source: trunk/samba-3.0.25pre1/docs/htmldocs/Samba3-Developers-Guide/unix-smb.html @ 1

Last change on this file since 1 was 1, checked in by Paul Smedley, 14 years ago

Initial code import

File size: 14.4 KB
Line 
1<html><head><meta http-equiv="Content-Type" content="text/html; charset=ISO-8859-1"><title>Chapter 1. NetBIOS in a Unix World</title><link rel="stylesheet" href="samba.css" type="text/css"><meta name="generator" content="DocBook XSL Stylesheets V1.68.1"><link rel="start" href="index.html" title="SAMBA Developers Guide"><link rel="up" href="pt01.html" title="Part I. The protocol"><link rel="prev" href="pt01.html" title="Part I. The protocol"><link rel="next" href="ntdomain.html" title="Chapter 2. NT Domain RPC's"></head><body bgcolor="white" text="black" link="#0000FF" vlink="#840084" alink="#0000FF"><div class="navheader"><table width="100%" summary="Navigation header"><tr><th colspan="3" align="center">Chapter 1. NetBIOS in a Unix World</th></tr><tr><td width="20%" align="left"><a accesskey="p" href="pt01.html">Prev</a> </td><th width="60%" align="center">Part I. The protocol</th><td width="20%" align="right"> <a accesskey="n" href="ntdomain.html">Next</a></td></tr></table><hr></div><div class="chapter" lang="en"><div class="titlepage"><div><div><h2 class="title"><a name="unix-smb"></a>Chapter 1. NetBIOS in a Unix World</h2></div><div><div class="author"><h3 class="author"><span class="firstname">Andrew</span> <span class="surname">Tridgell</span></h3></div></div><div><p class="pubdate">April 1995</p></div></div></div><div class="toc"><p><b>Table of Contents</b></p><dl><dt><span class="sect1"><a href="unix-smb.html#id244973">Introduction</a></span></dt><dt><span class="sect1"><a href="unix-smb.html#id244990">Usernames</a></span></dt><dt><span class="sect1"><a href="unix-smb.html#id244608">File Ownership</a></span></dt><dt><span class="sect1"><a href="unix-smb.html#id244632">Passwords</a></span></dt><dt><span class="sect1"><a href="unix-smb.html#id245752">Locking</a></span></dt><dt><span class="sect1"><a href="unix-smb.html#id245793">Deny Modes</a></span></dt><dt><span class="sect1"><a href="unix-smb.html#id245812">Trapdoor UIDs</a></span></dt><dt><span class="sect1"><a href="unix-smb.html#id245472">Port numbers</a></span></dt><dt><span class="sect1"><a href="unix-smb.html#id245503">Protocol Complexity</a></span></dt></dl></div><div class="sect1" lang="en"><div class="titlepage"><div><div><h2 class="title" style="clear: both"><a name="id244973"></a>Introduction</h2></div></div></div><p>
2This is a short document that describes some of the issues that
3confront a SMB implementation on unix, and how Samba copes with
4them. They may help people who are looking at unix&lt;-&gt;PC
5interoperability.
6</p><p>
7It was written to help out a person who was writing a paper on unix to
8PC connectivity.
9</p></div><div class="sect1" lang="en"><div class="titlepage"><div><div><h2 class="title" style="clear: both"><a name="id244990"></a>Usernames</h2></div></div></div><p>
10The SMB protocol has only a loose username concept. Early SMB
11protocols (such as CORE and COREPLUS) have no username concept at
12all. Even in later protocols clients often attempt operations
13(particularly printer operations) without first validating a username
14on the server.
15</p><p>
16Unix security is based around username/password pairs. A unix box
17should not allow clients to do any substantive operation without some
18sort of validation.
19</p><p>
20The problem mostly manifests itself when the unix server is in "share
21level" security mode. This is the default mode as the alternative
22"user level" security mode usually forces a client to connect to the
23server as the same user for each connected share, which is
24inconvenient in many sites.
25</p><p>
26In "share level" security the client normally gives a username in the
27"session setup" protocol, but does not supply an accompanying
28password. The client then connects to resources using the "tree
29connect" protocol, and supplies a password. The problem is that the
30user on the PC types the username and the password in different
31contexts, unaware that they need to go together to give access to the
32server. The username is normally the one the user typed in when they
33"logged onto" the PC (this assumes Windows for Workgroups). The
34password is the one they chose when connecting to the disk or printer.
35</p><p>
36The user often chooses a totally different username for their login as
37for the drive connection. Often they also want to access different
38drives as different usernames. The unix server needs some way of
39divining the correct username to combine with each password.
40</p><p>
41Samba tries to avoid this problem using several methods. These succeed
42in the vast majority of cases. The methods include username maps, the
43service%user syntax, the saving of session setup usernames for later
44validation and the derivation of the username from the service name
45(either directly or via the user= option).
46</p></div><div class="sect1" lang="en"><div class="titlepage"><div><div><h2 class="title" style="clear: both"><a name="id244608"></a>File Ownership</h2></div></div></div><p>
47The commonly used SMB protocols have no way of saying "you can't do
48that because you don't own the file". They have, in fact, no concept
49of file ownership at all.
50</p><p>
51This brings up all sorts of interesting problems. For example, when
52you copy a file to a unix drive, and the file is world writeable but
53owned by another user the file will transfer correctly but will
54receive the wrong date. This is because the utime() call under unix
55only succeeds for the owner of the file, or root, even if the file is
56world writeable. For security reasons Samba does all file operations
57as the validated user, not root, so the utime() fails. This can stuff
58up shared development diectories as programs like "make" will not get
59file time comparisons right.
60</p><p>
61There are several possible solutions to this problem, including
62username mapping, and forcing a specific username for particular
63shares.
64</p></div><div class="sect1" lang="en"><div class="titlepage"><div><div><h2 class="title" style="clear: both"><a name="id244632"></a>Passwords</h2></div></div></div><p>
65Many SMB clients uppercase passwords before sending them. I have no
66idea why they do this. Interestingly WfWg uppercases the password only
67if the server is running a protocol greater than COREPLUS, so
68obviously it isn't just the data entry routines that are to blame.
69</p><p>
70Unix passwords are case sensitive. So if users use mixed case
71passwords they are in trouble.
72</p><p>
73Samba can try to cope with this by either using the "password level"
74option which causes Samba to try the offered password with up to the
75specified number of case changes, or by using the "password server"
76option which allows Samba to do its validation via another machine
77(typically a WinNT server).
78</p><p>
79Samba supports the password encryption method used by SMB
80clients. Note that the use of password encryption in Microsoft
81networking leads to password hashes that are "plain text equivalent".
82This means that it is *VERY* important to ensure that the Samba
83smbpasswd file containing these password hashes is only readable
84by the root user. See the documentation ENCRYPTION.txt for more
85details.
86</p></div><div class="sect1" lang="en"><div class="titlepage"><div><div><h2 class="title" style="clear: both"><a name="id245752"></a>Locking</h2></div></div></div><p>
87Since samba 2.2, samba supports other types of locking as well. This
88section is outdated.
89</p><p>
90The locking calls available under a DOS/Windows environment are much
91richer than those available in unix. This means a unix server (like
92Samba) choosing to use the standard fcntl() based unix locking calls
93to implement SMB locking has to improvise a bit.
94</p><p>
95One major problem is that dos locks can be in a 32 bit (unsigned)
96range. Unix locking calls are 32 bits, but are signed, giving only a 31
97bit range. Unfortunately OLE2 clients use the top bit to select a
98locking range used for OLE semaphores.
99</p><p>
100To work around this problem Samba compresses the 32 bit range into 31
101bits by appropriate bit shifting. This seems to work but is not
102ideal. In a future version a separate SMB lockd may be added to cope
103with the problem.
104</p><p>
105It also doesn't help that many unix lockd daemons are very buggy and
106crash at the slightest provocation. They normally go mostly unused in
107a unix environment because few unix programs use byte range
108locking. The stress of huge numbers of lock requests from dos/windows
109clients can kill the daemon on some systems.
110</p><p>
111The second major problem is the "opportunistic locking" requested by
112some clients. If a client requests opportunistic locking then it is
113asking the server to notify it if anyone else tries to do something on
114the same file, at which time the client will say if it is willing to
115give up its lock. Unix has no simple way of implementing
116opportunistic locking, and currently Samba has no support for it.
117</p></div><div class="sect1" lang="en"><div class="titlepage"><div><div><h2 class="title" style="clear: both"><a name="id245793"></a>Deny Modes</h2></div></div></div><p>
118When a SMB client opens a file it asks for a particular "deny mode" to
119be placed on the file. These modes (DENY_NONE, DENY_READ, DENY_WRITE,
120DENY_ALL, DENY_FCB and DENY_DOS) specify what actions should be
121allowed by anyone else who tries to use the file at the same time. If
122DENY_READ is placed on the file, for example, then any attempt to open
123the file for reading should fail.
124</p><p>
125Unix has no equivalent notion. To implement this Samba uses either lock
126files based on the files inode and placed in a separate lock
127directory or a shared memory implementation. The lock file method
128is clumsy and consumes processing and file resources,
129the shared memory implementation is vastly prefered and is turned on
130by default for those systems that support it.
131</p></div><div class="sect1" lang="en"><div class="titlepage"><div><div><h2 class="title" style="clear: both"><a name="id245812"></a>Trapdoor UIDs</h2></div></div></div><p>
132A SMB session can run with several uids on the one socket. This
133happens when a user connects to two shares with different
134usernames. To cope with this the unix server needs to switch uids
135within the one process. On some unixes (such as SCO) this is not
136possible. This means that on those unixes the client is restricted to
137a single uid.
138</p><p>
139Note that you can also get the "trapdoor uid" message for other
140reasons. Please see the FAQ for details.
141</p></div><div class="sect1" lang="en"><div class="titlepage"><div><div><h2 class="title" style="clear: both"><a name="id245472"></a>Port numbers</h2></div></div></div><p>
142There is a convention that clients on sockets use high "unprivileged"
143port numbers (&gt;1000) and connect to servers on low "privilegedg" port
144numbers. This is enforced in Unix as non-root users can't open a
145socket for listening on port numbers less than 1000.
146</p><p>
147Most PC based SMB clients (such as WfWg and WinNT) don't follow this
148convention completely. The main culprit is the netbios nameserving on
149udp port 137. Name query requests come from a source port of 137. This
150is a problem when you combine it with the common firewalling technique
151of not allowing incoming packets on low port numbers. This means that
152these clients can't query a netbios nameserver on the other side of a
153low port based firewall.
154</p><p>
155The problem is more severe with netbios node status queries. I've
156found that WfWg, Win95 and WinNT3.5 all respond to netbios node status
157queries on port 137 no matter what the source port was in the
158request. This works between machines that are both using port 137, but
159it means it's not possible for a unix user to do a node status request
160to any of these OSes unless they are running as root. The answer comes
161back, but it goes to port 137 which the unix user can't listen
162on. Interestingly WinNT3.1 got this right - it sends node status
163responses back to the source port in the request.
164</p></div><div class="sect1" lang="en"><div class="titlepage"><div><div><h2 class="title" style="clear: both"><a name="id245503"></a>Protocol Complexity</h2></div></div></div><p>
165There are many "protocol levels" in the SMB protocol. It seems that
166each time new functionality was added to a Microsoft operating system,
167they added the equivalent functions in a new protocol level of the SMB
168protocol to "externalise" the new capabilities.
169</p><p>
170This means the protocol is very "rich", offering many ways of doing
171each file operation. This means SMB servers need to be complex and
172large. It also means it is very difficult to make them bug free. It is
173not just Samba that suffers from this problem, other servers such as
174WinNT don't support every variation of every call and it has almost
175certainly been a headache for MS developers to support the myriad of
176SMB calls that are available.
177</p><p>
178There are about 65 "top level" operations in the SMB protocol (things
179like SMBread and SMBwrite). Some of these include hundreds of
180sub-functions (SMBtrans has at least 120 sub-functions, like
181DosPrintQAdd and NetSessionEnum). All of them take several options
182that can change the way they work. Many take dozens of possible
183"information levels" that change the structures that need to be
184returned. Samba supports all but 2 of the "top level" functions. It
185supports only 8 (so far) of the SMBtrans sub-functions. Even NT
186doesn't support them all.
187</p><p>
188Samba currently supports up to the "NT LM 0.12" protocol, which is the
189one preferred by Win95 and WinNT3.5. Luckily this protocol level has a
190"capabilities" field which specifies which super-duper new-fangled
191options the server suports. This helps to make the implementation of
192this protocol level much easier.
193</p><p>
194There is also a problem with the SMB specications. SMB is a X/Open
195spec, but the X/Open book is far from ideal, and fails to cover many
196important issues, leaving much to the imagination. Microsoft recently
197renamed the SMB protocol CIFS (Common Internet File System) and have
198published new specifications. These are far superior to the old
199X/Open documents but there are still undocumented calls and features.
200This specification is actively being worked on by a CIFS developers
201mailing list hosted by Microsft.
202</p></div></div><div class="navfooter"><hr><table width="100%" summary="Navigation footer"><tr><td width="40%" align="left"><a accesskey="p" href="pt01.html">Prev</a> </td><td width="20%" align="center"><a accesskey="u" href="pt01.html">Up</a></td><td width="40%" align="right"> <a accesskey="n" href="ntdomain.html">Next</a></td></tr><tr><td width="40%" align="left" valign="top">Part I. The protocol </td><td width="20%" align="center"><a accesskey="h" href="index.html">Home</a></td><td width="40%" align="right" valign="top"> Chapter 2. NT Domain RPC's</td></tr></table></div></body></html>
Note: See TracBrowser for help on using the repository browser.