kmap.py

#!/usr/bin/env python3

"""
kmap.py - Kernel Source-to-Binary Mapping Tool

DESCRIPTION
    This tool parses Linux kernel build artifacts (.cmd files, archives) to
    create a JSON mapping of which source files contribute to which kernel
    binaries (modules and vmlinux). It supports merging data from multiple
    kernel build variants (e.g., stock, debug, rt) into a single output
    file using variant indices to track per-variant mappings.

    Supported RHEL versions: 7, 8, 9, upstream

FEATURES
    - Parses .cmd files generated by kbuild to extract compilation commands
    - Supports C (gcc/clang) and Rust (rustc) source files
    - Handles built-in objects via built-in.o.cmd (RHEL 7), built-in.a
      (RHEL 8/9), or vmlinux.a (upstream)
    - Merges multiple kernel variants into single output with variant indices
    - Maps modules to their containing RPM packages

OUTPUT FORMAT
    The output is a JSON file with the following structure:

    {
      "variants": ["stock", "rt", ...],
      "source-map": {
        "obj-src": {<object>: {<source>: [<variant_indices>]}},
        "src-obj": {<source>: {<object>: [<variant_indices>]}}
      },
      "module-map": {
        "module-rpm": {<module>: {<rpm_name>: [<variant_indices>]}},
        "rpm-modules": {<rpm_name>: {<module>: [<variant_indices>]}}
      }
    }

PARSING THE OUTPUT
    import json

    with open('kernel-map.json') as f:
        data = json.load(f)

    variants = data['variants']
    obj_src = data['source-map']['obj-src']
    src_obj = data['source-map']['src-obj']
    module_rpm = data['module-map']['module-rpm']
    rpm_modules = data['module-map']['rpm-modules']

    # Get sources for vmlinux in variant "stock" (index 0)
    vmlinux_sources = [src for src, indices in obj_src['vmlinux'].items()
                       if 0 in indices]

    # Find which object a source contributes to
    fork_objects = [obj for obj, indices in src_obj['kernel/fork.c'].items()
                    if 0 in indices]

    # Get RPM for a module
    def get_rpm(module, variant_idx):
        for rpm, indices in module_rpm.get(module, {}).items():
            if variant_idx in indices:
                return rpm
        return None

USAGE
    # Single variant
    ./kmap.py -d /build/stock -r upstream -v stock -o /output

    # Merge multiple variants
    ./kmap.py -d /build/stock -r upstream -v stock -o /output
    ./kmap.py -d /build/debug -r upstream -v debug -o /output -i kernel-map.json
    ./kmap.py -d /build/rt -r upstream -v rt -o /output -i kernel-map.json

    # With module-to-RPM mapping
    ./kmap.py -d /build -r upstream -v stock -o /output \\
        --vmlinux-rpm kernel-core-5.14.0.rpm \\
        --module-list kernel-modules-5.14.0.rpm:modules.list

CREDITS
    Inspired by gen_compile_commands.py and Joe Lawrence.
    Time measurement option added for Jan Stancek.
"""

import re
import os
import pathlib
import json
import argparse
import subprocess
import time


# =============================================================================
# Constants
# =============================================================================

VMLINUX = 'vmlinux'
KERNEL_PREFIX = 'kernel/'

# Supported compilers and linkers
COMPILERS = ('gcc', 'clang')
LINKER = 'ld'
RUST_COMPILER = 'rustc'


# =============================================================================
# Regex Patterns
#
# These patterns parse the kernel build .cmd files to extract:
# - What compiler/linker was used
# - What source files were compiled
# - What object files were linked
# =============================================================================

# Pattern for .ko.cmd files (module linking)
# Matches: cmd_path/to/module.ko := ld <params> -o <ko_name> <input_files>
# Note: upstream kernels use 'savedcmd_' prefix instead of 'cmd_'
_KO_CMD_PATTERN = (
    r'^(saved)?cmd_[^ ]*\.ko\s*:=\s*ld\s+'
    r'(?P<params>.*) -o (?P<ko_name>\S+) (?P<input_files>[^;]+?)\s*(?:;|$)'
)

# Pattern for .o.cmd files (object compilation)
# Matches multiple formats:
# - C/asm: cmd_*.o := gcc/clang <params> -o <output> <input>
# - Rust:  cmd_*.o := rustc <params> --emit=obj=<output> <input.rs>
# - Rust source mapping: source_*.o := <file.rs>
_O_CMD_PATTERN = (
    r'^(saved)?(?:cmd_|source_)[^ ]*\.o\s+:=\s+'
    r'(?:'
        r'(?P<prefix>(?:\S+=\S+\s+)*)?'
        r'(?P<command>gcc|clang|ld|rustc)\s+'
        r'(?P<params>.*?)'
        r'(?:'
            r'\s+-o\s+(?P<obj_out>\S+)\s+(?P<input_files>[^;]+?)'
            r'|'
            r'\s+--emit=obj=(?P<rust_obj_out>\S+).*?\s+(?P<rust_input_file>\S+\.rs)'
        r')'
        r'|'
        r'(?P<source_file>\S+\.rs)'
    r')\s*(?:;|$)'
)

ko_matcher = re.compile(_KO_CMD_PATTERN)
o_matcher = re.compile(_O_CMD_PATTERN)


# =============================================================================
# Utility Functions
# =============================================================================

def parse_args():
    """
    Parse command line arguments
    """
    parser = argparse.ArgumentParser(
            prog='Kernel Build Mapper',
            description='Scrape Linux Kernel .cmd files to map contribution '
                        'of individual source files to kernel binaries.',
            )

    parser.add_argument('-d', '--directory',
            help='Directory with kernel build artifacts.',
            default='',
            type=str)

    parser.add_argument('-o', '--outputdir',
            help='Directory where to put generated files.',
            default='',
            type=str)

    parser.add_argument('-p', '--prefix',
            help='Name prefix to the generated files.',
            default='',
            type=str)

    parser.add_argument('-r', '--rhel',
            help='Specify to what version of RHEL artifacts belong (upstream for RHEL 10+/kernel-ark)',
            choices=['7', '8', '9', 'upstream'],
            required=True)

    parser.add_argument('-m', '--no-modules',
            help='*SKIP* collecting information about modules',
            action='store_true')

    parser.add_argument('-b', '--no-builtin',
            help='*SKIP* collecting information about built-in',
            action='store_true')

    parser.add_argument('-t', '--time',
            help='Print elapsed time at the end',
            action='store_true')

    parser.add_argument('-v', '--variant',
            help='Name of this kernel variant (e.g., stock, debug, rt)',
            required=True,
            type=str)

    parser.add_argument('-i', '--input',
            help='Existing JSON file to merge with (from previous variant runs)',
            default='',
            type=str)

    parser.add_argument('--module-list',
            help='Module list in format <rpm-name>:<path> (can be specified multiple times)',
            action='append',
            default=[],
            metavar='RPM:PATH')

    parser.add_argument('--vmlinux-rpm',
            help='RPM package name for vmlinux (e.g., "kernel-5.14.0.rpm")',
            default='',
            type=str)

    return parser.parse_args()


def load_file(filename, build_dir=''):
    """
    Load a text file and return non-empty lines as a list
    """
    full_path = os.path.join(build_dir, filename)

    try:
        with open(full_path, 'rt', encoding='utf-8') as file:
            return [line.rstrip() for line in file if line.rstrip()]
    except FileNotFoundError:
        print(f'File "{full_path}" not found.')
        return []


def name_to_cmd(filename, suffix):
    """
    Convert object filename to corresponding .cmd script filename

    Example: kernel/fork.o -> kernel/.fork.o.cmd
    """
    path, _ = os.path.splitext(filename)
    dirname = os.path.dirname(path)
    basename = os.path.basename(path)

    return os.path.join(dirname, '.' + basename + suffix)


def strip_path_prefixes(path, build_dir):
    """
    Normalize path by removing build directory prefix and other artifacts
    """
    path = pathlib.Path(path)

    if build_dir:
        try:
            return str(path.resolve().relative_to(pathlib.Path(build_dir).resolve()))
        except ValueError:
            pass

    if path.is_absolute():
        path = pathlib.Path(*path.parts[1:])

    while path.parts and path.parts[0] == 'repos':
        path = pathlib.Path(*path.parts[1:])

    return str(path)


def is_external_source(path):
    """
    Check if path should be excluded from source mapping

    Filters out:
    - Rust toolchain sources (not part of kernel source)
    - Generated assembly files (e.g., initramfs_data.S)
    """
    if 'rustlib' in path:
        return True
    if path.endswith('_data.S'):
        return True
    return False


def is_mod_file_reference(ld_input):
    """
    Check if linker input is a .mod file reference (RHEL 9+ module linking)

    In RHEL 9+, module linking uses @path/to/module.mod files that contain
    the list of object files to link.
    """
    files = ld_input.split()
    return len(files) == 1 and files[0].startswith('@') and files[0].endswith('.mod')


# =============================================================================
# SourceMap Class
# =============================================================================

class SourceMap:
    """
    Collects and manages kernel source-to-binary mappings.

    This class parses kernel build artifacts (.cmd files, archives) to determine
    which source files contribute to which kernel binaries (modules and vmlinux).

    Data structures:
        obj_to_src: {object_name: {source_file: [variant_indices]}}
        src_to_obj: {source_file: {object_name: [variant_indices]}}
        module_rpm: {module_name: {rpm_name: [variant_indices]}}
        rpm_to_modules: {rpm_name: {module_name: [variant_indices]}}

    Variant indices correspond to positions in the 'variants' list.
    """

    # Dispatch table for RHEL-specific builtin collection methods
    BUILTIN_COLLECTORS = {
        '7': '_collect_builtin_rhel7',
        '8': '_collect_builtin_archive',
        '9': '_collect_builtin_archive',
        'upstream': '_collect_builtin_vmlinux_archive',
    }

    def __init__(self, build_dir='', output_dir='', outprefix='', rhel='7', variant=''):
        # Source mappings
        self.obj_to_src = {}
        self.src_to_obj = {}
        # Module-to-RPM mappings
        self.module_rpm = {}
        self.rpm_to_modules = {}

        # Configuration
        self.build_dir = build_dir
        self.output_dir = output_dir
        self.outprefix = outprefix
        self.rhel = rhel

        # Module list
        self.modlist = set()

        # Variant tracking for multi-variant merging
        self.variants = []
        self.variant = variant
        self.variant_idx = 0

    # -------------------------------------------------------------------------
    # Variant Management
    # -------------------------------------------------------------------------

    def load_existing(self, filename):
        """
        Load existing merged data from previous variant runs
        """
        if not filename:
            return

        full_path = os.path.join(self.output_dir, filename) if self.output_dir else filename

        if not os.path.exists(full_path):
            return

        with open(full_path, 'rt', encoding='utf-8') as f:
            data = json.load(f)

        self.variants = data.get('variants', [])

        module_map = data.get('module-map', {})
        self.module_rpm = module_map.get('module-rpm', {})
        self.rpm_to_modules = module_map.get('rpm-modules', {})

        source_map = data.get('source-map', {})
        self.obj_to_src = source_map.get('obj-src', {})
        self.src_to_obj = source_map.get('src-obj', {})

    def add_variant(self):
        """
        Register current variant and store its index
        """
        if self.variant in self.variants:
            self.variant_idx = self.variants.index(self.variant)
        else:
            self.variant_idx = len(self.variants)
            self.variants.append(self.variant)

    # -------------------------------------------------------------------------
    # Source Merging Helpers
    # -------------------------------------------------------------------------

    def _merge_sources(self, target, source):
        """
        Merge source dict into target dict by combining variant index lists
        """
        for src, indices in source.items():
            if src not in target:
                target[src] = []
            for idx in indices:
                if idx not in target[src]:
                    target[src].append(idx)

    def _add_source(self, sources, source_path):
        """
        Add a source file to the sources dict with current variant index

        Returns True if source was added, False if filtered out
        """
        source_path = os.path.normpath(source_path)
        source_path = strip_path_prefixes(source_path, self.build_dir)

        if is_external_source(source_path):
            return False

        sources[source_path] = [self.variant_idx]
        return True

    # -------------------------------------------------------------------------
    # .cmd File Parsing
    # -------------------------------------------------------------------------

    def _parse_ko_cmd(self, ko_name):
        """
        Parse .ko.cmd file to find the primary object file linked into the module
        """
        ko_script = name_to_cmd(ko_name, '.ko.cmd')
        lines = load_file(ko_script, build_dir=self.build_dir)

        for line in lines:
            result = ko_matcher.match(line)
            if result:
                input_files = result.group('input_files').split()
                return input_files[0]

        return None

    def _parse_o_cmd(self, o_name, translate=True):
        """
        Parse .o.cmd file to extract source files that contribute to the object

        Handles:
        - C/assembly compilation (gcc, clang)
        - Linking (ld) - recursively parses linked objects
        - Rust compilation (rustc)
        - Rust source file mappings

        Args:
            o_name: Object file name or .cmd file path
            translate: If True, convert object name to .cmd path (e.g., kernel/fork.o
                       becomes kernel/.fork.o.cmd). If False, use o_name as-is.
                       This parameter exists to handle different kbuild behaviors
                       across RHEL versions - some code paths may already have .cmd
                       paths while others have object names that need translation.
        """
        sources = {}

        o_script = name_to_cmd(o_name, '.o.cmd') if translate else o_name
        lines = load_file(o_script, build_dir=self.build_dir)

        for line in lines:
            result = o_matcher.match(line)
            if not result:
                continue

            command = result.group('command')

            if command in COMPILERS:
                self._handle_compiler(result, sources)
                break

            elif command == LINKER:
                self._handle_linker(result, sources)
                break

            elif command == RUST_COMPILER:
                self._handle_rustc(result, sources)
                break

            else:
                self._handle_rust_source_mapping(result, sources)
                break

        return sources

    def _handle_compiler(self, match_result, sources):
        """
        Handle gcc/clang compilation - extract the input source file
        """
        obj_input = match_result.group('input_files')
        if obj_input:
            self._add_source(sources, obj_input)

    def _handle_linker(self, match_result, sources):
        """
        Handle ld linking - recursively parse linked objects
        """
        obj_input = match_result.group('input_files')
        if not obj_input:
            return

        if is_mod_file_reference(obj_input):
            self._merge_sources(sources, self._parse_mod_cmd(obj_input))
        else:
            for obj_file in obj_input.split():
                self._merge_sources(sources, self._parse_o_cmd(obj_file))

    def _handle_rustc(self, match_result, sources):
        """
        Handle rustc compilation - extract the Rust source file
        """
        rust_input = match_result.group('rust_input_file')
        if rust_input:
            self._add_source(sources, rust_input)

    def _handle_rust_source_mapping(self, match_result, sources):
        """
        Handle Rust source file mapping (source_*.o := *.rs)
        """
        source_file = match_result.group('source_file')
        if source_file and source_file.endswith('.rs'):
            self._add_source(sources, source_file)

    def _parse_mod_cmd(self, mod_name):
        """
        Parse .mod file used by RHEL 9+ for linking modules

        These files contain a list of object files to link into the module.
        """
        sources = {}
        mod_name = mod_name.strip('@')
        lines = load_file(mod_name, build_dir=self.build_dir)

        for line in lines:
            self._merge_sources(sources, self._parse_o_cmd(line))

        return sources

    def _parse_a_cmd(self, a_name):
        """
        Parse archive .cmd file (RHEL 7 built-in.o.cmd files)
        """
        sources = {}
        lines = load_file(a_name, build_dir=self.build_dir)

        for line in lines:
            result = o_matcher.match(line)
            if not result:
                continue

            ar_input = result.group('input_files')
            for filename in ar_input.rsplit():
                if filename.endswith('.o'):
                    self._merge_sources(sources, self._parse_o_cmd(filename))
                elif filename.endswith('.a'):
                    self._merge_sources(sources, self._parse_a_cmd(filename))

        return sources

    # -------------------------------------------------------------------------
    # Module Collection
    # -------------------------------------------------------------------------

    def _load_module_list(self, modorder='modules.order'):
        """
        Load modules.order file and normalize paths

        RHEL 7, 8: modules.order has 'kernel/' prefix that needs to be stripped
        RHEL 9, upstream: modules.order paths are used as-is
        """
        lines = load_file(modorder, build_dir=self.build_dir)

        if not lines:
            print(f'Warning: modules.order is empty or not found')
            self.modlist = set()
            return

        if self.rhel in ['9', 'upstream']:
            self.modlist = set(lines)
        else:
            # RHEL 7, 8: strip fake 'kernel/' prefix added by kbuild
            self.modlist = {
                path[len(KERNEL_PREFIX):] if path.startswith(KERNEL_PREFIX) else path
                for path in lines
            }

    def _parse_module_objects(self):
        """
        Parse all module .cmd files and extract source mappings.

        Note on RHEL version differences:
        - RHEL 7/8/9: modules.order contains .ko paths (e.g., fs/exfat/exfat.ko)
        - upstream: modules.order contains .o paths (e.g., fs/exfat/exfat.o)

        The normalization below ensures consistent .ko keys in output regardless
        of the modules.order format.
        """
        for kobj in self.modlist:
            # Normalize to .ko - upstream has .o paths in modules.order
            if self.rhel == 'upstream' and kobj.endswith('.o'):
                module_name = kobj[:-2] + '.ko'
            else:
                module_name = kobj

            obj_name = self._parse_ko_cmd(module_name)
            if obj_name is None:
                print(f'Warning: Could not parse .ko.cmd for "{module_name}", skipping.')
                continue

            obj_source = self._parse_o_cmd(obj_name)
            if module_name not in self.obj_to_src:
                self.obj_to_src[module_name] = {}
            self._merge_sources(self.obj_to_src[module_name], obj_source)

    def collect_modules(self, no_modules=False, modorder='modules.order'):
        """
        Collect source files for kernel modules
        """
        if no_modules:
            return

        self._load_module_list(modorder)
        self._parse_module_objects()

    # -------------------------------------------------------------------------
    # Builtin Collection (RHEL version-specific)
    # -------------------------------------------------------------------------

    def _collect_builtin_rhel7(self):
        """
        Collect built-in objects for RHEL 7 using *built-in.o.cmd files
        """
        pattern = '*built-in.o.cmd'
        cmd_list = {
            strip_path_prefixes(path, self.build_dir)
            for path in pathlib.Path(self.build_dir).rglob(pattern)
        }

        for entry in cmd_list:
            sources = self._parse_a_cmd(entry)
            self._merge_sources(self.obj_to_src[VMLINUX], sources)

    def _collect_builtin_archive(self):
        """
        Collect built-in objects for RHEL 8/9 using *built-in.a archives
        """
        pattern = '*built-in.a'
        archive_list = {
            strip_path_prefixes(path, self.build_dir)
            for path in pathlib.Path(self.build_dir).rglob(pattern)
        }

        for entry in archive_list:
            path = os.path.join(self.build_dir, entry)
            objects = subprocess.check_output(['ar', 't', path]).decode().split()

            for obj in objects:
                obj_path = strip_path_prefixes(obj, build_dir=self.build_dir)
                sources = self._parse_o_cmd(obj_path)
                self._merge_sources(self.obj_to_src[VMLINUX], sources)

    def _collect_builtin_vmlinux_archive(self):
        """
        Collect built-in objects for upstream kernels using single vmlinux.a archive
        """
        path = os.path.join(self.build_dir, 'vmlinux.a')
        objects = subprocess.check_output(['ar', 't', path]).decode().split()

        for obj in objects:
            obj_path = strip_path_prefixes(obj, build_dir=self.build_dir)
            sources = self._parse_o_cmd(obj_path)
            self._merge_sources(self.obj_to_src[VMLINUX], sources)

    def collect_builtin(self, no_builtin=False):
        """
        Collect source files for vmlinux built-in objects

        Uses RHEL version-specific collection method via dispatch table.
        """
        if VMLINUX not in self.obj_to_src:
            self.obj_to_src[VMLINUX] = {}

        if no_builtin:
            return

        collector_name = self.BUILTIN_COLLECTORS.get(self.rhel, '_collect_builtin_archive')
        collector = getattr(self, collector_name)
        collector()

    # -------------------------------------------------------------------------
    # Module-to-RPM Mapping
    # -------------------------------------------------------------------------

    def _parse_module_list_file(self, filename, rpm_name):
        """
        Parse a module list file and add entries to module_rpm mapping

        Handles various formats:
        - Absolute paths: /lib/modules/5.14.0/kernel/drivers/net/e1000.ko.xz
        - Relative paths: kernel/drivers/net/e1000.ko
        - Skips RPM spec directives (%dir, %defattr, etc.)

        Args:
            filename: Path to the module list file
            rpm_name: Literal RPM package name (e.g., "kernel-modules-5.14.0.rpm")
        """
        if not filename or not os.path.exists(filename):
            return

        with open(filename, 'rt', encoding='utf-8') as f:
            for line in f:
                line = line.strip()

                # Skip empty lines, comments, and RPM spec directives
                if not line or line.startswith('#') or line.startswith('%'):
                    continue

                # Skip non-module files
                if '.ko' not in line:
                    continue

                # Extract module name from path
                module = os.path.basename(line)
                # Remove .ko and any compression suffix
                module = re.sub(r'\.ko(\.[^.]+)?$', '', module)

                if not module:
                    continue

                if module not in self.module_rpm:
                    self.module_rpm[module] = {}

                if rpm_name not in self.module_rpm[module]:
                    self.module_rpm[module][rpm_name] = []
                if self.variant_idx not in self.module_rpm[module][rpm_name]:
                    self.module_rpm[module][rpm_name].append(self.variant_idx)

    def collect_module_rpm(self, module_lists, vmlinux_rpm=''):
        """
        Collect module-to-RPM name mappings from module list files

        Args:
            module_lists: List of "rpm_name:path" strings
            vmlinux_rpm: RPM package name for vmlinux
        """
        if not module_lists and not vmlinux_rpm:
            return

        # Add vmlinux to specified RPM
        if vmlinux_rpm:
            if VMLINUX not in self.module_rpm:
                self.module_rpm[VMLINUX] = {}
            if vmlinux_rpm not in self.module_rpm[VMLINUX]:
                self.module_rpm[VMLINUX][vmlinux_rpm] = []
            if self.variant_idx not in self.module_rpm[VMLINUX][vmlinux_rpm]:
                self.module_rpm[VMLINUX][vmlinux_rpm].append(self.variant_idx)

        # Process module list files
        for entry in module_lists:
            if ':' not in entry:
                print(f'Warning: Invalid --module-list format "{entry}", expected RPM:PATH')
                continue

            rpm_name, path = entry.split(':', 1)
            self._parse_module_list_file(path, rpm_name)

    # -------------------------------------------------------------------------
    # Output
    # -------------------------------------------------------------------------

    def _build_reverse_map(self):
        """
        Create reverse mapping (source -> {obj: [indices]}) from obj_to_src
        """
        for obj, sources in self.obj_to_src.items():
            for source, indices in sources.items():
                if source not in self.src_to_obj:
                    self.src_to_obj[source] = {}
                if obj not in self.src_to_obj[source]:
                    self.src_to_obj[source][obj] = []
                for idx in indices:
                    if idx not in self.src_to_obj[source][obj]:
                        self.src_to_obj[source][obj].append(idx)

    def _build_rpm_reverse_map(self):
        """
        Create reverse mapping (rpm -> {module: [indices]}) from module_rpm
        """
        for module, rpms in self.module_rpm.items():
            for rpm, indices in rpms.items():
                if rpm not in self.rpm_to_modules:
                    self.rpm_to_modules[rpm] = {}
                if module not in self.rpm_to_modules[rpm]:
                    self.rpm_to_modules[rpm][module] = []
                for idx in indices:
                    if idx not in self.rpm_to_modules[rpm][module]:
                        self.rpm_to_modules[rpm][module].append(idx)

    def save_data(self):
        """
        Save collected data to JSON file

        Output structure:
          - variants: list of variant names
          - source-map: source file mappings
            - obj-src: object/module -> {source: [variant_indices]}
            - src-obj: source -> {object/module: [variant_indices]}
          - module-map: module to RPM mappings
            - module-rpm: module name -> {rpm-name: [variant_indices]}
            - rpm-modules: rpm-name -> {module name: [variant_indices]}

        Variant indices correspond to positions in the 'variants' list.
        """
        output = {
            'variants': self.variants,
            'source-map': {
                'obj-src': self.obj_to_src,
                'src-obj': self.src_to_obj
            },
            'module-map': {
                'module-rpm': self.module_rpm,
                'rpm-modules': self.rpm_to_modules
            }
        }

        filename = self.outprefix + 'kernel-map.json'
        with open(os.path.join(self.output_dir, filename), 'wt', encoding='utf-8') as f:
            json.dump(output, f, indent=2, sort_keys=True)

    # -------------------------------------------------------------------------
    # Main Entry Point
    # -------------------------------------------------------------------------

    def collect_data(self, no_modules=False, no_builtin=False, modorder='modules.order',
                     input_file='', module_lists=None, vmlinux_rpm=''):
        """
        Main entry point - collect all data and save to JSON
        """
        self.load_existing(input_file)
        self.add_variant()
        self.collect_modules(no_modules, modorder)
        self.collect_builtin(no_builtin)
        self.collect_module_rpm(module_lists or [], vmlinux_rpm)
        self._build_reverse_map()
        self._build_rpm_reverse_map()
        self.save_data()


# =============================================================================
# Main
# =============================================================================

def main():
    """
    Main entry point
    """
    args = parse_args()

    start = time.monotonic()

    source_map = SourceMap(
            build_dir=args.directory,
            output_dir=args.outputdir,
            outprefix=args.prefix,
            rhel=args.rhel,
            variant=args.variant)

    source_map.collect_data(
            no_modules=args.no_modules,
            no_builtin=args.no_builtin,
            input_file=args.input,
            module_lists=args.module_list,
            vmlinux_rpm=args.vmlinux_rpm)

    if args.time:
        elapsed = time.monotonic() - start
        minutes, seconds = divmod(int(elapsed), 60)
        print(f'Elapsed time: {minutes:02d}:{seconds:02d}')


if __name__ == '__main__':
    main()