10 - MDVSA-2012:110-1

Security issues were identified and fixed in mozilla firefox and thunderbird:

Mozilla developers identified and fixed several memory safety bugs in the browser engine used in Firefox and other Mozilla-based products. Some of these bugs showed evidence of memory corruption under certain circumstances, and we presume that with enough effort at least some of these could be exploited to run arbitrary code (CVE-2012-1949, CVE-2012-1948).

Security researcher Mario Gomes andresearch firm Code Audit Labs reported a mechanism to short-circuit page loads through drag and drop to the addressbar by canceling the page load. This causes the address of the previously site entered to be displayed in the addressbar instead of the currently loaded page. This could lead to potential phishing attacks on users (CVE-2012-1950).

Google security researcher Abhishek Arya used the Address Sanitizer tool to uncover four issues: two use-after-free problems, one out of bounds read bug, and a bad cast. The first use-after-free problem is caused when an array of nsSMILTimeValueSpec objects is destroyed but attempts are made to call into objects in this array later. The second use-after-free problem is in nsDocument::AdoptNode when it adopts into an empty document and then adopts into another document, emptying the first one. The heap buffer overflow is in ElementAnimations when data is read off of end of an array and then pointers are dereferenced. The bad cast happens when nsTableFrame::InsertFrames is called with frames in aFrameList that are a mix of row group frames and column group frames. AppendFrames is not able to handle this mix. All four of these issues are potentially exploitable (CVE-2012-1951, CVE-2012-1954, CVE-2012-1953, CVE-2012-1952).

Security researcher Mariusz Mlynski reported an issue with spoofing of the location property. In this issue, calls to history.forward and history.back are used to navigate to a site while displaying the previous site in the addressbar but changing the baseURI to the newer site. This can be used for phishing by allowing the user input form or other data on the newer, attacking, site while appearing to be on the older, displayed site (CVE-2012-1955).

Mozilla security researcher moz_bug_r_a4 reported a cross-site scripting (XSS) attack through the context menu using a data: URL. In this issue, context menu functionality (View Image, Show only this frame, and View background image) are disallowed in a javascript: URL but allowed in a data: URL, allowing for XSS. This can lead to arbitrary code execution (CVE-2012-1966).

Security researcher Mario Heiderich reported that javascript could be executed in the HTML feed-view using <embed> tag within the RSS <description>. This problem is due to <embed> tags not being filtered out during parsing and can lead to a potential cross-site scripting (XSS) attack. The flaw existed in a parser utility class and could affect other parts of the browser or add-ons which rely on that class to sanitize untrusted input (CVE-2012-1957).

Security researcher Arthur Gerkis used the Address Sanitizer tool to find a use-after-free in nsGlobalWindow::PageHidden when mFocusedContent is released and oldFocusedContent is used afterwards. This use-after-free could possibly allow for remote code execution (CVE-2012-1958).

Mozilla developer Bobby Holley found that same-compartment security wrappers (SCSW) can be bypassed by passing them to another compartment. Cross-compartment wrappers often do not go through SCSW, but have a filtering policy built into them. When an object is wrapped cross-compartment, the SCSW is stripped off and, when the object is read read back, it is not known that SCSW was previously present, resulting in a bypassing of SCSW. This could result in untrusted content having access to the XBL that implements browser functionality (CVE-2012-1959).

Google developer Tony Payne reported an out of bounds (OOB) read in QCMS, Mozillas color management library. With a carefully crafted color profile portions of a user's memory could be incorporated into a transformed image and possibly deciphered (CVE-2012-1960).

Bugzilla developer Fredric Buclin reported that the X-Frame-Options header is ignored when the value is duplicated, for example X-Frame-Options: SAMEORIGIN, SAMEORIGIN. This duplication occurs for unknown reasons on some websites and when it occurs results in Mozilla browsers not being protected against possible clickjacking attacks on those pages (CVE-2012-1961).

Security researcher Bill Keese reported a memory corruption. This is caused by JSDependentString::undepend changing a dependent string into a fixed string when there are additional dependent strings relying on the same base. When the undepend occurs during conversion, the base data is freed, leaving other dependent strings with dangling pointers. This can lead to a potentially exploitable crash (CVE-2012-1962).

Security researcher Karthikeyan Bhargavan of Prosecco at INRIA reported Content Security Policy (CSP) 1.0 implementation errors. CSP violation reports generated by Firefox and sent to the report-uri location include sensitive data within the blocked-uri parameter. These include fragment components and query strings even if the blocked-uri parameter has a different origin than the protected resource. This can be used to retrieve a user's OAuth 2.0 access tokens and OpenID credentials by malicious sites (CVE-2012-1963).

Security Researcher Matt McCutchen reported that a clickjacking attack using the certificate warning page. A man-in-the-middle (MITM) attacker can use an iframe to display its own certificate error warning page (about:certerror) with the Add Exception button of a real warning page from a malicious site. This can mislead users to adding a certificate exception for a different site than the perceived one. This can lead to compromised communications with the user perceived site through the MITM attack once the certificate exception has been added (CVE-2012-1964).

Security researchers Mario Gomes and Soroush Dalili reported that since Mozilla allows the pseudo-protocol feed: to prefix any valid URL, it is possible to construct feed:javascript: URLs that will execute scripts in some contexts. On some sites it may be possible to use this to evade output filtering that would otherwise strip javascript: URLs and thus contribute to cross-site scripting (XSS) problems on these sites (CVE-2012-1965).

Mozilla security researcher moz_bug_r_a4 reported a arbitrary code execution attack using a javascript: URL. The Gecko engine features a JavaScript sandbox utility that allows the browser or add-ons to safely execute script in the context of a web page. In certain cases, javascript: URLs are executed in such a sandbox with insufficient context that can allow those scripts to escape from the sandbox and run with elevated privilege. This can lead to arbitrary code execution (CVE-2012-1967).

The mozilla firefox and thunderbird packages has been upgraded to the latest respective versions which is unaffected by these security flaws.

Additionally the rootcerts packages has been upgraded to the latest version which brings updated root CA data.

Update:

Localization packages for firefox was missing with the MDVSA-2012:110 advisory and is being provided with this advisory.