[cap-talk] SANS Institute's "25 Most Dangerous Programming Errors"

[Bill Frantz]

erights at gmail.com (Mark Miller) on Monday, January 12, 2009 wrote:

>CWE-494: Download of Code Without Integrity Check
>
>CWE-404: Improper Resource Shutdown or Release
>?
E, Waterken/Joe-E, Caja: safe pointers/garbage collection

...

>CWE-285: Improper Access Control (Authorization)
E, Waterken/Joe-E, Caja: capabilities

>CWE-327: Use of a Broken or Risky Cryptographic Algorithm
E: VatTP uses 3DES, 1K bit Diffie-Hellman, SHA1, and DSA. 3DES is still
considered secure, if a bit quaint. SHA1 is falling to attack and NIST
wants if out of use by the federal government by 2010. 1K bit for the
Diffie-Hellman is now considered a bit small, but is still OK. DSA depends
on SHA1, but I don't know about possible replacements. Waterken/Joe-E and
Caja: Depend on algorithms implemented by servers, clients, and certificate
authorities (CAs). Unless tweaked by the server's sysops, web servers
probably will negotiate weak algorithms. May be vulnerable to the MD5
attacks on CAs.

Note that OpenSSL does not yet implement SHA2. Since OpenSSL is widely used
by open source software, it does not allow implementation of the
recommended fix for the weakening of SHA1 which is to use SHA2.

In the real world, SHA1 isn't publicly broken yet and NIST is running a
competition for a new secure hash algorithm. People will probably wait for
that algorithm to be selected and then upgrade.

Cheers - Bill

-----------------------------------------------------------------------
Bill Frantz        | There are also no libertar-  | Periwinkle
(408)356-8506      | ians in financial crises.    | 16345 Englewood Ave
www.pwpconsult.com |               - Jeff Frankel | Los Gatos, CA 95032