/*
* This is the protocol drain routine.
*
* No locks should be held when this is called. The drain routines have to
* presently acquire some locks which raises the possibility of lock order
* reversal.
*/
void
mb_reclaim(void *junk)
{
#if 0
struct domain *dp;
struct protosw *pr;
WITNESS_WARN(WARN_GIANTOK | WARN_SLEEPOK | WARN_PANIC, NULL,
"mb_reclaim()");
for (dp = domains; dp != NULL; dp = dp->dom_next)
for (pr = dp->dom_protosw; pr < dp->dom_protoswNPROTOSW; pr++)
if (pr->pr_drain != NULL)
(*pr->pr_drain)();
#endif
}
void
vm_smp_rendezvous(struct vm *vm, int vcpuid, cpuset_t dest,
vm_rendezvous_func_t func, void *arg)
{
int i;
/*
* Enforce that this function is called without any locks
*/
WITNESS_WARN(WARN_PANIC, NULL, "vm_smp_rendezvous");
KASSERT(vcpuid == -1 || (vcpuid >= 0 && vcpuid < VM_MAXCPU),
("vm_smp_rendezvous: invalid vcpuid %d", vcpuid));
restart:
mtx_lock(&vm->rendezvous_mtx);
if (vm->rendezvous_func != NULL) {
/*
* If a rendezvous is already in progress then we need to
* call the rendezvous handler in case this 'vcpuid' is one
* of the targets of the rendezvous.
*/
RENDEZVOUS_CTR0(vm, vcpuid, "Rendezvous already in progress");
mtx_unlock(&vm->rendezvous_mtx);
vm_handle_rendezvous(vm, vcpuid);
goto restart;
}
KASSERT(vm->rendezvous_func == NULL, ("vm_smp_rendezvous: previous "
"rendezvous is still in progress"));
RENDEZVOUS_CTR0(vm, vcpuid, "Initiating rendezvous");
vm->rendezvous_req_cpus = dest;
CPU_ZERO(&vm->rendezvous_done_cpus);
vm->rendezvous_arg = arg;
vm_set_rendezvous_func(vm, func);
mtx_unlock(&vm->rendezvous_mtx);
/*
* Wake up any sleeping vcpus and trigger a VM-exit in any running
* vcpus so they handle the rendezvous as soon as possible.
*/
for (i = 0; i < VM_MAXCPU; i++) {
if (CPU_ISSET(i, &dest))
vcpu_notify_event(vm, i, false);
}
vm_handle_rendezvous(vm, vcpuid);
}
/*
* Give next character to user as result of read.
*/
int
ureadc(int c, struct uio *uio)
{
struct iovec *iov;
char *iov_base;
WITNESS_WARN(WARN_GIANTOK | WARN_SLEEPOK, NULL,
"Calling ureadc()");
again:
if (uio->uio_iovcnt == 0 || uio->uio_resid == 0)
panic("ureadc");
iov = uio->uio_iov;
if (iov->iov_len == 0) {
uio->uio_iovcnt--;
uio->uio_iov++;
goto again;
}
switch (uio->uio_segflg) {
case UIO_USERSPACE:
if (subyte(iov->iov_base, c) < 0)
return (EFAULT);
break;
case UIO_SYSSPACE:
iov_base = iov->iov_base;
*iov_base = c;
iov->iov_base = iov_base;
break;
case UIO_NOCOPY:
break;
}
iov->iov_base = (char *)iov->iov_base + 1;
iov->iov_len--;
uio->uio_resid--;
uio->uio_offset++;
return (0);
}
/*
* Allocate a device specific dma_tag.
*/
int
bus_dma_tag_create(bus_dma_tag_t parent, bus_size_t alignment,
bus_addr_t boundary, bus_addr_t lowaddr, bus_addr_t highaddr,
bus_dma_filter_t *filter, void *filterarg, bus_size_t maxsize,
int nsegments, bus_size_t maxsegsz, int flags, bus_dma_lock_t *lockfunc,
void *lockfuncarg, bus_dma_tag_t *dmat)
{
struct bus_dma_tag_common *tc;
int error;
WITNESS_WARN(WARN_GIANTOK | WARN_SLEEPOK, NULL, "%s", __func__);
if (parent == NULL) {
error = bus_dma_bounce_impl.tag_create(parent, alignment,
boundary, lowaddr, highaddr, filter, filterarg, maxsize,
nsegments, maxsegsz, flags, lockfunc, lockfuncarg, dmat);
} else {
tc = (struct bus_dma_tag_common *)parent;
error = tc->impl->tag_create(parent, alignment,
boundary, lowaddr, highaddr, filter, filterarg, maxsize,
nsegments, maxsegsz, flags, lockfunc, lockfuncarg, dmat);
}
return (error);
}
/*
* Define the code needed before returning to user mode, for trap and
* syscall.
*/
void
userret(struct thread *td, struct trapframe *frame)
{
struct proc *p = td->td_proc;
CTR3(KTR_SYSC, "userret: thread %p (pid %d, %s)", td, p->p_pid,
td->td_name);
KASSERT((p->p_flag & P_WEXIT) == 0,
("Exiting process returns to usermode"));
#if 0
#ifdef DIAGNOSTIC
/* Check that we called signotify() enough. */
PROC_LOCK(p);
thread_lock(td);
if (SIGPENDING(td) && ((td->td_flags & TDF_NEEDSIGCHK) == 0 ||
(td->td_flags & TDF_ASTPENDING) == 0))
printf("failed to set signal flags properly for ast()\n");
thread_unlock(td);
PROC_UNLOCK(p);
#endif
#endif
#ifdef KTRACE
KTRUSERRET(td);
#endif
/*
* 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();
/*
* Charge system time if profiling.
*/
if (p->p_flag & P_PROFIL)
addupc_task(td, TRAPF_PC(frame), td->td_pticks * psratio);
/*
* Let the scheduler adjust our priority etc.
*/
sched_userret(td);
#ifdef XEN
PT_UPDATES_FLUSH();
#endif
/*
* Check for misbehavior.
*
* In case there is a callchain tracing ongoing because of
* hwpmc(4), skip the scheduler pinning check.
* hwpmc(4) subsystem, infact, will collect callchain informations
* at ast() checkpoint, which is past userret().
*/
WITNESS_WARN(WARN_PANIC, NULL, "userret: returning");
KASSERT(td->td_critnest == 0,
("userret: Returning in a critical section"));
KASSERT(td->td_locks == 0,
("userret: Returning with %d locks held", td->td_locks));
KASSERT(td->td_rw_rlocks == 0,
("userret: Returning with %d rwlocks held in read mode",
td->td_rw_rlocks));
KASSERT((td->td_pflags & TDP_NOFAULTING) == 0,
("userret: Returning with pagefaults disabled"));
KASSERT(td->td_no_sleeping == 0,
("userret: Returning with sleep disabled"));
KASSERT(td->td_pinned == 0 || (td->td_pflags & TDP_CALLCHAIN) != 0,
("userret: Returning with with pinned thread"));
KASSERT(td->td_vp_reserv == 0,
("userret: Returning while holding vnode reservation"));
KASSERT((td->td_flags & TDF_SBDRY) == 0,
("userret: Returning with stop signals deferred"));
#ifdef VIMAGE
/* Unfortunately td_vnet_lpush needs VNET_DEBUG. */
VNET_ASSERT(curvnet == NULL,
("%s: Returning on td %p (pid %d, %s) with vnet %p set in %s",
__func__, td, p->p_pid, td->td_name, curvnet,
(td->td_vnet_lpush != NULL) ? td->td_vnet_lpush : "N/A"));
#endif
#ifdef RACCT
PROC_LOCK(p);
while (p->p_throttled == 1)
msleep(p->p_racct, &p->p_mtx, 0, "racct", 0);
PROC_UNLOCK(p);
#endif
}
/*
* Process an asynchronous software trap.
* This is relatively easy.
* This function will return with preemption disabled.
*/
void
ast(struct trapframe *framep)
{
struct thread *td;
struct proc *p;
int flags;
int sig;
td = curthread;
p = td->td_proc;
CTR3(KTR_SYSC, "ast: thread %p (pid %d, %s)", td, p->p_pid,
p->p_comm);
KASSERT(TRAPF_USERMODE(framep), ("ast in kernel mode"));
WITNESS_WARN(WARN_PANIC, NULL, "Returning to user mode");
mtx_assert(&Giant, MA_NOTOWNED);
THREAD_LOCK_ASSERT(td, MA_NOTOWNED);
td->td_frame = framep;
td->td_pticks = 0;
/*
* This updates the td_flag's for the checks below in one
* "atomic" operation with turning off the astpending flag.
* If another AST is triggered while we are handling the
* AST's saved in flags, the astpending flag will be set and
* ast() will be called again.
*/
thread_lock(td);
flags = td->td_flags;
td->td_flags &= ~(TDF_ASTPENDING | TDF_NEEDSIGCHK | TDF_NEEDSUSPCHK |
TDF_NEEDRESCHED | TDF_ALRMPEND | TDF_PROFPEND | TDF_MACPEND);
thread_unlock(td);
PCPU_INC(cnt.v_trap);
if (td->td_ucred != p->p_ucred)
cred_update_thread(td);
if (td->td_pflags & TDP_OWEUPC && p->p_flag & P_PROFIL) {
addupc_task(td, td->td_profil_addr, td->td_profil_ticks);
td->td_profil_ticks = 0;
td->td_pflags &= ~TDP_OWEUPC;
}
#ifdef HWPMC_HOOKS
/* Handle Software PMC callchain capture. */
if (PMC_IS_PENDING_CALLCHAIN(td))
PMC_CALL_HOOK_UNLOCKED(td, PMC_FN_USER_CALLCHAIN_SOFT, (void *) framep);
#endif
if (flags & TDF_ALRMPEND) {
PROC_LOCK(p);
kern_psignal(p, SIGVTALRM);
PROC_UNLOCK(p);
}
if (flags & TDF_PROFPEND) {
PROC_LOCK(p);
kern_psignal(p, SIGPROF);
PROC_UNLOCK(p);
}
#ifdef MAC
if (flags & TDF_MACPEND)
mac_thread_userret(td);
#endif
if (flags & TDF_NEEDRESCHED) {
#ifdef KTRACE
if (KTRPOINT(td, KTR_CSW))
ktrcsw(1, 1, __func__);
#endif
thread_lock(td);
sched_prio(td, td->td_user_pri);
mi_switch(SW_INVOL | SWT_NEEDRESCHED, NULL);
thread_unlock(td);
#ifdef KTRACE
if (KTRPOINT(td, KTR_CSW))
ktrcsw(0, 1, __func__);
#endif
}
/*
* Check for signals. Unlocked reads of p_pendingcnt or
* p_siglist might cause process-directed signal to be handled
* later.
*/
if (flags & TDF_NEEDSIGCHK || p->p_pendingcnt > 0 ||
!SIGISEMPTY(p->p_siglist)) {
PROC_LOCK(p);
mtx_lock(&p->p_sigacts->ps_mtx);
while ((sig = cursig(td)) != 0)
postsig(sig);
mtx_unlock(&p->p_sigacts->ps_mtx);
PROC_UNLOCK(p);
}
/*
* We need to check to see if we have to exit or wait due to a
* single threading requirement or some other STOP condition.
*/
if (flags & TDF_NEEDSUSPCHK) {
PROC_LOCK(p);
thread_suspend_check(0);
PROC_UNLOCK(p);
}
if (td->td_pflags & TDP_OLDMASK) {
td->td_pflags &= ~TDP_OLDMASK;
kern_sigprocmask(td, SIG_SETMASK, &td->td_oldsigmask, NULL, 0);
}
userret(td, framep);
}
static int
uiomove_faultflag(void *cp, int n, struct uio *uio, int nofault)
{
#ifndef __rtems__
struct thread *td;
#endif /* __rtems__ */
struct iovec *iov;
u_int cnt;
int error, newflags, save;
#ifndef __rtems__
td = curthread;
#endif /* __rtems__ */
error = 0;
KASSERT(uio->uio_rw == UIO_READ || uio->uio_rw == UIO_WRITE,
("uiomove: mode"));
KASSERT(uio->uio_segflg != UIO_USERSPACE || uio->uio_td == td,
("uiomove proc"));
if (!nofault)
WITNESS_WARN(WARN_GIANTOK | WARN_SLEEPOK, NULL,
"Calling uiomove()");
#ifndef __rtems__
/* XXX does it make a sense to set TDP_DEADLKTREAT for UIO_SYSSPACE ? */
newflags = TDP_DEADLKTREAT;
if (uio->uio_segflg == UIO_USERSPACE && nofault) {
/*
* Fail if a non-spurious page fault occurs.
*/
newflags |= TDP_NOFAULTING | TDP_RESETSPUR;
}
save = curthread_pflags_set(newflags);
#else /* __rtems__ */
(void) newflags;
(void) save;
#endif /* __rtems__ */
while (n > 0 && uio->uio_resid) {
iov = uio->uio_iov;
cnt = iov->iov_len;
if (cnt == 0) {
uio->uio_iov++;
uio->uio_iovcnt--;
continue;
}
if (cnt > n)
cnt = n;
switch (uio->uio_segflg) {
case UIO_USERSPACE:
#ifndef __rtems__
if (ticks - PCPU_GET(switchticks) >= hogticks)
uio_yield();
#endif /* __rtems__ */
if (uio->uio_rw == UIO_READ)
error = copyout(cp, iov->iov_base, cnt);
else
error = copyin(iov->iov_base, cp, cnt);
if (error)
goto out;
break;
case UIO_SYSSPACE:
if (uio->uio_rw == UIO_READ)
bcopy(cp, iov->iov_base, cnt);
else
bcopy(iov->iov_base, cp, cnt);
break;
case UIO_NOCOPY:
break;
}
iov->iov_base = (char *)iov->iov_base + cnt;
iov->iov_len -= cnt;
uio->uio_resid -= cnt;
uio->uio_offset += cnt;
cp = (char *)cp + cnt;
n -= cnt;
}
out:
#ifndef __rtems__
curthread_pflags_restore(save);
#endif /* __rtems__ */
return (error);
}
/*
* 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();
}
int
uiomove(void *cp, int n, struct uio *uio)
{
struct thread *td = curthread;
struct iovec *iov;
u_int cnt;
int error = 0;
int save = 0;
KASSERT(uio->uio_rw == UIO_READ || uio->uio_rw == UIO_WRITE,
("uiomove: mode"));
KASSERT(uio->uio_segflg != UIO_USERSPACE || uio->uio_td == curthread,
("uiomove proc"));
WITNESS_WARN(WARN_GIANTOK | WARN_SLEEPOK, NULL,
"Calling uiomove()");
#ifndef __rtems__
save = td->td_pflags & TDP_DEADLKTREAT;
td->td_pflags |= TDP_DEADLKTREAT;
#endif
while (n > 0 && uio->uio_resid) {
iov = uio->uio_iov;
cnt = iov->iov_len;
if (cnt == 0) {
uio->uio_iov++;
uio->uio_iovcnt--;
continue;
}
if (cnt > n)
cnt = n;
switch (uio->uio_segflg) {
case UIO_USERSPACE:
if (ticks - PCPU_GET(switchticks) >= hogticks)
uio_yield();
if (uio->uio_rw == UIO_READ)
error = copyout(cp, iov->iov_base, cnt);
else
error = copyin(iov->iov_base, cp, cnt);
if (error)
goto out;
break;
case UIO_SYSSPACE:
if (uio->uio_rw == UIO_READ)
bcopy(cp, iov->iov_base, cnt);
else
bcopy(iov->iov_base, cp, cnt);
break;
case UIO_NOCOPY:
break;
}
iov->iov_base = (char *)iov->iov_base + cnt;
iov->iov_len -= cnt;
uio->uio_resid -= cnt;
uio->uio_offset += cnt;
cp = (char *)cp + cnt;
n -= cnt;
}
out:
#ifndef __rtems__
if (save == 0)
td->td_pflags &= ~TDP_DEADLKTREAT;
#endif
return (error);
}
int
uiofill(uint8_t val, int n, struct uio *uio)
{
struct thread *td;
struct iovec *iov;
size_t cnt;
int error, save;
#define UIOFILL_MAXBUF 256
uint8_t buf[UIOFILL_MAXBUF];
td = curthread;
error = 0;
KASSERT(uio->uio_rw == UIO_READ, ("uiofill: mode"));
KASSERT(uio->uio_segflg != UIO_USERSPACE || uio->uio_td == td,
("uiofill proc"));
WITNESS_WARN(WARN_GIANTOK | WARN_SLEEPOK, NULL,
"Calling uiofill()");
/* XXX does it make a sense to set TDP_DEADLKTREAT for UIO_SYSSPACE ? */
save = curthread_pflags_set(TDP_DEADLKTREAT);
if (uio->uio_segflg == UIO_USERSPACE)
memset(buf, val, imax(UIOFILL_MAXBUF, n));
while (n > 0 && uio->uio_resid) {
iov = uio->uio_iov;
cnt = iov->iov_len;
if (cnt == 0) {
uio->uio_iov++;
uio->uio_iovcnt--;
continue;
}
if (cnt > n)
cnt = n;
switch (uio->uio_segflg) {
case UIO_USERSPACE:
maybe_yield();
error = copyout(buf, iov->iov_base, cnt);
if (error)
goto out;
break;
case UIO_SYSSPACE:
memset(iov->iov_base, val, cnt);
break;
case UIO_NOCOPY:
break;
}
iov->iov_base = (char *)iov->iov_base + cnt;
iov->iov_len -= cnt;
uio->uio_resid -= cnt;
uio->uio_offset += cnt;
n -= cnt;
}
out:
curthread_pflags_restore(save);
return (error);
}
void
ia32_syscall(struct trapframe *frame)
{
caddr_t params;
int i;
struct sysent *callp;
struct thread *td = curthread;
struct proc *p = td->td_proc;
register_t orig_tf_rflags;
int error;
int narg;
u_int32_t args[8];
u_int64_t args64[8];
u_int code;
ksiginfo_t ksi;
PCPU_INC(cnt.v_syscall);
td->td_pticks = 0;
td->td_frame = frame;
if (td->td_ucred != p->p_ucred)
cred_update_thread(td);
params = (caddr_t)frame->tf_rsp + sizeof(u_int32_t);
code = frame->tf_rax;
orig_tf_rflags = frame->tf_rflags;
if (p->p_sysent->sv_prepsyscall) {
/*
* The prep code is MP aware.
*/
(*p->p_sysent->sv_prepsyscall)(frame, args, &code, ¶ms);
} else {
/*
* Need to check if this is a 32 bit or 64 bit syscall.
* fuword is MP aware.
*/
if (code == SYS_syscall) {
/*
* Code is first argument, followed by actual args.
*/
code = fuword32(params);
params += sizeof(int);
} else if (code == SYS___syscall) {
/*
* Like syscall, but code is a quad, so as to maintain
* quad alignment for the rest of the arguments.
* We use a 32-bit fetch in case params is not
* aligned.
*/
code = fuword32(params);
params += sizeof(quad_t);
}
}
if (p->p_sysent->sv_mask)
code &= p->p_sysent->sv_mask;
if (code >= p->p_sysent->sv_size)
callp = &p->p_sysent->sv_table[0];
else
callp = &p->p_sysent->sv_table[code];
narg = callp->sy_narg;
/*
* copyin and the ktrsyscall()/ktrsysret() code is MP-aware
*/
if (params != NULL && narg != 0)
error = copyin(params, (caddr_t)args,
(u_int)(narg * sizeof(int)));
else
error = 0;
for (i = 0; i < narg; i++)
args64[i] = args[i];
#ifdef KTRACE
if (KTRPOINT(td, KTR_SYSCALL))
ktrsyscall(code, narg, args64);
#endif
CTR4(KTR_SYSC, "syscall enter thread %p pid %d proc %s code %d", td,
td->td_proc->p_pid, td->td_proc->p_comm, code);
if (error == 0) {
td->td_retval[0] = 0;
td->td_retval[1] = frame->tf_rdx;
STOPEVENT(p, S_SCE, narg);
PTRACESTOP_SC(p, td, S_PT_SCE);
AUDIT_SYSCALL_ENTER(code, td);
error = (*callp->sy_call)(td, args64);
AUDIT_SYSCALL_EXIT(error, td);
}
switch (error) {
case 0:
frame->tf_rax = td->td_retval[0];
frame->tf_rdx = td->td_retval[1];
frame->tf_rflags &= ~PSL_C;
break;
case ERESTART:
/*
* Reconstruct pc, assuming lcall $X,y is 7 bytes,
* int 0x80 is 2 bytes. We saved this in tf_err.
*/
frame->tf_rip -= frame->tf_err;
break;
case EJUSTRETURN:
break;
default:
if (p->p_sysent->sv_errsize) {
if (error >= p->p_sysent->sv_errsize)
error = -1; /* XXX */
else
error = p->p_sysent->sv_errtbl[error];
}
frame->tf_rax = error;
frame->tf_rflags |= PSL_C;
break;
}
/*
* Traced syscall.
*/
if (orig_tf_rflags & PSL_T) {
frame->tf_rflags &= ~PSL_T;
ksiginfo_init_trap(&ksi);
ksi.ksi_signo = SIGTRAP;
ksi.ksi_code = TRAP_TRACE;
ksi.ksi_addr = (void *)frame->tf_rip;
trapsignal(td, &ksi);
}
/*
* Check for misbehavior.
*/
WITNESS_WARN(WARN_PANIC, NULL, "System call %s returning",
(code >= 0 && code < SYS_MAXSYSCALL) ? freebsd32_syscallnames[code] : "???");
KASSERT(td->td_critnest == 0,
("System call %s returning in a critical section",
(code >= 0 && code < SYS_MAXSYSCALL) ? freebsd32_syscallnames[code] : "???"));
KASSERT(td->td_locks == 0,
("System call %s returning with %d locks held",
(code >= 0 && code < SYS_MAXSYSCALL) ? freebsd32_syscallnames[code] : "???",
td->td_locks));
/*
* Handle reschedule and other end-of-syscall issues
*/
userret(td, frame);
CTR4(KTR_SYSC, "syscall exit thread %p pid %d proc %s code %d", td,
td->td_proc->p_pid, td->td_proc->p_comm, code);
#ifdef KTRACE
if (KTRPOINT(td, KTR_SYSRET))
ktrsysret(code, error, td->td_retval[0]);
#endif
/*
* This works because errno is findable through the
* register set. If we ever support an emulation where this
* is not the case, this code will need to be revisited.
*/
STOPEVENT(p, S_SCX, code);
PTRACESTOP_SC(p, td, S_PT_SCX);
}
