Sujet : Re: Chipsandcheese article on the CDC6600
De : anton (at) *nospam* mips.complang.tuwien.ac.at (Anton Ertl)
Groupes : comp.archDate : 26. Jul 2024, 17:36:07
Autres entêtes
Organisation : Institut fuer Computersprachen, Technische Universitaet Wien
Message-ID : <2024Jul26.183607@mips.complang.tuwien.ac.at>
References : 1 2 3 4 5 6 7 8 9 10 11 12 13
User-Agent : xrn 10.11
mitchalsup@aol.com (MitchAlsup1) writes:
On Thu, 25 Jul 2024 10:59:16 +0000, Anton Ertl wrote:
Now somebody will point out that sharing makes it possible for an
attacker to train branch predictors in one process to attack a
different process through Spectre and friends. While preventing
sharing would close that, it does not close training the predictors in
the same thread.
>
Not allowing a dependent AGEN to happen when the first AGEN takes
a fault ALSO prevents SPectré like attacks
Spectre does not need a fault. You are probably thinking of Meltdown.
That, at least has been fixed by Intel (and hopefully also ARM) in its
original variant pretty soon, although other variants have been
discovered since then (IIRC including some where the fault has nothing
to do with addresses).
Then not modifying
any cache prior to instruction retirement cements the door closed.
Not changing microarchitectural state (not just caches) through
misspeculation (invisible speculation) is a proper fix for Spectre,
and looks like the best fix to me.
- anton
-- 'Anyone trying for "industrial quality" ISA should avoid undefined behavior.' Mitch Alsup, <c17fcd89-f024-40e7-a594-88a85ac10d20o@googlegroups.com>