[PATCH] Simplify auth_check_ntlm_password

[Andrew Bartlett]

On Tue, 2017-03-07 at 13:52 -0800, Jeremy Allison wrote:
> On Wed, Mar 08, 2017 at 10:04:19AM +1300, Andrew Bartlett wrote:
> > On Tue, 2017-03-07 at 14:28 +0100, Volker Lendecke wrote:
> > > Hi!
> > > 
> > > While working on the auth code I had a hard time following the
> > > logic
> > > in auth_check_ntlm_password. The attached patchset is supposed to
> > > make
> > > that more traceable. It does not change behaviour, those patches
> > > will
> > > come at a later step and will make the routine even simpler. The
> > > behaviour change will be to always break out of the loop if
> > > 
> > > !NT_STATUS_EQUAL(nt_status,NT_STATUS_NOT_IMPLEMENTED);
> > > 
> > > Review appreciated!
> > 
> > Thanks Volker, these look like a helpful start to making this stuff
> > simpler.  
> > 
> > I looked at this set last night, to see how it works with the
> > authentication audit code (so I could integrate into that series it
> > as
> > promised).  Putting it in before causes rebase pain, so I'll take a
> > look at adding it after and see if it is any less difficult.
> > 
> > Otherwise I'll take the concepts and apply them once we are done.  
> 
> Really ? Does this cause much rebase pain ? Yeah, it'll cause
> some, but this really does clean up these code paths.

I'm not objecting to the patch, just saying I will try and add it on
top of the branch rather than underneath it.  I did actually try the
rebase last night, and perhaps you have better luck, but I've had
enough of these go bad on me that I would just like to apply it on top,
if that is OK.

> Looking at the previously posted logging patch I can only
> find two places where the logging patch touches source3/auth/auth.c:

We keep our repo public, but we haven't been spamming the list as we
go.  The current code is in auth-logging-ntlm in the catalyst git repo.

git://git.catalyst.net.nz/samba.git auth-logging-ntlm

> These should be a 2 minute clean-up in a rebase to add
> them to the now 2 authoritative places where this function
> now collects all the possible returns.
> 
> If you can point me at the current logging patchset I'll
> even do the rebase for you on top of this patch and send
> it as a new patchset to you.

Thanks, that would be most appreciated.

> I understand not wanting conflicts, but this one really
> should be trivial to solve.

I hope so - and the offer stands to apply it after:  I'm not objecting
to the patches, just trying to get as much audit logging done as
possible in the available time.  You know how long we have been waiting
for it :-)

Thanks,

Andrew Bartlett