3587 lines
90 KiB
C
3587 lines
90 KiB
C
|
/*
|
||
|
* Copyright (c) 1995-2016 Apple Inc. All rights reserved.
|
||
|
*
|
||
|
* @APPLE_OSREFERENCE_LICENSE_HEADER_START@
|
||
|
*
|
||
|
* This file contains Original Code and/or Modifications of Original Code
|
||
|
* as defined in and that are subject to the Apple Public Source License
|
||
|
* Version 2.0 (the 'License'). You may not use this file except in
|
||
|
* compliance with the License. The rights granted to you under the License
|
||
|
* may not be used to create, or enable the creation or redistribution of,
|
||
|
* unlawful or unlicensed copies of an Apple operating system, or to
|
||
|
* circumvent, violate, or enable the circumvention or violation of, any
|
||
|
* terms of an Apple operating system software license agreement.
|
||
|
*
|
||
|
* Please obtain a copy of the License at
|
||
|
* http://www.opensource.apple.com/apsl/ and read it before using this file.
|
||
|
*
|
||
|
* The Original Code and all software distributed under the License are
|
||
|
* distributed on an 'AS IS' basis, WITHOUT WARRANTY OF ANY KIND, EITHER
|
||
|
* EXPRESS OR IMPLIED, AND APPLE HEREBY DISCLAIMS ALL SUCH WARRANTIES,
|
||
|
* INCLUDING WITHOUT LIMITATION, ANY WARRANTIES OF MERCHANTABILITY,
|
||
|
* FITNESS FOR A PARTICULAR PURPOSE, QUIET ENJOYMENT OR NON-INFRINGEMENT.
|
||
|
* Please see the License for the specific language governing rights and
|
||
|
* limitations under the License.
|
||
|
*
|
||
|
* @APPLE_OSREFERENCE_LICENSE_HEADER_END@
|
||
|
*/
|
||
|
/*
|
||
|
* Copyright (c) 1982, 1986, 1989, 1991, 1993
|
||
|
* The Regents of the University of California. All rights reserved.
|
||
|
* (c) UNIX System Laboratories, Inc.
|
||
|
* All or some portions of this file are derived from material licensed
|
||
|
* to the University of California by American Telephone and Telegraph
|
||
|
* Co. or Unix System Laboratories, Inc. and are reproduced herein with
|
||
|
* the permission of UNIX System Laboratories, Inc.
|
||
|
*
|
||
|
* Redistribution and use in source and binary forms, with or without
|
||
|
* modification, are permitted provided that the following conditions
|
||
|
* are met:
|
||
|
* 1. Redistributions of source code must retain the above copyright
|
||
|
* notice, this list of conditions and the following disclaimer.
|
||
|
* 2. Redistributions in binary form must reproduce the above copyright
|
||
|
* notice, this list of conditions and the following disclaimer in the
|
||
|
* documentation and/or other materials provided with the distribution.
|
||
|
* 3. All advertising materials mentioning features or use of this software
|
||
|
* must display the following acknowledgement:
|
||
|
* This product includes software developed by the University of
|
||
|
* California, Berkeley and its contributors.
|
||
|
* 4. Neither the name of the University nor the names of its contributors
|
||
|
* may be used to endorse or promote products derived from this software
|
||
|
* without specific prior written permission.
|
||
|
*
|
||
|
* THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
|
||
|
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
|
||
|
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
|
||
|
* ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
|
||
|
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
|
||
|
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
|
||
|
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
|
||
|
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
|
||
|
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
|
||
|
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
|
||
|
* SUCH DAMAGE.
|
||
|
*
|
||
|
* @(#)kern_sig.c 8.7 (Berkeley) 4/18/94
|
||
|
*/
|
||
|
/*
|
||
|
* NOTICE: This file was modified by SPARTA, Inc. in 2005 to introduce
|
||
|
* support for mandatory and extensible security protections. This notice
|
||
|
* is included in support of clause 2.2 (b) of the Apple Public License,
|
||
|
* Version 2.0.
|
||
|
*/
|
||
|
|
||
|
#define SIGPROP /* include signal properties table */
|
||
|
#include <sys/param.h>
|
||
|
#include <sys/resourcevar.h>
|
||
|
#include <sys/proc_internal.h>
|
||
|
#include <sys/kauth.h>
|
||
|
#include <sys/systm.h>
|
||
|
#include <sys/timeb.h>
|
||
|
#include <sys/times.h>
|
||
|
#include <sys/acct.h>
|
||
|
#include <sys/file_internal.h>
|
||
|
#include <sys/kernel.h>
|
||
|
#include <sys/wait.h>
|
||
|
#include <sys/signalvar.h>
|
||
|
#include <sys/syslog.h>
|
||
|
#include <sys/stat.h>
|
||
|
#include <sys/lock.h>
|
||
|
#include <sys/kdebug.h>
|
||
|
#include <sys/reason.h>
|
||
|
|
||
|
#include <sys/mount.h>
|
||
|
#include <sys/sysproto.h>
|
||
|
|
||
|
#include <security/audit/audit.h>
|
||
|
|
||
|
#include <kern/cpu_number.h>
|
||
|
|
||
|
#include <sys/vm.h>
|
||
|
#include <sys/user.h> /* for coredump */
|
||
|
#include <kern/ast.h> /* for APC support */
|
||
|
#include <kern/kalloc.h>
|
||
|
#include <kern/task.h> /* extern void *get_bsdtask_info(task_t); */
|
||
|
#include <kern/thread.h>
|
||
|
#include <kern/sched_prim.h>
|
||
|
#include <kern/thread_call.h>
|
||
|
#include <kern/policy_internal.h>
|
||
|
#include <kern/sync_sema.h>
|
||
|
|
||
|
#include <os/log.h>
|
||
|
|
||
|
#include <mach/exception.h>
|
||
|
#include <mach/task.h>
|
||
|
#include <mach/thread_act.h>
|
||
|
#include <libkern/OSAtomic.h>
|
||
|
|
||
|
#include <sys/sdt.h>
|
||
|
#include <sys/codesign.h>
|
||
|
#include <sys/random.h>
|
||
|
#include <libkern/section_keywords.h>
|
||
|
|
||
|
#if CONFIG_MACF
|
||
|
#include <security/mac_framework.h>
|
||
|
#endif
|
||
|
|
||
|
/*
|
||
|
* Missing prototypes that Mach should export
|
||
|
*
|
||
|
* +++
|
||
|
*/
|
||
|
extern int thread_enable_fpe(thread_t act, int onoff);
|
||
|
extern kern_return_t get_signalact(task_t, thread_t *, int);
|
||
|
extern unsigned int get_useraddr(void);
|
||
|
extern boolean_t task_did_exec(task_t task);
|
||
|
extern boolean_t task_is_exec_copy(task_t task);
|
||
|
extern void vm_shared_region_reslide_stale(boolean_t driverkit);
|
||
|
|
||
|
/*
|
||
|
* ---
|
||
|
*/
|
||
|
|
||
|
extern void doexception(int exc, mach_exception_code_t code,
|
||
|
mach_exception_subcode_t sub);
|
||
|
|
||
|
static void stop(proc_t, proc_t);
|
||
|
static int cansignal_nomac(proc_t, kauth_cred_t, proc_t, int);
|
||
|
int cansignal(proc_t, kauth_cred_t, proc_t, int);
|
||
|
int killpg1(proc_t, int, int, int, int);
|
||
|
kern_return_t do_bsdexception(int, int, int);
|
||
|
void __posix_sem_syscall_return(kern_return_t);
|
||
|
char *proc_name_address(void *p);
|
||
|
|
||
|
static int filt_sigattach(struct knote *kn, struct kevent_qos_s *kev);
|
||
|
static void filt_sigdetach(struct knote *kn);
|
||
|
static int filt_signal(struct knote *kn, long hint);
|
||
|
static int filt_signaltouch(struct knote *kn, struct kevent_qos_s *kev);
|
||
|
static int filt_signalprocess(struct knote *kn, struct kevent_qos_s *kev);
|
||
|
|
||
|
SECURITY_READ_ONLY_EARLY(struct filterops) sig_filtops = {
|
||
|
.f_attach = filt_sigattach,
|
||
|
.f_detach = filt_sigdetach,
|
||
|
.f_event = filt_signal,
|
||
|
.f_touch = filt_signaltouch,
|
||
|
.f_process = filt_signalprocess,
|
||
|
};
|
||
|
|
||
|
/* structures and fns for killpg1 iterartion callback and filters */
|
||
|
struct killpg1_filtargs {
|
||
|
bool posix;
|
||
|
proc_t curproc;
|
||
|
};
|
||
|
|
||
|
struct killpg1_iterargs {
|
||
|
proc_t curproc;
|
||
|
kauth_cred_t uc;
|
||
|
int signum;
|
||
|
int nfound;
|
||
|
};
|
||
|
|
||
|
static int killpg1_allfilt(proc_t p, void * arg);
|
||
|
static int killpg1_callback(proc_t p, void * arg);
|
||
|
|
||
|
static int pgsignal_callback(proc_t p, void * arg);
|
||
|
static kern_return_t get_signalthread(proc_t, int, thread_t *);
|
||
|
|
||
|
|
||
|
/* flags for psignal_internal */
|
||
|
#define PSIG_LOCKED 0x1
|
||
|
#define PSIG_VFORK 0x2
|
||
|
#define PSIG_THREAD 0x4
|
||
|
#define PSIG_TRY_THREAD 0x8
|
||
|
|
||
|
static os_reason_t build_signal_reason(int signum, const char *procname);
|
||
|
static void psignal_internal(proc_t p, task_t task, thread_t thread, int flavor, int signum, os_reason_t signal_reason);
|
||
|
|
||
|
/*
|
||
|
* NOTE: Source and target may *NOT* overlap! (target is smaller)
|
||
|
*/
|
||
|
static void
|
||
|
sigaltstack_kern_to_user32(struct kern_sigaltstack *in, struct user32_sigaltstack *out)
|
||
|
{
|
||
|
out->ss_sp = CAST_DOWN_EXPLICIT(user32_addr_t, in->ss_sp);
|
||
|
out->ss_size = CAST_DOWN_EXPLICIT(user32_size_t, in->ss_size);
|
||
|
out->ss_flags = in->ss_flags;
|
||
|
}
|
||
|
|
||
|
static void
|
||
|
sigaltstack_kern_to_user64(struct kern_sigaltstack *in, struct user64_sigaltstack *out)
|
||
|
{
|
||
|
out->ss_sp = in->ss_sp;
|
||
|
out->ss_size = in->ss_size;
|
||
|
out->ss_flags = in->ss_flags;
|
||
|
}
|
||
|
|
||
|
/*
|
||
|
* NOTE: Source and target may are permitted to overlap! (source is smaller);
|
||
|
* this works because we copy fields in order from the end of the struct to
|
||
|
* the beginning.
|
||
|
*/
|
||
|
static void
|
||
|
sigaltstack_user32_to_kern(struct user32_sigaltstack *in, struct kern_sigaltstack *out)
|
||
|
{
|
||
|
out->ss_flags = in->ss_flags;
|
||
|
out->ss_size = in->ss_size;
|
||
|
out->ss_sp = CAST_USER_ADDR_T(in->ss_sp);
|
||
|
}
|
||
|
static void
|
||
|
sigaltstack_user64_to_kern(struct user64_sigaltstack *in, struct kern_sigaltstack *out)
|
||
|
{
|
||
|
out->ss_flags = in->ss_flags;
|
||
|
out->ss_size = (user_size_t)in->ss_size;
|
||
|
out->ss_sp = (user_addr_t)in->ss_sp;
|
||
|
}
|
||
|
|
||
|
static void
|
||
|
sigaction_kern_to_user32(struct kern_sigaction *in, struct user32_sigaction *out)
|
||
|
{
|
||
|
/* This assumes 32 bit __sa_handler is of type sig_t */
|
||
|
out->__sigaction_u.__sa_handler = CAST_DOWN_EXPLICIT(user32_addr_t, in->__sigaction_u.__sa_handler);
|
||
|
out->sa_mask = in->sa_mask;
|
||
|
out->sa_flags = in->sa_flags;
|
||
|
}
|
||
|
static void
|
||
|
sigaction_kern_to_user64(struct kern_sigaction *in, struct user64_sigaction *out)
|
||
|
{
|
||
|
/* This assumes 32 bit __sa_handler is of type sig_t */
|
||
|
out->__sigaction_u.__sa_handler = in->__sigaction_u.__sa_handler;
|
||
|
out->sa_mask = in->sa_mask;
|
||
|
out->sa_flags = in->sa_flags;
|
||
|
}
|
||
|
|
||
|
static void
|
||
|
__sigaction_user32_to_kern(struct __user32_sigaction *in, struct __kern_sigaction *out)
|
||
|
{
|
||
|
out->__sigaction_u.__sa_handler = CAST_USER_ADDR_T(in->__sigaction_u.__sa_handler);
|
||
|
out->sa_tramp = CAST_USER_ADDR_T(in->sa_tramp);
|
||
|
out->sa_mask = in->sa_mask;
|
||
|
out->sa_flags = in->sa_flags;
|
||
|
|
||
|
kern_return_t kr;
|
||
|
kr = machine_thread_function_pointers_convert_from_user(current_thread(),
|
||
|
&out->sa_tramp, 1);
|
||
|
assert(kr == KERN_SUCCESS);
|
||
|
}
|
||
|
|
||
|
static void
|
||
|
__sigaction_user64_to_kern(struct __user64_sigaction *in, struct __kern_sigaction *out)
|
||
|
{
|
||
|
out->__sigaction_u.__sa_handler = (user_addr_t)in->__sigaction_u.__sa_handler;
|
||
|
out->sa_tramp = (user_addr_t)in->sa_tramp;
|
||
|
out->sa_mask = in->sa_mask;
|
||
|
out->sa_flags = in->sa_flags;
|
||
|
|
||
|
kern_return_t kr;
|
||
|
kr = machine_thread_function_pointers_convert_from_user(current_thread(),
|
||
|
&out->sa_tramp, 1);
|
||
|
assert(kr == KERN_SUCCESS);
|
||
|
}
|
||
|
|
||
|
#if SIGNAL_DEBUG
|
||
|
void ram_printf(int);
|
||
|
int ram_debug = 0;
|
||
|
unsigned int rdebug_proc = 0;
|
||
|
void
|
||
|
ram_printf(int x)
|
||
|
{
|
||
|
printf("x is %d", x);
|
||
|
}
|
||
|
#endif /* SIGNAL_DEBUG */
|
||
|
|
||
|
|
||
|
void
|
||
|
signal_setast(thread_t sig_actthread)
|
||
|
{
|
||
|
act_set_astbsd(sig_actthread);
|
||
|
}
|
||
|
|
||
|
static int
|
||
|
cansignal_nomac(proc_t src, kauth_cred_t uc_src, proc_t dst, int signum)
|
||
|
{
|
||
|
/* you can signal yourself */
|
||
|
if (src == dst) {
|
||
|
return 1;
|
||
|
}
|
||
|
|
||
|
/* you can't send the init proc SIGKILL, even if root */
|
||
|
if (signum == SIGKILL && dst == initproc) {
|
||
|
return 0;
|
||
|
}
|
||
|
|
||
|
/* otherwise, root can always signal */
|
||
|
if (kauth_cred_issuser(uc_src)) {
|
||
|
return 1;
|
||
|
}
|
||
|
|
||
|
/* processes in the same session can send SIGCONT to each other */
|
||
|
if (signum == SIGCONT && proc_sessionid(src) == proc_sessionid(dst)) {
|
||
|
return 1;
|
||
|
}
|
||
|
|
||
|
#if XNU_TARGET_OS_IOS
|
||
|
// Allow debugging of third party drivers on iOS
|
||
|
if (proc_is_third_party_debuggable_driver(dst)) {
|
||
|
return 1;
|
||
|
}
|
||
|
#endif /* XNU_TARGET_OS_IOS */
|
||
|
|
||
|
/* the source process must be authorized to signal the target */
|
||
|
{
|
||
|
int allowed = 0;
|
||
|
kauth_cred_t uc_dst = NOCRED, uc_ref = NOCRED;
|
||
|
|
||
|
uc_dst = uc_ref = kauth_cred_proc_ref(dst);
|
||
|
|
||
|
/*
|
||
|
* If the real or effective UID of the sender matches the real or saved
|
||
|
* UID of the target, allow the signal to be sent.
|
||
|
*/
|
||
|
if (kauth_cred_getruid(uc_src) == kauth_cred_getruid(uc_dst) ||
|
||
|
kauth_cred_getruid(uc_src) == kauth_cred_getsvuid(uc_dst) ||
|
||
|
kauth_cred_getuid(uc_src) == kauth_cred_getruid(uc_dst) ||
|
||
|
kauth_cred_getuid(uc_src) == kauth_cred_getsvuid(uc_dst)) {
|
||
|
allowed = 1;
|
||
|
}
|
||
|
|
||
|
if (uc_ref != NOCRED) {
|
||
|
kauth_cred_unref(&uc_ref);
|
||
|
uc_ref = NOCRED;
|
||
|
}
|
||
|
|
||
|
return allowed;
|
||
|
}
|
||
|
}
|
||
|
|
||
|
/*
|
||
|
* Can process `src`, with ucred `uc_src`, send the signal `signum` to process
|
||
|
* `dst`? The ucred is referenced by the caller so internal fileds can be used
|
||
|
* safely.
|
||
|
*/
|
||
|
int
|
||
|
cansignal(proc_t src, kauth_cred_t uc_src, proc_t dst, int signum)
|
||
|
{
|
||
|
#if CONFIG_MACF
|
||
|
if (mac_proc_check_signal(src, dst, signum)) {
|
||
|
return 0;
|
||
|
}
|
||
|
#endif
|
||
|
|
||
|
return cansignal_nomac(src, uc_src, dst, signum);
|
||
|
}
|
||
|
|
||
|
/*
|
||
|
* <rdar://problem/21952708> Some signals can be restricted from being handled,
|
||
|
* forcing the default action for that signal. This behavior applies only to
|
||
|
* non-root (EUID != 0) processes, and is configured with the "sigrestrict=x"
|
||
|
* bootarg:
|
||
|
*
|
||
|
* 0 (default): Disallow use of restricted signals. Trying to register a handler
|
||
|
* returns ENOTSUP, which userspace may use to take special action (e.g. abort).
|
||
|
* 1: As above, but return EINVAL. Restricted signals behave similarly to SIGKILL.
|
||
|
* 2: Usual POSIX semantics.
|
||
|
*/
|
||
|
static TUNABLE(unsigned, sigrestrict_arg, "sigrestrict", 0);
|
||
|
|
||
|
#if XNU_PLATFORM_WatchOS
|
||
|
static int
|
||
|
sigrestrictmask(void)
|
||
|
{
|
||
|
if (kauth_getuid() != 0 && sigrestrict_arg != 2) {
|
||
|
return SIGRESTRICTMASK;
|
||
|
}
|
||
|
return 0;
|
||
|
}
|
||
|
|
||
|
static int
|
||
|
signal_is_restricted(proc_t p, int signum)
|
||
|
{
|
||
|
if (sigmask(signum) & sigrestrictmask()) {
|
||
|
if (sigrestrict_arg == 0 &&
|
||
|
task_get_apptype(proc_task(p)) == TASK_APPTYPE_APP_DEFAULT) {
|
||
|
return ENOTSUP;
|
||
|
} else {
|
||
|
return EINVAL;
|
||
|
}
|
||
|
}
|
||
|
return 0;
|
||
|
}
|
||
|
|
||
|
#else
|
||
|
|
||
|
static inline int
|
||
|
signal_is_restricted(proc_t p, int signum)
|
||
|
{
|
||
|
(void)p;
|
||
|
(void)signum;
|
||
|
return 0;
|
||
|
}
|
||
|
#endif /* !XNU_PLATFORM_WatchOS */
|
||
|
|
||
|
/*
|
||
|
* Returns: 0 Success
|
||
|
* EINVAL
|
||
|
* copyout:EFAULT
|
||
|
* copyin:EFAULT
|
||
|
*
|
||
|
* Notes: Uses current thread as a parameter to inform PPC to enable
|
||
|
* FPU exceptions via setsigvec(); this operation is not proxy
|
||
|
* safe!
|
||
|
*/
|
||
|
/* ARGSUSED */
|
||
|
int
|
||
|
sigaction(proc_t p, struct sigaction_args *uap, __unused int32_t *retval)
|
||
|
{
|
||
|
struct kern_sigaction vec;
|
||
|
struct __kern_sigaction __vec;
|
||
|
|
||
|
struct kern_sigaction *sa = &vec;
|
||
|
struct sigacts *ps = &p->p_sigacts;
|
||
|
|
||
|
int signum;
|
||
|
int bit, error = 0;
|
||
|
uint32_t sigreturn_validation = PS_SIGRETURN_VALIDATION_DEFAULT;
|
||
|
|
||
|
signum = uap->signum;
|
||
|
if (signum <= 0 || signum >= NSIG ||
|
||
|
signum == SIGKILL || signum == SIGSTOP) {
|
||
|
return EINVAL;
|
||
|
}
|
||
|
|
||
|
if (uap->nsa) {
|
||
|
if (IS_64BIT_PROCESS(p)) {
|
||
|
struct __user64_sigaction __vec64;
|
||
|
error = copyin(uap->nsa, &__vec64, sizeof(__vec64));
|
||
|
__sigaction_user64_to_kern(&__vec64, &__vec);
|
||
|
} else {
|
||
|
struct __user32_sigaction __vec32;
|
||
|
error = copyin(uap->nsa, &__vec32, sizeof(__vec32));
|
||
|
__sigaction_user32_to_kern(&__vec32, &__vec);
|
||
|
}
|
||
|
if (error) {
|
||
|
return error;
|
||
|
}
|
||
|
|
||
|
sigreturn_validation = (__vec.sa_flags & SA_VALIDATE_SIGRETURN_FROM_SIGTRAMP) ?
|
||
|
PS_SIGRETURN_VALIDATION_ENABLED : PS_SIGRETURN_VALIDATION_DISABLED;
|
||
|
__vec.sa_flags &= SA_USERSPACE_MASK; /* Only pass on valid sa_flags */
|
||
|
|
||
|
if ((__vec.sa_flags & SA_SIGINFO) || __vec.sa_handler != SIG_DFL) {
|
||
|
if ((error = signal_is_restricted(p, signum))) {
|
||
|
if (error == ENOTSUP) {
|
||
|
printf("%s(%d): denied attempt to register action for signal %d\n",
|
||
|
proc_name_address(p), proc_pid(p), signum);
|
||
|
}
|
||
|
return error;
|
||
|
}
|
||
|
}
|
||
|
}
|
||
|
|
||
|
if (uap->osa) {
|
||
|
sa->sa_handler = SIGACTION(p, signum);
|
||
|
sa->sa_mask = ps->ps_catchmask[signum];
|
||
|
bit = sigmask(signum);
|
||
|
sa->sa_flags = 0;
|
||
|
if ((ps->ps_sigonstack & bit) != 0) {
|
||
|
sa->sa_flags |= SA_ONSTACK;
|
||
|
}
|
||
|
if ((ps->ps_sigintr & bit) == 0) {
|
||
|
sa->sa_flags |= SA_RESTART;
|
||
|
}
|
||
|
if (ps->ps_siginfo & bit) {
|
||
|
sa->sa_flags |= SA_SIGINFO;
|
||
|
}
|
||
|
if (ps->ps_signodefer & bit) {
|
||
|
sa->sa_flags |= SA_NODEFER;
|
||
|
}
|
||
|
if ((signum == SIGCHLD) && (p->p_flag & P_NOCLDSTOP)) {
|
||
|
sa->sa_flags |= SA_NOCLDSTOP;
|
||
|
}
|
||
|
if ((signum == SIGCHLD) && (p->p_flag & P_NOCLDWAIT)) {
|
||
|
sa->sa_flags |= SA_NOCLDWAIT;
|
||
|
}
|
||
|
|
||
|
if (IS_64BIT_PROCESS(p)) {
|
||
|
struct user64_sigaction vec64 = {};
|
||
|
sigaction_kern_to_user64(sa, &vec64);
|
||
|
error = copyout(&vec64, uap->osa, sizeof(vec64));
|
||
|
} else {
|
||
|
struct user32_sigaction vec32 = {};
|
||
|
sigaction_kern_to_user32(sa, &vec32);
|
||
|
error = copyout(&vec32, uap->osa, sizeof(vec32));
|
||
|
}
|
||
|
if (error) {
|
||
|
return error;
|
||
|
}
|
||
|
}
|
||
|
|
||
|
if (uap->nsa) {
|
||
|
uint32_t old_sigreturn_validation = atomic_load_explicit(
|
||
|
&ps->ps_sigreturn_validation, memory_order_relaxed);
|
||
|
if (old_sigreturn_validation == PS_SIGRETURN_VALIDATION_DEFAULT) {
|
||
|
atomic_compare_exchange_strong_explicit(&ps->ps_sigreturn_validation,
|
||
|
&old_sigreturn_validation, sigreturn_validation,
|
||
|
memory_order_relaxed, memory_order_relaxed);
|
||
|
}
|
||
|
error = setsigvec(p, current_thread(), signum, &__vec, FALSE);
|
||
|
}
|
||
|
|
||
|
return error;
|
||
|
}
|
||
|
|
||
|
/* Routines to manipulate bits on all threads */
|
||
|
int
|
||
|
clear_procsiglist(proc_t p, int bit, boolean_t in_signalstart)
|
||
|
{
|
||
|
struct uthread * uth;
|
||
|
|
||
|
proc_lock(p);
|
||
|
if (!in_signalstart) {
|
||
|
proc_signalstart(p, 1);
|
||
|
}
|
||
|
|
||
|
|
||
|
TAILQ_FOREACH(uth, &p->p_uthlist, uu_list) {
|
||
|
uth->uu_siglist &= ~bit;
|
||
|
}
|
||
|
p->p_siglist &= ~bit;
|
||
|
if (!in_signalstart) {
|
||
|
proc_signalend(p, 1);
|
||
|
}
|
||
|
proc_unlock(p);
|
||
|
|
||
|
return 0;
|
||
|
}
|
||
|
|
||
|
|
||
|
static int
|
||
|
unblock_procsigmask(proc_t p, int bit)
|
||
|
{
|
||
|
struct uthread * uth;
|
||
|
|
||
|
proc_lock(p);
|
||
|
proc_signalstart(p, 1);
|
||
|
|
||
|
|
||
|
TAILQ_FOREACH(uth, &p->p_uthlist, uu_list) {
|
||
|
uth->uu_sigmask &= ~bit;
|
||
|
}
|
||
|
p->p_sigmask &= ~bit;
|
||
|
|
||
|
proc_signalend(p, 1);
|
||
|
proc_unlock(p);
|
||
|
return 0;
|
||
|
}
|
||
|
|
||
|
static int
|
||
|
block_procsigmask(proc_t p, int bit)
|
||
|
{
|
||
|
struct uthread * uth;
|
||
|
|
||
|
proc_lock(p);
|
||
|
proc_signalstart(p, 1);
|
||
|
|
||
|
|
||
|
TAILQ_FOREACH(uth, &p->p_uthlist, uu_list) {
|
||
|
uth->uu_sigmask |= bit;
|
||
|
}
|
||
|
p->p_sigmask |= bit;
|
||
|
|
||
|
proc_signalend(p, 1);
|
||
|
proc_unlock(p);
|
||
|
return 0;
|
||
|
}
|
||
|
|
||
|
int
|
||
|
set_procsigmask(proc_t p, int bit)
|
||
|
{
|
||
|
struct uthread * uth;
|
||
|
|
||
|
proc_lock(p);
|
||
|
proc_signalstart(p, 1);
|
||
|
|
||
|
|
||
|
TAILQ_FOREACH(uth, &p->p_uthlist, uu_list) {
|
||
|
uth->uu_sigmask = bit;
|
||
|
}
|
||
|
p->p_sigmask = bit;
|
||
|
proc_signalend(p, 1);
|
||
|
proc_unlock(p);
|
||
|
|
||
|
return 0;
|
||
|
}
|
||
|
|
||
|
/* XXX should be static? */
|
||
|
/*
|
||
|
* Notes: The thread parameter is used in the PPC case to select the
|
||
|
* thread on which the floating point exception will be enabled
|
||
|
* or disabled. We can't simply take current_thread(), since
|
||
|
* this is called from posix_spawn() on the not currently running
|
||
|
* process/thread pair.
|
||
|
*
|
||
|
* We mark thread as unused to alow compilation without warning
|
||
|
* on non-PPC platforms.
|
||
|
*/
|
||
|
int
|
||
|
setsigvec(proc_t p, __unused thread_t thread, int signum, struct __kern_sigaction *sa, boolean_t in_sigstart)
|
||
|
{
|
||
|
struct sigacts *ps = &p->p_sigacts;
|
||
|
int bit;
|
||
|
|
||
|
assert(signum < NSIG);
|
||
|
|
||
|
if ((signum == SIGKILL || signum == SIGSTOP) &&
|
||
|
sa->sa_handler != SIG_DFL) {
|
||
|
return EINVAL;
|
||
|
}
|
||
|
bit = sigmask(signum);
|
||
|
/*
|
||
|
* Change setting atomically.
|
||
|
*/
|
||
|
proc_set_sigact_trampact(p, signum, sa->sa_handler, sa->sa_tramp);
|
||
|
ps->ps_catchmask[signum] = sa->sa_mask & ~sigcantmask;
|
||
|
if (sa->sa_flags & SA_SIGINFO) {
|
||
|
ps->ps_siginfo |= bit;
|
||
|
} else {
|
||
|
ps->ps_siginfo &= ~bit;
|
||
|
}
|
||
|
if ((sa->sa_flags & SA_RESTART) == 0) {
|
||
|
ps->ps_sigintr |= bit;
|
||
|
} else {
|
||
|
ps->ps_sigintr &= ~bit;
|
||
|
}
|
||
|
if (sa->sa_flags & SA_ONSTACK) {
|
||
|
ps->ps_sigonstack |= bit;
|
||
|
} else {
|
||
|
ps->ps_sigonstack &= ~bit;
|
||
|
}
|
||
|
if (sa->sa_flags & SA_RESETHAND) {
|
||
|
ps->ps_sigreset |= bit;
|
||
|
} else {
|
||
|
ps->ps_sigreset &= ~bit;
|
||
|
}
|
||
|
if (sa->sa_flags & SA_NODEFER) {
|
||
|
ps->ps_signodefer |= bit;
|
||
|
} else {
|
||
|
ps->ps_signodefer &= ~bit;
|
||
|
}
|
||
|
if (signum == SIGCHLD) {
|
||
|
if (sa->sa_flags & SA_NOCLDSTOP) {
|
||
|
OSBitOrAtomic(P_NOCLDSTOP, &p->p_flag);
|
||
|
} else {
|
||
|
OSBitAndAtomic(~((uint32_t)P_NOCLDSTOP), &p->p_flag);
|
||
|
}
|
||
|
if ((sa->sa_flags & SA_NOCLDWAIT) || (sa->sa_handler == SIG_IGN)) {
|
||
|
OSBitOrAtomic(P_NOCLDWAIT, &p->p_flag);
|
||
|
} else {
|
||
|
OSBitAndAtomic(~((uint32_t)P_NOCLDWAIT), &p->p_flag);
|
||
|
}
|
||
|
}
|
||
|
|
||
|
/*
|
||
|
* Set bit in p_sigignore for signals that are set to SIG_IGN,
|
||
|
* and for signals set to SIG_DFL where the default is to ignore.
|
||
|
* However, don't put SIGCONT in p_sigignore,
|
||
|
* as we have to restart the process.
|
||
|
*/
|
||
|
if (sa->sa_handler == SIG_IGN ||
|
||
|
(sigprop[signum] & SA_IGNORE && sa->sa_handler == SIG_DFL)) {
|
||
|
clear_procsiglist(p, bit, in_sigstart);
|
||
|
if (signum != SIGCONT) {
|
||
|
p->p_sigignore |= bit; /* easier in psignal */
|
||
|
}
|
||
|
p->p_sigcatch &= ~bit;
|
||
|
} else {
|
||
|
p->p_sigignore &= ~bit;
|
||
|
if (sa->sa_handler == SIG_DFL) {
|
||
|
p->p_sigcatch &= ~bit;
|
||
|
} else {
|
||
|
p->p_sigcatch |= bit;
|
||
|
}
|
||
|
}
|
||
|
return 0;
|
||
|
}
|
||
|
|
||
|
/*
|
||
|
* Initialize signal state for process 0;
|
||
|
* set to ignore signals that are ignored by default.
|
||
|
*/
|
||
|
void
|
||
|
siginit(proc_t p)
|
||
|
{
|
||
|
int i;
|
||
|
|
||
|
for (i = 1; i < NSIG; i++) {
|
||
|
if (sigprop[i] & SA_IGNORE && i != SIGCONT) {
|
||
|
p->p_sigignore |= sigmask(i);
|
||
|
}
|
||
|
}
|
||
|
}
|
||
|
|
||
|
/*
|
||
|
* Reset signals for an exec of the specified process.
|
||
|
*/
|
||
|
void
|
||
|
execsigs(proc_t p, thread_t thread)
|
||
|
{
|
||
|
struct sigacts *ps = &p->p_sigacts;
|
||
|
int nc, mask;
|
||
|
struct uthread *ut;
|
||
|
|
||
|
ut = (struct uthread *)get_bsdthread_info(thread);
|
||
|
|
||
|
/*
|
||
|
* transfer saved signal states from the process
|
||
|
* back to the current thread.
|
||
|
*
|
||
|
* NOTE: We do this without the process locked,
|
||
|
* because we are guaranteed to be single-threaded
|
||
|
* by this point in exec and the p_siglist is
|
||
|
* only accessed by threads inside the process.
|
||
|
*/
|
||
|
ut->uu_siglist |= p->p_siglist;
|
||
|
p->p_siglist = 0;
|
||
|
|
||
|
/*
|
||
|
* Reset caught signals. Held signals remain held
|
||
|
* through p_sigmask (unless they were caught,
|
||
|
* and are now ignored by default).
|
||
|
*/
|
||
|
proc_reset_sigact(p, p->p_sigcatch);
|
||
|
while (p->p_sigcatch) {
|
||
|
nc = ffs((unsigned int)p->p_sigcatch);
|
||
|
mask = sigmask(nc);
|
||
|
p->p_sigcatch &= ~mask;
|
||
|
if (sigprop[nc] & SA_IGNORE) {
|
||
|
if (nc != SIGCONT) {
|
||
|
p->p_sigignore |= mask;
|
||
|
}
|
||
|
ut->uu_siglist &= ~mask;
|
||
|
}
|
||
|
}
|
||
|
|
||
|
atomic_store_explicit(&ps->ps_sigreturn_validation,
|
||
|
PS_SIGRETURN_VALIDATION_DEFAULT, memory_order_relaxed);
|
||
|
|
||
|
/*
|
||
|
* Reset stack state to the user stack.
|
||
|
* Clear set of signals caught on the signal stack.
|
||
|
*/
|
||
|
/* thread */
|
||
|
ut->uu_sigstk.ss_flags = SA_DISABLE;
|
||
|
ut->uu_sigstk.ss_size = 0;
|
||
|
ut->uu_sigstk.ss_sp = USER_ADDR_NULL;
|
||
|
ut->uu_flag &= ~UT_ALTSTACK;
|
||
|
/* process */
|
||
|
ps->ps_sigonstack = 0;
|
||
|
}
|
||
|
|
||
|
/*
|
||
|
* Manipulate signal mask.
|
||
|
* Note that we receive new mask, not pointer,
|
||
|
* and return old mask as return value;
|
||
|
* the library stub does the rest.
|
||
|
*/
|
||
|
int
|
||
|
sigprocmask(proc_t p, struct sigprocmask_args *uap, __unused int32_t *retval)
|
||
|
{
|
||
|
int error = 0;
|
||
|
sigset_t oldmask, nmask;
|
||
|
user_addr_t omask = uap->omask;
|
||
|
struct uthread *ut;
|
||
|
|
||
|
ut = current_uthread();
|
||
|
oldmask = ut->uu_sigmask;
|
||
|
|
||
|
if (uap->mask == USER_ADDR_NULL) {
|
||
|
/* just want old mask */
|
||
|
goto out;
|
||
|
}
|
||
|
error = copyin(uap->mask, &nmask, sizeof(sigset_t));
|
||
|
if (error) {
|
||
|
goto out;
|
||
|
}
|
||
|
|
||
|
switch (uap->how) {
|
||
|
case SIG_BLOCK:
|
||
|
block_procsigmask(p, (nmask & ~sigcantmask));
|
||
|
signal_setast(current_thread());
|
||
|
break;
|
||
|
|
||
|
case SIG_UNBLOCK:
|
||
|
unblock_procsigmask(p, (nmask & ~sigcantmask));
|
||
|
signal_setast(current_thread());
|
||
|
break;
|
||
|
|
||
|
case SIG_SETMASK:
|
||
|
set_procsigmask(p, (nmask & ~sigcantmask));
|
||
|
signal_setast(current_thread());
|
||
|
break;
|
||
|
|
||
|
default:
|
||
|
error = EINVAL;
|
||
|
break;
|
||
|
}
|
||
|
out:
|
||
|
if (!error && omask != USER_ADDR_NULL) {
|
||
|
copyout(&oldmask, omask, sizeof(sigset_t));
|
||
|
}
|
||
|
return error;
|
||
|
}
|
||
|
|
||
|
int
|
||
|
sigpending(__unused proc_t p, struct sigpending_args *uap, __unused int32_t *retval)
|
||
|
{
|
||
|
struct uthread *ut;
|
||
|
sigset_t pendlist;
|
||
|
|
||
|
ut = current_uthread();
|
||
|
pendlist = ut->uu_siglist;
|
||
|
|
||
|
if (uap->osv) {
|
||
|
copyout(&pendlist, uap->osv, sizeof(sigset_t));
|
||
|
}
|
||
|
return 0;
|
||
|
}
|
||
|
|
||
|
/*
|
||
|
* Suspend process until signal, providing mask to be set
|
||
|
* in the meantime. Note nonstandard calling convention:
|
||
|
* libc stub passes mask, not pointer, to save a copyin.
|
||
|
*/
|
||
|
|
||
|
static int
|
||
|
sigcontinue(__unused int error)
|
||
|
{
|
||
|
// struct uthread *ut = current_uthread();
|
||
|
unix_syscall_return(EINTR);
|
||
|
}
|
||
|
|
||
|
int
|
||
|
sigsuspend(proc_t p, struct sigsuspend_args *uap, int32_t *retval)
|
||
|
{
|
||
|
__pthread_testcancel(1);
|
||
|
return sigsuspend_nocancel(p, (struct sigsuspend_nocancel_args *)uap, retval);
|
||
|
}
|
||
|
|
||
|
int
|
||
|
sigsuspend_nocancel(proc_t p, struct sigsuspend_nocancel_args *uap, __unused int32_t *retval)
|
||
|
{
|
||
|
struct uthread *ut;
|
||
|
|
||
|
ut = current_uthread();
|
||
|
|
||
|
/*
|
||
|
* When returning from sigpause, we want
|
||
|
* the old mask to be restored after the
|
||
|
* signal handler has finished. Thus, we
|
||
|
* save it here and mark the sigacts structure
|
||
|
* to indicate this.
|
||
|
*/
|
||
|
ut->uu_oldmask = ut->uu_sigmask;
|
||
|
ut->uu_flag |= UT_SAS_OLDMASK;
|
||
|
ut->uu_sigmask = (uap->mask & ~sigcantmask);
|
||
|
(void) tsleep0((caddr_t) p, PPAUSE | PCATCH, "pause", 0, sigcontinue);
|
||
|
/* always return EINTR rather than ERESTART... */
|
||
|
return EINTR;
|
||
|
}
|
||
|
|
||
|
|
||
|
int
|
||
|
__disable_threadsignal(__unused proc_t p,
|
||
|
__unused struct __disable_threadsignal_args *uap,
|
||
|
__unused int32_t *retval)
|
||
|
{
|
||
|
struct uthread *uth;
|
||
|
|
||
|
uth = current_uthread();
|
||
|
|
||
|
/* No longer valid to have any signal delivered */
|
||
|
uth->uu_flag |= (UT_NO_SIGMASK | UT_CANCELDISABLE);
|
||
|
|
||
|
return 0;
|
||
|
}
|
||
|
|
||
|
void
|
||
|
__pthread_testcancel(int presyscall)
|
||
|
{
|
||
|
thread_t self = current_thread();
|
||
|
struct uthread * uthread;
|
||
|
|
||
|
uthread = (struct uthread *)get_bsdthread_info(self);
|
||
|
|
||
|
|
||
|
uthread->uu_flag &= ~UT_NOTCANCELPT;
|
||
|
|
||
|
if ((uthread->uu_flag & (UT_CANCELDISABLE | UT_CANCEL | UT_CANCELED)) == UT_CANCEL) {
|
||
|
if (presyscall != 0) {
|
||
|
unix_syscall_return(EINTR);
|
||
|
/* NOTREACHED */
|
||
|
} else {
|
||
|
thread_abort_safely(self);
|
||
|
}
|
||
|
}
|
||
|
}
|
||
|
|
||
|
|
||
|
|
||
|
int
|
||
|
__pthread_markcancel(__unused proc_t p,
|
||
|
struct __pthread_markcancel_args *uap, __unused int32_t *retval)
|
||
|
{
|
||
|
thread_act_t target_act;
|
||
|
int error = 0;
|
||
|
struct uthread *uth;
|
||
|
|
||
|
target_act = (thread_act_t)port_name_to_thread(uap->thread_port,
|
||
|
PORT_INTRANS_THREAD_IN_CURRENT_TASK);
|
||
|
|
||
|
if (target_act == THR_ACT_NULL) {
|
||
|
return ESRCH;
|
||
|
}
|
||
|
|
||
|
uth = (struct uthread *)get_bsdthread_info(target_act);
|
||
|
|
||
|
if ((uth->uu_flag & (UT_CANCEL | UT_CANCELED)) == 0) {
|
||
|
uth->uu_flag |= (UT_CANCEL | UT_NO_SIGMASK);
|
||
|
if (((uth->uu_flag & UT_NOTCANCELPT) == 0)
|
||
|
&& ((uth->uu_flag & UT_CANCELDISABLE) == 0)) {
|
||
|
thread_abort_safely(target_act);
|
||
|
}
|
||
|
}
|
||
|
|
||
|
thread_deallocate(target_act);
|
||
|
return error;
|
||
|
}
|
||
|
|
||
|
/* if action =0 ; return the cancellation state ,
|
||
|
* if marked for cancellation, make the thread canceled
|
||
|
* if action = 1 ; Enable the cancel handling
|
||
|
* if action = 2; Disable the cancel handling
|
||
|
*/
|
||
|
int
|
||
|
__pthread_canceled(__unused proc_t p,
|
||
|
struct __pthread_canceled_args *uap, __unused int32_t *retval)
|
||
|
{
|
||
|
thread_act_t thread;
|
||
|
struct uthread *uth;
|
||
|
int action = uap->action;
|
||
|
|
||
|
thread = current_thread();
|
||
|
uth = (struct uthread *)get_bsdthread_info(thread);
|
||
|
|
||
|
switch (action) {
|
||
|
case 1:
|
||
|
uth->uu_flag &= ~UT_CANCELDISABLE;
|
||
|
return 0;
|
||
|
case 2:
|
||
|
uth->uu_flag |= UT_CANCELDISABLE;
|
||
|
return 0;
|
||
|
case 0:
|
||
|
default:
|
||
|
if ((uth->uu_flag & (UT_CANCELDISABLE | UT_CANCEL | UT_CANCELED)) == UT_CANCEL) {
|
||
|
uth->uu_flag &= ~UT_CANCEL;
|
||
|
uth->uu_flag |= (UT_CANCELED | UT_NO_SIGMASK);
|
||
|
return 0;
|
||
|
}
|
||
|
return EINVAL;
|
||
|
}
|
||
|
return EINVAL;
|
||
|
}
|
||
|
|
||
|
__attribute__((noreturn))
|
||
|
void
|
||
|
__posix_sem_syscall_return(kern_return_t kern_result)
|
||
|
{
|
||
|
int error = 0;
|
||
|
|
||
|
if (kern_result == KERN_SUCCESS) {
|
||
|
error = 0;
|
||
|
} else if (kern_result == KERN_ABORTED) {
|
||
|
error = EINTR;
|
||
|
} else if (kern_result == KERN_OPERATION_TIMED_OUT) {
|
||
|
error = ETIMEDOUT;
|
||
|
} else {
|
||
|
error = EINVAL;
|
||
|
}
|
||
|
unix_syscall_return(error);
|
||
|
/* does not return */
|
||
|
}
|
||
|
|
||
|
/*
|
||
|
* Returns: 0 Success
|
||
|
* EINTR
|
||
|
* ETIMEDOUT
|
||
|
* EINVAL
|
||
|
* EFAULT if timespec is NULL
|
||
|
*/
|
||
|
int
|
||
|
__semwait_signal(proc_t p, struct __semwait_signal_args *uap,
|
||
|
int32_t *retval)
|
||
|
{
|
||
|
__pthread_testcancel(0);
|
||
|
return __semwait_signal_nocancel(p, (struct __semwait_signal_nocancel_args *)uap, retval);
|
||
|
}
|
||
|
|
||
|
int
|
||
|
__semwait_signal_nocancel(__unused proc_t p, struct __semwait_signal_nocancel_args *uap,
|
||
|
__unused int32_t *retval)
|
||
|
{
|
||
|
kern_return_t kern_result;
|
||
|
mach_timespec_t then;
|
||
|
struct timespec now;
|
||
|
struct user_timespec ts;
|
||
|
boolean_t truncated_timeout = FALSE;
|
||
|
|
||
|
if (uap->timeout) {
|
||
|
ts.tv_sec = (user_time_t)uap->tv_sec;
|
||
|
ts.tv_nsec = uap->tv_nsec;
|
||
|
|
||
|
if ((ts.tv_sec & 0xFFFFFFFF00000000ULL) != 0) {
|
||
|
ts.tv_sec = 0xFFFFFFFF;
|
||
|
ts.tv_nsec = 0;
|
||
|
truncated_timeout = TRUE;
|
||
|
}
|
||
|
|
||
|
if (uap->relative) {
|
||
|
then.tv_sec = (unsigned int)ts.tv_sec;
|
||
|
then.tv_nsec = (clock_res_t)ts.tv_nsec;
|
||
|
} else {
|
||
|
nanotime(&now);
|
||
|
|
||
|
/* if time has elapsed, set time to null timepsec to bailout rightaway */
|
||
|
if (now.tv_sec == ts.tv_sec ?
|
||
|
now.tv_nsec > ts.tv_nsec :
|
||
|
now.tv_sec > ts.tv_sec) {
|
||
|
then.tv_sec = 0;
|
||
|
then.tv_nsec = 0;
|
||
|
} else {
|
||
|
then.tv_sec = (unsigned int)(ts.tv_sec - now.tv_sec);
|
||
|
then.tv_nsec = (clock_res_t)(ts.tv_nsec - now.tv_nsec);
|
||
|
if (then.tv_nsec < 0) {
|
||
|
then.tv_nsec += NSEC_PER_SEC;
|
||
|
then.tv_sec--;
|
||
|
}
|
||
|
}
|
||
|
}
|
||
|
|
||
|
if (uap->mutex_sem == 0) {
|
||
|
kern_result = semaphore_timedwait_trap_internal((mach_port_name_t)uap->cond_sem, then.tv_sec, then.tv_nsec, __posix_sem_syscall_return);
|
||
|
} else {
|
||
|
kern_result = semaphore_timedwait_signal_trap_internal(uap->cond_sem, uap->mutex_sem, then.tv_sec, then.tv_nsec, __posix_sem_syscall_return);
|
||
|
}
|
||
|
} else {
|
||
|
if (uap->mutex_sem == 0) {
|
||
|
kern_result = semaphore_wait_trap_internal(uap->cond_sem, __posix_sem_syscall_return);
|
||
|
} else {
|
||
|
kern_result = semaphore_wait_signal_trap_internal(uap->cond_sem, uap->mutex_sem, __posix_sem_syscall_return);
|
||
|
}
|
||
|
}
|
||
|
|
||
|
if (kern_result == KERN_SUCCESS && !truncated_timeout) {
|
||
|
return 0;
|
||
|
} else if (kern_result == KERN_SUCCESS && truncated_timeout) {
|
||
|
return EINTR; /* simulate an exceptional condition because Mach doesn't support a longer timeout */
|
||
|
} else if (kern_result == KERN_ABORTED) {
|
||
|
return EINTR;
|
||
|
} else if (kern_result == KERN_OPERATION_TIMED_OUT) {
|
||
|
return ETIMEDOUT;
|
||
|
} else {
|
||
|
return EINVAL;
|
||
|
}
|
||
|
}
|
||
|
|
||
|
|
||
|
int
|
||
|
__pthread_kill(__unused proc_t p, struct __pthread_kill_args *uap,
|
||
|
__unused int32_t *retval)
|
||
|
{
|
||
|
thread_t target_act;
|
||
|
int error = 0;
|
||
|
int signum = uap->sig;
|
||
|
struct uthread *uth;
|
||
|
|
||
|
target_act = (thread_t)port_name_to_thread(uap->thread_port,
|
||
|
PORT_INTRANS_OPTIONS_NONE);
|
||
|
|
||
|
if (target_act == THREAD_NULL) {
|
||
|
return ESRCH;
|
||
|
}
|
||
|
if ((u_int)signum >= NSIG) {
|
||
|
error = EINVAL;
|
||
|
goto out;
|
||
|
}
|
||
|
|
||
|
uth = (struct uthread *)get_bsdthread_info(target_act);
|
||
|
|
||
|
if (uth->uu_flag & UT_NO_SIGMASK) {
|
||
|
error = ESRCH;
|
||
|
goto out;
|
||
|
}
|
||
|
|
||
|
/*
|
||
|
* workq threads must have kills enabled through either
|
||
|
* BSDTHREAD_CTL_WORKQ_ALLOW_KILL or BSDTHREAD_CTL_WORKQ_ALLOW_SIGMASK
|
||
|
*/
|
||
|
if ((thread_get_tag(target_act) & THREAD_TAG_WORKQUEUE) &&
|
||
|
!(uth->uu_workq_pthread_kill_allowed || p->p_workq_allow_sigmask)) {
|
||
|
error = ENOTSUP;
|
||
|
goto out;
|
||
|
}
|
||
|
|
||
|
if (signum) {
|
||
|
psignal_uthread(target_act, signum);
|
||
|
}
|
||
|
out:
|
||
|
thread_deallocate(target_act);
|
||
|
return error;
|
||
|
}
|
||
|
|
||
|
|
||
|
int
|
||
|
__pthread_sigmask(__unused proc_t p, struct __pthread_sigmask_args *uap,
|
||
|
__unused int32_t *retval)
|
||
|
{
|
||
|
user_addr_t set = uap->set;
|
||
|
user_addr_t oset = uap->oset;
|
||
|
sigset_t nset;
|
||
|
int error = 0;
|
||
|
struct uthread *ut;
|
||
|
sigset_t oldset;
|
||
|
|
||
|
ut = current_uthread();
|
||
|
oldset = ut->uu_sigmask;
|
||
|
|
||
|
if (set == USER_ADDR_NULL) {
|
||
|
/* need only old mask */
|
||
|
goto out;
|
||
|
}
|
||
|
|
||
|
error = copyin(set, &nset, sizeof(sigset_t));
|
||
|
if (error) {
|
||
|
goto out;
|
||
|
}
|
||
|
|
||
|
switch (uap->how) {
|
||
|
case SIG_BLOCK:
|
||
|
ut->uu_sigmask |= (nset & ~sigcantmask);
|
||
|
break;
|
||
|
|
||
|
case SIG_UNBLOCK:
|
||
|
ut->uu_sigmask &= ~(nset);
|
||
|
signal_setast(current_thread());
|
||
|
break;
|
||
|
|
||
|
case SIG_SETMASK:
|
||
|
ut->uu_sigmask = (nset & ~sigcantmask);
|
||
|
signal_setast(current_thread());
|
||
|
break;
|
||
|
|
||
|
default:
|
||
|
error = EINVAL;
|
||
|
}
|
||
|
out:
|
||
|
if (!error && oset != USER_ADDR_NULL) {
|
||
|
copyout(&oldset, oset, sizeof(sigset_t));
|
||
|
}
|
||
|
|
||
|
return error;
|
||
|
}
|
||
|
|
||
|
/*
|
||
|
* Returns: 0 Success
|
||
|
* EINVAL
|
||
|
* copyin:EFAULT
|
||
|
* copyout:EFAULT
|
||
|
*/
|
||
|
int
|
||
|
__sigwait(proc_t p, struct __sigwait_args *uap, int32_t *retval)
|
||
|
{
|
||
|
__pthread_testcancel(1);
|
||
|
return __sigwait_nocancel(p, (struct __sigwait_nocancel_args *)uap, retval);
|
||
|
}
|
||
|
|
||
|
int
|
||
|
__sigwait_nocancel(proc_t p, struct __sigwait_nocancel_args *uap, __unused int32_t *retval)
|
||
|
{
|
||
|
struct uthread *ut;
|
||
|
struct uthread *uth;
|
||
|
int error = 0;
|
||
|
sigset_t mask;
|
||
|
sigset_t siglist;
|
||
|
sigset_t sigw = 0;
|
||
|
int signum;
|
||
|
|
||
|
ut = current_uthread();
|
||
|
|
||
|
if (uap->set == USER_ADDR_NULL) {
|
||
|
return EINVAL;
|
||
|
}
|
||
|
|
||
|
error = copyin(uap->set, &mask, sizeof(sigset_t));
|
||
|
if (error) {
|
||
|
return error;
|
||
|
}
|
||
|
|
||
|
siglist = (mask & ~sigcantmask);
|
||
|
|
||
|
if (siglist == 0) {
|
||
|
return EINVAL;
|
||
|
}
|
||
|
|
||
|
proc_lock(p);
|
||
|
|
||
|
proc_signalstart(p, 1);
|
||
|
TAILQ_FOREACH(uth, &p->p_uthlist, uu_list) {
|
||
|
if ((sigw = uth->uu_siglist & siglist)) {
|
||
|
break;
|
||
|
}
|
||
|
}
|
||
|
proc_signalend(p, 1);
|
||
|
|
||
|
if (sigw) {
|
||
|
/* The signal was pending on a thread */
|
||
|
goto sigwait1;
|
||
|
}
|
||
|
/*
|
||
|
* When returning from sigwait, we want
|
||
|
* the old mask to be restored after the
|
||
|
* signal handler has finished. Thus, we
|
||
|
* save it here and mark the sigacts structure
|
||
|
* to indicate this.
|
||
|
*/
|
||
|
uth = ut; /* wait for it to be delivered to us */
|
||
|
ut->uu_oldmask = ut->uu_sigmask;
|
||
|
ut->uu_flag |= UT_SAS_OLDMASK;
|
||
|
if (siglist == (sigset_t)0) {
|
||
|
proc_unlock(p);
|
||
|
return EINVAL;
|
||
|
}
|
||
|
/* SIGKILL and SIGSTOP are not maskable as well */
|
||
|
ut->uu_sigmask = ~(siglist | sigcantmask);
|
||
|
ut->uu_sigwait = siglist;
|
||
|
|
||
|
/* No Continuations for now */
|
||
|
error = msleep((caddr_t)&ut->uu_sigwait, &p->p_mlock, PPAUSE | PCATCH, "pause", 0);
|
||
|
|
||
|
if (error == ERESTART) {
|
||
|
error = 0;
|
||
|
}
|
||
|
|
||
|
sigw = (ut->uu_sigwait & siglist);
|
||
|
ut->uu_sigmask = ut->uu_oldmask;
|
||
|
ut->uu_oldmask = 0;
|
||
|
ut->uu_flag &= ~UT_SAS_OLDMASK;
|
||
|
sigwait1:
|
||
|
ut->uu_sigwait = 0;
|
||
|
if (!error) {
|
||
|
signum = ffs((unsigned int)sigw);
|
||
|
if (!signum) {
|
||
|
panic("sigwait with no signal wakeup");
|
||
|
}
|
||
|
/* Clear the pending signal in the thread it was delivered */
|
||
|
uth->uu_siglist &= ~(sigmask(signum));
|
||
|
|
||
|
#if CONFIG_DTRACE
|
||
|
DTRACE_PROC2(signal__clear, int, signum, siginfo_t *, &(ut->t_dtrace_siginfo));
|
||
|
#endif
|
||
|
|
||
|
proc_unlock(p);
|
||
|
if (uap->sig != USER_ADDR_NULL) {
|
||
|
error = copyout(&signum, uap->sig, sizeof(int));
|
||
|
}
|
||
|
} else {
|
||
|
proc_unlock(p);
|
||
|
}
|
||
|
|
||
|
return error;
|
||
|
}
|
||
|
|
||
|
int
|
||
|
sigaltstack(__unused proc_t p, struct sigaltstack_args *uap, __unused int32_t *retval)
|
||
|
{
|
||
|
struct kern_sigaltstack ss;
|
||
|
struct kern_sigaltstack *pstk;
|
||
|
int error;
|
||
|
struct uthread *uth;
|
||
|
int onstack;
|
||
|
|
||
|
uth = current_uthread();
|
||
|
|
||
|
pstk = &uth->uu_sigstk;
|
||
|
if ((uth->uu_flag & UT_ALTSTACK) == 0) {
|
||
|
uth->uu_sigstk.ss_flags |= SA_DISABLE;
|
||
|
}
|
||
|
onstack = pstk->ss_flags & SA_ONSTACK;
|
||
|
if (uap->oss) {
|
||
|
if (IS_64BIT_PROCESS(p)) {
|
||
|
struct user64_sigaltstack ss64 = {};
|
||
|
sigaltstack_kern_to_user64(pstk, &ss64);
|
||
|
error = copyout(&ss64, uap->oss, sizeof(ss64));
|
||
|
} else {
|
||
|
struct user32_sigaltstack ss32 = {};
|
||
|
sigaltstack_kern_to_user32(pstk, &ss32);
|
||
|
error = copyout(&ss32, uap->oss, sizeof(ss32));
|
||
|
}
|
||
|
if (error) {
|
||
|
return error;
|
||
|
}
|
||
|
}
|
||
|
if (uap->nss == USER_ADDR_NULL) {
|
||
|
return 0;
|
||
|
}
|
||
|
if (IS_64BIT_PROCESS(p)) {
|
||
|
struct user64_sigaltstack ss64;
|
||
|
error = copyin(uap->nss, &ss64, sizeof(ss64));
|
||
|
sigaltstack_user64_to_kern(&ss64, &ss);
|
||
|
} else {
|
||
|
struct user32_sigaltstack ss32;
|
||
|
error = copyin(uap->nss, &ss32, sizeof(ss32));
|
||
|
sigaltstack_user32_to_kern(&ss32, &ss);
|
||
|
}
|
||
|
if (error) {
|
||
|
return error;
|
||
|
}
|
||
|
if ((ss.ss_flags & ~SA_DISABLE) != 0) {
|
||
|
return EINVAL;
|
||
|
}
|
||
|
|
||
|
if (ss.ss_flags & SA_DISABLE) {
|
||
|
/* if we are here we are not in the signal handler ;so no need to check */
|
||
|
if (uth->uu_sigstk.ss_flags & SA_ONSTACK) {
|
||
|
return EINVAL;
|
||
|
}
|
||
|
uth->uu_flag &= ~UT_ALTSTACK;
|
||
|
uth->uu_sigstk.ss_flags = ss.ss_flags;
|
||
|
return 0;
|
||
|
}
|
||
|
if (onstack) {
|
||
|
return EPERM;
|
||
|
}
|
||
|
/* The older stacksize was 8K, enforce that one so no compat problems */
|
||
|
#define OLDMINSIGSTKSZ 8*1024
|
||
|
if (ss.ss_size < OLDMINSIGSTKSZ) {
|
||
|
return ENOMEM;
|
||
|
}
|
||
|
uth->uu_flag |= UT_ALTSTACK;
|
||
|
uth->uu_sigstk = ss;
|
||
|
return 0;
|
||
|
}
|
||
|
|
||
|
int
|
||
|
kill(proc_t cp, struct kill_args *uap, __unused int32_t *retval)
|
||
|
{
|
||
|
proc_t p;
|
||
|
kauth_cred_t uc = kauth_cred_get();
|
||
|
int posix = uap->posix; /* !0 if posix behaviour desired */
|
||
|
|
||
|
AUDIT_ARG(pid, uap->pid);
|
||
|
AUDIT_ARG(signum, uap->signum);
|
||
|
|
||
|
if ((u_int)uap->signum >= NSIG) {
|
||
|
return EINVAL;
|
||
|
}
|
||
|
if (uap->pid > 0) {
|
||
|
/* kill single process */
|
||
|
if ((p = proc_find(uap->pid)) == NULL) {
|
||
|
if ((p = pzfind(uap->pid)) != NULL) {
|
||
|
/*
|
||
|
* POSIX 1003.1-2001 requires returning success when killing a
|
||
|
* zombie; see Rationale for kill(2).
|
||
|
*/
|
||
|
return 0;
|
||
|
}
|
||
|
return ESRCH;
|
||
|
}
|
||
|
AUDIT_ARG(process, p);
|
||
|
if (!cansignal(cp, uc, p, uap->signum)) {
|
||
|
proc_rele(p);
|
||
|
return EPERM;
|
||
|
}
|
||
|
if (uap->signum) {
|
||
|
psignal(p, uap->signum);
|
||
|
}
|
||
|
proc_rele(p);
|
||
|
return 0;
|
||
|
}
|
||
|
switch (uap->pid) {
|
||
|
case -1: /* broadcast signal */
|
||
|
return killpg1(cp, uap->signum, 0, 1, posix);
|
||
|
case 0: /* signal own process group */
|
||
|
return killpg1(cp, uap->signum, 0, 0, posix);
|
||
|
default: /* negative explicit process group */
|
||
|
return killpg1(cp, uap->signum, -(uap->pid), 0, posix);
|
||
|
}
|
||
|
/* NOTREACHED */
|
||
|
}
|
||
|
|
||
|
os_reason_t
|
||
|
build_userspace_exit_reason(uint32_t reason_namespace, uint64_t reason_code, user_addr_t payload, uint32_t payload_size,
|
||
|
user_addr_t reason_string, uint64_t reason_flags)
|
||
|
{
|
||
|
os_reason_t exit_reason = OS_REASON_NULL;
|
||
|
|
||
|
int error = 0;
|
||
|
int num_items_to_copy = 0;
|
||
|
uint32_t user_data_to_copy = 0;
|
||
|
char *reason_user_desc = NULL;
|
||
|
size_t reason_user_desc_len = 0;
|
||
|
|
||
|
exit_reason = os_reason_create(reason_namespace, reason_code);
|
||
|
if (exit_reason == OS_REASON_NULL) {
|
||
|
os_log(OS_LOG_DEFAULT, "build_userspace_exit_reason: failed to allocate exit reason\n");
|
||
|
return exit_reason;
|
||
|
}
|
||
|
|
||
|
exit_reason->osr_flags |= OS_REASON_FLAG_FROM_USERSPACE;
|
||
|
|
||
|
/*
|
||
|
* Only apply flags that are allowed to be passed from userspace.
|
||
|
*/
|
||
|
reason_flags = reason_flags & OS_REASON_FLAG_MASK_ALLOWED_FROM_USER;
|
||
|
exit_reason->osr_flags |= reason_flags;
|
||
|
|
||
|
if (!(exit_reason->osr_flags & OS_REASON_FLAG_NO_CRASH_REPORT)) {
|
||
|
exit_reason->osr_flags |= OS_REASON_FLAG_GENERATE_CRASH_REPORT;
|
||
|
}
|
||
|
|
||
|
if (payload != USER_ADDR_NULL) {
|
||
|
if (payload_size == 0) {
|
||
|
os_log(OS_LOG_DEFAULT, "build_userspace_exit_reason: exit reason with namespace %u,"
|
||
|
" nonzero payload but zero length\n", reason_namespace);
|
||
|
exit_reason->osr_flags |= OS_REASON_FLAG_BAD_PARAMS;
|
||
|
payload = USER_ADDR_NULL;
|
||
|
} else {
|
||
|
num_items_to_copy++;
|
||
|
|
||
|
if (payload_size > EXIT_REASON_PAYLOAD_MAX_LEN) {
|
||
|
exit_reason->osr_flags |= OS_REASON_FLAG_PAYLOAD_TRUNCATED;
|
||
|
payload_size = EXIT_REASON_PAYLOAD_MAX_LEN;
|
||
|
}
|
||
|
|
||
|
user_data_to_copy += payload_size;
|
||
|
}
|
||
|
}
|
||
|
|
||
|
if (reason_string != USER_ADDR_NULL) {
|
||
|
reason_user_desc = (char *)kalloc_data(EXIT_REASON_USER_DESC_MAX_LEN, Z_WAITOK);
|
||
|
|
||
|
if (reason_user_desc != NULL) {
|
||
|
error = copyinstr(reason_string, (void *) reason_user_desc,
|
||
|
EXIT_REASON_USER_DESC_MAX_LEN, &reason_user_desc_len);
|
||
|
|
||
|
if (error == 0) {
|
||
|
num_items_to_copy++;
|
||
|
user_data_to_copy += reason_user_desc_len;
|
||
|
} else if (error == ENAMETOOLONG) {
|
||
|
num_items_to_copy++;
|
||
|
reason_user_desc[EXIT_REASON_USER_DESC_MAX_LEN - 1] = '\0';
|
||
|
user_data_to_copy += reason_user_desc_len;
|
||
|
} else {
|
||
|
exit_reason->osr_flags |= OS_REASON_FLAG_FAILED_DATA_COPYIN;
|
||
|
kfree_data(reason_user_desc, EXIT_REASON_USER_DESC_MAX_LEN);
|
||
|
reason_user_desc = NULL;
|
||
|
reason_user_desc_len = 0;
|
||
|
}
|
||
|
}
|
||
|
}
|
||
|
|
||
|
if (num_items_to_copy != 0) {
|
||
|
uint32_t reason_buffer_size_estimate = 0;
|
||
|
mach_vm_address_t data_addr = 0;
|
||
|
|
||
|
reason_buffer_size_estimate = kcdata_estimate_required_buffer_size(num_items_to_copy, user_data_to_copy);
|
||
|
|
||
|
error = os_reason_alloc_buffer(exit_reason, reason_buffer_size_estimate);
|
||
|
if (error != 0) {
|
||
|
os_log(OS_LOG_DEFAULT, "build_userspace_exit_reason: failed to allocate signal reason buffer\n");
|
||
|
goto out_failed_copyin;
|
||
|
}
|
||
|
|
||
|
if (reason_user_desc != NULL && reason_user_desc_len != 0) {
|
||
|
if (KERN_SUCCESS == kcdata_get_memory_addr(&exit_reason->osr_kcd_descriptor,
|
||
|
EXIT_REASON_USER_DESC,
|
||
|
(uint32_t)reason_user_desc_len,
|
||
|
&data_addr)) {
|
||
|
kcdata_memcpy(&exit_reason->osr_kcd_descriptor, (mach_vm_address_t) data_addr,
|
||
|
reason_user_desc, (uint32_t)reason_user_desc_len);
|
||
|
} else {
|
||
|
os_log(OS_LOG_DEFAULT, "build_userspace_exit_reason: failed to allocate space for reason string\n");
|
||
|
goto out_failed_copyin;
|
||
|
}
|
||
|
}
|
||
|
|
||
|
if (payload != USER_ADDR_NULL) {
|
||
|
if (KERN_SUCCESS ==
|
||
|
kcdata_get_memory_addr(&exit_reason->osr_kcd_descriptor,
|
||
|
EXIT_REASON_USER_PAYLOAD,
|
||
|
payload_size,
|
||
|
&data_addr)) {
|
||
|
error = copyin(payload, (void *) data_addr, payload_size);
|
||
|
if (error) {
|
||
|
os_log(OS_LOG_DEFAULT, "build_userspace_exit_reason: failed to copy in payload data with error %d\n", error);
|
||
|
goto out_failed_copyin;
|
||
|
}
|
||
|
} else {
|
||
|
os_log(OS_LOG_DEFAULT, "build_userspace_exit_reason: failed to allocate space for payload data\n");
|
||
|
goto out_failed_copyin;
|
||
|
}
|
||
|
}
|
||
|
}
|
||
|
|
||
|
if (reason_user_desc != NULL) {
|
||
|
kfree_data(reason_user_desc, EXIT_REASON_USER_DESC_MAX_LEN);
|
||
|
reason_user_desc = NULL;
|
||
|
reason_user_desc_len = 0;
|
||
|
}
|
||
|
|
||
|
return exit_reason;
|
||
|
|
||
|
out_failed_copyin:
|
||
|
|
||
|
if (reason_user_desc != NULL) {
|
||
|
kfree_data(reason_user_desc, EXIT_REASON_USER_DESC_MAX_LEN);
|
||
|
reason_user_desc = NULL;
|
||
|
reason_user_desc_len = 0;
|
||
|
}
|
||
|
|
||
|
exit_reason->osr_flags |= OS_REASON_FLAG_FAILED_DATA_COPYIN;
|
||
|
os_reason_alloc_buffer(exit_reason, 0);
|
||
|
return exit_reason;
|
||
|
}
|
||
|
|
||
|
static int
|
||
|
terminate_with_payload_internal(struct proc *cur_proc, int target_pid, uint32_t reason_namespace,
|
||
|
uint64_t reason_code, user_addr_t payload, uint32_t payload_size,
|
||
|
user_addr_t reason_string, uint64_t reason_flags)
|
||
|
{
|
||
|
proc_t target_proc = PROC_NULL;
|
||
|
kauth_cred_t cur_cred = kauth_cred_get();
|
||
|
|
||
|
os_reason_t signal_reason = OS_REASON_NULL;
|
||
|
|
||
|
AUDIT_ARG(pid, target_pid);
|
||
|
if ((target_pid <= 0)) {
|
||
|
return EINVAL;
|
||
|
}
|
||
|
|
||
|
target_proc = proc_find(target_pid);
|
||
|
if (target_proc == PROC_NULL) {
|
||
|
return ESRCH;
|
||
|
}
|
||
|
|
||
|
AUDIT_ARG(process, target_proc);
|
||
|
|
||
|
if (!cansignal(cur_proc, cur_cred, target_proc, SIGKILL)) {
|
||
|
proc_rele(target_proc);
|
||
|
return EPERM;
|
||
|
}
|
||
|
|
||
|
if (target_pid != proc_getpid(cur_proc)) {
|
||
|
/*
|
||
|
* FLAG_ABORT should only be set on terminate_with_reason(getpid()) that
|
||
|
* was a fallback from an unsuccessful abort_with_reason(). In that case
|
||
|
* caller's pid matches the target one. Otherwise remove the flag.
|
||
|
*/
|
||
|
reason_flags &= ~((typeof(reason_flags))OS_REASON_FLAG_ABORT);
|
||
|
}
|
||
|
|
||
|
KERNEL_DEBUG_CONSTANT(BSDDBG_CODE(DBG_BSD_PROC, BSD_PROC_EXITREASON_CREATE) | DBG_FUNC_NONE,
|
||
|
proc_getpid(target_proc), reason_namespace,
|
||
|
reason_code, 0, 0);
|
||
|
|
||
|
signal_reason = build_userspace_exit_reason(reason_namespace, reason_code, payload, payload_size,
|
||
|
reason_string, (reason_flags | OS_REASON_FLAG_NO_CRASHED_TID));
|
||
|
|
||
|
if (target_pid == proc_getpid(cur_proc)) {
|
||
|
/*
|
||
|
* psignal_thread_with_reason() will pend a SIGKILL on the specified thread or
|
||
|
* return if the thread and/or task are already terminating. Either way, the
|
||
|
* current thread won't return to userspace.
|
||
|
*/
|
||
|
psignal_thread_with_reason(target_proc, current_thread(), SIGKILL, signal_reason);
|
||
|
} else {
|
||
|
psignal_with_reason(target_proc, SIGKILL, signal_reason);
|
||
|
}
|
||
|
|
||
|
proc_rele(target_proc);
|
||
|
|
||
|
return 0;
|
||
|
}
|
||
|
|
||
|
int
|
||
|
terminate_with_payload(struct proc *cur_proc, struct terminate_with_payload_args *args,
|
||
|
__unused int32_t *retval)
|
||
|
{
|
||
|
return terminate_with_payload_internal(cur_proc, args->pid, args->reason_namespace, args->reason_code, args->payload,
|
||
|
args->payload_size, args->reason_string, args->reason_flags);
|
||
|
}
|
||
|
|
||
|
static int
|
||
|
killpg1_allfilt(proc_t p, void * arg)
|
||
|
{
|
||
|
struct killpg1_filtargs * kfargp = (struct killpg1_filtargs *)arg;
|
||
|
|
||
|
/*
|
||
|
* Don't signal initproc, a system process, or the current process if POSIX
|
||
|
* isn't specified.
|
||
|
*/
|
||
|
return proc_getpid(p) > 1 && !(p->p_flag & P_SYSTEM) &&
|
||
|
(kfargp->posix ? true : p != kfargp->curproc);
|
||
|
}
|
||
|
|
||
|
static int
|
||
|
killpg1_callback(proc_t p, void *arg)
|
||
|
{
|
||
|
struct killpg1_iterargs *kargp = (struct killpg1_iterargs *)arg;
|
||
|
int signum = kargp->signum;
|
||
|
|
||
|
if (proc_list_exited(p)) {
|
||
|
/*
|
||
|
* Count zombies as found for the purposes of signalling, since POSIX
|
||
|
* 1003.1-2001 sees signalling zombies as successful. If killpg(2) or
|
||
|
* kill(2) with pid -1 only finds zombies that can be signalled, it
|
||
|
* shouldn't return ESRCH. See the Rationale for kill(2).
|
||
|
*
|
||
|
* Don't call into MAC -- it's not expecting signal checks for exited
|
||
|
* processes.
|
||
|
*/
|
||
|
if (cansignal_nomac(kargp->curproc, kargp->uc, p, signum)) {
|
||
|
kargp->nfound++;
|
||
|
}
|
||
|
} else if (cansignal(kargp->curproc, kargp->uc, p, signum)) {
|
||
|
kargp->nfound++;
|
||
|
|
||
|
if (signum != 0) {
|
||
|
psignal(p, signum);
|
||
|
}
|
||
|
}
|
||
|
|
||
|
return PROC_RETURNED;
|
||
|
}
|
||
|
|
||
|
/*
|
||
|
* Common code for kill process group/broadcast kill.
|
||
|
*/
|
||
|
int
|
||
|
killpg1(proc_t curproc, int signum, int pgid, int all, int posix)
|
||
|
{
|
||
|
kauth_cred_t uc;
|
||
|
struct pgrp *pgrp;
|
||
|
int error = 0;
|
||
|
|
||
|
uc = kauth_cred_proc_ref(curproc);
|
||
|
struct killpg1_iterargs karg = {
|
||
|
.curproc = curproc, .uc = uc, .nfound = 0, .signum = signum
|
||
|
};
|
||
|
|
||
|
if (all) {
|
||
|
/*
|
||
|
* Broadcast to all processes that the user can signal (pid was -1).
|
||
|
*/
|
||
|
struct killpg1_filtargs kfarg = {
|
||
|
.posix = posix, .curproc = curproc
|
||
|
};
|
||
|
proc_iterate(PROC_ALLPROCLIST | PROC_ZOMBPROCLIST, killpg1_callback,
|
||
|
&karg, killpg1_allfilt, &kfarg);
|
||
|
} else {
|
||
|
if (pgid == 0) {
|
||
|
/*
|
||
|
* Send to current the current process' process group.
|
||
|
*/
|
||
|
pgrp = proc_pgrp(curproc, NULL);
|
||
|
} else {
|
||
|
pgrp = pgrp_find(pgid);
|
||
|
if (pgrp == NULL) {
|
||
|
error = ESRCH;
|
||
|
goto out;
|
||
|
}
|
||
|
}
|
||
|
|
||
|
pgrp_iterate(pgrp, killpg1_callback, &karg, ^bool (proc_t p) {
|
||
|
if (p == kernproc || p == initproc) {
|
||
|
return false;
|
||
|
}
|
||
|
/* XXX shouldn't this allow signalling zombies? */
|
||
|
return !(p->p_flag & P_SYSTEM) && p->p_stat != SZOMB;
|
||
|
});
|
||
|
pgrp_rele(pgrp);
|
||
|
}
|
||
|
error = (karg.nfound > 0 ? 0 : (posix ? EPERM : ESRCH));
|
||
|
out:
|
||
|
kauth_cred_unref(&uc);
|
||
|
return error;
|
||
|
}
|
||
|
|
||
|
/*
|
||
|
* Send a signal to a process group.
|
||
|
*/
|
||
|
void
|
||
|
gsignal(int pgid, int signum)
|
||
|
{
|
||
|
struct pgrp *pgrp;
|
||
|
|
||
|
if (pgid && (pgrp = pgrp_find(pgid))) {
|
||
|
pgsignal(pgrp, signum, 0);
|
||
|
pgrp_rele(pgrp);
|
||
|
}
|
||
|
}
|
||
|
|
||
|
/*
|
||
|
* Send a signal to a process group. If checkctty is 1,
|
||
|
* limit to members which have a controlling terminal.
|
||
|
*/
|
||
|
|
||
|
static int
|
||
|
pgsignal_callback(proc_t p, void * arg)
|
||
|
{
|
||
|
int signum = *(int*)arg;
|
||
|
|
||
|
psignal(p, signum);
|
||
|
return PROC_RETURNED;
|
||
|
}
|
||
|
|
||
|
void
|
||
|
pgsignal(struct pgrp *pgrp, int signum, int checkctty)
|
||
|
{
|
||
|
if (pgrp == PGRP_NULL) {
|
||
|
return;
|
||
|
}
|
||
|
|
||
|
bool (^filter)(proc_t) = ^bool (proc_t p) {
|
||
|
return p->p_flag & P_CONTROLT;
|
||
|
};
|
||
|
|
||
|
pgrp_iterate(pgrp, pgsignal_callback, &signum, checkctty ? filter : NULL);
|
||
|
}
|
||
|
|
||
|
|
||
|
void
|
||
|
tty_pgsignal_locked(struct tty *tp, int signum, int checkctty)
|
||
|
{
|
||
|
struct pgrp * pg;
|
||
|
|
||
|
pg = tty_pgrp_locked(tp);
|
||
|
if (pg != PGRP_NULL) {
|
||
|
tty_unlock(tp);
|
||
|
pgsignal(pg, signum, checkctty);
|
||
|
pgrp_rele(pg);
|
||
|
tty_lock(tp);
|
||
|
}
|
||
|
}
|
||
|
/*
|
||
|
* Send a signal caused by a trap to a specific thread.
|
||
|
*/
|
||
|
void
|
||
|
threadsignal(thread_t sig_actthread, int signum, mach_exception_code_t code, boolean_t set_exitreason)
|
||
|
{
|
||
|
struct uthread *uth;
|
||
|
struct task * sig_task;
|
||
|
proc_t p;
|
||
|
int mask;
|
||
|
|
||
|
if ((u_int)signum >= NSIG || signum == 0) {
|
||
|
return;
|
||
|
}
|
||
|
|
||
|
mask = sigmask(signum);
|
||
|
if ((mask & threadmask) == 0) {
|
||
|
return;
|
||
|
}
|
||
|
sig_task = get_threadtask(sig_actthread);
|
||
|
p = (proc_t)(get_bsdtask_info(sig_task));
|
||
|
|
||
|
uth = get_bsdthread_info(sig_actthread);
|
||
|
|
||
|
proc_lock(p);
|
||
|
if (!(p->p_lflag & P_LTRACED) && (p->p_sigignore & mask)) {
|
||
|
proc_unlock(p);
|
||
|
return;
|
||
|
}
|
||
|
|
||
|
uth->uu_siglist |= mask;
|
||
|
uth->uu_code = code;
|
||
|
|
||
|
/* Attempt to establish whether the signal will be fatal (mirrors logic in psignal_internal()) */
|
||
|
if (set_exitreason && ((p->p_lflag & P_LTRACED) || (!(uth->uu_sigwait & mask)
|
||
|
&& !(uth->uu_sigmask & mask) && !(p->p_sigcatch & mask))) &&
|
||
|
!(mask & stopsigmask) && !(mask & contsigmask)) {
|
||
|
if (uth->uu_exit_reason == OS_REASON_NULL) {
|
||
|
KERNEL_DEBUG_CONSTANT(BSDDBG_CODE(DBG_BSD_PROC, BSD_PROC_EXITREASON_CREATE) | DBG_FUNC_NONE,
|
||
|
proc_getpid(p), OS_REASON_SIGNAL, signum, 0, 0);
|
||
|
|
||
|
os_reason_t signal_reason = build_signal_reason(signum, "exc handler");
|
||
|
|
||
|
set_thread_exit_reason(sig_actthread, signal_reason, TRUE);
|
||
|
|
||
|
/* We dropped/consumed the reference in set_thread_exit_reason() */
|
||
|
signal_reason = OS_REASON_NULL;
|
||
|
}
|
||
|
}
|
||
|
|
||
|
proc_unlock(p);
|
||
|
|
||
|
/* mark on process as well */
|
||
|
signal_setast(sig_actthread);
|
||
|
}
|
||
|
|
||
|
/* Called with proc locked */
|
||
|
static void
|
||
|
set_thread_extra_flags(task_t task, struct uthread *uth, os_reason_t reason)
|
||
|
{
|
||
|
extern int vm_shared_region_reslide_restrict;
|
||
|
boolean_t reslide_shared_region = FALSE;
|
||
|
boolean_t driver = task_is_driver(task);
|
||
|
assert(uth != NULL);
|
||
|
/*
|
||
|
* Check whether the userland fault address falls within the shared
|
||
|
* region and notify userland if so. To limit the occurrences of shared
|
||
|
* cache resliding - and its associated memory tax - only investigate the
|
||
|
* fault if it is consequence of accessing unmapped memory (SIGSEGV) or
|
||
|
* accessing with incorrect permissions (SIGBUS - KERN_PROTECTION_FAILURE).
|
||
|
*
|
||
|
* This allows launchd to apply special policies around this fault type.
|
||
|
*/
|
||
|
if (reason->osr_namespace == OS_REASON_SIGNAL &&
|
||
|
(reason->osr_code == SIGSEGV ||
|
||
|
(reason->osr_code == SIGBUS && uth->uu_code == KERN_PROTECTION_FAILURE))) {
|
||
|
mach_vm_address_t fault_address = uth->uu_subcode;
|
||
|
|
||
|
/* Address is in userland, so we hard clear any non-canonical bits to 0 here */
|
||
|
fault_address = VM_USER_STRIP_PTR(fault_address);
|
||
|
|
||
|
if (fault_address >= SHARED_REGION_BASE &&
|
||
|
fault_address <= SHARED_REGION_BASE + SHARED_REGION_SIZE) {
|
||
|
/*
|
||
|
* Always report whether the fault happened within the shared cache
|
||
|
* region, but only stale the slide if the resliding is extended
|
||
|
* to all processes or if the process faulting is a platform one.
|
||
|
*/
|
||
|
reason->osr_flags |= OS_REASON_FLAG_SHAREDREGION_FAULT;
|
||
|
|
||
|
#if __has_feature(ptrauth_calls)
|
||
|
if (!vm_shared_region_reslide_restrict || task_is_hardened_binary(current_task())) {
|
||
|
reslide_shared_region = TRUE;
|
||
|
}
|
||
|
#endif /* __has_feature(ptrauth_calls) */
|
||
|
}
|
||
|
|
||
|
if (driver) {
|
||
|
/*
|
||
|
* Always reslide the DriverKit shared region if the driver faulted.
|
||
|
* The memory cost is acceptable because the DriverKit shared cache is small
|
||
|
* and there are relatively few driver processes.
|
||
|
*/
|
||
|
reslide_shared_region = TRUE;
|
||
|
}
|
||
|
}
|
||
|
|
||
|
if (reslide_shared_region) {
|
||
|
vm_shared_region_reslide_stale(driver);
|
||
|
}
|
||
|
}
|
||
|
|
||
|
void
|
||
|
set_thread_exit_reason(void *th, void *reason, boolean_t proc_locked)
|
||
|
{
|
||
|
struct uthread *targ_uth = get_bsdthread_info(th);
|
||
|
struct task *targ_task = get_threadtask(th);
|
||
|
proc_t targ_proc = NULL;
|
||
|
|
||
|
os_reason_t exit_reason = (os_reason_t)reason;
|
||
|
|
||
|
if (exit_reason == OS_REASON_NULL) {
|
||
|
return;
|
||
|
}
|
||
|
|
||
|
if (!proc_locked) {
|
||
|
targ_proc = (proc_t)(get_bsdtask_info(targ_task));
|
||
|
|
||
|
proc_lock(targ_proc);
|
||
|
}
|
||
|
|
||
|
set_thread_extra_flags(targ_task, targ_uth, exit_reason);
|
||
|
|
||
|
if (targ_uth->uu_exit_reason == OS_REASON_NULL) {
|
||
|
targ_uth->uu_exit_reason = exit_reason;
|
||
|
} else {
|
||
|
/* The caller expects that we drop a reference on the exit reason */
|
||
|
os_reason_free(exit_reason);
|
||
|
}
|
||
|
|
||
|
if (!proc_locked) {
|
||
|
assert(targ_proc != NULL);
|
||
|
proc_unlock(targ_proc);
|
||
|
}
|
||
|
}
|
||
|
|
||
|
/*
|
||
|
* get_signalthread
|
||
|
*
|
||
|
* Picks an appropriate thread from a process to target with a signal.
|
||
|
*
|
||
|
* Called with proc locked.
|
||
|
* Returns thread with BSD ast set.
|
||
|
*
|
||
|
* We attempt to deliver a proc-wide signal to the first thread in the task.
|
||
|
* This allows single threaded applications which use signals to
|
||
|
* be able to be linked with multithreaded libraries.
|
||
|
*/
|
||
|
static kern_return_t
|
||
|
get_signalthread(proc_t p, int signum, thread_t * thr)
|
||
|
{
|
||
|
struct uthread *uth;
|
||
|
sigset_t mask = sigmask(signum);
|
||
|
bool skip_wqthreads = true;
|
||
|
|
||
|
*thr = THREAD_NULL;
|
||
|
|
||
|
|
||
|
again:
|
||
|
TAILQ_FOREACH(uth, &p->p_uthlist, uu_list) {
|
||
|
if (((uth->uu_flag & UT_NO_SIGMASK) == 0) &&
|
||
|
(((uth->uu_sigmask & mask) == 0) || (uth->uu_sigwait & mask))) {
|
||
|
thread_t th = get_machthread(uth);
|
||
|
if (skip_wqthreads && (thread_get_tag(th) & THREAD_TAG_WORKQUEUE)) {
|
||
|
/* Workqueue threads may be parked in the kernel unable to
|
||
|
* deliver signals for an extended period of time, so skip them
|
||
|
* in favor of pthreads in a first pass. (rdar://50054475). */
|
||
|
} else if (check_actforsig(proc_task(p), th, 1) == KERN_SUCCESS) {
|
||
|
*thr = th;
|
||
|
return KERN_SUCCESS;
|
||
|
}
|
||
|
}
|
||
|
}
|
||
|
if (skip_wqthreads) {
|
||
|
skip_wqthreads = false;
|
||
|
goto again;
|
||
|
}
|
||
|
if (get_signalact(proc_task(p), thr, 1) == KERN_SUCCESS) {
|
||
|
return KERN_SUCCESS;
|
||
|
}
|
||
|
|
||
|
return KERN_FAILURE;
|
||
|
}
|
||
|
|
||
|
static os_reason_t
|
||
|
build_signal_reason(int signum, const char *procname)
|
||
|
{
|
||
|
os_reason_t signal_reason = OS_REASON_NULL;
|
||
|
proc_t sender_proc = current_proc();
|
||
|
uint32_t reason_buffer_size_estimate = 0, proc_name_length = 0;
|
||
|
const char *default_sender_procname = "unknown";
|
||
|
mach_vm_address_t data_addr;
|
||
|
int ret;
|
||
|
|
||
|
signal_reason = os_reason_create(OS_REASON_SIGNAL, signum);
|
||
|
if (signal_reason == OS_REASON_NULL) {
|
||
|
printf("build_signal_reason: unable to allocate signal reason structure.\n");
|
||
|
return signal_reason;
|
||
|
}
|
||
|
|
||
|
reason_buffer_size_estimate = kcdata_estimate_required_buffer_size(2, sizeof(sender_proc->p_name) +
|
||
|
sizeof(pid_t));
|
||
|
|
||
|
ret = os_reason_alloc_buffer_noblock(signal_reason, reason_buffer_size_estimate);
|
||
|
if (ret != 0) {
|
||
|
printf("build_signal_reason: unable to allocate signal reason buffer.\n");
|
||
|
return signal_reason;
|
||
|
}
|
||
|
|
||
|
if (KERN_SUCCESS == kcdata_get_memory_addr(&signal_reason->osr_kcd_descriptor, KCDATA_TYPE_PID,
|
||
|
sizeof(pid_t), &data_addr)) {
|
||
|
pid_t pid = proc_getpid(sender_proc);
|
||
|
kcdata_memcpy(&signal_reason->osr_kcd_descriptor, data_addr, &pid, sizeof(pid));
|
||
|
} else {
|
||
|
printf("build_signal_reason: exceeded space in signal reason buf, unable to log PID\n");
|
||
|
}
|
||
|
|
||
|
proc_name_length = sizeof(sender_proc->p_name);
|
||
|
if (KERN_SUCCESS == kcdata_get_memory_addr(&signal_reason->osr_kcd_descriptor, KCDATA_TYPE_PROCNAME,
|
||
|
proc_name_length, &data_addr)) {
|
||
|
if (procname) {
|
||
|
char truncated_procname[proc_name_length];
|
||
|
strncpy((char *) &truncated_procname, procname, proc_name_length);
|
||
|
truncated_procname[proc_name_length - 1] = '\0';
|
||
|
|
||
|
kcdata_memcpy(&signal_reason->osr_kcd_descriptor, data_addr, truncated_procname,
|
||
|
(uint32_t)strlen((char *) &truncated_procname));
|
||
|
} else if (*sender_proc->p_name) {
|
||
|
kcdata_memcpy(&signal_reason->osr_kcd_descriptor, data_addr, &sender_proc->p_name,
|
||
|
sizeof(sender_proc->p_name));
|
||
|
} else {
|
||
|
kcdata_memcpy(&signal_reason->osr_kcd_descriptor, data_addr, &default_sender_procname,
|
||
|
(uint32_t)strlen(default_sender_procname) + 1);
|
||
|
}
|
||
|
} else {
|
||
|
printf("build_signal_reason: exceeded space in signal reason buf, unable to log procname\n");
|
||
|
}
|
||
|
|
||
|
return signal_reason;
|
||
|
}
|
||
|
|
||
|
/*
|
||
|
* Send the signal to the process. If the signal has an action, the action
|
||
|
* is usually performed by the target process rather than the caller; we add
|
||
|
* the signal to the set of pending signals for the process.
|
||
|
*
|
||
|
* Always drops a reference on a signal_reason if one is provided, whether via
|
||
|
* passing it to a thread or deallocating directly.
|
||
|
*
|
||
|
* Exceptions:
|
||
|
* o When a stop signal is sent to a sleeping process that takes the
|
||
|
* default action, the process is stopped without awakening it.
|
||
|
* o SIGCONT restarts stopped processes (or puts them back to sleep)
|
||
|
* regardless of the signal action (eg, blocked or ignored).
|
||
|
*
|
||
|
* Other ignored signals are discarded immediately.
|
||
|
*/
|
||
|
static void
|
||
|
psignal_internal(proc_t p, task_t task, thread_t thread, int flavor, int signum, os_reason_t signal_reason)
|
||
|
{
|
||
|
int prop;
|
||
|
user_addr_t action = USER_ADDR_NULL;
|
||
|
proc_t sig_proc;
|
||
|
thread_t sig_thread;
|
||
|
task_t sig_task;
|
||
|
int mask;
|
||
|
struct uthread *uth;
|
||
|
kern_return_t kret;
|
||
|
uid_t r_uid;
|
||
|
proc_t pp;
|
||
|
kauth_cred_t my_cred;
|
||
|
char *launchd_exit_reason_desc = NULL;
|
||
|
boolean_t update_thread_policy = FALSE;
|
||
|
|
||
|
if ((u_int)signum >= NSIG || signum == 0) {
|
||
|
panic("psignal: bad signal number %d", signum);
|
||
|
}
|
||
|
|
||
|
mask = sigmask(signum);
|
||
|
prop = sigprop[signum];
|
||
|
|
||
|
#if SIGNAL_DEBUG
|
||
|
if (rdebug_proc && (p != PROC_NULL) && (p == rdebug_proc)) {
|
||
|
ram_printf(3);
|
||
|
}
|
||
|
#endif /* SIGNAL_DEBUG */
|
||
|
|
||
|
/* catch unexpected initproc kills early for easier debuggging */
|
||
|
if (signum == SIGKILL && p == initproc) {
|
||
|
if (signal_reason == NULL) {
|
||
|
panic_plain("unexpected SIGKILL of %s %s (no reason provided)",
|
||
|
(p->p_name[0] != '\0' ? p->p_name : "initproc"),
|
||
|
((proc_getcsflags(p) & CS_KILLED) ? "(CS_KILLED)" : ""));
|
||
|
} else {
|
||
|
launchd_exit_reason_desc = exit_reason_get_string_desc(signal_reason);
|
||
|
panic_plain("unexpected SIGKILL of %s %s with reason -- namespace %d code 0x%llx description %." LAUNCHD_PANIC_REASON_STRING_MAXLEN "s",
|
||
|
(p->p_name[0] != '\0' ? p->p_name : "initproc"),
|
||
|
((proc_getcsflags(p) & CS_KILLED) ? "(CS_KILLED)" : ""),
|
||
|
signal_reason->osr_namespace, signal_reason->osr_code,
|
||
|
launchd_exit_reason_desc ? launchd_exit_reason_desc : "none");
|
||
|
}
|
||
|
}
|
||
|
|
||
|
/*
|
||
|
* We will need the task pointer later. Grab it now to
|
||
|
* check for a zombie process. Also don't send signals
|
||
|
* to kernel internal tasks.
|
||
|
*/
|
||
|
if (flavor & PSIG_VFORK) {
|
||
|
sig_task = task;
|
||
|
sig_thread = thread;
|
||
|
sig_proc = p;
|
||
|
} else if (flavor & PSIG_THREAD) {
|
||
|
sig_task = get_threadtask(thread);
|
||
|
sig_thread = thread;
|
||
|
sig_proc = (proc_t)get_bsdtask_info(sig_task);
|
||
|
} else if (flavor & PSIG_TRY_THREAD) {
|
||
|
assert((thread == current_thread()) && (p == current_proc()));
|
||
|
sig_task = proc_task(p);
|
||
|
sig_thread = thread;
|
||
|
sig_proc = p;
|
||
|
} else {
|
||
|
sig_task = proc_task(p);
|
||
|
sig_thread = THREAD_NULL;
|
||
|
sig_proc = p;
|
||
|
}
|
||
|
|
||
|
if ((sig_task == TASK_NULL) || is_kerneltask(sig_task)) {
|
||
|
os_reason_free(signal_reason);
|
||
|
return;
|
||
|
}
|
||
|
|
||
|
if ((flavor & (PSIG_VFORK | PSIG_THREAD)) == 0) {
|
||
|
proc_knote(sig_proc, NOTE_SIGNAL | signum);
|
||
|
}
|
||
|
|
||
|
if ((flavor & PSIG_LOCKED) == 0) {
|
||
|
proc_signalstart(sig_proc, 0);
|
||
|
}
|
||
|
|
||
|
/* Don't send signals to a process that has ignored them. */
|
||
|
if (((flavor & PSIG_VFORK) == 0) && ((sig_proc->p_lflag & P_LTRACED) == 0) && (sig_proc->p_sigignore & mask)) {
|
||
|
DTRACE_PROC3(signal__discard, thread_t, sig_thread, proc_t, sig_proc, int, signum);
|
||
|
goto sigout_unlocked;
|
||
|
}
|
||
|
|
||
|
/*
|
||
|
* The proc_lock prevents the targeted thread from being deallocated
|
||
|
* or handling the signal until we're done signaling it.
|
||
|
*
|
||
|
* Once the proc_lock is dropped, we have no guarantee the thread or uthread exists anymore.
|
||
|
*
|
||
|
* XXX: What if the thread goes inactive after the thread passes bsd ast point?
|
||
|
*/
|
||
|
proc_lock(sig_proc);
|
||
|
|
||
|
/*
|
||
|
* Don't send signals to a process which has already exited and thus
|
||
|
* committed to a particular p_xstat exit code.
|
||
|
* Additionally, don't abort the process running 'reboot'.
|
||
|
*/
|
||
|
if (ISSET(sig_proc->p_flag, P_REBOOT) || ISSET(sig_proc->p_lflag, P_LEXIT)) {
|
||
|
DTRACE_PROC3(signal__discard, thread_t, sig_thread, proc_t, sig_proc, int, signum);
|
||
|
goto sigout_locked;
|
||
|
}
|
||
|
|
||
|
if (flavor & PSIG_VFORK) {
|
||
|
action = SIG_DFL;
|
||
|
act_set_astbsd(sig_thread);
|
||
|
kret = KERN_SUCCESS;
|
||
|
} else if (flavor & PSIG_TRY_THREAD) {
|
||
|
uth = get_bsdthread_info(sig_thread);
|
||
|
if (((uth->uu_flag & UT_NO_SIGMASK) == 0) &&
|
||
|
(((uth->uu_sigmask & mask) == 0) || (uth->uu_sigwait & mask)) &&
|
||
|
((kret = check_actforsig(proc_task(sig_proc), sig_thread, 1)) == KERN_SUCCESS)) {
|
||
|
/* deliver to specified thread */
|
||
|
} else {
|
||
|
/* deliver to any willing thread */
|
||
|
kret = get_signalthread(sig_proc, signum, &sig_thread);
|
||
|
}
|
||
|
} else if (flavor & PSIG_THREAD) {
|
||
|
/* If successful return with ast set */
|
||
|
kret = check_actforsig(sig_task, sig_thread, 1);
|
||
|
} else {
|
||
|
/* If successful return with ast set */
|
||
|
kret = get_signalthread(sig_proc, signum, &sig_thread);
|
||
|
}
|
||
|
|
||
|
if (kret != KERN_SUCCESS) {
|
||
|
DTRACE_PROC3(signal__discard, thread_t, sig_thread, proc_t, sig_proc, int, signum);
|
||
|
proc_unlock(sig_proc);
|
||
|
goto sigout_unlocked;
|
||
|
}
|
||
|
|
||
|
uth = get_bsdthread_info(sig_thread);
|
||
|
|
||
|
/*
|
||
|
* If proc is traced, always give parent a chance.
|
||
|
*/
|
||
|
|
||
|
if ((flavor & PSIG_VFORK) == 0) {
|
||
|
if (sig_proc->p_lflag & P_LTRACED) {
|
||
|
action = SIG_DFL;
|
||
|
} else {
|
||
|
/*
|
||
|
* If the signal is being ignored,
|
||
|
* then we forget about it immediately.
|
||
|
* (Note: we don't set SIGCONT in p_sigignore,
|
||
|
* and if it is set to SIG_IGN,
|
||
|
* action will be SIG_DFL here.)
|
||
|
*/
|
||
|
if (sig_proc->p_sigignore & mask) {
|
||
|
goto sigout_locked;
|
||
|
}
|
||
|
|
||
|
if (uth->uu_sigwait & mask) {
|
||
|
action = KERN_SIG_WAIT;
|
||
|
} else if (uth->uu_sigmask & mask) {
|
||
|
action = KERN_SIG_HOLD;
|
||
|
} else if (sig_proc->p_sigcatch & mask) {
|
||
|
action = KERN_SIG_CATCH;
|
||
|
} else {
|
||
|
action = SIG_DFL;
|
||
|
}
|
||
|
}
|
||
|
}
|
||
|
|
||
|
/* TODO: p_nice isn't hooked up to the scheduler... */
|
||
|
if (sig_proc->p_nice > NZERO && action == SIG_DFL && (prop & SA_KILL) &&
|
||
|
(sig_proc->p_lflag & P_LTRACED) == 0) {
|
||
|
sig_proc->p_nice = NZERO;
|
||
|
}
|
||
|
|
||
|
if (prop & SA_CONT) {
|
||
|
uth->uu_siglist &= ~stopsigmask;
|
||
|
}
|
||
|
|
||
|
if (prop & SA_STOP) {
|
||
|
struct pgrp *pg;
|
||
|
/*
|
||
|
* If sending a tty stop signal to a member of an orphaned
|
||
|
* process group, discard the signal here if the action
|
||
|
* is default; don't stop the process below if sleeping,
|
||
|
* and don't clear any pending SIGCONT.
|
||
|
*/
|
||
|
pg = proc_pgrp(sig_proc, NULL);
|
||
|
if (prop & SA_TTYSTOP && pg->pg_jobc == 0 &&
|
||
|
action == SIG_DFL) {
|
||
|
pgrp_rele(pg);
|
||
|
goto sigout_locked;
|
||
|
}
|
||
|
pgrp_rele(pg);
|
||
|
uth->uu_siglist &= ~contsigmask;
|
||
|
}
|
||
|
|
||
|
uth->uu_siglist |= mask;
|
||
|
|
||
|
/*
|
||
|
* Defer further processing for signals which are held,
|
||
|
* except that stopped processes must be continued by SIGCONT.
|
||
|
*/
|
||
|
if ((action == KERN_SIG_HOLD) && ((prop & SA_CONT) == 0 || sig_proc->p_stat != SSTOP)) {
|
||
|
goto sigout_locked;
|
||
|
}
|
||
|
|
||
|
/*
|
||
|
* SIGKILL priority twiddling moved here from above because
|
||
|
* it needs sig_thread. Could merge it into large switch
|
||
|
* below if we didn't care about priority for tracing
|
||
|
* as SIGKILL's action is always SIG_DFL.
|
||
|
*
|
||
|
* TODO: p_nice isn't hooked up to the scheduler...
|
||
|
*/
|
||
|
if ((signum == SIGKILL) && (sig_proc->p_nice > NZERO)) {
|
||
|
sig_proc->p_nice = NZERO;
|
||
|
}
|
||
|
|
||
|
/*
|
||
|
* Process is traced - wake it up (if not already
|
||
|
* stopped) so that it can discover the signal in
|
||
|
* issig() and stop for the parent.
|
||
|
*/
|
||
|
if (sig_proc->p_lflag & P_LTRACED) {
|
||
|
if (sig_proc->p_stat != SSTOP) {
|
||
|
goto runlocked;
|
||
|
} else {
|
||
|
goto sigout_locked;
|
||
|
}
|
||
|
}
|
||
|
|
||
|
if ((flavor & PSIG_VFORK) != 0) {
|
||
|
goto runlocked;
|
||
|
}
|
||
|
|
||
|
if (action == KERN_SIG_WAIT) {
|
||
|
#if CONFIG_DTRACE
|
||
|
/*
|
||
|
* DTrace proc signal-clear returns a siginfo_t. Collect the needed info.
|
||
|
*/
|
||
|
r_uid = kauth_getruid(); /* per thread credential; protected by our thread context */
|
||
|
|
||
|
bzero((caddr_t)&(uth->t_dtrace_siginfo), sizeof(uth->t_dtrace_siginfo));
|
||
|
|
||
|
uth->t_dtrace_siginfo.si_signo = signum;
|
||
|
uth->t_dtrace_siginfo.si_pid = proc_getpid(current_proc());
|
||
|
uth->t_dtrace_siginfo.si_status = W_EXITCODE(signum, 0);
|
||
|
uth->t_dtrace_siginfo.si_uid = r_uid;
|
||
|
uth->t_dtrace_siginfo.si_code = 0;
|
||
|
#endif
|
||
|
uth->uu_sigwait = mask;
|
||
|
uth->uu_siglist &= ~mask;
|
||
|
wakeup(&uth->uu_sigwait);
|
||
|
/* if it is SIGCONT resume whole process */
|
||
|
if (prop & SA_CONT) {
|
||
|
OSBitOrAtomic(P_CONTINUED, &sig_proc->p_flag);
|
||
|
sig_proc->p_contproc = proc_getpid(current_proc());
|
||
|
(void) task_resume_internal(sig_task);
|
||
|
}
|
||
|
goto sigout_locked;
|
||
|
}
|
||
|
|
||
|
if (action != SIG_DFL) {
|
||
|
/*
|
||
|
* User wants to catch the signal.
|
||
|
* Wake up the thread, but don't un-suspend it
|
||
|
* (except for SIGCONT).
|
||
|
*/
|
||
|
if (prop & SA_CONT) {
|
||
|
OSBitOrAtomic(P_CONTINUED, &sig_proc->p_flag);
|
||
|
(void) task_resume_internal(sig_task);
|
||
|
sig_proc->p_stat = SRUN;
|
||
|
} else if (sig_proc->p_stat == SSTOP) {
|
||
|
goto sigout_locked;
|
||
|
}
|
||
|
/*
|
||
|
* Fill out siginfo structure information to pass to the
|
||
|
* signalled process/thread sigaction handler, when it
|
||
|
* wakes up. si_code is 0 because this is an ordinary
|
||
|
* signal, not a SIGCHLD, and so si_status is the signal
|
||
|
* number itself, instead of the child process exit status.
|
||
|
* We shift this left because it will be shifted right before
|
||
|
* it is passed to user space. kind of ugly to use W_EXITCODE
|
||
|
* this way, but it beats defining a new macro.
|
||
|
*
|
||
|
* Note: Avoid the SIGCHLD recursion case!
|
||
|
*/
|
||
|
if (signum != SIGCHLD) {
|
||
|
r_uid = kauth_getruid();
|
||
|
|
||
|
sig_proc->si_pid = proc_getpid(current_proc());
|
||
|
sig_proc->si_status = W_EXITCODE(signum, 0);
|
||
|
sig_proc->si_uid = r_uid;
|
||
|
sig_proc->si_code = 0;
|
||
|
}
|
||
|
|
||
|
goto runlocked;
|
||
|
} else {
|
||
|
/* Default action - varies */
|
||
|
if (mask & stopsigmask) {
|
||
|
assert(signal_reason == NULL);
|
||
|
/*
|
||
|
* These are the signals which by default
|
||
|
* stop a process.
|
||
|
*
|
||
|
* Don't clog system with children of init
|
||
|
* stopped from the keyboard.
|
||
|
*/
|
||
|
if (!(prop & SA_STOP) && sig_proc->p_pptr == initproc) {
|
||
|
uth->uu_siglist &= ~mask;
|
||
|
proc_unlock(sig_proc);
|
||
|
/* siglock still locked, proc_lock not locked */
|
||
|
psignal_locked(sig_proc, SIGKILL);
|
||
|
goto sigout_unlocked;
|
||
|
}
|
||
|
|
||
|
/*
|
||
|
* Stop the task
|
||
|
* if task hasn't already been stopped by
|
||
|
* a signal.
|
||
|
*/
|
||
|
uth->uu_siglist &= ~mask;
|
||
|
if (sig_proc->p_stat != SSTOP) {
|
||
|
sig_proc->p_xstat = signum;
|
||
|
sig_proc->p_stat = SSTOP;
|
||
|
OSBitAndAtomic(~((uint32_t)P_CONTINUED), &sig_proc->p_flag);
|
||
|
sig_proc->p_lflag &= ~P_LWAITED;
|
||
|
proc_signalend(sig_proc, 1);
|
||
|
proc_unlock(sig_proc);
|
||
|
|
||
|
pp = proc_parentholdref(sig_proc);
|
||
|
proc_signalstart(sig_proc, 0);
|
||
|
stop(sig_proc, pp);
|
||
|
if ((pp != PROC_NULL) && ((pp->p_flag & P_NOCLDSTOP) == 0)) {
|
||
|
my_cred = kauth_cred_proc_ref(sig_proc);
|
||
|
r_uid = kauth_cred_getruid(my_cred);
|
||
|
kauth_cred_unref(&my_cred);
|
||
|
|
||
|
proc_lock(sig_proc);
|
||
|
pp->si_pid = proc_getpid(sig_proc);
|
||
|
/*
|
||
|
* POSIX: sigaction for a stopped child
|
||
|
* when sent to the parent must set the
|
||
|
* child's signal number into si_status.
|
||
|
*/
|
||
|
if (signum != SIGSTOP) {
|
||
|
pp->si_status = WEXITSTATUS(sig_proc->p_xstat);
|
||
|
} else {
|
||
|
pp->si_status = W_EXITCODE(signum, signum);
|
||
|
}
|
||
|
pp->si_code = CLD_STOPPED;
|
||
|
pp->si_uid = r_uid;
|
||
|
proc_unlock(sig_proc);
|
||
|
|
||
|
psignal(pp, SIGCHLD);
|
||
|
}
|
||
|
if (pp != PROC_NULL) {
|
||
|
proc_parentdropref(pp, 0);
|
||
|
}
|
||
|
|
||
|
goto sigout_unlocked;
|
||
|
}
|
||
|
|
||
|
goto sigout_locked;
|
||
|
}
|
||
|
|
||
|
DTRACE_PROC3(signal__send, thread_t, sig_thread, proc_t, p, int, signum);
|
||
|
|
||
|
switch (signum) {
|
||
|
/*
|
||
|
* Signals ignored by default have been dealt
|
||
|
* with already, since their bits are on in
|
||
|
* p_sigignore.
|
||
|
*/
|
||
|
|
||
|
case SIGKILL:
|
||
|
/*
|
||
|
* Kill signal always sets process running and
|
||
|
* unsuspends it.
|
||
|
*/
|
||
|
/*
|
||
|
* Process will be running after 'run'
|
||
|
*/
|
||
|
sig_proc->p_stat = SRUN;
|
||
|
/*
|
||
|
* In scenarios where suspend/resume are racing
|
||
|
* the signal we are missing AST_BSD by the time
|
||
|
* we get here, set again to avoid races. This
|
||
|
* was the scenario with spindump enabled shutdowns.
|
||
|
* We would need to cover this approp down the line.
|
||
|
*/
|
||
|
act_set_astbsd(sig_thread);
|
||
|
kret = thread_abort(sig_thread);
|
||
|
update_thread_policy = (kret == KERN_SUCCESS);
|
||
|
|
||
|
if (uth->uu_exit_reason == OS_REASON_NULL) {
|
||
|
if (signal_reason == OS_REASON_NULL) {
|
||
|
KERNEL_DEBUG_CONSTANT(BSDDBG_CODE(DBG_BSD_PROC, BSD_PROC_EXITREASON_CREATE) | DBG_FUNC_NONE,
|
||
|
proc_getpid(sig_proc), OS_REASON_SIGNAL, signum, 0, 0);
|
||
|
|
||
|
signal_reason = build_signal_reason(signum, NULL);
|
||
|
}
|
||
|
|
||
|
os_reason_ref(signal_reason);
|
||
|
set_thread_exit_reason(sig_thread, signal_reason, TRUE);
|
||
|
}
|
||
|
|
||
|
goto sigout_locked;
|
||
|
|
||
|
case SIGCONT:
|
||
|
/*
|
||
|
* Let the process run. If it's sleeping on an
|
||
|
* event, it remains so.
|
||
|
*/
|
||
|
assert(signal_reason == NULL);
|
||
|
OSBitOrAtomic(P_CONTINUED, &sig_proc->p_flag);
|
||
|
sig_proc->p_contproc = proc_getpid(sig_proc);
|
||
|
sig_proc->p_xstat = signum;
|
||
|
|
||
|
(void) task_resume_internal(sig_task);
|
||
|
|
||
|
/*
|
||
|
* When processing a SIGCONT, we need to check
|
||
|
* to see if there are signals pending that
|
||
|
* were not delivered because we had been
|
||
|
* previously stopped. If that's the case,
|
||
|
* we need to thread_abort_safely() to trigger
|
||
|
* interruption of the current system call to
|
||
|
* cause their handlers to fire. If it's only
|
||
|
* the SIGCONT, then don't wake up.
|
||
|
*/
|
||
|
if (((flavor & (PSIG_VFORK | PSIG_THREAD)) == 0) && (((uth->uu_siglist & ~uth->uu_sigmask) & ~sig_proc->p_sigignore) & ~mask)) {
|
||
|
uth->uu_siglist &= ~mask;
|
||
|
sig_proc->p_stat = SRUN;
|
||
|
goto runlocked;
|
||
|
}
|
||
|
|
||
|
uth->uu_siglist &= ~mask;
|
||
|
sig_proc->p_stat = SRUN;
|
||
|
goto sigout_locked;
|
||
|
|
||
|
default:
|
||
|
{
|
||
|
/*
|
||
|
* A signal which has a default action of killing
|
||
|
* the process, and for which there is no handler,
|
||
|
* needs to act like SIGKILL
|
||
|
*
|
||
|
* The thread_sstop condition is a remnant of a fix
|
||
|
* where PSIG_THREAD exit reasons were not set
|
||
|
* correctly (93593933). We keep the behavior with
|
||
|
* SSTOP the same as before.
|
||
|
*/
|
||
|
const bool default_kill = (action == SIG_DFL) && (prop & SA_KILL);
|
||
|
const bool thread_sstop = (flavor & PSIG_THREAD) && (sig_proc->p_stat == SSTOP);
|
||
|
|
||
|
if (default_kill && !thread_sstop) {
|
||
|
sig_proc->p_stat = SRUN;
|
||
|
kret = thread_abort(sig_thread);
|
||
|
update_thread_policy = (kret == KERN_SUCCESS);
|
||
|
|
||
|
if (uth->uu_exit_reason == OS_REASON_NULL) {
|
||
|
if (signal_reason == OS_REASON_NULL) {
|
||
|
KERNEL_DEBUG_CONSTANT(BSDDBG_CODE(DBG_BSD_PROC, BSD_PROC_EXITREASON_CREATE) | DBG_FUNC_NONE,
|
||
|
proc_getpid(sig_proc), OS_REASON_SIGNAL, signum, 0, 0);
|
||
|
|
||
|
signal_reason = build_signal_reason(signum, NULL);
|
||
|
}
|
||
|
|
||
|
os_reason_ref(signal_reason);
|
||
|
set_thread_exit_reason(sig_thread, signal_reason, TRUE);
|
||
|
}
|
||
|
|
||
|
goto sigout_locked;
|
||
|
}
|
||
|
|
||
|
/*
|
||
|
* All other signals wake up the process, but don't
|
||
|
* resume it.
|
||
|
*/
|
||
|
if (sig_proc->p_stat == SSTOP) {
|
||
|
goto sigout_locked;
|
||
|
}
|
||
|
goto runlocked;
|
||
|
}
|
||
|
}
|
||
|
}
|
||
|
/*NOTREACHED*/
|
||
|
|
||
|
runlocked:
|
||
|
/*
|
||
|
* If we're being traced (possibly because someone attached us
|
||
|
* while we were stopped), check for a signal from the debugger.
|
||
|
*/
|
||
|
if (sig_proc->p_stat == SSTOP) {
|
||
|
if ((sig_proc->p_lflag & P_LTRACED) != 0 && sig_proc->p_xstat != 0) {
|
||
|
uth->uu_siglist |= sigmask(sig_proc->p_xstat);
|
||
|
}
|
||
|
|
||
|
if ((flavor & PSIG_VFORK) != 0) {
|
||
|
sig_proc->p_stat = SRUN;
|
||
|
}
|
||
|
} else {
|
||
|
/*
|
||
|
* setrunnable(p) in BSD and
|
||
|
* Wake up the thread if it is interruptible.
|
||
|
*/
|
||
|
sig_proc->p_stat = SRUN;
|
||
|
if ((flavor & PSIG_VFORK) == 0) {
|
||
|
thread_abort_safely(sig_thread);
|
||
|
}
|
||
|
}
|
||
|
|
||
|
sigout_locked:
|
||
|
if (update_thread_policy) {
|
||
|
/*
|
||
|
* Update the thread policy to heading to terminate, increase priority if
|
||
|
* necessary. This needs to be done before we drop the proc lock because the
|
||
|
* thread can take the fatal signal once it's dropped.
|
||
|
*/
|
||
|
proc_set_thread_policy(sig_thread, TASK_POLICY_ATTRIBUTE, TASK_POLICY_TERMINATED, TASK_POLICY_ENABLE);
|
||
|
}
|
||
|
|
||
|
proc_unlock(sig_proc);
|
||
|
|
||
|
sigout_unlocked:
|
||
|
os_reason_free(signal_reason);
|
||
|
if ((flavor & PSIG_LOCKED) == 0) {
|
||
|
proc_signalend(sig_proc, 0);
|
||
|
}
|
||
|
}
|
||
|
|
||
|
void
|
||
|
psignal(proc_t p, int signum)
|
||
|
{
|
||
|
psignal_internal(p, NULL, NULL, 0, signum, NULL);
|
||
|
}
|
||
|
|
||
|
void
|
||
|
psignal_with_reason(proc_t p, int signum, struct os_reason *signal_reason)
|
||
|
{
|
||
|
psignal_internal(p, NULL, NULL, 0, signum, signal_reason);
|
||
|
}
|
||
|
|
||
|
void
|
||
|
psignal_sigkill_with_reason(struct proc *p, struct os_reason *signal_reason)
|
||
|
{
|
||
|
psignal_internal(p, NULL, NULL, 0, SIGKILL, signal_reason);
|
||
|
}
|
||
|
|
||
|
void
|
||
|
psignal_locked(proc_t p, int signum)
|
||
|
{
|
||
|
psignal_internal(p, NULL, NULL, PSIG_LOCKED, signum, NULL);
|
||
|
}
|
||
|
|
||
|
void
|
||
|
psignal_vfork_with_reason(proc_t p, task_t new_task, thread_t thread, int signum, struct os_reason *signal_reason)
|
||
|
{
|
||
|
psignal_internal(p, new_task, thread, PSIG_VFORK, signum, signal_reason);
|
||
|
}
|
||
|
|
||
|
void
|
||
|
psignal_vfork(proc_t p, task_t new_task, thread_t thread, int signum)
|
||
|
{
|
||
|
psignal_internal(p, new_task, thread, PSIG_VFORK, signum, NULL);
|
||
|
}
|
||
|
|
||
|
void
|
||
|
psignal_uthread(thread_t thread, int signum)
|
||
|
{
|
||
|
psignal_internal(PROC_NULL, TASK_NULL, thread, PSIG_THREAD, signum, NULL);
|
||
|
}
|
||
|
|
||
|
/* same as psignal(), but prefer delivery to 'thread' if possible */
|
||
|
void
|
||
|
psignal_try_thread(proc_t p, thread_t thread, int signum)
|
||
|
{
|
||
|
psignal_internal(p, NULL, thread, PSIG_TRY_THREAD, signum, NULL);
|
||
|
}
|
||
|
|
||
|
void
|
||
|
psignal_try_thread_with_reason(proc_t p, thread_t thread, int signum, struct os_reason *signal_reason)
|
||
|
{
|
||
|
psignal_internal(p, TASK_NULL, thread, PSIG_TRY_THREAD, signum, signal_reason);
|
||
|
}
|
||
|
|
||
|
void
|
||
|
psignal_thread_with_reason(proc_t p, thread_t thread, int signum, struct os_reason *signal_reason)
|
||
|
{
|
||
|
psignal_internal(p, TASK_NULL, thread, PSIG_THREAD, signum, signal_reason);
|
||
|
}
|
||
|
|
||
|
/*
|
||
|
* If the current process has received a signal (should be caught or cause
|
||
|
* termination, should interrupt current syscall), return the signal number.
|
||
|
* Stop signals with default action are processed immediately, then cleared;
|
||
|
* they aren't returned. This is checked after each entry to the system for
|
||
|
* a syscall or trap (though this can usually be done without calling issignal
|
||
|
* by checking the pending signal masks in the CURSIG macro.) The normal call
|
||
|
* sequence is
|
||
|
*
|
||
|
* while (signum = CURSIG(curproc))
|
||
|
* postsig(signum);
|
||
|
*/
|
||
|
int
|
||
|
issignal_locked(proc_t p)
|
||
|
{
|
||
|
int signum, mask, prop, sigbits;
|
||
|
thread_t cur_act;
|
||
|
struct uthread * ut;
|
||
|
proc_t pp;
|
||
|
kauth_cred_t my_cred;
|
||
|
int retval = 0;
|
||
|
uid_t r_uid;
|
||
|
|
||
|
cur_act = current_thread();
|
||
|
|
||
|
#if SIGNAL_DEBUG
|
||
|
if (rdebug_proc && (p == rdebug_proc)) {
|
||
|
ram_printf(3);
|
||
|
}
|
||
|
#endif /* SIGNAL_DEBUG */
|
||
|
|
||
|
/*
|
||
|
* Try to grab the signal lock.
|
||
|
*/
|
||
|
if (sig_try_locked(p) <= 0) {
|
||
|
return 0;
|
||
|
}
|
||
|
|
||
|
proc_signalstart(p, 1);
|
||
|
|
||
|
ut = get_bsdthread_info(cur_act);
|
||
|
for (;;) {
|
||
|
sigbits = ut->uu_siglist & ~ut->uu_sigmask;
|
||
|
|
||
|
if (p->p_lflag & P_LPPWAIT) {
|
||
|
sigbits &= ~stopsigmask;
|
||
|
}
|
||
|
if (sigbits == 0) { /* no signal to send */
|
||
|
retval = 0;
|
||
|
goto out;
|
||
|
}
|
||
|
|
||
|
signum = ffs((unsigned int)sigbits);
|
||
|
mask = sigmask(signum);
|
||
|
prop = sigprop[signum];
|
||
|
|
||
|
/*
|
||
|
* We should see pending but ignored signals
|
||
|
* only if P_LTRACED was on when they were posted.
|
||
|
*/
|
||
|
if (mask & p->p_sigignore && (p->p_lflag & P_LTRACED) == 0) {
|
||
|
ut->uu_siglist &= ~mask;
|
||
|
continue;
|
||
|
}
|
||
|
|
||
|
if (p->p_lflag & P_LTRACED && (p->p_lflag & P_LPPWAIT) == 0) {
|
||
|
/*
|
||
|
* If traced, deliver the signal to the debugger, and wait to be
|
||
|
* released.
|
||
|
*/
|
||
|
task_t task;
|
||
|
p->p_xstat = signum;
|
||
|
|
||
|
if (p->p_lflag & P_LSIGEXC) {
|
||
|
p->sigwait = TRUE;
|
||
|
p->sigwait_thread = cur_act;
|
||
|
p->p_stat = SSTOP;
|
||
|
OSBitAndAtomic(~((uint32_t)P_CONTINUED), &p->p_flag);
|
||
|
p->p_lflag &= ~P_LWAITED;
|
||
|
ut->uu_siglist &= ~mask; /* clear the current signal from the pending list */
|
||
|
proc_signalend(p, 1);
|
||
|
proc_unlock(p);
|
||
|
do_bsdexception(EXC_SOFTWARE, EXC_SOFT_SIGNAL, signum);
|
||
|
proc_lock(p);
|
||
|
proc_signalstart(p, 1);
|
||
|
} else {
|
||
|
proc_unlock(p);
|
||
|
my_cred = kauth_cred_proc_ref(p);
|
||
|
r_uid = kauth_cred_getruid(my_cred);
|
||
|
kauth_cred_unref(&my_cred);
|
||
|
|
||
|
/*
|
||
|
* XXX Have to really stop for debuggers;
|
||
|
* XXX stop() doesn't do the right thing.
|
||
|
*/
|
||
|
task = proc_task(p);
|
||
|
task_suspend_internal(task);
|
||
|
|
||
|
proc_lock(p);
|
||
|
p->sigwait = TRUE;
|
||
|
p->sigwait_thread = cur_act;
|
||
|
p->p_stat = SSTOP;
|
||
|
OSBitAndAtomic(~((uint32_t)P_CONTINUED), &p->p_flag);
|
||
|
p->p_lflag &= ~P_LWAITED;
|
||
|
ut->uu_siglist &= ~mask;
|
||
|
|
||
|
proc_signalend(p, 1);
|
||
|
proc_unlock(p);
|
||
|
|
||
|
pp = proc_parentholdref(p);
|
||
|
if (pp != PROC_NULL) {
|
||
|
proc_lock(pp);
|
||
|
pp->si_pid = proc_getpid(p);
|
||
|
pp->p_xhighbits = p->p_xhighbits;
|
||
|
p->p_xhighbits = 0;
|
||
|
pp->si_status = p->p_xstat;
|
||
|
pp->si_code = CLD_TRAPPED;
|
||
|
pp->si_uid = r_uid;
|
||
|
proc_unlock(pp);
|
||
|
|
||
|
psignal(pp, SIGCHLD);
|
||
|
proc_list_lock();
|
||
|
wakeup((caddr_t)pp);
|
||
|
proc_parentdropref(pp, 1);
|
||
|
proc_list_unlock();
|
||
|
}
|
||
|
|
||
|
assert_wait((caddr_t)&p->sigwait, (THREAD_INTERRUPTIBLE));
|
||
|
thread_block(THREAD_CONTINUE_NULL);
|
||
|
proc_lock(p);
|
||
|
proc_signalstart(p, 1);
|
||
|
}
|
||
|
|
||
|
p->sigwait = FALSE;
|
||
|
p->sigwait_thread = NULL;
|
||
|
wakeup((caddr_t)&p->sigwait_thread);
|
||
|
|
||
|
if (signum == SIGKILL || ut->uu_siglist & sigmask(SIGKILL)) {
|
||
|
/*
|
||
|
* Deliver a pending sigkill even if it's not the current signal.
|
||
|
* Necessary for PT_KILL, which should not be delivered to the
|
||
|
* debugger, but we can't differentiate it from any other KILL.
|
||
|
*/
|
||
|
signum = SIGKILL;
|
||
|
goto deliver_sig;
|
||
|
}
|
||
|
|
||
|
/* We may have to quit. */
|
||
|
if (thread_should_abort(current_thread())) {
|
||
|
retval = 0;
|
||
|
goto out;
|
||
|
}
|
||
|
|
||
|
/*
|
||
|
* If parent wants us to take the signal,
|
||
|
* then it will leave it in p->p_xstat;
|
||
|
* otherwise we just look for signals again.
|
||
|
*/
|
||
|
signum = p->p_xstat;
|
||
|
if (signum == 0) {
|
||
|
continue;
|
||
|
}
|
||
|
|
||
|
/*
|
||
|
* Put the new signal into p_siglist. If the
|
||
|
* signal is being masked, look for other signals.
|
||
|
*/
|
||
|
mask = sigmask(signum);
|
||
|
ut->uu_siglist |= mask;
|
||
|
if (ut->uu_sigmask & mask) {
|
||
|
continue;
|
||
|
}
|
||
|
}
|
||
|
|
||
|
/*
|
||
|
* Decide whether the signal should be returned.
|
||
|
* Return the signal's number, or fall through
|
||
|
* to clear it from the pending mask.
|
||
|
*/
|
||
|
|
||
|
switch ((long)SIGACTION(p, signum)) {
|
||
|
case (long)SIG_DFL:
|
||
|
/*
|
||
|
* If there is a pending stop signal to process
|
||
|
* with default action, stop here,
|
||
|
* then clear the signal. However,
|
||
|
* if process is member of an orphaned
|
||
|
* process group, ignore tty stop signals.
|
||
|
*/
|
||
|
if (prop & SA_STOP) {
|
||
|
struct pgrp * pg;
|
||
|
|
||
|
proc_unlock(p);
|
||
|
pg = proc_pgrp(p, NULL);
|
||
|
if (p->p_lflag & P_LTRACED ||
|
||
|
(pg->pg_jobc == 0 &&
|
||
|
prop & SA_TTYSTOP)) {
|
||
|
proc_lock(p);
|
||
|
pgrp_rele(pg);
|
||
|
break; /* ignore signal */
|
||
|
}
|
||
|
pgrp_rele(pg);
|
||
|
if (p->p_stat != SSTOP) {
|
||
|
proc_lock(p);
|
||
|
p->p_xstat = signum;
|
||
|
p->p_stat = SSTOP;
|
||
|
p->p_lflag &= ~P_LWAITED;
|
||
|
proc_signalend(p, 1);
|
||
|
proc_unlock(p);
|
||
|
|
||
|
pp = proc_parentholdref(p);
|
||
|
proc_signalstart(p, 0);
|
||
|
stop(p, pp);
|
||
|
if ((pp != PROC_NULL) && ((pp->p_flag & P_NOCLDSTOP) == 0)) {
|
||
|
my_cred = kauth_cred_proc_ref(p);
|
||
|
r_uid = kauth_cred_getruid(my_cred);
|
||
|
kauth_cred_unref(&my_cred);
|
||
|
|
||
|
proc_lock(pp);
|
||
|
pp->si_pid = proc_getpid(p);
|
||
|
pp->si_status = WEXITSTATUS(p->p_xstat);
|
||
|
pp->si_code = CLD_STOPPED;
|
||
|
pp->si_uid = r_uid;
|
||
|
proc_unlock(pp);
|
||
|
|
||
|
psignal(pp, SIGCHLD);
|
||
|
}
|
||
|
if (pp != PROC_NULL) {
|
||
|
proc_parentdropref(pp, 0);
|
||
|
}
|
||
|
}
|
||
|
proc_lock(p);
|
||
|
break;
|
||
|
} else if (prop & SA_IGNORE) {
|
||
|
/*
|
||
|
* Except for SIGCONT, shouldn't get here.
|
||
|
* Default action is to ignore; drop it.
|
||
|
*/
|
||
|
break; /* ignore signal */
|
||
|
} else {
|
||
|
goto deliver_sig;
|
||
|
}
|
||
|
|
||
|
case (long)SIG_IGN:
|
||
|
/*
|
||
|
* Masking above should prevent us ever trying
|
||
|
* to take action on an ignored signal other
|
||
|
* than SIGCONT, unless process is traced.
|
||
|
*/
|
||
|
if ((prop & SA_CONT) == 0 &&
|
||
|
(p->p_lflag & P_LTRACED) == 0) {
|
||
|
printf("issignal\n");
|
||
|
}
|
||
|
break; /* ignore signal */
|
||
|
|
||
|
default:
|
||
|
/* This signal has an action - deliver it. */
|
||
|
goto deliver_sig;
|
||
|
}
|
||
|
|
||
|
/* If we dropped through, the signal was ignored - remove it from pending list. */
|
||
|
ut->uu_siglist &= ~mask;
|
||
|
} /* for(;;) */
|
||
|
|
||
|
/* NOTREACHED */
|
||
|
|
||
|
deliver_sig:
|
||
|
ut->uu_siglist &= ~mask;
|
||
|
retval = signum;
|
||
|
|
||
|
out:
|
||
|
proc_signalend(p, 1);
|
||
|
return retval;
|
||
|
}
|
||
|
|
||
|
/* called from _sleep */
|
||
|
int
|
||
|
CURSIG(proc_t p)
|
||
|
{
|
||
|
int signum, mask, prop, sigbits;
|
||
|
thread_t cur_act;
|
||
|
struct uthread * ut;
|
||
|
int retnum = 0;
|
||
|
|
||
|
|
||
|
cur_act = current_thread();
|
||
|
|
||
|
ut = get_bsdthread_info(cur_act);
|
||
|
|
||
|
if (ut->uu_siglist == 0) {
|
||
|
return 0;
|
||
|
}
|
||
|
|
||
|
if (((ut->uu_siglist & ~ut->uu_sigmask) == 0) && ((p->p_lflag & P_LTRACED) == 0)) {
|
||
|
return 0;
|
||
|
}
|
||
|
|
||
|
sigbits = ut->uu_siglist & ~ut->uu_sigmask;
|
||
|
|
||
|
for (;;) {
|
||
|
if (p->p_lflag & P_LPPWAIT) {
|
||
|
sigbits &= ~stopsigmask;
|
||
|
}
|
||
|
if (sigbits == 0) { /* no signal to send */
|
||
|
return retnum;
|
||
|
}
|
||
|
|
||
|
signum = ffs((unsigned int)sigbits);
|
||
|
mask = sigmask(signum);
|
||
|
prop = sigprop[signum];
|
||
|
sigbits &= ~mask; /* take the signal out */
|
||
|
|
||
|
/*
|
||
|
* We should see pending but ignored signals
|
||
|
* only if P_LTRACED was on when they were posted.
|
||
|
*/
|
||
|
if (mask & p->p_sigignore && (p->p_lflag & P_LTRACED) == 0) {
|
||
|
continue;
|
||
|
}
|
||
|
|
||
|
if (p->p_lflag & P_LTRACED && (p->p_lflag & P_LPPWAIT) == 0) {
|
||
|
return signum;
|
||
|
}
|
||
|
|
||
|
/*
|
||
|
* Decide whether the signal should be returned.
|
||
|
* Return the signal's number, or fall through
|
||
|
* to clear it from the pending mask.
|
||
|
*/
|
||
|
|
||
|
switch ((long)SIGACTION(p, signum)) {
|
||
|
case (long)SIG_DFL:
|
||
|
/*
|
||
|
* If there is a pending stop signal to process
|
||
|
* with default action, stop here,
|
||
|
* then clear the signal. However,
|
||
|
* if process is member of an orphaned
|
||
|
* process group, ignore tty stop signals.
|
||
|
*/
|
||
|
if (prop & SA_STOP) {
|
||
|
struct pgrp *pg;
|
||
|
|
||
|
pg = proc_pgrp(p, NULL);
|
||
|
|
||
|
if (p->p_lflag & P_LTRACED ||
|
||
|
(pg->pg_jobc == 0 &&
|
||
|
prop & SA_TTYSTOP)) {
|
||
|
pgrp_rele(pg);
|
||
|
break; /* == ignore */
|
||
|
}
|
||
|
pgrp_rele(pg);
|
||
|
retnum = signum;
|
||
|
break;
|
||
|
} else if (prop & SA_IGNORE) {
|
||
|
/*
|
||
|
* Except for SIGCONT, shouldn't get here.
|
||
|
* Default action is to ignore; drop it.
|
||
|
*/
|
||
|
break; /* == ignore */
|
||
|
} else {
|
||
|
return signum;
|
||
|
}
|
||
|
/*NOTREACHED*/
|
||
|
|
||
|
case (long)SIG_IGN:
|
||
|
/*
|
||
|
* Masking above should prevent us ever trying
|
||
|
* to take action on an ignored signal other
|
||
|
* than SIGCONT, unless process is traced.
|
||
|
*/
|
||
|
if ((prop & SA_CONT) == 0 &&
|
||
|
(p->p_lflag & P_LTRACED) == 0) {
|
||
|
printf("issignal\n");
|
||
|
}
|
||
|
break; /* == ignore */
|
||
|
|
||
|
default:
|
||
|
/*
|
||
|
* This signal has an action, let
|
||
|
* postsig() process it.
|
||
|
*/
|
||
|
return signum;
|
||
|
}
|
||
|
}
|
||
|
/* NOTREACHED */
|
||
|
}
|
||
|
|
||
|
/*
|
||
|
* Put the argument process into the stopped state and notify the parent
|
||
|
* via wakeup. Signals are handled elsewhere. The process must not be
|
||
|
* on the run queue.
|
||
|
*/
|
||
|
static void
|
||
|
stop(proc_t p, proc_t parent)
|
||
|
{
|
||
|
OSBitAndAtomic(~((uint32_t)P_CONTINUED), &p->p_flag);
|
||
|
if ((parent != PROC_NULL) && (parent->p_stat != SSTOP)) {
|
||
|
proc_list_lock();
|
||
|
wakeup((caddr_t)parent);
|
||
|
proc_list_unlock();
|
||
|
}
|
||
|
(void) task_suspend_internal(proc_task(p));
|
||
|
}
|
||
|
|
||
|
/*
|
||
|
* Take the action for the specified signal
|
||
|
* from the current set of pending signals.
|
||
|
*/
|
||
|
void
|
||
|
postsig_locked(int signum)
|
||
|
{
|
||
|
proc_t p = current_proc();
|
||
|
struct sigacts *ps = &p->p_sigacts;
|
||
|
user_addr_t catcher;
|
||
|
uint32_t code;
|
||
|
int mask, returnmask;
|
||
|
struct uthread * ut;
|
||
|
os_reason_t ut_exit_reason = OS_REASON_NULL;
|
||
|
int coredump_flags = 0;
|
||
|
|
||
|
#if DIAGNOSTIC
|
||
|
if (signum == 0) {
|
||
|
panic("postsig");
|
||
|
}
|
||
|
/*
|
||
|
* This must be called on master cpu
|
||
|
*/
|
||
|
if (cpu_number() != master_cpu) {
|
||
|
panic("psig not on master");
|
||
|
}
|
||
|
#endif
|
||
|
|
||
|
/*
|
||
|
* Try to grab the signal lock.
|
||
|
*/
|
||
|
if (sig_try_locked(p) <= 0) {
|
||
|
return;
|
||
|
}
|
||
|
|
||
|
proc_signalstart(p, 1);
|
||
|
|
||
|
ut = current_uthread();
|
||
|
mask = sigmask(signum);
|
||
|
ut->uu_siglist &= ~mask;
|
||
|
catcher = SIGACTION(p, signum);
|
||
|
if (catcher == SIG_DFL) {
|
||
|
/*
|
||
|
* Default catcher, where the default is to kill
|
||
|
* the process. (Other cases were ignored above.)
|
||
|
*/
|
||
|
|
||
|
/*
|
||
|
* exit_with_reason() below will consume a reference to the thread's exit reason, so we take another
|
||
|
* reference so the thread still has one even after we call exit_with_reason(). The thread's reference will
|
||
|
* ultimately be destroyed in uthread_cleanup().
|
||
|
*/
|
||
|
ut_exit_reason = ut->uu_exit_reason;
|
||
|
os_reason_ref(ut_exit_reason);
|
||
|
|
||
|
p->p_acflag |= AXSIG;
|
||
|
if (sigprop[signum] & SA_CORE) {
|
||
|
p->p_sigacts.ps_sig = signum;
|
||
|
proc_signalend(p, 1);
|
||
|
proc_unlock(p);
|
||
|
if (task_is_driver(proc_task(p))) {
|
||
|
coredump_flags |= COREDUMP_FULLFSYNC;
|
||
|
}
|
||
|
#if CONFIG_COREDUMP
|
||
|
if (coredump(p, 0, coredump_flags) == 0) {
|
||
|
signum |= WCOREFLAG;
|
||
|
}
|
||
|
#endif
|
||
|
} else {
|
||
|
proc_signalend(p, 1);
|
||
|
proc_unlock(p);
|
||
|
}
|
||
|
|
||
|
#if CONFIG_DTRACE
|
||
|
bzero((caddr_t)&(ut->t_dtrace_siginfo), sizeof(ut->t_dtrace_siginfo));
|
||
|
|
||
|
ut->t_dtrace_siginfo.si_signo = signum;
|
||
|
ut->t_dtrace_siginfo.si_pid = p->si_pid;
|
||
|
ut->t_dtrace_siginfo.si_uid = p->si_uid;
|
||
|
ut->t_dtrace_siginfo.si_status = WEXITSTATUS(p->si_status);
|
||
|
|
||
|
/* Fire DTrace proc:::fault probe when signal is generated by hardware. */
|
||
|
switch (signum) {
|
||
|
case SIGILL: case SIGBUS: case SIGSEGV: case SIGFPE: case SIGTRAP:
|
||
|
DTRACE_PROC2(fault, int, (int)(ut->uu_code), siginfo_t *, &(ut->t_dtrace_siginfo));
|
||
|
break;
|
||
|
default:
|
||
|
break;
|
||
|
}
|
||
|
|
||
|
|
||
|
DTRACE_PROC3(signal__handle, int, signum, siginfo_t *, &(ut->t_dtrace_siginfo),
|
||
|
void (*)(void), SIG_DFL);
|
||
|
#endif
|
||
|
|
||
|
KERNEL_DEBUG_CONSTANT(BSDDBG_CODE(DBG_BSD_PROC, BSD_PROC_FRCEXIT) | DBG_FUNC_NONE,
|
||
|
proc_getpid(p), W_EXITCODE(0, signum), 3, 0, 0);
|
||
|
|
||
|
exit_with_reason(p, W_EXITCODE(0, signum), (int *)NULL, TRUE, TRUE, 0, ut_exit_reason);
|
||
|
|
||
|
proc_lock(p);
|
||
|
return;
|
||
|
} else {
|
||
|
/*
|
||
|
* If we get here, the signal must be caught.
|
||
|
*/
|
||
|
#if DIAGNOSTIC
|
||
|
if (catcher == SIG_IGN || (ut->uu_sigmask & mask)) {
|
||
|
log(LOG_WARNING,
|
||
|
"postsig: processing masked or ignored signal\n");
|
||
|
}
|
||
|
#endif
|
||
|
|
||
|
/*
|
||
|
* Set the new mask value and also defer further
|
||
|
* occurences of this signal.
|
||
|
*
|
||
|
* Special case: user has done a sigpause. Here the
|
||
|
* current mask is not of interest, but rather the
|
||
|
* mask from before the sigpause is what we want
|
||
|
* restored after the signal processing is completed.
|
||
|
*/
|
||
|
if (ut->uu_flag & UT_SAS_OLDMASK) {
|
||
|
returnmask = ut->uu_oldmask;
|
||
|
ut->uu_flag &= ~UT_SAS_OLDMASK;
|
||
|
ut->uu_oldmask = 0;
|
||
|
} else {
|
||
|
returnmask = ut->uu_sigmask;
|
||
|
}
|
||
|
ut->uu_sigmask |= ps->ps_catchmask[signum];
|
||
|
if ((ps->ps_signodefer & mask) == 0) {
|
||
|
ut->uu_sigmask |= mask;
|
||
|
}
|
||
|
sigset_t siginfo = ps->ps_siginfo;
|
||
|
if ((signum != SIGILL) && (signum != SIGTRAP) && (ps->ps_sigreset & mask)) {
|
||
|
if ((signum != SIGCONT) && (sigprop[signum] & SA_IGNORE)) {
|
||
|
p->p_sigignore |= mask;
|
||
|
}
|
||
|
if (SIGACTION(p, signum) != SIG_DFL) {
|
||
|
proc_set_sigact(p, signum, SIG_DFL);
|
||
|
}
|
||
|
ps->ps_siginfo &= ~mask;
|
||
|
ps->ps_signodefer &= ~mask;
|
||
|
}
|
||
|
|
||
|
if (ps->ps_sig != signum) {
|
||
|
code = 0;
|
||
|
} else {
|
||
|
code = ps->ps_code;
|
||
|
ps->ps_code = 0;
|
||
|
}
|
||
|
OSIncrementAtomicLong(&p->p_stats->p_ru.ru_nsignals);
|
||
|
sendsig(p, catcher, signum, returnmask, code, siginfo);
|
||
|
}
|
||
|
proc_signalend(p, 1);
|
||
|
}
|
||
|
|
||
|
/*
|
||
|
* Attach a signal knote to the list of knotes for this process.
|
||
|
*
|
||
|
* Signal knotes share the knote list with proc knotes. This
|
||
|
* could be avoided by using a signal-specific knote list, but
|
||
|
* probably isn't worth the trouble.
|
||
|
*/
|
||
|
|
||
|
static int
|
||
|
filt_sigattach(struct knote *kn, __unused struct kevent_qos_s *kev)
|
||
|
{
|
||
|
proc_t p = current_proc(); /* can attach only to oneself */
|
||
|
|
||
|
proc_klist_lock();
|
||
|
|
||
|
kn->kn_proc = p;
|
||
|
kn->kn_flags |= EV_CLEAR; /* automatically set */
|
||
|
kn->kn_sdata = 0; /* incoming data is ignored */
|
||
|
|
||
|
KNOTE_ATTACH(&p->p_klist, kn);
|
||
|
|
||
|
proc_klist_unlock();
|
||
|
|
||
|
/* edge-triggered events can't have fired before we attached */
|
||
|
return 0;
|
||
|
}
|
||
|
|
||
|
/*
|
||
|
* remove the knote from the process list, if it hasn't already
|
||
|
* been removed by exit processing.
|
||
|
*/
|
||
|
|
||
|
static void
|
||
|
filt_sigdetach(struct knote *kn)
|
||
|
{
|
||
|
proc_t p;
|
||
|
|
||
|
proc_klist_lock();
|
||
|
p = kn->kn_proc;
|
||
|
if (p != NULL) {
|
||
|
kn->kn_proc = NULL;
|
||
|
KNOTE_DETACH(&p->p_klist, kn);
|
||
|
}
|
||
|
proc_klist_unlock();
|
||
|
}
|
||
|
|
||
|
/*
|
||
|
* Post an event to the signal filter. Because we share the same list
|
||
|
* as process knotes, we have to filter out and handle only signal events.
|
||
|
*
|
||
|
* We assume that we process fdt_invalidate() before we post the NOTE_EXIT for
|
||
|
* a process during exit. Therefore, since signal filters can only be
|
||
|
* set up "in-process", we should have already torn down the kqueue
|
||
|
* hosting the EVFILT_SIGNAL knote and should never see NOTE_EXIT.
|
||
|
*/
|
||
|
static int
|
||
|
filt_signal(struct knote *kn, long hint)
|
||
|
{
|
||
|
if (hint & NOTE_SIGNAL) {
|
||
|
hint &= ~NOTE_SIGNAL;
|
||
|
|
||
|
if (kn->kn_id == (unsigned int)hint) {
|
||
|
kn->kn_hook32++;
|
||
|
}
|
||
|
} else if (hint & NOTE_EXIT) {
|
||
|
panic("filt_signal: detected NOTE_EXIT event");
|
||
|
}
|
||
|
|
||
|
return kn->kn_hook32 != 0;
|
||
|
}
|
||
|
|
||
|
static int
|
||
|
filt_signaltouch(struct knote *kn, struct kevent_qos_s *kev)
|
||
|
{
|
||
|
#pragma unused(kev)
|
||
|
|
||
|
int res;
|
||
|
|
||
|
proc_klist_lock();
|
||
|
|
||
|
/*
|
||
|
* No data to save - just capture if it is already fired
|
||
|
*/
|
||
|
res = (kn->kn_hook32 > 0);
|
||
|
|
||
|
proc_klist_unlock();
|
||
|
|
||
|
return res;
|
||
|
}
|
||
|
|
||
|
static int
|
||
|
filt_signalprocess(struct knote *kn, struct kevent_qos_s *kev)
|
||
|
{
|
||
|
int res = 0;
|
||
|
|
||
|
/*
|
||
|
* Snapshot the event data.
|
||
|
*/
|
||
|
|
||
|
proc_klist_lock();
|
||
|
if (kn->kn_hook32) {
|
||
|
knote_fill_kevent(kn, kev, kn->kn_hook32);
|
||
|
kn->kn_hook32 = 0;
|
||
|
res = 1;
|
||
|
}
|
||
|
proc_klist_unlock();
|
||
|
return res;
|
||
|
}
|
||
|
|
||
|
void
|
||
|
bsd_ast(thread_t thread)
|
||
|
{
|
||
|
proc_t p = current_proc();
|
||
|
struct uthread *ut = get_bsdthread_info(thread);
|
||
|
int signum;
|
||
|
static int bsd_init_done = 0;
|
||
|
|
||
|
if (p == NULL) {
|
||
|
return;
|
||
|
}
|
||
|
|
||
|
if (timerisset(&p->p_vtimer_user.it_value)) {
|
||
|
uint32_t microsecs;
|
||
|
|
||
|
task_vtimer_update(proc_task(p), TASK_VTIMER_USER, µsecs);
|
||
|
|
||
|
if (!itimerdecr(p, &p->p_vtimer_user, microsecs)) {
|
||
|
if (timerisset(&p->p_vtimer_user.it_value)) {
|
||
|
task_vtimer_set(proc_task(p), TASK_VTIMER_USER);
|
||
|
} else {
|
||
|
task_vtimer_clear(proc_task(p), TASK_VTIMER_USER);
|
||
|
}
|
||
|
|
||
|
psignal_try_thread(p, thread, SIGVTALRM);
|
||
|
}
|
||
|
}
|
||
|
|
||
|
if (timerisset(&p->p_vtimer_prof.it_value)) {
|
||
|
uint32_t microsecs;
|
||
|
|
||
|
task_vtimer_update(proc_task(p), TASK_VTIMER_PROF, µsecs);
|
||
|
|
||
|
if (!itimerdecr(p, &p->p_vtimer_prof, microsecs)) {
|
||
|
if (timerisset(&p->p_vtimer_prof.it_value)) {
|
||
|
task_vtimer_set(proc_task(p), TASK_VTIMER_PROF);
|
||
|
} else {
|
||
|
task_vtimer_clear(proc_task(p), TASK_VTIMER_PROF);
|
||
|
}
|
||
|
|
||
|
psignal_try_thread(p, thread, SIGPROF);
|
||
|
}
|
||
|
}
|
||
|
|
||
|
if (timerisset(&p->p_rlim_cpu)) {
|
||
|
struct timeval tv;
|
||
|
|
||
|
task_vtimer_update(proc_task(p), TASK_VTIMER_RLIM, (uint32_t *) &tv.tv_usec);
|
||
|
|
||
|
proc_spinlock(p);
|
||
|
if (p->p_rlim_cpu.tv_sec > 0 || p->p_rlim_cpu.tv_usec > tv.tv_usec) {
|
||
|
tv.tv_sec = 0;
|
||
|
timersub(&p->p_rlim_cpu, &tv, &p->p_rlim_cpu);
|
||
|
proc_spinunlock(p);
|
||
|
} else {
|
||
|
timerclear(&p->p_rlim_cpu);
|
||
|
proc_spinunlock(p);
|
||
|
|
||
|
task_vtimer_clear(proc_task(p), TASK_VTIMER_RLIM);
|
||
|
|
||
|
psignal_try_thread(p, thread, SIGXCPU);
|
||
|
}
|
||
|
}
|
||
|
|
||
|
#if CONFIG_DTRACE
|
||
|
if (ut->t_dtrace_sig) {
|
||
|
uint8_t dt_action_sig = ut->t_dtrace_sig;
|
||
|
ut->t_dtrace_sig = 0;
|
||
|
psignal(p, dt_action_sig);
|
||
|
}
|
||
|
|
||
|
if (ut->t_dtrace_stop) {
|
||
|
ut->t_dtrace_stop = 0;
|
||
|
proc_lock(p);
|
||
|
p->p_dtrace_stop = 1;
|
||
|
proc_unlock(p);
|
||
|
(void)task_suspend_internal(proc_task(p));
|
||
|
}
|
||
|
|
||
|
if (ut->t_dtrace_resumepid) {
|
||
|
proc_t resumeproc = proc_find((int)ut->t_dtrace_resumepid);
|
||
|
ut->t_dtrace_resumepid = 0;
|
||
|
if (resumeproc != PROC_NULL) {
|
||
|
proc_lock(resumeproc);
|
||
|
/* We only act on processes stopped by dtrace */
|
||
|
if (resumeproc->p_dtrace_stop) {
|
||
|
resumeproc->p_dtrace_stop = 0;
|
||
|
proc_unlock(resumeproc);
|
||
|
task_resume_internal(proc_task(resumeproc));
|
||
|
} else {
|
||
|
proc_unlock(resumeproc);
|
||
|
}
|
||
|
proc_rele(resumeproc);
|
||
|
}
|
||
|
}
|
||
|
|
||
|
#endif /* CONFIG_DTRACE */
|
||
|
|
||
|
proc_lock(p);
|
||
|
if (CHECK_SIGNALS(p, current_thread(), ut)) {
|
||
|
while ((signum = issignal_locked(p))) {
|
||
|
postsig_locked(signum);
|
||
|
}
|
||
|
}
|
||
|
proc_unlock(p);
|
||
|
|
||
|
if (!bsd_init_done) {
|
||
|
bsd_init_done = 1;
|
||
|
bsdinit_task();
|
||
|
}
|
||
|
}
|
||
|
|
||
|
/* ptrace set runnable */
|
||
|
void
|
||
|
pt_setrunnable(proc_t p)
|
||
|
{
|
||
|
task_t task;
|
||
|
|
||
|
task = proc_task(p);
|
||
|
|
||
|
if (p->p_lflag & P_LTRACED) {
|
||
|
proc_lock(p);
|
||
|
p->p_stat = SRUN;
|
||
|
proc_unlock(p);
|
||
|
if (p->sigwait) {
|
||
|
wakeup((caddr_t)&(p->sigwait));
|
||
|
if ((p->p_lflag & P_LSIGEXC) == 0) { // 5878479
|
||
|
task_release(task);
|
||
|
}
|
||
|
}
|
||
|
}
|
||
|
}
|
||
|
|
||
|
kern_return_t
|
||
|
do_bsdexception(
|
||
|
int exc,
|
||
|
int code,
|
||
|
int sub)
|
||
|
{
|
||
|
mach_exception_data_type_t codes[EXCEPTION_CODE_MAX];
|
||
|
|
||
|
codes[0] = code;
|
||
|
codes[1] = sub;
|
||
|
return bsd_exception(exc, codes, 2);
|
||
|
}
|
||
|
|
||
|
int
|
||
|
proc_pendingsignals(proc_t p, sigset_t mask)
|
||
|
{
|
||
|
struct uthread * uth;
|
||
|
sigset_t bits = 0;
|
||
|
|
||
|
proc_lock(p);
|
||
|
/* If the process is in proc exit return no signal info */
|
||
|
if (p->p_lflag & P_LPEXIT) {
|
||
|
goto out;
|
||
|
}
|
||
|
|
||
|
|
||
|
bits = 0;
|
||
|
TAILQ_FOREACH(uth, &p->p_uthlist, uu_list) {
|
||
|
bits |= (((uth->uu_siglist & ~uth->uu_sigmask) & ~p->p_sigignore) & mask);
|
||
|
}
|
||
|
out:
|
||
|
proc_unlock(p);
|
||
|
return bits;
|
||
|
}
|
||
|
|
||
|
int
|
||
|
thread_issignal(proc_t p, thread_t th, sigset_t mask)
|
||
|
{
|
||
|
struct uthread * uth;
|
||
|
sigset_t bits = 0;
|
||
|
|
||
|
proc_lock(p);
|
||
|
uth = (struct uthread *)get_bsdthread_info(th);
|
||
|
if (uth) {
|
||
|
bits = (((uth->uu_siglist & ~uth->uu_sigmask) & ~p->p_sigignore) & mask);
|
||
|
}
|
||
|
proc_unlock(p);
|
||
|
return bits;
|
||
|
}
|
||
|
|
||
|
/*
|
||
|
* Allow external reads of the sigprop array.
|
||
|
*/
|
||
|
int
|
||
|
hassigprop(int sig, int prop)
|
||
|
{
|
||
|
return sigprop[sig] & prop;
|
||
|
}
|
||
|
|
||
|
void
|
||
|
pgsigio(pid_t pgid, int sig)
|
||
|
{
|
||
|
proc_t p = PROC_NULL;
|
||
|
|
||
|
if (pgid < 0) {
|
||
|
gsignal(-(pgid), sig);
|
||
|
} else if (pgid > 0 && (p = proc_find(pgid)) != 0) {
|
||
|
psignal(p, sig);
|
||
|
}
|
||
|
if (p != PROC_NULL) {
|
||
|
proc_rele(p);
|
||
|
}
|
||
|
}
|
||
|
|
||
|
void
|
||
|
proc_signalstart(proc_t p, int locked)
|
||
|
{
|
||
|
if (!locked) {
|
||
|
proc_lock(p);
|
||
|
}
|
||
|
|
||
|
if (p->p_signalholder == current_thread()) {
|
||
|
panic("proc_signalstart: thread attempting to signal a process for which it holds the signal lock");
|
||
|
}
|
||
|
|
||
|
p->p_sigwaitcnt++;
|
||
|
while ((p->p_lflag & P_LINSIGNAL) == P_LINSIGNAL) {
|
||
|
msleep(&p->p_sigmask, &p->p_mlock, 0, "proc_signstart", NULL);
|
||
|
}
|
||
|
p->p_sigwaitcnt--;
|
||
|
|
||
|
p->p_lflag |= P_LINSIGNAL;
|
||
|
p->p_signalholder = current_thread();
|
||
|
if (!locked) {
|
||
|
proc_unlock(p);
|
||
|
}
|
||
|
}
|
||
|
|
||
|
void
|
||
|
proc_signalend(proc_t p, int locked)
|
||
|
{
|
||
|
if (!locked) {
|
||
|
proc_lock(p);
|
||
|
}
|
||
|
p->p_lflag &= ~P_LINSIGNAL;
|
||
|
|
||
|
if (p->p_sigwaitcnt > 0) {
|
||
|
wakeup(&p->p_sigmask);
|
||
|
}
|
||
|
|
||
|
p->p_signalholder = NULL;
|
||
|
if (!locked) {
|
||
|
proc_unlock(p);
|
||
|
}
|
||
|
}
|
||
|
|
||
|
void
|
||
|
sig_lock_to_exit(proc_t p)
|
||
|
{
|
||
|
thread_t self = current_thread();
|
||
|
|
||
|
p->exit_thread = self;
|
||
|
proc_unlock(p);
|
||
|
|
||
|
task_hold_and_wait(proc_task(p));
|
||
|
|
||
|
proc_lock(p);
|
||
|
}
|
||
|
|
||
|
int
|
||
|
sig_try_locked(proc_t p)
|
||
|
{
|
||
|
thread_t self = current_thread();
|
||
|
|
||
|
while (p->sigwait || p->exit_thread) {
|
||
|
if (p->exit_thread) {
|
||
|
return 0;
|
||
|
}
|
||
|
msleep((caddr_t)&p->sigwait_thread, &p->p_mlock, PCATCH | PDROP, 0, 0);
|
||
|
if (thread_should_abort(self)) {
|
||
|
/*
|
||
|
* Terminate request - clean up.
|
||
|
*/
|
||
|
proc_lock(p);
|
||
|
return -1;
|
||
|
}
|
||
|
proc_lock(p);
|
||
|
}
|
||
|
return 1;
|
||
|
}
|