I've added a security bulletin to our official Struts 2 documentation to
begin to formalize this issue and its solution:
http://cwiki.apache.org/confluence/display/WW/S2-001+-+Remote+code+exploit+on+form+validation+error

Don

On 7/17/07, Don Brown <[EMAIL PROTECTED]> wrote:

The patch I commited is based on the original loopcount patch, but fixes
the problem where it wouldn't evaluate all non-recursive expressions.
Therefore, the issue has been fixed and all tests still pass.  I agree that
we should re-evaluate our usage of ognl down the road, but I believe the
committed fix will resolve the security issue.  I've back-ported the fix to
XWork 2.0 and XWork 1.2, and Rainer has promised XWork releases in the
next few days.

Don

On 7/17/07, Ing. Andrea Vettori < [EMAIL PROTECTED]> wrote:
>
>
> Il giorno 16/lug/07, alle ore 16:46, Antonio Petrelli ha scritto:
>
> > 2007/7/16, Ing. Andrea Vettori <[EMAIL PROTECTED]>:
> >>
> >> I suggested the value can be parametrized so if one
> >> known he use complex expression can use a higher value. (b) is solved
>
> >> using loopCount=1 by default when dealing with user input.
> >
> >
> >
> > OK! Thank you I think I got the point.
> > So you are saying that, with loopCount=1, the evaluation step stops at
>
> > evaluating the string as it is, right?
>
> ok !
>
> Now we should only understand what to do with expression like "%{foo}
> %{bar}" that has more than one expression at the "same" recursion level.
>
>
>
> --
> Ing. Andrea Vettori
> Consulente per l'Information Technology
>
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: [EMAIL PROTECTED]
> For additional commands, e-mail: [EMAIL PROTECTED]
>
>

Reply via email to