samba.git
24 years agoThis commit was manufactured by cvs2svn to create tag samba-misc-tags/tng/release-alpha-0-3
cvs2svn Import User [Sun, 13 Feb 2000 07:33:08 +0000 (07:33 +0000)]
This commit was manufactured by cvs2svn to create tag
'release-alpha-0-3'.

24 years agopreparing for release of alpha.0.3
Samba Release Account [Sun, 13 Feb 2000 07:33:07 +0000 (07:33 +0000)]
preparing for release of alpha.0.3

24 years agoremoved some client progs from explicit required makefile support while
Luke Leighton [Sun, 13 Feb 2000 07:28:18 +0000 (07:28 +0000)]
removed some client progs from explicit required makefile support while
i obtain access to a readline system.

24 years agonprocs not set properly
Luke Leighton [Sun, 13 Feb 2000 07:26:30 +0000 (07:26 +0000)]
nprocs not set properly

24 years agoadd connection was ok, delete wasn't. ncacn_np_use_del fixed; _use_ of
Luke Leighton [Sun, 13 Feb 2000 07:06:43 +0000 (07:06 +0000)]
add connection was ok, delete wasn't.  ncacn_np_use_del fixed; _use_ of
this fn fixed.

24 years agohmm, connections being closed, but use-list doesn't show that correctly.
Luke Leighton [Sun, 13 Feb 2000 06:35:22 +0000 (06:35 +0000)]
hmm, connections being closed, but use-list doesn't show that correctly.

24 years agoretired. wasn't a good name anyway.
Luke Leighton [Sun, 13 Feb 2000 06:04:31 +0000 (06:04 +0000)]
retired.  wasn't a good name anyway.

24 years agohooray, i think i finally have it. couple of bugs (like forgetting to
Luke Leighton [Sun, 13 Feb 2000 05:57:57 +0000 (05:57 +0000)]
hooray, i think i finally have it.  couple of bugs (like forgetting to
store the pipe name in the ncacn_np struct so that the re-use search
could FIND the damn thing to actually reuse it).

24 years agodebugging.
Luke Leighton [Sun, 13 Feb 2000 01:54:52 +0000 (01:54 +0000)]
debugging.

24 years agook. this introduces the following structures:
Luke Leighton [Sun, 13 Feb 2000 01:27:20 +0000 (01:27 +0000)]
ok.  this introduces the following structures:

msrpc_local and ncalrpc_l_use.
ncacn_np and ncacn_np_use.

basically, what is happening is that connections can be reused (ncalrpc
for unix socket loop-back and ncacn_np for named pipes over SMB).

so, that's what these are for.

unfortunately, what makes this so stupidly complicated is that ncacn_np
can reuse SMB connections, not just SMB pipes.  so it uses the following
structures:

cli_state and cli_use.

24 years agoFrom Elrond@Wunder-Nett.org Sun Feb 13 05:30:09 2000
Luke Leighton [Sat, 12 Feb 2000 18:31:31 +0000 (18:31 +0000)]
From Elrond@Wunder-Nett.org Sun Feb 13 05:30:09 2000
Date: Sat, 12 Feb 2000 19:00:20 +0100
From: Elrond <Elrond@Wunder-Nett.org>
To: Luke Kenneth Casson Leighton <lkcl@samba.org>
Subject: some cleanups

Some clean-ups, I made this morning:

- Re-added spool- and at-commands to rpcclient
- One of these "automated" xxx_io_xxx-return-checkers went
  wrong in srv_lsa.c
- instead of exporting the command-tables in
  rpcclient/*_cmds.c, those export now a
  add_xxx_commands(), that adds these commands and the
  table can now be static const (and so ends up in the
  shared readonly-data segment)

24 years agopatch from elrond - more srvsvc dynamic memory allocations.
Luke Leighton [Sat, 12 Feb 2000 18:30:05 +0000 (18:30 +0000)]
patch from elrond - more srvsvc dynamic memory allocations.

24 years agopatch from elrond starting on lsarpcd conversion
Luke Leighton [Sat, 12 Feb 2000 18:28:26 +0000 (18:28 +0000)]
patch from elrond starting on lsarpcd conversion

24 years agopatch from lars (clean-up)
Luke Leighton [Sat, 12 Feb 2000 09:53:43 +0000 (09:53 +0000)]
patch from lars (clean-up)

24 years agooops.
Luke Leighton [Sat, 12 Feb 2000 09:27:46 +0000 (09:27 +0000)]
oops.

24 years agothis one serious-attitude bug-fix.
Luke Leighton [Sat, 12 Feb 2000 08:28:10 +0000 (08:28 +0000)]
this one serious-attitude bug-fix.

ok.

rpctorture showed that msrpc connections not being reused wasn't good
enough.  results: one smbd process taking in several connections
to \PIPE\NETLOGON, and there's a deliberate hard-limit of 64 pipes
at the moment.

so....

i rewrote the msrpc connection code so that if there is a pre-existing
msrpc connection *with the same user-context (vuser_key)*, it is reused
that means NO bind-request because it will already have been done.

what took me so long was that the user-contexts (vuser_key) had to be
checked.  the architecture at the moment sends vuser_key on the dce/rpc
bind assoc_gid and dce/rpc header context_id fields.  unfortunately,
these were going over-the-wire on SMB connections, which they should NOT
do, it's supposed to only be used on ncalrpc (local rpc) unix-socket
loopback.

oops.

it's dog-slow... BUT IT WORKS!

i now have to find a way to call cli_net_use_add() and then actually
get the damn code to _use_ the client-side connection!

i think what is happening is that _every_ single new msrpc pipe initiated
gets its own SMB session established, which is not very good!

24 years agohow to make other people _majorly_ unhappy. i was informed of the existence
Luke Leighton [Sat, 12 Feb 2000 04:04:55 +0000 (04:04 +0000)]
how to make other people _majorly_ unhappy.  i was informed of the existence
of the "indent" command, today :) guess what?  i really like it :)

24 years agofix from doug (nazard@dragoninc.on.ca) - deleted extra }
Luke Leighton [Sat, 12 Feb 2000 02:18:41 +0000 (02:18 +0000)]
fix from doug (nazard@dragoninc.on.ca) - deleted extra }

24 years agorenamed msrpc_state to msrpc_local.
Luke Leighton [Sat, 12 Feb 2000 01:19:21 +0000 (01:19 +0000)]
renamed msrpc_state to msrpc_local.

24 years agolars kneschke's _reg_open_entry() conversion. you missed NT_STATUS_NOPROBLEMO
Luke Leighton [Sat, 12 Feb 2000 00:38:44 +0000 (00:38 +0000)]
lars kneschke's _reg_open_entry() conversion.  you missed NT_STATUS_NOPROBLEMO
on the end

24 years agoupdated rpctorture.
Luke Leighton [Fri, 11 Feb 2000 23:51:58 +0000 (23:51 +0000)]
updated rpctorture.

as a result, found a client-connection (non-reuse) problem.  it's going
to require a minor redesign of the MSRPC client-connection architecture.
basically, i'm going to have to split "establish-connection-and-rpc-bind"
from "get-me-a-connection".  at present, "get-me-a-connection" also
does the rpc-bind bit, thereby producing potentially hundreds of msrpc
connections.

given that the samba msrpcd architecture forks() per connection, this isn't
such a good idea!

24 years agoFunction* doesn't work as a typecast on some systems.
Luke Leighton [Fri, 11 Feb 2000 19:33:30 +0000 (19:33 +0000)]
Function* doesn't work as a typecast on some systems.

24 years agopatch from lars for copyright update.
Luke Leighton [Fri, 11 Feb 2000 18:21:32 +0000 (18:21 +0000)]
patch from lars for copyright update.

24 years agopreparing for release of alpha.0.2
Samba Release Account [Fri, 11 Feb 2000 08:23:45 +0000 (08:23 +0000)]
preparing for release of alpha.0.2

24 years agodid someone mention a code-freeze? oh well.
Luke Leighton [Fri, 11 Feb 2000 08:16:45 +0000 (08:16 +0000)]
did someone mention a code-freeze?  oh well.

this trivial set of cosmetic changes creates a set of client-side rpc commands:

net, regedit, samedit, eventlog, spoolss, lsa, cmdat, svccontrol.

rpcclient is still supported.  the -S server option has been dropped
(except on the command-line), in favour of specifying \\servername
or //servername as the first argument.

24 years agoGeneral cleanup of debugs.
Tim Potter [Fri, 11 Feb 2000 00:22:51 +0000 (00:22 +0000)]
General cleanup of debugs.

Implemented mapping for well-known RIDs.

Split up into different bits.

24 years agoyes, it's a biggie. another semi-automated job, sed script by sander striker.
Luke Leighton [Fri, 11 Feb 2000 00:13:06 +0000 (00:13 +0000)]
yes, it's a biggie.  another semi-automated job, sed script by sander striker.

i had to use :map = /_io_^[bX$xA &&^[^M6ddcf(  ^[ in vi to modify what you
did, sander, to change this:
if (!xxx_io_xxx(....))
{
return False;
}

/* ..... */
if (rpc_con_pipe_req(...))
{
....

into this:
if (xxx_io_xxx(....) &&
    rpc_con_pipe_req(...))
{
....

because the return False; would have resulted in memory leaks.

24 years agoRemoved debugs.
Tim Potter [Thu, 10 Feb 2000 23:49:34 +0000 (23:49 +0000)]
Removed debugs.

Added post-processing of calls to winbindd daemon.

24 years agopatch from elrond for dynamic UNICODE string management.
Luke Leighton [Thu, 10 Feb 2000 23:30:53 +0000 (23:30 +0000)]
patch from elrond for dynamic UNICODE string management.

24 years agopatch from elrond to do [important] dynamic memory allocation of
Luke Leighton [Thu, 10 Feb 2000 22:57:02 +0000 (22:57 +0000)]
patch from elrond to do [important] dynamic memory allocation of
net_share_enum, client and server.

24 years agopatch from lars for reg conversion
Luke Leighton [Thu, 10 Feb 2000 22:28:56 +0000 (22:28 +0000)]
patch from lars for reg conversion

24 years agomissed one - msrpc_trust_account_passwd takes $MACHINE.ACC as arg.
Luke Leighton [Thu, 10 Feb 2000 17:25:31 +0000 (17:25 +0000)]
missed one - msrpc_trust_account_passwd takes $MACHINE.ACC as arg.

24 years agopatch from lars for _srv_close conversion
Luke Leighton [Thu, 10 Feb 2000 17:13:37 +0000 (17:13 +0000)]
patch from lars for _srv_close conversion

24 years agomade sure that the secrets tdb was only initialised once!
Luke Leighton [Thu, 10 Feb 2000 15:25:01 +0000 (15:25 +0000)]
made sure that the secrets tdb was only initialised once!

24 years agomoving away from trust_get_passwd(), using lsa_query_secret() on either
Luke Leighton [Thu, 10 Feb 2000 15:23:23 +0000 (15:23 +0000)]
moving away from trust_get_passwd(), using lsa_query_secret() on either
$MACHINE.ACC or G$$TRUSTDOMAINNAME, instead.

the DOMAINNAME.SERVERNAME.mac file is pre-loaded into the
domainname.servername.tdb secrets file when it is first created.

24 years agoyess :) fixed the client / server reusage mixup by adding code to msrpc_find
Luke Leighton [Thu, 10 Feb 2000 14:57:55 +0000 (14:57 +0000)]
yess :) fixed the client / server reusage mixup by adding code to msrpc_find
that checks the vuser_key.  all client-side connections must now be separated
by the client's security context (vuser_key).

24 years agointeresting. this is quite a problem, i don't know the solution, yet.
Luke Leighton [Thu, 10 Feb 2000 14:13:35 +0000 (14:13 +0000)]
interesting.  this is quite a problem, i don't know the solution, yet.

client-side security contexts from msrpc functions *called* in smbd
are being confused with server-side security contexts in pass-through
(loop-back) msrpc functions!

24 years agoyess! rpcclient no longer links with libsmbpw.so, which was so grossly
Luke Leighton [Thu, 10 Feb 2000 12:22:48 +0000 (12:22 +0000)]
yess!  rpcclient no longer links with libsmbpw.so, which was so grossly
unnecessary.

24 years agopatch from lars kneschke for _reg_close conversion
Luke Leighton [Thu, 10 Feb 2000 11:55:43 +0000 (11:55 +0000)]
patch from lars kneschke for _reg_close conversion

24 years agoadded _lsa_set_secret implementation. fixed nt_decrypt_string() it
Luke Leighton [Thu, 10 Feb 2000 10:31:22 +0000 (10:31 +0000)]
added _lsa_set_secret implementation.  fixed nt_decrypt_string() it
shouldn't be leaving out the length bit, that's kinda-needed!

24 years agoadded _lsa_create_secret implementation.
Luke Leighton [Thu, 10 Feb 2000 09:58:37 +0000 (09:58 +0000)]
added _lsa_create_secret implementation.

24 years agoadded _lsa_query_secret implementation.
Luke Leighton [Thu, 10 Feb 2000 09:34:12 +0000 (09:34 +0000)]
added _lsa_query_secret implementation.

24 years agoadded implementation of _lsa_open_secret.
Luke Leighton [Thu, 10 Feb 2000 08:31:34 +0000 (08:31 +0000)]
added implementation of _lsa_open_secret.

24 years agostub _lsa_create_secret.
Luke Leighton [Thu, 10 Feb 2000 06:30:50 +0000 (06:30 +0000)]
stub _lsa_create_secret.

created LSA_SECRET parsing structure.

24 years agowell _that_ was unbelievably easy!!! switched off read-only in netmon,
Luke Leighton [Thu, 10 Feb 2000 05:50:11 +0000 (05:50 +0000)]
well _that_ was unbelievably easy!!!  switched off read-only in netmon,
edited the opcode until i got the right one, then it turns out that
lsa_create_secret is identical to lsa_open_secret except for the opcode.

anyway, upshot is: added rpcclient createsecret command.

why am i doing this? :) because i want to add lsa_create_secret(),
lsa_set_secret() to lsarpcd, which will then allow me to create
trust accounts ($MACHINE.ACC) *without* writing them direct to
/usr/local/samba/private/DOMAINNAME.MACHINENAME.mac.

i'm fed up with rpcclient having passdb/*.c and groupdb/*.c code linked
in to it.

24 years agofinally got it. lsa_set_secret(). now only need lsa_create_secret().
Luke Leighton [Thu, 10 Feb 2000 05:25:47 +0000 (05:25 +0000)]
finally got it.  lsa_set_secret().  now only need lsa_create_secret().

24 years agofinally getting somewhere with querysecret, setsecret
Luke Leighton [Thu, 10 Feb 2000 05:15:23 +0000 (05:15 +0000)]
finally getting somewhere with querysecret, setsecret

24 years agoattempting to add an lsa_set_secret(), client-side. consequently,
Luke Leighton [Thu, 10 Feb 2000 02:52:43 +0000 (02:52 +0000)]
attempting to add an lsa_set_secret(), client-side.  consequently,
found a bug in lsa_open_pol2().

24 years agotracking down memory leaks with yamd.
Luke Leighton [Thu, 10 Feb 2000 02:10:42 +0000 (02:10 +0000)]
tracking down memory leaks with yamd.

24 years agomake proto
Tim Potter [Wed, 9 Feb 2000 23:57:04 +0000 (23:57 +0000)]
make proto

24 years agoAdded *.po
Tim Potter [Wed, 9 Feb 2000 23:48:17 +0000 (23:48 +0000)]
Added *.po

24 years agoPreliminary commit of ntdom nss module allowing users and groups to be
Tim Potter [Wed, 9 Feb 2000 23:47:04 +0000 (23:47 +0000)]
Preliminary commit of ntdom nss module allowing users and groups to be
resolved from a remote sam database on a nt machine.  This is pretty
rough at the moment:

  - server name hardcoded into winbindd.c
  - single threaded, blocking server
  - must rm -f /tmp/winbindd before running server
  - no memory leak checking
  - only _nss_ntdom_getpwnam_r function fully implemented

It's still pretty funky though.  I'm only committing it for tridge and
jeremy to have a look.  (-:

24 years agoThis commit was manufactured by cvs2svn to create branch 'SAMBA_TNG'.
cvs2svn Import User [Wed, 9 Feb 2000 23:47:03 +0000 (23:47 +0000)]
This commit was manufactured by cvs2svn to create branch 'SAMBA_TNG'.

24 years agopatch from lars kneschke (the first!) for _reg_close() conversion
Luke Leighton [Wed, 9 Feb 2000 22:32:25 +0000 (22:32 +0000)]
patch from lars kneschke (the first!) for _reg_close() conversion

24 years agodisabled samtdb user alias lookup, it's broken.
Luke Leighton [Wed, 9 Feb 2000 20:59:31 +0000 (20:59 +0000)]
disabled samtdb user alias lookup, it's broken.

24 years agoreplace standard_sub_basic() and samlogon_user with standard_sub_vuser.
Luke Leighton [Wed, 9 Feb 2000 19:24:28 +0000 (19:24 +0000)]
replace standard_sub_basic() and samlogon_user with standard_sub_vuser.
currently code does NOT pick up correct user-context, so %U and %G will
NOT be substituted.

24 years agofix from luke howard for --with-ldap compile.
Luke Leighton [Wed, 9 Feb 2000 17:00:15 +0000 (17:00 +0000)]
fix from luke howard for --with-ldap compile.

24 years agofixing _samr_set_userinfo() - just added it to samtdb. and the other
Luke Leighton [Wed, 9 Feb 2000 07:08:22 +0000 (07:08 +0000)]
fixing _samr_set_userinfo() - just added it to samtdb.  and the other
samdb impls (as a stub).

24 years agoadded start of samtdb group database
Luke Leighton [Wed, 9 Feb 2000 07:01:23 +0000 (07:01 +0000)]
added start of samtdb group database

24 years agooops, fixing --with-sam-pwdb= types
Luke Leighton [Tue, 8 Feb 2000 22:09:04 +0000 (22:09 +0000)]
oops, fixing --with-sam-pwdb= types

24 years agoweird. another make proto out-of-sync.
Luke Leighton [Tue, 8 Feb 2000 20:46:06 +0000 (20:46 +0000)]
weird.  another make proto out-of-sync.

24 years agoadded samrpassd and netlogonpassd which have to be explicitly compiled.
Luke Leighton [Tue, 8 Feb 2000 20:41:40 +0000 (20:41 +0000)]
added samrpassd and netlogonpassd which have to be explicitly compiled.
i compile --with-sampwdb=tdb, so if i want to test passdb/ i couldn't!

24 years agopatch from elrond
Luke Leighton [Tue, 8 Feb 2000 20:15:52 +0000 (20:15 +0000)]
patch from elrond

- --with-sam-pwdb=tdb,passdb,nt5ldap.

- code cleanups (made some fns static)

- reduced linking options to lsarpcd, it's really simple, now!

24 years agopatch from elrond for lsarpc conversion. this one's taking a long time,
Luke Leighton [Tue, 8 Feb 2000 20:06:04 +0000 (20:06 +0000)]
patch from elrond for lsarpc conversion.  this one's taking a long time,
it's not trivial!

24 years agopreparing for release of alpha.0.1
Samba Release Account [Tue, 8 Feb 2000 17:40:35 +0000 (17:40 +0000)]
preparing for release of alpha.0.1

24 years agohm, someone changed open_socket_in and didn't do a make proto :)
Luke Leighton [Tue, 8 Feb 2000 16:34:46 +0000 (16:34 +0000)]
hm, someone changed open_socket_in and didn't do a make proto :)

24 years agocreating user alias groups and user domain groups.
Luke Leighton [Tue, 8 Feb 2000 09:10:47 +0000 (09:10 +0000)]
creating user alias groups and user domain groups.

24 years agodisabled but pwnotreq accounts were being allowed login access, i don't
Luke Leighton [Tue, 8 Feb 2000 05:48:25 +0000 (05:48 +0000)]
disabled but pwnotreq accounts were being allowed login access, i don't
think this should be allowed.  disabled is disabled, end of story.

24 years agopolicy link for rest of msrpc daemons.
Luke Leighton [Tue, 8 Feb 2000 05:45:48 +0000 (05:45 +0000)]
policy link for rest of msrpc daemons.

fixing createuser command to take a -p option.  currently, the default
action for samrtdbd if no password is specified is to set the account
to "DISABLED" and "NOPASSWORDREQUIRED".  setting this with the
samuserset2 rpcclient command is a minor pain in the neck.

24 years agothis weird-looking code uses the user_struct* that was obtained from
Luke Leighton [Tue, 8 Feb 2000 05:02:33 +0000 (05:02 +0000)]
this weird-looking code uses the user_struct* that was obtained from
an SMBsesssetupX to provide a response to api_WkstaUserLogon and
api_NetUserGetInfo.

IF my suspicions are correct, an api_NetUserGetInfo or api_WkstaUserLogon
against an anonymous IPC$ connection will produce a failure, resulting
in the Win9x client DROPPING the anonymous connection and re-establishing
an authenticated SMBsesssetupX.

this will cause the smbd server to get a NET_USER_INFO_3 structure from
either the PDC or from itself (MSRPC remote or MSRPC loop-back, doesn't
matter which, it's all the same to domain_client_validate()), thence the
api_Net... or api_Wksta... call to follow will be provided with the correct
user logon info.

it also means that we can start filling in some of the "stub" fields,
such as last_logon_time, with _real_ info.  well, real, if netlogond
bothered to fill it in, but you get the idea :)

24 years ago1) made standard_sub_vuser take a const user_struct* instead of a const
Luke Leighton [Tue, 8 Feb 2000 04:25:56 +0000 (04:25 +0000)]
1) made standard_sub_vuser take a const user_struct* instead of a const
   vuser_key*.  this stops stupid-amounts of linking to all sorts of
   crap in programs like make_codepage and testprns, which know _nothing_
   of users.  the original link was just a temporary hack to get binaries.

2) make vuid_free_user_struct() free the vuser malloc'd structure, too.

24 years agofirst stage of making some sort of "thread" context ... er, thing :) in
Luke Leighton [Tue, 8 Feb 2000 03:43:58 +0000 (03:43 +0000)]
first stage of making some sort of "thread" context ... er, thing :) in
the msrpc code.  it's not really threads, it's just a thread context,
so that if different thread-contexts are requested, then the msrpc daemon
will at least be able to switch user-security-context.

eventually, i will have to go so far as to be able to reconstruct PDUs
depending on the user context, but that will require one socket per
thread-context, and some means to set that up *inside* the Bind/Bind-Request
processing code [argh!], because that's what triggers a "new" user-context,
really.

i hope.

24 years agoPort of rebind argument to lib/util_sock:open_socket_in() function.
Tim Potter [Tue, 8 Feb 2000 03:13:38 +0000 (03:13 +0000)]
Port of rebind argument to lib/util_sock:open_socket_in() function.

24 years agoHOORAY! password setting is back. this through fixing policy handle
Luke Leighton [Tue, 8 Feb 2000 00:36:10 +0000 (00:36 +0000)]
HOORAY!  password setting is back.  this through fixing policy handle
chain linking _and_ an odd bug where password_ok() was ZERO_STRUCTING
the NET_USER_INFO_3 structure and it was being stored in the vuser.tdb
table, blanking out the NT user info!

i added code to authorise_login() to get and then store the vuser info
after seeing it set vuser->guest = True.

i'm not sure i like that code...

24 years agohmm, looks like there's a bug in become_unix_sec_ctx(), it does a
Luke Leighton [Tue, 8 Feb 2000 00:15:08 +0000 (00:15 +0000)]
hmm, looks like there's a bug in become_unix_sec_ctx(), it does a
become_initial_uid and _then_ checks to see if the context hasn't
changed.  odd.

24 years agofinally got bugs fixed in rpcclient samuserset command. hint: don't
Luke Leighton [Tue, 8 Feb 2000 00:02:15 +0000 (00:02 +0000)]
finally got bugs fixed in rpcclient samuserset command.  hint: don't
typecast things to void*...

24 years agotrying to sort out policy handle state info, and making a mess of it.
Luke Leighton [Mon, 7 Feb 2000 23:46:09 +0000 (23:46 +0000)]
trying to sort out policy handle state info, and making a mess of it.
how to set up user-security-context info on a per-connection basis:
each POLICY_HND that is created must have a vuser_key associated with
it.  this must be transferred from POLICY_HND to POLICY_HND (e.g
samr_open_domain(connect_pol, &domain_pol)
{
policy_link_key(connect_pol, domain_pol);
}
so that the user session key can be picked up from a chain of samr_connect(),
samr_open_domain(), samr_open_user(), samr_set_userinfo().

the only consistent info is the POLICY_HND, passed from one call to the next.

24 years agomissed one vuid to user_key conversion.
Luke Leighton [Mon, 7 Feb 2000 20:06:25 +0000 (20:06 +0000)]
missed one vuid to user_key conversion.

24 years agoAdded check for SGI IRIX brokenness with semaphores when using gcc.
Jeremy Allison [Mon, 7 Feb 2000 14:27:04 +0000 (14:27 +0000)]
Added check for SGI IRIX brokenness with semaphores when using gcc.
Code from Don Badrak <dbadrak@census.gov>
Jeremy.

24 years agoSpelling flame. (-:
Tim Potter [Mon, 7 Feb 2000 05:26:41 +0000 (05:26 +0000)]
Spelling flame.  (-:

24 years agoargh. had to add a "minimalist" creds interface back in. reason:
Luke Leighton [Mon, 7 Feb 2000 03:08:57 +0000 (03:08 +0000)]
argh.  had to add a "minimalist" creds interface back in.  reason:
pass-through (rpcclient->smbd->msrpcdaemon) doesn't know anything about
how to let msrpcdaemon "inherit" the credentials of the smbd process.

i forgot about that.

24 years agooops, i put the lm# in the trust account not the nt#.
Luke Leighton [Mon, 7 Feb 2000 02:41:56 +0000 (02:41 +0000)]
oops, i put the lm# in the trust account not the nt#.

24 years agoAdded *.lo to .cvsignore file.
Tim Potter [Mon, 7 Feb 2000 01:24:41 +0000 (01:24 +0000)]
Added *.lo to .cvsignore file.

24 years agoFix some more compile warnings.
Tim Potter [Mon, 7 Feb 2000 00:41:58 +0000 (00:41 +0000)]
Fix some more compile warnings.

24 years agoFixed compile warning in locking.c:traverse_fn()
Tim Potter [Mon, 7 Feb 2000 00:24:04 +0000 (00:24 +0000)]
Fixed compile warning in locking.c:traverse_fn()

24 years agonew loop-back authentication mechanism.
Luke Leighton [Sun, 6 Feb 2000 20:38:47 +0000 (20:38 +0000)]
new loop-back authentication mechanism.

24 years agowe're back full circle! i removed the "creds" interface. it turns out
Luke Leighton [Sun, 6 Feb 2000 18:39:28 +0000 (18:39 +0000)]
we're back full circle!  i removed the "creds" interface.  it turns out
that there is, in the DCE/RPC bind request, a 32-bit field (assoc_gid).
i'm overloading that with the CALLER's pid.  combined with the context_id
field, which is 16-bit, that can be overloaded with the smbd vuid, we
have sufficient info to make MSRPC calls over loop-back that reference
the correct vuid database.

24 years agomake proto
Luke Leighton [Sun, 6 Feb 2000 16:54:45 +0000 (16:54 +0000)]
make proto

24 years agopatch from sander for svcctl conversion.
Luke Leighton [Sun, 6 Feb 2000 16:50:56 +0000 (16:50 +0000)]
patch from sander for svcctl conversion.

24 years agoimplemented invalidate_vuid as a tdb_delete.
Luke Leighton [Sun, 6 Feb 2000 05:30:23 +0000 (05:30 +0000)]
implemented invalidate_vuid as a tdb_delete.

24 years agothis is probably one of THE most reluctant commits i've ever made.
Luke Leighton [Sun, 6 Feb 2000 05:17:28 +0000 (05:17 +0000)]
this is probably one of THE most reluctant commits i've ever made.

it's far-reaching, and necessary.

this adds a vuser_struct tdb database, with a key [smbd's pid, vuid].
smbd uses it in every instance of standard_sub() and standard_sub_vuser().
that's almost every single SMB call for any IPC$ access.

the next stage is to remove sesssetup_user, probably sessetup_user_list too,
and review all occurences of standard_sub_basic used by smbd because if
they use standard_sub_basic() they might be expecting to read sesssetup_user,
and if they do _that_, they should be using standard_sub_vuser() instead.

all i wanted was a means to get vuids across to msrpc daemons.

24 years ago*sigh*. libsamnt5ldap doesn't compile. libsamtdb shouldn't be compiled
Luke Leighton [Sat, 5 Feb 2000 23:37:02 +0000 (23:37 +0000)]
*sigh*.  libsamnt5ldap doesn't compile.  libsamtdb shouldn't be compiled
either - not on a make install, anyway!

24 years agoadded libsampass* to install-lib SHARED_LIBS.
Luke Leighton [Sat, 5 Feb 2000 20:33:30 +0000 (20:33 +0000)]
added libsampass* to install-lib SHARED_LIBS.

24 years agoadding functions that are common-code between samrtdbd and samrnt5ldapd
Luke Leighton [Sat, 5 Feb 2000 18:19:35 +0000 (18:19 +0000)]
adding functions that are common-code between samrtdbd and samrnt5ldapd
so as not to waste luke howard's time.

security checks in netlogond on _net_srv_pwset(): account-type checking.

24 years agopatch from elrond for _lsa_query_info_policy conversion.
Luke Leighton [Sat, 5 Feb 2000 17:43:03 +0000 (17:43 +0000)]
patch from elrond for _lsa_query_info_policy conversion.

elrond, could you please collapse reply_xxx() functions into api_xxx()
as they are static?  makes the code look cleaner!  thx!

24 years agonetlogontdbd now links. i had to disable net_login_general() from
Luke Leighton [Sat, 5 Feb 2000 15:24:13 +0000 (15:24 +0000)]
netlogontdbd now links.  i had to disable net_login_general() from
using "update encrypted" because the method used (update_smbpasswordfile)
is, as you can see, specific to the smbpasswd file!

i need to create a "create SAM user" function for this to be done properly.

24 years agoreplacing enumeration of passdb/*.c in _net_sam_sync with _samr* API.
Luke Leighton [Fri, 4 Feb 2000 23:41:30 +0000 (23:41 +0000)]
replacing enumeration of passdb/*.c in _net_sam_sync with _samr* API.
netlogontdbd fails to link (deliberately) because it's not linked
with the passdb/*.c API.  by the time i'm finished, it will, because
there won't _be_ any references to the passdb/*.c API.

24 years agofinally getting somewhere by creating an _samr* API as a library.
Luke Leighton [Fri, 4 Feb 2000 22:38:07 +0000 (22:38 +0000)]
finally getting somewhere by creating an _samr* API as a library.

the idea is to never, ever call passdb/*.c or groupdb/*.c functions
[which are an abomination] but to use _samr_*() instead.

e.g, smbpasswd calls _samr_set_userinfo() instead of mod_smbpwd_entry().

24 years agodummy functions need to be unique.
Luke Leighton [Fri, 4 Feb 2000 21:13:39 +0000 (21:13 +0000)]
dummy functions need to be unique.

24 years agoconsistency updates to luke howard's work. make sure it compiles
Luke Leighton [Fri, 4 Feb 2000 19:23:49 +0000 (19:23 +0000)]
consistency updates to luke howard's work.  make sure it compiles
in non-ldap environments.  luke, i did this by moving the typedefs
for LDAPDB and related typedefs to _outside_ the #ifdef WITH_NT5LDAP