document are to be interpreted as described in RFC 2119 [2].
2. The Label Stack
2.1. Encoding the Label Stack
The label stack is represented as a sequence of "label stack
entries". Each label stack entry is represented by 4 octets. This
is shown in Figure 1.
0 1 2 3
0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ Label
| Label | Exp |S| TTL | Stack
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ Entry
Label: Label Value, 20 bits
Exp: Experimental Use, 3 bits
S: Bottom of Stack, 1 bit
TTL: Time to Live, 8 bits
Figure 1
The label stack entries appear AFTER the data link layer headers, but
BEFORE any network layer headers. The top of the label stack appears
earliest in the packet, and the bottom appears latest. The network
layer packet immediately follows the label stack entry which has the
S bit set.
Each label stack entry is broken down into the following fields:
1. Bottom of Stack (S)
This bit is set to one for the last entry in the label stack
(i.e., for the bottom of the stack), and zero for all other
label stack entries.
2. Time to Live (TTL)
This eight-bit field is used to encode a time-to-live value.
The processing of this field is described in section 2.4.
3. Experimental Use
This three-bit field is reserved for experimental use.
4. Label Value
This 20-bit field carries the actual value of the Label.
When a labeled packet is received, the label value at the top
of the stack is looked up. As a result of a successful lookup
one learns:
a) the next hop to which the packet is to be forwarded;
b) the operation to be performed on the label stack before
forwarding; this operation may be to replace the top label
stack entry with another, or to pop an entry off the label
stack, or to replace the top label stack entry and then to
push one or more additional entries on the label stack.
In addition to learning the next hop and the label stack
operation, one may also learn the outgoing data link
encapsulation, and possibly other information which is needed
in order to properly forward the packet.
There are several reserved label values:
i. A value of 0 represents the "IPv4 Explicit NULL Label".
This label value is only legal at the bottom of the label
stack. It indicates that the label stack must be popped,
and the forwarding of the packet must then be based on the
IPv4 header.
ii. A value of 1 represents the "Router Alert Label". This
label value is legal anywhere in the label stack except at
the bottom. When a received packet contains this label
value at the top of the label stack, it is delivered to a
local software module for processing. The actual
forwarding of the packet is determined by the label
beneath it in the stack. However, if the packet is
forwarded further, the Router Alert Label should be pushed
back onto the label stack before forwarding. The use of
this label is analogous to the use of the "Router Alert
Option" in IP packets [5]. Since this label cannot occur
at the bottom of the stack, it is not associated with a
particular network layer protocol.
iii. A value of 2 represents the "IPv6 Explicit NULL Label".
This label value is only legal at the bottom of the label
stack. It indicates that the label stack must be popped,
and the forwarding of the packet must then be based on the
IPv6 header.
iv. A value of 3 represents the "Implicit NULL Label". This
is a label that an LSR may assign and distribute, but
which never actually appears in the encapsulation. When
an LSR would otherwise replace the label at the top of the
stack with a new label, but the new label is "Implicit
NULL", the LSR will pop the stack instead of doing the
replacement. Although this value may never appear in the
encapsulation, it needs to be specified in the Label
Distribution Protocol, so a value is reserved.
v. Values 4-15 are reserved.
2.2. Determining the Network Layer Protocol
When the last label is popped from a packet's label stack (resulting
in the stack being emptied), further processing of the packet is
based on the packet's network layer header. The LSR which pops the
last label off the stack must therefore be able to identify the
packet's network layer protocol. However, the label stack does not
contain any field which explicitly identifies the network layer
protocol. This means that the identity of the network layer protocol
must be inferable from the value of the label which is popped from
the bottom of the stack, possibly along with the contents of the
network layer header itself.
Therefore, when the first label is pushed onto a network layer
packet, either the label must be one which is used ONLY for packets
of a particular network layer, or the label must be one which is used
ONLY for a specified set of network layer protocols, where packets of
the specified network layers can be distinguished by inspection of
the network layer header. Furthermore, whenever that label is
replaced by another label value during a packet's transit, the new
value must also be one which meets the same criteria. If these
conditions are not met, the LSR which pops the last label off a
packet will not be able to identify the packet's network layer
protocol.
Adherence to these conditions does not necessarily enable
intermediate nodes to identify a packet's network layer protocol.
Under ordinary conditions, this is not necessary, but there are error
conditions under which it is desirable. For instance, if an
intermediate LSR determines that a labeled packet is undeliverable,
it may be desirable for that LSR to generate error messages which are
specific to the packet's network layer. The only means the
intermediate LSR has for identifying the network layer is inspection
of the top label and the network layer header. So if intermediate
nodes are to be able to generate protocol-specific error messages for
labeled packets, all labels in the stack must meet the criteria
specified above for labels which appear at the bottom of the stack.
If a packet cannot be forwarded for some reason (e.g., it exceeds the
data link MTU), and either its network layer protocol cannot be
identified, or there are no specified protocol-dependent rules for
handling the error condition, then the packet MUST be silently
discarded.
2.3. Generating ICMP Messages for Labeled IP Packets
Section 2.4 and section 3 discuss situations in which it is desirable
to generate ICMP messages for labeled IP packets. In order for a
particular LSR to be able to generate an ICMP packet and have that
packet sent to the source of the IP packet, two conditions must hold:
1. it must be possible for that LSR to determine that a particular
labeled packet is an IP packet;
2. it must be possible for that LSR to route to the packet's IP
source address.
Condition 1 is discussed in section 2.2. The following two
subsections discuss condition 2. However, there will be some cases
in which condition 2 does not hold at all, and in these cases it will
not be possible to generate the ICMP message.
2.3.1. Tunneling through a Transit Routing Domain
Suppose one is using MPLS to "tunnel" through a transit routing
domain, where the external routes are not leaked into the domain's
interior routers. For example, the interior routers may be running
OSPF, and may only know how to reach destinations within that OSPF
domain. The domain might contain several Autonomous System Border
Routers (ASBRs), which talk BGP to each other. However, in this
example the routes from BGP are not distributed into OSPF, and the
LSRs which are not ASBRs do not run BGP.
In this example, only an ASBR will know how to route to the source of
some arbitrary packet. If an interior router needs to send an ICMP
message to the source of an IP packet, it will not know how to route
the ICMP message.
One solution is to have one or more of the ASBRs inject "default"
into the IGP. (N.B.: this does NOT require that there be a "default"
carried by BGP.) This would then ensure that any unlabeled packet
which must leave the domain (such as an ICMP packet) gets sent to a
router which has full routing information. The routers with full
routing information will label the packets before sending them back
through the transit domain, so the use of default routing within the
transit domain does not cause any loops.
This solution only works for packets which have globally unique
addresses, and for networks in which all the ASBRs have complete
routing information. The next subsection describes a solution which
works when these conditions do not hold.
2.3.2. Tunneling Private Addresses through a Public Backbone
In some cases where MPLS is used to tunnel through a routing domain,
it may not be possible to route to the source address of a fragmented
packet at all. This would be the case, for example, if the IP
addresses carried in the packet were private (i.e., not globally
unique) addresses, and MPLS were being used to tunnel those packets
through a public backbone. Default routing to an ASBR will not work
in this environment.
In this environment, in order to send an ICMP message to the source
of a packet, one can copy the label stack from the original packet to
the ICMP message, and then label switch the ICMP message. This will
cause the message to proceed in the direction of the original
packet's destination, rather than its source. Unless the message is
label switched all the way to the destination host, it will end up,
unlabeled, in a router which does know how to route to the source of
original packet, at which point the message will be sent in the
proper direction.
This technique can be very useful if the ICMP message is a "Time
Exceeded" message or a "Destination Unreachable because fragmentation
needed and DF set" message.
When copying the label stack from the original packet to the ICMP
message, the label values must be copied exactly, but the TTL values
in the label stack should be set to the TTL value that is placed in
the IP header of the ICMP message. This TTL value should be long
enough to allow the circuitous route that the ICMP message will need
to follow.
Note that if a packet's TTL expiration is due to the presence of a
routing loop, then if this technique is used, the ICMP message may
loop as well. Since an ICMP message is never sent as a result of
receiving an ICMP message, and since many implementations throttle
the rate at which ICMP messages can be generated, this is not
expected to pose a problem.
2.4. Processing the Time to Live Field
2.4.1. Definitions
The "incoming TTL" of a labeled packet is defined to be the value of
the TTL field of the top label stack entry when the packet is
received.
The "outgoing TTL" of a labeled packet is defined to be the larger
of:
a) one less than the incoming TTL,
b) zero.
2.4.2. Protocol-independent rules
If the outgoing TTL of a labeled packet is 0, then the labeled packet
MUST NOT be further forwarded; nor may the label stack be stripped
off and the packet forwarded as an unlabeled packet. The packet's
lifetime in the network is considered to have expired.
Depending on the label value in the label stack entry, the packet MAY
be simply discarded, or it may be passed to the appropriate
"ordinary" network layer for error processing (e.g., for the
generation of an ICMP error message, see section 2.3).
When a labeled packet is forwarded, the TTL field of the label stack
entry at the top of the label stack MUST be set to the outgoing TTL
value.
Note that the outgoing TTL value is a function solely of the incoming
TTL value, and is independent of whether any labels are pushed or
popped before forwarding. There is no significance to the value of
the TTL field in any label stack entry which is not at the top of the
stack.
2.4.3. IP-dependent rules
We define the "IP TTL" field to be the value of the IPv4 TTL field,
or the value of the IPv6 Hop Limit field, whichever is applicable.
When an IP packet is first labeled, the TTL field of the label stack
entry MUST BE set to the value of the IP TTL field. (If the IP TTL
field needs to be decremented, as part of the IP processing, it is
assumed that this has already been done.)
When a label is popped, and the resulting label stack is empty, then
the value of the IP TTL field SHOULD BE replaced with the outgoing
TTL value, as defined above. In IPv4 this also requires modification
of the IP header checksum.
It is recognized that there may be situations where a network
administration prefers to decrement the IPv4 TTL by one as it
traverses an MPLS domain, instead of decrementing the IPv4 TTL by the
number of LSP hops within the domain.
2.4.4. Translating Between Different Encapsulations
Sometimes an LSR may receive a labeled packet over, e.g., a label
switching controlled ATM (LC-ATM) interface [9], and may need to send
it out over a PPP or LAN link. Then the incoming packet will not be
received using the encapsulation specified in this document, but the
outgoing packet will be sent using the encapsulation specified in
this document.
In this case, the value of the "incoming TTL" is determined by the
procedures used for carrying labeled packets on, e.g., LC-ATM
interfaces. TTL processing then proceeds as described above.