kern_proc.c revision 1.42 1 /* $NetBSD: kern_proc.c,v 1.42 2000/08/17 14:37:54 thorpej Exp $ */
2
3 /*-
4 * Copyright (c) 1999 The NetBSD Foundation, Inc.
5 * All rights reserved.
6 *
7 * This code is derived from software contributed to The NetBSD Foundation
8 * by Jason R. Thorpe of the Numerical Aerospace Simulation Facility,
9 * NASA Ames Research Center.
10 *
11 * Redistribution and use in source and binary forms, with or without
12 * modification, are permitted provided that the following conditions
13 * are met:
14 * 1. Redistributions of source code must retain the above copyright
15 * notice, this list of conditions and the following disclaimer.
16 * 2. Redistributions in binary form must reproduce the above copyright
17 * notice, this list of conditions and the following disclaimer in the
18 * documentation and/or other materials provided with the distribution.
19 * 3. All advertising materials mentioning features or use of this software
20 * must display the following acknowledgement:
21 * This product includes software developed by the NetBSD
22 * Foundation, Inc. and its contributors.
23 * 4. Neither the name of The NetBSD Foundation nor the names of its
24 * contributors may be used to endorse or promote products derived
25 * from this software without specific prior written permission.
26 *
27 * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS
28 * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
29 * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
30 * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS
31 * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
32 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
33 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
34 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
35 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
36 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
37 * POSSIBILITY OF SUCH DAMAGE.
38 */
39
40 /*
41 * Copyright (c) 1982, 1986, 1989, 1991, 1993
42 * The Regents of the University of California. All rights reserved.
43 *
44 * Redistribution and use in source and binary forms, with or without
45 * modification, are permitted provided that the following conditions
46 * are met:
47 * 1. Redistributions of source code must retain the above copyright
48 * notice, this list of conditions and the following disclaimer.
49 * 2. Redistributions in binary form must reproduce the above copyright
50 * notice, this list of conditions and the following disclaimer in the
51 * documentation and/or other materials provided with the distribution.
52 * 3. All advertising materials mentioning features or use of this software
53 * must display the following acknowledgement:
54 * This product includes software developed by the University of
55 * California, Berkeley and its contributors.
56 * 4. Neither the name of the University nor the names of its contributors
57 * may be used to endorse or promote products derived from this software
58 * without specific prior written permission.
59 *
60 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
61 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
62 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
63 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
64 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
65 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
66 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
67 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
68 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
69 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
70 * SUCH DAMAGE.
71 *
72 * @(#)kern_proc.c 8.7 (Berkeley) 2/14/95
73 */
74
75 #include <sys/param.h>
76 #include <sys/systm.h>
77 #include <sys/map.h>
78 #include <sys/kernel.h>
79 #include <sys/proc.h>
80 #include <sys/resourcevar.h>
81 #include <sys/buf.h>
82 #include <sys/acct.h>
83 #include <sys/wait.h>
84 #include <sys/file.h>
85 #include <ufs/ufs/quota.h>
86 #include <sys/uio.h>
87 #include <sys/malloc.h>
88 #include <sys/pool.h>
89 #include <sys/mbuf.h>
90 #include <sys/ioctl.h>
91 #include <sys/tty.h>
92 #include <sys/signalvar.h>
93
94 /*
95 * Structure associated with user cacheing.
96 */
97 struct uidinfo {
98 LIST_ENTRY(uidinfo) ui_hash;
99 uid_t ui_uid;
100 long ui_proccnt;
101 };
102 #define UIHASH(uid) (&uihashtbl[(uid) & uihash])
103 LIST_HEAD(uihashhead, uidinfo) *uihashtbl;
104 u_long uihash; /* size of hash table - 1 */
105
106 /*
107 * Other process lists
108 */
109 struct pidhashhead *pidhashtbl;
110 u_long pidhash;
111 struct pgrphashhead *pgrphashtbl;
112 u_long pgrphash;
113
114 struct proclist allproc;
115 struct proclist zombproc; /* resources have been freed */
116
117 /*
118 * Process list locking:
119 *
120 * We have two types of locks on the proclists: read locks and write
121 * locks. Read locks can be used in interrupt context, so while we
122 * hold the write lock, we must also block clock interrupts to
123 * lock out any scheduling changes that may happen in interrupt
124 * context.
125 *
126 * The proclist lock locks the following structures:
127 *
128 * allproc
129 * zombproc
130 * pidhashtbl
131 */
132 struct lock proclist_lock;
133
134 /*
135 * Locking of this proclist is special; it's accessed in a
136 * critical section of process exit, and thus locking it can't
137 * modify interrupt state. We use a simple spin lock for this
138 * proclist. Processes on this proclist are also on zombproc;
139 * we use the p_hash member to linkup to deadproc.
140 */
141 struct simplelock deadproc_slock;
142 struct proclist deadproc; /* dead, but not yet undead */
143
144 struct pool proc_pool;
145 struct pool pcred_pool;
146 struct pool plimit_pool;
147 struct pool pgrp_pool;
148 struct pool rusage_pool;
149
150 /*
151 * The process list descriptors, used during pid allocation and
152 * by sysctl. No locking on this data structure is needed since
153 * it is completely static.
154 */
155 const struct proclist_desc proclists[] = {
156 { &allproc },
157 { &zombproc },
158 { NULL },
159 };
160
161 static void orphanpg __P((struct pgrp *));
162 #ifdef DEBUG
163 void pgrpdump __P((void));
164 #endif
165
166 /*
167 * Initialize global process hashing structures.
168 */
169 void
170 procinit()
171 {
172 const struct proclist_desc *pd;
173
174 for (pd = proclists; pd->pd_list != NULL; pd++)
175 LIST_INIT(pd->pd_list);
176
177 spinlockinit(&proclist_lock, "proclk", 0);
178
179 LIST_INIT(&deadproc);
180 simple_lock_init(&deadproc_slock);
181
182 pidhashtbl = hashinit(maxproc / 4, M_PROC, M_WAITOK, &pidhash);
183 pgrphashtbl = hashinit(maxproc / 4, M_PROC, M_WAITOK, &pgrphash);
184 uihashtbl = hashinit(maxproc / 16, M_PROC, M_WAITOK, &uihash);
185
186 pool_init(&proc_pool, sizeof(struct proc), 0, 0, 0, "procpl",
187 0, pool_page_alloc_nointr, pool_page_free_nointr, M_PROC);
188 pool_init(&pgrp_pool, sizeof(struct pgrp), 0, 0, 0, "pgrppl",
189 0, pool_page_alloc_nointr, pool_page_free_nointr, M_PGRP);
190 pool_init(&pcred_pool, sizeof(struct pcred), 0, 0, 0, "pcredpl",
191 0, pool_page_alloc_nointr, pool_page_free_nointr, M_SUBPROC);
192 pool_init(&plimit_pool, sizeof(struct plimit), 0, 0, 0, "plimitpl",
193 0, pool_page_alloc_nointr, pool_page_free_nointr, M_SUBPROC);
194 pool_init(&rusage_pool, sizeof(struct rusage), 0, 0, 0, "rusgepl",
195 0, pool_page_alloc_nointr, pool_page_free_nointr, M_ZOMBIE);
196 }
197
198 /*
199 * Acquire a read lock on the proclist.
200 */
201 void
202 proclist_lock_read()
203 {
204 int error;
205
206 error = spinlockmgr(&proclist_lock, LK_SHARED, NULL);
207 #ifdef DIAGNOSTIC
208 if (__predict_false(error != 0))
209 panic("proclist_lock_read: failed to acquire lock");
210 #endif
211 }
212
213 /*
214 * Release a read lock on the proclist.
215 */
216 void
217 proclist_unlock_read()
218 {
219
220 (void) spinlockmgr(&proclist_lock, LK_RELEASE, NULL);
221 }
222
223 /*
224 * Acquire a write lock on the proclist.
225 */
226 int
227 proclist_lock_write()
228 {
229 int s, error;
230
231 s = splclock();
232 error = spinlockmgr(&proclist_lock, LK_EXCLUSIVE, NULL);
233 #ifdef DIAGNOSTIC
234 if (__predict_false(error != 0))
235 panic("proclist_lock: failed to acquire lock");
236 #endif
237 return (s);
238 }
239
240 /*
241 * Release a write lock on the proclist.
242 */
243 void
244 proclist_unlock_write(s)
245 int s;
246 {
247
248 (void) spinlockmgr(&proclist_lock, LK_RELEASE, NULL);
249 splx(s);
250 }
251
252 /*
253 * Change the count associated with number of processes
254 * a given user is using.
255 */
256 int
257 chgproccnt(uid, diff)
258 uid_t uid;
259 int diff;
260 {
261 struct uidinfo *uip;
262 struct uihashhead *uipp;
263
264 uipp = UIHASH(uid);
265 for (uip = uipp->lh_first; uip != 0; uip = uip->ui_hash.le_next)
266 if (uip->ui_uid == uid)
267 break;
268 if (uip) {
269 uip->ui_proccnt += diff;
270 if (uip->ui_proccnt > 0)
271 return (uip->ui_proccnt);
272 if (uip->ui_proccnt < 0)
273 panic("chgproccnt: procs < 0");
274 LIST_REMOVE(uip, ui_hash);
275 FREE(uip, M_PROC);
276 return (0);
277 }
278 if (diff <= 0) {
279 if (diff == 0)
280 return(0);
281 panic("chgproccnt: lost user");
282 }
283 MALLOC(uip, struct uidinfo *, sizeof(*uip), M_PROC, M_WAITOK);
284 LIST_INSERT_HEAD(uipp, uip, ui_hash);
285 uip->ui_uid = uid;
286 uip->ui_proccnt = diff;
287 return (diff);
288 }
289
290 /*
291 * Is p an inferior of q?
292 */
293 int
294 inferior(p, q)
295 struct proc *p;
296 struct proc *q;
297 {
298
299 for (; p != q; p = p->p_pptr)
300 if (p->p_pid == 0)
301 return (0);
302 return (1);
303 }
304
305 /*
306 * Locate a process by number
307 */
308 struct proc *
309 pfind(pid)
310 pid_t pid;
311 {
312 struct proc *p;
313
314 proclist_lock_read();
315 for (p = PIDHASH(pid)->lh_first; p != 0; p = p->p_hash.le_next)
316 if (p->p_pid == pid)
317 goto out;
318 out:
319 proclist_unlock_read();
320 return (p);
321 }
322
323 /*
324 * Locate a process group by number
325 */
326 struct pgrp *
327 pgfind(pgid)
328 pid_t pgid;
329 {
330 struct pgrp *pgrp;
331
332 for (pgrp = PGRPHASH(pgid)->lh_first; pgrp != 0; pgrp = pgrp->pg_hash.le_next)
333 if (pgrp->pg_id == pgid)
334 return (pgrp);
335 return (NULL);
336 }
337
338 /*
339 * Move p to a new or existing process group (and session)
340 */
341 int
342 enterpgrp(p, pgid, mksess)
343 struct proc *p;
344 pid_t pgid;
345 int mksess;
346 {
347 struct pgrp *pgrp = pgfind(pgid);
348
349 #ifdef DIAGNOSTIC
350 if (__predict_false(pgrp != NULL && mksess)) /* firewalls */
351 panic("enterpgrp: setsid into non-empty pgrp");
352 if (__predict_false(SESS_LEADER(p)))
353 panic("enterpgrp: session leader attempted setpgrp");
354 #endif
355 if (pgrp == NULL) {
356 pid_t savepid = p->p_pid;
357 struct proc *np;
358 /*
359 * new process group
360 */
361 #ifdef DIAGNOSTIC
362 if (__predict_false(p->p_pid != pgid))
363 panic("enterpgrp: new pgrp and pid != pgid");
364 #endif
365 pgrp = pool_get(&pgrp_pool, PR_WAITOK);
366 if ((np = pfind(savepid)) == NULL || np != p)
367 return (ESRCH);
368 if (mksess) {
369 struct session *sess;
370
371 /*
372 * new session
373 */
374 MALLOC(sess, struct session *, sizeof(struct session),
375 M_SESSION, M_WAITOK);
376 sess->s_sid = p->p_pid;
377 sess->s_leader = p;
378 sess->s_count = 1;
379 sess->s_ttyvp = NULL;
380 sess->s_ttyp = NULL;
381 memcpy(sess->s_login, p->p_session->s_login,
382 sizeof(sess->s_login));
383 p->p_flag &= ~P_CONTROLT;
384 pgrp->pg_session = sess;
385 #ifdef DIAGNOSTIC
386 if (__predict_false(p != curproc))
387 panic("enterpgrp: mksession and p != curproc");
388 #endif
389 } else {
390 pgrp->pg_session = p->p_session;
391 pgrp->pg_session->s_count++;
392 }
393 pgrp->pg_id = pgid;
394 LIST_INIT(&pgrp->pg_members);
395 LIST_INSERT_HEAD(PGRPHASH(pgid), pgrp, pg_hash);
396 pgrp->pg_jobc = 0;
397 } else if (pgrp == p->p_pgrp)
398 return (0);
399
400 /*
401 * Adjust eligibility of affected pgrps to participate in job control.
402 * Increment eligibility counts before decrementing, otherwise we
403 * could reach 0 spuriously during the first call.
404 */
405 fixjobc(p, pgrp, 1);
406 fixjobc(p, p->p_pgrp, 0);
407
408 LIST_REMOVE(p, p_pglist);
409 if (p->p_pgrp->pg_members.lh_first == 0)
410 pgdelete(p->p_pgrp);
411 p->p_pgrp = pgrp;
412 LIST_INSERT_HEAD(&pgrp->pg_members, p, p_pglist);
413 return (0);
414 }
415
416 /*
417 * remove process from process group
418 */
419 int
420 leavepgrp(p)
421 struct proc *p;
422 {
423
424 LIST_REMOVE(p, p_pglist);
425 if (p->p_pgrp->pg_members.lh_first == 0)
426 pgdelete(p->p_pgrp);
427 p->p_pgrp = 0;
428 return (0);
429 }
430
431 /*
432 * delete a process group
433 */
434 void
435 pgdelete(pgrp)
436 struct pgrp *pgrp;
437 {
438
439 if (pgrp->pg_session->s_ttyp != NULL &&
440 pgrp->pg_session->s_ttyp->t_pgrp == pgrp)
441 pgrp->pg_session->s_ttyp->t_pgrp = NULL;
442 LIST_REMOVE(pgrp, pg_hash);
443 if (--pgrp->pg_session->s_count == 0)
444 FREE(pgrp->pg_session, M_SESSION);
445 pool_put(&pgrp_pool, pgrp);
446 }
447
448 /*
449 * Adjust pgrp jobc counters when specified process changes process group.
450 * We count the number of processes in each process group that "qualify"
451 * the group for terminal job control (those with a parent in a different
452 * process group of the same session). If that count reaches zero, the
453 * process group becomes orphaned. Check both the specified process'
454 * process group and that of its children.
455 * entering == 0 => p is leaving specified group.
456 * entering == 1 => p is entering specified group.
457 */
458 void
459 fixjobc(p, pgrp, entering)
460 struct proc *p;
461 struct pgrp *pgrp;
462 int entering;
463 {
464 struct pgrp *hispgrp;
465 struct session *mysession = pgrp->pg_session;
466
467 /*
468 * Check p's parent to see whether p qualifies its own process
469 * group; if so, adjust count for p's process group.
470 */
471 if ((hispgrp = p->p_pptr->p_pgrp) != pgrp &&
472 hispgrp->pg_session == mysession) {
473 if (entering)
474 pgrp->pg_jobc++;
475 else if (--pgrp->pg_jobc == 0)
476 orphanpg(pgrp);
477 }
478
479 /*
480 * Check this process' children to see whether they qualify
481 * their process groups; if so, adjust counts for children's
482 * process groups.
483 */
484 for (p = p->p_children.lh_first; p != 0; p = p->p_sibling.le_next) {
485 if ((hispgrp = p->p_pgrp) != pgrp &&
486 hispgrp->pg_session == mysession &&
487 P_ZOMBIE(p) == 0) {
488 if (entering)
489 hispgrp->pg_jobc++;
490 else if (--hispgrp->pg_jobc == 0)
491 orphanpg(hispgrp);
492 }
493 }
494 }
495
496 /*
497 * A process group has become orphaned;
498 * if there are any stopped processes in the group,
499 * hang-up all process in that group.
500 */
501 static void
502 orphanpg(pg)
503 struct pgrp *pg;
504 {
505 struct proc *p;
506
507 for (p = pg->pg_members.lh_first; p != 0; p = p->p_pglist.le_next) {
508 if (p->p_stat == SSTOP) {
509 for (p = pg->pg_members.lh_first; p != 0;
510 p = p->p_pglist.le_next) {
511 psignal(p, SIGHUP);
512 psignal(p, SIGCONT);
513 }
514 return;
515 }
516 }
517 }
518
519 /* mark process as suid/sgid, reset some values do defaults */
520 void
521 p_sugid(p)
522 struct proc *p;
523 {
524 struct plimit *newlim;
525
526 p->p_flag |= P_SUGID;
527 /* reset what needs to be reset in plimit */
528 if (p->p_limit->pl_corename != defcorename) {
529 if (p->p_limit->p_refcnt > 1 &&
530 (p->p_limit->p_lflags & PL_SHAREMOD) == 0) {
531 newlim = limcopy(p->p_limit);
532 limfree(p->p_limit);
533 p->p_limit = newlim;
534 } else {
535 free(p->p_limit->pl_corename, M_TEMP);
536 }
537 p->p_limit->pl_corename = defcorename;
538 }
539 }
540
541
542 #ifdef DEBUG
543 void
544 pgrpdump()
545 {
546 struct pgrp *pgrp;
547 struct proc *p;
548 int i;
549
550 for (i = 0; i <= pgrphash; i++) {
551 if ((pgrp = pgrphashtbl[i].lh_first) != NULL) {
552 printf("\tindx %d\n", i);
553 for (; pgrp != 0; pgrp = pgrp->pg_hash.le_next) {
554 printf("\tpgrp %p, pgid %d, sess %p, sesscnt %d, mem %p\n",
555 pgrp, pgrp->pg_id, pgrp->pg_session,
556 pgrp->pg_session->s_count,
557 pgrp->pg_members.lh_first);
558 for (p = pgrp->pg_members.lh_first; p != 0;
559 p = p->p_pglist.le_next) {
560 printf("\t\tpid %d addr %p pgrp %p\n",
561 p->p_pid, p, p->p_pgrp);
562 }
563 }
564 }
565 }
566 }
567 #endif /* DEBUG */
568