pebble-firmware/sdk/tools/inject_metadata.py

#!/usr/bin/env python
# Copyright 2024 Google LLC
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#     http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.


from __future__ import with_statement
from struct import pack, unpack

import os
import os.path
import sys
import time

from subprocess import Popen, PIPE
from shutil import copy2
from binascii import crc32
from struct import pack
from pbpack import ResourcePack


import stm32_crc

# Pebble App Metadata Struct
# These are offsets of the PebbleProcessInfo struct in src/fw/app_management/pebble_process_info.h
HEADER_ADDR = 0x0                # 8 bytes
STRUCT_VERSION_ADDR = 0x8        # 2 bytes
SDK_VERSION_ADDR = 0xa           # 2 bytes
APP_VERSION_ADDR = 0xc           # 2 bytes
LOAD_SIZE_ADDR = 0xe             # 2 bytes
OFFSET_ADDR = 0x10               # 4 bytes
CRC_ADDR = 0x14                  # 4 bytes
NAME_ADDR = 0x18                 # 32 bytes
COMPANY_ADDR = 0x38              # 32 bytes
ICON_RES_ID_ADDR = 0x58          # 4 bytes
JUMP_TABLE_ADDR = 0x5c           # 4 bytes
FLAGS_ADDR = 0x60                # 4 bytes
NUM_RELOC_ENTRIES_ADDR = 0x64    # 4 bytes
UUID_ADDR = 0x68                 # 16 bytes
RESOURCE_CRC_ADDR = 0x78         # 4 bytes
RESOURCE_TIMESTAMP_ADDR = 0x7c   # 4 bytes
VIRTUAL_SIZE_ADDR = 0x80         # 2 bytes
STRUCT_SIZE_BYTES = 0x82

# Pebble App Flags
# These are PebbleAppFlags from src/fw/app_management/pebble_process_info.h
PROCESS_INFO_STANDARD_APP = (0)
PROCESS_INFO_WATCH_FACE = (1 << 0)
PROCESS_INFO_VISIBILITY_HIDDEN = (1 << 1)
PROCESS_INFO_VISIBILITY_SHOWN_ON_COMMUNICATION = (1 << 2)
PROCESS_INFO_ALLOW_JS = (1 << 3)
PROCESS_INFO_HAS_WORKER = (1 << 4)

# Max app size, including the struct and reloc table
# Note that even if the app is smaller than this, it still may be too big, as it needs to share this
# space with applib/ which changes in size from release to release.
MAX_APP_BINARY_SIZE = 0x10000

# This number is a rough estimate, but should not be less than the available space.
# Currently, app_state uses up a small part of the app space.
# See also APP_RAM in stm32f2xx_flash_fw.ld and APP in pebble_app.ld.
MAX_APP_MEMORY_SIZE = 24 * 1024

# This number is a rough estimate, but should not be less than the available space.
# Currently, worker_state uses up a small part of the worker space.
# See also WORKER_RAM in stm32f2xx_flash_fw.ld
MAX_WORKER_MEMORY_SIZE = 10 * 1024

ENTRY_PT_SYMBOL = 'main'
JUMP_TABLE_ADDR_SYMBOL = 'pbl_table_addr'
DEBUG = False


class InvalidBinaryError(Exception):
    pass


def inject_metadata(target_binary, target_elf, resources_file, timestamp, allow_js=False,
                    has_worker=False):

    if target_binary[-4:] != '.bin':
        raise Exception("Invalid filename <%s>! The filename should end in .bin" % target_binary)

    def get_nm_output(elf_file):
        nm_process = Popen(['arm-none-eabi-nm', elf_file], stdout=PIPE)
        # Popen.communicate returns a tuple of (stdout, stderr)
        nm_output = nm_process.communicate()[0].decode("utf8")

        if not nm_output:
            raise InvalidBinaryError()

        nm_output = [ line.split() for line in nm_output.splitlines() ]
        return nm_output

    def get_symbol_addr(nm_output, symbol):
        # nm output looks like the following...
        #
        #          U _ITM_registerTMCloneTable
        # 00000084 t jump_to_pbl_function
        #          U _Jv_RegisterClasses
        # 0000009c T main
        # 00000130 T memset
        #
        # We don't care about the lines that only have two columns, they're not functions.

        for sym in nm_output:
            if symbol == sym[-1] and len(sym) == 3:
                return int(sym[0], 16)

        raise Exception("Could not locate symbol <%s> in binary! Failed to inject app metadata" %
                        (symbol))

    def get_virtual_size(elf_file):
        """ returns the virtual size (static memory usage, .text + .data + .bss) in bytes """

        readelf_bss_process = Popen("arm-none-eabi-readelf -S '%s'" % elf_file, 
                                    shell=True, stdout=PIPE)
        readelf_bss_output = readelf_bss_process.communicate()[0].decode("utf8")

        # readelf -S output looks like the following...
        #
        # [Nr] Name              Type            Addr     Off    Size   ES Flg Lk Inf Al
        # [ 0]                   NULL            00000000 000000 000000 00      0   0  0
        # [ 1] .header           PROGBITS        00000000 008000 000082 00   A  0   0  1
        # [ 2] .text             PROGBITS        00000084 008084 0006be 00  AX  0   0  4
        # [ 3] .rel.text         REL             00000000 00b66c 0004d0 08     23   2  4
        # [ 4] .data             PROGBITS        00000744 008744 000004 00  WA  0   0  4
        # [ 5] .bss              NOBITS          00000748 008748 000054 00  WA  0   0  4

        last_section_end_addr = 0

        # Find the .bss section and calculate the size based on the end of the .bss section
        for line in readelf_bss_output.splitlines():
            if len(line) < 10:
                continue

            # Carve off the first column, since it sometimes has a space in it which screws up the
            # split.
            if not ']' in line:
                continue
            line = line[line.index(']') + 1:]

            columns = line.split()
            if len(columns) < 6:
                continue

            if columns[0] == '.bss':
                addr = int(columns[2], 16)
                size = int(columns[4], 16)
                last_section_end_addr = addr + size
            elif columns[0] == '.data' and last_section_end_addr == 0:
                addr = int(columns[2], 16)
                size = int(columns[4], 16)
                last_section_end_addr = addr + size

        if last_section_end_addr != 0:
            return last_section_end_addr

        sys.stderr.writeline("Failed to parse ELF sections while calculating the virtual size\n")
        sys.stderr.write(readelf_bss_output)
        raise Exception("Failed to parse ELF sections while calculating the virtual size")

    def get_relocate_entries(elf_file):
        """ returns a list of all the locations requiring an offset"""
        # TODO: insert link to the wiki page I'm about to write about PIC and relocatable values
        entries = []

        # get the .data locations
        readelf_relocs_process = Popen(['arm-none-eabi-readelf', '-r', elf_file], stdout=PIPE)
        readelf_relocs_output = readelf_relocs_process.communicate()[0].decode("utf8")
        lines = readelf_relocs_output.splitlines()

        i = 0
        reading_section = False
        while i < len(lines):
            if not reading_section:
                # look for the next section
                if lines[i].startswith("Relocation section '.rel.data"):
                    reading_section = True
                    i += 1 # skip the column title section
            else:
                if len(lines[i]) == 0:
                    # end of the section
                    reading_section = False
                else:
                    entries.append(int(lines[i].split(' ')[0], 16))
            i += 1

        # get any Global Offset Table (.got) entries
        readelf_relocs_process = Popen(['arm-none-eabi-readelf', '--sections', elf_file],
                                       stdout=PIPE)
        readelf_relocs_output = readelf_relocs_process.communicate()[0].decode("utf8")
        lines = readelf_relocs_output.splitlines()
        for line in lines:
            # We shouldn't need to do anything with the Procedure Linkage Table since we don't
            # actually export functions
            if '.got' in line and '.got.plt' not in line:
                words = line.split(' ')
                while '' in words:
                    words.remove('')
                section_label_idx = words.index('.got')
                addr = int(words[section_label_idx + 2], 16)
                length = int(words[section_label_idx + 4], 16)
                for i in range(addr, addr + length, 4):
                    entries.append(i)
                break

        return entries


    nm_output = get_nm_output(target_elf)

    try:
        app_entry_address = get_symbol_addr(nm_output, ENTRY_PT_SYMBOL)
    except:
        raise Exception("Missing app entry point! Must be `int main(void) { ... }` ")
    jump_table_address = get_symbol_addr(nm_output, JUMP_TABLE_ADDR_SYMBOL)

    reloc_entries = get_relocate_entries(target_elf)

    statinfo = os.stat(target_binary)
    app_load_size = statinfo.st_size

    if resources_file is not None:
        with open(resources_file, 'rb') as f:
            pbpack = ResourcePack.deserialize(f, is_system=False)
            resource_crc = pbpack.get_content_crc()
    else:
        resource_crc = 0

    if DEBUG:
        copy2(target_binary, target_binary + ".orig")

    with open(target_binary, 'r+b') as f:
        total_app_image_size = app_load_size + (len(reloc_entries) * 4)
        if total_app_image_size > MAX_APP_BINARY_SIZE:
            raise Exception("App image size is %u (app %u relocation table %u). Must be smaller "
                            "than %u bytes" % (total_app_image_size,
                                               app_load_size,
                                               len(reloc_entries) * 4,
                                               MAX_APP_BINARY_SIZE))

        def read_value_at_offset(offset, format_str, size):
            f.seek(offset)
            return unpack(format_str, f.read(size))

        app_bin = f.read()
        app_crc = stm32_crc.crc32(app_bin[STRUCT_SIZE_BYTES:])

        [app_flags] = read_value_at_offset(FLAGS_ADDR, '<L', 4)

        if allow_js:
            app_flags = app_flags | PROCESS_INFO_ALLOW_JS

        if has_worker:
            app_flags = app_flags | PROCESS_INFO_HAS_WORKER

        app_virtual_size = get_virtual_size(target_elf)

        struct_changes = {
            'load_size' : app_load_size,
            'entry_point' : "0x%08x" % app_entry_address,
            'symbol_table' : "0x%08x" % jump_table_address,
            'flags' : app_flags,
            'crc' : "0x%08x" % app_crc,
            'num_reloc_entries': "0x%08x" % len(reloc_entries),
            'resource_crc' : "0x%08x" % resource_crc,
            'timestamp' : timestamp,
            'virtual_size': app_virtual_size
        }

        def write_value_at_offset(offset, format_str, value):
            f.seek(offset)
            f.write(pack(format_str, value))

        write_value_at_offset(LOAD_SIZE_ADDR, '<H', app_load_size)
        write_value_at_offset(OFFSET_ADDR, '<L', app_entry_address)
        write_value_at_offset(CRC_ADDR, '<L', app_crc)

        write_value_at_offset(RESOURCE_CRC_ADDR, '<L', resource_crc)
        write_value_at_offset(RESOURCE_TIMESTAMP_ADDR, '<L', timestamp)

        write_value_at_offset(JUMP_TABLE_ADDR, '<L', jump_table_address)

        write_value_at_offset(FLAGS_ADDR, '<L', app_flags)

        write_value_at_offset(NUM_RELOC_ENTRIES_ADDR, '<L', len(reloc_entries))

        write_value_at_offset(VIRTUAL_SIZE_ADDR, "<H", app_virtual_size)

        # Write the reloc_entries past the end of the binary. This expands the size of the binary,
        # but this new stuff won't actually be loaded into ram.
        f.seek(app_load_size)
        for entry in reloc_entries:
            f.write(pack('<L', entry))

        f.flush()

    return struct_changes