Re: kernel: stop_machine: report (un)dead code (and feedback request)

[Ingo Molnar <mingo@kernel.org>]

* Chris Wilson <chris@chris-wilson.co.uk> wrote:

> On Thu, Nov 26, 2015 at 03:45:59PM +0100, Valentin Rothberg wrote:
> > Hi Chris,
> > 
> > your commit 4c477de14237 ("kernel: remove stop_machine() Kconfig
> > dependency") has shown up in today's linux-next tree (20151126).
> > The commit changes the #ifdef condition in kernel/stop_machine.c
> > from/to:
> > 
> >     -#if defined(CONFIG_STOP_MACHINE) && defined(CONFIG_SMP)
> >     +#if defined(CONFIG_SMP) || defined(CONFIG_HOTPLUG_CPU)
> > 
> > Although this change fixes certain configs on X86, the condition now is
> > a tautology since CONFIG_SMP is already required to compile the file:
> > 
> >     kernel/Makefile:65:obj-$(CONFIG_SMP) += stop_machine.o
> > 
> > AFAIU, we can safely remove this #ifdef?
> 
> That seems logical. The argument in favour of it would be to keep the
> ifdeffery around the function defintion the same as the function
> declaration in stop_machine.h.

But this would introduce a bit of fragility: we could re-introduce the same 
regression that the commit fixes, if we ever changed the SMP dependency for 
stop_machine.c.

Thanks,

	Ingo