2005-01-06 23:35:40 +00:00
|
|
|
/*-
|
1994-05-24 10:09:53 +00:00
|
|
|
* Copyright (c) 1982, 1986, 1989, 1991, 1993
|
|
|
|
* The Regents of the University of California. All rights reserved.
|
|
|
|
* (c) UNIX System Laboratories, Inc.
|
|
|
|
* All or some portions of this file are derived from material licensed
|
|
|
|
* to the University of California by American Telephone and Telegraph
|
|
|
|
* Co. or Unix System Laboratories, Inc. and are reproduced herein with
|
|
|
|
* the permission of UNIX System Laboratories, Inc.
|
|
|
|
*
|
|
|
|
* Redistribution and use in source and binary forms, with or without
|
|
|
|
* modification, are permitted provided that the following conditions
|
|
|
|
* are met:
|
|
|
|
* 1. Redistributions of source code must retain the above copyright
|
|
|
|
* notice, this list of conditions and the following disclaimer.
|
|
|
|
* 2. Redistributions in binary form must reproduce the above copyright
|
|
|
|
* notice, this list of conditions and the following disclaimer in the
|
|
|
|
* documentation and/or other materials provided with the distribution.
|
|
|
|
* 4. Neither the name of the University nor the names of its contributors
|
|
|
|
* may be used to endorse or promote products derived from this software
|
|
|
|
* without specific prior written permission.
|
|
|
|
*
|
|
|
|
* THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
|
|
|
|
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
|
|
|
|
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
|
|
|
|
* ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
|
|
|
|
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
|
|
|
|
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
|
|
|
|
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
|
|
|
|
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
|
|
|
|
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
|
|
|
|
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
|
|
|
|
* SUCH DAMAGE.
|
|
|
|
*
|
|
|
|
* @(#)kern_fork.c 8.6 (Berkeley) 4/8/94
|
|
|
|
*/
|
|
|
|
|
2003-06-11 00:56:59 +00:00
|
|
|
#include <sys/cdefs.h>
|
|
|
|
__FBSDID("$FreeBSD$");
|
|
|
|
|
1996-01-03 21:42:35 +00:00
|
|
|
#include "opt_ktrace.h"
|
2002-11-20 15:41:25 +00:00
|
|
|
#include "opt_mac.h"
|
1996-01-03 21:42:35 +00:00
|
|
|
|
1994-05-24 10:09:53 +00:00
|
|
|
#include <sys/param.h>
|
|
|
|
#include <sys/systm.h>
|
1995-11-12 06:43:28 +00:00
|
|
|
#include <sys/sysproto.h>
|
2003-03-24 21:15:35 +00:00
|
|
|
#include <sys/eventhandler.h>
|
1994-05-24 10:09:53 +00:00
|
|
|
#include <sys/filedesc.h>
|
|
|
|
#include <sys/kernel.h>
|
2003-08-15 21:29:06 +00:00
|
|
|
#include <sys/kthread.h>
|
1997-04-26 15:59:50 +00:00
|
|
|
#include <sys/sysctl.h>
|
Rework the witness code to work with sx locks as well as mutexes.
- Introduce lock classes and lock objects. Each lock class specifies a
name and set of flags (or properties) shared by all locks of a given
type. Currently there are three lock classes: spin mutexes, sleep
mutexes, and sx locks. A lock object specifies properties of an
additional lock along with a lock name and all of the extra stuff needed
to make witness work with a given lock. This abstract lock stuff is
defined in sys/lock.h. The lockmgr constants, types, and prototypes have
been moved to sys/lockmgr.h. For temporary backwards compatability,
sys/lock.h includes sys/lockmgr.h.
- Replace proc->p_spinlocks with a per-CPU list, PCPU(spinlocks), of spin
locks held. By making this per-cpu, we do not have to jump through
magic hoops to deal with sched_lock changing ownership during context
switches.
- Replace proc->p_heldmtx, formerly a list of held sleep mutexes, with
proc->p_sleeplocks, which is a list of held sleep locks including sleep
mutexes and sx locks.
- Add helper macros for logging lock events via the KTR_LOCK KTR logging
level so that the log messages are consistent.
- Add some new flags that can be passed to mtx_init():
- MTX_NOWITNESS - specifies that this lock should be ignored by witness.
This is used for the mutex that blocks a sx lock for example.
- MTX_QUIET - this is not new, but you can pass this to mtx_init() now
and no events will be logged for this lock, so that one doesn't have
to change all the individual mtx_lock/unlock() operations.
- All lock objects maintain an initialized flag. Use this flag to export
a mtx_initialized() macro that can be safely called from drivers. Also,
we on longer walk the all_mtx list if MUTEX_DEBUG is defined as witness
performs the corresponding checks using the initialized flag.
- The lock order reversal messages have been improved to output slightly
more accurate file and line numbers.
2001-03-28 09:03:24 +00:00
|
|
|
#include <sys/lock.h>
|
1994-05-24 10:09:53 +00:00
|
|
|
#include <sys/malloc.h>
|
2000-10-20 07:58:15 +00:00
|
|
|
#include <sys/mutex.h>
|
1994-05-24 10:09:53 +00:00
|
|
|
#include <sys/proc.h>
|
2002-08-04 01:07:02 +00:00
|
|
|
#include <sys/pioctl.h>
|
1994-05-24 10:09:53 +00:00
|
|
|
#include <sys/resourcevar.h>
|
2002-10-12 05:32:24 +00:00
|
|
|
#include <sys/sched.h>
|
2001-01-24 10:47:14 +00:00
|
|
|
#include <sys/syscall.h>
|
2003-08-15 21:29:06 +00:00
|
|
|
#include <sys/vmmeter.h>
|
1994-05-24 10:09:53 +00:00
|
|
|
#include <sys/vnode.h>
|
|
|
|
#include <sys/acct.h>
|
2002-11-20 15:41:25 +00:00
|
|
|
#include <sys/mac.h>
|
2000-09-07 01:33:02 +00:00
|
|
|
#include <sys/ktr.h>
|
1994-05-24 10:09:53 +00:00
|
|
|
#include <sys/ktrace.h>
|
1996-09-03 14:25:27 +00:00
|
|
|
#include <sys/unistd.h>
|
2001-03-07 02:30:39 +00:00
|
|
|
#include <sys/sx.h>
|
2004-11-27 06:51:39 +00:00
|
|
|
#include <sys/signalvar.h>
|
1994-05-24 10:09:53 +00:00
|
|
|
|
2006-02-02 01:32:58 +00:00
|
|
|
#include <security/audit/audit.h>
|
|
|
|
|
1994-10-10 01:00:49 +00:00
|
|
|
#include <vm/vm.h>
|
1996-02-23 18:49:25 +00:00
|
|
|
#include <vm/pmap.h>
|
|
|
|
#include <vm/vm_map.h>
|
1995-12-07 12:48:31 +00:00
|
|
|
#include <vm/vm_extern.h>
|
2002-03-20 04:09:59 +00:00
|
|
|
#include <vm/uma.h>
|
1994-10-10 01:00:49 +00:00
|
|
|
|
1999-01-26 02:38:12 +00:00
|
|
|
|
1995-11-12 06:43:28 +00:00
|
|
|
#ifndef _SYS_SYSPROTO_H_
|
1995-10-08 00:06:22 +00:00
|
|
|
struct fork_args {
|
1996-08-22 03:50:33 +00:00
|
|
|
int dummy;
|
1995-10-08 00:06:22 +00:00
|
|
|
};
|
1995-11-12 06:43:28 +00:00
|
|
|
#endif
|
1995-10-08 00:06:22 +00:00
|
|
|
|
2003-04-30 12:57:40 +00:00
|
|
|
static int forksleep; /* Place for fork1() to sleep on. */
|
2002-02-19 03:15:28 +00:00
|
|
|
|
2001-09-01 03:04:31 +00:00
|
|
|
/*
|
|
|
|
* MPSAFE
|
|
|
|
*/
|
1994-05-24 10:09:53 +00:00
|
|
|
/* ARGSUSED */
|
1994-05-25 09:21:21 +00:00
|
|
|
int
|
2001-09-12 08:38:13 +00:00
|
|
|
fork(td, uap)
|
|
|
|
struct thread *td;
|
1994-05-24 10:09:53 +00:00
|
|
|
struct fork_args *uap;
|
|
|
|
{
|
1999-06-30 15:33:41 +00:00
|
|
|
int error;
|
|
|
|
struct proc *p2;
|
1997-11-20 16:36:17 +00:00
|
|
|
|
2002-10-02 07:44:29 +00:00
|
|
|
error = fork1(td, RFFDG | RFPROC, 0, &p2);
|
1999-06-30 15:33:41 +00:00
|
|
|
if (error == 0) {
|
2001-09-12 08:38:13 +00:00
|
|
|
td->td_retval[0] = p2->p_pid;
|
|
|
|
td->td_retval[1] = 0;
|
1999-06-30 15:33:41 +00:00
|
|
|
}
|
2003-08-15 21:29:06 +00:00
|
|
|
return (error);
|
1994-05-24 10:09:53 +00:00
|
|
|
}
|
|
|
|
|
2001-09-01 03:04:31 +00:00
|
|
|
/*
|
|
|
|
* MPSAFE
|
|
|
|
*/
|
1994-05-24 10:09:53 +00:00
|
|
|
/* ARGSUSED */
|
1994-05-25 09:21:21 +00:00
|
|
|
int
|
2001-09-12 08:38:13 +00:00
|
|
|
vfork(td, uap)
|
|
|
|
struct thread *td;
|
1996-02-23 18:49:25 +00:00
|
|
|
struct vfork_args *uap;
|
1994-05-24 10:09:53 +00:00
|
|
|
{
|
1999-06-30 15:33:41 +00:00
|
|
|
int error;
|
|
|
|
struct proc *p2;
|
1997-11-20 16:36:17 +00:00
|
|
|
|
2002-10-02 07:44:29 +00:00
|
|
|
error = fork1(td, RFFDG | RFPROC | RFPPWAIT | RFMEM, 0, &p2);
|
1999-06-30 15:33:41 +00:00
|
|
|
if (error == 0) {
|
2001-09-12 08:38:13 +00:00
|
|
|
td->td_retval[0] = p2->p_pid;
|
|
|
|
td->td_retval[1] = 0;
|
1999-06-30 15:33:41 +00:00
|
|
|
}
|
2003-08-15 21:29:06 +00:00
|
|
|
return (error);
|
1996-02-23 18:49:25 +00:00
|
|
|
}
|
1994-05-24 10:09:53 +00:00
|
|
|
|
2001-09-01 03:04:31 +00:00
|
|
|
/*
|
|
|
|
* MPSAFE
|
|
|
|
*/
|
1996-02-23 18:49:25 +00:00
|
|
|
int
|
2001-09-12 08:38:13 +00:00
|
|
|
rfork(td, uap)
|
|
|
|
struct thread *td;
|
1996-02-23 18:49:25 +00:00
|
|
|
struct rfork_args *uap;
|
|
|
|
{
|
1999-06-30 15:33:41 +00:00
|
|
|
struct proc *p2;
|
2004-03-04 09:56:29 +00:00
|
|
|
int error;
|
1997-11-20 16:36:17 +00:00
|
|
|
|
2004-03-04 09:56:29 +00:00
|
|
|
/* Don't allow kernel-only flags. */
|
2001-12-19 00:53:23 +00:00
|
|
|
if ((uap->flags & RFKERNELONLY) != 0)
|
|
|
|
return (EINVAL);
|
2004-03-04 09:56:29 +00:00
|
|
|
|
2006-02-06 00:28:50 +00:00
|
|
|
AUDIT_ARG(fflags, uap->flags);
|
2002-10-02 07:44:29 +00:00
|
|
|
error = fork1(td, uap->flags, 0, &p2);
|
1999-06-30 15:33:41 +00:00
|
|
|
if (error == 0) {
|
2001-09-12 08:38:13 +00:00
|
|
|
td->td_retval[0] = p2 ? p2->p_pid : 0;
|
|
|
|
td->td_retval[1] = 0;
|
1999-06-30 15:33:41 +00:00
|
|
|
}
|
2003-08-15 21:29:06 +00:00
|
|
|
return (error);
|
1994-05-24 10:09:53 +00:00
|
|
|
}
|
|
|
|
|
2003-08-15 21:29:06 +00:00
|
|
|
int nprocs = 1; /* process 0 */
|
2001-06-11 21:54:19 +00:00
|
|
|
int lastpid = 0;
|
|
|
|
SYSCTL_INT(_kern, OID_AUTO, lastpid, CTLFLAG_RD, &lastpid, 0,
|
2001-02-12 17:59:01 +00:00
|
|
|
"Last used PID");
|
1994-05-24 10:09:53 +00:00
|
|
|
|
1999-12-06 11:13:50 +00:00
|
|
|
/*
|
2001-06-11 21:54:19 +00:00
|
|
|
* Random component to lastpid generation. We mix in a random factor to make
|
1999-12-06 11:13:50 +00:00
|
|
|
* it a little harder to predict. We sanity check the modulus value to avoid
|
|
|
|
* doing it in critical paths. Don't let it be too small or we pointlessly
|
|
|
|
* waste randomness entropy, and don't let it be impossibly large. Using a
|
|
|
|
* modulus that is too big causes a LOT more process table scans and slows
|
|
|
|
* down fork processing as the pidchecked caching is defeated.
|
|
|
|
*/
|
1999-11-28 17:51:09 +00:00
|
|
|
static int randompid = 0;
|
1999-12-06 11:13:50 +00:00
|
|
|
|
|
|
|
static int
|
2000-07-04 11:25:35 +00:00
|
|
|
sysctl_kern_randompid(SYSCTL_HANDLER_ARGS)
|
1999-12-06 11:13:50 +00:00
|
|
|
{
|
2000-12-04 10:23:29 +00:00
|
|
|
int error, pid;
|
|
|
|
|
2004-02-26 00:27:04 +00:00
|
|
|
error = sysctl_wire_old_buffer(req, sizeof(int));
|
|
|
|
if (error != 0)
|
|
|
|
return(error);
|
2002-05-02 15:13:45 +00:00
|
|
|
sx_xlock(&allproc_lock);
|
2000-12-04 10:23:29 +00:00
|
|
|
pid = randompid;
|
|
|
|
error = sysctl_handle_int(oidp, &pid, 0, req);
|
2002-05-02 15:13:45 +00:00
|
|
|
if (error == 0 && req->newptr != NULL) {
|
|
|
|
if (pid < 0 || pid > PID_MAX - 100) /* out of range */
|
|
|
|
pid = PID_MAX - 100;
|
|
|
|
else if (pid < 2) /* NOP */
|
|
|
|
pid = 0;
|
|
|
|
else if (pid < 100) /* Make it reasonable */
|
|
|
|
pid = 100;
|
|
|
|
randompid = pid;
|
|
|
|
}
|
|
|
|
sx_xunlock(&allproc_lock);
|
2000-12-04 10:23:29 +00:00
|
|
|
return (error);
|
1999-12-06 11:13:50 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
SYSCTL_PROC(_kern, OID_AUTO, randompid, CTLTYPE_INT|CTLFLAG_RW,
|
|
|
|
0, 0, sysctl_kern_randompid, "I", "Random PID modulus");
|
1999-11-28 17:51:09 +00:00
|
|
|
|
1997-12-12 04:00:59 +00:00
|
|
|
int
|
2002-10-02 07:44:29 +00:00
|
|
|
fork1(td, flags, pages, procp)
|
2003-08-15 21:29:06 +00:00
|
|
|
struct thread *td;
|
1996-04-17 17:05:08 +00:00
|
|
|
int flags;
|
2002-10-02 07:44:29 +00:00
|
|
|
int pages;
|
2003-08-15 21:29:06 +00:00
|
|
|
struct proc **procp;
|
1994-05-24 10:09:53 +00:00
|
|
|
{
|
2003-08-15 21:29:06 +00:00
|
|
|
struct proc *p1, *p2, *pptr;
|
1994-05-24 10:09:53 +00:00
|
|
|
struct proc *newproc;
|
2003-08-15 21:29:06 +00:00
|
|
|
int ok, trypid;
|
2003-06-19 05:57:25 +00:00
|
|
|
static int curfail, pidchecked = 0;
|
|
|
|
static struct timeval lastfail;
|
2001-03-07 05:21:47 +00:00
|
|
|
struct filedesc *fd;
|
2003-06-02 16:05:32 +00:00
|
|
|
struct filedesc_to_leader *fdtol;
|
2002-02-07 20:58:47 +00:00
|
|
|
struct thread *td2;
|
|
|
|
struct ksegrp *kg2;
|
2002-05-02 15:13:45 +00:00
|
|
|
struct sigacts *newsigacts;
|
2002-10-15 00:14:32 +00:00
|
|
|
int error;
|
1997-04-13 01:48:35 +00:00
|
|
|
|
2003-08-15 21:29:06 +00:00
|
|
|
/* Can't copy and clear. */
|
1996-04-17 17:05:08 +00:00
|
|
|
if ((flags & (RFFDG|RFCFDG)) == (RFFDG|RFCFDG))
|
|
|
|
return (EINVAL);
|
1994-05-24 10:09:53 +00:00
|
|
|
|
2003-08-15 21:29:06 +00:00
|
|
|
p1 = td->td_proc;
|
|
|
|
|
1997-04-13 01:48:35 +00:00
|
|
|
/*
|
|
|
|
* Here we don't create a new process, but we divorce
|
|
|
|
* certain parts of a process from itself.
|
|
|
|
*/
|
|
|
|
if ((flags & RFPROC) == 0) {
|
2006-03-15 23:24:14 +00:00
|
|
|
if ((p1->p_flag & P_HADTHREADS) &&
|
|
|
|
(flags & (RFCFDG | RFFDG))) {
|
|
|
|
PROC_LOCK(p1);
|
|
|
|
if (thread_single(SINGLE_BOUNDARY)) {
|
|
|
|
PROC_UNLOCK(p1);
|
|
|
|
return (ERESTART);
|
|
|
|
}
|
|
|
|
PROC_UNLOCK(p1);
|
|
|
|
}
|
|
|
|
|
2002-02-07 20:58:47 +00:00
|
|
|
vm_forkproc(td, NULL, NULL, flags);
|
1997-04-13 01:48:35 +00:00
|
|
|
|
|
|
|
/*
|
|
|
|
* Close all file descriptors.
|
|
|
|
*/
|
|
|
|
if (flags & RFCFDG) {
|
|
|
|
struct filedesc *fdtmp;
|
2003-01-01 01:01:14 +00:00
|
|
|
fdtmp = fdinit(td->td_proc->p_fd);
|
|
|
|
fdfree(td);
|
1997-04-13 01:48:35 +00:00
|
|
|
p1->p_fd = fdtmp;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
2004-03-04 09:56:29 +00:00
|
|
|
* Unshare file descriptors (from parent).
|
1997-04-13 01:48:35 +00:00
|
|
|
*/
|
2004-12-14 07:20:03 +00:00
|
|
|
if (flags & RFFDG)
|
|
|
|
fdunshare(p1, td);
|
2006-03-15 23:24:14 +00:00
|
|
|
|
|
|
|
if ((p1->p_flag & P_HADTHREADS) &&
|
|
|
|
(flags & (RFCFDG | RFFDG))) {
|
|
|
|
PROC_LOCK(p1);
|
|
|
|
thread_single_end();
|
|
|
|
PROC_UNLOCK(p1);
|
|
|
|
}
|
1999-07-03 20:58:44 +00:00
|
|
|
*procp = NULL;
|
1997-04-13 01:48:35 +00:00
|
|
|
return (0);
|
|
|
|
}
|
|
|
|
|
2003-04-01 01:26:20 +00:00
|
|
|
/*
|
|
|
|
* Note 1:1 allows for forking with one thread coming out on the
|
|
|
|
* other side with the expectation that the process is about to
|
|
|
|
* exec.
|
|
|
|
*/
|
2004-09-13 22:10:04 +00:00
|
|
|
if (p1->p_flag & P_HADTHREADS) {
|
Part 1 of KSE-III
The ability to schedule multiple threads per process
(one one cpu) by making ALL system calls optionally asynchronous.
to come: ia64 and power-pc patches, patches for gdb, test program (in tools)
Reviewed by: Almost everyone who counts
(at various times, peter, jhb, matt, alfred, mini, bernd,
and a cast of thousands)
NOTE: this is still Beta code, and contains lots of debugging stuff.
expect slight instability in signals..
2002-06-29 17:26:22 +00:00
|
|
|
/*
|
|
|
|
* Idle the other threads for a second.
|
|
|
|
* Since the user space is copied, it must remain stable.
|
|
|
|
* In addition, all threads (from the user perspective)
|
|
|
|
* need to either be suspended or in the kernel,
|
|
|
|
* where they will try restart in the parent and will
|
|
|
|
* be aborted in the child.
|
|
|
|
*/
|
|
|
|
PROC_LOCK(p1);
|
2002-09-05 07:30:18 +00:00
|
|
|
if (thread_single(SINGLE_NO_EXIT)) {
|
2004-03-04 09:56:29 +00:00
|
|
|
/* Abort. Someone else is single threading before us. */
|
Part 1 of KSE-III
The ability to schedule multiple threads per process
(one one cpu) by making ALL system calls optionally asynchronous.
to come: ia64 and power-pc patches, patches for gdb, test program (in tools)
Reviewed by: Almost everyone who counts
(at various times, peter, jhb, matt, alfred, mini, bernd,
and a cast of thousands)
NOTE: this is still Beta code, and contains lots of debugging stuff.
expect slight instability in signals..
2002-06-29 17:26:22 +00:00
|
|
|
PROC_UNLOCK(p1);
|
|
|
|
return (ERESTART);
|
|
|
|
}
|
|
|
|
PROC_UNLOCK(p1);
|
|
|
|
/*
|
|
|
|
* All other activity in this process
|
|
|
|
* is now suspended at the user boundary,
|
|
|
|
* (or other safe places if we think of any).
|
|
|
|
*/
|
|
|
|
}
|
|
|
|
|
2002-05-02 15:13:45 +00:00
|
|
|
/* Allocate new proc. */
|
2003-02-19 05:47:46 +00:00
|
|
|
newproc = uma_zalloc(proc_zone, M_WAITOK);
|
2002-11-20 15:41:25 +00:00
|
|
|
#ifdef MAC
|
|
|
|
mac_init_proc(newproc);
|
2006-02-02 01:32:58 +00:00
|
|
|
#endif
|
|
|
|
#ifdef AUDIT
|
|
|
|
audit_proc_alloc(newproc);
|
2002-11-20 15:41:25 +00:00
|
|
|
#endif
|
2005-07-01 16:28:32 +00:00
|
|
|
knlist_init(&newproc->p_klist, &newproc->p_mtx, NULL, NULL, NULL);
|
Moderate rewrite of kernel ktrace code to attempt to generally improve
reliability when tracing fast-moving processes or writing traces to
slow file systems by avoiding unbounded queueuing and dropped records.
Record loss was previously possible when the global pool of records
become depleted as a result of record generation outstripping record
commit, which occurred quickly in many common situations.
These changes partially restore the 4.x model of committing ktrace
records at the point of trace generation (synchronous), but maintain
the 5.x deferred record commit behavior (asynchronous) for situations
where entering VFS and sleeping is not possible (i.e., in the
scheduler). Records are now queued per-process as opposed to
globally, with processes responsible for committing records from their
own context as required.
- Eliminate the ktrace worker thread and global record queue, as they
are no longer used. Keep the global free record list, as records
are still used.
- Add a per-process record queue, which will hold any asynchronously
generated records, such as from context switches. This replaces the
global queue as the place to submit asynchronous records to.
- When a record is committed asynchronously, simply queue it to the
process.
- When a record is committed synchronously, first drain any pending
per-process records in order to maintain ordering as best we can.
Currently ordering between competing threads is provided via a global
ktrace_sx, but a per-process flag or lock may be desirable in the
future.
- When a process returns to user space following a system call, trap,
signal delivery, etc, flush any pending records.
- When a process exits, flush any pending records.
- Assert on process tear-down that there are no pending records.
- Slightly abstract the notion of being "in ktrace", which is used to
prevent the recursive generation of records, as well as generating
traces for ktrace events.
Future work here might look at changing the set of events marked for
synchronous and asynchronous record generation, re-balancing queue
depth, timeliness of commit to disk, and so on. I.e., performing a
drain every (n) records.
MFC after: 1 month
Discussed with: jhb
Requested by: Marc Olzheim <marcolz at stack dot nl>
2005-11-13 13:27:44 +00:00
|
|
|
STAILQ_INIT(&newproc->p_ktr);
|
2002-05-02 15:13:45 +00:00
|
|
|
|
2004-03-05 22:37:32 +00:00
|
|
|
/* We have to lock the process tree while we look for a pid. */
|
|
|
|
sx_slock(&proctree_lock);
|
|
|
|
|
1994-05-24 10:09:53 +00:00
|
|
|
/*
|
|
|
|
* Although process entries are dynamically created, we still keep
|
|
|
|
* a global limit on the maximum number we will create. Don't allow
|
2002-07-30 05:37:00 +00:00
|
|
|
* a nonprivileged user to use the last ten processes; don't let root
|
1994-05-24 10:09:53 +00:00
|
|
|
* exceed the limit. The variable nprocs is the current number of
|
|
|
|
* processes, maxproc is the limit.
|
|
|
|
*/
|
2002-05-02 15:13:45 +00:00
|
|
|
sx_xlock(&allproc_lock);
|
2004-07-26 07:54:39 +00:00
|
|
|
if ((nprocs >= maxproc - 10 &&
|
|
|
|
suser_cred(td->td_ucred, SUSER_RUID) != 0) ||
|
2004-07-13 13:10:07 +00:00
|
|
|
nprocs >= maxproc) {
|
2002-10-15 00:14:32 +00:00
|
|
|
error = EAGAIN;
|
|
|
|
goto fail;
|
1994-05-24 10:09:53 +00:00
|
|
|
}
|
2002-10-15 00:14:32 +00:00
|
|
|
|
1994-05-24 10:09:53 +00:00
|
|
|
/*
|
|
|
|
* Increment the count of procs running with this uid. Don't allow
|
|
|
|
* a nonprivileged user to exceed their current limit.
|
|
|
|
*/
|
2006-06-27 11:28:50 +00:00
|
|
|
error = suser_cred(td->td_ucred, SUSER_RUID | SUSER_ALLOWJAIL);
|
|
|
|
if (error == 0)
|
|
|
|
ok = chgproccnt(td->td_ucred->cr_ruidinfo, 1, 0);
|
|
|
|
else {
|
|
|
|
PROC_LOCK(p1);
|
|
|
|
ok = chgproccnt(td->td_ucred->cr_ruidinfo, 1,
|
|
|
|
lim_cur(p1, RLIMIT_NPROC));
|
|
|
|
PROC_UNLOCK(p1);
|
|
|
|
}
|
2000-09-14 23:07:39 +00:00
|
|
|
if (!ok) {
|
2002-10-15 00:14:32 +00:00
|
|
|
error = EAGAIN;
|
|
|
|
goto fail;
|
1994-05-24 10:09:53 +00:00
|
|
|
}
|
|
|
|
|
1999-06-30 15:33:41 +00:00
|
|
|
/*
|
2002-05-02 15:13:45 +00:00
|
|
|
* Increment the nprocs resource before blocking can occur. There
|
|
|
|
* are hard-limits as to the number of processes that can run.
|
1999-06-30 15:33:41 +00:00
|
|
|
*/
|
2002-05-02 15:13:45 +00:00
|
|
|
nprocs++;
|
1999-07-21 18:02:27 +00:00
|
|
|
|
1994-05-24 10:09:53 +00:00
|
|
|
/*
|
|
|
|
* Find an unused process ID. We remember a range of unused IDs
|
2001-06-11 21:54:19 +00:00
|
|
|
* ready to use (from lastpid+1 through pidchecked-1).
|
2000-09-07 01:33:02 +00:00
|
|
|
*
|
|
|
|
* If RFHIGHPID is set (used during system boot), do not allocate
|
|
|
|
* low-numbered pids.
|
1994-05-24 10:09:53 +00:00
|
|
|
*/
|
2001-06-11 21:54:19 +00:00
|
|
|
trypid = lastpid + 1;
|
2000-09-07 01:33:02 +00:00
|
|
|
if (flags & RFHIGHPID) {
|
2003-08-15 21:29:06 +00:00
|
|
|
if (trypid < 10)
|
2000-09-07 01:33:02 +00:00
|
|
|
trypid = 10;
|
|
|
|
} else {
|
2000-12-04 10:23:29 +00:00
|
|
|
if (randompid)
|
2000-09-07 01:33:02 +00:00
|
|
|
trypid += arc4random() % randompid;
|
|
|
|
}
|
1994-05-24 10:09:53 +00:00
|
|
|
retry:
|
|
|
|
/*
|
|
|
|
* If the process ID prototype has wrapped around,
|
|
|
|
* restart somewhat above 0, as the low-numbered procs
|
|
|
|
* tend to include daemons that don't exit.
|
|
|
|
*/
|
2000-09-07 01:33:02 +00:00
|
|
|
if (trypid >= PID_MAX) {
|
|
|
|
trypid = trypid % PID_MAX;
|
|
|
|
if (trypid < 100)
|
|
|
|
trypid += 100;
|
1994-05-24 10:09:53 +00:00
|
|
|
pidchecked = 0;
|
|
|
|
}
|
2000-09-07 01:33:02 +00:00
|
|
|
if (trypid >= pidchecked) {
|
1994-05-24 10:09:53 +00:00
|
|
|
int doingzomb = 0;
|
|
|
|
|
|
|
|
pidchecked = PID_MAX;
|
|
|
|
/*
|
|
|
|
* Scan the active and zombie procs to check whether this pid
|
|
|
|
* is in use. Remember the lowest pid that's greater
|
2000-09-07 01:33:02 +00:00
|
|
|
* than trypid, so we can avoid checking for a while.
|
1994-05-24 10:09:53 +00:00
|
|
|
*/
|
1999-11-16 10:56:05 +00:00
|
|
|
p2 = LIST_FIRST(&allproc);
|
1994-05-24 10:09:53 +00:00
|
|
|
again:
|
2001-01-24 10:47:14 +00:00
|
|
|
for (; p2 != NULL; p2 = LIST_NEXT(p2, p_list)) {
|
2000-09-07 01:33:02 +00:00
|
|
|
while (p2->p_pid == trypid ||
|
2004-03-05 22:37:32 +00:00
|
|
|
(p2->p_pgrp != NULL &&
|
|
|
|
(p2->p_pgrp->pg_id == trypid ||
|
|
|
|
(p2->p_session != NULL &&
|
|
|
|
p2->p_session->s_sid == trypid)))) {
|
2000-09-07 01:33:02 +00:00
|
|
|
trypid++;
|
2006-08-01 15:30:56 +00:00
|
|
|
if (trypid >= pidchecked)
|
1994-05-24 10:09:53 +00:00
|
|
|
goto retry;
|
|
|
|
}
|
2000-09-07 01:33:02 +00:00
|
|
|
if (p2->p_pid > trypid && pidchecked > p2->p_pid)
|
1994-05-24 10:09:53 +00:00
|
|
|
pidchecked = p2->p_pid;
|
2004-03-05 22:37:32 +00:00
|
|
|
if (p2->p_pgrp != NULL) {
|
|
|
|
if (p2->p_pgrp->pg_id > trypid &&
|
|
|
|
pidchecked > p2->p_pgrp->pg_id)
|
|
|
|
pidchecked = p2->p_pgrp->pg_id;
|
|
|
|
if (p2->p_session != NULL &&
|
|
|
|
p2->p_session->s_sid > trypid &&
|
|
|
|
pidchecked > p2->p_session->s_sid)
|
|
|
|
pidchecked = p2->p_session->s_sid;
|
|
|
|
}
|
1994-05-24 10:09:53 +00:00
|
|
|
}
|
|
|
|
if (!doingzomb) {
|
|
|
|
doingzomb = 1;
|
1999-11-16 10:56:05 +00:00
|
|
|
p2 = LIST_FIRST(&zombproc);
|
1994-05-24 10:09:53 +00:00
|
|
|
goto again;
|
|
|
|
}
|
|
|
|
}
|
2004-03-05 22:37:32 +00:00
|
|
|
sx_sunlock(&proctree_lock);
|
1994-05-24 10:09:53 +00:00
|
|
|
|
2000-09-07 01:33:02 +00:00
|
|
|
/*
|
2001-06-11 21:54:19 +00:00
|
|
|
* RFHIGHPID does not mess with the lastpid counter during boot.
|
2000-09-07 01:33:02 +00:00
|
|
|
*/
|
|
|
|
if (flags & RFHIGHPID)
|
|
|
|
pidchecked = 0;
|
|
|
|
else
|
2001-06-11 21:54:19 +00:00
|
|
|
lastpid = trypid;
|
2000-09-07 01:33:02 +00:00
|
|
|
|
2000-11-22 07:42:04 +00:00
|
|
|
p2 = newproc;
|
Part 1 of KSE-III
The ability to schedule multiple threads per process
(one one cpu) by making ALL system calls optionally asynchronous.
to come: ia64 and power-pc patches, patches for gdb, test program (in tools)
Reviewed by: Almost everyone who counts
(at various times, peter, jhb, matt, alfred, mini, bernd,
and a cast of thousands)
NOTE: this is still Beta code, and contains lots of debugging stuff.
expect slight instability in signals..
2002-06-29 17:26:22 +00:00
|
|
|
p2->p_state = PRS_NEW; /* protect against others */
|
2000-11-22 07:42:04 +00:00
|
|
|
p2->p_pid = trypid;
|
2006-02-06 00:28:50 +00:00
|
|
|
AUDIT_ARG(pid, p2->p_pid);
|
2000-11-22 07:42:04 +00:00
|
|
|
LIST_INSERT_HEAD(&allproc, p2, p_list);
|
|
|
|
LIST_INSERT_HEAD(PIDHASH(p2->p_pid), p2, p_hash);
|
2001-03-28 11:52:56 +00:00
|
|
|
sx_xunlock(&allproc_lock);
|
2000-11-22 07:42:04 +00:00
|
|
|
|
2002-05-02 15:13:45 +00:00
|
|
|
/*
|
|
|
|
* Malloc things while we don't hold any locks.
|
|
|
|
*/
|
- Merge struct procsig with struct sigacts.
- Move struct sigacts out of the u-area and malloc() it using the
M_SUBPROC malloc bucket.
- Add a small sigacts_*() API for managing sigacts structures: sigacts_alloc(),
sigacts_free(), sigacts_copy(), sigacts_share(), and sigacts_shared().
- Remove the p_sigignore, p_sigacts, and p_sigcatch macros.
- Add a mutex to struct sigacts that protects all the members of the struct.
- Add sigacts locking.
- Remove Giant from nosys(), kill(), killpg(), and kern_sigaction() now
that sigacts is locked.
- Several in-kernel functions such as psignal(), tdsignal(), trapsignal(),
and thread_stopped() are now MP safe.
Reviewed by: arch@
Approved by: re (rwatson)
2003-05-13 20:36:02 +00:00
|
|
|
if (flags & RFSIGSHARE)
|
2002-05-02 15:13:45 +00:00
|
|
|
newsigacts = NULL;
|
- Merge struct procsig with struct sigacts.
- Move struct sigacts out of the u-area and malloc() it using the
M_SUBPROC malloc bucket.
- Add a small sigacts_*() API for managing sigacts structures: sigacts_alloc(),
sigacts_free(), sigacts_copy(), sigacts_share(), and sigacts_shared().
- Remove the p_sigignore, p_sigacts, and p_sigcatch macros.
- Add a mutex to struct sigacts that protects all the members of the struct.
- Add sigacts locking.
- Remove Giant from nosys(), kill(), killpg(), and kern_sigaction() now
that sigacts is locked.
- Several in-kernel functions such as psignal(), tdsignal(), trapsignal(),
and thread_stopped() are now MP safe.
Reviewed by: arch@
Approved by: re (rwatson)
2003-05-13 20:36:02 +00:00
|
|
|
else
|
|
|
|
newsigacts = sigacts_alloc();
|
2002-05-02 15:13:45 +00:00
|
|
|
|
|
|
|
/*
|
|
|
|
* Copy filedesc.
|
|
|
|
*/
|
2003-06-02 16:05:32 +00:00
|
|
|
if (flags & RFCFDG) {
|
2004-11-08 12:43:23 +00:00
|
|
|
fd = fdinit(p1->p_fd);
|
2003-06-02 16:05:32 +00:00
|
|
|
fdtol = NULL;
|
|
|
|
} else if (flags & RFFDG) {
|
2004-11-08 12:43:23 +00:00
|
|
|
fd = fdcopy(p1->p_fd);
|
2003-06-02 16:05:32 +00:00
|
|
|
fdtol = NULL;
|
|
|
|
} else {
|
2003-01-01 01:01:14 +00:00
|
|
|
fd = fdshare(p1->p_fd);
|
2003-06-02 16:05:32 +00:00
|
|
|
if (p1->p_fdtol == NULL)
|
|
|
|
p1->p_fdtol =
|
|
|
|
filedesc_to_leader_alloc(NULL,
|
|
|
|
NULL,
|
|
|
|
p1->p_leader);
|
|
|
|
if ((flags & RFTHREAD) != 0) {
|
|
|
|
/*
|
|
|
|
* Shared file descriptor table and
|
|
|
|
* shared process leaders.
|
|
|
|
*/
|
|
|
|
fdtol = p1->p_fdtol;
|
2004-11-13 11:53:02 +00:00
|
|
|
FILEDESC_LOCK_FAST(p1->p_fd);
|
2003-06-02 16:05:32 +00:00
|
|
|
fdtol->fdl_refcount++;
|
2004-11-13 11:53:02 +00:00
|
|
|
FILEDESC_UNLOCK_FAST(p1->p_fd);
|
2003-06-02 16:05:32 +00:00
|
|
|
} else {
|
|
|
|
/*
|
|
|
|
* Shared file descriptor table, and
|
|
|
|
* different process leaders
|
|
|
|
*/
|
|
|
|
fdtol = filedesc_to_leader_alloc(p1->p_fdtol,
|
|
|
|
p1->p_fd,
|
|
|
|
p2);
|
|
|
|
}
|
|
|
|
}
|
1994-05-24 10:09:53 +00:00
|
|
|
/*
|
|
|
|
* Make a proc table entry for the new process.
|
|
|
|
* Start by zeroing the section of proc that is zero-initialized,
|
|
|
|
* then copy the section that is copied directly from the parent.
|
|
|
|
*/
|
2002-09-06 07:00:37 +00:00
|
|
|
td2 = FIRST_THREAD_IN_PROC(p2);
|
|
|
|
kg2 = FIRST_KSEGRP_IN_PROC(p2);
|
2001-09-12 08:38:13 +00:00
|
|
|
|
2004-03-04 09:56:29 +00:00
|
|
|
/* Allocate and switch to an alternate kstack if specified. */
|
2004-03-07 00:06:32 +00:00
|
|
|
if (pages != 0)
|
2003-06-14 06:20:25 +00:00
|
|
|
vm_thread_new_altkstack(td2, pages);
|
2002-10-02 07:44:29 +00:00
|
|
|
|
2003-05-01 21:16:38 +00:00
|
|
|
PROC_LOCK(p2);
|
|
|
|
PROC_LOCK(p1);
|
|
|
|
|
2002-02-07 20:58:47 +00:00
|
|
|
bzero(&p2->p_startzero,
|
2004-11-20 23:00:59 +00:00
|
|
|
__rangeof(struct proc, p_startzero, p_endzero));
|
2002-02-07 20:58:47 +00:00
|
|
|
bzero(&td2->td_startzero,
|
2004-11-20 23:00:59 +00:00
|
|
|
__rangeof(struct thread, td_startzero, td_endzero));
|
2002-02-07 20:58:47 +00:00
|
|
|
bzero(&kg2->kg_startzero,
|
2004-11-20 23:00:59 +00:00
|
|
|
__rangeof(struct ksegrp, kg_startzero, kg_endzero));
|
2001-09-12 08:38:13 +00:00
|
|
|
|
2002-02-07 20:58:47 +00:00
|
|
|
bcopy(&p1->p_startcopy, &p2->p_startcopy,
|
2004-11-20 23:00:59 +00:00
|
|
|
__rangeof(struct proc, p_startcopy, p_endcopy));
|
2002-02-07 20:58:47 +00:00
|
|
|
bcopy(&td->td_startcopy, &td2->td_startcopy,
|
2004-11-20 23:00:59 +00:00
|
|
|
__rangeof(struct thread, td_startcopy, td_endcopy));
|
2002-02-07 20:58:47 +00:00
|
|
|
bcopy(&td->td_ksegrp->kg_startcopy, &kg2->kg_startcopy,
|
2004-11-20 23:00:59 +00:00
|
|
|
__rangeof(struct ksegrp, kg_startcopy, kg_endcopy));
|
1994-05-24 10:09:53 +00:00
|
|
|
|
2004-01-03 02:02:26 +00:00
|
|
|
td2->td_sigstk = td->td_sigstk;
|
2005-04-20 13:14:52 +00:00
|
|
|
td2->td_sigmask = td->td_sigmask;
|
2004-01-03 02:02:26 +00:00
|
|
|
|
1994-05-24 10:09:53 +00:00
|
|
|
/*
|
|
|
|
* Duplicate sub-structures as needed.
|
|
|
|
* Increase reference counts on shared objects.
|
|
|
|
*/
|
2001-01-24 10:47:14 +00:00
|
|
|
p2->p_flag = 0;
|
2003-04-22 20:54:04 +00:00
|
|
|
if (p1->p_flag & P_PROFIL)
|
|
|
|
startprofclock(p2);
|
Change and clean the mutex lock interface.
mtx_enter(lock, type) becomes:
mtx_lock(lock) for sleep locks (MTX_DEF-initialized locks)
mtx_lock_spin(lock) for spin locks (MTX_SPIN-initialized)
similarily, for releasing a lock, we now have:
mtx_unlock(lock) for MTX_DEF and mtx_unlock_spin(lock) for MTX_SPIN.
We change the caller interface for the two different types of locks
because the semantics are entirely different for each case, and this
makes it explicitly clear and, at the same time, it rids us of the
extra `type' argument.
The enter->lock and exit->unlock change has been made with the idea
that we're "locking data" and not "entering locked code" in mind.
Further, remove all additional "flags" previously passed to the
lock acquire/release routines with the exception of two:
MTX_QUIET and MTX_NOSWITCH
The functionality of these flags is preserved and they can be passed
to the lock/unlock routines by calling the corresponding wrappers:
mtx_{lock, unlock}_flags(lock, flag(s)) and
mtx_{lock, unlock}_spin_flags(lock, flag(s)) for MTX_DEF and MTX_SPIN
locks, respectively.
Re-inline some lock acq/rel code; in the sleep lock case, we only
inline the _obtain_lock()s in order to ensure that the inlined code
fits into a cache line. In the spin lock case, we inline recursion and
actually only perform a function call if we need to spin. This change
has been made with the idea that we generally tend to avoid spin locks
and that also the spin locks that we do have and are heavily used
(i.e. sched_lock) do recurse, and therefore in an effort to reduce
function call overhead for some architectures (such as alpha), we
inline recursion for this case.
Create a new malloc type for the witness code and retire from using
the M_DEV type. The new type is called M_WITNESS and is only declared
if WITNESS is enabled.
Begin cleaning up some machdep/mutex.h code - specifically updated the
"optimized" inlined code in alpha/mutex.h and wrote MTX_LOCK_SPIN
and MTX_UNLOCK_SPIN asm macros for the i386/mutex.h as we presently
need those.
Finally, caught up to the interface changes in all sys code.
Contributors: jake, jhb, jasone (in no particular order)
2001-02-09 06:11:45 +00:00
|
|
|
mtx_lock_spin(&sched_lock);
|
2001-01-24 10:47:14 +00:00
|
|
|
p2->p_sflag = PS_INMEM;
|
2002-10-12 05:32:24 +00:00
|
|
|
/*
|
|
|
|
* Allow the scheduler to adjust the priority of the child and
|
|
|
|
* parent while we hold the sched_lock.
|
|
|
|
*/
|
2004-09-05 02:09:54 +00:00
|
|
|
sched_fork(td, td2);
|
2002-10-12 05:32:24 +00:00
|
|
|
|
Change and clean the mutex lock interface.
mtx_enter(lock, type) becomes:
mtx_lock(lock) for sleep locks (MTX_DEF-initialized locks)
mtx_lock_spin(lock) for spin locks (MTX_SPIN-initialized)
similarily, for releasing a lock, we now have:
mtx_unlock(lock) for MTX_DEF and mtx_unlock_spin(lock) for MTX_SPIN.
We change the caller interface for the two different types of locks
because the semantics are entirely different for each case, and this
makes it explicitly clear and, at the same time, it rids us of the
extra `type' argument.
The enter->lock and exit->unlock change has been made with the idea
that we're "locking data" and not "entering locked code" in mind.
Further, remove all additional "flags" previously passed to the
lock acquire/release routines with the exception of two:
MTX_QUIET and MTX_NOSWITCH
The functionality of these flags is preserved and they can be passed
to the lock/unlock routines by calling the corresponding wrappers:
mtx_{lock, unlock}_flags(lock, flag(s)) and
mtx_{lock, unlock}_spin_flags(lock, flag(s)) for MTX_DEF and MTX_SPIN
locks, respectively.
Re-inline some lock acq/rel code; in the sleep lock case, we only
inline the _obtain_lock()s in order to ensure that the inlined code
fits into a cache line. In the spin lock case, we inline recursion and
actually only perform a function call if we need to spin. This change
has been made with the idea that we generally tend to avoid spin locks
and that also the spin locks that we do have and are heavily used
(i.e. sched_lock) do recurse, and therefore in an effort to reduce
function call overhead for some architectures (such as alpha), we
inline recursion for this case.
Create a new malloc type for the witness code and retire from using
the M_DEV type. The new type is called M_WITNESS and is only declared
if WITNESS is enabled.
Begin cleaning up some machdep/mutex.h code - specifically updated the
"optimized" inlined code in alpha/mutex.h and wrote MTX_LOCK_SPIN
and MTX_UNLOCK_SPIN asm macros for the i386/mutex.h as we presently
need those.
Finally, caught up to the interface changes in all sys code.
Contributors: jake, jhb, jasone (in no particular order)
2001-02-09 06:11:45 +00:00
|
|
|
mtx_unlock_spin(&sched_lock);
|
2002-05-02 15:13:45 +00:00
|
|
|
p2->p_ucred = crhold(td->td_ucred);
|
2002-02-07 20:58:47 +00:00
|
|
|
td2->td_ucred = crhold(p2->p_ucred); /* XXXKSE */
|
2006-02-02 01:32:58 +00:00
|
|
|
#ifdef AUDIT
|
|
|
|
audit_proc_fork(p1, p2);
|
|
|
|
#endif
|
2002-03-27 21:36:18 +00:00
|
|
|
pargs_hold(p2->p_args);
|
1999-11-16 20:31:58 +00:00
|
|
|
|
1998-12-19 02:55:34 +00:00
|
|
|
if (flags & RFSIGSHARE) {
|
- Merge struct procsig with struct sigacts.
- Move struct sigacts out of the u-area and malloc() it using the
M_SUBPROC malloc bucket.
- Add a small sigacts_*() API for managing sigacts structures: sigacts_alloc(),
sigacts_free(), sigacts_copy(), sigacts_share(), and sigacts_shared().
- Remove the p_sigignore, p_sigacts, and p_sigcatch macros.
- Add a mutex to struct sigacts that protects all the members of the struct.
- Add sigacts locking.
- Remove Giant from nosys(), kill(), killpg(), and kern_sigaction() now
that sigacts is locked.
- Several in-kernel functions such as psignal(), tdsignal(), trapsignal(),
and thread_stopped() are now MP safe.
Reviewed by: arch@
Approved by: re (rwatson)
2003-05-13 20:36:02 +00:00
|
|
|
p2->p_sigacts = sigacts_hold(p1->p_sigacts);
|
1998-12-19 02:55:34 +00:00
|
|
|
} else {
|
- Merge struct procsig with struct sigacts.
- Move struct sigacts out of the u-area and malloc() it using the
M_SUBPROC malloc bucket.
- Add a small sigacts_*() API for managing sigacts structures: sigacts_alloc(),
sigacts_free(), sigacts_copy(), sigacts_share(), and sigacts_shared().
- Remove the p_sigignore, p_sigacts, and p_sigcatch macros.
- Add a mutex to struct sigacts that protects all the members of the struct.
- Add sigacts locking.
- Remove Giant from nosys(), kill(), killpg(), and kern_sigaction() now
that sigacts is locked.
- Several in-kernel functions such as psignal(), tdsignal(), trapsignal(),
and thread_stopped() are now MP safe.
Reviewed by: arch@
Approved by: re (rwatson)
2003-05-13 20:36:02 +00:00
|
|
|
sigacts_copy(newsigacts, p1->p_sigacts);
|
|
|
|
p2->p_sigacts = newsigacts;
|
1998-12-19 02:55:34 +00:00
|
|
|
}
|
1999-03-02 00:28:09 +00:00
|
|
|
if (flags & RFLINUXTHPN)
|
1998-12-19 02:55:34 +00:00
|
|
|
p2->p_sigparent = SIGUSR1;
|
1999-03-02 00:28:09 +00:00
|
|
|
else
|
|
|
|
p2->p_sigparent = SIGCHLD;
|
1999-01-26 02:38:12 +00:00
|
|
|
|
1994-05-24 10:09:53 +00:00
|
|
|
p2->p_textvp = p1->p_textvp;
|
2001-03-07 05:21:47 +00:00
|
|
|
p2->p_fd = fd;
|
2003-06-02 16:05:32 +00:00
|
|
|
p2->p_fdtol = fdtol;
|
1996-02-23 18:49:25 +00:00
|
|
|
|
1994-05-24 10:09:53 +00:00
|
|
|
/*
|
2004-03-04 09:56:29 +00:00
|
|
|
* p_limit is copy-on-write. Bump its refcount.
|
1994-05-24 10:09:53 +00:00
|
|
|
*/
|
2004-02-04 21:52:57 +00:00
|
|
|
p2->p_limit = lim_hold(p1->p_limit);
|
2004-11-20 02:28:48 +00:00
|
|
|
|
|
|
|
pstats_fork(p1->p_stats, p2->p_stats);
|
|
|
|
|
2004-10-01 05:01:29 +00:00
|
|
|
PROC_UNLOCK(p1);
|
2004-10-06 23:53:49 +00:00
|
|
|
PROC_UNLOCK(p2);
|
1994-05-24 10:09:53 +00:00
|
|
|
|
2004-03-08 00:32:34 +00:00
|
|
|
/* Bump references to the text vnode (for procfs) */
|
|
|
|
if (p2->p_textvp)
|
|
|
|
vref(p2->p_textvp);
|
|
|
|
|
2002-10-15 00:14:32 +00:00
|
|
|
/*
|
2004-03-04 09:56:29 +00:00
|
|
|
* Set up linkage for kernel based threading.
|
2002-10-15 00:14:32 +00:00
|
|
|
*/
|
2004-03-04 09:56:29 +00:00
|
|
|
if ((flags & RFTHREAD) != 0) {
|
2002-10-15 00:14:32 +00:00
|
|
|
mtx_lock(&ppeers_lock);
|
|
|
|
p2->p_peers = p1->p_peers;
|
|
|
|
p1->p_peers = p2;
|
|
|
|
p2->p_leader = p1->p_leader;
|
|
|
|
mtx_unlock(&ppeers_lock);
|
|
|
|
PROC_LOCK(p1->p_leader);
|
|
|
|
if ((p1->p_leader->p_flag & P_WEXIT) != 0) {
|
|
|
|
PROC_UNLOCK(p1->p_leader);
|
|
|
|
/*
|
|
|
|
* The task leader is exiting, so process p1 is
|
|
|
|
* going to be killed shortly. Since p1 obviously
|
|
|
|
* isn't dead yet, we know that the leader is either
|
|
|
|
* sending SIGKILL's to all the processes in this
|
|
|
|
* task or is sleeping waiting for all the peers to
|
|
|
|
* exit. We let p1 complete the fork, but we need
|
|
|
|
* to go ahead and kill the new process p2 since
|
|
|
|
* the task leader may not get a chance to send
|
|
|
|
* SIGKILL to it. We leave it on the list so that
|
|
|
|
* the task leader will wait for this new process
|
|
|
|
* to commit suicide.
|
|
|
|
*/
|
|
|
|
PROC_LOCK(p2);
|
|
|
|
psignal(p2, SIGKILL);
|
|
|
|
PROC_UNLOCK(p2);
|
2002-11-18 14:23:21 +00:00
|
|
|
} else
|
|
|
|
PROC_UNLOCK(p1->p_leader);
|
2002-10-15 00:14:32 +00:00
|
|
|
} else {
|
|
|
|
p2->p_peers = NULL;
|
|
|
|
p2->p_leader = p2;
|
|
|
|
}
|
|
|
|
|
2002-05-02 15:13:45 +00:00
|
|
|
sx_xlock(&proctree_lock);
|
|
|
|
PGRP_LOCK(p1->p_pgrp);
|
|
|
|
PROC_LOCK(p2);
|
|
|
|
PROC_LOCK(p1);
|
|
|
|
|
1997-02-17 10:58:46 +00:00
|
|
|
/*
|
2003-04-22 20:54:04 +00:00
|
|
|
* Preserve some more flags in subprocess. P_PROFIL has already
|
1997-11-20 16:36:17 +00:00
|
|
|
* been preserved.
|
1997-02-17 10:58:46 +00:00
|
|
|
*/
|
2004-01-03 02:02:26 +00:00
|
|
|
p2->p_flag |= p1->p_flag & P_SUGID;
|
|
|
|
td2->td_pflags |= td->td_pflags & TDP_ALTSTACK;
|
2002-02-23 11:12:57 +00:00
|
|
|
SESS_LOCK(p1->p_session);
|
1994-05-24 10:09:53 +00:00
|
|
|
if (p1->p_session->s_ttyvp != NULL && p1->p_flag & P_CONTROLT)
|
|
|
|
p2->p_flag |= P_CONTROLT;
|
2002-02-23 11:12:57 +00:00
|
|
|
SESS_UNLOCK(p1->p_session);
|
1996-04-17 17:05:08 +00:00
|
|
|
if (flags & RFPPWAIT)
|
1994-05-24 10:09:53 +00:00
|
|
|
p2->p_flag |= P_PPWAIT;
|
1997-11-20 16:36:17 +00:00
|
|
|
|
2004-01-09 23:42:36 +00:00
|
|
|
p2->p_pgrp = p1->p_pgrp;
|
1996-03-11 06:05:03 +00:00
|
|
|
LIST_INSERT_AFTER(p1, p2, p_pglist);
|
2002-04-02 17:12:06 +00:00
|
|
|
PGRP_UNLOCK(p1->p_pgrp);
|
1996-03-11 06:05:03 +00:00
|
|
|
LIST_INIT(&p2->p_children);
|
|
|
|
|
2003-08-19 17:51:11 +00:00
|
|
|
callout_init(&p2->p_itcallout, CALLOUT_MPSAFE);
|
2000-11-27 22:52:31 +00:00
|
|
|
|
1994-05-24 10:09:53 +00:00
|
|
|
#ifdef KTRACE
|
|
|
|
/*
|
2002-06-07 05:42:25 +00:00
|
|
|
* Copy traceflag and tracefile if enabled.
|
1994-05-24 10:09:53 +00:00
|
|
|
*/
|
2002-06-07 05:42:25 +00:00
|
|
|
mtx_lock(&ktrace_mtx);
|
2003-03-13 18:24:22 +00:00
|
|
|
KASSERT(p2->p_tracevp == NULL, ("new process has a ktrace vnode"));
|
2002-06-07 05:42:25 +00:00
|
|
|
if (p1->p_traceflag & KTRFAC_INHERIT) {
|
1994-05-24 10:09:53 +00:00
|
|
|
p2->p_traceflag = p1->p_traceflag;
|
2003-03-13 18:24:22 +00:00
|
|
|
if ((p2->p_tracevp = p1->p_tracevp) != NULL) {
|
|
|
|
VREF(p2->p_tracevp);
|
|
|
|
KASSERT(p1->p_tracecred != NULL,
|
|
|
|
("ktrace vnode with no cred"));
|
|
|
|
p2->p_tracecred = crhold(p1->p_tracecred);
|
|
|
|
}
|
1994-05-24 10:09:53 +00:00
|
|
|
}
|
2002-06-07 05:42:25 +00:00
|
|
|
mtx_unlock(&ktrace_mtx);
|
1994-05-24 10:09:53 +00:00
|
|
|
#endif
|
|
|
|
|
2002-08-07 11:31:45 +00:00
|
|
|
/*
|
|
|
|
* If PF_FORK is set, the child process inherits the
|
|
|
|
* procfs ioctl flags from its parent.
|
|
|
|
*/
|
|
|
|
if (p1->p_pfsflags & PF_FORK) {
|
|
|
|
p2->p_stops = p1->p_stops;
|
|
|
|
p2->p_pfsflags = p1->p_pfsflags;
|
|
|
|
}
|
|
|
|
|
1994-05-24 10:09:53 +00:00
|
|
|
/*
|
|
|
|
* This begins the section where we must prevent the parent
|
2004-10-06 23:53:49 +00:00
|
|
|
* from being swapped.
|
1994-05-24 10:09:53 +00:00
|
|
|
*/
|
2004-10-06 23:53:49 +00:00
|
|
|
_PHOLD(p1);
|
2001-03-07 02:30:39 +00:00
|
|
|
PROC_UNLOCK(p1);
|
2002-05-02 15:13:45 +00:00
|
|
|
|
|
|
|
/*
|
|
|
|
* Attach the new process to its parent.
|
|
|
|
*
|
|
|
|
* If RFNOWAIT is set, the newly created process becomes a child
|
|
|
|
* of init. This effectively disassociates the child from the
|
|
|
|
* parent.
|
|
|
|
*/
|
|
|
|
if (flags & RFNOWAIT)
|
|
|
|
pptr = initproc;
|
|
|
|
else
|
|
|
|
pptr = p1;
|
|
|
|
p2->p_pptr = pptr;
|
|
|
|
LIST_INSERT_HEAD(&pptr->p_children, p2, p_sibling);
|
|
|
|
sx_xunlock(&proctree_lock);
|
|
|
|
|
2003-04-17 22:24:59 +00:00
|
|
|
/* Inform accounting that we have forked. */
|
|
|
|
p2->p_acflag = AFORK;
|
|
|
|
PROC_UNLOCK(p2);
|
|
|
|
|
1996-02-23 18:49:25 +00:00
|
|
|
/*
|
1997-04-07 07:16:06 +00:00
|
|
|
* Finish creating the child process. It will return via a different
|
|
|
|
* execution path later. (ie: directly into user mode)
|
1996-02-23 18:49:25 +00:00
|
|
|
*/
|
2002-02-07 20:58:47 +00:00
|
|
|
vm_forkproc(td, p2, td2, flags);
|
1994-05-24 10:09:53 +00:00
|
|
|
|
2001-01-23 14:32:01 +00:00
|
|
|
if (flags == (RFFDG | RFPROC)) {
|
2004-09-03 05:11:32 +00:00
|
|
|
atomic_add_int(&cnt.v_forks, 1);
|
|
|
|
atomic_add_int(&cnt.v_forkpages, p2->p_vmspace->vm_dsize +
|
|
|
|
p2->p_vmspace->vm_ssize);
|
2001-01-23 14:32:01 +00:00
|
|
|
} else if (flags == (RFFDG | RFPROC | RFPPWAIT | RFMEM)) {
|
2004-09-03 05:11:32 +00:00
|
|
|
atomic_add_int(&cnt.v_vforks, 1);
|
|
|
|
atomic_add_int(&cnt.v_vforkpages, p2->p_vmspace->vm_dsize +
|
|
|
|
p2->p_vmspace->vm_ssize);
|
2001-01-23 14:32:01 +00:00
|
|
|
} else if (p1 == &proc0) {
|
2004-09-03 05:11:32 +00:00
|
|
|
atomic_add_int(&cnt.v_kthreads, 1);
|
|
|
|
atomic_add_int(&cnt.v_kthreadpages, p2->p_vmspace->vm_dsize +
|
|
|
|
p2->p_vmspace->vm_ssize);
|
2001-01-23 14:32:01 +00:00
|
|
|
} else {
|
2004-09-03 05:11:32 +00:00
|
|
|
atomic_add_int(&cnt.v_rforks, 1);
|
|
|
|
atomic_add_int(&cnt.v_rforkpages, p2->p_vmspace->vm_dsize +
|
|
|
|
p2->p_vmspace->vm_ssize);
|
2001-01-23 14:32:01 +00:00
|
|
|
}
|
|
|
|
|
1996-08-19 02:28:24 +00:00
|
|
|
/*
|
1999-04-17 08:36:07 +00:00
|
|
|
* Both processes are set up, now check if any loadable modules want
|
1996-08-22 03:50:33 +00:00
|
|
|
* to adjust anything.
|
|
|
|
* What if they have an error? XXX
|
1996-08-19 02:28:24 +00:00
|
|
|
*/
|
2003-03-24 21:15:35 +00:00
|
|
|
EVENTHANDLER_INVOKE(process_fork, p1, p2, flags);
|
1996-08-19 02:28:24 +00:00
|
|
|
|
2004-02-05 21:01:37 +00:00
|
|
|
/*
|
|
|
|
* Set the child start time and mark the process as being complete.
|
|
|
|
*/
|
|
|
|
microuptime(&p2->p_stats->p_start);
|
|
|
|
mtx_lock_spin(&sched_lock);
|
|
|
|
p2->p_state = PRS_NORMAL;
|
|
|
|
|
1994-05-24 10:09:53 +00:00
|
|
|
/*
|
2000-09-07 01:33:02 +00:00
|
|
|
* If RFSTOPPED not requested, make child runnable and add to
|
|
|
|
* run queue.
|
1994-05-24 10:09:53 +00:00
|
|
|
*/
|
2000-09-07 01:33:02 +00:00
|
|
|
if ((flags & RFSTOPPED) == 0) {
|
2002-09-11 08:13:56 +00:00
|
|
|
TD_SET_CAN_RUN(td2);
|
2004-09-01 02:11:28 +00:00
|
|
|
setrunqueue(td2, SRQ_BORING);
|
2000-09-07 01:33:02 +00:00
|
|
|
}
|
2004-02-05 21:01:37 +00:00
|
|
|
mtx_unlock_spin(&sched_lock);
|
1994-05-24 10:09:53 +00:00
|
|
|
|
|
|
|
/*
|
|
|
|
* Now can be swapped.
|
|
|
|
*/
|
2001-03-07 02:30:39 +00:00
|
|
|
PROC_LOCK(p1);
|
|
|
|
_PRELE(p1);
|
1994-05-24 10:09:53 +00:00
|
|
|
|
2000-04-16 18:53:38 +00:00
|
|
|
/*
|
2003-08-15 21:29:06 +00:00
|
|
|
* Tell any interested parties about the new process.
|
2000-04-16 18:53:38 +00:00
|
|
|
*/
|
2004-08-15 06:24:42 +00:00
|
|
|
KNOTE_LOCKED(&p1->p_klist, NOTE_FORK | p2->p_pid);
|
2003-08-15 21:29:06 +00:00
|
|
|
|
2001-03-07 02:30:39 +00:00
|
|
|
PROC_UNLOCK(p1);
|
2000-04-16 18:53:38 +00:00
|
|
|
|
1994-05-24 10:09:53 +00:00
|
|
|
/*
|
|
|
|
* Preserve synchronization semantics of vfork. If waiting for
|
|
|
|
* child to exec or exit, set P_PPWAIT on child, and sleep on our
|
|
|
|
* proc (in case of exit).
|
|
|
|
*/
|
2001-03-07 02:30:39 +00:00
|
|
|
PROC_LOCK(p2);
|
1996-04-17 17:05:08 +00:00
|
|
|
while (p2->p_flag & P_PPWAIT)
|
2001-03-07 02:30:39 +00:00
|
|
|
msleep(p1, &p2->p_mtx, PWAIT, "ppwait", 0);
|
|
|
|
PROC_UNLOCK(p2);
|
1994-05-24 10:09:53 +00:00
|
|
|
|
2002-08-22 21:45:58 +00:00
|
|
|
/*
|
2004-03-04 09:56:29 +00:00
|
|
|
* If other threads are waiting, let them continue now.
|
2002-08-22 21:45:58 +00:00
|
|
|
*/
|
2004-09-13 22:10:04 +00:00
|
|
|
if (p1->p_flag & P_HADTHREADS) {
|
2002-08-22 21:45:58 +00:00
|
|
|
PROC_LOCK(p1);
|
|
|
|
thread_single_end();
|
|
|
|
PROC_UNLOCK(p1);
|
|
|
|
}
|
|
|
|
|
1994-05-24 10:09:53 +00:00
|
|
|
/*
|
1999-06-30 15:33:41 +00:00
|
|
|
* Return child proc pointer to parent.
|
1994-05-24 10:09:53 +00:00
|
|
|
*/
|
1999-06-30 15:33:41 +00:00
|
|
|
*procp = p2;
|
1994-05-24 10:09:53 +00:00
|
|
|
return (0);
|
2002-10-15 00:14:32 +00:00
|
|
|
fail:
|
2004-03-05 22:37:32 +00:00
|
|
|
sx_sunlock(&proctree_lock);
|
2003-06-19 05:57:25 +00:00
|
|
|
if (ppsratecheck(&lastfail, &curfail, 1))
|
|
|
|
printf("maxproc limit exceeded by uid %i, please see tuning(7) and login.conf(5).\n",
|
2006-06-27 11:28:50 +00:00
|
|
|
td->td_ucred->cr_ruid);
|
2002-10-15 00:14:32 +00:00
|
|
|
sx_xunlock(&allproc_lock);
|
2004-01-25 18:42:18 +00:00
|
|
|
#ifdef MAC
|
|
|
|
mac_destroy_proc(newproc);
|
2006-02-02 01:32:58 +00:00
|
|
|
#endif
|
|
|
|
#ifdef AUDIT
|
|
|
|
audit_proc_free(newproc);
|
2004-01-25 18:42:18 +00:00
|
|
|
#endif
|
2002-10-15 00:14:32 +00:00
|
|
|
uma_zfree(proc_zone, newproc);
|
2004-09-13 22:10:04 +00:00
|
|
|
if (p1->p_flag & P_HADTHREADS) {
|
2002-10-15 00:14:32 +00:00
|
|
|
PROC_LOCK(p1);
|
|
|
|
thread_single_end();
|
|
|
|
PROC_UNLOCK(p1);
|
|
|
|
}
|
|
|
|
tsleep(&forksleep, PUSER, "fork", hz / 2);
|
|
|
|
return (error);
|
1994-05-24 10:09:53 +00:00
|
|
|
}
|
1996-08-19 02:28:24 +00:00
|
|
|
|
2001-01-24 10:47:14 +00:00
|
|
|
/*
|
|
|
|
* Handle the return of a child process from fork1(). This function
|
|
|
|
* is called from the MD fork_trampoline() entry point.
|
|
|
|
*/
|
|
|
|
void
|
|
|
|
fork_exit(callout, arg, frame)
|
2001-01-26 23:51:41 +00:00
|
|
|
void (*callout)(void *, struct trapframe *);
|
2001-01-24 10:47:14 +00:00
|
|
|
void *arg;
|
2001-01-24 21:59:25 +00:00
|
|
|
struct trapframe *frame;
|
2001-01-24 10:47:14 +00:00
|
|
|
{
|
2002-12-10 02:33:45 +00:00
|
|
|
struct proc *p;
|
2003-08-15 21:29:06 +00:00
|
|
|
struct thread *td;
|
|
|
|
|
|
|
|
/*
|
2004-07-27 03:46:31 +00:00
|
|
|
* Finish setting up thread glue so that it begins execution in a
|
|
|
|
* non-nested critical section with sched_lock held but not recursed.
|
2003-08-15 21:29:06 +00:00
|
|
|
*/
|
2002-12-10 02:33:45 +00:00
|
|
|
td = curthread;
|
|
|
|
p = td->td_proc;
|
2003-04-10 17:35:44 +00:00
|
|
|
td->td_oncpu = PCPU_GET(cpuid);
|
2004-02-05 21:01:37 +00:00
|
|
|
KASSERT(p->p_state == PRS_NORMAL, ("executing process is still new"));
|
2003-08-15 21:29:06 +00:00
|
|
|
|
2001-09-12 08:38:13 +00:00
|
|
|
sched_lock.mtx_lock = (uintptr_t)td;
|
2003-10-29 14:40:41 +00:00
|
|
|
mtx_assert(&sched_lock, MA_OWNED | MA_NOTRECURSED);
|
2004-08-09 18:21:12 +00:00
|
|
|
CTR4(KTR_PROC, "fork_exit: new thread %p (kse %p, pid %d, %s)",
|
2004-09-05 02:09:54 +00:00
|
|
|
td, td->td_sched, p->p_pid, p->p_comm);
|
2004-07-27 03:46:31 +00:00
|
|
|
|
|
|
|
/*
|
|
|
|
* Processes normally resume in mi_switch() after being
|
|
|
|
* cpu_switch()'ed to, but when children start up they arrive here
|
|
|
|
* instead, so we must do much the same things as mi_switch() would.
|
|
|
|
*/
|
|
|
|
|
|
|
|
if ((td = PCPU_GET(deadthread))) {
|
|
|
|
PCPU_SET(deadthread, NULL);
|
|
|
|
thread_stash(td);
|
|
|
|
}
|
|
|
|
td = curthread;
|
2001-12-14 23:37:35 +00:00
|
|
|
mtx_unlock_spin(&sched_lock);
|
2001-01-24 10:47:14 +00:00
|
|
|
|
|
|
|
/*
|
|
|
|
* cpu_set_fork_handler intercepts this function call to
|
2003-08-15 21:29:06 +00:00
|
|
|
* have this call a non-return function to stay in kernel mode.
|
|
|
|
* initproc has its own fork handler, but it does return.
|
|
|
|
*/
|
2001-02-20 05:26:15 +00:00
|
|
|
KASSERT(callout != NULL, ("NULL callout in fork_exit"));
|
2001-01-26 23:51:41 +00:00
|
|
|
callout(arg, frame);
|
2001-01-24 10:47:14 +00:00
|
|
|
|
|
|
|
/*
|
|
|
|
* Check if a kernel thread misbehaved and returned from its main
|
|
|
|
* function.
|
|
|
|
*/
|
|
|
|
if (p->p_flag & P_KTHREAD) {
|
|
|
|
printf("Kernel thread \"%s\" (pid %d) exited prematurely.\n",
|
|
|
|
p->p_comm, p->p_pid);
|
|
|
|
kthread_exit(0);
|
|
|
|
}
|
|
|
|
mtx_assert(&Giant, MA_NOTOWNED);
|
2006-08-15 12:10:57 +00:00
|
|
|
|
|
|
|
EVENTHANDLER_INVOKE(schedtail, p);
|
2001-01-24 10:47:14 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Simplified back end of syscall(), used when returning from fork()
|
|
|
|
* directly into user mode. Giant is not held on entry, and must not
|
|
|
|
* be held on return. This function is passed in to fork_exit() as the
|
|
|
|
* first parameter and is called when returning to a new userland process.
|
|
|
|
*/
|
|
|
|
void
|
2001-09-12 08:38:13 +00:00
|
|
|
fork_return(td, frame)
|
|
|
|
struct thread *td;
|
2001-01-24 10:47:14 +00:00
|
|
|
struct trapframe *frame;
|
|
|
|
{
|
|
|
|
|
2006-02-08 08:09:17 +00:00
|
|
|
userret(td, frame);
|
2001-01-24 10:47:14 +00:00
|
|
|
#ifdef KTRACE
|
2002-06-07 05:42:25 +00:00
|
|
|
if (KTRPOINT(td, KTR_SYSRET))
|
|
|
|
ktrsysret(SYS_fork, 0, 0);
|
2001-01-24 10:47:14 +00:00
|
|
|
#endif
|
|
|
|
mtx_assert(&Giant, MA_NOTOWNED);
|
|
|
|
}
|