Follow @Openwall on Twitter for new release announcements and other news
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Tue, 30 Oct 2012 13:34:07 -0400 (EDT)
From: "Steven M. Christey" <coley@...-smtp.mitre.org>
To: Kurt Seifried <kseifried@...hat.com>
cc: oss-security@...ts.openwall.com, Josh Bressers <bressers@...hat.com>
Subject: Re: Strange CVE situation (at least one ID should
 come of this)


> On 10/26/2012 01:54 PM, Josh Bressers wrote:
>>
>> If I was to list the security problems I found after a few minutes
>> of looking, they are:
>>
>> * It uses MD5 passwords * The shadow file is directly modified
>> without locking (which could lead to a race condition) * If you get
>> the password wrong, it doesn't unlink the empty temporary file.
>>
>> None are really a big deal, you *could* run this and probably never
>> notice these problems.
>>
>> Fundamentally though, this thing should get one CVE ID that
>> basically say "don't use this". How have situations like this been
>> handled in the past?

To have a CVE for "don't use this" is not consistent with long-existing 
practice.  I don't recall ever intentionally assigning a CVE for such a 
thing - after all, CVE is about vulnerabilities, and "don't use this" is 
awfully vague.

Deployment of risky software is effectively a configuration or asset 
management issue, which is well outside the scope of CVE. (Maybe it's more 
like a Common Configuration Enumeration (CCE) issue.)

In other words - we really shouldn't use CVE to handle this problem.  It 
is feature creep, and I believe that it WOULD become a huge mess.  Maybe 
this would work for some, but not for all of CVE's consumers, which is a 
wide variety of people and use cases.  I understand that there is a 
problem here, though.

It looks like Josh laid out at least 3 different security issues in your 
initial request.  Those can/should get CVEs assigned, even if there aren't 
full details.  The lack of a vendor CONFIRM reference or advisory, tells 
the consumer that the vendor hasn't addressed it.

Perhaps the OSS community could borrow an idea from one of the framework 
vendors with lots of third-party modules - I forget if it was Joomla or 
Drupal - who actively maintained a list of poorly maintained or obsolete 
software.

In the broadest sense, however, such old software is still useful for 
people who are starting in vulnerability research, or just doing it for 
fun; many people who audit what MITRE calls "phpGolf" applications, go on 
to do more substantive research.

Perhaps it is time to re-examine Crispin Cowan's Sardonix project, which 
tried to match vulnerability researchers with open source projects, in 
order to build reputations for both.

- Steve

Powered by blists - more mailing lists

Please check out the Open Source Software Security Wiki, which is counterpart to this mailing list.

Confused about mailing lists and their use? Read about mailing lists on Wikipedia and check out these guidelines on proper formatting of your messages.