diff --git a/flang/lib/Lower/ConvertCall.cpp b/flang/lib/Lower/ConvertCall.cpp
--- a/flang/lib/Lower/ConvertCall.cpp
+++ b/flang/lib/Lower/ConvertCall.cpp
@@ -396,9 +396,7 @@
           component->GetFirstSymbol(), &symMap);
       mlir::Value passObject = fir::getBase(pass);
       if (fir::isa_ref_type(passObject.getType()))
-        passObject = builder.create<fir::ConvertOp>(
-            loc, passObject.getType().dyn_cast<fir::ReferenceType>().getEleTy(),
-            passObject);
+        passObject = builder.create<fir::LoadOp>(loc, passObject);
       dispatch = builder.create<fir::DispatchOp>(
           loc, funcType.getResults(), builder.getStringAttr(procName),
           passObject, operands, nullptr);
diff --git a/flang/test/Lower/allocatable-polymorphic.f90 b/flang/test/Lower/allocatable-polymorphic.f90
--- a/flang/test/Lower/allocatable-polymorphic.f90
+++ b/flang/test/Lower/allocatable-polymorphic.f90
@@ -103,8 +103,8 @@
 ! CHECK: %{{.*}} = fir.call @_FortranAPointerAllocate(%[[P_DESC_CAST]], %{{.*}}, %{{.*}}, %{{.*}}, %{{.*}}) {{.*}}: (!fir.ref<!fir.box<none>>, i1, !fir.box<none>, !fir.ref<i8>, i32) -> i32
 
 ! call p%proc1()
-! CHECK: %[[P_CAST:.*]] = fir.convert %[[P_DESC:.*]] : (!fir.ref<!fir.class<!fir.ptr<!fir.type<_QMpolyTp1{a:i32,b:i32}>>>>) -> !fir.class<!fir.ptr<!fir.type<_QMpolyTp1{a:i32,b:i32}>>>
-! CHECK: fir.dispatch "proc1"(%[[P_CAST]] : !fir.class<!fir.ptr<!fir.type<_QMpolyTp1{a:i32,b:i32}>>>)
+! CHECK: %[[P_LOAD:.*]] = fir.load %[[P_DESC:.*]] : !fir.ref<!fir.class<!fir.ptr<!fir.type<_QMpolyTp1{a:i32,b:i32}>>>>
+! CHECK: fir.dispatch "proc1"(%[[P_LOAD]] : !fir.class<!fir.ptr<!fir.type<_QMpolyTp1{a:i32,b:i32}>>>)
 
 ! CHECK: %[[TYPE_DESC_P1:.*]] = fir.type_desc !fir.type<_QMpolyTp1{a:i32,b:i32}>
 ! CHECK: %[[C1_DESC_CAST:.*]] = fir.convert %[[C1_DESC:.*]] : (!fir.ref<!fir.class<!fir.ptr<!fir.type<_QMpolyTp1{a:i32,b:i32}>>>>) -> !fir.ref<!fir.box<none>>
@@ -125,12 +125,12 @@
 ! CHECK: %{{.*}} = fir.call @_FortranAPointerAllocate(%[[C2_DESC_CAST]], %{{.*}}, %{{.*}}, %{{.*}}, %{{.*}}) {{.*}}: (!fir.ref<!fir.box<none>>, i1, !fir.box<none>, !fir.ref<i8>, i32) -> i32
 
 ! call c1%proc1()
-! CHECK: %[[C1_DESC_CAST:.*]] = fir.convert %[[C1_DESC]] : (!fir.ref<!fir.class<!fir.ptr<!fir.type<_QMpolyTp1{a:i32,b:i32}>>>>) -> !fir.class<!fir.ptr<!fir.type<_QMpolyTp1{a:i32,b:i32}>>>
-! CHECK: fir.dispatch "proc1"(%[[C1_DESC_CAST]] : !fir.class<!fir.ptr<!fir.type<_QMpolyTp1{a:i32,b:i32}>>>)
+! CHECK: %[[C1_DESC_LOAD:.*]] = fir.load %[[C1_DESC]] : !fir.ref<!fir.class<!fir.ptr<!fir.type<_QMpolyTp1{a:i32,b:i32}>>>>
+! CHECK: fir.dispatch "proc1"(%[[C1_DESC_LOAD]] : !fir.class<!fir.ptr<!fir.type<_QMpolyTp1{a:i32,b:i32}>>>)
 
 ! call c2%proc1()
-! CHECK: %[[C2_DESC_CAST:.*]] = fir.convert %[[C2_DESC]] : (!fir.ref<!fir.class<!fir.ptr<!fir.type<_QMpolyTp1{a:i32,b:i32}>>>>) -> !fir.class<!fir.ptr<!fir.type<_QMpolyTp1{a:i32,b:i32}>>>
-! CHECK: fir.dispatch "proc1"(%[[C2_DESC_CAST]] : !fir.class<!fir.ptr<!fir.type<_QMpolyTp1{a:i32,b:i32}>>>)
+! CHECK: %[[C2_DESC_LOAD:.*]] = fir.load %[[C2_DESC]] : !fir.ref<!fir.class<!fir.ptr<!fir.type<_QMpolyTp1{a:i32,b:i32}>>>>
+! CHECK: fir.dispatch "proc1"(%[[C2_DESC_LOAD]] : !fir.class<!fir.ptr<!fir.type<_QMpolyTp1{a:i32,b:i32}>>>)
 
 ! call c1%proc2()
 ! CHECK: %[[C1_LOAD:.*]] = fir.load %[[C1_DESC]] : !fir.ref<!fir.class<!fir.ptr<!fir.type<_QMpolyTp1{a:i32,b:i32}>>>>
@@ -311,18 +311,18 @@
 ! CHECK: %[[C4_CAST:.*]] = fir.convert %[[C4]] : (!fir.ref<!fir.class<!fir.heap<!fir.array<?x!fir.type<_QMpolyTp1{a:i32,b:i32}>>>>>) -> !fir.ref<!fir.box<none>>
 ! CHECK: %{{.*}} = fir.call @_FortranAAllocatableAllocate(%[[C4_CAST]], %{{.*}}, %{{.*}}, %{{.*}}, %{{.*}}) {{.*}}: (!fir.ref<!fir.box<none>>, i1, !fir.box<none>, !fir.ref<i8>, i32) -> i32
 
-! CHECK: %[[C1_CAST:.*]] = fir.convert %[[C1_DESC]] : (!fir.ref<!fir.class<!fir.heap<!fir.type<_QMpolyTp1{a:i32,b:i32}>>>>) -> !fir.class<!fir.heap<!fir.type<_QMpolyTp1{a:i32,b:i32}>>>
-! CHECK: fir.dispatch "proc1"(%[[C1_CAST]] : !fir.class<!fir.heap<!fir.type<_QMpolyTp1{a:i32,b:i32}>>>)
+! CHECK: %[[C1_LOAD1:.*]] = fir.load %[[C1_DESC]] : !fir.ref<!fir.class<!fir.heap<!fir.type<_QMpolyTp1{a:i32,b:i32}>>>>
+! CHECK: fir.dispatch "proc1"(%[[C1_LOAD1]] : !fir.class<!fir.heap<!fir.type<_QMpolyTp1{a:i32,b:i32}>>>)
 
-! CHECK: %[[C2_CAST:.*]] = fir.convert %[[C2_DESC]] : (!fir.ref<!fir.class<!fir.heap<!fir.type<_QMpolyTp1{a:i32,b:i32}>>>>) -> !fir.class<!fir.heap<!fir.type<_QMpolyTp1{a:i32,b:i32}>>>
-! CHECK: fir.dispatch "proc1"(%[[C2_CAST]] : !fir.class<!fir.heap<!fir.type<_QMpolyTp1{a:i32,b:i32}>>>)
+! CHECK: %[[C2_LOAD1:.*]] = fir.load %[[C2_DESC]] : !fir.ref<!fir.class<!fir.heap<!fir.type<_QMpolyTp1{a:i32,b:i32}>>>>
+! CHECK: fir.dispatch "proc1"(%[[C2_LOAD1]] : !fir.class<!fir.heap<!fir.type<_QMpolyTp1{a:i32,b:i32}>>>)
 
-! CHECK: %[[C1_LOAD:.*]] = fir.load %[[C1_DESC]] : !fir.ref<!fir.class<!fir.heap<!fir.type<_QMpolyTp1{a:i32,b:i32}>>>>
-! CHECK: %[[C1_REBOX:.*]] = fir.rebox %[[C1_LOAD]] : (!fir.class<!fir.heap<!fir.type<_QMpolyTp1{a:i32,b:i32}>>>) -> !fir.class<!fir.type<_QMpolyTp1{a:i32,b:i32}>>
+! CHECK: %[[C1_LOAD2:.*]] = fir.load %[[C1_DESC]] : !fir.ref<!fir.class<!fir.heap<!fir.type<_QMpolyTp1{a:i32,b:i32}>>>>
+! CHECK: %[[C1_REBOX:.*]] = fir.rebox %[[C1_LOAD2]] : (!fir.class<!fir.heap<!fir.type<_QMpolyTp1{a:i32,b:i32}>>>) -> !fir.class<!fir.type<_QMpolyTp1{a:i32,b:i32}>>
 ! CHECK: fir.dispatch "proc2"(%[[C1_REBOX]] : !fir.class<!fir.type<_QMpolyTp1{a:i32,b:i32}>>) (%[[C1_REBOX]] : !fir.class<!fir.type<_QMpolyTp1{a:i32,b:i32}>>) {pass_arg_pos = 0 : i32}
 
-! CHECK: %[[C2_LOAD:.*]] = fir.load %[[C2_DESC]] : !fir.ref<!fir.class<!fir.heap<!fir.type<_QMpolyTp1{a:i32,b:i32}>>>>
-! CHECK: %[[C2_REBOX:.*]] = fir.rebox %[[C2_LOAD]] : (!fir.class<!fir.heap<!fir.type<_QMpolyTp1{a:i32,b:i32}>>>) -> !fir.class<!fir.type<_QMpolyTp1{a:i32,b:i32}>>
+! CHECK: %[[C2_LOAD2:.*]] = fir.load %[[C2_DESC]] : !fir.ref<!fir.class<!fir.heap<!fir.type<_QMpolyTp1{a:i32,b:i32}>>>>
+! CHECK: %[[C2_REBOX:.*]] = fir.rebox %[[C2_LOAD2]] : (!fir.class<!fir.heap<!fir.type<_QMpolyTp1{a:i32,b:i32}>>>) -> !fir.class<!fir.type<_QMpolyTp1{a:i32,b:i32}>>
 ! CHECK: fir.dispatch "proc2"(%[[C2_REBOX]] : !fir.class<!fir.type<_QMpolyTp1{a:i32,b:i32}>>) (%[[C2_REBOX]] : !fir.class<!fir.type<_QMpolyTp1{a:i32,b:i32}>>) {pass_arg_pos = 0 : i32}
 
 ! CHECK-LABEL: %{{.*}} = fir.do_loop