LAA: strip unnecessary getUniqueCastUse (#92119)

733b8b2 ([LAA] Simplify identification of speculatable strides [nfc])
refactored getStrideFromPointer() to compute directly on SCEVs, and
return an SCEV expression instead of a Value. However, it left behind a
call to getUniqueCastUse(), which is completely unnecessary. Remove
this, showing a positive test update, and simplify the surrounding
program logic.

Author: artagnon

llvm/lib/Analysis/LoopAccessAnalysis.cpp

@@ -2848,21 +2848,6 @@ static Value *stripGetElementPtr(Value *Ptr, ScalarEvolution *SE, Loop *Lp) {
   return GEP->getOperand(InductionOperand);
 }
 
-/// If a value has only one user that is a CastInst, return it.
-static Value *getUniqueCastUse(Value *Ptr, Loop *Lp, Type *Ty) {
-  Value *UniqueCast = nullptr;
-  for (User *U : Ptr->users()) {
-    CastInst *CI = dyn_cast<CastInst>(U);
-    if (CI && CI->getType() == Ty) {
-      if (!UniqueCast)
-        UniqueCast = CI;
-      else
-        return nullptr;
-    }
-  }
-  return UniqueCast;
-}
-
 /// Get the stride of a pointer access in a loop. Looks for symbolic
 /// strides "a[i*stride]". Returns the symbolic stride, or null otherwise.
 static const SCEV *getStrideFromPointer(Value *Ptr, ScalarEvolution *SE, Loop *Lp) {
@@ -2925,21 +2910,14 @@ static const SCEV *getStrideFromPointer(Value *Ptr, ScalarEvolution *SE, Loop *L
     return nullptr;
 
   // Look for the loop invariant symbolic value.
-  const SCEVUnknown *U = dyn_cast<SCEVUnknown>(V);
-  if (!U) {
-    const auto *C = dyn_cast<SCEVIntegralCastExpr>(V);
-    if (!C)
-      return nullptr;
-    U = dyn_cast<SCEVUnknown>(C->getOperand());
-    if (!U)
-      return nullptr;
+  if (isa<SCEVUnknown>(V))
+    return V;
 
-    // Match legacy behavior - this is not needed for correctness
-    if (!getUniqueCastUse(U->getValue(), Lp, V->getType()))
-      return nullptr;
-  }
+  if (const auto *C = dyn_cast<SCEVIntegralCastExpr>(V))
+    if (isa<SCEVUnknown>(C->getOperand()))
+      return V;
 
-  return V;
+  return nullptr;
 }
 
 void LoopAccessInfo::collectStridedAccess(Value *MemAccess) {

llvm/test/Analysis/LoopAccessAnalysis/symbolic-stride.ll

@@ -170,16 +170,23 @@ define void @single_stride_castexpr_multiuse(i32 %offset, ptr %src, ptr %dst, i1
 ; CHECK-NEXT:          %gep.src = getelementptr inbounds i32, ptr %src, i64 %iv.3
 ; CHECK-NEXT:      Grouped accesses:
 ; CHECK-NEXT:        Group [[GRP3]]:
-; CHECK-NEXT:          (Low: (((4 * %iv.1) + %dst) umin ((4 * %iv.1) + (4 * (sext i32 %offset to i64) * (200 + (-1 * (zext i32 %offset to i64))<nsw>)<nsw>) + %dst)) High: (4 + (((4 * %iv.1) + %dst) umax ((4 * %iv.1) + (4 * (sext i32 %offset to i64) * (200 + (-1 * (zext i32 %offset to i64))<nsw>)<nsw>) + %dst))))
-; CHECK-NEXT:            Member: {((4 * %iv.1) + %dst),+,(4 * (sext i32 %offset to i64))<nsw>}<%inner.loop>
+; CHECK-NEXT:          (Low: ((4 * %iv.1) + %dst) High: (804 + (4 * %iv.1) + (-4 * (zext i32 %offset to i64))<nsw> + %dst))
+; CHECK-NEXT:            Member: {((4 * %iv.1) + %dst),+,4}<%inner.loop>
 ; CHECK-NEXT:        Group [[GRP4]]:
-; CHECK-NEXT:          (Low: ((4 * (zext i32 %offset to i64))<nuw><nsw> + %src) High: (804 + %src))
-; CHECK-NEXT:            Member: {((4 * (zext i32 %offset to i64))<nuw><nsw> + %src),+,4}<%inner.loop>
+; CHECK-NEXT:          (Low: (4 + %src) High: (808 + (-4 * (zext i32 %offset to i64))<nsw> + %src))
+; CHECK-NEXT:            Member: {(4 + %src),+,4}<%inner.loop>
 ; CHECK-EMPTY:
 ; CHECK-NEXT:      Non vectorizable stores to invariant address were not found in loop.
 ; CHECK-NEXT:      SCEV assumptions:
+; CHECK-NEXT:      Equal predicate: %offset == 1
 ; CHECK-EMPTY:
 ; CHECK-NEXT:      Expressions re-written:
+; CHECK-NEXT:      [PSE] %gep.src = getelementptr inbounds i32, ptr %src, i64 %iv.3:
+; CHECK-NEXT:        {((4 * (zext i32 %offset to i64))<nuw><nsw> + %src),+,4}<%inner.loop>
+; CHECK-NEXT:        --> {(4 + %src),+,4}<%inner.loop>
+; CHECK-NEXT:      [PSE] %gep.dst = getelementptr i32, ptr %dst, i64 %iv.2:
+; CHECK-NEXT:        {((4 * %iv.1) + %dst),+,(4 * (sext i32 %offset to i64))<nsw>}<%inner.loop>
+; CHECK-NEXT:        --> {((4 * %iv.1) + %dst),+,4}<%inner.loop>
 ; CHECK-NEXT:    outer.header:
 ; CHECK-NEXT:      Report: loop is not the innermost loop
 ; CHECK-NEXT:      Dependences: