diff --git a/arm/main.cc b/arm/main.cc
index 815073d..ad2b54f 100644
--- a/arm/main.cc
+++ b/arm/main.cc
@@ -14,7 +14,8 @@ namespace {
 using async::AwaitableType;
 
 void Uart0Isr() {
-    tracing::trace(tracing::TraceEvent::kUartIsr);
+    ToggleLed(7);
+    //tracing::trace(tracing::TraceEvent::kUartIsr);
     HandleUartIsr();
 }
 
@@ -36,28 +37,73 @@ void SetupTimer() {
     NVIC_EnableIRQ(Timer0_IRQn);
 }
 
+
+#if 0 // https://gcc.gnu.org/bugzilla/show_bug.cgi?id=101133
 async::task<> echo() {
+    async::task<uint8_t> reader = UartReadLoop();
+    async::gimme<std::span<const std::byte>> feeder;
+    async::task<> writer = UartWriteLoop(feeder);
+    writer.h.resume();  // advance to first yield
     while (1) {
-        buffer buff = co_await UartRead(1);
-        co_await UartWrite(buff.data);
+        SetLed(1);
+        uint8_t c = co_await reader;
+        ClearLed(1);
+        ToggleLed(2);
+        co_await feeder.feed(std::as_bytes(std::span{&c, 1}));
+    }
+}
+#endif
+
+async::task<> echo() {
+    async::task<uint8_t> reader = UartReadLoop();
+    while (1) {
+        SetLed(1);
+        uint8_t c = co_await reader;
+        ClearLed(1);
+        ToggleLed(2);
+        co_await UartWrite(std::as_bytes(std::span{&c, 1}));
+    }
+}
+
+async::task<> dump_trace() {
+    while (1) {
+        co_await async::delay(5000);
+//        tracing::dump();
     }
 }
 }  // namespace
 
+#define XUL_SR_RX_FIFO_FULL 0x02       /* receive FIFO full */
+#define XUL_SR_RX_FIFO_VALID_DATA 0x01 /* data in receive FIFO */
+
 int main() {
     _vector_table[16 + HardFault_IRQn] =
         reinterpret_cast<uint32_t>(crash::HardFaultHandler);
 
     SetupUart();
+    UartWriteBlocking("uart setup done\r\n");
     SetupTimer();
+    UartWriteBlocking("timer setup done\r\n");
 
     async::schedule(echo().h);
+    async::schedule(dump_trace().h);
+
+    tracing::trace(tracing::TraceEvent::kUartIsr);
+    tracing::trace(tracing::TraceEvent::kUartIsr);
+
+    UartWriteBlocking("init done. starting main loop\r\n");
 
     async::main_loop([]() {
         static int cnt = 0;
         timer0->Pet();
         if ((cnt++ % 100000) == 0) {
             ToggleLed(0);
+            ClearLed(7);
+            ClearLed(2);
+            ClearLed(3);
+            ClearLed(4);
+
+            LogStuff();
         }
         return false;
     });
diff --git a/arm/prog.py b/arm/prog.py
index b86d6d5..d15370d 100644
--- a/arm/prog.py
+++ b/arm/prog.py
@@ -42,6 +42,12 @@ def main():
     write(s, offset, dat)
     jump(s, offset)
 
+    while True:
+        dat = s.read()
+        if not dat:
+            break
+        sys.stdout.buffer.write(dat)
+
 
 if __name__ == "__main__":
     main()