[mirror_ubuntu-bionic-kernel.git] / include / asm-x86 / string_32.h

#ifndef _I386_STRING_H_
#define _I386_STRING_H_

#ifdef __KERNEL__

/* Let gcc decide whether to inline or use the out of line functions */

#define __HAVE_ARCH_STRCPY
extern char *strcpy(char *dest, const char *src);

#define __HAVE_ARCH_STRNCPY
extern char *strncpy(char *dest, const char *src, size_t count);

#define __HAVE_ARCH_STRCAT
extern char *strcat(char *dest, const char *src);

#define __HAVE_ARCH_STRNCAT
extern char *strncat(char *dest, const char *src, size_t count);

#define __HAVE_ARCH_STRCMP
extern int strcmp(const char *cs, const char *ct);

#define __HAVE_ARCH_STRNCMP
extern int strncmp(const char *cs, const char *ct, size_t count);

#define __HAVE_ARCH_STRCHR
extern char *strchr(const char *s, int c);

#define __HAVE_ARCH_STRLEN
extern size_t strlen(const char *s);

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"
	"jz 1f\n\t"
	"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
extern void *memchr(const void * cs,int c,size_t count);

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
extern size_t strnlen(const char * s, size_t count);
/* 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 & 0xff;
			return s;
		case 2:
			*(unsigned short *)s = pattern & 0xffff;
			return s;
		case 3:
			*(unsigned short *)s = pattern & 0xffff;
			*(2+(unsigned char *)s) = pattern & 0xff;
			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
extern void *memscan(void * addr, int c, size_t size);

#endif /* __KERNEL__ */

#endif
Commit	Line	Data
1da177e4 LT	1	#ifndef _I386_STRING_H_
	2	#define _I386_STRING_H_
	3
	4	#ifdef __KERNEL__
1da177e4	5
06b0f574	6	/* Let gcc decide whether to inline or use the out of line functions */
1da177e4 LT	7
1da177e4 LT	8	#define __HAVE_ARCH_STRCPY
b520b85a	9	extern char strcpy(char dest, const char *src);
1da177e4 LT	10
1da177e4 LT	11	#define __HAVE_ARCH_STRNCPY
b520b85a	12	extern char strncpy(char dest, const char *src, size_t count);
1da177e4 LT	13
1da177e4 LT	14	#define __HAVE_ARCH_STRCAT
b520b85a	15	extern char strcat(char dest, const char *src);
1da177e4 LT	16
1da177e4 LT	17	#define __HAVE_ARCH_STRNCAT
b520b85a	18	extern char strncat(char dest, const char *src, size_t count);
1da177e4 LT	19
1da177e4 LT	20	#define __HAVE_ARCH_STRCMP
b520b85a	21	extern int strcmp(const char cs, const char ct);
1da177e4 LT	22
1da177e4 LT	23	#define __HAVE_ARCH_STRNCMP
b520b85a	24	extern int strncmp(const char cs, const char ct, size_t count);
1da177e4 LT	25
1da177e4 LT	26	#define __HAVE_ARCH_STRCHR
b520b85a	27	extern char strchr(const char s, int c);
1da177e4	28
1da177e4	29	#define __HAVE_ARCH_STRLEN
b520b85a	30	extern size_t strlen(const char *s);
1da177e4	31
652050ae	32	static __always_inline void * __memcpy(void * to, const void * from, size_t n)
1da177e4 LT	33	{
	34	int d0, d1, d2;
	35	__asm__ __volatile__(
	36	"rep ; movsl\n\t"
d5b63d78 DV	37	"movl %4,%%ecx\n\t"
d5b63d78 DV	38	"andl $3,%%ecx\n\t"
d5b63d78	39	"jz 1f\n\t"
d5b63d78 DV	40	"rep ; movsb\n\t"
d5b63d78 DV	41	"1:"
1da177e4	42	: "=&c" (d0), "=&D" (d1), "=&S" (d2)
d5b63d78	43	: "0" (n/4), "g" (n), "1" ((long) to), "2" ((long) from)
1da177e4 LT	44	: "memory");
	45	return (to);
	46	}
	47
	48	/*
d5b63d78	49	* This looks ugly, but the compiler can optimize it totally,
1da177e4 LT	50	* as the count is constant.
1da177e4 LT	51	*/
652050ae	52	static __always_inline void * __constant_memcpy(void * to, const void * from, size_t n)
1da177e4	53	{
d5b63d78 DV	54	long esi, edi;
	55	if (!n) return to;
	56	#if 1 /* want to do small copies with non-string ops? */
	57	switch (n) {
	58	case 1: (char)to = (char)from; return to;
	59	case 2: (short)to = (short)from; return to;
	60	case 4: (int)to = (int)from; return to;
	61	#if 1 /* including those doable with two moves? */
	62	case 3: (short)to = (short)from;
	63	((char)to+2) = ((char)from+2); return to;
	64	case 5: (int)to = (int)from;
	65	((char)to+4) = ((char)from+4); return to;
	66	case 6: (int)to = (int)from;
	67	((short)to+2) = ((short)from+2); return to;
	68	case 8: (int)to = (int)from;
	69	((int)to+1) = ((int)from+1); return to;
	70	#endif
	71	}
	72	#endif
	73	esi = (long) from;
	74	edi = (long) to;
	75	if (n >= 5*4) {
	76	/* large block: use rep prefix */
	77	int ecx;
	78	__asm__ __volatile__(
	79	"rep ; movsl"
	80	: "=&c" (ecx), "=&D" (edi), "=&S" (esi)
	81	: "0" (n/4), "1" (edi),"2" (esi)
	82	: "memory"
	83	);
	84	} else {
	85	/* small block: don't clobber ecx + smaller code */
	86	if (n >= 4*4) __asm__ __volatile__("movsl"
	87	:"=&D"(edi),"=&S"(esi):"0"(edi),"1"(esi):"memory");
	88	if (n >= 3*4) __asm__ __volatile__("movsl"
	89	:"=&D"(edi),"=&S"(esi):"0"(edi),"1"(esi):"memory");
	90	if (n >= 2*4) __asm__ __volatile__("movsl"
	91	:"=&D"(edi),"=&S"(esi):"0"(edi),"1"(esi):"memory");
	92	if (n >= 1*4) __asm__ __volatile__("movsl"
	93	:"=&D"(edi),"=&S"(esi):"0"(edi),"1"(esi):"memory");
	94	}
1da177e4	95	switch (n % 4) {
d5b63d78 DV	96	/* tail */
	97	case 0: return to;
	98	case 1: __asm__ __volatile__("movsb"
	99	:"=&D"(edi),"=&S"(esi):"0"(edi),"1"(esi):"memory");
	100	return to;
	101	case 2: __asm__ __volatile__("movsw"
	102	:"=&D"(edi),"=&S"(esi):"0"(edi),"1"(esi):"memory");
	103	return to;
	104	default: __asm__ __volatile__("movsw\n\tmovsb"
	105	:"=&D"(edi),"=&S"(esi):"0"(edi),"1"(esi):"memory");
	106	return to;
1da177e4 LT	107	}
1da177e4 LT	108	}
1da177e4 LT	109
	110	#define __HAVE_ARCH_MEMCPY
	111
	112	#ifdef CONFIG_X86_USE_3DNOW
	113
	114	#include <asm/mmx.h>
	115
	116	/*
	117	* This CPU favours 3DNow strongly (eg AMD Athlon)
	118	*/
	119
	120	static inline void * __constant_memcpy3d(void * to, const void * from, size_t len)
	121	{
	122	if (len < 512)
	123	return __constant_memcpy(to, from, len);
	124	return _mmx_memcpy(to, from, len);
	125	}
	126
	127	static __inline__ void __memcpy3d(void to, const void *from, size_t len)
	128	{
	129	if (len < 512)
	130	return __memcpy(to, from, len);
	131	return _mmx_memcpy(to, from, len);
	132	}
	133
	134	#define memcpy(t, f, n) \
	135	(__builtin_constant_p(n) ? \
	136	__constant_memcpy3d((t),(f),(n)) : \
	137	__memcpy3d((t),(f),(n)))
	138
	139	#else
	140
	141	/*
	142	* No 3D Now!
	143	*/
	144
	145	#define memcpy(t, f, n) \
	146	(__builtin_constant_p(n) ? \
	147	__constant_memcpy((t),(f),(n)) : \
	148	__memcpy((t),(f),(n)))
	149
	150	#endif
	151
	152	#define __HAVE_ARCH_MEMMOVE
	153	void memmove(void dest,const void * src, size_t n);
	154
	155	#define memcmp __builtin_memcmp
	156
	157	#define __HAVE_ARCH_MEMCHR
b520b85a	158	extern void memchr(const void cs,int c,size_t count);
1da177e4 LT	159
	160	static inline void * __memset_generic(void * s, char c,size_t count)
	161	{
	162	int d0, d1;
	163	__asm__ __volatile__(
	164	"rep\n\t"
	165	"stosb"
	166	: "=&c" (d0), "=&D" (d1)
	167	:"a" (c),"1" (s),"0" (count)
	168	:"memory");
	169	return s;
	170	}
	171
	172	/* we might want to write optimized versions of these later */
	173	#define __constant_count_memset(s,c,count) __memset_generic((s),(c),(count))
	174
	175	/*
	176	* memset(x,0,y) is a reasonably common thing to do, so we want to fill
	177	* things 32 bits at a time even when we don't know the size of the
	178	* area at compile-time..
	179	*/
652050ae	180	static __always_inline void * __constant_c_memset(void * s, unsigned long c, size_t count)
1da177e4 LT	181	{
	182	int d0, d1;
	183	__asm__ __volatile__(
	184	"rep ; stosl\n\t"
	185	"testb $2,%b3\n\t"
	186	"je 1f\n\t"
	187	"stosw\n"
	188	"1:\ttestb $1,%b3\n\t"
	189	"je 2f\n\t"
	190	"stosb\n"
	191	"2:"
793ae774	192	:"=&c" (d0), "=&D" (d1)
1da177e4 LT	193	:"a" (c), "q" (count), "0" (count/4), "1" ((long) s)
	194	:"memory");
	195	return (s);
	196	}
	197
	198	/* Added by Gertjan van Wingerde to make minix and sysv module work */
	199	#define __HAVE_ARCH_STRNLEN
b520b85a	200	extern size_t strnlen(const char * s, size_t count);
1da177e4 LT	201	/* end of additional stuff */
	202
	203	#define __HAVE_ARCH_STRSTR
1da177e4 LT	204	extern char strstr(const char cs, const char *ct);
	205
	206	/*
	207	* This looks horribly ugly, but the compiler can optimize it totally,
	208	* as we by now know that both pattern and count is constant..
	209	*/
652050ae	210	static __always_inline void * __constant_c_and_count_memset(void * s, unsigned long pattern, size_t count)
1da177e4 LT	211	{
	212	switch (count) {
	213	case 0:
	214	return s;
	215	case 1:
9b423934	216	(unsigned char )s = pattern & 0xff;
1da177e4 LT	217	return s;
1da177e4 LT	218	case 2:
9b423934	219	(unsigned short )s = pattern & 0xffff;
1da177e4 LT	220	return s;
1da177e4 LT	221	case 3:
9b423934 HH	222	(unsigned short )s = pattern & 0xffff;
9b423934 HH	223	(2+(unsigned char )s) = pattern & 0xff;
1da177e4 LT	224	return s;
	225	case 4:
	226	(unsigned long )s = pattern;
	227	return s;
	228	}
	229	#define COMMON(x) \
	230	__asm__ __volatile__( \
	231	"rep ; stosl" \
	232	x \
	233	: "=&c" (d0), "=&D" (d1) \
	234	: "a" (pattern),"0" (count/4),"1" ((long) s) \
	235	: "memory")
	236	{
	237	int d0, d1;
	238	switch (count % 4) {
	239	case 0: COMMON(""); return s;
	240	case 1: COMMON("\n\tstosb"); return s;
	241	case 2: COMMON("\n\tstosw"); return s;
	242	default: COMMON("\n\tstosw\n\tstosb"); return s;
	243	}
	244	}
	245
	246	#undef COMMON
	247	}
	248
	249	#define __constant_c_x_memset(s, c, count) \
	250	(__builtin_constant_p(count) ? \
	251	__constant_c_and_count_memset((s),(c),(count)) : \
	252	__constant_c_memset((s),(c),(count)))
	253
	254	#define __memset(s, c, count) \
	255	(__builtin_constant_p(count) ? \
	256	__constant_count_memset((s),(c),(count)) : \
	257	__memset_generic((s),(c),(count)))
	258
	259	#define __HAVE_ARCH_MEMSET
	260	#define memset(s, c, count) \
	261	(__builtin_constant_p(c) ? \
	262	__constant_c_x_memset((s),(0x01010101UL*(unsigned char)(c)),(count)) : \
	263	__memset((s),(c),(count)))
	264
	265	/*
	266	* find the first occurrence of byte 'c', or 1 past the area if none
	267	*/
	268	#define __HAVE_ARCH_MEMSCAN
b520b85a	269	extern void memscan(void addr, int c, size_t size);
1da177e4 LT	270
	271	#endif /* __KERNEL__ */
	272
	273	#endif