an issue relating to using current MIT krb code in kernel space

[Greg Hudson]

On Thu, 2011-01-20 at 16:32 -0500, Will Fiveash wrote:
> so I either have to add it
> back for Solaris or dynamically calculate the number of krb5_enctype
> entries there are in context->in_tkt_ktypes before I call
> FREE()/kmem_free().

Is dynamically calculating the length a problem?  We have a short
function krb5int_count_etypes (in lib/krb5/krb/etype_list.c) to conduct
this count, and the runtime overhead should be minimal since enctype
lists are pretty short.