gc: automatically write commit-graph files
[gitweb.git] / commit-graph.c
index 282cdb4aec654a875e529d94beb7c1970eba1573..212232e752a812c7707d174fcf19db5f0eb02010 100644 (file)
@@ -7,6 +7,7 @@
 #include "packfile.h"
 #include "commit.h"
 #include "object.h"
+#include "refs.h"
 #include "revision.h"
 #include "sha1-lookup.h"
 #include "commit-graph.h"
@@ -244,6 +245,9 @@ static struct commit_list **insert_parent_or_die(struct commit_graph *g,
        struct commit *c;
        struct object_id oid;
 
+       if (pos >= g->num_commits)
+               die("invalid parent position %"PRIu64, pos);
+
        hashcpy(oid.hash, g->chunk_oid_lookup + g->hash_len * pos);
        c = lookup_commit(&oid);
        if (!c)
@@ -653,11 +657,28 @@ static void compute_generation_numbers(struct packed_commit_list* commits)
        }
 }
 
+static int add_ref_to_list(const char *refname,
+                          const struct object_id *oid,
+                          int flags, void *cb_data)
+{
+       struct string_list *list = (struct string_list *)cb_data;
+
+       string_list_append(list, oid_to_hex(oid));
+       return 0;
+}
+
+void write_commit_graph_reachable(const char *obj_dir, int append)
+{
+       struct string_list list;
+
+       string_list_init(&list, 1);
+       for_each_ref(add_ref_to_list, &list);
+       write_commit_graph(obj_dir, NULL, &list, append);
+}
+
 void write_commit_graph(const char *obj_dir,
-                       const char **pack_indexes,
-                       int nr_packs,
-                       const char **commit_hex,
-                       int nr_commits,
+                       struct string_list *pack_indexes,
+                       struct string_list *commit_hex,
                        int append)
 {
        struct packed_oid_list oids;
@@ -698,10 +719,10 @@ void write_commit_graph(const char *obj_dir,
                int dirlen;
                strbuf_addf(&packname, "%s/pack/", obj_dir);
                dirlen = packname.len;
-               for (i = 0; i < nr_packs; i++) {
+               for (i = 0; i < pack_indexes->nr; i++) {
                        struct packed_git *p;
                        strbuf_setlen(&packname, dirlen);
-                       strbuf_addstr(&packname, pack_indexes[i]);
+                       strbuf_addstr(&packname, pack_indexes->items[i].string);
                        p = add_packed_git(packname.buf, packname.len, 1);
                        if (!p)
                                die("error adding pack %s", packname.buf);
@@ -714,12 +735,13 @@ void write_commit_graph(const char *obj_dir,
        }
 
        if (commit_hex) {
-               for (i = 0; i < nr_commits; i++) {
+               for (i = 0; i < commit_hex->nr; i++) {
                        const char *end;
                        struct object_id oid;
                        struct commit *result;
 
-                       if (commit_hex[i] && parse_oid_hex(commit_hex[i], &oid, &end))
+                       if (commit_hex->items[i].string &&
+                           parse_oid_hex(commit_hex->items[i].string, &oid, &end))
                                continue;
 
                        result = lookup_commit_reference_gently(&oid, 1);
@@ -830,6 +852,7 @@ void write_commit_graph(const char *obj_dir,
        oids.nr = 0;
 }
 
+#define VERIFY_COMMIT_GRAPH_ERROR_HASH 2
 static int verify_commit_graph_error;
 
 static void graph_report(const char *fmt, ...)
@@ -843,10 +866,16 @@ static void graph_report(const char *fmt, ...)
        va_end(ap);
 }
 
+#define GENERATION_ZERO_EXISTS 1
+#define GENERATION_NUMBER_EXISTS 2
+
 int verify_commit_graph(struct repository *r, struct commit_graph *g)
 {
        uint32_t i, cur_fanout_pos = 0;
-       struct object_id prev_oid, cur_oid;
+       struct object_id prev_oid, cur_oid, checksum;
+       int generation_zero = 0;
+       struct hashfile *f;
+       int devnull;
 
        if (!g) {
                graph_report("no commit-graph file loaded");
@@ -865,7 +894,18 @@ int verify_commit_graph(struct repository *r, struct commit_graph *g)
        if (verify_commit_graph_error)
                return verify_commit_graph_error;
 
+       devnull = open("/dev/null", O_WRONLY);
+       f = hashfd(devnull, NULL);
+       hashwrite(f, g->data, g->data_len - g->hash_len);
+       finalize_hashfile(f, checksum.hash, CSUM_CLOSE);
+       if (hashcmp(checksum.hash, g->data + g->data_len - g->hash_len)) {
+               graph_report(_("the commit-graph file has incorrect checksum and is likely corrupt"));
+               verify_commit_graph_error = VERIFY_COMMIT_GRAPH_ERROR_HASH;
+       }
+
        for (i = 0; i < g->num_commits; i++) {
+               struct commit *graph_commit;
+
                hashcpy(cur_oid.hash, g->chunk_oid_lookup + g->hash_len * i);
 
                if (i && oidcmp(&prev_oid, &cur_oid) >= 0)
@@ -883,6 +923,11 @@ int verify_commit_graph(struct repository *r, struct commit_graph *g)
                                             cur_fanout_pos, fanout_value, i);
                        cur_fanout_pos++;
                }
+
+               graph_commit = lookup_commit(&cur_oid);
+               if (!parse_commit_in_graph_one(g, graph_commit))
+                       graph_report("failed to parse %s from commit-graph",
+                                    oid_to_hex(&cur_oid));
        }
 
        while (cur_fanout_pos < 256) {
@@ -895,20 +940,89 @@ int verify_commit_graph(struct repository *r, struct commit_graph *g)
                cur_fanout_pos++;
        }
 
-       if (verify_commit_graph_error)
+       if (verify_commit_graph_error & ~VERIFY_COMMIT_GRAPH_ERROR_HASH)
                return verify_commit_graph_error;
 
        for (i = 0; i < g->num_commits; i++) {
-               struct commit *odb_commit;
+               struct commit *graph_commit, *odb_commit;
+               struct commit_list *graph_parents, *odb_parents;
+               uint32_t max_generation = 0;
 
                hashcpy(cur_oid.hash, g->chunk_oid_lookup + g->hash_len * i);
 
+               graph_commit = lookup_commit(&cur_oid);
                odb_commit = (struct commit *)create_object(r, cur_oid.hash, alloc_commit_node(r));
                if (parse_commit_internal(odb_commit, 0, 0)) {
                        graph_report("failed to parse %s from object database",
                                     oid_to_hex(&cur_oid));
                        continue;
                }
+
+               if (oidcmp(&get_commit_tree_in_graph_one(g, graph_commit)->object.oid,
+                          get_commit_tree_oid(odb_commit)))
+                       graph_report("root tree OID for commit %s in commit-graph is %s != %s",
+                                    oid_to_hex(&cur_oid),
+                                    oid_to_hex(get_commit_tree_oid(graph_commit)),
+                                    oid_to_hex(get_commit_tree_oid(odb_commit)));
+
+               graph_parents = graph_commit->parents;
+               odb_parents = odb_commit->parents;
+
+               while (graph_parents) {
+                       if (odb_parents == NULL) {
+                               graph_report("commit-graph parent list for commit %s is too long",
+                                            oid_to_hex(&cur_oid));
+                               break;
+                       }
+
+                       if (oidcmp(&graph_parents->item->object.oid, &odb_parents->item->object.oid))
+                               graph_report("commit-graph parent for %s is %s != %s",
+                                            oid_to_hex(&cur_oid),
+                                            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;
+
+                       graph_parents = graph_parents->next;
+                       odb_parents = odb_parents->next;
+               }
+
+               if (odb_parents != NULL)
+                       graph_report("commit-graph parent list for commit %s terminates early",
+                                    oid_to_hex(&cur_oid));
+
+               if (!graph_commit->generation) {
+                       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));
+                       generation_zero = GENERATION_ZERO_EXISTS;
+               } else if (generation_zero == GENERATION_ZERO_EXISTS)
+                       graph_report("commit-graph has non-zero generation number for commit %s, but zero elsewhere",
+                                    oid_to_hex(&cur_oid));
+
+               if (generation_zero == GENERATION_ZERO_EXISTS)
+                       continue;
+
+               /*
+                * If one of our parents has generation GENERATION_NUMBER_MAX, then
+                * our generation is also GENERATION_NUMBER_MAX. Decrement to avoid
+                * extra logic in the following condition.
+                */
+               if (max_generation == GENERATION_NUMBER_MAX)
+                       max_generation--;
+
+               if (graph_commit->generation != max_generation + 1)
+                       graph_report("commit-graph generation for commit %s is %u != %u",
+                                    oid_to_hex(&cur_oid),
+                                    graph_commit->generation,
+                                    max_generation + 1);
+
+               if (graph_commit->date != odb_commit->date)
+                       graph_report("commit date for commit %s in commit-graph is %"PRItime" != %"PRItime,
+                                    oid_to_hex(&cur_oid),
+                                    graph_commit->date,
+                                    odb_commit->date);
        }
 
        return verify_commit_graph_error;