Re: [PATCH] x86/mm: Clean up the __[PHYSICAL/VIRTUAL]_MASK_SHIFT definitions a bit

From: Ingo Molnar
Date: Thu May 09 2019 - 15:21:22 EST



* Yury Norov <yury.norov@xxxxxxxxx> wrote:

> On Thu, May 09, 2019 at 11:01:31AM +0200, Ingo Molnar wrote:
> >
> > * Yury Norov <yury.norov@xxxxxxxxx> wrote:
> >
> > > __VIRTUAL_MASK_SHIFT is defined twice to the same valie in
> > > arch/x86/include/asm/page_32_types.h. Fix it.
> > >
> > > Signed-off-by: Yury Norov <ynorov@xxxxxxxxxxx>
> > > ---
> > > arch/x86/include/asm/page_32_types.h | 5 ++---
> > > 1 file changed, 2 insertions(+), 3 deletions(-)
> > >
> > > diff --git a/arch/x86/include/asm/page_32_types.h b/arch/x86/include/asm/page_32_types.h
> > > index 0d5c739eebd7..9bfac5c80d89 100644
> > > --- a/arch/x86/include/asm/page_32_types.h
> > > +++ b/arch/x86/include/asm/page_32_types.h
> > > @@ -28,6 +28,8 @@
> > > #define MCE_STACK 0
> > > #define N_EXCEPTION_STACKS 1
> > >
> > > +#define __VIRTUAL_MASK_SHIFT 32
> > > +
> > > #ifdef CONFIG_X86_PAE
> > > /*
> > > * This is beyond the 44 bit limit imposed by the 32bit long pfns,
> > > @@ -36,11 +38,8 @@
> > > * The real limit is still 44 bits.
> > > */
> > > #define __PHYSICAL_MASK_SHIFT 52
> > > -#define __VIRTUAL_MASK_SHIFT 32
> > > -
> > > #else /* !CONFIG_X86_PAE */
> > > #define __PHYSICAL_MASK_SHIFT 32
> > > -#define __VIRTUAL_MASK_SHIFT 32
> > > #endif /* CONFIG_X86_PAE */
> >
> > I think it's clearer to see them defined where the physical mask shift is
> > defined.
> >
> > How about the patch below? It does away with the weird formatting and
> > cleans up both the comments and the style of the definition:
> >
> > /*
> > * 52 bits on PAE is beyond the 44-bit limit imposed by the
> > * 32-bit long PFNs, but we need the full mask to make sure
> > * inverted PROT_NONE entries have all the host bits set
> > * in a guest. The real limit is still 44 bits.
> > */
> > #ifdef CONFIG_X86_PAE
> > # define __PHYSICAL_MASK_SHIFT 52
> > # define __VIRTUAL_MASK_SHIFT 32
> > #else
> > # define __PHYSICAL_MASK_SHIFT 32
> > # define __VIRTUAL_MASK_SHIFT 32
> > #endif
> >
> > ?
>
> My main concern was about double definition. It pretty looks like a
> bug. But if it's intentional, I'm OK. In the patch below, could you
> please add some note to the comment that __VIRTUAL_MASK_SHIFT defined
> twice intentionally?

It's not defined "twice", it has values set in the PAE and the non-PAE
branch - just like __PHYSICAL_MASK_SHIFT.

__VIRTUAL_MASK_SHIFT happens to have the same value in both branches, but
that's OK and pretty standard and happens in other headers too.

Thanks,

Ingo