kern_proc.c revision 1.186
1 /* $NetBSD: kern_proc.c,v 1.186 2012/06/09 02:31:14 christos 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 1.186 2012/06/09 02:31:14 christos 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> 104 105 #ifdef COMPAT_NETBSD32 106 #include <compat/netbsd32/netbsd32.h> 107 #endif 108 109 /* 110 * Process lists. 111 */ 112 113 struct proclist allproc __cacheline_aligned; 114 struct proclist zombproc __cacheline_aligned; 115 116 kmutex_t * proc_lock __cacheline_aligned; 117 118 /* 119 * pid to proc lookup is done by indexing the pid_table array. 120 * Since pid numbers are only allocated when an empty slot 121 * has been found, there is no need to search any lists ever. 122 * (an orphaned pgrp will lock the slot, a session will lock 123 * the pgrp with the same number.) 124 * If the table is too small it is reallocated with twice the 125 * previous size and the entries 'unzipped' into the two halves. 126 * A linked list of free entries is passed through the pt_proc 127 * field of 'free' items - set odd to be an invalid ptr. 128 */ 129 130 struct pid_table { 131 struct proc *pt_proc; 132 struct pgrp *pt_pgrp; 133 pid_t pt_pid; 134 }; 135 #if 1 /* strongly typed cast - should be a noop */ 136 static inline uint p2u(struct proc *p) { return (uint)(uintptr_t)p; } 137 #else 138 #define p2u(p) ((uint)p) 139 #endif 140 #define P_VALID(p) (!(p2u(p) & 1)) 141 #define P_NEXT(p) (p2u(p) >> 1) 142 #define P_FREE(pid) ((struct proc *)(uintptr_t)((pid) << 1 | 1)) 143 144 /* 145 * Table of process IDs (PIDs). 146 */ 147 static struct pid_table *pid_table __read_mostly; 148 149 #define INITIAL_PID_TABLE_SIZE (1 << 5) 150 151 /* Table mask, threshold for growing and number of allocated PIDs. */ 152 static u_int pid_tbl_mask __read_mostly; 153 static u_int pid_alloc_lim __read_mostly; 154 static u_int pid_alloc_cnt __cacheline_aligned; 155 156 /* Next free, last free and maximum PIDs. */ 157 static u_int next_free_pt __cacheline_aligned; 158 static u_int last_free_pt __cacheline_aligned; 159 static pid_t pid_max __read_mostly; 160 161 /* Components of the first process -- never freed. */ 162 163 extern struct emul emul_netbsd; /* defined in kern_exec.c */ 164 165 struct session session0 = { 166 .s_count = 1, 167 .s_sid = 0, 168 }; 169 struct pgrp pgrp0 = { 170 .pg_members = LIST_HEAD_INITIALIZER(&pgrp0.pg_members), 171 .pg_session = &session0, 172 }; 173 filedesc_t filedesc0; 174 struct cwdinfo cwdi0 = { 175 .cwdi_cmask = CMASK, /* see cmask below */ 176 .cwdi_refcnt = 1, 177 }; 178 struct plimit limit0; 179 struct pstats pstat0; 180 struct vmspace vmspace0; 181 struct sigacts sigacts0; 182 struct proc proc0 = { 183 .p_lwps = LIST_HEAD_INITIALIZER(&proc0.p_lwps), 184 .p_sigwaiters = LIST_HEAD_INITIALIZER(&proc0.p_sigwaiters), 185 .p_nlwps = 1, 186 .p_nrlwps = 1, 187 .p_nlwpid = 1, /* must match lwp0.l_lid */ 188 .p_pgrp = &pgrp0, 189 .p_comm = "system", 190 /* 191 * Set P_NOCLDWAIT so that kernel threads are reparented to init(8) 192 * when they exit. init(8) can easily wait them out for us. 193 */ 194 .p_flag = PK_SYSTEM | PK_NOCLDWAIT, 195 .p_stat = SACTIVE, 196 .p_nice = NZERO, 197 .p_emul = &emul_netbsd, 198 .p_cwdi = &cwdi0, 199 .p_limit = &limit0, 200 .p_fd = &filedesc0, 201 .p_vmspace = &vmspace0, 202 .p_stats = &pstat0, 203 .p_sigacts = &sigacts0, 204 }; 205 kauth_cred_t cred0; 206 207 static const int nofile = NOFILE; 208 static const int maxuprc = MAXUPRC; 209 static const int cmask = CMASK; 210 211 static int sysctl_doeproc(SYSCTLFN_PROTO); 212 static int sysctl_kern_proc_args(SYSCTLFN_PROTO); 213 static void fill_kproc2(struct proc *, struct kinfo_proc2 *, bool); 214 215 /* 216 * The process list descriptors, used during pid allocation and 217 * by sysctl. No locking on this data structure is needed since 218 * it is completely static. 219 */ 220 const struct proclist_desc proclists[] = { 221 { &allproc }, 222 { &zombproc }, 223 { NULL }, 224 }; 225 226 static struct pgrp * pg_remove(pid_t); 227 static void pg_delete(pid_t); 228 static void orphanpg(struct pgrp *); 229 230 static specificdata_domain_t proc_specificdata_domain; 231 232 static pool_cache_t proc_cache; 233 234 static kauth_listener_t proc_listener; 235 236 static int 237 proc_listener_cb(kauth_cred_t cred, kauth_action_t action, void *cookie, 238 void *arg0, void *arg1, void *arg2, void *arg3) 239 { 240 struct proc *p; 241 int result; 242 243 result = KAUTH_RESULT_DEFER; 244 p = arg0; 245 246 switch (action) { 247 case KAUTH_PROCESS_CANSEE: { 248 enum kauth_process_req req; 249 250 req = (enum kauth_process_req)arg1; 251 252 switch (req) { 253 case KAUTH_REQ_PROCESS_CANSEE_ARGS: 254 case KAUTH_REQ_PROCESS_CANSEE_ENTRY: 255 case KAUTH_REQ_PROCESS_CANSEE_OPENFILES: 256 result = KAUTH_RESULT_ALLOW; 257 258 break; 259 260 case KAUTH_REQ_PROCESS_CANSEE_ENV: 261 if (kauth_cred_getuid(cred) != 262 kauth_cred_getuid(p->p_cred) || 263 kauth_cred_getuid(cred) != 264 kauth_cred_getsvuid(p->p_cred)) 265 break; 266 267 result = KAUTH_RESULT_ALLOW; 268 269 break; 270 271 default: 272 break; 273 } 274 275 break; 276 } 277 278 case KAUTH_PROCESS_FORK: { 279 int lnprocs = (int)(unsigned long)arg2; 280 281 /* 282 * Don't allow a nonprivileged user to use the last few 283 * processes. The variable lnprocs is the current number of 284 * processes, maxproc is the limit. 285 */ 286 if (__predict_false((lnprocs >= maxproc - 5))) 287 break; 288 289 result = KAUTH_RESULT_ALLOW; 290 291 break; 292 } 293 294 case KAUTH_PROCESS_CORENAME: 295 case KAUTH_PROCESS_STOPFLAG: 296 if (proc_uidmatch(cred, p->p_cred) == 0) 297 result = KAUTH_RESULT_ALLOW; 298 299 break; 300 301 default: 302 break; 303 } 304 305 return result; 306 } 307 308 /* 309 * Initialize global process hashing structures. 310 */ 311 void 312 procinit(void) 313 { 314 const struct proclist_desc *pd; 315 u_int i; 316 #define LINK_EMPTY ((PID_MAX + INITIAL_PID_TABLE_SIZE) & ~(INITIAL_PID_TABLE_SIZE - 1)) 317 318 for (pd = proclists; pd->pd_list != NULL; pd++) 319 LIST_INIT(pd->pd_list); 320 321 proc_lock = mutex_obj_alloc(MUTEX_DEFAULT, IPL_NONE); 322 pid_table = kmem_alloc(INITIAL_PID_TABLE_SIZE 323 * sizeof(struct pid_table), KM_SLEEP); 324 pid_tbl_mask = INITIAL_PID_TABLE_SIZE - 1; 325 pid_max = PID_MAX; 326 327 /* Set free list running through table... 328 Preset 'use count' above PID_MAX so we allocate pid 1 next. */ 329 for (i = 0; i <= pid_tbl_mask; i++) { 330 pid_table[i].pt_proc = P_FREE(LINK_EMPTY + i + 1); 331 pid_table[i].pt_pgrp = 0; 332 pid_table[i].pt_pid = 0; 333 } 334 /* slot 0 is just grabbed */ 335 next_free_pt = 1; 336 /* Need to fix last entry. */ 337 last_free_pt = pid_tbl_mask; 338 pid_table[last_free_pt].pt_proc = P_FREE(LINK_EMPTY); 339 /* point at which we grow table - to avoid reusing pids too often */ 340 pid_alloc_lim = pid_tbl_mask - 1; 341 #undef LINK_EMPTY 342 343 proc_specificdata_domain = specificdata_domain_create(); 344 KASSERT(proc_specificdata_domain != NULL); 345 346 proc_cache = pool_cache_init(sizeof(struct proc), 0, 0, 0, 347 "procpl", NULL, IPL_NONE, NULL, NULL, NULL); 348 349 proc_listener = kauth_listen_scope(KAUTH_SCOPE_PROCESS, 350 proc_listener_cb, NULL); 351 } 352 353 void 354 procinit_sysctl(void) 355 { 356 static struct sysctllog *clog; 357 358 sysctl_createv(&clog, 0, NULL, NULL, 359 CTLFLAG_PERMANENT, 360 CTLTYPE_NODE, "kern", NULL, 361 NULL, 0, NULL, 0, 362 CTL_KERN, CTL_EOL); 363 364 sysctl_createv(&clog, 0, NULL, NULL, 365 CTLFLAG_PERMANENT, 366 CTLTYPE_NODE, "proc", 367 SYSCTL_DESCR("System-wide process information"), 368 sysctl_doeproc, 0, NULL, 0, 369 CTL_KERN, KERN_PROC, CTL_EOL); 370 sysctl_createv(&clog, 0, NULL, NULL, 371 CTLFLAG_PERMANENT, 372 CTLTYPE_NODE, "proc2", 373 SYSCTL_DESCR("Machine-independent process information"), 374 sysctl_doeproc, 0, NULL, 0, 375 CTL_KERN, KERN_PROC2, CTL_EOL); 376 sysctl_createv(&clog, 0, NULL, NULL, 377 CTLFLAG_PERMANENT, 378 CTLTYPE_NODE, "proc_args", 379 SYSCTL_DESCR("Process argument information"), 380 sysctl_kern_proc_args, 0, NULL, 0, 381 CTL_KERN, KERN_PROC_ARGS, CTL_EOL); 382 383 /* 384 "nodes" under these: 385 386 KERN_PROC_ALL 387 KERN_PROC_PID pid 388 KERN_PROC_PGRP pgrp 389 KERN_PROC_SESSION sess 390 KERN_PROC_TTY tty 391 KERN_PROC_UID uid 392 KERN_PROC_RUID uid 393 KERN_PROC_GID gid 394 KERN_PROC_RGID gid 395 396 all in all, probably not worth the effort... 397 */ 398 } 399 400 /* 401 * Initialize process 0. 402 */ 403 void 404 proc0_init(void) 405 { 406 struct proc *p; 407 struct pgrp *pg; 408 struct rlimit *rlim; 409 rlim_t lim; 410 int i; 411 412 p = &proc0; 413 pg = &pgrp0; 414 415 mutex_init(&p->p_stmutex, MUTEX_DEFAULT, IPL_HIGH); 416 mutex_init(&p->p_auxlock, MUTEX_DEFAULT, IPL_NONE); 417 p->p_lock = mutex_obj_alloc(MUTEX_DEFAULT, IPL_NONE); 418 419 rw_init(&p->p_reflock); 420 cv_init(&p->p_waitcv, "wait"); 421 cv_init(&p->p_lwpcv, "lwpwait"); 422 423 LIST_INSERT_HEAD(&p->p_lwps, &lwp0, l_sibling); 424 425 pid_table[0].pt_proc = p; 426 LIST_INSERT_HEAD(&allproc, p, p_list); 427 428 pid_table[0].pt_pgrp = pg; 429 LIST_INSERT_HEAD(&pg->pg_members, p, p_pglist); 430 431 #ifdef __HAVE_SYSCALL_INTERN 432 (*p->p_emul->e_syscall_intern)(p); 433 #endif 434 435 /* Create credentials. */ 436 cred0 = kauth_cred_alloc(); 437 p->p_cred = cred0; 438 439 /* Create the CWD info. */ 440 rw_init(&cwdi0.cwdi_lock); 441 442 /* Create the limits structures. */ 443 mutex_init(&limit0.pl_lock, MUTEX_DEFAULT, IPL_NONE); 444 445 rlim = limit0.pl_rlimit; 446 for (i = 0; i < __arraycount(limit0.pl_rlimit); i++) { 447 rlim[i].rlim_cur = RLIM_INFINITY; 448 rlim[i].rlim_max = RLIM_INFINITY; 449 } 450 451 rlim[RLIMIT_NOFILE].rlim_max = maxfiles; 452 rlim[RLIMIT_NOFILE].rlim_cur = maxfiles < nofile ? maxfiles : nofile; 453 454 rlim[RLIMIT_NPROC].rlim_max = maxproc; 455 rlim[RLIMIT_NPROC].rlim_cur = maxproc < maxuprc ? maxproc : maxuprc; 456 457 lim = MIN(VM_MAXUSER_ADDRESS, ctob((rlim_t)uvmexp.free)); 458 rlim[RLIMIT_RSS].rlim_max = lim; 459 rlim[RLIMIT_MEMLOCK].rlim_max = lim; 460 rlim[RLIMIT_MEMLOCK].rlim_cur = lim / 3; 461 462 rlim[RLIMIT_NTHR].rlim_max = maxlwp; 463 rlim[RLIMIT_NTHR].rlim_cur = maxlwp < maxuprc ? maxlwp : maxuprc; 464 465 /* Note that default core name has zero length. */ 466 limit0.pl_corename = defcorename; 467 limit0.pl_cnlen = 0; 468 limit0.pl_refcnt = 1; 469 limit0.pl_writeable = false; 470 limit0.pl_sv_limit = NULL; 471 472 /* Configure virtual memory system, set vm rlimits. */ 473 uvm_init_limits(p); 474 475 /* Initialize file descriptor table for proc0. */ 476 fd_init(&filedesc0); 477 478 /* 479 * Initialize proc0's vmspace, which uses the kernel pmap. 480 * All kernel processes (which never have user space mappings) 481 * share proc0's vmspace, and thus, the kernel pmap. 482 */ 483 uvmspace_init(&vmspace0, pmap_kernel(), round_page(VM_MIN_ADDRESS), 484 trunc_page(VM_MAX_ADDRESS)); 485 486 /* Initialize signal state for proc0. XXX IPL_SCHED */ 487 mutex_init(&p->p_sigacts->sa_mutex, MUTEX_DEFAULT, IPL_SCHED); 488 siginit(p); 489 490 proc_initspecific(p); 491 kdtrace_proc_ctor(NULL, p); 492 } 493 494 /* 495 * Session reference counting. 496 */ 497 498 void 499 proc_sesshold(struct session *ss) 500 { 501 502 KASSERT(mutex_owned(proc_lock)); 503 ss->s_count++; 504 } 505 506 void 507 proc_sessrele(struct session *ss) 508 { 509 510 KASSERT(mutex_owned(proc_lock)); 511 /* 512 * We keep the pgrp with the same id as the session in order to 513 * stop a process being given the same pid. Since the pgrp holds 514 * a reference to the session, it must be a 'zombie' pgrp by now. 515 */ 516 if (--ss->s_count == 0) { 517 struct pgrp *pg; 518 519 pg = pg_remove(ss->s_sid); 520 mutex_exit(proc_lock); 521 522 kmem_free(pg, sizeof(struct pgrp)); 523 kmem_free(ss, sizeof(struct session)); 524 } else { 525 mutex_exit(proc_lock); 526 } 527 } 528 529 /* 530 * Check that the specified process group is in the session of the 531 * specified process. 532 * Treats -ve ids as process ids. 533 * Used to validate TIOCSPGRP requests. 534 */ 535 int 536 pgid_in_session(struct proc *p, pid_t pg_id) 537 { 538 struct pgrp *pgrp; 539 struct session *session; 540 int error; 541 542 mutex_enter(proc_lock); 543 if (pg_id < 0) { 544 struct proc *p1 = proc_find(-pg_id); 545 if (p1 == NULL) { 546 error = EINVAL; 547 goto fail; 548 } 549 pgrp = p1->p_pgrp; 550 } else { 551 pgrp = pgrp_find(pg_id); 552 if (pgrp == NULL) { 553 error = EINVAL; 554 goto fail; 555 } 556 } 557 session = pgrp->pg_session; 558 error = (session != p->p_pgrp->pg_session) ? EPERM : 0; 559 fail: 560 mutex_exit(proc_lock); 561 return error; 562 } 563 564 /* 565 * p_inferior: is p an inferior of q? 566 */ 567 static inline bool 568 p_inferior(struct proc *p, struct proc *q) 569 { 570 571 KASSERT(mutex_owned(proc_lock)); 572 573 for (; p != q; p = p->p_pptr) 574 if (p->p_pid == 0) 575 return false; 576 return true; 577 } 578 579 /* 580 * proc_find: locate a process by the ID. 581 * 582 * => Must be called with proc_lock held. 583 */ 584 proc_t * 585 proc_find_raw(pid_t pid) 586 { 587 struct pid_table *pt; 588 proc_t *p; 589 590 KASSERT(mutex_owned(proc_lock)); 591 pt = &pid_table[pid & pid_tbl_mask]; 592 p = pt->pt_proc; 593 if (__predict_false(!P_VALID(p) || pt->pt_pid != pid)) { 594 return NULL; 595 } 596 return p; 597 } 598 599 proc_t * 600 proc_find(pid_t pid) 601 { 602 proc_t *p; 603 604 p = proc_find_raw(pid); 605 if (__predict_false(p == NULL)) { 606 return NULL; 607 } 608 609 /* 610 * Only allow live processes to be found by PID. 611 * XXX: p_stat might change, since unlocked. 612 */ 613 if (__predict_true(p->p_stat == SACTIVE || p->p_stat == SSTOP)) { 614 return p; 615 } 616 return NULL; 617 } 618 619 /* 620 * pgrp_find: locate a process group by the ID. 621 * 622 * => Must be called with proc_lock held. 623 */ 624 struct pgrp * 625 pgrp_find(pid_t pgid) 626 { 627 struct pgrp *pg; 628 629 KASSERT(mutex_owned(proc_lock)); 630 631 pg = pid_table[pgid & pid_tbl_mask].pt_pgrp; 632 633 /* 634 * Cannot look up a process group that only exists because the 635 * session has not died yet (traditional). 636 */ 637 if (pg == NULL || pg->pg_id != pgid || LIST_EMPTY(&pg->pg_members)) { 638 return NULL; 639 } 640 return pg; 641 } 642 643 static void 644 expand_pid_table(void) 645 { 646 size_t pt_size, tsz; 647 struct pid_table *n_pt, *new_pt; 648 struct proc *proc; 649 struct pgrp *pgrp; 650 pid_t pid, rpid; 651 u_int i; 652 uint new_pt_mask; 653 654 pt_size = pid_tbl_mask + 1; 655 tsz = pt_size * 2 * sizeof(struct pid_table); 656 new_pt = kmem_alloc(tsz, KM_SLEEP); 657 new_pt_mask = pt_size * 2 - 1; 658 659 mutex_enter(proc_lock); 660 if (pt_size != pid_tbl_mask + 1) { 661 /* Another process beat us to it... */ 662 mutex_exit(proc_lock); 663 kmem_free(new_pt, tsz); 664 return; 665 } 666 667 /* 668 * Copy entries from old table into new one. 669 * If 'pid' is 'odd' we need to place in the upper half, 670 * even pid's to the lower half. 671 * Free items stay in the low half so we don't have to 672 * fixup the reference to them. 673 * We stuff free items on the front of the freelist 674 * because we can't write to unmodified entries. 675 * Processing the table backwards maintains a semblance 676 * of issuing pid numbers that increase with time. 677 */ 678 i = pt_size - 1; 679 n_pt = new_pt + i; 680 for (; ; i--, n_pt--) { 681 proc = pid_table[i].pt_proc; 682 pgrp = pid_table[i].pt_pgrp; 683 if (!P_VALID(proc)) { 684 /* Up 'use count' so that link is valid */ 685 pid = (P_NEXT(proc) + pt_size) & ~pt_size; 686 rpid = 0; 687 proc = P_FREE(pid); 688 if (pgrp) 689 pid = pgrp->pg_id; 690 } else { 691 pid = pid_table[i].pt_pid; 692 rpid = pid; 693 } 694 695 /* Save entry in appropriate half of table */ 696 n_pt[pid & pt_size].pt_proc = proc; 697 n_pt[pid & pt_size].pt_pgrp = pgrp; 698 n_pt[pid & pt_size].pt_pid = rpid; 699 700 /* Put other piece on start of free list */ 701 pid = (pid ^ pt_size) & ~pid_tbl_mask; 702 n_pt[pid & pt_size].pt_proc = 703 P_FREE((pid & ~pt_size) | next_free_pt); 704 n_pt[pid & pt_size].pt_pgrp = 0; 705 n_pt[pid & pt_size].pt_pid = 0; 706 707 next_free_pt = i | (pid & pt_size); 708 if (i == 0) 709 break; 710 } 711 712 /* Save old table size and switch tables */ 713 tsz = pt_size * sizeof(struct pid_table); 714 n_pt = pid_table; 715 pid_table = new_pt; 716 pid_tbl_mask = new_pt_mask; 717 718 /* 719 * pid_max starts as PID_MAX (= 30000), once we have 16384 720 * allocated pids we need it to be larger! 721 */ 722 if (pid_tbl_mask > PID_MAX) { 723 pid_max = pid_tbl_mask * 2 + 1; 724 pid_alloc_lim |= pid_alloc_lim << 1; 725 } else 726 pid_alloc_lim <<= 1; /* doubles number of free slots... */ 727 728 mutex_exit(proc_lock); 729 kmem_free(n_pt, tsz); 730 } 731 732 struct proc * 733 proc_alloc(void) 734 { 735 struct proc *p; 736 737 p = pool_cache_get(proc_cache, PR_WAITOK); 738 p->p_stat = SIDL; /* protect against others */ 739 proc_initspecific(p); 740 kdtrace_proc_ctor(NULL, p); 741 p->p_pid = -1; 742 proc_alloc_pid(p); 743 return p; 744 } 745 746 /* 747 * proc_alloc_pid: allocate PID and record the given proc 'p' so that 748 * proc_find_raw() can find it by the PID. 749 */ 750 751 pid_t 752 proc_alloc_pid(struct proc *p) 753 { 754 struct pid_table *pt; 755 pid_t pid; 756 int nxt; 757 758 for (;;expand_pid_table()) { 759 if (__predict_false(pid_alloc_cnt >= pid_alloc_lim)) 760 /* ensure pids cycle through 2000+ values */ 761 continue; 762 mutex_enter(proc_lock); 763 pt = &pid_table[next_free_pt]; 764 #ifdef DIAGNOSTIC 765 if (__predict_false(P_VALID(pt->pt_proc) || pt->pt_pgrp)) 766 panic("proc_alloc: slot busy"); 767 #endif 768 nxt = P_NEXT(pt->pt_proc); 769 if (nxt & pid_tbl_mask) 770 break; 771 /* Table full - expand (NB last entry not used....) */ 772 mutex_exit(proc_lock); 773 } 774 775 /* pid is 'saved use count' + 'size' + entry */ 776 pid = (nxt & ~pid_tbl_mask) + pid_tbl_mask + 1 + next_free_pt; 777 if ((uint)pid > (uint)pid_max) 778 pid &= pid_tbl_mask; 779 next_free_pt = nxt & pid_tbl_mask; 780 781 /* Grab table slot */ 782 pt->pt_proc = p; 783 784 KASSERT(pt->pt_pid == 0); 785 pt->pt_pid = pid; 786 if (p->p_pid == -1) { 787 p->p_pid = pid; 788 } 789 pid_alloc_cnt++; 790 mutex_exit(proc_lock); 791 792 return pid; 793 } 794 795 /* 796 * Free a process id - called from proc_free (in kern_exit.c) 797 * 798 * Called with the proc_lock held. 799 */ 800 void 801 proc_free_pid(pid_t pid) 802 { 803 struct pid_table *pt; 804 805 KASSERT(mutex_owned(proc_lock)); 806 807 pt = &pid_table[pid & pid_tbl_mask]; 808 809 /* save pid use count in slot */ 810 pt->pt_proc = P_FREE(pid & ~pid_tbl_mask); 811 KASSERT(pt->pt_pid == pid); 812 pt->pt_pid = 0; 813 814 if (pt->pt_pgrp == NULL) { 815 /* link last freed entry onto ours */ 816 pid &= pid_tbl_mask; 817 pt = &pid_table[last_free_pt]; 818 pt->pt_proc = P_FREE(P_NEXT(pt->pt_proc) | pid); 819 pt->pt_pid = 0; 820 last_free_pt = pid; 821 pid_alloc_cnt--; 822 } 823 824 atomic_dec_uint(&nprocs); 825 } 826 827 void 828 proc_free_mem(struct proc *p) 829 { 830 831 kdtrace_proc_dtor(NULL, p); 832 pool_cache_put(proc_cache, p); 833 } 834 835 /* 836 * proc_enterpgrp: move p to a new or existing process group (and session). 837 * 838 * If we are creating a new pgrp, the pgid should equal 839 * the calling process' pid. 840 * If is only valid to enter a process group that is in the session 841 * of the process. 842 * Also mksess should only be set if we are creating a process group 843 * 844 * Only called from sys_setsid, sys_setpgid and posix_spawn/spawn_return. 845 */ 846 int 847 proc_enterpgrp(struct proc *curp, pid_t pid, pid_t pgid, bool mksess) 848 { 849 struct pgrp *new_pgrp, *pgrp; 850 struct session *sess; 851 struct proc *p; 852 int rval; 853 pid_t pg_id = NO_PGID; 854 855 sess = mksess ? kmem_alloc(sizeof(*sess), KM_SLEEP) : NULL; 856 857 /* Allocate data areas we might need before doing any validity checks */ 858 mutex_enter(proc_lock); /* Because pid_table might change */ 859 if (pid_table[pgid & pid_tbl_mask].pt_pgrp == 0) { 860 mutex_exit(proc_lock); 861 new_pgrp = kmem_alloc(sizeof(*new_pgrp), KM_SLEEP); 862 mutex_enter(proc_lock); 863 } else 864 new_pgrp = NULL; 865 rval = EPERM; /* most common error (to save typing) */ 866 867 /* Check pgrp exists or can be created */ 868 pgrp = pid_table[pgid & pid_tbl_mask].pt_pgrp; 869 if (pgrp != NULL && pgrp->pg_id != pgid) 870 goto done; 871 872 /* Can only set another process under restricted circumstances. */ 873 if (pid != curp->p_pid) { 874 /* Must exist and be one of our children... */ 875 p = proc_find(pid); 876 if (p == NULL || !p_inferior(p, curp)) { 877 rval = ESRCH; 878 goto done; 879 } 880 /* ... in the same session... */ 881 if (sess != NULL || p->p_session != curp->p_session) 882 goto done; 883 /* ... existing pgid must be in same session ... */ 884 if (pgrp != NULL && pgrp->pg_session != p->p_session) 885 goto done; 886 /* ... and not done an exec. */ 887 if (p->p_flag & PK_EXEC) { 888 rval = EACCES; 889 goto done; 890 } 891 } else { 892 /* ... setsid() cannot re-enter a pgrp */ 893 if (mksess && (curp->p_pgid == curp->p_pid || 894 pgrp_find(curp->p_pid))) 895 goto done; 896 p = curp; 897 } 898 899 /* Changing the process group/session of a session 900 leader is definitely off limits. */ 901 if (SESS_LEADER(p)) { 902 if (sess == NULL && p->p_pgrp == pgrp) 903 /* unless it's a definite noop */ 904 rval = 0; 905 goto done; 906 } 907 908 /* Can only create a process group with id of process */ 909 if (pgrp == NULL && pgid != pid) 910 goto done; 911 912 /* Can only create a session if creating pgrp */ 913 if (sess != NULL && pgrp != NULL) 914 goto done; 915 916 /* Check we allocated memory for a pgrp... */ 917 if (pgrp == NULL && new_pgrp == NULL) 918 goto done; 919 920 /* Don't attach to 'zombie' pgrp */ 921 if (pgrp != NULL && LIST_EMPTY(&pgrp->pg_members)) 922 goto done; 923 924 /* Expect to succeed now */ 925 rval = 0; 926 927 if (pgrp == p->p_pgrp) 928 /* nothing to do */ 929 goto done; 930 931 /* Ok all setup, link up required structures */ 932 933 if (pgrp == NULL) { 934 pgrp = new_pgrp; 935 new_pgrp = NULL; 936 if (sess != NULL) { 937 sess->s_sid = p->p_pid; 938 sess->s_leader = p; 939 sess->s_count = 1; 940 sess->s_ttyvp = NULL; 941 sess->s_ttyp = NULL; 942 sess->s_flags = p->p_session->s_flags & ~S_LOGIN_SET; 943 memcpy(sess->s_login, p->p_session->s_login, 944 sizeof(sess->s_login)); 945 p->p_lflag &= ~PL_CONTROLT; 946 } else { 947 sess = p->p_pgrp->pg_session; 948 proc_sesshold(sess); 949 } 950 pgrp->pg_session = sess; 951 sess = NULL; 952 953 pgrp->pg_id = pgid; 954 LIST_INIT(&pgrp->pg_members); 955 #ifdef DIAGNOSTIC 956 if (__predict_false(pid_table[pgid & pid_tbl_mask].pt_pgrp)) 957 panic("enterpgrp: pgrp table slot in use"); 958 if (__predict_false(mksess && p != curp)) 959 panic("enterpgrp: mksession and p != curproc"); 960 #endif 961 pid_table[pgid & pid_tbl_mask].pt_pgrp = pgrp; 962 pgrp->pg_jobc = 0; 963 } 964 965 /* 966 * Adjust eligibility of affected pgrps to participate in job control. 967 * Increment eligibility counts before decrementing, otherwise we 968 * could reach 0 spuriously during the first call. 969 */ 970 fixjobc(p, pgrp, 1); 971 fixjobc(p, p->p_pgrp, 0); 972 973 /* Interlock with ttread(). */ 974 mutex_spin_enter(&tty_lock); 975 976 /* Move process to requested group. */ 977 LIST_REMOVE(p, p_pglist); 978 if (LIST_EMPTY(&p->p_pgrp->pg_members)) 979 /* defer delete until we've dumped the lock */ 980 pg_id = p->p_pgrp->pg_id; 981 p->p_pgrp = pgrp; 982 LIST_INSERT_HEAD(&pgrp->pg_members, p, p_pglist); 983 984 /* Done with the swap; we can release the tty mutex. */ 985 mutex_spin_exit(&tty_lock); 986 987 done: 988 if (pg_id != NO_PGID) { 989 /* Releases proc_lock. */ 990 pg_delete(pg_id); 991 } else { 992 mutex_exit(proc_lock); 993 } 994 if (sess != NULL) 995 kmem_free(sess, sizeof(*sess)); 996 if (new_pgrp != NULL) 997 kmem_free(new_pgrp, sizeof(*new_pgrp)); 998 #ifdef DEBUG_PGRP 999 if (__predict_false(rval)) 1000 printf("enterpgrp(%d,%d,%d), curproc %d, rval %d\n", 1001 pid, pgid, mksess, curp->p_pid, rval); 1002 #endif 1003 return rval; 1004 } 1005 1006 /* 1007 * proc_leavepgrp: remove a process from its process group. 1008 * => must be called with the proc_lock held, which will be released; 1009 */ 1010 void 1011 proc_leavepgrp(struct proc *p) 1012 { 1013 struct pgrp *pgrp; 1014 1015 KASSERT(mutex_owned(proc_lock)); 1016 1017 /* Interlock with ttread() */ 1018 mutex_spin_enter(&tty_lock); 1019 pgrp = p->p_pgrp; 1020 LIST_REMOVE(p, p_pglist); 1021 p->p_pgrp = NULL; 1022 mutex_spin_exit(&tty_lock); 1023 1024 if (LIST_EMPTY(&pgrp->pg_members)) { 1025 /* Releases proc_lock. */ 1026 pg_delete(pgrp->pg_id); 1027 } else { 1028 mutex_exit(proc_lock); 1029 } 1030 } 1031 1032 /* 1033 * pg_remove: remove a process group from the table. 1034 * => must be called with the proc_lock held; 1035 * => returns process group to free; 1036 */ 1037 static struct pgrp * 1038 pg_remove(pid_t pg_id) 1039 { 1040 struct pgrp *pgrp; 1041 struct pid_table *pt; 1042 1043 KASSERT(mutex_owned(proc_lock)); 1044 1045 pt = &pid_table[pg_id & pid_tbl_mask]; 1046 pgrp = pt->pt_pgrp; 1047 1048 KASSERT(pgrp != NULL); 1049 KASSERT(pgrp->pg_id == pg_id); 1050 KASSERT(LIST_EMPTY(&pgrp->pg_members)); 1051 1052 pt->pt_pgrp = NULL; 1053 1054 if (!P_VALID(pt->pt_proc)) { 1055 /* Orphaned pgrp, put slot onto free list. */ 1056 KASSERT((P_NEXT(pt->pt_proc) & pid_tbl_mask) == 0); 1057 pg_id &= pid_tbl_mask; 1058 pt = &pid_table[last_free_pt]; 1059 pt->pt_proc = P_FREE(P_NEXT(pt->pt_proc) | pg_id); 1060 KASSERT(pt->pt_pid == 0); 1061 last_free_pt = pg_id; 1062 pid_alloc_cnt--; 1063 } 1064 return pgrp; 1065 } 1066 1067 /* 1068 * pg_delete: delete and free a process group. 1069 * => must be called with the proc_lock held, which will be released. 1070 */ 1071 static void 1072 pg_delete(pid_t pg_id) 1073 { 1074 struct pgrp *pg; 1075 struct tty *ttyp; 1076 struct session *ss; 1077 1078 KASSERT(mutex_owned(proc_lock)); 1079 1080 pg = pid_table[pg_id & pid_tbl_mask].pt_pgrp; 1081 if (pg == NULL || pg->pg_id != pg_id || !LIST_EMPTY(&pg->pg_members)) { 1082 mutex_exit(proc_lock); 1083 return; 1084 } 1085 1086 ss = pg->pg_session; 1087 1088 /* Remove reference (if any) from tty to this process group */ 1089 mutex_spin_enter(&tty_lock); 1090 ttyp = ss->s_ttyp; 1091 if (ttyp != NULL && ttyp->t_pgrp == pg) { 1092 ttyp->t_pgrp = NULL; 1093 KASSERT(ttyp->t_session == ss); 1094 } 1095 mutex_spin_exit(&tty_lock); 1096 1097 /* 1098 * The leading process group in a session is freed by proc_sessrele(), 1099 * if last reference. Note: proc_sessrele() releases proc_lock. 1100 */ 1101 pg = (ss->s_sid != pg->pg_id) ? pg_remove(pg_id) : NULL; 1102 proc_sessrele(ss); 1103 1104 if (pg != NULL) { 1105 /* Free it, if was not done by proc_sessrele(). */ 1106 kmem_free(pg, sizeof(struct pgrp)); 1107 } 1108 } 1109 1110 /* 1111 * Adjust pgrp jobc counters when specified process changes process group. 1112 * We count the number of processes in each process group that "qualify" 1113 * the group for terminal job control (those with a parent in a different 1114 * process group of the same session). If that count reaches zero, the 1115 * process group becomes orphaned. Check both the specified process' 1116 * process group and that of its children. 1117 * entering == 0 => p is leaving specified group. 1118 * entering == 1 => p is entering specified group. 1119 * 1120 * Call with proc_lock held. 1121 */ 1122 void 1123 fixjobc(struct proc *p, struct pgrp *pgrp, int entering) 1124 { 1125 struct pgrp *hispgrp; 1126 struct session *mysession = pgrp->pg_session; 1127 struct proc *child; 1128 1129 KASSERT(mutex_owned(proc_lock)); 1130 1131 /* 1132 * Check p's parent to see whether p qualifies its own process 1133 * group; if so, adjust count for p's process group. 1134 */ 1135 hispgrp = p->p_pptr->p_pgrp; 1136 if (hispgrp != pgrp && hispgrp->pg_session == mysession) { 1137 if (entering) { 1138 pgrp->pg_jobc++; 1139 p->p_lflag &= ~PL_ORPHANPG; 1140 } else if (--pgrp->pg_jobc == 0) 1141 orphanpg(pgrp); 1142 } 1143 1144 /* 1145 * Check this process' children to see whether they qualify 1146 * their process groups; if so, adjust counts for children's 1147 * process groups. 1148 */ 1149 LIST_FOREACH(child, &p->p_children, p_sibling) { 1150 hispgrp = child->p_pgrp; 1151 if (hispgrp != pgrp && hispgrp->pg_session == mysession && 1152 !P_ZOMBIE(child)) { 1153 if (entering) { 1154 child->p_lflag &= ~PL_ORPHANPG; 1155 hispgrp->pg_jobc++; 1156 } else if (--hispgrp->pg_jobc == 0) 1157 orphanpg(hispgrp); 1158 } 1159 } 1160 } 1161 1162 /* 1163 * A process group has become orphaned; 1164 * if there are any stopped processes in the group, 1165 * hang-up all process in that group. 1166 * 1167 * Call with proc_lock held. 1168 */ 1169 static void 1170 orphanpg(struct pgrp *pg) 1171 { 1172 struct proc *p; 1173 1174 KASSERT(mutex_owned(proc_lock)); 1175 1176 LIST_FOREACH(p, &pg->pg_members, p_pglist) { 1177 if (p->p_stat == SSTOP) { 1178 p->p_lflag |= PL_ORPHANPG; 1179 psignal(p, SIGHUP); 1180 psignal(p, SIGCONT); 1181 } 1182 } 1183 } 1184 1185 #ifdef DDB 1186 #include <ddb/db_output.h> 1187 void pidtbl_dump(void); 1188 void 1189 pidtbl_dump(void) 1190 { 1191 struct pid_table *pt; 1192 struct proc *p; 1193 struct pgrp *pgrp; 1194 int id; 1195 1196 db_printf("pid table %p size %x, next %x, last %x\n", 1197 pid_table, pid_tbl_mask+1, 1198 next_free_pt, last_free_pt); 1199 for (pt = pid_table, id = 0; id <= pid_tbl_mask; id++, pt++) { 1200 p = pt->pt_proc; 1201 if (!P_VALID(p) && !pt->pt_pgrp) 1202 continue; 1203 db_printf(" id %x: ", id); 1204 if (P_VALID(p)) 1205 db_printf("slotpid %d proc %p id %d (0x%x) %s\n", 1206 pt->pt_pid, p, p->p_pid, p->p_pid, p->p_comm); 1207 else 1208 db_printf("next %x use %x\n", 1209 P_NEXT(p) & pid_tbl_mask, 1210 P_NEXT(p) & ~pid_tbl_mask); 1211 if ((pgrp = pt->pt_pgrp)) { 1212 db_printf("\tsession %p, sid %d, count %d, login %s\n", 1213 pgrp->pg_session, pgrp->pg_session->s_sid, 1214 pgrp->pg_session->s_count, 1215 pgrp->pg_session->s_login); 1216 db_printf("\tpgrp %p, pg_id %d, pg_jobc %d, members %p\n", 1217 pgrp, pgrp->pg_id, pgrp->pg_jobc, 1218 LIST_FIRST(&pgrp->pg_members)); 1219 LIST_FOREACH(p, &pgrp->pg_members, p_pglist) { 1220 db_printf("\t\tpid %d addr %p pgrp %p %s\n", 1221 p->p_pid, p, p->p_pgrp, p->p_comm); 1222 } 1223 } 1224 } 1225 } 1226 #endif /* DDB */ 1227 1228 #ifdef KSTACK_CHECK_MAGIC 1229 1230 #define KSTACK_MAGIC 0xdeadbeaf 1231 1232 /* XXX should be per process basis? */ 1233 static int kstackleftmin = KSTACK_SIZE; 1234 static int kstackleftthres = KSTACK_SIZE / 8; 1235 1236 void 1237 kstack_setup_magic(const struct lwp *l) 1238 { 1239 uint32_t *ip; 1240 uint32_t const *end; 1241 1242 KASSERT(l != NULL); 1243 KASSERT(l != &lwp0); 1244 1245 /* 1246 * fill all the stack with magic number 1247 * so that later modification on it can be detected. 1248 */ 1249 ip = (uint32_t *)KSTACK_LOWEST_ADDR(l); 1250 end = (uint32_t *)((char *)KSTACK_LOWEST_ADDR(l) + KSTACK_SIZE); 1251 for (; ip < end; ip++) { 1252 *ip = KSTACK_MAGIC; 1253 } 1254 } 1255 1256 void 1257 kstack_check_magic(const struct lwp *l) 1258 { 1259 uint32_t const *ip, *end; 1260 int stackleft; 1261 1262 KASSERT(l != NULL); 1263 1264 /* don't check proc0 */ /*XXX*/ 1265 if (l == &lwp0) 1266 return; 1267 1268 #ifdef __MACHINE_STACK_GROWS_UP 1269 /* stack grows upwards (eg. hppa) */ 1270 ip = (uint32_t *)((void *)KSTACK_LOWEST_ADDR(l) + KSTACK_SIZE); 1271 end = (uint32_t *)KSTACK_LOWEST_ADDR(l); 1272 for (ip--; ip >= end; ip--) 1273 if (*ip != KSTACK_MAGIC) 1274 break; 1275 1276 stackleft = (void *)KSTACK_LOWEST_ADDR(l) + KSTACK_SIZE - (void *)ip; 1277 #else /* __MACHINE_STACK_GROWS_UP */ 1278 /* stack grows downwards (eg. i386) */ 1279 ip = (uint32_t *)KSTACK_LOWEST_ADDR(l); 1280 end = (uint32_t *)((char *)KSTACK_LOWEST_ADDR(l) + KSTACK_SIZE); 1281 for (; ip < end; ip++) 1282 if (*ip != KSTACK_MAGIC) 1283 break; 1284 1285 stackleft = ((const char *)ip) - (const char *)KSTACK_LOWEST_ADDR(l); 1286 #endif /* __MACHINE_STACK_GROWS_UP */ 1287 1288 if (kstackleftmin > stackleft) { 1289 kstackleftmin = stackleft; 1290 if (stackleft < kstackleftthres) 1291 printf("warning: kernel stack left %d bytes" 1292 "(pid %u:lid %u)\n", stackleft, 1293 (u_int)l->l_proc->p_pid, (u_int)l->l_lid); 1294 } 1295 1296 if (stackleft <= 0) { 1297 panic("magic on the top of kernel stack changed for " 1298 "pid %u, lid %u: maybe kernel stack overflow", 1299 (u_int)l->l_proc->p_pid, (u_int)l->l_lid); 1300 } 1301 } 1302 #endif /* KSTACK_CHECK_MAGIC */ 1303 1304 int 1305 proclist_foreach_call(struct proclist *list, 1306 int (*callback)(struct proc *, void *arg), void *arg) 1307 { 1308 struct proc marker; 1309 struct proc *p; 1310 int ret = 0; 1311 1312 marker.p_flag = PK_MARKER; 1313 mutex_enter(proc_lock); 1314 for (p = LIST_FIRST(list); ret == 0 && p != NULL;) { 1315 if (p->p_flag & PK_MARKER) { 1316 p = LIST_NEXT(p, p_list); 1317 continue; 1318 } 1319 LIST_INSERT_AFTER(p, &marker, p_list); 1320 ret = (*callback)(p, arg); 1321 KASSERT(mutex_owned(proc_lock)); 1322 p = LIST_NEXT(&marker, p_list); 1323 LIST_REMOVE(&marker, p_list); 1324 } 1325 mutex_exit(proc_lock); 1326 1327 return ret; 1328 } 1329 1330 int 1331 proc_vmspace_getref(struct proc *p, struct vmspace **vm) 1332 { 1333 1334 /* XXXCDC: how should locking work here? */ 1335 1336 /* curproc exception is for coredump. */ 1337 1338 if ((p != curproc && (p->p_sflag & PS_WEXIT) != 0) || 1339 (p->p_vmspace->vm_refcnt < 1)) { /* XXX */ 1340 return EFAULT; 1341 } 1342 1343 uvmspace_addref(p->p_vmspace); 1344 *vm = p->p_vmspace; 1345 1346 return 0; 1347 } 1348 1349 /* 1350 * Acquire a write lock on the process credential. 1351 */ 1352 void 1353 proc_crmod_enter(void) 1354 { 1355 struct lwp *l = curlwp; 1356 struct proc *p = l->l_proc; 1357 kauth_cred_t oc; 1358 1359 /* Reset what needs to be reset in plimit. */ 1360 if (p->p_limit->pl_corename != defcorename) { 1361 lim_setcorename(p, defcorename, 0); 1362 } 1363 1364 mutex_enter(p->p_lock); 1365 1366 /* Ensure the LWP cached credentials are up to date. */ 1367 if ((oc = l->l_cred) != p->p_cred) { 1368 kauth_cred_hold(p->p_cred); 1369 l->l_cred = p->p_cred; 1370 kauth_cred_free(oc); 1371 } 1372 } 1373 1374 /* 1375 * Set in a new process credential, and drop the write lock. The credential 1376 * must have a reference already. Optionally, free a no-longer required 1377 * credential. The scheduler also needs to inspect p_cred, so we also 1378 * briefly acquire the sched state mutex. 1379 */ 1380 void 1381 proc_crmod_leave(kauth_cred_t scred, kauth_cred_t fcred, bool sugid) 1382 { 1383 struct lwp *l = curlwp, *l2; 1384 struct proc *p = l->l_proc; 1385 kauth_cred_t oc; 1386 1387 KASSERT(mutex_owned(p->p_lock)); 1388 1389 /* Is there a new credential to set in? */ 1390 if (scred != NULL) { 1391 p->p_cred = scred; 1392 LIST_FOREACH(l2, &p->p_lwps, l_sibling) { 1393 if (l2 != l) 1394 l2->l_prflag |= LPR_CRMOD; 1395 } 1396 1397 /* Ensure the LWP cached credentials are up to date. */ 1398 if ((oc = l->l_cred) != scred) { 1399 kauth_cred_hold(scred); 1400 l->l_cred = scred; 1401 } 1402 } else 1403 oc = NULL; /* XXXgcc */ 1404 1405 if (sugid) { 1406 /* 1407 * Mark process as having changed credentials, stops 1408 * tracing etc. 1409 */ 1410 p->p_flag |= PK_SUGID; 1411 } 1412 1413 mutex_exit(p->p_lock); 1414 1415 /* If there is a credential to be released, free it now. */ 1416 if (fcred != NULL) { 1417 KASSERT(scred != NULL); 1418 kauth_cred_free(fcred); 1419 if (oc != scred) 1420 kauth_cred_free(oc); 1421 } 1422 } 1423 1424 /* 1425 * proc_specific_key_create -- 1426 * Create a key for subsystem proc-specific data. 1427 */ 1428 int 1429 proc_specific_key_create(specificdata_key_t *keyp, specificdata_dtor_t dtor) 1430 { 1431 1432 return (specificdata_key_create(proc_specificdata_domain, keyp, dtor)); 1433 } 1434 1435 /* 1436 * proc_specific_key_delete -- 1437 * Delete a key for subsystem proc-specific data. 1438 */ 1439 void 1440 proc_specific_key_delete(specificdata_key_t key) 1441 { 1442 1443 specificdata_key_delete(proc_specificdata_domain, key); 1444 } 1445 1446 /* 1447 * proc_initspecific -- 1448 * Initialize a proc's specificdata container. 1449 */ 1450 void 1451 proc_initspecific(struct proc *p) 1452 { 1453 int error; 1454 1455 error = specificdata_init(proc_specificdata_domain, &p->p_specdataref); 1456 KASSERT(error == 0); 1457 } 1458 1459 /* 1460 * proc_finispecific -- 1461 * Finalize a proc's specificdata container. 1462 */ 1463 void 1464 proc_finispecific(struct proc *p) 1465 { 1466 1467 specificdata_fini(proc_specificdata_domain, &p->p_specdataref); 1468 } 1469 1470 /* 1471 * proc_getspecific -- 1472 * Return proc-specific data corresponding to the specified key. 1473 */ 1474 void * 1475 proc_getspecific(struct proc *p, specificdata_key_t key) 1476 { 1477 1478 return (specificdata_getspecific(proc_specificdata_domain, 1479 &p->p_specdataref, key)); 1480 } 1481 1482 /* 1483 * proc_setspecific -- 1484 * Set proc-specific data corresponding to the specified key. 1485 */ 1486 void 1487 proc_setspecific(struct proc *p, specificdata_key_t key, void *data) 1488 { 1489 1490 specificdata_setspecific(proc_specificdata_domain, 1491 &p->p_specdataref, key, data); 1492 } 1493 1494 int 1495 proc_uidmatch(kauth_cred_t cred, kauth_cred_t target) 1496 { 1497 int r = 0; 1498 1499 if (kauth_cred_getuid(cred) != kauth_cred_getuid(target) || 1500 kauth_cred_getuid(cred) != kauth_cred_getsvuid(target)) { 1501 /* 1502 * suid proc of ours or proc not ours 1503 */ 1504 r = EPERM; 1505 } else if (kauth_cred_getgid(target) != kauth_cred_getsvgid(target)) { 1506 /* 1507 * sgid proc has sgid back to us temporarily 1508 */ 1509 r = EPERM; 1510 } else { 1511 /* 1512 * our rgid must be in target's group list (ie, 1513 * sub-processes started by a sgid process) 1514 */ 1515 int ismember = 0; 1516 1517 if (kauth_cred_ismember_gid(cred, 1518 kauth_cred_getgid(target), &ismember) != 0 || 1519 !ismember) 1520 r = EPERM; 1521 } 1522 1523 return (r); 1524 } 1525 1526 /* 1527 * sysctl stuff 1528 */ 1529 1530 #define KERN_PROCSLOP (5 * sizeof(struct kinfo_proc)) 1531 1532 static const u_int sysctl_flagmap[] = { 1533 PK_ADVLOCK, P_ADVLOCK, 1534 PK_EXEC, P_EXEC, 1535 PK_NOCLDWAIT, P_NOCLDWAIT, 1536 PK_32, P_32, 1537 PK_CLDSIGIGN, P_CLDSIGIGN, 1538 PK_SUGID, P_SUGID, 1539 0 1540 }; 1541 1542 static const u_int sysctl_sflagmap[] = { 1543 PS_NOCLDSTOP, P_NOCLDSTOP, 1544 PS_WEXIT, P_WEXIT, 1545 PS_STOPFORK, P_STOPFORK, 1546 PS_STOPEXEC, P_STOPEXEC, 1547 PS_STOPEXIT, P_STOPEXIT, 1548 0 1549 }; 1550 1551 static const u_int sysctl_slflagmap[] = { 1552 PSL_TRACED, P_TRACED, 1553 PSL_FSTRACE, P_FSTRACE, 1554 PSL_CHTRACED, P_CHTRACED, 1555 PSL_SYSCALL, P_SYSCALL, 1556 0 1557 }; 1558 1559 static const u_int sysctl_lflagmap[] = { 1560 PL_CONTROLT, P_CONTROLT, 1561 PL_PPWAIT, P_PPWAIT, 1562 0 1563 }; 1564 1565 static const u_int sysctl_stflagmap[] = { 1566 PST_PROFIL, P_PROFIL, 1567 0 1568 1569 }; 1570 1571 /* used by kern_lwp also */ 1572 const u_int sysctl_lwpflagmap[] = { 1573 LW_SINTR, L_SINTR, 1574 LW_SYSTEM, L_SYSTEM, 1575 0 1576 }; 1577 1578 /* 1579 * Find the most ``active'' lwp of a process and return it for ps display 1580 * purposes 1581 */ 1582 static struct lwp * 1583 proc_active_lwp(struct proc *p) 1584 { 1585 static const int ostat[] = { 1586 0, 1587 2, /* LSIDL */ 1588 6, /* LSRUN */ 1589 5, /* LSSLEEP */ 1590 4, /* LSSTOP */ 1591 0, /* LSZOMB */ 1592 1, /* LSDEAD */ 1593 7, /* LSONPROC */ 1594 3 /* LSSUSPENDED */ 1595 }; 1596 1597 struct lwp *l, *lp = NULL; 1598 LIST_FOREACH(l, &p->p_lwps, l_sibling) { 1599 KASSERT(l->l_stat >= 0 && l->l_stat < __arraycount(ostat)); 1600 if (lp == NULL || 1601 ostat[l->l_stat] > ostat[lp->l_stat] || 1602 (ostat[l->l_stat] == ostat[lp->l_stat] && 1603 l->l_cpticks > lp->l_cpticks)) { 1604 lp = l; 1605 continue; 1606 } 1607 } 1608 return lp; 1609 } 1610 1611 static int 1612 sysctl_doeproc(SYSCTLFN_ARGS) 1613 { 1614 union { 1615 struct kinfo_proc kproc; 1616 struct kinfo_proc2 kproc2; 1617 } *kbuf; 1618 struct proc *p, *next, *marker; 1619 char *where, *dp; 1620 int type, op, arg, error; 1621 u_int elem_size, kelem_size, elem_count; 1622 size_t buflen, needed; 1623 bool match, zombie, mmmbrains; 1624 1625 if (namelen == 1 && name[0] == CTL_QUERY) 1626 return (sysctl_query(SYSCTLFN_CALL(rnode))); 1627 1628 dp = where = oldp; 1629 buflen = where != NULL ? *oldlenp : 0; 1630 error = 0; 1631 needed = 0; 1632 type = rnode->sysctl_num; 1633 1634 if (type == KERN_PROC) { 1635 if (namelen != 2 && !(namelen == 1 && name[0] == KERN_PROC_ALL)) 1636 return (EINVAL); 1637 op = name[0]; 1638 if (op != KERN_PROC_ALL) 1639 arg = name[1]; 1640 else 1641 arg = 0; /* Quell compiler warning */ 1642 elem_count = 0; /* Ditto */ 1643 kelem_size = elem_size = sizeof(kbuf->kproc); 1644 } else { 1645 if (namelen != 4) 1646 return (EINVAL); 1647 op = name[0]; 1648 arg = name[1]; 1649 elem_size = name[2]; 1650 elem_count = name[3]; 1651 kelem_size = sizeof(kbuf->kproc2); 1652 } 1653 1654 sysctl_unlock(); 1655 1656 kbuf = kmem_alloc(sizeof(*kbuf), KM_SLEEP); 1657 marker = kmem_alloc(sizeof(*marker), KM_SLEEP); 1658 marker->p_flag = PK_MARKER; 1659 1660 mutex_enter(proc_lock); 1661 mmmbrains = false; 1662 for (p = LIST_FIRST(&allproc);; p = next) { 1663 if (p == NULL) { 1664 if (!mmmbrains) { 1665 p = LIST_FIRST(&zombproc); 1666 mmmbrains = true; 1667 } 1668 if (p == NULL) 1669 break; 1670 } 1671 next = LIST_NEXT(p, p_list); 1672 if ((p->p_flag & PK_MARKER) != 0) 1673 continue; 1674 1675 /* 1676 * Skip embryonic processes. 1677 */ 1678 if (p->p_stat == SIDL) 1679 continue; 1680 1681 mutex_enter(p->p_lock); 1682 error = kauth_authorize_process(l->l_cred, 1683 KAUTH_PROCESS_CANSEE, p, 1684 KAUTH_ARG(KAUTH_REQ_PROCESS_CANSEE_ENTRY), NULL, NULL); 1685 if (error != 0) { 1686 mutex_exit(p->p_lock); 1687 continue; 1688 } 1689 1690 /* 1691 * TODO - make more efficient (see notes below). 1692 * do by session. 1693 */ 1694 switch (op) { 1695 case KERN_PROC_PID: 1696 /* could do this with just a lookup */ 1697 match = (p->p_pid == (pid_t)arg); 1698 break; 1699 1700 case KERN_PROC_PGRP: 1701 /* could do this by traversing pgrp */ 1702 match = (p->p_pgrp->pg_id == (pid_t)arg); 1703 break; 1704 1705 case KERN_PROC_SESSION: 1706 match = (p->p_session->s_sid == (pid_t)arg); 1707 break; 1708 1709 case KERN_PROC_TTY: 1710 match = true; 1711 if (arg == (int) KERN_PROC_TTY_REVOKE) { 1712 if ((p->p_lflag & PL_CONTROLT) == 0 || 1713 p->p_session->s_ttyp == NULL || 1714 p->p_session->s_ttyvp != NULL) { 1715 match = false; 1716 } 1717 } else if ((p->p_lflag & PL_CONTROLT) == 0 || 1718 p->p_session->s_ttyp == NULL) { 1719 if ((dev_t)arg != KERN_PROC_TTY_NODEV) { 1720 match = false; 1721 } 1722 } else if (p->p_session->s_ttyp->t_dev != (dev_t)arg) { 1723 match = false; 1724 } 1725 break; 1726 1727 case KERN_PROC_UID: 1728 match = (kauth_cred_geteuid(p->p_cred) == (uid_t)arg); 1729 break; 1730 1731 case KERN_PROC_RUID: 1732 match = (kauth_cred_getuid(p->p_cred) == (uid_t)arg); 1733 break; 1734 1735 case KERN_PROC_GID: 1736 match = (kauth_cred_getegid(p->p_cred) == (uid_t)arg); 1737 break; 1738 1739 case KERN_PROC_RGID: 1740 match = (kauth_cred_getgid(p->p_cred) == (uid_t)arg); 1741 break; 1742 1743 case KERN_PROC_ALL: 1744 match = true; 1745 /* allow everything */ 1746 break; 1747 1748 default: 1749 error = EINVAL; 1750 mutex_exit(p->p_lock); 1751 goto cleanup; 1752 } 1753 if (!match) { 1754 mutex_exit(p->p_lock); 1755 continue; 1756 } 1757 1758 /* 1759 * Grab a hold on the process. 1760 */ 1761 if (mmmbrains) { 1762 zombie = true; 1763 } else { 1764 zombie = !rw_tryenter(&p->p_reflock, RW_READER); 1765 } 1766 if (zombie) { 1767 LIST_INSERT_AFTER(p, marker, p_list); 1768 } 1769 1770 if (buflen >= elem_size && 1771 (type == KERN_PROC || elem_count > 0)) { 1772 if (type == KERN_PROC) { 1773 kbuf->kproc.kp_proc = *p; 1774 fill_eproc(p, &kbuf->kproc.kp_eproc, zombie); 1775 } else { 1776 fill_kproc2(p, &kbuf->kproc2, zombie); 1777 elem_count--; 1778 } 1779 mutex_exit(p->p_lock); 1780 mutex_exit(proc_lock); 1781 /* 1782 * Copy out elem_size, but not larger than kelem_size 1783 */ 1784 error = sysctl_copyout(l, kbuf, dp, 1785 min(kelem_size, elem_size)); 1786 mutex_enter(proc_lock); 1787 if (error) { 1788 goto bah; 1789 } 1790 dp += elem_size; 1791 buflen -= elem_size; 1792 } else { 1793 mutex_exit(p->p_lock); 1794 } 1795 needed += elem_size; 1796 1797 /* 1798 * Release reference to process. 1799 */ 1800 if (zombie) { 1801 next = LIST_NEXT(marker, p_list); 1802 LIST_REMOVE(marker, p_list); 1803 } else { 1804 rw_exit(&p->p_reflock); 1805 next = LIST_NEXT(p, p_list); 1806 } 1807 } 1808 mutex_exit(proc_lock); 1809 1810 if (where != NULL) { 1811 *oldlenp = dp - where; 1812 if (needed > *oldlenp) { 1813 error = ENOMEM; 1814 goto out; 1815 } 1816 } else { 1817 needed += KERN_PROCSLOP; 1818 *oldlenp = needed; 1819 } 1820 if (kbuf) 1821 kmem_free(kbuf, sizeof(*kbuf)); 1822 if (marker) 1823 kmem_free(marker, sizeof(*marker)); 1824 sysctl_relock(); 1825 return 0; 1826 bah: 1827 if (zombie) 1828 LIST_REMOVE(marker, p_list); 1829 else 1830 rw_exit(&p->p_reflock); 1831 cleanup: 1832 mutex_exit(proc_lock); 1833 out: 1834 if (kbuf) 1835 kmem_free(kbuf, sizeof(*kbuf)); 1836 if (marker) 1837 kmem_free(marker, sizeof(*marker)); 1838 sysctl_relock(); 1839 return error; 1840 } 1841 1842 int 1843 copyin_psstrings(struct proc *p, struct ps_strings *arginfo) 1844 { 1845 1846 #ifdef COMPAT_NETBSD32 1847 if (p->p_flag & PK_32) { 1848 struct ps_strings32 arginfo32; 1849 1850 int error = copyin_proc(p, (void *)p->p_psstrp, &arginfo32, 1851 sizeof(arginfo32)); 1852 if (error) 1853 return error; 1854 arginfo->ps_argvstr = (void *)(uintptr_t)arginfo32.ps_argvstr; 1855 arginfo->ps_nargvstr = arginfo32.ps_nargvstr; 1856 arginfo->ps_envstr = (void *)(uintptr_t)arginfo32.ps_envstr; 1857 arginfo->ps_nenvstr = arginfo32.ps_nenvstr; 1858 return 0; 1859 } 1860 #endif 1861 return copyin_proc(p, (void *)p->p_psstrp, arginfo, sizeof(*arginfo)); 1862 } 1863 1864 static int 1865 copy_procargs_sysctl_cb(void *cookie_, const void *src, size_t off, size_t len) 1866 { 1867 void **cookie = cookie_; 1868 struct lwp *l = cookie[0]; 1869 char *dst = cookie[1]; 1870 1871 return sysctl_copyout(l, src, dst + off, len); 1872 } 1873 1874 /* 1875 * sysctl helper routine for kern.proc_args pseudo-subtree. 1876 */ 1877 static int 1878 sysctl_kern_proc_args(SYSCTLFN_ARGS) 1879 { 1880 struct ps_strings pss; 1881 struct proc *p; 1882 pid_t pid; 1883 int type, error; 1884 void *cookie[2]; 1885 1886 if (namelen == 1 && name[0] == CTL_QUERY) 1887 return (sysctl_query(SYSCTLFN_CALL(rnode))); 1888 1889 if (newp != NULL || namelen != 2) 1890 return (EINVAL); 1891 pid = name[0]; 1892 type = name[1]; 1893 1894 switch (type) { 1895 case KERN_PROC_ARGV: 1896 case KERN_PROC_NARGV: 1897 case KERN_PROC_ENV: 1898 case KERN_PROC_NENV: 1899 /* ok */ 1900 break; 1901 default: 1902 return (EINVAL); 1903 } 1904 1905 sysctl_unlock(); 1906 1907 /* check pid */ 1908 mutex_enter(proc_lock); 1909 if ((p = proc_find(pid)) == NULL) { 1910 error = EINVAL; 1911 goto out_locked; 1912 } 1913 mutex_enter(p->p_lock); 1914 1915 /* Check permission. */ 1916 if (type == KERN_PROC_ARGV || type == KERN_PROC_NARGV) 1917 error = kauth_authorize_process(l->l_cred, KAUTH_PROCESS_CANSEE, 1918 p, KAUTH_ARG(KAUTH_REQ_PROCESS_CANSEE_ARGS), NULL, NULL); 1919 else if (type == KERN_PROC_ENV || type == KERN_PROC_NENV) 1920 error = kauth_authorize_process(l->l_cred, KAUTH_PROCESS_CANSEE, 1921 p, KAUTH_ARG(KAUTH_REQ_PROCESS_CANSEE_ENV), NULL, NULL); 1922 else 1923 error = EINVAL; /* XXXGCC */ 1924 if (error) { 1925 mutex_exit(p->p_lock); 1926 goto out_locked; 1927 } 1928 1929 if (oldp == NULL) { 1930 if (type == KERN_PROC_NARGV || type == KERN_PROC_NENV) 1931 *oldlenp = sizeof (int); 1932 else 1933 *oldlenp = ARG_MAX; /* XXX XXX XXX */ 1934 error = 0; 1935 mutex_exit(p->p_lock); 1936 goto out_locked; 1937 } 1938 1939 /* 1940 * Zombies don't have a stack, so we can't read their psstrings. 1941 * System processes also don't have a user stack. 1942 */ 1943 if (P_ZOMBIE(p) || (p->p_flag & PK_SYSTEM) != 0) { 1944 error = EINVAL; 1945 mutex_exit(p->p_lock); 1946 goto out_locked; 1947 } 1948 1949 error = rw_tryenter(&p->p_reflock, RW_READER) ? 0 : EBUSY; 1950 mutex_exit(p->p_lock); 1951 if (error) { 1952 goto out_locked; 1953 } 1954 mutex_exit(proc_lock); 1955 1956 if (type == KERN_PROC_NARGV || type == KERN_PROC_NENV) { 1957 int value; 1958 if ((error = copyin_psstrings(p, &pss)) == 0) { 1959 if (type == KERN_PROC_NARGV) 1960 value = pss.ps_nargvstr; 1961 else 1962 value = pss.ps_nenvstr; 1963 error = sysctl_copyout(l, &value, oldp, sizeof(value)); 1964 *oldlenp = sizeof(value); 1965 } 1966 } else { 1967 cookie[0] = l; 1968 cookie[1] = oldp; 1969 error = copy_procargs(p, type, oldlenp, 1970 copy_procargs_sysctl_cb, cookie); 1971 } 1972 rw_exit(&p->p_reflock); 1973 sysctl_relock(); 1974 return error; 1975 1976 out_locked: 1977 mutex_exit(proc_lock); 1978 sysctl_relock(); 1979 return error; 1980 } 1981 1982 int 1983 copy_procargs(struct proc *p, int oid, size_t *limit, 1984 int (*cb)(void *, const void *, size_t, size_t), void *cookie) 1985 { 1986 struct ps_strings pss; 1987 size_t len, i, loaded, entry_len; 1988 struct uio auio; 1989 struct iovec aiov; 1990 int error, argvlen; 1991 char *arg; 1992 char **argv; 1993 vaddr_t user_argv; 1994 struct vmspace *vmspace; 1995 1996 /* 1997 * Allocate a temporary buffer to hold the argument vector and 1998 * the arguments themselve. 1999 */ 2000 arg = kmem_alloc(PAGE_SIZE, KM_SLEEP); 2001 argv = kmem_alloc(PAGE_SIZE, KM_SLEEP); 2002 2003 /* 2004 * Lock the process down in memory. 2005 */ 2006 vmspace = p->p_vmspace; 2007 uvmspace_addref(vmspace); 2008 2009 /* 2010 * Read in the ps_strings structure. 2011 */ 2012 if ((error = copyin_psstrings(p, &pss)) != 0) 2013 goto done; 2014 2015 /* 2016 * Now read the address of the argument vector. 2017 */ 2018 switch (oid) { 2019 case KERN_PROC_ARGV: 2020 user_argv = (uintptr_t)pss.ps_argvstr; 2021 argvlen = pss.ps_nargvstr; 2022 break; 2023 case KERN_PROC_ENV: 2024 user_argv = (uintptr_t)pss.ps_envstr; 2025 argvlen = pss.ps_nenvstr; 2026 break; 2027 default: 2028 error = EINVAL; 2029 goto done; 2030 } 2031 2032 if (argvlen < 0) { 2033 error = EIO; 2034 goto done; 2035 } 2036 2037 #ifdef COMPAT_NETBSD32 2038 if (p->p_flag & PK_32) 2039 entry_len = sizeof(netbsd32_charp); 2040 else 2041 #endif 2042 entry_len = sizeof(char *); 2043 2044 /* 2045 * Now copy each string. 2046 */ 2047 len = 0; /* bytes written to user buffer */ 2048 loaded = 0; /* bytes from argv already processed */ 2049 i = 0; /* To make compiler happy */ 2050 2051 for (; argvlen; --argvlen) { 2052 int finished = 0; 2053 vaddr_t base; 2054 size_t xlen; 2055 int j; 2056 2057 if (loaded == 0) { 2058 size_t rem = entry_len * argvlen; 2059 loaded = MIN(rem, PAGE_SIZE); 2060 error = copyin_vmspace(vmspace, 2061 (const void *)user_argv, argv, loaded); 2062 if (error) 2063 break; 2064 user_argv += loaded; 2065 i = 0; 2066 } 2067 2068 #ifdef COMPAT_NETBSD32 2069 if (p->p_flag & PK_32) { 2070 netbsd32_charp *argv32; 2071 2072 argv32 = (netbsd32_charp *)argv; 2073 base = (vaddr_t)NETBSD32PTR64(argv32[i++]); 2074 } else 2075 #endif 2076 base = (vaddr_t)argv[i++]; 2077 loaded -= entry_len; 2078 2079 /* 2080 * The program has messed around with its arguments, 2081 * possibly deleting some, and replacing them with 2082 * NULL's. Treat this as the last argument and not 2083 * a failure. 2084 */ 2085 if (base == 0) 2086 break; 2087 2088 while (!finished) { 2089 xlen = PAGE_SIZE - (base & PAGE_MASK); 2090 2091 aiov.iov_base = arg; 2092 aiov.iov_len = PAGE_SIZE; 2093 auio.uio_iov = &aiov; 2094 auio.uio_iovcnt = 1; 2095 auio.uio_offset = base; 2096 auio.uio_resid = xlen; 2097 auio.uio_rw = UIO_READ; 2098 UIO_SETUP_SYSSPACE(&auio); 2099 error = uvm_io(&vmspace->vm_map, &auio); 2100 if (error) 2101 goto done; 2102 2103 /* Look for the end of the string */ 2104 for (j = 0; j < xlen; j++) { 2105 if (arg[j] == '\0') { 2106 xlen = j + 1; 2107 finished = 1; 2108 break; 2109 } 2110 } 2111 2112 /* Check for user buffer overflow */ 2113 if (len + xlen > *limit) { 2114 finished = 1; 2115 if (len > *limit) 2116 xlen = 0; 2117 else 2118 xlen = *limit - len; 2119 } 2120 2121 /* Copyout the page */ 2122 error = (*cb)(cookie, arg, len, xlen); 2123 if (error) 2124 goto done; 2125 2126 len += xlen; 2127 base += xlen; 2128 } 2129 } 2130 *limit = len; 2131 2132 done: 2133 kmem_free(argv, PAGE_SIZE); 2134 kmem_free(arg, PAGE_SIZE); 2135 uvmspace_free(vmspace); 2136 return error; 2137 } 2138 2139 /* 2140 * Fill in an eproc structure for the specified process. 2141 */ 2142 void 2143 fill_eproc(struct proc *p, struct eproc *ep, bool zombie) 2144 { 2145 struct tty *tp; 2146 struct lwp *l; 2147 2148 KASSERT(mutex_owned(proc_lock)); 2149 KASSERT(mutex_owned(p->p_lock)); 2150 2151 memset(ep, 0, sizeof(*ep)); 2152 2153 ep->e_paddr = p; 2154 ep->e_sess = p->p_session; 2155 if (p->p_cred) { 2156 kauth_cred_topcred(p->p_cred, &ep->e_pcred); 2157 kauth_cred_toucred(p->p_cred, &ep->e_ucred); 2158 } 2159 if (p->p_stat != SIDL && !P_ZOMBIE(p) && !zombie) { 2160 struct vmspace *vm = p->p_vmspace; 2161 2162 ep->e_vm.vm_rssize = vm_resident_count(vm); 2163 ep->e_vm.vm_tsize = vm->vm_tsize; 2164 ep->e_vm.vm_dsize = vm->vm_dsize; 2165 ep->e_vm.vm_ssize = vm->vm_ssize; 2166 ep->e_vm.vm_map.size = vm->vm_map.size; 2167 2168 /* Pick the primary (first) LWP */ 2169 l = proc_active_lwp(p); 2170 KASSERT(l != NULL); 2171 lwp_lock(l); 2172 if (l->l_wchan) 2173 strncpy(ep->e_wmesg, l->l_wmesg, WMESGLEN); 2174 lwp_unlock(l); 2175 } 2176 if (p->p_pptr) 2177 ep->e_ppid = p->p_pptr->p_pid; 2178 if (p->p_pgrp && p->p_session) { 2179 ep->e_pgid = p->p_pgrp->pg_id; 2180 ep->e_jobc = p->p_pgrp->pg_jobc; 2181 ep->e_sid = p->p_session->s_sid; 2182 if ((p->p_lflag & PL_CONTROLT) && 2183 (tp = ep->e_sess->s_ttyp)) { 2184 ep->e_tdev = tp->t_dev; 2185 ep->e_tpgid = tp->t_pgrp ? tp->t_pgrp->pg_id : NO_PGID; 2186 ep->e_tsess = tp->t_session; 2187 } else 2188 ep->e_tdev = (uint32_t)NODEV; 2189 ep->e_flag = ep->e_sess->s_ttyvp ? EPROC_CTTY : 0; 2190 if (SESS_LEADER(p)) 2191 ep->e_flag |= EPROC_SLEADER; 2192 strncpy(ep->e_login, ep->e_sess->s_login, MAXLOGNAME); 2193 } 2194 ep->e_xsize = ep->e_xrssize = 0; 2195 ep->e_xccount = ep->e_xswrss = 0; 2196 } 2197 2198 /* 2199 * Fill in a kinfo_proc2 structure for the specified process. 2200 */ 2201 static void 2202 fill_kproc2(struct proc *p, struct kinfo_proc2 *ki, bool zombie) 2203 { 2204 struct tty *tp; 2205 struct lwp *l, *l2; 2206 struct timeval ut, st, rt; 2207 sigset_t ss1, ss2; 2208 struct rusage ru; 2209 struct vmspace *vm; 2210 2211 KASSERT(mutex_owned(proc_lock)); 2212 KASSERT(mutex_owned(p->p_lock)); 2213 2214 sigemptyset(&ss1); 2215 sigemptyset(&ss2); 2216 memset(ki, 0, sizeof(*ki)); 2217 2218 ki->p_paddr = PTRTOUINT64(p); 2219 ki->p_fd = PTRTOUINT64(p->p_fd); 2220 ki->p_cwdi = PTRTOUINT64(p->p_cwdi); 2221 ki->p_stats = PTRTOUINT64(p->p_stats); 2222 ki->p_limit = PTRTOUINT64(p->p_limit); 2223 ki->p_vmspace = PTRTOUINT64(p->p_vmspace); 2224 ki->p_sigacts = PTRTOUINT64(p->p_sigacts); 2225 ki->p_sess = PTRTOUINT64(p->p_session); 2226 ki->p_tsess = 0; /* may be changed if controlling tty below */ 2227 ki->p_ru = PTRTOUINT64(&p->p_stats->p_ru); 2228 ki->p_eflag = 0; 2229 ki->p_exitsig = p->p_exitsig; 2230 ki->p_flag = L_INMEM; /* Process never swapped out */ 2231 ki->p_flag |= sysctl_map_flags(sysctl_flagmap, p->p_flag); 2232 ki->p_flag |= sysctl_map_flags(sysctl_sflagmap, p->p_sflag); 2233 ki->p_flag |= sysctl_map_flags(sysctl_slflagmap, p->p_slflag); 2234 ki->p_flag |= sysctl_map_flags(sysctl_lflagmap, p->p_lflag); 2235 ki->p_flag |= sysctl_map_flags(sysctl_stflagmap, p->p_stflag); 2236 ki->p_pid = p->p_pid; 2237 if (p->p_pptr) 2238 ki->p_ppid = p->p_pptr->p_pid; 2239 else 2240 ki->p_ppid = 0; 2241 ki->p_uid = kauth_cred_geteuid(p->p_cred); 2242 ki->p_ruid = kauth_cred_getuid(p->p_cred); 2243 ki->p_gid = kauth_cred_getegid(p->p_cred); 2244 ki->p_rgid = kauth_cred_getgid(p->p_cred); 2245 ki->p_svuid = kauth_cred_getsvuid(p->p_cred); 2246 ki->p_svgid = kauth_cred_getsvgid(p->p_cred); 2247 ki->p_ngroups = kauth_cred_ngroups(p->p_cred); 2248 kauth_cred_getgroups(p->p_cred, ki->p_groups, 2249 min(ki->p_ngroups, sizeof(ki->p_groups) / sizeof(ki->p_groups[0])), 2250 UIO_SYSSPACE); 2251 2252 ki->p_uticks = p->p_uticks; 2253 ki->p_sticks = p->p_sticks; 2254 ki->p_iticks = p->p_iticks; 2255 ki->p_tpgid = NO_PGID; /* may be changed if controlling tty below */ 2256 ki->p_tracep = PTRTOUINT64(p->p_tracep); 2257 ki->p_traceflag = p->p_traceflag; 2258 2259 memcpy(&ki->p_sigignore, &p->p_sigctx.ps_sigignore,sizeof(ki_sigset_t)); 2260 memcpy(&ki->p_sigcatch, &p->p_sigctx.ps_sigcatch, sizeof(ki_sigset_t)); 2261 2262 ki->p_cpticks = 0; 2263 ki->p_pctcpu = p->p_pctcpu; 2264 ki->p_estcpu = 0; 2265 ki->p_stat = p->p_stat; /* Will likely be overridden by LWP status */ 2266 ki->p_realstat = p->p_stat; 2267 ki->p_nice = p->p_nice; 2268 ki->p_xstat = p->p_xstat; 2269 ki->p_acflag = p->p_acflag; 2270 2271 strncpy(ki->p_comm, p->p_comm, 2272 min(sizeof(ki->p_comm), sizeof(p->p_comm))); 2273 strncpy(ki->p_ename, p->p_emul->e_name, sizeof(ki->p_ename)); 2274 2275 ki->p_nlwps = p->p_nlwps; 2276 ki->p_realflag = ki->p_flag; 2277 2278 if (p->p_stat != SIDL && !P_ZOMBIE(p) && !zombie) { 2279 vm = p->p_vmspace; 2280 ki->p_vm_rssize = vm_resident_count(vm); 2281 ki->p_vm_tsize = vm->vm_tsize; 2282 ki->p_vm_dsize = vm->vm_dsize; 2283 ki->p_vm_ssize = vm->vm_ssize; 2284 ki->p_vm_vsize = atop(vm->vm_map.size); 2285 /* 2286 * Since the stack is initially mapped mostly with 2287 * PROT_NONE and grown as needed, adjust the "mapped size" 2288 * to skip the unused stack portion. 2289 */ 2290 ki->p_vm_msize = 2291 atop(vm->vm_map.size) - vm->vm_issize + vm->vm_ssize; 2292 2293 /* Pick the primary (first) LWP */ 2294 l = proc_active_lwp(p); 2295 KASSERT(l != NULL); 2296 lwp_lock(l); 2297 ki->p_nrlwps = p->p_nrlwps; 2298 ki->p_forw = 0; 2299 ki->p_back = 0; 2300 ki->p_addr = PTRTOUINT64(l->l_addr); 2301 ki->p_stat = l->l_stat; 2302 ki->p_flag |= sysctl_map_flags(sysctl_lwpflagmap, l->l_flag); 2303 ki->p_swtime = l->l_swtime; 2304 ki->p_slptime = l->l_slptime; 2305 if (l->l_stat == LSONPROC) 2306 ki->p_schedflags = l->l_cpu->ci_schedstate.spc_flags; 2307 else 2308 ki->p_schedflags = 0; 2309 ki->p_priority = lwp_eprio(l); 2310 ki->p_usrpri = l->l_priority; 2311 if (l->l_wchan) 2312 strncpy(ki->p_wmesg, l->l_wmesg, sizeof(ki->p_wmesg)); 2313 ki->p_wchan = PTRTOUINT64(l->l_wchan); 2314 ki->p_cpuid = cpu_index(l->l_cpu); 2315 lwp_unlock(l); 2316 LIST_FOREACH(l, &p->p_lwps, l_sibling) { 2317 /* This is hardly correct, but... */ 2318 sigplusset(&l->l_sigpend.sp_set, &ss1); 2319 sigplusset(&l->l_sigmask, &ss2); 2320 ki->p_cpticks += l->l_cpticks; 2321 ki->p_pctcpu += l->l_pctcpu; 2322 ki->p_estcpu += l->l_estcpu; 2323 } 2324 } 2325 sigplusset(&p->p_sigpend.sp_set, &ss2); 2326 memcpy(&ki->p_siglist, &ss1, sizeof(ki_sigset_t)); 2327 memcpy(&ki->p_sigmask, &ss2, sizeof(ki_sigset_t)); 2328 2329 if (p->p_session != NULL) { 2330 ki->p_sid = p->p_session->s_sid; 2331 ki->p__pgid = p->p_pgrp->pg_id; 2332 if (p->p_session->s_ttyvp) 2333 ki->p_eflag |= EPROC_CTTY; 2334 if (SESS_LEADER(p)) 2335 ki->p_eflag |= EPROC_SLEADER; 2336 strncpy(ki->p_login, p->p_session->s_login, 2337 min(sizeof ki->p_login - 1, sizeof p->p_session->s_login)); 2338 ki->p_jobc = p->p_pgrp->pg_jobc; 2339 if ((p->p_lflag & PL_CONTROLT) && (tp = p->p_session->s_ttyp)) { 2340 ki->p_tdev = tp->t_dev; 2341 ki->p_tpgid = tp->t_pgrp ? tp->t_pgrp->pg_id : NO_PGID; 2342 ki->p_tsess = PTRTOUINT64(tp->t_session); 2343 } else { 2344 ki->p_tdev = (int32_t)NODEV; 2345 } 2346 } 2347 2348 if (!P_ZOMBIE(p) && !zombie) { 2349 ki->p_uvalid = 1; 2350 ki->p_ustart_sec = p->p_stats->p_start.tv_sec; 2351 ki->p_ustart_usec = p->p_stats->p_start.tv_usec; 2352 2353 calcru(p, &ut, &st, NULL, &rt); 2354 ki->p_rtime_sec = rt.tv_sec; 2355 ki->p_rtime_usec = rt.tv_usec; 2356 ki->p_uutime_sec = ut.tv_sec; 2357 ki->p_uutime_usec = ut.tv_usec; 2358 ki->p_ustime_sec = st.tv_sec; 2359 ki->p_ustime_usec = st.tv_usec; 2360 2361 memcpy(&ru, &p->p_stats->p_ru, sizeof(ru)); 2362 ki->p_uru_nvcsw = 0; 2363 ki->p_uru_nivcsw = 0; 2364 LIST_FOREACH(l2, &p->p_lwps, l_sibling) { 2365 ki->p_uru_nvcsw += (l2->l_ncsw - l2->l_nivcsw); 2366 ki->p_uru_nivcsw += l2->l_nivcsw; 2367 ruadd(&ru, &l2->l_ru); 2368 } 2369 ki->p_uru_maxrss = ru.ru_maxrss; 2370 ki->p_uru_ixrss = ru.ru_ixrss; 2371 ki->p_uru_idrss = ru.ru_idrss; 2372 ki->p_uru_isrss = ru.ru_isrss; 2373 ki->p_uru_minflt = ru.ru_minflt; 2374 ki->p_uru_majflt = ru.ru_majflt; 2375 ki->p_uru_nswap = ru.ru_nswap; 2376 ki->p_uru_inblock = ru.ru_inblock; 2377 ki->p_uru_oublock = ru.ru_oublock; 2378 ki->p_uru_msgsnd = ru.ru_msgsnd; 2379 ki->p_uru_msgrcv = ru.ru_msgrcv; 2380 ki->p_uru_nsignals = ru.ru_nsignals; 2381 2382 timeradd(&p->p_stats->p_cru.ru_utime, 2383 &p->p_stats->p_cru.ru_stime, &ut); 2384 ki->p_uctime_sec = ut.tv_sec; 2385 ki->p_uctime_usec = ut.tv_usec; 2386 } 2387 } 2388
Indexes created Sat Sep 20 22:09:52 GMT 2025