CVE-2020-12137

GNU Mailman 2.x before 2.1.30 uses the .obj extension for scrubbed application/octet-stream MIME parts. This behavior may contribute to XSS attacks against list-archive visitors, because an HTTP reply from an archive web server may lack a MIME type, and a web browser may perform MIME sniffing, conclude that the MIME type should have been text/html, and execute JavaScript code.

Updated mailman package fixes security vulnerability:

Up to mailman 2.1.29 when sending a file without a file extension (or an unknown file extension) then the file is stored in the list archive with the file extension .obj. Most web servers will try to assign a mime type based on the file extension and entries in /etc/mime.types, where .obj is usually not specified. This means the web server will send it out without a mime type. The browser will then try to guess the MIME type based on the file's content (MIME-sniffing). If the content is HTML then it will execute any javascript contained, leading to a potential cross-site scripting vulnerability.

The mailman package has been updated to version 2.1.30, fixing this bug and other issues. See the release announcement for details.

This update for mailman to version 2.1.34 fixes the following issues:

• The fix for lp#1859104 can result in ValueError being thrown on attempts to subscribe to a list. This is fixed and extended to apply REFUSE_SECOND_PENDING to unsubscription as well. (lp#1878458)
• DMARC mitigation no longer misses if the domain name returned by DNS contains upper case. (lp#1881035)
• A new WARN_MEMBER_OF_SUBSCRIBE setting can be set to No to prevent mailbombing of a member of a list with private rosters by repeated subscribe attempts. (lp#1883017)
• Very long filenames for scrubbed attachments are now truncated. (lp#1884456)
• A content injection vulnerability via the private login page has been fixed. CVE-2020-15011 (lp#1877379, bsc#1173369)
• A content injection vulnerability via the options login page has been discovered and reported by Vishal Singh. CVE-2020-12108 (lp#1873722, bsc#1171363)
• Bounce recognition for a non-compliant Yahoo format is added.
• Archiving workaround for non-ascii in string.lowercase in some Python packages is added.
• Thanks to Jim Popovitch, there is now a dmarc_moderation_addresses list setting that can be used to apply dmarc_moderation_action to mail From: addresses listed or matching listed regexps. This can be used to modify mail to addresses that don't accept external mail From: themselves.
• There is a new MAX_LISTNAME_LENGTH setting. The fix for lp#1780874 obtains a list of the names of all the all the lists in the installation in order to determine the maximum length of a legitimate list name. It does this on every web access and on sites with a very large number of lists, this can have performance implications. See the description in Defaults.py for more information.
• Thanks to Ralf Jung there is now the ability to add text based captchas (aka textchas) to the listinfo subscribe form. See the documentation for the new CAPTCHA setting in Defaults.py for how to enable this. Also note that if you have custom listinfo.html templates, you will have to add a <mm-captcha-ui> tag to those templates to make this work. This feature can be used in combination with or instead of the Google reCAPTCHA feature added in 2.1.26.
• Thanks to Ralf Hildebrandt the web admin Membership Management section now has a feature to sync the list's membership with a list of email addresses as with the bin/sync_members command.
• There is a new drop_cc list attribute set from DEFAULT_DROP_CC. This controls the dropping of addresses from the Cc: header in delivered messages by the duplicate avoidance process. (lp#1845751)
• There is a new REFUSE_SECOND_PENDING mm_cfg.py setting that will cause a second request to subscribe to a list when there is already a pending confirmation for that user. This can be set to Yes to prevent mailbombing of a third party by repeatedly posting the subscribe form. (lp#1859104)
• Fixed the confirm CGI to catch a rare TypeError on simultaneous confirmations of the same token. (lp#1785854)
• Scrubbed application/octet-stream MIME parts will now be given a .bin extension instead of .obj. CVE-2020-12137 (lp#1886117)
• Added bounce recognition for a non-compliant opensmtpd DSN with Action: error. (lp#1805137)
• Corrected and augmented some security log messages. (lp#1810098)
• Implemented use of QRUNNER_SLEEP_TIME for bin/qrunner --runner=All. (lp#1818205)
• Leading/trailing spaces in provided email addresses for login to private archives and the user options page are now ignored. (lp#1818872)
• Fixed the spelling of the --no-restart option for mailmanctl.
• Fixed an issue where certain combinations of charset and invalid characters in a list's description could produce a List-ID header without angle brackets. (lp#1831321)
• With the Postfix MTA and virtual domains, mappings for the site list -bounces and -request addresses in each…

This update for mailman to version 2.1.34 fixes the following issues:

• The fix for lp#1859104 can result in ValueError being thrown on attempts to subscribe to a list. This is fixed and extended to apply REFUSE_SECOND_PENDING to unsubscription as well. (lp#1878458)
• DMARC mitigation no longer misses if the domain name returned by DNS contains upper case. (lp#1881035)
• A new WARN_MEMBER_OF_SUBSCRIBE setting can be set to No to prevent mailbombing of a member of a list with private rosters by repeated subscribe attempts. (lp#1883017)
• Very long filenames for scrubbed attachments are now truncated. (lp#1884456)
• A content injection vulnerability via the private login page has been fixed. CVE-2020-15011 (lp#1877379, bsc#1173369)
• A content injection vulnerability via the options login page has been discovered and reported by Vishal Singh. CVE-2020-12108 (lp#1873722, bsc#1171363)
• Bounce recognition for a non-compliant Yahoo format is added.
• Archiving workaround for non-ascii in string.lowercase in some Python packages is added.
• Thanks to Jim Popovitch, there is now a dmarc_moderation_addresses list setting that can be used to apply dmarc_moderation_action to mail From: addresses listed or matching listed regexps. This can be used to modify mail to addresses that don't accept external mail From: themselves.
• There is a new MAX_LISTNAME_LENGTH setting. The fix for lp#1780874 obtains a list of the names of all the all the lists in the installation in order to determine the maximum length of a legitimate list name. It does this on every web access and on sites with a very large number of lists, this can have performance implications. See the description in Defaults.py for more information.
• Thanks to Ralf Jung there is now the ability to add text based captchas (aka textchas) to the listinfo subscribe form. See the documentation for the new CAPTCHA setting in Defaults.py for how to enable this. Also note that if you have custom listinfo.html templates, you will have to add a <mm-captcha-ui> tag to those templates to make this work. This feature can be used in combination with or instead of the Google reCAPTCHA feature added in 2.1.26.
• Thanks to Ralf Hildebrandt the web admin Membership Management section now has a feature to sync the list's membership with a list of email addresses as with the bin/sync_members command.
• There is a new drop_cc list attribute set from DEFAULT_DROP_CC. This controls the dropping of addresses from the Cc: header in delivered messages by the duplicate avoidance process. (lp#1845751)
• There is a new REFUSE_SECOND_PENDING mm_cfg.py setting that will cause a second request to subscribe to a list when there is already a pending confirmation for that user. This can be set to Yes to prevent mailbombing of a third party by repeatedly posting the subscribe form. (lp#1859104)
• Fixed the confirm CGI to catch a rare TypeError on simultaneous confirmations of the same token. (lp#1785854)
• Scrubbed application/octet-stream MIME parts will now be given a .bin extension instead of .obj. CVE-2020-12137 (lp#1886117)
• Added bounce recognition for a non-compliant opensmtpd DSN with Action: error. (lp#1805137)
• Corrected and augmented some security log messages. (lp#1810098)
• Implemented use of QRUNNER_SLEEP_TIME for bin/qrunner --runner=All. (lp#1818205)
• Leading/trailing spaces in provided email addresses for login to private archives and the user options page are now ignored. (lp#1818872)
• Fixed the spelling of the --no-restart option for mailmanctl.
• Fixed an issue where certain combinations of charset and invalid characters in a list's description could produce a List-ID header without angle brackets. (lp#1831321)
• With the Postfix MTA and virtual domains, mappings for the site list -bounces and -request addresses in each…

This update for mailman fixes the following issues:

Security issue fixed:

• CVE-2020-12108: Fixed a content injection bug (bsc#1171363).
• CVE-2020-12137: Fixed a XSS vulnerability caused by MIME type confusion (bsc#1170558).

Non-security issue fixed:

• Fixed rights and ownership on /var/lib/mailman/archives (bsc#1167068).
• Don't default to invalid hosts for DEFAULT_EMAIL_HOST (bsc#682920).

This update for mailman fixes the following issues:

Security issue fixed:

• CVE-2020-12137: Fixed a XSS vulnerability caused by MIME type confusion (bsc#1170558).

Non-security issue fixed:

• Fixed rights and ownership on /var/lib/mailman/archives (bsc#1167068).

It was discovered that Mailman incorrectly handled certain inputs. An attacker could possibly use this to issue execute arbitrary scripts or HTML. (CVE-2018-0618)

It was discovered that Mailman incorrectly handled certain inputs. An attacker could possibly use this issue to display arbitrary text on a web page. (CVE-2018-13796)

It was discovered that Mailman incorrectly handled certain files. An attacker could possibly use this issue to execute arbitrary code. (CVE-2020-12137)

USN-5009-1 fixed vulnerabilities in Mailman. This update provides the corresponding updates for Ubuntu 20.04 LTS. In addition, the following CVEs were fixed:

It was discovered that Mailman allows arbitrary content injection. An attacker could use this to inject malicious content. (CVE-2020-12108, CVE-2020-15011)

It was discovered that Mailman improperly sanitize the MIME content. An attacker could obtain sensitive information by sending a special type of attachment. (CVE-2020-12137)

Original advisory details:

Andre Protas, Richard Cloke, and Andy Nuttall discovered that Mailman did not properly associate cross-site request forgery (CSRF) tokens to specific accounts. A remote attacker could use this to perform a CSRF attack to gain access to another account. (CVE-2021-42097)

Andre Protas, Richard Cloke, and Andy Nuttall discovered that Mailman’s cross-site request forgery (CSRF) tokens for the options page are derived from the admin password. A remote attacker could possibly use this to assist in performing a brute force attack against the admin password. (CVE-2021-42096)