Re: [flexcoders] Re: CSSI Security Issues and Flex (Proof of concept exploit)

2005-05-06 Thread John Dowdell
Eric Raymond wrote:
> Here is a sample exploit.  It requires you to trick the user into
> clicking a "link". But if you can do that some percent of the time,
> with the aid of a flash decompiler to explore the app, you might be
> able to do all sorts of interesting things.

Thanks, Eric. I suspect one of my partners may already have forwarded 
this along to the Security Center silently, without an acknowledgment to 
the public conversation, so I won't put a lot of time into this here myself.

But it almost sounds like you're describing a situation of "evil 
author", rather than "evil site visitor", because you listed MXML rather 
than text which a visitor could type into a field for a classic 
script-injection approach. (And even then, "asfunction()" calls would be 
bounded by the same security sandbox which corrals all ActionScript calls.)

Hmm... or maybe your scenario is closer to "I relied on some type of 
web service which went rogue behind my back", that might be it...?

But I'd defer to the Security Team... if you haven't received an offlist 
advisory that someone sent this there for you already, then here's the 
"Please alert us" entry:
http://www.macromedia.com/devnet/security/security_zone/alertus.html

tx,
jd



-- 
John Dowdell . Macromedia Developer Support . San Francisco CA USA
Weblog: http://www.macromedia.com/go/blog_jd
Aggregator: http://www.macromedia.com/go/weblogs
Technotes: http://www.macromedia.com/support/
Spam killed my private email -- public record is best, thanks.


 
Yahoo! Groups Links

<*> To visit your group on the web, go to:
http://groups.yahoo.com/group/flexcoders/

<*> To unsubscribe from this group, send an email to:
[EMAIL PROTECTED]

<*> Your use of Yahoo! Groups is subject to:
http://docs.yahoo.com/info/terms/
 




[flexcoders] Re: CSSI Security Issues and Flex (Proof of concept exploit)

2005-05-06 Thread Eric Raymond
Two more points:

1) With a small number of ways to accept input and complete control
over the client implementation, Macromedia can plug this hole on the
client side.

2) But fixing things one the client may not prevent attack if the one
were to bypass the client in some way.  The flex proxy/remote objects
makes that a bit more difficult but not impossible.  One could
generate a post directly to the web service via a browser or other
tool.  This is analogous to performing input validation only in
javascript and not on the server side.

Is this a last point an argument for throwing our hands up in the air
and not doing anything (even if it's not complete)?  I don't think so

IMHO, having the four or five input methods in Flex by default
disallow entry of asfunction  and possibly "<" and ">" (or map them to
something like "[" and "]") would be a small step towards safety for
the masses.  Perhaps this would be turned off automatically on
"password" input.


So much for beating a dead horse




 
Yahoo! Groups Links

<*> To visit your group on the web, go to:
http://groups.yahoo.com/group/flexcoders/

<*> To unsubscribe from this group, send an email to:
[EMAIL PROTECTED]

<*> Your use of Yahoo! Groups is subject to:
http://docs.yahoo.com/info/terms/
 




[flexcoders] Re: CSSI Security Issues and Flex (Proof of concept exploit)

2005-05-05 Thread Eric Raymond
Semantics aside, flex applications tend to promote a type of
interaction that the majority of flash applications (in general) do not.  

FYI, I mispoke hugely.  The offending command is asfunction, not
fscommand.

Here is a sample exploit.  It requires you to trick the user into
clicking a "link". But if you can do that some percent of the time,
with the aid of a flash decompiler to explore the app, you might be
able to do all sorts of interesting things.


http://www.macromedia.com/2003/mxml";
xmlns="*">










--- In flexcoders@yahoogroups.com, John Dowdell <[EMAIL PROTECTED]> wrote:
> The title may be a bit of a misnomer, because Macromedia Flex lives on 
> the server, while cross-site scripting exploits would occur on the 
> client machines. This seems a sub-class of general security in the 
> Macromedia Flash Player rather than the development environment,
true...?
> 
> Here's general background info on security and privacy in the
Macromedia 
> Flash Player:
> http://www.macromedia.com/devnet/flashplayer/
> ... and here's background on recent security issues in the Macromedia 
> Flash Player:
> http://www.macromedia.com/devnet/security/security_zone/#flashplayer
> 
> As I understand the post, you're concerned about the possibility of a 
> command injection into a textfield of a SWF application. (I could be 
> wrong, but it sounded to me more like a script-injection issue than a 
> cross-site scripting issue.) Have you been able to see this happen yet? 
> have you typed "fscommand:()" into a textfield in a particular
component 
> to pop up an alert or such? If there's a recipe that could be
reproduced 
> in-house then we can work on it.
> 
> Or is it more a general curiosity, about whether there might be a way 
> that such a thing is possible?
> 
> tx,
> jd
> 
> 
> 
> 
> -- 
> John Dowdell . Macromedia Developer Support . San Francisco CA USA
> Weblog: http://www.macromedia.com/go/blog_jd
> Aggregator: http://www.macromedia.com/go/weblogs
> Technotes: http://www.macromedia.com/support/
> Spam killed my private email -- public record is best, thanks.




 
Yahoo! Groups Links

<*> To visit your group on the web, go to:
http://groups.yahoo.com/group/flexcoders/

<*> To unsubscribe from this group, send an email to:
[EMAIL PROTECTED]

<*> Your use of Yahoo! Groups is subject to:
http://docs.yahoo.com/info/terms/