[llvm] [VPlan] Explicitly handle scalar pointer inductions. (PR #83068)

[via llvm-commits]

================
@@ -537,21 +540,54 @@ static VPValue *createScalarIVSteps(VPlan &Plan, const InductionDescriptor &ID,
     VecPreheader->appendRecipe(Step->getDefiningRecipe());
   }
 
-  VPScalarIVStepsRecipe *Steps = new VPScalarIVStepsRecipe(ID, BaseIV, Step);
+  VPScalarIVStepsRecipe *Steps = new VPScalarIVStepsRecipe(
+      BaseIV, Step, InductionOpcode,
+      FPBinOp ? FPBinOp->getFastMathFlags() : FastMathFlags());
   HeaderVPBB->insert(Steps, IP);
   return Steps;
 }
 
-/// If any user of a VPWidenIntOrFpInductionRecipe needs scalar values,
-/// provide them by building scalar steps off of the canonical scalar IV and
+/// Legalize VPWidenPointerInductionRecipe, by replacing it with a PtrAdd
+/// (IndStart, ScalarIVSteps (0, Step)) if only its scalar values are used, as
+/// VPWidenPointerInductionRecipe cannot generate scalars. Also optimize
----------------
ayalz wrote:

```suggestion
/// VPWidenPointerInductionRecipe will generate vectors only. Also optimize
```
So if scalar are used **along with vectors** of widen pointer induction, the former will be extracted from the latter; as opposed to widening int or fp inductions, where scalar users are fed scalar steps even if there are also vector users. Right?

https://github.com/llvm/llvm-project/pull/83068