Merge pull request #1549 from elicn/refactor-qdb

Refactor qdb

Author: xwings

pyproject.toml

@@ -28,7 +28,7 @@ keywords = [
 
 [tool.poetry.dependencies]
 python = "^3.8"
-capstone = "^4"
+capstone = "^5"
 unicorn = "2.1.3"
 pefile = ">=2022.5.30"
 python-registry = "^1.3.1"

qiling/debugger/__init__.py

@@ -1,3 +1 @@
 from .debugger import QlDebugger
-# from .disassember import QlDisassember
-# from .utils import QlReadELF

qiling/debugger/disassember.py

@@ -3,7 +3,6 @@
 # Cross Platform and Multi Architecture Advanced Binary Emulation Framework
 #
 
-from .arch_x86 import ArchX86
-from .arch_mips import ArchMIPS
 from .arch_arm import ArchARM, ArchCORTEX_M
-from .arch_x8664 import ArchX8664
+from .arch_intel import ArchIntel, ArchX86, ArchX64
+from .arch_mips import ArchMIPS

qiling/debugger/qdb/arch/__init__.py

@@ -3,32 +3,81 @@
 # Cross Platform and Multi Architecture Advanced Binary Emulation Framework
 #
 
+from typing import Collection, Dict, Mapping, Optional, TypeVar
 
-from qiling.const import QL_ARCH
-from unicorn import UC_ERR_READ_UNMAPPED
-import unicorn
+T = TypeVar('T')
 
 
 class Arch:
+    """Arch base class.
     """
-    base class for arch
-    """
 
-    def __init__(self):
-        pass
+    def __init__(self, regs: Collection[str], swaps: Mapping[str, str], asize: int, isize: int) -> None:
+        """Initialize architecture instance.
+
+        Args:
+            regs  : collection of registers names to include in context
+            asize : native address size in bytes
+            isize : instruction size in bytes
+            swaps : readable register names alternatives, may be empty
+        """
+
+        self._regs = regs
+        self._swaps = swaps
+        self._asize = asize
+        self._isize = isize
 
     @property
-    def arch_insn_size(self):
-        return 4
+    def regs(self) -> Collection[str]:
+        """Collection of registers names.
+        """
+
+        return self._regs
 
     @property
-    def archbit(self):
-        return 4
+    def isize(self) -> int:
+        """Native instruction size.
+        """
+
+        return self._isize
+
+    @property
+    def asize(self) -> int:
+        """Native pointer size.
+        """
+
+        return self._asize
+
+    def swap_regs(self, mapping: Mapping[str, T]) -> Dict[str, T]:
+        """Swap default register names with their aliases.
+
+        Args:
+            mapping: regsiters names mapped to their values
+
+        Returns: a new dictionary where all swappable names were swapped with their aliases
+        """
+
+        return {self._swaps.get(k, k): v for k, v in mapping.items()}
+
+    def unalias(self, name: str) -> str:
+        """Get original register name for the specified alias.
+
+        Args:
+            name: aliaes register name
+
+        Returns: original name of aliased register, or same name if not an alias
+        """
+
+        # perform a reversed lookup in swaps to find the original name for given alias
+        return next((org for org, alt in self._swaps.items() if name == alt), name)
+
+    def read_insn(self, address: int) -> Optional[bytearray]:
+        """Read a single instruction from given address.
+
+        Args:
+            address: memory address to read from
 
-    def read_insn(self, address: int):
-        try:
-            result = self.read_mem(address, self.arch_insn_size)
-        except unicorn.unicorn.UcError as err:
-            result = None
+        Returns: instruction bytes, or None if memory could not be read
+        """
 
-        return result
+        return self.try_read_mem(address, self.isize)

qiling/debugger/qdb/arch/arch.py

@@ -3,105 +3,141 @@
 # Cross Platform and Multi Architecture Advanced Binary Emulation Framework
 #
 
-from typing import Mapping
+from typing import Dict, Optional
 
 from .arch import Arch
 
+
 class ArchARM(Arch):
-    def __init__(self):
-        super().__init__()
-        self._regs = (
-                "r0", "r1", "r2", "r3",
-                "r4", "r5", "r6", "r7",
-                "r8", "r9", "r10", "r11",
-                "r12", "sp", "lr", "pc",
-                )
+    def __init__(self) -> None:
+        regs = (
+            'r0', 'r1', 'r2', 'r3',
+            'r4', 'r5', 'r6', 'r7',
+            'r8', 'r9', 'r10', 'r11',
+            'r12', 'sp', 'lr', 'pc'
+        )
+
+        aliases = {
+            'r9' : 'sb',
+            'r10': 'sl',
+            'r12': 'ip',
+            'r11': 'fp'
+        }
+
+        asize = 4
+        isize = 4
+
+        super().__init__(regs, aliases, asize, isize)
+
+    @staticmethod
+    def get_flags(bits: int) -> Dict[str, bool]:
+        return {
+            'thumb':    bits & (0b1 <<  5) != 0,
+            'fiq':      bits & (0b1 <<  6) != 0,
+            'irq':      bits & (0b1 <<  7) != 0,
+            'overflow': bits & (0b1 << 28) != 0,
+            'carry':    bits & (0b1 << 29) != 0,
+            'zero':     bits & (0b1 << 30) != 0,
+            'neg':      bits & (0b1 << 31) != 0
+        }
+
+    @staticmethod
+    def get_mode(bits: int) -> str:
+        modes = {
+            0b10000: 'User',
+            0b10001: 'FIQ',
+            0b10010: 'IRQ',
+            0b10011: 'Supervisor',
+            0b10110: 'Monitor',
+            0b10111: 'Abort',
+            0b11010: 'Hypervisor',
+            0b11011: 'Undefined',
+            0b11111: 'System'
+        }
+
+        return modes.get(bits & 0b11111, '?')
 
     @property
-    def regs(self):
-        return self._regs
+    def is_thumb(self) -> bool:
+        """Query whether the processor is currently in thumb mode.
+        """
 
-    @regs.setter
-    def regs(self, regs):
-        self._regs += regs
+        return self.ql.arch.is_thumb
 
     @property
-    def regs_need_swapped(self):
-        return {
-                "sl": "r10",
-                "ip": "r12",
-                "fp": "r11",
-                }
+    def isize(self) -> int:
+        return 2 if self.is_thumb else self._isize
 
     @staticmethod
-    def get_flags(bits: int) -> Mapping[str, bool]:
-        """
-        get flags for ARM
+    def __is_wide_insn(data: bytes) -> bool:
+        """Determine whether a sequence of bytes respresents a wide thumb instruction.
         """
 
-        def get_mode(bits: int) -> int:
-            """
-            get operating mode for ARM
-            """
-            return {
-                    0b10000: "User",
-                    0b10001: "FIQ",
-                    0b10010: "IRQ",
-                    0b10011: "Supervisor",
-                    0b10110: "Monitor",
-                    0b10111: "Abort",
-                    0b11010: "Hypervisor",
-                    0b11011: "Undefined",
-                    0b11111: "System",
-                    }.get(bits & 0x00001f)
+        assert len(data) in (2, 4), f'unexpected instruction length: {len(data)}'
 
-        return {
-                "mode":     get_mode(bits),
-                "thumb":    bits & 0x00000020 != 0,
-                "fiq":      bits & 0x00000040 != 0,
-                "irq":      bits & 0x00000080 != 0,
-                "neg":      bits & 0x80000000 != 0,
-                "zero":     bits & 0x40000000 != 0,
-                "carry":    bits & 0x20000000 != 0,
-                "overflow": bits & 0x10000000 != 0,
-                }
+        # determine whether this is a wide instruction by inspecting the 5 most
+        # significant bits in the first half-word
+        return (data[1] >> 3) & 0b11111 in (0b11101, 0b11110, 0b11111)
 
-    @property
-    def thumb_mode(self) -> bool:
-        """
-        helper function for checking thumb mode
+    def __read_thumb_insn_fail(self, address: int) -> Optional[bytearray]:
+        """A failsafe method for reading thumb instructions. This method is needed for
+        rare cases in which a narrow instruction is on a page boundary where the next
+        page is unavailable.
         """
 
-        return self.ql.arch.is_thumb
+        lo_half = self.try_read_mem(address, 2)
 
+        if lo_half is None:
+            return None
 
-    def read_insn(self, address: int) -> bytes:
-        """
-        read instruction depending on current operating mode
+        data = lo_half
+
+        if ArchARM.__is_wide_insn(data):
+            hi_half = self.try_read_mem(address + 2, 2)
+
+            # fail if higher half-word was required but could not be read
+            if hi_half is None:
+                return None
+
+            data.extend(hi_half)
+
+        return data
+
+    def __read_thumb_insn(self, address: int) -> Optional[bytearray]:
+        """Read one instruction in thumb mode.
+
+        Thumb instructions may be either 2 or 4 bytes long, depending on encoding of
+        the first word. However, reading two chunks of two bytes each is slower. For
+        most cases reading all four bytes in advance will be safe and quicker.
         """
 
-        def thumb_read(address: int) -> bytes:
+        data = self.try_read_mem(address, 4)
 
-            first_two = self.ql.mem.read_ptr(address, 2)
-            result = self.ql.pack16(first_two)
+        if data is None:
+            # there is a slight chance we could not read 4 bytes because only 2
+            # are available. try the failsafe method to find out
+            return self.__read_thumb_insn_fail(address)
 
-            # to judge it's thumb mode or not
-            if any([
-                first_two & 0xf000 == 0xf000,
-                first_two & 0xf800 == 0xf800,
-                first_two & 0xe800 == 0xe800,
-                 ]):
+        if ArchARM.__is_wide_insn(data):
+            return data
 
-                latter_two = self.ql.mem.read_ptr(address+2, 2)
-                result += self.ql.pack16(latter_two)
+        return data[:2]
 
-            return result
+    def read_insn(self, address: int) -> Optional[bytearray]:
+        """Read one instruction worth of bytes.
+        """
 
-        return super().read_insn(address) if not self.thumb_mode else thumb_read(address)
+        if self.is_thumb:
+            return self.__read_thumb_insn(address)
 
+        return super().read_insn(address)
 
 
 class ArchCORTEX_M(ArchARM):
     def __init__(self):
         super().__init__()
-        self.regs += ("xpsr", "control", "primask", "basepri", "faultmask")
+
+        self._regs += (
+            'xpsr', 'control', 'primask',
+            'basepri', 'faultmask'
+        )