I have a newer version of the cifs-upcall patch which follows the linux coding standards. I have taken a look at the use of krb5_context through out all the code and reduced the usage of krb5_context to one call.
John Dey
On 8/22/16, 5:29 AM, "Jeff Layton" <jlayton@xxxxxxxxx> wrote:
>The handling of krb5 in cifs.upcall has always been pretty klunky. It
>rolls through /tmp, trying to find the latest credcache and has some
>hacks to allow it to use DIR: caches as well, but none of that really
>works for KEYRING:, which is pretty common these days.
>
>In practice, I doubt anyone relies on that behavior. What most people
>want is for cifs.upcall to find the default credcache for a user given
>krb5.conf -- full stop.
>
>This patchset rips out most of the unneeded machinery in cifs.upcall,
>and just has it find the default credcache and verify that it has a
>valid TGT. If not then we'll try to init it from the keytab as before.
>
>I think there's some more opportunity to clean up this code in the
>future as well. Currently we pass around strings that represent the
>credcache, and that could be made more efficient. It might also be
>good to just reimplement the whole thing with gssapi calls instead.
>
>Still, this is a good step in that direction I think.
>
>Jeff Layton (3):
> aclocal: fix typo in idmap.m4
> cifs.upcall: use krb5 routines to get default ccname
> cifs.upcall: make the krb5_context a static global variable
>
> aclocal/idmap.m4 | 2 +-
> cifs.upcall.c | 185 ++++++++++---------------------------------------------
> 2 files changed, 32 insertions(+), 155 deletions(-)
>
>--
>2.7.4
>
��.n��������+%������w��{.n�����{�����ܨ}���Ơz�j:+v�����w����ޙ��&�)ߡ�a����z�ޗ���ݢj��w�f
