[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))

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

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

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

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

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

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

	regs->efl &= ~(EFL_CF);

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

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

	vt = (void *)uthread->uu_arg;

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

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

		if (__probable(!code_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);
	}

	thread_exception_return();
	/* NOTREACHED */
}


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

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

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

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

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

	vt = (void *)uthread->uu_arg;

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

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

		args_in_regs = MIN(args_in_regs, callp->sy_narg);
		memcpy(vt, args_start_at_rdi ? &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);

	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
a1c7dba1 A	72	#define code_is_kdebug_trace(code) (((code) == SYS_kdebug_trace) \|\| ((code) == SYS_kdebug_trace64))
a1c7dba1 A	73
0c530ab8 A	74	/*
	75	* Function: unix_syscall
	76	*
	77	* Inputs: regs - pointer to i386 save area
	78	*
	79	* Outputs: none
	80	*/
	81	void
	82	unix_syscall(x86_saved_state_t *state)
	83	{
2d21ac55 A	84	thread_t thread;
	85	void *vt;
	86	unsigned int code;
	87	struct sysent *callp;
	88
	89	int error;
	90	vm_offset_t params;
	91	struct proc *p;
	92	struct uthread *uthread;
0c530ab8	93	x86_saved_state32_t *regs;
6d2010ae	94	boolean_t is_vfork;
0c530ab8 A	95
	96	assert(is_saved_state32(state));
	97	regs = saved_state32(state);
2d21ac55	98	#if DEBUG
0c530ab8 A	99	if (regs->eax == 0x800)
0c530ab8 A	100	thread_exception_return();
2d21ac55	101	#endif
0c530ab8 A	102	thread = current_thread();
	103	uthread = get_bsdthread_info(thread);
	104
	105	/* Get the approriate proc; may be different from task's for vfork() */
6d2010ae A	106	is_vfork = uthread->uu_flag & UT_VFORK;
6d2010ae A	107	if (__improbable(is_vfork != 0))
0c530ab8	108	p = current_proc();
6d2010ae A	109	else
6d2010ae A	110	p = (struct proc *)get_bsdtask_info(current_task());
0c530ab8 A	111
0c530ab8 A	112	/* Verify that we are not being called from a task without a proc */
6d2010ae	113	if (__improbable(p == NULL)) {
0c530ab8 A	114	regs->eax = EPERM;
	115	regs->efl \|= EFL_CF;
	116	task_terminate_internal(current_task());
	117	thread_exception_return();
	118	/* NOTREACHED */
	119	}
	120
2d21ac55	121	code = regs->eax & I386_SYSCALL_NUMBER_MASK;
b0d623f7 A	122	DEBUG_KPRINT_SYSCALL_UNIX("unix_syscall: code=%d(%s) eip=%u\n",
b0d623f7 A	123	code, syscallnames[code >= NUM_SYSENT ? 63 : code], (uint32_t)regs->eip);
b0d623f7	124	params = (vm_offset_t) (regs->uesp + sizeof (int));
2d21ac55 A	125
	126	regs->efl &= ~(EFL_CF);
	127
	128	callp = (code >= NUM_SYSENT) ? &sysent[63] : &sysent[code];
0c530ab8	129
6d2010ae	130	if (__improbable(callp == sysent)) {
0c530ab8	131	code = fuword(params);
2d21ac55 A	132	params += sizeof(int);
2d21ac55 A	133	callp = (code >= NUM_SYSENT) ? &sysent[63] : &sysent[code];
0c530ab8	134	}
2d21ac55	135
0c530ab8 A	136	vt = (void *)uthread->uu_arg;
0c530ab8 A	137
2d21ac55	138	if (callp->sy_arg_bytes != 0) {
fe8ab488	139	#if CONFIG_REQUIRES_U32_MUNGING
0c530ab8	140	sy_munge_t *mungerp;
fe8ab488 A	141	#else
	142	#error U32 syscalls on x86_64 kernel requires munging
	143	#endif
39236c6e	144	uint32_t nargs;
0c530ab8	145
2d21ac55	146	assert((unsigned) callp->sy_arg_bytes <= sizeof (uthread->uu_arg));
39236c6e A	147	nargs = callp->sy_arg_bytes;
	148	error = copyin((user_addr_t) params, (char *) vt, nargs);
	149	if (error) {
	150	regs->eax = error;
	151	regs->efl \|= EFL_CF;
	152	thread_exception_return();
	153	/* NOTREACHED */
0c530ab8	154	}
2d21ac55	155
a1c7dba1 A	156	if (__probable(!code_is_kdebug_trace(code))) {
a1c7dba1 A	157	int ip = (int )vt;
0c530ab8	158
316670eb A	159	KERNEL_DEBUG_CONSTANT_IST(KDEBUG_TRACE,
	160	BSDDBG_CODE(DBG_BSD_EXCP_SC, code) \| DBG_FUNC_START,
	161	ip, (ip+1), (ip+2), (ip+3), 0);
0c530ab8	162	}
fe8ab488 A	163
fe8ab488 A	164	#if CONFIG_REQUIRES_U32_MUNGING
0c530ab8 A	165	mungerp = callp->sy_arg_munge32;
0c530ab8 A	166
0c530ab8	167	if (mungerp != NULL)
fe8ab488 A	168	(*mungerp)(vt);
fe8ab488 A	169	#endif
0c530ab8	170	} else
316670eb A	171	KERNEL_DEBUG_CONSTANT_IST(KDEBUG_TRACE,
316670eb A	172	BSDDBG_CODE(DBG_BSD_EXCP_SC, code) \| DBG_FUNC_START,
2d21ac55 A	173	0, 0, 0, 0, 0);
2d21ac55 A	174
0c530ab8 A	175	/*
	176	* Delayed binding of thread credential to process credential, if we
	177	* are not running with an explicitly set thread credential.
	178	*/
2d21ac55	179	kauth_cred_uthread_update(uthread, p);
0c530ab8 A	180
0c530ab8 A	181	uthread->uu_rval[0] = 0;
fe8ab488	182	uthread->uu_rval[1] = 0;
2d21ac55	183	uthread->uu_flag \|= UT_NOTCANCELPT;
fe8ab488	184	uthread->syscall_code = code;
0c530ab8	185
2d21ac55 A	186	#ifdef JOE_DEBUG
	187	uthread->uu_iocount = 0;
	188	uthread->uu_vpindex = 0;
	189	#endif
0c530ab8 A	190
	191	AUDIT_SYSCALL_ENTER(code, p, uthread);
	192	error = ((callp->sy_call))((void ) p, (void *) vt, &(uthread->uu_rval[0]));
fe8ab488	193	AUDIT_SYSCALL_EXIT(code, p, uthread, error);
2d21ac55 A	194
	195	#ifdef JOE_DEBUG
	196	if (uthread->uu_iocount)
b0d623f7	197	printf("system call returned with uu_iocount != 0\n");
2d21ac55 A	198	#endif
	199	#if CONFIG_DTRACE
	200	uthread->t_dtrace_errno = error;
	201	#endif /* CONFIG_DTRACE */
	202
6d2010ae	203	if (__improbable(error == ERESTART)) {
0c530ab8 A	204	/*
	205	* Move the user's pc back to repeat the syscall:
	206	* 5 bytes for a sysenter, or 2 for an int 8x.
	207	* The SYSENTER_TF_CS covers single-stepping over a sysenter
	208	* - see debug trap handler in idt.s/idt64.s
	209	*/
b0d623f7	210
6d2010ae	211	pal_syscall_restart(thread, state);
0c530ab8 A	212	}
0c530ab8 A	213	else if (error != EJUSTRETURN) {
6d2010ae	214	if (__improbable(error)) {
0c530ab8 A	215	regs->eax = error;
	216	regs->efl \|= EFL_CF; /* carry bit */
	217	} else { /* (not error) */
fe8ab488 A	218	/*
	219	* We split retval across two registers, in case the
	220	* syscall had a 64-bit return value, in which case
	221	* eax/edx matches the function call ABI.
	222	*/
0c530ab8 A	223	regs->eax = uthread->uu_rval[0];
0c530ab8 A	224	regs->edx = uthread->uu_rval[1];
0c530ab8 A	225	}
	226	}
	227
b0d623f7 A	228	DEBUG_KPRINT_SYSCALL_UNIX(
	229	"unix_syscall: error=%d retval=(%u,%u)\n",
	230	error, regs->eax, regs->edx);
	231
2d21ac55	232	uthread->uu_flag &= ~UT_NOTCANCELPT;
6d2010ae A	233
6d2010ae A	234	if (__improbable(uthread->uu_lowpri_window)) {
0c530ab8 A	235	/*
	236	* task is marked as a low priority I/O type
	237	* and the I/O we issued while in this system call
	238	* collided with normal I/O operations... we'll
	239	* delay in order to mitigate the impact of this
	240	* task on the normal operation of the system
	241	*/
39236c6e	242	throttle_lowpri_io(1);
0c530ab8	243	}
a1c7dba1	244	if (__probable(!code_is_kdebug_trace(code)))
316670eb A	245	KERNEL_DEBUG_CONSTANT_IST(KDEBUG_TRACE,
	246	BSDDBG_CODE(DBG_BSD_EXCP_SC, code) \| DBG_FUNC_END,
	247	error, uthread->uu_rval[0], uthread->uu_rval[1], p->p_pid, 0);
b0d623f7	248
6d2010ae A	249	if (__improbable(!is_vfork && callp->sy_call == (sy_call_t *)execve && !error)) {
	250	pal_execve_return(thread);
	251	}
0c530ab8 A	252
	253	thread_exception_return();
	254	/* NOTREACHED */
	255	}
	256
	257
	258	void
	259	unix_syscall64(x86_saved_state_t *state)
	260	{
	261	thread_t thread;
fe8ab488	262	void *vt;
0c530ab8 A	263	unsigned int code;
0c530ab8 A	264	struct sysent *callp;
0c530ab8	265	int args_in_regs;
fe8ab488	266	boolean_t args_start_at_rdi;
0c530ab8	267	int error;
0c530ab8 A	268	struct proc *p;
0c530ab8 A	269	struct uthread *uthread;
0c530ab8 A	270	x86_saved_state64_t *regs;
	271
	272	assert(is_saved_state64(state));
	273	regs = saved_state64(state);
6d2010ae	274	#if DEBUG
0c530ab8 A	275	if (regs->rax == 0x2000800)
0c530ab8 A	276	thread_exception_return();
6d2010ae	277	#endif
0c530ab8 A	278	thread = current_thread();
	279	uthread = get_bsdthread_info(thread);
	280
	281	/* Get the approriate proc; may be different from task's for vfork() */
6d2010ae	282	if (__probable(!(uthread->uu_flag & UT_VFORK)))
0c530ab8 A	283	p = (struct proc *)get_bsdtask_info(current_task());
	284	else
	285	p = current_proc();
	286
	287	/* Verify that we are not being called from a task without a proc */
6d2010ae	288	if (__improbable(p == NULL)) {
0c530ab8 A	289	regs->rax = EPERM;
	290	regs->isf.rflags \|= EFL_CF;
	291	task_terminate_internal(current_task());
	292	thread_exception_return();
	293	/* NOTREACHED */
	294	}
0c530ab8 A	295
0c530ab8 A	296	code = regs->rax & SYSCALL_NUMBER_MASK;
b0d623f7 A	297	DEBUG_KPRINT_SYSCALL_UNIX(
	298	"unix_syscall64: code=%d(%s) rip=%llx\n",
	299	code, syscallnames[code >= NUM_SYSENT ? 63 : code], regs->isf.rip);
2d21ac55	300	callp = (code >= NUM_SYSENT) ? &sysent[63] : &sysent[code];
fe8ab488 A	301
fe8ab488 A	302	vt = (void *)uthread->uu_arg;
0c530ab8	303
6d2010ae	304	if (__improbable(callp == sysent)) {
0c530ab8 A	305	/*
	306	* indirect system call... system call number
	307	* passed as 'arg0'
	308	*/
fe8ab488	309	code = regs->rdi;
2d21ac55	310	callp = (code >= NUM_SYSENT) ? &sysent[63] : &sysent[code];
fe8ab488	311	args_start_at_rdi = FALSE;
0c530ab8	312	args_in_regs = 5;
fe8ab488 A	313	} else {
	314	args_start_at_rdi = TRUE;
	315	args_in_regs = 6;
0c530ab8 A	316	}
	317
	318	if (callp->sy_narg != 0) {
fe8ab488 A	319	assert(callp->sy_narg <= 8); /* size of uu_arg */
	320
	321	args_in_regs = MIN(args_in_regs, callp->sy_narg);
	322	memcpy(vt, args_start_at_rdi ? &regs->rdi : &regs->rsi, args_in_regs * sizeof(syscall_arg_t));
	323
	324
a1c7dba1	325	if (!code_is_kdebug_trace(code)) {
fe8ab488	326	uint64_t ip = (uint64_t )vt;
0c530ab8	327
316670eb A	328	KERNEL_DEBUG_CONSTANT_IST(KDEBUG_TRACE,
	329	BSDDBG_CODE(DBG_BSD_EXCP_SC, code) \| DBG_FUNC_START,
	330	(int)(ip), (int)((ip+1)), (int)((ip+2)), (int)((ip+3)), 0);
0c530ab8	331	}
0c530ab8	332
6d2010ae	333	if (__improbable(callp->sy_narg > args_in_regs)) {
2d21ac55	334	int copyin_count;
0c530ab8	335
fe8ab488	336	copyin_count = (callp->sy_narg - args_in_regs) * sizeof(syscall_arg_t);
0c530ab8	337
fe8ab488	338	error = copyin((user_addr_t)(regs->isf.rsp + sizeof(user_addr_t)), (char *)&uthread->uu_arg[args_in_regs], copyin_count);
0c530ab8	339	if (error) {
2d21ac55	340	regs->rax = error;
0c530ab8 A	341	regs->isf.rflags \|= EFL_CF;
	342	thread_exception_return();
	343	/* NOTREACHED */
	344	}
	345	}
0c530ab8	346	} else
316670eb A	347	KERNEL_DEBUG_CONSTANT_IST(KDEBUG_TRACE,
	348	BSDDBG_CODE(DBG_BSD_EXCP_SC, code) \| DBG_FUNC_START,
	349	0, 0, 0, 0, 0);
0c530ab8 A	350
	351	/*
	352	* Delayed binding of thread credential to process credential, if we
	353	* are not running with an explicitly set thread credential.
	354	*/
2d21ac55	355	kauth_cred_uthread_update(uthread, p);
0c530ab8 A	356
	357	uthread->uu_rval[0] = 0;
	358	uthread->uu_rval[1] = 0;
2d21ac55	359	uthread->uu_flag \|= UT_NOTCANCELPT;
fe8ab488	360	uthread->syscall_code = code;
0c530ab8	361
6d2010ae A	362	#ifdef JOE_DEBUG
	363	uthread->uu_iocount = 0;
	364	uthread->uu_vpindex = 0;
	365	#endif
0c530ab8 A	366
0c530ab8 A	367	AUDIT_SYSCALL_ENTER(code, p, uthread);
fe8ab488 A	368	error = ((callp->sy_call))((void ) p, vt, &(uthread->uu_rval[0]));
fe8ab488 A	369	AUDIT_SYSCALL_EXIT(code, p, uthread, error);
2d21ac55	370
6d2010ae A	371	#ifdef JOE_DEBUG
	372	if (uthread->uu_iocount)
	373	printf("system call returned with uu_iocount != 0\n");
	374	#endif
	375
2d21ac55 A	376	#if CONFIG_DTRACE
	377	uthread->t_dtrace_errno = error;
	378	#endif /* CONFIG_DTRACE */
0c530ab8	379
6d2010ae	380	if (__improbable(error == ERESTART)) {
0c530ab8 A	381	/*
	382	* all system calls come through via the syscall instruction
	383	* in 64 bit mode... its 2 bytes in length
	384	* move the user's pc back to repeat the syscall:
	385	*/
6d2010ae	386	pal_syscall_restart( thread, state );
0c530ab8 A	387	}
0c530ab8 A	388	else if (error != EJUSTRETURN) {
6d2010ae	389	if (__improbable(error)) {
2d21ac55	390	regs->rax = error;
0c530ab8 A	391	regs->isf.rflags \|= EFL_CF; /* carry bit */
	392	} else { /* (not error) */
	393
	394	switch (callp->sy_return_type) {
	395	case _SYSCALL_RET_INT_T:
	396	regs->rax = uthread->uu_rval[0];
	397	regs->rdx = uthread->uu_rval[1];
	398	break;
	399	case _SYSCALL_RET_UINT_T:
	400	regs->rax = ((u_int)uthread->uu_rval[0]);
	401	regs->rdx = ((u_int)uthread->uu_rval[1]);
	402	break;
	403	case _SYSCALL_RET_OFF_T:
	404	case _SYSCALL_RET_ADDR_T:
	405	case _SYSCALL_RET_SIZE_T:
	406	case _SYSCALL_RET_SSIZE_T:
d1ecb069	407	case _SYSCALL_RET_UINT64_T:
0c530ab8 A	408	regs->rax = ((uint64_t )(&uthread->uu_rval[0]));
	409	regs->rdx = 0;
	410	break;
	411	case _SYSCALL_RET_NONE:
	412	break;
	413	default:
	414	panic("unix_syscall: unknown return type");
	415	break;
	416	}
	417	regs->isf.rflags &= ~EFL_CF;
	418	}
	419	}
	420
b0d623f7 A	421	DEBUG_KPRINT_SYSCALL_UNIX(
	422	"unix_syscall64: error=%d retval=(%llu,%llu)\n",
	423	error, regs->rax, regs->rdx);
	424
2d21ac55	425	uthread->uu_flag &= ~UT_NOTCANCELPT;
0c530ab8	426
6d2010ae	427	if (__improbable(uthread->uu_lowpri_window)) {
0c530ab8 A	428	/*
	429	* task is marked as a low priority I/O type
	430	* and the I/O we issued while in this system call
	431	* collided with normal I/O operations... we'll
	432	* delay in order to mitigate the impact of this
	433	* task on the normal operation of the system
	434	*/
39236c6e	435	throttle_lowpri_io(1);
0c530ab8	436	}
a1c7dba1	437	if (__probable(!code_is_kdebug_trace(code)))
316670eb A	438	KERNEL_DEBUG_CONSTANT_IST(KDEBUG_TRACE,
	439	BSDDBG_CODE(DBG_BSD_EXCP_SC, code) \| DBG_FUNC_END,
	440	error, uthread->uu_rval[0], uthread->uu_rval[1], p->p_pid, 0);
0c530ab8 A	441
	442	thread_exception_return();
	443	/* NOTREACHED */
	444	}
	445
	446
	447	void
	448	unix_syscall_return(int error)
	449	{
	450	thread_t thread;
	451	struct uthread *uthread;
	452	struct proc *p;
	453	unsigned int code;
0c530ab8	454	struct sysent *callp;
0c530ab8 A	455
	456	thread = current_thread();
	457	uthread = get_bsdthread_info(thread);
	458
6d2010ae	459	pal_register_cache_state(thread, DIRTY);
b0d623f7	460
0c530ab8 A	461	p = current_proc();
	462
	463	if (proc_is64bit(p)) {
2d21ac55	464	x86_saved_state64_t *regs;
0c530ab8 A	465
	466	regs = saved_state64(find_user_regs(thread));
	467
fe8ab488	468	code = uthread->syscall_code;
2d21ac55	469	callp = (code >= NUM_SYSENT) ? &sysent[63] : &sysent[code];
0c530ab8	470
2d21ac55 A	471	#if CONFIG_DTRACE
	472	if (callp->sy_call == dtrace_systrace_syscall)
	473	dtrace_systrace_syscall_return( code, error, uthread->uu_rval );
	474	#endif /* CONFIG_DTRACE */
b0d623f7	475	AUDIT_SYSCALL_EXIT(code, p, uthread, error);
0c530ab8 A	476
0c530ab8 A	477	if (error == ERESTART) {
2d21ac55	478	/*
6d2010ae	479	* repeat the syscall
0c530ab8	480	*/
6d2010ae	481	pal_syscall_restart( thread, find_user_regs(thread) );
0c530ab8 A	482	}
0c530ab8 A	483	else if (error != EJUSTRETURN) {
2d21ac55 A	484	if (error) {
2d21ac55 A	485	regs->rax = error;
0c530ab8 A	486	regs->isf.rflags \|= EFL_CF; /* carry bit */
	487	} else { /* (not error) */
	488
2d21ac55	489	switch (callp->sy_return_type) {
0c530ab8	490	case _SYSCALL_RET_INT_T:
2d21ac55	491	regs->rax = uthread->uu_rval[0];
0c530ab8 A	492	regs->rdx = uthread->uu_rval[1];
	493	break;
	494	case _SYSCALL_RET_UINT_T:
2d21ac55	495	regs->rax = ((u_int)uthread->uu_rval[0]);
0c530ab8 A	496	regs->rdx = ((u_int)uthread->uu_rval[1]);
	497	break;
	498	case _SYSCALL_RET_OFF_T:
	499	case _SYSCALL_RET_ADDR_T:
	500	case _SYSCALL_RET_SIZE_T:
	501	case _SYSCALL_RET_SSIZE_T:
d1ecb069	502	case _SYSCALL_RET_UINT64_T:
2d21ac55	503	regs->rax = ((uint64_t )(&uthread->uu_rval[0]));
0c530ab8 A	504	regs->rdx = 0;
	505	break;
	506	case _SYSCALL_RET_NONE:
2d21ac55	507	break;
0c530ab8	508	default:
2d21ac55	509	panic("unix_syscall: unknown return type");
0c530ab8 A	510	break;
	511	}
	512	regs->isf.rflags &= ~EFL_CF;
	513	}
	514	}
b0d623f7 A	515	DEBUG_KPRINT_SYSCALL_UNIX(
	516	"unix_syscall_return: error=%d retval=(%llu,%llu)\n",
	517	error, regs->rax, regs->rdx);
0c530ab8	518	} else {
2d21ac55	519	x86_saved_state32_t *regs;
0c530ab8 A	520
	521	regs = saved_state32(find_user_regs(thread));
	522
2d21ac55	523	regs->efl &= ~(EFL_CF);
fe8ab488 A	524
fe8ab488 A	525	code = uthread->syscall_code;
2d21ac55 A	526	callp = (code >= NUM_SYSENT) ? &sysent[63] : &sysent[code];
	527
	528	#if CONFIG_DTRACE
	529	if (callp->sy_call == dtrace_systrace_syscall)
	530	dtrace_systrace_syscall_return( code, error, uthread->uu_rval );
	531	#endif /* CONFIG_DTRACE */
b0d623f7	532	AUDIT_SYSCALL_EXIT(code, p, uthread, error);
0c530ab8	533
0c530ab8	534	if (error == ERESTART) {
6d2010ae	535	pal_syscall_restart( thread, find_user_regs(thread) );
0c530ab8 A	536	}
0c530ab8 A	537	else if (error != EJUSTRETURN) {
2d21ac55 A	538	if (error) {
2d21ac55 A	539	regs->eax = error;
0c530ab8 A	540	regs->efl \|= EFL_CF; /* carry bit */
0c530ab8 A	541	} else { /* (not error) */
2d21ac55	542	regs->eax = uthread->uu_rval[0];
0c530ab8	543	regs->edx = uthread->uu_rval[1];
0c530ab8 A	544	}
0c530ab8 A	545	}
b0d623f7 A	546	DEBUG_KPRINT_SYSCALL_UNIX(
	547	"unix_syscall_return: error=%d retval=(%u,%u)\n",
	548	error, regs->eax, regs->edx);
0c530ab8	549	}
0c530ab8	550
0c530ab8	551
2d21ac55	552	uthread->uu_flag &= ~UT_NOTCANCELPT;
0c530ab8	553
593a1d5f	554	if (uthread->uu_lowpri_window) {
0c530ab8 A	555	/*
	556	* task is marked as a low priority I/O type
	557	* and the I/O we issued while in this system call
	558	* collided with normal I/O operations... we'll
	559	* delay in order to mitigate the impact of this
	560	* task on the normal operation of the system
	561	*/
39236c6e	562	throttle_lowpri_io(1);
0c530ab8	563	}
a1c7dba1	564	if (!code_is_kdebug_trace(code))
316670eb A	565	KERNEL_DEBUG_CONSTANT_IST(KDEBUG_TRACE,
	566	BSDDBG_CODE(DBG_BSD_EXCP_SC, code) \| DBG_FUNC_END,
	567	error, uthread->uu_rval[0], uthread->uu_rval[1], p->p_pid, 0);
0c530ab8 A	568
	569	thread_exception_return();
	570	/* NOTREACHED */
	571	}
	572