Atlassian Confluence < 5.10.6 - Persistent Cross-Site Scripting

  • 作者： Jodson Santos
    日期： 2017-01-04
  • 类别：
    平台：
  • 来源：https://www.exploit-db.com/exploits/40989/
  • =====[ Tempest Security Intelligence - ADV-3/2016 CVE-2016-6283 ]==============

Persisted Cross-Site Scripting (XSS) in Confluence Jira Software
----------------------------------------------------------------

Author(s):
- Jodson Santos
- jodson.santos@tempest.com.br

Tempest Security Intelligence - Recife, Pernambuco - Brazil

=====[Table of Contents]=====================================================

1. Overview
2. Detailed description
3. Affected versions & Solutions
4. Timeline of disclosure
5. Thanks & Acknowledgements
6. References

=====[1. Overview]============================================================

 * System affected: Atlassian Confluence
 * Software Version : 5.9.12
Other versions or models may also be affected.
 * Impact : This vulnerability allows an attacker to use
Confluence's
platform to deliver attacks against other users.

=====[2. Detailed description]================================================

Atlassian Confluence version 5.9.12 is vulnerable to persistent cross-site
scripting (XSS) because it fails to securely validate user controlled data,
thus making it possible for an attacker to supply crafted input in order to
harm users. The bug occurs at pages carrying attached files, even though
the attached file name parameter is correctly sanitized upon submission, it is
possible for an attacker to later edit the attached file name property and
supply crafted data (i.e HTML tags and script code) without the
occurrence of any security checks, resulting in an exploitable persistent XSS.

In order to reproduce the vulnerability, go to a page with an attached
file, click on "Attachments" in order to list the page's attachments, and then
click on "Properties" for the file of your choice. Edit the file name to, for
example, <script>alert(1)</script>test.pdf and then save the changes.
Albeit the XSS is not executed within the page display, it is possible to
trigger the execution of the supplied code while performing a search within
Confluence in which results include the attachment with crafted file name. For that
matter, the search terms " or * will promptly display the file and execute the
injected javascript code.

As a means to further enlighten this, the following excerpt demonstrates
a POST request with the malicious insertion within the newFileName field:

POST
/pages/doeditattachment.action?pageId={pageId}&attachmentBean.fileName={filename} HTTP/1.1
Host: {confluence host}
Cookie: mywork.tab.tasks=false; JSESSIONID={redacted};
confluence.browse.space.cookie=space-templates
Connection: keep-alive
Content-Type: application/x-www-form-urlencoded
Content-Length: {redacted}

atl_token={atl_token}&pageId={pageId}&isFromPageView=false&newFileName=<script>alert(1)</script>file&newComment=&newContentType=application%2Foctet-stream&newParentPage=&confirm=Save

It is worth noting that the issue may affect users regardless of privilege
levels, since the malicious page/attachment can be browsed by any user
within the Atlassian Confluence instance.

=====[3. Affected versions & Solutions]=======================================

This test was performed against Atlassian Confluence version 5.9.12.

According to vendor's response, the vulnerability is addressed and the
fix is part of the 5.10.6 release.

=====[4. Timeline of disclosure]==============================================

Jul/07/2016 - Vendor acknowledged the vulnerability.
Aug/04/2016 - Vendor released the fix for the vulnerability in version 5.10.6.

=====[5. Thanks & Acknowledgements]===========================================

- Tempest Security Intelligence / Tempest's Pentest Team [1]
- Joaquim Brasil
- Heyder Andrade
- Breno Cunha

=====[6. References]==========================================================

[1] https://en.wikipedia.org/wiki/Confluence_(software)