commit-graph: introduce commit_graph_data_slab
The struct commit is used in many contexts. However, members `generation` and `graph_pos` are only used for commit-graph related operations and otherwise waste memory. This wastage would have been more pronounced as we transition to generation number v2, which uses 64-bit generation number instead of current 32-bits. As they are often accessed together, let's introduce struct commit_graph_data and move them to a commit_graph_data slab. While the overall test suite runs just as fast as master, (series: 26m48s, master: 27m34s, faster by 2.87%), certain commands like `git merge-base --is-ancestor` were slowed by 40% as discovered by Szeder Gábor [1]. After minimizing commit-slab access, the slow down persists but is closer to 20%. Derrick Stolee believes the slow down is attributable to the underlying algorithm rather than the slowness of commit-slab access [2] and we will follow-up in a later series. [1]: https://lore.kernel.org/git/20200607195347.GA8232@szeder.dev/ [2]: https://lore.kernel.org/git/13db757a-9412-7f1e-805c-8a028c4ab2b1@gmail.com/ Signed-off-by: Abhishek Kumar <abhishekkumar8222@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Abhishek Kumar committed
Jun 17, 2020 at 14:44 UTC
4844812b9ec9bc6ffdc2da2c54d9146ff4c97d94
2 files changed
+78
-10
commit-graph.c
+68
-10
@@ -87,6 +87,58 @@ static int commit_pos_cmp(const void *va, const void *vb)
87
commit_pos_at(&commit_pos, b);
88
}
89
90
+define_commit_slab(commit_graph_data_slab, struct commit_graph_data);
91
+static struct commit_graph_data_slab commit_graph_data_slab =
92
+ COMMIT_SLAB_INIT(1, commit_graph_data_slab);
93
+
94
+uint32_t commit_graph_position(const struct commit *c)
95
+{
96
+ struct commit_graph_data *data =
97
+ commit_graph_data_slab_peek(&commit_graph_data_slab, c);
98
+
99
+ return data ? data->graph_pos : COMMIT_NOT_FROM_GRAPH;
100
+}
101
+
102
+uint32_t commit_graph_generation(const struct commit *c)
103
+{
104
+ struct commit_graph_data *data =
105
+ commit_graph_data_slab_peek(&commit_graph_data_slab, c);
106
+
107
+ if (!data)
108
+ return GENERATION_NUMBER_INFINITY;
109
+ else if (data->graph_pos == COMMIT_NOT_FROM_GRAPH)
110
+ return GENERATION_NUMBER_INFINITY;
111
+
112
+ return data->generation;
113
+}
114
+
115
+static struct commit_graph_data *commit_graph_data_at(const struct commit *c)
116
+{
117
+ unsigned int i, nth_slab;
118
+ struct commit_graph_data *data =
119
+ commit_graph_data_slab_peek(&commit_graph_data_slab, c);
120
+
121
+ if (data)
122
+ return data;
123
+
124
+ nth_slab = c->index / commit_graph_data_slab.slab_size;
125
+ data = commit_graph_data_slab_at(&commit_graph_data_slab, c);
126
+
127
+ /*
128
+ * commit-slab initializes elements with zero, overwrite this with
129
+ * COMMIT_NOT_FROM_GRAPH for graph_pos.
130
+ *
131
+ * We avoid initializing generation with checking if graph position
132
+ * is not COMMIT_NOT_FROM_GRAPH.
133
+ */
134
+ for (i = 0; i < commit_graph_data_slab.slab_size; i++) {
135
+ commit_graph_data_slab.slab[nth_slab][i].graph_pos =
136
+ COMMIT_NOT_FROM_GRAPH;
137
+ }
138
+
139
+ return data;
140
+}
141
+
142
static int commit_gen_cmp(const void *va, const void *vb)
143
{
144
const struct commit *a = *(const struct commit **)va;
@@ -1020,7 +1072,7 @@ static void write_graph_chunk_data(struct hashfile *f, int hash_len,
1072
else
1073
packedDate[0] = 0;
1074
1023
- packedDate[0] |= htonl((*list)->generation << 2);
1075
+ packedDate[0] |= htonl(commit_graph_data_at(*list)->generation << 2);
1076
1077
packedDate[1] = htonl((*list)->date);
1078
hashwrite(f, packedDate, 8);
@@ -1251,9 +1303,11 @@ static void compute_generation_numbers(struct write_commit_graph_context *ctx)
1303
_("Computing commit graph generation numbers"),
1304
ctx->commits.nr);
1305
for (i = 0; i < ctx->commits.nr; i++) {
1306
+ uint32_t generation = commit_graph_data_at(ctx->commits.list[i])->generation;
1307
+
1308
display_progress(ctx->progress, i + 1);
1255
- if (ctx->commits.list[i]->generation != GENERATION_NUMBER_INFINITY &&
1256
- ctx->commits.list[i]->generation != GENERATION_NUMBER_ZERO)
1309
+ if (generation != GENERATION_NUMBER_INFINITY &&
1310
+ generation != GENERATION_NUMBER_ZERO)
1311
continue;
1312
1313
commit_list_insert(ctx->commits.list[i], &list);
@@ -1264,22 +1318,26 @@ static void compute_generation_numbers(struct write_commit_graph_context *ctx)
1318
uint32_t max_generation = 0;
1319
1320
for (parent = current->parents; parent; parent = parent->next) {
1267
- if (parent->item->generation == GENERATION_NUMBER_INFINITY ||
1268
- parent->item->generation == GENERATION_NUMBER_ZERO) {
1321
+ generation = commit_graph_data_at(parent->item)->generation;
1322
+
1323
+ if (generation == GENERATION_NUMBER_INFINITY ||
1324
+ generation == GENERATION_NUMBER_ZERO) {
1325
all_parents_computed = 0;
1326
commit_list_insert(parent->item, &list);
1327
break;
1272
- } else if (parent->item->generation > max_generation) {
1273
- max_generation = parent->item->generation;
1328
+ } else if (generation > max_generation) {
1329
+ max_generation = generation;
1330
}
1331
}
1332
1333
if (all_parents_computed) {
1278
- current->generation = max_generation + 1;
1334
+ struct commit_graph_data *data = commit_graph_data_at(current);
1335
+
1336
+ data->generation = max_generation + 1;
1337
pop_commit(&list);
1338
1281
- if (current->generation > GENERATION_NUMBER_MAX)
1282
- current->generation = GENERATION_NUMBER_MAX;
1339
+ if (data->generation > GENERATION_NUMBER_MAX)
1340
+ data->generation = GENERATION_NUMBER_MAX;
1341
}
1342
}
1343
}
commit-graph.h
+10
@@ -135,4 +135,14 @@ void free_commit_graph(struct commit_graph *);
135
*/
136
void disable_commit_graph(struct repository *r);
137
138
+struct commit_graph_data {
139
+ uint32_t graph_pos;
140
+ uint32_t generation;
141
+};
142
+
143
+/*
144
+ * Commits should be parsed before accessing generation, graph positions.
145
+ */
146
+uint32_t commit_graph_generation(const struct commit *);
147
+uint32_t commit_graph_position(const struct commit *);
148
#endif