CUPS Browse Protocol

Introduction

This specification describes the CUPS browsing protocol which is used for printer sharing. The protocol is a UDP/IP-based broadcast service that operates on IP service port 631 by default. Each broadcast packet describes a single printer or class being shared.

For simple networks with a single subnet, a CUPS system sharing a printer (the server) will periodically broadcast that printer's availability and status information to the subnet. Every other CUPS system on the subnet (the clients) will receive the broadcast and make that printer available to local users. If a client stops receiving broadcasts from the server, or if the server sends a special "deleted" broadcast message, the client will remove its copy of the printer.

For larger networks with multiple subnets, a relay configuration can be used where one or more client systems poll the server and then broadcast the availability and status information for the server's shared printers to the clients' local subnets.

A key feature of CUPS printer sharing is support for implicit classes, which are automatically-created classes for printers that are shared by multiple servers. These implicit classes provide automatic load balancing and fail-safe printing functionality transparently to the user.

Security Considerations

Like most discovery protocols, CUPS browse packets are not encrypted or signed, so it is possible for malicious systems on a network to advertise or remove printers on that network to cause denial of service or information disclosure. In order to combat this, CUPS logs incoming browse packets and provides access controls to limit browse packet reception to known hosts.

Browse Packet Format

Each broadcast packet is an ASCII text string of up to 1450 bytes ending with a line feed (0x0a). The general format is:

printer-type printer-state printer-uri "printer-location" "printer-info" "printer-make-and-model" name=value name2=value2 ...

Each of the fields contains the value of the corresponding IPP attribute. The trailing "name=value" information is used to convey default job template attribute values (job-sheets-default, media-default, etc.), authentication requirements (auth-info-required), and additional IPP URI options that are requested by the server (ipp-options).

ABNF Definition

The following ABNF definition [RFC4234, RFC3986] defines the format of each browse packet:

PACKET         = TYPE WSP STATE WSP URI WSP LOCATION WSP INFO WSP
                 MAKE-AND-MODEL WSP *[ WSP ATTR-NAME "=" ATTR-VALUE ] LF

TYPE           = 1*HEXDIG

STATE          = "3" / "4" / "5"

URI            = "ipp://" ( 1*NAMECHAR / IP-literal / IPv4address )
                 [ ":" 1*DIGIT ] ( "/printers/" / "/classes/" ) 1*NAMECHAR
NAMECHAR       = %x21.22.24.26-2E.30-7E / %x25 HEXDIG HEXDIG
IP-literal     = See RFC 3986
IPv4address    = See RFC 3986

LOCATION       = QUOTED-STRING
INFO           = QUOTED-STRING
MAKE-AND-MODEL = QUOTED-STRING

ATTR-NAME      = 1*( ALPHA / DIGIT / "-" / "." )
ATTR-VALUE     = QUOTED-STRING / 1*UNQUOTE-CHAR

QUOTED-STRING  = DQUOTE *QUOTED-CHAR DQUOTE
QUOTED-CHAR    = %x20.21.23-5B.5D-7E / UTF8-CHAR / %x5C %x5C / %x5C %x22
UNQUOTE-CHAR   = %x21.23-26.28-5B.5D-7E / UTF8-CHAR
UTF8-CHAR      = %xC0.DF %x80.BF / %xE0.EF %x80.BF %x80.BF /
                 %xF0.F7 %x80.BF %x80.BF %x80.BF