u-boot/tools/patman/cros_subprocess.py

402 lines
15 KiB
Python
Raw Normal View History

# Copyright (c) 2012 The Chromium OS Authors.
# Use of this source code is governed by a BSD-style license that can be
# found in the LICENSE file.
#
# Copyright (c) 2003-2005 by Peter Astrand <astrand@lysator.liu.se>
# Licensed to PSF under a Contributor Agreement.
# See http://www.python.org/2.4/license for licensing details.
"""Subprocess execution
This module holds a subclass of subprocess.Popen with our own required
features, mainly that we get access to the subprocess output while it
is running rather than just at the end. This makes it easier to show
progress information and filter output in real time.
"""
import errno
import os
import pty
import select
import subprocess
import sys
import unittest
# Import these here so the caller does not need to import subprocess also.
PIPE = subprocess.PIPE
STDOUT = subprocess.STDOUT
PIPE_PTY = -3 # Pipe output through a pty
stay_alive = True
class Popen(subprocess.Popen):
"""Like subprocess.Popen with ptys and incremental output
This class deals with running a child process and filtering its output on
both stdout and stderr while it is running. We do this so we can monitor
progress, and possibly relay the output to the user if requested.
The class is similar to subprocess.Popen, the equivalent is something like:
Popen(args, stdout=subprocess.PIPE, stderr=subprocess.PIPE)
But this class has many fewer features, and two enhancement:
1. Rather than getting the output data only at the end, this class sends it
to a provided operation as it arrives.
2. We use pseudo terminals so that the child will hopefully flush its output
to us as soon as it is produced, rather than waiting for the end of a
line.
Use communicate_filter() to handle output from the subprocess.
"""
def __init__(self, args, stdin=None, stdout=PIPE_PTY, stderr=PIPE_PTY,
shell=False, cwd=None, env=None, **kwargs):
"""Cut-down constructor
Args:
args: Program and arguments for subprocess to execute.
stdin: See subprocess.Popen()
stdout: See subprocess.Popen(), except that we support the sentinel
value of cros_subprocess.PIPE_PTY.
stderr: See subprocess.Popen(), except that we support the sentinel
value of cros_subprocess.PIPE_PTY.
shell: See subprocess.Popen()
cwd: Working directory to change to for subprocess, or None if none.
env: Environment to use for this subprocess, or None to inherit parent.
kwargs: No other arguments are supported at the moment. Passing other
arguments will cause a ValueError to be raised.
"""
stdout_pty = None
stderr_pty = None
if stdout == PIPE_PTY:
stdout_pty = pty.openpty()
stdout = os.fdopen(stdout_pty[1])
if stderr == PIPE_PTY:
stderr_pty = pty.openpty()
stderr = os.fdopen(stderr_pty[1])
super(Popen, self).__init__(args, stdin=stdin,
stdout=stdout, stderr=stderr, shell=shell, cwd=cwd, env=env,
**kwargs)
# If we're on a PTY, we passed the slave half of the PTY to the subprocess.
# We want to use the master half on our end from now on. Setting this here
# does make some assumptions about the implementation of subprocess, but
# those assumptions are pretty minor.
# Note that if stderr is STDOUT, then self.stderr will be set to None by
# this constructor.
if stdout_pty is not None:
self.stdout = os.fdopen(stdout_pty[0])
if stderr_pty is not None:
self.stderr = os.fdopen(stderr_pty[0])
# Insist that unit tests exist for other arguments we don't support.
if kwargs:
raise ValueError("Unit tests do not test extra args - please add tests")
def convert_data(self, data):
"""Convert stdout/stderr data to the correct format for output
Args:
data: Data to convert, or None for ''
Returns:
Converted data, as bytes
"""
if data is None:
return b''
return data
def communicate_filter(self, output, input_buf=''):
"""Interact with process: Read data from stdout and stderr.
This method runs until end-of-file is reached, then waits for the
subprocess to terminate.
The output function is sent all output from the subprocess and must be
defined like this:
def output([self,] stream, data)
Args:
stream: the stream the output was received on, which will be
sys.stdout or sys.stderr.
data: a string containing the data
buildman: Detect Kconfig loops Hex and int Kconfig options are supposed to have defaults. This is so we can configure U-Boot without having to enter particular values for the items that don't have specific values in the board's defconfig file. If this rule is not followed, then introducing a new Kconfig can produce a loop like this: Break things (BREAK_ME) [] (NEW) Error in reading or end of file. Break things (BREAK_ME) [] (NEW) Error in reading or end of file. The continues forever since buildman passes /dev/null to 'conf', and the build system just tries again. Eventually there is so much output that buildman runs out of memory. We can detect this situation by looking for a symbol (like 'BREAK_ME') which has no default (the '[]' above) and is marked as new. If this appears multiple times in the output, we know something is wrong. Add a filter function for the output which detects this situation. Allow it to return True to terminate the process. Implement this termination in cros_subprocess. With this we get a nice message: buildman --board sandbox -T0 Building current source for 1 boards (0 threads, 32 jobs per thread) sandbox: w+ sandbox +.config:66:warning: symbol value '' invalid for BREAK_ME + +Error in reading or end of file. +make[3]: *** [scripts/kconfig/Makefile:75: syncconfig] Terminated +make[2]: *** [Makefile:569: syncconfig] Terminated +make: *** [Makefile:177: sub-make] Terminated +(** did you define an int/hex Kconfig with no default? **) Signed-off-by: Simon Glass <sjg@chromium.org>
2021-10-20 03:43:24 +00:00
Returns:
True to terminate the process
Note: The data read is buffered in memory, so do not use this
method if the data size is large or unlimited.
Args:
output: Function to call with each fragment of output.
Returns:
A tuple (stdout, stderr, combined) which is the data received on
stdout, stderr and the combined data (interleaved stdout and stderr).
Note that the interleaved output will only be sensible if you have
set both stdout and stderr to PIPE or PIPE_PTY. Even then it depends on
the timing of the output in the subprocess. If a subprocess flips
between stdout and stderr quickly in succession, by the time we come to
read the output from each we may see several lines in each, and will read
all the stdout lines, then all the stderr lines. So the interleaving
may not be correct. In this case you might want to pass
stderr=cros_subprocess.STDOUT to the constructor.
This feature is still useful for subprocesses where stderr is
rarely used and indicates an error.
Note also that if you set stderr to STDOUT, then stderr will be empty
and the combined output will just be the same as stdout.
"""
read_set = []
write_set = []
stdout = None # Return
stderr = None # Return
if self.stdin:
# Flush stdio buffer. This might block, if the user has
# been writing to .stdin in an uncontrolled fashion.
self.stdin.flush()
if input_buf:
write_set.append(self.stdin)
else:
self.stdin.close()
if self.stdout:
read_set.append(self.stdout)
stdout = bytearray()
if self.stderr and self.stderr != self.stdout:
read_set.append(self.stderr)
stderr = bytearray()
combined = bytearray()
buildman: Detect Kconfig loops Hex and int Kconfig options are supposed to have defaults. This is so we can configure U-Boot without having to enter particular values for the items that don't have specific values in the board's defconfig file. If this rule is not followed, then introducing a new Kconfig can produce a loop like this: Break things (BREAK_ME) [] (NEW) Error in reading or end of file. Break things (BREAK_ME) [] (NEW) Error in reading or end of file. The continues forever since buildman passes /dev/null to 'conf', and the build system just tries again. Eventually there is so much output that buildman runs out of memory. We can detect this situation by looking for a symbol (like 'BREAK_ME') which has no default (the '[]' above) and is marked as new. If this appears multiple times in the output, we know something is wrong. Add a filter function for the output which detects this situation. Allow it to return True to terminate the process. Implement this termination in cros_subprocess. With this we get a nice message: buildman --board sandbox -T0 Building current source for 1 boards (0 threads, 32 jobs per thread) sandbox: w+ sandbox +.config:66:warning: symbol value '' invalid for BREAK_ME + +Error in reading or end of file. +make[3]: *** [scripts/kconfig/Makefile:75: syncconfig] Terminated +make[2]: *** [Makefile:569: syncconfig] Terminated +make: *** [Makefile:177: sub-make] Terminated +(** did you define an int/hex Kconfig with no default? **) Signed-off-by: Simon Glass <sjg@chromium.org>
2021-10-20 03:43:24 +00:00
stop_now = False
input_offset = 0
while read_set or write_set:
try:
rlist, wlist, _ = select.select(read_set, write_set, [], 0.2)
except select.error as e:
if e.args[0] == errno.EINTR:
continue
raise
if not stay_alive:
self.terminate()
if self.stdin in wlist:
# When select has indicated that the file is writable,
# we can write up to PIPE_BUF bytes without risk
# blocking. POSIX defines PIPE_BUF >= 512
chunk = input_buf[input_offset : input_offset + 512]
bytes_written = os.write(self.stdin.fileno(), chunk)
input_offset += bytes_written
if input_offset >= len(input_buf):
self.stdin.close()
write_set.remove(self.stdin)
if self.stdout in rlist:
data = b''
# We will get an error on read if the pty is closed
try:
data = os.read(self.stdout.fileno(), 1024)
except OSError:
pass
if not len(data):
self.stdout.close()
read_set.remove(self.stdout)
else:
stdout += data
combined += data
if output:
buildman: Detect Kconfig loops Hex and int Kconfig options are supposed to have defaults. This is so we can configure U-Boot without having to enter particular values for the items that don't have specific values in the board's defconfig file. If this rule is not followed, then introducing a new Kconfig can produce a loop like this: Break things (BREAK_ME) [] (NEW) Error in reading or end of file. Break things (BREAK_ME) [] (NEW) Error in reading or end of file. The continues forever since buildman passes /dev/null to 'conf', and the build system just tries again. Eventually there is so much output that buildman runs out of memory. We can detect this situation by looking for a symbol (like 'BREAK_ME') which has no default (the '[]' above) and is marked as new. If this appears multiple times in the output, we know something is wrong. Add a filter function for the output which detects this situation. Allow it to return True to terminate the process. Implement this termination in cros_subprocess. With this we get a nice message: buildman --board sandbox -T0 Building current source for 1 boards (0 threads, 32 jobs per thread) sandbox: w+ sandbox +.config:66:warning: symbol value '' invalid for BREAK_ME + +Error in reading or end of file. +make[3]: *** [scripts/kconfig/Makefile:75: syncconfig] Terminated +make[2]: *** [Makefile:569: syncconfig] Terminated +make: *** [Makefile:177: sub-make] Terminated +(** did you define an int/hex Kconfig with no default? **) Signed-off-by: Simon Glass <sjg@chromium.org>
2021-10-20 03:43:24 +00:00
stop_now = output(sys.stdout, data)
if self.stderr in rlist:
data = b''
# We will get an error on read if the pty is closed
try:
data = os.read(self.stderr.fileno(), 1024)
except OSError:
pass
if not len(data):
self.stderr.close()
read_set.remove(self.stderr)
else:
stderr += data
combined += data
if output:
buildman: Detect Kconfig loops Hex and int Kconfig options are supposed to have defaults. This is so we can configure U-Boot without having to enter particular values for the items that don't have specific values in the board's defconfig file. If this rule is not followed, then introducing a new Kconfig can produce a loop like this: Break things (BREAK_ME) [] (NEW) Error in reading or end of file. Break things (BREAK_ME) [] (NEW) Error in reading or end of file. The continues forever since buildman passes /dev/null to 'conf', and the build system just tries again. Eventually there is so much output that buildman runs out of memory. We can detect this situation by looking for a symbol (like 'BREAK_ME') which has no default (the '[]' above) and is marked as new. If this appears multiple times in the output, we know something is wrong. Add a filter function for the output which detects this situation. Allow it to return True to terminate the process. Implement this termination in cros_subprocess. With this we get a nice message: buildman --board sandbox -T0 Building current source for 1 boards (0 threads, 32 jobs per thread) sandbox: w+ sandbox +.config:66:warning: symbol value '' invalid for BREAK_ME + +Error in reading or end of file. +make[3]: *** [scripts/kconfig/Makefile:75: syncconfig] Terminated +make[2]: *** [Makefile:569: syncconfig] Terminated +make: *** [Makefile:177: sub-make] Terminated +(** did you define an int/hex Kconfig with no default? **) Signed-off-by: Simon Glass <sjg@chromium.org>
2021-10-20 03:43:24 +00:00
stop_now = output(sys.stderr, data)
if stop_now:
self.terminate()
# All data exchanged. Translate lists into strings.
stdout = self.convert_data(stdout)
stderr = self.convert_data(stderr)
combined = self.convert_data(combined)
self.wait()
return (stdout, stderr, combined)
# Just being a unittest.TestCase gives us 14 public methods. Unless we
# disable this, we can only have 6 tests in a TestCase. That's not enough.
#
# pylint: disable=R0904
class TestSubprocess(unittest.TestCase):
"""Our simple unit test for this module"""
class MyOperation:
"""Provides a operation that we can pass to Popen"""
def __init__(self, input_to_send=None):
"""Constructor to set up the operation and possible input.
Args:
input_to_send: a text string to send when we first get input. We will
add \r\n to the string.
"""
self.stdout_data = ''
self.stderr_data = ''
self.combined_data = ''
self.stdin_pipe = None
self._input_to_send = input_to_send
if input_to_send:
pipe = os.pipe()
self.stdin_read_pipe = pipe[0]
self._stdin_write_pipe = os.fdopen(pipe[1], 'w')
def output(self, stream, data):
"""Output handler for Popen. Stores the data for later comparison"""
if stream == sys.stdout:
self.stdout_data += data
if stream == sys.stderr:
self.stderr_data += data
self.combined_data += data
# Output the input string if we have one.
if self._input_to_send:
self._stdin_write_pipe.write(self._input_to_send + '\r\n')
self._stdin_write_pipe.flush()
def _basic_check(self, plist, oper):
"""Basic checks that the output looks sane."""
self.assertEqual(plist[0], oper.stdout_data)
self.assertEqual(plist[1], oper.stderr_data)
self.assertEqual(plist[2], oper.combined_data)
# The total length of stdout and stderr should equal the combined length
self.assertEqual(len(plist[0]) + len(plist[1]), len(plist[2]))
def test_simple(self):
"""Simple redirection: Get process list"""
oper = TestSubprocess.MyOperation()
plist = Popen(['ps']).communicate_filter(oper.output)
self._basic_check(plist, oper)
def test_stderr(self):
"""Check stdout and stderr"""
oper = TestSubprocess.MyOperation()
cmd = 'echo fred >/dev/stderr && false || echo bad'
plist = Popen([cmd], shell=True).communicate_filter(oper.output)
self._basic_check(plist, oper)
self.assertEqual(plist [0], 'bad\r\n')
self.assertEqual(plist [1], 'fred\r\n')
def test_shell(self):
"""Check with and without shell works"""
oper = TestSubprocess.MyOperation()
cmd = 'echo test >/dev/stderr'
self.assertRaises(OSError, Popen, [cmd], shell=False)
plist = Popen([cmd], shell=True).communicate_filter(oper.output)
self._basic_check(plist, oper)
self.assertEqual(len(plist [0]), 0)
self.assertEqual(plist [1], 'test\r\n')
def test_list_args(self):
"""Check with and without shell works using list arguments"""
oper = TestSubprocess.MyOperation()
cmd = ['echo', 'test', '>/dev/stderr']
plist = Popen(cmd, shell=False).communicate_filter(oper.output)
self._basic_check(plist, oper)
self.assertEqual(plist [0], ' '.join(cmd[1:]) + '\r\n')
self.assertEqual(len(plist [1]), 0)
oper = TestSubprocess.MyOperation()
# this should be interpreted as 'echo' with the other args dropped
cmd = ['echo', 'test', '>/dev/stderr']
plist = Popen(cmd, shell=True).communicate_filter(oper.output)
self._basic_check(plist, oper)
self.assertEqual(plist [0], '\r\n')
def test_cwd(self):
"""Check we can change directory"""
for shell in (False, True):
oper = TestSubprocess.MyOperation()
plist = Popen('pwd', shell=shell, cwd='/tmp').communicate_filter(
oper.output)
self._basic_check(plist, oper)
self.assertEqual(plist [0], '/tmp\r\n')
def test_env(self):
"""Check we can change environment"""
for add in (False, True):
oper = TestSubprocess.MyOperation()
env = os.environ
if add:
env ['FRED'] = 'fred'
cmd = 'echo $FRED'
plist = Popen(cmd, shell=True, env=env).communicate_filter(oper.output)
self._basic_check(plist, oper)
self.assertEqual(plist [0], add and 'fred\r\n' or '\r\n')
def test_extra_args(self):
"""Check we can't add extra arguments"""
self.assertRaises(ValueError, Popen, 'true', close_fds=False)
def test_basic_input(self):
"""Check that incremental input works
We set up a subprocess which will prompt for name. When we see this prompt
we send the name as input to the process. It should then print the name
properly to stdout.
"""
oper = TestSubprocess.MyOperation('Flash')
prompt = 'What is your name?: '
cmd = 'echo -n "%s"; read name; echo Hello $name' % prompt
plist = Popen([cmd], stdin=oper.stdin_read_pipe,
shell=True).communicate_filter(oper.output)
self._basic_check(plist, oper)
self.assertEqual(len(plist [1]), 0)
self.assertEqual(plist [0], prompt + 'Hello Flash\r\r\n')
def test_isatty(self):
"""Check that ptys appear as terminals to the subprocess"""
oper = TestSubprocess.MyOperation()
cmd = ('if [ -t %d ]; then echo "terminal %d" >&%d; '
'else echo "not %d" >&%d; fi;')
both_cmds = ''
for fd in (1, 2):
both_cmds += cmd % (fd, fd, fd, fd, fd)
plist = Popen(both_cmds, shell=True).communicate_filter(oper.output)
self._basic_check(plist, oper)
self.assertEqual(plist [0], 'terminal 1\r\n')
self.assertEqual(plist [1], 'terminal 2\r\n')
# Now try with PIPE and make sure it is not a terminal
oper = TestSubprocess.MyOperation()
plist = Popen(both_cmds, stdout=subprocess.PIPE, stderr=subprocess.PIPE,
shell=True).communicate_filter(oper.output)
self._basic_check(plist, oper)
self.assertEqual(plist [0], 'not 1\n')
self.assertEqual(plist [1], 'not 2\n')
if __name__ == '__main__':
unittest.main()