Add tests exercising the interaction between http.emptyAuth and
servers that advertise Negotiate (SPNEGO) authentication.
Verify that auto mode gives Negotiate a chance via empty auth
(resulting in two 401 responses before falling through to
credential_fill with Basic credentials), and that false mode
strips Negotiate immediately (only one 401 response).
Signed-off-by: Matthew John Cheetham <mjcheetham@outlook.com>
The previous commits disabled NTLM authentication by default due to its
cryptographic weaknesses. Users can re-enable it via the config setting
http.<url>.allowNTLMAuth, but this requires manual intervention.
Credential helpers may have knowledge about which servers are trusted
for NTLM authentication (e.g., known on-prem Azure DevOps instances).
To allow them to signal this trust, introduce a simple negotiation:
when NTLM is suppressed and the server offered it, Git advertises
ntlm=suppressed to the credential helper. The helper can respond with
ntlm=allow to re-enable NTLM for this request.
This happens precisely at the point where we would otherwise warn the
user about NTLM being suppressed, ensuring the capability is only
advertised when relevant.
Helped-by: Matthew John Cheetham <mjcheetham@outlook.com>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
The new default of Git is to disable NTLM authentication by default.
To help users find the escape hatch of that config setting, should they
need it, suggest it when the authentication failed and the server had
offered NTLM, i.e. if re-enabling it would fix the problem.
Helped-by: Patrick Steinhardt <ps@pks.im>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
NTLM authentication is relatively weak. This is the case even with the
default setting of modern Windows versions, where NTLMv1 and LanManager
are disabled and only NTLMv2 is enabled: NTLMv2 hashes of even
reasonably complex 8-character passwords can be broken in a matter of
days, given enough compute resources.
Even worse: On Windows, NTLM authentication uses Security Support
Provider Interface ("SSPI"), which provides the credentials without
requiring the user to type them in.
Which means that an attacker could talk an unsuspecting user into
cloning from a server that is under the attacker's control and extracts
the user's NTLMv2 hash without their knowledge.
For that reason, let's disallow NTLM authentication by default.
NTLM authentication is quite simple to set up, though, and therefore
there are still some on-prem Azure DevOps setups out there whose users
and/or automation rely on this type of authentication. To give them an
escape hatch, introduce the `http.<url>.allowNTLMAuth` config setting
that can be set to `true` to opt back into using NTLM for a specific
remote repository.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Although NTLM authentication is considered weak (extending even to
NTLMv2, which purportedly allows brute-forcing reasonably complex
8-character passwords in a matter of days, given ample compute
resources), it _is_ one of the authentication methods supported by
libcurl.
Note: The added test case *cannot* reuse the existing `custom_auth`
facility. The reason is that that facility is backed by an NPH script
("No Parse Headers"), which does not allow handling the 3-phase NTLM
authentication correctly (in my hands, the NPH script would not even be
called upon the Type 3 message, a "200 OK" would be returned, but no
headers, let alone the `git http-backend` output as payload). Having a
separate NTLM authentication script makes the exact workings clearer and
more readable, anyway.
Co-authored-by: Matthew John Cheetham <mjcheetham@outlook.com>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
These fixes have been sent to the Git mailing list but have not been
picked up by the Git project yet.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
In some implementations, `regexec_buf()` assumes that it is fed lines;
Without `REG_NOTEOL` it thinks the end of the buffer is the end of a
line. Which makes sense, but trips up this case because we are not
feeding lines, but rather a whole buffer. So the final newline is not
the start of an empty line, but the true end of the buffer.
This causes an interesting bug:
$ echo content >file.txt
$ git grep --no-index -n '^$' file.txt
file.txt:2:
This bug is fixed by making the end of the buffer consistently the end
of the final line.
The patch was applied from
https://lore.kernel.org/git/20250113062601.GD767856@coredump.intra.peff.net/
Reported-by: Olly Betts <olly@survex.com>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
When a Unix socket is initialized, the current directory's path is
stored so that the cleanup code can `chdir()` back to where it was
before exit.
If the path that needs to be stored exceeds the default size of the
`sun_path` attribute of `struct sockaddr_un` (which is defined as a
108-sized byte array on Linux), a larger buffer needs to be allocated so
that it can hold the path, and it is the responsibility of the
`unix_sockaddr_cleanup()` function to release that allocated memory.
In Git's CI, this stack allocation is not necessary because the code is
checked out to `/home/runner/work/git/git`. Concatenate the path
`t/trash directory.t0301-credential-cache/.cache/git/credential/socket`
and a terminating NUL, and you end up with 96 bytes, 12 shy of the
default `sun_path` size.
However, I use worktrees with slightly longer paths:
`/home/me/projects/git/yes/i/nest/worktrees/to/organize/them/` is more
in line with what I have. When I recently tried to locally reproduce a
failure of the `linux-leaks` CI job, this t0301 test failed (where it
had not failed in CI).
The reason: When `credential-cache` tries to reach its daemon initially
by calling `unix_sockaddr_init()`, it is expected that the daemon cannot
be reached (the idea is to spin up the daemon in that case and try
again). However, when this first call to `unix_sockaddr_init()` fails,
the code returns early from the `unix_stream_connect()` function
_without_ giving the cleanup code a chance to run, skipping the
deallocation of above-mentioned path.
The fix is easy: do not return early but instead go directly to the
cleanup code.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
On Windows, symbolic links come in two flavors: file symlinks and
directory symlinks. Since Git was born on Linux where this distinction
does not exist, Git for Windows has to auto-detect the type by looking
at the target. When the target does not yet exist at symlink creation
time, Git for Windows creates a "phantom" file symlink and later, once
checkout is complete, calls `CreateFileW()` on the target to check
whether it is actually a directory.
If the symlink target is a UNC path (e.g. `\\attacker\share`), this
auto-detection triggers an SMB connection to the remote host. Windows
performs NTLM authentication by default for such connections, which
means a crafted repository can exfiltrate the cloning user's NTLMv2
hash to an attacker-controlled server without any user interaction
beyond `git clone -c core.symlinks=true <url>`.
There are ways to specify UNC paths that start with only a single
backslash (e.g. `\??\UNC\host\share`); All of them do start like
that, though, so let's use that as a tell-tale that we should skip
the auto-detection in `process_phantom_symlink()`. The symlink is
then left as a file symlink (the `mklink` default), and a warning is
emitted suggesting the user set the `symlink` gitattribute to `dir`
if a directory symlink is needed. When the attribute is already set,
auto-detection is never invoked in the first place, so that code path
is unaffected.
This is the same class of vulnerability as CVE-2025-66413
(https://github.com/git-for-windows/git/security/advisories/GHSA-hv9c-4jm9-jh3x)
and follows the same general mitigation pattern that MinTTY adopted for
ANSI escape sequences referencing network share paths
(https://github.com/mintty/mintty/security/advisories/GHSA-jf4m-m6rv-p6c5).
Note that there are legitimate paths starting with a single backslash
that are _not_ network paths: drive-less absolute paths are interpreted
as relative to the current working directory's drive. In practice, these
are highly uncommon (and brittle, just one working directory change
away from breaking). In any case, the only consequence is now that the
symlink type of those has to be specified via Git attributes, is all.
Reported-by: Justin Lee <jessdhoctor@gmail.com>
Addresses: CVE-2026-32631
Addresses: https://github.com/git-for-windows/git/security/advisories/GHSA-9j5h-h4m7-85hx
Assisted-by: Claude Opus 4.6
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Every once in a while I need to verify that Microsoft Git's test suite
passes for changes that are not yet meant for public consumption, and
since it was (made) too difficult to keep up a working Azure Pipeline
definition, I have to use GitHub Actions in a private GitHub repository
for that purpose.
In these tests, basically all Dockerized CI jobs fail consistently. The
symptom is something like:
error: cannot create async thread: Resource temporarily unavailable
in the middle of a test, typically in the t5xxx-t6xxx range. The first
such error is immediately followed by plenty more of these errors, and
not a single test succeeds afterwards.
At first, I thought that maybe the massive parallelism I enjoy there is
the problem, and I thought that the cgroups limits might be shared
between the many containers that run on essentially the same physical
machine. But even reducing the matrix to just a single of those
Dockerized jobs runs into the very same problems.
The underlying reason seems to be a substantial difference in the hosted
runners that execute these Dockerized jobs: forcing the PID limit of the
container to a high number lets the jobs pass, even when running the
complete matrix of all 13 Dockerized jobs concurrently. But that's not
the only difference: The jobs seem to take a lot longer in these
containers than, say, in the containers made available to
https://github.com/git/git.
When forcing a PID limit of 64k in that private repository, the jobs
completed successfully, but they also took a lot longer, between 2x to
2.5x longer, i.e. painfully much longer. Reducing the PID limit to 16k,
the CI jobs still passed, but took an equally long amount of time.
Reducing the PID limit to 8k caused the errors to reappear.
Here are the numbers from three example runs, the first one forcing the
PID and nproc limit to 65536, the second one to 16384, the third run is
from the public git/git repository:
Job | 64k | 16k | reference
------------------------------|---------|---------|---------
almalinux-8 | 19m 3s | 16m 0s | 9m 36s
debian-11 | 20m 31s | 20m 3s | 8m 5s
fedora-breaking-changes-meson | 16m 29s | 19m 19s | 9m 40s
linux-asan-ubsan | 1h 10m | 1h 11m | 34m 36s
linux-breaking-changes | 25m 39s | 25m 58s | 13m 15s
linux-leaks | 1h 9m | 1h 10m | 33m 30s
linux-meson | 28m 9s | 27m 4s | 13m 45s
linux-musl-meson | 16m 32s | 13m 39s | 8m 6s
linux-reftable-leaks | 1h 13m | 1h 13m | 34m 34s
linux-reftable | 26m 2s | 25m 48s | 13m 31s
linux-sha256 | 26m 12s | 26m 3s | 12m 36s
linux-TEST-vars | 26m 5s | 25m 21s | 13m 25s
linux32 | 21m 16s | 19m 57s | 10m 44s
It does not look as if the PID limit is the reason for the longer
runtime, seeing as the 64k vs 16k timings deviate no more than as is
usual with GitHub workflows. So let's go for 16k.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
* js/objects-larger-than-4gb-on-windows-more:
odb: use size_t for object_info.sizep and the size APIs
packfile,delta: drop the `cast_size_t_to_ulong()` wrappers
pack-objects: use size_t for in-core object sizes
packfile: widen unpack_entry()'s size out-parameter to size_t
pack-objects(check_pack_inflate()): use size_t instead of unsigned long
patch-delta: use size_t for sizes
compat/msvc: use _chsize_s for ftruncate
A new `diff.<driver>.process` configuration has been introduced to
allow a long-running external process to act as a hunk provider to
allows external tools to control which lines Git considers changed
while leaving all output formatting (word diff, color, blame, etc.) to
Git's standard pipeline.
* mm/diff-process-hunks:
blame: consult diff process for no-hunk detection
diff: bypass diff process with --no-ext-diff and in format-patch
diff: add long-running diff process via diff.<driver>.process
sub-process: separate process lifecycle from hashmap management
userdiff: add diff.<driver>.process config
xdiff: support external hunks via xpparam_t
"git log --follow" has been updated to handle non-linear history, in
which the path being tracked gets renamed differently in multiple
history lines, better.
* mv/log-follow-mergy:
log: improve --follow following renames for non-linear history
"git log --graph" has been modified to visually distinguish
parentless "root" commits (and commits that become roots due to
history simplification) by indenting them, preventing them from
appearing falsely related to unrelated commits rendered immediately
above them.
* ps/shift-root-in-graph:
graph: indent visual root in graph
lib-log-graph: move check_graph function
The `git multi-pack-index write --incremental` command has been
corrected to properly honor the `--base` option. Previously, the
custom base was ignored by the normal write path, and the pack
exclusion logic incorrectly skipped packs from layers above the
selected base, breaking reachability closure for bitmaps.
* tb/midx-incremental-custom-base:
midx-write: include packs above custom incremental base
midx: pass custom '--base' through incremental writes
t5334: expose shared `nth_line()` helper
The contributor guide has been updated to advise new contributors to
trim irrelevant quoted text when replying to review comments, matching
the existing advice given to reviewers.
* wy/doc-myfirstcontribution-trim-quotes:
MyFirstContribution: mention trimming quoted text in replies
The trailer sections in SubmittingPatches have been updated to
encourage use of standard trailers.
* kh/submittingpatches-trailers:
SubmittingPatches: note that trailer order matters
SubmittingPatches: be consistent with trailer markup
SubmittingPatches: document Based-on-patch-by trailer
SubmittingPatches: discourage common Linux trailers
SubmittingPatches: discuss non-ident trailers
SubmittingPatches: encourage trailer use for substantial help
The reference backends have been converted to always use absolute
paths internally. This allows dropping the calls to
`chdir_notify_reparent()` and fixes a memory leak in how the
reference database is constructed with an "onbranch" condition.
* ps/refs-avoid-chdir-notify-reparent:
refs: drop local buffer in `refs_compute_filesystem_location()`
refs: fix recursing `get_main_ref_store()` with "onbranch" config
repository: free main reference database
chdir-notify: drop unused `chdir_notify_reparent()`
refs: unregister reference stores from "chdir_notify"
setup: don't apply "GIT_REFERENCE_BACKEND" without a repository
setup: stop applying repository format twice
setup: inline `check_and_apply_repository_format()`
git replay learns --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: add helper to put entry into mapped_commits
replay: refactor enum replay_mode into a bool
'git branch --contains' and 'git for-each-ref --contains' have
been optimized to use the memoized commit traversal previously
used only by 'git tag --contains', significantly speeding up
connectivity checks across many candidate refs with shared
history.
* td/ref-filter-memoize-contains:
commit-reach: die on contains walk errors
ref-filter: memoize --contains with generations
commit-reach: reject cycles in contains walk
"git branch" command learned "--prune-merged" option to remove
local branches that have already been merged to the remote-tracking
branches they track.
* hn/branch-prune-merged:
branch: add --dry-run for --delete-merged
branch: add branch.<name>.deleteMerged opt-out
branch: add --delete-merged <branch>
branch: prepare delete_branches for a bulk caller
branch: let delete_branches skip unmerged branches on bulk refusal
branch: convert delete_branches() to a flags argument
branch: add --forked filter for --list mode
The lazy priority queue optimization pattern (deferring actual removal
in prio_queue_get() to allow get+put fusion) has been folded directly
into prio_queue itself, speeding up commit traversal workflows and
simplifying callers.
* kk/prio-queue-get-put-fusion:
prio-queue: fold lazy_queue into prio_queue for automatic get+put fusion
prio-queue: rename .nr to .nr_ and add accessor helpers
The `remote-object-info` command has been added to `git cat-file
--batch-command`, allowing clients to request object metadata
(currently size) from a remote server via protocol v2 without
downloading the entire object.
The client dynamically filters format placeholders based on
server-advertised capabilities and safely returns empty strings for
inapplicable or unsupported fields.
* ps/cat-file-remote-object-info:
cat-file: make remote-object-info allow-list dynamic
cat-file: validate remote atoms with allow_list
cat-file: add remote-object-info to batch-command
transport: add client support for object-info
serve: advertise object-info feature
fetch-pack: move fetch initialization
connect: refactor packet writing
fetch-pack: move function to connect.c
t1006: split test utility functions into new "lib-cat-file.sh"
cat-file: add declaration of variable i inside its for loop
git-compat-util: add strtoul_ul() with error handling
transport-helper: fix memory leak of helper on disconnect
Two new mechanisms, the GIT_CONFIG_INCLUDES environment variable and
the top-level --no-includes command-line option, have been introduced
to ignore configuration include directives.
Retracted.
cf. <539713c4-b291-42e6-8541-a16a454518f5@gmail.com>
* ds/config-no-includes:
git: add --no-includes top-level option
config: add GIT_CONFIG_INCLUDES
git-config.adoc: fix paragraph break
Commands that list branches and tags (like git branch and git tag)
have been optimized to pass the namespace prefix when initializing
their ref iterator, avoiding a loose-ref scaling regression in
repositories with many unrelated loose references.
* td/ref-filter-restore-prefix-iteration:
ref-filter: restore prefix-scoped iteration
The packed object source has been refactored into a proper struct
odb_source.
* ps/odb-source-packed:
odb/source-packed: drop pointer to "files" parent source
midx: refactor interfaces to work on "packed" source
odb/source-packed: stub out remaining functions
odb/source-packed: wire up `freshen_object()` callback
odb/source-packed: wire up `find_abbrev_len()` callback
odb/source-packed: wire up `count_objects()` callback
odb/source-packed: wire up `for_each_object()` callback
odb/source-packed: wire up `read_object_stream()` callback
odb/source-packed: wire up `read_object_info()` callback
packfile: use higher-level interface to implement `has_object_pack()`
odb/source-packed: wire up `reprepare()` callback
odb/source-packed: wire up `close()` callback
odb/source-packed: start converting to a proper `struct odb_source`
odb/source-packed: store pointer to "files" instead of generic source
packfile: move packed source into "odb/" subsystem
packfile: split out packfile list logic
packfile: rename `struct packfile_store` to `odb_source_packed`
When cURL follows a redirect, the WWW-Authenticate headers from the
redirect target were lost because credential_from_url() cleared the
credential state. This has been fixed by preserving the collected
headers across the redirect update.
* ap/http-redirect-wwwauth-fix:
http: preserve wwwauth_headers across redirects
A regression in the error diagnosis code for invalid .git files has
been fixed, avoiding a potential NULL-pointer crash when reporting
that a .git file does not point to a valid repository.
* jk/setup-gitfile-diag-fix:
read_gitfile_gently(): return non-repo path on error
The experimental "git history" command has been taught a new "drop"
subcommand to remove a commit and replay its descendants onto its
parent.
* ps/history-drop:
builtin/history: implement "drop" subcommand
builtin/history: split handling of ref updates into two phases
reset: stop assuming that the caller passes in a clean index
reset: allow the caller to specify the current HEAD object
reset: introduce ability to skip updating HEAD
reset: introduce dry-run mode
reset: modernize flags passed to `reset_working_tree()`
reset: rename `reset_head()`
reset: drop `USE_THE_REPOSITORY_VARIABLE`
read-cache: split out function to drop unmerged entries to stage 0
The "git repo info" command has been taught new keys to output both
absolute and relative paths for "gitdir" and "commondir", supported by
a new path-formatting helper extracted from "git rev-parse".
* jk/repo-info-path-keys:
repo: add path.gitdir with absolute and relative suffix formatting
repo: add path.commondir with absolute and relative suffix formatting
rev-parse: use append_formatted_path() for path formatting
path: introduce append_formatted_path() for shared path formatting
prio_queue_get() has been optimized by using a cascade-down approach
(promoting the smaller child at each level and sifting up the last
element from the leaf vacancy), which halves the number of comparisons
per extract-min operation in the common case.
* kk/prio-queue-cascade-sift:
prio-queue: use cascade-down for faster extract-min
The 'trust_executable_bit' (coming from 'core.filemode'
configuration) has been migrated into 'repo_config_values' to tie it
to a specific repository instance.
* ty/migrate-trust-executable-bit:
environment: move trust_executable_bit into repo_config_values
read-cache: move 'ce_mode_from_stat()' to 'read-cache.c'
read-cache: remove redundant extern declarations
The pack-objects command now supports using reachability bitmaps and
delta-islands concurrently with the `--path-walk` option, allowing
faster packaging by falling back to path-walk when bitmaps cannot
fully satisfy the request.
* tb/pack-path-walk-bitmap-delta-islands:
pack-objects: support `--delta-islands` with `--path-walk`
pack-objects: extract `record_tree_depth()` helper
pack-objects: support reachability bitmaps with `--path-walk`
t/perf: drop p5311's lookup-table permutation
"git rebase --update-refs" has been taught to resolve local branch
symrefs to their referents before queuing updates. This correctly
skips aliases of the current branch and avoids duplicate updates for
underlying real branches, fixing failures when branch aliases (like a
default branch rename) are present.
* sn/rebase-update-refs-symrefs:
rebase: skip branch symref aliases
The -m/-F/-c/-C options to supply commit log message from outside the
editor are now supported for all "git commit --fixup" variations.
* ec/commit-fixup-options:
commit: allow -c/-C for all kinds of --fixup
commit: allow -m/-F for all kinds of --fixup
The [includeIf "condition"] conditional inclusion facility for
configuration files has learned to use the location of worktree
in its condition.
* cl/conditional-config-on-worktree-path:
config: add "worktree" and "worktree/i" includeIf conditions
config: refactor include_by_gitdir() into include_by_path()
When fetching from a transport that provides a self-contained pack,
pass the transport pointer to the post-fetch `check_connected()` call
to optimize connectivity check.
Retracted.
cf. <CAL71e4MrVqC1=AR6x0_8S=8kVqPdDkhgCZRb4etFsxTzd6s_8Q@mail.gmail.com>
* kk/fetch-store-ref-optimization:
fetch: pass transport to post-fetch connectivity check
"git checkout --track=..." learned to optionally fetch the branch
from the remote the new branch will work with.
* hn/checkout-track-fetch:
checkout: extend --track with a "fetch" mode to refresh start-point
branch: expose helpers for finding the remote owning a tracking ref
Configuration file locking now retries for a short period, avoiding
failures when multiple processes attempt to update the configuration
simultaneously.
* jt/config-lock-timeout:
config: retry acquiring config.lock, configurable via core.configLockTimeout
The parse-options library learned to auto-correct misspelled
subcommand names.
* js/parseopt-subcommand-autocorrection:
SQUASH???
doc: document autocorrect API
parseopt: add tests for subcommand autocorrection
parseopt: enable subcommand autocorrection for git-remote and git-notes
parseopt: autocorrect mistyped subcommands
autocorrect: provide config resolution API
autocorrect: rename AUTOCORRECT_SHOW to AUTOCORRECT_HINT
autocorrect: use mode and delay instead of magic numbers
help: move tty check for autocorrection to autocorrect.c
help: make autocorrect handling reusable
parseopt: extract subcommand handling from parse_options_step()
The display of the rebase todo list in "git status" has been
improved to correctly abbreviate object IDs for more commands and
avoid misinterpreting refs as object IDs.
* pw/status-rebase-todo:
status: improve rebase todo list parsing
sequencer: factor out parsing of todo commands
When fetching objects into a lazily cloned repository, .promisor
files are created with information meant to help debugging. "git
repack" has been taught to carry this information forward to
packfiles that are newly created.
Retracted.
cf. <agx_GPfBKpkSc3Gx@lorenzo-VM>
* lp/repack-propagate-promisor-debugging-info:
repack-promisor: add missing headers
t7703: test for promisor file content after geometric repack
t7700: test for promisor file content after repack
repack-promisor: preserve content of promisor files after repack
repack-promisor add helper to fill promisor file after repack
pack-write: add explanation to promisor file content
Project-specific configuration for b4 has been introduced, and the
documentation has been updated to recommend using it as a
streamlined method for submitting patches.
* ps/doc-recommend-b4:
b4: introduce configuration for the Git project
MyFirstContribution: recommend the use of b4
MyFirstContribution: recommend shallow threading of cover letters
