On Wed, 2007-04-04 at 17:38 -0700, Chris Lattner wrote: > > Implement the llvm.bit.part_select.iN.iN.iN overloaded intrinsic. > > Should this check that the third type is smaller than the other two?
No, because that commit comment is in error. I was thinking of the
prototype for part_set (which I deleted from the implementation just
before commit). The prototype is actually like this:
declare i17 @llvm.bit.part_select.i17.i17 (i17 %val, i32 %loBit, i32 %
hiBit)
which only has two variable parts: result, 1st arg. And those two must
be the same width which is what this patch does. The other parameters
are i32.
Reid.
>
> -Chris
>
> >
> > ---
> > Diffs of the changes: (+2 -0)
> >
> > Verifier.cpp | 2 ++
> > 1 files changed, 2 insertions(+)
> >
> >
> > Index: llvm/lib/VMCore/Verifier.cpp
> > diff -u llvm/lib/VMCore/Verifier.cpp:1.198 llvm/lib/VMCore/
> > Verifier.cpp:1.199
> > --- llvm/lib/VMCore/Verifier.cpp:1.198 Sun Apr 1 02:22:57 2007
> > +++ llvm/lib/VMCore/Verifier.cpp Wed Apr 4 18:48:25 2007
> > @@ -1027,6 +1027,8 @@
> > case Intrinsic::bswap:
> > if (GotBits < 16 || GotBits % 16 != 0)
> > CheckFailed("Intrinsic requires even byte width
> > argument", F);
> > + /* FALL THROUGH */
> > + case Intrinsic::bit_part_select:
> > if (ArgNo == 1) {
> > unsigned ResultBits =
> > cast<IntegerType>(FTy->getReturnType())->getBitWidth();
> >
> >
> >
> > _______________________________________________
> > llvm-commits mailing list
> > llvm-commits@cs.uiuc.edu
> > http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
>
> _______________________________________________
> llvm-commits mailing list
> llvm-commits@cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
signature.asc
Description: This is a digitally signed message part
_______________________________________________ llvm-commits mailing list llvm-commits@cs.uiuc.edu http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
