13 years agor5366: added initial code for adding user accounts to libnet makefile.
Rafal Szczesniak [Sun, 13 Feb 2005 00:50:38 +0000 (00:50 +0000)]
r5366: added initial code for adding user accounts to libnet makefile.
It doesn't do anything useful yet, but at least doesn't break
the build.


13 years agor5365: Initial code for adding user accounts via libnet call.
Rafal Szczesniak [Sun, 13 Feb 2005 00:48:21 +0000 (00:48 +0000)]
r5365: Initial code for adding user accounts via libnet call.


13 years agor5364: Rename string fields called 'domain' and 'name' to be 'domain_name'.
Tim Potter [Sun, 13 Feb 2005 00:26:43 +0000 (00:26 +0000)]
r5364: Rename string fields called 'domain' and 'name' to be 'domain_name'.

13 years agor5363: - ndr_%_ptr -> ndr_%_unique_ptr
Jelmer Vernooij [Sun, 13 Feb 2005 00:07:17 +0000 (00:07 +0000)]
r5363: - ndr_%_ptr -> ndr_%_unique_ptr
- ndr_%_relative% -> ndr_%_relative_ptr%
- Change pointer_default() default to "unique"
  (DCE uses "ptr" as default, MIDL doesn't follow the standard and uses "unique")

13 years agor5362: Add pointer_default() support to pidl. pointer_default()
Jelmer Vernooij [Sat, 12 Feb 2005 23:03:26 +0000 (23:03 +0000)]
r5362: Add pointer_default() support to pidl. pointer_default()
is assumed to be "ptr" if not specified (just like midl).

The validator will warn when "ptr" is used at the moment, because
pidl only supports unique, ref and relative at the moment.

13 years agor5361: Rename some functions, add tests to validator.
Jelmer Vernooij [Sat, 12 Feb 2005 21:43:08 +0000 (21:43 +0000)]
r5361: Rename some functions, add tests to validator.

13 years agor5360: Remove a couple of unused functions.
Jelmer Vernooij [Sat, 12 Feb 2005 19:28:23 +0000 (19:28 +0000)]
r5360: Remove a couple of unused functions.

13 years agor5358: - added initial WINS server code. It passes most of the NBT-WINS test, but...
Andrew Tridgell [Sat, 12 Feb 2005 11:33:42 +0000 (11:33 +0000)]
r5358: - added initial WINS server code. It passes most of the NBT-WINS test, but doesn't yet
  do secure server WACK responses

- added a ldap_string_to_time() function, for converting a LDAP
  formatted time to a time_t

13 years agor5357: added ldb_msg_add_fmt(), for creating formatted ldb record values
Andrew Tridgell [Sat, 12 Feb 2005 11:30:33 +0000 (11:30 +0000)]
r5357: added ldb_msg_add_fmt(), for creating formatted ldb record values

13 years agor5356: fixed the hex coding for nbt names
Andrew Tridgell [Sat, 12 Feb 2005 03:10:50 +0000 (03:10 +0000)]
r5356: fixed the hex coding for nbt names

13 years agor5353: Throw a NTSTATUS exception if a NT error code is returned.
Tim Potter [Sat, 12 Feb 2005 01:05:16 +0000 (01:05 +0000)]
r5353: Throw a NTSTATUS exception if a NT error code is returned.

13 years agor5352: added a function nbt_name_string() that formats a nbt_name structure
Andrew Tridgell [Sat, 12 Feb 2005 01:00:15 +0000 (01:00 +0000)]
r5352: added a function nbt_name_string() that formats a nbt_name structure
as a human readable string. The format is designed to be able to be
used as the DN for the WINS database as well, while coping with
arbitrary bytes in the name (except nul bytes)

13 years agor5351: Add wrapper for nt_errstr and array functions for samr_ConnectInfo.
Tim Potter [Sat, 12 Feb 2005 00:54:11 +0000 (00:54 +0000)]
r5351: Add wrapper for nt_errstr and array functions for samr_ConnectInfo.

13 years agor5350: Implement an input typemap for NTSTATUS.
Tim Potter [Sat, 12 Feb 2005 00:53:32 +0000 (00:53 +0000)]
r5350: Implement an input typemap for NTSTATUS.

13 years agor5348: Implement the remaining samr connect functions and fix up setsecurity.
Tim Potter [Sat, 12 Feb 2005 00:40:00 +0000 (00:40 +0000)]
r5348: Implement the remaining samr connect functions and fix up setsecurity.

13 years agor5347: fixed the NBT-REGISTER test now that the nbt layer is case sensitive
Andrew Tridgell [Fri, 11 Feb 2005 23:55:15 +0000 (23:55 +0000)]
r5347: fixed the NBT-REGISTER test now that the nbt layer is case sensitive

13 years agor5346: - a bit more preparation for the WINS server going in
Andrew Tridgell [Fri, 11 Feb 2005 23:54:37 +0000 (23:54 +0000)]
r5346: - a bit more preparation for the WINS server going in

- more NBT packet asserts, to ensure that incoming requests have all
  the elements we depend on

- open the WINS database at startup if we are configured as a WINS server

- split out the nbtd server reply packet generation code so it can be
  shared by the WINS server

- re-did the logic of what is answered by the WINS server and what by
  the B node server. It now always tries to answer by the B node, and
  only "recurses" to the WINS server for names that are not found.

13 years agor5345: Some more minor code readability fixes.
Jelmer Vernooij [Fri, 11 Feb 2005 23:21:06 +0000 (23:21 +0000)]
r5345: Some more minor code readability fixes.

13 years agor5338: More efforts to handle types similarly.
Jelmer Vernooij [Fri, 11 Feb 2005 15:49:15 +0000 (15:49 +0000)]
r5338: More efforts to handle types similarly.

13 years agor5335: Add a Close() method to SamrHandle object.
Tim Potter [Fri, 11 Feb 2005 11:18:19 +0000 (11:18 +0000)]
r5335: Add a Close() method to SamrHandle object.

Add methods for {Query,Set}Security, Shutdown, CreateUser,
Open{User,Group,Alias} and RidToSid.

Time for a torture test I think!

13 years agor5334: Bah, it's all to hard...
Andrew Bartlett [Fri, 11 Feb 2005 10:50:54 +0000 (10:50 +0000)]
r5334: Bah, it's all to hard...

Andrew Bartlett

13 years agor5333: weird, w2k3 always sends a positive name release response, even for names...
Andrew Tridgell [Fri, 11 Feb 2005 10:37:59 +0000 (10:37 +0000)]
r5333: weird, w2k3 always sends a positive name release response, even for names that
have never been registered. I wonder if there is some reason?

13 years agor5330: Remove #include <sys/time.h> from includes.h.
Andrew Bartlett [Fri, 11 Feb 2005 10:15:56 +0000 (10:15 +0000)]
r5330: Remove #include <sys/time.h> from includes.h.

Add #include "system/time.h" back (it was removed in some of these
places because the definitions were provided by <sys/time.h> on tridge's

Andrew Bartlett

13 years agor5329: made the nbt server case sensitive
Andrew Tridgell [Fri, 11 Feb 2005 08:18:55 +0000 (08:18 +0000)]
r5329: made the nbt server case sensitive

13 years agor5328: - allow case sensitive nbt name lookups
Andrew Tridgell [Fri, 11 Feb 2005 07:54:20 +0000 (07:54 +0000)]
r5328:  - allow case sensitive nbt name lookups

 - added --case-sensitive option to nmblookup

 - added case sensitivity tests to the NBT-WINS test

13 years agor5326: removed the charset conversion from the nbtname code, so we no longer
Andrew Tridgell [Fri, 11 Feb 2005 07:20:16 +0000 (07:20 +0000)]
r5326: removed the charset conversion from the nbtname code, so we no longer
convert from/to DOS strings in NBT names. This will allow us to
support all foreign names as a WINS server, as long as they don't
contain a 0 byte.

13 years agor5325: - expanded the NBT-WINS test to include scopes
Andrew Tridgell [Fri, 11 Feb 2005 06:58:07 +0000 (06:58 +0000)]
r5325: - expanded the NBT-WINS test to include scopes

- fixed the bugs that the new test found

13 years agor5322: removed a whole bunch of #include lines that
Andrew Tridgell [Fri, 11 Feb 2005 02:08:39 +0000 (02:08 +0000)]
r5322: removed a whole bunch of #include lines that
thinks are not needed. Now to see how this fares on the build farm :)

13 years agor5321: added a program that works out the minimal set of #include lines
Andrew Tridgell [Fri, 11 Feb 2005 02:07:41 +0000 (02:07 +0000)]
r5321: added a program that works out the minimal set of #include lines
needed for a C file in Samba. It tries compiling without each #include
line in turn, and looks for any changes in the compiler output.

Note that the output is not guaranteed correct, it might be that an
include is needed onlyu on some platforms. To cope with this, it
doesn't consider lines with the word "needed" on them. So add a
comment like this:

#include "foo.h" /* needed by systems without kerberos */

and it will know to skip it

It also skips any include lines in a #if section.

If you are brave, you can give it the option --remove and it will
remove lines it thinks are not needed. Please review carefully before
committing the results, and watch the build farm for breakage.

13 years agor5320: Treat structs and unions somewhat more similarly:
Jelmer Vernooij [Fri, 11 Feb 2005 02:05:47 +0000 (02:05 +0000)]
r5320: Treat structs and unions somewhat more similarly:
 - use same names in hashes (DATA -> ELEMENTS, etc)
 - [case()] and [default] are no longer special case, they're just regular properties

13 years agor5319: Make some more alignment code generic, move NDR-specific stuff to
Jelmer Vernooij [Thu, 10 Feb 2005 23:34:59 +0000 (23:34 +0000)]
r5319: Make some more alignment code generic, move NDR-specific stuff to (now renamed to

13 years agor5310: allow for rounding errors in the sleep test
Andrew Tridgell [Thu, 10 Feb 2005 07:57:09 +0000 (07:57 +0000)]
r5310: allow for rounding errors in the sleep test

13 years agor5309: removed ads.h from includes.h
Andrew Tridgell [Thu, 10 Feb 2005 07:43:39 +0000 (07:43 +0000)]
r5309: removed ads.h from includes.h

13 years agor5308: trimmed back a lot of the old macros from smb_macros.h
Andrew Tridgell [Thu, 10 Feb 2005 07:39:14 +0000 (07:39 +0000)]
r5308: trimmed back a lot of the old macros from smb_macros.h

13 years agor5307: removed db_wrap.h from includes.h
Andrew Tridgell [Thu, 10 Feb 2005 07:22:25 +0000 (07:22 +0000)]
r5307: removed db_wrap.h from includes.h

13 years agor5306: removed all the unused mutex functions from mutex.c. When (if?) we
Andrew Tridgell [Thu, 10 Feb 2005 07:18:00 +0000 (07:18 +0000)]
r5306: removed all the unused mutex functions from mutex.c. When (if?) we
decide to reinstate the mutex code for the threads process model, I'd
like to do it a little differently. At least this gets it out of
includes.h for now.

13 years agor5305: removed libcli/ldap/ldap.h from includes.h
Andrew Tridgell [Thu, 10 Feb 2005 07:08:40 +0000 (07:08 +0000)]
r5305: removed libcli/ldap/ldap.h from includes.h

13 years agor5304: removed lib/socket/socket.h from includes.h
Andrew Tridgell [Thu, 10 Feb 2005 06:59:29 +0000 (06:59 +0000)]
r5304: removed lib/socket/socket.h from includes.h

13 years agor5303: fixed build of gconf registry backend
Andrew Tridgell [Thu, 10 Feb 2005 06:45:13 +0000 (06:45 +0000)]
r5303: fixed build of gconf registry backend

13 years agor5302: fixed a compilation problem on solaris caused by the recent include
Andrew Tridgell [Thu, 10 Feb 2005 06:36:30 +0000 (06:36 +0000)]
r5302: fixed a compilation problem on solaris caused by the recent include

13 years agor5301: fixed pthreads build
Andrew Tridgell [Thu, 10 Feb 2005 05:29:38 +0000 (05:29 +0000)]
r5301: fixed pthreads build

13 years agor5300: more uint32 and system/filesys.h build fixes when developer mode is enabled
Andrew Tridgell [Thu, 10 Feb 2005 05:22:53 +0000 (05:22 +0000)]
r5300: more uint32 and system/filesys.h build fixes when developer mode is enabled

13 years agor5299: fixed an include ordering problem
Andrew Tridgell [Thu, 10 Feb 2005 05:15:37 +0000 (05:15 +0000)]
r5299: fixed an include ordering problem

13 years agor5298: - got rid of pstring.h from includes.h. This at least makes it a bit
Andrew Tridgell [Thu, 10 Feb 2005 05:09:35 +0000 (05:09 +0000)]
r5298:  - got rid of pstring.h from includes.h. This at least makes it a bit
   less likely that anyone will use pstring for new code

 - got rid of winbind_client.h from includes.h. This one triggered a
   huge change, as winbind_client.h was including system/filesys.h and
   defining the old uint32 and uint16 types, as well as its own
   pstring and fstring.

13 years agor5297: ensure pstring is not in the generated prototypes
Andrew Tridgell [Thu, 10 Feb 2005 05:06:41 +0000 (05:06 +0000)]
r5297: ensure pstring is not in the generated prototypes

13 years agor5296: - only include the tdb headers where they are needed
Andrew Tridgell [Thu, 10 Feb 2005 03:48:43 +0000 (03:48 +0000)]
r5296: - only include the tdb headers where they are needed

- removed the u32 hack in events.c as I think this was only needed as
  tdb.h defines u32. Metze, can you check that this hack is indeed no
  longer needed on your suse system?

13 years agor5294: - added a separate NBT-WINS test for WINS operations (register, refresh, relea...
Andrew Tridgell [Thu, 10 Feb 2005 03:22:47 +0000 (03:22 +0000)]
r5294: - added a separate NBT-WINS test for WINS operations (register, refresh, release and query)

- change the iface_n_*() functions to return a "const char *" instead of a "struct ipv4_addr"
  I think that in general we should move towards "const char *" for
  all IP addresses, as this makes IPv6 much easier, and is also easier
  to debug. Andrew, when you get a chance, could you fix some of the
  auth code to use strings for IPs ?

- return a NTSTATUS error on bad name queries and node status instead
  of using rcode. This makes the calling code simpler.

- added low level name release code in libcli/nbt/

- use a real IP in the register and wins nbt torture tests, as w2k3
  WINS server silently rejects some operations that don't come from the
  IP being used (eg. it says "yes" to a release, but does not in fact
  release the name)

13 years agor5293: Map an empty policy handle to None when creating policy handles, and
Tim Potter [Thu, 10 Feb 2005 03:19:21 +0000 (03:19 +0000)]
r5293: Map an empty policy handle to None when creating policy handles, and
refuse to accept None as a policy handle when accepting them.  Now we
don't segfault after running the samr_Shutdown() test.

13 years agor5292: ensure we cleanup the epoll_fd on event context destruction
Andrew Tridgell [Thu, 10 Feb 2005 03:16:33 +0000 (03:16 +0000)]
r5292: ensure we cleanup the epoll_fd on event context destruction

13 years agor5291: fixed ncacn_ip_tcp against windows
Andrew Tridgell [Thu, 10 Feb 2005 01:19:26 +0000 (01:19 +0000)]
r5291: fixed ncacn_ip_tcp against windows

13 years agor5288: Make alignment code more generic
Jelmer Vernooij [Wed, 9 Feb 2005 21:53:25 +0000 (21:53 +0000)]
r5288: Make alignment code more generic

13 years agor5286: Some first steps in making the pidl code somewhat more generic for the
Jelmer Vernooij [Wed, 9 Feb 2005 21:10:23 +0000 (21:10 +0000)]
r5286: Some first steps in making the pidl code somewhat more generic for the
various data types:

Add ndr_flags argument to all ndr push/pull scalar functions

13 years agor5285: add a generic dlist_item struct which has a void *ptr for the data
Stefan Metzmacher [Wed, 9 Feb 2005 17:29:09 +0000 (17:29 +0000)]
r5285: add a generic dlist_item struct which has a void *ptr for the data


13 years agor5284: this is needed on my server
Stefan Metzmacher [Wed, 9 Feb 2005 13:51:14 +0000 (13:51 +0000)]
r5284: this is needed on my server

+/* this #undef is needed on my SuSE 9.2 Box with glibc-devel-2.3.3-118 */
+#undef u32

as sys/epoll.h has this:
typedef union epoll_data
  void *ptr;
  int fd;
  uint32_t u32;
  uint64_t u64;
} epoll_data_t;


13 years agor5282: merging autogen fixes from 3_0
Gerald Carter [Tue, 8 Feb 2005 22:55:24 +0000 (22:55 +0000)]
r5282: merging autogen fixes from 3_0

13 years agor5277: initialise the multi_homed flag in the name registration test
Andrew Tridgell [Tue, 8 Feb 2005 01:10:34 +0000 (01:10 +0000)]
r5277: initialise the multi_homed flag in the name registration test

13 years agor5276: - added support for NBT_OPCODE_REFRESH2 (type 0x9)
Andrew Tridgell [Tue, 8 Feb 2005 01:09:21 +0000 (01:09 +0000)]
r5276: - added support for NBT_OPCODE_REFRESH2 (type 0x9)

- when registering with a WINS server, initially use multi-homed
  registration, then switch to name refresh requests. Send refresh
  requests only to the WINS server that responded to our
  registration. If that server goes away, then start the registration
  from scratch. This makes registration more robust to WINS server

- send WINS registration requests out on our first interface rather
  than an unbound interface, to avoid the problem of WACK replies
  being sent to the wrong port (w2k3 WINS server does this)

13 years agor5275: - added support for NBT_OPCODE_MULTI_HOME_REG (opcode 0xf) for WINS name regis...
Andrew Tridgell [Tue, 8 Feb 2005 01:05:41 +0000 (01:05 +0000)]
r5275: - added support for NBT_OPCODE_MULTI_HOME_REG (opcode 0xf) for WINS name registrations

- fixed a bug in the send queue handling on timeouts

- added support for handling unexpected replies (replies to the wrong
  port) at the nbtsocket layer

- added separate layer 2 code for wins refresh and wins registration

13 years agor5274: fixed some const warnings by making the str_list_ functions return "const...
Andrew Tridgell [Tue, 8 Feb 2005 01:02:56 +0000 (01:02 +0000)]
r5274: fixed some const warnings by making the str_list_ functions return "const char **"

13 years agor5273: fixed another bug in the code that keeps timed events
Andrew Tridgell [Tue, 8 Feb 2005 01:01:22 +0000 (01:01 +0000)]
r5273: fixed another bug in the code that keeps timed events
sorted. Hopefully it really works now :-)

13 years agor5266: This is a nice typo ;-)
Günther Deschner [Mon, 7 Feb 2005 22:02:05 +0000 (22:02 +0000)]
r5266: This is a nice typo ;-)


13 years agor5261: translate nbt rcode errors to NTSTATUS codes
Andrew Tridgell [Mon, 7 Feb 2005 12:10:38 +0000 (12:10 +0000)]
r5261: translate nbt rcode errors to NTSTATUS codes

13 years agor5260: - show an error message on nmblookup failure
Andrew Tridgell [Mon, 7 Feb 2005 11:56:34 +0000 (11:56 +0000)]
r5260: - show an error message on nmblookup failure

- always try to enable broadcast on nbt name sockets (this matches
  samba3 behaviour better)

13 years agor5259: make sure we give the ip of the interface that a name query comes in
Andrew Tridgell [Mon, 7 Feb 2005 11:49:55 +0000 (11:49 +0000)]
r5259: make sure we give the ip of the interface that a name query comes in
on as the first IP in a multi-homed reply

13 years agor5256: More verbose description of functions (as I learn the code).
Rafal Szczesniak [Sun, 6 Feb 2005 23:06:27 +0000 (23:06 +0000)]
r5256: More verbose description of functions (as I learn the code).


13 years agor5253: need to pre-declare some structures
Andrew Tridgell [Sun, 6 Feb 2005 08:33:22 +0000 (08:33 +0000)]
r5253: need to pre-declare some structures

13 years agor5252: - fixed nmblookup for the nbt api changes
Andrew Tridgell [Sun, 6 Feb 2005 08:27:12 +0000 (08:27 +0000)]
r5252: - fixed nmblookup for the nbt api changes

- added a simple WINS server name registration and query test

13 years agor5251: - renamed the nbtd server side structures to have a nbtd_ prefix, to
Andrew Tridgell [Sun, 6 Feb 2005 08:25:53 +0000 (08:25 +0000)]
r5251: - renamed the nbtd server side structures to have a nbtd_ prefix, to
  be consistent with the function names

- added WINS client support to the NBT server. It will do initial WINS
  registration, and WINS refresh, automatically failing over to
  secondary WINS servers and handling multi-homed servers where we need
  to register multiple IPs.

- added support for multi-homed name query replies, which are
  essential for multi-homed registration as the WINS server will query
  us to ensure we have the names when doing the secondary IPs in
  multi-homed registration

13 years agor5250: - added low level support for retrying nbt name queries, rather than
Andrew Tridgell [Sun, 6 Feb 2005 08:22:18 +0000 (08:22 +0000)]
r5250: - added low level support for retrying nbt name queries, rather than
  having the 2nd layer functions do retries themselves. This makes the
  code simpler, and allows the TRN_ID to be reused in the retry (which
  is how it is supposed to work).

- added support for WACK replies to nbt name requests. A WACK reply
  specifies a timeout to wait for the real reply.

- added WINS name refresh async calls, supporting multiple wins
  servers and multiple IPs to register

13 years agor5249: don't include ';' in the default list separators for parsing
Andrew Tridgell [Sun, 6 Feb 2005 08:16:18 +0000 (08:16 +0000)]
r5249: don't include ';' in the default list separators for parsing
smb.conf. It was rarely (if ever?) used, and poses problems as it is
not multi-byte safe for character searches

13 years agor5248: fixed a silly bug in DLIST_ADD_AFTER()
Andrew Tridgell [Sun, 6 Feb 2005 08:14:44 +0000 (08:14 +0000)]
r5248: fixed a silly bug in DLIST_ADD_AFTER()

13 years agor5247: Fix rpcclient to work with new swig interface.
Tim Potter [Sun, 6 Feb 2005 04:34:29 +0000 (04:34 +0000)]
r5247: Fix rpcclient to work with new swig interface.

13 years agor5245: Remove dead code.
Tim Potter [Sun, 6 Feb 2005 00:50:55 +0000 (00:50 +0000)]
r5245: Remove dead code.

13 years agor5244: Convert this module to use the new structure mapping SWIG stuff instead of
Tim Potter [Sun, 6 Feb 2005 00:38:10 +0000 (00:38 +0000)]
r5244: Convert this module to use the new structure mapping SWIG stuff instead of
the old dictionary based routines.  The SAMR connect, close, enumdomains,
opendomain, enum{users,aliases,groups} functions now work again!

13 years agor5243: Woohoo - memory leak city! Comment out talloc_free() call until I
Tim Potter [Sun, 6 Feb 2005 00:35:58 +0000 (00:35 +0000)]
r5243: Woohoo - memory leak city!  Comment out talloc_free() call until I
figure out a better way to pass tallocated memory around with SWIG.

13 years agor5242: Check that argument is an integer or a long for uint32_t input
Tim Potter [Sun, 6 Feb 2005 00:34:44 +0000 (00:34 +0000)]
r5242: Check that argument is an integer or a long for uint32_t input

The uint32_t output typemap must return a Python long as an unsigned
uint32_t cannot be fully represented by a Python int.

Likewise for the NTSTATUS typemap.

13 years agor5241: Generate swig wrappers for unions as well as structures.
Tim Potter [Sun, 6 Feb 2005 00:18:46 +0000 (00:18 +0000)]
r5241: Generate swig wrappers for unions as well as structures.

13 years agor5240: Don't return WERR_OK when no values were found (reported by Matt Cobb)
Jelmer Vernooij [Sat, 5 Feb 2005 23:10:31 +0000 (23:10 +0000)]
r5240: Don't return WERR_OK when no values were found (reported by Matt Cobb)

13 years agor5237: Add error code for "class not registered"
Jelmer Vernooij [Sat, 5 Feb 2005 14:40:47 +0000 (14:40 +0000)]
r5237: Add error code for "class not registered"

13 years agor5224: Add in/out typemaps for resume handles. This saves us having to much
Tim Potter [Fri, 4 Feb 2005 06:35:45 +0000 (06:35 +0000)]
r5224: Add in/out typemaps for resume handles.  This saves us having to much
around with pointers to just one uint32.

Add an output typemap to copy a policy handle as the talloc context is
destroyed before the wrapper function returns.  More work here needed
to avoid memory leaks.

Use the swig carrays.i file to create accessor and setter functions
for fixed width integer types.  Also add functions for struct samr_SamEntry
as it's returned by the LookupDomain RPC.  This really needs to be
done by pidl so I don't have to go through and find all the structures
that are returned in arrays.

Include security.i to give us SIDs and security descriptors.

13 years agor5223: Rename dom_sid2 to dom_sid as we don't care about the difference
Tim Potter [Fri, 4 Feb 2005 06:28:50 +0000 (06:28 +0000)]
r5223: Rename dom_sid2 to dom_sid as we don't care about the difference
for the swig wrappers.

13 years agor5222: made the nbtd_self_packet() code more efficient
Andrew Tridgell [Fri, 4 Feb 2005 05:13:46 +0000 (05:13 +0000)]
r5222: made the nbtd_self_packet() code more efficient

13 years agor5221: replace the str_list_*() code with new code based on talloc(). This is
Andrew Tridgell [Fri, 4 Feb 2005 04:58:48 +0000 (04:58 +0000)]
r5221: replace the str_list_*() code with new code based on talloc(). This is
a precursor to adding the wins client code in the nbt server.

13 years agor5217: avoid epoll_ctl() if the event flags are already set correctly
Andrew Tridgell [Fri, 4 Feb 2005 02:35:13 +0000 (02:35 +0000)]
r5217: avoid epoll_ctl() if the event flags are already set correctly

13 years agor5216: don't defend group names against incoming name registration requests
Andrew Tridgell [Fri, 4 Feb 2005 02:31:04 +0000 (02:31 +0000)]
r5216: don't defend group names against incoming name registration requests

13 years agor5215: register aliases as both client and server node types, so nmblookup can see...
Andrew Tridgell [Fri, 4 Feb 2005 02:22:01 +0000 (02:22 +0000)]
r5215: register aliases as both client and server node types, so nmblookup can see them

13 years agor5214: added support for "netbios aliases" in smb.conf
Andrew Tridgell [Fri, 4 Feb 2005 02:20:55 +0000 (02:20 +0000)]
r5214: added support for "netbios aliases" in smb.conf

13 years agor5213: do our name broadcast refresh requests as register packets not refresh
Andrew Tridgell [Fri, 4 Feb 2005 02:16:59 +0000 (02:16 +0000)]
r5213: do our name broadcast refresh requests as register packets not refresh
packets, as w2k3 and Samba3 do not defend against broadcast name
refresh packets

13 years agor5212: added checking for receiving our own packets as broadcasts
Andrew Tridgell [Fri, 4 Feb 2005 02:05:27 +0000 (02:05 +0000)]
r5212: added checking for receiving our own packets as broadcasts

13 years agor5211: added broadcast name defense against both registration and refresh
Andrew Tridgell [Fri, 4 Feb 2005 01:55:50 +0000 (01:55 +0000)]
r5211: added broadcast name defense against both registration and refresh

13 years agor5210: changed server side nbt functions to be prefixed with nbtd_ instead of
Andrew Tridgell [Fri, 4 Feb 2005 01:39:10 +0000 (01:39 +0000)]
r5210: changed server side nbt functions to be prefixed with nbtd_ instead of
nbt_, so as to more clearly separate them from the client code in

13 years agor5209: Fix the endpoint mapper to work with IPX endpoints (which
Jelmer Vernooij [Fri, 4 Feb 2005 01:32:19 +0000 (01:32 +0000)]
r5209: Fix the endpoint mapper to work with IPX endpoints (which
accidently have the same protocol id as UUID's)
Before this, Samba would give NDR errors when contacting
a remote server that has IPX support enabled.

This one was on my long due bugs list.

13 years agor5199: fix some minor configure bugs
Stefan Metzmacher [Thu, 3 Feb 2005 13:00:16 +0000 (13:00 +0000)]
r5199: fix some minor configure bugs


13 years agor5198: don't consider failure to remove an epoll event as enough reason to
Andrew Tridgell [Thu, 3 Feb 2005 12:26:28 +0000 (12:26 +0000)]
r5198: don't consider failure to remove an epoll event as enough reason to
fallback to select(). This can happen in too many situations.

13 years agor5197: moved events code to lib/events/ (suggestion from metze)
Andrew Tridgell [Thu, 3 Feb 2005 11:56:03 +0000 (11:56 +0000)]
r5197: moved events code to lib/events/ (suggestion from metze)

13 years agor5196: fixed sily bug (that metze found)
Andrew Tridgell [Thu, 3 Feb 2005 11:30:36 +0000 (11:30 +0000)]
r5196: fixed sily bug (that metze found)

13 years agor5195: most events don't need the time of the event, so save a gettimeofday() call
Andrew Tridgell [Thu, 3 Feb 2005 11:25:52 +0000 (11:25 +0000)]
r5195: most events don't need the time of the event, so save a gettimeofday() call
and just use timeval_current() when its actually needed

13 years agor5194: added support for using epoll instead of select() on systems that have
Andrew Tridgell [Thu, 3 Feb 2005 08:24:08 +0000 (08:24 +0000)]
r5194: added support for using epoll instead of select() on systems that have
it. epoll is much more scalable than select(), but only exists on some
systems (such as Linux with the 2.6.x kernel). The code detects any
epoll system call failures at runtime and falls back to select() if
there is a problem, so it should be safe to compile this on a 2.6
kernel and run it on a 2.4.x kernel.

The speedup is quite large. It gains 20% in packet rate in the
BENCH-NBT test, on top of another 20% gain from the better timer
handling I added earlier. The really big gain will be when we are
dealing with large numbers of file descriptors. With epoll we can
handle hundreds of file descriptors all O(1), whereas with select it
is O(n).

13 years agor5193: make sure we mark the event dead when we free it on a dead connection
Andrew Tridgell [Thu, 3 Feb 2005 08:20:31 +0000 (08:20 +0000)]
r5193: make sure we mark the event dead when we free it on a dead connection

13 years agor5190: Grr - typo.
Tim Potter [Thu, 3 Feb 2005 04:22:37 +0000 (04:22 +0000)]
r5190: Grr - typo.