Re: Question regarding CVE applicability of missing HttpOnly flag

["Vincent Danen" <vdanen () redhat com>]

On 06/27/2014, at 12:42 PM, Kurt Seifried wrote:

> On 27/06/14 10:35 AM, Vincent Danen wrote:
> > On 06/26/2014, at 10:00 AM, Kurt Seifried wrote:
> >
> > > On 26/06/14 05:45 AM, Jamie Strandboge wrote:
> > > > Based on this email and the one this is in response to, I find
> > > > this comment unclear. Is MITRE saying that:
> > > >
> > > > a) lack of implementing SELinux, AppArmor, virus scanner,
> > > > firewall, <insert hardening software here> does not justify a
> > > > CVE because of the complexity? b) lack of implementing SELinux,
> > > > AppArmor, virus scanner, firewall, <insert hardening software
> > > > here> does not justify a CVE and also cannot be considered an
> > > > implementation error because of the complexity? c) implementing
> > > > SELinux, AppArmor, virus scanner, firewall, and/or <insert
> > > > hardening software here> is not worth it because the added
> > > > complexity intrinsically makes the system less secure? d)
> > > > something else?
> > > >
> > > > Thanks
> > >
> > > So one comment on this, replace the above with "DAC"
> > > (http://en.wikipedia.org/wiki/Discretionary_access_control) and I
> > > bet we'd hand it a CVE =).
> > >
> > > Security lines move, I would expect most modern system of any
> > > type (Windows, Linux, router, maybe not my bathroom scale that
> > > talks wifi... yet) to have some sort of firewall enabled by
> > > default and not simply leave everything exposed to the world. So
> > > in that case not having a fire enabled by default would
> > > definitely violate the principle of least surprise and maybe even
> > > qualify for a CVE.
> >
> > Wait.  You're saying that not having a firewall enabled by default
> > qualifies for a CVE?  I mean, firewalls are pretty common sense and
> > should definitely be used/available/whatever but to say that an
> > operating system or device doesn't have a firewall enabled by
> > default should have a CVE assigned seems... excessive, doesn't it?
>
> I'm saying in quite a few common situations it should probably qualify
> for a CVE. Not every single situation. Same for HTTPOnly.
> > How is not having a firewall enabled by default a _vulnerability_?
> > If we look at it this way, it's a good thing CVEs go past 9999 per
> > year because we need to change everything we used to call
> > "hardening" to be a vulnerability, do we not?
>
> How is not having DAC a _vulnerability_? and yet now DAC support is
> required....

That is my question.  If not having DAC is a vulnerability, why do I find absolutely _no_ CVEs for its absence?  
Whether it's required or not has no bearing on whether or not it's a flaw.  But now HttpOnly is not only _not_ 
required, but its absence is being considered a flaw.

This is where my concern/confusion is.  Are we calling things flaws because they have an actual vulnerability, or are 
we starting to call things "flaws" even tough they don't have vulnerabilities, don't fix vulnerabilities, but they're 
extremely useful things to have?  I don't think that's the definition of a flaw.



-- 
Vincent Danen / Red Hat Product Security

Attachment: signature.asc
Description: OpenPGP digital signature