Band Data}. |