diff --git a/ggml/include/ggml-cpp.h b/ggml/include/ggml-cpp.h
index a12342c25debe..48aa79682b65d 100644
--- a/ggml/include/ggml-cpp.h
+++ b/ggml/include/ggml-cpp.h
@@ -24,7 +24,7 @@ typedef std::unique_ptr<gguf_context, gguf_context_deleter> gguf_context_ptr;
 
 struct ggml_gallocr_deleter { void operator()(ggml_gallocr_t galloc) { ggml_gallocr_free(galloc); } };
 
-typedef std::unique_ptr<ggml_gallocr_t, ggml_gallocr_deleter> ggml_gallocr_ptr;
+typedef std::unique_ptr<ggml_gallocr, ggml_gallocr_deleter> ggml_gallocr_ptr;
 
 // ggml-backend
 
diff --git a/ggml/src/ggml-cuda/cpy.cu b/ggml/src/ggml-cuda/cpy.cu
index 2d46176eab344..d027271fcd932 100644
--- a/ggml/src/ggml-cuda/cpy.cu
+++ b/ggml/src/ggml-cuda/cpy.cu
@@ -592,6 +592,8 @@ void ggml_cuda_cpy(ggml_backend_cuda_context & ctx, const ggml_tensor * src0, gg
         dest_ptrs_d = ctx.cuda_graph->dest_ptrs_d;
         graph_cpynode_index = ctx.cuda_graph->graph_cpynode_index;
     }
+#else
+    GGML_UNUSED(disable_indirection_for_this_node);
 #endif
     if (src0->type == src1->type && ggml_is_contiguous(src0) && ggml_is_contiguous(src1)) {
         GGML_ASSERT(ggml_nbytes(src0) == ggml_nbytes(src1));