24 years agoSmall update to clitar.c to omit warnings about servers not
Richard Sharpe [Sat, 26 Sep 1998 00:41:20 +0000 (00:41 +0000)]
Small update to clitar.c to omit warnings about servers not
letting us change the date unless tar_real_noisy is True.

Also updated a few places where variables are declared but not set.

24 years Fixed bug with continuation line causing proto to fail.
Jeremy Allison [Fri, 25 Sep 1998 23:40:49 +0000 (23:40 +0000)] Fixed bug with continuation line causing proto to fail.
             Added $(PROGS) $(SPROGS) as targets for make clean.
acconfig.h: Added HAVE_IRIX_SPECIFIC_CAPABILITIES. Added sys/capability.h header check.
              Added function checks for srandom random srand rand.
includes.h: Added #include <sys/capability.h>.
ntdomain.h: Moved struct acct_info into here from smb.h
       Moved enum action_type into rpcclient.h
       Moved struct cli_state into client.h
       Moved struct nt_client_info, struct tar_client_info, struct client_info
             into rpcclient.h
lib/genrand.c: Changed to use sys_random() & friends.
lib/smbrun.c: Lose capabilities after fork.
lib/system.c: Added set_process_capability(), set_inherited_process_capability()
              sys_random(), sys_srandom().
lib/util.c: Added Ander's EFBIG lock check to fcntl_lock for 64 bit access to an
            32 bit mounted NFS filesystem.
nmbd/nmbd.c: Changed to use sys_random() & friends.
nmbd/nmbd_browsesync.c: Changed to use sys_random() & friends.
passdb/ldap.c: Missed one pdb_encode_acct_ctrl call.
passdb/passdb.c: Changed to Ander's code for ' ' characters.
passdb/smbpass.c: Added Ander's code to reset ACB_PWNOTREQ.
script/mkproto.awk: Added 'long' to prototypes.
smbd/chgpasswd.c: Lose capabilities after fork.
smbd/open.c: Do the mmap *after* the kernel oplock.
smbd/oplock.c: Removed stub code from kernel oplock path.
               Added set_process_capability(), set_inherited_process_capability() calls.
smbd/reply.c: Initialize count = 0, offset = 0.
smbd/server.c: Added set_process_capability(), set_inherited_process_capability() calls.
tests/summary.c: Ensure we have RANDOM or RAND.
utils/smbpasswd.c: Added Ander's code to reset ACB_PWNOTREQ.
utils/torture.c: Changed to use sys_random() & friends.


24 years agoAdded missing display enum.
Jeremy Allison [Fri, 25 Sep 1998 23:28:41 +0000 (23:28 +0000)]
Added missing display enum.

24 years agoFixed unused variable warning messages.
Jeremy Allison [Fri, 25 Sep 1998 23:24:28 +0000 (23:24 +0000)]
Fixed unused variable warning messages.

24 years agorpcclient: got samr command "enumusers -g -u" working
Luke Leighton [Fri, 25 Sep 1998 23:14:20 +0000 (23:14 +0000)]
rpcclient: got samr command "enumusers -g -u" working

24 years agoMoved the extra struct definitions Luke added to smb.h into rpcclient.h
Jeremy Allison [Fri, 25 Sep 1998 22:34:40 +0000 (22:34 +0000)]
Moved the extra struct definitions Luke added to smb.h into rpcclient.h
and client.h. They are still included from smb.h so nothing should
change (or break) but this keeps the new definitions easier to
maintain. This is similar to what I did with ntdomain.h

24 years agoadded in samr commands. assistance in returning the missing functions,
Luke Leighton [Fri, 25 Sep 1998 22:20:05 +0000 (22:20 +0000)]
added in samr commands.  assistance in returning the missing functions,
automatically removed because they were "unused", would be appreciated.

24 years agoadded wksinfo command to rpcclient, which don't work too good.
Luke Leighton [Fri, 25 Sep 1998 21:20:37 +0000 (21:20 +0000)]
added wksinfo command to rpcclient, which don't work too good.

24 years agoadded rpcclient program
Luke Leighton [Fri, 25 Sep 1998 21:01:52 +0000 (21:01 +0000)]
added rpcclient program

24 years agonot needed.
Luke Leighton [Fri, 25 Sep 1998 16:01:14 +0000 (16:01 +0000)]
not needed.

24 years agoI've disabled the conversion of null filenames to "." until we solve
Andrew Tridgell [Fri, 25 Sep 1998 03:52:29 +0000 (03:52 +0000)]
I've disabled the conversion of null filenames to "." until we solve
the win95 printing problem. I suspect it is a smbgetatr() problem with
a null name (which requires special behaviour).

This is an interim solution.

24 years agoUpdates to the scripts for the new 14-char space format of account
Jeremy Allison [Thu, 24 Sep 1998 22:35:16 +0000 (22:35 +0000)]
Updates to the scripts for the new 14-char space format of account

24 years agoIntegration of Anders Blomdell <>'s
Jeremy Allison [Thu, 24 Sep 1998 22:33:13 +0000 (22:33 +0000)]
Integration of Anders Blomdell <>'s
smbpasswd changes. Not exactly the same as his code - several

24 years agoqualifier name in session_enum was being assumed to exist. if NULL, the
Luke Leighton [Thu, 24 Sep 1998 20:19:31 +0000 (20:19 +0000)]
qualifier name in session_enum was being assumed to exist.  if NULL, the
name should not be read.  file_enum and connection_enum didn't do this:
wonder why only session_enum had this bug?

24 years agotook out #ifdef'd code in make_samr_r_query_usergroups that copied
Luke Leighton [Thu, 24 Sep 1998 20:17:19 +0000 (20:17 +0000)]
took out #ifdef'd code in make_samr_r_query_usergroups that copied
DOM_GID static-size array, because it's malloc'd now.

24 years agothis fixes the last of the issues where USRMGR.EXE and SRVMGR.EXE were
Luke Leighton [Thu, 24 Sep 1998 20:02:56 +0000 (20:02 +0000)]
this fixes the last of the issues where USRMGR.EXE and SRVMGR.EXE were
failing.  running these two programs on a samba pdc now work.


- removed __LINE__ from debug macros.

- removed call to SMB_ASSERT_ARRAY() in samr_r_query_usergroups.  the
  DOM_GID array is a malloc'd array (by make_dom_gids) not a static


- replaced all "struct smb_passwd"s with sam_passwds instead.  there
  were uid to rid confusion issues (assumptions that uids were equal
  to rids).

- #if 0'd the pdb_rid_is_user() call which, given the above corrections,
  would probably work now.


- replaced "struct smb_passwd" with sam_passwd in lookup_user_rid, as
  assumptions were being made that uids were equal to rids.

24 years agoremoved function names printed out in debug statements (duplicated in
Luke Leighton [Thu, 24 Sep 1998 19:57:02 +0000 (19:57 +0000)]
removed function names printed out in debug statements (duplicated in
DEBUG macro, automatically)

24 years agoreplaced one of the paragraphs describing nmbd problems. could someone
Luke Leighton [Thu, 24 Sep 1998 19:42:31 +0000 (19:42 +0000)]
replaced one of the paragraphs describing nmbd problems.  could someone
review this please?

24 years agoAdded a minor fix to clitar.c for a bug.
Richard Sharpe [Thu, 24 Sep 1998 13:43:36 +0000 (13:43 +0000)]
Added a minor fix to clitar.c for a bug.

Could not check that it compiles clean with Jeremy's -Wflags because
someone loaded some changes to reply.c that break in the locking area


24 years agocheck whether system type is the same as stored in the cache (full
Alexandre Oliva [Thu, 24 Sep 1998 08:34:21 +0000 (08:34 +0000)]
check whether system type is the same as stored in the cache (full
stand-alone macro, candidate for inclusion in the next release of autoconf)

create configure flag --enable-maintainer-mode, that enables automatic
rebuilding of configure from and aclocal.m4, and of based on acconfig.h.  This assumes autoconf and autoheader
are available.  Maintainer mode also introduces automatic dependency
tracking of include files, but it depends on features provided by gcc
and GNU make, so you must not enable maintainer mode if you are not
using any of these

24 years agoautomatically detect changes in the system type and exit, telling
Andrew Tridgell [Thu, 24 Sep 1998 04:52:40 +0000 (04:52 +0000)]
automatically detect changes in the system type and exit, telling
the user to remove config.cache

24 years agosome changes to the autoconf support
Andrew Tridgell [Thu, 24 Sep 1998 01:12:43 +0000 (01:12 +0000)]
some changes to the autoconf support

- added config.cache to "make clean". We get two many bug reports from
  people builing on multiple architectures who don't remove config.cache
  between compiles. Hopefully a future version of autoconf will do
  this automatically

- removed the targets that run autoconf and autoheader. There is far
  too great a risk of a enduser getting a error "autoheader not found"
  if the date on a file is changed. Samba developers will have to run
  autoconf and autoheader manually (as before)

24 years agonttrans.c:
Luke Leighton [Wed, 23 Sep 1998 21:49:09 +0000 (21:49 +0000)]

winreg was missing from the list of pipes.  advise using the array
already defined in... rpc_parse/parse_rpc.c's pipe_names[], but
writing code to strip "\PIPE\" from the front when making the

one location to update when adding new pipes, not two.


moved the ZERO_STRUCT(p) macro to _before_ the DLIST_ADD(Pipes, p) macro.


added { }s around the code inserted by DLIST_ADD and DLIST_REMOVE macros

24 years agoHoist by my own petard with warnings (forgot to set strict warnings
Jeremy Allison [Wed, 23 Sep 1998 17:56:34 +0000 (17:56 +0000)]
Hoist by my own petard with warnings (forgot to set strict warnings
in Makefile before compiling & checking in :-).

24 years agoFixed data corruption bugs in clitar.c with restores.
Richard Sharpe [Wed, 23 Sep 1998 14:37:01 +0000 (14:37 +0000)]
Fixed data corruption bugs in clitar.c with restores.

Have tested against samba with clitar using a hard-coded
max_xmit of 2920, since max smit = 2920 does not seem to work in
the smb.conf file.

Will have to test correctly against Win95 and WinNT now.

Have also compiled with -WJeremy'sFlags and get no more warnings
after I removed an unused variable.

24 years agosmbd/oplock.c: Use O_CREAT and O_TRUNC and correct mode flags
Jeremy Allison [Wed, 23 Sep 1998 01:58:27 +0000 (01:58 +0000)]
smbd/oplock.c: Use O_CREAT and O_TRUNC and correct mode flags
               when creating oplock test file.
smbd/server.c: Check for existance of kernel oplocks before
               becoming a daemon.

24 years agoFirst cut at kernel oplocks. This should have no effect unless runnin
Jeremy Allison [Wed, 23 Sep 1998 01:48:45 +0000 (01:48 +0000)]
First cut at kernel oplocks. This should have no effect unless runnin
on a machine that supports them in autoconf.

Move various functions out of lib/util.c into smbd/process.c
and smbd/oplock.c where they belong.


24 years agolook at the CAP_NT_SMBS bit in the client capabilities to determine if
Andrew Tridgell [Wed, 23 Sep 1998 01:25:33 +0000 (01:25 +0000)]
look at the CAP_NT_SMBS bit in the client capabilities to determine if
we should serve up volume labels as ascii or unicode.

NT wants ascii, W95 wants unicode. It's a crazy protocol!

24 years agoFixed up warnings in new client code.
Jeremy Allison [Wed, 23 Sep 1998 00:57:34 +0000 (00:57 +0000)]
Fixed up warnings in new client code.
Note to coders. If using gcc please use the compiler flags :
-Wall -Werror -Wshadow -Wstrict-prototypes -Wpointer-arith -Wcast-qual
*before* checking anything in to ensure a clean compile.

24 years agoreplace getpass() with getsmbpass() if getsmbpass.c compiles
Alexandre Oliva [Mon, 21 Sep 1998 11:34:44 +0000 (11:34 +0000)]
replace getpass() with getsmbpass() if getsmbpass.c compiles

24 years agomajor autoconf clean-up
Alexandre Oliva [Mon, 21 Sep 1998 09:07:08 +0000 (09:07 +0000)]
major autoconf clean-up
fix problems in builds with srcdir!=builddir

24 years agoimplemented du and tar -n
Alexandre Oliva [Mon, 21 Sep 1998 08:45:11 +0000 (08:45 +0000)]
implemented du and tar -n

24 years agolp_fstype() proto
Andrew Tridgell [Sun, 20 Sep 1998 15:48:35 +0000 (15:48 +0000)]
lp_fstype() proto

24 years ago3 changes:
Andrew Tridgell [Sun, 20 Sep 1998 15:48:10 +0000 (15:48 +0000)]
3 changes:

1) use lp_fstype() instead of FSTYPE_STRING
2) added SMB_SEARCH_BITS to the TconX reply options (in vwv3). I noted
   that NT sets this (undocumented) bit and setting it helped get
   autorun from exported cdroms working.
3) fixed volume labels in QFSINFO level 258.

I made these changes while getting the Encyclopadia Brittanica CD to
run from a Samba drive. (I bought it for Sue yesterday). The first
and second changes allowed Samba to export CDs with autorun info and
the client will autorun it when mounted. There are all sorts of nasty
implications in that that perhaps we can go into on
samba-technical. Think about creating some autorun info in /tmp/ then
waiting for people to mount it as scratch space ...

The last change was because EB wanted the right volume label. The code
we had used a non unicode volume label but tests with W95->NT4 showed
that it has to be unicode. There was a note in the code from Jeremy
saying that he thought it should _not_ be unicode. Jeremy, can you
explain why? It certainly didn't work as non-unicode (the client
displays a garbage volume label) and when I fixed it to use unicode it
all worked from Win95.

and in case anyone is interested EB98 now works fine from a Samba
drive :)

24 years agoadded a per-share parameter "fstype" that allows you to select the
Andrew Tridgell [Sun, 20 Sep 1998 15:40:52 +0000 (15:40 +0000)]
added a per-share parameter "fstype" that allows you to select the
filesystem type that will be reported to the client. If unspecified
then it defaults to FSTYPE_STRING (currently "Samba").

This is useful for cdroms where you want to set the fs type to CDFS.

24 years agoadd a define for SMB_SEARCH_BITS and change comment on FSTYPE_STRING
Andrew Tridgell [Sun, 20 Sep 1998 15:39:12 +0000 (15:39 +0000)]
add a define for SMB_SEARCH_BITS and change comment on FSTYPE_STRING

24 years agoFixed changes in clitar.c that Jeremey noted. One change was
Richard Sharpe [Sat, 19 Sep 1998 04:25:10 +0000 (04:25 +0000)]
Fixed changes in clitar.c that Jeremey noted. One change was
correct, the other needed slightly changing.

Now to test it is all OK :-)

24 years agoSmall bit of paranioa. Ensure that if the incoming name to
Jeremy Allison [Sat, 19 Sep 1998 03:34:12 +0000 (03:34 +0000)]
Small bit of paranioa. Ensure that if the incoming name to
unix_convert() was a single '\' (the base directory of the service)
that it gets translated to a '.', not a '\0'.

24 years agoRemove some byte-copying code in unix_format() that did nothing except
Jeremy Allison [Sat, 19 Sep 1998 02:35:43 +0000 (02:35 +0000)]
Remove some byte-copying code in unix_format() that did nothing except
slow down my benchmark :-).


24 years agoNow that Alexandre Oliva has explained the use of acconfig.h to
Jeremy Allison [Fri, 18 Sep 1998 21:57:12 +0000 (21:57 +0000)]
Now that Alexandre Oliva has explained the use of acconfig.h to
me, I'm adding the HAVE_LIBREADLINE and other stuff correctly.

Thanks Alexandre :-).


24 years agoFixed position of LARGE_SMB_OFF_T definition. The SOFF_T define was
Jeremy Allison [Fri, 18 Sep 1998 18:30:42 +0000 (18:30 +0000)]
Fixed position of LARGE_SMB_OFF_T definition. The SOFF_T define was
being done *before* the LARGE_SMB_OFF_T was being defined, but SOFF_T
depended on it :-).

This broke 64 bit file reporting.


24 years agoChanged variable TAB to htab as TAB is defined in a header file on RedHat5.1
Jeremy Allison [Fri, 18 Sep 1998 18:16:45 +0000 (18:16 +0000)]
Changed variable TAB to htab as TAB is defined in a header file on RedHat5.1

24 years agoFixed problems with premature kernel oplock checkin code.
Jeremy Allison [Fri, 18 Sep 1998 18:09:17 +0000 (18:09 +0000)]
Fixed problems with premature kernel oplock checkin code.
The ./configure & build now seem to work ok.


24 years agoconfigure include/ include/includes.h
Jeremy Allison [Fri, 18 Sep 1998 17:50:18 +0000 (17:50 +0000)]
configure include/ include/includes.h

   Fixed bugs in readline autoconf.

param/loadparm.c smbd/open.c smbd/oplock.c: Started on kernel oplock
        code - checking forced by above issue. Should not be used


24 years agoFixed compile errors in new code.
Jeremy Allison [Fri, 18 Sep 1998 17:32:35 +0000 (17:32 +0000)]
Fixed compile errors in new code.

24 years agopreparing for release of 2.0.0 alpha 6
Samba Release Account [Fri, 18 Sep 1998 14:05:05 +0000 (14:05 +0000)]
preparing for release of 2.0.0 alpha 6

24 years agooops ... I got the filenames wrong in my cleanup of the wins database
Andrew Tridgell [Fri, 18 Sep 1998 13:52:30 +0000 (13:52 +0000)]
oops ... I got the filenames wrong in my cleanup of the wins database
save, so it started saving weird filenames (from an unitialised

chaos reigns!

24 years agoAdding rewritten restore code ... Old code is still there
Richard Sharpe [Fri, 18 Sep 1998 12:47:46 +0000 (12:47 +0000)]
Adding rewritten restore code ... Old code is still there
surrounded by a OLD_DOTARPUT.

24 years agogot rid of SMB_STRUCT_STATVFS. I don't think we should be defining
Andrew Tridgell [Fri, 18 Sep 1998 03:53:14 +0000 (03:53 +0000)]
got rid of SMB_STRUCT_STATVFS. I don't think we should be defining
structures that only apply on some platforms.

24 years agogto ri of a bunch more #ifdef LARGE_SMB_OFF_T checks by introducing a
Andrew Tridgell [Fri, 18 Sep 1998 03:00:20 +0000 (03:00 +0000)]
gto ri of a bunch more #ifdef LARGE_SMB_OFF_T checks by introducing a
SOFF_T() macro for setting an SMB_OFF_T variable

also limited mmap based reads to MAX_MMAP_SIZE. We really can't mmap
2^50 bytes due to virtual address space problems.

24 years agoadded a SMB_OFF_T_BITS define, allowing us to get rid of most of the
Andrew Tridgell [Fri, 18 Sep 1998 02:30:03 +0000 (02:30 +0000)]
added a SMB_OFF_T_BITS define, allowing us to get rid of most of the
#ifdefs for LARGE_OFF_T.

24 years agofixed a typo (LLARGE_SMB_OFF_T instead of LARGE_SMB_OFF_T)
Andrew Tridgell [Fri, 18 Sep 1998 02:28:21 +0000 (02:28 +0000)]
fixed a typo (LLARGE_SMB_OFF_T instead of LARGE_SMB_OFF_T)

24 years agoremoved another use of the LL suffix. Hopefully this is the last one
Andrew Tridgell [Fri, 18 Sep 1998 02:27:24 +0000 (02:27 +0000)]
removed another use of the LL suffix. Hopefully this is the last one
(a grep doesn't show any more)

24 years agofixed a usage of off_t that should have been SMB_OFF_T
Andrew Tridgell [Fri, 18 Sep 1998 02:26:36 +0000 (02:26 +0000)]
fixed a usage of off_t that should have been SMB_OFF_T

24 years agogot rid of some #ifdef LARGE_XXXX stuff and got rid of non-portable LL
Andrew Tridgell [Fri, 18 Sep 1998 02:21:07 +0000 (02:21 +0000)]
got rid of some #ifdef LARGE_XXXX stuff and got rid of non-portable LL
suffix from some constants.

24 years agoinclude/includes.h:
Jeremy Allison [Fri, 18 Sep 1998 00:30:28 +0000 (00:30 +0000)]
lib/system.c: Can't assume every system has a statvfs varient.
              Return -1 for those that don't.
smbd/reply.c: Fixed printf warning.

24 years agonmbd/nmbd_winsserver.c: Fixed printf style warning.
Jeremy Allison [Fri, 18 Sep 1998 00:12:15 +0000 (00:12 +0000)]
nmbd/nmbd_winsserver.c: Fixed printf style warning.
script/mkproto.awk: Added SMB_BIG_UINT.

24 years agoconfigure Added checks for statvfs64. Last bit of 64 bit widening ...
Jeremy Allison [Thu, 17 Sep 1998 23:06:57 +0000 (23:06 +0000)]
configure Added checks for statvfs64. Last bit of 64 bit widening (I hope :-).
include/ Added #undef STAT_STATVFS64.
include/includes.h: Added SMB_STRUCT_STATVFS type, Changed SMB_BIG_INTEGER to
                    SMB_BIG_UINT and SMB_BIG_INT types.
include/smb.h: Added flag defines from CIFS spec.
lib/debug.c: Fixed one more mode_t issue.
lib/system.c: Added sys_statvfs wrapper.
lib/util.c: Changed trim_string to use size_t.
param/loadparm.c: Moved "blocking locks" into locking section. Alphabetised
                  locking options. Question - shuld we do this for all options ?
passdb/ldap.c: Changed SMB_BIG_INTEGER to SMB_BIG_UINT.
passdb/nispass.c: Changed SMB_BIG_INTEGER to SMB_BIG_UINT.
passdb/smbpass.c: Changed SMB_BIG_INTEGER to SMB_BIG_UINT.
smbd/dfree.c: Changed to use 64 bit types if available. Moved to use unsigned
smbd/dosmode.c: Fixed one more mode_t issue.
smbd/negprot.c: Changed literals to be FLAG_ #defines.
smbd/nttrans.c: Removed dead code.
smbd/open.c: Changed disk_free call.
smbd/process.c: Changed literals to be FLAG_ #defines.
smbd/reply.c: Changed disk_free call.
smbd/trans2.c: Fixed but in SMB_QUERY_FS_VOLUME_INFO call. Was using
               UNICODE - should use ascii.
tests/summary.c: Added STAT_STATVFS64 check.

24 years agoconfigure Added tests for fseek64 and ftell64.
Jeremy Allison [Thu, 17 Sep 1998 19:16:12 +0000 (19:16 +0000)]
configure Added tests for fseek64 and ftell64. Added fseek64 and ftell64.
includes.h: Added definition of SMB_BIG_INTEGER.
smb.h: Changed (*getsmbpwpos) and (*setsmbpwpos) to use SMB_BIG_INTEGER.
access.c: Tidyup of dbug statement.
system.c: Added sys_fseek and sys_ftell. Changed mode calls to use mode_t.
asyncdns.c: Tidyup of comment.
loadparm.c: Tidyup of set_default_server_announce_type() function definition.
ldap.c: Changed (*getsmbpwpos) and (*setsmbpwpos) to use SMB_BIG_INTEGER.
nispass.c: Changed (*getsmbpwpos) and (*setsmbpwpos) to use SMB_BIG_INTEGER.
smbpass.c: Changed (*getsmbpwpos) and (*setsmbpwpos) to use SMB_BIG_INTEGER.
smbpassfile.c: Use sys_fseek().
chgpasswd.c: Tidyup of debug statement.
dosmode.c: Changed mode calls to use mode_t.
ipc.c: Removal of dead code.
nttrans.c: Changed mode calls to use mode_t.
open.c: Changed mode calls to use mode_t.
pipes.c: Removal of dead code.
reply.c: Removal of dead code.
trans2.c: Removal of dead code. Changed mode calls to use mode_t.

24 years agoconfigure, Added fseek64 and ftell64.
Jeremy Allison [Thu, 17 Sep 1998 19:09:16 +0000 (19:09 +0000)]
configure, Added fseek64 and ftell64. Added fseek64 and ftell64.
includes.h: Added definition of SMB_BIG_INTEGER.
smb.h: Changed (*getsmbpwpos) and (*setsmbpwpos) to use SMB_BIG_INTEGER.
access.c: Tidyup of dbug statement.
system.c: Added sys_fseek and sys_ftell. Changed mode calls to use mode_t.
asyncdns.c: Tidyup of comment.
loadparm.c: Tidyup of set_default_server_announce_type() function definition.
ldap.c: Changed (*getsmbpwpos) and (*setsmbpwpos) to use SMB_BIG_INTEGER.
nispass.c: Changed (*getsmbpwpos) and (*setsmbpwpos) to use SMB_BIG_INTEGER.
smbpass.c: Changed (*getsmbpwpos) and (*setsmbpwpos) to use SMB_BIG_INTEGER.
smbpassfile.c: Use sys_fseek().
chgpasswd.c: Tidyup of debug statement.
dosmode.c: Changed mode calls to use mode_t.
ipc.c: Removal of dead code.
nttrans.c: Changed mode calls to use mode_t.
open.c: Changed mode calls to use mode_t.
pipes.c: Removal of dead code.
reply.c: Removal of dead code.
trans2.c: Removal of dead code. Changed mode calls to use mode_t.

24 years agopreparing for release of 2.0.0 alpha 5
Samba Release Account [Thu, 17 Sep 1998 08:35:56 +0000 (08:35 +0000)]
preparing for release of 2.0.0 alpha 5

24 years agofixed another potential fork bomb where the wins file becomes
Andrew Tridgell [Thu, 17 Sep 1998 08:35:07 +0000 (08:35 +0000)]
fixed another potential fork bomb where the wins file becomes
non-writeable for some reason.

24 years agofixed the nmbd fork bomb. It was a silly mistake, as
Andrew Tridgell [Thu, 17 Sep 1998 08:27:46 +0000 (08:27 +0000)]
fixed the nmbd fork bomb. It was a silly mistake, as
expected. wins_write_database() didn't exit after doing its stuff, so
when it returned you had two copies of nmbd :)

24 years agofixed a typo in my last commit
Andrew Tridgell [Thu, 17 Sep 1998 08:23:00 +0000 (08:23 +0000)]
fixed a typo in my last commit

24 years agofixed a potential problem with wins_write_database() child processes.
Andrew Tridgell [Thu, 17 Sep 1998 06:36:08 +0000 (06:36 +0000)]
fixed a potential problem with wins_write_database() child processes.

In sig_term() we were calling wins_write_database(0) which would fork a
child. This child might then get killed by the same process killing
off the parent. That process would then fork another child etc.

The solution is to pass a "background" flag to wins_write_database(0)
and only fork if this is set.

24 years agoRemoved hideous inefficiencies in old trim_string code. This was making
Jeremy Allison [Tue, 15 Sep 1998 01:38:10 +0000 (01:38 +0000)]
Removed hideous inefficiencies in old trim_string code. This was making
calls to strlen() a profiling hotspot.

24 years agoMapped correct open modes for READ_CONTROL access.
Jeremy Allison [Mon, 14 Sep 1998 20:43:50 +0000 (20:43 +0000)]
Mapped correct open modes for READ_CONTROL access.

24 years agoFixed problems people were having with creating profile
Jeremy Allison [Mon, 14 Sep 1998 19:49:55 +0000 (19:49 +0000)]
Fixed problems people were having with creating profile
directories (NTTrans/Create with Security Descriptor for
a directory). It turns out the CIFS spec is bogus (what a
suprise) and the 'is a directory' flag is actually embedded
in the create_options field.


24 years agoOk so with this bugfix 64 bit file access actually seems to work :-).
Jeremy Allison [Fri, 11 Sep 1998 21:42:18 +0000 (21:42 +0000)]
Ok so with this bugfix 64 bit file access actually seems to work :-).

Problems were just dumb bugs like (defining sys_lseek to return 'int' DOH !).


24 years agoOk - this is the 'expose 64 bit to the clients' checkin.
Jeremy Allison [Fri, 11 Sep 1998 19:14:27 +0000 (19:14 +0000)]
Ok - this is the 'expose 64 bit to the clients' checkin.

I have tested it by creating a 'holey' 20GB file - checking that
it shows up correctl in the NT file view (it does) and am busily
copying it to NULL: on the NT box. All good so far.... :-).

Also implemented NT 'delete on close' semantics.


24 years agoAdded ssize_t to configure code.
Jeremy Allison [Fri, 11 Sep 1998 01:24:30 +0000 (01:24 +0000)]
Added ssize_t to configure code.
Got 'religion' about using size_t and ssize_t for read/write stuff
as part of the code to expose 64 bits to the client.

This checkin does all the 'easy' stuff - such as all the read/write/lock
calls - but now comes the harder parts (open & friends) and all the
file enquiry functions.....


24 years agosmb.h: Removed fdnum from file_fd_struct. Not needed.
Jeremy Allison [Thu, 10 Sep 1998 18:57:06 +0000 (18:57 +0000)]
smb.h: Removed fdnum from file_fd_struct. Not needed.
files.c: Removed fd bitmap - not needed. Added code to do use arrays rather
         than linked list - disabled by default but can be enabled to check

24 years agofixed a bug in the wins database writer that caused the database to be
Andrew Tridgell [Thu, 10 Sep 1998 04:00:09 +0000 (04:00 +0000)]
fixed a bug in the wins database writer that caused the database to be
written continuously rather than once every 5 seconds (at most).

also changed it to 20 seconds :)

24 years agoImproved stat cache code by uppercasing any search name that gets added to
Jeremy Allison [Thu, 10 Sep 1998 00:35:10 +0000 (00:35 +0000)]
Improved stat cache code by uppercasing any search name that gets added to
it if we're in case insensitive mode, and then doing a memcmp rather
than a StrnCaseCmp (which is *horribly* slow) on every lookup.
Fixed bug with refusing NT SMB's (use *brackets* where needed :-).

24 years agoCopyright notices written to debug logs are now through 1998.
Christopher R. Hertel [Wed, 9 Sep 1998 16:37:54 +0000 (16:37 +0000)]
Copyright notices written to debug logs are now through 1998.

Chris -)-----

24 years agoAdded back groupname map stuff removed by Andrew's "slash 'n' burn"
Jeremy Allison [Tue, 8 Sep 1998 19:21:04 +0000 (19:21 +0000)]
Added back groupname map stuff removed by Andrew's "slash 'n' burn"
tactics :-). Protected by #ifdef until used.

Fixed bug in fd_attempt_close() where a pointer to potentially
free'd memory was returned. I hate that.

Added "blocking locks" as a per-share option for performance testing.

Changed is_mangled() so it will return true if called with a pathname
and any component of the pathname was mangled (it was already attempting
to do this, but not checking for a '/' as end-of-mangle).

This should be a better fix for the wierd stat cache bug Andrew identified.


24 years agoadded 0x10 to the frag length in the "bind ack".
Luke Leighton [Mon, 7 Sep 1998 17:01:52 +0000 (17:01 +0000)]
added 0x10 to the frag length in the "bind ack".

24 years agoAdding header and footer html files for SWAT, adding samba.gif file too.
John Terpstra [Sun, 6 Sep 1998 05:50:18 +0000 (05:50 +0000)]
Adding header and footer html files for SWAT, adding samba.gif file too.

24 years agoadd a "stat cache" boolean smb.conf option. (defaults to on)
Andrew Tridgell [Sun, 6 Sep 1998 01:37:14 +0000 (01:37 +0000)]
add a "stat cache" boolean smb.conf option. (defaults to on)

I think we need this so we can rule out stat cache bugs when dealing
with bug reports. If we ask a user to disable the stat cache and the
problem persists then we know it isn't a stat cache bug. The stat
cache code is sufficiently complicated that it can be pretty hard to
tell if it is causing problems or not.

24 years agofixed a stat cache bug (the one found by Matthew Geier).
Andrew Tridgell [Sat, 5 Sep 1998 15:31:10 +0000 (15:31 +0000)]
fixed a stat cache bug (the one found by Matthew Geier).

The fix I used is a very conservative fix. I'll leave it up to Jeremy
to put in a better fix. The problem was the detection of mangled

24 years agoadded a configuration summary at the end of ./configure. It also
Andrew Tridgell [Sat, 5 Sep 1998 13:53:13 +0000 (13:53 +0000)]
added a configuration summary at the end of ./configure. It also
aborts if essential functions are not available.

24 years agoexpand the sysv shmem test to look for semaphores as well as shared
Andrew Tridgell [Sat, 5 Sep 1998 13:35:54 +0000 (13:35 +0000)]
expand the sysv shmem test to look for semaphores as well as shared
memory. Some FreeBSD systems have sysv shared memory but no

24 years agosome cleanups to use ZERO_STRUCT() and friends
Andrew Tridgell [Sat, 5 Sep 1998 13:24:20 +0000 (13:24 +0000)]
some cleanups to use ZERO_STRUCT() and friends

24 years agosome people are foolishly running ./configure from other than the
Andrew Tridgell [Sat, 5 Sep 1998 13:21:57 +0000 (13:21 +0000)]
some people are foolishly running ./configure from other than the
source directory! This adds a new "trivial" test that makes sure that
the autoconf tests are working. It aborts the configure run if not.

24 years agoahh, the joy of deleting large chunks of code that someone else has
Andrew Tridgell [Sat, 5 Sep 1998 05:23:29 +0000 (05:23 +0000)]
ahh, the joy of deleting large chunks of code that someone else has
painstakingly put in :)

This gets rid of most of the #ifdef LARGE_SMB_INO_T ifdefs around
DEBUG() statements. We just use %.0f in all cases. Makes the code a
bit easier to read :)

24 years agotridge the destroyer returns!
Andrew Tridgell [Sat, 5 Sep 1998 05:07:05 +0000 (05:07 +0000)]
tridge the destroyer returns!

prompted by the interpret_security() dead code that Jean-Francois
pointed out I added a make target "finddead" that finds potentially
dead (ie. unused) code. It spat out 304 function names ...

I went through these are deleted many of them, making others static
(finddead also reports functions that are used only in the local

in doing this I have almost certainly deleted some useful code. I may
have even prevented compilation with some compile options. I
apologise. I decided it was better to get rid of this code now and add
back the one or two functions that are needed than to keep all this

So, if I have done a bit too much "destroying" then let me know. Keep
the swearing to a minimum :)

One bit I didn't do is the ubibt code. Chris, can you look at that?
Heaps of unused functions there. Can they be made static?

24 years agoFixed one more Debug problem with inode in non-64 bit case.
Jeremy Allison [Sat, 5 Sep 1998 03:38:50 +0000 (03:38 +0000)]
Fixed one more Debug problem with inode in non-64 bit case.

24 years agoBugfix for leak in reference counted file struct.
Jeremy Allison [Sat, 5 Sep 1998 03:14:40 +0000 (03:14 +0000)]
Bugfix for leak in reference counted file struct.
Added "nt smb support" parameter to allow NT SMB's to be turned off.

24 years agogot rid of interpret_security(). Thanks to Jean-Francois for pointing
Andrew Tridgell [Sat, 5 Sep 1998 01:12:07 +0000 (01:12 +0000)]
got rid of interpret_security(). Thanks to Jean-Francois for pointing
out it is no longer used (replaced by enumerated types in loadparm.c)

24 years agoModified dev_t and ino_t code to be 64 bit clean (including changes
Jeremy Allison [Fri, 4 Sep 1998 20:53:58 +0000 (20:53 +0000)]
Modified dev_t and ino_t code to be 64 bit clean (including changes
to oplock break message passing). I think that smbd/nmbd are now
inode and offset size independent (at least for 32 bit and 64 bit

Now to expose all this new functionality to NT clients.....


24 years agoFixed uppercasing of share name (pointed out by Jean-Francois).
Jeremy Allison [Fri, 4 Sep 1998 16:06:04 +0000 (16:06 +0000)]
Fixed uppercasing of share name (pointed out by Jean-Francois).

24 years agoMore 64 bit stuff - now the fcntl locks are 64 bit clean.
Jeremy Allison [Fri, 4 Sep 1998 00:23:28 +0000 (00:23 +0000)]
More 64 bit stuff - now the fcntl locks are 64 bit clean.
Nearly at the stage where I can expose the 64-bit-ness to
the NT clients....

24 years agoOk - this is the 64 bit widening check in. It changes the configure
Jeremy Allison [Thu, 3 Sep 1998 18:40:31 +0000 (18:40 +0000)]
Ok - this is the 64 bit widening check in. It changes the configure
to check for stat64 and friends, and then changes much of Samba
to use the data type SMB_OFF_T for file size information.

stat/fstat/lstat/lseek/ftruncate have now become sys_stat etc. to hide
the 64 bit calls if needed.

Note that this still does not expose 64 bit functionality to the
client, as the changes to the reply_xxx smb's are not yet done.

This code change should make these changes possible.

Still to do before full 64 bit-ness to the client:

fcntl lock code.
statfs code
widening of dev_t and ino_t (now possible due to SMB_DEV_T and SMB_OFF_T
types being in place).

Let me know if wierd things happen after this check-in and I'll
fix them :-).


24 years agochanged the SMBtrans reply code to align at the same alignment as
Andrew Tridgell [Thu, 3 Sep 1998 05:37:15 +0000 (05:37 +0000)]
changed the SMBtrans reply code to align at the same alignment as
observed from NT. We were aligning the data before but not the
parameters. This aligns both.

This change may break some other parts of ipc.c if we have relied on
the alignment somewhere, so if you think this broke something then let
me know.

24 years agowe were setting the strings 1 too long in make_srv_share_info1_str()
Andrew Tridgell [Thu, 3 Sep 1998 05:35:13 +0000 (05:35 +0000)]
we were setting the strings 1 too long in make_srv_share_info1_str()

24 years agofixed a bug in the name mangling code. It implicitly assumed that
Andrew Tridgell [Thu, 3 Sep 1998 03:14:31 +0000 (03:14 +0000)]
fixed a bug in the name mangling code. It implicitly assumed that
mangling a name can't increase it's size which isn't true. (imagine a
file called "L B" which mangles to "LB~XX")

The symptoms were that users couldn't run batch files from short
directory names that contained non 8.3 characters (such as spaces).

24 years agoFix for PR#9497 - not waiting for child.
Jeremy Allison [Wed, 2 Sep 1998 17:23:41 +0000 (17:23 +0000)]
Fix for PR#9497 - not waiting for child.

24 years agowe are never interested in SIGPIPE so just ignore (block) it
Andrew Tridgell [Wed, 2 Sep 1998 05:00:47 +0000 (05:00 +0000)]
we are never interested in SIGPIPE so just ignore (block) it
always. Don't even install a handler.

24 years agochanged ref to samba.gif to use /swat/ prefix
Andrew Tridgell [Wed, 2 Sep 1998 02:12:07 +0000 (02:12 +0000)]
changed ref to samba.gif to use /swat/ prefix

24 years agouse /swat/ prefix in both inetd and cgi modes, to enable a static header.html
Andrew Tridgell [Wed, 2 Sep 1998 02:02:30 +0000 (02:02 +0000)]
use /swat/ prefix in both inetd and cgi modes, to enable a static header.html