1 /* $NetBSD: kern_proc.c,v 1.223 2018/12/06 13:51:43 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 1.223 2018/12/06 13:51:43 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 #include "opt_kaslr.h" 73 #endif 74 75 #if defined(__HAVE_COMPAT_NETBSD32) && !defined(COMPAT_NETBSD32) \ 76 && !defined(_RUMPKERNEL) 77 #define COMPAT_NETBSD32 78 #endif 79 80 #include <sys/param.h> 81 #include <sys/systm.h> 82 #include <sys/kernel.h> 83 #include <sys/proc.h> 84 #include <sys/resourcevar.h> 85 #include <sys/buf.h> 86 #include <sys/acct.h> 87 #include <sys/wait.h> 88 #include <sys/file.h> 89 #include <ufs/ufs/quota.h> 90 #include <sys/uio.h> 91 #include <sys/pool.h> 92 #include <sys/pset.h> 93 #include <sys/mbuf.h> 94 #include <sys/ioctl.h> 95 #include <sys/tty.h> 96 #include <sys/signalvar.h> 97 #include <sys/ras.h> 98 #include <sys/filedesc.h> 99 #include <sys/syscall_stats.h> 100 #include <sys/kauth.h> 101 #include <sys/sleepq.h> 102 #include <sys/atomic.h> 103 #include <sys/kmem.h> 104 #include <sys/namei.h> 105 #include <sys/dtrace_bsd.h> 106 #include <sys/sysctl.h> 107 #include <sys/exec.h> 108 #include <sys/cpu.h> 109 110 #include <uvm/uvm_extern.h> 111 #include <uvm/uvm.h> 112 113 #ifdef COMPAT_NETBSD32 114 #include <compat/netbsd32/netbsd32.h> 115 #endif 116 117 /* 118 * Process lists. 119 */ 120 121 struct proclist allproc __cacheline_aligned; 122 struct proclist zombproc __cacheline_aligned; 123 124 kmutex_t * proc_lock __cacheline_aligned; 125 126 /* 127 * pid to proc lookup is done by indexing the pid_table array. 128 * Since pid numbers are only allocated when an empty slot 129 * has been found, there is no need to search any lists ever. 130 * (an orphaned pgrp will lock the slot, a session will lock 131 * the pgrp with the same number.) 132 * If the table is too small it is reallocated with twice the 133 * previous size and the entries 'unzipped' into the two halves. 134 * A linked list of free entries is passed through the pt_proc 135 * field of 'free' items - set odd to be an invalid ptr. 136 */ 137 138 struct pid_table { 139 struct proc *pt_proc; 140 struct pgrp *pt_pgrp; 141 pid_t pt_pid; 142 }; 143 #if 1 /* strongly typed cast - should be a noop */ 144 static inline uint p2u(struct proc *p) { return (uint)(uintptr_t)p; } 145 #else 146 #define p2u(p) ((uint)p) 147 #endif 148 #define P_VALID(p) (!(p2u(p) & 1)) 149 #define P_NEXT(p) (p2u(p) >> 1) 150 #define P_FREE(pid) ((struct proc *)(uintptr_t)((pid) << 1 | 1)) 151 152 /* 153 * Table of process IDs (PIDs). 154 */ 155 static struct pid_table *pid_table __read_mostly; 156 157 #define INITIAL_PID_TABLE_SIZE (1 << 5) 158 159 /* Table mask, threshold for growing and number of allocated PIDs. */ 160 static u_int pid_tbl_mask __read_mostly; 161 static u_int pid_alloc_lim __read_mostly; 162 static u_int pid_alloc_cnt __cacheline_aligned; 163 164 /* Next free, last free and maximum PIDs. */ 165 static u_int next_free_pt __cacheline_aligned; 166 static u_int last_free_pt __cacheline_aligned; 167 static pid_t pid_max __read_mostly; 168 169 /* Components of the first process -- never freed. */ 170 171 extern struct emul emul_netbsd; /* defined in kern_exec.c */ 172 173 struct session session0 = { 174 .s_count = 1, 175 .s_sid = 0, 176 }; 177 struct pgrp pgrp0 = { 178 .pg_members = LIST_HEAD_INITIALIZER(&pgrp0.pg_members), 179 .pg_session = &session0, 180 }; 181 filedesc_t filedesc0; 182 struct cwdinfo cwdi0 = { 183 .cwdi_cmask = CMASK, 184 .cwdi_refcnt = 1, 185 }; 186 struct plimit limit0; 187 struct pstats pstat0; 188 struct vmspace vmspace0; 189 struct sigacts sigacts0; 190 struct proc proc0 = { 191 .p_lwps = LIST_HEAD_INITIALIZER(&proc0.p_lwps), 192 .p_sigwaiters = LIST_HEAD_INITIALIZER(&proc0.p_sigwaiters), 193 .p_nlwps = 1, 194 .p_nrlwps = 1, 195 .p_nlwpid = 1, /* must match lwp0.l_lid */ 196 .p_pgrp = &pgrp0, 197 .p_comm = "system", 198 /* 199 * Set P_NOCLDWAIT so that kernel threads are reparented to init(8) 200 * when they exit. init(8) can easily wait them out for us. 201 */ 202 .p_flag = PK_SYSTEM | PK_NOCLDWAIT, 203 .p_stat = SACTIVE, 204 .p_nice = NZERO, 205 .p_emul = &emul_netbsd, 206 .p_cwdi = &cwdi0, 207 .p_limit = &limit0, 208 .p_fd = &filedesc0, 209 .p_vmspace = &vmspace0, 210 .p_stats = &pstat0, 211 .p_sigacts = &sigacts0, 212 #ifdef PROC0_MD_INITIALIZERS 213 PROC0_MD_INITIALIZERS 214 #endif 215 }; 216 kauth_cred_t cred0; 217 218 static const int nofile = NOFILE; 219 static const int maxuprc = MAXUPRC; 220 221 static int sysctl_doeproc(SYSCTLFN_PROTO); 222 static int sysctl_kern_proc_args(SYSCTLFN_PROTO); 223 static int sysctl_security_expose_address(SYSCTLFN_PROTO); 224 225 #ifdef KASLR 226 static int kern_expose_address = 0; 227 #else 228 static int kern_expose_address = 1; 229 #endif 230 /* 231 * The process list descriptors, used during pid allocation and 232 * by sysctl. No locking on this data structure is needed since 233 * it is completely static. 234 */ 235 const struct proclist_desc proclists[] = { 236 { &allproc }, 237 { &zombproc }, 238 { NULL }, 239 }; 240 241 static struct pgrp * pg_remove(pid_t); 242 static void pg_delete(pid_t); 243 static void orphanpg(struct pgrp *); 244 245 static specificdata_domain_t proc_specificdata_domain; 246 247 static pool_cache_t proc_cache; 248 249 static kauth_listener_t proc_listener; 250 251 static void fill_proc(const struct proc *, struct proc *, bool); 252 static int fill_pathname(struct lwp *, pid_t, void *, size_t *); 253 254 static int 255 proc_listener_cb(kauth_cred_t cred, kauth_action_t action, void *cookie, 256 void *arg0, void *arg1, void *arg2, void *arg3) 257 { 258 struct proc *p; 259 int result; 260 261 result = KAUTH_RESULT_DEFER; 262 p = arg0; 263 264 switch (action) { 265 case KAUTH_PROCESS_CANSEE: { 266 enum kauth_process_req req; 267 268 req = (enum kauth_process_req)arg1; 269 270 switch (req) { 271 case KAUTH_REQ_PROCESS_CANSEE_ARGS: 272 case KAUTH_REQ_PROCESS_CANSEE_ENTRY: 273 case KAUTH_REQ_PROCESS_CANSEE_OPENFILES: 274 case KAUTH_REQ_PROCESS_CANSEE_EPROC: 275 result = KAUTH_RESULT_ALLOW; 276 break; 277 278 case KAUTH_REQ_PROCESS_CANSEE_ENV: 279 if (kauth_cred_getuid(cred) != 280 kauth_cred_getuid(p->p_cred) || 281 kauth_cred_getuid(cred) != 282 kauth_cred_getsvuid(p->p_cred)) 283 break; 284 285 result = KAUTH_RESULT_ALLOW; 286 287 break; 288 289 case KAUTH_REQ_PROCESS_CANSEE_KPTR: 290 if (!kern_expose_address) 291 break; 292 293 if (kern_expose_address == 1 && !(p->p_flag & PK_KMEM)) 294 break; 295 296 result = KAUTH_RESULT_ALLOW; 297 298 break; 299 300 default: 301 break; 302 } 303 304 break; 305 } 306 307 case KAUTH_PROCESS_FORK: { 308 int lnprocs = (int)(unsigned long)arg2; 309 310 /* 311 * Don't allow a nonprivileged user to use the last few 312 * processes. The variable lnprocs is the current number of 313 * processes, maxproc is the limit. 314 */ 315 if (__predict_false((lnprocs >= maxproc - 5))) 316 break; 317 318 result = KAUTH_RESULT_ALLOW; 319 320 break; 321 } 322 323 case KAUTH_PROCESS_CORENAME: 324 case KAUTH_PROCESS_STOPFLAG: 325 if (proc_uidmatch(cred, p->p_cred) == 0) 326 result = KAUTH_RESULT_ALLOW; 327 328 break; 329 330 default: 331 break; 332 } 333 334 return result; 335 } 336 337 static int 338 proc_ctor(void *arg __unused, void *obj, int flags __unused) 339 { 340 memset(obj, 0, sizeof(struct proc)); 341 return 0; 342 } 343 344 /* 345 * Initialize global process hashing structures. 346 */ 347 void 348 procinit(void) 349 { 350 const struct proclist_desc *pd; 351 u_int i; 352 #define LINK_EMPTY ((PID_MAX + INITIAL_PID_TABLE_SIZE) & ~(INITIAL_PID_TABLE_SIZE - 1)) 353 354 for (pd = proclists; pd->pd_list != NULL; pd++) 355 LIST_INIT(pd->pd_list); 356 357 proc_lock = mutex_obj_alloc(MUTEX_DEFAULT, IPL_NONE); 358 pid_table = kmem_alloc(INITIAL_PID_TABLE_SIZE 359 * sizeof(struct pid_table), KM_SLEEP); 360 pid_tbl_mask = INITIAL_PID_TABLE_SIZE - 1; 361 pid_max = PID_MAX; 362 363 /* Set free list running through table... 364 Preset 'use count' above PID_MAX so we allocate pid 1 next. */ 365 for (i = 0; i <= pid_tbl_mask; i++) { 366 pid_table[i].pt_proc = P_FREE(LINK_EMPTY + i + 1); 367 pid_table[i].pt_pgrp = 0; 368 pid_table[i].pt_pid = 0; 369 } 370 /* slot 0 is just grabbed */ 371 next_free_pt = 1; 372 /* Need to fix last entry. */ 373 last_free_pt = pid_tbl_mask; 374 pid_table[last_free_pt].pt_proc = P_FREE(LINK_EMPTY); 375 /* point at which we grow table - to avoid reusing pids too often */ 376 pid_alloc_lim = pid_tbl_mask - 1; 377 #undef LINK_EMPTY 378 379 proc_specificdata_domain = specificdata_domain_create(); 380 KASSERT(proc_specificdata_domain != NULL); 381 382 proc_cache = pool_cache_init(sizeof(struct proc), 0, 0, 0, 383 "procpl", NULL, IPL_NONE, proc_ctor, NULL, NULL); 384 385 proc_listener = kauth_listen_scope(KAUTH_SCOPE_PROCESS, 386 proc_listener_cb, NULL); 387 } 388 389 void 390 procinit_sysctl(void) 391 { 392 static struct sysctllog *clog; 393 394 sysctl_createv(&clog, 0, NULL, NULL, 395 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 396 CTLTYPE_INT, "expose_address", 397 SYSCTL_DESCR("Enable exposing kernel addresses"), 398 sysctl_security_expose_address, 0, 399 &kern_expose_address, 0, CTL_KERN, CTL_CREATE, CTL_EOL); 400 sysctl_createv(&clog, 0, NULL, NULL, 401 CTLFLAG_PERMANENT, 402 CTLTYPE_NODE, "proc", 403 SYSCTL_DESCR("System-wide process information"), 404 sysctl_doeproc, 0, NULL, 0, 405 CTL_KERN, KERN_PROC, CTL_EOL); 406 sysctl_createv(&clog, 0, NULL, NULL, 407 CTLFLAG_PERMANENT, 408 CTLTYPE_NODE, "proc2", 409 SYSCTL_DESCR("Machine-independent process information"), 410 sysctl_doeproc, 0, NULL, 0, 411 CTL_KERN, KERN_PROC2, CTL_EOL); 412 sysctl_createv(&clog, 0, NULL, NULL, 413 CTLFLAG_PERMANENT, 414 CTLTYPE_NODE, "proc_args", 415 SYSCTL_DESCR("Process argument information"), 416 sysctl_kern_proc_args, 0, NULL, 0, 417 CTL_KERN, KERN_PROC_ARGS, CTL_EOL); 418 419 /* 420 "nodes" under these: 421 422 KERN_PROC_ALL 423 KERN_PROC_PID pid 424 KERN_PROC_PGRP pgrp 425 KERN_PROC_SESSION sess 426 KERN_PROC_TTY tty 427 KERN_PROC_UID uid 428 KERN_PROC_RUID uid 429 KERN_PROC_GID gid 430 KERN_PROC_RGID gid 431 432 all in all, probably not worth the effort... 433 */ 434 } 435 436 /* 437 * Initialize process 0. 438 */ 439 void 440 proc0_init(void) 441 { 442 struct proc *p; 443 struct pgrp *pg; 444 struct rlimit *rlim; 445 rlim_t lim; 446 int i; 447 448 p = &proc0; 449 pg = &pgrp0; 450 451 mutex_init(&p->p_stmutex, MUTEX_DEFAULT, IPL_HIGH); 452 mutex_init(&p->p_auxlock, MUTEX_DEFAULT, IPL_NONE); 453 p->p_lock = mutex_obj_alloc(MUTEX_DEFAULT, IPL_NONE); 454 455 rw_init(&p->p_reflock); 456 cv_init(&p->p_waitcv, "wait"); 457 cv_init(&p->p_lwpcv, "lwpwait"); 458 459 LIST_INSERT_HEAD(&p->p_lwps, &lwp0, l_sibling); 460 461 pid_table[0].pt_proc = p; 462 LIST_INSERT_HEAD(&allproc, p, p_list); 463 464 pid_table[0].pt_pgrp = pg; 465 LIST_INSERT_HEAD(&pg->pg_members, p, p_pglist); 466 467 #ifdef __HAVE_SYSCALL_INTERN 468 (*p->p_emul->e_syscall_intern)(p); 469 #endif 470 471 /* Create credentials. */ 472 cred0 = kauth_cred_alloc(); 473 p->p_cred = cred0; 474 475 /* Create the CWD info. */ 476 rw_init(&cwdi0.cwdi_lock); 477 478 /* Create the limits structures. */ 479 mutex_init(&limit0.pl_lock, MUTEX_DEFAULT, IPL_NONE); 480 481 rlim = limit0.pl_rlimit; 482 for (i = 0; i < __arraycount(limit0.pl_rlimit); i++) { 483 rlim[i].rlim_cur = RLIM_INFINITY; 484 rlim[i].rlim_max = RLIM_INFINITY; 485 } 486 487 rlim[RLIMIT_NOFILE].rlim_max = maxfiles; 488 rlim[RLIMIT_NOFILE].rlim_cur = maxfiles < nofile ? maxfiles : nofile; 489 490 rlim[RLIMIT_NPROC].rlim_max = maxproc; 491 rlim[RLIMIT_NPROC].rlim_cur = maxproc < maxuprc ? maxproc : maxuprc; 492 493 lim = MIN(VM_MAXUSER_ADDRESS, ctob((rlim_t)uvmexp.free)); 494 rlim[RLIMIT_RSS].rlim_max = lim; 495 rlim[RLIMIT_MEMLOCK].rlim_max = lim; 496 rlim[RLIMIT_MEMLOCK].rlim_cur = lim / 3; 497 498 rlim[RLIMIT_NTHR].rlim_max = maxlwp; 499 rlim[RLIMIT_NTHR].rlim_cur = maxlwp < maxuprc ? maxlwp : maxuprc; 500 501 /* Note that default core name has zero length. */ 502 limit0.pl_corename = defcorename; 503 limit0.pl_cnlen = 0; 504 limit0.pl_refcnt = 1; 505 limit0.pl_writeable = false; 506 limit0.pl_sv_limit = NULL; 507 508 /* Configure virtual memory system, set vm rlimits. */ 509 uvm_init_limits(p); 510 511 /* Initialize file descriptor table for proc0. */ 512 fd_init(&filedesc0); 513 514 /* 515 * Initialize proc0's vmspace, which uses the kernel pmap. 516 * All kernel processes (which never have user space mappings) 517 * share proc0's vmspace, and thus, the kernel pmap. 518 */ 519 uvmspace_init(&vmspace0, pmap_kernel(), round_page(VM_MIN_ADDRESS), 520 trunc_page(VM_MAXUSER_ADDRESS), 521 #ifdef __USE_TOPDOWN_VM 522 true 523 #else 524 false 525 #endif 526 ); 527 528 /* Initialize signal state for proc0. XXX IPL_SCHED */ 529 mutex_init(&p->p_sigacts->sa_mutex, MUTEX_DEFAULT, IPL_SCHED); 530 siginit(p); 531 532 proc_initspecific(p); 533 kdtrace_proc_ctor(NULL, p); 534 } 535 536 /* 537 * Session reference counting. 538 */ 539 540 void 541 proc_sesshold(struct session *ss) 542 { 543 544 KASSERT(mutex_owned(proc_lock)); 545 ss->s_count++; 546 } 547 548 void 549 proc_sessrele(struct session *ss) 550 { 551 552 KASSERT(mutex_owned(proc_lock)); 553 /* 554 * We keep the pgrp with the same id as the session in order to 555 * stop a process being given the same pid. Since the pgrp holds 556 * a reference to the session, it must be a 'zombie' pgrp by now. 557 */ 558 if (--ss->s_count == 0) { 559 struct pgrp *pg; 560 561 pg = pg_remove(ss->s_sid); 562 mutex_exit(proc_lock); 563 564 kmem_free(pg, sizeof(struct pgrp)); 565 kmem_free(ss, sizeof(struct session)); 566 } else { 567 mutex_exit(proc_lock); 568 } 569 } 570 571 /* 572 * Check that the specified process group is in the session of the 573 * specified process. 574 * Treats -ve ids as process ids. 575 * Used to validate TIOCSPGRP requests. 576 */ 577 int 578 pgid_in_session(struct proc *p, pid_t pg_id) 579 { 580 struct pgrp *pgrp; 581 struct session *session; 582 int error; 583 584 mutex_enter(proc_lock); 585 if (pg_id < 0) { 586 struct proc *p1 = proc_find(-pg_id); 587 if (p1 == NULL) { 588 error = EINVAL; 589 goto fail; 590 } 591 pgrp = p1->p_pgrp; 592 } else { 593 pgrp = pgrp_find(pg_id); 594 if (pgrp == NULL) { 595 error = EINVAL; 596 goto fail; 597 } 598 } 599 session = pgrp->pg_session; 600 error = (session != p->p_pgrp->pg_session) ? EPERM : 0; 601 fail: 602 mutex_exit(proc_lock); 603 return error; 604 } 605 606 /* 607 * p_inferior: is p an inferior of q? 608 */ 609 static inline bool 610 p_inferior(struct proc *p, struct proc *q) 611 { 612 613 KASSERT(mutex_owned(proc_lock)); 614 615 for (; p != q; p = p->p_pptr) 616 if (p->p_pid == 0) 617 return false; 618 return true; 619 } 620 621 /* 622 * proc_find: locate a process by the ID. 623 * 624 * => Must be called with proc_lock held. 625 */ 626 proc_t * 627 proc_find_raw(pid_t pid) 628 { 629 struct pid_table *pt; 630 proc_t *p; 631 632 KASSERT(mutex_owned(proc_lock)); 633 pt = &pid_table[pid & pid_tbl_mask]; 634 p = pt->pt_proc; 635 if (__predict_false(!P_VALID(p) || pt->pt_pid != pid)) { 636 return NULL; 637 } 638 return p; 639 } 640 641 proc_t * 642 proc_find(pid_t pid) 643 { 644 proc_t *p; 645 646 p = proc_find_raw(pid); 647 if (__predict_false(p == NULL)) { 648 return NULL; 649 } 650 651 /* 652 * Only allow live processes to be found by PID. 653 * XXX: p_stat might change, since unlocked. 654 */ 655 if (__predict_true(p->p_stat == SACTIVE || p->p_stat == SSTOP)) { 656 return p; 657 } 658 return NULL; 659 } 660 661 /* 662 * pgrp_find: locate a process group by the ID. 663 * 664 * => Must be called with proc_lock held. 665 */ 666 struct pgrp * 667 pgrp_find(pid_t pgid) 668 { 669 struct pgrp *pg; 670 671 KASSERT(mutex_owned(proc_lock)); 672 673 pg = pid_table[pgid & pid_tbl_mask].pt_pgrp; 674 675 /* 676 * Cannot look up a process group that only exists because the 677 * session has not died yet (traditional). 678 */ 679 if (pg == NULL || pg->pg_id != pgid || LIST_EMPTY(&pg->pg_members)) { 680 return NULL; 681 } 682 return pg; 683 } 684 685 static void 686 expand_pid_table(void) 687 { 688 size_t pt_size, tsz; 689 struct pid_table *n_pt, *new_pt; 690 struct proc *proc; 691 struct pgrp *pgrp; 692 pid_t pid, rpid; 693 u_int i; 694 uint new_pt_mask; 695 696 pt_size = pid_tbl_mask + 1; 697 tsz = pt_size * 2 * sizeof(struct pid_table); 698 new_pt = kmem_alloc(tsz, KM_SLEEP); 699 new_pt_mask = pt_size * 2 - 1; 700 701 mutex_enter(proc_lock); 702 if (pt_size != pid_tbl_mask + 1) { 703 /* Another process beat us to it... */ 704 mutex_exit(proc_lock); 705 kmem_free(new_pt, tsz); 706 return; 707 } 708 709 /* 710 * Copy entries from old table into new one. 711 * If 'pid' is 'odd' we need to place in the upper half, 712 * even pid's to the lower half. 713 * Free items stay in the low half so we don't have to 714 * fixup the reference to them. 715 * We stuff free items on the front of the freelist 716 * because we can't write to unmodified entries. 717 * Processing the table backwards maintains a semblance 718 * of issuing pid numbers that increase with time. 719 */ 720 i = pt_size - 1; 721 n_pt = new_pt + i; 722 for (; ; i--, n_pt--) { 723 proc = pid_table[i].pt_proc; 724 pgrp = pid_table[i].pt_pgrp; 725 if (!P_VALID(proc)) { 726 /* Up 'use count' so that link is valid */ 727 pid = (P_NEXT(proc) + pt_size) & ~pt_size; 728 rpid = 0; 729 proc = P_FREE(pid); 730 if (pgrp) 731 pid = pgrp->pg_id; 732 } else { 733 pid = pid_table[i].pt_pid; 734 rpid = pid; 735 } 736 737 /* Save entry in appropriate half of table */ 738 n_pt[pid & pt_size].pt_proc = proc; 739 n_pt[pid & pt_size].pt_pgrp = pgrp; 740 n_pt[pid & pt_size].pt_pid = rpid; 741 742 /* Put other piece on start of free list */ 743 pid = (pid ^ pt_size) & ~pid_tbl_mask; 744 n_pt[pid & pt_size].pt_proc = 745 P_FREE((pid & ~pt_size) | next_free_pt); 746 n_pt[pid & pt_size].pt_pgrp = 0; 747 n_pt[pid & pt_size].pt_pid = 0; 748 749 next_free_pt = i | (pid & pt_size); 750 if (i == 0) 751 break; 752 } 753 754 /* Save old table size and switch tables */ 755 tsz = pt_size * sizeof(struct pid_table); 756 n_pt = pid_table; 757 pid_table = new_pt; 758 pid_tbl_mask = new_pt_mask; 759 760 /* 761 * pid_max starts as PID_MAX (= 30000), once we have 16384 762 * allocated pids we need it to be larger! 763 */ 764 if (pid_tbl_mask > PID_MAX) { 765 pid_max = pid_tbl_mask * 2 + 1; 766 pid_alloc_lim |= pid_alloc_lim << 1; 767 } else 768 pid_alloc_lim <<= 1; /* doubles number of free slots... */ 769 770 mutex_exit(proc_lock); 771 kmem_free(n_pt, tsz); 772 } 773 774 struct proc * 775 proc_alloc(void) 776 { 777 struct proc *p; 778 779 p = pool_cache_get(proc_cache, PR_WAITOK); 780 p->p_stat = SIDL; /* protect against others */ 781 proc_initspecific(p); 782 kdtrace_proc_ctor(NULL, p); 783 p->p_pid = -1; 784 proc_alloc_pid(p); 785 return p; 786 } 787 788 /* 789 * proc_alloc_pid: allocate PID and record the given proc 'p' so that 790 * proc_find_raw() can find it by the PID. 791 */ 792 793 pid_t 794 proc_alloc_pid(struct proc *p) 795 { 796 struct pid_table *pt; 797 pid_t pid; 798 int nxt; 799 800 for (;;expand_pid_table()) { 801 if (__predict_false(pid_alloc_cnt >= pid_alloc_lim)) 802 /* ensure pids cycle through 2000+ values */ 803 continue; 804 mutex_enter(proc_lock); 805 pt = &pid_table[next_free_pt]; 806 #ifdef DIAGNOSTIC 807 if (__predict_false(P_VALID(pt->pt_proc) || pt->pt_pgrp)) 808 panic("proc_alloc: slot busy"); 809 #endif 810 nxt = P_NEXT(pt->pt_proc); 811 if (nxt & pid_tbl_mask) 812 break; 813 /* Table full - expand (NB last entry not used....) */ 814 mutex_exit(proc_lock); 815 } 816 817 /* pid is 'saved use count' + 'size' + entry */ 818 pid = (nxt & ~pid_tbl_mask) + pid_tbl_mask + 1 + next_free_pt; 819 if ((uint)pid > (uint)pid_max) 820 pid &= pid_tbl_mask; 821 next_free_pt = nxt & pid_tbl_mask; 822 823 /* Grab table slot */ 824 pt->pt_proc = p; 825 826 KASSERT(pt->pt_pid == 0); 827 pt->pt_pid = pid; 828 if (p->p_pid == -1) { 829 p->p_pid = pid; 830 } 831 pid_alloc_cnt++; 832 mutex_exit(proc_lock); 833 834 return pid; 835 } 836 837 /* 838 * Free a process id - called from proc_free (in kern_exit.c) 839 * 840 * Called with the proc_lock held. 841 */ 842 void 843 proc_free_pid(pid_t pid) 844 { 845 struct pid_table *pt; 846 847 KASSERT(mutex_owned(proc_lock)); 848 849 pt = &pid_table[pid & pid_tbl_mask]; 850 851 /* save pid use count in slot */ 852 pt->pt_proc = P_FREE(pid & ~pid_tbl_mask); 853 KASSERT(pt->pt_pid == pid); 854 pt->pt_pid = 0; 855 856 if (pt->pt_pgrp == NULL) { 857 /* link last freed entry onto ours */ 858 pid &= pid_tbl_mask; 859 pt = &pid_table[last_free_pt]; 860 pt->pt_proc = P_FREE(P_NEXT(pt->pt_proc) | pid); 861 pt->pt_pid = 0; 862 last_free_pt = pid; 863 pid_alloc_cnt--; 864 } 865 866 atomic_dec_uint(&nprocs); 867 } 868 869 void 870 proc_free_mem(struct proc *p) 871 { 872 873 kdtrace_proc_dtor(NULL, p); 874 pool_cache_put(proc_cache, p); 875 } 876 877 /* 878 * proc_enterpgrp: move p to a new or existing process group (and session). 879 * 880 * If we are creating a new pgrp, the pgid should equal 881 * the calling process' pid. 882 * If is only valid to enter a process group that is in the session 883 * of the process. 884 * Also mksess should only be set if we are creating a process group 885 * 886 * Only called from sys_setsid, sys_setpgid and posix_spawn/spawn_return. 887 */ 888 int 889 proc_enterpgrp(struct proc *curp, pid_t pid, pid_t pgid, bool mksess) 890 { 891 struct pgrp *new_pgrp, *pgrp; 892 struct session *sess; 893 struct proc *p; 894 int rval; 895 pid_t pg_id = NO_PGID; 896 897 sess = mksess ? kmem_alloc(sizeof(*sess), KM_SLEEP) : NULL; 898 899 /* Allocate data areas we might need before doing any validity checks */ 900 mutex_enter(proc_lock); /* Because pid_table might change */ 901 if (pid_table[pgid & pid_tbl_mask].pt_pgrp == 0) { 902 mutex_exit(proc_lock); 903 new_pgrp = kmem_alloc(sizeof(*new_pgrp), KM_SLEEP); 904 mutex_enter(proc_lock); 905 } else 906 new_pgrp = NULL; 907 rval = EPERM; /* most common error (to save typing) */ 908 909 /* Check pgrp exists or can be created */ 910 pgrp = pid_table[pgid & pid_tbl_mask].pt_pgrp; 911 if (pgrp != NULL && pgrp->pg_id != pgid) 912 goto done; 913 914 /* Can only set another process under restricted circumstances. */ 915 if (pid != curp->p_pid) { 916 /* Must exist and be one of our children... */ 917 p = proc_find(pid); 918 if (p == NULL || !p_inferior(p, curp)) { 919 rval = ESRCH; 920 goto done; 921 } 922 /* ... in the same session... */ 923 if (sess != NULL || p->p_session != curp->p_session) 924 goto done; 925 /* ... existing pgid must be in same session ... */ 926 if (pgrp != NULL && pgrp->pg_session != p->p_session) 927 goto done; 928 /* ... and not done an exec. */ 929 if (p->p_flag & PK_EXEC) { 930 rval = EACCES; 931 goto done; 932 } 933 } else { 934 /* ... setsid() cannot re-enter a pgrp */ 935 if (mksess && (curp->p_pgid == curp->p_pid || 936 pgrp_find(curp->p_pid))) 937 goto done; 938 p = curp; 939 } 940 941 /* Changing the process group/session of a session 942 leader is definitely off limits. */ 943 if (SESS_LEADER(p)) { 944 if (sess == NULL && p->p_pgrp == pgrp) 945 /* unless it's a definite noop */ 946 rval = 0; 947 goto done; 948 } 949 950 /* Can only create a process group with id of process */ 951 if (pgrp == NULL && pgid != pid) 952 goto done; 953 954 /* Can only create a session if creating pgrp */ 955 if (sess != NULL && pgrp != NULL) 956 goto done; 957 958 /* Check we allocated memory for a pgrp... */ 959 if (pgrp == NULL && new_pgrp == NULL) 960 goto done; 961 962 /* Don't attach to 'zombie' pgrp */ 963 if (pgrp != NULL && LIST_EMPTY(&pgrp->pg_members)) 964 goto done; 965 966 /* Expect to succeed now */ 967 rval = 0; 968 969 if (pgrp == p->p_pgrp) 970 /* nothing to do */ 971 goto done; 972 973 /* Ok all setup, link up required structures */ 974 975 if (pgrp == NULL) { 976 pgrp = new_pgrp; 977 new_pgrp = NULL; 978 if (sess != NULL) { 979 sess->s_sid = p->p_pid; 980 sess->s_leader = p; 981 sess->s_count = 1; 982 sess->s_ttyvp = NULL; 983 sess->s_ttyp = NULL; 984 sess->s_flags = p->p_session->s_flags & ~S_LOGIN_SET; 985 memcpy(sess->s_login, p->p_session->s_login, 986 sizeof(sess->s_login)); 987 p->p_lflag &= ~PL_CONTROLT; 988 } else { 989 sess = p->p_pgrp->pg_session; 990 proc_sesshold(sess); 991 } 992 pgrp->pg_session = sess; 993 sess = NULL; 994 995 pgrp->pg_id = pgid; 996 LIST_INIT(&pgrp->pg_members); 997 #ifdef DIAGNOSTIC 998 if (__predict_false(pid_table[pgid & pid_tbl_mask].pt_pgrp)) 999 panic("enterpgrp: pgrp table slot in use"); 1000 if (__predict_false(mksess && p != curp)) 1001 panic("enterpgrp: mksession and p != curproc"); 1002 #endif 1003 pid_table[pgid & pid_tbl_mask].pt_pgrp = pgrp; 1004 pgrp->pg_jobc = 0; 1005 } 1006 1007 /* 1008 * Adjust eligibility of affected pgrps to participate in job control. 1009 * Increment eligibility counts before decrementing, otherwise we 1010 * could reach 0 spuriously during the first call. 1011 */ 1012 fixjobc(p, pgrp, 1); 1013 fixjobc(p, p->p_pgrp, 0); 1014 1015 /* Interlock with ttread(). */ 1016 mutex_spin_enter(&tty_lock); 1017 1018 /* Move process to requested group. */ 1019 LIST_REMOVE(p, p_pglist); 1020 if (LIST_EMPTY(&p->p_pgrp->pg_members)) 1021 /* defer delete until we've dumped the lock */ 1022 pg_id = p->p_pgrp->pg_id; 1023 p->p_pgrp = pgrp; 1024 LIST_INSERT_HEAD(&pgrp->pg_members, p, p_pglist); 1025 1026 /* Done with the swap; we can release the tty mutex. */ 1027 mutex_spin_exit(&tty_lock); 1028 1029 done: 1030 if (pg_id != NO_PGID) { 1031 /* Releases proc_lock. */ 1032 pg_delete(pg_id); 1033 } else { 1034 mutex_exit(proc_lock); 1035 } 1036 if (sess != NULL) 1037 kmem_free(sess, sizeof(*sess)); 1038 if (new_pgrp != NULL) 1039 kmem_free(new_pgrp, sizeof(*new_pgrp)); 1040 #ifdef DEBUG_PGRP 1041 if (__predict_false(rval)) 1042 printf("enterpgrp(%d,%d,%d), curproc %d, rval %d\n", 1043 pid, pgid, mksess, curp->p_pid, rval); 1044 #endif 1045 return rval; 1046 } 1047 1048 /* 1049 * proc_leavepgrp: remove a process from its process group. 1050 * => must be called with the proc_lock held, which will be released; 1051 */ 1052 void 1053 proc_leavepgrp(struct proc *p) 1054 { 1055 struct pgrp *pgrp; 1056 1057 KASSERT(mutex_owned(proc_lock)); 1058 1059 /* Interlock with ttread() */ 1060 mutex_spin_enter(&tty_lock); 1061 pgrp = p->p_pgrp; 1062 LIST_REMOVE(p, p_pglist); 1063 p->p_pgrp = NULL; 1064 mutex_spin_exit(&tty_lock); 1065 1066 if (LIST_EMPTY(&pgrp->pg_members)) { 1067 /* Releases proc_lock. */ 1068 pg_delete(pgrp->pg_id); 1069 } else { 1070 mutex_exit(proc_lock); 1071 } 1072 } 1073 1074 /* 1075 * pg_remove: remove a process group from the table. 1076 * => must be called with the proc_lock held; 1077 * => returns process group to free; 1078 */ 1079 static struct pgrp * 1080 pg_remove(pid_t pg_id) 1081 { 1082 struct pgrp *pgrp; 1083 struct pid_table *pt; 1084 1085 KASSERT(mutex_owned(proc_lock)); 1086 1087 pt = &pid_table[pg_id & pid_tbl_mask]; 1088 pgrp = pt->pt_pgrp; 1089 1090 KASSERT(pgrp != NULL); 1091 KASSERT(pgrp->pg_id == pg_id); 1092 KASSERT(LIST_EMPTY(&pgrp->pg_members)); 1093 1094 pt->pt_pgrp = NULL; 1095 1096 if (!P_VALID(pt->pt_proc)) { 1097 /* Orphaned pgrp, put slot onto free list. */ 1098 KASSERT((P_NEXT(pt->pt_proc) & pid_tbl_mask) == 0); 1099 pg_id &= pid_tbl_mask; 1100 pt = &pid_table[last_free_pt]; 1101 pt->pt_proc = P_FREE(P_NEXT(pt->pt_proc) | pg_id); 1102 KASSERT(pt->pt_pid == 0); 1103 last_free_pt = pg_id; 1104 pid_alloc_cnt--; 1105 } 1106 return pgrp; 1107 } 1108 1109 /* 1110 * pg_delete: delete and free a process group. 1111 * => must be called with the proc_lock held, which will be released. 1112 */ 1113 static void 1114 pg_delete(pid_t pg_id) 1115 { 1116 struct pgrp *pg; 1117 struct tty *ttyp; 1118 struct session *ss; 1119 1120 KASSERT(mutex_owned(proc_lock)); 1121 1122 pg = pid_table[pg_id & pid_tbl_mask].pt_pgrp; 1123 if (pg == NULL || pg->pg_id != pg_id || !LIST_EMPTY(&pg->pg_members)) { 1124 mutex_exit(proc_lock); 1125 return; 1126 } 1127 1128 ss = pg->pg_session; 1129 1130 /* Remove reference (if any) from tty to this process group */ 1131 mutex_spin_enter(&tty_lock); 1132 ttyp = ss->s_ttyp; 1133 if (ttyp != NULL && ttyp->t_pgrp == pg) { 1134 ttyp->t_pgrp = NULL; 1135 KASSERT(ttyp->t_session == ss); 1136 } 1137 mutex_spin_exit(&tty_lock); 1138 1139 /* 1140 * The leading process group in a session is freed by proc_sessrele(), 1141 * if last reference. Note: proc_sessrele() releases proc_lock. 1142 */ 1143 pg = (ss->s_sid != pg->pg_id) ? pg_remove(pg_id) : NULL; 1144 proc_sessrele(ss); 1145 1146 if (pg != NULL) { 1147 /* Free it, if was not done by proc_sessrele(). */ 1148 kmem_free(pg, sizeof(struct pgrp)); 1149 } 1150 } 1151 1152 /* 1153 * Adjust pgrp jobc counters when specified process changes process group. 1154 * We count the number of processes in each process group that "qualify" 1155 * the group for terminal job control (those with a parent in a different 1156 * process group of the same session). If that count reaches zero, the 1157 * process group becomes orphaned. Check both the specified process' 1158 * process group and that of its children. 1159 * entering == 0 => p is leaving specified group. 1160 * entering == 1 => p is entering specified group. 1161 * 1162 * Call with proc_lock held. 1163 */ 1164 void 1165 fixjobc(struct proc *p, struct pgrp *pgrp, int entering) 1166 { 1167 struct pgrp *hispgrp; 1168 struct session *mysession = pgrp->pg_session; 1169 struct proc *child; 1170 1171 KASSERT(mutex_owned(proc_lock)); 1172 1173 /* 1174 * Check p's parent to see whether p qualifies its own process 1175 * group; if so, adjust count for p's process group. 1176 */ 1177 hispgrp = p->p_pptr->p_pgrp; 1178 if (hispgrp != pgrp && hispgrp->pg_session == mysession) { 1179 if (entering) { 1180 pgrp->pg_jobc++; 1181 p->p_lflag &= ~PL_ORPHANPG; 1182 } else if (--pgrp->pg_jobc == 0) 1183 orphanpg(pgrp); 1184 } 1185 1186 /* 1187 * Check this process' children to see whether they qualify 1188 * their process groups; if so, adjust counts for children's 1189 * process groups. 1190 */ 1191 LIST_FOREACH(child, &p->p_children, p_sibling) { 1192 hispgrp = child->p_pgrp; 1193 if (hispgrp != pgrp && hispgrp->pg_session == mysession && 1194 !P_ZOMBIE(child)) { 1195 if (entering) { 1196 child->p_lflag &= ~PL_ORPHANPG; 1197 hispgrp->pg_jobc++; 1198 } else if (--hispgrp->pg_jobc == 0) 1199 orphanpg(hispgrp); 1200 } 1201 } 1202 } 1203 1204 /* 1205 * A process group has become orphaned; 1206 * if there are any stopped processes in the group, 1207 * hang-up all process in that group. 1208 * 1209 * Call with proc_lock held. 1210 */ 1211 static void 1212 orphanpg(struct pgrp *pg) 1213 { 1214 struct proc *p; 1215 1216 KASSERT(mutex_owned(proc_lock)); 1217 1218 LIST_FOREACH(p, &pg->pg_members, p_pglist) { 1219 if (p->p_stat == SSTOP) { 1220 p->p_lflag |= PL_ORPHANPG; 1221 psignal(p, SIGHUP); 1222 psignal(p, SIGCONT); 1223 } 1224 } 1225 } 1226 1227 #ifdef DDB 1228 #include <ddb/db_output.h> 1229 void pidtbl_dump(void); 1230 void 1231 pidtbl_dump(void) 1232 { 1233 struct pid_table *pt; 1234 struct proc *p; 1235 struct pgrp *pgrp; 1236 int id; 1237 1238 db_printf("pid table %p size %x, next %x, last %x\n", 1239 pid_table, pid_tbl_mask+1, 1240 next_free_pt, last_free_pt); 1241 for (pt = pid_table, id = 0; id <= pid_tbl_mask; id++, pt++) { 1242 p = pt->pt_proc; 1243 if (!P_VALID(p) && !pt->pt_pgrp) 1244 continue; 1245 db_printf(" id %x: ", id); 1246 if (P_VALID(p)) 1247 db_printf("slotpid %d proc %p id %d (0x%x) %s\n", 1248 pt->pt_pid, p, p->p_pid, p->p_pid, p->p_comm); 1249 else 1250 db_printf("next %x use %x\n", 1251 P_NEXT(p) & pid_tbl_mask, 1252 P_NEXT(p) & ~pid_tbl_mask); 1253 if ((pgrp = pt->pt_pgrp)) { 1254 db_printf("\tsession %p, sid %d, count %d, login %s\n", 1255 pgrp->pg_session, pgrp->pg_session->s_sid, 1256 pgrp->pg_session->s_count, 1257 pgrp->pg_session->s_login); 1258 db_printf("\tpgrp %p, pg_id %d, pg_jobc %d, members %p\n", 1259 pgrp, pgrp->pg_id, pgrp->pg_jobc, 1260 LIST_FIRST(&pgrp->pg_members)); 1261 LIST_FOREACH(p, &pgrp->pg_members, p_pglist) { 1262 db_printf("\t\tpid %d addr %p pgrp %p %s\n", 1263 p->p_pid, p, p->p_pgrp, p->p_comm); 1264 } 1265 } 1266 } 1267 } 1268 #endif /* DDB */ 1269 1270 #ifdef KSTACK_CHECK_MAGIC 1271 1272 #define KSTACK_MAGIC 0xdeadbeaf 1273 1274 /* XXX should be per process basis? */ 1275 static int kstackleftmin = KSTACK_SIZE; 1276 static int kstackleftthres = KSTACK_SIZE / 8; 1277 1278 void 1279 kstack_setup_magic(const struct lwp *l) 1280 { 1281 uint32_t *ip; 1282 uint32_t const *end; 1283 1284 KASSERT(l != NULL); 1285 KASSERT(l != &lwp0); 1286 1287 /* 1288 * fill all the stack with magic number 1289 * so that later modification on it can be detected. 1290 */ 1291 ip = (uint32_t *)KSTACK_LOWEST_ADDR(l); 1292 end = (uint32_t *)((char *)KSTACK_LOWEST_ADDR(l) + KSTACK_SIZE); 1293 for (; ip < end; ip++) { 1294 *ip = KSTACK_MAGIC; 1295 } 1296 } 1297 1298 void 1299 kstack_check_magic(const struct lwp *l) 1300 { 1301 uint32_t const *ip, *end; 1302 int stackleft; 1303 1304 KASSERT(l != NULL); 1305 1306 /* don't check proc0 */ /*XXX*/ 1307 if (l == &lwp0) 1308 return; 1309 1310 #ifdef __MACHINE_STACK_GROWS_UP 1311 /* stack grows upwards (eg. hppa) */ 1312 ip = (uint32_t *)((void *)KSTACK_LOWEST_ADDR(l) + KSTACK_SIZE); 1313 end = (uint32_t *)KSTACK_LOWEST_ADDR(l); 1314 for (ip--; ip >= end; ip--) 1315 if (*ip != KSTACK_MAGIC) 1316 break; 1317 1318 stackleft = (void *)KSTACK_LOWEST_ADDR(l) + KSTACK_SIZE - (void *)ip; 1319 #else /* __MACHINE_STACK_GROWS_UP */ 1320 /* stack grows downwards (eg. i386) */ 1321 ip = (uint32_t *)KSTACK_LOWEST_ADDR(l); 1322 end = (uint32_t *)((char *)KSTACK_LOWEST_ADDR(l) + KSTACK_SIZE); 1323 for (; ip < end; ip++) 1324 if (*ip != KSTACK_MAGIC) 1325 break; 1326 1327 stackleft = ((const char *)ip) - (const char *)KSTACK_LOWEST_ADDR(l); 1328 #endif /* __MACHINE_STACK_GROWS_UP */ 1329 1330 if (kstackleftmin > stackleft) { 1331 kstackleftmin = stackleft; 1332 if (stackleft < kstackleftthres) 1333 printf("warning: kernel stack left %d bytes" 1334 "(pid %u:lid %u)\n", stackleft, 1335 (u_int)l->l_proc->p_pid, (u_int)l->l_lid); 1336 } 1337 1338 if (stackleft <= 0) { 1339 panic("magic on the top of kernel stack changed for " 1340 "pid %u, lid %u: maybe kernel stack overflow", 1341 (u_int)l->l_proc->p_pid, (u_int)l->l_lid); 1342 } 1343 } 1344 #endif /* KSTACK_CHECK_MAGIC */ 1345 1346 int 1347 proclist_foreach_call(struct proclist *list, 1348 int (*callback)(struct proc *, void *arg), void *arg) 1349 { 1350 struct proc marker; 1351 struct proc *p; 1352 int ret = 0; 1353 1354 marker.p_flag = PK_MARKER; 1355 mutex_enter(proc_lock); 1356 for (p = LIST_FIRST(list); ret == 0 && p != NULL;) { 1357 if (p->p_flag & PK_MARKER) { 1358 p = LIST_NEXT(p, p_list); 1359 continue; 1360 } 1361 LIST_INSERT_AFTER(p, &marker, p_list); 1362 ret = (*callback)(p, arg); 1363 KASSERT(mutex_owned(proc_lock)); 1364 p = LIST_NEXT(&marker, p_list); 1365 LIST_REMOVE(&marker, p_list); 1366 } 1367 mutex_exit(proc_lock); 1368 1369 return ret; 1370 } 1371 1372 int 1373 proc_vmspace_getref(struct proc *p, struct vmspace **vm) 1374 { 1375 1376 /* XXXCDC: how should locking work here? */ 1377 1378 /* curproc exception is for coredump. */ 1379 1380 if ((p != curproc && (p->p_sflag & PS_WEXIT) != 0) || 1381 (p->p_vmspace->vm_refcnt < 1)) { /* XXX */ 1382 return EFAULT; 1383 } 1384 1385 uvmspace_addref(p->p_vmspace); 1386 *vm = p->p_vmspace; 1387 1388 return 0; 1389 } 1390 1391 /* 1392 * Acquire a write lock on the process credential. 1393 */ 1394 void 1395 proc_crmod_enter(void) 1396 { 1397 struct lwp *l = curlwp; 1398 struct proc *p = l->l_proc; 1399 kauth_cred_t oc; 1400 1401 /* Reset what needs to be reset in plimit. */ 1402 if (p->p_limit->pl_corename != defcorename) { 1403 lim_setcorename(p, defcorename, 0); 1404 } 1405 1406 mutex_enter(p->p_lock); 1407 1408 /* Ensure the LWP cached credentials are up to date. */ 1409 if ((oc = l->l_cred) != p->p_cred) { 1410 kauth_cred_hold(p->p_cred); 1411 l->l_cred = p->p_cred; 1412 kauth_cred_free(oc); 1413 } 1414 } 1415 1416 /* 1417 * Set in a new process credential, and drop the write lock. The credential 1418 * must have a reference already. Optionally, free a no-longer required 1419 * credential. The scheduler also needs to inspect p_cred, so we also 1420 * briefly acquire the sched state mutex. 1421 */ 1422 void 1423 proc_crmod_leave(kauth_cred_t scred, kauth_cred_t fcred, bool sugid) 1424 { 1425 struct lwp *l = curlwp, *l2; 1426 struct proc *p = l->l_proc; 1427 kauth_cred_t oc; 1428 1429 KASSERT(mutex_owned(p->p_lock)); 1430 1431 /* Is there a new credential to set in? */ 1432 if (scred != NULL) { 1433 p->p_cred = scred; 1434 LIST_FOREACH(l2, &p->p_lwps, l_sibling) { 1435 if (l2 != l) 1436 l2->l_prflag |= LPR_CRMOD; 1437 } 1438 1439 /* Ensure the LWP cached credentials are up to date. */ 1440 if ((oc = l->l_cred) != scred) { 1441 kauth_cred_hold(scred); 1442 l->l_cred = scred; 1443 } 1444 } else 1445 oc = NULL; /* XXXgcc */ 1446 1447 if (sugid) { 1448 /* 1449 * Mark process as having changed credentials, stops 1450 * tracing etc. 1451 */ 1452 p->p_flag |= PK_SUGID; 1453 } 1454 1455 mutex_exit(p->p_lock); 1456 1457 /* If there is a credential to be released, free it now. */ 1458 if (fcred != NULL) { 1459 KASSERT(scred != NULL); 1460 kauth_cred_free(fcred); 1461 if (oc != scred) 1462 kauth_cred_free(oc); 1463 } 1464 } 1465 1466 /* 1467 * proc_specific_key_create -- 1468 * Create a key for subsystem proc-specific data. 1469 */ 1470 int 1471 proc_specific_key_create(specificdata_key_t *keyp, specificdata_dtor_t dtor) 1472 { 1473 1474 return (specificdata_key_create(proc_specificdata_domain, keyp, dtor)); 1475 } 1476 1477 /* 1478 * proc_specific_key_delete -- 1479 * Delete a key for subsystem proc-specific data. 1480 */ 1481 void 1482 proc_specific_key_delete(specificdata_key_t key) 1483 { 1484 1485 specificdata_key_delete(proc_specificdata_domain, key); 1486 } 1487 1488 /* 1489 * proc_initspecific -- 1490 * Initialize a proc's specificdata container. 1491 */ 1492 void 1493 proc_initspecific(struct proc *p) 1494 { 1495 int error __diagused; 1496 1497 error = specificdata_init(proc_specificdata_domain, &p->p_specdataref); 1498 KASSERT(error == 0); 1499 } 1500 1501 /* 1502 * proc_finispecific -- 1503 * Finalize a proc's specificdata container. 1504 */ 1505 void 1506 proc_finispecific(struct proc *p) 1507 { 1508 1509 specificdata_fini(proc_specificdata_domain, &p->p_specdataref); 1510 } 1511 1512 /* 1513 * proc_getspecific -- 1514 * Return proc-specific data corresponding to the specified key. 1515 */ 1516 void * 1517 proc_getspecific(struct proc *p, specificdata_key_t key) 1518 { 1519 1520 return (specificdata_getspecific(proc_specificdata_domain, 1521 &p->p_specdataref, key)); 1522 } 1523 1524 /* 1525 * proc_setspecific -- 1526 * Set proc-specific data corresponding to the specified key. 1527 */ 1528 void 1529 proc_setspecific(struct proc *p, specificdata_key_t key, void *data) 1530 { 1531 1532 specificdata_setspecific(proc_specificdata_domain, 1533 &p->p_specdataref, key, data); 1534 } 1535 1536 int 1537 proc_uidmatch(kauth_cred_t cred, kauth_cred_t target) 1538 { 1539 int r = 0; 1540 1541 if (kauth_cred_getuid(cred) != kauth_cred_getuid(target) || 1542 kauth_cred_getuid(cred) != kauth_cred_getsvuid(target)) { 1543 /* 1544 * suid proc of ours or proc not ours 1545 */ 1546 r = EPERM; 1547 } else if (kauth_cred_getgid(target) != kauth_cred_getsvgid(target)) { 1548 /* 1549 * sgid proc has sgid back to us temporarily 1550 */ 1551 r = EPERM; 1552 } else { 1553 /* 1554 * our rgid must be in target's group list (ie, 1555 * sub-processes started by a sgid process) 1556 */ 1557 int ismember = 0; 1558 1559 if (kauth_cred_ismember_gid(cred, 1560 kauth_cred_getgid(target), &ismember) != 0 || 1561 !ismember) 1562 r = EPERM; 1563 } 1564 1565 return (r); 1566 } 1567 1568 /* 1569 * sysctl stuff 1570 */ 1571 1572 #define KERN_PROCSLOP (5 * sizeof(struct kinfo_proc)) 1573 1574 static const u_int sysctl_flagmap[] = { 1575 PK_ADVLOCK, P_ADVLOCK, 1576 PK_EXEC, P_EXEC, 1577 PK_NOCLDWAIT, P_NOCLDWAIT, 1578 PK_32, P_32, 1579 PK_CLDSIGIGN, P_CLDSIGIGN, 1580 PK_SUGID, P_SUGID, 1581 0 1582 }; 1583 1584 static const u_int sysctl_sflagmap[] = { 1585 PS_NOCLDSTOP, P_NOCLDSTOP, 1586 PS_WEXIT, P_WEXIT, 1587 PS_STOPFORK, P_STOPFORK, 1588 PS_STOPEXEC, P_STOPEXEC, 1589 PS_STOPEXIT, P_STOPEXIT, 1590 0 1591 }; 1592 1593 static const u_int sysctl_slflagmap[] = { 1594 PSL_TRACED, P_TRACED, 1595 PSL_CHTRACED, P_CHTRACED, 1596 PSL_SYSCALL, P_SYSCALL, 1597 0 1598 }; 1599 1600 static const u_int sysctl_lflagmap[] = { 1601 PL_CONTROLT, P_CONTROLT, 1602 PL_PPWAIT, P_PPWAIT, 1603 0 1604 }; 1605 1606 static const u_int sysctl_stflagmap[] = { 1607 PST_PROFIL, P_PROFIL, 1608 0 1609 1610 }; 1611 1612 /* used by kern_lwp also */ 1613 const u_int sysctl_lwpflagmap[] = { 1614 LW_SINTR, L_SINTR, 1615 LW_SYSTEM, L_SYSTEM, 1616 0 1617 }; 1618 1619 /* 1620 * Find the most ``active'' lwp of a process and return it for ps display 1621 * purposes 1622 */ 1623 static struct lwp * 1624 proc_active_lwp(struct proc *p) 1625 { 1626 static const int ostat[] = { 1627 0, 1628 2, /* LSIDL */ 1629 6, /* LSRUN */ 1630 5, /* LSSLEEP */ 1631 4, /* LSSTOP */ 1632 0, /* LSZOMB */ 1633 1, /* LSDEAD */ 1634 7, /* LSONPROC */ 1635 3 /* LSSUSPENDED */ 1636 }; 1637 1638 struct lwp *l, *lp = NULL; 1639 LIST_FOREACH(l, &p->p_lwps, l_sibling) { 1640 KASSERT(l->l_stat >= 0 && l->l_stat < __arraycount(ostat)); 1641 if (lp == NULL || 1642 ostat[l->l_stat] > ostat[lp->l_stat] || 1643 (ostat[l->l_stat] == ostat[lp->l_stat] && 1644 l->l_cpticks > lp->l_cpticks)) { 1645 lp = l; 1646 continue; 1647 } 1648 } 1649 return lp; 1650 } 1651 1652 static int 1653 sysctl_doeproc(SYSCTLFN_ARGS) 1654 { 1655 union { 1656 struct kinfo_proc kproc; 1657 struct kinfo_proc2 kproc2; 1658 } *kbuf; 1659 struct proc *p, *next, *marker; 1660 char *where, *dp; 1661 int type, op, arg, error; 1662 u_int elem_size, kelem_size, elem_count; 1663 size_t buflen, needed; 1664 bool match, zombie, mmmbrains; 1665 const bool allowaddr = get_expose_address(curproc); 1666 1667 if (namelen == 1 && name[0] == CTL_QUERY) 1668 return (sysctl_query(SYSCTLFN_CALL(rnode))); 1669 1670 dp = where = oldp; 1671 buflen = where != NULL ? *oldlenp : 0; 1672 error = 0; 1673 needed = 0; 1674 type = rnode->sysctl_num; 1675 1676 if (type == KERN_PROC) { 1677 if (namelen == 0) 1678 return EINVAL; 1679 switch (op = name[0]) { 1680 case KERN_PROC_ALL: 1681 if (namelen != 1) 1682 return EINVAL; 1683 arg = 0; 1684 break; 1685 default: 1686 if (namelen != 2) 1687 return EINVAL; 1688 arg = name[1]; 1689 break; 1690 } 1691 elem_count = 0; /* Hush little compiler, don't you cry */ 1692 kelem_size = elem_size = sizeof(kbuf->kproc); 1693 } else { 1694 if (namelen != 4) 1695 return EINVAL; 1696 op = name[0]; 1697 arg = name[1]; 1698 elem_size = name[2]; 1699 elem_count = name[3]; 1700 kelem_size = sizeof(kbuf->kproc2); 1701 } 1702 1703 sysctl_unlock(); 1704 1705 kbuf = kmem_zalloc(sizeof(*kbuf), KM_SLEEP); 1706 marker = kmem_alloc(sizeof(*marker), KM_SLEEP); 1707 marker->p_flag = PK_MARKER; 1708 1709 mutex_enter(proc_lock); 1710 /* 1711 * Start with zombies to prevent reporting processes twice, in case they 1712 * are dying and being moved from the list of alive processes to zombies. 1713 */ 1714 mmmbrains = true; 1715 for (p = LIST_FIRST(&zombproc);; p = next) { 1716 if (p == NULL) { 1717 if (mmmbrains) { 1718 p = LIST_FIRST(&allproc); 1719 mmmbrains = false; 1720 } 1721 if (p == NULL) 1722 break; 1723 } 1724 next = LIST_NEXT(p, p_list); 1725 if ((p->p_flag & PK_MARKER) != 0) 1726 continue; 1727 1728 /* 1729 * Skip embryonic processes. 1730 */ 1731 if (p->p_stat == SIDL) 1732 continue; 1733 1734 mutex_enter(p->p_lock); 1735 error = kauth_authorize_process(l->l_cred, 1736 KAUTH_PROCESS_CANSEE, p, 1737 KAUTH_ARG(KAUTH_REQ_PROCESS_CANSEE_EPROC), NULL, NULL); 1738 if (error != 0) { 1739 mutex_exit(p->p_lock); 1740 continue; 1741 } 1742 1743 /* 1744 * Hande all the operations in one switch on the cost of 1745 * algorithm complexity is on purpose. The win splitting this 1746 * function into several similar copies makes maintenance burden 1747 * burden, code grow and boost is neglible in practical systems. 1748 */ 1749 switch (op) { 1750 case KERN_PROC_PID: 1751 match = (p->p_pid == (pid_t)arg); 1752 break; 1753 1754 case KERN_PROC_PGRP: 1755 match = (p->p_pgrp->pg_id == (pid_t)arg); 1756 break; 1757 1758 case KERN_PROC_SESSION: 1759 match = (p->p_session->s_sid == (pid_t)arg); 1760 break; 1761 1762 case KERN_PROC_TTY: 1763 match = true; 1764 if (arg == (int) KERN_PROC_TTY_REVOKE) { 1765 if ((p->p_lflag & PL_CONTROLT) == 0 || 1766 p->p_session->s_ttyp == NULL || 1767 p->p_session->s_ttyvp != NULL) { 1768 match = false; 1769 } 1770 } else if ((p->p_lflag & PL_CONTROLT) == 0 || 1771 p->p_session->s_ttyp == NULL) { 1772 if ((dev_t)arg != KERN_PROC_TTY_NODEV) { 1773 match = false; 1774 } 1775 } else if (p->p_session->s_ttyp->t_dev != (dev_t)arg) { 1776 match = false; 1777 } 1778 break; 1779 1780 case KERN_PROC_UID: 1781 match = (kauth_cred_geteuid(p->p_cred) == (uid_t)arg); 1782 break; 1783 1784 case KERN_PROC_RUID: 1785 match = (kauth_cred_getuid(p->p_cred) == (uid_t)arg); 1786 break; 1787 1788 case KERN_PROC_GID: 1789 match = (kauth_cred_getegid(p->p_cred) == (uid_t)arg); 1790 break; 1791 1792 case KERN_PROC_RGID: 1793 match = (kauth_cred_getgid(p->p_cred) == (uid_t)arg); 1794 break; 1795 1796 case KERN_PROC_ALL: 1797 match = true; 1798 /* allow everything */ 1799 break; 1800 1801 default: 1802 error = EINVAL; 1803 mutex_exit(p->p_lock); 1804 goto cleanup; 1805 } 1806 if (!match) { 1807 mutex_exit(p->p_lock); 1808 continue; 1809 } 1810 1811 /* 1812 * Grab a hold on the process. 1813 */ 1814 if (mmmbrains) { 1815 zombie = true; 1816 } else { 1817 zombie = !rw_tryenter(&p->p_reflock, RW_READER); 1818 } 1819 if (zombie) { 1820 LIST_INSERT_AFTER(p, marker, p_list); 1821 } 1822 1823 if (buflen >= elem_size && 1824 (type == KERN_PROC || elem_count > 0)) { 1825 if (type == KERN_PROC) { 1826 fill_proc(p, &kbuf->kproc.kp_proc, allowaddr); 1827 fill_eproc(p, &kbuf->kproc.kp_eproc, zombie, 1828 allowaddr); 1829 } else { 1830 fill_kproc2(p, &kbuf->kproc2, zombie, 1831 allowaddr); 1832 elem_count--; 1833 } 1834 mutex_exit(p->p_lock); 1835 mutex_exit(proc_lock); 1836 /* 1837 * Copy out elem_size, but not larger than kelem_size 1838 */ 1839 error = sysctl_copyout(l, kbuf, dp, 1840 uimin(kelem_size, elem_size)); 1841 mutex_enter(proc_lock); 1842 if (error) { 1843 goto bah; 1844 } 1845 dp += elem_size; 1846 buflen -= elem_size; 1847 } else { 1848 mutex_exit(p->p_lock); 1849 } 1850 needed += elem_size; 1851 1852 /* 1853 * Release reference to process. 1854 */ 1855 if (zombie) { 1856 next = LIST_NEXT(marker, p_list); 1857 LIST_REMOVE(marker, p_list); 1858 } else { 1859 rw_exit(&p->p_reflock); 1860 next = LIST_NEXT(p, p_list); 1861 } 1862 1863 /* 1864 * Short-circuit break quickly! 1865 */ 1866 if (op == KERN_PROC_PID) 1867 break; 1868 } 1869 mutex_exit(proc_lock); 1870 1871 if (where != NULL) { 1872 *oldlenp = dp - where; 1873 if (needed > *oldlenp) { 1874 error = ENOMEM; 1875 goto out; 1876 } 1877 } else { 1878 needed += KERN_PROCSLOP; 1879 *oldlenp = needed; 1880 } 1881 kmem_free(kbuf, sizeof(*kbuf)); 1882 kmem_free(marker, sizeof(*marker)); 1883 sysctl_relock(); 1884 return 0; 1885 bah: 1886 if (zombie) 1887 LIST_REMOVE(marker, p_list); 1888 else 1889 rw_exit(&p->p_reflock); 1890 cleanup: 1891 mutex_exit(proc_lock); 1892 out: 1893 kmem_free(kbuf, sizeof(*kbuf)); 1894 kmem_free(marker, sizeof(*marker)); 1895 sysctl_relock(); 1896 return error; 1897 } 1898 1899 int 1900 copyin_psstrings(struct proc *p, struct ps_strings *arginfo) 1901 { 1902 1903 #ifdef COMPAT_NETBSD32 1904 if (p->p_flag & PK_32) { 1905 struct ps_strings32 arginfo32; 1906 1907 int error = copyin_proc(p, (void *)p->p_psstrp, &arginfo32, 1908 sizeof(arginfo32)); 1909 if (error) 1910 return error; 1911 arginfo->ps_argvstr = (void *)(uintptr_t)arginfo32.ps_argvstr; 1912 arginfo->ps_nargvstr = arginfo32.ps_nargvstr; 1913 arginfo->ps_envstr = (void *)(uintptr_t)arginfo32.ps_envstr; 1914 arginfo->ps_nenvstr = arginfo32.ps_nenvstr; 1915 return 0; 1916 } 1917 #endif 1918 return copyin_proc(p, (void *)p->p_psstrp, arginfo, sizeof(*arginfo)); 1919 } 1920 1921 static int 1922 copy_procargs_sysctl_cb(void *cookie_, const void *src, size_t off, size_t len) 1923 { 1924 void **cookie = cookie_; 1925 struct lwp *l = cookie[0]; 1926 char *dst = cookie[1]; 1927 1928 return sysctl_copyout(l, src, dst + off, len); 1929 } 1930 1931 /* 1932 * sysctl helper routine for kern.proc_args pseudo-subtree. 1933 */ 1934 static int 1935 sysctl_kern_proc_args(SYSCTLFN_ARGS) 1936 { 1937 struct ps_strings pss; 1938 struct proc *p; 1939 pid_t pid; 1940 int type, error; 1941 void *cookie[2]; 1942 1943 if (namelen == 1 && name[0] == CTL_QUERY) 1944 return (sysctl_query(SYSCTLFN_CALL(rnode))); 1945 1946 if (newp != NULL || namelen != 2) 1947 return (EINVAL); 1948 pid = name[0]; 1949 type = name[1]; 1950 1951 switch (type) { 1952 case KERN_PROC_PATHNAME: 1953 sysctl_unlock(); 1954 error = fill_pathname(l, pid, oldp, oldlenp); 1955 sysctl_relock(); 1956 return error; 1957 1958 case KERN_PROC_ARGV: 1959 case KERN_PROC_NARGV: 1960 case KERN_PROC_ENV: 1961 case KERN_PROC_NENV: 1962 /* ok */ 1963 break; 1964 default: 1965 return (EINVAL); 1966 } 1967 1968 sysctl_unlock(); 1969 1970 /* check pid */ 1971 mutex_enter(proc_lock); 1972 if ((p = proc_find(pid)) == NULL) { 1973 error = EINVAL; 1974 goto out_locked; 1975 } 1976 mutex_enter(p->p_lock); 1977 1978 /* Check permission. */ 1979 if (type == KERN_PROC_ARGV || type == KERN_PROC_NARGV) 1980 error = kauth_authorize_process(l->l_cred, KAUTH_PROCESS_CANSEE, 1981 p, KAUTH_ARG(KAUTH_REQ_PROCESS_CANSEE_ARGS), NULL, NULL); 1982 else if (type == KERN_PROC_ENV || type == KERN_PROC_NENV) 1983 error = kauth_authorize_process(l->l_cred, KAUTH_PROCESS_CANSEE, 1984 p, KAUTH_ARG(KAUTH_REQ_PROCESS_CANSEE_ENV), NULL, NULL); 1985 else 1986 error = EINVAL; /* XXXGCC */ 1987 if (error) { 1988 mutex_exit(p->p_lock); 1989 goto out_locked; 1990 } 1991 1992 if (oldp == NULL) { 1993 if (type == KERN_PROC_NARGV || type == KERN_PROC_NENV) 1994 *oldlenp = sizeof (int); 1995 else 1996 *oldlenp = ARG_MAX; /* XXX XXX XXX */ 1997 error = 0; 1998 mutex_exit(p->p_lock); 1999 goto out_locked; 2000 } 2001 2002 /* 2003 * Zombies don't have a stack, so we can't read their psstrings. 2004 * System processes also don't have a user stack. 2005 */ 2006 if (P_ZOMBIE(p) || (p->p_flag & PK_SYSTEM) != 0) { 2007 error = EINVAL; 2008 mutex_exit(p->p_lock); 2009 goto out_locked; 2010 } 2011 2012 error = rw_tryenter(&p->p_reflock, RW_READER) ? 0 : EBUSY; 2013 mutex_exit(p->p_lock); 2014 if (error) { 2015 goto out_locked; 2016 } 2017 mutex_exit(proc_lock); 2018 2019 if (type == KERN_PROC_NARGV || type == KERN_PROC_NENV) { 2020 int value; 2021 if ((error = copyin_psstrings(p, &pss)) == 0) { 2022 if (type == KERN_PROC_NARGV) 2023 value = pss.ps_nargvstr; 2024 else 2025 value = pss.ps_nenvstr; 2026 error = sysctl_copyout(l, &value, oldp, sizeof(value)); 2027 *oldlenp = sizeof(value); 2028 } 2029 } else { 2030 cookie[0] = l; 2031 cookie[1] = oldp; 2032 error = copy_procargs(p, type, oldlenp, 2033 copy_procargs_sysctl_cb, cookie); 2034 } 2035 rw_exit(&p->p_reflock); 2036 sysctl_relock(); 2037 return error; 2038 2039 out_locked: 2040 mutex_exit(proc_lock); 2041 sysctl_relock(); 2042 return error; 2043 } 2044 2045 int 2046 copy_procargs(struct proc *p, int oid, size_t *limit, 2047 int (*cb)(void *, const void *, size_t, size_t), void *cookie) 2048 { 2049 struct ps_strings pss; 2050 size_t len, i, loaded, entry_len; 2051 struct uio auio; 2052 struct iovec aiov; 2053 int error, argvlen; 2054 char *arg; 2055 char **argv; 2056 vaddr_t user_argv; 2057 struct vmspace *vmspace; 2058 2059 /* 2060 * Allocate a temporary buffer to hold the argument vector and 2061 * the arguments themselve. 2062 */ 2063 arg = kmem_alloc(PAGE_SIZE, KM_SLEEP); 2064 argv = kmem_alloc(PAGE_SIZE, KM_SLEEP); 2065 2066 /* 2067 * Lock the process down in memory. 2068 */ 2069 vmspace = p->p_vmspace; 2070 uvmspace_addref(vmspace); 2071 2072 /* 2073 * Read in the ps_strings structure. 2074 */ 2075 if ((error = copyin_psstrings(p, &pss)) != 0) 2076 goto done; 2077 2078 /* 2079 * Now read the address of the argument vector. 2080 */ 2081 switch (oid) { 2082 case KERN_PROC_ARGV: 2083 user_argv = (uintptr_t)pss.ps_argvstr; 2084 argvlen = pss.ps_nargvstr; 2085 break; 2086 case KERN_PROC_ENV: 2087 user_argv = (uintptr_t)pss.ps_envstr; 2088 argvlen = pss.ps_nenvstr; 2089 break; 2090 default: 2091 error = EINVAL; 2092 goto done; 2093 } 2094 2095 if (argvlen < 0) { 2096 error = EIO; 2097 goto done; 2098 } 2099 2100 2101 /* 2102 * Now copy each string. 2103 */ 2104 len = 0; /* bytes written to user buffer */ 2105 loaded = 0; /* bytes from argv already processed */ 2106 i = 0; /* To make compiler happy */ 2107 entry_len = PROC_PTRSZ(p); 2108 2109 for (; argvlen; --argvlen) { 2110 int finished = 0; 2111 vaddr_t base; 2112 size_t xlen; 2113 int j; 2114 2115 if (loaded == 0) { 2116 size_t rem = entry_len * argvlen; 2117 loaded = MIN(rem, PAGE_SIZE); 2118 error = copyin_vmspace(vmspace, 2119 (const void *)user_argv, argv, loaded); 2120 if (error) 2121 break; 2122 user_argv += loaded; 2123 i = 0; 2124 } 2125 2126 #ifdef COMPAT_NETBSD32 2127 if (p->p_flag & PK_32) { 2128 netbsd32_charp *argv32; 2129 2130 argv32 = (netbsd32_charp *)argv; 2131 base = (vaddr_t)NETBSD32PTR64(argv32[i++]); 2132 } else 2133 #endif 2134 base = (vaddr_t)argv[i++]; 2135 loaded -= entry_len; 2136 2137 /* 2138 * The program has messed around with its arguments, 2139 * possibly deleting some, and replacing them with 2140 * NULL's. Treat this as the last argument and not 2141 * a failure. 2142 */ 2143 if (base == 0) 2144 break; 2145 2146 while (!finished) { 2147 xlen = PAGE_SIZE - (base & PAGE_MASK); 2148 2149 aiov.iov_base = arg; 2150 aiov.iov_len = PAGE_SIZE; 2151 auio.uio_iov = &aiov; 2152 auio.uio_iovcnt = 1; 2153 auio.uio_offset = base; 2154 auio.uio_resid = xlen; 2155 auio.uio_rw = UIO_READ; 2156 UIO_SETUP_SYSSPACE(&auio); 2157 error = uvm_io(&vmspace->vm_map, &auio, 0); 2158 if (error) 2159 goto done; 2160 2161 /* Look for the end of the string */ 2162 for (j = 0; j < xlen; j++) { 2163 if (arg[j] == '\0') { 2164 xlen = j + 1; 2165 finished = 1; 2166 break; 2167 } 2168 } 2169 2170 /* Check for user buffer overflow */ 2171 if (len + xlen > *limit) { 2172 finished = 1; 2173 if (len > *limit) 2174 xlen = 0; 2175 else 2176 xlen = *limit - len; 2177 } 2178 2179 /* Copyout the page */ 2180 error = (*cb)(cookie, arg, len, xlen); 2181 if (error) 2182 goto done; 2183 2184 len += xlen; 2185 base += xlen; 2186 } 2187 } 2188 *limit = len; 2189 2190 done: 2191 kmem_free(argv, PAGE_SIZE); 2192 kmem_free(arg, PAGE_SIZE); 2193 uvmspace_free(vmspace); 2194 return error; 2195 } 2196 2197 /* 2198 * Fill in a proc structure for the specified process. 2199 */ 2200 static void 2201 fill_proc(const struct proc *psrc, struct proc *p, bool allowaddr) 2202 { 2203 COND_SET_VALUE(p->p_list, psrc->p_list, allowaddr); 2204 COND_SET_VALUE(p->p_auxlock, psrc->p_auxlock, allowaddr); 2205 COND_SET_VALUE(p->p_lock, psrc->p_lock, allowaddr); 2206 COND_SET_VALUE(p->p_stmutex, psrc->p_stmutex, allowaddr); 2207 COND_SET_VALUE(p->p_reflock, psrc->p_reflock, allowaddr); 2208 COND_SET_VALUE(p->p_waitcv, psrc->p_waitcv, allowaddr); 2209 COND_SET_VALUE(p->p_lwpcv, psrc->p_lwpcv, allowaddr); 2210 COND_SET_VALUE(p->p_cred, psrc->p_cred, allowaddr); 2211 COND_SET_VALUE(p->p_fd, psrc->p_fd, allowaddr); 2212 COND_SET_VALUE(p->p_cwdi, psrc->p_cwdi, allowaddr); 2213 COND_SET_VALUE(p->p_stats, psrc->p_stats, allowaddr); 2214 COND_SET_VALUE(p->p_limit, psrc->p_limit, allowaddr); 2215 COND_SET_VALUE(p->p_vmspace, psrc->p_vmspace, allowaddr); 2216 COND_SET_VALUE(p->p_sigacts, psrc->p_sigacts, allowaddr); 2217 COND_SET_VALUE(p->p_aio, psrc->p_aio, allowaddr); 2218 p->p_mqueue_cnt = psrc->p_mqueue_cnt; 2219 COND_SET_VALUE(p->p_specdataref, psrc->p_specdataref, allowaddr); 2220 p->p_exitsig = psrc->p_exitsig; 2221 p->p_flag = psrc->p_flag; 2222 p->p_sflag = psrc->p_sflag; 2223 p->p_slflag = psrc->p_slflag; 2224 p->p_lflag = psrc->p_lflag; 2225 p->p_stflag = psrc->p_stflag; 2226 p->p_stat = psrc->p_stat; 2227 p->p_trace_enabled = psrc->p_trace_enabled; 2228 p->p_pid = psrc->p_pid; 2229 COND_SET_VALUE(p->p_pglist, psrc->p_pglist, allowaddr); 2230 COND_SET_VALUE(p->p_pptr, psrc->p_pptr, allowaddr); 2231 COND_SET_VALUE(p->p_sibling, psrc->p_sibling, allowaddr); 2232 COND_SET_VALUE(p->p_children, psrc->p_children, allowaddr); 2233 COND_SET_VALUE(p->p_lwps, psrc->p_lwps, allowaddr); 2234 COND_SET_VALUE(p->p_raslist, psrc->p_raslist, allowaddr); 2235 p->p_nlwps = psrc->p_nlwps; 2236 p->p_nzlwps = psrc->p_nzlwps; 2237 p->p_nrlwps = psrc->p_nrlwps; 2238 p->p_nlwpwait = psrc->p_nlwpwait; 2239 p->p_ndlwps = psrc->p_ndlwps; 2240 p->p_nlwpid = psrc->p_nlwpid; 2241 p->p_nstopchild = psrc->p_nstopchild; 2242 p->p_waited = psrc->p_waited; 2243 COND_SET_VALUE(p->p_zomblwp, psrc->p_zomblwp, allowaddr); 2244 COND_SET_VALUE(p->p_vforklwp, psrc->p_vforklwp, allowaddr); 2245 COND_SET_VALUE(p->p_sched_info, psrc->p_sched_info, allowaddr); 2246 p->p_estcpu = psrc->p_estcpu; 2247 p->p_estcpu_inherited = psrc->p_estcpu_inherited; 2248 p->p_forktime = psrc->p_forktime; 2249 p->p_pctcpu = psrc->p_pctcpu; 2250 COND_SET_VALUE(p->p_opptr, psrc->p_opptr, allowaddr); 2251 COND_SET_VALUE(p->p_timers, psrc->p_timers, allowaddr); 2252 p->p_rtime = psrc->p_rtime; 2253 p->p_uticks = psrc->p_uticks; 2254 p->p_sticks = psrc->p_sticks; 2255 p->p_iticks = psrc->p_iticks; 2256 p->p_xutime = psrc->p_xutime; 2257 p->p_xstime = psrc->p_xstime; 2258 p->p_traceflag = psrc->p_traceflag; 2259 COND_SET_VALUE(p->p_tracep, psrc->p_tracep, allowaddr); 2260 COND_SET_VALUE(p->p_textvp, psrc->p_textvp, allowaddr); 2261 COND_SET_VALUE(p->p_emul, psrc->p_emul, allowaddr); 2262 COND_SET_VALUE(p->p_emuldata, psrc->p_emuldata, allowaddr); 2263 COND_SET_VALUE(p->p_execsw, psrc->p_execsw, allowaddr); 2264 COND_SET_VALUE(p->p_klist, psrc->p_klist, allowaddr); 2265 COND_SET_VALUE(p->p_sigwaiters, psrc->p_sigwaiters, allowaddr); 2266 COND_SET_VALUE(p->p_sigpend, psrc->p_sigpend, allowaddr); 2267 COND_SET_VALUE(p->p_lwpctl, psrc->p_lwpctl, allowaddr); 2268 p->p_ppid = psrc->p_ppid; 2269 p->p_fpid = psrc->p_fpid; 2270 p->p_vfpid = psrc->p_vfpid; 2271 p->p_vfpid_done = psrc->p_vfpid_done; 2272 p->p_lwp_created = psrc->p_lwp_created; 2273 p->p_lwp_exited = psrc->p_lwp_exited; 2274 p->p_nsems = psrc->p_nsems; 2275 COND_SET_VALUE(p->p_path, psrc->p_path, allowaddr); 2276 COND_SET_VALUE(p->p_sigctx, psrc->p_sigctx, allowaddr); 2277 p->p_nice = psrc->p_nice; 2278 memcpy(p->p_comm, psrc->p_comm, sizeof(p->p_comm)); 2279 COND_SET_VALUE(p->p_pgrp, psrc->p_pgrp, allowaddr); 2280 COND_SET_VALUE(p->p_psstrp, psrc->p_psstrp, allowaddr); 2281 p->p_pax = psrc->p_pax; 2282 p->p_xexit = psrc->p_xexit; 2283 p->p_xsig = psrc->p_xsig; 2284 p->p_acflag = psrc->p_acflag; 2285 COND_SET_VALUE(p->p_md, psrc->p_md, allowaddr); 2286 p->p_stackbase = psrc->p_stackbase; 2287 COND_SET_VALUE(p->p_dtrace, psrc->p_dtrace, allowaddr); 2288 } 2289 2290 /* 2291 * Fill in an eproc structure for the specified process. 2292 */ 2293 void 2294 fill_eproc(struct proc *p, struct eproc *ep, bool zombie, bool allowaddr) 2295 { 2296 struct tty *tp; 2297 struct lwp *l; 2298 2299 KASSERT(mutex_owned(proc_lock)); 2300 KASSERT(mutex_owned(p->p_lock)); 2301 2302 COND_SET_VALUE(ep->e_paddr, p, allowaddr); 2303 COND_SET_VALUE(ep->e_sess, p->p_session, allowaddr); 2304 if (p->p_cred) { 2305 kauth_cred_topcred(p->p_cred, &ep->e_pcred); 2306 kauth_cred_toucred(p->p_cred, &ep->e_ucred); 2307 } 2308 if (p->p_stat != SIDL && !P_ZOMBIE(p) && !zombie) { 2309 struct vmspace *vm = p->p_vmspace; 2310 2311 ep->e_vm.vm_rssize = vm_resident_count(vm); 2312 ep->e_vm.vm_tsize = vm->vm_tsize; 2313 ep->e_vm.vm_dsize = vm->vm_dsize; 2314 ep->e_vm.vm_ssize = vm->vm_ssize; 2315 ep->e_vm.vm_map.size = vm->vm_map.size; 2316 2317 /* Pick the primary (first) LWP */ 2318 l = proc_active_lwp(p); 2319 KASSERT(l != NULL); 2320 lwp_lock(l); 2321 if (l->l_wchan) 2322 strncpy(ep->e_wmesg, l->l_wmesg, WMESGLEN); 2323 lwp_unlock(l); 2324 } 2325 ep->e_ppid = p->p_ppid; 2326 if (p->p_pgrp && p->p_session) { 2327 ep->e_pgid = p->p_pgrp->pg_id; 2328 ep->e_jobc = p->p_pgrp->pg_jobc; 2329 ep->e_sid = p->p_session->s_sid; 2330 if ((p->p_lflag & PL_CONTROLT) && 2331 (tp = p->p_session->s_ttyp)) { 2332 ep->e_tdev = tp->t_dev; 2333 ep->e_tpgid = tp->t_pgrp ? tp->t_pgrp->pg_id : NO_PGID; 2334 COND_SET_VALUE(ep->e_tsess, tp->t_session, allowaddr); 2335 } else 2336 ep->e_tdev = (uint32_t)NODEV; 2337 ep->e_flag = p->p_session->s_ttyvp ? EPROC_CTTY : 0; 2338 if (SESS_LEADER(p)) 2339 ep->e_flag |= EPROC_SLEADER; 2340 strncpy(ep->e_login, p->p_session->s_login, MAXLOGNAME); 2341 } 2342 ep->e_xsize = ep->e_xrssize = 0; 2343 ep->e_xccount = ep->e_xswrss = 0; 2344 } 2345 2346 /* 2347 * Fill in a kinfo_proc2 structure for the specified process. 2348 */ 2349 void 2350 fill_kproc2(struct proc *p, struct kinfo_proc2 *ki, bool zombie, bool allowaddr) 2351 { 2352 struct tty *tp; 2353 struct lwp *l, *l2; 2354 struct timeval ut, st, rt; 2355 sigset_t ss1, ss2; 2356 struct rusage ru; 2357 struct vmspace *vm; 2358 2359 KASSERT(mutex_owned(proc_lock)); 2360 KASSERT(mutex_owned(p->p_lock)); 2361 2362 sigemptyset(&ss1); 2363 sigemptyset(&ss2); 2364 2365 COND_SET_VALUE(ki->p_paddr, PTRTOUINT64(p), allowaddr); 2366 COND_SET_VALUE(ki->p_fd, PTRTOUINT64(p->p_fd), allowaddr); 2367 COND_SET_VALUE(ki->p_cwdi, PTRTOUINT64(p->p_cwdi), allowaddr); 2368 COND_SET_VALUE(ki->p_stats, PTRTOUINT64(p->p_stats), allowaddr); 2369 COND_SET_VALUE(ki->p_limit, PTRTOUINT64(p->p_limit), allowaddr); 2370 COND_SET_VALUE(ki->p_vmspace, PTRTOUINT64(p->p_vmspace), allowaddr); 2371 COND_SET_VALUE(ki->p_sigacts, PTRTOUINT64(p->p_sigacts), allowaddr); 2372 COND_SET_VALUE(ki->p_sess, PTRTOUINT64(p->p_session), allowaddr); 2373 ki->p_tsess = 0; /* may be changed if controlling tty below */ 2374 COND_SET_VALUE(ki->p_ru, PTRTOUINT64(&p->p_stats->p_ru), allowaddr); 2375 ki->p_eflag = 0; 2376 ki->p_exitsig = p->p_exitsig; 2377 ki->p_flag = L_INMEM; /* Process never swapped out */ 2378 ki->p_flag |= sysctl_map_flags(sysctl_flagmap, p->p_flag); 2379 ki->p_flag |= sysctl_map_flags(sysctl_sflagmap, p->p_sflag); 2380 ki->p_flag |= sysctl_map_flags(sysctl_slflagmap, p->p_slflag); 2381 ki->p_flag |= sysctl_map_flags(sysctl_lflagmap, p->p_lflag); 2382 ki->p_flag |= sysctl_map_flags(sysctl_stflagmap, p->p_stflag); 2383 ki->p_pid = p->p_pid; 2384 ki->p_ppid = p->p_ppid; 2385 ki->p_uid = kauth_cred_geteuid(p->p_cred); 2386 ki->p_ruid = kauth_cred_getuid(p->p_cred); 2387 ki->p_gid = kauth_cred_getegid(p->p_cred); 2388 ki->p_rgid = kauth_cred_getgid(p->p_cred); 2389 ki->p_svuid = kauth_cred_getsvuid(p->p_cred); 2390 ki->p_svgid = kauth_cred_getsvgid(p->p_cred); 2391 ki->p_ngroups = kauth_cred_ngroups(p->p_cred); 2392 kauth_cred_getgroups(p->p_cred, ki->p_groups, 2393 uimin(ki->p_ngroups, sizeof(ki->p_groups) / sizeof(ki->p_groups[0])), 2394 UIO_SYSSPACE); 2395 2396 ki->p_uticks = p->p_uticks; 2397 ki->p_sticks = p->p_sticks; 2398 ki->p_iticks = p->p_iticks; 2399 ki->p_tpgid = NO_PGID; /* may be changed if controlling tty below */ 2400 COND_SET_VALUE(ki->p_tracep, PTRTOUINT64(p->p_tracep), allowaddr); 2401 ki->p_traceflag = p->p_traceflag; 2402 2403 memcpy(&ki->p_sigignore, &p->p_sigctx.ps_sigignore,sizeof(ki_sigset_t)); 2404 memcpy(&ki->p_sigcatch, &p->p_sigctx.ps_sigcatch, sizeof(ki_sigset_t)); 2405 2406 ki->p_cpticks = 0; 2407 ki->p_pctcpu = p->p_pctcpu; 2408 ki->p_estcpu = 0; 2409 ki->p_stat = p->p_stat; /* Will likely be overridden by LWP status */ 2410 ki->p_realstat = p->p_stat; 2411 ki->p_nice = p->p_nice; 2412 ki->p_xstat = P_WAITSTATUS(p); 2413 ki->p_acflag = p->p_acflag; 2414 2415 strncpy(ki->p_comm, p->p_comm, 2416 uimin(sizeof(ki->p_comm), sizeof(p->p_comm))); 2417 strncpy(ki->p_ename, p->p_emul->e_name, sizeof(ki->p_ename)); 2418 2419 ki->p_nlwps = p->p_nlwps; 2420 ki->p_realflag = ki->p_flag; 2421 2422 if (p->p_stat != SIDL && !P_ZOMBIE(p) && !zombie) { 2423 vm = p->p_vmspace; 2424 ki->p_vm_rssize = vm_resident_count(vm); 2425 ki->p_vm_tsize = vm->vm_tsize; 2426 ki->p_vm_dsize = vm->vm_dsize; 2427 ki->p_vm_ssize = vm->vm_ssize; 2428 ki->p_vm_vsize = atop(vm->vm_map.size); 2429 /* 2430 * Since the stack is initially mapped mostly with 2431 * PROT_NONE and grown as needed, adjust the "mapped size" 2432 * to skip the unused stack portion. 2433 */ 2434 ki->p_vm_msize = 2435 atop(vm->vm_map.size) - vm->vm_issize + vm->vm_ssize; 2436 2437 /* Pick the primary (first) LWP */ 2438 l = proc_active_lwp(p); 2439 KASSERT(l != NULL); 2440 lwp_lock(l); 2441 ki->p_nrlwps = p->p_nrlwps; 2442 ki->p_forw = 0; 2443 ki->p_back = 0; 2444 COND_SET_VALUE(ki->p_addr, PTRTOUINT64(l->l_addr), allowaddr); 2445 ki->p_stat = l->l_stat; 2446 ki->p_flag |= sysctl_map_flags(sysctl_lwpflagmap, l->l_flag); 2447 ki->p_swtime = l->l_swtime; 2448 ki->p_slptime = l->l_slptime; 2449 if (l->l_stat == LSONPROC) 2450 ki->p_schedflags = l->l_cpu->ci_schedstate.spc_flags; 2451 else 2452 ki->p_schedflags = 0; 2453 ki->p_priority = lwp_eprio(l); 2454 ki->p_usrpri = l->l_priority; 2455 if (l->l_wchan) 2456 strncpy(ki->p_wmesg, l->l_wmesg, sizeof(ki->p_wmesg)); 2457 COND_SET_VALUE(ki->p_wchan, PTRTOUINT64(l->l_wchan), allowaddr); 2458 ki->p_cpuid = cpu_index(l->l_cpu); 2459 lwp_unlock(l); 2460 LIST_FOREACH(l, &p->p_lwps, l_sibling) { 2461 /* This is hardly correct, but... */ 2462 sigplusset(&l->l_sigpend.sp_set, &ss1); 2463 sigplusset(&l->l_sigmask, &ss2); 2464 ki->p_cpticks += l->l_cpticks; 2465 ki->p_pctcpu += l->l_pctcpu; 2466 ki->p_estcpu += l->l_estcpu; 2467 } 2468 } 2469 sigplusset(&p->p_sigpend.sp_set, &ss2); 2470 memcpy(&ki->p_siglist, &ss1, sizeof(ki_sigset_t)); 2471 memcpy(&ki->p_sigmask, &ss2, sizeof(ki_sigset_t)); 2472 2473 if (p->p_session != NULL) { 2474 ki->p_sid = p->p_session->s_sid; 2475 ki->p__pgid = p->p_pgrp->pg_id; 2476 if (p->p_session->s_ttyvp) 2477 ki->p_eflag |= EPROC_CTTY; 2478 if (SESS_LEADER(p)) 2479 ki->p_eflag |= EPROC_SLEADER; 2480 strncpy(ki->p_login, p->p_session->s_login, 2481 uimin(sizeof ki->p_login - 1, sizeof p->p_session->s_login)); 2482 ki->p_jobc = p->p_pgrp->pg_jobc; 2483 if ((p->p_lflag & PL_CONTROLT) && (tp = p->p_session->s_ttyp)) { 2484 ki->p_tdev = tp->t_dev; 2485 ki->p_tpgid = tp->t_pgrp ? tp->t_pgrp->pg_id : NO_PGID; 2486 COND_SET_VALUE(ki->p_tsess, PTRTOUINT64(tp->t_session), 2487 allowaddr); 2488 } else { 2489 ki->p_tdev = (int32_t)NODEV; 2490 } 2491 } 2492 2493 if (!P_ZOMBIE(p) && !zombie) { 2494 ki->p_uvalid = 1; 2495 ki->p_ustart_sec = p->p_stats->p_start.tv_sec; 2496 ki->p_ustart_usec = p->p_stats->p_start.tv_usec; 2497 2498 calcru(p, &ut, &st, NULL, &rt); 2499 ki->p_rtime_sec = rt.tv_sec; 2500 ki->p_rtime_usec = rt.tv_usec; 2501 ki->p_uutime_sec = ut.tv_sec; 2502 ki->p_uutime_usec = ut.tv_usec; 2503 ki->p_ustime_sec = st.tv_sec; 2504 ki->p_ustime_usec = st.tv_usec; 2505 2506 memcpy(&ru, &p->p_stats->p_ru, sizeof(ru)); 2507 ki->p_uru_nvcsw = 0; 2508 ki->p_uru_nivcsw = 0; 2509 LIST_FOREACH(l2, &p->p_lwps, l_sibling) { 2510 ki->p_uru_nvcsw += (l2->l_ncsw - l2->l_nivcsw); 2511 ki->p_uru_nivcsw += l2->l_nivcsw; 2512 ruadd(&ru, &l2->l_ru); 2513 } 2514 ki->p_uru_maxrss = ru.ru_maxrss; 2515 ki->p_uru_ixrss = ru.ru_ixrss; 2516 ki->p_uru_idrss = ru.ru_idrss; 2517 ki->p_uru_isrss = ru.ru_isrss; 2518 ki->p_uru_minflt = ru.ru_minflt; 2519 ki->p_uru_majflt = ru.ru_majflt; 2520 ki->p_uru_nswap = ru.ru_nswap; 2521 ki->p_uru_inblock = ru.ru_inblock; 2522 ki->p_uru_oublock = ru.ru_oublock; 2523 ki->p_uru_msgsnd = ru.ru_msgsnd; 2524 ki->p_uru_msgrcv = ru.ru_msgrcv; 2525 ki->p_uru_nsignals = ru.ru_nsignals; 2526 2527 timeradd(&p->p_stats->p_cru.ru_utime, 2528 &p->p_stats->p_cru.ru_stime, &ut); 2529 ki->p_uctime_sec = ut.tv_sec; 2530 ki->p_uctime_usec = ut.tv_usec; 2531 } 2532 } 2533 2534 2535 int 2536 proc_find_locked(struct lwp *l, struct proc **p, pid_t pid) 2537 { 2538 int error; 2539 2540 mutex_enter(proc_lock); 2541 if (pid == -1) 2542 *p = l->l_proc; 2543 else 2544 *p = proc_find(pid); 2545 2546 if (*p == NULL) { 2547 if (pid != -1) 2548 mutex_exit(proc_lock); 2549 return ESRCH; 2550 } 2551 if (pid != -1) 2552 mutex_enter((*p)->p_lock); 2553 mutex_exit(proc_lock); 2554 2555 error = kauth_authorize_process(l->l_cred, 2556 KAUTH_PROCESS_CANSEE, *p, 2557 KAUTH_ARG(KAUTH_REQ_PROCESS_CANSEE_ENTRY), NULL, NULL); 2558 if (error) { 2559 if (pid != -1) 2560 mutex_exit((*p)->p_lock); 2561 } 2562 return error; 2563 } 2564 2565 static int 2566 fill_pathname(struct lwp *l, pid_t pid, void *oldp, size_t *oldlenp) 2567 { 2568 int error; 2569 struct proc *p; 2570 2571 if ((error = proc_find_locked(l, &p, pid)) != 0) 2572 return error; 2573 2574 if (p->p_path == NULL) { 2575 if (pid != -1) 2576 mutex_exit(p->p_lock); 2577 return ENOENT; 2578 } 2579 2580 size_t len = strlen(p->p_path) + 1; 2581 if (oldp != NULL) { 2582 size_t copylen = uimin(len, *oldlenp); 2583 error = sysctl_copyout(l, p->p_path, oldp, copylen); 2584 if (error == 0 && *oldlenp < len) 2585 error = ENOSPC; 2586 } 2587 *oldlenp = len; 2588 if (pid != -1) 2589 mutex_exit(p->p_lock); 2590 return error; 2591 } 2592 2593 int 2594 proc_getauxv(struct proc *p, void **buf, size_t *len) 2595 { 2596 struct ps_strings pss; 2597 int error; 2598 void *uauxv, *kauxv; 2599 size_t size; 2600 2601 if ((error = copyin_psstrings(p, &pss)) != 0) 2602 return error; 2603 if (pss.ps_envstr == NULL) 2604 return EIO; 2605 2606 size = p->p_execsw->es_arglen; 2607 if (size == 0) 2608 return EIO; 2609 2610 size_t ptrsz = PROC_PTRSZ(p); 2611 uauxv = (void *)((char *)pss.ps_envstr + (pss.ps_nenvstr + 1) * ptrsz); 2612 2613 kauxv = kmem_alloc(size, KM_SLEEP); 2614 2615 error = copyin_proc(p, uauxv, kauxv, size); 2616 if (error) { 2617 kmem_free(kauxv, size); 2618 return error; 2619 } 2620 2621 *buf = kauxv; 2622 *len = size; 2623 2624 return 0; 2625 } 2626 2627 2628 static int 2629 sysctl_security_expose_address(SYSCTLFN_ARGS) 2630 { 2631 int expose_address, error; 2632 struct sysctlnode node; 2633 2634 node = *rnode; 2635 node.sysctl_data = &expose_address; 2636 expose_address = *(int *)rnode->sysctl_data; 2637 error = sysctl_lookup(SYSCTLFN_CALL(&node)); 2638 if (error || newp == NULL) 2639 return error; 2640 2641 if (kauth_authorize_system(l->l_cred, KAUTH_SYSTEM_KERNADDR, 2642 0, NULL, NULL, NULL)) 2643 return EPERM; 2644 2645 switch (expose_address) { 2646 case 0: 2647 case 1: 2648 case 2: 2649 break; 2650 default: 2651 return EINVAL; 2652 } 2653 2654 *(int *)rnode->sysctl_data = expose_address; 2655 2656 return 0; 2657 } 2658 2659 bool 2660 get_expose_address(struct proc *p) 2661 { 2662 /* allow only if sysctl variable is set or privileged */ 2663 return kauth_authorize_process(kauth_cred_get(), KAUTH_PROCESS_CANSEE, 2664 p, KAUTH_ARG(KAUTH_REQ_PROCESS_CANSEE_KPTR), NULL, NULL) == 0; 2665 } 2666