123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186 |
- import dataclasses
- import dis
- import sys
- from numbers import Real
- TERMINAL_OPCODES = {
- dis.opmap["RETURN_VALUE"],
- dis.opmap["JUMP_FORWARD"],
- dis.opmap["RAISE_VARARGS"],
- # TODO(jansel): double check exception handling
- }
- if sys.version_info >= (3, 9):
- TERMINAL_OPCODES.add(dis.opmap["RERAISE"])
- if sys.version_info >= (3, 11):
- TERMINAL_OPCODES.add(dis.opmap["JUMP_BACKWARD"])
- else:
- TERMINAL_OPCODES.add(dis.opmap["JUMP_ABSOLUTE"])
- JUMP_OPCODES = set(dis.hasjrel + dis.hasjabs)
- JUMP_OPNAMES = {dis.opname[opcode] for opcode in JUMP_OPCODES}
- HASLOCAL = set(dis.haslocal)
- HASFREE = set(dis.hasfree)
- stack_effect = dis.stack_effect
- def remove_dead_code(instructions):
- """Dead code elimination"""
- indexof = {id(inst): i for i, inst in enumerate(instructions)}
- live_code = set()
- def find_live_code(start):
- for i in range(start, len(instructions)):
- if i in live_code:
- return
- live_code.add(i)
- inst = instructions[i]
- if inst.opcode in JUMP_OPCODES:
- find_live_code(indexof[id(inst.target)])
- if inst.opcode in TERMINAL_OPCODES:
- return
- find_live_code(0)
- return [inst for i, inst in enumerate(instructions) if i in live_code]
- def remove_pointless_jumps(instructions):
- """Eliminate jumps to the next instruction"""
- pointless_jumps = {
- id(a)
- for a, b in zip(instructions, instructions[1:])
- if a.opname == "JUMP_ABSOLUTE" and a.target is b
- }
- return [inst for inst in instructions if id(inst) not in pointless_jumps]
- def propagate_line_nums(instructions):
- """Ensure every instruction has line number set in case some are removed"""
- cur_line_no = None
- def populate_line_num(inst):
- nonlocal cur_line_no
- if inst.starts_line:
- cur_line_no = inst.starts_line
- inst.starts_line = cur_line_no
- for inst in instructions:
- populate_line_num(inst)
- def remove_extra_line_nums(instructions):
- """Remove extra starts line properties before packing bytecode"""
- cur_line_no = None
- def remove_line_num(inst):
- nonlocal cur_line_no
- if inst.starts_line is None:
- return
- elif inst.starts_line == cur_line_no:
- inst.starts_line = None
- else:
- cur_line_no = inst.starts_line
- for inst in instructions:
- remove_line_num(inst)
- @dataclasses.dataclass
- class ReadsWrites:
- reads: set
- writes: set
- visited: set
- def livevars_analysis(instructions, instruction):
- indexof = {id(inst): i for i, inst in enumerate(instructions)}
- must = ReadsWrites(set(), set(), set())
- may = ReadsWrites(set(), set(), set())
- def walk(state, start):
- if start in state.visited:
- return
- state.visited.add(start)
- for i in range(start, len(instructions)):
- inst = instructions[i]
- if inst.opcode in HASLOCAL or inst.opcode in HASFREE:
- if "LOAD" in inst.opname or "DELETE" in inst.opname:
- if inst.argval not in must.writes:
- state.reads.add(inst.argval)
- elif "STORE" in inst.opname:
- state.writes.add(inst.argval)
- else:
- raise NotImplementedError(f"unhandled {inst.opname}")
- if inst.opcode in JUMP_OPCODES:
- walk(may, indexof[id(inst.target)])
- state = may
- if inst.opcode in TERMINAL_OPCODES:
- return
- walk(must, indexof[id(instruction)])
- return must.reads | may.reads
- @dataclasses.dataclass
- class FixedPointBox:
- value: bool = True
- @dataclasses.dataclass
- class StackSize:
- low: Real
- high: Real
- fixed_point: FixedPointBox
- def zero(self):
- self.low = 0
- self.high = 0
- self.fixed_point.value = False
- def offset_of(self, other, n):
- prior = (self.low, self.high)
- self.low = min(self.low, other.low + n)
- self.high = max(self.high, other.high + n)
- if (self.low, self.high) != prior:
- self.fixed_point.value = False
- def stacksize_analysis(instructions):
- assert instructions
- fixed_point = FixedPointBox()
- stack_sizes = {
- inst: StackSize(float("inf"), float("-inf"), fixed_point)
- for inst in instructions
- }
- stack_sizes[instructions[0]].zero()
- for _ in range(100):
- if fixed_point.value:
- break
- fixed_point.value = True
- for inst, next_inst in zip(instructions, instructions[1:] + [None]):
- stack_size = stack_sizes[inst]
- if inst.opcode not in TERMINAL_OPCODES:
- assert next_inst is not None, f"missing next inst: {inst}"
- stack_sizes[next_inst].offset_of(
- stack_size, stack_effect(inst.opcode, inst.arg, jump=False)
- )
- if inst.opcode in JUMP_OPCODES:
- stack_sizes[inst.target].offset_of(
- stack_size, stack_effect(inst.opcode, inst.arg, jump=True)
- )
- if False:
- for inst in instructions:
- stack_size = stack_sizes[inst]
- print(stack_size.low, stack_size.high, inst)
- low = min([x.low for x in stack_sizes.values()])
- high = max([x.high for x in stack_sizes.values()])
- assert fixed_point.value, "failed to reach fixed point"
- assert low >= 0
- return high
|