Split the linux syscall out to make the code easier to read and maintain.

This commit is contained in:
christos 2005-11-01 16:28:28 +00:00
parent 4f38beeb43
commit f1e915cc01
2 changed files with 268 additions and 75 deletions

View File

@ -1,7 +1,11 @@
/* $NetBSD: linux_syscall.c,v 1.2 2005/05/15 21:44:41 fvdl Exp $ */
/* $NetBSD: linux_syscall.c,v 1.3 2005/11/01 16:28:28 christos Exp $ */
/*-
* Copyright (c) 2005 Emmanuel Dreyfus, all rights reserved.
* Copyright (c) 1998, 2000 The NetBSD Foundation, Inc.
* All rights reserved.
*
* This code is derived from software contributed to The NetBSD Foundation
* by Charles M. Hannum.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
@ -13,15 +17,16 @@
* documentation and/or other materials provided with the distribution.
* 3. All advertising materials mentioning features or use of this software
* must display the following acknowledgement:
* This product includes software developed by Emmanuel Dreyfus
* 4. The name of the author may not be used to endorse or promote
* products derived from this software without specific prior written
* permission.
* This product includes software developed by the NetBSD
* Foundation, Inc. and its contributors.
* 4. Neither the name of The NetBSD Foundation nor the names of its
* contributors may be used to endorse or promote products derived
* from this software without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE THE AUTHOR AND CONTRIBUTORS ``AS IS''
* AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO,
* THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
* PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS
* THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS
* ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
* TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
* PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS
* BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
* CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
* SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
@ -31,16 +36,35 @@
* POSSIBILITY OF SUCH DAMAGE.
*/
#include "opt_compat_linux.h"
#include <sys/cdefs.h>
__KERNEL_RCSID(0, "$NetBSD: linux_syscall.c,v 1.3 2005/11/01 16:28:28 christos Exp $");
__KERNEL_RCSID(0, "$NetBSD: linux_syscall.c,v 1.2 2005/05/15 21:44:41 fvdl Exp $");
#include "opt_syscall_debug.h"
#include "opt_ktrace.h"
#include "opt_systrace.h"
#include "opt_compat_linux.h"
#include <sys/types.h>
#include <sys/param.h>
#include <sys/systm.h>
#include <sys/proc.h>
#include <sys/user.h>
#include <sys/signal.h>
#include <sys/proc.h>
#include <sys/sa.h>
#include <sys/savar.h>
#ifdef KTRACE
#include <sys/ktrace.h>
#endif
#ifdef SYSTRACE
#include <sys/systrace.h>
#endif
#include <sys/syscall.h>
#include <uvm/uvm_extern.h>
#include <machine/cpu.h>
#include <machine/psl.h>
#include <machine/userret.h>
#include <compat/linux/linux_syscall.h>
#include <compat/linux/common/linux_types.h>
#include <compat/linux/common/linux_errno.h>
@ -49,8 +73,211 @@ __KERNEL_RCSID(0, "$NetBSD: linux_syscall.c,v 1.2 2005/05/15 21:44:41 fvdl Exp $
#include <compat/linux/arch/amd64/linux_siginfo.h>
#include <compat/linux/arch/amd64/linux_syscall.h>
#include <compat/linux/arch/amd64/linux_machdep.h>
#include <compat/linux/common/linux_errno.h>
#define EMULNAME(x) __CONCAT(linux_,x)
#define EMULNAMEU(x) __CONCAT(LINUX_,x)
void linux_syscall_intern(struct proc *);
static void linux_syscall_plain(struct trapframe *);
static void linux_syscall_fancy(struct trapframe *);
#include "syscall.c"
void
linux_syscall_intern(struct proc *p)
{
#ifdef KTRACE
if (p->p_traceflag & (KTRFAC_SYSCALL | KTRFAC_SYSRET)) {
p->p_md.md_syscall = linux_syscall_fancy;
return;
}
#endif
#ifdef SYSTRACE
if (ISSET(p->p_flag, P_SYSTRACE)) {
p->p_md.md_syscall = linux_syscall_fancy;
return;
}
#endif
p->p_md.md_syscall = linux_syscall_plain;
}
/*
* syscall(frame):
* System call request from POSIX system call gate interface to kernel.
* Like trap(), argument is call by reference.
*/
static void
linux_syscall_plain(struct trapframe *frame)
{
caddr_t params;
const struct sysent *callp;
struct proc *p;
struct lwp *l;
int error;
size_t argsize, argoff;
register_t code, args[9], rval[2], *argp;
uvmexp.syscalls++;
l = curlwp;
p = l->l_proc;
code = frame->tf_rax;
callp = p->p_emul->e_sysent;
argoff = 0;
argp = &args[0];
code &= (LINUX_SYS_NSYSENT - 1);
callp += code;
argsize = (callp->sy_argsize >> 3) + argoff;
if (argsize) {
switch (MIN(argsize, 6)) {
case 6:
args[5] = frame->tf_r9;
case 5:
args[4] = frame->tf_r8;
case 4:
args[3] = frame->tf_r10;
case 3:
args[2] = frame->tf_rdx;
case 2:
args[1] = frame->tf_rsi;
case 1:
args[0] = frame->tf_rdi;
break;
default:
panic("impossible syscall argsize");
}
if (argsize > 6) {
argsize -= 6;
params = (caddr_t)frame->tf_rsp + sizeof(register_t);
error = copyin(params, (caddr_t)&args[6],
argsize << 3);
if (error != 0)
goto bad;
}
}
#ifdef SYSCALL_DEBUG
scdebug_call(l, code, argp);
#endif /* SYSCALL_DEBUG */
rval[0] = 0;
rval[1] = 0;
KERNEL_PROC_LOCK(l);
error = (*callp->sy_call)(l, argp, rval);
KERNEL_PROC_UNLOCK(l);
switch (error) {
case 0:
frame->tf_rax = rval[0];
frame->tf_rflags &= ~PSL_C; /* carry bit */
break;
case ERESTART:
/*
* The offset to adjust the PC by depends on whether we entered
* the kernel through the trap or call gate. We pushed the
* size of the instruction into tf_err on entry.
*/
frame->tf_rip -= frame->tf_err;
break;
case EJUSTRETURN:
/* nothing to do */
break;
default:
bad:
frame->tf_rax = native_to_linux_errno[error];
frame->tf_rflags |= PSL_C; /* carry bit */
break;
}
#ifdef SYSCALL_DEBUG
scdebug_ret(l, code, error, rval);
#endif /* SYSCALL_DEBUG */
userret(l);
}
static void
linux_syscall_fancy(struct trapframe *frame)
{
caddr_t params;
const struct sysent *callp;
struct proc *p;
struct lwp *l;
int error;
size_t argsize, argoff;
register_t code, args[9], rval[2], *argp;
uvmexp.syscalls++;
l = curlwp;
p = l->l_proc;
code = frame->tf_rax;
callp = p->p_emul->e_sysent;
argp = &args[0];
argoff = 0;
code &= (SYS_NSYSENT - 1);
callp += code;
argsize = (callp->sy_argsize >> 3) + argoff;
if (argsize) {
switch (MIN(argsize, 6)) {
case 6:
args[5] = frame->tf_r9;
case 5:
args[4] = frame->tf_r8;
case 4:
args[3] = frame->tf_r10;
case 3:
args[2] = frame->tf_rdx;
case 2:
args[1] = frame->tf_rsi;
case 1:
args[0] = frame->tf_rdi;
break;
default:
panic("impossible syscall argsize");
}
if (argsize > 6) {
argsize -= 6;
params = (caddr_t)frame->tf_rsp + sizeof(register_t);
error = copyin(params, (caddr_t)&args[6],
argsize << 3);
if (error != 0)
goto bad;
}
}
KERNEL_PROC_LOCK(l);
if ((error = trace_enter(l, code, code, NULL, argp)) != 0)
goto out;
rval[0] = 0;
rval[1] = 0;
error = (*callp->sy_call)(l, argp, rval);
out:
KERNEL_PROC_UNLOCK(l);
switch (error) {
case 0:
frame->tf_rax = rval[0];
frame->tf_rflags &= ~PSL_C; /* carry bit */
break;
case ERESTART:
/*
* The offset to adjust the PC by depends on whether we entered
* the kernel through the trap or call gate. We pushed the
* size of the instruction into tf_err on entry.
*/
frame->tf_rip -= frame->tf_err;
break;
case EJUSTRETURN:
/* nothing to do */
break;
default:
bad:
frame->tf_rax = native_to_linux_errno[error];
frame->tf_rflags |= PSL_C; /* carry bit */
break;
}
trace_exit(l, code, argp, rval, error);
userret(l);
}

View File

@ -1,4 +1,4 @@
/* $NetBSD: syscall.c,v 1.10 2005/11/01 09:13:48 manu Exp $ */
/* $NetBSD: syscall.c,v 1.11 2005/11/01 16:28:28 christos Exp $ */
/*-
* Copyright (c) 1998, 2000 The NetBSD Foundation, Inc.
@ -37,7 +37,7 @@
*/
#include <sys/cdefs.h>
__KERNEL_RCSID(0, "$NetBSD: syscall.c,v 1.10 2005/11/01 09:13:48 manu Exp $");
__KERNEL_RCSID(0, "$NetBSD: syscall.c,v 1.11 2005/11/01 16:28:28 christos Exp $");
#include "opt_syscall_debug.h"
#include "opt_ktrace.h"
@ -56,6 +56,7 @@ __KERNEL_RCSID(0, "$NetBSD: syscall.c,v 1.10 2005/11/01 09:13:48 manu Exp $");
#ifdef SYSTRACE
#include <sys/systrace.h>
#endif
#include <sys/syscall.h>
#include <uvm/uvm_extern.h>
@ -63,20 +64,12 @@ __KERNEL_RCSID(0, "$NetBSD: syscall.c,v 1.10 2005/11/01 09:13:48 manu Exp $");
#include <machine/psl.h>
#include <machine/userret.h>
#ifdef COMPAT_LINUX
#include <compat/linux/common/linux_errno.h>
#endif
#ifndef EMULNAME
#include <sys/syscall.h>
#define EMULNAME(x) (x)
#define EMULNAMEU(x) (x)
#include <sys/syscall.h>
void syscall_intern(struct proc *);
static void syscall_plain(struct trapframe *);
static void syscall_fancy(struct trapframe *);
void
child_return(arg)
void *arg;
child_return(void *arg)
{
struct lwp *l = arg;
struct trapframe *tf = l->l_md.md_regs;
@ -93,35 +86,28 @@ child_return(arg)
#ifdef KTRACE
if (KTRPOINT(p, KTR_SYSRET)) {
KERNEL_PROC_LOCK(l);
ktrsysret(p, EMULNAMEU(SYS_fork), 0, 0);
ktrsysret(p, SYS_fork, 0, 0);
KERNEL_PROC_UNLOCK(l);
}
#endif
}
#endif /* EMULNAME */
void EMULNAME(syscall_intern) __P((struct proc *));
void EMULNAME(syscall_plain) __P((struct trapframe *));
void EMULNAME(syscall_fancy) __P((struct trapframe *));
void
EMULNAME(syscall_intern)(p) /*
syscall_intern(p) */
struct proc *p;
syscall_intern(struct proc *p)
{
#ifdef KTRACE
if (p->p_traceflag & (KTRFAC_SYSCALL | KTRFAC_SYSRET)) {
p->p_md.md_syscall = EMULNAME(syscall_fancy);
p->p_md.md_syscall = syscall_fancy;
return;
}
#endif
#ifdef SYSTRACE
if (ISSET(p->p_flag, P_SYSTRACE)) {
p->p_md.md_syscall = EMULNAME(syscall_fancy);
p->p_md.md_syscall = syscall_fancy;
return;
}
#endif
p->p_md.md_syscall = EMULNAME(syscall_plain);
p->p_md.md_syscall = syscall_plain;
}
/*
@ -129,10 +115,8 @@ syscall_intern(p) */
* System call request from POSIX system call gate interface to kernel.
* Like trap(), argument is call by reference.
*/
void
EMULNAME(syscall_plain)(frame) /*
syscall_plain(frame) */
struct trapframe *frame;
static void
syscall_plain(struct trapframe *frame)
{
caddr_t params;
const struct sysent *callp;
@ -151,10 +135,9 @@ syscall_plain(frame) */
argoff = 0;
argp = &args[0];
#ifndef COMPAT_LINUX
switch (code) {
case EMULNAMEU(SYS_syscall):
case EMULNAMEU(SYS___syscall):
case SYS_syscall:
case SYS___syscall:
/*
* Code is first argument, followed by actual args.
*/
@ -165,9 +148,8 @@ syscall_plain(frame) */
default:
break;
}
#endif /* !COMPAT_LINUX */
code &= (EMULNAMEU(SYS_NSYSENT) - 1);
code &= (SYS_NSYSENT - 1);
callp += code;
argsize = (callp->sy_argsize >> 3) + argoff;
@ -212,9 +194,7 @@ syscall_plain(frame) */
switch (error) {
case 0:
frame->tf_rax = rval[0];
#ifndef COMPAT_LINUX
frame->tf_rdx = rval[1];
#endif
frame->tf_rflags &= ~PSL_C; /* carry bit */
break;
case ERESTART:
@ -230,11 +210,7 @@ syscall_plain(frame) */
break;
default:
bad:
#ifdef COMPAT_LINUX
frame->tf_rax = native_to_linux_errno[error];
#else
frame->tf_rax = error;
#endif
frame->tf_rflags |= PSL_C; /* carry bit */
break;
}
@ -245,10 +221,8 @@ syscall_plain(frame) */
userret(l);
}
void
EMULNAME(syscall_fancy)(frame) /*
syscall_fancy(frame) */
struct trapframe *frame;
static void
syscall_fancy(struct trapframe *frame)
{
caddr_t params;
const struct sysent *callp;
@ -267,10 +241,9 @@ syscall_fancy(frame) */
argp = &args[0];
argoff = 0;
#ifndef COMPAT_LINUX
switch (code) {
case EMULNAMEU(SYS_syscall):
case EMULNAMEU(SYS___syscall):
case SYS_syscall:
case SYS___syscall:
/*
* Code is first argument, followed by actual args.
*/
@ -280,9 +253,9 @@ syscall_fancy(frame) */
break;
default:
break;
}
#endif /* !COMPAT_LINUX */
code &= (EMULNAMEU(SYS_NSYSENT) - 1);
code &= (SYS_NSYSENT - 1);
callp += code;
argsize = (callp->sy_argsize >> 3) + argoff;
@ -326,9 +299,7 @@ out:
switch (error) {
case 0:
frame->tf_rax = rval[0];
#ifndef COMPAT_LINUX
frame->tf_rdx = rval[1];
#endif
frame->tf_rflags &= ~PSL_C; /* carry bit */
break;
case ERESTART:
@ -344,11 +315,7 @@ out:
break;
default:
bad:
#ifdef COMPAT_LINUX
frame->tf_rax = native_to_linux_errno[error];
#else
frame->tf_rax = error;
#endif
frame->tf_rflags |= PSL_C; /* carry bit */
break;
}
@ -357,4 +324,3 @@ out:
userret(l);
}