pull up r23712, r23714 from trunk
authorTom Yu <tlyu@mit.edu>
Fri, 12 Feb 2010 20:28:39 +0000 (20:28 +0000)
committerTom Yu <tlyu@mit.edu>
Fri, 12 Feb 2010 20:28:39 +0000 (20:28 +0000)
commitcf6de28a3bb42f4c999697ca7bbd32c182ec380f
tree81ce194b8511c388ea870666df2bc0cf81323265
parente9a60482b9d92c87a3762dba56a6e34a5d09d790
pull up r23712, r23714 from trunk

 ------------------------------------------------------------------------
 r23714 | ghudson | 2010-02-09 20:55:36 -0500 (Tue, 09 Feb 2010) | 13 lines

 ticket: 6656

 Followon fixes to r23712:
 * A few formatting fixes.
 * Fix unlikely leak in kdc_handle_protected_negotiation: if
   add_pa_data_element with copy == FALSE fails, it's still the
   caller's responsibility to free pa.contents.
 * Fix pre-existing (since r23465) leak of reply_encpart.enc_padata in
   process_as_req.
 * Call add_pa_data_element with copy == TRUE in
   return_referral_enc_padata since we are passing memory owned by the
   database entry.

 ------------------------------------------------------------------------
 r23712 | hartmans | 2010-02-09 14:15:07 -0500 (Tue, 09 Feb 2010) | 14 lines

 subject: enc_padata can include empty sequence
 ticket: 6656
 target_version: 1.8
 tags: pullup

 There are two issues with return_enc_padata.
 1)  It often will return an empty sequence of enc_padata rather than not including the field
 2) FAST negotiation is double supported in the referral tgs path and not supported in the non-referral path

 Rewrite the return_enc_padata logic to:

 * Split  out referral interactions with kdb into its own function
 * Use add_pa_data_element

ticket: 6656
version_fixed: 1.8
status: resolved

git-svn-id: svn://anonsvn.mit.edu/krb5/branches/krb5-1-8@23718 dc483132-0cff-0310-8789-dd5450dbe970
src/kdc/do_as_req.c
src/kdc/do_tgs_req.c
src/kdc/kdc_preauth.c
src/kdc/kdc_util.c
src/kdc/kdc_util.h