Age | Commit message (Collapse) | Author | Files | Lines |
|
username case conversion wrapper. It is now (very) explicit where we are
mapping between an incoming Windows username, and when we are doing a
UNIX password entry lookup, which may change the case of the given
username.
This makes things *much* clearer (IMHO:-) and will ease the adding of
the 'groupname map' parameter, and the addition of the special 'jeremy'
mode for Samba where unix users will not be needed. (We must think of
a better name for it :-).
Jeremy.
(This used to be commit fb6ed81844e7cb6049749e43ac9b4adfaf4ca2de)
|
|
Jeremy.
(This used to be commit 1d77728bbba49699a05eb70b685a4a97d7598122)
|
|
lib/rpc/include/rpc_lsa.h: Changed #defines for RPC calls - moved some, made LSA_LOOKUPNAMES correct.
lib/rpc/include/rpc_misc.h: Changed DOMAIN_ALIAS_xxx to BUILTIN_ALIAS_xxx.
Changed bitmasks for uid to rid to be 1 bit.
lib/rpc/parse/parse_misc.c: Changed make_unistr2 to put length as given, max length as one more.
lib/rpc/server/srv_netlog.c: Removed 'domain other sids' parameter.
lib/rpc/server/srv_samr.c: Changed DOMAIN_ALIAS_xxx to BUILTIN_ALIAS_xxx.
lib/rpc/server/srv_util.c: Changed DOMAIN_ALIAS_xxx to BUILTIN_ALIAS_xxx.
Jeremy.
(This used to be commit 34c91840a3d8c252715dc2f749b7a3b171a5b74f)
|
|
Chris's reformating of the (rather long named function) find_name_for_remote_broadcast_subnet
(moving the function name onto a line on it's own) caused the proto
awk script to miss it.
Jeremy.
(This used to be commit 17c758687f0ec6040633bc1815a52627b7e15f02)
|
|
split the name_record structure into pieces. The goal is that the key
(the name) be separate from the data associated with the key. Databases
such as gdbm store information in [key,content] pairs.
There is no functional change in with this update. It's just a step in
the direction that Jeremy and I have been discussing.
Chris -)-----
(This used to be commit e420a4bd7d368a0e910893400fb7b46ab8694a08)
|
|
seen by the nameserve.h header.
(This used to be commit 530b54b8a2234a72d7a12606bff38648e56c5c1c)
|
|
*Not* the usual. I did make proto to add the #ifndef..#endif block.
Just to be safe, I did a cvs diff and found no prototype changes.
ubiqx/sys_includes.h
More fudging. I want to use binary tree typdefs in nameserv.h, but
nameserv.h is included in includes.h which is included by all of the
ubiqx modules. The result is that the types are referenced before
they are declared. My solution is to prevent nameserv.h and proto.h
from being included in the ubiqx modules (where they are not needed
anyway).
Chris -)-----
(This used to be commit f128fda8e6abdbd9ae661dd17684580969ef5068)
|
|
as NT 4.x does directed broadcast node status requests for the *<0x0> name.
Jeremy.
(This used to be commit 8c6fe8870a72271a6acd1633efc362c59e283e19)
|
|
#ifndef _PROTO_H_
#define _PROTO_H_
:
:
#endif /* _PROTO_H_ */
in the appropriate places. This required that I add an END rule.
I generated a new proto.h and diff'ed it against the previous. There were
no differences except for the added lines.
Chris -)-----
(This used to be commit 6d4d0dc7e65f946f09d127969ccbb4b51dfd6a4e)
|
|
the return value was being ignored anyway. SGI's lint said:
function returns value which is sometimes ignored
add_name_to_subnet
Chris -)-----
(This used to be commit ae706bff10cc77f06b8069e637ec9768d6a46966)
|
|
preserve'.
Also removed 'domain allow/deny' parameters.
shmem_sysv.c: Added strerror code so I can see why sysV calls are failing.
lib/rpc/server/srv_netlog.c: Removed code that used 'domain allow/deny' parameters -
replaced with a comment so I will remember to fix this later.
Jeremy.
(This used to be commit 5f46c7c4b11a75f1ffbd806cde915b4bf28232db)
|
|
available) when scanning server/workgroup/share lists.
ipc.c: Removed redundent definition of ERROR_MORE_DATA (234) we already have
ERRmoredata in smb.h
Jeremy.
(This used to be commit ae3510e9fb309be343e245d8460c1ceb8164077b)
|
|
(This used to be commit 2cb94eec59bdcc1272063637bbc8b7a5449fdc62)
|
|
because "domain admin users" and "domain guest users" was overloaded.
incorrectly.
(This used to be commit 04b824007263ac4879c7282a2d230deaac7d2c7b)
|
|
cause nmbd to run out of fd's. Test case found by Eloy Paris.
Jeremy.
(This used to be commit 9e2570317138cc6a7ffdc603564f863ff20139b8)
|
|
smbpass.c: Fixed irix warnings.
Jeremy.
(This used to be commit 2caefb35ff56dba67968d04755d74238a38ba29d)
|
|
header files used by the ubiqx modules, instead of being hidden in the .c
files. This would ensure that anything in includes.h would be "seen" by
the ubiqx headers. I also had to put an #ifdef around the includes for
ubi_SplayTree.h and ubi_Cache.h in includes.h to prevent the header of the
descendant type from being included before its parent type.
Chris -)-----
(This used to be commit e30a7e023b7491c01e052ae3dbd1788f84075413)
|
|
gives a too-small buffer for share info. Fix from Gil Kloepfer <gil@arlut.utexas.edu>
smb.h:
server.c: Fix for a nastly little security problem with multi-user Windows NT
servers and Samba where the contents of the open-file cache can end
up being served out to users who shouldn't have access. This is some
*seriously* ugly code.
Jeremy.
(This used to be commit 05c85df3c7da982085615e5a1db6c71e164db4f5)
|
|
to something less hurried, simpler, and (I believe) more acceptable to
all. The ubi_*.c files all now #include sys_include.h which, for Samba,
contains only comments and the line
#include "../includes.h"
That will make sure that the Samba header is there, allows me to
distribute a different default header with the main ubiqx set, and allows
others to write their own sys_includes.h for their own purposes.
Thanks to Andrew and Jeremy for hammering this out with me.
Chris -)-----
(This used to be commit def161eeea4b430d785ec57150f96fae98a34bcb)
|
|
passdb.c:
lib/rpc/include/rpc_misc.h: First cut at automatic uid/gid to rid mapping.
We can change this at a later date to make more bits available if neccessary.
Jeremy.
(This used to be commit 34f40474aba97118e1e80fe6259c686e46dc16b4)
|
|
belong to -uUsername are printed when running 'smbstatus -uUsername'
PaulB
(This used to be commit da08e486871992043c2650e6aac304cd29ba693c)
|
|
Jeremy.
(This used to be commit 6721c28ea6bdb5210836352bb5ffc43d787907fa)
|
|
Manousaridis Haris <genesis@the.forthnet.gr>
and Magnús Gíslason <magnus@hi.is>
respectively.
Jeremy.
(This used to be commit 583aeaf350ede4ea3e28124b8d66ab0d335656d9)
|
|
(This used to be commit d455d194b7110d4bcbef58134d035bd55702b6f8)
|
|
Jeremy
(This used to be commit 6baeb4ad96bc58cf1bd53f6621067af8344c9556)
|
|
help of Ums Harald <Harald.Ums@pro-sieben.de>, who has been
testing our 8.3 wildcards with a test suite.
With his new code for 8.3 matching, this is the test done
(I'm quoting from his email)
"I tested it by generating a directory with about 7600 Files and run
automatc tests with about 4000 patterns. The result from Win95 -> WinNT
and Win95 -> Samba where identical according to diff."
I have also re-written the long filename wildcard code,
so that doing DIR a*z now matches files :
AAA.BBB.CCCC....ZZZZ
correctly, and other fixes besides.
I sincerely hope I can lay this (horrid) issue to rest
now :-).
Jeremy.
(This used to be commit 94e3f0d9b48c1ac6d9235eb6600aff1c47e024bc)
|
|
password.c: Added code to tell server.c when machine password needs changing.
server.c: Change machine password in idle cycles if it needs it.
smbpassfile.c: Fixed up length calculations for machine password file.
smbpasswd.c: Moved domain joining code/machine password changing code.
lib/rpc/client/cli_netlogon.c: And this is where it now lives.
Jeremy.
(This used to be commit b8fedca6191de96159df0d1d17082d82e8e44773)
|
|
smbpass.c: Do *NOT* make this function static. It breaks the compile on gcc
util.c: Getting closer to MS wildcard semantics. A trailing '*' matches
any trailing dot-separated components.
trans2.c: Removed hacks that change multiple '?' -> '*' as this
breaks things now.
trans2.h: Removed NT_FILE_ATTRIBUTE_NORMAL - now FILE_ATTRIBUTE_NORMAL is defined
in smb.h.
Jeremy.
(This used to be commit 42a65511068cd9006350c80bbed2f346f3f01cb0)
|
|
(This used to be commit dab9728a4376cc52d6e53b09b5f42d08712330d6)
|
|
not found. this gave me a clue...
#if 0 /* commented out until someone uses them jra */
...
#endif
(This used to be commit d553901b9a21c4917058fc6838aaf73eb9a1c8cc)
|
|
(This used to be commit eb157d6da4719b415a1f5202364aa77f2596bc03)
|
|
store passwords in ascii format
- dealt with lots of signed/unsigned char thingies spotted by Tim Winders.
(This used to be commit bd825f1ef15c4bf12aeba945f8bfdc7fd0e14d25)
|
|
Jeremy.
(This used to be commit 26c0176e104f7345f16f7fdb1115f32e6b0cfcdb)
|
|
(This used to be commit c74d0defcf8e9fca326af190730439f6451f826a)
|
|
going on.
(This used to be commit b7ccea8ee190a228cf479f325a2c8578e2bd90ce)
|
|
dir.c & mangle.c. This was possible after I checked the tree code to make
sure that the tree code did not reference functions by macro name. Also,
note that the AVL module has been removed to prevent conflict with the
SplayTree macro defines.
Chris -)-----
(This used to be commit a1f4c8567c3c5264cbfa99867fa894cafab7da81)
|
|
include scheme. Removing them works around the problem. If someone
does need them sometime (can't imagine why, since the splay tree is
typically faster), let me know and we can work something else out.
Chris -)-----
(This used to be commit b3c8b14a47e46e501e1684391e3706e143c2adbf)
|
|
for dir.h, which I've fixed. Andrew did not add includes.h to the ubiqx
headers, which is good because it would cause internal conflicts within the
ubiqx tree modules. It's also bad because the definitions and includes
that are part of includes.h are now in the ubiqx C files, but not in the
header files. So, if includes.h were to redefine int, for example, the
new definition would be in the ubiqx C files, but not in the headers.
So, until Andrew and I can work something out that we both agree upon,
there are three basic rules:
1) Don't include includes.h in the ubiqx headers. Problems may arise.
2) The ubiqx headers must follow includes.h in any Samba module that
uses them.
This can and should all be worked out. We just have to talk about it.
Chris -)-----
(This used to be commit 2b6be7084df4ddfca3a1ab7b73304007c5f9d0aa)
|
|
(This used to be commit f8d6aab70a02338c923220459eb12d17fa324e79)
|
|
test all the heuristics we know about for validating smb.conf
file.
Jeremy.
(This used to be commit 76dec5b04e9c3b4b727bf9a44d31acf32b4b6b4c)
|
|
#ifdef HAVE_INCLUDES_H
#include "../includes.h"
#endif
we want _all_ code in Samba to include this file to ensure that we can
override definitions, typedefs and includes in one place.
(This used to be commit 556745037ad7efffd8276b28fed445b007eba253)
|
|
mode.
Jeremy.
(This used to be commit 7727f09ea9055053ed1d3e2af1069ddae245efb4)
|
|
reply.c: Fixed password length modifiers to always be done
is none-encrypted mode used. This fixes Samba for
people who are using non-encrypted passwords with
security=server.
Jeremy.
(This used to be commit 720b565349e3467bd81d6d863b9ac54237edd3cf)
|
|
This header tries four different locations for a definition of NULL. If
NULL still hasn't been found, it defaults to ((void *)0). All of the
includes can be, essentially, overridden by defining NULL on the command
line, as in -DNULL=((void *)0). I have faith that this will avoid the
problem of NULL being in different places on different systems. If there
is a system out there that doesn't define NULL in any of the headers I've
included, then let me know *where* it's defined and I'll add another
header. Chris -)-----
(This used to be commit 05fcc678729999f249b97460a579bdc218028a47)
|
|
unnecessary "address of" warnings.
(This used to be commit ed0afe1e186ce53046e70d8a92103d4f59a2eed2)
|
|
does gmtime() not LocalTime() now.
(This used to be commit 64559e72f3028917059dbcafd0b3425266f7d567)
|
|
(This used to be commit 4aaf8b6a8bc521e738840fe72708add251eaaccb)
|
|
(This used to be commit 8e11d542eebe076d74ab264e22b87f7aed9bbe8f)
|
|
Jeremy.
(This used to be commit 67007d98b9daf5cccf049cb836dbb989516eb0ce)
|
|
Jeremy.
(This used to be commit 5d8ef7d0f1ecb358c841f382fc48e6eccbe8e171)
|