XPost: alt.comp.lang.pascal, comp.os.msdos.programmer   
   From: psharris.MAPS@magma.MAPS.ca   
      
   Michal Necasek wrote:   
   > P. Scott Harris wrote:   
   >   
   >> - there is a bit (10) in the command register of the PCI configuration   
   >> space related to enabling interrupts but this applies to the PCI 2.3   
   >> spec and doesn't apply to any of the PCI parallel port boards that I   
   >> have.   
   >  >   
   >  Doesn't apply? Why?   
      
   The bit is not read/write. I can't set it or clear it which I presume to mean   
   that the board doesn't implement the feature (which apparently applies to PCI   
   versions 2.3 and greater).   
      
   >   
   >> - I've reserved an ISA IRQ in the BIOS and then configured the PCI   
   >> parallel port board to use it (so IRQ sharing shouldn't be occurring).   
   >  You're doing it backwards, and your PCI board doesn't sound very   
   > PCI-compliant if what you say is true. You're *not* supposed to pull a   
   > number out of thin air and assign it to a board's IRQ. It is the   
   > system's responsibility to assign PCI resources (including IRQs), and   
   > PCI provides ample mechanisms to discover the configuration.   
   >   
   >  Reserving an ISA IRQ in the BIOS generally means that this IRQ will   
   > *not* be available to PCI boards. So, do you really know that your PCI   
   > board is hooked up to the IRQ you think it is? What does the PCI config   
   > space say in the IRQ register for the adapter?   
      
   I don't pull a value out of thin air. I just try to use the BIOS specified   
   parameters as pulled from the PCI config space registers e.g. base port=$BC00,   
   IRQ=12. IRQ12 can be shared. Using these values, I don't see an interrupt   
   generated. I was trying to force the PCI board to not share the interrupt.   
   Short   
   of reconfiguring all other boards in the system I though that I might be able   
   to   
   reserve an interrupt in the BIOS (ensuring nobody used it) then call the PCI   
   BIOS interrupt (Int 1A, AX=B10F) to set the interrupt to the reserved value.   
   This works in that I can read back the PCI registers and they show the new IRQ   
   setting (e.g. was 12, now shows 5) but it doesn't solve the problem.   
      
   I have 3 different PCI parallel port boards using different chipsets - Lava,   
   Netmos and Sun. Doesn't work with any of them   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)