svn commit: samba r6219 - in branches/SAMBA_4_0/source: librpc/rpc ntvfs/posix

[Richard Sharpe]

On Wed, 6 Apr 2005, Andrew Bartlett wrote:

> On Tue, 2005-04-05 at 19:53 +0000, sharpe at samba.org wrote:
> > Author: sharpe
> > Date: 2005-04-05 19:53:07 +0000 (Tue, 05 Apr 2005)
> > New Revision: 6219
> >
> > WebSVN: http://websvn.samba.org/cgi-bin/viewcvs.cgi?view=rev&root=samba&rev=6219
> >
> > Log:
> >
> > This change allows us to fall back to authenticating without
> > DCERPC_SCHANNEL_128 if we fail. Thus, it allows us to work with Windows
> > NT DCs ...
>
> That patch is fine for now, but I'll rework things to avoid as much of
> the reconnect as possible.

I tried doing it in dcerpc_schannel_key by simply trying the
ServerAuthenticate2 with a different set of negotiate_flags, but that
failed.

> More generally to the list - should 'quality of protection' be bundled
> in with the credentials code?  I'm wondering about abstracting the
> GENSEC 'want' and 'have' features back into credentials, the logic being
> 'if you want to use this password, you must meet this criteria'.  Being
> in the credentials code should also allow it to be used for 'basic'
> session setups.
>
> The defaults would of course come from the config file, where we would
> want 128 bit signed connections always, but accept 56.

I want to put the use client schannel = no stuff back as well, so I can
turn off schannel if needed ...

> Finally,
>
> > Modified: branches/SAMBA_4_0/source/ntvfs/posix/pvfs_search.c
> > ===================================================================
> > --- branches/SAMBA_4_0/source/ntvfs/posix/pvfs_search.c	2005-04-05 17:49:16 UTC (rev 6218)
> > +++ branches/SAMBA_4_0/source/ntvfs/posix/pvfs_search.c	2005-04-05 19:53:07 UTC (rev 6219)
> > @@ -266,7 +266,7 @@
> >  		talloc_free(file);
> >  	}
> >
> > -	pvfs_list_hibernate(dir);
> > +	/*pvfs_list_hibernate(dir);*/
> >
> >  	return NT_STATUS_OK;
> >  }
>
> Was this part intentional?

This relates to the issue that large directories cause client hangs. I
tried to avoid it ending up in the commit, but failed.

I have pledged to work on a fix for that, and have started looking at
Jeremy's fix in Samba 3 for FindFirst and FindNext and dir pointers as a
way to deal with this problem.

Regards
-----
Richard Sharpe, rsharpe[at]richardsharpe.com, rsharpe[at]samba.org,
sharpe[at]ethereal.com, http://www.richardsharpe.com