sfrench/samba-autobuild/.git
24 years agofirst pass at updating head branch to be to be the same as the SAMBA_2_0 branch
Andrew Tridgell [Mon, 13 Dec 1999 13:27:58 +0000 (13:27 +0000)]
first pass at updating head branch to be to be the same as the SAMBA_2_0 branch

24 years agochanged function name of get_home_dir() to get_unixhome_dir(), to stop
Luke Leighton [Sun, 12 Dec 1999 21:00:35 +0000 (21:00 +0000)]
changed function name of get_home_dir() to get_unixhome_dir(), to stop
clash with gnu readline library.

fixed issue with [homes] service not being there - call lp_add_home()
just before starting the msrpc processing.

24 years agofinal part of "first" phase converting over to msrpc daemon architecture.
Luke Leighton [Sun, 12 Dec 1999 20:03:42 +0000 (20:03 +0000)]
final part of "first" phase converting over to msrpc daemon architecture.
done a minimal amout of clean-up in the Makefile, removing unnecessary
modules from the link stage.  this is not complete, yet, and will
involve some changes, for example to smbd, to remove dependencies on
the password database API that shouldn't be there.  for example,
smbd should not ever call getsmbpwXXX() it should call the Samr or Lsa
API.

this first implementation has minor problems with not reinstantiating
the same services as the caller.  the "homes" service is a good example.

24 years agodelineation between smb and msrpc more marked. smbd now constructs
Luke Leighton [Sun, 12 Dec 1999 01:25:49 +0000 (01:25 +0000)]
delineation between smb and msrpc more marked.  smbd now constructs
pdus, and then feeds them over either a "local" function call or a "remote"
function call to an msrpc service.  the "remote" msrpc daemon, on the
other side of a unix socket, then calls the same "local" function that
smbd would, if the msrpc service were being run from inside smbd.

this allows a transition from local msrpc services (inside the same smbd
process) to remote (over a unix socket).

removed reference to pipes_struct in msrpc services.  all msrpc processing
functions take rpcsrv_struct which is a structure containing state info
for the msrpc functions to decode and create pdus.

created become_vuser() which does everything not related to connection_struct
that become_user() does.

removed, as best i could, connection_struct dependencies from the nt spoolss
printing code.

todo: remove dcinfo from rpcsrv_struct because this stores NETLOGON-specific
info on a per-connection basis, and if the connection dies then so does
the info, and that's a fairly serious problem.

had to put pretty much everything that is in user_struct into parse_creds.c
to feed unix user info over to the msrpc daemons.  why?  because it's
expensive to do unix password/group database lookups, and it's definitely
expensive to do nt user profile lookups, not to mention pretty difficult
and if you did either of these it would introduce a complication /
unnecessary interdependency.  so, send uid/gid/num_groups/gid_t* +
SID+num_rids+domain_group_rids* + unix username + nt username + nt domain
+ user session key etc.  this is the MINIMUM info identified so far that's
actually implemented.  missing bits include the called and calling
netbios names etc.  (basically, anything that can be loaded into
standard_sub() and standard_sub_basic()...)

24 years agoOK. This code works on a RedHat 6.0 system. However smbpasswd
Gerald Carter [Thu, 9 Dec 1999 07:06:12 +0000 (07:06 +0000)]
OK.  This code works on a RedHat 6.0 system.  However smbpasswd
time out of sending the session setup on Solaris 2.6.  No idea.
I'll work on it some tomorrow.  This is to fix the "Unable to
setup password vectors" thingy.

Also changed an inet_aton() to inet_addr() as the former is
not very portable :-)

Luke,  I set the redir flag to false because the connection to
the smb-agent was failing and smbpasswd bombed.  Double check me
on this one.

-jc

24 years agojerry spotted that get_domain_sids() was being called with the wrong
Luke Leighton [Wed, 8 Dec 1999 23:03:42 +0000 (23:03 +0000)]
jerry spotted that get_domain_sids() was being called with the wrong
arguments: get_any_dc_name() was being called with a server name
not a domain name.  oops.

24 years agomoved sys/un.h to after sys/socket.h. done a gnu autoconf around sys/un.h
Luke Leighton [Wed, 8 Dec 1999 22:57:34 +0000 (22:57 +0000)]
moved sys/un.h to after sys/socket.h.  done a gnu autoconf around sys/un.h

24 years agoparsing code for transferring unix and nt security credentials over-the-wire.
Luke Leighton [Wed, 8 Dec 1999 21:45:57 +0000 (21:45 +0000)]
parsing code for transferring unix and nt security credentials over-the-wire.
at present, a unix password is missing from the unix credentials, but is
not _actually_ expected to be needed.  weeelll... maybe :-)

this is used to transfer credentials between smbd and msrpc daemons, down
a unix socket, so that the unix and nt credentials can be inherited by
an msrpc daemon called up from smbd.

24 years agoABOUT TIME!!!!!!!!
Luke Leighton [Wed, 8 Dec 1999 21:43:03 +0000 (21:43 +0000)]
ABOUT TIME!!!!!!!!

damn, this one is bad.

started, at least two days ago, to add an authentication mechanism to
the smbd<->msrpc redirector/relay, such that sufficient unix / nt
information could be transferred across the unix socket to do a
become_user() on the other side of the socket.

it is necessary that the msrpc daemon inherit the same unix and nt
credentials as the smbd process from which it was spawned, until
such time as the msrpc daemon receives an authentication request
of its own, whereupon the msrpc daemon is responsible for authenticating
the new credentials and doing yet another become_user() etc sequence.

24 years agothe first independent msrpc daemon - lsarpcd.
Luke Leighton [Mon, 6 Dec 1999 00:44:32 +0000 (00:44 +0000)]
the first independent msrpc daemon - lsarpcd.

one horrible cut / paste job from smbd, plus a code split of shared
components between the two.

the job is not _yet_ complete, as i need to be able to do a become_user()
call for security reasons.  i picked lsarpcd first because you don't
_need_ security on it (microsoft botched so badly on this one, it's not
real.  at least they fixed this in nt5 with restrictanonymous=0x2).
fixing this involves sending the current smb and unix credentials down
the unix pipe so that the daemon it eventually goes to can pick them
up at the other end.

i can't believe this all worked!!!

24 years agoreenabled retry_time (commented out accidentally in name_status)
Luke Leighton [Sun, 5 Dec 1999 02:22:28 +0000 (02:22 +0000)]
reenabled retry_time (commented out accidentally in name_status)

24 years agofirst version of msrpc agent redirector code. client-side only.
Luke Leighton [Sun, 5 Dec 1999 01:34:34 +0000 (01:34 +0000)]
first version of msrpc agent redirector code.  client-side only.

24 years agocreated create_pipe_socket() function.
Luke Leighton [Sun, 5 Dec 1999 00:13:17 +0000 (00:13 +0000)]
created create_pipe_socket() function.

24 years agoargh! how horrible! spent ages working out why packets weren't being
Luke Leighton [Sat, 4 Dec 1999 23:40:21 +0000 (23:40 +0000)]
argh!  how horrible!  spent ages working out why packets weren't being
received properly when a UDP "retry" occurs.  it's because reads and
writes must be interleaved / matched.

scenario:

nmblookup connects to agent, sends request.

agent receives request, broadcasts it on 137.

agent RECEIVES 137 broadcast, sends it to nmblookup

agent receives RESPONSE to 137 broadcast, sends it to nmblookup.

if reads are not equally interspersed with writes, then second send
will fail.

if you think this is odd behaviour and that the agent should be filtering
its own UDP traffic, think again.

agent will be, potentially, redirecting nmbd traffic (including WINS
server) not just client programs.

24 years agoargh. trying to get the nmb agent code to filter out "self" packets.
Luke Leighton [Sat, 4 Dec 1999 22:22:37 +0000 (22:22 +0000)]
argh.  trying to get the nmb agent code to filter out "self" packets.

24 years agoeven more cool! nmb-agent moved over to new vagent style, too!
Luke Leighton [Sat, 4 Dec 1999 22:05:59 +0000 (22:05 +0000)]
even more cool!  nmb-agent moved over to new vagent style, too!

24 years agocool! created higher-order function table for agent redirection.
Luke Leighton [Sat, 4 Dec 1999 21:55:39 +0000 (21:55 +0000)]
cool!  created higher-order function table for agent redirection.

24 years agomore similar agent code
Luke Leighton [Sat, 4 Dec 1999 20:43:45 +0000 (20:43 +0000)]
more similar agent code

24 years agonmb agent memory free problems
Luke Leighton [Sat, 4 Dec 1999 20:29:03 +0000 (20:29 +0000)]
nmb agent memory free problems

24 years agotrying to make redirector agent code same base.
Luke Leighton [Sat, 4 Dec 1999 20:11:31 +0000 (20:11 +0000)]
trying to make redirector agent code same base.

24 years agojeremy is going to hate me for this.
Luke Leighton [Sat, 4 Dec 1999 19:14:37 +0000 (19:14 +0000)]
jeremy is going to hate me for this.

created an "nmb-agent" utility that, yes: it connects to the 137 socket
and accepts unix socket connections which it redirects onto port 137.

it uses the name_trn_id field to filter requests to the correct
location.

name_query() and name_status() are the first victims to use this
feature (by specifying a file descriptor of -1).

24 years agoargh! you wouldn't believe what i had to do: use the mid (multiplex id)
Luke Leighton [Sat, 4 Dec 1999 00:49:13 +0000 (00:49 +0000)]
argh!  you wouldn't believe what i had to do: use the mid (multiplex id)
to redirect multiple socket-based connnections onto a single client state.

argh!

24 years agoargh! smb-agent redirection client reusage is a nightmare!
Luke Leighton [Fri, 3 Dec 1999 23:36:53 +0000 (23:36 +0000)]
argh!  smb-agent redirection client reusage is a nightmare!
moved smb-agent over to a single-process model instead of fork()
in order to reuse client connections.  except, of course, you
can't do a select() on the same socket connections!  argh!

24 years agostarting "connection reuse" system in smb-agent. added version number
Luke Leighton [Fri, 3 Dec 1999 22:02:03 +0000 (22:02 +0000)]
starting "connection reuse" system in smb-agent.  added version number
which isn't actually used right now :-)

24 years agoi always get caught out with non-gnu-readline code mods :-)
Luke Leighton [Fri, 3 Dec 1999 19:59:44 +0000 (19:59 +0000)]
i always get caught out with non-gnu-readline code mods :-)

24 years agosmb-agent improvements. added -D (daemon) option. smb agent is
Luke Leighton [Fri, 3 Dec 1999 19:55:34 +0000 (19:55 +0000)]
smb-agent improvements.  added -D (daemon) option.  smb agent is
restricted to connections from the current user (socket is created
with current user uid).

24 years agocool! a unix socket smb redirector. code based on smbfilter and
Luke Leighton [Fri, 3 Dec 1999 18:16:08 +0000 (18:16 +0000)]
cool!  a unix socket smb redirector.  code based on smbfilter and
ideas from ssh-agent.

the intent is to be able to share smb sessions using cli_net_use_add()
across multiple processes, where one process knows the target server
name, user name and domain, but not the smb password.

24 years agopromptline undefined (pline)
Luke Leighton [Fri, 3 Dec 1999 17:02:47 +0000 (17:02 +0000)]
promptline undefined (pline)

24 years agonew get_any_dc_name() function allows lookups of trusted domains from
Luke Leighton [Thu, 2 Dec 1999 20:16:34 +0000 (20:16 +0000)]
new get_any_dc_name() function allows lookups of trusted domains from
lp_trusted_domains() parameter, so trusted domain logins should work,
right, if you put user = TRUSTED_DOMAIN\NTuser in "domain name map", right?

right - as _long_ as you're not using NTLMv2, because the damn NT username
gets mapped to the damn unix name too early, and NTLMv2 challenge-responses
are based on the client's user name, client's domain name, client's host name
etc damn etc.

so it becomes necessary to stop using char* username because this allows
for massive amounts of confusion as to which username is being referred to.
the underlying unix username on the local unix system that is associated with
the smbd process that represents the NT username?  or the NT username itself?

24 years agodomain_client_validate() no longer takes serverlist, it calls
Luke Leighton [Thu, 2 Dec 1999 19:07:13 +0000 (19:07 +0000)]
domain_client_validate() no longer takes serverlist, it calls
get_any_dc_name().

24 years agocleaning up: removing those horrible references to server list
Luke Leighton [Thu, 2 Dec 1999 19:03:23 +0000 (19:03 +0000)]
cleaning up: removing those horrible references to server list
functions (cli_net_use_addlist()).  needed originally because
there was no get_dc_any_name() function.

24 years agoadded get_any_dc_name() function.
Luke Leighton [Thu, 2 Dec 1999 18:49:28 +0000 (18:49 +0000)]
added get_any_dc_name() function.

24 years agoneed a domain resolving function, but get_trusted_serverlist() will do.
Luke Leighton [Thu, 2 Dec 1999 16:52:38 +0000 (16:52 +0000)]
need a domain resolving function, but get_trusted_serverlist() will do.
this is horrible.

24 years agooops, pwdb_initialise() called unnecessarily from here (failure to read
Luke Leighton [Thu, 2 Dec 1999 16:38:16 +0000 (16:38 +0000)]
oops, pwdb_initialise() called unnecessarily from here (failure to read
trusted domains may cause rpcclient to fail).

24 years agodefault SID map now reads in "trusted domains" from smb.conf.
Luke Leighton [Thu, 2 Dec 1999 16:31:24 +0000 (16:31 +0000)]
default SID map now reads in "trusted domains" from smb.conf.

24 years agoimproved enumdomains added -i option.
Luke Leighton [Thu, 2 Dec 1999 01:46:01 +0000 (01:46 +0000)]
improved enumdomains added -i option.

24 years agoclearing up connection-related stuff. password credentials were messing
Luke Leighton [Thu, 2 Dec 1999 01:16:05 +0000 (01:16 +0000)]
clearing up connection-related stuff.  password credentials were messing
up.

added a complicated prompt which i don't like, but it tells you
domain\user@hostname$

24 years agoadded net use (actually net -S srv -U user -W dom) and net del (actually
Luke Leighton [Wed, 1 Dec 1999 22:39:27 +0000 (22:39 +0000)]
added net use (actually net -S srv -U user -W dom) and net del (actually
same as net use but with -d and -f) command options

24 years agomore cli_session_setup() calls. what the heck are these doing???
Luke Leighton [Wed, 1 Dec 1999 22:06:53 +0000 (22:06 +0000)]
more cli_session_setup() calls.  what the heck are these doing???
they should all be replaced with cli_establish_connection().

created cli_use_wait_keyboard() which waits on multiple cli_states
and swallows session keepalives.

24 years agocli_session_setup() now takes an extra argument (host name). hey, what
Luke Leighton [Wed, 1 Dec 1999 21:47:30 +0000 (21:47 +0000)]
cli_session_setup() now takes an extra argument (host name).  hey, what
the heck is a cli_session_setup() call doing in here???  this should use
cli_establish_connection()server!

24 years agomake sure domain and name to generate trust account .mac file are upper case.
Luke Leighton [Wed, 1 Dec 1999 20:18:59 +0000 (20:18 +0000)]
make sure domain and name to generate trust account .mac file are upper case.

24 years agodamn, that took a while. nt login password was being stored incorrectly
Luke Leighton [Wed, 1 Dec 1999 20:18:21 +0000 (20:18 +0000)]
damn, that took a while.  nt login password was being stored incorrectly
in private .mac file (oops).  ntlogin test now works.

24 years agofixing joining to domain plus something weird going down with nt logins...
Luke Leighton [Wed, 1 Dec 1999 19:25:51 +0000 (19:25 +0000)]
fixing joining to domain plus something weird going down with nt logins...

24 years agoimproving createuser account command to be able to add workstations
Luke Leighton [Wed, 1 Dec 1999 18:47:29 +0000 (18:47 +0000)]
improving createuser account command to be able to add workstations
and then set a default random password.

24 years ago1) when no domain used in ntlogin test command, should use default one
Luke Leighton [Wed, 1 Dec 1999 16:39:51 +0000 (16:39 +0000)]
1) when no domain used in ntlogin test command, should use default one
   from previous lsaquery command.  over-ridden from DOMAIN\username

2) initialisation of cli_state is a little more specific: sets use_ntlmv2
   to Auto.  this can always be over-ridden.

3) fixed reusage of ntlmssp_cli_flgs which was being a pain

4) added pwd_compare() function then fixed bug in cli_use where NULL
   domain name was making connections multiply unfruitfully

5) type-casting of mallocs and Reallocs that cause ansi-c compilers to bitch

24 years agosys_select added one more argument (read, write selectors).
Luke Leighton [Wed, 1 Dec 1999 02:15:14 +0000 (02:15 +0000)]
sys_select added one more argument (read, write selectors).

24 years agosplit display.c into modules.
Luke Leighton [Wed, 1 Dec 1999 00:52:22 +0000 (00:52 +0000)]
split display.c into modules.

24 years agoadded failed connections to the net use array, even though they'd been
Luke Leighton [Tue, 30 Nov 1999 18:01:03 +0000 (18:01 +0000)]
added failed connections to the net use array, even though they'd been
freed / cleaned up.  oops, dat bad, cos they get freed again when u quit.

24 years agook. this is where it gets interesting. client states are now maintained
Luke Leighton [Tue, 30 Nov 1999 00:08:39 +0000 (00:08 +0000)]
ok.  this is where it gets interesting.  client states are now maintained
by cli_net_use_add() and cli_net_use_del().  MSRPC connections are
established with cli_connection_init(), and automatically unlinked with
cli_connection_unlink.  client states are _reused_ by cli_connection_init.

24 years agobug-fix
Luke Leighton [Mon, 29 Nov 1999 23:57:41 +0000 (23:57 +0000)]
bug-fix

24 years agothis is going to sound _really_ weird, ok, but i had to implement
Luke Leighton [Mon, 29 Nov 1999 23:56:09 +0000 (23:56 +0000)]
this is going to sound _really_ weird, ok, but i had to implement
equivalents of NetUseAdd and NetUseDel!

24 years agosam sync - one of the files that use multiple connection server list
Luke Leighton [Mon, 29 Nov 1999 21:48:41 +0000 (21:48 +0000)]
sam sync - one of the files that use multiple connection server list
to \PIPE\NETLOGON.

24 years agoattempting to resolve the issue that multiple servers often specified in
Luke Leighton [Mon, 29 Nov 1999 21:47:14 +0000 (21:47 +0000)]
attempting to resolve the issue that multiple servers often specified in
parameters to connect to \PIPE\NETLOGON.

24 years agook. got ntlogin command working. argh, it maintains a connection to
Luke Leighton [Mon, 29 Nov 1999 21:16:12 +0000 (21:16 +0000)]
ok.  got ntlogin command working.  argh, it maintains a connection to
the remote machine, because i don't know what to _do_ with it!!!!

argh!!!

24 years agofirst attempt at getting \PIPE\NETLOGON working. it's pretty horrible.
Luke Leighton [Mon, 29 Nov 1999 19:46:57 +0000 (19:46 +0000)]
first attempt at getting \PIPE\NETLOGON working.  it's pretty horrible.

24 years agorenamed PRINTER_HND to POLICY_HND.
Luke Leighton [Mon, 29 Nov 1999 17:45:47 +0000 (17:45 +0000)]
renamed PRINTER_HND to POLICY_HND.

24 years agocool! spooljobs works! this surprised me very much :-) helped to
Luke Leighton [Sat, 27 Nov 1999 23:31:45 +0000 (23:31 +0000)]
cool!  spooljobs works!  this surprised me very much :-)  helped to
specify \PIPE\spoolss instead of \PIPE\lsarpc...

24 years agowell, i stuffed up the spooler commands.
Luke Leighton [Sat, 27 Nov 1999 23:25:45 +0000 (23:25 +0000)]
well, i stuffed up the spooler commands.

24 years agomoved at command over to new abstract connection system. matthew, you
Luke Leighton [Sat, 27 Nov 1999 22:58:11 +0000 (22:58 +0000)]
moved at command over to new abstract connection system.  matthew, you
initialised dest_wks _after_ using it in at_soon() :-) so i fixed this :)

24 years agomoved browser command brsinfo over to new abstracted connection
Luke Leighton [Sat, 27 Nov 1999 22:53:28 +0000 (22:53 +0000)]
moved browser command brsinfo over to new abstracted connection

24 years agothis one's a handle-based one (missed in the first round).
Luke Leighton [Sat, 27 Nov 1999 22:47:17 +0000 (22:47 +0000)]
this one's a handle-based one (missed in the first round).

24 years agoremoved do_ prefix from srvsvc API
Luke Leighton [Sat, 27 Nov 1999 22:35:58 +0000 (22:35 +0000)]
removed do_ prefix from srvsvc API

24 years agoupdated \PIPE\wkssvc commands to use new abstracted connection system.
Luke Leighton [Sat, 27 Nov 1999 22:34:12 +0000 (22:34 +0000)]
updated \PIPE\wkssvc commands to use new abstracted connection system.

modified resolve_srv_name() to return dest host of *SMBSERVER if
server name is \\ip.add.ress.format

24 years agousing "abstracted" cli_connection, got \PIPE\srvsvc commands up and running
Luke Leighton [Sat, 27 Nov 1999 22:19:51 +0000 (22:19 +0000)]
using "abstracted" cli_connection, got \PIPE\srvsvc commands up and running
again.

24 years agomodified cli_connect_serverlist to take server list of format
Luke Leighton [Sat, 27 Nov 1999 22:14:37 +0000 (22:14 +0000)]
modified cli_connect_serverlist to take server list of format
\\server_name \\other_server etc.

24 years agofurther abstraction involving client states. main client-side code
Luke Leighton [Sat, 27 Nov 1999 21:50:11 +0000 (21:50 +0000)]
further abstraction involving client states.  main client-side code
is pretty much independent of SMB client states, which will make it
easier to add other transports.

24 years agoremove use of client_info lsa_info_pol.
Luke Leighton [Sat, 27 Nov 1999 20:34:32 +0000 (20:34 +0000)]
remove use of client_info lsa_info_pol.

24 years agoenhanced samuser command to do same thing as enumusers command (-g -u -a)
Luke Leighton [Sat, 27 Nov 1999 20:29:16 +0000 (20:29 +0000)]
enhanced samuser command to do same thing as enumusers command (-g -u -a)
except with only one user.  done by sharing same code.

24 years agobug-fixing registry commands and the rpcclient "rpcclient" command.
Luke Leighton [Sat, 27 Nov 1999 00:02:03 +0000 (00:02 +0000)]
bug-fixing registry commands and the rpcclient "rpcclient" command.
the rpcclient "rpcclient" command allows user options to be reset
(e.g the username / password) _without_ terminating rpcclient.
try this:

rpcclient -S srv1 -U% -l log
srv1$ rpcclient -S srv2
srv2$

:-)

24 years agowhoa. _major_ restructure of rpcclient. fixed some buuugs, created a few.
Luke Leighton [Fri, 26 Nov 1999 23:04:19 +0000 (23:04 +0000)]
whoa.  _major_ restructure of rpcclient.  fixed some buuugs, created a few.
found out that getopt() _must_ have optind set to 0 before reuse.

still haven't decided what to do with the net* api yet...

24 years agothis file manages client states associated with handles.
Luke Leighton [Fri, 26 Nov 1999 22:47:50 +0000 (22:47 +0000)]
this file manages client states associated with handles.

24 years agoprevious commit added an abstraction function that didn't even have
Luke Leighton [Thu, 25 Nov 1999 05:34:12 +0000 (05:34 +0000)]
previous commit added an abstraction function that didn't even have
struct cli_state, uint16 fnum into the code: rpc_hnd_api_req().
modified cli_lsarpc.c to use this.  the rest is const issues.

24 years agocool! completed a samr* API that _would_ look like an msdn samr* api...
Luke Leighton [Thu, 25 Nov 1999 05:26:48 +0000 (05:26 +0000)]
cool!  completed a samr* API that _would_ look like an msdn samr* api...
if microsoft bothered to publish it.  actually, there are good reasons
for not publishing it: people might write programs for it, and then
those programs wouldn't work on nt5, for example...

24 years agoregistry API moved over to new format. reg_connect() is the top-level
Luke Leighton [Wed, 24 Nov 1999 23:40:20 +0000 (23:40 +0000)]
registry API moved over to new format.  reg_connect() is the top-level
function, which takes \\server_name.

tested a _few_ functions.  found that regcreatekey receives a Fault PDU.

24 years agoservice control manager API completed. svcenum -i works, but does not
Luke Leighton [Wed, 24 Nov 1999 23:11:03 +0000 (23:11 +0000)]
service control manager API completed.  svcenum -i works, but does not
do so twice.  possible memory corruption, revolving around getopt().

24 years agook. *whew*. this is the first completed part of the restructure.
Luke Leighton [Wed, 24 Nov 1999 22:45:09 +0000 (22:45 +0000)]
ok.  *whew*.  this is the first completed part of the restructure.

verified that lsaquery, lsalookupsids work, and found some bugs in the
parameters of these commands :-)

soo... we now have an lsa_* api that has the same arguments as the nt
Lsa* api!  cool!

the only significant coding difference is the introduction of a
user_credentials structure, containing user, domain, pass and ntlmssp
flags.

24 years agofirst stages of removing struct cli_state* and uint16 fnum from all
Luke Leighton [Wed, 24 Nov 1999 20:24:33 +0000 (20:24 +0000)]
first stages of removing struct cli_state* and uint16 fnum from all
msrpc client code.  the intent is to hide / abstract / associate
connection info behind policy handles.

this makes the msrpc functions look more and more like their nt equivalents.

who-hou!

24 years agoWARNING! MOVED rpc_server/srv_lsa_hnd.c TO lib/util_hnd.c
Luke Leighton [Wed, 24 Nov 1999 18:15:50 +0000 (18:15 +0000)]
WARNING!  MOVED rpc_server/srv_lsa_hnd.c TO lib/util_hnd.c
CVS UPDATE MAY ISSUE WARNING ABOUT lib/util_hnd.c MODIFICATION
DATE BEING IN THE FUTURE.  CVS CHECKOUT A NEW REPOSITORY MAY BE
SAFER.

24 years agorewrote policy handle code to be generic (it's needed for client-side too)
Luke Leighton [Wed, 24 Nov 1999 18:09:33 +0000 (18:09 +0000)]
rewrote policy handle code to be generic (it's needed for client-side too)

attempted to fix regsetsec command

24 years agooops, freed argc,argv arguments twice.
Luke Leighton [Tue, 23 Nov 1999 23:14:10 +0000 (23:14 +0000)]
oops, freed argc,argv arguments twice.

24 years agoattempted a svcset command. password is encrypted / messed up, therefore
Luke Leighton [Tue, 23 Nov 1999 23:05:47 +0000 (23:05 +0000)]
attempted a svcset command.  password is encrypted / messed up, therefore
command fails.

24 years agooops!!!! wrong command!!!
Luke Leighton [Tue, 23 Nov 1999 20:32:52 +0000 (20:32 +0000)]
oops!!!! wrong command!!!

24 years agoadding svcctl 0x1b function
Luke Leighton [Tue, 23 Nov 1999 18:57:07 +0000 (18:57 +0000)]
adding svcctl 0x1b function

24 years agoshuffling msrpc code around so that it can be used independently of rpcclient
Luke Leighton [Tue, 23 Nov 1999 18:56:26 +0000 (18:56 +0000)]
shuffling msrpc code around so that it can be used independently of rpcclient

24 years agothe last one. that concludes the removal of all next_token() calls
Luke Leighton [Mon, 22 Nov 1999 20:14:13 +0000 (20:14 +0000)]
the last one.  that concludes the removal of all next_token() calls
from rpcclient/cmd_*.c.

24 years agoanother four next_token() removals (using getopt instead)
Luke Leighton [Mon, 22 Nov 1999 19:46:26 +0000 (19:46 +0000)]
another four next_token() removals (using getopt instead)

24 years agoanother two.
Luke Leighton [Mon, 22 Nov 1999 19:37:05 +0000 (19:37 +0000)]
another two.

24 years agoanother one
Luke Leighton [Mon, 22 Nov 1999 19:26:56 +0000 (19:26 +0000)]
another one

24 years agomoved two more commands over to getopt.
Luke Leighton [Mon, 22 Nov 1999 19:25:30 +0000 (19:25 +0000)]
moved two more commands over to getopt.

24 years agookay :) all cmd_() functions now take int argc, char **argv :) that
Luke Leighton [Mon, 22 Nov 1999 19:02:39 +0000 (19:02 +0000)]
okay :)  all cmd_() functions now take int argc, char **argv :)  that
means that some commands need more work, as they still use next_token(),
the use of which i wish to avoid.

plus, i was getting fed up of the poor command-line processing in some
of these commands.  i'm starting to need getopt() in them, especially
in samsetuser.

WARNING: only cmd_samr has been modded to use getopt() so far!  reg
commands won't work, esp.

24 years agoimplement server-side generation of NTLMv2 session key. YESSS :-)
Luke Leighton [Sun, 21 Nov 1999 19:59:56 +0000 (19:59 +0000)]
implement server-side generation of NTLMv2 session key.  YESSS :-)

24 years agoyou know what? this sort of thing makes me laugh. hmm, what functions
Luke Leighton [Sun, 21 Nov 1999 19:24:01 +0000 (19:24 +0000)]
you know what?  this sort of thing makes me laugh.  hmm, what functions
have we got.  and what data do we have.  hmm.. i wonder what the NTLMv2
user session key can be... hmmm... weell.... there's some hidden data
here, generated from the user password that doesn't go over-the-wire,
so that's _got_ to be involved.  and... that bit of data took a lot of
computation to produce, so it's probably _also_ involved... and md4 no, md5?
no, how about hmac_md5 yes let's try that one (the other's didn't work)
oh goodie, it worked!

i love it when this sort of thing happens.  took all of fifteen minutes to
guess it.  tried concatenating client and server challenges.  tried
concatenating _random_ bits of client and server challenges.  tried
md5 of the above.  tried hmac_md5 of the above.  eventually, it boils down
to this:

kr = MD4(NT#,username,domainname)
hmacntchal=hmac_md5(kr, nt server challenge)
sess_key = hmac_md5(kr, hmacntchal);

24 years agohmmm... have to add client-side support in domain_client_validate() to
Luke Leighton [Sun, 21 Nov 1999 17:27:20 +0000 (17:27 +0000)]
hmmm... have to add client-side support in domain_client_validate() to
_use_ user session key.

24 years agoadding user session key into network netlogon response.
Luke Leighton [Sun, 21 Nov 1999 17:11:00 +0000 (17:11 +0000)]
adding user session key into network netlogon response.

24 years agomoving create user function into msrpc_samr.c
Luke Leighton [Sun, 21 Nov 1999 17:09:20 +0000 (17:09 +0000)]
moving create user function into msrpc_samr.c

24 years agooops, #ifdef'd cli_shutdown out, as the fun has _already_ started:
Luke Leighton [Sat, 20 Nov 1999 22:05:31 +0000 (22:05 +0000)]
oops, #ifdef'd cli_shutdown out, as the fun has _already_ started:
NT refuses to play nice, and establish a trust relationship.

24 years agoattempting to establish inter-domain trust relationships. modified
Luke Leighton [Sat, 20 Nov 1999 21:59:16 +0000 (21:59 +0000)]
attempting to establish inter-domain trust relationships.  modified
smbpasswd so it can be used to set up inter-domain trust account.

24 years agomodified domain_client_validate to take trust account name / type. this
Luke Leighton [Sat, 20 Nov 1999 20:54:29 +0000 (20:54 +0000)]
modified domain_client_validate to take trust account name / type.  this
is to pass DOMAIN_NAME$ and SEC_CHAN_DOMAIN instead of WKSTA_NAME$ and
SEC_CHAN_WKSTA.

modified check_domain_security to determine if domain name is own domain,
and to use wksta trust account if so, otherwise check "trusting domains"
parameter and use inter-domain trust account if so, otherwise return
False.

24 years agodoing a code reshuffle. want to add code to establish trust relationships.
Luke Leighton [Sat, 20 Nov 1999 19:43:37 +0000 (19:43 +0000)]
doing a code reshuffle.  want to add code to establish trust relationships.

24 years agodynamic memory allocation i added a month ago: forgot to ZERO_STRUCT()
Luke Leighton [Sat, 20 Nov 1999 18:17:29 +0000 (18:17 +0000)]
dynamic memory allocation i added a month ago: forgot to ZERO_STRUCT()
some of the server-side stuff.  Realloc() was being used, so it
Realloc()d some random area of memory.  oops.dynamic memory allocation i added a month ago: forgot to ZERO_STRUCT()
some of the server-side stuff.  Realloc() was being used, so it
Realloc()d some random area of memory.  oops.dynamic memory allocation i added a month ago: forgot to ZERO_STRUCT()
some of the server-side stuff.  Realloc() was being used, so it
Realloc()d some random area of memory.  oops.