1 /* $OpenBSD: kern_fork.c,v 1.8 1997/06/05 10:15:26 deraadt 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 63 int nprocs = 1; /* process 0 */ 64 int randompid; /* when set to 1, pid's go random */ 65 pid_t lastpid; 66 67 #define ISFORK 0 68 #define ISVFORK 1 69 #define ISRFORK 2 70 71 int fork1 __P((struct proc *, int, int, register_t *)); 72 73 /*ARGSUSED*/ 74 int 75 sys_fork(p, v, retval) 76 struct proc *p; 77 void *v; 78 register_t *retval; 79 { 80 return (fork1(p, ISFORK, 0, retval)); 81 } 82 83 /*ARGSUSED*/ 84 int 85 sys_vfork(p, v, retval) 86 struct proc *p; 87 void *v; 88 register_t *retval; 89 { 90 return (fork1(p, ISVFORK, 0, retval)); 91 } 92 93 int 94 sys_rfork(p, v, retval) 95 struct proc *p; 96 void *v; 97 register_t *retval; 98 { 99 struct sys_rfork_args /* { 100 syscallarg(int) flags; 101 } */ *uap = v; 102 103 return (fork1(p, ISRFORK, SCARG(uap, flags), retval)); 104 } 105 106 int 107 fork1(p1, forktype, rforkflags, retval) 108 register struct proc *p1; 109 int forktype; 110 int rforkflags; 111 register_t *retval; 112 { 113 register struct proc *p2; 114 register uid_t uid; 115 struct proc *newproc; 116 int count; 117 static int pidchecked = 0; 118 int dupfd = 1, cleanfd = 0; 119 120 if (forktype == ISRFORK) { 121 dupfd = 0; 122 if ((rforkflags & RFPROC) == 0) 123 return (EINVAL); 124 if ((rforkflags & (RFFDG|RFCFDG)) == (RFFDG|RFCFDG)) 125 return (EINVAL); 126 if (rforkflags & RFFDG) 127 dupfd = 1; 128 if (rforkflags & RFNOWAIT) 129 return (EINVAL); /* XXX unimplimented */ 130 if (rforkflags & RFCFDG) 131 cleanfd = 1; 132 } 133 134 /* 135 * Although process entries are dynamically created, we still keep 136 * a global limit on the maximum number we will create. Don't allow 137 * a nonprivileged user to use the last process; don't let root 138 * exceed the limit. The variable nprocs is the current number of 139 * processes, maxproc is the limit. 140 */ 141 uid = p1->p_cred->p_ruid; 142 if ((nprocs >= maxproc - 1 && uid != 0) || nprocs >= maxproc) { 143 tablefull("proc"); 144 return (EAGAIN); 145 } 146 147 /* 148 * Increment the count of procs running with this uid. Don't allow 149 * a nonprivileged user to exceed their current limit. 150 */ 151 count = chgproccnt(uid, 1); 152 if (uid != 0 && count > p1->p_rlimit[RLIMIT_NPROC].rlim_cur) { 153 (void)chgproccnt(uid, -1); 154 return (EAGAIN); 155 } 156 157 /* Allocate new proc. */ 158 MALLOC(newproc, struct proc *, sizeof(struct proc), M_PROC, M_WAITOK); 159 160 lastpid++; 161 if (randompid) 162 lastpid = PID_MAX; 163 retry: 164 /* 165 * If the process ID prototype has wrapped around, 166 * restart somewhat above 0, as the low-numbered procs 167 * tend to include daemons that don't exit. 168 */ 169 if (lastpid >= PID_MAX) { 170 lastpid = arc4random() % PID_MAX; 171 pidchecked = 0; 172 } 173 if (lastpid >= pidchecked) { 174 int doingzomb = 0; 175 176 pidchecked = PID_MAX; 177 /* 178 * Scan the active and zombie procs to check whether this pid 179 * is in use. Remember the lowest pid that's greater 180 * than lastpid, so we can avoid checking for a while. 181 */ 182 p2 = allproc.lh_first; 183 again: 184 for (; p2 != 0; p2 = p2->p_list.le_next) { 185 while (p2->p_pid == lastpid || 186 p2->p_pgrp->pg_id == lastpid) { 187 lastpid++; 188 if (lastpid >= pidchecked) 189 goto retry; 190 } 191 if (p2->p_pid > lastpid && pidchecked > p2->p_pid) 192 pidchecked = p2->p_pid; 193 if (p2->p_pgrp->pg_id > lastpid && 194 pidchecked > p2->p_pgrp->pg_id) 195 pidchecked = p2->p_pgrp->pg_id; 196 } 197 if (!doingzomb) { 198 doingzomb = 1; 199 p2 = zombproc.lh_first; 200 goto again; 201 } 202 } 203 204 nprocs++; 205 p2 = newproc; 206 p2->p_stat = SIDL; /* protect against others */ 207 p2->p_pid = lastpid; 208 LIST_INSERT_HEAD(&allproc, p2, p_list); 209 p2->p_forw = p2->p_back = NULL; /* shouldn't be necessary */ 210 LIST_INSERT_HEAD(PIDHASH(p2->p_pid), p2, p_hash); 211 212 /* 213 * Make a proc table entry for the new process. 214 * Start by zeroing the section of proc that is zero-initialized, 215 * then copy the section that is copied directly from the parent. 216 */ 217 bzero(&p2->p_startzero, 218 (unsigned) ((caddr_t)&p2->p_endzero - (caddr_t)&p2->p_startzero)); 219 bcopy(&p1->p_startcopy, &p2->p_startcopy, 220 (unsigned) ((caddr_t)&p2->p_endcopy - (caddr_t)&p2->p_startcopy)); 221 222 /* 223 * Duplicate sub-structures as needed. 224 * Increase reference counts on shared objects. 225 * The p_stats and p_sigacts substructs are set in vm_fork. 226 */ 227 p2->p_flag = P_INMEM; 228 p2->p_emul = p1->p_emul; 229 if (p1->p_flag & P_PROFIL) 230 startprofclock(p2); 231 p2->p_flag |= (p1->p_flag & (P_SUGID | P_SUGIDEXEC)); 232 MALLOC(p2->p_cred, struct pcred *, sizeof(struct pcred), 233 M_SUBPROC, M_WAITOK); 234 bcopy(p1->p_cred, p2->p_cred, sizeof(*p2->p_cred)); 235 p2->p_cred->p_refcnt = 1; 236 crhold(p1->p_ucred); 237 238 /* bump references to the text vnode (for procfs) */ 239 p2->p_textvp = p1->p_textvp; 240 if (p2->p_textvp) 241 VREF(p2->p_textvp); 242 243 if (cleanfd) 244 p2->p_fd = fdinit(p1); 245 else if (dupfd) 246 p2->p_fd = fdcopy(p1); 247 else 248 p2->p_fd = fdshare(p1); 249 250 /* 251 * If p_limit is still copy-on-write, bump refcnt, 252 * otherwise get a copy that won't be modified. 253 * (If PL_SHAREMOD is clear, the structure is shared 254 * copy-on-write.) 255 */ 256 if (p1->p_limit->p_lflags & PL_SHAREMOD) 257 p2->p_limit = limcopy(p1->p_limit); 258 else { 259 p2->p_limit = p1->p_limit; 260 p2->p_limit->p_refcnt++; 261 } 262 263 if (p1->p_session->s_ttyvp != NULL && p1->p_flag & P_CONTROLT) 264 p2->p_flag |= P_CONTROLT; 265 if (forktype == ISVFORK) 266 p2->p_flag |= P_PPWAIT; 267 LIST_INSERT_AFTER(p1, p2, p_pglist); 268 p2->p_pptr = p1; 269 if (rforkflags & RFNOWAIT) { 270 /* XXX should we do anything? */ 271 } else { 272 LIST_INSERT_HEAD(&p1->p_children, p2, p_sibling); 273 } 274 LIST_INIT(&p2->p_children); 275 276 #ifdef KTRACE 277 /* 278 * Copy traceflag and tracefile if enabled. 279 * If not inherited, these were zeroed above. 280 */ 281 if (p1->p_traceflag&KTRFAC_INHERIT) { 282 p2->p_traceflag = p1->p_traceflag; 283 if ((p2->p_tracep = p1->p_tracep) != NULL) 284 VREF(p2->p_tracep); 285 } 286 #endif 287 288 /* 289 * This begins the section where we must prevent the parent 290 * from being swapped. 291 */ 292 p1->p_holdcnt++; 293 294 if (forktype == ISRFORK && (rforkflags & RFMEM)) { 295 /* share as much address space as possible */ 296 (void) vm_map_inherit(&p1->p_vmspace->vm_map, 297 VM_MIN_ADDRESS, VM_MAXUSER_ADDRESS - MAXSSIZ, 298 VM_INHERIT_SHARE); 299 } 300 301 #ifdef __FORK_BRAINDAMAGE 302 /* 303 * Set return values for child before vm_fork, 304 * so they can be copied to child stack. 305 * We return 0, rather than the traditional behaviour of modifying the 306 * return value in the system call stub. 307 * NOTE: the kernel stack may be at a different location in the child 308 * process, and thus addresses of automatic variables (including retval) 309 * may be invalid after vm_fork returns in the child process. 310 */ 311 retval[0] = 0; 312 retval[1] = 1; 313 if (vm_fork(p1, p2)) 314 return (0); 315 #else 316 /* 317 * Finish creating the child process. It will return through a 318 * different path later. 319 */ 320 vm_fork(p1, p2); 321 #endif 322 323 switch (forktype) { 324 case ISFORK: 325 forkstat.cntfork++; 326 break; 327 case ISVFORK: 328 forkstat.cntvfork++; 329 break; 330 case ISRFORK: 331 forkstat.cntrfork++; 332 break; 333 } 334 335 /* 336 * Make child runnable, set start time, and add to run queue. 337 */ 338 (void) splstatclock(); 339 p2->p_stats->p_start = time; 340 p2->p_acflag = AFORK; 341 p2->p_stat = SRUN; 342 setrunqueue(p2); 343 (void) spl0(); 344 345 /* 346 * Now can be swapped. 347 */ 348 p1->p_holdcnt--; 349 350 /* 351 * Preserve synchronization semantics of vfork. If waiting for 352 * child to exec or exit, set P_PPWAIT on child, and sleep on our 353 * proc (in case of exit). 354 */ 355 if (forktype == ISVFORK) 356 while (p2->p_flag & P_PPWAIT) 357 tsleep(p1, PWAIT, "ppwait", 0); 358 359 /* 360 * Return child pid to parent process, 361 * marking us as parent via retval[1]. 362 */ 363 retval[0] = p2->p_pid; 364 retval[1] = 0; 365 return (0); 366 } 367