Hi, I have a PC104 from kontron (mops_lcd) that I can't read IO ports from the ISA bus, The write to ports is ok but the read gets 0xff always. (even with datalines forced '0' all the time) Looks that IO input is comming from elsewhere as onboard devices like LPT or COM work well I suspect a bridge config or whatever... Kontron support doesn't really help and I wonder if anyone has stepped into this before... Luis C.
PC104 ISA bus.
Started by ●February 26, 2008
Reply by ●February 26, 20082008-02-26
On Feb 26, 11:26 am, LC <cupidoREM...@mail.ua.pt> wrote:> I have a PC104 from kontron (mops_lcd) that > I can't read IO ports from the ISA bus, > The write to ports is ok but the read gets > 0xff always. (even with datalines forced '0' all the time)There's probably a buffer somewhere that has to be enabled to let the PC104 data lines drive the processor local bus (which may very well be address/data multiplexed) at the appropriate time during a read cycle. While writes can be broadcast to multiple bus branches, reads can only have one such buffer enabled and whatever logic is supposed to do that probably hasn't been told that this address means a read from this bus. The fact that you can short the data lines to ground without crashing the system also points to your PC104 bus being somewhat isolated from the main processor bus. And remember the different control signals for IO vs memory access.
Reply by ●February 27, 20082008-02-27
Thanks, I suspect indeed of a bug on the isa-bridge configuration. inputs are probably coming from the PCI bus. How on a PC we (or them) select which ports are read from each bus ? This one has no such options on the bios config. lc. cs_posting@hotmail.com wrote:> On Feb 26, 11:26 am, LC <cupidoREM...@mail.ua.pt> wrote: > >> I have a PC104 from kontron (mops_lcd) that >> I can't read IO ports from the ISA bus, >> The write to ports is ok but the read gets >> 0xff always. (even with datalines forced '0' all the time) > > There's probably a buffer somewhere that has to be enabled to let the > PC104 data lines drive the processor local bus (which may very well > be address/data multiplexed) at the appropriate time during a read > cycle. While writes can be broadcast to multiple bus branches, reads > can only have one such buffer enabled and whatever logic is supposed > to do that probably hasn't been told that this address means a read > from this bus. > > The fact that you can short the data lines to ground without crashing > the system also points to your PC104 bus being somewhat isolated from > the main processor bus. > > And remember the different control signals for IO vs memory access.
Reply by ●February 28, 20082008-02-28
On Feb 27, 5:39 am, LC <cupidoREM...@mail.ua.pt> wrote:> Thanks, I suspect indeed of a bug on the isa-bridge > configuration. inputs are probably coming from the > PCI bus. > > How on a PC we (or them) select which ports are read from > each bus ? > This one has no such options on the bios config.Try to get the data sheet for the bridge chip, or more likely the motherboard chipset More directly, do you have a known-good pc104 card you can try?
Reply by ●February 28, 20082008-02-28
Will search the data on the isa bridge chip, maybe we can configure it after the boot at the beginning of our code... or maybe not :-( no other PC104, but many trivial PC's w/ isa and they do work beautifully. ;-) Opss I think we have an older P1@100MHz in PC104 format, will look. tks. lc cs_posting@hotmail.com wrote:> On Feb 27, 5:39 am, LC <cupidoREM...@mail.ua.pt> wrote: >> Thanks, I suspect indeed of a bug on the isa-bridge >> configuration. inputs are probably coming from the >> PCI bus. >> >> How on a PC we (or them) select which ports are read from >> each bus ? >> This one has no such options on the bios config. > > Try to get the data sheet for the bridge chip, or more likely the > motherboard chipset > > More directly, do you have a known-good pc104 card you can try?
