> Anybody have a real web app that relies on the integrity
> of content headers for security?

There have been various attacks using the UTF-7 charset (eg an XSS attack 
against Google in 2005). The attack works when the body of an HTTP response is 
misinterpreted by browsers as UTF-7, instead of UTF-8 or US-ASCII.
The charset is specified in the Content-Type header.

Another potentially relevant attack, named GIFAR, was reported a BlackHat USA 
this year. A single file (HTTP body) can easily be interpreted as a (harmless) 
GIF image or (potentially malicious) Java code (JAR file). One reason the 
attack worked was because a Java Virtual Machine (JVM) ignored the image/gif 
Content-Type. If the solution was for the JVM to care about the Content-Type, 
not protecting its integrity could undo the solution.

Both of these are real attacks. They might not be absolutely directly 
applicable to this OAuth body signing situation -- but they are close.


P.S. DKIM (DomainKeys Identifier Mail) [RFC 4871] defines a signature mechanism 
that covers header fields and the body of an email (that are very similar to 
header fields and the body of an HTTP message). It offers "simple" and 
"relaxed" canonicalization rules to cope with the complications of headers. It 
might be worth a look.


James Manger
[EMAIL PROTECTED]
Identity and security team — Chief Technology Office — Telstra

----------
From: [email protected] [mailto:[EMAIL PROTECTED] On Behalf Of Brian Eaton
Sent: Wednesday, 10 December 2008 1:25 PM
To: [EMAIL PROTECTED]
Cc: Marc Worrell; Louis Ryan; [email protected]; Scott Seely
Subject: [oauth] Re: [opensocial-and-gadgets-spec] Re: body signing for oauth 
and opensocial


On Tue, Dec 9, 2008 at 5:52 PM, Krishna Sankar (ksankar)
<[EMAIL PROTECTED]> wrote:
> Have been watching this discussion. Something is puzzling me - the spec
> [1] talks about integrity of the body nothing else. The only thing it
> says is that the body has not been altered since it left the consumer.
> As the header also has relevance to the body, we should guarantee the
> integrity of the headers as well, which means sign the header too.
>
> Looking from another perspective, whatever threat model requires a
> signature of the body also requires signature of the headers as well.
>
> Possible I am missing something obvious.

No, you're not.  In theory protecting the headers is vitally important
for security.

In practice it doesn't seem to matter.  No one has been able to point
to a single real web application whose security would be negatively
impacted by a bad guy who tampers with content headers but not the
body.

Several people, OTOH, have suggested reasons why trying to sign
headers will be complicated.  If it doesn't give security, but it does
make things complicated, I don't think we should do it.

Anybody have a real web app that relies on the integrity of content
headers for security?


--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups 
"OAuth" group.
To post to this group, send email to [email protected]
To unsubscribe from this group, send email to [EMAIL PROTECTED]
For more options, visit this group at http://groups.google.com/group/oauth?hl=en
-~----------~----~----~----~------~----~------~--~---

Reply via email to