[PATCH] powerpc: Optimise the 64bit optimised __clear_user
Michael Neuling
mikey at neuling.org
Thu Jun 7 16:12:09 EST 2012
Michael Ellerman <michael at ellerman.id.au> wrote:
> On Thu, 2012-06-07 at 16:05 +1000, Michael Neuling wrote:
> > Benjamin Herrenschmidt <benh at kernel.crashing.org> wrote:
> >
> > > On Wed, 2012-06-06 at 18:40 +0200, Segher Boessenkool wrote:
> > > > > +err1; dcbz r0,r3
> > > >
> > > > There is no such instruction, you probably meant "dcbz 0,r3"?
> > >
> > > This reminds me... what would happen if we changed all our
> > >
> > > #define r0 0
> > > #define r1 1
> > >
> > > etc... to:
> > >
> > > #define r0 %r0
> > > #define r1 %r1
> > >
> > > ?
> > >
> > > I'm thinking it might help catch that sort of nasties (and some of them
> > > can be really nasty, such as inverting mfspr/mtspr arguments, or vs ori,
> > > etc... ). I'm sure we'd have a problem with a few macros & inline
> > > constructs but nothing we can't fix..
> >
> > One problem with this is when we construct the instructions, like using
> > anything from ppc-opcode.h. eg. using PPC_POPCNTB would need to go from:
> > PPC_POPCNTB(r3,r3)
> > to:
> > PPC_POPCNTB(3,3)
>
> #define R(x) x
#define R(x) (x)
> #define PPC_POPCNTB(R(3), R(3))
Maybe, looks pretty gross but you're the maintainer! :-)
Mikey
More information about the Linuxppc-dev
mailing list