Interesting Compile Problem - 2.6 Kernel

[Lonni J Friedman]

On 10/9/07, Steve Jardine <sjardine at acm.org> wrote:
> All,
>
>     I have a real strange one to relate to you.
>
>     System: 4 Xeon Processor, Dual Core with Hyperthreading
>             PCI-E, PCI backplane
>
>     System A: CPU Family 15, Model 4, Stepping 1
>     System B: CPU Family 15, Model 4, Stepping 8
>
>     Now, System A compiles fine with the processor type of Pentium 4/Xeon, and SMP/Multicasting set. All 8 cores show up without a problem. Boots perfectly.
>
>     However, System B with the same settings cannot boot. Kernel panics when attempting to boot.
>
>     What I have found out: Stepping 1 is a true Pentium 4 core, works as advertised. System B can *only* have the kernel compiled as a CPU Type of Pentium PRO. It shows all 8 cores and will perform hyperthreading if the flags are set in the kernel configuration with the CPU type of Pentium PRO.
>
>     I have read that the stepping 8 Xeon does not have 64 bit cvapability for processing or for I/O to/from the cores. It works the way I compiled it, but I would like to know why it must be this way??

You're asking us why Intel made the design decisions that they made?

-- 
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
L. Friedman                                    netllama at gmail.com
LlamaLand                       https://netllama.linux-sxs.org