Re: [Hotp-toolkit-help] Possible bug in HOTP / support for TOTP, OCRA?

[Simon Josefsson]

Simon Josefsson <address@hidden> writes:

>> I had trouble getting it started, and the reason appears to be the "-" in the
>> proposed initial /etc/users.hotp file.  To get the PAM module working, I had 
>> to
>> uncomment these lines from usersfile.c:
>>
>>           if (strcmp (p, "-") == 0 && *p != '\0')
>>             return HOTP_BAD_PASSWORD;
>>           if (strcmp (p, passwd) != 0)
>>             return HOTP_BAD_PASSWORD;
>>
>> Without it, I always got the HOTP_BAD_PASSWORD return.  Looking at the code, 
>> I
>> was confused why the test *p != '\0' is done after strcmp (p, "0") == 0 as it
>> does not add anything?  Possibly this code is mistaken?
>>
>> I am uncertain about the purpose of this code, and am I the only one who 
>> cannot
>> get the PAM module going with the proposed initial /etc/users.hotp file?  I 
>> used
>> the README-proposed contents:
>>
>> HOTP root - 00
>
> I am looking at this now, first to see whether I can reproduce it
> locally.  Indeed the strcmp if case looks a bit strange.

Should be fixed in 1.0.1 now.  The code should have read:

          if (strcmp (p, "-") == 0 && *passwd != '\0')
            return HOTP_BAD_PASSWORD;
          if (strcmp (p, passwd) != 0)
            return HOTP_BAD_PASSWORD;

The intention being that if the user provided a non-empty password, and
the system did not know any password, authentication will fail.  This
helps to alert users that the security is only one-factor even if the
user thought that it would be two-factor because of the password.

Thanks,
/Simon