Re: [PATCH v5 6/6] mtrr, mm, x86: Enhance MTRR checks for KVA huge page mapping

[Toshi Kani]

On Mon, 2015-05-18 at 21:01 +0200, Borislav Petkov wrote:
> On Mon, May 18, 2015 at 11:22:39AM -0600, Toshi Kani wrote:
> > > diff --git a/arch/x86/mm/pgtable.c b/arch/x86/mm/pgtable.c
> > > index c30f9819786b..f1894daa79ee 100644
> > > --- a/arch/x86/mm/pgtable.c
> > > +++ b/arch/x86/mm/pgtable.c
> > > @@ -566,19 +566,24 @@ void native_set_fixmap(enum fixed_addresses idx, phys_addr_t phys,
> > >  /**
> > >   * pud_set_huge - setup kernel PUD mapping
> > >   *
> > > - * MTRR can override PAT memory types with 4KiB granularity.  Therefore,
> > > - * this function does not set up a huge page when the range is covered
> > > - * by a non-WB type of MTRR.  MTRR_TYPE_INVALID indicates that MTRR are
> > > - * disabled.
> > > + * MTRRs can override PAT memory types with 4KiB granularity. Therefore,
> > > + * this function sets up a huge page only if all of the following
> > > + * conditions are met:
> > 
> > It should be "if any of the following condition is met".  Or, does NOT
> > setup if all of ...
> > 
> > > + *
> > > + *  - MTRRs are disabled.
> > > + *  - The range is mapped uniformly by an MTRR, i.e. the range is
> > > + *    fully covered by a single MTRR entry or the default type.
> > > + *  - The MTRR memory type is WB.
> 
> Hmm, ok, so this is kinda like "any" but they also depend on each other.
> So it is
> 
> If
> 	- MTRRs are disabled
> 
> 	or
> 
> 	- MTRRs are enabled and the range is completely covered by a single MTRR
> 
> 	or
> 
> 	 - MTRRs are enabled and the range is not completely covered by a
> 	 single MTRR but the memory type of the range is WB, even if covered by
> 	 multiple MTRRs.
> 
> Right?

Well, #2 and #3 are independent. That is, uniform can be set regardless
of a type value, and WB can be returned regardless of a uniform value.  

#1 is a new condition added per your comment that uniform no longer
covers the MTRR disabled case.  Yes, #2 and #3 depend on #1 being false.

> So tell me this: why do we need to repeat that over those KVA helpers?
> It's not like the callers can do anything about it, can they?
>
> So maybe that comment - expanded into more detail - should be over
> mtrr_type_lookup() only. That'll be better, methinks.

The caller is responsible for verifying the conditions that are safe to
create huge page.  So, I think the comments are needed here to state
such conditions.

Thanks,
-Toshi

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/