1 /* $OpenBSD: kern_fork.c,v 1.14 1999/02/19 19:21:42 art Exp $ */ 2 /* $NetBSD: kern_fork.c,v 1.29 1996/02/09 18:59:34 christos Exp $ */ 3 4 /* 5 * Copyright (c) 1982, 1986, 1989, 1991, 1993 6 * The Regents of the University of California. All rights reserved. 7 * (c) UNIX System Laboratories, Inc. 8 * All or some portions of this file are derived from material licensed 9 * to the University of California by American Telephone and Telegraph 10 * Co. or Unix System Laboratories, Inc. and are reproduced herein with 11 * the permission of UNIX System Laboratories, Inc. 12 * 13 * Redistribution and use in source and binary forms, with or without 14 * modification, are permitted provided that the following conditions 15 * are met: 16 * 1. Redistributions of source code must retain the above copyright 17 * notice, this list of conditions and the following disclaimer. 18 * 2. Redistributions in binary form must reproduce the above copyright 19 * notice, this list of conditions and the following disclaimer in the 20 * documentation and/or other materials provided with the distribution. 21 * 3. All advertising materials mentioning features or use of this software 22 * must display the following acknowledgement: 23 * This product includes software developed by the University of 24 * California, Berkeley and its contributors. 25 * 4. Neither the name of the University nor the names of its contributors 26 * may be used to endorse or promote products derived from this software 27 * without specific prior written permission. 28 * 29 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 30 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 31 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 32 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 33 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 34 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 35 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 36 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 37 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 38 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 39 * SUCH DAMAGE. 40 * 41 * @(#)kern_fork.c 8.6 (Berkeley) 4/8/94 42 */ 43 44 #include <sys/param.h> 45 #include <sys/systm.h> 46 #include <sys/map.h> 47 #include <sys/filedesc.h> 48 #include <sys/kernel.h> 49 #include <sys/malloc.h> 50 #include <sys/mount.h> 51 #include <sys/proc.h> 52 #include <sys/resourcevar.h> 53 #include <sys/vnode.h> 54 #include <sys/file.h> 55 #include <sys/acct.h> 56 #include <sys/ktrace.h> 57 #include <dev/rndvar.h> 58 59 #include <sys/syscallargs.h> 60 61 #include <vm/vm.h> 62 #include <vm/vm_kern.h> 63 64 int nprocs = 1; /* process 0 */ 65 int randompid; /* when set to 1, pid's go random */ 66 pid_t lastpid; 67 68 /*ARGSUSED*/ 69 int 70 sys_fork(p, v, retval) 71 struct proc *p; 72 void *v; 73 register_t *retval; 74 { 75 return (fork1(p, ISFORK, 0, retval)); 76 } 77 78 /*ARGSUSED*/ 79 int 80 sys_vfork(p, v, retval) 81 struct proc *p; 82 void *v; 83 register_t *retval; 84 { 85 return (fork1(p, ISVFORK, 0, retval)); 86 } 87 88 int 89 sys_rfork(p, v, retval) 90 struct proc *p; 91 void *v; 92 register_t *retval; 93 { 94 struct sys_rfork_args /* { 95 syscallarg(int) flags; 96 } */ *uap = v; 97 98 return (fork1(p, ISRFORK, SCARG(uap, flags), retval)); 99 } 100 101 int 102 fork1(p1, forktype, rforkflags, retval) 103 register struct proc *p1; 104 int forktype; 105 int rforkflags; 106 register_t *retval; 107 { 108 register struct proc *p2; 109 register uid_t uid; 110 struct proc *newproc; 111 struct vmspace *vm; 112 int count; 113 static int pidchecked = 0; 114 int dupfd = 1, cleanfd = 0; 115 vm_offset_t uaddr; 116 117 /* 118 * Allocate a pcb and kernel stack for the process 119 */ 120 #if defined(arc) || defined(mips_cachealias) 121 uaddr = kmem_alloc_upage(kernel_map, USPACE); 122 #else 123 uaddr = kmem_alloc_pageable(kernel_map, USPACE); 124 #endif 125 if (uaddr == 0) 126 return ENOMEM; 127 128 if (forktype == ISRFORK) { 129 dupfd = 0; 130 if ((rforkflags & RFPROC) == 0) 131 return (EINVAL); 132 if ((rforkflags & (RFFDG|RFCFDG)) == (RFFDG|RFCFDG)) 133 return (EINVAL); 134 if (rforkflags & RFFDG) 135 dupfd = 1; 136 if (rforkflags & RFCFDG) 137 cleanfd = 1; 138 } 139 140 /* 141 * Although process entries are dynamically created, we still keep 142 * a global limit on the maximum number we will create. Don't allow 143 * a nonprivileged user to use the last process; don't let root 144 * exceed the limit. The variable nprocs is the current number of 145 * processes, maxproc is the limit. 146 */ 147 uid = p1->p_cred->p_ruid; 148 if ((nprocs >= maxproc - 1 && uid != 0) || nprocs >= maxproc) { 149 tablefull("proc"); 150 return (EAGAIN); 151 } 152 153 /* 154 * Increment the count of procs running with this uid. Don't allow 155 * a nonprivileged user to exceed their current limit. 156 */ 157 count = chgproccnt(uid, 1); 158 if (uid != 0 && count > p1->p_rlimit[RLIMIT_NPROC].rlim_cur) { 159 (void)chgproccnt(uid, -1); 160 return (EAGAIN); 161 } 162 163 /* Allocate new proc. */ 164 MALLOC(newproc, struct proc *, sizeof(struct proc), M_PROC, M_WAITOK); 165 166 lastpid++; 167 if (randompid) 168 lastpid = PID_MAX; 169 retry: 170 /* 171 * If the process ID prototype has wrapped around, 172 * restart somewhat above 0, as the low-numbered procs 173 * tend to include daemons that don't exit. 174 */ 175 if (lastpid >= PID_MAX) { 176 lastpid = arc4random() % PID_MAX; 177 pidchecked = 0; 178 } 179 if (lastpid >= pidchecked) { 180 int doingzomb = 0; 181 182 pidchecked = PID_MAX; 183 /* 184 * Scan the active and zombie procs to check whether this pid 185 * is in use. Remember the lowest pid that's greater 186 * than lastpid, so we can avoid checking for a while. 187 */ 188 p2 = allproc.lh_first; 189 again: 190 for (; p2 != 0; p2 = p2->p_list.le_next) { 191 while (p2->p_pid == lastpid || 192 p2->p_pgrp->pg_id == lastpid) { 193 lastpid++; 194 if (lastpid >= pidchecked) 195 goto retry; 196 } 197 if (p2->p_pid > lastpid && pidchecked > p2->p_pid) 198 pidchecked = p2->p_pid; 199 if (p2->p_pgrp->pg_id > lastpid && 200 pidchecked > p2->p_pgrp->pg_id) 201 pidchecked = p2->p_pgrp->pg_id; 202 } 203 if (!doingzomb) { 204 doingzomb = 1; 205 p2 = zombproc.lh_first; 206 goto again; 207 } 208 } 209 210 nprocs++; 211 p2 = newproc; 212 p2->p_stat = SIDL; /* protect against others */ 213 p2->p_pid = lastpid; 214 LIST_INSERT_HEAD(&allproc, p2, p_list); 215 p2->p_forw = p2->p_back = NULL; /* shouldn't be necessary */ 216 LIST_INSERT_HEAD(PIDHASH(p2->p_pid), p2, p_hash); 217 218 /* 219 * Make a proc table entry for the new process. 220 * Start by zeroing the section of proc that is zero-initialized, 221 * then copy the section that is copied directly from the parent. 222 */ 223 bzero(&p2->p_startzero, 224 (unsigned) ((caddr_t)&p2->p_endzero - (caddr_t)&p2->p_startzero)); 225 bcopy(&p1->p_startcopy, &p2->p_startcopy, 226 (unsigned) ((caddr_t)&p2->p_endcopy - (caddr_t)&p2->p_startcopy)); 227 228 /* 229 * Duplicate sub-structures as needed. 230 * Increase reference counts on shared objects. 231 * The p_stats and p_sigacts substructs are set in vm_fork. 232 */ 233 p2->p_flag = P_INMEM; 234 p2->p_emul = p1->p_emul; 235 if (p1->p_flag & P_PROFIL) 236 startprofclock(p2); 237 p2->p_flag |= (p1->p_flag & (P_SUGID | P_SUGIDEXEC)); 238 MALLOC(p2->p_cred, struct pcred *, sizeof(struct pcred), 239 M_SUBPROC, M_WAITOK); 240 bcopy(p1->p_cred, p2->p_cred, sizeof(*p2->p_cred)); 241 p2->p_cred->p_refcnt = 1; 242 crhold(p1->p_ucred); 243 244 /* bump references to the text vnode (for procfs) */ 245 p2->p_textvp = p1->p_textvp; 246 if (p2->p_textvp) 247 VREF(p2->p_textvp); 248 249 if (cleanfd) 250 p2->p_fd = fdinit(p1); 251 else if (dupfd) 252 p2->p_fd = fdcopy(p1); 253 else 254 p2->p_fd = fdshare(p1); 255 256 /* 257 * If p_limit is still copy-on-write, bump refcnt, 258 * otherwise get a copy that won't be modified. 259 * (If PL_SHAREMOD is clear, the structure is shared 260 * copy-on-write.) 261 */ 262 if (p1->p_limit->p_lflags & PL_SHAREMOD) 263 p2->p_limit = limcopy(p1->p_limit); 264 else { 265 p2->p_limit = p1->p_limit; 266 p2->p_limit->p_refcnt++; 267 } 268 269 if (p1->p_session->s_ttyvp != NULL && p1->p_flag & P_CONTROLT) 270 p2->p_flag |= P_CONTROLT; 271 if (forktype == ISVFORK) 272 p2->p_flag |= P_PPWAIT; 273 LIST_INSERT_AFTER(p1, p2, p_pglist); 274 p2->p_pptr = p1; 275 if (forktype == ISRFORK && (rforkflags & RFNOWAIT)) { 276 p2->p_flag |= P_NOZOMBIE; 277 } else { 278 LIST_INSERT_HEAD(&p1->p_children, p2, p_sibling); 279 } 280 LIST_INIT(&p2->p_children); 281 282 #ifdef KTRACE 283 /* 284 * Copy traceflag and tracefile if enabled. 285 * If not inherited, these were zeroed above. 286 */ 287 if (p1->p_traceflag&KTRFAC_INHERIT) { 288 p2->p_traceflag = p1->p_traceflag; 289 if ((p2->p_tracep = p1->p_tracep) != NULL) 290 VREF(p2->p_tracep); 291 } 292 #endif 293 294 /* 295 * This begins the section where we must prevent the parent 296 * from being swapped. 297 */ 298 p1->p_holdcnt++; 299 300 if (forktype == ISRFORK && (rforkflags & RFMEM)) { 301 /* share as much address space as possible */ 302 (void) vm_map_inherit(&p1->p_vmspace->vm_map, 303 VM_MIN_ADDRESS, VM_MAXUSER_ADDRESS - MAXSSIZ, 304 VM_INHERIT_SHARE); 305 } 306 307 p2->p_addr = (struct user *)uaddr; 308 309 #ifdef __FORK_BRAINDAMAGE 310 /* 311 * Set return values for child before vm_fork, 312 * so they can be copied to child stack. 313 * We return 0, rather than the traditional behaviour of modifying the 314 * return value in the system call stub. 315 * NOTE: the kernel stack may be at a different location in the child 316 * process, and thus addresses of automatic variables (including retval) 317 * may be invalid after vm_fork returns in the child process. 318 */ 319 retval[0] = 0; 320 retval[1] = 1; 321 if (vm_fork(p1, p2)) 322 return (0); 323 #else 324 /* 325 * Finish creating the child process. It will return through a 326 * different path later. 327 */ 328 vm_fork(p1, p2); 329 #endif 330 vm = p2->p_vmspace; 331 332 switch (forktype) { 333 case ISFORK: 334 forkstat.cntfork++; 335 forkstat.sizfork += vm->vm_dsize + vm->vm_ssize; 336 break; 337 case ISVFORK: 338 forkstat.cntvfork++; 339 forkstat.sizvfork += vm->vm_dsize + vm->vm_ssize; 340 break; 341 case ISRFORK: 342 forkstat.cntrfork++; 343 forkstat.sizrfork += vm->vm_dsize + vm->vm_ssize; 344 break; 345 } 346 347 /* 348 * Make child runnable, set start time, and add to run queue. 349 */ 350 (void) splstatclock(); 351 p2->p_stats->p_start = time; 352 p2->p_acflag = AFORK; 353 p2->p_stat = SRUN; 354 setrunqueue(p2); 355 (void) spl0(); 356 357 /* 358 * Now can be swapped. 359 */ 360 p1->p_holdcnt--; 361 362 /* 363 * Preserve synchronization semantics of vfork. If waiting for 364 * child to exec or exit, set P_PPWAIT on child, and sleep on our 365 * proc (in case of exit). 366 */ 367 if (forktype == ISVFORK) 368 while (p2->p_flag & P_PPWAIT) 369 tsleep(p1, PWAIT, "ppwait", 0); 370 371 /* 372 * Return child pid to parent process, 373 * marking us as parent via retval[1]. 374 */ 375 retval[0] = p2->p_pid; 376 retval[1] = 0; 377 return (0); 378 } 379