Re: Giant strides backward (re: Pentium LOCK bug)
ralf@localhost.localdomain
Tue, 11 Nov 1997 07:26:48 +0100
On Mon, Nov 10, 1997 at 09:57:46PM +0000, Paul Flinders wrote:
> I know that there have been lots of discussion about the Pentium LOCK bug so I
> started reading through the Pentium programming manual out of curiosity.
>
> I found this little gem
>
> On the Intel 286 processor, the LOCK prefix is sensitive to IOPL; if
> CPL is less privileged than the IOPL, a general protection exception
> is generated. On the Intel386 DX, Intel486, and Pentium processors,
> no check against IOPL is performed.
>
> If only they'd left that in.....
No, that'd be a design bug that would ruin all atomic operations as needed
for multithreading.
Ralf