mirror of https://github.com/zeldaret/mm.git
ABI FPR names (#767)
* Add o32 to disassembler and update macro.inc * Add a variable for asm processor to makefile and improve objdump flags * Update diff settings * git subrepo pull --force tools/asm-differ subrepo: subdir: "tools/asm-differ" merged: "1236288d1" upstream: origin: "https://github.com/simonlindholm/asm-differ" branch: "main" commit: "1236288d1" git-subrepo: version: "0.4.3" origin: "https://github.com/ingydotnet/git-subrepo.git" commit: "2f68596" * Remove * import and implement option * Fix some stuff in the makefile * Update asm-processor * Review * Fix old var name
This commit is contained in:
EllipticEllipsis
GitHub
parent
54a4d1eb12
commit
15dfaf0862
39
Makefile
39
Makefile
|
|
@ -21,9 +21,9 @@ ifeq ($(NON_MATCHING),1)
|
|||
COMPARE := 0
|
||||
endif
|
||||
|
||||
DISASM_FLAGS ?=
|
||||
DISASM_FLAGS := --reg-names=o32
|
||||
ifneq ($(FULL_DISASM), 0)
|
||||
DISASM_FLAGS += --full
|
||||
DISASM_FLAGS += --all
|
||||
endif
|
||||
|
||||
PROJECT_DIR := $(dir $(realpath $(firstword $(MAKEFILE_LIST))))
|
||||
|
|
@ -73,6 +73,9 @@ AS := $(MIPS_BINUTILS_PREFIX)as
|
|||
LD := $(MIPS_BINUTILS_PREFIX)ld
|
||||
OBJCOPY := $(MIPS_BINUTILS_PREFIX)objcopy
|
||||
OBJDUMP := $(MIPS_BINUTILS_PREFIX)objdump
|
||||
ASM_PROC := python3 tools/asm-processor/build.py
|
||||
|
||||
ASM_PROC_FLAGS := --input-enc=utf-8 --output-enc=euc-jp
|
||||
|
||||
IINC := -Iinclude -Isrc -Iassets -Ibuild -I.
|
||||
|
||||
|
|
@ -99,6 +102,9 @@ MIPS_VERSION := -mips2
|
|||
# we support Microsoft extensions such as anonymous structs, which the compiler does support but warns for their usage. Surpress the warnings with -woff.
|
||||
CFLAGS += -G 0 -non_shared -Xfullwarn -Xcpluscomm $(IINC) -nostdinc -Wab,-r4300_mul -woff 624,649,838,712
|
||||
|
||||
# Use relocations and abi fpr names in the dump
|
||||
OBJDUMP_FLAGS := -d -r -z -Mreg-names=32
|
||||
|
||||
ifeq ($(shell getconf LONG_BIT), 32)
|
||||
# Work around memory allocation bug in QEMU
|
||||
export QEMU_GUEST_BASE := 1
|
||||
|
|
@ -178,6 +184,7 @@ build/src/libultra/flash/%.o: MIPS_VERSION := -mips1
|
|||
build/src/code/audio/%.o: OPTFLAGS := -O2
|
||||
|
||||
build/assets/%.o: OPTFLAGS := -O1
|
||||
build/assets/%.o: ASM_PROC_FLAGS :=
|
||||
|
||||
# file flags
|
||||
build/src/boot_O2_g3/fault.o: CFLAGS += -trapuv
|
||||
|
|
@ -194,21 +201,19 @@ build/src/libultra/libc/llcvt.o: OPTFLAGS := -O1
|
|||
build/src/libultra/libc/llcvt.o: MIPS_VERSION := -mips3 -32
|
||||
|
||||
# cc & asm-processor
|
||||
build/src/boot_O2/%.o: CC := python3 tools/asm-processor/build.py $(CC) -- $(AS) $(ASFLAGS) --
|
||||
build/src/boot_O2_g3/%.o: CC := python3 tools/asm-processor/build.py $(CC) -- $(AS) $(ASFLAGS) --
|
||||
build/src/boot_O2/%.o: CC := $(ASM_PROC) $(ASM_PROC_FLAGS) $(CC) -- $(AS) $(ASFLAGS) --
|
||||
build/src/boot_O2_g3/%.o: CC := $(ASM_PROC) $(ASM_PROC_FLAGS) $(CC) -- $(AS) $(ASFLAGS) --
|
||||
|
||||
build/src/libultra/%.o: CC := python3 tools/asm-processor/build.py $(CC_OLD) -- $(AS) $(ASFLAGS) --
|
||||
build/src/libultra/%.o: CC := $(CC_OLD)
|
||||
# Needed at least until voice is decompiled
|
||||
build/src/libultra/voice/%.o: CC := $(ASM_PROC) $(ASM_PROC_FLAGS) $(CC_OLD) -- $(AS) $(ASFLAGS) --
|
||||
|
||||
build/src/code/%.o: CC := python3 tools/asm-processor/build.py $(CC) -- $(AS) $(ASFLAGS) --
|
||||
build/src/code/audio/%.o: CC := python3 tools/asm-processor/build.py $(CC) -- $(AS) $(ASFLAGS) --
|
||||
build/src/code/%.o: CC := $(ASM_PROC) $(ASM_PROC_FLAGS) $(CC) -- $(AS) $(ASFLAGS) --
|
||||
build/src/code/audio/%.o: CC := $(ASM_PROC) $(ASM_PROC_FLAGS) $(CC) -- $(AS) $(ASFLAGS) --
|
||||
|
||||
build/src/overlays/actors/%.o: CC := python3 tools/asm-processor/build.py $(CC) -- $(AS) $(ASFLAGS) --
|
||||
build/src/overlays/effects/%.o: CC := python3 tools/asm-processor/build.py $(CC) -- $(AS) $(ASFLAGS) --
|
||||
build/src/overlays/fbdemos/%.o: CC := python3 tools/asm-processor/build.py $(CC) -- $(AS) $(ASFLAGS) --
|
||||
build/src/overlays/gamestates/%.o: CC := python3 tools/asm-processor/build.py $(CC) -- $(AS) $(ASFLAGS) --
|
||||
build/src/overlays/kaleido_scope/%.o: CC := python3 tools/asm-processor/build.py $(CC) -- $(AS) $(ASFLAGS) --
|
||||
build/src/overlays/%.o: CC := $(ASM_PROC) $(ASM_PROC_FLAGS) $(CC) -- $(AS) $(ASFLAGS) --
|
||||
|
||||
build/assets/%.o: CC := python3 tools/asm-processor/build.py $(CC) -- $(AS) $(ASFLAGS) --
|
||||
build/assets/%.o: CC := $(ASM_PROC) $(ASM_PROC_FLAGS) $(CC) -- $(AS) $(ASFLAGS) --
|
||||
|
||||
#### Main Targets ###
|
||||
|
||||
|
|
@ -303,7 +308,7 @@ build/data/%.o: data/%.s
|
|||
build/src/overlays/%.o: src/overlays/%.c
|
||||
$(CC_CHECK) $<
|
||||
$(CC) -c $(CFLAGS) $(MIPS_VERSION) $(OPTFLAGS) -o $@ $<
|
||||
@$(OBJDUMP) -d $@ > $(@:.o=.s)
|
||||
@$(OBJDUMP) $(OBJDUMP_FLAGS) $@ > $(@:.o=.s)
|
||||
# TODO: `() || true` is currently necessary to suppress `Error 1 (ignored)` make warnings caused by `test`, but this will go away if
|
||||
# the following is moved to a separate rule that is only run once when all the required objects have been compiled.
|
||||
$(ZAPD) bovl -eh -i $@ -cfg $< --outputpath $(@D)/$(notdir $(@D))_reloc.s
|
||||
|
|
@ -313,21 +318,21 @@ build/src/overlays/%.o: src/overlays/%.c
|
|||
build/src/%.o: src/%.c
|
||||
$(CC_CHECK) $<
|
||||
$(CC) -c $(CFLAGS) $(MIPS_VERSION) $(OPTFLAGS) -o $@ $<
|
||||
@$(OBJDUMP) -d $@ > $(@:.o=.s)
|
||||
@$(OBJDUMP) $(OBJDUMP_FLAGS) $@ > $(@:.o=.s)
|
||||
$(RM_MDEBUG)
|
||||
|
||||
build/src/libultra/libc/ll.o: src/libultra/libc/ll.c
|
||||
$(CC_CHECK) $<
|
||||
$(CC) -c $(CFLAGS) $(MIPS_VERSION) $(OPTFLAGS) -o $@ $<
|
||||
python3 tools/set_o32abi_bit.py $@
|
||||
@$(OBJDUMP) -d $@ > $(@:.o=.s)
|
||||
@$(OBJDUMP) $(OBJDUMP_FLAGS) $@ > $(@:.o=.s)
|
||||
$(RM_MDEBUG)
|
||||
|
||||
build/src/libultra/libc/llcvt.o: src/libultra/libc/llcvt.c
|
||||
$(CC_CHECK) $<
|
||||
$(CC) -c $(CFLAGS) $(MIPS_VERSION) $(OPTFLAGS) -o $@ $<
|
||||
python3 tools/set_o32abi_bit.py $@
|
||||
@$(OBJDUMP) -d $@ > $(@:.o=.s)
|
||||
@$(OBJDUMP) $(OBJDUMP_FLAGS) $@ > $(@:.o=.s)
|
||||
$(RM_MDEBUG)
|
||||
|
||||
# Build C files from assets
|
||||
|
|
|
|||
|
|
@ -5,4 +5,5 @@ def apply(config, args):
|
|||
config['myimg'] = 'mm.us.rev1.rom_uncompressed.z64'
|
||||
config['mapfile'] = 'build/mm.map'
|
||||
config['source_directories'] = ['./src','./include']
|
||||
config['objdump_flags'] = ['-M','reg-names=32']
|
||||
config['makeflags'] = ['KEEP_MDEBUG=1']
|
||||
|
|
|
|||
|
|
@ -37,3 +37,39 @@
|
|||
.set TagHi, $29
|
||||
.set ErrorEPC, $30
|
||||
.set Reserved31, $31
|
||||
|
||||
|
||||
# Float register aliases (o32 ABI, odd ones are rarely used)
|
||||
|
||||
.set $fv0, $f0
|
||||
.set $fv0f, $f1
|
||||
.set $fv1, $f2
|
||||
.set $fv1f, $f3
|
||||
.set $ft0, $f4
|
||||
.set $ft0f, $f5
|
||||
.set $ft1, $f6
|
||||
.set $ft1f, $f7
|
||||
.set $ft2, $f8
|
||||
.set $ft2f, $f9
|
||||
.set $ft3, $f10
|
||||
.set $ft3f, $f11
|
||||
.set $fa0, $f12
|
||||
.set $fa0f, $f13
|
||||
.set $fa1, $f14
|
||||
.set $fa1f, $f15
|
||||
.set $ft4, $f16
|
||||
.set $ft4f, $f17
|
||||
.set $ft5, $f18
|
||||
.set $ft5f, $f19
|
||||
.set $fs0, $f20
|
||||
.set $fs0f, $f21
|
||||
.set $fs1, $f22
|
||||
.set $fs1f, $f23
|
||||
.set $fs2, $f24
|
||||
.set $fs2f, $f25
|
||||
.set $fs3, $f26
|
||||
.set $fs3f, $f27
|
||||
.set $fs4, $f28
|
||||
.set $fs4f, $f29
|
||||
.set $fs5, $f30
|
||||
.set $fs5f, $f31
|
||||
|
|
|
|||
|
|
@ -6,7 +6,7 @@
|
|||
[subrepo]
|
||||
remote = https://github.com/simonlindholm/asm-differ
|
||||
branch = main
|
||||
commit = 6f8f80b719359d018a2b734288c977aae6538870
|
||||
parent = 91a6e9f647d2035eba281d286c78f089f70f269f
|
||||
commit = 1236288d1520335c2bfb672078fec65084d7cb5c
|
||||
parent = 2c5690701a350c7e7c3d6252dff925ad65d59910
|
||||
method = merge
|
||||
cmdver = 0.4.3
|
||||
|
|
|
|||
|
|
@ -1,6 +1,5 @@
|
|||
repos:
|
||||
- repo: https://github.com/psf/black
|
||||
rev: 20.8b1
|
||||
rev: 22.1.0
|
||||
hooks:
|
||||
- id: black
|
||||
language_version: python3.6
|
||||
|
|
|
|||
|
|
@ -11,10 +11,10 @@ Nice differ for assembly code. Currently supports MIPS, PPC, AArch64, and ARM32;
|
|||
|
||||
## Usage
|
||||
|
||||
Create a file `diff_settings.sh` in some directory (see the one in this repo for an example). Then from that directory, run
|
||||
Create a file `diff_settings.py` in some directory (see the one in this repo for an example). Then from that directory, run
|
||||
|
||||
```bash
|
||||
/path/to/diff.sh [flags] (function|rom addr)
|
||||
/path/to/diff.py [flags] (function|rom addr)
|
||||
```
|
||||
|
||||
Recommended flags are `-mwo` (automatically run `make` on source file changes, and include symbols in diff). See `--help` for more details.
|
||||
|
|
|
|||
|
|
@ -109,6 +109,15 @@ if __name__ == "__main__":
|
|||
help="""Diff .o files rather than a whole binary. This makes it possible to
|
||||
see symbol names. (Recommended)""",
|
||||
)
|
||||
parser.add_argument(
|
||||
"-f",
|
||||
"--objfile",
|
||||
dest="objfile",
|
||||
type=str,
|
||||
help="""File path for an object file being diffed. When used
|
||||
the map file isn't searched for the function given. Useful for dynamically
|
||||
linked libraries.""",
|
||||
)
|
||||
parser.add_argument(
|
||||
"-e",
|
||||
"--elf",
|
||||
|
|
@ -132,6 +141,14 @@ if __name__ == "__main__":
|
|||
help="""Tweak --source handling to make it work with binutils < 2.33.
|
||||
Implies --source.""",
|
||||
)
|
||||
parser.add_argument(
|
||||
"-j",
|
||||
"--section",
|
||||
dest="diff_section",
|
||||
default=".text",
|
||||
metavar="SECTION",
|
||||
help="Diff restricted to a given output section.",
|
||||
)
|
||||
parser.add_argument(
|
||||
"-L",
|
||||
"--line-numbers",
|
||||
|
|
@ -356,6 +373,7 @@ except ModuleNotFoundError as e:
|
|||
class ProjectSettings:
|
||||
arch_str: str
|
||||
objdump_executable: str
|
||||
objdump_flags: List[str]
|
||||
build_command: List[str]
|
||||
map_format: str
|
||||
mw_build_dir: str
|
||||
|
|
@ -365,6 +383,7 @@ class ProjectSettings:
|
|||
source_directories: Optional[List[str]]
|
||||
source_extensions: List[str]
|
||||
show_line_numbers_default: bool
|
||||
disassemble_all: bool
|
||||
|
||||
|
||||
@dataclass
|
||||
|
|
@ -379,8 +398,10 @@ class Config:
|
|||
|
||||
# Build/objdump options
|
||||
diff_obj: bool
|
||||
objfile: Optional[str]
|
||||
make: bool
|
||||
source_old_binutils: bool
|
||||
diff_section: str
|
||||
inlines: bool
|
||||
max_function_size_lines: int
|
||||
max_function_size_bytes: int
|
||||
|
|
@ -422,9 +443,11 @@ def create_project_settings(settings: Dict[str, Any]) -> ProjectSettings:
|
|||
"source_extensions", [".c", ".h", ".cpp", ".hpp", ".s"]
|
||||
),
|
||||
objdump_executable=get_objdump_executable(settings.get("objdump_executable")),
|
||||
objdump_flags=settings.get("objdump_flags", []),
|
||||
map_format=settings.get("map_format", "gnu"),
|
||||
mw_build_dir=settings.get("mw_build_dir", "build/"),
|
||||
show_line_numbers_default=settings.get("show_line_numbers_default", True),
|
||||
disassemble_all=settings.get("disassemble_all", False),
|
||||
)
|
||||
|
||||
|
||||
|
|
@ -461,8 +484,10 @@ def create_config(args: argparse.Namespace, project: ProjectSettings) -> Config:
|
|||
arch=arch,
|
||||
# Build/objdump options
|
||||
diff_obj=args.diff_obj,
|
||||
objfile=args.objfile,
|
||||
make=args.make,
|
||||
source_old_binutils=args.source_old_binutils,
|
||||
diff_section=args.diff_section,
|
||||
inlines=args.inlines,
|
||||
max_function_size_lines=args.max_lines,
|
||||
max_function_size_bytes=args.max_lines * 4,
|
||||
|
|
@ -518,7 +543,7 @@ def get_arch(arch_str: str) -> "ArchSettings":
|
|||
raise ValueError(f"Unknown architecture: {arch_str}")
|
||||
|
||||
|
||||
BUFFER_CMD: List[str] = ["tail", "-c", str(10 ** 9)]
|
||||
BUFFER_CMD: List[str] = ["tail", "-c", str(10**9)]
|
||||
|
||||
# -S truncates long lines instead of wrapping them
|
||||
# -R interprets color escape sequences
|
||||
|
|
@ -809,12 +834,7 @@ class JsonFormatter(Formatter):
|
|||
return {"text": s, "format": f.name.lower()}
|
||||
elif isinstance(f, RotationFormat):
|
||||
attrs = asdict(f)
|
||||
attrs.update(
|
||||
{
|
||||
"text": s,
|
||||
"format": "rotation",
|
||||
}
|
||||
)
|
||||
attrs.update({"text": s, "format": "rotation"})
|
||||
return attrs
|
||||
else:
|
||||
static_assert_unreachable(f)
|
||||
|
|
@ -962,7 +982,7 @@ def restrict_to_function(dump: str, fn_name: str) -> str:
|
|||
return ""
|
||||
|
||||
|
||||
def serialize_data_references(references: List[Tuple[int, int, str]]) -> str:
|
||||
def serialize_rodata_references(references: List[Tuple[int, int, str]]) -> str:
|
||||
return "".join(
|
||||
f"DATAREF {text_offset} {from_offset} {from_section}\n"
|
||||
for (text_offset, from_offset, from_section) in references
|
||||
|
|
@ -991,7 +1011,11 @@ def run_objdump(cmd: ObjdumpCommand, config: Config, project: ProjectSettings) -
|
|||
flags, target, restrict = cmd
|
||||
try:
|
||||
out = subprocess.run(
|
||||
[project.objdump_executable] + config.arch.arch_flags + flags + [target],
|
||||
[project.objdump_executable]
|
||||
+ config.arch.arch_flags
|
||||
+ project.objdump_flags
|
||||
+ flags
|
||||
+ [target],
|
||||
check=True,
|
||||
stdout=subprocess.PIPE,
|
||||
stderr=subprocess.PIPE,
|
||||
|
|
@ -1009,11 +1033,11 @@ def run_objdump(cmd: ObjdumpCommand, config: Config, project: ProjectSettings) -
|
|||
with open(target, "rb") as f:
|
||||
obj_data = f.read()
|
||||
|
||||
return preprocess_objdump_out(restrict, obj_data, out)
|
||||
return preprocess_objdump_out(restrict, obj_data, out, config)
|
||||
|
||||
|
||||
def preprocess_objdump_out(
|
||||
restrict: Optional[str], obj_data: Optional[bytes], objdump_out: str
|
||||
restrict: Optional[str], obj_data: Optional[bytes], objdump_out: str, config: Config
|
||||
) -> str:
|
||||
"""
|
||||
Preprocess the output of objdump into a format that `process()` expects.
|
||||
|
|
@ -1033,13 +1057,16 @@ def preprocess_objdump_out(
|
|||
out = out.rstrip("\n")
|
||||
|
||||
if obj_data:
|
||||
out = serialize_data_references(parse_elf_data_references(obj_data)) + out
|
||||
out = (
|
||||
serialize_rodata_references(parse_elf_rodata_references(obj_data, config))
|
||||
+ out
|
||||
)
|
||||
|
||||
return out
|
||||
|
||||
|
||||
def search_map_file(
|
||||
fn_name: str, project: ProjectSettings
|
||||
fn_name: str, project: ProjectSettings, config: Config
|
||||
) -> Tuple[Optional[str], Optional[int]]:
|
||||
if not project.mapfile:
|
||||
fail(f"No map file configured; cannot find function {fn_name}.")
|
||||
|
|
@ -1059,7 +1086,7 @@ def search_map_file(
|
|||
cands = []
|
||||
last_line = ""
|
||||
for line in lines:
|
||||
if line.startswith(" .text"):
|
||||
if line.startswith(" " + config.diff_section):
|
||||
cur_objfile = line.split()[3]
|
||||
if "load address" in line:
|
||||
tokens = last_line.split() + line.split()
|
||||
|
|
@ -1084,9 +1111,12 @@ def search_map_file(
|
|||
re.compile(
|
||||
# ram elf rom
|
||||
r" \S+ \S+ (\S+) (\S+) . "
|
||||
+ fn_name
|
||||
# object name
|
||||
+ r"(?: \(entry of \.(?:init|text)\))? \t(\S+)"
|
||||
+ re.escape(fn_name)
|
||||
+ r"(?: \(entry of "
|
||||
+ re.escape(config.diff_section)
|
||||
+ r"\))? \t"
|
||||
# object name
|
||||
+ "(\S+)"
|
||||
),
|
||||
contents,
|
||||
)
|
||||
|
|
@ -1121,7 +1151,9 @@ def search_map_file(
|
|||
return None, None
|
||||
|
||||
|
||||
def parse_elf_data_references(data: bytes) -> List[Tuple[int, int, str]]:
|
||||
def parse_elf_rodata_references(
|
||||
data: bytes, config: Config
|
||||
) -> List[Tuple[int, int, str]]:
|
||||
e_ident = data[:16]
|
||||
if e_ident[:4] != b"\x7FELF":
|
||||
return []
|
||||
|
|
@ -1134,7 +1166,6 @@ def parse_elf_data_references(data: bytes) -> List[Tuple[int, int, str]]:
|
|||
is_little_endian = e_ident[5] == 1
|
||||
str_end = "<" if is_little_endian else ">"
|
||||
str_off = "I" if is_32bit else "Q"
|
||||
sym_size = {"B": 1, "H": 2, "I": 4, "Q": 8}
|
||||
|
||||
def read(spec: str, offset: int) -> Tuple[int, ...]:
|
||||
spec = spec.replace("P", str_off)
|
||||
|
|
@ -1186,7 +1217,12 @@ def parse_elf_data_references(data: bytes) -> List[Tuple[int, int, str]]:
|
|||
assert len(symtab_sections) == 1
|
||||
symtab = sections[symtab_sections[0]]
|
||||
|
||||
text_sections = [i for i in range(e_shnum) if sec_names[i] == b".text" and sections[i].sh_size != 0]
|
||||
section_name = config.diff_section.encode("utf-8")
|
||||
text_sections = [
|
||||
i
|
||||
for i in range(e_shnum)
|
||||
if sec_names[i] == section_name and sections[i].sh_size != 0
|
||||
]
|
||||
if len(text_sections) != 1:
|
||||
return []
|
||||
text_section = text_sections[0]
|
||||
|
|
@ -1195,11 +1231,10 @@ def parse_elf_data_references(data: bytes) -> List[Tuple[int, int, str]]:
|
|||
for s in sections:
|
||||
if s.sh_type == SHT_REL or s.sh_type == SHT_RELA:
|
||||
if s.sh_info == text_section:
|
||||
# Skip .text -> .text references
|
||||
# Skip section_name -> section_name references
|
||||
continue
|
||||
sec_name = sec_names[s.sh_info].decode("latin1")
|
||||
if sec_name == ".mwcats.text":
|
||||
# Skip Metrowerks CATS Utility section
|
||||
if sec_name != ".rodata":
|
||||
continue
|
||||
sec_base = sections[s.sh_info].sh_offset
|
||||
for i in range(0, s.sh_size, s.sh_entsize):
|
||||
|
|
@ -1257,11 +1292,16 @@ def dump_elf(
|
|||
f"--stop-address={end_addr}",
|
||||
]
|
||||
|
||||
if project.disassemble_all:
|
||||
disassemble_flag = "-D"
|
||||
else:
|
||||
disassemble_flag = "-d"
|
||||
|
||||
flags2 = [
|
||||
f"--disassemble={diff_elf_symbol}",
|
||||
]
|
||||
|
||||
objdump_flags = ["-drz", "-j", ".text"]
|
||||
objdump_flags = [disassemble_flag, "-rz", "-j", config.diff_section]
|
||||
return (
|
||||
project.myimg,
|
||||
(objdump_flags + flags1, project.baseimg, None),
|
||||
|
|
@ -1283,7 +1323,10 @@ def dump_objfile(
|
|||
if start.startswith("0"):
|
||||
fail("numerical start address not supported with -o; pass a function name")
|
||||
|
||||
objfile, _ = search_map_file(start, project)
|
||||
objfile = config.objfile
|
||||
if not objfile:
|
||||
objfile, _ = search_map_file(start, project, config)
|
||||
|
||||
if not objfile:
|
||||
fail("Not able to find .o file for function.")
|
||||
|
||||
|
|
@ -1297,7 +1340,12 @@ def dump_objfile(
|
|||
if not os.path.isfile(refobjfile):
|
||||
fail(f'Please ensure an OK .o file exists at "{refobjfile}".')
|
||||
|
||||
objdump_flags = ["-drz", "-j", ".text"]
|
||||
if project.disassemble_all:
|
||||
disassemble_flag = "-D"
|
||||
else:
|
||||
disassemble_flag = "-d"
|
||||
|
||||
objdump_flags = [disassemble_flag, "-rz", "-j", config.diff_section]
|
||||
return (
|
||||
objfile,
|
||||
(objdump_flags, refobjfile, start),
|
||||
|
|
@ -1314,7 +1362,7 @@ def dump_binary(
|
|||
run_make(project.myimg, project)
|
||||
start_addr = maybe_eval_int(start)
|
||||
if start_addr is None:
|
||||
_, start_addr = search_map_file(start, project)
|
||||
_, start_addr = search_map_file(start, project, config)
|
||||
if start_addr is None:
|
||||
fail("Not able to find function in map file.")
|
||||
if end is not None:
|
||||
|
|
@ -1333,6 +1381,7 @@ def dump_binary(
|
|||
(objdump_flags + flags2, project.myimg, None),
|
||||
)
|
||||
|
||||
|
||||
# Example: "ldr r4, [pc, #56] ; (4c <AddCoins+0x4c>)"
|
||||
ARM32_LOAD_POOL_PATTERN = r"(ldr\s+r([0-9]|1[0-3]),\s+\[pc,.*;\s*)(\([a-fA-F0-9]+.*\))"
|
||||
|
||||
|
|
@ -1363,7 +1412,7 @@ class AsmProcessor:
|
|||
class AsmProcessorMIPS(AsmProcessor):
|
||||
def process_reloc(self, row: str, prev: str) -> str:
|
||||
arch = self.config.arch
|
||||
if "R_MIPS_NONE" in row:
|
||||
if "R_MIPS_NONE" in row or "R_MIPS_JALR" in row:
|
||||
# GNU as emits no-op relocations immediately after real ones when
|
||||
# assembling with -mabi=64. Return without trying to parse 'imm' as an
|
||||
# integer.
|
||||
|
|
@ -1378,12 +1427,9 @@ class AsmProcessorMIPS(AsmProcessor):
|
|||
# TODO: handle unambiguous cases where all addends for a symbol are the
|
||||
# same, or show "+???".
|
||||
mnemonic = prev.split()[0]
|
||||
if (
|
||||
mnemonic in arch.instructions_with_address_immediates
|
||||
and not imm.startswith("0x")
|
||||
):
|
||||
imm = "0x" + imm
|
||||
repl += "+" + imm if int(imm, 0) > 0 else imm
|
||||
if mnemonic in arch.instructions_with_address_immediates:
|
||||
imm = hex(int(imm, 16))
|
||||
repl += ("" if imm.startswith("-") else "+") + imm
|
||||
if "R_MIPS_LO16" in row:
|
||||
repl = f"%lo({repl})"
|
||||
elif "R_MIPS_HI16" in row:
|
||||
|
|
@ -1398,6 +1444,12 @@ class AsmProcessorMIPS(AsmProcessor):
|
|||
# Branch to glabel. This gives confusing output, but there's not much
|
||||
# we can do here.
|
||||
pass
|
||||
elif "R_MIPS_GPREL16" in row:
|
||||
repl = f"%gp_rel({repl})"
|
||||
elif "R_MIPS_GOT16" in row:
|
||||
repl = f"%got({repl})"
|
||||
elif "R_MIPS_CALL16" in row:
|
||||
repl = f"%call16({repl})"
|
||||
else:
|
||||
assert False, f"unknown relocation type '{row}' for line '{prev}'"
|
||||
return before + repl + after
|
||||
|
|
@ -1544,6 +1596,7 @@ class ArchSettings:
|
|||
big_endian: Optional[bool] = True
|
||||
delay_slot_instructions: Set[str] = field(default_factory=set)
|
||||
|
||||
|
||||
MIPS_BRANCH_LIKELY_INSTRUCTIONS = {
|
||||
"beql",
|
||||
"bnel",
|
||||
|
|
@ -1651,10 +1704,13 @@ PPC_BRANCH_INSTRUCTIONS = {
|
|||
MIPS_SETTINGS = ArchSettings(
|
||||
name="mips",
|
||||
re_int=re.compile(r"[0-9]+"),
|
||||
re_comment=re.compile(r"<.*?>"),
|
||||
re_reg=re.compile(
|
||||
r"\$?\b(a[0-7]|t[0-9]|s[0-8]|at|v[01]|f[12]?[0-9]|f3[01]|kt?[01]|fp|ra|zero)\b"
|
||||
),
|
||||
re_comment=re.compile(r"<.*>"),
|
||||
# Includes:
|
||||
# - General purpose registers v0..1, a0..7, t0..9, s0..8, zero, at, fp, k0..1/kt0..1
|
||||
# - Float registers f0..31, or fv0..1, fa0..7, ft0..15, fs0..8 plus odd complements
|
||||
# (actually used number depends on ABI)
|
||||
# sp, gp should not be in this list
|
||||
re_reg=re.compile(r"\$?\b([astv][0-9]|at|f[astv]?[0-9]+f?|kt?[01]|fp|ra|zero)\b"),
|
||||
re_sprel=re.compile(r"(?<=,)([0-9]+|0x[0-9a-f]+)\(sp\)"),
|
||||
re_large_imm=re.compile(r"-?[1-9][0-9]{2,}|-?0x[0-9a-f]{3,}"),
|
||||
re_imm=re.compile(r"(\b|-)([0-9]+|0x[0-9a-fA-F]+)\b(?!\(sp)|%(lo|hi)\([^)]*\)"),
|
||||
|
|
@ -1672,7 +1728,7 @@ MIPSEL_SETTINGS = replace(MIPS_SETTINGS, name="mipsel", big_endian=False)
|
|||
ARM32_SETTINGS = ArchSettings(
|
||||
name="arm32",
|
||||
re_int=re.compile(r"[0-9]+"),
|
||||
re_comment=re.compile(r"(<.*?>|//.*$)"),
|
||||
re_comment=re.compile(r"(<.*>|//.*$)"),
|
||||
# Includes:
|
||||
# - General purpose registers: r0..13
|
||||
# - Frame pointer registers: lr (r14), pc (r15)
|
||||
|
|
@ -1690,26 +1746,33 @@ ARM32_SETTINGS = ArchSettings(
|
|||
proc=AsmProcessorARM32,
|
||||
)
|
||||
|
||||
ARMEL_SETTINGS = replace(ARM32_SETTINGS, name="armel", big_endian=False)
|
||||
|
||||
AARCH64_SETTINGS = ArchSettings(
|
||||
name="aarch64",
|
||||
re_int=re.compile(r"[0-9]+"),
|
||||
re_comment=re.compile(r"(<.*?>|//.*$)"),
|
||||
# GPRs and FP registers: X0-X30, W0-W30, [DSHQ]0..31
|
||||
re_comment=re.compile(r"(<.*>|//.*$)"),
|
||||
# GPRs and FP registers: X0-X30, W0-W30, [BHSDVQ]0..31
|
||||
# (FP registers may be followed by data width and number of elements, e.g. V0.4S)
|
||||
# The zero registers and SP should not be in this list.
|
||||
re_reg=re.compile(r"\$?\b([dshq][12]?[0-9]|[dshq]3[01]|[xw][12]?[0-9]|[xw]30)\b"),
|
||||
re_reg=re.compile(
|
||||
r"\$?\b([bhsdvq]([12]?[0-9]|3[01])(\.\d\d?[bhsdvq])?|[xw][12]?[0-9]|[xw]30)\b"
|
||||
),
|
||||
re_sprel=re.compile(r"sp, #-?(0x[0-9a-fA-F]+|[0-9]+)\b"),
|
||||
re_large_imm=re.compile(r"-?[1-9][0-9]{2,}|-?0x[0-9a-f]{3,}"),
|
||||
re_imm=re.compile(r"(?<!sp, )#-?(0x[0-9a-fA-F]+|[0-9]+)\b"),
|
||||
re_reloc=re.compile(r"R_AARCH64_"),
|
||||
branch_instructions=AARCH64_BRANCH_INSTRUCTIONS,
|
||||
instructions_with_address_immediates=AARCH64_BRANCH_INSTRUCTIONS.union({"bl", "adrp"}),
|
||||
instructions_with_address_immediates=AARCH64_BRANCH_INSTRUCTIONS.union(
|
||||
{"bl", "adrp"}
|
||||
),
|
||||
proc=AsmProcessorAArch64,
|
||||
)
|
||||
|
||||
PPC_SETTINGS = ArchSettings(
|
||||
name="ppc",
|
||||
re_int=re.compile(r"[0-9]+"),
|
||||
re_comment=re.compile(r"(<.*?>|//.*$)"),
|
||||
re_comment=re.compile(r"(<.*>|//.*$)"),
|
||||
re_reg=re.compile(r"\$?\b([rf][0-9]+)\b"),
|
||||
re_sprel=re.compile(r"(?<=,)(-?[0-9]+|-?0x[0-9a-f]+)\(r1\)"),
|
||||
re_large_imm=re.compile(r"-?[1-9][0-9]{2,}|-?0x[0-9a-f]{3,}"),
|
||||
|
|
@ -1724,6 +1787,7 @@ ARCH_SETTINGS = [
|
|||
MIPS_SETTINGS,
|
||||
MIPSEL_SETTINGS,
|
||||
ARM32_SETTINGS,
|
||||
ARMEL_SETTINGS,
|
||||
AARCH64_SETTINGS,
|
||||
PPC_SETTINGS,
|
||||
]
|
||||
|
|
@ -1743,6 +1807,7 @@ def hexify_int(row: str, pat: Match[str], arch: ArchSettings) -> str:
|
|||
|
||||
|
||||
def parse_relocated_line(line: str) -> Tuple[str, str, str]:
|
||||
# Pick out the last argument
|
||||
for c in ",\t ":
|
||||
if c in line:
|
||||
ind2 = line.rindex(c)
|
||||
|
|
@ -1751,6 +1816,7 @@ def parse_relocated_line(line: str) -> Tuple[str, str, str]:
|
|||
raise Exception(f"failed to parse relocated line: {line}")
|
||||
before = line[: ind2 + 1]
|
||||
after = line[ind2 + 1 :]
|
||||
# Move an optional ($reg) part of it to 'after'
|
||||
ind2 = after.find("(")
|
||||
if ind2 == -1:
|
||||
imm, after = after, ""
|
||||
|
|
@ -1879,8 +1945,15 @@ def process(dump: str, config: Config) -> List[Line]:
|
|||
row_parts = [part.lstrip() for part in row.split(" ", 1)]
|
||||
mnemonic = row_parts[0].strip()
|
||||
|
||||
if mnemonic not in arch.instructions_with_address_immediates:
|
||||
row = re.sub(arch.re_int, lambda m: hexify_int(row, m, arch), row)
|
||||
addr = ""
|
||||
if mnemonic in arch.instructions_with_address_immediates:
|
||||
row, addr = split_off_address(row)
|
||||
# objdump prefixes addresses with 0x/-0x if they don't resolve to some
|
||||
# symbol + offset. Strip that.
|
||||
addr = addr.replace("0x", "")
|
||||
|
||||
row = re.sub(arch.re_int, lambda m: hexify_int(row, m, arch), row)
|
||||
row += addr
|
||||
|
||||
# Let 'original' be 'row' with relocations applied, while we continue
|
||||
# transforming 'row' into a coarser version that ignores registers and
|
||||
|
|
@ -1900,9 +1973,6 @@ def process(dump: str, config: Config) -> List[Line]:
|
|||
scorable_line = normalized_original
|
||||
if not config.score_stack_differences:
|
||||
scorable_line = re.sub(arch.re_sprel, "addr(sp)", scorable_line)
|
||||
if mnemonic in arch.branch_instructions:
|
||||
# Replace the final argument with "<target>"
|
||||
scorable_line = re.sub(r"[^, \t]+$", "<target>", scorable_line)
|
||||
|
||||
if skip_next:
|
||||
skip_next = False
|
||||
|
|
@ -1925,8 +1995,6 @@ def process(dump: str, config: Config) -> List[Line]:
|
|||
branch_target = None
|
||||
if mnemonic in arch.branch_instructions:
|
||||
branch_target = int(row_parts[1].strip().split(",")[-1], 16)
|
||||
if mnemonic in arch.branch_likely_instructions:
|
||||
branch_target -= 4
|
||||
|
||||
output.append(
|
||||
Line(
|
||||
|
|
@ -1974,7 +2042,9 @@ def split_off_address(line: str) -> Tuple[str, str]:
|
|||
parts = line.split(",")
|
||||
if len(parts) < 2:
|
||||
parts = line.split(None, 1)
|
||||
off = len(line) - len(parts[-1])
|
||||
if len(parts) < 2:
|
||||
parts.append("")
|
||||
off = len(line) - len(parts[-1].strip())
|
||||
return line[:off], line[off:]
|
||||
|
||||
|
||||
|
|
@ -1990,7 +2060,7 @@ def diff_sequences(
|
|||
) -> List[Tuple[str, int, int, int, int]]:
|
||||
if (
|
||||
algorithm != "levenshtein"
|
||||
or len(seq1) * len(seq2) > 4 * 10 ** 8
|
||||
or len(seq1) * len(seq2) > 4 * 10**8
|
||||
or len(seq1) + len(seq2) >= 0x110000
|
||||
):
|
||||
return diff_sequences_difflib(seq1, seq2)
|
||||
|
|
@ -2177,10 +2247,15 @@ class Diff:
|
|||
|
||||
def trim_nops(lines: List[Line], arch: ArchSettings) -> List[Line]:
|
||||
lines = lines[:]
|
||||
while lines and lines[-1].mnemonic == "nop" and (len(lines) == 1 or lines[-2].mnemonic not in arch.delay_slot_instructions):
|
||||
while (
|
||||
lines
|
||||
and lines[-1].mnemonic == "nop"
|
||||
and (len(lines) == 1 or lines[-2].mnemonic not in arch.delay_slot_instructions)
|
||||
):
|
||||
lines.pop()
|
||||
return lines
|
||||
|
||||
|
||||
def do_diff(lines1: List[Line], lines2: List[Line], config: Config) -> Diff:
|
||||
if config.show_source:
|
||||
import cxxfilt
|
||||
|
|
@ -2212,7 +2287,6 @@ def do_diff(lines1: List[Line], lines2: List[Line], config: Config) -> Diff:
|
|||
lines2 = trim_nops(lines2, arch)
|
||||
|
||||
diffed_lines = diff_lines(lines1, lines2, config.algorithm)
|
||||
score = score_diff_lines(diffed_lines, config)
|
||||
max_score = len(lines1) * config.penalty_deletion
|
||||
|
||||
line_num_base = -1
|
||||
|
|
@ -2281,10 +2355,15 @@ def do_diff(lines1: List[Line], lines2: List[Line], config: Config) -> Diff:
|
|||
line2_line = line_num_2to1[line2.line_num]
|
||||
line2_target = (line2_line[0] + (target - line2.line_num), 0)
|
||||
|
||||
# Set the key for three-way diffing to a normalized version.
|
||||
# Adjust the branch target for scoring and three-way diffing.
|
||||
norm2, norm_branch2 = split_off_address(line2.normalized_original)
|
||||
if norm_branch2 != "<ign>":
|
||||
line2.normalized_original = norm2 + str(line2_target)
|
||||
if norm_branch2 != "<ignore>":
|
||||
retargetted = hex(line2_target[0]).replace("0x", "")
|
||||
if line2_target[1] != 0:
|
||||
retargetted += f"+{line2_target[1]}"
|
||||
line2.normalized_original = norm2 + retargetted
|
||||
sc_base, _ = split_off_address(line2.scorable_line)
|
||||
line2.scorable_line = sc_base + retargetted
|
||||
same_target = line2_target == (line1.branch_target, 0)
|
||||
else:
|
||||
# Do a naive comparison for non-branches (e.g. function calls).
|
||||
|
|
@ -2444,6 +2523,7 @@ def do_diff(lines1: List[Line], lines2: List[Line], config: Config) -> Diff:
|
|||
)
|
||||
)
|
||||
|
||||
score = score_diff_lines(diffed_lines, config)
|
||||
output = output[config.skip_lines :]
|
||||
return Diff(lines=output, score=score, max_score=max_score)
|
||||
|
||||
|
|
|
|||
|
|
@ -783,6 +783,13 @@ def parse_source(f, opt, framepointer, mips1, input_enc, output_enc, out_depende
|
|||
else:
|
||||
min_instr_count = 2
|
||||
skip_instr_count = 1
|
||||
elif opt == 'O0':
|
||||
if framepointer:
|
||||
min_instr_count = 8
|
||||
skip_instr_count = 8
|
||||
else:
|
||||
min_instr_count = 4
|
||||
skip_instr_count = 4
|
||||
elif opt == 'g':
|
||||
if framepointer:
|
||||
min_instr_count = 7
|
||||
|
|
@ -792,7 +799,7 @@ def parse_source(f, opt, framepointer, mips1, input_enc, output_enc, out_depende
|
|||
skip_instr_count = 4
|
||||
else:
|
||||
if opt != 'g3':
|
||||
raise Failure("must pass one of -g, -O1, -O2, -O2 -g3")
|
||||
raise Failure("must pass one of -g, -O0, -O1, -O2, -O2 -g3")
|
||||
if framepointer:
|
||||
min_instr_count = 4
|
||||
skip_instr_count = 4
|
||||
|
|
@ -813,6 +820,7 @@ def parse_source(f, opt, framepointer, mips1, input_enc, output_enc, out_depende
|
|||
]
|
||||
|
||||
is_cutscene_data = False
|
||||
is_early_include = False
|
||||
|
||||
for line_no, raw_line in enumerate(f, 1):
|
||||
raw_line = raw_line.rstrip()
|
||||
|
|
@ -832,44 +840,51 @@ def parse_source(f, opt, framepointer, mips1, input_enc, output_enc, out_depende
|
|||
global_asm = None
|
||||
else:
|
||||
global_asm.process_line(raw_line, output_enc)
|
||||
elif line in ['GLOBAL_ASM(', '#pragma GLOBAL_ASM(']:
|
||||
global_asm = GlobalAsmBlock("GLOBAL_ASM block at line " + str(line_no))
|
||||
start_index = len(output_lines)
|
||||
elif ((line.startswith('GLOBAL_ASM("') or line.startswith('#pragma GLOBAL_ASM("'))
|
||||
and line.endswith('")')):
|
||||
fname = line[line.index('(') + 2 : -2]
|
||||
out_dependencies.append(fname)
|
||||
global_asm = GlobalAsmBlock(fname)
|
||||
with open(fname, encoding=input_enc) as f:
|
||||
for line2 in f:
|
||||
global_asm.process_line(line2.rstrip(), output_enc)
|
||||
src, fn = global_asm.finish(state)
|
||||
output_lines[-1] = ''.join(src)
|
||||
asm_functions.append(fn)
|
||||
global_asm = None
|
||||
elif line == '#pragma asmproc recurse':
|
||||
# C includes qualified as
|
||||
# #pragma asmproc recurse
|
||||
# #include "file.c"
|
||||
# will be processed recursively when encountered
|
||||
is_early_include = True
|
||||
elif is_early_include:
|
||||
# Previous line was a #pragma asmproc recurse
|
||||
is_early_include = False
|
||||
if not line.startswith("#include "):
|
||||
raise Failure("#pragma asmproc recurse must be followed by an #include ")
|
||||
fpath = os.path.dirname(f.name)
|
||||
fname = os.path.join(fpath, line[line.index(' ') + 2 : -1])
|
||||
out_dependencies.append(fname)
|
||||
include_src = StringIO()
|
||||
with open(fname, encoding=input_enc) as include_file:
|
||||
parse_source(include_file, opt, framepointer, mips1, input_enc, output_enc, out_dependencies, include_src)
|
||||
include_src.write('#line ' + str(line_no + 1) + ' "' + f.name + '"')
|
||||
output_lines[-1] = include_src.getvalue()
|
||||
include_src.close()
|
||||
else:
|
||||
if line in ['GLOBAL_ASM(', '#pragma GLOBAL_ASM(']:
|
||||
global_asm = GlobalAsmBlock("GLOBAL_ASM block at line " + str(line_no))
|
||||
start_index = len(output_lines)
|
||||
elif ((line.startswith('GLOBAL_ASM("') or line.startswith('#pragma GLOBAL_ASM("'))
|
||||
and line.endswith('")')):
|
||||
fname = line[line.index('(') + 2 : -2]
|
||||
out_dependencies.append(fname)
|
||||
global_asm = GlobalAsmBlock(fname)
|
||||
with open(fname, encoding=input_enc) as f:
|
||||
for line2 in f:
|
||||
global_asm.process_line(line2.rstrip(), output_enc)
|
||||
src, fn = global_asm.finish(state)
|
||||
output_lines[-1] = ''.join(src)
|
||||
asm_functions.append(fn)
|
||||
global_asm = None
|
||||
elif line.startswith('#include "') and line.endswith('" EARLY'):
|
||||
# C includes qualified with EARLY (i.e. #include "file.c" EARLY) will be
|
||||
# processed recursively when encountered
|
||||
fpath = os.path.dirname(f.name)
|
||||
fname = os.path.join(fpath, line[line.index(' ') + 2 : -7])
|
||||
out_dependencies.append(fname)
|
||||
include_src = StringIO()
|
||||
with open(fname, encoding=input_enc) as include_file:
|
||||
parse_source(include_file, opt, framepointer, mips1, input_enc, output_enc, out_dependencies, include_src)
|
||||
include_src.write('#line ' + str(line_no + 1) + ' "' + f.name + '"')
|
||||
output_lines[-1] = include_src.getvalue()
|
||||
include_src.close()
|
||||
else:
|
||||
# This is a hack to replace all floating-point numbers in an array of a particular type
|
||||
# (in this case CutsceneData) with their corresponding IEEE-754 hexadecimal representation
|
||||
if cutscene_data_regexpr.search(line) is not None:
|
||||
is_cutscene_data = True
|
||||
elif line.endswith("};"):
|
||||
is_cutscene_data = False
|
||||
if is_cutscene_data:
|
||||
raw_line = re.sub(float_regexpr, repl_float_hex, raw_line)
|
||||
output_lines[-1] = raw_line
|
||||
# This is a hack to replace all floating-point numbers in an array of a particular type
|
||||
# (in this case CutsceneData) with their corresponding IEEE-754 hexadecimal representation
|
||||
if cutscene_data_regexpr.search(line) is not None:
|
||||
is_cutscene_data = True
|
||||
elif line.endswith("};"):
|
||||
is_cutscene_data = False
|
||||
if is_cutscene_data:
|
||||
raw_line = re.sub(float_regexpr, repl_float_hex, raw_line)
|
||||
output_lines[-1] = raw_line
|
||||
|
||||
if print_source:
|
||||
if isinstance(print_source, StringIO):
|
||||
|
|
@ -877,7 +892,14 @@ def parse_source(f, opt, framepointer, mips1, input_enc, output_enc, out_depende
|
|||
print_source.write(line + '\n')
|
||||
else:
|
||||
for line in output_lines:
|
||||
print_source.write(line.encode(output_enc) + b'\n')
|
||||
try:
|
||||
line_encoded = line.encode(output_enc)
|
||||
except UnicodeEncodeError:
|
||||
print("Failed to encode a line to", output_enc)
|
||||
print("The line:", line)
|
||||
print("The line, utf-8-encoded:", line.encode("utf-8"))
|
||||
raise
|
||||
print_source.write(line_encoded + b'\n')
|
||||
print_source.flush()
|
||||
if print_source != sys.stdout.buffer:
|
||||
print_source.close()
|
||||
|
|
@ -1239,6 +1261,7 @@ def run_wrapped(argv, outfile, functions):
|
|||
parser.add_argument('-mips1', dest='mips1', action='store_true')
|
||||
parser.add_argument('-g3', dest='g3', action='store_true')
|
||||
group = parser.add_mutually_exclusive_group(required=True)
|
||||
group.add_argument('-O0', dest='opt', action='store_const', const='O0')
|
||||
group.add_argument('-O1', dest='opt', action='store_const', const='O1')
|
||||
group.add_argument('-O2', dest='opt', action='store_const', const='O2')
|
||||
group.add_argument('-g', dest='opt', action='store_const', const='g')
|
||||
|
|
|
|||
|
|
@ -10,30 +10,49 @@ dir_path = os.path.dirname(os.path.realpath(__file__))
|
|||
prelude = os.path.join(dir_path, "prelude.inc")
|
||||
|
||||
all_args = sys.argv[1:]
|
||||
sep1 = all_args.index('--')
|
||||
sep2 = all_args.index('--', sep1+1)
|
||||
sep0 = [index for index, arg in enumerate(all_args) if not arg.startswith("-")][0]
|
||||
sep1 = all_args.index("--")
|
||||
sep2 = all_args.index("--", sep1 + 1)
|
||||
|
||||
compiler = all_args[:sep1]
|
||||
asmproc_flags = all_args[:sep0]
|
||||
compiler = all_args[sep0:sep1]
|
||||
|
||||
assembler = all_args[sep1+1:sep2]
|
||||
assembler_sh = ' '.join(shlex.quote(x) for x in assembler)
|
||||
assembler = all_args[sep1 + 1 : sep2]
|
||||
assembler_sh = " ".join(shlex.quote(x) for x in assembler)
|
||||
|
||||
compile_args = all_args[sep2+1:]
|
||||
compile_args = all_args[sep2 + 1 :]
|
||||
in_file = compile_args[-1]
|
||||
out_ind = compile_args.index('-o')
|
||||
out_ind = compile_args.index("-o")
|
||||
out_file = compile_args[out_ind + 1]
|
||||
del compile_args[-1]
|
||||
del compile_args[out_ind + 1]
|
||||
del compile_args[out_ind]
|
||||
|
||||
in_dir = os.path.split(os.path.realpath(in_file))[0]
|
||||
opt_flags = [x for x in compile_args if x in ['-g3', '-g', '-O1', '-O2', '-framepointer']]
|
||||
opt_flags = [
|
||||
x for x in compile_args if x in ["-g3", "-g", "-O0", "-O1", "-O2", "-framepointer"]
|
||||
]
|
||||
if "-mips2" not in compile_args:
|
||||
opt_flags.append("-mips1")
|
||||
|
||||
preprocessed_file = tempfile.NamedTemporaryFile(prefix='preprocessed', suffix='.c', delete=False)
|
||||
asmproc_flags += opt_flags + [in_file]
|
||||
|
||||
# Drop .mdebug and .gptab sections from resulting binaries. This makes
|
||||
# resulting .o files much smaller and speeds up builds, but loses line
|
||||
# number debug data.
|
||||
# asmproc_flags += ["--drop-mdebug-gptab"]
|
||||
|
||||
# Convert encoding before compiling.
|
||||
# asmproc_flags += ["--input-enc", "utf-8", "--output-enc", "euc-jp"]
|
||||
|
||||
preprocessed_file = tempfile.NamedTemporaryFile(
|
||||
prefix="preprocessed", suffix=".c", delete=False
|
||||
)
|
||||
|
||||
try:
|
||||
asmproc_flags = opt_flags + [in_file, '--input-enc', 'utf-8', '--output-enc', 'euc-jp']
|
||||
compile_cmdline = compiler + compile_args + ['-I', in_dir, '-o', out_file, preprocessed_file.name]
|
||||
compile_cmdline = (
|
||||
compiler + compile_args + ["-I", in_dir, "-o", out_file, preprocessed_file.name]
|
||||
)
|
||||
|
||||
functions, deps = asm_processor.run(asmproc_flags, outfile=preprocessed_file)
|
||||
try:
|
||||
|
|
@ -41,13 +60,24 @@ try:
|
|||
except subprocess.CalledProcessError as e:
|
||||
print("Failed to compile file " + in_file + ". Command line:")
|
||||
print()
|
||||
print(' '.join(shlex.quote(x) for x in compile_cmdline))
|
||||
print(" ".join(shlex.quote(x) for x in compile_cmdline))
|
||||
print()
|
||||
sys.exit(55)
|
||||
# To keep the preprocessed file:
|
||||
# os._exit(1)
|
||||
|
||||
asm_processor.run(asmproc_flags + ['--post-process', out_file, '--assembler', assembler_sh, '--asm-prelude', prelude], functions=functions)
|
||||
asm_processor.run(
|
||||
asmproc_flags
|
||||
+ [
|
||||
"--post-process",
|
||||
out_file,
|
||||
"--assembler",
|
||||
assembler_sh,
|
||||
"--asm-prelude",
|
||||
prelude,
|
||||
],
|
||||
functions=functions,
|
||||
)
|
||||
|
||||
deps_file = out_file[:-2] + ".asmproc.d"
|
||||
if deps:
|
||||
|
|
|
|||
|
|
@ -1,5 +1,43 @@
|
|||
.set noat
|
||||
.set noreorder
|
||||
.set gp=64
|
||||
.include "macro.inc"
|
||||
.macro glabel label
|
||||
.global \label
|
||||
\label:
|
||||
.endm
|
||||
|
||||
|
||||
# Float register aliases (o32 ABI, odd ones are rarely used)
|
||||
|
||||
.set $fv0, $f0
|
||||
.set $fv0f, $f1
|
||||
.set $fv1, $f2
|
||||
.set $fv1f, $f3
|
||||
.set $ft0, $f4
|
||||
.set $ft0f, $f5
|
||||
.set $ft1, $f6
|
||||
.set $ft1f, $f7
|
||||
.set $ft2, $f8
|
||||
.set $ft2f, $f9
|
||||
.set $ft3, $f10
|
||||
.set $ft3f, $f11
|
||||
.set $fa0, $f12
|
||||
.set $fa0f, $f13
|
||||
.set $fa1, $f14
|
||||
.set $fa1f, $f15
|
||||
.set $ft4, $f16
|
||||
.set $ft4f, $f17
|
||||
.set $ft5, $f18
|
||||
.set $ft5f, $f19
|
||||
.set $fs0, $f20
|
||||
.set $fs0f, $f21
|
||||
.set $fs1, $f22
|
||||
.set $fs1f, $f23
|
||||
.set $fs2, $f24
|
||||
.set $fs2f, $f25
|
||||
.set $fs3, $f26
|
||||
.set $fs3f, $f27
|
||||
.set $fs4, $f28
|
||||
.set $fs4f, $f29
|
||||
.set $fs5, $f30
|
||||
.set $fs5f, $f31
|
||||
|
|
|
|||
|
|
@ -2,34 +2,44 @@
|
|||
|
||||
import argparse, ast, math, os, re, struct
|
||||
import bisect
|
||||
from mips_isa import *
|
||||
from multiprocessing import *
|
||||
import multiprocessing
|
||||
from pathlib import Path
|
||||
import mips_isa
|
||||
|
||||
# Consider implementing gpr naming too, but already uses abi names by default
|
||||
fpr_name_options = {
|
||||
"numeric": mips_isa.numeric_fpr_names,
|
||||
"o32": mips_isa.o32_fpr_names,
|
||||
}
|
||||
|
||||
parser = argparse.ArgumentParser()
|
||||
parser.add_argument(
|
||||
"-j", dest="jobs", type=int, default=1, help="number of processes to run at once"
|
||||
)
|
||||
parser.add_argument(
|
||||
"--full",
|
||||
"-f",
|
||||
dest="full",
|
||||
"-a",
|
||||
"--all",
|
||||
dest="all",
|
||||
action="store_true",
|
||||
default=False,
|
||||
help="Decompile all files regardless of whether they are used or not",
|
||||
)
|
||||
parser.add_argument(
|
||||
"-files",
|
||||
"-f",
|
||||
"--files",
|
||||
dest="files",
|
||||
nargs="+",
|
||||
required=False,
|
||||
help="Optional list of files to diassemble separated by a space. This is a whitelist, all files will be skipped besides the ones listed here if used.",
|
||||
)
|
||||
parser.add_argument("--reg-names", choices=fpr_name_options.keys(), help="How to name registers in the output")
|
||||
|
||||
args = parser.parse_args()
|
||||
jobs = args.jobs
|
||||
|
||||
mips_isa.mips_fpr_names = fpr_name_options.get(args.reg_names, mips_isa.mips_fpr_names)
|
||||
|
||||
|
||||
ASM_OUT = "asm/"
|
||||
DATA_OUT = "data/"
|
||||
|
||||
|
|
@ -122,71 +132,71 @@ def discard_decomped_files(files_spec, include_files):
|
|||
|
||||
|
||||
MIPS_BRANCH_LIKELY_INSNS = [
|
||||
MIPS_INS_BEQL,
|
||||
MIPS_INS_BGEZALL,
|
||||
MIPS_INS_BGEZL,
|
||||
MIPS_INS_BGTZL,
|
||||
MIPS_INS_BLEZL,
|
||||
MIPS_INS_BLTZALL,
|
||||
MIPS_INS_BLTZL,
|
||||
MIPS_INS_BNEL,
|
||||
MIPS_INS_BC1TL,
|
||||
MIPS_INS_BC1FL,
|
||||
mips_isa.MIPS_INS_BEQL,
|
||||
mips_isa.MIPS_INS_BGEZALL,
|
||||
mips_isa.MIPS_INS_BGEZL,
|
||||
mips_isa.MIPS_INS_BGTZL,
|
||||
mips_isa.MIPS_INS_BLEZL,
|
||||
mips_isa.MIPS_INS_BLTZALL,
|
||||
mips_isa.MIPS_INS_BLTZL,
|
||||
mips_isa.MIPS_INS_BNEL,
|
||||
mips_isa.MIPS_INS_BC1TL,
|
||||
mips_isa.MIPS_INS_BC1FL,
|
||||
]
|
||||
|
||||
MIPS_BRANCH_INSNS = [
|
||||
*MIPS_BRANCH_LIKELY_INSNS,
|
||||
MIPS_INS_BEQ,
|
||||
MIPS_INS_BGEZ,
|
||||
MIPS_INS_BGEZAL,
|
||||
MIPS_INS_BGTZ,
|
||||
MIPS_INS_BNE,
|
||||
MIPS_INS_BLTZ,
|
||||
MIPS_INS_BLTZAL,
|
||||
MIPS_INS_BLEZ,
|
||||
MIPS_INS_BC1T,
|
||||
MIPS_INS_BC1F,
|
||||
MIPS_INS_BEQZ,
|
||||
MIPS_INS_BNEZ,
|
||||
MIPS_INS_B,
|
||||
mips_isa.MIPS_INS_BEQ,
|
||||
mips_isa.MIPS_INS_BGEZ,
|
||||
mips_isa.MIPS_INS_BGEZAL,
|
||||
mips_isa.MIPS_INS_BGTZ,
|
||||
mips_isa.MIPS_INS_BNE,
|
||||
mips_isa.MIPS_INS_BLTZ,
|
||||
mips_isa.MIPS_INS_BLTZAL,
|
||||
mips_isa.MIPS_INS_BLEZ,
|
||||
mips_isa.MIPS_INS_BC1T,
|
||||
mips_isa.MIPS_INS_BC1F,
|
||||
mips_isa.MIPS_INS_BEQZ,
|
||||
mips_isa.MIPS_INS_BNEZ,
|
||||
mips_isa.MIPS_INS_B,
|
||||
]
|
||||
|
||||
MIPS_JUMP_INSNS = [MIPS_INS_JAL, MIPS_INS_JALR, MIPS_INS_J, MIPS_INS_JR]
|
||||
MIPS_JUMP_INSNS = [mips_isa.MIPS_INS_JAL, mips_isa.MIPS_INS_JALR, mips_isa.MIPS_INS_J, mips_isa.MIPS_INS_JR]
|
||||
|
||||
MIPS_FP_LOAD_INSNS = [MIPS_INS_LWC1, MIPS_INS_LDC1]
|
||||
MIPS_FP_LOAD_INSNS = [mips_isa.MIPS_INS_LWC1, mips_isa.MIPS_INS_LDC1]
|
||||
|
||||
MIPS_FP_STORE_INSNS = [MIPS_INS_SWC1, MIPS_INS_SDC1]
|
||||
MIPS_FP_STORE_INSNS = [mips_isa.MIPS_INS_SWC1, mips_isa.MIPS_INS_SDC1]
|
||||
|
||||
MIPS_FP_LOAD_STORE_INSNS = [*MIPS_FP_LOAD_INSNS, *MIPS_FP_STORE_INSNS]
|
||||
|
||||
MIPS_STORE_INSNS = [
|
||||
MIPS_INS_SB,
|
||||
MIPS_INS_SH,
|
||||
MIPS_INS_SW,
|
||||
MIPS_INS_SWL,
|
||||
MIPS_INS_SWR,
|
||||
MIPS_INS_SD,
|
||||
MIPS_INS_SDL,
|
||||
MIPS_INS_SDR,
|
||||
MIPS_INS_SC,
|
||||
MIPS_INS_SCD,
|
||||
mips_isa.MIPS_INS_SB,
|
||||
mips_isa.MIPS_INS_SH,
|
||||
mips_isa.MIPS_INS_SW,
|
||||
mips_isa.MIPS_INS_SWL,
|
||||
mips_isa.MIPS_INS_SWR,
|
||||
mips_isa.MIPS_INS_SD,
|
||||
mips_isa.MIPS_INS_SDL,
|
||||
mips_isa.MIPS_INS_SDR,
|
||||
mips_isa.MIPS_INS_SC,
|
||||
mips_isa.MIPS_INS_SCD,
|
||||
*MIPS_FP_STORE_INSNS,
|
||||
]
|
||||
|
||||
MIPS_LOAD_STORE_INSNS = [
|
||||
MIPS_INS_LB,
|
||||
MIPS_INS_LBU,
|
||||
MIPS_INS_LH,
|
||||
MIPS_INS_LHU,
|
||||
MIPS_INS_LW,
|
||||
MIPS_INS_LWL,
|
||||
MIPS_INS_LWR,
|
||||
MIPS_INS_LWU,
|
||||
MIPS_INS_LD,
|
||||
MIPS_INS_LDL,
|
||||
MIPS_INS_LDR,
|
||||
MIPS_INS_LL,
|
||||
MIPS_INS_LLD,
|
||||
mips_isa.MIPS_INS_LB,
|
||||
mips_isa.MIPS_INS_LBU,
|
||||
mips_isa.MIPS_INS_LH,
|
||||
mips_isa.MIPS_INS_LHU,
|
||||
mips_isa.MIPS_INS_LW,
|
||||
mips_isa.MIPS_INS_LWL,
|
||||
mips_isa.MIPS_INS_LWR,
|
||||
mips_isa.MIPS_INS_LWU,
|
||||
mips_isa.MIPS_INS_LD,
|
||||
mips_isa.MIPS_INS_LDL,
|
||||
mips_isa.MIPS_INS_LDR,
|
||||
mips_isa.MIPS_INS_LL,
|
||||
mips_isa.MIPS_INS_LLD,
|
||||
*MIPS_STORE_INSNS,
|
||||
*MIPS_FP_LOAD_STORE_INSNS,
|
||||
]
|
||||
|
|
@ -615,7 +625,7 @@ def find_symbols_in_text(section, rodata_section, data_regions):
|
|||
|
||||
# assert insn.value_forname(insn.fields[0]) is not None
|
||||
# print(f"insn: {insn.mnemonic}, rt: {insn.rt}, first: {insn.value_forname(insn.fields[0])}")
|
||||
# assert insn.id not in [MIPS_INS_ORI, MIPS_INS_ADDIU, *MIPS_LOAD_STORE_INSNS] or insn.rt == insn.value_forname(insn.fields[0])
|
||||
# assert insn.id not in [MIPS_INS_ORI, mips_isa.MIPS_INS_ADDIU, *MIPS_LOAD_STORE_INSNS] or insn.rt == insn.value_forname(insn.fields[0])
|
||||
|
||||
if (
|
||||
delay_slot
|
||||
|
|
@ -636,9 +646,9 @@ def find_symbols_in_text(section, rodata_section, data_regions):
|
|||
|
||||
insns = []
|
||||
for i, raw_insn in enumerate(raw_insns, 0):
|
||||
insn = decode_insn(raw_insn, vram + i * 4)
|
||||
insn = mips_isa.decode_insn(raw_insn, vram + i * 4)
|
||||
|
||||
if insn.id == MIPS_INS_JR and insn.rs != MIPS_REG_RA:
|
||||
if insn.id == mips_isa.MIPS_INS_JR and insn.rs != mips_isa.MIPS_REG_RA:
|
||||
# It's hard to find when two jump tables next to each other end, so do a naive first pass
|
||||
# to try and find as many jump tables as possible.
|
||||
# Luckily IDO has a very homogeneous output for jump tables for which it is very unlikely
|
||||
|
|
@ -651,9 +661,9 @@ def find_symbols_in_text(section, rodata_section, data_regions):
|
|||
insn_m2 = insns[-2]
|
||||
insn_m3 = insns[-3]
|
||||
if (
|
||||
insn_m1.id == MIPS_INS_LW
|
||||
and insn_m2.id == MIPS_INS_ADDU
|
||||
and insn_m3.id == MIPS_INS_LUI
|
||||
insn_m1.id == mips_isa.MIPS_INS_LW
|
||||
and insn_m2.id == mips_isa.MIPS_INS_ADDU
|
||||
and insn_m3.id == mips_isa.MIPS_INS_LUI
|
||||
):
|
||||
prospective_jtbls.add((insn_m3.imm << 0x10) + insn_m1.imm)
|
||||
insns.append(insn)
|
||||
|
|
@ -674,7 +684,7 @@ def find_symbols_in_text(section, rodata_section, data_regions):
|
|||
Relocated jump targets give us functions in this section
|
||||
"""
|
||||
assert (
|
||||
insn.id == MIPS_INS_JAL
|
||||
insn.id == mips_isa.MIPS_INS_JAL
|
||||
), f"R_MIPS_26 applied to {insn.mnemonic} when it should be JAL"
|
||||
put_symbol(symbols_dict, "functions", insn.target)
|
||||
elif reloc[1] == 5: # R_MIPS_HI16
|
||||
|
|
@ -682,7 +692,7 @@ def find_symbols_in_text(section, rodata_section, data_regions):
|
|||
Relocated %hi gives us %hi values to match with associated %lo
|
||||
"""
|
||||
assert (
|
||||
insn.id == MIPS_INS_LUI
|
||||
insn.id == mips_isa.MIPS_INS_LUI
|
||||
), f"R_MIPS_HI16 applied to {insn.mnemonic} when it should be LUI"
|
||||
prev_hi = insn.imm
|
||||
hi_vram = vram + reloc[2]
|
||||
|
|
@ -691,7 +701,7 @@ def find_symbols_in_text(section, rodata_section, data_regions):
|
|||
Relocated %lo + a %hi to match with gives us relocated symbols in data sections
|
||||
"""
|
||||
assert (
|
||||
insn.id == MIPS_INS_ADDIU or insn.id in MIPS_LOAD_STORE_INSNS
|
||||
insn.id == mips_isa.MIPS_INS_ADDIU or insn.id in MIPS_LOAD_STORE_INSNS
|
||||
), f"R_MIPS_HI16 applied to {insn.mnemonic} when it should be ADDIU or a load/store"
|
||||
symbol_value = (prev_hi << 0x10) + insn.imm
|
||||
put_symbols(symbols_dict, "symbols", {hi_vram: symbol_value})
|
||||
|
|
@ -757,16 +767,16 @@ def find_symbols_in_text(section, rodata_section, data_regions):
|
|||
if insn.id in MIPS_BRANCH_INSNS:
|
||||
func_branch_labels.add(insn.offset)
|
||||
delayed_insn = insn
|
||||
elif insn.id == MIPS_INS_ERET:
|
||||
elif insn.id == mips_isa.MIPS_INS_ERET:
|
||||
put_symbol(symbols_dict, "functions", vaddr + 4)
|
||||
elif insn.id in MIPS_JUMP_INSNS:
|
||||
if insn.id == MIPS_INS_JAL:
|
||||
if insn.id == mips_isa.MIPS_INS_JAL:
|
||||
# mark function at target
|
||||
put_symbol(symbols_dict, "functions", insn.target)
|
||||
elif insn.id == MIPS_INS_J:
|
||||
elif insn.id == mips_isa.MIPS_INS_J:
|
||||
# mark label at target
|
||||
func_branch_labels.add(insn.target)
|
||||
elif insn.id == MIPS_INS_JR:
|
||||
elif insn.id == mips_isa.MIPS_INS_JR:
|
||||
# check if anything branches past it in either branch or jtbl labels
|
||||
if vaddr >= max(func_branch_labels, default=0) and vaddr >= max(
|
||||
func_jtbl_labels, default=0
|
||||
|
|
@ -792,7 +802,7 @@ def find_symbols_in_text(section, rodata_section, data_regions):
|
|||
symbols_dict, "functions", vaddr + 8 + n_padding * 4
|
||||
)
|
||||
delayed_insn = insn
|
||||
elif insn.id == MIPS_INS_LUI:
|
||||
elif insn.id == mips_isa.MIPS_INS_LUI:
|
||||
"""
|
||||
Process LUI instruction
|
||||
|
||||
|
|
@ -809,34 +819,34 @@ def find_symbols_in_text(section, rodata_section, data_regions):
|
|||
save_tracker(delayed_insn.offset, {insn.rt: (vaddr, insn.imm)})
|
||||
else:
|
||||
lui_tracker.update({insn.rt: (vaddr, insn.imm)})
|
||||
elif insn.id == MIPS_INS_ADDIU or insn.id in MIPS_LOAD_STORE_INSNS:
|
||||
elif insn.id == mips_isa.MIPS_INS_ADDIU or insn.id in MIPS_LOAD_STORE_INSNS:
|
||||
# try match with tracked lui and mark symbol
|
||||
hi_vram, imm_value = lui_tracker.get(
|
||||
insn.rs if insn.id == MIPS_INS_ADDIU else insn.base, (None, None)
|
||||
insn.rs if insn.id == mips_isa.MIPS_INS_ADDIU else insn.base, (None, None)
|
||||
)
|
||||
# if a match was found, validate and record the symbol, TODO improve validation
|
||||
if hi_vram != None and (
|
||||
(((imm_value >> 0x8) & 0xF) != 0 and imm_value < 0x1000)
|
||||
or (imm_value >= 0x8000 and imm_value < 0x80D0)
|
||||
or (imm_value >= 0xA400 and imm_value < 0xA480)
|
||||
or (imm_value < 0x0400 and insn.id == MIPS_INS_ADDIU)
|
||||
or (imm_value < 0x0400 and insn.id == mips_isa.MIPS_INS_ADDIU)
|
||||
):
|
||||
lo_vram = vaddr
|
||||
symbol_value = (imm_value << 0x10) + insn.imm
|
||||
put_symbols(symbols_dict, "symbols", {hi_vram: symbol_value})
|
||||
put_symbols(symbols_dict, "symbols", {lo_vram: symbol_value})
|
||||
if insn.id == MIPS_INS_LW:
|
||||
if insn.id == mips_isa.MIPS_INS_LW:
|
||||
# try find jr within the same block
|
||||
cur_idx = i
|
||||
lookahead_insn = insns[cur_idx] # TODO fix vaddr here
|
||||
# still in same block unless one of these instructions are found
|
||||
while lookahead_insn.id not in [
|
||||
*MIPS_BRANCH_INSNS,
|
||||
MIPS_INS_JAL,
|
||||
MIPS_INS_JALR,
|
||||
MIPS_INS_J,
|
||||
mips_isa.MIPS_INS_JAL,
|
||||
mips_isa.MIPS_INS_JALR,
|
||||
mips_isa.MIPS_INS_J,
|
||||
]:
|
||||
if lookahead_insn.id == MIPS_INS_JR:
|
||||
if lookahead_insn.id == mips_isa.MIPS_INS_JR:
|
||||
if lookahead_insn.rs == (
|
||||
insn.ft
|
||||
if insn.id in MIPS_FP_LOAD_STORE_INSNS
|
||||
|
|
@ -884,24 +894,24 @@ def find_symbols_in_text(section, rodata_section, data_regions):
|
|||
if cur_idx >= len(raw_insns):
|
||||
break
|
||||
lookahead_insn = insns[cur_idx] # TODO fix vaddr here
|
||||
elif insn.id == MIPS_INS_LD: # doubleword loads
|
||||
elif insn.id == mips_isa.MIPS_INS_LD: # doubleword loads
|
||||
put_symbol(symbols_dict, "dwords", symbol_value)
|
||||
elif insn.id in [MIPS_INS_LWC1, MIPS_INS_SWC1]: # float load/stores
|
||||
elif insn.id in [mips_isa.MIPS_INS_LWC1, mips_isa.MIPS_INS_SWC1]: # float load/stores
|
||||
# add float
|
||||
put_symbol(symbols_dict, "floats", symbol_value)
|
||||
elif insn.id in [MIPS_INS_LDC1, MIPS_INS_SDC1]: # double load/stores
|
||||
elif insn.id in [mips_isa.MIPS_INS_LDC1, mips_isa.MIPS_INS_SDC1]: # double load/stores
|
||||
# add double
|
||||
put_symbol(symbols_dict, "doubles", symbol_value)
|
||||
elif (
|
||||
insn.id == MIPS_INS_ADDIU and vaddr % 4 == 0
|
||||
insn.id == mips_isa.MIPS_INS_ADDIU and vaddr % 4 == 0
|
||||
): # strings seem to only ever be 4-byte aligned
|
||||
# add possible string
|
||||
put_symbol(symbols_dict, "prospective_strings", symbol_value)
|
||||
# clear lui tracking state if register is clobbered by the addiu/load instruction itself
|
||||
# insn.rt == (insn.rs if insn.id == MIPS_INS_ADDIU else insn.base) and
|
||||
# insn.rt == (insn.rs if insn.id == mips_isa.MIPS_INS_ADDIU else insn.base) and
|
||||
if insn.id not in MIPS_STORE_INSNS and insn.id not in MIPS_FP_LOAD_INSNS:
|
||||
clobber_conditionally(insn)
|
||||
elif insn.id == MIPS_INS_ORI:
|
||||
elif insn.id == mips_isa.MIPS_INS_ORI:
|
||||
# try match with tracked lui and mark constant
|
||||
hi_vram, imm_value = lui_tracker.get(insn.rs, (None, None))
|
||||
if hi_vram != None: # found match
|
||||
|
|
@ -916,7 +926,7 @@ def find_symbols_in_text(section, rodata_section, data_regions):
|
|||
else:
|
||||
# clear lui tracking if register is clobbered by something unrelated
|
||||
if (
|
||||
insn.id == MIPS_INS_ADDU
|
||||
insn.id == mips_isa.MIPS_INS_ADDU
|
||||
and insn.rs in lui_tracker.keys()
|
||||
and (insn.rd == insn.rs)
|
||||
):
|
||||
|
|
@ -929,35 +939,35 @@ def find_symbols_in_text(section, rodata_section, data_regions):
|
|||
pass
|
||||
# insns listed either write to fprs/cop0 or don't write to any
|
||||
elif insn.id not in [
|
||||
MIPS_INS_MTC0,
|
||||
MIPS_INS_MTC1,
|
||||
MIPS_INS_DMTC1,
|
||||
MIPS_INS_MULT,
|
||||
MIPS_INS_MULTU,
|
||||
MIPS_INS_DMULT,
|
||||
MIPS_INS_DMULTU,
|
||||
MIPS_INS_DIV,
|
||||
MIPS_INS_DIVU,
|
||||
MIPS_INS_DDIV,
|
||||
MIPS_INS_DDIVU,
|
||||
MIPS_INS_MTHI,
|
||||
MIPS_INS_MTLO,
|
||||
MIPS_INS_CTC1,
|
||||
MIPS_INS_NOP,
|
||||
MIPS_INS_BREAK,
|
||||
MIPS_INS_TLBP,
|
||||
MIPS_INS_TLBR,
|
||||
MIPS_INS_TLBWI,
|
||||
MIPS_INS_MOV_S,
|
||||
MIPS_INS_MOV_D,
|
||||
MIPS_INS_C_LT_S,
|
||||
MIPS_INS_C_LT_D,
|
||||
MIPS_INS_DIV_S,
|
||||
MIPS_INS_MUL_S,
|
||||
MIPS_INS_TRUNC_W_S,
|
||||
MIPS_INS_CVT_S_W,
|
||||
MIPS_INS_SUB_S,
|
||||
MIPS_INS_ADD_S,
|
||||
mips_isa.MIPS_INS_MTC0,
|
||||
mips_isa.MIPS_INS_MTC1,
|
||||
mips_isa.MIPS_INS_DMTC1,
|
||||
mips_isa.MIPS_INS_MULT,
|
||||
mips_isa.MIPS_INS_MULTU,
|
||||
mips_isa.MIPS_INS_DMULT,
|
||||
mips_isa.MIPS_INS_DMULTU,
|
||||
mips_isa.MIPS_INS_DIV,
|
||||
mips_isa.MIPS_INS_DIVU,
|
||||
mips_isa.MIPS_INS_DDIV,
|
||||
mips_isa.MIPS_INS_DDIVU,
|
||||
mips_isa.MIPS_INS_MTHI,
|
||||
mips_isa.MIPS_INS_MTLO,
|
||||
mips_isa.MIPS_INS_CTC1,
|
||||
mips_isa.MIPS_INS_NOP,
|
||||
mips_isa.MIPS_INS_BREAK,
|
||||
mips_isa.MIPS_INS_TLBP,
|
||||
mips_isa.MIPS_INS_TLBR,
|
||||
mips_isa.MIPS_INS_TLBWI,
|
||||
mips_isa.MIPS_INS_MOV_S,
|
||||
mips_isa.MIPS_INS_MOV_D,
|
||||
mips_isa.MIPS_INS_C_LT_S,
|
||||
mips_isa.MIPS_INS_C_LT_D,
|
||||
mips_isa.MIPS_INS_DIV_S,
|
||||
mips_isa.MIPS_INS_MUL_S,
|
||||
mips_isa.MIPS_INS_TRUNC_W_S,
|
||||
mips_isa.MIPS_INS_CVT_S_W,
|
||||
mips_isa.MIPS_INS_SUB_S,
|
||||
mips_isa.MIPS_INS_ADD_S,
|
||||
]:
|
||||
clobber_conditionally(insn)
|
||||
|
||||
|
|
@ -1000,18 +1010,18 @@ def find_symbols_in_text(section, rodata_section, data_regions):
|
|||
part += ", "
|
||||
op_str.append(part)
|
||||
|
||||
elif insn.id == MIPS_INS_LUI:
|
||||
elif insn.id == mips_isa.MIPS_INS_LUI:
|
||||
op_str = [op_str]
|
||||
instr["rt"] = insn.rt
|
||||
|
||||
elif insn.id == MIPS_INS_ADDIU or insn.id in MIPS_LOAD_STORE_INSNS:
|
||||
elif insn.id == mips_isa.MIPS_INS_ADDIU or insn.id in MIPS_LOAD_STORE_INSNS:
|
||||
op_str = [op_str]
|
||||
instr["rt"] = insn.rt
|
||||
instr["rs"] = insn.rs
|
||||
instr["ft"] = insn.ft
|
||||
instr["base"] = insn.base
|
||||
|
||||
elif insn.id == MIPS_INS_ORI:
|
||||
elif insn.id == mips_isa.MIPS_INS_ORI:
|
||||
op_str = [op_str]
|
||||
instr["rt"] = insn.rt
|
||||
instr["rs"] = insn.rs
|
||||
|
|
@ -1031,13 +1041,13 @@ def find_symbols_in_text(section, rodata_section, data_regions):
|
|||
|
||||
if delay_slot and delayed_insn is not None:
|
||||
if (
|
||||
delayed_insn.id == MIPS_INS_JAL
|
||||
or delayed_insn.id == MIPS_INS_JALR
|
||||
or (delayed_insn.id == MIPS_INS_JR and delayed_insn.rs == MIPS_REG_RA)
|
||||
delayed_insn.id == mips_isa.MIPS_INS_JAL
|
||||
or delayed_insn.id == mips_isa.MIPS_INS_JALR
|
||||
or (delayed_insn.id == mips_isa.MIPS_INS_JR and delayed_insn.rs == mips_isa.MIPS_REG_RA)
|
||||
):
|
||||
# destroy lui tracking state
|
||||
lui_tracker.clear()
|
||||
elif delayed_insn.id == MIPS_INS_JR and vaddr == next_jtbl_jr + 4:
|
||||
elif delayed_insn.id == mips_isa.MIPS_INS_JR and vaddr == next_jtbl_jr + 4:
|
||||
# save lui tracking state for each jtbl label
|
||||
for label in individual_jtbl_labels[next_jtbl_jr]:
|
||||
save_tracker(label, lui_tracker.copy())
|
||||
|
|
@ -1046,7 +1056,7 @@ def find_symbols_in_text(section, rodata_section, data_regions):
|
|||
# save lui tracking state
|
||||
save_tracker(delayed_insn.offset, lui_tracker.copy())
|
||||
# destroy current lui tracking state for unconditional branches
|
||||
if delayed_insn.id == MIPS_INS_B:
|
||||
if delayed_insn.id == mips_isa.MIPS_INS_B:
|
||||
lui_tracker.clear()
|
||||
|
||||
delayed_insn = None
|
||||
|
|
@ -1265,35 +1275,35 @@ def fixup_text_symbols(data, vram, data_regions, info):
|
|||
else:
|
||||
line += op_part
|
||||
|
||||
elif insn["id"] == MIPS_INS_LUI:
|
||||
elif insn["id"] == mips_isa.MIPS_INS_LUI:
|
||||
symbol_value = symbols.get(vaddr, None)
|
||||
if symbol_value is not None:
|
||||
line += (
|
||||
f"{mips_gpr_names[insn['rt']]}, %hi({proper_name(symbol_value)})"
|
||||
f"{ mips_isa.mips_gpr_names[insn['rt']]}, %hi({proper_name(symbol_value)})"
|
||||
)
|
||||
else:
|
||||
constant_value = constants.get(vaddr, None)
|
||||
if constant_value is not None:
|
||||
line += (
|
||||
f"{mips_gpr_names[insn['rt']]}, (0x{constant_value:08X} >> 16)"
|
||||
f"{ mips_isa.mips_gpr_names[insn['rt']]}, (0x{constant_value:08X} >> 16)"
|
||||
)
|
||||
else:
|
||||
line += entry["op"][0]
|
||||
|
||||
elif insn["id"] == MIPS_INS_ADDIU or insn["id"] in MIPS_LOAD_STORE_INSNS:
|
||||
elif insn["id"] == mips_isa.MIPS_INS_ADDIU or insn["id"] in MIPS_LOAD_STORE_INSNS:
|
||||
symbol_value = symbols.get(vaddr, None)
|
||||
if symbol_value is not None:
|
||||
if insn["id"] == MIPS_INS_ADDIU:
|
||||
line += f"{mips_gpr_names[insn['rt']]}, {mips_gpr_names[insn['rs']]}, %lo({proper_name(symbol_value)})"
|
||||
if insn["id"] == mips_isa.MIPS_INS_ADDIU:
|
||||
line += f"{ mips_isa.mips_gpr_names[insn['rt']]}, { mips_isa.mips_gpr_names[insn['rs']]}, %lo({proper_name(symbol_value)})"
|
||||
else:
|
||||
line += f"{mips_fpr_names[insn['ft']] if insn['id'] in MIPS_FP_LOAD_STORE_INSNS else mips_gpr_names[insn['rt']]}, %lo({proper_name(symbol_value)})({mips_gpr_names[insn['base']]})"
|
||||
line += f"{mips_isa.mips_fpr_names[insn['ft']] if insn['id'] in MIPS_FP_LOAD_STORE_INSNS else mips_isa.mips_gpr_names[insn['rt']]}, %lo({proper_name(symbol_value)})({ mips_isa.mips_gpr_names[insn['base']]})"
|
||||
else:
|
||||
line += entry["op"][0]
|
||||
|
||||
elif insn["id"] == MIPS_INS_ORI:
|
||||
elif insn["id"] == mips_isa.MIPS_INS_ORI:
|
||||
constant_value = constants.get(vaddr, None)
|
||||
if constant_value is not None:
|
||||
line += f"{mips_gpr_names[insn['rt']]}, {mips_gpr_names[insn['rs']]}, (0x{constant_value:08X} & 0xFFFF)"
|
||||
line += f"{ mips_isa.mips_gpr_names[insn['rt']]}, { mips_isa.mips_gpr_names[insn['rs']]}, (0x{constant_value:08X} & 0xFFFF)"
|
||||
else:
|
||||
line += entry["op"][0]
|
||||
|
||||
|
|
@ -1329,7 +1339,7 @@ def disassemble_text(data, vram, data_regions, info):
|
|||
for i, raw_insn in enumerate(raw_insns, 0):
|
||||
i *= 4
|
||||
vaddr = vram + i
|
||||
insn = decode_insn(raw_insns[i // 4], vaddr)
|
||||
insn = mips_isa.decode_insn(raw_insns[i // 4], vaddr)
|
||||
mnemonic = insn.mnemonic
|
||||
op_str = insn.op_str
|
||||
|
||||
|
|
@ -1389,27 +1399,27 @@ def disassemble_text(data, vram, data_regions, info):
|
|||
op_str_parts.append(insn.format_field(field))
|
||||
op_str = ", ".join(op_str_parts)
|
||||
delayed_insn = insn
|
||||
elif insn.id == MIPS_INS_LUI:
|
||||
elif insn.id == mips_isa.MIPS_INS_LUI:
|
||||
symbol_value = symbols.get(vaddr, None)
|
||||
if symbol_value is not None:
|
||||
op_str = f"{mips_gpr_names[insn.rt]}, %hi({proper_name(symbol_value)})"
|
||||
op_str = f"{ mips_isa.mips_gpr_names[insn.rt]}, %hi({proper_name(symbol_value)})"
|
||||
else:
|
||||
constant_value = constants.get(vaddr, None)
|
||||
if constant_value is not None:
|
||||
op_str = (
|
||||
f"{mips_gpr_names[insn.rt]}, (0x{constant_value:08X} >> 16)"
|
||||
f"{ mips_isa.mips_gpr_names[insn.rt]}, (0x{constant_value:08X} >> 16)"
|
||||
)
|
||||
elif insn.id == MIPS_INS_ADDIU or insn.id in MIPS_LOAD_STORE_INSNS:
|
||||
elif insn.id == mips_isa.MIPS_INS_ADDIU or insn.id in MIPS_LOAD_STORE_INSNS:
|
||||
symbol_value = symbols.get(vaddr, None)
|
||||
if symbol_value is not None:
|
||||
if insn.id == MIPS_INS_ADDIU:
|
||||
op_str = f"{mips_gpr_names[insn.rt]}, {mips_gpr_names[insn.rs]}, %lo({proper_name(symbol_value)})"
|
||||
if insn.id == mips_isa.MIPS_INS_ADDIU:
|
||||
op_str = f"{ mips_isa.mips_gpr_names[insn.rt]}, { mips_isa.mips_gpr_names[insn.rs]}, %lo({proper_name(symbol_value)})"
|
||||
else:
|
||||
op_str = f"{mips_fpr_names[insn.ft] if insn.id in MIPS_FP_LOAD_STORE_INSNS else mips_gpr_names[insn.rt]}, %lo({proper_name(symbol_value)})({mips_gpr_names[insn.base]})"
|
||||
elif insn.id == MIPS_INS_ORI:
|
||||
op_str = f"{mips_isa.mips_fpr_names[insn.ft] if insn.id in MIPS_FP_LOAD_STORE_INSNS else mips_isa.mips_gpr_names[insn.rt]}, %lo({proper_name(symbol_value)})({ mips_isa.mips_gpr_names[insn.base]})"
|
||||
elif insn.id == mips_isa.MIPS_INS_ORI:
|
||||
constant_value = constants.get(vaddr, None)
|
||||
if constant_value is not None:
|
||||
op_str = f"{mips_gpr_names[insn.rt]}, {mips_gpr_names[insn.rs]}, (0x{constant_value:08X} & 0xFFFF)"
|
||||
op_str = f"{ mips_isa.mips_gpr_names[insn.rt]}, { mips_isa.mips_gpr_names[insn.rs]}, (0x{constant_value:08X} & 0xFFFF)"
|
||||
|
||||
if delay_slot:
|
||||
mnemonic = " " + mnemonic
|
||||
|
|
@ -2205,7 +2215,7 @@ for segment in files_spec:
|
|||
new[offset] = name
|
||||
full_file_list[segment[0]] = new
|
||||
|
||||
if args.full:
|
||||
if args.all:
|
||||
new_spec = []
|
||||
for segment in files_spec:
|
||||
if args.files and not any(
|
||||
|
|
@ -2357,7 +2367,7 @@ for segment in files_spec:
|
|||
|
||||
del files_spec[:]
|
||||
|
||||
pool = get_context("fork").Pool(jobs)
|
||||
pool = multiprocessing.get_context("fork").Pool(jobs)
|
||||
# Find symbols for each segment
|
||||
for section in all_sections:
|
||||
if section[-1]["name"] == "makerom":
|
||||
|
|
@ -2393,7 +2403,7 @@ for section in all_sections:
|
|||
pool.close()
|
||||
pool.join()
|
||||
|
||||
pool = get_context("fork").Pool(jobs)
|
||||
pool = multiprocessing.get_context("fork").Pool(jobs)
|
||||
for section in all_sections:
|
||||
if section[-1]["type"] == "makerom":
|
||||
continue
|
||||
|
|
@ -2418,7 +2428,7 @@ for section in all_sections:
|
|||
vrom_addrs = {addr for _, addr in vrom_variables}
|
||||
|
||||
# Textual disassembly for each segment
|
||||
with get_context("fork").Pool(jobs) as p:
|
||||
with multiprocessing.get_context("fork").Pool(jobs) as p:
|
||||
p.map(
|
||||
disassemble_segment,
|
||||
[
|
||||
|
|
|
|||
|
|
@ -286,15 +286,26 @@ mips_cop0_names = [
|
|||
"TagLo" , "TagHi" , "ErrorEPC" , "Reserved31",
|
||||
]
|
||||
|
||||
mips_fpr_names = [
|
||||
numeric_fpr_names = [
|
||||
"$f0", "$f1", "$f2", "$f3",
|
||||
"$f4", "$f5", "$f6", "$f7", "$f8", "$f9", "$f10", "$f11",
|
||||
"$f12", "$f13", "$f14", "$f15",
|
||||
"$f16", "$f17", "$f18", "$f19",
|
||||
"$f20", "$f21", "$f22", "$f23", "$f24", "$f25", "$f26", "$f27", "$f28", "$f29", "$f30",
|
||||
"$31",
|
||||
"$31", # Floating-point control/status register
|
||||
]
|
||||
|
||||
o32_fpr_names = [
|
||||
"$fv0", "$fv0f", "$fv1", "$fv1f",
|
||||
"$ft0", "$ft0f", "$ft1", "$ft1f", "$ft2", "$ft2f", "$ft3", "$ft3f",
|
||||
"$fa0", "$fa0f", "$fa1", "$fa1f",
|
||||
"$ft4", "$ft4f", "$ft5", "$ft5f",
|
||||
"$fs0", "$fs0f", "$fs1", "$fs1f", "$fs2", "$fs2f", "$fs3", "$fs3f", "$fs4", "$fs4f", "$fs5",
|
||||
"$31", # Floating-point control/status register
|
||||
]
|
||||
|
||||
mips_fpr_names = numeric_fpr_names
|
||||
|
||||
# Instruction field fetching
|
||||
|
||||
def sign_extend_16(value):
|
||||
|
|
|
|||
Loading…
Reference in New Issue