kern_resource.c revision 1.60.2.2 1 /* $NetBSD: kern_resource.c,v 1.60.2.2 2001/11/14 19:16:37 nathanw Exp $ */
2
3 /*-
4 * Copyright (c) 1982, 1986, 1991, 1993
5 * The Regents of the University of California. All rights reserved.
6 * (c) UNIX System Laboratories, Inc.
7 * All or some portions of this file are derived from material licensed
8 * to the University of California by American Telephone and Telegraph
9 * Co. or Unix System Laboratories, Inc. and are reproduced herein with
10 * the permission of UNIX System Laboratories, Inc.
11 *
12 * Redistribution and use in source and binary forms, with or without
13 * modification, are permitted provided that the following conditions
14 * are met:
15 * 1. Redistributions of source code must retain the above copyright
16 * notice, this list of conditions and the following disclaimer.
17 * 2. Redistributions in binary form must reproduce the above copyright
18 * notice, this list of conditions and the following disclaimer in the
19 * documentation and/or other materials provided with the distribution.
20 * 3. All advertising materials mentioning features or use of this software
21 * must display the following acknowledgement:
22 * This product includes software developed by the University of
23 * California, Berkeley and its contributors.
24 * 4. Neither the name of the University nor the names of its contributors
25 * may be used to endorse or promote products derived from this software
26 * without specific prior written permission.
27 *
28 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
29 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
30 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
31 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
32 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
33 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
34 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
35 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
36 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
37 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
38 * SUCH DAMAGE.
39 *
40 * @(#)kern_resource.c 8.8 (Berkeley) 2/14/95
41 */
42
43 #include <sys/cdefs.h>
44 __KERNEL_RCSID(0, "$NetBSD: kern_resource.c,v 1.60.2.2 2001/11/14 19:16:37 nathanw Exp $");
45
46 #include <sys/param.h>
47 #include <sys/systm.h>
48 #include <sys/kernel.h>
49 #include <sys/file.h>
50 #include <sys/resourcevar.h>
51 #include <sys/malloc.h>
52 #include <sys/pool.h>
53 #include <sys/lwp.h>
54 #include <sys/proc.h>
55
56 #include <sys/mount.h>
57 #include <sys/syscallargs.h>
58
59 #include <uvm/uvm_extern.h>
60
61 /*
62 * Maximum process data and stack limits.
63 * They are variables so they are patchable.
64 *
65 * XXXX Do we really need them to be patchable?
66 */
67 rlim_t maxdmap = MAXDSIZ;
68 rlim_t maxsmap = MAXSSIZ;
69
70 /*
71 * Resource controls and accounting.
72 */
73
74 int
75 sys_getpriority(l, v, retval)
76 struct lwp *l;
77 void *v;
78 register_t *retval;
79 {
80 struct sys_getpriority_args /* {
81 syscallarg(int) which;
82 syscallarg(int) who;
83 } */ *uap = v;
84 struct proc *curp = l->l_proc, *p;
85 int low = NZERO + PRIO_MAX + 1;
86
87 switch (SCARG(uap, which)) {
88
89 case PRIO_PROCESS:
90 if (SCARG(uap, who) == 0)
91 p = curp;
92 else
93 p = pfind(SCARG(uap, who));
94 if (p == 0)
95 break;
96 low = p->p_nice;
97 break;
98
99 case PRIO_PGRP: {
100 struct pgrp *pg;
101
102 if (SCARG(uap, who) == 0)
103 pg = curp->p_pgrp;
104 else if ((pg = pgfind(SCARG(uap, who))) == NULL)
105 break;
106 for (p = pg->pg_members.lh_first; p != 0;
107 p = p->p_pglist.le_next) {
108 if (p->p_nice < low)
109 low = p->p_nice;
110 }
111 break;
112 }
113
114 case PRIO_USER:
115 if (SCARG(uap, who) == 0)
116 SCARG(uap, who) = curp->p_ucred->cr_uid;
117 proclist_lock_read();
118 for (p = allproc.lh_first; p != 0; p = p->p_list.le_next)
119 if (p->p_ucred->cr_uid == SCARG(uap, who) &&
120 p->p_nice < low)
121 low = p->p_nice;
122 proclist_unlock_read();
123 break;
124
125 default:
126 return (EINVAL);
127 }
128 if (low == NZERO + PRIO_MAX + 1)
129 return (ESRCH);
130 *retval = low - NZERO;
131 return (0);
132 }
133
134 /* ARGSUSED */
135 int
136 sys_setpriority(l, v, retval)
137 struct lwp *l;
138 void *v;
139 register_t *retval;
140 {
141 struct sys_setpriority_args /* {
142 syscallarg(int) which;
143 syscallarg(int) who;
144 syscallarg(int) prio;
145 } */ *uap = v;
146 struct proc *curp = l->l_proc, *p;
147 int found = 0, error = 0;
148
149 switch (SCARG(uap, which)) {
150
151 case PRIO_PROCESS:
152 if (SCARG(uap, who) == 0)
153 p = curp;
154 else
155 p = pfind(SCARG(uap, who));
156 if (p == 0)
157 break;
158 error = donice(curp, p, SCARG(uap, prio));
159 found++;
160 break;
161
162 case PRIO_PGRP: {
163 struct pgrp *pg;
164
165 if (SCARG(uap, who) == 0)
166 pg = curp->p_pgrp;
167 else if ((pg = pgfind(SCARG(uap, who))) == NULL)
168 break;
169 for (p = pg->pg_members.lh_first; p != 0;
170 p = p->p_pglist.le_next) {
171 error = donice(curp, p, SCARG(uap, prio));
172 found++;
173 }
174 break;
175 }
176
177 case PRIO_USER:
178 if (SCARG(uap, who) == 0)
179 SCARG(uap, who) = curp->p_ucred->cr_uid;
180 proclist_lock_read();
181 for (p = allproc.lh_first; p != 0; p = p->p_list.le_next)
182 if (p->p_ucred->cr_uid == SCARG(uap, who)) {
183 error = donice(curp, p, SCARG(uap, prio));
184 found++;
185 }
186 proclist_unlock_read();
187 break;
188
189 default:
190 return (EINVAL);
191 }
192 if (found == 0)
193 return (ESRCH);
194 return (error);
195 }
196
197 int
198 donice(curp, chgp, n)
199 struct proc *curp, *chgp;
200 int n;
201 {
202 struct pcred *pcred = curp->p_cred;
203 int s;
204
205 if (pcred->pc_ucred->cr_uid && pcred->p_ruid &&
206 pcred->pc_ucred->cr_uid != chgp->p_ucred->cr_uid &&
207 pcred->p_ruid != chgp->p_ucred->cr_uid)
208 return (EPERM);
209 if (n > PRIO_MAX)
210 n = PRIO_MAX;
211 if (n < PRIO_MIN)
212 n = PRIO_MIN;
213 n += NZERO;
214 if (n < chgp->p_nice && suser(pcred->pc_ucred, &curp->p_acflag))
215 return (EACCES);
216 chgp->p_nice = n;
217 SCHED_LOCK(s);
218 (void)resetprocpriority(chgp);
219 SCHED_UNLOCK(s);
220 return (0);
221 }
222
223 /* ARGSUSED */
224 int
225 sys_setrlimit(l, v, retval)
226 struct lwp *l;
227 void *v;
228 register_t *retval;
229 {
230 struct sys_setrlimit_args /* {
231 syscallarg(int) which;
232 syscallarg(const struct rlimit *) rlp;
233 } */ *uap = v;
234 struct proc *p = l->l_proc;
235 int which = SCARG(uap, which);
236 struct rlimit alim;
237 int error;
238
239 error = copyin(SCARG(uap, rlp), &alim, sizeof(struct rlimit));
240 if (error)
241 return (error);
242 return (dosetrlimit(p, p->p_cred, which, &alim));
243 }
244
245 int
246 dosetrlimit(p, cred, which, limp)
247 struct proc *p;
248 struct pcred *cred;
249 int which;
250 struct rlimit *limp;
251 {
252 struct rlimit *alimp;
253 struct plimit *newplim;
254 int error;
255
256 if ((u_int)which >= RLIM_NLIMITS)
257 return (EINVAL);
258
259 if (limp->rlim_cur < 0 || limp->rlim_max < 0)
260 return (EINVAL);
261
262 alimp = &p->p_rlimit[which];
263 /* if we don't change the value, no need to limcopy() */
264 if (limp->rlim_cur == alimp->rlim_cur &&
265 limp->rlim_max == alimp->rlim_max)
266 return 0;
267
268 if (limp->rlim_cur > alimp->rlim_max ||
269 limp->rlim_max > alimp->rlim_max)
270 if ((error = suser(cred->pc_ucred, &p->p_acflag)) != 0)
271 return (error);
272 if (limp->rlim_cur > limp->rlim_max)
273 limp->rlim_cur = limp->rlim_max;
274 if (p->p_limit->p_refcnt > 1 &&
275 (p->p_limit->p_lflags & PL_SHAREMOD) == 0) {
276 newplim = limcopy(p->p_limit);
277 limfree(p->p_limit);
278 p->p_limit = newplim;
279 alimp = &p->p_rlimit[which];
280 }
281
282 switch (which) {
283
284 case RLIMIT_DATA:
285 if (limp->rlim_cur > maxdmap)
286 limp->rlim_cur = maxdmap;
287 if (limp->rlim_max > maxdmap)
288 limp->rlim_max = maxdmap;
289 break;
290
291 case RLIMIT_STACK:
292 if (limp->rlim_cur > maxsmap)
293 limp->rlim_cur = maxsmap;
294 if (limp->rlim_max > maxsmap)
295 limp->rlim_max = maxsmap;
296
297 /*
298 * Stack is allocated to the max at exec time with
299 * only "rlim_cur" bytes accessible (In other words,
300 * allocates stack dividing two contiguous regions at
301 * "rlim_cur" bytes boundary).
302 *
303 * Since allocation is done in terms of page, roundup
304 * "rlim_cur" (otherwise, contiguous regions
305 * overlap). If stack limit is going up make more
306 * accessible, if going down make inaccessible.
307 */
308 limp->rlim_cur = round_page(limp->rlim_cur);
309 if (limp->rlim_cur != alimp->rlim_cur) {
310 vaddr_t addr;
311 vsize_t size;
312 vm_prot_t prot;
313
314 if (limp->rlim_cur > alimp->rlim_cur) {
315 prot = VM_PROT_ALL;
316 size = limp->rlim_cur - alimp->rlim_cur;
317 addr = USRSTACK - limp->rlim_cur;
318 } else {
319 prot = VM_PROT_NONE;
320 size = alimp->rlim_cur - limp->rlim_cur;
321 addr = USRSTACK - alimp->rlim_cur;
322 }
323 (void) uvm_map_protect(&p->p_vmspace->vm_map,
324 addr, addr+size, prot, FALSE);
325 }
326 break;
327
328 case RLIMIT_NOFILE:
329 if (limp->rlim_cur > maxfiles)
330 limp->rlim_cur = maxfiles;
331 if (limp->rlim_max > maxfiles)
332 limp->rlim_max = maxfiles;
333 break;
334
335 case RLIMIT_NPROC:
336 if (limp->rlim_cur > maxproc)
337 limp->rlim_cur = maxproc;
338 if (limp->rlim_max > maxproc)
339 limp->rlim_max = maxproc;
340 break;
341 }
342 *alimp = *limp;
343 return (0);
344 }
345
346 /* ARGSUSED */
347 int
348 sys_getrlimit(l, v, retval)
349 struct lwp *l;
350 void *v;
351 register_t *retval;
352 {
353 struct sys_getrlimit_args /* {
354 syscallarg(int) which;
355 syscallarg(struct rlimit *) rlp;
356 } */ *uap = v;
357 struct proc *p = l->l_proc;
358 int which = SCARG(uap, which);
359
360 if ((u_int)which >= RLIM_NLIMITS)
361 return (EINVAL);
362 return (copyout(&p->p_rlimit[which], SCARG(uap, rlp),
363 sizeof(struct rlimit)));
364 }
365
366 /*
367 * Transform the running time and tick information in proc p into user,
368 * system, and interrupt time usage.
369 */
370 void
371 calcru(p, up, sp, ip)
372 struct proc *p;
373 struct timeval *up;
374 struct timeval *sp;
375 struct timeval *ip;
376 {
377 u_quad_t u, st, ut, it, tot;
378 long sec, usec;
379 int s;
380 struct timeval tv;
381 struct lwp *l;
382
383 s = splstatclock();
384 st = p->p_sticks;
385 ut = p->p_uticks;
386 it = p->p_iticks;
387 splx(s);
388
389 tot = st + ut + it;
390 if (tot == 0) {
391 up->tv_sec = up->tv_usec = 0;
392 sp->tv_sec = sp->tv_usec = 0;
393 if (ip != NULL)
394 ip->tv_sec = ip->tv_usec = 0;
395 return;
396 }
397
398 sec = p->p_rtime.tv_sec;
399 usec = p->p_rtime.tv_usec;
400 for (l = LIST_FIRST(&p->p_lwps); l != NULL;
401 l = LIST_NEXT(l, l_sibling)) {
402 if (l->l_stat == LSONPROC) {
403 struct schedstate_percpu *spc;
404
405 KDASSERT(l->l_cpu != NULL);
406 spc = &l->l_cpu->ci_schedstate;
407
408 /*
409 * Adjust for the current time slice. This is
410 * actually fairly important since the error
411 * here is on the order of a time quantum,
412 * which is much greater than the sampling
413 * error.
414 */
415 microtime(&tv);
416 sec += tv.tv_sec - spc->spc_runtime.tv_sec;
417 usec += tv.tv_usec - spc->spc_runtime.tv_usec;
418
419 break;
420 }
421 }
422 u = (u_quad_t) sec * 1000000 + usec;
423 st = (u * st) / tot;
424 sp->tv_sec = st / 1000000;
425 sp->tv_usec = st % 1000000;
426 ut = (u * ut) / tot;
427 up->tv_sec = ut / 1000000;
428 up->tv_usec = ut % 1000000;
429 if (ip != NULL) {
430 it = (u * it) / tot;
431 ip->tv_sec = it / 1000000;
432 ip->tv_usec = it % 1000000;
433 }
434 }
435
436 /* ARGSUSED */
437 int
438 sys_getrusage(l, v, retval)
439 struct lwp *l;
440 void *v;
441 register_t *retval;
442 {
443 struct sys_getrusage_args /* {
444 syscallarg(int) who;
445 syscallarg(struct rusage *) rusage;
446 } */ *uap = v;
447 struct rusage *rup;
448 struct proc *p = l->l_proc;
449
450 switch (SCARG(uap, who)) {
451
452 case RUSAGE_SELF:
453 rup = &p->p_stats->p_ru;
454 calcru(p, &rup->ru_utime, &rup->ru_stime, NULL);
455 break;
456
457 case RUSAGE_CHILDREN:
458 rup = &p->p_stats->p_cru;
459 break;
460
461 default:
462 return (EINVAL);
463 }
464 return (copyout(rup, SCARG(uap, rusage), sizeof(struct rusage)));
465 }
466
467 void
468 ruadd(ru, ru2)
469 struct rusage *ru, *ru2;
470 {
471 long *ip, *ip2;
472 int i;
473
474 timeradd(&ru->ru_utime, &ru2->ru_utime, &ru->ru_utime);
475 timeradd(&ru->ru_stime, &ru2->ru_stime, &ru->ru_stime);
476 if (ru->ru_maxrss < ru2->ru_maxrss)
477 ru->ru_maxrss = ru2->ru_maxrss;
478 ip = &ru->ru_first; ip2 = &ru2->ru_first;
479 for (i = &ru->ru_last - &ru->ru_first; i >= 0; i--)
480 *ip++ += *ip2++;
481 }
482
483 /*
484 * Make a copy of the plimit structure.
485 * We share these structures copy-on-write after fork,
486 * and copy when a limit is changed.
487 */
488 struct plimit *
489 limcopy(lim)
490 struct plimit *lim;
491 {
492 struct plimit *newlim;
493
494 newlim = pool_get(&plimit_pool, PR_WAITOK);
495 memcpy(newlim->pl_rlimit, lim->pl_rlimit,
496 sizeof(struct rlimit) * RLIM_NLIMITS);
497 if (lim->pl_corename == defcorename) {
498 newlim->pl_corename = defcorename;
499 } else {
500 newlim->pl_corename = malloc(strlen(lim->pl_corename)+1,
501 M_TEMP, M_WAITOK);
502 strcpy(newlim->pl_corename, lim->pl_corename);
503 }
504 newlim->p_lflags = 0;
505 newlim->p_refcnt = 1;
506 return (newlim);
507 }
508
509 void
510 limfree(lim)
511 struct plimit *lim;
512 {
513
514 if (--lim->p_refcnt > 0)
515 return;
516 #ifdef DIAGNOSTIC
517 if (lim->p_refcnt < 0)
518 panic("limfree");
519 #endif
520 if (lim->pl_corename != defcorename)
521 free(lim->pl_corename, M_TEMP);
522 pool_put(&plimit_pool, lim);
523 }
524
525 struct pstats *
526 pstatscopy(ps)
527 struct pstats *ps;
528 {
529
530 struct pstats *newps;
531
532 newps = pool_get(&pstats_pool, PR_WAITOK);
533
534 memset(&newps->pstat_startzero, 0,
535 (unsigned) ((caddr_t)&newps->pstat_endzero -
536 (caddr_t)&newps->pstat_startzero));
537 memcpy(&newps->pstat_startcopy, &ps->pstat_startcopy,
538 ((caddr_t)&newps->pstat_endcopy -
539 (caddr_t)&newps->pstat_startcopy));
540
541 return (newps);
542
543 }
544
545 void
546 pstatsfree(ps)
547 struct pstats *ps;
548 {
549
550 pool_put(&pstats_pool, ps);
551 }
552