string_32.h 8.26 KB
Newer Older
1
/* SPDX-License-Identifier: GPL-2.0 */
2 3
#ifndef _ASM_X86_STRING_32_H
#define _ASM_X86_STRING_32_H
Linus Torvalds's avatar
Linus Torvalds committed
4 5 6

#ifdef __KERNEL__

7
/* Let gcc decide whether to inline or use the out of line functions */
Linus Torvalds's avatar
Linus Torvalds committed
8 9

#define __HAVE_ARCH_STRCPY
10
extern char *strcpy(char *dest, const char *src);
Linus Torvalds's avatar
Linus Torvalds committed
11 12

#define __HAVE_ARCH_STRNCPY
13
extern char *strncpy(char *dest, const char *src, size_t count);
Linus Torvalds's avatar
Linus Torvalds committed
14 15

#define __HAVE_ARCH_STRCAT
16
extern char *strcat(char *dest, const char *src);
Linus Torvalds's avatar
Linus Torvalds committed
17 18

#define __HAVE_ARCH_STRNCAT
19
extern char *strncat(char *dest, const char *src, size_t count);
Linus Torvalds's avatar
Linus Torvalds committed
20 21

#define __HAVE_ARCH_STRCMP
22
extern int strcmp(const char *cs, const char *ct);
Linus Torvalds's avatar
Linus Torvalds committed
23 24

#define __HAVE_ARCH_STRNCMP
25
extern int strncmp(const char *cs, const char *ct, size_t count);
Linus Torvalds's avatar
Linus Torvalds committed
26 27

#define __HAVE_ARCH_STRCHR
28
extern char *strchr(const char *s, int c);
Linus Torvalds's avatar
Linus Torvalds committed
29 30

#define __HAVE_ARCH_STRLEN
31
extern size_t strlen(const char *s);
Linus Torvalds's avatar
Linus Torvalds committed
32

33
static __always_inline void *__memcpy(void *to, const void *from, size_t n)
Linus Torvalds's avatar
Linus Torvalds committed
34
{
35 36 37 38 39 40 41 42 43 44 45
	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;
Linus Torvalds's avatar
Linus Torvalds committed
46 47 48
}

/*
Denis Vlasenko's avatar
Denis Vlasenko committed
49
 * This looks ugly, but the compiler can optimize it totally,
Linus Torvalds's avatar
Linus Torvalds committed
50 51
 * as the count is constant.
 */
52 53
static __always_inline void *__constant_memcpy(void *to, const void *from,
					       size_t n)
Linus Torvalds's avatar
Linus Torvalds committed
54
{
Denis Vlasenko's avatar
Denis Vlasenko committed
55
	long esi, edi;
56 57 58
	if (!n)
		return to;

Denis Vlasenko's avatar
Denis Vlasenko committed
59
	switch (n) {
60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84
	case 1:
		*(char *)to = *(char *)from;
		return to;
	case 2:
		*(short *)to = *(short *)from;
		return to;
	case 4:
		*(int *)to = *(int *)from;
		return to;
	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;
Denis Vlasenko's avatar
Denis Vlasenko committed
85
	}
86 87 88 89

	esi = (long)from;
	edi = (long)to;
	if (n >= 5 * 4) {
Denis Vlasenko's avatar
Denis Vlasenko committed
90 91
		/* large block: use rep prefix */
		int ecx;
92 93 94 95
		asm volatile("rep ; movsl"
			     : "=&c" (ecx), "=&D" (edi), "=&S" (esi)
			     : "0" (n / 4), "1" (edi), "2" (esi)
			     : "memory"
Denis Vlasenko's avatar
Denis Vlasenko committed
96 97 98
		);
	} else {
		/* small block: don't clobber ecx + smaller code */
99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118
		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");
Denis Vlasenko's avatar
Denis Vlasenko committed
119
	}
Linus Torvalds's avatar
Linus Torvalds committed
120
	switch (n % 4) {
Denis Vlasenko's avatar
Denis Vlasenko committed
121
		/* tail */
122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141
	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;
Linus Torvalds's avatar
Linus Torvalds committed
142 143 144 145
	}
}

#define __HAVE_ARCH_MEMCPY
146
extern void *memcpy(void *, const void *, size_t);
Linus Torvalds's avatar
Linus Torvalds committed
147

148
#ifndef CONFIG_FORTIFY_SOURCE
Linus Torvalds's avatar
Linus Torvalds committed
149 150 151 152 153 154 155 156
#ifdef CONFIG_X86_USE_3DNOW

#include <asm/mmx.h>

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

157
static inline void *__constant_memcpy3d(void *to, const void *from, size_t len)
Linus Torvalds's avatar
Linus Torvalds committed
158 159 160 161 162 163
{
	if (len < 512)
		return __constant_memcpy(to, from, len);
	return _mmx_memcpy(to, from, len);
}

164
static inline void *__memcpy3d(void *to, const void *from, size_t len)
Linus Torvalds's avatar
Linus Torvalds committed
165 166 167 168 169 170
{
	if (len < 512)
		return __memcpy(to, from, len);
	return _mmx_memcpy(to, from, len);
}

171 172 173 174
#define memcpy(t, f, n)				\
	(__builtin_constant_p((n))		\
	 ? __constant_memcpy3d((t), (f), (n))	\
	 : __memcpy3d((t), (f), (n)))
Linus Torvalds's avatar
Linus Torvalds committed
175 176 177 178 179 180

#else

/*
 *	No 3D Now!
 */
181

182 183 184
#if (__GNUC__ >= 4)
#define memcpy(t, f, n) __builtin_memcpy(t, f, n)
#else
185 186 187 188
#define memcpy(t, f, n)				\
	(__builtin_constant_p((n))		\
	 ? __constant_memcpy((t), (f), (n))	\
	 : __memcpy((t), (f), (n)))
189
#endif
Linus Torvalds's avatar
Linus Torvalds committed
190 191

#endif
192
#endif /* !CONFIG_FORTIFY_SOURCE */
Linus Torvalds's avatar
Linus Torvalds committed
193 194

#define __HAVE_ARCH_MEMMOVE
195
void *memmove(void *dest, const void *src, size_t n);
Linus Torvalds's avatar
Linus Torvalds committed
196

197 198
extern int memcmp(const void *, const void *, size_t);
#ifndef CONFIG_FORTIFY_SOURCE
Linus Torvalds's avatar
Linus Torvalds committed
199
#define memcmp __builtin_memcmp
200
#endif
Linus Torvalds's avatar
Linus Torvalds committed
201 202

#define __HAVE_ARCH_MEMCHR
203
extern void *memchr(const void *cs, int c, size_t count);
Linus Torvalds's avatar
Linus Torvalds committed
204

205
static inline void *__memset_generic(void *s, char c, size_t count)
Linus Torvalds's avatar
Linus Torvalds committed
206
{
207 208 209 210 211 212 213
	int d0, d1;
	asm volatile("rep\n\t"
		     "stosb"
		     : "=&c" (d0), "=&D" (d1)
		     : "a" (c), "1" (s), "0" (count)
		     : "memory");
	return s;
Linus Torvalds's avatar
Linus Torvalds committed
214 215 216
}

/* we might want to write optimized versions of these later */
217
#define __constant_count_memset(s, c, count) __memset_generic((s), (c), (count))
Linus Torvalds's avatar
Linus Torvalds committed
218 219

/*
220
 * memset(x, 0, y) is a reasonably common thing to do, so we want to fill
Linus Torvalds's avatar
Linus Torvalds committed
221 222 223
 * things 32 bits at a time even when we don't know the size of the
 * area at compile-time..
 */
224 225
static __always_inline
void *__constant_c_memset(void *s, unsigned long c, size_t count)
Linus Torvalds's avatar
Linus Torvalds committed
226
{
227 228 229 230 231 232 233 234 235 236 237 238 239
	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;
Linus Torvalds's avatar
Linus Torvalds committed
240 241 242 243
}

/* Added by Gertjan van Wingerde to make minix and sysv module work */
#define __HAVE_ARCH_STRNLEN
244
extern size_t strnlen(const char *s, size_t count);
Linus Torvalds's avatar
Linus Torvalds committed
245 246 247 248 249 250 251 252 253
/* 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..
 */
254 255 256
static __always_inline
void *__constant_c_and_count_memset(void *s, unsigned long pattern,
				    size_t count)
Linus Torvalds's avatar
Linus Torvalds committed
257 258
{
	switch (count) {
259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279
	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;
		*((unsigned char *)s + 2) = pattern & 0xff;
		return s;
	case 4:
		*(unsigned long *)s = pattern;
		return s;
	}

#define COMMON(x)							\
	asm volatile("rep ; stosl"					\
		     x							\
		     : "=&c" (d0), "=&D" (d1)				\
280
		     : "a" (eax), "0" (count/4), "1" ((long)s)	\
281 282 283 284
		     : "memory")

	{
		int d0, d1;
285 286 287 288 289 290 291
#if __GNUC__ == 4 && __GNUC_MINOR__ == 0
		/* Workaround for broken gcc 4.0 */
		register unsigned long eax asm("%eax") = pattern;
#else
		unsigned long eax = pattern;
#endif

292
		switch (count % 4) {
Linus Torvalds's avatar
Linus Torvalds committed
293
		case 0:
294
			COMMON("");
Linus Torvalds's avatar
Linus Torvalds committed
295 296
			return s;
		case 1:
297
			COMMON("\n\tstosb");
Linus Torvalds's avatar
Linus Torvalds committed
298 299
			return s;
		case 2:
300
			COMMON("\n\tstosw");
Linus Torvalds's avatar
Linus Torvalds committed
301
			return s;
302 303
		default:
			COMMON("\n\tstosw\n\tstosb");
Linus Torvalds's avatar
Linus Torvalds committed
304
			return s;
305
		}
Linus Torvalds's avatar
Linus Torvalds committed
306
	}
307

Linus Torvalds's avatar
Linus Torvalds committed
308 309 310
#undef COMMON
}

311 312 313 314
#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)))
Linus Torvalds's avatar
Linus Torvalds committed
315

316 317 318 319
#define __memset(s, c, count)				\
	(__builtin_constant_p(count)			\
	 ? __constant_count_memset((s), (c), (count))	\
	 : __memset_generic((s), (c), (count)))
Linus Torvalds's avatar
Linus Torvalds committed
320 321

#define __HAVE_ARCH_MEMSET
322 323
extern void *memset(void *, int, size_t);
#ifndef CONFIG_FORTIFY_SOURCE
324 325 326
#if (__GNUC__ >= 4)
#define memset(s, c, count) __builtin_memset(s, c, count)
#else
327 328 329 330 331
#define memset(s, c, count)						\
	(__builtin_constant_p(c)					\
	 ? __constant_c_x_memset((s), (0x01010101UL * (unsigned char)(c)), \
				 (count))				\
	 : __memset((s), (c), (count)))
332
#endif
333
#endif /* !CONFIG_FORTIFY_SOURCE */
Linus Torvalds's avatar
Linus Torvalds committed
334

335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358
#define __HAVE_ARCH_MEMSET16
static inline void *memset16(uint16_t *s, uint16_t v, size_t n)
{
	int d0, d1;
	asm volatile("rep\n\t"
		     "stosw"
		     : "=&c" (d0), "=&D" (d1)
		     : "a" (v), "1" (s), "0" (n)
		     : "memory");
	return s;
}

#define __HAVE_ARCH_MEMSET32
static inline void *memset32(uint32_t *s, uint32_t v, size_t n)
{
	int d0, d1;
	asm volatile("rep\n\t"
		     "stosl"
		     : "=&c" (d0), "=&D" (d1)
		     : "a" (v), "1" (s), "0" (n)
		     : "memory");
	return s;
}

Linus Torvalds's avatar
Linus Torvalds committed
359 360 361 362
/*
 * find the first occurrence of byte 'c', or 1 past the area if none
 */
#define __HAVE_ARCH_MEMSCAN
363
extern void *memscan(void *addr, int c, size_t size);
Linus Torvalds's avatar
Linus Torvalds committed
364 365 366

#endif /* __KERNEL__ */

367
#endif /* _ASM_X86_STRING_32_H */