13 years agos3: require tdb 1.1.7 with tdb_check()
Stefan Metzmacher [Fri, 4 Dec 2009 15:59:39 +0000 (16:59 +0100)]
s3: require tdb 1.1.7 with tdb_check()


13 years agos3:docs: Document "directory name cache size".
Karolin Seeger [Mon, 7 Dec 2009 15:28:30 +0000 (16:28 +0100)]
s3:docs: Document "directory name cache size".


13 years agos3: de-deprecate "write cache size" to prevent its removal without a proper alternative
Volker Lendecke [Mon, 7 Dec 2009 15:24:08 +0000 (16:24 +0100)]
s3: de-deprecate "write cache size" to prevent its removal without a proper alternative

13 years agoFix release script with newer versins of git
Simo Sorce [Mon, 7 Dec 2009 14:25:03 +0000 (09:25 -0500)]
Fix release script with newer versins of git

13 years agos4-smbtorture: disable some spoolss printerdata tests which s3 does not pass at the...
Günther Deschner [Mon, 7 Dec 2009 13:38:24 +0000 (14:38 +0100)]
s4-smbtorture: disable some spoolss printerdata tests which s3 does not pass at the moment.


13 years agos3: fix net and rpcclient after setprinterdataex changes.
Günther Deschner [Mon, 7 Dec 2009 10:54:26 +0000 (11:54 +0100)]
s3: fix net and rpcclient after setprinterdataex changes.


13 years agos4-smbtorture: fix RPC-SPOOLSS after _spoolss_SetPrinterDataEx IDL change.
Günther Deschner [Fri, 4 Dec 2009 23:58:45 +0000 (00:58 +0100)]
s4-smbtorture: fix RPC-SPOOLSS after _spoolss_SetPrinterDataEx IDL change.


13 years agos3-spoolss: let _spoolss_DeletePrinterData call into _spoolss_DeletePrinterDataEx.
Günther Deschner [Mon, 7 Dec 2009 10:28:56 +0000 (11:28 +0100)]
s3-spoolss: let _spoolss_DeletePrinterData call into _spoolss_DeletePrinterDataEx.


13 years agos3-spoolss: let _spoolss_SetPrinterData call into _spoolss_SetPrinterDataEx.
Günther Deschner [Fri, 4 Dec 2009 23:56:55 +0000 (00:56 +0100)]
s3-spoolss: let _spoolss_SetPrinterData call into _spoolss_SetPrinterDataEx.


13 years agos3-spoolss: let _spoolss_GetPrinterData call into _spoolss_GetPrinterDataEx.
Günther Deschner [Fri, 4 Dec 2009 23:52:28 +0000 (00:52 +0100)]
s3-spoolss: let _spoolss_GetPrinterData call into _spoolss_GetPrinterDataEx.


13 years agospoolss: rework spoolss_GetPrinterDataEx and spoolss_SetPrinterDataEx.
Günther Deschner [Fri, 4 Dec 2009 23:50:18 +0000 (00:50 +0100)]
spoolss: rework spoolss_GetPrinterDataEx and spoolss_SetPrinterDataEx.


13 years agos3-spoolss: a default printer should have at least a "PrintDriverData" key.
Günther Deschner [Fri, 4 Dec 2009 16:22:25 +0000 (17:22 +0100)]
s3-spoolss: a default printer should have at least a "PrintDriverData" key.


13 years agos3-winreg: fill in some inline comments.
Günther Deschner [Fri, 4 Dec 2009 12:33:49 +0000 (13:33 +0100)]
s3-winreg: fill in some inline comments.


13 years agos3-winreg: remove trailing whitespace.
Günther Deschner [Fri, 4 Dec 2009 12:29:08 +0000 (13:29 +0100)]
s3-winreg: remove trailing whitespace.


13 years agos3:auth: Fix typo in debug message.
Karolin Seeger [Mon, 7 Dec 2009 13:38:18 +0000 (14:38 +0100)]
s3:auth: Fix typo in debug message.


13 years agoexamples: Fix the build of the example VFS modules.
Karolin Seeger [Mon, 7 Dec 2009 09:15:49 +0000 (10:15 +0100)]
examples: Fix the build of the example VFS modules.


13 years agos4:dsdb Use ldb_match_msg_objectclass in operational.c
Andrew Bartlett [Mon, 7 Dec 2009 04:28:51 +0000 (15:28 +1100)]
s4:dsdb Use ldb_match_msg_objectclass in operational.c

This avoids not only a possibly non-portable compiler expression, but
also the need to look up the schema each time.

Andrew Bartlett

13 years agos4:ldb Add a function to match a message against an objectClass
Andrew Bartlett [Mon, 7 Dec 2009 04:27:43 +0000 (15:27 +1100)]
s4:ldb Add a function to match a message against an objectClass

(as objectClass will always be a case insensitive ascii string, we can
make a much simpler match function here than for the general case).

Andrew Bartlett

13 years Improve testsuite for primaryGroupToken behaviour
Andrew Bartlett [Mon, 7 Dec 2009 01:44:07 +0000 (12:44 +1100)] Improve testsuite for primaryGroupToken behaviour

This tries to show that the domain object should not have a
primaryGroupToken, for example.  (This passes against the old and new
code, as the failure case requires an object with an objectSid, and
exactly one group in it's subtree.  Sadly I don't know of a valid structure
that I can construct to test this).

Andrew Bartlett

13 years agos4:dsdb Make primaryGroupToken calculation more efficient and correct
Andrew Bartlett [Mon, 7 Dec 2009 01:41:43 +0000 (12:41 +1100)]
s4:dsdb Make primaryGroupToken calculation more efficient and correct

The original code here would do a subtree search under each object,
attempting to determine if it was a group.  This was incorrect, and
inefficient - we just need to ask for the objectClass attribute, and
check that value before returning the group's RID.

(Much of this patch reworks operational.c to allow a search for 2
attributes for this calculation).

Andrew Bartlett

13 years agos4:dsdb Make parentGUID handler use dsdb_module_search_dn()
Andrew Bartlett [Thu, 3 Dec 2009 23:15:15 +0000 (10:15 +1100)]
s4:dsdb Make parentGUID handler use dsdb_module_search_dn()

This avoids doing a new search from the top of the module stack.

This also removes the helper function dsdb_find_parentguid_by_dn()
which is now unused.

Andrew Bartlett

13 years agos4:build torturedir and -DTORTUREDIR is unused
Andrew Bartlett [Thu, 3 Dec 2009 20:29:29 +0000 (07:29 +1100)]
s4:build torturedir and -DTORTUREDIR is unused

(And made building RPMs much more difficult for JHT)

13 years agos4:dsdb Hide the LM password by default too
Andrew Bartlett [Tue, 1 Dec 2009 21:53:32 +0000 (08:53 +1100)]
s4:dsdb Hide the LM password by default too

13 years agostreamline some log levels for invalid servicenames
Christian Ambach [Wed, 2 Dec 2009 19:06:37 +0000 (20:06 +0100)]
streamline some log levels for invalid servicenames

I don't think we need to log the fact that a user gave a wrong sharename in Explorer with the highest log level.
The level of this was not very consistent:
service.c: DEBUG(3,("find_service() failed to find service %s\n", service));
service.c: DEBUG(0,("%s (%s) couldn't find service %s\n",
smb2_tcon.c: DEBUG(1,("smbd_smb2_tree_connect: couldn't find service %s\n",

This changes the last two to 3 as the first one.

Signed-off-by: Christian Ambach <>
13 years agos3: Fix a const discard warning
Volker Lendecke [Sun, 6 Dec 2009 11:50:51 +0000 (12:50 +0100)]
s3: Fix a const discard warning

13 years agos3: talloc_array_length deals file with a NULL argument
Volker Lendecke [Sun, 6 Dec 2009 11:50:25 +0000 (12:50 +0100)]
s3: talloc_array_length deals file with a NULL argument

13 years agos3: Small simplification: qsort deals fine with just one entry
Volker Lendecke [Sun, 6 Dec 2009 11:49:46 +0000 (12:49 +0100)]
s3: Small simplification: qsort deals fine with just one entry

13 years agos3: Simplify usr_info_cmp by calling timeval_compare()
Volker Lendecke [Sun, 6 Dec 2009 11:49:09 +0000 (12:49 +0100)]
s3: Simplify usr_info_cmp by calling timeval_compare()

13 years agos3: Add rpcclient wkssvc_enumerateusers
Volker Lendecke [Sun, 6 Dec 2009 11:40:44 +0000 (12:40 +0100)]
s3: Add rpcclient wkssvc_enumerateusers

13 years agos3: Complete support for NetWkstaGetInfo/NetWkstaEnumUsers
Ian Puleston [Sun, 22 Nov 2009 23:00:00 +0000 (00:00 +0100)]
s3: Complete support for NetWkstaGetInfo/NetWkstaEnumUsers

13 years agos3: First cut at _wkssvc_NetWkstaEnumUsers
Volker Lendecke [Mon, 3 Aug 2009 15:27:11 +0000 (17:27 +0200)]
s3: First cut at _wkssvc_NetWkstaEnumUsers

This needs access checks!

13 years agos3: Support NetWkstaGetInfo 101 and 102
Volker Lendecke [Mon, 3 Aug 2009 14:52:01 +0000 (16:52 +0200)]
s3: Support NetWkstaGetInfo 101 and 102

13 years agos4-ldif: Fix memory leek in ldb_ldif_write()
Kamen Mazdrashki [Sat, 5 Dec 2009 20:49:59 +0000 (22:49 +0200)]
s4-ldif: Fix memory leek in ldb_ldif_write()

13 years agopackaging(RHEL-CTDB): package dbwrap_tool and dbwrap_torture in the common pkg
Michael Adam [Fri, 4 Dec 2009 22:29:22 +0000 (23:29 +0100)]
packaging(RHEL-CTDB): package dbwrap_tool and dbwrap_torture in the common pkg


13 years agos3:dbwrap_ctdb: increase the number of commit retries 5-->100
Michael Adam [Fri, 4 Dec 2009 13:53:05 +0000 (14:53 +0100)]
s3:dbwrap_ctdb: increase the number of commit retries 5-->100

This is to cope with timeouts when recoveries and transactions collide.
Maybe 100 is too hight, but 10 or even 20 have been too low in a
very busy environment.


13 years agos3:dbwrap_ctdb: increase the rsn of the __transaction_lock__ when storing
Michael Adam [Fri, 4 Dec 2009 13:49:53 +0000 (14:49 +0100)]
s3:dbwrap_ctdb: increase the rsn of the __transaction_lock__ when storing

so that it is correctly handled by recoveries.
Also set the dmaster explicitly.


13 years agos3: Fix explicit stat64 support
Volker Lendecke [Sat, 5 Dec 2009 13:27:04 +0000 (14:27 +0100)]
s3: Fix explicit stat64 support

13 years agos3: check if glibc has broken posix_fallocate
Björn Jacke [Fri, 4 Dec 2009 22:26:47 +0000 (23:26 +0100)]
s3: check if glibc has broken posix_fallocate

13 years agoMake Samba3 pass the RAW-LOCK test as Windows.
Jeremy Allison [Fri, 4 Dec 2009 22:04:08 +0000 (14:04 -0800)]
Make Samba3 pass the RAW-LOCK test as Windows.

Make smbd behave as Windows does in canceling locks.


13 years agoMake smbtorture4 match Windows behavior.
Jeremy Allison [Fri, 4 Dec 2009 22:02:52 +0000 (14:02 -0800)]
Make smbtorture4 match Windows behavior.


13 years agos3: bug #6967: Prevent glibc error on net ads join:
Jim McDonough [Fri, 4 Dec 2009 17:31:53 +0000 (12:31 -0500)]
s3: bug #6967: Prevent glibc error on net ads join:
talloc()ed memory should not be SAFE_FREE()ed.

Signed-off-by: Jim McDonough <>
13 years agos4-smbtorture: heavily expand printerdata tests in RPC-SPOOLSS.
Günther Deschner [Fri, 4 Dec 2009 01:54:33 +0000 (02:54 +0100)]
s4-smbtorture: heavily expand printerdata tests in RPC-SPOOLSS.

Including new tests for:
- spoolss_SetPrinterDataEx
- spoolss_DeletePrinterDataEx
- spoolss_DeletePrinterKey


13 years agos3 aclocal.m4: Fix iconv checks, clean up m4 code
Kai Blin [Fri, 4 Dec 2009 08:47:25 +0000 (09:47 +0100)]
s3 aclocal.m4: Fix iconv checks, clean up m4 code

The check for iconv requiring giconv.h and libgiconv as well as
the check for iconv requiring biconv.h and libbiconv were using the wrong
variable to check for previous successful test results. This caused the checks
to always fall back to libbiconv on systems where that library was available.

In the course of fixing this, I had to clean up the indentation in that piece of
code, and I also rewrote/added some comments.

Many thanks to Tsurutani Naoki <> for the initial
patch and diagnosis.

13 years agos4-ldb: fixed show_deleted module not to corrupt parse trees
Andrew Tridgell [Fri, 4 Dec 2009 06:46:14 +0000 (17:46 +1100)]
s4-ldb: fixed show_deleted module not to corrupt parse trees

The show_deleted module was using a static private ptr in the module
to hold a parse tree to save on parsing. The code caused this
static ptr to change with each search, which caused incorrect
searches and numerous valgrind errors.

This patch replaces it with a hand-built parse tree.

13 years agos4-drsutil: fixed a memory leak in samdb_search_count
Andrew Tridgell [Fri, 4 Dec 2009 06:45:38 +0000 (17:45 +1100)]
s4-drsutil: fixed a memory leak in samdb_search_count

In general functions that don't return any memory should not take a memory context.
Otherwise it is too easy to have a bug like this where memory is leaked

13 years agos4 torture: Convert create_complex_file to use BASIC_INFO instead of deprecated command
Zachary Loafman [Tue, 24 Nov 2009 23:42:56 +0000 (23:42 +0000)]
s4 torture: Convert create_complex_file to use BASIC_INFO instead of deprecated command

Signed-off-by: Tim Prouty <>
13 years agos4 torture: Fix call to smbcli_open
Zachary Loafman [Wed, 18 Nov 2009 22:42:31 +0000 (22:42 +0000)]
s4 torture: Fix call to smbcli_open

Signed-off-by: Tim Prouty <>
13 years agos4 torture: Make RAW-SEARCH pass against win7
Zachary Loafman [Wed, 18 Nov 2009 00:29:23 +0000 (00:29 +0000)]
s4 torture: Make RAW-SEARCH pass against win7

Signed-off-by: Tim Prouty <>
13 years agos4 torture: Parameterize WRITE_AND_CLOSE support
Zachary Loafman [Wed, 18 Nov 2009 00:29:21 +0000 (00:29 +0000)]
s4 torture: Parameterize WRITE_AND_CLOSE support

Signed-off-by: Tim Prouty <>
13 years agos4 torture: Add lockread_supported based off of CAP_LOCK_AND_READ
Zachary Loafman [Tue, 17 Nov 2009 23:25:58 +0000 (23:25 +0000)]
s4 torture: Add lockread_supported based off of CAP_LOCK_AND_READ

Signed-off-by: Tim Prouty <>
13 years agos4 torture: Allow READ to be parameterized, add more readx tests
Zachary Loafman [Mon, 16 Nov 2009 21:13:26 +0000 (21:13 +0000)]
s4 torture: Allow READ to be parameterized, add more readx tests

Signed-off-by: Tim Prouty <>
13 years agos4-torture: Allow the legacy DENY_FCB/DENY_DOS share modes to be ignored
Zachary Loafman [Mon, 9 Nov 2009 18:32:06 +0000 (18:32 +0000)]
s4-torture: Allow the legacy DENY_FCB/DENY_DOS share modes to be ignored

Signed-off-by: Tim Prouty <>
13 years agos4 selftest/quicktest: Update known failures for the sfileinfo and oplock changes
Tim Prouty [Fri, 4 Dec 2009 01:01:53 +0000 (17:01 -0800)]
s4 selftest/quicktest: Update known failures for the sfileinfo and oplock changes

Two new samba4 bugs have been filed for the two corresponding known
failures.  For the short term raw.sfileinfo had to be removed from
quicktest.  It is no longer an individual test, but a test-suite and
quicktest can only run top level tests.

13 years agos4 torture: Add a new RAW-OPLOCK test: BATCH26
Tim Prouty [Thu, 3 Dec 2009 21:46:11 +0000 (13:46 -0800)]
s4 torture: Add a new RAW-OPLOCK test: BATCH26

Try a rename with a wide-open share mode on an already open file
and the there is still share mode contention.  For the reason why

Msft's anser:

   After further review, The reason for server to fail with sharing
   violation is that the windows server that executes a path-based
   rename request opens the file for DELETE access, but only with
   FILE_SHARED_READ as ShareAccess .  Therefore, the existing
   open(frame 76), which has shared read/write/delete , is compatible
   with the Windows servers access mode (DELETE), but Windows servers
   open is not compatible with access mode in existing open.

   Note that it is correct to state that the logic in Windows server
   could have been written to allow shared read/write/delete in which
   case it would succeed as you mention. The behavior here is
   historical based on the existing implementation.

13 years agos4 torture: Close the third handle in RAW-OPLOCK-BATCH22
Tim Prouty [Thu, 3 Dec 2009 21:16:18 +0000 (13:16 -0800)]
s4 torture: Close the third handle in RAW-OPLOCK-BATCH22

13 years agos4 torture: Convert an error to a warning in RAW-OPLOCK-BATCH22
Tim Prouty [Thu, 3 Dec 2009 02:39:05 +0000 (18:39 -0800)]
s4 torture: Convert an error to a warning in RAW-OPLOCK-BATCH22

Some servers choose to mark a client as bad if they fail an oplock
break request by timing out (win7 is an example).  Once the client is
marked as bad, future oplock requests will timeout instantly.  This
causes subsequent runs of this test to fail, so rather than erroring
out as a failure, a warning is printed instead.

There is also a bug in w2k3 where it was incorrectly returning
contending a share mode lock.  It worked in XP and has been re-fixed
in win7.

This can also now be run against samba3.

13 years agos4 torture: Add a few more windows target types
Tim Prouty [Wed, 2 Dec 2009 00:27:10 +0000 (16:27 -0800)]
s4 torture: Add a few more windows target types

13 years agos4 torture: Convert to a more modern version of read in RAW-OPLOCK-BATCH4
Tim Prouty [Fri, 20 Nov 2009 19:25:43 +0000 (11:25 -0800)]
s4 torture: Convert to a more modern version of read in RAW-OPLOCK-BATCH4

13 years agos4 torture: Allow some implementations to break from level1 to none with two breaks
Tim Prouty [Tue, 1 Dec 2009 23:29:00 +0000 (15:29 -0800)]
s4 torture: Allow some implementations to break from level1 to none with two breaks

13 years agoMake sure we're not using the old smb_fname data.
Jeremy Allison [Fri, 4 Dec 2009 01:44:38 +0000 (17:44 -0800)]
Make sure we're not using the old smb_fname data.


13 years agoRemove smb_fname duplicates that just keep the same information as in fsp->fsp_name.
Jeremy Allison [Fri, 4 Dec 2009 00:45:35 +0000 (16:45 -0800)]
Remove smb_fname duplicates that just keep the same information as in fsp->fsp_name.

Moving towards making VFS_OPEN/VFS_MKDIR/VFS_RMDIR
handle based...

13 years agos3-spoolss: handle SEC_FLAG_MAXIMUM_ALLOWED in spoolss_OpenPrinterEx.
Günther Deschner [Thu, 3 Dec 2009 22:23:36 +0000 (23:23 +0100)]
s3-spoolss: handle SEC_FLAG_MAXIMUM_ALLOWED in spoolss_OpenPrinterEx.

In case someone (smbtorture4) requests SEC_FLAG_MAXIMUM_ALLOWED, translate it


13 years agos4-smbtorture: check more WERRORs in RPC-SPOOLSS.
Günther Deschner [Thu, 3 Dec 2009 21:38:38 +0000 (22:38 +0100)]
s4-smbtorture: check more WERRORs in RPC-SPOOLSS.


13 years agos4-smbtorture: call test_EnumPrinterDrivers with architecture in RPC-SPOOLSS.
Günther Deschner [Thu, 3 Dec 2009 21:05:35 +0000 (22:05 +0100)]
s4-smbtorture: call test_EnumPrinterDrivers with architecture in RPC-SPOOLSS.


13 years agoRefactor reply_rmdir to use handle based code. All
Jeremy Allison [Thu, 3 Dec 2009 21:01:10 +0000 (13:01 -0800)]
Refactor reply_rmdir to use handle based code. All
calls are now handle based. Put rmdir into close.c
and make it private.

13 years agos4:operational LDB module - Fix usage of LDB constants
Matthias Dieter Wallnöfer [Thu, 3 Dec 2009 11:50:18 +0000 (12:50 +0100)]
s4:operational LDB module - Fix usage of LDB constants

13 years agos4:ntvfs/posix/pvfs_streams - Fix "discard const" warning
Matthias Dieter Wallnöfer [Fri, 20 Nov 2009 11:00:28 +0000 (12:00 +0100)]
s4:ntvfs/posix/pvfs_streams - Fix "discard const" warning

I removed one "const" in front of a string declaration to achieve this.

Signed-off-by: Andrew Tridgell <>
13 years agos4:ntvfs/posix/pvfs_resolve - Fix "discard const" warning
Matthias Dieter Wallnöfer [Fri, 20 Nov 2009 10:56:16 +0000 (11:56 +0100)]
s4:ntvfs/posix/pvfs_resolve - Fix "discard const" warning

I removed two "const"s in front of string declarations to achieve this.

Signed-off-by: Andrew Tridgell <>
13 years agos3: Fix crash in winbindd;(bug#6879).
Bo Yang [Fri, 4 Dec 2009 04:53:38 +0000 (12:53 +0800)]
s3: Fix crash in winbindd;(bug#6879).

Signed-off-by: Bo Yang <>
13 years agoRemove unneeded argument from can_set_delete_on_close(). Ensure
Jeremy Allison [Thu, 3 Dec 2009 02:06:40 +0000 (18:06 -0800)]
Remove unneeded argument from can_set_delete_on_close(). Ensure
can_set_delete_on_close() is correctly called before any setting
of the disposition bit (clean up the do_unlink() call).

13 years agos4/selftest: knownfail some locking tests until bugs are fixed
Steven Danneman [Thu, 3 Dec 2009 01:55:00 +0000 (17:55 -0800)]
s4/selftest: knownfail some locking tests until bugs are fixed

13 years agoѕ3: remove superfluous option check
Björn Jacke [Thu, 3 Dec 2009 01:32:47 +0000 (02:32 +0100)]
ѕ3: remove superfluous option check

this function is only called when strict alloc is set, no reason to check that twice.

13 years agos4/torture: add multiple lock cancel test
Steven Danneman [Thu, 26 Nov 2009 01:39:42 +0000 (17:39 -0800)]
s4/torture: add multiple lock cancel test

See what happens when we have multiple outstanding lock requests and
we try to cancel both of them within a single LockingAndX.

On Windows, it seems only the first lock in the array is cancelled,
and the second is left pending.  Though, this behavior goes against
the MS-CIFS spec.

13 years agos4/torture: add addition multiple lock tests
Steven Danneman [Tue, 1 Dec 2009 01:05:27 +0000 (17:05 -0800)]
s4/torture: add addition multiple lock tests

* test that 2 locks in a single LockAndX are transactional
* test that 1 unlock and 1 lock in a single LockAndX are not
* test that SMB2 doesn't like mixed lock/unlock in a single

13 years agos4/torture: fix build warnings by removing unecessary const
Steven Danneman [Wed, 25 Nov 2009 02:38:46 +0000 (18:38 -0800)]
s4/torture: fix build warnings by removing unecessary const

13 years agos4/torture: Add target functionality parameters to SMBv1 BRL tests
Steven Danneman [Wed, 25 Nov 2009 00:58:25 +0000 (16:58 -0800)]
s4/torture: Add target functionality parameters to SMBv1 BRL tests

Abstract the server requirements to pass some BRL tests.

* The new default for >64bit lock tests, is that the server should
* Add parameter for targets that don't implement DENY_DOS

13 years agos3-selftest: run LOCAL-NDR when running make selftest.
Günther Deschner [Wed, 2 Dec 2009 23:46:47 +0000 (00:46 +0100)]
s3-selftest: run LOCAL-NDR when running make selftest.


13 years agos4-smbtorture: be very strict in checking spools_EnumPrinterKey results.
Günther Deschner [Wed, 2 Dec 2009 22:38:05 +0000 (23:38 +0100)]
s4-smbtorture: be very strict in checking spools_EnumPrinterKey results.


13 years agospoolss: hand-marshall the spoolss_StringArray2 struct for spoolss_EnumPrinterKey.
Günther Deschner [Wed, 2 Dec 2009 20:26:10 +0000 (21:26 +0100)]
spoolss: hand-marshall the spoolss_StringArray2 struct for spoolss_EnumPrinterKey.

This call is just driving me nuts :-)


13 years agos4-drs: fixed UDV and overlapping sync calls in DRS
Andrew Tridgell [Wed, 2 Dec 2009 22:19:55 +0000 (09:19 +1100)]
s4-drs: fixed UDV and overlapping sync calls in DRS

When windows abandons a DRS sync, it will sometimes re-use the same bind handle for
a new sync. This means we need to check the DN of the sync and blank the getnc_state
if the DN has changed.

This also fixes the UDV to use the highest uSN for the partition, not for
the whole SAM.

13 years agos4-drs: fixed updating of uSNChanged in replmd_modify
Andrew Tridgell [Wed, 2 Dec 2009 06:14:40 +0000 (17:14 +1100)]
s4-drs: fixed updating of uSNChanged in replmd_modify

Updating of uSNChanged broke in a recent change

13 years agos3:build: remove redundant qnx block size definition
Björn Jacke [Wed, 2 Dec 2009 23:12:53 +0000 (00:12 +0100)]
s3:build: remove redundant qnx block size definition

13 years agoRestructure the ACL code some more, get the internal semantics
Jeremy Allison [Wed, 2 Dec 2009 23:02:28 +0000 (15:02 -0800)]
Restructure the ACL code some more, get the internal semantics
right. The previous bugs were due to the fact that get_nt_acl_internal()
could return an NTSTATUS error if there was no stored ACL blob, but
otherwise would return the underlying ACL from the filysystem. Fix
this so it always returns a valid acl if it can, and if it does not
its an error to be reported back to the client. This then changes
the inherit acl code. Previously we were trying to match Windows
by setting a minimal ACL on a new file that didn't inherit anything
from a parent directory. This is silly - the returned ACL wouldn't
match the underlying UNIX permissions. The current code will correctly
inherit from a parent if a parent has any inheritable ACE entries
that apply to the new object, but will return a mapping from the
underlying UNIX permissions if the parent has no inheritable entries.
This makes much more sense for new files/directories.

13 years agos3:build: fix shared library build on QNX
Björn Jacke [Wed, 2 Dec 2009 23:02:44 +0000 (00:02 +0100)]
s3:build: fix shared library build on QNX

Fixes #6860. Thanks to Matt Kraai!

13 years agoEnsure check_parent_acl_common() only looks at stored
Jeremy Allison [Wed, 2 Dec 2009 20:29:16 +0000 (12:29 -0800)]
Ensure check_parent_acl_common() only looks at stored
blobs - returns NT_STATUS_OK if there aren't any.

13 years agos3: prefer posix_fallocate for doing "strict allocate"
Björn Jacke [Wed, 2 Dec 2009 14:13:37 +0000 (15:13 +0100)]
s3: prefer posix_fallocate for doing "strict allocate"

posix_fallocate is more efficient than manual zero'ing the file. When
preallocation in kernel space is supported it's extremely fast. Support for
preallocation at fs layer via posix_fallocate and fallocate at kernel site
can be found in Linux kernel 2.6.23/glibc 2.10 with ext4, XFS and OCFS2. Other
systems that I know of which support fast preallocation in kernel space are
AIX 6.1 with JFS2 and recent Solaris versions with ZFS maybe UFS2, too.

People who have a system with preallocation in kernel space might want to set
"strict allocate = yes". This reduces file fragentation and it's also safer for
setups with quota being turned on.

As of today most systems still don't have preallocation in kernel space, and
that's why "strict allocate = no" will stay the default for now.

13 years agoEnsure get_nt_acl_internal() only looks at the ACL blobs, not
Jeremy Allison [Wed, 2 Dec 2009 20:09:48 +0000 (12:09 -0800)]
Ensure get_nt_acl_internal() only looks at the ACL blobs, not
reads off the underlying filesystem. Ensure that vfs_acl_tdb.c
when it can't find a blob matching the file.

13 years agoFix bug #6837 - "Too many open files" when trying to access large number of files
Jeremy Allison [Wed, 2 Dec 2009 18:01:14 +0000 (10:01 -0800)]
Fix bug #6837 - "Too many open files" when trying to access large number of files
from Windows 7. Original patch by me fixed up with the correct open files number

13 years agosamba-spoolss: use spoolss_StringArray2 in spoolss_EnumPrinterKey.
Günther Deschner [Wed, 2 Dec 2009 13:25:31 +0000 (14:25 +0100)]
samba-spoolss: use spoolss_StringArray2 in spoolss_EnumPrinterKey.

This should finally resolve the endian issues we were seeing on sparc and is
much cleaner for spoolss clients and servers.


13 years agospoolss: add spoolss_StringArray2.
Günther Deschner [Wed, 2 Dec 2009 13:22:52 +0000 (14:22 +0100)]
spoolss: add spoolss_StringArray2.

The difference to spoolss_StringArray is that in spoolss_StringArray2 the string
array is put into a subcontext of _ndr_size.


13 years agos3-spoolss: move MAXDEVICENAME to spoolss idl.
Günther Deschner [Tue, 1 Dec 2009 01:38:19 +0000 (02:38 +0100)]
s3-spoolss: move MAXDEVICENAME to spoolss idl.


13 years agos3: Cope with older external libtdb
Volker Lendecke [Mon, 30 Nov 2009 23:01:36 +0000 (00:01 +0100)]
s3: Cope with older external libtdb

13 years agoRevert "s3-build: taise tdb version when building against system libtdb library."
Günther Deschner [Tue, 1 Dec 2009 23:22:18 +0000 (00:22 +0100)]
Revert "s3-build: taise tdb version when building against system libtdb library."

This reverts commit ea20678c55fee9f4586630cdb5fe7f35457d309a.

Just one new error code does probably not justify raising the required tdb version.


13 years agoEnsure we don't see the xattr used to store NT security (visible when xattr_tdb
Jeremy Allison [Tue, 1 Dec 2009 22:08:16 +0000 (14:08 -0800)]
Ensure we don't see the xattr used to store NT security (visible when xattr_tdb
is used). Allows make test to pass with prepended to the vfs modules.

13 years agos3: Fix smbd to correctly return INVALID_LEVEL on set_file_end_of_file_info for paths
Tim Prouty [Tue, 1 Dec 2009 18:47:08 +0000 (10:47 -0800)]
s3: Fix smbd to correctly return INVALID_LEVEL on set_file_end_of_file_info for paths

This allows smbd to pass the freshly updated RAW-SFILEINFO-END-OF-FILE
torture test.

13 years agos4 torture: Change oplock to use passthrough for exclusive3/batch11
Tim Prouty [Tue, 1 Dec 2009 00:43:00 +0000 (16:43 -0800)]
s4 torture: Change oplock to use passthrough for exclusive3/batch11

In light of the INVALID_LEVEL that is seen for RAW_SFILEINFO_END_OF_FILE_INFO
requests on a path, I'm changing these back to using the passthrough
RAW_SFILEINFO_END_OF_FILE_INFORMATION to test the oplock break behavior as
originally intended

13 years agos4 torture: Update RAW-SFILEINFO-END-OF-FILE to test some additional corner cases
Tim Prouty [Mon, 30 Nov 2009 19:59:19 +0000 (11:59 -0800)]
s4 torture: Update RAW-SFILEINFO-END-OF-FILE to test some additional corner cases

It turns out setting the end-of-file with Trans2SetPathInfo using the
snia spec's info level will attempt to open the file, enforcing share
modes, but then subsequentlys fail the setpathinfo with a dos error of
INVALID_LEVEL.  Doing a Trans2SetFileInfo with either end-of-file info
level succeeds as expected.

13 years agoRevert "s4 torture: Allow onefs to be checked like samba3 and samba4"
Tim Prouty [Wed, 25 Nov 2009 22:26:18 +0000 (14:26 -0800)]
Revert "s4 torture: Allow onefs to be checked like samba3 and samba4"

This reverts commit 98f595036e196dd61340fef0faf63ca762a25307.

No longer necessary

13 years agos4 torture: Change RAW-SFILEINFO-END-OF-FILE to check for share modes by default
Tim Prouty [Wed, 25 Nov 2009 22:40:54 +0000 (14:40 -0800)]
s4 torture: Change RAW-SFILEINFO-END-OF-FILE to check for share modes by default

Since the windows behavior appears to be a bug, only check for
the windows-style share mode bug if target=<windows variant> is

13 years agos4 torture: Move target macros to a common header instead of redefining them in multi...
Tim Prouty [Wed, 25 Nov 2009 22:38:55 +0000 (14:38 -0800)]
s4 torture: Move target macros to a common header instead of redefining them in multiple files