Clean up unused tools

This commit is contained in:
Luke Street 2022-07-18 20:30:39 -04:00
parent 10e847fd07
commit 6acf3a1278
9 changed files with 2 additions and 1629 deletions

View File

@ -1,3 +0,0 @@
#!/bin/bash -e
python tools/deincbin.py "$1" > "$1.deincbin.s"
mv "$1.deincbin.s" "$1"

View File

@ -3,14 +3,10 @@ CFLAGS := -O3 -Wall -s
default: all
all: elf2dol elf2rel
all: elf2dol
elf2dol: elf2dol.c
$(CC) $(CFLAGS) -o $@ $^
elf2rel: elf2rel.c
$(CC) $(CFLAGS) -o $@ $^
clean:
$(RM) elf2dol
$(RM) elf2rel

View File

@ -1,8 +1,7 @@
#!/bin/bash -e
OBJDUMP="$DEVKITPPC/bin/powerpc-eabi-objdump -D -bbinary -EB -mpowerpc -M gekko"
if [ ! -z "$1" ]; then
OPTIONS="--start-address=$(($1)) --stop-address=$(($2))"
OPTIONS="--start-address=$(($1)) --stop-address=$(($2))"
fi
$OBJDUMP $OPTIONS baserom.dol > baserom.dump
$OBJDUMP $OPTIONS build/mp1.0/main.dol > main.dump

View File

@ -1,120 +0,0 @@
#!/usr/bin/env python3
#
# Usage: dump_common_data.py file.s
# Dumps all incbin data and prints the revised file to stdout.
import os
import re
import sys
# Reads a bytearray from baserom.dol
def read_baserom(start, size):
with open('baserom.dol', 'rb') as f:
f.seek(start, os.SEEK_SET)
return bytearray(f.read(size))
if len(sys.argv) != 2:
print('Usage: %s ASM_FILE' % sys.argv[0])
exit()
# reads a 32-bit big endian value starting at pos
def read_u32(data, pos):
return (data[pos]<<24) | (data[pos+1]<<16) | (data[pos+2]<<8) | (data[pos+3])
def is_ascii(code):
if code >= 0x20 and code <= 0x7E: # normal characters
return True
if code in [0x09, 0x0A]: # tab, newline
return True
return False
# reads a string starting at pos
def read_string(data, pos):
text = ''
while pos < len(data) and is_ascii(data[pos]):
text += chr(data[pos])
pos += 1
if pos < len(data) and data[pos] == 0:
return text
return ''
# escapes special characters in the string for use in a C string literal
def escape_string(text):
return text.replace('\\','\\\\').replace('"','\\"').replace('\n','\\n').replace('\t','\\t')
# returns True if value is 4-byte aligned
def is_aligned(num):
return num % 4 == 0
# returns True if value is a possible pointer
def is_pointer(num):
return num >= 0x80003100 and num <= 0x802F6C80
# returns True if all elements are zero
def is_all_zero(arr):
for val in arr:
if val != 0:
return False
return True
# returns string of comma-separated hex bytes
def hex_bytes(data):
return ', '.join('0x%02X' % n for n in data)
def convert_data(data, offset):
text = ''
size = len(data)
pos = 0
while pos < size:
# pad unaligned
pad = []
while not is_aligned(offset + pos) and pos < size:
pad.append(data[pos])
pos += 1
if len(pad) > 0:
if is_all_zero(pad):
text += '\t.balign 4\n'
else:
text += '\t.byte %s\n' % hex_bytes(pad)
# string?
string = read_string(data, pos)
if len(string) > 3:
text += '\t.asciz "%s"\n' % escape_string(string)
pos += len(string) + 1
continue
assert(is_aligned(offset + pos))
if pos + 4 <= size:
val = read_u32(data, pos)
if is_pointer(val):
text += '\t.4byte 0x%08X ;# ptr\n' % val
elif val == 0:
text += '\t.4byte 0\n'
else:
text += '\t.4byte 0x%08X\n' % val
pos += 4
return text
currSection = ''
with open(sys.argv[1], 'rt') as f:
for line in f.readlines():
line = line.rstrip()
# Section directive
m = re.match(r'\s*\.section\s+([\._A-Za-z0-9]+)', line)
if m:
currSection = m.groups()[0]
elif currSection in ['.rodata', '.data', '.sdata', '.sdata2', '.ctors', '.dtors', 'extab_', 'extabindex_']:
# Incbin directive
m = re.match(r'\s*\.incbin\s+"baserom.dol"\s*,\s*([^,]+),\s*([^,]+)', line)
if m:
g = m.groups()
start = int(g[0], 0)
size = int(g[1], 0)
data = read_baserom(start, size)
print('\t# ROM: 0x%X' % start)
print(convert_data(data, start))
continue
print(line)

View File

@ -1,836 +0,0 @@
#include <assert.h>
#include <errno.h>
#include <stdarg.h>
#include <stdint.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#ifndef MAX
#define MAX(a, b) (((a) > (b)) ? (a) : (b))
#endif
#ifndef MIN
#define MIN(a, b) (((a) < (b)) ? (a) : (b))
#endif
#define ARRAY_COUNT(arr) (sizeof(arr) / sizeof((arr)[0]))
#define IS_LITTLE_ENDIAN (((char*)((uint32_t[]){1}))[0] == 1)
// Relevant portions of elf.h
typedef uint32_t Elf32_Addr;
typedef uint32_t Elf32_Off;
typedef uint32_t Elf32_Word;
typedef int32_t Elf32_Sword;
typedef uint16_t Elf32_Half;
typedef uint16_t Elf32_Section;
#define EI_NIDENT (16)
typedef struct {
unsigned char e_ident[EI_NIDENT]; /* Magic number and other info */
Elf32_Half e_type; /* Object file type */
Elf32_Half e_machine; /* Architecture */
Elf32_Word e_version; /* Object file version */
Elf32_Addr e_entry; /* Entry point virtual address */
Elf32_Off e_phoff; /* Program header table file offset */
Elf32_Off e_shoff; /* Section header table file offset */
Elf32_Word e_flags; /* Processor-specific flags */
Elf32_Half e_ehsize; /* ELF header size in bytes */
Elf32_Half e_phentsize; /* Program header table entry size */
Elf32_Half e_phnum; /* Program header table entry count */
Elf32_Half e_shentsize; /* Section header table entry size */
Elf32_Half e_shnum; /* Section header table entry count */
Elf32_Half e_shstrndx; /* Section header string table index */
} Elf32_Ehdr;
#define EI_CLASS 4 /* File class byte index */
#define ELFCLASS32 1 /* 32-bit objects */
#define EI_DATA 5 /* Data encoding byte index */
#define ELFDATA2MSB 2 /* 2's complement, big endian */
#define EM_PPC 20 /* PowerPC */
typedef struct {
Elf32_Word sh_name; /* Section name (string tbl index) */
Elf32_Word sh_type; /* Section type */
Elf32_Word sh_flags; /* Section flags */
Elf32_Addr sh_addr; /* Section virtual addr at execution */
Elf32_Off sh_offset; /* Section file offset */
Elf32_Word sh_size; /* Section size in bytes */
Elf32_Word sh_link; /* Link to another section */
Elf32_Word sh_info; /* Additional section information */
Elf32_Word sh_addralign; /* Section alignment */
Elf32_Word sh_entsize; /* Entry size if section holds table */
} Elf32_Shdr;
#define SHT_PROGBITS 1 /* Program data */
#define SHT_SYMTAB 2 /* Symbol table */
#define SHT_STRTAB 3 /* String table */
#define SHT_RELA 4 /* Relocation entries with addends */
#define SHT_NOBITS 8 /* Program space with no data (bss) */
#define SHF_ALLOC (1 << 1) /* Occupies memory during execution */
#define SHF_EXECINSTR (1 << 2) /* Executable */
typedef struct {
Elf32_Word st_name; /* Symbol name (string tbl index) */
Elf32_Addr st_value; /* Symbol value */
Elf32_Word st_size; /* Symbol size */
unsigned char st_info; /* Symbol type and binding */
unsigned char st_other; /* Symbol visibility */
Elf32_Section st_shndx; /* Section index */
} Elf32_Sym;
#define ELF32_ST_TYPE(val) ((val)&0xf)
/* Legal values for ST_TYPE subfield of st_info (symbol type). */
#define STT_NOTYPE 0 /* Symbol type is unspecified */
#define STT_FUNC 2 /* Symbol is a code object */
typedef struct {
Elf32_Addr r_offset; /* Address */
Elf32_Word r_info; /* Relocation type and symbol index */
Elf32_Sword r_addend; /* Addend */
} Elf32_Rela;
/* How to extract and insert information held in the r_info field. */
#define ELF32_R_SYM(val) ((val) >> 8)
#define ELF32_R_TYPE(val) ((val)&0xff)
#define R_PPC_NONE 0
#define R_PPC_ADDR32 1 /* 32bit absolute address */
#define R_PPC_ADDR24 2 /* 26bit address, 2 bits ignored. */
#define R_PPC_ADDR16 3 /* 16bit absolute address */
#define R_PPC_ADDR16_LO 4 /* lower 16bit of absolute address */
#define R_PPC_ADDR16_HI 5 /* high 16bit of absolute address */
#define R_PPC_ADDR16_HA 6 /* adjusted high 16bit */
#define R_PPC_ADDR14 7 /* 16bit address, 2 bits ignored */
#define R_PPC_ADDR14_BRTAKEN 8
#define R_PPC_ADDR14_BRNTAKEN 9
#define R_PPC_REL24 10 /* PC relative 26 bit */
#define R_PPC_REL14 11 /* PC relative 16 bit */
#define R_DOLPHIN_SECTION 202
#define R_DOLPHIN_END 203
// end elf.h
struct RelHeader {
uint32_t moduleId; // unique module identifier
uint32_t nextModule; // always 0; filled in at runtime
uint32_t prevModule; // always 0; filled in at runtime
uint32_t sectionCount; // number of sections in the section table
uint32_t sectionTableOffset; // file position of section table
uint32_t moduleNameOffset; // offset of the module name in the string table
// (not in this file)
uint32_t moduleNameSize; // size of the module name in the string table (not
// in this file)
uint32_t formatVersion; // REL format version
uint32_t bssSize; // size of the BSS section
uint32_t relocationTableOffset; // file position of relocation entries
uint32_t importTableOffset; // file position of import table
uint32_t importTableSize; // size of import table
uint8_t prologSection; // section in which the _prolog function is in, or 0 if
// absent
uint8_t epilogSection; // section in which the _epilog function is in, or 0 if
// absent
uint8_t unresolvedSection; // section in which the _unresolved function is in,
// or 0 if absent
uint8_t pad33;
uint32_t prologOffset; // offset of the _prolog function in its section
uint32_t epilogOffset; // offset of the _epilog function in its section
uint32_t
unresolvedOffset; // offset of the _unresolved function in its section
};
struct RelRelocEntry {
int type; // relocation type
int patchSection; // section which relocation patch applies to
uint32_t patchOffset; // offset where this relocation patch applies
int symbolSection; // section of symbol
uint32_t symbolAddr; // for dol symbols, absolute address. for rel symbols,
// section offset
};
struct RelImportEntry {
int moduleId; // ID of module from which the relocation symbols are imported
// from
struct RelRelocEntry* relocs; // relocation entries
int relocsCount; // number of relocation entries
size_t relocsOffset; // file offset to relocation entries
};
struct Module {
int moduleId; // unique identifier of the module; the id of the DOL is always
// 0
const char* filename; // name of the module's ELF file
FILE* file; // ELF file
Elf32_Ehdr ehdr; // ELF header
Elf32_Sym* symtab; // ELF symbol table entries
int symtabCount; // number of ELF symbol table entries
char* strtab; // ELF string table
size_t strtabSize; // size of ELF string table
};
static struct Module dolModule;
static struct Module relModule;
static struct RelImportEntry* imports;
static int importsCount = 0;
static int minSectionCount = 0;
static int undefinedSymError = 0;
static void fatal_error(const char* msg, ...) {
va_list args;
va_start(args, msg);
vfprintf(stderr, msg, args);
va_end(args);
exit(1);
}
// Swaps byte order if the system is little endian
static void bswap32(uint32_t* ptr) {
if (IS_LITTLE_ENDIAN)
*ptr = (((*ptr >> 24) & 0xFF) << 0) | (((*ptr >> 16) & 0xFF) << 8) |
(((*ptr >> 8) & 0xFF) << 16) | (((*ptr >> 0) & 0xFF) << 24);
}
static void bswap16(uint16_t* ptr) {
if (IS_LITTLE_ENDIAN)
*ptr = (((*ptr >> 8) & 0xFF) << 0) | (((*ptr >> 0) & 0xFF) << 8);
}
static int read_checked(FILE* f, size_t offset, void* out, size_t size) {
return fseek(f, offset, SEEK_SET) == 0 && fread(out, size, 1, f) == 1;
}
static int write_checked(FILE* f, size_t offset, const void* in, size_t size) {
return fseek(f, offset, SEEK_SET) == 0 && fwrite(in, size, 1, f) == 1;
}
static uint32_t align(uint32_t n, unsigned int alignment) {
if (alignment == 0 || n % alignment == 0)
return n;
return n + alignment - (n % alignment);
}
static int is_supported_reloc_type(int type) {
switch (type) {
case R_PPC_ADDR32:
case R_PPC_ADDR24:
case R_PPC_ADDR16_LO:
case R_PPC_ADDR16_HA:
case R_PPC_REL24:
case R_PPC_REL14:
return 1;
}
return 0;
}
static const char* symbol_name(const struct Module* module,
const Elf32_Sym* sym) {
if (sym->st_name >= module->strtabSize)
return NULL;
return module->strtab + sym->st_name;
}
static int get_symtab_entry(const struct Module* module, int index,
Elf32_Sym* sym) {
if (index >= module->symtabCount)
return 0;
*sym = module->symtab[index];
bswap32(&sym->st_name);
bswap32(&sym->st_value);
bswap32(&sym->st_size);
bswap16(&sym->st_shndx);
return 1;
}
static int lookup_symbol_by_name(const struct Module* module, const char* name,
Elf32_Sym* sym) {
int i;
for (i = 0; i < module->symtabCount; i++) {
get_symtab_entry(module, i, sym);
const char* n = symbol_name(module, sym);
if (n != NULL && strcmp(name, n) == 0)
return 1;
}
return 0;
}
static struct RelImportEntry* get_import_for_module_id(int moduleId) {
int i;
for (i = 0; i < importsCount; i++) {
if (imports[i].moduleId == moduleId)
return &imports[i];
}
return NULL;
}
static void add_reloc_entry(const struct Module* module,
const Elf32_Rela* reloc, int relocSection) {
Elf32_Sym sym;
int symIdx = ELF32_R_SYM(reloc->r_info);
int relocType = ELF32_R_TYPE(reloc->r_info);
struct RelRelocEntry rentry;
struct RelImportEntry* import;
int moduleId; // module containing the symbol
if (!is_supported_reloc_type(relocType))
fatal_error("relocation type %i not supported\n", relocType);
rentry.patchSection = relocSection;
rentry.patchOffset = reloc->r_offset;
rentry.type = relocType;
// look for symbol in this module
if (!get_symtab_entry(module, symIdx, &sym))
fatal_error("couldn't find symbol index %i\n", symIdx);
if (sym.st_shndx != 0) // symbol is in this module
{
rentry.symbolSection = sym.st_shndx;
rentry.symbolAddr = sym.st_value + reloc->r_addend;
moduleId = module->moduleId;
} else // symbol is in another module (the DOL)
{
const char* name = symbol_name(&relModule, &sym);
if (!lookup_symbol_by_name(&dolModule, name, &sym)) {
undefinedSymError = 1;
fprintf(stderr, "could not find symbol '%s' in any module\n", name);
return;
}
if (sym.st_shndx >= dolModule.ehdr.e_shnum)
fatal_error("bad section index %i\n", sym.st_shndx);
rentry.symbolSection = sym.st_shndx;
rentry.symbolAddr = sym.st_value + reloc->r_addend;
moduleId = dolModule.moduleId;
}
import = get_import_for_module_id(moduleId);
// add import entry if it does not exist.
if (import == NULL) {
imports = realloc(imports, (importsCount + 1) * sizeof(*imports));
import = &imports[importsCount++];
import->moduleId = moduleId;
import->relocs = NULL;
import->relocsCount = 0;
}
// add relocation entry
import->relocs = realloc(import->relocs,
(import->relocsCount + 1) * sizeof(*import->relocs));
import->relocs[import->relocsCount++] = rentry;
}
static void module_get_section_header(const struct Module* module, int secNum,
Elf32_Shdr* shdr) {
size_t offset = module->ehdr.e_shoff + secNum * module->ehdr.e_shentsize;
if (secNum >= module->ehdr.e_shnum)
fatal_error("no such section index %i\n", secNum);
if (!read_checked(module->file, offset, shdr, sizeof(*shdr)))
fatal_error("error reading section header\n");
// convert from big endian
bswap32(&shdr->sh_name);
bswap32(&shdr->sh_type);
bswap32(&shdr->sh_flags);
bswap32(&shdr->sh_addr);
bswap32(&shdr->sh_offset);
bswap32(&shdr->sh_size);
bswap32(&shdr->sh_link);
bswap32(&shdr->sh_info);
bswap32(&shdr->sh_addralign);
bswap32(&shdr->sh_entsize);
}
static void module_read_relocs(struct Module* module) {
int i;
int j;
undefinedSymError = 0;
for (i = 0; i < (int)module->ehdr.e_shnum; i++) {
Elf32_Shdr shdr;
Elf32_Shdr forSection;
module_get_section_header(module, i, &shdr);
if (shdr.sh_type == SHT_RELA) {
module_get_section_header(module, shdr.sh_info, &forSection);
if (!(forSection.sh_flags & SHF_ALLOC))
continue;
for (j = 0; j < shdr.sh_size / sizeof(Elf32_Rela); j++) {
Elf32_Rela reloc;
read_checked(module->file, shdr.sh_offset + j * sizeof(Elf32_Rela),
&reloc, sizeof(reloc));
// convert from big endian
bswap32(&reloc.r_offset);
bswap32(&reloc.r_info);
bswap32((uint32_t*)&reloc.r_addend);
add_reloc_entry(&relModule, &reloc, shdr.sh_info);
}
}
}
if (undefinedSymError)
exit(1);
}
static int open_module(struct Module* module) {
int i;
// open file
module->file = fopen(module->filename, "rb");
if (module->file == NULL)
fatal_error("could not open %s for reading: %s\n", module->filename,
strerror(errno));
// read and verify ELF header
if (!read_checked(module->file, 0, &module->ehdr, sizeof(module->ehdr)))
fatal_error("error reading ELF header\n");
if (memcmp(module->ehdr.e_ident,
"\x7F"
"ELF",
4) != 0)
fatal_error("%s is not a valid ELF file\n", module->filename);
// convert from big endian
bswap16(&module->ehdr.e_type);
bswap16(&module->ehdr.e_machine);
bswap32(&module->ehdr.e_version);
bswap32(&module->ehdr.e_entry);
bswap32(&module->ehdr.e_phoff);
bswap32(&module->ehdr.e_shoff);
bswap32(&module->ehdr.e_flags);
bswap16(&module->ehdr.e_ehsize);
bswap16(&module->ehdr.e_phentsize);
bswap16(&module->ehdr.e_phnum);
bswap16(&module->ehdr.e_shentsize);
bswap16(&module->ehdr.e_shnum);
bswap16(&module->ehdr.e_shstrndx);
if (module->ehdr.e_shentsize < sizeof(Elf32_Shdr))
fatal_error("invalid section header size");
// Verify architecture
if (module->ehdr.e_ident[EI_CLASS] != ELFCLASS32 ||
module->ehdr.e_ident[EI_DATA] != ELFDATA2MSB ||
module->ehdr.e_machine != EM_PPC)
fatal_error("%s: wrong architecture. expected PowerPC 32-bit big endian.\n",
module->filename);
// Read symbol table and string table
for (i = 0; i < (int)module->ehdr.e_shnum; i++) {
Elf32_Shdr shdr;
module_get_section_header(module, i, &shdr);
if (shdr.sh_type == SHT_SYMTAB && module->symtab == NULL) {
module->symtabCount = shdr.sh_size / sizeof(Elf32_Sym);
module->symtab = malloc(shdr.sh_size);
if (!read_checked(module->file, shdr.sh_offset, module->symtab,
shdr.sh_size))
fatal_error("error reading symbol table\n");
} else if (shdr.sh_type == SHT_STRTAB && i != module->ehdr.e_shstrndx &&
module->strtab == NULL) {
module->strtabSize = shdr.sh_size;
module->strtab = malloc(shdr.sh_size);
if (!read_checked(module->file, shdr.sh_offset, module->strtab,
shdr.sh_size))
fatal_error("error reading string table\n");
}
}
if (module->symtab == NULL)
fatal_error("%s does not have a symbol table.\n", module->filename);
if (module->strtab == NULL)
fatal_error("%s does not have a string table.\n", module->filename);
return 1;
}
// searches for the special functions "_prolog", "_epliog", and "_unresolved"
static void find_rel_entry_functions(const struct Module* module,
struct RelHeader* relHdr) {
int i;
// puts("finding entry points");
for (i = 0; i < module->symtabCount; i++) {
Elf32_Sym sym;
Elf32_Shdr shdr;
const char* name;
get_symtab_entry(module, i, &sym);
name = symbol_name(module, &sym);
if (name == NULL)
continue;
if (strcmp(name, "_prolog") == 0) {
module_get_section_header(module, sym.st_shndx, &shdr);
relHdr->prologSection = sym.st_shndx;
relHdr->prologOffset = sym.st_value - shdr.sh_addr;
} else if (strcmp(name, "_epilog") == 0) {
module_get_section_header(module, sym.st_shndx, &shdr);
relHdr->epilogSection = sym.st_shndx;
relHdr->epilogOffset = sym.st_value - shdr.sh_addr;
} else if (strcmp(name, "_unresolved") == 0) {
module_get_section_header(module, sym.st_shndx, &shdr);
relHdr->unresolvedSection = sym.st_shndx;
relHdr->unresolvedOffset = sym.st_value - shdr.sh_addr;
}
}
}
// patches the bl instruction at insnp to jump to offset
static void patch_rel24_branch_offset(uint8_t* insnp, int32_t branchOffset) {
const uint32_t offsetMask =
0x03FFFFFC; // bits of instruction that contain the offset
uint32_t insn = (insnp[0] << 24) // read instruction
| (insnp[1] << 16) | (insnp[2] << 8) | (insnp[3] << 0);
assert(((insn >> 26) & 0x3F) ==
18); // TODO: do other instructions besides bl use R_PPC_REL24?
insn =
(insn & ~offsetMask) | (branchOffset & offsetMask); // patch instruction
// write instruction
insnp[0] = (insn >> 24) & 0xFF;
insnp[1] = (insn >> 16) & 0xFF;
insnp[2] = (insn >> 8) & 0xFF;
insnp[3] = (insn >> 0) & 0xFF;
}
static void patch_code_relocs(struct RelHeader* relHdr, int sectionId,
uint8_t* code, size_t size) {
struct RelImportEntry* import;
struct RelRelocEntry* reloc;
int i;
// Remove redundant R_PPC_REL24 relocations for calls to functions within
// the same module
import = get_import_for_module_id(relModule.moduleId);
assert(import != NULL);
for (i = 0, reloc = &import->relocs[0]; i < import->relocsCount;
i++, reloc++) {
if (reloc->patchSection == sectionId && reloc->type == R_PPC_REL24) {
assert(reloc->patchOffset < size);
patch_rel24_branch_offset(code + reloc->patchOffset,
reloc->symbolAddr - reloc->patchOffset);
// remove the relocation
reloc->type = R_PPC_NONE;
}
}
// Patch all calls to functions outside this module to jump to _unresolved
// by default.
if (relHdr->unresolvedSection == 0)
return;
import = get_import_for_module_id(0);
assert(import != NULL);
for (i = 0, reloc = &import->relocs[0]; i < import->relocsCount;
i++, reloc++) {
if (reloc->patchSection == sectionId && reloc->type == R_PPC_REL24) {
assert(reloc->patchOffset < size);
patch_rel24_branch_offset(code + reloc->patchOffset,
relHdr->unresolvedOffset - reloc->patchOffset);
}
}
}
static int compare_relocs(const void* a, const void* b) {
const struct RelRelocEntry* relocA = a;
const struct RelRelocEntry* relocB = b;
// Sort by sections to which these relocations apply
if (relocA->patchSection != relocB->patchSection)
return relocA->patchSection - relocB->patchSection;
// Sort by patch offset
if (relocA->patchOffset != relocB->patchOffset)
return relocA->patchOffset - relocB->patchOffset;
// Otherwise, leave the order alone
return (uintptr_t)a - (uintptr_t)b;
}
static int compare_imports(const void* a, const void* b) {
const struct RelImportEntry* impA = a;
const struct RelImportEntry* impB = b;
return impA->moduleId - impB->moduleId;
}
static void write_rel_file(struct Module* module, struct RelHeader* relHdr,
const char* filename) {
int i, j;
size_t filePos = sizeof(struct RelHeader); // skip over header for now
struct RelImportEntry* imp;
FILE* fout = fopen(filename, "wb");
if (fout == NULL)
fatal_error("could not open %s for writing: %s\n", filename,
strerror(errno));
relHdr->moduleId = module->moduleId;
relHdr->formatVersion = 1;
find_rel_entry_functions(module, relHdr);
// 1. Write section table and section contents
relHdr->sectionTableOffset = filePos;
relHdr->sectionCount = MAX(module->ehdr.e_shnum, minSectionCount);
// section contents follow section info table
filePos = relHdr->sectionTableOffset + relHdr->sectionCount * 8;
relHdr->bssSize = 0;
for (i = 0; i < module->ehdr.e_shnum; i++) {
Elf32_Shdr shdr;
struct {
uint32_t offset;
uint32_t size;
} secEntry = {0};
module_get_section_header(module, i, &shdr);
// write section contents
if (shdr.sh_type == SHT_PROGBITS && (shdr.sh_flags & SHF_ALLOC)) {
size_t sizeAligned = align(shdr.sh_size, 4);
uint32_t execflg = (shdr.sh_flags & SHF_EXECINSTR) ? 1 : 0;
filePos = align(filePos, shdr.sh_addralign);
if (shdr.sh_size > 0) {
uint8_t* data = calloc(sizeAligned, 1);
if (!read_checked(module->file, shdr.sh_offset, data, shdr.sh_size))
fatal_error("error reading section\n");
if (execflg)
patch_code_relocs(relHdr, i, data, sizeAligned);
if (!write_checked(fout, filePos, data, shdr.sh_size))
fatal_error("error writing rel section\n");
free(data);
}
secEntry.offset = filePos | execflg;
filePos += shdr.sh_size;
}
if (shdr.sh_flags & SHF_ALLOC)
secEntry.size = shdr.sh_size;
// write section table entry
bswap32(&secEntry.offset);
bswap32(&secEntry.size);
write_checked(fout, relHdr->sectionTableOffset + i * 8, &secEntry,
sizeof(secEntry));
// calculate total BSS size
if ((shdr.sh_flags & SHF_ALLOC) && shdr.sh_type == SHT_NOBITS)
relHdr->bssSize += shdr.sh_size;
}
// 2. Write relocation data
relHdr->relocationTableOffset = filePos;
// sort imports by module id
qsort(imports, importsCount, sizeof(struct RelImportEntry), compare_imports);
for (i = 0, imp = &imports[0]; i < importsCount; i++, imp++) {
struct RelRelocEntry* reloc;
int currSection = -1;
uint32_t prevOffset = 0;
struct {
uint16_t offset;
uint8_t type;
uint8_t section;
uint32_t symaddr;
} ent;
imp->relocsOffset = filePos;
// sort relocation entries by section
qsort(imp->relocs, imp->relocsCount, sizeof(struct RelRelocEntry),
compare_relocs);
for (j = 0, reloc = &imp->relocs[0]; j < imp->relocsCount; j++, reloc++) {
if (reloc->type == R_PPC_NONE) // ignore null relocations
continue;
if (reloc->patchSection != currSection) // section changed
{
currSection = reloc->patchSection;
// write section change
ent.offset = 0;
ent.type = R_DOLPHIN_SECTION;
ent.section = reloc->patchSection;
ent.symaddr = 0;
bswap16(&ent.offset);
bswap32(&ent.symaddr);
if (!write_checked(fout, filePos, &ent, sizeof(ent)))
fatal_error("error writing relocation entry");
filePos += sizeof(ent);
prevOffset = 0;
}
// write relocation
assert(reloc->patchOffset >= prevOffset);
ent.offset = reloc->patchOffset - prevOffset;
ent.type = reloc->type;
ent.section = reloc->symbolSection;
ent.symaddr = reloc->symbolAddr;
bswap16(&ent.offset);
bswap32(&ent.symaddr);
if (!write_checked(fout, filePos, &ent, sizeof(ent)))
fatal_error("error writing relocation entry");
filePos += sizeof(ent);
prevOffset = reloc->patchOffset;
}
// write end
ent.offset = 0;
ent.type = R_DOLPHIN_END;
ent.section = 0;
ent.symaddr = 0;
bswap16(&ent.offset);
bswap32(&ent.symaddr);
if (!write_checked(fout, filePos, &ent, sizeof(ent)))
fatal_error("error writing relocation entry");
filePos += sizeof(ent);
}
// 3. Write module import table
relHdr->importTableOffset = filePos;
for (i = 0, imp = &imports[0]; i < importsCount; i++, imp++) {
// write import table entry
struct {
uint32_t moduleId;
uint32_t relocsOffset;
} ent;
ent.moduleId = imp->moduleId;
ent.relocsOffset = imp->relocsOffset;
bswap32(&ent.moduleId);
bswap32(&ent.relocsOffset);
write_checked(fout, relHdr->importTableOffset + i * 8, &ent, sizeof(ent));
}
relHdr->importTableSize = importsCount * 8;
// 4. Write REL header.
// convert to big endian
bswap32(&relHdr->moduleId);
bswap32(&relHdr->sectionCount);
bswap32(&relHdr->sectionTableOffset);
bswap32(&relHdr->moduleNameOffset);
bswap32(&relHdr->moduleNameSize);
bswap32(&relHdr->formatVersion);
bswap32(&relHdr->bssSize);
bswap32(&relHdr->relocationTableOffset);
bswap32(&relHdr->importTableOffset);
bswap32(&relHdr->importTableSize);
bswap32(&relHdr->prologOffset);
bswap32(&relHdr->epilogOffset);
bswap32(&relHdr->unresolvedOffset);
write_checked(fout, 0, relHdr, sizeof(*relHdr));
fclose(fout);
}
static int parse_number(const char* str, int* n) {
char* end;
*n = strtol(str, &end, 0);
return end > str && *end == 0;
}
int main(int argc, char** argv) {
int i;
int moduleId = -1;
int nameOffset = 0;
int nameLen = 0;
const char* relName = NULL;
struct RelHeader relHdr = {0};
// Read command-line args
for (i = 1; i < argc; i++) {
if (strcmp(argv[i], "-c") == 0 ||
strcmp(argv[i], "--pad-section-count") == 0) {
if (i + 1 < argc && parse_number(argv[i + 1], &minSectionCount))
i++;
else
goto usage;
} else if (strcmp(argv[i], "-i") == 0 ||
strcmp(argv[i], "--module-id") == 0) {
if (i + 1 < argc && parse_number(argv[i + 1], &moduleId))
i++;
else
goto usage;
} else if (strcmp(argv[i], "-o") == 0 ||
strcmp(argv[i], "--name-offset") == 0) {
if (i + 1 < argc && parse_number(argv[i + 1], &nameOffset))
i++;
else
goto usage;
} else if (strcmp(argv[i], "-l") == 0 ||
strcmp(argv[i], "--name-length") == 0) {
if (i + 1 < argc && parse_number(argv[i + 1], &nameLen))
i++;
else
goto usage;
} else {
if (relModule.filename == NULL)
relModule.filename = argv[i];
else if (dolModule.filename == NULL)
dolModule.filename = argv[i];
else if (relName == NULL)
relName = argv[i];
else
goto usage;
}
}
if (relModule.filename == NULL || dolModule.filename == NULL ||
relName == NULL)
goto usage;
open_module(&relModule);
open_module(&dolModule);
dolModule.moduleId = 0;
relModule.moduleId = moduleId;
module_read_relocs(&relModule);
// TODO: Read this information from string table
relHdr.moduleNameOffset = nameOffset;
relHdr.moduleNameSize = nameLen;
write_rel_file(&relModule, &relHdr, relName);
fclose(relModule.file);
fclose(dolModule.file);
free(dolModule.strtab);
free(dolModule.symtab);
for (i = 0; i < importsCount; i++)
free(imports[i].relocs);
free(imports);
return 0;
usage:
fprintf(stderr,
"usage: %s reloc_elf static_elf rel_file\n"
"options:\n"
" -i, --module-id <n> sets the module ID in the rel header "
"to <n>\n"
" -c, --pad-section-count <n> ensures that the rel will have at "
"least <n>\n"
" sections\n"
" -o, --name-offset <offset> sets the name offset in the rel "
"header to\n"
" <offset>\n"
" -l, --name-length <len> sets the name length in the rel "
"header to <len>\n",
argv[0]);
return 1;
}

View File

@ -1,95 +0,0 @@
#! /usr/bin/env python3
# Written by Ethan Roseman (ethteck)
# MIT License
# Copyright 2021
# Modified by EpochFlame
import argparse
# Byte sequence that marks code size
CODESIZE_MAGIC = b"\x00\x00\x00\x06\x00\x00\x00\x00\x00\x00\x00\x34"
BLR_BYTE_SEQ = b"\x4E\x80\x00\x20"
MTLR_BYTE_SEQ = b"\x7C\x08\x03\xA6"
# Byte sequence array for branches to link register
BLR_BYTE_SEQ_ARRAY = [BLR_BYTE_SEQ,
b"\x4D\x80\x00\x20", b"\x4D\x80\x00\x21", b"\x4C\x81\x00\x20", b"\x4C\x81\x00\x21",
b"\x4D\x82\x00\x20", b"\x4D\x82\x00\x21", b"\x4C\x80\x00\x20", b"\x4C\x80\x00\x21",
b"\x4D\x81\x00\x20", b"\x4D\x81\x00\x21", b"\x4C\x80\x00\x20", b"\x4C\x80\x00\x21",
b"\x4C\x82\x00\x20", b"\x4C\x82\x00\x21", b"\x4C\x81\x00\x20", b"\x4C\x81\x00\x21",
b"\x4D\x83\x00\x20", b"\x4D\x83\x00\x21", b"\x4C\x83\x00\x20", b"\x4C\x83\x00\x21",
b"\x4D\x83\x00\x20", b"\x4D\x83\x00\x21", b"\x4C\x83\x00\x20", b"\x4C\x83\x00\x21"]
# Example invocation: ./frank.py vanilla.o profile.o output.o
parser = argparse.ArgumentParser()
parser.add_argument("vanilla", help="Path to the vanilla object", type=argparse.FileType('rb'))
parser.add_argument("profile", help="Path to the profile object", type=argparse.FileType('rb'))
parser.add_argument("target", help="Path to the target object (to write)")
args = parser.parse_args()
# Read contents into bytearrays and close files
vanilla_bytes = args.vanilla.read()
args.vanilla.close()
profile_bytes = args.profile.read()
args.profile.close()
# Remove byte sequence
stripped_bytes = profile_bytes.replace(b"\x48\x00\x00\x01\x60\x00\x00\x00", b"")
# Find end of code sections in vanilla and stripped bytes
code_size_offset = (vanilla_bytes.find(CODESIZE_MAGIC) + 12)
code_size_bytes = vanilla_bytes[code_size_offset:code_size_offset+4]
code_size = int.from_bytes(code_size_bytes, byteorder='big')
eoc_offset = 0x34 + code_size
# Break if the eoc is not found
assert(eoc_offset != len(vanilla_bytes))
# Replace 0x34 - eoc in vanilla with bytes from stripped
final_bytes = vanilla_bytes[:0x34] + stripped_bytes[0x34:eoc_offset] + vanilla_bytes[eoc_offset:]
# Fix branches to link register
for seq in BLR_BYTE_SEQ_ARRAY:
idx = 0
while idx < len(vanilla_bytes):
found_pos = vanilla_bytes.find(seq, idx)
if found_pos == -1:
break # break while loop when no targets remain
if found_pos % 4 != 0: # check 4-byte alignment
idx += 4
continue
final_bytes = final_bytes[:found_pos] + vanilla_bytes[found_pos:found_pos+4] + final_bytes[found_pos+4:]
idx = found_pos + len(seq)
# Reunify mtlr/blr instructions, shifting intermediary instructions up
idx = 0
while idx < len(final_bytes):
# Find mtlr position
mtlr_found_pos = final_bytes.find(MTLR_BYTE_SEQ, idx)
if mtlr_found_pos == -1:
break # break while loop when no targets remain
if mtlr_found_pos % 4 != 0: # check 4-byte alignment
idx += 4
continue
# Find paired blr position
blr_found_pos = final_bytes.find(BLR_BYTE_SEQ, mtlr_found_pos)
if blr_found_pos == -1:
break # break while loop when no targets remain
if blr_found_pos % 4 != 0: # check 4-byte alignment
idx += 4
continue
if mtlr_found_pos + 4 == blr_found_pos:
idx += 4
continue # continue if mtlr is followed directly by blr
final_bytes = final_bytes[:mtlr_found_pos] + final_bytes[mtlr_found_pos+4:blr_found_pos] + final_bytes[mtlr_found_pos:mtlr_found_pos+4] + final_bytes[blr_found_pos:]
idx = mtlr_found_pos + len(MTLR_BYTE_SEQ)
with open(args.target, "wb") as f:
f.write(final_bytes)

View File

@ -1,72 +0,0 @@
#!/usr/bin/env python3
# MIT License
# See https://github.com/ethteck/m2ctx/blob/main/m2ctx.py
import argparse
import os
import sys
import subprocess
import tempfile
script_dir = os.path.dirname(os.path.realpath(__file__))
root_dir = os.path.abspath(os.path.join(script_dir, ".."))
src_dir = root_dir + "src/"
CPP_FLAGS = [
"-Iinclude",
"-Isrc",
"-DGEKKO",
"-DHW2",
"-D__attribute__(...)=",
"-D__asm__(...)=",
"-ffreestanding",
"-DM2CTX",
]
def import_c_file(in_file) -> str:
in_file = os.path.relpath(in_file, root_dir)
cpp_command = ["gcc", "-E", "-P", "-dM", *CPP_FLAGS, in_file]
cpp_command2 = ["gcc", "-E", "-P", *CPP_FLAGS, in_file]
with tempfile.NamedTemporaryFile(suffix=".c") as tmp:
stock_macros = subprocess.check_output(["gcc", "-E", "-P", "-dM", tmp.name], cwd=root_dir, encoding="utf-8")
out_text = ""
try:
out_text += subprocess.check_output(cpp_command, cwd=root_dir, encoding="utf-8")
out_text += subprocess.check_output(cpp_command2, cwd=root_dir, encoding="utf-8")
except subprocess.CalledProcessError:
print(
"Failed to preprocess input file, when running command:\n"
+ cpp_command,
file=sys.stderr,
)
sys.exit(1)
if not out_text:
print("Output is empty - aborting")
sys.exit(1)
for line in stock_macros.strip().splitlines():
out_text = out_text.replace(line + "\n", "")
return out_text
def main():
parser = argparse.ArgumentParser(
description="""Create a context file which can be used for mips_to_c"""
)
parser.add_argument(
"c_file",
help="""File from which to create context""",
)
args = parser.parse_args()
output = import_c_file(args.c_file)
with open(os.path.join(root_dir, "ctx.c"), "w", encoding="UTF-8") as f:
f.write(output)
if __name__ == "__main__":
main()

View File

@ -1,496 +0,0 @@
#!/usr/bin/env python3
from capstone import *
from capstone.ppc import *
from elftools.elf.elffile import *
from elftools.elf.sections import *
import sys
# addr -> name
labels = {}
# fileOffset -> {addr, type}
relocations = {}
# index -> {offset, flags, length, is_bss, name}
sectionInfo = []
R_PPC_NONE = 0
R_PPC_ADDR32 = 1
R_PPC_ADDR24 = 2
R_PPC_ADDR16_LO = 4
R_PPC_ADDR16_HI = 5
R_PPC_ADDR16_HA = 6
R_PPC_REL24 = 10
R_PPC_REL14 = 11
R_DOLPHIN_SECTION = 202
R_DOLPHIN_END = 203
relocationTypeNames = {
R_PPC_NONE: 'R_PPC_NONE',
R_PPC_ADDR32: 'R_PPC_ADDR32',
R_PPC_ADDR24: 'R_PPC_ADDR24',
R_PPC_ADDR16_LO: 'R_PPC_ADDR16_LO',
R_PPC_ADDR16_HI: 'R_PPC_ADDR16_HI',
R_PPC_ADDR16_HA: 'R_PPC_ADDR16_HA',
R_PPC_REL24: 'R_PPC_REL24',
R_PPC_REL14: 'R_PPC_REL14',
R_DOLPHIN_SECTION: 'R_DOLPHIN_SECTION',
R_DOLPHIN_END: 'R_DOLPHIN_END'
}
def read_u8(offset):
return filecontent[offset]
def read_u16(offset):
return (filecontent[offset + 0] << 8) | filecontent[offset + 1]
def read_u32(offset):
return (filecontent[offset + 0] << 24) | (filecontent[offset + 1] << 16) | (filecontent[offset + 2] << 8) | filecontent[offset + 3]
def add_label(addr, name=None):
if addr in labels:
return labels[addr]
if name == None:
name = 'lbl_%08X' % addr
labels[addr] = name
return name
with open(sys.argv[1], 'rb') as file:
filecontent = bytearray(file.read())
if len(sys.argv) >= 3:
# Why is this so slow?
with open(sys.argv[2], 'rb') as f:
elf = ELFFile(f)
elfsymtab = elf.get_section_by_name('.symtab')
for i in range(0, elfsymtab.num_symbols()):
sym = elfsymtab.get_symbol(i)
if len(sym.name) > 0 and not sym.name[0] in {'.', '@'}:
add_label(sym['st_value'], sym.name)
id = read_u32(0)
numSections = read_u32(0x0C)
sectionInfoOffset = read_u32(0x10)
nameOffset = read_u32(0x14)
nameSize = read_u32(0x18)
version = read_u32(0x1C)
bssSize = read_u32(0x20)
relOffset = read_u32(0x24)
impOffset = read_u32(0x28)
impSize = read_u32(0x2C)
prologSection = read_u8(0x30)
epilogSection = read_u8(0x31)
unresolvedSection = read_u8(0x32)
prolog = read_u32(0x34)
epilog = read_u32(0x38)
unresolved = read_u32(0x3C)
print("# id: %i" % id)
print("# version: %i" % version)
print("# nameoffset: 0x%X, size: 0x%X" % (nameOffset, nameSize))
print("# section table: 0x%X, size: 0x%X" % (sectionInfoOffset, numSections*8))
print("# imp table: 0x%X" % impOffset)
print("# relocs offset: 0x%X" % relOffset)
print("# _prolog: %i:0x%X" % (prologSection, prolog))
print("# _epilog: %i:0x%X" % (epilogSection, epilog))
print("# _unresolved: %i:0x%X" % (unresolvedSection, unresolved))
print("# num sections: %i" % numSections)
print('.include "macros.inc"')
#print("%i sections:" % numSections)
# Read sections
for i in range(0, numSections):
o = sectionInfoOffset + i * 8
section = {
'offset': read_u32(o + 0) & ~3,
'flags': read_u32(o + 0) & 3,
'length': read_u32(o + 4)
}
if section['offset'] == 0 and section['length'] > 0:
section['is_bss'] = True
else:
section['is_bss'] = False
# Hack: if bss, then set file offset to something unique as to not
# clash with other symbols
if section['is_bss']:
section['offset'] = 0x10000000
# Determine name
if section['is_bss']:
section['name'] = '.bss%i' % i
elif section['flags'] & 1:
section['name'] = '.text%i' % i
else:
section['name'] = '.data%i' % i
sectionInfo.append(section)
print("# offset: 0x%08X\tlength: 0x%08X\tflags: %i" %
(section['offset'], section['length'], section['flags']))
sectionInfo[1]['name'] = '.text'
sectionInfo[2]['name'] = '.ctors'
sectionInfo[3]['name'] = '.dtors'
sectionInfo[4]['name'] = '.rodata'
sectionInfo[5]['name'] = '.data'
sectionInfo[6]['name'] = '.bss'
# Add labels for prologue and epilogue
if prologSection != 0:
labels[sectionInfo[prologSection]['offset'] + prolog] = '_prolog'
if epilogSection != 0:
labels[sectionInfo[epilogSection]['offset'] + epilog] = '_epilog'
if unresolvedSection != 0:
labels[sectionInfo[unresolvedSection]['offset'] + unresolved] = '_unresolved'
def read_relocation_info(module, o):
currSection = None
missingSymbols = False
while True:
offset = read_u16(o + 0)
type = read_u8(o + 2)
section = read_u8(o + 3)
addend = read_u32(o + 4)
# Get address of symbol and add label
symAddr = 0
if type == R_DOLPHIN_SECTION: # R_DOLPHIN_SECTION
currSection = sectionInfo[section]
relocOffset = currSection['offset']
if type < 200:
if module == 0: # dol
symAddr = addend
if symAddr not in labels:
print('error: symbol for 0x%08X not found' % symAddr)
missingSymbols = True
else: # rel
symAddr = sectionInfo[section]['offset'] + addend
labels[symAddr] = 'lbl_%08X' % symAddr
# Get file offset for relocation
relocOffset += offset
if type < 200:
reloc = {
'addr': symAddr,
'type': type,
}
relocations[relocOffset] = reloc
#print(" offset: 0x%04X(+0x%X)\ttype: %s\tsection: %i\tsym_addr: 0x%08X" % (relocOffset, offset, relocationTypeNames[type], section, symAddr))
#print(" offset: 0x%04X(+0x%X)\ttype: %s\tsection: %i\tsym_addr: ?" % (relocOffset, offset, relocationTypeNames[type], section))
if type == R_DOLPHIN_END:
break
o += 8
if missingSymbols:
exit(1)
numImpEntries = impSize / 8
#print("%i imports" % numImpEntries)
for i in range(0, int(numImpEntries)):
o = impOffset + i * 8
module = read_u32(o + 0)
relocation = read_u32(o + 4)
#print("module: %i, offset: 0x%08X" % (module, relocation))
read_relocation_info(module, relocation)
cs = Cs(CS_ARCH_PPC, CS_MODE_32 | CS_MODE_BIG_ENDIAN)
cs.detail = True
cs.imm_unsigned = False
def get_relocation_for_offset(o):
for i in range(o, o + 4):
if i in relocations:
return relocations[i]
return None
def get_label(addr):
if addr in labels:
return labels[addr]
return '0x%08X' % addr
def print_label(label):
if label in ['_prolog', '_epilog', '_unresolved']:
label = '.global %s\n%s' % (label, label)
print('%s:' % label)
def sign_extend_16(value):
if value > 0 and (value & 0x8000):
value -= 0x10000
return value
def disasm_fcmp(inst):
crd = (inst & 0x03800000) >> 23
a = (inst & 0x001f0000) >> 16
b = (inst & 0x0000f800) >> 11
return 'fcmpo cr%i, f%i, f%i' % (crd, a, b)
def disasm_mspr(inst, mode):
if (inst & 1):
return None
d = (inst & 0x03e00000) >> 21
a = (inst & 0x001f0000) >> 16
b = (inst & 0x0000f800) >>11
spr = (b << 5) + a
if mode:
return 'mtspr 0x%X, r%i' % (spr, d)
else:
return 'mfspr r%i, 0x%X' % (d, spr)
def disasm_mcrxr(inst):
if (inst & 0x007ff801):
return None
crd = (inst & 0x03800000) >> 23
return 'mcrxr cr%i' % crd
def disassemble_insn_that_capstone_cant_handle(o, reloc):
if reloc:
relocComment = '\t;# %s:%s' % (get_label(reloc['addr']), relocationTypeNames[reloc['type']])
else:
relocComment = ''
raw = read_u32(o)
asm = None
idx = (raw & 0xfc000000) >> 26
idx2 = (raw & 0x000007fe) >> 1
# mtspr
if idx == 31 and idx2 == 467:
asm = disasm_mspr(raw, 1)
# mfspr
elif idx == 31 and idx2 == 339:
asm = disasm_mspr(raw, 0)
# mcrxr
elif idx == 31 and idx2 == 512:
asm = disasm_mcrxr(raw)
# fcmpo
elif idx == 63 and idx2 == 32:
asm = disasm_fcmp(raw)
# Paired singles
elif idx == 4:
asm = disasm_ps(raw)
elif idx in {56, 57, 60, 61}:
asm = disasm_ps_mem(raw, idx)
if asm:
return asm
return '.4byte 0x%08X ;# (error: unknown instruction) %s' % (read_u32(o), relocComment)
def disassemble_insn(o, reloc):
if reloc:
relocComment = '\t;# %s:%s' % (get_label(reloc['addr']), relocationTypeNames[reloc['type']])
else:
relocComment = ''
try:
insn = next(cs.disasm(filecontent[o : o+4], o))
except StopIteration:
return disassemble_insn_that_capstone_cant_handle(o, reloc)
if reloc:
relocType = reloc['type']
else:
relocType = -1
# handle relocs label
if insn.id in {PPC_INS_BL, PPC_INS_BC} and relocType in {R_PPC_REL24, R_PPC_REL14}:
return '%s %s' % (insn.mnemonic, get_label(reloc['addr']))
if insn.id == PPC_INS_LIS and relocType == R_PPC_ADDR16_HA:
return '%s %s, %s@ha' % (insn.mnemonic, insn.reg_name(insn.operands[0].reg), get_label(reloc['addr']))
if insn.id == PPC_INS_LIS and relocType == R_PPC_ADDR16_HI:
return '%s %s, %s@h' % (insn.mnemonic, insn.reg_name(insn.operands[0].reg), get_label(reloc['addr']))
if insn.id in {PPC_INS_ADDI, PPC_INS_ORI} and relocType == R_PPC_ADDR16_LO:
return '%s %s, %s, %s@l' % (insn.mnemonic, insn.reg_name(insn.operands[0].reg), insn.reg_name(insn.operands[1].reg), get_label(reloc['addr']))
if insn.id in {
PPC_INS_LWZ, PPC_INS_LHZ, PPC_INS_LHA, PPC_INS_LBZ,
PPC_INS_LWZU, PPC_INS_LHZU, PPC_INS_LHAU, PPC_INS_LBZU,
PPC_INS_LFS, PPC_INS_LFD,
PPC_INS_LFSU, PPC_INS_LFDU,
PPC_INS_STW, PPC_INS_STH, PPC_INS_STB,
PPC_INS_STWU, PPC_INS_STHU, PPC_INS_STBU,
PPC_INS_STFS, PPC_INS_STFD,
PPC_INS_STFSU, PPC_INS_STFDU} \
and relocType == R_PPC_ADDR16_LO:
return '%s %s, %s@l(%s)' % (insn.mnemonic, insn.reg_name(insn.operands[0].reg), get_label(reloc['addr']), insn.reg_name(insn.operands[1].mem.base))
# branch target labels
if insn.id in {PPC_INS_B, PPC_INS_BL, PPC_INS_BDZ, PPC_INS_BDNZ, PPC_INS_BC}:
if reloc:
return '%s %s' % (insn.mnemonic, get_label(reloc['addr']))
#add_label(insn.operands[0].imm)
#label = labels[insn.operands[0].imm]
#if label:
# WTF, capstone?
if o == 0xAD8C:
return '%s lbl_0000ADB0' % insn.mnemonic
return '%s %s' % (insn.mnemonic, get_label(insn.operands[0].imm))
# misc. fixes
# Sign-extend immediate values because Capstone is an idiot and thinks all immediates are unsigned
if insn.id in {PPC_INS_ADDI, PPC_INS_ADDIC, PPC_INS_SUBFIC, PPC_INS_MULLI} and (insn.operands[2].imm & 0x8000):
return "%s %s, %s, %i ;# fixed addi" % (insn.mnemonic, insn.reg_name(insn.operands[0].reg), insn.reg_name(insn.operands[1].value.reg), insn.operands[2].imm - 0x10000)
if (insn.id == PPC_INS_LI or insn.id == PPC_INS_CMPWI) and (insn.operands[1].imm & 0x8000):
return "%s %s, %i" % (insn.mnemonic, insn.reg_name(insn.operands[0].reg), insn.operands[1].imm - 0x10000)
# cntlz -> cntlzw
if insn.id == PPC_INS_CNTLZW:
return "cntlzw %s" % insn.op_str
return '%s %s%s' % (insn.mnemonic, insn.op_str, relocComment)
def scan_local_labels(o, size):
end = o + size
while o < end:
reloc = get_relocation_for_offset(o)
if reloc:
pass
else:
try:
insn = next(cs.disasm(filecontent[o:o+4], o))
if insn.id in {PPC_INS_B, PPC_INS_BL, PPC_INS_BC, PPC_INS_BDZ, PPC_INS_BDNZ}:
for op in insn.operands:
if op.type == PPC_OP_IMM:
l = add_label(op.imm)
#print('adding local label %s(0x%X) from offset 0x%X' % (l, op.imm, o))
except StopIteration:
pass
o += 4
#for insn in cs.disasm(filecontent[o:o+size], o):
# # branch labels
# if insn.id in {PPC_INS_B, PPC_INS_BL, PPC_INS_BC, PPC_INS_BDZ, PPC_INS_BDNZ}:
# for op in insn.operands:
# if op.type == PPC_OP_IMM:
# l = add_label(op.imm)
# print('adding local label %s(0x%X) from offset 0x%X' % (l, op.imm, o))
def dump_code(o, size):
scan_local_labels(o, size)
end = o + size
code = filecontent[o : end]
while o < end:
if o in labels:
print_label(labels[o])
asm = disassemble_insn(o, get_relocation_for_offset(o))
print('/* %08X %08X */ %s' % (o, read_u32(o), asm))
#print('/* %08X */ %s' % (read_u32(o), asm))
o += 4
if o < end:
print('incomplete')
# returns True if value is 4-byte aligned
def is_aligned(num):
return num % 4 == 0
def align(num):
return (num + 3) & ~3
def is_ascii(code):
if code >= 0x20 and code <= 0x7E: # normal characters
return True
if code in [0x09, 0x0A]: # tab, newline
return True
return False
# returns True if all elements are zero
def is_all_zero(arr):
for val in arr:
if val != 0:
return False
return True
# returns string of comma-separated hex bytes
def hex_bytes(data):
return ', '.join('0x%02X' % n for n in data)
# reads a string starting at pos
def read_string(data, pos):
text = ''
while pos < len(data) and is_ascii(data[pos]):
text += chr(data[pos])
pos += 1
if pos < len(data) and data[pos] == 0:
return text
return ''
# escapes special characters in the string for use in a C string literal
def escape_string(text):
return text.replace('\\','\\\\').replace('"','\\"').replace('\n','\\n').replace('\t','\\t')
def output_data_range(secNum, o, end):
print(' # 0x%X' % o)
if not is_aligned(o):
print(' .byte ' + hex_bytes(filecontent[o:align(o)]))
o = align(o)
while o < (end & ~3):
# Try to see if this is a string.
string = read_string(filecontent, o)
if len(string) >= 4 and secNum == 5: # strings are only in .data
strEnd = o + len(string)+1
if is_aligned(strEnd) or is_all_zero(filecontent[strEnd : align(strEnd)-strEnd]):
print(' .asciz \"%s"' % escape_string(string))
if not is_aligned(strEnd):
print(' .balign 4')
o = align(strEnd)
continue
# Not a string
reloc = get_relocation_for_offset(o)
if reloc:
type = reloc['type']
if type == R_PPC_ADDR32:
value = labels[reloc['addr']]
else:
print('dunno what to do about %s here' % relocationTypeNames[type])
else:
value = '0x%08X' % read_u32(o)
print(' .4byte %s' % value)
o += 4
if o < end:
print(' .byte ' + hex_bytes(filecontent[o:end]))
return
def dump_data(secNum, o, size):
end = o + size
lastPos = o
while o < end:
if o in labels:
if o - lastPos > 0:
output_data_range(secNum, lastPos, o)
print_label(labels[o])
lastPos = o
o += 1
if o - lastPos > 0:
output_data_range(secNum, lastPos, o)
return
def output_bss_range(start, end):
print(' .skip 0x%X' % (end - start))
def dump_bss(o, size):
end = o + size
lastPos = o
while o < end:
if o in labels:
if o - lastPos > 0:
output_bss_range(lastPos, o)
print_label(labels[o])
lastPos = o
o += 1
if o - lastPos > 0:
output_bss_range(lastPos, o)
return
for i in range(0, numSections):
section = sectionInfo[i]
if section['offset'] == 0 and section['length'] == 0:
continue
print('# %i' % i)
print('.section %s' % section['name'])
if section['is_bss']:
# bss section
dump_bss(section['offset'], section['length'])
elif section['flags'] & 1:
# code section
dump_code(section['offset'], section['length'])
elif section['offset'] != 0:
# data section
dump_data(i, section['offset'], section['length'])
print('')