lpd - line printer daemon

lpd [-Llogfile] [-F] [-V] [-Ddebugopt] [-p port]

The lpd program is the printer server program of the LPRng software suite. This software is an enhanced and modified version of the Berkeley LPD software.

-L logfile specifies an alternate file to be used for logging error and debugging messages. The syslog(8) facility is used to log critical messages as well.

-F Under normal operation, the LPD server will run in background mode. The -F flag forces it to run in forground mode, where it is more easily debugged.

-V Print program version information.

-D debugopt Debugging is controlled using the -D option. This accepts a comma-separated list of debugging settings. These settings take one of two forms: facility=value , or value to set an overall default value. The available facilities can be determined by invoking LPD with the -D= parameter.

-p port Bind to the specified port rather than port 515 specfied by RFC1179.

Lpd is the line printer daemon (spool queue handler) and is normally invoked at boot time from the rc(8) file; it can also be started by a user. Note that the lpd server needs only run on systems where actual printing or spooling is taking place. lpr(1) and other related programs transfer files using network facilities to the lpd .

When started, lpd reads a configuration file to obtain basic operational parameters and then reads the printcap(5) database information to determine the which printers have spool queues and to start spool queue server processes. If running as a background server, it will disconnect from its control terminal and run in the background. It uses the system calls listen(2) and accept(2) to receive requests to print files in the queue, transfer files to the spooling area, display the queue, remove jobs from the queue, or perform a spool queue control function. In each case it creates one or more server processes to handle the request and the lpd process will listen for more requests.

Sending the server a SIGHUP signal causes the server to reread the various configuration and inititialization files. This action is similar to that of the INETD and other servers. The same action is taken when sent a reread command by the lpc(1) program. At an interveral specified by the poll_time configuration variable, lpd will check for spool queues with jobs and no printing activity, and start printing.

LPD access control is done using a rule set and match algorithm similar to a packet filter. Each request for printing, status, or control operations is matched against the rule set, and the first ACCEPT or REJECT value determines if the operation can be performed. The following is a typical permissions file:

# Set default permissions
DEFAULT ACCEPT
# Reject any connections from outside our subnet
REJECT SERVICE=X NOT IP=130.191.0.0/255.255.0.0
# Only accept Printing (P) and spooling (LPR) from
# the private network, the 10.0.0.0/8  network and fw
REJECT SERVICE=P,R NOT REMOTEHOST=*.private,10.0.0.0/8,fw.astart.com
# Do not accept forwarded jobs for printing
REJECT SERVICE=P FORWARD
# Allow only the administrators control access
ACCEPT SERVICE=C,M REMOTEHOST=spooler.astart.com USER=root,papowell
ACCEPT SERVICE=C,M SERVER REMOTEUSER=root,papowell
# Allow only the user on the same host who spooled job to remove it
ACCEPT SERVICE=M SAMEUSER SAMEHOST
REJECT SERVICE=M,C

Permission checking is done by using a set of keys (or fields) with associated values to check for permission. The SERVICE key has value P for printing (i.e.- unspooling), R for spooling (i.e.- LPR request), C and S for printer control and status respectively (i.e.- LPC request), M for removal (i.e.- LPRM request), Q for queue information (i.e.- LPRM request), and so forth. The key indicates the initial connection to the LPD spooler, and can be used to control connections from remote systems. The values of the USER, HOST, and IP keys taken from the control file which is being received or checked for permissions. The REMOTEUSER, REMOTEHOST and REMOTEIP keys are those either sent as part of a command, or derived from information about the current network connection. Each line of the permissions file is scanned for key names and values, and these are matched against the request keys information. When all matches on a line are made, then search terminates with the specified action (ACCEPT/REJECT). If no match is found the default permission value is used. The DEFAULT key is used to specify the current default permission to be used for successful matches or if there is no match after scanning the entire permissions database.

The GROUP entry is used to check that the USER name appears in a group entry in the system user group database. For example, GROUP=student*,staff* would check to see if any of the group name matching student* or staff* have the specified user name in them. If a system has the netgroups capability, a group name starting with a CW@ will be treated as a netgroup name, and current user name from the job file will bechecked to see if it is in the group. Similarly, the REMOTEGROUP entry will check a remote user name. The PORT entry can be used to ensure that a connection to the server orignates from a specified range of ports. For more details, see the lpd.perm(5) man page.

The permissions database is scanned in order of the fixed file entries and then by invoking the specified filters for each of the permissions lists. It is recommended that the filters be placed at the end of the permissions lists. The user name is one of the parameters passed to the filter, and can be used to determine if a user has permissions to print a file.

Key	Match	Connect	Job	Job	LPQ	LPRM	LPC
0	0	0	Spool	Print
SERVICE	S	'X'	'R'	'P'	'Q'	'M'	'C,S'
USER	S	-	JUSR	JUSR	JUSR	JUSR	JUSR
HOST	S	RH	JH	JH	JH	JH	JH
GROUP	S	-	JUSR	JUSR	JUSR	JUSR	JUSR
IP	IP	RIP	JIP	JIP	RIP	JIP	JIP
PORT	N	PORT	PORT	-	PORT	PORT	PORT
REMOTEUSER	S	-	JUSR	JUSR	JUSR	CUSR	CUSR
REMOTEHOST	S	RH	RH	JH	RH	RH	RH
REMOTEGROUP	S	-	JUSR	JUSR	JUSR	CUSR	CUSR
REMOTEIP	IP	RIP	RIP	JIP	RIP	RIP	RIP
CONTROLLINE	S	-	CL	CL	CL	CL	CL
PRINTER	S	-	PR	PR	PR	PR	PR
FORWARD	V	-	SA	-	-	SA	SA	SA
SAMEHOST	V	-	SA	-	SA	SA	SA
SAMEUSER	V	-	-	-	SU	SU	SU
SERVER	V	-	SV	-	SV	SV	SV
AUTH	V	-	AU	-	AU	AU	AU
AUTHTYPE	S	-	AU	-	AU	AU	AU
AUTHUSER	S	-	AU	-	AU	AU	AU
FWDUSER	S	-	AU	-	AU	AU	AU

KEY: JH = HOST host in control file RH = REMOTEHOST connecting host name JUSR = USER user in control file CUSR = REMOTEUSER user from control request JIP= IP IP address of host in control file RIP= REMOTEIP IP address of requesting host PORT= connecting host origination port CONTROLLINE= pattern match of control line in control file FW= IP of source of request = IP of host in control file SA= IP of source of request = IP of host in control file SU= user from request = user in control file SA= IP of source of request = IP of server host SV= matches if remote host is the server AU= authentication information IFIP= IP address of remote end of connection Match: S = string with wild card, IP = IPaddress[/netmask], N = low[-high] number range, V = exact value match SERVICE: 'X' - Connection request; 'R' - lpr request from remote host; 'P' - print job in queue; 'Q' - lpq request, 'M' - lprm request; 'C' - lpc spool control request; 'S' - lpc spool status request NOTE: when printing (P action), the remote and job check values (i.e. - RUSR, JUSR) are identical.

The special key letter=patterns searches the control file line starting with the (upper case) letter, and is usually used with printing and spooling checks. For example, C=A*,B* would check that the class information (i.e.- line in the control file starting with C) had a value starting with A or B.

There is a subtle problem with names and IP addresses which are obtained for 'multi-homed hosts', i.e. - those with multiple ethernet interfaces, and for IPV6 (IP Version 6), in which a host can have multiple addresses, and for the normal host which can have both a short name and a fully qualified domain name. In addition, a host can have multiple IP addresses, depending on the complexity of its configuration.

The IFIP (interface IP) field can be used to check the IP address of the origination of the request, as reported by the information returned by the accept() system call. Note that this information may be IPV4 or IPV6 information, depending on the origination of the system. This information is used by gethostbyaddr() to obtain the orginating host fully qualified domain name (FQDN) and set of IP addresses. Note that this FQDN will be for the originating interface, and may not be the cannonical host name. Some systems which use the Domain Name Server (DNS) system may add the cannonical system name as an alias.

When performing an IP address match, the entire list of IP addresses for a system will now be checked. If one of these matches, then success is reported. Similarly, the entire list of host names and aliases will be checked. If one of these matches, then success will be reported.

In addition, when checking for printing, if the name lookup for the host reported in the control file fails, then we assume that the host is unknown and all match checks for names or IP addresses will fail. You can determine if a host has an entry by using the following check, which will reject all requests from a remotehost which does not have a DNS entry.

REJECT NOT REMOTEHOST=*

Individual printer operations are controlled by values in the printcap database. See printcap(5) for details of the format and content of the various entries. The following are typical printer entries for a local and remote printer.

# main or shared printcap file - usually /etc/printcap
# remote postscript printer
fullpage
   |postscript
   :lp=postscript@farside.astart.com
# give access to (remote) hosts
t1|postscript2
   :cm=Test Printer 1
   :lp=postscript2@nearside.astart.com

# local printcap file # specification for local printer on nearside t1|postscript2 :oh=nearside.astart.com :cd=/usr/spool/LPD/safe :sd=/usr/spool/LPD/t1 # # /usr/spool/LPD/t1/printcap file - t1: :lp=/dev/pr :if=/usr/lib/pr/if :of=/usr/lib/pr/if

Printcap information can be distributed by individual files or shared using NSF, YP, or other methods; see lpd.conf(5) for the exact details of the location of printcap files and programs, given by the printcap_path and lpd_printcap_path configuration information. The usual printcap configuration is to have a main (shared) printcap database which is used by all hosts. The printcap information is usually extremely simple, consisting only of the printer name and host (i.e. - fullpage printer entry).

On hosts which have printers attached or which are to provide spooling queue directories, more extensive printcap information is needed. In the shared database, oh (options for specified host only) field restricts use of this entry to the specified host. This entry can contain host specific information, such as the location of the spool queue and/or actual device to be used for output.

In the above example, the main printcap file, //etc/printcap has entries for all printers. Note that these entries do not specify the spool directories (sd and cd fields), but this could be provided. On a host with a printer specific information can be provided in serveral ways. The simplest is to simply put an additional entry in the shared printcap file, with the oh field set to the support host name. An alternative would be to specify the spool directories (sd and cd fields) in the shared information, and to put the printer specific information in a printcap file.

In addition to the oh flag, the server flag indicates that this entry is for a the LPD server only. This can be used to simplify the management of client and server entries.

The printcap information is obtained in the following order. If the lpd_printcap_path configuration value is nonblank then the lpd server will proccess only this information otherwise it uses the printcap_path information. All client programs use the contents of the configuration printcap_path variable to get a list of locations of printcap files. Each of these entries in the path lists are processed, and the printcap information is extracted. Entries which have oh fields are only used by the specified host. The files and information is processed in linear order, later entries overriding preceeding ones.

When processing jobs or performing spool queue specific requests, the LPD server will check to see if there is a printcap file in the control directory for the spool queue and the contents will be processed. Since only the LPD server has access to the spool and control queues, this information is processed only by the server.

In addition to files, printcap information can be obtained from programs or filters. For example, the printcap_path of the form will use the contents of the file, and then use the program to get information about a specific printer. When information about a particular spool queue is needed and one or more filters are specified as the source of printcap information, then the filter will be started and the printer name written on its standard input. The filter must provide a printcap entry for the requested printer on its standard output.

The filter can be used to interface to databases or nonstandard information sources which do not produce printcap information in an acceptible form.

Each spool queue has a spool directory (sd) and optional control directory (cd) where job and control information is kept. Under normal operation the spool and control directories are identical, but if the spool directory is NFS exported for use by other printer spoolers which write files directly into the spool queue, then it is recommended that the control directory be a separate directory and not NFS mounted. The following files are used for printer operations. Per job entries are marked with an asterisk (*).

File Name	Dir	Purpose
printer	CD	lock file and server process PID
unspooler.printer	CD	subserver process PID
control.printer	CD	queue control information
*hfAnnn	SD	job hold file
*cfAnnnHOST	SD	job control file
*dfAnnnHOST	SD	job data file
*bfAnnn.*	SD	tempory files

The nnn in the file names stands for the job number. RFC1179 requires this to be a 3 digit number, but the longnumber printcap flag or a nonzero longnumber configuration variable will enable 6 digit numbers.

The lock file is used to prevent multiple job queue servers from becoming active simultaneously, and to store the server process id. The lock file name is the name as the printer name; all other control files have the printer name appended as indicated above.

The printer spool control file contains information controlling the queue operations. It consists of a series of lines with keywords and values to control printing, spooling, and automatic job holding operations. The following is an example of a typical spool control file.

spooling_disabled 0
printing_disabled 1
holdall 0
redirect p1@host2
debug 10,log=/tmp/log
class A

The spooling_disabled and printing_disabled entries control spooling and printing; the lpc enable, disable, start, and stop command alter these values. The holdall entry will prevent jobs from being processed until released with the lpc hold or release comands; the lpc holdall and noholdall commands alter these values.

The redirect entry causes the lpd server to forward jobs to the specified remote printer; the lpc redirect command alters this field. The class field controls the class of jobs being printed. By default, the class value is a pattern that matches the class entry in a job file; however a entry of the form letter=patterns will print jobs whose control file line starting with letter matches one of the patterns. The debug line provides a set of debugging parameters for diagnostic information for the particular spool queue.

Each print job consists of a control file and one or more data files. Lines in the control file file specify the job data files or parameters for the job and the general format of the file is specified by RFC1179. Each line consists of a flag character and a parameter; upper case and digit characters specify options and lower case letters specify the printing format and names of data files. The following is a list of the control file flag characters.

A

Identifier A job identifier to be used when displaying job information and/or status. The insertion of this line is controlled by the use_identifier printcap/configuration variable.

C

Class String to be used for the class line on the burst page.

H

Host Name. Name of the machine where lpr was invoked.

I

Indent. The number of characters to indent the output by (in ascii).

J

Job Name. String to be used for the job name on the burst page.

L

Banner user name. Information for banner page.

P

Person. Login name of the person who invoked lpr. This is used to verify ownership by lprm.

M

Send mail to the specified user when the current print job completes.

N

File name. The original name of a data file which is in the job.

T

Title. String to be used as the title for pr(1) when the LPR -p option was specified.

U

Unlink. Job file to remove when printing completed.

W

Width. The page width (in characters) to used for printing.

Z

zoptions. Options passed by lpr -Zzoptions. These are passed to output filters to aid in printing.

f

Formatted File. Name of a file to print which is already formatted.

l

Like ``f'' but passes control characters and does not make page breaks.

p

Name of a file to print using pr(1) as a filter.

t

Troff File. The file contains troff(1) output (cat phototypesetter commands).

d

DVI File. The file contains Tex(l) output (DVI format from Stanford).

g

Graph File. The file contains data produced by plot(3X).

c

Cifplot File. The file contains data produced by cifplot.

v

The file contains a raster image.

r

The file contains text data with FORTRAN carriage control characters.

1

Troff Font R. Name of the font file to use instead of the default. (Obsolete)

2

Troff Font I. Name of the font file to use instead of the default. (Obsolete)

3

Troff Font B. Name of the font file to use instead of the default. (Obsolete)

4

Troff Font S. Name of the font file to use instead of the default. (Obsolete)

Each job in the spool queue can have an associated job hold file which is used by the server process to control the printing of the job. The status file contains information controlling the job hold status and error status. The spool server will attempt to print a job a limited number of times before abandoning it or setting an error status in the job status file. The following is a typical job hold file. hold 0 priority 0 active 2135 redirect remove 0 error

A nonzero hold entry will prevent the job from being processed; the lpc hold and release commands update this field. The priority field overrides the normal first-in first-out printing priority; jobs with non-zero priority fields are printed first. The lpc topq command updates this field. If the active field is non-zero, the job is being printed by the server with the specified process id. The redirect field allows individual jobs to be forwarded to a different printer; the lpc move command updates this field. Finally, the remove and error fields are used to control printing of problem jobs. The remove field is set when a job should be removed; the error field records information that would prevent a job from being printed.

The LPR program is used to submit a job to the LPRng system. The LPR program opens a connection to the LPD server and then transfer the job control file and data files. The LPD server checks to see if the remote host and user has permissions to spool to the requested printer, and then checks to see if the printer is accepting jobs. If both conditions are met, the job is accepted and the control and data files are placed int the spool directory. The LPRng software sends the control file first, followed by the data files.

If the LPR program is acting as a filter, it is not necessary to temporarily store the print job on the local machine. The input data can be sent directly to the LPD server for spooling using an implicit job size of 0 and sending data until the connection is terminated to the server. However, some LPD servers do not accept 0 size jobs, even though it is specified by the RFC1179, so by default LPR will create a temporary file. The LPR -k (seKure) option specifies this direct transmission mode be used.

When LPR is to send a job to the server, it must determine the location of the server. It does this by examining the values of the specified printer and host.

If the printer and host are explicitly specified in the form then the LPR program will send the job to the specified spool queue and to the server running on This can be explicitly specified by the PRINTER environment variable or by the LPR -P option.

If the printer is specified only by a name, then the information in the printcap database is used. The printcap entry for the printer is searched for and the remote host and printer information extracted. The job is sent to the server running on the specified host.

This action can be modified by the following printcap or configuration tags.

1. default_host=host

(Configuration) If there is no printcap entry for the printer, the job is sent to the LPD server running on

2. force_localhost

(Configuration or printcap) If this flag is specified, then LPR and other client programs will send the job to the server running on the localhost. This overrides the default_host information.

The LPD system can forward jobs from one spool directory to another. This is controlled by the following options.

1.

The forward field in the spool control file has a value rp@rm. This can be set using the LPC forward command.

2.

The lp (line printer) printcap entry has the form rp@rm. There is a rm (remote machine) and optional rp (remote printer) printcap entry.

The first of the above conditions to be met will determine the destination. If printing is enabled, then jobs will be forwarded to the remote destination. Example:

# using lp=rp@host
test:sd=/usr/spool/test
  :lp=test@host
test:sd=/usr/spool/test
  :lp=test@host%port
# using :rp:rm:
test:sd=/usr/spool/test
  :rp=test:rm=host

3.

The LPD server uses the same algorithm for sending jobs as the LPR program. A connection is made to the remote server and the files are copied to the server. A set of timeouts is used to control error recover and retry operations. The printcap and configuration variables connect_timeout, connect_interval, connect_grace, and send_try control connecting to the remote host. A connection is attempted to the remote server from a random port in the range of ports specified by the originate_port variable. If a connection is not completed within connect_timeout seconds, the connection is aborted, and then after the connect_interval seconds it is retried. The procedure repeated indefinately for printing, but only once for status or control operations. A connect_timeout value of 0 indicates no timeout; a value of 0 specifies infinite timeout After a job has been successfully printed, the connection is closed and the server waits for connect_grace seconds before trying to reconnect.

Normally job files are forwarded to a printer without modification. The flag makes the queue a bounce queue and allows banners to be generated and data files to passed through the appropriate format filter. The entire output of this process is then passed to the destination with the format specified by the bq_format option (default l or binary). See PRINTING OPERATIONS for details about filters. For example, the following printcap entry will filter format f files.

testbq:sd=/usr/spool/testbq:
  :lpd_bounce
  :bq_format=l
  :lp=final@host
  :if=/usr/lib/filter_for_f
  :mf=/usr/lib/filter_for_m
  :pf=/usr/lib/filter_for_pr

Sometimes only the indicated format of the data files needs to be changed. This can be done using the translate_format option. This entry consists of pairs of lower case characters of the form SdSd...; S is the original and d is the translated format.

changeformat:
  :sd=/usr/spool/changeformat:
  :translate_format=mfpf
  :lp=final@host

In the example above, the m format is processed by a filter, and then its format type is changed to f; the p format is processed similarly. Note that the lpr -p option specifies that the job will be processed by the command - the filter must do both the pr processing and any necessary format conversions.

The printcap flag will cause LPR to do bounce queue filtering before sending the job to the remote queue. This can have unexpected effects if the filters are not available on the local host.

A typical entry which will cause LPR to do filtering is shown below.

testbq:lpr_bounce
  :lp=printer@host
  :if=/usr/lib/filter_for_f
  :vf=/usr/lib/filter_for_v
  :mf=/usr/lib/filter_for_m
  :translate_format=mfvf

This entry will force LPR to run jobs with formats f, m, and v
through the appropriate filter.
It will also rename the formats to the f format.

When a job is submitted for printing, sometimes it is desirable to have it dynamically rerouted to another spool queue, or multiple copies send to various destination. This can be done by using a

When a job is accepted by the LPD server, part of the processing includes passing it to a program specified by the printcap entry. This filter is invoked with the original control file as STDIN, and the default set of filter options. The output of the routing filter will be a set of directives used by LPD when forwarding the job to another printer or in processing the job. The environment and options flags are set as for a standard filter. (See "FILTERS" for details.) Here is a sample printcap entry:

t2|Test Printer 2
    :sd=/var/spool/LPD/t2
    :lf=log
    :lp=t2@printserver
    :bq=t1@localhost
    :destinations=t1@localhost,t2@localhost
    :router=/usr/local/libexec/filters/router

The routing filter exit status is used as follows:
	0  (JSUCC) - normal processing
	37 (JHOLD) - job is held
	any other value - job is deleted from queue


The router filter returns one or more routing entries with the
following format.  Note that entry order is not important, but each
entry will end with the 'end' tag.
dest <destination queue>
copies <number of copies to be made>
X<controlfile modifications>
end

Example of router output:
dest t1@localhost
copies 2
CA
end
dest t2@localhost
CZ
end

The above routing information will have copies of the job sent to
the t1 and t2 spool queue servers.  If no valid routing information
is returned by the router filter the job will be sent to the default
bounce queue destination.

Sometimes it is desirable to reformat a control file before sending to a remote destination. If the printcap entry is present, then the control file is passed through the filter. If the filter exits with status JSUCC, then the job is process normally; status JABORT causes the job processing to be aborted, status JREMOVE causes the job processing to be removed, and any other status is treated as JFAIL.

After passing the control file through the control_filter, the LPD server will reread it, and transfer only the data files specified in the new control file to the destination.

The printcap entry and the configuration entry causes the name of the spool queue to be placed in the job control file. This value can be used by the filter to determine how to process a job. When combined with the use of the Bounce Queue, this can be used to reformat jobs before sending to another printer spooler system.

When printing is enabled, the LPD server will create a spool server process to carry out printing operations. For each job in the queue, the spool server process will create a subserver process to carry out the actual printing operations. If the subserver process fails, the server process will initiate recovery operations. Job will be attempted to be printed until all are done or a subserver returns an ABORT indication; the server will then terminate operations.

The server process normally scans the queue once, at initiation; if the spool control file is modified, usually by using the lpc command, the spool queue is rescanned. The overall algorithm for job printing is:

open the print device;
send some initialization strings;
send a banner to the device;
send the job data files to the device;
send some termination strings;
close the print device;

In order to handle the various device requirements, the subserver process in turn uses 'filter' programs specified in the printcap entry to carry out the individual steps.

OF Filter

The 'of' filter is used for initialization, banner printing and the termination strings. It has the peculiar property of suspending itself when sent a special escape string, allowing other filters to be used to print the individual job files.

Data Filters

Each data file in a job has format specified by a lower case character and an associated filter specified in the printcap file. For example, the 'g' format is printed by the 'gf' filter, and so forth. By convention, the 'if' filter is used to print 'f' (ordinary text) and 'l' (binary) format jobs.

lp-pipe Filters

If the printcap device specification has the form |program then the output device is accessed by the specified program. This allows the program to take care of any required initialization or communication requirements.

The following is a concise summary of the actual algorithm used to print files. Note that LP stands for the printer device or filter specified by the 'lp' printcap entry; OF stands for the 'of' printcap filter; IF is the default 'if' filter; BP is the banner printing filter; and ?F stands for the filter for data file. The '??' values stand for entries from the printcap file.

LP = open( 'lp' );	// open device, filter, or network connection
OF = IF = LP;		// set defaults
set up accounting according to 'af' entry;
if( 'of' ) OF = filter( 'of' ) -> LP;	// make OF filter
if 'as' then record start of job accounting information.
if 'achk' then check for accounting limits.
if( leader on open 'ld' ) `ld` -> OF	// send leader
if( FF on open 'fo' ) `fo` -> OF	// send leader

// print a banner // first check to see if required // and then to see if not suppressed by printcap // or by user do_banner = (always banner 'ab' || (!suppress banner 'sb' && job has banner )); if( ! header last 'hl' && do_banner ){ if( banner program 'bp' ){ fork and exec bp to generate banner, but into temp file. cat temp file -> OF; } else { short banner info -> OF; } }

// now we suspend the OF filter, use other filters if( OF != LP ) suspend OF filter;

for each data file df in job do // send FF between files of job if( !first job && ! no FF separator 'sf' ){ if( OF != LP ) wake up OF filter; 'ff' -> OF; if( OF != LP ) suspend OF filter; }

// get filter for job format = jobformat; if( jobformat == 'f' or jobformat = 'l' ){ format = 'f'; } filter = check pc for filter for format; ?F = LP; // default - no filter if( filter ){ ?F = filter( filter ) -> LP; }

data file -> ?F; // note: if :direct_read: flag set, filter input // is directly from the file, otherwise the // file contents are written to the filter input.

if( ?F != LP ) close( ?F ) endfor

// finish printing

if( OF != LP ) wake up OF filter; if( header last 'hl' && do_banner ){ if( ! no FF separator 'sf' ){ 'ff' -> OF; } if( banner program 'bp' ){ fork and exec bp to generate banner, but into temp file. cat temp file -> OF; } else { short banner info -> OF; } }

if( ff on close 'fq' ){ 'ff' -> OF; }

if( trailer on close 'tr' ){ tr -> OF; }

if 'ae' then record end of job accounting information.

if( OF != LP ) close( OF ); close( LP );

When printing or transferring a job to a spool queue fails, it is retried the number of times specified by the rt (or send_try ) printcap variable. A 0 value specifies an infinite number or retries. When the retry count is exceeded, then the send_failure_action printcap variable determines the action to be taken. The variable can be the values succ , fail , abort , remove , ignore , or hold , which will cause the job to be treated as normally completed, retried, aborted, removed, or ignored and retried at a later time respectively. These names correspond to the JSUCC , JFAIL , etc. error codes returned by filters. If the variable has the form |/filter , then the filter is run and passed the number of attempts on the standard input. The filter must exits with a JSUCC, JFAIL, etc., error code and the server will take the appropriate action as listed above.

The print filters normally have their input provided by a process via a pipe. However, if the direct_read printcap flag is set, then the filter input is taken directly from the job file. This is compatible with the vintage BSD method, but loses the ability to track the job progress.

After the job print or transfer attempt, if the job is to be removed and the printcap variable save_on_error is true, the job will not be removed from the spool queue but only flagged with an error. The job can then be retried at a later time. If the job is successfully printed it is usually removed from the spool queue. However, if the printcap variable save_when_done is true the job will merely be marked as completed and not removed from the queue.

As described in the previous section, filters are created to handle output to devices or other filters. The command line to invoke a filter is generated in the following manner.

1.

The printcap entry or configuration value defining the filter command is obtained.

2.

The file to be printed or the banner line/file generated by the banner printer will be written to STDIN (file descriptor 0) of the filter. The output device (or /dev/null if this is not a printing filter) will be be STDOUT (file descriptor 1) and STDERR (file descriptor 2) will be connected to the error logging file. If this is a printing filter, the error log will be determined by the :af: printcap field and FD 3 will be opened and set to the either the file, remote host, or input of the filter program.

3.

Filter specifications starting with ROOT will be run as root (EUID = 0). This can be a serious security loophole and should only be used as a last resort for specific problems.

4.

The options for the filter command line will be replaced by appropriate values. Option specifications have the form $[0| ][-][']X. The default option expansion has the form $X -> -X'value'; the form $0X or $(space)X adds a space after the -X, i.e.- $0X -> -X 'value'; the form $-X suppresses the -X, i.e. - $-X -> 'value'; and the form $'X' suppresses the quotes around the value. Note that the 0,-, and ' can be combined. For example, $-'X -> value. The options will be exanded as follows:

Key	Value
a	Accounting file (printcap 'af' entry)
b	Job size, i.e.- total data file size, in bytes
c	if binary (format 'l') expands to -c
d	Control directory
e	job data file
f	original print file name (control file N field)
h	Control file hostname
i	Control file indent (I) field
j	job number from control file name
k	Control file name
l	printcap Page length (pl) value
m	printcap Cost factor (co) value
n	Control file user logname (P) field
p	Remote Printer name for forwarded jobs
r	Remote Host name for forwarded jobs
s	printer Status file (ps) value
t	current time in simple format
w	printcap Page width (pw) value
x	printcap x dimension (px) value
y	printcap y dimension (py) value
F	data file format character
P	Printer name
S	printcap Comment tag (cm) value
Upper Case	control file line starting with letter
Digit control file line starting with digit

5.

The options specified by the filter_options (for none OF filters) or of_filter_options (for the OF filter) will be appended to the command line and expanded. To suppress adding options, you can use the form '-$ filter', i.e. - of=-$/bin/cat. If the 'bkf' (backwards compatible filter options) printcap flag is set, the of filter is given the options specified by bk_of_filter_options and other filters those by bk_filter_options. The following shows the various combinations possible, and typical values for the options.

Options
filter_options	$C $F $H $J $L $P $Q $R $Z $a $c $d $e $f $h $i \
	$j $k $l $n $s $w $x $y $-a
bk_filter_options $P $w $l $x $y $F $c $L $i $J $C $0n $0h $-a
bk_of_filter_options $w $l $x $y
6.

A printing filter which executes correctly and completely should
exit with a 0 error status.
A nonzero error status will be interpreted as follows:
JFAIL	32	 failed - retry later
JABORT	33	 aborted - do not try again, but keep job
JREMOVE	34	 failed - remove job


The JFAIL will cause the job to be retried at a later time.
A limit can be placed on the number of retries using the
:rt: or :send_try: printcap entry.
A retry value of 0 will cause infinite retries.
The JABORT indicates serious problems and will cause printing operations
on the job to stop until restarted by operator intervention.
The JREMOVE status indicates problems,
and the job should be removed from the spool queue.

The environment variables for filters are highly restricted,
due to the possibility for abuse by users.
The following variables are set:
USER and LOGNAME

user name or daemon name.
LOGDIR

home directory of user or daemon.
PATH

from the
filter_path
configuration variable.
LD_LIBRARY_PATH

from the
filter_ld_path
configuration variable.
SHELL

set to
/bin/sh
IFS

set to blank and tab.
TZ

the TZ environment variable.
SPOOL_DIR

the spool directory for the printer
CONTROL_DIR

the control directory for the printer
PRINTCAP_ENTRY

the printcap entry for the printer
CONTROL

the control file for the print job
pass_env environment variables

Values of environment variables listed in the pass_env configuration variable.

The LPRng software provides several methods of performing accounting. The printcap af (accounting field), as and ae (accounting start and end), and achk (accounting check) provide a basic set of facilities. The af field specifies a file, filter, or TCP network connection to an accounting server. If af has the form |filter or |-$ filter then a filter will be started and all accounting information will be sent to the filter. The first form passes the filter the command line options specified by the filter_options configuration variable and the second suppresses option passing. If af has the form host%port then a TCP connection will be opened to the port on the specified host and accounting information sent there. All other forms will be treated as a pathname relative to the queue spool directory.

If af specifies a file, then the accounting information is appended to an existing file; the accounting file will not be created.

When af specifies a filter or network connection and the achk flag is set, then after writing the initial accounting information (see as printcap field below) the server will wait for a reply of the form ACCEPT from the filter or server. If not received, the job will not be printed.

The as (accounting start) and ae (accounting end) fields can specify a string to be printed or a filter. Options in the string will be expanded as for filters, and the strings printed to either the accounting information destination. If the as field specifies a filter, then the print server will wait for the filter to exit before printing the job. If the exit status is 0 (successful), the job will be printed. A non-zero JREMOVE status will remove the job, while any other status will terminate queue printing operations. After printing the job, the ae filter will be started and the server will wait for it to complete before printing the next job.

The as and ae filters will have STDOUT set to the printing device and or filter, and the STDERR set to the error log file for the print queue, and fild descriptor 3 set to the destination specified by the af field.

As a convenience, all format filters for printing will be started with file descriptor 3 set to the destination (file or filter) specified by the printcap af field. This allows special filters which can query devices for page counts to pass their information directly to an accounting program. The descriptor will READ/WRITE, allowing filters to query the accounting program and/or update the information directly.

In order to provide a centrallized method to track job status and information, the printcap/configurtion variable logger_destination enable the send of status and other information to a remote destination. The logger_destination value has the form

Examples:

IDENTIFIER jobid [PRINTER name] at timestamp \
   STATUS | TRACE | FILTER_STATUS PID nnn
[ status information]

The status information format consists of an identifier line, followed by a specifier of the status type. The logging information entry is terminated by a line with a single period on it. Lines with a starting period have the period duplicated.

Rather than building authentication facilties into LPRng, an interface to authentication programs is defined, and will be used as follows. The printcap and configuration entries auth, auth_client_filter, auth_forward, auth_forward_id, auth_forward_filter, auth_receive_filter, and auth_server_id entries control authentication. The auth value specifies the type of authentication to be used for client to server authentication. Typical values would be pgp, kerberos, etc. If the authentication type is not built-in, the client programs use the auth_client_filter program to perform authentication. When a server gets and an authentication request, it will use the auth_receive_filter program to perform authentication. The auth_server_id is the remote server id used when a client is sending jobs to the server or when the server is originating a request. When a server forwards a request, it uses auth_forward value to determine if authentication is to be done, and the auth_forward_id as the destination server id.

The client will open a connection to the server and sends a command with the following format. The REQ_SECURE field in the command corresponds to the one-byte command type used by the LPR protocol.

Commands:
	\REQ_SECUREprinter C user\n
Print job transfers:
	\REQ_SECUREprinter C user controfilename\n

ACK_SUCCESS	0	success, no error
ACK_STOP_Q	1	failed; no spooling to the remote queue
ACK_RETRY	2	failed; retry later
ACK_FAIL	3	failed; job rejected, no retry

FD	Options	Purpose
0	R/W	socket connection to remote host (R/W)
1	W	pipe or file descriptor
		for information for server
2	W	error log
3	R	pipe or file descriptor
		for responses to client
The command line arguments will have the form:
program -S -Pprinter -nuser -Rserver_user -Ttempfile
The printer and user information will be obtained from the
command line sent to the server.
The authenticator can create additional temporary or working
files with the pathnames tempfile.ext;  these should be deleted
after the authentication process has been completed.
After receiving \ACK_SUCCESS, the client starts an authenticator
process, and provides the following open file descriptors for
it.  The authenticator process will run UID user.
FD	Options	Purpose
0	R/W	socket connection to remote host (R/W)
1	W	pipe or file descriptor
		for responses to client
2	W	error log
The command line arguments will have the form:
program -C -Pprinter -nuser -Rserver_user -Ttempfile

The authenticator can create additional temporary or working
files with the pathnames tempfile.ext;  these will be deleted
after the authentication process has been completed.
The client authenticator will be running as the client user.
After exchanging authentication information,
the client authenticator will transfer the contents of the
temporary file to the server authenticator, using FD 0.  It will
then wait for reply status on FD 0.   If the transfer step fails,
or there is no reply status of the correct format,  the client
authenticator will print any recieved information on FD 1,
error information on FD 2, and then exit with error code JFAIL.
After receiving the files on FD 0,  the server authenticator
will perform the required authentication procedures and leave
the results in tempfile.  The server authentictor will write
the following to FD 1,  for use by the server:
authentication_info\n
If the transfer step or authentication fails,  then the server
will write an error message to FD 2 and exit with error code
JFAIL.
The server will use this authentication information to
determine if the remote user has permission to access the system.
The server authentication process will read input from FD 3
until and end of file,
and then proceed to transfer the input to the client authenticator.
If the data transfer fails,  then the process will exit with
error code JFAIL, otherwise it will exit with error code JSUCC.
The client authenticator will read the status information from
FD 0,  and after performing authentication will write it to FD 1.
If data transfer or authentication fails,  the authenticator
will write an error message to FD 2 and exit with error code
JFAIL, otherwise it will exit with error code JSUCC.

The Server to Server authentication procedure is used by one server to forward jobs or commands to another server. It should be noted that this forwarding operation puts an implicit trust in the security of the client to server to server chain. In the description below, src and dst are the userid of the source and destination servers respectively. The originating host takes the part of the client, and will transfer a job acting like the client. The initial information transfer from the originating (src) server will have the format:

Commands:
	\REQ_SECUREprinter F user\n
Print job transfers:
	\REQ_SECUREprinter F user controfilename\n
After receiving a 0 acknowlegement byte,
the src server will invoke its authenticator with the arguments
below.
The forward_user value will default to the
server_user value if not explicitly provided.
program -C -Pprinter -nserver_user \
	-Rforward_user -Ttempfile
On the destintation server the authenticator is invoked with the arguments:
program -S -Pprinter -nserver_user \
	-Rforward_user -Ttempfile
The authentication is performed to determine that the transfer was
between the two servers,  rather than the user to server.

As a convenience, Kerberos 5 authentication has been built into the LPD clients and servers. If you are not familiar with Kerberos, then you should obtain other documentation and/or assistance before attempting to use this. The following facilities/configuration values are used to support Kerberos.

A Kerberos principal is the name used for authentication purposes by Kerberos. For example, user principals have the form user@REALM; for example, papowell@ASTART.COM. Services and/or servers have the form service/host@REALM; for example, the lpd server on dickory would have the form: lpr/astart2.astart.com@ASTART.COM

User to server authenticaiton process will use the user's principal name, and generate a service name for the server. The name generation is controlled by the following configuration and/or printcap values.

service

The name of the service to be used to identify the service. This is usually "lpr".

kerberos_keytab

The location of the server keytab file. The keytab file corresponds to the user password, and must be considered a security risk. It should be owned by the LPD server user, and readable/writable only by the server.

kerberos_life

The lifetime of the authentication ticket used by the server. This usually defaults to 10 hours.

kerberos_renew

The renewal time of the ticket.

In addition to the default values, an explicit server principal can be specified in the printcap file using the kerberos_server_principal This allows cross domain authentication to be done.

When setting up Kerberos authentication, you will need to establish principals for each server, and to distribute and install the keytab files on each server.

The following permissions tags are available to check on authentication procedures.

AUTH=[NONE,USER,FWD]	- authentication
	AUTH=NONE	- no authentication
	AUTH=USER	- authentication from a client
	AUTH=FWD	- forwarded authentication from a lpd server
AUTHTYPE=globmatch
AUTHUSER=globmatch
FWDUSER=globmatch

REJECT AUTH=NONE
To reject server forwarded authentication as well, we use REJECT AUTH=NONE,FWD.
If a remote server with name
serverhost
has id information FFEDBEEFDEAF,  then the
following will accept only forwarded jobs from this server.
ACCEPT FWDUSER=FFEDBEEFDEAF REMOTEHOST=serverhost
REJECT AUTH=FWD

The files used by LPRng are set by values in the printer configuration file. The following are a commonly used set of default values.

//etc/lprng/lpd.conf	LPRng configuration file
${HOME}/.printcap	user printer description file
//etc/printcap	printer description file
//etc/lprng/lpd.perms	permissions
/var/run/lprng/lpd	lock file for queue control
/var/spool/lpd		spool directories
/var/spool/lpd/QUEUE/control	queue control
/var/spool/lpd/QUEUE/log	trace or debug log file
/var/spool/lpd/QUEUE/acct	accounting file
/var/spool/lpd/QUEUE/status	status file

lpd.conf(5), lpc(8), checkpc(8), lpr(1), lpq(1), lprm(1), printcap(5), lpd.perms(5), pr(1), lprng_certs(1), lprng_index_certs(1).

Most of the diagnostics are self explanatory.
If you are puzzled over the exact cause of failure,
set the debugging level on (-D5) and run again.
The debugging information will 
help you to pinpoint the exact cause of failure.

LPRng is a enhanced printer spooler system with functionality similar to the Berkeley LPR software. The LPRng mailing list is lprng@lprng.com; subscribe by sending mail to lprng-request@lprng.com with the word subscribe in the body. The software is available from ftp://ftp.lprng.com/pub/LPRng.

Patrick Powell <papowell@lprng.com>.