pack-revindex: prepare for incremental MIDX bitmaps

Prepare the reverse index machinery to handle object lookups in an incremental MIDX bitmap. These changes are broken out across a few functions: - load_midx_revindex() learns to use the appropriate MIDX filename depending on whether the given 'struct multi_pack_index *' is incremental or not. - pack_pos_to_midx() and midx_to_pack_pos() now both take in a global object position in the MIDX pseudo-pack order, and find the earliest containing MIDX (similar to midx.c::midx_for_object(). - midx_pack_order_cmp() adjusts its call to pack_pos_to_midx() by the number of objects in the base (since 'vb - midx->revindx_data' is relative to the containing MIDX, and pack_pos_to_midx() expects a global position). Likewise, this function adjusts its output by adding m->num_objects_in_base to return a global position out through the `*pos` pointer. Together, these changes are sufficient to use the multi-pack index's reverse index format for incremental multi-pack reachability bitmaps. Signed-off-by: Taylor Blau <me@ttaylorr.com> Acked-by: Elijah Newren <newren@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>

Taylor Blau committed Mar 20, 2025 at 13:56 UTC 8331c17b7961736ac69b74d26af579048ad77866
2 files changed +56 -21
pack-bitmap.c
+31 -12
@@ -170,6 +170,15 @@ static struct ewah_bitmap *read_bitmap_1(struct bitmap_index *index)
170 return read_bitmap(index->map, index->map_size, &index->map_pos);
171 }
172
173 +static uint32_t bitmap_num_objects_total(struct bitmap_index *index)
174 +{
175 + if (index->midx) {
176 + struct multi_pack_index *m = index->midx;
177 + return m->num_objects + m->num_objects_in_base;
178 + }
179 + return index->pack->num_objects;
180 +}
181 +
182 static uint32_t bitmap_num_objects(struct bitmap_index *index)
183 {
184 if (index->midx)
@@ -924,7 +933,7 @@ static inline int bitmap_position_extended(struct bitmap_index *bitmap_git,
933
934 if (pos < kh_end(positions)) {
935 int bitmap_pos = kh_value(positions, pos);
927 - return bitmap_pos + bitmap_num_objects(bitmap_git);
936 + return bitmap_pos + bitmap_num_objects_total(bitmap_git);
937 }
938
939 return -1;
@@ -992,7 +1001,7 @@ static int ext_index_add_object(struct bitmap_index *bitmap_git,
1001 bitmap_pos = kh_value(eindex->positions, hash_pos);
1002 }
1003
995 - return bitmap_pos + bitmap_num_objects(bitmap_git);
1004 + return bitmap_pos + bitmap_num_objects_total(bitmap_git);
1005 }
1006
1007 struct bitmap_show_data {
@@ -1342,11 +1351,17 @@ struct ewah_bitmap *pseudo_merge_bitmap_for_commit(struct bitmap_index *bitmap_g
1351 if (pos < 0 || pos >= bitmap_num_objects(bitmap_git))
1352 goto done;
1353
1354 + /*
1355 + * Use bitmap-relative positions instead of offsetting
1356 + * by bitmap_git->num_objects_in_base because we use
1357 + * this to find a match in pseudo_merge_for_parents(),
1358 + * and pseudo-merge groups cannot span multiple bitmap
1359 + * layers.
1360 + */
1361 bitmap_set(parents, pos);
1362 }
1363
1348 - match = pseudo_merge_for_parents(&bitmap_git->pseudo_merges,
1349 - parents);
1364 + match = pseudo_merge_for_parents(&bitmap_git->pseudo_merges, parents);
1365
1366 done:
1367 bitmap_free(parents);
@@ -1500,7 +1515,9 @@ static void show_extended_objects(struct bitmap_index *bitmap_git,
1515 for (i = 0; i < eindex->count; ++i) {
1516 struct object *obj;
1517
1503 - if (!bitmap_get(objects, st_add(bitmap_num_objects(bitmap_git), i)))
1518 + if (!bitmap_get(objects,
1519 + st_add(bitmap_num_objects_total(bitmap_git),
1520 + i)))
1521 continue;
1522
1523 obj = eindex->objects[i];
@@ -1679,7 +1696,7 @@ static void filter_bitmap_exclude_type(struct bitmap_index *bitmap_git,
1696 * them individually.
1697 */
1698 for (i = 0; i < eindex->count; i++) {
1682 - size_t pos = st_add(i, bitmap_num_objects(bitmap_git));
1699 + size_t pos = st_add(i, bitmap_num_objects_total(bitmap_git));
1700 if (eindex->objects[i]->type == type &&
1701 bitmap_get(to_filter, pos) &&
1702 !bitmap_get(tips, pos))
@@ -1705,7 +1722,7 @@ static unsigned long get_size_by_pos(struct bitmap_index *bitmap_git,
1722
1723 oi.sizep = &size;
1724
1708 - if (pos < bitmap_num_objects(bitmap_git)) {
1725 + if (pos < bitmap_num_objects_total(bitmap_git)) {
1726 struct packed_git *pack;
1727 off_t ofs;
1728
@@ -1728,8 +1745,9 @@ static unsigned long get_size_by_pos(struct bitmap_index *bitmap_git,
1745 die(_("unable to get size of %s"), oid_to_hex(&oid));
1746 }
1747 } else {
1748 + size_t eindex_pos = pos - bitmap_num_objects_total(bitmap_git);
1749 struct eindex *eindex = &bitmap_git->ext_index;
1732 - struct object *obj = eindex->objects[pos - bitmap_num_objects(bitmap_git)];
1750 + struct object *obj = eindex->objects[eindex_pos];
1751 if (oid_object_info_extended(bitmap_repo(bitmap_git), &obj->oid,
1752 &oi, 0) < 0)
1753 die(_("unable to get size of %s"), oid_to_hex(&obj->oid));
@@ -1882,7 +1900,7 @@ static void filter_packed_objects_from_bitmap(struct bitmap_index *bitmap_git,
1900 uint32_t objects_nr;
1901 size_t i, pos;
1902
1885 - objects_nr = bitmap_num_objects(bitmap_git);
1903 + objects_nr = bitmap_num_objects_total(bitmap_git);
1904 pos = objects_nr / BITS_IN_EWORD;
1905
1906 if (pos > result->word_alloc)
@@ -2419,7 +2437,7 @@ static uint32_t count_object_type(struct bitmap_index *bitmap_git,
2437 for (i = 0; i < eindex->count; ++i) {
2438 if (eindex->objects[i]->type == type &&
2439 bitmap_get(objects,
2422 - st_add(bitmap_num_objects(bitmap_git), i)))
2440 + st_add(bitmap_num_objects_total(bitmap_git), i)))
2441 count++;
2442 }
2443
@@ -2820,7 +2838,7 @@ uint32_t *create_bitmap_mapping(struct bitmap_index *bitmap_git,
2838 BUG("rebuild_existing_bitmaps: missing required rev-cache "
2839 "extension");
2840
2823 - num_objects = bitmap_num_objects(bitmap_git);
2841 + num_objects = bitmap_num_objects_total(bitmap_git);
2842 CALLOC_ARRAY(reposition, num_objects);
2843
2844 for (i = 0; i < num_objects; ++i) {
@@ -2963,7 +2981,8 @@ static off_t get_disk_usage_for_extended(struct bitmap_index *bitmap_git)
2981 struct object *obj = eindex->objects[i];
2982
2983 if (!bitmap_get(result,
2966 - st_add(bitmap_num_objects(bitmap_git), i)))
2984 + st_add(bitmap_num_objects_total(bitmap_git),
2985 + i)))
2986 continue;
2987
2988 if (oid_object_info_extended(bitmap_repo(bitmap_git), &obj->oid,
pack-revindex.c
+25 -9
@@ -383,8 +383,14 @@ int load_midx_revindex(struct multi_pack_index *m)
383 trace2_data_string("load_midx_revindex", the_repository,
384 "source", "rev");
385
386 - get_midx_filename_ext(m->repo->hash_algo, &revindex_name, m->object_dir,
387 - get_midx_checksum(m), MIDX_EXT_REV);
386 + if (m->has_chain)
387 + get_split_midx_filename_ext(m->repo->hash_algo, &revindex_name,
388 + m->object_dir, get_midx_checksum(m),
389 + MIDX_EXT_REV);
390 + else
391 + get_midx_filename_ext(m->repo->hash_algo, &revindex_name,
392 + m->object_dir, get_midx_checksum(m),
393 + MIDX_EXT_REV);
394
395 ret = load_revindex_from_disk(revindex_name.buf,
396 m->num_objects,
@@ -471,11 +477,15 @@ off_t pack_pos_to_offset(struct packed_git *p, uint32_t pos)
477
478 uint32_t pack_pos_to_midx(struct multi_pack_index *m, uint32_t pos)
479 {
480 + while (m && pos < m->num_objects_in_base)
481 + m = m->base_midx;
482 + if (!m)
483 + BUG("NULL multi-pack-index for object position: %"PRIu32, pos);
484 if (!m->revindex_data)
485 BUG("pack_pos_to_midx: reverse index not yet loaded");
476 - if (m->num_objects <= pos)
486 + if (m->num_objects + m->num_objects_in_base <= pos)
487 BUG("pack_pos_to_midx: out-of-bounds object at %"PRIu32, pos);
478 - return get_be32(m->revindex_data + pos);
488 + return get_be32(m->revindex_data + pos - m->num_objects_in_base);
489 }
490
491 struct midx_pack_key {
@@ -491,7 +501,8 @@ static int midx_pack_order_cmp(const void *va, const void *vb)
501 const struct midx_pack_key *key = va;
502 struct multi_pack_index *midx = key->midx;
503
494 - uint32_t versus = pack_pos_to_midx(midx, (uint32_t*)vb - (const uint32_t *)midx->revindex_data);
504 + size_t pos = (uint32_t *)vb - (const uint32_t *)midx->revindex_data;
505 + uint32_t versus = pack_pos_to_midx(midx, pos + midx->num_objects_in_base);
506 uint32_t versus_pack = nth_midxed_pack_int_id(midx, versus);
507 off_t versus_offset;
508
@@ -529,9 +540,9 @@ static int midx_key_to_pack_pos(struct multi_pack_index *m,
540 {
541 uint32_t *found;
542
532 - if (key->pack >= m->num_packs)
543 + if (key->pack >= m->num_packs + m->num_packs_in_base)
544 BUG("MIDX pack lookup out of bounds (%"PRIu32" >= %"PRIu32")",
534 - key->pack, m->num_packs);
545 + key->pack, m->num_packs + m->num_packs_in_base);
546 /*
547 * The preferred pack sorts first, so determine its identifier by
548 * looking at the first object in pseudo-pack order.
@@ -551,7 +562,8 @@ static int midx_key_to_pack_pos(struct multi_pack_index *m,
562 if (!found)
563 return -1;
564
554 - *pos = found - m->revindex_data;
565 + *pos = (found - m->revindex_data) + m->num_objects_in_base;
566 +
567 return 0;
568 }
569
@@ -559,9 +571,13 @@ int midx_to_pack_pos(struct multi_pack_index *m, uint32_t at, uint32_t *pos)
571 {
572 struct midx_pack_key key;
573
574 + while (m && at < m->num_objects_in_base)
575 + m = m->base_midx;
576 + if (!m)
577 + BUG("NULL multi-pack-index for object position: %"PRIu32, at);
578 if (!m->revindex_data)
579 BUG("midx_to_pack_pos: reverse index not yet loaded");
564 - if (m->num_objects <= at)
580 + if (m->num_objects + m->num_objects_in_base <= at)
581 BUG("midx_to_pack_pos: out-of-bounds object at %"PRIu32, at);
582
583 key.pack = nth_midxed_pack_int_id(m, at);