Parallel Port Driver

The hal_parport driver is a real time component so it must be loaded into the real time thread with loadrt. The configuration string describes the parallel ports to be used, and (optionally) their types. If the configuration string does not describe at least one port, it is an error.

Numbers below 16 refer to parallel ports detected by the system. This is the simplest way to configure the hal_parport driver and cooperates with the Linux parport_pc driver if it is loaded. A port of 0 is the first parallel port detected on the system, 1 is the next and so on.

This will use the first parallel port Linux detects:

Instead, the port address may be specified using the hex notation with the 0x prefix.

The config string represents the hexadecimal address of the port, optionally followed by a direction, all repeated for each port. The directions are in, out, or x, and determine the direction of the physical pins 2 to 9 of the D-Sub 25 connector. If the direction is not specified, the data group will by default be configured as outputs. For example:

This example installs the drivers for a port 0x0278, with pins 2 to 9 as outputs (by default, since neither in nor out is specified), a port 0x0378, with pins 2 to 9 as inputs and a 0x20A0 port, with pins 2 to 9 explicitly specified as outputs. Note that you must know the base address of the parallel ports to configure the drivers correctly. For ISA bus ports, this is usually not a problem, since the ports are almost always at a well-known address, such as 0x278 or 0x378 which are typically configured in the BIOS. The addresses of PCI bus cards are usually found with lspci -v in an I/O ports line, or in a kernel message after running sudo modprobe -a parport_pc. There is no default address, so if <config-string> does not contain at least one address, it is an error.

For each parallel port handled by the hal_parport driver, a type can optionally be specified. The type is one of in, out, epp, or x.

If the type is not specified, the default is out.

A type of epp is the same as out, but the hal_parport driver requests that the port switch into EPP mode. The hal_parport driver does not use the EPP bus protocol, but on some systems EPP mode changes the electrical characteristics of the port in a way that may make some marginal hardware work better. The Gecko G540’s charge pump is known to require this on some parallel ports.

See the Note above about mode x.

This will enable two system-detected parallel ports, the first in output mode and the second in input mode:

You must also direct LinuxCNC to run the read and write functions.

One good PCI parport card is made with the Netmos 9815 chipset. It has good +5 V signals, and can come in a single or dual ports.

To find the I/O addresses for PCI cards open a terminal window and use the list pci command:

Look for the entry with "Netmos" in it. Example of a 2-port card:

From experimentation, I’ve found the first port (the on-card port) uses the third address listed (c000), and the second port (the one that attaches with a ribbon cable) uses the first address listed (b800). The following example shows the onboard parallel port and a PCI parallel port using the default out direction.

Please note that your values will differ. The Netmos cards are Plug-N-Play, and might change their settings depending on which slot you put them into, so if you like to get under the hood and re-arrange things, be sure to check these values before you start LinuxCNC.

For each pin, <p> is the port number, and <n> is the physical pin number in the 25 pin D-shell connector.

For each physical output pin, the driver creates a single HAL pin, for example: parport.0.pin-14-out.

For each physical input pin, the driver creates two HAL pins, for example: parport.0.pin-12-in and parport.0.pin-12-in-not.

The -in HAL pin is TRUE if the physical pin is high, and FALSE if the physical pin is low. The -in-not HAL pin is inverted and is FALSE if the physical pin is high.

The -invert parameter determines whether an output pin is active high or active low. If -invert is FALSE, setting the HAL -out pin TRUE drives the physical pin high, and FALSE drives it low. If -invert is TRUE, then setting the HAL -out pin TRUE will drive the physical pin low.

The individual functions are provided for situations where one port needs to be updated in a very fast thread, but other ports can be updated in a slower thread to save CPU time. It is probably not a good idea to use both an -all function and an individual function at the same time.

If loading the module reports

then ensure that the standard kernel module parport_pc is not loaded
[In the LinuxCNC packages for Ubuntu, the file /etc/modprobe.d/emc2 generally prevents parport_pc from being automatically loaded.]
and that no other device in the system has claimed the I/O ports.

If the module loads but does not appear to function, then the port address is incorrect.

To setup DoubleStep on the parallel port you must add the function parport.n.reset after parport.n.write and configure stepspace to 0 and the reset time wanted. So that step can be asserted on every period in HAL and then toggled off by parport after being asserted for time specified by parport.`__n__.reset-time`.

For example:

More information on DoubleStep can be found on the wiki.

In today’s PCs, parallel ports may require a plug and play (PNP) configuration before they can be used. The kernel module probe_parport configures all PNP ports present. It must be loaded before hal_parport. On machines without a PNP port, it can be loaded but will have no effect.