[deliverable/linux.git] / include / asm-i386 / string.h

#ifndef _I386_STRING_H_
#define _I386_STRING_H_

#ifdef __KERNEL__
/*
 * On a 486 or Pentium, we are better off not using the
 * byte string operations. But on a 386 or a PPro the
 * byte string ops are faster than doing it by hand
 * (MUCH faster on a Pentium).
 */

/*
 * This string-include defines all string functions as inline
 * functions. Use gcc. It also assumes ds=es=data space, this should be
 * normal. Most of the string-functions are rather heavily hand-optimized,
 * see especially strsep,strstr,str[c]spn. They should work, but are not
 * very easy to understand. Everything is done entirely within the register
 * set, making the functions fast and clean. String instructions have been
 * used through-out, making for "slightly" unclear code :-)
 *
 *		NO Copyright (C) 1991, 1992 Linus Torvalds,
 *		consider these trivial functions to be PD.
 */

/* AK: in fact I bet it would be better to move this stuff all out of line.
 */

#define __HAVE_ARCH_STRCPY
static inline char * strcpy(char * dest,const char *src)
{
int d0, d1, d2;
__asm__ __volatile__(
	"1:\tlodsb\n\t"
	"stosb\n\t"
	"testb %%al,%%al\n\t"
	"jne 1b"
	: "=&S" (d0), "=&D" (d1), "=&a" (d2)
	:"0" (src),"1" (dest) : "memory");
return dest;
}

#define __HAVE_ARCH_STRNCPY
static inline char * strncpy(char * dest,const char *src,size_t count)
{
int d0, d1, d2, d3;
__asm__ __volatile__(
	"1:\tdecl %2\n\t"
	"js 2f\n\t"
	"lodsb\n\t"
	"stosb\n\t"
	"testb %%al,%%al\n\t"
	"jne 1b\n\t"
	"rep\n\t"
	"stosb\n"
	"2:"
	: "=&S" (d0), "=&D" (d1), "=&c" (d2), "=&a" (d3)
	:"0" (src),"1" (dest),"2" (count) : "memory");
return dest;
}

#define __HAVE_ARCH_STRCAT
static inline char * strcat(char * dest,const char * src)
{
int d0, d1, d2, d3;
__asm__ __volatile__(
	"repne\n\t"
	"scasb\n\t"
	"decl %1\n"
	"1:\tlodsb\n\t"
	"stosb\n\t"
	"testb %%al,%%al\n\t"
	"jne 1b"
	: "=&S" (d0), "=&D" (d1), "=&a" (d2), "=&c" (d3)
	: "0" (src), "1" (dest), "2" (0), "3" (0xffffffffu):"memory");
return dest;
}

#define __HAVE_ARCH_STRNCAT
static inline char * strncat(char * dest,const char * src,size_t count)
{
int d0, d1, d2, d3;
__asm__ __volatile__(
	"repne\n\t"
	"scasb\n\t"
	"decl %1\n\t"
	"movl %8,%3\n"
	"1:\tdecl %3\n\t"
	"js 2f\n\t"
	"lodsb\n\t"
	"stosb\n\t"
	"testb %%al,%%al\n\t"
	"jne 1b\n"
	"2:\txorl %2,%2\n\t"
	"stosb"
	: "=&S" (d0), "=&D" (d1), "=&a" (d2), "=&c" (d3)
	: "0" (src),"1" (dest),"2" (0),"3" (0xffffffffu), "g" (count)
	: "memory");
return dest;
}

#define __HAVE_ARCH_STRCMP
static inline int strcmp(const char * cs,const char * ct)
{
int d0, d1;
register int __res;
__asm__ __volatile__(
	"1:\tlodsb\n\t"
	"scasb\n\t"
	"jne 2f\n\t"
	"testb %%al,%%al\n\t"
	"jne 1b\n\t"
	"xorl %%eax,%%eax\n\t"
	"jmp 3f\n"
	"2:\tsbbl %%eax,%%eax\n\t"
	"orb $1,%%al\n"
	"3:"
	:"=a" (__res), "=&S" (d0), "=&D" (d1)
	:"1" (cs),"2" (ct)
	:"memory");
return __res;
}

#define __HAVE_ARCH_STRNCMP
static inline int strncmp(const char * cs,const char * ct,size_t count)
{
register int __res;
int d0, d1, d2;
__asm__ __volatile__(
	"1:\tdecl %3\n\t"
	"js 2f\n\t"
	"lodsb\n\t"
	"scasb\n\t"
	"jne 3f\n\t"
	"testb %%al,%%al\n\t"
	"jne 1b\n"
	"2:\txorl %%eax,%%eax\n\t"
	"jmp 4f\n"
	"3:\tsbbl %%eax,%%eax\n\t"
	"orb $1,%%al\n"
	"4:"
	:"=a" (__res), "=&S" (d0), "=&D" (d1), "=&c" (d2)
	:"1" (cs),"2" (ct),"3" (count)
	:"memory");
return __res;
}

#define __HAVE_ARCH_STRCHR
static inline char * strchr(const char * s, int c)
{
int d0;
register char * __res;
__asm__ __volatile__(
	"movb %%al,%%ah\n"
	"1:\tlodsb\n\t"
	"cmpb %%ah,%%al\n\t"
	"je 2f\n\t"
	"testb %%al,%%al\n\t"
	"jne 1b\n\t"
	"movl $1,%1\n"
	"2:\tmovl %1,%0\n\t"
	"decl %0"
	:"=a" (__res), "=&S" (d0)
	:"1" (s),"0" (c)
	:"memory");
return __res;
}

#define __HAVE_ARCH_STRRCHR
static inline char * strrchr(const char * s, int c)
{
int d0, d1;
register char * __res;
__asm__ __volatile__(
	"movb %%al,%%ah\n"
	"1:\tlodsb\n\t"
	"cmpb %%ah,%%al\n\t"
	"jne 2f\n\t"
	"leal -1(%%esi),%0\n"
	"2:\ttestb %%al,%%al\n\t"
	"jne 1b"
	:"=g" (__res), "=&S" (d0), "=&a" (d1)
	:"0" (0),"1" (s),"2" (c)
	:"memory");
return __res;
}

#define __HAVE_ARCH_STRLEN
static inline size_t strlen(const char * s)
{
int d0;
register int __res;
__asm__ __volatile__(
	"repne\n\t"
	"scasb\n\t"
	"notl %0\n\t"
	"decl %0"
	:"=c" (__res), "=&D" (d0)
	:"1" (s),"a" (0), "0" (0xffffffffu)
	:"memory");
return __res;
}

static __always_inline void * __memcpy(void * to, const void * from, size_t n)
{
int d0, d1, d2;
__asm__ __volatile__(
	"rep ; movsl\n\t"
	"movl %4,%%ecx\n\t"
	"andl $3,%%ecx\n\t"
#if 1	/* want to pay 2 byte penalty for a chance to skip microcoded rep? */
	"jz 1f\n\t"
#endif
	"rep ; movsb\n\t"
	"1:"
	: "=&c" (d0), "=&D" (d1), "=&S" (d2)
	: "0" (n/4), "g" (n), "1" ((long) to), "2" ((long) from)
	: "memory");
return (to);
}

/*
 * This looks ugly, but the compiler can optimize it totally,
 * as the count is constant.
 */
static __always_inline void * __constant_memcpy(void * to, const void * from, size_t n)
{
	long esi, edi;
	if (!n) return to;
#if 1	/* want to do small copies with non-string ops? */
	switch (n) {
		case 1: *(char*)to = *(char*)from; return to;
		case 2: *(short*)to = *(short*)from; return to;
		case 4: *(int*)to = *(int*)from; return to;
#if 1	/* including those doable with two moves? */
		case 3: *(short*)to = *(short*)from;
			*((char*)to+2) = *((char*)from+2); return to;
		case 5: *(int*)to = *(int*)from;
			*((char*)to+4) = *((char*)from+4); return to;
		case 6: *(int*)to = *(int*)from;
			*((short*)to+2) = *((short*)from+2); return to;
		case 8: *(int*)to = *(int*)from;
			*((int*)to+1) = *((int*)from+1); return to;
#endif
	}
#endif
	esi = (long) from;
	edi = (long) to;
	if (n >= 5*4) {
		/* large block: use rep prefix */
		int ecx;
		__asm__ __volatile__(
			"rep ; movsl"
			: "=&c" (ecx), "=&D" (edi), "=&S" (esi)
			: "0" (n/4), "1" (edi),"2" (esi)
			: "memory"
		);
	} else {
		/* small block: don't clobber ecx + smaller code */
		if (n >= 4*4) __asm__ __volatile__("movsl"
			:"=&D"(edi),"=&S"(esi):"0"(edi),"1"(esi):"memory");
		if (n >= 3*4) __asm__ __volatile__("movsl"
			:"=&D"(edi),"=&S"(esi):"0"(edi),"1"(esi):"memory");
		if (n >= 2*4) __asm__ __volatile__("movsl"
			:"=&D"(edi),"=&S"(esi):"0"(edi),"1"(esi):"memory");
		if (n >= 1*4) __asm__ __volatile__("movsl"
			:"=&D"(edi),"=&S"(esi):"0"(edi),"1"(esi):"memory");
	}
	switch (n % 4) {
		/* tail */
		case 0: return to;
		case 1: __asm__ __volatile__("movsb"
			:"=&D"(edi),"=&S"(esi):"0"(edi),"1"(esi):"memory");
			return to;
		case 2: __asm__ __volatile__("movsw"
			:"=&D"(edi),"=&S"(esi):"0"(edi),"1"(esi):"memory");
			return to;
		default: __asm__ __volatile__("movsw\n\tmovsb"
			:"=&D"(edi),"=&S"(esi):"0"(edi),"1"(esi):"memory");
			return to;
	}
}

#define __HAVE_ARCH_MEMCPY

#ifdef CONFIG_X86_USE_3DNOW

#include <asm/mmx.h>

/*
 *	This CPU favours 3DNow strongly (eg AMD Athlon)
 */

static inline void * __constant_memcpy3d(void * to, const void * from, size_t len)
{
	if (len < 512)
		return __constant_memcpy(to, from, len);
	return _mmx_memcpy(to, from, len);
}

static __inline__ void *__memcpy3d(void *to, const void *from, size_t len)
{
	if (len < 512)
		return __memcpy(to, from, len);
	return _mmx_memcpy(to, from, len);
}

#define memcpy(t, f, n) \
(__builtin_constant_p(n) ? \
 __constant_memcpy3d((t),(f),(n)) : \
 __memcpy3d((t),(f),(n)))

#else

/*
 *	No 3D Now!
 */
 
#define memcpy(t, f, n) \
(__builtin_constant_p(n) ? \
 __constant_memcpy((t),(f),(n)) : \
 __memcpy((t),(f),(n)))

#endif

#define __HAVE_ARCH_MEMMOVE
void *memmove(void * dest,const void * src, size_t n);

#define memcmp __builtin_memcmp

#define __HAVE_ARCH_MEMCHR
static inline void * memchr(const void * cs,int c,size_t count)
{
int d0;
register void * __res;
if (!count)
	return NULL;
__asm__ __volatile__(
	"repne\n\t"
	"scasb\n\t"
	"je 1f\n\t"
	"movl $1,%0\n"
	"1:\tdecl %0"
	:"=D" (__res), "=&c" (d0)
	:"a" (c),"0" (cs),"1" (count)
	:"memory");
return __res;
}

static inline void * __memset_generic(void * s, char c,size_t count)
{
int d0, d1;
__asm__ __volatile__(
	"rep\n\t"
	"stosb"
	: "=&c" (d0), "=&D" (d1)
	:"a" (c),"1" (s),"0" (count)
	:"memory");
return s;
}

/* we might want to write optimized versions of these later */
#define __constant_count_memset(s,c,count) __memset_generic((s),(c),(count))

/*
 * memset(x,0,y) is a reasonably common thing to do, so we want to fill
 * things 32 bits at a time even when we don't know the size of the
 * area at compile-time..
 */
static __always_inline void * __constant_c_memset(void * s, unsigned long c, size_t count)
{
int d0, d1;
__asm__ __volatile__(
	"rep ; stosl\n\t"
	"testb $2,%b3\n\t"
	"je 1f\n\t"
	"stosw\n"
	"1:\ttestb $1,%b3\n\t"
	"je 2f\n\t"
	"stosb\n"
	"2:"
	:"=&c" (d0), "=&D" (d1)
	:"a" (c), "q" (count), "0" (count/4), "1" ((long) s)
	:"memory");
return (s);	
}

/* Added by Gertjan van Wingerde to make minix and sysv module work */
#define __HAVE_ARCH_STRNLEN
static inline size_t strnlen(const char * s, size_t count)
{
int d0;
register int __res;
__asm__ __volatile__(
	"movl %2,%0\n\t"
	"jmp 2f\n"
	"1:\tcmpb $0,(%0)\n\t"
	"je 3f\n\t"
	"incl %0\n"
	"2:\tdecl %1\n\t"
	"cmpl $-1,%1\n\t"
	"jne 1b\n"
	"3:\tsubl %2,%0"
	:"=a" (__res), "=&d" (d0)
	:"c" (s),"1" (count)
	:"memory");
return __res;
}
/* end of additional stuff */

#define __HAVE_ARCH_STRSTR

extern char *strstr(const char *cs, const char *ct);

/*
 * This looks horribly ugly, but the compiler can optimize it totally,
 * as we by now know that both pattern and count is constant..
 */
static __always_inline void * __constant_c_and_count_memset(void * s, unsigned long pattern, size_t count)
{
	switch (count) {
		case 0:
			return s;
		case 1:
			*(unsigned char *)s = pattern;
			return s;
		case 2:
			*(unsigned short *)s = pattern;
			return s;
		case 3:
			*(unsigned short *)s = pattern;
			*(2+(unsigned char *)s) = pattern;
			return s;
		case 4:
			*(unsigned long *)s = pattern;
			return s;
	}
#define COMMON(x) \
__asm__  __volatile__( \
	"rep ; stosl" \
	x \
	: "=&c" (d0), "=&D" (d1) \
	: "a" (pattern),"0" (count/4),"1" ((long) s) \
	: "memory")
{
	int d0, d1;
	switch (count % 4) {
		case 0: COMMON(""); return s;
		case 1: COMMON("\n\tstosb"); return s;
		case 2: COMMON("\n\tstosw"); return s;
		default: COMMON("\n\tstosw\n\tstosb"); return s;
	}
}
  
#undef COMMON
}

#define __constant_c_x_memset(s, c, count) \
(__builtin_constant_p(count) ? \
 __constant_c_and_count_memset((s),(c),(count)) : \
 __constant_c_memset((s),(c),(count)))

#define __memset(s, c, count) \
(__builtin_constant_p(count) ? \
 __constant_count_memset((s),(c),(count)) : \
 __memset_generic((s),(c),(count)))

#define __HAVE_ARCH_MEMSET
#define memset(s, c, count) \
(__builtin_constant_p(c) ? \
 __constant_c_x_memset((s),(0x01010101UL*(unsigned char)(c)),(count)) : \
 __memset((s),(c),(count)))

/*
 * find the first occurrence of byte 'c', or 1 past the area if none
 */
#define __HAVE_ARCH_MEMSCAN
static inline void * memscan(void * addr, int c, size_t size)
{
	if (!size)
		return addr;
	__asm__("repnz; scasb\n\t"
		"jnz 1f\n\t"
		"dec %%edi\n"
		"1:"
		: "=D" (addr), "=c" (size)
		: "0" (addr), "1" (size), "a" (c)
		: "memory");
	return addr;
}

#endif /* __KERNEL__ */

#endif
Commit	Line	Data
1da177e4 LT	1	#ifndef _I386_STRING_H_
	2	#define _I386_STRING_H_
	3
	4	#ifdef __KERNEL__
1da177e4 LT	5	/*
	6	* On a 486 or Pentium, we are better off not using the
	7	* byte string operations. But on a 386 or a PPro the
	8	* byte string ops are faster than doing it by hand
	9	* (MUCH faster on a Pentium).
	10	*/
	11
	12	/*
	13	* This string-include defines all string functions as inline
	14	* functions. Use gcc. It also assumes ds=es=data space, this should be
	15	* normal. Most of the string-functions are rather heavily hand-optimized,
	16	* see especially strsep,strstr,str[c]spn. They should work, but are not
	17	* very easy to understand. Everything is done entirely within the register
	18	* set, making the functions fast and clean. String instructions have been
	19	* used through-out, making for "slightly" unclear code :-)
	20	*
	21	* NO Copyright (C) 1991, 1992 Linus Torvalds,
	22	* consider these trivial functions to be PD.
	23	*/
	24
	25	/* AK: in fact I bet it would be better to move this stuff all out of line.
	26	*/
	27
	28	#define __HAVE_ARCH_STRCPY
	29	static inline char * strcpy(char * dest,const char *src)
	30	{
	31	int d0, d1, d2;
	32	__asm__ __volatile__(
	33	"1:\tlodsb\n\t"
	34	"stosb\n\t"
	35	"testb %%al,%%al\n\t"
	36	"jne 1b"
	37	: "=&S" (d0), "=&D" (d1), "=&a" (d2)
	38	:"0" (src),"1" (dest) : "memory");
	39	return dest;
	40	}
	41
	42	#define __HAVE_ARCH_STRNCPY
	43	static inline char * strncpy(char * dest,const char *src,size_t count)
	44	{
	45	int d0, d1, d2, d3;
	46	__asm__ __volatile__(
	47	"1:\tdecl %2\n\t"
	48	"js 2f\n\t"
	49	"lodsb\n\t"
	50	"stosb\n\t"
	51	"testb %%al,%%al\n\t"
	52	"jne 1b\n\t"
	53	"rep\n\t"
	54	"stosb\n"
	55	"2:"
	56	: "=&S" (d0), "=&D" (d1), "=&c" (d2), "=&a" (d3)
	57	:"0" (src),"1" (dest),"2" (count) : "memory");
	58	return dest;
	59	}
	60
	61	#define __HAVE_ARCH_STRCAT
	62	static inline char * strcat(char * dest,const char * src)
	63	{
	64	int d0, d1, d2, d3;
	65	__asm__ __volatile__(
	66	"repne\n\t"
	67	"scasb\n\t"
	68	"decl %1\n"
69	"1:\tlodsb\n\t"
70	"stosb\n\t"
71	"testb %%al,%%al\n\t"
72	"jne 1b"
73	: "=&S" (d0), "=&D" (d1), "=&a" (d2), "=&c" (d3)
74	: "0" (src), "1" (dest), "2" (0), "3" (0xffffffffu):"memory");
75	return dest;
76	}
77
78	#define __HAVE_ARCH_STRNCAT
79	static inline char * strncat(char * dest,const char * src,size_t count)
80	{
81	int d0, d1, d2, d3;
82	__asm__ __volatile__(
83	"repne\n\t"
84	"scasb\n\t"
85	"decl %1\n\t"
86	"movl %8,%3\n"
87	"1:\tdecl %3\n\t"
88	"js 2f\n\t"
89	"lodsb\n\t"
90	"stosb\n\t"
91	"testb %%al,%%al\n\t"
92	"jne 1b\n"
93	"2:\txorl %2,%2\n\t"
94	"stosb"
95	: "=&S" (d0), "=&D" (d1), "=&a" (d2), "=&c" (d3)
96	: "0" (src),"1" (dest),"2" (0),"3" (0xffffffffu), "g" (count)
97	: "memory");
98	return dest;
99	}
100
101	#define __HAVE_ARCH_STRCMP
102	static inline int strcmp(const char * cs,const char * ct)
103	{
104	int d0, d1;
105	register int __res;
106	__asm__ __volatile__(
107	"1:\tlodsb\n\t"
108	"scasb\n\t"
109	"jne 2f\n\t"
110	"testb %%al,%%al\n\t"
111	"jne 1b\n\t"
112	"xorl %%eax,%%eax\n\t"
113	"jmp 3f\n"
114	"2:\tsbbl %%eax,%%eax\n\t"
115	"orb $1,%%al\n"
116	"3:"
117	:"=a" (__res), "=&S" (d0), "=&D" (d1)
793ae774 LT	118	:"1" (cs),"2" (ct)
793ae774 LT	119	:"memory");
1da177e4 LT	120	return __res;
	121	}
	122
	123	#define __HAVE_ARCH_STRNCMP
	124	static inline int strncmp(const char * cs,const char * ct,size_t count)
	125	{
	126	register int __res;
	127	int d0, d1, d2;
	128	__asm__ __volatile__(
	129	"1:\tdecl %3\n\t"
	130	"js 2f\n\t"
	131	"lodsb\n\t"
	132	"scasb\n\t"
	133	"jne 3f\n\t"
	134	"testb %%al,%%al\n\t"
	135	"jne 1b\n"
	136	"2:\txorl %%eax,%%eax\n\t"
	137	"jmp 4f\n"
	138	"3:\tsbbl %%eax,%%eax\n\t"
	139	"orb $1,%%al\n"
	140	"4:"
793ae774 LT	141	:"=a" (__res), "=&S" (d0), "=&D" (d1), "=&c" (d2)
	142	:"1" (cs),"2" (ct),"3" (count)
	143	:"memory");
1da177e4 LT	144	return __res;
	145	}
	146
	147	#define __HAVE_ARCH_STRCHR
	148	static inline char * strchr(const char * s, int c)
	149	{
	150	int d0;
	151	register char * __res;
	152	__asm__ __volatile__(
	153	"movb %%al,%%ah\n"
	154	"1:\tlodsb\n\t"
	155	"cmpb %%ah,%%al\n\t"
	156	"je 2f\n\t"
	157	"testb %%al,%%al\n\t"
	158	"jne 1b\n\t"
	159	"movl $1,%1\n"
	160	"2:\tmovl %1,%0\n\t"
	161	"decl %0"
793ae774 LT	162	:"=a" (__res), "=&S" (d0)
	163	:"1" (s),"0" (c)
	164	:"memory");
1da177e4 LT	165	return __res;
	166	}
	167
	168	#define __HAVE_ARCH_STRRCHR
	169	static inline char * strrchr(const char * s, int c)
	170	{
	171	int d0, d1;
	172	register char * __res;
	173	__asm__ __volatile__(
	174	"movb %%al,%%ah\n"
	175	"1:\tlodsb\n\t"
	176	"cmpb %%ah,%%al\n\t"
	177	"jne 2f\n\t"
	178	"leal -1(%%esi),%0\n"
	179	"2:\ttestb %%al,%%al\n\t"
	180	"jne 1b"
793ae774 LT	181	:"=g" (__res), "=&S" (d0), "=&a" (d1)
	182	:"0" (0),"1" (s),"2" (c)
	183	:"memory");
1da177e4 LT	184	return __res;
	185	}
	186
	187	#define __HAVE_ARCH_STRLEN
	188	static inline size_t strlen(const char * s)
	189	{
	190	int d0;
	191	register int __res;
	192	__asm__ __volatile__(
	193	"repne\n\t"
	194	"scasb\n\t"
	195	"notl %0\n\t"
	196	"decl %0"
793ae774 LT	197	:"=c" (__res), "=&D" (d0)
	198	:"1" (s),"a" (0), "0" (0xffffffffu)
	199	:"memory");
1da177e4 LT	200	return __res;
	201	}
	202
652050ae	203	static __always_inline void * __memcpy(void * to, const void * from, size_t n)
1da177e4 LT	204	{
	205	int d0, d1, d2;
	206	__asm__ __volatile__(
	207	"rep ; movsl\n\t"
d5b63d78 DV	208	"movl %4,%%ecx\n\t"
	209	"andl $3,%%ecx\n\t"
	210	#if 1 /* want to pay 2 byte penalty for a chance to skip microcoded rep? */
	211	"jz 1f\n\t"
	212	#endif
	213	"rep ; movsb\n\t"
	214	"1:"
1da177e4	215	: "=&c" (d0), "=&D" (d1), "=&S" (d2)
d5b63d78	216	: "0" (n/4), "g" (n), "1" ((long) to), "2" ((long) from)
1da177e4 LT	217	: "memory");
	218	return (to);
	219	}
	220
	221	/*
d5b63d78	222	* This looks ugly, but the compiler can optimize it totally,
1da177e4 LT	223	* as the count is constant.
1da177e4 LT	224	*/
652050ae	225	static __always_inline void * __constant_memcpy(void * to, const void * from, size_t n)
1da177e4	226	{
d5b63d78 DV	227	long esi, edi;
	228	if (!n) return to;
	229	#if 1 /* want to do small copies with non-string ops? */
	230	switch (n) {
	231	case 1: (char)to = (char)from; return to;
	232	case 2: (short)to = (short)from; return to;
	233	case 4: (int)to = (int)from; return to;
	234	#if 1 /* including those doable with two moves? */
	235	case 3: (short)to = (short)from;
	236	((char)to+2) = ((char)from+2); return to;
	237	case 5: (int)to = (int)from;
	238	((char)to+4) = ((char)from+4); return to;
	239	case 6: (int)to = (int)from;
	240	((short)to+2) = ((short)from+2); return to;
	241	case 8: (int)to = (int)from;
	242	((int)to+1) = ((int)from+1); return to;
	243	#endif
	244	}
	245	#endif
	246	esi = (long) from;
	247	edi = (long) to;
	248	if (n >= 5*4) {
	249	/* large block: use rep prefix */
	250	int ecx;
	251	__asm__ __volatile__(
	252	"rep ; movsl"
	253	: "=&c" (ecx), "=&D" (edi), "=&S" (esi)
	254	: "0" (n/4), "1" (edi),"2" (esi)
	255	: "memory"
	256	);
	257	} else {
	258	/* small block: don't clobber ecx + smaller code */
	259	if (n >= 4*4) __asm__ __volatile__("movsl"
	260	:"=&D"(edi),"=&S"(esi):"0"(edi),"1"(esi):"memory");
	261	if (n >= 3*4) __asm__ __volatile__("movsl"
	262	:"=&D"(edi),"=&S"(esi):"0"(edi),"1"(esi):"memory");
	263	if (n >= 2*4) __asm__ __volatile__("movsl"
	264	:"=&D"(edi),"=&S"(esi):"0"(edi),"1"(esi):"memory");
	265	if (n >= 1*4) __asm__ __volatile__("movsl"
	266	:"=&D"(edi),"=&S"(esi):"0"(edi),"1"(esi):"memory");
	267	}
1da177e4	268	switch (n % 4) {
d5b63d78 DV	269	/* tail */
	270	case 0: return to;
	271	case 1: __asm__ __volatile__("movsb"
	272	:"=&D"(edi),"=&S"(esi):"0"(edi),"1"(esi):"memory");
	273	return to;
	274	case 2: __asm__ __volatile__("movsw"
	275	:"=&D"(edi),"=&S"(esi):"0"(edi),"1"(esi):"memory");
	276	return to;
	277	default: __asm__ __volatile__("movsw\n\tmovsb"
	278	:"=&D"(edi),"=&S"(esi):"0"(edi),"1"(esi):"memory");
	279	return to;
1da177e4 LT	280	}
1da177e4 LT	281	}
1da177e4 LT	282
	283	#define __HAVE_ARCH_MEMCPY
	284
	285	#ifdef CONFIG_X86_USE_3DNOW
	286
	287	#include <asm/mmx.h>
	288
	289	/*
	290	* This CPU favours 3DNow strongly (eg AMD Athlon)
	291	*/
	292
	293	static inline void * __constant_memcpy3d(void * to, const void * from, size_t len)
	294	{
	295	if (len < 512)
	296	return __constant_memcpy(to, from, len);
	297	return _mmx_memcpy(to, from, len);
	298	}
	299
	300	static __inline__ void __memcpy3d(void to, const void *from, size_t len)
	301	{
	302	if (len < 512)
	303	return __memcpy(to, from, len);
	304	return _mmx_memcpy(to, from, len);
	305	}
	306
	307	#define memcpy(t, f, n) \
	308	(__builtin_constant_p(n) ? \
	309	__constant_memcpy3d((t),(f),(n)) : \
	310	__memcpy3d((t),(f),(n)))
	311
	312	#else
	313
	314	/*
	315	* No 3D Now!
	316	*/
	317
	318	#define memcpy(t, f, n) \
	319	(__builtin_constant_p(n) ? \
	320	__constant_memcpy((t),(f),(n)) : \
	321	__memcpy((t),(f),(n)))
	322
	323	#endif
	324
	325	#define __HAVE_ARCH_MEMMOVE
	326	void memmove(void dest,const void * src, size_t n);
	327
	328	#define memcmp __builtin_memcmp
	329
	330	#define __HAVE_ARCH_MEMCHR
	331	static inline void * memchr(const void * cs,int c,size_t count)
	332	{
	333	int d0;
	334	register void * __res;
	335	if (!count)
	336	return NULL;
	337	__asm__ __volatile__(
	338	"repne\n\t"
	339	"scasb\n\t"
	340	"je 1f\n\t"
	341	"movl $1,%0\n"
	342	"1:\tdecl %0"
793ae774 LT	343	:"=D" (__res), "=&c" (d0)
	344	:"a" (c),"0" (cs),"1" (count)
	345	:"memory");
1da177e4 LT	346	return __res;
	347	}
	348
	349	static inline void * __memset_generic(void * s, char c,size_t count)
	350	{
	351	int d0, d1;
	352	__asm__ __volatile__(
	353	"rep\n\t"
	354	"stosb"
	355	: "=&c" (d0), "=&D" (d1)
	356	:"a" (c),"1" (s),"0" (count)
	357	:"memory");
	358	return s;
	359	}
	360
	361	/* we might want to write optimized versions of these later */
	362	#define __constant_count_memset(s,c,count) __memset_generic((s),(c),(count))
	363
	364	/*
	365	* memset(x,0,y) is a reasonably common thing to do, so we want to fill
	366	* things 32 bits at a time even when we don't know the size of the
	367	* area at compile-time..
	368	*/
652050ae	369	static __always_inline void * __constant_c_memset(void * s, unsigned long c, size_t count)
1da177e4 LT	370	{
	371	int d0, d1;
	372	__asm__ __volatile__(
	373	"rep ; stosl\n\t"
	374	"testb $2,%b3\n\t"
	375	"je 1f\n\t"
	376	"stosw\n"
	377	"1:\ttestb $1,%b3\n\t"
	378	"je 2f\n\t"
	379	"stosb\n"
	380	"2:"
793ae774	381	:"=&c" (d0), "=&D" (d1)
1da177e4 LT	382	:"a" (c), "q" (count), "0" (count/4), "1" ((long) s)
	383	:"memory");
	384	return (s);
	385	}
	386
	387	/* Added by Gertjan van Wingerde to make minix and sysv module work */
	388	#define __HAVE_ARCH_STRNLEN
	389	static inline size_t strnlen(const char * s, size_t count)
	390	{
	391	int d0;
	392	register int __res;
	393	__asm__ __volatile__(
	394	"movl %2,%0\n\t"
	395	"jmp 2f\n"
	396	"1:\tcmpb $0,(%0)\n\t"
	397	"je 3f\n\t"
	398	"incl %0\n"
	399	"2:\tdecl %1\n\t"
	400	"cmpl $-1,%1\n\t"
	401	"jne 1b\n"
	402	"3:\tsubl %2,%0"
	403	:"=a" (__res), "=&d" (d0)
793ae774 LT	404	:"c" (s),"1" (count)
793ae774 LT	405	:"memory");
1da177e4 LT	406	return __res;
	407	}
	408	/* end of additional stuff */
	409
	410	#define __HAVE_ARCH_STRSTR
	411
	412	extern char strstr(const char cs, const char *ct);
	413
	414	/*
	415	* This looks horribly ugly, but the compiler can optimize it totally,
	416	* as we by now know that both pattern and count is constant..
	417	*/
652050ae	418	static __always_inline void * __constant_c_and_count_memset(void * s, unsigned long pattern, size_t count)
1da177e4 LT	419	{
	420	switch (count) {
	421	case 0:
	422	return s;
	423	case 1:
	424	(unsigned char )s = pattern;
	425	return s;
	426	case 2:
	427	(unsigned short )s = pattern;
	428	return s;
	429	case 3:
	430	(unsigned short )s = pattern;
	431	(2+(unsigned char )s) = pattern;
	432	return s;
	433	case 4:
	434	(unsigned long )s = pattern;
	435	return s;
	436	}
	437	#define COMMON(x) \
	438	__asm__ __volatile__( \
	439	"rep ; stosl" \
	440	x \
	441	: "=&c" (d0), "=&D" (d1) \
	442	: "a" (pattern),"0" (count/4),"1" ((long) s) \
	443	: "memory")
	444	{
	445	int d0, d1;
	446	switch (count % 4) {
	447	case 0: COMMON(""); return s;
	448	case 1: COMMON("\n\tstosb"); return s;
	449	case 2: COMMON("\n\tstosw"); return s;
	450	default: COMMON("\n\tstosw\n\tstosb"); return s;
	451	}
	452	}
	453
	454	#undef COMMON
	455	}
	456
	457	#define __constant_c_x_memset(s, c, count) \
	458	(__builtin_constant_p(count) ? \
	459	__constant_c_and_count_memset((s),(c),(count)) : \
	460	__constant_c_memset((s),(c),(count)))
	461
	462	#define __memset(s, c, count) \
	463	(__builtin_constant_p(count) ? \
	464	__constant_count_memset((s),(c),(count)) : \
	465	__memset_generic((s),(c),(count)))
	466
	467	#define __HAVE_ARCH_MEMSET
	468	#define memset(s, c, count) \
	469	(__builtin_constant_p(c) ? \
	470	__constant_c_x_memset((s),(0x01010101UL*(unsigned char)(c)),(count)) : \
	471	__memset((s),(c),(count)))
	472
	473	/*
	474	* find the first occurrence of byte 'c', or 1 past the area if none
	475	*/
	476	#define __HAVE_ARCH_MEMSCAN
	477	static inline void * memscan(void * addr, int c, size_t size)
	478	{
	479	if (!size)
	480	return addr;
	481	__asm__("repnz; scasb\n\t"
	482	"jnz 1f\n\t"
483	"dec %%edi\n"
484	"1:"
485	: "=D" (addr), "=c" (size)
793ae774 LT	486	: "0" (addr), "1" (size), "a" (c)
793ae774 LT	487	: "memory");
1da177e4 LT	488	return addr;
	489	}
	490
	491	#endif /* __KERNEL__ */
	492
	493	#endif