Andreas Bogk mentions:
[about the perl script that searches for the specific
sequence:]> Do *not* rely on this program. The program could create the sequence
> at runtime.
Right. The sequence is not at all hard to hide from such a per
script. Some people mentioned that there are now 4 known opcodes that
stop a pentium.
Some claim that "it stops the clock" because the temp of the processor
drops dramatically. I''m not sure. How much of the power that a Pentium
draws would be dissipated in the clock distribution circuitery? I
expect not a seriously large part, so the temp of the processor indeed
should drop dramatically if it simply halts, and no more changes are
incurred in most of the processor.
At the moment it seems that:
- All Intel-socket7 processors are affected: Pentium, Pentium-MMX.
- Cyrix (= IBM) processors fall for a closely similar prank, but
that can be configured away. (Use set6x86 for this).
- AMDs are immune to this.
- 486s, Pentium Pros and Pentium IIs don''t fall for this.
- I expect 386''s to behave, Overdrive processors to fail.
Some claim this is really bad. But keep in mind that anybody who "can
execute arbitrary code on your machine" can already cause lots of
trouble by starving the system of lots of resources at the same time.
(last time I checked, fsck-ing was faster than waiting for a Linux
system to recover from "out of memory").
Regards,
Roger Wolff.
