test: fix pty test hangs on aix

Some pty tests persistently hung on the AIX CI buildbots. Fix that by
adding a helper script that properly sets up the pty before spawning
the script under test.

On investigation I discovered that the test runner hung when it tried
to close the slave pty's file descriptor, probably due to a bug in
AIX's pty implementation. I could reproduce it with a short C program.
The test runner also leaked file descriptors to the child process.

I couldn't convince python's `subprocess.Popen()` to do what I wanted
it to do so I opted to move the logic to a helper script that can do
fork/setsid/etc. without having to worry about stomping on state in
tools/test.py.

In the process I also uncovered some bugs in the pty module of the
python distro that ships with macOS 10.14, leading me to reimplement
a sizable chunk of the functionality of that module.

And last but not least, of course there are differences between ptys
on different platforms and the helper script has to paper over that.
Of course.

Really, this commit took me longer to put together than I care to admit.

Caveat emptor: this commit takes the hacky ^D feeding to the slave out
of tools/test.py and puts it in the *.in input files. You can also feed
other control characters to tests, like ^C or ^Z, simply by inserting
them into the corresponding input file. I think that's nice.

Fixes: nodejs/build#1820
Fixes: nodejs#28489

PR-URL: nodejs#28600
Backport-PR-URL: nodejs#28826
Reviewed-By: Richard Lau <riclau@uk.ibm.com>
Reviewed-By: Sam Roberts <vieuxtech@gmail.com>
Reviewed-By: Anna Henningsen <anna@addaleax.net>

test/pseudo-tty/pseudo-tty.status

@@ -1,13 +1,5 @@
 prefix pseudo-tty
 
-[$system==aix]
-# being investigated under https://github.com/nodejs/node/issues/9728
-test-tty-wrap              : FAIL, PASS
-# https://github.com/nodejs/build/issues/1820#issuecomment-505998851
-# https://github.com/nodejs/node/pull/28469
-console-dumb-tty: SKIP
-test-fatal-error: SKIP
-
 [$system==solaris]
 # https://github.com/nodejs/node/pull/16225 - `ioctl(fd, TIOCGWINSZ)` seems
 # to fail with EINVAL on SmartOS when `fd` is a pty from python's pty module.

test/pseudo-tty/pty_helper.py

@@ -0,0 +1,98 @@
+import errno
+import os
+import pty
+import select
+import signal
+import sys
+import termios
+
+STDIN = 0
+STDOUT = 1
+STDERR = 2
+
+
+def pipe(sfd, dfd):
+  try:
+    data = os.read(sfd, 256)
+  except OSError as e:
+    if e.errno != errno.EIO:
+      raise
+    return True  # EOF
+
+  if not data:
+    return True  # EOF
+
+  if dfd == STDOUT:
+    # Work around platform quirks. Some platforms echo ^D as \x04
+    # (AIX, BSDs) and some don't (Linux).
+    filt = lambda c: ord(c) > 31 or c in '\t\n\r\f'
+    data = filter(filt, data)
+
+  while data:
+    try:
+      n = os.write(dfd, data)
+    except OSError as e:
+      if e.errno != errno.EIO:
+        raise
+      return True  # EOF
+    data = data[n:]
+
+
+if __name__ == '__main__':
+  argv = sys.argv[1:]
+
+  # Make select() interruptable by SIGCHLD.
+  signal.signal(signal.SIGCHLD, lambda nr, _: None)
+
+  master_fd, slave_fd = pty.openpty()
+  assert master_fd > STDIN
+
+  mode = termios.tcgetattr(slave_fd)
+  # Don't translate \n to \r\n.
+  mode[1] = mode[1] & ~termios.ONLCR  # oflag
+  # Disable ECHOCTL. It's a BSD-ism that echoes e.g. \x04 as ^D but it
+  # doesn't work on platforms like AIX and Linux. I checked Linux's tty
+  # driver and it's a no-op, the driver is just oblivious to the flag.
+  mode[3] = mode[3] & ~termios.ECHOCTL  # lflag
+  termios.tcsetattr(slave_fd, termios.TCSANOW, mode)
+
+  pid = os.fork()
+  if not pid:
+    os.setsid()
+    os.close(master_fd)
+
+    # Ensure the pty is a controlling tty.
+    name = os.ttyname(slave_fd)
+    fd = os.open(name, os.O_RDWR)
+    os.dup2(fd, slave_fd)
+    os.close(fd)
+
+    os.dup2(slave_fd, STDIN)
+    os.dup2(slave_fd, STDOUT)
+    os.dup2(slave_fd, STDERR)
+
+    if slave_fd > STDERR:
+      os.close(slave_fd)
+
+    os.execve(argv[0], argv, os.environ)
+    raise Exception('unreachable')
+
+  os.close(slave_fd)
+
+  fds = [STDIN, master_fd]
+  while fds:
+    try:
+      rfds, _, _ = select.select(fds, [], [])
+    except select.error as e:
+      if e[0] != errno.EINTR:
+        raise
+      if pid == os.waitpid(pid, os.WNOHANG)[0]:
+        break
+
+    if STDIN in rfds:
+      if pipe(STDIN, master_fd):
+        fds.remove(STDIN)
+
+    if master_fd in rfds:
+      if pipe(master_fd, STDOUT):
+        break

test/pseudo-tty/test-stdout-read.in

@@ -1 +1,2 @@
 Hello!
+

test/pseudo-tty/test-stdout-read.out

@@ -1 +1,2 @@
+Hello!
 <Buffer 48 65 6c 6c 6f 21 0a>

test/pseudo-tty/testcfg.py

@@ -29,8 +29,9 @@
 
 import test
 import os
-from os.path import join, exists, basename, isdir
+from os.path import join, exists, basename, dirname, isdir
 import re
+import sys
 import utils
 
 try:
@@ -44,6 +45,7 @@
   xrange = range  # Python 3
 
 FLAGS_PATTERN = re.compile(r"//\s+Flags:(.*)")
+PTY_HELPER = join(dirname(__file__), 'pty_helper.py')
 
 class TTYTestCase(test.TestCase):
 
@@ -117,16 +119,18 @@ def GetSource(self):
           + open(self.expected).read())
 
   def RunCommand(self, command, env):
-    input = None
+    fd = None
     if self.input is not None and exists(self.input):
-      input = open(self.input).read()
+      fd = os.open(self.input, os.O_RDONLY)
     full_command = self.context.processor(command)
+    full_command = [sys.executable, PTY_HELPER] + full_command
     output = test.Execute(full_command,
                      self.context,
                      self.context.GetTimeout(self.mode),
                      env,
-                     faketty=True,
-                     input=input)
+                     stdin=fd)
+    if fd is not None:
+      os.close(fd)
     return test.TestOutput(self,
                       full_command,
                       output,

tools/test.py

@@ -648,15 +648,10 @@ def RunProcess(context, timeout, args, **rest):
       prev_error_mode = Win32SetErrorMode(error_mode)
       Win32SetErrorMode(error_mode | prev_error_mode)
 
-  faketty = rest.pop('faketty', False)
-  pty_out = rest.pop('pty_out')
-
   process = subprocess.Popen(
     args = popen_args,
     **rest
   )
-  if faketty:
-    os.close(rest['stdout'])
   if utils.IsWindows() and context.suppress_dialogs and prev_error_mode != SEM_INVALID_VALUE:
     Win32SetErrorMode(prev_error_mode)
   # Compute the end time - if the process crosses this limit we
@@ -668,28 +663,6 @@ def RunProcess(context, timeout, args, **rest):
   # loop and keep track of whether or not it times out.
   exit_code = None
   sleep_time = INITIAL_SLEEP_TIME
-  output = ''
-  if faketty:
-    while True:
-      if time.time() >= end_time:
-        # Kill the process and wait for it to exit.
-        KillTimedOutProcess(context, process.pid)
-        exit_code = process.wait()
-        timed_out = True
-        break
-
-      # source: http://stackoverflow.com/a/12471855/1903116
-      # related: http://stackoverflow.com/q/11165521/1903116
-      try:
-        data = os.read(pty_out, 9999)
-      except OSError as e:
-        if e.errno != errno.EIO:
-          raise
-        break # EIO means EOF on some systems
-      else:
-        if not data: # EOF
-          break
-        output += data
 
   while exit_code is None:
     if (not end_time is None) and (time.time() >= end_time):
@@ -703,7 +676,7 @@ def RunProcess(context, timeout, args, **rest):
       sleep_time = sleep_time * SLEEP_TIME_FACTOR
       if sleep_time > MAX_SLEEP_TIME:
         sleep_time = MAX_SLEEP_TIME
-  return (process, exit_code, timed_out, output)
+  return (process, exit_code, timed_out)
 
 
 def PrintError(str):
@@ -725,29 +698,12 @@ def CheckedUnlink(name):
       PrintError("os.unlink() " + str(e))
     break
 
-def Execute(args, context, timeout=None, env={}, faketty=False, disable_core_files=False, input=None):
-  if faketty:
-    import pty
-    (out_master, fd_out) = pty.openpty()
-    fd_in = fd_err = fd_out
-    pty_out = out_master
-
-    if input is not None:
-      # Before writing input data, disable echo so the input doesn't show
-      # up as part of the output.
-      import termios
-      attr = termios.tcgetattr(fd_in)
-      attr[3] = attr[3] & ~termios.ECHO
-      termios.tcsetattr(fd_in, termios.TCSADRAIN, attr)
-
-      os.write(pty_out, input)
-      os.write(pty_out, '\x04') # End-of-file marker (Ctrl+D)
-  else:
-    (fd_out, outname) = tempfile.mkstemp()
-    (fd_err, errname) = tempfile.mkstemp()
-    fd_in = 0
-    pty_out = None
+def Execute(args, context, timeout=None, env=None, disable_core_files=False, stdin=None):
+  (fd_out, outname) = tempfile.mkstemp()
+  (fd_err, errname) = tempfile.mkstemp()
 
+  if env is None:
+    env = {}
   env_copy = os.environ.copy()
 
   # Remove NODE_PATH
@@ -766,28 +722,22 @@ def disableCoreFiles():
       resource.setrlimit(resource.RLIMIT_CORE, (0,0))
     preexec_fn = disableCoreFiles
 
-  (process, exit_code, timed_out, output) = RunProcess(
+  (process, exit_code, timed_out) = RunProcess(
     context,
     timeout,
     args = args,
-    stdin = fd_in,
+    stdin = stdin,
     stdout = fd_out,
     stderr = fd_err,
     env = env_copy,
-    faketty = faketty,
-    pty_out = pty_out,
     preexec_fn = preexec_fn
   )
-  if faketty:
-    os.close(out_master)
-    errors = ''
-  else:
-    os.close(fd_out)
-    os.close(fd_err)
-    output = open(outname).read()
-    errors = open(errname).read()
-    CheckedUnlink(outname)
-    CheckedUnlink(errname)
+  os.close(fd_out)
+  os.close(fd_err)
+  output = open(outname).read()
+  errors = open(errname).read()
+  CheckedUnlink(outname)
+  CheckedUnlink(errname)
 
   return CommandOutput(exit_code, timed_out, output, errors)