kern_proc.c revision 1.180.8.3
1 /* $NetBSD: kern_proc.c,v 1.180.8.3 2012/04/29 23:05:04 mrg Exp $ */ 2 3 /*- 4 * Copyright (c) 1999, 2006, 2007, 2008 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, and by Andrew Doran. 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 * 20 * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS 21 * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED 22 * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR 23 * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS 24 * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR 25 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF 26 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS 27 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN 28 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) 29 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE 30 * POSSIBILITY OF SUCH DAMAGE. 31 */ 32 33 /* 34 * Copyright (c) 1982, 1986, 1989, 1991, 1993 35 * The Regents of the University of California. All rights reserved. 36 * 37 * Redistribution and use in source and binary forms, with or without 38 * modification, are permitted provided that the following conditions 39 * are met: 40 * 1. Redistributions of source code must retain the above copyright 41 * notice, this list of conditions and the following disclaimer. 42 * 2. Redistributions in binary form must reproduce the above copyright 43 * notice, this list of conditions and the following disclaimer in the 44 * documentation and/or other materials provided with the distribution. 45 * 3. Neither the name of the University nor the names of its contributors 46 * may be used to endorse or promote products derived from this software 47 * without specific prior written permission. 48 * 49 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 50 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 51 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 52 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 53 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 54 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 55 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 56 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 57 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 58 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 59 * SUCH DAMAGE. 60 * 61 * @(#)kern_proc.c 8.7 (Berkeley) 2/14/95 62 */ 63 64 #include <sys/cdefs.h> 65 __KERNEL_RCSID(0, "$NetBSD: kern_proc.c,v 1.180.8.3 2012/04/29 23:05:04 mrg Exp $"); 66 67 #ifdef _KERNEL_OPT 68 #include "opt_kstack.h" 69 #include "opt_maxuprc.h" 70 #include "opt_dtrace.h" 71 #include "opt_compat_netbsd32.h" 72 #endif 73 74 #include <sys/param.h> 75 #include <sys/systm.h> 76 #include <sys/kernel.h> 77 #include <sys/proc.h> 78 #include <sys/resourcevar.h> 79 #include <sys/buf.h> 80 #include <sys/acct.h> 81 #include <sys/wait.h> 82 #include <sys/file.h> 83 #include <ufs/ufs/quota.h> 84 #include <sys/uio.h> 85 #include <sys/pool.h> 86 #include <sys/pset.h> 87 #include <sys/mbuf.h> 88 #include <sys/ioctl.h> 89 #include <sys/tty.h> 90 #include <sys/signalvar.h> 91 #include <sys/ras.h> 92 #include <sys/filedesc.h> 93 #include "sys/syscall_stats.h" 94 #include <sys/kauth.h> 95 #include <sys/sleepq.h> 96 #include <sys/atomic.h> 97 #include <sys/kmem.h> 98 #include <sys/dtrace_bsd.h> 99 #include <sys/sysctl.h> 100 #include <sys/exec.h> 101 #include <sys/cpu.h> 102 103 #include <uvm/uvm_extern.h> 104 #include <uvm/uvm_extern.h> 105 106 #ifdef COMPAT_NETBSD32 107 #include <compat/netbsd32/netbsd32.h> 108 #endif 109 110 /* 111 * Process lists. 112 */ 113 114 struct proclist allproc __cacheline_aligned; 115 struct proclist zombproc __cacheline_aligned; 116 117 kmutex_t * proc_lock __cacheline_aligned; 118 119 /* 120 * pid to proc lookup is done by indexing the pid_table array. 121 * Since pid numbers are only allocated when an empty slot 122 * has been found, there is no need to search any lists ever. 123 * (an orphaned pgrp will lock the slot, a session will lock 124 * the pgrp with the same number.) 125 * If the table is too small it is reallocated with twice the 126 * previous size and the entries 'unzipped' into the two halves. 127 * A linked list of free entries is passed through the pt_proc 128 * field of 'free' items - set odd to be an invalid ptr. 129 */ 130 131 struct pid_table { 132 struct proc *pt_proc; 133 struct pgrp *pt_pgrp; 134 pid_t pt_pid; 135 }; 136 #if 1 /* strongly typed cast - should be a noop */ 137 static inline uint p2u(struct proc *p) { return (uint)(uintptr_t)p; } 138 #else 139 #define p2u(p) ((uint)p) 140 #endif 141 #define P_VALID(p) (!(p2u(p) & 1)) 142 #define P_NEXT(p) (p2u(p) >> 1) 143 #define P_FREE(pid) ((struct proc *)(uintptr_t)((pid) << 1 | 1)) 144 145 /* 146 * Table of process IDs (PIDs). 147 */ 148 static struct pid_table *pid_table __read_mostly; 149 150 #define INITIAL_PID_TABLE_SIZE (1 << 5) 151 152 /* Table mask, threshold for growing and number of allocated PIDs. */ 153 static u_int pid_tbl_mask __read_mostly; 154 static u_int pid_alloc_lim __read_mostly; 155 static u_int pid_alloc_cnt __cacheline_aligned; 156 157 /* Next free, last free and maximum PIDs. */ 158 static u_int next_free_pt __cacheline_aligned; 159 static u_int last_free_pt __cacheline_aligned; 160 static pid_t pid_max __read_mostly; 161 162 /* Components of the first process -- never freed. */ 163 164 extern struct emul emul_netbsd; /* defined in kern_exec.c */ 165 166 struct session session0 = { 167 .s_count = 1, 168 .s_sid = 0, 169 }; 170 struct pgrp pgrp0 = { 171 .pg_members = LIST_HEAD_INITIALIZER(&pgrp0.pg_members), 172 .pg_session = &session0, 173 }; 174 filedesc_t filedesc0; 175 struct cwdinfo cwdi0 = { 176 .cwdi_cmask = CMASK, /* see cmask below */ 177 .cwdi_refcnt = 1, 178 }; 179 struct plimit limit0; 180 struct pstats pstat0; 181 struct vmspace vmspace0; 182 struct sigacts sigacts0; 183 struct proc proc0 = { 184 .p_lwps = LIST_HEAD_INITIALIZER(&proc0.p_lwps), 185 .p_sigwaiters = LIST_HEAD_INITIALIZER(&proc0.p_sigwaiters), 186 .p_nlwps = 1, 187 .p_nrlwps = 1, 188 .p_nlwpid = 1, /* must match lwp0.l_lid */ 189 .p_pgrp = &pgrp0, 190 .p_comm = "system", 191 /* 192 * Set P_NOCLDWAIT so that kernel threads are reparented to init(8) 193 * when they exit. init(8) can easily wait them out for us. 194 */ 195 .p_flag = PK_SYSTEM | PK_NOCLDWAIT, 196 .p_stat = SACTIVE, 197 .p_nice = NZERO, 198 .p_emul = &emul_netbsd, 199 .p_cwdi = &cwdi0, 200 .p_limit = &limit0, 201 .p_fd = &filedesc0, 202 .p_vmspace = &vmspace0, 203 .p_stats = &pstat0, 204 .p_sigacts = &sigacts0, 205 }; 206 kauth_cred_t cred0; 207 208 static const int nofile = NOFILE; 209 static const int maxuprc = MAXUPRC; 210 static const int cmask = CMASK; 211 212 static int sysctl_doeproc(SYSCTLFN_PROTO); 213 static int sysctl_kern_proc_args(SYSCTLFN_PROTO); 214 static void fill_kproc2(struct proc *, struct kinfo_proc2 *, bool); 215 216 /* 217 * The process list descriptors, used during pid allocation and 218 * by sysctl. No locking on this data structure is needed since 219 * it is completely static. 220 */ 221 const struct proclist_desc proclists[] = { 222 { &allproc }, 223 { &zombproc }, 224 { NULL }, 225 }; 226 227 static struct pgrp * pg_remove(pid_t); 228 static void pg_delete(pid_t); 229 static void orphanpg(struct pgrp *); 230 231 static specificdata_domain_t proc_specificdata_domain; 232 233 static pool_cache_t proc_cache; 234 235 static kauth_listener_t proc_listener; 236 237 static int 238 proc_listener_cb(kauth_cred_t cred, kauth_action_t action, void *cookie, 239 void *arg0, void *arg1, void *arg2, void *arg3) 240 { 241 struct proc *p; 242 int result; 243 244 result = KAUTH_RESULT_DEFER; 245 p = arg0; 246 247 switch (action) { 248 case KAUTH_PROCESS_CANSEE: { 249 enum kauth_process_req req; 250 251 req = (enum kauth_process_req)arg1; 252 253 switch (req) { 254 case KAUTH_REQ_PROCESS_CANSEE_ARGS: 255 case KAUTH_REQ_PROCESS_CANSEE_ENTRY: 256 case KAUTH_REQ_PROCESS_CANSEE_OPENFILES: 257 result = KAUTH_RESULT_ALLOW; 258 259 break; 260 261 case KAUTH_REQ_PROCESS_CANSEE_ENV: 262 if (kauth_cred_getuid(cred) != 263 kauth_cred_getuid(p->p_cred) || 264 kauth_cred_getuid(cred) != 265 kauth_cred_getsvuid(p->p_cred)) 266 break; 267 268 result = KAUTH_RESULT_ALLOW; 269 270 break; 271 272 default: 273 break; 274 } 275 276 break; 277 } 278 279 case KAUTH_PROCESS_FORK: { 280 int lnprocs = (int)(unsigned long)arg2; 281 282 /* 283 * Don't allow a nonprivileged user to use the last few 284 * processes. The variable lnprocs is the current number of 285 * processes, maxproc is the limit. 286 */ 287 if (__predict_false((lnprocs >= maxproc - 5))) 288 break; 289 290 result = KAUTH_RESULT_ALLOW; 291 292 break; 293 } 294 295 case KAUTH_PROCESS_CORENAME: 296 case KAUTH_PROCESS_STOPFLAG: 297 if (proc_uidmatch(cred, p->p_cred) == 0) 298 result = KAUTH_RESULT_ALLOW; 299 300 break; 301 302 default: 303 break; 304 } 305 306 return result; 307 } 308 309 /* 310 * Initialize global process hashing structures. 311 */ 312 void 313 procinit(void) 314 { 315 const struct proclist_desc *pd; 316 u_int i; 317 #define LINK_EMPTY ((PID_MAX + INITIAL_PID_TABLE_SIZE) & ~(INITIAL_PID_TABLE_SIZE - 1)) 318 319 for (pd = proclists; pd->pd_list != NULL; pd++) 320 LIST_INIT(pd->pd_list); 321 322 proc_lock = mutex_obj_alloc(MUTEX_DEFAULT, IPL_NONE); 323 pid_table = kmem_alloc(INITIAL_PID_TABLE_SIZE 324 * sizeof(struct pid_table), KM_SLEEP); 325 pid_tbl_mask = INITIAL_PID_TABLE_SIZE - 1; 326 pid_max = PID_MAX; 327 328 /* Set free list running through table... 329 Preset 'use count' above PID_MAX so we allocate pid 1 next. */ 330 for (i = 0; i <= pid_tbl_mask; i++) { 331 pid_table[i].pt_proc = P_FREE(LINK_EMPTY + i + 1); 332 pid_table[i].pt_pgrp = 0; 333 pid_table[i].pt_pid = 0; 334 } 335 /* slot 0 is just grabbed */ 336 next_free_pt = 1; 337 /* Need to fix last entry. */ 338 last_free_pt = pid_tbl_mask; 339 pid_table[last_free_pt].pt_proc = P_FREE(LINK_EMPTY); 340 /* point at which we grow table - to avoid reusing pids too often */ 341 pid_alloc_lim = pid_tbl_mask - 1; 342 #undef LINK_EMPTY 343 344 proc_specificdata_domain = specificdata_domain_create(); 345 KASSERT(proc_specificdata_domain != NULL); 346 347 proc_cache = pool_cache_init(sizeof(struct proc), 0, 0, 0, 348 "procpl", NULL, IPL_NONE, NULL, NULL, NULL); 349 350 proc_listener = kauth_listen_scope(KAUTH_SCOPE_PROCESS, 351 proc_listener_cb, NULL); 352 } 353 354 void 355 procinit_sysctl(void) 356 { 357 static struct sysctllog *clog; 358 359 sysctl_createv(&clog, 0, NULL, NULL, 360 CTLFLAG_PERMANENT, 361 CTLTYPE_NODE, "kern", NULL, 362 NULL, 0, NULL, 0, 363 CTL_KERN, CTL_EOL); 364 365 sysctl_createv(&clog, 0, NULL, NULL, 366 CTLFLAG_PERMANENT, 367 CTLTYPE_NODE, "proc", 368 SYSCTL_DESCR("System-wide process information"), 369 sysctl_doeproc, 0, NULL, 0, 370 CTL_KERN, KERN_PROC, CTL_EOL); 371 sysctl_createv(&clog, 0, NULL, NULL, 372 CTLFLAG_PERMANENT, 373 CTLTYPE_NODE, "proc2", 374 SYSCTL_DESCR("Machine-independent process information"), 375 sysctl_doeproc, 0, NULL, 0, 376 CTL_KERN, KERN_PROC2, CTL_EOL); 377 sysctl_createv(&clog, 0, NULL, NULL, 378 CTLFLAG_PERMANENT, 379 CTLTYPE_NODE, "proc_args", 380 SYSCTL_DESCR("Process argument information"), 381 sysctl_kern_proc_args, 0, NULL, 0, 382 CTL_KERN, KERN_PROC_ARGS, CTL_EOL); 383 384 /* 385 "nodes" under these: 386 387 KERN_PROC_ALL 388 KERN_PROC_PID pid 389 KERN_PROC_PGRP pgrp 390 KERN_PROC_SESSION sess 391 KERN_PROC_TTY tty 392 KERN_PROC_UID uid 393 KERN_PROC_RUID uid 394 KERN_PROC_GID gid 395 KERN_PROC_RGID gid 396 397 all in all, probably not worth the effort... 398 */ 399 } 400 401 /* 402 * Initialize process 0. 403 */ 404 void 405 proc0_init(void) 406 { 407 struct proc *p; 408 struct pgrp *pg; 409 struct rlimit *rlim; 410 rlim_t lim; 411 int i; 412 413 p = &proc0; 414 pg = &pgrp0; 415 416 mutex_init(&p->p_stmutex, MUTEX_DEFAULT, IPL_HIGH); 417 mutex_init(&p->p_auxlock, MUTEX_DEFAULT, IPL_NONE); 418 p->p_lock = mutex_obj_alloc(MUTEX_DEFAULT, IPL_NONE); 419 420 rw_init(&p->p_reflock); 421 cv_init(&p->p_waitcv, "wait"); 422 cv_init(&p->p_lwpcv, "lwpwait"); 423 424 LIST_INSERT_HEAD(&p->p_lwps, &lwp0, l_sibling); 425 426 pid_table[0].pt_proc = p; 427 LIST_INSERT_HEAD(&allproc, p, p_list); 428 429 pid_table[0].pt_pgrp = pg; 430 LIST_INSERT_HEAD(&pg->pg_members, p, p_pglist); 431 432 #ifdef __HAVE_SYSCALL_INTERN 433 (*p->p_emul->e_syscall_intern)(p); 434 #endif 435 436 /* Create credentials. */ 437 cred0 = kauth_cred_alloc(); 438 p->p_cred = cred0; 439 440 /* Create the CWD info. */ 441 rw_init(&cwdi0.cwdi_lock); 442 443 /* Create the limits structures. */ 444 mutex_init(&limit0.pl_lock, MUTEX_DEFAULT, IPL_NONE); 445 446 rlim = limit0.pl_rlimit; 447 for (i = 0; i < __arraycount(limit0.pl_rlimit); i++) { 448 rlim[i].rlim_cur = RLIM_INFINITY; 449 rlim[i].rlim_max = RLIM_INFINITY; 450 } 451 452 rlim[RLIMIT_NOFILE].rlim_max = maxfiles; 453 rlim[RLIMIT_NOFILE].rlim_cur = maxfiles < nofile ? maxfiles : nofile; 454 455 rlim[RLIMIT_NPROC].rlim_max = maxproc; 456 rlim[RLIMIT_NPROC].rlim_cur = maxproc < maxuprc ? maxproc : maxuprc; 457 458 lim = MIN(VM_MAXUSER_ADDRESS, ctob((rlim_t)uvmexp.free)); 459 rlim[RLIMIT_RSS].rlim_max = lim; 460 rlim[RLIMIT_MEMLOCK].rlim_max = lim; 461 rlim[RLIMIT_MEMLOCK].rlim_cur = lim / 3; 462 463 /* Note that default core name has zero length. */ 464 limit0.pl_corename = defcorename; 465 limit0.pl_cnlen = 0; 466 limit0.pl_refcnt = 1; 467 limit0.pl_writeable = false; 468 limit0.pl_sv_limit = NULL; 469 470 /* Configure virtual memory system, set vm rlimits. */ 471 uvm_init_limits(p); 472 473 /* Initialize file descriptor table for proc0. */ 474 fd_init(&filedesc0); 475 476 /* 477 * Initialize proc0's vmspace, which uses the kernel pmap. 478 * All kernel processes (which never have user space mappings) 479 * share proc0's vmspace, and thus, the kernel pmap. 480 */ 481 uvmspace_init(&vmspace0, pmap_kernel(), round_page(VM_MIN_ADDRESS), 482 trunc_page(VM_MAX_ADDRESS)); 483 484 /* Initialize signal state for proc0. XXX IPL_SCHED */ 485 mutex_init(&p->p_sigacts->sa_mutex, MUTEX_DEFAULT, IPL_SCHED); 486 siginit(p); 487 488 proc_initspecific(p); 489 kdtrace_proc_ctor(NULL, p); 490 } 491 492 /* 493 * Session reference counting. 494 */ 495 496 void 497 proc_sesshold(struct session *ss) 498 { 499 500 KASSERT(mutex_owned(proc_lock)); 501 ss->s_count++; 502 } 503 504 void 505 proc_sessrele(struct session *ss) 506 { 507 508 KASSERT(mutex_owned(proc_lock)); 509 /* 510 * We keep the pgrp with the same id as the session in order to 511 * stop a process being given the same pid. Since the pgrp holds 512 * a reference to the session, it must be a 'zombie' pgrp by now. 513 */ 514 if (--ss->s_count == 0) { 515 struct pgrp *pg; 516 517 pg = pg_remove(ss->s_sid); 518 mutex_exit(proc_lock); 519 520 kmem_free(pg, sizeof(struct pgrp)); 521 kmem_free(ss, sizeof(struct session)); 522 } else { 523 mutex_exit(proc_lock); 524 } 525 } 526 527 /* 528 * Check that the specified process group is in the session of the 529 * specified process. 530 * Treats -ve ids as process ids. 531 * Used to validate TIOCSPGRP requests. 532 */ 533 int 534 pgid_in_session(struct proc *p, pid_t pg_id) 535 { 536 struct pgrp *pgrp; 537 struct session *session; 538 int error; 539 540 mutex_enter(proc_lock); 541 if (pg_id < 0) { 542 struct proc *p1 = proc_find(-pg_id); 543 if (p1 == NULL) { 544 error = EINVAL; 545 goto fail; 546 } 547 pgrp = p1->p_pgrp; 548 } else { 549 pgrp = pgrp_find(pg_id); 550 if (pgrp == NULL) { 551 error = EINVAL; 552 goto fail; 553 } 554 } 555 session = pgrp->pg_session; 556 error = (session != p->p_pgrp->pg_session) ? EPERM : 0; 557 fail: 558 mutex_exit(proc_lock); 559 return error; 560 } 561 562 /* 563 * p_inferior: is p an inferior of q? 564 */ 565 static inline bool 566 p_inferior(struct proc *p, struct proc *q) 567 { 568 569 KASSERT(mutex_owned(proc_lock)); 570 571 for (; p != q; p = p->p_pptr) 572 if (p->p_pid == 0) 573 return false; 574 return true; 575 } 576 577 /* 578 * proc_find: locate a process by the ID. 579 * 580 * => Must be called with proc_lock held. 581 */ 582 proc_t * 583 proc_find_raw(pid_t pid) 584 { 585 struct pid_table *pt; 586 proc_t *p; 587 588 KASSERT(mutex_owned(proc_lock)); 589 pt = &pid_table[pid & pid_tbl_mask]; 590 p = pt->pt_proc; 591 if (__predict_false(!P_VALID(p) || pt->pt_pid != pid)) { 592 return NULL; 593 } 594 return p; 595 } 596 597 proc_t * 598 proc_find(pid_t pid) 599 { 600 proc_t *p; 601 602 p = proc_find_raw(pid); 603 if (__predict_false(p == NULL)) { 604 return NULL; 605 } 606 607 /* 608 * Only allow live processes to be found by PID. 609 * XXX: p_stat might change, since unlocked. 610 */ 611 if (__predict_true(p->p_stat == SACTIVE || p->p_stat == SSTOP)) { 612 return p; 613 } 614 return NULL; 615 } 616 617 /* 618 * pgrp_find: locate a process group by the ID. 619 * 620 * => Must be called with proc_lock held. 621 */ 622 struct pgrp * 623 pgrp_find(pid_t pgid) 624 { 625 struct pgrp *pg; 626 627 KASSERT(mutex_owned(proc_lock)); 628 629 pg = pid_table[pgid & pid_tbl_mask].pt_pgrp; 630 631 /* 632 * Cannot look up a process group that only exists because the 633 * session has not died yet (traditional). 634 */ 635 if (pg == NULL || pg->pg_id != pgid || LIST_EMPTY(&pg->pg_members)) { 636 return NULL; 637 } 638 return pg; 639 } 640 641 static void 642 expand_pid_table(void) 643 { 644 size_t pt_size, tsz; 645 struct pid_table *n_pt, *new_pt; 646 struct proc *proc; 647 struct pgrp *pgrp; 648 pid_t pid, rpid; 649 u_int i; 650 uint new_pt_mask; 651 652 pt_size = pid_tbl_mask + 1; 653 tsz = pt_size * 2 * sizeof(struct pid_table); 654 new_pt = kmem_alloc(tsz, KM_SLEEP); 655 new_pt_mask = pt_size * 2 - 1; 656 657 mutex_enter(proc_lock); 658 if (pt_size != pid_tbl_mask + 1) { 659 /* Another process beat us to it... */ 660 mutex_exit(proc_lock); 661 kmem_free(new_pt, tsz); 662 return; 663 } 664 665 /* 666 * Copy entries from old table into new one. 667 * If 'pid' is 'odd' we need to place in the upper half, 668 * even pid's to the lower half. 669 * Free items stay in the low half so we don't have to 670 * fixup the reference to them. 671 * We stuff free items on the front of the freelist 672 * because we can't write to unmodified entries. 673 * Processing the table backwards maintains a semblance 674 * of issuing pid numbers that increase with time. 675 */ 676 i = pt_size - 1; 677 n_pt = new_pt + i; 678 for (; ; i--, n_pt--) { 679 proc = pid_table[i].pt_proc; 680 pgrp = pid_table[i].pt_pgrp; 681 if (!P_VALID(proc)) { 682 /* Up 'use count' so that link is valid */ 683 pid = (P_NEXT(proc) + pt_size) & ~pt_size; 684 rpid = 0; 685 proc = P_FREE(pid); 686 if (pgrp) 687 pid = pgrp->pg_id; 688 } else { 689 pid = pid_table[i].pt_pid; 690 rpid = pid; 691 } 692 693 /* Save entry in appropriate half of table */ 694 n_pt[pid & pt_size].pt_proc = proc; 695 n_pt[pid & pt_size].pt_pgrp = pgrp; 696 n_pt[pid & pt_size].pt_pid = rpid; 697 698 /* Put other piece on start of free list */ 699 pid = (pid ^ pt_size) & ~pid_tbl_mask; 700 n_pt[pid & pt_size].pt_proc = 701 P_FREE((pid & ~pt_size) | next_free_pt); 702 n_pt[pid & pt_size].pt_pgrp = 0; 703 n_pt[pid & pt_size].pt_pid = 0; 704 705 next_free_pt = i | (pid & pt_size); 706 if (i == 0) 707 break; 708 } 709 710 /* Save old table size and switch tables */ 711 tsz = pt_size * sizeof(struct pid_table); 712 n_pt = pid_table; 713 pid_table = new_pt; 714 pid_tbl_mask = new_pt_mask; 715 716 /* 717 * pid_max starts as PID_MAX (= 30000), once we have 16384 718 * allocated pids we need it to be larger! 719 */ 720 if (pid_tbl_mask > PID_MAX) { 721 pid_max = pid_tbl_mask * 2 + 1; 722 pid_alloc_lim |= pid_alloc_lim << 1; 723 } else 724 pid_alloc_lim <<= 1; /* doubles number of free slots... */ 725 726 mutex_exit(proc_lock); 727 kmem_free(n_pt, tsz); 728 } 729 730 struct proc * 731 proc_alloc(void) 732 { 733 struct proc *p; 734 735 p = pool_cache_get(proc_cache, PR_WAITOK); 736 p->p_stat = SIDL; /* protect against others */ 737 proc_initspecific(p); 738 kdtrace_proc_ctor(NULL, p); 739 p->p_pid = -1; 740 proc_alloc_pid(p); 741 return p; 742 } 743 744 /* 745 * proc_alloc_pid: allocate PID and record the given proc 'p' so that 746 * proc_find_raw() can find it by the PID. 747 */ 748 749 pid_t 750 proc_alloc_pid(struct proc *p) 751 { 752 struct pid_table *pt; 753 pid_t pid; 754 int nxt; 755 756 for (;;expand_pid_table()) { 757 if (__predict_false(pid_alloc_cnt >= pid_alloc_lim)) 758 /* ensure pids cycle through 2000+ values */ 759 continue; 760 mutex_enter(proc_lock); 761 pt = &pid_table[next_free_pt]; 762 #ifdef DIAGNOSTIC 763 if (__predict_false(P_VALID(pt->pt_proc) || pt->pt_pgrp)) 764 panic("proc_alloc: slot busy"); 765 #endif 766 nxt = P_NEXT(pt->pt_proc); 767 if (nxt & pid_tbl_mask) 768 break; 769 /* Table full - expand (NB last entry not used....) */ 770 mutex_exit(proc_lock); 771 } 772 773 /* pid is 'saved use count' + 'size' + entry */ 774 pid = (nxt & ~pid_tbl_mask) + pid_tbl_mask + 1 + next_free_pt; 775 if ((uint)pid > (uint)pid_max) 776 pid &= pid_tbl_mask; 777 next_free_pt = nxt & pid_tbl_mask; 778 779 /* Grab table slot */ 780 pt->pt_proc = p; 781 782 KASSERT(pt->pt_pid == 0); 783 pt->pt_pid = pid; 784 if (p->p_pid == -1) { 785 p->p_pid = pid; 786 } 787 pid_alloc_cnt++; 788 mutex_exit(proc_lock); 789 790 return pid; 791 } 792 793 /* 794 * Free a process id - called from proc_free (in kern_exit.c) 795 * 796 * Called with the proc_lock held. 797 */ 798 void 799 proc_free_pid(pid_t pid) 800 { 801 struct pid_table *pt; 802 803 KASSERT(mutex_owned(proc_lock)); 804 805 pt = &pid_table[pid & pid_tbl_mask]; 806 807 /* save pid use count in slot */ 808 pt->pt_proc = P_FREE(pid & ~pid_tbl_mask); 809 KASSERT(pt->pt_pid == pid); 810 pt->pt_pid = 0; 811 812 if (pt->pt_pgrp == NULL) { 813 /* link last freed entry onto ours */ 814 pid &= pid_tbl_mask; 815 pt = &pid_table[last_free_pt]; 816 pt->pt_proc = P_FREE(P_NEXT(pt->pt_proc) | pid); 817 pt->pt_pid = 0; 818 last_free_pt = pid; 819 pid_alloc_cnt--; 820 } 821 822 atomic_dec_uint(&nprocs); 823 } 824 825 void 826 proc_free_mem(struct proc *p) 827 { 828 829 kdtrace_proc_dtor(NULL, p); 830 pool_cache_put(proc_cache, p); 831 } 832 833 /* 834 * proc_enterpgrp: move p to a new or existing process group (and session). 835 * 836 * If we are creating a new pgrp, the pgid should equal 837 * the calling process' pid. 838 * If is only valid to enter a process group that is in the session 839 * of the process. 840 * Also mksess should only be set if we are creating a process group 841 * 842 * Only called from sys_setsid, sys_setpgid and posix_spawn/spawn_return. 843 */ 844 int 845 proc_enterpgrp(struct proc *curp, pid_t pid, pid_t pgid, bool mksess) 846 { 847 struct pgrp *new_pgrp, *pgrp; 848 struct session *sess; 849 struct proc *p; 850 int rval; 851 pid_t pg_id = NO_PGID; 852 853 sess = mksess ? kmem_alloc(sizeof(*sess), KM_SLEEP) : NULL; 854 855 /* Allocate data areas we might need before doing any validity checks */ 856 mutex_enter(proc_lock); /* Because pid_table might change */ 857 if (pid_table[pgid & pid_tbl_mask].pt_pgrp == 0) { 858 mutex_exit(proc_lock); 859 new_pgrp = kmem_alloc(sizeof(*new_pgrp), KM_SLEEP); 860 mutex_enter(proc_lock); 861 } else 862 new_pgrp = NULL; 863 rval = EPERM; /* most common error (to save typing) */ 864 865 /* Check pgrp exists or can be created */ 866 pgrp = pid_table[pgid & pid_tbl_mask].pt_pgrp; 867 if (pgrp != NULL && pgrp->pg_id != pgid) 868 goto done; 869 870 /* Can only set another process under restricted circumstances. */ 871 if (pid != curp->p_pid) { 872 /* Must exist and be one of our children... */ 873 p = proc_find(pid); 874 if (p == NULL || !p_inferior(p, curp)) { 875 rval = ESRCH; 876 goto done; 877 } 878 /* ... in the same session... */ 879 if (sess != NULL || p->p_session != curp->p_session) 880 goto done; 881 /* ... existing pgid must be in same session ... */ 882 if (pgrp != NULL && pgrp->pg_session != p->p_session) 883 goto done; 884 /* ... and not done an exec. */ 885 if (p->p_flag & PK_EXEC) { 886 rval = EACCES; 887 goto done; 888 } 889 } else { 890 /* ... setsid() cannot re-enter a pgrp */ 891 if (mksess && (curp->p_pgid == curp->p_pid || 892 pgrp_find(curp->p_pid))) 893 goto done; 894 p = curp; 895 } 896 897 /* Changing the process group/session of a session 898 leader is definitely off limits. */ 899 if (SESS_LEADER(p)) { 900 if (sess == NULL && p->p_pgrp == pgrp) 901 /* unless it's a definite noop */ 902 rval = 0; 903 goto done; 904 } 905 906 /* Can only create a process group with id of process */ 907 if (pgrp == NULL && pgid != pid) 908 goto done; 909 910 /* Can only create a session if creating pgrp */ 911 if (sess != NULL && pgrp != NULL) 912 goto done; 913 914 /* Check we allocated memory for a pgrp... */ 915 if (pgrp == NULL && new_pgrp == NULL) 916 goto done; 917 918 /* Don't attach to 'zombie' pgrp */ 919 if (pgrp != NULL && LIST_EMPTY(&pgrp->pg_members)) 920 goto done; 921 922 /* Expect to succeed now */ 923 rval = 0; 924 925 if (pgrp == p->p_pgrp) 926 /* nothing to do */ 927 goto done; 928 929 /* Ok all setup, link up required structures */ 930 931 if (pgrp == NULL) { 932 pgrp = new_pgrp; 933 new_pgrp = NULL; 934 if (sess != NULL) { 935 sess->s_sid = p->p_pid; 936 sess->s_leader = p; 937 sess->s_count = 1; 938 sess->s_ttyvp = NULL; 939 sess->s_ttyp = NULL; 940 sess->s_flags = p->p_session->s_flags & ~S_LOGIN_SET; 941 memcpy(sess->s_login, p->p_session->s_login, 942 sizeof(sess->s_login)); 943 p->p_lflag &= ~PL_CONTROLT; 944 } else { 945 sess = p->p_pgrp->pg_session; 946 proc_sesshold(sess); 947 } 948 pgrp->pg_session = sess; 949 sess = NULL; 950 951 pgrp->pg_id = pgid; 952 LIST_INIT(&pgrp->pg_members); 953 #ifdef DIAGNOSTIC 954 if (__predict_false(pid_table[pgid & pid_tbl_mask].pt_pgrp)) 955 panic("enterpgrp: pgrp table slot in use"); 956 if (__predict_false(mksess && p != curp)) 957 panic("enterpgrp: mksession and p != curproc"); 958 #endif 959 pid_table[pgid & pid_tbl_mask].pt_pgrp = pgrp; 960 pgrp->pg_jobc = 0; 961 } 962 963 /* 964 * Adjust eligibility of affected pgrps to participate in job control. 965 * Increment eligibility counts before decrementing, otherwise we 966 * could reach 0 spuriously during the first call. 967 */ 968 fixjobc(p, pgrp, 1); 969 fixjobc(p, p->p_pgrp, 0); 970 971 /* Interlock with ttread(). */ 972 mutex_spin_enter(&tty_lock); 973 974 /* Move process to requested group. */ 975 LIST_REMOVE(p, p_pglist); 976 if (LIST_EMPTY(&p->p_pgrp->pg_members)) 977 /* defer delete until we've dumped the lock */ 978 pg_id = p->p_pgrp->pg_id; 979 p->p_pgrp = pgrp; 980 LIST_INSERT_HEAD(&pgrp->pg_members, p, p_pglist); 981 982 /* Done with the swap; we can release the tty mutex. */ 983 mutex_spin_exit(&tty_lock); 984 985 done: 986 if (pg_id != NO_PGID) { 987 /* Releases proc_lock. */ 988 pg_delete(pg_id); 989 } else { 990 mutex_exit(proc_lock); 991 } 992 if (sess != NULL) 993 kmem_free(sess, sizeof(*sess)); 994 if (new_pgrp != NULL) 995 kmem_free(new_pgrp, sizeof(*new_pgrp)); 996 #ifdef DEBUG_PGRP 997 if (__predict_false(rval)) 998 printf("enterpgrp(%d,%d,%d), curproc %d, rval %d\n", 999 pid, pgid, mksess, curp->p_pid, rval); 1000 #endif 1001 return rval; 1002 } 1003 1004 /* 1005 * proc_leavepgrp: remove a process from its process group. 1006 * => must be called with the proc_lock held, which will be released; 1007 */ 1008 void 1009 proc_leavepgrp(struct proc *p) 1010 { 1011 struct pgrp *pgrp; 1012 1013 KASSERT(mutex_owned(proc_lock)); 1014 1015 /* Interlock with ttread() */ 1016 mutex_spin_enter(&tty_lock); 1017 pgrp = p->p_pgrp; 1018 LIST_REMOVE(p, p_pglist); 1019 p->p_pgrp = NULL; 1020 mutex_spin_exit(&tty_lock); 1021 1022 if (LIST_EMPTY(&pgrp->pg_members)) { 1023 /* Releases proc_lock. */ 1024 pg_delete(pgrp->pg_id); 1025 } else { 1026 mutex_exit(proc_lock); 1027 } 1028 } 1029 1030 /* 1031 * pg_remove: remove a process group from the table. 1032 * => must be called with the proc_lock held; 1033 * => returns process group to free; 1034 */ 1035 static struct pgrp * 1036 pg_remove(pid_t pg_id) 1037 { 1038 struct pgrp *pgrp; 1039 struct pid_table *pt; 1040 1041 KASSERT(mutex_owned(proc_lock)); 1042 1043 pt = &pid_table[pg_id & pid_tbl_mask]; 1044 pgrp = pt->pt_pgrp; 1045 1046 KASSERT(pgrp != NULL); 1047 KASSERT(pgrp->pg_id == pg_id); 1048 KASSERT(LIST_EMPTY(&pgrp->pg_members)); 1049 1050 pt->pt_pgrp = NULL; 1051 1052 if (!P_VALID(pt->pt_proc)) { 1053 /* Orphaned pgrp, put slot onto free list. */ 1054 KASSERT((P_NEXT(pt->pt_proc) & pid_tbl_mask) == 0); 1055 pg_id &= pid_tbl_mask; 1056 pt = &pid_table[last_free_pt]; 1057 pt->pt_proc = P_FREE(P_NEXT(pt->pt_proc) | pg_id); 1058 KASSERT(pt->pt_pid == 0); 1059 last_free_pt = pg_id; 1060 pid_alloc_cnt--; 1061 } 1062 return pgrp; 1063 } 1064 1065 /* 1066 * pg_delete: delete and free a process group. 1067 * => must be called with the proc_lock held, which will be released. 1068 */ 1069 static void 1070 pg_delete(pid_t pg_id) 1071 { 1072 struct pgrp *pg; 1073 struct tty *ttyp; 1074 struct session *ss; 1075 1076 KASSERT(mutex_owned(proc_lock)); 1077 1078 pg = pid_table[pg_id & pid_tbl_mask].pt_pgrp; 1079 if (pg == NULL || pg->pg_id != pg_id || !LIST_EMPTY(&pg->pg_members)) { 1080 mutex_exit(proc_lock); 1081 return; 1082 } 1083 1084 ss = pg->pg_session; 1085 1086 /* Remove reference (if any) from tty to this process group */ 1087 mutex_spin_enter(&tty_lock); 1088 ttyp = ss->s_ttyp; 1089 if (ttyp != NULL && ttyp->t_pgrp == pg) { 1090 ttyp->t_pgrp = NULL; 1091 KASSERT(ttyp->t_session == ss); 1092 } 1093 mutex_spin_exit(&tty_lock); 1094 1095 /* 1096 * The leading process group in a session is freed by proc_sessrele(), 1097 * if last reference. Note: proc_sessrele() releases proc_lock. 1098 */ 1099 pg = (ss->s_sid != pg->pg_id) ? pg_remove(pg_id) : NULL; 1100 proc_sessrele(ss); 1101 1102 if (pg != NULL) { 1103 /* Free it, if was not done by proc_sessrele(). */ 1104 kmem_free(pg, sizeof(struct pgrp)); 1105 } 1106 } 1107 1108 /* 1109 * Adjust pgrp jobc counters when specified process changes process group. 1110 * We count the number of processes in each process group that "qualify" 1111 * the group for terminal job control (those with a parent in a different 1112 * process group of the same session). If that count reaches zero, the 1113 * process group becomes orphaned. Check both the specified process' 1114 * process group and that of its children. 1115 * entering == 0 => p is leaving specified group. 1116 * entering == 1 => p is entering specified group. 1117 * 1118 * Call with proc_lock held. 1119 */ 1120 void 1121 fixjobc(struct proc *p, struct pgrp *pgrp, int entering) 1122 { 1123 struct pgrp *hispgrp; 1124 struct session *mysession = pgrp->pg_session; 1125 struct proc *child; 1126 1127 KASSERT(mutex_owned(proc_lock)); 1128 1129 /* 1130 * Check p's parent to see whether p qualifies its own process 1131 * group; if so, adjust count for p's process group. 1132 */ 1133 hispgrp = p->p_pptr->p_pgrp; 1134 if (hispgrp != pgrp && hispgrp->pg_session == mysession) { 1135 if (entering) { 1136 pgrp->pg_jobc++; 1137 p->p_lflag &= ~PL_ORPHANPG; 1138 } else if (--pgrp->pg_jobc == 0) 1139 orphanpg(pgrp); 1140 } 1141 1142 /* 1143 * Check this process' children to see whether they qualify 1144 * their process groups; if so, adjust counts for children's 1145 * process groups. 1146 */ 1147 LIST_FOREACH(child, &p->p_children, p_sibling) { 1148 hispgrp = child->p_pgrp; 1149 if (hispgrp != pgrp && hispgrp->pg_session == mysession && 1150 !P_ZOMBIE(child)) { 1151 if (entering) { 1152 child->p_lflag &= ~PL_ORPHANPG; 1153 hispgrp->pg_jobc++; 1154 } else if (--hispgrp->pg_jobc == 0) 1155 orphanpg(hispgrp); 1156 } 1157 } 1158 } 1159 1160 /* 1161 * A process group has become orphaned; 1162 * if there are any stopped processes in the group, 1163 * hang-up all process in that group. 1164 * 1165 * Call with proc_lock held. 1166 */ 1167 static void 1168 orphanpg(struct pgrp *pg) 1169 { 1170 struct proc *p; 1171 1172 KASSERT(mutex_owned(proc_lock)); 1173 1174 LIST_FOREACH(p, &pg->pg_members, p_pglist) { 1175 if (p->p_stat == SSTOP) { 1176 p->p_lflag |= PL_ORPHANPG; 1177 psignal(p, SIGHUP); 1178 psignal(p, SIGCONT); 1179 } 1180 } 1181 } 1182 1183 #ifdef DDB 1184 #include <ddb/db_output.h> 1185 void pidtbl_dump(void); 1186 void 1187 pidtbl_dump(void) 1188 { 1189 struct pid_table *pt; 1190 struct proc *p; 1191 struct pgrp *pgrp; 1192 int id; 1193 1194 db_printf("pid table %p size %x, next %x, last %x\n", 1195 pid_table, pid_tbl_mask+1, 1196 next_free_pt, last_free_pt); 1197 for (pt = pid_table, id = 0; id <= pid_tbl_mask; id++, pt++) { 1198 p = pt->pt_proc; 1199 if (!P_VALID(p) && !pt->pt_pgrp) 1200 continue; 1201 db_printf(" id %x: ", id); 1202 if (P_VALID(p)) 1203 db_printf("slotpid %d proc %p id %d (0x%x) %s\n", 1204 pt->pt_pid, p, p->p_pid, p->p_pid, p->p_comm); 1205 else 1206 db_printf("next %x use %x\n", 1207 P_NEXT(p) & pid_tbl_mask, 1208 P_NEXT(p) & ~pid_tbl_mask); 1209 if ((pgrp = pt->pt_pgrp)) { 1210 db_printf("\tsession %p, sid %d, count %d, login %s\n", 1211 pgrp->pg_session, pgrp->pg_session->s_sid, 1212 pgrp->pg_session->s_count, 1213 pgrp->pg_session->s_login); 1214 db_printf("\tpgrp %p, pg_id %d, pg_jobc %d, members %p\n", 1215 pgrp, pgrp->pg_id, pgrp->pg_jobc, 1216 LIST_FIRST(&pgrp->pg_members)); 1217 LIST_FOREACH(p, &pgrp->pg_members, p_pglist) { 1218 db_printf("\t\tpid %d addr %p pgrp %p %s\n", 1219 p->p_pid, p, p->p_pgrp, p->p_comm); 1220 } 1221 } 1222 } 1223 } 1224 #endif /* DDB */ 1225 1226 #ifdef KSTACK_CHECK_MAGIC 1227 1228 #define KSTACK_MAGIC 0xdeadbeaf 1229 1230 /* XXX should be per process basis? */ 1231 static int kstackleftmin = KSTACK_SIZE; 1232 static int kstackleftthres = KSTACK_SIZE / 8; 1233 1234 void 1235 kstack_setup_magic(const struct lwp *l) 1236 { 1237 uint32_t *ip; 1238 uint32_t const *end; 1239 1240 KASSERT(l != NULL); 1241 KASSERT(l != &lwp0); 1242 1243 /* 1244 * fill all the stack with magic number 1245 * so that later modification on it can be detected. 1246 */ 1247 ip = (uint32_t *)KSTACK_LOWEST_ADDR(l); 1248 end = (uint32_t *)((char *)KSTACK_LOWEST_ADDR(l) + KSTACK_SIZE); 1249 for (; ip < end; ip++) { 1250 *ip = KSTACK_MAGIC; 1251 } 1252 } 1253 1254 void 1255 kstack_check_magic(const struct lwp *l) 1256 { 1257 uint32_t const *ip, *end; 1258 int stackleft; 1259 1260 KASSERT(l != NULL); 1261 1262 /* don't check proc0 */ /*XXX*/ 1263 if (l == &lwp0) 1264 return; 1265 1266 #ifdef __MACHINE_STACK_GROWS_UP 1267 /* stack grows upwards (eg. hppa) */ 1268 ip = (uint32_t *)((void *)KSTACK_LOWEST_ADDR(l) + KSTACK_SIZE); 1269 end = (uint32_t *)KSTACK_LOWEST_ADDR(l); 1270 for (ip--; ip >= end; ip--) 1271 if (*ip != KSTACK_MAGIC) 1272 break; 1273 1274 stackleft = (void *)KSTACK_LOWEST_ADDR(l) + KSTACK_SIZE - (void *)ip; 1275 #else /* __MACHINE_STACK_GROWS_UP */ 1276 /* stack grows downwards (eg. i386) */ 1277 ip = (uint32_t *)KSTACK_LOWEST_ADDR(l); 1278 end = (uint32_t *)((char *)KSTACK_LOWEST_ADDR(l) + KSTACK_SIZE); 1279 for (; ip < end; ip++) 1280 if (*ip != KSTACK_MAGIC) 1281 break; 1282 1283 stackleft = ((const char *)ip) - (const char *)KSTACK_LOWEST_ADDR(l); 1284 #endif /* __MACHINE_STACK_GROWS_UP */ 1285 1286 if (kstackleftmin > stackleft) { 1287 kstackleftmin = stackleft; 1288 if (stackleft < kstackleftthres) 1289 printf("warning: kernel stack left %d bytes" 1290 "(pid %u:lid %u)\n", stackleft, 1291 (u_int)l->l_proc->p_pid, (u_int)l->l_lid); 1292 } 1293 1294 if (stackleft <= 0) { 1295 panic("magic on the top of kernel stack changed for " 1296 "pid %u, lid %u: maybe kernel stack overflow", 1297 (u_int)l->l_proc->p_pid, (u_int)l->l_lid); 1298 } 1299 } 1300 #endif /* KSTACK_CHECK_MAGIC */ 1301 1302 int 1303 proclist_foreach_call(struct proclist *list, 1304 int (*callback)(struct proc *, void *arg), void *arg) 1305 { 1306 struct proc marker; 1307 struct proc *p; 1308 int ret = 0; 1309 1310 marker.p_flag = PK_MARKER; 1311 mutex_enter(proc_lock); 1312 for (p = LIST_FIRST(list); ret == 0 && p != NULL;) { 1313 if (p->p_flag & PK_MARKER) { 1314 p = LIST_NEXT(p, p_list); 1315 continue; 1316 } 1317 LIST_INSERT_AFTER(p, &marker, p_list); 1318 ret = (*callback)(p, arg); 1319 KASSERT(mutex_owned(proc_lock)); 1320 p = LIST_NEXT(&marker, p_list); 1321 LIST_REMOVE(&marker, p_list); 1322 } 1323 mutex_exit(proc_lock); 1324 1325 return ret; 1326 } 1327 1328 int 1329 proc_vmspace_getref(struct proc *p, struct vmspace **vm) 1330 { 1331 1332 /* XXXCDC: how should locking work here? */ 1333 1334 /* curproc exception is for coredump. */ 1335 1336 if ((p != curproc && (p->p_sflag & PS_WEXIT) != 0) || 1337 (p->p_vmspace->vm_refcnt < 1)) { /* XXX */ 1338 return EFAULT; 1339 } 1340 1341 uvmspace_addref(p->p_vmspace); 1342 *vm = p->p_vmspace; 1343 1344 return 0; 1345 } 1346 1347 /* 1348 * Acquire a write lock on the process credential. 1349 */ 1350 void 1351 proc_crmod_enter(void) 1352 { 1353 struct lwp *l = curlwp; 1354 struct proc *p = l->l_proc; 1355 kauth_cred_t oc; 1356 1357 /* Reset what needs to be reset in plimit. */ 1358 if (p->p_limit->pl_corename != defcorename) { 1359 lim_setcorename(p, defcorename, 0); 1360 } 1361 1362 mutex_enter(p->p_lock); 1363 1364 /* Ensure the LWP cached credentials are up to date. */ 1365 if ((oc = l->l_cred) != p->p_cred) { 1366 kauth_cred_hold(p->p_cred); 1367 l->l_cred = p->p_cred; 1368 kauth_cred_free(oc); 1369 } 1370 } 1371 1372 /* 1373 * Set in a new process credential, and drop the write lock. The credential 1374 * must have a reference already. Optionally, free a no-longer required 1375 * credential. The scheduler also needs to inspect p_cred, so we also 1376 * briefly acquire the sched state mutex. 1377 */ 1378 void 1379 proc_crmod_leave(kauth_cred_t scred, kauth_cred_t fcred, bool sugid) 1380 { 1381 struct lwp *l = curlwp, *l2; 1382 struct proc *p = l->l_proc; 1383 kauth_cred_t oc; 1384 1385 KASSERT(mutex_owned(p->p_lock)); 1386 1387 /* Is there a new credential to set in? */ 1388 if (scred != NULL) { 1389 p->p_cred = scred; 1390 LIST_FOREACH(l2, &p->p_lwps, l_sibling) { 1391 if (l2 != l) 1392 l2->l_prflag |= LPR_CRMOD; 1393 } 1394 1395 /* Ensure the LWP cached credentials are up to date. */ 1396 if ((oc = l->l_cred) != scred) { 1397 kauth_cred_hold(scred); 1398 l->l_cred = scred; 1399 } 1400 } else 1401 oc = NULL; /* XXXgcc */ 1402 1403 if (sugid) { 1404 /* 1405 * Mark process as having changed credentials, stops 1406 * tracing etc. 1407 */ 1408 p->p_flag |= PK_SUGID; 1409 } 1410 1411 mutex_exit(p->p_lock); 1412 1413 /* If there is a credential to be released, free it now. */ 1414 if (fcred != NULL) { 1415 KASSERT(scred != NULL); 1416 kauth_cred_free(fcred); 1417 if (oc != scred) 1418 kauth_cred_free(oc); 1419 } 1420 } 1421 1422 /* 1423 * proc_specific_key_create -- 1424 * Create a key for subsystem proc-specific data. 1425 */ 1426 int 1427 proc_specific_key_create(specificdata_key_t *keyp, specificdata_dtor_t dtor) 1428 { 1429 1430 return (specificdata_key_create(proc_specificdata_domain, keyp, dtor)); 1431 } 1432 1433 /* 1434 * proc_specific_key_delete -- 1435 * Delete a key for subsystem proc-specific data. 1436 */ 1437 void 1438 proc_specific_key_delete(specificdata_key_t key) 1439 { 1440 1441 specificdata_key_delete(proc_specificdata_domain, key); 1442 } 1443 1444 /* 1445 * proc_initspecific -- 1446 * Initialize a proc's specificdata container. 1447 */ 1448 void 1449 proc_initspecific(struct proc *p) 1450 { 1451 int error; 1452 1453 error = specificdata_init(proc_specificdata_domain, &p->p_specdataref); 1454 KASSERT(error == 0); 1455 } 1456 1457 /* 1458 * proc_finispecific -- 1459 * Finalize a proc's specificdata container. 1460 */ 1461 void 1462 proc_finispecific(struct proc *p) 1463 { 1464 1465 specificdata_fini(proc_specificdata_domain, &p->p_specdataref); 1466 } 1467 1468 /* 1469 * proc_getspecific -- 1470 * Return proc-specific data corresponding to the specified key. 1471 */ 1472 void * 1473 proc_getspecific(struct proc *p, specificdata_key_t key) 1474 { 1475 1476 return (specificdata_getspecific(proc_specificdata_domain, 1477 &p->p_specdataref, key)); 1478 } 1479 1480 /* 1481 * proc_setspecific -- 1482 * Set proc-specific data corresponding to the specified key. 1483 */ 1484 void 1485 proc_setspecific(struct proc *p, specificdata_key_t key, void *data) 1486 { 1487 1488 specificdata_setspecific(proc_specificdata_domain, 1489 &p->p_specdataref, key, data); 1490 } 1491 1492 int 1493 proc_uidmatch(kauth_cred_t cred, kauth_cred_t target) 1494 { 1495 int r = 0; 1496 1497 if (kauth_cred_getuid(cred) != kauth_cred_getuid(target) || 1498 kauth_cred_getuid(cred) != kauth_cred_getsvuid(target)) { 1499 /* 1500 * suid proc of ours or proc not ours 1501 */ 1502 r = EPERM; 1503 } else if (kauth_cred_getgid(target) != kauth_cred_getsvgid(target)) { 1504 /* 1505 * sgid proc has sgid back to us temporarily 1506 */ 1507 r = EPERM; 1508 } else { 1509 /* 1510 * our rgid must be in target's group list (ie, 1511 * sub-processes started by a sgid process) 1512 */ 1513 int ismember = 0; 1514 1515 if (kauth_cred_ismember_gid(cred, 1516 kauth_cred_getgid(target), &ismember) != 0 || 1517 !ismember) 1518 r = EPERM; 1519 } 1520 1521 return (r); 1522 } 1523 1524 /* 1525 * sysctl stuff 1526 */ 1527 1528 #define KERN_PROCSLOP (5 * sizeof(struct kinfo_proc)) 1529 1530 static const u_int sysctl_flagmap[] = { 1531 PK_ADVLOCK, P_ADVLOCK, 1532 PK_EXEC, P_EXEC, 1533 PK_NOCLDWAIT, P_NOCLDWAIT, 1534 PK_32, P_32, 1535 PK_CLDSIGIGN, P_CLDSIGIGN, 1536 PK_SUGID, P_SUGID, 1537 0 1538 }; 1539 1540 static const u_int sysctl_sflagmap[] = { 1541 PS_NOCLDSTOP, P_NOCLDSTOP, 1542 PS_WEXIT, P_WEXIT, 1543 PS_STOPFORK, P_STOPFORK, 1544 PS_STOPEXEC, P_STOPEXEC, 1545 PS_STOPEXIT, P_STOPEXIT, 1546 0 1547 }; 1548 1549 static const u_int sysctl_slflagmap[] = { 1550 PSL_TRACED, P_TRACED, 1551 PSL_FSTRACE, P_FSTRACE, 1552 PSL_CHTRACED, P_CHTRACED, 1553 PSL_SYSCALL, P_SYSCALL, 1554 0 1555 }; 1556 1557 static const u_int sysctl_lflagmap[] = { 1558 PL_CONTROLT, P_CONTROLT, 1559 PL_PPWAIT, P_PPWAIT, 1560 0 1561 }; 1562 1563 static const u_int sysctl_stflagmap[] = { 1564 PST_PROFIL, P_PROFIL, 1565 0 1566 1567 }; 1568 1569 /* used by kern_lwp also */ 1570 const u_int sysctl_lwpflagmap[] = { 1571 LW_SINTR, L_SINTR, 1572 LW_SYSTEM, L_SYSTEM, 1573 0 1574 }; 1575 1576 /* 1577 * Find the most ``active'' lwp of a process and return it for ps display 1578 * purposes 1579 */ 1580 static struct lwp * 1581 proc_active_lwp(struct proc *p) 1582 { 1583 static const int ostat[] = { 1584 0, 1585 2, /* LSIDL */ 1586 6, /* LSRUN */ 1587 5, /* LSSLEEP */ 1588 4, /* LSSTOP */ 1589 0, /* LSZOMB */ 1590 1, /* LSDEAD */ 1591 7, /* LSONPROC */ 1592 3 /* LSSUSPENDED */ 1593 }; 1594 1595 struct lwp *l, *lp = NULL; 1596 LIST_FOREACH(l, &p->p_lwps, l_sibling) { 1597 KASSERT(l->l_stat >= 0 && l->l_stat < __arraycount(ostat)); 1598 if (lp == NULL || 1599 ostat[l->l_stat] > ostat[lp->l_stat] || 1600 (ostat[l->l_stat] == ostat[lp->l_stat] && 1601 l->l_cpticks > lp->l_cpticks)) { 1602 lp = l; 1603 continue; 1604 } 1605 } 1606 return lp; 1607 } 1608 1609 static int 1610 sysctl_doeproc(SYSCTLFN_ARGS) 1611 { 1612 union { 1613 struct kinfo_proc kproc; 1614 struct kinfo_proc2 kproc2; 1615 } *kbuf; 1616 struct proc *p, *next, *marker; 1617 char *where, *dp; 1618 int type, op, arg, error; 1619 u_int elem_size, kelem_size, elem_count; 1620 size_t buflen, needed; 1621 bool match, zombie, mmmbrains; 1622 1623 if (namelen == 1 && name[0] == CTL_QUERY) 1624 return (sysctl_query(SYSCTLFN_CALL(rnode))); 1625 1626 dp = where = oldp; 1627 buflen = where != NULL ? *oldlenp : 0; 1628 error = 0; 1629 needed = 0; 1630 type = rnode->sysctl_num; 1631 1632 if (type == KERN_PROC) { 1633 if (namelen != 2 && !(namelen == 1 && name[0] == KERN_PROC_ALL)) 1634 return (EINVAL); 1635 op = name[0]; 1636 if (op != KERN_PROC_ALL) 1637 arg = name[1]; 1638 else 1639 arg = 0; /* Quell compiler warning */ 1640 elem_count = 0; /* Ditto */ 1641 kelem_size = elem_size = sizeof(kbuf->kproc); 1642 } else { 1643 if (namelen != 4) 1644 return (EINVAL); 1645 op = name[0]; 1646 arg = name[1]; 1647 elem_size = name[2]; 1648 elem_count = name[3]; 1649 kelem_size = sizeof(kbuf->kproc2); 1650 } 1651 1652 sysctl_unlock(); 1653 1654 kbuf = kmem_alloc(sizeof(*kbuf), KM_SLEEP); 1655 marker = kmem_alloc(sizeof(*marker), KM_SLEEP); 1656 marker->p_flag = PK_MARKER; 1657 1658 mutex_enter(proc_lock); 1659 mmmbrains = false; 1660 for (p = LIST_FIRST(&allproc);; p = next) { 1661 if (p == NULL) { 1662 if (!mmmbrains) { 1663 p = LIST_FIRST(&zombproc); 1664 mmmbrains = true; 1665 } 1666 if (p == NULL) 1667 break; 1668 } 1669 next = LIST_NEXT(p, p_list); 1670 if ((p->p_flag & PK_MARKER) != 0) 1671 continue; 1672 1673 /* 1674 * Skip embryonic processes. 1675 */ 1676 if (p->p_stat == SIDL) 1677 continue; 1678 1679 mutex_enter(p->p_lock); 1680 error = kauth_authorize_process(l->l_cred, 1681 KAUTH_PROCESS_CANSEE, p, 1682 KAUTH_ARG(KAUTH_REQ_PROCESS_CANSEE_ENTRY), NULL, NULL); 1683 if (error != 0) { 1684 mutex_exit(p->p_lock); 1685 continue; 1686 } 1687 1688 /* 1689 * TODO - make more efficient (see notes below). 1690 * do by session. 1691 */ 1692 switch (op) { 1693 case KERN_PROC_PID: 1694 /* could do this with just a lookup */ 1695 match = (p->p_pid == (pid_t)arg); 1696 break; 1697 1698 case KERN_PROC_PGRP: 1699 /* could do this by traversing pgrp */ 1700 match = (p->p_pgrp->pg_id == (pid_t)arg); 1701 break; 1702 1703 case KERN_PROC_SESSION: 1704 match = (p->p_session->s_sid == (pid_t)arg); 1705 break; 1706 1707 case KERN_PROC_TTY: 1708 match = true; 1709 if (arg == (int) KERN_PROC_TTY_REVOKE) { 1710 if ((p->p_lflag & PL_CONTROLT) == 0 || 1711 p->p_session->s_ttyp == NULL || 1712 p->p_session->s_ttyvp != NULL) { 1713 match = false; 1714 } 1715 } else if ((p->p_lflag & PL_CONTROLT) == 0 || 1716 p->p_session->s_ttyp == NULL) { 1717 if ((dev_t)arg != KERN_PROC_TTY_NODEV) { 1718 match = false; 1719 } 1720 } else if (p->p_session->s_ttyp->t_dev != (dev_t)arg) { 1721 match = false; 1722 } 1723 break; 1724 1725 case KERN_PROC_UID: 1726 match = (kauth_cred_geteuid(p->p_cred) == (uid_t)arg); 1727 break; 1728 1729 case KERN_PROC_RUID: 1730 match = (kauth_cred_getuid(p->p_cred) == (uid_t)arg); 1731 break; 1732 1733 case KERN_PROC_GID: 1734 match = (kauth_cred_getegid(p->p_cred) == (uid_t)arg); 1735 break; 1736 1737 case KERN_PROC_RGID: 1738 match = (kauth_cred_getgid(p->p_cred) == (uid_t)arg); 1739 break; 1740 1741 case KERN_PROC_ALL: 1742 match = true; 1743 /* allow everything */ 1744 break; 1745 1746 default: 1747 error = EINVAL; 1748 mutex_exit(p->p_lock); 1749 goto cleanup; 1750 } 1751 if (!match) { 1752 mutex_exit(p->p_lock); 1753 continue; 1754 } 1755 1756 /* 1757 * Grab a hold on the process. 1758 */ 1759 if (mmmbrains) { 1760 zombie = true; 1761 } else { 1762 zombie = !rw_tryenter(&p->p_reflock, RW_READER); 1763 } 1764 if (zombie) { 1765 LIST_INSERT_AFTER(p, marker, p_list); 1766 } 1767 1768 if (buflen >= elem_size && 1769 (type == KERN_PROC || elem_count > 0)) { 1770 if (type == KERN_PROC) { 1771 kbuf->kproc.kp_proc = *p; 1772 fill_eproc(p, &kbuf->kproc.kp_eproc, zombie); 1773 } else { 1774 fill_kproc2(p, &kbuf->kproc2, zombie); 1775 elem_count--; 1776 } 1777 mutex_exit(p->p_lock); 1778 mutex_exit(proc_lock); 1779 /* 1780 * Copy out elem_size, but not larger than kelem_size 1781 */ 1782 error = sysctl_copyout(l, kbuf, dp, 1783 min(kelem_size, elem_size)); 1784 mutex_enter(proc_lock); 1785 if (error) { 1786 goto bah; 1787 } 1788 dp += elem_size; 1789 buflen -= elem_size; 1790 } else { 1791 mutex_exit(p->p_lock); 1792 } 1793 needed += elem_size; 1794 1795 /* 1796 * Release reference to process. 1797 */ 1798 if (zombie) { 1799 next = LIST_NEXT(marker, p_list); 1800 LIST_REMOVE(marker, p_list); 1801 } else { 1802 rw_exit(&p->p_reflock); 1803 next = LIST_NEXT(p, p_list); 1804 } 1805 } 1806 mutex_exit(proc_lock); 1807 1808 if (where != NULL) { 1809 *oldlenp = dp - where; 1810 if (needed > *oldlenp) { 1811 error = ENOMEM; 1812 goto out; 1813 } 1814 } else { 1815 needed += KERN_PROCSLOP; 1816 *oldlenp = needed; 1817 } 1818 if (kbuf) 1819 kmem_free(kbuf, sizeof(*kbuf)); 1820 if (marker) 1821 kmem_free(marker, sizeof(*marker)); 1822 sysctl_relock(); 1823 return 0; 1824 bah: 1825 if (zombie) 1826 LIST_REMOVE(marker, p_list); 1827 else 1828 rw_exit(&p->p_reflock); 1829 cleanup: 1830 mutex_exit(proc_lock); 1831 out: 1832 if (kbuf) 1833 kmem_free(kbuf, sizeof(*kbuf)); 1834 if (marker) 1835 kmem_free(marker, sizeof(*marker)); 1836 sysctl_relock(); 1837 return error; 1838 } 1839 1840 int 1841 copyin_psstrings(struct proc *p, struct ps_strings *arginfo) 1842 { 1843 1844 #ifdef COMPAT_NETBSD32 1845 if (p->p_flag & PK_32) { 1846 struct ps_strings32 arginfo32; 1847 1848 int error = copyin_proc(p, (void *)p->p_psstrp, &arginfo32, 1849 sizeof(arginfo32)); 1850 if (error) 1851 return error; 1852 arginfo->ps_argvstr = (void *)(uintptr_t)arginfo32.ps_argvstr; 1853 arginfo->ps_nargvstr = arginfo32.ps_nargvstr; 1854 arginfo->ps_envstr = (void *)(uintptr_t)arginfo32.ps_envstr; 1855 arginfo->ps_nenvstr = arginfo32.ps_nenvstr; 1856 return 0; 1857 } 1858 #endif 1859 return copyin_proc(p, (void *)p->p_psstrp, arginfo, sizeof(*arginfo)); 1860 } 1861 1862 static int 1863 copy_procargs_sysctl_cb(void *cookie_, const void *src, size_t off, size_t len) 1864 { 1865 void **cookie = cookie_; 1866 struct lwp *l = cookie[0]; 1867 char *dst = cookie[1]; 1868 1869 return sysctl_copyout(l, src, dst + off, len); 1870 } 1871 1872 /* 1873 * sysctl helper routine for kern.proc_args pseudo-subtree. 1874 */ 1875 static int 1876 sysctl_kern_proc_args(SYSCTLFN_ARGS) 1877 { 1878 struct ps_strings pss; 1879 struct proc *p; 1880 pid_t pid; 1881 int type, error; 1882 void *cookie[2]; 1883 1884 if (namelen == 1 && name[0] == CTL_QUERY) 1885 return (sysctl_query(SYSCTLFN_CALL(rnode))); 1886 1887 if (newp != NULL || namelen != 2) 1888 return (EINVAL); 1889 pid = name[0]; 1890 type = name[1]; 1891 1892 switch (type) { 1893 case KERN_PROC_ARGV: 1894 case KERN_PROC_NARGV: 1895 case KERN_PROC_ENV: 1896 case KERN_PROC_NENV: 1897 /* ok */ 1898 break; 1899 default: 1900 return (EINVAL); 1901 } 1902 1903 sysctl_unlock(); 1904 1905 /* check pid */ 1906 mutex_enter(proc_lock); 1907 if ((p = proc_find(pid)) == NULL) { 1908 error = EINVAL; 1909 goto out_locked; 1910 } 1911 mutex_enter(p->p_lock); 1912 1913 /* Check permission. */ 1914 if (type == KERN_PROC_ARGV || type == KERN_PROC_NARGV) 1915 error = kauth_authorize_process(l->l_cred, KAUTH_PROCESS_CANSEE, 1916 p, KAUTH_ARG(KAUTH_REQ_PROCESS_CANSEE_ARGS), NULL, NULL); 1917 else if (type == KERN_PROC_ENV || type == KERN_PROC_NENV) 1918 error = kauth_authorize_process(l->l_cred, KAUTH_PROCESS_CANSEE, 1919 p, KAUTH_ARG(KAUTH_REQ_PROCESS_CANSEE_ENV), NULL, NULL); 1920 else 1921 error = EINVAL; /* XXXGCC */ 1922 if (error) { 1923 mutex_exit(p->p_lock); 1924 goto out_locked; 1925 } 1926 1927 if (oldp == NULL) { 1928 if (type == KERN_PROC_NARGV || type == KERN_PROC_NENV) 1929 *oldlenp = sizeof (int); 1930 else 1931 *oldlenp = ARG_MAX; /* XXX XXX XXX */ 1932 error = 0; 1933 mutex_exit(p->p_lock); 1934 goto out_locked; 1935 } 1936 1937 /* 1938 * Zombies don't have a stack, so we can't read their psstrings. 1939 * System processes also don't have a user stack. 1940 */ 1941 if (P_ZOMBIE(p) || (p->p_flag & PK_SYSTEM) != 0) { 1942 error = EINVAL; 1943 mutex_exit(p->p_lock); 1944 goto out_locked; 1945 } 1946 1947 error = rw_tryenter(&p->p_reflock, RW_READER) ? 0 : EBUSY; 1948 mutex_exit(p->p_lock); 1949 if (error) { 1950 goto out_locked; 1951 } 1952 mutex_exit(proc_lock); 1953 1954 if (type == KERN_PROC_NARGV || type == KERN_PROC_NENV) { 1955 int value; 1956 if ((error = copyin_psstrings(p, &pss)) == 0) { 1957 if (type == KERN_PROC_NARGV) 1958 value = pss.ps_nargvstr; 1959 else 1960 value = pss.ps_nenvstr; 1961 error = sysctl_copyout(l, &value, oldp, sizeof(value)); 1962 *oldlenp = sizeof(value); 1963 } 1964 } else { 1965 cookie[0] = l; 1966 cookie[1] = oldp; 1967 error = copy_procargs(p, type, oldlenp, 1968 copy_procargs_sysctl_cb, cookie); 1969 } 1970 rw_exit(&p->p_reflock); 1971 sysctl_relock(); 1972 return error; 1973 1974 out_locked: 1975 mutex_exit(proc_lock); 1976 sysctl_relock(); 1977 return error; 1978 } 1979 1980 int 1981 copy_procargs(struct proc *p, int oid, size_t *limit, 1982 int (*cb)(void *, const void *, size_t, size_t), void *cookie) 1983 { 1984 struct ps_strings pss; 1985 size_t len, i, loaded, entry_len; 1986 struct uio auio; 1987 struct iovec aiov; 1988 int error, argvlen; 1989 char *arg; 1990 char **argv; 1991 vaddr_t user_argv; 1992 struct vmspace *vmspace; 1993 1994 /* 1995 * Allocate a temporary buffer to hold the argument vector and 1996 * the arguments themselve. 1997 */ 1998 arg = kmem_alloc(PAGE_SIZE, KM_SLEEP); 1999 argv = kmem_alloc(PAGE_SIZE, KM_SLEEP); 2000 2001 /* 2002 * Lock the process down in memory. 2003 */ 2004 vmspace = p->p_vmspace; 2005 uvmspace_addref(vmspace); 2006 2007 /* 2008 * Read in the ps_strings structure. 2009 */ 2010 if ((error = copyin_psstrings(p, &pss)) != 0) 2011 goto done; 2012 2013 /* 2014 * Now read the address of the argument vector. 2015 */ 2016 switch (oid) { 2017 case KERN_PROC_ARGV: 2018 user_argv = (uintptr_t)pss.ps_argvstr; 2019 argvlen = pss.ps_nargvstr; 2020 break; 2021 case KERN_PROC_ENV: 2022 user_argv = (uintptr_t)pss.ps_envstr; 2023 argvlen = pss.ps_nenvstr; 2024 break; 2025 default: 2026 error = EINVAL; 2027 goto done; 2028 } 2029 2030 if (argvlen < 0) { 2031 error = EIO; 2032 goto done; 2033 } 2034 2035 #ifdef COMPAT_NETBSD32 2036 if (p->p_flag & PK_32) 2037 entry_len = sizeof(netbsd32_charp); 2038 else 2039 #endif 2040 entry_len = sizeof(char *); 2041 2042 /* 2043 * Now copy each string. 2044 */ 2045 len = 0; /* bytes written to user buffer */ 2046 loaded = 0; /* bytes from argv already processed */ 2047 i = 0; /* To make compiler happy */ 2048 2049 for (; argvlen; --argvlen) { 2050 int finished = 0; 2051 vaddr_t base; 2052 size_t xlen; 2053 int j; 2054 2055 if (loaded == 0) { 2056 size_t rem = entry_len * argvlen; 2057 loaded = MIN(rem, PAGE_SIZE); 2058 error = copyin_vmspace(vmspace, 2059 (const void *)user_argv, argv, loaded); 2060 if (error) 2061 break; 2062 user_argv += loaded; 2063 i = 0; 2064 } 2065 2066 #ifdef COMPAT_NETBSD32 2067 if (p->p_flag & PK_32) { 2068 netbsd32_charp *argv32; 2069 2070 argv32 = (netbsd32_charp *)argv; 2071 base = (vaddr_t)NETBSD32PTR64(argv32[i++]); 2072 } else 2073 #endif 2074 base = (vaddr_t)argv[i++]; 2075 loaded -= entry_len; 2076 2077 /* 2078 * The program has messed around with its arguments, 2079 * possibly deleting some, and replacing them with 2080 * NULL's. Treat this as the last argument and not 2081 * a failure. 2082 */ 2083 if (base == 0) 2084 break; 2085 2086 while (!finished) { 2087 xlen = PAGE_SIZE - (base & PAGE_MASK); 2088 2089 aiov.iov_base = arg; 2090 aiov.iov_len = PAGE_SIZE; 2091 auio.uio_iov = &aiov; 2092 auio.uio_iovcnt = 1; 2093 auio.uio_offset = base; 2094 auio.uio_resid = xlen; 2095 auio.uio_rw = UIO_READ; 2096 UIO_SETUP_SYSSPACE(&auio); 2097 error = uvm_io(&vmspace->vm_map, &auio); 2098 if (error) 2099 goto done; 2100 2101 /* Look for the end of the string */ 2102 for (j = 0; j < xlen; j++) { 2103 if (arg[j] == '\0') { 2104 xlen = j + 1; 2105 finished = 1; 2106 break; 2107 } 2108 } 2109 2110 /* Check for user buffer overflow */ 2111 if (len + xlen > *limit) { 2112 finished = 1; 2113 if (len > *limit) 2114 xlen = 0; 2115 else 2116 xlen = *limit - len; 2117 } 2118 2119 /* Copyout the page */ 2120 error = (*cb)(cookie, arg, len, xlen); 2121 if (error) 2122 goto done; 2123 2124 len += xlen; 2125 base += xlen; 2126 } 2127 } 2128 *limit = len; 2129 2130 done: 2131 kmem_free(argv, PAGE_SIZE); 2132 kmem_free(arg, PAGE_SIZE); 2133 uvmspace_free(vmspace); 2134 return error; 2135 } 2136 2137 /* 2138 * Fill in an eproc structure for the specified process. 2139 */ 2140 void 2141 fill_eproc(struct proc *p, struct eproc *ep, bool zombie) 2142 { 2143 struct tty *tp; 2144 struct lwp *l; 2145 2146 KASSERT(mutex_owned(proc_lock)); 2147 KASSERT(mutex_owned(p->p_lock)); 2148 2149 memset(ep, 0, sizeof(*ep)); 2150 2151 ep->e_paddr = p; 2152 ep->e_sess = p->p_session; 2153 if (p->p_cred) { 2154 kauth_cred_topcred(p->p_cred, &ep->e_pcred); 2155 kauth_cred_toucred(p->p_cred, &ep->e_ucred); 2156 } 2157 if (p->p_stat != SIDL && !P_ZOMBIE(p) && !zombie) { 2158 struct vmspace *vm = p->p_vmspace; 2159 2160 ep->e_vm.vm_rssize = vm_resident_count(vm); 2161 ep->e_vm.vm_tsize = vm->vm_tsize; 2162 ep->e_vm.vm_dsize = vm->vm_dsize; 2163 ep->e_vm.vm_ssize = vm->vm_ssize; 2164 ep->e_vm.vm_map.size = vm->vm_map.size; 2165 2166 /* Pick the primary (first) LWP */ 2167 l = proc_active_lwp(p); 2168 KASSERT(l != NULL); 2169 lwp_lock(l); 2170 if (l->l_wchan) 2171 strncpy(ep->e_wmesg, l->l_wmesg, WMESGLEN); 2172 lwp_unlock(l); 2173 } 2174 if (p->p_pptr) 2175 ep->e_ppid = p->p_pptr->p_pid; 2176 if (p->p_pgrp && p->p_session) { 2177 ep->e_pgid = p->p_pgrp->pg_id; 2178 ep->e_jobc = p->p_pgrp->pg_jobc; 2179 ep->e_sid = p->p_session->s_sid; 2180 if ((p->p_lflag & PL_CONTROLT) && 2181 (tp = ep->e_sess->s_ttyp)) { 2182 ep->e_tdev = tp->t_dev; 2183 ep->e_tpgid = tp->t_pgrp ? tp->t_pgrp->pg_id : NO_PGID; 2184 ep->e_tsess = tp->t_session; 2185 } else 2186 ep->e_tdev = (uint32_t)NODEV; 2187 ep->e_flag = ep->e_sess->s_ttyvp ? EPROC_CTTY : 0; 2188 if (SESS_LEADER(p)) 2189 ep->e_flag |= EPROC_SLEADER; 2190 strncpy(ep->e_login, ep->e_sess->s_login, MAXLOGNAME); 2191 } 2192 ep->e_xsize = ep->e_xrssize = 0; 2193 ep->e_xccount = ep->e_xswrss = 0; 2194 } 2195 2196 /* 2197 * Fill in a kinfo_proc2 structure for the specified process. 2198 */ 2199 static void 2200 fill_kproc2(struct proc *p, struct kinfo_proc2 *ki, bool zombie) 2201 { 2202 struct tty *tp; 2203 struct lwp *l, *l2; 2204 struct timeval ut, st, rt; 2205 sigset_t ss1, ss2; 2206 struct rusage ru; 2207 struct vmspace *vm; 2208 2209 KASSERT(mutex_owned(proc_lock)); 2210 KASSERT(mutex_owned(p->p_lock)); 2211 2212 sigemptyset(&ss1); 2213 sigemptyset(&ss2); 2214 memset(ki, 0, sizeof(*ki)); 2215 2216 ki->p_paddr = PTRTOUINT64(p); 2217 ki->p_fd = PTRTOUINT64(p->p_fd); 2218 ki->p_cwdi = PTRTOUINT64(p->p_cwdi); 2219 ki->p_stats = PTRTOUINT64(p->p_stats); 2220 ki->p_limit = PTRTOUINT64(p->p_limit); 2221 ki->p_vmspace = PTRTOUINT64(p->p_vmspace); 2222 ki->p_sigacts = PTRTOUINT64(p->p_sigacts); 2223 ki->p_sess = PTRTOUINT64(p->p_session); 2224 ki->p_tsess = 0; /* may be changed if controlling tty below */ 2225 ki->p_ru = PTRTOUINT64(&p->p_stats->p_ru); 2226 ki->p_eflag = 0; 2227 ki->p_exitsig = p->p_exitsig; 2228 ki->p_flag = L_INMEM; /* Process never swapped out */ 2229 ki->p_flag |= sysctl_map_flags(sysctl_flagmap, p->p_flag); 2230 ki->p_flag |= sysctl_map_flags(sysctl_sflagmap, p->p_sflag); 2231 ki->p_flag |= sysctl_map_flags(sysctl_slflagmap, p->p_slflag); 2232 ki->p_flag |= sysctl_map_flags(sysctl_lflagmap, p->p_lflag); 2233 ki->p_flag |= sysctl_map_flags(sysctl_stflagmap, p->p_stflag); 2234 ki->p_pid = p->p_pid; 2235 if (p->p_pptr) 2236 ki->p_ppid = p->p_pptr->p_pid; 2237 else 2238 ki->p_ppid = 0; 2239 ki->p_uid = kauth_cred_geteuid(p->p_cred); 2240 ki->p_ruid = kauth_cred_getuid(p->p_cred); 2241 ki->p_gid = kauth_cred_getegid(p->p_cred); 2242 ki->p_rgid = kauth_cred_getgid(p->p_cred); 2243 ki->p_svuid = kauth_cred_getsvuid(p->p_cred); 2244 ki->p_svgid = kauth_cred_getsvgid(p->p_cred); 2245 ki->p_ngroups = kauth_cred_ngroups(p->p_cred); 2246 kauth_cred_getgroups(p->p_cred, ki->p_groups, 2247 min(ki->p_ngroups, sizeof(ki->p_groups) / sizeof(ki->p_groups[0])), 2248 UIO_SYSSPACE); 2249 2250 ki->p_uticks = p->p_uticks; 2251 ki->p_sticks = p->p_sticks; 2252 ki->p_iticks = p->p_iticks; 2253 ki->p_tpgid = NO_PGID; /* may be changed if controlling tty below */ 2254 ki->p_tracep = PTRTOUINT64(p->p_tracep); 2255 ki->p_traceflag = p->p_traceflag; 2256 2257 memcpy(&ki->p_sigignore, &p->p_sigctx.ps_sigignore,sizeof(ki_sigset_t)); 2258 memcpy(&ki->p_sigcatch, &p->p_sigctx.ps_sigcatch, sizeof(ki_sigset_t)); 2259 2260 ki->p_cpticks = 0; 2261 ki->p_pctcpu = p->p_pctcpu; 2262 ki->p_estcpu = 0; 2263 ki->p_stat = p->p_stat; /* Will likely be overridden by LWP status */ 2264 ki->p_realstat = p->p_stat; 2265 ki->p_nice = p->p_nice; 2266 ki->p_xstat = p->p_xstat; 2267 ki->p_acflag = p->p_acflag; 2268 2269 strncpy(ki->p_comm, p->p_comm, 2270 min(sizeof(ki->p_comm), sizeof(p->p_comm))); 2271 strncpy(ki->p_ename, p->p_emul->e_name, sizeof(ki->p_ename)); 2272 2273 ki->p_nlwps = p->p_nlwps; 2274 ki->p_realflag = ki->p_flag; 2275 2276 if (p->p_stat != SIDL && !P_ZOMBIE(p) && !zombie) { 2277 vm = p->p_vmspace; 2278 ki->p_vm_rssize = vm_resident_count(vm); 2279 ki->p_vm_tsize = vm->vm_tsize; 2280 ki->p_vm_dsize = vm->vm_dsize; 2281 ki->p_vm_ssize = vm->vm_ssize; 2282 ki->p_vm_vsize = vm->vm_map.size; 2283 /* 2284 * Since the stack is initially mapped mostly with 2285 * PROT_NONE and grown as needed, adjust the "mapped size" 2286 * to skip the unused stack portion. 2287 */ 2288 ki->p_vm_msize = 2289 atop(vm->vm_map.size) - vm->vm_issize + vm->vm_ssize; 2290 2291 /* Pick the primary (first) LWP */ 2292 l = proc_active_lwp(p); 2293 KASSERT(l != NULL); 2294 lwp_lock(l); 2295 ki->p_nrlwps = p->p_nrlwps; 2296 ki->p_forw = 0; 2297 ki->p_back = 0; 2298 ki->p_addr = PTRTOUINT64(l->l_addr); 2299 ki->p_stat = l->l_stat; 2300 ki->p_flag |= sysctl_map_flags(sysctl_lwpflagmap, l->l_flag); 2301 ki->p_swtime = l->l_swtime; 2302 ki->p_slptime = l->l_slptime; 2303 if (l->l_stat == LSONPROC) 2304 ki->p_schedflags = l->l_cpu->ci_schedstate.spc_flags; 2305 else 2306 ki->p_schedflags = 0; 2307 ki->p_priority = lwp_eprio(l); 2308 ki->p_usrpri = l->l_priority; 2309 if (l->l_wchan) 2310 strncpy(ki->p_wmesg, l->l_wmesg, sizeof(ki->p_wmesg)); 2311 ki->p_wchan = PTRTOUINT64(l->l_wchan); 2312 ki->p_cpuid = cpu_index(l->l_cpu); 2313 lwp_unlock(l); 2314 LIST_FOREACH(l, &p->p_lwps, l_sibling) { 2315 /* This is hardly correct, but... */ 2316 sigplusset(&l->l_sigpend.sp_set, &ss1); 2317 sigplusset(&l->l_sigmask, &ss2); 2318 ki->p_cpticks += l->l_cpticks; 2319 ki->p_pctcpu += l->l_pctcpu; 2320 ki->p_estcpu += l->l_estcpu; 2321 } 2322 } 2323 sigplusset(&p->p_sigpend.sp_set, &ss2); 2324 memcpy(&ki->p_siglist, &ss1, sizeof(ki_sigset_t)); 2325 memcpy(&ki->p_sigmask, &ss2, sizeof(ki_sigset_t)); 2326 2327 if (p->p_session != NULL) { 2328 ki->p_sid = p->p_session->s_sid; 2329 ki->p__pgid = p->p_pgrp->pg_id; 2330 if (p->p_session->s_ttyvp) 2331 ki->p_eflag |= EPROC_CTTY; 2332 if (SESS_LEADER(p)) 2333 ki->p_eflag |= EPROC_SLEADER; 2334 strncpy(ki->p_login, p->p_session->s_login, 2335 min(sizeof ki->p_login - 1, sizeof p->p_session->s_login)); 2336 ki->p_jobc = p->p_pgrp->pg_jobc; 2337 if ((p->p_lflag & PL_CONTROLT) && (tp = p->p_session->s_ttyp)) { 2338 ki->p_tdev = tp->t_dev; 2339 ki->p_tpgid = tp->t_pgrp ? tp->t_pgrp->pg_id : NO_PGID; 2340 ki->p_tsess = PTRTOUINT64(tp->t_session); 2341 } else { 2342 ki->p_tdev = (int32_t)NODEV; 2343 } 2344 } 2345 2346 if (!P_ZOMBIE(p) && !zombie) { 2347 ki->p_uvalid = 1; 2348 ki->p_ustart_sec = p->p_stats->p_start.tv_sec; 2349 ki->p_ustart_usec = p->p_stats->p_start.tv_usec; 2350 2351 calcru(p, &ut, &st, NULL, &rt); 2352 ki->p_rtime_sec = rt.tv_sec; 2353 ki->p_rtime_usec = rt.tv_usec; 2354 ki->p_uutime_sec = ut.tv_sec; 2355 ki->p_uutime_usec = ut.tv_usec; 2356 ki->p_ustime_sec = st.tv_sec; 2357 ki->p_ustime_usec = st.tv_usec; 2358 2359 memcpy(&ru, &p->p_stats->p_ru, sizeof(ru)); 2360 ki->p_uru_nvcsw = 0; 2361 ki->p_uru_nivcsw = 0; 2362 LIST_FOREACH(l2, &p->p_lwps, l_sibling) { 2363 ki->p_uru_nvcsw += (l2->l_ncsw - l2->l_nivcsw); 2364 ki->p_uru_nivcsw += l2->l_nivcsw; 2365 ruadd(&ru, &l2->l_ru); 2366 } 2367 ki->p_uru_maxrss = ru.ru_maxrss; 2368 ki->p_uru_ixrss = ru.ru_ixrss; 2369 ki->p_uru_idrss = ru.ru_idrss; 2370 ki->p_uru_isrss = ru.ru_isrss; 2371 ki->p_uru_minflt = ru.ru_minflt; 2372 ki->p_uru_majflt = ru.ru_majflt; 2373 ki->p_uru_nswap = ru.ru_nswap; 2374 ki->p_uru_inblock = ru.ru_inblock; 2375 ki->p_uru_oublock = ru.ru_oublock; 2376 ki->p_uru_msgsnd = ru.ru_msgsnd; 2377 ki->p_uru_msgrcv = ru.ru_msgrcv; 2378 ki->p_uru_nsignals = ru.ru_nsignals; 2379 2380 timeradd(&p->p_stats->p_cru.ru_utime, 2381 &p->p_stats->p_cru.ru_stime, &ut); 2382 ki->p_uctime_sec = ut.tv_sec; 2383 ki->p_uctime_usec = ut.tv_usec; 2384 } 2385 } 2386
Indexes created Sun Sep 21 20:09:37 GMT 2025