[apple/xnu.git] / bsd / dev / i386 / systemcalls.c

/*
 * Copyright (c) 2000-2008 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@
 */
#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 <mach/branch_predicates.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>

#include <machine/pal_routines.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);

/* dynamically generated at build time based on syscalls.master */
extern const char *syscallnames[];

#define code_is_kdebug_trace(code) (((code) == SYS_kdebug_trace) ||   \
                                    ((code) == SYS_kdebug_trace64) || \
                                    ((code) == SYS_kdebug_trace_string))

/*
 * Function:	unix_syscall
 *
 * Inputs:	regs	- pointer to i386 save area
 *
 * Outputs:	none
 */
void
unix_syscall(x86_saved_state_t *state)
{
	thread_t		thread;
	void			*vt;
	unsigned int		code;
	struct sysent		*callp;

	int			error;
	vm_offset_t		params;
	struct proc		*p;
	struct uthread		*uthread;
	x86_saved_state32_t	*regs;
	boolean_t		is_vfork;

	assert(is_saved_state32(state));
	regs = saved_state32(state);
#if DEBUG
	if (regs->eax == 0x800)
		thread_exception_return();
#endif
	thread = current_thread();
	uthread = get_bsdthread_info(thread);

#if PROC_REF_DEBUG
	uthread_reset_proc_refcount(uthread);
#endif

	/* Get the approriate proc; may be different from task's for vfork() */
	is_vfork = uthread->uu_flag & UT_VFORK;
	if (__improbable(is_vfork != 0))
		p = current_proc();
	else 
		p = (struct proc *)get_bsdtask_info(current_task());

	/* Verify that we are not being called from a task without a proc */
	if (__improbable(p == NULL)) {
		regs->eax = EPERM;
		regs->efl |= EFL_CF;
		task_terminate_internal(current_task());
		thread_exception_return();
		/* NOTREACHED */
	}

	code = regs->eax & I386_SYSCALL_NUMBER_MASK;
	DEBUG_KPRINT_SYSCALL_UNIX("unix_syscall: code=%d(%s) eip=%u\n",
							  code, syscallnames[code >= NUM_SYSENT ? 63 : code], (uint32_t)regs->eip);
	params = (vm_offset_t) (regs->uesp + sizeof (int));

	regs->efl &= ~(EFL_CF);

	callp = (code >= NUM_SYSENT) ? &sysent[63] : &sysent[code];

	if (__improbable(callp == sysent)) {
		code = fuword(params);
		params += sizeof(int);
		callp = (code >= NUM_SYSENT) ? &sysent[63] : &sysent[code];
	}

	vt = (void *)uthread->uu_arg;

	if (callp->sy_arg_bytes != 0) {
#if CONFIG_REQUIRES_U32_MUNGING
		sy_munge_t	*mungerp;
#else
#error U32 syscalls on x86_64 kernel requires munging
#endif
		uint32_t	 nargs;

		assert((unsigned) callp->sy_arg_bytes <= sizeof (uthread->uu_arg));
		nargs = callp->sy_arg_bytes;
		error = copyin((user_addr_t) params, (char *) vt, nargs);
		if (error) {
			regs->eax = error;
			regs->efl |= EFL_CF;
			thread_exception_return();
			/* NOTREACHED */
		}

		if (__probable(!code_is_kdebug_trace(code))) {
			int *ip = (int *)vt;

			KERNEL_DEBUG_CONSTANT_IST(KDEBUG_TRACE,
				BSDDBG_CODE(DBG_BSD_EXCP_SC, code) | DBG_FUNC_START,
				*ip, *(ip+1), *(ip+2), *(ip+3), 0);
		}

#if CONFIG_REQUIRES_U32_MUNGING
		mungerp = callp->sy_arg_munge32;

		if (mungerp != NULL)
			(*mungerp)(vt);
#endif
	} else
		KERNEL_DEBUG_CONSTANT_IST(KDEBUG_TRACE, 
			BSDDBG_CODE(DBG_BSD_EXCP_SC, code) | DBG_FUNC_START,
			0, 0, 0, 0, 0);

	/*
	 * Delayed binding of thread credential to process credential, if we
	 * are not running with an explicitly set thread credential.
	 */
	kauth_cred_uthread_update(uthread, p);

	uthread->uu_rval[0] = 0;
	uthread->uu_rval[1] = 0;
	uthread->uu_flag |= UT_NOTCANCELPT;
	uthread->syscall_code = code;

#ifdef JOE_DEBUG
        uthread->uu_iocount = 0;
        uthread->uu_vpindex = 0;
#endif

	AUDIT_SYSCALL_ENTER(code, p, uthread);
	error = (*(callp->sy_call))((void *) p, (void *) vt, &(uthread->uu_rval[0]));
	AUDIT_SYSCALL_EXIT(code, p, uthread, error);

#ifdef JOE_DEBUG
        if (uthread->uu_iocount)
                printf("system call returned with uu_iocount != 0\n");
#endif
#if CONFIG_DTRACE
	uthread->t_dtrace_errno = error;
#endif /* CONFIG_DTRACE */

	if (__improbable(error == ERESTART)) {
		/*
		 * Move the user's pc back to repeat the syscall:
		 * 5 bytes for a sysenter, or 2 for an int 8x.
		 * The SYSENTER_TF_CS covers single-stepping over a sysenter
		 * - see debug trap handler in idt.s/idt64.s
		 */

		pal_syscall_restart(thread, state);
	}
	else if (error != EJUSTRETURN) {
		if (__improbable(error)) {
		    regs->eax = error;
		    regs->efl |= EFL_CF;	/* carry bit */
		} else { /* (not error) */
			/*
			 * We split retval across two registers, in case the
			 * syscall had a 64-bit return value, in which case
			 * eax/edx matches the function call ABI.
			 */
		    regs->eax = uthread->uu_rval[0];
		    regs->edx = uthread->uu_rval[1];
		} 
	}

	DEBUG_KPRINT_SYSCALL_UNIX(
		"unix_syscall: error=%d retval=(%u,%u)\n",
		error, regs->eax, regs->edx);

	uthread->uu_flag &= ~UT_NOTCANCELPT;

	if (__improbable(uthread->uu_lowpri_window)) {
	        /*
		 * task is marked as a low priority I/O type
		 * and the I/O we issued while in this system call
		 * collided with normal I/O operations... we'll
		 * delay in order to mitigate the impact of this
		 * task on the normal operation of the system
		 */
		throttle_lowpri_io(1);
	}
	if (__probable(!code_is_kdebug_trace(code)))
		KERNEL_DEBUG_CONSTANT_IST(KDEBUG_TRACE, 
			BSDDBG_CODE(DBG_BSD_EXCP_SC, code) | DBG_FUNC_END,
			error, uthread->uu_rval[0], uthread->uu_rval[1], p->p_pid, 0);

	if (__improbable(!is_vfork && callp->sy_call == (sy_call_t *)execve && !error)) {
		pal_execve_return(thread);
	}

#if PROC_REF_DEBUG
	if (__improbable(uthread_get_proc_refcount(uthread) != 0)) {
		panic("system call returned with uu_proc_refcount != 0");
	}
#endif

	thread_exception_return();
	/* NOTREACHED */
}


void
unix_syscall64(x86_saved_state_t *state)
{
	thread_t	thread;
	void			*vt;
	unsigned int	code;
	struct sysent	*callp;
	int		args_in_regs;
	boolean_t	args_start_at_rdi;
	int		error;
	struct proc	*p;
	struct uthread	*uthread;
	x86_saved_state64_t *regs;

	assert(is_saved_state64(state));
	regs = saved_state64(state);
#if	DEBUG
	if (regs->rax == 0x2000800)
		thread_exception_return();
#endif
	thread = current_thread();
	uthread = get_bsdthread_info(thread);

#if PROC_REF_DEBUG
	uthread_reset_proc_refcount(uthread);
#endif

	/* Get the approriate proc; may be different from task's for vfork() */
	if (__probable(!(uthread->uu_flag & UT_VFORK)))
		p = (struct proc *)get_bsdtask_info(current_task());
	else 
		p = current_proc();

	/* Verify that we are not being called from a task without a proc */
	if (__improbable(p == NULL)) {
		regs->rax = EPERM;
		regs->isf.rflags |= EFL_CF;
		task_terminate_internal(current_task());
		thread_exception_return();
		/* NOTREACHED */
	}

	code = regs->rax & SYSCALL_NUMBER_MASK;
	DEBUG_KPRINT_SYSCALL_UNIX(
		"unix_syscall64: code=%d(%s) rip=%llx\n",
		code, syscallnames[code >= NUM_SYSENT ? 63 : code], regs->isf.rip);
	callp = (code >= NUM_SYSENT) ? &sysent[63] : &sysent[code];

	vt = (void *)uthread->uu_arg;

	if (__improbable(callp == sysent)) {
	        /*
		 * indirect system call... system call number
		 * passed as 'arg0'
		 */
		code = regs->rdi;
		callp = (code >= NUM_SYSENT) ? &sysent[63] : &sysent[code];
		args_start_at_rdi = FALSE;
		args_in_regs = 5;
	} else {
		args_start_at_rdi = TRUE;
		args_in_regs = 6;
	}

	if (callp->sy_narg != 0) {
		assert(callp->sy_narg <= 8); /* size of uu_arg */

		args_in_regs = MIN(args_in_regs, callp->sy_narg);
		memcpy(vt, args_start_at_rdi ? &regs->rdi : &regs->rsi, args_in_regs * sizeof(syscall_arg_t));


		if (!code_is_kdebug_trace(code)) {
			uint64_t *ip = (uint64_t *)vt;

			KERNEL_DEBUG_CONSTANT_IST(KDEBUG_TRACE, 
				BSDDBG_CODE(DBG_BSD_EXCP_SC, code) | DBG_FUNC_START,
				(int)(*ip), (int)(*(ip+1)), (int)(*(ip+2)), (int)(*(ip+3)), 0);
		}

		if (__improbable(callp->sy_narg > args_in_regs)) {
			int copyin_count;

			copyin_count = (callp->sy_narg - args_in_regs) * sizeof(syscall_arg_t);

			error = copyin((user_addr_t)(regs->isf.rsp + sizeof(user_addr_t)), (char *)&uthread->uu_arg[args_in_regs], copyin_count);
			if (error) {
				regs->rax = error;
				regs->isf.rflags |= EFL_CF;
				thread_exception_return();
				/* NOTREACHED */
			}
		}
	} else
		KERNEL_DEBUG_CONSTANT_IST(KDEBUG_TRACE,
			BSDDBG_CODE(DBG_BSD_EXCP_SC, code) | DBG_FUNC_START,
			0, 0, 0, 0, 0);

	/*
	 * Delayed binding of thread credential to process credential, if we
	 * are not running with an explicitly set thread credential.
	 */
	kauth_cred_uthread_update(uthread, p);

	uthread->uu_rval[0] = 0;
	uthread->uu_rval[1] = 0;
	uthread->uu_flag |= UT_NOTCANCELPT;
	uthread->syscall_code = code;

#ifdef JOE_DEBUG
        uthread->uu_iocount = 0;
        uthread->uu_vpindex = 0;
#endif

	AUDIT_SYSCALL_ENTER(code, p, uthread);
	error = (*(callp->sy_call))((void *) p, vt, &(uthread->uu_rval[0]));
	AUDIT_SYSCALL_EXIT(code, p, uthread, error);

#ifdef JOE_DEBUG
        if (uthread->uu_iocount)
               printf("system call returned with uu_iocount != 0\n");
#endif

#if CONFIG_DTRACE
	uthread->t_dtrace_errno = error;
#endif /* CONFIG_DTRACE */
	
	if (__improbable(error == ERESTART)) {
		/*
		 * all system calls come through via the syscall instruction
		 * in 64 bit mode... its 2 bytes in length
		 * move the user's pc back to repeat the syscall:
		 */
		pal_syscall_restart( thread, state );
	}
	else if (error != EJUSTRETURN) {
		if (__improbable(error)) {
			regs->rax = error;
			regs->isf.rflags |= EFL_CF;	/* carry bit */
		} else { /* (not error) */

			switch (callp->sy_return_type) {
			case _SYSCALL_RET_INT_T:
				regs->rax = uthread->uu_rval[0];
				regs->rdx = uthread->uu_rval[1];
				break;
			case _SYSCALL_RET_UINT_T:
				regs->rax = ((u_int)uthread->uu_rval[0]);
				regs->rdx = ((u_int)uthread->uu_rval[1]);
				break;
			case _SYSCALL_RET_OFF_T:
			case _SYSCALL_RET_ADDR_T:
			case _SYSCALL_RET_SIZE_T:
			case _SYSCALL_RET_SSIZE_T:
			case _SYSCALL_RET_UINT64_T:
			        regs->rax = *((uint64_t *)(&uthread->uu_rval[0]));
				regs->rdx = 0;
				break;
			case _SYSCALL_RET_NONE:
				break;
			default:
				panic("unix_syscall: unknown return type");
				break;
			}
			regs->isf.rflags &= ~EFL_CF;
		} 
	}

	DEBUG_KPRINT_SYSCALL_UNIX(
		"unix_syscall64: error=%d retval=(%llu,%llu)\n",
		error, regs->rax, regs->rdx);
	
	uthread->uu_flag &= ~UT_NOTCANCELPT;

	if (__improbable(uthread->uu_lowpri_window)) {
	        /*
		 * task is marked as a low priority I/O type
		 * and the I/O we issued while in this system call
		 * collided with normal I/O operations... we'll
		 * delay in order to mitigate the impact of this
		 * task on the normal operation of the system
		 */
		throttle_lowpri_io(1);
	}
	if (__probable(!code_is_kdebug_trace(code)))
		KERNEL_DEBUG_CONSTANT_IST(KDEBUG_TRACE, 
			BSDDBG_CODE(DBG_BSD_EXCP_SC, code) | DBG_FUNC_END,
			error, uthread->uu_rval[0], uthread->uu_rval[1], p->p_pid, 0);

#if PROC_REF_DEBUG
	if (__improbable(uthread_get_proc_refcount(uthread))) {
		panic("system call returned with uu_proc_refcount != 0");
	}
#endif

	thread_exception_return();
	/* NOTREACHED */
}


void
unix_syscall_return(int error)
{
	thread_t		thread;
	struct uthread		*uthread;
	struct proc *p;
	unsigned int code;
	struct sysent *callp;

	thread = current_thread();
	uthread = get_bsdthread_info(thread);

	pal_register_cache_state(thread, DIRTY);

	p = current_proc();

	if (proc_is64bit(p)) {
		x86_saved_state64_t *regs;

		regs = saved_state64(find_user_regs(thread));

		code = uthread->syscall_code;
		callp = (code >= NUM_SYSENT) ? &sysent[63] : &sysent[code];

#if CONFIG_DTRACE
		if (callp->sy_call == dtrace_systrace_syscall)
			dtrace_systrace_syscall_return( code, error, uthread->uu_rval );
#endif /* CONFIG_DTRACE */
		AUDIT_SYSCALL_EXIT(code, p, uthread, error);

		if (error == ERESTART) {
			/*
			 * repeat the syscall
			 */
			pal_syscall_restart( thread, find_user_regs(thread) );
		}
		else if (error != EJUSTRETURN) {
			if (error) {
				regs->rax = error;
				regs->isf.rflags |= EFL_CF;	/* carry bit */
			} else { /* (not error) */

				switch (callp->sy_return_type) {
				case _SYSCALL_RET_INT_T:
					regs->rax = uthread->uu_rval[0];
					regs->rdx = uthread->uu_rval[1];
					break;
				case _SYSCALL_RET_UINT_T:
					regs->rax = ((u_int)uthread->uu_rval[0]);
					regs->rdx = ((u_int)uthread->uu_rval[1]);
					break;
				case _SYSCALL_RET_OFF_T:
				case _SYSCALL_RET_ADDR_T:
				case _SYSCALL_RET_SIZE_T:
				case _SYSCALL_RET_SSIZE_T:
				case _SYSCALL_RET_UINT64_T:
					regs->rax = *((uint64_t *)(&uthread->uu_rval[0]));
					regs->rdx = 0;
					break;
				case _SYSCALL_RET_NONE:
					break;
				default:
					panic("unix_syscall: unknown return type");
					break;
				}
				regs->isf.rflags &= ~EFL_CF;
			} 
		}
		DEBUG_KPRINT_SYSCALL_UNIX(
			"unix_syscall_return: error=%d retval=(%llu,%llu)\n",
			error, regs->rax, regs->rdx);
	} else {
		x86_saved_state32_t	*regs;

		regs = saved_state32(find_user_regs(thread));

		regs->efl &= ~(EFL_CF);

		code = uthread->syscall_code;
		callp = (code >= NUM_SYSENT) ? &sysent[63] : &sysent[code];

#if CONFIG_DTRACE
		if (callp->sy_call == dtrace_systrace_syscall)
			dtrace_systrace_syscall_return( code, error, uthread->uu_rval );
#endif /* CONFIG_DTRACE */
		AUDIT_SYSCALL_EXIT(code, p, uthread, error);

		if (error == ERESTART) {
			pal_syscall_restart( thread, find_user_regs(thread) );
		}
		else if (error != EJUSTRETURN) {
			if (error) {
				regs->eax = error;
				regs->efl |= EFL_CF;	/* carry bit */
			} else { /* (not error) */
				regs->eax = uthread->uu_rval[0];
				regs->edx = uthread->uu_rval[1];
			} 
		}
		DEBUG_KPRINT_SYSCALL_UNIX(
			"unix_syscall_return: error=%d retval=(%u,%u)\n",
			error, regs->eax, regs->edx);
	}


	uthread->uu_flag &= ~UT_NOTCANCELPT;

	if (uthread->uu_lowpri_window) {
	        /*
		 * task is marked as a low priority I/O type
		 * and the I/O we issued while in this system call
		 * collided with normal I/O operations... we'll
		 * delay in order to mitigate the impact of this
		 * task on the normal operation of the system
		 */
		throttle_lowpri_io(1);
	}
	if (!code_is_kdebug_trace(code))
		KERNEL_DEBUG_CONSTANT_IST(KDEBUG_TRACE, 
			BSDDBG_CODE(DBG_BSD_EXCP_SC, code) | DBG_FUNC_END,
			error, uthread->uu_rval[0], uthread->uu_rval[1], p->p_pid, 0);

	thread_exception_return();
	/* NOTREACHED */
}
Commit	Line	Data
0c530ab8	1	/*
b0d623f7	2	* Copyright (c) 2000-2008 Apple Inc. All rights reserved.
0c530ab8	3	*
2d21ac55	4	* @APPLE_OSREFERENCE_LICENSE_HEADER_START@
0c530ab8	5	*
2d21ac55 A	6	* This file contains Original Code and/or Modifications of Original Code
	7	* as defined in and that are subject to the Apple Public Source License
	8	* Version 2.0 (the 'License'). You may not use this file except in
	9	* compliance with the License. The rights granted to you under the License
	10	* may not be used to create, or enable the creation or redistribution of,
	11	* unlawful or unlicensed copies of an Apple operating system, or to
	12	* circumvent, violate, or enable the circumvention or violation of, any
	13	* terms of an Apple operating system software license agreement.
0c530ab8	14	*
2d21ac55 A	15	* Please obtain a copy of the License at
	16	* http://www.opensource.apple.com/apsl/ and read it before using this file.
	17	*
	18	* The Original Code and all software distributed under the License are
	19	* distributed on an 'AS IS' basis, WITHOUT WARRANTY OF ANY KIND, EITHER
0c530ab8 A	20	* EXPRESS OR IMPLIED, AND APPLE HEREBY DISCLAIMS ALL SUCH WARRANTIES,
0c530ab8 A	21	* INCLUDING WITHOUT LIMITATION, ANY WARRANTIES OF MERCHANTABILITY,
2d21ac55 A	22	* FITNESS FOR A PARTICULAR PURPOSE, QUIET ENJOYMENT OR NON-INFRINGEMENT.
	23	* Please see the License for the specific language governing rights and
	24	* limitations under the License.
0c530ab8	25	*
2d21ac55	26	* @APPLE_OSREFERENCE_LICENSE_HEADER_END@
0c530ab8 A	27	*/
	28	#include <kern/task.h>
	29	#include <kern/thread.h>
	30	#include <kern/assert.h>
	31	#include <kern/clock.h>
	32	#include <kern/locks.h>
	33	#include <kern/sched_prim.h>
b0d623f7	34	#include <kern/debug.h>
0c530ab8 A	35	#include <mach/machine/thread_status.h>
0c530ab8 A	36	#include <mach/thread_act.h>
6d2010ae	37	#include <mach/branch_predicates.h>
0c530ab8 A	38
	39	#include <sys/kernel.h>
	40	#include <sys/vm.h>
	41	#include <sys/proc_internal.h>
	42	#include <sys/syscall.h>
	43	#include <sys/systm.h>
	44	#include <sys/user.h>
	45	#include <sys/errno.h>
0c530ab8 A	46	#include <sys/kdebug.h>
	47	#include <sys/sysent.h>
	48	#include <sys/sysproto.h>
	49	#include <sys/kauth.h>
	50	#include <sys/systm.h>
	51
b0d623f7	52	#include <security/audit/audit.h>
0c530ab8 A	53
	54	#include <i386/seg.h>
	55	#include <i386/machine_routines.h>
	56	#include <mach/i386/syscall_sw.h>
	57
6d2010ae A	58	#include <machine/pal_routines.h>
6d2010ae A	59
2d21ac55 A	60	#if CONFIG_DTRACE
	61	extern int32_t dtrace_systrace_syscall(struct proc , void , int *);
	62	extern void dtrace_systrace_syscall_return(unsigned short, int, int *);
	63	#endif
	64
0c530ab8 A	65	extern void unix_syscall(x86_saved_state_t *);
0c530ab8 A	66	extern void unix_syscall64(x86_saved_state_t *);
0c530ab8	67	extern void *find_user_regs(thread_t);
0c530ab8	68
b0d623f7 A	69	/* dynamically generated at build time based on syscalls.master */
	70	extern const char *syscallnames[];
	71
3e170ce0 A	72	#define code_is_kdebug_trace(code) (((code) == SYS_kdebug_trace) \|\| \
	73	((code) == SYS_kdebug_trace64) \|\| \
	74	((code) == SYS_kdebug_trace_string))
a1c7dba1	75
0c530ab8 A	76	/*
	77	* Function: unix_syscall
	78	*
	79	* Inputs: regs - pointer to i386 save area
	80	*
	81	* Outputs: none
	82	*/
	83	void
	84	unix_syscall(x86_saved_state_t *state)
	85	{
2d21ac55 A	86	thread_t thread;
	87	void *vt;
	88	unsigned int code;
	89	struct sysent *callp;
	90
	91	int error;
	92	vm_offset_t params;
	93	struct proc *p;
	94	struct uthread *uthread;
0c530ab8	95	x86_saved_state32_t *regs;
6d2010ae	96	boolean_t is_vfork;
0c530ab8 A	97
	98	assert(is_saved_state32(state));
	99	regs = saved_state32(state);
2d21ac55	100	#if DEBUG
0c530ab8 A	101	if (regs->eax == 0x800)
0c530ab8 A	102	thread_exception_return();
2d21ac55	103	#endif
0c530ab8 A	104	thread = current_thread();
	105	uthread = get_bsdthread_info(thread);
	106
3e170ce0 A	107	#if PROC_REF_DEBUG
	108	uthread_reset_proc_refcount(uthread);
	109	#endif
	110
0c530ab8	111	/* Get the approriate proc; may be different from task's for vfork() */
6d2010ae A	112	is_vfork = uthread->uu_flag & UT_VFORK;
6d2010ae A	113	if (__improbable(is_vfork != 0))
0c530ab8	114	p = current_proc();
6d2010ae A	115	else
6d2010ae A	116	p = (struct proc *)get_bsdtask_info(current_task());
0c530ab8 A	117
0c530ab8 A	118	/* Verify that we are not being called from a task without a proc */
6d2010ae	119	if (__improbable(p == NULL)) {
0c530ab8 A	120	regs->eax = EPERM;
	121	regs->efl \|= EFL_CF;
	122	task_terminate_internal(current_task());
	123	thread_exception_return();
	124	/* NOTREACHED */
	125	}
	126
2d21ac55	127	code = regs->eax & I386_SYSCALL_NUMBER_MASK;
b0d623f7 A	128	DEBUG_KPRINT_SYSCALL_UNIX("unix_syscall: code=%d(%s) eip=%u\n",
b0d623f7 A	129	code, syscallnames[code >= NUM_SYSENT ? 63 : code], (uint32_t)regs->eip);
b0d623f7	130	params = (vm_offset_t) (regs->uesp + sizeof (int));
2d21ac55 A	131
	132	regs->efl &= ~(EFL_CF);
	133
	134	callp = (code >= NUM_SYSENT) ? &sysent[63] : &sysent[code];
0c530ab8	135
6d2010ae	136	if (__improbable(callp == sysent)) {
0c530ab8	137	code = fuword(params);
2d21ac55 A	138	params += sizeof(int);
2d21ac55 A	139	callp = (code >= NUM_SYSENT) ? &sysent[63] : &sysent[code];
0c530ab8	140	}
2d21ac55	141
0c530ab8 A	142	vt = (void *)uthread->uu_arg;
0c530ab8 A	143
2d21ac55	144	if (callp->sy_arg_bytes != 0) {
fe8ab488	145	#if CONFIG_REQUIRES_U32_MUNGING
0c530ab8	146	sy_munge_t *mungerp;
fe8ab488 A	147	#else
	148	#error U32 syscalls on x86_64 kernel requires munging
	149	#endif
39236c6e	150	uint32_t nargs;
0c530ab8	151
2d21ac55	152	assert((unsigned) callp->sy_arg_bytes <= sizeof (uthread->uu_arg));
39236c6e A	153	nargs = callp->sy_arg_bytes;
	154	error = copyin((user_addr_t) params, (char *) vt, nargs);
	155	if (error) {
	156	regs->eax = error;
	157	regs->efl \|= EFL_CF;
	158	thread_exception_return();
	159	/* NOTREACHED */
0c530ab8	160	}
2d21ac55	161
a1c7dba1 A	162	if (__probable(!code_is_kdebug_trace(code))) {
a1c7dba1 A	163	int ip = (int )vt;
0c530ab8	164
316670eb A	165	KERNEL_DEBUG_CONSTANT_IST(KDEBUG_TRACE,
	166	BSDDBG_CODE(DBG_BSD_EXCP_SC, code) \| DBG_FUNC_START,
	167	ip, (ip+1), (ip+2), (ip+3), 0);
0c530ab8	168	}
fe8ab488 A	169
fe8ab488 A	170	#if CONFIG_REQUIRES_U32_MUNGING
0c530ab8 A	171	mungerp = callp->sy_arg_munge32;
0c530ab8 A	172
0c530ab8	173	if (mungerp != NULL)
fe8ab488 A	174	(*mungerp)(vt);
fe8ab488 A	175	#endif
0c530ab8	176	} else
316670eb A	177	KERNEL_DEBUG_CONSTANT_IST(KDEBUG_TRACE,
316670eb A	178	BSDDBG_CODE(DBG_BSD_EXCP_SC, code) \| DBG_FUNC_START,
2d21ac55 A	179	0, 0, 0, 0, 0);
2d21ac55 A	180
0c530ab8 A	181	/*
	182	* Delayed binding of thread credential to process credential, if we
	183	* are not running with an explicitly set thread credential.
	184	*/
2d21ac55	185	kauth_cred_uthread_update(uthread, p);
0c530ab8 A	186
0c530ab8 A	187	uthread->uu_rval[0] = 0;
fe8ab488	188	uthread->uu_rval[1] = 0;
2d21ac55	189	uthread->uu_flag \|= UT_NOTCANCELPT;
fe8ab488	190	uthread->syscall_code = code;
0c530ab8	191
2d21ac55 A	192	#ifdef JOE_DEBUG
	193	uthread->uu_iocount = 0;
	194	uthread->uu_vpindex = 0;
	195	#endif
0c530ab8 A	196
	197	AUDIT_SYSCALL_ENTER(code, p, uthread);
	198	error = ((callp->sy_call))((void ) p, (void *) vt, &(uthread->uu_rval[0]));
fe8ab488	199	AUDIT_SYSCALL_EXIT(code, p, uthread, error);
2d21ac55 A	200
	201	#ifdef JOE_DEBUG
	202	if (uthread->uu_iocount)
b0d623f7	203	printf("system call returned with uu_iocount != 0\n");
2d21ac55 A	204	#endif
	205	#if CONFIG_DTRACE
	206	uthread->t_dtrace_errno = error;
	207	#endif /* CONFIG_DTRACE */
	208
6d2010ae	209	if (__improbable(error == ERESTART)) {
0c530ab8 A	210	/*
	211	* Move the user's pc back to repeat the syscall:
	212	* 5 bytes for a sysenter, or 2 for an int 8x.
	213	* The SYSENTER_TF_CS covers single-stepping over a sysenter
	214	* - see debug trap handler in idt.s/idt64.s
	215	*/
b0d623f7	216
6d2010ae	217	pal_syscall_restart(thread, state);
0c530ab8 A	218	}
0c530ab8 A	219	else if (error != EJUSTRETURN) {
6d2010ae	220	if (__improbable(error)) {
0c530ab8 A	221	regs->eax = error;
	222	regs->efl \|= EFL_CF; /* carry bit */
	223	} else { /* (not error) */
fe8ab488 A	224	/*
	225	* We split retval across two registers, in case the
	226	* syscall had a 64-bit return value, in which case
	227	* eax/edx matches the function call ABI.
	228	*/
0c530ab8 A	229	regs->eax = uthread->uu_rval[0];
0c530ab8 A	230	regs->edx = uthread->uu_rval[1];
0c530ab8 A	231	}
	232	}
	233
b0d623f7 A	234	DEBUG_KPRINT_SYSCALL_UNIX(
	235	"unix_syscall: error=%d retval=(%u,%u)\n",
	236	error, regs->eax, regs->edx);
	237
2d21ac55	238	uthread->uu_flag &= ~UT_NOTCANCELPT;
6d2010ae A	239
6d2010ae A	240	if (__improbable(uthread->uu_lowpri_window)) {
0c530ab8 A	241	/*
	242	* task is marked as a low priority I/O type
	243	* and the I/O we issued while in this system call
	244	* collided with normal I/O operations... we'll
	245	* delay in order to mitigate the impact of this
	246	* task on the normal operation of the system
	247	*/
39236c6e	248	throttle_lowpri_io(1);
0c530ab8	249	}
a1c7dba1	250	if (__probable(!code_is_kdebug_trace(code)))
316670eb A	251	KERNEL_DEBUG_CONSTANT_IST(KDEBUG_TRACE,
	252	BSDDBG_CODE(DBG_BSD_EXCP_SC, code) \| DBG_FUNC_END,
	253	error, uthread->uu_rval[0], uthread->uu_rval[1], p->p_pid, 0);
b0d623f7	254
6d2010ae A	255	if (__improbable(!is_vfork && callp->sy_call == (sy_call_t *)execve && !error)) {
	256	pal_execve_return(thread);
	257	}
0c530ab8	258
3e170ce0 A	259	#if PROC_REF_DEBUG
	260	if (__improbable(uthread_get_proc_refcount(uthread) != 0)) {
	261	panic("system call returned with uu_proc_refcount != 0");
	262	}
	263	#endif
	264
0c530ab8 A	265	thread_exception_return();
	266	/* NOTREACHED */
	267	}
	268
	269
	270	void
	271	unix_syscall64(x86_saved_state_t *state)
	272	{
	273	thread_t thread;
fe8ab488	274	void *vt;
0c530ab8 A	275	unsigned int code;
0c530ab8 A	276	struct sysent *callp;
0c530ab8	277	int args_in_regs;
fe8ab488	278	boolean_t args_start_at_rdi;
0c530ab8	279	int error;
0c530ab8 A	280	struct proc *p;
0c530ab8 A	281	struct uthread *uthread;
0c530ab8 A	282	x86_saved_state64_t *regs;
	283
	284	assert(is_saved_state64(state));
	285	regs = saved_state64(state);
6d2010ae	286	#if DEBUG
0c530ab8 A	287	if (regs->rax == 0x2000800)
0c530ab8 A	288	thread_exception_return();
6d2010ae	289	#endif
0c530ab8 A	290	thread = current_thread();
	291	uthread = get_bsdthread_info(thread);
	292
3e170ce0 A	293	#if PROC_REF_DEBUG
	294	uthread_reset_proc_refcount(uthread);
	295	#endif
	296
0c530ab8	297	/* Get the approriate proc; may be different from task's for vfork() */
6d2010ae	298	if (__probable(!(uthread->uu_flag & UT_VFORK)))
0c530ab8 A	299	p = (struct proc *)get_bsdtask_info(current_task());
	300	else
	301	p = current_proc();
	302
	303	/* Verify that we are not being called from a task without a proc */
6d2010ae	304	if (__improbable(p == NULL)) {
0c530ab8 A	305	regs->rax = EPERM;
	306	regs->isf.rflags \|= EFL_CF;
	307	task_terminate_internal(current_task());
	308	thread_exception_return();
	309	/* NOTREACHED */
	310	}
0c530ab8 A	311
0c530ab8 A	312	code = regs->rax & SYSCALL_NUMBER_MASK;
b0d623f7 A	313	DEBUG_KPRINT_SYSCALL_UNIX(
	314	"unix_syscall64: code=%d(%s) rip=%llx\n",
	315	code, syscallnames[code >= NUM_SYSENT ? 63 : code], regs->isf.rip);
2d21ac55	316	callp = (code >= NUM_SYSENT) ? &sysent[63] : &sysent[code];
fe8ab488 A	317
fe8ab488 A	318	vt = (void *)uthread->uu_arg;
0c530ab8	319
6d2010ae	320	if (__improbable(callp == sysent)) {
0c530ab8 A	321	/*
	322	* indirect system call... system call number
	323	* passed as 'arg0'
	324	*/
fe8ab488	325	code = regs->rdi;
2d21ac55	326	callp = (code >= NUM_SYSENT) ? &sysent[63] : &sysent[code];
fe8ab488	327	args_start_at_rdi = FALSE;
0c530ab8	328	args_in_regs = 5;
fe8ab488 A	329	} else {
	330	args_start_at_rdi = TRUE;
	331	args_in_regs = 6;
0c530ab8 A	332	}
	333
	334	if (callp->sy_narg != 0) {
fe8ab488 A	335	assert(callp->sy_narg <= 8); /* size of uu_arg */
	336
	337	args_in_regs = MIN(args_in_regs, callp->sy_narg);
	338	memcpy(vt, args_start_at_rdi ? &regs->rdi : &regs->rsi, args_in_regs * sizeof(syscall_arg_t));
	339
	340
a1c7dba1	341	if (!code_is_kdebug_trace(code)) {
fe8ab488	342	uint64_t ip = (uint64_t )vt;
0c530ab8	343
316670eb A	344	KERNEL_DEBUG_CONSTANT_IST(KDEBUG_TRACE,
	345	BSDDBG_CODE(DBG_BSD_EXCP_SC, code) \| DBG_FUNC_START,
	346	(int)(ip), (int)((ip+1)), (int)((ip+2)), (int)((ip+3)), 0);
0c530ab8	347	}
0c530ab8	348
6d2010ae	349	if (__improbable(callp->sy_narg > args_in_regs)) {
2d21ac55	350	int copyin_count;
0c530ab8	351
fe8ab488	352	copyin_count = (callp->sy_narg - args_in_regs) * sizeof(syscall_arg_t);
0c530ab8	353
fe8ab488	354	error = copyin((user_addr_t)(regs->isf.rsp + sizeof(user_addr_t)), (char *)&uthread->uu_arg[args_in_regs], copyin_count);
0c530ab8	355	if (error) {
2d21ac55	356	regs->rax = error;
0c530ab8 A	357	regs->isf.rflags \|= EFL_CF;
	358	thread_exception_return();
	359	/* NOTREACHED */
	360	}
	361	}
0c530ab8	362	} else
316670eb A	363	KERNEL_DEBUG_CONSTANT_IST(KDEBUG_TRACE,
	364	BSDDBG_CODE(DBG_BSD_EXCP_SC, code) \| DBG_FUNC_START,
	365	0, 0, 0, 0, 0);
0c530ab8 A	366
	367	/*
	368	* Delayed binding of thread credential to process credential, if we
	369	* are not running with an explicitly set thread credential.
	370	*/
2d21ac55	371	kauth_cred_uthread_update(uthread, p);
0c530ab8 A	372
	373	uthread->uu_rval[0] = 0;
	374	uthread->uu_rval[1] = 0;
2d21ac55	375	uthread->uu_flag \|= UT_NOTCANCELPT;
fe8ab488	376	uthread->syscall_code = code;
0c530ab8	377
6d2010ae A	378	#ifdef JOE_DEBUG
	379	uthread->uu_iocount = 0;
	380	uthread->uu_vpindex = 0;
	381	#endif
0c530ab8 A	382
0c530ab8 A	383	AUDIT_SYSCALL_ENTER(code, p, uthread);
fe8ab488 A	384	error = ((callp->sy_call))((void ) p, vt, &(uthread->uu_rval[0]));
fe8ab488 A	385	AUDIT_SYSCALL_EXIT(code, p, uthread, error);
2d21ac55	386
6d2010ae A	387	#ifdef JOE_DEBUG
	388	if (uthread->uu_iocount)
	389	printf("system call returned with uu_iocount != 0\n");
	390	#endif
	391
2d21ac55 A	392	#if CONFIG_DTRACE
	393	uthread->t_dtrace_errno = error;
	394	#endif /* CONFIG_DTRACE */
0c530ab8	395
6d2010ae	396	if (__improbable(error == ERESTART)) {
0c530ab8 A	397	/*
	398	* all system calls come through via the syscall instruction
	399	* in 64 bit mode... its 2 bytes in length
	400	* move the user's pc back to repeat the syscall:
	401	*/
6d2010ae	402	pal_syscall_restart( thread, state );
0c530ab8 A	403	}
0c530ab8 A	404	else if (error != EJUSTRETURN) {
6d2010ae	405	if (__improbable(error)) {
2d21ac55	406	regs->rax = error;
0c530ab8 A	407	regs->isf.rflags \|= EFL_CF; /* carry bit */
	408	} else { /* (not error) */
	409
	410	switch (callp->sy_return_type) {
	411	case _SYSCALL_RET_INT_T:
	412	regs->rax = uthread->uu_rval[0];
	413	regs->rdx = uthread->uu_rval[1];
	414	break;
	415	case _SYSCALL_RET_UINT_T:
	416	regs->rax = ((u_int)uthread->uu_rval[0]);
	417	regs->rdx = ((u_int)uthread->uu_rval[1]);
	418	break;
	419	case _SYSCALL_RET_OFF_T:
	420	case _SYSCALL_RET_ADDR_T:
	421	case _SYSCALL_RET_SIZE_T:
	422	case _SYSCALL_RET_SSIZE_T:
d1ecb069	423	case _SYSCALL_RET_UINT64_T:
0c530ab8 A	424	regs->rax = ((uint64_t )(&uthread->uu_rval[0]));
	425	regs->rdx = 0;
	426	break;
	427	case _SYSCALL_RET_NONE:
	428	break;
	429	default:
	430	panic("unix_syscall: unknown return type");
	431	break;
	432	}
	433	regs->isf.rflags &= ~EFL_CF;
	434	}
	435	}
	436
b0d623f7 A	437	DEBUG_KPRINT_SYSCALL_UNIX(
	438	"unix_syscall64: error=%d retval=(%llu,%llu)\n",
	439	error, regs->rax, regs->rdx);
	440
2d21ac55	441	uthread->uu_flag &= ~UT_NOTCANCELPT;
0c530ab8	442
6d2010ae	443	if (__improbable(uthread->uu_lowpri_window)) {
0c530ab8 A	444	/*
	445	* task is marked as a low priority I/O type
	446	* and the I/O we issued while in this system call
	447	* collided with normal I/O operations... we'll
	448	* delay in order to mitigate the impact of this
	449	* task on the normal operation of the system
	450	*/
39236c6e	451	throttle_lowpri_io(1);
0c530ab8	452	}
a1c7dba1	453	if (__probable(!code_is_kdebug_trace(code)))
316670eb A	454	KERNEL_DEBUG_CONSTANT_IST(KDEBUG_TRACE,
	455	BSDDBG_CODE(DBG_BSD_EXCP_SC, code) \| DBG_FUNC_END,
	456	error, uthread->uu_rval[0], uthread->uu_rval[1], p->p_pid, 0);
0c530ab8	457
3e170ce0 A	458	#if PROC_REF_DEBUG
	459	if (__improbable(uthread_get_proc_refcount(uthread))) {
	460	panic("system call returned with uu_proc_refcount != 0");
	461	}
	462	#endif
	463
0c530ab8 A	464	thread_exception_return();
	465	/* NOTREACHED */
	466	}
	467
	468
	469	void
	470	unix_syscall_return(int error)
	471	{
	472	thread_t thread;
	473	struct uthread *uthread;
	474	struct proc *p;
	475	unsigned int code;
0c530ab8	476	struct sysent *callp;
0c530ab8 A	477
	478	thread = current_thread();
	479	uthread = get_bsdthread_info(thread);
	480
6d2010ae	481	pal_register_cache_state(thread, DIRTY);
b0d623f7	482
0c530ab8 A	483	p = current_proc();
	484
	485	if (proc_is64bit(p)) {
2d21ac55	486	x86_saved_state64_t *regs;
0c530ab8 A	487
	488	regs = saved_state64(find_user_regs(thread));
	489
fe8ab488	490	code = uthread->syscall_code;
2d21ac55	491	callp = (code >= NUM_SYSENT) ? &sysent[63] : &sysent[code];
0c530ab8	492
2d21ac55 A	493	#if CONFIG_DTRACE
	494	if (callp->sy_call == dtrace_systrace_syscall)
	495	dtrace_systrace_syscall_return( code, error, uthread->uu_rval );
	496	#endif /* CONFIG_DTRACE */
b0d623f7	497	AUDIT_SYSCALL_EXIT(code, p, uthread, error);
0c530ab8 A	498
0c530ab8 A	499	if (error == ERESTART) {
2d21ac55	500	/*
6d2010ae	501	* repeat the syscall
0c530ab8	502	*/
6d2010ae	503	pal_syscall_restart( thread, find_user_regs(thread) );
0c530ab8 A	504	}
0c530ab8 A	505	else if (error != EJUSTRETURN) {
2d21ac55 A	506	if (error) {
2d21ac55 A	507	regs->rax = error;
0c530ab8 A	508	regs->isf.rflags \|= EFL_CF; /* carry bit */
	509	} else { /* (not error) */
	510
2d21ac55	511	switch (callp->sy_return_type) {
0c530ab8	512	case _SYSCALL_RET_INT_T:
2d21ac55	513	regs->rax = uthread->uu_rval[0];
0c530ab8 A	514	regs->rdx = uthread->uu_rval[1];
	515	break;
	516	case _SYSCALL_RET_UINT_T:
2d21ac55	517	regs->rax = ((u_int)uthread->uu_rval[0]);
0c530ab8 A	518	regs->rdx = ((u_int)uthread->uu_rval[1]);
	519	break;
	520	case _SYSCALL_RET_OFF_T:
	521	case _SYSCALL_RET_ADDR_T:
	522	case _SYSCALL_RET_SIZE_T:
	523	case _SYSCALL_RET_SSIZE_T:
d1ecb069	524	case _SYSCALL_RET_UINT64_T:
2d21ac55	525	regs->rax = ((uint64_t )(&uthread->uu_rval[0]));
0c530ab8 A	526	regs->rdx = 0;
	527	break;
	528	case _SYSCALL_RET_NONE:
2d21ac55	529	break;
0c530ab8	530	default:
2d21ac55	531	panic("unix_syscall: unknown return type");
0c530ab8 A	532	break;
	533	}
	534	regs->isf.rflags &= ~EFL_CF;
	535	}
	536	}
b0d623f7 A	537	DEBUG_KPRINT_SYSCALL_UNIX(
	538	"unix_syscall_return: error=%d retval=(%llu,%llu)\n",
	539	error, regs->rax, regs->rdx);
0c530ab8	540	} else {
2d21ac55	541	x86_saved_state32_t *regs;
0c530ab8 A	542
	543	regs = saved_state32(find_user_regs(thread));
	544
2d21ac55	545	regs->efl &= ~(EFL_CF);
fe8ab488 A	546
fe8ab488 A	547	code = uthread->syscall_code;
2d21ac55 A	548	callp = (code >= NUM_SYSENT) ? &sysent[63] : &sysent[code];
	549
	550	#if CONFIG_DTRACE
	551	if (callp->sy_call == dtrace_systrace_syscall)
	552	dtrace_systrace_syscall_return( code, error, uthread->uu_rval );
	553	#endif /* CONFIG_DTRACE */
b0d623f7	554	AUDIT_SYSCALL_EXIT(code, p, uthread, error);
0c530ab8	555
0c530ab8	556	if (error == ERESTART) {
6d2010ae	557	pal_syscall_restart( thread, find_user_regs(thread) );
0c530ab8 A	558	}
0c530ab8 A	559	else if (error != EJUSTRETURN) {
2d21ac55 A	560	if (error) {
2d21ac55 A	561	regs->eax = error;
0c530ab8 A	562	regs->efl \|= EFL_CF; /* carry bit */
0c530ab8 A	563	} else { /* (not error) */
2d21ac55	564	regs->eax = uthread->uu_rval[0];
0c530ab8	565	regs->edx = uthread->uu_rval[1];
0c530ab8 A	566	}
0c530ab8 A	567	}
b0d623f7 A	568	DEBUG_KPRINT_SYSCALL_UNIX(
	569	"unix_syscall_return: error=%d retval=(%u,%u)\n",
	570	error, regs->eax, regs->edx);
0c530ab8	571	}
0c530ab8	572
0c530ab8	573
2d21ac55	574	uthread->uu_flag &= ~UT_NOTCANCELPT;
0c530ab8	575
593a1d5f	576	if (uthread->uu_lowpri_window) {
0c530ab8 A	577	/*
	578	* task is marked as a low priority I/O type
	579	* and the I/O we issued while in this system call
	580	* collided with normal I/O operations... we'll
	581	* delay in order to mitigate the impact of this
	582	* task on the normal operation of the system
	583	*/
39236c6e	584	throttle_lowpri_io(1);
0c530ab8	585	}
a1c7dba1	586	if (!code_is_kdebug_trace(code))
316670eb A	587	KERNEL_DEBUG_CONSTANT_IST(KDEBUG_TRACE,
	588	BSDDBG_CODE(DBG_BSD_EXCP_SC, code) \| DBG_FUNC_END,
	589	error, uthread->uu_rval[0], uthread->uu_rval[1], p->p_pid, 0);
0c530ab8 A	590
	591	thread_exception_return();
	592	/* NOTREACHED */
	593	}
	594