I would like to report an ACPI/PNP Bios problem which prevents
the correct detection of all parallel port parameters.
I've patched the latest BIOS (MOPNV10J.86A.0175) onto the system, but still
neither Linux nor Windows are able to detect the correct parallel port setting and
will only operate in compatibility, unidirectional PCSPP mode.
This is very unfortunate, because some hardware (purposely-built stuff and
some old scanner/streamer too) still rely on this type of interface.
The reason for this screwup is pretty simple to spot. The BIOS does not
generate the necessary ACPI/PNP0401 entry, once the parallel port mode
is set to ECP. Instead, it continues to report a simple PNP0400 device.
This in turn will force the operating system to operate parallel port interface
in "1980 mode".
This is still a problem as of 2010/10/20 and BIOS rev. MOPNV10J.86A.0311.2010.0802.2346.
The BIOS incorrectly reports the parallel port as "PCSPP" irrespective of the selection made in system setup (ECP, EPP, bi-directional,...). Some recent tests done in the CNC / EMC2 machine control community suggest the port is actually being set properly but is not being reported properly. The work-around is not pretty.
These integrated Atom motherboards look very attractive to CNC integrators but for many applications they require fully functional parallel ports. Word is getting around that the D510MO does not work and this may poison the well for future Intel MB usage. There are, after all, alternatives from ASUS, Jetway, etc.
I had a similar problem on my D425KT board. The BIOS would not switch the parallel port to bidirectional. I finally got it to work in bidirectional mode by setting bit 5 in the ECP control register at the base port address + &h402. The software application was written in VB6 under WindowsXP SP3 and used the Inpout32.dll to communicate with the port at address 378.