[Cryptography] Speculation re Intel HW cockup; reqs. OS rewrites & slow execution

[Tom Mitchell]

On Thu, Jan 4, 2018 at 2:23 PM, Henry Baker <hbaker1 at pipeline.com> wrote:

> At 01:32 PM 1/4/2018, jamesd at echeque.com wrote:
> >The fix is not to abandon speculative execution, but to do it right, and
> chances are that doing it right is going to be faster and more efficient,
> not slower and less efficient.
>
> OK, I'll bite.
>
> How would you do speculative execution "right" ?
>



One  is to make sure we know which of the multitude of problems
is being discussed.

Two is to have more (or less) registers to speculate into.  Speculation
depths of 200+ instructions touches at a lot of stuff and has
a lot of results that may need to be canceled and unwound.

In this mailing list the one that is going to be interesting next month is
the side channel that allows
processes at different mandatory security levels inside a machine to
communicate
at bit rates much faster than 1 bps.   This one may well lurk behind door
#2 for security experts
long after the big nasty bug that makes soggy pudding of cloud computing
gets addressed.

To do speculative execution "right" is complex and hard.  It solves some
memory and
latency problems so one solution is to remove the need with better memory
and I/O.

Registers... some will recall the Z80 with two sets of registers. Something
like that
might be applied in hardware but then the architecture and perhaps the ISA
changes
would break way too much stuff.

No easy answer.   Time to dust off my system and computer architecture
library.
The old books may have clues that the modern "kids" no longer see behind
a logic block by another team.

Most system bugs I got involved in over the years were at team boundarys
both hardware and software.  i.e. organizational and management.





-- 
  T o m    M i t c h e l l
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.metzdowd.com/pipermail/cryptography/attachments/20180104/31c67ab6/attachment.html>