rovka added inline comments.
================
Comment at: include/llvm/IR/DerivedTypes.h:515
+ else
+ VTy = VectorType::getTruncatedElementVectorType(VTy);
+ }
----------------
Why not move this logic to getTruncatedElementVectorType and drop
getNarrowerFpElementVectorType altogether?
================
Comment at: include/llvm/IR/Intrinsics.h:130
+ Kind == PtrToElt || Kind == VecElementArgument ||
+ Kind == Subdivide2Argument || Kind == Subdivide4Argument);
return Argument_Info >> 3;
----------------
Not directly related to your change: Is there a reason why VecOfAnyPtrsToElt
isn't in this assert? If not, maybe this is a good time to add a
First/LastArgumentKind and just check that Kind is in the range.
================
Comment at: lib/IR/Function.cpp:989
+ Type *Ty = Tys[D.getArgumentNumber()];
+ if (VectorType *VTy = dyn_cast<VectorType>(Ty))
+ return VectorType::getSubdividedVectorType(VTy, 1);
----------------
Maybe replace this with an assert and drop the unreachable.
================
Comment at: lib/IR/Function.cpp:997
+ return VectorType::getSubdividedVectorType(VTy, 2);
+ llvm_unreachable("unhandled");
+ }
----------------
Ditto,
================
Comment at: lib/IR/Function.cpp:1312
+ case IITDescriptor::Subdivide4Argument: {
+ // This may only be used when referring to a previous vector argument.
+ if (D.getArgumentNumber() >= ArgTys.size())
----------------
Can you share this code? Either a small helper or a fallthrough with a repeat
check on D.Kind to get the number of subdivisions would work.
Repository:
rL LLVM
CHANGES SINCE LAST ACTION
https://reviews.llvm.org/D67549/new/
https://reviews.llvm.org/D67549
_______________________________________________
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
