diff --git a/vhost-user-backend/src/backend.rs b/vhost-user-backend/src/backend.rs
index 07a3de9..9121b2e 100644
--- a/vhost-user-backend/src/backend.rs
+++ b/vhost-user-backend/src/backend.rs
@@ -89,9 +89,14 @@ pub trait VhostUserBackend: Send + Sync {
     /// Set handler for communicating with the frontend by the gpu specific backend communication
     /// channel.
     ///
-    /// This method only exits when the crate feature gpu-socket is enabled, because this is only
-    /// useful for a gpu device.
-    fn set_gpu_socket(&self, _gpu_backend: GpuBackend);
+    /// This function returns a `Result`, returning an error if the backend does not implement this
+    /// function.
+    fn set_gpu_socket(&self, _gpu_backend: GpuBackend) -> Result<()> {
+        Err(std::io::Error::new(
+            std::io::ErrorKind::Unsupported,
+            "backend does not support set_gpu_socket() / VHOST_USER_GPU_SET_SOCKET",
+        ))
+    }
 
     /// Get the map to map queue index to worker thread index.
     ///
@@ -206,9 +211,14 @@ pub trait VhostUserBackendMut: Send + Sync {
     /// Set handler for communicating with the frontend by the gpu specific backend communication
     /// channel.
     ///
-    /// This method only exits when the crate feature gpu-socket is enabled, because this is only
-    /// useful for a gpu device.
-    fn set_gpu_socket(&mut self, gpu_backend: GpuBackend);
+    /// This function returns a `Result`, returning an error if the backend does not implement this
+    /// function.
+    fn set_gpu_socket(&mut self, _gpu_backend: GpuBackend) -> Result<()> {
+        Err(std::io::Error::new(
+            std::io::ErrorKind::Unsupported,
+            "backend does not support set_gpu_socket() / VHOST_USER_GPU_SET_SOCKET",
+        ))
+    }
 
     /// Get the map to map queue index to worker thread index.
     ///
@@ -315,7 +325,7 @@ impl<T: VhostUserBackend> VhostUserBackend for Arc<T> {
         self.deref().set_backend_req_fd(backend)
     }
 
-    fn set_gpu_socket(&self, gpu_backend: GpuBackend) {
+    fn set_gpu_socket(&self, gpu_backend: GpuBackend) -> Result<()> {
         self.deref().set_gpu_socket(gpu_backend)
     }
 
@@ -396,7 +406,7 @@ impl<T: VhostUserBackendMut> VhostUserBackend for Mutex<T> {
         self.lock().unwrap().set_backend_req_fd(backend)
     }
 
-    fn set_gpu_socket(&self, gpu_backend: GpuBackend) {
+    fn set_gpu_socket(&self, gpu_backend: GpuBackend) -> Result<()> {
         self.lock().unwrap().set_gpu_socket(gpu_backend)
     }
 
@@ -480,7 +490,7 @@ impl<T: VhostUserBackendMut> VhostUserBackend for RwLock<T> {
         self.write().unwrap().set_backend_req_fd(backend)
     }
 
-    fn set_gpu_socket(&self, gpu_backend: GpuBackend) {
+    fn set_gpu_socket(&self, gpu_backend: GpuBackend) -> Result<()> {
         self.write().unwrap().set_gpu_socket(gpu_backend)
     }
 
@@ -604,8 +614,6 @@ pub mod tests {
 
         fn set_backend_req_fd(&mut self, _backend: Backend) {}
 
-        fn set_gpu_socket(&mut self, _gpu_backend: GpuBackend) {}
-
         fn queues_per_thread(&self) -> Vec<u64> {
             vec![1, 1]
         }
diff --git a/vhost-user-backend/src/handler.rs b/vhost-user-backend/src/handler.rs
index 54fbc47..3986731 100644
--- a/vhost-user-backend/src/handler.rs
+++ b/vhost-user-backend/src/handler.rs
@@ -555,8 +555,10 @@ where
         self.backend.set_backend_req_fd(backend);
     }
 
-    fn set_gpu_socket(&mut self, gpu_backend: GpuBackend) {
-        self.backend.set_gpu_socket(gpu_backend);
+    fn set_gpu_socket(&mut self, gpu_backend: GpuBackend) -> VhostUserResult<()> {
+        self.backend
+            .set_gpu_socket(gpu_backend)
+            .map_err(VhostUserError::ReqHandlerError)
     }
 
     fn get_inflight_fd(
diff --git a/vhost-user-backend/tests/vhost-user-server.rs b/vhost-user-backend/tests/vhost-user-server.rs
index 213748b..7f78e2a 100644
--- a/vhost-user-backend/tests/vhost-user-server.rs
+++ b/vhost-user-backend/tests/vhost-user-server.rs
@@ -10,7 +10,6 @@ use std::thread;
 use vhost::vhost_user::message::{
     VhostUserConfigFlags, VhostUserHeaderFlag, VhostUserInflight, VhostUserProtocolFeatures,
 };
-use vhost::vhost_user::GpuBackend;
 use vhost::vhost_user::{Backend, Frontend, Listener, VhostUserFrontend};
 use vhost::{VhostBackend, VhostUserMemoryRegionInfo, VringConfigData};
 use vhost_user_backend::{VhostUserBackendMut, VhostUserDaemon, VringRwLock};
@@ -78,8 +77,6 @@ impl VhostUserBackendMut for MockVhostBackend {
         Ok(())
     }
 
-    fn set_gpu_socket(&mut self, _gpu_backend: GpuBackend) {}
-
     fn update_memory(&mut self, atomic_mem: GuestMemoryAtomic<GuestMemoryMmap>) -> Result<()> {
         let mem = atomic_mem.memory();
         let region = mem.find_region(GuestAddress(0x100000)).unwrap();
diff --git a/vhost/src/vhost_user/backend_req_handler.rs b/vhost/src/vhost_user/backend_req_handler.rs
index ec0c9ac..22e893b 100644
--- a/vhost/src/vhost_user/backend_req_handler.rs
+++ b/vhost/src/vhost_user/backend_req_handler.rs
@@ -66,7 +66,7 @@ pub trait VhostUserBackendReqHandler {
     fn get_config(&self, offset: u32, size: u32, flags: VhostUserConfigFlags) -> Result<Vec<u8>>;
     fn set_config(&self, offset: u32, buf: &[u8], flags: VhostUserConfigFlags) -> Result<()>;
     fn set_backend_req_fd(&self, _backend: Backend) {}
-    fn set_gpu_socket(&self, gpu_backend: GpuBackend);
+    fn set_gpu_socket(&self, _gpu_backend: GpuBackend) -> Result<()>;
     fn get_inflight_fd(&self, inflight: &VhostUserInflight) -> Result<(VhostUserInflight, File)>;
     fn set_inflight_fd(&self, inflight: &VhostUserInflight, file: File) -> Result<()>;
     fn get_max_mem_slots(&self) -> Result<u64>;
@@ -126,7 +126,7 @@ pub trait VhostUserBackendReqHandlerMut {
     ) -> Result<Vec<u8>>;
     fn set_config(&mut self, offset: u32, buf: &[u8], flags: VhostUserConfigFlags) -> Result<()>;
     fn set_backend_req_fd(&mut self, _backend: Backend) {}
-    fn set_gpu_socket(&mut self, _gpu_backend: GpuBackend);
+    fn set_gpu_socket(&mut self, _gpu_backend: GpuBackend) -> Result<()>;
     fn get_inflight_fd(
         &mut self,
         inflight: &VhostUserInflight,
@@ -238,8 +238,8 @@ impl<T: VhostUserBackendReqHandlerMut> VhostUserBackendReqHandler for Mutex<T> {
         self.lock().unwrap().set_backend_req_fd(backend)
     }
 
-    fn set_gpu_socket(&self, gpu_backend: GpuBackend) {
-        self.lock().unwrap().set_gpu_socket(gpu_backend);
+    fn set_gpu_socket(&self, gpu_backend: GpuBackend) -> Result<()> {
+        self.lock().unwrap().set_gpu_socket(gpu_backend)
     }
 
     fn get_inflight_fd(&self, inflight: &VhostUserInflight) -> Result<(VhostUserInflight, File)> {
@@ -814,8 +814,7 @@ impl<S: VhostUserBackendReqHandler> BackendReqHandler<S> {
         // since we have no way to check this. If not, it will fail later.
         let sock = unsafe { UnixStream::from_raw_fd(file.into_raw_fd()) };
         let gpu_backend = GpuBackend::from_stream(sock);
-        self.backend.set_gpu_socket(gpu_backend);
-        Ok(())
+        self.backend.set_gpu_socket(gpu_backend)
     }
 
     fn handle_vring_fd_request(
diff --git a/vhost/src/vhost_user/dummy_backend.rs b/vhost/src/vhost_user/dummy_backend.rs
index e671f19..b83279b 100644
--- a/vhost/src/vhost_user/dummy_backend.rs
+++ b/vhost/src/vhost_user/dummy_backend.rs
@@ -259,7 +259,9 @@ impl VhostUserBackendReqHandlerMut for DummyBackendReqHandler {
         Ok(())
     }
 
-    fn set_gpu_socket(&mut self, _gpu_backend: GpuBackend) {}
+    fn set_gpu_socket(&mut self, _gpu_backend: GpuBackend) -> Result<()> {
+        Ok(())
+    }
 
     fn get_inflight_fd(
         &mut self,