ira/wip.git
16 years agor24339: Move output functions to separate files.
Jelmer Vernooij [Sun, 12 Aug 2007 00:50:25 +0000 (00:50 +0000)]
r24339: Move output functions to separate files.

16 years agor24338: Add callbacks for starting and finishing tests.
Jelmer Vernooij [Sat, 11 Aug 2007 22:14:07 +0000 (22:14 +0000)]
r24338: Add callbacks for starting and finishing tests.

16 years agor24337: Make libreplace provide offsetof.
Jelmer Vernooij [Sat, 11 Aug 2007 21:30:19 +0000 (21:30 +0000)]
r24337: Make libreplace provide offsetof.

16 years agor24336: Use standard data type uint32_t rather than tdb-specific u32.
Jelmer Vernooij [Sat, 11 Aug 2007 21:19:24 +0000 (21:19 +0000)]
r24336: Use standard data type uint32_t rather than tdb-specific u32.

16 years agor24335: Fix -N support.
Jelmer Vernooij [Sat, 11 Aug 2007 18:31:27 +0000 (18:31 +0000)]
r24335: Fix -N support.

16 years agor24333: Add convenience function for checking the contents of a file in tests.
Jelmer Vernooij [Sat, 11 Aug 2007 17:08:22 +0000 (17:08 +0000)]
r24333: Add convenience function for checking the contents of a file in tests.

16 years agor24313: Merge r23377 from 3_2
Volker Lendecke [Fri, 10 Aug 2007 11:53:52 +0000 (11:53 +0000)]
r24313: Merge r23377 from 3_2

Patch from Bjoern Jacke <bjoern@j3e.de> to deal
with non-GNU implementations of tr in autogen.sh

16 years agor24301: The less escape characters the better. This changes the winbind
Andrew Bartlett [Fri, 10 Aug 2007 08:46:29 +0000 (08:46 +0000)]
r24301: The less escape characters the better.  This changes the winbind
seperator to /, which is not an escape character, and uses the fact
that we always support / as the DOMAIN/username seperator.

Andrew Bartlett

16 years agor24300: Try to get the LOCAL-EVENT test to pass on hosts with epoll() in
Andrew Bartlett [Fri, 10 Aug 2007 07:40:50 +0000 (07:40 +0000)]
r24300: Try to get the LOCAL-EVENT test to pass on hosts with epoll() in
glibc, but not in the host kernel.

Andrew Bartlett

16 years agor24286: (missed from previous commit)
Andrew Bartlett [Thu, 9 Aug 2007 08:43:15 +0000 (08:43 +0000)]
r24286: (missed from previous commit)

A number of machines on the build farm fail because while they use
SOCKET_WRAPPER, they don't compile with --enable-developer.  As such
the winbind tests fail.

We need to ensure that on production systems we never have
SOCKET_WRAPPER set, or this could be a security issue.

Andrew Bartlett

16 years agor24285: A number of machines on the build farm fail because while they use
Andrew Bartlett [Thu, 9 Aug 2007 08:08:09 +0000 (08:08 +0000)]
r24285: A number of machines on the build farm fail because while they use
SOCKET_WRAPPER, they don't compile with --enable-developer.  As such
the winbind tests fail.

We need to ensure that on production systems we never have
SOCKET_WRAPPER set, or this could be a security issue.

Andrew Bartlett

16 years agor24284: change brlock_tdb.c to use the dbwrap API. This actually makes the
Andrew Tridgell [Thu, 9 Aug 2007 06:36:16 +0000 (06:36 +0000)]
r24284: change brlock_tdb.c to use the dbwrap API. This actually makes the
backend abstraction for brlock pointless, but I have left it in place
for now. It would be useful for other clustering systems that can't
map to dbwrap, and would also be useful if we wanted to keep the
remote function call capabilities in ctdb instead of the less
efficient fetch_locked() call in dbwrap

16 years agor24283: added a dbwrap API for temporary databases, based on the one from the
Andrew Tridgell [Thu, 9 Aug 2007 06:34:20 +0000 (06:34 +0000)]
r24283: added a dbwrap API for temporary databases, based on the one from the
s3 clustering code.

This commit just establishes the API. The ctdb backend for this will
come soon.

16 years agor24282: Try to fix the occasional Samba4 crash in BASE-BENCH-READWRITE, as
Andrew Bartlett [Thu, 9 Aug 2007 06:26:19 +0000 (06:26 +0000)]
r24282: Try to fix the occasional Samba4 crash in BASE-BENCH-READWRITE, as
seen in particular on opi.

This looked like a Heimdal problem, but I think it was simply that we
didn't do a talloc_reference() to keep tabs on the memory we were
using, and in between obtaining the pointer and using it, it was
assigned to unrelated memory.

Andrew Bartlett

16 years agor24277: Tidyup as requested by metze.
Andrew Bartlett [Wed, 8 Aug 2007 06:37:37 +0000 (06:37 +0000)]
r24277: Tidyup as requested by metze.

Andrew Bartlett

16 years agor24276: don't mark a stream readable until after the messaging setup is
Andrew Tridgell [Wed, 8 Aug 2007 03:20:37 +0000 (03:20 +0000)]
r24276: don't mark a stream readable until after the messaging setup is
complete. This is needed because messaging setup with ctdb involve
events, and we don't want a SMB packet to be processed on this stream
until after the stream structure is fully setup

16 years agor24275: - setup the connection to the ctdb daemon
Andrew Tridgell [Wed, 8 Aug 2007 03:19:38 +0000 (03:19 +0000)]
r24275: - setup the connection to the ctdb daemon

- disable the brlock ctdb backend for now

16 years agor24274: - merge from ctdb bzr tree
Andrew Tridgell [Wed, 8 Aug 2007 03:18:51 +0000 (03:18 +0000)]
r24274: - merge from ctdb bzr tree

- use ctdb_attach() instead of ctdb_db_handle()

- make ctdb_attach() return an existing db handle if it exists

16 years agor24273: Fix bug #4817 by <mwallnoefer@yahoo.de>. (Unable to add a computer
Andrew Bartlett [Wed, 8 Aug 2007 02:41:12 +0000 (02:41 +0000)]
r24273: Fix bug #4817 by <mwallnoefer@yahoo.de>.  (Unable to add a computer
from MMC Active Directory Users and Computers).

Windows sets a 14 UCS2 char buffer as the password in this case.

We need to allow random buffers to be accepted as complex passwords,
even if they don't have ASCII upper or lower case characters.  (If
half the bytes are > 127, then it's likely a random buffer).

Also make the test match the documented windows behaviour of '3 of the
4 classes: upper, lower, digit, special'.

Andrew Bartlett

16 years agor24263: Fix bug 4846 (unable to copy users in MMC Active Directory Users and
Andrew Bartlett [Tue, 7 Aug 2007 09:01:08 +0000 (09:01 +0000)]
r24263: Fix bug 4846 (unable to copy users in MMC Active Directory Users and
Computers).

We now generate a security descriptor for each object, when it is
created.  This seems to keep MMC happy.  The next step is to honour
it.

Andrew Bartlett

16 years agor24262: Set the objectCategory by default in the objectclass module, rather than...
Andrew Bartlett [Tue, 7 Aug 2007 05:58:47 +0000 (05:58 +0000)]
r24262: Set the objectCategory by default in the objectclass module, rather than using templates.

Modify the samba3sam test to be less fussy, and not use the
objectclass module (which requires proper schema stuff now).

Andrew Bartlett

16 years agor24261: Fix the standalone ldb build after I moved the objectclass module out.
Andrew Bartlett [Tue, 7 Aug 2007 04:29:42 +0000 (04:29 +0000)]
r24261: Fix the standalone ldb build after I moved the objectclass module out.

Andrew Bartlett

16 years agor24260: Ensure we always override any existing values for these generated
Andrew Bartlett [Tue, 7 Aug 2007 03:35:45 +0000 (03:35 +0000)]
r24260: Ensure we always override any existing values for these generated
attributes.  Anything else leads to madness.

Andrew Bartlett

16 years agor24259: Rework the objectclass module to use the new schema, rather than the
Andrew Bartlett [Tue, 7 Aug 2007 03:33:26 +0000 (03:33 +0000)]
r24259: Rework the objectclass module to use the new schema, rather than the
ldb_subclass list.

Next step will be to have this module also set the objectCategory and
default ntSecurityDescriptor

Andrew Bartlett

16 years agor24249: Thse generated attributes should not be pushed this far down the stack
Andrew Bartlett [Mon, 6 Aug 2007 05:43:54 +0000 (05:43 +0000)]
r24249: Thse generated attributes should not be pushed this far down the stack
in any cse.

Andrew Bartlett

16 years agor24248: Attempt to fix bug #4830 by <mwallnoefer@yahoo.de>. If there is no
Andrew Bartlett [Mon, 6 Aug 2007 04:07:05 +0000 (04:07 +0000)]
r24248: Attempt to fix bug #4830 by <mwallnoefer@yahoo.de>.  If there is no
payload to the control, we still need to inialise *value, as otherwise
we read uninitialised data later.

Andrew Bartlett

16 years agor24247: Remove extra newlines from ldb_debug() calls - it already adds one.
Andrew Bartlett [Mon, 6 Aug 2007 03:48:56 +0000 (03:48 +0000)]
r24247: Remove extra newlines from ldb_debug() calls - it already adds one.

Andrew Bartlett

16 years agor24246: Avoid the annoying 'probable memory leak in ldb' messages, by fixing
Andrew Bartlett [Mon, 6 Aug 2007 03:01:59 +0000 (03:01 +0000)]
r24246: Avoid the annoying 'probable memory leak in ldb' messages, by fixing
some issues in the NBT server (this was a false positive, but easily
worked around) and DRSUAPI server.

We should take care not to use the ldb_context as a talloc pool, and
to always ensure that any results from ldb_search() are moved off that
pool with talloc_steal or talloc_free().

To work around the issue in provision, for which I can find no fault
(other than a lot of work being done in provision), I've moved the
detector trigger to 400 additional blocks.

This fixes Bug #4810 by <mwallnoefer@yahoo.de>

Andrew Bartlett

16 years agor24245: Fix bug #4828 - we need to set the samba LDB debug handler early, so
Andrew Bartlett [Mon, 6 Aug 2007 01:36:49 +0000 (01:36 +0000)]
r24245: Fix bug #4828 - we need to set the samba LDB debug handler early, so
we catch messages pushed out by modules during startup.

This also sets an explict mapping between ldb and Samba debug levels.

Andrew Bartlett

16 years agor24157: Merge from kai/samba4-gsoc.git;h=728deba680f8cf85cab168a6278a2cf657f65fdb
Kai Blin [Fri, 3 Aug 2007 13:51:12 +0000 (13:51 +0000)]
r24157: Merge from kai/samba4-gsoc.git;h=728deba680f8cf85cab168a6278a2cf657f65fdb

Make WBSRV_SAMBA3_SET_STRING use safe_strcpy instead of strncpy.

16 years agor24146: It is not an error for a Win2k3-only server not to support the NT4
Andrew Bartlett [Fri, 3 Aug 2007 08:25:15 +0000 (08:25 +0000)]
r24146: It is not an error for a Win2k3-only server not to support the NT4
replication call.

Andrew Bartlett

16 years agor24127: Set the Domain SID into the libnet context, and have libnet_UserInfo
Andrew Bartlett [Thu, 2 Aug 2007 13:08:39 +0000 (13:08 +0000)]
r24127: Set the Domain SID into the libnet context, and have libnet_UserInfo
return full SIDs for the user SID and primary group sid.

This should help kai with his getpwnam work in winbind.

Andrew Bartlett

16 years agor24118: Start fixing #4842 (usrmgr polcies menu not working) by removing range
Andrew Bartlett [Thu, 2 Aug 2007 01:30:03 +0000 (01:30 +0000)]
r24118: Start fixing #4842 (usrmgr polcies menu not working) by removing range
restriction on the maximum returned size.  There isn't a good reason
to have a limit on this one.

Andrew Bartlett

16 years agor24112: Complete initialistion of the libnet_ctx when setting up the domain.
Andrew Bartlett [Wed, 1 Aug 2007 04:05:06 +0000 (04:05 +0000)]
r24112: Complete initialistion of the libnet_ctx when setting up the domain.
We need to set the access_mask and the domain name, or else libnet
will try to do this itself.

This seems to fix the issues Kai was having.

Andrew Bartlett

16 years agor24111: Untested code is broken code, untested code is broken code...
Andrew Bartlett [Wed, 1 Aug 2007 01:22:53 +0000 (01:22 +0000)]
r24111: Untested code is broken code, untested code is broken code...

Apologies for my previous commit, which should never have been
commited untested.

Andrew Bartlett

16 years agor24110: I hate seeing callers manually filling in the composite context. Use
Andrew Bartlett [Wed, 1 Aug 2007 00:38:53 +0000 (00:38 +0000)]
r24110: I hate seeing callers manually filling in the composite context.  Use
the helper functions instead (and in kai's new code, which just copied
the previous bad practice).

Andrew Bartlett

16 years agor24109: Add a wb_name2domain call
Kai Blin [Tue, 31 Jul 2007 23:49:04 +0000 (23:49 +0000)]
r24109: Add a wb_name2domain call

16 years agor24108: Split out samba3_parse_domuser to a seperate file and rename, so it can be
Kai Blin [Tue, 31 Jul 2007 23:43:59 +0000 (23:43 +0000)]
r24108: Split out samba3_parse_domuser to a seperate file and rename, so it can be
used for a name2domain call.

16 years agor24083: Don't fail the test (looking for the user in the enum) if we didn't
Andrew Bartlett [Mon, 30 Jul 2007 10:43:50 +0000 (10:43 +0000)]
r24083: Don't fail the test (looking for the user in the enum) if we didn't
create the user in the first place.

Andrew Bartlett

16 years agor24082: Following the removal of a fanstsy condition from the SAMR testsuite,
Andrew Bartlett [Mon, 30 Jul 2007 10:37:33 +0000 (10:37 +0000)]
r24082: Following the removal of a fanstsy condition from the SAMR testsuite,
allow the server side to enumerate all domain controllers and domain
members...

Andrew Bartlett

16 years agor24081: Domain Controllers are also shown in this enumeration.
Andrew Bartlett [Mon, 30 Jul 2007 10:34:23 +0000 (10:34 +0000)]
r24081: Domain Controllers are also shown in this enumeration.

Andrew Bartlett

16 years agor24080: Set the primary group (matching windows) when creating new users in
Andrew Bartlett [Mon, 30 Jul 2007 10:30:34 +0000 (10:30 +0000)]
r24080: Set the primary group (matching windows) when creating new users in
SAMR.  This can't be done in the ldb templates code, as it doesn't
happen over direct LDAP.

As noted in bug #4829.

Andrew Bartlett

16 years agor24076: Make ldap.js pass against Win2k3 again (looks like we don't match AD
Andrew Bartlett [Mon, 30 Jul 2007 10:05:47 +0000 (10:05 +0000)]
r24076: Make ldap.js pass against Win2k3 again (looks like we don't match AD
on this error code, but allow both for now).

Also prove that bug #4829 needs a different solution: we can't fix
this by changing the template.  I think this fix needs to be in the
SAMR server.

Andrew Bartlett

16 years agor24075: As suggested by metze, match the behaviour of ntvfs_posix, and remove
Andrew Bartlett [Mon, 30 Jul 2007 09:04:28 +0000 (09:04 +0000)]
r24075: As suggested by metze, match the behaviour of ntvfs_posix, and remove
the backend data (effectivly closing the handle) when we close an IPC
FD.

This should fix #4821.

Andrew Bartlett

16 years agor24074: Test both permitted logon hours and permitted workstations in the
Andrew Bartlett [Mon, 30 Jul 2007 08:58:39 +0000 (08:58 +0000)]
r24074: Test both permitted logon hours and permitted workstations in the
RPC-SAMLOGON test.

This showed that, as noted by bug #4823, we didn't test for invalid
workstations.  In fact, the code had been ported across, but because
untested code is broken code, it never worked...

Andrew Bartlett

16 years agor24061: Anther part of bug #4823, which is that until now Samba4 didn't parse
Andrew Bartlett [Fri, 27 Jul 2007 06:31:12 +0000 (06:31 +0000)]
r24061: Anther part of bug #4823, which is that until now Samba4 didn't parse
the logon hours, even if set.

This code happily stolen from the great work in Samba3 :-)

Andrew Bartlett

16 years agor24060: Fix bug #4806 by Matthias Wallnöfer <mwallnoefer@yahoo.de>: We need to
Andrew Bartlett [Fri, 27 Jul 2007 03:08:15 +0000 (03:08 +0000)]
r24060: Fix bug #4806 by Matthias Wallnöfer <mwallnoefer@yahoo.de>: We need to
include the attribute allowedChildClassesEffective for MMC to allow
the creation of containers.

This may need further refinement, but it seems to work for now.

Andrew Bartlett

16 years agor24059: Fix bug 4822 reported by Matthias Wallnöfer <mwallnoefer@yahoo.de>.
Andrew Bartlett [Fri, 27 Jul 2007 02:07:17 +0000 (02:07 +0000)]
r24059: Fix bug 4822 reported by Matthias Wallnöfer <mwallnoefer@yahoo.de>.

Any SAMR client (usrmgr.exe in this case) that attempted to set a
property to a zero length string found instead the the old value was
kept.

In fixing this, rework the macros to be cleaner (add the
always-present .string) to every macro, and remove the use of the
samdb_modify() and samdb_replace() wrappers where possible.

Andrew Bartlett

16 years agor24054: Fix some warnings
Volker Lendecke [Thu, 26 Jul 2007 07:48:14 +0000 (07:48 +0000)]
r24054: Fix some warnings

16 years agor24053: Ensure we filter EnumDomainUsers with the supplied mask.
Andrew Bartlett [Thu, 26 Jul 2007 07:27:46 +0000 (07:27 +0000)]
r24053: Ensure we filter EnumDomainUsers with the supplied mask.

Should fix another part (list of domains in usrmgr incorrectly
including accounts) of bug #4815 by mwallnoefer@yahoo.de.

Andrew Bartlett

16 years agor24052: Fix some of the NT4 usrmgr.exe portions of bug 4815.
Andrew Bartlett [Thu, 26 Jul 2007 03:50:24 +0000 (03:50 +0000)]
r24052: Fix some of the NT4 usrmgr.exe portions of bug 4815.

 - The icons in usermgr were incorrect, because the acct_flags were
   not filled in (due to missing attribute in ldb query)

 - The Full name was missing, and the description used as the full
   name (due to missing attributes in ldb query and incorrect IDL)

To prove the correctness of these fixes, I added a substantial new
test to RPC-SAMR-USERS, to ensure cross-consistancy between
QueryDisplayInfo and QueryUserInfo on each user.

This showed that for some reason, we must add ACB_NORMAL to the
acct_flags on level 2 queries (for machine trust accounts)...

Getting this right is important, because Samba3's RPC winbind methods
uses these queries.

Andrew Bartlett

16 years agor24051: more monitor function calls and monitor msg names
Rafal Szczesniak [Wed, 25 Jul 2007 23:17:02 +0000 (23:17 +0000)]
r24051: more monitor function calls and monitor msg names
convention change.

rafal

16 years agor24041: After metze's ack, check in torture_comment from Zack Krisch
Volker Lendecke [Wed, 25 Jul 2007 09:57:14 +0000 (09:57 +0000)]
r24041: After metze's ack, check in torture_comment from Zack Krisch
<zack.kirsch@isilon.com>

16 years agor24012: Remove duplicate code block (from bad merge).
Andrew Bartlett [Tue, 24 Jul 2007 06:03:03 +0000 (06:03 +0000)]
r24012: Remove duplicate code block (from bad merge).

Andrew Bartlett

16 years agor24011: Keep the connect handle around in libnet, in case we want it.
Andrew Bartlett [Tue, 24 Jul 2007 06:02:32 +0000 (06:02 +0000)]
r24011: Keep the connect handle around in libnet, in case we want it.

Andrew Bartlett

16 years agor24010: Fix warning for the function paramter to qsort().
Andrew Bartlett [Tue, 24 Jul 2007 06:01:30 +0000 (06:01 +0000)]
r24010: Fix warning for the function paramter to qsort().

Andrew Bartlett

16 years agor23995: Work to allow mimir's libnet code to be called from winbind.
Andrew Bartlett [Mon, 23 Jul 2007 02:56:51 +0000 (02:56 +0000)]
r23995: Work to allow mimir's libnet code to be called from winbind.

We now setup a libnet_ctx for each domain.  We should then be able to
replace/merge some more of the winbind code with libnet calls,
referencing domain->libnet_ctx.

Andrew Bartlett

16 years agor23994: Finish my work to ensure that non-root and non-administrator users
Andrew Bartlett [Mon, 23 Jul 2007 02:10:11 +0000 (02:10 +0000)]
r23994: Finish my work to ensure that non-root and non-administrator users
cannot vampire, provision or upgrade a Samba4 server via SWAT.

(The previous commit was an accident, and not complete).

This should get Samba4 closer to being 'secure' for an alpha release.

Andrew Bartlett

16 years agor23993: Attempt to fix bug #4808, reported by mwallnoefer@yahoo.de. The issue
Andrew Bartlett [Mon, 23 Jul 2007 01:46:39 +0000 (01:46 +0000)]
r23993: Attempt to fix bug #4808, reported by mwallnoefer@yahoo.de.  The issue
is that when we all ldb_msg_add_empty(), we might realloc() the
msg->elements array.  We need to ensure the source pointer (when
copying an element from the same msg) is still valid, or the data
copied.

Andrew Bartlett

16 years agor23982: Fix use-after-realloc() found by valgrind and mwallnoefer@yahoo.de.
Andrew Bartlett [Sat, 21 Jul 2007 10:14:46 +0000 (10:14 +0000)]
r23982: Fix use-after-realloc() found by valgrind and mwallnoefer@yahoo.de.

Should fix bug #4804.

Andrew Bartlett

16 years agor23980: Fix one more use of pwrite in expand_file.
Michael Adam [Fri, 20 Jul 2007 16:31:32 +0000 (16:31 +0000)]
r23980: Fix one more use of pwrite in expand_file.

Michael

16 years agor23979: Fix another occurence of (written != requested) as an
Michael Adam [Fri, 20 Jul 2007 15:00:58 +0000 (15:00 +0000)]
r23979: Fix another occurence of (written != requested) as an
error condition to write. This is in tdb_new_database.

Fix one call to tdb_new_database in tdb_open_ex to not
overwrite the newly propagated errno (typically ENOSPC).

Michael

16 years agor23978: Merge r23161 from Samba3:
Michael Adam [Fri, 20 Jul 2007 14:52:47 +0000 (14:52 +0000)]
r23978: Merge r23161 from Samba3:

Add TDB_VOLATILE as open_flag to activate the per-hashchain dead record
optimization.

16 years agor23977: Im prove the pwrite-patch to tdb_expand_file of r23972:
Michael Adam [Fri, 20 Jul 2007 14:23:12 +0000 (14:23 +0000)]
r23977: Im prove the pwrite-patch to tdb_expand_file of r23972:

* prevent infinite loops due to 0 bytes written:
  try once more. if we still get 0 as return,
  set errno to ENOSPC and return -1 (error)

* replace int by correct types (ssize_t and size_t).

* print a warning log message in case "written < requested to write"
  usually this means, that the next call to pwrite will fail
  with return value -1 and set errno accordingly.

  Note that the former error condition "written != requested to write"
  is not a correct error condition of write/pwrite. If this is due
  to an error, a subsequent call to (p)write will reveal the cause
  (typically "no space left on device" - ENOSPC).

Michael

16 years agor23972: Fix a bug in pwrite error detection in tdb_expand_file():
Michael Adam [Thu, 19 Jul 2007 13:46:26 +0000 (13:46 +0000)]
r23972: Fix a bug in pwrite error detection in tdb_expand_file():
The proper error condition is (ret == -1) instead of
(ret != number_of_byte_told_to_write).

Michael

16 years agor23966: It isn't great, but at least now we have some access control in SWAT
Andrew Bartlett [Thu, 19 Jul 2007 07:48:26 +0000 (07:48 +0000)]
r23966: It isn't great, but at least now we have some access control in SWAT

This patch prevents non-root and non-administrator users from running
the provision, upgrade and vampire pages.  *I think* the rest of SWAT
is LDB operations, or otherwise authenticated, so we should now be
secure.

I wish I had a better way to 'prove' we got this right, but this is better than nothing, and moves us closer to an alpha.

Andrew Bartlett

16 years agor23965: Add testing the 'net time' command to the script.
Andrew Bartlett [Thu, 19 Jul 2007 06:44:18 +0000 (06:44 +0000)]
r23965: Add testing the 'net time' command to the script.

Andrew Bartlett

16 years agor23964: Update blackbox selftest scripts to cover more code, and to more
Andrew Bartlett [Thu, 19 Jul 2007 06:40:42 +0000 (06:40 +0000)]
r23964: Update blackbox selftest scripts to cover more code, and to more
consistantly report errors.  (Some were being lost due to the "echo
foo | cmd" calling convention).

Andrew Bartlett

16 years agor23961: Allow SWAT to operate on x86_64 machines.
Andrew Bartlett [Thu, 19 Jul 2007 04:00:32 +0000 (04:00 +0000)]
r23961: Allow SWAT to operate on x86_64 machines.

On machines with a 4 byte int, and a 8 byte pointer, the ESP could would fail.

The problem is that 0 != NULL.  0 is an int (4 bytes) and NULL is a
pointer (8), and this matters critically to varargs functions.

If a 0 was passed as the 'terminating' argument, then only 4 bytes
would be written to the stack, but va_arg(ap, char *) would try and
pull 8, reading uninitalised memory.

Andrew Bartlett

16 years agor23960: Don't destory the 'reason' for terminating the service before printing it.
Andrew Bartlett [Thu, 19 Jul 2007 03:57:44 +0000 (03:57 +0000)]
r23960: Don't destory the 'reason' for terminating the service before printing it.

Andrew Bartlett

16 years agor23959: add more monitor messages support that's been sitting around on my
Rafal Szczesniak [Wed, 18 Jul 2007 21:24:37 +0000 (21:24 +0000)]
r23959: add more monitor messages support that's been sitting around on my
laptop for a while.

rafal

16 years agor23950: unlink before rename is superfluous.
Michael Adam [Wed, 18 Jul 2007 08:29:00 +0000 (08:29 +0000)]
r23950: unlink before rename is superfluous.

Michael

16 years agor23925: Use NULL instead of 0 for a void * argument.
Michael Adam [Tue, 17 Jul 2007 10:30:13 +0000 (10:30 +0000)]
r23925: Use NULL instead of 0 for a void * argument.

16 years agor23912: We always accept / as a seperator, and it is far less confusing
Andrew Bartlett [Tue, 17 Jul 2007 05:40:36 +0000 (05:40 +0000)]
r23912: We always accept / as a seperator, and it is far less confusing
compared with a shell-escape (\).

Fixes bug #4765

Andrew Bartlett

16 years agor23907: Fix bug 4790 reported by mwallnoefer@yahoo.de:
Andrew Bartlett [Tue, 17 Jul 2007 00:10:53 +0000 (00:10 +0000)]
r23907: Fix bug 4790 reported by mwallnoefer@yahoo.de:

Before the provisioning enters to the function provision_default_paths (in
scripting/libjs/provision.js), the variable subobj.DNSDOMAIN isn't properly set
(for example for the filename of the DNS zonefile).

Andrew Bartlett

16 years agor23905: SATOH Fumiyasu <fumiyas@osstech.jp> points out that we want &&, not ; here...
Andrew Bartlett [Mon, 16 Jul 2007 23:38:29 +0000 (23:38 +0000)]
r23905: SATOH Fumiyasu <fumiyas@osstech.jp> points out that we want &&, not ; here...

(We don't want to make a distclean of the main user tree, just because
they don't have the parent directory checked out).

Andrew Bartlett

16 years agor23895: reapply rev 23493:
Stefan Metzmacher [Mon, 16 Jul 2007 15:03:03 +0000 (15:03 +0000)]
r23895: reapply rev 23493:

regenerate lex.c files with flex 2.5.33
this makes sure we include config.h as first header

hopefully fixes the build on SerNet-aix

abartlet: please don't revert that again with your next
          heimdal merge...:-)

metze

16 years agor23890: Allow wbinfo -a to work against Samba4's winbind.
Andrew Bartlett [Mon, 16 Jul 2007 11:27:29 +0000 (11:27 +0000)]
r23890: Allow wbinfo -a to work against Samba4's winbind.

Add a test for wbinfo -a to test_member.sh

Reimplement the server-side 'pam_auth' and 'pam_auth_crap' calls to
use the same SamLogon code as auth_winbind uses.

In my previous code, we did not bind to the LSA and SAMR pipes, before
attempting operations.  We now do this (how we passed any tests before
is beyond me).

This required some rework, particularly to make it easier to setup
secondary connections.  The new rpc_secondary_auth_connection()
function also performs the bind.

The dcerpc_connect.c file was getting to big, so things have been
merged into dcerpc_secondary.c.

Andrew Bartlett

16 years agor23881: A quick fix from davecb@spamcop.net to be more portable to non-GNU
Andrew Bartlett [Mon, 16 Jul 2007 01:48:07 +0000 (01:48 +0000)]
r23881: A quick fix from davecb@spamcop.net to be more portable to non-GNU
make in autogen.sh.

Andrew Bartlett

16 years agor23880: Don't crash when we run wbinfo -a against our own winbind when we are a DC.
Andrew Bartlett [Sun, 15 Jul 2007 10:46:34 +0000 (10:46 +0000)]
r23880: Don't crash when we run wbinfo -a against our own winbind when we are a DC.

Next step is to make it work...

Andrew Bartlett

16 years agor23876: Prepare to run nsstest from make test, just not add it to tests_all.sh yet.
Kai Blin [Sun, 15 Jul 2007 07:51:31 +0000 (07:51 +0000)]
r23876: Prepare to run nsstest from make test, just not add it to tests_all.sh yet.

16 years agor23875: As pointed out by mwallnoefer@yahoo.de:
Andrew Bartlett [Sat, 14 Jul 2007 00:12:25 +0000 (00:12 +0000)]
r23875: As pointed out by mwallnoefer@yahoo.de:
On default Active Directory installations, the NETLOGON share isn't
an indipendent directory. In fact it's mapped to the subdirectory
"scripts" from the share SYSVOL under <Domain name>.

Andrew Bartlett

16 years agor23862: Explain who requested unknown dependency. Helps a lot in chasing dependency...
Alexander Bokovoy [Fri, 13 Jul 2007 10:02:44 +0000 (10:02 +0000)]
r23862: Explain who requested unknown dependency. Helps a lot in chasing dependency hell when trimming down s4 platform

16 years agor23860: export WINBINDD_SOCKET_DIR into the testenv
Stefan Metzmacher [Fri, 13 Jul 2007 08:38:51 +0000 (08:38 +0000)]
r23860: export WINBINDD_SOCKET_DIR into the testenv

metze

16 years agor23859: Work to have Group Policy work 'out of the box' in Samba4.
Andrew Bartlett [Fri, 13 Jul 2007 08:01:36 +0000 (08:01 +0000)]
r23859: Work to have Group Policy work 'out of the box' in Samba4.

This involves creating the SYSVOL and NETLOGON shares at provision
time, and creating the right subdirectories.

This also changes the behaviour of lp.get("foo") in ejs - we now
return undefined, rather than syntax error, if the parameter doesn't
exist (perhaps because the share isn't defined).

Andrew Bartlett

16 years agor23853: Fix a very misleading error message in tdbbackup.
Michael Adam [Thu, 12 Jul 2007 13:41:34 +0000 (13:41 +0000)]
r23853: Fix a very misleading error message in tdbbackup.

Michael

16 years agor23852: Merge Samba 3.2's wbinfo into Samba4, so Kai can use it for testing.
Andrew Bartlett [Thu, 12 Jul 2007 10:16:59 +0000 (10:16 +0000)]
r23852: Merge Samba 3.2's wbinfo into Samba4, so Kai can use it for testing.

Andrew Bartlett

16 years agor23851: export _ABS versions of PREFIX and SRCDIR
Stefan Metzmacher [Thu, 12 Jul 2007 09:52:32 +0000 (09:52 +0000)]
r23851: export _ABS versions of PREFIX and SRCDIR

metze

16 years agor23849: ldap_server: Provide more info in debug traces
Andrew Bartlett [Thu, 12 Jul 2007 06:15:47 +0000 (06:15 +0000)]
r23849: ldap_server:  Provide more info in debug traces

blackbox tests:  increase test coverage by running more options.

Andrew Bartlett

16 years agor23848: Thanks to derrell for pointing out that I had not finished my patch to
Andrew Bartlett [Thu, 12 Jul 2007 04:56:33 +0000 (04:56 +0000)]
r23848: Thanks to derrell for pointing out that I had not finished my patch to
split out the auth methods.

This caused all SWAT logins to fail, except when using local system
authentication.

Andrew Bartlett

16 years agor23821: Extend the winreg test by a test for DeleteKey on a
Michael Adam [Tue, 10 Jul 2007 21:35:20 +0000 (21:35 +0000)]
r23821: Extend the winreg test by a test for DeleteKey on a
key that has subkeys. This should return WERR_ACCESS_DENIED.

Michael

16 years agor23816: A little more static, but leave the dead code testjoin.c as documentation.
Andrew Bartlett [Tue, 10 Jul 2007 13:41:00 +0000 (13:41 +0000)]
r23816: A little more static, but leave the dead code testjoin.c as documentation.

Andrew Bartlett

16 years agor23815: Thanks to Matthias Wallnoefer <mwallnoefer@yahoo.de> for pointing out
Andrew Bartlett [Tue, 10 Jul 2007 13:26:10 +0000 (13:26 +0000)]
r23815: Thanks to Matthias Wallnoefer <mwallnoefer@yahoo.de> for pointing out
that we had the wrong objectClass for OU=Domain
Controllers,${DOMAINDN} (was CN=Domain Controllers,${DOMAINDN})

This fixes both the SAMR server and the LDIF templates.

Andrew Bartlett

16 years agor23812: Remove more code found as dead by the find_static script, and make
Andrew Bartlett [Tue, 10 Jul 2007 11:56:01 +0000 (11:56 +0000)]
r23812: Remove more code found as dead by the find_static script, and make
other functions just static.

Andrew Bartlett

16 years agor23811: Try to ensure struct nbt_name is always pre-declared. Might fix the
Andrew Bartlett [Tue, 10 Jul 2007 11:54:49 +0000 (11:54 +0000)]
r23811: Try to ensure struct nbt_name is always pre-declared.  Might fix the
build on aico.

Andrew Bartlett

16 years agor23810: Make things static, and remove unsued code.
Andrew Bartlett [Tue, 10 Jul 2007 11:37:30 +0000 (11:37 +0000)]
r23810: Make things static, and remove unsued code.

This includes some of the original ildap ldap client API.  ldb
provides a much easier abstraction on this to use, and doesn't use
these functions.

Andrew Bartlett

16 years agor23809: Don't give users the fantasy that we can control choice of GENSEC
Andrew Bartlett [Tue, 10 Jul 2007 10:50:44 +0000 (10:50 +0000)]
r23809: Don't give users the fantasy that we can control choice of GENSEC
security mechanisms at the moment.  I'll put this back when I
implement the functionality.

Andrew Bartlett

16 years agor23807: added hex_encode_talloc()
Andrew Tridgell [Tue, 10 Jul 2007 08:07:42 +0000 (08:07 +0000)]
r23807: added hex_encode_talloc()

16 years agor23806: update Samba4 with the latest ctdb code.
Andrew Tridgell [Tue, 10 Jul 2007 08:06:51 +0000 (08:06 +0000)]
r23806: update Samba4 with the latest ctdb code.

This doesn't get the ctdb code fully working in Samba4, it just gets
it building and not breaking non-clustered use of Samba. It will take
a bit longer to update some of the calling ctdb_cluster.c code to make
it work correctly in Samba4.

Note also that Samba4 now only links to the client portion of
ctdb. For the moment I am leaving the ctdbd as a separate daemon,
which you install separately from http://ctdb.samba.org/.

16 years agor23805: this directory is not used any more
Andrew Tridgell [Tue, 10 Jul 2007 08:06:42 +0000 (08:06 +0000)]
r23805: this directory is not used any more