Jeremy Allison [Fri, 25 Jan 2013 18:21:48 +0000 (10:21 -0800)]
Fix bug #9588 - ACLs are not inherited to directories for DFS shares.
We can return with NT_STATUS_OK in an error code path. This
has a really strange effect in that it prevents the ACL editor
in Windows XP from recursively changing ACE entries on sub-directories
after a change in a DFS-root share (we end up returning a path
that looks like: \\IPV4\share1\xptest/testdir with a mixture
of Windows and POSIX pathname separators).
Signed-off-by: Jeremy Allison <jra@samba.org> Reviewed-by: David Disseldorp <ddiss@samba.org>
Autobuild-User(master): David Disseldorp <ddiss@samba.org>
Autobuild-Date(master): Mon Jan 28 13:48:13 CET 2013 on sn-devel-104
Reviewed-by: Andreas Schneider <asn@samba.org>
Autobuild-User(master): Andreas Schneider <asn@cryptomilk.org>
Autobuild-Date(master): Mon Jan 28 11:51:12 CET 2013 on sn-devel-104
s4:service_task: add missing imessaging_cleanup() to task_server_terminate()
Signed-off-by: Stefan Metzmacher <metze@samba.org> Reviewed-by: Andrew Bartlett <abartlet@samba.org>
Autobuild-User(master): Stefan Metzmacher <metze@samba.org>
Autobuild-Date(master): Sun Jan 27 15:50:30 CET 2013 on sn-devel-104
selftest: rename 'promoted_vampire_dc' to 'promoted_dc'
Unix domain socket are limited to 104 characters on Linux.
Using something like this fails as it uses more than 104 characters:
'/memdisk/autobuild/flakey/b232141/samba/bin/ab/promoted_vampire_dc/private/smbd.tmp/msg/msg.482379.2147483647'
Signed-off-by: Stefan Metzmacher <metze@samba.org> Reviewed-by: Andrew Bartlett <abartlet@samba.org>
Andrew Bartlett [Tue, 22 Jan 2013 12:39:15 +0000 (23:39 +1100)]
selftest: Add test of upgradeprovision using the old alpha13 tree
This ensures that upgradeprovision works as expected on a known good old database.
Andrew Bartlett
Reviewed-by: Stefan Metzmacher <metze@samba.org>
Autobuild-User(master): Andrew Bartlett <abartlet@samba.org>
Autobuild-Date(master): Sun Jan 27 11:55:54 CET 2013 on sn-devel-104
samba_upgradeprovision: detect dns_backend for the reference provision
If we have a DomainDnsZone partition, we use BIND9_DLZ as backend
and fix errors in the ForestDnsZone and DomainDnsZone partitions.
Note: this should work fine also for SAMBA_INTERNAL.
If the current setup doesn't use dns specific partitions (e.g. alpha13 setups)
we pass dns_backend=BIND9_FLATFILE.
Signed-off-by: Stefan Metzmacher <metze@samba.org> Reviewed-by: Andrew Bartlett <abartlet@samba.org>
If we have a DomainDnsZone partition:
- we use BIND9_DLZ as backend if a dns-<netbiosname> account is available
- otherwise, we use SAMBA_INTERNAL
else:
- we use BIND9_FLATFILE if a dns or dns-<netbiosname> account is available
- otherwise, we use NONE
Signed-off-by: Stefan Metzmacher <metze@samba.org> Reviewed-by: Andrew Bartlett <abartlet@samba.org>
Andrew Bartlett [Fri, 25 Jan 2013 12:00:12 +0000 (23:00 +1100)]
s4-process_single: Use pid,task_id as cluster_id in process_single just like process_prefork
This avoids two different process single task servers (eg the drepl
server) sharing the same server id. The task id starts at 2^31 to
avoid collision with the fd based scheme for connections.
Reported-by: Matthieu Patou <mat@matws.net> Reviewed-by: Stefan Metzmacher <metze@samba.org>
Autobuild-User(master): Stefan Metzmacher <metze@samba.org>
Autobuild-Date(master): Sat Jan 26 16:13:05 CET 2013 on sn-devel-104
Andrew Bartlett [Fri, 25 Jan 2013 22:35:21 +0000 (09:35 +1100)]
ldb: Ensure to decrement the transaction_active whenever we delete a transaction
This is in the error path for prepare_commit, which rarely fails, but
when it does we need to ensure that when a new transaction is opened,
that it really starts a new transaction.
We bump the version to recognise critical fix for the AD DC
Without this fix, a single invalid inbound replicated link disables
all subsequent replication as we operate without a transaction (which
is refused by ldb_tdb).
Andrew Bartlett
Reviewed-by: Matthieu Patou <mat@matws.net> Reviewed-by: Stefan Metzmacher <metze@samba.org>
Jeremy Allison [Thu, 24 Jan 2013 20:33:53 +0000 (12:33 -0800)]
Regression test for bug #9587 - archive flag is always set on directories.
Ensure we get the correct attributes on files
and directories after a rename.
Signed-off-by: Jeremy Allison <jra@samba.org> Reviewed-by: David Disseldorp <ddiss@samba.org>
Autobuild-User(master): David Disseldorp <ddiss@samba.org>
Autobuild-Date(master): Fri Jan 25 13:42:40 CET 2013 on sn-devel-104
Jeremy Allison [Thu, 24 Jan 2013 19:02:30 +0000 (11:02 -0800)]
Fix bug #9587 - archive flag is always set on directories.
Creating a directory to a Samba share sets the attributes to 'D' only
(correct) - only when creating a new file should the 'A' attribute
be set.
However, doing a rename of that directory sets the 'A' attribute in error.
This should only be done on a file rename. smbclient regression test to follow.
Signed-off-by: Jeremy Allison <jra@samba.org> Reviewed-by: David Disseldorp <ddiss@samba.org>
Reported-by: Matthieu Patou <mat@matws.net> Reviewed-by: Matthieu Patou <mat@matws.net> Signed-off-by: Andrew Bartlett <abartlett@samba.org>
Autobuild-User(master): Andrew Bartlett <abartlet@samba.org>
Autobuild-Date(master): Fri Jan 25 12:00:04 CET 2013 on sn-devel-104
Volker Lendecke [Thu, 24 Jan 2013 15:39:05 +0000 (16:39 +0100)]
Avoid a very small memleak on talloc_tos()
"fname" did leak on talloc_tos(). Not really a bad memleak, but as I
just came across it I thought I might just fix it
Reviewed-by: Jeremy Allison <jra@samba.org>
Autobuild-User(master): Jeremy Allison <jra@samba.org>
Autobuild-Date(master): Fri Jan 25 00:54:01 CET 2013 on sn-devel-104
Jeremy Allison [Wed, 23 Jan 2013 17:57:50 +0000 (09:57 -0800)]
Fix bug #9572 - File corruption during SMB1 read by Mac OSX 10.8.2 clients.
Accept a large read if we told the client we have UNIX extensions
and the client sent a non-zero upper 16-bit size.
Do the non-zero upper 16-bit size check first to save a function
call in what is a hot path.
Signed-off-by: Jeremy Allison <jra@samba.org> Reviewed-by: Andrew Bartlett <abartlet@samba.org> Reviewed-by: Stefan Metzmacher <metze@samba.org>
Autobuild-User(master): Jeremy Allison <jra@samba.org>
Autobuild-Date(master): Thu Jan 24 21:01:51 CET 2013 on sn-devel-104
Günther Deschner [Wed, 23 Jan 2013 09:33:21 +0000 (10:33 +0100)]
s4-torture: add some basic tests for PlayGDIScriptOnPrinterIC.
Guenther
Signed-off-by: Günther Deschner <gd@samba.org> Reviewed-by: Andreas Schneider <asn@samba.org>
Autobuild-User(master): Andreas Schneider <asn@cryptomilk.org>
Autobuild-Date(master): Thu Jan 24 19:20:52 CET 2013 on sn-devel-104
Reviewed-by: Stefan Metzmacher <metze@samba.org>
Autobuild-User(master): Stefan Metzmacher <metze@samba.org>
Autobuild-Date(master): Thu Jan 24 16:09:37 CET 2013 on sn-devel-104
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Autobuild-User(master): Stefan Metzmacher <metze@samba.org>
Autobuild-Date(master): Wed Jan 23 20:04:09 CET 2013 on sn-devel-104
waf: Fix pdb_ldap which cannot be built as a module.
The module has two init functions, pdb_ldap_init() and
pdb_ldapsam_init(). As a shared module only one can be found until we
create a symlink.
Reviewed-by: Günther Deschner <gd@samba.org>
Autobuild-User(master): Andreas Schneider <asn@cryptomilk.org>
Autobuild-Date(master): Wed Jan 23 10:51:59 CET 2013 on sn-devel-104
Andrew Bartlett [Sun, 20 Jan 2013 23:45:10 +0000 (10:45 +1100)]
libcli/auth: fix void function cannot return value error
Reviewed-by: Jeremy Allison <jra@samba.org>
Autobuild-User(master): Jeremy Allison <jra@samba.org>
Autobuild-Date(master): Tue Jan 22 22:32:31 CET 2013 on sn-devel-104
Matthieu Patou [Thu, 3 Jan 2013 22:33:45 +0000 (14:33 -0800)]
Tests: Fix the display of test vars in screen --testenv
The form bash -c echo "important stuff blabla bla" && LD_LIBARY_PATH bash
is not working in screen when it's working in xterm and the in_screen
script already wrap all the command within a bash shell so there is no
need to re-force bash as the echo will execute in a bash shell
Signed-off-by: Matthieu Patou <mat@matws.net> Reviewed-by: Andrew Bartlett <abartlet@samba.org>
Autobuild-User(master): Andrew Bartlett <abartlet@samba.org>
Autobuild-Date(master): Tue Jan 22 13:03:52 CET 2013 on sn-devel-104
Matthieu Patou [Tue, 30 Oct 2012 05:12:33 +0000 (22:12 -0700)]
devel-script: add options for RODC and partial replica for replicate flags
Reviewed-by: Andrew Bartlett <abartlet@samba.org>
Autobuild-User(master): Andrew Bartlett <abartlet@samba.org>
Autobuild-Date(master): Tue Jan 22 00:12:17 CET 2013 on sn-devel-104
Matthieu Patou [Sun, 30 Dec 2012 00:43:44 +0000 (16:43 -0800)]
dsdb: Fix warning about unused var
Reviewed-by: Stefan Metzmacher <metze@samba.org>
Autobuild-User(master): Stefan Metzmacher <metze@samba.org>
Autobuild-Date(master): Mon Jan 21 17:51:16 CET 2013 on sn-devel-104
Andrew Bartlett [Sat, 29 Dec 2012 04:13:54 +0000 (15:13 +1100)]
dsdb: Ensure "authenticated users" is processed for group memberships
This change moves the addition of "Authenticated Users" from the very end of the
token processing to the start. The reason is that we need to see if
"Authenticated Users" is a member of other builtin groups, just as we
would for any other SID. This picks up the "Pre-Windows 2000 Compatible Access"
group, which is in turn often used in ACLs on LDAP objects.
Without this change, the eventual token does not contain S-1-5-32-554
and users other than "Administrator" are unable to read uidNumber
(in particular).
Andrew Bartlett
Reviewed-by: Stefan Metzmacher <metze@samba.org> Reviewed-by: Andrew Bartlett <abartlet@samba.org>
Andrew Bartlett [Thu, 3 Jan 2013 09:40:32 +0000 (20:40 +1100)]
libcli/security: handle node initialisation in one spot in insert_in_object_tree()
This removes special-case for initalising the children array in
insert_in_object_tree(). talloc_realloc() handles the intial allocate
case perfectly well, so there is no need to have this duplicated.
This also restores having just one place were the rest of the elements
are intialised, to ensure uniform behaviour.
To do this, we have to rework insert_in_object_tree to have only one
output variable, both because having both root and new_node as output
variables was too confusing, and because otherwise the two pointers
were being allowed to point at the same memory.
Andrew Bartlett
Signed-off-by: Stefan Metzmacher <metze@samba.org> Reviewed-by: Stefan Metzmacher <metze@samba.org> Reviewed-by: Andrew Bartlett <abartlet@samba.org>
Andrew Bartlett [Wed, 2 Jan 2013 04:01:23 +0000 (15:01 +1100)]
dsdb-acl: Use the structural objectClass in acl_check_access_on_attribute()
This commit enters the GUID into the object tree so that that access
rights assigned to the structural objectClass are also available, as
well as rights assigned to the attribute property groups.
Andrew Bartlett
Signed-off-by: Stefan Metzmacher <metze@samba.org> Reviewed-by: Stefan Metzmacher <metze@samba.org> Reviewed-by: Andrew Bartlett <abartlet@samba.org>
Andrew Bartlett [Wed, 2 Jan 2013 04:01:00 +0000 (15:01 +1100)]
dsdb-acl: Pass the structural objectClass into acl_check_access_on_attribute
This will, when the GUID is entered into the object tree (not in this
commit) ensure that access rights assigned to the structural
objectClass are also available, as well as rights assigned to the
attribute property groups.
Andrew Bartlett
Signed-off-by: Stefan Metzmacher <metze@samba.org> Reviewed-by: Stefan Metzmacher <metze@samba.org> Reviewed-by: Andrew Bartlett <abartlet@samba.org>
Andrew Bartlett [Tue, 1 Jan 2013 22:26:15 +0000 (09:26 +1100)]
dsdb-acl: use dsdb_get_structural_oc_from_msg() rather than class_schemaid_guid_by_lDAPDisplayName
This uses dsdb_get_last_structural_objectclass(), which encodes this ordering
knowledge in one place in the code, rather than using this uncommented
magic expression:
(char *)oc_el->values[oc_el->num_values-1].data
Andrew Bartlett
Reviewed-by: Stefan Metzmacher <metze@samba.org> Reviewed-by: Andrew Bartlett <abartlet@samba.org>