Re: [PATCH 1/5] ACPI: NUMA: Add 'nohmat' option

[Dan Williams]

On Wed, Mar 18, 2020 at 1:24 AM Rafael J. Wysocki <rafael@xxxxxxxxxx> wrote:
>
> On Wed, Mar 18, 2020 at 1:09 AM Dan Williams <dan.j.williams@xxxxxxxxx> wrote:
> >
> > On Mon, Mar 2, 2020 at 2:36 PM Dan Williams <dan.j.williams@xxxxxxxxx> wrote:
> > >
> > > Disable parsing of the HMAT for debug, to workaround broken platform
> > > instances, or cases where it is otherwise not wanted.
> >
> > Rafael, any heartburn with this change to the numa= option?
> >
> > ...as I look at this I realize I failed to also update
> > Documentation/x86/x86_64/boot-options.rst, will fix.
>
> Thanks!
>
> Apart from this just a minor nit below.
>
> > >
> > > Cc: x86@xxxxxxxxxx
> > > Cc: "Rafael J. Wysocki" <rjw@xxxxxxxxxxxxx>
> > > Cc: Dave Hansen <dave.hansen@xxxxxxxxxxxxxxx>
> > > Cc: Andy Lutomirski <luto@xxxxxxxxxx>
> > > Cc: Peter Zijlstra <peterz@xxxxxxxxxxxxx>
> > > Cc: Thomas Gleixner <tglx@xxxxxxxxxxxxx>
> > > Cc: Ingo Molnar <mingo@xxxxxxxxxx>
> > > Cc: Borislav Petkov <bp@xxxxxxxxx>
> > > Cc: "H. Peter Anvin" <hpa@xxxxxxxxx>
> > > Signed-off-by: Dan Williams <dan.j.williams@xxxxxxxxx>
> > > ---
> > >  arch/x86/mm/numa.c       |    4 ++++
> > >  drivers/acpi/numa/hmat.c |    3 ++-
> > >  include/acpi/acpi_numa.h |    1 +
> > >  3 files changed, 7 insertions(+), 1 deletion(-)
> > >
> > > diff --git a/arch/x86/mm/numa.c b/arch/x86/mm/numa.c
> > > index 59ba008504dc..22de2e2610c1 100644
> > > --- a/arch/x86/mm/numa.c
> > > +++ b/arch/x86/mm/numa.c
> > > @@ -44,6 +44,10 @@ static __init int numa_setup(char *opt)
> > >  #ifdef CONFIG_ACPI_NUMA
> > >         if (!strncmp(opt, "noacpi", 6))
> > >                 acpi_numa = -1;
> > > +#ifdef CONFIG_ACPI_HMAT
> > > +       if (!strncmp(opt, "nohmat", 6))
> > > +               hmat_disable = 1;
> > > +#endif
>
> I wonder if IS_ENABLED() would work here?

I took a look. hmat_disable, acpi_numa, and numa_emu_cmdline() are in
other compilation units. I could wrap writing those variables with
helper functions, and change numa_emu_cmdline(), to compile away when
their respective configuration options are not present.

Should we do that in general to have a touch point to report "you
specified an option that is invalid for your current kernel
configuration"? I'm happy to do that as a follow-on if you think it's
worthwhile.