1 /* $NetBSD: kern_proc.c,v 1.268 2022/07/01 01:06:40 riastradh Exp $ */ 2 3 /*- 4 * Copyright (c) 1999, 2006, 2007, 2008, 2020 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 1.268 2022/07/01 01:06:40 riastradh 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/ioctl.h> 94 #include <sys/tty.h> 95 #include <sys/signalvar.h> 96 #include <sys/ras.h> 97 #include <sys/filedesc.h> 98 #include <sys/syscall_stats.h> 99 #include <sys/kauth.h> 100 #include <sys/sleepq.h> 101 #include <sys/atomic.h> 102 #include <sys/kmem.h> 103 #include <sys/namei.h> 104 #include <sys/dtrace_bsd.h> 105 #include <sys/sysctl.h> 106 #include <sys/exec.h> 107 #include <sys/cpu.h> 108 #include <sys/compat_stub.h> 109 #include <sys/futex.h> 110 #include <sys/pserialize.h> 111 112 #include <uvm/uvm_extern.h> 113 114 /* 115 * Process lists. 116 */ 117 118 struct proclist allproc __cacheline_aligned; 119 struct proclist zombproc __cacheline_aligned; 120 121 kmutex_t proc_lock __cacheline_aligned; 122 static pserialize_t proc_psz; 123 124 /* 125 * pid to lwp/proc lookup is done by indexing the pid_table array. 126 * Since pid numbers are only allocated when an empty slot 127 * has been found, there is no need to search any lists ever. 128 * (an orphaned pgrp will lock the slot, a session will lock 129 * the pgrp with the same number.) 130 * If the table is too small it is reallocated with twice the 131 * previous size and the entries 'unzipped' into the two halves. 132 * A linked list of free entries is passed through the pt_lwp 133 * field of 'free' items - set odd to be an invalid ptr. Two 134 * additional bits are also used to indicate if the slot is 135 * currently occupied by a proc or lwp, and if the PID is 136 * hidden from certain kinds of lookups. We thus require a 137 * minimum alignment for proc and lwp structures (LWPs are 138 * at least 32-byte aligned). 139 */ 140 141 struct pid_table { 142 uintptr_t pt_slot; 143 struct pgrp *pt_pgrp; 144 pid_t pt_pid; 145 }; 146 147 #define PT_F_FREE ((uintptr_t)__BIT(0)) 148 #define PT_F_LWP 0 /* pseudo-flag */ 149 #define PT_F_PROC ((uintptr_t)__BIT(1)) 150 151 #define PT_F_TYPEBITS (PT_F_FREE|PT_F_PROC) 152 #define PT_F_ALLBITS (PT_F_FREE|PT_F_PROC) 153 154 #define PT_VALID(s) (((s) & PT_F_FREE) == 0) 155 #define PT_RESERVED(s) ((s) == 0) 156 #define PT_NEXT(s) ((u_int)(s) >> 1) 157 #define PT_SET_FREE(pid) (((pid) << 1) | PT_F_FREE) 158 #define PT_SET_LWP(l) ((uintptr_t)(l)) 159 #define PT_SET_PROC(p) (((uintptr_t)(p)) | PT_F_PROC) 160 #define PT_SET_RESERVED 0 161 #define PT_GET_LWP(s) ((struct lwp *)((s) & ~PT_F_ALLBITS)) 162 #define PT_GET_PROC(s) ((struct proc *)((s) & ~PT_F_ALLBITS)) 163 #define PT_GET_TYPE(s) ((s) & PT_F_TYPEBITS) 164 #define PT_IS_LWP(s) (PT_GET_TYPE(s) == PT_F_LWP && (s) != 0) 165 #define PT_IS_PROC(s) (PT_GET_TYPE(s) == PT_F_PROC) 166 167 #define MIN_PROC_ALIGNMENT (PT_F_ALLBITS + 1) 168 169 /* 170 * Table of process IDs (PIDs). 171 */ 172 static struct pid_table *pid_table __read_mostly; 173 174 #define INITIAL_PID_TABLE_SIZE (1 << 5) 175 176 /* Table mask, threshold for growing and number of allocated PIDs. */ 177 static u_int pid_tbl_mask __read_mostly; 178 static u_int pid_alloc_lim __read_mostly; 179 static u_int pid_alloc_cnt __cacheline_aligned; 180 181 /* Next free, last free and maximum PIDs. */ 182 static u_int next_free_pt __cacheline_aligned; 183 static u_int last_free_pt __cacheline_aligned; 184 static pid_t pid_max __read_mostly; 185 186 /* Components of the first process -- never freed. */ 187 188 extern struct emul emul_netbsd; /* defined in kern_exec.c */ 189 190 struct session session0 = { 191 .s_count = 1, 192 .s_sid = 0, 193 }; 194 struct pgrp pgrp0 = { 195 .pg_members = LIST_HEAD_INITIALIZER(&pgrp0.pg_members), 196 .pg_session = &session0, 197 }; 198 filedesc_t filedesc0; 199 struct cwdinfo cwdi0 = { 200 .cwdi_cmask = CMASK, 201 .cwdi_refcnt = 1, 202 }; 203 struct plimit limit0; 204 struct pstats pstat0; 205 struct vmspace vmspace0; 206 struct sigacts sigacts0; 207 struct proc proc0 = { 208 .p_lwps = LIST_HEAD_INITIALIZER(&proc0.p_lwps), 209 .p_sigwaiters = LIST_HEAD_INITIALIZER(&proc0.p_sigwaiters), 210 .p_nlwps = 1, 211 .p_nrlwps = 1, 212 .p_pgrp = &pgrp0, 213 .p_comm = "system", 214 /* 215 * Set P_NOCLDWAIT so that kernel threads are reparented to init(8) 216 * when they exit. init(8) can easily wait them out for us. 217 */ 218 .p_flag = PK_SYSTEM | PK_NOCLDWAIT, 219 .p_stat = SACTIVE, 220 .p_nice = NZERO, 221 .p_emul = &emul_netbsd, 222 .p_cwdi = &cwdi0, 223 .p_limit = &limit0, 224 .p_fd = &filedesc0, 225 .p_vmspace = &vmspace0, 226 .p_stats = &pstat0, 227 .p_sigacts = &sigacts0, 228 #ifdef PROC0_MD_INITIALIZERS 229 PROC0_MD_INITIALIZERS 230 #endif 231 }; 232 kauth_cred_t cred0; 233 234 static const int nofile = NOFILE; 235 static const int maxuprc = MAXUPRC; 236 237 static int sysctl_doeproc(SYSCTLFN_PROTO); 238 static int sysctl_kern_proc_args(SYSCTLFN_PROTO); 239 static int sysctl_security_expose_address(SYSCTLFN_PROTO); 240 241 #ifdef KASLR 242 static int kern_expose_address = 0; 243 #else 244 static int kern_expose_address = 1; 245 #endif 246 /* 247 * The process list descriptors, used during pid allocation and 248 * by sysctl. No locking on this data structure is needed since 249 * it is completely static. 250 */ 251 const struct proclist_desc proclists[] = { 252 { &allproc }, 253 { &zombproc }, 254 { NULL }, 255 }; 256 257 static struct pgrp * pg_remove(pid_t); 258 static void pg_delete(pid_t); 259 static void orphanpg(struct pgrp *); 260 261 static specificdata_domain_t proc_specificdata_domain; 262 263 static pool_cache_t proc_cache; 264 265 static kauth_listener_t proc_listener; 266 267 static void fill_proc(const struct proc *, struct proc *, bool); 268 static int fill_pathname(struct lwp *, pid_t, void *, size_t *); 269 static int fill_cwd(struct lwp *, pid_t, void *, size_t *); 270 271 static int 272 proc_listener_cb(kauth_cred_t cred, kauth_action_t action, void *cookie, 273 void *arg0, void *arg1, void *arg2, void *arg3) 274 { 275 struct proc *p; 276 int result; 277 278 result = KAUTH_RESULT_DEFER; 279 p = arg0; 280 281 switch (action) { 282 case KAUTH_PROCESS_CANSEE: { 283 enum kauth_process_req req; 284 285 req = (enum kauth_process_req)(uintptr_t)arg1; 286 287 switch (req) { 288 case KAUTH_REQ_PROCESS_CANSEE_ARGS: 289 case KAUTH_REQ_PROCESS_CANSEE_ENTRY: 290 case KAUTH_REQ_PROCESS_CANSEE_OPENFILES: 291 case KAUTH_REQ_PROCESS_CANSEE_EPROC: 292 result = KAUTH_RESULT_ALLOW; 293 break; 294 295 case KAUTH_REQ_PROCESS_CANSEE_ENV: 296 if (kauth_cred_getuid(cred) != 297 kauth_cred_getuid(p->p_cred) || 298 kauth_cred_getuid(cred) != 299 kauth_cred_getsvuid(p->p_cred)) 300 break; 301 302 result = KAUTH_RESULT_ALLOW; 303 304 break; 305 306 case KAUTH_REQ_PROCESS_CANSEE_KPTR: 307 if (!kern_expose_address) 308 break; 309 310 if (kern_expose_address == 1 && !(p->p_flag & PK_KMEM)) 311 break; 312 313 result = KAUTH_RESULT_ALLOW; 314 315 break; 316 317 default: 318 break; 319 } 320 321 break; 322 } 323 324 case KAUTH_PROCESS_FORK: { 325 int lnprocs = (int)(unsigned long)arg2; 326 327 /* 328 * Don't allow a nonprivileged user to use the last few 329 * processes. The variable lnprocs is the current number of 330 * processes, maxproc is the limit. 331 */ 332 if (__predict_false((lnprocs >= maxproc - 5))) 333 break; 334 335 result = KAUTH_RESULT_ALLOW; 336 337 break; 338 } 339 340 case KAUTH_PROCESS_CORENAME: 341 case KAUTH_PROCESS_STOPFLAG: 342 if (proc_uidmatch(cred, p->p_cred) == 0) 343 result = KAUTH_RESULT_ALLOW; 344 345 break; 346 347 default: 348 break; 349 } 350 351 return result; 352 } 353 354 static int 355 proc_ctor(void *arg __unused, void *obj, int flags __unused) 356 { 357 struct proc *p = obj; 358 359 memset(p, 0, sizeof(*p)); 360 klist_init(&p->p_klist); 361 362 /* 363 * There is no need for a proc_dtor() to do a klist_fini(), 364 * since knote_proc_exit() ensures that p->p_klist is empty 365 * when a process exits. 366 */ 367 368 return 0; 369 } 370 371 static pid_t proc_alloc_pid_slot(struct proc *, uintptr_t); 372 373 /* 374 * Initialize global process hashing structures. 375 */ 376 void 377 procinit(void) 378 { 379 const struct proclist_desc *pd; 380 u_int i; 381 #define LINK_EMPTY ((PID_MAX + INITIAL_PID_TABLE_SIZE) & ~(INITIAL_PID_TABLE_SIZE - 1)) 382 383 for (pd = proclists; pd->pd_list != NULL; pd++) 384 LIST_INIT(pd->pd_list); 385 386 mutex_init(&proc_lock, MUTEX_DEFAULT, IPL_NONE); 387 388 proc_psz = pserialize_create(); 389 390 pid_table = kmem_alloc(INITIAL_PID_TABLE_SIZE 391 * sizeof(struct pid_table), KM_SLEEP); 392 pid_tbl_mask = INITIAL_PID_TABLE_SIZE - 1; 393 pid_max = PID_MAX; 394 395 /* Set free list running through table... 396 Preset 'use count' above PID_MAX so we allocate pid 1 next. */ 397 for (i = 0; i <= pid_tbl_mask; i++) { 398 pid_table[i].pt_slot = PT_SET_FREE(LINK_EMPTY + i + 1); 399 pid_table[i].pt_pgrp = 0; 400 pid_table[i].pt_pid = 0; 401 } 402 /* slot 0 is just grabbed */ 403 next_free_pt = 1; 404 /* Need to fix last entry. */ 405 last_free_pt = pid_tbl_mask; 406 pid_table[last_free_pt].pt_slot = PT_SET_FREE(LINK_EMPTY); 407 /* point at which we grow table - to avoid reusing pids too often */ 408 pid_alloc_lim = pid_tbl_mask - 1; 409 #undef LINK_EMPTY 410 411 /* Reserve PID 1 for init(8). */ /* XXX slightly gross */ 412 mutex_enter(&proc_lock); 413 if (proc_alloc_pid_slot(&proc0, PT_SET_RESERVED) != 1) 414 panic("failed to reserve PID 1 for init(8)"); 415 mutex_exit(&proc_lock); 416 417 proc_specificdata_domain = specificdata_domain_create(); 418 KASSERT(proc_specificdata_domain != NULL); 419 420 size_t proc_alignment = coherency_unit; 421 if (proc_alignment < MIN_PROC_ALIGNMENT) 422 proc_alignment = MIN_PROC_ALIGNMENT; 423 424 proc_cache = pool_cache_init(sizeof(struct proc), proc_alignment, 0, 0, 425 "procpl", NULL, IPL_NONE, proc_ctor, NULL, NULL); 426 427 proc_listener = kauth_listen_scope(KAUTH_SCOPE_PROCESS, 428 proc_listener_cb, NULL); 429 } 430 431 void 432 procinit_sysctl(void) 433 { 434 static struct sysctllog *clog; 435 436 sysctl_createv(&clog, 0, NULL, NULL, 437 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 438 CTLTYPE_INT, "expose_address", 439 SYSCTL_DESCR("Enable exposing kernel addresses"), 440 sysctl_security_expose_address, 0, 441 &kern_expose_address, 0, CTL_KERN, CTL_CREATE, CTL_EOL); 442 sysctl_createv(&clog, 0, NULL, NULL, 443 CTLFLAG_PERMANENT, 444 CTLTYPE_NODE, "proc", 445 SYSCTL_DESCR("System-wide process information"), 446 sysctl_doeproc, 0, NULL, 0, 447 CTL_KERN, KERN_PROC, CTL_EOL); 448 sysctl_createv(&clog, 0, NULL, NULL, 449 CTLFLAG_PERMANENT, 450 CTLTYPE_NODE, "proc2", 451 SYSCTL_DESCR("Machine-independent process information"), 452 sysctl_doeproc, 0, NULL, 0, 453 CTL_KERN, KERN_PROC2, CTL_EOL); 454 sysctl_createv(&clog, 0, NULL, NULL, 455 CTLFLAG_PERMANENT, 456 CTLTYPE_NODE, "proc_args", 457 SYSCTL_DESCR("Process argument information"), 458 sysctl_kern_proc_args, 0, NULL, 0, 459 CTL_KERN, KERN_PROC_ARGS, CTL_EOL); 460 461 /* 462 "nodes" under these: 463 464 KERN_PROC_ALL 465 KERN_PROC_PID pid 466 KERN_PROC_PGRP pgrp 467 KERN_PROC_SESSION sess 468 KERN_PROC_TTY tty 469 KERN_PROC_UID uid 470 KERN_PROC_RUID uid 471 KERN_PROC_GID gid 472 KERN_PROC_RGID gid 473 474 all in all, probably not worth the effort... 475 */ 476 } 477 478 /* 479 * Initialize process 0. 480 */ 481 void 482 proc0_init(void) 483 { 484 struct proc *p; 485 struct pgrp *pg; 486 struct rlimit *rlim; 487 rlim_t lim; 488 int i; 489 490 p = &proc0; 491 pg = &pgrp0; 492 493 mutex_init(&p->p_stmutex, MUTEX_DEFAULT, IPL_HIGH); 494 mutex_init(&p->p_auxlock, MUTEX_DEFAULT, IPL_NONE); 495 p->p_lock = mutex_obj_alloc(MUTEX_DEFAULT, IPL_NONE); 496 497 rw_init(&p->p_reflock); 498 cv_init(&p->p_waitcv, "wait"); 499 cv_init(&p->p_lwpcv, "lwpwait"); 500 501 LIST_INSERT_HEAD(&p->p_lwps, &lwp0, l_sibling); 502 503 KASSERT(lwp0.l_lid == 0); 504 pid_table[lwp0.l_lid].pt_slot = PT_SET_LWP(&lwp0); 505 LIST_INSERT_HEAD(&allproc, p, p_list); 506 507 pid_table[lwp0.l_lid].pt_pgrp = pg; 508 LIST_INSERT_HEAD(&pg->pg_members, p, p_pglist); 509 510 #ifdef __HAVE_SYSCALL_INTERN 511 (*p->p_emul->e_syscall_intern)(p); 512 #endif 513 514 /* Create credentials. */ 515 cred0 = kauth_cred_alloc(); 516 p->p_cred = cred0; 517 518 /* Create the CWD info. */ 519 rw_init(&cwdi0.cwdi_lock); 520 521 /* Create the limits structures. */ 522 mutex_init(&limit0.pl_lock, MUTEX_DEFAULT, IPL_NONE); 523 524 rlim = limit0.pl_rlimit; 525 for (i = 0; i < __arraycount(limit0.pl_rlimit); i++) { 526 rlim[i].rlim_cur = RLIM_INFINITY; 527 rlim[i].rlim_max = RLIM_INFINITY; 528 } 529 530 rlim[RLIMIT_NOFILE].rlim_max = maxfiles; 531 rlim[RLIMIT_NOFILE].rlim_cur = maxfiles < nofile ? maxfiles : nofile; 532 533 rlim[RLIMIT_NPROC].rlim_max = maxproc; 534 rlim[RLIMIT_NPROC].rlim_cur = maxproc < maxuprc ? maxproc : maxuprc; 535 536 lim = MIN(VM_MAXUSER_ADDRESS, ctob((rlim_t)uvm_availmem(false))); 537 rlim[RLIMIT_RSS].rlim_max = lim; 538 rlim[RLIMIT_MEMLOCK].rlim_max = lim; 539 rlim[RLIMIT_MEMLOCK].rlim_cur = lim / 3; 540 541 rlim[RLIMIT_NTHR].rlim_max = maxlwp; 542 rlim[RLIMIT_NTHR].rlim_cur = maxlwp / 2; 543 544 /* Note that default core name has zero length. */ 545 limit0.pl_corename = defcorename; 546 limit0.pl_cnlen = 0; 547 limit0.pl_refcnt = 1; 548 limit0.pl_writeable = false; 549 limit0.pl_sv_limit = NULL; 550 551 /* Configure virtual memory system, set vm rlimits. */ 552 uvm_init_limits(p); 553 554 /* Initialize file descriptor table for proc0. */ 555 fd_init(&filedesc0); 556 557 /* 558 * Initialize proc0's vmspace, which uses the kernel pmap. 559 * All kernel processes (which never have user space mappings) 560 * share proc0's vmspace, and thus, the kernel pmap. 561 */ 562 uvmspace_init(&vmspace0, pmap_kernel(), round_page(VM_MIN_ADDRESS), 563 trunc_page(VM_MAXUSER_ADDRESS), 564 #ifdef __USE_TOPDOWN_VM 565 true 566 #else 567 false 568 #endif 569 ); 570 571 /* Initialize signal state for proc0. XXX IPL_SCHED */ 572 mutex_init(&p->p_sigacts->sa_mutex, MUTEX_DEFAULT, IPL_SCHED); 573 siginit(p); 574 575 proc_initspecific(p); 576 kdtrace_proc_ctor(NULL, p); 577 } 578 579 /* 580 * Session reference counting. 581 */ 582 583 void 584 proc_sesshold(struct session *ss) 585 { 586 587 KASSERT(mutex_owned(&proc_lock)); 588 ss->s_count++; 589 } 590 591 void 592 proc_sessrele(struct session *ss) 593 { 594 struct pgrp *pg; 595 596 KASSERT(mutex_owned(&proc_lock)); 597 KASSERT(ss->s_count > 0); 598 599 /* 600 * We keep the pgrp with the same id as the session in order to 601 * stop a process being given the same pid. Since the pgrp holds 602 * a reference to the session, it must be a 'zombie' pgrp by now. 603 */ 604 if (--ss->s_count == 0) { 605 pg = pg_remove(ss->s_sid); 606 } else { 607 pg = NULL; 608 ss = NULL; 609 } 610 611 mutex_exit(&proc_lock); 612 613 if (pg) 614 kmem_free(pg, sizeof(struct pgrp)); 615 if (ss) 616 kmem_free(ss, sizeof(struct session)); 617 } 618 619 /* 620 * Check that the specified process group is in the session of the 621 * specified process. 622 * Treats -ve ids as process ids. 623 * Used to validate TIOCSPGRP requests. 624 */ 625 int 626 pgid_in_session(struct proc *p, pid_t pg_id) 627 { 628 struct pgrp *pgrp; 629 struct session *session; 630 int error; 631 632 if (pg_id == INT_MIN) 633 return EINVAL; 634 635 mutex_enter(&proc_lock); 636 if (pg_id < 0) { 637 struct proc *p1 = proc_find(-pg_id); 638 if (p1 == NULL) { 639 error = EINVAL; 640 goto fail; 641 } 642 pgrp = p1->p_pgrp; 643 } else { 644 pgrp = pgrp_find(pg_id); 645 if (pgrp == NULL) { 646 error = EINVAL; 647 goto fail; 648 } 649 } 650 session = pgrp->pg_session; 651 error = (session != p->p_pgrp->pg_session) ? EPERM : 0; 652 fail: 653 mutex_exit(&proc_lock); 654 return error; 655 } 656 657 /* 658 * p_inferior: is p an inferior of q? 659 */ 660 static inline bool 661 p_inferior(struct proc *p, struct proc *q) 662 { 663 664 KASSERT(mutex_owned(&proc_lock)); 665 666 for (; p != q; p = p->p_pptr) 667 if (p->p_pid == 0) 668 return false; 669 return true; 670 } 671 672 /* 673 * proc_find_lwp: locate an lwp in said proc by the ID. 674 * 675 * => Must be called with p::p_lock held. 676 * => LSIDL lwps are not returned because they are only partially 677 * constructed while occupying the slot. 678 * => Callers need to be careful about lwp::l_stat of the returned 679 * lwp. 680 */ 681 struct lwp * 682 proc_find_lwp(proc_t *p, pid_t pid) 683 { 684 struct pid_table *pt; 685 unsigned pt_mask; 686 struct lwp *l = NULL; 687 uintptr_t slot; 688 int s; 689 690 KASSERT(mutex_owned(p->p_lock)); 691 692 /* 693 * Look in the pid_table. This is done unlocked inside a 694 * pserialize read section covering pid_table's memory 695 * allocation only, so take care to read things in the correct 696 * order: 697 * 698 * 1. First read the table mask -- this only ever increases, in 699 * expand_pid_table, so a stale value is safely 700 * conservative. 701 * 702 * 2. Next read the pid table -- this is always set _before_ 703 * the mask increases, so if we see a new table and stale 704 * mask, the mask is still valid for the table. 705 */ 706 s = pserialize_read_enter(); 707 pt_mask = atomic_load_acquire(&pid_tbl_mask); 708 pt = &atomic_load_consume(&pid_table)[pid & pt_mask]; 709 slot = atomic_load_consume(&pt->pt_slot); 710 if (__predict_false(!PT_IS_LWP(slot))) { 711 pserialize_read_exit(s); 712 return NULL; 713 } 714 715 /* 716 * Check to see if the LWP is from the correct process. We won't 717 * see entries in pid_table from a prior process that also used "p", 718 * by virtue of the fact that allocating "p" means all prior updates 719 * to dependant data structures are visible to this thread. 720 */ 721 l = PT_GET_LWP(slot); 722 if (__predict_false(atomic_load_relaxed(&l->l_proc) != p)) { 723 pserialize_read_exit(s); 724 return NULL; 725 } 726 727 /* 728 * We now know that p->p_lock holds this LWP stable. 729 * 730 * If the status is not LSIDL, it means the LWP is intended to be 731 * findable by LID and l_lid cannot change behind us. 732 * 733 * No need to acquire the LWP's lock to check for LSIDL, as 734 * p->p_lock must be held to transition in and out of LSIDL. 735 * Any other observed state of is no particular interest. 736 */ 737 pserialize_read_exit(s); 738 return l->l_stat != LSIDL && l->l_lid == pid ? l : NULL; 739 } 740 741 /* 742 * proc_find_lwp_unlocked: locate an lwp in said proc by the ID. 743 * 744 * => Called in a pserialize read section with no locks held. 745 * => LSIDL lwps are not returned because they are only partially 746 * constructed while occupying the slot. 747 * => Callers need to be careful about lwp::l_stat of the returned 748 * lwp. 749 * => If an LWP is found, it's returned locked. 750 */ 751 struct lwp * 752 proc_find_lwp_unlocked(proc_t *p, pid_t pid) 753 { 754 struct pid_table *pt; 755 unsigned pt_mask; 756 struct lwp *l = NULL; 757 uintptr_t slot; 758 759 KASSERT(pserialize_in_read_section()); 760 761 /* 762 * Look in the pid_table. This is done unlocked inside a 763 * pserialize read section covering pid_table's memory 764 * allocation only, so take care to read things in the correct 765 * order: 766 * 767 * 1. First read the table mask -- this only ever increases, in 768 * expand_pid_table, so a stale value is safely 769 * conservative. 770 * 771 * 2. Next read the pid table -- this is always set _before_ 772 * the mask increases, so if we see a new table and stale 773 * mask, the mask is still valid for the table. 774 */ 775 pt_mask = atomic_load_acquire(&pid_tbl_mask); 776 pt = &atomic_load_consume(&pid_table)[pid & pt_mask]; 777 slot = atomic_load_consume(&pt->pt_slot); 778 if (__predict_false(!PT_IS_LWP(slot))) { 779 return NULL; 780 } 781 782 /* 783 * Lock the LWP we found to get it stable. If it's embryonic or 784 * reaped (LSIDL) then none of the other fields can safely be 785 * checked. 786 */ 787 l = PT_GET_LWP(slot); 788 lwp_lock(l); 789 if (__predict_false(l->l_stat == LSIDL)) { 790 lwp_unlock(l); 791 return NULL; 792 } 793 794 /* 795 * l_proc and l_lid are now known stable because the LWP is not 796 * LSIDL, so check those fields too to make sure we found the 797 * right thing. 798 */ 799 if (__predict_false(l->l_proc != p || l->l_lid != pid)) { 800 lwp_unlock(l); 801 return NULL; 802 } 803 804 /* Everything checks out, return it locked. */ 805 return l; 806 } 807 808 /* 809 * proc_find_lwp_acquire_proc: locate an lwp and acquire a lock 810 * on its containing proc. 811 * 812 * => Similar to proc_find_lwp(), but does not require you to have 813 * the proc a priori. 814 * => Also returns proc * to caller, with p::p_lock held. 815 * => Same caveats apply. 816 */ 817 struct lwp * 818 proc_find_lwp_acquire_proc(pid_t pid, struct proc **pp) 819 { 820 struct pid_table *pt; 821 struct proc *p = NULL; 822 struct lwp *l = NULL; 823 uintptr_t slot; 824 825 KASSERT(pp != NULL); 826 mutex_enter(&proc_lock); 827 pt = &pid_table[pid & pid_tbl_mask]; 828 829 slot = pt->pt_slot; 830 if (__predict_true(PT_IS_LWP(slot) && pt->pt_pid == pid)) { 831 l = PT_GET_LWP(slot); 832 p = l->l_proc; 833 mutex_enter(p->p_lock); 834 if (__predict_false(l->l_stat == LSIDL)) { 835 mutex_exit(p->p_lock); 836 l = NULL; 837 p = NULL; 838 } 839 } 840 mutex_exit(&proc_lock); 841 842 KASSERT(p == NULL || mutex_owned(p->p_lock)); 843 *pp = p; 844 return l; 845 } 846 847 /* 848 * proc_find_raw_pid_table_locked: locate a process by the ID. 849 * 850 * => Must be called with proc_lock held. 851 */ 852 static proc_t * 853 proc_find_raw_pid_table_locked(pid_t pid, bool any_lwpid) 854 { 855 struct pid_table *pt; 856 proc_t *p = NULL; 857 uintptr_t slot; 858 859 /* No - used by DDB. KASSERT(mutex_owned(&proc_lock)); */ 860 pt = &pid_table[pid & pid_tbl_mask]; 861 862 slot = pt->pt_slot; 863 if (__predict_true(PT_IS_LWP(slot) && pt->pt_pid == pid)) { 864 /* 865 * When looking up processes, require a direct match 866 * on the PID assigned to the proc, not just one of 867 * its LWPs. 868 * 869 * N.B. We require lwp::l_proc of LSIDL LWPs to be 870 * valid here. 871 */ 872 p = PT_GET_LWP(slot)->l_proc; 873 if (__predict_false(p->p_pid != pid && !any_lwpid)) 874 p = NULL; 875 } else if (PT_IS_PROC(slot) && pt->pt_pid == pid) { 876 p = PT_GET_PROC(slot); 877 } 878 return p; 879 } 880 881 proc_t * 882 proc_find_raw(pid_t pid) 883 { 884 885 return proc_find_raw_pid_table_locked(pid, false); 886 } 887 888 static proc_t * 889 proc_find_internal(pid_t pid, bool any_lwpid) 890 { 891 proc_t *p; 892 893 KASSERT(mutex_owned(&proc_lock)); 894 895 p = proc_find_raw_pid_table_locked(pid, any_lwpid); 896 if (__predict_false(p == NULL)) { 897 return NULL; 898 } 899 900 /* 901 * Only allow live processes to be found by PID. 902 * XXX: p_stat might change, since proc unlocked. 903 */ 904 if (__predict_true(p->p_stat == SACTIVE || p->p_stat == SSTOP)) { 905 return p; 906 } 907 return NULL; 908 } 909 910 proc_t * 911 proc_find(pid_t pid) 912 { 913 return proc_find_internal(pid, false); 914 } 915 916 proc_t * 917 proc_find_lwpid(pid_t pid) 918 { 919 return proc_find_internal(pid, true); 920 } 921 922 /* 923 * pgrp_find: locate a process group by the ID. 924 * 925 * => Must be called with proc_lock held. 926 */ 927 struct pgrp * 928 pgrp_find(pid_t pgid) 929 { 930 struct pgrp *pg; 931 932 KASSERT(mutex_owned(&proc_lock)); 933 934 pg = pid_table[pgid & pid_tbl_mask].pt_pgrp; 935 936 /* 937 * Cannot look up a process group that only exists because the 938 * session has not died yet (traditional). 939 */ 940 if (pg == NULL || pg->pg_id != pgid || LIST_EMPTY(&pg->pg_members)) { 941 return NULL; 942 } 943 return pg; 944 } 945 946 static void 947 expand_pid_table(void) 948 { 949 size_t pt_size, tsz; 950 struct pid_table *n_pt, *new_pt; 951 uintptr_t slot; 952 struct pgrp *pgrp; 953 pid_t pid, rpid; 954 u_int i; 955 uint new_pt_mask; 956 957 KASSERT(mutex_owned(&proc_lock)); 958 959 /* Unlock the pid_table briefly to allocate memory. */ 960 pt_size = pid_tbl_mask + 1; 961 mutex_exit(&proc_lock); 962 963 tsz = pt_size * 2 * sizeof(struct pid_table); 964 new_pt = kmem_alloc(tsz, KM_SLEEP); 965 new_pt_mask = pt_size * 2 - 1; 966 967 /* XXX For now. The pratical limit is much lower anyway. */ 968 KASSERT(new_pt_mask <= FUTEX_TID_MASK); 969 970 mutex_enter(&proc_lock); 971 if (pt_size != pid_tbl_mask + 1) { 972 /* Another process beat us to it... */ 973 mutex_exit(&proc_lock); 974 kmem_free(new_pt, tsz); 975 goto out; 976 } 977 978 /* 979 * Copy entries from old table into new one. 980 * If 'pid' is 'odd' we need to place in the upper half, 981 * even pid's to the lower half. 982 * Free items stay in the low half so we don't have to 983 * fixup the reference to them. 984 * We stuff free items on the front of the freelist 985 * because we can't write to unmodified entries. 986 * Processing the table backwards maintains a semblance 987 * of issuing pid numbers that increase with time. 988 */ 989 i = pt_size - 1; 990 n_pt = new_pt + i; 991 for (; ; i--, n_pt--) { 992 slot = pid_table[i].pt_slot; 993 pgrp = pid_table[i].pt_pgrp; 994 if (!PT_VALID(slot)) { 995 /* Up 'use count' so that link is valid */ 996 pid = (PT_NEXT(slot) + pt_size) & ~pt_size; 997 rpid = 0; 998 slot = PT_SET_FREE(pid); 999 if (pgrp) 1000 pid = pgrp->pg_id; 1001 } else { 1002 pid = pid_table[i].pt_pid; 1003 rpid = pid; 1004 } 1005 1006 /* Save entry in appropriate half of table */ 1007 n_pt[pid & pt_size].pt_slot = slot; 1008 n_pt[pid & pt_size].pt_pgrp = pgrp; 1009 n_pt[pid & pt_size].pt_pid = rpid; 1010 1011 /* Put other piece on start of free list */ 1012 pid = (pid ^ pt_size) & ~pid_tbl_mask; 1013 n_pt[pid & pt_size].pt_slot = 1014 PT_SET_FREE((pid & ~pt_size) | next_free_pt); 1015 n_pt[pid & pt_size].pt_pgrp = 0; 1016 n_pt[pid & pt_size].pt_pid = 0; 1017 1018 next_free_pt = i | (pid & pt_size); 1019 if (i == 0) 1020 break; 1021 } 1022 1023 /* Save old table size and switch tables */ 1024 tsz = pt_size * sizeof(struct pid_table); 1025 n_pt = pid_table; 1026 atomic_store_release(&pid_table, new_pt); 1027 KASSERT(new_pt_mask >= pid_tbl_mask); 1028 atomic_store_release(&pid_tbl_mask, new_pt_mask); 1029 1030 /* 1031 * pid_max starts as PID_MAX (= 30000), once we have 16384 1032 * allocated pids we need it to be larger! 1033 */ 1034 if (pid_tbl_mask > PID_MAX) { 1035 pid_max = pid_tbl_mask * 2 + 1; 1036 pid_alloc_lim |= pid_alloc_lim << 1; 1037 } else 1038 pid_alloc_lim <<= 1; /* doubles number of free slots... */ 1039 1040 mutex_exit(&proc_lock); 1041 1042 /* 1043 * Make sure that unlocked access to the old pid_table is complete 1044 * and then free it. 1045 */ 1046 pserialize_perform(proc_psz); 1047 kmem_free(n_pt, tsz); 1048 1049 out: /* Return with proc_lock held again. */ 1050 mutex_enter(&proc_lock); 1051 } 1052 1053 struct proc * 1054 proc_alloc(void) 1055 { 1056 struct proc *p; 1057 1058 p = pool_cache_get(proc_cache, PR_WAITOK); 1059 p->p_stat = SIDL; /* protect against others */ 1060 proc_initspecific(p); 1061 kdtrace_proc_ctor(NULL, p); 1062 1063 /* 1064 * Allocate a placeholder in the pid_table. When we create the 1065 * first LWP for this process, it will take ownership of the 1066 * slot. 1067 */ 1068 if (__predict_false(proc_alloc_pid(p) == -1)) { 1069 /* Allocating the PID failed; unwind. */ 1070 proc_finispecific(p); 1071 proc_free_mem(p); 1072 p = NULL; 1073 } 1074 return p; 1075 } 1076 1077 /* 1078 * proc_alloc_pid_slot: allocate PID and record the occcupant so that 1079 * proc_find_raw() can find it by the PID. 1080 */ 1081 static pid_t __noinline 1082 proc_alloc_pid_slot(struct proc *p, uintptr_t slot) 1083 { 1084 struct pid_table *pt; 1085 pid_t pid; 1086 int nxt; 1087 1088 KASSERT(mutex_owned(&proc_lock)); 1089 1090 for (;;expand_pid_table()) { 1091 if (__predict_false(pid_alloc_cnt >= pid_alloc_lim)) { 1092 /* ensure pids cycle through 2000+ values */ 1093 continue; 1094 } 1095 /* 1096 * The first user process *must* be given PID 1. 1097 * it has already been reserved for us. This 1098 * will be coming in from the proc_alloc() call 1099 * above, and the entry will be usurped later when 1100 * the first user LWP is created. 1101 * XXX this is slightly gross. 1102 */ 1103 if (__predict_false(PT_RESERVED(pid_table[1].pt_slot) && 1104 p != &proc0)) { 1105 KASSERT(PT_IS_PROC(slot)); 1106 pt = &pid_table[1]; 1107 pt->pt_slot = slot; 1108 return 1; 1109 } 1110 pt = &pid_table[next_free_pt]; 1111 #ifdef DIAGNOSTIC 1112 if (__predict_false(PT_VALID(pt->pt_slot) || pt->pt_pgrp)) 1113 panic("proc_alloc: slot busy"); 1114 #endif 1115 nxt = PT_NEXT(pt->pt_slot); 1116 if (nxt & pid_tbl_mask) 1117 break; 1118 /* Table full - expand (NB last entry not used....) */ 1119 } 1120 1121 /* pid is 'saved use count' + 'size' + entry */ 1122 pid = (nxt & ~pid_tbl_mask) + pid_tbl_mask + 1 + next_free_pt; 1123 if ((uint)pid > (uint)pid_max) 1124 pid &= pid_tbl_mask; 1125 next_free_pt = nxt & pid_tbl_mask; 1126 1127 /* XXX For now. The pratical limit is much lower anyway. */ 1128 KASSERT(pid <= FUTEX_TID_MASK); 1129 1130 /* Grab table slot */ 1131 pt->pt_slot = slot; 1132 1133 KASSERT(pt->pt_pid == 0); 1134 pt->pt_pid = pid; 1135 pid_alloc_cnt++; 1136 1137 return pid; 1138 } 1139 1140 pid_t 1141 proc_alloc_pid(struct proc *p) 1142 { 1143 pid_t pid; 1144 1145 KASSERT((((uintptr_t)p) & PT_F_ALLBITS) == 0); 1146 KASSERT(p->p_stat == SIDL); 1147 1148 mutex_enter(&proc_lock); 1149 pid = proc_alloc_pid_slot(p, PT_SET_PROC(p)); 1150 if (pid != -1) 1151 p->p_pid = pid; 1152 mutex_exit(&proc_lock); 1153 1154 return pid; 1155 } 1156 1157 pid_t 1158 proc_alloc_lwpid(struct proc *p, struct lwp *l) 1159 { 1160 struct pid_table *pt; 1161 pid_t pid; 1162 1163 KASSERT((((uintptr_t)l) & PT_F_ALLBITS) == 0); 1164 KASSERT(l->l_proc == p); 1165 KASSERT(l->l_stat == LSIDL); 1166 1167 /* 1168 * For unlocked lookup in proc_find_lwp(), make sure l->l_proc 1169 * is globally visible before the LWP becomes visible via the 1170 * pid_table. 1171 */ 1172 #ifndef __HAVE_ATOMIC_AS_MEMBAR 1173 membar_producer(); 1174 #endif 1175 1176 /* 1177 * If the slot for p->p_pid currently points to the proc, 1178 * then we should usurp this ID for the LWP. This happens 1179 * at least once per process (for the first LWP), and can 1180 * happen again if the first LWP for a process exits and 1181 * before the process creates another. 1182 */ 1183 mutex_enter(&proc_lock); 1184 pid = p->p_pid; 1185 pt = &pid_table[pid & pid_tbl_mask]; 1186 KASSERT(pt->pt_pid == pid); 1187 if (PT_IS_PROC(pt->pt_slot)) { 1188 KASSERT(PT_GET_PROC(pt->pt_slot) == p); 1189 l->l_lid = pid; 1190 pt->pt_slot = PT_SET_LWP(l); 1191 } else { 1192 /* Need to allocate a new slot. */ 1193 pid = proc_alloc_pid_slot(p, PT_SET_LWP(l)); 1194 if (pid != -1) 1195 l->l_lid = pid; 1196 } 1197 mutex_exit(&proc_lock); 1198 1199 return pid; 1200 } 1201 1202 static void __noinline 1203 proc_free_pid_internal(pid_t pid, uintptr_t type __diagused) 1204 { 1205 struct pid_table *pt; 1206 1207 KASSERT(mutex_owned(&proc_lock)); 1208 1209 pt = &pid_table[pid & pid_tbl_mask]; 1210 1211 KASSERT(PT_GET_TYPE(pt->pt_slot) == type); 1212 KASSERT(pt->pt_pid == pid); 1213 1214 /* save pid use count in slot */ 1215 pt->pt_slot = PT_SET_FREE(pid & ~pid_tbl_mask); 1216 pt->pt_pid = 0; 1217 1218 if (pt->pt_pgrp == NULL) { 1219 /* link last freed entry onto ours */ 1220 pid &= pid_tbl_mask; 1221 pt = &pid_table[last_free_pt]; 1222 pt->pt_slot = PT_SET_FREE(PT_NEXT(pt->pt_slot) | pid); 1223 pt->pt_pid = 0; 1224 last_free_pt = pid; 1225 pid_alloc_cnt--; 1226 } 1227 } 1228 1229 /* 1230 * Free a process id - called from proc_free (in kern_exit.c) 1231 * 1232 * Called with the proc_lock held. 1233 */ 1234 void 1235 proc_free_pid(pid_t pid) 1236 { 1237 1238 KASSERT(mutex_owned(&proc_lock)); 1239 proc_free_pid_internal(pid, PT_F_PROC); 1240 } 1241 1242 /* 1243 * Free a process id used by an LWP. If this was the process's 1244 * first LWP, we convert the slot to point to the process; the 1245 * entry will get cleaned up later when the process finishes exiting. 1246 * 1247 * If not, then it's the same as proc_free_pid(). 1248 */ 1249 void 1250 proc_free_lwpid(struct proc *p, pid_t pid) 1251 { 1252 1253 KASSERT(mutex_owned(&proc_lock)); 1254 1255 if (__predict_true(p->p_pid == pid)) { 1256 struct pid_table *pt; 1257 1258 pt = &pid_table[pid & pid_tbl_mask]; 1259 1260 KASSERT(pt->pt_pid == pid); 1261 KASSERT(PT_IS_LWP(pt->pt_slot)); 1262 KASSERT(PT_GET_LWP(pt->pt_slot)->l_proc == p); 1263 1264 pt->pt_slot = PT_SET_PROC(p); 1265 return; 1266 } 1267 proc_free_pid_internal(pid, PT_F_LWP); 1268 } 1269 1270 void 1271 proc_free_mem(struct proc *p) 1272 { 1273 1274 kdtrace_proc_dtor(NULL, p); 1275 pool_cache_put(proc_cache, p); 1276 } 1277 1278 /* 1279 * proc_enterpgrp: move p to a new or existing process group (and session). 1280 * 1281 * If we are creating a new pgrp, the pgid should equal 1282 * the calling process' pid. 1283 * If is only valid to enter a process group that is in the session 1284 * of the process. 1285 * Also mksess should only be set if we are creating a process group 1286 * 1287 * Only called from sys_setsid, sys_setpgid and posix_spawn/spawn_return. 1288 */ 1289 int 1290 proc_enterpgrp(struct proc *curp, pid_t pid, pid_t pgid, bool mksess) 1291 { 1292 struct pgrp *new_pgrp, *pgrp; 1293 struct session *sess; 1294 struct proc *p; 1295 int rval; 1296 pid_t pg_id = NO_PGID; 1297 1298 /* Allocate data areas we might need before doing any validity checks */ 1299 sess = mksess ? kmem_alloc(sizeof(*sess), KM_SLEEP) : NULL; 1300 new_pgrp = kmem_alloc(sizeof(*new_pgrp), KM_SLEEP); 1301 1302 mutex_enter(&proc_lock); 1303 rval = EPERM; /* most common error (to save typing) */ 1304 1305 /* Check pgrp exists or can be created */ 1306 pgrp = pid_table[pgid & pid_tbl_mask].pt_pgrp; 1307 if (pgrp != NULL && pgrp->pg_id != pgid) 1308 goto done; 1309 1310 /* Can only set another process under restricted circumstances. */ 1311 if (pid != curp->p_pid) { 1312 /* Must exist and be one of our children... */ 1313 p = proc_find_internal(pid, false); 1314 if (p == NULL || !p_inferior(p, curp)) { 1315 rval = ESRCH; 1316 goto done; 1317 } 1318 /* ... in the same session... */ 1319 if (sess != NULL || p->p_session != curp->p_session) 1320 goto done; 1321 /* ... existing pgid must be in same session ... */ 1322 if (pgrp != NULL && pgrp->pg_session != p->p_session) 1323 goto done; 1324 /* ... and not done an exec. */ 1325 if (p->p_flag & PK_EXEC) { 1326 rval = EACCES; 1327 goto done; 1328 } 1329 } else { 1330 /* ... setsid() cannot re-enter a pgrp */ 1331 if (mksess && (curp->p_pgid == curp->p_pid || 1332 pgrp_find(curp->p_pid))) 1333 goto done; 1334 p = curp; 1335 } 1336 1337 /* Changing the process group/session of a session 1338 leader is definitely off limits. */ 1339 if (SESS_LEADER(p)) { 1340 if (sess == NULL && p->p_pgrp == pgrp) 1341 /* unless it's a definite noop */ 1342 rval = 0; 1343 goto done; 1344 } 1345 1346 /* Can only create a process group with id of process */ 1347 if (pgrp == NULL && pgid != pid) 1348 goto done; 1349 1350 /* Can only create a session if creating pgrp */ 1351 if (sess != NULL && pgrp != NULL) 1352 goto done; 1353 1354 /* Check we allocated memory for a pgrp... */ 1355 if (pgrp == NULL && new_pgrp == NULL) 1356 goto done; 1357 1358 /* Don't attach to 'zombie' pgrp */ 1359 if (pgrp != NULL && LIST_EMPTY(&pgrp->pg_members)) 1360 goto done; 1361 1362 /* Expect to succeed now */ 1363 rval = 0; 1364 1365 if (pgrp == p->p_pgrp) 1366 /* nothing to do */ 1367 goto done; 1368 1369 /* Ok all setup, link up required structures */ 1370 1371 if (pgrp == NULL) { 1372 pgrp = new_pgrp; 1373 new_pgrp = NULL; 1374 if (sess != NULL) { 1375 sess->s_sid = p->p_pid; 1376 sess->s_leader = p; 1377 sess->s_count = 1; 1378 sess->s_ttyvp = NULL; 1379 sess->s_ttyp = NULL; 1380 sess->s_flags = p->p_session->s_flags & ~S_LOGIN_SET; 1381 memcpy(sess->s_login, p->p_session->s_login, 1382 sizeof(sess->s_login)); 1383 p->p_lflag &= ~PL_CONTROLT; 1384 } else { 1385 sess = p->p_pgrp->pg_session; 1386 proc_sesshold(sess); 1387 } 1388 pgrp->pg_session = sess; 1389 sess = NULL; 1390 1391 pgrp->pg_id = pgid; 1392 LIST_INIT(&pgrp->pg_members); 1393 #ifdef DIAGNOSTIC 1394 if (__predict_false(pid_table[pgid & pid_tbl_mask].pt_pgrp)) 1395 panic("enterpgrp: pgrp table slot in use"); 1396 if (__predict_false(mksess && p != curp)) 1397 panic("enterpgrp: mksession and p != curproc"); 1398 #endif 1399 pid_table[pgid & pid_tbl_mask].pt_pgrp = pgrp; 1400 pgrp->pg_jobc = 0; 1401 } 1402 1403 /* 1404 * Adjust eligibility of affected pgrps to participate in job control. 1405 * Increment eligibility counts before decrementing, otherwise we 1406 * could reach 0 spuriously during the first call. 1407 */ 1408 fixjobc(p, pgrp, 1); 1409 fixjobc(p, p->p_pgrp, 0); 1410 1411 /* Interlock with ttread(). */ 1412 mutex_spin_enter(&tty_lock); 1413 1414 /* Move process to requested group. */ 1415 LIST_REMOVE(p, p_pglist); 1416 if (LIST_EMPTY(&p->p_pgrp->pg_members)) 1417 /* defer delete until we've dumped the lock */ 1418 pg_id = p->p_pgrp->pg_id; 1419 p->p_pgrp = pgrp; 1420 LIST_INSERT_HEAD(&pgrp->pg_members, p, p_pglist); 1421 1422 /* Done with the swap; we can release the tty mutex. */ 1423 mutex_spin_exit(&tty_lock); 1424 1425 done: 1426 if (pg_id != NO_PGID) { 1427 /* Releases proc_lock. */ 1428 pg_delete(pg_id); 1429 } else { 1430 mutex_exit(&proc_lock); 1431 } 1432 if (sess != NULL) 1433 kmem_free(sess, sizeof(*sess)); 1434 if (new_pgrp != NULL) 1435 kmem_free(new_pgrp, sizeof(*new_pgrp)); 1436 #ifdef DEBUG_PGRP 1437 if (__predict_false(rval)) 1438 printf("enterpgrp(%d,%d,%d), curproc %d, rval %d\n", 1439 pid, pgid, mksess, curp->p_pid, rval); 1440 #endif 1441 return rval; 1442 } 1443 1444 /* 1445 * proc_leavepgrp: remove a process from its process group. 1446 * => must be called with the proc_lock held, which will be released; 1447 */ 1448 void 1449 proc_leavepgrp(struct proc *p) 1450 { 1451 struct pgrp *pgrp; 1452 1453 KASSERT(mutex_owned(&proc_lock)); 1454 1455 /* Interlock with ttread() */ 1456 mutex_spin_enter(&tty_lock); 1457 pgrp = p->p_pgrp; 1458 LIST_REMOVE(p, p_pglist); 1459 p->p_pgrp = NULL; 1460 mutex_spin_exit(&tty_lock); 1461 1462 if (LIST_EMPTY(&pgrp->pg_members)) { 1463 /* Releases proc_lock. */ 1464 pg_delete(pgrp->pg_id); 1465 } else { 1466 mutex_exit(&proc_lock); 1467 } 1468 } 1469 1470 /* 1471 * pg_remove: remove a process group from the table. 1472 * => must be called with the proc_lock held; 1473 * => returns process group to free; 1474 */ 1475 static struct pgrp * 1476 pg_remove(pid_t pg_id) 1477 { 1478 struct pgrp *pgrp; 1479 struct pid_table *pt; 1480 1481 KASSERT(mutex_owned(&proc_lock)); 1482 1483 pt = &pid_table[pg_id & pid_tbl_mask]; 1484 pgrp = pt->pt_pgrp; 1485 1486 KASSERT(pgrp != NULL); 1487 KASSERT(pgrp->pg_id == pg_id); 1488 KASSERT(LIST_EMPTY(&pgrp->pg_members)); 1489 1490 pt->pt_pgrp = NULL; 1491 1492 if (!PT_VALID(pt->pt_slot)) { 1493 /* Orphaned pgrp, put slot onto free list. */ 1494 KASSERT((PT_NEXT(pt->pt_slot) & pid_tbl_mask) == 0); 1495 pg_id &= pid_tbl_mask; 1496 pt = &pid_table[last_free_pt]; 1497 pt->pt_slot = PT_SET_FREE(PT_NEXT(pt->pt_slot) | pg_id); 1498 KASSERT(pt->pt_pid == 0); 1499 last_free_pt = pg_id; 1500 pid_alloc_cnt--; 1501 } 1502 return pgrp; 1503 } 1504 1505 /* 1506 * pg_delete: delete and free a process group. 1507 * => must be called with the proc_lock held, which will be released. 1508 */ 1509 static void 1510 pg_delete(pid_t pg_id) 1511 { 1512 struct pgrp *pg; 1513 struct tty *ttyp; 1514 struct session *ss; 1515 1516 KASSERT(mutex_owned(&proc_lock)); 1517 1518 pg = pid_table[pg_id & pid_tbl_mask].pt_pgrp; 1519 if (pg == NULL || pg->pg_id != pg_id || !LIST_EMPTY(&pg->pg_members)) { 1520 mutex_exit(&proc_lock); 1521 return; 1522 } 1523 1524 ss = pg->pg_session; 1525 1526 /* Remove reference (if any) from tty to this process group */ 1527 mutex_spin_enter(&tty_lock); 1528 ttyp = ss->s_ttyp; 1529 if (ttyp != NULL && ttyp->t_pgrp == pg) { 1530 ttyp->t_pgrp = NULL; 1531 KASSERT(ttyp->t_session == ss); 1532 } 1533 mutex_spin_exit(&tty_lock); 1534 1535 /* 1536 * The leading process group in a session is freed by proc_sessrele(), 1537 * if last reference. It will also release the locks. 1538 */ 1539 pg = (ss->s_sid != pg->pg_id) ? pg_remove(pg_id) : NULL; 1540 proc_sessrele(ss); 1541 1542 if (pg != NULL) { 1543 /* Free it, if was not done above. */ 1544 kmem_free(pg, sizeof(struct pgrp)); 1545 } 1546 } 1547 1548 /* 1549 * Adjust pgrp jobc counters when specified process changes process group. 1550 * We count the number of processes in each process group that "qualify" 1551 * the group for terminal job control (those with a parent in a different 1552 * process group of the same session). If that count reaches zero, the 1553 * process group becomes orphaned. Check both the specified process' 1554 * process group and that of its children. 1555 * entering == 0 => p is leaving specified group. 1556 * entering == 1 => p is entering specified group. 1557 * 1558 * Call with proc_lock held. 1559 */ 1560 void 1561 fixjobc(struct proc *p, struct pgrp *pgrp, int entering) 1562 { 1563 struct pgrp *hispgrp; 1564 struct session *mysession = pgrp->pg_session; 1565 struct proc *child; 1566 1567 KASSERT(mutex_owned(&proc_lock)); 1568 1569 /* 1570 * Check p's parent to see whether p qualifies its own process 1571 * group; if so, adjust count for p's process group. 1572 */ 1573 hispgrp = p->p_pptr->p_pgrp; 1574 if (hispgrp != pgrp && hispgrp->pg_session == mysession) { 1575 if (entering) { 1576 pgrp->pg_jobc++; 1577 p->p_lflag &= ~PL_ORPHANPG; 1578 } else { 1579 /* KASSERT(pgrp->pg_jobc > 0); */ 1580 if (--pgrp->pg_jobc == 0) 1581 orphanpg(pgrp); 1582 } 1583 } 1584 1585 /* 1586 * Check this process' children to see whether they qualify 1587 * their process groups; if so, adjust counts for children's 1588 * process groups. 1589 */ 1590 LIST_FOREACH(child, &p->p_children, p_sibling) { 1591 hispgrp = child->p_pgrp; 1592 if (hispgrp != pgrp && hispgrp->pg_session == mysession && 1593 !P_ZOMBIE(child)) { 1594 if (entering) { 1595 child->p_lflag &= ~PL_ORPHANPG; 1596 hispgrp->pg_jobc++; 1597 } else { 1598 KASSERT(hispgrp->pg_jobc > 0); 1599 if (--hispgrp->pg_jobc == 0) 1600 orphanpg(hispgrp); 1601 } 1602 } 1603 } 1604 } 1605 1606 /* 1607 * A process group has become orphaned; 1608 * if there are any stopped processes in the group, 1609 * hang-up all process in that group. 1610 * 1611 * Call with proc_lock held. 1612 */ 1613 static void 1614 orphanpg(struct pgrp *pg) 1615 { 1616 struct proc *p; 1617 1618 KASSERT(mutex_owned(&proc_lock)); 1619 1620 LIST_FOREACH(p, &pg->pg_members, p_pglist) { 1621 if (p->p_stat == SSTOP) { 1622 p->p_lflag |= PL_ORPHANPG; 1623 psignal(p, SIGHUP); 1624 psignal(p, SIGCONT); 1625 } 1626 } 1627 } 1628 1629 #ifdef DDB 1630 #include <ddb/db_output.h> 1631 void pidtbl_dump(void); 1632 void 1633 pidtbl_dump(void) 1634 { 1635 struct pid_table *pt; 1636 struct proc *p; 1637 struct pgrp *pgrp; 1638 uintptr_t slot; 1639 int id; 1640 1641 db_printf("pid table %p size %x, next %x, last %x\n", 1642 pid_table, pid_tbl_mask+1, 1643 next_free_pt, last_free_pt); 1644 for (pt = pid_table, id = 0; id <= pid_tbl_mask; id++, pt++) { 1645 slot = pt->pt_slot; 1646 if (!PT_VALID(slot) && !pt->pt_pgrp) 1647 continue; 1648 if (PT_IS_LWP(slot)) { 1649 p = PT_GET_LWP(slot)->l_proc; 1650 } else if (PT_IS_PROC(slot)) { 1651 p = PT_GET_PROC(slot); 1652 } else { 1653 p = NULL; 1654 } 1655 db_printf(" id %x: ", id); 1656 if (p != NULL) 1657 db_printf("slotpid %d proc %p id %d (0x%x) %s\n", 1658 pt->pt_pid, p, p->p_pid, p->p_pid, p->p_comm); 1659 else 1660 db_printf("next %x use %x\n", 1661 PT_NEXT(slot) & pid_tbl_mask, 1662 PT_NEXT(slot) & ~pid_tbl_mask); 1663 if ((pgrp = pt->pt_pgrp)) { 1664 db_printf("\tsession %p, sid %d, count %d, login %s\n", 1665 pgrp->pg_session, pgrp->pg_session->s_sid, 1666 pgrp->pg_session->s_count, 1667 pgrp->pg_session->s_login); 1668 db_printf("\tpgrp %p, pg_id %d, pg_jobc %d, members %p\n", 1669 pgrp, pgrp->pg_id, pgrp->pg_jobc, 1670 LIST_FIRST(&pgrp->pg_members)); 1671 LIST_FOREACH(p, &pgrp->pg_members, p_pglist) { 1672 db_printf("\t\tpid %d addr %p pgrp %p %s\n", 1673 p->p_pid, p, p->p_pgrp, p->p_comm); 1674 } 1675 } 1676 } 1677 } 1678 #endif /* DDB */ 1679 1680 #ifdef KSTACK_CHECK_MAGIC 1681 1682 #define KSTACK_MAGIC 0xdeadbeaf 1683 1684 /* XXX should be per process basis? */ 1685 static int kstackleftmin = KSTACK_SIZE; 1686 static int kstackleftthres = KSTACK_SIZE / 8; 1687 1688 void 1689 kstack_setup_magic(const struct lwp *l) 1690 { 1691 uint32_t *ip; 1692 uint32_t const *end; 1693 1694 KASSERT(l != NULL); 1695 KASSERT(l != &lwp0); 1696 1697 /* 1698 * fill all the stack with magic number 1699 * so that later modification on it can be detected. 1700 */ 1701 ip = (uint32_t *)KSTACK_LOWEST_ADDR(l); 1702 end = (uint32_t *)((char *)KSTACK_LOWEST_ADDR(l) + KSTACK_SIZE); 1703 for (; ip < end; ip++) { 1704 *ip = KSTACK_MAGIC; 1705 } 1706 } 1707 1708 void 1709 kstack_check_magic(const struct lwp *l) 1710 { 1711 uint32_t const *ip, *end; 1712 int stackleft; 1713 1714 KASSERT(l != NULL); 1715 1716 /* don't check proc0 */ /*XXX*/ 1717 if (l == &lwp0) 1718 return; 1719 1720 #ifdef __MACHINE_STACK_GROWS_UP 1721 /* stack grows upwards (eg. hppa) */ 1722 ip = (uint32_t *)((void *)KSTACK_LOWEST_ADDR(l) + KSTACK_SIZE); 1723 end = (uint32_t *)KSTACK_LOWEST_ADDR(l); 1724 for (ip--; ip >= end; ip--) 1725 if (*ip != KSTACK_MAGIC) 1726 break; 1727 1728 stackleft = (void *)KSTACK_LOWEST_ADDR(l) + KSTACK_SIZE - (void *)ip; 1729 #else /* __MACHINE_STACK_GROWS_UP */ 1730 /* stack grows downwards (eg. i386) */ 1731 ip = (uint32_t *)KSTACK_LOWEST_ADDR(l); 1732 end = (uint32_t *)((char *)KSTACK_LOWEST_ADDR(l) + KSTACK_SIZE); 1733 for (; ip < end; ip++) 1734 if (*ip != KSTACK_MAGIC) 1735 break; 1736 1737 stackleft = ((const char *)ip) - (const char *)KSTACK_LOWEST_ADDR(l); 1738 #endif /* __MACHINE_STACK_GROWS_UP */ 1739 1740 if (kstackleftmin > stackleft) { 1741 kstackleftmin = stackleft; 1742 if (stackleft < kstackleftthres) 1743 printf("warning: kernel stack left %d bytes" 1744 "(pid %u:lid %u)\n", stackleft, 1745 (u_int)l->l_proc->p_pid, (u_int)l->l_lid); 1746 } 1747 1748 if (stackleft <= 0) { 1749 panic("magic on the top of kernel stack changed for " 1750 "pid %u, lid %u: maybe kernel stack overflow", 1751 (u_int)l->l_proc->p_pid, (u_int)l->l_lid); 1752 } 1753 } 1754 #endif /* KSTACK_CHECK_MAGIC */ 1755 1756 int 1757 proclist_foreach_call(struct proclist *list, 1758 int (*callback)(struct proc *, void *arg), void *arg) 1759 { 1760 struct proc marker; 1761 struct proc *p; 1762 int ret = 0; 1763 1764 marker.p_flag = PK_MARKER; 1765 mutex_enter(&proc_lock); 1766 for (p = LIST_FIRST(list); ret == 0 && p != NULL;) { 1767 if (p->p_flag & PK_MARKER) { 1768 p = LIST_NEXT(p, p_list); 1769 continue; 1770 } 1771 LIST_INSERT_AFTER(p, &marker, p_list); 1772 ret = (*callback)(p, arg); 1773 KASSERT(mutex_owned(&proc_lock)); 1774 p = LIST_NEXT(&marker, p_list); 1775 LIST_REMOVE(&marker, p_list); 1776 } 1777 mutex_exit(&proc_lock); 1778 1779 return ret; 1780 } 1781 1782 int 1783 proc_vmspace_getref(struct proc *p, struct vmspace **vm) 1784 { 1785 1786 /* XXXCDC: how should locking work here? */ 1787 1788 /* curproc exception is for coredump. */ 1789 1790 if ((p != curproc && (p->p_sflag & PS_WEXIT) != 0) || 1791 (p->p_vmspace->vm_refcnt < 1)) { 1792 return EFAULT; 1793 } 1794 1795 uvmspace_addref(p->p_vmspace); 1796 *vm = p->p_vmspace; 1797 1798 return 0; 1799 } 1800 1801 /* 1802 * Acquire a write lock on the process credential. 1803 */ 1804 void 1805 proc_crmod_enter(void) 1806 { 1807 struct lwp *l = curlwp; 1808 struct proc *p = l->l_proc; 1809 kauth_cred_t oc; 1810 1811 /* Reset what needs to be reset in plimit. */ 1812 if (p->p_limit->pl_corename != defcorename) { 1813 lim_setcorename(p, defcorename, 0); 1814 } 1815 1816 mutex_enter(p->p_lock); 1817 1818 /* Ensure the LWP cached credentials are up to date. */ 1819 if ((oc = l->l_cred) != p->p_cred) { 1820 kauth_cred_hold(p->p_cred); 1821 l->l_cred = p->p_cred; 1822 kauth_cred_free(oc); 1823 } 1824 } 1825 1826 /* 1827 * Set in a new process credential, and drop the write lock. The credential 1828 * must have a reference already. Optionally, free a no-longer required 1829 * credential. 1830 */ 1831 void 1832 proc_crmod_leave(kauth_cred_t scred, kauth_cred_t fcred, bool sugid) 1833 { 1834 struct lwp *l = curlwp, *l2; 1835 struct proc *p = l->l_proc; 1836 kauth_cred_t oc; 1837 1838 KASSERT(mutex_owned(p->p_lock)); 1839 1840 /* Is there a new credential to set in? */ 1841 if (scred != NULL) { 1842 p->p_cred = scred; 1843 LIST_FOREACH(l2, &p->p_lwps, l_sibling) { 1844 if (l2 != l) 1845 l2->l_prflag |= LPR_CRMOD; 1846 } 1847 1848 /* Ensure the LWP cached credentials are up to date. */ 1849 if ((oc = l->l_cred) != scred) { 1850 kauth_cred_hold(scred); 1851 l->l_cred = scred; 1852 } 1853 } else 1854 oc = NULL; /* XXXgcc */ 1855 1856 if (sugid) { 1857 /* 1858 * Mark process as having changed credentials, stops 1859 * tracing etc. 1860 */ 1861 p->p_flag |= PK_SUGID; 1862 } 1863 1864 mutex_exit(p->p_lock); 1865 1866 /* If there is a credential to be released, free it now. */ 1867 if (fcred != NULL) { 1868 KASSERT(scred != NULL); 1869 kauth_cred_free(fcred); 1870 if (oc != scred) 1871 kauth_cred_free(oc); 1872 } 1873 } 1874 1875 /* 1876 * proc_specific_key_create -- 1877 * Create a key for subsystem proc-specific data. 1878 */ 1879 int 1880 proc_specific_key_create(specificdata_key_t *keyp, specificdata_dtor_t dtor) 1881 { 1882 1883 return (specificdata_key_create(proc_specificdata_domain, keyp, dtor)); 1884 } 1885 1886 /* 1887 * proc_specific_key_delete -- 1888 * Delete a key for subsystem proc-specific data. 1889 */ 1890 void 1891 proc_specific_key_delete(specificdata_key_t key) 1892 { 1893 1894 specificdata_key_delete(proc_specificdata_domain, key); 1895 } 1896 1897 /* 1898 * proc_initspecific -- 1899 * Initialize a proc's specificdata container. 1900 */ 1901 void 1902 proc_initspecific(struct proc *p) 1903 { 1904 int error __diagused; 1905 1906 error = specificdata_init(proc_specificdata_domain, &p->p_specdataref); 1907 KASSERT(error == 0); 1908 } 1909 1910 /* 1911 * proc_finispecific -- 1912 * Finalize a proc's specificdata container. 1913 */ 1914 void 1915 proc_finispecific(struct proc *p) 1916 { 1917 1918 specificdata_fini(proc_specificdata_domain, &p->p_specdataref); 1919 } 1920 1921 /* 1922 * proc_getspecific -- 1923 * Return proc-specific data corresponding to the specified key. 1924 */ 1925 void * 1926 proc_getspecific(struct proc *p, specificdata_key_t key) 1927 { 1928 1929 return (specificdata_getspecific(proc_specificdata_domain, 1930 &p->p_specdataref, key)); 1931 } 1932 1933 /* 1934 * proc_setspecific -- 1935 * Set proc-specific data corresponding to the specified key. 1936 */ 1937 void 1938 proc_setspecific(struct proc *p, specificdata_key_t key, void *data) 1939 { 1940 1941 specificdata_setspecific(proc_specificdata_domain, 1942 &p->p_specdataref, key, data); 1943 } 1944 1945 int 1946 proc_uidmatch(kauth_cred_t cred, kauth_cred_t target) 1947 { 1948 int r = 0; 1949 1950 if (kauth_cred_getuid(cred) != kauth_cred_getuid(target) || 1951 kauth_cred_getuid(cred) != kauth_cred_getsvuid(target)) { 1952 /* 1953 * suid proc of ours or proc not ours 1954 */ 1955 r = EPERM; 1956 } else if (kauth_cred_getgid(target) != kauth_cred_getsvgid(target)) { 1957 /* 1958 * sgid proc has sgid back to us temporarily 1959 */ 1960 r = EPERM; 1961 } else { 1962 /* 1963 * our rgid must be in target's group list (ie, 1964 * sub-processes started by a sgid process) 1965 */ 1966 int ismember = 0; 1967 1968 if (kauth_cred_ismember_gid(cred, 1969 kauth_cred_getgid(target), &ismember) != 0 || 1970 !ismember) 1971 r = EPERM; 1972 } 1973 1974 return (r); 1975 } 1976 1977 /* 1978 * sysctl stuff 1979 */ 1980 1981 #define KERN_PROCSLOP (5 * sizeof(struct kinfo_proc)) 1982 1983 static const u_int sysctl_flagmap[] = { 1984 PK_ADVLOCK, P_ADVLOCK, 1985 PK_EXEC, P_EXEC, 1986 PK_NOCLDWAIT, P_NOCLDWAIT, 1987 PK_32, P_32, 1988 PK_CLDSIGIGN, P_CLDSIGIGN, 1989 PK_SUGID, P_SUGID, 1990 0 1991 }; 1992 1993 static const u_int sysctl_sflagmap[] = { 1994 PS_NOCLDSTOP, P_NOCLDSTOP, 1995 PS_WEXIT, P_WEXIT, 1996 PS_STOPFORK, P_STOPFORK, 1997 PS_STOPEXEC, P_STOPEXEC, 1998 PS_STOPEXIT, P_STOPEXIT, 1999 0 2000 }; 2001 2002 static const u_int sysctl_slflagmap[] = { 2003 PSL_TRACED, P_TRACED, 2004 PSL_CHTRACED, P_CHTRACED, 2005 PSL_SYSCALL, P_SYSCALL, 2006 0 2007 }; 2008 2009 static const u_int sysctl_lflagmap[] = { 2010 PL_CONTROLT, P_CONTROLT, 2011 PL_PPWAIT, P_PPWAIT, 2012 0 2013 }; 2014 2015 static const u_int sysctl_stflagmap[] = { 2016 PST_PROFIL, P_PROFIL, 2017 0 2018 2019 }; 2020 2021 /* used by kern_lwp also */ 2022 const u_int sysctl_lwpflagmap[] = { 2023 LW_SINTR, L_SINTR, 2024 LW_SYSTEM, L_SYSTEM, 2025 0 2026 }; 2027 2028 /* 2029 * Find the most ``active'' lwp of a process and return it for ps display 2030 * purposes 2031 */ 2032 static struct lwp * 2033 proc_active_lwp(struct proc *p) 2034 { 2035 static const int ostat[] = { 2036 0, 2037 2, /* LSIDL */ 2038 6, /* LSRUN */ 2039 5, /* LSSLEEP */ 2040 4, /* LSSTOP */ 2041 0, /* LSZOMB */ 2042 1, /* LSDEAD */ 2043 7, /* LSONPROC */ 2044 3 /* LSSUSPENDED */ 2045 }; 2046 2047 struct lwp *l, *lp = NULL; 2048 LIST_FOREACH(l, &p->p_lwps, l_sibling) { 2049 KASSERT(l->l_stat >= 0 && l->l_stat < __arraycount(ostat)); 2050 if (lp == NULL || 2051 ostat[l->l_stat] > ostat[lp->l_stat] || 2052 (ostat[l->l_stat] == ostat[lp->l_stat] && 2053 l->l_cpticks > lp->l_cpticks)) { 2054 lp = l; 2055 continue; 2056 } 2057 } 2058 return lp; 2059 } 2060 2061 static int 2062 sysctl_doeproc(SYSCTLFN_ARGS) 2063 { 2064 union { 2065 struct kinfo_proc kproc; 2066 struct kinfo_proc2 kproc2; 2067 } *kbuf; 2068 struct proc *p, *next, *marker; 2069 char *where, *dp; 2070 int type, op, arg, error; 2071 u_int elem_size, kelem_size, elem_count; 2072 size_t buflen, needed; 2073 bool match, zombie, mmmbrains; 2074 const bool allowaddr = get_expose_address(curproc); 2075 2076 if (namelen == 1 && name[0] == CTL_QUERY) 2077 return (sysctl_query(SYSCTLFN_CALL(rnode))); 2078 2079 dp = where = oldp; 2080 buflen = where != NULL ? *oldlenp : 0; 2081 error = 0; 2082 needed = 0; 2083 type = rnode->sysctl_num; 2084 2085 if (type == KERN_PROC) { 2086 if (namelen == 0) 2087 return EINVAL; 2088 switch (op = name[0]) { 2089 case KERN_PROC_ALL: 2090 if (namelen != 1) 2091 return EINVAL; 2092 arg = 0; 2093 break; 2094 default: 2095 if (namelen != 2) 2096 return EINVAL; 2097 arg = name[1]; 2098 break; 2099 } 2100 elem_count = 0; /* Hush little compiler, don't you cry */ 2101 kelem_size = elem_size = sizeof(kbuf->kproc); 2102 } else { 2103 if (namelen != 4) 2104 return EINVAL; 2105 op = name[0]; 2106 arg = name[1]; 2107 elem_size = name[2]; 2108 elem_count = name[3]; 2109 kelem_size = sizeof(kbuf->kproc2); 2110 } 2111 2112 sysctl_unlock(); 2113 2114 kbuf = kmem_zalloc(sizeof(*kbuf), KM_SLEEP); 2115 marker = kmem_alloc(sizeof(*marker), KM_SLEEP); 2116 marker->p_flag = PK_MARKER; 2117 2118 mutex_enter(&proc_lock); 2119 /* 2120 * Start with zombies to prevent reporting processes twice, in case they 2121 * are dying and being moved from the list of alive processes to zombies. 2122 */ 2123 mmmbrains = true; 2124 for (p = LIST_FIRST(&zombproc);; p = next) { 2125 if (p == NULL) { 2126 if (mmmbrains) { 2127 p = LIST_FIRST(&allproc); 2128 mmmbrains = false; 2129 } 2130 if (p == NULL) 2131 break; 2132 } 2133 next = LIST_NEXT(p, p_list); 2134 if ((p->p_flag & PK_MARKER) != 0) 2135 continue; 2136 2137 /* 2138 * Skip embryonic processes. 2139 */ 2140 if (p->p_stat == SIDL) 2141 continue; 2142 2143 mutex_enter(p->p_lock); 2144 error = kauth_authorize_process(l->l_cred, 2145 KAUTH_PROCESS_CANSEE, p, 2146 KAUTH_ARG(KAUTH_REQ_PROCESS_CANSEE_EPROC), NULL, NULL); 2147 if (error != 0) { 2148 mutex_exit(p->p_lock); 2149 continue; 2150 } 2151 2152 /* 2153 * Hande all the operations in one switch on the cost of 2154 * algorithm complexity is on purpose. The win splitting this 2155 * function into several similar copies makes maintenance 2156 * burden, code grow and boost is negligible in practical 2157 * systems. 2158 */ 2159 switch (op) { 2160 case KERN_PROC_PID: 2161 match = (p->p_pid == (pid_t)arg); 2162 break; 2163 2164 case KERN_PROC_PGRP: 2165 match = (p->p_pgrp->pg_id == (pid_t)arg); 2166 break; 2167 2168 case KERN_PROC_SESSION: 2169 match = (p->p_session->s_sid == (pid_t)arg); 2170 break; 2171 2172 case KERN_PROC_TTY: 2173 match = true; 2174 if (arg == (int) KERN_PROC_TTY_REVOKE) { 2175 if ((p->p_lflag & PL_CONTROLT) == 0 || 2176 p->p_session->s_ttyp == NULL || 2177 p->p_session->s_ttyvp != NULL) { 2178 match = false; 2179 } 2180 } else if ((p->p_lflag & PL_CONTROLT) == 0 || 2181 p->p_session->s_ttyp == NULL) { 2182 if ((dev_t)arg != KERN_PROC_TTY_NODEV) { 2183 match = false; 2184 } 2185 } else if (p->p_session->s_ttyp->t_dev != (dev_t)arg) { 2186 match = false; 2187 } 2188 break; 2189 2190 case KERN_PROC_UID: 2191 match = (kauth_cred_geteuid(p->p_cred) == (uid_t)arg); 2192 break; 2193 2194 case KERN_PROC_RUID: 2195 match = (kauth_cred_getuid(p->p_cred) == (uid_t)arg); 2196 break; 2197 2198 case KERN_PROC_GID: 2199 match = (kauth_cred_getegid(p->p_cred) == (uid_t)arg); 2200 break; 2201 2202 case KERN_PROC_RGID: 2203 match = (kauth_cred_getgid(p->p_cred) == (uid_t)arg); 2204 break; 2205 2206 case KERN_PROC_ALL: 2207 match = true; 2208 /* allow everything */ 2209 break; 2210 2211 default: 2212 error = EINVAL; 2213 mutex_exit(p->p_lock); 2214 goto cleanup; 2215 } 2216 if (!match) { 2217 mutex_exit(p->p_lock); 2218 continue; 2219 } 2220 2221 /* 2222 * Grab a hold on the process. 2223 */ 2224 if (mmmbrains) { 2225 zombie = true; 2226 } else { 2227 zombie = !rw_tryenter(&p->p_reflock, RW_READER); 2228 } 2229 if (zombie) { 2230 LIST_INSERT_AFTER(p, marker, p_list); 2231 } 2232 2233 if (buflen >= elem_size && 2234 (type == KERN_PROC || elem_count > 0)) { 2235 ruspace(p); /* Update process vm resource use */ 2236 2237 if (type == KERN_PROC) { 2238 fill_proc(p, &kbuf->kproc.kp_proc, allowaddr); 2239 fill_eproc(p, &kbuf->kproc.kp_eproc, zombie, 2240 allowaddr); 2241 } else { 2242 fill_kproc2(p, &kbuf->kproc2, zombie, 2243 allowaddr); 2244 elem_count--; 2245 } 2246 mutex_exit(p->p_lock); 2247 mutex_exit(&proc_lock); 2248 /* 2249 * Copy out elem_size, but not larger than kelem_size 2250 */ 2251 error = sysctl_copyout(l, kbuf, dp, 2252 uimin(kelem_size, elem_size)); 2253 mutex_enter(&proc_lock); 2254 if (error) { 2255 goto bah; 2256 } 2257 dp += elem_size; 2258 buflen -= elem_size; 2259 } else { 2260 mutex_exit(p->p_lock); 2261 } 2262 needed += elem_size; 2263 2264 /* 2265 * Release reference to process. 2266 */ 2267 if (zombie) { 2268 next = LIST_NEXT(marker, p_list); 2269 LIST_REMOVE(marker, p_list); 2270 } else { 2271 rw_exit(&p->p_reflock); 2272 next = LIST_NEXT(p, p_list); 2273 } 2274 2275 /* 2276 * Short-circuit break quickly! 2277 */ 2278 if (op == KERN_PROC_PID) 2279 break; 2280 } 2281 mutex_exit(&proc_lock); 2282 2283 if (where != NULL) { 2284 *oldlenp = dp - where; 2285 if (needed > *oldlenp) { 2286 error = ENOMEM; 2287 goto out; 2288 } 2289 } else { 2290 needed += KERN_PROCSLOP; 2291 *oldlenp = needed; 2292 } 2293 kmem_free(kbuf, sizeof(*kbuf)); 2294 kmem_free(marker, sizeof(*marker)); 2295 sysctl_relock(); 2296 return 0; 2297 bah: 2298 if (zombie) 2299 LIST_REMOVE(marker, p_list); 2300 else 2301 rw_exit(&p->p_reflock); 2302 cleanup: 2303 mutex_exit(&proc_lock); 2304 out: 2305 kmem_free(kbuf, sizeof(*kbuf)); 2306 kmem_free(marker, sizeof(*marker)); 2307 sysctl_relock(); 2308 return error; 2309 } 2310 2311 int 2312 copyin_psstrings(struct proc *p, struct ps_strings *arginfo) 2313 { 2314 #if !defined(_RUMPKERNEL) 2315 int retval; 2316 2317 if (p->p_flag & PK_32) { 2318 MODULE_HOOK_CALL(kern_proc32_copyin_hook, (p, arginfo), 2319 enosys(), retval); 2320 return retval; 2321 } 2322 #endif /* !defined(_RUMPKERNEL) */ 2323 2324 return copyin_proc(p, (void *)p->p_psstrp, arginfo, sizeof(*arginfo)); 2325 } 2326 2327 static int 2328 copy_procargs_sysctl_cb(void *cookie_, const void *src, size_t off, size_t len) 2329 { 2330 void **cookie = cookie_; 2331 struct lwp *l = cookie[0]; 2332 char *dst = cookie[1]; 2333 2334 return sysctl_copyout(l, src, dst + off, len); 2335 } 2336 2337 /* 2338 * sysctl helper routine for kern.proc_args pseudo-subtree. 2339 */ 2340 static int 2341 sysctl_kern_proc_args(SYSCTLFN_ARGS) 2342 { 2343 struct ps_strings pss; 2344 struct proc *p; 2345 pid_t pid; 2346 int type, error; 2347 void *cookie[2]; 2348 2349 if (namelen == 1 && name[0] == CTL_QUERY) 2350 return (sysctl_query(SYSCTLFN_CALL(rnode))); 2351 2352 if (newp != NULL || namelen != 2) 2353 return (EINVAL); 2354 pid = name[0]; 2355 type = name[1]; 2356 2357 switch (type) { 2358 case KERN_PROC_PATHNAME: 2359 sysctl_unlock(); 2360 error = fill_pathname(l, pid, oldp, oldlenp); 2361 sysctl_relock(); 2362 return error; 2363 2364 case KERN_PROC_CWD: 2365 sysctl_unlock(); 2366 error = fill_cwd(l, pid, oldp, oldlenp); 2367 sysctl_relock(); 2368 return error; 2369 2370 case KERN_PROC_ARGV: 2371 case KERN_PROC_NARGV: 2372 case KERN_PROC_ENV: 2373 case KERN_PROC_NENV: 2374 /* ok */ 2375 break; 2376 default: 2377 return (EINVAL); 2378 } 2379 2380 sysctl_unlock(); 2381 2382 /* check pid */ 2383 mutex_enter(&proc_lock); 2384 if ((p = proc_find(pid)) == NULL) { 2385 error = EINVAL; 2386 goto out_locked; 2387 } 2388 mutex_enter(p->p_lock); 2389 2390 /* Check permission. */ 2391 if (type == KERN_PROC_ARGV || type == KERN_PROC_NARGV) 2392 error = kauth_authorize_process(l->l_cred, KAUTH_PROCESS_CANSEE, 2393 p, KAUTH_ARG(KAUTH_REQ_PROCESS_CANSEE_ARGS), NULL, NULL); 2394 else if (type == KERN_PROC_ENV || type == KERN_PROC_NENV) 2395 error = kauth_authorize_process(l->l_cred, KAUTH_PROCESS_CANSEE, 2396 p, KAUTH_ARG(KAUTH_REQ_PROCESS_CANSEE_ENV), NULL, NULL); 2397 else 2398 error = EINVAL; /* XXXGCC */ 2399 if (error) { 2400 mutex_exit(p->p_lock); 2401 goto out_locked; 2402 } 2403 2404 if (oldp == NULL) { 2405 if (type == KERN_PROC_NARGV || type == KERN_PROC_NENV) 2406 *oldlenp = sizeof (int); 2407 else 2408 *oldlenp = ARG_MAX; /* XXX XXX XXX */ 2409 error = 0; 2410 mutex_exit(p->p_lock); 2411 goto out_locked; 2412 } 2413 2414 /* 2415 * Zombies don't have a stack, so we can't read their psstrings. 2416 * System processes also don't have a user stack. 2417 */ 2418 if (P_ZOMBIE(p) || (p->p_flag & PK_SYSTEM) != 0) { 2419 error = EINVAL; 2420 mutex_exit(p->p_lock); 2421 goto out_locked; 2422 } 2423 2424 error = rw_tryenter(&p->p_reflock, RW_READER) ? 0 : EBUSY; 2425 mutex_exit(p->p_lock); 2426 if (error) { 2427 goto out_locked; 2428 } 2429 mutex_exit(&proc_lock); 2430 2431 if (type == KERN_PROC_NARGV || type == KERN_PROC_NENV) { 2432 int value; 2433 if ((error = copyin_psstrings(p, &pss)) == 0) { 2434 if (type == KERN_PROC_NARGV) 2435 value = pss.ps_nargvstr; 2436 else 2437 value = pss.ps_nenvstr; 2438 error = sysctl_copyout(l, &value, oldp, sizeof(value)); 2439 *oldlenp = sizeof(value); 2440 } 2441 } else { 2442 cookie[0] = l; 2443 cookie[1] = oldp; 2444 error = copy_procargs(p, type, oldlenp, 2445 copy_procargs_sysctl_cb, cookie); 2446 } 2447 rw_exit(&p->p_reflock); 2448 sysctl_relock(); 2449 return error; 2450 2451 out_locked: 2452 mutex_exit(&proc_lock); 2453 sysctl_relock(); 2454 return error; 2455 } 2456 2457 int 2458 copy_procargs(struct proc *p, int oid, size_t *limit, 2459 int (*cb)(void *, const void *, size_t, size_t), void *cookie) 2460 { 2461 struct ps_strings pss; 2462 size_t len, i, loaded, entry_len; 2463 struct uio auio; 2464 struct iovec aiov; 2465 int error, argvlen; 2466 char *arg; 2467 char **argv; 2468 vaddr_t user_argv; 2469 struct vmspace *vmspace; 2470 2471 /* 2472 * Allocate a temporary buffer to hold the argument vector and 2473 * the arguments themselve. 2474 */ 2475 arg = kmem_alloc(PAGE_SIZE, KM_SLEEP); 2476 argv = kmem_alloc(PAGE_SIZE, KM_SLEEP); 2477 2478 /* 2479 * Lock the process down in memory. 2480 */ 2481 vmspace = p->p_vmspace; 2482 uvmspace_addref(vmspace); 2483 2484 /* 2485 * Read in the ps_strings structure. 2486 */ 2487 if ((error = copyin_psstrings(p, &pss)) != 0) 2488 goto done; 2489 2490 /* 2491 * Now read the address of the argument vector. 2492 */ 2493 switch (oid) { 2494 case KERN_PROC_ARGV: 2495 user_argv = (uintptr_t)pss.ps_argvstr; 2496 argvlen = pss.ps_nargvstr; 2497 break; 2498 case KERN_PROC_ENV: 2499 user_argv = (uintptr_t)pss.ps_envstr; 2500 argvlen = pss.ps_nenvstr; 2501 break; 2502 default: 2503 error = EINVAL; 2504 goto done; 2505 } 2506 2507 if (argvlen < 0) { 2508 error = EIO; 2509 goto done; 2510 } 2511 2512 2513 /* 2514 * Now copy each string. 2515 */ 2516 len = 0; /* bytes written to user buffer */ 2517 loaded = 0; /* bytes from argv already processed */ 2518 i = 0; /* To make compiler happy */ 2519 entry_len = PROC_PTRSZ(p); 2520 2521 for (; argvlen; --argvlen) { 2522 int finished = 0; 2523 vaddr_t base; 2524 size_t xlen; 2525 int j; 2526 2527 if (loaded == 0) { 2528 size_t rem = entry_len * argvlen; 2529 loaded = MIN(rem, PAGE_SIZE); 2530 error = copyin_vmspace(vmspace, 2531 (const void *)user_argv, argv, loaded); 2532 if (error) 2533 break; 2534 user_argv += loaded; 2535 i = 0; 2536 } 2537 2538 #if !defined(_RUMPKERNEL) 2539 if (p->p_flag & PK_32) 2540 MODULE_HOOK_CALL(kern_proc32_base_hook, 2541 (argv, i++), 0, base); 2542 else 2543 #endif /* !defined(_RUMPKERNEL) */ 2544 base = (vaddr_t)argv[i++]; 2545 loaded -= entry_len; 2546 2547 /* 2548 * The program has messed around with its arguments, 2549 * possibly deleting some, and replacing them with 2550 * NULL's. Treat this as the last argument and not 2551 * a failure. 2552 */ 2553 if (base == 0) 2554 break; 2555 2556 while (!finished) { 2557 xlen = PAGE_SIZE - (base & PAGE_MASK); 2558 2559 aiov.iov_base = arg; 2560 aiov.iov_len = PAGE_SIZE; 2561 auio.uio_iov = &aiov; 2562 auio.uio_iovcnt = 1; 2563 auio.uio_offset = base; 2564 auio.uio_resid = xlen; 2565 auio.uio_rw = UIO_READ; 2566 UIO_SETUP_SYSSPACE(&auio); 2567 error = uvm_io(&vmspace->vm_map, &auio, 0); 2568 if (error) 2569 goto done; 2570 2571 /* Look for the end of the string */ 2572 for (j = 0; j < xlen; j++) { 2573 if (arg[j] == '\0') { 2574 xlen = j + 1; 2575 finished = 1; 2576 break; 2577 } 2578 } 2579 2580 /* Check for user buffer overflow */ 2581 if (len + xlen > *limit) { 2582 finished = 1; 2583 if (len > *limit) 2584 xlen = 0; 2585 else 2586 xlen = *limit - len; 2587 } 2588 2589 /* Copyout the page */ 2590 error = (*cb)(cookie, arg, len, xlen); 2591 if (error) 2592 goto done; 2593 2594 len += xlen; 2595 base += xlen; 2596 } 2597 } 2598 *limit = len; 2599 2600 done: 2601 kmem_free(argv, PAGE_SIZE); 2602 kmem_free(arg, PAGE_SIZE); 2603 uvmspace_free(vmspace); 2604 return error; 2605 } 2606 2607 /* 2608 * Fill in a proc structure for the specified process. 2609 */ 2610 static void 2611 fill_proc(const struct proc *psrc, struct proc *p, bool allowaddr) 2612 { 2613 COND_SET_STRUCT(p->p_list, psrc->p_list, allowaddr); 2614 memset(&p->p_auxlock, 0, sizeof(p->p_auxlock)); 2615 COND_SET_STRUCT(p->p_lock, psrc->p_lock, allowaddr); 2616 memset(&p->p_stmutex, 0, sizeof(p->p_stmutex)); 2617 memset(&p->p_reflock, 0, sizeof(p->p_reflock)); 2618 COND_SET_STRUCT(p->p_waitcv, psrc->p_waitcv, allowaddr); 2619 COND_SET_STRUCT(p->p_lwpcv, psrc->p_lwpcv, allowaddr); 2620 COND_SET_PTR(p->p_cred, psrc->p_cred, allowaddr); 2621 COND_SET_PTR(p->p_fd, psrc->p_fd, allowaddr); 2622 COND_SET_PTR(p->p_cwdi, psrc->p_cwdi, allowaddr); 2623 COND_SET_PTR(p->p_stats, psrc->p_stats, allowaddr); 2624 COND_SET_PTR(p->p_limit, psrc->p_limit, allowaddr); 2625 COND_SET_PTR(p->p_vmspace, psrc->p_vmspace, allowaddr); 2626 COND_SET_PTR(p->p_sigacts, psrc->p_sigacts, allowaddr); 2627 COND_SET_PTR(p->p_aio, psrc->p_aio, allowaddr); 2628 p->p_mqueue_cnt = psrc->p_mqueue_cnt; 2629 memset(&p->p_specdataref, 0, sizeof(p->p_specdataref)); 2630 p->p_exitsig = psrc->p_exitsig; 2631 p->p_flag = psrc->p_flag; 2632 p->p_sflag = psrc->p_sflag; 2633 p->p_slflag = psrc->p_slflag; 2634 p->p_lflag = psrc->p_lflag; 2635 p->p_stflag = psrc->p_stflag; 2636 p->p_stat = psrc->p_stat; 2637 p->p_trace_enabled = psrc->p_trace_enabled; 2638 p->p_pid = psrc->p_pid; 2639 COND_SET_STRUCT(p->p_pglist, psrc->p_pglist, allowaddr); 2640 COND_SET_PTR(p->p_pptr, psrc->p_pptr, allowaddr); 2641 COND_SET_STRUCT(p->p_sibling, psrc->p_sibling, allowaddr); 2642 COND_SET_STRUCT(p->p_children, psrc->p_children, allowaddr); 2643 COND_SET_STRUCT(p->p_lwps, psrc->p_lwps, allowaddr); 2644 COND_SET_PTR(p->p_raslist, psrc->p_raslist, allowaddr); 2645 p->p_nlwps = psrc->p_nlwps; 2646 p->p_nzlwps = psrc->p_nzlwps; 2647 p->p_nrlwps = psrc->p_nrlwps; 2648 p->p_nlwpwait = psrc->p_nlwpwait; 2649 p->p_ndlwps = psrc->p_ndlwps; 2650 p->p_nstopchild = psrc->p_nstopchild; 2651 p->p_waited = psrc->p_waited; 2652 COND_SET_PTR(p->p_zomblwp, psrc->p_zomblwp, allowaddr); 2653 COND_SET_PTR(p->p_vforklwp, psrc->p_vforklwp, allowaddr); 2654 COND_SET_PTR(p->p_sched_info, psrc->p_sched_info, allowaddr); 2655 p->p_estcpu = psrc->p_estcpu; 2656 p->p_estcpu_inherited = psrc->p_estcpu_inherited; 2657 p->p_forktime = psrc->p_forktime; 2658 p->p_pctcpu = psrc->p_pctcpu; 2659 COND_SET_PTR(p->p_opptr, psrc->p_opptr, allowaddr); 2660 COND_SET_PTR(p->p_timers, psrc->p_timers, allowaddr); 2661 p->p_rtime = psrc->p_rtime; 2662 p->p_uticks = psrc->p_uticks; 2663 p->p_sticks = psrc->p_sticks; 2664 p->p_iticks = psrc->p_iticks; 2665 p->p_xutime = psrc->p_xutime; 2666 p->p_xstime = psrc->p_xstime; 2667 p->p_traceflag = psrc->p_traceflag; 2668 COND_SET_PTR(p->p_tracep, psrc->p_tracep, allowaddr); 2669 COND_SET_PTR(p->p_textvp, psrc->p_textvp, allowaddr); 2670 COND_SET_PTR(p->p_emul, psrc->p_emul, allowaddr); 2671 COND_SET_PTR(p->p_emuldata, psrc->p_emuldata, allowaddr); 2672 COND_SET_CPTR(p->p_execsw, psrc->p_execsw, allowaddr); 2673 COND_SET_STRUCT(p->p_klist, psrc->p_klist, allowaddr); 2674 COND_SET_STRUCT(p->p_sigwaiters, psrc->p_sigwaiters, allowaddr); 2675 COND_SET_STRUCT(p->p_sigpend.sp_info, psrc->p_sigpend.sp_info, 2676 allowaddr); 2677 p->p_sigpend.sp_set = psrc->p_sigpend.sp_set; 2678 COND_SET_PTR(p->p_lwpctl, psrc->p_lwpctl, allowaddr); 2679 p->p_ppid = psrc->p_ppid; 2680 p->p_oppid = psrc->p_oppid; 2681 COND_SET_PTR(p->p_path, psrc->p_path, allowaddr); 2682 p->p_sigctx = psrc->p_sigctx; 2683 p->p_nice = psrc->p_nice; 2684 memcpy(p->p_comm, psrc->p_comm, sizeof(p->p_comm)); 2685 COND_SET_PTR(p->p_pgrp, psrc->p_pgrp, allowaddr); 2686 COND_SET_VALUE(p->p_psstrp, psrc->p_psstrp, allowaddr); 2687 p->p_pax = psrc->p_pax; 2688 p->p_xexit = psrc->p_xexit; 2689 p->p_xsig = psrc->p_xsig; 2690 p->p_acflag = psrc->p_acflag; 2691 COND_SET_STRUCT(p->p_md, psrc->p_md, allowaddr); 2692 p->p_stackbase = psrc->p_stackbase; 2693 COND_SET_PTR(p->p_dtrace, psrc->p_dtrace, allowaddr); 2694 } 2695 2696 /* 2697 * Fill in an eproc structure for the specified process. 2698 */ 2699 void 2700 fill_eproc(struct proc *p, struct eproc *ep, bool zombie, bool allowaddr) 2701 { 2702 struct tty *tp; 2703 struct lwp *l; 2704 2705 KASSERT(mutex_owned(&proc_lock)); 2706 KASSERT(mutex_owned(p->p_lock)); 2707 2708 COND_SET_PTR(ep->e_paddr, p, allowaddr); 2709 COND_SET_PTR(ep->e_sess, p->p_session, allowaddr); 2710 if (p->p_cred) { 2711 kauth_cred_topcred(p->p_cred, &ep->e_pcred); 2712 kauth_cred_toucred(p->p_cred, &ep->e_ucred); 2713 } 2714 if (p->p_stat != SIDL && !P_ZOMBIE(p) && !zombie) { 2715 struct vmspace *vm = p->p_vmspace; 2716 2717 ep->e_vm.vm_rssize = vm_resident_count(vm); 2718 ep->e_vm.vm_tsize = vm->vm_tsize; 2719 ep->e_vm.vm_dsize = vm->vm_dsize; 2720 ep->e_vm.vm_ssize = vm->vm_ssize; 2721 ep->e_vm.vm_map.size = vm->vm_map.size; 2722 2723 /* Pick the primary (first) LWP */ 2724 l = proc_active_lwp(p); 2725 KASSERT(l != NULL); 2726 lwp_lock(l); 2727 if (l->l_wchan) 2728 strncpy(ep->e_wmesg, l->l_wmesg, WMESGLEN); 2729 lwp_unlock(l); 2730 } 2731 ep->e_ppid = p->p_ppid; 2732 if (p->p_pgrp && p->p_session) { 2733 ep->e_pgid = p->p_pgrp->pg_id; 2734 ep->e_jobc = p->p_pgrp->pg_jobc; 2735 ep->e_sid = p->p_session->s_sid; 2736 if ((p->p_lflag & PL_CONTROLT) && 2737 (tp = p->p_session->s_ttyp)) { 2738 ep->e_tdev = tp->t_dev; 2739 ep->e_tpgid = tp->t_pgrp ? tp->t_pgrp->pg_id : NO_PGID; 2740 COND_SET_PTR(ep->e_tsess, tp->t_session, allowaddr); 2741 } else 2742 ep->e_tdev = (uint32_t)NODEV; 2743 ep->e_flag = p->p_session->s_ttyvp ? EPROC_CTTY : 0; 2744 if (SESS_LEADER(p)) 2745 ep->e_flag |= EPROC_SLEADER; 2746 strncpy(ep->e_login, p->p_session->s_login, MAXLOGNAME); 2747 } 2748 ep->e_xsize = ep->e_xrssize = 0; 2749 ep->e_xccount = ep->e_xswrss = 0; 2750 } 2751 2752 /* 2753 * Fill in a kinfo_proc2 structure for the specified process. 2754 */ 2755 void 2756 fill_kproc2(struct proc *p, struct kinfo_proc2 *ki, bool zombie, bool allowaddr) 2757 { 2758 struct tty *tp; 2759 struct lwp *l, *l2; 2760 struct timeval ut, st, rt; 2761 sigset_t ss1, ss2; 2762 struct rusage ru; 2763 struct vmspace *vm; 2764 2765 KASSERT(mutex_owned(&proc_lock)); 2766 KASSERT(mutex_owned(p->p_lock)); 2767 2768 sigemptyset(&ss1); 2769 sigemptyset(&ss2); 2770 2771 COND_SET_VALUE(ki->p_paddr, PTRTOUINT64(p), allowaddr); 2772 COND_SET_VALUE(ki->p_fd, PTRTOUINT64(p->p_fd), allowaddr); 2773 COND_SET_VALUE(ki->p_cwdi, PTRTOUINT64(p->p_cwdi), allowaddr); 2774 COND_SET_VALUE(ki->p_stats, PTRTOUINT64(p->p_stats), allowaddr); 2775 COND_SET_VALUE(ki->p_limit, PTRTOUINT64(p->p_limit), allowaddr); 2776 COND_SET_VALUE(ki->p_vmspace, PTRTOUINT64(p->p_vmspace), allowaddr); 2777 COND_SET_VALUE(ki->p_sigacts, PTRTOUINT64(p->p_sigacts), allowaddr); 2778 COND_SET_VALUE(ki->p_sess, PTRTOUINT64(p->p_session), allowaddr); 2779 ki->p_tsess = 0; /* may be changed if controlling tty below */ 2780 COND_SET_VALUE(ki->p_ru, PTRTOUINT64(&p->p_stats->p_ru), allowaddr); 2781 ki->p_eflag = 0; 2782 ki->p_exitsig = p->p_exitsig; 2783 ki->p_flag = L_INMEM; /* Process never swapped out */ 2784 ki->p_flag |= sysctl_map_flags(sysctl_flagmap, p->p_flag); 2785 ki->p_flag |= sysctl_map_flags(sysctl_sflagmap, p->p_sflag); 2786 ki->p_flag |= sysctl_map_flags(sysctl_slflagmap, p->p_slflag); 2787 ki->p_flag |= sysctl_map_flags(sysctl_lflagmap, p->p_lflag); 2788 ki->p_flag |= sysctl_map_flags(sysctl_stflagmap, p->p_stflag); 2789 ki->p_pid = p->p_pid; 2790 ki->p_ppid = p->p_ppid; 2791 ki->p_uid = kauth_cred_geteuid(p->p_cred); 2792 ki->p_ruid = kauth_cred_getuid(p->p_cred); 2793 ki->p_gid = kauth_cred_getegid(p->p_cred); 2794 ki->p_rgid = kauth_cred_getgid(p->p_cred); 2795 ki->p_svuid = kauth_cred_getsvuid(p->p_cred); 2796 ki->p_svgid = kauth_cred_getsvgid(p->p_cred); 2797 ki->p_ngroups = kauth_cred_ngroups(p->p_cred); 2798 kauth_cred_getgroups(p->p_cred, ki->p_groups, 2799 uimin(ki->p_ngroups, sizeof(ki->p_groups) / sizeof(ki->p_groups[0])), 2800 UIO_SYSSPACE); 2801 2802 ki->p_uticks = p->p_uticks; 2803 ki->p_sticks = p->p_sticks; 2804 ki->p_iticks = p->p_iticks; 2805 ki->p_tpgid = NO_PGID; /* may be changed if controlling tty below */ 2806 COND_SET_VALUE(ki->p_tracep, PTRTOUINT64(p->p_tracep), allowaddr); 2807 ki->p_traceflag = p->p_traceflag; 2808 2809 memcpy(&ki->p_sigignore, &p->p_sigctx.ps_sigignore,sizeof(ki_sigset_t)); 2810 memcpy(&ki->p_sigcatch, &p->p_sigctx.ps_sigcatch, sizeof(ki_sigset_t)); 2811 2812 ki->p_cpticks = 0; 2813 ki->p_pctcpu = p->p_pctcpu; 2814 ki->p_estcpu = 0; 2815 ki->p_stat = p->p_stat; /* Will likely be overridden by LWP status */ 2816 ki->p_realstat = p->p_stat; 2817 ki->p_nice = p->p_nice; 2818 ki->p_xstat = P_WAITSTATUS(p); 2819 ki->p_acflag = p->p_acflag; 2820 2821 strncpy(ki->p_comm, p->p_comm, 2822 uimin(sizeof(ki->p_comm), sizeof(p->p_comm))); 2823 strncpy(ki->p_ename, p->p_emul->e_name, sizeof(ki->p_ename)); 2824 2825 ki->p_nlwps = p->p_nlwps; 2826 ki->p_realflag = ki->p_flag; 2827 2828 if (p->p_stat != SIDL && !P_ZOMBIE(p) && !zombie) { 2829 vm = p->p_vmspace; 2830 ki->p_vm_rssize = vm_resident_count(vm); 2831 ki->p_vm_tsize = vm->vm_tsize; 2832 ki->p_vm_dsize = vm->vm_dsize; 2833 ki->p_vm_ssize = vm->vm_ssize; 2834 ki->p_vm_vsize = atop(vm->vm_map.size); 2835 /* 2836 * Since the stack is initially mapped mostly with 2837 * PROT_NONE and grown as needed, adjust the "mapped size" 2838 * to skip the unused stack portion. 2839 */ 2840 ki->p_vm_msize = 2841 atop(vm->vm_map.size) - vm->vm_issize + vm->vm_ssize; 2842 2843 /* Pick the primary (first) LWP */ 2844 l = proc_active_lwp(p); 2845 KASSERT(l != NULL); 2846 lwp_lock(l); 2847 ki->p_nrlwps = p->p_nrlwps; 2848 ki->p_forw = 0; 2849 ki->p_back = 0; 2850 COND_SET_VALUE(ki->p_addr, PTRTOUINT64(l->l_addr), allowaddr); 2851 ki->p_stat = l->l_stat; 2852 ki->p_flag |= sysctl_map_flags(sysctl_lwpflagmap, l->l_flag); 2853 ki->p_swtime = l->l_swtime; 2854 ki->p_slptime = l->l_slptime; 2855 if (l->l_stat == LSONPROC) 2856 ki->p_schedflags = l->l_cpu->ci_schedstate.spc_flags; 2857 else 2858 ki->p_schedflags = 0; 2859 ki->p_priority = lwp_eprio(l); 2860 ki->p_usrpri = l->l_priority; 2861 if (l->l_wchan) 2862 strncpy(ki->p_wmesg, l->l_wmesg, sizeof(ki->p_wmesg)); 2863 COND_SET_VALUE(ki->p_wchan, PTRTOUINT64(l->l_wchan), allowaddr); 2864 ki->p_cpuid = cpu_index(l->l_cpu); 2865 lwp_unlock(l); 2866 LIST_FOREACH(l, &p->p_lwps, l_sibling) { 2867 /* This is hardly correct, but... */ 2868 sigplusset(&l->l_sigpend.sp_set, &ss1); 2869 sigplusset(&l->l_sigmask, &ss2); 2870 ki->p_cpticks += l->l_cpticks; 2871 ki->p_pctcpu += l->l_pctcpu; 2872 ki->p_estcpu += l->l_estcpu; 2873 } 2874 } 2875 sigplusset(&p->p_sigpend.sp_set, &ss1); 2876 memcpy(&ki->p_siglist, &ss1, sizeof(ki_sigset_t)); 2877 memcpy(&ki->p_sigmask, &ss2, sizeof(ki_sigset_t)); 2878 2879 if (p->p_session != NULL) { 2880 ki->p_sid = p->p_session->s_sid; 2881 ki->p__pgid = p->p_pgrp->pg_id; 2882 if (p->p_session->s_ttyvp) 2883 ki->p_eflag |= EPROC_CTTY; 2884 if (SESS_LEADER(p)) 2885 ki->p_eflag |= EPROC_SLEADER; 2886 strncpy(ki->p_login, p->p_session->s_login, 2887 uimin(sizeof ki->p_login - 1, sizeof p->p_session->s_login)); 2888 ki->p_jobc = p->p_pgrp->pg_jobc; 2889 if ((p->p_lflag & PL_CONTROLT) && (tp = p->p_session->s_ttyp)) { 2890 ki->p_tdev = tp->t_dev; 2891 ki->p_tpgid = tp->t_pgrp ? tp->t_pgrp->pg_id : NO_PGID; 2892 COND_SET_VALUE(ki->p_tsess, PTRTOUINT64(tp->t_session), 2893 allowaddr); 2894 } else { 2895 ki->p_tdev = (int32_t)NODEV; 2896 } 2897 } 2898 2899 if (!P_ZOMBIE(p) && !zombie) { 2900 ki->p_uvalid = 1; 2901 ki->p_ustart_sec = p->p_stats->p_start.tv_sec; 2902 ki->p_ustart_usec = p->p_stats->p_start.tv_usec; 2903 2904 calcru(p, &ut, &st, NULL, &rt); 2905 ki->p_rtime_sec = rt.tv_sec; 2906 ki->p_rtime_usec = rt.tv_usec; 2907 ki->p_uutime_sec = ut.tv_sec; 2908 ki->p_uutime_usec = ut.tv_usec; 2909 ki->p_ustime_sec = st.tv_sec; 2910 ki->p_ustime_usec = st.tv_usec; 2911 2912 memcpy(&ru, &p->p_stats->p_ru, sizeof(ru)); 2913 ki->p_uru_nvcsw = 0; 2914 ki->p_uru_nivcsw = 0; 2915 LIST_FOREACH(l2, &p->p_lwps, l_sibling) { 2916 ki->p_uru_nvcsw += (l2->l_ncsw - l2->l_nivcsw); 2917 ki->p_uru_nivcsw += l2->l_nivcsw; 2918 ruadd(&ru, &l2->l_ru); 2919 } 2920 ki->p_uru_maxrss = ru.ru_maxrss; 2921 ki->p_uru_ixrss = ru.ru_ixrss; 2922 ki->p_uru_idrss = ru.ru_idrss; 2923 ki->p_uru_isrss = ru.ru_isrss; 2924 ki->p_uru_minflt = ru.ru_minflt; 2925 ki->p_uru_majflt = ru.ru_majflt; 2926 ki->p_uru_nswap = ru.ru_nswap; 2927 ki->p_uru_inblock = ru.ru_inblock; 2928 ki->p_uru_oublock = ru.ru_oublock; 2929 ki->p_uru_msgsnd = ru.ru_msgsnd; 2930 ki->p_uru_msgrcv = ru.ru_msgrcv; 2931 ki->p_uru_nsignals = ru.ru_nsignals; 2932 2933 timeradd(&p->p_stats->p_cru.ru_utime, 2934 &p->p_stats->p_cru.ru_stime, &ut); 2935 ki->p_uctime_sec = ut.tv_sec; 2936 ki->p_uctime_usec = ut.tv_usec; 2937 } 2938 } 2939 2940 2941 int 2942 proc_find_locked(struct lwp *l, struct proc **p, pid_t pid) 2943 { 2944 int error; 2945 2946 mutex_enter(&proc_lock); 2947 if (pid == -1) 2948 *p = l->l_proc; 2949 else 2950 *p = proc_find(pid); 2951 2952 if (*p == NULL) { 2953 if (pid != -1) 2954 mutex_exit(&proc_lock); 2955 return ESRCH; 2956 } 2957 if (pid != -1) 2958 mutex_enter((*p)->p_lock); 2959 mutex_exit(&proc_lock); 2960 2961 error = kauth_authorize_process(l->l_cred, 2962 KAUTH_PROCESS_CANSEE, *p, 2963 KAUTH_ARG(KAUTH_REQ_PROCESS_CANSEE_ENTRY), NULL, NULL); 2964 if (error) { 2965 if (pid != -1) 2966 mutex_exit((*p)->p_lock); 2967 } 2968 return error; 2969 } 2970 2971 static int 2972 fill_pathname(struct lwp *l, pid_t pid, void *oldp, size_t *oldlenp) 2973 { 2974 int error; 2975 struct proc *p; 2976 2977 if ((error = proc_find_locked(l, &p, pid)) != 0) 2978 return error; 2979 2980 if (p->p_path == NULL) { 2981 if (pid != -1) 2982 mutex_exit(p->p_lock); 2983 return ENOENT; 2984 } 2985 2986 size_t len = strlen(p->p_path) + 1; 2987 if (oldp != NULL) { 2988 size_t copylen = uimin(len, *oldlenp); 2989 error = sysctl_copyout(l, p->p_path, oldp, copylen); 2990 if (error == 0 && *oldlenp < len) 2991 error = ENOSPC; 2992 } 2993 *oldlenp = len; 2994 if (pid != -1) 2995 mutex_exit(p->p_lock); 2996 return error; 2997 } 2998 2999 static int 3000 fill_cwd(struct lwp *l, pid_t pid, void *oldp, size_t *oldlenp) 3001 { 3002 int error; 3003 struct proc *p; 3004 char *path; 3005 char *bp, *bend; 3006 struct cwdinfo *cwdi; 3007 struct vnode *vp; 3008 size_t len, lenused; 3009 3010 if ((error = proc_find_locked(l, &p, pid)) != 0) 3011 return error; 3012 3013 len = MAXPATHLEN * 4; 3014 3015 path = kmem_alloc(len, KM_SLEEP); 3016 3017 bp = &path[len]; 3018 bend = bp; 3019 *(--bp) = '\0'; 3020 3021 cwdi = p->p_cwdi; 3022 rw_enter(&cwdi->cwdi_lock, RW_READER); 3023 vp = cwdi->cwdi_cdir; 3024 error = getcwd_common(vp, NULL, &bp, path, len/2, 0, l); 3025 rw_exit(&cwdi->cwdi_lock); 3026 3027 if (error) 3028 goto out; 3029 3030 lenused = bend - bp; 3031 3032 if (oldp != NULL) { 3033 size_t copylen = uimin(lenused, *oldlenp); 3034 error = sysctl_copyout(l, bp, oldp, copylen); 3035 if (error == 0 && *oldlenp < lenused) 3036 error = ENOSPC; 3037 } 3038 *oldlenp = lenused; 3039 out: 3040 if (pid != -1) 3041 mutex_exit(p->p_lock); 3042 kmem_free(path, len); 3043 return error; 3044 } 3045 3046 int 3047 proc_getauxv(struct proc *p, void **buf, size_t *len) 3048 { 3049 struct ps_strings pss; 3050 int error; 3051 void *uauxv, *kauxv; 3052 size_t size; 3053 3054 if ((error = copyin_psstrings(p, &pss)) != 0) 3055 return error; 3056 if (pss.ps_envstr == NULL) 3057 return EIO; 3058 3059 size = p->p_execsw->es_arglen; 3060 if (size == 0) 3061 return EIO; 3062 3063 size_t ptrsz = PROC_PTRSZ(p); 3064 uauxv = (void *)((char *)pss.ps_envstr + (pss.ps_nenvstr + 1) * ptrsz); 3065 3066 kauxv = kmem_alloc(size, KM_SLEEP); 3067 3068 error = copyin_proc(p, uauxv, kauxv, size); 3069 if (error) { 3070 kmem_free(kauxv, size); 3071 return error; 3072 } 3073 3074 *buf = kauxv; 3075 *len = size; 3076 3077 return 0; 3078 } 3079 3080 3081 static int 3082 sysctl_security_expose_address(SYSCTLFN_ARGS) 3083 { 3084 int expose_address, error; 3085 struct sysctlnode node; 3086 3087 node = *rnode; 3088 node.sysctl_data = &expose_address; 3089 expose_address = *(int *)rnode->sysctl_data; 3090 error = sysctl_lookup(SYSCTLFN_CALL(&node)); 3091 if (error || newp == NULL) 3092 return error; 3093 3094 if (kauth_authorize_system(l->l_cred, KAUTH_SYSTEM_KERNADDR, 3095 0, NULL, NULL, NULL)) 3096 return EPERM; 3097 3098 switch (expose_address) { 3099 case 0: 3100 case 1: 3101 case 2: 3102 break; 3103 default: 3104 return EINVAL; 3105 } 3106 3107 *(int *)rnode->sysctl_data = expose_address; 3108 3109 return 0; 3110 } 3111 3112 bool 3113 get_expose_address(struct proc *p) 3114 { 3115 /* allow only if sysctl variable is set or privileged */ 3116 return kauth_authorize_process(kauth_cred_get(), KAUTH_PROCESS_CANSEE, 3117 p, KAUTH_ARG(KAUTH_REQ_PROCESS_CANSEE_KPTR), NULL, NULL) == 0; 3118 } 3119