XPCNativeWrapper: Difference between revisions
No edit summary |
|||
Line 26: | Line 26: | ||
document._content.document.title = 'Hello World'; | document._content.document.title = 'Hello World'; | ||
</pre> | </pre> | ||
On the surface it looks harmless enough. However, an attacker could execute malicious code if they added a | On the surface it looks harmless enough. However, an attacker could execute malicious code if they added a getter to the title property: | ||
<pre> | <pre> | ||
document.title getter = function() { | document.title getter = function() { |
Revision as of 09:13, 22 March 2005
XPCNativeWrapper is a Javascript object that should be used whenever privileged code is used to access unprivileged code. It is used to create a security wrapper that is analogous to a sandbox.
Basic Usage
Using XPCNativeWrapper is very simple. You create an instance of an XPCNativeWrapper with the untrusted object as the first parameter and the desired properties and methods of the untrusted object as the additional parameters.
Example
This example creates a wrapper around the document object so that it can safely read the title of the document and call one of its methods:
doc = window._content; doc2 = new XPCNativeWrapper(doc, 'title', 'getElementsByTagName()');
Notice how the desired properties are specified by their name and the desired methods are specified by their name followed a pair of empty brackets.
The returned object can be used just like the untrusted object. Methods and properties can be called from it directly.
var tt = doc2.title; var bd = doc2.getElementsByTagName('body')[0];
Attack Scenarios
To illustrate the importance of using XPCNativeWrapper, consider the following scenario:
An extension is written to modify the title property of the document.
document._content.document.title = 'Hello World';
On the surface it looks harmless enough. However, an attacker could execute malicious code if they added a getter to the title property:
document.title getter = function() { alert('code executed'); //malicious code... };
The same analogy applies to calling methods on an untrusted object. Consider the following scenario:
An extension tries to get access to the body element of a document.
var bd = document._content.document.getElementsByTagName('body')[0];
It may also look harmless, but an attacker could execute malicious code if they overwrote the getElementsByTagName method with another:
document.getElementsByTagName = function() { alert('code executed'); //malicious code... };