[ANNOUNCE] Git v2.50.1 and friends

A maintenance release Git v2.50.1 and others for older maintenance tracks down to maint-2.43 are now available at the usual places.

This is a set of coordinated security fix releases. Please update at your earliest convenience.

The tarballs are found at:

    https://www.kernel.org/pub/software/scm/git/

The following public repositories all have a copy of the v2.43.7, v2.44.4, v2.45.4, v2.46.4, v2.47.3, v2.48.2, v2.49.1 and v2.50.1 tags:

  url = https://git.kernel.org/pub/scm/git/git
  url = https://kernel.googlesource.com/pub/scm/git/git
  url = git://repo.or.cz/alt-git.git
  url = https://github.com/gitster/git

----------------------------------------------------------------

Git v2.50.1 Release Notes =========================

This release merges up the fixes that appear in v2.43.7, v2.44.4, v2.45.4, v2.46.4, v2.47.3, v2.48.2, and v2.49.1 to address the following CVEs: CVE-2025-27613, CVE-2025-27614, CVE-2025-46334, CVE-2025-46835, CVE-2025-48384, CVE-2025-48385, and CVE-2025-48386. See the release notes for v2.43.7 for details.

----------------------------------------------------------------

Git v2.43.7 Release Notes =========================

This release includes fixes for CVE-2025-27613, CVE-2025-27614, CVE-2025-46334, CVE-2025-46835, CVE-2025-48384, CVE-2025-48385, and CVE-2025-48386.

Fixes since v2.43.6 -------------------

 * CVE-2025-27613, Gitk:

   When a user clones an untrusted repository and runs Gitk without
   additional command arguments, any writable file can be created and
   truncated. The option "Support per-file encoding" must have been
   enabled. The operation "Show origin of this line" is affected as
   well, regardless of the option being enabled or not.

 * CVE-2025-27614, Gitk:

   A Git repository can be crafted in such a way that a user who has
   cloned the repository can be tricked into running any script
   supplied by the attacker by invoking `gitk filename`, where
   `filename` has a particular structure.

 * CVE-2025-46334, Git GUI (Windows only):

   A malicious repository can ship versions of sh.exe or typical
   textconv filter programs such as astextplain. On Windows, path
   lookup can find such executables in the worktree. These programs
   are invoked when the user selects "Git Bash" or "Browse Files" from
   the menu.

 * CVE-2025-46835, Git GUI:

   When a user clones an untrusted repository and is tricked into
   editing a file located in a maliciously named directory in the
   repository, then Git GUI can create and overwrite any writable
   file.

 * CVE-2025-48384, Git:

   When reading a config value, Git strips any trailing carriage
   return and line feed (CRLF). When writing a config entry, values
   with a trailing CR are not quoted, causing the CR to be lost when
   the config is later read.  When initializing a submodule, if the
   submodule path contains a trailing CR, the altered path is read
   resulting in the submodule being checked out to an incorrect
   location. If a symlink exists that points the altered path to the
   submodule hooks directory, and the submodule contains an executable
   post-checkout hook, the script may be unintentionally executed
   after checkout.

 * CVE-2025-48385, Git:

   When cloning a repository Git knows to optionally fetch a bundle
   advertised by the remote server, which allows the server-side to
   offload parts of the clone to a CDN. The Git client does not
   perform sufficient validation of the advertised bundles, which
   allows the remote side to perform protocol injection.

   This protocol injection can cause the client to write the fetched
   bundle to a location controlled by the adversary. The fetched
   content is fully controlled by the server, which can in the worst
   case lead to arbitrary code execution.

 * CVE-2025-48386, Git:

   The wincred credential helper uses a static buffer (`target`) as a
   unique key for storing and comparing against internal storage. This
   credential helper does not properly bounds check the available
   space remaining in the buffer before appending to it with
   `wcsncat()`, leading to potential buffer overflows.

----------------------------------------------------------------

Changes since v2.50.0 are as follows:

Avi Halachmi (:avih) (1):
      gitk: encode arguments correctly with "open"

Johannes Sixt (27):
      gitk: treat file names beginning with "|" as relative paths
      gitk: have callers of diffcmd supply pipe symbol when necessary
      gitk: sanitize 'exec' arguments: simple cases
      gitk: sanitize 'exec' arguments: 'eval exec'
      gitk: sanitize 'exec' arguments: redirections
      gitk: sanitize 'exec' arguments: redirections and background
      gitk: sanitize 'exec' arguments: redirect to process
      gitk: sanitize 'open' arguments: simple commands
      gitk: sanitize 'open' arguments: simple commands with redirections
      gitk: sanitize 'open' arguments: simple commands, readable and writable
      gitk: collect construction of blameargs into a single conditional
      gitk: sanitize 'open' arguments: command pipeline
      git-gui: remove special treatment of Windows from open_cmd_pipe
      git-gui: remove git config --list handling for git < 1.5.3
      git-gui: treat file names beginning with "|" as relative paths
      git-gui: sanitize 'exec' arguments: simple cases
      git-gui: sanitize 'exec' arguments: background
      git-gui: remove option --stderr from git_read
      git-gui: break out a separate function git_read_nice
      git-gui: use git_read in githook_read
      gitk: sanitize 'open' arguments: revisit recently updated 'open' calls
      git-gui: convert git_read*, git_write to be non-variadic
      git-gui: pass redirections as separate argument to _open_stdout_stderr
      git-gui: pass redirections as separate argument to git_read
      git-gui: introduce function git_redir for git calls with redirections
      git-gui: do not mistake command arguments as redirection operators
      git-gui: sanitize 'exec' arguments: convert new 'cygpath' calls

Junio C Hamano (2):
      Git 2.49.1
      Git 2.50.1

Justin Tobler (1):
      config: quote values containing CR character

Mark Levedahl (13):
      git-gui: _which, only add .exe suffix if not present
      git-gui: use [is_Windows], not bad _shellpath
      git-gui: make _shellpath usable on startup
      git-gui: remove Tcl 8.4 workaround on 2>@1 redirection
      git-gui: use only the configured shell
      git-gui: remove HEAD detachment implementation for git < 1.5.3
      git-gui: remove unused proc is_shellscript
      git-gui: avoid auto_execok for git-bash menu item
      git-gui: avoid auto_execok in do_windows_shortcut
      git-gui: cleanup git-bash menu item
      git-gui: assure PATH has only absolute elements.
      git-gui: sanitize $PATH on all platforms
      git-gui: override exec and open only on Windows

Patrick Steinhardt' via Git Security (1):
      bundle-uri: fix arbitrary file writes via parameter injection

Taylor Blau (7):
      wincred: avoid buffer overflow in wcsncat()
      Git 2.43.7
      Git 2.44.4
      Git 2.45.4
      Git 2.46.4
      Git 2.47.3
      Git 2.48.2

Junio C Hamano wrote:

I was building v2.45.4 on an older system and two tests failed:

    t1300.237 writing value with trailing CR not stripped on read
    t7450.50  submodule must not checkout into different directory

These both look like they use new git config syntax (get, set, unset) which is not available in the older releases.

I haven't done enough digging to determine when the syntax changed to prepare a proper patch, but I thought I'd mention it, in case anyone else runs into a similar issue or is highly motivated to work up a patch against which ever older releases need it.

The fix looks something like this, I believe:

-- >8 --

In a recent security release, 05e9cd64ee (config: quote values containing CR character, 2025-05-19) added calls to `git config get`, `git config set`, and `git config unset` which are not present on the maint-2.43 branch.

These subcommands were added in the following commits, released in git-2.46.0:

  4e51389000 (builtin/config: introduce "get" subcommand, 2024-05-06),
  00bbdde141 (builtin/config: introduce "set" subcommand, 2024-05-06),
  95ea69c67b (builtin/config: introduce "unset" subcommand, 2024-05-06)

Revert to the previous `git config` syntax for older maintenance branches.

Signed-off-by: Todd Zullinger <tmz@pobox.com>
---

This is based on maint-2.43. I tested that it merges cleanly to 2.44 and 2.45. I have only run a build with 2.45.4 so far. I pushed this to a github fork, which looks like it will run the CI for each branch:

    https://github.com/tmzullinger/git/actions

 t/t1300-config.sh           | 4 ++--
 t/t7450-bad-git-dotfiles.sh | 4 ++--
 2 files changed, 4 insertions(+), 4 deletions(-)

Todd Zullinger <tmz@pobox.com> writes:

Thanks, will queue.

On 25/07/08 05:05PM, Todd Zullinger wrote:

Thanks for raising this fix! Apologies, it slipped my mind that the config subsommands wouldn't be available on older versions.

The changes here look good to me. :)

-Justin

Justin Tobler wrote:

No worries.  Thanks for working on these CVE fixes!

I thought that the CI might catch those failures, but I imagine they slipped past others as easily as some other failures slipped past me last night.

There are unrelated test failures on those older maint branches which make noticing any new test failures less likely. Even more so when you're working on a security issue and there is less time and fewer eyes on some of the less important details. :)

I looked at a couple of the failed jobs at the time and they were unrelated to the tests I'd changed. I incorrectly presumed that was the reason for all of the failures.

A few (linux-leaks (ubuntu-latest), linux-reftable-leaks (ubuntu-latest) failed t7450.50 (submodule must not checkout into different directory) with:

  + printf sub\r
  + git -C repo mv sub sub
  Aborted (core dumped)
  error: last command exited with $?=134
  + rm -rf sub repo bad-clone
  + exit 134
  + eval_ret=134
  + :
  not ok 50 - submodule must not checkout into different directory

I don't know if that is worth looking at, given the age of these maint branches.

While I was poking though, here are the other failures from the maint-2.45 branch runs at GitHub:

Several jobs (linux-musl (alpine), osx-gcc (macos-13)) failed in imap-send.c and http.c, with:

    call to '_curl_easy_setopt_err_long' declared with
    attribute warning: curl_easy_setopt expects a long
    argument [-Werror=attribute-warning]

I think this has been fixed in later releases. It rings a tiny bell, anyway.

The 'pedantic (fedora)' and 'win build' jobs both failed with:

  Error: object-file.c:58:9: initializer-string for array of
  'unsigned char' truncates NUL terminator but destination
  lacks 'nonstring' attribute (33 chars into 32 available)
  [-Werror=unterminated-string-initialization]
     58 |         "\x47\x3a\x0f\x4c\x3b\xe8\xa9\x36\x81\xa2" \
        |         ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
  object-file.c:80:17: note: in expansion of macro 'EMPTY_BLOB_SHA256_BIN_LITERAL'
     80 |         .hash = EMPTY_BLOB_SHA256_BIN_LITERAL,
        |                 ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~

which I presume is also fixed in later releases and may not be worth fixing on these old maintenance branches.

Thanks for the extra eyes.