Your company here — click to reach over 10,000 unique daily visitors

tc-flow - Man Page

flow based traffic control filter


Mapping mode:

tc filter ... flow map key KEY [ OPS ] [ Options ]

Hashing mode:

tc filter ... flow hash keys KEY_LIST [ perturb secs ] [ Options ]

OPS := [ OPS ] OP

Options := [ divisor NUM ] [ baseclass ID ] [ match EMATCH_TREE ] [ action ACTION_SPEC ]


OP := { or | and | xor | rshift | addend } NUM

ID := X:Y

KEY := { src | dst | proto | proto-src | proto-dst | iif | priority | mark | nfct | nfct-src | nfct-dst | nfct-proto-src | nfct-proto-dst | rt-classid | sk-uid | sk-gid | vlan-tag | rxhash }


The flow classifier is meant to extend the SFQ hashing capabilities without hard-coding new hash functions. It also allows deterministic mappings of keys to classes.



Apply an action from the generic actions framework on matching packets.

baseclass ID

An offset for the resulting class ID. ID may be root, none or a hexadecimal class ID in the form [X:]Y. X must match qdisc's/class's major handle (if omitted, the correct value is chosen automatically). If the whole baseclass is omitted, Y defaults to 1.

divisor NUM

Number of buckets to use for sorting into. Keys are calculated modulo NUM.

hash keys KEY-LIST

Perform a jhash2 operation over the keys in KEY-LIST, the result (modulo the divisor if given) is taken as class ID, optionally offset by the value of baseclass. It is possible to specify an interval (in seconds) after which jhash2's entropy source is recreated using the perturb parameter.

map key KEY

Packet data identified by KEY is translated into class IDs to push the packet into. The value may be mangled by OPS before using it for the mapping. They are applied in the order listed here:

and NUM

Perform bitwise AND operation with numeric value NUM.

or NUM

Perform bitwise OR operation with numeric value NUM.

xor NUM

Perform bitwise XOR operation with numeric value NUM.

rshift NUM

Shift the value of KEY to the right by NUM bits.

addend NUM

Add NUM to the value of KEY.

For the or, and, xor and rshift operations, NUM is assumed to be an unsigned, 32bit integer value. For the addend operation, NUM may be much more complex: It may be prefixed by a minus ('-') sign to cause subtraction instead of addition and for keys of src, dst, nfct-src and nfct-dst it may be given in IP address notation. See below for an illustrating example.


Match packets using the extended match infrastructure. See tc-ematch(8) for a detailed description of the allowed syntax in EMATCH_TREE.


In mapping mode, a single key is used (after optional permutation) to build a class ID. The resulting ID is deducible in most cases. In hashing more, a number of keys may be specified which are then hashed and the output used as class ID. This ID is not deducible in beforehand, and may even change over time for a given flow if a perturb interval has been given.

The range of class IDs can be limited by the divisor option, which is used for a modulus.

src,  dst

Use source or destination address as key. In case of IPv4 and TIPC, this is the actual address value. For IPv6, the 128bit address is folded into a 32bit value by XOR'ing the four 32bit words. In all other cases, the kernel-internal socket address is used (after folding into 32bits on 64bit systems).


Use the layer four protocol number as key.


Use the layer four source port as key. If not available, the kernel-internal socket address is used instead.


Use the layer four destination port as key. If not available, the associated kernel-internal dst_entry address is used after XOR'ing with the packet's layer three protocol number.


Use the incoming interface index as key.


Use the packet's priority as key. Usually this is the IP header's DSCP/ECN value.


Use the netfilter fwmark as key.


Use the associated conntrack entry address as key.

nfct-src,  nfct-dst,  nfct-proto-src,  nfct-proto-dst

These are conntrack-aware variants of src, dst, proto-src and proto-dst. In case of NAT, these are basically the packet header's values before NAT was applied.


Use the packet's destination routing table entry's realm as key.


For locally generated packets, use the user or group ID the originating socket belongs to as key.


Use the packet's vlan ID as key.


Use the flow hash as key.


Classic SFQ hash:
tc filter add ... flow hash \
	keys src,dst,proto,proto-src,proto-dst divisor 1024
Classic SFQ hash, but using information from conntrack to work properly in combination with NAT:
tc filter add ... flow hash \
	keys nfct-src,nfct-dst,proto,nfct-proto-src,nfct-proto-dst \
	divisor 1024
Map destination IPs of to classids 1-256:
tc filter add ... flow map \
	key dst addend - divisor 256
Alternative to the above:
tc filter add ... flow map \
	key dst and 0xff
The same, but in reverse order:
tc filter add ... flow map \
	key dst and 0xff xor 0xff

See Also

tc(8), tc-ematch(8), tc-sfq(8)

Referenced By

tc(8), tc-flower(8).

20 Oct 2015 iproute2 Linux