Index: llvm/lib/Analysis/Loads.cpp
===================================================================
--- llvm/lib/Analysis/Loads.cpp
+++ llvm/lib/Analysis/Loads.cpp
@@ -498,12 +498,15 @@
     if (!AreEquivalentAddressValues(StorePtr, Ptr))
       return nullptr;
 
+    if (IsLoadCSE)
+      *IsLoadCSE = false;
+
     Value *Val = SI->getValueOperand();
-    if (CastInst::isBitOrNoopPointerCastable(Val->getType(), AccessTy, DL)) {
-      if (IsLoadCSE)
-        *IsLoadCSE = false;
+    if (CastInst::isBitOrNoopPointerCastable(Val->getType(), AccessTy, DL))
       return Val;
-    }
+
+    if (auto *C = dyn_cast<Constant>(Val))
+      return ConstantFoldLoadThroughBitcast(C, AccessTy, DL);
   }
 
   return nullptr;
Index: llvm/test/Transforms/InstCombine/load-store-forward.ll
===================================================================
--- llvm/test/Transforms/InstCombine/load-store-forward.ll
+++ llvm/test/Transforms/InstCombine/load-store-forward.ll
@@ -22,8 +22,7 @@
 ; CHECK-LABEL: @vec_store_load_first(
 ; CHECK-NEXT:    [[P2:%.*]] = bitcast i32* [[P:%.*]] to <2 x i32>*
 ; CHECK-NEXT:    store <2 x i32> <i32 1, i32 2>, <2 x i32>* [[P2]], align 8
-; CHECK-NEXT:    [[LOAD:%.*]] = load i32, i32* [[P]], align 4
-; CHECK-NEXT:    ret i32 [[LOAD]]
+; CHECK-NEXT:    ret i32 1
 ;
   %p2 = bitcast i32* %p to <2 x i32>*
   store <2 x i32> <i32 1, i32 2>, <2 x i32>* %p2