Implement storedevs, prepare to port littlefs

kernel/fs/littlefs/scripts/changeprefix.py

@@ -0,0 +1,181 @@
+#!/usr/bin/env python3
+#
+# Change prefixes in files/filenames. Useful for creating different versions
+# of a codebase that don't conflict at compile time.
+#
+# Example:
+# $ ./scripts/changeprefix.py lfs lfs3
+#
+# Copyright (c) 2022, The littlefs authors.
+# Copyright (c) 2019, Arm Limited. All rights reserved.
+# SPDX-License-Identifier: BSD-3-Clause
+#
+
+import glob
+import itertools
+import os
+import os.path
+import re
+import shlex
+import shutil
+import subprocess
+import tempfile
+
+GIT_PATH = ['git']
+
+
+def openio(path, mode='r', buffering=-1):
+    # allow '-' for stdin/stdout
+    if path == '-':
+        if mode == 'r':
+            return os.fdopen(os.dup(sys.stdin.fileno()), mode, buffering)
+        else:
+            return os.fdopen(os.dup(sys.stdout.fileno()), mode, buffering)
+    else:
+        return open(path, mode, buffering)
+
+def changeprefix(from_prefix, to_prefix, line):
+    line, count1 = re.subn(
+        '\\b'+from_prefix,
+        to_prefix,
+        line)
+    line, count2 = re.subn(
+        '\\b'+from_prefix.upper(),
+        to_prefix.upper(),
+        line)
+    line, count3 = re.subn(
+        '\\B-D'+from_prefix.upper(),
+        '-D'+to_prefix.upper(),
+        line)
+    return line, count1+count2+count3
+
+def changefile(from_prefix, to_prefix, from_path, to_path, *,
+        no_replacements=False):
+    # rename any prefixes in file
+    count = 0
+
+    # create a temporary file to avoid overwriting ourself
+    if from_path == to_path and to_path != '-':
+        to_path_temp = tempfile.NamedTemporaryFile('w', delete=False)
+        to_path = to_path_temp.name
+    else:
+        to_path_temp = None
+
+    with openio(from_path) as from_f:
+        with openio(to_path, 'w') as to_f:
+            for line in from_f:
+                if not no_replacements:
+                    line, n = changeprefix(from_prefix, to_prefix, line)
+                    count += n
+                to_f.write(line)
+
+    if from_path != '-' and to_path != '-':
+        shutil.copystat(from_path, to_path)
+
+    if to_path_temp:
+        shutil.move(to_path, from_path)
+    elif from_path != '-':
+        os.remove(from_path)
+
+    # Summary
+    print('%s: %d replacements' % (
+        '%s -> %s' % (from_path, to_path) if not to_path_temp else from_path,
+        count))
+
+def main(from_prefix, to_prefix, paths=[], *,
+        verbose=False,
+        output=None,
+        no_replacements=False,
+        no_renames=False,
+        git=False,
+        no_stage=False,
+        git_path=GIT_PATH):
+    if not paths:
+        if git:
+            cmd = git_path + ['ls-tree', '-r', '--name-only', 'HEAD']
+            if verbose:
+                print(' '.join(shlex.quote(c) for c in cmd))
+            paths = subprocess.check_output(cmd, encoding='utf8').split()
+        else:
+            print('no paths?', file=sys.stderr)
+            sys.exit(1)
+        
+    for from_path in paths:
+        # rename filename?
+        if output:
+            to_path = output
+        elif no_renames:
+            to_path = from_path
+        else:
+            to_path = os.path.join(
+                os.path.dirname(from_path),
+                changeprefix(from_prefix, to_prefix,
+                    os.path.basename(from_path))[0])
+
+        # rename contents
+        changefile(from_prefix, to_prefix, from_path, to_path,
+            no_replacements=no_replacements)
+
+        # stage?
+        if git and not no_stage:
+            if from_path != to_path:
+                cmd = git_path + ['rm', '-q', from_path]
+                if verbose:
+                    print(' '.join(shlex.quote(c) for c in cmd))
+                subprocess.check_call(cmd)
+            cmd = git_path + ['add', to_path]
+            if verbose:
+                print(' '.join(shlex.quote(c) for c in cmd))
+            subprocess.check_call(cmd)
+
+
+if __name__ == "__main__":
+    import argparse
+    import sys
+    parser = argparse.ArgumentParser(
+        description="Change prefixes in files/filenames. Useful for creating "
+            "different versions of a codebase that don't conflict at compile "
+            "time.",
+        allow_abbrev=False)
+    parser.add_argument(
+        'from_prefix',
+        help="Prefix to replace.")
+    parser.add_argument(
+        'to_prefix',
+        help="Prefix to replace with.")
+    parser.add_argument(
+        'paths',
+        nargs='*',
+        help="Files to operate on.")
+    parser.add_argument(
+        '-v', '--verbose',
+        action='store_true',
+        help="Output commands that run behind the scenes.")
+    parser.add_argument(
+        '-o', '--output',
+        help="Output file.")
+    parser.add_argument(
+        '-N', '--no-replacements',
+        action='store_true',
+        help="Don't change prefixes in files")
+    parser.add_argument(
+        '-R', '--no-renames',
+        action='store_true',
+        help="Don't rename files")
+    parser.add_argument(
+        '--git',
+        action='store_true',
+        help="Use git to find/update files.")
+    parser.add_argument(
+        '--no-stage',
+        action='store_true',
+        help="Don't stage changes with git.")
+    parser.add_argument(
+        '--git-path',
+        type=lambda x: x.split(),
+        default=GIT_PATH,
+        help="Path to git executable, may include flags. "
+            "Defaults to %r." % GIT_PATH)
+    sys.exit(main(**{k: v
+        for k, v in vars(parser.parse_intermixed_args()).items()
+        if v is not None}))

kernel/fs/littlefs/scripts/code.py

@@ -0,0 +1,707 @@
+#!/usr/bin/env python3
+#
+# Script to find code size at the function level. Basically just a big wrapper
+# around nm with some extra conveniences for comparing builds. Heavily inspired
+# by Linux's Bloat-O-Meter.
+#
+# Example:
+# ./scripts/code.py lfs.o lfs_util.o -Ssize
+#
+# Copyright (c) 2022, The littlefs authors.
+# Copyright (c) 2020, Arm Limited. All rights reserved.
+# SPDX-License-Identifier: BSD-3-Clause
+#
+
+import collections as co
+import csv
+import difflib
+import itertools as it
+import math as m
+import os
+import re
+import shlex
+import subprocess as sp
+
+
+NM_PATH = ['nm']
+NM_TYPES = 'tTrRdD'
+OBJDUMP_PATH = ['objdump']
+
+
+# integer fields
+class Int(co.namedtuple('Int', 'x')):
+    __slots__ = ()
+    def __new__(cls, x=0):
+        if isinstance(x, Int):
+            return x
+        if isinstance(x, str):
+            try:
+                x = int(x, 0)
+            except ValueError:
+                # also accept +-∞ and +-inf
+                if re.match('^\s*\+?\s*(?:∞|inf)\s*$', x):
+                    x = m.inf
+                elif re.match('^\s*-\s*(?:∞|inf)\s*$', x):
+                    x = -m.inf
+                else:
+                    raise
+        assert isinstance(x, int) or m.isinf(x), x
+        return super().__new__(cls, x)
+
+    def __str__(self):
+        if self.x == m.inf:
+            return '∞'
+        elif self.x == -m.inf:
+            return '-∞'
+        else:
+            return str(self.x)
+
+    def __int__(self):
+        assert not m.isinf(self.x)
+        return self.x
+
+    def __float__(self):
+        return float(self.x)
+
+    none = '%7s' % '-'
+    def table(self):
+        return '%7s' % (self,)
+
+    diff_none = '%7s' % '-'
+    diff_table = table
+
+    def diff_diff(self, other):
+        new = self.x if self else 0
+        old = other.x if other else 0
+        diff = new - old
+        if diff == +m.inf:
+            return '%7s' % '+∞'
+        elif diff == -m.inf:
+            return '%7s' % '-∞'
+        else:
+            return '%+7d' % diff
+
+    def ratio(self, other):
+        new = self.x if self else 0
+        old = other.x if other else 0
+        if m.isinf(new) and m.isinf(old):
+            return 0.0
+        elif m.isinf(new):
+            return +m.inf
+        elif m.isinf(old):
+            return -m.inf
+        elif not old and not new:
+            return 0.0
+        elif not old:
+            return 1.0
+        else:
+            return (new-old) / old
+
+    def __add__(self, other):
+        return self.__class__(self.x + other.x)
+
+    def __sub__(self, other):
+        return self.__class__(self.x - other.x)
+
+    def __mul__(self, other):
+        return self.__class__(self.x * other.x)
+
+# code size results
+class CodeResult(co.namedtuple('CodeResult', [
+        'file', 'function',
+        'size'])):
+    _by = ['file', 'function']
+    _fields = ['size']
+    _sort = ['size']
+    _types = {'size': Int}
+
+    __slots__ = ()
+    def __new__(cls, file='', function='', size=0):
+        return super().__new__(cls, file, function,
+            Int(size))
+
+    def __add__(self, other):
+        return CodeResult(self.file, self.function,
+            self.size + other.size)
+
+
+def openio(path, mode='r', buffering=-1):
+    # allow '-' for stdin/stdout
+    if path == '-':
+        if mode == 'r':
+            return os.fdopen(os.dup(sys.stdin.fileno()), mode, buffering)
+        else:
+            return os.fdopen(os.dup(sys.stdout.fileno()), mode, buffering)
+    else:
+        return open(path, mode, buffering)
+
+def collect(obj_paths, *,
+        nm_path=NM_PATH,
+        nm_types=NM_TYPES,
+        objdump_path=OBJDUMP_PATH,
+        sources=None,
+        everything=False,
+        **args):
+    size_pattern = re.compile(
+        '^(?P<size>[0-9a-fA-F]+)' +
+        ' (?P<type>[%s])' % re.escape(nm_types) +
+        ' (?P<func>.+?)$')
+    line_pattern = re.compile(
+        '^\s+(?P<no>[0-9]+)'
+            '(?:\s+(?P<dir>[0-9]+))?'
+            '\s+.*'
+            '\s+(?P<path>[^\s]+)$')
+    info_pattern = re.compile(
+        '^(?:.*(?P<tag>DW_TAG_[a-z_]+).*'
+            '|.*DW_AT_name.*:\s*(?P<name>[^:\s]+)\s*'
+            '|.*DW_AT_decl_file.*:\s*(?P<file>[0-9]+)\s*)$')
+
+    results = []
+    for path in obj_paths:
+        # guess the source, if we have debug-info we'll replace this later
+        file = re.sub('(\.o)?$', '.c', path, 1)
+
+        # find symbol sizes
+        results_ = []
+        # note nm-path may contain extra args
+        cmd = nm_path + ['--size-sort', path]
+        if args.get('verbose'):
+            print(' '.join(shlex.quote(c) for c in cmd))
+        proc = sp.Popen(cmd,
+            stdout=sp.PIPE,
+            stderr=sp.PIPE if not args.get('verbose') else None,
+            universal_newlines=True,
+            errors='replace',
+            close_fds=False)
+        for line in proc.stdout:
+            m = size_pattern.match(line)
+            if m:
+                func = m.group('func')
+                # discard internal functions
+                if not everything and func.startswith('__'):
+                    continue
+                results_.append(CodeResult(
+                    file, func,
+                    int(m.group('size'), 16)))
+        proc.wait()
+        if proc.returncode != 0:
+            if not args.get('verbose'):
+                for line in proc.stderr:
+                    sys.stdout.write(line)
+            sys.exit(-1)
+
+
+        # try to figure out the source file if we have debug-info
+        dirs = {}
+        files = {}
+        # note objdump-path may contain extra args
+        cmd = objdump_path + ['--dwarf=rawline', path]
+        if args.get('verbose'):
+            print(' '.join(shlex.quote(c) for c in cmd))
+        proc = sp.Popen(cmd,
+            stdout=sp.PIPE,
+            stderr=sp.PIPE if not args.get('verbose') else None,
+            universal_newlines=True,
+            errors='replace',
+            close_fds=False)
+        for line in proc.stdout:
+            # note that files contain references to dirs, which we
+            # dereference as soon as we see them as each file table follows a
+            # dir table
+            m = line_pattern.match(line)
+            if m:
+                if not m.group('dir'):
+                    # found a directory entry
+                    dirs[int(m.group('no'))] = m.group('path')
+                else:
+                    # found a file entry
+                    dir = int(m.group('dir'))
+                    if dir in dirs:
+                        files[int(m.group('no'))] = os.path.join(
+                            dirs[dir],
+                            m.group('path'))
+                    else:
+                        files[int(m.group('no'))] = m.group('path')
+        proc.wait()
+        if proc.returncode != 0:
+            if not args.get('verbose'):
+                for line in proc.stderr:
+                    sys.stdout.write(line)
+            # do nothing on error, we don't need objdump to work, source files
+            # may just be inaccurate
+            pass
+
+        defs = {}
+        is_func = False
+        f_name = None
+        f_file = None
+        # note objdump-path may contain extra args
+        cmd = objdump_path + ['--dwarf=info', path]
+        if args.get('verbose'):
+            print(' '.join(shlex.quote(c) for c in cmd))
+        proc = sp.Popen(cmd,
+            stdout=sp.PIPE,
+            stderr=sp.PIPE if not args.get('verbose') else None,
+            universal_newlines=True,
+            errors='replace',
+            close_fds=False)
+        for line in proc.stdout:
+            # state machine here to find definitions
+            m = info_pattern.match(line)
+            if m:
+                if m.group('tag'):
+                    if is_func:
+                        defs[f_name] = files.get(f_file, '?')
+                    is_func = (m.group('tag') == 'DW_TAG_subprogram')
+                elif m.group('name'):
+                    f_name = m.group('name')
+                elif m.group('file'):
+                    f_file = int(m.group('file'))
+        if is_func:
+            defs[f_name] = files.get(f_file, '?')
+        proc.wait()
+        if proc.returncode != 0:
+            if not args.get('verbose'):
+                for line in proc.stderr:
+                    sys.stdout.write(line)
+            # do nothing on error, we don't need objdump to work, source files
+            # may just be inaccurate
+            pass
+
+        for r in results_:
+            # find best matching debug symbol, this may be slightly different
+            # due to optimizations
+            if defs:
+                # exact match? avoid difflib if we can for speed
+                if r.function in defs:
+                    file = defs[r.function]
+                else:
+                    _, file = max(
+                        defs.items(),
+                        key=lambda d: difflib.SequenceMatcher(None,
+                            d[0],
+                            r.function, False).ratio())
+            else:
+                file = r.file
+
+            # ignore filtered sources
+            if sources is not None:
+                if not any(
+                        os.path.abspath(file) == os.path.abspath(s)
+                        for s in sources):
+                    continue
+            else:
+                # default to only cwd
+                if not everything and not os.path.commonpath([
+                        os.getcwd(),
+                        os.path.abspath(file)]) == os.getcwd():
+                    continue
+
+            # simplify path
+            if os.path.commonpath([
+                    os.getcwd(),
+                    os.path.abspath(file)]) == os.getcwd():
+                file = os.path.relpath(file)
+            else:
+                file = os.path.abspath(file)
+
+            results.append(r._replace(file=file))
+
+    return results
+
+
+def fold(Result, results, *,
+        by=None,
+        defines=None,
+        **_):
+    if by is None:
+        by = Result._by
+
+    for k in it.chain(by or [], (k for k, _ in defines or [])):
+        if k not in Result._by and k not in Result._fields:
+            print("error: could not find field %r?" % k)
+            sys.exit(-1)
+
+    # filter by matching defines
+    if defines is not None:
+        results_ = []
+        for r in results:
+            if all(getattr(r, k) in vs for k, vs in defines):
+                results_.append(r)
+        results = results_
+
+    # organize results into conflicts
+    folding = co.OrderedDict()
+    for r in results:
+        name = tuple(getattr(r, k) for k in by)
+        if name not in folding:
+            folding[name] = []
+        folding[name].append(r)
+
+    # merge conflicts
+    folded = []
+    for name, rs in folding.items():
+        folded.append(sum(rs[1:], start=rs[0]))
+
+    return folded
+
+def table(Result, results, diff_results=None, *,
+        by=None,
+        fields=None,
+        sort=None,
+        summary=False,
+        all=False,
+        percent=False,
+        **_):
+    all_, all = all, __builtins__.all
+
+    if by is None:
+        by = Result._by
+    if fields is None:
+        fields = Result._fields
+    types = Result._types
+
+    # fold again
+    results = fold(Result, results, by=by)
+    if diff_results is not None:
+        diff_results = fold(Result, diff_results, by=by)
+
+    # organize by name
+    table = {
+        ','.join(str(getattr(r, k) or '') for k in by): r
+        for r in results}
+    diff_table = {
+        ','.join(str(getattr(r, k) or '') for k in by): r
+        for r in diff_results or []}
+    names = list(table.keys() | diff_table.keys())
+
+    # sort again, now with diff info, note that python's sort is stable
+    names.sort()
+    if diff_results is not None:
+        names.sort(key=lambda n: tuple(
+            types[k].ratio(
+                getattr(table.get(n), k, None),
+                getattr(diff_table.get(n), k, None))
+            for k in fields),
+            reverse=True)
+    if sort:
+        for k, reverse in reversed(sort):
+            names.sort(
+                key=lambda n: tuple(
+                    (getattr(table[n], k),)
+                    if getattr(table.get(n), k, None) is not None else ()
+                    for k in ([k] if k else [
+                        k for k in Result._sort if k in fields])),
+                reverse=reverse ^ (not k or k in Result._fields))
+
+
+    # build up our lines
+    lines = []
+
+    # header
+    header = []
+    header.append('%s%s' % (
+        ','.join(by),
+        ' (%d added, %d removed)' % (
+            sum(1 for n in table if n not in diff_table),
+            sum(1 for n in diff_table if n not in table))
+            if diff_results is not None and not percent else '')
+        if not summary else '')
+    if diff_results is None:
+        for k in fields:
+            header.append(k)
+    elif percent:
+        for k in fields:
+            header.append(k)
+    else:
+        for k in fields:
+            header.append('o'+k)
+        for k in fields:
+            header.append('n'+k)
+        for k in fields:
+            header.append('d'+k)
+    header.append('')
+    lines.append(header)
+
+    def table_entry(name, r, diff_r=None, ratios=[]):
+        entry = []
+        entry.append(name)
+        if diff_results is None:
+            for k in fields:
+                entry.append(getattr(r, k).table()
+                    if getattr(r, k, None) is not None
+                    else types[k].none)
+        elif percent:
+            for k in fields:
+                entry.append(getattr(r, k).diff_table()
+                    if getattr(r, k, None) is not None
+                    else types[k].diff_none)
+        else:
+            for k in fields:
+                entry.append(getattr(diff_r, k).diff_table()
+                    if getattr(diff_r, k, None) is not None
+                    else types[k].diff_none)
+            for k in fields:
+                entry.append(getattr(r, k).diff_table()
+                    if getattr(r, k, None) is not None
+                    else types[k].diff_none)
+            for k in fields:
+                entry.append(types[k].diff_diff(
+                        getattr(r, k, None),
+                        getattr(diff_r, k, None)))
+        if diff_results is None:
+            entry.append('')
+        elif percent:
+            entry.append(' (%s)' % ', '.join(
+                '+∞%' if t == +m.inf
+                else '-∞%' if t == -m.inf
+                else '%+.1f%%' % (100*t)
+                for t in ratios))
+        else:
+            entry.append(' (%s)' % ', '.join(
+                    '+∞%' if t == +m.inf
+                    else '-∞%' if t == -m.inf
+                    else '%+.1f%%' % (100*t)
+                    for t in ratios
+                    if t)
+                if any(ratios) else '')
+        return entry
+
+    # entries
+    if not summary:
+        for name in names:
+            r = table.get(name)
+            if diff_results is None:
+                diff_r = None
+                ratios = None
+            else:
+                diff_r = diff_table.get(name)
+                ratios = [
+                    types[k].ratio(
+                        getattr(r, k, None),
+                        getattr(diff_r, k, None))
+                    for k in fields]
+                if not all_ and not any(ratios):
+                    continue
+            lines.append(table_entry(name, r, diff_r, ratios))
+
+    # total
+    r = next(iter(fold(Result, results, by=[])), None)
+    if diff_results is None:
+        diff_r = None
+        ratios = None
+    else:
+        diff_r = next(iter(fold(Result, diff_results, by=[])), None)
+        ratios = [
+            types[k].ratio(
+                getattr(r, k, None),
+                getattr(diff_r, k, None))
+            for k in fields]
+    lines.append(table_entry('TOTAL', r, diff_r, ratios))
+
+    # find the best widths, note that column 0 contains the names and column -1
+    # the ratios, so those are handled a bit differently
+    widths = [
+        ((max(it.chain([w], (len(l[i]) for l in lines)))+1+4-1)//4)*4-1
+        for w, i in zip(
+            it.chain([23], it.repeat(7)),
+            range(len(lines[0])-1))]
+
+    # print our table
+    for line in lines:
+        print('%-*s  %s%s' % (
+            widths[0], line[0],
+            ' '.join('%*s' % (w, x)
+                for w, x in zip(widths[1:], line[1:-1])),
+            line[-1]))
+
+
+def main(obj_paths, *,
+        by=None,
+        fields=None,
+        defines=None,
+        sort=None,
+        **args):
+    # find sizes
+    if not args.get('use', None):
+        results = collect(obj_paths, **args)
+    else:
+        results = []
+        with openio(args['use']) as f:
+            reader = csv.DictReader(f, restval='')
+            for r in reader:
+                if not any('code_'+k in r and r['code_'+k].strip()
+                        for k in CodeResult._fields):
+                    continue
+                try:
+                    results.append(CodeResult(
+                        **{k: r[k] for k in CodeResult._by
+                            if k in r and r[k].strip()},
+                        **{k: r['code_'+k] for k in CodeResult._fields
+                            if 'code_'+k in r and r['code_'+k].strip()}))
+                except TypeError:
+                    pass
+
+    # fold
+    results = fold(CodeResult, results, by=by, defines=defines)
+
+    # sort, note that python's sort is stable
+    results.sort()
+    if sort:
+        for k, reverse in reversed(sort):
+            results.sort(
+                key=lambda r: tuple(
+                    (getattr(r, k),) if getattr(r, k) is not None else ()
+                    for k in ([k] if k else CodeResult._sort)),
+                reverse=reverse ^ (not k or k in CodeResult._fields))
+
+    # write results to CSV
+    if args.get('output'):
+        with openio(args['output'], 'w') as f:
+            writer = csv.DictWriter(f,
+                (by if by is not None else CodeResult._by)
+                + ['code_'+k for k in (
+                    fields if fields is not None else CodeResult._fields)])
+            writer.writeheader()
+            for r in results:
+                writer.writerow(
+                    {k: getattr(r, k) for k in (
+                        by if by is not None else CodeResult._by)}
+                    | {'code_'+k: getattr(r, k) for k in (
+                        fields if fields is not None else CodeResult._fields)})
+
+    # find previous results?
+    if args.get('diff'):
+        diff_results = []
+        try:
+            with openio(args['diff']) as f:
+                reader = csv.DictReader(f, restval='')
+                for r in reader:
+                    if not any('code_'+k in r and r['code_'+k].strip()
+                            for k in CodeResult._fields):
+                        continue
+                    try:
+                        diff_results.append(CodeResult(
+                            **{k: r[k] for k in CodeResult._by
+                                if k in r and r[k].strip()},
+                            **{k: r['code_'+k] for k in CodeResult._fields
+                                if 'code_'+k in r and r['code_'+k].strip()}))
+                    except TypeError:
+                        pass
+        except FileNotFoundError:
+            pass
+
+        # fold
+        diff_results = fold(CodeResult, diff_results, by=by, defines=defines)
+
+    # print table
+    if not args.get('quiet'):
+        table(CodeResult, results,
+            diff_results if args.get('diff') else None,
+            by=by if by is not None else ['function'],
+            fields=fields,
+            sort=sort,
+            **args)
+
+
+if __name__ == "__main__":
+    import argparse
+    import sys
+    parser = argparse.ArgumentParser(
+        description="Find code size at the function level.",
+        allow_abbrev=False)
+    parser.add_argument(
+        'obj_paths',
+        nargs='*',
+        help="Input *.o files.")
+    parser.add_argument(
+        '-v', '--verbose',
+        action='store_true',
+        help="Output commands that run behind the scenes.")
+    parser.add_argument(
+        '-q', '--quiet',
+        action='store_true',
+        help="Don't show anything, useful with -o.")
+    parser.add_argument(
+        '-o', '--output',
+        help="Specify CSV file to store results.")
+    parser.add_argument(
+        '-u', '--use',
+        help="Don't parse anything, use this CSV file.")
+    parser.add_argument(
+        '-d', '--diff',
+        help="Specify CSV file to diff against.")
+    parser.add_argument(
+        '-a', '--all',
+        action='store_true',
+        help="Show all, not just the ones that changed.")
+    parser.add_argument(
+        '-p', '--percent',
+        action='store_true',
+        help="Only show percentage change, not a full diff.")
+    parser.add_argument(
+        '-b', '--by',
+        action='append',
+        choices=CodeResult._by,
+        help="Group by this field.")
+    parser.add_argument(
+        '-f', '--field',
+        dest='fields',
+        action='append',
+        choices=CodeResult._fields,
+        help="Show this field.")
+    parser.add_argument(
+        '-D', '--define',
+        dest='defines',
+        action='append',
+        type=lambda x: (lambda k,v: (k, set(v.split(','))))(*x.split('=', 1)),
+        help="Only include results where this field is this value.")
+    class AppendSort(argparse.Action):
+        def __call__(self, parser, namespace, value, option):
+            if namespace.sort is None:
+                namespace.sort = []
+            namespace.sort.append((value, True if option == '-S' else False))
+    parser.add_argument(
+        '-s', '--sort',
+        nargs='?',
+        action=AppendSort,
+        help="Sort by this field.")
+    parser.add_argument(
+        '-S', '--reverse-sort',
+        nargs='?',
+        action=AppendSort,
+        help="Sort by this field, but backwards.")
+    parser.add_argument(
+        '-Y', '--summary',
+        action='store_true',
+        help="Only show the total.")
+    parser.add_argument(
+        '-F', '--source',
+        dest='sources',
+        action='append',
+        help="Only consider definitions in this file. Defaults to anything "
+            "in the current directory.")
+    parser.add_argument(
+        '--everything',
+        action='store_true',
+        help="Include builtin and libc specific symbols.")
+    parser.add_argument(
+        '--nm-types',
+        default=NM_TYPES,
+        help="Type of symbols to report, this uses the same single-character "
+            "type-names emitted by nm. Defaults to %r." % NM_TYPES)
+    parser.add_argument(
+        '--nm-path',
+        type=lambda x: x.split(),
+        default=NM_PATH,
+        help="Path to the nm executable, may include flags. "
+            "Defaults to %r." % NM_PATH)
+    parser.add_argument(
+        '--objdump-path',
+        type=lambda x: x.split(),
+        default=OBJDUMP_PATH,
+        help="Path to the objdump executable, may include flags. "
+            "Defaults to %r." % OBJDUMP_PATH)
+    sys.exit(main(**{k: v
+        for k, v in vars(parser.parse_intermixed_args()).items()
+        if v is not None}))

kernel/fs/littlefs/scripts/cov.py

@@ -0,0 +1,828 @@
+#!/usr/bin/env python3
+#
+# Script to find coverage info after running tests.
+#
+# Example:
+# ./scripts/cov.py \
+#     lfs.t.a.gcda lfs_util.t.a.gcda \
+#     -Flfs.c -Flfs_util.c -slines
+#
+# Copyright (c) 2022, The littlefs authors.
+# Copyright (c) 2020, Arm Limited. All rights reserved.
+# SPDX-License-Identifier: BSD-3-Clause
+#
+
+import collections as co
+import csv
+import itertools as it
+import json
+import math as m
+import os
+import re
+import shlex
+import subprocess as sp
+
+# TODO use explode_asserts to avoid counting assert branches?
+# TODO use dwarf=info to find functions for inline functions?
+
+GCOV_PATH = ['gcov']
+
+
+# integer fields
+class Int(co.namedtuple('Int', 'x')):
+    __slots__ = ()
+    def __new__(cls, x=0):
+        if isinstance(x, Int):
+            return x
+        if isinstance(x, str):
+            try:
+                x = int(x, 0)
+            except ValueError:
+                # also accept +-∞ and +-inf
+                if re.match('^\s*\+?\s*(?:∞|inf)\s*$', x):
+                    x = m.inf
+                elif re.match('^\s*-\s*(?:∞|inf)\s*$', x):
+                    x = -m.inf
+                else:
+                    raise
+        assert isinstance(x, int) or m.isinf(x), x
+        return super().__new__(cls, x)
+
+    def __str__(self):
+        if self.x == m.inf:
+            return '∞'
+        elif self.x == -m.inf:
+            return '-∞'
+        else:
+            return str(self.x)
+
+    def __int__(self):
+        assert not m.isinf(self.x)
+        return self.x
+
+    def __float__(self):
+        return float(self.x)
+
+    none = '%7s' % '-'
+    def table(self):
+        return '%7s' % (self,)
+
+    diff_none = '%7s' % '-'
+    diff_table = table
+
+    def diff_diff(self, other):
+        new = self.x if self else 0
+        old = other.x if other else 0
+        diff = new - old
+        if diff == +m.inf:
+            return '%7s' % '+∞'
+        elif diff == -m.inf:
+            return '%7s' % '-∞'
+        else:
+            return '%+7d' % diff
+
+    def ratio(self, other):
+        new = self.x if self else 0
+        old = other.x if other else 0
+        if m.isinf(new) and m.isinf(old):
+            return 0.0
+        elif m.isinf(new):
+            return +m.inf
+        elif m.isinf(old):
+            return -m.inf
+        elif not old and not new:
+            return 0.0
+        elif not old:
+            return 1.0
+        else:
+            return (new-old) / old
+
+    def __add__(self, other):
+        return self.__class__(self.x + other.x)
+
+    def __sub__(self, other):
+        return self.__class__(self.x - other.x)
+
+    def __mul__(self, other):
+        return self.__class__(self.x * other.x)
+
+# fractional fields, a/b
+class Frac(co.namedtuple('Frac', 'a,b')):
+    __slots__ = ()
+    def __new__(cls, a=0, b=None):
+        if isinstance(a, Frac) and b is None:
+            return a
+        if isinstance(a, str) and b is None:
+            a, b = a.split('/', 1)
+        if b is None:
+            b = a
+        return super().__new__(cls, Int(a), Int(b))
+
+    def __str__(self):
+        return '%s/%s' % (self.a, self.b)
+
+    def __float__(self):
+        return float(self.a)
+
+    none = '%11s %7s' % ('-', '-')
+    def table(self):
+        t = self.a.x/self.b.x if self.b.x else 1.0
+        return '%11s %7s' % (
+            self,
+            '∞%' if t == +m.inf
+            else '-∞%' if t == -m.inf
+            else '%.1f%%' % (100*t))
+
+    diff_none = '%11s' % '-'
+    def diff_table(self):
+        return '%11s' % (self,)
+
+    def diff_diff(self, other):
+        new_a, new_b = self if self else (Int(0), Int(0))
+        old_a, old_b = other if other else (Int(0), Int(0))
+        return '%11s' % ('%s/%s' % (
+            new_a.diff_diff(old_a).strip(),
+            new_b.diff_diff(old_b).strip()))
+
+    def ratio(self, other):
+        new_a, new_b = self if self else (Int(0), Int(0))
+        old_a, old_b = other if other else (Int(0), Int(0))
+        new = new_a.x/new_b.x if new_b.x else 1.0
+        old = old_a.x/old_b.x if old_b.x else 1.0
+        return new - old
+
+    def __add__(self, other):
+        return self.__class__(self.a + other.a, self.b + other.b)
+
+    def __sub__(self, other):
+        return self.__class__(self.a - other.a, self.b - other.b)
+
+    def __mul__(self, other):
+        return self.__class__(self.a * other.a, self.b + other.b)
+
+    def __lt__(self, other):
+        self_t = self.a.x/self.b.x if self.b.x else 1.0
+        other_t = other.a.x/other.b.x if other.b.x else 1.0
+        return (self_t, self.a.x) < (other_t, other.a.x)
+
+    def __gt__(self, other):
+        return self.__class__.__lt__(other, self)
+
+    def __le__(self, other):
+        return not self.__gt__(other)
+
+    def __ge__(self, other):
+        return not self.__lt__(other)
+
+# coverage results
+class CovResult(co.namedtuple('CovResult', [
+        'file', 'function', 'line',
+        'calls', 'hits', 'funcs', 'lines', 'branches'])):
+    _by = ['file', 'function', 'line']
+    _fields = ['calls', 'hits', 'funcs', 'lines', 'branches']
+    _sort = ['funcs', 'lines', 'branches', 'hits', 'calls']
+    _types = {
+        'calls': Int, 'hits': Int,
+        'funcs': Frac, 'lines': Frac, 'branches': Frac}
+
+    __slots__ = ()
+    def __new__(cls, file='', function='', line=0,
+            calls=0, hits=0, funcs=0, lines=0, branches=0):
+        return super().__new__(cls, file, function, int(Int(line)),
+            Int(calls), Int(hits), Frac(funcs), Frac(lines), Frac(branches))
+
+    def __add__(self, other):
+        return CovResult(self.file, self.function, self.line,
+            max(self.calls, other.calls),
+            max(self.hits, other.hits),
+            self.funcs + other.funcs,
+            self.lines + other.lines,
+            self.branches + other.branches)
+
+
+def openio(path, mode='r', buffering=-1):
+    # allow '-' for stdin/stdout
+    if path == '-':
+        if mode == 'r':
+            return os.fdopen(os.dup(sys.stdin.fileno()), mode, buffering)
+        else:
+            return os.fdopen(os.dup(sys.stdout.fileno()), mode, buffering)
+    else:
+        return open(path, mode, buffering)
+
+def collect(gcda_paths, *,
+        gcov_path=GCOV_PATH,
+        sources=None,
+        everything=False,
+        **args):
+    results = []
+    for path in gcda_paths:
+        # get coverage info through gcov's json output
+        # note, gcov-path may contain extra args
+        cmd = GCOV_PATH + ['-b', '-t', '--json-format', path]
+        if args.get('verbose'):
+            print(' '.join(shlex.quote(c) for c in cmd))
+        proc = sp.Popen(cmd,
+            stdout=sp.PIPE,
+            stderr=sp.PIPE if not args.get('verbose') else None,
+            universal_newlines=True,
+            errors='replace',
+            close_fds=False)
+        data = json.load(proc.stdout)
+        proc.wait()
+        if proc.returncode != 0:
+            if not args.get('verbose'):
+                for line in proc.stderr:
+                    sys.stdout.write(line)
+            sys.exit(-1)
+
+        # collect line/branch coverage
+        for file in data['files']:
+            # ignore filtered sources
+            if sources is not None:
+                if not any(
+                        os.path.abspath(file['file']) == os.path.abspath(s)
+                        for s in sources):
+                    continue
+            else:
+                # default to only cwd
+                if not everything and not os.path.commonpath([
+                        os.getcwd(),
+                        os.path.abspath(file['file'])]) == os.getcwd():
+                    continue
+
+            # simplify path
+            if os.path.commonpath([
+                    os.getcwd(),
+                    os.path.abspath(file['file'])]) == os.getcwd():
+                file_name = os.path.relpath(file['file'])
+            else:
+                file_name = os.path.abspath(file['file'])
+
+            for func in file['functions']:
+                func_name = func.get('name', '(inlined)')
+                # discard internal functions (this includes injected test cases)
+                if not everything:
+                    if func_name.startswith('__'):
+                        continue
+
+                # go ahead and add functions, later folding will merge this if
+                # there are other hits on this line
+                results.append(CovResult(
+                    file_name, func_name, func['start_line'],
+                    func['execution_count'], 0,
+                    Frac(1 if func['execution_count'] > 0 else 0, 1),
+                    0,
+                    0))
+
+            for line in file['lines']:
+                func_name = line.get('function_name', '(inlined)')
+                # discard internal function (this includes injected test cases)
+                if not everything:
+                    if func_name.startswith('__'):
+                        continue
+
+                # go ahead and add lines, later folding will merge this if
+                # there are other hits on this line
+                results.append(CovResult(
+                    file_name, func_name, line['line_number'],
+                    0, line['count'],
+                    0,
+                    Frac(1 if line['count'] > 0 else 0, 1),
+                    Frac(
+                        sum(1 if branch['count'] > 0 else 0
+                            for branch in line['branches']),
+                        len(line['branches']))))
+
+    return results
+
+
+def fold(Result, results, *,
+        by=None,
+        defines=None,
+        **_):
+    if by is None:
+        by = Result._by
+
+    for k in it.chain(by or [], (k for k, _ in defines or [])):
+        if k not in Result._by and k not in Result._fields:
+            print("error: could not find field %r?" % k)
+            sys.exit(-1)
+
+    # filter by matching defines
+    if defines is not None:
+        results_ = []
+        for r in results:
+            if all(getattr(r, k) in vs for k, vs in defines):
+                results_.append(r)
+        results = results_
+
+    # organize results into conflicts
+    folding = co.OrderedDict()
+    for r in results:
+        name = tuple(getattr(r, k) for k in by)
+        if name not in folding:
+            folding[name] = []
+        folding[name].append(r)
+
+    # merge conflicts
+    folded = []
+    for name, rs in folding.items():
+        folded.append(sum(rs[1:], start=rs[0]))
+
+    return folded
+
+def table(Result, results, diff_results=None, *,
+        by=None,
+        fields=None,
+        sort=None,
+        summary=False,
+        all=False,
+        percent=False,
+        **_):
+    all_, all = all, __builtins__.all
+
+    if by is None:
+        by = Result._by
+    if fields is None:
+        fields = Result._fields
+    types = Result._types
+
+    # fold again
+    results = fold(Result, results, by=by)
+    if diff_results is not None:
+        diff_results = fold(Result, diff_results, by=by)
+
+    # organize by name
+    table = {
+        ','.join(str(getattr(r, k) or '') for k in by): r
+        for r in results}
+    diff_table = {
+        ','.join(str(getattr(r, k) or '') for k in by): r
+        for r in diff_results or []}
+    names = list(table.keys() | diff_table.keys())
+
+    # sort again, now with diff info, note that python's sort is stable
+    names.sort()
+    if diff_results is not None:
+        names.sort(key=lambda n: tuple(
+            types[k].ratio(
+                getattr(table.get(n), k, None),
+                getattr(diff_table.get(n), k, None))
+            for k in fields),
+            reverse=True)
+    if sort:
+        for k, reverse in reversed(sort):
+            names.sort(
+                key=lambda n: tuple(
+                    (getattr(table[n], k),)
+                    if getattr(table.get(n), k, None) is not None else ()
+                    for k in ([k] if k else [
+                        k for k in Result._sort if k in fields])),
+                reverse=reverse ^ (not k or k in Result._fields))
+
+
+    # build up our lines
+    lines = []
+
+    # header
+    header = []
+    header.append('%s%s' % (
+        ','.join(by),
+        ' (%d added, %d removed)' % (
+            sum(1 for n in table if n not in diff_table),
+            sum(1 for n in diff_table if n not in table))
+            if diff_results is not None and not percent else '')
+        if not summary else '')
+    if diff_results is None:
+        for k in fields:
+            header.append(k)
+    elif percent:
+        for k in fields:
+            header.append(k)
+    else:
+        for k in fields:
+            header.append('o'+k)
+        for k in fields:
+            header.append('n'+k)
+        for k in fields:
+            header.append('d'+k)
+    header.append('')
+    lines.append(header)
+
+    def table_entry(name, r, diff_r=None, ratios=[]):
+        entry = []
+        entry.append(name)
+        if diff_results is None:
+            for k in fields:
+                entry.append(getattr(r, k).table()
+                    if getattr(r, k, None) is not None
+                    else types[k].none)
+        elif percent:
+            for k in fields:
+                entry.append(getattr(r, k).diff_table()
+                    if getattr(r, k, None) is not None
+                    else types[k].diff_none)
+        else:
+            for k in fields:
+                entry.append(getattr(diff_r, k).diff_table()
+                    if getattr(diff_r, k, None) is not None
+                    else types[k].diff_none)
+            for k in fields:
+                entry.append(getattr(r, k).diff_table()
+                    if getattr(r, k, None) is not None
+                    else types[k].diff_none)
+            for k in fields:
+                entry.append(types[k].diff_diff(
+                        getattr(r, k, None),
+                        getattr(diff_r, k, None)))
+        if diff_results is None:
+            entry.append('')
+        elif percent:
+            entry.append(' (%s)' % ', '.join(
+                '+∞%' if t == +m.inf
+                else '-∞%' if t == -m.inf
+                else '%+.1f%%' % (100*t)
+                for t in ratios))
+        else:
+            entry.append(' (%s)' % ', '.join(
+                    '+∞%' if t == +m.inf
+                    else '-∞%' if t == -m.inf
+                    else '%+.1f%%' % (100*t)
+                    for t in ratios
+                    if t)
+                if any(ratios) else '')
+        return entry
+
+    # entries
+    if not summary:
+        for name in names:
+            r = table.get(name)
+            if diff_results is None:
+                diff_r = None
+                ratios = None
+            else:
+                diff_r = diff_table.get(name)
+                ratios = [
+                    types[k].ratio(
+                        getattr(r, k, None),
+                        getattr(diff_r, k, None))
+                    for k in fields]
+                if not all_ and not any(ratios):
+                    continue
+            lines.append(table_entry(name, r, diff_r, ratios))
+
+    # total
+    r = next(iter(fold(Result, results, by=[])), None)
+    if diff_results is None:
+        diff_r = None
+        ratios = None
+    else:
+        diff_r = next(iter(fold(Result, diff_results, by=[])), None)
+        ratios = [
+            types[k].ratio(
+                getattr(r, k, None),
+                getattr(diff_r, k, None))
+            for k in fields]
+    lines.append(table_entry('TOTAL', r, diff_r, ratios))
+
+    # find the best widths, note that column 0 contains the names and column -1
+    # the ratios, so those are handled a bit differently
+    widths = [
+        ((max(it.chain([w], (len(l[i]) for l in lines)))+1+4-1)//4)*4-1
+        for w, i in zip(
+            it.chain([23], it.repeat(7)),
+            range(len(lines[0])-1))]
+
+    # print our table
+    for line in lines:
+        print('%-*s  %s%s' % (
+            widths[0], line[0],
+            ' '.join('%*s' % (w, x)
+                for w, x in zip(widths[1:], line[1:-1])),
+            line[-1]))
+
+
+def annotate(Result, results, *,
+        annotate=False,
+        lines=False,
+        branches=False,
+        **args):
+    # if neither branches/lines specified, color both
+    if annotate and not lines and not branches:
+        lines, branches = True, True
+
+    for path in co.OrderedDict.fromkeys(r.file for r in results).keys():
+        # flatten to line info
+        results = fold(Result, results, by=['file', 'line'])
+        table = {r.line: r for r in results if r.file == path}
+
+        # calculate spans to show
+        if not annotate:
+            spans = []
+            last = None
+            func = None
+            for line, r in sorted(table.items()):
+                if ((lines and int(r.hits) == 0)
+                        or (branches and r.branches.a < r.branches.b)):
+                    if last is not None and line - last.stop <= args['context']:
+                        last = range(
+                            last.start,
+                            line+1+args['context'])
+                    else:
+                        if last is not None:
+                            spans.append((last, func))
+                        last = range(
+                            line-args['context'],
+                            line+1+args['context'])
+                        func = r.function
+            if last is not None:
+                spans.append((last, func))
+
+        with open(path) as f:
+            skipped = False
+            for i, line in enumerate(f):
+                # skip lines not in spans?
+                if not annotate and not any(i+1 in s for s, _ in spans):
+                    skipped = True
+                    continue
+
+                if skipped:
+                    skipped = False
+                    print('%s@@ %s:%d: %s @@%s' % (
+                        '\x1b[36m' if args['color'] else '',
+                        path,
+                        i+1,
+                        next(iter(f for _, f in spans)),
+                        '\x1b[m' if args['color'] else ''))
+
+                # build line
+                if line.endswith('\n'):
+                    line = line[:-1]
+
+                if i+1 in table:
+                    r = table[i+1]
+                    line = '%-*s // %s hits%s' % (
+                        args['width'],
+                        line,
+                        r.hits,
+                        ', %s branches' % (r.branches,)
+                            if int(r.branches.b) else '')
+
+                    if args['color']:
+                        if lines and int(r.hits) == 0:
+                            line = '\x1b[1;31m%s\x1b[m' % line
+                        elif branches and r.branches.a < r.branches.b:
+                            line = '\x1b[35m%s\x1b[m' % line
+
+                print(line)
+
+
+def main(gcda_paths, *,
+        by=None,
+        fields=None,
+        defines=None,
+        sort=None,
+        hits=False,
+        **args):
+    # figure out what color should be
+    if args.get('color') == 'auto':
+        args['color'] = sys.stdout.isatty()
+    elif args.get('color') == 'always':
+        args['color'] = True
+    else:
+        args['color'] = False
+
+    # find sizes
+    if not args.get('use', None):
+        results = collect(gcda_paths, **args)
+    else:
+        results = []
+        with openio(args['use']) as f:
+            reader = csv.DictReader(f, restval='')
+            for r in reader:
+                if not any('cov_'+k in r and r['cov_'+k].strip()
+                        for k in CovResult._fields):
+                    continue
+                try:
+                    results.append(CovResult(
+                        **{k: r[k] for k in CovResult._by
+                            if k in r and r[k].strip()},
+                        **{k: r['cov_'+k]
+                            for k in CovResult._fields
+                            if 'cov_'+k in r
+                                and r['cov_'+k].strip()}))
+                except TypeError:
+                    pass
+
+    # fold
+    results = fold(CovResult, results, by=by, defines=defines)
+
+    # sort, note that python's sort is stable
+    results.sort()
+    if sort:
+        for k, reverse in reversed(sort):
+            results.sort(
+                key=lambda r: tuple(
+                    (getattr(r, k),) if getattr(r, k) is not None else ()
+                    for k in ([k] if k else CovResult._sort)),
+                reverse=reverse ^ (not k or k in CovResult._fields))
+
+    # write results to CSV
+    if args.get('output'):
+        with openio(args['output'], 'w') as f:
+            writer = csv.DictWriter(f,
+                (by if by is not None else CovResult._by)
+                + ['cov_'+k for k in (
+                    fields if fields is not None else CovResult._fields)])
+            writer.writeheader()
+            for r in results:
+                writer.writerow(
+                    {k: getattr(r, k) for k in (
+                        by if by is not None else CovResult._by)}
+                    | {'cov_'+k: getattr(r, k) for k in (
+                        fields if fields is not None else CovResult._fields)})
+
+    # find previous results?
+    if args.get('diff'):
+        diff_results = []
+        try:
+            with openio(args['diff']) as f:
+                reader = csv.DictReader(f, restval='')
+                for r in reader:
+                    if not any('cov_'+k in r and r['cov_'+k].strip()
+                            for k in CovResult._fields):
+                        continue
+                    try:
+                        diff_results.append(CovResult(
+                            **{k: r[k] for k in CovResult._by
+                                if k in r and r[k].strip()},
+                            **{k: r['cov_'+k]
+                                for k in CovResult._fields
+                                if 'cov_'+k in r
+                                    and r['cov_'+k].strip()}))
+                    except TypeError:
+                        pass
+        except FileNotFoundError:
+            pass
+
+        # fold
+        diff_results = fold(CovResult, diff_results,
+            by=by, defines=defines)
+
+    # print table
+    if not args.get('quiet'):
+        if (args.get('annotate')
+                or args.get('lines')
+                or args.get('branches')):
+            # annotate sources
+            annotate(CovResult, results, **args)
+        else:
+            # print table
+            table(CovResult, results,
+                diff_results if args.get('diff') else None,
+                by=by if by is not None else ['function'],
+                fields=fields if fields is not None
+                    else ['lines', 'branches'] if not hits
+                    else ['calls', 'hits'],
+                sort=sort,
+                **args)
+
+    # catch lack of coverage
+    if args.get('error_on_lines') and any(
+            r.lines.a < r.lines.b for r in results):
+        sys.exit(2)
+    elif args.get('error_on_branches') and any(
+            r.branches.a < r.branches.b for r in results):
+        sys.exit(3)
+
+
+if __name__ == "__main__":
+    import argparse
+    import sys
+    parser = argparse.ArgumentParser(
+        description="Find coverage info after running tests.",
+        allow_abbrev=False)
+    parser.add_argument(
+        'gcda_paths',
+        nargs='*',
+        help="Input *.gcda files.")
+    parser.add_argument(
+        '-v', '--verbose',
+        action='store_true',
+        help="Output commands that run behind the scenes.")
+    parser.add_argument(
+        '-q', '--quiet',
+        action='store_true',
+        help="Don't show anything, useful with -o.")
+    parser.add_argument(
+        '-o', '--output',
+        help="Specify CSV file to store results.")
+    parser.add_argument(
+        '-u', '--use',
+        help="Don't parse anything, use this CSV file.")
+    parser.add_argument(
+        '-d', '--diff',
+        help="Specify CSV file to diff against.")
+    parser.add_argument(
+        '-a', '--all',
+        action='store_true',
+        help="Show all, not just the ones that changed.")
+    parser.add_argument(
+        '-p', '--percent',
+        action='store_true',
+        help="Only show percentage change, not a full diff.")
+    parser.add_argument(
+        '-b', '--by',
+        action='append',
+        choices=CovResult._by,
+        help="Group by this field.")
+    parser.add_argument(
+        '-f', '--field',
+        dest='fields',
+        action='append',
+        choices=CovResult._fields,
+        help="Show this field.")
+    parser.add_argument(
+        '-D', '--define',
+        dest='defines',
+        action='append',
+        type=lambda x: (lambda k,v: (k, set(v.split(','))))(*x.split('=', 1)),
+        help="Only include results where this field is this value.")
+    class AppendSort(argparse.Action):
+        def __call__(self, parser, namespace, value, option):
+            if namespace.sort is None:
+                namespace.sort = []
+            namespace.sort.append((value, True if option == '-S' else False))
+    parser.add_argument(
+        '-s', '--sort',
+        nargs='?',
+        action=AppendSort,
+        help="Sort by this field.")
+    parser.add_argument(
+        '-S', '--reverse-sort',
+        nargs='?',
+        action=AppendSort,
+        help="Sort by this field, but backwards.")
+    parser.add_argument(
+        '-Y', '--summary',
+        action='store_true',
+        help="Only show the total.")
+    parser.add_argument(
+        '-F', '--source',
+        dest='sources',
+        action='append',
+        help="Only consider definitions in this file. Defaults to anything "
+            "in the current directory.")
+    parser.add_argument(
+        '--everything',
+        action='store_true',
+        help="Include builtin and libc specific symbols.")
+    parser.add_argument(
+        '--hits',
+        action='store_true',
+        help="Show total hits instead of coverage.")
+    parser.add_argument(
+        '-A', '--annotate',
+        action='store_true',
+        help="Show source files annotated with coverage info.")
+    parser.add_argument(
+        '-L', '--lines',
+        action='store_true',
+        help="Show uncovered lines.")
+    parser.add_argument(
+        '-B', '--branches',
+        action='store_true',
+        help="Show uncovered branches.")
+    parser.add_argument(
+        '-c', '--context',
+        type=lambda x: int(x, 0),
+        default=3,
+        help="Show n additional lines of context. Defaults to 3.")
+    parser.add_argument(
+        '-W', '--width',
+        type=lambda x: int(x, 0),
+        default=80,
+        help="Assume source is styled with this many columns. Defaults to 80.")
+    parser.add_argument(
+        '--color',
+        choices=['never', 'always', 'auto'],
+        default='auto',
+        help="When to use terminal colors. Defaults to 'auto'.")
+    parser.add_argument(
+        '-e', '--error-on-lines',
+        action='store_true',
+        help="Error if any lines are not covered.")
+    parser.add_argument(
+        '-E', '--error-on-branches',
+        action='store_true',
+        help="Error if any branches are not covered.")
+    parser.add_argument(
+        '--gcov-path',
+        default=GCOV_PATH,
+        type=lambda x: x.split(),
+        help="Path to the gcov executable, may include paths. "
+            "Defaults to %r." % GCOV_PATH)
+    sys.exit(main(**{k: v
+        for k, v in vars(parser.parse_intermixed_args()).items()
+        if v is not None}))

kernel/fs/littlefs/scripts/data.py

@@ -0,0 +1,704 @@
+#!/usr/bin/env python3
+#
+# Script to find data size at the function level. Basically just a big wrapper
+# around nm with some extra conveniences for comparing builds. Heavily inspired
+# by Linux's Bloat-O-Meter.
+#
+# Example:
+# ./scripts/data.py lfs.o lfs_util.o -Ssize
+#
+# Copyright (c) 2022, The littlefs authors.
+# Copyright (c) 2020, Arm Limited. All rights reserved.
+# SPDX-License-Identifier: BSD-3-Clause
+#
+
+import collections as co
+import csv
+import difflib
+import itertools as it
+import math as m
+import os
+import re
+import shlex
+import subprocess as sp
+
+
+NM_PATH = ['nm']
+NM_TYPES = 'dDbB'
+OBJDUMP_PATH = ['objdump']
+
+
+# integer fields
+class Int(co.namedtuple('Int', 'x')):
+    __slots__ = ()
+    def __new__(cls, x=0):
+        if isinstance(x, Int):
+            return x
+        if isinstance(x, str):
+            try:
+                x = int(x, 0)
+            except ValueError:
+                # also accept +-∞ and +-inf
+                if re.match('^\s*\+?\s*(?:∞|inf)\s*$', x):
+                    x = m.inf
+                elif re.match('^\s*-\s*(?:∞|inf)\s*$', x):
+                    x = -m.inf
+                else:
+                    raise
+        assert isinstance(x, int) or m.isinf(x), x
+        return super().__new__(cls, x)
+
+    def __str__(self):
+        if self.x == m.inf:
+            return '∞'
+        elif self.x == -m.inf:
+            return '-∞'
+        else:
+            return str(self.x)
+
+    def __int__(self):
+        assert not m.isinf(self.x)
+        return self.x
+
+    def __float__(self):
+        return float(self.x)
+
+    none = '%7s' % '-'
+    def table(self):
+        return '%7s' % (self,)
+
+    diff_none = '%7s' % '-'
+    diff_table = table
+
+    def diff_diff(self, other):
+        new = self.x if self else 0
+        old = other.x if other else 0
+        diff = new - old
+        if diff == +m.inf:
+            return '%7s' % '+∞'
+        elif diff == -m.inf:
+            return '%7s' % '-∞'
+        else:
+            return '%+7d' % diff
+
+    def ratio(self, other):
+        new = self.x if self else 0
+        old = other.x if other else 0
+        if m.isinf(new) and m.isinf(old):
+            return 0.0
+        elif m.isinf(new):
+            return +m.inf
+        elif m.isinf(old):
+            return -m.inf
+        elif not old and not new:
+            return 0.0
+        elif not old:
+            return 1.0
+        else:
+            return (new-old) / old
+
+    def __add__(self, other):
+        return self.__class__(self.x + other.x)
+
+    def __sub__(self, other):
+        return self.__class__(self.x - other.x)
+
+    def __mul__(self, other):
+        return self.__class__(self.x * other.x)
+
+# data size results
+class DataResult(co.namedtuple('DataResult', [
+        'file', 'function',
+        'size'])):
+    _by = ['file', 'function']
+    _fields = ['size']
+    _sort = ['size']
+    _types = {'size': Int}
+
+    __slots__ = ()
+    def __new__(cls, file='', function='', size=0):
+        return super().__new__(cls, file, function,
+            Int(size))
+
+    def __add__(self, other):
+        return DataResult(self.file, self.function,
+            self.size + other.size)
+
+
+def openio(path, mode='r', buffering=-1):
+    # allow '-' for stdin/stdout
+    if path == '-':
+        if mode == 'r':
+            return os.fdopen(os.dup(sys.stdin.fileno()), mode, buffering)
+        else:
+            return os.fdopen(os.dup(sys.stdout.fileno()), mode, buffering)
+    else:
+        return open(path, mode, buffering)
+
+def collect(obj_paths, *,
+        nm_path=NM_PATH,
+        nm_types=NM_TYPES,
+        objdump_path=OBJDUMP_PATH,
+        sources=None,
+        everything=False,
+        **args):
+    size_pattern = re.compile(
+        '^(?P<size>[0-9a-fA-F]+)' +
+        ' (?P<type>[%s])' % re.escape(nm_types) +
+        ' (?P<func>.+?)$')
+    line_pattern = re.compile(
+        '^\s+(?P<no>[0-9]+)'
+            '(?:\s+(?P<dir>[0-9]+))?'
+            '\s+.*'
+            '\s+(?P<path>[^\s]+)$')
+    info_pattern = re.compile(
+        '^(?:.*(?P<tag>DW_TAG_[a-z_]+).*'
+            '|.*DW_AT_name.*:\s*(?P<name>[^:\s]+)\s*'
+            '|.*DW_AT_decl_file.*:\s*(?P<file>[0-9]+)\s*)$')
+
+    results = []
+    for path in obj_paths:
+        # guess the source, if we have debug-info we'll replace this later
+        file = re.sub('(\.o)?$', '.c', path, 1)
+
+        # find symbol sizes
+        results_ = []
+        # note nm-path may contain extra args
+        cmd = nm_path + ['--size-sort', path]
+        if args.get('verbose'):
+            print(' '.join(shlex.quote(c) for c in cmd))
+        proc = sp.Popen(cmd,
+            stdout=sp.PIPE,
+            stderr=sp.PIPE if not args.get('verbose') else None,
+            universal_newlines=True,
+            errors='replace',
+            close_fds=False)
+        for line in proc.stdout:
+            m = size_pattern.match(line)
+            if m:
+                func = m.group('func')
+                # discard internal functions
+                if not everything and func.startswith('__'):
+                    continue
+                results_.append(DataResult(
+                    file, func,
+                    int(m.group('size'), 16)))
+        proc.wait()
+        if proc.returncode != 0:
+            if not args.get('verbose'):
+                for line in proc.stderr:
+                    sys.stdout.write(line)
+            sys.exit(-1)
+
+
+        # try to figure out the source file if we have debug-info
+        dirs = {}
+        files = {}
+        # note objdump-path may contain extra args
+        cmd = objdump_path + ['--dwarf=rawline', path]
+        if args.get('verbose'):
+            print(' '.join(shlex.quote(c) for c in cmd))
+        proc = sp.Popen(cmd,
+            stdout=sp.PIPE,
+            stderr=sp.PIPE if not args.get('verbose') else None,
+            universal_newlines=True,
+            errors='replace',
+            close_fds=False)
+        for line in proc.stdout:
+            # note that files contain references to dirs, which we
+            # dereference as soon as we see them as each file table follows a
+            # dir table
+            m = line_pattern.match(line)
+            if m:
+                if not m.group('dir'):
+                    # found a directory entry
+                    dirs[int(m.group('no'))] = m.group('path')
+                else:
+                    # found a file entry
+                    dir = int(m.group('dir'))
+                    if dir in dirs:
+                        files[int(m.group('no'))] = os.path.join(
+                            dirs[dir],
+                            m.group('path'))
+                    else:
+                        files[int(m.group('no'))] = m.group('path')
+        proc.wait()
+        if proc.returncode != 0:
+            if not args.get('verbose'):
+                for line in proc.stderr:
+                    sys.stdout.write(line)
+            # do nothing on error, we don't need objdump to work, source files
+            # may just be inaccurate
+            pass
+
+        defs = {}
+        is_func = False
+        f_name = None
+        f_file = None
+        # note objdump-path may contain extra args
+        cmd = objdump_path + ['--dwarf=info', path]
+        if args.get('verbose'):
+            print(' '.join(shlex.quote(c) for c in cmd))
+        proc = sp.Popen(cmd,
+            stdout=sp.PIPE,
+            stderr=sp.PIPE if not args.get('verbose') else None,
+            universal_newlines=True,
+            errors='replace',
+            close_fds=False)
+        for line in proc.stdout:
+            # state machine here to find definitions
+            m = info_pattern.match(line)
+            if m:
+                if m.group('tag'):
+                    if is_func:
+                        defs[f_name] = files.get(f_file, '?')
+                    is_func = (m.group('tag') == 'DW_TAG_subprogram')
+                elif m.group('name'):
+                    f_name = m.group('name')
+                elif m.group('file'):
+                    f_file = int(m.group('file'))
+        if is_func:
+            defs[f_name] = files.get(f_file, '?')
+        proc.wait()
+        if proc.returncode != 0:
+            if not args.get('verbose'):
+                for line in proc.stderr:
+                    sys.stdout.write(line)
+            # do nothing on error, we don't need objdump to work, source files
+            # may just be inaccurate
+            pass
+
+        for r in results_:
+            # find best matching debug symbol, this may be slightly different
+            # due to optimizations
+            if defs:
+                # exact match? avoid difflib if we can for speed
+                if r.function in defs:
+                    file = defs[r.function]
+                else:
+                    _, file = max(
+                        defs.items(),
+                        key=lambda d: difflib.SequenceMatcher(None,
+                            d[0],
+                            r.function, False).ratio())
+            else:
+                file = r.file
+
+            # ignore filtered sources
+            if sources is not None:
+                if not any(
+                        os.path.abspath(file) == os.path.abspath(s)
+                        for s in sources):
+                    continue
+            else:
+                # default to only cwd
+                if not everything and not os.path.commonpath([
+                        os.getcwd(),
+                        os.path.abspath(file)]) == os.getcwd():
+                    continue
+
+            # simplify path
+            if os.path.commonpath([
+                    os.getcwd(),
+                    os.path.abspath(file)]) == os.getcwd():
+                file = os.path.relpath(file)
+            else:
+                file = os.path.abspath(file)
+
+            results.append(r._replace(file=file))
+
+    return results
+
+
+def fold(Result, results, *,
+        by=None,
+        defines=None,
+        **_):
+    if by is None:
+        by = Result._by
+
+    for k in it.chain(by or [], (k for k, _ in defines or [])):
+        if k not in Result._by and k not in Result._fields:
+            print("error: could not find field %r?" % k)
+            sys.exit(-1)
+
+    # filter by matching defines
+    if defines is not None:
+        results_ = []
+        for r in results:
+            if all(getattr(r, k) in vs for k, vs in defines):
+                results_.append(r)
+        results = results_
+
+    # organize results into conflicts
+    folding = co.OrderedDict()
+    for r in results:
+        name = tuple(getattr(r, k) for k in by)
+        if name not in folding:
+            folding[name] = []
+        folding[name].append(r)
+
+    # merge conflicts
+    folded = []
+    for name, rs in folding.items():
+        folded.append(sum(rs[1:], start=rs[0]))
+
+    return folded
+
+def table(Result, results, diff_results=None, *,
+        by=None,
+        fields=None,
+        sort=None,
+        summary=False,
+        all=False,
+        percent=False,
+        **_):
+    all_, all = all, __builtins__.all
+
+    if by is None:
+        by = Result._by
+    if fields is None:
+        fields = Result._fields
+    types = Result._types
+
+    # fold again
+    results = fold(Result, results, by=by)
+    if diff_results is not None:
+        diff_results = fold(Result, diff_results, by=by)
+
+    # organize by name
+    table = {
+        ','.join(str(getattr(r, k) or '') for k in by): r
+        for r in results}
+    diff_table = {
+        ','.join(str(getattr(r, k) or '') for k in by): r
+        for r in diff_results or []}
+    names = list(table.keys() | diff_table.keys())
+
+    # sort again, now with diff info, note that python's sort is stable
+    names.sort()
+    if diff_results is not None:
+        names.sort(key=lambda n: tuple(
+            types[k].ratio(
+                getattr(table.get(n), k, None),
+                getattr(diff_table.get(n), k, None))
+            for k in fields),
+            reverse=True)
+    if sort:
+        for k, reverse in reversed(sort):
+            names.sort(
+                key=lambda n: tuple(
+                    (getattr(table[n], k),)
+                    if getattr(table.get(n), k, None) is not None else ()
+                    for k in ([k] if k else [
+                        k for k in Result._sort if k in fields])),
+                reverse=reverse ^ (not k or k in Result._fields))
+
+
+    # build up our lines
+    lines = []
+
+    # header
+    header = []
+    header.append('%s%s' % (
+        ','.join(by),
+        ' (%d added, %d removed)' % (
+            sum(1 for n in table if n not in diff_table),
+            sum(1 for n in diff_table if n not in table))
+            if diff_results is not None and not percent else '')
+        if not summary else '')
+    if diff_results is None:
+        for k in fields:
+            header.append(k)
+    elif percent:
+        for k in fields:
+            header.append(k)
+    else:
+        for k in fields:
+            header.append('o'+k)
+        for k in fields:
+            header.append('n'+k)
+        for k in fields:
+            header.append('d'+k)
+    header.append('')
+    lines.append(header)
+
+    def table_entry(name, r, diff_r=None, ratios=[]):
+        entry = []
+        entry.append(name)
+        if diff_results is None:
+            for k in fields:
+                entry.append(getattr(r, k).table()
+                    if getattr(r, k, None) is not None
+                    else types[k].none)
+        elif percent:
+            for k in fields:
+                entry.append(getattr(r, k).diff_table()
+                    if getattr(r, k, None) is not None
+                    else types[k].diff_none)
+        else:
+            for k in fields:
+                entry.append(getattr(diff_r, k).diff_table()
+                    if getattr(diff_r, k, None) is not None
+                    else types[k].diff_none)
+            for k in fields:
+                entry.append(getattr(r, k).diff_table()
+                    if getattr(r, k, None) is not None
+                    else types[k].diff_none)
+            for k in fields:
+                entry.append(types[k].diff_diff(
+                        getattr(r, k, None),
+                        getattr(diff_r, k, None)))
+        if diff_results is None:
+            entry.append('')
+        elif percent:
+            entry.append(' (%s)' % ', '.join(
+                '+∞%' if t == +m.inf
+                else '-∞%' if t == -m.inf
+                else '%+.1f%%' % (100*t)
+                for t in ratios))
+        else:
+            entry.append(' (%s)' % ', '.join(
+                    '+∞%' if t == +m.inf
+                    else '-∞%' if t == -m.inf
+                    else '%+.1f%%' % (100*t)
+                    for t in ratios
+                    if t)
+                if any(ratios) else '')
+        return entry
+
+    # entries
+    if not summary:
+        for name in names:
+            r = table.get(name)
+            if diff_results is None:
+                diff_r = None
+                ratios = None
+            else:
+                diff_r = diff_table.get(name)
+                ratios = [
+                    types[k].ratio(
+                        getattr(r, k, None),
+                        getattr(diff_r, k, None))
+                    for k in fields]
+                if not all_ and not any(ratios):
+                    continue
+            lines.append(table_entry(name, r, diff_r, ratios))
+
+    # total
+    r = next(iter(fold(Result, results, by=[])), None)
+    if diff_results is None:
+        diff_r = None
+        ratios = None
+    else:
+        diff_r = next(iter(fold(Result, diff_results, by=[])), None)
+        ratios = [
+            types[k].ratio(
+                getattr(r, k, None),
+                getattr(diff_r, k, None))
+            for k in fields]
+    lines.append(table_entry('TOTAL', r, diff_r, ratios))
+
+    # find the best widths, note that column 0 contains the names and column -1
+    # the ratios, so those are handled a bit differently
+    widths = [
+        ((max(it.chain([w], (len(l[i]) for l in lines)))+1+4-1)//4)*4-1
+        for w, i in zip(
+            it.chain([23], it.repeat(7)),
+            range(len(lines[0])-1))]
+
+    # print our table
+    for line in lines:
+        print('%-*s  %s%s' % (
+            widths[0], line[0],
+            ' '.join('%*s' % (w, x)
+                for w, x in zip(widths[1:], line[1:-1])),
+            line[-1]))
+
+
+def main(obj_paths, *,
+        by=None,
+        fields=None,
+        defines=None,
+        sort=None,
+        **args):
+    # find sizes
+    if not args.get('use', None):
+        results = collect(obj_paths, **args)
+    else:
+        results = []
+        with openio(args['use']) as f:
+            reader = csv.DictReader(f, restval='')
+            for r in reader:
+                try:
+                    results.append(DataResult(
+                        **{k: r[k] for k in DataResult._by
+                            if k in r and r[k].strip()},
+                        **{k: r['data_'+k] for k in DataResult._fields
+                            if 'data_'+k in r and r['data_'+k].strip()}))
+                except TypeError:
+                    pass
+
+    # fold
+    results = fold(DataResult, results, by=by, defines=defines)
+
+    # sort, note that python's sort is stable
+    results.sort()
+    if sort:
+        for k, reverse in reversed(sort):
+            results.sort(
+                key=lambda r: tuple(
+                    (getattr(r, k),) if getattr(r, k) is not None else ()
+                    for k in ([k] if k else DataResult._sort)),
+                reverse=reverse ^ (not k or k in DataResult._fields))
+
+    # write results to CSV
+    if args.get('output'):
+        with openio(args['output'], 'w') as f:
+            writer = csv.DictWriter(f,
+                (by if by is not None else DataResult._by)
+                + ['data_'+k for k in (
+                    fields if fields is not None else DataResult._fields)])
+            writer.writeheader()
+            for r in results:
+                writer.writerow(
+                    {k: getattr(r, k) for k in (
+                        by if by is not None else DataResult._by)}
+                    | {'data_'+k: getattr(r, k) for k in (
+                        fields if fields is not None else DataResult._fields)})
+
+    # find previous results?
+    if args.get('diff'):
+        diff_results = []
+        try:
+            with openio(args['diff']) as f:
+                reader = csv.DictReader(f, restval='')
+                for r in reader:
+                    if not any('data_'+k in r and r['data_'+k].strip()
+                            for k in DataResult._fields):
+                        continue
+                    try:
+                        diff_results.append(DataResult(
+                            **{k: r[k] for k in DataResult._by
+                                if k in r and r[k].strip()},
+                            **{k: r['data_'+k] for k in DataResult._fields
+                                if 'data_'+k in r and r['data_'+k].strip()}))
+                    except TypeError:
+                        pass
+        except FileNotFoundError:
+            pass
+
+        # fold
+        diff_results = fold(DataResult, diff_results, by=by, defines=defines)
+
+    # print table
+    if not args.get('quiet'):
+        table(DataResult, results,
+            diff_results if args.get('diff') else None,
+            by=by if by is not None else ['function'],
+            fields=fields,
+            sort=sort,
+            **args)
+
+
+if __name__ == "__main__":
+    import argparse
+    import sys
+    parser = argparse.ArgumentParser(
+        description="Find data size at the function level.",
+        allow_abbrev=False)
+    parser.add_argument(
+        'obj_paths',
+        nargs='*',
+        help="Input *.o files.")
+    parser.add_argument(
+        '-v', '--verbose',
+        action='store_true',
+        help="Output commands that run behind the scenes.")
+    parser.add_argument(
+        '-q', '--quiet',
+        action='store_true',
+        help="Don't show anything, useful with -o.")
+    parser.add_argument(
+        '-o', '--output',
+        help="Specify CSV file to store results.")
+    parser.add_argument(
+        '-u', '--use',
+        help="Don't parse anything, use this CSV file.")
+    parser.add_argument(
+        '-d', '--diff',
+        help="Specify CSV file to diff against.")
+    parser.add_argument(
+        '-a', '--all',
+        action='store_true',
+        help="Show all, not just the ones that changed.")
+    parser.add_argument(
+        '-p', '--percent',
+        action='store_true',
+        help="Only show percentage change, not a full diff.")
+    parser.add_argument(
+        '-b', '--by',
+        action='append',
+        choices=DataResult._by,
+        help="Group by this field.")
+    parser.add_argument(
+        '-f', '--field',
+        dest='fields',
+        action='append',
+        choices=DataResult._fields,
+        help="Show this field.")
+    parser.add_argument(
+        '-D', '--define',
+        dest='defines',
+        action='append',
+        type=lambda x: (lambda k,v: (k, set(v.split(','))))(*x.split('=', 1)),
+        help="Only include results where this field is this value.")
+    class AppendSort(argparse.Action):
+        def __call__(self, parser, namespace, value, option):
+            if namespace.sort is None:
+                namespace.sort = []
+            namespace.sort.append((value, True if option == '-S' else False))
+    parser.add_argument(
+        '-s', '--sort',
+        nargs='?',
+        action=AppendSort,
+        help="Sort by this field.")
+    parser.add_argument(
+        '-S', '--reverse-sort',
+        nargs='?',
+        action=AppendSort,
+        help="Sort by this field, but backwards.")
+    parser.add_argument(
+        '-Y', '--summary',
+        action='store_true',
+        help="Only show the total.")
+    parser.add_argument(
+        '-F', '--source',
+        dest='sources',
+        action='append',
+        help="Only consider definitions in this file. Defaults to anything "
+            "in the current directory.")
+    parser.add_argument(
+        '--everything',
+        action='store_true',
+        help="Include builtin and libc specific symbols.")
+    parser.add_argument(
+        '--nm-types',
+        default=NM_TYPES,
+        help="Type of symbols to report, this uses the same single-character "
+            "type-names emitted by nm. Defaults to %r." % NM_TYPES)
+    parser.add_argument(
+        '--nm-path',
+        type=lambda x: x.split(),
+        default=NM_PATH,
+        help="Path to the nm executable, may include flags. "
+            "Defaults to %r." % NM_PATH)
+    parser.add_argument(
+        '--objdump-path',
+        type=lambda x: x.split(),
+        default=OBJDUMP_PATH,
+        help="Path to the objdump executable, may include flags. "
+            "Defaults to %r." % OBJDUMP_PATH)
+    sys.exit(main(**{k: v
+        for k, v in vars(parser.parse_intermixed_args()).items()
+        if v is not None}))

kernel/fs/littlefs/scripts/prettyasserts.py

@@ -0,0 +1,478 @@
+#!/usr/bin/env python3
+#
+# Preprocessor that makes asserts easier to debug.
+#
+# Example:
+# ./scripts/prettyasserts.py -p LFS_ASSERT lfs.c -o lfs.a.c
+#
+# Copyright (c) 2022, The littlefs authors.
+# Copyright (c) 2020, Arm Limited. All rights reserved.
+# SPDX-License-Identifier: BSD-3-Clause
+#
+
+import re
+import sys
+
+# NOTE the use of macros here helps keep a consistent stack depth which
+# tools may rely on.
+#
+# If compilation errors are noisy consider using -ftrack-macro-expansion=0.
+#
+
+LIMIT = 16
+
+CMP = {
+    '==': 'eq',
+    '!=': 'ne',
+    '<=': 'le',
+    '>=': 'ge',
+    '<':  'lt',
+    '>':  'gt',
+}
+
+LEXEMES = {
+    'ws':       [r'(?:\s|\n|#.*?\n|//.*?\n|/\*.*?\*/)+'],
+    'assert':   ['assert'],
+    'arrow':    ['=>'],
+    'string':   [r'"(?:\\.|[^"])*"', r"'(?:\\.|[^'])\'"],
+    'paren':    [r'\(', r'\)'],
+    'cmp':      CMP.keys(),
+    'logic':    [r'\&\&', r'\|\|'],
+    'sep':      [':', ';', r'\{', r'\}', ','],
+    'op':       ['->'], # specifically ops that conflict with cmp
+}
+
+
+def openio(path, mode='r', buffering=-1):
+    # allow '-' for stdin/stdout
+    if path == '-':
+        if mode == 'r':
+            return os.fdopen(os.dup(sys.stdin.fileno()), mode, buffering)
+        else:
+            return os.fdopen(os.dup(sys.stdout.fileno()), mode, buffering)
+    else:
+        return open(path, mode, buffering)
+
+def write_header(f, limit=LIMIT):
+    f.writeln("// Generated by %s:" % sys.argv[0])
+    f.writeln("//")
+    f.writeln("// %s" % ' '.join(sys.argv))
+    f.writeln("//")
+    f.writeln()
+
+    f.writeln("#include <stdbool.h>")
+    f.writeln("#include <stdint.h>")
+    f.writeln("#include <inttypes.h>")
+    f.writeln("#include <stdio.h>")
+    f.writeln("#include <string.h>")
+    f.writeln("#include <signal.h>")
+    # give source a chance to define feature macros
+    f.writeln("#undef _FEATURES_H")
+    f.writeln()
+
+    # write print macros
+    f.writeln("__attribute__((unused))")
+    f.writeln("static void __pretty_assert_print_bool(")
+    f.writeln("        const void *v, size_t size) {")
+    f.writeln("    (void)size;")
+    f.writeln("    printf(\"%s\", *(const bool*)v ? \"true\" : \"false\");")
+    f.writeln("}")
+    f.writeln()
+    f.writeln("__attribute__((unused))")
+    f.writeln("static void __pretty_assert_print_int(")
+    f.writeln("        const void *v, size_t size) {")
+    f.writeln("    (void)size;")
+    f.writeln("    printf(\"%\"PRIiMAX, *(const intmax_t*)v);")
+    f.writeln("}")
+    f.writeln()
+    f.writeln("__attribute__((unused))")
+    f.writeln("static void __pretty_assert_print_ptr(")
+    f.writeln("        const void *v, size_t size) {")
+    f.writeln("    (void)size;")
+    f.writeln("    printf(\"%p\", v);")
+    f.writeln("}")
+    f.writeln()
+    f.writeln("__attribute__((unused))")
+    f.writeln("static void __pretty_assert_print_mem(")
+    f.writeln("        const void *v, size_t size) {")
+    f.writeln("    const uint8_t *v_ = v;")
+    f.writeln("    printf(\"\\\"\");")
+    f.writeln("    for (size_t i = 0; i < size && i < %d; i++) {" % limit)
+    f.writeln("        if (v_[i] >= ' ' && v_[i] <= '~') {")
+    f.writeln("            printf(\"%c\", v_[i]);")
+    f.writeln("        } else {")
+    f.writeln("            printf(\"\\\\x%02x\", v_[i]);")
+    f.writeln("        }")
+    f.writeln("    }")
+    f.writeln("    if (size > %d) {" % limit)
+    f.writeln("        printf(\"...\");")
+    f.writeln("    }")
+    f.writeln("    printf(\"\\\"\");")
+    f.writeln("}")
+    f.writeln()
+    f.writeln("__attribute__((unused))")
+    f.writeln("static void __pretty_assert_print_str(")
+    f.writeln("        const void *v, size_t size) {")
+    f.writeln("    __pretty_assert_print_mem(v, size);")
+    f.writeln("}")
+    f.writeln()
+    f.writeln("__attribute__((unused, noinline))")
+    f.writeln("static void __pretty_assert_fail(")
+    f.writeln("        const char *file, int line,")
+    f.writeln("        void (*type_print_cb)(const void*, size_t),")
+    f.writeln("        const char *cmp,")
+    f.writeln("        const void *lh, size_t lsize,")
+    f.writeln("        const void *rh, size_t rsize) {")
+    f.writeln("    printf(\"%s:%d:assert: assert failed with \", file, line);")
+    f.writeln("    type_print_cb(lh, lsize);")
+    f.writeln("    printf(\", expected %s \", cmp);")
+    f.writeln("    type_print_cb(rh, rsize);")
+    f.writeln("    printf(\"\\n\");")
+    f.writeln("    fflush(NULL);")
+    f.writeln("    raise(SIGABRT);")
+    f.writeln("}")
+    f.writeln()
+
+    # write assert macros
+    for op, cmp in sorted(CMP.items()):
+        f.writeln("#define __PRETTY_ASSERT_BOOL_%s(lh, rh) do { \\"
+            % cmp.upper())
+        f.writeln("    bool _lh = !!(lh); \\")
+        f.writeln("    bool _rh = !!(rh); \\")
+        f.writeln("    if (!(_lh %s _rh)) { \\" % op)
+        f.writeln("        __pretty_assert_fail( \\")
+        f.writeln("                __FILE__, __LINE__, \\")
+        f.writeln("                __pretty_assert_print_bool, \"%s\", \\"
+            % cmp)
+        f.writeln("                &_lh, 0, \\")
+        f.writeln("                &_rh, 0); \\")
+        f.writeln("    } \\")
+        f.writeln("} while (0)")
+    for op, cmp in sorted(CMP.items()):
+        f.writeln("#define __PRETTY_ASSERT_INT_%s(lh, rh) do { \\"
+            % cmp.upper())
+        f.writeln("    __typeof__(lh) _lh = lh; \\")
+        f.writeln("    __typeof__(lh) _rh = rh; \\")
+        f.writeln("    if (!(_lh %s _rh)) { \\" % op)
+        f.writeln("        __pretty_assert_fail( \\")
+        f.writeln("                __FILE__, __LINE__, \\")
+        f.writeln("                __pretty_assert_print_int, \"%s\", \\"
+            % cmp)
+        f.writeln("                &(intmax_t){_lh}, 0, \\")
+        f.writeln("                &(intmax_t){_rh}, 0); \\")
+        f.writeln("    } \\")
+        f.writeln("} while (0)")
+    for op, cmp in sorted(CMP.items()):
+        f.writeln("#define __PRETTY_ASSERT_MEM_%s(lh, rh, size) do { \\"
+            % cmp.upper())
+        f.writeln("    const void *_lh = lh; \\")
+        f.writeln("    const void *_rh = rh; \\")
+        f.writeln("    if (!(memcmp(_lh, _rh, size) %s 0)) { \\" % op)
+        f.writeln("        __pretty_assert_fail( \\")
+        f.writeln("                __FILE__, __LINE__, \\")
+        f.writeln("                __pretty_assert_print_mem, \"%s\", \\"
+            % cmp)
+        f.writeln("                _lh, size, \\")
+        f.writeln("                _rh, size); \\")
+        f.writeln("    } \\")
+        f.writeln("} while (0)")
+    for op, cmp in sorted(CMP.items()):
+        f.writeln("#define __PRETTY_ASSERT_STR_%s(lh, rh) do { \\"
+            % cmp.upper())
+        f.writeln("    const char *_lh = lh; \\")
+        f.writeln("    const char *_rh = rh; \\")
+        f.writeln("    if (!(strcmp(_lh, _rh) %s 0)) { \\" % op)
+        f.writeln("        __pretty_assert_fail( \\")
+        f.writeln("                __FILE__, __LINE__, \\")
+        f.writeln("                __pretty_assert_print_str, \"%s\", \\"
+            % cmp)
+        f.writeln("                _lh, strlen(_lh), \\")
+        f.writeln("                _rh, strlen(_rh)); \\")
+        f.writeln("    } \\")
+        f.writeln("} while (0)")
+    for op, cmp in sorted(CMP.items()):
+        # Only EQ and NE are supported when compared to NULL.
+        if cmp not in ['eq', 'ne']:
+            continue
+        f.writeln("#define __PRETTY_ASSERT_PTR_%s(lh, rh) do { \\"
+            % cmp.upper())
+        f.writeln("    const void *_lh = (const void*)(uintptr_t)lh; \\")
+        f.writeln("    const void *_rh = (const void*)(uintptr_t)rh; \\")
+        f.writeln("    if (!(_lh %s _rh)) { \\" % op)
+        f.writeln("        __pretty_assert_fail( \\")
+        f.writeln("                __FILE__, __LINE__, \\")
+        f.writeln("                __pretty_assert_print_ptr, \"%s\", \\"
+            % cmp)
+        f.writeln("                (const void*){_lh}, 0, \\")
+        f.writeln("                (const void*){_rh}, 0); \\")
+        f.writeln("    } \\")
+        f.writeln("} while (0)")
+    f.writeln()
+    f.writeln()
+
+def mkassert(type, cmp, lh, rh, size=None):
+    if size is not None:
+        return ("__PRETTY_ASSERT_%s_%s(%s, %s, %s)"
+            % (type.upper(), cmp.upper(), lh, rh, size))
+    else:
+        return ("__PRETTY_ASSERT_%s_%s(%s, %s)"
+            % (type.upper(), cmp.upper(), lh, rh))
+
+
+# simple recursive descent parser
+class ParseFailure(Exception):
+    def __init__(self, expected, found):
+        self.expected = expected
+        self.found = found
+
+    def __str__(self):
+        return "expected %r, found %s..." % (
+            self.expected, repr(self.found)[:70])
+
+class Parser:
+    def __init__(self, in_f, lexemes=LEXEMES):
+        p = '|'.join('(?P<%s>%s)' % (n, '|'.join(l))
+            for n, l in lexemes.items())
+        p = re.compile(p, re.DOTALL)
+        data = in_f.read()
+        tokens = []
+        line = 1
+        col = 0
+        while True:
+            m = p.search(data)
+            if m:
+                if m.start() > 0:
+                    tokens.append((None, data[:m.start()], line, col))
+                tokens.append((m.lastgroup, m.group(), line, col))
+                data = data[m.end():]
+            else:
+                tokens.append((None, data, line, col))
+                break
+        self.tokens = tokens
+        self.off = 0
+
+    def lookahead(self, *pattern):
+        if self.off < len(self.tokens):
+            token = self.tokens[self.off]
+            if token[0] in pattern or token[1] in pattern:
+                self.m = token[1]
+                return self.m
+        self.m = None
+        return self.m
+
+    def accept(self, *patterns):
+        m = self.lookahead(*patterns)
+        if m is not None:
+            self.off += 1
+        return m
+
+    def expect(self, *patterns):
+        m = self.accept(*patterns)
+        if not m:
+            raise ParseFailure(patterns, self.tokens[self.off:])
+        return m
+
+    def push(self):
+        return self.off
+
+    def pop(self, state):
+        self.off = state
+
+def p_assert(p):
+    state = p.push()
+
+    # assert(memcmp(a,b,size) cmp 0)?
+    try:
+        p.expect('assert') ; p.accept('ws')
+        p.expect('(') ; p.accept('ws')
+        p.expect('memcmp') ; p.accept('ws')
+        p.expect('(') ; p.accept('ws')
+        lh = p_expr(p) ; p.accept('ws')
+        p.expect(',') ; p.accept('ws')
+        rh = p_expr(p) ; p.accept('ws')
+        p.expect(',') ; p.accept('ws')
+        size = p_expr(p) ; p.accept('ws')
+        p.expect(')') ; p.accept('ws')
+        cmp = p.expect('cmp') ; p.accept('ws')
+        p.expect('0') ; p.accept('ws')
+        p.expect(')')
+        return mkassert('mem', CMP[cmp], lh, rh, size)
+    except ParseFailure:
+        p.pop(state)
+
+    # assert(strcmp(a,b) cmp 0)?
+    try:
+        p.expect('assert') ; p.accept('ws')
+        p.expect('(') ; p.accept('ws')
+        p.expect('strcmp') ; p.accept('ws')
+        p.expect('(') ; p.accept('ws')
+        lh = p_expr(p) ; p.accept('ws')
+        p.expect(',') ; p.accept('ws')
+        rh = p_expr(p) ; p.accept('ws')
+        p.expect(')') ; p.accept('ws')
+        cmp = p.expect('cmp') ; p.accept('ws')
+        p.expect('0') ; p.accept('ws')
+        p.expect(')')
+        return mkassert('str', CMP[cmp], lh, rh)
+    except ParseFailure:
+        p.pop(state)
+
+    # assert(a cmp b)?
+    try:
+        p.expect('assert') ; p.accept('ws')
+        p.expect('(') ; p.accept('ws')
+        lh = p_expr(p) ; p.accept('ws')
+        cmp = p.expect('cmp') ; p.accept('ws')
+        rh = p_expr(p) ; p.accept('ws')
+        p.expect(')')
+        if rh == 'NULL' or lh == 'NULL':
+            return mkassert('ptr', CMP[cmp], lh, rh)
+        return mkassert('int', CMP[cmp], lh, rh)
+    except ParseFailure:
+        p.pop(state)
+
+    # assert(a)?
+    p.expect('assert') ; p.accept('ws')
+    p.expect('(') ; p.accept('ws')
+    lh = p_exprs(p) ; p.accept('ws')
+    p.expect(')')
+    return mkassert('bool', 'eq', lh, 'true')
+
+def p_expr(p):
+    res = []
+    while True:
+        if p.accept('('):
+            res.append(p.m)
+            while True:
+                res.append(p_exprs(p))
+                if p.accept('sep'):
+                    res.append(p.m)
+                else:
+                    break
+            res.append(p.expect(')'))
+        elif p.lookahead('assert'):
+            state = p.push()
+            try:
+                res.append(p_assert(p))
+            except ParseFailure:
+                p.pop(state)
+                res.append(p.expect('assert'))
+        elif p.accept('string', 'op', 'ws', None):
+            res.append(p.m)
+        else:
+            return ''.join(res)
+
+def p_exprs(p):
+    res = []
+    while True:
+        res.append(p_expr(p))
+        if p.accept('cmp', 'logic', ','):
+            res.append(p.m)
+        else:
+            return ''.join(res)
+
+def p_stmt(p):
+    ws = p.accept('ws') or ''
+
+    # memcmp(lh,rh,size) => 0?
+    if p.lookahead('memcmp'):
+        state = p.push()
+        try:
+            p.expect('memcmp') ; p.accept('ws')
+            p.expect('(') ; p.accept('ws')
+            lh = p_expr(p) ; p.accept('ws')
+            p.expect(',') ; p.accept('ws')
+            rh = p_expr(p) ; p.accept('ws')
+            p.expect(',') ; p.accept('ws')
+            size = p_expr(p) ; p.accept('ws')
+            p.expect(')') ; p.accept('ws')
+            p.expect('=>') ; p.accept('ws')
+            p.expect('0') ; p.accept('ws')
+            return ws + mkassert('mem', 'eq', lh, rh, size)
+        except ParseFailure:
+            p.pop(state)
+
+    # strcmp(lh,rh) => 0?
+    if p.lookahead('strcmp'):
+        state = p.push()
+        try:
+            p.expect('strcmp') ; p.accept('ws') ; p.expect('(') ; p.accept('ws')
+            lh = p_expr(p) ; p.accept('ws')
+            p.expect(',') ; p.accept('ws')
+            rh = p_expr(p) ; p.accept('ws')
+            p.expect(')') ; p.accept('ws')
+            p.expect('=>') ; p.accept('ws')
+            p.expect('0') ; p.accept('ws')
+            return ws + mkassert('str', 'eq', lh, rh)
+        except ParseFailure:
+            p.pop(state)
+
+    # lh => rh?
+    lh = p_exprs(p)
+    if p.accept('=>'):
+        rh = p_exprs(p)
+        return ws + mkassert('int', 'eq', lh, rh)
+    else:
+        return ws + lh
+
+def main(input=None, output=None, pattern=[], limit=LIMIT):
+    with openio(input or '-', 'r') as in_f:
+        # create parser
+        lexemes = LEXEMES.copy()
+        lexemes['assert'] += pattern
+        p = Parser(in_f, lexemes)
+
+        with openio(output or '-', 'w') as f:
+            def writeln(s=''):
+                f.write(s)
+                f.write('\n')
+            f.writeln = writeln
+
+            # write extra verbose asserts
+            write_header(f, limit=limit)
+            if input is not None:
+                f.writeln("#line %d \"%s\"" % (1, input))
+
+            # parse and write out stmt at a time
+            try:
+                while True:
+                    f.write(p_stmt(p))
+                    if p.accept('sep'):
+                        f.write(p.m)
+                    else:
+                        break
+            except ParseFailure as e:
+                print('warning: %s' % e)
+                pass
+
+            for i in range(p.off, len(p.tokens)):
+                f.write(p.tokens[i][1])
+
+
+if __name__ == "__main__":
+    import argparse
+    import sys
+    parser = argparse.ArgumentParser(
+        description="Preprocessor that makes asserts easier to debug.",
+        allow_abbrev=False)
+    parser.add_argument(
+        'input',
+        help="Input C file.")
+    parser.add_argument(
+        '-o', '--output',
+        required=True,
+        help="Output C file.")
+    parser.add_argument(
+        '-p', '--pattern',
+        action='append',
+        help="Regex patterns to search for starting an assert statement. This"
+            " implicitly includes \"assert\" and \"=>\".")
+    parser.add_argument(
+        '-l', '--limit',
+        type=lambda x: int(x, 0),
+        default=LIMIT,
+        help="Maximum number of characters to display in strcmp and memcmp. "
+            "Defaults to %r." % LIMIT)
+    sys.exit(main(**{k: v
+        for k, v in vars(parser.parse_intermixed_args()).items()
+        if v is not None}))

kernel/fs/littlefs/scripts/readblock.py

@@ -0,0 +1,26 @@
+#!/usr/bin/env python3
+
+import subprocess as sp
+
+def main(args):
+    with open(args.disk, 'rb') as f:
+        f.seek(args.block * args.block_size)
+        block = (f.read(args.block_size)
+            .ljust(args.block_size, b'\xff'))
+
+    # what did you expect?
+    print("%-8s  %-s" % ('off', 'data'))
+    return sp.run(['xxd', '-g1', '-'], input=block).returncode
+
+if __name__ == "__main__":
+    import argparse
+    import sys
+    parser = argparse.ArgumentParser(
+        description="Hex dump a specific block in a disk.")
+    parser.add_argument('disk',
+        help="File representing the block device.")
+    parser.add_argument('block_size', type=lambda x: int(x, 0),
+        help="Size of a block in bytes.")
+    parser.add_argument('block', type=lambda x: int(x, 0),
+        help="Address of block to dump.")
+    sys.exit(main(parser.parse_args()))

kernel/fs/littlefs/scripts/readmdir.py

@@ -0,0 +1,399 @@
+#!/usr/bin/env python3
+
+import struct
+import binascii
+import sys
+import itertools as it
+
+TAG_TYPES = {
+    'splice':       (0x700, 0x400),
+    'create':       (0x7ff, 0x401),
+    'delete':       (0x7ff, 0x4ff),
+    'name':         (0x700, 0x000),
+    'reg':          (0x7ff, 0x001),
+    'dir':          (0x7ff, 0x002),
+    'superblock':   (0x7ff, 0x0ff),
+    'struct':       (0x700, 0x200),
+    'dirstruct':    (0x7ff, 0x200),
+    'ctzstruct':    (0x7ff, 0x202),
+    'inlinestruct': (0x7ff, 0x201),
+    'userattr':     (0x700, 0x300),
+    'tail':         (0x700, 0x600),
+    'softtail':     (0x7ff, 0x600),
+    'hardtail':     (0x7ff, 0x601),
+    'gstate':       (0x700, 0x700),
+    'movestate':    (0x7ff, 0x7ff),
+    'crc':          (0x700, 0x500),
+    'ccrc':         (0x780, 0x500),
+    'fcrc':         (0x7ff, 0x5ff),
+}
+
+class Tag:
+    def __init__(self, *args):
+        if len(args) == 1:
+            self.tag = args[0]
+        elif len(args) == 3:
+            if isinstance(args[0], str):
+                type = TAG_TYPES[args[0]][1]
+            else:
+                type = args[0]
+
+            if isinstance(args[1], str):
+                id = int(args[1], 0) if args[1] not in 'x.' else 0x3ff
+            else:
+                id = args[1]
+
+            if isinstance(args[2], str):
+                size = int(args[2], str) if args[2] not in 'x.' else 0x3ff
+            else:
+                size = args[2]
+
+            self.tag = (type << 20) | (id << 10) | size
+        else:
+            assert False
+
+    @property
+    def isvalid(self):
+        return not bool(self.tag & 0x80000000)
+
+    @property
+    def isattr(self):
+        return not bool(self.tag & 0x40000000)
+
+    @property
+    def iscompactable(self):
+        return bool(self.tag & 0x20000000)
+
+    @property
+    def isunique(self):
+        return not bool(self.tag & 0x10000000)
+
+    @property
+    def type(self):
+        return (self.tag & 0x7ff00000) >> 20
+
+    @property
+    def type1(self):
+        return (self.tag & 0x70000000) >> 20
+
+    @property
+    def type3(self):
+        return (self.tag & 0x7ff00000) >> 20
+
+    @property
+    def id(self):
+        return (self.tag & 0x000ffc00) >> 10
+
+    @property
+    def size(self):
+        return (self.tag & 0x000003ff) >> 0
+
+    @property
+    def dsize(self):
+        return 4 + (self.size if self.size != 0x3ff else 0)
+
+    @property
+    def chunk(self):
+        return self.type & 0xff
+
+    @property
+    def schunk(self):
+        return struct.unpack('b', struct.pack('B', self.chunk))[0]
+
+    def is_(self, type):
+        try:
+            if ' ' in type:
+                type1, type3 = type.split()
+                return (self.is_(type1) and
+                    (self.type & ~TAG_TYPES[type1][0]) == int(type3, 0))
+
+            return self.type == int(type, 0)
+
+        except (ValueError, KeyError):
+            return (self.type & TAG_TYPES[type][0]) == TAG_TYPES[type][1]
+
+    def mkmask(self):
+        return Tag(
+            0x700 if self.isunique else 0x7ff,
+            0x3ff if self.isattr else 0,
+            0)
+
+    def chid(self, nid):
+        ntag = Tag(self.type, nid, self.size)
+        if hasattr(self, 'off'):    ntag.off    = self.off
+        if hasattr(self, 'data'):   ntag.data   = self.data
+        if hasattr(self, 'ccrc'):   ntag.crc    = self.crc
+        if hasattr(self, 'erased'): ntag.erased = self.erased
+        return ntag
+
+    def typerepr(self):
+        if (self.is_('ccrc')
+                and getattr(self, 'ccrc', 0xffffffff) != 0xffffffff):
+            crc_status = ' (bad)'
+        elif self.is_('fcrc') and getattr(self, 'erased', False):
+            crc_status = ' (era)'
+        else:
+            crc_status = ''
+
+        reverse_types = {v: k for k, v in TAG_TYPES.items()}
+        for prefix in range(12):
+            mask = 0x7ff & ~((1 << prefix)-1)
+            if (mask, self.type & mask) in reverse_types:
+                type = reverse_types[mask, self.type & mask]
+                if prefix > 0:
+                    return '%s %#x%s' % (
+                        type, self.type & ((1 << prefix)-1), crc_status)
+                else:
+                    return '%s%s' % (type, crc_status)
+        else:
+            return '%02x%s' % (self.type, crc_status)
+
+    def idrepr(self):
+        return repr(self.id) if self.id != 0x3ff else '.'
+
+    def sizerepr(self):
+        return repr(self.size) if self.size != 0x3ff else 'x'
+
+    def __repr__(self):
+        return 'Tag(%r, %d, %d)' % (self.typerepr(), self.id, self.size)
+
+    def __lt__(self, other):
+        return (self.id, self.type) < (other.id, other.type)
+
+    def __bool__(self):
+        return self.isvalid
+
+    def __int__(self):
+        return self.tag
+
+    def __index__(self):
+        return self.tag
+
+class MetadataPair:
+    def __init__(self, blocks):
+        if len(blocks) > 1:
+            self.pair = [MetadataPair([block]) for block in blocks]
+            self.pair = sorted(self.pair, reverse=True)
+
+            self.data = self.pair[0].data
+            self.rev  = self.pair[0].rev
+            self.tags = self.pair[0].tags
+            self.ids  = self.pair[0].ids
+            self.log  = self.pair[0].log
+            self.all_ = self.pair[0].all_
+            return
+
+        self.pair = [self]
+        self.data = blocks[0]
+        block = self.data
+
+        self.rev, = struct.unpack('<I', block[0:4])
+        crc = binascii.crc32(block[0:4])
+        fcrctag = None
+        fcrcdata = None
+
+        # parse tags
+        corrupt = False
+        tag = Tag(0xffffffff)
+        off = 4
+        self.log = []
+        self.all_ = []
+        while len(block) - off >= 4:
+            ntag, = struct.unpack('>I', block[off:off+4])
+
+            tag = Tag((int(tag) ^ ntag) & 0x7fffffff)
+            tag.off = off + 4
+            tag.data = block[off+4:off+tag.dsize]
+            if tag.is_('ccrc'):
+                crc = binascii.crc32(block[off:off+2*4], crc)
+            else:
+                crc = binascii.crc32(block[off:off+tag.dsize], crc)
+            tag.crc = crc
+            off += tag.dsize
+
+            self.all_.append(tag)
+
+            if tag.is_('fcrc') and len(tag.data) == 8:
+                fcrctag = tag
+                fcrcdata = struct.unpack('<II', tag.data)
+            elif tag.is_('ccrc'):
+                # is valid commit?
+                if crc != 0xffffffff:
+                    corrupt = True
+                if not corrupt:
+                    self.log = self.all_.copy()
+                    # end of commit?
+                    if fcrcdata:
+                        fcrcsize, fcrc = fcrcdata
+                        fcrc_ = 0xffffffff ^ binascii.crc32(
+                            block[off:off+fcrcsize])
+                        if fcrc_ == fcrc:
+                            fcrctag.erased = True
+                            corrupt = True
+
+                # reset tag parsing
+                crc = 0
+                tag = Tag(int(tag) ^ ((tag.type & 1) << 31))
+                fcrctag = None
+                fcrcdata = None
+
+        # find active ids
+        self.ids = list(it.takewhile(
+            lambda id: Tag('name', id, 0) in self,
+            it.count()))
+
+        # find most recent tags
+        self.tags = []
+        for tag in self.log:
+            if tag.is_('crc') or tag.is_('splice'):
+                continue
+            elif tag.id == 0x3ff:
+                if tag in self and self[tag] is tag:
+                    self.tags.append(tag)
+            else:
+                # id could have change, I know this is messy and slow
+                # but it works
+                for id in self.ids:
+                    ntag = tag.chid(id)
+                    if ntag in self and self[ntag] is tag:
+                        self.tags.append(ntag)
+
+        self.tags = sorted(self.tags)
+
+    def __bool__(self):
+        return bool(self.log)
+
+    def __lt__(self, other):
+        # corrupt blocks don't count
+        if not self or not other:
+            return bool(other)
+
+        # use sequence arithmetic to avoid overflow
+        return not ((other.rev - self.rev) & 0x80000000)
+
+    def __contains__(self, args):
+        try:
+            self[args]
+            return True
+        except KeyError:
+            return False
+
+    def __getitem__(self, args):
+        if isinstance(args, tuple):
+            gmask, gtag = args
+        else:
+            gmask, gtag = args.mkmask(), args
+
+        gdiff = 0
+        for tag in reversed(self.log):
+            if (gmask.id != 0 and tag.is_('splice') and
+                    tag.id <= gtag.id - gdiff):
+                if tag.is_('create') and tag.id == gtag.id - gdiff:
+                    # creation point
+                    break
+
+                gdiff += tag.schunk
+
+            if ((int(gmask) & int(tag)) ==
+                    (int(gmask) & int(gtag.chid(gtag.id - gdiff)))):
+                if tag.size == 0x3ff:
+                    # deleted
+                    break
+
+                return tag
+
+        raise KeyError(gmask, gtag)
+
+    def _dump_tags(self, tags, f=sys.stdout, truncate=True):
+        f.write("%-8s  %-8s  %-13s %4s %4s" % (
+            'off', 'tag', 'type', 'id', 'len'))
+        if truncate:
+            f.write('  data (truncated)')
+        f.write('\n')
+
+        for tag in tags:
+            f.write("%08x: %08x  %-14s %3s %4s" % (
+                tag.off, tag,
+                tag.typerepr(), tag.idrepr(), tag.sizerepr()))
+            if truncate:
+                f.write("  %-23s  %-8s\n" % (
+                    ' '.join('%02x' % c for c in tag.data[:8]),
+                    ''.join(c if c >= ' ' and c <= '~' else '.'
+                        for c in map(chr, tag.data[:8]))))
+            else:
+                f.write("\n")
+                for i in range(0, len(tag.data), 16):
+                    f.write("  %08x: %-47s  %-16s\n" % (
+                        tag.off+i,
+                        ' '.join('%02x' % c for c in tag.data[i:i+16]),
+                        ''.join(c if c >= ' ' and c <= '~' else '.'
+                            for c in map(chr, tag.data[i:i+16]))))
+
+    def dump_tags(self, f=sys.stdout, truncate=True):
+        self._dump_tags(self.tags, f=f, truncate=truncate)
+
+    def dump_log(self, f=sys.stdout, truncate=True):
+        self._dump_tags(self.log, f=f, truncate=truncate)
+
+    def dump_all(self, f=sys.stdout, truncate=True):
+        self._dump_tags(self.all_, f=f, truncate=truncate)
+
+def main(args):
+    blocks = []
+    with open(args.disk, 'rb') as f:
+        for block in [args.block1, args.block2]:
+            if block is None:
+                continue
+            f.seek(block * args.block_size)
+            blocks.append(f.read(args.block_size)
+                .ljust(args.block_size, b'\xff'))
+
+    # find most recent pair
+    mdir = MetadataPair(blocks)
+
+    try:
+        mdir.tail = mdir[Tag('tail', 0, 0)]
+        if mdir.tail.size != 8 or mdir.tail.data == 8*b'\xff':
+            mdir.tail = None
+    except KeyError:
+        mdir.tail = None
+
+    print("mdir {%s} rev %d%s%s%s" % (
+        ', '.join('%#x' % b
+            for b in [args.block1, args.block2]
+            if b is not None),
+        mdir.rev,
+        ' (was %s)' % ', '.join('%d' % m.rev for m in mdir.pair[1:])
+        if len(mdir.pair) > 1 else '',
+        ' (corrupted!)' if not mdir else '',
+        ' -> {%#x, %#x}' % struct.unpack('<II', mdir.tail.data)
+        if mdir.tail else ''))
+    if args.all:
+        mdir.dump_all(truncate=not args.no_truncate)
+    elif args.log:
+        mdir.dump_log(truncate=not args.no_truncate)
+    else:
+        mdir.dump_tags(truncate=not args.no_truncate)
+
+    return 0 if mdir else 1
+
+if __name__ == "__main__":
+    import argparse
+    import sys
+    parser = argparse.ArgumentParser(
+        description="Dump useful info about metadata pairs in littlefs.")
+    parser.add_argument('disk',
+        help="File representing the block device.")
+    parser.add_argument('block_size', type=lambda x: int(x, 0),
+        help="Size of a block in bytes.")
+    parser.add_argument('block1', type=lambda x: int(x, 0),
+        help="First block address for finding the metadata pair.")
+    parser.add_argument('block2', nargs='?', type=lambda x: int(x, 0),
+        help="Second block address for finding the metadata pair.")
+    parser.add_argument('-l', '--log', action='store_true',
+        help="Show tags in log.")
+    parser.add_argument('-a', '--all', action='store_true',
+        help="Show all tags in log, included tags in corrupted commits.")
+    parser.add_argument('-T', '--no-truncate', action='store_true',
+        help="Don't truncate large amounts of data.")
+    sys.exit(main(parser.parse_args()))

kernel/fs/littlefs/scripts/readtree.py

@@ -0,0 +1,183 @@
+#!/usr/bin/env python3
+
+import struct
+import sys
+import json
+import io
+import itertools as it
+from readmdir import Tag, MetadataPair
+
+def main(args):
+    superblock = None
+    gstate = b'\0\0\0\0\0\0\0\0\0\0\0\0'
+    dirs = []
+    mdirs = []
+    corrupted = []
+    cycle = False
+    with open(args.disk, 'rb') as f:
+        tail = (args.block1, args.block2)
+        hard = False
+        while True:
+            for m in it.chain((m for d in dirs for m in d), mdirs):
+                if set(m.blocks) == set(tail):
+                    # cycle detected
+                    cycle = m.blocks
+            if cycle:
+                break
+
+            # load mdir
+            data = []
+            blocks = {}
+            for block in tail:
+                f.seek(block * args.block_size)
+                data.append(f.read(args.block_size)
+                    .ljust(args.block_size, b'\xff'))
+                blocks[id(data[-1])] = block
+
+            mdir = MetadataPair(data)
+            mdir.blocks = tuple(blocks[id(p.data)] for p in mdir.pair)
+
+            # fetch some key metadata as a we scan
+            try:
+                mdir.tail = mdir[Tag('tail', 0, 0)]
+                if mdir.tail.size != 8 or mdir.tail.data == 8*b'\xff':
+                    mdir.tail = None
+            except KeyError:
+                mdir.tail = None
+
+            # have superblock?
+            try:
+                nsuperblock = mdir[
+                    Tag(0x7ff, 0x3ff, 0), Tag('superblock', 0, 0)]
+                superblock = nsuperblock, mdir[Tag('inlinestruct', 0, 0)]
+            except KeyError:
+                pass
+
+            # have gstate?
+            try:
+                ngstate = mdir[Tag('movestate', 0, 0)]
+                gstate = bytes((a or 0) ^ (b or 0)
+                    for a,b in it.zip_longest(gstate, ngstate.data))
+            except KeyError:
+                pass
+
+            # corrupted?
+            if not mdir:
+                corrupted.append(mdir)
+
+            # add to directories
+            mdirs.append(mdir)
+            if mdir.tail is None or not mdir.tail.is_('hardtail'):
+                dirs.append(mdirs)
+                mdirs = []
+
+            if mdir.tail is None:
+                break
+
+            tail = struct.unpack('<II', mdir.tail.data)
+            hard = mdir.tail.is_('hardtail')
+
+    # find paths
+    dirtable = {}
+    for dir in dirs:
+        dirtable[frozenset(dir[0].blocks)] = dir
+
+    pending = [("/", dirs[0])]
+    while pending:
+        path, dir = pending.pop(0)
+        for mdir in dir:
+            for tag in mdir.tags:
+                if tag.is_('dir'):
+                    try:
+                        npath = tag.data.decode('utf8')
+                        dirstruct = mdir[Tag('dirstruct', tag.id, 0)]
+                        nblocks = struct.unpack('<II', dirstruct.data)
+                        nmdir = dirtable[frozenset(nblocks)]
+                        pending.append(((path + '/' + npath), nmdir))
+                    except KeyError:
+                        pass
+
+        dir[0].path = path.replace('//', '/')
+
+    # print littlefs + version info
+    version = ('?', '?')
+    if superblock:
+        version = tuple(reversed(
+            struct.unpack('<HH', superblock[1].data[0:4].ljust(4, b'\xff'))))
+    print("%-47s%s" % ("littlefs v%s.%s" % version,
+        "data (truncated, if it fits)"
+        if not any([args.no_truncate, args.log, args.all]) else ""))
+
+    # print gstate
+    print("gstate 0x%s" % ''.join('%02x' % c for c in gstate))
+    tag = Tag(struct.unpack('<I', gstate[0:4].ljust(4, b'\xff'))[0])
+    blocks = struct.unpack('<II', gstate[4:4+8].ljust(8, b'\xff'))
+    if tag.size or not tag.isvalid:
+        print("  orphans >=%d" % max(tag.size, 1))
+    if tag.type:
+        print("  move dir {%#x, %#x} id %d" % (
+            blocks[0], blocks[1], tag.id))
+
+    # print mdir info
+    for i, dir in enumerate(dirs):
+        print("dir %s" % (json.dumps(dir[0].path)
+            if hasattr(dir[0], 'path') else '(orphan)'))
+
+        for j, mdir in enumerate(dir):
+            print("mdir {%#x, %#x} rev %d (was %d)%s%s" % (
+                mdir.blocks[0], mdir.blocks[1], mdir.rev, mdir.pair[1].rev,
+                ' (corrupted!)' if not mdir else '',
+                ' -> {%#x, %#x}' % struct.unpack('<II', mdir.tail.data)
+                if mdir.tail else ''))
+
+            f = io.StringIO()
+            if args.log:
+                mdir.dump_log(f, truncate=not args.no_truncate)
+            elif args.all:
+                mdir.dump_all(f, truncate=not args.no_truncate)
+            else:
+                mdir.dump_tags(f, truncate=not args.no_truncate)
+
+            lines = list(filter(None, f.getvalue().split('\n')))
+            for k, line in enumerate(lines):
+                print("%s %s" % (
+                    ' ' if j == len(dir)-1 else
+                    'v' if k == len(lines)-1 else
+                    '|',
+                    line))
+
+    errcode = 0
+    for mdir in corrupted:
+        errcode = errcode or 1
+        print("*** corrupted mdir {%#x, %#x}! ***" % (
+            mdir.blocks[0], mdir.blocks[1]))
+
+    if cycle:
+        errcode = errcode or 2
+        print("*** cycle detected {%#x, %#x}! ***" % (
+            cycle[0], cycle[1]))
+
+    return errcode
+
+if __name__ == "__main__":
+    import argparse
+    import sys
+    parser = argparse.ArgumentParser(
+        description="Dump semantic info about the metadata tree in littlefs")
+    parser.add_argument('disk',
+        help="File representing the block device.")
+    parser.add_argument('block_size', type=lambda x: int(x, 0),
+        help="Size of a block in bytes.")
+    parser.add_argument('block1', nargs='?', default=0,
+        type=lambda x: int(x, 0),
+        help="Optional first block address for finding the superblock.")
+    parser.add_argument('block2', nargs='?', default=1,
+        type=lambda x: int(x, 0),
+        help="Optional second block address for finding the superblock.")
+    parser.add_argument('-l', '--log', action='store_true',
+        help="Show tags in log.")
+    parser.add_argument('-a', '--all', action='store_true',
+        help="Show all tags in log, included tags in corrupted commits.")
+    parser.add_argument('-T', '--no-truncate', action='store_true',
+        help="Show the full contents of files/attrs/tags.")
+    sys.exit(main(parser.parse_args()))

kernel/fs/littlefs/scripts/stack.py

@@ -0,0 +1,735 @@
+#!/usr/bin/env python3
+#
+# Script to find stack usage at the function level. Will detect recursion and
+# report as infinite stack usage.
+#
+# Example:
+# ./scripts/stack.py lfs.ci lfs_util.ci -Slimit
+#
+# Copyright (c) 2022, The littlefs authors.
+# SPDX-License-Identifier: BSD-3-Clause
+#
+
+import collections as co
+import csv
+import itertools as it
+import math as m
+import os
+import re
+
+
+
+# integer fields
+class Int(co.namedtuple('Int', 'x')):
+    __slots__ = ()
+    def __new__(cls, x=0):
+        if isinstance(x, Int):
+            return x
+        if isinstance(x, str):
+            try:
+                x = int(x, 0)
+            except ValueError:
+                # also accept +-∞ and +-inf
+                if re.match('^\s*\+?\s*(?:∞|inf)\s*$', x):
+                    x = m.inf
+                elif re.match('^\s*-\s*(?:∞|inf)\s*$', x):
+                    x = -m.inf
+                else:
+                    raise
+        assert isinstance(x, int) or m.isinf(x), x
+        return super().__new__(cls, x)
+
+    def __str__(self):
+        if self.x == m.inf:
+            return '∞'
+        elif self.x == -m.inf:
+            return '-∞'
+        else:
+            return str(self.x)
+
+    def __int__(self):
+        assert not m.isinf(self.x)
+        return self.x
+
+    def __float__(self):
+        return float(self.x)
+
+    none = '%7s' % '-'
+    def table(self):
+        return '%7s' % (self,)
+
+    diff_none = '%7s' % '-'
+    diff_table = table
+
+    def diff_diff(self, other):
+        new = self.x if self else 0
+        old = other.x if other else 0
+        diff = new - old
+        if diff == +m.inf:
+            return '%7s' % '+∞'
+        elif diff == -m.inf:
+            return '%7s' % '-∞'
+        else:
+            return '%+7d' % diff
+
+    def ratio(self, other):
+        new = self.x if self else 0
+        old = other.x if other else 0
+        if m.isinf(new) and m.isinf(old):
+            return 0.0
+        elif m.isinf(new):
+            return +m.inf
+        elif m.isinf(old):
+            return -m.inf
+        elif not old and not new:
+            return 0.0
+        elif not old:
+            return 1.0
+        else:
+            return (new-old) / old
+
+    def __add__(self, other):
+        return self.__class__(self.x + other.x)
+
+    def __sub__(self, other):
+        return self.__class__(self.x - other.x)
+
+    def __mul__(self, other):
+        return self.__class__(self.x * other.x)
+
+# size results
+class StackResult(co.namedtuple('StackResult', [
+        'file', 'function', 'frame', 'limit', 'children'])):
+    _by = ['file', 'function']
+    _fields = ['frame', 'limit']
+    _sort = ['limit', 'frame']
+    _types = {'frame': Int, 'limit': Int}
+
+    __slots__ = ()
+    def __new__(cls, file='', function='',
+            frame=0, limit=0, children=set()):
+        return super().__new__(cls, file, function,
+            Int(frame), Int(limit),
+            children)
+
+    def __add__(self, other):
+        return StackResult(self.file, self.function,
+            self.frame + other.frame,
+            max(self.limit, other.limit),
+            self.children | other.children)
+
+
+def openio(path, mode='r', buffering=-1):
+    # allow '-' for stdin/stdout
+    if path == '-':
+        if mode == 'r':
+            return os.fdopen(os.dup(sys.stdin.fileno()), mode, buffering)
+        else:
+            return os.fdopen(os.dup(sys.stdout.fileno()), mode, buffering)
+    else:
+        return open(path, mode, buffering)
+
+def collect(ci_paths, *,
+        sources=None,
+        everything=False,
+        **args):
+    # parse the vcg format
+    k_pattern = re.compile('([a-z]+)\s*:', re.DOTALL)
+    v_pattern = re.compile('(?:"(.*?)"|([a-z]+))', re.DOTALL)
+    def parse_vcg(rest):
+        def parse_vcg(rest):
+            node = []
+            while True:
+                rest = rest.lstrip()
+                m_ = k_pattern.match(rest)
+                if not m_:
+                    return (node, rest)
+                k, rest = m_.group(1), rest[m_.end(0):]
+
+                rest = rest.lstrip()
+                if rest.startswith('{'):
+                    v, rest = parse_vcg(rest[1:])
+                    assert rest[0] == '}', "unexpected %r" % rest[0:1]
+                    rest = rest[1:]
+                    node.append((k, v))
+                else:
+                    m_ = v_pattern.match(rest)
+                    assert m_, "unexpected %r" % rest[0:1]
+                    v, rest = m_.group(1) or m_.group(2), rest[m_.end(0):]
+                    node.append((k, v))
+
+        node, rest = parse_vcg(rest)
+        assert rest == '', "unexpected %r" % rest[0:1]
+        return node
+
+    # collect into functions
+    callgraph = co.defaultdict(lambda: (None, None, 0, set()))
+    f_pattern = re.compile(
+        r'([^\\]*)\\n([^:]*)[^\\]*\\n([0-9]+) bytes \((.*)\)')
+    for path in ci_paths:
+        with open(path) as f:
+            vcg = parse_vcg(f.read())
+        for k, graph in vcg:
+            if k != 'graph':
+                continue
+            for k, info in graph:
+                if k == 'node':
+                    info = dict(info)
+                    m_ = f_pattern.match(info['label'])
+                    if m_:
+                        function, file, size, type = m_.groups()
+                        if (not args.get('quiet')
+                                and 'static' not in type
+                                and 'bounded' not in type):
+                            print("warning: "
+                                "found non-static stack for %s (%s, %s)" % (
+                                function, type, size))
+                        _, _, _, targets = callgraph[info['title']]
+                        callgraph[info['title']] = (
+                            file, function, int(size), targets)
+                elif k == 'edge':
+                    info = dict(info)
+                    _, _, _, targets = callgraph[info['sourcename']]
+                    targets.add(info['targetname'])
+                else:
+                    continue
+
+    callgraph_ = co.defaultdict(lambda: (None, None, 0, set()))
+    for source, (s_file, s_function, frame, targets) in callgraph.items():
+        # discard internal functions
+        if not everything and s_function.startswith('__'):
+            continue
+        # ignore filtered sources
+        if sources is not None:
+            if not any(
+                    os.path.abspath(s_file) == os.path.abspath(s)
+                    for s in sources):
+                continue
+        else:
+            # default to only cwd
+            if not everything and not os.path.commonpath([
+                    os.getcwd(),
+                    os.path.abspath(s_file)]) == os.getcwd():
+                continue
+
+        # smiplify path
+        if os.path.commonpath([
+                os.getcwd(),
+                os.path.abspath(s_file)]) == os.getcwd():
+            s_file = os.path.relpath(s_file)
+        else:
+            s_file = os.path.abspath(s_file)
+
+        callgraph_[source] = (s_file, s_function, frame, targets)
+    callgraph = callgraph_
+
+    if not everything:
+        callgraph_ = co.defaultdict(lambda: (None, None, 0, set()))
+        for source, (s_file, s_function, frame, targets) in callgraph.items():
+            # discard filtered sources
+            if sources is not None and not any(
+                    os.path.abspath(s_file) == os.path.abspath(s)
+                    for s in sources):
+                continue
+            # discard internal functions
+            if s_function.startswith('__'):
+                continue
+            callgraph_[source] = (s_file, s_function, frame, targets)
+        callgraph = callgraph_
+
+    # find maximum stack size recursively, this requires also detecting cycles
+    # (in case of recursion)
+    def find_limit(source, seen=None):
+        seen = seen or set()
+        if source not in callgraph:
+            return 0
+        _, _, frame, targets = callgraph[source]
+
+        limit = 0
+        for target in targets:
+            if target in seen:
+                # found a cycle
+                return m.inf
+            limit_ = find_limit(target, seen | {target})
+            limit = max(limit, limit_)
+
+        return frame + limit
+
+    def find_children(targets):
+        children = set()
+        for target in targets:
+            if target in callgraph:
+                t_file, t_function, _, _ = callgraph[target]
+                children.add((t_file, t_function))
+        return children
+
+    # build results
+    results = []
+    for source, (s_file, s_function, frame, targets) in callgraph.items():
+        limit = find_limit(source)
+        children = find_children(targets)
+        results.append(StackResult(s_file, s_function, frame, limit, children))
+
+    return results
+
+
+def fold(Result, results, *,
+        by=None,
+        defines=None,
+        **_):
+    if by is None:
+        by = Result._by
+
+    for k in it.chain(by or [], (k for k, _ in defines or [])):
+        if k not in Result._by and k not in Result._fields:
+            print("error: could not find field %r?" % k)
+            sys.exit(-1)
+
+    # filter by matching defines
+    if defines is not None:
+        results_ = []
+        for r in results:
+            if all(getattr(r, k) in vs for k, vs in defines):
+                results_.append(r)
+        results = results_
+
+    # organize results into conflicts
+    folding = co.OrderedDict()
+    for r in results:
+        name = tuple(getattr(r, k) for k in by)
+        if name not in folding:
+            folding[name] = []
+        folding[name].append(r)
+
+    # merge conflicts
+    folded = []
+    for name, rs in folding.items():
+        folded.append(sum(rs[1:], start=rs[0]))
+
+    return folded
+
+def table(Result, results, diff_results=None, *,
+        by=None,
+        fields=None,
+        sort=None,
+        summary=False,
+        all=False,
+        percent=False,
+        tree=False,
+        depth=1,
+        **_):
+    all_, all = all, __builtins__.all
+
+    if by is None:
+        by = Result._by
+    if fields is None:
+        fields = Result._fields
+    types = Result._types
+
+    # fold again
+    results = fold(Result, results, by=by)
+    if diff_results is not None:
+        diff_results = fold(Result, diff_results, by=by)
+
+    # organize by name
+    table = {
+        ','.join(str(getattr(r, k) or '') for k in by): r
+        for r in results}
+    diff_table = {
+        ','.join(str(getattr(r, k) or '') for k in by): r
+        for r in diff_results or []}
+    names = list(table.keys() | diff_table.keys())
+
+    # sort again, now with diff info, note that python's sort is stable
+    names.sort()
+    if diff_results is not None:
+        names.sort(key=lambda n: tuple(
+            types[k].ratio(
+                getattr(table.get(n), k, None),
+                getattr(diff_table.get(n), k, None))
+            for k in fields),
+            reverse=True)
+    if sort:
+        for k, reverse in reversed(sort):
+            names.sort(
+                key=lambda n: tuple(
+                    (getattr(table[n], k),)
+                    if getattr(table.get(n), k, None) is not None else ()
+                    for k in ([k] if k else [
+                        k for k in Result._sort if k in fields])),
+                reverse=reverse ^ (not k or k in Result._fields))
+
+
+    # build up our lines
+    lines = []
+
+    # header
+    header = []
+    header.append('%s%s' % (
+        ','.join(by),
+        ' (%d added, %d removed)' % (
+            sum(1 for n in table if n not in diff_table),
+            sum(1 for n in diff_table if n not in table))
+            if diff_results is not None and not percent else '')
+        if not summary else '')
+    if diff_results is None:
+        for k in fields:
+            header.append(k)
+    elif percent:
+        for k in fields:
+            header.append(k)
+    else:
+        for k in fields:
+            header.append('o'+k)
+        for k in fields:
+            header.append('n'+k)
+        for k in fields:
+            header.append('d'+k)
+    header.append('')
+    lines.append(header)
+
+    def table_entry(name, r, diff_r=None, ratios=[]):
+        entry = []
+        entry.append(name)
+        if diff_results is None:
+            for k in fields:
+                entry.append(getattr(r, k).table()
+                    if getattr(r, k, None) is not None
+                    else types[k].none)
+        elif percent:
+            for k in fields:
+                entry.append(getattr(r, k).diff_table()
+                    if getattr(r, k, None) is not None
+                    else types[k].diff_none)
+        else:
+            for k in fields:
+                entry.append(getattr(diff_r, k).diff_table()
+                    if getattr(diff_r, k, None) is not None
+                    else types[k].diff_none)
+            for k in fields:
+                entry.append(getattr(r, k).diff_table()
+                    if getattr(r, k, None) is not None
+                    else types[k].diff_none)
+            for k in fields:
+                entry.append(types[k].diff_diff(
+                        getattr(r, k, None),
+                        getattr(diff_r, k, None)))
+        if diff_results is None:
+            entry.append('')
+        elif percent:
+            entry.append(' (%s)' % ', '.join(
+                '+∞%' if t == +m.inf
+                else '-∞%' if t == -m.inf
+                else '%+.1f%%' % (100*t)
+                for t in ratios))
+        else:
+            entry.append(' (%s)' % ', '.join(
+                    '+∞%' if t == +m.inf
+                    else '-∞%' if t == -m.inf
+                    else '%+.1f%%' % (100*t)
+                    for t in ratios
+                    if t)
+                if any(ratios) else '')
+        return entry
+
+    # entries
+    if not summary:
+        for name in names:
+            r = table.get(name)
+            if diff_results is None:
+                diff_r = None
+                ratios = None
+            else:
+                diff_r = diff_table.get(name)
+                ratios = [
+                    types[k].ratio(
+                        getattr(r, k, None),
+                        getattr(diff_r, k, None))
+                    for k in fields]
+                if not all_ and not any(ratios):
+                    continue
+            lines.append(table_entry(name, r, diff_r, ratios))
+
+    # total
+    r = next(iter(fold(Result, results, by=[])), None)
+    if diff_results is None:
+        diff_r = None
+        ratios = None
+    else:
+        diff_r = next(iter(fold(Result, diff_results, by=[])), None)
+        ratios = [
+            types[k].ratio(
+                getattr(r, k, None),
+                getattr(diff_r, k, None))
+            for k in fields]
+    lines.append(table_entry('TOTAL', r, diff_r, ratios))
+
+    # find the best widths, note that column 0 contains the names and column -1
+    # the ratios, so those are handled a bit differently
+    widths = [
+        ((max(it.chain([w], (len(l[i]) for l in lines)))+1+4-1)//4)*4-1
+        for w, i in zip(
+            it.chain([23], it.repeat(7)),
+            range(len(lines[0])-1))]
+
+    # adjust the name width based on the expected call depth, though
+    # note this doesn't really work with unbounded recursion
+    if not summary and not m.isinf(depth):
+        widths[0] += 4*(depth-1)
+
+    # print the tree recursively
+    if not tree:
+        print('%-*s  %s%s' % (
+            widths[0], lines[0][0],
+            ' '.join('%*s' % (w, x)
+                for w, x in zip(widths[1:], lines[0][1:-1])),
+            lines[0][-1]))
+
+    if not summary:
+        line_table = {n: l for n, l in zip(names, lines[1:-1])}
+
+        def recurse(names_, depth_, prefixes=('', '', '', '')):
+            for i, name in enumerate(names_):
+                if name not in line_table:
+                    continue
+                line = line_table[name]
+                is_last = (i == len(names_)-1)
+
+                print('%s%-*s ' % (
+                    prefixes[0+is_last],
+                    widths[0] - (
+                        len(prefixes[0+is_last])
+                        if not m.isinf(depth) else 0),
+                    line[0]),
+                    end='')
+                if not tree:
+                    print(' %s%s' % (
+                        ' '.join('%*s' % (w, x)
+                            for w, x in zip(widths[1:], line[1:-1])),
+                        line[-1]),
+                        end='')
+                print() 
+
+                # recurse?
+                if name in table and depth_ > 1:
+                    children = {
+                        ','.join(str(getattr(Result(*c), k) or '') for k in by)
+                        for c in table[name].children}
+                    recurse(
+                        # note we're maintaining sort order
+                        [n for n in names if n in children],
+                        depth_-1,
+                        (prefixes[2+is_last] + "|-> ",
+                         prefixes[2+is_last] + "'-> ",
+                         prefixes[2+is_last] + "|   ",
+                         prefixes[2+is_last] + "    "))
+
+        recurse(names, depth)
+
+    if not tree:
+        print('%-*s  %s%s' % (
+            widths[0], lines[-1][0],
+            ' '.join('%*s' % (w, x)
+                for w, x in zip(widths[1:], lines[-1][1:-1])),
+            lines[-1][-1]))
+
+
+def main(ci_paths,
+        by=None,
+        fields=None,
+        defines=None,
+        sort=None,
+        **args):
+    # it doesn't really make sense to not have a depth with tree,
+    # so assume depth=inf if tree by default
+    if args.get('depth') is None:
+        args['depth'] = m.inf if args['tree'] else 1
+    elif args.get('depth') == 0:
+        args['depth'] = m.inf
+
+    # find sizes
+    if not args.get('use', None):
+        results = collect(ci_paths, **args)
+    else:
+        results = []
+        with openio(args['use']) as f:
+            reader = csv.DictReader(f, restval='')
+            for r in reader:
+                if not any('stack_'+k in r and r['stack_'+k].strip()
+                        for k in StackResult._fields):
+                    continue
+                try:
+                    results.append(StackResult(
+                        **{k: r[k] for k in StackResult._by
+                            if k in r and r[k].strip()},
+                        **{k: r['stack_'+k] for k in StackResult._fields
+                            if 'stack_'+k in r and r['stack_'+k].strip()}))
+                except TypeError:
+                    pass
+
+    # fold
+    results = fold(StackResult, results, by=by, defines=defines)
+
+    # sort, note that python's sort is stable
+    results.sort()
+    if sort:
+        for k, reverse in reversed(sort):
+            results.sort(
+                key=lambda r: tuple(
+                    (getattr(r, k),) if getattr(r, k) is not None else ()
+                    for k in ([k] if k else StackResult._sort)),
+                reverse=reverse ^ (not k or k in StackResult._fields))
+
+    # write results to CSV
+    if args.get('output'):
+        with openio(args['output'], 'w') as f:
+            writer = csv.DictWriter(f,
+                (by if by is not None else StackResult._by)
+                + ['stack_'+k for k in (
+                    fields if fields is not None else StackResult._fields)])
+            writer.writeheader()
+            for r in results:
+                writer.writerow(
+                    {k: getattr(r, k) for k in (
+                        by if by is not None else StackResult._by)}
+                    | {'stack_'+k: getattr(r, k) for k in (
+                        fields if fields is not None else StackResult._fields)})
+
+    # find previous results?
+    if args.get('diff'):
+        diff_results = []
+        try:
+            with openio(args['diff']) as f:
+                reader = csv.DictReader(f, restval='')
+                for r in reader:
+                    if not any('stack_'+k in r and r['stack_'+k].strip()
+                            for k in StackResult._fields):
+                        continue
+                    try:
+                        diff_results.append(StackResult(
+                            **{k: r[k] for k in StackResult._by
+                                if k in r and r[k].strip()},
+                            **{k: r['stack_'+k] for k in StackResult._fields
+                                if 'stack_'+k in r and r['stack_'+k].strip()}))
+                    except TypeError:
+                        raise
+        except FileNotFoundError:
+            pass
+
+        # fold
+        diff_results = fold(StackResult, diff_results, by=by, defines=defines)
+
+    # print table
+    if not args.get('quiet'):
+        table(StackResult, results,
+            diff_results if args.get('diff') else None,
+            by=by if by is not None else ['function'],
+            fields=fields,
+            sort=sort,
+            **args)
+
+    # error on recursion
+    if args.get('error_on_recursion') and any(
+            m.isinf(float(r.limit)) for r in results):
+        sys.exit(2)
+
+
+if __name__ == "__main__":
+    import argparse
+    import sys
+    parser = argparse.ArgumentParser(
+        description="Find stack usage at the function level.",
+        allow_abbrev=False)
+    parser.add_argument(
+        'ci_paths',
+        nargs='*',
+        help="Input *.ci files.")
+    parser.add_argument(
+        '-v', '--verbose',
+        action='store_true',
+        help="Output commands that run behind the scenes.")
+    parser.add_argument(
+        '-q', '--quiet',
+        action='store_true',
+        help="Don't show anything, useful with -o.")
+    parser.add_argument(
+        '-o', '--output',
+        help="Specify CSV file to store results.")
+    parser.add_argument(
+        '-u', '--use',
+        help="Don't parse anything, use this CSV file.")
+    parser.add_argument(
+        '-d', '--diff',
+        help="Specify CSV file to diff against.")
+    parser.add_argument(
+        '-a', '--all',
+        action='store_true',
+        help="Show all, not just the ones that changed.")
+    parser.add_argument(
+        '-p', '--percent',
+        action='store_true',
+        help="Only show percentage change, not a full diff.")
+    parser.add_argument(
+        '-b', '--by',
+        action='append',
+        choices=StackResult._by,
+        help="Group by this field.")
+    parser.add_argument(
+        '-f', '--field',
+        dest='fields',
+        action='append',
+        choices=StackResult._fields,
+        help="Show this field.")
+    parser.add_argument(
+        '-D', '--define',
+        dest='defines',
+        action='append',
+        type=lambda x: (lambda k,v: (k, set(v.split(','))))(*x.split('=', 1)),
+        help="Only include results where this field is this value.")
+    class AppendSort(argparse.Action):
+        def __call__(self, parser, namespace, value, option):
+            if namespace.sort is None:
+                namespace.sort = []
+            namespace.sort.append((value, True if option == '-S' else False))
+    parser.add_argument(
+        '-s', '--sort',
+        nargs='?',
+        action=AppendSort,
+        help="Sort by this field.")
+    parser.add_argument(
+        '-S', '--reverse-sort',
+        nargs='?',
+        action=AppendSort,
+        help="Sort by this field, but backwards.")
+    parser.add_argument(
+        '-Y', '--summary',
+        action='store_true',
+        help="Only show the total.")
+    parser.add_argument(
+        '-F', '--source',
+        dest='sources',
+        action='append',
+        help="Only consider definitions in this file. Defaults to anything "
+            "in the current directory.")
+    parser.add_argument(
+        '--everything',
+        action='store_true',
+        help="Include builtin and libc specific symbols.")
+    parser.add_argument(
+        '--tree',
+        action='store_true',
+        help="Only show the function call tree.")
+    parser.add_argument(
+        '-Z', '--depth',
+        nargs='?',
+        type=lambda x: int(x, 0),
+        const=0,
+        help="Depth of function calls to show. 0 shows all calls but may not "
+            "terminate!")
+    parser.add_argument(
+        '-e', '--error-on-recursion',
+        action='store_true',
+        help="Error if any functions are recursive.")
+    sys.exit(main(**{k: v
+        for k, v in vars(parser.parse_intermixed_args()).items()
+        if v is not None}))

kernel/fs/littlefs/scripts/structs.py

@@ -0,0 +1,652 @@
+#!/usr/bin/env python3
+#
+# Script to find struct sizes.
+#
+# Example:
+# ./scripts/structs.py lfs.o lfs_util.o -Ssize
+#
+# Copyright (c) 2022, The littlefs authors.
+# SPDX-License-Identifier: BSD-3-Clause
+#
+
+import collections as co
+import csv
+import difflib
+import itertools as it
+import math as m
+import os
+import re
+import shlex
+import subprocess as sp
+
+
+OBJDUMP_PATH = ['objdump']
+
+
+
+# integer fields
+class Int(co.namedtuple('Int', 'x')):
+    __slots__ = ()
+    def __new__(cls, x=0):
+        if isinstance(x, Int):
+            return x
+        if isinstance(x, str):
+            try:
+                x = int(x, 0)
+            except ValueError:
+                # also accept +-∞ and +-inf
+                if re.match('^\s*\+?\s*(?:∞|inf)\s*$', x):
+                    x = m.inf
+                elif re.match('^\s*-\s*(?:∞|inf)\s*$', x):
+                    x = -m.inf
+                else:
+                    raise
+        assert isinstance(x, int) or m.isinf(x), x
+        return super().__new__(cls, x)
+
+    def __str__(self):
+        if self.x == m.inf:
+            return '∞'
+        elif self.x == -m.inf:
+            return '-∞'
+        else:
+            return str(self.x)
+
+    def __int__(self):
+        assert not m.isinf(self.x)
+        return self.x
+
+    def __float__(self):
+        return float(self.x)
+
+    none = '%7s' % '-'
+    def table(self):
+        return '%7s' % (self,)
+
+    diff_none = '%7s' % '-'
+    diff_table = table
+
+    def diff_diff(self, other):
+        new = self.x if self else 0
+        old = other.x if other else 0
+        diff = new - old
+        if diff == +m.inf:
+            return '%7s' % '+∞'
+        elif diff == -m.inf:
+            return '%7s' % '-∞'
+        else:
+            return '%+7d' % diff
+
+    def ratio(self, other):
+        new = self.x if self else 0
+        old = other.x if other else 0
+        if m.isinf(new) and m.isinf(old):
+            return 0.0
+        elif m.isinf(new):
+            return +m.inf
+        elif m.isinf(old):
+            return -m.inf
+        elif not old and not new:
+            return 0.0
+        elif not old:
+            return 1.0
+        else:
+            return (new-old) / old
+
+    def __add__(self, other):
+        return self.__class__(self.x + other.x)
+
+    def __sub__(self, other):
+        return self.__class__(self.x - other.x)
+
+    def __mul__(self, other):
+        return self.__class__(self.x * other.x)
+
+# struct size results
+class StructResult(co.namedtuple('StructResult', ['file', 'struct', 'size'])):
+    _by = ['file', 'struct']
+    _fields = ['size']
+    _sort = ['size']
+    _types = {'size': Int}
+
+    __slots__ = ()
+    def __new__(cls, file='', struct='', size=0):
+        return super().__new__(cls, file, struct,
+            Int(size))
+
+    def __add__(self, other):
+        return StructResult(self.file, self.struct,
+            self.size + other.size)
+
+
+def openio(path, mode='r', buffering=-1):
+    # allow '-' for stdin/stdout
+    if path == '-':
+        if mode == 'r':
+            return os.fdopen(os.dup(sys.stdin.fileno()), mode, buffering)
+        else:
+            return os.fdopen(os.dup(sys.stdout.fileno()), mode, buffering)
+    else:
+        return open(path, mode, buffering)
+
+def collect(obj_paths, *,
+        objdump_path=OBJDUMP_PATH,
+        sources=None,
+        everything=False,
+        internal=False,
+        **args):
+    line_pattern = re.compile(
+        '^\s+(?P<no>[0-9]+)'
+            '(?:\s+(?P<dir>[0-9]+))?'
+            '\s+.*'
+            '\s+(?P<path>[^\s]+)$')
+    info_pattern = re.compile(
+        '^(?:.*(?P<tag>DW_TAG_[a-z_]+).*'
+            '|.*DW_AT_name.*:\s*(?P<name>[^:\s]+)\s*'
+            '|.*DW_AT_decl_file.*:\s*(?P<file>[0-9]+)\s*'
+            '|.*DW_AT_byte_size.*:\s*(?P<size>[0-9]+)\s*)$')
+
+    results = []
+    for path in obj_paths:
+        # find files, we want to filter by structs in .h files
+        dirs = {}
+        files = {}
+        # note objdump-path may contain extra args
+        cmd = objdump_path + ['--dwarf=rawline', path]
+        if args.get('verbose'):
+            print(' '.join(shlex.quote(c) for c in cmd))
+        proc = sp.Popen(cmd,
+            stdout=sp.PIPE,
+            stderr=sp.PIPE if not args.get('verbose') else None,
+            universal_newlines=True,
+            errors='replace',
+            close_fds=False)
+        for line in proc.stdout:
+            # note that files contain references to dirs, which we
+            # dereference as soon as we see them as each file table follows a
+            # dir table
+            m = line_pattern.match(line)
+            if m:
+                if not m.group('dir'):
+                    # found a directory entry
+                    dirs[int(m.group('no'))] = m.group('path')
+                else:
+                    # found a file entry
+                    dir = int(m.group('dir'))
+                    if dir in dirs:
+                        files[int(m.group('no'))] = os.path.join(
+                            dirs[dir],
+                            m.group('path'))
+                    else:
+                        files[int(m.group('no'))] = m.group('path')
+        proc.wait()
+        if proc.returncode != 0:
+            if not args.get('verbose'):
+                for line in proc.stderr:
+                    sys.stdout.write(line)
+            sys.exit(-1)
+
+        # collect structs as we parse dwarf info
+        results_ = []
+        is_struct = False
+        s_name = None
+        s_file = None
+        s_size = None
+        # note objdump-path may contain extra args
+        cmd = objdump_path + ['--dwarf=info', path]
+        if args.get('verbose'):
+            print(' '.join(shlex.quote(c) for c in cmd))
+        proc = sp.Popen(cmd,
+            stdout=sp.PIPE,
+            stderr=sp.PIPE if not args.get('verbose') else None,
+            universal_newlines=True,
+            errors='replace',
+            close_fds=False)
+        for line in proc.stdout:
+            # state machine here to find structs
+            m = info_pattern.match(line)
+            if m:
+                if m.group('tag'):
+                    if is_struct:
+                        file = files.get(s_file, '?')
+                        results_.append(StructResult(file, s_name, s_size))
+                    is_struct = (m.group('tag') == 'DW_TAG_structure_type')
+                elif m.group('name'):
+                    s_name = m.group('name')
+                elif m.group('file'):
+                    s_file = int(m.group('file'))
+                elif m.group('size'):
+                    s_size = int(m.group('size'))
+        if is_struct:
+            file = files.get(s_file, '?')
+            results_.append(StructResult(file, s_name, s_size))
+        proc.wait()
+        if proc.returncode != 0:
+            if not args.get('verbose'):
+                for line in proc.stderr:
+                    sys.stdout.write(line)
+            sys.exit(-1)
+
+        for r in results_:
+            # ignore filtered sources
+            if sources is not None:
+                if not any(
+                        os.path.abspath(r.file) == os.path.abspath(s)
+                        for s in sources):
+                    continue
+            else:
+                # default to only cwd
+                if not everything and not os.path.commonpath([
+                        os.getcwd(),
+                        os.path.abspath(r.file)]) == os.getcwd():
+                    continue
+
+                # limit to .h files unless --internal
+                if not internal and not r.file.endswith('.h'):
+                    continue
+
+            # simplify path
+            if os.path.commonpath([
+                    os.getcwd(),
+                    os.path.abspath(r.file)]) == os.getcwd():
+                file = os.path.relpath(r.file)
+            else:
+                file = os.path.abspath(r.file)
+
+            results.append(r._replace(file=file))
+
+    return results
+
+
+def fold(Result, results, *,
+        by=None,
+        defines=None,
+        **_):
+    if by is None:
+        by = Result._by
+
+    for k in it.chain(by or [], (k for k, _ in defines or [])):
+        if k not in Result._by and k not in Result._fields:
+            print("error: could not find field %r?" % k)
+            sys.exit(-1)
+
+    # filter by matching defines
+    if defines is not None:
+        results_ = []
+        for r in results:
+            if all(getattr(r, k) in vs for k, vs in defines):
+                results_.append(r)
+        results = results_
+
+    # organize results into conflicts
+    folding = co.OrderedDict()
+    for r in results:
+        name = tuple(getattr(r, k) for k in by)
+        if name not in folding:
+            folding[name] = []
+        folding[name].append(r)
+
+    # merge conflicts
+    folded = []
+    for name, rs in folding.items():
+        folded.append(sum(rs[1:], start=rs[0]))
+
+    return folded
+
+def table(Result, results, diff_results=None, *,
+        by=None,
+        fields=None,
+        sort=None,
+        summary=False,
+        all=False,
+        percent=False,
+        **_):
+    all_, all = all, __builtins__.all
+
+    if by is None:
+        by = Result._by
+    if fields is None:
+        fields = Result._fields
+    types = Result._types
+
+    # fold again
+    results = fold(Result, results, by=by)
+    if diff_results is not None:
+        diff_results = fold(Result, diff_results, by=by)
+
+    # organize by name
+    table = {
+        ','.join(str(getattr(r, k) or '') for k in by): r
+        for r in results}
+    diff_table = {
+        ','.join(str(getattr(r, k) or '') for k in by): r
+        for r in diff_results or []}
+    names = list(table.keys() | diff_table.keys())
+
+    # sort again, now with diff info, note that python's sort is stable
+    names.sort()
+    if diff_results is not None:
+        names.sort(key=lambda n: tuple(
+            types[k].ratio(
+                getattr(table.get(n), k, None),
+                getattr(diff_table.get(n), k, None))
+            for k in fields),
+            reverse=True)
+    if sort:
+        for k, reverse in reversed(sort):
+            names.sort(
+                key=lambda n: tuple(
+                    (getattr(table[n], k),)
+                    if getattr(table.get(n), k, None) is not None else ()
+                    for k in ([k] if k else [
+                        k for k in Result._sort if k in fields])),
+                reverse=reverse ^ (not k or k in Result._fields))
+
+
+    # build up our lines
+    lines = []
+
+    # header
+    header = []
+    header.append('%s%s' % (
+        ','.join(by),
+        ' (%d added, %d removed)' % (
+            sum(1 for n in table if n not in diff_table),
+            sum(1 for n in diff_table if n not in table))
+            if diff_results is not None and not percent else '')
+        if not summary else '')
+    if diff_results is None:
+        for k in fields:
+            header.append(k)
+    elif percent:
+        for k in fields:
+            header.append(k)
+    else:
+        for k in fields:
+            header.append('o'+k)
+        for k in fields:
+            header.append('n'+k)
+        for k in fields:
+            header.append('d'+k)
+    header.append('')
+    lines.append(header)
+
+    def table_entry(name, r, diff_r=None, ratios=[]):
+        entry = []
+        entry.append(name)
+        if diff_results is None:
+            for k in fields:
+                entry.append(getattr(r, k).table()
+                    if getattr(r, k, None) is not None
+                    else types[k].none)
+        elif percent:
+            for k in fields:
+                entry.append(getattr(r, k).diff_table()
+                    if getattr(r, k, None) is not None
+                    else types[k].diff_none)
+        else:
+            for k in fields:
+                entry.append(getattr(diff_r, k).diff_table()
+                    if getattr(diff_r, k, None) is not None
+                    else types[k].diff_none)
+            for k in fields:
+                entry.append(getattr(r, k).diff_table()
+                    if getattr(r, k, None) is not None
+                    else types[k].diff_none)
+            for k in fields:
+                entry.append(types[k].diff_diff(
+                        getattr(r, k, None),
+                        getattr(diff_r, k, None)))
+        if diff_results is None:
+            entry.append('')
+        elif percent:
+            entry.append(' (%s)' % ', '.join(
+                '+∞%' if t == +m.inf
+                else '-∞%' if t == -m.inf
+                else '%+.1f%%' % (100*t)
+                for t in ratios))
+        else:
+            entry.append(' (%s)' % ', '.join(
+                    '+∞%' if t == +m.inf
+                    else '-∞%' if t == -m.inf
+                    else '%+.1f%%' % (100*t)
+                    for t in ratios
+                    if t)
+                if any(ratios) else '')
+        return entry
+
+    # entries
+    if not summary:
+        for name in names:
+            r = table.get(name)
+            if diff_results is None:
+                diff_r = None
+                ratios = None
+            else:
+                diff_r = diff_table.get(name)
+                ratios = [
+                    types[k].ratio(
+                        getattr(r, k, None),
+                        getattr(diff_r, k, None))
+                    for k in fields]
+                if not all_ and not any(ratios):
+                    continue
+            lines.append(table_entry(name, r, diff_r, ratios))
+
+    # total
+    r = next(iter(fold(Result, results, by=[])), None)
+    if diff_results is None:
+        diff_r = None
+        ratios = None
+    else:
+        diff_r = next(iter(fold(Result, diff_results, by=[])), None)
+        ratios = [
+            types[k].ratio(
+                getattr(r, k, None),
+                getattr(diff_r, k, None))
+            for k in fields]
+    lines.append(table_entry('TOTAL', r, diff_r, ratios))
+
+    # find the best widths, note that column 0 contains the names and column -1
+    # the ratios, so those are handled a bit differently
+    widths = [
+        ((max(it.chain([w], (len(l[i]) for l in lines)))+1+4-1)//4)*4-1
+        for w, i in zip(
+            it.chain([23], it.repeat(7)),
+            range(len(lines[0])-1))]
+
+    # print our table
+    for line in lines:
+        print('%-*s  %s%s' % (
+            widths[0], line[0],
+            ' '.join('%*s' % (w, x)
+                for w, x in zip(widths[1:], line[1:-1])),
+            line[-1]))
+
+
+def main(obj_paths, *,
+        by=None,
+        fields=None,
+        defines=None,
+        sort=None,
+        **args):
+    # find sizes
+    if not args.get('use', None):
+        results = collect(obj_paths, **args)
+    else:
+        results = []
+        with openio(args['use']) as f:
+            reader = csv.DictReader(f, restval='')
+            for r in reader:
+                if not any('struct_'+k in r and r['struct_'+k].strip()
+                        for k in StructResult._fields):
+                    continue
+                try:
+                    results.append(StructResult(
+                        **{k: r[k] for k in StructResult._by
+                            if k in r and r[k].strip()},
+                        **{k: r['struct_'+k]
+                            for k in StructResult._fields
+                            if 'struct_'+k in r
+                                and r['struct_'+k].strip()}))
+                except TypeError:
+                    pass
+
+    # fold
+    results = fold(StructResult, results, by=by, defines=defines)
+
+    # sort, note that python's sort is stable
+    results.sort()
+    if sort:
+        for k, reverse in reversed(sort):
+            results.sort(
+                key=lambda r: tuple(
+                    (getattr(r, k),) if getattr(r, k) is not None else ()
+                    for k in ([k] if k else StructResult._sort)),
+                reverse=reverse ^ (not k or k in StructResult._fields))
+
+    # write results to CSV
+    if args.get('output'):
+        with openio(args['output'], 'w') as f:
+            writer = csv.DictWriter(f,
+                (by if by is not None else StructResult._by)
+                + ['struct_'+k for k in (
+                    fields if fields is not None else StructResult._fields)])
+            writer.writeheader()
+            for r in results:
+                writer.writerow(
+                    {k: getattr(r, k) for k in (
+                        by if by is not None else StructResult._by)}
+                    | {'struct_'+k: getattr(r, k) for k in (
+                        fields if fields is not None else StructResult._fields)})
+
+    # find previous results?
+    if args.get('diff'):
+        diff_results = []
+        try:
+            with openio(args['diff']) as f:
+                reader = csv.DictReader(f, restval='')
+                for r in reader:
+                    if not any('struct_'+k in r and r['struct_'+k].strip()
+                            for k in StructResult._fields):
+                        continue
+                    try:
+                        diff_results.append(StructResult(
+                            **{k: r[k] for k in StructResult._by
+                                if k in r and r[k].strip()},
+                            **{k: r['struct_'+k]
+                                for k in StructResult._fields
+                                if 'struct_'+k in r
+                                    and r['struct_'+k].strip()}))
+                    except TypeError:
+                        pass
+        except FileNotFoundError:
+            pass
+
+        # fold
+        diff_results = fold(StructResult, diff_results, by=by, defines=defines)
+
+    # print table
+    if not args.get('quiet'):
+        table(StructResult, results,
+            diff_results if args.get('diff') else None,
+            by=by if by is not None else ['struct'],
+            fields=fields,
+            sort=sort,
+            **args)
+
+
+if __name__ == "__main__":
+    import argparse
+    import sys
+    parser = argparse.ArgumentParser(
+        description="Find struct sizes.",
+        allow_abbrev=False)
+    parser.add_argument(
+        'obj_paths',
+        nargs='*',
+        help="Input *.o files.")
+    parser.add_argument(
+        '-v', '--verbose',
+        action='store_true',
+        help="Output commands that run behind the scenes.")
+    parser.add_argument(
+        '-q', '--quiet',
+        action='store_true',
+        help="Don't show anything, useful with -o.")
+    parser.add_argument(
+        '-o', '--output',
+        help="Specify CSV file to store results.")
+    parser.add_argument(
+        '-u', '--use',
+        help="Don't parse anything, use this CSV file.")
+    parser.add_argument(
+        '-d', '--diff',
+        help="Specify CSV file to diff against.")
+    parser.add_argument(
+        '-a', '--all',
+        action='store_true',
+        help="Show all, not just the ones that changed.")
+    parser.add_argument(
+        '-p', '--percent',
+        action='store_true',
+        help="Only show percentage change, not a full diff.")
+    parser.add_argument(
+        '-b', '--by',
+        action='append',
+        choices=StructResult._by,
+        help="Group by this field.")
+    parser.add_argument(
+        '-f', '--field',
+        dest='fields',
+        action='append',
+        choices=StructResult._fields,
+        help="Show this field.")
+    parser.add_argument(
+        '-D', '--define',
+        dest='defines',
+        action='append',
+        type=lambda x: (lambda k,v: (k, set(v.split(','))))(*x.split('=', 1)),
+        help="Only include results where this field is this value.")
+    class AppendSort(argparse.Action):
+        def __call__(self, parser, namespace, value, option):
+            if namespace.sort is None:
+                namespace.sort = []
+            namespace.sort.append((value, True if option == '-S' else False))
+    parser.add_argument(
+        '-s', '--sort',
+        nargs='?',
+        action=AppendSort,
+        help="Sort by this field.")
+    parser.add_argument(
+        '-S', '--reverse-sort',
+        nargs='?',
+        action=AppendSort,
+        help="Sort by this field, but backwards.")
+    parser.add_argument(
+        '-Y', '--summary',
+        action='store_true',
+        help="Only show the total.")
+    parser.add_argument(
+        '-F', '--source',
+        dest='sources',
+        action='append',
+        help="Only consider definitions in this file. Defaults to anything "
+            "in the current directory.")
+    parser.add_argument(
+        '--everything',
+        action='store_true',
+        help="Include builtin and libc specific symbols.")
+    parser.add_argument(
+        '--internal',
+        action='store_true',
+        help="Also show structs in .c files.")
+    parser.add_argument(
+        '--objdump-path',
+        type=lambda x: x.split(),
+        default=OBJDUMP_PATH,
+        help="Path to the objdump executable, may include flags. "
+            "Defaults to %r." % OBJDUMP_PATH)
+    sys.exit(main(**{k: v
+        for k, v in vars(parser.parse_intermixed_args()).items()
+        if v is not None}))

kernel/fs/littlefs/scripts/summary.py

@@ -0,0 +1,829 @@
+#!/usr/bin/env python3
+#
+# Script to summarize the outputs of other scripts. Operates on CSV files.
+#
+# Example:
+# ./scripts/code.py lfs.o lfs_util.o -q -o lfs.code.csv
+# ./scripts/data.py lfs.o lfs_util.o -q -o lfs.data.csv
+# ./scripts/summary.py lfs.code.csv lfs.data.csv -q -o lfs.csv
+# ./scripts/summary.py -Y lfs.csv -f code=code_size,data=data_size
+#
+# Copyright (c) 2022, The littlefs authors.
+# SPDX-License-Identifier: BSD-3-Clause
+#
+
+import collections as co
+import csv
+import functools as ft
+import itertools as it
+import math as m
+import os
+import re
+
+
+# supported merge operations
+#
+# this is a terrible way to express these
+#
+OPS = {
+    'sum':     lambda xs: sum(xs[1:], start=xs[0]),
+    'prod':    lambda xs: m.prod(xs[1:], start=xs[0]),
+    'min':     min,
+    'max':     max,
+    'mean':    lambda xs: Float(sum(float(x) for x in xs) / len(xs)),
+    'stddev':  lambda xs: (
+        lambda mean: Float(
+            m.sqrt(sum((float(x) - mean)**2 for x in xs) / len(xs)))
+        )(sum(float(x) for x in xs) / len(xs)),
+    'gmean':   lambda xs: Float(m.prod(float(x) for x in xs)**(1/len(xs))),
+    'gstddev': lambda xs: (
+        lambda gmean: Float(
+            m.exp(m.sqrt(sum(m.log(float(x)/gmean)**2 for x in xs) / len(xs)))
+            if gmean else m.inf)
+        )(m.prod(float(x) for x in xs)**(1/len(xs))),
+}
+
+
+# integer fields
+class Int(co.namedtuple('Int', 'x')):
+    __slots__ = ()
+    def __new__(cls, x=0):
+        if isinstance(x, Int):
+            return x
+        if isinstance(x, str):
+            try:
+                x = int(x, 0)
+            except ValueError:
+                # also accept +-∞ and +-inf
+                if re.match('^\s*\+?\s*(?:∞|inf)\s*$', x):
+                    x = m.inf
+                elif re.match('^\s*-\s*(?:∞|inf)\s*$', x):
+                    x = -m.inf
+                else:
+                    raise
+        assert isinstance(x, int) or m.isinf(x), x
+        return super().__new__(cls, x)
+
+    def __str__(self):
+        if self.x == m.inf:
+            return '∞'
+        elif self.x == -m.inf:
+            return '-∞'
+        else:
+            return str(self.x)
+
+    def __int__(self):
+        assert not m.isinf(self.x)
+        return self.x
+
+    def __float__(self):
+        return float(self.x)
+
+    none = '%7s' % '-'
+    def table(self):
+        return '%7s' % (self,)
+
+    diff_none = '%7s' % '-'
+    diff_table = table
+
+    def diff_diff(self, other):
+        new = self.x if self else 0
+        old = other.x if other else 0
+        diff = new - old
+        if diff == +m.inf:
+            return '%7s' % '+∞'
+        elif diff == -m.inf:
+            return '%7s' % '-∞'
+        else:
+            return '%+7d' % diff
+
+    def ratio(self, other):
+        new = self.x if self else 0
+        old = other.x if other else 0
+        if m.isinf(new) and m.isinf(old):
+            return 0.0
+        elif m.isinf(new):
+            return +m.inf
+        elif m.isinf(old):
+            return -m.inf
+        elif not old and not new:
+            return 0.0
+        elif not old:
+            return 1.0
+        else:
+            return (new-old) / old
+
+    def __add__(self, other):
+        return self.__class__(self.x + other.x)
+
+    def __sub__(self, other):
+        return self.__class__(self.x - other.x)
+
+    def __mul__(self, other):
+        return self.__class__(self.x * other.x)
+
+# float fields
+class Float(co.namedtuple('Float', 'x')):
+    __slots__ = ()
+    def __new__(cls, x=0.0):
+        if isinstance(x, Float):
+            return x
+        if isinstance(x, str):
+            try:
+                x = float(x)
+            except ValueError:
+                # also accept +-∞ and +-inf
+                if re.match('^\s*\+?\s*(?:∞|inf)\s*$', x):
+                    x = m.inf
+                elif re.match('^\s*-\s*(?:∞|inf)\s*$', x):
+                    x = -m.inf
+                else:
+                    raise
+        assert isinstance(x, float), x
+        return super().__new__(cls, x)
+
+    def __str__(self):
+        if self.x == m.inf:
+            return '∞'
+        elif self.x == -m.inf:
+            return '-∞'
+        else:
+            return '%.1f' % self.x
+
+    def __float__(self):
+        return float(self.x)
+
+    none = Int.none
+    table = Int.table
+    diff_none = Int.diff_none
+    diff_table = Int.diff_table
+    diff_diff = Int.diff_diff
+    ratio = Int.ratio
+    __add__ = Int.__add__
+    __sub__ = Int.__sub__
+    __mul__ = Int.__mul__
+
+# fractional fields, a/b
+class Frac(co.namedtuple('Frac', 'a,b')):
+    __slots__ = ()
+    def __new__(cls, a=0, b=None):
+        if isinstance(a, Frac) and b is None:
+            return a
+        if isinstance(a, str) and b is None:
+            a, b = a.split('/', 1)
+        if b is None:
+            b = a
+        return super().__new__(cls, Int(a), Int(b))
+
+    def __str__(self):
+        return '%s/%s' % (self.a, self.b)
+
+    def __float__(self):
+        return float(self.a)
+
+    none = '%11s %7s' % ('-', '-')
+    def table(self):
+        t = self.a.x/self.b.x if self.b.x else 1.0
+        return '%11s %7s' % (
+            self,
+            '∞%' if t == +m.inf
+            else '-∞%' if t == -m.inf
+            else '%.1f%%' % (100*t))
+
+    diff_none = '%11s' % '-'
+    def diff_table(self):
+        return '%11s' % (self,)
+
+    def diff_diff(self, other):
+        new_a, new_b = self if self else (Int(0), Int(0))
+        old_a, old_b = other if other else (Int(0), Int(0))
+        return '%11s' % ('%s/%s' % (
+            new_a.diff_diff(old_a).strip(),
+            new_b.diff_diff(old_b).strip()))
+
+    def ratio(self, other):
+        new_a, new_b = self if self else (Int(0), Int(0))
+        old_a, old_b = other if other else (Int(0), Int(0))
+        new = new_a.x/new_b.x if new_b.x else 1.0
+        old = old_a.x/old_b.x if old_b.x else 1.0
+        return new - old
+
+    def __add__(self, other):
+        return self.__class__(self.a + other.a, self.b + other.b)
+
+    def __sub__(self, other):
+        return self.__class__(self.a - other.a, self.b - other.b)
+
+    def __mul__(self, other):
+        return self.__class__(self.a * other.a, self.b + other.b)
+
+    def __lt__(self, other):
+        self_t = self.a.x/self.b.x if self.b.x else 1.0
+        other_t = other.a.x/other.b.x if other.b.x else 1.0
+        return (self_t, self.a.x) < (other_t, other.a.x)
+
+    def __gt__(self, other):
+        return self.__class__.__lt__(other, self)
+
+    def __le__(self, other):
+        return not self.__gt__(other)
+
+    def __ge__(self, other):
+        return not self.__lt__(other)
+
+# available types
+TYPES = co.OrderedDict([
+    ('int',   Int),
+    ('float', Float),
+    ('frac',  Frac)
+])
+
+
+def infer(results, *,
+        by=None,
+        fields=None,
+        types={},
+        ops={},
+        renames=[],
+        **_):
+    # if fields not specified, try to guess from data
+    if fields is None:
+        fields = co.OrderedDict()
+        for r in results:
+            for k, v in r.items():
+                if (by is None or k not in by) and v.strip():
+                    types_ = []
+                    for t in fields.get(k, TYPES.values()):
+                        try:
+                            t(v)
+                            types_.append(t)
+                        except ValueError:
+                            pass
+                    fields[k] = types_
+        fields = list(k for k, v in fields.items() if v)
+
+    # deduplicate fields
+    fields = list(co.OrderedDict.fromkeys(fields).keys())
+
+    # if by not specified, guess it's anything not in fields and not a
+    # source of a rename
+    if by is None:
+        by = co.OrderedDict()
+        for r in results:
+            # also ignore None keys, these are introduced by csv.DictReader
+            # when header + row mismatch
+            by.update((k, True) for k in r.keys()
+                if k is not None
+                    and k not in fields
+                    and not any(k == old_k for _, old_k in renames))
+        by = list(by.keys())
+
+    # deduplicate fields
+    by = list(co.OrderedDict.fromkeys(by).keys())
+
+    # find best type for all fields
+    types_ = {}
+    for k in fields:
+        if k in types:
+            types_[k] = types[k]
+        else:
+            for t in TYPES.values():
+                for r in results:
+                    if k in r and r[k].strip():
+                        try:
+                            t(r[k])
+                        except ValueError:
+                            break
+                else:
+                    types_[k] = t
+                    break
+            else:
+                print("error: no type matches field %r?" % k)
+                sys.exit(-1)
+    types = types_
+
+    # does folding change the type?
+    types_ = {}
+    for k, t in types.items():
+        types_[k] = ops.get(k, OPS['sum'])([t()]).__class__
+
+
+    # create result class
+    def __new__(cls, **r):
+        return cls.__mro__[1].__new__(cls,
+            **{k: r.get(k, '') for k in by},
+            **{k: r[k] if k in r and isinstance(r[k], list)
+                else [types[k](r[k])] if k in r
+                else []
+                for k in fields})
+
+    def __add__(self, other):
+        return self.__class__(
+            **{k: getattr(self, k) for k in by},
+            **{k: object.__getattribute__(self, k)
+                + object.__getattribute__(other, k)
+                for k in fields})
+
+    def __getattribute__(self, k):
+        if k in fields:
+            if object.__getattribute__(self, k):
+                return ops.get(k, OPS['sum'])(object.__getattribute__(self, k))
+            else:
+                return None
+        return object.__getattribute__(self, k)
+
+    return type('Result', (co.namedtuple('Result', by + fields),), {
+        '__slots__': (),
+        '__new__': __new__,
+        '__add__': __add__,
+        '__getattribute__': __getattribute__,
+        '_by': by,
+        '_fields': fields,
+        '_sort': fields,
+        '_types': types_,
+    })
+
+
+def fold(Result, results, *,
+        by=None,
+        defines=None,
+        **_):
+    if by is None:
+        by = Result._by
+
+    for k in it.chain(by or [], (k for k, _ in defines or [])):
+        if k not in Result._by and k not in Result._fields:
+            print("error: could not find field %r?" % k)
+            sys.exit(-1)
+
+    # filter by matching defines
+    if defines is not None:
+        results_ = []
+        for r in results:
+            if all(getattr(r, k) in vs for k, vs in defines):
+                results_.append(r)
+        results = results_
+
+    # organize results into conflicts
+    folding = co.OrderedDict()
+    for r in results:
+        name = tuple(getattr(r, k) for k in by)
+        if name not in folding:
+            folding[name] = []
+        folding[name].append(r)
+
+    # merge conflicts
+    folded = []
+    for name, rs in folding.items():
+        folded.append(sum(rs[1:], start=rs[0]))
+
+    return folded
+
+def table(Result, results, diff_results=None, *,
+        by=None,
+        fields=None,
+        sort=None,
+        summary=False,
+        all=False,
+        percent=False,
+        **_):
+    all_, all = all, __builtins__.all
+
+    if by is None:
+        by = Result._by
+    if fields is None:
+        fields = Result._fields
+    types = Result._types
+
+    # fold again
+    results = fold(Result, results, by=by)
+    if diff_results is not None:
+        diff_results = fold(Result, diff_results, by=by)
+
+    # organize by name
+    table = {
+        ','.join(str(getattr(r, k) or '') for k in by): r
+        for r in results}
+    diff_table = {
+        ','.join(str(getattr(r, k) or '') for k in by): r
+        for r in diff_results or []}
+    names = list(table.keys() | diff_table.keys())
+
+    # sort again, now with diff info, note that python's sort is stable
+    names.sort()
+    if diff_results is not None:
+        names.sort(key=lambda n: tuple(
+            types[k].ratio(
+                getattr(table.get(n), k, None),
+                getattr(diff_table.get(n), k, None))
+            for k in fields),
+            reverse=True)
+    if sort:
+        for k, reverse in reversed(sort):
+            names.sort(
+                key=lambda n: tuple(
+                    (getattr(table[n], k),)
+                    if getattr(table.get(n), k, None) is not None else ()
+                    for k in ([k] if k else [
+                        k for k in Result._sort if k in fields])),
+                reverse=reverse ^ (not k or k in Result._fields))
+
+
+    # build up our lines
+    lines = []
+
+    # header
+    header = []
+    header.append('%s%s' % (
+        ','.join(by),
+        ' (%d added, %d removed)' % (
+            sum(1 for n in table if n not in diff_table),
+            sum(1 for n in diff_table if n not in table))
+            if diff_results is not None and not percent else '')
+        if not summary else '')
+    if diff_results is None:
+        for k in fields:
+            header.append(k)
+    elif percent:
+        for k in fields:
+            header.append(k)
+    else:
+        for k in fields:
+            header.append('o'+k)
+        for k in fields:
+            header.append('n'+k)
+        for k in fields:
+            header.append('d'+k)
+    header.append('')
+    lines.append(header)
+
+    def table_entry(name, r, diff_r=None, ratios=[]):
+        entry = []
+        entry.append(name)
+        if diff_results is None:
+            for k in fields:
+                entry.append(getattr(r, k).table()
+                    if getattr(r, k, None) is not None
+                    else types[k].none)
+        elif percent:
+            for k in fields:
+                entry.append(getattr(r, k).diff_table()
+                    if getattr(r, k, None) is not None
+                    else types[k].diff_none)
+        else:
+            for k in fields:
+                entry.append(getattr(diff_r, k).diff_table()
+                    if getattr(diff_r, k, None) is not None
+                    else types[k].diff_none)
+            for k in fields:
+                entry.append(getattr(r, k).diff_table()
+                    if getattr(r, k, None) is not None
+                    else types[k].diff_none)
+            for k in fields:
+                entry.append(types[k].diff_diff(
+                        getattr(r, k, None),
+                        getattr(diff_r, k, None)))
+        if diff_results is None:
+            entry.append('')
+        elif percent:
+            entry.append(' (%s)' % ', '.join(
+                '+∞%' if t == +m.inf
+                else '-∞%' if t == -m.inf
+                else '%+.1f%%' % (100*t)
+                for t in ratios))
+        else:
+            entry.append(' (%s)' % ', '.join(
+                    '+∞%' if t == +m.inf
+                    else '-∞%' if t == -m.inf
+                    else '%+.1f%%' % (100*t)
+                    for t in ratios
+                    if t)
+                if any(ratios) else '')
+        return entry
+
+    # entries
+    if not summary:
+        for name in names:
+            r = table.get(name)
+            if diff_results is None:
+                diff_r = None
+                ratios = None
+            else:
+                diff_r = diff_table.get(name)
+                ratios = [
+                    types[k].ratio(
+                        getattr(r, k, None),
+                        getattr(diff_r, k, None))
+                    for k in fields]
+                if not all_ and not any(ratios):
+                    continue
+            lines.append(table_entry(name, r, diff_r, ratios))
+
+    # total
+    r = next(iter(fold(Result, results, by=[])), None)
+    if diff_results is None:
+        diff_r = None
+        ratios = None
+    else:
+        diff_r = next(iter(fold(Result, diff_results, by=[])), None)
+        ratios = [
+            types[k].ratio(
+                getattr(r, k, None),
+                getattr(diff_r, k, None))
+            for k in fields]
+    lines.append(table_entry('TOTAL', r, diff_r, ratios))
+
+    # find the best widths, note that column 0 contains the names and column -1
+    # the ratios, so those are handled a bit differently
+    widths = [
+        ((max(it.chain([w], (len(l[i]) for l in lines)))+1+4-1)//4)*4-1
+        for w, i in zip(
+            it.chain([23], it.repeat(7)),
+            range(len(lines[0])-1))]
+
+    # print our table
+    for line in lines:
+        print('%-*s  %s%s' % (
+            widths[0], line[0],
+            ' '.join('%*s' % (w, x)
+                for w, x in zip(widths[1:], line[1:-1])),
+            line[-1]))
+
+
+def openio(path, mode='r', buffering=-1):
+    # allow '-' for stdin/stdout
+    if path == '-':
+        if mode == 'r':
+            return os.fdopen(os.dup(sys.stdin.fileno()), mode, buffering)
+        else:
+            return os.fdopen(os.dup(sys.stdout.fileno()), mode, buffering)
+    else:
+        return open(path, mode, buffering)
+
+def main(csv_paths, *,
+        by=None,
+        fields=None,
+        defines=None,
+        sort=None,
+        **args):
+    # separate out renames
+    renames = list(it.chain.from_iterable(
+        ((k, v) for v in vs)
+        for k, vs in it.chain(by or [], fields or [])))
+    if by is not None:
+        by = [k for k, _ in by]
+    if fields is not None:
+        fields = [k for k, _ in fields]
+
+    # figure out types
+    types = {}
+    for t in TYPES.keys():
+        for k in args.get(t, []):
+            if k in types:
+                print("error: conflicting type for field %r?" % k)
+                sys.exit(-1)
+            types[k] = TYPES[t]
+    # rename types?
+    if renames:
+        types_ = {}
+        for new_k, old_k in renames:
+            if old_k in types:
+                types_[new_k] = types[old_k]
+        types.update(types_)
+
+    # figure out merge operations
+    ops = {}
+    for o in OPS.keys():
+        for k in args.get(o, []):
+            if k in ops:
+                print("error: conflicting op for field %r?" % k)
+                sys.exit(-1)
+            ops[k] = OPS[o]
+    # rename ops?
+    if renames:
+        ops_ = {}
+        for new_k, old_k in renames:
+            if old_k in ops:
+                ops_[new_k] = ops[old_k]
+        ops.update(ops_)
+
+    # find CSV files
+    results = []
+    for path in csv_paths:
+        try:
+            with openio(path) as f:
+                reader = csv.DictReader(f, restval='')
+                for r in reader:
+                    # rename fields?
+                    if renames:
+                        # make a copy so renames can overlap
+                        r_ = {}
+                        for new_k, old_k in renames:
+                            if old_k in r:
+                                r_[new_k] = r[old_k]
+                        r.update(r_)
+
+                    results.append(r)
+        except FileNotFoundError:
+            pass
+
+    # homogenize
+    Result = infer(results,
+        by=by,
+        fields=fields,
+        types=types,
+        ops=ops,
+        renames=renames)
+    results_ = []
+    for r in results:
+        if not any(k in r and r[k].strip()
+                for k in Result._fields):
+            continue
+        try:
+            results_.append(Result(**{
+                k: r[k] for k in Result._by + Result._fields
+                if k in r and r[k].strip()}))
+        except TypeError:
+            pass
+    results = results_
+
+    # fold
+    results = fold(Result, results, by=by, defines=defines)
+
+    # sort, note that python's sort is stable
+    results.sort()
+    if sort:
+        for k, reverse in reversed(sort):
+            results.sort(
+                key=lambda r: tuple(
+                    (getattr(r, k),) if getattr(r, k) is not None else ()
+                    for k in ([k] if k else Result._sort)),
+                reverse=reverse ^ (not k or k in Result._fields))
+
+    # write results to CSV
+    if args.get('output'):
+        with openio(args['output'], 'w') as f:
+            writer = csv.DictWriter(f, Result._by + Result._fields)
+            writer.writeheader()
+            for r in results:
+                # note we need to go through getattr to resolve lazy fields
+                writer.writerow({
+                    k: getattr(r, k) for k in Result._by + Result._fields})
+
+    # find previous results?
+    if args.get('diff'):
+        diff_results = []
+        try:
+            with openio(args['diff']) as f:
+                reader = csv.DictReader(f, restval='')
+                for r in reader:
+                    # rename fields?
+                    if renames:
+                        # make a copy so renames can overlap
+                        r_ = {}
+                        for new_k, old_k in renames:
+                            if old_k in r:
+                                r_[new_k] = r[old_k]
+                        r.update(r_)
+
+                    if not any(k in r and r[k].strip()
+                            for k in Result._fields):
+                        continue
+                    try:
+                        diff_results.append(Result(**{
+                            k: r[k] for k in Result._by + Result._fields
+                            if k in r and r[k].strip()}))
+                    except TypeError:
+                        pass
+        except FileNotFoundError:
+            pass
+
+        # fold
+        diff_results = fold(Result, diff_results, by=by, defines=defines)
+
+    # print table
+    if not args.get('quiet'):
+        table(Result, results,
+            diff_results if args.get('diff') else None,
+            by=by,
+            fields=fields,
+            sort=sort,
+            **args)
+
+
+if __name__ == "__main__":
+    import argparse
+    import sys
+    parser = argparse.ArgumentParser(
+        description="Summarize measurements in CSV files.",
+        allow_abbrev=False)
+    parser.add_argument(
+        'csv_paths',
+        nargs='*',
+        help="Input *.csv files.")
+    parser.add_argument(
+        '-q', '--quiet',
+        action='store_true',
+        help="Don't show anything, useful with -o.")
+    parser.add_argument(
+        '-o', '--output',
+        help="Specify CSV file to store results.")
+    parser.add_argument(
+        '-d', '--diff',
+        help="Specify CSV file to diff against.")
+    parser.add_argument(
+        '-a', '--all',
+        action='store_true',
+        help="Show all, not just the ones that changed.")
+    parser.add_argument(
+        '-p', '--percent',
+        action='store_true',
+        help="Only show percentage change, not a full diff.")
+    parser.add_argument(
+        '-b', '--by',
+        action='append',
+        type=lambda x: (
+            lambda k,v=None: (k, v.split(',') if v is not None else ())
+            )(*x.split('=', 1)),
+        help="Group by this field. Can rename fields with new_name=old_name.")
+    parser.add_argument(
+        '-f', '--field',
+        dest='fields',
+        action='append',
+        type=lambda x: (
+            lambda k,v=None: (k, v.split(',') if v is not None else ())
+            )(*x.split('=', 1)),
+        help="Show this field. Can rename fields with new_name=old_name.")
+    parser.add_argument(
+        '-D', '--define',
+        dest='defines',
+        action='append',
+        type=lambda x: (lambda k,v: (k, set(v.split(','))))(*x.split('=', 1)),
+        help="Only include results where this field is this value. May include "
+            "comma-separated options.")
+    class AppendSort(argparse.Action):
+        def __call__(self, parser, namespace, value, option):
+            if namespace.sort is None:
+                namespace.sort = []
+            namespace.sort.append((value, True if option == '-S' else False))
+    parser.add_argument(
+        '-s', '--sort',
+        nargs='?',
+        action=AppendSort,
+        help="Sort by this field.")
+    parser.add_argument(
+        '-S', '--reverse-sort',
+        nargs='?',
+        action=AppendSort,
+        help="Sort by this field, but backwards.")
+    parser.add_argument(
+        '-Y', '--summary',
+        action='store_true',
+        help="Only show the total.")
+    parser.add_argument(
+        '--int',
+        action='append',
+        help="Treat these fields as ints.")
+    parser.add_argument(
+        '--float',
+        action='append',
+        help="Treat these fields as floats.")
+    parser.add_argument(
+        '--frac',
+        action='append',
+        help="Treat these fields as fractions.")
+    parser.add_argument(
+        '--sum',
+        action='append',
+        help="Add these fields (the default).")
+    parser.add_argument(
+        '--prod',
+        action='append',
+        help="Multiply these fields.")
+    parser.add_argument(
+        '--min',
+        action='append',
+        help="Take the minimum of these fields.")
+    parser.add_argument(
+        '--max',
+        action='append',
+        help="Take the maximum of these fields.")
+    parser.add_argument(
+        '--mean',
+        action='append',
+        help="Average these fields.")
+    parser.add_argument(
+        '--stddev',
+        action='append',
+        help="Find the standard deviation of these fields.")
+    parser.add_argument(
+        '--gmean',
+        action='append',
+        help="Find the geometric mean of these fields.")
+    parser.add_argument(
+        '--gstddev',
+        action='append',
+        help="Find the geometric standard deviation of these fields.")
+    sys.exit(main(**{k: v
+        for k, v in vars(parser.parse_intermixed_args()).items()
+        if v is not None}))

kernel/fs/littlefs/scripts/tailpipe.py

@@ -0,0 +1,177 @@
+#!/usr/bin/env python3
+#
+# Efficiently displays the last n lines of a file/pipe.
+#
+# Example:
+# ./scripts/tailpipe.py trace -n5
+#
+# Copyright (c) 2022, The littlefs authors.
+# SPDX-License-Identifier: BSD-3-Clause
+#
+
+import collections as co
+import io
+import os
+import select
+import shutil
+import sys
+import threading as th
+import time
+
+
+def openio(path, mode='r', buffering=-1):
+    # allow '-' for stdin/stdout
+    if path == '-':
+        if mode == 'r':
+            return os.fdopen(os.dup(sys.stdin.fileno()), mode, buffering)
+        else:
+            return os.fdopen(os.dup(sys.stdout.fileno()), mode, buffering)
+    else:
+        return open(path, mode, buffering)
+
+class LinesIO:
+    def __init__(self, maxlen=None):
+        self.maxlen = maxlen
+        self.lines = co.deque(maxlen=maxlen)
+        self.tail = io.StringIO()
+
+        # trigger automatic sizing
+        if maxlen == 0:
+            self.resize(0)
+
+    def write(self, s):
+        # note using split here ensures the trailing string has no newline
+        lines = s.split('\n')
+
+        if len(lines) > 1 and self.tail.getvalue():
+            self.tail.write(lines[0])
+            lines[0] = self.tail.getvalue()
+            self.tail = io.StringIO()
+
+        self.lines.extend(lines[:-1])
+
+        if lines[-1]:
+            self.tail.write(lines[-1])
+
+    def resize(self, maxlen):
+        self.maxlen = maxlen
+        if maxlen == 0:
+            maxlen = shutil.get_terminal_size((80, 5))[1]
+        if maxlen != self.lines.maxlen:
+            self.lines = co.deque(self.lines, maxlen=maxlen)
+
+    canvas_lines = 1
+    def draw(self):
+        # did terminal size change?
+        if self.maxlen == 0:
+            self.resize(0)
+
+        # first thing first, give ourself a canvas
+        while LinesIO.canvas_lines < len(self.lines):
+            sys.stdout.write('\n')
+            LinesIO.canvas_lines += 1
+
+        # clear the bottom of the canvas if we shrink
+        shrink = LinesIO.canvas_lines - len(self.lines)
+        if shrink > 0:
+            for i in range(shrink):
+                sys.stdout.write('\r')
+                if shrink-1-i > 0:
+                    sys.stdout.write('\x1b[%dA' % (shrink-1-i))
+                sys.stdout.write('\x1b[K')
+                if shrink-1-i > 0:
+                    sys.stdout.write('\x1b[%dB' % (shrink-1-i))
+            sys.stdout.write('\x1b[%dA' % shrink)
+            LinesIO.canvas_lines = len(self.lines)
+
+        for i, line in enumerate(self.lines):
+            # move cursor, clear line, disable/reenable line wrapping
+            sys.stdout.write('\r')
+            if len(self.lines)-1-i > 0:
+                sys.stdout.write('\x1b[%dA' % (len(self.lines)-1-i))
+            sys.stdout.write('\x1b[K')
+            sys.stdout.write('\x1b[?7l')
+            sys.stdout.write(line)
+            sys.stdout.write('\x1b[?7h')
+            if len(self.lines)-1-i > 0:
+                sys.stdout.write('\x1b[%dB' % (len(self.lines)-1-i))
+        sys.stdout.flush()
+
+
+def main(path='-', *, lines=5, cat=False, sleep=None, keep_open=False):
+    if cat:
+        ring = sys.stdout
+    else:
+        ring = LinesIO(lines)
+
+    # if sleep print in background thread to avoid getting stuck in a read call
+    event = th.Event()
+    lock = th.Lock()
+    if not cat:
+        done = False
+        def background():
+            while not done:
+                event.wait()
+                event.clear()
+                with lock:
+                    ring.draw()
+                time.sleep(sleep or 0.01)
+        th.Thread(target=background, daemon=True).start()
+
+    try:
+        while True:
+            with openio(path) as f:
+                for line in f:
+                    with lock:
+                        ring.write(line)
+                        event.set()
+
+            if not keep_open:
+                break
+            # don't just flood open calls
+            time.sleep(sleep or 0.1)
+    except FileNotFoundError as e:
+        print("error: file not found %r" % path)
+        sys.exit(-1)
+    except KeyboardInterrupt:
+        pass
+
+    if not cat:
+        done = True
+        lock.acquire() # avoids https://bugs.python.org/issue42717
+        sys.stdout.write('\n')
+
+
+if __name__ == "__main__":
+    import sys
+    import argparse
+    parser = argparse.ArgumentParser(
+        description="Efficiently displays the last n lines of a file/pipe.",
+        allow_abbrev=False)
+    parser.add_argument(
+        'path',
+        nargs='?',
+        help="Path to read from.")
+    parser.add_argument(
+        '-n', '--lines',
+        nargs='?',
+        type=lambda x: int(x, 0),
+        const=0,
+        help="Show this many lines of history. 0 uses the terminal height. "
+            "Defaults to 5.")
+    parser.add_argument(
+        '-z', '--cat',
+        action='store_true',
+        help="Pipe directly to stdout.")
+    parser.add_argument(
+        '-s', '--sleep',
+        type=float,
+        help="Seconds to sleep between reads. Defaults to 0.01.")
+    parser.add_argument(
+        '-k', '--keep-open',
+        action='store_true',
+        help="Reopen the pipe on EOF, useful when multiple "
+            "processes are writing.")
+    sys.exit(main(**{k: v
+        for k, v in vars(parser.parse_intermixed_args()).items()
+        if v is not None}))

kernel/fs/littlefs/scripts/teepipe.py

@@ -0,0 +1,73 @@
+#!/usr/bin/env python3
+#
+# tee, but for pipes
+#
+# Example:
+# ./scripts/tee.py in_pipe out_pipe1 out_pipe2
+#
+# Copyright (c) 2022, The littlefs authors.
+# SPDX-License-Identifier: BSD-3-Clause
+#
+
+import os
+import io
+import time
+import sys
+
+
+def openio(path, mode='r', buffering=-1):
+    # allow '-' for stdin/stdout
+    if path == '-':
+        if mode == 'r':
+            return os.fdopen(os.dup(sys.stdin.fileno()), mode, buffering)
+        else:
+            return os.fdopen(os.dup(sys.stdout.fileno()), mode, buffering)
+    else:
+        return open(path, mode, buffering)
+
+def main(in_path, out_paths, *, keep_open=False):
+    out_pipes = [openio(p, 'wb', 0) for p in out_paths]
+    try:
+        with openio(in_path, 'rb', 0) as f:
+            while True:
+                buf = f.read(io.DEFAULT_BUFFER_SIZE)
+                if not buf:
+                    if not keep_open:
+                        break
+                    # don't just flood reads
+                    time.sleep(0.1)
+                    continue
+
+                for p in out_pipes:
+                    try:
+                        p.write(buf)
+                    except BrokenPipeError:
+                        pass
+    except FileNotFoundError as e:
+        print("error: file not found %r" % in_path)
+        sys.exit(-1)
+    except KeyboardInterrupt:
+        pass
+
+
+if __name__ == "__main__":
+    import sys
+    import argparse
+    parser = argparse.ArgumentParser(
+        description="tee, but for pipes.",
+        allow_abbrev=False)
+    parser.add_argument(
+        'in_path',
+        help="Path to read from.")
+    parser.add_argument(
+        'out_paths',
+        nargs='+',
+        help="Path to write to.")
+    parser.add_argument(
+        '-k', '--keep-open',
+        action='store_true',
+        help="Reopen the pipe on EOF, useful when multiple "
+            "processes are writing.")
+    sys.exit(main(**{k: v
+        for k, v in vars(parser.parse_intermixed_args()).items()
+        if v is not None}))

kernel/fs/littlefs/scripts/watch.py

@@ -0,0 +1,265 @@
+#!/usr/bin/env python3
+#
+# Traditional watch command, but with higher resolution updates and a bit
+# different options/output format
+#
+# Example:
+# ./scripts/watch.py -s0.1 date
+#
+# Copyright (c) 2022, The littlefs authors.
+# SPDX-License-Identifier: BSD-3-Clause
+#
+
+import collections as co
+import errno
+import fcntl
+import io
+import os
+import pty
+import re
+import shutil
+import struct
+import subprocess as sp
+import sys
+import termios
+import time
+
+try:
+    import inotify_simple
+except ModuleNotFoundError:
+    inotify_simple = None
+
+
+def openio(path, mode='r', buffering=-1):
+    # allow '-' for stdin/stdout
+    if path == '-':
+        if mode == 'r':
+            return os.fdopen(os.dup(sys.stdin.fileno()), mode, buffering)
+        else:
+            return os.fdopen(os.dup(sys.stdout.fileno()), mode, buffering)
+    else:
+        return open(path, mode, buffering)
+
+def inotifywait(paths):
+    # wait for interesting events
+    inotify = inotify_simple.INotify()
+    flags = (inotify_simple.flags.ATTRIB
+        | inotify_simple.flags.CREATE
+        | inotify_simple.flags.DELETE
+        | inotify_simple.flags.DELETE_SELF
+        | inotify_simple.flags.MODIFY
+        | inotify_simple.flags.MOVED_FROM
+        | inotify_simple.flags.MOVED_TO
+        | inotify_simple.flags.MOVE_SELF)
+
+    # recurse into directories
+    for path in paths:
+        if os.path.isdir(path):
+            for dir, _, files in os.walk(path):
+                inotify.add_watch(dir, flags)
+                for f in files:
+                    inotify.add_watch(os.path.join(dir, f), flags)
+        else:
+            inotify.add_watch(path, flags)
+
+    # wait for event
+    inotify.read()
+
+class LinesIO:
+    def __init__(self, maxlen=None):
+        self.maxlen = maxlen
+        self.lines = co.deque(maxlen=maxlen)
+        self.tail = io.StringIO()
+
+        # trigger automatic sizing
+        if maxlen == 0:
+            self.resize(0)
+
+    def write(self, s):
+        # note using split here ensures the trailing string has no newline
+        lines = s.split('\n')
+
+        if len(lines) > 1 and self.tail.getvalue():
+            self.tail.write(lines[0])
+            lines[0] = self.tail.getvalue()
+            self.tail = io.StringIO()
+
+        self.lines.extend(lines[:-1])
+
+        if lines[-1]:
+            self.tail.write(lines[-1])
+
+    def resize(self, maxlen):
+        self.maxlen = maxlen
+        if maxlen == 0:
+            maxlen = shutil.get_terminal_size((80, 5))[1]
+        if maxlen != self.lines.maxlen:
+            self.lines = co.deque(self.lines, maxlen=maxlen)
+
+    canvas_lines = 1
+    def draw(self):
+        # did terminal size change?
+        if self.maxlen == 0:
+            self.resize(0)
+
+        # first thing first, give ourself a canvas
+        while LinesIO.canvas_lines < len(self.lines):
+            sys.stdout.write('\n')
+            LinesIO.canvas_lines += 1
+
+        # clear the bottom of the canvas if we shrink
+        shrink = LinesIO.canvas_lines - len(self.lines)
+        if shrink > 0:
+            for i in range(shrink):
+                sys.stdout.write('\r')
+                if shrink-1-i > 0:
+                    sys.stdout.write('\x1b[%dA' % (shrink-1-i))
+                sys.stdout.write('\x1b[K')
+                if shrink-1-i > 0:
+                    sys.stdout.write('\x1b[%dB' % (shrink-1-i))
+            sys.stdout.write('\x1b[%dA' % shrink)
+            LinesIO.canvas_lines = len(self.lines)
+
+        for i, line in enumerate(self.lines):
+            # move cursor, clear line, disable/reenable line wrapping
+            sys.stdout.write('\r')
+            if len(self.lines)-1-i > 0:
+                sys.stdout.write('\x1b[%dA' % (len(self.lines)-1-i))
+            sys.stdout.write('\x1b[K')
+            sys.stdout.write('\x1b[?7l')
+            sys.stdout.write(line)
+            sys.stdout.write('\x1b[?7h')
+            if len(self.lines)-1-i > 0:
+                sys.stdout.write('\x1b[%dB' % (len(self.lines)-1-i))
+        sys.stdout.flush()
+
+
+def main(command, *,
+        lines=0,
+        cat=False,
+        sleep=None,
+        keep_open=False,
+        keep_open_paths=None,
+        exit_on_error=False):
+    returncode = 0
+    try:
+        while True:
+            # reset ring each run
+            if cat:
+                ring = sys.stdout
+            else:
+                ring = LinesIO(lines)
+
+            try:
+                # run the command under a pseudoterminal 
+                mpty, spty = pty.openpty()
+
+                # forward terminal size
+                w, h = shutil.get_terminal_size((80, 5))
+                if lines:
+                    h = lines
+                fcntl.ioctl(spty, termios.TIOCSWINSZ,
+                    struct.pack('HHHH', h, w, 0, 0))
+
+                proc = sp.Popen(command,
+                    stdout=spty,
+                    stderr=spty,
+                    close_fds=False)
+                os.close(spty)
+                mpty = os.fdopen(mpty, 'r', 1)
+
+                while True:
+                    try:
+                        line = mpty.readline()
+                    except OSError as e:
+                        if e.errno != errno.EIO:
+                            raise
+                        break
+                    if not line:
+                        break
+
+                    ring.write(line)
+                    if not cat:
+                        ring.draw()
+
+                mpty.close()
+                proc.wait()
+                if exit_on_error and proc.returncode != 0:
+                    returncode = proc.returncode
+                    break
+            except OSError as e:
+                if e.errno != errno.ETXTBSY:
+                    raise
+                pass
+
+            # try to inotifywait
+            if keep_open and inotify_simple is not None:
+                if keep_open_paths:
+                    paths = set(keep_paths)
+                else:
+                    # guess inotify paths from command
+                    paths = set()
+                    for p in command:
+                        for p in {
+                                p,
+                                re.sub('^-.', '', p),
+                                re.sub('^--[^=]+=', '', p)}:
+                            if p and os.path.exists(p):
+                                paths.add(p)
+                ptime = time.time()
+                inotifywait(paths)
+                # sleep for a minimum amount of time, this helps issues around
+                # rapidly updating files
+                time.sleep(max(0, (sleep or 0.1) - (time.time()-ptime)))
+            else:
+                time.sleep(sleep or 0.1)
+    except KeyboardInterrupt:
+        pass
+
+    if not cat:
+        sys.stdout.write('\n')
+    sys.exit(returncode)
+
+
+if __name__ == "__main__":
+    import sys
+    import argparse
+    parser = argparse.ArgumentParser(
+        description="Traditional watch command, but with higher resolution "
+            "updates and a bit different options/output format.",
+        allow_abbrev=False)
+    parser.add_argument(
+        'command',
+        nargs=argparse.REMAINDER,
+        help="Command to run.")
+    parser.add_argument(
+        '-n', '--lines',
+        nargs='?',
+        type=lambda x: int(x, 0),
+        const=0,
+        help="Show this many lines of history. 0 uses the terminal height. "
+            "Defaults to 0.")
+    parser.add_argument(
+        '-z', '--cat',
+        action='store_true',
+        help="Pipe directly to stdout.")
+    parser.add_argument(
+        '-s', '--sleep',
+        type=float,
+        help="Seconds to sleep between runs. Defaults to 0.1.")
+    parser.add_argument(
+        '-k', '--keep-open',
+        action='store_true',
+        help="Try to use inotify to wait for changes.")
+    parser.add_argument(
+        '-K', '--keep-open-path',
+        dest='keep_open_paths',
+        action='append',
+        help="Use this path for inotify. Defaults to guessing.")
+    parser.add_argument(
+        '-e', '--exit-on-error',
+        action='store_true',
+        help="Exit on error.")
+    sys.exit(main(**{k: v
+        for k, v in vars(parser.parse_args()).items()
+        if v is not None}))