diff --git a/llm/ext_server/ext_server.cpp b/llm/ext_server/ext_server.cpp
index d0a036b3..c30d4220 100644
--- a/llm/ext_server/ext_server.cpp
+++ b/llm/ext_server/ext_server.cpp
@@ -111,6 +111,10 @@ void llama_server_stop() {
   // TODO - too verbose, remove once things are solid
   LOG_TEE("requesting llama server shutdown\n");
   ext_server_running = false;
+
+  // unblocks the update_slots() loop so it can clean up and exit
+  llama->request_cancel(0);
+
   ext_server_thread.join();
   delete llama;
   llama = NULL;
diff --git a/llm/llama.cpp b/llm/llama.cpp
index 328b83de..6efb8eb3 160000
--- a/llm/llama.cpp
+++ b/llm/llama.cpp
@@ -1 +1 @@
-Subproject commit 328b83de23b33240e28f4e74900d1d06726f5eb1
+Subproject commit 6efb8eb30e7025b168f3fda3ff83b9b386428ad6