12 years agos4:operational - Remove some outdated comments
Matthias Dieter Wallnöfer [Tue, 11 Aug 2009 09:08:05 +0000 (11:08 +0200)]
s4:operational - Remove some outdated comments

12 years agoldb: Don't break the standalone LDB build (operational module removed)
Matthias Dieter Wallnöfer [Tue, 11 Aug 2009 09:02:03 +0000 (11:02 +0200)]
ldb: Don't break the standalone LDB build (operational module removed)

12 years agos4:test for "primaryGroupToken"
Matthias Dieter Wallnöfer [Tue, 11 Aug 2009 10:25:13 +0000 (12:25 +0200)]
s4:test for "primaryGroupToken"

Tests for the right behaviour of this introduced constructed attribute.
Since we don't support the read-only-ness of those attributes yet, I commented
some lines out.
Also I had to add a function for python which converts domain SIDs in RIDs.
And a small fix for the "groupType" test.

12 years agos4:samldb module - Remove duplicate line
Matthias Dieter Wallnöfer [Mon, 10 Aug 2009 11:19:18 +0000 (13:19 +0200)]
s4:samldb module - Remove duplicate line

12 years agos4:operational module - move and enhancements
Matthias Dieter Wallnöfer [Mon, 10 Aug 2009 11:16:41 +0000 (13:16 +0200)]
s4:operational module - move and enhancements

This moves the "operational" LDB module to the right place under "dsdb/samdb/ldb_modules"
(suggested by abartlet) and enhances it for supporting dynamic generated
"primaryGroupToken" for AD groups. This should fix bug #6466.

12 years agos4:pwsettings script - Fix a small glitch
Matthias Dieter Wallnöfer [Mon, 10 Aug 2009 09:06:33 +0000 (11:06 +0200)]
s4:pwsettings script - Fix a small glitch

This fixes the problem with the setting and getting of the "minPwdAge" and
"maxPwdAge" attributes. I wanted to handle them in days but forgot to add
conversions (from "ticks" (tenth of microsecond) -> "days" and backwards).

12 years agoRename ASN1_BITFIELD to ASN1_BIT_STRING.
Kouhei Sutou [Mon, 10 Aug 2009 03:18:58 +0000 (12:18 +0900)]

X.690 uses "BIT STRING" not "BIT FIELD".

12 years agotorture/basic: in run_derefopen() the file could have been deleted before the last...
Matthias Dieter Wallnöfer [Mon, 10 Aug 2009 08:12:51 +0000 (10:12 +0200)]
torture/basic: in run_derefopen() the file could have been deleted before the last unlink

Through a suggestion pointed out in bug #6622 the test file sometimes doesn't exist on
the last turn anymore. So we haven't to fail here since it could have been deleted by
a concurrent process (e.g. when the same test runs multiple times). Therefore also
NT_STATUS_OBJECT_NAME_NOT_FOUND is an acceptable result.

12 years agos4:AD LDIFs - More refactoring
Matthias Dieter Wallnöfer [Sat, 8 Aug 2009 11:50:10 +0000 (13:50 +0200)]
s4:AD LDIFs - More refactoring

This commit includes:
- Additional static object data in SAMBA 4's AD to start supporting of
  - forest updates, - lost and found, - quotas on DS, - physical locations,
  - licensing of sites, - subnets, - policies for WMI, - DNS entries in AD
- Reordering of provision*.ldif files to be able to find entries and make future
  additions easier
- Add comments in provision*.ldif files to point out where subentries are located
  when they are based in other LDIFs
- Removations of autogenerated "cn" attributes

12 years agopidl: fix parse_idl test after idl.yp changes
Stefan Metzmacher [Wed, 12 Aug 2009 08:24:01 +0000 (10:24 +0200)]
pidl: fix parse_idl test after idl.yp changes

We now generate 'FILE' and 'LINE' elements for each layer.

This change makes the expected PIDL trees human readable:-)


12 years agopidl: fix samba3-cli test after the async function were added
Stefan Metzmacher [Wed, 12 Aug 2009 08:01:12 +0000 (10:01 +0200)]
pidl: fix samba3-cli test after the async function were added


12 years agotry to give some hint as to what is causing NDR string errors
Andrew Tridgell [Wed, 12 Aug 2009 05:34:03 +0000 (15:34 +1000)]
try to give some hint as to what is causing NDR string errors

12 years agowhen we get an NDR error in the logs, it is useful to know where it happened
Andrew Tridgell [Wed, 12 Aug 2009 05:33:37 +0000 (15:33 +1000)]
when we get an NDR error in the logs, it is useful to know where it happened

12 years agono need to shout about getting an oplock
Andrew Tridgell [Wed, 12 Aug 2009 05:20:02 +0000 (15:20 +1000)]
no need to shout about getting an oplock

12 years agoraise the debug level for a common message
Andrew Tridgell [Wed, 12 Aug 2009 05:19:42 +0000 (15:19 +1000)]
raise the debug level for a common message

when a client disconnects we expect this to happen, so don't print an
error each time

12 years agoAllow selection of the samba4 process model for 'make test'
Andrew Tridgell [Wed, 12 Aug 2009 04:41:44 +0000 (14:41 +1000)]
Allow selection of the samba4 process model for 'make test'

To choose the process model, set the environment variable
SAMBA_PROCESS_MODEL to the desired model. This will allow us to enable
the standard process model for some machines in the build farm without
enabling it for all of them. I don't want to just enable it
universally as I am concerned with total memory using during some of
the tests.

12 years agoprint server log on stderr as well as to log file
Andrew Tridgell [Wed, 12 Aug 2009 04:38:55 +0000 (14:38 +1000)]
print server log on stderr as well as to log file

I think we are missing some important messages from the server during
'make test' because we don't show the log file contents during
runs. This patch uses tee to put the log messages to stderr so we can
see any server messages associated with the test that caused them.

12 years agos4:provision Allow provision-backend to not run slapd for 'make test'
Andrew Bartlett [Wed, 12 Aug 2009 01:09:50 +0000 (11:09 +1000)]
s4:provision Allow provision-backend to not run slapd for 'make test'

As the version of OpenLDAP required for Samba4 is fairly new, we don't
want to make it a requirement before this python code is run in 'make

As such, skip over the actual starting of slapd, but check the rest
runs alright (which still validates syntax and other modules).

Andrew Bartlett

12 years agos4:selftest Don't start the slapd for the provision, only for the run
Andrew Bartlett [Mon, 10 Aug 2009 12:01:28 +0000 (22:01 +1000)]
s4:selftest Don't start the slapd for the provision, only for the run

The provision-backend script now starts slapd for us

Andrew Bartlett

12 years agos4:selftest Make OpenLDAP guess it's own modules from now on
Andrew Bartlett [Mon, 10 Aug 2009 11:51:08 +0000 (21:51 +1000)]
s4:selftest Make OpenLDAP guess it's own modules from now on

In the releases of OpenLDAP we require (2.4.17) it can guess this much
itself, so no need for us to do it.

Andrew Bartlett

12 years agos4:provision Make the --ol-slapd paramter take the full path to slapd
Andrew Bartlett [Mon, 10 Aug 2009 11:46:20 +0000 (21:46 +1000)]
s4:provision Make the --ol-slapd paramter take the full path to slapd

12 years agos4:provision Assume the OpenLDAP backend can find it's own modules
Andrew Bartlett [Mon, 10 Aug 2009 11:43:11 +0000 (21:43 +1000)]
s4:provision Assume the OpenLDAP backend can find it's own modules

12 years agos4:provision Rework and further automate setup of OpenLDAP backend
Oliver Liebel [Sun, 9 Aug 2009 23:45:01 +0000 (09:45 +1000)]
s4:provision Rework and further automate setup of OpenLDAP backend

heres the summary of all changes/extensions:

- Andrew Bartlett's patch to generate indext
- Howard Chu's idea to use nosync on the DB included, but made optional

- slaptest-path is not needed any more (slapd -Ttest is used instead)
and is therefore removed. slapd-path is now recommended when
openldap-backend is chosen.
its also used for olc-conversion

- slapd-detection is now always done by ldapsearch (ldb module),
looking anonymous for objectClass: OpenLDAProotDSE via our ldapi_uri.

- if ldapsearch was not successfull, (no slapd listening on our socket)
slapd is
started via special generated slapdcommand_prov  (ldapi_uri only)

- slapd-"provision-process" startup is done via pythons subprocess.

- the slapd-provision-pid is stored under paths.ldapdir/slapd_provision_pid.

- after provision-backend is finished:
--- is compared with our stored slapd_provision_pid.
if the are unique, will be read out, and the
slapd "provison"-process will be shut down.
--- proper slapd-shutdown is verified again with ldb-search -> ldapi_uri
-> rootDSE.
--- if the pids are different or one of the pid-files is missing, slapd
will not be shut down,
instead an error message is displayed to locate slapd manually
--- extended help-messages (relevant to slapd) are always displayed,
e.g. the commandline with which slapd has to be started when everythings
(slapd-commandline is stored under paths.ldapdir/slapd_command_file.txt))

- upgraded the content of the mini-howto (howto-ol-backend-s4.txt)

12 years agos3:smbcacls: forbid change of debug level from config file
Michael Adam [Tue, 11 Aug 2009 21:35:48 +0000 (23:35 +0200)]
s3:smbcacls: forbid change of debug level from config file


12 years agos3: refomat definitions of some generic access rights for better readability
Michael Adam [Tue, 11 Aug 2009 12:28:31 +0000 (14:28 +0200)]
s3: refomat definitions of some generic access rights for better readability


12 years agoselftest: configure "ncalrpc dir" for the client
Stefan Metzmacher [Fri, 1 May 2009 16:18:31 +0000 (18:18 +0200)]
selftest: configure "ncalrpc dir" for the client

Some torture tests will need this soon.


12 years agoselftest: fix typo lock => lockdir
Stefan Metzmacher [Fri, 1 May 2009 16:17:55 +0000 (18:17 +0200)]
selftest: fix typo lock => lockdir


12 years agos3:winbindd: raise the timeout for lsa_Lookup*() calls from 10 to 35 seconds.
Stefan Metzmacher [Tue, 11 Aug 2009 11:50:16 +0000 (13:50 +0200)]
s3:winbindd: raise the timeout for lsa_Lookup*() calls from 10 to 35 seconds.


12 years agos4:torture Add test for the NTP signd server
Andrew Bartlett [Tue, 11 Aug 2009 08:20:39 +0000 (18:20 +1000)]
s4:torture Add test for the NTP signd server

This is used by at patch to the NTP project to supply authenticated
time as required by MS-SNTP. (ie, to keep windows clients in time sync
in the domain)

Andrew Bartlett

12 years agoi18n: Update s3 net German translation.
Kai Blin [Tue, 11 Aug 2009 07:33:51 +0000 (09:33 +0200)]
i18n: Update s3 net German translation.

12 years agos3 net: i18n support for net utility functions
Kai Blin [Tue, 11 Aug 2009 07:01:10 +0000 (09:01 +0200)]
s3 net: i18n support for net utility functions

12 years agos3 net: i18n support for net usershare
Kai Blin [Tue, 11 Aug 2009 06:57:01 +0000 (08:57 +0200)]
s3 net: i18n support for net usershare

12 years agos3 net: i18n support for net user
Kai Blin [Tue, 11 Aug 2009 06:40:48 +0000 (08:40 +0200)]
s3 net: i18n support for net user

12 years agos3 net: i18n support for net time
Kai Blin [Tue, 11 Aug 2009 06:38:42 +0000 (08:38 +0200)]
s3 net: i18n support for net time

12 years agos3 net: i18n support for net status
Kai Blin [Tue, 11 Aug 2009 06:35:46 +0000 (08:35 +0200)]
s3 net: i18n support for net status

12 years agos3 net: i18n support for net share
Kai Blin [Tue, 11 Aug 2009 06:32:10 +0000 (08:32 +0200)]
s3 net: i18n support for net share

12 years agos3 net: i18n support for net sam
Kai Blin [Tue, 11 Aug 2009 06:31:04 +0000 (08:31 +0200)]
s3 net: i18n support for net sam

12 years agos3 net: i18n support for net rpc shell
Kai Blin [Mon, 10 Aug 2009 18:38:33 +0000 (20:38 +0200)]
s3 net: i18n support for net rpc shell

12 years agos3 net: i18n support for net rpc service
Kai Blin [Mon, 10 Aug 2009 18:23:41 +0000 (20:23 +0200)]
s3 net: i18n support for net rpc service

12 years agos3 net: i18n support for net rpc samsync
Kai Blin [Mon, 10 Aug 2009 17:54:27 +0000 (19:54 +0200)]
s3 net: i18n support for net rpc samsync

12 years agoEnsure the fsp->fsp_name stat_ex struct is always
Jeremy Allison [Mon, 10 Aug 2009 23:39:13 +0000 (16:39 -0700)]
Ensure the fsp->fsp_name stat_ex struct is always
up to date after the open.

12 years agoRefactor the use of create_time and change time to go
Jeremy Allison [Mon, 10 Aug 2009 22:07:53 +0000 (15:07 -0700)]
Refactor the use of create_time and change time to go
through functions. Will aid in making us pass RAW-SETFILEINFO.

12 years agos3 net: i18n support for net rpc rights
Kai Blin [Mon, 10 Aug 2009 16:52:24 +0000 (18:52 +0200)]
s3 net: i18n support for net rpc rights

12 years agos3 net: i18n support net rpc registry
Kai Blin [Mon, 10 Aug 2009 16:29:50 +0000 (18:29 +0200)]
s3 net: i18n support net rpc registry

12 years agos3 net: i18n support for net rpc printer
Kai Blin [Mon, 10 Aug 2009 16:24:57 +0000 (18:24 +0200)]
s3 net: i18n support for net rpc printer

12 years agos3 net: i18n support for net rpc join
Kai Blin [Sun, 9 Aug 2009 14:57:51 +0000 (16:57 +0200)]
s3 net: i18n support for net rpc join

12 years agos3-spoolss: remove device mode defines from nt_printing.h.
Günther Deschner [Mon, 10 Aug 2009 09:25:52 +0000 (11:25 +0200)]
s3-spoolss: remove device mode defines from nt_printing.h.


12 years agos3: re-run make samba3-idl.
Günther Deschner [Mon, 10 Aug 2009 09:25:09 +0000 (11:25 +0200)]
s3: re-run make samba3-idl.


12 years agospoolss: fill in more details of the spoolss_DeviceMode.
Günther Deschner [Mon, 10 Aug 2009 09:24:40 +0000 (11:24 +0200)]
spoolss: fill in more details of the spoolss_DeviceMode.


12 years agoImplement db_rbt_traverse
Volker Lendecke [Sat, 8 Aug 2009 11:43:41 +0000 (13:43 +0200)]
Implement db_rbt_traverse

12 years agoMake wbinfo also print the members when querying a group
Volker Lendecke [Sat, 8 Aug 2009 11:42:55 +0000 (13:42 +0200)]
Make wbinfo also print the members when querying a group

12 years agolibreplace: set SONAMEFLAG to "-Wl,-soname=" on QNX
Matt Kraai [Fri, 7 Aug 2009 20:58:52 +0000 (13:58 -0700)]
libreplace: set SONAMEFLAG to "-Wl,-soname=" on QNX

12 years agos3:smbd: convert '\\' into '/' in SMB2 Create
Stefan Metzmacher [Sat, 8 Aug 2009 08:47:11 +0000 (10:47 +0200)]
s3:smbd: convert '\\' into '/' in SMB2 Create


12 years agos3:smbd: rename conn => sconn for smbd_server_connection structs
Stefan Metzmacher [Fri, 7 Aug 2009 13:21:07 +0000 (15:21 +0200)]
s3:smbd: rename conn => sconn for smbd_server_connection structs

This should avoid confusion between smbd_server_connection
and connection_struct variables.


12 years agoMerge branch 'master' of ssh://
Alexander Bokovoy [Sat, 8 Aug 2009 05:07:24 +0000 (08:07 +0300)]
Merge branch 'master' of ssh://

12 years agoPass absolute file paths to Inkscape when transforming .svg files
Alexander Bokovoy [Sat, 8 Aug 2009 05:02:52 +0000 (08:02 +0300)]
Pass absolute file paths to Inkscape when transforming .svg files

Some recent versions of Inkscape (0.47 or around) have bug when export file name
is treated as relative against the directory of original .svg if it wasn't specified
as an absolute path. Fix it by always using absolute paths during conversion.

12 years agoMove the checks for null timestamps down below the VFS_NTIMES
Jeremy Allison [Fri, 7 Aug 2009 19:38:31 +0000 (12:38 -0700)]
Move the checks for null timestamps down below the VFS_NTIMES
layer (as it's done in onefs). This simplifies greatly the
code in smb_set_file_time() w.r.t. changenotify messages.

12 years agoCorrectly send out notify messages for timestamp changes.
Jeremy Allison [Fri, 7 Aug 2009 17:19:20 +0000 (10:19 -0700)]
Correctly send out notify messages for timestamp changes.
"change time" has no notify message, so don't send anything
out when we change it. Use FILE_NOTIFY_CHANGE_CREATION correctly
when changing the create time.

12 years agos4: Add a new script for setting password properties for a domain in a easier way
Matthias Dieter Wallnöfer [Wed, 5 Aug 2009 21:18:47 +0000 (23:18 +0200)]
s4: Add a new script for setting password properties for a domain in a easier way

12 years agos4:libcli/smb2: move SMB2_GETINFO_* flags into smb2_constants.h
Stefan Metzmacher [Fri, 7 Aug 2009 12:12:19 +0000 (14:12 +0200)]
s4:libcli/smb2: move SMB2_GETINFO_* flags into smb2_constants.h


12 years agos4:libcli/smb2: remove unused and redundant SMB2 security flags
Stefan Metzmacher [Fri, 7 Aug 2009 12:11:11 +0000 (14:11 +0200)]
s4:libcli/smb2: remove unused and redundant SMB2 security flags


12 years agos4:libcli: move SMB2 Find constants to smb2_constants.h
Stefan Metzmacher [Fri, 7 Aug 2009 12:09:42 +0000 (14:09 +0200)]
s4:libcli: move SMB2 Find constants to smb2_constants.h


12 years agos4:libcli/raw: we don't need to include "smb.h" explicit
Stefan Metzmacher [Fri, 7 Aug 2009 12:55:37 +0000 (14:55 +0200)]
s4:libcli/raw: we don't need to include "smb.h" explicit


12 years agos4:libcli/raw: also include smb2_constants.h into interfaces.h
Stefan Metzmacher [Fri, 7 Aug 2009 12:55:01 +0000 (14:55 +0200)]
s4:libcli/raw: also include smb2_constants.h into interfaces.h


12 years agos3:smbd: split the dir entry marshalling into smbd_marshall_dir_entry()
Stefan Metzmacher [Thu, 6 Aug 2009 10:15:51 +0000 (12:15 +0200)]
s3:smbd: split the dir entry marshalling into smbd_marshall_dir_entry()

So that we can reuse it for SMB2 Find.


12 years agos3:smbd: move dptr globals into struct smbd_server_connection
Stefan Metzmacher [Thu, 6 Aug 2009 10:15:51 +0000 (12:15 +0200)]
s3:smbd: move dptr globals into struct smbd_server_connection


12 years agos3:smbd: remove dirptr and dirpath from connection_struct
Stefan Metzmacher [Fri, 7 Aug 2009 07:31:45 +0000 (09:31 +0200)]
s3:smbd: remove dirptr and dirpath from connection_struct

They're both only used in the context of a function,
so we can make them stack variables.


12 years agos3:smbd: implement get_lanman2_dir_entry() on top of smbd_dirptr_get_entry()
Stefan Metzmacher [Fri, 7 Aug 2009 06:55:35 +0000 (08:55 +0200)]
s3:smbd: implement get_lanman2_dir_entry() on top of smbd_dirptr_get_entry()


12 years agos3:smbd: let get_dir_entry() use smbd_dirptr_get_entry()
Stefan Metzmacher [Fri, 7 Aug 2009 06:54:06 +0000 (08:54 +0200)]
s3:smbd: let get_dir_entry() use smbd_dirptr_get_entry()


12 years agos3:smbd: add generic smbd_dirptr_get_entry()
Stefan Metzmacher [Thu, 6 Aug 2009 18:53:13 +0000 (20:53 +0200)]
s3:smbd: add generic smbd_dirptr_get_entry()


12 years agos3:smbd: conn_free_internal() can be static now
Stefan Metzmacher [Fri, 7 Aug 2009 09:48:03 +0000 (11:48 +0200)]
s3:smbd: conn_free_internal() can be static now


12 years agos3:smbd: add a smbd_server_connection pointer to connection_struct
Stefan Metzmacher [Thu, 6 Aug 2009 11:22:33 +0000 (13:22 +0200)]
s3:smbd: add a smbd_server_connection pointer to connection_struct

This can be NULL for faked connection structs used in the rpc server
or printing code.


12 years agoFix some nonempty blank lines
Volker Lendecke [Fri, 7 Aug 2009 10:09:21 +0000 (12:09 +0200)]
Fix some nonempty blank lines

12 years agopidl: parse idl 'pipe' typedefs, but print out a not supported message for now
Stefan Metzmacher [Wed, 5 Aug 2009 11:40:38 +0000 (13:40 +0200)]
pidl: parse idl 'pipe' typedefs, but print out a not supported message for now


12 years agopidl: fix formatting in expr.yp
Stefan Metzmacher [Tue, 4 Aug 2009 13:00:13 +0000 (15:00 +0200)]
pidl: fix formatting in expr.yp


12 years agopidl: fix formatting in idl.yp
Stefan Metzmacher [Tue, 4 Aug 2009 12:59:33 +0000 (14:59 +0200)]
pidl: fix formatting in idl.yp


12 years agolibrpc: fix callers after epmapper.idl changes
Stefan Metzmacher [Wed, 5 Aug 2009 11:03:37 +0000 (13:03 +0200)]
librpc: fix callers after epmapper.idl changes


12 years agolibrpc: rerun 'make idl' after epmapper.idl changes
Stefan Metzmacher [Wed, 5 Aug 2009 11:05:20 +0000 (13:05 +0200)]
librpc: rerun 'make idl' after epmapper.idl changes


12 years agoepmapper.idl: avoid usage of 'pipe', this is a reserved word in IDL
Stefan Metzmacher [Wed, 5 Aug 2009 11:04:59 +0000 (13:04 +0200)]
epmapper.idl: avoid usage of 'pipe', this is a reserved word in IDL

We now use 'named_pipe' instead.


12 years agos3:smbd: don't ignore check_descend in get_dir_entry()
Stefan Metzmacher [Thu, 6 Aug 2009 19:16:38 +0000 (21:16 +0200)]
s3:smbd: don't ignore check_descend in get_dir_entry()


12 years agos3-wkssvcs: fix check for BUILTIN\Administrators membership in wkssvc_Netr{Join,Unjoi...
Günther Deschner [Thu, 6 Aug 2009 18:59:32 +0000 (20:59 +0200)]
s3-wkssvcs: fix check for BUILTIN\Administrators membership in wkssvc_Netr{Join,Unjoin}Domain2.


12 years agos3 net: i18n support for net rpc audit
Kai Blin [Fri, 7 Aug 2009 07:34:06 +0000 (09:34 +0200)]
s3 net: i18n support for net rpc audit

12 years agos3 net: net i18n in it's own module
Kai Blin [Thu, 6 Aug 2009 18:52:58 +0000 (20:52 +0200)]
s3 net: net i18n in it's own module

12 years agoRevert "s3: Update i18n files"
Kai Blin [Thu, 6 Aug 2009 18:38:39 +0000 (20:38 +0200)]
Revert "s3: Update i18n files"

This reverts commit 57b5269038624b70e117025af9b977e0addcdc8e,
6e044a22cf232798efbd40a2b55fd9556483d403 and

The correct location for net .po files is in locale/net/

12 years agos3 net: i18n support for net rpc
Kai Blin [Tue, 4 Aug 2009 06:28:50 +0000 (08:28 +0200)]
s3 net: i18n support for net rpc

12 years agos3 net: i18n support for net registry
Kai Blin [Sun, 2 Aug 2009 08:26:48 +0000 (10:26 +0200)]
s3 net: i18n support for net registry

12 years agos3 net: i18n support for net rap
Kai Blin [Sun, 2 Aug 2009 08:25:26 +0000 (10:25 +0200)]
s3 net: i18n support for net rap

12 years agofixed another ambiguous talloc call
Andrew Tridgell [Fri, 7 Aug 2009 07:23:52 +0000 (17:23 +1000)]
fixed another ambiguous talloc call

During the creation of the 3 RPC pipes in winbind we try to steal the
RPC binding structure to be a child of the pipe once the pipe is
established. This fails with a talloc warning as the rpc connection
code already holds a reference to the binding.

The fix is to use talloc_reparent() instead.

12 years agoensure that child tasks die when the parent dies
Andrew Tridgell [Fri, 7 Aug 2009 07:21:54 +0000 (17:21 +1000)]
ensure that child tasks die when the parent dies

Previously we relied on process groups and SIGTERM to ensure that
child tasks died in the standard process model when the parent task
died. This doesn't work when the server is run in interactive mode, as
in that case we don't call become_daemon() and don't get a separate
process group.

The fix is to have a pipe held open by the parent server process, and
inherited by child tasks. If the parent exits then the write side of
the pipe is implicitly closed, which causes an event in the child
tasks that causes them to exit

12 years agoprime the sam ldb schema in the parent samba process
Andrew Tridgell [Fri, 7 Aug 2009 07:19:39 +0000 (17:19 +1000)]
prime the sam ldb schema in the parent samba process

While testing the use of the standard process model with 'make test' I
found that testing was much slower (by several times) with the
standard model than with the single model. The primary problem was
that each SMB connection would open a new sam ldb context, and all of
those would reload the full AD schema.

The fix is to pre-open the SAM during server startup, before any child
processes are forked. This sets up the global schema context which is
inherited by all connections.

The standard model is still slower at make test than the single model,
but not by nearly as much. I am working on further reducing the gap.

12 years agouse talloc with the global schema consistently
Andrew Tridgell [Fri, 7 Aug 2009 07:16:26 +0000 (17:16 +1000)]
use talloc with the global schema consistently

Before this change, the first opener of the sam ldb context would
become the owner of the global schema, then the autofree context got a
reference to the schema. Any subsequent opens of the sam ldb also got
a reference. This meant that the talloc hierarchy was inconsistent
between the first sam ldb open and subsequent opens. With this change
the autofree context becomes the owner of the global schema, and all
ldb contexts get a reference.

12 years agofixed several places that unnecessarily take a reference to the event context
Andrew Tridgell [Fri, 7 Aug 2009 07:14:13 +0000 (17:14 +1000)]
fixed several places that unnecessarily take a reference to the event context

These references were triggering the ambiguous talloc_free errors from
the recent talloc changes when the server is run using the 'standard'
process model instead of the 'single' process model. I am aiming to
move the build farm to use the 'standard' process model soon, as part
of an effort to make our test environment better match the real
deployment of Samba4.

The references are not needed as the way that the event context is
used is as the 'top parent', so when the event context is freed then
all of the structures that were taking a reference to the event
context were actually freed as well, thus making the references

12 years agos3: add a test to test libsmbclient
Bo Yang [Fri, 7 Aug 2009 22:14:33 +0000 (06:14 +0800)]
s3: add a test to test libsmbclient

Signed-off-by: Bo Yang <>
12 years agos3: Fix nss info substitution
Bo Yang [Fri, 7 Aug 2009 22:10:37 +0000 (06:10 +0800)]
s3: Fix nss info substitution

Signed-off-by: Bo Yang <>
12 years agomake sure we never look past the end of either string in ldb_comparison_fold()
Andrew Tridgell [Fri, 7 Aug 2009 04:34:01 +0000 (14:34 +1000)]
make sure we never look past the end of either string in ldb_comparison_fold()

This fixes a bug in the samba3sam test with the python libraries as
noticed by abartlet

12 years agos4:ldb Make error message in rnd_name more useful
Andrew Bartlett [Fri, 7 Aug 2009 04:22:54 +0000 (14:22 +1000)]
s4:ldb Make error message in rnd_name more useful

12 years agoMerge branch 'master' of ssh:// into master-devel
Andrew Bartlett [Fri, 7 Aug 2009 03:08:35 +0000 (13:08 +1000)]
Merge branch 'master' of ssh:// into master-devel

12 years agos4:setup Remove extra newlines that break OpenLDAP backend
Oliver Liebel [Fri, 7 Aug 2009 02:38:51 +0000 (12:38 +1000)]
s4:setup Remove extra newlines that break OpenLDAP backend

12 years agos3: Fix a bug in renames of directories
Tim Prouty [Thu, 6 Aug 2009 22:53:33 +0000 (15:53 -0700)]
s3: Fix a bug in renames of directories

Recently code was added to match windows semantics of denying the
rename of a directory if there are open files underneath it.  This
does partly match windows semantics, but it turns out the rename
should be allowed if the open file handle is for the directory being
renamed, or for a stream on the directory being renamed.  This patch
refines the check to better follow these rename semantics.

12 years agos4 torture: Extend the RAW-RENAME test to more fully test directory renames.
Tim Prouty [Thu, 6 Aug 2009 18:23:23 +0000 (11:23 -0700)]
s4 torture: Extend the RAW-RENAME test to more fully test directory renames.

The existing test was only covering files opened underneath the
directory that was being renamed.  It is not uncommon for windows
clients to actually hold a read-only handle to a directory open across
the rename, which it turns out doesn't return NT_STATUS_ACCESS_DENIED.
Additionally, holding a handle open to a stream on the directory is
also allowed.