Re: Pentium DEATH in user-mode

Richard B. Johnson (root@chaos.analogic.com)
Sat, 8 Nov 1997 21:51:01 -0500 (EST)


On Sat, 8 Nov 1997, Billy Harvey wrote:

> Do the unimplemented opcodes perhaps raise a flag anywhere in the CPU
> that we could test for?
>
> Billy Harvey
>
Illegal opcodes, or instructions that are not appropriate in user-mode
__should__ cause a trap, which is caught by priviliged kernel code.
That code would normally kill the task with a "core-dump". It could
also choose to ignore the instructions. For instance, user mode code
can issue a CLI instruction followed by a HLT and nothing will happen
(last time I checked). This allows a user to write and test sections
of code in "user-land" that may eventually exist within the kernel.

Cheers,
Dick Johnson

Richard B. Johnson
Project Engineer
Analogic Corporation
Penguin : Linux version 2.1.60 on an i586 machine (66.15 BogoMips).
Warning : It's hard to remain at the trailing edge of technology.