xref: /freebsd/contrib/xz/src/liblzma/common/memcmplen.h (revision 09a53ad8f1318c5daae6cfb19d97f4f6459f0013)
1 ///////////////////////////////////////////////////////////////////////////////
2 //
3 /// \file       memcmplen.h
4 /// \brief      Optimized comparison of two buffers
5 //
6 //  Author:     Lasse Collin
7 //
8 //  This file has been put into the public domain.
9 //  You can do whatever you want with this file.
10 //
11 ///////////////////////////////////////////////////////////////////////////////
12 
13 #ifndef LZMA_MEMCMPLEN_H
14 #define LZMA_MEMCMPLEN_H
15 
16 #include "common.h"
17 
18 #ifdef HAVE_IMMINTRIN_H
19 #	include <immintrin.h>
20 #endif
21 
22 
23 /// Find out how many equal bytes the two buffers have.
24 ///
25 /// \param      buf1    First buffer
26 /// \param      buf2    Second buffer
27 /// \param      len     How many bytes have already been compared and will
28 ///                     be assumed to match
29 /// \param      limit   How many bytes to compare at most, including the
30 ///                     already-compared bytes. This must be significantly
31 ///                     smaller than UINT32_MAX to avoid integer overflows.
32 ///                     Up to LZMA_MEMCMPLEN_EXTRA bytes may be read past
33 ///                     the specified limit from both buf1 and buf2.
34 ///
35 /// \return     Number of equal bytes in the buffers is returned.
36 ///             This is always at least len and at most limit.
37 ///
38 /// \note       LZMA_MEMCMPLEN_EXTRA defines how many extra bytes may be read.
39 ///             It's rounded up to 2^n. This extra amount needs to be
40 ///             allocated in the buffers being used. It needs to be
41 ///             initialized too to keep Valgrind quiet.
42 static inline uint32_t lzma_attribute((__always_inline__))
43 lzma_memcmplen(const uint8_t *buf1, const uint8_t *buf2,
44 		uint32_t len, uint32_t limit)
45 {
46 	assert(len <= limit);
47 	assert(limit <= UINT32_MAX / 2);
48 
49 #if defined(TUKLIB_FAST_UNALIGNED_ACCESS) \
50 		&& ((TUKLIB_GNUC_REQ(3, 4) && defined(__x86_64__)) \
51 			|| (defined(__INTEL_COMPILER) && defined(__x86_64__)) \
52 			|| (defined(__INTEL_COMPILER) && defined(_M_X64)) \
53 			|| (defined(_MSC_VER) && defined(_M_X64)))
54 	// NOTE: This will use 64-bit unaligned access which
55 	// TUKLIB_FAST_UNALIGNED_ACCESS wasn't meant to permit, but
56 	// it's convenient here at least as long as it's x86-64 only.
57 	//
58 	// I keep this x86-64 only for now since that's where I know this
59 	// to be a good method. This may be fine on other 64-bit CPUs too.
60 	// On big endian one should use xor instead of subtraction and switch
61 	// to __builtin_clzll().
62 #define LZMA_MEMCMPLEN_EXTRA 8
63 	while (len < limit) {
64 		const uint64_t x = *(const uint64_t *)(buf1 + len)
65 				- *(const uint64_t *)(buf2 + len);
66 		if (x != 0) {
67 #	if defined(_M_X64) // MSVC or Intel C compiler on Windows
68 			unsigned long tmp;
69 			_BitScanForward64(&tmp, x);
70 			len += (uint32_t)tmp >> 3;
71 #	else // GCC, clang, or Intel C compiler
72 			len += (uint32_t)__builtin_ctzll(x) >> 3;
73 #	endif
74 			return my_min(len, limit);
75 		}
76 
77 		len += 8;
78 	}
79 
80 	return limit;
81 
82 #elif defined(TUKLIB_FAST_UNALIGNED_ACCESS) \
83 		&& defined(HAVE__MM_MOVEMASK_EPI8) \
84 		&& ((defined(__GNUC__) && defined(__SSE2_MATH__)) \
85 			|| (defined(__INTEL_COMPILER) && defined(__SSE2__)) \
86 			|| (defined(_MSC_VER) && defined(_M_IX86_FP) \
87 				&& _M_IX86_FP >= 2))
88 	// NOTE: Like above, this will use 128-bit unaligned access which
89 	// TUKLIB_FAST_UNALIGNED_ACCESS wasn't meant to permit.
90 	//
91 	// SSE2 version for 32-bit and 64-bit x86. On x86-64 the above
92 	// version is sometimes significantly faster and sometimes
93 	// slightly slower than this SSE2 version, so this SSE2
94 	// version isn't used on x86-64.
95 #	define LZMA_MEMCMPLEN_EXTRA 16
96 	while (len < limit) {
97 		const uint32_t x = 0xFFFF ^ _mm_movemask_epi8(_mm_cmpeq_epi8(
98 			_mm_loadu_si128((const __m128i *)(buf1 + len)),
99 			_mm_loadu_si128((const __m128i *)(buf2 + len))));
100 
101 		if (x != 0) {
102 #	if defined(__INTEL_COMPILER)
103 			len += _bit_scan_forward(x);
104 #	elif defined(_MSC_VER)
105 			unsigned long tmp;
106 			_BitScanForward(&tmp, x);
107 			len += tmp;
108 #	else
109 			len += __builtin_ctz(x);
110 #	endif
111 			return my_min(len, limit);
112 		}
113 
114 		len += 16;
115 	}
116 
117 	return limit;
118 
119 #elif defined(TUKLIB_FAST_UNALIGNED_ACCESS) && !defined(WORDS_BIGENDIAN)
120 	// Generic 32-bit little endian method
121 #	define LZMA_MEMCMPLEN_EXTRA 4
122 	while (len < limit) {
123 		uint32_t x = *(const uint32_t *)(buf1 + len)
124 				- *(const uint32_t *)(buf2 + len);
125 		if (x != 0) {
126 			if ((x & 0xFFFF) == 0) {
127 				len += 2;
128 				x >>= 16;
129 			}
130 
131 			if ((x & 0xFF) == 0)
132 				++len;
133 
134 			return my_min(len, limit);
135 		}
136 
137 		len += 4;
138 	}
139 
140 	return limit;
141 
142 #elif defined(TUKLIB_FAST_UNALIGNED_ACCESS) && defined(WORDS_BIGENDIAN)
143 	// Generic 32-bit big endian method
144 #	define LZMA_MEMCMPLEN_EXTRA 4
145 	while (len < limit) {
146 		uint32_t x = *(const uint32_t *)(buf1 + len)
147 				^ *(const uint32_t *)(buf2 + len);
148 		if (x != 0) {
149 			if ((x & 0xFFFF0000) == 0) {
150 				len += 2;
151 				x <<= 16;
152 			}
153 
154 			if ((x & 0xFF000000) == 0)
155 				++len;
156 
157 			return my_min(len, limit);
158 		}
159 
160 		len += 4;
161 	}
162 
163 	return limit;
164 
165 #else
166 	// Simple portable version that doesn't use unaligned access.
167 #	define LZMA_MEMCMPLEN_EXTRA 0
168 	while (len < limit && buf1[len] == buf2[len])
169 		++len;
170 
171 	return len;
172 #endif
173 }
174 
175 #endif
176