1 /*- 2 * Copyright (c) 1982, 1986, 1991, 1993 3 * The Regents of the University of California. All rights reserved. 4 * (c) UNIX System Laboratories, Inc. 5 * All or some portions of this file are derived from material licensed 6 * to the University of California by American Telephone and Telegraph 7 * Co. or Unix System Laboratories, Inc. and are reproduced herein with 8 * the permission of UNIX System Laboratories, Inc. 9 * 10 * Redistribution and use in source and binary forms, with or without 11 * modification, are permitted provided that the following conditions 12 * are met: 13 * 1. Redistributions of source code must retain the above copyright 14 * notice, this list of conditions and the following disclaimer. 15 * 2. Redistributions in binary form must reproduce the above copyright 16 * notice, this list of conditions and the following disclaimer in the 17 * documentation and/or other materials provided with the distribution. 18 * 3. All advertising materials mentioning features or use of this software 19 * must display the following acknowledgement: 20 * This product includes software developed by the University of 21 * California, Berkeley and its contributors. 22 * 4. Neither the name of the University nor the names of its contributors 23 * may be used to endorse or promote products derived from this software 24 * without specific prior written permission. 25 * 26 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 27 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 28 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 29 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 30 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 31 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 32 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 33 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 34 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 35 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 36 * SUCH DAMAGE. 37 * 38 * @(#)kern_resource.c 8.5 (Berkeley) 1/21/94 39 * $FreeBSD: src/sys/kern/kern_resource.c,v 1.55.2.5 2001/11/03 01:41:08 ps Exp $ 40 * $DragonFly: src/sys/kern/kern_resource.c,v 1.31 2007/02/03 17:05:58 corecode Exp $ 41 */ 42 43 #include "opt_compat.h" 44 45 #include <sys/param.h> 46 #include <sys/systm.h> 47 #include <sys/sysproto.h> 48 #include <sys/file.h> 49 #include <sys/kern_syscall.h> 50 #include <sys/kernel.h> 51 #include <sys/resourcevar.h> 52 #include <sys/malloc.h> 53 #include <sys/proc.h> 54 #include <sys/time.h> 55 #include <sys/lockf.h> 56 57 #include <vm/vm.h> 58 #include <vm/vm_param.h> 59 #include <sys/lock.h> 60 #include <vm/pmap.h> 61 #include <vm/vm_map.h> 62 63 #include <sys/thread2.h> 64 65 static int donice (struct proc *chgp, int n); 66 67 static MALLOC_DEFINE(M_UIDINFO, "uidinfo", "uidinfo structures"); 68 #define UIHASH(uid) (&uihashtbl[(uid) & uihash]) 69 static LIST_HEAD(uihashhead, uidinfo) *uihashtbl; 70 static u_long uihash; /* size of hash table - 1 */ 71 72 static struct uidinfo *uicreate (uid_t uid); 73 static struct uidinfo *uilookup (uid_t uid); 74 75 /* 76 * Resource controls and accounting. 77 */ 78 79 struct getpriority_info { 80 int low; 81 int who; 82 }; 83 84 static int getpriority_callback(struct proc *p, void *data); 85 86 int 87 sys_getpriority(struct getpriority_args *uap) 88 { 89 struct getpriority_info info; 90 struct proc *curp = curproc; 91 struct proc *p; 92 int low = PRIO_MAX + 1; 93 94 switch (uap->which) { 95 case PRIO_PROCESS: 96 if (uap->who == 0) 97 p = curp; 98 else 99 p = pfind(uap->who); 100 if (p == 0) 101 break; 102 if (!PRISON_CHECK(curp->p_ucred, p->p_ucred)) 103 break; 104 low = p->p_nice; 105 break; 106 107 case PRIO_PGRP: 108 { 109 struct pgrp *pg; 110 111 if (uap->who == 0) 112 pg = curp->p_pgrp; 113 else if ((pg = pgfind(uap->who)) == NULL) 114 break; 115 LIST_FOREACH(p, &pg->pg_members, p_pglist) { 116 if ((PRISON_CHECK(curp->p_ucred, p->p_ucred) && p->p_nice < low)) 117 low = p->p_nice; 118 } 119 break; 120 } 121 case PRIO_USER: 122 if (uap->who == 0) 123 uap->who = curp->p_ucred->cr_uid; 124 info.low = low; 125 info.who = uap->who; 126 allproc_scan(getpriority_callback, &info); 127 low = info.low; 128 break; 129 130 default: 131 return (EINVAL); 132 } 133 if (low == PRIO_MAX + 1) 134 return (ESRCH); 135 uap->sysmsg_result = low; 136 return (0); 137 } 138 139 /* 140 * Figure out the current lowest nice priority for processes owned 141 * by the specified user. 142 */ 143 static 144 int 145 getpriority_callback(struct proc *p, void *data) 146 { 147 struct getpriority_info *info = data; 148 149 if (PRISON_CHECK(curproc->p_ucred, p->p_ucred) && 150 p->p_ucred->cr_uid == info->who && 151 p->p_nice < info->low) { 152 info->low = p->p_nice; 153 } 154 return(0); 155 } 156 157 struct setpriority_info { 158 int prio; 159 int who; 160 int error; 161 int found; 162 }; 163 164 static int setpriority_callback(struct proc *p, void *data); 165 166 int 167 sys_setpriority(struct setpriority_args *uap) 168 { 169 struct setpriority_info info; 170 struct proc *curp = curproc; 171 struct proc *p; 172 int found = 0, error = 0; 173 174 switch (uap->which) { 175 case PRIO_PROCESS: 176 if (uap->who == 0) 177 p = curp; 178 else 179 p = pfind(uap->who); 180 if (p == 0) 181 break; 182 if (!PRISON_CHECK(curp->p_ucred, p->p_ucred)) 183 break; 184 error = donice(p, uap->prio); 185 found++; 186 break; 187 188 case PRIO_PGRP: 189 { 190 struct pgrp *pg; 191 192 if (uap->who == 0) 193 pg = curp->p_pgrp; 194 else if ((pg = pgfind(uap->who)) == NULL) 195 break; 196 LIST_FOREACH(p, &pg->pg_members, p_pglist) { 197 if (PRISON_CHECK(curp->p_ucred, p->p_ucred)) { 198 error = donice(p, uap->prio); 199 found++; 200 } 201 } 202 break; 203 } 204 case PRIO_USER: 205 if (uap->who == 0) 206 uap->who = curp->p_ucred->cr_uid; 207 info.prio = uap->prio; 208 info.who = uap->who; 209 info.error = 0; 210 info.found = 0; 211 allproc_scan(setpriority_callback, &info); 212 error = info.error; 213 found = info.found; 214 break; 215 216 default: 217 return (EINVAL); 218 } 219 if (found == 0) 220 return (ESRCH); 221 return (error); 222 } 223 224 static 225 int 226 setpriority_callback(struct proc *p, void *data) 227 { 228 struct setpriority_info *info = data; 229 int error; 230 231 if (p->p_ucred->cr_uid == info->who && 232 PRISON_CHECK(curproc->p_ucred, p->p_ucred)) { 233 error = donice(p, info->prio); 234 if (error) 235 info->error = error; 236 ++info->found; 237 } 238 return(0); 239 } 240 241 static int 242 donice(struct proc *chgp, int n) 243 { 244 struct proc *curp = curproc; 245 struct ucred *cr = curp->p_ucred; 246 struct lwp *lp; 247 248 if (cr->cr_uid && cr->cr_ruid && 249 cr->cr_uid != chgp->p_ucred->cr_uid && 250 cr->cr_ruid != chgp->p_ucred->cr_uid) 251 return (EPERM); 252 if (n > PRIO_MAX) 253 n = PRIO_MAX; 254 if (n < PRIO_MIN) 255 n = PRIO_MIN; 256 if (n < chgp->p_nice && suser_cred(cr, 0)) 257 return (EACCES); 258 chgp->p_nice = n; 259 FOREACH_LWP_IN_PROC(lp, chgp) 260 chgp->p_usched->resetpriority(lp); 261 return (0); 262 } 263 264 /* 265 * Set realtime priority 266 */ 267 /* ARGSUSED */ 268 int 269 sys_rtprio(struct rtprio_args *uap) 270 { 271 struct proc *curp = curproc; 272 struct proc *p; 273 struct lwp *lp; 274 struct ucred *cr = curp->p_ucred; 275 struct rtprio rtp; 276 int error; 277 278 error = copyin(uap->rtp, &rtp, sizeof(struct rtprio)); 279 if (error) 280 return (error); 281 282 if (uap->pid == 0) 283 p = curp; 284 else 285 p = pfind(uap->pid); 286 287 if (p == 0) 288 return (ESRCH); 289 290 /* XXX lwp */ 291 lp = FIRST_LWP_IN_PROC(p); 292 switch (uap->function) { 293 case RTP_LOOKUP: 294 return (copyout(&lp->lwp_rtprio, uap->rtp, sizeof(struct rtprio))); 295 case RTP_SET: 296 if (cr->cr_uid && cr->cr_ruid && 297 cr->cr_uid != p->p_ucred->cr_uid && 298 cr->cr_ruid != p->p_ucred->cr_uid) 299 return (EPERM); 300 /* disallow setting rtprio in most cases if not superuser */ 301 if (suser_cred(cr, 0)) { 302 /* can't set someone else's */ 303 if (uap->pid) 304 return (EPERM); 305 /* can't set realtime priority */ 306 /* 307 * Realtime priority has to be restricted for reasons which should be 308 * obvious. However, for idle priority, there is a potential for 309 * system deadlock if an idleprio process gains a lock on a resource 310 * that other processes need (and the idleprio process can't run 311 * due to a CPU-bound normal process). Fix me! XXX 312 */ 313 if (RTP_PRIO_IS_REALTIME(rtp.type)) 314 return (EPERM); 315 } 316 switch (rtp.type) { 317 #ifdef RTP_PRIO_FIFO 318 case RTP_PRIO_FIFO: 319 #endif 320 case RTP_PRIO_REALTIME: 321 case RTP_PRIO_NORMAL: 322 case RTP_PRIO_IDLE: 323 if (rtp.prio > RTP_PRIO_MAX) 324 return (EINVAL); 325 lp->lwp_rtprio = rtp; 326 return (0); 327 default: 328 return (EINVAL); 329 } 330 331 default: 332 return (EINVAL); 333 } 334 } 335 336 int 337 sys_setrlimit(struct __setrlimit_args *uap) 338 { 339 struct rlimit alim; 340 int error; 341 342 error = copyin(uap->rlp, &alim, sizeof(alim)); 343 if (error) 344 return (error); 345 346 error = kern_setrlimit(uap->which, &alim); 347 348 return (error); 349 } 350 351 int 352 sys_getrlimit(struct __getrlimit_args *uap) 353 { 354 struct rlimit lim; 355 int error; 356 357 error = kern_getrlimit(uap->which, &lim); 358 359 if (error == 0) 360 error = copyout(&lim, uap->rlp, sizeof(*uap->rlp)); 361 return error; 362 } 363 364 /* 365 * Transform the running time and tick information in lwp lp's thread into user, 366 * system, and interrupt time usage. 367 * 368 * Since we are limited to statclock tick granularity this is a statisical 369 * calculation which will be correct over the long haul, but should not be 370 * expected to measure fine grained deltas. 371 */ 372 void 373 calcru(struct lwp *lp, struct timeval *up, struct timeval *sp) 374 { 375 struct thread *td = lp->lwp_thread; 376 377 /* 378 * Calculate at the statclock level. YYY if the thread is owned by 379 * another cpu we need to forward the request to the other cpu, or 380 * have a token to interlock the information. 381 */ 382 crit_enter(); 383 up->tv_sec = td->td_uticks / 1000000; 384 up->tv_usec = td->td_uticks % 1000000; 385 sp->tv_sec = td->td_sticks / 1000000; 386 sp->tv_usec = td->td_sticks % 1000000; 387 crit_exit(); 388 } 389 390 /* 391 * Aggregate resource statistics of all lwps of a process. 392 * 393 * proc.p_ru keeps track of all statistics directly related to a proc. This 394 * consists of RSS usage and nswap information and aggregate numbers for all 395 * former lwps of this proc. 396 * 397 * proc.p_cru is the sum of all stats of reaped children. 398 * 399 * lwp.lwp_ru contains the stats directly related to one specific lwp, meaning 400 * packet, scheduler switch or page fault counts, etc. This information gets 401 * added to lwp.lwp_proc.p_ru when the lwp exits. 402 */ 403 void 404 calcru_proc(struct proc *p, struct rusage *ru) 405 { 406 struct timeval upt, spt; 407 long *rip1, *rip2; 408 struct lwp *lp; 409 410 *ru = p->p_ru; 411 412 FOREACH_LWP_IN_PROC(lp, p) { 413 calcru(lp, &upt, &spt); 414 timevaladd(&ru->ru_utime, &upt); 415 timevaladd(&ru->ru_stime, &spt); 416 for (rip1 = &ru->ru_first, rip2 = &lp->lwp_ru.ru_first; 417 rip1 <= &ru->ru_last; 418 rip1++, rip2++) 419 *rip1 += *rip2; 420 } 421 } 422 423 424 /* ARGSUSED */ 425 int 426 sys_getrusage(struct getrusage_args *uap) 427 { 428 struct rusage ru; 429 struct rusage *rup; 430 431 switch (uap->who) { 432 433 case RUSAGE_SELF: 434 rup = &ru; 435 calcru_proc(curproc, rup); 436 break; 437 438 case RUSAGE_CHILDREN: 439 rup = &curproc->p_cru; 440 break; 441 442 default: 443 return (EINVAL); 444 } 445 return (copyout((caddr_t)rup, (caddr_t)uap->rusage, 446 sizeof (struct rusage))); 447 } 448 449 void 450 ruadd(struct rusage *ru, struct rusage *ru2) 451 { 452 long *ip, *ip2; 453 int i; 454 455 timevaladd(&ru->ru_utime, &ru2->ru_utime); 456 timevaladd(&ru->ru_stime, &ru2->ru_stime); 457 if (ru->ru_maxrss < ru2->ru_maxrss) 458 ru->ru_maxrss = ru2->ru_maxrss; 459 ip = &ru->ru_first; ip2 = &ru2->ru_first; 460 for (i = &ru->ru_last - &ru->ru_first; i >= 0; i--) 461 *ip++ += *ip2++; 462 } 463 464 /* 465 * Find the uidinfo structure for a uid. This structure is used to 466 * track the total resource consumption (process count, socket buffer 467 * size, etc.) for the uid and impose limits. 468 */ 469 void 470 uihashinit(void) 471 { 472 uihashtbl = hashinit(maxproc / 16, M_UIDINFO, &uihash); 473 } 474 475 static struct uidinfo * 476 uilookup(uid_t uid) 477 { 478 struct uihashhead *uipp; 479 struct uidinfo *uip; 480 481 uipp = UIHASH(uid); 482 LIST_FOREACH(uip, uipp, ui_hash) { 483 if (uip->ui_uid == uid) 484 break; 485 } 486 return (uip); 487 } 488 489 static struct uidinfo * 490 uicreate(uid_t uid) 491 { 492 struct uidinfo *uip, *norace; 493 494 /* 495 * Allocate space and check for a race 496 */ 497 MALLOC(uip, struct uidinfo *, sizeof(*uip), M_UIDINFO, M_WAITOK); 498 norace = uilookup(uid); 499 if (norace != NULL) { 500 FREE(uip, M_UIDINFO); 501 return (norace); 502 } 503 504 /* 505 * Initialize structure and enter it into the hash table 506 */ 507 LIST_INSERT_HEAD(UIHASH(uid), uip, ui_hash); 508 uip->ui_uid = uid; 509 uip->ui_proccnt = 0; 510 uip->ui_sbsize = 0; 511 uip->ui_ref = 0; 512 uip->ui_posixlocks = 0; 513 varsymset_init(&uip->ui_varsymset, NULL); 514 return (uip); 515 } 516 517 struct uidinfo * 518 uifind(uid_t uid) 519 { 520 struct uidinfo *uip; 521 522 uip = uilookup(uid); 523 if (uip == NULL) 524 uip = uicreate(uid); 525 uip->ui_ref++; 526 return (uip); 527 } 528 529 static __inline void 530 uifree(struct uidinfo *uip) 531 { 532 if (uip->ui_sbsize != 0) 533 /* XXX no %qd in kernel. Truncate. */ 534 kprintf("freeing uidinfo: uid = %d, sbsize = %ld\n", 535 uip->ui_uid, (long)uip->ui_sbsize); 536 if (uip->ui_proccnt != 0) 537 kprintf("freeing uidinfo: uid = %d, proccnt = %ld\n", 538 uip->ui_uid, uip->ui_proccnt); 539 LIST_REMOVE(uip, ui_hash); 540 varsymset_clean(&uip->ui_varsymset); 541 FREE(uip, M_UIDINFO); 542 } 543 544 void 545 uihold(struct uidinfo *uip) 546 { 547 ++uip->ui_ref; 548 KKASSERT(uip->ui_ref > 0); 549 } 550 551 void 552 uidrop(struct uidinfo *uip) 553 { 554 KKASSERT(uip->ui_ref > 0); 555 if (--uip->ui_ref == 0) 556 uifree(uip); 557 } 558 559 void 560 uireplace(struct uidinfo **puip, struct uidinfo *nuip) 561 { 562 uidrop(*puip); 563 *puip = nuip; 564 } 565 566 /* 567 * Change the count associated with number of processes 568 * a given user is using. When 'max' is 0, don't enforce a limit 569 */ 570 int 571 chgproccnt(struct uidinfo *uip, int diff, int max) 572 { 573 /* don't allow them to exceed max, but allow subtraction */ 574 if (diff > 0 && uip->ui_proccnt + diff > max && max != 0) 575 return (0); 576 uip->ui_proccnt += diff; 577 if (uip->ui_proccnt < 0) 578 kprintf("negative proccnt for uid = %d\n", uip->ui_uid); 579 return (1); 580 } 581 582 /* 583 * Change the total socket buffer size a user has used. 584 */ 585 int 586 chgsbsize(struct uidinfo *uip, u_long *hiwat, u_long to, rlim_t max) 587 { 588 rlim_t new; 589 590 crit_enter(); 591 new = uip->ui_sbsize + to - *hiwat; 592 /* don't allow them to exceed max, but allow subtraction */ 593 if (to > *hiwat && new > max) { 594 crit_exit(); 595 return (0); 596 } 597 uip->ui_sbsize = new; 598 *hiwat = to; 599 if (uip->ui_sbsize < 0) 600 kprintf("negative sbsize for uid = %d\n", uip->ui_uid); 601 crit_exit(); 602 return (1); 603 } 604 605