[llvm] [LV] Fix runtime-VF logic when generating RT-checks (PR #130118)

[David Sherwood via llvm-commits]

================
@@ -1924,21 +1924,17 @@ class GeneratedRTChecks {
                                  "vector.memcheck");
 
       auto DiffChecks = RtPtrChecking.getDiffChecks();
-      if (DiffChecks) {
-        Value *RuntimeVF = nullptr;
-        MemRuntimeCheckCond = addDiffRuntimeChecks(
-            MemCheckBlock->getTerminator(), *DiffChecks, MemCheckExp,
-            [VF, &RuntimeVF](IRBuilderBase &B, unsigned Bits) {
-              if (!RuntimeVF)
-                RuntimeVF = getRuntimeVF(B, B.getIntNTy(Bits), VF);
-              return RuntimeVF;
-            },
-            IC);
-      } else {
-        MemRuntimeCheckCond = addRuntimeChecks(
-            MemCheckBlock->getTerminator(), L, RtPtrChecking.getChecks(),
-            MemCheckExp, VectorizerParams::HoistRuntimeChecks);
-      }
+      MemRuntimeCheckCond =
+          DiffChecks
+              ? addDiffRuntimeChecks(
+                    MemCheckBlock->getTerminator(), *DiffChecks, MemCheckExp,
+                    [VF](IRBuilderBase &B, unsigned Bits) {
----------------
david-arm wrote:

Then it gets worse with this code:

```
    // Check if the same compare has already been created earlier. In that case,
    // there is no need to check it again.
    Value *IsConflict = SeenCompares.lookup({Diff, VFTimesUFTimesSize});
```

because even though `SrcStart, SinkStart, AccessSize, NeedsFreeze` may all be different we still perform the lookup based on the first runtime VF, potentially getting the value `IsConflict` wrong.

https://github.com/llvm/llvm-project/pull/130118