]>
Commit | Line | Data |
---|---|---|
5ce8f203 UD |
1 | @node Resource Usage And Limitation, Non-Local Exits, Date and Time, Top |
2 | @c %MENU% Functions for examining resource usage and getting and setting limits | |
3 | @chapter Resource Usage And Limitation | |
4 | This chapter describes functions for examining how much of various kinds of | |
5 | resources (CPU time, memory, etc.) a process has used and getting and setting | |
6 | limits on future usage. | |
7 | ||
8 | @menu | |
9 | * Resource Usage:: Measuring various resources used. | |
10 | * Limits on Resources:: Specifying limits on resource usage. | |
11 | * Priority:: Reading or setting process run priority. | |
b642f101 UD |
12 | * Memory Resources:: Querying memory available resources. |
13 | * Processor Resources:: Learn about the processors available. | |
5ce8f203 UD |
14 | @end menu |
15 | ||
16 | ||
17 | @node Resource Usage | |
18 | @section Resource Usage | |
19 | ||
20 | @pindex sys/resource.h | |
21 | The function @code{getrusage} and the data type @code{struct rusage} | |
22 | are used to examine the resource usage of a process. They are declared | |
23 | in @file{sys/resource.h}. | |
24 | ||
25 | @comment sys/resource.h | |
26 | @comment BSD | |
27 | @deftypefun int getrusage (int @var{processes}, struct rusage *@var{rusage}) | |
28 | This function reports resource usage totals for processes specified by | |
29 | @var{processes}, storing the information in @code{*@var{rusage}}. | |
30 | ||
31 | In most systems, @var{processes} has only two valid values: | |
32 | ||
33 | @table @code | |
34 | @comment sys/resource.h | |
35 | @comment BSD | |
36 | @item RUSAGE_SELF | |
37 | Just the current process. | |
38 | ||
39 | @comment sys/resource.h | |
40 | @comment BSD | |
41 | @item RUSAGE_CHILDREN | |
42 | All child processes (direct and indirect) that have already terminated. | |
43 | @end table | |
44 | ||
45 | In the GNU system, you can also inquire about a particular child process | |
46 | by specifying its process ID. | |
47 | ||
48 | The return value of @code{getrusage} is zero for success, and @code{-1} | |
49 | for failure. | |
50 | ||
51 | @table @code | |
52 | @item EINVAL | |
53 | The argument @var{processes} is not valid. | |
54 | @end table | |
55 | @end deftypefun | |
56 | ||
57 | One way of getting resource usage for a particular child process is with | |
58 | the function @code{wait4}, which returns totals for a child when it | |
59 | terminates. @xref{BSD Wait Functions}. | |
60 | ||
61 | @comment sys/resource.h | |
62 | @comment BSD | |
63 | @deftp {Data Type} {struct rusage} | |
64 | This data type stores various resource usage statistics. It has the | |
65 | following members, and possibly others: | |
66 | ||
67 | @table @code | |
68 | @item struct timeval ru_utime | |
69 | Time spent executing user instructions. | |
70 | ||
71 | @item struct timeval ru_stime | |
72 | Time spent in operating system code on behalf of @var{processes}. | |
73 | ||
74 | @item long int ru_maxrss | |
75 | The maximum resident set size used, in kilobytes. That is, the maximum | |
76 | number of kilobytes of physical memory that @var{processes} used | |
77 | simultaneously. | |
78 | ||
79 | @item long int ru_ixrss | |
80 | An integral value expressed in kilobytes times ticks of execution, which | |
81 | indicates the amount of memory used by text that was shared with other | |
82 | processes. | |
83 | ||
84 | @item long int ru_idrss | |
85 | An integral value expressed the same way, which is the amount of | |
86 | unshared memory used for data. | |
87 | ||
88 | @item long int ru_isrss | |
89 | An integral value expressed the same way, which is the amount of | |
90 | unshared memory used for stack space. | |
91 | ||
92 | @item long int ru_minflt | |
93 | The number of page faults which were serviced without requiring any I/O. | |
94 | ||
95 | @item long int ru_majflt | |
96 | The number of page faults which were serviced by doing I/O. | |
97 | ||
98 | @item long int ru_nswap | |
99 | The number of times @var{processes} was swapped entirely out of main memory. | |
100 | ||
101 | @item long int ru_inblock | |
102 | The number of times the file system had to read from the disk on behalf | |
103 | of @var{processes}. | |
104 | ||
105 | @item long int ru_oublock | |
106 | The number of times the file system had to write to the disk on behalf | |
107 | of @var{processes}. | |
108 | ||
109 | @item long int ru_msgsnd | |
110 | Number of IPC messages sent. | |
111 | ||
112 | @item long int ru_msgrcv | |
113 | Number of IPC messages received. | |
114 | ||
115 | @item long int ru_nsignals | |
116 | Number of signals received. | |
117 | ||
118 | @item long int ru_nvcsw | |
119 | The number of times @var{processes} voluntarily invoked a context switch | |
120 | (usually to wait for some service). | |
121 | ||
122 | @item long int ru_nivcsw | |
123 | The number of times an involuntary context switch took place (because | |
124 | a time slice expired, or another process of higher priority was | |
125 | scheduled). | |
126 | @end table | |
127 | @end deftp | |
128 | ||
129 | @code{vtimes} is a historical function that does some of what | |
130 | @code{getrusage} does. @code{getrusage} is a better choice. | |
131 | ||
132 | @code{vtimes} and its @code{vtimes} data structure are declared in | |
133 | @file{sys/vtimes.h}. | |
134 | @pindex sys/vtimes.h | |
135 | @comment vtimes.h | |
136 | ||
137 | @deftypefun int vtimes (struct vtimes @var{current}, struct vtimes @var{child}) | |
138 | ||
139 | @code{vtimes} reports resource usage totals for a process. | |
140 | ||
141 | If @var{current} is non-null, @code{vtimes} stores resource usage totals for | |
142 | the invoking process alone in the structure to which it points. If | |
143 | @var{child} is non-null, @code{vtimes} stores resource usage totals for all | |
144 | past children (which have terminated) of the invoking process in the structure | |
145 | to which it points. | |
146 | ||
147 | @deftp {Data Type} {struct vtimes} | |
148 | This data type contains information about the resource usage of a process. | |
149 | Each member corresponds to a member of the @code{struct rusage} data type | |
150 | described above. | |
151 | ||
152 | @table @code | |
153 | @item vm_utime | |
154 | User CPU time. Analogous to @code{ru_utime} in @code{struct rusage} | |
155 | @item vm_stime | |
156 | System CPU time. Analogous to @code{ru_stime} in @code{struct rusage} | |
157 | @item vm_idsrss | |
158 | Data and stack memory. The sum of the values that would be reported as | |
159 | @code{ru_idrss} and @code{ru_isrss} in @code{struct rusage} | |
160 | @item vm_ixrss | |
161 | Shared memory. Analogous to @code{ru_ixrss} in @code{struct rusage} | |
162 | @item vm_maxrss | |
163 | Maximent resident set size. Analogous to @code{ru_maxrss} in | |
164 | @code{struct rusage} | |
165 | @item vm_majflt | |
166 | Major page faults. Analogous to @code{ru_majflt} in @code{struct rusage} | |
167 | @item vm_minflt | |
168 | Minor page faults. Analogous to @code{ru_minflt} in @code{struct rusage} | |
169 | @item vm_nswap | |
170 | Swap count. Analogous to @code{ru_nswap} in @code{struct rusage} | |
171 | @item vm_inblk | |
172 | Disk reads. Analogous to @code{ru_inblk} in @code{struct rusage} | |
173 | @item vm_oublk | |
174 | Disk writes. Analogous to @code{ru_oublk} in @code{struct rusage} | |
175 | @end table | |
176 | @end deftp | |
177 | ||
178 | ||
179 | The return value is zero if the function succeeds; @code{-1} otherwise. | |
180 | ||
181 | ||
182 | ||
183 | @end deftypefun | |
184 | An additional historical function for examining resource usage, | |
185 | @code{vtimes}, is supported but not documented here. It is declared in | |
186 | @file{sys/vtimes.h}. | |
187 | ||
188 | @node Limits on Resources | |
189 | @section Limiting Resource Usage | |
190 | @cindex resource limits | |
191 | @cindex limits on resource usage | |
192 | @cindex usage limits | |
193 | ||
194 | You can specify limits for the resource usage of a process. When the | |
195 | process tries to exceed a limit, it may get a signal, or the system call | |
196 | by which it tried to do so may fail, depending on the resource. Each | |
197 | process initially inherits its limit values from its parent, but it can | |
198 | subsequently change them. | |
199 | ||
200 | There are two per-process limits associated with a resource: | |
201 | @cindex limit | |
202 | ||
203 | @table @dfn | |
204 | @item current limit | |
205 | The current limit is the value the system will not allow usage to | |
206 | exceed. It is also called the ``soft limit'' because the process being | |
207 | limited can generally raise the current limit at will. | |
208 | @cindex current limit | |
209 | @cindex soft limit | |
210 | ||
211 | @item maximum limit | |
212 | The maximum limit is the maximum value to which a process is allowed to | |
213 | set its current limit. It is also called the ``hard limit'' because | |
214 | there is no way for a process to get around it. A process may lower | |
215 | its own maximum limit, but only the superuser may increase a maximum | |
216 | limit. | |
217 | @cindex maximum limit | |
218 | @cindex hard limit | |
219 | @end table | |
220 | ||
221 | @pindex sys/resource.h | |
222 | The symbols for use with @code{getrlimit}, @code{setrlimit}, | |
0bc93a2f | 223 | @code{getrlimit64}, and @code{setrlimit64} are defined in |
5ce8f203 UD |
224 | @file{sys/resource.h}. |
225 | ||
226 | @comment sys/resource.h | |
227 | @comment BSD | |
228 | @deftypefun int getrlimit (int @var{resource}, struct rlimit *@var{rlp}) | |
229 | Read the current and maximum limits for the resource @var{resource} | |
230 | and store them in @code{*@var{rlp}}. | |
231 | ||
232 | The return value is @code{0} on success and @code{-1} on failure. The | |
233 | only possible @code{errno} error condition is @code{EFAULT}. | |
234 | ||
235 | When the sources are compiled with @code{_FILE_OFFSET_BITS == 64} on a | |
236 | 32-bit system this function is in fact @code{getrlimit64}. Thus, the | |
237 | LFS interface transparently replaces the old interface. | |
238 | @end deftypefun | |
239 | ||
240 | @comment sys/resource.h | |
241 | @comment Unix98 | |
242 | @deftypefun int getrlimit64 (int @var{resource}, struct rlimit64 *@var{rlp}) | |
243 | This function is similar to @code{getrlimit} but its second parameter is | |
244 | a pointer to a variable of type @code{struct rlimit64}, which allows it | |
245 | to read values which wouldn't fit in the member of a @code{struct | |
246 | rlimit}. | |
247 | ||
248 | If the sources are compiled with @code{_FILE_OFFSET_BITS == 64} on a | |
249 | 32-bit machine, this function is available under the name | |
250 | @code{getrlimit} and so transparently replaces the old interface. | |
251 | @end deftypefun | |
252 | ||
253 | @comment sys/resource.h | |
254 | @comment BSD | |
255 | @deftypefun int setrlimit (int @var{resource}, const struct rlimit *@var{rlp}) | |
256 | Store the current and maximum limits for the resource @var{resource} | |
257 | in @code{*@var{rlp}}. | |
258 | ||
259 | The return value is @code{0} on success and @code{-1} on failure. The | |
260 | following @code{errno} error condition is possible: | |
261 | ||
262 | @table @code | |
263 | @item EPERM | |
264 | @itemize @bullet | |
265 | @item | |
266 | The process tried to raise a current limit beyond the maximum limit. | |
267 | ||
268 | @item | |
269 | The process tried to raise a maximum limit, but is not superuser. | |
270 | @end itemize | |
271 | @end table | |
272 | ||
273 | When the sources are compiled with @code{_FILE_OFFSET_BITS == 64} on a | |
274 | 32-bit system this function is in fact @code{setrlimit64}. Thus, the | |
275 | LFS interface transparently replaces the old interface. | |
276 | @end deftypefun | |
277 | ||
278 | @comment sys/resource.h | |
279 | @comment Unix98 | |
280 | @deftypefun int setrlimit64 (int @var{resource}, const struct rlimit64 *@var{rlp}) | |
281 | This function is similar to @code{setrlimit} but its second parameter is | |
282 | a pointer to a variable of type @code{struct rlimit64} which allows it | |
283 | to set values which wouldn't fit in the member of a @code{struct | |
284 | rlimit}. | |
285 | ||
286 | If the sources are compiled with @code{_FILE_OFFSET_BITS == 64} on a | |
287 | 32-bit machine this function is available under the name | |
288 | @code{setrlimit} and so transparently replaces the old interface. | |
289 | @end deftypefun | |
290 | ||
291 | @comment sys/resource.h | |
292 | @comment BSD | |
293 | @deftp {Data Type} {struct rlimit} | |
294 | This structure is used with @code{getrlimit} to receive limit values, | |
295 | and with @code{setrlimit} to specify limit values for a particular process | |
296 | and resource. It has two fields: | |
297 | ||
298 | @table @code | |
299 | @item rlim_t rlim_cur | |
300 | The current limit | |
301 | ||
302 | @item rlim_t rlim_max | |
303 | The maximum limit. | |
304 | @end table | |
305 | ||
306 | For @code{getrlimit}, the structure is an output; it receives the current | |
307 | values. For @code{setrlimit}, it specifies the new values. | |
308 | @end deftp | |
309 | ||
310 | For the LFS functions a similar type is defined in @file{sys/resource.h}. | |
311 | ||
312 | @comment sys/resource.h | |
313 | @comment Unix98 | |
314 | @deftp {Data Type} {struct rlimit64} | |
315 | This structure is analogous to the @code{rlimit} structure above, but | |
316 | its components have wider ranges. It has two fields: | |
317 | ||
318 | @table @code | |
319 | @item rlim64_t rlim_cur | |
320 | This is analogous to @code{rlimit.rlim_cur}, but with a different type. | |
321 | ||
322 | @item rlim64_t rlim_max | |
323 | This is analogous to @code{rlimit.rlim_max}, but with a different type. | |
324 | @end table | |
325 | ||
326 | @end deftp | |
327 | ||
328 | Here is a list of resources for which you can specify a limit. Memory | |
329 | and file sizes are measured in bytes. | |
330 | ||
331 | @table @code | |
332 | @comment sys/resource.h | |
333 | @comment BSD | |
334 | @item RLIMIT_CPU | |
335 | @vindex RLIMIT_CPU | |
336 | The maximum amount of CPU time the process can use. If it runs for | |
337 | longer than this, it gets a signal: @code{SIGXCPU}. The value is | |
338 | measured in seconds. @xref{Operation Error Signals}. | |
339 | ||
340 | @comment sys/resource.h | |
341 | @comment BSD | |
342 | @item RLIMIT_FSIZE | |
343 | @vindex RLIMIT_FSIZE | |
344 | The maximum size of file the process can create. Trying to write a | |
345 | larger file causes a signal: @code{SIGXFSZ}. @xref{Operation Error | |
346 | Signals}. | |
347 | ||
348 | @comment sys/resource.h | |
349 | @comment BSD | |
350 | @item RLIMIT_DATA | |
351 | @vindex RLIMIT_DATA | |
352 | The maximum size of data memory for the process. If the process tries | |
353 | to allocate data memory beyond this amount, the allocation function | |
354 | fails. | |
355 | ||
356 | @comment sys/resource.h | |
357 | @comment BSD | |
358 | @item RLIMIT_STACK | |
359 | @vindex RLIMIT_STACK | |
360 | The maximum stack size for the process. If the process tries to extend | |
361 | its stack past this size, it gets a @code{SIGSEGV} signal. | |
362 | @xref{Program Error Signals}. | |
363 | ||
364 | @comment sys/resource.h | |
365 | @comment BSD | |
366 | @item RLIMIT_CORE | |
367 | @vindex RLIMIT_CORE | |
368 | The maximum size core file that this process can create. If the process | |
369 | terminates and would dump a core file larger than this, then no core | |
370 | file is created. So setting this limit to zero prevents core files from | |
371 | ever being created. | |
372 | ||
373 | @comment sys/resource.h | |
374 | @comment BSD | |
375 | @item RLIMIT_RSS | |
376 | @vindex RLIMIT_RSS | |
377 | The maximum amount of physical memory that this process should get. | |
378 | This parameter is a guide for the system's scheduler and memory | |
379 | allocator; the system may give the process more memory when there is a | |
380 | surplus. | |
381 | ||
382 | @comment sys/resource.h | |
383 | @comment BSD | |
384 | @item RLIMIT_MEMLOCK | |
385 | The maximum amount of memory that can be locked into physical memory (so | |
386 | it will never be paged out). | |
387 | ||
388 | @comment sys/resource.h | |
389 | @comment BSD | |
390 | @item RLIMIT_NPROC | |
391 | The maximum number of processes that can be created with the same user ID. | |
392 | If you have reached the limit for your user ID, @code{fork} will fail | |
393 | with @code{EAGAIN}. @xref{Creating a Process}. | |
394 | ||
395 | @comment sys/resource.h | |
396 | @comment BSD | |
397 | @item RLIMIT_NOFILE | |
398 | @vindex RLIMIT_NOFILE | |
399 | @itemx RLIMIT_OFILE | |
400 | @vindex RLIMIT_OFILE | |
401 | The maximum number of files that the process can open. If it tries to | |
402 | open more files than this, its open attempt fails with @code{errno} | |
403 | @code{EMFILE}. @xref{Error Codes}. Not all systems support this limit; | |
404 | GNU does, and 4.4 BSD does. | |
405 | ||
406 | @comment sys/resource.h | |
407 | @comment Unix98 | |
408 | @item RLIMIT_AS | |
409 | @vindex RLIMIT_AS | |
410 | The maximum size of total memory that this process should get. If the | |
411 | process tries to allocate more memory beyond this amount with, for | |
412 | example, @code{brk}, @code{malloc}, @code{mmap} or @code{sbrk}, the | |
413 | allocation function fails. | |
414 | ||
415 | @comment sys/resource.h | |
416 | @comment BSD | |
417 | @item RLIM_NLIMITS | |
418 | @vindex RLIM_NLIMITS | |
419 | The number of different resource limits. Any valid @var{resource} | |
420 | operand must be less than @code{RLIM_NLIMITS}. | |
421 | @end table | |
422 | ||
423 | @comment sys/resource.h | |
424 | @comment BSD | |
425 | @deftypevr Constant int RLIM_INFINITY | |
426 | This constant stands for a value of ``infinity'' when supplied as | |
427 | the limit value in @code{setrlimit}. | |
428 | @end deftypevr | |
429 | ||
430 | ||
431 | The following are historical functions to do some of what the functions | |
432 | above do. The functions above are better choices. | |
433 | ||
434 | @code{ulimit} and the command symbols are declared in @file{ulimit.h}. | |
435 | @pindex ulimit.h | |
5ce8f203 | 436 | |
b642f101 UD |
437 | @comment ulimit.h |
438 | @comment BSD | |
5ce8f203 UD |
439 | @deftypefun int ulimit (int @var{cmd}, ...) |
440 | ||
441 | @code{ulimit} gets the current limit or sets the current and maximum | |
442 | limit for a particular resource for the calling process according to the | |
443 | command @var{cmd}.a | |
444 | ||
445 | If you are getting a limit, the command argument is the only argument. | |
446 | If you are setting a limit, there is a second argument: | |
447 | @code{long int} @var{limit} which is the value to which you are setting | |
448 | the limit. | |
449 | ||
450 | The @var{cmd} values and the operations they specify are: | |
451 | @table @code | |
452 | ||
453 | @item GETFSIZE | |
454 | Get the current limit on the size of a file, in units of 512 bytes. | |
455 | ||
456 | @item SETFSIZE | |
457 | Set the current and maximum limit on the size of a file to @var{limit} * | |
458 | 512 bytes. | |
459 | ||
460 | @end table | |
461 | ||
462 | There are also some other @var{cmd} values that may do things on some | |
463 | systems, but they are not supported. | |
464 | ||
465 | Only the superuser may increase a maximum limit. | |
466 | ||
467 | When you successfully get a limit, the return value of @code{ulimit} is | |
468 | that limit, which is never negative. When you successfully set a limit, | |
469 | the return value is zero. When the function fails, the return value is | |
470 | @code{-1} and @code{errno} is set according to the reason: | |
471 | ||
472 | @table @code | |
473 | @item EPERM | |
474 | A process tried to increase a maximum limit, but is not superuser. | |
475 | @end table | |
476 | ||
477 | ||
478 | @end deftypefun | |
479 | ||
480 | @code{vlimit} and its resource symbols are declared in @file{sys/vlimit.h}. | |
5ce8f203 | 481 | @pindex sys/vlimit.h |
5ce8f203 | 482 | |
b642f101 UD |
483 | @comment sys/vlimit.h |
484 | @comment BSD | |
5ce8f203 UD |
485 | @deftypefun int vlimit (int @var{resource}, int @var{limit}) |
486 | ||
487 | @code{vlimit} sets the current limit for a resource for a process. | |
488 | ||
489 | @var{resource} identifies the resource: | |
490 | ||
491 | @table @code | |
492 | @item LIM_CPU | |
493 | Maximum CPU time. Same as @code{RLIMIT_CPU} for @code{setrlimit}. | |
494 | @item LIM_FSIZE | |
495 | Maximum file size. Same as @code{RLIMIT_FSIZE} for @code{setrlimit}. | |
496 | @item LIM_DATA | |
497 | Maximum data memory. Same as @code{RLIMIT_DATA} for @code{setrlimit}. | |
498 | @item LIM_STACK | |
499 | Maximum stack size. Same as @code{RLIMIT_STACK} for @code{setrlimit}. | |
500 | @item LIM_CORE | |
501 | Maximum core file size. Same as @code{RLIMIT_COR} for @code{setrlimit}. | |
502 | @item LIM_MAXRSS | |
503 | Maximum physical memory. Same as @code{RLIMIT_RSS} for @code{setrlimit}. | |
504 | @end table | |
505 | ||
506 | The return value is zero for success, and @code{-1} with @code{errno} set | |
507 | accordingly for failure: | |
508 | ||
509 | @table @code | |
510 | @item EPERM | |
511 | The process tried to set its current limit beyond its maximum limit. | |
512 | @end table | |
513 | ||
514 | @end deftypefun | |
515 | ||
516 | @node Priority | |
639c6286 | 517 | @section Process CPU Priority And Scheduling |
5ce8f203 | 518 | @cindex process priority |
639c6286 | 519 | @cindex cpu priority |
5ce8f203 UD |
520 | @cindex priority of a process |
521 | ||
639c6286 UD |
522 | When multiple processes simultaneously require CPU time, the system's |
523 | scheduling policy and process CPU priorities determine which processes | |
524 | get it. This section describes how that determination is made and | |
525 | GNU C library functions to control it. | |
526 | ||
527 | It is common to refer to CPU scheduling simply as scheduling and a | |
528 | process' CPU priority simply as the process' priority, with the CPU | |
529 | resource being implied. Bear in mind, though, that CPU time is not the | |
530 | only resource a process uses or that processes contend for. In some | |
531 | cases, it is not even particularly important. Giving a process a high | |
532 | ``priority'' may have very little effect on how fast a process runs with | |
533 | respect to other processes. The priorities discussed in this section | |
534 | apply only to CPU time. | |
535 | ||
536 | CPU scheduling is a complex issue and different systems do it in wildly | |
537 | different ways. New ideas continually develop and find their way into | |
538 | the intricacies of the various systems' scheduling algorithms. This | |
87b56f36 | 539 | section discusses the general concepts, some specifics of systems |
639c6286 UD |
540 | that commonly use the GNU C library, and some standards. |
541 | ||
542 | For simplicity, we talk about CPU contention as if there is only one CPU | |
543 | in the system. But all the same principles apply when a processor has | |
544 | multiple CPUs, and knowing that the number of processes that can run at | |
545 | any one time is equal to the number of CPUs, you can easily extrapolate | |
546 | the information. | |
547 | ||
548 | The functions described in this section are all defined by the POSIX.1 | |
95fdc6a0 | 549 | and POSIX.1b standards (the @code{sched@dots{}} functions are POSIX.1b). |
639c6286 UD |
550 | However, POSIX does not define any semantics for the values that these |
551 | functions get and set. In this chapter, the semantics are based on the | |
552 | Linux kernel's implementation of the POSIX standard. As you will see, | |
553 | the Linux implementation is quite the inverse of what the authors of the | |
554 | POSIX syntax had in mind. | |
555 | ||
556 | @menu | |
557 | * Absolute Priority:: The first tier of priority. Posix | |
558 | * Realtime Scheduling:: Scheduling among the process nobility | |
559 | * Basic Scheduling Functions:: Get/set scheduling policy, priority | |
560 | * Traditional Scheduling:: Scheduling among the vulgar masses | |
d9997a45 | 561 | * CPU Affinity:: Limiting execution to certain CPUs |
639c6286 UD |
562 | @end menu |
563 | ||
564 | ||
565 | ||
566 | @node Absolute Priority | |
567 | @subsection Absolute Priority | |
568 | @cindex absolute priority | |
569 | @cindex priority, absolute | |
570 | ||
571 | Every process has an absolute priority, and it is represented by a number. | |
572 | The higher the number, the higher the absolute priority. | |
573 | ||
574 | @cindex realtime CPU scheduling | |
575 | On systems of the past, and most systems today, all processes have | |
576 | absolute priority 0 and this section is irrelevant. In that case, | |
577 | @xref{Traditional Scheduling}. Absolute priorities were invented to | |
0bc93a2f | 578 | accommodate realtime systems, in which it is vital that certain processes |
639c6286 UD |
579 | be able to respond to external events happening in real time, which |
580 | means they cannot wait around while some other process that @emph{wants | |
581 | to}, but doesn't @emph{need to} run occupies the CPU. | |
582 | ||
583 | @cindex ready to run | |
584 | @cindex preemptive scheduling | |
585 | When two processes are in contention to use the CPU at any instant, the | |
586 | one with the higher absolute priority always gets it. This is true even if the | |
11bf311e | 587 | process with the lower priority is already using the CPU (i.e., the |
639c6286 UD |
588 | scheduling is preemptive). Of course, we're only talking about |
589 | processes that are running or ``ready to run,'' which means they are | |
590 | ready to execute instructions right now. When a process blocks to wait | |
591 | for something like I/O, its absolute priority is irrelevant. | |
592 | ||
593 | @cindex runnable process | |
48b22986 | 594 | @strong{NB:} The term ``runnable'' is a synonym for ``ready to run.'' |
639c6286 UD |
595 | |
596 | When two processes are running or ready to run and both have the same | |
597 | absolute priority, it's more interesting. In that case, who gets the | |
0bc93a2f | 598 | CPU is determined by the scheduling policy. If the processes have |
639c6286 UD |
599 | absolute priority 0, the traditional scheduling policy described in |
600 | @ref{Traditional Scheduling} applies. Otherwise, the policies described | |
601 | in @ref{Realtime Scheduling} apply. | |
602 | ||
603 | You normally give an absolute priority above 0 only to a process that | |
604 | can be trusted not to hog the CPU. Such processes are designed to block | |
605 | (or terminate) after relatively short CPU runs. | |
606 | ||
607 | A process begins life with the same absolute priority as its parent | |
608 | process. Functions described in @ref{Basic Scheduling Functions} can | |
609 | change it. | |
610 | ||
611 | Only a privileged process can change a process' absolute priority to | |
612 | something other than @code{0}. Only a privileged process or the | |
613 | target process' owner can change its absolute priority at all. | |
614 | ||
615 | POSIX requires absolute priority values used with the realtime | |
616 | scheduling policies to be consecutive with a range of at least 32. On | |
617 | Linux, they are 1 through 99. The functions | |
618 | @code{sched_get_priority_max} and @code{sched_set_priority_min} portably | |
619 | tell you what the range is on a particular system. | |
620 | ||
621 | ||
622 | @subsubsection Using Absolute Priority | |
623 | ||
624 | One thing you must keep in mind when designing real time applications is | |
625 | that having higher absolute priority than any other process doesn't | |
626 | guarantee the process can run continuously. Two things that can wreck a | |
87b56f36 | 627 | good CPU run are interrupts and page faults. |
639c6286 UD |
628 | |
629 | Interrupt handlers live in that limbo between processes. The CPU is | |
630 | executing instructions, but they aren't part of any process. An | |
631 | interrupt will stop even the highest priority process. So you must | |
632 | allow for slight delays and make sure that no device in the system has | |
633 | an interrupt handler that could cause too long a delay between | |
634 | instructions for your process. | |
635 | ||
636 | Similarly, a page fault causes what looks like a straightforward | |
637 | sequence of instructions to take a long time. The fact that other | |
638 | processes get to run while the page faults in is of no consequence, | |
639 | because as soon as the I/O is complete, the high priority process will | |
640 | kick them out and run again, but the wait for the I/O itself could be a | |
641 | problem. To neutralize this threat, use @code{mlock} or | |
642 | @code{mlockall}. | |
643 | ||
644 | There are a few ramifications of the absoluteness of this priority on a | |
645 | single-CPU system that you need to keep in mind when you choose to set a | |
646 | priority and also when you're working on a program that runs with high | |
647 | absolute priority. Consider a process that has higher absolute priority | |
648 | than any other process in the system and due to a bug in its program, it | |
649 | gets into an infinite loop. It will never cede the CPU. You can't run | |
650 | a command to kill it because your command would need to get the CPU in | |
651 | order to run. The errant program is in complete control. It controls | |
652 | the vertical, it controls the horizontal. | |
653 | ||
654 | There are two ways to avoid this: 1) keep a shell running somewhere with | |
655 | a higher absolute priority. 2) keep a controlling terminal attached to | |
656 | the high priority process group. All the priority in the world won't | |
657 | stop an interrupt handler from running and delivering a signal to the | |
658 | process if you hit Control-C. | |
659 | ||
95fdc6a0 | 660 | Some systems use absolute priority as a means of allocating a fixed |
0bc93a2f | 661 | percentage of CPU time to a process. To do this, a super high priority |
639c6286 UD |
662 | privileged process constantly monitors the process' CPU usage and raises |
663 | its absolute priority when the process isn't getting its entitled share | |
664 | and lowers it when the process is exceeding it. | |
665 | ||
48b22986 | 666 | @strong{NB:} The absolute priority is sometimes called the ``static |
639c6286 UD |
667 | priority.'' We don't use that term in this manual because it misses the |
668 | most important feature of the absolute priority: its absoluteness. | |
669 | ||
670 | ||
671 | @node Realtime Scheduling | |
672 | @subsection Realtime Scheduling | |
b642f101 | 673 | @cindex realtime scheduling |
639c6286 UD |
674 | |
675 | Whenever two processes with the same absolute priority are ready to run, | |
676 | the kernel has a decision to make, because only one can run at a time. | |
677 | If the processes have absolute priority 0, the kernel makes this decision | |
678 | as described in @ref{Traditional Scheduling}. Otherwise, the decision | |
679 | is as described in this section. | |
680 | ||
681 | If two processes are ready to run but have different absolute priorities, | |
682 | the decision is much simpler, and is described in @ref{Absolute | |
683 | Priority}. | |
684 | ||
87b56f36 | 685 | Each process has a scheduling policy. For processes with absolute |
639c6286 UD |
686 | priority other than zero, there are two available: |
687 | ||
688 | @enumerate | |
689 | @item | |
690 | First Come First Served | |
691 | @item | |
692 | Round Robin | |
693 | @end enumerate | |
694 | ||
695 | The most sensible case is where all the processes with a certain | |
696 | absolute priority have the same scheduling policy. We'll discuss that | |
697 | first. | |
698 | ||
699 | In Round Robin, processes share the CPU, each one running for a small | |
700 | quantum of time (``time slice'') and then yielding to another in a | |
701 | circular fashion. Of course, only processes that are ready to run and | |
702 | have the same absolute priority are in this circle. | |
703 | ||
704 | In First Come First Served, the process that has been waiting the | |
705 | longest to run gets the CPU, and it keeps it until it voluntarily | |
706 | relinquishes the CPU, runs out of things to do (blocks), or gets | |
707 | preempted by a higher priority process. | |
708 | ||
709 | First Come First Served, along with maximal absolute priority and | |
710 | careful control of interrupts and page faults, is the one to use when a | |
711 | process absolutely, positively has to run at full CPU speed or not at | |
712 | all. | |
713 | ||
714 | Judicious use of @code{sched_yield} function invocations by processes | |
715 | with First Come First Served scheduling policy forms a good compromise | |
716 | between Round Robin and First Come First Served. | |
717 | ||
718 | To understand how scheduling works when processes of different scheduling | |
719 | policies occupy the same absolute priority, you have to know the nitty | |
720 | gritty details of how processes enter and exit the ready to run list: | |
721 | ||
722 | In both cases, the ready to run list is organized as a true queue, where | |
723 | a process gets pushed onto the tail when it becomes ready to run and is | |
724 | popped off the head when the scheduler decides to run it. Note that | |
725 | ready to run and running are two mutually exclusive states. When the | |
726 | scheduler runs a process, that process is no longer ready to run and no | |
727 | longer in the ready to run list. When the process stops running, it | |
728 | may go back to being ready to run again. | |
729 | ||
730 | The only difference between a process that is assigned the Round Robin | |
731 | scheduling policy and a process that is assigned First Come First Serve | |
732 | is that in the former case, the process is automatically booted off the | |
733 | CPU after a certain amount of time. When that happens, the process goes | |
734 | back to being ready to run, which means it enters the queue at the tail. | |
735 | The time quantum we're talking about is small. Really small. This is | |
736 | not your father's timesharing. For example, with the Linux kernel, the | |
737 | round robin time slice is a thousand times shorter than its typical | |
738 | time slice for traditional scheduling. | |
739 | ||
740 | A process begins life with the same scheduling policy as its parent process. | |
741 | Functions described in @ref{Basic Scheduling Functions} can change it. | |
742 | ||
743 | Only a privileged process can set the scheduling policy of a process | |
744 | that has absolute priority higher than 0. | |
745 | ||
746 | @node Basic Scheduling Functions | |
747 | @subsection Basic Scheduling Functions | |
748 | ||
749 | This section describes functions in the GNU C library for setting the | |
750 | absolute priority and scheduling policy of a process. | |
751 | ||
752 | @strong{Portability Note:} On systems that have the functions in this | |
753 | section, the macro _POSIX_PRIORITY_SCHEDULING is defined in | |
754 | @file{<unistd.h>}. | |
755 | ||
756 | For the case that the scheduling policy is traditional scheduling, more | |
757 | functions to fine tune the scheduling are in @ref{Traditional Scheduling}. | |
758 | ||
759 | Don't try to make too much out of the naming and structure of these | |
760 | functions. They don't match the concepts described in this manual | |
761 | because the functions are as defined by POSIX.1b, but the implementation | |
762 | on systems that use the GNU C library is the inverse of what the POSIX | |
763 | structure contemplates. The POSIX scheme assumes that the primary | |
764 | scheduling parameter is the scheduling policy and that the priority | |
765 | value, if any, is a parameter of the scheduling policy. In the | |
766 | implementation, though, the priority value is king and the scheduling | |
767 | policy, if anything, only fine tunes the effect of that priority. | |
768 | ||
769 | The symbols in this section are declared by including file @file{sched.h}. | |
770 | ||
771 | @comment sched.h | |
772 | @comment POSIX | |
773 | @deftp {Data Type} {struct sched_param} | |
774 | This structure describes an absolute priority. | |
775 | @table @code | |
776 | @item int sched_priority | |
777 | absolute priority value | |
778 | @end table | |
779 | @end deftp | |
780 | ||
781 | @comment sched.h | |
782 | @comment POSIX | |
783 | @deftypefun int sched_setscheduler (pid_t @var{pid}, int @var{policy}, const struct sched_param *@var{param}) | |
784 | ||
785 | This function sets both the absolute priority and the scheduling policy | |
786 | for a process. | |
787 | ||
788 | It assigns the absolute priority value given by @var{param} and the | |
789 | scheduling policy @var{policy} to the process with Process ID @var{pid}, | |
790 | or the calling process if @var{pid} is zero. If @var{policy} is | |
0bc93a2f | 791 | negative, @code{sched_setscheduler} keeps the existing scheduling policy. |
639c6286 UD |
792 | |
793 | The following macros represent the valid values for @var{policy}: | |
794 | ||
795 | @table @code | |
796 | @item SCHED_OTHER | |
797 | Traditional Scheduling | |
798 | @item SCHED_FIFO | |
87b56f36 | 799 | First In First Out |
639c6286 UD |
800 | @item SCHED_RR |
801 | Round Robin | |
802 | @end table | |
803 | ||
804 | @c The Linux kernel code (in sched.c) actually reschedules the process, | |
805 | @c but it puts it at the head of the run queue, so I'm not sure just what | |
806 | @c the effect is, but it must be subtle. | |
807 | ||
808 | On success, the return value is @code{0}. Otherwise, it is @code{-1} | |
809 | and @code{ERRNO} is set accordingly. The @code{errno} values specific | |
810 | to this function are: | |
811 | ||
812 | @table @code | |
813 | @item EPERM | |
814 | @itemize @bullet | |
815 | @item | |
816 | The calling process does not have @code{CAP_SYS_NICE} permission and | |
817 | @var{policy} is not @code{SCHED_OTHER} (or it's negative and the | |
818 | existing policy is not @code{SCHED_OTHER}. | |
819 | ||
820 | @item | |
821 | The calling process does not have @code{CAP_SYS_NICE} permission and its | |
11bf311e | 822 | owner is not the target process' owner. I.e., the effective uid of the |
639c6286 UD |
823 | calling process is neither the effective nor the real uid of process |
824 | @var{pid}. | |
825 | @c We need a cross reference to the capabilities section, when written. | |
826 | @end itemize | |
827 | ||
828 | @item ESRCH | |
829 | There is no process with pid @var{pid} and @var{pid} is not zero. | |
830 | ||
831 | @item EINVAL | |
832 | @itemize @bullet | |
833 | @item | |
834 | @var{policy} does not identify an existing scheduling policy. | |
835 | ||
836 | @item | |
837 | The absolute priority value identified by *@var{param} is outside the | |
838 | valid range for the scheduling policy @var{policy} (or the existing | |
839 | scheduling policy if @var{policy} is negative) or @var{param} is | |
840 | null. @code{sched_get_priority_max} and @code{sched_get_priority_min} | |
841 | tell you what the valid range is. | |
842 | ||
843 | @item | |
844 | @var{pid} is negative. | |
845 | @end itemize | |
846 | @end table | |
847 | ||
848 | @end deftypefun | |
849 | ||
850 | ||
851 | @comment sched.h | |
852 | @comment POSIX | |
853 | @deftypefun int sched_getscheduler (pid_t @var{pid}) | |
854 | ||
855 | This function returns the scheduling policy assigned to the process with | |
856 | Process ID (pid) @var{pid}, or the calling process if @var{pid} is zero. | |
857 | ||
858 | The return value is the scheduling policy. See | |
859 | @code{sched_setscheduler} for the possible values. | |
860 | ||
861 | If the function fails, the return value is instead @code{-1} and | |
862 | @code{errno} is set accordingly. | |
863 | ||
864 | The @code{errno} values specific to this function are: | |
865 | ||
866 | @table @code | |
867 | ||
868 | @item ESRCH | |
869 | There is no process with pid @var{pid} and it is not zero. | |
870 | ||
871 | @item EINVAL | |
872 | @var{pid} is negative. | |
873 | ||
874 | @end table | |
875 | ||
876 | Note that this function is not an exact mate to @code{sched_setscheduler} | |
877 | because while that function sets the scheduling policy and the absolute | |
878 | priority, this function gets only the scheduling policy. To get the | |
879 | absolute priority, use @code{sched_getparam}. | |
880 | ||
881 | @end deftypefun | |
882 | ||
883 | ||
884 | @comment sched.h | |
885 | @comment POSIX | |
886 | @deftypefun int sched_setparam (pid_t @var{pid}, const struct sched_param *@var{param}) | |
887 | ||
888 | This function sets a process' absolute priority. | |
889 | ||
890 | It is functionally identical to @code{sched_setscheduler} with | |
891 | @var{policy} = @code{-1}. | |
892 | ||
893 | @c in fact, that's how it's implemented in Linux. | |
894 | ||
895 | @end deftypefun | |
896 | ||
897 | @comment sched.h | |
898 | @comment POSIX | |
899 | @deftypefun int sched_getparam (pid_t @var{pid}, const struct sched_param *@var{param}) | |
900 | ||
901 | This function returns a process' absolute priority. | |
902 | ||
903 | @var{pid} is the Process ID (pid) of the process whose absolute priority | |
904 | you want to know. | |
905 | ||
906 | @var{param} is a pointer to a structure in which the function stores the | |
907 | absolute priority of the process. | |
908 | ||
909 | On success, the return value is @code{0}. Otherwise, it is @code{-1} | |
910 | and @code{ERRNO} is set accordingly. The @code{errno} values specific | |
911 | to this function are: | |
912 | ||
913 | @table @code | |
914 | ||
915 | @item ESRCH | |
916 | There is no process with pid @var{pid} and it is not zero. | |
917 | ||
918 | @item EINVAL | |
919 | @var{pid} is negative. | |
920 | ||
921 | @end table | |
922 | ||
923 | @end deftypefun | |
924 | ||
925 | ||
926 | @comment sched.h | |
927 | @comment POSIX | |
928 | @deftypefun int sched_get_priority_min (int *@var{policy}); | |
929 | ||
930 | This function returns the lowest absolute priority value that is | |
931 | allowable for a process with scheduling policy @var{policy}. | |
932 | ||
933 | On Linux, it is 0 for SCHED_OTHER and 1 for everything else. | |
934 | ||
935 | On success, the return value is @code{0}. Otherwise, it is @code{-1} | |
936 | and @code{ERRNO} is set accordingly. The @code{errno} values specific | |
937 | to this function are: | |
938 | ||
939 | @table @code | |
940 | @item EINVAL | |
941 | @var{policy} does not identify an existing scheduling policy. | |
942 | @end table | |
943 | ||
944 | @end deftypefun | |
945 | ||
946 | @comment sched.h | |
947 | @comment POSIX | |
87b56f36 | 948 | @deftypefun int sched_get_priority_max (int *@var{policy}); |
639c6286 UD |
949 | |
950 | This function returns the highest absolute priority value that is | |
951 | allowable for a process that with scheduling policy @var{policy}. | |
952 | ||
953 | On Linux, it is 0 for SCHED_OTHER and 99 for everything else. | |
954 | ||
955 | On success, the return value is @code{0}. Otherwise, it is @code{-1} | |
956 | and @code{ERRNO} is set accordingly. The @code{errno} values specific | |
957 | to this function are: | |
958 | ||
959 | @table @code | |
960 | @item EINVAL | |
961 | @var{policy} does not identify an existing scheduling policy. | |
962 | @end table | |
963 | ||
964 | @end deftypefun | |
965 | ||
966 | @comment sched.h | |
967 | @comment POSIX | |
968 | @deftypefun int sched_rr_get_interval (pid_t @var{pid}, struct timespec *@var{interval}) | |
969 | ||
87b56f36 | 970 | This function returns the length of the quantum (time slice) used with |
639c6286 UD |
971 | the Round Robin scheduling policy, if it is used, for the process with |
972 | Process ID @var{pid}. | |
973 | ||
87b56f36 | 974 | It returns the length of time as @var{interval}. |
639c6286 UD |
975 | @c We need a cross-reference to where timespec is explained. But that |
976 | @c section doesn't exist yet, and the time chapter needs to be slightly | |
977 | @c reorganized so there is a place to put it (which will be right next | |
978 | @c to timeval, which is presently misplaced). 2000.05.07. | |
979 | ||
980 | With a Linux kernel, the round robin time slice is always 150 | |
981 | microseconds, and @var{pid} need not even be a real pid. | |
982 | ||
983 | The return value is @code{0} on success and in the pathological case | |
984 | that it fails, the return value is @code{-1} and @code{errno} is set | |
985 | accordingly. There is nothing specific that can go wrong with this | |
986 | function, so there are no specific @code{errno} values. | |
987 | ||
988 | @end deftypefun | |
989 | ||
990 | @comment sched.h | |
991 | @comment POSIX | |
3c44837c | 992 | @deftypefun int sched_yield (void) |
639c6286 UD |
993 | |
994 | This function voluntarily gives up the process' claim on the CPU. | |
995 | ||
996 | Technically, @code{sched_yield} causes the calling process to be made | |
997 | immediately ready to run (as opposed to running, which is what it was | |
998 | before). This means that if it has absolute priority higher than 0, it | |
999 | gets pushed onto the tail of the queue of processes that share its | |
1000 | absolute priority and are ready to run, and it will run again when its | |
1001 | turn next arrives. If its absolute priority is 0, it is more | |
1002 | complicated, but still has the effect of yielding the CPU to other | |
1003 | processes. | |
1004 | ||
1005 | If there are no other processes that share the calling process' absolute | |
1006 | priority, this function doesn't have any effect. | |
1007 | ||
1008 | To the extent that the containing program is oblivious to what other | |
1009 | processes in the system are doing and how fast it executes, this | |
1010 | function appears as a no-op. | |
1011 | ||
1012 | The return value is @code{0} on success and in the pathological case | |
1013 | that it fails, the return value is @code{-1} and @code{errno} is set | |
1014 | accordingly. There is nothing specific that can go wrong with this | |
1015 | function, so there are no specific @code{errno} values. | |
1016 | ||
1017 | @end deftypefun | |
1018 | ||
1019 | @node Traditional Scheduling | |
1020 | @subsection Traditional Scheduling | |
1021 | @cindex scheduling, traditional | |
1022 | ||
1023 | This section is about the scheduling among processes whose absolute | |
1024 | priority is 0. When the system hands out the scraps of CPU time that | |
0bc93a2f | 1025 | are left over after the processes with higher absolute priority have |
639c6286 UD |
1026 | taken all they want, the scheduling described herein determines who |
1027 | among the great unwashed processes gets them. | |
1028 | ||
1029 | @menu | |
1030 | * Traditional Scheduling Intro:: | |
1031 | * Traditional Scheduling Functions:: | |
1032 | @end menu | |
1033 | ||
1034 | @node Traditional Scheduling Intro | |
1035 | @subsubsection Introduction To Traditional Scheduling | |
1036 | ||
1037 | Long before there was absolute priority (See @ref{Absolute Priority}), | |
1038 | Unix systems were scheduling the CPU using this system. When Posix came | |
0bc93a2f | 1039 | in like the Romans and imposed absolute priorities to accommodate the |
639c6286 UD |
1040 | needs of realtime processing, it left the indigenous Absolute Priority |
1041 | Zero processes to govern themselves by their own familiar scheduling | |
1042 | policy. | |
1043 | ||
1044 | Indeed, absolute priorities higher than zero are not available on many | |
1045 | systems today and are not typically used when they are, being intended | |
1046 | mainly for computers that do realtime processing. So this section | |
1047 | describes the only scheduling many programmers need to be concerned | |
1048 | about. | |
1049 | ||
1050 | But just to be clear about the scope of this scheduling: Any time a | |
1051 | process with a absolute priority of 0 and a process with an absolute | |
1052 | priority higher than 0 are ready to run at the same time, the one with | |
1053 | absolute priority 0 does not run. If it's already running when the | |
1054 | higher priority ready-to-run process comes into existence, it stops | |
1055 | immediately. | |
1056 | ||
1057 | In addition to its absolute priority of zero, every process has another | |
1058 | priority, which we will refer to as "dynamic priority" because it changes | |
87b56f36 | 1059 | over time. The dynamic priority is meaningless for processes with |
639c6286 UD |
1060 | an absolute priority higher than zero. |
1061 | ||
1062 | The dynamic priority sometimes determines who gets the next turn on the | |
1063 | CPU. Sometimes it determines how long turns last. Sometimes it | |
1064 | determines whether a process can kick another off the CPU. | |
1065 | ||
1066 | In Linux, the value is a combination of these things, but mostly it is | |
1067 | just determines the length of the time slice. The higher a process' | |
1068 | dynamic priority, the longer a shot it gets on the CPU when it gets one. | |
1069 | If it doesn't use up its time slice before giving up the CPU to do | |
1070 | something like wait for I/O, it is favored for getting the CPU back when | |
1071 | it's ready for it, to finish out its time slice. Other than that, | |
1072 | selection of processes for new time slices is basically round robin. | |
1073 | But the scheduler does throw a bone to the low priority processes: A | |
1074 | process' dynamic priority rises every time it is snubbed in the | |
1075 | scheduling process. In Linux, even the fat kid gets to play. | |
1076 | ||
1077 | The fluctuation of a process' dynamic priority is regulated by another | |
1078 | value: The ``nice'' value. The nice value is an integer, usually in the | |
1079 | range -20 to 20, and represents an upper limit on a process' dynamic | |
1080 | priority. The higher the nice number, the lower that limit. | |
1081 | ||
1082 | On a typical Linux system, for example, a process with a nice value of | |
1083 | 20 can get only 10 milliseconds on the CPU at a time, whereas a process | |
1084 | with a nice value of -20 can achieve a high enough priority to get 400 | |
1085 | milliseconds. | |
1086 | ||
1087 | The idea of the nice value is deferential courtesy. In the beginning, | |
1088 | in the Unix garden of Eden, all processes shared equally in the bounty | |
1089 | of the computer system. But not all processes really need the same | |
1090 | share of CPU time, so the nice value gave a courteous process the | |
1091 | ability to refuse its equal share of CPU time that others might prosper. | |
1092 | Hence, the higher a process' nice value, the nicer the process is. | |
1093 | (Then a snake came along and offered some process a negative nice value | |
1094 | and the system became the crass resource allocation system we know | |
1095 | today). | |
1096 | ||
1097 | Dynamic priorities tend upward and downward with an objective of | |
1098 | smoothing out allocation of CPU time and giving quick response time to | |
1099 | infrequent requests. But they never exceed their nice limits, so on a | |
1100 | heavily loaded CPU, the nice value effectively determines how fast a | |
1101 | process runs. | |
1102 | ||
1103 | In keeping with the socialistic heritage of Unix process priority, a | |
1104 | process begins life with the same nice value as its parent process and | |
1105 | can raise it at will. A process can also raise the nice value of any | |
1106 | other process owned by the same user (or effective user). But only a | |
1107 | privileged process can lower its nice value. A privileged process can | |
1108 | also raise or lower another process' nice value. | |
1109 | ||
87b56f36 | 1110 | GNU C Library functions for getting and setting nice values are described in |
639c6286 UD |
1111 | @xref{Traditional Scheduling Functions}. |
1112 | ||
1113 | @node Traditional Scheduling Functions | |
1114 | @subsubsection Functions For Traditional Scheduling | |
1115 | ||
5ce8f203 | 1116 | @pindex sys/resource.h |
639c6286 UD |
1117 | This section describes how you can read and set the nice value of a |
1118 | process. All these symbols are declared in @file{sys/resource.h}. | |
1119 | ||
1120 | The function and macro names are defined by POSIX, and refer to | |
1121 | "priority," but the functions actually have to do with nice values, as | |
1122 | the terms are used both in the manual and POSIX. | |
1123 | ||
1124 | The range of valid nice values depends on the kernel, but typically it | |
1125 | runs from @code{-20} to @code{20}. A lower nice value corresponds to | |
1126 | higher priority for the process. These constants describe the range of | |
5ce8f203 UD |
1127 | priority values: |
1128 | ||
b642f101 | 1129 | @vtable @code |
5ce8f203 UD |
1130 | @comment sys/resource.h |
1131 | @comment BSD | |
1132 | @item PRIO_MIN | |
639c6286 | 1133 | The lowest valid nice value. |
5ce8f203 UD |
1134 | |
1135 | @comment sys/resource.h | |
1136 | @comment BSD | |
1137 | @item PRIO_MAX | |
639c6286 | 1138 | The highest valid nice value. |
b642f101 | 1139 | @end vtable |
5ce8f203 UD |
1140 | |
1141 | @comment sys/resource.h | |
639c6286 | 1142 | @comment BSD,POSIX |
5ce8f203 | 1143 | @deftypefun int getpriority (int @var{class}, int @var{id}) |
639c6286 | 1144 | Return the nice value of a set of processes; @var{class} and @var{id} |
5ce8f203 | 1145 | specify which ones (see below). If the processes specified do not all |
639c6286 | 1146 | have the same nice value, this returns the lowest value that any of them |
5ce8f203 UD |
1147 | has. |
1148 | ||
639c6286 UD |
1149 | On success, the return value is @code{0}. Otherwise, it is @code{-1} |
1150 | and @code{ERRNO} is set accordingly. The @code{errno} values specific | |
1151 | to this function are: | |
5ce8f203 UD |
1152 | |
1153 | @table @code | |
1154 | @item ESRCH | |
1155 | The combination of @var{class} and @var{id} does not match any existing | |
1156 | process. | |
1157 | ||
1158 | @item EINVAL | |
1159 | The value of @var{class} is not valid. | |
1160 | @end table | |
1161 | ||
639c6286 UD |
1162 | If the return value is @code{-1}, it could indicate failure, or it could |
1163 | be the nice value. The only way to make certain is to set @code{errno = | |
1164 | 0} before calling @code{getpriority}, then use @code{errno != 0} | |
1165 | afterward as the criterion for failure. | |
5ce8f203 UD |
1166 | @end deftypefun |
1167 | ||
1168 | @comment sys/resource.h | |
639c6286 UD |
1169 | @comment BSD,POSIX |
1170 | @deftypefun int setpriority (int @var{class}, int @var{id}, int @var{niceval}) | |
1171 | Set the nice value of a set of processes to @var{niceval}; @var{class} | |
5ce8f203 UD |
1172 | and @var{id} specify which ones (see below). |
1173 | ||
6a7a8b22 | 1174 | The return value is @code{0} on success, and @code{-1} on |
639c6286 UD |
1175 | failure. The following @code{errno} error condition are possible for |
1176 | this function: | |
5ce8f203 UD |
1177 | |
1178 | @table @code | |
1179 | @item ESRCH | |
1180 | The combination of @var{class} and @var{id} does not match any existing | |
1181 | process. | |
1182 | ||
1183 | @item EINVAL | |
1184 | The value of @var{class} is not valid. | |
1185 | ||
1186 | @item EPERM | |
639c6286 | 1187 | The call would set the nice value of a process which is owned by a different |
11bf311e | 1188 | user than the calling process (i.e., the target process' real or effective |
639c6286 UD |
1189 | uid does not match the calling process' effective uid) and the calling |
1190 | process does not have @code{CAP_SYS_NICE} permission. | |
5ce8f203 UD |
1191 | |
1192 | @item EACCES | |
639c6286 UD |
1193 | The call would lower the process' nice value and the process does not have |
1194 | @code{CAP_SYS_NICE} permission. | |
5ce8f203 | 1195 | @end table |
639c6286 | 1196 | |
5ce8f203 UD |
1197 | @end deftypefun |
1198 | ||
1199 | The arguments @var{class} and @var{id} together specify a set of | |
1200 | processes in which you are interested. These are the possible values of | |
1201 | @var{class}: | |
1202 | ||
b642f101 | 1203 | @vtable @code |
5ce8f203 UD |
1204 | @comment sys/resource.h |
1205 | @comment BSD | |
1206 | @item PRIO_PROCESS | |
639c6286 | 1207 | One particular process. The argument @var{id} is a process ID (pid). |
5ce8f203 UD |
1208 | |
1209 | @comment sys/resource.h | |
1210 | @comment BSD | |
1211 | @item PRIO_PGRP | |
639c6286 UD |
1212 | All the processes in a particular process group. The argument @var{id} is |
1213 | a process group ID (pgid). | |
5ce8f203 UD |
1214 | |
1215 | @comment sys/resource.h | |
1216 | @comment BSD | |
1217 | @item PRIO_USER | |
11bf311e | 1218 | All the processes owned by a particular user (i.e., whose real uid |
639c6286 | 1219 | indicates the user). The argument @var{id} is a user ID (uid). |
b642f101 | 1220 | @end vtable |
5ce8f203 | 1221 | |
639c6286 UD |
1222 | If the argument @var{id} is 0, it stands for the calling process, its |
1223 | process group, or its owner (real uid), according to @var{class}. | |
5ce8f203 | 1224 | |
b642f101 UD |
1225 | @comment unistd.h |
1226 | @comment BSD | |
5ce8f203 | 1227 | @deftypefun int nice (int @var{increment}) |
639c6286 | 1228 | Increment the nice value of the calling process by @var{increment}. |
6a7a8b22 AJ |
1229 | The return value is the new nice value on success, and @code{-1} on |
1230 | failure. In the case of failure, @code{errno} will be set to the | |
1231 | same values as for @code{setpriority}. | |
1232 | ||
5ce8f203 UD |
1233 | |
1234 | Here is an equivalent definition of @code{nice}: | |
1235 | ||
1236 | @smallexample | |
1237 | int | |
1238 | nice (int increment) | |
1239 | @{ | |
6a7a8b22 AJ |
1240 | int result, old = getpriority (PRIO_PROCESS, 0); |
1241 | result = setpriority (PRIO_PROCESS, 0, old + increment); | |
1242 | if (result != -1) | |
1243 | return old + increment; | |
1244 | else | |
1245 | return -1; | |
5ce8f203 UD |
1246 | @} |
1247 | @end smallexample | |
1248 | @end deftypefun | |
b642f101 | 1249 | |
d9997a45 UD |
1250 | |
1251 | @node CPU Affinity | |
1252 | @subsection Limiting execution to certain CPUs | |
1253 | ||
1254 | On a multi-processor system the operating system usually distributes | |
1255 | the different processes which are runnable on all available CPUs in a | |
1256 | way which allows the system to work most efficiently. Which processes | |
1257 | and threads run can be to some extend be control with the scheduling | |
1258 | functionality described in the last sections. But which CPU finally | |
1259 | executes which process or thread is not covered. | |
1260 | ||
1261 | There are a number of reasons why a program might want to have control | |
1262 | over this aspect of the system as well: | |
1263 | ||
1264 | @itemize @bullet | |
1265 | @item | |
1266 | One thread or process is responsible for absolutely critical work | |
1267 | which under no circumstances must be interrupted or hindered from | |
1268 | making process by other process or threads using CPU resources. In | |
1269 | this case the special process would be confined to a CPU which no | |
1270 | other process or thread is allowed to use. | |
1271 | ||
1272 | @item | |
1273 | The access to certain resources (RAM, I/O ports) has different costs | |
1274 | from different CPUs. This is the case in NUMA (Non-Uniform Memory | |
11bf311e | 1275 | Architecture) machines. Preferably memory should be accessed locally |
d9997a45 UD |
1276 | but this requirement is usually not visible to the scheduler. |
1277 | Therefore forcing a process or thread to the CPUs which have local | |
1278 | access to the mostly used memory helps to significantly boost the | |
1279 | performance. | |
1280 | ||
1281 | @item | |
1282 | In controlled runtimes resource allocation and book-keeping work (for | |
1283 | instance garbage collection) is performance local to processors. This | |
1284 | can help to reduce locking costs if the resources do not have to be | |
1285 | protected from concurrent accesses from different processors. | |
1286 | @end itemize | |
1287 | ||
1288 | The POSIX standard up to this date is of not much help to solve this | |
1289 | problem. The Linux kernel provides a set of interfaces to allow | |
1290 | specifying @emph{affinity sets} for a process. The scheduler will | |
1291 | schedule the thread or process on on CPUs specified by the affinity | |
1292 | masks. The interfaces which the GNU C library define follow to some | |
1293 | extend the Linux kernel interface. | |
1294 | ||
1295 | @comment sched.h | |
1296 | @comment GNU | |
1297 | @deftp {Data Type} cpu_set_t | |
1298 | This data set is a bitset where each bit represents a CPU. How the | |
1299 | system's CPUs are mapped to bits in the bitset is system dependent. | |
1300 | The data type has a fixed size; in the unlikely case that the number | |
1301 | of bits are not sufficient to describe the CPUs of the system a | |
1302 | different interface has to be used. | |
1303 | ||
1304 | This type is a GNU extension and is defined in @file{sched.h}. | |
1305 | @end deftp | |
1306 | ||
1307 | To manipulate the bitset, to set and reset bits, a number of macros is | |
1308 | defined. Some of the macros take a CPU number as a parameter. Here | |
1309 | it is important to never exceed the size of the bitset. The following | |
1310 | macro specifies the number of bits in the @code{cpu_set_t} bitset. | |
1311 | ||
1312 | @comment sched.h | |
1313 | @comment GNU | |
1314 | @deftypevr Macro int CPU_SETSIZE | |
1315 | The value of this macro is the maximum number of CPUs which can be | |
1316 | handled with a @code{cpu_set_t} object. | |
1317 | @end deftypevr | |
1318 | ||
1319 | The type @code{cpu_set_t} should be considered opaque; all | |
1320 | manipulation should happen via the next four macros. | |
1321 | ||
1322 | @comment sched.h | |
1323 | @comment GNU | |
1324 | @deftypefn Macro void CPU_ZERO (cpu_set_t *@var{set}) | |
1325 | This macro initializes the CPU set @var{set} to be the empty set. | |
1326 | ||
1327 | This macro is a GNU extension and is defined in @file{sched.h}. | |
1328 | @end deftypefn | |
1329 | ||
1330 | @comment sched.h | |
1331 | @comment GNU | |
1332 | @deftypefn Macro void CPU_SET (int @var{cpu}, cpu_set_t *@var{set}) | |
1333 | This macro adds @var{cpu} to the CPU set @var{set}. | |
1334 | ||
1335 | The @var{cpu} parameter must not have side effects since it is | |
1336 | evaluated more than once. | |
1337 | ||
1338 | This macro is a GNU extension and is defined in @file{sched.h}. | |
1339 | @end deftypefn | |
1340 | ||
1341 | @comment sched.h | |
1342 | @comment GNU | |
1343 | @deftypefn Macro void CPU_CLR (int @var{cpu}, cpu_set_t *@var{set}) | |
1344 | This macro removes @var{cpu} from the CPU set @var{set}. | |
1345 | ||
1346 | The @var{cpu} parameter must not have side effects since it is | |
1347 | evaluated more than once. | |
1348 | ||
1349 | This macro is a GNU extension and is defined in @file{sched.h}. | |
1350 | @end deftypefn | |
1351 | ||
1352 | @comment sched.h | |
1353 | @comment GNU | |
1354 | @deftypefn Macro int CPU_ISSET (int @var{cpu}, const cpu_set_t *@var{set}) | |
1355 | This macro returns a nonzero value (true) if @var{cpu} is a member | |
1356 | of the CPU set @var{set}, and zero (false) otherwise. | |
1357 | ||
1358 | The @var{cpu} parameter must not have side effects since it is | |
1359 | evaluated more than once. | |
1360 | ||
1361 | This macro is a GNU extension and is defined in @file{sched.h}. | |
1362 | @end deftypefn | |
1363 | ||
1364 | ||
1365 | CPU bitsets can be constructed from scratch or the currently installed | |
1366 | affinity mask can be retrieved from the system. | |
1367 | ||
1368 | @comment sched.h | |
1369 | @comment GNU | |
6f0b2e1f | 1370 | @deftypefun int sched_getaffinity (pid_t @var{pid}, size_t @var{cpusetsize}, cpu_set_t *@var{cpuset}) |
d9997a45 UD |
1371 | |
1372 | This functions stores the CPU affinity mask for the process or thread | |
6f0b2e1f RM |
1373 | with the ID @var{pid} in the @var{cpusetsize} bytes long bitmap |
1374 | pointed to by @var{cpuset}. If successful, the function always | |
1375 | initializes all bits in the @code{cpu_set_t} object and returns zero. | |
d9997a45 UD |
1376 | |
1377 | If @var{pid} does not correspond to a process or thread on the system | |
1378 | the or the function fails for some other reason, it returns @code{-1} | |
1379 | and @code{errno} is set to represent the error condition. | |
1380 | ||
1381 | @table @code | |
1382 | @item ESRCH | |
1383 | No process or thread with the given ID found. | |
1384 | ||
1385 | @item EFAULT | |
1386 | The pointer @var{cpuset} is does not point to a valid object. | |
1387 | @end table | |
1388 | ||
1389 | This function is a GNU extension and is declared in @file{sched.h}. | |
1390 | @end deftypefun | |
1391 | ||
1392 | Note that it is not portably possible to use this information to | |
1393 | retrieve the information for different POSIX threads. A separate | |
1394 | interface must be provided for that. | |
1395 | ||
1396 | @comment sched.h | |
1397 | @comment GNU | |
6f0b2e1f | 1398 | @deftypefun int sched_setaffinity (pid_t @var{pid}, size_t @var{cpusetsize}, const cpu_set_t *@var{cpuset}) |
d9997a45 | 1399 | |
6f0b2e1f RM |
1400 | This function installs the @var{cpusetsize} bytes long affinity mask |
1401 | pointed to by @var{cpuset} for the process or thread with the ID @var{pid}. | |
1402 | If successful the function returns zero and the scheduler will in future | |
1403 | take the affinity information into account. | |
d9997a45 UD |
1404 | |
1405 | If the function fails it will return @code{-1} and @code{errno} is set | |
1406 | to the error code: | |
1407 | ||
1408 | @table @code | |
1409 | @item ESRCH | |
1410 | No process or thread with the given ID found. | |
1411 | ||
1412 | @item EFAULT | |
1413 | The pointer @var{cpuset} is does not point to a valid object. | |
1414 | ||
1415 | @item EINVAL | |
1416 | The bitset is not valid. This might mean that the affinity set might | |
1417 | not leave a processor for the process or thread to run on. | |
1418 | @end table | |
1419 | ||
1420 | This function is a GNU extension and is declared in @file{sched.h}. | |
1421 | @end deftypefun | |
1422 | ||
1423 | ||
b642f101 UD |
1424 | @node Memory Resources |
1425 | @section Querying memory available resources | |
1426 | ||
1427 | The amount of memory available in the system and the way it is organized | |
1428 | determines oftentimes the way programs can and have to work. For | |
5a7eedfb | 1429 | functions like @code{mmap} it is necessary to know about the size of |
b642f101 UD |
1430 | individual memory pages and knowing how much memory is available enables |
1431 | a program to select appropriate sizes for, say, caches. Before we get | |
1432 | into these details a few words about memory subsystems in traditional | |
5a7eedfb | 1433 | Unix systems will be given. |
b642f101 UD |
1434 | |
1435 | @menu | |
1436 | * Memory Subsystem:: Overview about traditional Unix memory handling. | |
1437 | * Query Memory Parameters:: How to get information about the memory | |
1438 | subsystem? | |
1439 | @end menu | |
1440 | ||
1441 | @node Memory Subsystem | |
1442 | @subsection Overview about traditional Unix memory handling | |
1443 | ||
1444 | @cindex address space | |
1445 | @cindex physical memory | |
1446 | @cindex physical address | |
1447 | Unix systems normally provide processes virtual address spaces. This | |
1448 | means that the addresses of the memory regions do not have to correspond | |
1449 | directly to the addresses of the actual physical memory which stores the | |
1450 | data. An extra level of indirection is introduced which translates | |
1451 | virtual addresses into physical addresses. This is normally done by the | |
1452 | hardware of the processor. | |
1453 | ||
1454 | @cindex shared memory | |
1455 | Using a virtual address space has several advantage. The most important | |
1456 | is process isolation. The different processes running on the system | |
1457 | cannot interfere directly with each other. No process can write into | |
1458 | the address space of another process (except when shared memory is used | |
1459 | but then it is wanted and controlled). | |
1460 | ||
1461 | Another advantage of virtual memory is that the address space the | |
1462 | processes see can actually be larger than the physical memory available. | |
1463 | The physical memory can be extended by storage on an external media | |
1464 | where the content of currently unused memory regions is stored. The | |
1465 | address translation can then intercept accesses to these memory regions | |
1466 | and make memory content available again by loading the data back into | |
1467 | memory. This concept makes it necessary that programs which have to use | |
1468 | lots of memory know the difference between available virtual address | |
1469 | space and available physical memory. If the working set of virtual | |
1470 | memory of all the processes is larger than the available physical memory | |
1471 | the system will slow down dramatically due to constant swapping of | |
1472 | memory content from the memory to the storage media and back. This is | |
1473 | called ``thrashing''. | |
1474 | @cindex thrashing | |
1475 | ||
1476 | @cindex memory page | |
1477 | @cindex page, memory | |
1478 | A final aspect of virtual memory which is important and follows from | |
1479 | what is said in the last paragraph is the granularity of the virtual | |
1480 | address space handling. When we said that the virtual address handling | |
1481 | stores memory content externally it cannot do this on a byte-by-byte | |
1482 | basis. The administrative overhead does not allow this (leaving alone | |
1483 | the processor hardware). Instead several thousand bytes are handled | |
1484 | together and form a @dfn{page}. The size of each page is always a power | |
1485 | of two byte. The smallest page size in use today is 4096, with 8192, | |
1486 | 16384, and 65536 being other popular sizes. | |
1487 | ||
1488 | @node Query Memory Parameters | |
1489 | @subsection How to get information about the memory subsystem? | |
1490 | ||
1491 | The page size of the virtual memory the process sees is essential to | |
1492 | know in several situations. Some programming interface (e.g., | |
1493 | @code{mmap}, @pxref{Memory-mapped I/O}) require the user to provide | |
1494 | information adjusted to the page size. In the case of @code{mmap} is it | |
1495 | necessary to provide a length argument which is a multiple of the page | |
1496 | size. Another place where the knowledge about the page size is useful | |
1497 | is in memory allocation. If one allocates pieces of memory in larger | |
1498 | chunks which are then subdivided by the application code it is useful to | |
1499 | adjust the size of the larger blocks to the page size. If the total | |
1500 | memory requirement for the block is close (but not larger) to a multiple | |
1501 | of the page size the kernel's memory handling can work more effectively | |
1502 | since it only has to allocate memory pages which are fully used. (To do | |
1503 | this optimization it is necessary to know a bit about the memory | |
1504 | allocator which will require a bit of memory itself for each block and | |
1505 | this overhead must not push the total size over the page size multiple. | |
1506 | ||
1507 | The page size traditionally was a compile time constant. But recent | |
1508 | development of processors changed this. Processors now support | |
1509 | different page sizes and they can possibly even vary among different | |
1510 | processes on the same system. Therefore the system should be queried at | |
1511 | runtime about the current page size and no assumptions (except about it | |
1512 | being a power of two) should be made. | |
1513 | ||
1514 | @vindex _SC_PAGESIZE | |
1515 | The correct interface to query about the page size is @code{sysconf} | |
1516 | (@pxref{Sysconf Definition}) with the parameter @code{_SC_PAGESIZE}. | |
1517 | There is a much older interface available, too. | |
1518 | ||
1519 | @comment unistd.h | |
1520 | @comment BSD | |
1521 | @deftypefun int getpagesize (void) | |
1522 | The @code{getpagesize} function returns the page size of the process. | |
1523 | This value is fixed for the runtime of the process but can vary in | |
1524 | different runs of the application. | |
1525 | ||
1526 | The function is declared in @file{unistd.h}. | |
1527 | @end deftypefun | |
1528 | ||
1529 | Widely available on @w{System V} derived systems is a method to get | |
1530 | information about the physical memory the system has. The call | |
1531 | ||
1532 | @vindex _SC_PHYS_PAGES | |
1533 | @cindex sysconf | |
1534 | @smallexample | |
1535 | sysconf (_SC_PHYS_PAGES) | |
1536 | @end smallexample | |
1537 | ||
cb4fe8a2 UD |
1538 | @noindent |
1539 | returns the total number of pages of physical the system has. | |
b642f101 UD |
1540 | This does not mean all this memory is available. This information can |
1541 | be found using | |
1542 | ||
1543 | @vindex _SC_AVPHYS_PAGES | |
1544 | @cindex sysconf | |
1545 | @smallexample | |
1546 | sysconf (_SC_AVPHYS_PAGES) | |
1547 | @end smallexample | |
1548 | ||
1549 | These two values help to optimize applications. The value returned for | |
1550 | @code{_SC_AVPHYS_PAGES} is the amount of memory the application can use | |
1551 | without hindering any other process (given that no other process | |
1552 | increases its memory usage). The value returned for | |
1553 | @code{_SC_PHYS_PAGES} is more or less a hard limit for the working set. | |
1554 | If all applications together constantly use more than that amount of | |
1555 | memory the system is in trouble. | |
1556 | ||
cb4fe8a2 UD |
1557 | The GNU C library provides in addition to these already described way to |
1558 | get this information two functions. They are declared in the file | |
1559 | @file{sys/sysinfo.h}. Programmers should prefer to use the | |
1560 | @code{sysconf} method described above. | |
1561 | ||
1562 | @comment sys/sysinfo.h | |
1563 | @comment GNU | |
4c78249d | 1564 | @deftypefun {long int} get_phys_pages (void) |
cb4fe8a2 UD |
1565 | The @code{get_phys_pages} function returns the total number of pages of |
1566 | physical the system has. To get the amount of memory this number has to | |
1567 | be multiplied by the page size. | |
1568 | ||
1569 | This function is a GNU extension. | |
1570 | @end deftypefun | |
1571 | ||
1572 | @comment sys/sysinfo.h | |
1573 | @comment GNU | |
4c78249d | 1574 | @deftypefun {long int} get_avphys_pages (void) |
cb4fe8a2 UD |
1575 | The @code{get_phys_pages} function returns the number of available pages of |
1576 | physical the system has. To get the amount of memory this number has to | |
1577 | be multiplied by the page size. | |
1578 | ||
1579 | This function is a GNU extension. | |
1580 | @end deftypefun | |
1581 | ||
b642f101 UD |
1582 | @node Processor Resources |
1583 | @section Learn about the processors available | |
1584 | ||
1585 | The use of threads or processes with shared memory allows an application | |
1586 | to take advantage of all the processing power a system can provide. If | |
1587 | the task can be parallelized the optimal way to write an application is | |
1588 | to have at any time as many processes running as there are processors. | |
1589 | To determine the number of processors available to the system one can | |
1590 | run | |
1591 | ||
1592 | @vindex _SC_NPROCESSORS_CONF | |
1593 | @cindex sysconf | |
1594 | @smallexample | |
1595 | sysconf (_SC_NPROCESSORS_CONF) | |
1596 | @end smallexample | |
1597 | ||
1598 | @noindent | |
1599 | which returns the number of processors the operating system configured. | |
1600 | But it might be possible for the operating system to disable individual | |
1601 | processors and so the call | |
1602 | ||
1603 | @vindex _SC_NPROCESSORS_ONLN | |
1604 | @cindex sysconf | |
1605 | @smallexample | |
1606 | sysconf (_SC_NPROCESSORS_ONLN) | |
1607 | @end smallexample | |
1608 | ||
1609 | @noindent | |
1610 | returns the number of processors which are currently inline (i.e., | |
1611 | available). | |
e4cf5229 | 1612 | |
cb4fe8a2 UD |
1613 | For these two pieces of information the GNU C library also provides |
1614 | functions to get the information directly. The functions are declared | |
1615 | in @file{sys/sysinfo.h}. | |
1616 | ||
1617 | @comment sys/sysinfo.h | |
1618 | @comment GNU | |
1619 | @deftypefun int get_nprocs_conf (void) | |
1620 | The @code{get_nprocs_conf} function returns the number of processors the | |
1621 | operating system configured. | |
1622 | ||
1623 | This function is a GNU extension. | |
1624 | @end deftypefun | |
1625 | ||
1626 | @comment sys/sysinfo.h | |
1627 | @comment GNU | |
1628 | @deftypefun int get_nprocs (void) | |
1629 | The @code{get_nprocs} function returns the number of available processors. | |
1630 | ||
1631 | This function is a GNU extension. | |
1632 | @end deftypefun | |
1633 | ||
e4cf5229 UD |
1634 | @cindex load average |
1635 | Before starting more threads it should be checked whether the processors | |
1636 | are not already overused. Unix systems calculate something called the | |
1637 | @dfn{load average}. This is a number indicating how many processes were | |
1638 | running. This number is average over different periods of times | |
1639 | (normally 1, 5, and 15 minutes). | |
1640 | ||
1641 | @comment stdlib.h | |
1642 | @comment BSD | |
1643 | @deftypefun int getloadavg (double @var{loadavg}[], int @var{nelem}) | |
1644 | This function gets the 1, 5 and 15 minute load averages of the | |
1645 | system. The values are placed in @var{loadavg}. @code{getloadavg} will | |
1646 | place at most @var{nelem} elements into the array but never more than | |
1647 | three elements. The return value is the number of elements written to | |
1648 | @var{loadavg}, or -1 on error. | |
1649 | ||
1650 | This function is declared in @file{stdlib.h}. | |
1651 | @end deftypefun |