Line data Source code
1 : // Copyright (c) 2012-2014 The Bitcoin developers
2 : // Copyright (c) 2017-2020 The PIVX Core developers
3 : // Distributed under the MIT software license, see the accompanying
4 : // file COPYING or http://www.opensource.org/licenses/mit-license.php.
5 :
6 : #include "bloom.h"
7 :
8 : #include "hash.h"
9 : #include "primitives/transaction.h"
10 : #include "script/script.h"
11 : #include "script/standard.h"
12 : #include "random.h"
13 : #include "streams.h"
14 :
15 : #include <math.h>
16 : #include <stdlib.h>
17 :
18 :
19 : #define LN2SQUARED 0.4804530139182014246671025263266649717305529515945455
20 : #define LN2 0.6931471805599453094172321214581765680755001343602552
21 :
22 502 : CBloomFilter::CBloomFilter(const unsigned int nElements, const double nFPRate, const unsigned int nTweakIn, unsigned char nFlagsIn) :
23 : /**
24 : * The ideal size for a bloom filter with a given number of elements and false positive rate is:
25 : * - nElements * log(fp rate) / ln(2)^2
26 : * We ignore filter parameters which will create a bloom filter larger than the protocol limits
27 : */
28 1004 : vData(std::min((unsigned int)(-1 / LN2SQUARED * nElements * log(nFPRate)), MAX_BLOOM_FILTER_SIZE * 8) / 8),
29 : /**
30 : * The ideal number of hash functions is filter size * ln(2) / number of elements
31 : * Again, we ignore filter parameters which will create a bloom filter with more hash functions than the protocol limits
32 : * See https://en.wikipedia.org/wiki/Bloom_filter for an explanation of these formulas
33 : */
34 : isFull(false),
35 : isEmpty(false),
36 1004 : nHashFuncs(std::min((unsigned int)(vData.size() * 8 / nElements * LN2), MAX_HASH_FUNCS)),
37 : nTweak(nTweakIn),
38 502 : nFlags(nFlagsIn)
39 : {
40 502 : }
41 :
42 7131 : inline unsigned int CBloomFilter::Hash(unsigned int nHashNum, const std::vector<unsigned char>& vDataToHash) const
43 : {
44 : // 0xFBA4C795 chosen as it guarantees a reasonable bit difference between nHashNum values.
45 7131 : return MurmurHash3(nHashNum * 0xFBA4C795 + nTweak, vDataToHash) % (vData.size() * 8);
46 : }
47 :
48 343 : void CBloomFilter::insert(const std::vector<unsigned char>& vKey)
49 : {
50 343 : if (isFull)
51 : return;
52 3704 : for (unsigned int i = 0; i < nHashFuncs; i++) {
53 3361 : unsigned int nIndex = Hash(i, vKey);
54 : // Sets bit nIndex of vData
55 3361 : vData[nIndex >> 3] |= (1 << (7 & nIndex));
56 : }
57 343 : isEmpty = false;
58 : }
59 :
60 8 : void CBloomFilter::insert(const COutPoint& outpoint)
61 : {
62 8 : CDataStream stream(SER_NETWORK, PROTOCOL_VERSION);
63 8 : stream << outpoint;
64 16 : std::vector<unsigned char> data(stream.begin(), stream.end());
65 8 : insert(data);
66 8 : }
67 :
68 9 : void CBloomFilter::insert(const uint256& hash)
69 : {
70 9 : std::vector<unsigned char> data(hash.begin(), hash.end());
71 9 : insert(data);
72 9 : }
73 :
74 737 : bool CBloomFilter::contains(const std::vector<unsigned char>& vKey) const
75 : {
76 737 : if (isFull) {
77 : return true;
78 : }
79 737 : if (isEmpty) {
80 : return false;
81 : }
82 4111 : for (unsigned int i = 0; i < nHashFuncs; i++) {
83 3770 : unsigned int nIndex = Hash(i, vKey);
84 : // Checks bit nIndex of vData
85 3770 : if (!(vData[nIndex >> 3] & (1 << (7 & nIndex))))
86 : return false;
87 : }
88 : return true;
89 : }
90 :
91 88 : bool CBloomFilter::contains(const COutPoint& outpoint) const
92 : {
93 88 : CDataStream stream(SER_NETWORK, PROTOCOL_VERSION);
94 88 : stream << outpoint;
95 176 : std::vector<unsigned char> data(stream.begin(), stream.end());
96 176 : return contains(data);
97 : }
98 :
99 76 : bool CBloomFilter::contains(const uint256& hash) const
100 : {
101 76 : std::vector<unsigned char> data(hash.begin(), hash.end());
102 152 : return contains(data);
103 : }
104 :
105 109 : void CBloomFilter::clear()
106 : {
107 109 : vData.assign(vData.size(), 0);
108 109 : isFull = false;
109 109 : isEmpty = true;
110 109 : }
111 :
112 0 : void CBloomFilter::reset(const unsigned int nNewTweak)
113 : {
114 0 : clear();
115 0 : nTweak = nNewTweak;
116 0 : }
117 :
118 0 : bool CBloomFilter::IsWithinSizeConstraints() const
119 : {
120 0 : return vData.size() <= MAX_BLOOM_FILTER_SIZE && nHashFuncs <= MAX_HASH_FUNCS;
121 : }
122 :
123 : /**
124 : * Returns true if this filter will match anything. See {@link org.pivxj.core.BloomFilter#setMatchAll()}
125 : * for when this can be a useful thing to do.
126 : */
127 0 : bool CBloomFilter::MatchesAll() const {
128 0 : for (unsigned char b : vData)
129 0 : if (b != 0xff)
130 0 : return false;
131 0 : return true;
132 : }
133 :
134 : /**
135 : * Copies filter into this. Filter must have the same size, hash function count and nTweak or an
136 : * IllegalArgumentException will be thrown.
137 : */
138 0 : bool CBloomFilter::Merge(const CBloomFilter& filter) {
139 0 : if (!this->MatchesAll() && !filter.MatchesAll()) {
140 0 : if(! (filter.vData.size() == this->vData.size() &&
141 0 : filter.nHashFuncs == this->nHashFuncs &&
142 0 : filter.nTweak == this->nTweak)){
143 : return false;
144 : }
145 0 : for (unsigned int i = 0; i < vData.size(); i++)
146 0 : this->vData[i] |= filter.vData[i];
147 : } else {
148 : // TODO: Check this.
149 0 : this->vData.clear();
150 0 : this->vData[0] = 0xff;
151 : }
152 : return true;
153 : }
154 :
155 13262 : bool CBloomFilter::IsRelevantAndUpdate(const CTransaction& tx)
156 : {
157 13262 : bool fFound = false;
158 : // Match if the filter contains the hash of tx
159 : // for finding tx when they appear in a block
160 13262 : if (isFull)
161 : return true;
162 76 : if (isEmpty)
163 : return false;
164 76 : const uint256& hash = tx.GetHash();
165 76 : if (contains(hash))
166 13 : fFound = true;
167 :
168 199 : for (unsigned int i = 0; i < tx.vout.size(); i++) {
169 123 : const CTxOut& txout = tx.vout[i];
170 : // Match if the filter contains any arbitrary script data element in any scriptPubKey in tx
171 : // If this matches, also add the specific output that was matched.
172 : // This means clients don't have to update the filter themselves when a new relevant tx
173 : // is discovered in order to find spending transactions, which avoids round-tripping and race conditions.
174 123 : CScript::const_iterator pc = txout.scriptPubKey.begin();
175 246 : std::vector<unsigned char> data;
176 1206 : while (pc < txout.scriptPubKey.end()) {
177 490 : opcodetype opcode;
178 490 : if (!txout.scriptPubKey.GetOp(pc, opcode, data)){
179 : break;
180 : }
181 :
182 490 : if (data.size() != 0 && contains(data)) {
183 10 : fFound = true;
184 10 : if ((nFlags & BLOOM_UPDATE_MASK) == BLOOM_UPDATE_ALL)
185 4 : insert(COutPoint(hash, i));
186 6 : else if ((nFlags & BLOOM_UPDATE_MASK) == BLOOM_UPDATE_P2PUBKEY_ONLY) {
187 2 : txnouttype type;
188 4 : std::vector<std::vector<unsigned char> > vSolutions;
189 2 : if (Solver(txout.scriptPubKey, type, vSolutions) &&
190 2 : (type == TX_PUBKEY || type == TX_MULTISIG))
191 1 : insert(COutPoint(hash, i));
192 : }
193 : break;
194 : }
195 : }
196 : }
197 :
198 76 : if (fFound)
199 : return true;
200 :
201 131 : for (const CTxIn& txin : tx.vin) {
202 : // Match if the filter contains an outpoint tx spends
203 84 : if (contains(txin.prevout))
204 6 : return true;
205 :
206 : // Match if the filter contains any arbitrary script data element in any scriptSig in tx
207 80 : CScript::const_iterator pc = txin.scriptSig.begin();
208 158 : std::vector<unsigned char> data;
209 416 : while (pc < txin.scriptSig.end()) {
210 130 : opcodetype opcode;
211 130 : if (!txin.scriptSig.GetOp(pc, opcode, data))
212 : break;
213 130 : if (data.size() != 0 && contains(data)) {
214 2 : return true;
215 : }
216 : }
217 : }
218 :
219 47 : return false;
220 : }
221 :
222 0 : void CBloomFilter::UpdateEmptyFull()
223 : {
224 0 : bool full = true;
225 0 : bool empty = true;
226 0 : for (unsigned int i = 0; i < vData.size(); i++) {
227 0 : full &= vData[i] == 0xff;
228 0 : empty &= vData[i] == 0;
229 : }
230 0 : isFull = full;
231 0 : isEmpty = empty;
232 0 : }
233 :
234 3358 : CRollingBloomFilter::CRollingBloomFilter(const unsigned int nElements, const double fpRate)
235 : {
236 3358 : double logFpRate = log(fpRate);
237 : /* The optimal number of hash functions is log(fpRate) / log(0.5), but
238 : * restrict it to the range 1-50. */
239 3358 : nHashFuncs = std::max(1, std::min((int)round(logFpRate / log(0.5)), 50));
240 : /* In this rolling bloom filter, we'll store between 2 and 3 generations of nElements / 2 entries. */
241 3358 : nEntriesPerGeneration = (nElements + 1) / 2;
242 3358 : uint32_t nMaxElements = nEntriesPerGeneration * 3;
243 : /* The maximum fpRate = pow(1.0 - exp(-nHashFuncs * nMaxElements / nFilterBits), nHashFuncs)
244 : * => pow(fpRate, 1.0 / nHashFuncs) = 1.0 - exp(-nHashFuncs * nMaxElements / nFilterBits)
245 : * => 1.0 - pow(fpRate, 1.0 / nHashFuncs) = exp(-nHashFuncs * nMaxElements / nFilterBits)
246 : * => log(1.0 - pow(fpRate, 1.0 / nHashFuncs)) = -nHashFuncs * nMaxElements / nFilterBits
247 : * => nFilterBits = -nHashFuncs * nMaxElements / log(1.0 - pow(fpRate, 1.0 / nHashFuncs))
248 : * => nFilterBits = -nHashFuncs * nMaxElements / log(1.0 - exp(logFpRate / nHashFuncs))
249 : */
250 3358 : uint32_t nFilterBits = (uint32_t)ceil(-1.0 * nHashFuncs * nMaxElements / log(1.0 - exp(logFpRate / nHashFuncs)));
251 3358 : data.clear();
252 : /* For each data element we need to store 2 bits. If both bits are 0, the
253 : * bit is treated as unset. If the bits are (01), (10), or (11), the bit is
254 : * treated as set in generation 1, 2, or 3 respectively.
255 : * These bits are stored in separate integers: position P corresponds to bit
256 : * (P & 63) of the integers data[(P >> 6) * 2] and data[(P >> 6) * 2 + 1]. */
257 3358 : data.resize(((nFilterBits + 63) / 64) << 1);
258 3358 : reset();
259 3358 : }
260 :
261 : /* Similar to CBloomFilter::Hash */
262 7553205 : static inline uint32_t RollingBloomHash(unsigned int nHashNum, uint32_t nTweak, const std::vector<unsigned char>& vDataToHash) {
263 7553205 : return MurmurHash3(nHashNum * 0xFBA4C795 + nTweak, vDataToHash);
264 : }
265 :
266 :
267 : // A replacement for x % n. This assumes that x and n are 32bit integers, and x is a uniformly random distributed 32bit value
268 : // which should be the case for a good hash.
269 : // See https://lemire.me/blog/2016/06/27/a-fast-alternative-to-the-modulo-reduction/
270 7553205 : static inline uint32_t FastMod(uint32_t x, size_t n) {
271 7553205 : return ((uint64_t)x * (uint64_t)n) >> 32;
272 : }
273 :
274 360996 : void CRollingBloomFilter::insert(const std::vector<unsigned char>& vKey)
275 : {
276 360996 : if (nEntriesThisGeneration == nEntriesPerGeneration) {
277 36 : nEntriesThisGeneration = 0;
278 36 : nGeneration++;
279 36 : if (nGeneration == 4) {
280 11 : nGeneration = 1;
281 : }
282 36 : uint64_t nGenerationMask1 = -(uint64_t)(nGeneration & 1);
283 36 : uint64_t nGenerationMask2 = -(uint64_t)(nGeneration >> 1);
284 : /* Wipe old entries that used this generation number. */
285 68214 : for (uint32_t p = 0; p < data.size(); p += 2) {
286 68178 : uint64_t p1 = data[p], p2 = data[p + 1];
287 68178 : uint64_t mask = (p1 ^ nGenerationMask1) | (p2 ^ nGenerationMask2);
288 68178 : data[p] = p1 & mask;
289 68178 : data[p + 1] = p2 & mask;
290 : }
291 : }
292 360996 : nEntriesThisGeneration++;
293 :
294 7551945 : for (int n = 0; n < nHashFuncs; n++) {
295 7190949 : uint32_t h = RollingBloomHash(n, nTweak, vKey);
296 7190949 : int bit = h & 0x3F;
297 : /* FastMod works with the upper bits of h, so it is safe to ignore that the lower bits of h are already used for bit. */
298 7190949 : uint32_t pos = FastMod(h, data.size());
299 : /* The lowest bit of pos is ignored, and set to zero for the first bit, and to one for the second. */
300 7190949 : data[pos & ~1] = (data[pos & ~1] & ~(((uint64_t)1) << bit)) | ((uint64_t)(nGeneration & 1)) << bit;
301 7190949 : data[pos | 1] = (data[pos | 1] & ~(((uint64_t)1) << bit)) | ((uint64_t)(nGeneration >> 1)) << bit;
302 : }
303 360996 : }
304 :
305 358638 : void CRollingBloomFilter::insert(const uint256& hash)
306 : {
307 358638 : std::vector<unsigned char> data(hash.begin(), hash.end());
308 358638 : insert(data);
309 358638 : }
310 :
311 87217 : bool CRollingBloomFilter::contains(const std::vector<unsigned char>& vKey) const
312 : {
313 377700 : for (int n = 0; n < nHashFuncs; n++) {
314 362257 : uint32_t h = RollingBloomHash(n, nTweak, vKey);
315 362257 : int bit = h & 0x3F;
316 362257 : uint32_t pos = FastMod(h, data.size());
317 : /* If the relevant bit is not set in either data[pos & ~1] or data[pos | 1], the filter does not contain vKey */
318 362257 : if (!(((data[pos & ~1] | data[pos | 1]) >> bit) & 1)) {
319 : return false;
320 : }
321 : }
322 : return true;
323 : }
324 :
325 74558 : bool CRollingBloomFilter::contains(const uint256& hash) const
326 : {
327 74558 : std::vector<unsigned char> data(hash.begin(), hash.end());
328 149116 : return contains(data);
329 : }
330 :
331 5440 : void CRollingBloomFilter::reset()
332 : {
333 5440 : nTweak = GetRand(std::numeric_limits<unsigned int>::max());
334 5440 : nEntriesThisGeneration = 0;
335 5440 : nGeneration = 1;
336 381022700 : for (std::vector<uint64_t>::iterator it = data.begin(); it != data.end(); it++) {
337 381017600 : *it = 0;
338 : }
339 5440 : }
|
