Re: [PATCH v4 1/7] x86, mm, pat: Set WT to PA7 slot of PAT MSR

[Toshi Kani]

On Mon, 2014-11-03 at 19:15 +0100, Thomas Gleixner wrote:
> On Mon, 3 Nov 2014, Andy Lutomirski wrote:
> > On Mon, Nov 3, 2014 at 9:47 AM, Toshi Kani <toshi.kani@xxxxxx> wrote:
> > > On Mon, 2014-11-03 at 18:14 +0100, Thomas Gleixner wrote:
> > >> On Mon, 27 Oct 2014, Toshi Kani wrote:
> > >> > +   } else {
> > >> > +           /*
> > >> > +            * PAT full support. WT is set to slot 7, which minimizes
> > >> > +            * the risk of using the PAT bit as slot 3 is UC and is
> > >> > +            * currently unused. Slot 4 should remain as reserved.
> > >>
> > >> This comment makes no sense. What minimizes which risk and what has
> > >> this to do with slot 3 and slot 4?
> > >
> > > This is for precaution.  Since the patch enables the PAT bit the first
> > > time, it was suggested that we keep slot 4 reserved and set it to WB.
> > > The PAT bit still has no effect to slot 0/1/2 (WB/WC/UC-) after this
> > > patch.  Slot 7 is the safest slot since slot 3 (UC) is unused today.
> > >
> > > https://lkml.org/lkml/2014/9/4/691
> > > https://lkml.org/lkml/2014/9/5/394
> > >
> > 
> > I would clarify the comment, since this really has nothing to do with
> > slot 3 being unused.  How about:
> > 
> > We put WT in slot 7 to improve robustness in the presence of errata
> > that might cause the high PAT bit to be ignored.  This way a buggy
> > slot 7 access will hit slot 3, and slot 3 is UC, so at worst we lose
> > performance without causing a correctness issue.  Pentium 4 erratum
> > N46 is an example of such an erratum, although we try not to use PAT
> > at all on affected CPUs.
> 
> Indeed. That makes a lot more sense.
>  
> > >> > +            *
> > >> > +            *  PTE encoding used in Linux:
> > >> > +            *      PAT
> > >> > +            *      |PCD
> > >> > +            *      ||PWT  PAT
> > >> > +            *      |||    slot
> > >> > +            *      000    0    WB : _PAGE_CACHE_MODE_WB
> > >> > +            *      001    1    WC : _PAGE_CACHE_MODE_WC
> > >> > +            *      010    2    UC-: _PAGE_CACHE_MODE_UC_MINUS
> > >> > +            *      011    3    UC : _PAGE_CACHE_MODE_UC
> > >> > +            *      100    4    <reserved>
> > >> > +            *      101    5    <reserved>
> > >> > +            *      110    6    <reserved>
> > >>
> > >> Well, they are still mapped to WB/WC/UC_MINUS ....
> > >
> > > Right, the reserved slots are also initialized with their safe values.
> > > However, the macros _PAGE_CACHE_MODE_XXX only refer to the slots
> > > specified above.
> 
> Then the table should reflect this, i.e.: reserved, but mapped to XX
> 
> And a comment below that explaining WHY we map the reserved slots.

Yes, I will update the table and add a comment.

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/