[net-next-2.6.git] / arch / xtensa / kernel / syscalls.c

/*
 * arch/xtensa/kernel/syscalls.c
 *
 * This file is subject to the terms and conditions of the GNU General Public
 * License.  See the file "COPYING" in the main directory of this archive
 * for more details.
 *
 * Copyright (C) 2001 - 2005 Tensilica Inc.
 * Copyright (C) 2000 Silicon Graphics, Inc.
 * Copyright (C) 1995 - 2000 by Ralf Baechle
 *
 * Joe Taylor <joe@tensilica.com, joetylr@yahoo.com>
 * Marc Gauthier <marc@tensilica.com, marc@alumni.uwaterloo.ca>
 * Chris Zankel <chris@zankel.net>
 * Kevin Chea
 *
 */

#define DEBUG	0

#include <linux/linkage.h>
#include <linux/mm.h>
#include <linux/smp.h>
#include <linux/smp_lock.h>
#include <linux/mman.h>
#include <linux/sched.h>
#include <linux/file.h>
#include <linux/slab.h>
#include <linux/utsname.h>
#include <linux/unistd.h>
#include <linux/stringify.h>
#include <linux/syscalls.h>
#include <linux/sem.h>
#include <linux/msg.h>
#include <linux/shm.h>
#include <linux/errno.h>
#include <asm/ptrace.h>
#include <asm/signal.h>
#include <asm/uaccess.h>
#include <asm/hardirq.h>
#include <asm/mman.h>
#include <asm/shmparam.h>
#include <asm/page.h>

extern void do_syscall_trace(void);
typedef int (*syscall_t)(void *a0,...);
extern syscall_t sys_call_table[];
extern unsigned char sys_narg_table[];

/*
 * sys_pipe() is the normal C calling standard for creating a pipe. It's not
 * the way unix traditional does this, though.
 */

int sys_pipe(int __user *userfds)
{
	int fd[2];
	int error;

	error = do_pipe(fd);
	if (!error) {
		if (copy_to_user(userfds, fd, 2 * sizeof(int)))
			error = -EFAULT;
	}
	return error;
}

/*
 * Common code for old and new mmaps.
 */
long sys_mmap(unsigned long addr, unsigned long len, unsigned long prot,
	      unsigned long flags, unsigned long fd, unsigned long pgoff)
{
	int error = -EBADF;
	struct file * file = NULL;

	flags &= ~(MAP_EXECUTABLE | MAP_DENYWRITE);
	if (!(flags & MAP_ANONYMOUS)) {
		file = fget(fd);
		if (!file)
			goto out;
	}

	down_write(&current->mm->mmap_sem);
	error = do_mmap_pgoff(file, addr, len, prot, flags, pgoff);
	up_write(&current->mm->mmap_sem);

	if (file)
		fput(file);
out:
	return error;
}

int sys_clone(struct pt_regs *regs)
{
	unsigned long clone_flags;
	unsigned long newsp;
	int __user *parent_tidptr, *child_tidptr;
	clone_flags = regs->areg[4];
	newsp = regs->areg[3];
	parent_tidptr = (int __user *)regs->areg[5];
	child_tidptr = (int __user *)regs->areg[6];
	if (!newsp)
		newsp = regs->areg[1];
	return do_fork(clone_flags,newsp,regs,0,parent_tidptr,child_tidptr);
}

/*
 * sys_execve() executes a new program.
 */

int sys_execve(struct pt_regs *regs)
{
	int error;
	char * filename;

	filename = getname((char *) (long)regs->areg[5]);
	error = PTR_ERR(filename);
	if (IS_ERR(filename))
		goto out;
	error = do_execve(filename, (char **) (long)regs->areg[3],
	                  (char **) (long)regs->areg[4], regs);
	putname(filename);

out:
	return error;
}

int sys_uname(struct old_utsname * name)
{
	if (name && !copy_to_user(name, utsname(), sizeof (*name)))
		return 0;
	return -EFAULT;
}

/*
 * Build the string table for the builtin "poor man's strace".
 */

#if DEBUG
#define SYSCALL(fun, narg) #fun,
static char *sfnames[] = {
#include "syscalls.h"
};
#undef SYS
#endif

void system_call (struct pt_regs *regs)
{
	syscall_t syscall;
	unsigned long parm0, parm1, parm2, parm3, parm4, parm5;
	int nargs, res;
	unsigned int syscallnr;
	int ps;

#if DEBUG
	int i;
	unsigned long parms[6];
	char *sysname;
#endif

	regs->syscall = regs->areg[2];

	do_syscall_trace();

	/* Have to load after syscall_trace because strace
	 * sometimes changes regs->syscall.
	 */
	syscallnr = regs->syscall;

	parm0 = parm1 = parm2 = parm3 = parm4 = parm5 = 0;

	/* Restore interrupt level to syscall invoker's.
	 * If this were in assembly, we wouldn't disable
	 * interrupts in the first place:
	 */
	local_save_flags (ps);
	local_irq_restore((ps & ~PS_INTLEVEL_MASK) |
			  (regs->ps & PS_INTLEVEL_MASK) );

	if (syscallnr > __NR_Linux_syscalls) {
		regs->areg[2] = -ENOSYS;
		return;
	}

	syscall = sys_call_table[syscallnr];
	nargs = sys_narg_table[syscallnr];

	if (syscall == NULL) {
		regs->areg[2] = -ENOSYS;
		return;
	}

	/* There shouldn't be more than six arguments in the table! */

	if (nargs > 6)
		panic("Internal error - too many syscall arguments (%d)!\n",
		      nargs);

	/* Linux takes system-call arguments in registers.  The ABI
         * and Xtensa software conventions require the system-call
         * number in a2.  If an argument exists in a2, we move it to
         * the next available register.  Note that for improved
         * efficiency, we do NOT shift all parameters down one
         * register to maintain the original order.
	 *
         * At best case (zero arguments), we just write the syscall
         * number to a2.  At worst case (1 to 6 arguments), we move
         * the argument in a2 to the next available register, then
         * write the syscall number to a2.
	 *
         * For clarity, the following truth table enumerates all
         * possibilities.
	 *
         * arguments	syscall number	arg0, arg1, arg2, arg3, arg4, arg5
         * ---------	--------------	----------------------------------
	 *	0	      a2
	 *	1	      a2	a3
	 *	2	      a2	a4,   a3
	 *	3	      a2	a5,   a3,   a4
	 *	4	      a2	a6,   a3,   a4,   a5
	 *	5	      a2	a7,   a3,   a4,   a5,   a6
	 *	6	      a2	a8,   a3,   a4,   a5,   a6,   a7
	 */
	if (nargs) {
		parm0 = regs->areg[nargs+2];
		parm1 = regs->areg[3];
		parm2 = regs->areg[4];
		parm3 = regs->areg[5];
		parm4 = regs->areg[6];
		parm5 = regs->areg[7];
	} else /* nargs == 0 */
		parm0 = (unsigned long) regs;

#if DEBUG
	parms[0] = parm0;
	parms[1] = parm1;
	parms[2] = parm2;
	parms[3] = parm3;
	parms[4] = parm4;
	parms[5] = parm5;

	sysname = sfnames[syscallnr];
	if (strncmp(sysname, "sys_", 4) == 0)
		sysname = sysname + 4;

	printk("\017SYSCALL:I:%x:%d:%s  %s(", regs->pc, current->pid,
	       current->comm, sysname);
	for (i = 0; i < nargs; i++)
		printk((i>0) ? ", %#lx" : "%#lx", parms[i]);
	printk(")\n");
#endif

	res = syscall((void *)parm0, parm1, parm2, parm3, parm4, parm5);

#if DEBUG
	printk("\017SYSCALL:O:%d:%s  %s(",current->pid, current->comm, sysname);
	for (i = 0; i < nargs; i++)
		printk((i>0) ? ", %#lx" : "%#lx", parms[i]);
	if (res < 4096)
		printk(") = %d\n", res);
	else
		printk(") = %#x\n", res);
#endif /* DEBUG */

	regs->areg[2] = res;
	do_syscall_trace();
}

/*
 * Do a system call from kernel instead of calling sys_execve so we
 * end up with proper pt_regs.
 */
int kernel_execve(const char *filename, char *const argv[], char *const envp[])
{
	long __res;
	asm volatile (
		"  mov   a5, %2 \n"
		"  mov   a4, %4 \n"
		"  mov   a3, %3 \n"
		"  movi  a2, %1 \n"
		"  syscall      \n"
		"  mov   %0, a2 \n"
		: "=a" (__res)
		: "i" (__NR_execve), "a" (filename), "a" (argv), "a" (envp)
		: "a2", "a3", "a4", "a5");
	return __res;
}
Commit	Line	Data
5a0015d6	1	/*
f30c2269	2	* arch/xtensa/kernel/syscalls.c
5a0015d6 CZ	3	*
	4	* This file is subject to the terms and conditions of the GNU General Public
	5	* License. See the file "COPYING" in the main directory of this archive
	6	* for more details.
	7	*
	8	* Copyright (C) 2001 - 2005 Tensilica Inc.
	9	* Copyright (C) 2000 Silicon Graphics, Inc.
	10	* Copyright (C) 1995 - 2000 by Ralf Baechle
	11	*
	12	* Joe Taylor <joe@tensilica.com, joetylr@yahoo.com>
	13	* Marc Gauthier <marc@tensilica.com, marc@alumni.uwaterloo.ca>
	14	* Chris Zankel <chris@zankel.net>
	15	* Kevin Chea
	16	*
	17	*/
	18
	19	#define DEBUG 0
	20
5a0015d6 CZ	21	#include <linux/linkage.h>
	22	#include <linux/mm.h>
	23	#include <linux/smp.h>
	24	#include <linux/smp_lock.h>
	25	#include <linux/mman.h>
	26	#include <linux/sched.h>
	27	#include <linux/file.h>
	28	#include <linux/slab.h>
	29	#include <linux/utsname.h>
	30	#include <linux/unistd.h>
	31	#include <linux/stringify.h>
	32	#include <linux/syscalls.h>
	33	#include <linux/sem.h>
	34	#include <linux/msg.h>
	35	#include <linux/shm.h>
	36	#include <linux/errno.h>
	37	#include <asm/ptrace.h>
	38	#include <asm/signal.h>
	39	#include <asm/uaccess.h>
	40	#include <asm/hardirq.h>
	41	#include <asm/mman.h>
	42	#include <asm/shmparam.h>
	43	#include <asm/page.h>
5a0015d6 CZ	44
	45	extern void do_syscall_trace(void);
	46	typedef int (syscall_t)(void a0,...);
5a0015d6 CZ	47	extern syscall_t sys_call_table[];
	48	extern unsigned char sys_narg_table[];
	49
	50	/*
	51	* sys_pipe() is the normal C calling standard for creating a pipe. It's not
	52	* the way unix traditional does this, though.
	53	*/
	54
	55	int sys_pipe(int __user *userfds)
	56	{
	57	int fd[2];
	58	int error;
	59
	60	error = do_pipe(fd);
	61	if (!error) {
	62	if (copy_to_user(userfds, fd, 2 * sizeof(int)))
	63	error = -EFAULT;
	64	}
	65	return error;
	66	}
	67
	68	/*
	69	* Common code for old and new mmaps.
	70	*/
813e6783 CZ	71	long sys_mmap(unsigned long addr, unsigned long len, unsigned long prot,
813e6783 CZ	72	unsigned long flags, unsigned long fd, unsigned long pgoff)
5a0015d6 CZ	73	{
	74	int error = -EBADF;
	75	struct file * file = NULL;
	76
	77	flags &= ~(MAP_EXECUTABLE \| MAP_DENYWRITE);
	78	if (!(flags & MAP_ANONYMOUS)) {
	79	file = fget(fd);
	80	if (!file)
	81	goto out;
	82	}
	83
	84	down_write(&current->mm->mmap_sem);
	85	error = do_mmap_pgoff(file, addr, len, prot, flags, pgoff);
	86	up_write(&current->mm->mmap_sem);
	87
	88	if (file)
	89	fput(file);
	90	out:
	91	return error;
	92	}
	93
5a0015d6 CZ	94	int sys_clone(struct pt_regs *regs)
	95	{
	96	unsigned long clone_flags;
	97	unsigned long newsp;
	98	int __user parent_tidptr, child_tidptr;
	99	clone_flags = regs->areg[4];
	100	newsp = regs->areg[3];
	101	parent_tidptr = (int __user *)regs->areg[5];
	102	child_tidptr = (int __user *)regs->areg[6];
	103	if (!newsp)
	104	newsp = regs->areg[1];
	105	return do_fork(clone_flags,newsp,regs,0,parent_tidptr,child_tidptr);
	106	}
	107
	108	/*
	109	* sys_execve() executes a new program.
	110	*/
	111
	112	int sys_execve(struct pt_regs *regs)
	113	{
	114	int error;
	115	char * filename;
	116
	117	filename = getname((char *) (long)regs->areg[5]);
	118	error = PTR_ERR(filename);
	119	if (IS_ERR(filename))
	120	goto out;
	121	error = do_execve(filename, (char **) (long)regs->areg[3],
	122	(char **) (long)regs->areg[4], regs);
	123	putname(filename);
	124
	125	out:
	126	return error;
	127	}
	128
	129	int sys_uname(struct old_utsname * name)
	130	{
e9ff3990	131	if (name && !copy_to_user(name, utsname(), sizeof (*name)))
5a0015d6 CZ	132	return 0;
	133	return -EFAULT;
	134	}
	135
5a0015d6 CZ	136	/*
	137	* Build the string table for the builtin "poor man's strace".
	138	*/
	139
	140	#if DEBUG
	141	#define SYSCALL(fun, narg) #fun,
	142	static char *sfnames[] = {
	143	#include "syscalls.h"
	144	};
	145	#undef SYS
	146	#endif
	147
	148	void system_call (struct pt_regs *regs)
	149	{
	150	syscall_t syscall;
	151	unsigned long parm0, parm1, parm2, parm3, parm4, parm5;
	152	int nargs, res;
	153	unsigned int syscallnr;
	154	int ps;
	155
	156	#if DEBUG
	157	int i;
	158	unsigned long parms[6];
	159	char *sysname;
	160	#endif
	161
	162	regs->syscall = regs->areg[2];
	163
	164	do_syscall_trace();
	165
	166	/* Have to load after syscall_trace because strace
	167	* sometimes changes regs->syscall.
	168	*/
	169	syscallnr = regs->syscall;
	170
	171	parm0 = parm1 = parm2 = parm3 = parm4 = parm5 = 0;
	172
	173	/* Restore interrupt level to syscall invoker's.
	174	* If this were in assembly, we wouldn't disable
	175	* interrupts in the first place:
	176	*/
	177	local_save_flags (ps);
173d6681 CZ	178	local_irq_restore((ps & ~PS_INTLEVEL_MASK) \|
173d6681 CZ	179	(regs->ps & PS_INTLEVEL_MASK) );
5a0015d6 CZ	180
	181	if (syscallnr > __NR_Linux_syscalls) {
	182	regs->areg[2] = -ENOSYS;
	183	return;
	184	}
	185
	186	syscall = sys_call_table[syscallnr];
	187	nargs = sys_narg_table[syscallnr];
	188
	189	if (syscall == NULL) {
	190	regs->areg[2] = -ENOSYS;
	191	return;
	192	}
	193
	194	/* There shouldn't be more than six arguments in the table! */
	195
	196	if (nargs > 6)
	197	panic("Internal error - too many syscall arguments (%d)!\n",
	198	nargs);
	199
	200	/* Linux takes system-call arguments in registers. The ABI
	201	* and Xtensa software conventions require the system-call
	202	* number in a2. If an argument exists in a2, we move it to
	203	* the next available register. Note that for improved
	204	* efficiency, we do NOT shift all parameters down one
	205	* register to maintain the original order.
	206	*
	207	* At best case (zero arguments), we just write the syscall
	208	* number to a2. At worst case (1 to 6 arguments), we move
	209	* the argument in a2 to the next available register, then
	210	* write the syscall number to a2.
	211	*
	212	* For clarity, the following truth table enumerates all
	213	* possibilities.
	214	*
	215	* arguments syscall number arg0, arg1, arg2, arg3, arg4, arg5
	216	* --------- -------------- ----------------------------------
	217	* 0 a2
	218	* 1 a2 a3
	219	* 2 a2 a4, a3
	220	* 3 a2 a5, a3, a4
	221	* 4 a2 a6, a3, a4, a5
	222	* 5 a2 a7, a3, a4, a5, a6
	223	* 6 a2 a8, a3, a4, a5, a6, a7
	224	*/
	225	if (nargs) {
	226	parm0 = regs->areg[nargs+2];
	227	parm1 = regs->areg[3];
	228	parm2 = regs->areg[4];
	229	parm3 = regs->areg[5];
	230	parm4 = regs->areg[6];
	231	parm5 = regs->areg[7];
	232	} else /* nargs == 0 */
	233	parm0 = (unsigned long) regs;
	234
	235	#if DEBUG
	236	parms[0] = parm0;
	237	parms[1] = parm1;
	238	parms[2] = parm2;
	239	parms[3] = parm3;
	240	parms[4] = parm4;
	241	parms[5] = parm5;
	242
	243	sysname = sfnames[syscallnr];
244	if (strncmp(sysname, "sys_", 4) == 0)
245	sysname = sysname + 4;
246
247	printk("\017SYSCALL:I:%x:%d:%s %s(", regs->pc, current->pid,
248	current->comm, sysname);
249	for (i = 0; i < nargs; i++)
250	printk((i>0) ? ", %#lx" : "%#lx", parms[i]);
251	printk(")\n");
252	#endif
253
254	res = syscall((void *)parm0, parm1, parm2, parm3, parm4, parm5);
255
256	#if DEBUG
257	printk("\017SYSCALL:O:%d:%s %s(",current->pid, current->comm, sysname);
258	for (i = 0; i < nargs; i++)
259	printk((i>0) ? ", %#lx" : "%#lx", parms[i]);
260	if (res < 4096)
261	printk(") = %d\n", res);
262	else
263	printk(") = %#x\n", res);
264	#endif /* DEBUG */
265
266	regs->areg[2] = res;
267	do_syscall_trace();
268	}
fe74290d AB	269
	270	/*
	271	* Do a system call from kernel instead of calling sys_execve so we
	272	* end up with proper pt_regs.
	273	*/
	274	int kernel_execve(const char filename, char const argv[], char *const envp[])
	275	{
	276	long __res;
	277	asm volatile (
	278	" mov a5, %2 \n"
	279	" mov a4, %4 \n"
	280	" mov a3, %3 \n"
	281	" movi a2, %1 \n"
	282	" syscall \n"
	283	" mov %0, a2 \n"
	284	: "=a" (__res)
	285	: "i" (__NR_execve), "a" (filename), "a" (argv), "a" (envp)
	286	: "a2", "a3", "a4", "a5");
	287	return __res;
	288	}