]> git.saurik.com Git - apple/xnu.git/blob - bsd/kern/kern_resource.c
projects / apple / xnu.git / blob
? search:
summary | shortlog | log | commit | commitdiff | tree
blame | history | raw | HEAD
xnu-1228.0.2.tar.gz
[apple/xnu.git] / bsd / kern / kern_resource.c
1 /*
2 * Copyright (c) 2000-2007 Apple Inc. All rights reserved.
3 *
4 * @APPLE_OSREFERENCE_LICENSE_HEADER_START@
5 *
6 * This file contains Original Code and/or Modifications of Original Code
7 * as defined in and that are subject to the Apple Public Source License
8 * Version 2.0 (the 'License'). You may not use this file except in
9 * compliance with the License. The rights granted to you under the License
10 * may not be used to create, or enable the creation or redistribution of,
11 * unlawful or unlicensed copies of an Apple operating system, or to
12 * circumvent, violate, or enable the circumvention or violation of, any
13 * terms of an Apple operating system software license agreement.
14 *
15 * Please obtain a copy of the License at
16 * http://www.opensource.apple.com/apsl/ and read it before using this file.
17 *
18 * The Original Code and all software distributed under the License are
19 * distributed on an 'AS IS' basis, WITHOUT WARRANTY OF ANY KIND, EITHER
20 * EXPRESS OR IMPLIED, AND APPLE HEREBY DISCLAIMS ALL SUCH WARRANTIES,
21 * INCLUDING WITHOUT LIMITATION, ANY WARRANTIES OF MERCHANTABILITY,
22 * FITNESS FOR A PARTICULAR PURPOSE, QUIET ENJOYMENT OR NON-INFRINGEMENT.
23 * Please see the License for the specific language governing rights and
24 * limitations under the License.
25 *
26 * @APPLE_OSREFERENCE_LICENSE_HEADER_END@
27 */
28 /* Copyright (c) 1995, 1997 Apple Computer, Inc. All Rights Reserved */
29 /*-
30 * Copyright (c) 1982, 1986, 1991, 1993
31 * The Regents of the University of California. All rights reserved.
32 * (c) UNIX System Laboratories, Inc.
33 * All or some portions of this file are derived from material licensed
34 * to the University of California by American Telephone and Telegraph
35 * Co. or Unix System Laboratories, Inc. and are reproduced herein with
36 * the permission of UNIX System Laboratories, Inc.
37 *
38 * Redistribution and use in source and binary forms, with or without
39 * modification, are permitted provided that the following conditions
40 * are met:
41 * 1. Redistributions of source code must retain the above copyright
42 * notice, this list of conditions and the following disclaimer.
43 * 2. Redistributions in binary form must reproduce the above copyright
44 * notice, this list of conditions and the following disclaimer in the
45 * documentation and/or other materials provided with the distribution.
46 * 3. All advertising materials mentioning features or use of this software
47 * must display the following acknowledgement:
48 * This product includes software developed by the University of
49 * California, Berkeley and its contributors.
50 * 4. Neither the name of the University nor the names of its contributors
51 * may be used to endorse or promote products derived from this software
52 * without specific prior written permission.
53 *
54 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
55 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
56 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
57 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
58 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
59 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
60 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
61 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
62 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
63 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
64 * SUCH DAMAGE.
65 *
66 * @(#)kern_resource.c 8.5 (Berkeley) 1/21/94
67 */
68 /*
69 * NOTICE: This file was modified by SPARTA, Inc. in 2005 to introduce
70 * support for mandatory and extensible security protections. This notice
71 * is included in support of clause 2.2 (b) of the Apple Public License,
72 * Version 2.0.
73 */
74
75 #include <sys/param.h>
76 #include <sys/systm.h>
77 #include <sys/sysctl.h>
78 #include <sys/kernel.h>
79 #include <sys/file_internal.h>
80 #include <sys/resourcevar.h>
81 #include <sys/malloc.h>
82 #include <sys/proc_internal.h>
83 #include <sys/kauth.h>
84 #include <machine/spl.h>
85
86 #include <sys/mount_internal.h>
87 #include <sys/sysproto.h>
88
89 #include <bsm/audit_kernel.h>
90
91 #include <machine/vmparam.h>
92
93 #include <mach/mach_types.h>
94 #include <mach/time_value.h>
95 #include <mach/task.h>
96 #include <mach/task_info.h>
97 #include <mach/vm_map.h>
98 #include <mach/mach_vm.h>
99 #include <mach/thread_act.h> /* for thread_policy_set( ) */
100 #include <kern/lock.h>
101 #include <kern/thread.h>
102
103 #include <kern/task.h>
104 #include <kern/clock.h> /* for absolutetime_to_microtime() */
105 #include <netinet/in.h> /* for TRAFFIC_MGT_SO_BACKGROUND */
106 #include <sys/socketvar.h> /* for struct socket */
107
108 #include <vm/vm_map.h>
109
110 int donice(struct proc *curp, struct proc *chgp, int n);
111 int dosetrlimit(struct proc *p, u_int which, struct rlimit *limp);
112 static int do_background_thread(struct proc *curp, int priority);
113
114 rlim_t maxdmap = MAXDSIZ; /* XXX */
115 rlim_t maxsmap = MAXSSIZ - PAGE_SIZE; /* XXX */
116
117 /*
118 * Limits on the number of open files per process, and the number
119 * of child processes per process.
120 *
121 * Note: would be in kern/subr_param.c in FreeBSD.
122 */
123 int maxfilesperproc = OPEN_MAX; /* per-proc open files limit */
124
125 SYSCTL_INT( _kern, KERN_MAXPROCPERUID, maxprocperuid, CTLFLAG_RW,
126 &maxprocperuid, 0, "Maximum processes allowed per userid" );
127
128 SYSCTL_INT( _kern, KERN_MAXFILESPERPROC, maxfilesperproc, CTLFLAG_RW,
129 &maxfilesperproc, 0, "Maximum files allowed open per process" );
130
131 /* Args and fn for proc_iteration callback used in setpriority */
132 struct puser_nice_args {
133 proc_t curp;
134 int prio;
135 id_t who;
136 int * foundp;
137 int * errorp;
138 };
139 static int puser_donice_callback(proc_t p, void * arg);
140
141
142 /* Args and fn for proc_iteration callback used in setpriority */
143 struct ppgrp_nice_args {
144 proc_t curp;
145 int prio;
146 int * foundp;
147 int * errorp;
148 };
149 static int ppgrp_donice_callback(proc_t p, void * arg);
150
151 /*
152 * Resource controls and accounting.
153 */
154 int
155 getpriority(struct proc *curp, struct getpriority_args *uap, register_t *retval)
156 {
157 struct proc *p;
158 int low = PRIO_MAX + 1;
159 kauth_cred_t my_cred;
160
161 /* would also test (uap->who < 0), but id_t is unsigned */
162 if (uap->who > 0x7fffffff)
163 return (EINVAL);
164
165 switch (uap->which) {
166
167 case PRIO_PROCESS:
168 if (uap->who == 0) {
169 p = curp;
170 low = p->p_nice;
171 } else {
172 p = proc_find(uap->who);
173 if (p == 0)
174 break;
175 low = p->p_nice;
176 proc_rele(p);
177
178 }
179 break;
180
181 case PRIO_PGRP: {
182 struct pgrp *pg = PGRP_NULL;
183
184 if (uap->who == 0) {
185 /* returns the pgrp to ref */
186 pg = proc_pgrp(curp);
187 } else if ((pg = pgfind(uap->who)) == PGRP_NULL) {
188 break;
189 }
190 /* No need for iteration as it is a simple scan */
191 pgrp_lock(pg);
192 for (p = pg->pg_members.lh_first; p != 0; p = p->p_pglist.le_next) {
193 if (p->p_nice < low)
194 low = p->p_nice;
195 }
196 pgrp_unlock(pg);
197 pg_rele(pg);
198 break;
199 }
200
201 case PRIO_USER:
202 if (uap->who == 0)
203 uap->who = kauth_cred_getuid(kauth_cred_get());
204
205 proc_list_lock();
206
207 for (p = allproc.lh_first; p != 0; p = p->p_list.le_next) {
208 my_cred = kauth_cred_proc_ref(p);
209 if (kauth_cred_getuid(my_cred) == uap->who &&
210 p->p_nice < low)
211 low = p->p_nice;
212 kauth_cred_unref(&my_cred);
213 }
214
215 proc_list_unlock();
216
217 break;
218
219 case PRIO_DARWIN_THREAD: {
220 thread_t thread;
221 struct uthread *ut;
222
223 /* we currently only support the current thread */
224 if (uap->who != 0) {
225 return (EINVAL);
226 }
227
228 thread = current_thread();
229 ut = get_bsdthread_info(thread);
230
231 low = 0;
232 if ( (ut->uu_flag & UT_BACKGROUND) != 0 ) {
233 low = 1;
234 }
235 break;
236 }
237
238 default:
239 return (EINVAL);
240 }
241 if (low == PRIO_MAX + 1)
242 return (ESRCH);
243 *retval = low;
244 return (0);
245 }
246
247 /* call back function used for proc iteration in PRIO_USER */
248 static int
249 puser_donice_callback(proc_t p, void * arg)
250 {
251 int error, n;
252 struct puser_nice_args * pun = (struct puser_nice_args *)arg;
253 kauth_cred_t my_cred;
254
255 my_cred = kauth_cred_proc_ref(p);
256 if (kauth_cred_getuid(my_cred) == pun->who) {
257 error = donice(pun->curp, p, pun->prio);
258 if (pun->errorp != NULL)
259 *pun->errorp = error;
260 if (pun->foundp != NULL) {
261 n = *pun->foundp;
262 *pun->foundp = n+1;
263 }
264 }
265 kauth_cred_unref(&my_cred);
266
267 return(PROC_RETURNED);
268 }
269
270 /* call back function used for proc iteration in PRIO_PGRP */
271 static int
272 ppgrp_donice_callback(proc_t p, void * arg)
273 {
274 int error;
275 struct ppgrp_nice_args * pun = (struct ppgrp_nice_args *)arg;
276 int n;
277
278 error = donice(pun->curp, p, pun->prio);
279 if (pun->errorp != NULL)
280 *pun->errorp = error;
281 if (pun->foundp!= NULL) {
282 n = *pun->foundp;
283 *pun->foundp = n+1;
284 }
285
286 return(PROC_RETURNED);
287 }
288
289 /*
290 * Returns: 0 Success
291 * EINVAL
292 * ESRCH
293 * donice:EPERM
294 * donice:EACCES
295 */
296 /* ARGSUSED */
297 int
298 setpriority(struct proc *curp, struct setpriority_args *uap, __unused register_t *retval)
299 {
300 struct proc *p;
301 int found = 0, error = 0;
302 int refheld = 0;
303
304 AUDIT_ARG(cmd, uap->which);
305 AUDIT_ARG(owner, uap->who, 0);
306 AUDIT_ARG(value, uap->prio);
307
308 /* would also test (uap->who < 0), but id_t is unsigned */
309 if (uap->who > 0x7fffffff)
310 return (EINVAL);
311
312 switch (uap->which) {
313
314 case PRIO_PROCESS:
315 if (uap->who == 0)
316 p = curp;
317 else {
318 p = proc_find(uap->who);
319 if (p == 0)
320 break;
321 refheld = 1;
322 }
323 error = donice(curp, p, uap->prio);
324 found++;
325 if (refheld != 0)
326 proc_rele(p);
327 break;
328
329 case PRIO_PGRP: {
330 struct pgrp *pg = PGRP_NULL;
331 struct ppgrp_nice_args ppgrp;
332
333 if (uap->who == 0) {
334 pg = proc_pgrp(curp);
335 } else if ((pg = pgfind(uap->who)) == PGRP_NULL)
336 break;
337
338 ppgrp.curp = curp;
339 ppgrp.prio = uap->prio;
340 ppgrp.foundp = &found;
341 ppgrp.errorp = &error;
342
343 /* PGRP_DROPREF drops the reference on process group */
344 pgrp_iterate(pg, PGRP_DROPREF, ppgrp_donice_callback, (void *)&ppgrp, NULL, NULL);
345
346 break;
347 }
348
349 case PRIO_USER: {
350 struct puser_nice_args punice;
351
352 if (uap->who == 0)
353 uap->who = kauth_cred_getuid(kauth_cred_get());
354
355 punice.curp = curp;
356 punice.prio = uap->prio;
357 punice.who = uap->who;
358 punice.foundp = &found;
359 error = 0;
360 punice.errorp = &error;
361 proc_iterate(PROC_ALLPROCLIST, puser_donice_callback, (void *)&punice, NULL, NULL);
362
363 break;
364 }
365
366 case PRIO_DARWIN_THREAD: {
367 /* we currently only support the current thread */
368 if (uap->who != 0) {
369 return (EINVAL);
370 }
371 error = do_background_thread(curp, uap->prio);
372 found++;
373 break;
374 }
375
376 default:
377 return (EINVAL);
378 }
379 if (found == 0)
380 return (ESRCH);
381 return (error);
382 }
383
384
385 /*
386 * Returns: 0 Success
387 * EPERM
388 * EACCES
389 * mac_check_proc_sched:???
390 */
391 int
392 donice(struct proc *curp, struct proc *chgp, int n)
393 {
394 int error = 0;
395 kauth_cred_t ucred;
396 kauth_cred_t my_cred;
397
398 ucred = kauth_cred_proc_ref(curp);
399 my_cred = kauth_cred_proc_ref(chgp);
400
401 if (suser(ucred, NULL) && ucred->cr_ruid &&
402 kauth_cred_getuid(ucred) != kauth_cred_getuid(my_cred) &&
403 ucred->cr_ruid != kauth_cred_getuid(my_cred)) {
404 error = EPERM;
405 goto out;
406 }
407 if (n > PRIO_MAX)
408 n = PRIO_MAX;
409 if (n < PRIO_MIN)
410 n = PRIO_MIN;
411 if (n < chgp->p_nice && suser(ucred, &curp->p_acflag)) {
412 error = EACCES;
413 goto out;
414 }
415 #if CONFIG_MACF
416 error = mac_proc_check_sched(curp, chgp);
417 if (error)
418 goto out;
419 #endif
420 proc_lock(chgp);
421 chgp->p_nice = n;
422 proc_unlock(chgp);
423 (void)resetpriority(chgp);
424 out:
425 kauth_cred_unref(&ucred);
426 kauth_cred_unref(&my_cred);
427 return (error);
428 }
429
430 /*
431 * do_background_thread
432 * Returns: 0 Success
433 * XXX - todo - does this need a MACF hook?
434 */
435 static int
436 do_background_thread(struct proc *curp, int priority)
437 {
438 int i;
439 thread_t thread;
440 struct uthread *ut;
441 thread_precedence_policy_data_t policy;
442 struct filedesc *fdp;
443 struct fileproc *fp;
444
445 thread = current_thread();
446 ut = get_bsdthread_info(thread);
447
448 if ( (priority & PRIO_DARWIN_BG) == 0 ) {
449 /* turn off backgrounding of thread */
450 if ( (ut->uu_flag & UT_BACKGROUND) == 0 ) {
451 /* already off */
452 return(0);
453 }
454
455 /* clear background bit in thread and disable disk IO throttle */
456 ut->uu_flag &= ~UT_BACKGROUND;
457 ut->uu_iopol_disk = IOPOL_NORMAL;
458
459 /* reset thread priority (we did not save previous value) */
460 policy.importance = 0;
461 thread_policy_set( thread, THREAD_PRECEDENCE_POLICY,
462 (thread_policy_t)&policy,
463 THREAD_PRECEDENCE_POLICY_COUNT );
464
465 /* disable networking IO throttle.
466 * NOTE - It is a known limitation of the current design that we
467 * could potentially clear TRAFFIC_MGT_SO_BACKGROUND bit for
468 * sockets created by other threads within this process.
469 */
470 proc_fdlock(curp);
471 fdp = curp->p_fd;
472 for ( i = 0; i < fdp->fd_nfiles; i++ ) {
473 struct socket *sockp;
474
475 fp = fdp->fd_ofiles[ i ];
476 if ( fp == NULL || (fdp->fd_ofileflags[ i ] & UF_RESERVED) != 0 ||
477 fp->f_fglob->fg_type != DTYPE_SOCKET ) {
478 continue;
479 }
480 sockp = (struct socket *)fp->f_fglob->fg_data;
481 if ( sockp->so_background_thread != thread ) {
482 continue;
483 }
484 sockp->so_traffic_mgt_flags &= ~TRAFFIC_MGT_SO_BACKGROUND;
485 sockp->so_background_thread = NULL;
486 }
487 proc_fdunlock(curp);
488
489 return(0);
490 }
491
492 /* background this thread */
493 if ( (ut->uu_flag & UT_BACKGROUND) != 0 ) {
494 /* already backgrounded */
495 return(0);
496 }
497
498 /* tag thread as background and throttle disk IO */
499 ut->uu_flag |= UT_BACKGROUND;
500 ut->uu_iopol_disk = IOPOL_THROTTLE;
501
502 policy.importance = INT_MIN;
503 thread_policy_set( thread, THREAD_PRECEDENCE_POLICY,
504 (thread_policy_t)&policy,
505 THREAD_PRECEDENCE_POLICY_COUNT );
506
507 /* throttle networking IO happens in socket( ) syscall.
508 * If UT_BACKGROUND is set in the current thread then
509 * TRAFFIC_MGT_SO_BACKGROUND socket option is set.
510 */
511 return(0);
512 }
513
514
515 /*
516 * Returns: 0 Success
517 * copyin:EFAULT
518 * dosetrlimit:
519 */
520 /* ARGSUSED */
521 int
522 setrlimit(struct proc *p, struct setrlimit_args *uap, __unused register_t *retval)
523 {
524 struct rlimit alim;
525 int error;
526
527 if ((error = copyin(uap->rlp, (caddr_t)&alim,
528 sizeof (struct rlimit))))
529 return (error);
530
531 return (dosetrlimit(p, uap->which, &alim));
532 }
533
534 /*
535 * Returns: 0 Success
536 * EINVAL
537 * ENOMEM Cannot copy limit structure
538 * suser:EPERM
539 *
540 * Notes: EINVAL is returned both for invalid arguments, and in the
541 * case that the current usage (e.g. RLIMIT_STACK) is already
542 * in excess of the requested limit.
543 */
544 int
545 dosetrlimit(struct proc *p, u_int which, struct rlimit *limp)
546 {
547 struct rlimit *alimp;
548 int error;
549 kern_return_t kr;
550 int posix = (which & _RLIMIT_POSIX_FLAG) ? 1 : 0;
551
552 /* Mask out POSIX flag, saved above */
553 which &= ~_RLIMIT_POSIX_FLAG;
554
555 if (which >= RLIM_NLIMITS)
556 return (EINVAL);
557
558 alimp = &p->p_rlimit[which];
559 if (limp->rlim_cur > limp->rlim_max)
560 return EINVAL;
561
562 if (limp->rlim_cur > alimp->rlim_max ||
563 limp->rlim_max > alimp->rlim_max)
564 if ((error = suser(kauth_cred_get(), &p->p_acflag))) {
565 return (error);
566 }
567
568 proc_limitblock(p);
569
570 if ((error = proc_limitreplace(p)) != 0) {
571 proc_limitunblock(p);
572 return(error);
573 }
574
575 alimp = &p->p_rlimit[which];
576
577 switch (which) {
578
579 case RLIMIT_CPU:
580 if (limp->rlim_cur == RLIM_INFINITY) {
581 task_vtimer_clear(p->task, TASK_VTIMER_RLIM);
582 timerclear(&p->p_rlim_cpu);
583 }
584 else {
585 task_absolutetime_info_data_t tinfo;
586 mach_msg_type_number_t count;
587 struct timeval ttv, tv;
588
589 count = TASK_ABSOLUTETIME_INFO_COUNT;
590 task_info(p->task, TASK_ABSOLUTETIME_INFO,
591 (task_info_t)&tinfo, &count);
592 absolutetime_to_microtime(tinfo.total_user + tinfo.total_system,
593 (uint32_t *) &ttv.tv_sec, (uint32_t *) &ttv.tv_usec);
594
595 tv.tv_sec = (limp->rlim_cur > __INT_MAX__ ? __INT_MAX__ : limp->rlim_cur);
596 tv.tv_usec = 0;
597 timersub(&tv, &ttv, &p->p_rlim_cpu);
598
599 timerclear(&tv);
600 if (timercmp(&p->p_rlim_cpu, &tv, >))
601 task_vtimer_set(p->task, TASK_VTIMER_RLIM);
602 else {
603 task_vtimer_clear(p->task, TASK_VTIMER_RLIM);
604
605 timerclear(&p->p_rlim_cpu);
606
607 psignal(p, SIGXCPU);
608 }
609 }
610 break;
611
612 case RLIMIT_DATA:
613 if (limp->rlim_cur > maxdmap)
614 limp->rlim_cur = maxdmap;
615 if (limp->rlim_max > maxdmap)
616 limp->rlim_max = maxdmap;
617 break;
618
619 case RLIMIT_STACK:
620 /* Disallow illegal stack size instead of clipping */
621 if (limp->rlim_cur > maxsmap ||
622 limp->rlim_max > maxsmap) {
623 if (posix) {
624 error = EINVAL;
625 goto out;
626 }
627 else {
628 /*
629 * 4797860 - workaround poorly written installers by
630 * doing previous implementation (< 10.5) when caller
631 * is non-POSIX conforming.
632 */
633 if (limp->rlim_cur > maxsmap)
634 limp->rlim_cur = maxsmap;
635 if (limp->rlim_max > maxsmap)
636 limp->rlim_max = maxsmap;
637 }
638 }
639
640 /*
641 * Stack is allocated to the max at exec time with only
642 * "rlim_cur" bytes accessible. If stack limit is going
643 * up make more accessible, if going down make inaccessible.
644 */
645 if (limp->rlim_cur > alimp->rlim_cur) {
646 user_addr_t addr;
647 user_size_t size;
648
649 /* grow stack */
650 size = round_page_64(limp->rlim_cur);
651 size -= round_page_64(alimp->rlim_cur);
652
653 #if STACK_GROWTH_UP
654 /* go to top of current stack */
655 addr = p->user_stack + round_page_64(alimp->rlim_cur);
656 #else /* STACK_GROWTH_UP */
657 addr = p->user_stack - round_page_64(limp->rlim_cur);
658 #endif /* STACK_GROWTH_UP */
659 kr = mach_vm_protect(current_map(),
660 addr, size,
661 FALSE, VM_PROT_DEFAULT);
662 if (kr != KERN_SUCCESS) {
663 error = EINVAL;
664 goto out;
665 }
666 } else if (limp->rlim_cur < alimp->rlim_cur) {
667 user_addr_t addr;
668 user_size_t size;
669 user_addr_t cur_sp;
670
671 /* shrink stack */
672
673 /*
674 * First check if new stack limit would agree
675 * with current stack usage.
676 * Get the current thread's stack pointer...
677 */
678 cur_sp = thread_adjuserstack(current_thread(),
679 0);
680 #if STACK_GROWTH_UP
681 if (cur_sp >= p->user_stack &&
682 cur_sp < (p->user_stack +
683 round_page_64(alimp->rlim_cur))) {
684 /* current stack pointer is in main stack */
685 if (cur_sp >= (p->user_stack +
686 round_page_64(limp->rlim_cur))) {
687 /*
688 * New limit would cause
689 * current usage to be invalid:
690 * reject new limit.
691 */
692 error = EINVAL;
693 goto out;
694 }
695 } else {
696 /* not on the main stack: reject */
697 error = EINVAL;
698 goto out;
699 }
700
701 #else /* STACK_GROWTH_UP */
702 if (cur_sp <= p->user_stack &&
703 cur_sp > (p->user_stack -
704 round_page_64(alimp->rlim_cur))) {
705 /* stack pointer is in main stack */
706 if (cur_sp <= (p->user_stack -
707 round_page_64(limp->rlim_cur))) {
708 /*
709 * New limit would cause
710 * current usage to be invalid:
711 * reject new limit.
712 */
713 error = EINVAL;
714 goto out;
715 }
716 } else {
717 /* not on the main stack: reject */
718 error = EINVAL;
719 goto out;
720 }
721 #endif /* STACK_GROWTH_UP */
722
723 size = round_page_64(alimp->rlim_cur);
724 size -= round_page_64(limp->rlim_cur);
725
726 #if STACK_GROWTH_UP
727 addr = p->user_stack + round_page_64(limp->rlim_cur);
728 #else /* STACK_GROWTH_UP */
729 addr = p->user_stack - round_page_64(alimp->rlim_cur);
730 #endif /* STACK_GROWTH_UP */
731
732 kr = mach_vm_protect(current_map(),
733 addr, size,
734 FALSE, VM_PROT_NONE);
735 if (kr != KERN_SUCCESS) {
736 error = EINVAL;
737 goto out;
738 }
739 } else {
740 /* no change ... */
741 }
742 break;
743
744 case RLIMIT_NOFILE:
745 /*
746 * Only root can set the maxfiles limits, as it is
747 * systemwide resource. If we are expecting POSIX behavior,
748 * instead of clamping the value, return EINVAL. We do this
749 * because historically, people have been able to attempt to
750 * set RLIM_INFINITY to get "whatever the maximum is".
751 */
752 if ( is_suser() ) {
753 if (limp->rlim_cur != alimp->rlim_cur &&
754 limp->rlim_cur > (rlim_t)maxfiles) {
755 if (posix) {
756 error = EINVAL;
757 goto out;
758 }
759 limp->rlim_cur = maxfiles;
760 }
761 if (limp->rlim_max != alimp->rlim_max &&
762 limp->rlim_max > (rlim_t)maxfiles)
763 limp->rlim_max = maxfiles;
764 }
765 else {
766 if (limp->rlim_cur != alimp->rlim_cur &&
767 limp->rlim_cur > (rlim_t)maxfilesperproc) {
768 if (posix) {
769 error = EINVAL;
770 goto out;
771 }
772 limp->rlim_cur = maxfilesperproc;
773 }
774 if (limp->rlim_max != alimp->rlim_max &&
775 limp->rlim_max > (rlim_t)maxfilesperproc)
776 limp->rlim_max = maxfilesperproc;
777 }
778 break;
779
780 case RLIMIT_NPROC:
781 /*
782 * Only root can set to the maxproc limits, as it is
783 * systemwide resource; all others are limited to
784 * maxprocperuid (presumably less than maxproc).
785 */
786 if ( is_suser() ) {
787 if (limp->rlim_cur > (rlim_t)maxproc)
788 limp->rlim_cur = maxproc;
789 if (limp->rlim_max > (rlim_t)maxproc)
790 limp->rlim_max = maxproc;
791 }
792 else {
793 if (limp->rlim_cur > (rlim_t)maxprocperuid)
794 limp->rlim_cur = maxprocperuid;
795 if (limp->rlim_max > (rlim_t)maxprocperuid)
796 limp->rlim_max = maxprocperuid;
797 }
798 break;
799
800 case RLIMIT_MEMLOCK:
801 /*
802 * Tell the Mach VM layer about the new limit value.
803 */
804
805 vm_map_set_user_wire_limit(current_map(), limp->rlim_cur);
806 break;
807
808 } /* switch... */
809 proc_lock(p);
810 *alimp = *limp;
811 proc_unlock(p);
812 error = 0;
813 out:
814 proc_limitunblock(p);
815 return (error);
816 }
817
818 /* ARGSUSED */
819 int
820 getrlimit(struct proc *p, struct getrlimit_args *uap, __unused register_t *retval)
821 {
822 struct rlimit lim;
823
824 /*
825 * Take out flag now in case we need to use it to trigger variant
826 * behaviour later.
827 */
828 uap->which &= ~_RLIMIT_POSIX_FLAG;
829
830 if (uap->which >= RLIM_NLIMITS)
831 return (EINVAL);
832 proc_limitget(p, uap->which, &lim);
833 return (copyout((caddr_t)&lim,
834 uap->rlp, sizeof (struct rlimit)));
835 }
836
837 /*
838 * Transform the running time and tick information in proc p into user,
839 * system, and interrupt time usage.
840 */
841 /* No lock on proc is held for this.. */
842 void
843 calcru(struct proc *p, struct timeval *up, struct timeval *sp, struct timeval *ip)
844 {
845 task_t task;
846
847 timerclear(up);
848 timerclear(sp);
849 if (ip != NULL)
850 timerclear(ip);
851
852 task = p->task;
853 if (task) {
854 task_basic_info_data_t tinfo;
855 task_thread_times_info_data_t ttimesinfo;
856 mach_msg_type_number_t task_info_stuff, task_ttimes_stuff;
857 struct timeval ut,st;
858
859 task_info_stuff = TASK_BASIC_INFO_COUNT;
860 task_info(task, TASK_BASIC_INFO,
861 (task_info_t)&tinfo, &task_info_stuff);
862 ut.tv_sec = tinfo.user_time.seconds;
863 ut.tv_usec = tinfo.user_time.microseconds;
864 st.tv_sec = tinfo.system_time.seconds;
865 st.tv_usec = tinfo.system_time.microseconds;
866 timeradd(&ut, up, up);
867 timeradd(&st, sp, sp);
868
869 task_ttimes_stuff = TASK_THREAD_TIMES_INFO_COUNT;
870 task_info(task, TASK_THREAD_TIMES_INFO,
871 (task_info_t)&ttimesinfo, &task_ttimes_stuff);
872
873 ut.tv_sec = ttimesinfo.user_time.seconds;
874 ut.tv_usec = ttimesinfo.user_time.microseconds;
875 st.tv_sec = ttimesinfo.system_time.seconds;
876 st.tv_usec = ttimesinfo.system_time.microseconds;
877 timeradd(&ut, up, up);
878 timeradd(&st, sp, sp);
879 }
880 }
881
882 __private_extern__ void munge_rusage(struct rusage *a_rusage_p, struct user_rusage *a_user_rusage_p);
883
884 /* ARGSUSED */
885 int
886 getrusage(struct proc *p, struct getrusage_args *uap, __unused register_t *retval)
887 {
888 struct rusage *rup, rubuf;
889 struct user_rusage rubuf64;
890 size_t retsize = sizeof(rubuf); /* default: 32 bits */
891 caddr_t retbuf = (caddr_t)&rubuf; /* default: 32 bits */
892 struct timeval utime;
893 struct timeval stime;
894
895
896 switch (uap->who) {
897 case RUSAGE_SELF:
898 calcru(p, &utime, &stime, NULL);
899 // LP64todo: proc struct should have 64 bit version of struct
900 proc_lock(p);
901 rup = &p->p_stats->p_ru;
902 rup->ru_utime = utime;
903 rup->ru_stime = stime;
904
905 rubuf = *rup;
906 proc_unlock(p);
907
908 break;
909
910 case RUSAGE_CHILDREN:
911 proc_lock(p);
912 rup = &p->p_stats->p_cru;
913 rubuf = *rup;
914 proc_unlock(p);
915 break;
916
917 default:
918 return (EINVAL);
919 }
920 if (IS_64BIT_PROCESS(p)) {
921 retsize = sizeof(rubuf64);
922 retbuf = (caddr_t)&rubuf64;
923 munge_rusage(&rubuf, &rubuf64);
924 }
925 return (copyout(retbuf, uap->rusage, retsize));
926 }
927
928 void
929 ruadd(struct rusage *ru, struct rusage *ru2)
930 {
931 long *ip, *ip2;
932 int i;
933
934 timeradd(&ru->ru_utime, &ru2->ru_utime, &ru->ru_utime);
935 timeradd(&ru->ru_stime, &ru2->ru_stime, &ru->ru_stime);
936 if (ru->ru_maxrss < ru2->ru_maxrss)
937 ru->ru_maxrss = ru2->ru_maxrss;
938 ip = &ru->ru_first; ip2 = &ru2->ru_first;
939 for (i = &ru->ru_last - &ru->ru_first; i >= 0; i--)
940 *ip++ += *ip2++;
941 }
942
943 void
944 proc_limitget(proc_t p, int which, struct rlimit * limp)
945 {
946 proc_list_lock();
947 limp->rlim_cur = p->p_rlimit[which].rlim_cur;
948 limp->rlim_max = p->p_rlimit[which].rlim_max;
949 proc_list_unlock();
950 }
951
952
953 void
954 proc_limitdrop(proc_t p, int exiting)
955 {
956 struct plimit * freelim = NULL;
957 struct plimit * freeoldlim = NULL;
958
959 proc_list_lock();
960
961 if (--p->p_limit->pl_refcnt == 0) {
962 freelim = p->p_limit;
963 p->p_limit = NULL;
964 }
965 if ((exiting != 0) && (p->p_olimit != NULL) && (--p->p_olimit->pl_refcnt == 0)) {
966 freeoldlim = p->p_olimit;
967 p->p_olimit = NULL;
968 }
969
970 proc_list_unlock();
971 if (freelim != NULL)
972 FREE_ZONE(freelim, sizeof *p->p_limit, M_PLIMIT);
973 if (freeoldlim != NULL)
974 FREE_ZONE(freeoldlim, sizeof *p->p_olimit, M_PLIMIT);
975 }
976
977
978 void
979 proc_limitfork(proc_t parent, proc_t child)
980 {
981 proc_list_lock();
982 child->p_limit = parent->p_limit;
983 child->p_limit->pl_refcnt++;
984 child->p_olimit = NULL;
985 proc_list_unlock();
986 }
987
988 void
989 proc_limitblock(proc_t p)
990 {
991 proc_lock(p);
992 while (p->p_lflag & P_LLIMCHANGE) {
993 p->p_lflag |= P_LLIMWAIT;
994 msleep(&p->p_olimit, &p->p_mlock, 0, "proc_limitblock", NULL);
995 }
996 p->p_lflag |= P_LLIMCHANGE;
997 proc_unlock(p);
998
999 }
1000
1001
1002 void
1003 proc_limitunblock(proc_t p)
1004 {
1005 proc_lock(p);
1006 p->p_lflag &= ~P_LLIMCHANGE;
1007 if (p->p_lflag & P_LLIMWAIT) {
1008 p->p_lflag &= ~P_LLIMWAIT;
1009 wakeup(&p->p_olimit);
1010 }
1011 proc_unlock(p);
1012 }
1013
1014 /* This is called behind serialization provided by proc_limitblock/unlbock */
1015 int
1016 proc_limitreplace(proc_t p)
1017 {
1018 struct plimit *copy;
1019
1020
1021 proc_list_lock();
1022
1023 if (p->p_limit->pl_refcnt == 1) {
1024 proc_list_unlock();
1025 return(0);
1026 }
1027
1028 proc_list_unlock();
1029
1030 MALLOC_ZONE(copy, struct plimit *,
1031 sizeof(struct plimit), M_PLIMIT, M_WAITOK);
1032 if (copy == NULL) {
1033 return(ENOMEM);
1034 }
1035
1036 proc_list_lock();
1037 bcopy(p->p_limit->pl_rlimit, copy->pl_rlimit,
1038 sizeof(struct rlimit) * RLIM_NLIMITS);
1039 copy->pl_refcnt = 1;
1040 /* hang on to reference to old till process exits */
1041 p->p_olimit = p->p_limit;
1042 p->p_limit = copy;
1043 proc_list_unlock();
1044
1045 return(0);
1046 }
1047
1048
1049 /*
1050 * iopolicysys
1051 *
1052 * Description: System call MUX for use in manipulating I/O policy attributes of the current process or thread
1053 *
1054 * Parameters: cmd Policy command
1055 * arg Pointer to policy arguments
1056 *
1057 * Returns: 0 Success
1058 * EINVAL Invalid command or invalid policy arguments
1059 *
1060 */
1061 int
1062 iopolicysys(__unused struct proc *p, __unused struct iopolicysys_args *uap, __unused register_t *retval)
1063 {
1064 int error = 0;
1065 thread_t thread = THREAD_NULL;
1066 int *policy;
1067 struct uthread *ut = NULL;
1068 struct _iopol_param_t iop_param;
1069
1070 if ((error = copyin(uap->arg, &iop_param, sizeof(iop_param))) != 0)
1071 goto exit;
1072
1073 if (iop_param.iop_iotype != IOPOL_TYPE_DISK) {
1074 error = EINVAL;
1075 goto exit;
1076 }
1077
1078 switch (iop_param.iop_scope) {
1079 case IOPOL_SCOPE_PROCESS:
1080 policy = &p->p_iopol_disk;
1081 break;
1082 case IOPOL_SCOPE_THREAD:
1083 thread = current_thread();
1084 ut = get_bsdthread_info(thread);
1085 policy = &ut->uu_iopol_disk;
1086 break;
1087 default:
1088 error = EINVAL;
1089 goto exit;
1090 }
1091
1092 switch(uap->cmd) {
1093 case IOPOL_CMD_SET:
1094 switch (iop_param.iop_policy) {
1095 case IOPOL_DEFAULT:
1096 case IOPOL_NORMAL:
1097 case IOPOL_THROTTLE:
1098 case IOPOL_PASSIVE:
1099 proc_lock(p);
1100 *policy = iop_param.iop_policy;
1101 proc_unlock(p);
1102 break;
1103 default:
1104 error = EINVAL;
1105 goto exit;
1106 }
1107 break;
1108 case IOPOL_CMD_GET:
1109 switch (*policy) {
1110 case IOPOL_DEFAULT:
1111 case IOPOL_NORMAL:
1112 case IOPOL_THROTTLE:
1113 case IOPOL_PASSIVE:
1114 iop_param.iop_policy = *policy;
1115 break;
1116 default: // in-kernel
1117 // this should never happen
1118 printf("%s: unknown I/O policy %d\n", __func__, *policy);
1119 // restore to default value
1120 *policy = IOPOL_DEFAULT;
1121 iop_param.iop_policy = *policy;
1122 }
1123
1124 error = copyout((caddr_t)&iop_param, uap->arg, sizeof(iop_param));
1125 break;
1126 default:
1127 error = EINVAL; // unknown command
1128 break;
1129 }
1130
1131 exit:
1132 *retval = error;
1133 return (error);
1134 }
mirror of XNU