Please accept my appologies for this, it does look like I messed up the MI code. The following change may mess up the pcib code a little, but I think that the MD code now does the right thing.
Does this fix things for you? Warner Index: pcivar.h =================================================================== RCS file: /cache/ncvs/src/sys/dev/pci/pcivar.h,v retrieving revision 1.56 diff -u -r1.56 pcivar.h --- pcivar.h 1 Jun 2002 05:40:33 -0000 1.56 +++ pcivar.h 7 Jun 2002 02:35:52 -0000 @@ -246,10 +246,12 @@ #undef PCIB_ACCESSOR /* - * PCI interrupt validation. + * PCI interrupt validation. Invalid interrupt values such as 0 or 128 + * on i386 or other platforms should be mapped out in the MD pcireadconf + * code and not here, since the only MI invalid IRQ is 255. */ #define PCI_INVALID_IRQ 255 -#define PCI_INTERRUPT_VALID(x) (((x) != 0) && ((x) != PCI_INVALID_IRQ)) +#define PCI_INTERRUPT_VALID(x) ((x) != PCI_INVALID_IRQ) /* * Convenience functions. To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
