Generic packet editor action in tc(8) Linux Generic packet editor action in tc(8) NAME pedit - generic packet editor action SYNOPSIS tc ... action pedit munge { RAW_OP | LAYERED_OP } [ CONTROL ] RAW_OP := offset OFFSET { u8 | u16 | u32 } [ AT_SPEC ] CMD_SPEC AT_SPEC := at AT offmask MASK shift SHIFT LAYERED_OP := { ip IPHDR_FIELD | ip6 IP6HDR_FIELD | udp UDPHDR_FIELD | tcp TCPHDR_FIELD | icmp ICMPHDR_FIELD } CMD_SPEC IPHDR_FIELD := { src | dst | tos | dsfield | ihl | protocol | precedence | nofrag | firstfrag | ce | df | mf | dport | sport | icmp_type | icmp_code } CMD_SPEC := { clear | invert | set VAL | preserve } [ retain RVAL ] CONTROL := { reclassify | pipe | drop | shot | continue | pass } DESCRIPTION The pedit action can be used to change arbitrary packet data. The location of data to change can either be specified by giving an offset and size as in RAW_OP, or for header values by naming the header and field to edit the size is then chosen automatically based on the header field size. Currently this is supported only for IPv4 headers. OPTIONS offset OFFSET { u32 | u16 | u8 } Specify the offset at which to change data. OFFSET is a signed integer, it's base is automatically chosen (e.g. hex if prefixed by 0x or octal if prefixed by 0). The second argument specifies the length of data to change, that is four bytes (u32), two bytes (u16) or a single byte (u8). at AT offmask MASK shift SHIFT This is an optional part of RAW_OP which allows to have a variable OFFSET depending on packet data at offset AT, which is binary ANDed with MASK and right-shifted by SHIFT before adding it to OFFSET. ip IPHDR_FIELD Change an IPv4 header field. The supported keywords for IPHDR_FIELD are: src dst Source or destination IP address, a four-byte value. tos dsfield precedence Type Of Service field, an eight-bit value. ihl Change the IP Header Length field, a four-bit value. protocol Next-layer Protocol field, an eight-bit value. nofrag firstfrag ce df present immediately following the (minimal sized) IP header. If it is not or the latter is big‐ ger than the minimum of 20 bytes, this will do unexpected things. These fields are eight-bit values. clear Clear the addressed data (i.e., set it to zero). invert Swap every bit in the addressed data. set VAL Set the addressed data to a specific value. The size of VAL is defined by either one of the u32, u16 or u8 keywords in RAW_OP, or the size of the addressed header field in LAYERED_OP. preserve Keep the addressed data as is. retain RVAL This optional extra part of CMD_SPEC allows to exclude bits from being changed. CONTROL The following keywords allow to control how the tree of qdisc, classes, filters and actions is further traversed after this action. reclassify Restart with the first filter in the current list. pipe Continue with the next action attached to the same filter. drop shot Drop the packet. continue Continue classification with the next filter in line. pass Finish classification process and return to calling qdisc for further packet processing. This is the default. EXAMPLES Being able to edit packet data, one could do all kinds of things, such as e.g. implementing port redirection. Certainly not the most useful application, but as an example it should do: First, qdiscs need to be set up to attach filters to. For the receive path, a simple ingress qdisc will do, for transmit path a classful qdisc (HTB in this case) is necessary: tc qdisc replace dev eth0 root handle 1: htb tc qdisc add dev eth0 ingress handle ffff: Finally, a filter with pedit action can be added for each direction. In this case, u32 is used matching on the port number to redirect from, while pedit then does the actual rewriting: tc filter add dev eth0 parent 1: u32 \ match ip dport 23 0xffff \ action pedit pedit munge ip dport set 22 tc filter add dev eth0 parent ffff: u32 \ match ip sport 22 0xffff \