third_party: Update pam_wrapper to version 1.0.7

Signed-off-by: Andreas Schneider <asn@samba.org>
Reviewed-by: Guenther Deschner <gd@samba.org>

python/tests: turn GraphError into failure inside of test_verify()

That test was already marked as flapping in commit
5a73f904e192c44e304850287ac439d0b52f7be5.

However the test generates an UNEXPECTED(error)
instead of an UNEXPECTED(failure).

  [67(594)/75 at 36m28s] samba.tests.kcc.python3(vampire_dc)
  repsFrom source DSA guid (62f01940-a784-4e60-947a-a661c787c8cc) not found
  repsFrom source DSA guid (62f01940-a784-4e60-947a-a661c787c8cc) not found
  repsFrom source DSA guid (62f01940-a784-4e60-947a-a661c787c8cc) not found
  repsFrom source DSA guid (62f01940-a784-4e60-947a-a661c787c8cc) not found
  repsFrom source DSA guid (62f01940-a784-4e60-947a-a661c787c8cc) not found
  UNEXPECTED(error): samba.tests.kcc.python3.samba.tests.kcc.KCCTests.test_verify(vampire_dc)
  REASON: Exception: Exception: Traceback (most recent call last):
    File "bin/python/samba/tests/kcc/__init__.py", line 80, in test_verify
      attempt_live_connections=False)
    File "bin/python/samba/kcc/__init__.py", line 2659, in run
      ('connected',))
    File "bin/python/samba/kcc/__init__.py", line 2513, in plot_all_connections
      vertex_colors=vertex_colours)
    File "bin/python/samba/kcc/graph_utils.py", line 334, in verify_and_dot
      for p, e, doc in errors)))
  samba.kcc.graph_utils.GraphError: The 'dsa_final CN=NTDS Settings,CN=LOCALVAMPIREDC,CN=Servers,CN=Default-First-Site-Name,CN=Sites,CN=Configuration,DC=samba,DC=example,DC=com' graph lacks the following properties:
  connected: the graph is not connected, as the following vertices are unreachable:

Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Andreas Schneider <asn@samba.org>

smb2_server: set req->do_encryption = true earlier

The STATUS_SESSION_EXPIRED error was returned unencrypted,
if the request was encrypted.

If clients use SMB3 encryption and the kerberos authenticated session
expires, clients disconnect the connection instead of doing a reauthentication.

From https://blogs.msdn.microsoft.com/openspecification/2012/10/05/encryption-in-smb-3-0-a-protocol-perspective/

  The sender encrypts the message if any of the following conditions is
  satisfied:

    - If the sender is sending a response to an encrypted request.
    - If Session.EncryptData is TRUE and the request or response being
      sent is not NEGOTIATE.
    - If Session.EncryptData is FALSE, the request or response being sent
      is not NEGOTIATE or SESSION_SETUP or TREE_CONNECT, and
      <TreeConnect|Share>.EncryptData is TRUE.

[MS-SMB2] 3.3.4.1.4 Encrypting the Message

 If Connection.Dialect belongs to the SMB 3.x dialect family and
 Connection.ClientCapabilities includes the SMB2_GLOBAL_CAP_ENCRYPTION
 bit, the server MUST encrypt the message before sending, if any of the
 following conditions are satisfied:

 - If the message being sent is any response to a client request for which
   Request.IsEncrypted is TRUE.

 - If Session.EncryptData is TRUE and the response being sent is not
   SMB2_NEGOTIATE or SMB2 SESSION_SETUP.

 - If Session.EncryptData is FALSE, the response being sent is not
   SMB2_NEGOTIATE or SMB2 SESSION_SETUP or SMB2 TREE_CONNECT, and
   Share.EncryptData for the share associated with the TreeId in the SMB2
   header of the response is TRUE.

 The server MUST encrypt the message as specified in section 3.1.4.3,
 before sending it to the client.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=13624

Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
Autobuild-User(master): Volker Lendecke <vl@samba.org>
Autobuild-Date(master): Tue Oct  2 14:11:30 CEST 2018 on sn-devel-144

s4:torture: split smb2.session.expire{1,2} to run with signing and encryptpion

This reproduces the problem we have with expired encrypted sessions.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=13624

Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>

s3/locking: Fix logging of lock reference count

lock reference count is always increased and reduced by a value of 1.
But lock_ref_count variable holds the old value prior to change and
was being logged wrongly under debug level 10. DEBUG statement must
log lock_ref_count+1 and lock_ref_count-1 respectively when value
gets increased and decreased.

Signed-off-by: Anoop C S <anoopcs@redhat.com>
Reviewed-by: Andreas Schneider <asn@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>

libsmb: Remove smb_share_modes.[ch]

This was declared nonfunctional in 2014. Finally remove it.

Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>

s3: smbd: Prevent valgrind errors in smbtorture3 POSIX test.

Missing fsp talloc free and linked list delete in error
paths in close_directory(). Now matches close_normal_file()
and close_fake_file().

BUG: https://bugzilla.samba.org/show_bug.cgi?id=13633

Signed-off-by: Jeremy Allison <jra@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
Autobuild-User(master): Jeremy Allison <jra@samba.org>
Autobuild-Date(master): Sat Sep 29 05:32:41 CEST 2018 on sn-devel-144

libcli: Add debug message if fail to negoatiate SMB protocol

Currently if the client and server can't negotiate an SMB protocol, you
just get the followiing error on the client-side, which doesn't tell you
much.
ERROR(runtime): uncaught exception - (3221225667, 'The network responded
incorrectly.')

This patch adds a debug message to help highlight what's actually going
wrong.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=13621

Signed-off-by: Tim Beale <timbeale@catalyst.net.nz>
Reviewed-by: Ralph Boehme <slow@samba.org>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>
Autobuild-User(master): Andrew Bartlett <abartlet@samba.org>
Autobuild-Date(master): Fri Sep 28 11:25:29 CEST 2018 on sn-devel-144

s3/smbd: Server responds incorrectly if no SMB protocol chosen

The SMBnegprot response from the server contains the DialectIndex of the
selected protocol from the client's request message. Currently, if no
protocol is selected, the server is responding with a DialectIndex=zero,
which is a valid index (PROTOCOL_CORE by default). The Windows spec, and
historically the code, should return DialectIndex=0xffff if no protocol
is chosen. The following commit changed it recently (presumably
inadvertently), so that it now returns DialectIndex=zero.

06940155f315529c5b5 s3:smbd: Fix size types in reply_negprot()

This results in somewhat confusing error messages on the client side:
ERROR(runtime): uncaught exception - (3221225997, 'The transport
connection has been reset.')

or, when signing is configured as mandatory:
smbXcli_negprot: SMB signing is mandatory and the selected protocol
level (1) doesn't support it.
ERROR(runtime): uncaught exception - (3221225506, '{Access Denied} A
process has requested access to an object but has not been granted those
access rights.')

This patch restores the old behaviour of returning 0xffff.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=13621

Pair-Programmed-With: Ralph Boehme <slow@samba.org>
Signed-off-by: Tim Beale <timbeale@catalyst.net.nz>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>

netcmd: Make sure SMB connection is signed when backing up sysvol

i.e. protect the client against man-in-the-middle attacks by default.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=13621

Signed-off-by: Tim Beale <timbeale@catalyst.net.nz>
Reviewed-by: Ralph Boehme <slow@samba.org>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>

s4/web_server: fix panic from wrong type extracted from conn

enabling
  server services = +web

and we get a panic

web_server.c:251: Type mismatch: name[struct web_server_data] expected[struct task_server]
smb_panic_default: PANIC (pid 29137): ../source4/web_server/web_server.c:251: Type mismatch: name[struct web_server_data] expected[struct task_server]

Signed-off-by: Noel Power <noel.power@suse.com>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>
Autobuild-User(master): Andrew Bartlett <abartlet@samba.org>
Autobuild-Date(master): Fri Sep 28 03:57:47 CEST 2018 on sn-devel-144

add new gitlab CI job for building pure python3

Signed-off-by: Noel Power <noel.power@suse.com>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>

script: add new autobuild task for building pure python3

Signed-off-by: Noel Power <noel.power@suse.com>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>

PY3: We support python3 now, remove error when python3 detected

Signed-off-by: Noel Power <noel.power@suse.com>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>

buildtools/wafsamba: Finally fix reference to basestring PY3 error.

While a previous attempt squashed the error on the config & make
phase, make install threw up this error again.

Signed-off-by: Noel Power <noel.power@suse.com>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>

dlz_bind9: torture tests exploring rndc reload behaviour

These tests establish that the process triggered by the command 'rndc reload'
does not cause samba's bind9 dlz plugin to crash.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=13214

Signed-off-by: Aaron Haslett <aaronhaslett@catalyst.net.nz>
Reviewed-by: Douglas Bagnall <douglas.bagnall@catalyst.net.nz>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>
Autobuild-User(master): Andrew Bartlett <abartlet@samba.org>
Autobuild-Date(master): Thu Sep 27 04:47:16 CEST 2018 on sn-devel-144

netcmd: domain backup offline bug fix - ignore sock files

Ignoring autogenerated .sock files and directories during file system based
offline domain backup.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=13604
Signed-off-by: Aaron Haslett <aaronhaslett@catalyst.net.nz>
Reviewed-by: Douglas Bagnall <douglas.bagnall@catalyst.net.nz>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>

buildtools/wafsamba: add_manual_dependency needs bytes for value

Signed-off-by: Noel Power <noel.power@suse.com>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>

buildtools/wafsamba: fix basestring not defined error in PY3

Test for str first (which exists in py3 & py2) this avoids
the undefined runtime error.

Signed-off-by: Noel Power <noel.power@suse.com>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>

s4/web_server: PY3: port to python3

Note: Unlike other libraries this library is been only built for the
configured python version. It depends on availability of 'swat' python
module.

The swat module is hosted externally, it seems not to have been modified
for a number of years, I don't think swat is python3 compatabile. These
changes are enough to get allow the samba binary to launch a web server
(which will just display a placeholder page announcing you need to install
swat). It maybe that removing this functionality is what we should do, but
that is a decision that can be made at a later time.

Signed-off-by: Noel Power <noel.power@suse.com>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>

python/samba: PY3 add compat function urllib_join to replace urllib.urljoin

Signed-off-by: Noel Power <noel.power@suse.com>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>

PY3: md5 related functions need to be passed bytes

Signed-off-by: Noel Power <noel.power@suse.com>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>

PY3: decode bytes in py3 where strings are needed

Signed-off-by: Noel Power <noel.power@suse.com>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>

PY3: fix "TabError: inconsistent use of tabs and spaces"

Signed-off-by: Noel Power <noel.power@suse.com>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>

lib/ldb: Ensure ldb.Dn can accept utf8 encoded unicode

Additionally remove the associated known fail.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=13616
Signed-off-by: Noel Power <noel.power@suse.com>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>

lib/ldb/tests: add test for ldb.Dn passed utf8 unicode

object dn format should be a utf8 encoded string
Note: Currently this fails in python2 as the c python binding for
      the dn string param uses PyArg_ParseTupleAndKeywords() with 's'
      format, this will accept str *or* unicode in the default encoding.
      The default encoding in python2 is... ascii.

Also adding here a knownfail to squash the error produced by the test.

Signed-off-by: Noel Power <noel.power@suse.com>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>

lib/ldb: Test correct variable for no mem condition

Signed-off-by: Noel Power <noel.power@suse.com>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>

py3 kcc tests: mark verify test as flapping

Signed-off-by: Douglas Bagnall <douglas.bagnall@catalyst.net.nz>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>
Autobuild-User(master): Andrew Bartlett <abartlet@samba.org>
Autobuild-Date(master): Wed Sep 26 23:40:58 CEST 2018 on sn-devel-144

turn --with-json-audit into global --with-json

Fold the build option --with-json-audit into the toplevel wscript
to reflect the fact that JSON support is no longer local to the
audit subsystem.

Signed-off-by: Philipp Gesang <philipp.gesang@intra2net.com>
Reviewed-by: Douglas Bagnall <douglas.bagnall@catalyst.net.nz>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>

docs: Only build vfs manpages if the module is enabled

Signed-off-by: Andreas Schneider <asn@samba.org>
Reviewed-by: Guenther Deschner <gd@samba.org>
Autobuild-User(master): Günther Deschner <gd@samba.org>
Autobuild-Date(master): Wed Sep 26 20:10:07 CEST 2018 on sn-devel-144

samba-tool domain backup: fix py2.6 incompatible format

Signed-off-by: Douglas Bagnall <douglas.bagnall@catalyst.net.nz>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>
Autobuild-User(master): Andrew Bartlett <abartlet@samba.org>
Autobuild-Date(master): Wed Sep 26 10:40:49 CEST 2018 on sn-devel-144

domain_backup test: fix py2.6 incompatible format

Signed-off-by: Douglas Bagnall <douglas.bagnall@catalyst.net.nz>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>

join: Support site detection when --server is specified

When a new DC is joined to the domain, samba-tool would automatically
detect an appropriate site for the new DC. However, it only did this if
the --server option wasn't specified. The new DC's site got
automatically updated as part of the finddc() work, however, this step
gets skipped if we already know the server DC to join to.

In other words, if Default-First-Site-Name doesn't exist and you specify
--server in the join, then you have to also specify --site manually,
otherwise the command fails. This is precisely what's happening in the
join_ldapcmp.sh test, now that the backupfromdc testenv no longer has the
Default-First-Site-Name present.

This patch adds a new find_dc_site() function which uses the same
net.finddc() API (except based on the server-address rather than
domain-name). Assigning DEFAULTSITE has been moved so that it only
gets done if finddc() can't determine the site.

Signed-off-by: Tim Beale <timbeale@catalyst.net.nz>
Reviewed-by: Douglas Bagnall <douglas.bagnall@catalyst.net.nz>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>

join: Avoid duplicating "Default-First-Site-Name" string

The provision code already defines "Default-First-Site-Name" so we might
as well reuse it.

The join.py already uses a suitable default, so assigning the default in
the domain netcmd code is unnecessary.

Signed-off-by: Tim Beale <timbeale@catalyst.net.nz>
Reviewed-by: Douglas Bagnall <douglas.bagnall@catalyst.net.nz>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>

selftest: Test join against DC with non-default site

Recent changes around restoring a domain that lacked
Default-First-Site-Name highlighted a problem. Normally when you join a
DC to a domain, samba-tool works out the correct site to use
automatically. However, if the join uses '--server' to select a DC, then
this doesn't work. It defaults back to Default-First-Site-Name, and the
join command fails if this site doesn't exist.

All the testenvs had Default-First-Site-Name present, so this was never
tested. Now the backupfromdc no longer has a Default-First-Site-Name
site, so running a simple join against that DC fails, highlighting the
problem.

Signed-off-by: Tim Beale <timbeale@catalyst.net.nz>
Reviewed-by: Douglas Bagnall <douglas.bagnall@catalyst.net.nz>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>

selftest: Change backup testenvs to use non-default site

Previously (i.e. up until the last patch) the backup/restore commands
only worked if the Default-First-Site-Name site was present. If this
site didn't exist, then the various restore testenvs would fail to
start. This is now fixed, but this patch changes the backupfrom testenv
so that it uses a non-default site. This will detect the problem if it
is ever re-introduced.

To do this we need to change provision_ad_dc() so the
extra_provision_options can be specified as an argument. (Note that Perl
treats undef the same as an empty array).

By default, the restore will add the new DC into the
Default-First-Site-Name site. This means the backupfromdc and restored
testenvs will now have different sites, so we need to update the ldapcmp
filters to exclude site-specific attributes.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=13621

Signed-off-by: Tim Beale <timbeale@catalyst.net.nz>
Reviewed-by: Douglas Bagnall <douglas.bagnall@catalyst.net.nz>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>

netcmd: Re-create default site for backup-restore (if missing)

Normally when a new DC joins a domain, samba-tool works out the new
DC's site automatically. However, it does this by querying the existing
DC using CLDAP. In the restore case, there is no DC running. We could
still query the DB on disk and work out the correct site based on the
new DC's IP, however:
- comparing between the CN=Subnet DNs and an IP-address string seems
  like it'd be non-trivial to write, and
- in the lab-domain rename case, chances are the user will want a
  completely different subnet to what's already in the DB.

The restore command now has a --site option so the user can specify an
appropriate site for the restored DC. This patch makes the restore
command work by default (i.e. without a --site option) even if the
default Default-First-Site-Name doesn't exist. Basically the solution is
to just check Default-First-Site-Name exists and create it if it
doesn't. As the recommended workflow is to use the restored DC as a
temporary seed that you'll later throw away, this approach seems
acceptable. Subsequent DCs will then be joined to the running restored
DC, so an appropriate site will be determined using CLDAP. The only
side-effect is potentially an extra Site object.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=13621

Signed-off-by: Tim Beale <timbeale@catalyst.net.nz>
Reviewed-by: Douglas Bagnall <douglas.bagnall@catalyst.net.nz>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>

tests: Add test-case for restore into non-default site

Add a test-case that exercises the new '--site' restore option and
ensures the restored DC gets added to the correct site.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=13621

Signed-off-by: Tim Beale <timbeale@catalyst.net.nz>
Reviewed-by: Douglas Bagnall <douglas.bagnall@catalyst.net.nz>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>

netcmd: Add --site option when restoring a domain

Restoring a backup only worked if the Default-First-Site-Name site was
still present. When the new restored DC account is created, it was
trying to add the new server's DN under CN=Default-First-Site-Name.
However, if the original domain was setup using a different site, then
the restore would fail because the DN didn't exist.

When running the restore command, you should be able to specify the
site that you want the new/restored DC to be in (same as during a
DC 'join'). Passing the correct --site argument is one way to avoid
this problem. (A subsequent patch will further improve the tool so it
can work around non-default sites automatically).

Note we also need to pass the site through to where the new DNS entries
get registered (in the rename case).

BUG: https://bugzilla.samba.org/show_bug.cgi?id=13621

Signed-off-by: Tim Beale <timbeale@catalyst.net.nz>
Reviewed-by: Douglas Bagnall <douglas.bagnall@catalyst.net.nz>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>

netcmd: Tweak backup-offline output to avoid subunit truncation

Currently a backup-offline test is occasionally flapping in autobuild,
however, the output is truncated so we can't see what the actual problem
is. The output only ever contains the list of backup dirs. I suspect
that the ']' character printed at the end of the python list might be
getting interpretted by subunit as the end of *all* the output.

If so, we should be able to avoid the problem by printing the list items
without the '['/']'s, i.e. join the list into a single string.

Signed-off-by: Tim Beale <timbeale@catalyst.net.nz>
Reviewed-by: Douglas Bagnall <douglas.bagnall@catalyst.net.nz>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>

s3: Fix the build when compiling without JSON support

[3762/3895] Compiling source3/utils/net_ads_gpo.c
../source3/utils/net_ads.c: In function ‘net_ads_cldap_netlogon_json’:
../source3/utils/net_ads.c:311:2: error: parameter name omitted
  (ADS_STRUCT *, const char *,
  ^
../source3/utils/net_ads.c:311:2: error: parameter name omitted
../source3/utils/net_ads.c:312:16: error: parameter name omitted
   const struct NETLOGON_SAM_LOGON_RESPONSE_EX *)
                ^
../source3/utils/net_ads.c: In function ‘net_ads_info_json’:
../source3/utils/net_ads.c:520:1: error: parameter name omitted
 static int net_ads_info_json(ADS_STRUCT *)
 ^

Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Ralph Boehme <slow@samba.org>
Autobuild-User(master): Ralph Böhme <slow@samba.org>
Autobuild-Date(master): Mon Sep 24 07:03:09 CEST 2018 on sn-devel-144

tests/blackbox: add test for net ads JSON output

Implement blackbox tests for

    $ net ads info --json
    $ net ads lookup --json

that validate

    a) JSON wellformedness (by feeding it into the JSON library
       that ships with Python), and
    b) equality of the set of keys printed to that of the
       non-JSON version.

Reviewed-by: Gary Lockyer <gary@catalyst.net.nz>
Signed-off-by: Philipp Gesang <philipp.gesang@intra2net.com>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>
Autobuild-User(master): Andrew Bartlett <abartlet@samba.org>
Autobuild-Date(master): Sat Sep 22 09:20:09 CEST 2018 on sn-devel-144

s3: net: normalize output of lookup subcommand

Use spaces and tabs consistently following the majority of the
printed output: tabs only for indenting, no space before the
colon separator, a single space after the separator.

The irregularities in formatting date back to the original commit
2c029a8b96..

Reviewed-by: Gary Lockyer <gary@catalyst.net.nz>
Signed-off-by: Philipp Gesang <philipp.gesang@intra2net.com>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>

s3: net: implement json output for ads lookup

Add JSON printer (option '--json') for the 'net ads lookup'
command. This outputs the same information as the plain version,
with integral ({LMNT,LM20} Token, NT Version) and boolean values
(Flags) not stringified.

Reviewed-by: Gary Lockyer <gary@catalyst.net.nz>
Signed-off-by: Philipp Gesang <philipp.gesang@intra2net.com>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>

s3: net: implement json output for ads info

Add the switch '--json' to 'net' to format the output as JSON.

The rationale is to supply the information in a machine-readable
fashion to complement the text version of the output which is
neither particularly well defined nor locale-safe.

The output differs from that of plain 'info' in that times are
not formatted as timestamps.

Currently affects only the 'net ads info' subcommand.

Reviewed-by: Gary Lockyer <gary@catalyst.net.nz>
Signed-off-by: Philipp Gesang <philipp.gesang@intra2net.com>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>

lib/audit_logging: make json_{is_invalid,to_string}() accept a const*

Allow for json_is_invalid() and json_to_string() to be used on a
const pointer. Neither function requires for the json object to
be mutable so constraining them to non-const* is unnecessary.

Reviewed-by: Gary Lockyer <gary@catalyst.net.nz>
Signed-off-by: Philipp Gesang <philipp.gesang@intra2net.com>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>

selftest: Tweak PSO test-suite name

There are 2 different PSO tests:
- make test TESTS=ldap.password_settings
- make test TESTS=samba_tool.passwordsettings

There's also another test that's completely unrelated to PSOs:
- make test TESTS=blackbox.password_settings

This patch renames ldap.password_settings --> ldap.passwordsettings.
This means 'make test TESTS=passwordsettings' will run both PSO tests,
but not the unrelated blackbox test.

Signed-off-by: Tim Beale <timbeale@catalyst.net.nz>
Reviewed-by: Douglas Bagnall <douglas.bagnall@catalyst.net.nz>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>
Autobuild-User(master): Andrew Bartlett <abartlet@samba.org>
Autobuild-Date(master): Fri Sep 21 22:58:17 CEST 2018 on sn-devel-144

tests: test ldap search requesting non-existent attribute

As an after-thought to commit 563e454e8c55e94a950, we thought it
might be a good idea to add a test case that requests an non-existent
attribute in the attribute-filter as well the search-filter.

Signed-off-by: Tim Beale <timbeale@catalyst.net.nz>
Reviewed-by: Douglas Bagnall <douglas.bagnall@catalyst.net.nz>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>

acl_read: Rework Samba code to reflect Windows logic

This patch should not alter functionality. It is just updating the Samba
code to better match the Windows specification docs.

When fixing Samba BUG #13434, the Microsoft behaviour wasn't clearly
documented, so we made a best guess based on observed behaviour.
The problem was an exception was made to allow "objectClass=*" searches
to return objects, even if you didn't have Read Property rights for the
object's objectClass attribute. However, the logic behind what
attributes were and weren't covered by this exception wasn't clear.

I made a guess that it was attributes belonging to the Public Info
property-set that also have the systemOnly flag set.

Microsoft have confirmed the object visibility behaviour. It turns out
that an optimization is made for the 4 attributes that are always
present for every object (i.e. objectClass, distinguishedName,
name, objectGUID). They're updating their Docs to reflect this.

Now that we know the Windows logic, we can update the Samba code.
This simplifies the code somewhat.

Signed-off-by: Tim Beale <timbeale@catalyst.net.nz>
Reviewed-by: Douglas Bagnall <douglas.bagnall@catalyst.net.nz>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>

tests: Minor code cleanups to confidential_attr test

+ fix a couple of flake8 warnings
+ add some extra code comments (particularly around the cases where the
  child class overrides a particular method, to avoid confusion when
  browsing the code).
+ assert_not_in_result() was duplicated (it's only needed for the deny
  ACL tests)
+ skip redundant if in dirsync's assert_search_result() (it always has
  to use the base-DN - we never pass it this as an args).

Signed-off-by: Tim Beale <timbeale@catalyst.net.nz>
Reviewed-by: Douglas Bagnall <douglas.bagnall@catalyst.net.nz>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>

python tests: fix format() strings for Python 2.6

Python 2.6 wants "{0}".format(x), not "{}".format(x).

Signed-off-by: Douglas Bagnall <douglas.bagnall@catalyst.net.nz>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>

ldb: Fix dependency of ldb module

BUG: https://bugzilla.samba.org/show_bug.cgi?id=13626
Signed-off-by: Amitay Isaacs <amitay@gmail.com>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>

heimdal-build: Avoid using python str.format()

In python 2.6, the format fields need to be numbered explicitly.  This
causes the build to fail on RHEL6/Centos6 with following error:

 File "/home/amitay/samba.git/source4/heimdal_build/wscript_build", line 87, in HEIMDAL_ASN1
    os.path.join(bld.path.abspath(), option_file)))
ValueError: zero length field name in format

To use str.format() in HEIMDAL_ASN1(),

   "--option-file='{}'"  needs to be   "--options-file='{0}'"

Or this fix which avoids str.format() completely.

Signed-off-by: Amitay Isaacs <amitay@gmail.com>
Reviewed-by: Martin Schwenke <martin@meltin.net>
Autobuild-User(master): Martin Schwenke <martins@samba.org>
Autobuild-Date(master): Fri Sep 21 03:06:44 CEST 2018 on sn-devel-144

Fix pdb_samba_dsdb build on Ubuntu 16.04 with -O3 -Werror=strict-overflow

../source3/passdb/pdb_samba_dsdb.c: In function ‘pdb_samba_dsdb_set_trusteddom_pw’:
../source3/passdb/pdb_samba_dsdb.c:2778:6: error: assuming signed overflow does not occur when simplifying conditional to constant [-Werror=strict-overflow]
   if (i == 0) {
      ^
cc1: all warnings being treated as errors

gcc version 5.4.0 20160609 (Ubuntu 5.4.0-6ubuntu1~16.04.10)

Signed-off-by: Andrew Bartlett <abartlet@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
Autobuild-User(master): Andrew Bartlett <abartlet@samba.org>
Autobuild-Date(master): Thu Sep 20 01:29:48 CEST 2018 on sn-devel-144

s3: nmbd: Stop nmbd network announce storm.

Correct fix for. On announce, work->lastannounce_time is set
to current time t, so we must check that 't >= work->lastannounce_time',
not 't > work->lastannounce_time' otherwise we end up not
doing the comparison, and always doing the announce.

Reported by Reuben Farrelly

BUG: https://bugzilla.samba.org/show_bug.cgi?id=13620

Signed-off-by: Andrew Bartlett <abartlet@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
Revviewe-by: Andreas Schneider <asn@samba.org>

PY3: make sure print stmt is enclosed by '(' & ')'

Signed-off-by: Noel Power <noel.power@suse.com>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>

s3/locking: Fix comments to reflect code flow

* posix_pending_close_db is an in-memory database and not a tdb.
  Therefore adjusting comments to convey the correct meaning of the
  database.
* we do not have posix_locking_close_file() any more which got renamed
  to locking_close_file(). Thus fixing comment to mention the new name.

Signed-off-by: Anoop C S <anoopcs@redhat.com>
Reviewed-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
Autobuild-User(master): Jeremy Allison <jra@samba.org>
Autobuild-Date(master): Wed Sep 19 22:23:28 CEST 2018 on sn-devel-144

s3: Don't copy back the stat struct from stat_cache_lookup if we did not get a hit on the full path.

Signed-off-by: Richard Sharpe <realrichardsharpe@gmail.com>
Reviewed-by: Jeremy Allison <jra@samba.org>

uid_wrapper: Fix Tab/indent error with python3

Signed-off-by: Noel Power <noel.power@suse.com>
Reviewed-by: Andreas Schneider <asn@samba.org>
Autobuild-User(master): Noel Power <npower@samba.org>
Autobuild-Date(master): Wed Sep 19 17:49:33 CEST 2018 on sn-devel-144

docs: Add gpfs:check_fstype to vfs_gpfs manpage

Signed-off-by; Christof Schmit <cs@samba.org>
Reviewed-by: Volker Lendecke <vl@samba.org>
Autobuild-User(master): Christof Schmitt <cs@samba.org>
Autobuild-Date(master): Wed Sep 19 03:56:47 CEST 2018 on sn-devel-144

vfs_gpfs: Check for GPFS file system on connect

The vfs_gpfs modules uses GPFS API calls that only succeed when using
the module with the GPFS file system. Add an explicit statfs check for
the file system type on connect, to make it obvious when the file system
is missing or not mounted. The check can be skipped by setting
gpfs:check_fstype to 'no'.

Signed-off-by: Christof Schmitt <cs@samba.org>
Reviewed-by: Volker Lendecke <vl@samba.org>

gitlab-ci: Correct the ctdb tasks

They're swapped.

Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>
Autobuild-User(master): Martin Schwenke <martins@samba.org>
Autobuild-Date(master): Tue Sep 18 10:33:57 CEST 2018 on sn-devel-144

ctdb-recoverd: Set recovery lock handle at start of attempt

This allows the attempt to be cancelled if an election is lost and an
unlock is done before the attempt is completed.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=13617

Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>
Autobuild-User(master): Martin Schwenke <martins@samba.org>
Autobuild-Date(master): Tue Sep 18 02:18:30 CEST 2018 on sn-devel-144

ctdb-recoverd: Handle cancellation when releasing recovery lock

If the recovery lock is in the process of being taken then free the
cluster mutex handle but leave the recovery lock handle in place.
This allows ctdb_recovery_lock() to fail.

Note that this isn't yet live because rec->recovery_lock_handle is
still only set at the completion of the attempt to take the lock.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=13617

Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>

ctdb-recoverd: Return early when the recovery lock is not held

This makes upcoming changes simpler.

Update to modern debug macro while touching relevant line.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=13617

Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>

ctdb-recoverd: Store recovery lock handle

... not just cluster mutex handle.

This makes the recovery lock handle long-lived and with allow the
releasing code to cancel an in-progress attempt to take the recovery
lock.

The cluster mutex handle is now allocated off the recovery lock
handle.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=13617

Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>

ctdb-recoverd: Use talloc() to allocate recovery lock handle

At the moment this is still local and is freed after the mutex is
successfully taken.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=13617

Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>

ctdb-recoverd: Rename hold_reclock_state to ctdb_recovery_lock_handle

This will be a longer lived structure.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=13617

Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>

ctdb-recoverd: Re-check master on failure to take recovery lock

If the master changed while trying to take the lock then fail gracefully.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=13617

Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>

ctdb-recoverd: Clean up taking of recovery lock

No functional changes, just coding style cleanups and debug message
tweaks.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=13617

Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>

ctdb-cluster-mutex: Block signals around fork

If SIGTERM is received and the tevent signal handler setup in the
recovery daemon is still enabled then the signal is handled and a
corresponding event is queued.  The child never runs an event loop so
the signal is effectively ignored.

Resetting the SIGTERM handler isn't enough.  A signal can arrive
before that.

Block SIGTERM before forking and then immediately unblock it in the
parent.

In the child, unblock SIGTERM after the signal handler is reset.  An
explicit unblock is needed because according to sigprocmask(2) "the
signal mask is preserved across execve(2)".

BUG: https://bugzilla.samba.org/show_bug.cgi?id=13617

Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>

ctdb-cluster-mutex: Reset SIGTERM handler in cluster mutex child

If SIGTERM is received and the tevent signal handler setup in the
recovery daemon is still enabled then the signal is handled and a
corresponding event is queued.  The child never runs an event loop so
the signal is effectively ignored.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=13617

Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>

s4/librpc: Fix py2 dependecies leaking into py3 libraries

Caused strange crashes when using python3

Signed-off-by: Noel Power <noel.power@suse.com>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>
Autobuild-User(master): Andrew Bartlett <abartlet@samba.org>
Autobuild-Date(master): Sun Sep 16 09:21:59 CEST 2018 on sn-devel-144

PY3: wrap filter calls with list where list is expected

filter in PY2 returns list in PY3 it returns an iterator

Signed-off-by: Noel Power <noel.power@suse.com>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>

PY3: ensure output of subprocess.Popen is decoded

Signed-off-by: Noel Power <noel.power@suse.com>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>

PY3: detect talloc_utils (for version of python running the build)

It seems all combinations have potential to fail here :/ however for
the moment making the assumption
  *  if python2 driving the build you are building against libpython2 (with the option of having python3 as 'extra-python'
  *  if python3 driving the build you are building against libpython3 (with the option of having python3 as 'extra-python' NOTE: this isn't inforced it probably should be !!!! In a system with python2 and python3 installed it seems waf needs PYTHON env variable to decide which libpython to build against. It's also an option that configure should use that too (to figure out which talloc_util to use or build)

However.... right now I just want the thing to build

Signed-off-by: Noel Power <noel.power@suse.com>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>

PY3: use pyembed_libname for library names (otherwise missing deps)

Signed-off-by: Noel Power <noel.power@suse.com>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>

PY3: xrange->range

Signed-off-by: Noel Power <noel.power@suse.com>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>

PY3: string.upper not in PY3

Signed-off-by: Noel Power <noel.power@suse.com>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>

PY3: decode output of cmd_output for easier string manip

Signed-off-by: Noel Power <noel.power@suse.com>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>

PY3: string.translate string.upper don't exist in PY3

Signed-off-by: Noel Power <noel.power@suse.com>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>

PY3: bytes.maketrans, string.maketrans is a PY2 only function

Signed-off-by: Noel Power <noel.power@suse.com>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>

PY3: enclose filter with list as len on result of filter fails

filter returns an iterator in PY3 (and a list in PY2)

Signed-off-by: Noel Power <noel.power@suse.com>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>

PY3: fix string.rsplit usage (doens't exist in PY3)

Signed-off-by: Noel Power <noel.power@suse.com>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>

py2/py3 enclose map with list

Signed-off-by: Noel Power <noel.power@suse.com>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>

PY3: dict_keys doesn't have a sort method

Signed-off-by: Noel Power <noel.power@suse.com>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>

PY3: dict doesn't have has_key

Signed-off-by: Noel Power <noel.power@suse.com>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>

PY3: iterkeys -> keys

Signed-off-by: Noel Power <noel.power@suse.com>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>

PY3: iterkeys -> keys

Signed-off-by: Noel Power <noel.power@suse.com>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>

PY3: dict has no iterkeys method

Signed-off-by: Noel Power <noel.power@suse.com>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>

PY3: fix iteritems usage, not supported in python3

Signed-off-by: Noel Power <noel.power@suse.com>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>

PY3: fix some octal literals

Signed-off-by: Noel Power <noel.power@suse.com>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>

s4/selftest: enable samba.tests.ntlmdisabled for py3

Signed-off-by: Noel Power <noel.power@suse.com>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>
Autobuild-User(master): Andrew Bartlett <abartlet@samba.org>
Autobuild-Date(master): Sat Sep 15 18:10:42 CEST 2018 on sn-devel-144

selftest/knownfail: Add PY3 entries for samba.tests.ntlmdisabled

Signed-off-by: Noel Power <noel.power@suse.com>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>

s4/selftest: enable samba.tests.netlogonsvc for py3

Signed-off-by: Noel Power <noel.power@suse.com>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>

s4/selftest: enable samba.tests.kcc.kcc_utils for py3

Signed-off-by: Noel Power <noel.power@suse.com>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>

s4/selftest: enable samba.tests.kcc for py3

Signed-off-by: Noel Power <noel.power@suse.com>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>

s4/selftetst: enable samba4.drs.cracknames for py3

Signed-off-by: Noel Power <noel.power@suse.com>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>

s4/selftest: enable samba.tests.join for py3

Signed-off-by: Noel Power <noel.power@suse.com>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>

python/samba/tests: py_gensec_update takes bytes as param

Signed-off-by: Noel Power <noel.power@suse.com>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>