[PATCH] samba-tool ntacl do a proper VFS connect

[Jeremy Allison]

On Fri, Jan 04, 2013 at 12:48:28PM +1100, Andrew Bartlett wrote:
> On Fri, 2012-12-28 at 17:37 -0800, Jeremy Allison wrote:
> > On Fri, Dec 28, 2012 at 04:35:00PM +1100, Andrew Bartlett wrote:
> > > On Fri, 2012-12-28 at 15:07 +1100, Andrew Bartlett wrote:
> > > > Attached for review is the small collection of patches I've been working
> > > > on over the holidays, now that I have written some tests for them.
> > > > 
> > > > Most of the substantive changes have been posted to one or other list,
> > > > or in bugzilla already, but what I hadn't asked for yet is specific
> > > > review (and optionally push). 
> > > > 
> > > > Here are the patches to have samba-tool ntacl do a proper VFS
> > > > connect to a specific share, which makes it give less warnings and uses
> > > > the VFS layer more correctly.
> > > > 
> > > > I know most folks are busy with family over this holiday time, but if I
> > > > could get some review on these changes, i would appreciate it.
> > > 
> > > On Simo's request, the attached revised patches try to clarify why the
> > > crate_conn_struct() function was split into create_conn_struct() and
> > > create_conn_struct_cwd().  
> > > 
> > > Simo, I would rather not make it create_conn_struct_nocwd() instead,
> > > because in the long term this is clearer, and it makes it clear in the
> > > callers that this code changes the current working directory (which
> > > seems to have surprised Jeremy, who wrote in the function header
> > > comment:
> > > 
> > > Note: this performs a vfs connect and CHANGES CWD !!!! JRA.
> > 
> > Actually it didn't surprise me -  I wrote that comment to
> > make sure that anyone else looking at it knew that it had
> > a BIG side effect.
> > 
> > > If we have to have two names (which we do), encoding this clearly in the
> > > name seems the most reasonable thing.
> > 
> > I'd like to look these patches over very carefully, but
> > I won't be back in at work fulltime until Jan 3rd (Thursday).
> > 
> > Sorry for the delay, family vacation has to come first :-).
> 
> Family time is important, and I also really appreciate you looking into
> the share access issue.   If you could look over these patches when you
> get a chance, it will help us not only clean up part of the annoying
> messages during make test, it will make it easier to finish and land the
> other VFS ACL changes (because Christian will be able to use samba-tool
> ntacl for testing his GPFS changes). 
> 
> Besides the choice of function names, I really don't there is anything
> particularly special in here, just one step towards making this all work
> a little better now the pressure of 4.0.0 is behind us. 

This will be second on my list once I've gotten the share access issue
fixed. That's a tricky one to get right as I'm sure you appreciate, so
I'm still working on code and testing for the patch.

Jeremy.