[llvm] [LoopVectorize] Add cost of generating tail-folding mask to the loop (PR #130565)

[Florian Hahn via llvm-commits]

================
@@ -4601,8 +4601,32 @@ VectorizationFactor LoopVectorizationPlanner::selectVectorizationFactor() {
         continue;
 
       InstructionCost C = CM.expectedCost(VF);
-      VectorizationFactor Candidate(VF, C, ScalarCost.ScalarCost);
 
+      // Add on other costs that are modelled in VPlan, but not in the legacy
+      // cost model.
+      VPCostContext CostCtx(CM.TTI, *CM.TLI, CM.Legal->getWidestInductionType(),
+                            CM, CM.CostKind);
+      if (VPRegionBlock *VectorRegion = P->getVectorLoopRegion()) {
+        for (VPBlockBase *Block :
+             vp_depth_first_shallow(VectorRegion->getEntry())) {
----------------
fhahn wrote:

could be slightly simplified using `blocksOnly`:
```suggestion
  for (VPBasicBlock *VPBB : VPBlockUtils::blocksOnly<VPBasicBlock>(vp_depth_first_shallow(VectorRegion->getEntry())) {
```

https://github.com/llvm/llvm-project/pull/130565