bpo-36719: regrtest -jN no longer stops on crash (GH-13231)

"python3 -m test -jN ..." now continues the execution of next tests
when a worker process crash (CHILD_ERROR state). Previously, the test
suite stopped immediately. Use --failfast to stop at the first error.

Moreover, --forever now also implies --failfast.

5 files changed, 49 insertions, 18 deletions

diff --git a/Lib/test/libregrtest/cmdline.py b/Lib/test/libregrtest/cmdline.py
index cb09ee0..dc0d880 100644
--- a/Lib/test/libregrtest/cmdline.py
+++ b/Lib/test/libregrtest/cmdline.py
@@ -256,7 +256,7 @@ def _create_parser():
                        help='suppress error message boxes on Windows')
     group.add_argument('-F', '--forever', action='store_true',
                        help='run the specified tests in a loop, until an '
-                            'error happens')
+                            'error happens; imply --failfast')
     group.add_argument('--list-tests', action='store_true',
                        help="only write the name of tests that will be run, "
                             "don't execute them")
@@ -389,5 +389,8 @@ def _parse_args(args, **kwargs):
         with open(ns.match_filename) as fp:
             for line in fp:
                 ns.match_tests.append(line.strip())
+    if ns.forever:
+        # --forever implies --failfast
+        ns.failfast = True
 
     return ns
diff --git a/Lib/test/libregrtest/main.py b/Lib/test/libregrtest/main.py
index c19ea44..02717d8c 100644
--- a/Lib/test/libregrtest/main.py
+++ b/Lib/test/libregrtest/main.py
@@ -16,7 +16,7 @@ from test.libregrtest.runtest import (
     findtests, runtest, get_abs_module,
     STDTESTS, NOTTESTS, PASSED, FAILED, ENV_CHANGED, SKIPPED, RESOURCE_DENIED,
     INTERRUPTED, CHILD_ERROR, TEST_DID_NOT_RUN,
-    PROGRESS_MIN_TIME, format_test_result)
+    PROGRESS_MIN_TIME, format_test_result, is_failed)
 from test.libregrtest.setup import setup_tests
 from test.libregrtest.utils import removepy, count, format_duration, printlist
 from test import support
@@ -404,7 +404,7 @@ class Regrtest:
             test_time = time.monotonic() - start_time
             if test_time >= PROGRESS_MIN_TIME:
                 previous_test = "%s in %s" % (previous_test, format_duration(test_time))
-            elif result[0] == PASSED:
+            elif result.result == PASSED:
                 # be quiet: say nothing if the test passed shortly
                 previous_test = None
 
@@ -413,6 +413,9 @@ class Regrtest:
                 if module not in save_modules and module.startswith("test."):
                     support.unload(module)
 
+            if self.ns.failfast and is_failed(result, self.ns):
+                break
+
         if previous_test:
             print(previous_test)
 
diff --git a/Lib/test/libregrtest/runtest.py b/Lib/test/libregrtest/runtest.py
index a957492..a43b766 100644
--- a/Lib/test/libregrtest/runtest.py
+++ b/Lib/test/libregrtest/runtest.py
@@ -24,7 +24,7 @@ SKIPPED = -2
 RESOURCE_DENIED = -3
 INTERRUPTED = -4
 CHILD_ERROR = -5   # error in a child process
-TEST_DID_NOT_RUN = -6   # error in a child process
+TEST_DID_NOT_RUN = -6
 
 _FORMAT_TEST_RESULT = {
     PASSED: '%s passed',
@@ -64,6 +64,15 @@ NOTTESTS = set()
 FOUND_GARBAGE = []
 
 
+def is_failed(result, ns):
+    ok = result.result
+    if ok in (PASSED, RESOURCE_DENIED, SKIPPED, TEST_DID_NOT_RUN):
+        return False
+    if ok == ENV_CHANGED:
+        return ns.fail_env_changed
+    return True
+
+
 def format_test_result(result):
     fmt = _FORMAT_TEST_RESULT.get(result.result, "%s")
     return fmt % result.test_name
diff --git a/Lib/test/libregrtest/runtest_mp.py b/Lib/test/libregrtest/runtest_mp.py
index dbab695..ced7f86 100644
--- a/Lib/test/libregrtest/runtest_mp.py
+++ b/Lib/test/libregrtest/runtest_mp.py
@@ -13,7 +13,7 @@ from test import support
 
 from test.libregrtest.runtest import (
     runtest, INTERRUPTED, CHILD_ERROR, PROGRESS_MIN_TIME,
-    format_test_result, TestResult)
+    format_test_result, TestResult, is_failed)
 from test.libregrtest.setup import setup_tests
 from test.libregrtest.utils import format_duration
 
@@ -22,8 +22,12 @@ from test.libregrtest.utils import format_duration
 PROGRESS_UPDATE = 30.0   # seconds
 
 
-def must_stop(result):
-    return result.result in (INTERRUPTED, CHILD_ERROR)
+def must_stop(result, ns):
+    if result.result == INTERRUPTED:
+        return True
+    if ns.failfast and is_failed(result, ns):
+        return True
+    return False
 
 
 def run_test_in_subprocess(testname, ns):
@@ -66,16 +70,22 @@ class MultiprocessIterator:
 
     """A thread-safe iterator over tests for multiprocess mode."""
 
-    def __init__(self, tests):
+    def __init__(self, tests_iter):
         self.lock = threading.Lock()
-        self.tests = tests
+        self.tests_iter = tests_iter
 
     def __iter__(self):
         return self
 
     def __next__(self):
         with self.lock:
-            return next(self.tests)
+            if self.tests_iter is None:
+                raise StopIteration
+            return next(self.tests_iter)
+
+    def stop(self):
+        with self.lock:
+            self.tests_iter = None
 
 
 MultiprocessResult = collections.namedtuple('MultiprocessResult',
@@ -92,23 +102,24 @@ class MultiprocessThread(threading.Thread):
         self._popen = None
 
     def kill(self):
-        if not self.is_alive():
+        popen = self._popen
+        if popen is None:
             return
-        if self._popen is not None:
-            self._popen.kill()
+        print("Kill regrtest worker process %s" % popen.pid)
+        popen.kill()
 
     def _runtest(self, test_name):
         try:
             self.start_time = time.monotonic()
             self.current_test_name = test_name
 
-            popen = run_test_in_subprocess(test_name, self.ns)
-            self._popen = popen
+            self._popen = run_test_in_subprocess(test_name, self.ns)
+            popen = self._popen
             with popen:
                 try:
                     stdout, stderr = popen.communicate()
                 except:
-                    popen.kill()
+                    self.kill()
                     popen.wait()
                     raise
 
@@ -153,7 +164,7 @@ class MultiprocessThread(threading.Thread):
                 mp_result = self._runtest(test_name)
                 self.output.put((False, mp_result))
 
-                if must_stop(mp_result.result):
+                if must_stop(mp_result.result, self.ns):
                     break
             except BaseException:
                 self.output.put((True, traceback.format_exc()))
@@ -255,7 +266,7 @@ class MultiprocessRunner:
         if mp_result.stderr and not self.ns.pgo:
             print(mp_result.stderr, file=sys.stderr, flush=True)
 
-        if must_stop(mp_result.result):
+        if must_stop(mp_result.result, self.ns):
             return True
 
         return False
@@ -280,6 +291,8 @@ class MultiprocessRunner:
             if self.test_timeout is not None:
                 faulthandler.cancel_dump_traceback_later()
 
+        # a test failed (and --failfast is set) or all tests completed
+        self.pending.stop()
         self.wait_workers()
 
 
diff --git a/Misc/NEWS.d/next/Tests/2019-05-10-01-50-30.bpo-36719.O84ZWv.rst b/Misc/NEWS.d/next/Tests/2019-05-10-01-50-30.bpo-36719.O84ZWv.rst
new file mode 100644
index 0000000..9f60145
--- /dev/null
+++ b/Misc/NEWS.d/next/Tests/2019-05-10-01-50-30.bpo-36719.O84ZWv.rst
@@ -0,0 +1,3 @@
+"python3 -m test -jN ..." now continues the execution of next tests when a
+worker process crash (CHILD_ERROR state). Previously, the test suite stopped
+immediately. Use --failfast to stop at the first error.