Re: [PATCH] kernel: clean up USE_GENERIC_SMP_HELPERS

[Peter Zijlstra <a.p.zijlstra@chello.nl>]

On Thu, 2010-12-23 at 18:02 +0800, Cong Wang wrote:
> 于 2010年12月22日 21:08, Peter Zijlstra 写道:
> > On Mon, 2010-11-29 at 14:06 +0800, Amerigo Wang wrote:
> >> For arch which needs USE_GENERIC_SMP_HELPERS, it has to
> >> select USE_GENERIC_SMP_HELPERS, rather than leaving a choice
> >> to user, since they don't provide their own implementions.
> >>
> >
> >>   config MN10300_CURRENT_IN_E2
> >> diff --git a/arch/x86/Kconfig b/arch/x86/Kconfig
> >> index e330da2..736beea 100644
> >> --- a/arch/x86/Kconfig
> >> +++ b/arch/x86/Kconfig
> >> @@ -65,6 +65,7 @@ config X86
> >>   	select HAVE_SPARSE_IRQ
> >>   	select GENERIC_IRQ_PROBE
> >>   	select GENERIC_PENDING_IRQ if SMP
> >> +	select USE_GENERIC_SMP_HELPERS if SMP
> >>
> >>   config INSTRUCTION_DECODER
> >>   	def_bool (KPROBES || PERF_EVENTS)
> >> @@ -203,10 +204,6 @@ config HAVE_INTEL_TXT
> >>   	def_bool y
> >>   	depends on EXPERIMENTAL&&  DMAR&&  ACPI
> >>
> >> -config USE_GENERIC_SMP_HELPERS
> >> -	def_bool y
> >> -	depends on SMP
> >> -
> >>   config X86_32_SMP
> >>   	def_bool y
> >>   	depends on X86_32&&  SMP
> >
> > That changelog and the patch don't match, the patch is a total NOP,
> > users never see that config option.
> 
> Ok, but isn't 'select' is supposed to be better to express this here? :)

It sure is shorter, so on that ground one can argue so indeed. Its just
that the discrepancy between the Changelog (describing an actual change
in semantics) and the patch (being a NOP) bothered me.

> This patch also moves on_each_cpu() to kernel/smp.c, btw.

Right, it does that, it would probably have been better if that were a
patch on its own, seeing as its a completely orthogonal change.