shared/tinyusb: Only run TinyUSB on the main thread if GIL is disabled.

If GIL is disabled then there's threat of a race condition if some other
code specifically requests USB processing (i.e. to unblock stdio), while
a scheduled TinyUSB callback is already running on another thread.

Relies on the change in the parent commit, where scheduler is restricted
to main thread if GIL is disabled.

Fixes #15390 - "TinyUSB callback can't recurse" exceptions on rp2 when
using _thread module and USB serial I/O.

Adds a unit test for stdin functioning correctly in threads (fails on rp2
port without this fix).

This work was funded through GitHub Sponsors.

Signed-off-by: Angus Gratton <angus@redyak.com.au>

shared/tinyusb/mp_usbd_runtime.c

@@ -501,6 +501,15 @@ void mp_usbd_task_callback(mp_sched_node_t *node) {
 // Task function can be called manually to force processing of USB events
 // (mostly from USB-CDC serial port when blocking.)
 void mp_usbd_task(void) {
+    #if MICROPY_PY_THREAD && !MICROPY_PY_THREAD_GIL
+    if (!mp_thread_is_main_thread()) {
+        // Avoid race with the scheduler callback by scheduling TinyUSB to run
+        // on the main thread.
+        mp_usbd_schedule_task();
+        return;
+    }
+    #endif
+
     if (in_usbd_task) {
         // If this exception triggers, it means a USB callback tried to do
         // something that itself became blocked on TinyUSB (most likely: read or

tests/thread/thread_stdin.py

@@ -0,0 +1,44 @@
+# Test that having multiple threads block on stdin doesn't cause any issues.
+#
+# The test doesn't expect any input on stdin.
+#
+# This is a regression test for https://github.com/micropython/micropython/issues/15230
+# on rp2, but doubles as a general property to test across all ports.
+import sys
+import _thread
+
+try:
+    import select
+except ImportError:
+    print("SKIP")
+    raise SystemExit
+
+
+class StdinWaiter:
+    def __init__(self):
+        self._done = False
+
+    def wait_stdin(self, timeout_ms):
+        poller = select.poll()
+        poller.register(sys.stdin, select.POLLIN)
+        poller.poll(timeout_ms)
+        # Ignoring the poll result as we don't expect any input
+        self._done = True
+
+    def is_done(self):
+        return self._done
+
+
+thread_waiter = StdinWaiter()
+_thread.start_new_thread(thread_waiter.wait_stdin, (1000,))
+StdinWaiter().wait_stdin(1000)
+
+# Spinning here is mostly not necessary but there is some inconsistency waking
+# the two threads, especially on CPython CI runners where the thread may not
+# have run yet. The actual delay is <20ms but spinning here instead of
+# sleep(0.1) means the test can run on MP builds without float support.
+while not thread_waiter.is_done():
+    pass
+
+# The background thread should have completed its wait by now.
+print(thread_waiter.is_done())