configure

#!/usr/bin/env python3
"""
Configure the build of klee-uclibc. It can be built
as a native library or as an LLVM bitcode archive.

Information about configuration is written to console and
more verbose information is writen to a log file (usually
config.log)

You can use the CC environment variable (or flag --with-cc)
to force a particular C compiler. e.g.

$ CC=/usr/bin/clang ./configure

If the forced compiler is a tool name (e.g. CC=clang) then
the PATH is searched. Otherwise it is assumed to be a
relative or absolute path.

If building an LLVM bitcode archive and CC is not set
then this script will try and find a LLVM bitcode compiler
in your path in the following order

1. clang built inside `llvm-config --bindir`
2. clang in PATH

"""

import argparse
import fileinput
import logging
import os
import platform
import pprint
import shutil
import subprocess
import sys

uclibcRoot=os.path.dirname( os.path.abspath(__file__) )
templateTarget= os.path.join( uclibcRoot, 'Makefile.klee')
templateFile= os.path.join( uclibcRoot, 'Makefile.klee.in')
configLogFile= os.path.join( uclibcRoot, 'config.log')


def main(args):
    # Log everything to file
    logging.basicConfig(level=logging.DEBUG,
                        format='%(levelname)s:: %(funcName)s() at line %(lineno)d ::%(message)s',
                        filename=configLogFile,
                        filemode='w'
                        )

    # Log the arguments used so that it is easier to reinvoke later.
    logging.debug('Executing with command line:\n{} {}\n'.format( __file__, ' '.join(args)))

    # Log to console too but by default be less verbose
    consoleHandler = logging.StreamHandler()
    consoleHandler.setLevel(logging.INFO)
    consoleFormatter = logging.Formatter('%(levelname)s:%(message)s')
    consoleHandler.setFormatter(consoleFormatter)
    logging.getLogger().addHandler(consoleHandler) # Add extra handler to root logger

    parser = argparse.ArgumentParser(description=__doc__)

    # Two different modes
    meo = parser.add_mutually_exclusive_group(required=True)
    meo.add_argument('-n','--make-native', help='Produce native binary library.', action='store_true')
    meo.add_argument('-l','--make-llvm-lib', help='Produce library compiled as LLVM bitcode.', action='store_true')

    parser.add_argument('--with-llvm-config', help='Path to llvm-config executable to use. If not set then llvm-config in the PATH environment will be used.', type=str)

    parser.add_argument('--with-cc', help='Force C compiler to be used. If this and CC environment variable is not set then auto-detection will be detected.', type=str)
    parser.add_argument('--enable-assertions', default=False, action='store_true', help='Enable assertions. (Default %(default)s)')
    parser.add_argument('--enable-release', default=False, action='store_true', help='Enable Release mode. (Default %(default)s)')

    parser.add_argument('--disable-prebuilt-config', default=False, action='store_true',
                        help='Do not use pre-built config (+ any needed patches) to emulate legacy klee-uclibc. If you want to generate your own config use this option then run `make menuconfig`. (Default %(default)s)')


    parser.add_argument('--log-level',type=str, default='info',choices=['debug','info','warning','error','critical'],
                        help='Set console logging level output (Default %(default)s). Logging to "{}" is unaffected.'.format(configLogFile))


    pargs = parser.parse_args(args)

    # Configure the level for console output but leave file output untouched
    consoleHandler.setLevel(level=getattr(logging, pargs.log_level.upper(), None))

    # Force compiler if requested
    cc = pargs.with_cc
    if not cc:
        cc = os.getenv('CC')
    if cc:
        logging.info('Forcing C compiler to be...{}'.format(cc))
        if os.sep in cc:
            # Absolute or relative path
            cc = os.path.abspath(cc)
        else:
            # Search for tool in PATH
            ccAbs = shutil.which(cc)
            if ccAbs is None:
                logging.error('"{0}" is not in your path.'.format(cc))
                sys.exit(1)
            else:
                cc = ccAbs

        if not os.path.exists(cc):
            logging.error('"{}" does not exist.'.format(cc))
            sys.exit(1)

        logging.info('Absolute path to compiler...{}'.format(cc))


    if pargs.make_native:
        handleNativeConfig(pargs, cc)
    else:
        handleLLVMConfig(pargs, cc)

    uclibcConfigFile = os.path.join(uclibcRoot, '.config')

    if pargs.disable_prebuilt_config:
        logging.info('Not using pre-built config. You should run `make menuconfig` or `make config`')
        if os.path.exists(uclibcConfigFile):
            logging.warning('A pre-existing .config file was detected. You should probably remove it.')
    else:
        if os.path.exists(uclibcConfigFile):
            logging.warning('Removing existing config file...{}'.format(uclibcConfigFile))
            os.remove(uclibcConfigFile)
        installPrebuiltConfig()

def installPrebuiltConfig():
    """ This function installs pre-made .config files
        and any necessary patches for a particular architecture.

        FIXME: Remove this for upstream klee-uclibc. People should
               just run `make menuconfig` themselves. It's not hard!
    """
    p = platform.machine()
    logging.info('Setting up pre-made configure for...{}'.format(p))

    patchDir = os.path.join(uclibcRoot, 'klee-premade-configs', p)
    if not os.path.exists(patchDir):
        logging.error('"{}" does not exist. Cannot install pre-made .config file'.format(patchDir))
        sys.exit(1)

    configFile = os.path.join(patchDir, 'config')
    if not os.path.exists(configFile):
        logging.error('"{}" does not exist. Cannot install pre-made .config file'.format(patchDir))
        sys.exit(1)


    if p == 'x86_64' and platform.architecture()[0] == '64bit':
        logging.info('Installing .config file')
        shutil.copy(configFile, os.path.join(uclibcRoot, '.config'))
    elif p == 'i686' and platform.architecture()[0] == '32bit':
        logging.info('Installing .config file')
        shutil.copy(configFile, os.path.join(uclibcRoot, '.config'))
    else:
        logging.error('Your architecture is not supported. You will need to run `make menuconfig` manually')

    logging.info('Looking for kernel include path...')
    path = findKernelIncludePath()
    logging.info('Found "{}"'.format(path))

    # Patch the .config file as necessary
    for line in fileinput.input(os.path.join(uclibcRoot, '.config'),inplace=True):
        if "KERNEL_HEADERS" in line:
            line = 'KERNEL_HEADERS="' + path + '"\n'
            logging.debug('Patching KERNEL_HEADERS with path "{}"'.format(path))
        sys.stdout.write(line)

def findKernelIncludePath():
    """ This function searches for Kernel include files
        which are needed to build uclibc
    """

    std_include = os.environ.get("UCLIBC_KERNEL_HEADERS", "/usr/include")
    test_file = "asm/unistd.h"

    p = platform.machine()

    if p == 'x86_64':
        if os.path.exists( os.path.join(std_include, "x86_64-linux-gnu", test_file) ):
            return os.path.join(std_include, "x86_64-linux-gnu")
    if p == 'i686':
        if os.path.exists( os.path.join(std_include, "i386-linux-gnu", test_file) ):
            return os.path.join(std_include, "i386-linux-gnu")

    test_path = os.path.join(std_include, test_file)
    if not os.path.exists( test_path ):
        msg = ("Kernel header files not found at '%s': '%s' is absent."
               "Export the UCLIBC_KERNEL_HEADERS environment variable to change the"
               "default path ('/usr/include')")
        logging.error(msg % (std_include, test_path))
    return std_include

def runTool(cmd):
    """
    return (returnCode, output) from tool
    """
    retCode=0
    output=""
    try:
        logging.debug('Executing {0}'.format(pprint.pformat(cmd)))
        output = subprocess.check_output(cmd, stderr=subprocess.STDOUT)
    except subprocess.CalledProcessError as e:
        retCode = e.returncode
        output = e.output

    # Clean up output
    output = output.decode()

    logging.debug('Execution completed with return code:{}'.format(retCode))
    logging.debug('Execution completed with output:"{}"'.format(output))
    return (retCode, output)

def handleCommonOptions(pargs, subs):
    if pargs.enable_assertions:
        subs['ASSERTS'] = 'y'
        logging.info('Enabling assertions')
    else:
        logging.info('Disabling assertions')
        subs['ASSERTS'] = 'n'

    if pargs.enable_release:
        logging.info('Configuring for Release build')
        subs['DEBUG'] = 'n'
    else:
        logging.info('Configuring for Debug build')
        subs['DEBUG'] = 'y'

def handleNativeConfig(pargs, cc=None):
    logging.info('Configuring for native archive')

    subs = { 'EMIT_LLVM':'', # Building natively so we don't want LLVM Bitcode
             'TOOLDIR':'' # Native compiler tools should be in PATH
           }
    handleCommonOptions(pargs, subs)


    def searchPath(cc, lookFor):
        if not cc:
            logging.info('Looking for...{}'.format(lookFor))
            ccNew = shutil.which(lookFor)

            if ccNew:
                logging.info('Found...{}'.format(ccNew))
            else:
                logging.info('Could not find {}'.format(lookFor))

            return ccNew

        return cc


    # Detect compiler if not forced
    if not cc:
        cc=searchPath(cc, 'clang')
        cc=searchPath(cc, 'cc')

    # Test compiler
    #FIXME
    if cc is None:
        logging.error('Could not find a C compiler')
        sys.exit(1)

    logging.info('Using CC...{}'.format(cc))
    subs['CC'] = cc
    subs['HOSTCC'] = cc

    if not checkForNCurses(subs['HOSTCC']):
        sys.exit(1)

    # Check tools are present
    tools = { 'LINKER':'ld',
              'ARCHIVER':'ar',
              'NM':'nm',
              'OBJDUMP':'objdump'
            }
    for (name, executable) in tools.items():
        if not shutil.which(executable):
            logging.error('Could not find {} in PATH'.format(executable))
            sys.exit(1)

        subs[name] = executable

    # Do substitution
    logging.debug('Using template substitutions \n{} \nfor file "{}"'.format(
                  pprint.pformat(subs),
                  templateTarget)
                 )

    doTemplate(subs, templateFile, templateTarget)

def handleLLVMConfig(pargs, cc=None):
    # Substitutions
    subs = { 'EMIT_LLVM':'-emit-llvm'}
    handleCommonOptions(pargs, subs)

    logging.info('Configuring for LLVM bitcode archive')

    llvmConfigTool = pargs.with_llvm_config
    if not llvmConfigTool:
        llvmConfigTool = shutil.which('llvm-config')
    else:
        if os.sep not in llvmConfigTool:
            llvmConfigTool = shutil.which(llvmConfigTool)
        else:
            llvmConfigTool = os.path.abspath(llvmConfigTool)
            if not os.path.exists(llvmConfigTool):
                logging.error('"{}" does not exist. Cannot use as llvm-config tool'.format(llvmConfigTool))
                sys.exit(1)

    logging.info('Using llvm-config at...{}'.format(llvmConfigTool))

    if llvmConfigTool is None:
        logging.error('llvm-config cannot be found')
        sys.exit(1)

    # Detect tool directory
    (retCode, output) = runTool([llvmConfigTool, '--bindir'])
    if retCode != 0:
        logging.error('Failed to execute llvm-config:\n{}'.format(output))
        sys.exit(1)

    llvmToolDir = output.replace('\n','').lstrip().rstrip()
    if not os.path.exists(llvmToolDir):
        logging.error('The tool directory ({})reported by llvm-config does not exist.'.format(llvmToolDir))
        sys.exit(1)

    logging.info('Using llvm tool dir...{}'.format(llvmToolDir))
    subs['TOOLDIR'] = llvmToolDir + os.sep # Trailing slash is needed in Makefile

    # Check for the needed llvm tools
    llvmDeps = {
        'NM' : ['llvm-nm'],
        'ARCHIVER' : ['llvm-ar'],
        'LINKER': ['llvm-link', 'llvm-ld' ], # Two possible tools to use as linker
        'OBJDUMP': ['llvm-objdump']
    }

    for (name,toolNames) in llvmDeps.items():
        found=False
        for t in toolNames:
            absToolPath = os.path.join( llvmToolDir, t)
            if os.path.exists(absToolPath):
                logging.info('Found "{}".'.format(absToolPath))
                found=True
                subs[name] = t # Add substitution
                break

        if not found:
            logging.error('Could not find needed tool. Tried {}'.format(llvmDeps))
            sys.exit(1)

    if not cc:
        cc = findBitCodeCompiler(llvmToolDir)
        if not cc:
            logging.error('Failed to find a working LLVM bitcode compiler')
            sys.exit(1)
        logging.info('Using LLVM Bitcode Compiler...{}'.format(cc))
    else:
        logging.info('Using LLVM Bitcode Compiler specified by CC ...{}'.format(cc))
        if not testBitCodeCompiler(cc, llvmToolDir):
            logging.error('LLVM Bitcode compiler does not work')
            sys.exit(1)

    # Add compiler to substitutions
    subs['CC'] = cc
    subs['HOSTCC'] = cc

    if not checkForNCurses(subs['HOSTCC']):
        sys.exit(1)

    logging.debug('Using template substitutions \n{} \nfor file "{}"'.format(
                  pprint.pformat(subs),
                  templateTarget)
                 )

    doTemplate(subs, templateFile, templateTarget)

def doTemplate(subs, src, dest):
    """
        Do a Template substitution using @KEY@ syntax.

        subs : Dictionary of substitutions
        src  : File to apply template to
        dest : The destination for templated file.
    """

    # Remove old destination
    if os.path.exists(dest):
        logging.info('Removing template destination "{}"'.format(dest))
        os.remove(dest)

    if not os.path.exists(src):
        logging.error('Template source "{}" does not exist'.format(src))
        sys.exit(1)

    # Read file into string and do replacements
    with open(src,'r') as f:
        srcString = f.read()

        # Do replacements
        for (oldString, replacement) in subs.items():
            srcString = srcString.replace('@' + oldString + '@', replacement)

    # Write templated string to file
    logging.info('Writing templated file to "{}"'.format(dest))
    with open(dest,'w') as f:
        f.write(srcString)

def findBitCodeCompiler(llvmToolDir):
    """
        Search for LLVM Bitcode compiler.

        Returns absolute path to compiler. The compiler will
        be tested against the tools in llvmToolDir
    """
    logging.info('Searching for LLVM Bitcode compiler...')

    # First try clang in LLVM Build directory
    ccPath = os.path.join(llvmToolDir, 'clang')
    if os.path.exists( ccPath):
        logging.info('Found clang in LLVM Build dir...{}'.format(ccPath))

        if not testBitCodeCompiler(ccPath, llvmToolDir):
            ccPath=None
    else:
        ccPath=None

    # If that failed try clang in PATH
    if not ccPath:
        clang = shutil.which('clang')
        if clang:
            logging.info('Found clang in PATH...{}'.format(clang))
            if testBitCodeCompiler(clang, llvmToolDir):
                ccPath = clang

    return ccPath

def testBitCodeCompiler(cc, llvmToolDir):
    """
        Returns true if cc is working LLVM bitcode
        compiler.

        This works by compiling a small program to LLVM bitcode using cc
        then this is converted back to LLVM assembly using llvm-dis.
        If there is an incompatibility between llvm-dis and the bitcode
        compiler it will hopefully be detected here.

        cc : Absolute path to bitcode compiler
    """
    import tempfile
    logging.info('Testing LLVM Bitcode compiler...{}'.format(cc))

    # Get temporary file name for output
    bitCodeFileName=None
    cProgramFileName=None
    llvmAsFileName=None

    def cleanUp():
        if bitCodeFileName and os.path.exists(bitCodeFileName): os.remove(bitCodeFileName)
        if cProgramFileName and os.path.exists(cProgramFileName): os.remove(cProgramFileName)
        if llvmAsFileName and os.path.exists(llvmAsFileName): os.remove(llvmAsFileName)

    try:
        with tempfile.NamedTemporaryFile(suffix='.bc', delete=False) as f:
            bitCodeFileName = f.name

        if not bitCodeFileName:
            logging.error('Failed to generated temporary file name')
            sys.exit(1)

        with tempfile.NamedTemporaryFile(mode='w', suffix='.c', delete=False) as f:
            # Write a simple C program to tempfile
            f.write('int main() { return 0; }')
            cProgramFileName = f.name

        if not cProgramFileName:
            msg='Failed to generated temporary file name'
            logging.error(msg)
            raise Exception(msg)

        (retCode, ccOutput) = runTool([cc,
                                       '-c',
                                       '-g',
                                       '-emit-llvm',
                                       cProgramFileName,
                                       '-o', bitCodeFileName]
                                      )

        if retCode != 0:
            logging.info('Compiler failed with output:\n{}'.format(ccOutput))
            cleanUp()
            return False

        logging.debug('{} succeeded'.format(cc))
        llvmDis = os.path.join(llvmToolDir, 'llvm-dis')

        if not os.path.exists(llvmDis):
            logging.error('Cannot find "{}"'.format(llvmDis))
            cleanUp()
            return False

        # Get temporary file name for LLVM assembly
        with tempfile.NamedTemporaryFile(suffix='.ll', delete=False) as f:
            llvmAsFileName = f.name

        if not llvmAsFileName:
            msg='Failed to generated temporary file name'
            logging.error(msg)
            raise Exception(msg)

        (retCode, llvmDisOutput) = runTool([ llvmDis, '-o=' + llvmAsFileName, bitCodeFileName])

        if retCode != 0:
            logging.info('Conversion of LLVM Bitcode to LLVM Assembly failed with output:\n{}'.format(llvmDisOutput))
            cleanUp()
            return False

    except Exception:
        cleanUp()
        raise

    cleanUp()
    logging.info('Compiler {} works'.format(cc))
    return True

def checkForNCurses(cc):
    """
        uClibc needs ncurses for make menuconfig
        This function returns true if it is available.
        Other wise it returns false.

        cc is the compiler to use
    """
    import tempfile

    # Simple ncurses program
    src="""
#include <ncurses.h>

int main()
{
    initscr();
    return 0;
}

    """
    logging.info('Checking for ncurses...')
    outputName='a.out'
    with tempfile.NamedTemporaryFile(mode='w+', suffix='.c') as f:
        f.write(src)
        f.flush()
        (returnCode, output) = runTool([cc, f.name, '-lncurses', '-o', outputName])

        if returnCode == 0:
            os.remove(outputName)
            return True

        logging.error('Failed to find ncurses. Compiler said:\n{}'.format(output))
        logging.error('You should install the ncurses library and development headers.')
        return False


if __name__ == '__main__':
    main(sys.argv[1:])