HACK smb2 session setup STATUS_PENDING after 10 secs

Revert "TMP: torture:smb2.session: only run bind1 test"

This reverts commit 75db98a26be2b11f35c29459f7ce9d4701e4b051.

Revert "HACK only test_session_reconnect1"

This reverts commit 3dbd1092c912f9436e0376250b9c33407672aff9.

HACK only test_session_reconnect1

TMP: torture:smb2.session: only run bind1 test

Revert "TMP DEBUG smb2 server"

This reverts commit 60ae322fb176cc32c888262f6eefa744e94445ae.

TMP DEBUG smb2 server

was commit 4153765d4c741f6b5fe8285313422f169156af75

Revert "TMP: DEBUG libcli/smb/smbXcli_base.c"

This reverts commit 938a99a2e327db1bf3d8fb110449d3b55c8b8f93.

TMP: DEBUG libcli/smb/smbXcli_base.c

Revert "TODO make it possible to run smb2.session.bind1 test with -U% against IPC$ (windows)"

This reverts commit f5eaec43d48bb939723500536012951f1730717b.

TODO make it possible to run smb2.session.bind1 test with -U% against IPC$ (windows)

STATUS_INVALID_PARAMETER is returned

--- 00-STEPS-multi-channel: STEP07b - session bind [DEBUG] ---

STEP07 - TMP/DEBUG security >= user

STEP07 - TMP/DEBUG debug binding

STEP07 - DEBUG - smb2:session bind

STEP07 - TMP/DEBUG BIND

--- 00-STEPS-multi-channel: STEP02c - SMB connection passing [tests] ---

Revert "msgtest fd-passing"

This reverts commit b3acb9b155309f5459cd74c884a03e8b0a9b97f5.

msgtest fd-passing

-- 00-STEPS-multi-channel: STEP02b - SMB connection passing [DEBUG] ---

STEP02: smbXsrv_client: debug filter_pass

--- 00-STEPS-multi-channel: STEP02x - SMB connection passing - extra ---

STEP02: TODO smbXsrv_client: add commented out stuff for future purposes..

STEP02: smbXsrv_client: add commented out code related to local db_rec

STEP02: smbXsrv_client: add commented out code in ..pass_filter"

(prep for verifying client guid..)

STEP02: smbXsrv_client: add commented out store op in smbXsrv_client_create

STEP02: smbXsrv_client: add commented out table var to smbXsrv_client_connection_pass_loop"

--- 00-STEPS-multi-channel: STEP02a - SMB connection passing HACK ---

--- 00-STEPS-multi-channel: STEP07a enable multi channel ---

--- 00-STEPS-multi-channel: START, STEPs 01,05,06,07,10,02: [UPSTREAM] ---

update TODO.multi-channel

selftest: samba3.smb2.replay.replay8 fails (incomplete test)

selftest: samba3.smb2.replay.replay3 succeeds now

selftest: samba3.smb2.replay.replay4 succeeds now (channel sequence)

Signed-off-by: Michael Adam <obnox@samba.org>

TODO selftest: enable multi channel in the nt4_dc env

Signed-off-by: Michael Adam <obnox@samba.org>

TMP/DEBUG: debugging in smb2 dispatcher

SQ: torture:smb2: replay8 needs multi-channel

LATER? HACK add smb2.replay.replay8 test

SQ: torture:smb2: replay7 needs multi-channel

LATER? s4:torture: Add smb2.replay.replay7 test to examine server behaviour when Multiple Channels are utilized.

Signed-off-by: Anubhav Rakshit <anubhav.rakshit@gmail.com>
LATER? sq replay7 doesn't work (on a single CPU server?)

LATER? bla replay7...

s4-torture: add CHECK_CREATE_OUT_WITHOUT_HANDLE macro for checking two smb2_creates w/o file handle.

Guenther

Signed-off-by: Günther Deschner <gd@samba.org>

s3:smbXsrv.idl: add 8 byte channel_sequence number and request counters to IDL.

Guenther

Signed-off-by: Guenther Deschner <gd@samba.org>

selftest:knownfail: smb2.lock.*replay succeed now

Signed-off-by: Michael Adam <obnox@samba.org>

s3-smbd: enable SMB2 lock sequence checking when SMB3 is enabled.

While windows enables it only for resilient and persistent handles a SMB server
SHOULD (according to MS-SMB2 section 3.3.5.14 ) activate processing of lock
sequence numbers for EITHER
- protocol dialect 2.1 and resilient handles OR
- protocol dialect 3.x.

Guenther

Signed-off-by: Guenther Deschner <gd@samba.org>

s4-torture: in smb2.lock.replay try testing lock sequence checking against SMB3 samba servers.

Guenther

Signed-off-by: Guenther Deschner <gd@samba.org>

s4-torture: make smb2.lock.replay test work against Windows 2012 R2.

Guenther

Signed-off-by: Guenther Deschner <gd@samba.org>

s4-torture: make test_session_reauth5 work against Windows 2012R2.

The unlink of the file as anonymous is expected to fail.

Guenther

Signed-off-by: Guenther Deschner <gd@samba.org>

s3:smb2_lock: implement lock_sequence replay detection

Pair-Programmed-With: Michael Adam <obnox@samba.org>

Signed-off-by: Stefan Metzmacher <metze@samba.org>
Signed-off-by: Michael Adam <obnox@samba.org>
Reviewed-by: Guenther Deschner <gd@samba.org>

s3:smb2_lock: pass in_lock_sequence to smbd_smb2_lock_send()

Take the value from the client if the dialect is SMB2_10 or higher,
otherwise default to 0.

Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Guenther Deschner <gd@samba.org>

s3:smbXsrv_open: initialize smbXsrv_open_global->lock_sequence_array with 0xFF

This does not match the current documentation, but is very likely the
right thing to do.

If we would match the documentation and initialize with 0x00,
we would return STATUS_SUCCESS without doing any locks.
If the client also follows the documentation and starts
with a lock_sequence of 0 for the first operation.

Signed-off-by: Stefan Metzmacher <metze@samba.org>
Signed-off-by: Michael Adam <obnox@samba.org>
Reviewed-by: Guenther Deschner <gd@samba.org>

s3:smbXsrv.idl: add lock_sequence_array to smbXsrv_open_global0

This is needed for lock replay detection.

Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Guenther Deschner <gd@samba.org>

TODO.multi-channel: update plans for oplock breaks

Update TODO.multi-channel

TODO.multi-channel

STEP02a - HACK (avoid swrap fdpass..) accept unix sockets... source3/lib/util_sock.c

This lets selftest work without socket-wrapper supporting fd-passing...

ctdb-daemon: Log a message when vfork() takes long time

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): Mon Sep 19 12:18:15 CEST 2016 on sn-devel-144

ctdb-daemon: Log a message when fork() takes long time

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

ctdb-daemon: Avoid extra condition in tevent trace callback

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

ctdb-recovery-helper: Add missing initialisation of ban_credits

BUG: https://bugzilla.samba.org/show_bug.cgi?id=12275

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

lib: poll_funcs : poll_funcs_context_slot_find can select the wrong slot to replace.

Look for an exact match first, before a free slot.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=12272

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): Mon Sep 19 07:12:00 CEST 2016 on sn-devel-144

s4/messaging: let the imessaging ctx destructor free msg_dgm_ref

BUG: https://bugzilla.samba.org/show_bug.cgi?id=12272

Signed-off-by: Ralph Boehme <slow@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>

lib/poll_funcs: free contexts in poll_funcs_state_destructor()

This ensures the destructors get called in the proper order.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=12272

Signed-off-by: Ralph Boehme <slow@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>

lib/poll_funcs: free timers in poll_funcs_state_destructor()

BUG: https://bugzilla.samba.org/show_bug.cgi?id=12272

Signed-off-by: Ralph Boehme <slow@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>

s3/smbd: remove a misleading error message

It can happen that we get 0 cleanup events, so remove this error
message.

Signed-off-by: Ralph Boehme <slow@samba.org>
Reviewed-by: Andreas Schneider <asn@samba.org>
Autobuild-User(master): Ralph Böhme <slow@samba.org>
Autobuild-Date(master): Fri Sep 16 16:43:16 CEST 2016 on sn-devel-144

tevent_tutorial: Fix tevent_thread referencing

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): Fri Sep 16 05:16:02 CEST 2016 on sn-devel-144

tevent_tutorial: Fix typos

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

s3/smbd: set FILE_ATTRIBUTE_DIRECTORY as necessary

Some VFS modules like GPFS will always return success from
SMB_VFS_GET_DOS_ATTRIBUTES() but only set a subset of the attributes. It
neither sets FILE_ATTRIBUTE_NORMAL nor FILE_ATTRIBUTE_DIRECTORY.

We already handle the case that the VFS stack returns with result==0 and
then add the FILE_ATTRIBUTE_NORMAL, regardless of the type of the
filesystem object. If we want to handle result==0 situation in the
SMB_VFS_GET_DOS_ATTRIBUTES() caller, then do it right by either setting
FILE_ATTRIBUTE_NORMAL or FILE_ATTRIBUTE_DIRECTORY.

Bug: https://bugzilla.samba.org/show_bug.cgi?id=12261

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): Fri Sep 16 00:34:43 CEST 2016 on sn-devel-144

nfs4acl: Fix owner mapping with ID_TYPE_BOTH

This fixes a corner case when using NFS4 ACLs with ID_TYPE_BOTH.  Before
this patch, the owner entry in the ACL would be mapped to a gid entry in
the NFSv4 ACL, and not the expected special owner entry. This is caused
by the id mapping returning a valid gid and the nfs4 mapping assumed
that this was actually a group.

Fix this by asking for the uid first, and explicitly checking if the
mapped uid matches the owner. That creates a uid entry in the NFSv4 ACL
that can be changed later in smbacl4_substitute_{simple,special} to the
expected special owner entry.

Signed-off-by: Christof Schmitt <cs@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>

s3-spoolss: fix winreg_printer_ver_to_qword

We were reporting the OS minor number as the driver version number in all
GetDriver/EnumDriver calls.

Guenther

Signed-off-by: Guenther Deschner <gd@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>

smbd: Reset O_NONBLOCK on open files

See the comment inline :-)

Bug: https://bugzilla.samba.org/show_bug.cgi?id=12268
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Simo <simo@samba.org>
Autobuild-User(master): Volker Lendecke <vl@samba.org>
Autobuild-Date(master): Thu Sep 15 20:21:41 CEST 2016 on sn-devel-144

s4:samba_spnupdate: do not attempt to parse log level, use parsed value

The log level parameter can contain debug class specific entries.
Do not attempt to parse this as int, but use the values that the
debugging system already parsed

BUG: https://bugzilla.samba.org/show_bug.cgi?id=9945
Signed-off-by: Christian Ambach <ambi@samba.org>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>
Autobuild-User(master): Christian Ambach <ambi@samba.org>
Autobuild-Date(master): Wed Sep 14 23:15:18 CEST 2016 on sn-devel-144

python/join: do not attempt to parse log level, use parsed value

The log level parameter can contain debug class specific entries.
Do not attempt to parse this as int, but use the values that the
debugging system already parsed

BUG: https://bugzilla.samba.org/show_bug.cgi?id=9945
Signed-off-by: Christian Ambach <ambi@samba.org>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>

python/drs_utils: do not attempt to parse log level, use parsed value

The log level parameter can contain debug class specific entries.
Do not attempt to parse this as int, but use the values that the
debugging system already parsed

BUG: https://bugzilla.samba.org/show_bug.cgi?id=9945
Signed-off-by: Christian Ambach <ambi@samba.org>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>

tests/param add a test for LoadParm.log_level

Signed-off-by: Christian Ambach <ambi@samba.org>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>

s4:param add log_level function to retrieve log level in Python code

Signed-off-by: Christian Ambach <ambi@samba.org>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>

unix_msg: Fix unix_dgram_send_queue_init

We should not free a DLIST element

Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Ralph Böhme <slow@samba.org>
Autobuild-User(master): Volker Lendecke <vl@samba.org>
Autobuild-Date(master): Wed Sep 14 19:30:28 CEST 2016 on sn-devel-144

unix_msg: Fix CID 1372875 Double close

The two error callers of unix_dgram_sendq_schedule_free already run down the
sendq in the error case.

Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Ralph Böhme <slow@samba.org>

ctdb-tests: Add new public IP takeover no-op test

Test with DisableIPFailover=1 and with no public IP addresses
configured.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=12254

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): Wed Sep 14 12:30:30 CEST 2016 on sn-devel-144

ctdb-tests: Factor out new local daemons functions ps_ctdbd

Useful for being able to ensure that tests are doing what is expected.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=12254

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

ctdb-ipalloc: ipalloc_set_public_ips() can't fail

So make it a void function.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=12254

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

ctdb-ipalloc: Move merged IP list creation to ipalloc()

BUG: https://bugzilla.samba.org/show_bug.cgi?id=12254

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

ctdb-ipalloc: Drop known_ips argument from merged IP list creation

This is available in the IP allocation state.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=12254

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

ctdb-ipalloc: Optimise check to see if IPs can be hosted

Add an early return if there are no known IP addresses.

Also add an extra comment for clarification.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=12254

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

ctdb-ipalloc: Whether IPs can be hosted need not depend on merged IP list

Merged IP list won't be available here...

BUG: https://bugzilla.samba.org/show_bug.cgi?id=12254

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

ctdb-ipalloc: Store known public IPs in IP allocation state

This was dropped because it wasn't used, but it will be needed again.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=12254

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

ctdb-packaging: Fix systemd network dependency

https://www.freedesktop.org/wiki/Software/systemd/NetworkTarget/ says:

  network.target has very little meaning during start-up. [...]
  Whether any network interfaces are already configured when it is
  reached is undefined. [...]

  network-online.target is a target that actively waits until the
  ne[t]work is "up",

CTDB expects to be able to bind a socket to a node address and expects
interfaces for public IP addresses to exist.  CTDB also doesn't expect
time to jump, so also wait until time is synchronised.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=12255

Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Steve French <sfrench@samba.org>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>
Tested-by: Steve French <sfrench@samba.org>

ctdb-daemon: Remove NUM_DB_PRIORITIES

It's not used anymore.

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

ctdb-recoverd: Drop code to freeze databases from set_recovery_mode()

This function is called only once from force_election() and does not
require freezing of databases.

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

s3-ctdb: Use correct db_id size in marshalling record buffer

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

ctdb-common: Use correct db_id size in marshalling record buffer

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

ctdb-protocol: Fix marshalling for GET_DB_SEQNUM control request

BUG: https://bugzilla.samba.org/show_bug.cgi?id=12259

Even though database id is 32-bit, it's sent on wire as 64-bits.
The database id is the first 32-bits on the wire.  This needs fixing
eventually, but for now keep the same wire format.

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

s3:libsmb fix a typo

Signed-off-by: Christian Ambach <ambi@samba.org>
Reviewed-by: Volker Lendecke <vl@samba.org>
Autobuild-User(master): Volker Lendecke <vl@samba.org>
Autobuild-Date(master): Wed Sep 14 02:06:18 CEST 2016 on sn-devel-144

s4-kdc: Do not leak memory on error in kpasswd_make_error_reply()

CID 1372874

Signed-off-by: Andreas Schneider <asn@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
Autobuild-User(master): Jeremy Allison <jra@samba.org>
Autobuild-Date(master): Tue Sep 13 22:13:31 CEST 2016 on sn-devel-144

unix_msg: add a test for dgram socket caching

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): Tue Sep 13 04:11:30 CEST 2016 on sn-devel-144

unix_msg: always create a send queue for a peer

Previously, we only created a send queue for a peer if the initial send
to the non-blocking non-connected socket reported EWOULDBOCK (because
the channel was full).

With this change, we now always create a send queue and use a connected,
non-blocking datagram socket from the beginning.

Initially, the socket of the send queue is set to non-blocking mode and
we attempt a direct send via sendmsg(). If that returns EWOULDBOCK, we
set the send queue to blocking mode and let the threadpool handle the
IO.

When a send queue becomes empty, we set the send queue socket back to
non-blocking.

Signed-off-by: Ralph Boehme <slow@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>

unix_msg: introduce send queue caching

This introduces caching of unix datagram send queues. Right now send
queues are only created for peers if the channel to the peer is full and
a send reported EWOULDBLOCK.

At this stage, performance will actually be slightly worse, because now
if there's a cached queue for a peer without queued messages, we don't
attempt direct send anymore until the send queue is removed from the
cache.

The next commit will modify unix_msg to always create a send queue with
the datagram socket in connected mode and again attempt an non-blocking
send on the connected socket first. Then only if that returns
EWOULDBLOCK, the send has to go through the threadpool.

Signed-off-by: Ralph Boehme <slow@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>

unix_msg: add flag to prepare_socket_nonblock()

This allows prepare_socket_nonblock() to be called to set a socket to
non-blocking (as before) as well as blocking. This will be used in a
subsequent commit.

Signed-off-by: Ralph Boehme <slow@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>

messaging: Call messaging_dgm_send under become_root only if necessary

Signed-off-by: Ralph Boehme <slow@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>

unix_msg: Return errno from find_send_queue

Signed-off-by: : Ralph Boehme <slow@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>

unix_msg: modify find_send_queue() to take a struct sockaddr_un

In one of the next commits unix_dgram_send_queue_init() will be moved
into find_send_queue and that takes a struct sockaddr_un.

Signed-off-by: Ralph Boehme <slow@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>