freebsd-skq/sys/kern/kern_fork.c

1053 lines
26 KiB
C
Raw Normal View History

/*-
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("$FreeBSD$");
#include "opt_kdtrace.h"
#include "opt_ktrace.h"
#include "opt_kstack_pages.h"
#include "opt_procdesc.h"
1994-05-24 10:09:53 +00:00
#include <sys/param.h>
#include <sys/systm.h>
#include <sys/sysproto.h>
#include <sys/eventhandler.h>
#include <sys/fcntl.h>
1994-05-24 10:09:53 +00:00
#include <sys/filedesc.h>
#include <sys/jail.h>
1994-05-24 10:09:53 +00:00
#include <sys/kernel.h>
#include <sys/kthread.h>
#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>
#include <sys/mutex.h>
#include <sys/priv.h>
1994-05-24 10:09:53 +00:00
#include <sys/proc.h>
#include <sys/procdesc.h>
#include <sys/pioctl.h>
#include <sys/racct.h>
1994-05-24 10:09:53 +00:00
#include <sys/resourcevar.h>
#include <sys/sched.h>
#include <sys/syscall.h>
#include <sys/vmmeter.h>
1994-05-24 10:09:53 +00:00
#include <sys/vnode.h>
#include <sys/acct.h>
#include <sys/ktr.h>
1994-05-24 10:09:53 +00:00
#include <sys/ktrace.h>
#include <sys/unistd.h>
#include <sys/sdt.h>
#include <sys/sx.h>
#include <sys/sysent.h>
#include <sys/signalvar.h>
1994-05-24 10:09:53 +00:00
#include <security/audit/audit.h>
#include <security/mac/mac_framework.h>
#include <vm/vm.h>
#include <vm/pmap.h>
#include <vm/vm_map.h>
#include <vm/vm_extern.h>
#include <vm/uma.h>
#ifdef KDTRACE_HOOKS
#include <sys/dtrace_bsd.h>
dtrace_fork_func_t dtrace_fasttrap_fork;
#endif
SDT_PROVIDER_DECLARE(proc);
SDT_PROBE_DEFINE3(proc, kernel, , create, create, "struct proc *",
"struct proc *", "int");
#ifndef _SYS_SYSPROTO_H_
struct fork_args {
int dummy;
};
#endif
1994-05-24 10:09:53 +00:00
/* ARGSUSED */
int
sys_fork(struct thread *td, struct fork_args *uap)
1994-05-24 10:09:53 +00:00
{
int error;
struct proc *p2;
error = fork1(td, RFFDG | RFPROC, 0, &p2, NULL, 0);
if (error == 0) {
td->td_retval[0] = p2->p_pid;
td->td_retval[1] = 0;
}
return (error);
1994-05-24 10:09:53 +00:00
}
/* ARGUSED */
int
sys_pdfork(td, uap)
struct thread *td;
struct pdfork_args *uap;
{
#ifdef PROCDESC
int error, fd;
struct proc *p2;
/*
* It is necessary to return fd by reference because 0 is a valid file
* descriptor number, and the child needs to be able to distinguish
* itself from the parent using the return value.
*/
error = fork1(td, RFFDG | RFPROC | RFPROCDESC, 0, &p2,
&fd, uap->flags);
if (error == 0) {
td->td_retval[0] = p2->p_pid;
td->td_retval[1] = 0;
error = copyout(&fd, uap->fdp, sizeof(fd));
}
return (error);
#else
return (ENOSYS);
#endif
}
1994-05-24 10:09:53 +00:00
/* ARGSUSED */
int
sys_vfork(struct thread *td, struct vfork_args *uap)
1994-05-24 10:09:53 +00:00
{
int error, flags;
struct proc *p2;
flags = RFFDG | RFPROC | RFPPWAIT | RFMEM;
error = fork1(td, flags, 0, &p2, NULL, 0);
if (error == 0) {
td->td_retval[0] = p2->p_pid;
td->td_retval[1] = 0;
}
return (error);
}
1994-05-24 10:09:53 +00:00
int
sys_rfork(struct thread *td, struct rfork_args *uap)
{
struct proc *p2;
int error;
/* Don't allow kernel-only flags. */
if ((uap->flags & RFKERNELONLY) != 0)
return (EINVAL);
AUDIT_ARG_FFLAGS(uap->flags);
error = fork1(td, uap->flags, 0, &p2, NULL, 0);
if (error == 0) {
td->td_retval[0] = p2 ? p2->p_pid : 0;
td->td_retval[1] = 0;
}
return (error);
1994-05-24 10:09:53 +00:00
}
int nprocs = 1; /* process 0 */
int lastpid = 0;
SYSCTL_INT(_kern, OID_AUTO, lastpid, CTLFLAG_RD, &lastpid, 0,
"Last used PID");
1994-05-24 10:09:53 +00:00
/*
* Random component to lastpid generation. We mix in a random factor to make
* 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.
*/
static int randompid = 0;
static int
sysctl_kern_randompid(SYSCTL_HANDLER_ARGS)
{
int error, pid;
error = sysctl_wire_old_buffer(req, sizeof(int));
if (error != 0)
return(error);
sx_xlock(&allproc_lock);
pid = randompid;
error = sysctl_handle_int(oidp, &pid, 0, req);
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);
return (error);
}
SYSCTL_PROC(_kern, OID_AUTO, randompid, CTLTYPE_INT|CTLFLAG_RW,
0, 0, sysctl_kern_randompid, "I", "Random PID modulus");
static int
fork_findpid(int flags)
{
struct proc *p;
int trypid;
static int pidchecked = 0;
/*
* Requires allproc_lock in order to iterate over the list
* of processes, and proctree_lock to access p_pgrp.
*/
sx_assert(&allproc_lock, SX_LOCKED);
sx_assert(&proctree_lock, SX_LOCKED);
/*
* Find an unused process ID. We remember a range of unused IDs
* ready to use (from lastpid+1 through pidchecked-1).
*
* If RFHIGHPID is set (used during system boot), do not allocate
* low-numbered pids.
*/
trypid = lastpid + 1;
if (flags & RFHIGHPID) {
if (trypid < 10)
trypid = 10;
} else {
if (randompid)
trypid += arc4random() % randompid;
}
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.
*/
if (trypid >= pid_max) {
trypid = trypid % pid_max;
if (trypid < 100)
trypid += 100;
pidchecked = 0;
}
if (trypid >= pidchecked) {
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
* than trypid, so we can avoid checking for a while.
*/
p = LIST_FIRST(&allproc);
again:
for (; p != NULL; p = LIST_NEXT(p, p_list)) {
while (p->p_pid == trypid ||
(p->p_pgrp != NULL &&
(p->p_pgrp->pg_id == trypid ||
(p->p_session != NULL &&
p->p_session->s_sid == trypid)))) {
trypid++;
if (trypid >= pidchecked)
goto retry;
}
if (p->p_pid > trypid && pidchecked > p->p_pid)
pidchecked = p->p_pid;
if (p->p_pgrp != NULL) {
if (p->p_pgrp->pg_id > trypid &&
pidchecked > p->p_pgrp->pg_id)
pidchecked = p->p_pgrp->pg_id;
if (p->p_session != NULL &&
p->p_session->s_sid > trypid &&
pidchecked > p->p_session->s_sid)
pidchecked = p->p_session->s_sid;
}
}
if (!doingzomb) {
doingzomb = 1;
p = LIST_FIRST(&zombproc);
goto again;
}
}
/*
* RFHIGHPID does not mess with the lastpid counter during boot.
*/
if (flags & RFHIGHPID)
pidchecked = 0;
else
lastpid = trypid;
return (trypid);
}
static int
fork_norfproc(struct thread *td, int flags)
{
int error;
struct proc *p1;
KASSERT((flags & RFPROC) == 0,
("fork_norfproc called with RFPROC set"));
p1 = td->td_proc;
if (((p1->p_flag & (P_HADTHREADS|P_SYSTEM)) == P_HADTHREADS) &&
(flags & (RFCFDG | RFFDG))) {
PROC_LOCK(p1);
if (thread_single(SINGLE_BOUNDARY)) {
PROC_UNLOCK(p1);
return (ERESTART);
}
PROC_UNLOCK(p1);
}
error = vm_forkproc(td, NULL, NULL, NULL, flags);
if (error)
goto fail;
/*
* Close all file descriptors.
*/
if (flags & RFCFDG) {
struct filedesc *fdtmp;
fdtmp = fdinit(td->td_proc->p_fd);
Merge Capsicum overhaul: - Capability is no longer separate descriptor type. Now every descriptor has set of its own capability rights. - The cap_new(2) system call is left, but it is no longer documented and should not be used in new code. - The new syscall cap_rights_limit(2) should be used instead of cap_new(2), which limits capability rights of the given descriptor without creating a new one. - The cap_getrights(2) syscall is renamed to cap_rights_get(2). - If CAP_IOCTL capability right is present we can further reduce allowed ioctls list with the new cap_ioctls_limit(2) syscall. List of allowed ioctls can be retrived with cap_ioctls_get(2) syscall. - If CAP_FCNTL capability right is present we can further reduce fcntls that can be used with the new cap_fcntls_limit(2) syscall and retrive them with cap_fcntls_get(2). - To support ioctl and fcntl white-listing the filedesc structure was heavly modified. - The audit subsystem, kdump and procstat tools were updated to recognize new syscalls. - Capability rights were revised and eventhough I tried hard to provide backward API and ABI compatibility there are some incompatible changes that are described in detail below: CAP_CREATE old behaviour: - Allow for openat(2)+O_CREAT. - Allow for linkat(2). - Allow for symlinkat(2). CAP_CREATE new behaviour: - Allow for openat(2)+O_CREAT. Added CAP_LINKAT: - Allow for linkat(2). ABI: Reuses CAP_RMDIR bit. - Allow to be target for renameat(2). Added CAP_SYMLINKAT: - Allow for symlinkat(2). Removed CAP_DELETE. Old behaviour: - Allow for unlinkat(2) when removing non-directory object. - Allow to be source for renameat(2). Removed CAP_RMDIR. Old behaviour: - Allow for unlinkat(2) when removing directory. Added CAP_RENAMEAT: - Required for source directory for the renameat(2) syscall. Added CAP_UNLINKAT (effectively it replaces CAP_DELETE and CAP_RMDIR): - Allow for unlinkat(2) on any object. - Required if target of renameat(2) exists and will be removed by this call. Removed CAP_MAPEXEC. CAP_MMAP old behaviour: - Allow for mmap(2) with any combination of PROT_NONE, PROT_READ and PROT_WRITE. CAP_MMAP new behaviour: - Allow for mmap(2)+PROT_NONE. Added CAP_MMAP_R: - Allow for mmap(PROT_READ). Added CAP_MMAP_W: - Allow for mmap(PROT_WRITE). Added CAP_MMAP_X: - Allow for mmap(PROT_EXEC). Added CAP_MMAP_RW: - Allow for mmap(PROT_READ | PROT_WRITE). Added CAP_MMAP_RX: - Allow for mmap(PROT_READ | PROT_EXEC). Added CAP_MMAP_WX: - Allow for mmap(PROT_WRITE | PROT_EXEC). Added CAP_MMAP_RWX: - Allow for mmap(PROT_READ | PROT_WRITE | PROT_EXEC). Renamed CAP_MKDIR to CAP_MKDIRAT. Renamed CAP_MKFIFO to CAP_MKFIFOAT. Renamed CAP_MKNODE to CAP_MKNODEAT. CAP_READ old behaviour: - Allow pread(2). - Disallow read(2), readv(2) (if there is no CAP_SEEK). CAP_READ new behaviour: - Allow read(2), readv(2). - Disallow pread(2) (CAP_SEEK was also required). CAP_WRITE old behaviour: - Allow pwrite(2). - Disallow write(2), writev(2) (if there is no CAP_SEEK). CAP_WRITE new behaviour: - Allow write(2), writev(2). - Disallow pwrite(2) (CAP_SEEK was also required). Added convinient defines: #define CAP_PREAD (CAP_SEEK | CAP_READ) #define CAP_PWRITE (CAP_SEEK | CAP_WRITE) #define CAP_MMAP_R (CAP_MMAP | CAP_SEEK | CAP_READ) #define CAP_MMAP_W (CAP_MMAP | CAP_SEEK | CAP_WRITE) #define CAP_MMAP_X (CAP_MMAP | CAP_SEEK | 0x0000000000000008ULL) #define CAP_MMAP_RW (CAP_MMAP_R | CAP_MMAP_W) #define CAP_MMAP_RX (CAP_MMAP_R | CAP_MMAP_X) #define CAP_MMAP_WX (CAP_MMAP_W | CAP_MMAP_X) #define CAP_MMAP_RWX (CAP_MMAP_R | CAP_MMAP_W | CAP_MMAP_X) #define CAP_RECV CAP_READ #define CAP_SEND CAP_WRITE #define CAP_SOCK_CLIENT \ (CAP_CONNECT | CAP_GETPEERNAME | CAP_GETSOCKNAME | CAP_GETSOCKOPT | \ CAP_PEELOFF | CAP_RECV | CAP_SEND | CAP_SETSOCKOPT | CAP_SHUTDOWN) #define CAP_SOCK_SERVER \ (CAP_ACCEPT | CAP_BIND | CAP_GETPEERNAME | CAP_GETSOCKNAME | \ CAP_GETSOCKOPT | CAP_LISTEN | CAP_PEELOFF | CAP_RECV | CAP_SEND | \ CAP_SETSOCKOPT | CAP_SHUTDOWN) Added defines for backward API compatibility: #define CAP_MAPEXEC CAP_MMAP_X #define CAP_DELETE CAP_UNLINKAT #define CAP_MKDIR CAP_MKDIRAT #define CAP_RMDIR CAP_UNLINKAT #define CAP_MKFIFO CAP_MKFIFOAT #define CAP_MKNOD CAP_MKNODAT #define CAP_SOCK_ALL (CAP_SOCK_CLIENT | CAP_SOCK_SERVER) Sponsored by: The FreeBSD Foundation Reviewed by: Christoph Mallon <christoph.mallon@gmx.de> Many aspects discussed with: rwatson, benl, jonathan ABI compatibility discussed with: kib
2013-03-02 00:53:12 +00:00
fdescfree(td);
p1->p_fd = fdtmp;
}
/*
* Unshare file descriptors (from parent).
*/
if (flags & RFFDG)
fdunshare(p1, td);
fail:
if (((p1->p_flag & (P_HADTHREADS|P_SYSTEM)) == P_HADTHREADS) &&
(flags & (RFCFDG | RFFDG))) {
PROC_LOCK(p1);
thread_single_end();
PROC_UNLOCK(p1);
}
return (error);
}
static void
do_fork(struct thread *td, int flags, struct proc *p2, struct thread *td2,
struct vmspace *vm2, int pdflags)
1994-05-24 10:09:53 +00:00
{
struct proc *p1, *pptr;
int p2_held, trypid;
struct filedesc *fd;
struct filedesc_to_leader *fdtol;
struct sigacts *newsigacts;
sx_assert(&proctree_lock, SX_SLOCKED);
sx_assert(&allproc_lock, SX_XLOCKED);
1994-05-24 10:09:53 +00:00
p2_held = 0;
p1 = td->td_proc;
/*
* Increment the nprocs resource before blocking can occur. There
* are hard-limits as to the number of processes that can run.
*/
nprocs++;
trypid = fork_findpid(flags);
1994-05-24 10:09:53 +00:00
sx_sunlock(&proctree_lock);
1994-05-24 10:09:53 +00:00
p2->p_state = PRS_NEW; /* protect against others */
p2->p_pid = trypid;
AUDIT_ARG_PID(p2->p_pid);
LIST_INSERT_HEAD(&allproc, p2, p_list);
LIST_INSERT_HEAD(PIDHASH(p2->p_pid), p2, p_hash);
tidhash_add(td2);
Close race conditions between fork() and [sg]etpriority()'s PRIO_USER case, possibly also other places that deferences p_ucred. In the past, we insert a new process into the allproc list right after PID allocation, and release the allproc_lock sx. Because most content in new proc's structure is not yet initialized, this could lead to undefined result if we do not handle PRS_NEW with care. The problem with PRS_NEW state is that it does not provide fine grained information about how much initialization is done for a new process. By defination, after PRIO_USER setpriority(), all processes that belongs to given user should have their nice value set to the specified value. Therefore, if p_{start,end}copy section was done for a PRS_NEW process, we can not safely ignore it because p_nice is in this area. On the other hand, we should be careful on PRS_NEW processes because we do not allow non-root users to lower their nice values, and without a successful copy of the copy section, we can get stale values that is inherted from the uninitialized area of the process structure. This commit tries to close the race condition by grabbing proc mutex *before* we release allproc_lock xlock, and do copy as well as zero immediately after the allproc_lock xunlock. This guarantees that the new process would have its p_copy and p_zero sections, as well as user credential informaion initialized. In getpriority() case, instead of grabbing PROC_LOCK for a PRS_NEW process, we just skip the process in question, because it does not affect the final result of the call, as the p_nice value would be copied from its parent, and we will see it during allproc traverse. Other potential solutions are still under evaluation. Discussed with: davidxu, jhb, rwatson PR: kern/108071 MFC after: 2 weeks
2007-02-26 03:38:09 +00:00
PROC_LOCK(p2);
PROC_LOCK(p1);
sx_xunlock(&allproc_lock);
Close race conditions between fork() and [sg]etpriority()'s PRIO_USER case, possibly also other places that deferences p_ucred. In the past, we insert a new process into the allproc list right after PID allocation, and release the allproc_lock sx. Because most content in new proc's structure is not yet initialized, this could lead to undefined result if we do not handle PRS_NEW with care. The problem with PRS_NEW state is that it does not provide fine grained information about how much initialization is done for a new process. By defination, after PRIO_USER setpriority(), all processes that belongs to given user should have their nice value set to the specified value. Therefore, if p_{start,end}copy section was done for a PRS_NEW process, we can not safely ignore it because p_nice is in this area. On the other hand, we should be careful on PRS_NEW processes because we do not allow non-root users to lower their nice values, and without a successful copy of the copy section, we can get stale values that is inherted from the uninitialized area of the process structure. This commit tries to close the race condition by grabbing proc mutex *before* we release allproc_lock xlock, and do copy as well as zero immediately after the allproc_lock xunlock. This guarantees that the new process would have its p_copy and p_zero sections, as well as user credential informaion initialized. In getpriority() case, instead of grabbing PROC_LOCK for a PRS_NEW process, we just skip the process in question, because it does not affect the final result of the call, as the p_nice value would be copied from its parent, and we will see it during allproc traverse. Other potential solutions are still under evaluation. Discussed with: davidxu, jhb, rwatson PR: kern/108071 MFC after: 2 weeks
2007-02-26 03:38:09 +00:00
bcopy(&p1->p_startcopy, &p2->p_startcopy,
__rangeof(struct proc, p_startcopy, p_endcopy));
pargs_hold(p2->p_args);
Close race conditions between fork() and [sg]etpriority()'s PRIO_USER case, possibly also other places that deferences p_ucred. In the past, we insert a new process into the allproc list right after PID allocation, and release the allproc_lock sx. Because most content in new proc's structure is not yet initialized, this could lead to undefined result if we do not handle PRS_NEW with care. The problem with PRS_NEW state is that it does not provide fine grained information about how much initialization is done for a new process. By defination, after PRIO_USER setpriority(), all processes that belongs to given user should have their nice value set to the specified value. Therefore, if p_{start,end}copy section was done for a PRS_NEW process, we can not safely ignore it because p_nice is in this area. On the other hand, we should be careful on PRS_NEW processes because we do not allow non-root users to lower their nice values, and without a successful copy of the copy section, we can get stale values that is inherted from the uninitialized area of the process structure. This commit tries to close the race condition by grabbing proc mutex *before* we release allproc_lock xlock, and do copy as well as zero immediately after the allproc_lock xunlock. This guarantees that the new process would have its p_copy and p_zero sections, as well as user credential informaion initialized. In getpriority() case, instead of grabbing PROC_LOCK for a PRS_NEW process, we just skip the process in question, because it does not affect the final result of the call, as the p_nice value would be copied from its parent, and we will see it during allproc traverse. Other potential solutions are still under evaluation. Discussed with: davidxu, jhb, rwatson PR: kern/108071 MFC after: 2 weeks
2007-02-26 03:38:09 +00:00
PROC_UNLOCK(p1);
bzero(&p2->p_startzero,
__rangeof(struct proc, p_startzero, p_endzero));
p2->p_ucred = crhold(td->td_ucred);
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
/* Tell the prison that we exist. */
prison_proc_hold(p2->p_ucred->cr_prison);
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
Close race conditions between fork() and [sg]etpriority()'s PRIO_USER case, possibly also other places that deferences p_ucred. In the past, we insert a new process into the allproc list right after PID allocation, and release the allproc_lock sx. Because most content in new proc's structure is not yet initialized, this could lead to undefined result if we do not handle PRS_NEW with care. The problem with PRS_NEW state is that it does not provide fine grained information about how much initialization is done for a new process. By defination, after PRIO_USER setpriority(), all processes that belongs to given user should have their nice value set to the specified value. Therefore, if p_{start,end}copy section was done for a PRS_NEW process, we can not safely ignore it because p_nice is in this area. On the other hand, we should be careful on PRS_NEW processes because we do not allow non-root users to lower their nice values, and without a successful copy of the copy section, we can get stale values that is inherted from the uninitialized area of the process structure. This commit tries to close the race condition by grabbing proc mutex *before* we release allproc_lock xlock, and do copy as well as zero immediately after the allproc_lock xunlock. This guarantees that the new process would have its p_copy and p_zero sections, as well as user credential informaion initialized. In getpriority() case, instead of grabbing PROC_LOCK for a PRS_NEW process, we just skip the process in question, because it does not affect the final result of the call, as the p_nice value would be copied from its parent, and we will see it during allproc traverse. Other potential solutions are still under evaluation. Discussed with: davidxu, jhb, rwatson PR: kern/108071 MFC after: 2 weeks
2007-02-26 03:38:09 +00:00
PROC_UNLOCK(p2);
/*
* Malloc things while we don't hold any locks.
*/
if (flags & RFSIGSHARE)
newsigacts = NULL;
else
newsigacts = sigacts_alloc();
/*
* Copy filedesc.
*/
if (flags & RFCFDG) {
fd = fdinit(p1->p_fd);
fdtol = NULL;
} else if (flags & RFFDG) {
fd = fdcopy(p1->p_fd);
fdtol = NULL;
} else {
fd = fdshare(p1->p_fd);
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;
Replace custom file descriptor array sleep lock constructed using a mutex and flags with an sxlock. This leads to a significant and measurable performance improvement as a result of access to shared locking for frequent lookup operations, reduced general overhead, and reduced overhead in the event of contention. All of these are imported for threaded applications where simultaneous access to a shared file descriptor array occurs frequently. Kris has reported 2x-4x transaction rate improvements on 8-core MySQL benchmarks; smaller improvements can be expected for many workloads as a result of reduced overhead. - Generally eliminate the distinction between "fast" and regular acquisisition of the filedesc lock; the plan is that they will now all be fast. Change all locking instances to either shared or exclusive locks. - Correct a bug (pointed out by kib) in fdfree() where previously msleep() was called without the mutex held; sx_sleep() is now always called with the sxlock held exclusively. - Universally hold the struct file lock over changes to struct file, rather than the filedesc lock or no lock. Always update the f_ops field last. A further memory barrier is required here in the future (discussed with jhb). - Improve locking and reference management in linux_at(), which fails to properly acquire vnode references before using vnode pointers. Annotate improper use of vn_fullpath(), which will be replaced at a future date. In fcntl(), we conservatively acquire an exclusive lock, even though in some cases a shared lock may be sufficient, which should be revisited. The dropping of the filedesc lock in fdgrowtable() is no longer required as the sxlock can be held over the sleep operation; we should consider removing that (pointed out by attilio). Tested by: kris Discussed with: jhb, kris, attilio, jeff
2007-04-04 09:11:34 +00:00
FILEDESC_XLOCK(p1->p_fd);
fdtol->fdl_refcount++;
Replace custom file descriptor array sleep lock constructed using a mutex and flags with an sxlock. This leads to a significant and measurable performance improvement as a result of access to shared locking for frequent lookup operations, reduced general overhead, and reduced overhead in the event of contention. All of these are imported for threaded applications where simultaneous access to a shared file descriptor array occurs frequently. Kris has reported 2x-4x transaction rate improvements on 8-core MySQL benchmarks; smaller improvements can be expected for many workloads as a result of reduced overhead. - Generally eliminate the distinction between "fast" and regular acquisisition of the filedesc lock; the plan is that they will now all be fast. Change all locking instances to either shared or exclusive locks. - Correct a bug (pointed out by kib) in fdfree() where previously msleep() was called without the mutex held; sx_sleep() is now always called with the sxlock held exclusively. - Universally hold the struct file lock over changes to struct file, rather than the filedesc lock or no lock. Always update the f_ops field last. A further memory barrier is required here in the future (discussed with jhb). - Improve locking and reference management in linux_at(), which fails to properly acquire vnode references before using vnode pointers. Annotate improper use of vn_fullpath(), which will be replaced at a future date. In fcntl(), we conservatively acquire an exclusive lock, even though in some cases a shared lock may be sufficient, which should be revisited. The dropping of the filedesc lock in fdgrowtable() is no longer required as the sxlock can be held over the sleep operation; we should consider removing that (pointed out by attilio). Tested by: kris Discussed with: jhb, kris, attilio, jeff
2007-04-04 09:11:34 +00:00
FILEDESC_XUNLOCK(p1->p_fd);
} 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.
*/
PROC_LOCK(p2);
PROC_LOCK(p1);
bzero(&td2->td_startzero,
__rangeof(struct thread, td_startzero, td_endzero));
bcopy(&td->td_startcopy, &td2->td_startcopy,
__rangeof(struct thread, td_startcopy, td_endcopy));
1994-05-24 10:09:53 +00:00
bcopy(&p2->p_comm, &td2->td_name, sizeof(td2->td_name));
td2->td_sigstk = td->td_sigstk;
td2->td_flags = TDF_INMEM;
td2->td_lend_user_pri = PRI_MAX;
Change the curvnet variable from a global const struct vnet *, previously always pointing to the default vnet context, to a dynamically changing thread-local one. The currvnet context should be set on entry to networking code via CURVNET_SET() macros, and reverted to previous state via CURVNET_RESTORE(). Recursions on curvnet are permitted, though strongly discuouraged. This change should have no functional impact on nooptions VIMAGE kernel builds, where CURVNET_* macros expand to whitespace. The curthread->td_vnet (aka curvnet) variable's purpose is to be an indicator of the vnet context in which the current network-related operation takes place, in case we cannot deduce the current vnet context from any other source, such as by looking at mbuf's m->m_pkthdr.rcvif->if_vnet, sockets's so->so_vnet etc. Moreover, so far curvnet has turned out to be an invaluable consistency checking aid: it helps to catch cases when sockets, ifnets or any other vnet-aware structures may have leaked from one vnet to another. The exact placement of the CURVNET_SET() / CURVNET_RESTORE() macros was a result of an empirical iterative process, whith an aim to reduce recursions on CURVNET_SET() to a minimum, while still reducing the scope of CURVNET_SET() to networking only operations - the alternative would be calling CURVNET_SET() on each system call entry. In general, curvnet has to be set in three typicall cases: when processing socket-related requests from userspace or from within the kernel; when processing inbound traffic flowing from device drivers to upper layers of the networking stack, and when executing timer-driven networking functions. This change also introduces a DDB subcommand to show the list of all vnet instances. Approved by: julian (mentor)
2009-05-05 10:56:12 +00:00
#ifdef VIMAGE
td2->td_vnet = NULL;
td2->td_vnet_lpush = NULL;
#endif
/*
* Allow the scheduler to initialize the child.
*/
thread_lock(td);
sched_fork(td, td2);
thread_unlock(td);
1994-05-24 10:09:53 +00:00
/*
* Duplicate sub-structures as needed.
* Increase reference counts on shared objects.
*/
p2->p_flag = P_INMEM;
p2->p_swtick = ticks;
if (p1->p_flag & P_PROFIL)
startprofclock(p2);
td2->td_ucred = crhold(p2->p_ucred);
if (flags & RFSIGSHARE) {
p2->p_sigacts = sigacts_hold(p1->p_sigacts);
} else {
sigacts_copy(newsigacts, p1->p_sigacts);
p2->p_sigacts = newsigacts;
}
if (flags & RFTSIGZMB)
p2->p_sigparent = RFTSIGNUM(flags);
else if (flags & RFLINUXTHPN)
p2->p_sigparent = SIGUSR1;
else
p2->p_sigparent = SIGCHLD;
1994-05-24 10:09:53 +00:00
p2->p_textvp = p1->p_textvp;
p2->p_fd = fd;
p2->p_fdtol = fdtol;
1994-05-24 10:09:53 +00:00
/*
* p_limit is copy-on-write. Bump its refcount.
1994-05-24 10:09:53 +00:00
*/
lim_fork(p1, p2);
pstats_fork(p1->p_stats, p2->p_stats);
PROC_UNLOCK(p1);
PROC_UNLOCK(p2);
1994-05-24 10:09:53 +00:00
/* Bump references to the text vnode (for procfs). */
if (p2->p_textvp)
vref(p2->p_textvp);
/*
* Set up linkage for kernel based threading.
*/
if ((flags & RFTHREAD) != 0) {
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);
kern_psignal(p2, SIGKILL);
PROC_UNLOCK(p2);
} else
PROC_UNLOCK(p1->p_leader);
} else {
p2->p_peers = NULL;
p2->p_leader = p2;
}
sx_xlock(&proctree_lock);
PGRP_LOCK(p1->p_pgrp);
PROC_LOCK(p2);
PROC_LOCK(p1);
/*
* Preserve some more flags in subprocess. P_PROFIL has already
* been preserved.
*/
p2->p_flag |= p1->p_flag & P_SUGID;
td2->td_pflags |= td->td_pflags & TDP_ALTSTACK;
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;
SESS_UNLOCK(p1->p_session);
if (flags & RFPPWAIT)
1994-05-24 10:09:53 +00:00
p2->p_flag |= P_PPWAIT;
p2->p_pgrp = p1->p_pgrp;
LIST_INSERT_AFTER(p1, p2, p_pglist);
2002-04-02 17:12:06 +00:00
PGRP_UNLOCK(p1->p_pgrp);
LIST_INIT(&p2->p_children);
LIST_INIT(&p2->p_orphans);
callout_init_mtx(&p2->p_itcallout, &p2->p_mtx, 0);
/*
* 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
* from being swapped.
1994-05-24 10:09:53 +00:00
*/
_PHOLD(p1);
PROC_UNLOCK(p1);
/*
* 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);
/* Inform accounting that we have forked. */
p2->p_acflag = AFORK;
PROC_UNLOCK(p2);
#ifdef KTRACE
ktrprocfork(p1, p2);
#endif
/*
The biggie: Get rid of the UPAGES from the top of the per-process address space. (!) Have each process use the kernel stack and pcb in the kvm space. Since the stacks are at a different address, we cannot copy the stack at fork() and allow the child to return up through the function call tree to return to user mode - create a new execution context and have the new process begin executing from cpu_switch() and go to user mode directly. In theory this should speed up fork a bit. Context switch the tss_esp0 pointer in the common tss. This is a lot simpler since than swithching the gdt[GPROC0_SEL].sd.sd_base pointer to each process's tss since the esp0 pointer is a 32 bit pointer, and the sd_base setting is split into three different bit sections at non-aligned boundaries and requires a lot of twiddling to reset. The 8K of memory at the top of the process space is now empty, and unmapped (and unmappable, it's higher than VM_MAXUSER_ADDRESS). Simplity the pmap code to manage process contexts, we no longer have to double map the UPAGES, this simplifies and should measuably speed up fork(). The following parts came from John Dyson: Set PG_G on the UPAGES that are now in kernel context, and invalidate them when swapping them out. Move the upages object (upobj) from the vmspace to the proc structure. Now that the UPAGES (pcb and kernel stack) are out of user space, make rfork(..RFMEM..) do what was intended by sharing the vmspace entirely via reference counting rather than simply inheriting the mappings.
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)
*/
vm_forkproc(td, p2, td2, vm2, flags);
1994-05-24 10:09:53 +00:00
if (flags == (RFFDG | RFPROC)) {
PCPU_INC(cnt.v_forks);
PCPU_ADD(cnt.v_forkpages, p2->p_vmspace->vm_dsize +
p2->p_vmspace->vm_ssize);
} else if (flags == (RFFDG | RFPROC | RFPPWAIT | RFMEM)) {
PCPU_INC(cnt.v_vforks);
PCPU_ADD(cnt.v_vforkpages, p2->p_vmspace->vm_dsize +
p2->p_vmspace->vm_ssize);
} else if (p1 == &proc0) {
PCPU_INC(cnt.v_kthreads);
PCPU_ADD(cnt.v_kthreadpages, p2->p_vmspace->vm_dsize +
p2->p_vmspace->vm_ssize);
} else {
PCPU_INC(cnt.v_rforks);
PCPU_ADD(cnt.v_rforkpages, p2->p_vmspace->vm_dsize +
p2->p_vmspace->vm_ssize);
}
#ifdef PROCDESC
/*
* Associate the process descriptor with the process before anything
* can happen that might cause that process to need the descriptor.
* However, don't do this until after fork(2) can no longer fail.
*/
if (flags & RFPROCDESC)
procdesc_new(p2, pdflags);
#endif
/*
* Both processes are set up, now check if any loadable modules want
* to adjust anything.
*/
EVENTHANDLER_INVOKE(process_fork, p1, p2, flags);
/*
* Set the child start time and mark the process as being complete.
*/
PROC_LOCK(p2);
PROC_LOCK(p1);
microuptime(&p2->p_stats->p_start);
PROC_SLOCK(p2);
p2->p_state = PRS_NORMAL;
PROC_SUNLOCK(p2);
#ifdef KDTRACE_HOOKS
/*
* Tell the DTrace fasttrap provider about the new process
* if it has registered an interest. We have to do this only after
* p_state is PRS_NORMAL since the fasttrap module will use pfind()
* later on.
*/
if (dtrace_fasttrap_fork)
dtrace_fasttrap_fork(p1, p2);
#endif
if ((p1->p_flag & (P_TRACED | P_FOLLOWFORK)) == (P_TRACED |
P_FOLLOWFORK)) {
/*
* Arrange for debugger to receive the fork event.
*
* We can report PL_FLAG_FORKED regardless of
* P_FOLLOWFORK settings, but it does not make a sense
* for runaway child.
*/
td->td_dbgflags |= TDB_FORK;
td->td_dbg_forked = p2->p_pid;
td2->td_dbgflags |= TDB_STOPATFORK;
_PHOLD(p2);
p2_held = 1;
}
if (flags & RFPPWAIT) {
td->td_pflags |= TDP_RFPPWAIT;
td->td_rfppwait_p = p2;
}
PROC_UNLOCK(p2);
if ((flags & RFSTOPPED) == 0) {
/*
* If RFSTOPPED not requested, make child runnable and
* add to run queue.
*/
thread_lock(td2);
TD_SET_CAN_RUN(td2);
sched_add(td2, SRQ_BORING);
thread_unlock(td2);
}
1994-05-24 10:09:53 +00:00
/*
* Now can be swapped.
*/
_PRELE(p1);
PROC_UNLOCK(p1);
1994-05-24 10:09:53 +00:00
/*
* Tell any interested parties about the new process.
*/
knote_fork(&p1->p_klist, p2->p_pid);
SDT_PROBE(proc, kernel, , create, p2, p1, flags, 0, 0);
/*
* Wait until debugger is attached to child.
*/
PROC_LOCK(p2);
while ((td2->td_dbgflags & TDB_STOPATFORK) != 0)
cv_wait(&p2->p_dbgwait, &p2->p_mtx);
if (p2_held)
_PRELE(p2);
PROC_UNLOCK(p2);
}
int
fork1(struct thread *td, int flags, int pages, struct proc **procp,
int *procdescp, int pdflags)
{
struct proc *p1;
struct proc *newproc;
int ok;
struct thread *td2;
struct vmspace *vm2;
vm_ooffset_t mem_charged;
int error;
static int curfail;
static struct timeval lastfail;
#ifdef PROCDESC
struct file *fp_procdesc = NULL;
#endif
/* Check for the undefined or unimplemented flags. */
if ((flags & ~(RFFLAGS | RFTSIGFLAGS(RFTSIGMASK))) != 0)
return (EINVAL);
/* Signal value requires RFTSIGZMB. */
if ((flags & RFTSIGFLAGS(RFTSIGMASK)) != 0 && (flags & RFTSIGZMB) == 0)
return (EINVAL);
/* Can't copy and clear. */
if ((flags & (RFFDG|RFCFDG)) == (RFFDG|RFCFDG))
return (EINVAL);
/* Check the validity of the signal number. */
if ((flags & RFTSIGZMB) != 0 && (u_int)RFTSIGNUM(flags) > _SIG_MAXSIG)
return (EINVAL);
#ifdef PROCDESC
if ((flags & RFPROCDESC) != 0) {
/* Can't not create a process yet get a process descriptor. */
if ((flags & RFPROC) == 0)
return (EINVAL);
/* Must provide a place to put a procdesc if creating one. */
if (procdescp == NULL)
return (EINVAL);
}
#endif
p1 = td->td_proc;
/*
* Here we don't create a new process, but we divorce
* certain parts of a process from itself.
*/
if ((flags & RFPROC) == 0) {
*procp = NULL;
return (fork_norfproc(td, flags));
}
#ifdef PROCDESC
/*
* If required, create a process descriptor in the parent first; we
* will abandon it if something goes wrong. We don't finit() until
* later.
*/
if (flags & RFPROCDESC) {
error = falloc(td, &fp_procdesc, procdescp, 0);
if (error != 0)
return (error);
}
#endif
mem_charged = 0;
vm2 = NULL;
if (pages == 0)
pages = KSTACK_PAGES;
/* Allocate new proc. */
newproc = uma_zalloc(proc_zone, M_WAITOK);
td2 = FIRST_THREAD_IN_PROC(newproc);
if (td2 == NULL) {
td2 = thread_alloc(pages);
if (td2 == NULL) {
error = ENOMEM;
goto fail1;
}
proc_linkup(newproc, td2);
} else {
if (td2->td_kstack == 0 || td2->td_kstack_pages != pages) {
if (td2->td_kstack != 0)
vm_thread_dispose(td2);
if (!thread_alloc_stack(td2, pages)) {
error = ENOMEM;
goto fail1;
}
}
}
if ((flags & RFMEM) == 0) {
vm2 = vmspace_fork(p1->p_vmspace, &mem_charged);
if (vm2 == NULL) {
error = ENOMEM;
goto fail1;
}
if (!swap_reserve(mem_charged)) {
/*
* The swap reservation failed. The accounting
* from the entries of the copied vm2 will be
* substracted in vmspace_free(), so force the
* reservation there.
*/
swap_reserve_force(mem_charged);
error = ENOMEM;
goto fail1;
}
} else
vm2 = NULL;
/*
* XXX: This is ugly; when we copy resource usage, we need to bump
* per-cred resource counters.
*/
newproc->p_ucred = p1->p_ucred;
/*
* Initialize resource accounting for the child process.
*/
error = racct_proc_fork(p1, newproc);
if (error != 0) {
error = EAGAIN;
goto fail1;
}
#ifdef MAC
mac_proc_init(newproc);
#endif
knlist_init_mtx(&newproc->p_klist, &newproc->p_mtx);
STAILQ_INIT(&newproc->p_ktr);
/* 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
* a nonprivileged user to use the last ten processes; don't let root
* exceed the limit. The variable nprocs is the current number of
* processes, maxproc is the limit.
1994-05-24 10:09:53 +00:00
*/
sx_xlock(&allproc_lock);
if ((nprocs >= maxproc - 10 && priv_check_cred(td->td_ucred,
PRIV_MAXPROC, 0) != 0) || nprocs >= maxproc) {
error = EAGAIN;
goto fail;
}
/*
* Increment the count of procs running with this uid. Don't allow
* a nonprivileged user to exceed their current limit.
*
* XXXRW: Can we avoid privilege here if it's not needed?
*/
error = priv_check_cred(td->td_ucred, PRIV_PROC_LIMIT, 0);
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);
}
if (ok) {
do_fork(td, flags, newproc, td2, vm2, pdflags);
/*
* Return child proc pointer to parent.
*/
*procp = newproc;
#ifdef PROCDESC
if (flags & RFPROCDESC) {
procdesc_finit(newproc->p_procdesc, fp_procdesc);
fdrop(fp_procdesc, td);
}
#endif
racct_proc_fork_done(newproc);
return (0);
}
error = EAGAIN;
fail:
sx_sunlock(&proctree_lock);
if (ppsratecheck(&lastfail, &curfail, 1))
printf("maxproc limit exceeded by uid %u (pid %d); see tuning(7) and login.conf(5)\n",
td->td_ucred->cr_ruid, p1->p_pid);
sx_xunlock(&allproc_lock);
#ifdef MAC
mac_proc_destroy(newproc);
#endif
racct_proc_exit(newproc);
fail1:
if (vm2 != NULL)
vmspace_free(vm2);
uma_zfree(proc_zone, newproc);
#ifdef PROCDESC
2013-02-17 11:47:01 +00:00
if ((flags & RFPROCDESC) != 0 && fp_procdesc != NULL) {
fdclose(td->td_proc->p_fd, fp_procdesc, *procdescp, td);
fdrop(fp_procdesc, td);
}
#endif
pause("fork", hz / 2);
return (error);
1994-05-24 10:09:53 +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(void (*callout)(void *, struct trapframe *), void *arg,
struct trapframe *frame)
{
struct proc *p;
struct thread *td;
struct thread *dtd;
td = curthread;
p = td->td_proc;
KASSERT(p->p_state == PRS_NORMAL, ("executing process is still new"));
CTR4(KTR_PROC, "fork_exit: new thread %p (td_sched %p, pid %d, %s)",
td, td->td_sched, p->p_pid, td->td_name);
sched_fork_exit(td);
/*
* 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 ((dtd = PCPU_GET(deadthread))) {
PCPU_SET(deadthread, NULL);
thread_stash(dtd);
}
thread_unlock(td);
/*
* cpu_set_fork_handler intercepts this function call to
* have this call a non-return function to stay in kernel mode.
* initproc has its own fork handler, but it does return.
*/
KASSERT(callout != NULL, ("NULL callout in fork_exit"));
callout(arg, frame);
/*
* 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",
td->td_name, p->p_pid);
kproc_exit(0);
}
mtx_assert(&Giant, MA_NOTOWNED);
if (p->p_sysent->sv_schedtail != NULL)
(p->p_sysent->sv_schedtail)(td);
}
/*
* 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
fork_return(struct thread *td, struct trapframe *frame)
{
struct proc *p, *dbg;
if (td->td_dbgflags & TDB_STOPATFORK) {
p = td->td_proc;
sx_xlock(&proctree_lock);
PROC_LOCK(p);
if ((p->p_pptr->p_flag & (P_TRACED | P_FOLLOWFORK)) ==
(P_TRACED | P_FOLLOWFORK)) {
/*
* If debugger still wants auto-attach for the
* parent's children, do it now.
*/
dbg = p->p_pptr->p_pptr;
p->p_flag |= P_TRACED;
p->p_oppid = p->p_pptr->p_pid;
proc_reparent(p, dbg);
sx_xunlock(&proctree_lock);
td->td_dbgflags |= TDB_CHILD;
ptracestop(td, SIGSTOP);
td->td_dbgflags &= ~TDB_CHILD;
} else {
/*
* ... otherwise clear the request.
*/
sx_xunlock(&proctree_lock);
td->td_dbgflags &= ~TDB_STOPATFORK;
cv_broadcast(&p->p_dbgwait);
}
PROC_UNLOCK(p);
}
userret(td, frame);
#ifdef KTRACE
if (KTRPOINT(td, KTR_SYSRET))
ktrsysret(SYS_fork, 0, 0);
#endif
}