mcu/tools/mertos/mer_utility.py

#!/usr/bin/python
# ****************************************************************************
#  Copyright Statement:
#  --------------------
#  This software is protected by Copyright and the information contained
#  herein is confidential. The software may not be copied and the information
#  contained herein may not be used or disclosed except with the written
#  permission of MediaTek Inc. (C) 2018
#
#  BY OPENING THIS FILE, BUYER HEREBY UNEQUIVOCALLY ACKNOWLEDGES AND AGREES
#  THAT THE SOFTWARE/FIRMWARE AND ITS DOCUMENTATIONS ("MEDIATEK SOFTWARE")
#  RECEIVED FROM MEDIATEK AND/OR ITS REPRESENTATIVES ARE PROVIDED TO BUYER ON
#  AN "AS-IS" BASIS ONLY. MEDIATEK EXPRESSLY DISCLAIMS ANY AND ALL WARRANTIES,
#  EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE IMPLIED WARRANTIES OF
#  MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE OR NONINFRINGEMENT.
#  NEITHER DOES MEDIATEK PROVIDE ANY WARRANTY WHATSOEVER WITH RESPECT TO THE
#  SOFTWARE OF ANY THIRD PARTY WHICH MAY BE USED BY, INCORPORATED IN, OR
#  SUPPLIED WITH THE MEDIATEK SOFTWARE, AND BUYER AGREES TO LOOK ONLY TO SUCH
#  THIRD PARTY FOR ANY WARRANTY CLAIM RELATING THERETO. MEDIATEK SHALL ALSO
#  NOT BE RESPONSIBLE FOR ANY MEDIATEK SOFTWARE RELEASES MADE TO BUYER'S
#  SPECIFICATION OR TO CONFORM TO A PARTICULAR STANDARD OR OPEN FORUM.
#
#  BUYER'S SOLE AND EXCLUSIVE REMEDY AND MEDIATEK'S ENTIRE AND CUMULATIVE
#  LIABILITY WITH RESPECT TO THE MEDIATEK SOFTWARE RELEASED HEREUNDER WILL BE,
#  AT MEDIATEK'S OPTION, TO REVISE OR REPLACE THE MEDIATEK SOFTWARE AT ISSUE,
#  OR REFUND ANY SOFTWARE LICENSE FEES OR SERVICE CHARGE PAID BY BUYER TO
#  MEDIATEK FOR SUCH MEDIATEK SOFTWARE AT ISSUE.
#
#  THE TRANSACTION CONTEMPLATED HEREUNDER SHALL BE CONSTRUED IN ACCORDANCE
#  WITH THE LAWS OF THE STATE OF CALIFORNIA, USA, EXCLUDING ITS CONFLICT OF
#  LAWS PRINCIPLES.  ANY DISPUTES, CONTROVERSIES OR CLAIMS ARISING THEREOF AND
#  RELATED THERETO SHALL BE SETTLED BY ARBITRATION IN SAN FRANCISCO, CA, UNDER
#  THE RULES OF THE INTERNATIONAL CHAMBER OF COMMERCE (ICC).
#
# ****************************************************************************
#
# ****************************************************************************
#
#  Filename:
#  ---------
#    mer_utility.py
#
#  Project:
#  --------
#    MERTOS
#
#  Description:
#  ------------
#    Provide utilities used by MERTOS code generating.
#
#
#  Author:
#  -------
#    Ke-Ting Chen (mtk03141)
#
# ****************************************************************************

import struct
import subprocess


class ElfDataExtractor(object):
    def __init__(self, readelf_exec, elf_path):
        self.elf_path = elf_path

        # Get symbol table
        # Example lines:
        #   Num:    Value  Size Type    Bind   Vis      Ndx Name
        #    20: 00000b1c  5216 OBJECT  GLOBAL DEFAULT    9 sys_comp_config_tbl
        self.symbols = {}
        symbol_text_lines = subprocess.check_output([readelf_exec, '--symbols',
                                                     '--wide', elf_path]).splitlines()
        for line in symbol_text_lines:
            if line and line.lstrip()[0].isdigit():
                contents = line.split()
                if len(contents) == 7:
                    contents.append(contents[0][:-1])
                    #print contents[0][:-1]
                self.symbols[contents[7]] = {
                    'name': contents[7],
                    'type': contents[3],
                    'section_number': (int(contents[6]) if contents[6].isdigit() else contents[6]),
                    'address': int(contents[1], 16),
                    'size': int(contents[2]),
                    'number': int(contents[0][:-1])}

        # Assumes there is no hole in the number
        self.symbol_list = sorted(self.symbols.viewvalues(), key=lambda x: x['number'])

        # Get section info
        # Example lines:
        #   [Nr] Name    Type     Addr     Off    Size   ES Flg Lk Inf Al
        #   [ 9] .rodata PROGBITS 00000000 000aa0 00231c 00   A  0   0  4
        self.sections = {}
        section_text_lines = subprocess.check_output([readelf_exec, '--sections',
                                                      '--wide', elf_path]).splitlines()
        for line in section_text_lines:
            line = line.translate(None, '[]').lstrip()
            if line and line[0].isdigit():
                contents = line.split()
                if contents[0] == '0':
                    self.sections['NULL'] = {'name': 'NULL', 'number': 0,
                                             'address': 0, 'offset': 0, 'size': 0}
                else:
                    self.sections[contents[1]] = {
                        'name': contents[1], 'number': int(contents[0]),
                        'address': int(contents[3], 16), 'offset': int(contents[4], 16),
                        'size': int(contents[5], 16)}

        # Assumes there is no hole in the section number
        self.section_list = sorted(self.sections.viewvalues(), key=lambda x: x['number'])

        # Get relocation tables
        self.relocations = {}
        with open(elf_path, 'rb') as elf_file:
            for name, section in self.sections.viewitems():
                if name.startswith('.rel.') or name.startswith('.rela.'):
                    elf_file.seek(section['offset'])
                    raw_data = elf_file.read(section['size'])
                    if name.startswith('.rel.'):
                        #print 'processing ' + name
                        relocation = self.unpack_data_struct_array(
                            raw_data, '=IxHx', ['offset', 'symbol_number'])
                        for reloc_info in relocation:
                            reloc_info['addend'] = 0
                        name = name[len('.rel'):]
                    else:
                        relocation = self.unpack_data_struct_array(
                            raw_data, '=IBHxI', ['offset', 'type', 'symbol_number', 'addend'])
                        #print 'processing ' + name
                        name = name[len('.rela'):]
                    relocation.sort(key=lambda x: x['offset'])
                    self.relocations[name] = relocation

        # Read in the elf and perform relocation to simplify query
        elf_data = list(open(elf_path, 'rb').read())
        for section_name, relocation in self.relocations.viewitems():
            section_offset = self.sections[section_name]['offset']
            for reloc_info in relocation:
                elf_address = reloc_info['offset'] + section_offset
                symbol = self.symbol_list[reloc_info['symbol_number']]
                #print symbol['name'], reloc_info['symbol_number'], symbol['type']
                if symbol['type'] == 'SECTION':
                    section = self.section_list[symbol['section_number']]
                    original_value = struct.unpack_from(
                        'I', ''.join(elf_data[elf_address:elf_address + 4]))[0]
                    fill_data = original_value + section['offset'] + reloc_info['addend']
                    #print 'SECTION', symbol['name'], elf_data[elf_address:elf_address+4], hex(fill_data), hex(original_value)
                else:
                    fill_data = reloc_info['symbol_number'] | 0x80000000
                    #print 'OTHER', symbol['name'], elf_data[elf_address:elf_address+4], hex(fill_data)
                elf_data[elf_address:elf_address + 4] = struct.pack('I', fill_data)
        elf_data = ''.join(elf_data)

        # Get symbol data from elf
        for symbol in self.symbols.viewvalues():
            if symbol['size'] != 0:
                # Get the content address in elf
                section = self.section_list[symbol['section_number']]
                address_offset = symbol['address'] - section['address']
                elf_address = address_offset + section['offset']
                # Get content from elf file
                symbol['data'] = elf_data[elf_address:elf_address + symbol['size']]

    def get_symbol_data(self, symbol_name):
        return self.symbols[symbol_name]['data']

    def get_symbol_section_name(self, symbol_name):
        return self.section_list[
            self.symbols[symbol_name]['section_number']]['name']

    def get_symbol_name(self, target_address):
        if target_address | 0x80000000:
            symbol_number = target_address & 0x7FFFFFFF
            return self.symbol_list[symbol_number]['name']
        else:
            for symbol in self.symbol_list:
                if symbol['offset'] == target_address:
                    return symbol['name']
            else:
                return 'SYMBOL_NOT_FOUND'

    def get_string_data(self, target_address):
        elf_address = target_address

        with open(self.elf_path, 'rb') as elf_file:
            elf_file.seek(elf_address)
            contents = ''
            find_begin_index = 0
            while 1:
                contents += elf_file.read(32)
                string_end_index = contents.find('\x00', find_begin_index)
                if string_end_index == -1:
                    find_begin_index = len(contents)
                else:
                    contents = contents[:string_end_index+1]
                    break
        return contents

    def unpack_symbol_data(self, format_string, symbol_name):
        return struct.unpack_from(format_string, self.get_symbol_data(symbol_name))

    def unpack_data_struct_array(self, raw_data, struct_format_string, field_names):
        struct_size = struct.calcsize(struct_format_string)
        array_length = len(raw_data) / struct_size
        array = []
        #print len(raw_data), struct_size, struct_format_string, field_names

        for index in xrange(array_length):
            data = struct.unpack_from(struct_format_string, raw_data, struct_size * index)
            entry = dict(zip(field_names, data))
            array.append(entry)

        return array

    def unpack_symbol_data_struct_array(self, symbol_name, struct_format_string, field_names):
        raw_data = self.get_symbol_data(symbol_name)
        return self.unpack_data_struct_array(raw_data, struct_format_string, field_names)