reachable.c

   1 #include "cache.h"
   2 #include "hex.h"
   3 #include "refs.h"
   4 #include "tag.h"
   5 #include "commit.h"
   6 #include "blob.h"
   7 #include "diff.h"
   8 #include "revision.h"
   9 #include "reachable.h"
  10 #include "cache-tree.h"
  11 #include "progress.h"
  12 #include "list-objects.h"
  13 #include "packfile.h"
  14 #include "worktree.h"
  15 #include "object-store.h"
  16 #include "pack-bitmap.h"
  17 #include "pack-mtimes.h"
  18
  19 struct connectivity_progress {
  20         struct progress *progress;
  21         unsigned long count;
  22 };
  23
  24 static void update_progress(struct connectivity_progress *cp)
  25 {
  26         cp->count++;
  27         if ((cp->count & 1023) == 0)
  28                 display_progress(cp->progress, cp->count);
  29 }
  30
  31 static int add_one_ref(const char *path, const struct object_id *oid,
  32                        int flag, void *cb_data)
  33 {
  34         struct rev_info *revs = (struct rev_info *)cb_data;
  35         struct object *object;
  36
  37         if ((flag & REF_ISSYMREF) && (flag & REF_ISBROKEN)) {
  38                 warning("symbolic ref is dangling: %s", path);
  39                 return 0;
  40         }
  41
  42         object = parse_object_or_die(oid, path);
  43         add_pending_object(revs, object, "");
  44
  45         return 0;
  46 }
  47
  48 /*
  49  * The traversal will have already marked us as SEEN, so we
  50  * only need to handle any progress reporting here.
  51  */
  52 static void mark_object(struct object *obj, const char *name, void *data)
  53 {
  54         update_progress(data);
  55 }
  56
  57 static void mark_commit(struct commit *c, void *data)
  58 {
  59         mark_object(&c->object, NULL, data);
  60 }
  61
  62 struct recent_data {
  63         struct rev_info *revs;
  64         timestamp_t timestamp;
  65         report_recent_object_fn *cb;
  66         int ignore_in_core_kept_packs;
  67 };
  68
  69 static void add_recent_object(const struct object_id *oid,
  70                               struct packed_git *pack,
  71                               off_t offset,
  72                               timestamp_t mtime,
  73                               struct recent_data *data)
  74 {
  75         struct object *obj;
  76         enum object_type type;
  77
  78         if (mtime <= data->timestamp)
  79                 return;
  80
  81         /*
  82          * We do not want to call parse_object here, because
  83          * inflating blobs and trees could be very expensive.
  84          * However, we do need to know the correct type for
  85          * later processing, and the revision machinery expects
  86          * commits and tags to have been parsed.
  87          */
  88         type = oid_object_info(the_repository, oid, NULL);
  89         if (type < 0)
  90                 die("unable to get object info for %s", oid_to_hex(oid));
  91
  92         switch (type) {
  93         case OBJ_TAG:
  94         case OBJ_COMMIT:
  95                 obj = parse_object_or_die(oid, NULL);
  96                 break;
  97         case OBJ_TREE:
  98                 obj = (struct object *)lookup_tree(the_repository, oid);
  99                 break;
 100         case OBJ_BLOB:
 101                 obj = (struct object *)lookup_blob(the_repository, oid);
 102                 break;
 103         default:
 104                 die("unknown object type for %s: %s",
 105                     oid_to_hex(oid), type_name(type));
 106         }
 107
 108         if (!obj)
 109                 die("unable to lookup %s", oid_to_hex(oid));
 110
 111         add_pending_object(data->revs, obj, "");
 112         if (data->cb)
 113                 data->cb(obj, pack, offset, mtime);
 114 }
 115
 116 static int want_recent_object(struct recent_data *data,
 117                               const struct object_id *oid)
 118 {
 119         if (data->ignore_in_core_kept_packs &&
 120             has_object_kept_pack(oid, IN_CORE_KEEP_PACKS))
 121                 return 0;
 122         return 1;
 123 }
 124
 125 static int add_recent_loose(const struct object_id *oid,
 126                             const char *path, void *data)
 127 {
 128         struct stat st;
 129         struct object *obj;
 130
 131         if (!want_recent_object(data, oid))
 132                 return 0;
 133
 134         obj = lookup_object(the_repository, oid);
 135
 136         if (obj && obj->flags & SEEN)
 137                 return 0;
 138
 139         if (stat(path, &st) < 0) {
 140                 /*
 141                  * It's OK if an object went away during our iteration; this
 142                  * could be due to a simultaneous repack. But anything else
 143                  * we should abort, since we might then fail to mark objects
 144                  * which should not be pruned.
 145                  */
 146                 if (errno == ENOENT)
 147                         return 0;
 148                 return error_errno("unable to stat %s", oid_to_hex(oid));
 149         }
 150
 151         add_recent_object(oid, NULL, 0, st.st_mtime, data);
 152         return 0;
 153 }
 154
 155 static int add_recent_packed(const struct object_id *oid,
 156                              struct packed_git *p, uint32_t pos,
 157                              void *data)
 158 {
 159         struct object *obj;
 160         timestamp_t mtime = p->mtime;
 161
 162         if (!want_recent_object(data, oid))
 163                 return 0;
 164
 165         obj = lookup_object(the_repository, oid);
 166
 167         if (obj && obj->flags & SEEN)
 168                 return 0;
 169         if (p->is_cruft) {
 170                 if (load_pack_mtimes(p) < 0)
 171                         die(_("could not load cruft pack .mtimes"));
 172                 mtime = nth_packed_mtime(p, pos);
 173         }
 174         add_recent_object(oid, p, nth_packed_object_offset(p, pos), mtime, data);
 175         return 0;
 176 }
 177
 178 int add_unseen_recent_objects_to_traversal(struct rev_info *revs,
 179                                            timestamp_t timestamp,
 180                                            report_recent_object_fn *cb,
 181                                            int ignore_in_core_kept_packs)
 182 {
 183         struct recent_data data;
 184         enum for_each_object_flags flags;
 185         int r;
 186
 187         data.revs = revs;
 188         data.timestamp = timestamp;
 189         data.cb = cb;
 190         data.ignore_in_core_kept_packs = ignore_in_core_kept_packs;
 191
 192         r = for_each_loose_object(add_recent_loose, &data,
 193                                   FOR_EACH_OBJECT_LOCAL_ONLY);
 194         if (r)
 195                 return r;
 196
 197         flags = FOR_EACH_OBJECT_LOCAL_ONLY | FOR_EACH_OBJECT_PACK_ORDER;
 198         if (ignore_in_core_kept_packs)
 199                 flags |= FOR_EACH_OBJECT_SKIP_IN_CORE_KEPT_PACKS;
 200
 201         return for_each_packed_object(add_recent_packed, &data, flags);
 202 }
 203
 204 static int mark_object_seen(const struct object_id *oid,
 205                              enum object_type type,
 206                              int exclude,
 207                              uint32_t name_hash,
 208                              struct packed_git *found_pack,
 209                              off_t found_offset)
 210 {
 211         struct object *obj = lookup_object_by_type(the_repository, oid, type);
 212         if (!obj)
 213                 die("unable to create object '%s'", oid_to_hex(oid));
 214
 215         obj->flags |= SEEN;
 216         return 0;
 217 }
 218
 219 void mark_reachable_objects(struct rev_info *revs, int mark_reflog,
 220                             timestamp_t mark_recent, struct progress *progress)
 221 {
 222         struct connectivity_progress cp;
 223         struct bitmap_index *bitmap_git;
 224
 225         /*
 226          * Set up revision parsing, and mark us as being interested
 227          * in all object types, not just commits.
 228          */
 229         revs->tag_objects = 1;
 230         revs->blob_objects = 1;
 231         revs->tree_objects = 1;
 232
 233         /* Add all refs from the index file */
 234         add_index_objects_to_pending(revs, 0);
 235
 236         /* Add all external refs */
 237         for_each_ref(add_one_ref, revs);
 238
 239         /* detached HEAD is not included in the list above */
 240         head_ref(add_one_ref, revs);
 241         other_head_refs(add_one_ref, revs);
 242
 243         /* Add all reflog info */
 244         if (mark_reflog)
 245                 add_reflogs_to_pending(revs, 0);
 246
 247         cp.progress = progress;
 248         cp.count = 0;
 249
 250         bitmap_git = prepare_bitmap_walk(revs, 0);
 251         if (bitmap_git) {
 252                 traverse_bitmap_commit_list(bitmap_git, revs, mark_object_seen);
 253                 free_bitmap_index(bitmap_git);
 254         } else {
 255                 if (prepare_revision_walk(revs))
 256                         die("revision walk setup failed");
 257                 traverse_commit_list(revs, mark_commit, mark_object, &cp);
 258         }
 259
 260         if (mark_recent) {
 261                 revs->ignore_missing_links = 1;
 262                 if (add_unseen_recent_objects_to_traversal(revs, mark_recent,
 263                                                            NULL, 0))
 264                         die("unable to mark recent objects");
 265                 if (prepare_revision_walk(revs))
 266                         die("revision walk setup failed");
 267                 traverse_commit_list(revs, mark_commit, mark_object, &cp);
 268         }
 269
 270         display_progress(cp.progress, cp.count);
 271 }