test_signal: Signal handling on the Tru64 buildbot

appears to be utterly insane.  Plug some theoretical
insecurities in the test script:

- Verify that the SIGALRM handler was actually installed.

- Don't call alarm() before the handler is installed.

- Move everything that can fail inside the try/finally,
  so the test cleans up after itself more often.

- Try sending all the expected signals in
  force_test_exit(), not just SIGALRM.  Since that was
  fixed to actually send SIGALRM (instead of invisibly
  dying with an AttributeError), we've seen that sending
  SIGALRM alone does not stop this from hanging.

- Move the "kill the child" business into the finally
  clause, so the child doesn't survive test failure
  to send SIGALRM to other tests later (there are also
  baffling SIGALRM-related failures in test_socket).

- Cancel the alarm in the finally clause -- if the
  test dies early, we again don't want SIGALRM showing
  up to confuse a later test.

Alas, this still relies on timing luck wrt the spawned
script that sends the test signals, but it's hard to see
how waiting for seconds can so often be so unlucky.

test_threadedsignals:  curiously, this test never fails
on Tru64, but doesn't normally signal SIGALRM.  Anyway,
fixed an obvious (but probably inconsequential) logic
error.

1 files changed, 84 insertions, 64 deletions

diff --git a/Lib/test/test_signal.py b/Lib/test/test_signal.py
index 694e3f0..55cd978 100644
--- a/Lib/test/test_signal.py
+++ b/Lib/test/test_signal.py
@@ -6,6 +6,8 @@ import os, sys, time
 if sys.platform[:3] in ('win', 'os2') or sys.platform=='riscos':
     raise TestSkipped, "Can't test signal on %s" % sys.platform
 
+MAX_DURATION = 20   # Entire test should last at most 20 sec.
+
 if verbose:
     x = '-x'
 else:
@@ -34,7 +36,7 @@ def handlerA(*args):
     global a_called
     a_called = True
     if verbose:
-        print "handlerA", args
+        print "handlerA invoked", args
 
 class HandlerBCalled(Exception):
     pass
@@ -43,88 +45,111 @@ def handlerB(*args):
     global b_called
     b_called = True
     if verbose:
-        print "handlerB", args
+        print "handlerB invoked", args
     raise HandlerBCalled, args
 
-MAX_DURATION = 20
-signal.alarm(MAX_DURATION)   # Entire test should last at most 20 sec.
+# Set up a child to send signals to us (the parent) after waiting long
+# enough to receive the alarm.  It seems we miss the alarm for some
+# reason.  This will hopefully stop the hangs on Tru64/Alpha.
+# Alas, it doesn't.  Tru64 appears to miss all the signals at times, or
+# seemingly random subsets of them, and nothing done in force_test_exit
+# so far has actually helped.
+def force_test_exit():
+    # Sigh, both imports seem necessary to avoid errors.
+    import os
+    fork_pid = os.fork()
+    if fork_pid:
+        # In parent.
+        return fork_pid
+
+    # In child.
+    import os, time
+    try:
+        # Wait 5 seconds longer than the expected alarm to give enough
+        # time for the normal sequence of events to occur.  This is
+        # just a stop-gap to try to prevent the test from hanging.
+        time.sleep(MAX_DURATION + 5)
+        print >> sys.__stdout__, '  child should not have to kill parent'
+        for signame in "SIGHUP", "SIGUSR1", "SIGUSR2", "SIGALRM":
+            os.kill(pid, getattr(signal, signame))
+            print >> sys.__stdout__, "    child sent", signame, "to", pid
+            time.sleep(1)
+    finally:
+        os._exit(0)
+
+# Install handlers.
 hup = signal.signal(signal.SIGHUP, handlerA)
 usr1 = signal.signal(signal.SIGUSR1, handlerB)
 usr2 = signal.signal(signal.SIGUSR2, signal.SIG_IGN)
 alrm = signal.signal(signal.SIGALRM, signal.default_int_handler)
 
-vereq(signal.getsignal(signal.SIGHUP), handlerA)
-vereq(signal.getsignal(signal.SIGUSR1), handlerB)
-vereq(signal.getsignal(signal.SIGUSR2), signal.SIG_IGN)
-
 try:
-    signal.signal(4242, handlerB)
-    raise TestFailed, 'expected ValueError for invalid signal # to signal()'
-except ValueError:
-    pass
 
-try:
-    signal.getsignal(4242)
-    raise TestFailed, 'expected ValueError for invalid signal # to getsignal()'
-except ValueError:
-    pass
+    signal.alarm(MAX_DURATION)
+    vereq(signal.getsignal(signal.SIGHUP), handlerA)
+    vereq(signal.getsignal(signal.SIGUSR1), handlerB)
+    vereq(signal.getsignal(signal.SIGUSR2), signal.SIG_IGN)
+    vereq(signal.getsignal(signal.SIGALRM), signal.default_int_handler)
 
-try:
-    signal.signal(signal.SIGUSR1, None)
-    raise TestFailed, 'expected TypeError for non-callable'
-except TypeError:
-    pass
+    # Try to ensure this test exits even if there is some problem with alarm.
+    # Tru64/Alpha often hangs and is ultimately killed by the buildbot.
+    fork_pid = force_test_exit()
 
-# Set up a child to send an alarm signal to us (the parent) after waiting
-# long enough to receive the alarm.  It seems we miss the alarm for some
-# reason.  This will hopefully stop the hangs on Tru64/Alpha.
-def force_test_exit():
-    # Sigh, both imports seem necessary to avoid errors.
-    import os
-    fork_pid = os.fork()
-    if fork_pid == 0:
-        # In child
-        import os, time
-        try:
-            # Wait 5 seconds longer than the expected alarm to give enough
-            # time for the normal sequence of events to occur.  This is
-            # just a stop-gap to prevent the test from hanging.
-            time.sleep(MAX_DURATION + 5)
-            print >> sys.__stdout__, '  child should not have to kill parent'
-            for i in range(3):
-                os.kill(pid, signal.SIGALRM)
-                print >> sys.__stdout__, "    child sent SIGALRM to", pid
-        finally:
-            os._exit(0)
-    # In parent (or error)
-    return fork_pid
+    try:
+        signal.getsignal(4242)
+        raise TestFailed('expected ValueError for invalid signal # to '
+                         'getsignal()')
+    except ValueError:
+        pass
 
-try:
-    os.system(script)
+    try:
+        signal.signal(4242, handlerB)
+        raise TestFailed('expected ValueError for invalid signal # to '
+                         'signal()')
+    except ValueError:
+        pass
 
-    # Try to ensure this test exits even if there is some problem with alarm.
-    # Tru64/Alpha sometimes hangs and is ultimately killed by the buildbot.
-    fork_pid = force_test_exit()
-    print "starting pause() loop..."
+    try:
+        signal.signal(signal.SIGUSR1, None)
+        raise TestFailed('expected TypeError for non-callable')
+    except TypeError:
+        pass
 
+    # Launch an external script to send us signals.
+    # We expect the external script to:
+    #    send HUP, which invokes handlerA to set a_called
+    #    send USR1, which invokes handlerB to set b_called and raise
+    #               HandlerBCalled
+    #    send USR2, which is ignored
+    #
+    # Then we expect the alarm to go off, and its handler raises
+    # KeyboardInterrupt, finally getting us out of the loop.
+    os.system(script)
     try:
+        if verbose:
+            print "starting pause() loop..."
         while 1:
-            if verbose:
-                print "call pause()..."
             try:
+                if verbose:
+                    print "call pause()..."
                 signal.pause()
                 if verbose:
                     print "pause() returned"
             except HandlerBCalled:
                 if verbose:
                     print "HandlerBCalled exception caught"
-                else:
-                    pass
 
     except KeyboardInterrupt:
         if verbose:
-            print "KeyboardInterrupt (assume the alarm() went off)"
+            print "KeyboardInterrupt (the alarm() went off)"
+
+    if not a_called:
+        print 'HandlerA not called'
+
+    if not b_called:
+        print 'HandlerB not called'
 
+finally:
     # Forcibly kill the child we created to ping us if there was a test error.
     try:
         # Make sure we don't kill ourself if there was a fork error.
@@ -132,16 +157,11 @@ try:
             os.kill(fork_pid, signal.SIGKILL)
     except:
         # If the child killed us, it has probably exited.  Killing a
-        # non-existant process will raise an error which we don't care about.
+        # non-existent process will raise an error which we don't care about.
         pass
 
-    if not a_called:
-        print 'HandlerA not called'
-
-    if not b_called:
-        print 'HandlerB not called'
-
-finally:
+    # Restore handlers.
+    signal.alarm(0) # cancel alarm in case we died early
     signal.signal(signal.SIGHUP, hup)
     signal.signal(signal.SIGUSR1, usr1)
     signal.signal(signal.SIGUSR2, usr2)