ira/wip.git
25 years agoThis commit was manufactured by cvs2svn to create tag samba-2.0.4
cvs2svn Import User [Tue, 18 May 1999 00:39:42 +0000 (00:39 +0000)]
This commit was manufactured by cvs2svn to create tag
'release-2-0-4'.

25 years agopreparing for release of 2.0.4
Samba Release Account [Tue, 18 May 1999 00:39:41 +0000 (00:39 +0000)]
preparing for release of 2.0.4

25 years agoFixed bug where copymap was referenced if malloc fails.
Jeremy Allison [Mon, 17 May 1999 23:37:24 +0000 (23:37 +0000)]
Fixed bug where copymap was referenced if malloc fails.
Jeremy.

25 years agoRemoved the sockethack code (sorry Andrew :-). Added the recursive directory
Jeremy Allison [Mon, 17 May 1999 22:28:11 +0000 (22:28 +0000)]
Removed the sockethack code (sorry Andrew :-). Added the recursive directory
smbclient bugfix (sorely needed).
Jeremy.

25 years agoadded a --with-sockethack option that should give us an _enormous_
Andrew Tridgell [Sun, 16 May 1999 12:14:31 +0000 (12:14 +0000)]
added a --with-sockethack option that should give us an _enormous_
boost in netbench performance on Linux/SMP if the corresponding 4 line
change is made in the kernel. Note: this is a experimental performance
hack only at the moment. Do _not_ enable it until we find a way to
make it safe in the kernel, which may not be for a while.

25 years agoadded full netbench simulation in smbtorture. It now reads a 4MB load
Andrew Tridgell [Sat, 15 May 1999 06:20:33 +0000 (06:20 +0000)]
added full netbench simulation in smbtorture. It now reads a 4MB load
description file that was derived from a packet capture of a real
netbench run. That file describes every SMB operation that a netbench
client performs on the wire and smbtorture uses that to simulate any
number of simultaneous netbench clients, giving results on the same
scale that netbench does.

the nice thing is that it all seems to work. I'm actually getting
results pretty close to what a real netbench lab gets.

25 years agoUpdated Roadmap as we have met one our our objectives
Samba Release Account [Sat, 15 May 1999 02:03:14 +0000 (02:03 +0000)]
Updated Roadmap as we have met one our our objectives
(the "NT Security release" *is* the 2.0.4 release).
Jeremy.

25 years agoFixed typo.
Samba Release Account [Sat, 15 May 1999 02:00:00 +0000 (02:00 +0000)]
Fixed typo.
Jeremy.

25 years agoFixes due to DEBUG format %d needing (int) casting for
Jeremy Allison [Sat, 15 May 1999 00:06:34 +0000 (00:06 +0000)]
Fixes due to DEBUG format %d needing (int) casting for
arguments on AIX. This fixes many cases where we are
printing a size_t and using a format of %d.
Jeremy.

25 years agoUpdated for 2.0.4
Jeremy Allison [Thu, 13 May 1999 18:03:58 +0000 (18:03 +0000)]
Updated for 2.0.4
Jeremy.

25 years agoModified post install script so xtoi is run only if /etc/xinetd.conf exists.
John Terpstra [Wed, 12 May 1999 20:11:29 +0000 (20:11 +0000)]
Modified post install script so xtoi is run only if /etc/xinetd.conf exists.

25 years agoExplicit cast to int before mask...
Jeremy Allison [Wed, 12 May 1999 19:48:04 +0000 (19:48 +0000)]
Explicit cast to int before mask...
Jeremy.

25 years agoEnsure that when a pipe number is extracted from the incoming buffer not
Jeremy Allison [Wed, 12 May 1999 19:34:09 +0000 (19:34 +0000)]
Ensure that when a pipe number is extracted from the incoming buffer not
using SVAL, that we mask off the lower 16 bits to ensure correct comparisons.
Jeremy.

25 years agoAdded new package components (not for general distribution).
John Terpstra [Tue, 11 May 1999 22:53:21 +0000 (22:53 +0000)]
Added new package components (not for general distribution).

25 years agouse a separate directory for each cleint in netbench tests
Andrew Tridgell [Mon, 10 May 1999 23:19:16 +0000 (23:19 +0000)]
use a separate directory for each cleint in netbench tests

25 years agoprint a + at the start of the unlink phase, so we can see it more
Andrew Tridgell [Sun, 9 May 1999 11:07:47 +0000 (11:07 +0000)]
print a + at the start of the unlink phase, so we can see it more
clearly

25 years agofixed typo
Andrew Tridgell [Sun, 9 May 1999 09:55:24 +0000 (09:55 +0000)]
fixed typo

25 years agoset a 2 minute timeout on SMB packets in the torture code.
Andrew Tridgell [Sun, 9 May 1999 09:53:08 +0000 (09:53 +0000)]
set a 2 minute timeout on SMB packets in the torture code.

25 years agobetter filename choice
Andrew Tridgell [Sun, 9 May 1999 08:40:25 +0000 (08:40 +0000)]
better filename choice

25 years agouse a better synchronisation method at startup
Andrew Tridgell [Sun, 9 May 1999 08:30:13 +0000 (08:30 +0000)]
use a better synchronisation method at startup

25 years agodon't delete files till all are written. This stresses the disk
Andrew Tridgell [Sun, 9 May 1999 05:02:14 +0000 (05:02 +0000)]
don't delete files till all are written. This stresses the disk
subsystem in a more similar way to a netbench run.

25 years agoadded synchronisation code at startup so all clients start at the same
Andrew Tridgell [Sun, 9 May 1999 04:45:09 +0000 (04:45 +0000)]
added synchronisation code at startup so all clients start at the same
time. That gives a more accurate result.

25 years agoimproved the netbench simulator a bit. give results as throughput in
Andrew Tridgell [Sun, 9 May 1999 03:51:44 +0000 (03:51 +0000)]
improved the netbench simulator a bit. give results as throughput in
MB/sec

25 years agoadded a very simple netbench simulator to smbtorture. The tests are
Andrew Tridgell [Sun, 9 May 1999 03:00:10 +0000 (03:00 +0000)]
added a very simple netbench simulator to smbtorture. The tests are
called NBW95 and NBWNT.

I don't have the right sort of hardware (100MBit etc) to test whether
the results of this similator are anything like what netbench sees,
but I'm hoping Jeremy will run that test for me.

A typical run would be:

smbtorture //server/share -Uuser%pass -o 10 -N 20 NBWNT NBW95

that would run a NT and a 95 client test with 20 clients and 10 files
per client. The result is given as seconds for the run.

I think that for best results you would use a few copies of smbtorture
running on different clients and each similating maybe 10 clients to
build up a large client load with a small number of client machines.

25 years agomodified smbtorture so you can run individual tests using command line
Andrew Tridgell [Sun, 9 May 1999 00:37:56 +0000 (00:37 +0000)]
modified smbtorture so you can run individual tests using command line
parameters. For example:

smbtorture //fjall/tmp RANDOMIPC LOCK2 ATTR

would run those 3 tests. ALL will run all tests. The default is ALL.

25 years agoThe install instructions file is UNIX_INSTALL.txt not UNIX-INSTALL.txt.
Matthew Chapman [Sat, 8 May 1999 06:23:11 +0000 (06:23 +0000)]
The install instructions file is UNIX_INSTALL.txt not UNIX-INSTALL.txt.
Never underestimate the stupidity of SysAdmins.

25 years agoFix check return for sys_open.
Jeremy Allison [Wed, 5 May 1999 18:32:52 +0000 (18:32 +0000)]
Fix check return for sys_open.
Jeremy.

25 years agoTidyups caught by running gcc with -Wall. It gets a different set from the IRIX compiler.
Jeremy Allison [Wed, 5 May 1999 17:06:54 +0000 (17:06 +0000)]
Tidyups caught by running gcc with -Wall. It gets a different set from the IRIX compiler.
Jeremy.

25 years agoSize bugfix for smbsh (patch from Dave Smith dsmith@tqs.com).
Jeremy Allison [Tue, 4 May 1999 01:05:08 +0000 (01:05 +0000)]
Size bugfix for smbsh (patch from Dave Smith dsmith@tqs.com).
Jeremy.

25 years agoRemoved the null packet send. This code worked perfectly with IRIX
Jeremy Allison [Mon, 3 May 1999 19:31:25 +0000 (19:31 +0000)]
Removed the null packet send. This code worked perfectly with IRIX
and kills Win9x clients with a Linux server. I need to go back and
check the WinNT traces to make sure of this.....
Jeremy.

25 years agoAdded "oplock break wait time" parameter to help with Win9x oplock related
Jeremy Allison [Mon, 3 May 1999 17:33:10 +0000 (17:33 +0000)]
Added "oplock break wait time" parameter to help with Win9x oplock related
problems. Default set to 10ms.
Also documented it :-).
Jeremy.

25 years agoFixed (I think) the last outstanding bug before 2.0.4. This was the lotus
Jeremy Allison [Thu, 29 Apr 1999 01:53:15 +0000 (01:53 +0000)]
Fixed (I think) the last outstanding bug before 2.0.4. This was the lotus
notes client problem where we were too generous in allowing a fallback
to opening a directory as a file (PR#15717). I added code to ensure that only non
oplock opens with no read/write access requested fall back to a directory
open.
This seems to work with the NT rename problem as well as correctly
stopping "type \directory".

Jeremy.

25 years agoRe-wrote prs_force_grow() as my fist implementation was crap. Now only
Jeremy Allison [Wed, 28 Apr 1999 22:21:43 +0000 (22:21 +0000)]
Re-wrote prs_force_grow() as my fist implementation was crap. Now only
grow by the requested amount in this case.
Jeremy.

25 years agoFixed the security=domain which seemed to have been broken. Hmmm. Wonder
Jeremy Allison [Wed, 28 Apr 1999 21:44:59 +0000 (21:44 +0000)]
Fixed the security=domain which seemed to have been broken. Hmmm. Wonder
when. The code in rpc_client/cli_pipe.c wasn't coping with a ERRmoredata
error correctly, and the rpc_read code wasn't growing the parse buffer
correctly either.....

Also made smbd and nmbd ignore SIGFPE for systems that have it.

Jeremy.

25 years agogot rid of /tmp time based random source. I saw a system with a huge
Andrew Tridgell [Wed, 28 Apr 1999 02:01:33 +0000 (02:01 +0000)]
got rid of /tmp time based random source. I saw a system with a huge
number of files in /tmp and logging in was taking a _long_ time.

25 years agoCIFS spec states (and I've seen on the wire, so I believe it now :-) that
Jeremy Allison [Tue, 27 Apr 1999 01:23:01 +0000 (01:23 +0000)]
CIFS spec states (and I've seen on the wire, so I believe it now :-) that
the offset in an SMBlseek is a *signed* value on send, and an unsigned
value on return. In addition, setting the position to before the start
of file should implicitly set the file pointer to zero (this wasn't being
done correctly before).

As SMBlseek is only used to determine file length this shouldn't be a
problem but I have a nice smbd log file showing an SMBlseek set of offset
0xFFFFFFFF failing that caused this change :-).

Jeremy.

25 years agoAdded SO_RCVBUF and SO_SNDBUF options to socket options.
John Terpstra [Mon, 26 Apr 1999 23:08:18 +0000 (23:08 +0000)]
Added SO_RCVBUF and SO_SNDBUF options to socket options.

25 years agoAdded SO_RCVBUF and SO_SNDBUF to socket options.
John Terpstra [Mon, 26 Apr 1999 23:07:31 +0000 (23:07 +0000)]
Added SO_RCVBUF and SO_SNDBUF to socket options.

25 years agoFixed memory leak in do_get (oh, the shame :-).
Jeremy Allison [Mon, 26 Apr 1999 23:02:06 +0000 (23:02 +0000)]
Fixed memory leak in do_get (oh, the shame :-).
Jeremy.

25 years agoAdded documentation for "mangle lock" parameter.
Jeremy Allison [Mon, 26 Apr 1999 21:23:13 +0000 (21:23 +0000)]
Added documentation for "mangle lock" parameter.
Jeremy.

25 years agoinclude/rpc_dce.h:
Jeremy Allison [Mon, 26 Apr 1999 21:17:52 +0000 (21:17 +0000)]
include/rpc_dce.h:
rpc_server/srv_pipe.c: Caused bind_nak to be sent to pipe bind request that we
                       don't support (WINREG). Stops "smb too short" errors from
                       appearing in the NT eventlog.
param/loadparm.c:
smbd/blocking.c:
smbd/reply.c: Changed "truncate locks" to "mangle locks" to match documentation.
passdb/smbpasschange.c: Fix SWAT crash bug when local UNIX user doesn't exist when adding
                        a new smbpasswd user.
Jeremy.

25 years agoENCRYPTION.txt: Updated descriptive format of smbpasswd to actuallly match code.
Jeremy Allison [Mon, 26 Apr 1999 17:20:41 +0000 (17:20 +0000)]
ENCRYPTION.txt: Updated descriptive format of smbpasswd to actuallly match code.
Passwords.txt: Added "OR security=domain".
Jeremy.

25 years agoTidy up only.
John Terpstra [Sun, 25 Apr 1999 12:15:44 +0000 (12:15 +0000)]
Tidy up only.

25 years agoAdded smbsh.1 man page.
John Terpstra [Sun, 25 Apr 1999 12:10:45 +0000 (12:10 +0000)]
Added smbsh.1 man page.

25 years agoAdded smbsh.1 man page.
John Terpstra [Sun, 25 Apr 1999 11:48:49 +0000 (11:48 +0000)]
Added smbsh.1 man page.

25 years agoUpdated samba2.spec.tmpl to store nmblokkup.1 and swat.8 man pages as well
John Terpstra [Sun, 25 Apr 1999 11:15:56 +0000 (11:15 +0000)]
Updated samba2.spec.tmpl to store nmblokkup.1 and swat.8 man pages as well
as to create an /etc/xinetd.conf file.

25 years agoFixed the bug that causes the master parent smbd to crash rather than
Jeremy Allison [Fri, 23 Apr 1999 23:46:51 +0000 (23:46 +0000)]
Fixed the bug that causes the master parent smbd to crash rather than
shutdown cleanly. It was one of the oplock fixes that broke it.
Jeremy.

25 years agoFixed postun bug reported by Markus Hillig.
John Terpstra [Fri, 23 Apr 1999 22:51:49 +0000 (22:51 +0000)]
Fixed postun bug reported by Markus Hillig.

25 years agoFixed postun handling bug reported by Markus Hillig.
John Terpstra [Fri, 23 Apr 1999 22:50:34 +0000 (22:50 +0000)]
Fixed postun handling bug reported by Markus Hillig.

25 years agoFixing compile warnings reported by Henrik Lassen <henrik@lassen.dk>.
Matthew Chapman [Fri, 23 Apr 1999 16:35:06 +0000 (16:35 +0000)]
Fixing compile warnings reported by Henrik Lassen <henrik@lassen.dk>.

25 years agoAdded 'truncate locks' parameter that only have meaning on 32-bit systems.
Jeremy Allison [Wed, 21 Apr 1999 20:27:51 +0000 (20:27 +0000)]
Added 'truncate locks' parameter that only have meaning on 32-bit systems.
NT (Access to be precise) is sending 64 bit lock offsets and expecting
them to be honored. I now do some tricky mangling stuff on 32-bit systems
to try and map the 64-bit lock offset range into the 31-bit one mandated
by 32-bit POSIX. Roll-on 64-bit Linux.....

Jeremy.

25 years agoTurns out that the logon/logoff NET_Q_SAM_LOGON/NET_Q_SAM_LOGOFF
Jeremy Allison [Tue, 20 Apr 1999 01:01:44 +0000 (01:01 +0000)]
Turns out that the logon/logoff NET_Q_SAM_LOGON/NET_Q_SAM_LOGOFF
packets are different. Logon contains a uint16 validation_level,
logoff does not. This may explain the error that the older Samba
logoff code would get from an NT PDC (sending a packet 2 bytes
too long).
Jeremy.

25 years agoFixed some stupid crash bugs in the PDC user login code. Added
Jeremy Allison [Mon, 19 Apr 1999 21:49:22 +0000 (21:49 +0000)]
Fixed some stupid crash bugs in the PDC user login code. Added
return checks to all user login functions.
Jeremy.

25 years agoFix for delete veto files from oota@pes.com1.fc.nec.co.jp. Oota Toshiya.
Jeremy Allison [Mon, 19 Apr 1999 20:03:43 +0000 (20:03 +0000)]
Fix for delete veto files from oota@pes.com1.fc.nec.co.jp. Oota Toshiya.
Jeremy.

25 years agoI finally wrote that smb filter program that Jeremy and I have talked
Andrew Tridgell [Sat, 17 Apr 1999 10:51:56 +0000 (10:51 +0000)]
I finally wrote that smb filter program that Jeremy and I have talked
about so often. It is a simple plug proxy that listens on port 139 and
redirects connections to another machine on the same port. The nice
thing is that you can filter the packets on the way through.

So, for example, if you want to see what NT server does with a NT
client when unicode isn't negotiated just set CAPABILITY_MASK and
CLI_CAPABILITY_MASK to CAP_UNICODE and run "smbfilter desthost
DESTNAME". Then connect to desthost with the NT client. The filter
will create a connection to desthost, set the destination netbios name
to DESTNAME in the session request and change the negotiated
capabilities on the connection to exclude unicode. Then you just fire
up a sniffer and you have a perfect record of how a NT server responds
to a NT client when it doesn't negotiate unicode.

There are lots of nice things we can do with this. I'm especially
looking forward to confirming some suspected bugs in NT clients by
reproducing them with a MS server with different negotiated
options.

to build it do "make smbfilter". The only documentation is the source
code and this commit message.

25 years agoUpdated docs. Added NT_Security info document and referenced it in SWAT.
Jeremy Allison [Fri, 16 Apr 1999 23:20:24 +0000 (23:20 +0000)]
Updated docs. Added NT_Security info document and referenced it in SWAT.
Added comment about "wide linke" negatively affecting performance.
Jeremy.

25 years agoNew documentation explaining how the security dialog stuff works.
Jeremy Allison [Fri, 16 Apr 1999 22:22:38 +0000 (22:22 +0000)]
New documentation explaining how the security dialog stuff works.
Jeremy.

25 years agoMaking chown work broke recursive set, due to the owner_sid not explicitly
Jeremy Allison [Wed, 14 Apr 1999 22:50:33 +0000 (22:50 +0000)]
Making chown work broke recursive set, due to the owner_sid not explicitly
being referenced in the SECURITY_INFORMATION field in a set packet. Now I
allow the name lookup to sometimes fail to map to a UNIX id (for chown
where the group SID is bogus) and yet still reference the owner and group
fields in the SID even if they're not explicitly referenced in the SECURITY_INFORMATION field
(for set).
Jeremy.

25 years agolib/util_sid.c: Fixed a crash bug blooper.
Jeremy Allison [Wed, 14 Apr 1999 22:10:50 +0000 (22:10 +0000)]
lib/util_sid.c: Fixed a crash bug blooper.
rpc_parse/parse_sec.c smbd/nttrans.c: Fixed code so that a command line NT "chown"
     command now works against a Samba share.
Jeremy.

25 years agoModified permission mappings to ignore "take owner" and "delete attribute"
Jeremy Allison [Wed, 14 Apr 1999 00:04:32 +0000 (00:04 +0000)]
Modified permission mappings to ignore "take owner" and "delete attribute"
and other things that cannot be mapped correctly on SD set.
Jeremy.

25 years agoFixed testparm errors.
Jeremy Allison [Tue, 13 Apr 1999 22:54:19 +0000 (22:54 +0000)]
Fixed testparm errors.
Jeremy.

25 years agoCorrectly handle "CONTAINER_INHERIT" & "OBJECT_INHERIT" bits. Set debug
Jeremy Allison [Tue, 13 Apr 1999 20:10:33 +0000 (20:10 +0000)]
Correctly handle "CONTAINER_INHERIT" & "OBJECT_INHERIT" bits. Set debug
level of message when invalid bits sent to 1.
Jeremy.

25 years agoMakefile.in: Added installservers target at user request.
Jeremy Allison [Tue, 13 Apr 1999 18:36:35 +0000 (18:36 +0000)]
Makefile.in: Added installservers target at user request.
smbd/nttrans.c: Fixed get/set of directory permissions so that the modification of
                permissions of files within a directory and in a tree can be done
                correctly.
utils/testparm.c: Added password server checks.
Jeremy.

25 years agocli_write needs to accept replies in any order - remember NT is a
Matthew Chapman [Mon, 12 Apr 1999 14:52:41 +0000 (14:52 +0000)]
cli_write needs to accept replies in any order - remember NT is a
multithreaded monster.

25 years agotest for libdl rather than just always including it
Andrew Tridgell [Mon, 12 Apr 1999 13:11:19 +0000 (13:11 +0000)]
test for libdl rather than just always including it

25 years agoFixed the problem where is a file had no permissions allowed for its
Jeremy Allison [Sat, 10 Apr 1999 00:33:39 +0000 (00:33 +0000)]
Fixed the problem where is a file had no permissions allowed for its
owner, the owner couldn't modify the permissions. This was caused
by NT needing an open file handle to access file permissions, which
of course was being denied by the UNIX permissions. Fixed this by
allowing a pseudo "stat" file handle, similar to the "directory"
handle, which allows NT clients to get/set permissions.
Also fixed a couple of small purify-found memory leaks.
Jeremy.

25 years agoFixed a couple of purify unuitialized memory read bugs.
Jeremy Allison [Fri, 9 Apr 1999 18:28:13 +0000 (18:28 +0000)]
Fixed a couple of purify unuitialized memory read bugs.
Jeremy.

25 years agoPut this back - it dropped off.
John Terpstra [Fri, 9 Apr 1999 13:29:02 +0000 (13:29 +0000)]
Put this back - it dropped off.

25 years agowe need libdl when linking with libpam on some systems (eg PHT TL), so
Andrew Tridgell [Fri, 9 Apr 1999 13:02:59 +0000 (13:02 +0000)]
we need libdl when linking with libpam on some systems (eg PHT TL), so
do it on all as it seems harmless

25 years agoOverload the WRITE_OWNER bit to mean UNIX "no access".
Jeremy Allison [Fri, 9 Apr 1999 00:03:48 +0000 (00:03 +0000)]
Overload the WRITE_OWNER bit to mean UNIX "no access".
This works rather well, if a little unintuitive. It allows
a fixed set of permissions to always be seen, even if an
NT user deletes permissions then they appear as "O" when
the permissions are next viewed. It allows the owner of
a file to always be able to manipulate the group and world
permissions.
Now I need to write a document explaining how all of this
works for the 2.0.4 release :-).
Jeremy.

25 years agoStop sub files/directories getting 0 permissions when
Jeremy Allison [Thu, 8 Apr 1999 22:14:36 +0000 (22:14 +0000)]
Stop sub files/directories getting 0 permissions when
an ACL with zero ACE entries is sent on the "set permissions on all
sub-directories" dialog box. Hmmmm. Currently return "permission denied".
May have to cope with this in future.
Jeremy.

25 years agoAdded modification of security descriptors. This is somewhat restricted
Jeremy Allison [Thu, 8 Apr 1999 21:13:02 +0000 (21:13 +0000)]
Added modification of security descriptors. This is somewhat restricted
in that you can only modify what permissions are there to start with,
and one a catagory (user/group/other) is removed you cannot add them
back again, but it allows usefull modification of existing permsissions.
Jeremy.

25 years agoUpdated docs as NT acl support is now on by default.
Jeremy Allison [Thu, 8 Apr 1999 01:48:53 +0000 (01:48 +0000)]
Updated docs as NT acl support is now on by default.
param/loadparm.c: NT acl support is now on by default.
source/smbd/reply.c: Only set global_client_caps on first sessionsetup.
source/smbd/trans2.c: Use global_client_caps instead of client type to
determine if we do the unicode hack.
Jeremy.

25 years agoclient/client.c: Fixed problem where doing smbclient -L //name did a lookup
Jeremy Allison [Thu, 8 Apr 1999 01:30:33 +0000 (01:30 +0000)]
client/client.c: Fixed problem where doing smbclient -L //name did a lookup
on "//name".
smbd/nttrans.c: Return correct "changenotify cancelled" error.
smbd/trans2.c: Fixed display of security descriptors on mapped drives.
This was *PAINFULL* to find & fix.
Jeremy.

25 years agosmbd/chgpasswd.c smbd/reply.c: Tidied up int -> uid_t or gid_t types.
Jeremy Allison [Wed, 7 Apr 1999 22:36:56 +0000 (22:36 +0000)]
smbd/chgpasswd.c smbd/reply.c: Tidied up int -> uid_t or gid_t types.
smbd/trans2.c: Fixed *nasty* little bug where NT does a setfileinfo with
a mode of 0 on *viewing* a file properties - this would change the UNIX
mode on a file, not what was desired.
Jeremy.

25 years agoFixed a couple of bugs in displaying a UNIX permission as an NT
Jeremy Allison [Wed, 7 Apr 1999 20:00:37 +0000 (20:00 +0000)]
Fixed a couple of bugs in displaying a UNIX permission as an NT
ACL. Firstly, UNIX groups should be returned as SID_NAME_ALIAS
not SID_NAME_DOM_GRP. Secondly I was mapping the group entry of
a UNIX permission to a user rid, not a group rid (that was just
dumb :-). Now the full permissions are seen correctly.
Jeremy.

25 years agoDomain logon bugfix from Dejan Ilic <svedja@lysator.liu.se>.
Jeremy Allison [Wed, 7 Apr 1999 19:26:03 +0000 (19:26 +0000)]
Domain logon bugfix from Dejan Ilic <svedja@lysator.liu.se>.
Jeremy.

25 years agoAdded "restrict anonymous" parameter from Andy Bakun <abakun@reac.com>.
Jeremy Allison [Wed, 7 Apr 1999 19:22:01 +0000 (19:22 +0000)]
Added "restrict anonymous" parameter from Andy Bakun <abakun@reac.com>.
Jeremy.

25 years agoFixed the debug messages about buffer overrun (removed unneeded
Jeremy Allison [Tue, 6 Apr 1999 23:02:02 +0000 (23:02 +0000)]
Fixed the debug messages about buffer overrun (removed unneeded
prs_align calls) when doing a lookup of user info on reading a
NT acl.
Jeremy.

25 years agoThis code implements the NT right click -> security -> ownership and
Jeremy Allison [Tue, 6 Apr 1999 00:39:49 +0000 (00:39 +0000)]
This code implements the NT right click -> security -> ownership and
permissions tabs.
Using this code you can now see UNIX users and groups as owners and
have the standard UNIX permissions mapped into NT ACLs.
There are still a couple of rough edges, but it's a significant
improvement on the code that was in the 2.0.x tree.

Jeremy.

25 years agoRemoved the 'ALLOW_IN_KERNEL_OPLOCK' code - it was causing problems in
Jeremy Allison [Fri, 2 Apr 1999 02:02:50 +0000 (02:02 +0000)]
Removed the 'ALLOW_IN_KERNEL_OPLOCK' code - it was causing problems in
NetBench runs plus it didn't cure the underlying race condition anyway.
rpc_parse/parse_sec.c: Corrected code for calculating SD size.
Jeremy.

25 years agoFixed the share enum code to be dynamic. I have now tested this with
Jeremy Allison [Thu, 1 Apr 1999 20:00:47 +0000 (20:00 +0000)]
Fixed the share enum code to be dynamic. I have now tested this with
returning >1200 share to NT and it works correctly.

I also now have access to Purify (yipee!) and have run it at the
highest warning level on Samba. It fixed a couple of uninitialized
memory reads and has caused some tidyups.

Jeremy.

25 years agorpc_server/srv_pipe.c: Fixed signed & sealed RPC calls.
Jeremy Allison [Wed, 31 Mar 1999 00:09:22 +0000 (00:09 +0000)]
rpc_server/srv_pipe.c: Fixed signed & sealed RPC calls.
This one line typo took over 6 hours to track down. ARGGHHHHH!!!!!!

include/trans2.h: Preparing for correct SMB_QUERY_FS_DEVICE_INFO reply.
All other changes - fixed casts to dump_data when compiling with
-DDEBUG_PASSWORDS
Jeremy.

25 years agodeclare ttarf as double, as in client.c
Alexandre Oliva [Tue, 30 Mar 1999 10:41:06 +0000 (10:41 +0000)]
declare ttarf as double, as in client.c

25 years agomade dir_total a double so it can be larger than 2^32 on all systems.
Andrew Tridgell [Tue, 30 Mar 1999 06:35:03 +0000 (06:35 +0000)]
made dir_total a double so it can be larger than 2^32 on all systems.

25 years agoclient/client.c: Fixed format problem with %d -> %.0f.
Jeremy Allison [Tue, 30 Mar 1999 06:13:42 +0000 (06:13 +0000)]
client/client.c: Fixed format problem with %d -> %.0f.
rpc_client/cli_srvsvc.c
rpcclient/cmd_lsarpc.c
rpcclient/cmd_reg.c
rpcclient/cmd_samr.c: Fixed rpcclient compile problems introduced by
the new RPC code.
Jeremy.

25 years agofix a couple of minor glitches in the new rpc code. Just little things
Andrew Tridgell [Tue, 30 Mar 1999 03:32:18 +0000 (03:32 +0000)]
fix a couple of minor glitches in the new rpc code. Just little things
like unused variables and wrong number of parameters for a debug
format.

25 years agoThis file is no longer needed (thank goodness....).
Jeremy Allison [Tue, 30 Mar 1999 02:21:11 +0000 (02:21 +0000)]
This file is no longer needed (thank goodness....).
Jeremy.

25 years agoWell, here are 28,000 lines of changes that re-write the
Jeremy Allison [Tue, 30 Mar 1999 02:19:10 +0000 (02:19 +0000)]
Well, here are 28,000 lines of changes that re-write the
RPC subsystem.
But they're really good :-).
Jeremy.

25 years agoFixed the issue with HPUX 10.20 and bigcrypt. HPUX has bigcrypt in
Jeremy Allison [Fri, 26 Mar 1999 21:25:40 +0000 (21:25 +0000)]
Fixed the issue with HPUX 10.20 and bigcrypt. HPUX has bigcrypt in
its C library, but the password hashes don't use it. I fixed this
by adding a 'USE_BOTH_CRYPT_CALLS' define for HPUX and causing
this to check first bigcrypt then crypt.
I'm not immensely happy with this, but can't think of a good
way to test if the password file uses crypt or bigcrypt.
Jeremy.

25 years agoAdded PAM support back in now that it defaults to off.
John Terpstra [Fri, 26 Mar 1999 11:54:52 +0000 (11:54 +0000)]
Added PAM support back in now that it defaults to off.

25 years agoAdded PAM support to configure now that it defaults to off.
John Terpstra [Fri, 26 Mar 1999 11:54:03 +0000 (11:54 +0000)]
Added PAM support to configure now that it defaults to off.

25 years agoadded --with-pam to redhat spec file. This may be needed in other
Andrew Tridgell [Fri, 26 Mar 1999 11:31:19 +0000 (11:31 +0000)]
added --with-pam to redhat spec file. This may be needed in other
packaging files, but I don't know which ones.

25 years agoonly use PAM if it is explicitly configured, rather than
Andrew Tridgell [Fri, 26 Mar 1999 11:28:35 +0000 (11:28 +0000)]
only use PAM if it is explicitly configured, rather than
auto-detecting it and sing it by default. Too many OSes now have
"sleeping" PAM implementations where they have the PAM calls but don't
use PAM for their authentication. This confuses the hell out of people
who install Samba and because they don't have a /etc/pam.d/samba file
find they can't authenticate.

25 years agochanged from %llu to %.0f in order to remain portable while retaining
Andrew Tridgell [Fri, 26 Mar 1999 11:10:26 +0000 (11:10 +0000)]
changed from %llu to %.0f in order to remain portable while retaining
the ability to handle very large numbers. (double is good for about 51
bits with 1 byte accuracy. That is plenty).

25 years agoTighten up the smb requests we can process when using kernel oplocks
Jeremy Allison [Fri, 26 Mar 1999 01:23:12 +0000 (01:23 +0000)]
Tighten up the smb requests we can process when using kernel oplocks
in order to prevent deadlock.
This cures most deadlock problems but there is still one remaining
deadlock condition that will have to be fixed in the kernel code -
shame :-(.
Jeremy.

25 years agofirst cut at a simple man page for smbsh
Herb Lewis [Thu, 25 Mar 1999 21:49:53 +0000 (21:49 +0000)]
first cut at a simple man page for smbsh

25 years agoother formats for swat man page
Herb Lewis [Thu, 25 Mar 1999 21:48:18 +0000 (21:48 +0000)]
other formats for swat man page

25 years ago* client/client.c (dir_total): use SMB_BIG_UINT
Alexandre Oliva [Thu, 25 Mar 1999 19:52:45 +0000 (19:52 +0000)]
* client/client.c (dir_total): use SMB_BIG_UINT
* client/clitar.c (ttarf): ditto
* lib/snprintf.c: support long longs; adapted from Cloyce D. Spradling's
patch <cloyce@headgear.org>

25 years agofix minor useage problem in manpagename macro
Herb Lewis [Thu, 25 Mar 1999 17:51:06 +0000 (17:51 +0000)]
fix minor useage problem in manpagename macro