Vulnerabilities > CVE-2023-25815 - Use of Externally-Controlled Format String vulnerability in multiple products
Summary
In Git for Windows, the Windows port of Git, no localized messages are shipped with the installer. As a consequence, Git is expected not to localize messages at all, and skips the gettext initialization. However, due to a change in MINGW-packages, the `gettext()` function's implicit initialization no longer uses the runtime prefix but uses the hard-coded path `C:\mingw64\share\locale` to look for localized messages. And since any authenticated user has the permission to create folders in `C:\` (and since `C:\mingw64` does not typically exist), it is possible for low-privilege users to place fake messages in that location where `git.exe` will pick them up in version 2.40.1. This vulnerability is relatively hard to exploit and requires social engineering. For example, a legitimate message at the end of a clone could be maliciously modified to ask the user to direct their web browser to a malicious website, and the user might think that the message comes from Git and is legitimate. It does require local write access by the attacker, though, which makes this attack vector less likely. Version 2.40.1 contains a patch for this issue. Some workarounds are available. Do not work on a Windows machine with shared accounts, or alternatively create a `C:\mingw64` folder and leave it empty. Users who have administrative rights may remove the permission to create folders in `C:\`.
Vulnerable Configurations
Common Weakness Enumeration (CWE)
Common Attack Pattern Enumeration and Classification (CAPEC)
- Format String Injection An attacker includes formatting characters in a string input field on the target application. Most applications assume that users will provide static text and may respond unpredictably to the presence of formatting character. For example, in certain functions of the C programming languages such as printf, the formatting character %s will print the contents of a memory location expecting this location to identify a string and the formatting character %n prints the number of DWORD written in the memory. An attacker can use this to read or write to memory locations or files, or simply to manipulate the value of the resulting text in unexpected ways. Reading or writing memory may result in program crashes and writing memory could result in the execution of arbitrary code if the attacker can write to the program stack.
- String Format Overflow in syslog() This attack targets the format string vulnerabilities in the syslog() function. An attacker would typically inject malicious input in the format string parameter of the syslog function. This is a common problem, and many public vulnerabilities and associated exploits have been posted.
References
- https://github.com/git-for-windows/git/releases/tag/v2.40.1.windows.1
- https://axcheron.github.io/exploit-101-format-strings/#writing-to-the-stack
- https://github.com/msys2/MINGW-packages/pull/10461
- https://pubs.opengroup.org/onlinepubs/9699919799/functions/printf.html
- https://github.com/git-for-windows/git/security/advisories/GHSA-9w66-8mq8-5vm8
- http://www.openwall.com/lists/oss-security/2023/04/25/2
- https://lists.fedoraproject.org/archives/list/[email protected]/message/PI7FZ4NNR5S5J5K6AMVQBH2JFP6NE4L7/
- https://lists.fedoraproject.org/archives/list/[email protected]/message/YFZWGQKB6MM5MNF2DLFTD7KS2KWPICKL/
- https://lists.fedoraproject.org/archives/list/[email protected]/message/RKOXOAZ42HLXHXTW6JZI4L5DAIYDTYCU/
- https://security.gentoo.org/glsa/202312-15
- https://lists.debian.org/debian-lts-announce/2024/06/msg00018.html