#include <kern/task.h>
#include <kern/thread.h>
#include <kern/assert.h>
#include <kern/clock.h>
#include <kern/locks.h>
#include <kern/sched_prim.h>
#include <kern/debug.h>
#include <mach/machine/thread_status.h>
#include <mach/thread_act.h>
#include <sys/kernel.h>
#include <sys/vm.h>
#include <sys/proc_internal.h>
#include <sys/syscall.h>
#include <sys/systm.h>
#include <sys/user.h>
#include <sys/errno.h>
#include <sys/kdebug.h>
#include <sys/sysent.h>
#include <sys/sysproto.h>
#include <sys/kauth.h>
#include <sys/systm.h>
#include <security/audit/audit.h>
#include <i386/seg.h>
#include <i386/machine_routines.h>
#include <mach/i386/syscall_sw.h>
#if CONFIG_DTRACE
extern int32_t dtrace_systrace_syscall(struct proc *, void *, int *);
extern void dtrace_systrace_syscall_return(unsigned short, int, int *);
#endif
extern void unix_syscall(x86_saved_state_t *);
extern void unix_syscall64(x86_saved_state_t *);
extern void *find_user_regs(thread_t);
extern void x86_toggle_sysenter_arg_store(thread_t thread, boolean_t valid);
extern boolean_t x86_sysenter_arg_store_isvalid(thread_t thread);
extern const char *syscallnames[];
void
unix_syscall(x86_saved_state_t *state)
{
thread_t thread;
void *vt;
unsigned int code;
struct sysent *callp;
int error;
vm_offset_t params;
struct proc *p;
struct uthread *uthread;
x86_saved_state32_t *regs;
boolean_t args_in_uthread;
assert(is_saved_state32(state));
regs = saved_state32(state);
#if DEBUG
if (regs->eax == 0x800)
thread_exception_return();
#endif
thread = current_thread();
uthread = get_bsdthread_info(thread);
if (!(uthread->uu_flag & UT_VFORK))
p = (struct proc *)get_bsdtask_info(current_task());
else
p = current_proc();
if (p == NULL) {
regs->eax = EPERM;
regs->efl |= EFL_CF;
task_terminate_internal(current_task());
thread_exception_return();
}
code = regs->eax & I386_SYSCALL_NUMBER_MASK;
DEBUG_KPRINT_SYSCALL_UNIX("unix_syscall: code=%d(%s) eip=%u\n",
code, syscallnames[code >= NUM_SYSENT ? 63 : code], (uint32_t)regs->eip);
args_in_uthread = ((regs->eax & I386_SYSCALL_ARG_BYTES_MASK) != 0) && x86_sysenter_arg_store_isvalid(thread);
params = (vm_offset_t) (regs->uesp + sizeof (int));
regs->efl &= ~(EFL_CF);
callp = (code >= NUM_SYSENT) ? &sysent[63] : &sysent[code];
if (callp == sysent) {
code = fuword(params);
params += sizeof(int);
callp = (code >= NUM_SYSENT) ? &sysent[63] : &sysent[code];
}
vt = (void *)uthread->uu_arg;
if (callp->sy_arg_bytes != 0) {
sy_munge_t *mungerp;
assert((unsigned) callp->sy_arg_bytes <= sizeof (uthread->uu_arg));
if (!args_in_uthread)
{
uint32_t nargs;
nargs = callp->sy_arg_bytes;
error = copyin((user_addr_t) params, (char *) vt, nargs);
if (error) {
regs->eax = error;
regs->efl |= EFL_CF;
thread_exception_return();
}
}
if (code != 180) {
int *ip = (int *)vt;
KERNEL_DEBUG_CONSTANT(BSDDBG_CODE(DBG_BSD_EXCP_SC, code) | DBG_FUNC_START,
*ip, *(ip+1), *(ip+2), *(ip+3), 0);
}
mungerp = callp->sy_arg_munge32;
if (mungerp != NULL)
(*mungerp)(NULL, vt);
} else
KERNEL_DEBUG_CONSTANT(BSDDBG_CODE(DBG_BSD_EXCP_SC, code) | DBG_FUNC_START,
0, 0, 0, 0, 0);
kauth_cred_uthread_update(uthread, p);
uthread->uu_rval[0] = 0;
uthread->uu_rval[1] = regs->edx;
uthread->uu_flag |= UT_NOTCANCELPT;
#ifdef JOE_DEBUG
uthread->uu_iocount = 0;
uthread->uu_vpindex = 0;
#endif
AUDIT_SYSCALL_ENTER(code, p, uthread);
error = (*(callp->sy_call))((void *) p, (void *) vt, &(uthread->uu_rval[0]));
AUDIT_SYSCALL_EXIT(code, p, uthread, error);
#if CONFIG_MACF
mac_thread_userret(code, error, thread);
#endif
#ifdef JOE_DEBUG
if (uthread->uu_iocount)
printf("system call returned with uu_iocount != 0\n");
#endif
#if CONFIG_DTRACE
uthread->t_dtrace_errno = error;
#endif
if (error == ERESTART) {
if (regs->cs == SYSENTER_CS || regs->cs == SYSENTER_TF_CS) {
regs->eip -= 5;
}
else
regs->eip -= 2;
}
else if (error != EJUSTRETURN) {
if (error) {
regs->eax = error;
regs->efl |= EFL_CF;
} else {
regs->eax = uthread->uu_rval[0];
regs->edx = uthread->uu_rval[1];
}
}
DEBUG_KPRINT_SYSCALL_UNIX(
"unix_syscall: error=%d retval=(%u,%u)\n",
error, regs->eax, regs->edx);
uthread->uu_flag &= ~UT_NOTCANCELPT;
#if DEBUG
syscall_exit_funnelcheck();
#endif
if (uthread->uu_lowpri_window) {
throttle_lowpri_io(TRUE);
}
if (code != 180)
KERNEL_DEBUG_CONSTANT(BSDDBG_CODE(DBG_BSD_EXCP_SC, code) | DBG_FUNC_END,
error, uthread->uu_rval[0], uthread->uu_rval[1], p->p_pid, 0);
thread_exception_return();
}
void
unix_syscall64(x86_saved_state_t *state)
{
thread_t thread;
unsigned int code;
struct sysent *callp;
void *uargp;
int args_in_regs;
int error;
struct proc *p;
struct uthread *uthread;
x86_saved_state64_t *regs;
assert(is_saved_state64(state));
regs = saved_state64(state);
if (regs->rax == 0x2000800)
thread_exception_return();
thread = current_thread();
uthread = get_bsdthread_info(thread);
if (!(uthread->uu_flag & UT_VFORK))
p = (struct proc *)get_bsdtask_info(current_task());
else
p = current_proc();
if (p == NULL) {
regs->rax = EPERM;
regs->isf.rflags |= EFL_CF;
task_terminate_internal(current_task());
thread_exception_return();
}
args_in_regs = 6;
code = regs->rax & SYSCALL_NUMBER_MASK;
DEBUG_KPRINT_SYSCALL_UNIX(
"unix_syscall64: code=%d(%s) rip=%llx\n",
code, syscallnames[code >= NUM_SYSENT ? 63 : code], regs->isf.rip);
callp = (code >= NUM_SYSENT) ? &sysent[63] : &sysent[code];
uargp = (void *)(®s->rdi);
if (callp == sysent) {
code = regs->rdi;
callp = (code >= NUM_SYSENT) ? &sysent[63] : &sysent[code];
uargp = (void *)(®s->rsi);
args_in_regs = 5;
}
if (callp->sy_narg != 0) {
if (code != 180) {
uint64_t *ip = (uint64_t *)uargp;
KERNEL_DEBUG_CONSTANT(BSDDBG_CODE(DBG_BSD_EXCP_SC, code) | DBG_FUNC_START,
(int)(*ip), (int)(*(ip+1)), (int)(*(ip+2)), (int)(*(ip+3)), 0);
}
assert(callp->sy_narg <= 8);
if (callp->sy_narg > args_in_regs) {
int copyin_count;
copyin_count = (callp->sy_narg - args_in_regs) * sizeof(uint64_t);
error = copyin((user_addr_t)(regs->isf.rsp + sizeof(user_addr_t)), (char *)®s->v_arg6, copyin_count);
if (error) {
regs->rax = error;
regs->isf.rflags |= EFL_CF;
thread_exception_return();
}
}
if (callp->sy_flags & UNSAFE_64BIT) {
callp = &sysent[63];
goto unsafe;
}
} else
KERNEL_DEBUG_CONSTANT(BSDDBG_CODE(DBG_BSD_EXCP_SC, code) | DBG_FUNC_START,
0, 0, 0, 0, 0);
unsafe:
kauth_cred_uthread_update(uthread, p);
uthread->uu_rval[0] = 0;
uthread->uu_rval[1] = 0;
uthread->uu_flag |= UT_NOTCANCELPT;
AUDIT_SYSCALL_ENTER(code, p, uthread);
error = (*(callp->sy_call))((void *) p, uargp, &(uthread->uu_rval[0]));
AUDIT_SYSCALL_EXIT(code, p, uthread, error);
#if CONFIG_DTRACE
uthread->t_dtrace_errno = error;
#endif
if (error == ERESTART) {
regs->isf.rip -= 2;
}
else if (error != EJUSTRETURN) {
if (error) {
regs->rax = error;
regs->isf.rflags |= EFL_CF;
} else {
switch (callp->sy_return_type) {
case _SYSCALL_RET_INT_T:
regs->rax = uthread->uu_rval[0];
regs->rdx = uthread->uu_rval[1];
break;
case _SYSCALL_RET_UINT_T:
regs->rax = ((u_int)uthread->uu_rval[0]);
regs->rdx = ((u_int)uthread->uu_rval[1]);
break;
case _SYSCALL_RET_OFF_T:
case _SYSCALL_RET_ADDR_T:
case _SYSCALL_RET_SIZE_T:
case _SYSCALL_RET_SSIZE_T:
case _SYSCALL_RET_UINT64_T:
regs->rax = *((uint64_t *)(&uthread->uu_rval[0]));
regs->rdx = 0;
break;
case _SYSCALL_RET_NONE:
break;
default:
panic("unix_syscall: unknown return type");
break;
}
regs->isf.rflags &= ~EFL_CF;
}
}
DEBUG_KPRINT_SYSCALL_UNIX(
"unix_syscall64: error=%d retval=(%llu,%llu)\n",
error, regs->rax, regs->rdx);
uthread->uu_flag &= ~UT_NOTCANCELPT;
syscall_exit_funnelcheck();
if (uthread->uu_lowpri_window) {
throttle_lowpri_io(TRUE);
}
if (code != 180)
KERNEL_DEBUG_CONSTANT(BSDDBG_CODE(DBG_BSD_EXCP_SC, code) | DBG_FUNC_END,
error, uthread->uu_rval[0], uthread->uu_rval[1], p->p_pid, 0);
thread_exception_return();
}
void
unix_syscall_return(int error)
{
thread_t thread;
struct uthread *uthread;
struct proc *p;
unsigned int code;
vm_offset_t params;
struct sysent *callp;
thread = current_thread();
uthread = get_bsdthread_info(thread);
p = current_proc();
if (proc_is64bit(p)) {
x86_saved_state64_t *regs;
regs = saved_state64(find_user_regs(thread));
code = regs->rax & SYSCALL_NUMBER_MASK;
callp = (code >= NUM_SYSENT) ? &sysent[63] : &sysent[code];
if (callp == sysent)
code = regs->rdi;
#if CONFIG_DTRACE
if (callp->sy_call == dtrace_systrace_syscall)
dtrace_systrace_syscall_return( code, error, uthread->uu_rval );
#endif
AUDIT_SYSCALL_EXIT(code, p, uthread, error);
if (error == ERESTART) {
regs->isf.rip -= 2;
}
else if (error != EJUSTRETURN) {
if (error) {
regs->rax = error;
regs->isf.rflags |= EFL_CF;
} else {
switch (callp->sy_return_type) {
case _SYSCALL_RET_INT_T:
regs->rax = uthread->uu_rval[0];
regs->rdx = uthread->uu_rval[1];
break;
case _SYSCALL_RET_UINT_T:
regs->rax = ((u_int)uthread->uu_rval[0]);
regs->rdx = ((u_int)uthread->uu_rval[1]);
break;
case _SYSCALL_RET_OFF_T:
case _SYSCALL_RET_ADDR_T:
case _SYSCALL_RET_SIZE_T:
case _SYSCALL_RET_SSIZE_T:
case _SYSCALL_RET_UINT64_T:
regs->rax = *((uint64_t *)(&uthread->uu_rval[0]));
regs->rdx = 0;
break;
case _SYSCALL_RET_NONE:
break;
default:
panic("unix_syscall: unknown return type");
break;
}
regs->isf.rflags &= ~EFL_CF;
}
}
DEBUG_KPRINT_SYSCALL_UNIX(
"unix_syscall_return: error=%d retval=(%llu,%llu)\n",
error, regs->rax, regs->rdx);
} else {
x86_saved_state32_t *regs;
regs = saved_state32(find_user_regs(thread));
regs->efl &= ~(EFL_CF);
code = regs->eax & I386_SYSCALL_NUMBER_MASK;
callp = (code >= NUM_SYSENT) ? &sysent[63] : &sysent[code];
#if CONFIG_DTRACE
if (callp->sy_call == dtrace_systrace_syscall)
dtrace_systrace_syscall_return( code, error, uthread->uu_rval );
#endif
AUDIT_SYSCALL_EXIT(code, p, uthread, error);
if (callp == sysent) {
params = (vm_offset_t) (regs->uesp + sizeof (int));
code = fuword(params);
}
if (error == ERESTART) {
regs->eip -= ((regs->cs & 0xffff) == SYSENTER_CS) ? 5 : 2;
}
else if (error != EJUSTRETURN) {
if (error) {
regs->eax = error;
regs->efl |= EFL_CF;
} else {
regs->eax = uthread->uu_rval[0];
regs->edx = uthread->uu_rval[1];
}
}
DEBUG_KPRINT_SYSCALL_UNIX(
"unix_syscall_return: error=%d retval=(%u,%u)\n",
error, regs->eax, regs->edx);
}
uthread->uu_flag &= ~UT_NOTCANCELPT;
syscall_exit_funnelcheck();
if (uthread->uu_lowpri_window) {
throttle_lowpri_io(TRUE);
}
if (code != 180)
KERNEL_DEBUG_CONSTANT(BSDDBG_CODE(DBG_BSD_EXCP_SC, code) | DBG_FUNC_END,
error, uthread->uu_rval[0], uthread->uu_rval[1], p->p_pid, 0);
thread_exception_return();
}
void
munge_wwwlww(
__unused const void *in32,
void *out64)
{
uint32_t *arg32;
uint64_t *arg64;
arg32 = (uint32_t *) out64;
arg64 = (uint64_t *) out64;
arg64[5] = arg32[6];
arg64[4] = arg32[5];
arg32[7] = arg32[4];
arg32[6] = arg32[3];
arg64[2] = arg32[2];
arg64[1] = arg32[1];
arg64[0] = arg32[0];
}
void
munge_wwlwww(
__unused const void *in32,
void *out64)
{
uint32_t *arg32;
uint64_t *arg64;
arg32 = (uint32_t *) out64;
arg64 = (uint64_t *) out64;
arg64[5] = arg32[6];
arg64[4] = arg32[5];
arg64[3] = arg32[4];
arg32[5] = arg32[3];
arg32[4] = arg32[2];
arg64[1] = arg32[1];
arg64[0] = arg32[0];
}