I misremembered Alistair's PR in my last message.  The device-level
problem is that the Buslogic probe falsely recognizes an Adaptec card,
not vice versa.  Alistair's patch to prevent this has *not* been
checked in.  Even if it were, it would still true that either probe
will smash the state of the other card if it runs after the other card
has been detected.

Alistair's PR is #1418; it attempts to fix both problems, but
(empirically) only the part that prevents false positives on the
Buslogic probe actually worked.