Re: [RFT/RFC PATCH v3 3/5] cpu-topology: Move cpu topology code to common code.

[Atish Patra]

On 4/16/19 6:23 AM, Sudeep Holla wrote:
> On Mon, Apr 15, 2019 at 03:08:45PM -0700, Atish Patra wrote:
> > On 4/15/19 8:27 AM, Sudeep Holla wrote:
> > > Hi Atish,
> > >
> > > Thanks again for doing this. Overall changes look good except a couple
> > > of minor nit, see below.
> > >
> > > On Wed, Mar 20, 2019 at 04:48:04PM -0700, Atish Patra wrote:
> > > > Both RISC-V & ARM64 are using cpu-map device tree to describe
> > > > their cpu topology. It's better to move the relevant code to
> > > > a common place instead of duplicate code.
> > > >
> > > > Signed-off-by: Atish Patra <atish.patra@xxxxxxx>
> > > > Tested-by: Jeffrey Hugo <jhugo@xxxxxxxxxxxxxx>
> > > > ---
> > > >    arch/arm64/include/asm/topology.h |  23 ---
> > > >    arch/arm64/kernel/topology.c      | 303 +-----------------------------
> > > >    drivers/base/arch_topology.c      | 298 ++++++++++++++++++++++++++++-
> > > >    drivers/base/topology.c           |   1 +
> > > >    include/linux/arch_topology.h     |  28 +++
> > > >    5 files changed, 330 insertions(+), 323 deletions(-)
> > >
> > > [...]
> > >
> > > > diff --git a/drivers/base/arch_topology.c b/drivers/base/arch_topology.c
> > > > index edfcf8d9..6cc6a860 100644
> > > > --- a/drivers/base/arch_topology.c
> > > > +++ b/drivers/base/arch_topology.c
> > > > @@ -6,8 +6,8 @@
> > > >     * Written by: Juri Lelli, ARM Ltd.
> > > >     */
> > > > -#include <linux/acpi.h>
> > > >    #include <linux/arch_topology.h>
> > > > +#include <linux/acpi.h>
> > > >    #include <linux/cpu.h>
> > > >    #include <linux/cpufreq.h>
> > > >    #include <linux/device.h>
> > > > @@ -16,6 +16,11 @@
> > > >    #include <linux/string.h>
> > > >    #include <linux/sched/topology.h>
> > > >    #include <linux/cpuset.h>
> > > > +#include <linux/cpumask.h>
> > > > +#include <linux/init.h>
> > > > +#include <linux/percpu.h>
> > > > +#include <linux/sched.h>
> > > > +#include <linux/smp.h>
> > > >    DEFINE_PER_CPU(unsigned long, freq_scale) = SCHED_CAPACITY_SCALE;
> > > > @@ -278,3 +283,294 @@ static void parsing_done_workfn(struct work_struct *work)
> > > >    #else
> > > >    core_initcall(free_raw_capacity);
> > > >    #endif
> > > > +
> > > > +#if defined(CONFIG_ARM64) || defined(CONFIG_RISCV)
> > >
> > > Why can't the above one be just GENERIC_ARCH_TOPOLOGY ?
> > > I may be missing to find it myself, but would like to know.
> > GENERIC_ARCH_TOPOLOGY is now used for both RISCV, ARM & ARM64.
> > The below functions under this #ifdef have different implementation for ARM
> > and ARM64.
> >
> > parse_dt_topology
> > cpu_coregroup_mask
> > update_siblings_masks
> >
> > While we can combine the later two functions and move them to common code as
> > well, parse_dt_topology is significantly different.
>
> Sure, had a quick glance and indeed they may look different, but won't
> it defeat the purpose of this binding consolidation ?
I didn't want change too much at first go.

> > That's why we need some kind of #ifdef or renaming of parse_dt_topology for
> > ARM32 code.
>
> I am fine if we want to take this up later to keep the impact minimum.
> But cpu_coregroup_mask and update_siblings_masks can and must be unified.

Sure. I will just leave parse_dt_topology as it is for now and unify 
other two functions.

I think we should unify parse_dt_topology in separate series.

Regards,
Atish
> In fact the existing generic version must work on ARM32 too.
>
> > Thanks for the review!!
>
> You are welcome.
>
> --
> Regards,
> Sudeep