xref: /openbsd-src/sys/kern/kern_fork.c (revision df930be708d50e9715f173caa26ffe1b7599b157)
1 /*	$NetBSD: kern_fork.c,v 1.25 1995/10/07 06:28:15 mycroft Exp $	*/
2 
3 /*
4  * Copyright (c) 1982, 1986, 1989, 1991, 1993
5  *	The Regents of the University of California.  All rights reserved.
6  * (c) UNIX System Laboratories, Inc.
7  * All or some portions of this file are derived from material licensed
8  * to the University of California by American Telephone and Telegraph
9  * Co. or Unix System Laboratories, Inc. and are reproduced herein with
10  * the permission of UNIX System Laboratories, Inc.
11  *
12  * Redistribution and use in source and binary forms, with or without
13  * modification, are permitted provided that the following conditions
14  * are met:
15  * 1. Redistributions of source code must retain the above copyright
16  *    notice, this list of conditions and the following disclaimer.
17  * 2. Redistributions in binary form must reproduce the above copyright
18  *    notice, this list of conditions and the following disclaimer in the
19  *    documentation and/or other materials provided with the distribution.
20  * 3. All advertising materials mentioning features or use of this software
21  *    must display the following acknowledgement:
22  *	This product includes software developed by the University of
23  *	California, Berkeley and its contributors.
24  * 4. Neither the name of the University nor the names of its contributors
25  *    may be used to endorse or promote products derived from this software
26  *    without specific prior written permission.
27  *
28  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
29  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
30  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
31  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
32  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
33  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
34  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
35  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
36  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
37  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
38  * SUCH DAMAGE.
39  *
40  *	@(#)kern_fork.c	8.6 (Berkeley) 4/8/94
41  */
42 
43 #include <sys/param.h>
44 #include <sys/systm.h>
45 #include <sys/map.h>
46 #include <sys/filedesc.h>
47 #include <sys/kernel.h>
48 #include <sys/malloc.h>
49 #include <sys/proc.h>
50 #include <sys/resourcevar.h>
51 #include <sys/vnode.h>
52 #include <sys/file.h>
53 #include <sys/acct.h>
54 #include <sys/ktrace.h>
55 
56 /* ARGSUSED */
57 sys_fork(p, v, retval)
58 	struct proc *p;
59 	void *v;
60 	register_t *retval;
61 {
62 
63 	return (fork1(p, 0, retval));
64 }
65 
66 /* ARGSUSED */
67 sys_vfork(p, v, retval)
68 	struct proc *p;
69 	void *v;
70 	register_t *retval;
71 {
72 
73 	return (fork1(p, 1, retval));
74 }
75 
76 int	nprocs = 1;		/* process 0 */
77 
78 fork1(p1, isvfork, retval)
79 	register struct proc *p1;
80 	int isvfork;
81 	register_t *retval;
82 {
83 	register struct proc *p2;
84 	register uid_t uid;
85 	struct proc *newproc;
86 	struct proc **hash;
87 	int count;
88 	static int nextpid, pidchecked = 0;
89 
90 	/*
91 	 * Although process entries are dynamically created, we still keep
92 	 * a global limit on the maximum number we will create.  Don't allow
93 	 * a nonprivileged user to use the last process; don't let root
94 	 * exceed the limit. The variable nprocs is the current number of
95 	 * processes, maxproc is the limit.
96 	 */
97 	uid = p1->p_cred->p_ruid;
98 	if ((nprocs >= maxproc - 1 && uid != 0) || nprocs >= maxproc) {
99 		tablefull("proc");
100 		return (EAGAIN);
101 	}
102 
103 	/*
104 	 * Increment the count of procs running with this uid. Don't allow
105 	 * a nonprivileged user to exceed their current limit.
106 	 */
107 	count = chgproccnt(uid, 1);
108 	if (uid != 0 && count > p1->p_rlimit[RLIMIT_NPROC].rlim_cur) {
109 		(void)chgproccnt(uid, -1);
110 		return (EAGAIN);
111 	}
112 
113 	/* Allocate new proc. */
114 	MALLOC(newproc, struct proc *, sizeof(struct proc), M_PROC, M_WAITOK);
115 
116 	/*
117 	 * Find an unused process ID.  We remember a range of unused IDs
118 	 * ready to use (from nextpid+1 through pidchecked-1).
119 	 */
120 	nextpid++;
121 retry:
122 	/*
123 	 * If the process ID prototype has wrapped around,
124 	 * restart somewhat above 0, as the low-numbered procs
125 	 * tend to include daemons that don't exit.
126 	 */
127 	if (nextpid >= PID_MAX) {
128 		nextpid = 100;
129 		pidchecked = 0;
130 	}
131 	if (nextpid >= pidchecked) {
132 		int doingzomb = 0;
133 
134 		pidchecked = PID_MAX;
135 		/*
136 		 * Scan the active and zombie procs to check whether this pid
137 		 * is in use.  Remember the lowest pid that's greater
138 		 * than nextpid, so we can avoid checking for a while.
139 		 */
140 		p2 = allproc.lh_first;
141 again:
142 		for (; p2 != 0; p2 = p2->p_list.le_next) {
143 			while (p2->p_pid == nextpid ||
144 			    p2->p_pgrp->pg_id == nextpid) {
145 				nextpid++;
146 				if (nextpid >= pidchecked)
147 					goto retry;
148 			}
149 			if (p2->p_pid > nextpid && pidchecked > p2->p_pid)
150 				pidchecked = p2->p_pid;
151 			if (p2->p_pgrp->pg_id > nextpid &&
152 			    pidchecked > p2->p_pgrp->pg_id)
153 				pidchecked = p2->p_pgrp->pg_id;
154 		}
155 		if (!doingzomb) {
156 			doingzomb = 1;
157 			p2 = zombproc.lh_first;
158 			goto again;
159 		}
160 	}
161 
162 	nprocs++;
163 	p2 = newproc;
164 	p2->p_stat = SIDL;			/* protect against others */
165 	p2->p_pid = nextpid;
166 	LIST_INSERT_HEAD(&allproc, p2, p_list);
167 	p2->p_forw = p2->p_back = NULL;		/* shouldn't be necessary */
168 	LIST_INSERT_HEAD(PIDHASH(p2->p_pid), p2, p_hash);
169 
170 	/*
171 	 * Make a proc table entry for the new process.
172 	 * Start by zeroing the section of proc that is zero-initialized,
173 	 * then copy the section that is copied directly from the parent.
174 	 */
175 	bzero(&p2->p_startzero,
176 	    (unsigned) ((caddr_t)&p2->p_endzero - (caddr_t)&p2->p_startzero));
177 	bcopy(&p1->p_startcopy, &p2->p_startcopy,
178 	    (unsigned) ((caddr_t)&p2->p_endcopy - (caddr_t)&p2->p_startcopy));
179 
180 	/*
181 	 * Duplicate sub-structures as needed.
182 	 * Increase reference counts on shared objects.
183 	 * The p_stats and p_sigacts substructs are set in vm_fork.
184 	 */
185 	p2->p_flag = P_INMEM;
186 	p2->p_emul = p1->p_emul;
187 	if (p1->p_flag & P_PROFIL)
188 		startprofclock(p2);
189 	MALLOC(p2->p_cred, struct pcred *, sizeof(struct pcred),
190 	    M_SUBPROC, M_WAITOK);
191 	bcopy(p1->p_cred, p2->p_cred, sizeof(*p2->p_cred));
192 	p2->p_cred->p_refcnt = 1;
193 	crhold(p1->p_ucred);
194 
195 	/* bump references to the text vnode (for procfs) */
196 	p2->p_textvp = p1->p_textvp;
197 	if (p2->p_textvp)
198 		VREF(p2->p_textvp);
199 
200 	p2->p_fd = fdcopy(p1);
201 	/*
202 	 * If p_limit is still copy-on-write, bump refcnt,
203 	 * otherwise get a copy that won't be modified.
204 	 * (If PL_SHAREMOD is clear, the structure is shared
205 	 * copy-on-write.)
206 	 */
207 	if (p1->p_limit->p_lflags & PL_SHAREMOD)
208 		p2->p_limit = limcopy(p1->p_limit);
209 	else {
210 		p2->p_limit = p1->p_limit;
211 		p2->p_limit->p_refcnt++;
212 	}
213 
214 	if (p1->p_session->s_ttyvp != NULL && p1->p_flag & P_CONTROLT)
215 		p2->p_flag |= P_CONTROLT;
216 	if (isvfork)
217 		p2->p_flag |= P_PPWAIT;
218 	LIST_INSERT_AFTER(p1, p2, p_pglist);
219 	p2->p_pptr = p1;
220 	LIST_INSERT_HEAD(&p1->p_children, p2, p_sibling);
221 	LIST_INIT(&p2->p_children);
222 
223 #ifdef KTRACE
224 	/*
225 	 * Copy traceflag and tracefile if enabled.
226 	 * If not inherited, these were zeroed above.
227 	 */
228 	if (p1->p_traceflag&KTRFAC_INHERIT) {
229 		p2->p_traceflag = p1->p_traceflag;
230 		if ((p2->p_tracep = p1->p_tracep) != NULL)
231 			VREF(p2->p_tracep);
232 	}
233 #endif
234 
235 	/*
236 	 * This begins the section where we must prevent the parent
237 	 * from being swapped.
238 	 */
239 	p1->p_holdcnt++;
240 	/*
241 	 * Set return values for child before vm_fork,
242 	 * so they can be copied to child stack.
243 	 * We return parent pid, and mark as child in retval[1].
244 	 * NOTE: the kernel stack may be at a different location in the child
245 	 * process, and thus addresses of automatic variables (including retval)
246 	 * may be invalid after vm_fork returns in the child process.
247 	 */
248 	retval[0] = p1->p_pid;
249 	retval[1] = 1;
250 	if (vm_fork(p1, p2, isvfork))
251 		return (0);
252 
253 	/*
254 	 * Make child runnable, set start time, and add to run queue.
255 	 */
256 	(void) splhigh();
257 	p2->p_stats->p_start = time;
258 	p2->p_acflag = AFORK;
259 	p2->p_stat = SRUN;
260 	setrunqueue(p2);
261 	(void) spl0();
262 
263 	/*
264 	 * Now can be swapped.
265 	 */
266 	p1->p_holdcnt--;
267 
268 	/*
269 	 * Preserve synchronization semantics of vfork.  If waiting for
270 	 * child to exec or exit, set P_PPWAIT on child, and sleep on our
271 	 * proc (in case of exit).
272 	 */
273 	if (isvfork)
274 		while (p2->p_flag & P_PPWAIT)
275 			tsleep(p1, PWAIT, "ppwait", 0);
276 
277 	/*
278 	 * Return child pid to parent process,
279 	 * marking us as parent via retval[1].
280 	 */
281 	retval[0] = p2->p_pid;
282 	retval[1] = 0;
283 	return (0);
284 }
285