| Age | Commit message (Collapse) | Author | Files | Lines |
|---|
|
A collection of patches from Git for Windows has been upstreamed,
mostly focusing on simplifying and robustifying build configurations
for MinGW/MSYS2, dropping obsolete compatibility options, and allowing
the main 'git.exe' to be used directly without the extra wrapper
process on Windows.
* js/mingw-build-updates:
t0060: adjust the code style
mingw: allow `git.exe` to be used instead of the "Git wrapper"
mingw: ensure valid CTYPE
mingw: always define `ETC_*` for MSYS2 environments
windows: skip linking `git-<command>` for built-ins
mingw: rely on MSYS2's metadata instead of hard-coding it
mingw: only enable the MSYS2-specific stuff when compiling in MSYS2
mingw: set the prefix and HOST_CPU as per MSYS2's settings
mingw: avoid over-specifying `--pic-executable`
mingw: only use -Wl,--large-address-aware for 32-bit builds
mingw: drop the -D_USE_32BIT_TIME_T option
mingw: stop hard-coding `CC = gcc`
mingw: include the Python parts in the build
|
|
These days, the test cases are less free-form than in the wild old days
of the Git project.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
|
Git for Windows wants to add `git.exe` to the users' `PATH`, without
cluttering the latter with unnecessary executables such as `wish.exe`.
To that end, it invented the concept of its "Git wrapper", i.e. a tiny
executable located in `C:\Program Files\Git\cmd\git.exe` (originally a
CMD script) whose sole purpose is to set up a couple of environment
variables and then spawn the _actual_ `git.exe` (which nowadays lives in
`C:\Program Files\Git\mingw64\bin\git.exe` for 64-bit, and the obvious
equivalent for 32-bit installations).
Currently, the following environment variables are set unless already
initialized:
- `MSYSTEM`, to make sure that the MSYS2 Bash and the MSYS2 Perl
interpreter behave as expected, and
- `PLINK_PROTOCOL`, to force PuTTY's `plink.exe` to use the SSH
protocol instead of Telnet,
- `PATH`, to make sure that the `bin` folder in the user's home
directory, as well as the `/mingw64/bin` and the `/usr/bin`
directories are included. The trick here is that the `/mingw64/bin/`
and `/usr/bin/` directories are relative to the top-level installation
directory of Git for Windows (which the included Bash interprets as
`/`, i.e. as the MSYS pseudo root directory).
Using the absence of `MSYSTEM` as a tell-tale, we can detect in
`git.exe` whether these environment variables have been initialized
properly. Therefore we can call `C:\Program Files\Git\mingw64\bin\git`
in-place after this change, without having to call Git through the Git
wrapper.
Obviously, above-mentioned directories must be _prepended_ to the `PATH`
variable, otherwise we risk picking up executables from unrelated Git
installations. We do that by constructing the new `PATH` value from
scratch, appending `$HOME/bin` (if `HOME` is set), then the MSYS2 system
directories, and then appending the original `PATH`.
Side note: this modification of the `PATH` variable is independent of
the modification necessary to reach the executables and scripts in
`/mingw64/libexec/git-core/`, i.e. the `GIT_EXEC_PATH`. That
modification is still performed by Git, elsewhere, long after making the
changes described above.
While we _still_ cannot simply hard-link `mingw64\bin\git.exe` to `cmd`
(because the former depends on a couple of `.dll` files that are only in
`mingw64\bin`, i.e. calling `...\cmd\git.exe` would fail to load due to
missing dependencies), at least we can now avoid that extra process of
running the Git wrapper (which then has to wait for the spawned
`git.exe` to finish) by calling `...\mingw64\bin\git.exe` directly, via
its absolute path.
Testing this is in Git's test suite tricky: we set up a "new" MSYS
pseudo-root and copy the `git.exe` file into the appropriate location,
then verify that `MSYSTEM` is set properly, and also that the `PATH` is
modified so that scripts can be found in `$HOME/bin`, `/mingw64/bin/`
and `/usr/bin/`.
This addresses https://github.com/git-for-windows/git/issues/2283
Note: This keeps the same, hard-coded MSYSTEM platform support for CMake
as before, and introduces an `msystem' and `mingw-prefix` knob for Meson
(read: neither CMake nor Meson will automatically inherit the setting
from the current build environment).
Helped-by: Johannes Sixt <j6t@kdbg.org>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
|
The documentation for 'git cherry-pick' has been updated to clarify
that the '--no-commit' option intentionally skips setting the
'CHERRY_PICK_HEAD' ref. A test has also been added to ensure this
behavior holds even when the operation stops for conflicts.
* as/cherry-pick-no-commit-doc:
doc: cherry-pick: note --no-commit skips CHERRY_PICK_HEAD
t3507: check no CHERRY_PICK_HEAD after conflicting --no-commit
|
|
The process of downloading packfile URIs in protocol v2 has been
instrumented with a Trace2 region. This visibility allows tracking
the cumulative time spent downloading external packs and the number
of advertised URIs without emitting a separate event per pack.
* tn/fetch-pack-trace-packfile-uri:
fetch-pack: trace packfile URI downloads
|
|
The git rev-list command has been augmented with a '--missing-only'
option that filters the output to only show missing objects,
stripping the leading '?' character and suppressing present objects,
which is useful when used in combination with '--missing=print' or
'--missing=print-info'.
* sa/rev-list-missing-only:
rev-list: add --missing-only option to filter output
|
|
"git maintenance" triggered "rerere gc" in unappropriate times and
interfered with "git rebase" etc. too much. The conditions "rerere
gc" gets triggered have been tweaked.
* ps/tune-rerere-gc:
builtin/maintenance: improve heuristic for "rerere gc"
rerere: extract logic to determine whether entries are stale
|
|
The autostash fallback in 'git checkout -m' has been refined to only
retry when there are local changes. Additionally, a blank line now
visually separates autostash conflict advice from the subsequent
branch-switch message.
* hn/checkout-m-autostash-refine:
checkout: separate autostash conflict advice from branch-switch message
stash: reserve exit status 1 for conflicts
|
|
When a protocol v2 fetch includes packfile URIs, the client downloads
each advertised pack in a separate http-fetch process. Existing Trace2
regions cover negotiation, but not the time spent downloading these
packs or the number of advertised URIs.
Add a Trace2 region around the packfile URI download loop and record the
number of URIs. This makes the cost of downloading external packs
visible without emitting an event for each pack.
Extend the existing packfile URI test to verify the region and count.
Signed-off-by: Ted Nyman <tnyman@openai.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
|
The error path in 'git submodule--helper' has been updated to plug a
memory leak when a repository handle could not be obtained,
leveraging an updated idempotent repo_clear().
* jk/submodule-error-leak:
submodule--helper: free URL when repository setup fails
repository: make repo_clear() idempotent
|
|
Teach 'am', 'revert', and 'rebase' that running 'commit --amend' or a
partial 'commit <paths>' makes no sense during operations that stop
and return control to the user to resolve conflicts left in the
working tree, just like 'cherry-pick' and 'merge' do.
* en/no-amend-during-conflicts:
commit: refuse partial commits during conflict resolution
commit: refuse to amend during conflict resolution
commit: reword the empty-commit rebase amend error
commit: allow a partial commit when a rebase pick becomes empty
commit: clarify FROM_REBASE_PICK and is_from_rebase() names
|
|
CGI helper scripts used by HTTP-related test scripts have been updated
to use atomic filesystem operations, preventing race conditions when
Apache handles concurrent requests.
* mm/lib-httpd-cgi-safe:
t/lib-httpd: document writing concurrency-safe CGI helpers
t/lib-httpd: make http-429 first-request check atomic
t/lib-httpd: fix apply-one-time-script race under concurrent requests
|
|
Whether CHERRY_PICK_HEAD is written depends on the command, on whether
the merge started, and on --no-commit, all in one condition in
do_pick_commit(). The suite checks the clean --no-commit pick; nothing
checks the conflicting one.
The test that already runs a conflicting --no-commit pick compares the
advice the command prints, which is what tells us it stopped on a
conflict. Assert the ref is missing there too.
Signed-off-by: Aleksei Sviridkin <f@lex.la>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
|
The object lookup machinery has been taught to gracefully recover
when a multi-pack-index points to an owning pack that was removed
during a concurrent geometric repack, and 'git replay' has been
fixed to not segfault when reading such missing objects.
* en/midx-missing-pack-fallback:
packfile: recover when a multi-pack-index names a removed pack
mktree: do not use OBJECT_INFO_QUICK when checking objects
mktree: plug per-tree leak in --batch mode
replay: fail gracefully when a merge input is unreadable
|
|
The memory ownership of argv elements passed to the revision
machinery has been made more robust by keeping logically "freed"
elements alive until the rev_info struct is released, preventing
use-after-free bugs when options store references to them.
* jk/rev-info-argv-to-free:
revision: simplify mark_argv_for_free() callers
revision: hang on to "freed" argv elements
|
|
The 'git replay' command has been taught the '--linearize' option to
drop merge commits and linearize the replayed history, mimicking 'git
rebase --no-rebase-merges'.
* tc/replay-linearize:
replay: offer an option to linearize the commit topology
replay: resolve the replay base outside pick_regular_commit()
replay: add helper to put entry into replayed_commits
|
|
The "rerere-gc" maintenance task is responsible for pruning rerere
entries older than a certain configurable cutoff point. Whether or not
the task gets run during auto-maintenance can be configured via
"maintenance.rerere-gc.auto":
- A negative value indicates that maintenance should always run.
- A zero value indicates that maintenance should never run.
- Otherwise, a positive value indicates that maintenance should always
run in case we have at least a single rerere entry.
While the first two conditions are sensible, the last one is less so as
it does not account for whether we would even prune old entries in the
first place. Instead, it effectively implies that we unconditionally
spawn "git rerere gc" when rerere is enabled. Chances are high though
that there is nothing to prune, as the default cutoff dates are 60 days
for resolved rerere entries and 15 days for unresolved ones.
Besides being a waste of compute, it also obstructs concurrent processes
that want to write new resolutions as garbage collection takes a central
lock file, as reported in [1]. That race is a longstanding one that
existed even before we introduced fine-grained maintenance tasks, and
the proper fix is to use a locking timeout in the writing processes. But
the race is made worse by us performing garbage collection a lot more
often.
Refine the heuristic to take into account whether any entries can be
pruned in the first place. This ensures that we'll only ever run this
task in situations where it will do anything, and should thus result in
a lot less frequent invocations of "git rerere gc".
Furthermore, tweak the meaning of "maintenance.rerere-gc.auto" so that
positive values allow the user to configure the number of prunable
entries that need to exist before we run it and set the default value to
512. This number is pulled out of thin air, but it ensures that we know
to batch-delete entries instead of pruning every single entry that is
older than the cutoff point.
Note that this now requires us to actually open the rerere-entry
directories and stat the individual files in there, which does add a bit
of overhead when one has lots of rerere entries. To counteract this
overhead, we thus use the same sampling heuristic as we do for loose
objects, where we only consider those entries that start with a "17".
[1]: <pull.2214.git.1788337897490.gitgitgadget@gmail.com>
Reported-by: Thomas Bachem <mail@thomasbachem.com>
Signed-off-by: Patrick Steinhardt <ps@pks.im>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
|
Various spelling mistakes in comments and test descriptions have
been corrected.
* hk/typofix:
versioncmp: fix typo in versioncmp.c, t/t0022-crlf-rename.sh
|
|
The string extraction logic for the branch name and worktree name
from the given path in 'git worktree add' has been corrected and
simplified to avoid out-of-bounds reads and improper handling of
trailing slashes.
* rs/worktree-add-basename-fixes:
worktree add: let worktree_basename() return string copy
worktree add: trim slashes when deriving branch name from path
worktree add: reject separator-only path
worktree add: don't read out of bounds in worktree_basename()
|
|
When working with partial clones, callers often need only the missing
object IDs. Today that means post-processing --missing=print to drop
present objects and strip the leading '?':
git rev-list --objects --all --missing=print | perl -ne 'print if s/^[?]//'
This is for a one-shot walk, not a fetch loop. Callers already have
--missing=print and strip the leading '?'. Gitaly does that when packing
a quarantine: '?' lines are objects that must already exist in the main
repo. Tests do the same (is this blob still missing). --missing-only is
just that list without the prefix.
Add --missing-only. Use it with --missing=print or --missing=print-info
to print only missing objects. --missing= still picks the format;
--missing-only only filters. The leading '?' is omitted. With
print-info, path= and type= are still shown.
Require --missing=print or --missing=print-info. Reject --count and
--disk-usage.
Signed-off-by: Siddharth Asthana <siddharthasthana31@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
|
A few tests for the reference handling subsystem have been added to
exercise the handling of forbidden characters and symbolic references.
* ns/ref-symref-additional-tests:
t1402: test forbidden characters in refnames
t1401: check symbolic-ref failure and --quiet silence on a non-symbolic ref
|
|
"git checkout -m" stashes the user's local changes when it cannot
perform the checkout, and then applies the stash. When applying the
stash results in conflicts, the advice on how to deal with them is
printed directly on top of the branch-switch message ("Switched to
branch ..."), making the two hard to tell apart. Print a blank line
in between so that the advice and the branch-switch message are
visually distinct.
apply_autostash_ref() reports whether applying the stash resulted in
conflicts via its enum stash_apply_result return value, so only print
the blank line in the conflicted case.
Signed-off-by: Harald Nordgren <haraldnordgren@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
|
"git stash apply", "pop" and "branch" exit with status 1 both when
applying the stash entry resulted in conflicts and when they fail for
other reasons, so callers cannot tell the two apart.
Follow the convention of "git merge-tree" and the merge strategies,
which exit with status 1 to indicate conflicts and with a different
non-zero status for errors: those subcommands now exit with status 1
only when applying the stash entry resulted in conflicts, in which
case the stash entry is left in place, and exit with status 128, the
status die() uses, when they fail for other reasons. Document the
exit statuses.
The only subcommand implementations that can return a positive value
are "apply", "pop" and "branch", which return the value of
do_apply_stash(): "apply" returns it directly, and "pop" and "branch"
drop the stash entry, via do_drop_stash(), which always returns 0,
only when the application succeeded. do_apply_stash() only returns a
positive value when the three-way merge was unclean. cmd_stash() now
maps negative values to 128 and passes positive values through as the
exit status, so exit status 1 unambiguously indicates conflicts.
enum stash_apply_result makes the convention explicit, and the
autostash helpers use it to tell users that their stashed changes
were saved when applying them fails.
Signed-off-by: Harald Nordgren <haraldnordgren@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
|
If repo setup fails, we'll return an error without freeing the allocated
url string, leaking the memory. The test suite does trigger this error,
but never with the leak. We only allocate a url if submodule_from_path()
returned something, but our tests use other situations, like totally
nonexistent submodules.
We can cover this case by asking about a submodule that exists but which
has not been initialized. The new test fails with SANITIZE=leak.
The smallest fix would just be a call to free(url), but I think it's a
little nicer to set up a dedicated out-path for cleanup here. The
previous commit made it safe to call repo_clear() even if
repo_submodule_init() fails.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
|
Similar to the previous commit, just as `git commit --amend` is a
foot-gun during conflict resolution, so is a partial commit (`git commit
<paths>`). Recording a conflict resolution is about capturing the state
of the entire tree on top of HEAD, not a subset of paths. For many years
we have rejected partial commits in the middle of
- a merge
- a cherry-pick
but, just like amending, this was never extended to the other operations
that can also leave conflicts to resolve:
- an `am` operation
- a revert
- a rebase that stopped for conflict resolution
Reuse sequencer_ongoing_operation(), introduced for the analogous
`--amend` check, to detect these and refuse the partial commit. A rebase
that stopped because a pick became empty is not conflict resolution and,
as an earlier patch established, is deliberately left permitted.
Signed-off-by: Elijah Newren <newren@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
|
Running `git commit --amend` during conflict resolution is an ugly
foot-gun. For many years, we have rejected amending during conflict
resolution in the middle of
- a merge
- a cherry-pick
However, this was never extended to other operations that can also
produce conflicts:
- an `am` operation
- a revert
- a rebase
Extend it to handle these other cases now.
Extending to `am`, revert, and the apply backend of rebase are fairly
straightforward. However, with the merge backend of rebase we have to
be more careful, since it powers interactive rebases and
- the interactive machinery internally uses `git commit --amend` for
`squash` and `reword` directives
- users are expected to `git commit --amend` after hitting an `edit`
or `break` directive
So, we need to be careful with rebase to only reject amending when doing
conflict resolution.
A few files under the rebase-merge/ directory provide us the necessary
information:
- stopped-sha is written only when the rebase stops and hands control
back to the user, so its presence marks a genuine stop -- as opposed
to the sequencer's own internal `git commit --amend` while applying
a squash, fixup, or reword, during which no stopped-sha exists.
- amend is written only when the rebase stops with HEAD already
pointing at the commit the user is meant to amend: a clean `edit`,
or a fast-forward `reword`. Its absence at a stop therefore means
the commit did not apply, so HEAD is the previously-applied commit
rather than the one being rebased -- exactly the case we refuse.
So for the merge backend we die when stopped-sha exists and amend does
not. This covers a plain conflicted pick as well as a conflicted `edit`
(both leave HEAD on the previously-applied commit), while still allowing
a clean `edit` or `reword` stop and a `break` stop (no stopped-sha).
stopped-sha is unlinked at the start of the resume loop, so a resumed
squash's internal amend is unaffected.
Signed-off-by: Elijah Newren <newren@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
|
When a rebase applies a commit that becomes empty, it stops and asks the
user to decide whether to keep it or drop it. HEAD still points at the
previously-applied commit at that point, so amending is refused, with:
You are in the middle of a rebase -- cannot amend.
That message would suggest that amending is not allowed during an 'edit'
or 'break' stop, which is misleading, plus it lacks the specificity that
might help the user know why their particular case is a problem: the
commit they intended to amend became empty and was dropped, so amending
would affect the wrong commit. Reword the error accordingly.
Signed-off-by: Elijah Newren <newren@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
|
For years, we disallowed partial commits during merges or cherry-picks.
In commit 430b75f7209c (commit: give correct advice for empty commit
during a rebase, 2019-12-06) it was noted that the "cannot do a partial
commit during a cherry-pick" message was also printed when rebasing a
commit that became empty, and rather than drop the check in that case,
that commit opted to make the message print the actual operation that
was in progress.
Since a commit that has become empty comes without conflicts, a new
partial commit poses no problems; remove the error in that case.
Signed-off-by: Elijah Newren <newren@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
|
In setup_revisions() we rewrite the incoming argv array, losing
references to the strings it contains. For a synthetic argv array
constructed from heap strings, that traditionally meant we leaked those
allocated strings.
We fixed the leak in cd43948798 (revision: manage memory ownership of
argv in setup_revisions(), 2025-09-19). Now callers can tell the
revision code that argv entries are allocated and should be freed, which
it will do before overwriting them.
But this introduced a new bug! The overwritten entries go away as soon
as option parsing is finished, but a few options may actually create new
references to those strings. And once we free the strings, those stale
references become use-after-free bugs. For example, running:
git stash show --src-prefix=foo/
demonstrates the problem:
1. The stash command generates its own synthetic argv (because it has
to treat the stash specifiers specially) which it then passes to
setup_revisions().
2. Parsing will create a reference to the partial string "foo/" in
revs.diffopt.a_prefix.
3. When setup_revisions() finishes, we rewrite argv to throw away
parsed strings. This frees the entry holding "--src-prefix=foo",
at which point we have a dangling reference in revs.diffopt.
4. We generate an actual diff, accessing garbage memory via
revs.diffopt.a_prefix. The output is usually garbled, but ASan also
detects this reliably.
One obvious fix here is to allocate new strings when we pull data out of
the argv array. But doing so is error prone (every string option must
remember to do it or risk a subtle bug), and creates more questions
about memory ownership (e.g., some callers assign string literals
directly to a_prefix, and we would not want to free those).
Instead we can fix this centrally by delaying the free() calls. We'll
collect any "freed" strings in a new array, hold on to it for the life
of the rev_info struct, and then release it at the end. We can easily
use a strvec for this, since it handles growth and cleanup for us.
This fixes the prefix case above (which is now tested in t3903), and
should fix any other stray cases. Though I could not find any; we use
OPT_STRING only in the prefix diff options, and very few revision opts
store strings. Those that do (like --format and --encoding) already make
a copy of the string. They do not need for us to hold on to the memory
longer, but it does not hurt them if we do.
One may note that combined with cd43948798 we have approached a simpler
solution in a roundabout way. We are still hacking up argv, but now
carefully constructing a parallel argv of old strings we've overwritten
(and will eventually free). In an alternate universe, we could instead
leave the original argv pristine and return a new reduced-size argv.
This is conceptually simpler, though it does mean that every caller must
free that new argv array itself (not the entries). That's not something
they traditionally had to do, so it would mean tweaking every caller.
So even though the combination of this cd43948798 and this patch is a
little convoluted, it should make things just work (no leaks and no
use-after-free) without modifying any callers.
Reported-by: Nicolas Le Cam <niko.lecam@gmail.com>
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
|
Update t/lib-httpd.sh to document the fixes applied to
apply-one-time-script.sh and http-429.sh for future developers working
on helper scripts. Add concrete examples of patterns and anti-patterns
that should be considered when handling state management.
Signed-off-by: Michael Montalbo <mmontalbo@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
|
http-429.sh is a helper for testing retry logic. It uses "test -f" to
check for the existence of a state file and later uses "touch" or
"rm -f" on that file to determine if it should return a 429. This method
of managing state can fail if the helper script is invoked concurrently.
However, this failure does not currently manifest itself since the
helper is invoked sequentially.
As a preventive measure, fix the state management logic so it relies on
an atomic mkdir operation to mark that a 429 was returned. When
$retry_after is "permanent", always return 429 now that we do not rely
on a state file that is "touch"ed and "rm"ed to indicate when to respond
with a 429.
Signed-off-by: Michael Montalbo <mmontalbo@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
|
apply-one-time-script.sh is a test helper that executes a
"one-time-script" responsible for modifying the response normally
returned by git-http-backend. apply-one-time-script.sh should run
"one-time-script" once and return a modified response once. However,
sometimes a race between multiple concurrent requests causes
apply-one-time-script.sh to misbehave and return multiple modified
responses or an empty response that results in:
fatal: ... The requested URL returned error: 500
fatal: could not fetch <oid> from promisor remote
This can be seen in the flaky failure of t5616.47 on the macOS CI
runners.
Fix the logic that checks if "one-time-script" has returned its modified
response by chaining "rm one-time-script" with its execution. This
ensures a racing script does not also have the opportunity to execute
"one-time-script".
Add t/t5567-one-time-script.sh to verify the race is fixed. Implement a
stub "git-http-backend" that intentionally invokes a concurrent request,
and check that only one modified response is returned without error.
Signed-off-by: Michael Montalbo <mmontalbo@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
|
The patch fixes two typos in two places.
versioncmp.c: "fractionnal" -> "fractional"
t/t0022-crlf-rename.sh: "similiarity" -> "similarity"
No functional changes, only update a comment and a test_description.
Signed-off-by: Hardik Kumar <hardikxk@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
|
One of the stated goals of git-replay(1) is to allow implementing the
git-rebase(1) functionality on the server side.
The default mode of git-rebase(1) is to act as if `--no-rebase-merges`
was given. This mode drops merge commits instead of replaying them, and
linearizes the history into a sequence of regular (single-parent)
commits.
Add option `--linearize` to git-replay(1) to do the same. Each replayed
commit is stacked on top of the previously replayed one. When a merge is
encountered, the commits reachable from all of its sides are replayed
into the single line and the merge itself is dropped.
If a ref was pointing to a merge commit, that ref is updated to the
merge's last replayed ancestor.
git-replay(1) accepts multiple branches, for example:
$ git replay --onto main topic1 topic2
Without `--linearize` this replays 'topic1' and 'topic2' onto 'main'
(keeping shared portions of history shared and divergent parts
divergent) and updates both refs.
Due to current implementation limitations, replaying multiple branches
with `--linearize` is disallowed to avoid concatenating unrelated
histories into a single line. For the same reason disallow the use of
`--contained` with `--linearize`.
Users who want to linearize multiple branches are advised to do this in
separate git-replay(1) invocations. Linearizing multiple branches at
once might be added later.
Note that `--linearize` is not modeled after git-rebase(1)'s
`--rebase-merges[=<mode>]` interface. Recreating merges, by preserving
their topology, is a distinct operation that would be a separate mode.
`--linearize` only drops merges and replays commits linearly. So
git-replay(1) uses its own option rather than reusing that interface.
Based-on-patches-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Toon Claes <toon@iotcl.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
|
The git worktree repair command failed to rewrite the .git file of
a working tree from a relative path to an absolute path when the
command was run in the working tree itself. The
read_gitfile_gently() function was modified to also return whether
the path originally recorded in the file was absolute, and this new
capability is used to correctly detect such mismatches.
* yn/worktree-repair-relative:
worktree repair: detect relative path in .git file correctly
|
|
The application of the edited patch in 'git add -e' has been
refactored to use the internal apply API directly, avoiding the need
to spawn a 'git apply' subprocess.
* gr/add-e-use-apply-api:
builtin/add.c: replace run_command() with direct apply_all_patches() call
|
|
The pack-objects command has been updated to record the total bytes
written to pack files in trace2 output, allowing performance
analysis of different compression settings by comparing the
resulting pack sizes.
* fr/pack-objects-trace-pack-bytes:
pack-objects: trace pack bytes written
|
|
The mechanism to generate a packfile corresponding to the result of
a fetch/push has been made pluggable through a set of object
database callback functions, removing hardcoded references to
'pack-objects' and enabling alternative ODBs to serve packfiles
themselves.
* ps/odb-pluggable-pack-generation:
bundle: generate packfiles via the object database
bundle: get (mostly) rid of `the_repository`
builtin/bundle: refactor option handling for progress meter
send-pack: generate packfiles via the object database
upload-pack: generate packfiles via the object database
odb: introduce interface to generate packfiles
|
|
The 'git receive-pack' command has been updated to use a new ODB
transaction interface for writing incoming packfiles, making it more
backend-agnostic.
* jt/receive-pack-pluggable-writes:
odb/transaction: add transaction interface to write packfiles
odb: return temporary ODB source when set
builtin/receive-pack: explicitly pass packfile fd
builtin/receive-pack: report unpack errors via strbuf
builtin/receive-pack: lift global state out of unpack()
builtin/receive-pack: read unpack limit config lazily
builtin/receive-pack: pass shallow file explicitly
odb/transaction: add transaction finalize interface
builtin/receive-pack: properly clean up keep files
|
|
The trailer parsing machinery has been updated to avoid mistaking
lines that begin with a URL (e.g., 'https://...') as trailer lines.
This prevents intended textual URLs from being mangled or mistakenly
treated as metadata keys.
* kh/trailers-no-urls:
trailers: stop recognizing URLs as trailers
|
|
The command line completion (in contrib/) has been taught to handle
the experimental 'git history' command.
* vm/complete-history:
completion: complete 'git history split' pathspecs
completion: complete 'git history --update-refs' values
completion: complete 'git history --empty' values
completion: add 'git history' subcommands
|
|
The object database (odb) API has been refactored to distinguish
between missing objects and corrupt ones by returning more
descriptive error statuses. Both the packed and loose backends now
faithfully propagate error details using a generic strbuf error
mechanism, removing backend-specific leakage from central lookup
paths.
* ps/odb-generic-corrupt-objects:
odb: handle `OBJECT_INFO_DIE_IF_CORRUPT` generically
odb/source: allow `read_object_info()` to bubble up error messages
odb/source: let callers discern missing and corrupt objects
odb/source: introduce error status when reading objects
odb/source-packed: flag known-bad objects as corrupt and not missing
|
|
The DWIM logic in 'git worktree add' sometimes tried to infer a
remote-tracking branch when an explicit '-b' or '-B' option was
given to create a new branch, causing the explicit branch name to
be ignored, which has been corrected.
* yn/worktree-add-no-dwim-with-b:
worktree add: shouldn't dwim if -b or -B is given
|
|
The instructions for deprecated commands emitted by
you_still_use_that() have been reworded to clarify that the removal
decision is final and to provide more assertive guidance on finding
a replacement.
* jc/you-still-use-that:
you_still_use_that(): reword the instructions
|
|
'git worktree add' did not prevent DWIM behavior when '-b' or '-B' was
specified, which has been corrected.
* yn/worktree-ambiguous-remote-advice:
worktree add: treat multiple matches with --guess-remote as an error
worktree add: improve message for ambiguous remote branch name
checkout: improve message for ambiguous remote branch name
checkout: extract function to display advice for ambiguous remotes
|
|
A geometric repack writes a new pack and multi-pack-index and then
deletes the packs the new one subsumes. A process still using the
previous MIDX keeps seeing a removed pack listed as the owner of some
objects. Since a MIDX attributes each object to exactly one pack, such
an object is served only through its recorded owner; if that owner was
just removed, find_pack_entry() cannot serve it -- the MIDX lookup routes
to the missing pack, and the regular pack fallback deliberately skips
every MIDX-covered pack, so a surviving copy in another covered pack
(e.g. a kept base pack) is never consulted.
Unlike the ordinary "a pack's .idx is mapped but its .pack is gone"
race, the second read does not rescue us. Reloading the on-disk pack set
does not reload the borrowed, cached MIDX (freeing it under the code that
caches the "struct multi_pack_index *" would be a use-after-free), so the
stale MIDX keeps routing to the removed pack and the surviving copy stays
hidden behind the covered-pack skip. cat-file, rev-list and pack-objects
can thus all spuriously fail with "unable to read object".
Teach find_pack_entry() to recover. The MIDX lookup now returns a
tri-state, distinguishing an object absent from the MIDX from one it owns
via a pack that can no longer be opened; in the latter case, once the
regular fallback has also missed, scan the MIDX's packs directly for a
surviving copy. Because the return value is no longer a boolean, rename
fill_midx_entry() to midx_fill_entry() so callers must reckon with the
new enum rather than silently treat MIDX_FILL_OWNER_UNAVAILABLE as a hit.
Do the scan only on the second read (OBJECT_INFO_SECOND_READ): by then
the cheaper on-disk reload has run, so an object merely relocated into a
new (uncovered) pack has already been found by the regular fallback, and
only a genuine hidden duplicate reaches the rescan. A QUICK caller that
skips the second read simply accepts the false negative, as QUICK is
designed to.
Reloading the stale MIDX would be a more complete fix but is much more
involved (the borrowers above need proper invalidation), so leave that
for later.
Assisted-by: Claude Opus 4.8 & GPT-5.6 Sol
Helped-by: Jeff King <peff@peff.net>
Signed-off-by: Elijah Newren <newren@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
|
mktree_line() checks each referenced object's type with
odb_read_object_info_extended() under OBJECT_INFO_QUICK. QUICK skips the
reprepare-and-retry that reloads the on-disk pack set, so a resident
"git mktree --batch" reader reports an object that a concurrent repack
just relocated into a new pack as missing, and rejects the entry.
QUICK entered this lookup in 817b0f602710 (mktree: do not check type of
remote objects, 2022-06-21) only to avoid lazily fetching promisor
objects; OBJECT_INFO_SKIP_FETCH_OBJECT already provides that. Drop
OBJECT_INFO_QUICK and keep OBJECT_INFO_SKIP_FETCH_OBJECT, so mktree still
avoids a promisor fetch but recovers an object that was merely repacked.
Add a regression test driving a resident mktree --batch reader across a
concurrent repack that retires a pack.
Assisted-by: Claude Opus 4.8 & GPT-5.6 Sol
Signed-off-by: Elijah Newren <newren@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
|
When objects involved in the merge cannot be read, the merge machinery
will return early with result.clean = -1, and result.tree left as NULL.
pick_regular_commit() tested only "if (!result->clean)", ignoring the
case where "clean < 0". That causes the code to try to use
result->tree, resulting in a SIGSEGV.
Handle clean < 0 explicitly; the merge machinery will already have printed
messages such as "Could not read <object>" and "collecting merge info
failed for trees...", so we don't need to add much detail beyond the
fact that the merge failed.
Signed-off-by: Elijah Newren <newren@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
|
The reftable code has been optimized to avoid an unnecessary
stat/reload of the stack when an addition already holds the
list_file lock, reducing the number of newfstatat syscalls from
linear to constant when writing refs.
* kn/reftable-optimize-reloading:
reftable/stack: avoid reloading the stack when already locked
reftable/stack: move list lock to `struct reftable_stack`
reftable/stack: rename reftable_stack_new_addition()
reftable/stack: remove `REFTABLE_STACK_NEW_ADDITION_RELOAD`
|
|
A heap-use-after-free bug in the object name parsing code when
reporting failures with a relative path to a sparse directory has
been corrected.
* sk/object-name-use-after-free:
object-name: avoid use-after-free in get_oid_with_context_1()
|
