Pointer validation in the gss krb5 mech

Nico Williams nico at cryptonector.com
Tue Apr 12 13:47:36 EDT 2011


+1.

If we wanted any sort of validation I'd rather we intern handles and return
small integers (cast as pointers) instead of pointers.  Also, I'm not really
a fan of magic fields.  If garbage is passed in, better to segfault (or
assert) than to return an error that is likely to get ignored (thus masking
a bug).  Rip it out.

Nico
--



More information about the krbdev mailing list