Re: [PATCH v6 1/5] x86: Move per cpu cpu_llc_shared_map to a field in struct cpuinfo_x86

[Ingo Molnar <mingo@kernel.org>]

* Kevin Winchester <kjwinchester@gmail.com> wrote:

> On 29 April 2012 20:37, H. Peter Anvin <hpa@zytor.com> wrote:
> > On 04/29/2012 04:33 PM, Kevin Winchester wrote:
> >> Commit 141168c36cde ("x86: Simplify code by removing a !SMP #ifdefs from
> >> 'struct cpuinfo_x86'") caused the compilation error:
> >>
> >> mce_amd.c:(.cpuinit.text+0x4723): undefined reference to 'cpu_llc_shared_map'
> >>
> >> by removing an #ifdef CONFIG_SMP around a block containing a reference
> >> to cpu_llc_shared_map.  Rather than replace the #ifdef, move
> >> cpu_llc_shared_map to be a new cpumask_t field llc_shared_map in
> >> struct cpuinfo_x86 and adjust all references to cpu_llc_shared_map.
> >>
> >
> > Okay...  I must not get this.
> >
> > Why is this better than just moving the DEFINE_PER_CPU to a place which
> > isn't dependent on SMP?
> >
> 
> To be honest, the idea was suggested by Ingo a while back, and 
> I just volunteered to implement it.  I believe the idea was to 
> work towards gathering all CPU-specific data for SMP and !SMP 
> into one place.  That said, moving the DEFINE_PER_CPU 
> elsewhere would likely still give the same benefits as my 
> patch in terms of ifdef reduction.  I cannot really see a 
> benefit/downside either way, really.
> 
> Perhaps Ingo will chime in, and you and he (and anyone else 
> with an opinion) can figure out the best way of accomplishing 
> this simplification, and then I can adjust my series 
> accordingly.

Well, cpu_llc_shared_map is really a CPU attribute and as such I 
think the logical place for it is "struct cpuinfo_x86". Peter, 
any objections to that?

Thanks,

	Ingo