ztdm: inhfd: handle child hanging

Let's kill the child when the test is hanging.  Due to PEP 475 the
SIGALRM handler needs to throw an exception to be able to interrupt
wait().  To improve debuggig, close the fd in the child after reading
it and detect that in the parent to show whether the child hung
part way.

Signed-off-by: Michał Mirosław <[email protected]>

test/zdtm.py

@@ -40,6 +40,7 @@
 
 def alarm(*args):
     print("==== ALARM ====")
+    raise InterruptedError("ALARM")
 
 
 def traceit(f, e, a):
@@ -710,11 +711,9 @@ def start(self):
                     # regular files, so we loop.
                     data = b''
                     while not data:
-                        # In python 2.7, peer_file.read() doesn't call the read
-                        # system call if it's read file to the end once. The
-                        # next seek allows to workaround this problem.
-                        data = os.read(peer_file.fileno(), 16)
+                        data = os.read(peer_file.fileno(), len(msg) + 16)
                         time.sleep(0.1)
+                    peer_file.close()
                 except Exception as e:
                     print("Unable to read a peer file: %s" % e)
                     sys.exit(1)
@@ -745,7 +744,16 @@ def stop(self):
             my_file.write(msg)
             my_file.flush()
             i += 1
-        pid, status = os.waitpid(self.__peer_pid, 0)
+        signal.alarm(10)
+        try:
+            pid, status = os.waitpid(self.__peer_pid, 0)
+        except InterruptedError:
+            fds = set(os.listdir("/proc/%s/fd" % self.__peer_pid))
+            self.kill()
+            pid, status = os.waitpid(self.__peer_pid, 0)
+        fds = self.__fds.difference(fds)
+        if fds:
+            print("before SIGKILL, child managed to close fds: " + str(fds))
         with open(self.__name + ".out") as output:
             print(output.read())
         self.__peer_pid = 0