• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1 #ifndef Py_HASH_H
2 
3 #define Py_HASH_H
4 #ifdef __cplusplus
5 extern "C" {
6 #endif
7 
8 /* Helpers for hash functions */
9 #ifndef Py_LIMITED_API
10 PyAPI_FUNC(Py_hash_t) _Py_HashDouble(PyObject *, double);
11 PyAPI_FUNC(Py_hash_t) _Py_HashPointer(const void*);
12 // Similar to _Py_HashPointer(), but don't replace -1 with -2
13 PyAPI_FUNC(Py_hash_t) _Py_HashPointerRaw(const void*);
14 PyAPI_FUNC(Py_hash_t) _Py_HashBytes(const void*, Py_ssize_t);
15 #endif
16 
17 /* Prime multiplier used in string and various other hashes. */
18 #define _PyHASH_MULTIPLIER 1000003UL  /* 0xf4243 */
19 
20 /* Parameters used for the numeric hash implementation.  See notes for
21    _Py_HashDouble in Python/pyhash.c.  Numeric hashes are based on
22    reduction modulo the prime 2**_PyHASH_BITS - 1. */
23 
24 #if SIZEOF_VOID_P >= 8
25 #  define _PyHASH_BITS 61
26 #else
27 #  define _PyHASH_BITS 31
28 #endif
29 
30 #define _PyHASH_MODULUS (((size_t)1 << _PyHASH_BITS) - 1)
31 #define _PyHASH_INF 314159
32 #define _PyHASH_IMAG _PyHASH_MULTIPLIER
33 
34 
35 /* hash secret
36  *
37  * memory layout on 64 bit systems
38  *   cccccccc cccccccc cccccccc  uc -- unsigned char[24]
39  *   pppppppp ssssssss ........  fnv -- two Py_hash_t
40  *   k0k0k0k0 k1k1k1k1 ........  siphash -- two uint64_t
41  *   ........ ........ ssssssss  djbx33a -- 16 bytes padding + one Py_hash_t
42  *   ........ ........ eeeeeeee  pyexpat XML hash salt
43  *
44  * memory layout on 32 bit systems
45  *   cccccccc cccccccc cccccccc  uc
46  *   ppppssss ........ ........  fnv -- two Py_hash_t
47  *   k0k0k0k0 k1k1k1k1 ........  siphash -- two uint64_t (*)
48  *   ........ ........ ssss....  djbx33a -- 16 bytes padding + one Py_hash_t
49  *   ........ ........ eeee....  pyexpat XML hash salt
50  *
51  * (*) The siphash member may not be available on 32 bit platforms without
52  *     an unsigned int64 data type.
53  */
54 #ifndef Py_LIMITED_API
55 typedef union {
56     /* ensure 24 bytes */
57     unsigned char uc[24];
58     /* two Py_hash_t for FNV */
59     struct {
60         Py_hash_t prefix;
61         Py_hash_t suffix;
62     } fnv;
63     /* two uint64 for SipHash24 */
64     struct {
65         uint64_t k0;
66         uint64_t k1;
67     } siphash;
68     /* a different (!) Py_hash_t for small string optimization */
69     struct {
70         unsigned char padding[16];
71         Py_hash_t suffix;
72     } djbx33a;
73     struct {
74         unsigned char padding[16];
75         Py_hash_t hashsalt;
76     } expat;
77 } _Py_HashSecret_t;
78 PyAPI_DATA(_Py_HashSecret_t) _Py_HashSecret;
79 
80 #ifdef Py_DEBUG
81 PyAPI_DATA(int) _Py_HashSecret_Initialized;
82 #endif
83 
84 
85 /* hash function definition */
86 typedef struct {
87     Py_hash_t (*const hash)(const void *, Py_ssize_t);
88     const char *name;
89     const int hash_bits;
90     const int seed_bits;
91 } PyHash_FuncDef;
92 
93 PyAPI_FUNC(PyHash_FuncDef*) PyHash_GetFuncDef(void);
94 #endif
95 
96 
97 /* cutoff for small string DJBX33A optimization in range [1, cutoff).
98  *
99  * About 50% of the strings in a typical Python application are smaller than
100  * 6 to 7 chars. However DJBX33A is vulnerable to hash collision attacks.
101  * NEVER use DJBX33A for long strings!
102  *
103  * A Py_HASH_CUTOFF of 0 disables small string optimization. 32 bit platforms
104  * should use a smaller cutoff because it is easier to create colliding
105  * strings. A cutoff of 7 on 64bit platforms and 5 on 32bit platforms should
106  * provide a decent safety margin.
107  */
108 #ifndef Py_HASH_CUTOFF
109 #  define Py_HASH_CUTOFF 0
110 #elif (Py_HASH_CUTOFF > 7 || Py_HASH_CUTOFF < 0)
111 #  error Py_HASH_CUTOFF must in range 0...7.
112 #endif /* Py_HASH_CUTOFF */
113 
114 
115 /* hash algorithm selection
116  *
117  * The values for Py_HASH_SIPHASH24 and Py_HASH_FNV are hard-coded in the
118  * configure script.
119  *
120  * - FNV is available on all platforms and architectures.
121  * - SIPHASH24 only works on platforms that don't require aligned memory for integers.
122  * - With EXTERNAL embedders can provide an alternative implementation with::
123  *
124  *     PyHash_FuncDef PyHash_Func = {...};
125  *
126  * XXX: Figure out __declspec() for extern PyHash_FuncDef.
127  */
128 #define Py_HASH_EXTERNAL 0
129 #define Py_HASH_SIPHASH24 1
130 #define Py_HASH_FNV 2
131 
132 #ifndef Py_HASH_ALGORITHM
133 #  ifndef HAVE_ALIGNED_REQUIRED
134 #    define Py_HASH_ALGORITHM Py_HASH_SIPHASH24
135 #  else
136 #    define Py_HASH_ALGORITHM Py_HASH_FNV
137 #  endif /* uint64_t && uint32_t && aligned */
138 #endif /* Py_HASH_ALGORITHM */
139 
140 #ifdef __cplusplus
141 }
142 #endif
143 
144 #endif /* !Py_HASH_H */
145