]> git.saurik.com Git - apple/xnu.git/blob - bsd/kern/kern_resource.c
projects / apple / xnu.git / blob
? search:


b51c4ecbe35ae692fcbd3d52c6db058e8382134d
[apple/xnu.git] / bsd / kern / kern_resource.c
1 /*
2 * Copyright (c) 2000-2008 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 <security/audit/audit.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 __private_extern__ 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, int32_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 int32_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(value32, 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 int32_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 clock_sec_t tv_sec;
589 clock_usec_t tv_usec;
590
591 count = TASK_ABSOLUTETIME_INFO_COUNT;
592 task_info(p->task, TASK_ABSOLUTETIME_INFO,
593 (task_info_t)&tinfo, &count);
594 absolutetime_to_microtime(tinfo.total_user + tinfo.total_system,
595 &tv_sec, &tv_usec);
596 ttv.tv_sec = tv_sec;
597 ttv.tv_usec = tv_usec;
598
599 tv.tv_sec = (limp->rlim_cur > __INT_MAX__ ? __INT_MAX__ : limp->rlim_cur);
600 tv.tv_usec = 0;
601 timersub(&tv, &ttv, &p->p_rlim_cpu);
602
603 timerclear(&tv);
604 if (timercmp(&p->p_rlim_cpu, &tv, >))
605 task_vtimer_set(p->task, TASK_VTIMER_RLIM);
606 else {
607 task_vtimer_clear(p->task, TASK_VTIMER_RLIM);
608
609 timerclear(&p->p_rlim_cpu);
610
611 psignal(p, SIGXCPU);
612 }
613 }
614 break;
615
616 case RLIMIT_DATA:
617 if (limp->rlim_cur > maxdmap)
618 limp->rlim_cur = maxdmap;
619 if (limp->rlim_max > maxdmap)
620 limp->rlim_max = maxdmap;
621 break;
622
623 case RLIMIT_STACK:
624 /* Disallow illegal stack size instead of clipping */
625 if (limp->rlim_cur > maxsmap ||
626 limp->rlim_max > maxsmap) {
627 if (posix) {
628 error = EINVAL;
629 goto out;
630 }
631 else {
632 /*
633 * 4797860 - workaround poorly written installers by
634 * doing previous implementation (< 10.5) when caller
635 * is non-POSIX conforming.
636 */
637 if (limp->rlim_cur > maxsmap)
638 limp->rlim_cur = maxsmap;
639 if (limp->rlim_max > maxsmap)
640 limp->rlim_max = maxsmap;
641 }
642 }
643
644 /*
645 * Stack is allocated to the max at exec time with only
646 * "rlim_cur" bytes accessible. If stack limit is going
647 * up make more accessible, if going down make inaccessible.
648 */
649 if (limp->rlim_cur > alimp->rlim_cur) {
650 user_addr_t addr;
651 user_size_t size;
652
653 /* grow stack */
654 size = round_page_64(limp->rlim_cur);
655 size -= round_page_64(alimp->rlim_cur);
656
657 #if STACK_GROWTH_UP
658 /* go to top of current stack */
659 addr = p->user_stack + round_page_64(alimp->rlim_cur);
660 #else /* STACK_GROWTH_UP */
661 addr = p->user_stack - round_page_64(limp->rlim_cur);
662 #endif /* STACK_GROWTH_UP */
663 kr = mach_vm_protect(current_map(),
664 addr, size,
665 FALSE, VM_PROT_DEFAULT);
666 if (kr != KERN_SUCCESS) {
667 error = EINVAL;
668 goto out;
669 }
670 } else if (limp->rlim_cur < alimp->rlim_cur) {
671 user_addr_t addr;
672 user_size_t size;
673 user_addr_t cur_sp;
674
675 /* shrink stack */
676
677 /*
678 * First check if new stack limit would agree
679 * with current stack usage.
680 * Get the current thread's stack pointer...
681 */
682 cur_sp = thread_adjuserstack(current_thread(),
683 0);
684 #if STACK_GROWTH_UP
685 if (cur_sp >= p->user_stack &&
686 cur_sp < (p->user_stack +
687 round_page_64(alimp->rlim_cur))) {
688 /* current stack pointer is in main stack */
689 if (cur_sp >= (p->user_stack +
690 round_page_64(limp->rlim_cur))) {
691 /*
692 * New limit would cause
693 * current usage to be invalid:
694 * reject new limit.
695 */
696 error = EINVAL;
697 goto out;
698 }
699 } else {
700 /* not on the main stack: reject */
701 error = EINVAL;
702 goto out;
703 }
704
705 #else /* STACK_GROWTH_UP */
706 if (cur_sp <= p->user_stack &&
707 cur_sp > (p->user_stack -
708 round_page_64(alimp->rlim_cur))) {
709 /* stack pointer is in main stack */
710 if (cur_sp <= (p->user_stack -
711 round_page_64(limp->rlim_cur))) {
712 /*
713 * New limit would cause
714 * current usage to be invalid:
715 * reject new limit.
716 */
717 error = EINVAL;
718 goto out;
719 }
720 } else {
721 /* not on the main stack: reject */
722 error = EINVAL;
723 goto out;
724 }
725 #endif /* STACK_GROWTH_UP */
726
727 size = round_page_64(alimp->rlim_cur);
728 size -= round_page_64(limp->rlim_cur);
729
730 #if STACK_GROWTH_UP
731 addr = p->user_stack + round_page_64(limp->rlim_cur);
732 #else /* STACK_GROWTH_UP */
733 addr = p->user_stack - round_page_64(alimp->rlim_cur);
734 #endif /* STACK_GROWTH_UP */
735
736 kr = mach_vm_protect(current_map(),
737 addr, size,
738 FALSE, VM_PROT_NONE);
739 if (kr != KERN_SUCCESS) {
740 error = EINVAL;
741 goto out;
742 }
743 } else {
744 /* no change ... */
745 }
746 break;
747
748 case RLIMIT_NOFILE:
749 /*
750 * Only root can set the maxfiles limits, as it is
751 * systemwide resource. If we are expecting POSIX behavior,
752 * instead of clamping the value, return EINVAL. We do this
753 * because historically, people have been able to attempt to
754 * set RLIM_INFINITY to get "whatever the maximum is".
755 */
756 if ( is_suser() ) {
757 if (limp->rlim_cur != alimp->rlim_cur &&
758 limp->rlim_cur > (rlim_t)maxfiles) {
759 if (posix) {
760 error = EINVAL;
761 goto out;
762 }
763 limp->rlim_cur = maxfiles;
764 }
765 if (limp->rlim_max != alimp->rlim_max &&
766 limp->rlim_max > (rlim_t)maxfiles)
767 limp->rlim_max = maxfiles;
768 }
769 else {
770 if (limp->rlim_cur != alimp->rlim_cur &&
771 limp->rlim_cur > (rlim_t)maxfilesperproc) {
772 if (posix) {
773 error = EINVAL;
774 goto out;
775 }
776 limp->rlim_cur = maxfilesperproc;
777 }
778 if (limp->rlim_max != alimp->rlim_max &&
779 limp->rlim_max > (rlim_t)maxfilesperproc)
780 limp->rlim_max = maxfilesperproc;
781 }
782 break;
783
784 case RLIMIT_NPROC:
785 /*
786 * Only root can set to the maxproc limits, as it is
787 * systemwide resource; all others are limited to
788 * maxprocperuid (presumably less than maxproc).
789 */
790 if ( is_suser() ) {
791 if (limp->rlim_cur > (rlim_t)maxproc)
792 limp->rlim_cur = maxproc;
793 if (limp->rlim_max > (rlim_t)maxproc)
794 limp->rlim_max = maxproc;
795 }
796 else {
797 if (limp->rlim_cur > (rlim_t)maxprocperuid)
798 limp->rlim_cur = maxprocperuid;
799 if (limp->rlim_max > (rlim_t)maxprocperuid)
800 limp->rlim_max = maxprocperuid;
801 }
802 break;
803
804 case RLIMIT_MEMLOCK:
805 /*
806 * Tell the Mach VM layer about the new limit value.
807 */
808
809 vm_map_set_user_wire_limit(current_map(), limp->rlim_cur);
810 break;
811
812 } /* switch... */
813 proc_lock(p);
814 *alimp = *limp;
815 proc_unlock(p);
816 error = 0;
817 out:
818 proc_limitunblock(p);
819 return (error);
820 }
821
822 /* ARGSUSED */
823 int
824 getrlimit(struct proc *p, struct getrlimit_args *uap, __unused int32_t *retval)
825 {
826 struct rlimit lim;
827
828 /*
829 * Take out flag now in case we need to use it to trigger variant
830 * behaviour later.
831 */
832 uap->which &= ~_RLIMIT_POSIX_FLAG;
833
834 if (uap->which >= RLIM_NLIMITS)
835 return (EINVAL);
836 proc_limitget(p, uap->which, &lim);
837 return (copyout((caddr_t)&lim,
838 uap->rlp, sizeof (struct rlimit)));
839 }
840
841 /*
842 * Transform the running time and tick information in proc p into user,
843 * system, and interrupt time usage.
844 */
845 /* No lock on proc is held for this.. */
846 void
847 calcru(struct proc *p, struct timeval *up, struct timeval *sp, struct timeval *ip)
848 {
849 task_t task;
850
851 timerclear(up);
852 timerclear(sp);
853 if (ip != NULL)
854 timerclear(ip);
855
856 task = p->task;
857 if (task) {
858 task_basic_info_32_data_t tinfo;
859 task_thread_times_info_data_t ttimesinfo;
860 task_events_info_data_t teventsinfo;
861 mach_msg_type_number_t task_info_count, task_ttimes_count;
862 mach_msg_type_number_t task_events_count;
863 struct timeval ut,st;
864
865 task_info_count = TASK_BASIC_INFO_32_COUNT;
866 task_info(task, TASK_BASIC2_INFO_32,
867 (task_info_t)&tinfo, &task_info_count);
868 ut.tv_sec = tinfo.user_time.seconds;
869 ut.tv_usec = tinfo.user_time.microseconds;
870 st.tv_sec = tinfo.system_time.seconds;
871 st.tv_usec = tinfo.system_time.microseconds;
872 timeradd(&ut, up, up);
873 timeradd(&st, sp, sp);
874
875 task_ttimes_count = TASK_THREAD_TIMES_INFO_COUNT;
876 task_info(task, TASK_THREAD_TIMES_INFO,
877 (task_info_t)&ttimesinfo, &task_ttimes_count);
878
879 ut.tv_sec = ttimesinfo.user_time.seconds;
880 ut.tv_usec = ttimesinfo.user_time.microseconds;
881 st.tv_sec = ttimesinfo.system_time.seconds;
882 st.tv_usec = ttimesinfo.system_time.microseconds;
883 timeradd(&ut, up, up);
884 timeradd(&st, sp, sp);
885
886 task_events_count = TASK_EVENTS_INFO_COUNT;
887 task_info(task, TASK_EVENTS_INFO,
888 (task_info_t)&teventsinfo, &task_events_count);
889
890 /*
891 * No need to lock "p": this does not need to be
892 * completely consistent, right ?
893 */
894 p->p_stats->p_ru.ru_minflt = (teventsinfo.faults -
895 teventsinfo.pageins);
896 p->p_stats->p_ru.ru_majflt = teventsinfo.pageins;
897 p->p_stats->p_ru.ru_nivcsw = (teventsinfo.csw -
898 p->p_stats->p_ru.ru_nvcsw);
899 if (p->p_stats->p_ru.ru_nivcsw < 0)
900 p->p_stats->p_ru.ru_nivcsw = 0;
901
902 p->p_stats->p_ru.ru_maxrss = tinfo.resident_size;
903 }
904 }
905
906 __private_extern__ void munge_user64_rusage(struct rusage *a_rusage_p, struct user64_rusage *a_user_rusage_p);
907 __private_extern__ void munge_user32_rusage(struct rusage *a_rusage_p, struct user32_rusage *a_user_rusage_p);
908
909 /* ARGSUSED */
910 int
911 getrusage(struct proc *p, struct getrusage_args *uap, __unused int32_t *retval)
912 {
913 struct rusage *rup, rubuf;
914 struct user64_rusage rubuf64;
915 struct user32_rusage rubuf32;
916 size_t retsize = sizeof(rubuf); /* default: 32 bits */
917 caddr_t retbuf = (caddr_t)&rubuf; /* default: 32 bits */
918 struct timeval utime;
919 struct timeval stime;
920
921
922 switch (uap->who) {
923 case RUSAGE_SELF:
924 calcru(p, &utime, &stime, NULL);
925 proc_lock(p);
926 rup = &p->p_stats->p_ru;
927 rup->ru_utime = utime;
928 rup->ru_stime = stime;
929
930 rubuf = *rup;
931 proc_unlock(p);
932
933 break;
934
935 case RUSAGE_CHILDREN:
936 proc_lock(p);
937 rup = &p->p_stats->p_cru;
938 rubuf = *rup;
939 proc_unlock(p);
940 break;
941
942 default:
943 return (EINVAL);
944 }
945 if (IS_64BIT_PROCESS(p)) {
946 retsize = sizeof(rubuf64);
947 retbuf = (caddr_t)&rubuf64;
948 munge_user64_rusage(&rubuf, &rubuf64);
949 } else {
950 retsize = sizeof(rubuf32);
951 retbuf = (caddr_t)&rubuf32;
952 munge_user32_rusage(&rubuf, &rubuf32);
953 }
954
955 return (copyout(retbuf, uap->rusage, retsize));
956 }
957
958 void
959 ruadd(struct rusage *ru, struct rusage *ru2)
960 {
961 long *ip, *ip2;
962 long i;
963
964 timeradd(&ru->ru_utime, &ru2->ru_utime, &ru->ru_utime);
965 timeradd(&ru->ru_stime, &ru2->ru_stime, &ru->ru_stime);
966 if (ru->ru_maxrss < ru2->ru_maxrss)
967 ru->ru_maxrss = ru2->ru_maxrss;
968 ip = &ru->ru_first; ip2 = &ru2->ru_first;
969 for (i = &ru->ru_last - &ru->ru_first; i >= 0; i--)
970 *ip++ += *ip2++;
971 }
972
973 void
974 proc_limitget(proc_t p, int which, struct rlimit * limp)
975 {
976 proc_list_lock();
977 limp->rlim_cur = p->p_rlimit[which].rlim_cur;
978 limp->rlim_max = p->p_rlimit[which].rlim_max;
979 proc_list_unlock();
980 }
981
982
983 void
984 proc_limitdrop(proc_t p, int exiting)
985 {
986 struct plimit * freelim = NULL;
987 struct plimit * freeoldlim = NULL;
988
989 proc_list_lock();
990
991 if (--p->p_limit->pl_refcnt == 0) {
992 freelim = p->p_limit;
993 p->p_limit = NULL;
994 }
995 if ((exiting != 0) && (p->p_olimit != NULL) && (--p->p_olimit->pl_refcnt == 0)) {
996 freeoldlim = p->p_olimit;
997 p->p_olimit = NULL;
998 }
999
1000 proc_list_unlock();
1001 if (freelim != NULL)
1002 FREE_ZONE(freelim, sizeof *p->p_limit, M_PLIMIT);
1003 if (freeoldlim != NULL)
1004 FREE_ZONE(freeoldlim, sizeof *p->p_olimit, M_PLIMIT);
1005 }
1006
1007
1008 void
1009 proc_limitfork(proc_t parent, proc_t child)
1010 {
1011 proc_list_lock();
1012 child->p_limit = parent->p_limit;
1013 child->p_limit->pl_refcnt++;
1014 child->p_olimit = NULL;
1015 proc_list_unlock();
1016 }
1017
1018 void
1019 proc_limitblock(proc_t p)
1020 {
1021 proc_lock(p);
1022 while (p->p_lflag & P_LLIMCHANGE) {
1023 p->p_lflag |= P_LLIMWAIT;
1024 msleep(&p->p_olimit, &p->p_mlock, 0, "proc_limitblock", NULL);
1025 }
1026 p->p_lflag |= P_LLIMCHANGE;
1027 proc_unlock(p);
1028
1029 }
1030
1031
1032 void
1033 proc_limitunblock(proc_t p)
1034 {
1035 proc_lock(p);
1036 p->p_lflag &= ~P_LLIMCHANGE;
1037 if (p->p_lflag & P_LLIMWAIT) {
1038 p->p_lflag &= ~P_LLIMWAIT;
1039 wakeup(&p->p_olimit);
1040 }
1041 proc_unlock(p);
1042 }
1043
1044 /* This is called behind serialization provided by proc_limitblock/unlbock */
1045 int
1046 proc_limitreplace(proc_t p)
1047 {
1048 struct plimit *copy;
1049
1050
1051 proc_list_lock();
1052
1053 if (p->p_limit->pl_refcnt == 1) {
1054 proc_list_unlock();
1055 return(0);
1056 }
1057
1058 proc_list_unlock();
1059
1060 MALLOC_ZONE(copy, struct plimit *,
1061 sizeof(struct plimit), M_PLIMIT, M_WAITOK);
1062 if (copy == NULL) {
1063 return(ENOMEM);
1064 }
1065
1066 proc_list_lock();
1067 bcopy(p->p_limit->pl_rlimit, copy->pl_rlimit,
1068 sizeof(struct rlimit) * RLIM_NLIMITS);
1069 copy->pl_refcnt = 1;
1070 /* hang on to reference to old till process exits */
1071 p->p_olimit = p->p_limit;
1072 p->p_limit = copy;
1073 proc_list_unlock();
1074
1075 return(0);
1076 }
1077
1078
1079 /*
1080 * iopolicysys
1081 *
1082 * Description: System call MUX for use in manipulating I/O policy attributes of the current process or thread
1083 *
1084 * Parameters: cmd Policy command
1085 * arg Pointer to policy arguments
1086 *
1087 * Returns: 0 Success
1088 * EINVAL Invalid command or invalid policy arguments
1089 *
1090 */
1091 int
1092 iopolicysys(__unused struct proc *p, __unused struct iopolicysys_args *uap, __unused int32_t *retval)
1093 {
1094 int error = 0;
1095 thread_t thread = THREAD_NULL;
1096 int *policy;
1097 struct uthread *ut = NULL;
1098 struct _iopol_param_t iop_param;
1099
1100 if ((error = copyin(uap->arg, &iop_param, sizeof(iop_param))) != 0)
1101 goto exit;
1102
1103 if (iop_param.iop_iotype != IOPOL_TYPE_DISK) {
1104 error = EINVAL;
1105 goto exit;
1106 }
1107
1108 switch (iop_param.iop_scope) {
1109 case IOPOL_SCOPE_PROCESS:
1110 policy = &p->p_iopol_disk;
1111 break;
1112 case IOPOL_SCOPE_THREAD:
1113 thread = current_thread();
1114 ut = get_bsdthread_info(thread);
1115 policy = &ut->uu_iopol_disk;
1116 break;
1117 default:
1118 error = EINVAL;
1119 goto exit;
1120 }
1121
1122 switch(uap->cmd) {
1123 case IOPOL_CMD_SET:
1124 switch (iop_param.iop_policy) {
1125 case IOPOL_DEFAULT:
1126 case IOPOL_NORMAL:
1127 case IOPOL_THROTTLE:
1128 case IOPOL_PASSIVE:
1129 proc_lock(p);
1130 *policy = iop_param.iop_policy;
1131 proc_unlock(p);
1132 break;
1133 default:
1134 error = EINVAL;
1135 goto exit;
1136 }
1137 break;
1138 case IOPOL_CMD_GET:
1139 switch (*policy) {
1140 case IOPOL_DEFAULT:
1141 case IOPOL_NORMAL:
1142 case IOPOL_THROTTLE:
1143 case IOPOL_PASSIVE:
1144 iop_param.iop_policy = *policy;
1145 break;
1146 default: // in-kernel
1147 // this should never happen
1148 printf("%s: unknown I/O policy %d\n", __func__, *policy);
1149 // restore to default value
1150 *policy = IOPOL_DEFAULT;
1151 iop_param.iop_policy = *policy;
1152 }
1153
1154 error = copyout((caddr_t)&iop_param, uap->arg, sizeof(iop_param));
1155 break;
1156 default:
1157 error = EINVAL; // unknown command
1158 break;
1159 }
1160
1161 exit:
1162 *retval = error;
1163 return (error);
1164 }
1165
1166
1167 boolean_t thread_is_io_throttled(void);
1168
1169 boolean_t
1170 thread_is_io_throttled(void) {
1171
1172 int policy;
1173 struct uthread *ut;
1174
1175 policy = current_proc()->p_iopol_disk;
1176
1177 ut = get_bsdthread_info(current_thread());
1178
1179 if (ut->uu_iopol_disk != IOPOL_DEFAULT)
1180 policy = ut->uu_iopol_disk;
1181
1182 if (policy == IOPOL_THROTTLE)
1183 return TRUE;
1184
1185 return FALSE;
1186 }
mirror of XNU