11 years agoutil: rewrite dlinklist.h so that DLIST_ADD_END() is O(1)
Andrew Tridgell [Wed, 10 Feb 2010 23:53:58 +0000 (15:53 -0800)]
util: rewrite dlinklist.h so that DLIST_ADD_END() is O(1)

This changes the meaning of the ->prev pointer in our doubly linked
lists to point at the end of the list from the front of the list. That
allows us to implement DLIST_ADD_END() and related functions in O(1)
time, which can be a huge saving in many places in Samba.

This also means that the 'type' argument to various DLIST_*() macros
is no longer needed, but I have left it in for now to keep the
patchset small, which will make it easier to revert if any problems
are found. In the future we should remove the 'type' arguments.

(jra. Move the one use of DLIST_TAIL over to the new macros).

11 years agos3-smbd: update to use new DLIST macros
Andrew Tridgell [Sat, 6 Feb 2010 01:42:29 +0000 (12:42 +1100)]
s3-smbd: update to use new DLIST macros
(cherry picked from commit 365b408c458c848a818637d9b36a0423aeb1ba54)

11 years agos3-registry: update to use new DLIST macros
Andrew Tridgell [Sat, 6 Feb 2010 01:41:57 +0000 (12:41 +1100)]
s3-registry: update to use new DLIST macros
(cherry picked from commit 3437713ad7e5bccafde30553a8232119fd2a9eb9)

11 years agos3-perfcount: update to use new DLIST macros
Andrew Tridgell [Sat, 6 Feb 2010 01:41:39 +0000 (12:41 +1100)]
s3-perfcount: update to use new DLIST macros
(cherry picked from commit a13b507f2d8be7f90c8872094cd0732926a6fcbb)

11 years agos3-locking: update to use DLIST_ADD_AFTER()
Andrew Tridgell [Sat, 6 Feb 2010 01:41:07 +0000 (12:41 +1100)]
s3-locking: update to use DLIST_ADD_AFTER()
(cherry picked from commit 6c6df527e14514027cbcaa6deac25adf04363926)

11 years agos3-libsmb: update libsmb to use new DLIST macros
Andrew Tridgell [Sat, 6 Feb 2010 01:40:38 +0000 (12:40 +1100)]
s3-libsmb: update libsmb to use new DLIST macros

manipulating p->prev directly is not safe any more
(cherry picked from commit 3c650ac1e3e1cdbbabecfddcd29325f20b5dcb48)

11 years agos3-memcache: update memcache to use new DLIST macros
Andrew Tridgell [Sat, 6 Feb 2010 01:40:03 +0000 (12:40 +1100)]
s3-memcache: update memcache to use new DLIST macros

we don't need a separate lru pointer any more
(cherry picked from commit 4ffd7aca3e38728077bd80c2a65c4efbcfd216fc)

11 years agos3-ldb: update the old ldb in s3 to use new DLIST macros
Andrew Tridgell [Sat, 6 Feb 2010 01:38:47 +0000 (12:38 +1100)]
s3-ldb: update the old ldb in s3 to use new DLIST macros
(cherry picked from commit a7d8bfd373392eecf4fff33d39b85e1b55ad901d)

11 years agos3-nmbd: update nmbd to use new DLIST_ macros
Andrew Tridgell [Sat, 6 Feb 2010 01:38:24 +0000 (12:38 +1100)]
s3-nmbd: update nmbd to use new DLIST_ macros
(cherry picked from commit 4d23d777bc6d4fad20d0f3084fe658635812bee9)

11 years agos4-ldb: update ldb_tdb to use new DLIST_ macros
Andrew Tridgell [Sat, 6 Feb 2010 01:37:48 +0000 (12:37 +1100)]
s4-ldb: update ldb_tdb to use new DLIST_ macros
(cherry picked from commit 6ede911654566a7e21ded70dacac68df3b6a1bca)

11 years agos4-dsdb: update repl_meta_data.c to use new DLIST_ macros
Jeremy Allison [Wed, 10 Feb 2010 23:35:38 +0000 (15:35 -0800)]
s4-dsdb: update repl_meta_data.c to use new DLIST_ macros

11 years agoTemporary changes to dlinklist to keep the implementation static whilst
Jeremy Allison [Wed, 10 Feb 2010 23:30:59 +0000 (15:30 -0800)]
Temporary changes to dlinklist to keep the implementation static whilst
uses of (list)->prev are moved over to DLIST_PREV. This will be replaced
when the final (new) version of the dlinklist.h header is added.


11 years agoRevert "Change the default of "nmbd bind explicit broadcast" to "no""
Jeremy Allison [Wed, 10 Feb 2010 20:42:27 +0000 (12:42 -0800)]
Revert "Change the default of "nmbd bind explicit broadcast" to "no""

This reverts commit 84fba3c1bc962804259f201d465acfdf0cd3c6a8.

Now we have a "processed packet queue" in nmbd we can go back
to doing this by default.


11 years agoMore of the fix for bug #7118 - nmbd problems with socket address.
Jeremy Allison [Wed, 10 Feb 2010 20:32:05 +0000 (12:32 -0800)]
More of the fix for bug #7118 - nmbd problems with socket address.

Add a simple "processed packet queue" cache to stop nmbd responding to
packets received on the broadcast and non-broadcast socket (which
it has opened when "nmbd bind explicit broadcast = yes").

This is a very simple packet queue - it only keeps the packets
processed during a single call to listen_for_packets() (i.e. one
select call). This means that if the delivery notification for a
packet received on both broadcast and non-broadcast addresses
is done in two different select calls, the packet will still be
processed twice. This is a very rare occurrance and we can just
live with it when it does as the protocol is stateless. If this
is ever flagged as a repeatable problem then we can add a longer
lived cache, using timeout processing to clear etc. etc. But without
storing all packets processed we can never be *sure* we've eliminated
the race condition so I'm going to go with this simple solution until
someone proves a more complex one is needed :-).


11 years agoChange the default of "nmbd bind explicit broadcast" to "no"
Jeremy Allison [Wed, 10 Feb 2010 16:39:37 +0000 (08:39 -0800)]
Change the default of "nmbd bind explicit broadcast" to "no"
until the double processing problem in bug #7118 is fixed.


11 years agotestprogs: add more error codes for spoolss test.
Günther Deschner [Wed, 10 Feb 2010 12:17:36 +0000 (13:17 +0100)]
testprogs: add more error codes for spoolss test.


11 years agos3: change ldap filter to what really was intended
Björn Jacke [Wed, 10 Feb 2010 12:48:11 +0000 (13:48 +0100)]
s3: change ldap filter to what really was intended

11 years agotdb: fix recovery reuse after crash
Rusty Russell [Thu, 4 Feb 2010 13:29:54 +0000 (23:59 +1030)]
tdb: fix recovery reuse after crash

If a process (or the machine) dies after just after writing the
recovery head (pointing at the end of file), the recovery record will filled
with 0x42.  This will not invoke a recovery on open, since rec.magic

Unfortunately, the first transaction commit will happily reuse that
area: tdb_recovery_allocate() doesn't check the magic.  The recovery
record has length 0x42424242, and it writes that back into the
now-valid-looking transaction header) for the next comer (which
happens to be tdb_wipe_all in my tests).

Signed-off-by: Rusty Russell <>
11 years agotdb: give a name to the invalid recovery area constant (0)
Rusty Russell [Thu, 4 Feb 2010 13:30:24 +0000 (00:00 +1030)]
tdb: give a name to the invalid recovery area constant (0)

Signed-off-by: Rusty Russell <>
11 years agos4:provision Just 'do the right thing' with empty smb.conf files
Andrew Bartlett [Wed, 10 Feb 2010 02:56:24 +0000 (13:56 +1100)]
s4:provision Just 'do the right thing' with empty smb.conf files

For some reason, JHT keeps on creating an empty smb.conf file,
expecting it to be the same as a non-existant one.  It is easier to
just realise what he meant.

Andrew Bartlett

11 years agos4-smbtorture: use test_GetPrinter_level in RPC-SPOOLSS-NOTIFY.
Günther Deschner [Tue, 9 Feb 2010 23:43:51 +0000 (00:43 +0100)]
s4-smbtorture: use test_GetPrinter_level in RPC-SPOOLSS-NOTIFY.


11 years agoFix bug #7122 - Reading a large browselist fails (server returns invalid values in...
Jeremy Allison [Tue, 9 Feb 2010 23:14:38 +0000 (15:14 -0800)]
Fix bug #7122 - Reading a large browselist fails (server returns invalid values in subsequent SMBtrans replies)

There are two problems:

1). The server is off-by-one in the end of buffer space test.
2). The server returns 0 in the totaldata (smb_vwv1) and totalparams (smb_vwv0)
fields in the second and subsequent SMBtrans replies.

This patch fixes both.


11 years agoMissed one check on the memcpy for bug #7063.
Jeremy Allison [Tue, 9 Feb 2010 22:56:12 +0000 (14:56 -0800)]
Missed one check on the memcpy for bug #7063.


11 years agoSecond part of fix for bug 7063 - Samba 3.4.5 on ubuntu 8.04 64 bit - Core dumps.
Jeremy Allison [Tue, 9 Feb 2010 22:48:15 +0000 (14:48 -0800)]
Second part of fix for bug 7063 - Samba 3.4.5 on ubuntu 8.04 64 bit - Core dumps.

Ensure we have no naked memcpy calls. This isn't a crash bug (it's
already checked in the data_blob_talloc_zero() above, but I want to
get into the pattern of having all memcpy's covered by safety checks.


11 years agoFix off-by-one error in working out the limit of the NetServerEnum comment.
Jeremy Allison [Tue, 9 Feb 2010 20:17:08 +0000 (12:17 -0800)]
Fix off-by-one error in working out the limit of the NetServerEnum comment.


11 years agos3:smbd: use StrCaseCmp() instead of strcasecmp
Stefan Metzmacher [Tue, 9 Feb 2010 17:58:36 +0000 (18:58 +0100)]
s3:smbd: use StrCaseCmp() instead of strcasecmp


11 years agos3:smbd: Fix really ugly bool vs. int bug!!!
Stefan Metzmacher [Tue, 9 Feb 2010 17:54:41 +0000 (18:54 +0100)]
s3:smbd: Fix really ugly bool vs. int bug!!!

A comparison function for qsort needs to return an 'int'!
Otherwise you'll get random results depending on the compiler
and the architecture...


11 years agocheck_python.m4 - Raise the minimum python version up to 2.4
Matthias Dieter Wallnöfer [Tue, 9 Feb 2010 16:48:44 +0000 (17:48 +0100)]
check_python.m4 - Raise the minimum python version up to 2.4

This is needed since we at the s4 side have some code which requires this. I
think everybody should be fine since we got no complaints on the mailing list
about this change.

Patch template: Jelmer Vernooij <>

11 years agopidl - Remove "Py_RETURN_NONE" compatibility code
Matthias Dieter Wallnöfer [Thu, 4 Feb 2010 10:25:06 +0000 (11:25 +0100)]
pidl - Remove "Py_RETURN_NONE" compatibility code

This was needed only by Python 2.3 which we no longer support.

11 years agos4:Remove "Py_RETURN_NONE" compatibility code
Matthias Dieter Wallnöfer [Thu, 4 Feb 2010 10:17:40 +0000 (11:17 +0100)]
s4:Remove "Py_RETURN_NONE" compatibility code

This was needed only by Python 2.3 which we no longer support.

11 years agos3-docs: Fix make manpages3.
Karolin Seeger [Tue, 9 Feb 2010 16:21:01 +0000 (17:21 +0100)]
s3-docs: Fix make manpages3.


11 years agos3-docs: Add missing para end tag.
Karolin Seeger [Tue, 9 Feb 2010 15:36:39 +0000 (16:36 +0100)]
s3-docs: Add missing para end tag.

(cherry picked from commit b78de63ef3cde53e3aabbe46654aac5a335f16a8)

11 years agotestprogs: regenerate printlib prototypes for spoolss test.
Günther Deschner [Tue, 9 Feb 2010 15:16:54 +0000 (16:16 +0100)]
testprogs: regenerate printlib prototypes for spoolss test.


11 years agotestprogs: print job info levels in EnumJobs() spoolss tests.
Günther Deschner [Tue, 9 Feb 2010 15:14:45 +0000 (16:14 +0100)]
testprogs: print job info levels in EnumJobs() spoolss tests.


11 years agotestprogs: print datatypes info levels in EnumPrintProcessorDatatypes() spoolss tests.
Günther Deschner [Tue, 9 Feb 2010 14:59:29 +0000 (15:59 +0100)]
testprogs: print datatypes info levels in EnumPrintProcessorDatatypes() spoolss tests.


11 years agotestprogs: print printprocessor info levels in EnumPrintProcessors() spoolss tests.
Günther Deschner [Tue, 9 Feb 2010 14:54:25 +0000 (15:54 +0100)]
testprogs: print printprocessor info levels in EnumPrintProcessors() spoolss tests.


11 years agotestprogs: print monitor info levels in EnumMonitors() spoolss tests.
Günther Deschner [Tue, 9 Feb 2010 14:24:34 +0000 (15:24 +0100)]
testprogs: print monitor info levels in EnumMonitors() spoolss tests.


11 years agotestprogs: print port info levels in EnumPorts() spoolss tests.
Günther Deschner [Tue, 9 Feb 2010 14:00:54 +0000 (15:00 +0100)]
testprogs: print port info levels in EnumPorts() spoolss tests.


11 years agotestprogs: print form info levels in EnumForms(),GetForm() spoolss tests.
Günther Deschner [Tue, 9 Feb 2010 13:49:04 +0000 (14:49 +0100)]
testprogs: print form info levels in EnumForms(),GetForm() spoolss tests.


11 years agos3:selftest: make selftest requires bash for now
Stefan Metzmacher [Tue, 9 Feb 2010 11:58:29 +0000 (12:58 +0100)]
s3:selftest: make selftest requires bash for now

So force bash until we removed the dependency to bash.


11 years agos3:passdb: only use gid_to_sid() result if the result is a group of our local sam
Stefan Metzmacher [Tue, 9 Feb 2010 11:57:01 +0000 (12:57 +0100)]
s3:passdb: only use gid_to_sid() result if the result is a group of our local sam

Otherwise retry with pdb_gid_to_sid().


11 years agos3: fix some wrong newlines in de translation strings
Björn Jacke [Tue, 9 Feb 2010 08:07:54 +0000 (09:07 +0100)]
s3: fix some wrong newlines in de translation strings

11 years agovfs_catia: fix return type warnings
Björn Jacke [Tue, 9 Feb 2010 08:09:57 +0000 (09:09 +0100)]
vfs_catia: fix return type warnings

11 years agos4-torture: allow host-only in unc lists in smbtorture
Andrew Tridgell [Tue, 9 Feb 2010 02:18:31 +0000 (13:18 +1100)]
s4-torture: allow host-only in unc lists in smbtorture

Allow UNC lists like this:

the share name will be taken from the command line

11 years agoMake "nmbd bind explicit broadcast" on by default.
Jeremy Allison [Mon, 8 Feb 2010 22:35:02 +0000 (14:35 -0800)]
Make "nmbd bind explicit broadcast" on by default.

Fix a comment typo.


11 years agoFix bug #6876 for acl_tdb module.
Jeremy Allison [Mon, 8 Feb 2010 19:04:38 +0000 (11:04 -0800)]
Fix bug #6876 for acl_tdb module.

As pointed out by, the rmdir module initializer was
duplicated. Fix this properly.


11 years agos3:libsmb: fix NetServerEnum3 rap calls.
Stefan Metzmacher [Mon, 8 Feb 2010 17:38:03 +0000 (18:38 +0100)]
s3:libsmb: fix NetServerEnum3 rap calls.


11 years agos3:smbd: implement api_RNetServerEnum3
Stefan Metzmacher [Fri, 5 Feb 2010 17:08:46 +0000 (18:08 +0100)]
s3:smbd: implement api_RNetServerEnum3

This is needed to support large browse lists.


11 years agos3:smbd: add/improve some DEBUG messages in api_RNetServerEnum2()
Stefan Metzmacher [Mon, 8 Feb 2010 17:45:18 +0000 (18:45 +0100)]
s3:smbd: add/improve some DEBUG messages in api_RNetServerEnum2()


11 years agos3:smbd: rename api_RNetServerEnum => api_RNetServerEnum2
Stefan Metzmacher [Fri, 5 Feb 2010 15:55:15 +0000 (16:55 +0100)]
s3:smbd: rename api_RNetServerEnum => api_RNetServerEnum2


11 years agos3:docs-xml: document "nmbd bind explicit broadcast"
Stefan Metzmacher [Mon, 8 Feb 2010 11:59:13 +0000 (12:59 +0100)]
s3:docs-xml: document "nmbd bind explicit broadcast"


11 years agos3:nmbd: change "nmbd:bind explicit broadcast" into "nmbd bind explicit broadcast"
Stefan Metzmacher [Mon, 8 Feb 2010 11:51:29 +0000 (12:51 +0100)]
s3:nmbd: change "nmbd:bind explicit broadcast" into "nmbd bind explicit broadcast"


11 years agos3:nmbd: also listen explicit on the subnet broadcast addresses
Stefan Metzmacher [Thu, 28 Jan 2010 10:04:05 +0000 (11:04 +0100)]
s3:nmbd: also listen explicit on the subnet broadcast addresses

And send replies always via the unicast address of the subnet.

This behavior is off by default (as before)
and can be enabled with "nmbd:bind explicit broadcast = yes".


11 years agorelease-scripts: parametrize scripts
Simo Sorce [Mon, 8 Feb 2010 15:25:13 +0000 (10:25 -0500)]
release-scripts: parametrize scripts

This should make it easier to keep all release scripts alined as it will reduce
the difference between them to ideally a few variables

Also moves the tdb script in the scripts directory.

11 years agotalloc: Fix abi checks in release script
Simo Sorce [Mon, 8 Feb 2010 15:05:24 +0000 (10:05 -0500)]
talloc: Fix abi checks in release script

We must perform abi checks against the version we are going to release.
Not against the current tree we are in.

11 years agotalloc: Fix abi checks and ifdefs
Simo Sorce [Mon, 8 Feb 2010 15:02:49 +0000 (10:02 -0500)]
talloc: Fix abi checks and ifdefs

teach the abi check scripts to skip the DOXYGEN sections
fix the header to use #ifdef DOXYGEN and not #if DOXYGEN

11 years agos4-smbtorture: skip over ValidatePassword if DCERPC error is returned.
Günther Deschner [Wed, 9 Dec 2009 10:21:08 +0000 (11:21 +0100)]
s4-smbtorture: skip over ValidatePassword if DCERPC error is returned.


11 years agos4:subtree_delete - "NULL" as format string isn't allowed on FreeBSD
Matthias Dieter Wallnöfer [Mon, 8 Feb 2010 11:08:36 +0000 (12:08 +0100)]
s4:subtree_delete - "NULL" as format string isn't allowed on FreeBSD

I changed the format string into "(objectClass=*)" which should be an
equivalent expression for choosing all available objects.

Consider bug 7115 for the issue.

11 years agos3-net: fix net ads dns usage calls.
Günther Deschner [Mon, 8 Feb 2010 10:53:59 +0000 (11:53 +0100)]
s3-net: fix net ads dns usage calls.

Bjoern, please check.


11 years agos3:pdb_ldap: don't search for the users primary group, if we already know it
Stefan Metzmacher [Fri, 5 Feb 2010 15:20:21 +0000 (16:20 +0100)]
s3:pdb_ldap: don't search for the users primary group, if we already know it


11 years agos3:pdb_ldap: optimize ldapsam_alias_memberships() and cache ldap searches.
Stefan Metzmacher [Wed, 3 Feb 2010 10:32:41 +0000 (11:32 +0100)]
s3:pdb_ldap: optimize ldapsam_alias_memberships() and cache ldap searches.

ldapsam_alias_memberships() does the same LDAP search twice, triggered
via add_aliases() from create_local_nt_token().

This happens when no domain aliases are used.


11 years agos3:pdb_ldap: try to build the full unix_pw structure with ldapsam:trusted support
Stefan Metzmacher [Thu, 4 Feb 2010 16:19:57 +0000 (17:19 +0100)]
s3:pdb_ldap: try to build the full unix_pw structure with ldapsam:trusted support

And also store the gid_to_sid mappings in the idmap_cache.


11 years agos3:passdb: speed up pdb_get_group_sid()
Stefan Metzmacher [Thu, 4 Feb 2010 16:16:59 +0000 (17:16 +0100)]
s3:passdb: speed up pdb_get_group_sid()

Use the cached version gid_to_sid() instead
of pdb_gid_to_sid().

And also avoid the expensive lookup_sid() call
for wellkown domain groups.


11 years agos3: Make pdb_copy_sam_account also copy the group sid
Volker Lendecke [Sun, 31 Jan 2010 14:18:51 +0000 (15:18 +0100)]
s3: Make pdb_copy_sam_account also copy the group sid

Signed-off-by: Stefan Metzmacher <>
11 years agos3: shortcut gid_to_sid when "ldapsam:trusted = yes"
Stefan Metzmacher [Sat, 30 Jan 2010 21:28:19 +0000 (22:28 +0100)]
s3: shortcut gid_to_sid when "ldapsam:trusted = yes"

The normal gid_to_sid behaviour is to call sys_getgrgid()
to get the name for the given gid and then call the
getsamgrnam passdb method for the resulting name.

In the ldapsam:trusted case we can reduce the gid_to_sid
operation to one simple search for the gidNumber attribute
and only get the sambaSID attribute from the correspoinding
LDAP object. This reduces the number of ldap roundtrips
for this operation.


11 years agos4:heimdal: regerenate files
Stefan Metzmacher [Mon, 8 Feb 2010 08:57:31 +0000 (09:57 +0100)]
s4:heimdal: regerenate files

Andrew using cp like in commit ca12e7bc8ff4a91f2044c0a60550fec902e97a78
is wrong as that removes #include "config.h" and breaks the build on AIX.


11 years agos4:heimdal_build: allow flex-2.5.35:bison-2.4.1 to regenerate files
Stefan Metzmacher [Mon, 8 Feb 2010 08:29:22 +0000 (09:29 +0100)]
s4:heimdal_build: allow flex-2.5.35:bison-2.4.1 to regenerate files


11 years agolibreplace: add check for NSS_WRAPPER_DISABLE similar to SOCKET_WRAPPER_DISABLE
Stefan Metzmacher [Mon, 8 Feb 2010 07:55:44 +0000 (08:55 +0100)]
libreplace: add check for NSS_WRAPPER_DISABLE similar to SOCKET_WRAPPER_DISABLE


11 years agolibreplace: Attempt to fix bug 5910, detect a broken getaddrinfo
Volker Lendecke [Sun, 7 Feb 2010 17:09:11 +0000 (18:09 +0100)]
libreplace: Attempt to fix bug 5910, detect a broken getaddrinfo

Signed-off-by: Stefan Metzmacher <>
11 years agos3: Don't invalidate cache for uninitialized domains.
Bo Yang [Tue, 9 Feb 2010 08:35:40 +0000 (16:35 +0800)]
s3: Don't invalidate cache for uninitialized domains.

Signed-off-by: Bo Yang <>
11 years agos3: signals are processed twice in child.
Bo Yang [Tue, 9 Feb 2010 09:02:20 +0000 (17:02 +0800)]
s3: signals are processed twice in child.

Signed-off-by: Bo Yang <>
11 years agos3: quiet the unresolved symbol warnings of the Compaq compiler when linking our...
Björn Jacke [Sun, 7 Feb 2010 23:08:38 +0000 (00:08 +0100)]
s3: quiet the unresolved symbol warnings of the Compaq compiler when linking our vfs modules

11 years agos4-torture: more useful error message in SMB2-DIR
Andrew Tridgell [Sun, 7 Feb 2010 10:12:14 +0000 (21:12 +1100)]
s4-torture: more useful error message in SMB2-DIR

11 years agonbt: don't reference the event_ctx in nbtsock
Andrew Tridgell [Sun, 7 Feb 2010 10:08:56 +0000 (21:08 +1100)]
nbt: don't reference the event_ctx in nbtsock

This causes talloc_free with references errors

11 years agotalloc: fixed doc typo
Andrew Tridgell [Sun, 7 Feb 2010 23:58:48 +0000 (10:58 +1100)]
talloc: fixed doc typo

11 years agoMove the talloc details to the mainpage.
Andreas Schneider [Sun, 7 Feb 2010 23:40:07 +0000 (00:40 +0100)]
Move the talloc details to the mainpage.

Signed-off-by: Andrew Tridgell <>
11 years agotalloc: Added doxygen config file.
Andreas Schneider [Sun, 17 Jan 2010 10:25:34 +0000 (11:25 +0100)]
talloc: Added doxygen config file.

Signed-off-by: Andrew Tridgell <>
11 years agotalloc: Documented talloc with doxygen.
Andreas Schneider [Sun, 17 Jan 2010 10:25:02 +0000 (11:25 +0100)]
talloc: Documented talloc with doxygen.

Signed-off-by: Andrew Tridgell <>
11 years agos3: fix build issue on Tru64
Björn Jacke [Sun, 7 Feb 2010 19:10:57 +0000 (20:10 +0100)]
s3: fix build issue on Tru64

Thanks, Volker for the hint - acl_type is a macro on Tru64. Renamed it
to acltype. This fixes #7103.

11 years agos3: Fix some DEBUG statements
Volker Lendecke [Sun, 7 Feb 2010 19:55:13 +0000 (20:55 +0100)]
s3: Fix some DEBUG statements

11 years agos3: Make cli_get_fs_volume_info() use cli_trans()
Volker Lendecke [Sun, 7 Feb 2010 11:08:39 +0000 (12:08 +0100)]
s3: Make cli_get_fs_volume_info() use cli_trans()

11 years agos3: Remove some unused code
Volker Lendecke [Sun, 7 Feb 2010 11:01:36 +0000 (12:01 +0100)]
s3: Remove some unused code

11 years agos3: Fix some comments
Volker Lendecke [Sun, 7 Feb 2010 10:14:22 +0000 (11:14 +0100)]
s3: Fix some comments

11 years agos3: Fix some nonempty blank lines
Volker Lendecke [Sun, 7 Feb 2010 10:05:07 +0000 (11:05 +0100)]
s3: Fix some nonempty blank lines

11 years agolibwbclient: wbc_create_logon_info is always called with mem_ctx==NULL
Volker Lendecke [Sat, 6 Feb 2010 15:56:29 +0000 (16:56 +0100)]
libwbclient: wbc_create_logon_info is always called with mem_ctx==NULL

11 years agolibwbclient: wbc_create_password_policy_info is always called with mem_ctx==NULL
Volker Lendecke [Sat, 6 Feb 2010 15:54:05 +0000 (16:54 +0100)]
libwbclient: wbc_create_password_policy_info is always called with mem_ctx==NULL

11 years agolibwbclient: wbc_create_error_info is always called with mem_ctx==NULL
Volker Lendecke [Sat, 6 Feb 2010 15:53:07 +0000 (16:53 +0100)]
libwbclient: wbc_create_error_info is always called with mem_ctx==NULL

11 years agos4-registry: fixed byte order assumptions
Andrew Tridgell [Sun, 7 Feb 2010 07:11:42 +0000 (18:11 +1100)]
s4-registry: fixed byte order assumptions

the registry tests were broken on big-endian systems

11 years agotdb: raise version to 1.2.1 tdb-1.2.1
Simo Sorce [Sat, 6 Feb 2010 15:48:39 +0000 (10:48 -0500)]
tdb: raise version to 1.2.1

after recent fixes we need to raise the version to 1.2.1 so that
we can require also the right patched version.

11 years agopam_winbind: Remove a nested if-statement. Bo, please check!
Volker Lendecke [Sat, 6 Feb 2010 11:56:19 +0000 (12:56 +0100)]
pam_winbind: Remove a nested if-statement. Bo, please check!

11 years agopam_winbind: Use strchr instead of strstr for a single character
Volker Lendecke [Sat, 6 Feb 2010 11:55:00 +0000 (12:55 +0100)]
pam_winbind: Use strchr instead of strstr for a single character

11 years agoFix trailing whitespace errors I added (sorry).
Jeremy Allison [Sat, 6 Feb 2010 06:51:11 +0000 (22:51 -0800)]
Fix trailing whitespace errors I added (sorry).


11 years agos3: Fix malformed require_membership_of_sid.
Bo Yang [Sun, 7 Feb 2010 06:45:42 +0000 (14:45 +0800)]
s3: Fix malformed require_membership_of_sid.

Signed-off-by: Bo Yang <>
11 years agos3-events: make the old timed events compatible with tevent
Andrew Tridgell [Sat, 6 Feb 2010 03:14:45 +0000 (19:14 -0800)]
s3-events: make the old timed events compatible with tevent

tevent ensures that a timed event is only called once. The old events
code relied on the called handler removing the event itself. If the
handler removed the event after calling a function which invoked the
event loop then the timed event could loop forever.

This change makes the two timed event systems more compatible, by
allowing the handler to free the te if it wants to, but ensuring it is
off the linked list of events before the handler is called, and
ensuring it is freed even if the handler doesn't free it.

11 years agos3-smbd: add a rate limited cleanup of brl, connections and locking db
Andrew Tridgell [Sat, 6 Feb 2010 05:08:56 +0000 (21:08 -0800)]
s3-smbd: add a rate limited cleanup of brl, connections and locking db

On unclean shutdown we can end up with stale entries in the brlock,
connections and locking db. Previously we would do the cleanup on
every unclean exit, but that can cause smbd to be completely
unavailable for several minutes when a large number of child smbd
processes exit.

This adds a rate limited cleanup of the databases, with the default
that cleanup happens at most every 20s

11 years agos3-brlock: we don't need these MSG_SMB_UNLOCK calls now
Andrew Tridgell [Sat, 6 Feb 2010 05:02:24 +0000 (21:02 -0800)]
s3-brlock: we don't need these MSG_SMB_UNLOCK calls now

These have been replaced with the min timeout in blocking.c

11 years agos3-brlock: add a minimim retry time for pending blocking locks
Andrew Tridgell [Sat, 6 Feb 2010 04:59:43 +0000 (20:59 -0800)]
s3-brlock: add a minimim retry time for pending blocking locks

When we are waiting on a pending byte range lock, another smbd might
exit uncleanly, and therefore not notify us of the removal of the
lock, and thus not trigger the lock to be retried.

We coped with this up to now by adding a message_send_all() in the
SIGCHLD and cluster reconfigure handlers to send a MSG_SMB_UNLOCK to
all smbd processes. That would generate O(N^2) work when a large
number of clients disconnected at once (such as on a network outage),
which could leave the whole system unusable for a very long time (many
minutes, or even longer).

By adding a minimum re-check time for pending byte range locks we
avoid this problem by ensuring that pending locks are retried at a
more regular interval.

11 years agos4-ldb: fixed selftest
Andrew Tridgell [Sat, 6 Feb 2010 03:26:22 +0000 (14:26 +1100)]
s4-ldb: fixed selftest

subunitrun seemed to be calling a test from its own source, instead of
the samba test. Fix this by calling the test directly.

11 years agoRemove now unused variable.
Jeremy Allison [Sat, 6 Feb 2010 00:20:34 +0000 (16:20 -0800)]
Remove now unused variable.

11 years agoFix bug 7104 - "wide links" and "unix extensions" are incompatible.
Jeremy Allison [Fri, 5 Feb 2010 23:20:18 +0000 (15:20 -0800)]
Fix bug 7104 - "wide links" and "unix extensions" are incompatible.

Change parameter "wide links" to default to "no".
Ensure "wide links = no" if "unix extensions = yes" on a share.
Fix man pages to refect this.

Remove "within share" checks for a UNIX symlink set - even if
widelinks = no. The server will not follow that link anyway.

Correct DEBUG message in check_reduced_name() to add missing "\n"
so it's really clear when a path is being denied as it's outside
the enclosing share path.