static void
acpi_lid_notify_status_changed(void *arg)
{
struct acpi_lid_softc *sc;
struct acpi_softc *acpi_sc;
ACPI_FUNCTION_TRACE((char *)(uintptr_t)__func__);
sc = (struct acpi_lid_softc *)arg;
/*
* Evaluate _LID and check the return value, update lid status.
* Zero: The lid is closed
* Non-zero: The lid is open
*/
if (ACPI_FAILURE(acpi_EvaluateInteger(sc->lid_handle, "_LID", &sc->lid_status)))
return_VOID;
acpi_sc = acpi_device_get_parent_softc(sc->lid_dev);
if (acpi_sc == NULL) {
return_VOID;
}
ACPI_VPRINT(sc->lid_dev, acpi_sc,
"Lid %s\n", sc->lid_status ? "opened" : "closed");
if (sc->lid_status == 0) {
EVENTHANDLER_INVOKE(acpi_sleep_event, acpi_sc->acpi_lid_switch_sx);
} else {
EVENTHANDLER_INVOKE(acpi_wakeup_event, acpi_sc->acpi_lid_switch_sx);
}
return_VOID;
}
/*
* Shutdown the system cleanly to prepare for reboot, halt, or power off.
*/
void
kern_reboot(int howto)
{
static int once = 0;
#if defined(SMP)
/*
* Bind us to CPU 0 so that all shutdown code runs there. Some
* systems don't shutdown properly (i.e., ACPI power off) if we
* run on another processor.
*/
if (!SCHEDULER_STOPPED()) {
thread_lock(curthread);
sched_bind(curthread, 0);
thread_unlock(curthread);
KASSERT(PCPU_GET(cpuid) == 0, ("boot: not running on cpu 0"));
}
#endif
/* We're in the process of rebooting. */
rebooting = 1;
/* We are out of the debugger now. */
kdb_active = 0;
/*
* Do any callouts that should be done BEFORE syncing the filesystems.
*/
EVENTHANDLER_INVOKE(shutdown_pre_sync, howto);
/*
* Now sync filesystems
*/
if (!cold && (howto & RB_NOSYNC) == 0 && once == 0) {
once = 1;
bufshutdown(show_busybufs);
}
print_uptime();
cngrab();
/*
* Ok, now do things that assume all filesystem activity has
* been completed.
*/
EVENTHANDLER_INVOKE(shutdown_post_sync, howto);
if ((howto & (RB_HALT|RB_DUMP)) == RB_DUMP && !cold && !dumping)
doadump(TRUE);
/* Now that we're going to really halt the system... */
EVENTHANDLER_INVOKE(shutdown_final, howto);
for(;;) ; /* safety against shutdown_reset not working */
/* NOTREACHED */
}
static void
set_rootvnode(void)
{
struct proc *p;
if (VFS_ROOT(TAILQ_FIRST(&mountlist), LK_EXCLUSIVE, &rootvnode))
panic("Cannot find root vnode");
VOP_UNLOCK(rootvnode, 0);
p = curthread->td_proc;
FILEDESC_XLOCK(p->p_fd);
if (p->p_fd->fd_cdir != NULL)
vrele(p->p_fd->fd_cdir);
p->p_fd->fd_cdir = rootvnode;
VREF(rootvnode);
if (p->p_fd->fd_rdir != NULL)
vrele(p->p_fd->fd_rdir);
p->p_fd->fd_rdir = rootvnode;
VREF(rootvnode);
FILEDESC_XUNLOCK(p->p_fd);
EVENTHANDLER_INVOKE(mountroot);
}
static int
kern_do_pat(u_int utim)
{
int error;
if ((utim & WD_LASTVAL) != 0 && (utim & WD_INTERVAL) > 0)
return (EINVAL);
if ((utim & WD_LASTVAL) != 0) {
MPASS((wd_last_u & ~WD_INTERVAL) == 0);
utim &= ~WD_LASTVAL;
utim |= wd_last_u;
} else
wd_last_u = (utim & WD_INTERVAL);
if ((utim & WD_INTERVAL) == WD_TO_NEVER) {
utim = 0;
/* Assume all is well; watchdog signals failure. */
error = 0;
} else {
/* Assume no watchdog available; watchdog flags success */
error = EOPNOTSUPP;
}
EVENTHANDLER_INVOKE(watchdog_list, utim, &error);
return (error);
}
/* HVM mode suspension. */
static void
xctrl_suspend()
{
int suspend_cancelled;
EVENTHANDLER_INVOKE(power_suspend);
/*
* Be sure to hold Giant across DEVICE_SUSPEND/RESUME since non-MPSAFE
* drivers need this.
*/
mtx_lock(&Giant);
if (DEVICE_SUSPEND(root_bus) != 0) {
mtx_unlock(&Giant);
printf("%s: device_suspend failed\n", __func__);
return;
}
mtx_unlock(&Giant);
/*
* Prevent any races with evtchn_interrupt() handler.
*/
disable_intr();
irq_suspend();
suspend_cancelled = HYPERVISOR_suspend(0);
if (suspend_cancelled)
irq_resume();
else
xenpci_resume();
/*
* Re-enable interrupts and put the scheduler back to normal.
*/
enable_intr();
/*
* FreeBSD really needs to add DEVICE_SUSPEND_CANCEL or
* similar.
*/
mtx_lock(&Giant);
if (!suspend_cancelled)
DEVICE_RESUME(root_bus);
mtx_unlock(&Giant);
EVENTHANDLER_INVOKE(power_resume);
}
void
tcp_offload_listen_start(struct tcpcb *tp)
{
INP_WLOCK_ASSERT(tp->t_inpcb);
EVENTHANDLER_INVOKE(tcp_offload_listen_start, tp);
}
void
__tesla_event_function_prologue_syscallret(void **tesla_data,
struct thread *td, int error, struct syscall_args *sa)
{
EVENTHANDLER_INVOKE(tesla_event_function_prologue_syscallret,
tesla_data, td, error, sa);
}
static void
if_low_power_nwk_ev_callback(void *arg)
{
struct if_low_power_ev_args *if_low_power_ev_args =
(struct if_low_power_ev_args *)arg;
EVENTHANDLER_INVOKE(&if_low_power_evhdlr_ctx,
if_low_power_event,
if_low_power_ev_args->ifp,
if_low_power_ev_args->event_code);
}
/*
* Prepare a proc for use.
*/
static int
proc_ctor(void *mem, int size, void *arg, int flags)
{
struct proc *p;
p = (struct proc *)mem;
SDT_PROBE(proc, kernel, ctor , entry, p, size, arg, flags, 0);
EVENTHANDLER_INVOKE(process_ctor, p);
SDT_PROBE(proc, kernel, ctor , return, p, size, arg, flags, 0);
return (0);
}
static void
acpi_lid_notify_status_changed(void *arg)
{
struct acpi_lid_softc *sc;
struct acpi_softc *acpi_sc;
ACPI_STATUS status;
ACPI_FUNCTION_TRACE((char *)(uintptr_t)__func__);
sc = (struct acpi_lid_softc *)arg;
ACPI_SERIAL_BEGIN(lid);
/*
* Evaluate _LID and check the return value, update lid status.
* Zero: The lid is closed
* Non-zero: The lid is open
*/
status = acpi_GetInteger(sc->lid_handle, "_LID", &sc->lid_status);
if (ACPI_FAILURE(status))
goto out;
acpi_sc = acpi_device_get_parent_softc(sc->lid_dev);
if (acpi_sc == NULL)
goto out;
ACPI_VPRINT(sc->lid_dev, acpi_sc, "Lid %s\n",
sc->lid_status ? "opened" : "closed");
acpi_UserNotify("Lid", sc->lid_handle, sc->lid_status);
if (sc->lid_status == 0)
EVENTHANDLER_INVOKE(acpi_sleep_event, acpi_sc->acpi_lid_switch_sx);
else
EVENTHANDLER_INVOKE(acpi_wakeup_event, acpi_sc->acpi_lid_switch_sx);
out:
ACPI_SERIAL_END(lid);
return_VOID;
}
/*
* Allow userspace to directly trigger the VM drain routine for testing
* purposes.
*/
static int
debug_vm_lowmem(SYSCTL_HANDLER_ARGS)
{
int error, i;
i = 0;
error = sysctl_handle_int(oidp, &i, 0, req);
if (error)
return (error);
if ((i & ~(VM_LOW_KMEM | VM_LOW_PAGES)) != 0)
return (EINVAL);
if (i != 0)
EVENTHANDLER_INVOKE(vm_lowmem, i);
return (0);
}
/*
* UMA should ensure that this function is never called.
* Freeing a proc structure would violate type stability.
*/
static void
proc_fini(void *mem, int size)
{
#ifdef notnow
struct proc *p;
p = (struct proc *)mem;
EVENTHANDLER_INVOKE(process_fini, p);
pstats_free(p->p_stats);
thread_free(FIRST_THREAD_IN_PROC(p));
mtx_destroy(&p->p_mtx);
if (p->p_ksi != NULL)
ksiginfo_free(p->p_ksi);
#else
panic("proc reclaimed");
#endif
}
void
power_profile_set_state(int state)
{
int changed;
if (state != power_profile_state) {
power_profile_state = state;
changed = 1;
kprintf("system power profile changed to '%s'\n",
(state == POWER_PROFILE_PERFORMANCE) ? "performance" : "economy");
} else {
changed = 0;
}
if (changed)
EVENTHANDLER_INVOKE(power_profile_change, 0);
}
static int
sysctl_nmbufs(SYSCTL_HANDLER_ARGS)
{
int error, newnmbufs;
newnmbufs = nmbufs;
error = sysctl_handle_int(oidp, &newnmbufs, 0, req);
if (error == 0 && req->newptr && newnmbufs != nmbufs) {
if (newnmbufs > nmbufs) {
nmbufs = newnmbufs;
nmbufs = uma_zone_set_max(zone_mbuf, nmbufs);
EVENTHANDLER_INVOKE(nmbufs_change);
} else
error = EINVAL;
}
return (error);
}
static int
sysctl_nmbclusters(SYSCTL_HANDLER_ARGS)
{
int error, newnmbclusters;
newnmbclusters = nmbclusters;
error = sysctl_handle_int(oidp, &newnmbclusters, 0, req);
if (error == 0 && req->newptr) {
if (newnmbclusters > nmbclusters) {
nmbclusters = newnmbclusters;
uma_zone_set_max(zone_clust, nmbclusters);
EVENTHANDLER_INVOKE(nmbclusters_change);
} else
error = EINVAL;
}
return (error);
}
/*
* Initialize type-stable parts of a proc (when newly created).
*/
static int
proc_init(void *mem, int size, int flags)
{
struct proc *p;
p = (struct proc *)mem;
SDT_PROBE(proc, kernel, init, entry, p, size, flags, 0, 0);
p->p_sched = (struct p_sched *)&p[1];
bzero(&p->p_mtx, sizeof(struct mtx));
mtx_init(&p->p_mtx, "process lock", NULL, MTX_DEF | MTX_DUPOK);
mtx_init(&p->p_slock, "process slock", NULL, MTX_SPIN | MTX_RECURSE);
cv_init(&p->p_pwait, "ppwait");
TAILQ_INIT(&p->p_threads); /* all threads in proc */
EVENTHANDLER_INVOKE(process_init, p);
p->p_stats = pstats_alloc();
SDT_PROBE(proc, kernel, init, return, p, size, flags, 0, 0);
return (0);
}
static int
sysctl_nmbclusters(SYSCTL_HANDLER_ARGS)
{
int error, newnmbclusters;
newnmbclusters = nmbclusters;
error = sysctl_handle_int(oidp, &newnmbclusters, 0, req);
if (error == 0 && req->newptr && newnmbclusters != nmbclusters) {
if (newnmbclusters > nmbclusters &&
nmbufs >= nmbclusters + nmbjumbop + nmbjumbo9 + nmbjumbo16) {
nmbclusters = newnmbclusters;
nmbclusters = uma_zone_set_max(zone_clust, nmbclusters);
#ifndef __rtems__
EVENTHANDLER_INVOKE(nmbclusters_change);
#endif /* __rtems__ */
} else
error = EINVAL;
}
return (error);
}
/*
* Reclaim a proc after use.
*/
static void
proc_dtor(void *mem, int size, void *arg)
{
struct proc *p;
struct thread *td;
/* INVARIANTS checks go here */
p = (struct proc *)mem;
td = FIRST_THREAD_IN_PROC(p);
SDT_PROBE(proc, kernel, dtor, entry, p, size, arg, td, 0);
if (td != NULL) {
#ifdef INVARIANTS
KASSERT((p->p_numthreads == 1),
("bad number of threads in exiting process"));
KASSERT(STAILQ_EMPTY(&p->p_ktr), ("proc_dtor: non-empty p_ktr"));
#endif
/* Free all OSD associated to this thread. */
osd_thread_exit(td);
}
EVENTHANDLER_INVOKE(process_dtor, p);
if (p->p_ksi != NULL)
KASSERT(! KSI_ONQ(p->p_ksi), ("SIGCHLD queue"));
SDT_PROBE(proc, kernel, dtor, return, p, size, arg, 0, 0);
}
/*
* Go through the rigmarole of shutting down..
* this used to be in machdep.c but I'll be dammned if I could see
* anything machine dependant in it.
*/
static void
boot(int howto)
{
/*
* Get rid of any user scheduler baggage and then give
* us a high priority.
*/
if (curthread->td_release)
curthread->td_release(curthread);
lwkt_setpri_self(TDPRI_MAX);
/* collect extra flags that shutdown_nice might have set */
howto |= shutdown_howto;
#ifdef SMP
/*
* We really want to shutdown on the BSP. Subsystems such as ACPI
* can't power-down the box otherwise.
*/
if (smp_active_mask > 1) {
kprintf("boot() called on cpu#%d\n", mycpu->gd_cpuid);
}
if (panicstr == NULL && mycpu->gd_cpuid != 0) {
kprintf("Switching to cpu #0 for shutdown\n");
lwkt_setcpu_self(globaldata_find(0));
}
#endif
/*
* Do any callouts that should be done BEFORE syncing the filesystems.
*/
EVENTHANDLER_INVOKE(shutdown_pre_sync, howto);
/*
* Try to get rid of any remaining FS references. The calling
* process, proc0, and init may still hold references. The
* VFS cache subsystem may still hold a root reference to root.
*
* XXX this needs work. We really need to SIGSTOP all remaining
* processes in order to avoid blowups due to proc0's filesystem
* references going away. For now just make sure that the init
* process is stopped.
*/
if (panicstr == NULL) {
shutdown_cleanup_proc(curproc);
shutdown_cleanup_proc(&proc0);
if (initproc) {
if (initproc != curproc) {
ksignal(initproc, SIGSTOP);
tsleep(boot, 0, "shutdn", hz / 20);
}
shutdown_cleanup_proc(initproc);
}
vfs_cache_setroot(NULL, NULL);
}
/*
* Now sync filesystems
*/
if (!cold && (howto & RB_NOSYNC) == 0 && waittime < 0) {
int iter, nbusy, pbusy;
waittime = 0;
kprintf("\nsyncing disks... ");
sys_sync(NULL); /* YYY was sync(&proc0, NULL). why proc0 ? */
/*
* With soft updates, some buffers that are
* written will be remarked as dirty until other
* buffers are written.
*/
for (iter = pbusy = 0; iter < 20; iter++) {
nbusy = scan_all_buffers(shutdown_busycount1, NULL);
if (nbusy == 0)
break;
kprintf("%d ", nbusy);
if (nbusy < pbusy)
iter = 0;
pbusy = nbusy;
/*
* XXX:
* Process soft update work queue if buffers don't sync
* after 6 iterations by permitting the syncer to run.
*/
if (iter > 5)
bio_ops_sync(NULL);
sys_sync(NULL); /* YYY was sync(&proc0, NULL). why proc0 ? */
tsleep(boot, 0, "shutdn", hz * iter / 20 + 1);
}
kprintf("\n");
/*
* Count only busy local buffers to prevent forcing
* a fsck if we're just a client of a wedged NFS server
*/
nbusy = scan_all_buffers(shutdown_busycount2, NULL);
if (nbusy) {
/*
* Failed to sync all blocks. Indicate this and don't
* unmount filesystems (thus forcing an fsck on reboot).
*/
kprintf("giving up on %d buffers\n", nbusy);
#ifdef DDB
if (debugger_on_panic)
Debugger("busy buffer problem");
#endif /* DDB */
tsleep(boot, 0, "shutdn", hz * 5 + 1);
} else {
kprintf("done\n");
/*
* Unmount filesystems
*/
if (panicstr == NULL)
vfs_unmountall();
}
tsleep(boot, 0, "shutdn", hz / 10 + 1);
}
print_uptime();
/*
* Dump before doing post_sync shutdown ops
*/
crit_enter();
if ((howto & (RB_HALT|RB_DUMP)) == RB_DUMP && !cold) {
dumpsys();
}
/*
* Ok, now do things that assume all filesystem activity has
* been completed. This will also call the device shutdown
* methods.
*/
EVENTHANDLER_INVOKE(shutdown_post_sync, howto);
/* Now that we're going to really halt the system... */
EVENTHANDLER_INVOKE(shutdown_final, howto);
for(;;) ; /* safety against shutdown_reset not working */
/* NOTREACHED */
}
static void
xctrl_suspend()
{
#ifdef SMP
cpuset_t cpu_suspend_map;
#endif
int suspend_cancelled;
EVENTHANDLER_INVOKE(power_suspend);
if (smp_started) {
thread_lock(curthread);
sched_bind(curthread, 0);
thread_unlock(curthread);
}
KASSERT((PCPU_GET(cpuid) == 0), ("Not running on CPU#0"));
/*
* Clear our XenStore node so the toolstack knows we are
* responding to the suspend request.
*/
xs_write(XST_NIL, "control", "shutdown", "");
/*
* Be sure to hold Giant across DEVICE_SUSPEND/RESUME since non-MPSAFE
* drivers need this.
*/
mtx_lock(&Giant);
if (DEVICE_SUSPEND(root_bus) != 0) {
mtx_unlock(&Giant);
printf("%s: device_suspend failed\n", __func__);
return;
}
mtx_unlock(&Giant);
#ifdef SMP
CPU_ZERO(&cpu_suspend_map); /* silence gcc */
if (smp_started) {
/*
* Suspend other CPUs. This prevents IPIs while we
* are resuming, and will allow us to reset per-cpu
* vcpu_info on resume.
*/
cpu_suspend_map = all_cpus;
CPU_CLR(PCPU_GET(cpuid), &cpu_suspend_map);
if (!CPU_EMPTY(&cpu_suspend_map))
suspend_cpus(cpu_suspend_map);
}
#endif
/*
* Prevent any races with evtchn_interrupt() handler.
*/
disable_intr();
intr_suspend();
xen_hvm_suspend();
suspend_cancelled = HYPERVISOR_suspend(0);
xen_hvm_resume(suspend_cancelled != 0);
intr_resume(suspend_cancelled != 0);
enable_intr();
/*
* Reset grant table info.
*/
gnttab_resume(NULL);
#ifdef SMP
/* Send an IPI_BITMAP in case there are pending bitmap IPIs. */
lapic_ipi_vectored(IPI_BITMAP_VECTOR, APIC_IPI_DEST_ALL);
if (smp_started && !CPU_EMPTY(&cpu_suspend_map)) {
/*
* Now that event channels have been initialized,
* resume CPUs.
*/
resume_cpus(cpu_suspend_map);
}
#endif
/*
* FreeBSD really needs to add DEVICE_SUSPEND_CANCEL or
* similar.
*/
mtx_lock(&Giant);
DEVICE_RESUME(root_bus);
mtx_unlock(&Giant);
if (smp_started) {
thread_lock(curthread);
sched_unbind(curthread);
thread_unlock(curthread);
}
EVENTHANDLER_INVOKE(power_resume);
if (bootverbose)
printf("System resumed after suspension\n");
}
/*
* Exit: deallocate address space and other resources, change proc state to
* zombie, and unlink proc from allproc and parent's lists. Save exit status
* and rusage for wait(). Check for child processes and orphan them.
*/
void
exit1(struct thread *td, int rv)
{
struct proc *p, *nq, *q;
struct vnode *vtmp;
struct vnode *ttyvp = NULL;
struct plimit *plim;
mtx_assert(&Giant, MA_NOTOWNED);
p = td->td_proc;
/*
* XXX in case we're rebooting we just let init die in order to
* work around an unsolved stack overflow seen very late during
* shutdown on sparc64 when the gmirror worker process exists.
*/
if (p == initproc && rebooting == 0) {
printf("init died (signal %d, exit %d)\n",
WTERMSIG(rv), WEXITSTATUS(rv));
panic("Going nowhere without my init!");
}
/*
* MUST abort all other threads before proceeding past here.
*/
PROC_LOCK(p);
while (p->p_flag & P_HADTHREADS) {
/*
* First check if some other thread got here before us.
* If so, act appropriately: exit or suspend.
*/
thread_suspend_check(0);
/*
* Kill off the other threads. This requires
* some co-operation from other parts of the kernel
* so it may not be instantaneous. With this state set
* any thread entering the kernel from userspace will
* thread_exit() in trap(). Any thread attempting to
* sleep will return immediately with EINTR or EWOULDBLOCK
* which will hopefully force them to back out to userland
* freeing resources as they go. Any thread attempting
* to return to userland will thread_exit() from userret().
* thread_exit() will unsuspend us when the last of the
* other threads exits.
* If there is already a thread singler after resumption,
* calling thread_single will fail; in that case, we just
* re-check all suspension request, the thread should
* either be suspended there or exit.
*/
if (!thread_single(SINGLE_EXIT))
break;
/*
* All other activity in this process is now stopped.
* Threading support has been turned off.
*/
}
KASSERT(p->p_numthreads == 1,
("exit1: proc %p exiting with %d threads", p, p->p_numthreads));
racct_sub(p, RACCT_NTHR, 1);
/*
* Wakeup anyone in procfs' PIOCWAIT. They should have a hold
* on our vmspace, so we should block below until they have
* released their reference to us. Note that if they have
* requested S_EXIT stops we will block here until they ack
* via PIOCCONT.
*/
_STOPEVENT(p, S_EXIT, rv);
/*
* Ignore any pending request to stop due to a stop signal.
* Once P_WEXIT is set, future requests will be ignored as
* well.
*/
p->p_flag &= ~P_STOPPED_SIG;
KASSERT(!P_SHOULDSTOP(p), ("exiting process is stopped"));
/*
* Note that we are exiting and do another wakeup of anyone in
* PIOCWAIT in case they aren't listening for S_EXIT stops or
* decided to wait again after we told them we are exiting.
*/
p->p_flag |= P_WEXIT;
wakeup(&p->p_stype);
/*
* Wait for any processes that have a hold on our vmspace to
* release their reference.
*/
while (p->p_lock > 0)
msleep(&p->p_lock, &p->p_mtx, PWAIT, "exithold", 0);
p->p_xstat = rv; /* Let event handler change exit status */
PROC_UNLOCK(p);
/* Drain the limit callout while we don't have the proc locked */
callout_drain(&p->p_limco);
#ifdef AUDIT
/*
* The Sun BSM exit token contains two components: an exit status as
* passed to exit(), and a return value to indicate what sort of exit
* it was. The exit status is WEXITSTATUS(rv), but it's not clear
* what the return value is.
*/
AUDIT_ARG_EXIT(WEXITSTATUS(rv), 0);
AUDIT_SYSCALL_EXIT(0, td);
#endif
/* Are we a task leader? */
if (p == p->p_leader) {
mtx_lock(&ppeers_lock);
q = p->p_peers;
while (q != NULL) {
PROC_LOCK(q);
kern_psignal(q, SIGKILL);
PROC_UNLOCK(q);
q = q->p_peers;
}
while (p->p_peers != NULL)
msleep(p, &ppeers_lock, PWAIT, "exit1", 0);
mtx_unlock(&ppeers_lock);
}
/*
* Check if any loadable modules need anything done at process exit.
* E.g. SYSV IPC stuff
* XXX what if one of these generates an error?
*/
EVENTHANDLER_INVOKE(process_exit, p);
/*
* If parent is waiting for us to exit or exec,
* P_PPWAIT is set; we will wakeup the parent below.
*/
PROC_LOCK(p);
rv = p->p_xstat; /* Event handler could change exit status */
stopprofclock(p);
p->p_flag &= ~(P_TRACED | P_PPWAIT | P_PPTRACE);
/*
* Stop the real interval timer. If the handler is currently
* executing, prevent it from rearming itself and let it finish.
*/
if (timevalisset(&p->p_realtimer.it_value) &&
callout_stop(&p->p_itcallout) == 0) {
timevalclear(&p->p_realtimer.it_interval);
msleep(&p->p_itcallout, &p->p_mtx, PWAIT, "ritwait", 0);
KASSERT(!timevalisset(&p->p_realtimer.it_value),
("realtime timer is still armed"));
}
PROC_UNLOCK(p);
/*
* Reset any sigio structures pointing to us as a result of
* F_SETOWN with our pid.
*/
funsetownlst(&p->p_sigiolst);
/*
* If this process has an nlminfo data area (for lockd), release it
*/
if (nlminfo_release_p != NULL && p->p_nlminfo != NULL)
(*nlminfo_release_p)(p);
/*
* Close open files and release open-file table.
* This may block!
*/
fdescfree(td);
/*
* If this thread tickled GEOM, we need to wait for the giggling to
* stop before we return to userland
*/
if (td->td_pflags & TDP_GEOM)
g_waitidle();
/*
* Remove ourself from our leader's peer list and wake our leader.
*/
mtx_lock(&ppeers_lock);
if (p->p_leader->p_peers) {
q = p->p_leader;
while (q->p_peers != p)
q = q->p_peers;
q->p_peers = p->p_peers;
wakeup(p->p_leader);
}
mtx_unlock(&ppeers_lock);
vmspace_exit(td);
sx_xlock(&proctree_lock);
if (SESS_LEADER(p)) {
struct session *sp = p->p_session;
struct tty *tp;
/*
* s_ttyp is not zero'd; we use this to indicate that
* the session once had a controlling terminal. (for
* logging and informational purposes)
*/
SESS_LOCK(sp);
ttyvp = sp->s_ttyvp;
tp = sp->s_ttyp;
sp->s_ttyvp = NULL;
sp->s_ttydp = NULL;
sp->s_leader = NULL;
SESS_UNLOCK(sp);
/*
* Signal foreground pgrp and revoke access to
* controlling terminal if it has not been revoked
* already.
*
* Because the TTY may have been revoked in the mean
* time and could already have a new session associated
* with it, make sure we don't send a SIGHUP to a
* foreground process group that does not belong to this
* session.
*/
if (tp != NULL) {
tty_lock(tp);
if (tp->t_session == sp)
tty_signal_pgrp(tp, SIGHUP);
tty_unlock(tp);
}
if (ttyvp != NULL) {
sx_xunlock(&proctree_lock);
if (vn_lock(ttyvp, LK_EXCLUSIVE) == 0) {
VOP_REVOKE(ttyvp, REVOKEALL);
VOP_UNLOCK(ttyvp, 0);
}
sx_xlock(&proctree_lock);
}
}
fixjobc(p, p->p_pgrp, 0);
sx_xunlock(&proctree_lock);
(void)acct_process(td);
/* Release the TTY now we've unlocked everything. */
if (ttyvp != NULL)
vrele(ttyvp);
#ifdef KTRACE
ktrprocexit(td);
#endif
/*
* Release reference to text vnode
*/
if ((vtmp = p->p_textvp) != NULL) {
p->p_textvp = NULL;
vrele(vtmp);
}
/*
* Release our limits structure.
*/
plim = p->p_limit;
p->p_limit = NULL;
lim_free(plim);
tidhash_remove(td);
/*
* Remove proc from allproc queue and pidhash chain.
* Place onto zombproc. Unlink from parent's child list.
*/
sx_xlock(&allproc_lock);
LIST_REMOVE(p, p_list);
LIST_INSERT_HEAD(&zombproc, p, p_list);
LIST_REMOVE(p, p_hash);
sx_xunlock(&allproc_lock);
/*
* Call machine-dependent code to release any
* machine-dependent resources other than the address space.
* The address space is released by "vmspace_exitfree(p)" in
* vm_waitproc().
*/
cpu_exit(td);
WITNESS_WARN(WARN_PANIC, NULL, "process (pid %d) exiting", p->p_pid);
/*
* Reparent all of our children to init.
*/
sx_xlock(&proctree_lock);
q = LIST_FIRST(&p->p_children);
if (q != NULL) /* only need this if any child is S_ZOMB */
wakeup(initproc);
for (; q != NULL; q = nq) {
nq = LIST_NEXT(q, p_sibling);
PROC_LOCK(q);
proc_reparent(q, initproc);
q->p_sigparent = SIGCHLD;
/*
* Traced processes are killed
* since their existence means someone is screwing up.
*/
if (q->p_flag & P_TRACED) {
struct thread *temp;
/*
* Since q was found on our children list, the
* proc_reparent() call moved q to the orphan
* list due to present P_TRACED flag. Clear
* orphan link for q now while q is locked.
*/
clear_orphan(q);
q->p_flag &= ~(P_TRACED | P_STOPPED_TRACE);
FOREACH_THREAD_IN_PROC(q, temp)
temp->td_dbgflags &= ~TDB_SUSPEND;
kern_psignal(q, SIGKILL);
}
PROC_UNLOCK(q);
}
/*
* Also get rid of our orphans.
*/
while ((q = LIST_FIRST(&p->p_orphans)) != NULL) {
PROC_LOCK(q);
clear_orphan(q);
PROC_UNLOCK(q);
}
/* Save exit status. */
PROC_LOCK(p);
p->p_xthread = td;
/* Tell the prison that we are gone. */
prison_proc_free(p->p_ucred->cr_prison);
#ifdef KDTRACE_HOOKS
/*
* Tell the DTrace fasttrap provider about the exit if it
* has declared an interest.
*/
if (dtrace_fasttrap_exit)
dtrace_fasttrap_exit(p);
#endif
/*
* Notify interested parties of our demise.
*/
KNOTE_LOCKED(&p->p_klist, NOTE_EXIT);
#ifdef KDTRACE_HOOKS
int reason = CLD_EXITED;
if (WCOREDUMP(rv))
reason = CLD_DUMPED;
else if (WIFSIGNALED(rv))
reason = CLD_KILLED;
SDT_PROBE(proc, kernel, , exit, reason, 0, 0, 0, 0);
#endif
/*
* Just delete all entries in the p_klist. At this point we won't
* report any more events, and there are nasty race conditions that
* can beat us if we don't.
*/
knlist_clear(&p->p_klist, 1);
/*
* If this is a process with a descriptor, we may not need to deliver
* a signal to the parent. proctree_lock is held over
* procdesc_exit() to serialize concurrent calls to close() and
* exit().
*/
if (p->p_procdesc == NULL || procdesc_exit(p)) {
/*
* Notify parent that we're gone. If parent has the
* PS_NOCLDWAIT flag set, or if the handler is set to SIG_IGN,
* notify process 1 instead (and hope it will handle this
* situation).
*/
PROC_LOCK(p->p_pptr);
mtx_lock(&p->p_pptr->p_sigacts->ps_mtx);
if (p->p_pptr->p_sigacts->ps_flag &
(PS_NOCLDWAIT | PS_CLDSIGIGN)) {
struct proc *pp;
mtx_unlock(&p->p_pptr->p_sigacts->ps_mtx);
pp = p->p_pptr;
PROC_UNLOCK(pp);
proc_reparent(p, initproc);
p->p_sigparent = SIGCHLD;
PROC_LOCK(p->p_pptr);
/*
* Notify parent, so in case he was wait(2)ing or
* executing waitpid(2) with our pid, he will
* continue.
*/
wakeup(pp);
} else
mtx_unlock(&p->p_pptr->p_sigacts->ps_mtx);
if (p->p_pptr == initproc)
kern_psignal(p->p_pptr, SIGCHLD);
else if (p->p_sigparent != 0) {
if (p->p_sigparent == SIGCHLD)
childproc_exited(p);
else /* LINUX thread */
kern_psignal(p->p_pptr, p->p_sigparent);
}
} else
PROC_LOCK(p->p_pptr);
sx_xunlock(&proctree_lock);
/*
* The state PRS_ZOMBIE prevents other proesses from sending
* signal to the process, to avoid memory leak, we free memory
* for signal queue at the time when the state is set.
*/
sigqueue_flush(&p->p_sigqueue);
sigqueue_flush(&td->td_sigqueue);
/*
* We have to wait until after acquiring all locks before
* changing p_state. We need to avoid all possible context
* switches (including ones from blocking on a mutex) while
* marked as a zombie. We also have to set the zombie state
* before we release the parent process' proc lock to avoid
* a lost wakeup. So, we first call wakeup, then we grab the
* sched lock, update the state, and release the parent process'
* proc lock.
*/
wakeup(p->p_pptr);
cv_broadcast(&p->p_pwait);
sched_exit(p->p_pptr, td);
PROC_SLOCK(p);
p->p_state = PRS_ZOMBIE;
PROC_UNLOCK(p->p_pptr);
/*
* Hopefully no one will try to deliver a signal to the process this
* late in the game.
*/
knlist_destroy(&p->p_klist);
/*
* Save our children's rusage information in our exit rusage.
*/
ruadd(&p->p_ru, &p->p_rux, &p->p_stats->p_cru, &p->p_crux);
/*
* Make sure the scheduler takes this thread out of its tables etc.
* This will also release this thread's reference to the ucred.
* Other thread parts to release include pcb bits and such.
*/
thread_exit();
}
void
vfs_mountroot(void)
{
struct mount *mp;
struct sbuf *sb;
struct thread *td;
time_t timebase;
int error;
td = curthread;
vfs_mountroot_wait();
sb = sbuf_new_auto();
vfs_mountroot_conf0(sb);
sbuf_finish(sb);
error = vfs_mountroot_devfs(td, &mp);
while (!error) {
error = vfs_mountroot_parse(sb, mp);
if (!error) {
error = vfs_mountroot_shuffle(td, mp);
if (!error) {
sbuf_clear(sb);
error = vfs_mountroot_readconf(td, sb);
sbuf_finish(sb);
}
}
}
sbuf_delete(sb);
/*
* Iterate over all currently mounted file systems and use
* the time stamp found to check and/or initialize the RTC.
* Call inittodr() only once and pass it the largest of the
* timestamps we encounter.
*/
timebase = 0;
mtx_lock(&mountlist_mtx);
mp = TAILQ_FIRST(&mountlist);
while (mp != NULL) {
if (mp->mnt_time > timebase)
timebase = mp->mnt_time;
mp = TAILQ_NEXT(mp, mnt_list);
}
mtx_unlock(&mountlist_mtx);
inittodr(timebase);
/* Keep prison0's root in sync with the global rootvnode. */
mtx_lock(&prison0.pr_mtx);
prison0.pr_root = rootvnode;
vref(prison0.pr_root);
mtx_unlock(&prison0.pr_mtx);
mtx_lock(&mountlist_mtx);
atomic_store_rel_int(&root_mount_complete, 1);
wakeup(&root_mount_complete);
mtx_unlock(&mountlist_mtx);
EVENTHANDLER_INVOKE(mountroot);
}
/*
* Shutdown the system cleanly to prepare for reboot, halt, or power off.
*/
void
kern_reboot(int howto)
{
static int first_buf_printf = 1;
#if defined(SMP)
/*
* Bind us to CPU 0 so that all shutdown code runs there. Some
* systems don't shutdown properly (i.e., ACPI power off) if we
* run on another processor.
*/
if (!SCHEDULER_STOPPED()) {
thread_lock(curthread);
sched_bind(curthread, 0);
thread_unlock(curthread);
KASSERT(PCPU_GET(cpuid) == 0, ("boot: not running on cpu 0"));
}
#endif
/* We're in the process of rebooting. */
rebooting = 1;
/* collect extra flags that shutdown_nice might have set */
howto |= shutdown_howto;
/* We are out of the debugger now. */
kdb_active = 0;
/*
* Do any callouts that should be done BEFORE syncing the filesystems.
*/
EVENTHANDLER_INVOKE(shutdown_pre_sync, howto);
/*
* Now sync filesystems
*/
if (!cold && (howto & RB_NOSYNC) == 0 && waittime < 0) {
register struct buf *bp;
int iter, nbusy, pbusy;
#ifndef PREEMPTION
int subiter;
#endif
waittime = 0;
wdog_kern_pat(WD_LASTVAL);
sys_sync(curthread, NULL);
/*
* With soft updates, some buffers that are
* written will be remarked as dirty until other
* buffers are written.
*/
for (iter = pbusy = 0; iter < 20; iter++) {
nbusy = 0;
for (bp = &buf[nbuf]; --bp >= buf; )
if (isbufbusy(bp))
nbusy++;
if (nbusy == 0) {
if (first_buf_printf)
printf("All buffers synced.");
break;
}
if (first_buf_printf) {
printf("Syncing disks, buffers remaining... ");
first_buf_printf = 0;
}
printf("%d ", nbusy);
if (nbusy < pbusy)
iter = 0;
pbusy = nbusy;
wdog_kern_pat(WD_LASTVAL);
sys_sync(curthread, NULL);
#ifdef PREEMPTION
/*
* Drop Giant and spin for a while to allow
* interrupt threads to run.
*/
DROP_GIANT();
DELAY(50000 * iter);
PICKUP_GIANT();
#else
/*
* Drop Giant and context switch several times to
* allow interrupt threads to run.
*/
DROP_GIANT();
for (subiter = 0; subiter < 50 * iter; subiter++) {
thread_lock(curthread);
mi_switch(SW_VOL, NULL);
thread_unlock(curthread);
DELAY(1000);
}
PICKUP_GIANT();
#endif
}
printf("\n");
/*
* Count only busy local buffers to prevent forcing
* a fsck if we're just a client of a wedged NFS server
*/
nbusy = 0;
for (bp = &buf[nbuf]; --bp >= buf; ) {
if (isbufbusy(bp)) {
#if 0
/* XXX: This is bogus. We should probably have a BO_REMOTE flag instead */
if (bp->b_dev == NULL) {
TAILQ_REMOVE(&mountlist,
bp->b_vp->v_mount, mnt_list);
continue;
}
#endif
nbusy++;
if (show_busybufs > 0) {
printf(
"%d: buf:%p, vnode:%p, flags:%0x, blkno:%jd, lblkno:%jd, buflock:",
nbusy, bp, bp->b_vp, bp->b_flags,
(intmax_t)bp->b_blkno,
(intmax_t)bp->b_lblkno);
BUF_LOCKPRINTINFO(bp);
if (show_busybufs > 1)
vn_printf(bp->b_vp,
"vnode content: ");
}
}
}
if (nbusy) {
/*
* Failed to sync all blocks. Indicate this and don't
* unmount filesystems (thus forcing an fsck on reboot).
*/
printf("Giving up on %d buffers\n", nbusy);
DELAY(5000000); /* 5 seconds */
} else {
if (!first_buf_printf)
printf("Final sync complete\n");
/*
* Unmount filesystems
*/
if (panicstr == 0)
vfs_unmountall();
}
swapoff_all();
DELAY(100000); /* wait for console output to finish */
}
print_uptime();
cngrab();
/*
* Ok, now do things that assume all filesystem activity has
* been completed.
*/
EVENTHANDLER_INVOKE(shutdown_post_sync, howto);
if ((howto & (RB_HALT|RB_DUMP)) == RB_DUMP && !cold && !dumping)
doadump(TRUE);
/* Now that we're going to really halt the system... */
EVENTHANDLER_INVOKE(shutdown_final, howto);
for(;;) ; /* safety against shutdown_reset not working */
/* NOTREACHED */
}
/* Full PV mode suspension. */
static void
xctrl_suspend()
{
int i, j, k, fpp, suspend_cancelled;
unsigned long max_pfn, start_info_mfn;
EVENTHANDLER_INVOKE(power_suspend);
#ifdef SMP
struct thread *td;
cpuset_t map;
u_int cpuid;
/*
* Bind us to CPU 0 and stop any other VCPUs.
*/
td = curthread;
thread_lock(td);
sched_bind(td, 0);
thread_unlock(td);
cpuid = PCPU_GET(cpuid);
KASSERT(cpuid == 0, ("xen_suspend: not running on cpu 0"));
map = all_cpus;
CPU_CLR(cpuid, &map);
CPU_NAND(&map, &stopped_cpus);
if (!CPU_EMPTY(&map))
stop_cpus(map);
#endif
/*
* Be sure to hold Giant across DEVICE_SUSPEND/RESUME since non-MPSAFE
* drivers need this.
*/
mtx_lock(&Giant);
if (DEVICE_SUSPEND(root_bus) != 0) {
mtx_unlock(&Giant);
printf("%s: device_suspend failed\n", __func__);
#ifdef SMP
if (!CPU_EMPTY(&map))
restart_cpus(map);
#endif
return;
}
mtx_unlock(&Giant);
local_irq_disable();
xencons_suspend();
gnttab_suspend();
intr_suspend();
max_pfn = HYPERVISOR_shared_info->arch.max_pfn;
void *shared_info = HYPERVISOR_shared_info;
HYPERVISOR_shared_info = NULL;
pmap_kremove((vm_offset_t) shared_info);
PT_UPDATES_FLUSH();
xen_start_info->store_mfn = MFNTOPFN(xen_start_info->store_mfn);
xen_start_info->console.domU.mfn = MFNTOPFN(xen_start_info->console.domU.mfn);
/*
* We'll stop somewhere inside this hypercall. When it returns,
* we'll start resuming after the restore.
*/
start_info_mfn = VTOMFN(xen_start_info);
pmap_suspend();
suspend_cancelled = HYPERVISOR_suspend(start_info_mfn);
pmap_resume();
pmap_kenter_ma((vm_offset_t) shared_info, xen_start_info->shared_info);
HYPERVISOR_shared_info = shared_info;
HYPERVISOR_shared_info->arch.pfn_to_mfn_frame_list_list =
VTOMFN(xen_pfn_to_mfn_frame_list_list);
fpp = PAGE_SIZE/sizeof(unsigned long);
for (i = 0, j = 0, k = -1; i < max_pfn; i += fpp, j++) {
if ((j % fpp) == 0) {
k++;
xen_pfn_to_mfn_frame_list_list[k] =
VTOMFN(xen_pfn_to_mfn_frame_list[k]);
j = 0;
}
xen_pfn_to_mfn_frame_list[k][j] =
VTOMFN(&xen_phys_machine[i]);
}
HYPERVISOR_shared_info->arch.max_pfn = max_pfn;
gnttab_resume();
intr_resume(suspend_cancelled != 0);
local_irq_enable();
xencons_resume();
#ifdef CONFIG_SMP
for_each_cpu(i)
vcpu_prepare(i);
#endif
/*
* Only resume xenbus /after/ we've prepared our VCPUs; otherwise
* the VCPU hotplug callback can race with our vcpu_prepare
*/
mtx_lock(&Giant);
DEVICE_RESUME(root_bus);
mtx_unlock(&Giant);
#ifdef SMP
thread_lock(curthread);
sched_unbind(curthread);
thread_unlock(curthread);
if (!CPU_EMPTY(&map))
restart_cpus(map);
#endif
EVENTHANDLER_INVOKE(power_resume);
}
int
wdog_kern_pat(u_int utim)
{
int error;
if ((utim & WD_LASTVAL) != 0 && (utim & WD_INTERVAL) > 0)
return (EINVAL);
if ((utim & WD_LASTVAL) != 0) {
/*
* if WD_LASTVAL is set, fill in the bits for timeout
* from the saved value in wd_last_u.
*/
MPASS((wd_last_u & ~WD_INTERVAL) == 0);
utim &= ~WD_LASTVAL;
utim |= wd_last_u;
} else {
/*
* Otherwise save the new interval.
* This can be zero (to disable the watchdog)
*/
wd_last_u = (utim & WD_INTERVAL);
wd_last_u_sysctl = wd_last_u;
wd_last_u_sysctl_secs = pow2ns_to_ticks(wd_last_u) / hz;
}
if ((utim & WD_INTERVAL) == WD_TO_NEVER) {
utim = 0;
/* Assume all is well; watchdog signals failure. */
error = 0;
} else {
/* Assume no watchdog available; watchdog flags success */
error = EOPNOTSUPP;
}
if (wd_softtimer) {
if (utim == 0) {
callout_stop(&wd_softtimeo_handle);
} else {
(void) callout_reset(&wd_softtimeo_handle,
pow2ns_to_ticks(utim), wd_timeout_cb, "soft");
}
error = 0;
} else {
EVENTHANDLER_INVOKE(watchdog_list, utim, &error);
}
wd_set_pretimeout(wd_pretimeout, true);
/*
* If we were able to arm/strobe the watchdog, then
* update the last time it was strobed for WDIOC_GETTIMELEFT
*/
if (!error) {
struct timespec ts;
error = kern_clock_gettime(curthread /* XXX */,
CLOCK_MONOTONIC_FAST, &ts);
if (!error) {
wd_lastpat = ts.tv_sec;
wd_lastpat_valid = 1;
}
}
return (error);
}
static void
do_fork(struct thread *td, struct fork_req *fr, struct proc *p2, struct thread *td2,
struct vmspace *vm2, struct file *fp_procdesc)
{
struct proc *p1, *pptr;
int trypid;
struct filedesc *fd;
struct filedesc_to_leader *fdtol;
struct sigacts *newsigacts;
sx_assert(&proctree_lock, SX_SLOCKED);
sx_assert(&allproc_lock, SX_XLOCKED);
p1 = td->td_proc;
trypid = fork_findpid(fr->fr_flags);
sx_sunlock(&proctree_lock);
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);
allproc_gen++;
LIST_INSERT_HEAD(PIDHASH(p2->p_pid), p2, p_hash);
tidhash_add(td2);
PROC_LOCK(p2);
PROC_LOCK(p1);
sx_xunlock(&allproc_lock);
bcopy(&p1->p_startcopy, &p2->p_startcopy,
__rangeof(struct proc, p_startcopy, p_endcopy));
pargs_hold(p2->p_args);
PROC_UNLOCK(p1);
bzero(&p2->p_startzero,
__rangeof(struct proc, p_startzero, p_endzero));
/* Tell the prison that we exist. */
prison_proc_hold(p2->p_ucred->cr_prison);
PROC_UNLOCK(p2);
/*
* Malloc things while we don't hold any locks.
*/
if (fr->fr_flags & RFSIGSHARE)
newsigacts = NULL;
else
newsigacts = sigacts_alloc();
/*
* Copy filedesc.
*/
if (fr->fr_flags & RFCFDG) {
fd = fdinit(p1->p_fd, false);
fdtol = NULL;
} else if (fr->fr_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 ((fr->fr_flags & RFTHREAD) != 0) {
/*
* Shared file descriptor table, and shared
* process leaders.
*/
fdtol = p1->p_fdtol;
FILEDESC_XLOCK(p1->p_fd);
fdtol->fdl_refcount++;
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);
}
}
/*
* 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));
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;
#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);
/*
* Duplicate sub-structures as needed.
* Increase reference counts on shared objects.
*/
p2->p_flag = P_INMEM;
p2->p_flag2 = p1->p_flag2 & (P2_NOTRACE | P2_NOTRACE_EXEC | P2_TRAPCAP);
p2->p_swtick = ticks;
if (p1->p_flag & P_PROFIL)
startprofclock(p2);
/*
* Whilst the proc lock is held, copy the VM domain data out
* using the VM domain method.
*/
vm_domain_policy_init(&p2->p_vm_dom_policy);
vm_domain_policy_localcopy(&p2->p_vm_dom_policy,
&p1->p_vm_dom_policy);
if (fr->fr_flags & RFSIGSHARE) {
p2->p_sigacts = sigacts_hold(p1->p_sigacts);
} else {
sigacts_copy(newsigacts, p1->p_sigacts);
p2->p_sigacts = newsigacts;
}
if (fr->fr_flags & RFTSIGZMB)
p2->p_sigparent = RFTSIGNUM(fr->fr_flags);
else if (fr->fr_flags & RFLINUXTHPN)
p2->p_sigparent = SIGUSR1;
else
p2->p_sigparent = SIGCHLD;
p2->p_textvp = p1->p_textvp;
p2->p_fd = fd;
p2->p_fdtol = fdtol;
if (p1->p_flag2 & P2_INHERIT_PROTECTED) {
p2->p_flag |= P_PROTECTED;
p2->p_flag2 |= P2_INHERIT_PROTECTED;
}
/*
* p_limit is copy-on-write. Bump its refcount.
*/
lim_fork(p1, p2);
thread_cow_get_proc(td2, p2);
pstats_fork(p1->p_stats, p2->p_stats);
PROC_UNLOCK(p1);
PROC_UNLOCK(p2);
/* Bump references to the text vnode (for procfs). */
if (p2->p_textvp)
vrefact(p2->p_textvp);
/*
* Set up linkage for kernel based threading.
*/
if ((fr->fr_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) | TDP_FORKING;
SESS_LOCK(p1->p_session);
if (p1->p_session->s_ttyvp != NULL && p1->p_flag & P_CONTROLT)
p2->p_flag |= P_CONTROLT;
SESS_UNLOCK(p1->p_session);
if (fr->fr_flags & RFPPWAIT)
p2->p_flag |= P_PPWAIT;
p2->p_pgrp = p1->p_pgrp;
LIST_INSERT_AFTER(p1, p2, p_pglist);
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;
}
/*
* This begins the section where we must prevent the parent
* from being swapped.
*/
_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 ((fr->fr_flags & RFNOWAIT) != 0) {
pptr = p1->p_reaper;
p2->p_reaper = pptr;
} else {
p2->p_reaper = (p1->p_treeflag & P_TREE_REAPER) != 0 ?
p1 : p1->p_reaper;
pptr = p1;
}
p2->p_pptr = pptr;
LIST_INSERT_HEAD(&pptr->p_children, p2, p_sibling);
LIST_INIT(&p2->p_reaplist);
LIST_INSERT_HEAD(&p2->p_reaper->p_reaplist, p2, p_reapsibling);
if (p2->p_reaper == p1)
p2->p_reapsubtree = p2->p_pid;
sx_xunlock(&proctree_lock);
/* Inform accounting that we have forked. */
p2->p_acflag = AFORK;
PROC_UNLOCK(p2);
#ifdef KTRACE
ktrprocfork(p1, p2);
#endif
/*
* 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, fr->fr_flags);
if (fr->fr_flags == (RFFDG | RFPROC)) {
VM_CNT_INC(v_forks);
VM_CNT_ADD(v_forkpages, p2->p_vmspace->vm_dsize +
p2->p_vmspace->vm_ssize);
} else if (fr->fr_flags == (RFFDG | RFPROC | RFPPWAIT | RFMEM)) {
VM_CNT_INC(v_vforks);
VM_CNT_ADD(v_vforkpages, p2->p_vmspace->vm_dsize +
p2->p_vmspace->vm_ssize);
} else if (p1 == &proc0) {
VM_CNT_INC(v_kthreads);
VM_CNT_ADD(v_kthreadpages, p2->p_vmspace->vm_dsize +
p2->p_vmspace->vm_ssize);
} else {
VM_CNT_INC(v_rforks);
VM_CNT_ADD(v_rforkpages, p2->p_vmspace->vm_dsize +
p2->p_vmspace->vm_ssize);
}
/*
* 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 (fr->fr_flags & RFPROCDESC)
procdesc_new(p2, fr->fr_pd_flags);
/*
* Both processes are set up, now check if any loadable modules want
* to adjust anything.
*/
EVENTHANDLER_INVOKE(process_fork, p1, p2, fr->fr_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 so that any
* tracepoints inherited from the parent can be removed. We have to do
* this only after p_state is PRS_NORMAL since the fasttrap module will
* use pfind() later on.
*/
if ((fr->fr_flags & RFMEM) == 0 && dtrace_fasttrap_fork)
dtrace_fasttrap_fork(p1, p2);
#endif
/*
* Hold the process so that it cannot exit after we make it runnable,
* but before we wait for the debugger.
*/
_PHOLD(p2);
if (p1->p_ptevents & PTRACE_FORK) {
/*
* 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;
}
if (fr->fr_flags & RFPPWAIT) {
td->td_pflags |= TDP_RFPPWAIT;
td->td_rfppwait_p = p2;
td->td_dbgflags |= TDB_VFORK;
}
PROC_UNLOCK(p2);
/*
* Now can be swapped.
*/
_PRELE(p1);
PROC_UNLOCK(p1);
/*
* Tell any interested parties about the new process.
*/
knote_fork(p1->p_klist, p2->p_pid);
SDT_PROBE3(proc, , , create, p2, p1, fr->fr_flags);
if (fr->fr_flags & RFPROCDESC) {
procdesc_finit(p2->p_procdesc, fp_procdesc);
fdrop(fp_procdesc, td);
}
if ((fr->fr_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);
if (fr->fr_pidp != NULL)
*fr->fr_pidp = p2->p_pid;
} else {
*fr->fr_procp = p2;
}
PROC_LOCK(p2);
/*
* Wait until debugger is attached to child.
*/
while (td2->td_proc == p2 && (td2->td_dbgflags & TDB_STOPATFORK) != 0)
cv_wait(&p2->p_dbgwait, &p2->p_mtx);
_PRELE(p2);
racct_proc_fork_done(p2);
PROC_UNLOCK(p2);
}
static int
dcons_os_checkc(struct dcons_softc *dc)
{
EVENTHANDLER_INVOKE(dcons_poll, 0);
return (dcons_os_checkc_nopoll(dc));
}
