[llvm] LAA: be less conservative in isNoWrap (PR #112553)

[Florian Hahn via llvm-commits]

================
@@ -825,14 +825,13 @@ static bool hasComputableBounds(PredicatedScalarEvolution &PSE, Value *Ptr,
 
 /// Check whether a pointer address cannot wrap.
 static bool isNoWrap(PredicatedScalarEvolution &PSE,
-                     const DenseMap<Value *, const SCEV *> &Strides, Value *Ptr, Type *AccessTy,
-                     Loop *L) {
+                     const DenseMap<Value *, const SCEV *> &Strides, Value *Ptr,
+                     Type *AccessTy, Loop *L, bool Assume) {
   const SCEV *PtrScev = PSE.getSCEV(Ptr);
   if (PSE.getSE()->isLoopInvariant(PtrScev, L))
     return true;
 
-  int64_t Stride = getPtrStride(PSE, AccessTy, Ptr, L, Strides).value_or(0);
-  return Stride == 1 ||
+  return getPtrStride(PSE, AccessTy, Ptr, L, Strides, Assume).has_value() ||
----------------
fhahn wrote:

> getPtrStride returns Strides == 1 or -1, except when isNoWrapAddRec or Assume are true; we can include the case of -1 Stride, and when isNoWrapAddRec is true. 

I am not sure this is true for all cases;  I think unless  `ShouldCheckWrap` is true, we may return strides > 1 or < -1 even if they may wrap; we should probably pass  `ShouldCheckWrap = true` (for most parts, checks in other places should catch issues, but better be safe here)

https://github.com/llvm/llvm-project/pull/112553