2023-02-27 18:54:48 +03:00
#!/usr/bin/env python3
# SPDX-License-Identifier: LGPL-2.1-or-later
# Convert ELF static PIE to PE/EFI image.
# To do so we simply copy desired ELF sections while preserving their memory layout to ensure that
# code still runs as expected. We then translate ELF relocations to PE relocations so that the EFI
# loader/firmware can properly load the binary to any address at runtime.
#
# To make this as painless as possible we only operate on static PIEs as they should only contain
# base relocations that are easy to handle as they have a one-to-one mapping to PE relocations.
#
# EDK2 does a similar process using their GenFw tool. The main difference is that they use the
# --emit-relocs linker flag, which emits a lot of different (static) ELF relocation types that have
# to be handled differently for each architecture and is overall more work than its worth.
#
# Note that on arches where binutils has PE support (x86/x86_64 mostly, aarch64 only recently)
# objcopy can be used to convert ELF to PE. But this will still not convert ELF relocations, making
# the resulting binary useless. gnu-efi relies on this method and contains a stub that performs the
# ELF dynamic relocations at runtime.
2023-08-09 22:57:47 +03:00
# pylint: disable=attribute-defined-outside-init
2023-02-27 18:54:48 +03:00
import argparse
import hashlib
import io
import os
import pathlib
import time
2023-05-23 20:00:52 +03:00
import typing
2023-02-27 18:54:48 +03:00
from ctypes import (
c_char ,
c_uint8 ,
c_uint16 ,
c_uint32 ,
c_uint64 ,
LittleEndianStructure ,
sizeof ,
)
from elftools . elf . constants import SH_FLAGS
2023-09-28 17:09:42 +03:00
from elftools . elf . elffile import ELFFile
2023-02-27 18:54:48 +03:00
from elftools . elf . enums import (
ENUM_DT_FLAGS_1 ,
ENUM_RELOC_TYPE_AARCH64 ,
ENUM_RELOC_TYPE_ARM ,
ENUM_RELOC_TYPE_i386 ,
ENUM_RELOC_TYPE_x64 ,
)
from elftools . elf . relocation import (
Relocation as ElfRelocation ,
RelocationTable as ElfRelocationTable ,
)
class PeCoffHeader ( LittleEndianStructure ) :
_fields_ = (
tools/elf2efi: align columns in tables, unify formatting
For tables which represent binary data structures, readability is greatly
enhanced if the part which shows field size and type is aligned. This follows
the usual style for tables in the rest of the systemd codebase.
Also, use the same style for functions: if the function signature is too long
to fit in one line, put each parameter on a separate line.
Also, for comprehension expressions, if they are split, use the usual Python
style.
Also, drop format annotations, since the code isn't automatically formatted
anymore, and automatic formatting is neither feasible nor a goal for the
systemd codebase.
2024-03-13 15:16:17 +03:00
( " Machine " , c_uint16 ) ,
( " NumberOfSections " , c_uint16 ) ,
( " TimeDateStamp " , c_uint32 ) ,
2023-02-27 18:54:48 +03:00
( " PointerToSymbolTable " , c_uint32 ) ,
tools/elf2efi: align columns in tables, unify formatting
For tables which represent binary data structures, readability is greatly
enhanced if the part which shows field size and type is aligned. This follows
the usual style for tables in the rest of the systemd codebase.
Also, use the same style for functions: if the function signature is too long
to fit in one line, put each parameter on a separate line.
Also, for comprehension expressions, if they are split, use the usual Python
style.
Also, drop format annotations, since the code isn't automatically formatted
anymore, and automatic formatting is neither feasible nor a goal for the
systemd codebase.
2024-03-13 15:16:17 +03:00
( " NumberOfSymbols " , c_uint32 ) ,
2023-02-27 18:54:48 +03:00
( " SizeOfOptionalHeader " , c_uint16 ) ,
tools/elf2efi: align columns in tables, unify formatting
For tables which represent binary data structures, readability is greatly
enhanced if the part which shows field size and type is aligned. This follows
the usual style for tables in the rest of the systemd codebase.
Also, use the same style for functions: if the function signature is too long
to fit in one line, put each parameter on a separate line.
Also, for comprehension expressions, if they are split, use the usual Python
style.
Also, drop format annotations, since the code isn't automatically formatted
anymore, and automatic formatting is neither feasible nor a goal for the
systemd codebase.
2024-03-13 15:16:17 +03:00
( " Characteristics " , c_uint16 ) ,
2023-02-27 18:54:48 +03:00
)
class PeDataDirectory ( LittleEndianStructure ) :
_fields_ = (
( " VirtualAddress " , c_uint32 ) ,
tools/elf2efi: align columns in tables, unify formatting
For tables which represent binary data structures, readability is greatly
enhanced if the part which shows field size and type is aligned. This follows
the usual style for tables in the rest of the systemd codebase.
Also, use the same style for functions: if the function signature is too long
to fit in one line, put each parameter on a separate line.
Also, for comprehension expressions, if they are split, use the usual Python
style.
Also, drop format annotations, since the code isn't automatically formatted
anymore, and automatic formatting is neither feasible nor a goal for the
systemd codebase.
2024-03-13 15:16:17 +03:00
( " Size " , c_uint32 ) ,
2023-02-27 18:54:48 +03:00
)
class PeRelocationBlock ( LittleEndianStructure ) :
_fields_ = (
tools/elf2efi: align columns in tables, unify formatting
For tables which represent binary data structures, readability is greatly
enhanced if the part which shows field size and type is aligned. This follows
the usual style for tables in the rest of the systemd codebase.
Also, use the same style for functions: if the function signature is too long
to fit in one line, put each parameter on a separate line.
Also, for comprehension expressions, if they are split, use the usual Python
style.
Also, drop format annotations, since the code isn't automatically formatted
anymore, and automatic formatting is neither feasible nor a goal for the
systemd codebase.
2024-03-13 15:16:17 +03:00
( " PageRVA " , c_uint32 ) ,
2023-02-27 18:54:48 +03:00
( " BlockSize " , c_uint32 ) ,
)
def __init__ ( self , PageRVA : int ) :
super ( ) . __init__ ( PageRVA )
2023-07-14 15:16:44 +03:00
self . entries : typing . List [ PeRelocationEntry ] = [ ]
2023-02-27 18:54:48 +03:00
class PeRelocationEntry ( LittleEndianStructure ) :
_fields_ = (
( " Offset " , c_uint16 , 12 ) ,
tools/elf2efi: align columns in tables, unify formatting
For tables which represent binary data structures, readability is greatly
enhanced if the part which shows field size and type is aligned. This follows
the usual style for tables in the rest of the systemd codebase.
Also, use the same style for functions: if the function signature is too long
to fit in one line, put each parameter on a separate line.
Also, for comprehension expressions, if they are split, use the usual Python
style.
Also, drop format annotations, since the code isn't automatically formatted
anymore, and automatic formatting is neither feasible nor a goal for the
systemd codebase.
2024-03-13 15:16:17 +03:00
( " Type " , c_uint16 , 4 ) ,
2023-02-27 18:54:48 +03:00
)
class PeOptionalHeaderStart ( LittleEndianStructure ) :
_fields_ = (
tools/elf2efi: align columns in tables, unify formatting
For tables which represent binary data structures, readability is greatly
enhanced if the part which shows field size and type is aligned. This follows
the usual style for tables in the rest of the systemd codebase.
Also, use the same style for functions: if the function signature is too long
to fit in one line, put each parameter on a separate line.
Also, for comprehension expressions, if they are split, use the usual Python
style.
Also, drop format annotations, since the code isn't automatically formatted
anymore, and automatic formatting is neither feasible nor a goal for the
systemd codebase.
2024-03-13 15:16:17 +03:00
( " Magic " , c_uint16 ) ,
( " MajorLinkerVersion " , c_uint8 ) ,
( " MinorLinkerVersion " , c_uint8 ) ,
( " SizeOfCode " , c_uint32 ) ,
( " SizeOfInitializedData " , c_uint32 ) ,
2023-02-27 18:54:48 +03:00
( " SizeOfUninitializedData " , c_uint32 ) ,
tools/elf2efi: align columns in tables, unify formatting
For tables which represent binary data structures, readability is greatly
enhanced if the part which shows field size and type is aligned. This follows
the usual style for tables in the rest of the systemd codebase.
Also, use the same style for functions: if the function signature is too long
to fit in one line, put each parameter on a separate line.
Also, for comprehension expressions, if they are split, use the usual Python
style.
Also, drop format annotations, since the code isn't automatically formatted
anymore, and automatic formatting is neither feasible nor a goal for the
systemd codebase.
2024-03-13 15:16:17 +03:00
( " AddressOfEntryPoint " , c_uint32 ) ,
( " BaseOfCode " , c_uint32 ) ,
2023-02-27 18:54:48 +03:00
)
class PeOptionalHeaderMiddle ( LittleEndianStructure ) :
_fields_ = (
tools/elf2efi: align columns in tables, unify formatting
For tables which represent binary data structures, readability is greatly
enhanced if the part which shows field size and type is aligned. This follows
the usual style for tables in the rest of the systemd codebase.
Also, use the same style for functions: if the function signature is too long
to fit in one line, put each parameter on a separate line.
Also, for comprehension expressions, if they are split, use the usual Python
style.
Also, drop format annotations, since the code isn't automatically formatted
anymore, and automatic formatting is neither feasible nor a goal for the
systemd codebase.
2024-03-13 15:16:17 +03:00
( " SectionAlignment " , c_uint32 ) ,
( " FileAlignment " , c_uint32 ) ,
2023-02-27 18:54:48 +03:00
( " MajorOperatingSystemVersion " , c_uint16 ) ,
( " MinorOperatingSystemVersion " , c_uint16 ) ,
tools/elf2efi: align columns in tables, unify formatting
For tables which represent binary data structures, readability is greatly
enhanced if the part which shows field size and type is aligned. This follows
the usual style for tables in the rest of the systemd codebase.
Also, use the same style for functions: if the function signature is too long
to fit in one line, put each parameter on a separate line.
Also, for comprehension expressions, if they are split, use the usual Python
style.
Also, drop format annotations, since the code isn't automatically formatted
anymore, and automatic formatting is neither feasible nor a goal for the
systemd codebase.
2024-03-13 15:16:17 +03:00
( " MajorImageVersion " , c_uint16 ) ,
( " MinorImageVersion " , c_uint16 ) ,
( " MajorSubsystemVersion " , c_uint16 ) ,
( " MinorSubsystemVersion " , c_uint16 ) ,
( " Win32VersionValue " , c_uint32 ) ,
( " SizeOfImage " , c_uint32 ) ,
( " SizeOfHeaders " , c_uint32 ) ,
( " CheckSum " , c_uint32 ) ,
( " Subsystem " , c_uint16 ) ,
( " DllCharacteristics " , c_uint16 ) ,
2023-02-27 18:54:48 +03:00
)
class PeOptionalHeaderEnd ( LittleEndianStructure ) :
_fields_ = (
tools/elf2efi: align columns in tables, unify formatting
For tables which represent binary data structures, readability is greatly
enhanced if the part which shows field size and type is aligned. This follows
the usual style for tables in the rest of the systemd codebase.
Also, use the same style for functions: if the function signature is too long
to fit in one line, put each parameter on a separate line.
Also, for comprehension expressions, if they are split, use the usual Python
style.
Also, drop format annotations, since the code isn't automatically formatted
anymore, and automatic formatting is neither feasible nor a goal for the
systemd codebase.
2024-03-13 15:16:17 +03:00
( " LoaderFlags " , c_uint32 ) ,
( " NumberOfRvaAndSizes " , c_uint32 ) ,
( " ExportTable " , PeDataDirectory ) ,
( " ImportTable " , PeDataDirectory ) ,
( " ResourceTable " , PeDataDirectory ) ,
( " ExceptionTable " , PeDataDirectory ) ,
( " CertificateTable " , PeDataDirectory ) ,
( " BaseRelocationTable " , PeDataDirectory ) ,
( " Debug " , PeDataDirectory ) ,
( " Architecture " , PeDataDirectory ) ,
( " GlobalPtr " , PeDataDirectory ) ,
( " TLSTable " , PeDataDirectory ) ,
( " LoadConfigTable " , PeDataDirectory ) ,
( " BoundImport " , PeDataDirectory ) ,
( " IAT " , PeDataDirectory ) ,
2023-02-27 18:54:48 +03:00
( " DelayImportDescriptor " , PeDataDirectory ) ,
tools/elf2efi: align columns in tables, unify formatting
For tables which represent binary data structures, readability is greatly
enhanced if the part which shows field size and type is aligned. This follows
the usual style for tables in the rest of the systemd codebase.
Also, use the same style for functions: if the function signature is too long
to fit in one line, put each parameter on a separate line.
Also, for comprehension expressions, if they are split, use the usual Python
style.
Also, drop format annotations, since the code isn't automatically formatted
anymore, and automatic formatting is neither feasible nor a goal for the
systemd codebase.
2024-03-13 15:16:17 +03:00
( " CLRRuntimeHeader " , PeDataDirectory ) ,
( " Reserved " , PeDataDirectory ) ,
2023-02-27 18:54:48 +03:00
)
class PeOptionalHeader ( LittleEndianStructure ) :
pass
class PeOptionalHeader32 ( PeOptionalHeader ) :
_anonymous_ = ( " Start " , " Middle " , " End " )
_fields_ = (
tools/elf2efi: align columns in tables, unify formatting
For tables which represent binary data structures, readability is greatly
enhanced if the part which shows field size and type is aligned. This follows
the usual style for tables in the rest of the systemd codebase.
Also, use the same style for functions: if the function signature is too long
to fit in one line, put each parameter on a separate line.
Also, for comprehension expressions, if they are split, use the usual Python
style.
Also, drop format annotations, since the code isn't automatically formatted
anymore, and automatic formatting is neither feasible nor a goal for the
systemd codebase.
2024-03-13 15:16:17 +03:00
( " Start " , PeOptionalHeaderStart ) ,
( " BaseOfData " , c_uint32 ) ,
( " ImageBase " , c_uint32 ) ,
( " Middle " , PeOptionalHeaderMiddle ) ,
2023-02-27 18:54:48 +03:00
( " SizeOfStackReserve " , c_uint32 ) ,
tools/elf2efi: align columns in tables, unify formatting
For tables which represent binary data structures, readability is greatly
enhanced if the part which shows field size and type is aligned. This follows
the usual style for tables in the rest of the systemd codebase.
Also, use the same style for functions: if the function signature is too long
to fit in one line, put each parameter on a separate line.
Also, for comprehension expressions, if they are split, use the usual Python
style.
Also, drop format annotations, since the code isn't automatically formatted
anymore, and automatic formatting is neither feasible nor a goal for the
systemd codebase.
2024-03-13 15:16:17 +03:00
( " SizeOfStackCommit " , c_uint32 ) ,
( " SizeOfHeapReserve " , c_uint32 ) ,
( " SizeOfHeapCommit " , c_uint32 ) ,
( " End " , PeOptionalHeaderEnd ) ,
2023-02-27 18:54:48 +03:00
)
class PeOptionalHeader32Plus ( PeOptionalHeader ) :
_anonymous_ = ( " Start " , " Middle " , " End " )
_fields_ = (
tools/elf2efi: align columns in tables, unify formatting
For tables which represent binary data structures, readability is greatly
enhanced if the part which shows field size and type is aligned. This follows
the usual style for tables in the rest of the systemd codebase.
Also, use the same style for functions: if the function signature is too long
to fit in one line, put each parameter on a separate line.
Also, for comprehension expressions, if they are split, use the usual Python
style.
Also, drop format annotations, since the code isn't automatically formatted
anymore, and automatic formatting is neither feasible nor a goal for the
systemd codebase.
2024-03-13 15:16:17 +03:00
( " Start " , PeOptionalHeaderStart ) ,
( " ImageBase " , c_uint64 ) ,
( " Middle " , PeOptionalHeaderMiddle ) ,
2023-02-27 18:54:48 +03:00
( " SizeOfStackReserve " , c_uint64 ) ,
tools/elf2efi: align columns in tables, unify formatting
For tables which represent binary data structures, readability is greatly
enhanced if the part which shows field size and type is aligned. This follows
the usual style for tables in the rest of the systemd codebase.
Also, use the same style for functions: if the function signature is too long
to fit in one line, put each parameter on a separate line.
Also, for comprehension expressions, if they are split, use the usual Python
style.
Also, drop format annotations, since the code isn't automatically formatted
anymore, and automatic formatting is neither feasible nor a goal for the
systemd codebase.
2024-03-13 15:16:17 +03:00
( " SizeOfStackCommit " , c_uint64 ) ,
( " SizeOfHeapReserve " , c_uint64 ) ,
( " SizeOfHeapCommit " , c_uint64 ) ,
( " End " , PeOptionalHeaderEnd ) ,
2023-02-27 18:54:48 +03:00
)
class PeSection ( LittleEndianStructure ) :
_fields_ = (
tools/elf2efi: align columns in tables, unify formatting
For tables which represent binary data structures, readability is greatly
enhanced if the part which shows field size and type is aligned. This follows
the usual style for tables in the rest of the systemd codebase.
Also, use the same style for functions: if the function signature is too long
to fit in one line, put each parameter on a separate line.
Also, for comprehension expressions, if they are split, use the usual Python
style.
Also, drop format annotations, since the code isn't automatically formatted
anymore, and automatic formatting is neither feasible nor a goal for the
systemd codebase.
2024-03-13 15:16:17 +03:00
( " Name " , c_char * 8 ) ,
( " VirtualSize " , c_uint32 ) ,
( " VirtualAddress " , c_uint32 ) ,
( " SizeOfRawData " , c_uint32 ) ,
( " PointerToRawData " , c_uint32 ) ,
2023-02-27 18:54:48 +03:00
( " PointerToRelocations " , c_uint32 ) ,
( " PointerToLinenumbers " , c_uint32 ) ,
tools/elf2efi: align columns in tables, unify formatting
For tables which represent binary data structures, readability is greatly
enhanced if the part which shows field size and type is aligned. This follows
the usual style for tables in the rest of the systemd codebase.
Also, use the same style for functions: if the function signature is too long
to fit in one line, put each parameter on a separate line.
Also, for comprehension expressions, if they are split, use the usual Python
style.
Also, drop format annotations, since the code isn't automatically formatted
anymore, and automatic formatting is neither feasible nor a goal for the
systemd codebase.
2024-03-13 15:16:17 +03:00
( " NumberOfRelocations " , c_uint16 ) ,
( " NumberOfLinenumbers " , c_uint16 ) ,
( " Characteristics " , c_uint32 ) ,
2023-02-27 18:54:48 +03:00
)
def __init__ ( self ) :
super ( ) . __init__ ( )
self . data = bytearray ( )
N_DATA_DIRECTORY_ENTRIES = 16
assert sizeof ( PeSection ) == 40
assert sizeof ( PeCoffHeader ) == 20
assert sizeof ( PeOptionalHeader32 ) == 224
assert sizeof ( PeOptionalHeader32Plus ) == 240
2023-09-28 17:09:42 +03:00
PE_CHARACTERISTICS_RX = 0x60000020 # CNT_CODE|MEM_READ|MEM_EXECUTE
PE_CHARACTERISTICS_RW = 0xC0000040 # CNT_INITIALIZED_DATA|MEM_READ|MEM_WRITE
tools/elf2efi: align columns in tables, unify formatting
For tables which represent binary data structures, readability is greatly
enhanced if the part which shows field size and type is aligned. This follows
the usual style for tables in the rest of the systemd codebase.
Also, use the same style for functions: if the function signature is too long
to fit in one line, put each parameter on a separate line.
Also, for comprehension expressions, if they are split, use the usual Python
style.
Also, drop format annotations, since the code isn't automatically formatted
anymore, and automatic formatting is neither feasible nor a goal for the
systemd codebase.
2024-03-13 15:16:17 +03:00
PE_CHARACTERISTICS_R = 0x40000040 # CNT_INITIALIZED_DATA|MEM_READ
2023-09-28 17:09:42 +03:00
IGNORE_SECTIONS = [
" .eh_frame " ,
" .eh_frame_hdr " ,
" .ARM.exidx " ,
]
IGNORE_SECTION_TYPES = [
" SHT_DYNAMIC " ,
" SHT_DYNSYM " ,
" SHT_GNU_ATTRIBUTES " ,
" SHT_GNU_HASH " ,
" SHT_HASH " ,
" SHT_NOTE " ,
" SHT_REL " ,
" SHT_RELA " ,
" SHT_RELR " ,
" SHT_STRTAB " ,
" SHT_SYMTAB " ,
]
2023-02-27 18:54:48 +03:00
# EFI mandates 4KiB memory pages.
SECTION_ALIGNMENT = 4096
FILE_ALIGNMENT = 512
# Nobody cares about DOS headers, so put the PE header right after.
PE_OFFSET = 64
2023-07-30 21:59:04 +03:00
PE_MAGIC = b " PE \0 \0 "
2023-02-27 18:54:48 +03:00
def align_to ( x : int , align : int ) - > int :
return ( x + align - 1 ) & ~ ( align - 1 )
2023-09-28 17:09:42 +03:00
def align_down ( x : int , align : int ) - > int :
return x & ~ ( align - 1 )
2023-02-27 18:54:48 +03:00
2023-09-22 13:13:09 +03:00
def next_section_address ( sections : typing . List [ PeSection ] ) - > int :
tools/elf2efi: align columns in tables, unify formatting
For tables which represent binary data structures, readability is greatly
enhanced if the part which shows field size and type is aligned. This follows
the usual style for tables in the rest of the systemd codebase.
Also, use the same style for functions: if the function signature is too long
to fit in one line, put each parameter on a separate line.
Also, for comprehension expressions, if they are split, use the usual Python
style.
Also, drop format annotations, since the code isn't automatically formatted
anymore, and automatic formatting is neither feasible nor a goal for the
systemd codebase.
2024-03-13 15:16:17 +03:00
return align_to ( sections [ - 1 ] . VirtualAddress + sections [ - 1 ] . VirtualSize ,
SECTION_ALIGNMENT )
2023-09-22 13:13:09 +03:00
2023-09-28 17:09:42 +03:00
def iter_copy_sections ( elf : ELFFile ) - > typing . Iterator [ PeSection ] :
pe_s = None
2023-02-27 18:54:48 +03:00
2023-09-28 17:09:42 +03:00
# This is essentially the same as copying by ELF load segments, except that we assemble them
# manually, so that we can easily strip unwanted sections. We try to only discard things we know
# about so that there are no surprises.
2023-02-27 18:54:48 +03:00
2023-09-28 18:54:35 +03:00
relro = None
for elf_seg in elf . iter_segments ( ) :
if elf_seg [ " p_type " ] == " PT_LOAD " and elf_seg [ " p_align " ] != SECTION_ALIGNMENT :
raise RuntimeError ( " ELF segments are not properly aligned. " )
elif elf_seg [ " p_type " ] == " PT_GNU_RELRO " :
relro = elf_seg
2023-09-28 17:09:42 +03:00
for elf_s in elf . iter_sections ( ) :
if (
elf_s [ " sh_flags " ] & SH_FLAGS . SHF_ALLOC == 0
or elf_s [ " sh_type " ] in IGNORE_SECTION_TYPES
or elf_s . name in IGNORE_SECTIONS
) :
continue
if elf_s [ " sh_type " ] not in [ " SHT_PROGBITS " , " SHT_NOBITS " ] :
raise RuntimeError ( f " Unknown section { elf_s . name } . " )
2023-02-27 18:54:48 +03:00
2023-09-28 17:09:42 +03:00
if elf_s [ " sh_flags " ] & SH_FLAGS . SHF_EXECINSTR :
rwx = PE_CHARACTERISTICS_RX
elif elf_s [ " sh_flags " ] & SH_FLAGS . SHF_WRITE :
rwx = PE_CHARACTERISTICS_RW
else :
rwx = PE_CHARACTERISTICS_R
2023-02-27 18:54:48 +03:00
2023-09-28 18:54:35 +03:00
# PE images are always relro.
if relro and relro . section_in_segment ( elf_s ) :
rwx = PE_CHARACTERISTICS_R
2023-09-28 17:09:42 +03:00
if pe_s and pe_s . Characteristics != rwx :
yield pe_s
pe_s = None
2023-02-27 18:54:48 +03:00
2023-09-28 17:09:42 +03:00
if pe_s :
# Insert padding to properly align the section.
pad_len = elf_s [ " sh_addr " ] - pe_s . VirtualAddress - len ( pe_s . data )
pe_s . data + = bytearray ( pad_len ) + elf_s . data ( )
else :
pe_s = PeSection ( )
pe_s . VirtualAddress = elf_s [ " sh_addr " ]
pe_s . Characteristics = rwx
pe_s . data = elf_s . data ( )
2023-02-27 18:54:48 +03:00
2023-09-28 17:09:42 +03:00
if pe_s :
yield pe_s
2023-02-27 18:54:48 +03:00
2023-09-28 17:09:42 +03:00
def convert_sections ( elf : ELFFile , opt : PeOptionalHeader ) - > typing . List [ PeSection ] :
last_vma = 0
2023-02-27 18:54:48 +03:00
sections = [ ]
2023-09-28 17:09:42 +03:00
for pe_s in iter_copy_sections ( elf ) :
# Truncate the VMA to the nearest page and insert appropriate padding. This should not
# cause any overlap as this is pretty much how ELF *segments* are loaded/mmapped anyways.
# The ELF sections inside should also be properly aligned as we reuse the ELF VMA layout
# for the PE image.
vma = pe_s . VirtualAddress
pe_s . VirtualAddress = align_down ( vma , SECTION_ALIGNMENT )
pe_s . data = bytearray ( vma - pe_s . VirtualAddress ) + pe_s . data
pe_s . VirtualSize = len ( pe_s . data )
pe_s . SizeOfRawData = align_to ( len ( pe_s . data ) , FILE_ALIGNMENT )
pe_s . Name = {
PE_CHARACTERISTICS_RX : b " .text " ,
PE_CHARACTERISTICS_RW : b " .data " ,
PE_CHARACTERISTICS_R : b " .rodata " ,
} [ pe_s . Characteristics ]
# This can happen if not building with `-z separate-code`.
if pe_s . VirtualAddress < last_vma :
raise RuntimeError ( " Overlapping PE sections. " )
last_vma = pe_s . VirtualAddress + pe_s . VirtualSize
2023-02-27 18:54:48 +03:00
if pe_s . Name == b " .text " :
opt . BaseOfCode = pe_s . VirtualAddress
opt . SizeOfCode + = pe_s . VirtualSize
else :
opt . SizeOfInitializedData + = pe_s . VirtualSize
if pe_s . Name == b " .data " and isinstance ( opt , PeOptionalHeader32 ) :
opt . BaseOfData = pe_s . VirtualAddress
sections . append ( pe_s )
return sections
2023-09-22 13:15:55 +03:00
def copy_sections (
elf : ELFFile ,
opt : PeOptionalHeader ,
input_names : str ,
sections : typing . List [ PeSection ] ,
) :
for name in input_names . split ( " , " ) :
elf_s = elf . get_section_by_name ( name )
if not elf_s :
continue
if elf_s . data_alignment > 1 and SECTION_ALIGNMENT % elf_s . data_alignment != 0 :
raise RuntimeError ( f " ELF section { name } is not aligned. " )
if elf_s [ " sh_flags " ] & ( SH_FLAGS . SHF_EXECINSTR | SH_FLAGS . SHF_WRITE ) != 0 :
raise RuntimeError ( f " ELF section { name } is not read-only data. " )
pe_s = PeSection ( )
pe_s . Name = name . encode ( )
pe_s . data = elf_s . data ( )
pe_s . VirtualAddress = next_section_address ( sections )
pe_s . VirtualSize = len ( elf_s . data ( ) )
pe_s . SizeOfRawData = align_to ( len ( elf_s . data ( ) ) , FILE_ALIGNMENT )
pe_s . Characteristics = PE_CHARACTERISTICS_R
opt . SizeOfInitializedData + = pe_s . VirtualSize
sections . append ( pe_s )
2023-02-27 18:54:48 +03:00
def apply_elf_relative_relocation (
2023-07-30 21:59:04 +03:00
reloc : ElfRelocation ,
image_base : int ,
sections : typing . List [ PeSection ] ,
addend_size : int ,
2023-02-27 18:54:48 +03:00
) :
tools/elf2efi: align columns in tables, unify formatting
For tables which represent binary data structures, readability is greatly
enhanced if the part which shows field size and type is aligned. This follows
the usual style for tables in the rest of the systemd codebase.
Also, use the same style for functions: if the function signature is too long
to fit in one line, put each parameter on a separate line.
Also, for comprehension expressions, if they are split, use the usual Python
style.
Also, drop format annotations, since the code isn't automatically formatted
anymore, and automatic formatting is neither feasible nor a goal for the
systemd codebase.
2024-03-13 15:16:17 +03:00
[ target ] = [ pe_s for pe_s in sections
if pe_s . VirtualAddress < = reloc [ " r_offset " ] < pe_s . VirtualAddress + len ( pe_s . data ) ]
2023-02-27 18:54:48 +03:00
addend_offset = reloc [ " r_offset " ] - target . VirtualAddress
if reloc . is_RELA ( ) :
addend = reloc [ " r_addend " ]
else :
addend = target . data [ addend_offset : addend_offset + addend_size ]
addend = int . from_bytes ( addend , byteorder = " little " )
value = ( image_base + addend ) . to_bytes ( addend_size , byteorder = " little " )
target . data [ addend_offset : addend_offset + addend_size ] = value
def convert_elf_reloc_table (
elf : ELFFile ,
elf_reloc_table : ElfRelocationTable ,
2023-09-28 17:09:42 +03:00
elf_image_base : int ,
2023-07-14 15:16:44 +03:00
sections : typing . List [ PeSection ] ,
pe_reloc_blocks : typing . Dict [ int , PeRelocationBlock ] ,
2023-02-27 18:54:48 +03:00
) :
NONE_RELOC = {
" EM_386 " : ENUM_RELOC_TYPE_i386 [ " R_386_NONE " ] ,
" EM_AARCH64 " : ENUM_RELOC_TYPE_AARCH64 [ " R_AARCH64_NONE " ] ,
" EM_ARM " : ENUM_RELOC_TYPE_ARM [ " R_ARM_NONE " ] ,
2023-02-28 20:05:18 +03:00
" EM_LOONGARCH " : 0 ,
2023-02-27 18:54:48 +03:00
" EM_RISCV " : 0 ,
" EM_X86_64 " : ENUM_RELOC_TYPE_x64 [ " R_X86_64_NONE " ] ,
} [ elf [ " e_machine " ] ]
RELATIVE_RELOC = {
" EM_386 " : ENUM_RELOC_TYPE_i386 [ " R_386_RELATIVE " ] ,
" EM_AARCH64 " : ENUM_RELOC_TYPE_AARCH64 [ " R_AARCH64_RELATIVE " ] ,
" EM_ARM " : ENUM_RELOC_TYPE_ARM [ " R_ARM_RELATIVE " ] ,
2023-02-28 20:05:18 +03:00
" EM_LOONGARCH " : 3 ,
2023-02-27 18:54:48 +03:00
" EM_RISCV " : 3 ,
" EM_X86_64 " : ENUM_RELOC_TYPE_x64 [ " R_X86_64_RELATIVE " ] ,
} [ elf [ " e_machine " ] ]
for reloc in elf_reloc_table . iter_relocations ( ) :
if reloc [ " r_info_type " ] == NONE_RELOC :
continue
if reloc [ " r_info_type " ] == RELATIVE_RELOC :
tools/elf2efi: align columns in tables, unify formatting
For tables which represent binary data structures, readability is greatly
enhanced if the part which shows field size and type is aligned. This follows
the usual style for tables in the rest of the systemd codebase.
Also, use the same style for functions: if the function signature is too long
to fit in one line, put each parameter on a separate line.
Also, for comprehension expressions, if they are split, use the usual Python
style.
Also, drop format annotations, since the code isn't automatically formatted
anymore, and automatic formatting is neither feasible nor a goal for the
systemd codebase.
2024-03-13 15:16:17 +03:00
apply_elf_relative_relocation ( reloc ,
elf_image_base ,
sections ,
elf . elfclass / / 8 )
2023-02-27 18:54:48 +03:00
# Now that the ELF relocation has been applied, we can create a PE relocation.
block_rva = reloc [ " r_offset " ] & ~ 0xFFF
if block_rva not in pe_reloc_blocks :
pe_reloc_blocks [ block_rva ] = PeRelocationBlock ( block_rva )
entry = PeRelocationEntry ( )
entry . Offset = reloc [ " r_offset " ] & 0xFFF
# REL_BASED_HIGHLOW or REL_BASED_DIR64
entry . Type = 3 if elf . elfclass == 32 else 10
pe_reloc_blocks [ block_rva ] . entries . append ( entry )
continue
raise RuntimeError ( f " Unsupported relocation { reloc } " )
def convert_elf_relocations (
2023-09-28 17:09:42 +03:00
elf : ELFFile ,
opt : PeOptionalHeader ,
sections : typing . List [ PeSection ] ,
minimum_sections : int ,
2023-05-23 20:00:52 +03:00
) - > typing . Optional [ PeSection ] :
2023-02-27 18:54:48 +03:00
dynamic = elf . get_section_by_name ( " .dynamic " )
if dynamic is None :
raise RuntimeError ( " ELF .dynamic section is missing. " )
[ flags_tag ] = dynamic . iter_tags ( " DT_FLAGS_1 " )
if not flags_tag [ " d_val " ] & ENUM_DT_FLAGS_1 [ " DF_1_PIE " ] :
raise RuntimeError ( " ELF file is not a PIE. " )
2023-09-28 17:22:13 +03:00
# This checks that the ELF image base is 0.
symtab = elf . get_section_by_name ( " .symtab " )
if symtab :
exe_start = symtab . get_symbol_by_name ( " __executable_start " )
if exe_start and exe_start [ 0 ] [ " st_value " ] != 0 :
raise RuntimeError ( " Unexpected ELF image base. " )
tools/elf2efi: align columns in tables, unify formatting
For tables which represent binary data structures, readability is greatly
enhanced if the part which shows field size and type is aligned. This follows
the usual style for tables in the rest of the systemd codebase.
Also, use the same style for functions: if the function signature is too long
to fit in one line, put each parameter on a separate line.
Also, for comprehension expressions, if they are split, use the usual Python
style.
Also, drop format annotations, since the code isn't automatically formatted
anymore, and automatic formatting is neither feasible nor a goal for the
systemd codebase.
2024-03-13 15:16:17 +03:00
opt . SizeOfHeaders = align_to ( PE_OFFSET
+ len ( PE_MAGIC )
+ sizeof ( PeCoffHeader )
+ sizeof ( opt )
+ sizeof ( PeSection ) * max ( len ( sections ) + 1 , minimum_sections ) ,
FILE_ALIGNMENT )
2023-09-28 17:09:42 +03:00
# We use the basic VMA layout from the ELF image in the PE image. This could cause the first
# section to overlap the PE image headers during runtime at VMA 0. We can simply apply a fixed
# offset relative to the PE image base when applying/converting ELF relocations. Afterwards we
# just have to apply the offset to the PE addresses so that the PE relocations work correctly on
# the ELF portions of the image.
segment_offset = 0
if sections [ 0 ] . VirtualAddress < opt . SizeOfHeaders :
tools/elf2efi: align columns in tables, unify formatting
For tables which represent binary data structures, readability is greatly
enhanced if the part which shows field size and type is aligned. This follows
the usual style for tables in the rest of the systemd codebase.
Also, use the same style for functions: if the function signature is too long
to fit in one line, put each parameter on a separate line.
Also, for comprehension expressions, if they are split, use the usual Python
style.
Also, drop format annotations, since the code isn't automatically formatted
anymore, and automatic formatting is neither feasible nor a goal for the
systemd codebase.
2024-03-13 15:16:17 +03:00
segment_offset = align_to ( opt . SizeOfHeaders - sections [ 0 ] . VirtualAddress ,
SECTION_ALIGNMENT )
2023-09-28 17:09:42 +03:00
opt . AddressOfEntryPoint = elf [ " e_entry " ] + segment_offset
opt . BaseOfCode + = segment_offset
if isinstance ( opt , PeOptionalHeader32 ) :
opt . BaseOfData + = segment_offset
2023-07-14 15:16:44 +03:00
pe_reloc_blocks : typing . Dict [ int , PeRelocationBlock ] = { }
2023-02-27 18:54:48 +03:00
for reloc_type , reloc_table in dynamic . get_relocation_tables ( ) . items ( ) :
if reloc_type not in [ " REL " , " RELA " ] :
raise RuntimeError ( " Unsupported relocation type {elf_reloc_type} . " )
tools/elf2efi: align columns in tables, unify formatting
For tables which represent binary data structures, readability is greatly
enhanced if the part which shows field size and type is aligned. This follows
the usual style for tables in the rest of the systemd codebase.
Also, use the same style for functions: if the function signature is too long
to fit in one line, put each parameter on a separate line.
Also, for comprehension expressions, if they are split, use the usual Python
style.
Also, drop format annotations, since the code isn't automatically formatted
anymore, and automatic formatting is neither feasible nor a goal for the
systemd codebase.
2024-03-13 15:16:17 +03:00
convert_elf_reloc_table ( elf ,
reloc_table ,
opt . ImageBase + segment_offset ,
sections ,
pe_reloc_blocks )
2023-02-27 18:54:48 +03:00
2023-09-28 17:09:42 +03:00
for pe_s in sections :
pe_s . VirtualAddress + = segment_offset
2023-05-23 20:00:52 +03:00
if len ( pe_reloc_blocks ) == 0 :
return None
2023-02-27 18:54:48 +03:00
data = bytearray ( )
for rva in sorted ( pe_reloc_blocks ) :
block = pe_reloc_blocks [ rva ]
n_relocs = len ( block . entries )
# Each block must start on a 32-bit boundary. Because each entry is 16 bits
# the len has to be even. We pad by adding a none relocation.
if n_relocs % 2 != 0 :
n_relocs + = 1
block . entries . append ( PeRelocationEntry ( ) )
2023-09-28 17:09:42 +03:00
block . PageRVA + = segment_offset
tools/elf2efi: align columns in tables, unify formatting
For tables which represent binary data structures, readability is greatly
enhanced if the part which shows field size and type is aligned. This follows
the usual style for tables in the rest of the systemd codebase.
Also, use the same style for functions: if the function signature is too long
to fit in one line, put each parameter on a separate line.
Also, for comprehension expressions, if they are split, use the usual Python
style.
Also, drop format annotations, since the code isn't automatically formatted
anymore, and automatic formatting is neither feasible nor a goal for the
systemd codebase.
2024-03-13 15:16:17 +03:00
block . BlockSize = sizeof ( PeRelocationBlock ) + sizeof ( PeRelocationEntry ) * n_relocs
2023-02-27 18:54:48 +03:00
data + = block
for entry in sorted ( block . entries , key = lambda e : e . Offset ) :
data + = entry
pe_reloc_s = PeSection ( )
pe_reloc_s . Name = b " .reloc "
pe_reloc_s . data = data
2023-09-22 13:13:09 +03:00
pe_reloc_s . VirtualAddress = next_section_address ( sections )
2023-02-27 18:54:48 +03:00
pe_reloc_s . VirtualSize = len ( data )
pe_reloc_s . SizeOfRawData = align_to ( len ( data ) , FILE_ALIGNMENT )
# CNT_INITIALIZED_DATA|MEM_READ|MEM_DISCARDABLE
pe_reloc_s . Characteristics = 0x42000040
sections . append ( pe_reloc_s )
opt . SizeOfInitializedData + = pe_reloc_s . VirtualSize
return pe_reloc_s
def write_pe (
tools/elf2efi: align columns in tables, unify formatting
For tables which represent binary data structures, readability is greatly
enhanced if the part which shows field size and type is aligned. This follows
the usual style for tables in the rest of the systemd codebase.
Also, use the same style for functions: if the function signature is too long
to fit in one line, put each parameter on a separate line.
Also, for comprehension expressions, if they are split, use the usual Python
style.
Also, drop format annotations, since the code isn't automatically formatted
anymore, and automatic formatting is neither feasible nor a goal for the
systemd codebase.
2024-03-13 15:16:17 +03:00
file ,
coff : PeCoffHeader ,
opt : PeOptionalHeader ,
sections : typing . List [ PeSection ] ,
2023-02-27 18:54:48 +03:00
) :
file . write ( b " MZ " )
file . seek ( 0x3C , io . SEEK_SET )
file . write ( PE_OFFSET . to_bytes ( 2 , byteorder = " little " ) )
file . seek ( PE_OFFSET , io . SEEK_SET )
2023-07-30 21:59:04 +03:00
file . write ( PE_MAGIC )
2023-02-27 18:54:48 +03:00
file . write ( coff )
file . write ( opt )
offset = opt . SizeOfHeaders
for pe_s in sorted ( sections , key = lambda s : s . VirtualAddress ) :
if pe_s . VirtualAddress < opt . SizeOfHeaders :
raise RuntimeError ( f " Section { pe_s . Name } overlapping PE headers. " )
pe_s . PointerToRawData = offset
file . write ( pe_s )
offset = align_to ( offset + len ( pe_s . data ) , FILE_ALIGNMENT )
2023-07-30 21:59:04 +03:00
assert file . tell ( ) < = opt . SizeOfHeaders
2023-02-27 18:54:48 +03:00
for pe_s in sections :
file . seek ( pe_s . PointerToRawData , io . SEEK_SET )
file . write ( pe_s . data )
file . truncate ( offset )
def elf2efi ( args : argparse . Namespace ) :
elf = ELFFile ( args . ELF )
if not elf . little_endian :
raise RuntimeError ( " ELF file is not little-endian. " )
if elf [ " e_type " ] not in [ " ET_DYN " , " ET_EXEC " ] :
raise RuntimeError ( " Unsupported ELF type. " )
pe_arch = {
" EM_386 " : 0x014C ,
" EM_AARCH64 " : 0xAA64 ,
" EM_ARM " : 0x01C2 ,
2023-02-28 20:05:18 +03:00
" EM_LOONGARCH " : 0x6232 if elf . elfclass == 32 else 0x6264 ,
" EM_RISCV " : 0x5032 if elf . elfclass == 32 else 0x5064 ,
2023-02-27 18:54:48 +03:00
" EM_X86_64 " : 0x8664 ,
} . get ( elf [ " e_machine " ] )
if pe_arch is None :
2023-09-21 20:06:55 +03:00
raise RuntimeError ( f " Unsupported ELF arch { elf [ ' e_machine ' ] } " )
2023-02-27 18:54:48 +03:00
coff = PeCoffHeader ( )
opt = PeOptionalHeader32 ( ) if elf . elfclass == 32 else PeOptionalHeader32Plus ( )
# We relocate to a unique image base to reduce the chances for runtime relocation to occur.
base_name = pathlib . Path ( args . PE . name ) . name . encode ( )
opt . ImageBase = int ( hashlib . sha1 ( base_name ) . hexdigest ( ) [ 0 : 8 ] , 16 )
if elf . elfclass == 32 :
opt . ImageBase = ( 0x400000 + opt . ImageBase ) & 0xFFFF0000
else :
opt . ImageBase = ( 0x100000000 + opt . ImageBase ) & 0x1FFFF0000
2023-09-28 17:09:42 +03:00
sections = convert_sections ( elf , opt )
2023-09-22 13:15:55 +03:00
copy_sections ( elf , opt , args . copy_sections , sections )
2023-09-28 17:09:42 +03:00
pe_reloc_s = convert_elf_relocations ( elf , opt , sections , args . minimum_sections )
2023-02-27 18:54:48 +03:00
coff . Machine = pe_arch
coff . NumberOfSections = len ( sections )
coff . TimeDateStamp = int ( os . environ . get ( " SOURCE_DATE_EPOCH " , time . time ( ) ) )
coff . SizeOfOptionalHeader = sizeof ( opt )
# EXECUTABLE_IMAGE|LINE_NUMS_STRIPPED|LOCAL_SYMS_STRIPPED|DEBUG_STRIPPED
# and (32BIT_MACHINE or LARGE_ADDRESS_AWARE)
coff . Characteristics = 0x30E if elf . elfclass == 32 else 0x22E
opt . SectionAlignment = SECTION_ALIGNMENT
opt . FileAlignment = FILE_ALIGNMENT
opt . MajorImageVersion = args . version_major
opt . MinorImageVersion = args . version_minor
opt . MajorSubsystemVersion = args . efi_major
opt . MinorSubsystemVersion = args . efi_minor
opt . Subsystem = args . subsystem
opt . Magic = 0x10B if elf . elfclass == 32 else 0x20B
2023-09-22 13:13:09 +03:00
opt . SizeOfImage = next_section_address ( sections )
2023-02-27 18:54:48 +03:00
# DYNAMIC_BASE|NX_COMPAT|HIGH_ENTROPY_VA or DYNAMIC_BASE|NX_COMPAT
opt . DllCharacteristics = 0x160 if elf . elfclass == 64 else 0x140
# These values are taken from a natively built PE binary (although, unused by EDK2/EFI).
opt . SizeOfStackReserve = 0x100000
opt . SizeOfStackCommit = 0x001000
opt . SizeOfHeapReserve = 0x100000
opt . SizeOfHeapCommit = 0x001000
opt . NumberOfRvaAndSizes = N_DATA_DIRECTORY_ENTRIES
2023-05-23 20:00:52 +03:00
if pe_reloc_s :
opt . BaseRelocationTable = PeDataDirectory (
pe_reloc_s . VirtualAddress , pe_reloc_s . VirtualSize
)
2023-02-27 18:54:48 +03:00
write_pe ( args . PE , coff , opt , sections )
2024-03-13 15:21:51 +03:00
def create_parser ( ) - > argparse . ArgumentParser :
2023-02-27 18:54:48 +03:00
parser = argparse . ArgumentParser ( description = " Convert ELF binaries to PE/EFI " )
parser . add_argument (
" --version-major " ,
type = int ,
default = 0 ,
help = " Major image version of EFI image " ,
)
parser . add_argument (
" --version-minor " ,
type = int ,
default = 0 ,
help = " Minor image version of EFI image " ,
)
parser . add_argument (
" --efi-major " ,
type = int ,
default = 0 ,
help = " Minimum major EFI subsystem version " ,
)
parser . add_argument (
" --efi-minor " ,
type = int ,
default = 0 ,
help = " Minimum minor EFI subsystem version " ,
)
parser . add_argument (
" --subsystem " ,
type = int ,
default = 10 ,
help = " PE subsystem " ,
)
parser . add_argument (
" ELF " ,
type = argparse . FileType ( " rb " ) ,
help = " Input ELF file " ,
)
parser . add_argument (
" PE " ,
type = argparse . FileType ( " wb " ) ,
help = " Output PE/EFI file " ,
)
2023-05-23 03:43:59 +03:00
parser . add_argument (
" --minimum-sections " ,
type = int ,
default = 0 ,
help = " Minimum number of sections to leave space for " ,
)
2023-09-22 13:15:55 +03:00
parser . add_argument (
" --copy-sections " ,
type = str ,
default = " " ,
help = " Copy these sections if found " ,
)
2024-03-13 15:21:51 +03:00
return parser
2023-02-27 18:54:48 +03:00
2024-03-13 15:21:51 +03:00
def main ( ) :
parser = create_parser ( )
2023-02-27 18:54:48 +03:00
elf2efi ( parser . parse_args ( ) )
if __name__ == " __main__ " :
main ( )