The "git pack-objects --path-walk" traversal has been integrated
with several object filters, including blobless and sparse filters.
* ds/path-walk-filters:
path-walk: support `combine` filter
path-walk: support `object:type` filter
path-walk: support `tree:0` filter
t6601: tag otherwise-unreachable trees
pack-objects: support sparse:oid filter with path-walk
path-walk: add pl_sparse_trees to control tree pruning
path-walk: support blob size limit filter
backfill: die on incompatible filter options
path-walk: support blobless filter
path-walk: always emit directly-requested objects
t/perf: add pack-objects filter and path-walk benchmark
pack-objects: pass --objects with --path-walk
t5620: make test work with path-walk var
The --filter=sparse:<oid> option to 'git pack-objects' allows focusing
an object set to a sparse-checkout definition. This reduces the set of
matching blobs while retaining all reachable trees. No server currently
supports fetching with this filter because it is expensive to compute
and reachability bitmaps do not help without a significant effort to
extend the bitmap feature to store bitmaps for each supported sparse-
checkout definition.
Without focusing on serving fetches and clones with these filters, there
are still benefits that could be realized by making this faster. With
the sparse index, it's more realistic now than ever to be able to
operate a local clone that was bootstrapped by a packfile created with
a sparse filter, because the missing trees are not needed to move a
sparse-checkout from one commit to another or to view the history of any
path in scope. Such clones could perhaps be bootstrapped by partial
bundles.
Previously, constructing these sparse packs has been incredibly
computationally inefficient. The revision walk that explores which
objects are in scope spends a lot of time checking each object to see if
it matches the sparse-checkout patterns, causing quadratic behavior
(number of objects times number of sparse-checkout patterns). This
improves somewhat when using cone-mode sparse-checkout patterns that can
use hashtables and prefix matches to determine containment. However, the
check per object is still too expensive for most cases.
This is where the path-walk feature comes in. We can proceed as normal
by placing objects in bins by path and _then_ check a group of objects
all at once. Since sparse:<oid> only restricts blobs, the path-walk must
include all reachable trees while using the cone-mode patterns to skip
blobs at paths outside the sparse scope. This establishes a baseline for
a potential future "treesparse:<oid>" filter that would also restrict
trees, but introducing such a new filter is deferred to a later change.
The implementation here is focused around loading the sparse-checkout
patterns from the provided object ID and checking that the patterns are
indeed cone-mode patterns. We can then load the correct pattern list
into the path walk context and use the logic that already exists from
bff4555767 (backfill: add --sparse option, 2025-02-03), though that
feature loads sparse-checkout patterns from the worktree's local
settings and also restricts tree objects. We use a combination of errors
and warnings to signal problems during this load. The difference is that
errors are likely fatal for the non-path-walk version while the warnings
are probably just implementation details for the path-walk version and
the 'git pack-objects' command can fall back to the revision walk
version.
Now that the SEEN flag is deferred until after pattern checks (from the
previous commit), handle the case where a tree with a shared OID appears
at both an out-of-cone and in-cone path. When trees are not being pruned
(pl_sparse_trees == 0), the path-walk re-walks the tree at the in-cone
path so that in-cone blobs within it are discovered. The new tests in
t5317 and t6601 demonstrate this behavior and would fail without these
changes.
The performance test p5315 shows the impact of this change when using
sparse filters:
Test HEAD~1 HEAD
----------------------------------------------------------------------
5315.10: repack (sparse:oid) 77.98 77.47 -0.7%
5315.11: repack size (sparse:oid) 187.5M 187.4M -0.0%
5315.12: repack (sparse:oid, --path-walk) 77.91 31.41 -59.7%
5315.13: repack size (sparse:oid, --path-walk) 187.5M 161.1M -14.1%
These performance tests were run on the Git repository. The --path-walk
feature shows meaningful space savings (14% smaller for sparse packs)
and dramatic time savings (60% faster) by leveraging the path-walk's
ability to skip blobs outside the sparse scope.
Co-authored-by: Taylor Blau <me@ttaylorr.com>
Signed-off-by: Taylor Blaue <me@ttaylorr.com>
Signed-off-by: Derrick Stolee <stolee@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The 'git pack-objects' command can opt-in to using the path-walk API for
scanning the objects. Currently, this option is dynamically disabled if
combined with '--filter=<X>', even when using a simple filter such as
'blob:none' to signal a blobless packfile. This is a common scenario for
repos at scale, so is worth integrating.
Also, users can opt-in to the '--path-walk' option by default through
the pack.usePathWalk=true config option. When using that in a blobless
partial clone, the following warning can appear even though the user did
not specify either option directly:
warning: cannot use --filter with --path-walk
Teach the path-walk API to handle the 'blob:none' object filter
natively. When revs->filter.choice is LOFC_BLOB_NONE, the path-walk
sets info->blobs to 0 (skipping all blob objects) and clears the
filter from revs so that prepare_revision_walk() does not reject the
configuration.
This check is implemented in the static prepare_filters() method, which
will simultaneously check if the input filters are compatible and will
make the appropriate mutations to the path_walk_info and filters if the
path_walk_info is non-NULL. This allows us to use this logic both in the
API method path_walk_filter_compatible() for use in
builtin/pack-objects.c and as a prep step in walk_objects_by_path().
Update the test helper (test-path-walk) to accept --filter=<spec>
as a test-tool option (before '--'), applying it to revs after
setup_revisions() to avoid the --objects requirement check. We can also
revert recent GIT_TEST_PACK_PATH_WALK overrides in t5620.
Also switch test-path-walk from REV_INFO_INIT with manual repo
assignment to repo_init_revisions(), which properly initializes
the filter_spec strbuf needed for filter parsing.
Add tests for blob:none with --all and with a single branch.
The performance test p5315 shows the impact of this change when using
blobless filters:
Test HEAD~1 HEAD
---------------------------------------------------------------------
5315.6: repack (blob:none) 13.53 13.87 +2.5%
5315.7: repack size (blob:none) 137.7M 137.8M +0.1%
5315.8: repack (blob:none, --path-walk) 13.51 23.43 +73.4%
5315.9: repack size (blob:none, --path-walk) 137.7M 115.2M -16.3%
These performance tests were run on the Git repository. The --path-walk
feature shows meaningful space savings (16% smaller for blobless packs)
at the cost of increased computation time due to the two compression
passes. This data demonstrates that the feature is engaged and provides
real compression benefits when --no-reuse-delta forces fresh deltas.
Co-Authored-by: Taylor Blau <me@ttaylorr.com>
Signed-off-by: Taylor Blau <me@ttaylorr.com>
Signed-off-by: Derrick Stolee <stolee@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When 'git pack-objects' has the --path-walk option enabled, it uses a
different set of revision walk parameters than normal. For one,
--objects was previously assumed by the path-walk API and could be
omitted. We also needed --boundary to allow discovering UNINTERESTING
objects to use as delta bases.
We will be updating the path-walk API soon to work with some filter
options. However, the revision machinery will trigger a fatal error:
fatal: object filtering requires --objects
The fix is easy: add the --objects option as an argument. This has no
effect on the path-walk API but does simplify the revision option
parsing for the objects filter.
We can remove the comment about "removing" the options because they were
never removed and instead not added. We still need to disable using
bitmaps.
Signed-off-by: Derrick Stolee <stolee@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The odb_read_stream structure uses unsigned long for the size field,
which is 32-bit on Windows even in 64-bit builds. When streaming
objects larger than 4GB, the size would be truncated to zero or an
incorrect value, resulting in empty files being written to disk.
Change the size field in odb_read_stream to size_t and introduce
unpack_object_header_sz() to return sizes via size_t pointer. Since
object_info.sizep remains unsigned long for API compatibility, use
temporary variables where the types differ, with comments noting the
truncation limitation for code paths that still use unsigned long.
Widening the producers to size_t in this way introduces a handful of
silent size_t -> unsigned long narrowings on Windows, all in
builtin/pack-objects.c, where the consumers are still typed
unsigned long. Make those narrowings explicit with
cast_size_t_to_ulong() so they assert loudly the moment an object
actually exceeds ULONG_MAX bytes:
- oe_get_size_slow() returns unsigned long but holds a size_t
locally; cast at the return.
- write_reuse_object() passes a size_t into check_pack_inflate(),
whose expect parameter is unsigned long; cast at the call.
- check_object() routes a size_t through SET_SIZE() and
SET_DELTA_SIZE(), both of which take unsigned long via
oe_set_size() / oe_set_delta_size(); cast at the three call
sites in the OBJ_OFS_DELTA / OBJ_REF_DELTA branches and in the
non-delta default arm.
The cast-only treatment is deliberately a stop-gap. Properly
widening oe_set_size, oe_get_size_slow's return type,
check_pack_inflate's expect parameter, object_info.sizep,
patch_delta, and the OE_SIZE_BITS bit-fields cascades into a series
that is too large to be reviewable, so the proper widening is
deferred to a follow-up topic. Until then,
cast_size_t_to_ulong() at least makes the truncation explicit at
the source: it documents the boundary, and on a 64-bit non-Windows
platform it is a no-op.
This was originally authored by LordKiRon <https://github.com/LordKiRon>,
who preferred not to reveal their real name and therefore agreed that I
take over authorship.
Helped-by: Torsten Bögershausen <tboegi@web.de>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
pack-objects's --stdin-packs=follow mode learns to handle
excluded-but-open packs.
* tb/stdin-packs-excluded-but-open:
repack: mark non-MIDX packs above the split as excluded-open
pack-objects: support excluded-open packs with --stdin-packs
t7704: demonstrate failure with once-cruft objects above the geometric split
pack-objects: refactor `read_packs_list_from_stdin()` to use `strmap`
pack-objects: plug leak in `read_stdin_packs()`
Object name handling (disambiguation and abbreviation) has been
refactored to be backend-generic, moving logic into the respective
object database backends.
* ps/odb-generic-object-name-handling:
odb: introduce generic `odb_find_abbrev_len()`
object-file: move logic to compute packed abbreviation length
object-name: move logic to compute loose abbreviation length
object-name: simplify computing common prefixes
object-name: abbreviate loose object names without `disambiguate_state`
object-name: merge `update_candidates()` and `match_prefix()`
object-name: backend-generic `get_short_oid()`
object-name: backend-generic `repo_collect_ambiguous()`
object-name: extract function to parse object ID prefixes
object-name: move logic to iterate through packed prefixed objects
object-name: move logic to iterate through loose prefixed objects
odb: introduce `struct odb_for_each_object_options`
oidtree: extend iteration to allow for arbitrary return codes
oidtree: modernize the code a bit
object-file: fix sparse 'plain integer as NULL pointer' error
In cd846bacc7 (pack-objects: introduce '--stdin-packs=follow',
2025-06-23), pack-objects learned to traverse through commits in
included packs when using '--stdin-packs=follow', rescuing reachable
objects from unlisted packs into the output.
When we encounter a commit in an excluded pack during this rescuing
phase we will traverse through its parents. But because we set
`revs.no_kept_objects = 1`, commit simplification will prevent us from
showing it via `get_revision()`. (In practice, `--stdin-packs=follow`
walks commits down to the roots, but only opens up trees for ones that
do not appear in an excluded pack.)
But there are certain cases where we *do* need to see the parents of an
object in an excluded pack. Namely, if an object is rescue-able, but
only reachable from object(s) which appear in excluded packs, then
commit simplification will exclude those commits from the object
traversal, and we will never see a copy of that object, and thus not
rescue it.
This is what causes the failure in the previous commit during repacking.
When performing a geometric repack, packs above the geometric split that
weren't part of the previous MIDX (e.g., packs pushed directly into
`$GIT_DIR/objects/pack`) may not have full object closure. When those
packs are listed as excluded via the '^' marker, the reachability
traversal encounters the sequence described above, and may miss objects
which we expect to rescue with `--stdin-packs=follow`.
Introduce a new "excluded-open" pack prefix, '!'. Like '^'-prefixed
packs, objects from '!'-prefixed packs are excluded from the resulting
pack. But unlike '^', commits in '!'-prefixed packs *are* used as
starting points for the follow traversal, and the traversal does not
treat them as a closure boundary.
In order to distinguish excluded-closed from excluded-open packs during
the traversal, introduce a new `pack_keep_in_core_open` bit on
`struct packed_git`, along with a corresponding `KEPT_PACK_IN_CORE_OPEN`
flag for the kept-pack cache.
In `add_object_entry_from_pack()`, move the `want_object_in_pack()`
check to *after* `add_pending_oid()`. This is necessary so that commits
from excluded-open packs are added as traversal tips even though their
objects won't appear in the output. As a consequence, the caller
`for_each_object_in_pack()` will always provide a non-NULL 'p', hence we
are able to drop the "if (p)" conditional.
The `include_check` and `include_check_obj` callbacks on `rev_info` are
used to halt the walk at closed-excluded packs, since objects behind a
'^' boundary are guaranteed to have closure and need not be rescued.
The following commit will make use of this new functionality within the
repack layer to resolve the test failure demonstrated in the previous
commit.
Signed-off-by: Taylor Blau <me@ttaylorr.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The '--stdin-packs' mode of pack-objects maintains two separate
string_lists: one for included packs, and one for excluded packs. Each
list stores the pack basename as a string and the corresponding
`packed_git` pointer in its `->util` field.
This works, but makes it awkward to extend the set of pack "kinds" that
pack-objects can accept via stdin, since each new kind would need its
own string_list and duplicated handling. A future commit will want to do
just this, so prepare for that change by handling the various "kinds" of
packs specified over stdin in a more generic fashion.
Namely, replace the two `string_list`s with a single `strmap` keyed on
the pack basename, with values pointing to a new `struct
stdin_pack_info`. This struct tracks both the `packed_git` pointer and a
`kind` bitfield indicating whether the pack was specified as included or
excluded.
Extract the logic for sorting packs by mtime and adding their objects
into a separate `stdin_packs_add_pack_entries()` helper.
While we could have used a `string_list`, we must handle the case where
the same pack is specified more than once. With a `string_list` only, we
would have to pay a quadratic cost to either (a) insert elements into
their sorted positions, or (b) a repeated linear search, which is
accidentally quadratic. For that reason, use a strmap instead.
This patch does not include any functional changes.
Signed-off-by: Taylor Blau <me@ttaylorr.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The `read_stdin_packs()` function added originally via 339bce27f4
(builtin/pack-objects.c: add '--stdin-packs' option, 2021-02-22)
declares a `rev_info` struct but neglects to call `release_revisions()`
on it before returning, creating the potential for a leak.
The related change in 97ec43247c (pack-objects: declare 'rev_info' for
'--stdin-packs' earlier, 2025-06-23) carried forward this oversight and
did not address it.
Ensure that we call `release_revisions()` appropriately to prevent a
potential leak from this function. Note that in practice our `rev_info`
here does not have a present leak, hence t5331 passes cleanly before
this commit, even when built with SANITIZE=leak.
Signed-off-by: Taylor Blau <me@ttaylorr.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Reduce system overhead "git upload-pack" spends on relaying "git
pack-objects" output to the "git fetch" running on the other end of
the connection.
* ps/upload-pack-buffer-more-writes:
builtin/pack-objects: reduce lock contention when writing packfile data
csum-file: drop `hashfd_throughput()`
csum-file: introduce `hashfd_ext()`
sideband: use writev(3p) to send pktlines
wrapper: introduce writev(3p) wrappers
compat/posix: introduce writev(3p) wrapper
upload-pack: reduce lock contention when writing packfile data
upload-pack: prefer flushing data over sending keepalive
upload-pack: adapt keepalives based on buffering
upload-pack: fix debug statement when flushing packfile data
The `odb_for_each_object()` function only accepts a bitset of flags. In
a subsequent commit we'll want to change object iteration to also
support iterating over only those objects that have a specific prefix.
While we could of course add the prefix to the function signature, or
alternatively introduce a new function, both of these options don't
really seem to be that sensible.
Instead, introduce a new `struct odb_for_each_object_options` that can
be passed to a new `odb_for_each_object_ext()` function. Splice through
the options structure into the respective object database sources.
Signed-off-by: Patrick Steinhardt <ps@pks.im>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When running `git pack-objects --stdout` we feed the data through
`hashfd_ext()` with a progress meter and a smaller-than-usual buffer
length of 8kB so that we can track throughput more granularly. But as
packfiles tend to be on the larger side, this small buffer size may
cause a ton of write(3p) syscalls.
Originally, the buffer we used in `hashfd()` was 8kB for all use cases.
This was changed though in 2ca245f8be (csum-file.h: increase hashfile
buffer size, 2021-05-18) because we noticed that the number of writes
can have an impact on performance. So the buffer size was increased to
128kB, which improved performance a bit for some use cases.
But the commit didn't touch the buffer size for `hashd_throughput()`.
The reasoning here was that callers expect the progress indicator to
update frequently, and a larger buffer size would of course reduce the
update frequency especially on slow networks.
While that is of course true, there was (and still is, even though it's
now a call to `hashfd_ext()`) only a single caller of this function in
git-pack-objects(1). This command is responsible for writing packfiles,
and those packfiles are often on the bigger side. So arguably:
- The user won't care about increments of 8kB when packfiles tend to
be megabytes or even gigabytes in size.
- Reducing the number of syscalls would be even more valuable here
than it would be for multi-pack indices, which was the benchmark
done in the mentioned commit, as MIDXs are typically significantly
smaller than packfiles.
- Nowadays, many internet connections should be able to transfer data
at a rate significantly higher than 8kB per second.
Update the buffer to instead have a size of `LARGE_PACKET_DATA_MAX - 1`,
which translates to ~64kB. This limit was chosen because `git
pack-objects --stdout` is most often used when sending packfiles via
git-upload-pack(1), where packfile data is chunked into pktlines when
using the sideband. Furthermore, most internet connections should have a
bandwidth signifcantly higher than 64kB/s, so we'd still be able to
observe progress updates at a rate of at least once per second.
This change significantly reduces the number of write(3p) syscalls from
355,000 to 44,000 when packing the Linux repository. While this results
in a small performance improvement on an otherwise-unused system, this
improvement is mostly negligible. More importantly though, it will
reduce lock contention in the kernel on an extremely busy system where
we have many processes writing data at once.
Suggested-by: Jeff King <peff@peff.net>
Signed-off-by: Patrick Steinhardt <ps@pks.im>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The `hashfd_throughput()` function is used by a single callsite in
git-pack-objects(1). In contrast to `hashfd()`, this function uses a
progress meter to measure throughput and a smaller buffer length so that
the progress meter can provide more granular metrics.
We're going to change that caller in the next commit to be a bit more
specific to packing objects. As such, `hashfd_throughput()` will be a
somewhat unfitting mechanism for any potential new callers.
Drop the function and replace it with a call to `hashfd_ext()`.
Signed-off-by: Patrick Steinhardt <ps@pks.im>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The object source API is getting restructured to allow plugging new
backends.
* ps/odb-sources:
odb/source: make `begin_transaction()` function pluggable
odb/source: make `write_alternate()` function pluggable
odb/source: make `read_alternates()` function pluggable
odb/source: make `write_object_stream()` function pluggable
odb/source: make `write_object()` function pluggable
odb/source: make `freshen_object()` function pluggable
odb/source: make `for_each_object()` function pluggable
odb/source: make `read_object_stream()` function pluggable
odb/source: make `read_object_info()` function pluggable
odb/source: make `close()` function pluggable
odb/source: make `reprepare()` function pluggable
odb/source: make `free()` function pluggable
odb/source: introduce source type for robustness
odb: move reparenting logic into respective subsystems
odb: embed base source in the "files" backend
odb: introduce "files" source
odb: split `struct odb_source` into separate header
The parse-options API learned to notice an options[] array with
duplicated long options.
* rs/parse-options-duplicated-long-options:
parseopt: check for duplicate long names and numerical options
pack-objects: remove duplicate --stdin-packs definition
The "files" backend is implemented as a pointer in the `struct
odb_source`. This contradicts our typical pattern for pluggable backends
like we use it for example in the ref store or for object database
streams, where we typically embed the generic base structure in the
specialized implementation. This pattern has a couple of small benefits:
- We avoid an extra allocation.
- We hide implementation details in the generic structure.
- We can easily downcast from a generic backend to the specialized
structure and vice versa because the offsets are known at compile
time.
- It becomes trivial to identify locations where we depend on backend
specific logic because the cast needs to be explicit.
Refactor our "files" object database source to do the same and embed the
`struct odb_source` in the `struct odb_source_files`.
There are still a bunch of sites in our code base where we do have to
access internals of the "files" backend. The intent is that those will
go away over time, but this will certainly take a while. Meanwhile,
provide a `odb_source_files_downcast()` function that can convert a
generic source into a "files" source.
As we only have a single source the downcast succeeds unconditionally
for now. Eventually though the intent is to make the cast `BUG()` in
case the caller requests to downcast a non-"files" backend to a "files"
backend.
Signed-off-by: Patrick Steinhardt <ps@pks.im>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Introduce a new "files" object database source. This source encapsulates
access to both loose object files and the packfile store, similar to how
the "files" backend for refs encapsulates access to loose refs and the
packed-refs file.
Note that for now the "files" source is still a direct member of a
`struct odb_source`. This architecture will be reversed in the next
commit so that the files source contains a `struct odb_source`.
Signed-off-by: Patrick Steinhardt <ps@pks.im>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Revamp object enumeration API around odb.
* ps/odb-for-each-object:
odb: drop unused `for_each_{loose,packed}_object()` functions
reachable: convert to use `odb_for_each_object()`
builtin/pack-objects: use `packfile_store_for_each_object()`
odb: introduce mtime fields for object info requests
treewide: drop uses of `for_each_{loose,packed}_object()`
treewide: enumerate promisor objects via `odb_for_each_object()`
builtin/fsck: refactor to use `odb_for_each_object()`
odb: introduce `odb_for_each_object()`
packfile: introduce function to iterate through objects
packfile: extract function to iterate through objects of a store
object-file: introduce function to iterate through objects
object-file: extract function to read object info from path
odb: fix flags parameter to be unsigned
odb: rename `FOR_EACH_OBJECT_*` flags
A handful of places used refs_for_each_ref_in() API incorrectly,
which has been corrected.
* ps/for-each-ref-in-fixes:
bisect: simplify string_list memory handling
bisect: fix misuse of `refs_for_each_ref_in()`
pack-bitmap: fix bug with exact ref match in "pack.preferBitmapTips"
pack-bitmap: deduplicate logic to iterate over preferred bitmap tips
cd846bacc7 (pack-objects: introduce '--stdin-packs=follow', 2025-06-23)
added a new definition of the option --stdin-packs that accepts an
argument. It kept the old definition, which still shows up in the short
help, but is shadowed by the new one. Remove it.
Hinted-at-by: Junio C Hamano <gitster@pobox.com>
Signed-off-by: René Scharfe <l.s.r@web.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
"git pack-objects --stdin-packs" with "--exclude-promisor-objects"
fetched objects that are promised, which was not wanted. This has
been fixed.
* ps/pack-concat-wo-backfill:
builtin/pack-objects: don't fetch objects when merging packs
* ps/odb-for-each-object:
odb: drop unused `for_each_{loose,packed}_object()` functions
reachable: convert to use `odb_for_each_object()`
builtin/pack-objects: use `packfile_store_for_each_object()`
odb: introduce mtime fields for object info requests
treewide: drop uses of `for_each_{loose,packed}_object()`
treewide: enumerate promisor objects via `odb_for_each_object()`
builtin/fsck: refactor to use `odb_for_each_object()`
odb: introduce `odb_for_each_object()`
packfile: introduce function to iterate through objects
packfile: extract function to iterate through objects of a store
object-file: introduce function to iterate through objects
object-file: extract function to read object info from path
odb: fix flags parameter to be unsigned
odb: rename `FOR_EACH_OBJECT_*` flags
We have two locations that iterate over the preferred bitmap tips as
configured by the user via "pack.preferBitmapTips". Both of these
callsites are subtly wrong: when the preferred bitmap tips contain an
exact refname match, then we will hit a `BUG()`.
Prepare for the fix by unifying the two callsites into a new
`for_each_preferred_bitmap_tip()` function.
This removes the last callsite of `bitmap_preferred_tips()` outside of
"pack-bitmap.c". As such, convert the function to be local to that file
only. Note that the function is still used by a second caller, so we
cannot just inline it.
Signed-off-by: Patrick Steinhardt <ps@pks.im>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The "--stdin-packs" option can be used to merge objects from multiple
packfiles given via stdin into a new packfile. One big upside of this
option is that we don't have to perform a complete rev walk to enumerate
objects. Instead, we can simply enumerate all objects that are part of
the specified packfiles, which can be significantly faster in very large
repositories.
There is one downside though: when we don't perform a rev walk we also
don't have a good way to learn about the respective object's names. As a
consequence, we cannot use the name hashes as a heuristic to get better
delta selection.
We try to offset this downside though by performing a localized rev
walk: we queue all objects that we're about to repack as interesting,
and all objects from excluded packfiles as uninteresting. We then
perform a best-effort rev walk that allows us to fill in object names.
There is one gotcha here though: when "--exclude-promisor-objects" has
not been given we will perform backfill fetches for any promised objects
that are missing. This used to not be an issue though as this option was
mutually exclusive with "--stdin-packs". But that has changed recently,
and starting with dcc9c7ef47 (builtin/repack: handle promisor packs with
geometric repacking, 2026-01-05) we will now repack promisor packs
during geometric compaction. The consequence is that a geometric repack
may now perform a bunch of backfill fetches.
We of course cannot pass "--exclude-promisor-objects" to fix this
issue -- after all, the whole intent is to repack objects part of a
promisor pack. But arguably we don't have to: the rev walk is intended
as best effort, and we already configure it to ignore missing links to
other objects. So we can adapt the walk to unconditionally disable
fetching any missing objects.
Do so and add a test that verifies we don't backfill any objects.
Reported-by: Lukas Wanko <lwanko@gitlab.com>
Signed-off-by: Patrick Steinhardt <ps@pks.im>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The following command reports that some header files are included twice:
$ git grep '#include' '*.c' | sort | uniq -cd
Remove the second #include line in each case, as it has no effect.
Signed-off-by: René Scharfe <l.s.r@web.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Many callsites of string_list_remove_duplicates() call it
immdediately after calling string_list_sort(), understandably
as the former requires string-list to be sorted, it is clear
that these places are sorting only to remove duplicates and
for no other reason.
Introduce a helper function string_list_sort_u that combines
these two calls that often appear together, to simplify
these callsites. Replace the current calls of those methods with
string_list_sort_u().
Signed-off-by: Amisha Chhajed <amishhhaaaa@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When enumerating objects that are supposed to be stored in a new cruft
pack we use `for_each_packed_object()` and then derive each object's
mtime individually. Refactor this logic to instead use the new
`packfile_store_for_each_object()` function with an object info request
that asks for the respective mtimes.
Signed-off-by: Patrick Steinhardt <ps@pks.im>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Rename the `FOR_EACH_OBJECT_*` flags to have an `ODB_` prefix. This
prepares us for a new upcoming `odb_for_each_object()` function and
ensures that both the function and its flags have the same prefix.
Signed-off-by: Patrick Steinhardt <ps@pks.im>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
"git repack --geometric" did not work with promisor packs, which
has been corrected.
* ps/geometric-repacking-with-promisor-remotes:
builtin/repack: handle promisor packs with geometric repacking
repack-promisor: extract function to remove redundant packs
repack-promisor: extract function to finalize repacking
repack-geometry: extract function to compute repacking split
builtin/pack-objects: exclude promisor objects with "--stdin-packs"
The object-info API has been cleaned up.
* ps/read-object-info-improvements:
packfile: drop repository parameter from `packed_object_info()`
packfile: skip unpacking object header for disk size requests
packfile: disentangle return value of `packed_object_info()`
packfile: always populate pack-specific info when reading object info
packfile: extend `is_delta` field to allow for "unknown" state
packfile: always declare object info to be OI_PACKED
object-file: always set OI_LOOSE when reading object info
The packfile_store data structure is moved from object store to odb
source.
* ps/packfile-store-in-odb-source:
packfile: move MIDX into packfile store
packfile: refactor `find_pack_entry()` to work on the packfile store
packfile: inline `find_kept_pack_entry()`
packfile: only prepare owning store in `packfile_store_prepare()`
packfile: only prepare owning store in `packfile_store_get_packs()`
packfile: move packfile store into object source
packfile: refactor misleading code when unusing pack windows
packfile: refactor kept-pack cache to work with packfile stores
packfile: pass source to `prepare_pack()`
packfile: create store via its owning source
* ps/packfile-store-in-odb-source:
packfile: move MIDX into packfile store
packfile: refactor `find_pack_entry()` to work on the packfile store
packfile: inline `find_kept_pack_entry()`
packfile: only prepare owning store in `packfile_store_prepare()`
packfile: only prepare owning store in `packfile_store_get_packs()`
packfile: move packfile store into object source
packfile: refactor misleading code when unusing pack windows
packfile: refactor kept-pack cache to work with packfile stores
packfile: pass source to `prepare_pack()`
packfile: create store via its owning source
odb: properly close sources before freeing them
builtin/gc: fix condition for whether to write commit graphs
* ps/read-object-info-improvements:
packfile: drop repository parameter from `packed_object_info()`
packfile: skip unpacking object header for disk size requests
packfile: disentangle return value of `packed_object_info()`
packfile: always populate pack-specific info when reading object info
packfile: extend `is_delta` field to allow for "unknown" state
packfile: always declare object info to be OI_PACKED
object-file: always set OI_LOOSE when reading object info
It is currently not possible to combine "--exclude-promisor-objects"
with "--stdin-packs" because both flags want to set up a revision walk
to enumerate the objects to pack. In a subsequent commit though we want
to extend geometric repacks to support promisor objects, and for that we
need to handle the combination of both flags.
There are two cases we have to think about here:
- "--stdin-packs" asks us to pack exactly the objects part of the
specified packfiles. It is somewhat questionable what to do in the
case where the user asks us to exclude promisor objects, but at the
same time explicitly passes a promisor pack to us. For now, we
simply abort the request as it is self-contradicting. As we have
also been dying before this commit there is no regression here.
- "--stdin-packs=follow" does the same as the first flag, but it also
asks us to include all objects transitively reachable from any
object in the packs we are about to repack. This is done by doing
the revision walk mentioned further up. Luckily, fixing this case is
trivial: we only need to modify the revision walk to also set the
`exclude_promisor_objects` field.
Note that we do not support the "--exclude-promisor-objects-best-effort"
flag for now as we don't need it to support geometric repacking with
promisor objects.
Signed-off-by: Patrick Steinhardt <ps@pks.im>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The function `packed_object_info()` takes a packfile and offset and
returns the object info for the corresponding object. Despite these two
parameters though it also takes a repository pointer. This is redundant
information though, as `struct packed_git` already has a repository
pointer that is always populated.
Drop the redundant parameter.
Signed-off-by: Patrick Steinhardt <ps@pks.im>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The packfile store is a member of `struct object_database`, which means
that we have a single store per database. This doesn't really make much
sense though: each source connected to the database has its own set of
packfiles, so there is a conceptual mismatch here. This hasn't really
caused much of a problem in the past, but with the advent of pluggable
object databases this is becoming more of a problem because some of the
sources may not even use packfiles in the first place.
Move the packfile store down by one level from the object database into
the object database source. This ensures that each source now has its
own packfile store, and we can eventually start to abstract it away
entirely so that the caller doesn't even know what kind of store it
uses.
Note that we only need to adjust a relatively small number of callers,
way less than one might expect. This is because most callers are using
`repo_for_each_pack()`, which handles enumeration of all packfiles that
exist in the repository. So for now, none of these callers need to be
adapted. The remaining callers that iterate through the packfiles
directly and that need adjustment are those that are a bit more tangled
with packfiles. These will be adjusted over time.
Note that this patch only moves the packfile store, and there is still a
bunch of functions that seemingly operate on a packfile store but that
end up iterating over all sources. These will be adjusted in subsequent
commits.
Signed-off-by: Patrick Steinhardt <ps@pks.im>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The kept pack cache is a cache of packfiles that are marked as kept
either via an accompanying ".kept" file or via an in-memory flag. The
cache can be retrieved via `kept_pack_cache()`, where one needs to pass
in a repository.
Ultimately though the kept-pack cache is a property of the packfile
store, and this causes problems in a subsequent commit where we want to
move down the packfile store to be a per-object-source entity.
Prepare for this and refactor the kept-pack cache to work on top of a
packfile store instead. While at it, rename both the function and flags
specific to the kept-pack cache so that they can be properly attributed
to the respective subsystems.
Signed-off-by: Patrick Steinhardt <ps@pks.im>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Allow callers of parse_tag() pass in the repository to use. Let most of
them pass in the_repository to get the same result as before. One of
them has stopped using the_repository in ef9b0370da (sha1-name.c: store
and use repo in struct disambiguate_state, 2019-04-16); let it pass in
its stored repository.
Signed-off-by: René Scharfe <l.s.r@web.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The "git_istream" abstraction has been revamped to make it easier
to interface with pluggable object database design.
* ps/object-read-stream:
streaming: drop redundant type and size pointers
streaming: move into object database subsystem
streaming: refactor interface to be object-database-centric
streaming: move logic to read packed objects streams into backend
streaming: move logic to read loose objects streams into backend
streaming: make the `odb_read_stream` definition public
streaming: get rid of `the_repository`
streaming: rely on object sources to create object stream
packfile: introduce function to read object info from a store
streaming: move zlib stream into backends
streaming: create structure for filtered object streams
streaming: create structure for packed object streams
streaming: create structure for loose object streams
streaming: create structure for in-core object streams
streaming: allocate stream inside the backend-specific logic
streaming: explicitly pass packfile info when streaming a packed object
streaming: propagate final object type via the stream
streaming: drop the `open()` callback function
streaming: rename `git_istream` into `odb_read_stream`
* ps/object-read-stream:
streaming: drop redundant type and size pointers
streaming: move into object database subsystem
streaming: refactor interface to be object-database-centric
streaming: move logic to read packed objects streams into backend
streaming: move logic to read loose objects streams into backend
streaming: make the `odb_read_stream` definition public
streaming: get rid of `the_repository`
streaming: rely on object sources to create object stream
packfile: introduce function to read object info from a store
streaming: move zlib stream into backends
streaming: create structure for filtered object streams
streaming: create structure for packed object streams
streaming: create structure for loose object streams
streaming: create structure for in-core object streams
streaming: allocate stream inside the backend-specific logic
streaming: explicitly pass packfile info when streaming a packed object
streaming: propagate final object type via the stream
streaming: drop the `open()` callback function
streaming: rename `git_istream` into `odb_read_stream`
A part of code paths that deals with loose objects has been cleaned
up.
* ps/object-source-loose:
object-file: refactor writing objects via a stream
object-file: rename `write_object_file()`
object-file: refactor freshening of objects
object-file: rename `has_loose_object()`
object-file: read objects via the loose object source
object-file: move loose object map into loose source
object-file: hide internals when we need to reprepare loose sources
object-file: move loose object cache into loose source
object-file: introduce `struct odb_source_loose`
object-file: move `fetch_if_missing`
odb: adjust naming to free object sources
odb: introduce `odb_source_new()`
odb: fix subtle logic to check whether an alternate is usable
In the preceding commits we have turned `struct odb_read_stream` into a
publicly visible structure. Furthermore, this structure now contains the
type and size of the object that we are about to stream. Consequently,
the out-pointers that we used before to propagate the type and size of
the streamed object are now somewhat redundant with the data contained
in the structure itself.
Drop these out-pointers and adapt callers accordingly.
Signed-off-by: Patrick Steinhardt <ps@pks.im>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The "streaming" terminology is somewhat generic, so it may not be
immediately obvious that "streaming.{c,h}" is specific to the object
database. Rectify this by moving it into the "odb/" directory so that it
can be immediately attributed to the object subsystem.
Signed-off-by: Patrick Steinhardt <ps@pks.im>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Refactor the streaming interface to be centered around object databases
instead of centered around the repository. Rename the functions
accordingly.
Signed-off-by: Patrick Steinhardt <ps@pks.im>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
In the following patches we are about to make the `git_istream` more
generic so that it becomes fully controlled by the specific object
source that wants to create it. As part of these refactorings we'll
fully move the structure into the object database subsystem.
Prepare for this change by renaming the structure from `git_istream`
to `odb_read_stream`. This mirrors the `odb_write_stream` structure that
we already have.
Signed-off-by: Patrick Steinhardt <ps@pks.im>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Some ref backend storage can hold not just the object name of an
annotated tag, but the object name of the object the tag points at.
The code to handle this information has been streamlined.
* ps/ref-peeled-tags:
t7004: do not chdir around in the main process
ref-filter: fix stale parsed objects
ref-filter: parse objects on demand
ref-filter: detect broken tags when dereferencing them
refs: don't store peeled object IDs for invalid tags
object: add flag to `peel_object()` to verify object type
refs: drop infrastructure to peel via iterators
refs: drop `current_ref_iter` hack
builtin/show-ref: convert to use `reference_get_peeled_oid()`
ref-filter: propagate peeled object ID
upload-pack: convert to use `reference_get_peeled_oid()`
refs: expose peeled object ID via the iterator
refs: refactor reference status flags
refs: fully reset `struct ref_iterator::ref` on iteration
refs: introduce `.ref` field for the base iterator
refs: introduce wrapper struct for `each_ref_fn`
Both the "files" and "reftable" backend are able to store peeled values
for tags in the respective formats. This allows for a more efficient
lookup of the target object of such a tag without having to manually
peel via the object database.
The infrastructure to access these peeled object IDs is somewhat funky
though. When iterating through objects, we store a pointer reference to
the current iterator in a global variable. The callbacks invoked by that
iterator are then expected to call `peel_iterated_oid()`, which checks
whether the globally-stored iterator's current reference refers to the
one handed into that function. If so, we ask the iterator to peel the
object, otherwise we manually peel the object via the object database.
Depending on global state like this is somewhat weird and also quite
fragile.
Introduce a new `struct reference::peeled_oid` field that can be
populated by the reference backends. This field can be accessed via a
new function `reference_get_peeled_oid()` that either uses that value,
if set, or alternatively peels via the ODB. With this change we don't
have to rely on global state anymore, but make the peeled object ID
available to the callback functions directly.
Adjust trivial callers that already have a `struct reference` available.
Remaining callers will be adjusted in subsequent commits.
Signed-off-by: Patrick Steinhardt <ps@pks.im>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The `each_ref_fn` callback function type is used across our code base
for several different functions that iterate through reference. There's
a bunch of callbacks implementing this type, which makes any changes to
the callback signature extremely noisy. An example of the required churn
is e8207717f1 (refs: add referent to each_ref_fn, 2024-08-09): adding a
single argument required us to change 48 files.
It was already proposed back then [1] that we might want to introduce a
wrapper structure to alleviate the pain going forward. While this of
course requires the same kind of global refactoring as just introducing
a new parameter, it at least allows us to more change the callback type
afterwards by just extending the wrapper structure.
One counterargument to this refactoring is that it makes the structure
more opaque. While it is obvious which callsites need to be fixed up
when we change the function type, it's not obvious anymore once we use
a structure. That being said, we only have a handful of sites that
actually need to populate this wrapper structure: our ref backends,
"refs/iterator.c" as well as very few sites that invoke the iterator
callback functions directly.
Introduce this wrapper structure so that we can adapt the iterator
interfaces more readily.
[1]: <ZmarVcF5JjsZx0dl@tanuki>
Signed-off-by: Patrick Steinhardt <ps@pks.im>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Junio C Hamano