reachable.c

   1 #include "cache.h"
   2 #include "refs.h"
   3 #include "tag.h"
   4 #include "commit.h"
   5 #include "blob.h"
   6 #include "diff.h"
   7 #include "revision.h"
   8 #include "reachable.h"
   9 #include "cache-tree.h"
  10 #include "progress.h"
  11 #include "list-objects.h"
  12 #include "packfile.h"
  13 #include "worktree.h"
  14 #include "object-store.h"
  15 #include "pack-bitmap.h"
  16 #include "pack-mtimes.h"
  17
  18 struct connectivity_progress {
  19         struct progress *progress;
  20         unsigned long count;
  21 };
  22
  23 static void update_progress(struct connectivity_progress *cp)
  24 {
  25         cp->count++;
  26         if ((cp->count & 1023) == 0)
  27                 display_progress(cp->progress, cp->count);
  28 }
  29
  30 static int add_one_ref(const char *path, const struct object_id *oid,
  31                        int flag, void *cb_data)
  32 {
  33         struct rev_info *revs = (struct rev_info *)cb_data;
  34         struct object *object;
  35
  36         if ((flag & REF_ISSYMREF) && (flag & REF_ISBROKEN)) {
  37                 warning("symbolic ref is dangling: %s", path);
  38                 return 0;
  39         }
  40
  41         object = parse_object_or_die(oid, path);
  42         add_pending_object(revs, object, "");
  43
  44         return 0;
  45 }
  46
  47 /*
  48  * The traversal will have already marked us as SEEN, so we
  49  * only need to handle any progress reporting here.
  50  */
  51 static void mark_object(struct object *obj, const char *name, void *data)
  52 {
  53         update_progress(data);
  54 }
  55
  56 static void mark_commit(struct commit *c, void *data)
  57 {
  58         mark_object(&c->object, NULL, data);
  59 }
  60
  61 struct recent_data {
  62         struct rev_info *revs;
  63         timestamp_t timestamp;
  64         report_recent_object_fn *cb;
  65         int ignore_in_core_kept_packs;
  66 };
  67
  68 static void add_recent_object(const struct object_id *oid,
  69                               struct packed_git *pack,
  70                               off_t offset,
  71                               timestamp_t mtime,
  72                               struct recent_data *data)
  73 {
  74         struct object *obj;
  75         enum object_type type;
  76
  77         if (mtime <= data->timestamp)
  78                 return;
  79
  80         /*
  81          * We do not want to call parse_object here, because
  82          * inflating blobs and trees could be very expensive.
  83          * However, we do need to know the correct type for
  84          * later processing, and the revision machinery expects
  85          * commits and tags to have been parsed.
  86          */
  87         type = oid_object_info(the_repository, oid, NULL);
  88         if (type < 0)
  89                 die("unable to get object info for %s", oid_to_hex(oid));
  90
  91         switch (type) {
  92         case OBJ_TAG:
  93         case OBJ_COMMIT:
  94                 obj = parse_object_or_die(oid, NULL);
  95                 break;
  96         case OBJ_TREE:
  97                 obj = (struct object *)lookup_tree(the_repository, oid);
  98                 break;
  99         case OBJ_BLOB:
 100                 obj = (struct object *)lookup_blob(the_repository, oid);
 101                 break;
 102         default:
 103                 die("unknown object type for %s: %s",
 104                     oid_to_hex(oid), type_name(type));
 105         }
 106
 107         if (!obj)
 108                 die("unable to lookup %s", oid_to_hex(oid));
 109
 110         add_pending_object(data->revs, obj, "");
 111         if (data->cb)
 112                 data->cb(obj, pack, offset, mtime);
 113 }
 114
 115 static int want_recent_object(struct recent_data *data,
 116                               const struct object_id *oid)
 117 {
 118         if (data->ignore_in_core_kept_packs &&
 119             has_object_kept_pack(oid, IN_CORE_KEEP_PACKS))
 120                 return 0;
 121         return 1;
 122 }
 123
 124 static int add_recent_loose(const struct object_id *oid,
 125                             const char *path, void *data)
 126 {
 127         struct stat st;
 128         struct object *obj;
 129
 130         if (!want_recent_object(data, oid))
 131                 return 0;
 132
 133         obj = lookup_object(the_repository, oid);
 134
 135         if (obj && obj->flags & SEEN)
 136                 return 0;
 137
 138         if (stat(path, &st) < 0) {
 139                 /*
 140                  * It's OK if an object went away during our iteration; this
 141                  * could be due to a simultaneous repack. But anything else
 142                  * we should abort, since we might then fail to mark objects
 143                  * which should not be pruned.
 144                  */
 145                 if (errno == ENOENT)
 146                         return 0;
 147                 return error_errno("unable to stat %s", oid_to_hex(oid));
 148         }
 149
 150         add_recent_object(oid, NULL, 0, st.st_mtime, data);
 151         return 0;
 152 }
 153
 154 static int add_recent_packed(const struct object_id *oid,
 155                              struct packed_git *p, uint32_t pos,
 156                              void *data)
 157 {
 158         struct object *obj;
 159         timestamp_t mtime = p->mtime;
 160
 161         if (!want_recent_object(data, oid))
 162                 return 0;
 163
 164         obj = lookup_object(the_repository, oid);
 165
 166         if (obj && obj->flags & SEEN)
 167                 return 0;
 168         if (p->is_cruft) {
 169                 if (load_pack_mtimes(p) < 0)
 170                         die(_("could not load cruft pack .mtimes"));
 171                 mtime = nth_packed_mtime(p, pos);
 172         }
 173         add_recent_object(oid, p, nth_packed_object_offset(p, pos), mtime, data);
 174         return 0;
 175 }
 176
 177 int add_unseen_recent_objects_to_traversal(struct rev_info *revs,
 178                                            timestamp_t timestamp,
 179                                            report_recent_object_fn *cb,
 180                                            int ignore_in_core_kept_packs)
 181 {
 182         struct recent_data data;
 183         enum for_each_object_flags flags;
 184         int r;
 185
 186         data.revs = revs;
 187         data.timestamp = timestamp;
 188         data.cb = cb;
 189         data.ignore_in_core_kept_packs = ignore_in_core_kept_packs;
 190
 191         r = for_each_loose_object(add_recent_loose, &data,
 192                                   FOR_EACH_OBJECT_LOCAL_ONLY);
 193         if (r)
 194                 return r;
 195
 196         flags = FOR_EACH_OBJECT_LOCAL_ONLY | FOR_EACH_OBJECT_PACK_ORDER;
 197         if (ignore_in_core_kept_packs)
 198                 flags |= FOR_EACH_OBJECT_SKIP_IN_CORE_KEPT_PACKS;
 199
 200         return for_each_packed_object(add_recent_packed, &data, flags);
 201 }
 202
 203 static int mark_object_seen(const struct object_id *oid,
 204                              enum object_type type,
 205                              int exclude,
 206                              uint32_t name_hash,
 207                              struct packed_git *found_pack,
 208                              off_t found_offset)
 209 {
 210         struct object *obj = lookup_object_by_type(the_repository, oid, type);
 211         if (!obj)
 212                 die("unable to create object '%s'", oid_to_hex(oid));
 213
 214         obj->flags |= SEEN;
 215         return 0;
 216 }
 217
 218 void mark_reachable_objects(struct rev_info *revs, int mark_reflog,
 219                             timestamp_t mark_recent, struct progress *progress)
 220 {
 221         struct connectivity_progress cp;
 222         struct bitmap_index *bitmap_git;
 223
 224         /*
 225          * Set up revision parsing, and mark us as being interested
 226          * in all object types, not just commits.
 227          */
 228         revs->tag_objects = 1;
 229         revs->blob_objects = 1;
 230         revs->tree_objects = 1;
 231
 232         /* Add all refs from the index file */
 233         add_index_objects_to_pending(revs, 0);
 234
 235         /* Add all external refs */
 236         for_each_ref(add_one_ref, revs);
 237
 238         /* detached HEAD is not included in the list above */
 239         head_ref(add_one_ref, revs);
 240         other_head_refs(add_one_ref, revs);
 241
 242         /* Add all reflog info */
 243         if (mark_reflog)
 244                 add_reflogs_to_pending(revs, 0);
 245
 246         cp.progress = progress;
 247         cp.count = 0;
 248
 249         bitmap_git = prepare_bitmap_walk(revs, 0);
 250         if (bitmap_git) {
 251                 traverse_bitmap_commit_list(bitmap_git, revs, mark_object_seen);
 252                 free_bitmap_index(bitmap_git);
 253         } else {
 254                 if (prepare_revision_walk(revs))
 255                         die("revision walk setup failed");
 256                 traverse_commit_list(revs, mark_commit, mark_object, &cp);
 257         }
 258
 259         if (mark_recent) {
 260                 revs->ignore_missing_links = 1;
 261                 if (add_unseen_recent_objects_to_traversal(revs, mark_recent,
 262                                                            NULL, 0))
 263                         die("unable to mark recent objects");
 264                 if (prepare_revision_walk(revs))
 265                         die("revision walk setup failed");
 266                 traverse_commit_list(revs, mark_commit, mark_object, &cp);
 267         }
 268
 269         display_progress(cp.progress, cp.count);
 270 }