In 6aec8d38fd (t: refactor tests depending on Perl to print data,
2025-04-03) we have changed some of the tests in t4150 to use sed(1)
instead of Perl. One of the conversions is broken though:
sed: -e expression #1, char 41: unterminated `s' command
Curiously enough, the test itself still passes. This is caused by a
sequence of failures:
1. The output of sed(1) is piped into git-update-ref(1), and because
sed(1) is the upstream command we don't notice that it fails.
2. git-update-ref(1) does not receive any input and thus won't create
any references.
3. We then repack the repository with the configured pseudo merges
pattern, but as we didn't create any references the pattern doesn't
match anything.
4. We use `test_pseudo_merges()` to compute the list of pseudo-merges
and write it into a file. This file is empty as there are none.
5. The loop over the pseudo-merges becomes a no-op.
6. The final test succeeds as well because the number of lines in an
empty file is obviously the same as the number of unique lines,
namely zero.
Fix the issue by adding the terminating '|' to the sed(1) command.
Furthermore, make the test a tiny bit more robust by not using it as
part of a pipe.
Reported-by: SZEDER Gábor <szeder.dev@gmail.com>
Signed-off-by: Patrick Steinhardt <ps@pks.im>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
In 6aec8d38fd (t: refactor tests depending on Perl to print data,
2025-04-03) we have changed one of the tests in t4150 to use awk(1)
instead of Perl. The test works, but at least gawk(1) prints a warning
now:
awk: cmd. line:3: warning: escape sequence `\@' treated as plain `@'
Fix this by removing the backslash.
Reported-by: SZEDER Gábor <szeder.dev@gmail.com>
Signed-off-by: Patrick Steinhardt <ps@pks.im>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Tests that compare $HOME and $(pwd), which should be the same
directory unless the tests chdir's around, would fail when the user
enters the test directory via symbolic links, which has been
corrected.
* mm/test-in-absolute-home:
t: run tests from a normalized working directory
In the GitHub workflow used in Git's CI builds, the `vs test` jobs use a
subset of a specific revision of Git for Windows' SDK to run Git's test
suite. This revision is validated by another CI workflow to ensure that
said revision _can_ run Git's test suite successfully, skipping buggy
updates in Git for Windows' SDK.
The `win+Meson test` jobs do things differently, quite differently. They
use the Bash of the Git for Windows version that is installed on the
runners to run Git's test suite.
This difference has consequences.
When 68cb0b5253 (builtin/receive-pack: add option to skip connectivity
check, 2025-05-20) introduced a test case that uses `tee <file> | git
receive-pack` as `--receive-pack` parameter (imitating an existing
pattern in the same test script), it hit just the sweet spot to trigger
a bug in the MSYS2 runtime shipped in Git for Windows v2.49.0. This
version is the one currently installed on GitHub's runners.
The problem is that the `git receive-pack` process finishes while the
`tee` process does not need to write anything anymore and therefore does
not receive an EOF. Instead, it should receive a SIGPIPE, but the bug in
the MSYS2 runtime prevents that from working as intended. As a
consequence, the `tee` process waits for more input from the `git.exe
send-pack` process but none is coming, and the test script patiently
waits until the 6h timeout hits.
Only every once in a while, the `git receive-pack` process manages to
send an EOF to the `tee` process and no hang occurs. Therefore, the
problem can be worked around by cancelling the clearly-hanging job after
twenty or so minutes and re-running it, repeating the process about half
a dozen times, until the hang was successfully avoided.
This bug in the MSYS2 runtime has been fixed in the meantime, which is
the reason why the same test case causes no problems in the `win test`
and the `vs test` jobs.
This will continue to be the case until the Git for Windows version on
the GitHub runners is upgraded to a version that distributes a newer
MSYS2 runtime version. However, as of time of writing, this _is_ the
latest Git for Windows version, and will be for another 1.5 weeks, until
Git v2.50.0 is scheduled to appear (and shortly thereafter Git for
Windows v2.50.0). Traditionally it takes a while before the runners pick
up the new version.
We could just wait it out, six hours at a time.
Here, I opt for an alternative: Detect the buggy MSYS2 runtime and
simply skip the test case. It's not like the `receive-pack` test cases
are specific to Windows, and even then, to my chagrin the CI runs in
git-for-windows/git spend around ten hours of compute time each and
every time to run the entire test suite on all the platforms, even the
tests that cover cross-platform code, and for Windows alone we do that
three times: with GCC, with MSVC, and with MSVC via Meson. Therefore, I
deem it more than acceptable to skip this test case in one of those
matrices.
For good luck, also the preceding test case is skipped in that scenario,
as it uses the same `--receive-pack=tee <file> | git receive-pack`
pattern, even though I never observed that test case to hang in
practice.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
"git verify-refs" (and hence "git fsck --reference") started
erroring out in a repository in which secondary worktrees were
prepared with Git 2.43 or lower.
* sj/ref-contents-check-fix:
fsck: ignore missing "refs" directory for linked worktrees
"git refs verify" doesn't work if there are worktrees created on Git
v2.43.0 or older versions. These versions don't automatically create the
"refs" directory, causing the error:
error: cannot open directory .git/worktrees/<worktree name>/refs:
No such file or directory
Since 8f4c00de95 (builtin/worktree: create refdb via ref backend,
2024-01-08), we automatically create the "refs" directory for new
worktrees. And in 7c78d819e6 (ref: support multiple worktrees check for
refs, 2024-11-20), we assume that all linked worktrees have this
directory and would wrongly report an error to the user, thus
introducing compatibility issue.
Check for ENOENT errno before reporting directory access errors for
linked worktrees to maintain backward compatibility.
Reported-by: Kristoffer Haugsbakk <code@khaugsbakk.name>
Signed-off-by: shejialuo <shejialuo@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Mark a new feature added during this cycle as experimental and fix
its default so that existing users of the fast-export command is
not broken.
* jc/signed-fast-export-is-experimental:
fast-export: --signed-commits is experimental
Some tests make git perform actions that produce observable pathnames,
and have expectations on those paths. Tests run with $HOME set to a
$TRASH_DIRECTORY, and with their working directory the same
$TRASH_DIRECTORY, although these paths are logically identical, they do
not observe the same pathname canonicalization rules and thus might not
be represented by strings that compare equal. In particular, no pathname
normalization is applied to $TRASH_DIRECTORY or $HOME, while tests
change their working directory with `cd -P`, which normalizes the
working directory's path by fully resolving symbolic links.
t7900's macOS maintenance tests (which are not limited to running on
macOS) have an expectation on a path that `git maintenance` forms by
using abspath.c strbuf_realpath() to resolve a canonical absolute path
based on $HOME. When t7900 runs from a working directory that contains
symbolic links in its pathname, $HOME will also contain symbolic links,
which `git maintenance` resolves but the test's expectation does not,
causing a test failure.
Align $TRASH_DIRECTORY and $HOME with the normalized path as used for
the working directory by resetting them to match the working directory
after it's established by `cd -P`. With all paths in agreement and
symbolic links resolved, pathname expectations can be set and met based
on string comparison without regard to external environmental factors
such as the presence of symbolic links in a path.
Suggested-by: Junio C Hamano <gitster@pobox.com>
Signed-off-by: Mark Mentovai <mark@chromium.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
"git apply --index/--cached" when applying a deletion patch in
reverse failed to give the mode bits of the path "removed" by the
patch to the file it creates, which has been corrected.
* mm/apply-reverse-mode-of-deleted-path:
apply: set file mode when --reverse creates a deleted file
t4129: test that git apply warns for unexpected mode changes
As the design of signature handling is still being discussed, it is
likely that the data stream produced by the code in Git 2.50 would
have to be changed in such a way that is not backward compatible.
Mark the feature as experimental and discourge its use for now.
Also flip the default on the generation side to "strip"; users of
existing versions would not have passed --signed-commits=strip and
will be broken by this change if the default is made to abort, and
will be encouraged by the error message to produce data stream with
future breakage guarantees by passing --signed-commits option.
As we tone down the default behaviour, we no longer need the
FAST_EXPORT_SIGNED_COMMITS_NOABORT environment variable, which was
not discoverable enough.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
"git receive-pack" optionally learns not to care about connectivity
check, which can be useful when the repository arranges to ensure
connectivity by some other means.
* jt/receive-pack-skip-connectivity-check:
builtin/receive-pack: add option to skip connectivity check
t5410: test receive-pack connectivity check
Remove the leftover hints to the test framework to mark tests that
do not pass the leak checker tests, as they should no longer be
needed.
* kn/passing-leak-tests:
t: remove unexpected SANITIZE_LEAK variables
Prefix '#' to the commit title in the "rebase -i" todo file, just
like a merge commit being replayed.
* en/sequencer-comment-messages:
sequencer: make it clearer that commit descriptions are just comments
The code path to access the "packed-refs" file while "fsck" is
taught to mmap the file, instead of reading the whole file in the
memory.
* sj/use-mmap-to-check-packed-refs:
packed-backend: mmap large "packed-refs" file during fsck
packed-backend: extract snapshot allocation in `load_contents`
packed-backend: fsck should warn when "packed-refs" file is empty
"git apply" and "git add -i/-p" code paths no longer unnecessarily
expand sparse-index while working.
* ds/sparse-apply-add-p:
p2000: add performance test for patch-mode commands
reset: integrate sparse index with --patch
git add: make -p/-i aware of sparse index
apply: integrate with the sparse index
"git merge-tree" learned an option to see if it resolves cleanly
without actually creating a result.
* en/merge-tree-check:
merge-tree: add a new --quiet flag
merge-ort: add a new mergeability_only option
Support to create a loose object file with unknown object type has
been dropped.
* jk/no-funny-object-types:
object-file: drop support for writing objects with unknown types
hash-object: handle --literally with OPT_NEGBIT
hash-object: merge HASH_* and INDEX_* flags
hash-object: stop allowing unknown types
t: add lib-loose.sh
t/helper: add zlib test-tool
oid_object_info(): drop type_name strbuf
fsck: stop using object_info->type_name strbuf
oid_object_info_convert(): stop using string for object type
cat-file: use type enum instead of buffer for -t option
object-file: drop OBJECT_INFO_ALLOW_UNKNOWN_TYPE flag
cat-file: make --allow-unknown-type a noop
object-file.h: fix typo in variable declaration
The userdiff pattern for shell scripts has been updated to cope
with more bash-isms.
* md/userdiff-bash-shell-function:
userdiff: extend Bash pattern to cover more shell function forms
Commit 01aff0a (apply: correctly reverse patch's pre- and post-image
mode bits, 2023-12-26) revised reverse_patches() to maintain the desired
property that when only one of patch::old_mode and patch::new_mode is
set, the mode will be carried in old_mode. That property is generally
correct, with one notable exception: when creating a file, only new_mode
will be set. Since reversing a deletion results in a creation, new_mode
must be set in that case.
Omitting handling for this case means that reversing a patch that
removes an executable file will not result in the executable permission
being set on the re-created file. Existing test coverage for file modes
focuses only on mode changes of existing files.
Swap old_mode and new_mode in reverse_patches() for what's represented
in the patch as a file deletion, as it is transformed into a file
creation under reversal. This causes git apply --reverse to set the
executable permission properly when re-creating a deleted executable
file.
Add tests ensuring that git apply sets file modes correctly on file
creation, both in the forward and reverse directions.
Signed-off-by: Mark Mentovai <mark@chromium.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
There is no test covering what commit 01aff0a (apply: correctly reverse
patch's pre- and post-image mode bits, 2023-12-26) addressed. Prior to
that commit, git apply was erroneously unaware of a file's expected mode
while reverse-patching a file whose mode was not changing.
Add the missing test coverage to assure that git apply is aware of the
expected mode of a file being patched when the patch does not indicate
that the file's mode is changing. This is achieved by arranging a file
mode so that it doesn't agree with patch being applied, and checking git
apply's output for the warning it's supposed to raise in this situation.
Test in both reverse and normal (forward) directions.
Signed-off-by: Mark Mentovai <mark@chromium.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Two "scalar" subcommands that adds a repository that hasn't been
under "scalar"'s control are taught an option not to enable the
scheduled maintenance on it.
* ds/scalar-no-maintenance:
scalar reconfigure: improve --maintenance docs
scalar reconfigure: add --maintenance=<mode> option
scalar clone: add --no-maintenance option
scalar register: add --no-maintenance option
scalar: customize register_dir()'s behavior
As of 1fc7ddf35b (test-lib: unconditionally enable leak checking,
2024-11-20), both the `GIT_TEST_PASSING_SANITIZE_LEAK` and
`TEST_PASSES_SANITIZE_LEAK` variables no longer have any meaning, the
leak checks are enabled by default. However, some newly added tests
include them by mistake. Let's clean this up.
Signed-off-by: Karthik Nayak <karthik.188@gmail.com>
Acked-by: Justin Tobler <jltobler@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
During git-receive-pack(1), connectivity of the object graph is
validated to ensure that the received packfile does not leave the
repository in a broken state. This is done via git-rev-list(1) and
walking the objects, which can be expensive for large repositories.
Generally, this check is critical to avoid an incomplete received
packfile from corrupting a repository. Server operators may have
additional knowledge though around exactly how Git is being used on the
server-side which can be used to facilitate more efficient connectivity
computation of incoming objects.
For example, if it can be ensured that all objects in a repository are
connected and do not depend on any missing objects, the connectivity of
newly written objects can be checked by walking the object graph
containing only the new objects from the updated tips and identifying
the missing objects which represent the boundary between the new objects
and the repository. These boundary objects can be checked in the
canonical repository to ensure the new objects connect as expected and
thus avoid walking the rest of the object graph.
Git itself cannot make the guarantees required for such an optimization
as it is possible for a repository to contain an unreachable object that
references a missing object without the repository being considered
corrupt.
Introduce the --skip-connectivity-check option for git-receive-pack(1)
which bypasses this connectivity check to give more control to the
server-side. Note that without proper server-side validation of newly
received objects handled outside of Git, usage of this option risks
corrupting a repository.
Signed-off-by: Justin Tobler <jltobler@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
As part of git-recieve-pack(1), the connectivity of objects is checked.
Add a test validating that git-receive-pack(1) fails due to an incoming
packfile that would leave the repository with missing objects. Instead
of creating a new test file, "t5410" is generalized for receive-pack
testing.
Signed-off-by: Justin Tobler <jltobler@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Performance regression in not-yet-released code has been corrected.
* ps/reftable-read-block-perffix:
reftable: fix perf regression when reading blocks of unwanted type
Code cleanup.
* jk/oidmap-cleanup:
raw_object_store: drop extra pointer to replace_map
oidmap: add size function
oidmap: rename oidmap_free() to oidmap_clear()
Bundle-URI feature did not use refs recorded in the bundle other
than normal branches as anchoring points to optimize the follow-up
fetch during "git clone"; now it is told to utilize all.
* sc/bundle-uri-use-all-refs-in-bundle:
bundle-uri: add test for bundle-uri clones with tags
bundle-uri: copy all bundle references ino the refs/bundle space
Git Forges may be interested in whether two branches can be merged while
not being interested in what the resulting merge tree is nor which files
conflicted. For such cases, add a new --quiet flag which
will make use of the new mergeability_only flag added to merge-ort in
the previous commit. This option allows the merge machinery to, in the
outer layer of the merge:
* exit early when a conflict is detected
* avoid writing (most) merged blobs/trees to the object store
Signed-off-by: Elijah Newren <newren@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Every once in a while, users report that editing the commit summaries
in the todo list does not get reflected in the rebase operation,
suggesting that users are (a) only using one-line commit messages, and
(b) not understanding that the commit summaries are merely helpful
comments to help them find the right hashes.
It may be difficult to correct users' poor commit messages, but we can
at least try to make it clearer that the commit summaries are not
directives of some sort by inserting a comment character. Hopefully
that leads to them looking a little further and noticing the hints at
the bottom to use 'reword' or 'edit' directives.
Yes, this change may look funny at first since it hardcodes '#' rather
than using comment_line_str. However:
* comment_line_str exists to allow disambiguation between lines in
a commit message and lines that are instructions to users editing
the commit message. No such disambiguation is needed for these
comments that occur on the same line after existing directives
* the exact "comment" character(s) on regular pick lines used aren't
actually important; I could have used anything, including completely
random variable length text for each line and it'd work because we
ignore everything after 'pick' and the hash.
* The whole point of this change is to signal to users that they
should NOT be editing any part of the line after the hash (and if
they do so, their edits will be ignored), while the whole point of
comment_line_str is to allow highly flexible editing. So making
it more general by using comment_line_str actually feels
counterproductive.
* The character for merge directives absolutely must be '#'; that
has been deeply hardcoded for a long time (see below), and will
break if some other comment character is used instead. In a
desire to have pick and merge directives be similar, I use the
same comment character for both.
* Perhaps merge directives could be fixed to not be inflexible about
the comment character used, if someone feels highly motivated, but
I think that should be done in a separate follow-on patch.
Here are (some of?) the locations where '#' has already been hardcoded
for a long time for merges:
1) In check_label_or_ref_arg():
case TODO_LABEL:
/*
* '#' is not a valid label as the merge command uses it to
* separate merge parents from the commit subject.
*/
2) In do_merge():
/*
* For octopus merges, the arg starts with the list of revisions to be
* merged. The list is optionally followed by '#' and the oneline.
*/
merge_arg_len = oneline_offset = arg_len;
for (p = arg; p - arg < arg_len; p += strspn(p, " \t\n")) {
if (!*p)
break;
if (*p == '#' && (!p[1] || isspace(p[1]))) {
3) In label_oid():
if ((buf->len == the_hash_algo->hexsz &&
!get_oid_hex(label, &dummy)) ||
(buf->len == 1 && *label == '#') ||
hashmap_get_from_hash(&state->labels,
strihash(label), label)) {
/*
* If the label already exists, or if the label is a
* valid full OID, or the label is a '#' (which we use
* as a separator between merge heads and oneline), we
* append a dash and a number to make it unique.
*/
Signed-off-by: Elijah Newren <newren@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The previous three changes contributed performance improvements to 'git
apply', 'git add -p', and 'git reset -p' when using a sparse index. The
improvement to 'git apply' also improved 'git checkout -p'. Add
performance tests to demonstrate this (and to help validate that
performance remains good in the future).
In the truncated test output below, we see that the full checkout
performance changes within noise expectations, but the sparse index
cases improve 33% and then 96% for 'git add -p' and 41% and then 95% for
'git reset -p'. 'git checkout -p' improves immediatley by 91% because it
does not need any change to its builtin.
Test HEAD~4 HEAD~3 HEAD~2 HEAD~1
-------------------------------------------------------------------------------------
2000.118: ... git add -p (full-v3) 0.79 0.79 +0.0% 0.82 +3.8% 0.82 +3.8%
2000.119: ... git add -p (full-v4) 0.74 0.76 +2.7% 0.74 +0.0% 0.76 +2.7%
2000.120: ... git add -p (sparse-v3) 1.94 1.28 -34.0% 0.07 -96.4% 0.07 -96.4%
2000.121: ... git add -p (sparse-v4) 1.93 1.28 -33.7% 0.06 -96.9% 0.06 -96.9%
2000.122: ... git checkout -p (full-v3) 1.18 1.18 +0.0% 1.18 +0.0% 1.19 +0.8%
2000.123: ... git checkout -p (full-v4) 1.10 1.12 +1.8% 1.11 +0.9% 1.11 +0.9%
2000.124: ... git checkout -p (sparse-v3) 1.31 0.11 -91.6% 0.11 -91.6% 0.11 -91.6%
2000.125: ... git checkout -p (sparse-v4) 1.29 0.11 -91.5% 0.11 -91.5% 0.11 -91.5%
2000.126: ... git reset -p (full-v3) 0.81 0.80 -1.2% 0.83 +2.5% 0.83 +2.5%
2000.127: ... git reset -p (full-v4) 0.78 0.77 -1.3% 0.77 -1.3% 0.78 +0.0%
2000.128: ... git reset -p (sparse-v3) 1.58 0.92 -41.8% 0.91 -42.4% 0.07 -95.6%
2000.129: ... git reset -p (sparse-v4) 1.58 0.92 -41.8% 0.92 -41.8% 0.07 -95.6%
It is worth noting that if our test was more involved and had multiple
hunks to evaluate, then the time spent in 'git apply' would dominate due
to multiple index loads and writes. As it stands, we need the sparse
index improvement in 'git add -p' itself to confirm this performance
improvement.
Since the change for 'git add -i' is identical, we avoid a second test
case for that similar operation.
Signed-off-by: Derrick Stolee <stolee@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Similar to the previous change for 'git add -p', the reset builtin
checked for integration with the sparse index after possibly redirecting
its logic toward the interactive logic. This means that the builtin
would expand the sparse index to a full one upon read.
Move this check earlier within cmd_reset() to improve performance here.
Add tests to guarantee that we are not universally expanding the index.
Add behavior tests to check that we are doing the same operations as a
full index.
Signed-off-by: Derrick Stolee <stolee@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
It is slow to expand a sparse index in-memory due to parsing of trees.
We aim to minimize that performance cost when possible. 'git add -p'
uses 'git apply' child processes to modify the index, but still there
are some expansions that occur.
It turns out that control flows out of cmd_add() in the interactive
cases before the lines that confirm that the builtin is integrated with
the sparse index.
Moving that integration point earlier in cmd_add() allows 'git add -i'
and 'git add -p' to operate without expanding a sparse index to a full
one.
Add test cases that confirm that these interactive add options work with
the sparse index.
Signed-off-by: Derrick Stolee <stolee@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The sparse index allows storing directory entries in the index, marked
with the skip-wortkree bit and pointing to a tree object. This may be an
unexpected data shape for some implementation areas, so we are rolling
it out incrementally on a builtin-per-builtin basis.
This change enables the sparse index for 'git apply'. The main
motivation for this change is that 'git apply' is used as a child
process of 'git add -p' and expanding the sparse index for each of those
child processes can lead to significant performance issues.
The good news is that the actual index manipulation code used by 'git
apply' is already integrated with the sparse index, so the only product
change is to mark the builtin as allowing the sparse index so it isn't
inflated on read.
The more involved part of this change is around adding tests that verify
how 'git apply' behaves in a sparse-checkout environment and whether or
not the index expands in certain operations.
Signed-off-by: Derrick Stolee <stolee@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The previous function regex required explicit matching of function
bodies using `{`, `(`, `((`, or `[[`, which caused several issues:
- It failed to capture valid functions where `{` was on the next line
due to line continuation (`\`).
- It did not recognize functions with single command body, such as
`x () echo hello`.
Replacing the function body matching logic with `.*$`, ensures
that everything on the function definition line is captured.
Additionally, the word regex is refined to better recognize shell
syntax, including additional parameter expansion operators and
command-line options.
Signed-off-by: Moumita Dhar <dhar61595@gmail.com>
Acked-by: Johannes Sixt <j6t@kdbg.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When passed the "--literally" option, hash-object will allow any
arbitrary string for its "-t" type option. Such objects are only useful
for testing or debugging, as they cannot be used in the normal way
(e.g., you cannot fetch their contents!).
Let's drop this feature, which will eventually let us simplify the
object-writing code. This is technically backwards incompatible, but
since such objects were never really functional, it seems unlikely that
anybody will notice.
We will retain the --literally flag, as it also instructs hash-object
not to worry about other format issues (e.g., type-specific things that
fsck would complain about). The documentation does not need to be
updated, as it was always vague about which checks we're loosening (it
uses only the phrase "any garbage").
The code change is a bit hard to verify from just the patch text. We can
drop our local hash_literally() helper, but it was really just wrapping
write_object_file_literally(). We now replace that with calling
index_fd(), as we do for the non-literal code path, but dropping the
INDEX_FORMAT_CHECK flag. This ends up being the same semantically as
what the _literally() code path was doing (modulo handling unknown
types, which is our goal).
We'll be able to clean up these code paths a bit more in subsequent
patches.
The existing test is flipped to show that we now reject the unknown
type. The additional "extra-long type" test is now redundant, as we bail
early upon seeing a bogus type.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
This commit adds a shell library for writing raw loose objects into the
object database. Normally this is done with hash-object, but the
specific intent here is to allow broken objects that hash-object may not
support.
We'll convert several cases that use "hash-object --literally" to write
objects with invalid types. That works currently, but dropping this
dependency will allow us to remove that feature and simplify the
object-writing code.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
It's occasionally useful when testing or debugging to be able to do raw
zlib inflate/deflate operations (e.g., to check the bytes of a specific
loose or packed object).
Even though zlib's deflate algorithm is used by many other programs,
this is surprisingly hard to do in a portable way. E.g., gzip can do
this if you manually munge some header bytes. But the result is somewhat
arcane, and we don't assume gzip is available anyway. Likewise, pigz
will handle raw zlib, but we can't assume it is available.
So let's introduce a short test helper for just doing zlib operations.
We'll use it in subsequent patches to add some new tests, but it would
also have come in handy a few times in the past:
- The hard-coded pack data from 3b910d0c5e (add tests for indexing
packs with delta cycles, 2013-08-23) could probably be generated on
the fly.
- Likewise we could avoid the hard-coded data from 0b1493c2d4
(git_inflate(): skip zlib_post_call() sanity check on Z_NEED_DICT,
2025-02-25). Though note this would require support for more zlib
options.
- It would have helped with the debugging documented in 41dfbb2dbe
(howto: add article on recovering a corrupted object, 2013-10-25).
I'll leave refactoring existing tests for another day, but I hope the
examples above show the general utility.
I aimed for simplicity in the code. In particular, it will read all
input into a memory buffer, rather than streaming. That makes the zlib
loops harder to get wrong (which has been a source of subtle bugs in the
past).
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When fsck-ing a loose object, we use object_info's type_name strbuf to
record the parsed object type as a string. For most objects this is
redundant with the object_type enum, but it does let us report the
string when we encounter an object with an unknown type (for which there
is no matching enum value).
There are a few downsides, though:
1. The code to report these cases is not actually robust. Since we did
not pass a strbuf to unpack_loose_header(), we only retrieved types
from headers up to 32 bytes. In longer cases, we'd simply say
"object corrupt or missing".
2. This is the last caller that uses object_info's type_name strbuf
support. It would be nice to refactor it so that we can simplify
that code.
3. Likewise, we'll check the hash of the object using its unknown type
(again, as long as that type is short enough). That depends on the
hash_object_file_literally() code, which we'd eventually like to
get rid of.
So we can simplify things by bailing immediately in read_loose_object()
when we encounter an unknown type. This has a few user-visible effects:
a. Instead of producing a single line of error output like this:
error: 26ed13ce3564fbbb44e35bde42c7da717ea004a6: object is of unknown type 'bogus': .git/objects/26/ed13ce3564fbbb44e35bde42c7da717ea004a6
we'll now issue two lines (the first from read_loose_object() when
we see the unparsable header, and the second from the fsck code,
since we couldn't read the object):
error: unable to parse type from header 'bogus 4' of .git/objects/26/ed13ce3564fbbb44e35bde42c7da717ea004a6
error: 26ed13ce3564fbbb44e35bde42c7da717ea004a6: object corrupt or missing: .git/objects/26/ed13ce3564fbbb44e35bde42c7da717ea004a6
This is a little more verbose, but this sort of error should be
rare (such objects are almost impossible to work with, and cannot
be transferred between repositories as they are not representable
in packfiles). And as a bonus, reporting the broken header in full
could help with debugging other cases (e.g., a header like "blob
xyzzy\0" would fail in parsing the size, but previously we'd not
have showed the offending bytes).
b. An object with an unknown type will be reported as corrupt, without
actually doing a hash check. Again, I think this is unlikely to
matter in practice since such objects are totally unusable.
We'll update one fsck test to match the new error strings. And we can
remove another test that covered the case of an object with an unknown
type _and_ a hash corruption. Since we'll skip the hash check now in
this case, the test is no longer interesting.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The cat-file command has some minor support for handling objects with
"unknown" types. I.e., strings that are not "blob", "commit", "tree", or
"tag".
In theory this could be used for debugging or experimenting with
extensions to Git. But in practice this support is not very useful:
1. You can get the type and size of such objects, but nothing else.
Not even the contents!
2. Only loose objects are supported, since packfiles use numeric ids
for the types, rather than strings.
3. Likewise you cannot ever transfer objects between repositories,
because they cannot be represented in the packfiles used for the
on-the-wire protocol.
The support for these unknown types complicates the object-parsing code,
and has led to bugs such as b748ddb7a4 (unpack_loose_header(): fix
infinite loop on broken zlib input, 2025-02-25). So let's drop it.
The first step is to remove the user-facing parts, which are accessible
only via cat-file. This is technically backwards-incompatible, but given
the limitations listed above, these objects couldn't possibly be useful
in any workflow.
However, we can't just rip out the option entirely. That would hurt a
caller who ran:
git cat-file -t --allow-unknown-object <oid>
and fed it normal, well-formed objects. There --allow-unknown-type was
doing nothing, but we wouldn't want to start bailing with an error. So
to protect any such callers, we'll retain --allow-unknown-type as a
noop.
The code change is fairly small (but we'll able to clean up more code in
follow-on patches). The test updates drop any use of the option. We
still retain tests that feed the broken objects to cat-file without
--allow-unknown-type, as we should continue to confirm that those
objects are rejected. Note that in one spot we can drop a layer of loop,
re-indenting the body; viewing the diff with "-w" helps there.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Make repository clean-up tasks "gc" can do available to "git
maintenance" front-end.
* ps/maintenance-missing-tasks:
builtin/maintenance: introduce "rerere-gc" task
builtin/gc: move rerere garbage collection into separate function
builtin/maintenance: introduce "worktree-prune" task
builtin/gc: move pruning of worktrees into a separate function
builtin/gc: remove global variables where it is trivial to do
builtin/gc: fix indentation of `cmd_gc()` parameters
We assume the "packed-refs" won't be empty and instead has at least one
line in it (even when there are no refs packed, there is the file header
line). Because there is no terminating LF in the empty file, we will
report "packedRefEntryNotTerminated(ERROR)" to the user.
However, the runtime code paths would accept an empty "packed-refs"
file, for example, "create_snapshot" would simply return the "snapshot"
without checking the content of "packed-refs". So, we should skip
checking the content of "packed-refs" when it is empty during fsck.
After 694b7a1999 (repack_without_ref(): write peeled refs in the
rewritten file, 2013-04-22), we would always write a header into the
"packed-refs" file. So, versions of Git that are not too ancient never
write such an empty "packed-refs" file.
As an empty file often indicates a sign of a filesystem-level issue, the
way we want to resolve this inconsistency is not make everybody totally
silent but notice and report the anomaly.
Let's create a "FSCK_INFO" message id "EMPTY_PACKED_REFS_FILE" to report
to the users that "packed-refs" is empty.
Signed-off-by: shejialuo <shejialuo@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
"git add 'f?o'" did not add 'foo' if 'f?o', an unusual pathname,
also existed on the working tree, which has been corrected.
* kj/glob-path-with-special-char:
dir.c: literal match with wildcard in pathspec should still glob
"git diff --minimal" used to give non-minimal output when its
optimization kicked in, which has been disabled.
* ng/xdiff-truly-minimal:
xdiff: disable cleanup_records heuristic with --minimal
"git index-pack --fix-thin" used to abort to prevent a cycle in
delta chains from forming in a corner case even when there is no
such cycle.
* ds/fix-thin-fix:
index-pack: allow revisiting REF_DELTA chains
t5309: create failing test for 'git index-pack'
test-tool: add pack-deltas helper
This function does not free the oidmap struct itself; it just drops all
items from the map (using hashmap_clear_() internally). It should be
called oidmap_clear(), per CodingGuidelines.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
In fd888311fb (reftable/table: move reading block into block reader,
2025-04-07), we have refactored how reftable blocks are read so that
most of the logic is contained in the "block.c" subsystem itself. Most
importantly, the whole logic to read the data itself is now contained in
that subsystem.
This change caused a significant performance regression though when
reading blocks that aren't of the specific type one is searching for:
Benchmark 1: update-ref: create 100k refs (revision = fd888311fbc~)
Time (mean ± σ): 2.171 s ± 0.028 s [User: 1.189 s, System: 0.977 s]
Range (min … max): 2.117 s … 2.206 s 10 runs
Benchmark 2: update-ref: create 100k refs (revision = fd888311fb)
Time (mean ± σ): 3.418 s ± 0.030 s [User: 2.371 s, System: 1.037 s]
Range (min … max): 3.377 s … 3.473 s 10 runs
Summary
update-ref: create 100k refs (revision = fd888311fbc~) ran
1.57 ± 0.02 times faster than update-ref: create 100k refs (revision = fd888311fb)
The root caute of the performance regression is that we changed when
exactly blocks of an uninteresting type are being discarded. Previous to
the refactoring in the mentioned commit we'd load the block data, read
its type, notice that it's not the wanted type and discard the block.
After the commit though we don't discard the block immediately, but we
fully decode it only to realize that it's not the desired type. We then
discard the block again, but have already performed a bunch of pointless
work.
Fix the regression by making `reftable_block_init()` return early in
case the block is not of the desired type. This fixes the performance
hit:
Benchmark 1: update-ref: create 100k refs (revision = HEAD~)
Time (mean ± σ): 2.712 s ± 0.018 s [User: 1.990 s, System: 0.716 s]
Range (min … max): 2.682 s … 2.741 s 10 runs
Benchmark 2: update-ref: create 100k refs (revision = HEAD)
Time (mean ± σ): 1.670 s ± 0.012 s [User: 0.991 s, System: 0.676 s]
Range (min … max): 1.652 s … 1.693 s 10 runs
Summary
update-ref: create 100k refs (revision = HEAD) ran
1.62 ± 0.02 times faster than update-ref: create 100k refs (revision = HEAD~)
Note that the baseline performance is lower than in the original due to
a couple of unrelated performance improvements that have landed since
the original commit.
Signed-off-by: Patrick Steinhardt <ps@pks.im>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
