1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
|
<pre>Network Working Group Ross Finlayson
Request for Comments: 906 Stanford University
June 1984
<span class="h1">Bootstrap Loading using TFTP</span>
Status of this Memo
It is often convenient to be able to bootstrap a computer system from
a communications network. This RFC proposes the use of the IP TFTP
protocol for bootstrap loading in this case.
This RFC specifies a proposed protocol for the ARPA Internet
community, and requests discussion and suggestions for improvements.
Introduction
Many computer systems, such as diskless workstations, are
bootstrapped by loading one or more code files across a network.
Unfortunately, the protocol used to load these initial files has not
been standardized - numerous methods have been employed by different
computer manufacturers. This can make it difficult, for example, for
an installation to support several different kinds of systems on a
local-area network. Each different booting mechanism that is used
must be supported, for example by implementing a number of servers on
one or more host machines. This is in spite of the fact that these
heterogeneous systems may be able to communicate freely (all using
the same protocol) once they have been booted.
We propose that TFTP (Trivial File Transfer Protocol) [<a href="#ref-6" title=""The TFTP Protocol (Revision 2)"">6</a>] be used as
a standard protocol for bootstrap loading. This protocol is
well-suited for our purpose, being based on the standard Internet
Protocol (IP) [<a href="#ref-4" title=""Internet Protocol - DARPA Internet Program Protocol Specification"">4</a>]. It is easily implemented, both in the machines to
be booted, and in bootstrap servers elsewhere on the net. (In
addition, many popular operating systems already support TFTP
servers.) The fact that TFTP is a rather slow protocol is not a
serious concern, due to the fact that it need be used only for the
primary bootstrap. A secondary bootstrap could use a faster
protocol.
This RFC describes how system to be booted (called the "booter"
below) would use TFTP to load a desired code file. It also describes
an existing implementation (in ROM) for Ethernet.
Note that we are specifying only the network protocols that would be
used by the booting system. We do not attempt to mandate the method
by which a user actually boots a system (such as the format of a
command typed at the console). In addition, our proposal does not
<span class="grey">Finlayson [Page 1]</span></pre>
<hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-2" ></span>
<span class="grey"><a href="./rfc906">RFC 906</a> June 1984</span>
presuppose the use of any particular data-link level network
architecture (although the example that we describe below uses
Ethernet).
Network Protocols used by the Booting System
To load a file, the booter sends a standard TFTP read request (RRQ)
packet, containing the name of the file to be loaded. The file name
should not assume any operating system dependent naming conventions
(file names containing only alphanumeric characters should suffice).
Thereafter, the system receives TFTP DATA packets, and sends TFTP ACK
and/or ERROR packets, in accordance with the TFTP specification [<a href="#ref-6" title=""The TFTP Protocol (Revision 2)"">6</a>].
TFTP is implemented using the User Datagram Protocol (UDP) [<a href="#ref-5" title=""User Datagram Protocol"">5</a>], which
is in turn implemented using IP. Thus, the booter must be able to
receive IP datagrams containing up to 524 octets (excluding the IP
header), since TFTP DATA packets can be up to 516 octets long, and
UDP headers are 8 octets long. The booting machine is not required
to respond to incoming TFTP read or write requests.
We allow for the use of two additional protocols. These are ARP
(Address Resolution Protocol) [<a href="#ref-3" title=""An Ethernet Address Resolution Protocol"">3</a>], and RARP (Reverse Address
Resolution Protocol) [<a href="#ref-1" title=""A Reverse Address Resolution Protocol"">1</a>]. The possible use of these protocols is
described below. The booter could also use other protocols (such as
for name lookup), but they should be IP-based, and an internet
standard.
The IP datagram containing the initial TFTP RRQ (and all other IP
datagrams sent by the booter) must of course contain both a source
internet address and a destination internet address in its IP header.
It is frequently the case, however, that the booter does not
initially know its own internet address, but only a lower-level (e.g.
Ethernet) address. The Reverse Address Resolution Protocol
(RARP) [<a href="#ref-1" title=""A Reverse Address Resolution Protocol"">1</a>] may be used by the booter to find its internet address
(prior to sending the TFTP RRQ). RARP was motivated by Plummer's
Address Resolution Protocol (ARP) [<a href="#ref-3" title=""An Ethernet Address Resolution Protocol"">3</a>]. Unlike ARP, which is used to
find the 'hardware' address corresponding to a known higher-level
protocol (e.g. internet) address, RARP is used to determine a
higher-level protocol address, given a known hardware address. RARP
uses the same packet format as ARP, and like ARP, can be used for a
wide variety of data-link protocols.
ARP may also be used. If the destination internet address is known,
then an ARP request containing this address may be broadcast, to find
a corresponding hardware address to which to send the subsequent TFTP
RRQ. It may not matter if this request should fail, because the RRQ
can also be broadcast (at the data-link level). However, because
such an ARP request packet also contains the sender's (that is, the
<span class="grey">Finlayson [Page 2]</span></pre>
<hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-3" ></span>
<span class="grey"><a href="./rfc906">RFC 906</a> June 1984</span>
booter's) internet and hardware addresses, this information is made
available to the rest of the local subnet, and could be useful for
routing, for instance.
If a single destination internet address is not known, then a special
'broadcast' internet address could be used as the destination address
in the TFTP RRQ, so that it will be received by all 'local' internet
hosts. (At this time, however, no standard for internet broadcasting
has been officially adopted. [**])
An Example Implementation
The author has implemented TFTP booting as specified above. The
resulting code resides in ROM. (This implementation is for a
Motorola 68000 based workstation, booting over an Ethernet.) A user
wishing to boot such a machine types a file name, and (optionally)
the internet address of the workstation, and/or the internet address
of a server machine from which the file is to be loaded. The
bootstrap code proceeds as follows:
(1) The workstation's Ethernet address is found (by querying the
Ethernet interface).
(2) If the internet address of the workstation was not given, then
a RARP request is broadcast, in order to find it. If this request
fails (that is, times out), then the bootstrap fails.
(3) If the internet address of a server host was given, then
broadcast an ARP request to try to find a corresponding Ethernet
address. If this fails, or if a server internet address was not
given, then the Ethernet broadcast address is used.
(4) If the internet address of a server host was not given, then
we use a special internet address that represents a broadcast on
the "local subnet", as described in [<a href="#ref-2" title=""Internet Broadcasting"">2</a>]. (This is not an internet
standard.)
(5) A TFTP RRQ for the requested file is sent to the Ethernet
address found in step (3). The source internet address is that
found in step (2), and the destination internet address is that
found in step (4).
Note that because several TFTP servers may, in general, reply to the
RRQ, we do not abort if a TFTP ERROR packet is received, because this
does not preclude the possibility of some other server replying later
with the first data packet of the requested file. When the first
valid TFTP DATA packet is received in response to the RRQ, the source
internet and Ethernet addresses of this packet are used as the
<span class="grey">Finlayson [Page 3]</span></pre>
<hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-4" ></span>
<span class="grey"><a href="./rfc906">RFC 906</a> June 1984</span>
destination addresses in subsequent TFTP ACK packets. Should another
server later respond with a DATA packet, an ERROR packet is sent back
in response.
An implementation of TFTP booting can take up a lot of space if care
is not taken. This can be a significant problem if the code is to
fit in a limited amount of ROM. However, the implementation
described above consists of less than 4K bytes of code (not counting
the Ethernet device driver).
Acknowledgements
The ideas presented here are the result of discussions with several
other people, in particular Jeff Mogul.
References
[<a id="ref-1">1</a>] Finlayson, R., Mann, T., Mogul, J. & Theimer, M., "A Reverse
Address Resolution Protocol", <a href="./rfc903">RFC 903</a> Stanford University,
June 1984.
[<a id="ref-2">2</a>] Mogul, J., "Internet Broadcasting", Proposed RFC, January 1984.
[<a id="ref-3">3</a>] Plummer, D., "An Ethernet Address Resolution Protocol",
<a href="./rfc826">RFC 826</a>, MIT-LCS, November 1982.
[<a id="ref-4">4</a>] Postel, J., ed., "Internet Protocol - DARPA Internet Program
Protocol Specification", <a href="./rfc791">RFC 791</a>, USC/Information Sciences
Institute, September 1981.
[<a id="ref-5">5</a>] Postel, J., "User Datagram Protocol", <a href="./rfc768">RFC 768</a> USC/Information
Sciences Institute, August 1980.
[<a id="ref-6">6</a>] Sollins, K., "The TFTP Protocol (Revision 2)", <a href="./rfc783">RFC 783</a>, MIT/LCS,
June 1981.
[**] Editor's Note: While there is no standard for an Internet wide
broadcast or multicast address, it is strongly recommended that
the "all ones" local part of the Internet address be used to
indicate a broadcast in a particular network. That is, in class
A network 1 the broadcast address would be 1.255.255.255, in
class B network 128.1 the broadcast address would be
128.1.255.255, and in class C network 192.1.1 the broadcast
address would be 192.1.1.255.
Finlayson [Page 4]
</pre>
|
