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
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
|
import dataclasses
import typing
from flags import variable_used_unspecialized
from formatting import (
Formatter,
UNUSED,
maybe_parenthesize,
parenthesize_cond,
)
from instructions import (
ActiveCacheEffect,
Instruction,
MacroInstruction,
Component,
Tiers,
TIER_ONE,
)
from parsing import StackEffect, CacheEffect, Family
@dataclasses.dataclass
class StackOffset:
"""Represent the stack offset for a PEEK or POKE.
- At stack_pointer[0], deep and high are both empty.
(Note that that is an invalid stack reference.)
- Below stack top, only deep is non-empty.
- Above stack top, only high is non-empty.
- In complex cases, both deep and high may be non-empty.
All this would be much simpler if all stack entries were the same
size, but with conditional and array effects, they aren't.
The offsets are each represented by a list of StackEffect objects.
The name in the StackEffects is unused.
"""
deep: list[StackEffect] = dataclasses.field(default_factory=list)
high: list[StackEffect] = dataclasses.field(default_factory=list)
def clone(self) -> "StackOffset":
return StackOffset(list(self.deep), list(self.high))
def negate(self) -> "StackOffset":
return StackOffset(list(self.high), list(self.deep))
def deeper(self, eff: StackEffect) -> None:
if eff in self.high:
self.high.remove(eff)
else:
self.deep.append(eff)
def higher(self, eff: StackEffect) -> None:
if eff in self.deep:
self.deep.remove(eff)
else:
self.high.append(eff)
def as_terms(self) -> list[tuple[str, str]]:
num = 0
terms: list[tuple[str, str]] = []
for eff in self.deep:
if eff.size:
terms.append(("-", maybe_parenthesize(eff.size)))
elif eff.cond and eff.cond not in ("0", "1"):
terms.append(("-", f"({parenthesize_cond(eff.cond)} ? 1 : 0)"))
elif eff.cond != "0":
num -= 1
for eff in self.high:
if eff.size:
terms.append(("+", maybe_parenthesize(eff.size)))
elif eff.cond and eff.cond not in ("0", "1"):
terms.append(("+", f"({parenthesize_cond(eff.cond)} ? 1 : 0)"))
elif eff.cond != "0":
num += 1
if num < 0:
terms.insert(0, ("-", str(-num)))
elif num > 0:
terms.append(("+", str(num)))
return terms
def as_index(self) -> str:
terms = self.as_terms()
return make_index(terms)
def equivalent_to(self, other: "StackOffset") -> bool:
if self.deep == other.deep and self.high == other.high:
return True
deep = list(self.deep)
for x in other.deep:
try:
deep.remove(x)
except ValueError:
return False
if deep:
return False
high = list(self.high)
for x in other.high:
try:
high.remove(x)
except ValueError:
return False
if high:
return False
return True
def make_index(terms: list[tuple[str, str]]) -> str:
# Produce an index expression from the terms honoring PEP 8,
# surrounding binary ops with spaces but not unary minus
index = ""
for sign, term in terms:
if index:
index += f" {sign} {term}"
elif sign == "+":
index = term
else:
index = sign + term
return index or "0"
@dataclasses.dataclass
class StackItem:
offset: StackOffset
effect: StackEffect
def as_variable(self, lax: bool = False) -> str:
"""Return e.g. stack_pointer[-1]."""
terms = self.offset.as_terms()
if self.effect.size:
terms.insert(0, ("+", "stack_pointer"))
index = make_index(terms)
if self.effect.size:
res = index
else:
res = f"stack_pointer[{index}]"
if not lax:
# Check that we're not reading or writing above stack top.
# Skip this for output variable initialization (lax=True).
assert (
self.effect in self.offset.deep and not self.offset.high
), f"Push or pop above current stack level: {res}"
return res
def as_stack_effect(self, lax: bool = False) -> StackEffect:
return StackEffect(
self.as_variable(lax=lax),
self.effect.type if self.effect.size else "",
self.effect.cond,
self.effect.size,
)
@dataclasses.dataclass
class CopyItem:
src: StackItem
dst: StackItem
class EffectManager:
"""Manage stack effects and offsets for an instruction."""
instr: Instruction
active_caches: list[ActiveCacheEffect]
peeks: list[StackItem]
pokes: list[StackItem]
copies: list[CopyItem] # See merge()
# Track offsets from stack pointer
min_offset: StackOffset
final_offset: StackOffset
# Link to previous manager
pred: "EffectManager | None" = None
def __init__(
self,
instr: Instruction,
active_caches: list[ActiveCacheEffect],
pred: "EffectManager | None" = None,
):
self.instr = instr
self.active_caches = active_caches
self.peeks = []
self.pokes = []
self.copies = []
self.final_offset = pred.final_offset.clone() if pred else StackOffset()
for eff in reversed(instr.input_effects):
self.final_offset.deeper(eff)
self.peeks.append(StackItem(offset=self.final_offset.clone(), effect=eff))
self.min_offset = self.final_offset.clone()
for eff in instr.output_effects:
self.pokes.append(StackItem(offset=self.final_offset.clone(), effect=eff))
self.final_offset.higher(eff)
self.pred = pred
while pred:
# Replace push(x) + pop(y) with copy(x, y).
# Check that the sources and destinations are disjoint.
sources: set[str] = set()
destinations: set[str] = set()
while (
pred.pokes
and self.peeks
and pred.pokes[-1].effect == self.peeks[0].effect
):
src = pred.pokes.pop(-1)
dst = self.peeks.pop(0)
assert src.offset.equivalent_to(dst.offset), (src, dst)
pred.final_offset.deeper(src.effect)
if dst.effect.name != src.effect.name:
if dst.effect.name != UNUSED:
destinations.add(dst.effect.name)
if src.effect.name != UNUSED:
sources.add(src.effect.name)
self.copies.append(CopyItem(src, dst))
# TODO: Turn this into an error (pass an Analyzer instance?)
assert sources & destinations == set(), (
pred.instr.name,
self.instr.name,
sources,
destinations,
)
# See if we can get more copies of a earlier predecessor.
if self.peeks and not pred.pokes and not pred.peeks:
pred = pred.pred
else:
pred = None # Break
# Fix up patterns of copies through UNUSED,
# e.g. cp(a, UNUSED) + cp(UNUSED, b) -> cp(a, b).
if any(copy.src.effect.name == UNUSED for copy in self.copies):
pred = self.pred
while pred is not None:
for copy in self.copies:
if copy.src.effect.name == UNUSED:
for pred_copy in pred.copies:
if pred_copy.dst == copy.src:
copy.src = pred_copy.src
break
pred = pred.pred
def adjust_deeper(self, eff: StackEffect) -> None:
for peek in self.peeks:
peek.offset.deeper(eff)
for poke in self.pokes:
poke.offset.deeper(eff)
for copy in self.copies:
copy.src.offset.deeper(eff)
copy.dst.offset.deeper(eff)
self.min_offset.deeper(eff)
self.final_offset.deeper(eff)
def adjust_higher(self, eff: StackEffect) -> None:
for peek in self.peeks:
peek.offset.higher(eff)
for poke in self.pokes:
poke.offset.higher(eff)
for copy in self.copies:
copy.src.offset.higher(eff)
copy.dst.offset.higher(eff)
self.min_offset.higher(eff)
self.final_offset.higher(eff)
def adjust(self, offset: StackOffset) -> None:
deep = list(offset.deep)
high = list(offset.high)
for down in deep:
self.adjust_deeper(down)
for up in high:
self.adjust_higher(up)
def adjust_inverse(self, offset: StackOffset) -> None:
deep = list(offset.deep)
high = list(offset.high)
for down in deep:
self.adjust_higher(down)
for up in high:
self.adjust_deeper(up)
def collect_vars(self) -> dict[str, StackEffect]:
"""Collect all variables, skipping unused ones."""
vars: dict[str, StackEffect] = {}
def add(eff: StackEffect) -> None:
if eff.name != UNUSED:
if eff.name in vars:
# TODO: Make this an error
assert vars[eff.name] == eff, (
self.instr.name,
eff.name,
vars[eff.name],
eff,
)
else:
vars[eff.name] = eff
for copy in self.copies:
add(copy.src.effect)
add(copy.dst.effect)
for peek in self.peeks:
add(peek.effect)
for poke in self.pokes:
add(poke.effect)
return vars
def less_than(a: StackOffset, b: StackOffset) -> bool:
# TODO: Handle more cases
if a.high != b.high:
return False
return a.deep[: len(b.deep)] == b.deep
def get_managers(parts: list[Component]) -> list[EffectManager]:
managers: list[EffectManager] = []
pred: EffectManager | None = None
for part in parts:
mgr = EffectManager(part.instr, part.active_caches, pred)
managers.append(mgr)
pred = mgr
return managers
def get_stack_effect_info_for_macro(mac: MacroInstruction) -> tuple[str, str]:
"""Get the stack effect info for a macro instruction.
Returns a tuple (popped, pushed) where each is a string giving a
symbolic expression for the number of values popped/pushed.
"""
parts = [part for part in mac.parts if isinstance(part, Component)]
managers = get_managers(parts)
popped = StackOffset()
for mgr in managers:
if less_than(mgr.min_offset, popped):
popped = mgr.min_offset.clone()
# Compute pushed = final - popped
pushed = managers[-1].final_offset.clone()
for effect in popped.deep:
pushed.higher(effect)
for effect in popped.high:
pushed.deeper(effect)
return popped.negate().as_index(), pushed.as_index()
def write_single_instr(
instr: Instruction, out: Formatter, tier: Tiers = TIER_ONE
) -> None:
try:
write_components(
[Component(instr, instr.active_caches)],
out,
tier,
0,
instr.family,
)
except AssertionError as err:
raise AssertionError(f"Error writing instruction {instr.name}") from err
def write_macro_instr(mac: MacroInstruction, out: Formatter) -> None:
parts = [
part
for part in mac.parts
if isinstance(part, Component) and part.instr.name != "_SET_IP"
]
out.emit("")
with out.block(f"TARGET({mac.name})"):
needs_this = any(part.instr.needs_this_instr for part in parts)
if needs_this and not mac.predicted:
out.emit(f"_Py_CODEUNIT *this_instr = frame->instr_ptr = next_instr;")
else:
out.emit(f"frame->instr_ptr = next_instr;")
out.emit(f"next_instr += {mac.cache_offset+1};")
out.emit(f"INSTRUCTION_STATS({mac.name});")
if mac.predicted:
out.emit(f"PREDICTED({mac.name});")
if needs_this:
out.emit(f"_Py_CODEUNIT *this_instr = next_instr - {mac.cache_offset+1};")
out.static_assert_family_size(mac.name, mac.family, mac.cache_offset)
try:
next_instr_is_set = write_components(
parts, out, TIER_ONE, mac.cache_offset, mac.family
)
except AssertionError as err:
raise AssertionError(f"Error writing macro {mac.name}") from err
if not parts[-1].instr.always_exits:
if parts[-1].instr.check_eval_breaker:
out.emit("CHECK_EVAL_BREAKER();")
out.emit("DISPATCH();")
def write_components(
parts: list[Component],
out: Formatter,
tier: Tiers,
cache_offset: int,
family: Family | None,
) -> bool:
managers = get_managers(parts)
all_vars: dict[str, StackEffect] = {}
for mgr in managers:
for name, eff in mgr.collect_vars().items():
if name in all_vars:
# TODO: Turn this into an error -- variable conflict
assert all_vars[name] == eff, (
name,
mgr.instr.name,
all_vars[name],
eff,
)
else:
all_vars[name] = eff
# Declare all variables
for name, eff in all_vars.items():
out.declare(eff, None)
next_instr_is_set = False
for mgr in managers:
if len(parts) > 1:
out.emit(f"// {mgr.instr.name}")
for copy in mgr.copies:
copy_src_effect = copy.src.effect
if copy_src_effect.name != copy.dst.effect.name:
if copy_src_effect.name == UNUSED:
copy_src_effect = copy.src.as_stack_effect()
out.assign(copy.dst.effect, copy_src_effect)
for peek in mgr.peeks:
out.assign(
peek.effect,
peek.as_stack_effect(),
)
# Initialize array outputs
for poke in mgr.pokes:
if poke.effect.size and poke.effect.name not in mgr.instr.unmoved_names:
out.assign(
poke.effect,
poke.as_stack_effect(lax=True),
)
if mgr.instr.name in ("_PUSH_FRAME", "_POP_FRAME"):
# Adjust stack to min_offset.
# This means that all input effects of this instruction
# are materialized, but not its output effects.
# That's as intended, since these two are so special.
out.stack_adjust(mgr.min_offset.deep, mgr.min_offset.high)
# However, for tier 2, pretend the stack is at final offset.
mgr.adjust_inverse(mgr.final_offset)
if tier == TIER_ONE:
# TODO: Check in analyzer that _{PUSH,POP}_FRAME is last.
assert (
mgr is managers[-1]
), f"Expected {mgr.instr.name!r} to be the last uop"
assert_no_pokes(managers)
if mgr.instr.name == "_SAVE_RETURN_OFFSET":
next_instr_is_set = True
if tier == TIER_ONE:
assert_no_pokes(managers)
if len(parts) == 1:
mgr.instr.write_body(out, 0, mgr.active_caches, tier, family)
else:
with out.block(""):
mgr.instr.write_body(out, -4, mgr.active_caches, tier, family)
if mgr is managers[-1] and not next_instr_is_set and not mgr.instr.always_exits:
# Adjust the stack to its final depth, *then* write the
# pokes for all preceding uops.
# Note that for array output effects we may still write
# past the stack top.
out.stack_adjust(mgr.final_offset.deep, mgr.final_offset.high)
write_all_pokes(mgr.final_offset, managers, out)
return next_instr_is_set
def assert_no_pokes(managers: list[EffectManager]) -> None:
for mgr in managers:
for poke in mgr.pokes:
if not poke.effect.size and poke.effect.name not in mgr.instr.unmoved_names:
assert (
poke.effect.name == UNUSED
), f"Unexpected poke of {poke.effect.name} in {mgr.instr.name!r}"
def write_all_pokes(
offset: StackOffset, managers: list[EffectManager], out: Formatter
) -> None:
# Emit all remaining pushes (pokes)
for m in managers:
m.adjust_inverse(offset)
write_pokes(m, out)
def write_pokes(mgr: EffectManager, out: Formatter) -> None:
for poke in mgr.pokes:
if not poke.effect.size and poke.effect.name not in mgr.instr.unmoved_names:
out.assign(
poke.as_stack_effect(),
poke.effect,
)
def write_single_instr_for_abstract_interp(instr: Instruction, out: Formatter) -> None:
try:
_write_components_for_abstract_interp(
[Component(instr, instr.active_caches)],
out,
)
except AssertionError as err:
raise AssertionError(
f"Error writing abstract instruction {instr.name}"
) from err
def _write_components_for_abstract_interp(
parts: list[Component],
out: Formatter,
) -> None:
managers = get_managers(parts)
for mgr in managers:
if mgr is managers[-1]:
out.stack_adjust(mgr.final_offset.deep, mgr.final_offset.high)
mgr.adjust_inverse(mgr.final_offset)
# NULL out the output stack effects
for poke in mgr.pokes:
if not poke.effect.size and poke.effect.name not in mgr.instr.unmoved_names:
out.emit(
f"PARTITIONNODE_OVERWRITE((_Py_PARTITIONNODE_t *)"
f"PARTITIONNODE_NULLROOT, PEEK(-({poke.offset.as_index()})), true);"
)
|