Re: [RFC 00/15] x86_64: Optimize percpu accesses

[Eric W. Biederman]

Christoph Lameter <cl@xxxxxxxxxxxxxxxxxxxx> writes:

> H. Peter Anvin wrote:
>
>> And how much is that, especially on *small* systems?
>
> i386?
>
> i386 uses 4K mappings. There are just a few cpus supported, there is scarcity of
> ZONE_NORMAL memory so the per cpu areas really cannot get that big. See the
> cpu_alloc patchsets for i386.

i386 is fundamentally resource constrained.  However x86_32 should support a
strict superset of the machines the x86_64 kernel supports.

Because it is resource constrained in the lowmem zone you should not
be able to bring up all of the cpus on a huge cpu box.   But you should still
be able to boot and run the kernel.  So for percpu data we have effectively
same size constraints.

Eric
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/