[PATCH] D17834: [AArch64] fold 'isPositive' vector integer operations (PR26819)

[Tim Northover via llvm-commits]

t.p.northover added inline comments.

================
Comment at: lib/Target/AArch64/AArch64ISelLowering.cpp:7447
@@ +7446,3 @@
+    case MVT::v4i32:
+    case MVT::v2i64: if (!Subtarget->hasNEON()) return SDValue(); break;
+  }
----------------
t.p.northover wrote:
> This looks like a vestige of x86's different levels of vector support. On AArch64 all vector types are legal or none are.
> 
> So I'd suggest making this *just* the NEON check (or even an assertion), and moving the foldVectorXorShiftIntoCmp call below the isBeforeLegalizeOps check.
Oh, actually you probably do need an isVectorType check to avoid doing nasty things to scalars.


http://reviews.llvm.org/D17834