In message <v690utaprm.fsf@kechara.flame.org>, Michael Graff writes:
>Here are the contents as I see them:

>Files changed in mod M310-hangfix.kobj:
>        cd KERNEL
>        I i386/OBJ/locore.o
>        I i386/OBJ/machdep.o

That's my understanding.  I'm told there will also be patches for 3.0 and
2.1 of BSDI.  The engineer who did it says "we are not at liberty to
disuss the mechanism".  Another report I saw says that disabling L1 cache
prevents the hang, but I don't really have *any* idea what this does; it
could be as simple as there being some other undocumented opcode that
makes the chip not wedge on this, or it could be some pretty deep magic.

I would love to know what the deal is, unfortunately, I'm afraid any
BSDI answer may end up being under our NDA.

(I will, however, try to find out whatever I can from the guy I know
at Intel.  Maybe there's *something* they can say.)

-s