[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[FD] CVE-2017-5344 : dotCMS Blind Boolean SQL Injection in dotCMS <= 3.6.1

# Blind Boolean SQL Injection in dotCMS <= 3.6.1 (CVE-2017-5344)

## Product Description

dotCMS is a scalable, java based, open source content management system
(CMS) that has been designed to manage and deliver personalized, permission
based content experiences across multiple channels. dotCMS can serve as the
plaform for sites, mobile apps, mini-sites, portals, intranets or as a
headless CMS (content is consumed via RESTful APIs). dotCMS is used
everywhere, from running small sites to powering multi-node installations
for governemnts, Fortune 100 companies, Universities and Global Brands. A
dotCMS environment can scale to support hundreds of editors managing
thousands of sites with millions of content objects.

## Vulnerability Type

Blind Boolean SQL injection

## Vulnerability Description

dotCMS versions up to 3.6.1 (and possibly others) are vulnerable to blind
boolean SQL injection in the q and inode parameters at the
/categoriesServlet path. This servlet is a remotely accessible,
unauthenticated function of default dotCMS installations and can be
exploited to exfiltrate sensitive information from databases accessible to
the DMBS user configured with the product.

Exploitation of the vulnerability is limited to the MySQL DMBS in 3.5 -
3.6.1 as SQL escaping controls were added to address a similar
vulnerability discovered in previous versions of the product. The means of
bypassing these features which realise this vulnerability have only been
successfully tested with MySQL 5.5, 5.6 and 5.7 and it is believed other
DMBS's are not affected. Versions prior to 3.6 do not have these controls
and can be exploited directly on a greater number of paired DMBS's.
PostgreSQL is vulnerable in all described versions of dotCMS when
PostgreSQL standard_confirming_strings setting is disabled (enabled by

The vulnerability is the result of string interpolation and directly SQL
statement execution without sanitising user input. The intermediate
resolution for a previous SQLi vulnerability was to whitelist and partially
filter user input before interpolation. This vulnerability overcomes this
filtering to perform blind boolean SQL injection. The resolution to this
vulnerability was to implement the use of prepared statements in the
affected locations.

This vulnerability has been present in dotCMS since at least since version

## Exploit

A proof of concept is available here:

## Versions

dotCMS <= 3.3.2 and MYSQL, MSSQL, H2, PostgreSQL

dotCMS 3.5 - 3.6.1 and (MYSQL or PostgreSQL w/ standard_confirming_strings

## Attack Type

Unauthenticated, Remote

## Impact

The SQL injection vulnerability can be used to exfiltrate sensitive
information from the DBMS used with dotCMS. Depending of the DBMS
configuration and type, the issue could be as severe as establishing a
remote shell (such as by using xp_exec on MSSQL servers) or in the most
limited cases, restricted only to exfiltration of data in dotCMS database

## Credit

This vulnerability was discovered by Ben Nott <pajexali AT gmail.com>.

Credit goes to Erlar Lang for discovering similar SQL injection
vulnerabilities in nearby code and for inspiring this discovery.

## Disclosure Timeline

  * Jan 2, 2017 - Issue discovered.
  * Jan 2, 2017 - Vendor advised of discovery and contact requested for
full disclosure.
  * Jan 4, 2017 - Provided full disclosure to vendor.
  * Jan 5, 2017 - Vendor acknowledged disclosure and confirmed finding
  * Jan 14, 2017 - Vendor advised patch developed and preparing for release.
  * Jan 24, 2017 - Vendor advised patching in progress.
  * Feb 15, 2017 - Vendor advises ready for public disclosure.

## References

Vendor advisory: http://dotcms.com/security/SI-39
CVE: http://cve.mitre.org/cgi-bin/cvename.cgi?name=2017-5344

Sent through the Full Disclosure mailing list
Web Archives & RSS: http://seclists.org/fulldisclosure/