1
<?xml version="1.0" encoding="iso-8859-1"?>
2
<!DOCTYPE chapter PUBLIC "-//Samba-Team//DTD DocBook V4.2-Based Variant V1.0//EN" "http://www.samba.org/samba/DTD/samba-doc">
3
<chapter id="unix-smb">
6
<firstname>Andrew</firstname><surname>Tridgell</surname>
8
<pubdate>April 1995</pubdate>
11
<title>NetBIOS in a Unix World</title>
14
<title>Introduction</title>
16
This is a short document that describes some of the issues that
17
confront a SMB implementation on unix, and how Samba copes with
18
them. They may help people who are looking at unix<->PC
23
It was written to help out a person who was writing a paper on unix to
30
<title>Usernames</title>
32
The SMB protocol has only a loose username concept. Early SMB
33
protocols (such as CORE and COREPLUS) have no username concept at
34
all. Even in later protocols clients often attempt operations
35
(particularly printer operations) without first validating a username
40
Unix security is based around username/password pairs. A unix box
41
should not allow clients to do any substantive operation without some
46
The problem mostly manifests itself when the unix server is in "share
47
level" security mode. This is the default mode as the alternative
48
"user level" security mode usually forces a client to connect to the
49
server as the same user for each connected share, which is
50
inconvenient in many sites.
54
In "share level" security the client normally gives a username in the
55
"session setup" protocol, but does not supply an accompanying
56
password. The client then connects to resources using the "tree
57
connect" protocol, and supplies a password. The problem is that the
58
user on the PC types the username and the password in different
59
contexts, unaware that they need to go together to give access to the
60
server. The username is normally the one the user typed in when they
61
"logged onto" the PC (this assumes Windows for Workgroups). The
62
password is the one they chose when connecting to the disk or printer.
66
The user often chooses a totally different username for their login as
67
for the drive connection. Often they also want to access different
68
drives as different usernames. The unix server needs some way of
69
divining the correct username to combine with each password.
73
Samba tries to avoid this problem using several methods. These succeed
74
in the vast majority of cases. The methods include username maps, the
75
service%user syntax, the saving of session setup usernames for later
76
validation and the derivation of the username from the service name
77
(either directly or via the user= option).
83
<title>File Ownership</title>
86
The commonly used SMB protocols have no way of saying "you can't do
87
that because you don't own the file". They have, in fact, no concept
88
of file ownership at all.
92
This brings up all sorts of interesting problems. For example, when
93
you copy a file to a unix drive, and the file is world writeable but
94
owned by another user the file will transfer correctly but will
95
receive the wrong date. This is because the utime() call under unix
96
only succeeds for the owner of the file, or root, even if the file is
97
world writeable. For security reasons Samba does all file operations
98
as the validated user, not root, so the utime() fails. This can stuff
99
up shared development diectories as programs like "make" will not get
100
file time comparisons right.
104
There are several possible solutions to this problem, including
105
username mapping, and forcing a specific username for particular
112
<title>Passwords</title>
115
Many SMB clients uppercase passwords before sending them. I have no
116
idea why they do this. Interestingly WfWg uppercases the password only
117
if the server is running a protocol greater than COREPLUS, so
118
obviously it isn't just the data entry routines that are to blame.
122
Unix passwords are case sensitive. So if users use mixed case
123
passwords they are in trouble.
127
Samba can try to cope with this by either using the "password level"
128
option which causes Samba to try the offered password with up to the
129
specified number of case changes, or by using the "password server"
130
option which allows Samba to do its validation via another machine
131
(typically a WinNT server).
135
Samba supports the password encryption method used by SMB
136
clients. Note that the use of password encryption in Microsoft
137
networking leads to password hashes that are "plain text equivalent".
138
This means that it is *VERY* important to ensure that the Samba
139
smbpasswd file containing these password hashes is only readable
140
by the root user. See the documentation ENCRYPTION.txt for more
147
<title>Locking</title>
149
Since samba 2.2, samba supports other types of locking as well. This
154
The locking calls available under a DOS/Windows environment are much
155
richer than those available in unix. This means a unix server (like
156
Samba) choosing to use the standard fcntl() based unix locking calls
157
to implement SMB locking has to improvise a bit.
161
One major problem is that dos locks can be in a 32 bit (unsigned)
162
range. Unix locking calls are 32 bits, but are signed, giving only a 31
163
bit range. Unfortunately OLE2 clients use the top bit to select a
164
locking range used for OLE semaphores.
168
To work around this problem Samba compresses the 32 bit range into 31
169
bits by appropriate bit shifting. This seems to work but is not
170
ideal. In a future version a separate SMB lockd may be added to cope
175
It also doesn't help that many unix lockd daemons are very buggy and
176
crash at the slightest provocation. They normally go mostly unused in
177
a unix environment because few unix programs use byte range
178
locking. The stress of huge numbers of lock requests from dos/windows
179
clients can kill the daemon on some systems.
183
The second major problem is the "opportunistic locking" requested by
184
some clients. If a client requests opportunistic locking then it is
185
asking the server to notify it if anyone else tries to do something on
186
the same file, at which time the client will say if it is willing to
187
give up its lock. Unix has no simple way of implementing
188
opportunistic locking, and currently Samba has no support for it.
194
<title>Deny Modes</title>
197
When a SMB client opens a file it asks for a particular "deny mode" to
198
be placed on the file. These modes (DENY_NONE, DENY_READ, DENY_WRITE,
199
DENY_ALL, DENY_FCB and DENY_DOS) specify what actions should be
200
allowed by anyone else who tries to use the file at the same time. If
201
DENY_READ is placed on the file, for example, then any attempt to open
202
the file for reading should fail.
206
Unix has no equivalent notion. To implement this Samba uses either lock
207
files based on the files inode and placed in a separate lock
208
directory or a shared memory implementation. The lock file method
209
is clumsy and consumes processing and file resources,
210
the shared memory implementation is vastly prefered and is turned on
211
by default for those systems that support it.
217
<title>Trapdoor UIDs</title>
219
A SMB session can run with several uids on the one socket. This
220
happens when a user connects to two shares with different
221
usernames. To cope with this the unix server needs to switch uids
222
within the one process. On some unixes (such as SCO) this is not
223
possible. This means that on those unixes the client is restricted to
228
Note that you can also get the "trapdoor uid" message for other
229
reasons. Please see the FAQ for details.
235
<title>Port numbers</title>
237
There is a convention that clients on sockets use high "unprivileged"
238
port numbers (>1000) and connect to servers on low "privilegedg" port
239
numbers. This is enforced in Unix as non-root users can't open a
240
socket for listening on port numbers less than 1000.
244
Most PC based SMB clients (such as WfWg and WinNT) don't follow this
245
convention completely. The main culprit is the netbios nameserving on
246
udp port 137. Name query requests come from a source port of 137. This
247
is a problem when you combine it with the common firewalling technique
248
of not allowing incoming packets on low port numbers. This means that
249
these clients can't query a netbios nameserver on the other side of a
250
low port based firewall.
254
The problem is more severe with netbios node status queries. I've
255
found that WfWg, Win95 and WinNT3.5 all respond to netbios node status
256
queries on port 137 no matter what the source port was in the
257
request. This works between machines that are both using port 137, but
258
it means it's not possible for a unix user to do a node status request
259
to any of these OSes unless they are running as root. The answer comes
260
back, but it goes to port 137 which the unix user can't listen
261
on. Interestingly WinNT3.1 got this right - it sends node status
262
responses back to the source port in the request.
268
<title>Protocol Complexity</title>
270
There are many "protocol levels" in the SMB protocol. It seems that
271
each time new functionality was added to a Microsoft operating system,
272
they added the equivalent functions in a new protocol level of the SMB
273
protocol to "externalise" the new capabilities.
277
This means the protocol is very "rich", offering many ways of doing
278
each file operation. This means SMB servers need to be complex and
279
large. It also means it is very difficult to make them bug free. It is
280
not just Samba that suffers from this problem, other servers such as
281
WinNT don't support every variation of every call and it has almost
282
certainly been a headache for MS developers to support the myriad of
283
SMB calls that are available.
287
There are about 65 "top level" operations in the SMB protocol (things
288
like SMBread and SMBwrite). Some of these include hundreds of
289
sub-functions (SMBtrans has at least 120 sub-functions, like
290
DosPrintQAdd and NetSessionEnum). All of them take several options
291
that can change the way they work. Many take dozens of possible
292
"information levels" that change the structures that need to be
293
returned. Samba supports all but 2 of the "top level" functions. It
294
supports only 8 (so far) of the SMBtrans sub-functions. Even NT
295
doesn't support them all.
299
Samba currently supports up to the "NT LM 0.12" protocol, which is the
300
one preferred by Win95 and WinNT3.5. Luckily this protocol level has a
301
"capabilities" field which specifies which super-duper new-fangled
302
options the server suports. This helps to make the implementation of
303
this protocol level much easier.
307
There is also a problem with the SMB specications. SMB is a X/Open
308
spec, but the X/Open book is far from ideal, and fails to cover many
309
important issues, leaving much to the imagination. Microsoft recently
310
renamed the SMB protocol CIFS (Common Internet File System) and have
311
published new specifications. These are far superior to the old
312
X/Open documents but there are still undocumented calls and features.
313
This specification is actively being worked on by a CIFS developers
314
mailing list hosted by Microsft.