/*
* Function: unix_syscall
*
* Inputs: regs - pointer to i386 save area
*
* Outputs: none
*/
void
unix_syscall(x86_saved_state_t *state)
{
thread_t thread;
void *vt;
unsigned int code;
struct sysent *callp;
int error;
vm_offset_t params;
struct proc *p;
struct uthread *uthread;
x86_saved_state32_t *regs;
boolean_t is_vfork;
assert(is_saved_state32(state));
regs = saved_state32(state);
#if DEBUG
if (regs->eax == 0x800)
thread_exception_return();
#endif
thread = current_thread();
uthread = get_bsdthread_info(thread);
/* Get the approriate proc; may be different from task's for vfork() */
is_vfork = uthread->uu_flag & UT_VFORK;
if (__improbable(is_vfork != 0))
p = current_proc();
else
p = (struct proc *)get_bsdtask_info(current_task());
/* Verify that we are not being called from a task without a proc */
if (__improbable(p == NULL)) {
regs->eax = EPERM;
regs->efl |= EFL_CF;
task_terminate_internal(current_task());
thread_exception_return();
/* NOTREACHED */
}
code = regs->eax & I386_SYSCALL_NUMBER_MASK;
DEBUG_KPRINT_SYSCALL_UNIX("unix_syscall: code=%d(%s) eip=%u\n",
code, syscallnames[code >= NUM_SYSENT ? 63 : code], (uint32_t)regs->eip);
params = (vm_offset_t) (regs->uesp + sizeof (int));
regs->efl &= ~(EFL_CF);
callp = (code >= NUM_SYSENT) ? &sysent[63] : &sysent[code];
if (__improbable(callp == sysent)) {
code = fuword(params);
params += sizeof(int);
callp = (code >= NUM_SYSENT) ? &sysent[63] : &sysent[code];
}
vt = (void *)uthread->uu_arg;
if (callp->sy_arg_bytes != 0) {
#if CONFIG_REQUIRES_U32_MUNGING
sy_munge_t *mungerp;
#else
#error U32 syscalls on x86_64 kernel requires munging
#endif
uint32_t nargs;
assert((unsigned) callp->sy_arg_bytes <= sizeof (uthread->uu_arg));
nargs = callp->sy_arg_bytes;
error = copyin((user_addr_t) params, (char *) vt, nargs);
if (error) {
regs->eax = error;
regs->efl |= EFL_CF;
thread_exception_return();
/* NOTREACHED */
}
if (__probable(code != 180)) {
int *ip = (int *)vt;
KERNEL_DEBUG_CONSTANT_IST(KDEBUG_TRACE,
BSDDBG_CODE(DBG_BSD_EXCP_SC, code) | DBG_FUNC_START,
*ip, *(ip+1), *(ip+2), *(ip+3), 0);
}
#if CONFIG_REQUIRES_U32_MUNGING
mungerp = callp->sy_arg_munge32;
if (mungerp != NULL)
(*mungerp)(vt);
#endif
} else
KERNEL_DEBUG_CONSTANT_IST(KDEBUG_TRACE,
BSDDBG_CODE(DBG_BSD_EXCP_SC, code) | DBG_FUNC_START,
0, 0, 0, 0, 0);
/*
* Delayed binding of thread credential to process credential, if we
* are not running with an explicitly set thread credential.
*/
kauth_cred_uthread_update(uthread, p);
uthread->uu_rval[0] = 0;
uthread->uu_rval[1] = 0;
uthread->uu_flag |= UT_NOTCANCELPT;
uthread->syscall_code = code;
#ifdef JOE_DEBUG
uthread->uu_iocount = 0;
uthread->uu_vpindex = 0;
#endif
AUDIT_SYSCALL_ENTER(code, p, uthread);
error = (*(callp->sy_call))((void *) p, (void *) vt, &(uthread->uu_rval[0]));
AUDIT_SYSCALL_EXIT(code, p, uthread, error);
#ifdef JOE_DEBUG
if (uthread->uu_iocount)
printf("system call returned with uu_iocount != 0\n");
#endif
#if CONFIG_DTRACE
uthread->t_dtrace_errno = error;
#endif /* CONFIG_DTRACE */
if (__improbable(error == ERESTART)) {
/*
* Move the user's pc back to repeat the syscall:
* 5 bytes for a sysenter, or 2 for an int 8x.
* The SYSENTER_TF_CS covers single-stepping over a sysenter
* - see debug trap handler in idt.s/idt64.s
*/
pal_syscall_restart(thread, state);
}
else if (error != EJUSTRETURN) {
if (__improbable(error)) {
regs->eax = error;
regs->efl |= EFL_CF; /* carry bit */
} else { /* (not error) */
/*
* We split retval across two registers, in case the
* syscall had a 64-bit return value, in which case
* eax/edx matches the function call ABI.
*/
regs->eax = uthread->uu_rval[0];
regs->edx = uthread->uu_rval[1];
}
}
DEBUG_KPRINT_SYSCALL_UNIX(
"unix_syscall: error=%d retval=(%u,%u)\n",
error, regs->eax, regs->edx);
uthread->uu_flag &= ~UT_NOTCANCELPT;
if (__improbable(uthread->uu_lowpri_window)) {
/*
* task is marked as a low priority I/O type
* and the I/O we issued while in this system call
* collided with normal I/O operations... we'll
* delay in order to mitigate the impact of this
* task on the normal operation of the system
*/
throttle_lowpri_io(1);
}
if (__probable(code != 180))
KERNEL_DEBUG_CONSTANT_IST(KDEBUG_TRACE,
BSDDBG_CODE(DBG_BSD_EXCP_SC, code) | DBG_FUNC_END,
error, uthread->uu_rval[0], uthread->uu_rval[1], p->p_pid, 0);
if (__improbable(!is_vfork && callp->sy_call == (sy_call_t *)execve && !error)) {
pal_execve_return(thread);
}
thread_exception_return();
/* NOTREACHED */
}
void
unix_syscall_return(int error)
{
thread_t thread;
struct uthread *uthread;
struct proc *p;
unsigned int code;
struct sysent *callp;
thread = current_thread();
uthread = get_bsdthread_info(thread);
pal_register_cache_state(thread, DIRTY);
p = current_proc();
if (proc_is64bit(p)) {
x86_saved_state64_t *regs;
regs = saved_state64(find_user_regs(thread));
code = uthread->syscall_code;
callp = (code >= NUM_SYSENT) ? &sysent[63] : &sysent[code];
#if CONFIG_DTRACE
if (callp->sy_call == dtrace_systrace_syscall)
dtrace_systrace_syscall_return( code, error, uthread->uu_rval );
#endif /* CONFIG_DTRACE */
AUDIT_SYSCALL_EXIT(code, p, uthread, error);
if (error == ERESTART) {
/*
* repeat the syscall
*/
pal_syscall_restart( thread, find_user_regs(thread) );
}
else if (error != EJUSTRETURN) {
if (error) {
regs->rax = error;
regs->isf.rflags |= EFL_CF; /* carry bit */
} else { /* (not error) */
switch (callp->sy_return_type) {
case _SYSCALL_RET_INT_T:
regs->rax = uthread->uu_rval[0];
regs->rdx = uthread->uu_rval[1];
break;
case _SYSCALL_RET_UINT_T:
regs->rax = ((u_int)uthread->uu_rval[0]);
regs->rdx = ((u_int)uthread->uu_rval[1]);
break;
case _SYSCALL_RET_OFF_T:
case _SYSCALL_RET_ADDR_T:
case _SYSCALL_RET_SIZE_T:
case _SYSCALL_RET_SSIZE_T:
case _SYSCALL_RET_UINT64_T:
regs->rax = *((uint64_t *)(&uthread->uu_rval[0]));
regs->rdx = 0;
break;
case _SYSCALL_RET_NONE:
break;
default:
panic("unix_syscall: unknown return type");
break;
}
regs->isf.rflags &= ~EFL_CF;
}
}
DEBUG_KPRINT_SYSCALL_UNIX(
"unix_syscall_return: error=%d retval=(%llu,%llu)\n",
error, regs->rax, regs->rdx);
} else {
x86_saved_state32_t *regs;
regs = saved_state32(find_user_regs(thread));
regs->efl &= ~(EFL_CF);
code = uthread->syscall_code;
callp = (code >= NUM_SYSENT) ? &sysent[63] : &sysent[code];
#if CONFIG_DTRACE
if (callp->sy_call == dtrace_systrace_syscall)
dtrace_systrace_syscall_return( code, error, uthread->uu_rval );
#endif /* CONFIG_DTRACE */
AUDIT_SYSCALL_EXIT(code, p, uthread, error);
if (error == ERESTART) {
pal_syscall_restart( thread, find_user_regs(thread) );
}
else if (error != EJUSTRETURN) {
if (error) {
regs->eax = error;
regs->efl |= EFL_CF; /* carry bit */
} else { /* (not error) */
regs->eax = uthread->uu_rval[0];
regs->edx = uthread->uu_rval[1];
}
}
DEBUG_KPRINT_SYSCALL_UNIX(
"unix_syscall_return: error=%d retval=(%u,%u)\n",
error, regs->eax, regs->edx);
}
uthread->uu_flag &= ~UT_NOTCANCELPT;
if (uthread->uu_lowpri_window) {
/*
* task is marked as a low priority I/O type
* and the I/O we issued while in this system call
* collided with normal I/O operations... we'll
* delay in order to mitigate the impact of this
* task on the normal operation of the system
*/
throttle_lowpri_io(1);
}
if (code != 180)
KERNEL_DEBUG_CONSTANT_IST(KDEBUG_TRACE,
BSDDBG_CODE(DBG_BSD_EXCP_SC, code) | DBG_FUNC_END,
error, uthread->uu_rval[0], uthread->uu_rval[1], p->p_pid, 0);
thread_exception_return();
/* NOTREACHED */
}
void
unix_syscall64(x86_saved_state_t *state)
{
thread_t thread;
void *vt;
unsigned int code;
struct sysent *callp;
int args_in_regs;
boolean_t args_start_at_rdi;
int error;
struct proc *p;
struct uthread *uthread;
x86_saved_state64_t *regs;
assert(is_saved_state64(state));
regs = saved_state64(state);
#if DEBUG
if (regs->rax == 0x2000800)
thread_exception_return();
#endif
thread = current_thread();
uthread = get_bsdthread_info(thread);
/* Get the approriate proc; may be different from task's for vfork() */
if (__probable(!(uthread->uu_flag & UT_VFORK)))
p = (struct proc *)get_bsdtask_info(current_task());
else
p = current_proc();
/* Verify that we are not being called from a task without a proc */
if (__improbable(p == NULL)) {
regs->rax = EPERM;
regs->isf.rflags |= EFL_CF;
task_terminate_internal(current_task());
thread_exception_return();
/* NOTREACHED */
}
code = regs->rax & SYSCALL_NUMBER_MASK;
DEBUG_KPRINT_SYSCALL_UNIX(
"unix_syscall64: code=%d(%s) rip=%llx\n",
code, syscallnames[code >= NUM_SYSENT ? 63 : code], regs->isf.rip);
callp = (code >= NUM_SYSENT) ? &sysent[63] : &sysent[code];
vt = (void *)uthread->uu_arg;
if (__improbable(callp == sysent)) {
/*
* indirect system call... system call number
* passed as 'arg0'
*/
code = regs->rdi;
callp = (code >= NUM_SYSENT) ? &sysent[63] : &sysent[code];
args_start_at_rdi = FALSE;
args_in_regs = 5;
} else {
args_start_at_rdi = TRUE;
args_in_regs = 6;
}
if (callp->sy_narg != 0) {
assert(callp->sy_narg <= 8); /* size of uu_arg */
args_in_regs = MIN(args_in_regs, callp->sy_narg);
memcpy(vt, args_start_at_rdi ? ®s->rdi : ®s->rsi, args_in_regs * sizeof(syscall_arg_t));
if (code != 180) {
uint64_t *ip = (uint64_t *)vt;
KERNEL_DEBUG_CONSTANT_IST(KDEBUG_TRACE,
BSDDBG_CODE(DBG_BSD_EXCP_SC, code) | DBG_FUNC_START,
(int)(*ip), (int)(*(ip+1)), (int)(*(ip+2)), (int)(*(ip+3)), 0);
}
if (__improbable(callp->sy_narg > args_in_regs)) {
int copyin_count;
copyin_count = (callp->sy_narg - args_in_regs) * sizeof(syscall_arg_t);
error = copyin((user_addr_t)(regs->isf.rsp + sizeof(user_addr_t)), (char *)&uthread->uu_arg[args_in_regs], copyin_count);
if (error) {
regs->rax = error;
regs->isf.rflags |= EFL_CF;
thread_exception_return();
/* NOTREACHED */
}
}
} else
KERNEL_DEBUG_CONSTANT_IST(KDEBUG_TRACE,
BSDDBG_CODE(DBG_BSD_EXCP_SC, code) | DBG_FUNC_START,
0, 0, 0, 0, 0);
/*
* Delayed binding of thread credential to process credential, if we
* are not running with an explicitly set thread credential.
*/
kauth_cred_uthread_update(uthread, p);
uthread->uu_rval[0] = 0;
uthread->uu_rval[1] = 0;
uthread->uu_flag |= UT_NOTCANCELPT;
uthread->syscall_code = code;
#ifdef JOE_DEBUG
uthread->uu_iocount = 0;
uthread->uu_vpindex = 0;
#endif
AUDIT_SYSCALL_ENTER(code, p, uthread);
error = (*(callp->sy_call))((void *) p, vt, &(uthread->uu_rval[0]));
AUDIT_SYSCALL_EXIT(code, p, uthread, error);
#ifdef JOE_DEBUG
if (uthread->uu_iocount)
printf("system call returned with uu_iocount != 0\n");
#endif
#if CONFIG_DTRACE
uthread->t_dtrace_errno = error;
#endif /* CONFIG_DTRACE */
if (__improbable(error == ERESTART)) {
/*
* all system calls come through via the syscall instruction
* in 64 bit mode... its 2 bytes in length
* move the user's pc back to repeat the syscall:
*/
pal_syscall_restart( thread, state );
}
else if (error != EJUSTRETURN) {
if (__improbable(error)) {
regs->rax = error;
regs->isf.rflags |= EFL_CF; /* carry bit */
} else { /* (not error) */
switch (callp->sy_return_type) {
case _SYSCALL_RET_INT_T:
regs->rax = uthread->uu_rval[0];
regs->rdx = uthread->uu_rval[1];
break;
case _SYSCALL_RET_UINT_T:
regs->rax = ((u_int)uthread->uu_rval[0]);
regs->rdx = ((u_int)uthread->uu_rval[1]);
break;
case _SYSCALL_RET_OFF_T:
case _SYSCALL_RET_ADDR_T:
case _SYSCALL_RET_SIZE_T:
case _SYSCALL_RET_SSIZE_T:
case _SYSCALL_RET_UINT64_T:
regs->rax = *((uint64_t *)(&uthread->uu_rval[0]));
regs->rdx = 0;
break;
case _SYSCALL_RET_NONE:
break;
default:
panic("unix_syscall: unknown return type");
break;
}
regs->isf.rflags &= ~EFL_CF;
}
}
DEBUG_KPRINT_SYSCALL_UNIX(
"unix_syscall64: error=%d retval=(%llu,%llu)\n",
error, regs->rax, regs->rdx);
uthread->uu_flag &= ~UT_NOTCANCELPT;
if (__improbable(uthread->uu_lowpri_window)) {
/*
* task is marked as a low priority I/O type
* and the I/O we issued while in this system call
* collided with normal I/O operations... we'll
* delay in order to mitigate the impact of this
* task on the normal operation of the system
*/
throttle_lowpri_io(1);
}
if (__probable(code != 180))
KERNEL_DEBUG_CONSTANT_IST(KDEBUG_TRACE,
BSDDBG_CODE(DBG_BSD_EXCP_SC, code) | DBG_FUNC_END,
error, uthread->uu_rval[0], uthread->uu_rval[1], p->p_pid, 0);
thread_exception_return();
/* NOTREACHED */
}
/*
* 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();
}
static inline int
syscallenter(struct thread *td, struct syscall_args *sa)
{
struct proc *p;
int error, traced;
PCPU_INC(cnt.v_syscall);
p = td->td_proc;
td->td_pticks = 0;
if (td->td_ucred != p->p_ucred)
cred_update_thread(td);
if (p->p_flag & P_TRACED) {
traced = 1;
PROC_LOCK(p);
td->td_dbgflags &= ~TDB_USERWR;
td->td_dbgflags |= TDB_SCE;
PROC_UNLOCK(p);
} else
traced = 0;
error = (p->p_sysent->sv_fetch_syscall_args)(td, sa);
#ifdef KTRACE
if (KTRPOINT(td, KTR_SYSCALL))
ktrsyscall(sa->code, sa->narg, sa->args);
#endif
CTR6(KTR_SYSC,
"syscall: td=%p pid %d %s (%#lx, %#lx, %#lx)",
td, td->td_proc->p_pid, syscallname(p, sa->code),
sa->args[0], sa->args[1], sa->args[2]);
if (error == 0) {
STOPEVENT(p, S_SCE, sa->narg);
if (p->p_flag & P_TRACED && p->p_stops & S_PT_SCE) {
PROC_LOCK(p);
ptracestop((td), SIGTRAP);
PROC_UNLOCK(p);
}
if (td->td_dbgflags & TDB_USERWR) {
/*
* Reread syscall number and arguments if
* debugger modified registers or memory.
*/
error = (p->p_sysent->sv_fetch_syscall_args)(td, sa);
#ifdef KTRACE
if (KTRPOINT(td, KTR_SYSCALL))
ktrsyscall(sa->code, sa->narg, sa->args);
#endif
if (error != 0)
goto retval;
}
#ifdef CAPABILITY_MODE
/*
* In capability mode, we only allow access to system calls
* flagged with SYF_CAPENABLED.
*/
if (IN_CAPABILITY_MODE(td) &&
!(sa->callp->sy_flags & SYF_CAPENABLED)) {
error = ECAPMODE;
goto retval;
}
#endif
error = syscall_thread_enter(td, sa->callp);
if (error != 0)
goto retval;
#ifdef KDTRACE_HOOKS
/*
* If the systrace module has registered it's probe
* callback and if there is a probe active for the
* syscall 'entry', process the probe.
*/
if (systrace_probe_func != NULL && sa->callp->sy_entry != 0)
(*systrace_probe_func)(sa->callp->sy_entry, sa->code,
sa->callp, sa->args, 0);
#endif
AUDIT_SYSCALL_ENTER(sa->code, td);
error = (sa->callp->sy_call)(td, sa->args);
AUDIT_SYSCALL_EXIT(error, td);
/* Save the latest error return value. */
td->td_errno = error;
#ifdef KDTRACE_HOOKS
/*
* If the systrace module has registered it's probe
* callback and if there is a probe active for the
* syscall 'return', process the probe.
*/
if (systrace_probe_func != NULL && sa->callp->sy_return != 0)
(*systrace_probe_func)(sa->callp->sy_return, sa->code,
sa->callp, NULL, (error) ? -1 : td->td_retval[0]);
#endif
syscall_thread_exit(td, sa->callp);
CTR4(KTR_SYSC, "syscall: p=%p error=%d return %#lx %#lx",
p, error, td->td_retval[0], td->td_retval[1]);
}
retval:
if (traced) {
PROC_LOCK(p);
td->td_dbgflags &= ~TDB_SCE;
PROC_UNLOCK(p);
}
(p->p_sysent->sv_set_syscall_retval)(td, error);
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);
}
