Bug in canon_name.c?

[Sam Hartman]

>>>>> "Michael" == Michael Tautschnig <michael.tautschnig at zt-consulting.com> writes:

    Michael> Hi!  I tested the IPSec-daemon "racoon" and got segfaults
    Michael> when using gssapi_krb - authentication. I was able to
    Michael> track that down to the call of gss_canonicalize_name,
    Michael> which is called with GSS_C_NO_OID as the mech_type.
    Michael> Please correct me if that is wrong, but some
    Michael> documentation said that this would even be preferred -
    Michael> but gss_canonicalize_name checks the mech_type versus
    Michael> gss_mech_krb5 using the macro g_OID_equal which will
    Michael> immediatly segfault.

    Michael> Is racoon doing something unwanted or is this a
    Michael> mit-kerberos-bug?

Probably both.  I'm not sure why it needs to call
gss_canonicalize_name, but it probably should not segfault in the MIT
code.