]>
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 | ||
f32dde8c DS |
141 | void clear_bloom_key(struct bloom_key *key) |
142 | { | |
143 | FREE_AND_NULL(key->hashes); | |
144 | } | |
145 | ||
f1294eaf | 146 | void add_key_to_filter(const struct bloom_key *key, |
eb591e42 DS |
147 | struct bloom_filter *filter, |
148 | const struct bloom_filter_settings *settings) | |
f1294eaf GS |
149 | { |
150 | int i; | |
151 | uint64_t mod = filter->len * BITS_PER_WORD; | |
152 | ||
153 | for (i = 0; i < settings->num_hashes; i++) { | |
154 | uint64_t hash_mod = key->hashes[i] % mod; | |
155 | uint64_t block_pos = hash_mod / BITS_PER_WORD; | |
156 | ||
157 | filter->data[block_pos] |= get_bitmask(hash_mod); | |
158 | } | |
159 | } | |
ed591feb GS |
160 | |
161 | void init_bloom_filters(void) | |
162 | { | |
163 | init_bloom_filter_slab(&bloom_filters); | |
164 | } | |
165 | ||
65c1a28b DS |
166 | static int pathmap_cmp(const void *hashmap_cmp_fn_data, |
167 | const struct hashmap_entry *eptr, | |
168 | const struct hashmap_entry *entry_or_key, | |
169 | const void *keydata) | |
170 | { | |
171 | const struct pathmap_hash_entry *e1, *e2; | |
172 | ||
173 | e1 = container_of(eptr, const struct pathmap_hash_entry, entry); | |
174 | e2 = container_of(entry_or_key, const struct pathmap_hash_entry, entry); | |
175 | ||
176 | return strcmp(e1->path, e2->path); | |
177 | } | |
178 | ||
ed591feb | 179 | struct bloom_filter *get_bloom_filter(struct repository *r, |
1217c03e | 180 | struct commit *c, |
eb591e42 | 181 | int compute_if_not_present) |
ed591feb GS |
182 | { |
183 | struct bloom_filter *filter; | |
184 | struct bloom_filter_settings settings = DEFAULT_BLOOM_FILTER_SETTINGS; | |
185 | int i; | |
186 | struct diff_options diffopt; | |
e3696980 | 187 | int max_changes = 512; |
ed591feb GS |
188 | |
189 | if (bloom_filters.slab_size == 0) | |
190 | return NULL; | |
191 | ||
192 | filter = bloom_filter_slab_at(&bloom_filters, c); | |
193 | ||
1217c03e GS |
194 | if (!filter->data) { |
195 | load_commit_graph_info(r, c); | |
196 | if (c->graph_pos != COMMIT_NOT_FROM_GRAPH && | |
197 | r->objects->commit_graph->chunk_bloom_indexes) { | |
198 | if (load_bloom_filter_from_graph(r->objects->commit_graph, filter, c)) | |
199 | return filter; | |
200 | else | |
201 | return NULL; | |
202 | } | |
203 | } | |
204 | ||
205 | if (filter->data || !compute_if_not_present) | |
206 | return filter; | |
207 | ||
ed591feb GS |
208 | repo_diff_setup(r, &diffopt); |
209 | diffopt.flags.recursive = 1; | |
caf388ca | 210 | diffopt.detect_rename = 0; |
e3696980 | 211 | diffopt.max_changes = max_changes; |
ed591feb GS |
212 | diff_setup_done(&diffopt); |
213 | ||
891c17c9 DS |
214 | /* ensure commit is parsed so we have parent information */ |
215 | repo_parse_commit(r, c); | |
216 | ||
ed591feb GS |
217 | if (c->parents) |
218 | diff_tree_oid(&c->parents->item->object.oid, &c->object.oid, "", &diffopt); | |
219 | else | |
220 | diff_tree_oid(NULL, &c->object.oid, "", &diffopt); | |
221 | diffcore_std(&diffopt); | |
222 | ||
2f6775f0 | 223 | if (diffopt.num_changes <= max_changes) { |
ed591feb GS |
224 | struct hashmap pathmap; |
225 | struct pathmap_hash_entry *e; | |
226 | struct hashmap_iter iter; | |
65c1a28b | 227 | hashmap_init(&pathmap, pathmap_cmp, NULL, 0); |
ed591feb GS |
228 | |
229 | for (i = 0; i < diff_queued_diff.nr; i++) { | |
230 | const char *path = diff_queued_diff.queue[i]->two->path; | |
231 | ||
232 | /* | |
65c1a28b DS |
233 | * Add each leading directory of the changed file, i.e. for |
234 | * 'dir/subdir/file' add 'dir' and 'dir/subdir' as well, so | |
235 | * the Bloom filter could be used to speed up commands like | |
236 | * 'git log dir/subdir', too. | |
237 | * | |
238 | * Note that directories are added without the trailing '/'. | |
239 | */ | |
ed591feb GS |
240 | do { |
241 | char *last_slash = strrchr(path, '/'); | |
242 | ||
243 | FLEX_ALLOC_STR(e, path, path); | |
244 | hashmap_entry_init(&e->entry, strhash(path)); | |
65c1a28b DS |
245 | |
246 | if (!hashmap_get(&pathmap, &e->entry, NULL)) | |
247 | hashmap_add(&pathmap, &e->entry); | |
248 | else | |
249 | free(e); | |
ed591feb GS |
250 | |
251 | if (!last_slash) | |
252 | last_slash = (char*)path; | |
253 | *last_slash = '\0'; | |
254 | ||
255 | } while (*path); | |
256 | ||
257 | diff_free_filepair(diff_queued_diff.queue[i]); | |
258 | } | |
259 | ||
260 | filter->len = (hashmap_get_size(&pathmap) * settings.bits_per_entry + BITS_PER_WORD - 1) / BITS_PER_WORD; | |
261 | filter->data = xcalloc(filter->len, sizeof(unsigned char)); | |
262 | ||
263 | hashmap_for_each_entry(&pathmap, &iter, e, entry) { | |
264 | struct bloom_key key; | |
265 | fill_bloom_key(e->path, strlen(e->path), &key, &settings); | |
266 | add_key_to_filter(&key, filter, &settings); | |
267 | } | |
268 | ||
269 | hashmap_free_entries(&pathmap, struct pathmap_hash_entry, entry); | |
270 | } else { | |
271 | for (i = 0; i < diff_queued_diff.nr; i++) | |
272 | diff_free_filepair(diff_queued_diff.queue[i]); | |
273 | filter->data = NULL; | |
274 | filter->len = 0; | |
275 | } | |
276 | ||
277 | free(diff_queued_diff.queue); | |
278 | DIFF_QUEUE_CLEAR(&diff_queued_diff); | |
279 | ||
280 | return filter; | |
281 | } | |
a56b9464 GS |
282 | |
283 | int bloom_filter_contains(const struct bloom_filter *filter, | |
284 | const struct bloom_key *key, | |
285 | const struct bloom_filter_settings *settings) | |
286 | { | |
287 | int i; | |
288 | uint64_t mod = filter->len * BITS_PER_WORD; | |
289 | ||
290 | if (!mod) | |
291 | return -1; | |
292 | ||
293 | for (i = 0; i < settings->num_hashes; i++) { | |
294 | uint64_t hash_mod = key->hashes[i] % mod; | |
295 | uint64_t block_pos = hash_mod / BITS_PER_WORD; | |
296 | if (!(filter->data[block_pos] & get_bitmask(hash_mod))) | |
297 | return 0; | |
298 | } | |
299 | ||
300 | return 1; | |
066b70ae | 301 | } |