Bug#296687: More info on Parallel Port bug
I looked at:
URL: http://www.ibiblio.org/peanut/Kernel-2.6.10/parport.txt
Then in 2.6 I checked up on my parallel port in
/proc/sys/dev/parport/parport0 and looked at the "modes" file per the
above. It shows: PCSPP,TRISTATE,COMPAT,ECP. It doesn't show DMA since
I've disabled it by dma=none as an option to the parport_pc module.
According to the document at ibiblio, this means that all these mode are
being used. Before disabling DMA, it was being used to. Also dmesg
shows that FIFO is apparently being used too.
This is definitely wrong. Since my ECP port is connected by a cable to
a 25-year-old legacy parallel port on my printer, the ECP port must
operate in the old SPP mode and not in any of the other advanced modes.
This may be the cause of the problem. The original mode requires that the
driver do the handshaking, but in the ECP mode the hardware does it and
the cable control lines are used for different purposes (the pinout
changes). So I now suspect that the problem is that the wrong modes are
used on the port.
Now if both ends of the parallel cable were connected to ECP ports, all
would work OK. So Linux needs to detect what kind of port is on the
other end of the cable. Note that my printer is not on at boot-time so
there's no way to probe it then. Perhaps a module option is needed to
tell the driver what the capabilities are of the other parallel port.
David Lawyer
Reply to: