-
Notifications
You must be signed in to change notification settings - Fork 0
/
analysis.py
192 lines (164 loc) · 6.16 KB
/
analysis.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
# Data flow analysis and stuff.
import math
import typing
from copy import deepcopy
from dataclasses import dataclass
from pythia.analysis_domain import iteration_strategy, InstructionLattice
import pythia.graph_utils as gu
import pythia.type_system as ts
from pythia import disassemble, ast_transform
from pythia import tac
from pythia.analysis_domain import InvariantMap
from pythia.analysis_liveness import LivenessVarLattice
from pythia import analysis_typed_pointer as typed_pointer
from pythia.graph_utils import Location
type Cfg = gu.Cfg[tac.Tac]
TYPE_INV_NAME = typed_pointer.TypedPointerLattice.name()
LIVENESS_INV_NAME = LivenessVarLattice.name()
@dataclass
class InvariantTriple[Inv]:
pre: InvariantMap[Inv]
intermediate: dict[Location, Inv]
post: InvariantMap[Inv]
def analyze[
Inv
](
_cfg: Cfg, analysis: InstructionLattice[Inv], keep_intermediate=False
) -> InvariantTriple[Inv]:
pre_result: InvariantMap[Inv] = {}
post_result: InvariantMap[Inv] = {}
intermediate_result: dict[Location, Inv] = {}
cfg = iteration_strategy(_cfg, analysis.backward)
wl = [entry] = {cfg.entry_label}
pre_result[entry] = analysis.initial()
while wl:
label = min(wl)
wl.remove(label)
block = cfg[label]
invariant = deepcopy(pre_result[label])
for i, (index, ins) in enumerate(block.items()):
location = (label, index)
if keep_intermediate:
intermediate_result[location] = deepcopy(invariant)
try:
invariant = analysis.transfer(invariant, ins, location)
except Exception as e:
e.add_note(f"At {location}")
e.add_note(f"from {ins}")
e.add_note(f"original command: {_cfg.annotator(location, ins)}")
e.add_note(f"pre: {invariant}")
raise e
post_result[label] = invariant
for next_label in cfg.successors(label):
next_pre_invariants = [
post_result.get(n, analysis.bottom())
for n in cfg.predecessors(next_label)
]
next_pre = pre_result.get(next_label, analysis.bottom())
new_next_pre = analysis.join_all(invariant, *next_pre_invariants)
if not analysis.is_less_than(new_next_pre, next_pre):
pre_result[next_label] = new_next_pre
wl.add(next_label)
pre_result, post_result = cfg.order((pre_result, post_result))
return InvariantTriple(pre_result, intermediate_result, post_result)
def print_analysis(
cfg: Cfg,
invariants: dict[str, InvariantTriple],
loop_end: typing.Optional[Location],
dirty_locals: set[str],
print_invariants: bool = True,
) -> None:
for label, block in sorted(cfg.items()):
if math.isinf(label):
continue
if print_invariants:
print("Pre:")
for name, invariant_pair in invariants.items():
pre_invariant = invariant_pair.pre[label]
print(f"\t{name}: ", end="")
print(str(pre_invariant))
gu.print_block(label, block, cfg.annotator)
if print_invariants:
print("Post:")
for name, invariant_pair in invariants.items():
post_invariant = invariant_pair.post[label]
print(f"\t{name}:", end="")
print(str(post_invariant))
print()
if loop_end is not None and label == loop_end:
print(f"Dirty Locals:", ", ".join(dirty_locals))
print()
print("Successors:", list(cfg.successors(label)))
print()
def run(
cfg: Cfg,
for_location: typing.Optional[Location],
module_type: ts.Module,
function_name: str,
) -> dict[str, InvariantTriple]:
liveness_invariants = analyze(cfg, LivenessVarLattice(), keep_intermediate=True)
typed_pointer_analysis = typed_pointer.TypedPointerLattice(
liveness_invariants.intermediate, function_name, module_type, for_location
)
typed_pointer_invariants = analyze(
cfg, typed_pointer_analysis, keep_intermediate=False
)
invariant_pairs: dict[str, InvariantTriple] = {
LIVENESS_INV_NAME: liveness_invariants,
TYPE_INV_NAME: typed_pointer_invariants,
}
return invariant_pairs
def find_dirty_roots(
invariants: dict[str, InvariantTriple], loop_end: typing.Optional[gu.Label]
) -> set[str]:
if loop_end is None:
return set()
return set(
typed_pointer.find_dirty_roots(
invariants[TYPE_INV_NAME].post[loop_end],
invariants[LIVENESS_INV_NAME].post[loop_end],
)
)
def analyze_function(
filename: str,
*function_names: str,
print_invariants: bool,
outfile: str,
simplify: bool,
) -> None:
print(filename, function_names, print_invariants, outfile, simplify)
functions, imports = disassemble.read_file(
filename, filter_for_loops=not bool(function_names)
)
module_type = ts.parse_file(filename)
if not function_names:
if not functions:
raise ValueError("No functions with for loops found")
function_names = tuple(functions.keys())
dirty_map: dict[str, set[str]] = {}
for function_name in function_names:
print(function_name)
f = functions[function_name]
cfg = tac.make_tac_cfg(f)
cfg = gu.simplify_cfg(cfg)
if not simplify:
cfg = gu.refine_to_chain(cfg)
# gu.pretty_print_cfg(cfg)
try:
for_location, loop_end = gu.find_first_for_loop(
cfg, lambda b: isinstance(b, tac.For)
)
except ValueError:
for_location, loop_end = None, None
invariant_pairs = run(
cfg, for_location, module_type=module_type, function_name=function_name
)
dirty_map[function_name] = find_dirty_roots(invariant_pairs, loop_end)
if print_invariants:
print_analysis(cfg, invariant_pairs, loop_end, dirty_map[function_name])
output = ast_transform.transform(filename, dirty_map)
if outfile is None:
print(output)
else:
with open(outfile, "w", encoding="utf-8") as f:
print(output, file=f)