llm, mlxrunner: fix done channel value consumed by first receiver

Receiving from a buffered chan error consumes the value, so only the
first caller (WaitUntilRunning, HasExited, or Close) sees the signal.
Subsequent receivers block or take the wrong branch. Replace with a
closed chan struct{} which can be received from any number of times,
and store the error in a separate field.

llm/server.go

@@ -87,7 +87,8 @@ type LlamaServer interface {
 type llmServer struct {
 	port      int
 	cmd       *exec.Cmd
-	done      chan error // Channel to signal when the process exits
+	done      chan struct{} // closed when the process exits
+	doneErr   error         // valid after done is closed
 	status    *StatusWriter
 	options   api.Options
 	modelPath string
@@ -280,7 +281,7 @@ func NewLlamaServer(systemInfo ml.SystemInfo, gpus []ml.DeviceInfo, modelPath st
 		sem:            semaphore.NewWeighted(int64(numParallel)),
 		totalLayers:    f.KV().BlockCount() + 1,
 		loadStart:      time.Now(),
-		done:           make(chan error, 1),
+		done:           make(chan struct{}),
 	}
 
 	if err != nil {
@@ -304,10 +305,11 @@ func NewLlamaServer(systemInfo ml.SystemInfo, gpus []ml.DeviceInfo, modelPath st
 			if strings.Contains(s.status.LastErrMsg, "unknown model") {
 				s.status.LastErrMsg = "this model is not supported by your version of Ollama. You may need to upgrade"
 			}
-			s.done <- errors.New(s.status.LastErrMsg)
+			s.doneErr = errors.New(s.status.LastErrMsg)
 		} else {
-			s.done <- err
+			s.doneErr = err
 		}
+		close(s.done)
 	}()
 
 	if tok != nil {
@@ -1356,8 +1358,8 @@ func (s *llmServer) WaitUntilRunning(ctx context.Context) error {
 		case <-ctx.Done():
 			slog.Warn("client connection closed before server finished loading, aborting load")
 			return fmt.Errorf("timed out waiting for llama runner to start: %w", ctx.Err())
-		case err := <-s.done:
-			return fmt.Errorf("llama runner process has terminated: %w", err)
+		case <-s.done:
+			return fmt.Errorf("llama runner process has terminated: %w", s.doneErr)
 		default:
 		}
 		if time.Now().After(stallTimer) {