Re: NT status/error codes instead of DOS error codes - bug 3.0.0

On Sun, Nov 30, 2003 at 01:12:40PM -0800, John Kleven wrote:
> K -- attached is trace of the problem from ethereal,
> plus a copy of logs in /var/log/samba.  I ran smbd
> with debug level 10.
> 
> It is weird because if an error actually occurs during
> sending packets (i.e., bad password or invalid share,
> etc) then smbd WILL return a DOS error code.  But for
> succesful responses it always returns NT status code.
> 
> BTW, it might be 'ok' for smbd to return an NT status
> code "success" to the negotiate protocol response
> because at that point, the client (my code) has not
> been able to even send the capabilities int yet.  But
> after the session setup w/ the capabilities, it should
> be illegal to return the NT status code - i think.

Success NT status codes are identical to DOS success
codes. Remember - success is eclass (16 bit) zero, ecode
(16 bit) zero.  NT status success is just (32 bit) zero
in the same field. Ah - I see. We're setting the NT status
codes bit on success return. I can easily fix this...

Jeremy.

Re: NT status/error codes instead of DOS error codes - bug 3.0.0

On Sun, Nov 30, 2003 at 01:12:40PM -0800, John Kleven wrote:
> K -- attached is trace of the problem from ethereal,
> plus a copy of logs in /var/log/samba.  I ran smbd
> with debug level 10.
> 
> It is weird because if an error actually occurs during
> sending packets (i.e., bad password or invalid share,
> etc) then smbd WILL return a DOS error code.  But for
> succesful responses it always returns NT status code.
> 
> BTW, it might be 'ok' for smbd to return an NT status
> code "success" to the negotiate protocol response
> because at that point, the client (my code) has not
> been able to even send the capabilities int yet.  But
> after the session setup w/ the capabilities, it should
> be illegal to return the NT status code - i think.

Ok, can you try this patch please ? I think this will
fix it.

Jeremy.

? client_oplocks.txt
? security_server_segfault.patch
? lib/smbldap.c.mod
Index: smbd/process.c
===================================================================
RCS file: /data/cvs/samba/source/smbd/process.c,v
retrieving revision 1.92.2.27

Bug 660 (smbd crash)

Can anyone with a Solaris 9 box reproduce

  https://bugzilla.samba.org/show_bug.cgi?id=660

Thanks.

cheers, jerry
 ----------------------------------------------------------------------
 Hewlett-Packard            ------------------------- http://www.hp.com
 SAMBA Team                 ---------------------- http://www.samba.org
 GnuPG Key                  ---- http://www.plainjoe.org/gpg_public.asc
 "If we're adding to the noise, turn off this song" --Switchfoot (2003)

Re: CVS update: samba4/source/ntvfs

Jelmer Vernooij wrote:

>On Thu, Nov 27, 2003 at 01:37:44PM +0100, Stefan Metzmacher wrote about 'Re: CVS update: samba4/source/ntvfs':
>  
>
>>Jelmer Vernooij wrote:
>>    
>>
>>>Isn't that what the ntvfs_interface_version function is supposed to
>>>be used for? A backend can check only the sizes of the structs it uses
>>>and ignore the other ones.
>>>      
>>>
>>hmm, should we really let the module decide that?
>>    
>>
>
>  
>
>>I think the module should be able to check the sizes and then decide 
>>which funtions it want to register.
>>But a module can also not check for sizes and just register it's 
>>functions for an old samba version.
>>And then we segfault. (tridge I know that segfaulting is good to find 
>>bugs but I think this is no bug! and we should check this)
>>also we should check the the version in
>>ntvfs_interface_version()
>>    
>>
>Even then, you shouldn't store the interface version in the ops

initgroups() system call in smbd child process doesn't get suppliementary group info from LDAP

Hi Experts,

Here is my setup:
================
* OS:
	Solaris 9 with Native LDAP client enabled

* Samba: 
	2.2.8a compiled with OpenLDAP 2.1.22:
	# CPPFLAGS="-I/usr/local/openldap_22/include" LDFLAGS="-L/usr/local/openldap_22/lib"
./configure --prefix=/usr/local/samba_2.2.8_ldap22 --with-ldapsam

* OpenLDAP 2.1.22

* iPlanet Directory Server 5.1 is setup to as SAM

The problem is:
===============
With patch 112960-03 and below, everything works fine. With patch newer than 112960-03, Samba cann't get
the supplementary group information for a user from directory server. Therefore, the user gets access
denied when access to those files with supplementary group permission.

More testing and data below:
============================

There is a user which primary group id is 513 and the supplementary group id is 512.

==> With patch 112960-03 and below

	a) Debug log indicates that 2 groups are returned by sys_getgroups()

Port 445 and multiple NBSS-Session request and Multiple smbd process

Hi,


We are facing below problem in the point and print scenario.Uploading
the drivers onto samba server and downloading onto windows client is
successful

    1.But while doing large printing(around 100 jobs) from one
PC,multiple smbds are getting spawned for  that one PC.
        -What we could observe is that, there are multiple NBSS-Session
request coming from windows client and for one session request there is
smbd process.
        -One NBSS-session request is always  happening after a RST had
happened for port 445.(As far as we know port 445 is supported only in
3.0)

      I would like to know why there is multiple NBSS-Session request
for single PC while doing pandp printing.
      
    Our undertanding is that there will be one smbd process for each PC
when connecting with IP and if we are connection with netbiosname there
will be one more smbd process.

    We would like to know why this is happening and how  to avoid
this?ie we want to have only one process running for one PC.We have "max
smbd process" configured due to our system restrictions.If multiple smbd
process are coming up for single PC,then it will create a problem to our
system.

Re: NT status/error codes instead of DOS error codes - bug 3.0.0

> The FLAGS2 bit indicates what the client or server *has done* in the given
> packet, not what the client or server *can do*.  There are some odd cases
> in
> which a Windows server will send NT error codes even though the client
> asked
> for DOS codes.  These situations are rare (jCIFS doesn't do NT codes and
> has
> not yet--as far as I know--excersized any of the odd cases in which
> Windows
> will send NT codes when it shouldn't), but they do exists.

I don't recall if I have ever seen that. If I have I could not claim that
it  wasn't Samba. Here are two useless facts that you can add to your
collection of shifty SMB behavior:

1) With NT 4.0, a redundant negotiate with Flags2 NTSTATUS32 on returns
with Flags2 NTSTATUS32 on but with a dos errorcode ERRSVR/ERRerror (which
is what you would get had DOS error codes been requested).

2) With NT 4.0, with an STATUS_DFS_PATH_NOT_COVERED condition, if DOS
error codes were negotiated, an ERRSRV/ERRreserved is returned which near
as I can tell works just fine for triggering DFS referrals from a wide
variety of operations.

Mike

--

-- 
A program should be written to  model the concepts of the task it
performs rather than the physical world or a process because this
maximizes the  potential for it  to be applied  to tasks that are

wbinfo -c duplicate entry

Hi,

With "wbinfo -c user", a entry is created in winbind_idmap.tdb. But if I
retype the same command with same username a second entry (with a
different uid) is created in tdb file.

Is behavior normal ?

See you,

-- 
Raphaël Berghmans <rberghmans <at> arafox.com>

--

-- 
To unsubscribe from this list go to the following URL and read the
instructions:  http://lists.samba.org/mailman/listinfo/samba

Re: CVS update: samba4/source/ntvfs

On Mon, Dec 01, 2003 at 07:51:21AM +0100, Stefan Metzmacher wrote about 'Re: CVS update: samba4/source/ntvfs':
> Jelmer Vernooij wrote:
> >On Thu, Nov 27, 2003 at 01:37:44PM +0100, Stefan Metzmacher wrote about 
> >'Re: CVS update: samba4/source/ntvfs':
> >>Jelmer Vernooij wrote:
> >>>Isn't that what the ntvfs_interface_version function is supposed to
> >>>be used for? A backend can check only the sizes of the structs it uses
> >>>and ignore the other ones.
> >>hmm, should we really let the module decide that?
Yes. The backend nows best what API changes would break it. The whole
version_info() function becomes obsolete if we force anything.

> >Even then, you shouldn't store the interface version in the ops
> >struct.
> then I would also not store the name on the ops struct!
Why not? The name is relevant to the subsystem (it has to know which
module to use), while the version is only interesting to the module
itself.

It's about time I start to do some actual Samba programming again,
instead of discussing it 

Jelmer

--

-- 
Jelmer Vernooij                                              <jelmer <at> samba.org>
http://samba.org/~jelmer/                             http://samba.vernstok.nl/
Bugs in bugzilla: 28 

Re: crash bugs 809, 812, & 830

Hello Jerry!

I have got a similar problem to what is described in the bugs..... After 
  searching, I found that printing may be a problem.

-> When connecting / printing via Windows 2003 Server (english) things 
are fine.

-> When connecting / printing via Windows 2000 Pro (german) Samba 
crashes. It crashes if you either just print a file or if you just open 
the print spooler to view the print-queue. Samba seems to work fine 
after the crash, and the file is printed, too.

-> Sadly I do not have a  Windows 2000 Pro in english.

I have not yet had time to make a level 10 logfile for analysis, but I 
may do this today / tomorrow. Unfortunately I do not have time to 
reproduce the error on a test-environment.

If I can help you with anything else, beside from the level-10 log and 
the smb.conf, please let me know.

Thanks,

	Torsten.

Gerald (Jerry) Carter wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>