Add types and accessors for working with half-aligned 64-bit values.

[mirror_ovs.git] / build-aux / check-structs

#! /usr/bin/python

import os.path
import sys
import re

macros = {}

anyWarnings = False

types = {}
types['char'] = {"size": 1, "alignment": 1}
types['uint8_t'] = {"size": 1, "alignment": 1}
types['uint16_t'] = {"size": 2, "alignment": 2}
types['uint32_t'] = {"size": 4, "alignment": 4}
types['uint64_t'] = {"size": 8, "alignment": 8}
types['ovs_be16'] = {"size": 2, "alignment": 2}
types['ovs_be32'] = {"size": 4, "alignment": 4}
types['ovs_be64'] = {"size": 8, "alignment": 8}

token = None
line = ""
idRe = "[a-zA-Z_][a-zA-Z_0-9]*"
tokenRe = "#?" + idRe + "|[0-9]+|."
inComment = False
inDirective = False
def getToken():
    global token
    global line
    global inComment
    global inDirective
    while True:
        line = line.lstrip()
        if line != "":
            if line.startswith("/*"):
                inComment = True
                line = line[2:]
            elif inComment:
                commentEnd = line.find("*/")
                if commentEnd < 0:
                    line = ""
                else:
                    inComment = False
                    line = line[commentEnd + 2:]
            else:
                match = re.match(tokenRe, line)
                token = match.group(0)
                line = line[len(token):]
                if token.startswith('#'):
                    inDirective = True
                elif token in macros and not inDirective:
                    line = macros[token] + line
                    continue
                return True
        elif inDirective:
            token = "$"
            inDirective = False
            return True
        else:
            global lineNumber
            line = inputFile.readline()
            lineNumber += 1
            while line.endswith("\\\n"):
                line = line[:-2] + inputFile.readline()
                lineNumber += 1
            if line == "":
                if token == None:
                    fatal("unexpected end of input")
                token = None
                return False
    
def fatal(msg):
    sys.stderr.write("%s:%d: error at \"%s\": %s\n" % (fileName, lineNumber, token, msg))
    sys.exit(1)
    
def warn(msg):
    global anyWarnings
    anyWarnings = True
    sys.stderr.write("%s:%d: warning: %s\n" % (fileName, lineNumber, msg))

def skipDirective():
    getToken()
    while token != '$':
        getToken()

def isId(s):
    return re.match(idRe + "$", s) != None

def forceId():
    if not isId(token):
        fatal("identifier expected")

def forceInteger():
    if not re.match('[0-9]+$', token):
        fatal("integer expected")

def match(t):
    if token == t:
        getToken()
        return True
    else:
        return False

def forceMatch(t):
    if not match(t):
        fatal("%s expected" % t)

def parseTaggedName():
    assert token in ('struct', 'union')
    name = token
    getToken()
    forceId()
    name = "%s %s" % (name, token)
    getToken()
    return name

def parseTypeName():
    if token in ('struct', 'union'):
        name = parseTaggedName()
    elif isId(token):
        name = token
        getToken()
    else:
        fatal("type name expected")

    if name in types:
        return name
    else:
        fatal("unknown type \"%s\"" % name)

def parseStruct():
    isStruct = token == 'struct'
    structName = parseTaggedName()
    if token == ";":
        return

    ofs = size = 0
    alignment = 4               # ARM has minimum 32-bit alignment
    forceMatch('{')
    while not match('}'):
        typeName = parseTypeName()
        typeSize = types[typeName]['size']
        typeAlignment = types[typeName]['alignment']

        forceId()
        memberName = token
        getToken()

        if match('['):
            if token == ']':
                count = 0
            else:
                forceInteger()
                count = int(token)
                getToken()
            forceMatch(']')
        else:
            count = 1

        nBytes = typeSize * count
        if isStruct:
            if ofs % typeAlignment:
                shortage = typeAlignment - (ofs % typeAlignment)
                warn("%s member %s is %d bytes short of %d-byte alignment"
                     % (structName, memberName, shortage, typeAlignment))
                size += shortage
                ofs += shortage
            size += nBytes
            ofs += nBytes
        else:
            if nBytes > size:
                size = nBytes
        if typeAlignment > alignment:
            alignment = typeAlignment

        forceMatch(';')
    if size % alignment:
        shortage = alignment - (size % alignment)
        if (structName == "struct ofp_packet_in" and
            shortage == 2 and
            memberName == 'data' and
            count == 0):
            # This is intentional
            pass
        else:
            warn("%s needs %d bytes of tail padding" % (structName, shortage))
        size += shortage
    types[structName] = {"size": size, "alignment": alignment}

def checkStructs():
    if len(sys.argv) < 2:
        sys.stderr.write("at least one non-option argument required; "
                         "use --help for help")
        sys.exit(1)

    if '--help' in sys.argv:
        argv0 = os.path.basename(sys.argv[0])
        print '''\
%(argv0)s, for checking struct and struct member alignment
usage: %(argv0)s HEADER [HEADER]...

This program reads the header files specified on the command line and
verifies that all struct members are aligned on natural boundaries
without any need for the compiler to add additional padding.  It also
verifies that each struct's size is a multiple of 32 bits (because
some ABIs for ARM require all structs to be a multiple of 32 bits), or
64 bits if the struct has any 64-bit members, again without the
compiler adding additional padding.  Finally, it checks struct size
assertions using OFP_ASSERT.

Header files are read in the order specified.  #include directives are
not processed, so specify them in dependency order.

This program is specialized for reading include/openflow/openflow.h
and include/openflow/nicira-ext.h.  It will not work on arbitrary
header files without extensions.''' % {"argv0": argv0}
        sys.exit(0)

    global fileName
    for fileName in sys.argv[1:]:
        global inputFile
        global lineNumber
        inputFile = open(fileName)
        lineNumber = 0
        while getToken():
            if token in ("#ifdef", "#ifndef", "#include",
                         "#endif", "#elif", "#else"):
                skipDirective()
            elif token == "#define":
                getToken()
                name = token
                if line.startswith('('):
                    skipDirective()
                else:
                    definition = ""
                    getToken()
                    while token != '$':
                        definition += token
                        getToken()
                    macros[name] = definition
            elif token == "enum":
                while token != ';':
                    getToken()
            elif token in ('struct', 'union'):
                parseStruct()
            elif match('OFP_ASSERT') or match('BOOST_STATIC_ASSERT'):
                forceMatch('(')
                forceMatch('sizeof')
                forceMatch('(')
                typeName = parseTypeName()
                forceMatch(')')
                forceMatch('=')
                forceMatch('=')
                forceInteger()
                size = int(token)
                getToken()
                forceMatch(')')
                if types[typeName]['size'] != size:
                    warn("%s is %d bytes long but declared as %d" % (
                            typeName, types[typeName]['size'], size))
            else:
                fatal("parse error")
        inputFile.close()
    if anyWarnings:
        sys.exit(1)

if __name__ == '__main__':
    checkStructs()