sfrench/samba-autobuild/.git
20 years agoFix swatdir location for --with-fhs
Jelmer Vernooij [Sat, 10 Jan 2004 22:05:38 +0000 (22:05 +0000)]
Fix swatdir location for --with-fhs

20 years agostarting new version of release notes -- post 3.0.2pre1
Gerald Carter [Fri, 9 Jan 2004 16:02:42 +0000 (16:02 +0000)]
starting new version of release notes -- post 3.0.2pre1

20 years agofix some warnings from the Sun compiler
Gerald Carter [Fri, 9 Jan 2004 14:54:33 +0000 (14:54 +0000)]
fix some warnings from the Sun compiler

20 years agoRomve debugging assertions (oops...)
Andrew Bartlett [Fri, 9 Jan 2004 02:47:35 +0000 (02:47 +0000)]
Romve debugging assertions (oops...)

Andrew Bartlett

20 years agoAdd a new type of name lookup 'ads'. This seperates this from normal
Andrew Bartlett [Fri, 9 Jan 2004 02:38:58 +0000 (02:38 +0000)]
Add a new type of name lookup 'ads'.  This seperates this from normal
hostname lookups, and ensures that we don't lookup 'short' (ie NetBIOS)
domain names in DNS.

Andrew Bartlett

20 years agoUse StrCaseCmp, not strcasecmp. Should fix winbind build on IRIX
Jelmer Vernooij [Thu, 8 Jan 2004 22:53:37 +0000 (22:53 +0000)]
Use StrCaseCmp, not strcasecmp. Should fix winbind build on IRIX

20 years agofix segfault when sid_ptr == 0 in DsEnumDomainTrusts() reply
Gerald Carter [Thu, 8 Jan 2004 22:21:29 +0000 (22:21 +0000)]
fix segfault when sid_ptr == 0 in DsEnumDomainTrusts() reply

20 years agofix a seg fault caused by abartlet's last checkin; there's no way this could have...
Gerald Carter [Thu, 8 Jan 2004 19:56:40 +0000 (19:56 +0000)]
fix a seg fault caused by abartlet's last checkin; there's no way this could have been tested against an NT4 DC

20 years agoWe might not have the 'samba' directory in the samba_3_0 build.
Andrew Bartlett [Thu, 8 Jan 2004 08:44:39 +0000 (08:44 +0000)]
We might not have the 'samba' directory in the samba_3_0 build.

Andrew Bartlett

20 years agouse SAFE_FREE(), not free().
Andrew Bartlett [Thu, 8 Jan 2004 08:41:26 +0000 (08:41 +0000)]
use SAFE_FREE(), not free().

Andrew Bartlett

20 years agoThis merges in my 'always use ADS' patch. Tested on a mix of NT and ADS
Andrew Bartlett [Thu, 8 Jan 2004 08:19:18 +0000 (08:19 +0000)]
This merges in my 'always use ADS' patch.  Tested on a mix of NT and ADS
domains, this patch ensures that we always use the ADS backend when
security=ADS, and the remote server is capable.

The routines used for this behaviour have been upgraded to modern Samba
codeing standards.

This is a change in behaviour for mixed mode domains, and if the trusted
domain cannot be reached with our current krb5.conf file, we will show
that domain as disconnected.

This is in line with existing behaviour for native mode domains, and for
our primary domain.

As a consequence of testing this patch, I found that our kerberos error
handling was well below par - we would often throw away useful error
values.  These changes move more routines to ADS_STATUS to return
kerberos errors.

Also found when valgrinding the setup, fix a few memory leaks.

While sniffing the resultant connections, I noticed we would query our
list of trusted domains twice - so I have reworked some of the code to
avoid that.

Andrew Bartlett

20 years agoIn tdb_allocate(), we would create a new record by writing a local variable
Andrew Bartlett [Thu, 8 Jan 2004 05:37:23 +0000 (05:37 +0000)]
In tdb_allocate(), we would create a new record by writing a local variable
'newrec' into the tdb.

This was not initialised, so valgrind warned about it.

(Note:  valgrind only makes sense on tdbs with 'mmap = no' in your smb.conf)

Andrew Bartlett

20 years agoMake it clearer that the domain here is the domain of the user for
Andrew Bartlett [Thu, 8 Jan 2004 02:57:42 +0000 (02:57 +0000)]
Make it clearer that the domain here is the domain of the user for
authentication.

Andrew Bartlett

20 years agoMove more of winbind to use 'find_our_domain()' rather than the dangerous
Andrew Bartlett [Thu, 8 Jan 2004 02:15:46 +0000 (02:15 +0000)]
Move more of winbind to use 'find_our_domain()' rather than the dangerous
find_domain_from_name(lp_workgroup()).

(as find_domain_from_name() can change the data in lp_workgroup())

Andrew Bartlett

20 years agoThe correct test for 'is our primary domain' is domain->primary
Andrew Bartlett [Thu, 8 Jan 2004 00:55:13 +0000 (00:55 +0000)]
The correct test for 'is our primary domain' is domain->primary

20 years agoMachines are people too!
Andrew Bartlett [Wed, 7 Jan 2004 23:46:47 +0000 (23:46 +0000)]
Machines are people too!

While machine accounts cannot use an NTLM login (NT4 style), they are
otherwise full and valid members of the domain, and expect to be able to
use kerberos to connect to CIFS servers.

This means that the LocalSystem account, used by various services, can
perform things like backups, without the admin needing to enter further
passwords.

This particular issue (bug 722) has started to come up a lot on the lists.

I have only enabled it for winbindd-based systems, as the macros use use
to call the 'add user script' will strip the $ from the username for
security reasons.

Andrew Bartlett

20 years agoFix for bug #922. Fast path not called for strlower_m() and strupper_m().
Jeremy Allison [Wed, 7 Jan 2004 23:21:36 +0000 (23:21 +0000)]
Fix for bug #922. Fast path not called for strlower_m() and strupper_m().
From ab@samba.org (Alexander Bokovoy).
Jeremy.

20 years agoTypo fix.
Rafal Szczesniak [Wed, 7 Jan 2004 22:44:28 +0000 (22:44 +0000)]
Typo fix.

rafal

20 years agoDoxygen comment fix.
Rafal Szczesniak [Wed, 7 Jan 2004 22:43:36 +0000 (22:43 +0000)]
Doxygen comment fix.

rafal

20 years agoFix from Justin Baugh <justin.baugh@request.com> for bugid #948 for
Jeremy Allison [Wed, 7 Jan 2004 21:08:42 +0000 (21:08 +0000)]
Fix from Justin Baugh <justin.baugh@request.com> for bugid #948 for
FreeBSD winbindd.
Jeremy.

20 years agocommiting jra's fix for Exchange clear test auth
Gerald Carter [Wed, 7 Jan 2004 19:57:08 +0000 (19:57 +0000)]
commiting jra's fix for Exchange clear test auth

20 years agoFix from Luke Howard <lukeh@PADL.COM> for incorrect early free().
Jeremy Allison [Wed, 7 Jan 2004 19:55:01 +0000 (19:55 +0000)]
Fix from Luke Howard <lukeh@PADL.COM> for incorrect early free().
Jeremy.

20 years agoDon't duplicate pulling the 'IPC' username from secrets.tdb, instead
Andrew Bartlett [Wed, 7 Jan 2004 10:11:24 +0000 (10:11 +0000)]
Don't duplicate pulling the 'IPC' username from secrets.tdb, instead
just use one function for both places.

Andrew Bartlett

20 years agoThere is a German translation of swat -- surprise :-)
Volker Lendecke [Wed, 7 Jan 2004 10:02:10 +0000 (10:02 +0000)]
There is a German translation of swat -- surprise :-)

Fix some msgs

Volker

20 years agoAdd smbget utility, a simple wget-like utility that uses libsmbclient.
Jelmer Vernooij [Wed, 7 Jan 2004 00:43:52 +0000 (00:43 +0000)]
Add smbget utility, a simple wget-like utility that uses libsmbclient.
Supports recursive downloads and resume, progress indication and shows
estimated time remaining.

20 years agoFix segfualt caused by incorrect configuration. If lp_realm() was not set,
Andrew Bartlett [Tue, 6 Jan 2004 23:57:12 +0000 (23:57 +0000)]
Fix segfualt caused by incorrect configuration.  If lp_realm() was not set,
but security=ADS, we would attempt to free the principal name that krb5
never allocated.

Also fix the dump_data() of the session key, now that we use a data_blob to
store that.

Andrew Bartlett

20 years agoPatch penguin. Cleaning out old mbp patch.
Jeremy Allison [Tue, 6 Jan 2004 22:34:06 +0000 (22:34 +0000)]
Patch penguin. Cleaning out old mbp patch.
Jeremy.

20 years agomore commits logged
Gerald Carter [Tue, 6 Jan 2004 20:56:59 +0000 (20:56 +0000)]
more commits logged

20 years agoWrite bug number like in the rest of the file
Jelmer Vernooij [Tue, 6 Jan 2004 20:03:34 +0000 (20:03 +0000)]
Write bug number like in the rest of the file

20 years agoFix -s option to smbcontrol (#908)
Jelmer Vernooij [Tue, 6 Jan 2004 20:01:48 +0000 (20:01 +0000)]
Fix -s option to smbcontrol (#908)

20 years agoremove unused seek_file(); don't hardcode '\' when printing the auth-user
Gerald Carter [Tue, 6 Jan 2004 19:57:14 +0000 (19:57 +0000)]
remove unused seek_file(); don't hardcode '\' when printing the auth-user

20 years agobumping to 3.0.2pre2
Gerald Carter [Tue, 6 Jan 2004 19:14:22 +0000 (19:14 +0000)]
bumping to 3.0.2pre2

20 years agomore commit logs
Gerald Carter [Tue, 6 Jan 2004 19:05:23 +0000 (19:05 +0000)]
more commit logs

20 years ago* making sure contributors are listed in alphabetical order
Gerald Carter [Tue, 6 Jan 2004 18:42:09 +0000 (18:42 +0000)]
* making sure contributors are listed in alphabetical order
* adding jra's fix for bug 815

20 years agoisolate ldap debug messages to the common smbldap_XXX() functions
Gerald Carter [Tue, 6 Jan 2004 18:26:18 +0000 (18:26 +0000)]
isolate ldap debug messages to the common smbldap_XXX() functions

20 years agoXFS quota patch from Stefan Metzmacher <metze@metzemix.de>.
Jeremy Allison [Tue, 6 Jan 2004 18:13:32 +0000 (18:13 +0000)]
XFS quota patch from Stefan Metzmacher <metze@metzemix.de>.
Jeremy.

20 years agoUpdates for pread/pwrite code.
Jeremy Allison [Tue, 6 Jan 2004 17:53:34 +0000 (17:53 +0000)]
Updates for pread/pwrite code.
Jeremy.

20 years agoCorrectly detect AFS headers on SuSE in /usr/include/afs/afs/
Volker Lendecke [Tue, 6 Jan 2004 15:41:32 +0000 (15:41 +0000)]
Correctly detect AFS headers on SuSE in /usr/include/afs/afs/

Volker

20 years agofix case in objectclass name (not that it really matters); patch from Darren Chew...
Gerald Carter [Tue, 6 Jan 2004 14:40:35 +0000 (14:40 +0000)]
fix case in objectclass name (not that it really matters); patch from Darren Chew <darrenc@vicscouts.asn.au>

20 years agoPatch by Stefan Metzmacher <metze@metzemix.de>:
Andrew Bartlett [Tue, 6 Jan 2004 10:22:13 +0000 (10:22 +0000)]
Patch by Stefan Metzmacher <metze@metzemix.de>:

here's a small fix that fixes the new quota system on irix.

I need to reanable XFS quotas on irix for the new quota system
(Jerry do you want to wait for this for the release ?)

But the old system works and is the default on irix!

20 years agoFix typo..
Volker Lendecke [Tue, 6 Jan 2004 07:57:35 +0000 (07:57 +0000)]
Fix typo..

Volker

20 years agoPatch based on work from James Peach <jpeach@sgi.com> to convert over to
Jeremy Allison [Tue, 6 Jan 2004 01:22:14 +0000 (01:22 +0000)]
Patch based on work from James Peach <jpeach@sgi.com> to convert over to
using pread/pwrite. Modified a little to ensure fsp->pos is correct.
Fix for #889.
Jeremy.

20 years agoEnsure that for wbinfo --set-auth-user, we actually use the domain.
Andrew Bartlett [Tue, 6 Jan 2004 00:32:24 +0000 (00:32 +0000)]
Ensure that for wbinfo --set-auth-user, we actually use the domain.

Andrew Bartlett

20 years agocifs mount helper merge
Steve French [Mon, 5 Jan 2004 22:18:44 +0000 (22:18 +0000)]
cifs mount helper merge

20 years agoworking on new format for relerase notes in 3.0.2pre1
Gerald Carter [Mon, 5 Jan 2004 21:51:01 +0000 (21:51 +0000)]
working on new format for relerase notes in 3.0.2pre1

20 years agoFix more cases to ensure that as a server, we don't complain to the client
Andrew Bartlett [Mon, 5 Jan 2004 21:24:27 +0000 (21:24 +0000)]
Fix more cases to ensure that as a server, we don't complain to the client
about our server-side lack of session key.

Andrew Bartlett

20 years agoAdded last missing file.
Jeremy Allison [Mon, 5 Jan 2004 21:03:12 +0000 (21:03 +0000)]
Added last missing file.
Jeremy.

20 years agoOops. Broke the build. Added missing files.
Jeremy Allison [Mon, 5 Jan 2004 21:02:37 +0000 (21:02 +0000)]
Oops. Broke the build. Added missing files.
Jeremy.

20 years agoFix from James Flemer <jflemer@uvm.edu> to make HAVE_ATTR_LIST linked to
Jeremy Allison [Mon, 5 Jan 2004 21:01:08 +0000 (21:01 +0000)]
Fix from James Flemer <jflemer@uvm.edu> to make HAVE_ATTR_LIST linked to
HAVE_SYS_ATTRIBUTES_H to fix AIX compile.
Jeremy.

20 years agofix inverted check using krb5_kt_resolve() and HAVE_MEMORY_KEYTAB; bug 912
Gerald Carter [Mon, 5 Jan 2004 20:23:56 +0000 (20:23 +0000)]
fix inverted check using krb5_kt_resolve() and HAVE_MEMORY_KEYTAB; bug 912

20 years agoPatch from Stefan (metze) Metzmacher <metze at metzemix.de> to revert to 2.2.x quota...
Jeremy Allison [Mon, 5 Jan 2004 19:36:02 +0000 (19:36 +0000)]
Patch from Stefan (metze) Metzmacher <metze at metzemix.de> to revert to 2.2.x quota methods.

:-).

"here's a patch which ports the samba 2.2 samba_linux_quota.h stuff to 3_0.

This is needed because of so many broken quota files outthere.

Please, test this with old, new kernels
(strucr dqblk, struct mem_dqblk, and struct if_dqblk)
, quota.user, aquota.user formats

what is when a user is over soft quota and over hard quotas..."

Jeremy.

20 years agoEnsure we set "always sign" flag if set. We don't currently do anything with
Jeremy Allison [Mon, 5 Jan 2004 19:21:06 +0000 (19:21 +0000)]
Ensure we set "always sign" flag if set. We don't currently do anything with
this but we should log the fact it was negotiated.
Jeremy.

20 years agoFix warning
Volker Lendecke [Mon, 5 Jan 2004 16:58:37 +0000 (16:58 +0000)]
Fix warning

Volker

20 years agoDon't free the encrypted_session_key early - that causes the subsequent
Andrew Bartlett [Mon, 5 Jan 2004 12:36:21 +0000 (12:36 +0000)]
Don't free the encrypted_session_key early - that causes the subsequent
test for a valid length to fail...

This should fix 'security=server' and hosts-equiv failures picked up by
the build farm.

Andrew Bartlett

20 years agoshorten some more lines.
Andrew Bartlett [Mon, 5 Jan 2004 12:21:04 +0000 (12:21 +0000)]
shorten some more lines.

20 years agoTry to keep vl happy - shorten some of these lines.
Andrew Bartlett [Mon, 5 Jan 2004 12:20:15 +0000 (12:20 +0000)]
Try to keep vl happy - shorten some of these lines.

20 years agoGrumble... grumble... fix the build...
Andrew Bartlett [Mon, 5 Jan 2004 05:07:59 +0000 (05:07 +0000)]
Grumble... grumble... fix the build...

20 years agoShow the sid type in name->sid translatons in a way that can be easily
Andrew Bartlett [Mon, 5 Jan 2004 04:26:35 +0000 (04:26 +0000)]
Show the sid type in name->sid translatons in a way that can be easily
understood by humans.

Andrew Bartlett

20 years agoAlways call the auto-init funciton - this avoids tdb segfaulting under
Andrew Bartlett [Mon, 5 Jan 2004 04:15:55 +0000 (04:15 +0000)]
Always call the auto-init funciton - this avoids tdb segfaulting under
us if we failed to open it earlier.

Andrew Bartlett

20 years agoCorrectly handle per-pipe NTLMSSP inside a NULL session. Previously we
Andrew Bartlett [Mon, 5 Jan 2004 04:12:40 +0000 (04:12 +0000)]
Correctly handle per-pipe NTLMSSP inside a NULL session.  Previously we
would attempt to supply a password to the 'inside' NTLMSSP, which the
remote side naturally rejected.

Andrew Bartlett

20 years agoChange our Domain controller lookup routines to more carefully seperate
Andrew Bartlett [Mon, 5 Jan 2004 04:10:28 +0000 (04:10 +0000)]
Change our Domain controller lookup routines to more carefully seperate
DNS names (realms) from NetBIOS domain names.

Until now, we would experience delays as we broadcast lookups for DNS names
onto the local network segments.

Now if DNS comes back negative, we fall straight back to looking up the
short name.

Andrew Bartlett

20 years agoFix typo in RW2 torture test. Closes bugzilla bug #924.
Tim Potter [Mon, 5 Jan 2004 02:57:33 +0000 (02:57 +0000)]
Fix typo in RW2 torture test.  Closes bugzilla bug #924.

20 years agoAdd const.
Andrew Bartlett [Mon, 5 Jan 2004 02:16:51 +0000 (02:16 +0000)]
Add const.

20 years agoThere is some memory corruption hidden somewhere in our winbind code. If I
Andrew Bartlett [Mon, 5 Jan 2004 02:12:38 +0000 (02:12 +0000)]
There is some memory corruption hidden somewhere in our winbind code.  If I
could reproduce it, I would fix it, but for now just make sure we always
SAFE_FREE() and set our starting pointers to NULL.

Andrew Bartlett

20 years agoChange (unused) structure parameter for cli_ds_enum_domain_trusts() cleanup.
Andrew Bartlett [Mon, 5 Jan 2004 02:05:19 +0000 (02:05 +0000)]
Change (unused) structure parameter for cli_ds_enum_domain_trusts() cleanup.

20 years agorpc_client/cli_lsarpc.c:
Andrew Bartlett [Mon, 5 Jan 2004 02:04:37 +0000 (02:04 +0000)]
rpc_client/cli_lsarpc.c:
rpc_parse/parse_lsa.c:
nsswitch/winbindd_rpc.c:
nsswitch/winbindd.h:
 - Add const

libads/ads_ldap.c:
 - Cleanup function for use

nsswitch/winbindd_ads.c:
 - Use new utility function ads_sid_to_dn
 - Don't search for 'dn=', rather call the ads_search_retry_dn()

nsswitch/winbindd_ads.c:
include/rpc_ds.h:
rpc_client/cli_ds.c:
 - Fixup braindamage in cli_ds_enum_domain_trusts():
    - This function was returning a UNISTR2 up to the caller, and
      was doing nasty (invalid, per valgrind) things with memcpy()
    - Create a new structure that represents this informaiton in a useful way
      and use talloc.

Andrew Bartlett

20 years agoFix for bug 707, getent group for huge ads groups (>1500 members)
Andrew Bartlett [Mon, 5 Jan 2004 01:48:21 +0000 (01:48 +0000)]
Fix for bug 707, getent group for huge ads groups (>1500 members)
This introduces range retrieval of ADS attributes.

VL rewrote most of Günther's patch, partly to remove code duplication and
partly to get the retrieval of members in one rush, not interrupted by the
lookups for the DN.

I rewrote that patch, to ensure that we can keep an eye on the USN
(sequence number) of the entry - this allows us to ensure the read was
atomic.

In particular, the range retrieval is now generic, for strings.  It
could easily be made generic for any attribute type, if need be.

Andrew Bartlett

20 years agoI'm not quite sure what happened here - but replace the ads_sid_to_dn
Andrew Bartlett [Mon, 5 Jan 2004 01:06:56 +0000 (01:06 +0000)]
I'm not quite sure what happened here - but replace the ads_sid_to_dn
function with one that compiles.

Andrew Bartlett

20 years agoWe can't possilby get 'ok' here, as the if statement above just checked for it.
Andrew Bartlett [Mon, 5 Jan 2004 00:15:34 +0000 (00:15 +0000)]
We can't possilby get 'ok' here, as the if statement above just checked for it.

20 years agoMake arbitary binary data unsigned char.
Andrew Bartlett [Mon, 5 Jan 2004 00:14:12 +0000 (00:14 +0000)]
Make arbitary binary data unsigned char.

20 years agoAdd a utilty function for converting a sid to a DN.
Andrew Bartlett [Mon, 5 Jan 2004 00:13:00 +0000 (00:13 +0000)]
Add a utilty function for converting a sid to a DN.

Andrew Bartlett

20 years agoMake it clear that we cannot sign if we don't have a session key.
Andrew Bartlett [Mon, 5 Jan 2004 00:11:35 +0000 (00:11 +0000)]
Make it clear that we cannot sign if we don't have a session key.

20 years agoAutomaticly initialise the signing engine, if we have a session key.
Andrew Bartlett [Mon, 5 Jan 2004 00:11:02 +0000 (00:11 +0000)]
Automaticly initialise the signing engine, if we have a session key.

20 years ago- Put functions for generating SQL queries in pdb_sql.c
Jelmer Vernooij [Sun, 4 Jan 2004 21:09:42 +0000 (21:09 +0000)]
- Put functions for generating SQL queries in pdb_sql.c
- Add pgSQL backend (based on patch by Hamish Friedlander)
- Use query generate functions from pdb_mysql and pdb_pgsql
- Only pdb_pgsql.c needs to be changed whenever the fields in SAM_ACCOUNT change

20 years agoCommit the translation of the realm to the netbios domain name in the kerberos
Volker Lendecke [Sun, 4 Jan 2004 11:51:31 +0000 (11:51 +0000)]
Commit the translation of the realm to the netbios domain name in the kerberos
session setup. After talking to jht and abartlet I made this unconditional, no
additional parameter.

Jerry: This is a change in behaviour, but I think it is necessary.

Volker

20 years agoEven if the 'device type' is always an ascii string, use push_string to get
Andrew Bartlett [Sun, 4 Jan 2004 11:05:30 +0000 (11:05 +0000)]
Even if the 'device type' is always an ascii string, use push_string to get
it out onto the wire.  Avoids valgrind warnings because the fstrcpy() causes
part of the wire buffer to be 'marked'.

Andrew Bartlett

20 years agoAnd yet another const
Volker Lendecke [Sat, 3 Jan 2004 20:20:59 +0000 (20:20 +0000)]
And yet another const

Volker

20 years agoThere is not a particularly good excuse for complaining to the *client* that
Andrew Bartlett [Sat, 3 Jan 2004 01:12:56 +0000 (01:12 +0000)]
There is not a particularly good excuse for complaining to the *client* that
it sent 'INVALID_PARAMETER', when it was us as the server that could not
come up with a session key.  Instead, allow normal authentication to take
place, but do not setup a session key.

Andrew Bartlett

20 years agoMatch Win2k, and return NT_STATUS_INVALID_PARAMETER
Andrew Bartlett [Fri, 2 Jan 2004 23:55:44 +0000 (23:55 +0000)]
Match Win2k, and return NT_STATUS_INVALID_PARAMETER
if this parameter is not an account type

Andrew Bartlett

20 years agoUnder certain error conditions (a talloc() failure above) this would cause
Andrew Bartlett [Fri, 2 Jan 2004 11:39:07 +0000 (11:39 +0000)]
Under certain error conditions (a talloc() failure above) this would cause
a double-free(), and the resultant malloc heap corruption.

This may be one of our lurking winbind segfaults.

Andrew Bartlett

20 years agoHaving no members of a group is a perfectly valid (if unusual) situation.
Andrew Bartlett [Fri, 2 Jan 2004 05:33:14 +0000 (05:33 +0000)]
Having no members of a group is a perfectly valid (if unusual) situation.

Andrew Bartlett

20 years agoJHT came up with a nasty (broken) torture case in preparing examples for
Andrew Bartlett [Fri, 2 Jan 2004 05:32:07 +0000 (05:32 +0000)]
JHT came up with a nasty (broken) torture case in preparing examples for
his book.

This prompted me to look at the code that reads the unix group list.  This
code did a lot of name -> uid -> name -> sid translations, which caused
problems.  Instead, we now do just name->sid

I also cleaned up some interfaces, and client tools.

Andrew Bartlett

20 years agoAfter talking with abartlet remove the fix for bug 707 again.
Volker Lendecke [Thu, 1 Jan 2004 21:10:35 +0000 (21:10 +0000)]
After talking with abartlet remove the fix for bug 707 again.

Volker

20 years agoFix for bug 707, getent group for huge ads groups (>1500 members)
Volker Lendecke [Thu, 1 Jan 2004 20:30:50 +0000 (20:30 +0000)]
Fix for bug 707, getent group for huge ads groups (>1500 members)
This introduces range retrieval of ADS attributes.

I've rewritten most of Günther's patch, partly to remove code duplication and
partly to get the retrieval of members in one rush, not interrupted by the
lookups for the DN.

Andrew, you told me that you would like to see a check whether the AD sequence
number is the same before and after the retrieval to achieve atomicity. This
would be trivial to add, but I'm not sure that we want this, as this adds two
roundtrips to every membership query. We can not know before the first query
whether we get additional range values, and at that point it's too late to ask
for the USN.

Tested with a group of 4000 members along with lots of small groups.

Volker

20 years agoChanges to our PAM code to cope with the fact that we can't handle some
Andrew Bartlett [Wed, 31 Dec 2003 08:45:03 +0000 (08:45 +0000)]
Changes to our PAM code to cope with the fact that we can't handle some
domains (in particular, the domain of the current machine, if it is not a PDC)

By changing the error codes, we now return values that PAM can correctly
use for better stacking of PAM modules - in particular of the password change
module.

This allows pam_winbind to co-exist with other pam modules for password changes.

Andrew Bartlett

20 years agoForgot to commit this for the 'get our primary domain' change.
Andrew Bartlett [Wed, 31 Dec 2003 08:42:22 +0000 (08:42 +0000)]
Forgot to commit this for the 'get our primary domain' change.

20 years agoJerry rightly complained that we can't assume that the first domain is
Andrew Bartlett [Wed, 31 Dec 2003 05:26:29 +0000 (05:26 +0000)]
Jerry rightly complained that we can't assume that the first domain is
our primary domain - new domains are added to the front of the list. :-(

Use a much more reliable 'flag test' instead.  (note:  changes winbind structures, make clean).

Andrew Bartlett

20 years agoauth/auth_util.c:
Andrew Bartlett [Wed, 31 Dec 2003 00:31:43 +0000 (00:31 +0000)]
auth/auth_util.c:
 - Fill in the 'backup' idea of a domain, if the DC didn't supply one.  This
   doesn't seem to occour in reality, hence why we missed the typo.

lib/charcnv.c:
lib/smbldap.c:
libads/ldap.c:
libsmb/libsmbclient.c:
printing/nt_printing.c:
 - all the callers to pull_utf8_allocate() pass a char ** as the first
   parammeter, so don't make them all cast it to a void **

nsswitch/winbind_util.c:
 - Allow for a more 'correct' view of when usernames should be qualified
   in winbindd.  If we are a PDC, or have 'winbind trusted domains only',
   then for the authentication returns stip the domain portion.
 - Fix valgrind warning about use of free()ed name when looking up our
   local domain.  lp_workgroup() is maniplated inside a procedure that
   uses it's former value.  Instead, use the fact that our local domain is
   always the first in the list.

Andrew Bartlett

20 years agoGet the DOMAIN\username around the right way (I had username\domain...)
Andrew Bartlett [Tue, 30 Dec 2003 22:27:33 +0000 (22:27 +0000)]
Get the DOMAIN\username around the right way (I had username\domain...)

Push the unix username into utf8 for it's trip across the socket.

Andrew Bartlett

20 years agoMove to short lived TALLOC_CTX* for allocating printer
Gerald Carter [Tue, 30 Dec 2003 22:17:14 +0000 (22:17 +0000)]
Move to short lived TALLOC_CTX* for allocating printer
objects from the print handle cache.   Fixes bug that
caused smbd to consume large amounts of RAM when

(a) a printer handle was kept open over an extended
    period of time, and
(b) the client issued frequent requests that resulted
    in a call to get_a_printer()

20 years agoAnother little one: Make pdb_test.c at least compile, although its way out of
Volker Lendecke [Tue, 30 Dec 2003 21:12:36 +0000 (21:12 +0000)]
Another little one: Make pdb_test.c at least compile, although its way out of
date.

Volker

20 years agoThe AFS pts command always generates completely lower-case user names. As case
Volker Lendecke [Tue, 30 Dec 2003 16:00:56 +0000 (16:00 +0000)]
The AFS pts command always generates completely lower-case user names. As case
is not significant in windows user names we should not lose information by
lower-casing the name before handing it to AFS.

Volker

20 years agoFix Bug # 924
Volker Lendecke [Tue, 30 Dec 2003 15:18:25 +0000 (15:18 +0000)]
Fix Bug # 924

Volker

20 years agoTry to gain a bit more consistancy in the output of usernames from ntlm_auth:
Andrew Bartlett [Tue, 30 Dec 2003 13:20:39 +0000 (13:20 +0000)]
Try to gain a bit more consistancy in the output of usernames from ntlm_auth:

Instead of returning a name in DOMAIN\user format, we now return it in the
same way that nsswtich does - following the rules of 'winbind use default
domain', in the correct case and with the correct seperator.

This should help sites who are using Squid or the new SASL code I'm working
on, to match back to their unix usernames.

Andrew Bartlett

20 years agoMake the name of the NTLMSSP client more consistant before we lock it in stone.
Andrew Bartlett [Tue, 30 Dec 2003 08:52:46 +0000 (08:52 +0000)]
Make the name of the NTLMSSP client more consistant before we lock it in stone.

20 years agoRemove testing hack
Andrew Bartlett [Tue, 30 Dec 2003 07:38:32 +0000 (07:38 +0000)]
Remove testing hack

20 years agoMove our basic password checking code from inside the authentication
Andrew Bartlett [Tue, 30 Dec 2003 07:33:58 +0000 (07:33 +0000)]
Move our basic password checking code from inside the authentication
subsystem into a seperate file - ntlm_check.c.

This allows us to call these routines from ntlm_auth.  The purpose of this
exercise is to allow ntlm_auth (when operating as an NTLMSSP server) to
avoid talking to winbind.  This should allow for easier debugging.

ntlm_auth itself has been reorgainised, so as to share more code between
the SPNEGO-wrapped and 'raw' NTLMSSP modes.  A new 'client' NTLMSSP mode
has been added, for use with a Cyrus-SASL module I am writing (based on vl's
work)

Andrew Bartlett

20 years agoRefactor our authentication and authentication testing code.
Andrew Bartlett [Tue, 30 Dec 2003 05:02:32 +0000 (05:02 +0000)]
Refactor our authentication and authentication testing code.

The next move will be to remove our password checking code from the SAM
authentication backend, and into a file where other parts of samba can use
it.

The ntlm_auth changes provide for better use of common code.

Andrew Bartlett

20 years agoAdd the alignment required before all 2-byte quantities in NDR. Allows us
Andrew Bartlett [Mon, 29 Dec 2003 04:21:32 +0000 (04:21 +0000)]
Add the alignment required before all 2-byte quantities in NDR.  Allows us
to correctly parse plaintext netlogon calls with odd-length passwords

Andrew Bartlett

20 years agoShutting down the connection closes outstanding sessions, so we don't need
Andrew Bartlett [Sun, 28 Dec 2003 09:57:29 +0000 (09:57 +0000)]
Shutting down the connection closes outstanding sessions, so we don't need
to do it twice...

Amdrew Bartlett