CVE-2014-7844

BSD mailx 8.1.2 and earlier allows remote attackers to execute arbitrary commands via a crafted email address.

Updated nail package fixes security vulnerabilities:

A flaw was found in the way mailx handled the parsing of email addresses. A syntactically valid email address could allow a local attacker to cause mailx to execute arbitrary shell commands through shell meta-characters and the direct command execution functionality (CVE-2004-2771, CVE-2014-7844).

These are all security issues fixed in the mailx-12.5-24.4 package on the GA media of openSUSE Tumbleweed.

This update for mailx enables IPv6 support and includes the following fixes:

* Crop off the brackets of an ipv6 address if found. (bnc#853246)
* Enable mailx to parse IPv6 addresses including a port ([ipv6]:port).
  (bnc#853246)
* Do not pseudo detect Latin nor UTF-8 in binary attachments.
  (bnc#827010)

It was discovered that bsd-mailx contained a feature that allowed syntactically valid email addresses to be treated as shell commands. A remote attacker could possibly use this issue with a valid email address to execute arbitrary commands.

This functionality has now been disabled by default, and can be re-enabled with the "expandaddr" configuration option. This update alone does not remove all possibilities of command execution. In environments where scripts use mailx to process arbitrary email addresses, it is recommended to modify them to use a "--" separator before the address to properly handle those that begin with "-". In addition, specifying sendmail options after the "--" separator is no longer supported, existing scripts may need to be modified to use the "-a" option instead.