Full Disclosure mailing list archives

Re: Hilariously Bad SQRL Implementation


From: Scott Arciszewski <scott () arciszewski me>
Date: Thu, 21 Aug 2014 22:13:10 -0400

Why would any sane rational human being implement something from Gibson?

On that note, I did come across the attrition charlatan page after I
started this project. But as far as I'm concerned, I'm only implementing an
Ed25519 signature. That the specific client application is one of Steven
Gibson's brain-children is sort of irrelevant to the scope of my
initiative, but I will certainly treat anything he produces with due
skepticism. That being said, just because he has a history of
fraud/incompetence doesn't mean I will immediately discredit SQRL without
evaluating the spec/src, whenever they're available. It deserves a fair
chance and an uncensored review.

I would implement time proven solutions based on real world testing not
an experimental solution from a rather dubious source.

I suppose libsodium/NaCl would fail to meet some peoples' definitions for
"time proven", but the underlying crypto (Curve25519/Ed25519) is being
implemented in various high profile applications (Tor, Cryptocat, etc.) so
if it fails to pass the test of time, a lot of people will end up with egg
on their face.

I suppose, as an interesting side project, it might be interesting to
explore but for production, I wouldn't touch with a million foot stick.

I fully agree. I wouldn't recommend either the library I'm building nor the
program/protocol that Steven Gibson produces be used in any production
environment until they have been reviewed by competent hackers and
cryptographers.


On Wed, Aug 20, 2014 at 11:04 PM, Sanguinarious <
Sanguinarious () occultusterra com> wrote:

Why would any sane rational human being implement something from
Gibson? I still remember him saying how implementing raw sockets in
Windows XP will totally and utterly destroy the entire internet. I
would implement time proven solutions based on real world testing not
an experimental solution from a rather dubious source.

I suppose, as an interesting side project, it might be interesting to
explore but for production, I wouldn't touch with a million foot
stick.


On Sun, Aug 17, 2014 at 8:22 PM, Scott Arciszewski
<kobrasrealm () gmail com> wrote:
If any of you are familiar with Stephen Gibson's SQRL protocol for user
authentication (really neat idea), you might have come across this PHP
implementation before: https://github.com/geir54/php-sqrl

Unfortunately, this library is actually pretty terrible. Not only does it
pass all of the data off to a Heroku app to perform the signature
verification, it is also vulnerable to SQL Injection:


https://github.com/geir54/php-sqrl/blob/0fa574520a1843a33a84c3985f934e84af6f2042/sqrl_verify.php#L39-59

I thought about submitting a pull request to fix this, but I don't
believe
there is honestly much here to salvage. So, I'm writing my own
implementation here:

https://github.com/darkitecht/php-sqrl <- Not ready, at all, for even
beta
testing.

P.S. Also, it uses mt_rand() for challenge generation in a crypto
library.
Tsk tsk.

_______________________________________________
Sent through the Full Disclosure mailing list
http://nmap.org/mailman/listinfo/fulldisclosure
Web Archives & RSS: http://seclists.org/fulldisclosure/

_______________________________________________
Sent through the Full Disclosure mailing list
http://nmap.org/mailman/listinfo/fulldisclosure
Web Archives & RSS: http://seclists.org/fulldisclosure/


_______________________________________________
Sent through the Full Disclosure mailing list
http://nmap.org/mailman/listinfo/fulldisclosure
Web Archives & RSS: http://seclists.org/fulldisclosure/


Current thread: