s3:libsmb: Pass domain to remote_password_change()

BUG: https://bugzilla.samba.org/show_bug.cgi?id=12975

Signed-off-by: Andreas Schneider <asn@samba.org>
Reviewed-by: Andrew Bartlet <abartlet@samba.org>

s3:utils: Do not report an invalid range for AD DC role

BUG: https://bugzilla.samba.org/show_bug.cgi?id=12629

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): Wed Aug 23 03:23:55 CEST 2017 on sn-devel-144

tdb: fix tbdtool list freelist output

Due to the non-fixable bug in the BUCKET macro tdbtool list printed some
other hash chainlist, not the freelist.

Bug: https://bugzilla.samba.org/show_bug.cgi?id=12888

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

tdb: document hashtable bucket offset calculation madness

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

tbd: BUCKET(-1) returns wrong offset because tdb->hash_size is an unsigned int

The following C program demonstrates the issue:

  #include <stdio.h>
  #include <stdlib.h>
  #include <stdarg.h>
  #include <stdbool.h>
  #include <string.h>
  #include <unistd.h>
  #include <errno.h>
  #include <dirent.h>
  #include <sys/types.h>

  int main(int argc, char **argv)
  {
      int hash = -1;
      int tsize_signed = 10;
      unsigned int tsize_unsigned = 10;
      int bucket;

  #define BUCKET(hash, tsize) ((hash) % (tsize))

      bucket = BUCKET(hash, tsize_unsigned);
      printf("hash [%d] tsize [%d] bucket [%d]\n", hash, tsize_unsigned, bucket);

      bucket = BUCKET(hash, tsize_signed);
      printf("hash [%d] tsize [%d] bucket [%d]\n", hash, tsize_signed, bucket);

      return 0;
  }

Output:

$ ./tmp
hash [-1] tsize [10] bucket [5]
hash [-1] tsize [10] bucket [-1]

The first version is what the current BUCKET() macro does. As a result
we lock the hashtable chain in bucket 5, NOT the freelist.

-1 is sign converted to an unsigned int 4294967295 and

  4294967295 % 10 = 5.

As all callers will lock the same wrong list consistently locking is
still consistent.

Stumpled across this when looking at the output of `tdbtool DB list`
which always printed some other hashchain and not the freelist.

The freelist bucket offset computation doesn't use the BUCKET macro in
freelist.c (directly or indirectly) when working on the freelist, it
just directly uses the FREELIST_TOP define, so this problem only affects
tdbtool list.

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

tdb: rename struct tdb_traverse_lock hash member to list

The variable stores the hashtable bucket, not the hash. No change in
behaviour.

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

python: Make generated modules samba.ntstatus and samba.werror Python 3 compatible.

Signed-off-by: Lumir Balhar <lbalhar@redhat.com>
Reviewed-by: Andreas Schneider <asn@samba.org>
Reviewed-by: Andrew Bartlet <abartlet@samba.org>
Autobuild-User(master): Andreas Schneider <asn@cryptomilk.org>
Autobuild-Date(master): Tue Aug 22 17:38:17 CEST 2017 on sn-devel-144

python: scripting: Port ntstatus and werror generators to Python 3 compatible form.

Signed-off-by: Lumir Balhar <lbalhar@redhat.com>
Reviewed-by: Andreas Schneider <asn@samba.org>
Reviewed-by: Andrew Bartlet <abartlet@samba.org>

s3:printing: Add NULL check for state_path()

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): Sat Aug 19 05:33:41 CEST 2017 on sn-devel-144

s3:libsmb: let do_connect() debug the negotiation result similar to "session request ok"

Also modify non-specified max_protocol to be PROTOCOL_LATEST
(currently PROTOCOL_SMB3_11).

BUG: https://bugzilla.samba.org/show_bug.cgi?id=12881

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

s3:libsmb: don't call cli_NetServerEnum() on SMB2/3 connections in SMBC_opendir_ctx()

This is all we can do with when using we allow SMB2/3 and the server supports
it, 'smb://' can't work unless we implement LLMNR and maybe WSD.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=12876

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

s3:smbclient: don't try any workgroup listing with "client min protocol = SMB2"

BUG: https://bugzilla.samba.org/show_bug.cgi?id=12863

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

s3:smbclient: improve the error messages for smbclient -L

BUG: https://bugzilla.samba.org/show_bug.cgi?id=12863

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

s3:libsmb: let get_ipc_connect() use CLI_FULL_CONNECTION_FORCE_SMB1

get_ipc_connect() is only used in code paths that require cli_NetServerEnum()
to work, so it must already require SMB1 only.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=12876

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

xattr.id: Fix a typo

Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Andreas Schneider <asn@samba.org>
Autobuild-User(master): Volker Lendecke <vl@samba.org>
Autobuild-Date(master): Fri Aug 18 20:49:42 CEST 2017 on sn-devel-144

s3:utils: Allow to run smbpasswd as user

BUG: https://bugzilla.samba.org/show_bug.cgi?id=12974

Signed-off-by: Andreas Schneider <asn@samba.org>
Reviewed-by: Volker Lendecke <vl@samba.org>
Autobuild-User(master): Andreas Schneider <asn@cryptomilk.org>
Autobuild-Date(master): Fri Aug 18 14:01:27 CEST 2017 on sn-devel-144

s3:gse_krb5: make use of precalculated krb5 keys in fill_mem_keytab_from_secrets()

This avoids a lot of cpu cycles, which were wasted for each single smb
connection, even if the client didn't use kerberos.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=12973

Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>
Autobuild-User(master): Andrew Bartlett <abartlet@samba.org>
Autobuild-Date(master): Fri Aug 18 10:04:57 CEST 2017 on sn-devel-144

s3:secrets: allow secrets_fetch_or_upgrade_domain_info() on an AD DC

The reason for the check is for write access as secrets.ldb is the
master database.

But secrets_fetch_or_upgrade_domain_info() just syncs the values
we got from if they got overwritten by secrets_store_machine_pw_sync().

BUG: https://bugzilla.samba.org/show_bug.cgi?id=12973

Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>

getncchanges.py: Add test for GET_ANC and linked attributes

Add a basic test that when we use GET_ANC and the parents have linked
attributes, then we receive all the expected links and all the expected
objects by the end of the test.

This extends the test code to track what linked attributes get received
and check whether they match what's present on the DC.

Also made some minor cleanups to store the received objects/links each
time we successfully receive a GETNCChanges response (this saves the
test case having to repeat this code every time).

Note that although this test involves linked attributes, it shouldn't
exercise the GET_TGT case at all.

Signed-off-by: Tim Beale <timbeale@catalyst.net.nz>
Reviewed-by: Garming Sam <garming@samba.org>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>
BUG: https://bugzilla.samba.org/show_bug.cgi?id=12972

getncchanges.py: Add GET_ANC replication test case

This test:
- creates blocks of parent/child objects
- modifies the parents, so the child gets received first in the
  replication (which means the client has to use GET_ANC)
- checks that we always receive the parent before the child (if not, it
  either retries with GET_ANC, or asserts if GET_ANC is already set)
- modifies the parent objects to change their USN while the
  replication is in progress
- checks that all expected objects are received by the end of the
  test

I've added a repl_get_next() function to help simulate a client's
behaviour - if it encounters an object it doesn't know the parent of,
then it retries with GET_ANC.

Also added some debug to drs_base.py that developers can turn on to make
it easier to see what objects we're actually receiving in the
responses.

Signed-off-by: Tim Beale <timbeale@catalyst.net.nz>
Reviewed-by: Garming Sam <garming@samba.org>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>
BUG: https://bugzilla.samba.org/show_bug.cgi?id=12972

getncchanges.py: Add a new test for replication

This adds a new test to check that if objects are modified during a
replication, then those objects don't wind up missing from the
replication data.

Note that when this scenario occurs, samba returns the objects in a
different order to Windows. This test doesn't care what order the
replicated objects get returned in, so long as they all have been
received by the end of the test.

As part of this, I've refactored _check_replication() in drs_base.py so
it can be reused in new tests. In these cases, the objects are split up
over multiple different chunks. So asserting that the objects are returned
in a specific order makes it difficult to run the same test on both Samba
and Windows.

Signed-off-by: Tim Beale <timbeale@catalyst.net.nz>
Reviewed-by: Garming Sam <garming@samba.org>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>
BUG: https://bugzilla.samba.org/show_bug.cgi?id=12972

replmd: Try to add forward-link for unknown cross-partition links

Previously Samba would just drop cross-partition links where the link
target object is unknown. Instead, what we want to do is try to add the
forward link for the GUID specified. We can't add the backlink because
we don't know the target, however, dbcheck should be able to fix any
missing backlinks.

The new behaviour should now mean dbcheck will detect the problem and be
able to fix it. It's still not ideal, but it's better than dropping the
link completely.

I've updated the log so that it has higher severity and tells the user
what they need to do to fix it.

These changes now mean that the selftests now detect an error - instead
of completely dropping the serverReference, we now have a missing
backlink. I've updated the selftests to fix up any missing
serverReference backlinks before running dbcheck.

Signed-off-by: Tim Beale <timbeale@catalyst.net.nz>
Reviewed-by: Garming Sam <garming@samba.org>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>
BUG: https://bugzilla.samba.org/show_bug.cgi?id=12972

replmd: Don't fail cycle if we get link for deleted object with GET_TGT

We are going to end up supporting 2 different server schemes:
A. the old/default behaviour of sending all the linked attributes last,
   at the end of the replication cycle.
B. the new/Microsoft way of sending the linked attributes interleaved
   with the source/target objects.

Normally if we're talking to a server using the old scheme-A, we won't
ever use the GET_TGT flag. However, there are a couple of cases where
it can happen:
- A link to a new object was added during the replication cycle.
- An object was deleted while the replication was in progress (and
the linked attribute got queued before the object was deleted).

Talking to an Samba DC running the old scheme will just cause it to
start the replication cycle from scratch again, which is fairly
harmless. However, there is a chance that the same thing can happen
again, in which case the replication cycle will fail (because GET_TGT
was already set).

Even if we're using the new scheme (B), we could still potentially hit
this case, as we can still queue up linked attributes between requests
(group memberships can be larger than what can fit into a single
replication chunk).

If GET_TGT is set in the GetNcChanges request, then the local copy of
the target object should always be up-to-date when we process the linked
attribute. So if we still think the target object is deleted/recycled at
this point, then it's safe to ignore the linked attribute (because we
know our local copy is up-to-date). This logic matches the MS spec logic
in ProcessLinkValue().

Not failing the replication cycle may be beneficial if we're trying to
do a full-sync of a large database. Otherwise it might be time-consuming
and frustrating to repeat the sync unnecessarily.

Signed-off-by: Tim Beale <timbeale@catalyst.net.nz>
Reviewed-by: Garming Sam <garming@samba.org>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>
BUG: https://bugzilla.samba.org/show_bug.cgi?id=12972

replmd: Avoid dropping links if link target is deleted

The server-side can potentially send the linked attribute before the
target-object. This happens on Microsoft, and will happen on Samba once
server-side GET_TGT support is added. In these cases there is a hole
where the Samba client can silently drop the linked attribute.

If the old copy of the target object was deleted/recycled, then the
client can receive the new linked attribute before it realizes the target
has now been reincarnated. It silently ignores the linked attribute,
thinking its receiving out of date information, when really it's the
client's copy of the target object that's out of date.

In this case we want to retry with the GET_TGT flag set, which will
force the updated version of the target object to be sent along with the
linked attribute. This deleted/recycled target case is the main reason
that Windows added the GET_TGT flag.

If the server sends all the links at the end, instead of along with the
source object, then this case can still be hit. If so, it will cause the
server to restart the replication from the beginning again. This is
probably preferential to silently dropping links.

Signed-off-by: Tim Beale <timbeale@catalyst.net.nz>
Reviewed-by: Garming Sam <garming@samba.org>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>
BUG: https://bugzilla.samba.org/show_bug.cgi?id=12972

replmd: Move where we store linked attributes

There was a bug in my previous patch where the code would verify
*all* links in the list, rather than just the ones that are new. And it
would do this for every replication chunk it received, regardless of
whether there were actually any links in that chunk.

The problem is by the time we want to verify the attributes, we don't
actually know which attributes are new. We can fix this by moving where
we store the linked attributes from the start of processing the
replication chunk to the end of processing the chunk. We can then verify
the new linked attributes at the same time we store them.

Longer-term we may want to try to apply the linked attribute at this
point. This would save looking up the source/target objects twice, but
it makes things a bit more complicated (attributes will usually apply at
this point *most* of the time, but not *all* the time).

Signed-off-by: Tim Beale <timbeale@catalyst.net.nz>
Reviewed-by: Garming Sam <garming@samba.org>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>
BUG: https://bugzilla.samba.org/show_bug.cgi?id=12972

drs_utils: Add GET_TGT support to 'samba-tool drs replicate --local'

Update drs_Replicate.replicate() so it handles being passed the GET_TGT
flag (more_flags). To do this, we need to always use a v10 GetNCChanges
request (v8 and v10 are essentially the same except for the more_flags).

If the replicate_chunk() call into the C bindings throws an error, check
to see whether the error could be fixed by setting the GET_TGT flag, and
re-send the request if so.

Unfortunately because WERR_DS_DRA_RECYCLED_TARGET isn't documented with
the other AD error codes, I've left it hardcoded for now (Microsoft
should be fixing up their Docs).

Signed-off-by: Tim Beale <timbeale@catalyst.net.nz>
Reviewed-by: Garming Sam <garming@samba.org>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>
BUG: https://bugzilla.samba.org/show_bug.cgi?id=12972

replmd: Set GET_ANC if Windows sends a link with unknown source object

Windows replication can send the linked attribute before it sends the
source object. The MS-DRSR spec says that in this case the client should
resend the GetNCChanges request with the GET_ANC flag set. In my testing
this resolves the problem - Windows will include the source object for the
linked attribute in the same replication chunk.

This problem doesn't happen with Samba-to-Samba replication, because the
source object for the linked attribute is guaranteed to have already been
sent.

Signed-off-by: Tim Beale <timbeale@catalyst.net.nz>
Reviewed-by: Garming Sam <garming@samba.org>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>
BUG: https://bugzilla.samba.org/show_bug.cgi?id=12972

drepl: Support GET_TGT on periodic replication client

- Update IDL comments to include Microsoft reference doc
- Add support for sending v10 GetNCChanges request (needed for the
  GET_TGT flag, which is in the new 'more_flags' field)
- Update to also set the GET_TGT flag in the same place we were setting
  GET_ANC (I split this logic out into a separate function).
- The state struct now needs to hold a 'more_flags' field as well (this
  flag is different to the GET_ANC replica flag)

Note that using the GET_TGT when replicating from a Windows DC could be
highly inefficient. Because Samba keeps the GET_TGT flag set throughout
the replication cycle, it will basically receive a repeated object from
Windows for every single linked attribute that it receives.

I believe Windows behaviour only expects the client to set the GET_TGT
flag when it actually needs to (i.e. when it receives a target object it
doesn't know about), rather than throughout the replication cycle.
However, this approach won't work with Samba-to-Samba replication,
because when the server receives the GET_TGT flag it restarts the
replication cycle from scratch. So if we only set the GET_TGT flag when
the client encountered an unknown target then Samba-to-Samba could
potentially get into an endless replication loop.

Signed-off-by: Tim Beale <timbeale@catalyst.net.nz>
Reviewed-by: Garming Sam <garming@samba.org>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>
BUG: https://bugzilla.samba.org/show_bug.cgi?id=12972

drs: Check target object is known after applying objects

Currently we only check that the target object is known at the end of
the transaction (i.e. the .prepare_commit hook). It's too late at this
point to resend the request with GET_TGT. Move this processing earlier
on, after we've applied all the objects (i.e. off the .extended hook).

In reality, we need to perform the checks at both points. I've
split the common code that gets the source/target details out of the
la_entry into a helper function. It's not the greatest function ever,
but seemed to make more sense than duplicating the code.

Signed-off-by: Tim Beale <timbeale@catalyst.net.nz>
Reviewed-by: Garming Sam <garming@samba.org>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>
BUG: https://bugzilla.samba.org/show_bug.cgi?id=12972

drs: Fail replication transaction instead of dropping links

If the DRS client received a linked attribute that it couldn't resolve
the target for, then it would just ignore that link and keep going. That
link would then be lost forever (although a full-sync would resolve
this). Instead of silently ignoring the link, fail the transaction.

This *can* happen on Samba, but it is unusual. The target object and
linked-attribute would need to be added while a replication is still in
progress. It can also happen fairly easily when talking to a Windows DC.

There are two import exceptions to this:

1). Linked attributes that span partitions. We can never guarantee that
we will have received the target object, because it may be in a partition
we haven't replicated yet. Samba doesn't have a great way of handling
this currently, but we shouldn't fail the replication (because that breaks
basic join tests). Just skip that linked attribute and hope that a
subsequent full-sync will fix it.
(I queried Microsoft and they said resolving cross-partition linked
attributes is a implementation-specific problem to solve. GET_TGT won't
resolve it)

2). When the replication involves a subset of objects, e.g.
critical-only. In these cases, we don't increase the highwater-mark, so
it is probably not such a dire problem if we don't add the link. In the
case of critical-only, we will do a subsequent full sync which will then
add the links.

Signed-off-by: Tim Beale <timbeale@catalyst.net.nz>
Reviewed-by: Garming Sam <garming@samba.org>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>
BUG: https://bugzilla.samba.org/show_bug.cgi?id=12972

replmd: Split checking link attr target into new function

We want to re-use this code to check that the linked attribute's target
object exists *before* we try to commit the transaction. This will allow
us to re-request the block with the GET_TGT flag set.

This splits checking the target object exists into a separate function.

Minor changes of note:
- the 'parent' argument was passed to replmd_process_linked_attribute()
  as NULL, so I've just replaced where it was used in the refactored code
  with NULL.
- I've tweaked the "Failed to find GUID" error message slightly to display
  the attribute ID rather than the attribute name (saves repeating
  lookups and/or passing extra arguments).
- Tweaked the replmd_deletion_state() logic - it only made sense to call
  it in the code block where we actually found the target

Signed-off-by: Tim Beale <timbeale@catalyst.net.nz>
Reviewed-by: Garming Sam <garming@samba.org>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>
BUG: https://bugzilla.samba.org/show_bug.cgi?id=12972

werror: Add WERR_DS_DRA_RECYCLED_TARGET

When the DRS client encounters a linked attribute with an unknown target
object, it should return a RECYCLED_TARGET error, which should result in
the client resending the GETNCChanges request with the GET_TGT flag set.

This error code is currently documented by Microsoft under System Error
Codes (8200-8999). I contacted them and they will also add it to the
MS-ERREF doc in future.

Signed-off-by: Tim Beale <timbeale@catalyst.net.nz>
Reviewed-by: Garming Sam <garming@samba.org>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>
BUG: https://bugzilla.samba.org/show_bug.cgi?id=12972

libcli/smb: debug an error if smb1cli_req_writev_submit() is called for SMB2/3

BUG: https://bugzilla.samba.org/show_bug.cgi?id=12968

Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
Autobuild-User(master): Jeremy Allison <jra@samba.org>
Autobuild-Date(master): Fri Aug 18 04:45:03 CEST 2017 on sn-devel-144

s3: libsmb: Add cli_smb2_chkpath() and use from cli_chkpath().

BUG: https://bugzilla.samba.org/show_bug.cgi?id=12968

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

util: Add documentation for PID file handling

Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>
Autobuild-User(master): Amitay Isaacs <amitay@samba.org>
Autobuild-Date(master): Thu Aug 17 19:45:32 CEST 2017 on sn-devel-144

ctdb-daemon: Fix a return value

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

ctdb-daemon: Fix default socket location

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

gencache: Simplify gencache_stabilize

The only record that must remain in gencache_notrans.tdb is the last_stabilize
marker. Use tdb_wipe_all and store the marker under the allrecord lock.

Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Andreas Schneider <asn@samba.org>
Autobuild-User(master): Andreas Schneider <asn@cryptomilk.org>
Autobuild-Date(master): Thu Aug 17 15:49:00 CEST 2017 on sn-devel-144

gencache: Remove tdb_check from gencache_init()

This was legacy from times when we had just one non-transactioned gencache.tdb.
With the split into transactioned gencache.tdb and fast, non-transactioned,
mutexed clear-if-first gencache_notrans.tdb this has become unnecessary.

Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Andreas Schneider <asn@samba.org>

tdb: Clarify the CLEAR_IF_FIRST locked logic

This is another level of indentation, but it took me a while staring at the
if-condition to find that "locked" was assigned the result of "==0", not the
return value of tdb_nest_lock().

Best viewed with "git show -b".

Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Andreas Schneider <asn@samba.org>

util: Add error handling to become_daemon()

Log failure and exit if fork() or setsid() fails.

Leave the logic in the non-setsid() code as it is.  This is probably
meant to fall through on failure of either opening /dev/tty or
ioctl().  Documentation for the ioctl() failure case is far from
clear.

Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Andreas Schneider <asn@samba.org>
Autobuild-User(master): Andreas Schneider <asn@cryptomilk.org>
Autobuild-Date(master): Thu Aug 17 11:48:32 CEST 2017 on sn-devel-144

util: Move become_daemon.c to samba-util-core

So that CTDB can use it.

Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Andreas Schneider <asn@samba.org>

util: Modernise logging

Switch to using DBG_ERR(), wrap logging/sd_notifyf() lines.

Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Andreas Schneider <asn@samba.org>

util: Make function definitions consistent with header file

no_process_group -> no_session, name -> daemon, drop _PUBLIC_.

Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Andreas Schneider <asn@samba.org>

util: Add become_daemon.h

Rename argument no_process_group to no_session to describe what it
actually does.  Consistently use "daemon" for name of daemon argument.
Add documentation.

Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Andreas Schneider <asn@samba.org>

util: Avoid use of includes.h

Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Andreas Schneider <asn@samba.org>

tests: replace traffic_summary test with python blackbox test

Replace the shell subunit test for script/traffic_summary.pl with a
python black box test.

This involves moving the test files to more standard locations.

Signed-off-by: Gary Lockyer <gary@catalyst.net.nz>
Reviewed-by: Douglas Bagnall <douglas.bagnall@catalyst.net.nz>
Reviewed-by: Garming Sam <garming@catalyst.net.nz>
Autobuild-User(master): Douglas Bagnall <dbagnall@samba.org>
Autobuild-Date(master): Thu Aug 17 07:59:38 CEST 2017 on sn-devel-144

scripts: Scripts to replay and generate samba traffic

Scripts to generate representative network traffic and replay this to a
samba instance.  For load testing, performance profiling and capacity
planning.

traffic_learner  process a file generated by traffic_summary and
                 generate a model that can be used by traffic_replay to
                 generate samba network traffic.

traffic_replay   Replay a summary file generated by traffic_summary, or
                 use a model created by traffic_learner to generate
                 network traffic.

Signed-off-by: Gary Lockyer <gary@catalyst.net.nz>
Signed-off-by: Douglas Bagnall <douglas.bagnall@catalyst.net.nz>
Reviewed-by: Douglas Bagnall <douglas.bagnall@catalyst.net.nz>
Reviewed-by: Garming Sam <garming@catalyst.net.nz>
Pair-programmed-with: Garming Sam <garming@catalyst.net.nz>
Pair-programmed-with: Douglas Bagnall <douglas.bagnall@catalyst.net.nz>
Pair-Programmed-With: Andrew Bartlett <abartlet@samba.org>
Pair-Programmed-With: Tim Beale <timbeale@catalyst.net.nz>

blackbox tests: method to check specific exit codes

Signed-off-by: Gary Lockyer <gary@catalyst.net.nz>
Reviewed-by: Douglas Bagnall <douglas.bagnall@catalyst.net.nz>
Reviewed-by: Garming Sam <garming@catalyst.net.nz>

traffic_summary: avoid uninitialised variable warning

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

lib: talloc: Use the system <talloc.h> include.

Signed-off-by: Jeremy Allison <jra@samba.org>
Reviewed-by: Andreas Schneider <asn@samba.org>
Autobuild-User(master): Jeremy Allison <jra@samba.org>
Autobuild-Date(master): Thu Aug 17 00:53:48 CEST 2017 on sn-devel-144

lib: tevent: Use system <tevent.h>, not internal header path (except in self-test).

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

s3: messaging: Add background job send failure message

Signed-off-by: Anoop C S <anoopcs@redhat.com>
Reviewed-by: Andreas Schneider <asn@samba.org>
Reviewed-by: Martin Schwenke <martin@meltin.net>
Autobuild-User(master): Martin Schwenke <martins@samba.org>
Autobuild-Date(master): Wed Aug 16 16:26:12 CEST 2017 on sn-devel-144

libhttp: Remove an unneeded include

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 Aug 16 04:11:47 CEST 2017 on sn-devel-144

dbwrap: Simplify dbwrap_unmarshall_fn

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

dbwrap: Simplify dbwrap_trans_delete

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

dbwrap: Simplify dbwrap_trans_store

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

dbwrap: Simplify dbwrap_store_uint32_bystring

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

dbwrap: Simplify dbwrap_store_int32_bystring

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

dbwrap: Convert dbwrap_delete to dbwrap_do_locked

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

dbwrap: Convert dbwrap_store to dbwrap_do_locked

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

py-librpc: Strictly check the type of the incoming sid pointer

This avoids casting another type of object to a void* and then to a SID

Signed-off-by: Andrew Bartlett <abartlet@samba.org>
Reviewed-by: Garming Sam <garming@catalyst.net.nz>
Autobuild-User(master): Garming Sam <garming@samba.org>
Autobuild-Date(master): Tue Aug 15 12:00:58 CEST 2017 on sn-devel-144

s4/lib/tls: Use SHA256 to sign the TLS certificates

The use of SHA-1 has been on the "do not" list for a while now, so make our
self-signed certificates use SHA256 using the new
gnutls_x509_crt_sign2 provided since GNUTLS 1.2.0

Signed-off-by: Andrew Bartlett <abartlet@samba.org>
Reviewed-by: Garming Sam <garming@catalyst.net.nz>
BUG: https://bugzilla.samba.org/show_bug.cgi?id=12953

samba-tool dns query: Allow '*' in names

As DNS wild cards are now supported we need to allow '*' characters in
the domain names.

Signed-off-by: Gary Lockyer <gary@catalyst.net.nz>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>
Reviewed-by: Garming Sam <garming@catalyst.net.nz>
BUG: https://bugzilla.samba.org/show_bug.cgi?id=12952

samba-tool dns: Test support of DNS wild card in names

As DNS wild cards are now supported we need to allow '*' characters in
the domain names.

Signed-off-by: Gary Lockyer <gary@catalyst.net.nz>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>
Reviewed-by: Garming Sam <garming@catalyst.net.nz>
BUG: https://bugzilla.samba.org/show_bug.cgi?id=12952

dnsserver: Add support for dns wildcards

Add support for dns wildcard records. i.e. if the following records
exist

  exact.samba.example.com 3600 A 1.1.1.1
  *.samba.example.com     3600 A 1.1.1.2

look up on exact.samba.example.com will return 1.1.1.1
look up on *.samba.example.com     will return 1.1.1.2
look up on other.samba.example.com will return 1.1.1.2

Signed-off-by: Gary Lockyer <gary@catalyst.net.nz>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>
Reviewed-by: Garming Sam <garming@catalyst.net.nz>
BUG: https://bugzilla.samba.org/show_bug.cgi?id=12952

dnsserver: Tighten DNS name checking

Add checks for the maximum permitted length, maximum number of labels
and the maximum label length.  These extra checks will be used by the
DNS wild card handling.

Signed-off-by: Gary Lockyer <gary@catalyst.net.nz>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>
Reviewed-by: Garming Sam <garming@catalyst.net.nz>
BUG: https://bugzilla.samba.org/show_bug.cgi?id=12952

dnsserver: Tests for dns wildcard entries

Add tests for dns wildcards.
Tests validated against Windows Server 2012 R2

Signed-off-by: Gary Lockyer <gary@catalyst.net.nz>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>
Reviewed-by: Garming Sam <garming@catalyst.net.nz>
BUG: https://bugzilla.samba.org/show_bug.cgi?id=12952

s4: com: Replace erroneous inclusion of internal talloc.h header with external.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=12932

Signed-off-by: Jeremy Allison <jra@samba.org>
Reviewed-by: Stefan Metzmacher <metze@samba.org>
Autobuild-User(master): Jeremy Allison <jra@samba.org>
Autobuild-Date(master): Tue Aug 15 08:06:40 CEST 2017 on sn-devel-144

ctdb-daemon: Drop the implementation of CHECK_SRVIDS control

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 Aug 14 13:00:16 CEST 2017 on sn-devel-144

ctdb-protocol: Drop unused protocol data structures

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

ctdb-protocol: Drop marshalling code for CHECK_SRVIDS control

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

ctdb-client: Drop client code to send CHECK_SRVIDS control

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

ctdb-tools: Drop check_srvids command from ctdb tool

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

ctdb-protocol: Mark CTDB_CONTROL_CHECK_SRVIDS obsolete

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

ctdb-client: Server id exists should check if the pid still exists

This matches what the older client code and samba does.

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

ctdb-tools: Remove duplicate code

These lines are duplicates of those above.  It has always been this
way...

Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>
Autobuild-User(master): Amitay Isaacs <amitay@samba.org>
Autobuild-Date(master): Mon Aug 14 09:00:45 CEST 2017 on sn-devel-144

ctdb-scripts: Ignore shellcheck SC2181 warning (use of $?)

Given the size of the command substitutions it would be less clear to
embed the assignments and substitutions inside a conditional.  It is
clearer if the exit code is checked afterwards.

However, do fix some untidy uses of != instead of -ne when comparing
with $?.  Make the code easier to understand by reversing the logic
and using -eq and ||.

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

ctdb-tools: Avoid shellcheck SC2181 warnings (use of $?) in onnode

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

ctdb-tools: Use a clear and readable if-statement

This is consistent with the if-statement above.

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

ctdb-tools: Reformat and explain complex code

There are multiple command groups and redirects on very long lines.
Reformat the long lines to break them up and add a comment to explain
what is happening.

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

ctdb-tools: Avoid shellcheck SC2188 warning (redirect without command)

Shellcheck found a bug!

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

ctdb-scripts: Avoid shellcheck warning SC2188 (redirect without command)

This makes the code look deliberate instead like something has been
accidentally omitted.

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

ctdb-tests: Indentation fixups

The rest of the code in this file now matches the coding guidelines,
so clean up the rest.

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

ctdb-tests: Fix logic to handle PATH additions for tests

When using non-standard test subdirectories, the current code can fail
to find the test bin directory and stupidly just adds /bin to PATH.

Switch to using CTDB_TESTS_ARE_INSTALLED along with some sanity checks
to determine the mode of operation.

With this change, test directories can now be created as
subdirectories of arbitrary component directories.  Tests can then be
run directly, either by specifying the subdirectory or individual test
cases.

Integration into the top-level tests/ directory is then done via a
symbolic link, which enables 2 things:

* Ability to run a directory of test cases from top level by simply
  specifying the link name.

* Ease of installation - the installation code just works with the
  symbolic link.

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

ctdb-tests: Move die() function to top of script

So it can be called within the script instead of just by scripts that
include it.

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

ctdb-tests: run_tests.sh sets evironment variable CTDB_TEST_DIR

Instead of just local variable test_dir.  The environment variable can
be accessed from other test infrastructure scripts.

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

blackbox: Add test for 'net ads changetrustpw'

BUG: BUG: https://bugzilla.samba.org/show_bug.cgi?id=12956

Signed-off-by: Andreas Schneider <asn@samba.org>
Reviewed-by: Richard Sharpe <realrichardsharpe@gmail.com>
Autobuild-User(master): Andreas Schneider <asn@cryptomilk.org>
Autobuild-Date(master): Fri Aug 11 22:09:27 CEST 2017 on sn-devel-144

s3:libads: Fix changing passwords with Kerberos

BUG: https://bugzilla.samba.org/show_bug.cgi?id=12956

Signed-off-by: Andreas Schneider <asn@samba.org>
Reviewed-by: Richard Sharpe <realrichardsharpe@gmail.com>

s4:rpc_server:backupkey: Move variable into scope

CID: #1415510

BUG: https://bugzilla.samba.org/show_bug.cgi?id=12959

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

heimdal: Fix printing a short int into a string

The size of portstr is too small to print an integer and we should print
a short anyway.

This fixes building with GCC 7.1

BUG: https://bugzilla.samba.org/show_bug.cgi?id=12930

Signed-off-by: Andreas Schneider <asn@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
Autobuild-User(master): Andreas Schneider <asn@cryptomilk.org>
Autobuild-Date(master): Fri Aug 11 18:08:04 CEST 2017 on sn-devel-144

examples: add cache effectiveness stats to gencache.stp

Signed-off-by: Ralph Boehme <slow@samba.org>
Reviewed-by: Stefan Metzmacher <metze@samba.org>
Autobuild-User(master): Ralph Böhme <slow@samba.org>
Autobuild-Date(master): Fri Aug 11 14:19:24 CEST 2017 on sn-devel-144

README.Coding: add "Error and out logic"

Signed-off-by: Ralph Boehme <slow@samba.org>
Reviewed-by: Simo <simo@samba.org>
Autobuild-User(master): Ralph Böhme <slow@samba.org>
Autobuild-Date(master): Thu Aug 10 14:36:01 CEST 2017 on sn-devel-144

ctdb-tests: Add a big no-op LCP2 IP takeover test

Although this tests correctness it is most useful for testing that
changes to IP takeover algorithm do not cause obvious performance
regressions.

Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>
Autobuild-User(master): Amitay Isaacs <amitay@samba.org>
Autobuild-Date(master): Thu Aug 10 10:30:58 CEST 2017 on sn-devel-144

ctdb-takeover: Do not call ctdb_announce_vnn_iface() for updateip

This causes any tracked connections for the IP address to be lost.

When doing a takeip, the server sends a tickle ACK to the client, the
client responds with a valid ACK and the server's TCP stack responds
with a reset because the connection does not exist.  However, in the
updateip, case the connection *does* exist, so the tickle *does not*
cause the connection to be reset.

ctdb_announce_vnn_iface() clears the list of tracked TCP connections
while sending the tickle ACKs.  So, if there are no reconnects as in
the takeip case, then the list of connections is simply lost.

The "updateip" event in the 10.interface event script already sends
gratuitous ARPs and tickles connections in both directions.  This
ensures that traffic continues after packets may have been dropped
when the script temporarily blocks traffic to the IP address.

All of this means that the call to ctdb_announce_vnn_iface() can just
be deleted.

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

build: Do not recurse on symlinks to directories when building tarballs

DIST_FILES() causes all files in any specified directory to be
recursively added to the tarball.  However, a symbolic link to a
directory is detected as a regular directory so is also subject to
recursion.  This means that a symbolic link to a directory is
dereferenced and the directory of files beyond it are added to the
tarball under a directory corresponding to the link.  This is almost
certainly not what is intended because it will usually result in
duplicate files.  This is because the contents of a symbolic link's
target directory will already be present in the tarball.

Instead, do not treat symbolic links to directories as directories,
but add them to the tarball like normal files.

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

build: Do not ignore symlinks to directories when building tarballs

Tarballs currently do not contain symbolic links to directories even
if they are committed in git.  This means that CTDB tests fail when
run in-tree from a tarball, due to a couple of missing links needed by
unit tests:

  ERROR: Directory .../ctdb/tests/var/unit_eventscripts/etc-ctdb/events.d does not exist.

Subdirectories of directories specified via DIST_DIRS() are ignored,
since all the files within them are separately added to the tarball.
Symbolic links to directories are detected as directories, so they are
also ignored, causing them to be missing from the tarball.

Instead, do not treat symbolic links to directories as directories,
but add them to the tarball like normal files.

It is easy to confirm that this change causes no difference to current
tarballs other than causing the missing CTDB test links to be added:

  $ diff -u samba-4.8.0pre1-GIT-eb691cd0242.tar.gz.contents samba-4.8.0pre1-GIT-dfb16de0149.tar.gz.contents
  --- samba-4.8.0pre1-GIT-eb691cd0242.tar.gz.contents 2017-08-08 20:21:40.022993091 +1000
  +++ samba-4.8.0pre1-GIT-dfb16de0149.tar.gz.contents 2017-08-08 20:35:11.001580747 +1000
  @@ -578,7 +578,9 @@
   ctdb/tests/eventscripts/91.lvs.startup.001.sh
   ctdb/tests/eventscripts/91.lvs.startup.002.sh
   ctdb/tests/eventscripts/README
  +ctdb/tests/eventscripts/etc-ctdb/events.d
   ctdb/tests/eventscripts/etc-ctdb/functions
  +ctdb/tests/eventscripts/etc-ctdb/nfs-checks.d
   ctdb/tests/eventscripts/etc-ctdb/nfs-linux-kernel-callout
   ctdb/tests/eventscripts/etc-ctdb/public_addresses
   ctdb/tests/eventscripts/etc-ctdb/rc.local

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

tdb: Do not allow to pass NULL as the buffer to transaction_write()

This fixes a GCC warning.

Signed-off-by: Andreas Schneider <asn@samba.org>
Reviewed-by: Volker Lendecke <vl@samba.org>
Autobuild-User(master): Andreas Schneider <asn@cryptomilk.org>
Autobuild-Date(master): Thu Aug 10 02:26:09 CEST 2017 on sn-devel-144

tdb: Write zero data using 8k buffer in transaction_expand_file()

Signed-off-by: Andreas Schneider <asn@samba.org>
Reviewed-by: Volker Lendecke <vl@samba.org>

tdb: Avoid NULL tdb_write

Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Andreas Schneider <asn@samba.org>