The pingback system is an optional mechanism which reports statistical data to civicrm.org. The pingback URL specified an unencrypted protocol (HTTP), and well-positioned eavesdropper could potentially intercept statistical data. The pingback URL should specify an encrypted protocol (HTTPS) to prevent eavesdropping.
To receive future CiviCRM security notices, subscribe to our notifications. Check here for details of our security policy and how to report a suspected security issue.
After successfully calling the "Contact.create" API, the caller could receive a list of all fields relating to the contact -- including a sensitive field that normally has restricted access. In some contexts, leaking the sensitive field could allow an attacker to access CiviCRM as the targeted user.
In the "Recently Viewed" block, the title field of the hyperlink was not properly escaped.
In CiviContribute forms which combine the "On Behalf Of" feature with "Organization" records, some data was not properly escaped.
In the "Search Results" screen, some elements were not properly escaped.
Drupal Views allows an administrator to produce a screen with data from CiviCRM's custom-fields. Certain custom-fields could potentially be manipulated to inject SQL.
When viewing the CiviCRM "Mailing" report, a logged-in user could modify the URL to access the report for another mailing -- even if
they ordinarily would not have access that information.
Unfortunately, we could not obtain sufficient information about these issues to determine whether they cause actual vulnerabilities in CiviCRM.
When displaying entity reference fields, the labels were not properly being escaped.
This issue affects your site if it is hosted on WordPress, and you use ACLs to restrict access to contact data.
It was identified that CiviCRM on WordPress CMS did not correctly trigger ACL checks when viewing CiviCRM profile URLs via checksum. This might lead sites to disclose some contact data via profile pages.
When generating an API query, the ORDER BY clause for some entities was not correctly validated and escaped. This may have permitted data disclosure via time-based blind SQL attacks.
This is mitigated by the fact that attacks would require API access to exploit the vulnerability.
It was identified that inputs were not correctly validated when viewing an activity related to a case, due to custom group title not being properly escaped for SQL generation.
This is mitigated by the fact that an attacker would need to have the "administer CiviCRM" permission, and that the issue only affects sites with CiviCase enabled.
It was identified that CRM_Contact_BAO_Query::apiQuery did not correctly validate contact ID inputs. This could expose contact data via SQL injection.
This is mitigated by permissions restrictions meaning that anonymous users would not typically be able to exploit this vector.
Sites which use the Drupal 6 "devel" module with CiviCRM to log SQL queries may be vulnerable to a SQL injection. However, it is not clear if this vulnerability is exploitable.
CiviCRM allows administrators to define custom profile-forms in which constituents enter their names, addresses, custom data, etc. CiviCRM is designed to embed all its forms within a CMS (such as Drupal, Joomla, or WordPress), but some administrators also need to embed profile-forms in an external site or custom HTML document. This has sometimes been accomplished with an "HTML Snippet" technique -- the bare, literal HTML code for a profile-form is manually copied and pasted to an external web site.
The CiviCRM log file is stored in data folder determined by the CMS. In all supported CMS's, this data folder defaults to world-readable, but CiviCRM needs to store logs confidentially. CiviCRM relies on two redundant protections to ensure that log files remain confidential:
CiviCRM includes a handful of backend scripts (bin/migrate/*.php and bin/encryptDB.php) which facilitate some special workflows (such as migrating site-configurations and obfuscating the database). These scripts include security protections, but -- depending on your organizational policies -- these protections may be inadequate. CiviCRM v4.7.11+ tightens access to these scripts.
Who is impacted?
In older versions, the security of these scripts rests on three things: a username, a password, and the site-key.
An automated security audit (based on static code analysis of the CiviCRM codebase) indicated that a dependency (PEAR CLI from the "packages" folder) could potentially reveal semi-sensitive backtrace data if an attacker could run it and provoke an error.
An exploit of this has not been identified.
As a precautionary measure, CiviCRM v4.7.11 removes PEAR CLI.
CiviCRM allows users to import contacts using CSV or SQL. Prior to 4.7.11 (or 4.6.21), the permission "import contacts" allowed users to import by any means -- either CSV or SQL. A user with this permission could use it to bypass ACL rules. Beginning with 4.7.11+ (or 4.6.21+), there is now a separate permission "import SQL datasource". If you want your users to be able to import contacts using SQL, you must now grant both permissions ("import contacts" and "import SQL datasource").
CiviCRM previously did not set secure flags to restrict cookies to SSL where appropriate. This was not a security risk by itself, but the change is being made and notified in security release information as part of a wider "defense in depth" process within CiviCRM.
A SQL injection vulnerability in CiviCRM's API was identified, where an API parameter was identified as being passed directly to SQL.
This is mitigated by the fact that the remote user must have some elevated permissions to exploit the vulnerability. CiviCRM recommends that all sites upgrade to obtain this and other recent fixes.
An access bypass was identified where if a user was permitted only the "View own contact" permission in the CMS, they were also able to edit their own contact. This bypass of permissions checking did not extend to other contacts in CiviCRM.
A potential for information disclosure was identified in a packaged library, HTML TreeBuilder.
CiviCRM now patches the TreeBuilder library to direct debug output to the CiviCRM debug log, rather than to screen.
This release addresses an issue where it was possible to deliver a cross-site scripting attack through the CiviCRM backend.
To exploit this vulnerability, both the attacker and victim need permission to access the CiviCRM backend, and the victim must visit a specific screen.
For more information about this type of vulnerability, see OWASP's page on Cross Site Scripting.
Multiple SQL injections have been identified in AJAX helpers supporting backend forms. An exploit has been demonstrated. Executing an exploit requires a user account with some kind of CiviCRM permission (such as "access CiviCRM" or "view my contact").