[llvm] [LV] Don't vectorize epilogue with scalable VF if no iterations remain. (PR #149789)

[Paul Walker via llvm-commits]

================
@@ -4479,6 +4479,28 @@ VectorizationFactor LoopVectorizationPlanner::selectEpilogueVectorizationFactor(
   Type *TCType = Legal->getWidestInductionType();
   const SCEV *RemainingIterations = nullptr;
   unsigned MaxTripCount = 0;
+  if (MainLoopVF.isFixed()) {
----------------
paulwalker-arm wrote:

I've a few SCEV patches, one at review and one delayed by https://github.com/llvm/llvm-project/pull/146102 that should fix this.

That said, we should not be unnecessarily restricting things to fixed length vectors.  As a minimum I think the `MainLoopVF.isFixed()` check can be move to just prior to the point where `MaxTripCount` is recalculated, with the prior call to `getURemExpr()` using `SE.getElementCount(TCType, MainLoopVF * IC)`.

https://github.com/llvm/llvm-project/pull/149789