]>
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}) | |
c8ce789c AO |
28 | @safety{@prelim{}@mtsafe{}@assafe{}@acsafe{}} |
29 | @c On HURD, this calls task_info 3 times. On UNIX, it's a syscall. | |
5ce8f203 UD |
30 | This function reports resource usage totals for processes specified by |
31 | @var{processes}, storing the information in @code{*@var{rusage}}. | |
32 | ||
33 | In most systems, @var{processes} has only two valid values: | |
34 | ||
35 | @table @code | |
36 | @comment sys/resource.h | |
37 | @comment BSD | |
38 | @item RUSAGE_SELF | |
39 | Just the current process. | |
40 | ||
41 | @comment sys/resource.h | |
42 | @comment BSD | |
43 | @item RUSAGE_CHILDREN | |
44 | All child processes (direct and indirect) that have already terminated. | |
45 | @end table | |
46 | ||
5ce8f203 UD |
47 | The return value of @code{getrusage} is zero for success, and @code{-1} |
48 | for failure. | |
49 | ||
50 | @table @code | |
51 | @item EINVAL | |
52 | The argument @var{processes} is not valid. | |
53 | @end table | |
54 | @end deftypefun | |
55 | ||
56 | One way of getting resource usage for a particular child process is with | |
57 | the function @code{wait4}, which returns totals for a child when it | |
58 | terminates. @xref{BSD Wait Functions}. | |
59 | ||
60 | @comment sys/resource.h | |
61 | @comment BSD | |
62 | @deftp {Data Type} {struct rusage} | |
63 | This data type stores various resource usage statistics. It has the | |
64 | following members, and possibly others: | |
65 | ||
66 | @table @code | |
67 | @item struct timeval ru_utime | |
68 | Time spent executing user instructions. | |
69 | ||
70 | @item struct timeval ru_stime | |
71 | Time spent in operating system code on behalf of @var{processes}. | |
72 | ||
73 | @item long int ru_maxrss | |
74 | The maximum resident set size used, in kilobytes. That is, the maximum | |
75 | number of kilobytes of physical memory that @var{processes} used | |
76 | simultaneously. | |
77 | ||
78 | @item long int ru_ixrss | |
79 | An integral value expressed in kilobytes times ticks of execution, which | |
80 | indicates the amount of memory used by text that was shared with other | |
81 | processes. | |
82 | ||
83 | @item long int ru_idrss | |
84 | An integral value expressed the same way, which is the amount of | |
85 | unshared memory used for data. | |
86 | ||
87 | @item long int ru_isrss | |
88 | An integral value expressed the same way, which is the amount of | |
89 | unshared memory used for stack space. | |
90 | ||
91 | @item long int ru_minflt | |
92 | The number of page faults which were serviced without requiring any I/O. | |
93 | ||
94 | @item long int ru_majflt | |
95 | The number of page faults which were serviced by doing I/O. | |
96 | ||
97 | @item long int ru_nswap | |
98 | The number of times @var{processes} was swapped entirely out of main memory. | |
99 | ||
100 | @item long int ru_inblock | |
101 | The number of times the file system had to read from the disk on behalf | |
102 | of @var{processes}. | |
103 | ||
104 | @item long int ru_oublock | |
105 | The number of times the file system had to write to the disk on behalf | |
106 | of @var{processes}. | |
107 | ||
108 | @item long int ru_msgsnd | |
109 | Number of IPC messages sent. | |
110 | ||
111 | @item long int ru_msgrcv | |
112 | Number of IPC messages received. | |
113 | ||
114 | @item long int ru_nsignals | |
115 | Number of signals received. | |
116 | ||
117 | @item long int ru_nvcsw | |
118 | The number of times @var{processes} voluntarily invoked a context switch | |
119 | (usually to wait for some service). | |
120 | ||
121 | @item long int ru_nivcsw | |
122 | The number of times an involuntary context switch took place (because | |
123 | a time slice expired, or another process of higher priority was | |
124 | scheduled). | |
125 | @end table | |
126 | @end deftp | |
127 | ||
128 | @code{vtimes} is a historical function that does some of what | |
129 | @code{getrusage} does. @code{getrusage} is a better choice. | |
130 | ||
131 | @code{vtimes} and its @code{vtimes} data structure are declared in | |
132 | @file{sys/vtimes.h}. | |
133 | @pindex sys/vtimes.h | |
5ce8f203 | 134 | |
8ded91fb RM |
135 | @comment sys/vtimes.h |
136 | @deftypefun int vtimes (struct vtimes *@var{current}, struct vtimes *@var{child}) | |
c8ce789c AO |
137 | @safety{@prelim{}@mtsafe{}@assafe{}@acsafe{}} |
138 | @c Calls getrusage twice. | |
5ce8f203 UD |
139 | |
140 | @code{vtimes} reports resource usage totals for a process. | |
141 | ||
142 | If @var{current} is non-null, @code{vtimes} stores resource usage totals for | |
143 | the invoking process alone in the structure to which it points. If | |
144 | @var{child} is non-null, @code{vtimes} stores resource usage totals for all | |
145 | past children (which have terminated) of the invoking process in the structure | |
146 | to which it points. | |
147 | ||
148 | @deftp {Data Type} {struct vtimes} | |
149 | This data type contains information about the resource usage of a process. | |
150 | Each member corresponds to a member of the @code{struct rusage} data type | |
151 | described above. | |
152 | ||
153 | @table @code | |
154 | @item vm_utime | |
155 | User CPU time. Analogous to @code{ru_utime} in @code{struct rusage} | |
156 | @item vm_stime | |
157 | System CPU time. Analogous to @code{ru_stime} in @code{struct rusage} | |
158 | @item vm_idsrss | |
159 | Data and stack memory. The sum of the values that would be reported as | |
160 | @code{ru_idrss} and @code{ru_isrss} in @code{struct rusage} | |
161 | @item vm_ixrss | |
162 | Shared memory. Analogous to @code{ru_ixrss} in @code{struct rusage} | |
163 | @item vm_maxrss | |
164 | Maximent resident set size. Analogous to @code{ru_maxrss} in | |
165 | @code{struct rusage} | |
166 | @item vm_majflt | |
167 | Major page faults. Analogous to @code{ru_majflt} in @code{struct rusage} | |
168 | @item vm_minflt | |
169 | Minor page faults. Analogous to @code{ru_minflt} in @code{struct rusage} | |
170 | @item vm_nswap | |
171 | Swap count. Analogous to @code{ru_nswap} in @code{struct rusage} | |
172 | @item vm_inblk | |
173 | Disk reads. Analogous to @code{ru_inblk} in @code{struct rusage} | |
174 | @item vm_oublk | |
175 | Disk writes. Analogous to @code{ru_oublk} in @code{struct rusage} | |
176 | @end table | |
177 | @end deftp | |
178 | ||
179 | ||
180 | The return value is zero if the function succeeds; @code{-1} otherwise. | |
181 | ||
182 | ||
183 | ||
184 | @end deftypefun | |
185 | An additional historical function for examining resource usage, | |
186 | @code{vtimes}, is supported but not documented here. It is declared in | |
187 | @file{sys/vtimes.h}. | |
188 | ||
189 | @node Limits on Resources | |
190 | @section Limiting Resource Usage | |
191 | @cindex resource limits | |
192 | @cindex limits on resource usage | |
193 | @cindex usage limits | |
194 | ||
195 | You can specify limits for the resource usage of a process. When the | |
196 | process tries to exceed a limit, it may get a signal, or the system call | |
197 | by which it tried to do so may fail, depending on the resource. Each | |
198 | process initially inherits its limit values from its parent, but it can | |
199 | subsequently change them. | |
200 | ||
201 | There are two per-process limits associated with a resource: | |
202 | @cindex limit | |
203 | ||
204 | @table @dfn | |
205 | @item current limit | |
206 | The current limit is the value the system will not allow usage to | |
207 | exceed. It is also called the ``soft limit'' because the process being | |
208 | limited can generally raise the current limit at will. | |
209 | @cindex current limit | |
210 | @cindex soft limit | |
211 | ||
212 | @item maximum limit | |
213 | The maximum limit is the maximum value to which a process is allowed to | |
214 | set its current limit. It is also called the ``hard limit'' because | |
215 | there is no way for a process to get around it. A process may lower | |
216 | its own maximum limit, but only the superuser may increase a maximum | |
217 | limit. | |
218 | @cindex maximum limit | |
219 | @cindex hard limit | |
220 | @end table | |
221 | ||
222 | @pindex sys/resource.h | |
223 | The symbols for use with @code{getrlimit}, @code{setrlimit}, | |
0bc93a2f | 224 | @code{getrlimit64}, and @code{setrlimit64} are defined in |
5ce8f203 UD |
225 | @file{sys/resource.h}. |
226 | ||
227 | @comment sys/resource.h | |
228 | @comment BSD | |
229 | @deftypefun int getrlimit (int @var{resource}, struct rlimit *@var{rlp}) | |
c8ce789c AO |
230 | @safety{@prelim{}@mtsafe{}@assafe{}@acsafe{}} |
231 | @c Direct syscall on most systems. | |
5ce8f203 UD |
232 | Read the current and maximum limits for the resource @var{resource} |
233 | and store them in @code{*@var{rlp}}. | |
234 | ||
235 | The return value is @code{0} on success and @code{-1} on failure. The | |
236 | only possible @code{errno} error condition is @code{EFAULT}. | |
237 | ||
238 | When the sources are compiled with @code{_FILE_OFFSET_BITS == 64} on a | |
239 | 32-bit system this function is in fact @code{getrlimit64}. Thus, the | |
240 | LFS interface transparently replaces the old interface. | |
241 | @end deftypefun | |
242 | ||
243 | @comment sys/resource.h | |
244 | @comment Unix98 | |
245 | @deftypefun int getrlimit64 (int @var{resource}, struct rlimit64 *@var{rlp}) | |
c8ce789c AO |
246 | @safety{@prelim{}@mtsafe{}@assafe{}@acsafe{}} |
247 | @c Direct syscall on most systems, wrapper to getrlimit otherwise. | |
5ce8f203 UD |
248 | This function is similar to @code{getrlimit} but its second parameter is |
249 | a pointer to a variable of type @code{struct rlimit64}, which allows it | |
250 | to read values which wouldn't fit in the member of a @code{struct | |
251 | rlimit}. | |
252 | ||
253 | If the sources are compiled with @code{_FILE_OFFSET_BITS == 64} on a | |
254 | 32-bit machine, this function is available under the name | |
255 | @code{getrlimit} and so transparently replaces the old interface. | |
256 | @end deftypefun | |
257 | ||
258 | @comment sys/resource.h | |
259 | @comment BSD | |
260 | @deftypefun int setrlimit (int @var{resource}, const struct rlimit *@var{rlp}) | |
c8ce789c AO |
261 | @safety{@prelim{}@mtsafe{}@assafe{}@acsafe{}} |
262 | @c Direct syscall on most systems; lock-taking critical section on HURD. | |
5ce8f203 UD |
263 | Store the current and maximum limits for the resource @var{resource} |
264 | in @code{*@var{rlp}}. | |
265 | ||
266 | The return value is @code{0} on success and @code{-1} on failure. The | |
267 | following @code{errno} error condition is possible: | |
268 | ||
269 | @table @code | |
270 | @item EPERM | |
271 | @itemize @bullet | |
272 | @item | |
273 | The process tried to raise a current limit beyond the maximum limit. | |
274 | ||
275 | @item | |
276 | The process tried to raise a maximum limit, but is not superuser. | |
277 | @end itemize | |
278 | @end table | |
279 | ||
280 | When the sources are compiled with @code{_FILE_OFFSET_BITS == 64} on a | |
281 | 32-bit system this function is in fact @code{setrlimit64}. Thus, the | |
282 | LFS interface transparently replaces the old interface. | |
283 | @end deftypefun | |
284 | ||
285 | @comment sys/resource.h | |
286 | @comment Unix98 | |
287 | @deftypefun int setrlimit64 (int @var{resource}, const struct rlimit64 *@var{rlp}) | |
c8ce789c AO |
288 | @safety{@prelim{}@mtsafe{}@assafe{}@acsafe{}} |
289 | @c Wrapper for setrlimit or direct syscall. | |
5ce8f203 UD |
290 | This function is similar to @code{setrlimit} but its second parameter is |
291 | a pointer to a variable of type @code{struct rlimit64} which allows it | |
292 | to set values which wouldn't fit in the member of a @code{struct | |
293 | rlimit}. | |
294 | ||
295 | If the sources are compiled with @code{_FILE_OFFSET_BITS == 64} on a | |
296 | 32-bit machine this function is available under the name | |
297 | @code{setrlimit} and so transparently replaces the old interface. | |
298 | @end deftypefun | |
299 | ||
300 | @comment sys/resource.h | |
301 | @comment BSD | |
302 | @deftp {Data Type} {struct rlimit} | |
303 | This structure is used with @code{getrlimit} to receive limit values, | |
304 | and with @code{setrlimit} to specify limit values for a particular process | |
305 | and resource. It has two fields: | |
306 | ||
307 | @table @code | |
308 | @item rlim_t rlim_cur | |
309 | The current limit | |
310 | ||
311 | @item rlim_t rlim_max | |
312 | The maximum limit. | |
313 | @end table | |
314 | ||
315 | For @code{getrlimit}, the structure is an output; it receives the current | |
316 | values. For @code{setrlimit}, it specifies the new values. | |
317 | @end deftp | |
318 | ||
319 | For the LFS functions a similar type is defined in @file{sys/resource.h}. | |
320 | ||
321 | @comment sys/resource.h | |
322 | @comment Unix98 | |
323 | @deftp {Data Type} {struct rlimit64} | |
324 | This structure is analogous to the @code{rlimit} structure above, but | |
325 | its components have wider ranges. It has two fields: | |
326 | ||
327 | @table @code | |
328 | @item rlim64_t rlim_cur | |
329 | This is analogous to @code{rlimit.rlim_cur}, but with a different type. | |
330 | ||
331 | @item rlim64_t rlim_max | |
332 | This is analogous to @code{rlimit.rlim_max}, but with a different type. | |
333 | @end table | |
334 | ||
335 | @end deftp | |
336 | ||
337 | Here is a list of resources for which you can specify a limit. Memory | |
338 | and file sizes are measured in bytes. | |
339 | ||
2fe82ca6 | 340 | @vtable @code |
5ce8f203 UD |
341 | @comment sys/resource.h |
342 | @comment BSD | |
343 | @item RLIMIT_CPU | |
5ce8f203 UD |
344 | The maximum amount of CPU time the process can use. If it runs for |
345 | longer than this, it gets a signal: @code{SIGXCPU}. The value is | |
346 | measured in seconds. @xref{Operation Error Signals}. | |
347 | ||
348 | @comment sys/resource.h | |
349 | @comment BSD | |
350 | @item RLIMIT_FSIZE | |
5ce8f203 UD |
351 | The maximum size of file the process can create. Trying to write a |
352 | larger file causes a signal: @code{SIGXFSZ}. @xref{Operation Error | |
353 | Signals}. | |
354 | ||
355 | @comment sys/resource.h | |
356 | @comment BSD | |
357 | @item RLIMIT_DATA | |
5ce8f203 UD |
358 | The maximum size of data memory for the process. If the process tries |
359 | to allocate data memory beyond this amount, the allocation function | |
360 | fails. | |
361 | ||
362 | @comment sys/resource.h | |
363 | @comment BSD | |
364 | @item RLIMIT_STACK | |
5ce8f203 UD |
365 | The maximum stack size for the process. If the process tries to extend |
366 | its stack past this size, it gets a @code{SIGSEGV} signal. | |
367 | @xref{Program Error Signals}. | |
368 | ||
369 | @comment sys/resource.h | |
370 | @comment BSD | |
371 | @item RLIMIT_CORE | |
5ce8f203 UD |
372 | The maximum size core file that this process can create. If the process |
373 | terminates and would dump a core file larger than this, then no core | |
374 | file is created. So setting this limit to zero prevents core files from | |
375 | ever being created. | |
376 | ||
377 | @comment sys/resource.h | |
378 | @comment BSD | |
379 | @item RLIMIT_RSS | |
5ce8f203 UD |
380 | The maximum amount of physical memory that this process should get. |
381 | This parameter is a guide for the system's scheduler and memory | |
382 | allocator; the system may give the process more memory when there is a | |
383 | surplus. | |
384 | ||
385 | @comment sys/resource.h | |
386 | @comment BSD | |
387 | @item RLIMIT_MEMLOCK | |
388 | The maximum amount of memory that can be locked into physical memory (so | |
389 | it will never be paged out). | |
390 | ||
391 | @comment sys/resource.h | |
392 | @comment BSD | |
393 | @item RLIMIT_NPROC | |
394 | The maximum number of processes that can be created with the same user ID. | |
395 | If you have reached the limit for your user ID, @code{fork} will fail | |
396 | with @code{EAGAIN}. @xref{Creating a Process}. | |
397 | ||
398 | @comment sys/resource.h | |
399 | @comment BSD | |
400 | @item RLIMIT_NOFILE | |
5ce8f203 | 401 | @itemx RLIMIT_OFILE |
5ce8f203 UD |
402 | The maximum number of files that the process can open. If it tries to |
403 | open more files than this, its open attempt fails with @code{errno} | |
404 | @code{EMFILE}. @xref{Error Codes}. Not all systems support this limit; | |
405 | GNU does, and 4.4 BSD does. | |
406 | ||
407 | @comment sys/resource.h | |
408 | @comment Unix98 | |
409 | @item RLIMIT_AS | |
5ce8f203 UD |
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 | |
5ce8f203 UD |
418 | The number of different resource limits. Any valid @var{resource} |
419 | operand must be less than @code{RLIM_NLIMITS}. | |
2fe82ca6 | 420 | @end vtable |
5ce8f203 UD |
421 | |
422 | @comment sys/resource.h | |
423 | @comment BSD | |
8ded91fb | 424 | @deftypevr Constant rlim_t RLIM_INFINITY |
5ce8f203 UD |
425 | This constant stands for a value of ``infinity'' when supplied as |
426 | the limit value in @code{setrlimit}. | |
427 | @end deftypevr | |
428 | ||
429 | ||
430 | The following are historical functions to do some of what the functions | |
431 | above do. The functions above are better choices. | |
432 | ||
433 | @code{ulimit} and the command symbols are declared in @file{ulimit.h}. | |
434 | @pindex ulimit.h | |
5ce8f203 | 435 | |
b642f101 UD |
436 | @comment ulimit.h |
437 | @comment BSD | |
8ded91fb | 438 | @deftypefun {long int} ulimit (int @var{cmd}, @dots{}) |
c8ce789c AO |
439 | @safety{@prelim{}@mtsafe{}@assafe{}@acsafe{}} |
440 | @c Wrapper for getrlimit, setrlimit or | |
441 | @c sysconf(_SC_OPEN_MAX)->getdtablesize->getrlimit. | |
5ce8f203 UD |
442 | |
443 | @code{ulimit} gets the current limit or sets the current and maximum | |
444 | limit for a particular resource for the calling process according to the | |
d3e22d59 | 445 | command @var{cmd}. |
5ce8f203 UD |
446 | |
447 | If you are getting a limit, the command argument is the only argument. | |
448 | If you are setting a limit, there is a second argument: | |
449 | @code{long int} @var{limit} which is the value to which you are setting | |
450 | the limit. | |
451 | ||
452 | The @var{cmd} values and the operations they specify are: | |
2fe82ca6 | 453 | @vtable @code |
5ce8f203 UD |
454 | |
455 | @item GETFSIZE | |
456 | Get the current limit on the size of a file, in units of 512 bytes. | |
457 | ||
458 | @item SETFSIZE | |
459 | Set the current and maximum limit on the size of a file to @var{limit} * | |
460 | 512 bytes. | |
461 | ||
2fe82ca6 | 462 | @end vtable |
5ce8f203 UD |
463 | |
464 | There are also some other @var{cmd} values that may do things on some | |
465 | systems, but they are not supported. | |
466 | ||
467 | Only the superuser may increase a maximum limit. | |
468 | ||
469 | When you successfully get a limit, the return value of @code{ulimit} is | |
470 | that limit, which is never negative. When you successfully set a limit, | |
471 | the return value is zero. When the function fails, the return value is | |
472 | @code{-1} and @code{errno} is set according to the reason: | |
473 | ||
474 | @table @code | |
475 | @item EPERM | |
476 | A process tried to increase a maximum limit, but is not superuser. | |
477 | @end table | |
478 | ||
479 | ||
480 | @end deftypefun | |
481 | ||
482 | @code{vlimit} and its resource symbols are declared in @file{sys/vlimit.h}. | |
5ce8f203 | 483 | @pindex sys/vlimit.h |
5ce8f203 | 484 | |
b642f101 UD |
485 | @comment sys/vlimit.h |
486 | @comment BSD | |
5ce8f203 | 487 | @deftypefun int vlimit (int @var{resource}, int @var{limit}) |
c8ce789c AO |
488 | @safety{@prelim{}@mtunsafe{@mtasurace{:setrlimit}}@asunsafe{}@acsafe{}} |
489 | @c It calls getrlimit and modifies the rlim_cur field before calling | |
490 | @c setrlimit. There's a window for a concurrent call to setrlimit that | |
491 | @c modifies e.g. rlim_max, which will be lost if running as super-user. | |
5ce8f203 UD |
492 | |
493 | @code{vlimit} sets the current limit for a resource for a process. | |
494 | ||
495 | @var{resource} identifies the resource: | |
496 | ||
2fe82ca6 | 497 | @vtable @code |
5ce8f203 UD |
498 | @item LIM_CPU |
499 | Maximum CPU time. Same as @code{RLIMIT_CPU} for @code{setrlimit}. | |
500 | @item LIM_FSIZE | |
501 | Maximum file size. Same as @code{RLIMIT_FSIZE} for @code{setrlimit}. | |
502 | @item LIM_DATA | |
503 | Maximum data memory. Same as @code{RLIMIT_DATA} for @code{setrlimit}. | |
504 | @item LIM_STACK | |
505 | Maximum stack size. Same as @code{RLIMIT_STACK} for @code{setrlimit}. | |
506 | @item LIM_CORE | |
507 | Maximum core file size. Same as @code{RLIMIT_COR} for @code{setrlimit}. | |
508 | @item LIM_MAXRSS | |
509 | Maximum physical memory. Same as @code{RLIMIT_RSS} for @code{setrlimit}. | |
2fe82ca6 | 510 | @end vtable |
5ce8f203 UD |
511 | |
512 | The return value is zero for success, and @code{-1} with @code{errno} set | |
513 | accordingly for failure: | |
514 | ||
515 | @table @code | |
516 | @item EPERM | |
517 | The process tried to set its current limit beyond its maximum limit. | |
518 | @end table | |
519 | ||
520 | @end deftypefun | |
521 | ||
522 | @node Priority | |
639c6286 | 523 | @section Process CPU Priority And Scheduling |
5ce8f203 | 524 | @cindex process priority |
639c6286 | 525 | @cindex cpu priority |
5ce8f203 UD |
526 | @cindex priority of a process |
527 | ||
639c6286 UD |
528 | When multiple processes simultaneously require CPU time, the system's |
529 | scheduling policy and process CPU priorities determine which processes | |
530 | get it. This section describes how that determination is made and | |
1f77f049 | 531 | @glibcadj{} functions to control it. |
639c6286 UD |
532 | |
533 | It is common to refer to CPU scheduling simply as scheduling and a | |
534 | process' CPU priority simply as the process' priority, with the CPU | |
535 | resource being implied. Bear in mind, though, that CPU time is not the | |
536 | only resource a process uses or that processes contend for. In some | |
537 | cases, it is not even particularly important. Giving a process a high | |
538 | ``priority'' may have very little effect on how fast a process runs with | |
539 | respect to other processes. The priorities discussed in this section | |
540 | apply only to CPU time. | |
541 | ||
542 | CPU scheduling is a complex issue and different systems do it in wildly | |
543 | different ways. New ideas continually develop and find their way into | |
544 | the intricacies of the various systems' scheduling algorithms. This | |
87b56f36 | 545 | section discusses the general concepts, some specifics of systems |
1f77f049 | 546 | that commonly use @theglibc{}, and some standards. |
639c6286 UD |
547 | |
548 | For simplicity, we talk about CPU contention as if there is only one CPU | |
549 | in the system. But all the same principles apply when a processor has | |
550 | multiple CPUs, and knowing that the number of processes that can run at | |
551 | any one time is equal to the number of CPUs, you can easily extrapolate | |
552 | the information. | |
553 | ||
554 | The functions described in this section are all defined by the POSIX.1 | |
95fdc6a0 | 555 | and POSIX.1b standards (the @code{sched@dots{}} functions are POSIX.1b). |
639c6286 UD |
556 | However, POSIX does not define any semantics for the values that these |
557 | functions get and set. In this chapter, the semantics are based on the | |
558 | Linux kernel's implementation of the POSIX standard. As you will see, | |
559 | the Linux implementation is quite the inverse of what the authors of the | |
560 | POSIX syntax had in mind. | |
561 | ||
562 | @menu | |
563 | * Absolute Priority:: The first tier of priority. Posix | |
564 | * Realtime Scheduling:: Scheduling among the process nobility | |
565 | * Basic Scheduling Functions:: Get/set scheduling policy, priority | |
566 | * Traditional Scheduling:: Scheduling among the vulgar masses | |
d9997a45 | 567 | * CPU Affinity:: Limiting execution to certain CPUs |
639c6286 UD |
568 | @end menu |
569 | ||
570 | ||
571 | ||
572 | @node Absolute Priority | |
573 | @subsection Absolute Priority | |
574 | @cindex absolute priority | |
575 | @cindex priority, absolute | |
576 | ||
577 | Every process has an absolute priority, and it is represented by a number. | |
578 | The higher the number, the higher the absolute priority. | |
579 | ||
580 | @cindex realtime CPU scheduling | |
581 | On systems of the past, and most systems today, all processes have | |
582 | absolute priority 0 and this section is irrelevant. In that case, | |
583 | @xref{Traditional Scheduling}. Absolute priorities were invented to | |
0bc93a2f | 584 | accommodate realtime systems, in which it is vital that certain processes |
639c6286 UD |
585 | be able to respond to external events happening in real time, which |
586 | means they cannot wait around while some other process that @emph{wants | |
587 | to}, but doesn't @emph{need to} run occupies the CPU. | |
588 | ||
589 | @cindex ready to run | |
590 | @cindex preemptive scheduling | |
591 | When two processes are in contention to use the CPU at any instant, the | |
592 | one with the higher absolute priority always gets it. This is true even if the | |
11bf311e | 593 | process with the lower priority is already using the CPU (i.e., the |
639c6286 UD |
594 | scheduling is preemptive). Of course, we're only talking about |
595 | processes that are running or ``ready to run,'' which means they are | |
596 | ready to execute instructions right now. When a process blocks to wait | |
597 | for something like I/O, its absolute priority is irrelevant. | |
598 | ||
599 | @cindex runnable process | |
48b22986 | 600 | @strong{NB:} The term ``runnable'' is a synonym for ``ready to run.'' |
639c6286 UD |
601 | |
602 | When two processes are running or ready to run and both have the same | |
603 | absolute priority, it's more interesting. In that case, who gets the | |
0bc93a2f | 604 | CPU is determined by the scheduling policy. If the processes have |
639c6286 UD |
605 | absolute priority 0, the traditional scheduling policy described in |
606 | @ref{Traditional Scheduling} applies. Otherwise, the policies described | |
607 | in @ref{Realtime Scheduling} apply. | |
608 | ||
609 | You normally give an absolute priority above 0 only to a process that | |
610 | can be trusted not to hog the CPU. Such processes are designed to block | |
611 | (or terminate) after relatively short CPU runs. | |
612 | ||
613 | A process begins life with the same absolute priority as its parent | |
614 | process. Functions described in @ref{Basic Scheduling Functions} can | |
615 | change it. | |
616 | ||
617 | Only a privileged process can change a process' absolute priority to | |
618 | something other than @code{0}. Only a privileged process or the | |
619 | target process' owner can change its absolute priority at all. | |
620 | ||
621 | POSIX requires absolute priority values used with the realtime | |
622 | scheduling policies to be consecutive with a range of at least 32. On | |
623 | Linux, they are 1 through 99. The functions | |
624 | @code{sched_get_priority_max} and @code{sched_set_priority_min} portably | |
625 | tell you what the range is on a particular system. | |
626 | ||
627 | ||
628 | @subsubsection Using Absolute Priority | |
629 | ||
630 | One thing you must keep in mind when designing real time applications is | |
631 | that having higher absolute priority than any other process doesn't | |
632 | guarantee the process can run continuously. Two things that can wreck a | |
87b56f36 | 633 | good CPU run are interrupts and page faults. |
639c6286 UD |
634 | |
635 | Interrupt handlers live in that limbo between processes. The CPU is | |
636 | executing instructions, but they aren't part of any process. An | |
637 | interrupt will stop even the highest priority process. So you must | |
638 | allow for slight delays and make sure that no device in the system has | |
639 | an interrupt handler that could cause too long a delay between | |
640 | instructions for your process. | |
641 | ||
642 | Similarly, a page fault causes what looks like a straightforward | |
643 | sequence of instructions to take a long time. The fact that other | |
644 | processes get to run while the page faults in is of no consequence, | |
d3e22d59 | 645 | because as soon as the I/O is complete, the higher priority process will |
639c6286 UD |
646 | kick them out and run again, but the wait for the I/O itself could be a |
647 | problem. To neutralize this threat, use @code{mlock} or | |
648 | @code{mlockall}. | |
649 | ||
650 | There are a few ramifications of the absoluteness of this priority on a | |
651 | single-CPU system that you need to keep in mind when you choose to set a | |
652 | priority and also when you're working on a program that runs with high | |
653 | absolute priority. Consider a process that has higher absolute priority | |
654 | than any other process in the system and due to a bug in its program, it | |
655 | gets into an infinite loop. It will never cede the CPU. You can't run | |
656 | a command to kill it because your command would need to get the CPU in | |
657 | order to run. The errant program is in complete control. It controls | |
658 | the vertical, it controls the horizontal. | |
659 | ||
660 | There are two ways to avoid this: 1) keep a shell running somewhere with | |
d3e22d59 | 661 | a higher absolute priority or 2) keep a controlling terminal attached to |
639c6286 UD |
662 | the high priority process group. All the priority in the world won't |
663 | stop an interrupt handler from running and delivering a signal to the | |
664 | process if you hit Control-C. | |
665 | ||
95fdc6a0 | 666 | Some systems use absolute priority as a means of allocating a fixed |
0bc93a2f | 667 | percentage of CPU time to a process. To do this, a super high priority |
639c6286 UD |
668 | privileged process constantly monitors the process' CPU usage and raises |
669 | its absolute priority when the process isn't getting its entitled share | |
670 | and lowers it when the process is exceeding it. | |
671 | ||
48b22986 | 672 | @strong{NB:} The absolute priority is sometimes called the ``static |
639c6286 UD |
673 | priority.'' We don't use that term in this manual because it misses the |
674 | most important feature of the absolute priority: its absoluteness. | |
675 | ||
676 | ||
677 | @node Realtime Scheduling | |
678 | @subsection Realtime Scheduling | |
b642f101 | 679 | @cindex realtime scheduling |
639c6286 UD |
680 | |
681 | Whenever two processes with the same absolute priority are ready to run, | |
682 | the kernel has a decision to make, because only one can run at a time. | |
683 | If the processes have absolute priority 0, the kernel makes this decision | |
684 | as described in @ref{Traditional Scheduling}. Otherwise, the decision | |
685 | is as described in this section. | |
686 | ||
687 | If two processes are ready to run but have different absolute priorities, | |
688 | the decision is much simpler, and is described in @ref{Absolute | |
689 | Priority}. | |
690 | ||
87b56f36 | 691 | Each process has a scheduling policy. For processes with absolute |
639c6286 UD |
692 | priority other than zero, there are two available: |
693 | ||
694 | @enumerate | |
695 | @item | |
696 | First Come First Served | |
697 | @item | |
698 | Round Robin | |
699 | @end enumerate | |
700 | ||
701 | The most sensible case is where all the processes with a certain | |
702 | absolute priority have the same scheduling policy. We'll discuss that | |
703 | first. | |
704 | ||
705 | In Round Robin, processes share the CPU, each one running for a small | |
706 | quantum of time (``time slice'') and then yielding to another in a | |
707 | circular fashion. Of course, only processes that are ready to run and | |
708 | have the same absolute priority are in this circle. | |
709 | ||
710 | In First Come First Served, the process that has been waiting the | |
711 | longest to run gets the CPU, and it keeps it until it voluntarily | |
712 | relinquishes the CPU, runs out of things to do (blocks), or gets | |
713 | preempted by a higher priority process. | |
714 | ||
715 | First Come First Served, along with maximal absolute priority and | |
716 | careful control of interrupts and page faults, is the one to use when a | |
717 | process absolutely, positively has to run at full CPU speed or not at | |
718 | all. | |
719 | ||
720 | Judicious use of @code{sched_yield} function invocations by processes | |
721 | with First Come First Served scheduling policy forms a good compromise | |
722 | between Round Robin and First Come First Served. | |
723 | ||
724 | To understand how scheduling works when processes of different scheduling | |
725 | policies occupy the same absolute priority, you have to know the nitty | |
d3e22d59 | 726 | gritty details of how processes enter and exit the ready to run list. |
639c6286 UD |
727 | |
728 | In both cases, the ready to run list is organized as a true queue, where | |
729 | a process gets pushed onto the tail when it becomes ready to run and is | |
730 | popped off the head when the scheduler decides to run it. Note that | |
731 | ready to run and running are two mutually exclusive states. When the | |
732 | scheduler runs a process, that process is no longer ready to run and no | |
733 | longer in the ready to run list. When the process stops running, it | |
734 | may go back to being ready to run again. | |
735 | ||
736 | The only difference between a process that is assigned the Round Robin | |
737 | scheduling policy and a process that is assigned First Come First Serve | |
738 | is that in the former case, the process is automatically booted off the | |
739 | CPU after a certain amount of time. When that happens, the process goes | |
740 | back to being ready to run, which means it enters the queue at the tail. | |
741 | The time quantum we're talking about is small. Really small. This is | |
742 | not your father's timesharing. For example, with the Linux kernel, the | |
743 | round robin time slice is a thousand times shorter than its typical | |
744 | time slice for traditional scheduling. | |
745 | ||
746 | A process begins life with the same scheduling policy as its parent process. | |
747 | Functions described in @ref{Basic Scheduling Functions} can change it. | |
748 | ||
749 | Only a privileged process can set the scheduling policy of a process | |
750 | that has absolute priority higher than 0. | |
751 | ||
752 | @node Basic Scheduling Functions | |
753 | @subsection Basic Scheduling Functions | |
754 | ||
1f77f049 | 755 | This section describes functions in @theglibc{} for setting the |
639c6286 UD |
756 | absolute priority and scheduling policy of a process. |
757 | ||
758 | @strong{Portability Note:} On systems that have the functions in this | |
759 | section, the macro _POSIX_PRIORITY_SCHEDULING is defined in | |
760 | @file{<unistd.h>}. | |
761 | ||
762 | For the case that the scheduling policy is traditional scheduling, more | |
763 | functions to fine tune the scheduling are in @ref{Traditional Scheduling}. | |
764 | ||
765 | Don't try to make too much out of the naming and structure of these | |
766 | functions. They don't match the concepts described in this manual | |
767 | because the functions are as defined by POSIX.1b, but the implementation | |
1f77f049 | 768 | on systems that use @theglibc{} is the inverse of what the POSIX |
639c6286 UD |
769 | structure contemplates. The POSIX scheme assumes that the primary |
770 | scheduling parameter is the scheduling policy and that the priority | |
771 | value, if any, is a parameter of the scheduling policy. In the | |
772 | implementation, though, the priority value is king and the scheduling | |
773 | policy, if anything, only fine tunes the effect of that priority. | |
774 | ||
775 | The symbols in this section are declared by including file @file{sched.h}. | |
776 | ||
777 | @comment sched.h | |
778 | @comment POSIX | |
779 | @deftp {Data Type} {struct sched_param} | |
780 | This structure describes an absolute priority. | |
781 | @table @code | |
782 | @item int sched_priority | |
783 | absolute priority value | |
784 | @end table | |
785 | @end deftp | |
786 | ||
787 | @comment sched.h | |
788 | @comment POSIX | |
789 | @deftypefun int sched_setscheduler (pid_t @var{pid}, int @var{policy}, const struct sched_param *@var{param}) | |
c8ce789c AO |
790 | @safety{@prelim{}@mtsafe{}@assafe{}@acsafe{}} |
791 | @c Direct syscall, Linux only. | |
639c6286 UD |
792 | |
793 | This function sets both the absolute priority and the scheduling policy | |
794 | for a process. | |
795 | ||
796 | It assigns the absolute priority value given by @var{param} and the | |
797 | scheduling policy @var{policy} to the process with Process ID @var{pid}, | |
798 | or the calling process if @var{pid} is zero. If @var{policy} is | |
0bc93a2f | 799 | negative, @code{sched_setscheduler} keeps the existing scheduling policy. |
639c6286 UD |
800 | |
801 | The following macros represent the valid values for @var{policy}: | |
802 | ||
2fe82ca6 | 803 | @vtable @code |
639c6286 UD |
804 | @item SCHED_OTHER |
805 | Traditional Scheduling | |
806 | @item SCHED_FIFO | |
87b56f36 | 807 | First In First Out |
639c6286 UD |
808 | @item SCHED_RR |
809 | Round Robin | |
2fe82ca6 | 810 | @end vtable |
639c6286 UD |
811 | |
812 | @c The Linux kernel code (in sched.c) actually reschedules the process, | |
813 | @c but it puts it at the head of the run queue, so I'm not sure just what | |
814 | @c the effect is, but it must be subtle. | |
815 | ||
816 | On success, the return value is @code{0}. Otherwise, it is @code{-1} | |
817 | and @code{ERRNO} is set accordingly. The @code{errno} values specific | |
818 | to this function are: | |
819 | ||
820 | @table @code | |
821 | @item EPERM | |
822 | @itemize @bullet | |
823 | @item | |
824 | The calling process does not have @code{CAP_SYS_NICE} permission and | |
825 | @var{policy} is not @code{SCHED_OTHER} (or it's negative and the | |
826 | existing policy is not @code{SCHED_OTHER}. | |
827 | ||
828 | @item | |
829 | The calling process does not have @code{CAP_SYS_NICE} permission and its | |
11bf311e | 830 | owner is not the target process' owner. I.e., the effective uid of the |
639c6286 UD |
831 | calling process is neither the effective nor the real uid of process |
832 | @var{pid}. | |
833 | @c We need a cross reference to the capabilities section, when written. | |
834 | @end itemize | |
835 | ||
836 | @item ESRCH | |
837 | There is no process with pid @var{pid} and @var{pid} is not zero. | |
838 | ||
839 | @item EINVAL | |
840 | @itemize @bullet | |
841 | @item | |
842 | @var{policy} does not identify an existing scheduling policy. | |
843 | ||
844 | @item | |
845 | The absolute priority value identified by *@var{param} is outside the | |
846 | valid range for the scheduling policy @var{policy} (or the existing | |
847 | scheduling policy if @var{policy} is negative) or @var{param} is | |
848 | null. @code{sched_get_priority_max} and @code{sched_get_priority_min} | |
849 | tell you what the valid range is. | |
850 | ||
851 | @item | |
852 | @var{pid} is negative. | |
853 | @end itemize | |
854 | @end table | |
855 | ||
856 | @end deftypefun | |
857 | ||
858 | ||
859 | @comment sched.h | |
860 | @comment POSIX | |
861 | @deftypefun int sched_getscheduler (pid_t @var{pid}) | |
c8ce789c AO |
862 | @safety{@prelim{}@mtsafe{}@assafe{}@acsafe{}} |
863 | @c Direct syscall, Linux only. | |
639c6286 UD |
864 | |
865 | This function returns the scheduling policy assigned to the process with | |
866 | Process ID (pid) @var{pid}, or the calling process if @var{pid} is zero. | |
867 | ||
868 | The return value is the scheduling policy. See | |
869 | @code{sched_setscheduler} for the possible values. | |
870 | ||
871 | If the function fails, the return value is instead @code{-1} and | |
872 | @code{errno} is set accordingly. | |
873 | ||
874 | The @code{errno} values specific to this function are: | |
875 | ||
876 | @table @code | |
877 | ||
878 | @item ESRCH | |
879 | There is no process with pid @var{pid} and it is not zero. | |
880 | ||
881 | @item EINVAL | |
882 | @var{pid} is negative. | |
883 | ||
884 | @end table | |
885 | ||
886 | Note that this function is not an exact mate to @code{sched_setscheduler} | |
887 | because while that function sets the scheduling policy and the absolute | |
888 | priority, this function gets only the scheduling policy. To get the | |
889 | absolute priority, use @code{sched_getparam}. | |
890 | ||
891 | @end deftypefun | |
892 | ||
893 | ||
894 | @comment sched.h | |
895 | @comment POSIX | |
896 | @deftypefun int sched_setparam (pid_t @var{pid}, const struct sched_param *@var{param}) | |
c8ce789c AO |
897 | @safety{@prelim{}@mtsafe{}@assafe{}@acsafe{}} |
898 | @c Direct syscall, Linux only. | |
639c6286 UD |
899 | |
900 | This function sets a process' absolute priority. | |
901 | ||
902 | It is functionally identical to @code{sched_setscheduler} with | |
903 | @var{policy} = @code{-1}. | |
904 | ||
905 | @c in fact, that's how it's implemented in Linux. | |
906 | ||
907 | @end deftypefun | |
908 | ||
909 | @comment sched.h | |
910 | @comment POSIX | |
8ded91fb | 911 | @deftypefun int sched_getparam (pid_t @var{pid}, struct sched_param *@var{param}) |
c8ce789c AO |
912 | @safety{@prelim{}@mtsafe{}@assafe{}@acsafe{}} |
913 | @c Direct syscall, Linux only. | |
639c6286 UD |
914 | |
915 | This function returns a process' absolute priority. | |
916 | ||
917 | @var{pid} is the Process ID (pid) of the process whose absolute priority | |
918 | you want to know. | |
919 | ||
920 | @var{param} is a pointer to a structure in which the function stores the | |
921 | absolute priority of the process. | |
922 | ||
923 | On success, the return value is @code{0}. Otherwise, it is @code{-1} | |
d3e22d59 | 924 | and @code{errno} is set accordingly. The @code{errno} values specific |
639c6286 UD |
925 | to this function are: |
926 | ||
927 | @table @code | |
928 | ||
929 | @item ESRCH | |
930 | There is no process with pid @var{pid} and it is not zero. | |
931 | ||
932 | @item EINVAL | |
933 | @var{pid} is negative. | |
934 | ||
935 | @end table | |
936 | ||
937 | @end deftypefun | |
938 | ||
939 | ||
940 | @comment sched.h | |
941 | @comment POSIX | |
8ded91fb | 942 | @deftypefun int sched_get_priority_min (int @var{policy}) |
c8ce789c AO |
943 | @safety{@prelim{}@mtsafe{}@assafe{}@acsafe{}} |
944 | @c Direct syscall, Linux only. | |
639c6286 UD |
945 | |
946 | This function returns the lowest absolute priority value that is | |
947 | allowable for a process with scheduling policy @var{policy}. | |
948 | ||
949 | On Linux, it is 0 for SCHED_OTHER and 1 for everything else. | |
950 | ||
951 | On success, the return value is @code{0}. Otherwise, it is @code{-1} | |
952 | and @code{ERRNO} is set accordingly. The @code{errno} values specific | |
953 | to this function are: | |
954 | ||
955 | @table @code | |
956 | @item EINVAL | |
957 | @var{policy} does not identify an existing scheduling policy. | |
958 | @end table | |
959 | ||
960 | @end deftypefun | |
961 | ||
962 | @comment sched.h | |
963 | @comment POSIX | |
8ded91fb | 964 | @deftypefun int sched_get_priority_max (int @var{policy}) |
c8ce789c AO |
965 | @safety{@prelim{}@mtsafe{}@assafe{}@acsafe{}} |
966 | @c Direct syscall, Linux only. | |
639c6286 UD |
967 | |
968 | This function returns the highest absolute priority value that is | |
969 | allowable for a process that with scheduling policy @var{policy}. | |
970 | ||
971 | On Linux, it is 0 for SCHED_OTHER and 99 for everything else. | |
972 | ||
973 | On success, the return value is @code{0}. Otherwise, it is @code{-1} | |
974 | and @code{ERRNO} is set accordingly. The @code{errno} values specific | |
975 | to this function are: | |
976 | ||
977 | @table @code | |
978 | @item EINVAL | |
979 | @var{policy} does not identify an existing scheduling policy. | |
980 | @end table | |
981 | ||
982 | @end deftypefun | |
983 | ||
984 | @comment sched.h | |
985 | @comment POSIX | |
986 | @deftypefun int sched_rr_get_interval (pid_t @var{pid}, struct timespec *@var{interval}) | |
c8ce789c AO |
987 | @safety{@prelim{}@mtsafe{}@assafe{}@acsafe{}} |
988 | @c Direct syscall, Linux only. | |
639c6286 | 989 | |
87b56f36 | 990 | This function returns the length of the quantum (time slice) used with |
639c6286 UD |
991 | the Round Robin scheduling policy, if it is used, for the process with |
992 | Process ID @var{pid}. | |
993 | ||
87b56f36 | 994 | It returns the length of time as @var{interval}. |
639c6286 UD |
995 | @c We need a cross-reference to where timespec is explained. But that |
996 | @c section doesn't exist yet, and the time chapter needs to be slightly | |
997 | @c reorganized so there is a place to put it (which will be right next | |
998 | @c to timeval, which is presently misplaced). 2000.05.07. | |
999 | ||
1000 | With a Linux kernel, the round robin time slice is always 150 | |
1001 | microseconds, and @var{pid} need not even be a real pid. | |
1002 | ||
1003 | The return value is @code{0} on success and in the pathological case | |
1004 | that it fails, the return value is @code{-1} and @code{errno} is set | |
1005 | accordingly. There is nothing specific that can go wrong with this | |
1006 | function, so there are no specific @code{errno} values. | |
1007 | ||
1008 | @end deftypefun | |
1009 | ||
1010 | @comment sched.h | |
1011 | @comment POSIX | |
3c44837c | 1012 | @deftypefun int sched_yield (void) |
c8ce789c AO |
1013 | @safety{@prelim{}@mtsafe{}@assafe{}@acsafe{}} |
1014 | @c Direct syscall on Linux; alias to swtch on HURD. | |
639c6286 UD |
1015 | |
1016 | This function voluntarily gives up the process' claim on the CPU. | |
1017 | ||
1018 | Technically, @code{sched_yield} causes the calling process to be made | |
1019 | immediately ready to run (as opposed to running, which is what it was | |
1020 | before). This means that if it has absolute priority higher than 0, it | |
1021 | gets pushed onto the tail of the queue of processes that share its | |
1022 | absolute priority and are ready to run, and it will run again when its | |
1023 | turn next arrives. If its absolute priority is 0, it is more | |
1024 | complicated, but still has the effect of yielding the CPU to other | |
1025 | processes. | |
1026 | ||
1027 | If there are no other processes that share the calling process' absolute | |
1028 | priority, this function doesn't have any effect. | |
1029 | ||
1030 | To the extent that the containing program is oblivious to what other | |
1031 | processes in the system are doing and how fast it executes, this | |
1032 | function appears as a no-op. | |
1033 | ||
1034 | The return value is @code{0} on success and in the pathological case | |
1035 | that it fails, the return value is @code{-1} and @code{errno} is set | |
1036 | accordingly. There is nothing specific that can go wrong with this | |
1037 | function, so there are no specific @code{errno} values. | |
1038 | ||
1039 | @end deftypefun | |
1040 | ||
1041 | @node Traditional Scheduling | |
1042 | @subsection Traditional Scheduling | |
1043 | @cindex scheduling, traditional | |
1044 | ||
1045 | This section is about the scheduling among processes whose absolute | |
1046 | priority is 0. When the system hands out the scraps of CPU time that | |
0bc93a2f | 1047 | are left over after the processes with higher absolute priority have |
639c6286 UD |
1048 | taken all they want, the scheduling described herein determines who |
1049 | among the great unwashed processes gets them. | |
1050 | ||
1051 | @menu | |
1052 | * Traditional Scheduling Intro:: | |
1053 | * Traditional Scheduling Functions:: | |
1054 | @end menu | |
1055 | ||
1056 | @node Traditional Scheduling Intro | |
1057 | @subsubsection Introduction To Traditional Scheduling | |
1058 | ||
1059 | Long before there was absolute priority (See @ref{Absolute Priority}), | |
d3e22d59 | 1060 | Unix systems were scheduling the CPU using this system. When POSIX came |
0bc93a2f | 1061 | in like the Romans and imposed absolute priorities to accommodate the |
639c6286 UD |
1062 | needs of realtime processing, it left the indigenous Absolute Priority |
1063 | Zero processes to govern themselves by their own familiar scheduling | |
1064 | policy. | |
1065 | ||
1066 | Indeed, absolute priorities higher than zero are not available on many | |
1067 | systems today and are not typically used when they are, being intended | |
1068 | mainly for computers that do realtime processing. So this section | |
1069 | describes the only scheduling many programmers need to be concerned | |
1070 | about. | |
1071 | ||
1072 | But just to be clear about the scope of this scheduling: Any time a | |
9dcc8f11 | 1073 | process with an absolute priority of 0 and a process with an absolute |
639c6286 UD |
1074 | priority higher than 0 are ready to run at the same time, the one with |
1075 | absolute priority 0 does not run. If it's already running when the | |
1076 | higher priority ready-to-run process comes into existence, it stops | |
1077 | immediately. | |
1078 | ||
1079 | In addition to its absolute priority of zero, every process has another | |
1080 | priority, which we will refer to as "dynamic priority" because it changes | |
87b56f36 | 1081 | over time. The dynamic priority is meaningless for processes with |
639c6286 UD |
1082 | an absolute priority higher than zero. |
1083 | ||
1084 | The dynamic priority sometimes determines who gets the next turn on the | |
1085 | CPU. Sometimes it determines how long turns last. Sometimes it | |
1086 | determines whether a process can kick another off the CPU. | |
1087 | ||
d3e22d59 | 1088 | In Linux, the value is a combination of these things, but mostly it |
639c6286 UD |
1089 | just determines the length of the time slice. The higher a process' |
1090 | dynamic priority, the longer a shot it gets on the CPU when it gets one. | |
1091 | If it doesn't use up its time slice before giving up the CPU to do | |
1092 | something like wait for I/O, it is favored for getting the CPU back when | |
1093 | it's ready for it, to finish out its time slice. Other than that, | |
1094 | selection of processes for new time slices is basically round robin. | |
1095 | But the scheduler does throw a bone to the low priority processes: A | |
1096 | process' dynamic priority rises every time it is snubbed in the | |
1097 | scheduling process. In Linux, even the fat kid gets to play. | |
1098 | ||
1099 | The fluctuation of a process' dynamic priority is regulated by another | |
1100 | value: The ``nice'' value. The nice value is an integer, usually in the | |
1101 | range -20 to 20, and represents an upper limit on a process' dynamic | |
1102 | priority. The higher the nice number, the lower that limit. | |
1103 | ||
1104 | On a typical Linux system, for example, a process with a nice value of | |
1105 | 20 can get only 10 milliseconds on the CPU at a time, whereas a process | |
1106 | with a nice value of -20 can achieve a high enough priority to get 400 | |
1107 | milliseconds. | |
1108 | ||
1109 | The idea of the nice value is deferential courtesy. In the beginning, | |
1110 | in the Unix garden of Eden, all processes shared equally in the bounty | |
1111 | of the computer system. But not all processes really need the same | |
1112 | share of CPU time, so the nice value gave a courteous process the | |
1113 | ability to refuse its equal share of CPU time that others might prosper. | |
1114 | Hence, the higher a process' nice value, the nicer the process is. | |
1115 | (Then a snake came along and offered some process a negative nice value | |
1116 | and the system became the crass resource allocation system we know | |
d3e22d59 | 1117 | today.) |
639c6286 UD |
1118 | |
1119 | Dynamic priorities tend upward and downward with an objective of | |
1120 | smoothing out allocation of CPU time and giving quick response time to | |
1121 | infrequent requests. But they never exceed their nice limits, so on a | |
1122 | heavily loaded CPU, the nice value effectively determines how fast a | |
1123 | process runs. | |
1124 | ||
1125 | In keeping with the socialistic heritage of Unix process priority, a | |
1126 | process begins life with the same nice value as its parent process and | |
1127 | can raise it at will. A process can also raise the nice value of any | |
1128 | other process owned by the same user (or effective user). But only a | |
1129 | privileged process can lower its nice value. A privileged process can | |
1130 | also raise or lower another process' nice value. | |
1131 | ||
1f77f049 | 1132 | @glibcadj{} functions for getting and setting nice values are described in |
639c6286 UD |
1133 | @xref{Traditional Scheduling Functions}. |
1134 | ||
1135 | @node Traditional Scheduling Functions | |
1136 | @subsubsection Functions For Traditional Scheduling | |
1137 | ||
5ce8f203 | 1138 | @pindex sys/resource.h |
639c6286 UD |
1139 | This section describes how you can read and set the nice value of a |
1140 | process. All these symbols are declared in @file{sys/resource.h}. | |
1141 | ||
1142 | The function and macro names are defined by POSIX, and refer to | |
1143 | "priority," but the functions actually have to do with nice values, as | |
1144 | the terms are used both in the manual and POSIX. | |
1145 | ||
1146 | The range of valid nice values depends on the kernel, but typically it | |
1147 | runs from @code{-20} to @code{20}. A lower nice value corresponds to | |
1148 | higher priority for the process. These constants describe the range of | |
5ce8f203 UD |
1149 | priority values: |
1150 | ||
b642f101 | 1151 | @vtable @code |
5ce8f203 UD |
1152 | @comment sys/resource.h |
1153 | @comment BSD | |
1154 | @item PRIO_MIN | |
639c6286 | 1155 | The lowest valid nice value. |
5ce8f203 UD |
1156 | |
1157 | @comment sys/resource.h | |
1158 | @comment BSD | |
1159 | @item PRIO_MAX | |
639c6286 | 1160 | The highest valid nice value. |
b642f101 | 1161 | @end vtable |
5ce8f203 UD |
1162 | |
1163 | @comment sys/resource.h | |
f227c3e0 | 1164 | @comment BSD, POSIX |
5ce8f203 | 1165 | @deftypefun int getpriority (int @var{class}, int @var{id}) |
c8ce789c AO |
1166 | @safety{@prelim{}@mtsafe{}@assafe{}@acsafe{}} |
1167 | @c Direct syscall on UNIX. On HURD, calls _hurd_priority_which_map. | |
639c6286 | 1168 | Return the nice value of a set of processes; @var{class} and @var{id} |
5ce8f203 | 1169 | specify which ones (see below). If the processes specified do not all |
639c6286 | 1170 | have the same nice value, this returns the lowest value that any of them |
5ce8f203 UD |
1171 | has. |
1172 | ||
639c6286 | 1173 | On success, the return value is @code{0}. Otherwise, it is @code{-1} |
d3e22d59 | 1174 | and @code{errno} is set accordingly. The @code{errno} values specific |
639c6286 | 1175 | to this function are: |
5ce8f203 UD |
1176 | |
1177 | @table @code | |
1178 | @item ESRCH | |
1179 | The combination of @var{class} and @var{id} does not match any existing | |
1180 | process. | |
1181 | ||
1182 | @item EINVAL | |
1183 | The value of @var{class} is not valid. | |
1184 | @end table | |
1185 | ||
639c6286 UD |
1186 | If the return value is @code{-1}, it could indicate failure, or it could |
1187 | be the nice value. The only way to make certain is to set @code{errno = | |
1188 | 0} before calling @code{getpriority}, then use @code{errno != 0} | |
1189 | afterward as the criterion for failure. | |
5ce8f203 UD |
1190 | @end deftypefun |
1191 | ||
1192 | @comment sys/resource.h | |
f227c3e0 | 1193 | @comment BSD, POSIX |
639c6286 | 1194 | @deftypefun int setpriority (int @var{class}, int @var{id}, int @var{niceval}) |
c8ce789c AO |
1195 | @safety{@prelim{}@mtsafe{}@assafe{}@acsafe{}} |
1196 | @c Direct syscall on UNIX. On HURD, calls _hurd_priority_which_map. | |
639c6286 | 1197 | Set the nice value of a set of processes to @var{niceval}; @var{class} |
5ce8f203 UD |
1198 | and @var{id} specify which ones (see below). |
1199 | ||
6a7a8b22 | 1200 | The return value is @code{0} on success, and @code{-1} on |
639c6286 UD |
1201 | failure. The following @code{errno} error condition are possible for |
1202 | this function: | |
5ce8f203 UD |
1203 | |
1204 | @table @code | |
1205 | @item ESRCH | |
1206 | The combination of @var{class} and @var{id} does not match any existing | |
1207 | process. | |
1208 | ||
1209 | @item EINVAL | |
1210 | The value of @var{class} is not valid. | |
1211 | ||
1212 | @item EPERM | |
639c6286 | 1213 | The call would set the nice value of a process which is owned by a different |
11bf311e | 1214 | user than the calling process (i.e., the target process' real or effective |
639c6286 UD |
1215 | uid does not match the calling process' effective uid) and the calling |
1216 | process does not have @code{CAP_SYS_NICE} permission. | |
5ce8f203 UD |
1217 | |
1218 | @item EACCES | |
639c6286 UD |
1219 | The call would lower the process' nice value and the process does not have |
1220 | @code{CAP_SYS_NICE} permission. | |
5ce8f203 | 1221 | @end table |
639c6286 | 1222 | |
5ce8f203 UD |
1223 | @end deftypefun |
1224 | ||
1225 | The arguments @var{class} and @var{id} together specify a set of | |
1226 | processes in which you are interested. These are the possible values of | |
1227 | @var{class}: | |
1228 | ||
b642f101 | 1229 | @vtable @code |
5ce8f203 UD |
1230 | @comment sys/resource.h |
1231 | @comment BSD | |
1232 | @item PRIO_PROCESS | |
639c6286 | 1233 | One particular process. The argument @var{id} is a process ID (pid). |
5ce8f203 UD |
1234 | |
1235 | @comment sys/resource.h | |
1236 | @comment BSD | |
1237 | @item PRIO_PGRP | |
639c6286 UD |
1238 | All the processes in a particular process group. The argument @var{id} is |
1239 | a process group ID (pgid). | |
5ce8f203 UD |
1240 | |
1241 | @comment sys/resource.h | |
1242 | @comment BSD | |
1243 | @item PRIO_USER | |
11bf311e | 1244 | All the processes owned by a particular user (i.e., whose real uid |
639c6286 | 1245 | indicates the user). The argument @var{id} is a user ID (uid). |
b642f101 | 1246 | @end vtable |
5ce8f203 | 1247 | |
639c6286 UD |
1248 | If the argument @var{id} is 0, it stands for the calling process, its |
1249 | process group, or its owner (real uid), according to @var{class}. | |
5ce8f203 | 1250 | |
b642f101 UD |
1251 | @comment unistd.h |
1252 | @comment BSD | |
5ce8f203 | 1253 | @deftypefun int nice (int @var{increment}) |
c8ce789c AO |
1254 | @safety{@prelim{}@mtunsafe{@mtasurace{:setpriority}}@asunsafe{}@acsafe{}} |
1255 | @c Calls getpriority before and after setpriority, using the result of | |
1256 | @c the first call to compute the argument for setpriority. This creates | |
1257 | @c a window for a concurrent setpriority (or nice) call to be lost or | |
1258 | @c exhibit surprising behavior. | |
639c6286 | 1259 | Increment the nice value of the calling process by @var{increment}. |
6a7a8b22 AJ |
1260 | The return value is the new nice value on success, and @code{-1} on |
1261 | failure. In the case of failure, @code{errno} will be set to the | |
1262 | same values as for @code{setpriority}. | |
1263 | ||
5ce8f203 UD |
1264 | |
1265 | Here is an equivalent definition of @code{nice}: | |
1266 | ||
1267 | @smallexample | |
1268 | int | |
1269 | nice (int increment) | |
1270 | @{ | |
6a7a8b22 AJ |
1271 | int result, old = getpriority (PRIO_PROCESS, 0); |
1272 | result = setpriority (PRIO_PROCESS, 0, old + increment); | |
1273 | if (result != -1) | |
1274 | return old + increment; | |
1275 | else | |
1276 | return -1; | |
5ce8f203 UD |
1277 | @} |
1278 | @end smallexample | |
1279 | @end deftypefun | |
b642f101 | 1280 | |
d9997a45 UD |
1281 | |
1282 | @node CPU Affinity | |
1283 | @subsection Limiting execution to certain CPUs | |
1284 | ||
1285 | On a multi-processor system the operating system usually distributes | |
1286 | the different processes which are runnable on all available CPUs in a | |
1287 | way which allows the system to work most efficiently. Which processes | |
1288 | and threads run can be to some extend be control with the scheduling | |
1289 | functionality described in the last sections. But which CPU finally | |
1290 | executes which process or thread is not covered. | |
1291 | ||
1292 | There are a number of reasons why a program might want to have control | |
1293 | over this aspect of the system as well: | |
1294 | ||
1295 | @itemize @bullet | |
1296 | @item | |
1297 | One thread or process is responsible for absolutely critical work | |
1298 | which under no circumstances must be interrupted or hindered from | |
d3e22d59 | 1299 | making progress by other processes or threads using CPU resources. In |
d9997a45 UD |
1300 | this case the special process would be confined to a CPU which no |
1301 | other process or thread is allowed to use. | |
1302 | ||
1303 | @item | |
1304 | The access to certain resources (RAM, I/O ports) has different costs | |
1305 | from different CPUs. This is the case in NUMA (Non-Uniform Memory | |
11bf311e | 1306 | Architecture) machines. Preferably memory should be accessed locally |
d9997a45 UD |
1307 | but this requirement is usually not visible to the scheduler. |
1308 | Therefore forcing a process or thread to the CPUs which have local | |
d3e22d59 | 1309 | access to the most-used memory helps to significantly boost the |
d9997a45 UD |
1310 | performance. |
1311 | ||
1312 | @item | |
1313 | In controlled runtimes resource allocation and book-keeping work (for | |
1314 | instance garbage collection) is performance local to processors. This | |
1315 | can help to reduce locking costs if the resources do not have to be | |
1316 | protected from concurrent accesses from different processors. | |
1317 | @end itemize | |
1318 | ||
1319 | The POSIX standard up to this date is of not much help to solve this | |
1320 | problem. The Linux kernel provides a set of interfaces to allow | |
1321 | specifying @emph{affinity sets} for a process. The scheduler will | |
bbf70ae9 | 1322 | schedule the thread or process on CPUs specified by the affinity |
1f77f049 | 1323 | masks. The interfaces which @theglibc{} define follow to some |
d3e22d59 | 1324 | extent the Linux kernel interface. |
d9997a45 UD |
1325 | |
1326 | @comment sched.h | |
1327 | @comment GNU | |
1328 | @deftp {Data Type} cpu_set_t | |
1329 | This data set is a bitset where each bit represents a CPU. How the | |
1330 | system's CPUs are mapped to bits in the bitset is system dependent. | |
1331 | The data type has a fixed size; in the unlikely case that the number | |
1332 | of bits are not sufficient to describe the CPUs of the system a | |
1333 | different interface has to be used. | |
1334 | ||
1335 | This type is a GNU extension and is defined in @file{sched.h}. | |
1336 | @end deftp | |
1337 | ||
d3e22d59 | 1338 | To manipulate the bitset, to set and reset bits, a number of macros are |
d9997a45 UD |
1339 | defined. Some of the macros take a CPU number as a parameter. Here |
1340 | it is important to never exceed the size of the bitset. The following | |
1341 | macro specifies the number of bits in the @code{cpu_set_t} bitset. | |
1342 | ||
1343 | @comment sched.h | |
1344 | @comment GNU | |
1345 | @deftypevr Macro int CPU_SETSIZE | |
1346 | The value of this macro is the maximum number of CPUs which can be | |
1347 | handled with a @code{cpu_set_t} object. | |
1348 | @end deftypevr | |
1349 | ||
1350 | The type @code{cpu_set_t} should be considered opaque; all | |
1351 | manipulation should happen via the next four macros. | |
1352 | ||
1353 | @comment sched.h | |
1354 | @comment GNU | |
1355 | @deftypefn Macro void CPU_ZERO (cpu_set_t *@var{set}) | |
c8ce789c AO |
1356 | @safety{@prelim{}@mtsafe{}@assafe{}@acsafe{}} |
1357 | @c CPU_ZERO ok | |
1358 | @c __CPU_ZERO_S ok | |
1359 | @c memset dup ok | |
d9997a45 UD |
1360 | This macro initializes the CPU set @var{set} to be the empty set. |
1361 | ||
1362 | This macro is a GNU extension and is defined in @file{sched.h}. | |
1363 | @end deftypefn | |
1364 | ||
1365 | @comment sched.h | |
1366 | @comment GNU | |
1367 | @deftypefn Macro void CPU_SET (int @var{cpu}, cpu_set_t *@var{set}) | |
c8ce789c AO |
1368 | @safety{@prelim{}@mtsafe{}@assafe{}@acsafe{}} |
1369 | @c CPU_SET ok | |
1370 | @c __CPU_SET_S ok | |
1371 | @c __CPUELT ok | |
1372 | @c __CPUMASK ok | |
d9997a45 UD |
1373 | This macro adds @var{cpu} to the CPU set @var{set}. |
1374 | ||
1375 | The @var{cpu} parameter must not have side effects since it is | |
1376 | evaluated more than once. | |
1377 | ||
1378 | This macro is a GNU extension and is defined in @file{sched.h}. | |
1379 | @end deftypefn | |
1380 | ||
1381 | @comment sched.h | |
1382 | @comment GNU | |
1383 | @deftypefn Macro void CPU_CLR (int @var{cpu}, cpu_set_t *@var{set}) | |
c8ce789c AO |
1384 | @safety{@prelim{}@mtsafe{}@assafe{}@acsafe{}} |
1385 | @c CPU_CLR ok | |
1386 | @c __CPU_CLR_S ok | |
1387 | @c __CPUELT dup ok | |
1388 | @c __CPUMASK dup ok | |
d9997a45 UD |
1389 | This macro removes @var{cpu} from the CPU set @var{set}. |
1390 | ||
1391 | The @var{cpu} parameter must not have side effects since it is | |
1392 | evaluated more than once. | |
1393 | ||
1394 | This macro is a GNU extension and is defined in @file{sched.h}. | |
1395 | @end deftypefn | |
1396 | ||
1397 | @comment sched.h | |
1398 | @comment GNU | |
1399 | @deftypefn Macro int CPU_ISSET (int @var{cpu}, const cpu_set_t *@var{set}) | |
c8ce789c AO |
1400 | @safety{@prelim{}@mtsafe{}@assafe{}@acsafe{}} |
1401 | @c CPU_ISSET ok | |
1402 | @c __CPU_ISSET_S ok | |
1403 | @c __CPUELT dup ok | |
1404 | @c __CPUMASK dup ok | |
d9997a45 UD |
1405 | This macro returns a nonzero value (true) if @var{cpu} is a member |
1406 | of the CPU set @var{set}, and zero (false) otherwise. | |
1407 | ||
1408 | The @var{cpu} parameter must not have side effects since it is | |
1409 | evaluated more than once. | |
1410 | ||
1411 | This macro is a GNU extension and is defined in @file{sched.h}. | |
1412 | @end deftypefn | |
1413 | ||
1414 | ||
1415 | CPU bitsets can be constructed from scratch or the currently installed | |
1416 | affinity mask can be retrieved from the system. | |
1417 | ||
1418 | @comment sched.h | |
1419 | @comment GNU | |
6f0b2e1f | 1420 | @deftypefun int sched_getaffinity (pid_t @var{pid}, size_t @var{cpusetsize}, cpu_set_t *@var{cpuset}) |
c8ce789c AO |
1421 | @safety{@prelim{}@mtsafe{}@assafe{}@acsafe{}} |
1422 | @c Wrapped syscall to zero out past the kernel cpu set size; Linux | |
1423 | @c only. | |
d9997a45 | 1424 | |
d3e22d59 | 1425 | This function stores the CPU affinity mask for the process or thread |
6f0b2e1f RM |
1426 | with the ID @var{pid} in the @var{cpusetsize} bytes long bitmap |
1427 | pointed to by @var{cpuset}. If successful, the function always | |
1428 | initializes all bits in the @code{cpu_set_t} object and returns zero. | |
d9997a45 UD |
1429 | |
1430 | If @var{pid} does not correspond to a process or thread on the system | |
1431 | the or the function fails for some other reason, it returns @code{-1} | |
1432 | and @code{errno} is set to represent the error condition. | |
1433 | ||
1434 | @table @code | |
1435 | @item ESRCH | |
1436 | No process or thread with the given ID found. | |
1437 | ||
1438 | @item EFAULT | |
d3e22d59 | 1439 | The pointer @var{cpuset} does not point to a valid object. |
d9997a45 UD |
1440 | @end table |
1441 | ||
1442 | This function is a GNU extension and is declared in @file{sched.h}. | |
1443 | @end deftypefun | |
1444 | ||
1445 | Note that it is not portably possible to use this information to | |
1446 | retrieve the information for different POSIX threads. A separate | |
1447 | interface must be provided for that. | |
1448 | ||
1449 | @comment sched.h | |
1450 | @comment GNU | |
6f0b2e1f | 1451 | @deftypefun int sched_setaffinity (pid_t @var{pid}, size_t @var{cpusetsize}, const cpu_set_t *@var{cpuset}) |
c8ce789c AO |
1452 | @safety{@prelim{}@mtsafe{}@assafe{}@acsafe{}} |
1453 | @c Wrapped syscall to detect attempts to set bits past the kernel cpu | |
1454 | @c set size; Linux only. | |
d9997a45 | 1455 | |
6f0b2e1f RM |
1456 | This function installs the @var{cpusetsize} bytes long affinity mask |
1457 | pointed to by @var{cpuset} for the process or thread with the ID @var{pid}. | |
d3e22d59 | 1458 | If successful the function returns zero and the scheduler will in the future |
6f0b2e1f | 1459 | take the affinity information into account. |
d9997a45 UD |
1460 | |
1461 | If the function fails it will return @code{-1} and @code{errno} is set | |
1462 | to the error code: | |
1463 | ||
1464 | @table @code | |
1465 | @item ESRCH | |
1466 | No process or thread with the given ID found. | |
1467 | ||
1468 | @item EFAULT | |
d3e22d59 | 1469 | The pointer @var{cpuset} does not point to a valid object. |
d9997a45 UD |
1470 | |
1471 | @item EINVAL | |
1472 | The bitset is not valid. This might mean that the affinity set might | |
1473 | not leave a processor for the process or thread to run on. | |
1474 | @end table | |
1475 | ||
1476 | This function is a GNU extension and is declared in @file{sched.h}. | |
1477 | @end deftypefun | |
1478 | ||
1479 | ||
b642f101 UD |
1480 | @node Memory Resources |
1481 | @section Querying memory available resources | |
1482 | ||
1483 | The amount of memory available in the system and the way it is organized | |
1484 | determines oftentimes the way programs can and have to work. For | |
5a7eedfb | 1485 | functions like @code{mmap} it is necessary to know about the size of |
b642f101 UD |
1486 | individual memory pages and knowing how much memory is available enables |
1487 | a program to select appropriate sizes for, say, caches. Before we get | |
1488 | into these details a few words about memory subsystems in traditional | |
5a7eedfb | 1489 | Unix systems will be given. |
b642f101 UD |
1490 | |
1491 | @menu | |
1492 | * Memory Subsystem:: Overview about traditional Unix memory handling. | |
1493 | * Query Memory Parameters:: How to get information about the memory | |
1494 | subsystem? | |
1495 | @end menu | |
1496 | ||
1497 | @node Memory Subsystem | |
1498 | @subsection Overview about traditional Unix memory handling | |
1499 | ||
1500 | @cindex address space | |
1501 | @cindex physical memory | |
1502 | @cindex physical address | |
1503 | Unix systems normally provide processes virtual address spaces. This | |
1504 | means that the addresses of the memory regions do not have to correspond | |
1505 | directly to the addresses of the actual physical memory which stores the | |
1506 | data. An extra level of indirection is introduced which translates | |
1507 | virtual addresses into physical addresses. This is normally done by the | |
1508 | hardware of the processor. | |
1509 | ||
1510 | @cindex shared memory | |
d3e22d59 | 1511 | Using a virtual address space has several advantages. The most important |
b642f101 UD |
1512 | is process isolation. The different processes running on the system |
1513 | cannot interfere directly with each other. No process can write into | |
1514 | the address space of another process (except when shared memory is used | |
1515 | but then it is wanted and controlled). | |
1516 | ||
1517 | Another advantage of virtual memory is that the address space the | |
1518 | processes see can actually be larger than the physical memory available. | |
1519 | The physical memory can be extended by storage on an external media | |
1520 | where the content of currently unused memory regions is stored. The | |
1521 | address translation can then intercept accesses to these memory regions | |
1522 | and make memory content available again by loading the data back into | |
1523 | memory. This concept makes it necessary that programs which have to use | |
1524 | lots of memory know the difference between available virtual address | |
1525 | space and available physical memory. If the working set of virtual | |
1526 | memory of all the processes is larger than the available physical memory | |
1527 | the system will slow down dramatically due to constant swapping of | |
1528 | memory content from the memory to the storage media and back. This is | |
1529 | called ``thrashing''. | |
1530 | @cindex thrashing | |
1531 | ||
1532 | @cindex memory page | |
1533 | @cindex page, memory | |
1534 | A final aspect of virtual memory which is important and follows from | |
1535 | what is said in the last paragraph is the granularity of the virtual | |
1536 | address space handling. When we said that the virtual address handling | |
1537 | stores memory content externally it cannot do this on a byte-by-byte | |
1538 | basis. The administrative overhead does not allow this (leaving alone | |
1539 | the processor hardware). Instead several thousand bytes are handled | |
1540 | together and form a @dfn{page}. The size of each page is always a power | |
d3e22d59 | 1541 | of two bytes. The smallest page size in use today is 4096, with 8192, |
b642f101 UD |
1542 | 16384, and 65536 being other popular sizes. |
1543 | ||
1544 | @node Query Memory Parameters | |
1545 | @subsection How to get information about the memory subsystem? | |
1546 | ||
1547 | The page size of the virtual memory the process sees is essential to | |
d3e22d59 | 1548 | know in several situations. Some programming interfaces (e.g., |
b642f101 | 1549 | @code{mmap}, @pxref{Memory-mapped I/O}) require the user to provide |
d3e22d59 | 1550 | information adjusted to the page size. In the case of @code{mmap} it is |
b642f101 UD |
1551 | necessary to provide a length argument which is a multiple of the page |
1552 | size. Another place where the knowledge about the page size is useful | |
1553 | is in memory allocation. If one allocates pieces of memory in larger | |
1554 | chunks which are then subdivided by the application code it is useful to | |
1555 | adjust the size of the larger blocks to the page size. If the total | |
1556 | memory requirement for the block is close (but not larger) to a multiple | |
1557 | of the page size the kernel's memory handling can work more effectively | |
1558 | since it only has to allocate memory pages which are fully used. (To do | |
1559 | this optimization it is necessary to know a bit about the memory | |
1560 | allocator which will require a bit of memory itself for each block and | |
d3e22d59 | 1561 | this overhead must not push the total size over the page size multiple.) |
b642f101 UD |
1562 | |
1563 | The page size traditionally was a compile time constant. But recent | |
1564 | development of processors changed this. Processors now support | |
1565 | different page sizes and they can possibly even vary among different | |
1566 | processes on the same system. Therefore the system should be queried at | |
1567 | runtime about the current page size and no assumptions (except about it | |
1568 | being a power of two) should be made. | |
1569 | ||
1570 | @vindex _SC_PAGESIZE | |
1571 | The correct interface to query about the page size is @code{sysconf} | |
1572 | (@pxref{Sysconf Definition}) with the parameter @code{_SC_PAGESIZE}. | |
1573 | There is a much older interface available, too. | |
1574 | ||
1575 | @comment unistd.h | |
1576 | @comment BSD | |
1577 | @deftypefun int getpagesize (void) | |
c8ce789c AO |
1578 | @safety{@prelim{}@mtsafe{}@assafe{}@acsafe{}} |
1579 | @c Obtained from the aux vec at program startup time. GNU/Linux/m68k is | |
1580 | @c the exception, with the possibility of a syscall. | |
b642f101 UD |
1581 | The @code{getpagesize} function returns the page size of the process. |
1582 | This value is fixed for the runtime of the process but can vary in | |
1583 | different runs of the application. | |
1584 | ||
1585 | The function is declared in @file{unistd.h}. | |
1586 | @end deftypefun | |
1587 | ||
1588 | Widely available on @w{System V} derived systems is a method to get | |
1589 | information about the physical memory the system has. The call | |
1590 | ||
1591 | @vindex _SC_PHYS_PAGES | |
1592 | @cindex sysconf | |
1593 | @smallexample | |
1594 | sysconf (_SC_PHYS_PAGES) | |
1595 | @end smallexample | |
1596 | ||
cb4fe8a2 | 1597 | @noindent |
d3e22d59 | 1598 | returns the total number of pages of physical memory the system has. |
b642f101 UD |
1599 | This does not mean all this memory is available. This information can |
1600 | be found using | |
1601 | ||
1602 | @vindex _SC_AVPHYS_PAGES | |
1603 | @cindex sysconf | |
1604 | @smallexample | |
1605 | sysconf (_SC_AVPHYS_PAGES) | |
1606 | @end smallexample | |
1607 | ||
1608 | These two values help to optimize applications. The value returned for | |
1609 | @code{_SC_AVPHYS_PAGES} is the amount of memory the application can use | |
1610 | without hindering any other process (given that no other process | |
1611 | increases its memory usage). The value returned for | |
1612 | @code{_SC_PHYS_PAGES} is more or less a hard limit for the working set. | |
1613 | If all applications together constantly use more than that amount of | |
1614 | memory the system is in trouble. | |
1615 | ||
1f77f049 | 1616 | @Theglibc{} provides in addition to these already described way to |
cb4fe8a2 UD |
1617 | get this information two functions. They are declared in the file |
1618 | @file{sys/sysinfo.h}. Programmers should prefer to use the | |
1619 | @code{sysconf} method described above. | |
1620 | ||
1621 | @comment sys/sysinfo.h | |
1622 | @comment GNU | |
4c78249d | 1623 | @deftypefun {long int} get_phys_pages (void) |
c8ce789c AO |
1624 | @safety{@prelim{}@mtsafe{}@asunsafe{@ascuheap{} @asulock{}}@acunsafe{@aculock{} @acsfd{} @acsmem{}}} |
1625 | @c This fopens a /proc file and scans it for the requested information. | |
cb4fe8a2 | 1626 | The @code{get_phys_pages} function returns the total number of pages of |
d3e22d59 | 1627 | physical memory the system has. To get the amount of memory this number has to |
cb4fe8a2 UD |
1628 | be multiplied by the page size. |
1629 | ||
1630 | This function is a GNU extension. | |
1631 | @end deftypefun | |
1632 | ||
1633 | @comment sys/sysinfo.h | |
1634 | @comment GNU | |
4c78249d | 1635 | @deftypefun {long int} get_avphys_pages (void) |
c8ce789c | 1636 | @safety{@prelim{}@mtsafe{}@asunsafe{@ascuheap{} @asulock{}}@acunsafe{@aculock{} @acsfd{} @acsmem{}}} |
cd1fb604 | 1637 | The @code{get_avphys_pages} function returns the number of available pages of |
d3e22d59 | 1638 | physical memory the system has. To get the amount of memory this number has to |
cb4fe8a2 UD |
1639 | be multiplied by the page size. |
1640 | ||
1641 | This function is a GNU extension. | |
1642 | @end deftypefun | |
1643 | ||
b642f101 UD |
1644 | @node Processor Resources |
1645 | @section Learn about the processors available | |
1646 | ||
1647 | The use of threads or processes with shared memory allows an application | |
1648 | to take advantage of all the processing power a system can provide. If | |
1649 | the task can be parallelized the optimal way to write an application is | |
1650 | to have at any time as many processes running as there are processors. | |
1651 | To determine the number of processors available to the system one can | |
1652 | run | |
1653 | ||
1654 | @vindex _SC_NPROCESSORS_CONF | |
1655 | @cindex sysconf | |
1656 | @smallexample | |
1657 | sysconf (_SC_NPROCESSORS_CONF) | |
1658 | @end smallexample | |
1659 | ||
1660 | @noindent | |
1661 | which returns the number of processors the operating system configured. | |
1662 | But it might be possible for the operating system to disable individual | |
1663 | processors and so the call | |
1664 | ||
1665 | @vindex _SC_NPROCESSORS_ONLN | |
1666 | @cindex sysconf | |
1667 | @smallexample | |
1668 | sysconf (_SC_NPROCESSORS_ONLN) | |
1669 | @end smallexample | |
1670 | ||
1671 | @noindent | |
26428b7c | 1672 | returns the number of processors which are currently online (i.e., |
b642f101 | 1673 | available). |
e4cf5229 | 1674 | |
1f77f049 | 1675 | For these two pieces of information @theglibc{} also provides |
cb4fe8a2 UD |
1676 | functions to get the information directly. The functions are declared |
1677 | in @file{sys/sysinfo.h}. | |
1678 | ||
1679 | @comment sys/sysinfo.h | |
1680 | @comment GNU | |
1681 | @deftypefun int get_nprocs_conf (void) | |
c8ce789c AO |
1682 | @safety{@prelim{}@mtsafe{}@asunsafe{@ascuheap{} @asulock{}}@acunsafe{@aculock{} @acsfd{} @acsmem{}}} |
1683 | @c This function reads from from /sys using dir streams (single user, so | |
1684 | @c no @mtasurace issue), and on some arches, from /proc using streams. | |
cb4fe8a2 UD |
1685 | The @code{get_nprocs_conf} function returns the number of processors the |
1686 | operating system configured. | |
1687 | ||
1688 | This function is a GNU extension. | |
1689 | @end deftypefun | |
1690 | ||
1691 | @comment sys/sysinfo.h | |
1692 | @comment GNU | |
1693 | @deftypefun int get_nprocs (void) | |
c8ce789c AO |
1694 | @safety{@prelim{}@mtsafe{}@assafe{}@acsafe{@acsfd{}}} |
1695 | @c This function reads from /proc using file descriptor I/O. | |
cb4fe8a2 UD |
1696 | The @code{get_nprocs} function returns the number of available processors. |
1697 | ||
1698 | This function is a GNU extension. | |
1699 | @end deftypefun | |
1700 | ||
e4cf5229 UD |
1701 | @cindex load average |
1702 | Before starting more threads it should be checked whether the processors | |
1703 | are not already overused. Unix systems calculate something called the | |
1704 | @dfn{load average}. This is a number indicating how many processes were | |
d3e22d59 | 1705 | running. This number is an average over different periods of time |
e4cf5229 UD |
1706 | (normally 1, 5, and 15 minutes). |
1707 | ||
1708 | @comment stdlib.h | |
1709 | @comment BSD | |
1710 | @deftypefun int getloadavg (double @var{loadavg}[], int @var{nelem}) | |
c8ce789c AO |
1711 | @safety{@prelim{}@mtsafe{}@assafe{}@acsafe{@acsfd{}}} |
1712 | @c Calls host_info on HURD; on Linux, opens /proc/loadavg, reads from | |
1713 | @c it, closes it, without cancellation point, and calls strtod_l with | |
1714 | @c the C locale to convert the strings to doubles. | |
e4cf5229 | 1715 | This function gets the 1, 5 and 15 minute load averages of the |
cf822e3c | 1716 | system. The values are placed in @var{loadavg}. @code{getloadavg} will |
e4cf5229 UD |
1717 | place at most @var{nelem} elements into the array but never more than |
1718 | three elements. The return value is the number of elements written to | |
1719 | @var{loadavg}, or -1 on error. | |
1720 | ||
1721 | This function is declared in @file{stdlib.h}. | |
1722 | @end deftypefun |