Introduce Virtio-loopback epsilon release:

Epsilon release introduces a new compatibility layer which make virtio-loopback design to work with QEMU and rust-vmm vhost-user backend without require any changes. Signed-off-by: Timos Ampelikiotis <t.ampelikiotis@virtualopensystems.com> Change-Id: I52e57563e08a7d0bdc002f8e928ee61ba0c53dd9
author: Timos Ampelikiotis <t.ampelikiotis@virtualopensystems.com> 2023-10-10 11:40:56 +0000
committer: Timos Ampelikiotis <t.ampelikiotis@virtualopensystems.com> 2023-10-10 11:40:56 +0000
commit: e02cda008591317b1625707ff8e115a4841aa889 (patch)
tree: aee302e3cf8b59ec2d32ec481be3d1afddfc8968 /scripts/codeconverter
parent: cc668e6b7e0ffd8c9d130513d12053cf5eda1d3b (diff)
9 files changed, 2995 insertions, 0 deletions
diff --git a/scripts/codeconverter/codeconverter/__init__.py b/scripts/codeconverter/codeconverter/__init__.py
new file mode 100644
index 000000000..e69de29bb
--- /dev/null
+++ b/scripts/codeconverter/codeconverter/__init__.py
diff --git a/scripts/codeconverter/codeconverter/patching.py b/scripts/codeconverter/codeconverter/patching.py
new file mode 100644
index 000000000..9e92505d3
--- /dev/null
+++ b/scripts/codeconverter/codeconverter/patching.py
@@ -0,0 +1,466 @@
+# Copyright (C) 2020 Red Hat Inc.
+#
+# Authors:
+#  Eduardo Habkost <ehabkost@redhat.com>
+#
+# This work is licensed under the terms of the GNU GPL, version 2.  See
+# the COPYING file in the top-level directory.
+from typing import IO, Match, NamedTuple, Optional, Literal, Iterable, Type, Dict, List, Any, TypeVar, NewType, Tuple, Union
+from pathlib import Path
+from itertools import chain
+from tempfile import NamedTemporaryFile
+import os
+import re
+import subprocess
+from io import StringIO
+
+import logging
+logger = logging.getLogger(__name__)
+DBG = logger.debug
+INFO = logger.info
+WARN = logger.warning
+ERROR = logger.error
+
+from .utils import *
+
+T = TypeVar('T')
+
+class Patch(NamedTuple):
+    # start inside file.original_content
+    start: int
+    # end position inside file.original_content
+    end: int
+    # replacement string for file.original_content[start:end]
+    replacement: str
+
+IdentifierType = Literal['type', 'symbol', 'include', 'constant']
+class RequiredIdentifier(NamedTuple):
+    type: IdentifierType
+    name: str
+
+class FileMatch:
+    """Base class for regex matches
+
+    Subclasses just need to set the `regexp` class attribute
+    """
+    regexp: Optional[str] = None
+
+    def __init__(self, f: 'FileInfo', m: Match) -> None:
+        self.file: 'FileInfo' = f
+        self.match: Match[str] = m
+
+    @property
+    def name(self) -> str:
+        if 'name' not in self.match.groupdict():
+            return '[no name]'
+        return self.group('name')
+
+    @classmethod
+    def compiled_re(klass):
+        return re.compile(klass.regexp, re.MULTILINE)
+
+    def start(self) -> int:
+        return self.match.start()
+
+    def end(self) -> int:
+        return self.match.end()
+
+    def line_col(self) -> LineAndColumn:
+        return self.file.line_col(self.start())
+
+    def group(self, group: Union[int, str]) -> str:
+        return self.match.group(group)
+
+    def getgroup(self, group: str) -> Optional[str]:
+        if group not in self.match.groupdict():
+            return None
+        return self.match.group(group)
+
+    def log(self, level, fmt, *args) -> None:
+        pos = self.line_col()
+        logger.log(level, '%s:%d:%d: '+fmt, self.file.filename, pos.line, pos.col, *args)
+
+    def debug(self, fmt, *args) -> None:
+        self.log(logging.DEBUG, fmt, *args)
+
+    def info(self, fmt, *args) -> None:
+        self.log(logging.INFO, fmt, *args)
+
+    def warn(self, fmt, *args) -> None:
+        self.log(logging.WARNING, fmt, *args)
+
+    def error(self, fmt, *args) -> None:
+        self.log(logging.ERROR, fmt, *args)
+
+    def sub(self, original: str, replacement: str) -> str:
+        """Replace content
+
+        XXX: this won't use the match position, but will just
+        replace all strings that look like the original match.
+        This should be enough for all the patterns used in this
+        script.
+        """
+        return original.replace(self.group(0), replacement)
+
+    def sanity_check(self) -> None:
+        """Sanity check match, and print warnings if necessary"""
+        pass
+
+    def replacement(self) -> Optional[str]:
+        """Return replacement text for pattern, to use new code conventions"""
+        return None
+
+    def make_patch(self, replacement: str) -> 'Patch':
+        """Make patch replacing the content of this match"""
+        return Patch(self.start(), self.end(), replacement)
+
+    def make_subpatch(self, start: int, end: int, replacement: str) -> 'Patch':
+        return Patch(self.start() + start, self.start() + end, replacement)
+
+    def make_removal_patch(self) -> 'Patch':
+        """Make patch removing contents of match completely"""
+        return self.make_patch('')
+
+    def append(self, s: str) -> 'Patch':
+        """Make patch appending string after this match"""
+        return Patch(self.end(), self.end(), s)
+
+    def prepend(self, s: str) -> 'Patch':
+        """Make patch prepending string before this match"""
+        return Patch(self.start(), self.start(), s)
+
+    def gen_patches(self) -> Iterable['Patch']:
+        """Patch source code contents to use new code patterns"""
+        replacement = self.replacement()
+        if replacement is not None:
+            yield self.make_patch(replacement)
+
+    @classmethod
+    def has_replacement_rule(klass) -> bool:
+        return (klass.gen_patches is not FileMatch.gen_patches
+                or klass.replacement is not FileMatch.replacement)
+
+    def contains(self, other: 'FileMatch') -> bool:
+        return other.start() >= self.start() and other.end() <= self.end()
+
+    def __repr__(self) -> str:
+        start = self.file.line_col(self.start())
+        end = self.file.line_col(self.end() - 1)
+        return '<%s %s at %d:%d-%d:%d: %r>' % (self.__class__.__name__,
+                                                    self.name,
+                                                    start.line, start.col,
+                                                    end.line, end.col, self.group(0)[:100])
+
+    def required_identifiers(self) -> Iterable[RequiredIdentifier]:
+        """Can be implemented by subclasses to keep track of identifier references
+
+        This method will be used by the code that moves declarations around the file,
+        to make sure we find the right spot for them.
+        """
+        raise NotImplementedError()
+
+    def provided_identifiers(self) -> Iterable[RequiredIdentifier]:
+        """Can be implemented by subclasses to keep track of identifier references
+
+        This method will be used by the code that moves declarations around the file,
+        to make sure we find the right spot for them.
+        """
+        raise NotImplementedError()
+
+    @classmethod
+    def finditer(klass, content: str, pos=0, endpos=-1) -> Iterable[Match]:
+        """Helper for re.finditer()"""
+        if endpos >= 0:
+            content = content[:endpos]
+        return klass.compiled_re().finditer(content, pos)
+
+    @classmethod
+    def domatch(klass, content: str, pos=0, endpos=-1) -> Optional[Match]:
+        """Helper for re.match()"""
+        if endpos >= 0:
+            content = content[:endpos]
+        return klass.compiled_re().match(content, pos)
+
+    def group_finditer(self, klass: Type['FileMatch'], group: Union[str, int]) -> Iterable['FileMatch']:
+        assert self.file.original_content
+        return (klass(self.file, m)
+                for m in klass.finditer(self.file.original_content,
+                                        self.match.start(group),
+                                        self.match.end(group)))
+
+    def try_group_match(self, klass: Type['FileMatch'], group: Union[str, int]) -> Optional['FileMatch']:
+        assert self.file.original_content
+        m = klass.domatch(self.file.original_content,
+                          self.match.start(group),
+                          self.match.end(group))
+        if not m:
+            return None
+        else:
+            return klass(self.file, m)
+
+    def group_match(self, group: Union[str, int]) -> 'FileMatch':
+        m = self.try_group_match(FullMatch, group)
+        assert m
+        return m
+
+    @property
+    def allfiles(self) -> 'FileList':
+        return self.file.allfiles
+
+class FullMatch(FileMatch):
+    """Regexp that will match all contents of string
+    Useful when used with group_match()
+    """
+    regexp = r'(?s).*' # (?s) is re.DOTALL
+
+def all_subclasses(c: Type[FileMatch]) -> Iterable[Type[FileMatch]]:
+    for sc in c.__subclasses__():
+        yield sc
+        yield from all_subclasses(sc)
+
+def match_class_dict() -> Dict[str, Type[FileMatch]]:
+    d = dict((t.__name__, t) for t in all_subclasses(FileMatch))
+    return d
+
+def names(matches: Iterable[FileMatch]) -> Iterable[str]:
+    return [m.name for m in matches]
+
+class PatchingError(Exception):
+    pass
+
+class OverLappingPatchesError(PatchingError):
+    pass
+
+def apply_patches(s: str, patches: Iterable[Patch]) -> str:
+    """Apply a sequence of patches to string
+
+    >>> apply_patches('abcdefg', [Patch(2,2,'xxx'), Patch(0, 1, 'yy')])
+    'yybxxxcdefg'
+    """
+    r = StringIO()
+    last = 0
+    def patch_sort_key(item: Tuple[int, Patch]) -> Tuple[int, int, int]:
+        """Patches are sorted by byte position,
+        patches at the same byte position are applied in the order
+        they were generated.
+        """
+        i,p = item
+        return (p.start, p.end, i)
+
+    for i,p in sorted(enumerate(patches), key=patch_sort_key):
+        DBG("Applying patch at position %d (%s) - %d (%s): %r",
+            p.start, line_col(s, p.start),
+            p.end, line_col(s, p.end),
+            p.replacement)
+        if last > p.start:
+            raise OverLappingPatchesError("Overlapping patch at position %d (%s), last patch at %d (%s)" % \
+                (p.start, line_col(s, p.start), last, line_col(s, last)))
+        r.write(s[last:p.start])
+        r.write(p.replacement)
+        last = p.end
+    r.write(s[last:])
+    return r.getvalue()
+
+class RegexpScanner:
+    def __init__(self) -> None:
+        self.match_index: Dict[Type[Any], List[FileMatch]] = {}
+        self.match_name_index: Dict[Tuple[Type[Any], str, str], Optional[FileMatch]] = {}
+
+    def _matches_of_type(self, klass: Type[Any]) -> Iterable[FileMatch]:
+        raise NotImplementedError()
+
+    def matches_of_type(self, t: Type[T]) -> List[T]:
+        if t not in self.match_index:
+            self.match_index[t] = list(self._matches_of_type(t))
+        return self.match_index[t] # type: ignore
+
+    def find_matches(self, t: Type[T], name: str, group: str='name') -> List[T]:
+        indexkey = (t, name, group)
+        if indexkey in self.match_name_index:
+            return self.match_name_index[indexkey] # type: ignore
+        r: List[T] = []
+        for m in self.matches_of_type(t):
+            assert isinstance(m, FileMatch)
+            if m.getgroup(group) == name:
+                r.append(m) # type: ignore
+        self.match_name_index[indexkey] = r # type: ignore
+        return r
+
+    def find_match(self, t: Type[T], name: str, group: str='name') -> Optional[T]:
+        l = self.find_matches(t, name, group)
+        if not l:
+            return None
+        if len(l) > 1:
+            logger.warn("multiple matches found for %r (%s=%r)", t, group, name)
+            return None
+        return l[0]
+
+    def reset_index(self) -> None:
+        self.match_index.clear()
+        self.match_name_index.clear()
+
+class FileInfo(RegexpScanner):
+    filename: Path
+    original_content: Optional[str] = None
+
+    def __init__(self, files: 'FileList', filename: os.PathLike, force:bool=False) -> None:
+        super().__init__()
+        self.allfiles = files
+        self.filename = Path(filename)
+        self.patches: List[Patch] = []
+        self.force = force
+
+    def __repr__(self) -> str:
+        return f'<FileInfo {repr(self.filename)}>'
+
+    def filename_matches(self, name: str) -> bool:
+        nameparts = Path(name).parts
+        return self.filename.parts[-len(nameparts):] == nameparts
+
+    def line_col(self, start: int) -> LineAndColumn:
+        """Return line and column for a match object inside original_content"""
+        return line_col(self.original_content, start)
+
+    def _matches_of_type(self, klass: Type[Any]) -> List[FileMatch]:
+        """Build FileMatch objects for each match of regexp"""
+        if not hasattr(klass, 'regexp') or klass.regexp is None:
+            return []
+        assert hasattr(klass, 'regexp')
+        DBG("%s: scanning for %s", self.filename, klass.__name__)
+        DBG("regexp: %s", klass.regexp)
+        matches = [klass(self, m) for m in klass.finditer(self.original_content)]
+        DBG('%s: %d matches found for %s: %s', self.filename, len(matches),
+            klass.__name__,' '.join(names(matches)))
+        return matches
+
+    def find_match(self, t: Type[T], name: str, group: str='name') -> Optional[T]:
+        for m in self.matches_of_type(t):
+            assert isinstance(m, FileMatch)
+            if m.getgroup(group) == name:
+                return m # type: ignore
+        return None
+
+    def reset_content(self, s:str):
+        self.original_content = s
+        self.patches.clear()
+        self.reset_index()
+        self.allfiles.reset_index()
+
+    def load(self) -> None:
+        if self.original_content is not None:
+            return
+        with open(self.filename, 'rt') as f:
+            self.reset_content(f.read())
+
+    @property
+    def all_matches(self) -> Iterable[FileMatch]:
+        lists = list(self.match_index.values())
+        return (m for l in lists
+                  for m in l)
+
+    def gen_patches(self, matches: List[FileMatch]) -> None:
+        for m in matches:
+            DBG("Generating patches for %r", m)
+            for i,p in enumerate(m.gen_patches()):
+                DBG("patch %d generated by %r:", i, m)
+                DBG("replace contents at %s-%s with %r",
+                    self.line_col(p.start), self.line_col(p.end), p.replacement)
+                self.patches.append(p)
+
+    def scan_for_matches(self, class_names: Optional[List[str]]=None) -> Iterable[FileMatch]:
+        DBG("class names: %r", class_names)
+        class_dict = match_class_dict()
+        if class_names is None:
+            DBG("default class names")
+            class_names = list(name for name,klass in class_dict.items()
+                               if klass.has_replacement_rule())
+        DBG("class_names: %r", class_names)
+        for cn in class_names:
+            matches = self.matches_of_type(class_dict[cn])
+            DBG('%d matches found for %s: %s',
+                    len(matches), cn, ' '.join(names(matches)))
+            yield from matches
+
+    def apply_patches(self) -> None:
+        """Replace self.original_content after applying patches from self.patches"""
+        self.reset_content(self.get_patched_content())
+
+    def get_patched_content(self) -> str:
+        assert self.original_content is not None
+        return apply_patches(self.original_content, self.patches)
+
+    def write_to_file(self, f: IO[str]) -> None:
+        f.write(self.get_patched_content())
+
+    def write_to_filename(self, filename: os.PathLike) -> None:
+        with open(filename, 'wt') as of:
+            self.write_to_file(of)
+
+    def patch_inplace(self) -> None:
+        newfile = self.filename.with_suffix('.changed')
+        self.write_to_filename(newfile)
+        os.rename(newfile, self.filename)
+
+    def show_diff(self) -> None:
+        with NamedTemporaryFile('wt') as f:
+            self.write_to_file(f)
+            f.flush()
+            subprocess.call(['diff', '-u', self.filename, f.name])
+
+    def ref(self):
+        return TypeInfoReference
+
+class FileList(RegexpScanner):
+    def __init__(self):
+        super().__init__()
+        self.files: List[FileInfo] = []
+
+    def extend(self, *args, **kwargs):
+        self.files.extend(*args, **kwargs)
+
+    def __iter__(self):
+        return iter(self.files)
+
+    def _matches_of_type(self, klass: Type[Any]) -> Iterable[FileMatch]:
+        return chain(*(f._matches_of_type(klass) for f in self.files))
+
+    def find_file(self, name: str) -> Optional[FileInfo]:
+        """Get file with path ending with @name"""
+        for f in self.files:
+            if f.filename_matches(name):
+                return f
+        else:
+            return None
+
+    def one_pass(self, class_names: List[str]) -> int:
+        total_patches = 0
+        for f in self.files:
+            INFO("Scanning file %s", f.filename)
+            matches = list(f.scan_for_matches(class_names))
+            INFO("Generating patches for file %s", f.filename)
+            f.gen_patches(matches)
+            total_patches += len(f.patches)
+        if total_patches:
+            for f in self.files:
+                try:
+                    f.apply_patches()
+                except PatchingError:
+                    logger.exception("%s: failed to patch file", f.filename)
+        return total_patches
+
+    def patch_content(self, max_passes, class_names: List[str]) -> None:
+        """Multi-pass content patching loop
+
+        We run multiple passes because there are rules that will
+        delete init functions once they become empty.
+        """
+        passes = 0
+        total_patches  = 0
+        DBG("max_passes: %r", max_passes)
+        while not max_passes or max_passes <= 0 or passes < max_passes:
+            passes += 1
+            INFO("Running pass: %d", passes)
+            count = self.one_pass(class_names)
+            DBG("patch content: pass %d: %d patches generated", passes, count)
+            total_patches += count
+        DBG("%d patches applied total in %d passes", total_patches, passes)
diff --git a/scripts/codeconverter/codeconverter/qom_macros.py b/scripts/codeconverter/codeconverter/qom_macros.py
new file mode 100644
index 000000000..2d2f2055a
--- /dev/null
+++ b/scripts/codeconverter/codeconverter/qom_macros.py
@@ -0,0 +1,861 @@
+# Copyright (C) 2020 Red Hat Inc.
+#
+# Authors:
+#  Eduardo Habkost <ehabkost@redhat.com>
+#
+# This work is licensed under the terms of the GNU GPL, version 2.  See
+# the COPYING file in the top-level directory.
+import re
+from itertools import chain
+from typing import *
+
+from .regexps import *
+from .patching import *
+from .utils import *
+
+import logging
+logger = logging.getLogger(__name__)
+DBG = logger.debug
+INFO = logger.info
+WARN = logger.warning
+
+# simple expressions:
+
+RE_CONSTANT = OR(RE_STRING, RE_NUMBER)
+
+class DefineDirective(FileMatch):
+    """Match any #define directive"""
+    regexp = S(r'^[ \t]*#[ \t]*define', CPP_SPACE, NAMED('name', RE_IDENTIFIER), r'\b')
+
+class ExpressionDefine(FileMatch):
+    """Simple #define preprocessor directive for an expression"""
+    regexp = S(r'^[ \t]*#[ \t]*define', CPP_SPACE, NAMED('name', RE_IDENTIFIER),
+               CPP_SPACE, NAMED('value', RE_EXPRESSION), r'[ \t]*\n')
+
+    def provided_identifiers(self) -> Iterable[RequiredIdentifier]:
+        yield RequiredIdentifier('constant', self.group('name'))
+
+class ConstantDefine(ExpressionDefine):
+    """Simple #define preprocessor directive for a number or string constant"""
+    regexp = S(r'^[ \t]*#[ \t]*define', CPP_SPACE, NAMED('name', RE_IDENTIFIER),
+               CPP_SPACE, NAMED('value', RE_CONSTANT), r'[ \t]*\n')
+
+
+class TypeIdentifiers(NamedTuple):
+    """Type names found in type declarations"""
+    # TYPE_MYDEVICE
+    typename: Optional[str]
+    # MYDEVICE
+    uppercase: Optional[str] = None
+    # MyDevice
+    instancetype: Optional[str] = None
+    # MyDeviceClass
+    classtype: Optional[str] = None
+    # my_device
+    lowercase: Optional[str] = None
+
+    def allfields(self):
+        return tuple(getattr(self, f) for f in self._fields)
+
+    def merge(self, other: 'TypeIdentifiers') -> Optional['TypeIdentifiers']:
+        """Check if identifiers match, return new identifier with complete list"""
+        if any(not opt_compare(a, b) for a,b in zip(self, other)):
+            return None
+        return TypeIdentifiers(*(merge(a, b) for a,b in zip(self, other)))
+
+    def __str__(self) -> str:
+        values = ((f, getattr(self, f)) for f in self._fields)
+        s = ', '.join('%s=%s' % (f,v) for f,v in values if v is not None)
+        return f'{s}'
+
+    def check_consistency(self) -> List[str]:
+        """Check if identifiers are consistent with each other,
+        return list of problems (or empty list if everything seems consistent)
+        """
+        r = []
+        if self.typename is None:
+            r.append("typename (TYPE_MYDEVICE) is unavailable")
+
+        if self.uppercase is None:
+            r.append("uppercase name is unavailable")
+
+        if (self.instancetype is not None
+            and self.classtype is not None
+            and self.classtype != f'{self.instancetype}Class'):
+                r.append("class typedef %s doesn't match instance typedef %s" %
+                         (self.classtype, self.instancetype))
+
+        if (self.uppercase is not None
+            and self.typename is not None
+            and f'TYPE_{self.uppercase}' != self.typename):
+            r.append("uppercase name (%s) doesn't match type name (%s)" %
+                     (self.uppercase, self.typename))
+
+        return r
+
+class TypedefMatch(FileMatch):
+    """typedef declaration"""
+    def provided_identifiers(self) -> Iterable[RequiredIdentifier]:
+        yield RequiredIdentifier('type', self.group('name'))
+
+class SimpleTypedefMatch(TypedefMatch):
+    """Simple typedef declaration
+    (no replacement rules)"""
+    regexp = S(r'^[ \t]*typedef', SP,
+               NAMED('typedef_type', RE_TYPE), SP,
+               NAMED('name', RE_IDENTIFIER), r'\s*;[ \t]*\n')
+
+RE_MACRO_DEFINE = S(r'^[ \t]*#\s*define\s+', NAMED('name', RE_IDENTIFIER),
+                    r'\s*\(\s*', RE_IDENTIFIER, r'\s*\)', CPP_SPACE)
+
+RE_STRUCT_ATTRIBUTE = r'QEMU_PACKED'
+
+# This doesn't parse the struct definitions completely, it just assumes
+# the closing brackets are going to be in an unindented line:
+RE_FULL_STRUCT = S('struct', SP, M(RE_IDENTIFIER, n='?', name='structname'), SP,
+                   NAMED('body', r'{\n',
+                         # acceptable inside the struct body:
+                         # - lines starting with space or tab
+                         # - empty lines
+                         # - preprocessor directives
+                         # - comments
+                         OR(r'[ \t][^\n]*\n',
+                            r'#[^\n]*\n',
+                            r'\n',
+                            S(r'[ \t]*', RE_COMMENT, r'[ \t]*\n'),
+                            repeat='*?'),
+                         r'}', M(RE_STRUCT_ATTRIBUTE, SP, n='*')))
+RE_STRUCT_TYPEDEF = S(r'^[ \t]*typedef', SP, RE_FULL_STRUCT, SP,
+                      NAMED('name', RE_IDENTIFIER), r'\s*;[ \t]*\n')
+
+class FullStructTypedefMatch(TypedefMatch):
+    """typedef struct [SomeStruct] { ...} SomeType
+    Will be replaced by separate struct declaration + typedef
+    """
+    regexp = RE_STRUCT_TYPEDEF
+
+    def make_structname(self) -> str:
+        """Make struct name for struct+typedef split"""
+        name = self.group('structname')
+        if not name:
+            name = self.name
+        return name
+
+    def strip_typedef(self) -> Patch:
+        """generate patch that will strip typedef from the struct declartion
+
+        The caller is responsible for readding the typedef somewhere else.
+        """
+        name = self.make_structname()
+        body = self.group('body')
+        return self.make_patch(f'struct {name} {body};\n')
+
+    def make_simple_typedef(self) -> str:
+        structname = self.make_structname()
+        name = self.name
+        return f'typedef struct {structname} {name};\n'
+
+    def move_typedef(self, position) -> Iterator[Patch]:
+        """Generate patches to move typedef elsewhere"""
+        yield self.strip_typedef()
+        yield Patch(position, position, self.make_simple_typedef())
+
+    def split_typedef(self) -> Iterator[Patch]:
+        """Split into struct definition + typedef in-place"""
+        yield self.strip_typedef()
+        yield self.append(self.make_simple_typedef())
+
+class StructTypedefSplit(FullStructTypedefMatch):
+    """split struct+typedef declaration"""
+    def gen_patches(self) -> Iterator[Patch]:
+        if self.group('structname'):
+            yield from self.split_typedef()
+
+class DuplicatedTypedefs(SimpleTypedefMatch):
+    """Delete ALL duplicate typedefs (unsafe)"""
+    def gen_patches(self) -> Iterable[Patch]:
+        other_td = [td for td in chain(self.file.matches_of_type(SimpleTypedefMatch),
+                                       self.file.matches_of_type(FullStructTypedefMatch))
+                    if td.name == self.name]
+        DBG("other_td: %r", other_td)
+        if any(td.start() < self.start() for td in other_td):
+            # patch only if handling the first typedef
+            return
+        for td in other_td:
+            if isinstance(td, SimpleTypedefMatch):
+                DBG("other td: %r", td.match.groupdict())
+                if td.group('typedef_type') != self.group('typedef_type'):
+                    yield td.make_removal_patch()
+            elif isinstance(td, FullStructTypedefMatch):
+                DBG("other td: %r", td.match.groupdict())
+                if self.group('typedef_type') == 'struct '+td.group('structname'):
+                    yield td.strip_typedef()
+
+class QOMDuplicatedTypedefs(DuplicatedTypedefs):
+    """Delete duplicate typedefs if used by QOM type"""
+    def gen_patches(self) -> Iterable[Patch]:
+        qom_macros = [TypeCheckMacro, DeclareInstanceChecker, DeclareClassCheckers, DeclareObjCheckers]
+        qom_matches = chain(*(self.file.matches_of_type(t) for t in qom_macros))
+        in_use = any(RequiredIdentifier('type', self.name) in m.required_identifiers()
+                     for m in qom_matches)
+        if in_use:
+            yield from DuplicatedTypedefs.gen_patches(self)
+
+class QOMStructTypedefSplit(FullStructTypedefMatch):
+    """split struct+typedef declaration if used by QOM type"""
+    def gen_patches(self) -> Iterator[Patch]:
+        qom_macros = [TypeCheckMacro, DeclareInstanceChecker, DeclareClassCheckers, DeclareObjCheckers]
+        qom_matches = chain(*(self.file.matches_of_type(t) for t in qom_macros))
+        in_use = any(RequiredIdentifier('type', self.name) in m.required_identifiers()
+                     for m in qom_matches)
+        if in_use:
+            yield from self.split_typedef()
+
+def typedefs(file: FileInfo) -> Iterable[TypedefMatch]:
+    return (cast(TypedefMatch, m)
+            for m in chain(file.matches_of_type(SimpleTypedefMatch),
+                           file.matches_of_type(FullStructTypedefMatch)))
+
+def find_typedef(f: FileInfo, name: Optional[str]) -> Optional[TypedefMatch]:
+    if not name:
+        return None
+    for td in typedefs(f):
+        if td.name == name:
+            return td
+    return None
+
+CHECKER_MACROS = ['OBJECT_CHECK', 'OBJECT_CLASS_CHECK', 'OBJECT_GET_CLASS']
+CheckerMacroName = Literal['OBJECT_CHECK', 'OBJECT_CLASS_CHECK', 'OBJECT_GET_CLASS']
+
+RE_CHECK_MACRO = \
+    S(RE_MACRO_DEFINE,
+      OR(*CHECKER_MACROS, name='checker'),
+      M(r'\s*\(\s*', OR(NAMED('typedefname', RE_IDENTIFIER), RE_TYPE, name='c_type'), r'\s*,', CPP_SPACE,
+        OPTIONAL_PARS(RE_IDENTIFIER), r',', CPP_SPACE,
+        NAMED('qom_typename', RE_IDENTIFIER), r'\s*\)\n',
+        n='?', name='check_args'))
+
+EXPECTED_CHECKER_SUFFIXES: List[Tuple[CheckerMacroName, str]] = [
+    ('OBJECT_GET_CLASS', '_GET_CLASS'),
+    ('OBJECT_CLASS_CHECK', '_CLASS'),
+]
+
+class TypeCheckMacro(FileMatch):
+    """OBJECT_CHECK/OBJECT_CLASS_CHECK/OBJECT_GET_CLASS macro definitions
+    Will be replaced by DECLARE_*_CHECKERS macro
+    """
+    regexp = RE_CHECK_MACRO
+
+    @property
+    def checker(self) -> CheckerMacroName:
+        """Name of checker macro being used"""
+        return self.group('checker') # type: ignore
+
+    @property
+    def typedefname(self) -> Optional[str]:
+        return self.group('typedefname')
+
+    def find_typedef(self) -> Optional[TypedefMatch]:
+        return find_typedef(self.file, self.typedefname)
+
+    def sanity_check(self) -> None:
+        DBG("groups: %r", self.match.groups())
+        if not self.group('check_args'):
+            self.warn("type check macro not parsed completely: %s", self.name)
+            return
+        DBG("type identifiers: %r", self.type_identifiers)
+        if self.typedefname and self.find_typedef() is None:
+            self.warn("typedef used by %s not found", self.name)
+
+    def find_matching_macros(self) -> List['TypeCheckMacro']:
+        """Find other check macros that generate the same macro names
+
+        The returned list will always be sorted.
+        """
+        my_ids = self.type_identifiers
+        assert my_ids
+        return [m for m in self.file.matches_of_type(TypeCheckMacro)
+                if m.type_identifiers is not None
+                   and my_ids.uppercase is not None
+                   and (my_ids.uppercase == m.type_identifiers.uppercase
+                        or my_ids.typename == m.type_identifiers.typename)]
+
+    def merge_ids(self, matches: List['TypeCheckMacro']) -> Optional[TypeIdentifiers]:
+        """Try to merge info about type identifiers from all matches in a list"""
+        if not matches:
+            return None
+        r = matches[0].type_identifiers
+        if r is None:
+            return None
+        for m in matches[1:]:
+            assert m.type_identifiers
+            new = r.merge(m.type_identifiers)
+            if new is None:
+                self.warn("macro %s identifiers (%s) don't match macro %s (%s)",
+                          matches[0].name, r, m.name, m.type_identifiers)
+                return None
+            r = new
+        return r
+
+    def required_identifiers(self) -> Iterable[RequiredIdentifier]:
+        yield RequiredIdentifier('include', '"qom/object.h"')
+        if self.type_identifiers is None:
+            return
+        # to make sure typedefs will be moved above all related macros,
+        # return dependencies from all of them, not just this match
+        for m in self.find_matching_macros():
+            yield RequiredIdentifier('type', m.group('c_type'))
+            yield RequiredIdentifier('constant', m.group('qom_typename'))
+
+    @property
+    def type_identifiers(self) -> Optional[TypeIdentifiers]:
+        """Extract type identifier information from match"""
+        typename = self.group('qom_typename')
+        c_type = self.group('c_type')
+        if not typename or not c_type:
+            return None
+        typedef = self.group('typedefname')
+        classtype = None
+        instancetype = None
+        uppercase = None
+        expected_suffix = dict(EXPECTED_CHECKER_SUFFIXES).get(self.checker)
+
+        # here the available data depends on the checker macro being called:
+        # - we need to remove the suffix from the macro name
+        # - depending on the macro type, we know the class type name, or
+        #   the instance type name
+        if self.checker in ('OBJECT_GET_CLASS', 'OBJECT_CLASS_CHECK'):
+            classtype = c_type
+        elif self.checker == 'OBJECT_CHECK':
+            instancetype = c_type
+            uppercase = self.name
+        else:
+            assert False
+        if expected_suffix and self.name.endswith(expected_suffix):
+            uppercase = self.name[:-len(expected_suffix)]
+        return TypeIdentifiers(typename=typename, classtype=classtype,
+                               instancetype=instancetype, uppercase=uppercase)
+
+    def gen_patches(self) -> Iterable[Patch]:
+        # the implementation is a bit tricky because we need to group
+        # macros dealing with the same type into a single declaration
+        if self.type_identifiers is None:
+            self.warn("couldn't extract type information from macro %s", self.name)
+            return
+
+        if self.name == 'INTERFACE_CLASS':
+            # INTERFACE_CLASS is special and won't be patched
+            return
+
+        for checker,suffix in EXPECTED_CHECKER_SUFFIXES:
+            if self.name.endswith(suffix):
+                if self.checker != checker:
+                    self.warn("macro %s is using macro %s instead of %s", self.name, self.checker, checker)
+                    return
+                break
+
+        matches = self.find_matching_macros()
+        DBG("found %d matching macros: %s", len(matches), ' '.join(m.name for m in matches))
+        # we will generate patches only when processing the first macro:
+        if matches[0].start != self.start:
+            DBG("skipping %s (will patch when handling %s)", self.name, matches[0].name)
+            return
+
+
+        ids = self.merge_ids(matches)
+        if ids is None:
+            DBG("type identifier mismatch, won't patch %s", self.name)
+            return
+
+        if not ids.uppercase:
+            self.warn("macro %s doesn't follow the expected name pattern", self.name)
+            return
+        if not ids.typename:
+            self.warn("macro %s: couldn't extract type name", self.name)
+            return
+
+        #issues = ids.check_consistency()
+        #if issues:
+        #    for i in issues:
+        #        self.warn("inconsistent identifiers: %s", i)
+
+        names = [n for n in (ids.instancetype, ids.classtype, ids.uppercase, ids.typename)
+                 if n is not None]
+        if len(set(names)) != len(names):
+            self.warn("duplicate names used by macro: %r", ids)
+            return
+
+        assert ids.classtype or ids.instancetype
+        assert ids.typename
+        assert ids.uppercase
+        if ids.classtype and ids.instancetype:
+            new_decl = (f'DECLARE_OBJ_CHECKERS({ids.instancetype}, {ids.classtype},\n'
+                        f'                     {ids.uppercase}, {ids.typename})\n')
+        elif ids.classtype:
+            new_decl = (f'DECLARE_CLASS_CHECKERS({ids.classtype}, {ids.uppercase},\n'
+                        f'                       {ids.typename})\n')
+        elif ids.instancetype:
+            new_decl = (f'DECLARE_INSTANCE_CHECKER({ids.instancetype}, {ids.uppercase},\n'
+                        f'                         {ids.typename})\n')
+        else:
+            assert False
+
+        # we need to ensure the typedefs are already available
+        issues = []
+        for t in [ids.instancetype, ids.classtype]:
+            if not t:
+                continue
+            if re.fullmatch(RE_STRUCT_TYPE, t):
+                self.info("type %s is not a typedef", t)
+                continue
+            td = find_typedef(self.file, t)
+            #if not td and self.allfiles.find_file('include/qemu/typedefs.h'):
+            #
+            if not td:
+                # it is OK if the typedef is in typedefs.h
+                f = self.allfiles.find_file('include/qemu/typedefs.h')
+                if f and find_typedef(f, t):
+                    self.info("typedef %s found in typedefs.h", t)
+                    continue
+
+                issues.append("couldn't find typedef %s" % (t))
+            elif td.start() > self.start():
+                issues.append("typedef %s need to be moved earlier in the file" % (td.name))
+
+        for issue in issues:
+            self.warn(issue)
+
+        if issues and not self.file.force:
+            return
+
+        # delete all matching macros and add new declaration:
+        for m in matches:
+            yield m.make_patch('')
+        for issue in issues:
+            yield self.prepend("/* FIXME: %s */\n" % (issue))
+        yield self.append(new_decl)
+
+class InterfaceCheckMacro(FileMatch):
+    """Type checking macro using INTERFACE_CHECK
+    Will be replaced by DECLARE_INTERFACE_CHECKER
+    """
+    regexp = S(RE_MACRO_DEFINE,
+               'INTERFACE_CHECK',
+               r'\s*\(\s*', OR(NAMED('instancetype', RE_IDENTIFIER), RE_TYPE, name='c_type'),
+               r'\s*,', CPP_SPACE,
+               OPTIONAL_PARS(RE_IDENTIFIER), r',', CPP_SPACE,
+               NAMED('qom_typename', RE_IDENTIFIER), r'\s*\)\n')
+
+    def required_identifiers(self) -> Iterable[RequiredIdentifier]:
+        yield RequiredIdentifier('include', '"qom/object.h"')
+        yield RequiredIdentifier('type', self.group('instancetype'))
+        yield RequiredIdentifier('constant', self.group('qom_typename'))
+
+    def gen_patches(self) -> Iterable[Patch]:
+        if self.file.filename_matches('qom/object.h'):
+            self.debug("skipping object.h")
+            return
+
+        typename = self.group('qom_typename')
+        uppercase = self.name
+        instancetype = self.group('instancetype')
+        c = f"DECLARE_INTERFACE_CHECKER({instancetype}, {uppercase},\n"+\
+            f"                          {typename})\n"
+        yield self.make_patch(c)
+
+
+class TypeDeclaration(FileMatch):
+    """Parent class to all type declarations"""
+    @property
+    def instancetype(self) -> Optional[str]:
+        return self.getgroup('instancetype')
+
+    @property
+    def classtype(self) -> Optional[str]:
+        return self.getgroup('classtype')
+
+    @property
+    def typename(self) -> Optional[str]:
+        return self.getgroup('typename')
+
+class TypeCheckerDeclaration(TypeDeclaration):
+    """Parent class to all type checker declarations"""
+    @property
+    def typename(self) -> str:
+        return self.group('typename')
+
+    @property
+    def uppercase(self) -> str:
+        return self.group('uppercase')
+
+class DeclareInstanceChecker(TypeCheckerDeclaration):
+    """DECLARE_INSTANCE_CHECKER use"""
+    #TODO: replace lonely DECLARE_INSTANCE_CHECKER with DECLARE_OBJ_CHECKERS
+    #      if all types are found.
+    #      This will require looking up the correct class type in the TypeInfo
+    #      structs in another file
+    regexp = S(r'^[ \t]*DECLARE_INSTANCE_CHECKER\s*\(\s*',
+               NAMED('instancetype', RE_TYPE), r'\s*,\s*',
+               NAMED('uppercase', RE_IDENTIFIER), r'\s*,\s*',
+               OR(RE_IDENTIFIER, RE_STRING, RE_MACRO_CONCAT, RE_FUN_CALL, name='typename'), SP,
+               r'\)[ \t]*;?[ \t]*\n')
+
+    def required_identifiers(self) -> Iterable[RequiredIdentifier]:
+        yield RequiredIdentifier('include', '"qom/object.h"')
+        yield RequiredIdentifier('constant', self.group('typename'))
+        yield RequiredIdentifier('type', self.group('instancetype'))
+
+class DeclareInterfaceChecker(TypeCheckerDeclaration):
+    """DECLARE_INTERFACE_CHECKER use"""
+    regexp = S(r'^[ \t]*DECLARE_INTERFACE_CHECKER\s*\(\s*',
+               NAMED('instancetype', RE_TYPE), r'\s*,\s*',
+               NAMED('uppercase', RE_IDENTIFIER), r'\s*,\s*',
+               OR(RE_IDENTIFIER, RE_STRING, RE_MACRO_CONCAT, RE_FUN_CALL, name='typename'), SP,
+               r'\)[ \t]*;?[ \t]*\n')
+
+    def required_identifiers(self) -> Iterable[RequiredIdentifier]:
+        yield RequiredIdentifier('include', '"qom/object.h"')
+        yield RequiredIdentifier('constant', self.group('typename'))
+        yield RequiredIdentifier('type', self.group('instancetype'))
+
+class DeclareInstanceType(TypeDeclaration):
+    """DECLARE_INSTANCE_TYPE use"""
+    regexp = S(r'^[ \t]*DECLARE_INSTANCE_TYPE\s*\(\s*',
+               NAMED('uppercase', RE_IDENTIFIER), r'\s*,\s*',
+               NAMED('instancetype', RE_TYPE), SP,
+               r'\)[ \t]*;?[ \t]*\n')
+
+    def required_identifiers(self) -> Iterable[RequiredIdentifier]:
+        yield RequiredIdentifier('include', '"qom/object.h"')
+        yield RequiredIdentifier('type', self.group('instancetype'))
+
+class DeclareClassType(TypeDeclaration):
+    """DECLARE_CLASS_TYPE use"""
+    regexp = S(r'^[ \t]*DECLARE_CLASS_TYPE\s*\(\s*',
+               NAMED('uppercase', RE_IDENTIFIER), r'\s*,\s*',
+               NAMED('classtype', RE_TYPE), SP,
+               r'\)[ \t]*;?[ \t]*\n')
+
+    def required_identifiers(self) -> Iterable[RequiredIdentifier]:
+        yield RequiredIdentifier('include', '"qom/object.h"')
+        yield RequiredIdentifier('type', self.group('classtype'))
+
+
+
+class DeclareClassCheckers(TypeCheckerDeclaration):
+    """DECLARE_CLASS_CHECKER use"""
+    regexp = S(r'^[ \t]*DECLARE_CLASS_CHECKERS\s*\(\s*',
+               NAMED('classtype', RE_TYPE), r'\s*,\s*',
+               NAMED('uppercase', RE_IDENTIFIER), r'\s*,\s*',
+               OR(RE_IDENTIFIER, RE_STRING, RE_MACRO_CONCAT, RE_FUN_CALL, name='typename'), SP,
+               r'\)[ \t]*;?[ \t]*\n')
+
+    def required_identifiers(self) -> Iterable[RequiredIdentifier]:
+        yield RequiredIdentifier('include', '"qom/object.h"')
+        yield RequiredIdentifier('constant', self.group('typename'))
+        yield RequiredIdentifier('type', self.group('classtype'))
+
+class DeclareObjCheckers(TypeCheckerDeclaration):
+    """DECLARE_OBJ_CHECKERS use"""
+    #TODO: detect when OBJECT_DECLARE_SIMPLE_TYPE can be used
+    regexp = S(r'^[ \t]*DECLARE_OBJ_CHECKERS\s*\(\s*',
+               NAMED('instancetype', RE_TYPE), r'\s*,\s*',
+               NAMED('classtype', RE_TYPE), r'\s*,\s*',
+               NAMED('uppercase', RE_IDENTIFIER), r'\s*,\s*',
+               OR(RE_IDENTIFIER, RE_STRING, RE_MACRO_CONCAT, RE_FUN_CALL, name='typename'), SP,
+               r'\)[ \t]*;?[ \t]*\n')
+
+    def required_identifiers(self) -> Iterable[RequiredIdentifier]:
+        yield RequiredIdentifier('include', '"qom/object.h"')
+        yield RequiredIdentifier('constant', self.group('typename'))
+        yield RequiredIdentifier('type', self.group('classtype'))
+        yield RequiredIdentifier('type', self.group('instancetype'))
+
+class TypeDeclarationFixup(FileMatch):
+    """Common base class for code that will look at a set of type declarations"""
+    regexp = RE_FILE_BEGIN
+    def gen_patches(self) -> Iterable[Patch]:
+        if self.file.filename_matches('qom/object.h'):
+            self.debug("skipping object.h")
+            return
+
+        # group checkers by uppercase name:
+        decl_types: List[Type[TypeDeclaration]] = [DeclareInstanceChecker, DeclareInstanceType,
+                                                   DeclareClassCheckers, DeclareClassType,
+                                                   DeclareObjCheckers]
+        checker_dict: Dict[str, List[TypeDeclaration]] = {}
+        for t in decl_types:
+            for m in self.file.matches_of_type(t):
+                checker_dict.setdefault(m.group('uppercase'), []).append(m)
+        self.debug("checker_dict: %r", checker_dict)
+        for uppercase,checkers in checker_dict.items():
+            fields = ('instancetype', 'classtype', 'uppercase', 'typename')
+            fvalues = dict((field, set(getattr(m, field) for m in checkers
+                                       if getattr(m, field, None) is not None))
+                            for field in fields)
+            for field,values in fvalues.items():
+                if len(values) > 1:
+                    for c in checkers:
+                        c.warn("%s mismatch (%s)", field, ' '.join(values))
+                    return
+
+            field_dict = dict((f, v.pop() if v else None) for f,v in fvalues.items())
+            yield from self.gen_patches_for_type(uppercase, checkers, field_dict)
+
+    def find_conflicts(self, uppercase: str, checkers: List[TypeDeclaration]) -> bool:
+        """Look for conflicting declarations that would make it unsafe to add new ones"""
+        conflicting: List[FileMatch] = []
+        # conflicts in the same file:
+        conflicting.extend(chain(self.file.find_matches(DefineDirective, uppercase),
+                                 self.file.find_matches(DeclareInterfaceChecker, uppercase, 'uppercase'),
+                                 self.file.find_matches(DeclareClassType, uppercase, 'uppercase'),
+                                 self.file.find_matches(DeclareInstanceType, uppercase, 'uppercase')))
+
+        # conflicts in another file:
+        conflicting.extend(o for o in chain(self.allfiles.find_matches(DeclareInstanceChecker, uppercase, 'uppercase'),
+                                            self.allfiles.find_matches(DeclareClassCheckers, uppercase, 'uppercase'),
+                                            self.allfiles.find_matches(DeclareInterfaceChecker, uppercase, 'uppercase'),
+                                            self.allfiles.find_matches(DefineDirective, uppercase))
+                           if o is not None and o.file != self.file
+                               # if both are .c files, there's no conflict at all:
+                               and not (o.file.filename.suffix == '.c' and
+                                       self.file.filename.suffix == '.c'))
+
+        if conflicting:
+            for c in checkers:
+                c.warn("skipping due to conflicting %s macro", uppercase)
+            for o in conflicting:
+                if o is None:
+                    continue
+                o.warn("conflicting %s macro is here", uppercase)
+            return True
+
+        return False
+
+    def gen_patches_for_type(self, uppercase: str,
+                             checkers: List[TypeDeclaration],
+                             fields: Dict[str, Optional[str]]) -> Iterable[Patch]:
+        """Should be reimplemented by subclasses"""
+        return
+        yield
+
+class DeclareVoidTypes(TypeDeclarationFixup):
+    """Add DECLARE_*_TYPE(..., void) when there's no declared type"""
+    regexp = RE_FILE_BEGIN
+    def gen_patches_for_type(self, uppercase: str,
+                             checkers: List[TypeDeclaration],
+                             fields: Dict[str, Optional[str]]) -> Iterable[Patch]:
+        if self.find_conflicts(uppercase, checkers):
+            return
+
+        #_,last_checker = max((m.start(), m) for m in checkers)
+        _,first_checker = min((m.start(), m) for m in checkers)
+
+        if not any(m.instancetype for m in checkers):
+            yield first_checker.prepend(f'DECLARE_INSTANCE_TYPE({uppercase}, void)\n')
+        if not any(m.classtype for m in checkers):
+            yield first_checker.prepend(f'DECLARE_CLASS_TYPE({uppercase}, void)\n')
+
+        #if not all(len(v) == 1 for v in fvalues.values()):
+        #    return
+        #
+        #final_values = dict((field, values.pop())
+        #                    for field,values in fvalues.items())
+        #s = (f"DECLARE_OBJ_CHECKERS({final_values['instancetype']}, {final_values['classtype']},\n"+
+        #        f"                     {final_values['uppercase']}, {final_values['typename']})\n")
+        #for c in checkers:
+        #    yield c.make_removal_patch()
+        #yield last_checker.append(s)
+
+
+class AddDeclareTypeName(TypeDeclarationFixup):
+    """Add DECLARE_TYPE_NAME declarations if necessary"""
+    def gen_patches_for_type(self, uppercase: str,
+                             checkers: List[TypeDeclaration],
+                             fields: Dict[str, Optional[str]]) -> Iterable[Patch]:
+        typename = fields.get('typename')
+        if typename is None:
+            self.warn("typename unavailable")
+            return
+        if typename == f'TYPE_{uppercase}':
+            self.info("already using TYPE_%s as type name", uppercase)
+            return
+        if self.file.find_match(DeclareTypeName, uppercase, 'uppercase'):
+            self.info("type name for %s already declared", uppercase)
+            return
+        _,first_checker = min((m.start(), m) for m in checkers)
+        s = f'DECLARE_TYPE_NAME({uppercase}, {typename})\n'
+        yield first_checker.prepend(s)
+
+class TrivialClassStruct(FileMatch):
+    """Trivial class struct"""
+    regexp = S(r'^[ \t]*struct\s*', NAMED('name', RE_IDENTIFIER),
+               r'\s*{\s*', NAMED('parent_struct', RE_IDENTIFIER), r'\s*parent(_class)?\s*;\s*};\n')
+
+class DeclareTypeName(FileMatch):
+    """DECLARE_TYPE_NAME usage"""
+    regexp = S(r'^[ \t]*DECLARE_TYPE_NAME\s*\(',
+               NAMED('uppercase', RE_IDENTIFIER), r'\s*,\s*',
+               OR(RE_IDENTIFIER, RE_STRING, RE_MACRO_CONCAT, RE_FUN_CALL, name='typename'),
+               r'\s*\);?[ \t]*\n')
+
+class ObjectDeclareType(TypeCheckerDeclaration):
+    """OBJECT_DECLARE_TYPE usage
+    Will be replaced with OBJECT_DECLARE_SIMPLE_TYPE if possible
+    """
+    regexp = S(r'^[ \t]*OBJECT_DECLARE_TYPE\s*\(',
+               NAMED('instancetype', RE_TYPE), r'\s*,\s*',
+               NAMED('classtype', RE_TYPE), r'\s*,\s*',
+               NAMED('uppercase', RE_IDENTIFIER), SP,
+               r'\)[ \t]*;?[ \t]*\n')
+
+    def gen_patches(self):
+        DBG("groups: %r", self.match.groupdict())
+        trivial_struct = self.file.find_match(TrivialClassStruct, self.group('classtype'))
+        if trivial_struct:
+            d = self.match.groupdict().copy()
+            d['parent_struct'] = trivial_struct.group("parent_struct")
+            yield trivial_struct.make_removal_patch()
+            c = ("OBJECT_DECLARE_SIMPLE_TYPE(%(instancetype)s, %(lowercase)s,\n"
+                 "                           %(uppercase)s, %(parent_struct)s)\n" % d)
+            yield self.make_patch(c)
+
+class ObjectDeclareSimpleType(TypeCheckerDeclaration):
+    """OBJECT_DECLARE_SIMPLE_TYPE usage"""
+    regexp = S(r'^[ \t]*OBJECT_DECLARE_SIMPLE_TYPE\s*\(',
+               NAMED('instancetype', RE_TYPE), r'\s*,\s*',
+               NAMED('uppercase', RE_IDENTIFIER), SP,
+               r'\)[ \t]*;?[ \t]*\n')
+
+class OldStyleObjectDeclareSimpleType(TypeCheckerDeclaration):
+    """OBJECT_DECLARE_SIMPLE_TYPE usage (old API)"""
+    regexp = S(r'^[ \t]*OBJECT_DECLARE_SIMPLE_TYPE\s*\(',
+               NAMED('instancetype', RE_TYPE), r'\s*,\s*',
+               NAMED('lowercase', RE_IDENTIFIER), r'\s*,\s*',
+               NAMED('uppercase', RE_IDENTIFIER), r'\s*,\s*',
+               NAMED('parent_classtype', RE_TYPE), SP,
+               r'\)[ \t]*;?[ \t]*\n')
+
+    @property
+    def classtype(self) -> Optional[str]:
+        instancetype = self.instancetype
+        assert instancetype
+        return f"{instancetype}Class"
+
+def find_typename_uppercase(files: FileList, typename: str) -> Optional[str]:
+    """Try to find what's the right MODULE_OBJ_NAME for a given type name"""
+    decl = files.find_match(DeclareTypeName, name=typename, group='typename')
+    if decl:
+        return decl.group('uppercase')
+    if typename.startswith('TYPE_'):
+        return typename[len('TYPE_'):]
+    return None
+
+def find_type_checkers(files:FileList, name:str, group:str='uppercase') -> Iterable[TypeCheckerDeclaration]:
+    """Find usage of DECLARE*CHECKER macro"""
+    c: Type[TypeCheckerDeclaration]
+    for c in (DeclareInstanceChecker, DeclareClassCheckers, DeclareObjCheckers, ObjectDeclareType, ObjectDeclareSimpleType):
+        yield from files.find_matches(c, name=name, group=group)
+
+class Include(FileMatch):
+    """#include directive"""
+    regexp = RE_INCLUDE
+    def provided_identifiers(self) -> Iterable[RequiredIdentifier]:
+        yield RequiredIdentifier('include', self.group('includepath'))
+
+class InitialIncludes(FileMatch):
+    """Initial #include block"""
+    regexp = S(RE_FILE_BEGIN,
+               M(SP, RE_COMMENTS,
+                 r'^[ \t]*#[ \t]*ifndef[ \t]+', RE_IDENTIFIER, r'[ \t]*\n',
+                 n='?', name='ifndef_block'),
+               M(SP, RE_COMMENTS,
+                 OR(RE_INCLUDE, RE_SIMPLEDEFINE),
+                 n='*', name='includes'))
+
+class SymbolUserList(NamedTuple):
+    definitions: List[FileMatch]
+    users: List[FileMatch]
+
+class MoveSymbols(FileMatch):
+    """Handle missing symbols
+    - Move typedefs and defines when necessary
+    - Add missing #include lines when necessary
+    """
+    regexp = RE_FILE_BEGIN
+
+    def gen_patches(self) -> Iterator[Patch]:
+        if self.file.filename_matches('qom/object.h'):
+            self.debug("skipping object.h")
+            return
+
+        index: Dict[RequiredIdentifier, SymbolUserList] = {}
+        definition_classes = [SimpleTypedefMatch, FullStructTypedefMatch, ConstantDefine, Include]
+        user_classes = [TypeCheckMacro, DeclareObjCheckers, DeclareInstanceChecker, DeclareClassCheckers, InterfaceCheckMacro]
+
+        # first we scan for all symbol definitions and usage:
+        for dc in definition_classes:
+            defs = self.file.matches_of_type(dc)
+            for d in defs:
+                DBG("scanning %r", d)
+                for i in d.provided_identifiers():
+                    index.setdefault(i, SymbolUserList([], [])).definitions.append(d)
+        DBG("index: %r", list(index.keys()))
+        for uc in user_classes:
+            users = self.file.matches_of_type(uc)
+            for u in users:
+                for i in u.required_identifiers():
+                    index.setdefault(i, SymbolUserList([], [])).users.append(u)
+
+        # validate all symbols:
+        for i,ul in index.items():
+            if not ul.users:
+                # unused symbol
+                continue
+
+            # symbol not defined
+            if len(ul.definitions) == 0:
+                if i.type == 'include':
+                   includes, = self.file.matches_of_type(InitialIncludes)
+                   #FIXME: don't do this if we're already inside qom/object.h
+                   yield includes.append(f'#include {i.name}\n')
+                else:
+                    u.warn("definition of %s %s not found in file", i.type, i.name)
+                continue
+
+            # symbol defined twice:
+            if len(ul.definitions) > 1:
+                ul.definitions[1].warn("%s defined twice", i.name)
+                ul.definitions[0].warn("previously defined here")
+                continue
+
+            # symbol defined.  check if all users are after its definition:
+            assert len(ul.definitions) == 1
+            definition = ul.definitions[0]
+            DBG("handling repositioning of %r", definition)
+            earliest = min(ul.users, key=lambda u: u.start())
+            if earliest.start() > definition.start():
+                DBG("%r is OK", definition)
+                continue
+
+            DBG("%r needs to be moved", definition)
+            if isinstance(definition, SimpleTypedefMatch) \
+               or isinstance(definition, ConstantDefine):
+                # simple typedef or define can be moved directly:
+                yield definition.make_removal_patch()
+                yield earliest.prepend(definition.group(0))
+            elif isinstance(definition, FullStructTypedefMatch) \
+                 and definition.group('structname'):
+                # full struct typedef is more complex: we need to remove
+                # the typedef
+                yield from definition.move_typedef(earliest.start())
+            else:
+                definition.warn("definition of %s %s needs to be moved earlier in the file", i.type, i.name)
+                earliest.warn("definition of %s %s is used here", i.type, i.name)
+
+
+class EmptyPreprocessorConditional(FileMatch):
+    """Delete empty preprocessor conditionals"""
+    regexp = r'^[ \t]*#(if|ifdef)[ \t].*\n+[ \t]*#endif[ \t]*\n'
+    def gen_patches(self) -> Iterable[Patch]:
+        yield self.make_removal_patch()
diff --git a/scripts/codeconverter/codeconverter/qom_type_info.py b/scripts/codeconverter/codeconverter/qom_type_info.py
new file mode 100644
index 000000000..255cb5992
--- /dev/null
+++ b/scripts/codeconverter/codeconverter/qom_type_info.py
@@ -0,0 +1,969 @@
+# Copyright (C) 2020 Red Hat Inc.
+#
+# Authors:
+#  Eduardo Habkost <ehabkost@redhat.com>
+#
+# This work is licensed under the terms of the GNU GPL, version 2.  See
+# the COPYING file in the top-level directory.
+import re
+from .regexps import *
+from .patching import *
+from .utils import *
+from .qom_macros import *
+
+TI_FIELDS = [ 'name', 'parent', 'abstract', 'interfaces',
+    'instance_size', 'instance_init', 'instance_post_init', 'instance_finalize',
+    'class_size', 'class_init', 'class_base_init', 'class_data']
+
+RE_TI_FIELD_NAME = OR(*TI_FIELDS)
+
+RE_TI_FIELD_INIT = S(r'[ \t]*', NAMED('comments', RE_COMMENTS),
+                     r'\.', NAMED('field', RE_TI_FIELD_NAME), r'\s*=\s*',
+                     NAMED('value', RE_EXPRESSION), r'[ \t]*,?[ \t]*\n')
+RE_TI_FIELDS = M(RE_TI_FIELD_INIT)
+
+RE_TYPEINFO_START = S(r'^[ \t]*', M(r'(static|const)\s+', name='modifiers'), r'TypeInfo\s+',
+                      NAMED('name', RE_IDENTIFIER), r'\s*=\s*{[ \t]*\n')
+
+ParsedArray = List[str]
+ParsedInitializerValue = Union[str, ParsedArray]
+class InitializerValue(NamedTuple):
+    raw: str
+    parsed: Optional[ParsedInitializerValue]
+    match: Optional[Match]
+
+class ArrayItem(FileMatch):
+    regexp = RE_ARRAY_ITEM
+
+class ArrayInitializer(FileMatch):
+    regexp = RE_ARRAY
+
+    def parsed(self) -> ParsedArray:
+        #DBG('parse_array: %r', m.group(0))
+        return [m.group('arrayitem') for m in self.group_finditer(ArrayItem, 'arrayitems')]
+
+class FieldInitializer(FileMatch):
+    regexp = RE_TI_FIELD_INIT
+
+    @property
+    def raw(self) -> str:
+        return self.group('value')
+
+    @property
+    def parsed(self) -> ParsedInitializerValue:
+        parsed: ParsedInitializerValue = self.raw
+        #DBG("parse_initializer_value: %r", s)
+        array = self.try_group_match(ArrayInitializer, 'value')
+        if array:
+            assert isinstance(array, ArrayInitializer)
+            return array.parsed()
+        return parsed
+
+TypeInfoInitializers = Dict[str, FieldInitializer]
+
+class TypeDefinition(FileMatch):
+    """
+    Common base class for type definitions (TypeInfo variables or OBJECT_DEFINE* macros)
+    """
+    @property
+    def instancetype(self) -> Optional[str]:
+        return self.group('instancetype')
+
+    @property
+    def classtype(self) -> Optional[str]:
+        return self.group('classtype')
+
+    @property
+    def uppercase(self) -> Optional[str]:
+        return self.group('uppercase')
+
+    @property
+    def parent_uppercase(self) -> str:
+        return self.group('parent_uppercase')
+
+    @property
+    def initializers(self) -> Optional[TypeInfoInitializers]:
+        if getattr(self, '_inititalizers', None):
+            self._initializers: TypeInfoInitializers
+            return self._initializers
+        fields = self.group('fields')
+        if fields is None:
+            return None
+        d = dict((fm.group('field'), fm)
+                  for fm in self.group_finditer(FieldInitializer, 'fields'))
+        self._initializers = d # type: ignore
+        return self._initializers
+
+
+class TypeInfoVar(TypeDefinition):
+    """TypeInfo variable declaration with initializer"""
+    regexp = S(NAMED('begin', RE_TYPEINFO_START),
+               M(NAMED('fields', RE_TI_FIELDS),
+                 NAMED('endcomments', SP, RE_COMMENTS),
+                 NAMED('end', r'};?\n'),
+                 n='?', name='fullspec'))
+
+    def is_static(self) -> bool:
+        return 'static' in self.group('modifiers')
+
+    def is_const(self) -> bool:
+        return 'const' in self.group('modifiers')
+
+    def is_full(self) -> bool:
+        return bool(self.group('fullspec'))
+
+    def get_initializers(self) -> TypeInfoInitializers:
+        """Helper for code that needs to deal with missing initializer info"""
+        if self.initializers is None:
+            return {}
+        return self.initializers
+
+    def get_raw_initializer_value(self, field: str, default: str = '') -> str:
+        initializers = self.get_initializers()
+        if field in initializers:
+            return initializers[field].raw
+        else:
+            return default
+
+    @property
+    def typename(self) -> Optional[str]:
+        return self.get_raw_initializer_value('name')
+
+    @property
+    def uppercase(self) -> Optional[str]:
+        typename = self.typename
+        if not typename:
+            return None
+        if not typename.startswith('TYPE_'):
+            return None
+        return typename[len('TYPE_'):]
+
+    @property
+    def classtype(self) -> Optional[str]:
+        class_size = self.get_raw_initializer_value('class_size')
+        if not class_size:
+            return None
+        m = re.fullmatch(RE_SIZEOF, class_size)
+        if not m:
+            return None
+        return m.group('sizeoftype')
+
+    @property
+    def instancetype(self) -> Optional[str]:
+        instance_size = self.get_raw_initializer_value('instance_size')
+        if not instance_size:
+            return None
+        m = re.fullmatch(RE_SIZEOF, instance_size)
+        if not m:
+            return None
+        return m.group('sizeoftype')
+
+
+    #def extract_identifiers(self) -> Optional[TypeIdentifiers]:
+    #    """Try to extract identifiers from names being used"""
+    #    DBG("extracting idenfiers from %s", self.name)
+        #uppercase = None
+        #if typename and re.fullmatch(RE_IDENTIFIER, typename) and typename.startswith("TYPE_"):
+        #    uppercase = typename[len('TYPE_'):]
+        #lowercase = None
+        #funcs = set()
+        #prefixes = set()
+        #for field,suffix in [('instance_init', '_init'),
+        #                     ('instance_finalize', '_finalize'),
+        #                     ('class_init', '_class_init')]:
+        #    if field not in values:
+        #        continue
+        #    func = values[field].raw
+        #    funcs.add(func)
+        #    if func.endswith(suffix):
+        #        prefixes.add(func[:-len(suffix)])
+        #    else:
+        #        self.warn("function name %s doesn't have expected %s suffix",
+        #                  func, suffix)
+        #if len(prefixes) == 1:
+        #    lowercase = prefixes.pop()
+        #elif len(prefixes) > 1:
+        #    self.warn("inconsistent function names: %s", ' '.join(funcs))
+
+        #.parent = TYPE_##PARENT_MODULE_OBJ_NAME, \
+        #return TypeIdentifiers(typename=typename,
+        #                       uppercase=uppercase, lowercase=lowercase,
+        #                       instancetype=instancetype, classtype=classtype)
+
+    def append_field(self, field: str, value: str) -> Patch:
+        """Generate patch appending a field initializer"""
+        content = f'    .{field} = {value},\n'
+        fm = self.group_match('fields')
+        assert fm
+        return fm.append(content)
+
+    def patch_field(self, field: str, replacement: str) -> Patch:
+        """Generate patch replacing a field initializer"""
+        initializers = self.initializers
+        assert initializers
+        value = initializers.get(field)
+        assert value
+        return value.make_patch(replacement)
+
+    def remove_field(self, field: str) -> Iterable[Patch]:
+        initializers = self.initializers
+        assert initializers
+        if field in initializers:
+            yield self.patch_field(field, '')
+
+    def remove_fields(self, *fields: str) -> Iterable[Patch]:
+        for f in fields:
+            yield from self.remove_field(f)
+
+    def patch_field_value(self, field: str, replacement: str) -> Patch:
+        """Replace just the value of a field initializer"""
+        initializers = self.initializers
+        assert initializers
+        value = initializers.get(field)
+        assert value
+        vm = value.group_match('value')
+        assert vm
+        return vm.make_patch(replacement)
+
+
+class RemoveRedundantClassSize(TypeInfoVar):
+    """Remove class_size when using OBJECT_DECLARE_SIMPLE_TYPE"""
+    def gen_patches(self) -> Iterable[Patch]:
+        initializers = self.initializers
+        if initializers is None:
+            return
+        if 'class_size' not in initializers:
+            return
+
+        self.debug("Handling %s", self.name)
+        m = re.fullmatch(RE_SIZEOF, initializers['class_size'].raw)
+        if not m:
+            self.warn("%s class_size is not sizeof?", self.name)
+            return
+        classtype = m.group('sizeoftype')
+        if not classtype.endswith('Class'):
+            self.warn("%s class size type (%s) is not *Class?", self.name, classtype)
+            return
+        self.debug("classtype is %s", classtype)
+        instancetype = classtype[:-len('Class')]
+        self.debug("intanceypte is %s", instancetype)
+        self.debug("searching for simpletype declaration using %s as InstanceType", instancetype)
+        decl = self.allfiles.find_match(OldStyleObjectDeclareSimpleType,
+                                        instancetype, 'instancetype')
+        if not decl:
+            self.debug("No simpletype declaration found for %s", instancetype)
+            return
+        self.debug("Found simple type declaration")
+        decl.debug("declaration is here")
+        yield from self.remove_field('class_size')
+
+class RemoveDeclareSimpleTypeArg(OldStyleObjectDeclareSimpleType):
+    """Remove class_size when using OBJECT_DECLARE_SIMPLE_TYPE"""
+    def gen_patches(self) -> Iterable[Patch]:
+        c = (f'OBJECT_DECLARE_SIMPLE_TYPE({self.group("instancetype")}, {self.group("lowercase")},\n'
+             f'                           {self.group("uppercase")})\n')
+        yield self.make_patch(c)
+
+class UseDeclareTypeExtended(TypeInfoVar):
+    """Replace TypeInfo variable with OBJECT_DEFINE_TYPE_EXTENDED"""
+    def gen_patches(self) -> Iterable[Patch]:
+        # this will just ensure the caches for find_match() and matches_for_type()
+        # will be loaded in advance:
+        find_type_checkers(self.allfiles, 'xxxxxxxxxxxxxxxxx')
+
+        if not self.is_static():
+            self.info("Skipping non-static TypeInfo variable")
+            return
+
+        type_info_macro = self.file.find_match(TypeInfoMacro, self.name)
+        if not type_info_macro:
+            self.warn("TYPE_INFO(%s) line not found", self.name)
+            return
+
+        values = self.initializers
+        if values is None:
+            return
+        if 'name' not in values:
+            self.warn("name not set in TypeInfo variable %s", self.name)
+            return
+
+        typename = values['name'].raw
+
+        if 'parent' not in values:
+            self.warn("parent not set in TypeInfo variable %s", self.name)
+            return
+        parent_typename = values['parent'].raw
+
+        instancetype = None
+        if 'instance_size' in values:
+            m = re.fullmatch(RE_SIZEOF, values['instance_size'].raw)
+            if m:
+                instancetype = m.group('sizeoftype')
+            else:
+                self.warn("can't extract instance type in TypeInfo variable %s", self.name)
+                self.warn("instance_size is set to: %r", values['instance_size'].raw)
+                return
+
+        classtype = None
+        if 'class_size' in values:
+            m = re.fullmatch(RE_SIZEOF, values['class_size'].raw)
+            if m:
+                classtype = m.group('sizeoftype')
+            else:
+                self.warn("can't extract class type in TypeInfo variable %s", self.name)
+                self.warn("class_size is set to: %r", values['class_size'].raw)
+                return
+
+        #for t in (typename, parent_typename):
+        #    if not re.fullmatch(RE_IDENTIFIER, t):
+        #        self.info("type name is not a macro/constant")
+        #        if instancetype or classtype:
+        #            self.warn("macro/constant type name is required for instance/class type")
+        #        if not self.file.force:
+        #            return
+
+        # Now, the challenge is to find out the right MODULE_OBJ_NAME for the
+        # type and for the parent type
+        self.info("TypeInfo variable for %s is here", typename)
+        uppercase = find_typename_uppercase(self.allfiles, typename)
+        if not uppercase:
+            self.info("Can't find right uppercase name for %s", typename)
+            if instancetype or classtype:
+                self.warn("Can't find right uppercase name for %s", typename)
+                self.warn("This will make type validation difficult in the future")
+            return
+
+        parent_uppercase = find_typename_uppercase(self.allfiles, parent_typename)
+        if not parent_uppercase:
+            self.info("Can't find right uppercase name for parent type (%s)", parent_typename)
+            if instancetype or classtype:
+                self.warn("Can't find right uppercase name for parent type (%s)", parent_typename)
+                self.warn("This will make type validation difficult in the future")
+            return
+
+        ok = True
+
+        #checkers: List[TypeCheckerDeclaration] = list(find_type_checkers(self.allfiles, uppercase))
+        #for c in checkers:
+        #    c.info("instance type checker declaration (%s) is here", c.group('uppercase'))
+        #if not checkers:
+        #    self.info("No type checkers declared for %s", uppercase)
+        #    if instancetype or classtype:
+        #        self.warn("Can't find where type checkers for %s (%s) are declared.  We will need them to validate sizes of %s",
+        #                  typename, uppercase, self.name)
+
+        if not instancetype:
+            instancetype = 'void'
+        if not classtype:
+            classtype = 'void'
+
+        #checker_instancetypes = set(c.instancetype for c in checkers
+        #                            if c.instancetype is not None)
+        #if len(checker_instancetypes) > 1:
+        #    self.warn("ambiguous set of type checkers")
+        #    for c in checkers:
+        #        c.warn("instancetype is %s here", c.instancetype)
+        #    ok = False
+        #elif len(checker_instancetypes) == 1:
+        #    checker_instancetype = checker_instancetypes.pop()
+        #    DBG("checker instance type: %r", checker_instancetype)
+        #    if instancetype != checker_instancetype:
+        #        self.warn("type at instance_size is %r.  Should instance_size be set to sizeof(%s) ?",
+        #                instancetype, checker_instancetype)
+        #        ok = False
+        #else:
+        #    if instancetype != 'void':
+        #        self.warn("instance type checker for %s (%s) not found", typename, instancetype)
+        #        ok = False
+
+        #checker_classtypes = set(c.classtype for c in checkers
+        #                         if c.classtype is not None)
+        #if len(checker_classtypes) > 1:
+        #    self.warn("ambiguous set of type checkers")
+        #    for c in checkers:
+        #        c.warn("classtype is %s here", c.classtype)
+        #    ok = False
+        #elif len(checker_classtypes) == 1:
+        #    checker_classtype = checker_classtypes.pop()
+        #    DBG("checker class type: %r", checker_classtype)
+        #    if classtype != checker_classtype:
+        #        self.warn("type at class_size is %r.  Should class_size be set to sizeof(%s) ?",
+        #                classtype, checker_classtype)
+        #        ok = False
+        #else:
+        #    if classtype != 'void':
+        #        self.warn("class type checker for %s (%s) not found", typename, classtype)
+        #        ok = False
+
+        #if not ok:
+        #    for c in checkers:
+        #        c.warn("Type checker declaration for %s (%s) is here",
+        #                           typename, type(c).__name__)
+        #    return
+
+        #if parent_decl is None:
+        #    self.warn("Can't find where parent type %s is declared", parent_typename)
+
+        #yield self.prepend(f'DECLARE_TYPE_NAME({uppercase}, {typename})\n')
+        #if not instancetype:
+        #    yield self.prepend(f'DECLARE_INSTANCE_TYPE({uppercase}, void)\n')
+        #if not classtype:
+        #    yield self.prepend(f'DECLARE_CLASS_TYPE({uppercase}, void)\n')
+        self.info("%s can be patched!", self.name)
+        replaced_fields = ['name', 'parent', 'instance_size', 'class_size']
+        begin = self.group_match('begin')
+        newbegin =  f'OBJECT_DEFINE_TYPE_EXTENDED({self.name},\n'
+        newbegin += f'                            {instancetype}, {classtype},\n'
+        newbegin += f'                            {uppercase}, {parent_uppercase}'
+        if set(values.keys()) - set(replaced_fields):
+            newbegin += ',\n'
+        yield begin.make_patch(newbegin)
+        yield from self.remove_fields(*replaced_fields)
+        end = self.group_match('end')
+        yield end.make_patch(')\n')
+        yield type_info_macro.make_removal_patch()
+
+class ObjectDefineTypeExtended(TypeDefinition):
+    """OBJECT_DEFINE_TYPE_EXTENDED usage"""
+    regexp = S(r'^[ \t]*OBJECT_DEFINE_TYPE_EXTENDED\s*\(\s*',
+               NAMED('name', RE_IDENTIFIER), r'\s*,\s*',
+               NAMED('instancetype', RE_IDENTIFIER), r'\s*,\s*',
+               NAMED('classtype', RE_IDENTIFIER), r'\s*,\s*',
+               NAMED('uppercase', RE_IDENTIFIER), r'\s*,\s*',
+               NAMED('parent_uppercase', RE_IDENTIFIER),
+               M(r',\s*\n',
+                 NAMED('fields', RE_TI_FIELDS),
+                 n='?'),
+               r'\s*\);?\n?')
+
+class ObjectDefineType(TypeDefinition):
+    """OBJECT_DEFINE_TYPE usage"""
+    regexp = S(r'^[ \t]*OBJECT_DEFINE_TYPE\s*\(\s*',
+               NAMED('lowercase', RE_IDENTIFIER), r'\s*,\s*',
+               NAMED('uppercase', RE_IDENTIFIER), r'\s*,\s*',
+               NAMED('parent_uppercase', RE_IDENTIFIER),
+               M(r',\s*\n',
+                 NAMED('fields', RE_TI_FIELDS),
+                 n='?'),
+               r'\s*\);?\n?')
+
+def find_type_definitions(files: FileList, uppercase: str) -> Iterable[TypeDefinition]:
+    types: List[Type[TypeDefinition]] = [TypeInfoVar, ObjectDefineType, ObjectDefineTypeExtended]
+    for t in types:
+        for m in files.matches_of_type(t):
+            m.debug("uppercase: %s", m.uppercase)
+    yield from (m for t in types
+                  for m in files.matches_of_type(t)
+                if m.uppercase == uppercase)
+
+class AddDeclareVoidClassType(TypeDeclarationFixup):
+    """Will add DECLARE_CLASS_TYPE(..., void) if possible"""
+    def gen_patches_for_type(self, uppercase: str,
+                             checkers: List[TypeDeclaration],
+                             fields: Dict[str, Optional[str]]) -> Iterable[Patch]:
+        defs = list(find_type_definitions(self.allfiles, uppercase))
+        if len(defs) > 1:
+            self.warn("multiple definitions for %s", uppercase)
+            for d in defs:
+                d.warn("definition found here")
+            return
+        elif len(defs) == 0:
+            self.warn("type definition for %s not found", uppercase)
+            return
+        d = defs[0]
+        if d.classtype is None:
+            d.info("definition for %s has classtype, skipping", uppercase)
+            return
+        class_type_checkers = [c for c in checkers
+                               if c.classtype is not None]
+        if class_type_checkers:
+            for c in class_type_checkers:
+                c.warn("class type checker for %s is present here", uppercase)
+            return
+
+        _,last_checker = max((m.start(), m) for m in checkers)
+        s = f'DECLARE_CLASS_TYPE({uppercase}, void)\n'
+        yield last_checker.append(s)
+
+class AddDeclareVoidInstanceType(FileMatch):
+    """Will add DECLARE_INSTANCE_TYPE(..., void) if possible"""
+    regexp = S(r'^[ \t]*#[ \t]*define', CPP_SPACE,
+               NAMED('name', r'TYPE_[a-zA-Z0-9_]+\b'),
+               CPP_SPACE, r'.*\n')
+
+    def gen_patches(self) -> Iterable[Patch]:
+        assert self.name.startswith('TYPE_')
+        uppercase = self.name[len('TYPE_'):]
+        defs = list(find_type_definitions(self.allfiles, uppercase))
+        if len(defs) > 1:
+            self.warn("multiple definitions for %s", uppercase)
+            for d in defs:
+                d.warn("definition found here")
+            return
+        elif len(defs) == 0:
+            self.warn("type definition for %s not found", uppercase)
+            return
+        d = defs[0]
+        instancetype = d.instancetype
+        if instancetype is not None and instancetype != 'void':
+            return
+
+        instance_checkers = [c for c in find_type_checkers(self.allfiles, uppercase)
+                             if c.instancetype]
+        if instance_checkers:
+            d.warn("instance type checker for %s already declared", uppercase)
+            for c in instance_checkers:
+                c.warn("instance checker for %s is here", uppercase)
+            return
+
+        s = f'DECLARE_INSTANCE_TYPE({uppercase}, void)\n'
+        yield self.append(s)
+
+class AddObjectDeclareType(DeclareObjCheckers):
+    """Will add OBJECT_DECLARE_TYPE(...) if possible"""
+    def gen_patches(self) -> Iterable[Patch]:
+        uppercase = self.uppercase
+        typename = self.group('typename')
+        instancetype = self.group('instancetype')
+        classtype = self.group('classtype')
+
+        if typename != f'TYPE_{uppercase}':
+            self.warn("type name mismatch: %s vs %s", typename, uppercase)
+            return
+
+        typedefs = [(t,self.allfiles.find_matches(SimpleTypedefMatch, t))
+                    for t in (instancetype, classtype)]
+        for t,tds in typedefs:
+            if not tds:
+                self.warn("typedef %s not found", t)
+                return
+            for td in tds:
+                td_type = td.group('typedef_type')
+                if td_type != f'struct {t}':
+                    self.warn("typedef mismatch: %s is defined as %s", t, td_type)
+                    td.warn("typedef is here")
+                    return
+
+        # look for reuse of same struct type
+        other_instance_checkers = [c for c in find_type_checkers(self.allfiles, instancetype, 'instancetype')
+                                if c.uppercase != uppercase]
+        if other_instance_checkers:
+            self.warn("typedef %s is being reused", instancetype)
+            for ic in other_instance_checkers:
+                ic.warn("%s is reused here", instancetype)
+            if not self.file.force:
+                return
+
+        decl_types: List[Type[TypeDeclaration]] = [DeclareClassCheckers, DeclareObjCheckers]
+        class_decls = [m for t in decl_types
+                       for m in self.allfiles.find_matches(t, uppercase, 'uppercase')]
+
+        defs = list(find_type_definitions(self.allfiles, uppercase))
+        if len(defs) > 1:
+            self.warn("multiple definitions for %s", uppercase)
+            for d in defs:
+                d.warn("definition found here")
+            if not self.file.force:
+                return
+        elif len(defs) == 0:
+            self.warn("type definition for %s not found", uppercase)
+            if not self.file.force:
+                return
+        else:
+            d = defs[0]
+            if d.instancetype != instancetype:
+                self.warn("mismatching instance type for %s (%s)", uppercase, instancetype)
+                d.warn("instance type declared here (%s)", d.instancetype)
+                if not self.file.force:
+                    return
+            if d.classtype != classtype:
+                self.warn("mismatching class type for %s (%s)", uppercase, classtype)
+                d.warn("class type declared here (%s)", d.classtype)
+                if not self.file.force:
+                    return
+
+        assert self.file.original_content
+        for t,tds in typedefs:
+            assert tds
+            for td in tds:
+                if td.file is not self.file:
+                    continue
+
+                # delete typedefs that are truly redundant:
+                # 1) defined after DECLARE_OBJ_CHECKERS
+                if td.start() > self.start():
+                    yield td.make_removal_patch()
+                # 2) defined before DECLARE_OBJ_CHECKERS, but unused
+                elif not re.search(r'\b'+t+r'\b', self.file.original_content[td.end():self.start()]):
+                    yield td.make_removal_patch()
+
+        c = (f'OBJECT_DECLARE_TYPE({instancetype}, {classtype}, {uppercase})\n')
+        yield self.make_patch(c)
+
+class AddObjectDeclareSimpleType(DeclareInstanceChecker):
+    """Will add OBJECT_DECLARE_SIMPLE_TYPE(...) if possible"""
+    def gen_patches(self) -> Iterable[Patch]:
+        uppercase = self.uppercase
+        typename = self.group('typename')
+        instancetype = self.group('instancetype')
+
+        if typename != f'TYPE_{uppercase}':
+            self.warn("type name mismatch: %s vs %s", typename, uppercase)
+            return
+
+        typedefs = [(t,self.allfiles.find_matches(SimpleTypedefMatch, t))
+                    for t in (instancetype,)]
+        for t,tds in typedefs:
+            if not tds:
+                self.warn("typedef %s not found", t)
+                return
+            for td in tds:
+                td_type = td.group('typedef_type')
+                if td_type != f'struct {t}':
+                    self.warn("typedef mismatch: %s is defined as %s", t, td_type)
+                    td.warn("typedef is here")
+                    return
+
+        # look for reuse of same struct type
+        other_instance_checkers = [c for c in find_type_checkers(self.allfiles, instancetype, 'instancetype')
+                                if c.uppercase != uppercase]
+        if other_instance_checkers:
+            self.warn("typedef %s is being reused", instancetype)
+            for ic in other_instance_checkers:
+                ic.warn("%s is reused here", instancetype)
+            if not self.file.force:
+                return
+
+        decl_types: List[Type[TypeDeclaration]] = [DeclareClassCheckers, DeclareObjCheckers]
+        class_decls = [m for t in decl_types
+                       for m in self.allfiles.find_matches(t, uppercase, 'uppercase')]
+        if class_decls:
+            self.warn("class type declared for %s", uppercase)
+            for cd in class_decls:
+                cd.warn("class declaration found here")
+            return
+
+        defs = list(find_type_definitions(self.allfiles, uppercase))
+        if len(defs) > 1:
+            self.warn("multiple definitions for %s", uppercase)
+            for d in defs:
+                d.warn("definition found here")
+            if not self.file.force:
+                return
+        elif len(defs) == 0:
+            self.warn("type definition for %s not found", uppercase)
+            if not self.file.force:
+                return
+        else:
+            d = defs[0]
+            if d.instancetype != instancetype:
+                self.warn("mismatching instance type for %s (%s)", uppercase, instancetype)
+                d.warn("instance type declared here (%s)", d.instancetype)
+                if not self.file.force:
+                    return
+            if d.classtype:
+                self.warn("class type set for %s", uppercase)
+                d.warn("class type declared here")
+                if not self.file.force:
+                    return
+
+        assert self.file.original_content
+        for t,tds in typedefs:
+            assert tds
+            for td in tds:
+                if td.file is not self.file:
+                    continue
+
+                # delete typedefs that are truly redundant:
+                # 1) defined after DECLARE_OBJ_CHECKERS
+                if td.start() > self.start():
+                    yield td.make_removal_patch()
+                # 2) defined before DECLARE_OBJ_CHECKERS, but unused
+                elif not re.search(r'\b'+t+r'\b', self.file.original_content[td.end():self.start()]):
+                    yield td.make_removal_patch()
+
+        c = (f'OBJECT_DECLARE_SIMPLE_TYPE({instancetype}, {uppercase})\n')
+        yield self.make_patch(c)
+
+
+class TypeInfoStringName(TypeInfoVar):
+    """Replace hardcoded type names with TYPE_ constant"""
+    def gen_patches(self) -> Iterable[Patch]:
+        values = self.initializers
+        if values is None:
+            return
+        if 'name' not in values:
+            self.warn("name not set in TypeInfo variable %s", self.name)
+            return
+        typename = values['name'].raw
+        if re.fullmatch(RE_IDENTIFIER, typename):
+            return
+
+        self.warn("name %s is not an identifier", typename)
+        #all_defines = [m for m in self.allfiles.matches_of_type(ExpressionDefine)]
+        #self.debug("all_defines: %r", all_defines)
+        constants = [m for m in self.allfiles.matches_of_type(ExpressionDefine)
+                     if m.group('value').strip() == typename.strip()]
+        if not constants:
+            self.warn("No macro for %s found", typename)
+            return
+        if len(constants) > 1:
+            self.warn("I don't know which macro to use: %r", constants)
+            return
+        yield self.patch_field_value('name', constants[0].name)
+
+class RedundantTypeSizes(TypeInfoVar):
+    """Remove redundant instance_size/class_size from TypeInfo vars"""
+    def gen_patches(self) -> Iterable[Patch]:
+        values = self.initializers
+        if values is None:
+            return
+        if 'name' not in values:
+            self.warn("name not set in TypeInfo variable %s", self.name)
+            return
+        typename = values['name'].raw
+        if 'parent' not in values:
+            self.warn("parent not set in TypeInfo variable %s", self.name)
+            return
+        parent_typename = values['parent'].raw
+
+        if 'instance_size' not in values and 'class_size' not in values:
+            self.debug("no need to validate %s", self.name)
+            return
+
+        instance_decls = find_type_checkers(self.allfiles, typename)
+        if instance_decls:
+            self.debug("won't touch TypeInfo var that has type checkers")
+            return
+
+        parent = find_type_info(self.allfiles, parent_typename)
+        if not parent:
+            self.warn("Can't find TypeInfo for %s", parent_typename)
+            return
+
+        if 'instance_size' in values and parent.get_raw_initializer_value('instance_size') != values['instance_size'].raw:
+            self.info("instance_size mismatch")
+            parent.info("parent type declared here")
+            return
+
+        if 'class_size' in values and parent.get_raw_initializer_value('class_size') != values['class_size'].raw:
+            self.info("class_size mismatch")
+            parent.info("parent type declared here")
+            return
+
+        self.debug("will patch variable %s", self.name)
+
+        if 'instance_size' in values:
+            self.debug("deleting instance_size")
+            yield self.patch_field('instance_size', '')
+
+        if 'class_size' in values:
+            self.debug("deleting class_size")
+            yield self.patch_field('class_size', '')
+
+
+#class TypeInfoVarInitFuncs(TypeInfoVar):
+#    """TypeInfo variable
+#    Will create missing init functions
+#    """
+#    def gen_patches(self) -> Iterable[Patch]:
+#        values = self.initializers
+#        if values is None:
+#            self.warn("type not parsed completely: %s", self.name)
+#            return
+#
+#        macro = self.file.find_match(TypeInfoVar, self.name)
+#        if macro is None:
+#            self.warn("No TYPE_INFO macro for %s", self.name)
+#            return
+#
+#        ids = self.extract_identifiers()
+#        if ids is None:
+#            return
+#
+#        DBG("identifiers extracted: %r", ids)
+#        fields = set(values.keys())
+#        if ids.lowercase:
+#            if 'instance_init' not in fields:
+#                yield self.prepend(('static void %s_init(Object *obj)\n'
+#                                    '{\n'
+#                                    '}\n\n') % (ids.lowercase))
+#                yield self.append_field('instance_init', ids.lowercase+'_init')
+#
+#            if 'instance_finalize' not in fields:
+#                yield self.prepend(('static void %s_finalize(Object *obj)\n'
+#                                    '{\n'
+#                                    '}\n\n') % (ids.lowercase))
+#                yield self.append_field('instance_finalize', ids.lowercase+'_finalize')
+#
+#
+#            if 'class_init' not in fields:
+#                yield self.prepend(('static void %s_class_init(ObjectClass *oc, void *data)\n'
+#                                    '{\n'
+#                                    '}\n\n') % (ids.lowercase))
+#                yield self.append_field('class_init', ids.lowercase+'_class_init')
+
+class TypeInitMacro(FileMatch):
+    """Use of type_init(...) macro"""
+    regexp = S(r'^[ \t]*type_init\s*\(\s*', NAMED('name', RE_IDENTIFIER), r'\s*\);?[ \t]*\n')
+
+class DeleteEmptyTypeInitFunc(TypeInitMacro):
+    """Delete empty function declared using type_init(...)"""
+    def gen_patches(self) -> Iterable[Patch]:
+        fn = self.file.find_match(StaticVoidFunction, self.name)
+        DBG("function for %s: %s", self.name, fn)
+        if fn and fn.body == '':
+            yield fn.make_patch('')
+            yield self.make_patch('')
+
+class StaticVoidFunction(FileMatch):
+    """simple static void function
+    (no replacement rules)
+    """
+    #NOTE: just like RE_FULL_STRUCT, this doesn't parse any of the body contents
+    #      of the function.  Tt will just look for "}" in the beginning of a line
+    regexp = S(r'static\s+void\s+', NAMED('name', RE_IDENTIFIER), r'\s*\(\s*void\s*\)\n',
+               r'{\n',
+               NAMED('body',
+                     # acceptable inside the function body:
+                     # - lines starting with space or tab
+                     # - empty lines
+                     # - preprocessor directives
+                     OR(r'[ \t][^\n]*\n',
+                        r'#[^\n]*\n',
+                        r'\n',
+                        repeat='*')),
+               r'};?\n')
+
+    @property
+    def body(self) -> str:
+        return self.group('body')
+
+    def has_preprocessor_directive(self) -> bool:
+        return bool(re.search(r'^[ \t]*#', self.body, re.MULTILINE))
+
+def find_containing_func(m: FileMatch) -> Optional['StaticVoidFunction']:
+    """Return function containing this match"""
+    for fn in m.file.matches_of_type(StaticVoidFunction):
+        if fn.contains(m):
+            return fn
+    return None
+
+class TypeRegisterStaticCall(FileMatch):
+    """type_register_static() call
+    Will be replaced by TYPE_INFO() macro
+    """
+    regexp = S(r'^[ \t]*', NAMED('func_name', 'type_register_static'),
+               r'\s*\(&\s*', NAMED('name', RE_IDENTIFIER), r'\s*\);[ \t]*\n')
+
+class UseTypeInfo(TypeRegisterStaticCall):
+    """Replace type_register_static() call with TYPE_INFO declaration"""
+    def gen_patches(self) -> Iterable[Patch]:
+        fn = find_containing_func(self)
+        if fn:
+            DBG("%r is inside %r", self, fn)
+            type_init = self.file.find_match(TypeInitMacro, fn.name)
+            if type_init is None:
+                self.warn("can't find type_init(%s) line", fn.name)
+                if not self.file.force:
+                    return
+        else:
+            self.warn("can't identify the function where type_register_static(&%s) is called", self.name)
+            if not self.file.force:
+                return
+
+        #if fn.has_preprocessor_directive() and not self.file.force:
+        #    self.warn("function %s has preprocessor directives, this requires --force", fn.name)
+        #    return
+
+        var = self.file.find_match(TypeInfoVar, self.name)
+        if var is None:
+            self.warn("can't find TypeInfo var declaration for %s", self.name)
+            return
+
+        if not var.is_full():
+            self.warn("variable declaration %s wasn't parsed fully", var.name)
+            if not self.file.force:
+                return
+
+        if fn and fn.contains(var):
+            self.warn("TypeInfo %s variable is inside a function", self.name)
+            if not self.file.force:
+                return
+
+        # delete type_register_static() call:
+        yield self.make_patch('')
+        # append TYPE_REGISTER(...) after variable declaration:
+        yield var.append(f'TYPE_INFO({self.name})\n')
+
+class TypeRegisterCall(FileMatch):
+    """type_register_static() call"""
+    regexp = S(r'^[ \t]*', NAMED('func_name', 'type_register'),
+               r'\s*\(&\s*', NAMED('name', RE_IDENTIFIER), r'\s*\);[ \t]*\n')
+
+class MakeTypeRegisterStatic(TypeRegisterCall):
+    """Make type_register() call static if variable is static const"""
+    def gen_patches(self):
+        var = self.file.find_match(TypeInfoVar, self.name)
+        if var is None:
+            self.warn("can't find TypeInfo var declaration for %s", self.name)
+            return
+        if var.is_static() and var.is_const():
+            yield self.group_match('func_name').make_patch('type_register_static')
+
+class MakeTypeRegisterNotStatic(TypeRegisterStaticCall):
+    """Make type_register() call static if variable is static const"""
+    def gen_patches(self):
+        var = self.file.find_match(TypeInfoVar, self.name)
+        if var is None:
+            self.warn("can't find TypeInfo var declaration for %s", self.name)
+            return
+        if not var.is_static() or not var.is_const():
+            yield self.group_match('func_name').make_patch('type_register')
+
+class TypeInfoMacro(FileMatch):
+    """TYPE_INFO macro usage"""
+    regexp = S(r'^[ \t]*TYPE_INFO\s*\(\s*', NAMED('name', RE_IDENTIFIER), r'\s*\)[ \t]*;?[ \t]*\n')
+
+def find_type_info(files: RegexpScanner, name: str) -> Optional[TypeInfoVar]:
+    ti = [ti for ti in files.matches_of_type(TypeInfoVar)
+            if ti.get_raw_initializer_value('name') == name]
+    DBG("type info vars: %r", ti)
+    if len(ti) > 1:
+        DBG("multiple TypeInfo vars found for %s", name)
+        return None
+    if len(ti) == 0:
+        DBG("no TypeInfo var found for %s", name)
+        return None
+    return ti[0]
+
+class CreateClassStruct(DeclareInstanceChecker):
+    """Replace DECLARE_INSTANCE_CHECKER with OBJECT_DECLARE_SIMPLE_TYPE"""
+    def gen_patches(self) -> Iterable[Patch]:
+        typename = self.group('typename')
+        DBG("looking for TypeInfo variable for %s", typename)
+        var = find_type_info(self.allfiles, typename)
+        if var is None:
+            self.warn("no TypeInfo var found for %s", typename)
+            return
+        assert var.initializers
+        if 'class_size' in var.initializers:
+            self.warn("class size already set for TypeInfo %s", var.name)
+            return
+        classtype = self.group('instancetype')+'Class'
+        return
+        yield
+        #TODO: need to find out what's the parent class type...
+        #yield var.append_field('class_size', f'sizeof({classtype})')
+        #c = (f'OBJECT_DECLARE_SIMPLE_TYPE({instancetype}, {lowercase},\n'
+        #     f'                           MODULE_OBJ_NAME, ParentClassType)\n')
+        #yield self.make_patch(c)
+
+def type_infos(file: FileInfo) -> Iterable[TypeInfoVar]:
+    return file.matches_of_type(TypeInfoVar)
+
+def full_types(file: FileInfo) -> Iterable[TypeInfoVar]:
+    return [t for t in type_infos(file) if t.is_full()]
+
+def partial_types(file: FileInfo) -> Iterable[TypeInfoVar]:
+    return [t for t in type_infos(file) if not t.is_full()]
diff --git a/scripts/codeconverter/codeconverter/regexps.py b/scripts/codeconverter/codeconverter/regexps.py
new file mode 100644
index 000000000..77993cc3b
--- /dev/null
+++ b/scripts/codeconverter/codeconverter/regexps.py
@@ -0,0 +1,118 @@
+# Copyright (C) 2020 Red Hat Inc.
+#
+# Authors:
+#  Eduardo Habkost <ehabkost@redhat.com>
+#
+# This work is licensed under the terms of the GNU GPL, version 2.  See
+# the COPYING file in the top-level directory.
+"""Helpers for creation of regular expressions"""
+import re
+
+import logging
+logger = logging.getLogger(__name__)
+DBG = logger.debug
+INFO = logger.info
+WARN = logger.warning
+
+def S(*regexps) -> str:
+    """Just a shortcut to concatenate multiple regexps more easily"""
+    return ''.join(regexps)
+
+def P(*regexps, name=None, capture=False, repeat='') -> str:
+    """Just add parenthesis around regexp(s), with optional name or repeat suffix"""
+    s = S(*regexps)
+    if name:
+        return f'(?P<{name}>{s}){repeat}'
+    elif capture:
+        return f'({s}){repeat}'
+    else:
+        return f'(?:{s}){repeat}'
+
+def NAMED(name, *regexps) -> str:
+    """Make named group using <P<name>...) syntax
+
+    >>> NAMED('mygroup', 'xyz', 'abc')
+    '(?P<mygroup>xyzabc)'
+    """
+    return P(*regexps, name=name)
+
+def OR(*regexps, **kwargs) -> str:
+    """Build (a|b|c) regexp"""
+    return P('|'.join(regexps), **kwargs)
+
+def M(*regexps, n='*', name=None) -> str:
+    """Add repetition qualifier to regexp(s)
+
+    >>> M('a', 'b')
+    '(?:ab)*'
+    >>> M('a' , 'b', n='+')
+    '(?:ab)+'
+    >>> M('a' , 'b', n='{2,3}', name='name')
+    '(?P<name>(?:ab){2,3})'
+    """
+    r = P(*regexps, repeat=n)
+    if name:
+        r = NAMED(name, r)
+    return r
+
+# helper to make parenthesis optional around regexp
+OPTIONAL_PARS = lambda R: OR(S(r'\(\s*', R, r'\s*\)'), R)
+def test_optional_pars():
+    r = OPTIONAL_PARS('abc')+'$'
+    assert re.match(r, 'abc')
+    assert re.match(r, '(abc)')
+    assert not re.match(r, '(abcd)')
+    assert not re.match(r, '(abc')
+    assert not re.match(r, 'abc)')
+
+
+# this disables the MULTILINE flag, so it will match at the
+# beginning of the file:
+RE_FILE_BEGIN = r'(?-m:^)'
+
+# C primitives:
+
+SP = r'\s*'
+
+RE_COMMENT = r'//[^\n]*$|/\*([^*]|\*[^/])*\*/'
+RE_COMMENTS = M(RE_COMMENT + SP)
+
+RE_IDENTIFIER = r'[a-zA-Z_][a-zA-Z0-9_]*(?![a-zA-Z0-9])'
+RE_STRING = r'\"([^\"\\]|\\[a-z\"])*\"'
+RE_NUMBER = r'[0-9]+|0x[0-9a-fA-F]+'
+
+# space or escaped newlines:
+CPP_SPACE = OR(r'\s', r'\\\n', repeat='+')
+
+RE_PATH = '[a-zA-Z0-9/_.-]+'
+
+RE_INCLUDEPATH = OR(S(r'\"', RE_PATH, r'\"'),
+                    S(r'<', RE_PATH, r'>'))
+
+RE_INCLUDE = S(r'^[ \t]*#[ \t]*include[ \t]+', NAMED('includepath', RE_INCLUDEPATH), r'[ \t]*\n')
+RE_SIMPLEDEFINE = S(r'^[ \t]*#[ \t]*define[ \t]+', RE_IDENTIFIER, r'[ \t]*\n')
+
+RE_STRUCT_TYPE = S(r'struct\s+', RE_IDENTIFIER)
+RE_TYPE = OR(RE_IDENTIFIER, RE_STRUCT_TYPE)
+
+RE_MACRO_CONCAT = M(S(OR(RE_IDENTIFIER, RE_STRING), SP), n='{2,}')
+
+RE_SIMPLE_VALUE = OR(RE_IDENTIFIER, RE_STRING, RE_NUMBER)
+
+RE_FUN_CALL = S(RE_IDENTIFIER, r'\s*\(\s*', RE_SIMPLE_VALUE, r'\s*\)')
+RE_SIZEOF = S(r'sizeof\s*\(\s*', NAMED('sizeoftype', RE_TYPE), r'\s*\)')
+
+RE_ADDRESS = S(r'&\s*', RE_IDENTIFIER)
+
+RE_ARRAY_ITEM = S(r'{\s*', NAMED('arrayitem', M(RE_SIMPLE_VALUE, n='?')), r'\s*}\s*,?')
+RE_ARRAY_CAST = S(r'\(\s*', RE_IDENTIFIER, r'\s*\[\s*\]\)')
+RE_ARRAY_ITEMS = M(S(RE_ARRAY_ITEM, SP))
+RE_ARRAY = S(M(RE_ARRAY_CAST, n='?'), r'\s*{\s*',
+             NAMED('arrayitems', RE_ARRAY_ITEMS),
+             r'}')
+
+# NOTE: this covers a very small subset of valid expressions
+
+RE_EXPRESSION = OR(RE_SIZEOF, RE_FUN_CALL, RE_MACRO_CONCAT, RE_SIMPLE_VALUE,
+                   RE_ARRAY, RE_ADDRESS)
+
diff --git a/scripts/codeconverter/codeconverter/test_patching.py b/scripts/codeconverter/codeconverter/test_patching.py
new file mode 100644
index 000000000..71dfbd47e
--- /dev/null
+++ b/scripts/codeconverter/codeconverter/test_patching.py
@@ -0,0 +1,104 @@
+# Copyright (C) 2020 Red Hat Inc.
+#
+# Authors:
+#  Eduardo Habkost <ehabkost@redhat.com>
+#
+# This work is licensed under the terms of the GNU GPL, version 2.  See
+# the COPYING file in the top-level directory.
+from tempfile import NamedTemporaryFile
+from .patching import FileInfo, FileMatch, Patch, FileList
+from .regexps import *
+
+class BasicPattern(FileMatch):
+    regexp = '[abc]{3}'
+
+    @property
+    def name(self):
+        return self.group(0)
+
+    def replacement(self) -> str:
+        # replace match with the middle character repeated 5 times
+        return self.group(0)[1].upper()*5
+
+def test_pattern_patching():
+    of = NamedTemporaryFile('wt')
+    of.writelines(['one line\n',
+                  'this pattern will be patched: defbbahij\n',
+                  'third line\n',
+                  'another pattern: jihaabfed'])
+    of.flush()
+
+    files = FileList()
+    f = FileInfo(files, of.name)
+    f.load()
+    matches = f.matches_of_type(BasicPattern)
+    assert len(matches) == 2
+    p2 = matches[1]
+
+    # manually add patch, to see if .append() works:
+    f.patches.append(p2.append('XXX'))
+
+    # apply all patches:
+    f.gen_patches(matches)
+    patched = f.get_patched_content()
+    assert patched == ('one line\n'+
+                       'this pattern will be patched: defBBBBBhij\n'+
+                       'third line\n'+
+                       'another pattern: jihAAAAAXXXfed')
+
+class Function(FileMatch):
+    regexp = S(r'BEGIN\s+', NAMED('name', RE_IDENTIFIER), r'\n',
+               r'(.*\n)*?END\n')
+
+class Statement(FileMatch):
+    regexp = S(r'^\s*', NAMED('name', RE_IDENTIFIER), r'\(\)\n')
+
+def test_container_match():
+    of = NamedTemporaryFile('wt')
+    of.writelines(['statement1()\n',
+                   'statement2()\n',
+                   'BEGIN function1\n',
+                   '  statement3()\n',
+                   '  statement4()\n',
+                   'END\n',
+                   'BEGIN function2\n',
+                   '  statement5()\n',
+                   '  statement6()\n',
+                   'END\n',
+                   'statement7()\n'])
+    of.flush()
+
+    files = FileList()
+    f = FileInfo(files, of.name)
+    f.load()
+    assert len(f.matches_of_type(Function)) == 2
+    print(' '.join(m.name for m in f.matches_of_type(Statement)))
+    assert len(f.matches_of_type(Statement)) == 7
+
+    f1 = f.find_match(Function, 'function1')
+    f2 = f.find_match(Function, 'function2')
+    st1 = f.find_match(Statement, 'statement1')
+    st2 = f.find_match(Statement, 'statement2')
+    st3 = f.find_match(Statement, 'statement3')
+    st4 = f.find_match(Statement, 'statement4')
+    st5 = f.find_match(Statement, 'statement5')
+    st6 = f.find_match(Statement, 'statement6')
+    st7 = f.find_match(Statement, 'statement7')
+
+    assert not f1.contains(st1)
+    assert not f1.contains(st2)
+    assert not f1.contains(st2)
+    assert f1.contains(st3)
+    assert f1.contains(st4)
+    assert not f1.contains(st5)
+    assert not f1.contains(st6)
+    assert not f1.contains(st7)
+
+    assert not f2.contains(st1)
+    assert not f2.contains(st2)
+    assert not f2.contains(st2)
+    assert not f2.contains(st3)
+    assert not f2.contains(st4)
+    assert f2.contains(st5)
+    assert f2.contains(st6)
+    assert not f2.contains(st7)
diff --git a/scripts/codeconverter/codeconverter/test_regexps.py b/scripts/codeconverter/codeconverter/test_regexps.py
new file mode 100644
index 000000000..a445634d8
--- /dev/null
+++ b/scripts/codeconverter/codeconverter/test_regexps.py
@@ -0,0 +1,282 @@
+# Copyright (C) 2020 Red Hat Inc.
+#
+# Authors:
+#  Eduardo Habkost <ehabkost@redhat.com>
+#
+# This work is licensed under the terms of the GNU GPL, version 2.  See
+# the COPYING file in the top-level directory.
+from .regexps import *
+from .qom_macros import *
+from .qom_type_info import *
+
+def test_res() -> None:
+    def fullmatch(regexp, s):
+        return re.fullmatch(regexp, s, re.MULTILINE)
+
+    assert fullmatch(RE_IDENTIFIER, 'sizeof')
+    assert fullmatch(RE_IDENTIFIER, 'X86CPU')
+    assert fullmatch(RE_FUN_CALL, 'sizeof(X86CPU)')
+    assert fullmatch(RE_IDENTIFIER, 'X86_CPU_TYPE_NAME')
+    assert fullmatch(RE_SIMPLE_VALUE, '"base"')
+    print(RE_FUN_CALL)
+    assert fullmatch(RE_FUN_CALL, 'X86_CPU_TYPE_NAME("base")')
+    print(RE_TI_FIELD_INIT)
+    assert fullmatch(RE_TI_FIELD_INIT, '.name = X86_CPU_TYPE_NAME("base"),\n')
+
+
+    assert fullmatch(RE_MACRO_CONCAT, 'TYPE_ASPEED_GPIO "-ast2600"')
+    assert fullmatch(RE_EXPRESSION, 'TYPE_ASPEED_GPIO "-ast2600"')
+
+    print(RE_MACRO_DEFINE)
+    assert re.search(RE_MACRO_DEFINE, r'''
+    #define OFFSET_CHECK(c)                     \
+    do {                                        \
+        if (!(c)) {                             \
+            goto bad_offset;                    \
+        }                                       \
+    } while (0)
+    ''', re.MULTILINE)
+
+    print(RE_CHECK_MACRO)
+    print(CPP_SPACE)
+    assert not re.match(RE_CHECK_MACRO, r'''
+    #define OFFSET_CHECK(c)                     \
+    do {                                        \
+        if (!(c)) {                             \
+            goto bad_offset;                    \
+        }                                       \
+    } while (0)''', re.MULTILINE)
+
+    print(RE_CHECK_MACRO)
+    assert fullmatch(RE_CHECK_MACRO, r'''#define PCI_DEVICE(obj) \
+                     OBJECT_CHECK(PCIDevice, (obj), TYPE_PCI_DEVICE)
+''')
+    assert fullmatch(RE_CHECK_MACRO, r'''#define COLLIE_MACHINE(obj) \
+                     OBJECT_CHECK(CollieMachineState, obj, TYPE_COLLIE_MACHINE)
+''')
+
+    print(RE_TYPEINFO_START)
+    assert re.search(RE_TYPEINFO_START, r'''
+    cc->open = qmp_chardev_open_file;
+}
+
+static const TypeInfo char_file_type_info = {
+    .name = TYPE_CHARDEV_FILE,
+#ifdef _WIN32
+    .parent = TYPE_CHARDEV_WIN,
+''', re.MULTILINE)
+    assert re.search(RE_TYPEINFO_START, r'''
+        TypeInfo ti = {
+            .name = armsse_variants[i].name,
+            .parent = TYPE_ARMSSE,
+            .class_init = armsse_class_init,
+            .class_data = (void *)&armsse_variants[i],
+        };''', re.MULTILINE)
+
+    print(RE_ARRAY_ITEM)
+    assert fullmatch(RE_ARRAY_ITEM, '{ TYPE_HOTPLUG_HANDLER },')
+    assert fullmatch(RE_ARRAY_ITEM, '{ TYPE_ACPI_DEVICE_IF },')
+    assert fullmatch(RE_ARRAY_ITEM, '{ }')
+    assert fullmatch(RE_ARRAY_CAST, '(InterfaceInfo[])')
+    assert fullmatch(RE_ARRAY, '''(InterfaceInfo[]) {
+            { TYPE_HOTPLUG_HANDLER },
+            { TYPE_ACPI_DEVICE_IF },
+            { }
+    }''')
+    print(RE_COMMENT)
+    assert fullmatch(RE_COMMENT, r'''/* multi-line
+                                      * comment
+                                      */''')
+
+    print(RE_TI_FIELDS)
+    assert fullmatch(RE_TI_FIELDS,
+    r'''/* could be TYPE_SYS_BUS_DEVICE (or LPC etc) */
+        .parent = TYPE_DEVICE,
+''')
+    assert fullmatch(RE_TI_FIELDS, r'''.name = TYPE_TPM_CRB,
+        /* could be TYPE_SYS_BUS_DEVICE (or LPC etc) */
+        .parent = TYPE_DEVICE,
+        .instance_size = sizeof(CRBState),
+        .class_init  = tpm_crb_class_init,
+        .interfaces = (InterfaceInfo[]) {
+            { TYPE_TPM_IF },
+            { }
+        }
+''')
+    assert fullmatch(RE_TI_FIELDS + SP + RE_COMMENTS,
+        r'''.name = TYPE_PALM_MISC_GPIO,
+            .parent = TYPE_SYS_BUS_DEVICE,
+            .instance_size = sizeof(PalmMiscGPIOState),
+            .instance_init = palm_misc_gpio_init,
+            /*
+             * No class init required: device has no internal state so does not
+             * need to set up reset or vmstate, and has no realize method.
+             */''')
+
+    print(TypeInfoVar.regexp)
+    test_empty = 'static const TypeInfo x86_base_cpu_type_info = {\n'+\
+                 '};\n';
+    assert fullmatch(TypeInfoVar.regexp, test_empty)
+
+    test_simple = r'''
+    static const TypeInfo x86_base_cpu_type_info = {
+        .name = X86_CPU_TYPE_NAME("base"),
+        .parent = TYPE_X86_CPU,
+        .class_init = x86_cpu_base_class_init,
+    };
+    '''
+    assert re.search(TypeInfoVar.regexp, test_simple, re.MULTILINE)
+
+    test_interfaces = r'''
+    static const TypeInfo acpi_ged_info = {
+        .name          = TYPE_ACPI_GED,
+        .parent        = TYPE_SYS_BUS_DEVICE,
+        .instance_size = sizeof(AcpiGedState),
+        .instance_init  = acpi_ged_initfn,
+        .class_init    = acpi_ged_class_init,
+        .interfaces = (InterfaceInfo[]) {
+            { TYPE_HOTPLUG_HANDLER },
+            { TYPE_ACPI_DEVICE_IF },
+            { }
+        }
+    };
+    '''
+    assert re.search(TypeInfoVar.regexp, test_interfaces, re.MULTILINE)
+
+    test_comments = r'''
+    static const TypeInfo palm_misc_gpio_info = {
+        .name = TYPE_PALM_MISC_GPIO,
+        .parent = TYPE_SYS_BUS_DEVICE,
+        .instance_size = sizeof(PalmMiscGPIOState),
+        .instance_init = palm_misc_gpio_init,
+        /*
+         * No class init required: device has no internal state so does not
+         * need to set up reset or vmstate, and has no realize method.
+         */
+    };
+    '''
+    assert re.search(TypeInfoVar.regexp, test_comments, re.MULTILINE)
+
+    test_comments = r'''
+    static const TypeInfo tpm_crb_info = {
+        .name = TYPE_TPM_CRB,
+        /* could be TYPE_SYS_BUS_DEVICE (or LPC etc) */
+        .parent = TYPE_DEVICE,
+        .instance_size = sizeof(CRBState),
+        .class_init  = tpm_crb_class_init,
+        .interfaces = (InterfaceInfo[]) {
+            { TYPE_TPM_IF },
+            { }
+        }
+    };
+    '''
+    assert re.search(TypeInfoVar.regexp, test_comments, re.MULTILINE)
+
+def test_struct_re():
+    print('---')
+    print(RE_STRUCT_TYPEDEF)
+    assert re.search(RE_STRUCT_TYPEDEF, r'''
+typedef struct TCGState {
+    AccelState parent_obj;
+
+    bool mttcg_enabled;
+    unsigned long tb_size;
+} TCGState;
+''', re.MULTILINE)
+
+    assert re.search(RE_STRUCT_TYPEDEF, r'''
+typedef struct {
+    ISADevice parent_obj;
+
+    QEMUSoundCard card;
+    uint32_t freq;
+    uint32_t port;
+    int ticking[2];
+    int enabled;
+    int active;
+    int bufpos;
+#ifdef DEBUG
+    int64_t exp[2];
+#endif
+    int16_t *mixbuf;
+    uint64_t dexp[2];
+    SWVoiceOut *voice;
+    int left, pos, samples;
+    QEMUAudioTimeStamp ats;
+    FM_OPL *opl;
+    PortioList port_list;
+} AdlibState;
+''', re.MULTILINE)
+
+    false_positive = r'''
+typedef struct dma_pagetable_entry {
+    int32_t frame;
+    int32_t owner;
+} A B C D E;
+struct foo {
+    int x;
+} some_variable;
+'''
+    assert not re.search(RE_STRUCT_TYPEDEF, false_positive, re.MULTILINE)
+
+def test_initial_includes():
+    print(InitialIncludes.regexp)
+    c = '''
+#ifndef HW_FLASH_H
+#define HW_FLASH_H
+
+/* NOR flash devices */
+
+#include "qom/object.h"
+#include "exec/hwaddr.h"
+
+/* pflash_cfi01.c */
+'''
+    print(repr(list(m.groupdict() for m in InitialIncludes.finditer(c))))
+    m = InitialIncludes.domatch(c)
+    assert m
+    print(repr(m.group(0)))
+    assert m.group(0).endswith('#include "exec/hwaddr.h"\n')
+
+    c = '''#ifndef QEMU_VIRTIO_9P_H
+#define QEMU_VIRTIO_9P_H
+
+#include "standard-headers/linux/virtio_9p.h"
+#include "hw/virtio/virtio.h"
+#include "9p.h"
+
+
+'''
+    print(repr(list(m.groupdict() for m in InitialIncludes.finditer(c))))
+    m = InitialIncludes.domatch(c)
+    assert m
+    print(repr(m.group(0)))
+    assert m.group(0).endswith('#include "9p.h"\n')
+
+    c = '''#include "qom/object.h"
+/*
+ * QEMU ES1370 emulation
+...
+ */
+
+/* #define DEBUG_ES1370 */
+/* #define VERBOSE_ES1370 */
+#define SILENT_ES1370
+
+#include "qemu/osdep.h"
+#include "hw/audio/soundhw.h"
+#include "audio/audio.h"
+#include "hw/pci/pci.h"
+#include "migration/vmstate.h"
+#include "qemu/module.h"
+#include "sysemu/dma.h"
+
+/* Missing stuff:
+   SCTRL_P[12](END|ST)INC
+'''
+    print(repr(list(m.groupdict() for m in InitialIncludes.finditer(c))))
+    m = InitialIncludes.domatch(c)
+    assert m
+    print(repr(m.group(0)))
+    assert m.group(0).endswith('#include "sysemu/dma.h"\n')
+
diff --git a/scripts/codeconverter/codeconverter/utils.py b/scripts/codeconverter/codeconverter/utils.py
new file mode 100644
index 000000000..760ab7eec
--- /dev/null
+++ b/scripts/codeconverter/codeconverter/utils.py
@@ -0,0 +1,72 @@
+# Copyright (C) 2020 Red Hat Inc.
+#
+# Authors:
+#  Eduardo Habkost <ehabkost@redhat.com>
+#
+# This work is licensed under the terms of the GNU GPL, version 2.  See
+# the COPYING file in the top-level directory.
+from typing import *
+
+import logging
+logger = logging.getLogger(__name__)
+DBG = logger.debug
+INFO = logger.info
+WARN = logger.warning
+
+T = TypeVar('T')
+def opt_compare(a: T, b: T) -> bool:
+    """Compare two values, ignoring mismatches if one of them is None"""
+    return (a is None) or (b is None) or (a == b)
+
+def merge(a: T, b: T) -> T:
+    """Merge two values if they matched using opt_compare()"""
+    assert opt_compare(a, b)
+    if a is None:
+        return b
+    else:
+        return a
+
+def test_comp_merge():
+    assert opt_compare(None, 1) == True
+    assert opt_compare(2, None) == True
+    assert opt_compare(1, 1) == True
+    assert opt_compare(1, 2) == False
+
+    assert merge(None, None) is None
+    assert merge(None, 10) == 10
+    assert merge(10, None) == 10
+    assert merge(10, 10) == 10
+
+
+LineNumber = NewType('LineNumber', int)
+ColumnNumber = NewType('ColumnNumber', int)
+class LineAndColumn(NamedTuple):
+    line: int
+    col: int
+
+    def __str__(self):
+        return '%d:%d' % (self.line, self.col)
+
+def line_col(s, position: int) -> LineAndColumn:
+    """Return line and column for a char position in string
+
+    Character position starts in 0, but lines and columns start in 1.
+    """
+    before = s[:position]
+    lines = before.split('\n')
+    line = len(lines)
+    col = len(lines[-1]) + 1
+    return LineAndColumn(line, col)
+
+def test_line_col():
+    assert line_col('abc\ndefg\nhijkl', 0) == (1, 1)
+    assert line_col('abc\ndefg\nhijkl', 2) == (1, 3)
+    assert line_col('abc\ndefg\nhijkl', 3) == (1, 4)
+    assert line_col('abc\ndefg\nhijkl', 4) == (2, 1)
+    assert line_col('abc\ndefg\nhijkl', 10) == (3, 2)
+
+def not_optional(arg: Optional[T]) -> T:
+    assert arg is not None
+    return arg
+
+__all__ = ['not_optional', 'opt_compare', 'merge', 'line_col', 'LineAndColumn']
+\ No newline at end of file
diff --git a/scripts/codeconverter/converter.py b/scripts/codeconverter/converter.py
new file mode 100755
index 000000000..75cb515d9
--- /dev/null
+++ b/scripts/codeconverter/converter.py
@@ -0,0 +1,123 @@
+#!/usr/bin/env python3
+# QEMU library
+#
+# Copyright (C) 2020 Red Hat Inc.
+#
+# Authors:
+#  Eduardo Habkost <ehabkost@redhat.com>
+#
+# This work is licensed under the terms of the GNU GPL, version 2.  See
+# the COPYING file in the top-level directory.
+#
+import sys
+import argparse
+import os
+import os.path
+import re
+from typing import *
+
+from codeconverter.patching import FileInfo, match_class_dict, FileList
+import codeconverter.qom_macros
+from codeconverter.qom_type_info import TI_FIELDS, type_infos, TypeInfoVar
+
+import logging
+logger = logging.getLogger(__name__)
+DBG = logger.debug
+INFO = logger.info
+WARN = logger.warning
+
+def process_all_files(parser: argparse.ArgumentParser, args: argparse.Namespace) -> None:
+    DBG("filenames: %r", args.filenames)
+
+    files = FileList()
+    files.extend(FileInfo(files, fn, args.force) for fn in args.filenames)
+    for f in files:
+        DBG('opening %s', f.filename)
+        f.load()
+
+    if args.table:
+        fields = ['filename', 'variable_name'] + TI_FIELDS
+        print('\t'.join(fields))
+        for f in files:
+            for t in f.matches_of_type(TypeInfoVar):
+                assert isinstance(t, TypeInfoVar)
+                values = [f.filename, t.name] + \
+                         [t.get_raw_initializer_value(f)
+                          for f in TI_FIELDS]
+                DBG('values: %r', values)
+                assert all('\t' not in v for v in values)
+                values = [v.replace('\n', ' ').replace('"', '') for v in values]
+                print('\t'.join(values))
+        return
+
+    match_classes = match_class_dict()
+    if not args.patterns:
+        parser.error("--pattern is required")
+
+    classes = [p for arg in args.patterns
+               for p in re.split(r'[\s,]', arg)
+               if p.strip()]
+    for c in classes:
+        if c not in match_classes \
+           or not match_classes[c].regexp:
+            print("Invalid pattern name: %s" % (c), file=sys.stderr)
+            print("Valid patterns:", file=sys.stderr)
+            print(PATTERN_HELP, file=sys.stderr)
+            sys.exit(1)
+
+    DBG("classes: %r", classes)
+    files.patch_content(max_passes=args.passes, class_names=classes)
+
+    for f in files:
+        #alltypes.extend(f.type_infos)
+        #full_types.extend(f.full_types())
+
+        if not args.dry_run:
+            if args.inplace:
+                f.patch_inplace()
+            if args.diff:
+                f.show_diff()
+            if not args.diff and not args.inplace:
+                f.write_to_file(sys.stdout)
+                sys.stdout.flush()
+
+
+PATTERN_HELP = ('\n'.join("  %s: %s" % (n, str(c.__doc__).strip())
+                for (n,c) in sorted(match_class_dict().items())
+                if c.has_replacement_rule()))
+
+def main() -> None:
+    p = argparse.ArgumentParser(formatter_class=argparse.RawDescriptionHelpFormatter)
+    p.add_argument('filenames', nargs='+')
+    p.add_argument('--passes', type=int, default=1,
+                   help="Number of passes (0 means unlimited)")
+    p.add_argument('--pattern', required=True, action='append',
+                   default=[], dest='patterns',
+                   help="Pattern to scan for")
+    p.add_argument('--inplace', '-i', action='store_true',
+                   help="Patch file in place")
+    p.add_argument('--dry-run', action='store_true',
+                   help="Don't patch files or print patching results")
+    p.add_argument('--force', '-f', action='store_true',
+                   help="Perform changes even if not completely safe")
+    p.add_argument('--diff', action='store_true',
+                   help="Print diff output on stdout")
+    p.add_argument('--debug', '-d', action='store_true',
+                   help="Enable debugging")
+    p.add_argument('--verbose', '-v', action='store_true',
+                   help="Verbose logging on stderr")
+    p.add_argument('--table', action='store_true',
+                   help="Print CSV table of type information")
+    p.add_argument_group("Valid pattern names",
+                         PATTERN_HELP)
+    args = p.parse_args()
+
+    loglevel = (logging.DEBUG if args.debug
+             else logging.INFO if args.verbose
+             else logging.WARN)
+    logging.basicConfig(format='%(levelname)s: %(message)s', level=loglevel)
+    DBG("args: %r", args)
+    process_all_files(p, args)
+
+if __name__ == '__main__':
+    main()
+\ No newline at end of file
author	Timos Ampelikiotis <t.ampelikiotis@virtualopensystems.com>	2023-10-10 11:40:56 +0000
committer	Timos Ampelikiotis <t.ampelikiotis@virtualopensystems.com>	2023-10-10 11:40:56 +0000
commit	e02cda008591317b1625707ff8e115a4841aa889 (patch)
tree	aee302e3cf8b59ec2d32ec481be3d1afddfc8968 /scripts/codeconverter
parent	cc668e6b7e0ffd8c9d130513d12053cf5eda1d3b (diff)