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

/*
 * Copyright (c) 2000-2016 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@
 */
#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
 */
__attribute__((noreturn))
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;
	pid_t			pid;

	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());

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

	regs->efl &= ~(EFL_CF);

	callp = (code >= nsysent) ? &sysent[SYS_invalid] : &sysent[code];

	if (__improbable(callp == sysent)) {
		code = fuword(params);
		params += sizeof(int);
		callp = (code >= nsysent) ? &sysent[SYS_invalid] : &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;
	pid = proc_pid(p);

#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], 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 */
}

__attribute__((noreturn))
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;
	pid_t		pid;

	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 >= nsysent ? SYS_invalid : code], regs->isf.rip);
	callp = (code >= nsysent) ? &sysent[SYS_invalid] : &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 >= nsysent) ? &sysent[SYS_invalid] : &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;
	pid = proc_pid(p);

#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], 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 >= nsysent) ? &sysent[SYS_invalid] : &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 >= nsysent) ? &sysent[SYS_invalid] : &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	/*
39037602	2	* Copyright (c) 2000-2016 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	*/
39037602	83	__attribute__((noreturn))
0c530ab8 A	84	void
	85	unix_syscall(x86_saved_state_t *state)
	86	{
2d21ac55 A	87	thread_t thread;
	88	void *vt;
	89	unsigned int code;
	90	struct sysent *callp;
	91
	92	int error;
	93	vm_offset_t params;
	94	struct proc *p;
	95	struct uthread *uthread;
0c530ab8	96	x86_saved_state32_t *regs;
6d2010ae	97	boolean_t is_vfork;
743345f9	98	pid_t pid;
0c530ab8 A	99
	100	assert(is_saved_state32(state));
	101	regs = saved_state32(state);
2d21ac55	102	#if DEBUG
0c530ab8 A	103	if (regs->eax == 0x800)
0c530ab8 A	104	thread_exception_return();
2d21ac55	105	#endif
0c530ab8 A	106	thread = current_thread();
	107	uthread = get_bsdthread_info(thread);
	108
3e170ce0 A	109	#if PROC_REF_DEBUG
	110	uthread_reset_proc_refcount(uthread);
	111	#endif
	112
0c530ab8	113	/* Get the approriate proc; may be different from task's for vfork() */
6d2010ae A	114	is_vfork = uthread->uu_flag & UT_VFORK;
6d2010ae A	115	if (__improbable(is_vfork != 0))
0c530ab8	116	p = current_proc();
6d2010ae A	117	else
6d2010ae A	118	p = (struct proc *)get_bsdtask_info(current_task());
0c530ab8	119
2d21ac55	120	code = regs->eax & I386_SYSCALL_NUMBER_MASK;
b0d623f7	121	DEBUG_KPRINT_SYSCALL_UNIX("unix_syscall: code=%d(%s) eip=%u\n",
39037602	122	code, syscallnames[code >= nsysent ? SYS_invalid : code], (uint32_t)regs->eip);
b0d623f7	123	params = (vm_offset_t) (regs->uesp + sizeof (int));
2d21ac55 A	124
	125	regs->efl &= ~(EFL_CF);
	126
39037602	127	callp = (code >= nsysent) ? &sysent[SYS_invalid] : &sysent[code];
0c530ab8	128
6d2010ae	129	if (__improbable(callp == sysent)) {
0c530ab8	130	code = fuword(params);
2d21ac55	131	params += sizeof(int);
39037602	132	callp = (code >= nsysent) ? &sysent[SYS_invalid] : &sysent[code];
0c530ab8	133	}
2d21ac55	134
0c530ab8 A	135	vt = (void *)uthread->uu_arg;
0c530ab8 A	136
2d21ac55	137	if (callp->sy_arg_bytes != 0) {
fe8ab488	138	#if CONFIG_REQUIRES_U32_MUNGING
0c530ab8	139	sy_munge_t *mungerp;
fe8ab488 A	140	#else
	141	#error U32 syscalls on x86_64 kernel requires munging
	142	#endif
39236c6e	143	uint32_t nargs;
0c530ab8	144
2d21ac55	145	assert((unsigned) callp->sy_arg_bytes <= sizeof (uthread->uu_arg));
39236c6e A	146	nargs = callp->sy_arg_bytes;
	147	error = copyin((user_addr_t) params, (char *) vt, nargs);
	148	if (error) {
	149	regs->eax = error;
	150	regs->efl \|= EFL_CF;
	151	thread_exception_return();
	152	/* NOTREACHED */
0c530ab8	153	}
2d21ac55	154
a1c7dba1 A	155	if (__probable(!code_is_kdebug_trace(code))) {
a1c7dba1 A	156	int ip = (int )vt;
0c530ab8	157
316670eb A	158	KERNEL_DEBUG_CONSTANT_IST(KDEBUG_TRACE,
	159	BSDDBG_CODE(DBG_BSD_EXCP_SC, code) \| DBG_FUNC_START,
	160	ip, (ip+1), (ip+2), (ip+3), 0);
0c530ab8	161	}
fe8ab488 A	162
fe8ab488 A	163	#if CONFIG_REQUIRES_U32_MUNGING
0c530ab8 A	164	mungerp = callp->sy_arg_munge32;
0c530ab8 A	165
0c530ab8	166	if (mungerp != NULL)
fe8ab488 A	167	(*mungerp)(vt);
fe8ab488 A	168	#endif
0c530ab8	169	} else
316670eb A	170	KERNEL_DEBUG_CONSTANT_IST(KDEBUG_TRACE,
316670eb A	171	BSDDBG_CODE(DBG_BSD_EXCP_SC, code) \| DBG_FUNC_START,
2d21ac55 A	172	0, 0, 0, 0, 0);
2d21ac55 A	173
0c530ab8 A	174	/*
	175	* Delayed binding of thread credential to process credential, if we
	176	* are not running with an explicitly set thread credential.
	177	*/
2d21ac55	178	kauth_cred_uthread_update(uthread, p);
0c530ab8 A	179
0c530ab8 A	180	uthread->uu_rval[0] = 0;
fe8ab488	181	uthread->uu_rval[1] = 0;
2d21ac55	182	uthread->uu_flag \|= UT_NOTCANCELPT;
fe8ab488	183	uthread->syscall_code = code;
743345f9	184	pid = proc_pid(p);
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,
316670eb A	246	BSDDBG_CODE(DBG_BSD_EXCP_SC, code) \| DBG_FUNC_END,
743345f9	247	error, uthread->uu_rval[0], uthread->uu_rval[1], 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	252
3e170ce0 A	253	#if PROC_REF_DEBUG
	254	if (__improbable(uthread_get_proc_refcount(uthread) != 0)) {
	255	panic("system call returned with uu_proc_refcount != 0");
	256	}
	257	#endif
	258
0c530ab8 A	259	thread_exception_return();
	260	/* NOTREACHED */
	261	}
	262
39037602	263	__attribute__((noreturn))
0c530ab8 A	264	void
	265	unix_syscall64(x86_saved_state_t *state)
	266	{
	267	thread_t thread;
fe8ab488	268	void *vt;
0c530ab8 A	269	unsigned int code;
0c530ab8 A	270	struct sysent *callp;
0c530ab8	271	int args_in_regs;
fe8ab488	272	boolean_t args_start_at_rdi;
0c530ab8	273	int error;
0c530ab8 A	274	struct proc *p;
0c530ab8 A	275	struct uthread *uthread;
0c530ab8	276	x86_saved_state64_t *regs;
743345f9	277	pid_t pid;
0c530ab8 A	278
	279	assert(is_saved_state64(state));
	280	regs = saved_state64(state);
6d2010ae	281	#if DEBUG
0c530ab8 A	282	if (regs->rax == 0x2000800)
0c530ab8 A	283	thread_exception_return();
6d2010ae	284	#endif
0c530ab8 A	285	thread = current_thread();
	286	uthread = get_bsdthread_info(thread);
	287
3e170ce0 A	288	#if PROC_REF_DEBUG
	289	uthread_reset_proc_refcount(uthread);
	290	#endif
	291
0c530ab8	292	/* Get the approriate proc; may be different from task's for vfork() */
6d2010ae	293	if (__probable(!(uthread->uu_flag & UT_VFORK)))
0c530ab8 A	294	p = (struct proc *)get_bsdtask_info(current_task());
	295	else
	296	p = current_proc();
	297
	298	/* Verify that we are not being called from a task without a proc */
6d2010ae	299	if (__improbable(p == NULL)) {
0c530ab8 A	300	regs->rax = EPERM;
	301	regs->isf.rflags \|= EFL_CF;
	302	task_terminate_internal(current_task());
	303	thread_exception_return();
	304	/* NOTREACHED */
	305	}
0c530ab8 A	306
0c530ab8 A	307	code = regs->rax & SYSCALL_NUMBER_MASK;
b0d623f7 A	308	DEBUG_KPRINT_SYSCALL_UNIX(
b0d623f7 A	309	"unix_syscall64: code=%d(%s) rip=%llx\n",
39037602 A	310	code, syscallnames[code >= nsysent ? SYS_invalid : code], regs->isf.rip);
39037602 A	311	callp = (code >= nsysent) ? &sysent[SYS_invalid] : &sysent[code];
fe8ab488 A	312
fe8ab488 A	313	vt = (void *)uthread->uu_arg;
0c530ab8	314
6d2010ae	315	if (__improbable(callp == sysent)) {
0c530ab8 A	316	/*
	317	* indirect system call... system call number
	318	* passed as 'arg0'
	319	*/
fe8ab488	320	code = regs->rdi;
39037602	321	callp = (code >= nsysent) ? &sysent[SYS_invalid] : &sysent[code];
fe8ab488	322	args_start_at_rdi = FALSE;
0c530ab8	323	args_in_regs = 5;
fe8ab488 A	324	} else {
	325	args_start_at_rdi = TRUE;
	326	args_in_regs = 6;
0c530ab8 A	327	}
	328
	329	if (callp->sy_narg != 0) {
fe8ab488 A	330	assert(callp->sy_narg <= 8); /* size of uu_arg */
	331
	332	args_in_regs = MIN(args_in_regs, callp->sy_narg);
	333	memcpy(vt, args_start_at_rdi ? &regs->rdi : &regs->rsi, args_in_regs * sizeof(syscall_arg_t));
	334
	335
a1c7dba1	336	if (!code_is_kdebug_trace(code)) {
fe8ab488	337	uint64_t ip = (uint64_t )vt;
0c530ab8	338
316670eb A	339	KERNEL_DEBUG_CONSTANT_IST(KDEBUG_TRACE,
	340	BSDDBG_CODE(DBG_BSD_EXCP_SC, code) \| DBG_FUNC_START,
	341	(int)(ip), (int)((ip+1)), (int)((ip+2)), (int)((ip+3)), 0);
0c530ab8	342	}
0c530ab8	343
6d2010ae	344	if (__improbable(callp->sy_narg > args_in_regs)) {
2d21ac55	345	int copyin_count;
0c530ab8	346
fe8ab488	347	copyin_count = (callp->sy_narg - args_in_regs) * sizeof(syscall_arg_t);
0c530ab8	348
fe8ab488	349	error = copyin((user_addr_t)(regs->isf.rsp + sizeof(user_addr_t)), (char *)&uthread->uu_arg[args_in_regs], copyin_count);
0c530ab8	350	if (error) {
2d21ac55	351	regs->rax = error;
0c530ab8 A	352	regs->isf.rflags \|= EFL_CF;
	353	thread_exception_return();
	354	/* NOTREACHED */
	355	}
	356	}
0c530ab8	357	} else
316670eb A	358	KERNEL_DEBUG_CONSTANT_IST(KDEBUG_TRACE,
	359	BSDDBG_CODE(DBG_BSD_EXCP_SC, code) \| DBG_FUNC_START,
	360	0, 0, 0, 0, 0);
0c530ab8 A	361
	362	/*
	363	* Delayed binding of thread credential to process credential, if we
	364	* are not running with an explicitly set thread credential.
	365	*/
2d21ac55	366	kauth_cred_uthread_update(uthread, p);
0c530ab8 A	367
	368	uthread->uu_rval[0] = 0;
	369	uthread->uu_rval[1] = 0;
2d21ac55	370	uthread->uu_flag \|= UT_NOTCANCELPT;
fe8ab488	371	uthread->syscall_code = code;
743345f9	372	pid = proc_pid(p);
0c530ab8	373
6d2010ae A	374	#ifdef JOE_DEBUG
	375	uthread->uu_iocount = 0;
	376	uthread->uu_vpindex = 0;
	377	#endif
0c530ab8 A	378
0c530ab8 A	379	AUDIT_SYSCALL_ENTER(code, p, uthread);
fe8ab488 A	380	error = ((callp->sy_call))((void ) p, vt, &(uthread->uu_rval[0]));
fe8ab488 A	381	AUDIT_SYSCALL_EXIT(code, p, uthread, error);
2d21ac55	382
6d2010ae A	383	#ifdef JOE_DEBUG
	384	if (uthread->uu_iocount)
	385	printf("system call returned with uu_iocount != 0\n");
	386	#endif
	387
2d21ac55 A	388	#if CONFIG_DTRACE
	389	uthread->t_dtrace_errno = error;
	390	#endif /* CONFIG_DTRACE */
0c530ab8	391
6d2010ae	392	if (__improbable(error == ERESTART)) {
0c530ab8 A	393	/*
	394	* all system calls come through via the syscall instruction
	395	* in 64 bit mode... its 2 bytes in length
	396	* move the user's pc back to repeat the syscall:
	397	*/
6d2010ae	398	pal_syscall_restart( thread, state );
0c530ab8 A	399	}
0c530ab8 A	400	else if (error != EJUSTRETURN) {
6d2010ae	401	if (__improbable(error)) {
2d21ac55	402	regs->rax = error;
0c530ab8 A	403	regs->isf.rflags \|= EFL_CF; /* carry bit */
	404	} else { /* (not error) */
	405
	406	switch (callp->sy_return_type) {
	407	case _SYSCALL_RET_INT_T:
	408	regs->rax = uthread->uu_rval[0];
	409	regs->rdx = uthread->uu_rval[1];
	410	break;
	411	case _SYSCALL_RET_UINT_T:
	412	regs->rax = ((u_int)uthread->uu_rval[0]);
	413	regs->rdx = ((u_int)uthread->uu_rval[1]);
	414	break;
	415	case _SYSCALL_RET_OFF_T:
	416	case _SYSCALL_RET_ADDR_T:
	417	case _SYSCALL_RET_SIZE_T:
	418	case _SYSCALL_RET_SSIZE_T:
d1ecb069	419	case _SYSCALL_RET_UINT64_T:
0c530ab8 A	420	regs->rax = ((uint64_t )(&uthread->uu_rval[0]));
	421	regs->rdx = 0;
	422	break;
	423	case _SYSCALL_RET_NONE:
	424	break;
	425	default:
	426	panic("unix_syscall: unknown return type");
	427	break;
	428	}
	429	regs->isf.rflags &= ~EFL_CF;
	430	}
	431	}
	432
b0d623f7 A	433	DEBUG_KPRINT_SYSCALL_UNIX(
	434	"unix_syscall64: error=%d retval=(%llu,%llu)\n",
	435	error, regs->rax, regs->rdx);
	436
2d21ac55	437	uthread->uu_flag &= ~UT_NOTCANCELPT;
0c530ab8	438
6d2010ae	439	if (__improbable(uthread->uu_lowpri_window)) {
0c530ab8 A	440	/*
	441	* task is marked as a low priority I/O type
	442	* and the I/O we issued while in this system call
	443	* collided with normal I/O operations... we'll
	444	* delay in order to mitigate the impact of this
	445	* task on the normal operation of the system
	446	*/
39236c6e	447	throttle_lowpri_io(1);
0c530ab8	448	}
a1c7dba1	449	if (__probable(!code_is_kdebug_trace(code)))
316670eb A	450	KERNEL_DEBUG_CONSTANT_IST(KDEBUG_TRACE,
316670eb A	451	BSDDBG_CODE(DBG_BSD_EXCP_SC, code) \| DBG_FUNC_END,
743345f9	452	error, uthread->uu_rval[0], uthread->uu_rval[1], pid, 0);
0c530ab8	453
3e170ce0 A	454	#if PROC_REF_DEBUG
	455	if (__improbable(uthread_get_proc_refcount(uthread))) {
	456	panic("system call returned with uu_proc_refcount != 0");
	457	}
	458	#endif
	459
0c530ab8 A	460	thread_exception_return();
	461	/* NOTREACHED */
	462	}
	463
	464
	465	void
	466	unix_syscall_return(int error)
	467	{
	468	thread_t thread;
	469	struct uthread *uthread;
	470	struct proc *p;
	471	unsigned int code;
0c530ab8	472	struct sysent *callp;
0c530ab8 A	473
	474	thread = current_thread();
	475	uthread = get_bsdthread_info(thread);
	476
6d2010ae	477	pal_register_cache_state(thread, DIRTY);
b0d623f7	478
0c530ab8 A	479	p = current_proc();
	480
	481	if (proc_is64bit(p)) {
2d21ac55	482	x86_saved_state64_t *regs;
0c530ab8 A	483
	484	regs = saved_state64(find_user_regs(thread));
	485
fe8ab488	486	code = uthread->syscall_code;
39037602	487	callp = (code >= nsysent) ? &sysent[SYS_invalid] : &sysent[code];
0c530ab8	488
2d21ac55 A	489	#if CONFIG_DTRACE
	490	if (callp->sy_call == dtrace_systrace_syscall)
	491	dtrace_systrace_syscall_return( code, error, uthread->uu_rval );
	492	#endif /* CONFIG_DTRACE */
b0d623f7	493	AUDIT_SYSCALL_EXIT(code, p, uthread, error);
0c530ab8 A	494
0c530ab8 A	495	if (error == ERESTART) {
2d21ac55	496	/*
6d2010ae	497	* repeat the syscall
0c530ab8	498	*/
6d2010ae	499	pal_syscall_restart( thread, find_user_regs(thread) );
0c530ab8 A	500	}
0c530ab8 A	501	else if (error != EJUSTRETURN) {
2d21ac55 A	502	if (error) {
2d21ac55 A	503	regs->rax = error;
0c530ab8 A	504	regs->isf.rflags \|= EFL_CF; /* carry bit */
	505	} else { /* (not error) */
	506
2d21ac55	507	switch (callp->sy_return_type) {
0c530ab8	508	case _SYSCALL_RET_INT_T:
2d21ac55	509	regs->rax = uthread->uu_rval[0];
0c530ab8 A	510	regs->rdx = uthread->uu_rval[1];
	511	break;
	512	case _SYSCALL_RET_UINT_T:
2d21ac55	513	regs->rax = ((u_int)uthread->uu_rval[0]);
0c530ab8 A	514	regs->rdx = ((u_int)uthread->uu_rval[1]);
	515	break;
	516	case _SYSCALL_RET_OFF_T:
	517	case _SYSCALL_RET_ADDR_T:
	518	case _SYSCALL_RET_SIZE_T:
	519	case _SYSCALL_RET_SSIZE_T:
d1ecb069	520	case _SYSCALL_RET_UINT64_T:
2d21ac55	521	regs->rax = ((uint64_t )(&uthread->uu_rval[0]));
0c530ab8 A	522	regs->rdx = 0;
	523	break;
	524	case _SYSCALL_RET_NONE:
2d21ac55	525	break;
0c530ab8	526	default:
2d21ac55	527	panic("unix_syscall: unknown return type");
0c530ab8 A	528	break;
	529	}
	530	regs->isf.rflags &= ~EFL_CF;
	531	}
	532	}
b0d623f7 A	533	DEBUG_KPRINT_SYSCALL_UNIX(
	534	"unix_syscall_return: error=%d retval=(%llu,%llu)\n",
	535	error, regs->rax, regs->rdx);
0c530ab8	536	} else {
2d21ac55	537	x86_saved_state32_t *regs;
0c530ab8 A	538
	539	regs = saved_state32(find_user_regs(thread));
	540
2d21ac55	541	regs->efl &= ~(EFL_CF);
fe8ab488 A	542
fe8ab488 A	543	code = uthread->syscall_code;
39037602	544	callp = (code >= nsysent) ? &sysent[SYS_invalid] : &sysent[code];
2d21ac55 A	545
	546	#if CONFIG_DTRACE
	547	if (callp->sy_call == dtrace_systrace_syscall)
	548	dtrace_systrace_syscall_return( code, error, uthread->uu_rval );
	549	#endif /* CONFIG_DTRACE */
b0d623f7	550	AUDIT_SYSCALL_EXIT(code, p, uthread, error);
0c530ab8	551
0c530ab8	552	if (error == ERESTART) {
6d2010ae	553	pal_syscall_restart( thread, find_user_regs(thread) );
0c530ab8 A	554	}
0c530ab8 A	555	else if (error != EJUSTRETURN) {
2d21ac55 A	556	if (error) {
2d21ac55 A	557	regs->eax = error;
0c530ab8 A	558	regs->efl \|= EFL_CF; /* carry bit */
0c530ab8 A	559	} else { /* (not error) */
2d21ac55	560	regs->eax = uthread->uu_rval[0];
0c530ab8	561	regs->edx = uthread->uu_rval[1];
0c530ab8 A	562	}
0c530ab8 A	563	}
b0d623f7 A	564	DEBUG_KPRINT_SYSCALL_UNIX(
	565	"unix_syscall_return: error=%d retval=(%u,%u)\n",
	566	error, regs->eax, regs->edx);
0c530ab8	567	}
0c530ab8	568
0c530ab8	569
2d21ac55	570	uthread->uu_flag &= ~UT_NOTCANCELPT;
0c530ab8	571
593a1d5f	572	if (uthread->uu_lowpri_window) {
0c530ab8 A	573	/*
	574	* task is marked as a low priority I/O type
	575	* and the I/O we issued while in this system call
	576	* collided with normal I/O operations... we'll
	577	* delay in order to mitigate the impact of this
	578	* task on the normal operation of the system
	579	*/
39236c6e	580	throttle_lowpri_io(1);
0c530ab8	581	}
a1c7dba1	582	if (!code_is_kdebug_trace(code))
316670eb A	583	KERNEL_DEBUG_CONSTANT_IST(KDEBUG_TRACE,
	584	BSDDBG_CODE(DBG_BSD_EXCP_SC, code) \| DBG_FUNC_END,
	585	error, uthread->uu_rval[0], uthread->uu_rval[1], p->p_pid, 0);
0c530ab8 A	586
	587	thread_exception_return();
	588	/* NOTREACHED */
	589	}