1 /* $OpenBSD: kern_fork.c,v 1.28 2000/02/21 20:00:09 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 <sys/sched.h> 58 #include <dev/rndvar.h> 59 60 #include <sys/syscallargs.h> 61 62 #include <vm/vm.h> 63 #include <vm/vm_kern.h> 64 65 #if defined(UVM) 66 #include <uvm/uvm_extern.h> 67 #include <uvm/uvm_map.h> 68 #endif 69 70 int nprocs = 1; /* process 0 */ 71 int randompid; /* when set to 1, pid's go random */ 72 pid_t lastpid; 73 struct forkstat forkstat; 74 75 76 /*ARGSUSED*/ 77 int 78 sys_fork(p, v, retval) 79 struct proc *p; 80 void *v; 81 register_t *retval; 82 { 83 return (fork1(p, FORK_FORK, NULL, 0, retval)); 84 } 85 86 /*ARGSUSED*/ 87 int 88 sys_vfork(p, v, retval) 89 struct proc *p; 90 void *v; 91 register_t *retval; 92 { 93 return (fork1(p, FORK_VFORK|FORK_PPWAIT, NULL, 0, retval)); 94 } 95 96 int 97 sys_rfork(p, v, retval) 98 struct proc *p; 99 void *v; 100 register_t *retval; 101 { 102 struct sys_rfork_args /* { 103 syscallarg(int) flags; 104 } */ *uap = v; 105 int rforkflags; 106 int flags; 107 108 flags = FORK_RFORK; 109 rforkflags = SCARG(uap, flags); 110 111 if ((rforkflags & RFPROC) == 0) 112 return (EINVAL); 113 114 switch(rforkflags & (RFFDG|RFCFDG)) { 115 case (RFFDG|RFCFDG): 116 return EINVAL; 117 case RFCFDG: 118 flags |= FORK_CLEANFILES; 119 break; 120 case RFFDG: 121 break; 122 default: 123 flags |= FORK_SHAREFILES; 124 break; 125 } 126 127 if (rforkflags & RFNOWAIT) 128 flags |= FORK_NOZOMBIE; 129 130 if (rforkflags & RFMEM) 131 flags |= FORK_SHAREVM; 132 133 return (fork1(p, flags, NULL, 0, retval)); 134 } 135 136 int 137 fork1(p1, flags, stack, stacksize, retval) 138 register struct proc *p1; 139 int flags; 140 void *stack; 141 size_t stacksize; 142 register_t *retval; 143 { 144 register struct proc *p2; 145 register uid_t uid; 146 struct proc *newproc; 147 struct vmspace *vm; 148 int count; 149 static int pidchecked = 0; 150 vaddr_t uaddr; 151 152 /* 153 * Although process entries are dynamically created, we still keep 154 * a global limit on the maximum number we will create. We reserve 155 * the last 5 processes to root. The variable nprocs is the current 156 * number of processes, maxproc is the limit. 157 */ 158 uid = p1->p_cred->p_ruid; 159 if ((nprocs >= maxproc - 5 && uid != 0) || nprocs >= maxproc) { 160 tablefull("proc"); 161 return (EAGAIN); 162 } 163 164 /* 165 * Increment the count of procs running with this uid. Don't allow 166 * a nonprivileged user to exceed their current limit. 167 */ 168 count = chgproccnt(uid, 1); 169 if (uid != 0 && count > p1->p_rlimit[RLIMIT_NPROC].rlim_cur) { 170 (void)chgproccnt(uid, -1); 171 return (EAGAIN); 172 } 173 174 /* 175 * Allocate a pcb and kernel stack for the process 176 */ 177 #if defined(arc) || defined(mips_cachealias) 178 uaddr = kmem_alloc_upage(kernel_map, USPACE); 179 #else 180 #if defined(UVM) 181 uaddr = uvm_km_valloc(kernel_map, USPACE); 182 #else 183 uaddr = kmem_alloc_pageable(kernel_map, USPACE); 184 #endif 185 #endif 186 if (uaddr == 0) 187 return ENOMEM; 188 189 /* Allocate new proc. */ 190 MALLOC(newproc, struct proc *, sizeof(struct proc), M_PROC, M_WAITOK); 191 192 lastpid++; 193 if (randompid) 194 lastpid = PID_MAX; 195 retry: 196 /* 197 * If the process ID prototype has wrapped around, 198 * restart somewhat above 0, as the low-numbered procs 199 * tend to include daemons that don't exit. 200 */ 201 if (lastpid >= PID_MAX) { 202 lastpid = arc4random() % PID_MAX; 203 pidchecked = 0; 204 } 205 if (lastpid >= pidchecked) { 206 int doingzomb = 0; 207 208 pidchecked = PID_MAX; 209 /* 210 * Scan the active and zombie procs to check whether this pid 211 * is in use. Remember the lowest pid that's greater 212 * than lastpid, so we can avoid checking for a while. 213 */ 214 p2 = allproc.lh_first; 215 again: 216 for (; p2 != 0; p2 = p2->p_list.le_next) { 217 while (p2->p_pid == lastpid || 218 p2->p_pgrp->pg_id == lastpid) { 219 lastpid++; 220 if (lastpid >= pidchecked) 221 goto retry; 222 } 223 if (p2->p_pid > lastpid && pidchecked > p2->p_pid) 224 pidchecked = p2->p_pid; 225 if (p2->p_pgrp->pg_id > lastpid && 226 pidchecked > p2->p_pgrp->pg_id) 227 pidchecked = p2->p_pgrp->pg_id; 228 } 229 if (!doingzomb) { 230 doingzomb = 1; 231 p2 = zombproc.lh_first; 232 goto again; 233 } 234 } 235 236 nprocs++; 237 p2 = newproc; 238 p2->p_stat = SIDL; /* protect against others */ 239 p2->p_pid = lastpid; 240 LIST_INSERT_HEAD(&allproc, p2, p_list); 241 p2->p_forw = p2->p_back = NULL; /* shouldn't be necessary */ 242 LIST_INSERT_HEAD(PIDHASH(p2->p_pid), p2, p_hash); 243 244 /* 245 * Make a proc table entry for the new process. 246 * Start by zeroing the section of proc that is zero-initialized, 247 * then copy the section that is copied directly from the parent. 248 */ 249 bzero(&p2->p_startzero, 250 (unsigned) ((caddr_t)&p2->p_endzero - (caddr_t)&p2->p_startzero)); 251 bcopy(&p1->p_startcopy, &p2->p_startcopy, 252 (unsigned) ((caddr_t)&p2->p_endcopy - (caddr_t)&p2->p_startcopy)); 253 254 /* 255 * Duplicate sub-structures as needed. 256 * Increase reference counts on shared objects. 257 * The p_stats and p_sigacts substructs are set in vm_fork. 258 */ 259 p2->p_flag = P_INMEM; 260 p2->p_emul = p1->p_emul; 261 if (p1->p_flag & P_PROFIL) 262 startprofclock(p2); 263 p2->p_flag |= (p1->p_flag & (P_SUGID | P_SUGIDEXEC)); 264 MALLOC(p2->p_cred, struct pcred *, sizeof(struct pcred), 265 M_SUBPROC, M_WAITOK); 266 bcopy(p1->p_cred, p2->p_cred, sizeof(*p2->p_cred)); 267 p2->p_cred->p_refcnt = 1; 268 crhold(p1->p_ucred); 269 270 /* bump references to the text vnode (for procfs) */ 271 p2->p_textvp = p1->p_textvp; 272 if (p2->p_textvp) 273 VREF(p2->p_textvp); 274 275 if (flags & FORK_CLEANFILES) 276 p2->p_fd = fdinit(p1); 277 else if (flags & FORK_SHAREFILES) 278 p2->p_fd = fdshare(p1); 279 else 280 p2->p_fd = fdcopy(p1); 281 282 /* 283 * If p_limit is still copy-on-write, bump refcnt, 284 * otherwise get a copy that won't be modified. 285 * (If PL_SHAREMOD is clear, the structure is shared 286 * copy-on-write.) 287 */ 288 if (p1->p_limit->p_lflags & PL_SHAREMOD) 289 p2->p_limit = limcopy(p1->p_limit); 290 else { 291 p2->p_limit = p1->p_limit; 292 p2->p_limit->p_refcnt++; 293 } 294 295 if (p1->p_session->s_ttyvp != NULL && p1->p_flag & P_CONTROLT) 296 p2->p_flag |= P_CONTROLT; 297 if (flags & FORK_PPWAIT) 298 p2->p_flag |= P_PPWAIT; 299 LIST_INSERT_AFTER(p1, p2, p_pglist); 300 p2->p_pptr = p1; 301 if (flags & FORK_NOZOMBIE) 302 p2->p_flag |= P_NOZOMBIE; 303 LIST_INSERT_HEAD(&p1->p_children, p2, p_sibling); 304 LIST_INIT(&p2->p_children); 305 306 #ifdef KTRACE 307 /* 308 * Copy traceflag and tracefile if enabled. 309 * If not inherited, these were zeroed above. 310 */ 311 if (p1->p_traceflag & KTRFAC_INHERIT) { 312 p2->p_traceflag = p1->p_traceflag; 313 if ((p2->p_tracep = p1->p_tracep) != NULL) 314 VREF(p2->p_tracep); 315 } 316 #endif 317 318 /* 319 * set priority of child to be that of parent 320 * XXX should move p_estcpu into the region of struct proc which gets 321 * copied. 322 */ 323 scheduler_fork_hook(p1, p2); 324 325 /* 326 * This begins the section where we must prevent the parent 327 * from being swapped. 328 */ 329 p1->p_holdcnt++; 330 331 #if !defined(UVM) /* We do this later for UVM */ 332 if (flags & FORK_SHAREVM) { 333 /* share as much address space as possible */ 334 (void) vm_map_inherit(&p1->p_vmspace->vm_map, 335 VM_MIN_ADDRESS, VM_MAXUSER_ADDRESS - MAXSSIZ, 336 VM_INHERIT_SHARE); 337 } 338 #endif 339 340 p2->p_addr = (struct user *)uaddr; 341 342 #ifdef __FORK_BRAINDAMAGE 343 /* 344 * Set return values for child before vm_fork, 345 * so they can be copied to child stack. 346 * We return 0, rather than the traditional behaviour of modifying the 347 * return value in the system call stub. 348 * NOTE: the kernel stack may be at a different location in the child 349 * process, and thus addresses of automatic variables (including retval) 350 * may be invalid after vm_fork returns in the child process. 351 */ 352 retval[0] = 0; 353 retval[1] = 1; 354 if (vm_fork(p1, p2, stack, stacksize)) 355 return (0); 356 #else 357 /* 358 * Finish creating the child process. It will return through a 359 * different path later. 360 */ 361 #if defined(UVM) 362 uvm_fork(p1, p2, ((flags & FORK_SHAREVM) ? TRUE : FALSE), stack, 363 stacksize); 364 #else /* UVM */ 365 vm_fork(p1, p2, stack, stacksize); 366 #endif /* UVM */ 367 #endif 368 vm = p2->p_vmspace; 369 370 if (flags & FORK_FORK) { 371 forkstat.cntfork++; 372 forkstat.sizfork += vm->vm_dsize + vm->vm_ssize; 373 } else if (flags & FORK_VFORK) { 374 forkstat.cntvfork++; 375 forkstat.sizvfork += vm->vm_dsize + vm->vm_ssize; 376 } else if (flags & FORK_RFORK) { 377 forkstat.cntrfork++; 378 forkstat.sizrfork += vm->vm_dsize + vm->vm_ssize; 379 } 380 381 /* 382 * Make child runnable, set start time, and add to run queue. 383 */ 384 (void) splstatclock(); 385 p2->p_stats->p_start = time; 386 p2->p_acflag = AFORK; 387 p2->p_stat = SRUN; 388 setrunqueue(p2); 389 (void) spl0(); 390 391 /* 392 * Now can be swapped. 393 */ 394 p1->p_holdcnt--; 395 396 #if defined(UVM) 397 uvmexp.forks++; 398 if (flags & FORK_PPWAIT) 399 uvmexp.forks_ppwait++; 400 if (flags & FORK_SHAREVM) 401 uvmexp.forks_sharevm++; 402 #endif 403 404 /* 405 * Preserve synchronization semantics of vfork. If waiting for 406 * child to exec or exit, set P_PPWAIT on child, and sleep on our 407 * proc (in case of exit). 408 */ 409 if (flags & FORK_PPWAIT) 410 while (p2->p_flag & P_PPWAIT) 411 tsleep(p1, PWAIT, "ppwait", 0); 412 413 /* 414 * Return child pid to parent process, 415 * marking us as parent via retval[1]. 416 */ 417 retval[0] = p2->p_pid; 418 retval[1] = 0; 419 return (0); 420 } 421 422