]>
git.saurik.com Git - apple/xnu.git/blob - bsd/kern/kern_fork.c
2 * Copyright (c) 2000-2002 Apple Computer, Inc. All rights reserved.
4 * @APPLE_LICENSE_HEADER_START@
6 * The contents of this file constitute Original Code as defined in and
7 * are subject to the Apple Public Source License Version 1.1 (the
8 * "License"). You may not use this file except in compliance with the
9 * License. Please obtain a copy of the License at
10 * http://www.apple.com/publicsource and read it before using this file.
12 * This Original Code and all software distributed under the License are
13 * distributed on an "AS IS" basis, WITHOUT WARRANTY OF ANY KIND, EITHER
14 * EXPRESS OR IMPLIED, AND APPLE HEREBY DISCLAIMS ALL SUCH WARRANTIES,
15 * INCLUDING WITHOUT LIMITATION, ANY WARRANTIES OF MERCHANTABILITY,
16 * FITNESS FOR A PARTICULAR PURPOSE OR NON-INFRINGEMENT. Please see the
17 * License for the specific language governing rights and limitations
20 * @APPLE_LICENSE_HEADER_END@
22 /* Copyright (c) 1995, 1997 Apple Computer, Inc. All Rights Reserved */
24 * Copyright (c) 1982, 1986, 1989, 1991, 1993
25 * The Regents of the University of California. All rights reserved.
26 * (c) UNIX System Laboratories, Inc.
27 * All or some portions of this file are derived from material licensed
28 * to the University of California by American Telephone and Telegraph
29 * Co. or Unix System Laboratories, Inc. and are reproduced herein with
30 * the permission of UNIX System Laboratories, Inc.
32 * Redistribution and use in source and binary forms, with or without
33 * modification, are permitted provided that the following conditions
35 * 1. Redistributions of source code must retain the above copyright
36 * notice, this list of conditions and the following disclaimer.
37 * 2. Redistributions in binary form must reproduce the above copyright
38 * notice, this list of conditions and the following disclaimer in the
39 * documentation and/or other materials provided with the distribution.
40 * 3. All advertising materials mentioning features or use of this software
41 * must display the following acknowledgement:
42 * This product includes software developed by the University of
43 * California, Berkeley and its contributors.
44 * 4. Neither the name of the University nor the names of its contributors
45 * may be used to endorse or promote products derived from this software
46 * without specific prior written permission.
48 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
49 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
50 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
51 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
52 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
53 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
54 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
55 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
56 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
57 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
60 * @(#)kern_fork.c 8.8 (Berkeley) 2/14/95
63 #include <sys/param.h>
64 #include <sys/systm.h>
65 #include <sys/filedesc.h>
66 #include <sys/kernel.h>
67 #include <sys/malloc.h>
70 #include <sys/resourcevar.h>
71 #include <sys/vnode.h>
75 #include <sys/ktrace.h>
78 #include <mach/mach_types.h>
79 #include <kern/mach_param.h>
81 #include <machine/spl.h>
83 thread_act_t
cloneproc(struct proc
*, int);
84 struct proc
* forkproc(struct proc
*, int);
85 thread_act_t
procdup();
87 #define DOFORK 0x1 /* fork() system call */
88 #define DOVFORK 0x2 /* vfork() system call */
89 static int fork1(struct proc
*, long, register_t
*);
100 return (fork1(p
, (long)DOFORK
, retval
));
107 vfork(p
, uap
, retval
)
112 register struct proc
* newproc
;
114 thread_act_t cur_act
= (thread_act_t
)current_act();
120 * Although process entries are dynamically created, we still keep
121 * a global limit on the maximum number we will create. Don't allow
122 * a nonprivileged user to use the last process; don't let root
123 * exceed the limit. The variable nprocs is the current number of
124 * processes, maxproc is the limit.
126 uid
= p
->p_cred
->p_ruid
;
127 if ((nprocs
>= maxproc
- 1 && uid
!= 0) || nprocs
>= maxproc
) {
134 * Increment the count of procs running with this uid. Don't allow
135 * a nonprivileged user to exceed their current limit.
137 count
= chgproccnt(uid
, 1);
138 if (uid
!= 0 && count
> p
->p_rlimit
[RLIMIT_NPROC
].rlim_cur
) {
139 (void)chgproccnt(uid
, -1);
143 ut
= (struct uthread
*)get_bsdthread_info(cur_act
);
144 if (ut
->uu_flag
& P_VFORK
) {
145 printf("vfork called recursively by %s\n", p
->p_comm
);
148 p
->p_flag
|= P_VFORK
;
151 /* The newly created process comes with signal lock held */
152 newproc
= (struct proc
*)forkproc(p
,1);
154 LIST_INSERT_AFTER(p
, newproc
, p_pglist
);
156 newproc
->task
= p
->task
;
157 LIST_INSERT_HEAD(&p
->p_children
, newproc
, p_sibling
);
158 LIST_INIT(&newproc
->p_children
);
159 LIST_INSERT_HEAD(&allproc
, newproc
, p_list
);
160 LIST_INSERT_HEAD(PIDHASH(newproc
->p_pid
), newproc
, p_hash
);
161 TAILQ_INIT(& newproc
->p_evlist
);
162 newproc
->p_stat
= SRUN
;
163 newproc
->p_flag
|= P_INVFORK
;
164 newproc
->p_vforkact
= cur_act
;
166 ut
->uu_flag
|= P_VFORK
;
167 ut
->uu_proc
= newproc
;
168 ut
->uu_userstate
= (void *)act_thread_csave();
169 ut
->uu_vforkmask
= ut
->uu_sigmask
;
171 thread_set_child(cur_act
, newproc
->p_pid
);
173 newproc
->p_stats
->p_start
= time
;
174 newproc
->p_acflag
= AFORK
;
177 * Preserve synchronization semantics of vfork. If waiting for
178 * child to exec or exit, set P_PPWAIT on child, and sleep on our
179 * proc (in case of exit).
181 newproc
->p_flag
|= P_PPWAIT
;
183 /* drop the signal lock on the child */
184 signal_unlock(newproc
);
186 retval
[0] = newproc
->p_pid
;
187 retval
[1] = 1; /* mark child */
193 * Return to parent vfork ehread()
196 vfork_return(th_act
, p
, p2
, retval
)
204 thread_t newth
, self
= current_thread();
205 thread_act_t cur_act
= (thread_act_t
)current_act();
210 ut
= (struct uthread
*)get_bsdthread_info(cur_act
);
212 act_thread_catt(ut
->uu_userstate
);
214 /* Make sure only one at this time */
216 if (p
->p_vforkcnt
<0)
217 panic("vfork cnt is -ve");
218 if (p
->p_vforkcnt
<=0)
219 p
->p_flag
&= ~P_VFORK
;
220 ut
->uu_userstate
= 0;
221 ut
->uu_flag
&= ~P_VFORK
;
223 ut
->uu_sigmask
= ut
->uu_vforkmask
;
224 p2
->p_flag
&= ~P_INVFORK
;
225 p2
->p_vforkact
= (void *)0;
227 thread_set_parent(cur_act
, p2
->p_pid
);
230 retval
[0] = p2
->p_pid
;
231 retval
[1] = 0; /* mark parent */
244 kern_return_t result
;
245 extern task_t kernel_task
;
247 if (parent
->task
== kernel_task
)
248 result
= task_create_local(TASK_NULL
, FALSE
, FALSE
, &task
);
250 result
= task_create_local(parent
->task
, TRUE
, FALSE
, &task
);
251 if (result
!= KERN_SUCCESS
)
252 printf("fork/procdup: task_create failed. Code: 0x%x\n", result
);
254 /* task->proc = child; */
255 set_bsdtask_info(task
, child
);
256 if (child
->p_nice
!= 0)
257 resetpriority(child
);
258 result
= thread_create(task
, &thread
);
259 if (result
!= KERN_SUCCESS
)
260 printf("fork/procdup: thread_create failed. Code: 0x%x\n", result
);
267 fork1(p1
, flags
, retval
)
272 register struct proc
*p2
;
279 * Although process entries are dynamically created, we still keep
280 * a global limit on the maximum number we will create. Don't allow
281 * a nonprivileged user to use the last process; don't let root
282 * exceed the limit. The variable nprocs is the current number of
283 * processes, maxproc is the limit.
285 uid
= p1
->p_cred
->p_ruid
;
286 if ((nprocs
>= maxproc
- 1 && uid
!= 0) || nprocs
>= maxproc
) {
293 * Increment the count of procs running with this uid. Don't allow
294 * a nonprivileged user to exceed their current limit.
296 count
= chgproccnt(uid
, 1);
297 if (uid
!= 0 && count
> p1
->p_rlimit
[RLIMIT_NPROC
].rlim_cur
) {
298 (void)chgproccnt(uid
, -1);
302 /* The newly created process comes with signal lock held */
303 newth
= cloneproc(p1
, 1);
305 /* p2 = newth->task->proc; */
306 p2
= (struct proc
*)(get_bsdtask_info(get_threadtask(newth
)));
308 thread_set_child(newth
, p2
->p_pid
);
311 p2
->p_stats
->p_start
= time
;
313 p2
->p_acflag
= AFORK
;
316 * Preserve synchronization semantics of vfork. If waiting for
317 * child to exec or exit, set P_PPWAIT on child, and sleep on our
318 * proc (in case of exit).
320 if (flags
== DOVFORK
)
321 p2
->p_flag
|= P_PPWAIT
;
322 /* drop the signal lock on the child */
325 (void) thread_resume(newth
);
327 /* drop the extra references we got during the creation */
328 if (t
= (task_t
)get_threadtask(newth
)) {
331 act_deallocate(newth
);
333 while (p2
->p_flag
& P_PPWAIT
)
334 tsleep(p1
, PWAIT
, "ppwait", 0);
336 retval
[0] = p2
->p_pid
;
337 retval
[1] = 0; /* mark parent */
345 * Create a new process from a specified process.
346 * On return newly created child process has signal
347 * lock held to block delivery of signal to it if called with
348 * lock set. fork() code needs to explicity remove this lock
349 * before signals can be delivered
353 register struct proc
*p1
;
356 register struct proc
*p2
;
359 p2
= (struct proc
*)forkproc(p1
,lock
);
362 th
= procdup(p2
, p1
); /* child, parent */
364 LIST_INSERT_AFTER(p1
, p2
, p_pglist
);
366 LIST_INSERT_HEAD(&p1
->p_children
, p2
, p_sibling
);
367 LIST_INIT(&p2
->p_children
);
368 LIST_INSERT_HEAD(&allproc
, p2
, p_list
);
369 LIST_INSERT_HEAD(PIDHASH(p2
->p_pid
), p2
, p_hash
);
370 TAILQ_INIT(&p2
->p_evlist
);
372 * Make child runnable, set start time.
381 register struct proc
*p1
;
384 register struct proc
*p2
, *newproc
;
385 static int nextpid
= 0, pidchecked
= 0;
388 /* Allocate new proc. */
389 MALLOC_ZONE(newproc
, struct proc
*,
390 sizeof *newproc
, M_PROC
, M_WAITOK
);
391 MALLOC_ZONE(newproc
->p_cred
, struct pcred
*,
392 sizeof *newproc
->p_cred
, M_SUBPROC
, M_WAITOK
);
393 MALLOC_ZONE(newproc
->p_stats
, struct pstats
*,
394 sizeof *newproc
->p_stats
, M_SUBPROC
, M_WAITOK
);
395 MALLOC_ZONE(newproc
->p_sigacts
, struct sigacts
*,
396 sizeof *newproc
->p_sigacts
, M_SUBPROC
, M_WAITOK
);
399 * Find an unused process ID. We remember a range of unused IDs
400 * ready to use (from nextpid+1 through pidchecked-1).
405 * If the process ID prototype has wrapped around,
406 * restart somewhat above 0, as the low-numbered procs
407 * tend to include daemons that don't exit.
409 if (nextpid
>= PID_MAX
) {
413 if (nextpid
>= pidchecked
) {
416 pidchecked
= PID_MAX
;
418 * Scan the active and zombie procs to check whether this pid
419 * is in use. Remember the lowest pid that's greater
420 * than nextpid, so we can avoid checking for a while.
422 p2
= allproc
.lh_first
;
424 for (; p2
!= 0; p2
= p2
->p_list
.le_next
) {
425 while (p2
->p_pid
== nextpid
||
426 p2
->p_pgrp
->pg_id
== nextpid
||
427 p2
->p_session
->s_sid
== nextpid
) {
429 if (nextpid
>= pidchecked
)
432 if (p2
->p_pid
> nextpid
&& pidchecked
> p2
->p_pid
)
433 pidchecked
= p2
->p_pid
;
434 if (p2
->p_pgrp
&& p2
->p_pgrp
->pg_id
> nextpid
&&
435 pidchecked
> p2
->p_pgrp
->pg_id
)
436 pidchecked
= p2
->p_pgrp
->pg_id
;
437 if (p2
->p_session
->s_sid
> nextpid
&&
438 pidchecked
> p2
->p_session
->s_sid
)
439 pidchecked
= p2
->p_session
->s_sid
;
443 p2
= zombproc
.lh_first
;
454 * Make a proc table entry for the new process.
455 * Start by zeroing the section of proc that is zero-initialized,
456 * then copy the section that is copied directly from the parent.
458 bzero(&p2
->p_startzero
,
459 (unsigned) ((caddr_t
)&p2
->p_endzero
- (caddr_t
)&p2
->p_startzero
));
460 bcopy(&p1
->p_startcopy
, &p2
->p_startcopy
,
461 (unsigned) ((caddr_t
)&p2
->p_endcopy
- (caddr_t
)&p2
->p_startcopy
));
462 p2
->vm_shm
= (void *)NULL
; /* Make sure it is zero */
465 * Duplicate sub-structures as needed.
466 * Increase reference counts on shared objects.
467 * The p_stats and p_sigacts substructs are set in vm_fork.
469 p2
->p_flag
= P_INMEM
;
470 if (p1
->p_flag
& P_PROFIL
)
472 bcopy(p1
->p_cred
, p2
->p_cred
, sizeof(*p2
->p_cred
));
473 p2
->p_cred
->p_refcnt
= 1;
475 lockinit(&p2
->p_cred
->pc_lock
, PLOCK
, "proc cred", 0, 0);
477 /* bump references to the text vnode */
478 p2
->p_textvp
= p1
->p_textvp
;
482 p2
->p_fd
= fdcopy(p1
);
487 * If p_limit is still copy-on-write, bump refcnt,
488 * otherwise get a copy that won't be modified.
489 * (If PL_SHAREMOD is clear, the structure is shared
492 if (p1
->p_limit
->p_lflags
& PL_SHAREMOD
)
493 p2
->p_limit
= limcopy(p1
->p_limit
);
495 p2
->p_limit
= p1
->p_limit
;
496 p2
->p_limit
->p_refcnt
++;
499 bzero(&p2
->p_stats
->pstat_startzero
,
500 (unsigned) ((caddr_t
)&p2
->p_stats
->pstat_endzero
-
501 (caddr_t
)&p2
->p_stats
->pstat_startzero
));
502 bcopy(&p1
->p_stats
->pstat_startcopy
, &p2
->p_stats
->pstat_startcopy
,
503 ((caddr_t
)&p2
->p_stats
->pstat_endcopy
-
504 (caddr_t
)&p2
->p_stats
->pstat_startcopy
));
506 if (p1
->p_sigacts
!= NULL
)
507 (void)memcpy(p2
->p_sigacts
,
508 p1
->p_sigacts
, sizeof *p2
->p_sigacts
);
510 (void)memset(p2
->p_sigacts
, 0, sizeof *p2
->p_sigacts
);
512 if (p1
->p_session
->s_ttyvp
!= NULL
&& p1
->p_flag
& P_CONTROLT
)
513 p2
->p_flag
|= P_CONTROLT
;
518 p2
->p_debugger
= 0; /* don't inherit */
519 lockinit(&p2
->signal_lock
, PVM
, "signal", 0, 0);
520 /* block all signals to reach the process */
524 p2
->sigwait_thread
= NULL
;
525 p2
->exit_thread
= NULL
;
526 p2
->user_stack
= p1
->user_stack
;
527 p2
->p_xxxsigpending
= 0;
530 TAILQ_INIT(&p2
->p_uthlist
);
534 * Copy traceflag and tracefile if enabled.
535 * If not inherited, these were zeroed above.
537 if (p1
->p_traceflag
&KTRFAC_INHERIT
) {
538 p2
->p_traceflag
= p1
->p_traceflag
;
539 if ((p2
->p_tracep
= p1
->p_tracep
) != NULL
)
547 #include <kern/zalloc.h>
549 struct zone
*uthread_zone
;
550 int uthread_zone_inited
= 0;
555 if (!uthread_zone_inited
) {
556 uthread_zone
= zinit(sizeof(struct uthread
),
557 THREAD_MAX
* sizeof(struct uthread
),
558 THREAD_CHUNK
* sizeof(struct uthread
),
560 uthread_zone_inited
= 1;
565 uthread_alloc(task_t task
, thread_act_t thr_act
)
568 struct uthread
*uth
, *uth_parent
;
570 extern task_t kernel_task
;
571 boolean_t funnel_state
;
573 if (!uthread_zone_inited
)
576 ut
= (void *)zalloc(uthread_zone
);
577 bzero(ut
, sizeof(struct uthread
));
579 if (task
!= kernel_task
) {
580 uth
= (struct uthread
*)ut
;
581 p
= get_bsdtask_info(task
);
583 funnel_state
= thread_funnel_set(kernel_flock
, TRUE
);
584 uth_parent
= (struct uthread
*)get_bsdthread_info(current_act());
586 if (uth_parent
->uu_flag
& USAS_OLDMASK
)
587 uth
->uu_sigmask
= uth_parent
->uu_oldmask
;
589 uth
->uu_sigmask
= uth_parent
->uu_sigmask
;
591 uth
->uu_act
= thr_act
;
594 TAILQ_INSERT_TAIL(&p
->p_uthlist
, uth
, uu_list
);
596 (void)thread_funnel_set(kernel_flock
, funnel_state
);
604 uthread_free(task_t task
, void *uthread
, void * bsd_info
)
607 struct uthread
*uth
= (struct uthread
*)uthread
;
608 struct proc
* p
= (struct proc
*)bsd_info
;
609 extern task_t kernel_task
;
611 boolean_t funnel_state
;
613 sel
= &uth
->uu_state
.ss_select
;
614 /* cleanup the select bit space */
616 FREE(sel
->ibits
, M_TEMP
);
617 FREE(sel
->obits
, M_TEMP
);
620 if (sel
->allocsize
&& uth
->uu_wqsub
){
621 kfree(uth
->uu_wqsub
, sel
->allocsize
);
622 sel
->count
= sel
->nfcount
= 0;
628 if ((task
!= kernel_task
) && p
) {
629 funnel_state
= thread_funnel_set(kernel_flock
, TRUE
);
631 TAILQ_REMOVE(&p
->p_uthlist
, uth
, uu_list
);
633 (void)thread_funnel_set(kernel_flock
, funnel_state
);
635 /* and free the uthread itself */
636 zfree(uthread_zone
, (vm_offset_t
)uthread
);