]>
Commit | Line | Data |
---|---|---|
f52207a4 GS |
1 | #include "git-compat-util.h" |
2 | #include "bloom.h" | |
ed591feb GS |
3 | #include "diff.h" |
4 | #include "diffcore.h" | |
5 | #include "revision.h" | |
6 | #include "hashmap.h" | |
1217c03e GS |
7 | #include "commit-graph.h" |
8 | #include "commit.h" | |
ed591feb GS |
9 | |
10 | define_commit_slab(bloom_filter_slab, struct bloom_filter); | |
11 | ||
066b70ae | 12 | static struct bloom_filter_slab bloom_filters; |
ed591feb GS |
13 | |
14 | struct pathmap_hash_entry { | |
15 | struct hashmap_entry entry; | |
16 | const char path[FLEX_ARRAY]; | |
17 | }; | |
f52207a4 GS |
18 | |
19 | static uint32_t rotate_left(uint32_t value, int32_t count) | |
20 | { | |
21 | uint32_t mask = 8 * sizeof(uint32_t) - 1; | |
22 | count &= mask; | |
23 | return ((value << count) | (value >> ((-count) & mask))); | |
24 | } | |
25 | ||
f1294eaf GS |
26 | static inline unsigned char get_bitmask(uint32_t pos) |
27 | { | |
28 | return ((unsigned char)1) << (pos & (BITS_PER_WORD - 1)); | |
29 | } | |
30 | ||
1217c03e | 31 | static int load_bloom_filter_from_graph(struct commit_graph *g, |
eb591e42 DS |
32 | struct bloom_filter *filter, |
33 | struct commit *c) | |
1217c03e GS |
34 | { |
35 | uint32_t lex_pos, start_index, end_index; | |
36 | ||
37 | while (c->graph_pos < g->num_commits_in_base) | |
38 | g = g->base_graph; | |
39 | ||
40 | /* The commit graph commit 'c' lives in doesn't carry bloom filters. */ | |
41 | if (!g->chunk_bloom_indexes) | |
42 | return 0; | |
43 | ||
44 | lex_pos = c->graph_pos - g->num_commits_in_base; | |
45 | ||
46 | end_index = get_be32(g->chunk_bloom_indexes + 4 * lex_pos); | |
47 | ||
48 | if (lex_pos > 0) | |
49 | start_index = get_be32(g->chunk_bloom_indexes + 4 * (lex_pos - 1)); | |
50 | else | |
51 | start_index = 0; | |
52 | ||
53 | filter->len = end_index - start_index; | |
54 | filter->data = (unsigned char *)(g->chunk_bloom_data + | |
55 | sizeof(unsigned char) * start_index + | |
56 | BLOOMDATA_CHUNK_HEADER_SIZE); | |
57 | ||
58 | return 1; | |
59 | } | |
60 | ||
f52207a4 GS |
61 | /* |
62 | * Calculate the murmur3 32-bit hash value for the given data | |
63 | * using the given seed. | |
64 | * Produces a uniformly distributed hash value. | |
65 | * Not considered to be cryptographically secure. | |
66 | * Implemented as described in https://en.wikipedia.org/wiki/MurmurHash#Algorithm | |
67 | */ | |
68 | uint32_t murmur3_seeded(uint32_t seed, const char *data, size_t len) | |
69 | { | |
70 | const uint32_t c1 = 0xcc9e2d51; | |
71 | const uint32_t c2 = 0x1b873593; | |
72 | const uint32_t r1 = 15; | |
73 | const uint32_t r2 = 13; | |
74 | const uint32_t m = 5; | |
75 | const uint32_t n = 0xe6546b64; | |
76 | int i; | |
77 | uint32_t k1 = 0; | |
78 | const char *tail; | |
79 | ||
80 | int len4 = len / sizeof(uint32_t); | |
81 | ||
82 | uint32_t k; | |
83 | for (i = 0; i < len4; i++) { | |
84 | uint32_t byte1 = (uint32_t)data[4*i]; | |
85 | uint32_t byte2 = ((uint32_t)data[4*i + 1]) << 8; | |
86 | uint32_t byte3 = ((uint32_t)data[4*i + 2]) << 16; | |
87 | uint32_t byte4 = ((uint32_t)data[4*i + 3]) << 24; | |
88 | k = byte1 | byte2 | byte3 | byte4; | |
89 | k *= c1; | |
90 | k = rotate_left(k, r1); | |
91 | k *= c2; | |
92 | ||
93 | seed ^= k; | |
94 | seed = rotate_left(seed, r2) * m + n; | |
95 | } | |
96 | ||
97 | tail = (data + len4 * sizeof(uint32_t)); | |
98 | ||
99 | switch (len & (sizeof(uint32_t) - 1)) { | |
100 | case 3: | |
101 | k1 ^= ((uint32_t)tail[2]) << 16; | |
102 | /*-fallthrough*/ | |
103 | case 2: | |
104 | k1 ^= ((uint32_t)tail[1]) << 8; | |
105 | /*-fallthrough*/ | |
106 | case 1: | |
107 | k1 ^= ((uint32_t)tail[0]) << 0; | |
108 | k1 *= c1; | |
109 | k1 = rotate_left(k1, r1); | |
110 | k1 *= c2; | |
111 | seed ^= k1; | |
112 | break; | |
113 | } | |
114 | ||
115 | seed ^= (uint32_t)len; | |
116 | seed ^= (seed >> 16); | |
117 | seed *= 0x85ebca6b; | |
118 | seed ^= (seed >> 13); | |
119 | seed *= 0xc2b2ae35; | |
120 | seed ^= (seed >> 16); | |
121 | ||
122 | return seed; | |
f1294eaf GS |
123 | } |
124 | ||
125 | void fill_bloom_key(const char *data, | |
eb591e42 DS |
126 | size_t len, |
127 | struct bloom_key *key, | |
128 | const struct bloom_filter_settings *settings) | |
f1294eaf GS |
129 | { |
130 | int i; | |
131 | const uint32_t seed0 = 0x293ae76f; | |
132 | const uint32_t seed1 = 0x7e646e2c; | |
133 | const uint32_t hash0 = murmur3_seeded(seed0, data, len); | |
134 | const uint32_t hash1 = murmur3_seeded(seed1, data, len); | |
135 | ||
136 | key->hashes = (uint32_t *)xcalloc(settings->num_hashes, sizeof(uint32_t)); | |
137 | for (i = 0; i < settings->num_hashes; i++) | |
138 | key->hashes[i] = hash0 + i * hash1; | |
139 | } | |
140 | ||
141 | void add_key_to_filter(const struct bloom_key *key, | |
eb591e42 DS |
142 | struct bloom_filter *filter, |
143 | const struct bloom_filter_settings *settings) | |
f1294eaf GS |
144 | { |
145 | int i; | |
146 | uint64_t mod = filter->len * BITS_PER_WORD; | |
147 | ||
148 | for (i = 0; i < settings->num_hashes; i++) { | |
149 | uint64_t hash_mod = key->hashes[i] % mod; | |
150 | uint64_t block_pos = hash_mod / BITS_PER_WORD; | |
151 | ||
152 | filter->data[block_pos] |= get_bitmask(hash_mod); | |
153 | } | |
154 | } | |
ed591feb GS |
155 | |
156 | void init_bloom_filters(void) | |
157 | { | |
158 | init_bloom_filter_slab(&bloom_filters); | |
159 | } | |
160 | ||
65c1a28b DS |
161 | static int pathmap_cmp(const void *hashmap_cmp_fn_data, |
162 | const struct hashmap_entry *eptr, | |
163 | const struct hashmap_entry *entry_or_key, | |
164 | const void *keydata) | |
165 | { | |
166 | const struct pathmap_hash_entry *e1, *e2; | |
167 | ||
168 | e1 = container_of(eptr, const struct pathmap_hash_entry, entry); | |
169 | e2 = container_of(entry_or_key, const struct pathmap_hash_entry, entry); | |
170 | ||
171 | return strcmp(e1->path, e2->path); | |
172 | } | |
173 | ||
ed591feb | 174 | struct bloom_filter *get_bloom_filter(struct repository *r, |
1217c03e | 175 | struct commit *c, |
eb591e42 | 176 | int compute_if_not_present) |
ed591feb GS |
177 | { |
178 | struct bloom_filter *filter; | |
179 | struct bloom_filter_settings settings = DEFAULT_BLOOM_FILTER_SETTINGS; | |
180 | int i; | |
181 | struct diff_options diffopt; | |
e3696980 | 182 | int max_changes = 512; |
ed591feb GS |
183 | |
184 | if (bloom_filters.slab_size == 0) | |
185 | return NULL; | |
186 | ||
187 | filter = bloom_filter_slab_at(&bloom_filters, c); | |
188 | ||
1217c03e GS |
189 | if (!filter->data) { |
190 | load_commit_graph_info(r, c); | |
191 | if (c->graph_pos != COMMIT_NOT_FROM_GRAPH && | |
192 | r->objects->commit_graph->chunk_bloom_indexes) { | |
193 | if (load_bloom_filter_from_graph(r->objects->commit_graph, filter, c)) | |
194 | return filter; | |
195 | else | |
196 | return NULL; | |
197 | } | |
198 | } | |
199 | ||
200 | if (filter->data || !compute_if_not_present) | |
201 | return filter; | |
202 | ||
ed591feb GS |
203 | repo_diff_setup(r, &diffopt); |
204 | diffopt.flags.recursive = 1; | |
caf388ca | 205 | diffopt.detect_rename = 0; |
e3696980 | 206 | diffopt.max_changes = max_changes; |
ed591feb GS |
207 | diff_setup_done(&diffopt); |
208 | ||
891c17c9 DS |
209 | /* ensure commit is parsed so we have parent information */ |
210 | repo_parse_commit(r, c); | |
211 | ||
ed591feb GS |
212 | if (c->parents) |
213 | diff_tree_oid(&c->parents->item->object.oid, &c->object.oid, "", &diffopt); | |
214 | else | |
215 | diff_tree_oid(NULL, &c->object.oid, "", &diffopt); | |
216 | diffcore_std(&diffopt); | |
217 | ||
2f6775f0 | 218 | if (diffopt.num_changes <= max_changes) { |
ed591feb GS |
219 | struct hashmap pathmap; |
220 | struct pathmap_hash_entry *e; | |
221 | struct hashmap_iter iter; | |
65c1a28b | 222 | hashmap_init(&pathmap, pathmap_cmp, NULL, 0); |
ed591feb GS |
223 | |
224 | for (i = 0; i < diff_queued_diff.nr; i++) { | |
225 | const char *path = diff_queued_diff.queue[i]->two->path; | |
226 | ||
227 | /* | |
65c1a28b DS |
228 | * Add each leading directory of the changed file, i.e. for |
229 | * 'dir/subdir/file' add 'dir' and 'dir/subdir' as well, so | |
230 | * the Bloom filter could be used to speed up commands like | |
231 | * 'git log dir/subdir', too. | |
232 | * | |
233 | * Note that directories are added without the trailing '/'. | |
234 | */ | |
ed591feb GS |
235 | do { |
236 | char *last_slash = strrchr(path, '/'); | |
237 | ||
238 | FLEX_ALLOC_STR(e, path, path); | |
239 | hashmap_entry_init(&e->entry, strhash(path)); | |
65c1a28b DS |
240 | |
241 | if (!hashmap_get(&pathmap, &e->entry, NULL)) | |
242 | hashmap_add(&pathmap, &e->entry); | |
243 | else | |
244 | free(e); | |
ed591feb GS |
245 | |
246 | if (!last_slash) | |
247 | last_slash = (char*)path; | |
248 | *last_slash = '\0'; | |
249 | ||
250 | } while (*path); | |
251 | ||
252 | diff_free_filepair(diff_queued_diff.queue[i]); | |
253 | } | |
254 | ||
255 | filter->len = (hashmap_get_size(&pathmap) * settings.bits_per_entry + BITS_PER_WORD - 1) / BITS_PER_WORD; | |
256 | filter->data = xcalloc(filter->len, sizeof(unsigned char)); | |
257 | ||
258 | hashmap_for_each_entry(&pathmap, &iter, e, entry) { | |
259 | struct bloom_key key; | |
260 | fill_bloom_key(e->path, strlen(e->path), &key, &settings); | |
261 | add_key_to_filter(&key, filter, &settings); | |
262 | } | |
263 | ||
264 | hashmap_free_entries(&pathmap, struct pathmap_hash_entry, entry); | |
265 | } else { | |
266 | for (i = 0; i < diff_queued_diff.nr; i++) | |
267 | diff_free_filepair(diff_queued_diff.queue[i]); | |
268 | filter->data = NULL; | |
269 | filter->len = 0; | |
270 | } | |
271 | ||
272 | free(diff_queued_diff.queue); | |
273 | DIFF_QUEUE_CLEAR(&diff_queued_diff); | |
274 | ||
275 | return filter; | |
276 | } | |
a56b9464 GS |
277 | |
278 | int bloom_filter_contains(const struct bloom_filter *filter, | |
279 | const struct bloom_key *key, | |
280 | const struct bloom_filter_settings *settings) | |
281 | { | |
282 | int i; | |
283 | uint64_t mod = filter->len * BITS_PER_WORD; | |
284 | ||
285 | if (!mod) | |
286 | return -1; | |
287 | ||
288 | for (i = 0; i < settings->num_hashes; i++) { | |
289 | uint64_t hash_mod = key->hashes[i] % mod; | |
290 | uint64_t block_pos = hash_mod / BITS_PER_WORD; | |
291 | if (!(filter->data[block_pos] & get_bitmask(hash_mod))) | |
292 | return 0; | |
293 | } | |
294 | ||
295 | return 1; | |
066b70ae | 296 | } |