NTLM buffer overflow via integer overflow
Project curl Security Advisory, November 29th 2017 - Permalink
libcurl contains a buffer overrun flaw in the NTLM authentication code.
The internal function
Curl_ntlm_core_mk_ntlmv2_hash sums up the lengths of
the user name + password (= SUM) and multiplies the sum by two (= SIZE) to
figure out how large storage to allocate from the heap.
The SUM value is subsequently used to iterate over the input and generate
output into the storage buffer. On systems with a 32 bit
size_t, the math to
calculate SIZE triggers an integer overflow when the combined lengths of the
user name and password is larger than 2GB (2^31 bytes). This integer overflow
usually causes a very small buffer to actually get allocated instead of the
intended very huge one, making the use of that buffer end up in a buffer
We are not aware of any exploit of this flaw.
This bug was introduced in commit 86724581b6c02d160b5, January 2014.
The Common Vulnerabilities and Exposures (CVE) project has assigned the name CVE-2017-8816 to this issue.
CWE-131: Incorrect Calculation of Buffer Size
This is only an issue on 32 bit systems. It also requires the user and password fields to use more than 2GB of memory combined, which in itself should be rare.
- Affected versions: libcurl 7.36.0 to and including 7.56.1
- Not affected versions: libcurl < 7.36.0 and >= 7.57.0
curl is used by many applications, but not always advertised as such.
In libcurl version 7.57.0, the integer overflow is avoided.
A patch for CVE-2017-8816 is available.
We suggest you take one of the following actions immediately, in order of preference:
A - Upgrade curl to version 7.57.0
B - Apply the patch to your version and rebuild
C - Put length restrictions on the user name and passwords you can pass to libcurl
It was reported to the curl project on November 6, 2017. We contacted distros@openwall on November 21.
curl 7.57.10 was released on November 29 2017, coordinated with the publication of this advisory.
Reported by Alex Nichols. Patch by Daniel Stenberg.
Thanks a lot!