bsd/kern/kern_fork.c

   1 /*
   2  * Copyright (c) 2006 Apple Computer, Inc. All Rights Reserved.
   3  *
   4  * @APPLE_LICENSE_OSREFERENCE_HEADER_START@
   5  *
   6  * This file contains Original Code and/or Modifications of Original Code
   7  * as defined in and that are subject to the Apple Public Source License
   8  * Version 2.0 (the 'License'). You may not use this file except in
   9  * compliance with the License.  The rights granted to you under the
  10  * License may not be used to create, or enable the creation or
  11  * redistribution of, unlawful or unlicensed copies of an Apple operating
  12  * system, or to circumvent, violate, or enable the circumvention or
  13  * violation of, any terms of an Apple operating system software license
  14  * agreement.
  15  *
  16  * Please obtain a copy of the License at
  17  * http://www.opensource.apple.com/apsl/ and read it before using this
  18  * file.
  19  *
  20  * The Original Code and all software distributed under the License are
  21  * distributed on an 'AS IS' basis, WITHOUT WARRANTY OF ANY KIND, EITHER
  22  * EXPRESS OR IMPLIED, AND APPLE HEREBY DISCLAIMS ALL SUCH WARRANTIES,
  23  * INCLUDING WITHOUT LIMITATION, ANY WARRANTIES OF MERCHANTABILITY,
  24  * FITNESS FOR A PARTICULAR PURPOSE, QUIET ENJOYMENT OR NON-INFRINGEMENT.
  25  * Please see the License for the specific language governing rights and
  26  * limitations under the License.
  27  *
  28  * @APPLE_LICENSE_OSREFERENCE_HEADER_END@
  29  */
  30 /* Copyright (c) 1995, 1997 Apple Computer, Inc. All Rights Reserved */
  31 /*
  32  * Copyright (c) 1982, 1986, 1989, 1991, 1993
  33  *      The Regents of the University of California.  All rights reserved.
  34  * (c) UNIX System Laboratories, Inc.
  35  * All or some portions of this file are derived from material licensed
  36  * to the University of California by American Telephone and Telegraph
  37  * Co. or Unix System Laboratories, Inc. and are reproduced herein with
  38  * the permission of UNIX System Laboratories, Inc.
  39  *
  40  * Redistribution and use in source and binary forms, with or without
  41  * modification, are permitted provided that the following conditions
  42  * are met:
  43  * 1. Redistributions of source code must retain the above copyright
  44  *    notice, this list of conditions and the following disclaimer.
  45  * 2. Redistributions in binary form must reproduce the above copyright
  46  *    notice, this list of conditions and the following disclaimer in the
  47  *    documentation and/or other materials provided with the distribution.
  48  * 3. All advertising materials mentioning features or use of this software
  49  *    must display the following acknowledgement:
  50  *      This product includes software developed by the University of
  51  *      California, Berkeley and its contributors.
  52  * 4. Neither the name of the University nor the names of its contributors
  53  *    may be used to endorse or promote products derived from this software
  54  *    without specific prior written permission.
  55  *
  56  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
  57  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
  58  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
  59  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
  60  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
  61  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
  62  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
  63  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
  64  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
  65  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
  66  * SUCH DAMAGE.
  67  *
  68  *      @(#)kern_fork.c 8.8 (Berkeley) 2/14/95
  69  */
  70
  71 #include <kern/assert.h>
  72 #include <sys/param.h>
  73 #include <sys/systm.h>
  74 #include <sys/filedesc.h>
  75 #include <sys/kernel.h>
  76 #include <sys/malloc.h>
  77 #include <sys/proc_internal.h>
  78 #include <sys/kauth.h>
  79 #include <sys/user.h>
  80 #include <sys/resourcevar.h>
  81 #include <sys/vnode_internal.h>
  82 #include <sys/file_internal.h>
  83 #include <sys/acct.h>
  84 #if KTRACE
  85 #include <sys/ktrace.h>
  86 #endif
  87
  88 #include <bsm/audit_kernel.h>
  89
  90 #include <mach/mach_types.h>
  91 #include <kern/kern_types.h>
  92 #include <kern/kalloc.h>
  93 #include <kern/mach_param.h>
  94 #include <kern/task.h>
  95 #include <kern/zalloc.h>
  96
  97 #include <machine/spl.h>
  98
  99 #include <vm/vm_protos.h>       // for vm_map_commpage64
 100
 101 thread_t cloneproc(struct proc *, int);
 102 struct proc * forkproc(struct proc *, int);
 103 thread_t procdup(struct proc *child, struct proc *parent);
 104
 105 #define DOFORK  0x1     /* fork() system call */
 106 #define DOVFORK 0x2     /* vfork() system call */
 107 static int fork1(struct proc *, long, register_t *);
 108
 109 /*
 110  * fork system call.
 111  */
 112 int
 113 fork(struct proc *p, __unused void *uap, register_t *retval)
 114 {
 115         return (fork1(p, (long)DOFORK, retval));
 116 }
 117
 118 /*
 119  * vfork system call
 120  */
 121 int
 122 vfork(struct proc *p, void *uap, register_t *retval)
 123 {
 124         register struct proc * newproc;
 125         register uid_t uid;
 126         thread_t cur_act = (thread_t)current_thread();
 127         int count;
 128         task_t t;
 129         uthread_t ut;
 130
 131         /*
 132          * Although process entries are dynamically created, we still keep
 133          * a global limit on the maximum number we will create.  Don't allow
 134          * a nonprivileged user to use the last process; don't let root
 135          * exceed the limit. The variable nprocs is the current number of
 136          * processes, maxproc is the limit.
 137          */
 138         uid = kauth_cred_get()->cr_ruid;
 139         if ((nprocs >= maxproc - 1 && uid != 0) || nprocs >= maxproc) {
 140                 tablefull("proc");
 141                 retval[1] = 0;
 142                 return (EAGAIN);
 143         }
 144
 145         /*
 146          * Increment the count of procs running with this uid. Don't allow
 147          * a nonprivileged user to exceed their current limit.
 148          */
 149         count = chgproccnt(uid, 1);
 150         if (uid != 0 && count > p->p_rlimit[RLIMIT_NPROC].rlim_cur) {
 151                 (void)chgproccnt(uid, -1);
 152                 return (EAGAIN);
 153         }
 154
 155         ut = (struct uthread *)get_bsdthread_info(cur_act);
 156         if (ut->uu_flag & UT_VFORK) {
 157                 printf("vfork called recursively by %s\n", p->p_comm);
 158                 (void)chgproccnt(uid, -1);
 159                 return (EINVAL);
 160         }
 161         p->p_flag  |= P_VFORK;
 162         p->p_vforkcnt++;
 163
 164         /* The newly created process comes with signal lock held */
 165         newproc = (struct proc *)forkproc(p,1);
 166
 167         AUDIT_ARG(pid, newproc->p_pid);
 168
 169         LIST_INSERT_AFTER(p, newproc, p_pglist);
 170         newproc->p_pptr = p;
 171         newproc->task = p->task;
 172         LIST_INSERT_HEAD(&p->p_children, newproc, p_sibling);
 173         LIST_INIT(&newproc->p_children);
 174         LIST_INSERT_HEAD(&allproc, newproc, p_list);
 175         LIST_INSERT_HEAD(PIDHASH(newproc->p_pid), newproc, p_hash);
 176         TAILQ_INIT(& newproc->p_evlist);
 177         newproc->p_stat = SRUN;
 178         newproc->p_flag  |= P_INVFORK;
 179         newproc->p_vforkact = cur_act;
 180
 181         ut->uu_flag |= UT_VFORK;
 182         ut->uu_proc = newproc;
 183         ut->uu_userstate = (void *)act_thread_csave();
 184         ut->uu_vforkmask = ut->uu_sigmask;
 185
 186         /* temporarily drop thread-set-id state */
 187         if (ut->uu_flag & UT_SETUID) {
 188                 ut->uu_flag |= UT_WASSETUID;
 189                 ut->uu_flag &= ~UT_SETUID;
 190         }
 191
 192         thread_set_child(cur_act, newproc->p_pid);
 193
 194         microtime(&newproc->p_stats->p_start);
 195         newproc->p_acflag = AFORK;
 196
 197         /*
 198          * Preserve synchronization semantics of vfork.  If waiting for
 199          * child to exec or exit, set P_PPWAIT on child, and sleep on our
 200          * proc (in case of exit).
 201          */
 202         newproc->p_flag |= P_PPWAIT;
 203
 204         /* drop the signal lock on the child */
 205         signal_unlock(newproc);
 206
 207         retval[0] = newproc->p_pid;
 208         retval[1] = 1;                  /* mark child */
 209
 210         return (0);
 211 }
 212
 213 /*
 214  * Return to parent vfork ehread()
 215  */
 216 void
 217 vfork_return(__unused thread_t th_act, struct proc *p, struct proc *p2,
 218         register_t *retval)
 219 {
 220         thread_t cur_act = (thread_t)current_thread();
 221         uthread_t ut;
 222
 223         ut = (struct uthread *)get_bsdthread_info(cur_act);
 224
 225         act_thread_catt(ut->uu_userstate);
 226
 227         /* Make sure only one at this time */
 228         p->p_vforkcnt--;
 229         if (p->p_vforkcnt <0)
 230                 panic("vfork cnt is -ve");
 231         if (p->p_vforkcnt <=0)
 232                 p->p_flag  &= ~P_VFORK;
 233         ut->uu_userstate = 0;
 234         ut->uu_flag &= ~UT_VFORK;
 235         /* restore thread-set-id state */
 236         if (ut->uu_flag & UT_WASSETUID) {
 237                 ut->uu_flag |= UT_SETUID;
 238                 ut->uu_flag &= UT_WASSETUID;
 239         }
 240         ut->uu_proc = 0;
 241         ut->uu_sigmask = ut->uu_vforkmask;
 242         p2->p_flag  &= ~P_INVFORK;
 243         p2->p_vforkact = (void *)0;
 244
 245         thread_set_parent(cur_act, p2->p_pid);
 246
 247         if (retval) {
 248                 retval[0] = p2->p_pid;
 249                 retval[1] = 0;                  /* mark parent */
 250         }
 251
 252         return;
 253 }
 254
 255 thread_t
 256 procdup(struct proc *child, struct proc *parent)
 257 {
 258         thread_t                thread;
 259         task_t                  task;
 260         kern_return_t   result;
 261
 262         if (parent->task == kernel_task)
 263                 result = task_create_internal(TASK_NULL, FALSE, &task);
 264         else
 265                 result = task_create_internal(parent->task, TRUE, &task);
 266         if (result != KERN_SUCCESS)
 267             printf("fork/procdup: task_create failed. Code: 0x%x\n", result);
 268         child->task = task;
 269         /* task->proc = child; */
 270         set_bsdtask_info(task, child);
 271         if (parent->p_flag & P_LP64) {
 272                 task_set_64bit(task, TRUE);
 273                 child->p_flag |= P_LP64;
 274 #ifdef __PPC__
 275                 /* LP64todo - clean up this hacked mapping of commpage */
 276                 pmap_map_sharedpage(task, get_map_pmap(get_task_map(task)));
 277                 vm_map_commpage64(get_task_map(task));
 278 #endif  /* __PPC__ */
 279         } else {
 280                 task_set_64bit(task, FALSE);
 281                 child->p_flag &= ~P_LP64;
 282         }
 283         if (child->p_nice != 0)
 284                 resetpriority(child);
 285
 286         result = thread_create(task, &thread);
 287         if (result != KERN_SUCCESS)
 288             printf("fork/procdup: thread_create failed. Code: 0x%x\n", result);
 289
 290         return(thread);
 291 }
 292
 293
 294 static int
 295 fork1(p1, flags, retval)
 296         struct proc *p1;
 297         long flags;
 298         register_t *retval;
 299 {
 300         register struct proc *p2;
 301         register uid_t uid;
 302         thread_t newth;
 303         int count;
 304         task_t t;
 305
 306         /*
 307          * Although process entries are dynamically created, we still keep
 308          * a global limit on the maximum number we will create.  Don't allow
 309          * a nonprivileged user to use the last process; don't let root
 310          * exceed the limit. The variable nprocs is the current number of
 311          * processes, maxproc is the limit.
 312          */
 313         uid = kauth_cred_get()->cr_ruid;
 314         if ((nprocs >= maxproc - 1 && uid != 0) || nprocs >= maxproc) {
 315                 tablefull("proc");
 316                 retval[1] = 0;
 317                 return (EAGAIN);
 318         }
 319
 320         /*
 321          * Increment the count of procs running with this uid. Don't allow
 322          * a nonprivileged user to exceed their current limit.
 323          */
 324         count = chgproccnt(uid, 1);
 325         if (uid != 0 && count > p1->p_rlimit[RLIMIT_NPROC].rlim_cur) {
 326                 (void)chgproccnt(uid, -1);
 327                 return (EAGAIN);
 328         }
 329
 330         /* The newly created process comes with signal lock held */
 331         newth = cloneproc(p1, 1);
 332         thread_dup(newth);
 333         /* p2 = newth->task->proc; */
 334         p2 = (struct proc *)(get_bsdtask_info(get_threadtask(newth)));
 335         set_security_token(p2);         /* propagate change of PID */
 336
 337         AUDIT_ARG(pid, p2->p_pid);
 338
 339         thread_set_child(newth, p2->p_pid);
 340
 341         microtime(&p2->p_stats->p_start);
 342         p2->p_acflag = AFORK;
 343
 344         /*
 345          * Preserve synchronization semantics of vfork.  If waiting for
 346          * child to exec or exit, set P_PPWAIT on child, and sleep on our
 347          * proc (in case of exit).
 348          */
 349         if (flags == DOVFORK)
 350                 p2->p_flag |= P_PPWAIT;
 351         /* drop the signal lock on the child */
 352         signal_unlock(p2);
 353
 354         (void) thread_resume(newth);
 355
 356         /* drop the extra references we got during the creation */
 357         if ((t = (task_t)get_threadtask(newth)) != NULL) {
 358                 task_deallocate(t);
 359         }
 360         thread_deallocate(newth);
 361
 362         KNOTE(&p1->p_klist, NOTE_FORK | p2->p_pid);
 363
 364         while (p2->p_flag & P_PPWAIT)
 365                 tsleep(p1, PWAIT, "ppwait", 0);
 366
 367         retval[0] = p2->p_pid;
 368         retval[1] = 0;                  /* mark parent */
 369
 370         return (0);
 371 }
 372
 373 /*
 374  * cloneproc()
 375  *
 376  * Create a new process from a specified process.
 377  * On return newly created child process has signal
 378  * lock held to block delivery of signal to it if called with
 379  * lock set. fork() code needs to explicity remove this lock
 380  * before signals can be delivered
 381  */
 382 thread_t
 383 cloneproc(p1, lock)
 384         register struct proc *p1;
 385         register int lock;
 386 {
 387         register struct proc *p2;
 388         thread_t th;
 389
 390         p2 = (struct proc *)forkproc(p1,lock);
 391
 392
 393         th = procdup(p2, p1);   /* child, parent */
 394
 395         LIST_INSERT_AFTER(p1, p2, p_pglist);
 396         p2->p_pptr = p1;
 397         LIST_INSERT_HEAD(&p1->p_children, p2, p_sibling);
 398         LIST_INIT(&p2->p_children);
 399         LIST_INSERT_HEAD(&allproc, p2, p_list);
 400         LIST_INSERT_HEAD(PIDHASH(p2->p_pid), p2, p_hash);
 401         TAILQ_INIT(&p2->p_evlist);
 402         /*
 403          * Make child runnable, set start time.
 404          */
 405         p2->p_stat = SRUN;
 406
 407         return(th);
 408 }
 409
 410 struct proc *
 411 forkproc(p1, lock)
 412         register struct proc *p1;
 413         register int lock;
 414 {
 415         register struct proc *p2, *newproc;
 416         static int nextpid = 0, pidchecked = 0;
 417
 418         /* Allocate new proc. */
 419         MALLOC_ZONE(newproc, struct proc *,
 420                         sizeof *newproc, M_PROC, M_WAITOK);
 421         if (newproc == NULL)
 422                 panic("forkproc: M_PROC zone exhausted");
 423         MALLOC_ZONE(newproc->p_stats, struct pstats *,
 424                         sizeof *newproc->p_stats, M_SUBPROC, M_WAITOK);
 425         if (newproc->p_stats == NULL)
 426                 panic("forkproc: M_SUBPROC zone exhausted (p_stats)");
 427         MALLOC_ZONE(newproc->p_sigacts, struct sigacts *,
 428                         sizeof *newproc->p_sigacts, M_SUBPROC, M_WAITOK);
 429         if (newproc->p_sigacts == NULL)
 430                 panic("forkproc: M_SUBPROC zone exhausted (p_sigacts)");
 431
 432         /*
 433          * Find an unused process ID.  We remember a range of unused IDs
 434          * ready to use (from nextpid+1 through pidchecked-1).
 435          */
 436         nextpid++;
 437 retry:
 438         /*
 439          * If the process ID prototype has wrapped around,
 440          * restart somewhat above 0, as the low-numbered procs
 441          * tend to include daemons that don't exit.
 442          */
 443         if (nextpid >= PID_MAX) {
 444                 nextpid = 100;
 445                 pidchecked = 0;
 446         }
 447         if (nextpid >= pidchecked) {
 448                 int doingzomb = 0;
 449
 450                 pidchecked = PID_MAX;
 451                 /*
 452                  * Scan the active and zombie procs to check whether this pid
 453                  * is in use.  Remember the lowest pid that's greater
 454                  * than nextpid, so we can avoid checking for a while.
 455                  */
 456                 p2 = allproc.lh_first;
 457 again:
 458                 for (; p2 != 0; p2 = p2->p_list.le_next) {
 459                         while (p2->p_pid == nextpid ||
 460                             p2->p_pgrp->pg_id == nextpid ||
 461                                 p2->p_session->s_sid == nextpid) {
 462                                 nextpid++;
 463                                 if (nextpid >= pidchecked)
 464                                         goto retry;
 465                         }
 466                         if (p2->p_pid > nextpid && pidchecked > p2->p_pid)
 467                                 pidchecked = p2->p_pid;
 468                         if (p2->p_pgrp && p2->p_pgrp->pg_id > nextpid &&
 469                             pidchecked > p2->p_pgrp->pg_id)
 470                                 pidchecked = p2->p_pgrp->pg_id;
 471                         if (p2->p_session->s_sid > nextpid &&
 472                                 pidchecked > p2->p_session->s_sid)
 473                                 pidchecked = p2->p_session->s_sid;
 474                 }
 475                 if (!doingzomb) {
 476                         doingzomb = 1;
 477                         p2 = zombproc.lh_first;
 478                         goto again;
 479                 }
 480         }
 481
 482         nprocs++;
 483         p2 = newproc;
 484         p2->p_stat = SIDL;
 485         p2->p_shutdownstate = 0;
 486         p2->p_pid = nextpid;
 487
 488         /*
 489          * Make a proc table entry for the new process.
 490          * Start by zeroing the section of proc that is zero-initialized,
 491          * then copy the section that is copied directly from the parent.
 492          */
 493         bzero(&p2->p_startzero,
 494             (unsigned) ((caddr_t)&p2->p_endzero - (caddr_t)&p2->p_startzero));
 495         bcopy(&p1->p_startcopy, &p2->p_startcopy,
 496             (unsigned) ((caddr_t)&p2->p_endcopy - (caddr_t)&p2->p_startcopy));
 497         p2->vm_shm = (void *)NULL; /* Make sure it is zero */
 498
 499         /*
 500          * Some flags are inherited from the parent.
 501          * Duplicate sub-structures as needed.
 502          * Increase reference counts on shared objects.
 503          * The p_stats and p_sigacts substructs are set in vm_fork.
 504          */
 505         p2->p_flag = (p1->p_flag & (P_LP64 | P_CLASSIC | P_AFFINITY));
 506         if (p1->p_flag & P_PROFIL)
 507                 startprofclock(p2);
 508         /*
 509          * Note that if the current thread has an assumed identity, this
 510          * credential will be granted to the new process.
 511          */
 512         p2->p_ucred = kauth_cred_get_with_ref();
 513
 514         lck_mtx_init(&p2->p_mlock, proc_lck_grp, proc_lck_attr);
 515         lck_mtx_init(&p2->p_fdmlock, proc_lck_grp, proc_lck_attr);
 516         klist_init(&p2->p_klist);
 517
 518         /* bump references to the text vnode */
 519         p2->p_textvp = p1->p_textvp;
 520         if (p2->p_textvp) {
 521                 vnode_rele(p2->p_textvp);
 522         }
 523         /* XXX may fail to copy descriptors to child */
 524         p2->p_fd = fdcopy(p1);
 525
 526         if (p1->vm_shm) {
 527                 /* XXX may fail to attach shm to child */
 528                 (void)shmfork(p1,p2);
 529         }
 530         /*
 531          * If p_limit is still copy-on-write, bump refcnt,
 532          * otherwise get a copy that won't be modified.
 533          * (If PL_SHAREMOD is clear, the structure is shared
 534          * copy-on-write.)
 535          */
 536         if (p1->p_limit->p_lflags & PL_SHAREMOD)
 537                 p2->p_limit = limcopy(p1->p_limit);
 538         else {
 539                 p2->p_limit = p1->p_limit;
 540                 p2->p_limit->p_refcnt++;
 541         }
 542
 543         bzero(&p2->p_stats->pstat_startzero,
 544             (unsigned) ((caddr_t)&p2->p_stats->pstat_endzero -
 545             (caddr_t)&p2->p_stats->pstat_startzero));
 546         bcopy(&p1->p_stats->pstat_startcopy, &p2->p_stats->pstat_startcopy,
 547             ((caddr_t)&p2->p_stats->pstat_endcopy -
 548              (caddr_t)&p2->p_stats->pstat_startcopy));
 549
 550         bzero(&p2->p_stats->user_p_prof, sizeof(struct user_uprof));
 551
 552         if (p1->p_sigacts != NULL)
 553                 (void)memcpy(p2->p_sigacts,
 554                                 p1->p_sigacts, sizeof *p2->p_sigacts);
 555         else
 556                 (void)memset(p2->p_sigacts, 0, sizeof *p2->p_sigacts);
 557
 558         if (p1->p_session->s_ttyvp != NULL && p1->p_flag & P_CONTROLT)
 559                 p2->p_flag |= P_CONTROLT;
 560
 561         p2->p_argslen = p1->p_argslen;
 562         p2->p_argc = p1->p_argc;
 563         p2->p_xstat = 0;
 564         p2->p_ru = NULL;
 565
 566         p2->p_debugger = 0;     /* don't inherit */
 567         lockinit(&p2->signal_lock, PVM, "signal", 0, 0);
 568         /* block all signals to reach the process */
 569         if (lock)
 570                 signal_lock(p2);
 571         p2->sigwait = FALSE;
 572         p2->sigwait_thread = NULL;
 573         p2->exit_thread = NULL;
 574         p2->user_stack = p1->user_stack;
 575         p2->p_vforkcnt = 0;
 576         p2->p_vforkact = 0;
 577         p2->p_lflag  = 0;
 578         p2->p_ladvflag  = 0;
 579         p2->p_internalref  = 0;
 580         TAILQ_INIT(&p2->p_uthlist);
 581         TAILQ_INIT(&p2->aio_activeq);
 582         TAILQ_INIT(&p2->aio_doneq);
 583         p2->aio_active_count = 0;
 584         p2->aio_done_count = 0;
 585
 586 #if KTRACE
 587         /*
 588          * Copy traceflag and tracefile if enabled.
 589          * If not inherited, these were zeroed above.
 590          */
 591         if (p1->p_traceflag&KTRFAC_INHERIT) {
 592                 p2->p_traceflag = p1->p_traceflag;
 593                 if ((p2->p_tracep = p1->p_tracep) != NULL) {
 594                         vnode_ref(p2->p_tracep);
 595                 }
 596         }
 597 #endif
 598         return(p2);
 599
 600 }
 601
 602 void
 603 proc_lock(proc_t p)
 604 {
 605         lck_mtx_lock(&p->p_mlock);
 606 }
 607
 608 void
 609 proc_unlock(proc_t p)
 610 {
 611         lck_mtx_unlock(&p->p_mlock);
 612 }
 613
 614 #include <kern/zalloc.h>
 615
 616 struct zone     *uthread_zone;
 617 int uthread_zone_inited = 0;
 618
 619 void
 620 uthread_zone_init(void)
 621 {
 622         if (!uthread_zone_inited) {
 623                 uthread_zone = zinit(sizeof(struct uthread),
 624                                         THREAD_MAX * sizeof(struct uthread),
 625                                         THREAD_CHUNK * sizeof(struct uthread),
 626                                         "uthreads");
 627                 uthread_zone_inited = 1;
 628         }
 629 }
 630
 631 void *
 632 uthread_alloc(task_t task, thread_t thr_act )
 633 {
 634         struct proc *p;
 635         struct uthread *uth, *uth_parent;
 636         void *ut;
 637         boolean_t funnel_state;
 638
 639         if (!uthread_zone_inited)
 640                 uthread_zone_init();
 641
 642         ut = (void *)zalloc(uthread_zone);
 643         bzero(ut, sizeof(struct uthread));
 644
 645         p = (struct proc *) get_bsdtask_info(task);
 646         uth = (struct uthread *)ut;
 647
 648         /*
 649          * Thread inherits credential from the creating thread, if both
 650          * are in the same task.
 651          *
 652          * If the creating thread has no credential or is from another
 653          * task we can leave the new thread credential NULL.  If it needs
 654          * one later, it will be lazily assigned from the task's process.
 655          */
 656         uth_parent = (struct uthread *)get_bsdthread_info(current_thread());
 657         if ((task == current_task()) &&
 658             (uth_parent != NULL) &&
 659             (uth_parent->uu_ucred != NOCRED)) {
 660                 uth->uu_ucred = uth_parent->uu_ucred;
 661                 kauth_cred_ref(uth->uu_ucred);
 662                 /* the credential we just inherited is an assumed credential */
 663                 if (uth_parent->uu_flag & UT_SETUID)
 664                         uth->uu_flag |= UT_SETUID;
 665         } else {
 666                 uth->uu_ucred = NOCRED;
 667         }
 668
 669         if (task != kernel_task) {
 670
 671                 funnel_state = thread_funnel_set(kernel_flock, TRUE);
 672                 if (uth_parent) {
 673                         if (uth_parent->uu_flag & UT_SAS_OLDMASK)
 674                                 uth->uu_sigmask = uth_parent->uu_oldmask;
 675                         else
 676                                 uth->uu_sigmask = uth_parent->uu_sigmask;
 677                 }
 678                 uth->uu_act = thr_act;
 679                 //signal_lock(p);
 680                 if (p) {
 681                         TAILQ_INSERT_TAIL(&p->p_uthlist, uth, uu_list);
 682                 }
 683                 //signal_unlock(p);
 684                 (void)thread_funnel_set(kernel_flock, funnel_state);
 685         }
 686
 687         return (ut);
 688 }
 689
 690
 691 void
 692 uthread_free(task_t task, void *uthread, void * bsd_info)
 693 {
 694         struct _select *sel;
 695         struct uthread *uth = (struct uthread *)uthread;
 696         struct proc * p = (struct proc *)bsd_info;
 697         boolean_t funnel_state;
 698
 699         /*
 700          * Per-thread audit state should never last beyond system
 701          * call return.  Since we don't audit the thread creation/
 702          * removal, the thread state pointer should never be
 703          * non-NULL when we get here.
 704          */
 705         assert(uth->uu_ar == NULL);
 706
 707         sel = &uth->uu_select;
 708         /* cleanup the select bit space */
 709         if (sel->nbytes) {
 710                 FREE(sel->ibits, M_TEMP);
 711                 FREE(sel->obits, M_TEMP);
 712         }
 713
 714         if (sel->allocsize && sel->wqset){
 715                 kfree(sel->wqset, sel->allocsize);
 716                 sel->count = 0;
 717                 sel->allocsize = 0;
 718                 sel->wqset = 0;
 719                 sel->wql = 0;
 720         }
 721
 722         if (uth->uu_ucred != NOCRED)
 723                 kauth_cred_rele(uth->uu_ucred);
 724
 725         if ((task != kernel_task) && p) {
 726                 funnel_state = thread_funnel_set(kernel_flock, TRUE);
 727                 //signal_lock(p);
 728                 TAILQ_REMOVE(&p->p_uthlist, uth, uu_list);
 729                 //signal_unlock(p);
 730                 (void)thread_funnel_set(kernel_flock, funnel_state);
 731         }
 732         /* and free the uthread itself */
 733         zfree(uthread_zone, uthread);
 734 }