VERSION: Disable GIT_SNAPSHOT for the 4.18.7 release.

Signed-off-by: Jule Anger <janger@samba.org>

WHATSNEW: Add release notes for Samba 4.18.7.

Signed-off-by: Jule Anger <janger@samba.org>

smbd: Fix BZ15481

Bug: https://bugzilla.samba.org/show_bug.cgi?id=15481

Signed-off-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 20 22:42:48 UTC 2023 on atb-devel-224

(cherry picked from commit 3481bbfede5127e3664bcf464a0ae3dec9247ab7)

Autobuild-User(v4-18-test): Jule Anger <janger@samba.org>
Autobuild-Date(v4-18-test): Fri Sep 22 21:07:52 UTC 2023 on atb-devel-224

tests: Add reproducer for BZ15481

Bug: https://bugzilla.samba.org/show_bug.cgi?id=15481

Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
(cherry picked from commit 56df75d44795582dcecb8676a0d80d6f4a46c7e9)

s4:kdc: Add correct Asserted Identity SID in response to an S4U2Self request

I’m not sure exactly how this check was supposed to work. But in any
case, within fast_unwrap_request() the Heimdal KDC replaces the outer
padata with the padata from the inner FAST request. Hence, this check
does not accomplish anything useful: at no point should the KDC plugin
see the outer padata.

A couple of unwanted consequences resulted from this check. One was that
a client who sent empty FX‐FAST padata within the inner FAST request
would receive the *Authentication Authority* Asserted Identity SID
instead of the *Service* Asserted Identity SID. Another consequence was
that a client could in the same manner bypass the restriction on
performing S4U2Self with an RODC‐issued TGT.

Overall, samba_wdc_is_s4u2self_req() is somewhat of a hack. But the
Heimdal plugin API gives us nothing better to work with.

Signed-off-by: Joseph Sutton <josephsutton@catalyst.net.nz>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>
(cherry picked from commit 5c580dbdb3e6a70c8d2f5059e2b7293a7e780414)

BUG: https://bugzilla.samba.org/show_bug.cgi?id=15477

s4:kdc: Avoid copying data if not needed

krb5_pac_add_buffer() makes its own copy of the data we pass in. We
don't need to make yet another copy.

Signed-off-by: Joseph Sutton <josephsutton@catalyst.net.nz>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>
(cherry picked from commit fa901e7346d36ae64a7ceab5dcf76bc210a67c93)

BUG: https://bugzilla.samba.org/show_bug.cgi?id=15476

s4:kdc: Don't pass a NULL pointer into krb5_pac_add_buffer()

Heimdal contains an assertion that the data pointer is not NULL. We need
to pass in a pointer to some dummy data instead.

Signed-off-by: Joseph Sutton <josephsutton@catalyst.net.nz>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>
(cherry picked from commit 47ef49fd91f050ce4a79a8471b3e66c808f48752)

BUG: https://bugzilla.samba.org/show_bug.cgi?id=15476

s3: smbd: Ensure we remove any pending aio values for named pipes on forced shutdown.

Matches file and directory closes.

Remove knownfail.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=15423

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): Wed Sep 20 02:43:18 UTC 2023 on atb-devel-224

(cherry picked from commit 11280f1705c0faa1729f5aeaa1b6a1f79ab5a199)

Autobuild-User(v4-18-test): Jule Anger <janger@samba.org>
Autobuild-Date(v4-18-test): Wed Sep 20 21:38:55 UTC 2023 on atb-devel-224

s3: torture: Add a new SMB2 test: SMB2-PIPE-READ-ASYNC-DISCONNECT

Shows the server crashes if we open a named pipe, do an async read
and then disconnect.

Adds knownfail:

BUG: https://bugzilla.samba.org/show_bug.cgi?id=15423

Signed-off-by: Jeremy Allison <jra@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
(cherry picked from commit 66398dd03c46633b474438dddb771caa2d245e64)

s3: smbd: named pipe writes are async. Use the same logic as for named pipe transacts to avoid crashes on shutdown.

Noticed by Metze.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=15423

Signed-off-by: Jeremy Allison <jra@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
(cherry picked from commit ea062c3b0d4dbb1f0682f808ac893bf36a6fb194)

s3: smbd: named pipe reads are async. Use the same logic as for named pipe transacts to avoid crashes on shutdown.

Noticed by Metze.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=15423

Signed-off-by: Jeremy Allison <jra@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
(cherry picked from commit 3f32bf887d4425655e81da0b2234cbca3b1d56e6)

s3: smbd: Add some DEVELOPER-only code to panic if the destructor for an aio_lnk is called and the associated fsp doesn't exist.

Make this DEVELOPER-only as it walks the entire open
file list on every file close (with associated aio).

This helps catch really subtle problems with orphaned
aio lnk structs.

Reproducer test case to follow.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=15423

Signed-off-by: Jeremy Allison <jra@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
(cherry picked from commit 82e88f70f181300f6f98691f6680839a94470e13)

nsswitch/wb_common.c: fix socket fd and memory leaks of global state

When we are called in wb_atfork_child() or winbind_destructor(),
wb_thread_ctx_destructor() is not called for the global state
of the current nor any other thread, which means we would
leak the related memory and socket fds.

Now we maintain a global list protected by a global mutex.
We traverse the list and close all socket fds, which are no
longer used (winbind_destructor) or no longer valid in the
current process (wb_atfork_child), in addition we 'autofree'
the ones, which are only visible internally as global (per thread)
context.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=15464

Tested-by: Krzysztof Piotr Oledzki <ole@ans.pl>
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
Autobuild-User(master): Stefan Metzmacher <metze@samba.org>
Autobuild-Date(master): Thu Sep 14 18:53:07 UTC 2023 on atb-devel-224

(cherry picked from commit 4af3faace481d23869b64485b791bdd43d8972c5)

Autobuild-User(v4-18-test): Jule Anger <janger@samba.org>
Autobuild-Date(v4-18-test): Mon Sep 18 17:25:43 UTC 2023 on atb-devel-224

nsswitch/wb_common.c: don't operate on a stale wb_global_ctx.key

If nss_winbind is loaded into a process that uses fork multiple times
without any further calls into nss_winbind, wb_atfork_child handler
was using a wb_global_ctx.key that was no longer registered in the
pthread library, so we operated on a slot that was potentially
reused by other libraries or the main application. Which is likely
to cause memory corruption.

So we better don't call pthread_key_delete() in wb_atfork_child().

BUG: https://bugzilla.samba.org/show_bug.cgi?id=15464

Reported-by: Krzysztof Piotr Oledzki <ole@ans.pl>
Tested-by: Krzysztof Piotr Oledzki <ole@ans.pl>
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
(cherry picked from commit 91b30a7261e6455d3a4f31728c23e4849e3945b9)

nsswitch/wb_common.c: winbind_destructor can always use get_wb_global_ctx()

The HAVE_PTHREAD logic inside of get_wb_global_ctx() will do all
required magic.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=15464

Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
(cherry picked from commit 836823e5047d0eb18e66707386ba03b812adfaf8)

nsswitch/wb_common.c: fix build without HAVE_PTHREAD

BUG: https://bugzilla.samba.org/show_bug.cgi?id=15464

Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
(cherry picked from commit 4faf806412c4408db25448b1f67c09359ec2f81f)

nsswitch: add test for pthread_key_delete missuse (bug 15464)

This is based on https://bugzilla.samba.org/attachment.cgi?id=18081
written by Krzysztof Piotr Oledzki <ole@ans.pl>

BUG: https://bugzilla.samba.org/show_bug.cgi?id=15464

Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
(cherry picked from commit 62af25d44e542548d8cdecb061a6001e0071ee76)

libsmb: Fix test for smbc_getxattr

Bug: https://bugzilla.samba.org/show_bug.cgi?id=14808
Signed-off-by: Remi Collet <rcollet@redhat.com>
Reviewed-by: Jeremy Allison <jra@samba.org>
Reviewed-by: Volker Lendecke <vl@samba.org>
Autobuild-User(master): Volker Lendecke <vl@samba.org>
Autobuild-Date(master): Sun Apr  9 10:44:38 UTC 2023 on atb-devel-224

(cherry picked from commit 0cd66fe6bd4ac2aa0b302ddf3eb5068fc9c522ec)

Autobuild-User(v4-18-test): Jule Anger <janger@samba.org>
Autobuild-Date(v4-18-test): Wed Sep  6 09:28:02 UTC 2023 on atb-devel-224

libsmb: fix regression on smbc_getxattr and fix doc

Bug: https://bugzilla.samba.org/show_bug.cgi?id=14808
Signed-off-by: Remi Collet <rcollet@redhat.com>
Reviewed-by: Jeremy Allison <jra@samba.org>
Reviewed-by: Volker Lendecke <vl@samba.org>
(cherry picked from commit 4fc166628fda160d1cd38c140a9664defc5844ab)

mdssvc: better support for search with mdfind from Macs

When searching Samba via Spotlight from a Mac with mdfind, only 50 results are
returned.

It seems the changes for bug #15342 where one step in the right
direction. There, a status indicator meaning "search is still being processed"
was implemented, returning a special status indicator (0x23) in a response, when
the query was still running in the backend, eg Elasticsearch, and we haven't got
any result when when the Mac already comes along asking for results of a query.

Turns out, we should also return 0x23, ie "search is still being processed" when
we have some initial search results from the backend. Otherwise mdfind will stop
querying for more results. It works in Finder, as the Finder by default employs
a "live" search where it just keeps polling for more results even after the
server returned an empty result set. The Finder just keeps on querying in some
interval, typically 4 seconds, and a Mac server Spotlight server might return
new results if new files where created that match the query, hence "live"
search.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=15463

Signed-off-by: Ralph Boehme <slow@samba.org>
Reviewed-by: Noel Power <noel.power@suse.com>
Autobuild-User(master): Ralph Böhme <slow@samba.org>
Autobuild-Date(master): Wed Aug 30 10:51:40 UTC 2023 on atb-devel-224

(cherry picked from commit 3afa27a01ca0b81b0044d5d543a999936582e67a)

Autobuild-User(v4-18-test): Jule Anger <janger@samba.org>
Autobuild-Date(v4-18-test): Tue Sep  5 09:28:33 UTC 2023 on atb-devel-224

vfs_aio_pthread: use SMB_VFS_NEXT_OPENAT() in aio_pthread_openat_fn()

1. Set 'aio_allow_open' to false if fsp->fsp_flags.is_pathref
2. Move !(how->flags & O_CREAT) and !(how->flags & O_EXCL) up and set 'aio_allow_open' to false
3. Use SMB_VFS_NEXT_OPENAT() instead of openat() for disable async opens case.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=15453

Signed-off-by: MikeLiu <mikeliu@qnap.com>
Reviewed-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
Autobuild-User(master): Jeremy Allison <jra@samba.org>
Autobuild-Date(master): Tue Aug 22 17:44:00 UTC 2023 on atb-devel-224

Autobuild-User(v4-18-test): Jule Anger <janger@samba.org>
Autobuild-Date(v4-18-test): Mon Sep  4 09:49:30 UTC 2023 on atb-devel-224

s3: smbd: Sanitize any "server" and "share" components of SMB1 DFS paths to remove UNIX separators.

Remove knownfail.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=15419

Signed-off-by: Jeremy Allison <jra@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
Autobuild-User(master): Ralph Böhme <slow@samba.org>
Autobuild-Date(master): Thu Jul 27 10:52:50 UTC 2023 on atb-devel-224

(cherry picked from commit 20df26b908182f0455f301a51aeb54b6044af580)

Autobuild-User(v4-18-test): Jule Anger <janger@samba.org>
Autobuild-Date(v4-18-test): Thu Aug 31 09:38:21 UTC 2023 on atb-devel-224

s3: torture: Add test to show an SMB1 DFS path of "\\x//\\/" crashes smbd.

Adds knownfail.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=15419

Signed-off-by: Jeremy Allison <jra@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
(cherry picked from commit 2aa9ffa2f0fc79599efbfe0c37aac4ef5160f712)

ctdb-common: Set immediate mode for pcap capture

Fix a problem where ctdb_killtcp (almost always) fails to capture
packets with --enable-pcap and libpcap ≥ 1.9.1.  The problem is due to
a gradual change in libpcap semantics when using
pcap_get_selectable_fd(3PCAP) to get a file descriptor and then using
that file descriptor in non-blocking mode.

pcap_set_immediate_mode(3PCAP) says:

  pcap_set_immediate_mode() sets whether immediate mode should be set
  on a capture handle when the handle is activated.  In immediate
  mode, packets are always delivered as soon as they arrive, with no
  buffering.

and

  On Linux, with previous releases of libpcap, capture devices are
  always in immediate mode; however, in 1.5.0 and later, they are, by
  default, not in immediate mode, so if pcap_set_immediate_mode() is
  available, it should be used.

However, it wasn't until libpcap commit
2ade7676101366983bd4f86bc039ffd25da8c126 (before libpcap 1.9.1) that
it became a requirement to use pcap_set_immediate_mode(), even with a
timeout of 0.

More explanation in this libpcap issue comment:

  https://github.com/the-tcpdump-group/libpcap/issues/860#issuecomment-541204548

Do a configure check for pcap_set_immediate_mode() even though it has
existed for 10 years.  It is easy enough.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=15451

Signed-off-by: Martin Schwenke <mschwenke@ddn.com>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>
Autobuild-User(master): Amitay Isaacs <amitay@samba.org>
Autobuild-Date(master): Tue Aug 15 10:53:52 UTC 2023 on atb-devel-224

(cherry picked from commit dc7b48c404337891b5105df4d6751cf549a533eb)

Autobuild-User(v4-18-test): Jule Anger <janger@samba.org>
Autobuild-Date(v4-18-test): Tue Aug 29 12:27:35 UTC 2023 on atb-devel-224

ctdb-common: Replace pcap_open_live() by lower level calls

A subsequent commit will insert an additional call before
pcap_activate().

This sequence of calls is taken from the source for pcap_open_live(),
so there should be no change in behaviour.

Given the defaults set by pcap_create_common(), it would be possible
to omit the calls to pcap_set_promisc() and pcap_set_timeout().
However, those defaults don't seem to be well documented, so continue
to explicitly set everything that was set before.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=15451

Signed-off-by: Martin Schwenke <mschwenke@ddn.com>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>
(cherry picked from commit ffc2ae616d8fab7528fbdfd8c6b94c5b9a0e3a7c)

ctdb-common: Improve error handling

Factor out a failure label, which will get more use in subsequent
commits, and only set private_data when success is certain.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=15451

Signed-off-by: Martin Schwenke <mschwenke@ddn.com>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>
(cherry picked from commit d87041d8968e91db9d257445321b85693303f95e)

s4-rpc_server/drsupai: Avoid looping with Azure AD Connect by not incrementing temp_highest_usn for the NC root

We send the NC root first, as a special case for every chunk
that we send until the natural point where it belongs.

We do not bump the tmp_highest_usn in the highwatermark that
the client and server use (it is meant to be an opauqe cookie)
until the 'natural' point where the object appears, similar
to the cache for GET_ANC.

The issue is that without this, because the NC root was sorted
first in whatever chunk it appeared in but could have a 'high'
highwatermark, Azure AD Connect will send back the same
new_highwatermark->tmp_highest_usn, and due to a bug,
a zero reserved_usn, which makes Samba discard it.

The reserved_usn is now much less likely to ever be set because
the tmp_higest_usn is now always advancing.

RN: Avoid infinite loop in initial user sync with Azure AD Connect
 when synchronising a large Samba AD domain.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=15401

Signed-off-by: Andrew Bartlett <abartlet@samba.org>
Reviewed-by: Stefan Metzmacher <metze@samba.org>
(cherry picked from commit 79ca6ef28a6f94965cb030c4a7da8c1b9db7150b)

Autobuild-User(v4-18-test): Jule Anger <janger@samba.org>
Autobuild-Date(v4-18-test): Mon Aug 21 09:12:14 UTC 2023 on atb-devel-224

s4-rpc_server/drsuapi: Ensure logs show DN for replicated objects, not (null)

BUG: https://bugzilla.samba.org/show_bug.cgi?id=15407
BUG: https://bugzilla.samba.org/show_bug.cgi?id=15401

Signed-off-by: Andrew Bartlett <abartlet@samba.org>
Reviewed-by: Stefan Metzmacher <metze@samba.org>
(cherry picked from commit 17359afa627a3086ec8d6862f007a3479574a8b4)

s4-rpc_server/drsuapi: Update getnc_state to be != NULL

This is closer to our READDME.Coding style

BUG: https://bugzilla.samba.org/show_bug.cgi?id=15401

Signed-off-by: Andrew Bartlett <abartlet@samba.org>
Reviewed-by: Stefan Metzmacher <metze@samba.org>
(cherry picked from commit 2aba9e230ea62efcbd829f6f073894dfa3180c91)

s4-rpc_server/drsuapi: Rename ncRoot -> untrusted_ncRoot to avoid misuse

Because of the requirement to echo back the original string, we can
not force this to be a trustworthy value.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=15401

Signed-off-by: Andrew Bartlett <abartlet@samba.org>
Reviewed-by: Stefan Metzmacher <metze@samba.org>
(cherry picked from commit 2ed9815eeacfcf3a58871bafe0212398cc34c39e)

s4-rpc_server/drsuapi: Avoid modification to ncRoot input variable in GetNCChanges

This tries to avoid it appearing that ncRoot is a value that can
be trusted and used internally by not updating it and instead leaving
it just as an input/echo-back value.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=15401

Signed-off-by: Andrew Bartlett <abartlet@samba.org>
Reviewed-by: Stefan Metzmacher <metze@samba.org>
(cherry picked from commit 548f141f11e89d335d8f9d74ab6925fa6b90fb84)

s4-rpc_server/drsuapi: Fix indentation in GetNCChanges()

This avoids the indentation correction being in the previous patch.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=15401

Signed-off-by: Andrew Bartlett <abartlet@samba.org>
Reviewed-by: Stefan Metzmacher <metze@samba.org>
(cherry picked from commit fe7418e1765b79f60945b787536b4d84a548fe02)

s4-rpc_server/drsuapi: Only keep and invalidate replication cycle state for normal replication

This changes the GetNCChanges server to use a per-call state for
extended operations like RID_ALLOC or REPL_OBJ and only maintain
and (more importantly) invalidate the state during normal replication.

This allows REPL_OBJ to be called during a normal replication cycle
that continues using after that call, continuing with the same
highwatermark cookie.

Azure AD will do a sequence of (roughly)

* Normal replication (objects 1..100)
* REPL_OBJ (of 1 object)
* Normal replication (objects 101..200)

However, if there are more than 100 (in this example) objects in the
domain, and the second replication is required, the objects 1..100
are sent, as the replication state was invalidated by the REPL_OBJ call.

RN: Improve GetNChanges to address some (but not all "Azure AD Connect")
syncronisation tool looping during the initial user sync phase.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=15401

Signed-off-by: Andrew Bartlett <abartlet@samba.org>
Reviewed-by: Stefan Metzmacher <metze@samba.org>
(cherry picked from commit 99579e706312192f46df33d55949db7f1475d0d0)

s4-torture/drs: Add test showing that if present in the set the NC root leads and tmp_highest_usn moves

The NC root, on any replication when it appears, is the first object to be
replicated, including for all subsequent chunks in the replication.

However the tmp_highest_usn is not updated by that USN, it must
only be updated for the non-NC changes (to match Windows exactly),
or at least only updated with the non-NC changes until it would
naturally appear.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=15401

Signed-off-by: Andrew Bartlett <abartlet@samba.org>
Reviewed-by: Stefan Metzmacher <metze@samba.org>
(cherry picked from commit 87414955212143b8502b4c02aca150bc72cb8de5)

s4-torture/drs: Add test demonstrating that a GetNCChanges REPL_OBJ will not reset the replication cookie

This demonstrates the behaviour used by the "Azure AD Connect" cloud sync tool.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=15401

Signed-off-by: Andrew Bartlett <abartlet@samba.org>
Reviewed-by: Stefan Metzmacher <metze@samba.org>
(cherry picked from commit b323169d6ff8357f7c999ae346137166c98218ac)

s4-torture/drs: Add a test matching Azure AD Connect REPL_OBJ behaviour

Azure AD Connect will send a GUID but no DummyDN.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=15401

Signed-off-by: Andrew Bartlett <abartlet@samba.org>
Reviewed-by: Stefan Metzmacher <metze@samba.org>
(cherry picked from commit db16366b0bbefcdb91a0b36c903ed63456a081b8)

s4-torture/drs: Use addCleanup() in getchanges.py for OU handling

BUG: https://bugzilla.samba.org/show_bug.cgi?id=15401

Signed-off-by: Andrew Bartlett <abartlet@samba.org>
Reviewed-by: Stefan Metzmacher <metze@samba.org>
(cherry picked from commit 40f831e67e1f312b1db52c74c119899245d03e32)

s4-torture/drs: Create temp OU with a unique name per test

It is always better to keep the testing OUs unique if possible.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=15401

Signed-off-by: Andrew Bartlett <abartlet@samba.org>
Reviewed-by: Stefan Metzmacher <metze@samba.org>
(cherry picked from commit 628eab11b3c2e82875bf602e363b781d3e5eb96d)

s4-torture/drs: Save the server dnsname on the DcConnection object

This object is used to hold one of many possible connections and
it is helpful for debugging and uniqueness to know which DC is being
connected to.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=15401

Signed-off-by: Andrew Bartlett <abartlet@samba.org>
Reviewed-by: Stefan Metzmacher <metze@samba.org>
(cherry picked from commit c30bb8769ff2c4eba2d8f8a2bd3a56946b7d9d5e)

s4-rpc_server/drsuapi: Remove rudundant check for valid and non-NULL ncRoot_dn

This check was valuable before aee2039e63ceeb5e69a0461fb77e0f18278e4dc4
but now only checks things we know to be true, as the value has come
from Samba via drs_ObjectIdentifier_to_dn_and_nc_root() either on this
or a previous call.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=15401

Signed-off-by: Andrew Bartlett <abartlet@samba.org>
Reviewed-by: Stefan Metzmacher <metze@samba.org>
(cherry picked from commit 0550e469eda4022659718ae9a56f5deaa9f9a307)

s4-dsdb: Improve logging for drs_ObjectIdentifier_to_dn_and_nc_root()

At this layer we can make a reasonable assumption about being able
to read ldb_errstring() to print that for extra useful debugging.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=15401

Signed-off-by: Andrew Bartlett <abartlet@samba.org>
Reviewed-by: Stefan Metzmacher <metze@samba.org>
(cherry picked from commit 63843a22c8db73d459bee61e73bb1f0d31e3d427)

s4-rpc_server/drsuapi: Improve debug message for drs_ObjectIdentifier_to_dn_and_nc_root() failure

BUG: https://bugzilla.samba.org/show_bug.cgi?id=15401

Signed-off-by: Andrew Bartlett <abartlet@samba.org>
Reviewed-by: Stefan Metzmacher <metze@samba.org>
(cherry picked from commit a12bcce89d26ae05bbaeed560cf8fcc7b5bcfdab)

s4-rpc_server/drsuapi: Improve debugging of invalid DNs

This is still unreachable, so but improve the logging
to give more detail in this area anyway.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=15401

Signed-off-by: Andrew Bartlett <abartlet@samba.org>
Reviewed-by: Stefan Metzmacher <metze@samba.org>
(cherry picked from commit d0c1ce53add2fd3b3a4186581f4e214029cbcf1a)

s4-rpc_server/drsuapi: Add tmp_highest_usn tracking to replication log

BUG: https://bugzilla.samba.org/show_bug.cgi?id=15401

Signed-off-by: Andrew Bartlett <abartlet@samba.org>
Reviewed-by: Stefan Metzmacher <metze@samba.org>
(cherry picked from commit 0d9ea6c559317e19642662220c089e2d59ef3ecd)

s3: smbd: Ensure init_smb1_request() zeros out what the incoming pointer points to.

Remove the now unneeded req->xxx = NULL assignments (and the
deliberately bogus req->session = (void *)0xDEADBEEF one
used to demonstrate the bug).

Remove knownfail.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=15432

Signed-off-by: Jeremy Allison <jra@samba.org>
Reviewed-by: Noel Power <npower@samba.org>
Autobuild-User(master): Noel Power <npower@samba.org>
Autobuild-Date(master): Tue Aug 15 12:06:36 UTC 2023 on atb-devel-224

(Back-ported from commit 4145bfb1b5a3639caf26a310d612aec29fc00117)

Autobuild-User(v4-18-test): Jule Anger <janger@samba.org>
Autobuild-Date(v4-18-test): Fri Aug 18 14:07:39 UTC 2023 on atb-devel-224

s3: torture: Add SMB1-NEGOTIATE-TCON that shows the SMB1 server crashes on the uninitialized req->session.

Found by Robert Morris <rtm@lcs.mit.edu>.

Adds knownfail.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=15432

Signed-off-by: Jeremy Allison <jra@samba.org>
Reviewed-by: Noel Power <nopower@samba.org>
(Back-ported from commit c32df3bb31ce6275cfb91107e34e2d6b3c2fba1b)

s3: smbd: init_smb1_request() isn't being passed zero'ed memory from any codepath.

If a client does a SMB1 NEGPROT followed by SMB1 TCON
then req->session is left uninitialized.

Show this causes a crash by deliberately initializing
req->session to an invalid pointer. This will be removed
once the test shows the crash, and the fix is added to
cause init_smb1_request() to zero the memory passed in.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=15432

Signed-off-by: Jeremy Allison <jra@samba.org>
Reviewed-by: Noel Power <npower@samba.org>
(Back-ported from commit f02f74e931f5821c7b7c1be2b8f0fb60c9a69b19)

VERSION: Bump version up to Samba 4.18.7...

and re-enable GIT_SNAPSHOT.

Signed-off-by: Jule Anger <janger@samba.org>

VERSION: Disable GIT_SNAPSHOT for the 4.18.6 release.

Signed-off-by: Jule Anger <janger@samba.org>

WHATSNEW: Add release notes for Samba 4.18.6.

Signed-off-by: Jule Anger <janger@samba.org>

s3: smbd: Add missing 'return;'s in exit paths in reply_exit_done().

Remove knownfail.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=15430

Signed-off-by: Jeremy Allison <jra@samba.org>
Reviewed-by: Noel Power <npower@samba.org>
Autobuild-User(master): Noel Power <npower@samba.org>
Autobuild-Date(master): Mon Aug 14 19:52:49 UTC 2023 on atb-devel-224

(cherry picked from commit d79d0508a4b8bdc4582a350d109181ecae0bf1e2)

Autobuild-User(v4-18-test): Jule Anger <janger@samba.org>
Autobuild-Date(v4-18-test): Wed Aug 16 11:49:39 UTC 2023 on atb-devel-224

s3: torture: Add a test doing an SMB1 negotiate+exit.

Robert Morris <rtm@lcs.mit.edu> noticed a missing
return in reply_exit_done().

Adds knownfail.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=15430

Signed-off-by: Jeremy Allison <jra@samba.org>
Reviewed-by: Noel Power <npower@samba.org>
(cherry picked from commit 63895e03c4e8ed79a3b2cda928f58ec278cd6608)

s3: smbd: Ensure all callers to srvstr_pull_req_talloc() pass a zeroed-out dest pointer.

Now we've fixed srvstr_pull_req_talloc() this isn't
strictly needed, but ensuring pointers are initialized
is best practice to avoid future bugs.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=15420

Signed-off-by: Jeremy Allison <jra@samba.org>
Reviewed-by: Volker Lendecke <vl@samba.org>
Autobuild-User(master): Volker Lendecke <vl@samba.org>
Autobuild-Date(master): Mon Aug 14 15:55:43 UTC 2023 on atb-devel-224

(cherry picked from commit 5379b8d557a9a16b81eafb87b60b81debc4bfccb)

s3: smbd: Uncorrupt the pointer we were using to prove a crash.

Rather than restore to uninitialized, set to NULL as per
modern coding practices.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=15420
Reviewed-by: Volker Lendecke <vl@samba.org>
Signed-off-by: Jeremy Allison <jra@samba.org>
Reviewed-by: Volker Lendecke <vl@samba.org>
(cherry picked from commit 5bc50d2ea4444244721e72b4264311c7005d2f3c)

s3: smbd: Ensure srvstr_pull_req_talloc() always NULLs out *dest.

Robert Morris <rtm@lcs.mit.edu> noticed that in the case
where srvstr_pull_req_talloc() is being called with
buffer remaining == 0, we don't NULL out the destination
pointed which is *always* done in the codepaths inside
pull_string_talloc(). This prevents a crash in the caller.

Remove knownfail.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=15420

Signed-off-by: Jeremy Allison <jra@samba.org>
Reviewed-by: Volker Lendecke <vl@samba.org>
(cherry picked from commit 9220c45cc191b34e293190f6a923ba463edd5db9)

s3: torture: Add SMB1-TRUNCATED-SESSSETUP test.

Shows that we indirect through an uninitialized pointer and the client crashes
it's own smbd.

Add knownfail.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=15420

Signed-off-by: Jeremy Allison <jra@samba.org>
Reviewed-by: Volker Lendecke <vl@samba.org>
(cherry picked from commit 963fd8aa9b76361ab9aeb63307773f2498b17879)

s3: smbd: Deliberately currupt an uninitialized pointer.

We will need this to show smbd crashing in the test code.
This will be removed once we're passing the test.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=15420

Signed-off-by: Jeremy Allison <jra@samba.org>
Reviewed-by: Volker Lendecke <vl@samba.org>
(cherry picked from commit e7bf94b4e3a7f994aa6f0b859089c5add2ad380f)

mdssvc: Do an early talloc_free() in _mdssvc_open()

Environment setup:
When macOS Finder connect to a samba server with 'spotlight = yes',
macOS would issue mdssvc open (mdssvc.opnum == 0) to samba and it goes
through api _mdssvc_open().

After applied 578e434a94147dc2d7dbfc006d2ab84807859c1d,
(this is reported by jaywei@qnap.com)
this line 'talloc_free(path);' is deleted if _mdssvc_open() normal exit,
so memory is lazy de-allocate: delayed to
smbd_tevent_trace_callback() @ smb2_process.c. [1]

Supposed to explicitly free 'path' in _mdssvc_open() @ srv_mdssvc_nt.c[2]
just like abnormal exit, do not wait for main loop to free 'path' which is
no longer used, this is more consistent while reading source code.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=15449

[1] gdb tracing 'path' address 0x56204ccc67e0 to know how it is freed.
Breakpoint 2, _tc_free_children_internal (tc=0x56204ccc6780, ptr=0x56204ccc67e0, location=0x7ff430d96410 "../../lib/talloc/talloc.c:1714") at ../../lib/talloc/talloc.c:1656
1656            while (tc->child) {
(gdb) bt
0  _tc_free_children_internal (tc=0x56204ccc6780, ptr=0x56204ccc67e0, location=0x7ff430d96410 "../../lib/talloc/talloc.c:1714") at ../../lib/talloc/talloc.c:1656
1  0x00007ff430d92b14 in _tc_free_internal (tc=0x56204ccc6780, location=0x7ff430d96410 "../../lib/talloc/talloc.c:1714") at ../../lib/talloc/talloc.c:1183
2  0x00007ff430d93b71 in _tc_free_children_internal (tc=0x56204ccc6720, ptr=0x56204ccc6780, location=0x7ff430d96410 "../../lib/talloc/talloc.c:1714") at ../../lib/talloc/talloc.c:1668
3  0x00007ff430d93d66 in talloc_free_children (ptr=0x56204ccc6780) at ../../lib/talloc/talloc.c:1714
4  0x00007ff432235aca in talloc_pop (frame=0x56204ccc6780) at ../../lib/util/talloc_stack.c:125
5  0x00007ff430d92959 in _tc_free_internal (tc=0x56204ccc6720, location=0x7ff431f358d0 "../../source3/smbd/process.c:3726") at ../../lib/talloc/talloc.c:1157
6  0x00007ff430d92cd5 in _talloc_free_internal (ptr=0x56204ccc6780, location=0x7ff431f358d0 "../../source3/smbd/process.c:3726") at ../../lib/talloc/talloc.c:1247
7  0x00007ff430d93f96 in _talloc_free (ptr=0x56204ccc6780, location=0x7ff431f358d0 "../../source3/smbd/process.c:3726") at ../../lib/talloc/talloc.c:1791
8  0x00007ff431d81292 in smbd_tevent_trace_callback (point=TEVENT_TRACE_AFTER_LOOP_ONCE, private_data=0x7ffe46591e30) at ../../source3/smbd/process.c:3726
<...cut...>

[2] gdb tracing 'path' address 0x55a6d66deed0 to know how it is freed.
Breakpoint 2, _tc_free_children_internal (tc=0x55a6d66deed0, ptr=0x55a6d66def30, location=0x7fc4cca84040 "../../source3/rpc_server/mdssvc/srv_mdssvc_nt.c:189") at ../../lib/talloc/talloc.c:1656
1656            while (tc->child) {
(gdb) bt
0  _tc_free_children_internal (tc=0x55a6d66deed0, ptr=0x55a6d66def30, location=0x7fc4cca84040 "../../source3/rpc_server/mdssvc/srv_mdssvc_nt.c:189") at ../../lib/talloc/talloc.c:1656
1  0x00007fc4cb892b14 in _tc_free_internal (tc=0x55a6d66deed0, location=0x7fc4cca84040 "../../source3/rpc_server/mdssvc/srv_mdssvc_nt.c:189") at ../../lib/talloc/talloc.c:1183
2  0x00007fc4cb892cd5 in _talloc_free_internal (ptr=0x55a6d66def30, location=0x7fc4cca84040 "../../source3/rpc_server/mdssvc/srv_mdssvc_nt.c:189") at ../../lib/talloc/talloc.c:1247
3  0x00007fc4cb893f96 in _talloc_free (ptr=0x55a6d66def30, location=0x7fc4cca84040 "../../source3/rpc_server/mdssvc/srv_mdssvc_nt.c:189") at ../../lib/talloc/talloc.c:1791
4  0x00007fc4cc9396e4 in _mdssvc_open (p=0x55a6d66d5600, r=0x55a6d66edc60) at ../../source3/rpc_server/mdssvc/srv_mdssvc_nt.c:189
<...cut...>

Signed-off-by: Jones Syue <jonessyue@qnap.com>
Reviewed-by: Noel Power <npower@samba.org>
Reviewed-by: Volker Lendecke <vl@samba.org>
Autobuild-User(master): Noel Power <npower@samba.org>
Autobuild-Date(master): Mon Aug 14 18:11:37 UTC 2023 on atb-devel-224

(cherry picked from commit 044cb8f9d558bfcd7658cae0f05ff36330538748)

s3:smbd: fix multichannel connection passing race

If a client opens multiple connection with the same
client guid in parallel, our connection passing is likely
to hit a race.

Assume we have 3 processes:

smbdA: This process already handles all connections for
       a given client guid
smbdB: This just received a new connection with an
       SMB2 neprot for the same client guid
smbdC: This also received a new connection with an
       SMB2 neprot for the same client guid

Now both smbdB and smbdC send a MSG_SMBXSRV_CONNECTION_PASS
message to smbdA. These messages contain the socket fd
for each connection.

While waiting for a MSG_SMBXSRV_CONNECTION_PASSED message
from smbdA, both smbdB and smbdC watch the smbXcli_client.tdb
record for changes (that also verifies smbdA stays alive).

Once one of them say smbdB received the MSG_SMBXSRV_CONNECTION_PASSED
message, the dbwrap_watch logic will wakeup smbdC in order to
let it recheck the smbXcli_client.tdb record in order to
handle the case where smbdA died or deleted its record.

Now smbdC rechecks the smbXcli_client.tdb record, but it
was not woken because of a problem with smbdA. It meant
that smbdC sends a MSG_SMBXSRV_CONNECTION_PASS message
including the socket fd again.

As a result smbdA got the socket fd from smbdC twice (or even more),
and creates two (or more) smbXsrv_connection structures for the
same low level tcp connection. And it also sends more than one
SMB2 negprot response. Depending on the tevent logic, it will
use different smbXsrv_connection structures to process incoming
requests. And this will almost immediately result in errors.

The typicall error is:
 smb2_validate_sequence_number: smb2_validate_sequence_number: bad message_id 2 (sequence id 2) (granted = 1, low = 1, range = 1)

But other errors would also be possible.

The detail that leads to the long delays on the client side is
that our smbd_server_connection_terminate_ex() code will close
only the fd of a single smbXsrv_connection, but the refcount
on the socket fd in the kernel is still not 0, so the tcp
connection is still alive...

Now we remember the server_id of the process that we send
the MSG_SMBXSRV_CONNECTION_PASS message to. And just keep
watching the smbXcli_client.tdb record if the server_id
don't change. As we just need more patience to wait for
the MSG_SMBXSRV_CONNECTION_PASSED message.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=15346

Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Andreas Schneider <asn@samba.org>
Autobuild-User(master): Andreas Schneider <asn@cryptomilk.org>
Autobuild-Date(master): Tue Aug  8 13:59:58 UTC 2023 on atb-devel-224

(cherry picked from commit f348b84fbcf203ab1ba92840cf7aecd55dbf9aa0)

Autobuild-User(v4-18-test): Jule Anger <janger@samba.org>
Autobuild-Date(v4-18-test): Fri Aug 11 09:49:53 UTC 2023 on atb-devel-224

s3:smbd: always clear filter_subreq in smb2srv_client_mc_negprot_next()

Commit 5d66d5b84f87267243dcd5223210906ce589af91 introduced a
'verify_again:' target, if we ever hit that, we would leak
the existing filter_subreq.

Moving it just above a possible messaging_filtered_read_send()
will allow us to only clear it if we actually create a new
request. That will help us in the next commits.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=15346

Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Andreas Schneider <asn@samba.org>
(cherry picked from commit 50d61e5300250922bf36bb699306f82dff6a00b9)

s4:torture/smb2: add smb2.multichannel.bugs.bug_15346

This demonstrates the race quite easily against
Samba and works fine against Windows Server 2022.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=15346

Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Andreas Schneider <asn@samba.org>
(cherry picked from commit 4028d6582907cf582730ceec56872d8584ad02e6)

s4:torture/smb2: make it possible to pass existing_conn to smb2_connect_ext()

BUG: https://bugzilla.samba.org/show_bug.cgi?id=15346

Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Andreas Schneider <asn@samba.org>
(cherry picked from commit 2b93058be3f6e5eaee239ad3b0e707c62089d18e)

s4:torture/smb2: let us have a common torture_smb2_con_share()

BUG: https://bugzilla.samba.org/show_bug.cgi?id=15346

Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Andreas Schneider <asn@samba.org>
(cherry picked from commit dc5a500f0a76720b2a5cb5b1142cf4c35cb6bdea)

s4:torture/smb2: let torture_smb2_con_sopt() use smb2_connect()

There's no need for smb2_connect_ext().

BUG: https://bugzilla.samba.org/show_bug.cgi?id=15346

Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Andreas Schneider <asn@samba.org>
(cherry picked from commit ade663ee6ca1a2813b203ea667d933f4dab9e7b7)

dcerpc.idl: fix definitions for DCERPC_PKT_CO_CANCEL and DCERPC_PKT_ORPHANED payload

It seems commit 259129e8f4bc8cacd1850eba3f6551134835d079 was partly just
fantasy...

Windows clients just use 16 bytes for DCERPC_PKT_CO_CANCEL and
DCERPC_PKT_ORPHANED pdus.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=15446

Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Volker Lendecke <vl@samba.org>
Autobuild-User(master): Volker Lendecke <vl@samba.org>
Autobuild-Date(master): Tue Aug  8 08:57:46 UTC 2023 on atb-devel-224

(cherry picked from commit 9ec22e680249cfde06fb1a0a34fcc94d1f47002d)

Autobuild-User(v4-18-test): Jule Anger <janger@samba.org>
Autobuild-Date(v4-18-test): Tue Aug  8 12:56:30 UTC 2023 on atb-devel-224

librpc/rpc: let dcerpc_read_ncacn_packet_next_vector() handle fragments without any payload

DCERPC_PKT_CO_CANCEL and DCERPC_PKT_ORPHANED don't have any payload by
default. In order to receive them via dcerpc_read_ncacn_packet_send/recv
we need to allow fragments with frag_len == DCERPC_NCACN_PAYLOAD_OFFSET.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=15446

Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Volker Lendecke <vl@samba.org>
(cherry picked from commit 5c724a3e156ae734e4d187bf9639d895bb011834)

s4:torture/ndr: add tests for DCERPC_PKT_CO_CANCEL and DCERPC_PKT_ORPHANED

The PDUs were generated by Windows clients.

And we fail to parse them currently.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=15446

Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Volker Lendecke <vl@samba.org>
(cherry picked from commit c37adb762640b7df9731d6a60edce808aa8787f8)

vfs_aio_pthread: fix segfault if samba-tool ntacl get

If configured as AD DC and aio_pthread appended into 'vfs objects'[1],
run these commands would get segfault:
1. sudo samba-tool ntacl get .
2. sudo net vfs getntacl sysvol .
gdb said it goes through aio_pthread_openat_fn() @ vfs_aio_pthread.c[2],
and the fsp->conn->sconn->client is null (0x0).

'sconn->client' memory is allocated when a new connection is accpeted:
smbd_accept_connection > smbd_process > smbXsrv_client_create
While running local commands looks like it would not go through
smbXsrv_client_create so the 'client' is null, segfault might happen.
We should not dereference 'client->server_multi_channel_enabled',
if 'client' is null.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=15441

[1] smb.conf example, samba-4.18.5, ubuntu 22.04.2
[global]
        dns forwarder = 127.0.0.53
        netbios name = U22-JONES-88X1
        realm = U22-JONES-88X1.X88X1.JONES
        server role = active directory domain controller
        workgroup = X88X1
        idmap_ldb:use rfc2307 = yes
        vfs objects = dfs_samba4 acl_xattr aio_pthread

[sysvol]
        path = /var/lib/samba/sysvol
        read only = No

[netlogon]
        path = /var/lib/samba/sysvol/u22-jones-88x1.x88x1.jones/scripts
        read only = No

[2] gdb
(gdb) run /usr/local/samba/bin/samba-tool ntacl get .
Starting program: /usr/local/Python3/bin/python3 /usr/local/samba/bin/samba-tool ntacl get .
[Thread debugging using libthread_db enabled]
Using host libthread_db library "/lib/libthread_db.so.1".

Program received signal SIGSEGV, Segmentation fault.
0x00007fffd0eb809e in aio_pthread_openat_fn (handle=0x8d5cc0, dirfsp=0x8c3070, smb_fname=0x18ab4f0, fsp=0x1af3550, flags=196608, mode=0)
    at ../../source3/modules/vfs_aio_pthread.c:467
warning: Source file is more recent than executable.
467             if (fsp->conn->sconn->client->server_multi_channel_enabled) {
(gdb) bt
    at ../../source3/modules/vfs_aio_pthread.c:467
    at ../../source3/smbd/pysmbd.c:320
---Type <return> to continue, or q <return> to quit---
(gdb) f
    at ../../source3/modules/vfs_aio_pthread.c:467
467             if (fsp->conn->sconn->client->server_multi_channel_enabled) {
(gdb) p fsp->conn->sconn->client
$1 = (struct smbXsrv_client *) 0x0
(gdb)

Signed-off-by: Jones Syue <jonessyue@qnap.com>
Reviewed-by: Ralph Boehme <slow@samba.org>
Reviewed-by: Stefan Metzmacher <metze@samba.org>
(cherry picked from commit 8f4c1c67b4f118a9a47b09ac7908cd3d969b19c2)

vfs_aio_pthread: don't crash without a pthreadpool

During 'samba-tool ntacl sysvolreset' and similar.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=15441

Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>
(cherry picked from commit 3694f2ce6205a647eb5dab2115785fb45decaf0b)

lib/cmdline: Also redact --newpassword in samba_cmdline_burn()

BUG: https://bugzilla.samba.org/show_bug.cgi?id=15289

Signed-off-by: Andrew Bartlett <abartlet@samba.org>
Reviewed-by: Douglas Bagnall <douglas.bagnall@catalyst.net.nz>
Autobuild-User(master): Andrew Bartlett <abartlet@samba.org>
Autobuild-Date(master): Fri Jul 21 06:16:30 UTC 2023 on atb-devel-224

(cherry picked from commit 76ad44f446c42832e87b2c60a4731a8de3a0018f)

RN: post-exec password redaction for samba-tool is more reliable for
 fully random passwords as it no longer uses regular expressions
 containing the password value itself.

Autobuild-User(v4-18-test): Jule Anger <janger@samba.org>
Autobuild-Date(v4-18-test): Fri Aug  4 08:05:00 UTC 2023 on atb-devel-224

lib/cmdline: Also burn the --password2 parameter if given

BUG: https://bugzilla.samba.org/show_bug.cgi?id=15289

Signed-off-by: Andrew Bartlett <abartlet@samba.org>
Reviewed-by: Douglas Bagnall <douglas.bagnall@catalyst.net.nz>
(cherry picked from commit 414b3803bb6a1b12c44b52ab1ff64a8b7f61fd03)

samba-tool: Use samba.glue.get_burnt_cmdline rather than regex

This use avoids having two different methods to match on command-line
passwords.  We already have a dependency on the setproctitle python
module, and this does not change as the (C) libbsd setproctitle()
can't be run from within a python module.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=15289

Signed-off-by: Andrew Bartlett <abartlet@samba.org>
Reviewed-by: Douglas Bagnall <douglas.bagnall@catalyst.net.nz>
(cherry picked from commit a53ebc288f47329c997d52325eeeb5e91ce43b75)

python: Add glue.burn_commandline() method

This uses samba_cmdline_burn() to as to have common
command line redaction code.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=15289

Signed-off-by: Andrew Bartlett <abartlet@samba.org>
Reviewed-by: Douglas Bagnall <douglas.bagnall@catalyst.net.nz>
(cherry picked from commit 3f9e455898554b726bf1689f743b2d9cb6b59537)

python: Remove const from PyList_AsStringList()

The returned strings are not owned by python, so need not be const.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=15289

Signed-off-by: Andrew Bartlett <abartlet@samba.org>
Reviewed-by: Douglas Bagnall <douglas.bagnall@catalyst.net.nz>
(cherry picked from commit 5afd206d1d8f0344a2f1fa7a238204d1fb164eda)

python: Move PyList_AsStringList to common code so we can reuse

BUG: https://bugzilla.samba.org/show_bug.cgi?id=15289

Signed-off-by: Andrew Bartlett <abartlet@samba.org>
Reviewed-by: Douglas Bagnall <douglas.bagnall@catalyst.net.nz>
(cherry picked from commit fd81759e2ed44cac3bc67243a39256f953969103)

lib/cmdline: Return if the commandline was redacted in samba_cmdline_burn()

BUG: https://bugzilla.samba.org/show_bug.cgi?id=15289

Signed-off-by: Andrew Bartlett <abartlet@samba.org>
Reviewed-by: Douglas Bagnall <douglas.bagnall@catalyst.net.nz>
(cherry picked from commit 848fea1a01a4ddc1598150823d5d0784d3ef0be4)

s3/modules: Fix DFS links when widelinks = yes

In openat(), even if we fail to open the file,
propagate stat if and only if the object is a link in
a DFS share. This allows calling code to further process
the link.

Also remove knownfail

Pair-Programmed-With: Jeremy Alison <jra@samba.org>

BUG: https://bugzilla.samba.org/show_bug.cgi?id=15435

Signed-off-by: Noel Power <noel.power@suse.com>
Reviewed-by: Jeremy Allison <jra@samba.org>
Autobuild-User(master): Jeremy Allison <jra@samba.org>
Autobuild-Date(master): Sat Jul 29 00:43:52 UTC 2023 on atb-devel-224

(cherry picked from commit 0bf8b25aacdf2f5c746922320b32e3f0886c81f5)

Autobuild-User(v4-18-test): Jule Anger <janger@samba.org>
Autobuild-Date(v4-18-test): Thu Aug  3 08:44:47 UTC 2023 on atb-devel-224

s3/modules: Add flag indicating if connected share is a dfs share

Not used yet, will be used in the next commit to avoid testing
if the connected share is a dfs one.

Pair-Programmed-With: Jeremy Alison <jra@samba.org>
BUG: https://bugzilla.samba.org/show_bug.cgi?id=15435

Signed-off-by: Noel Power <noel.power@suse.com>
Reviewed-by: Jeremy Allison <jra@samba.org>
(cherry picked from commit 2668dcd0968133cca4f8410bf8c41ed0483f5d87)

sefltest: Add new regression test dfs with widelinks = yes

Adds a new test trying to cd into dfs path on share with
widelinks enabled, should generate an error (see BUG:)

Add a knownfail so CI continues

BUG: https://bugzilla.samba.org/show_bug.cgi?id=15435

Signed-off-by: Noel Power <noel.power@suse.com>
Reviewed-by: Jeremy Allison <jra@samba.org>
(cherry picked from commit 3d2e9db8b95f9f45d486f8272e53584975f177fa)

selftest: Add new dfs share (with widelinks enabled)

Adds share (to be used in later test) that has dfs node
but additionally has widelinks set to yes

BUG: https://bugzilla.samba.org/show_bug.cgi?id=15435

Signed-off-by: Noel Power <noel.power@suse.com>
Reviewed-by: Jeremy Allison <jra@samba.org>
(cherry picked from commit b57cdfd7efb161cf96b3a39dc7a1652db817e602)

dsdb: Use samdb_system_container_dn() in pdb_samba_dsdb_*()

This makes more calls to add children, but avoids the cn=system string in the
codebase which makes it easier to audit that this is always being built
correctly.

Signed-off-by: Andrew Bartlett <abartlet@samba.org>
Reviewed-by: Stefan Metzmacher <metze@samba.org>
Autobuild-User(master): Stefan Metzmacher <metze@samba.org>
Autobuild-Date(master): Mon Jul 31 07:20:21 UTC 2023 on atb-devel-224

(cherry picked from commit 5571ce9619d856d3c9545099366f4e0259aee8ef)

RN: A second container with name CN=System would disable the operation
 of the Samba AD DC.  Samba now finds the CN=System container by exact
 DN and not a search.

Autobuild-User(v4-18-test): Jule Anger <janger@samba.org>
Autobuild-Date(v4-18-test): Tue Aug  1 10:57:04 UTC 2023 on atb-devel-224

dsdb: Use samdb_system_container_dn() in dsdb_trust_*()

This is now exactly the same actions, but just uses common code to do it.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=9959

Signed-off-by: Andrew Bartlett <abartlet@samba.org>
Reviewed-by: Stefan Metzmacher <metze@samba.org>
(cherry picked from commit 4250d07e4dcd43bf7450b1ae603ff46fdc892d02)

s4-rpc_server/backupkey: Use samdb_system_container_dn() in get_lsa_secret()

This is now exactly the same actions, but just uses common code to do it.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=9959

Signed-off-by: Andrew Bartlett <abartlet@samba.org>
Reviewed-by: Stefan Metzmacher <metze@samba.org>
(cherry picked from commit 9b4f3f3cb4ed17bb233d3b5ccd191be63f01f3f4)

s4-rpc_server/backupkey: Use samdb_system_container_dn() in set_lsa_secret()

This is now exactly the same actions, but just uses common code to do it.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=9959

Signed-off-by: Andrew Bartlett <abartlet@samba.org>
Reviewed-by: Stefan Metzmacher <metze@samba.org>
(cherry picked from commit 13eed1e0e7d0bdef6b5cdb6b858f124b812adbea)

s4-rpc_server/netlogon: Use samdb_system_container_dn() in fill_trusted_domains_array()

This is now exactly the same actions, but just uses common code to do it.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=9959

Signed-off-by: Andrew Bartlett <abartlet@samba.org>
Reviewed-by: Stefan Metzmacher <metze@samba.org>
(cherry picked from commit a900f6aa5d909d912ee3ca529baa4047c9c4da87)

s4-rpc_server/lsa: Use samdb_system_container_dn() in dcesrv_lsa_get_policy_state()

This is now exactly the same actions, but just uses common code to do it.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=9959

Signed-off-by: Andrew Bartlett <abartlet@samba.org>
Reviewed-by: Stefan Metzmacher <metze@samba.org>
(cherry picked from commit 4e18066fa243da1c505f782ba87187c3bb1078ee)

dsdb: Use samdb_get_system_container_dn() to get Password Settings Container

By doing this we use the common samdb_get_system_container_dn() routine and we
avoid doing a linerize and parse step on the main DN, instead using the
already stored parse of the DN.  This is more hygenic.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=9959

Signed-off-by: Andrew Bartlett <abartlet@samba.org>
Reviewed-by: Stefan Metzmacher <metze@samba.org>
(cherry picked from commit 3669caa97f76d3e893ac6a1ab88341057929ee6a)

dsdb: Use samdb_system_container_dn() in samldb.c

BUG: https://bugzilla.samba.org/show_bug.cgi?id=9959

Signed-off-by: Andrew Bartlett <abartlet@samba.org>
Reviewed-by: Stefan Metzmacher <metze@samba.org>
(cherry picked from commit 97b682e0eb0450513dcecb74be672e18e84fe7a2)

dsdb: Add new function samdb_system_container_dn()

This will replace many calls crafting or searching for this DN
elsewhere in the code.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=9959

Pair-Programmed-With: Stefan Metzmacher <metze@samba.org>

Signed-off-by: Andrew Bartlett <abartlet@samba.org>
Signed-off-by: Stefan Metzmacher <metze@samba.org>
(cherry picked from commit 25b0e1102e1a502152d2695aeddf7c65555b16fb)

Bug #9959: Don't search for CN=System

BUG: https://bugzilla.samba.org/show_bug.cgi?id=9959

Signed-off-by: Arvid Requate <requate@univention.de>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>
Reviewed-by: Stefan Metzmacher <metze@samba.org>
(cherry picked from commit 2d461844a201fbca55ebc9a46a15e1d16048055b)

For Bug #9959: local talloc frame for next commit

BUG: https://bugzilla.samba.org/show_bug.cgi?id=9959

Signed-off-by: Arvid Requate <requate@univention.de>
[abartlet@samba.org Added additional talloc_free() in failure paths]
Reviewed-by: Stefan Metzmacher <metze@samba.org>
(cherry picked from commit b6e80733c3a589f9d784eec86fc713f1ec9c1049)

mdssvc: fix returning file modification date for older Mac releases

Mac 10.10 uses kMDItemContentModificationDate instead of
kMDItemFSContentChangeDate.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=15427

Signed-off-by: Ralph Boehme <slow@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
Autobuild-User(master): Jeremy Allison <jra@samba.org>
Autobuild-Date(master): Wed Jul 26 23:42:44 UTC 2023 on atb-devel-224

(cherry picked from commit c2e83ebe726b7bc42b329198214c784936f19888)

Autobuild-User(v4-18-test): Jule Anger <janger@samba.org>
Autobuild-Date(v4-18-test): Mon Jul 31 09:11:21 UTC 2023 on atb-devel-224

mdssvc: fix date marshalling

Did this ever work? Possible just copied over from Netatalk and was always
broken... The Mac client expects the timevalue as seconds relative to
2001-01-01 00:00:00 UTC, packed as IEEE float.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=15427

Signed-off-by: Ralph Boehme <slow@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
(cherry picked from commit 620ca1e68d02be45a94aa41217a141d211fceb1f)

mdssvc: prepare for returning timestamps with sub-seconds granularity

BUG: https://bugzilla.samba.org/show_bug.cgi?id=15427

Signed-off-by: Ralph Boehme <slow@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
(cherry picked from commit 9dc66fecf7c1743d264c5c4f8978b77bab75ed86)

mdssvc: reduce pagesize to 50

Lastest macOS queries additional file metadata per search result, which causes
the mashalled paged result set including metadata to exceed the 64 KB result
fragment buffer.

Lacking fragementation support in mdssvc (it's supported by the protocol), for
now just reduce the maximum number of results per search page.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=15342

Signed-off-by: Ralph Boehme <slow@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
(cherry picked from commit 724a0518c901589fe1171d94648391832e056f4d)

tests/mdssvc: match hits:total:value to be the actual amount of entries in hits

BUG: https://bugzilla.samba.org/show_bug.cgi?id=15342

Signed-off-by: Ralph Boehme <slow@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
(cherry picked from commit 7f5e4edf64f7e4175f652bf8762d4edc110ad6b1)

mdssvc: fix enforcement of "elasticsearch:max results"

This wasn't enforced at all thus a query would return all available matches
without limit.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=15342

Signed-off-by: Ralph Boehme <slow@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
(cherry picked from commit d8fa5c8e2a1794ea8dc663485315ebd9401b2628)

mdssvc: add and use SL_PAGESIZE

SL_PAGESIZE is the number of entries we want to process per paged search result
set. This is different from MAX_SL_RESULTS which ought to be a default maximum
value for total number of results returned for a search query.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=15342

Signed-off-by: Ralph Boehme <slow@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
(cherry picked from commit 086c2602d074d4dc0d44f5534857e5f59a8690b2)

mdssvc: fix long running backend queries

If a query is still running in the backend and we have no results yet, returning
0 triggers a search termination by the client in latest macOS releases. macOS
returns 0x23 in this case.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=15342

Signed-off-by: Ralph Boehme <slow@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
(cherry picked from commit 925fefae20e52a3c89a56bdd0cd5b98cc951db5f)

mdssvc: set query state for continued queries to SLQ_STATE_RUNNING

SLQ_STATE_RESULTS implies that there are already results attached to the slq
which is not the case. Instead the backend will start processing from where it
left off when it hits the maximum result limit and had set the state to
SLQ_STATE_FULL.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=15342

Signed-off-by: Ralph Boehme <slow@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
(cherry picked from commit 4149ef97e5906604be1587622f390f121db183e2)