@matheusbsbs / matheus-bagliano / commits / 7b7cbaef27

odb: introduce mtime fields for object info requests

There are some use cases where we need to figure out the mtime for objects. Most importantly, this is the case when we want to prune unreachable objects. But getting at that data requires users to manually derive the info either via the loose object's mtime, the packfiles' mtime or via the ".mtimes" file. Introduce a new `struct object_info::mtimep` pointer that allows callers to request an object's mtime. This new field will be used in a subsequent commit. Note that the concept of "mtime" is ambiguous: given an object, it may be stored multiple times in the object database, and each of these instances may have a different mtime. Disambiguating these mtimes is nothing that can happen on the generic ODB layer: the caller may search for the oldest object, the newest object, or even the relation of object mtimes depending on the specific source they are located in. As such, it is the responsibility of the caller to disambiguate mtimes. A consequence of this is that it's most likely incorrect to look up the mtime via `odb_read_object_info()`, as this interface does not give us enough information to disambiguate the mtime. Document this accordingly and tell users to use `odb_for_each_object()` instead. Even with this gotcha though it's sensible to have this request as part of the object info, as the mtime is a property of the object storage format. If we for example had a "black-box" storage backend, we'd still need to be able to query it for the mtime info in a generic way. We could introduce a safety mechanism that for example calls `BUG()` in case we look up the mtime outside of `odb_for_each_object()`. But that feels somewhat heavy-handed. Signed-off-by: Patrick Steinhardt <ps@pks.im> Signed-off-by: Junio C Hamano <gitster@pobox.com>

Patrick Steinhardt committed Jan 26, 2026 at 10:51 UTC 7b7cbaef2781cf755bc900e871964ae62ad532c5

4 files changed +74 -11

object-file.c

+25 -4

@@ -409,6 +409,7 @@ static int read_object_info_from_path(struct odb_source *source,

409 char hdr[MAX_HEADER_LEN];

410 unsigned long size_scratch;

411 enum object_type type_scratch;

412 + struct stat st;

413

414 /*

415 * If we don't care about type or size, then we don't

@@ -421,7 +422,7 @@ static int read_object_info_from_path(struct odb_source *source,

422 if (!oi || (!oi->typep && !oi->sizep && !oi->contentp)) {

423 struct stat st;

424

424 - if ((!oi || !oi->disk_sizep) && (flags & OBJECT_INFO_QUICK)) {

425 + if ((!oi || (!oi->disk_sizep && !oi->mtimep)) && (flags & OBJECT_INFO_QUICK)) {

426 ret = quick_has_loose(source->loose, oid) ? 0 : -1;

427 goto out;

428 }

@@ -431,8 +432,12 @@ static int read_object_info_from_path(struct odb_source *source,

432 goto out;

433 }

434

434 - if (oi && oi->disk_sizep)

435 - *oi->disk_sizep = st.st_size;

435 + if (oi) {

436 + if (oi->disk_sizep)

437 + *oi->disk_sizep = st.st_size;

438 + if (oi->mtimep)

439 + *oi->mtimep = st.st_mtime;

440 + }

441

442 ret = 0;

443 goto out;

@@ -446,7 +451,21 @@ static int read_object_info_from_path(struct odb_source *source,

451 goto out;

452 }

453

449 - map = map_fd(fd, path, &mapsize);

454 + if (fstat(fd, &st)) {

455 + close(fd);

456 + ret = -1;

457 + goto out;

458 + }

459 +

460 + mapsize = xsize_t(st.st_size);

461 + if (!mapsize) {

462 + close(fd);

463 + ret = error(_("object file %s is empty"), path);

464 + goto out;

465 + }

466 +

467 + map = xmmap(NULL, mapsize, PROT_READ, MAP_PRIVATE, fd, 0);

468 + close(fd);

469 if (!map) {

470 ret = -1;

471 goto out;

@@ -454,6 +473,8 @@ static int read_object_info_from_path(struct odb_source *source,

473

474 if (oi->disk_sizep)

475 *oi->disk_sizep = mapsize;

476 + if (oi->mtimep)

477 + *oi->mtimep = st.st_mtime;

478

479 stream_to_end = &stream;

480

odb.c

@@ -702,6 +702,8 @@ static int do_oid_object_info_extended(struct object_database *odb,

702 oidclr(oi->delta_base_oid, odb->repo->hash_algo);

703 if (oi->contentp)

704 *oi->contentp = xmemdupz(co->buf, co->size);

705 + if (oi->mtimep)

706 + *oi->mtimep = 0;

707 oi->whence = OI_CACHED;

708 }

709 return 0;

odb.h

+13

@@ -318,6 +318,19 @@ struct object_info {

318 struct object_id *delta_base_oid;

319 void **contentp;

320

321 + /*

322 + * The time the given looked-up object has been last modified.

323 + *

324 + * Note: the mtime may be ambiguous in case the object exists multiple

325 + * times in the object database. It is thus _not_ recommended to use

326 + * this field outside of contexts where you would read every instance

327 + * of the object, like for example with `odb_for_each_object()`. As it

328 + * is impossible to say at the ODB level what the intent of the caller

329 + * is (e.g. whether to find the oldest or newest object), it is the

330 + * responsibility of the caller to disambiguate the mtimes.

331 + */

332 + time_t *mtimep;

333 +

334 /* Response */

335 enum {

336 OI_CACHED,

packfile.c

+34 -7

@@ -1578,13 +1578,14 @@ static void add_delta_base_cache(struct packed_git *p, off_t base_offset,

1578 hashmap_add(&delta_base_cache, &ent->ent);

1579 }

1580

1581 -int packed_object_info(struct packed_git *p,

1582 - off_t obj_offset, struct object_info *oi)

1581 +static int packed_object_info_with_index_pos(struct packed_git *p, off_t obj_offset,

1582 + uint32_t *maybe_index_pos, struct object_info *oi)

1583 {

1584 struct pack_window *w_curs = NULL;

1585 unsigned long size;

1586 off_t curpos = obj_offset;

1587 enum object_type type = OBJ_NONE;

1588 + uint32_t pack_pos;

1589 int ret;

1590

1591 /*

@@ -1619,16 +1620,35 @@ int packed_object_info(struct packed_git *p,

1620 }

1621 }

1622

1622 - if (oi->disk_sizep) {

1623 - uint32_t pos;

1624 - if (offset_to_pack_pos(p, obj_offset, &pos) < 0) {

1623 + if (oi->disk_sizep || (oi->mtimep && p->is_cruft)) {

1624 + if (offset_to_pack_pos(p, obj_offset, &pack_pos) < 0) {

1625 error("could not find object at offset %"PRIuMAX" "

1626 "in pack %s", (uintmax_t)obj_offset, p->pack_name);

1627 ret = -1;

1628 goto out;

1629 }

1630 + }

1631 +

1632 + if (oi->disk_sizep)

1633 + *oi->disk_sizep = pack_pos_to_offset(p, pack_pos + 1) - obj_offset;

1634 +

1635 + if (oi->mtimep) {

1636 + if (p->is_cruft) {

1637 + uint32_t index_pos;

1638 +

1639 + if (load_pack_mtimes(p) < 0)

1640 + die(_("could not load .mtimes for cruft pack '%s'"),

1641 + pack_basename(p));

1642 +

1643 + if (maybe_index_pos)

1644 + index_pos = *maybe_index_pos;

1645 + else

1646 + index_pos = pack_pos_to_index(p, pack_pos);

1647

1631 - *oi->disk_sizep = pack_pos_to_offset(p, pos + 1) - obj_offset;

1648 + *oi->mtimep = nth_packed_mtime(p, index_pos);

1649 + } else {

1650 + *oi->mtimep = p->mtime;

1651 + }

1652 }

1653

1654 if (oi->typep) {

@@ -1681,6 +1701,12 @@ out:

1701 return ret;

1702 }

1703

1704 +int packed_object_info(struct packed_git *p, off_t obj_offset,

1705 + struct object_info *oi)

1706 +{

1707 + return packed_object_info_with_index_pos(p, obj_offset, NULL, oi);

1708 +}

1709 +

1710 static void *unpack_compressed_entry(struct packed_git *p,

1711 struct pack_window **w_curs,

1712 off_t curpos,

@@ -2378,7 +2404,8 @@ static int packfile_store_for_each_object_wrapper(const struct object_id *oid,

2404 off_t offset = nth_packed_object_offset(pack, index_pos);

2405 struct object_info oi = *data->request;

2406

2381 - if (packed_object_info(pack, offset, &oi) < 0) {

2407 + if (packed_object_info_with_index_pos(pack, offset,

2408 + &index_pos, &oi) < 0) {

2409 mark_bad_packed_object(pack, oid);

2410 return -1;

2411 }