gh-102795: Fix use of poll in test_epoll's test_control_and_wait (#102796)

This test can fail unnecessarily. In the test we wait for events on two
file descriptors. This is done in a single call to select.epoll's poll()
function. However, it is valid for the OS to return only one event via
poll() and the next via a subsequent call to poll(). This rarely
happens, but it can cause the test to fail despite properly functioning
polling.

Instead, we poll a second time when necessary.

Lib/test/test_epoll.py

@@ -27,6 +27,7 @@ import select
 import socket
 import time
 import unittest
+from test import support
 
 if not hasattr(select, "epoll"):
     raise unittest.SkipTest("test works only on Linux 2.6")
@@ -186,10 +187,16 @@ class TestEPoll(unittest.TestCase):
         client.sendall(b"Hello!")
         server.sendall(b"world!!!")
 
+        # we might receive events one at a time, necessitating multiple calls to
+        # poll
+        events = []
+        for _ in support.busy_retry(support.SHORT_TIMEOUT):
             now = time.monotonic()
-        events = ep.poll(1.0, 4)
+            events += ep.poll(1.0, 4)
             then = time.monotonic()
             self.assertFalse(then - now > 0.01)
+            if len(events) >= 2:
+                break
 
         expected = [(client.fileno(), select.EPOLLIN | select.EPOLLOUT),
                     (server.fileno(), select.EPOLLIN | select.EPOLLOUT)]

Misc/NEWS.d/next/Tests/2023-03-17-22-00-47.gh-issue-102795.z21EoC.rst

@@ -0,0 +1 @@
+fix use of poll in test_epoll's test_control_and_wait