Hi,

In the absence of any objections, we'll update the guidelines to reflect the
edits below (in bold).
The focus here is to bring us inline with industry ratings (having spoken to
Microsoft and looking at Mozilla's guidelines). The key change is to rate
sandbox escape issues as "High".


Cheers
Chris

Severity Guidelines for Security Issues
Although the Chromium project itself does not rate security vulnerabilities,
vendors shipping products based on Chromium might wish to rate the severity
of security issues in the products they release. This document contains
guidelines for how to rate these issues.  We recommend vendors rate security
vulnerabilities using one of four severity levels: critical, high, medium,
and low.Critical
A vulnerability is critical if the vulnerability lets an attacker run
arbitrary code with the user's privileges in the normal course of browsing.
 For example, an uncontrolled buffer overflow in the browser process should
typically be rated critical, especially if a malicious web site can directly
control the contents of the buffer.

We recommend rating most memory safety issues in the browser process as
critical unless the possibility of arbitrary code execution can be ruled
out.  However, not all crashes indicate a critical vulnerability.  For
example, Chromium is designed to crash in a controlled manner (e.g., with a
__debugBreak) when memory is exhausted or in other exceptional
circumstances.  Also, an arbitrary code execution vulnerability that
requires an unusual user action (such as printing a certificate error
message) should typically not be rated as critical.
High
A vulnerability is of high severity if the vulnerability lets an attacker
read or modify confidential data belonging to other web sites.  For example,
an issue that lets the attacker circumvent the same-origin policy should
typically be rated high.  Additionally, we recommend rating issues that let
an attacker execute arbitrary code within the confines of the sandbox as
high because the sandbox is designed to limit the privileges of a
compromised rendering engine.

Vulnerabilities that interfere with browser security features are also high
severity.  For example, issues that let attackers suppress SafeBrowsing
warnings warrant a high rating because they can be combined with other
vulnerabilities to cause serious harm to users.  Similarly, vulnerabilities
that disrupt other browser security indicators, such as the location bar and
lock icon, are also high severity.  (Note that the status bubble is not a
security indicator.)

*We also recommend rating vulnerabilities in the implementation of the
sandbox as high even though these vulnerabilities are not typically
accessible unless the attacker can also compromise the rendering engine.*

*Medium*

A vulnerability is of medium severity if the vulnerability lets an attacker
obtain only limited amounts or kinds of information.  For example, an issue
that lets the attacker enumerate recently visited URLs should typically be
rated medium.  An issue that is not harmful in and of itself but that can be
combined with other vulnerabilities to cause harm will usually warrant a
medium severity.  For example, ignoring a "do not cache" directive might not
itself be harmful but might facilitate other attacks.  Additionally,
otherwise *higher* severity issues that require some unusual user action
(such as terminating a tab's process while in full-screen mode) will
typically be rated as medium severity.
Low
A vulnerability is of low severity if the vulnerability grants the attacker
only temporary control over non-critical browser features.  For example, an
issue that lets the attacker hang the browser is a low severity issue.
 (Note that tab hangs are not security issues if they can be resolved simply
by closing the tab.)  Also, security improvements that are not necessarily
vulnerabilities (like not saving the password portion of URLs in browser
history) should also typically be rated as low severity.  In general, users
should be able to use a browser with low severity issues for extended
periods of time without many adverse effects.

-- 
Chromium Developers mailing list: [email protected] 
View archives, change email options, or unsubscribe: 
    http://groups.google.com/group/chromium-dev

Reply via email to