[llvm] [VPlan] Compute induction end values in VPlan. (PR #112145)

[via llvm-commits]

================
@@ -9762,13 +9732,18 @@ void VPDerivedIVRecipe::execute(VPTransformState &State) {
     State.Builder.setFastMathFlags(FPBinOp->getFastMathFlags());
 
   Value *Step = State.get(getStepValue(), VPLane(0));
-  Value *CanonicalIV = State.get(getOperand(1), VPLane(0));
+  Value *Index = State.get(getOperand(1), VPLane(0));
   Value *DerivedIV = emitTransformedIndex(
-      State.Builder, CanonicalIV, getStartValue()->getLiveInIRValue(), Step,
-      Kind, cast_if_present<BinaryOperator>(FPBinOp));
+      State.Builder, Index, getStartValue()->getLiveInIRValue(), Step, Kind,
+      cast_if_present<BinaryOperator>(FPBinOp));
   DerivedIV->setName(Name);
-  assert(DerivedIV != CanonicalIV && "IV didn't need transforming?");
-
+  // If index is the vector trip count, the concrete value will only be set in
+  // prepareToExecute, leading to missed simplifications, e.g. if it is 0.
+  // TODO: Remove the special case for the vector trip count once it is computed
----------------
ayalz wrote:

Thanks, this is clear.

Knowing VTC is zero at compile-time suggests more simplification than eliminating redundant DeriveIV's ... ideally VF's (and UF's) should be restricted so as to prevent this?

https://github.com/llvm/llvm-project/pull/112145