Re: arcfour-hmac checksum salt value

["Kevin Coffman" <kwc@citi.umich.edu>]

On Fri, Mar 14, 2008 at 11:03 AM, Love Hörnquist Åstrand <lha@kth.se> wrote:
>
>  11 mar 2008 kl. 12.42 skrev Kevin Coffman:
>
>
>
>  > While implementing arcfour-hmac for Linux Kernel NFS use, I have run
>  > into the following issue:
>  >
>  > According to rfc4757 (sections 7.2 and 7.3), the salt value when
>  > generating the checksum for both MIC and WRAP tokens is 15.  However,
>  > the MIT, Heimdal, and Java implementations all seem to map the usage
>  > values (used while creating the checksum) in WRAP tokens to a salt
>  > value of 13 instead.
>  >
>  > Can someone verify that either I'm confused, or the spec is wrong in
>  > the case of the checksum salt value that Microsoft used for WRAP
>  > tokens?
>
>   From where do you get 13 in heimdal ?
>
>   From what I can read, heimdal uses KRB5_KU_USAGE_SIGN that later in
>  the crypto layer is mapped to 15 for the mic checksum.
>
>  Love

Function usage2arcfour() when given KRB5_KU_USAGE_SEAL.

_gsskrb5_wrap()
  --> _gssapi_wrap_arcfour()
    --> arcfour_mic_cksum()
      --> krb5_create_checksum()
        --> usage2arcfour()

I was only looking at your code (and testing my code against Solaris).

Let me know if I'm misreading this...

K.C.