]> git.ipfire.org Git - thirdparty/git.git/blobdiff - commit-graph.c
commit-graph: minimize commit_graph_data_slab access
[thirdparty/git.git] / commit-graph.c
index 6dc777e2f395ea12dc1a362b21f3b180c07e72bb..fdd1c4fa7c54510006fb9eb7722b822470897fd3 100644 (file)
@@ -18,6 +18,7 @@
 #include "progress.h"
 #include "bloom.h"
 #include "commit-slab.h"
+#include "shallow.h"
 
 void git_test_write_commit_graph_or_die(void)
 {
@@ -86,15 +87,69 @@ static int commit_pos_cmp(const void *va, const void *vb)
               commit_pos_at(&commit_pos, b);
 }
 
+define_commit_slab(commit_graph_data_slab, struct commit_graph_data);
+static struct commit_graph_data_slab commit_graph_data_slab =
+       COMMIT_SLAB_INIT(1, commit_graph_data_slab);
+
+uint32_t commit_graph_position(const struct commit *c)
+{
+       struct commit_graph_data *data =
+               commit_graph_data_slab_peek(&commit_graph_data_slab, c);
+
+       return data ? data->graph_pos : COMMIT_NOT_FROM_GRAPH;
+}
+
+uint32_t commit_graph_generation(const struct commit *c)
+{
+       struct commit_graph_data *data =
+               commit_graph_data_slab_peek(&commit_graph_data_slab, c);
+
+       if (!data)
+               return GENERATION_NUMBER_INFINITY;
+       else if (data->graph_pos == COMMIT_NOT_FROM_GRAPH)
+               return GENERATION_NUMBER_INFINITY;
+
+       return data->generation;
+}
+
+static struct commit_graph_data *commit_graph_data_at(const struct commit *c)
+{
+       unsigned int i, nth_slab;
+       struct commit_graph_data *data =
+               commit_graph_data_slab_peek(&commit_graph_data_slab, c);
+
+       if (data)
+               return data;
+
+       nth_slab = c->index / commit_graph_data_slab.slab_size;
+       data = commit_graph_data_slab_at(&commit_graph_data_slab, c);
+
+       /*
+        * commit-slab initializes elements with zero, overwrite this with
+        * COMMIT_NOT_FROM_GRAPH for graph_pos.
+        *
+        * We avoid initializing generation with checking if graph position
+        * is not COMMIT_NOT_FROM_GRAPH.
+        */
+       for (i = 0; i < commit_graph_data_slab.slab_size; i++) {
+               commit_graph_data_slab.slab[nth_slab][i].graph_pos =
+                       COMMIT_NOT_FROM_GRAPH;
+       }
+
+       return data;
+}
+
 static int commit_gen_cmp(const void *va, const void *vb)
 {
        const struct commit *a = *(const struct commit **)va;
        const struct commit *b = *(const struct commit **)vb;
 
+       uint32_t generation_a = commit_graph_generation(a);
+       uint32_t generation_b = commit_graph_generation(b);
        /* lower generation commits first */
-       if (a->generation < b->generation)
+       if (generation_a < generation_b)
                return -1;
-       else if (a->generation > b->generation)
+       else if (generation_a > generation_b)
                return 1;
 
        /* use date as a heuristic when generations are equal */
@@ -281,8 +336,7 @@ struct commit_graph *parse_commit_graph(void *graph_map, size_t graph_size)
                if (data + graph_size - chunk_lookup <
                    GRAPH_CHUNKLOOKUP_WIDTH) {
                        error(_("commit-graph chunk lookup table entry missing; file may be incomplete"));
-                       free(graph);
-                       return NULL;
+                       goto free_and_return;
                }
 
                chunk_id = get_be32(chunk_lookup + 0);
@@ -293,8 +347,7 @@ struct commit_graph *parse_commit_graph(void *graph_map, size_t graph_size)
                if (chunk_offset > graph_size - the_hash_algo->rawsz) {
                        error(_("commit-graph improper chunk offset %08x%08x"), (uint32_t)(chunk_offset >> 32),
                              (uint32_t)chunk_offset);
-                       free(graph);
-                       return NULL;
+                       goto free_and_return;
                }
 
                switch (chunk_id) {
@@ -361,8 +414,7 @@ struct commit_graph *parse_commit_graph(void *graph_map, size_t graph_size)
 
                if (chunk_repeated) {
                        error(_("commit-graph chunk id %08x appears multiple times"), chunk_id);
-                       free(graph);
-                       return NULL;
+                       goto free_and_return;
                }
 
                if (last_chunk_id == GRAPH_CHUNKID_OIDLOOKUP)
@@ -381,17 +433,20 @@ struct commit_graph *parse_commit_graph(void *graph_map, size_t graph_size)
                /* We need both the bloom chunks to exist together. Else ignore the data */
                graph->chunk_bloom_indexes = NULL;
                graph->chunk_bloom_data = NULL;
-               graph->bloom_filter_settings = NULL;
+               FREE_AND_NULL(graph->bloom_filter_settings);
        }
 
        hashcpy(graph->oid.hash, graph->data + graph->data_len - graph->hash_len);
 
-       if (verify_commit_graph_lite(graph)) {
-               free(graph);
-               return NULL;
-       }
+       if (verify_commit_graph_lite(graph))
+               goto free_and_return;
 
        return graph;
+
+free_and_return:
+       free(graph->bloom_filter_settings);
+       free(graph);
+       return NULL;
 }
 
 static struct commit_graph *load_commit_graph_one(const char *graph_file,
@@ -669,13 +724,14 @@ static struct commit_list **insert_parent_or_die(struct repository *r,
        c = lookup_commit(r, &oid);
        if (!c)
                die(_("could not find commit %s"), oid_to_hex(&oid));
-       c->graph_pos = pos;
+       commit_graph_data_at(c)->graph_pos = pos;
        return &commit_list_insert(c, pptr)->next;
 }
 
 static void fill_commit_graph_info(struct commit *item, struct commit_graph *g, uint32_t pos)
 {
        const unsigned char *commit_data;
+       struct commit_graph_data *graph_data;
        uint32_t lex_index;
 
        while (pos < g->num_commits_in_base)
@@ -683,8 +739,10 @@ static void fill_commit_graph_info(struct commit *item, struct commit_graph *g,
 
        lex_index = pos - g->num_commits_in_base;
        commit_data = g->chunk_commit_data + GRAPH_DATA_WIDTH * lex_index;
-       item->graph_pos = pos;
-       item->generation = get_be32(commit_data + g->hash_len + 8) >> 2;
+
+       graph_data = commit_graph_data_at(item);
+       graph_data->graph_pos = pos;
+       graph_data->generation = get_be32(commit_data + g->hash_len + 8) >> 2;
 }
 
 static inline void set_commit_tree(struct commit *c, struct tree *t)
@@ -700,6 +758,7 @@ static int fill_commit_in_graph(struct repository *r,
        uint32_t *parent_data_ptr;
        uint64_t date_low, date_high;
        struct commit_list **pptr;
+       struct commit_graph_data *graph_data;
        const unsigned char *commit_data;
        uint32_t lex_index;
 
@@ -713,7 +772,8 @@ static int fill_commit_in_graph(struct repository *r,
         * Store the "full" position, but then use the
         * "local" position for the rest of the calculation.
         */
-       item->graph_pos = pos;
+       graph_data = commit_graph_data_at(item);
+       graph_data->graph_pos = pos;
        lex_index = pos - g->num_commits_in_base;
 
        commit_data = g->chunk_commit_data + (g->hash_len + 16) * lex_index;
@@ -726,7 +786,7 @@ static int fill_commit_in_graph(struct repository *r,
        date_low = get_be32(commit_data + g->hash_len + 12);
        item->date = (timestamp_t)((date_high << 32) | date_low);
 
-       item->generation = get_be32(commit_data + g->hash_len + 8) >> 2;
+       graph_data->generation = get_be32(commit_data + g->hash_len + 8) >> 2;
 
        pptr = &item->parents;
 
@@ -758,8 +818,9 @@ static int fill_commit_in_graph(struct repository *r,
 
 static int find_commit_in_graph(struct commit *item, struct commit_graph *g, uint32_t *pos)
 {
-       if (item->graph_pos != COMMIT_NOT_FROM_GRAPH) {
-               *pos = item->graph_pos;
+       uint32_t graph_pos = commit_graph_position(item);
+       if (graph_pos != COMMIT_NOT_FROM_GRAPH) {
+               *pos = graph_pos;
                return 1;
        } else {
                struct commit_graph *cur_g = g;
@@ -814,12 +875,13 @@ static struct tree *load_tree_for_commit(struct repository *r,
 {
        struct object_id oid;
        const unsigned char *commit_data;
+       uint32_t graph_pos = commit_graph_position(c);
 
-       while (c->graph_pos < g->num_commits_in_base)
+       while (graph_pos < g->num_commits_in_base)
                g = g->base_graph;
 
        commit_data = g->chunk_commit_data +
-                       GRAPH_DATA_WIDTH * (c->graph_pos - g->num_commits_in_base);
+                       GRAPH_DATA_WIDTH * (graph_pos - g->num_commits_in_base);
 
        hashcpy(oid.hash, commit_data);
        set_commit_tree(c, lookup_tree(r, &oid));
@@ -833,7 +895,7 @@ static struct tree *get_commit_tree_in_graph_one(struct repository *r,
 {
        if (c->maybe_tree)
                return c->maybe_tree;
-       if (c->graph_pos == COMMIT_NOT_FROM_GRAPH)
+       if (commit_graph_position(c) == COMMIT_NOT_FROM_GRAPH)
                BUG("get_commit_tree_in_graph_one called from non-commit-graph commit");
 
        return load_tree_for_commit(r, g, (struct commit *)c);
@@ -880,7 +942,6 @@ struct write_commit_graph_context {
        unsigned append:1,
                 report_progress:1,
                 split:1,
-                check_oids:1,
                 changed_paths:1,
                 order_by_pack:1;
 
@@ -1020,7 +1081,7 @@ static void write_graph_chunk_data(struct hashfile *f, int hash_len,
                else
                        packedDate[0] = 0;
 
-               packedDate[0] |= htonl((*list)->generation << 2);
+               packedDate[0] |= htonl(commit_graph_data_at(*list)->generation << 2);
 
                packedDate[1] = htonl((*list)->date);
                hashwrite(f, packedDate, 8);
@@ -1219,7 +1280,7 @@ static void close_reachable(struct write_commit_graph_context *ctx)
                        continue;
                if (ctx->split) {
                        if ((!parse_commit(commit) &&
-                            commit->graph_pos == COMMIT_NOT_FROM_GRAPH) ||
+                            commit_graph_position(commit) == COMMIT_NOT_FROM_GRAPH) ||
                            flags == COMMIT_GRAPH_SPLIT_REPLACE)
                                add_missing_parents(ctx, commit);
                } else if (!parse_commit_no_graph(commit))
@@ -1251,9 +1312,11 @@ static void compute_generation_numbers(struct write_commit_graph_context *ctx)
                                        _("Computing commit graph generation numbers"),
                                        ctx->commits.nr);
        for (i = 0; i < ctx->commits.nr; i++) {
+               uint32_t generation = commit_graph_data_at(ctx->commits.list[i])->generation;
+
                display_progress(ctx->progress, i + 1);
-               if (ctx->commits.list[i]->generation != GENERATION_NUMBER_INFINITY &&
-                   ctx->commits.list[i]->generation != GENERATION_NUMBER_ZERO)
+               if (generation != GENERATION_NUMBER_INFINITY &&
+                   generation != GENERATION_NUMBER_ZERO)
                        continue;
 
                commit_list_insert(ctx->commits.list[i], &list);
@@ -1264,22 +1327,26 @@ static void compute_generation_numbers(struct write_commit_graph_context *ctx)
                        uint32_t max_generation = 0;
 
                        for (parent = current->parents; parent; parent = parent->next) {
-                               if (parent->item->generation == GENERATION_NUMBER_INFINITY ||
-                                   parent->item->generation == GENERATION_NUMBER_ZERO) {
+                               generation = commit_graph_data_at(parent->item)->generation;
+
+                               if (generation == GENERATION_NUMBER_INFINITY ||
+                                   generation == GENERATION_NUMBER_ZERO) {
                                        all_parents_computed = 0;
                                        commit_list_insert(parent->item, &list);
                                        break;
-                               } else if (parent->item->generation > max_generation) {
-                                       max_generation = parent->item->generation;
+                               } else if (generation > max_generation) {
+                                       max_generation = generation;
                                }
                        }
 
                        if (all_parents_computed) {
-                               current->generation = max_generation + 1;
+                               struct commit_graph_data *data = commit_graph_data_at(current);
+
+                               data->generation = max_generation + 1;
                                pop_commit(&list);
 
-                               if (current->generation > GENERATION_NUMBER_MAX)
-                                       current->generation = GENERATION_NUMBER_MAX;
+                               if (data->generation > GENERATION_NUMBER_MAX)
+                                       data->generation = GENERATION_NUMBER_MAX;
                        }
                }
        }
@@ -1318,13 +1385,25 @@ static void compute_bloom_filters(struct write_commit_graph_context *ctx)
        stop_progress(&progress);
 }
 
+struct refs_cb_data {
+       struct oidset *commits;
+       struct progress *progress;
+};
+
 static int add_ref_to_set(const char *refname,
                          const struct object_id *oid,
                          int flags, void *cb_data)
 {
-       struct oidset *commits = (struct oidset *)cb_data;
+       struct object_id peeled;
+       struct refs_cb_data *data = (struct refs_cb_data *)cb_data;
+
+       if (!peel_ref(refname, &peeled))
+               oid = &peeled;
+       if (oid_object_info(the_repository, oid, NULL) == OBJ_COMMIT)
+               oidset_insert(data->commits, oid);
+
+       display_progress(data->progress, oidset_size(data->commits));
 
-       oidset_insert(commits, oid);
        return 0;
 }
 
@@ -1333,13 +1412,22 @@ int write_commit_graph_reachable(struct object_directory *odb,
                                 const struct split_commit_graph_opts *split_opts)
 {
        struct oidset commits = OIDSET_INIT;
+       struct refs_cb_data data;
        int result;
 
-       for_each_ref(add_ref_to_set, &commits);
+       memset(&data, 0, sizeof(data));
+       data.commits = &commits;
+       if (flags & COMMIT_GRAPH_WRITE_PROGRESS)
+               data.progress = start_delayed_progress(
+                       _("Collecting referenced commits"), 0);
+
+       for_each_ref(add_ref_to_set, &data);
        result = write_commit_graph(odb, NULL, &commits,
                                    flags, split_opts);
 
        oidset_clear(&commits);
+       if (data.progress)
+               stop_progress(&data.progress);
        return result;
 }
 
@@ -1391,46 +1479,19 @@ static int fill_oids_from_packs(struct write_commit_graph_context *ctx,
 static int fill_oids_from_commits(struct write_commit_graph_context *ctx,
                                  struct oidset *commits)
 {
-       uint32_t i = 0;
-       struct strbuf progress_title = STRBUF_INIT;
        struct oidset_iter iter;
        struct object_id *oid;
 
        if (!oidset_size(commits))
                return 0;
 
-       if (ctx->report_progress) {
-               strbuf_addf(&progress_title,
-                           Q_("Finding commits for commit graph from %d ref",
-                              "Finding commits for commit graph from %d refs",
-                              oidset_size(commits)),
-                           oidset_size(commits));
-               ctx->progress = start_delayed_progress(
-                                       progress_title.buf,
-                                       oidset_size(commits));
-       }
-
        oidset_iter_init(commits, &iter);
        while ((oid = oidset_iter_next(&iter))) {
-               struct commit *result;
-
-               display_progress(ctx->progress, ++i);
-
-               result = lookup_commit_reference_gently(ctx->r, oid, 1);
-               if (result) {
-                       ALLOC_GROW(ctx->oids.list, ctx->oids.nr + 1, ctx->oids.alloc);
-                       oidcpy(&ctx->oids.list[ctx->oids.nr], &(result->object.oid));
-                       ctx->oids.nr++;
-               } else if (ctx->check_oids) {
-                       error(_("invalid commit object id: %s"),
-                             oid_to_hex(oid));
-                       return -1;
-               }
+               ALLOC_GROW(ctx->oids.list, ctx->oids.nr + 1, ctx->oids.alloc);
+               oidcpy(&ctx->oids.list[ctx->oids.nr], oid);
+               ctx->oids.nr++;
        }
 
-       stop_progress(&ctx->progress);
-       strbuf_release(&progress_title);
-
        return 0;
 }
 
@@ -1464,7 +1525,7 @@ static uint32_t count_distinct_commits(struct write_commit_graph_context *ctx)
                        if (ctx->split) {
                                struct commit *c = lookup_commit(ctx->r, &ctx->oids.list[i]);
 
-                               if (!c || c->graph_pos != COMMIT_NOT_FROM_GRAPH)
+                               if (!c || commit_graph_position(c) != COMMIT_NOT_FROM_GRAPH)
                                        continue;
                        }
 
@@ -1498,7 +1559,7 @@ static void copy_oids_to_commits(struct write_commit_graph_context *ctx)
                ctx->commits.list[ctx->commits.nr] = lookup_commit(ctx->r, &ctx->oids.list[i]);
 
                if (ctx->split && flags != COMMIT_GRAPH_SPLIT_REPLACE &&
-                   ctx->commits.list[ctx->commits.nr]->graph_pos != COMMIT_NOT_FROM_GRAPH)
+                   commit_graph_position(ctx->commits.list[ctx->commits.nr]) != COMMIT_NOT_FROM_GRAPH)
                        continue;
 
                if (ctx->split && flags == COMMIT_GRAPH_SPLIT_REPLACE)
@@ -1576,7 +1637,8 @@ static int write_commit_graph_file(struct write_commit_graph_context *ctx)
        if (ctx->split) {
                char *lock_name = get_chain_filename(ctx->odb);
 
-               hold_lock_file_for_update(&lk, lock_name, LOCK_DIE_ON_ERROR);
+               hold_lock_file_for_update_mode(&lk, lock_name,
+                                              LOCK_DIE_ON_ERROR, 0444);
 
                fd = git_mkstemp_mode(ctx->graph_name, 0444);
                if (fd < 0) {
@@ -1584,9 +1646,16 @@ static int write_commit_graph_file(struct write_commit_graph_context *ctx)
                        return -1;
                }
 
+               if (adjust_shared_perm(ctx->graph_name)) {
+                       error(_("unable to adjust shared permissions for '%s'"),
+                             ctx->graph_name);
+                       return -1;
+               }
+
                f = hashfd(fd, ctx->graph_name);
        } else {
-               hold_lock_file_for_update(&lk, ctx->graph_name, LOCK_DIE_ON_ERROR);
+               hold_lock_file_for_update_mode(&lk, ctx->graph_name,
+                                              LOCK_DIE_ON_ERROR, 0444);
                fd = lk.tempfile->fd;
                f = hashfd(lk.tempfile->fd, lk.tempfile->filename.buf);
        }
@@ -2008,7 +2077,6 @@ int write_commit_graph(struct object_directory *odb,
        ctx->append = flags & COMMIT_GRAPH_WRITE_APPEND ? 1 : 0;
        ctx->report_progress = flags & COMMIT_GRAPH_WRITE_PROGRESS ? 1 : 0;
        ctx->split = flags & COMMIT_GRAPH_WRITE_SPLIT ? 1 : 0;
-       ctx->check_oids = flags & COMMIT_GRAPH_WRITE_CHECK_OIDS ? 1 : 0;
        ctx->split_opts = split_opts;
        ctx->changed_paths = flags & COMMIT_GRAPH_WRITE_BLOOM_FILTERS ? 1 : 0;
        ctx->total_bloom_filter_data_size = 0;
@@ -2240,6 +2308,7 @@ int verify_commit_graph(struct repository *r, struct commit_graph *g, int flags)
                struct commit *graph_commit, *odb_commit;
                struct commit_list *graph_parents, *odb_parents;
                uint32_t max_generation = 0;
+               uint32_t generation;
 
                display_progress(progress, i + 1);
                hashcpy(cur_oid.hash, g->chunk_oid_lookup + g->hash_len * i);
@@ -2278,8 +2347,9 @@ int verify_commit_graph(struct repository *r, struct commit_graph *g, int flags)
                                             oid_to_hex(&graph_parents->item->object.oid),
                                             oid_to_hex(&odb_parents->item->object.oid));
 
-                       if (graph_parents->item->generation > max_generation)
-                               max_generation = graph_parents->item->generation;
+                       generation = commit_graph_generation(graph_parents->item);
+                       if (generation > max_generation)
+                               max_generation = generation;
 
                        graph_parents = graph_parents->next;
                        odb_parents = odb_parents->next;
@@ -2289,7 +2359,7 @@ int verify_commit_graph(struct repository *r, struct commit_graph *g, int flags)
                        graph_report(_("commit-graph parent list for commit %s terminates early"),
                                     oid_to_hex(&cur_oid));
 
-               if (!graph_commit->generation) {
+               if (!commit_graph_generation(graph_commit)) {
                        if (generation_zero == GENERATION_NUMBER_EXISTS)
                                graph_report(_("commit-graph has generation number zero for commit %s, but non-zero elsewhere"),
                                             oid_to_hex(&cur_oid));
@@ -2309,10 +2379,11 @@ int verify_commit_graph(struct repository *r, struct commit_graph *g, int flags)
                if (max_generation == GENERATION_NUMBER_MAX)
                        max_generation--;
 
-               if (graph_commit->generation != max_generation + 1)
+               generation = commit_graph_generation(graph_commit);
+               if (generation != max_generation + 1)
                        graph_report(_("commit-graph generation for commit %s is %u != %u"),
                                     oid_to_hex(&cur_oid),
-                                    graph_commit->generation,
+                                    generation,
                                     max_generation + 1);
 
                if (graph_commit->date != odb_commit->date)