I think there may be something amiss with the snapshot.

I just added a PNP ISA card to my new system and it crashed in ym_attach
calling a routine at address 0.  The stack arguments (and the fact that
it was halfway through a printf) argues that it was calling

        sc->sc_ad1848.sc_ih = isa_intr_establish(sc->sc_ic, sc->ym_irq,
                                                 IST_EDGE,
                                                 IPL_AUDIO, ad1848_isa_intr,
                                                 &sc->sc_ad1848);

though it seems truly odd that it could have successfully linked without
properly resolving that address...


There's also a bug in db_disasm:  when trying to disassemble the instructions
at address 0, it gets a page fault and crashes (back into ddb, fortunately).
ddb should probably catch page-faults when fetching data from suspicious
addresses.