Re: [RFC PATCH 4/9] powerpc/mm/book3s-64: Move _PAGE_PRESENT to the most significant bit

[Paul Mackerras <paulus@ozlabs.org>]

On Sat, Feb 20, 2016 at 10:11:14PM +0530, Aneesh Kumar K.V wrote:
> Paul Mackerras <paulus@samba.org> writes:
> 
> > This changes _PAGE_PRESENT for 64-bit Book 3S processors from 0x2 to
> > 0x8000_0000_0000_0000, because that is where PowerISA v3.0 CPUs in
> > radix mode will expect to find it.
> 
> All the changes in this patch related to _PAGE_PRESENT movement or are
> they cleanup that got added to this patch. I am looking at the hpte slot
> array changes and wondering how that is related to _PAGE_PRESENT.

I was preserving the property mentioned in this comment:

> >  /*
> >   * The linux hugepage PMD now include the pmd entries followed by the address
> >   * to the stashed pgtable_t. The stashed pgtable_t contains the hpte bits.
> > - * [ 1 bit secondary | 3 bit hidx | 1 bit valid | 000]. We use one byte per
> > + * [ 000 | 1 bit secondary | 3 bit hidx | 1 bit valid]. We use one byte per
> >   * each HPTE entry. With 16MB hugepage and 64K HPTE we need 256 entries and
> >   * with 4K HPTE we need 4096 entries. Both will fit in a 4K pgtable_t.
> >   *
> > - * The last three bits are intentionally left to zero. This memory location
> > + * The top three bits are intentionally left as zero. This memory location
> >   * are also used as normal page PTE pointers. So if we have any pointers
> >   * left around while we collapse a hugepage, we need to make sure
> >   * _PAGE_PRESENT bit of that is zero when we look at them

I don't know if this comment still applies, but now that _PAGE_PRESENT
is the top bit of a byte rather than one of the low bits, then to avoid
having _PAGE_PRESENT overlap these HPTE location bits in a byte, we
need to move the location bits.

It seems pretty bogus to me that we might interpret an array of these
bytes as a PTE, and if you're sure we never do that, we can drop this
change (and the now-misleading comment).

Paul.