Developer Documentation
snappy-internal.h
1 // Copyright 2008 Google Inc. All Rights Reserved.
2 //
3 // Redistribution and use in source and binary forms, with or without
4 // modification, are permitted provided that the following conditions are
5 // met:
6 //
7 // * Redistributions of source code must retain the above copyright
8 // notice, this list of conditions and the following disclaimer.
9 // * Redistributions in binary form must reproduce the above
10 // copyright notice, this list of conditions and the following disclaimer
11 // in the documentation and/or other materials provided with the
12 // distribution.
13 // * Neither the name of Google Inc. nor the names of its
14 // contributors may be used to endorse or promote products derived from
15 // this software without specific prior written permission.
16 //
17 // THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
18 // "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
19 // LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
20 // A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
21 // OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
22 // SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
23 // LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
24 // DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
25 // THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
26 // (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
27 // OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
28 //
29 // Internals shared between the Snappy implementation and its unittest.
30 
31 #ifndef THIRD_PARTY_SNAPPY_SNAPPY_INTERNAL_H_
32 #define THIRD_PARTY_SNAPPY_SNAPPY_INTERNAL_H_
33 
34 #include "snappy-stubs-internal.h"
35 
36 namespace snappy {
37 namespace internal {
38 
40  public:
41  WorkingMemory() : large_table_(NULL) { }
42  ~WorkingMemory() { delete[] large_table_; }
43 
44  // Allocates and clears a hash table using memory in "*this",
45  // stores the number of buckets in "*table_size" and returns a pointer to
46  // the base of the hash table.
47  uint16* GetHashTable(size_t input_size, int* table_size);
48 
49  private:
50  uint16 small_table_[1<<10]; // 2KB
51  uint16* large_table_; // Allocated only when needed
52 
53  DISALLOW_COPY_AND_ASSIGN(WorkingMemory);
54 };
55 
56 // Flat array compression that does not emit the "uncompressed length"
57 // prefix. Compresses "input" string to the "*op" buffer.
58 //
59 // REQUIRES: "input_length <= kBlockSize"
60 // REQUIRES: "op" points to an array of memory that is at least
61 // "MaxCompressedLength(input_length)" in size.
62 // REQUIRES: All elements in "table[0..table_size-1]" are initialized to zero.
63 // REQUIRES: "table_size" is a power of two
64 //
65 // Returns an "end" pointer into "op" buffer.
66 // "end - op" is the compressed size of "input".
67 char* CompressFragment(const char* input,
68  size_t input_length,
69  char* op,
70  uint16* table,
71  const int table_size);
72 
73 // Return the largest n such that
74 //
75 // s1[0,n-1] == s2[0,n-1]
76 // and n <= (s2_limit - s2).
77 //
78 // Does not read *s2_limit or beyond.
79 // Does not read *(s1 + (s2_limit - s2)) or beyond.
80 // Requires that s2_limit >= s2.
81 //
82 // Separate implementation for x86_64, for speed. Uses the fact that
83 // x86_64 is little endian.
84 #if defined(ARCH_K8)
85 static inline int FindMatchLength(const char* s1,
86  const char* s2,
87  const char* s2_limit) {
88  assert(s2_limit >= s2);
89  int matched = 0;
90 
91  // Find out how long the match is. We loop over the data 64 bits at a
92  // time until we find a 64-bit block that doesn't match; then we find
93  // the first non-matching bit and use that to calculate the total
94  // length of the match.
95  while (PREDICT_TRUE(s2 <= s2_limit - 8)) {
96  if (UNALIGNED_LOAD64(s2) == UNALIGNED_LOAD64(s1 + matched)) {
97  s2 += 8;
98  matched += 8;
99  } else {
100  // On current (mid-2008) Opteron models there is a 3% more
101  // efficient code sequence to find the first non-matching byte.
102  // However, what follows is ~10% better on Intel Core 2 and newer,
103  // and we expect AMD's bsf instruction to improve.
104  uint64 x = UNALIGNED_LOAD64(s2) ^ UNALIGNED_LOAD64(s1 + matched);
105  int matching_bits = Bits::FindLSBSetNonZero64(x);
106  matched += matching_bits >> 3;
107  return matched;
108  }
109  }
110  while (PREDICT_TRUE(s2 < s2_limit)) {
111  if (s1[matched] == *s2) {
112  ++s2;
113  ++matched;
114  } else {
115  return matched;
116  }
117  }
118  return matched;
119 }
120 #else
121 static inline int FindMatchLength(const char* s1,
122  const char* s2,
123  const char* s2_limit) {
124  // Implementation based on the x86-64 version, above.
125  assert(s2_limit >= s2);
126  int matched = 0;
127 
128  while (s2 <= s2_limit - 4 &&
129  UNALIGNED_LOAD32(s2) == UNALIGNED_LOAD32(s1 + matched)) {
130  s2 += 4;
131  matched += 4;
132  }
133  if (LittleEndian::IsLittleEndian() && s2 <= s2_limit - 4) {
134  uint32 x = UNALIGNED_LOAD32(s2) ^ UNALIGNED_LOAD32(s1 + matched);
135  int matching_bits = Bits::FindLSBSetNonZero(x);
136  matched += matching_bits >> 3;
137  } else {
138  while ((s2 < s2_limit) && (s1[matched] == *s2)) {
139  ++s2;
140  ++matched;
141  }
142  }
143  return matched;
144 }
145 #endif
146 
147 } // end namespace internal
148 } // end namespace snappy
149 
150 #endif // THIRD_PARTY_SNAPPY_SNAPPY_INTERNAL_H_