CVE-2016-9586 log

Source
Severity Medium
Remote Yes
Type Arbitrary code execution
Description
libcurl's implementation of the printf() functions triggers a buffer overflow when doing a large floating point output. The bug occurs when the conversion outputs more than 255 bytes. The flaw happens because the floating point conversion is using system functions without the correct boundary checks.
The functions have been documented as deprecated for a long time and users are discouraged from using them in "new programs" as they are planned to get removed at a future point. But as the functions are present and there's nothing preventing users from using them, we expect there to be a certain amount of existing users in the wild.
If there are any application that accepts a format string from the outside without necessary input filtering, it could allow remote attacks.
Group Package Affected Fixed Severity Status Ticket
AVG-117 lib32-libcurl-gnutls 7.51.0-1 7.52.1-1 Medium Fixed FS#52247
AVG-116 lib32-libcurl-compat 7.51.0-1 7.52.1-1 Medium Fixed FS#52247
AVG-115 lib32-curl 7.51.0-1 7.52.1-1 Medium Fixed FS#52247
AVG-114 libcurl-gnutls 7.51.0-1 7.52.1-1 Medium Fixed FS#52247
AVG-113 libcurl-compat 7.51.0-1 7.52.1-1 Medium Fixed FS#52247
AVG-112 curl 7.51.0-1 7.52.1-1 Medium Fixed FS#52247
Date Advisory Group Package Severity Description
03 Jan 2017 ASA-201701-9 AVG-115 lib32-curl Medium multiple issues
03 Jan 2017 ASA-201701-8 AVG-114 libcurl-gnutls Medium multiple issues
03 Jan 2017 ASA-201701-7 AVG-113 libcurl-compat Medium multiple issues
03 Jan 2017 ASA-201701-11 AVG-117 lib32-libcurl-gnutls Medium multiple issues
03 Jan 2017 ASA-201701-10 AVG-116 lib32-libcurl-compat Medium multiple issues
27 Dec 2016 ASA-201612-22 AVG-112 curl Medium multiple issues
References
https://curl.haxx.se/docs/adv_20161221A.html
Notes
This flaw is present in the curl_*printf (curlx_*printf) family of functions, which are not used by curl but are exposed from libcurl.
To be exposed, third-party code would need to be using these long-deprecated functions, with a floating-point specifier and user-controlled (floating-point) input. The overflow itself is of a 256-byte stack-allocated buffer, when the decimal expansion of the float exceeds that by up to 70 bytes. Beyond about 16 digits for a double, the decimal expansion is effectively random so the attacker has very little control over precisely what bytes are written.