| Age | Commit message (Collapse) | Author | Files | Lines |
|---|
|
The iconv library is used by Git to reencode files, commit messages and
other things. As such it is a rather integral part, but given that many
platforms nowadays use UTF-8 everywhere you can live without support for
reencoding in many situations. It is thus optional to build Git with
iconv, and some of our platforms wired up in "config.mak.uname" disable
it. But while we support building without it, running our test suite
with "NO_ICONV=Yes" causes many test failures.
Wire up a new test prerequisite ICONV that gets populated via our
GIT-BUILD-OPTIONS. Annotate failing tests accordingly.
Note that this commit does not do a deep dive into every single test to
assess whether the failure is expected or not. Most of the tests do
smell like the expected kind of failure though.
Signed-off-by: Patrick Steinhardt <ps@pks.im>
Signed-off-by: Taylor Blau <me@ttaylorr.com>
|
|
When assembling our LSAN_OPTIONS that configure the leak sanitizer we
end up prepending the string with various different colon-separated
options via calls to `prepend_var`. One of the settings we add is the
path where the sanitizer should store logs, which can be an arbitrary
filesystem path.
Naturally, filesystem paths may contain whitespace characters. And while
it does seem as if we were quoting the value, we use escaped quotes and
consequently split up the value if it does contain spaces. This leads to
the following error in t0000 when having a value with whitespaces:
.../t/test-lib.sh: eval: line 64: unexpected EOF while looking for matching `"'
++ return 1
error: last command exited with $?=1
not ok 5 - subtest: 3 passing tests
The error itself is a bit puzzling at first. The basic problem is that
the code sees the leading escaped quote during eval, but because we
truncate everything after the space character it doesn't see the
trailing escaped quote and thus fails to parse the string.
Properly quote the value to fix the issue while using single-quotes to
quote the inner value passed to eval. The issue can be reproduced by
t0000 with such a path that contains spaces.
Signed-off-by: Patrick Steinhardt <ps@pks.im>
Signed-off-by: Taylor Blau <me@ttaylorr.com>
|
|
Signed-off-by: Taylor Blau <me@ttaylorr.com>
|
|
On macOS, fsmonitor can fall into a race condition that results in
a client waiting forever to be notified for an event that have
already happened. This problem has been corrected.
* jk/fsmonitor-event-listener-race-fix:
fsmonitor: initialize fs event listener before accepting clients
simple-ipc: split async server initialization and running
|
|
A new configuration variable remote.<name>.serverOption makes the
transport layer act as if the --serverOption=<value> option is
given from the command line.
* xx/remote-server-option-config:
ls-remote: leakfix for not clearing server_options
fetch: respect --server-option when fetching multiple remotes
transport.c::handshake: make use of server options from remote
remote: introduce remote.<name>.serverOption configuration
transport: introduce parse_transport_option() method
|
|
Docfix.
* js/doc-platform-support-link-fix:
docs: fix the `maintain-git` links in `technical/platform-support`
|
|
Docfix.
* jh/config-unset-doc-fix:
git-config.1: remove value from positional args in unset usage
|
|
Deprecate the "trailer_info" struct name and replace it with
"trailer_block". This is more readable, for two reasons:
1. "trailer_info" on the surface sounds like it's about a single
trailer when in reality it is a collection of one or more trailers,
and
2. the "*_block" suffix is more informative than "*_info", because it
describes a block (or region) of contiguous text which has trailers
in it, which has been parsed into the trailer_block structure.
Rename the
size_t trailer_block_start, trailer_block_end;
members of trailer_info to just "start" and "end". Rename the "info"
pointer to "trailer_block" because it is more descriptive. Update
comments accordingly.
Signed-off-by: Linus Arver <linus@ucla.edu>
Signed-off-by: Taylor Blau <me@ttaylorr.com>
|
|
Refactor t3404 to replace instances of `test` with `test_line_count()`
for checking line counts. This improves readability and aligns with Git's
current test practices.
Signed-off-by: Usman Akinyemi <usmanakinyemi202@gmail.com>
Signed-off-by: Taylor Blau <me@ttaylorr.com>
|
|
The exit code of the preceding command in a pipe is disregarded. So
if that preceding command is a Git command that fails, the test would
not fail. Instead, by saving the output of that Git command to a file,
and removing the pipe, we make sure the test will fail if that Git
command fails. This particular patch focuses on all `git show` and
some instances of `git cat-file`.
Signed-off-by: Usman Akinyemi <usmanakinyemi202@gmail.com>
Signed-off-by: Taylor Blau <me@ttaylorr.com>
|
|
Discussing the desire to make breaking changes, declaring that
breaking changes are made at a certain version boundary, and
recording these decisions in this document, are necessary but not
sufficient. We need to make sure that we can implement, test, and
deploy such impactful changes.
Earlier we considered to guard the breaking changes with a run-time
check of the `feature.git<version>` configuration to allow brave
users and developers to opt into them as early adoptors. But the
engineering cost to support such a run-time switch, covering new and
disappearing git subcommands and how "git help" would adjust the
documentation to the run-time switch, would be unrealistically high
to be worth it.
Formalize the mechanism based on a compile-time switch to allow
early adopters to opt into the breaking change in a version of Git
before the planned version for the breaking change.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
|
The documentation only lists "files" as a possible value, but
"reftable" is also valid.
Signed-off-by: Bence Ferdinandy <bence@ferdinandy.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
|
The clar source file '$(UNIT_TEST_DIR)/clar/clar.c' includes the
generated 'clar.suite', but this dependency is not taken into account by
our Makefile, so that it is possible for a parallel build to fail if
Make tries to build 'clar.o' before 'clar.suite' is generated.
Correctly specify the dependency.
Signed-off-by: Philippe Blain <levraiphilippeblain@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
|
As part of the effort to get rid of global state due to the global
the_repository variable, replace the_repository with the repository
argument that gets passed down through the builtin function.
The repo might be NULL, but we should be safe in write_archive() because
it detects if we are outside of a repository and calls
setup_git_directory() which will error.
Signed-off-by: John Cai <johncai86@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
|
As part of the effort to get rid of global state due to the_repository
variable, remove the the_repository with the repository argument that
gets passed down through the builtin function.
Signed-off-by: John Cai <johncai86@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
|
The current code in run_builtin() passes in a repository to the builtin
based on whether cmd_struct's option flag has RUN_SETUP.
This is incorrect, however, since some builtins that only have
RUN_SETUP_GENTLY can potentially take a repository.
setup_git_directory_gently() tells us whether or not a command is being
run inside of a repository.
Use the output of setup_git_directory_gently() to help determine whether
or not there is a repository to pass to the builtin. If not, then we
just pass NULL.
As part of this patch, we need to modify add to check for a NULL repo
before calling repo_git_config(), since add -h can be run outside of a
repository.
Signed-off-by: John Cai <johncai86@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
|
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
|
Code clean-up.
* jk/output-prefix-cleanup:
diff: store graph prefix buf in git_graph struct
diff: return line_prefix directly when possible
diff: return const char from output_prefix callback
diff: drop line_prefix_length field
line-log: use diff_line_prefix() instead of custom helper
|
|
More leakfixes.
* ps/leakfixes-part-8: (23 commits)
builtin/send-pack: fix leaking list of push options
remote: fix leaking push reports
t/helper: fix leaks in proc-receive helper
pack-write: fix return parameter of `write_rev_file_order()`
revision: fix leaking saved parents
revision: fix memory leaks when rewriting parents
midx-write: fix leaking buffer
pack-bitmap-write: fix leaking OID array
pseudo-merge: fix leaking strmap keys
pseudo-merge: fix various memory leaks
line-log: fix several memory leaks
diff: improve lifecycle management of diff queues
builtin/revert: fix leaking `gpg_sign` and `strategy` config
t/helper: fix leaking repository in partial-clone helper
builtin/clone: fix leaking repo state when cloning with bundle URIs
builtin/pack-redundant: fix various memory leaks
builtin/stash: fix leaking `pathspec_from_file`
submodule: fix leaking submodule entry list
wt-status: fix leaking buffer with sparse directories
shell: fix leaking strings
...
|
|
Use after free and double freeing at the end in "git log -L... -p"
had been identified and fixed.
* ds/line-log-asan-fix:
line-log: protect inner strbuf from free
|
|
Doc update to clarify how periodical maintenance are scheduled,
spread across time to avoid thundering hurds.
* sk/doc-maintenance-schedule:
doc: add a note about staggering of maintenance
|
|
Document "amlog" notes.
* tb/notes-amlog-doc:
Documentation: mention the amlog in howto/maintain-git.txt
|
|
The reftable library is now prepared to expect that the memory
allocation function given to it may fail to allocate and to deal
with such an error.
* ps/reftable-alloc-failures: (26 commits)
reftable/basics: fix segfault when growing `names` array fails
reftable/basics: ban standard allocator functions
reftable: introduce `REFTABLE_FREE_AND_NULL()`
reftable: fix calls to free(3P)
reftable: handle trivial allocation failures
reftable/tree: handle allocation failures
reftable/pq: handle allocation failures when adding entries
reftable/block: handle allocation failures
reftable/blocksource: handle allocation failures
reftable/iter: handle allocation failures when creating indexed table iter
reftable/stack: handle allocation failures in auto compaction
reftable/stack: handle allocation failures in `stack_compact_range()`
reftable/stack: handle allocation failures in `reftable_new_stack()`
reftable/stack: handle allocation failures on reload
reftable/reader: handle allocation failures in `reader_init_iter()`
reftable/reader: handle allocation failures for unindexed reader
reftable/merged: handle allocation failures in `merged_table_init_iter()`
reftable/writer: handle allocation failures in `reftable_new_writer()`
reftable/writer: handle allocation failures in `writer_index_hash()`
reftable/record: handle allocation failures when decoding records
...
|
|
The way AsciiDoc is used for SYNOPSIS part of the manual pages has
been revamped. The sources, at least for the simple cases, got
vastly pleasant to work with.
* ja/doc-synopsis-markup:
doc: apply synopsis simplification on git-clone and git-init
doc: update the guidelines to reflect the current formatting rules
doc: introduce a synopsis typesetting
|
|
Fix typos via codespell.
Signed-off-by: Andrew Kreimer <algonell@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
|
Fix a typo in comments: bellow -> below.
Signed-off-by: Andrew Kreimer <algonell@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
|
Fix typos via codespell.
Signed-off-by: Andrew Kreimer <algonell@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
|
Fix typos via codespell.
Signed-off-by: Andrew Kreimer <algonell@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
|
Fix typos via codespell.
Signed-off-by: Andrew Kreimer <algonell@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
|
Fix typos and grammar.
Reported-by: Patrick Steinhardt <ps@pks.im>
Signed-off-by: Andrew Kreimer <algonell@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
|
We can only check out commits or branches, not refs in general. And the
problem here is if another worktree is using the branch that we want to
check out.
Let’s be more direct and just talk about branches instead of refs.
Also replace “be held” with “in use”. Further, “in use” is not
restricted to a branch being checked out (e.g. the branch could be busy
on a rebase), hence generalize to “or otherwise in use” in the option
description.
Signed-off-by: Kristoffer Haugsbakk <code@khaugsbakk.name>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
|
Signed-off-by: Xing Xin <xingxin.xx@bytedance.com>
Acked-by: Kristoffer Haugsbakk <kristofferhaugsbakk@fastmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
|
The function `unbundle_from_file()` has two memory leaks:
- We do not release the `struct bundle_header header` when hitting
errors because we return early without any cleanup.
- We do not release the `struct strbuf bundle_ref` at all.
Plug these leaks by creating a common exit path where both of these
variables are released.
While at it, refactor the code such that the variable assignments do not
happen inside the conditional statement itself according to our coding
style.
Signed-off-by: Toon Claes <toon@iotcl.com>
Acked-by: Patrick Steinhardt <ps@pks.im>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
|
It was reported on the mailing list that running `git maintenance start`
immediately segfaults starting with b6c3f8e12c (builtin/maintenance: fix
leak in `get_schedule_cmd()`, 2024-09-26). And indeed, this segfault is
trivial to reproduce up to a point where one is scratching their head
why we didn't catch this regression in our test suite.
The root cause of this error is `get_schedule_cmd()`, which does not
populate the `out` parameter in all cases anymore starting with the
mentioned commit. Callers do assume it to always be populated though and
will e.g. call `strvec_split()` on the returned value, which will of
course segfault when the variable is uninitialized.
So why didn't we catch this trivial regression? The reason is that our
tests always set up the "GIT_TEST_MAINT_SCHEDULER" environment variable
via "t/test-lib.sh", which allows us to override the scheduler command
with a custom one so that we don't accidentally modify the developer's
system. But the faulty code where we don't set the `out` parameter will
only get hit in case that environment variable is _not_ set, which is
never the case when executing our tests.
Fix the regression by again unconditionally allocating the value in the
`out` parameter, if provided. Add a test that unsets the environment
variable to catch future regressions in this area.
Reported-by: Shubham Kanodia <shubham.kanodia10@gmail.com>
Signed-off-by: Patrick Steinhardt <ps@pks.im>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
|
We explicitly avoid saying "ref <src>" when introducing the source
side of a refspec, because it can be a fully-spelled hexadecimal
object name, and it also can be a pattern that is not quite a "ref".
But we are loose when we introduce <dst> and say "ref <dst>", even
though it can also be a pattern. Let's omit "ref" also from the
destination side.
Clarify that <src> can be a ref, a (limited glob) pattern, or an
object name.
Even though the very original design of refspec expected that '*'
was used only at the end (e.g., "refs/heads/*" was expected, but not
"refs/heads/*-wip"), the code and its use evolved to handle a single
'*' anywhere in the pattern. Update the text to remove the mention
of "the same prefix". Anything that matches the pattern are named
by such a (limited glob) pattern in <src>.
Also put a bit more stress on the fact that we accept only one '*'
in the pattern by saying "one and only one `*`".
Helped-by: Monika Kairaitytė <monika@kibit.lt>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
|
This test script uses "test - [def]", but when a test fails because
the file passed to it does not exist,
it fails silently without an error message.
Use test_path_* helper functions, which are designed to give better
error messages when their expectations are not met.
I have added a mechanical validation that applies the same transformation
done in this patch, when the test script is passed to a sed script as shown
below.
sed -e 's/^\( *\)test -f /\1test_path_is_file /' \
-e 's/^\( *\)test -d /\1test_path_is_dir /' \
-e 's/^\( *\)test -e /\1test_path_exists /' \
-e 's/^\( *\)! test -[edf] /\1test_path_is_missing /' \
-e 's/^\( *\)test ! -[edf] /\1test_path_is_missing /' \
"$1" >foo.sh
Reviewers can use the sed script to tranform the original test script and
compare the result in foo.sh with the results of applying the patch.
You will see an instance of "!(test -e 3)" which was manually replaced with
""test_path_is_missing 3", and everything else should match.
Careful and deliberate observation was done to check instances where
"test ! - [df] foo" was used in the test script to make sure that the test
instances were expecting foo to EITHER be a file or a directory, and NOT a
possibility of being both as this would make replacing "test ! -f foo" with
"test_path_is_missing foo" unreasonable.
In the tests control flow, foo has been created as EITHER a
reguar file OR a directory and should NOT exist
after "git clean" or "git clean -d", as the case maybe, has been called.
This made it reasonable to replace
"test ! -[df] foo" with "test_path_is_missing foo".
Signed-off-by: Abraham Samuel Adekunle <abrahamadekunle50@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
|
In `loose.c`, we rely on the global variable `the_hash_algo`. As such we
have guarded the file with the 'USE_THE_REPOSITORY_VARIABLE' definition.
Let's derive the hash algorithm from the available repository variable
and remove this guard. This brings us one step closer to removing the
global 'the_repository' variable.
Signed-off-by: Karthik Nayak <karthik.188@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
|
Add jobs that exercise Git on Windows. Unfortunately, building and
especially testing Git on Windows is inherently slower compared to other
Unix-like systems, mostly because spawning processes is way slower. We
thus use the same layout as we use in GitHub Actions, where we have one
build job, and then pass on the resulting build artifacts to ten test
jobs that split up the work across each other.
Unfortunately, the GitLab runners for Windows machines are embarassingly
slow by themselves. So while this strategy leads to around 20 minutes of
build time in GitHub Actions, the same pipeline takes around an hour in
GitLab CI. Still, having late coverage is certainly better than having
none at all.
Signed-off-by: Patrick Steinhardt <ps@pks.im>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
|
We're about to add a couple of jobs for Windows. As the Windows runners
are quite slow, we will split those up across two stages: one stage to
build the artifacts, and one stage that runs test slices in parallel.
Introduce stages and "needs" dependencies for the preexisting jobs as a
preparatory step. The stages will lead to a more natural representation
of jobs in the UI, whereas the "needs" dependency ensures that jobs do
not have to wait for all jobs in the preceding stage to finish.
Signed-off-by: Patrick Steinhardt <ps@pks.im>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
|
We try to abstract away any differences between different CI platforms
in "ci/lib.sh", such that knowledge specific to e.g. GitHub Actions or
GitLab CI is neatly encapsulated in a single place. Next to some generic
variables, we also set up some variables that are specific to the actual
platform that the CI operates on, e.g. Linux or macOS.
We do not yet support Windows runners on GitLab CI. Unfortunately, those
systems do not use the same "CI_JOB_IMAGE" environment variable as both
Linux and macOS do. Instead, we can use the "OS" variable, which should
have a value of "Windows_NT" on Windows platforms.
Handle the combination of "$OS,$CI_JOB_IMAGE" and introduce support for
Windows.
Signed-off-by: Patrick Steinhardt <ps@pks.im>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
|
In order to build and test Git, we have to first set up the Git for
Windows SDK, which contains various required tools and libraries. The
SDK is basically a clone of [1], but that repository is quite large due
to all the binaries it contains. We thus use both shallow clones and
sparse checkouts to speed up the setup. To handle this complexity we use
a GitHub action that is hosted externally at [2].
Unfortunately, this makes it rather hard to reuse the logic for CI
platforms other than GitHub Actions. After chatting with Johannes
Schindelin we came to the conclusion that it would be nice if the Git
for Windows SDK would regularly publish releases that one can easily
download and extract, thus moving all of the complexity into that single
step. Like this, all that a CI job needs to do is to fetch and extract
the resulting archive. This published release comes in the form of a new
"ci-artifacts" tag that gets updated regularly [3].
Implement a new script that knows how to fetch and extract that script
and convert GitHub Actions to use it.
[1]: https://github.com/git-for-windows/git-sdk-64/
[2]: https://github.com/git-for-windows/setup-git-for-windows-sdk/
[3]: https://github.com/git-for-windows/git-sdk-64/releases/tag/ci-artifacts/
Helped-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Patrick Steinhardt <ps@pks.im>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
|
Windows by default has a restriction in place to only allow paths up to
260 characters. This restriction can nowadays be lifted by setting a
registry key, but is still active by default.
In t7300 we have one test that exercises the behaviour of git-clean(1)
with such long paths. Interestingly enough, this test fails on my system
that uses Windows 10 with mingw-w64 installed via MSYS2: instead of
observing ENAMETOOLONG, we observe ENOENT. This behaviour is consistent
across multiple different environments I have tried.
I cannot say why exactly we observe a different error here, but I would
not be surprised if this was either dependent on the Windows version,
the version of MinGW, the current working directory of Git or any kind
of combination of these.
Work around the issue by handling both errors.
Signed-off-by: Patrick Steinhardt <ps@pks.im>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
|
When interactively rebasing merge commits, the commit message is parsed to
extract a probably meaningful label name. For instance if the merge commit
is “Merge branch 'feature0'”, then the rebase script will have thes lines:
```
label feature0
merge -C $sha feature0 # “Merge branch 'feature0'
```
This heuristic fails in the case of octopus merges or when the merge commit
message is actually unrelated to the parent commits.
An example that combines both is:
```
*---. 967bfa4 (HEAD -> integration) Integration
|\ \ \
| | | * 2135be1 (feature2, feat2) Feature 2
| |_|/
|/| |
| | * c88b01a Feature 1
| |/
|/|
| * 75f3139 (feat0) Feature 0
|/
* 25c86d0 (main) Initial commit
```
yields the labels Integration, Integration-2 and Integration-3.
Fix this by using a branch name for each merge commit's parent that is the
tip of at least one branch, and falling back to a label derived from the
merge commit message otherwise.
In the example above, the labels become feat0, Integration and feature2.
Signed-off-by: Nicolas Guichard <nicolas@guichard.eu>
Acked-by: Phillip Wood <phillip.wood@dunelm.org.uk>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
|
Extract load_branch_decorations from todo_list_add_update_ref_commands so
it can be re-used in make_script_with_merges.
Since it can now be called multiple times, use non-static lists and place
it next to load_ref_decorations to re-use the decoration_loaded guard.
Signed-off-by: Nicolas Guichard <nicolas@guichard.eu>
Acked-by: Phillip Wood <phillip.wood@dunelm.org.uk>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
|
load_branch_decorations calls normalize_glob_ref on each string of filter's
string_lists. This effectively replaces the potentially non-owning char* of
those items with an owning char*.
Set the strdup_string flag on those string_lists.
This was not caught until now because:
- when passing string_lists already with the strdup_string already set, the
behaviour was correct
- when passing static string_lists, the new char* remain reachable until
program exit
Signed-off-by: Nicolas Guichard <nicolas@guichard.eu>
Acked-by: Phillip Wood <phillip.wood@dunelm.org.uk>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
|
The code fetches the submodules remote based on the superproject remote name
instead of the submodule remote name[1].
Instead of grabbing the default remote of the superproject repository, ask
the default remote of the submodule we are going to run 'git fetch' in.
1. https://lore.kernel.org/git/ZJR5SPDj4Wt_gmRO@pweza/
Signed-off-by: Daniel Black <daniel@mariadb.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
|
• Provide a commit message in the example command.
The command will hang since it is waiting for a commit message on
stdin. Which is usable but not straightforward enough since this is
example code.
• Use `||` directly since that is more straightforward than checking the
last exit status.
Also use `echo` and `exit` since `die` is not defined.
• Expose variable declarations.
Signed-off-by: Kristoffer Haugsbakk <code@khaugsbakk.name>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
|
The synopsis for `git config unset` mentions two positional arguments:
`<name>` and `<value>`. While the first argument is correct, the second
is not. Users are expected to provide the value via `--value=<value>`.
Remove the positional argument. The `--value=<value>` option is already
documented correctly, so this is all we need to do to fix the
documentation.
Signed-off-by: Josh Heinrichs <joshiheinrichs@gmail.com>
Reviewed-by: Patrick Steinhardt <ps@pks.im>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
|
There's a racy hang in fsmonitor on macOS that we sometimes see in CI.
When we serve a client, what's supposed to happen is:
1. The client thread calls with_lock__wait_for_cookie() in which we
create a cookie file and then wait for a pthread_cond event
2. The filesystem event listener sees the cookie file creation, does
some internal book-keeping, and then triggers the pthread_cond.
But there's a problem: we start the listener that accepts client threads
before we start the fs event thread. So it's possible for us to accept a
client which creates the cookie file and starts waiting before the fs
event thread is initialized, and we miss those filesystem events
entirely. That leaves the client thread hanging forever.
In CI, the symptom is that t9210 (which is testing scalar, which always
enables fsmonitor under the hood) may hang forever in "scalar clone". It
is waiting on "git fetch" which is waiting on the fsmonitor daemon.
The race happens more frequently under load, but you can trigger it
predictably with a sleep like this, which delays the start of the fs
event thread:
--- a/compat/fsmonitor/fsm-listen-darwin.c
+++ b/compat/fsmonitor/fsm-listen-darwin.c
@@ -510,6 +510,7 @@ void fsm_listen__loop(struct fsmonitor_daemon_state *state)
FSEventStreamSetDispatchQueue(data->stream, data->dq);
data->stream_scheduled = 1;
+ sleep(1);
if (!FSEventStreamStart(data->stream)) {
error(_("Failed to start the FSEventStream"));
goto force_error_stop_without_loop;
One solution might be to reverse the order of initialization: start the
fs event thread before we start the thread listening for clients. But
the fsmonitor code explicitly does it in the opposite direction. The fs
event thread wants to refer to the ipc_server_data struct, so we need it
to be initialized first.
A further complication is that we need a signal from the fs event thread
that it is actually ready and listening. And those details happen within
backend-specific fsmonitor code, whereas the initialization is in the
shared code.
So instead, let's use the ipc_server init/start split added in the
previous commit. The generic fsmonitor code will init the ipc_server but
_not_ start it, leaving that to the backend specific code, which now
needs to call ipc_server_start_async() at the right time.
For macOS, that is right after we start the FSEventStream that you can
see in the diff above.
It's not clear to me if Windows suffers from the same problem (and we
simply don't trigger it in CI), or if it is immune. Regardless, the
obvious place to start accepting clients there is right after we've
established the ReadDirectoryChanges watch.
This makes the hangs go away in our macOS CI environment, even when
compiled with the sleep() above.
Helped-by: Koji Nakamaru <koji.nakamaru@gree.net>
Signed-off-by: Jeff King <peff@peff.net>
Acked-by: Koji Nakamaru <koji.nakamaru@gree.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
|
To start an async ipc server, you call ipc_server_run_async(). That
initializes the ipc_server_data object, and starts all of the threads
running, which may immediately start serving clients.
This can create some awkward timing problems, though. In the fsmonitor
daemon (the sole user of the simple-ipc system), we want to create the
ipc server early in the process, which means we may start serving
clients before the rest of the daemon is fully initialized.
To solve this, let's break run_async() into two parts: an initialization
which allocates all data and spawns the threads (without letting them
run), and a start function which actually lets them begin work. Since we
have two simple-ipc implementations, we have to handle this twice:
- in ipc-unix-socket.c, we have a central listener thread which hands
connections off to worker threads using a work_available mutex. We
can hold that mutex after init, and release it when we're ready to
start.
We do need an extra "started" flag so that we know whether the main
thread is holding the mutex or not (e.g., if we prematurely stop the
server, we want to make sure all of the worker threads are released
to hear about the shutdown).
- in ipc-win32.c, we don't have a central mutex. So we'll introduce a
new startup_barrier mutex, which we'll similarly hold until we're
ready to let the threads proceed.
We again need a "started" flag here to make sure that we release the
barrier mutex when shutting down, so that the sub-threads can
proceed to the finish.
I've renamed the run_async() function to init_async() to make sure we
catch all callers, since they'll now need to call the matching
start_async().
We could leave run_async() as a wrapper that does both, but there's not
much point. There are only two callers, one of which is fsmonitor, which
will want to actually do work between the two calls. And the other is
just a test-tool wrapper.
For now I've added the start_async() calls in fsmonitor where they would
otherwise have happened, so there should be no behavior change with this
patch.
Signed-off-by: Jeff King <peff@peff.net>
Acked-by: Koji Nakamaru <koji.nakamaru@gree.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
