jdc_core/convert.rs
1//! Region tree → Stmt tree conversion.
2//!
3//! Responsibilities:
4//! * weave in per-block statements and terminals (return/throw),
5//! * resolve `Goto{target}` nodes against the enclosing loop/switch scope
6//! stack into `break` / `continue` / labeled goto,
7//! * classify loops into while / do-while / infinite forms,
8//! * assemble try/catch with exception parameter variables.
9
10use std::collections::{HashMap, HashSet};
11
12use crate::cfg::Cfg;
13use crate::ir::build::{BlockResult, Term};
14use crate::ir::expr::{BinOp, ConstVal, Expr, UnOp};
15use crate::ir::stmt::{Catch, Stmt};
16use crate::structure::Region;
17
18/// A named Java statement for `break`/`continue` targets.
19#[derive(Debug)]
20pub enum Jump {
21 Break(Option<String>),
22 Continue(Option<String>),
23 /// Continue of the loop at the given depth-from-innermost, prefixed by
24 /// an emission of the target block's own statements (the jump re-runs
25 /// a shared statement-bearing tail that flows back into the header;
26 /// a bare Continue would skip those statements).
27 ContinueVia(usize, Option<String>),
28 /// Unstructured: emit `label:` + goto as comments (best effort).
29 RawGoto(usize),
30}
31
32#[derive(Debug)]
33struct LoopCtx {
34 header: usize,
35 exits: HashSet<usize>,
36 label: String,
37}
38
39#[derive(Debug)]
40struct SwitchCtx {
41 follow: Option<usize>,
42 label: String,
43 /// Loop-stack depth when the switch was entered: a `break` from a case
44 /// body only needs a label when the jump sits inside a loop that was
45 /// opened WITHIN the switch (an enclosing loop does not intercept it).
46 loops_depth: usize,
47}
48
49pub struct Converter<'a> {
50 pub cfg: &'a Cfg,
51 pub results: &'a Vec<BlockResult>,
52 pub groups: Vec<crate::structure::TryGroup>,
53 /// Dominators over the full method graph (for goto fallback decisions).
54 pub dom: crate::structure::DomInfo,
55 /// Block whose statements are currently being converted (for postdom
56 /// checks in goto resolution).
57 cur_block: usize,
58 loops: Vec<LoopCtx>,
59 switches: Vec<SwitchCtx>,
60 /// Labels actually referenced by break/continue statements.
61 used_labels: HashSet<String>,
62 /// Follow block of each enclosing If region; a goto to the innermost
63 /// follow is natural structured flow and disappears.
64 if_follows: Vec<usize>,
65 label_counter: usize,
66 /// True while converting the last element of a Seq (a trailing Goto
67 /// there may be inlined as a copy instead of an unemittable jump).
68 goto_is_last: bool,
69 /// The block whose emission IMMEDIATELY follows the current position:
70 /// the innermost enclosing If's follow while converting its arms, or
71 /// the next Seq sibling's head. A RawGoto may elide to fallthrough
72 /// ONLY when its target is this block — the old stack-wide
73 /// `if_follows.contains` check elided jumps to OUTER follows whose
74 /// emission was separated by intervening follow/copy emissions
75 /// (jdk17 DatagramChannelImpl.receive: the b15 fall arm's Goto{SET}
76 /// elided into cur=14's follow emission, skipping
77 /// `sender = sourceSocketAddress()` on the n==0&&isOpen path;
78 /// sun.net.www.protocol.http HttpURLConnection.getInputStream0: the
79 /// clone arms' Goto{addToCache} fell into the post-loop tail — the
80 /// documented addToCache-skip residual).
81 expect_next: std::cell::Cell<Option<usize>>,
82 /// Set while converting a Seq element that DIRECTLY follows a
83 /// CopyStmts of that same block: the back-edge-stub pattern
84 /// [CopyStmts{b}, Goto{b}] already emitted b's statements, so the
85 /// Goto must not inline them again (jdk11 HttpURLConnection
86 /// getInputStream0: the clone tail rendered the clone/removeFromCache
87 /// block twice — the second removeFromCache removed the just-cloned
88 /// entry from the auth cache; a header-adjacent b would also make
89 /// ContinueVia re-emit b's statements after the copy).
90 prev_copy: std::cell::Cell<Option<usize>>,
91 /// Heads of copy-walked shared tails (from the structurer).
92 copied_tails: std::collections::HashSet<usize>,
93 /// Collected label emissions: block target -> label name (for `Label` stmts).
94 pub pending_labels: HashMap<usize, String>,
95 /// Names of FINAL fields of the class being decompiled. A shared
96 /// terminator block whose statements write one of them must never be
97 /// inlined/copied at extra arrival sites: javac rejects a second
98 /// assignment to a final field (sun.security.util.Debug `hexDigits`).
99 final_fields: HashSet<String>,
100}
101
102/// Remove a trailing `Goto` whose target is the natural continuation.
103fn strip_trailing_goto(s: &mut Stmt, follow: Option<usize>) {
104 let Some(f) = follow else { return };
105 let items = match s {
106 Stmt::Block(v) => v,
107 Stmt::Goto(t) if *t as usize == f => {
108 *s = Stmt::Block(vec![]);
109 return;
110 }
111 _ => return,
112 };
113 while let Some(last) = items.last_mut() {
114 match last {
115 Stmt::Goto(t) if *t as usize == f => {
116 items.pop();
117 return;
118 }
119 Stmt::Block(inner) if !inner.is_empty() => {
120 // descend into trailing block
121 let n = inner.len();
122 if matches!(inner[n - 1], Stmt::Goto(t) if t as usize == f) {
123 inner.pop();
124 return;
125 }
126 break;
127 }
128 _ => break,
129 }
130 }
131}
132
133impl<'a> Converter<'a> {
134 pub fn with_copied_tails(mut self, tails: std::collections::HashSet<usize>) -> Self {
135 self.copied_tails = tails;
136 self
137 }
138
139 pub fn with_final_fields(mut self, finals: HashSet<String>) -> Self {
140 self.final_fields = finals;
141 self
142 }
143
144 /// True when any statement assigns a final field of this class.
145 fn stmts_write_final(&self, v: &[Stmt]) -> bool {
146 v.iter().any(|s| match s {
147 Stmt::ExprStmt(Expr::Assign { target, .. }) => {
148 matches!(&**target, Expr::Field { name, .. } if self.final_fields.contains(name))
149 }
150 _ => false,
151 })
152 }
153
154 /// Chain-aware twin of stmts_write_final for a tail HEAD: walks the
155 /// linear single-succ Fall/Goto chain (bounded) so a blank-final write
156 /// PAST the head still rejects the inline term-copy (URICertStore
157 /// clinit's two-field tail; mirrors structure's
158 /// terminator_writes_final chain probe).
159 fn tail_chain_writes_final(&self, t: usize) -> bool {
160 if self.final_fields.is_empty() {
161 return false;
162 }
163 let mut x = t;
164 let mut seen: std::collections::HashSet<usize> = std::collections::HashSet::new();
165 for _ in 0..8 {
166 if !seen.insert(x) {
167 return false;
168 }
169 if self.stmts_write_final(&self.results[x].stmts) {
170 return true;
171 }
172 if !matches!(self.results[x].term, Term::Fallthrough | Term::Goto) {
173 return false;
174 }
175 let succs = &self.cfg.blocks[x].succ;
176 if succs.len() != 1 {
177 return false;
178 }
179 x = succs[0];
180 }
181 false
182 }
183
184 pub fn new(cfg: &'a Cfg, results: &'a Vec<BlockResult>) -> Self {
185 let universe: HashSet<usize> = (0..cfg.blocks.len()).collect();
186 let dom = crate::structure::compute_dominators(cfg, &universe, cfg.entry);
187 let groups = crate::structure::group_exceptions_with(cfg, Some(results));
188 Self::from_parts(cfg, results, groups, dom)
189 }
190
191 /// `new` with the group set and dominator tree precomputed and shared
192 /// with the Structurer (see `Structurer::with_precomputed_groups`).
193 pub fn with_precomputed(
194 cfg: &'a Cfg,
195 results: &'a Vec<BlockResult>,
196 groups: Vec<crate::structure::TryGroup>,
197 dom: crate::structure::DomInfo,
198 ) -> Self {
199 Self::from_parts(cfg, results, groups, dom)
200 }
201
202 fn from_parts(
203 cfg: &'a Cfg,
204 results: &'a Vec<BlockResult>,
205 groups: Vec<crate::structure::TryGroup>,
206 dom: crate::structure::DomInfo,
207 ) -> Self {
208 Converter {
209 cfg,
210 results,
211 groups,
212 dom,
213 cur_block: usize::MAX,
214 loops: Vec::new(),
215 switches: Vec::new(),
216 if_follows: Vec::new(),
217 used_labels: HashSet::new(),
218 label_counter: 0,
219 goto_is_last: false,
220 expect_next: std::cell::Cell::new(None),
221 prev_copy: std::cell::Cell::new(None),
222 copied_tails: std::collections::HashSet::new(),
223 pending_labels: HashMap::new(),
224 final_fields: HashSet::new(),
225 }
226 }
227
228 fn next_label(&mut self) -> String {
229 self.label_counter += 1;
230 format!("L{}", self.label_counter)
231 }
232
233 fn groups_end(&self, gi: usize) -> Option<u32> {
234 self.groups.get(gi).map(|g| g.end)
235 }
236
237 pub fn convert(&mut self, r: Region) -> Stmt {
238 self.conv(r)
239 }
240
241 fn conv(&mut self, r: Region) -> Stmt {
242 match r {
243 Region::Empty => Stmt::Block(vec![]),
244 Region::Basic { block } => {
245 self.cur_block = block;
246 if crate::dbg_flag!("JCDC_DBG_BLOCKS") {
247 eprintln!("conv Basic {} ({} stmts)", block, self.results[block].stmts.len());
248 }
249 self.block_stmts(block)
250 }
251 Region::CopyStmts { block } => {
252 // Copies include the block's terminator when it is a
253 // return/throw (shared terminator blocks are inlined at
254 // every arrival site).
255 let mut v = self.results[block].stmts.clone();
256 match &self.results[block].term {
257 Term::Return(e) => v.push(Stmt::Return(e.clone())),
258 Term::Throw(e) => v.push(Stmt::Throw(e.clone())),
259 _ => {}
260 }
261 if v.len() == 1 {
262 v.into_iter().next().unwrap()
263 } else {
264 Stmt::Block(v)
265 }
266 }
267 Region::Seq(v) => {
268 let mut out = Vec::new();
269 let n = v.len();
270 let heads: Vec<usize> = v
271 .iter()
272 .map(crate::structure::region_head_block)
273 .collect();
274 let prev_copies: Vec<Option<usize>> = (0..n)
275 .map(|k| {
276 if k > 0 {
277 match &v[k - 1] {
278 Region::CopyStmts { block } => Some(*block),
279 _ => None,
280 }
281 } else {
282 None
283 }
284 })
285 .collect();
286 for (k, x) in v.into_iter().enumerate() {
287 let save = self.goto_is_last;
288 self.goto_is_last = k + 1 == n;
289 let save_pc = self.prev_copy.get();
290 self.prev_copy.set(prev_copies[k]);
291 let save_exp = self.expect_next.get();
292 // Only override when the next emission is KNOWN: a
293 // headless tail (Goto/Empty — a jump out) must keep
294 // the inherited expectation, or a nested arm's Goto
295 // would elide against a phantom (jdk17 dci: Basic17's
296 // MAX head erased the follow expectation and Goto{16}
297 // elided past the SET).
298 if k + 1 < n && heads[k + 1] != usize::MAX {
299 self.expect_next.set(Some(heads[k + 1]));
300 }
301 out.push(self.conv(x));
302 self.expect_next.set(save_exp);
303 self.prev_copy.set(save_pc);
304 self.goto_is_last = save;
305 }
306 Stmt::Block(out)
307 }
308 Region::If { block, cond, then_r, else_r, follow, ternary } => {
309 self.cur_block = block;
310 let mut head = self.block_stmts_no_term(block);
311 if let Some((_tv, _fv)) = ternary {
312 // Pure value diamond: the folded conditional flows into
313 // the merge block's rebuilt input stack; nothing to emit
314 // here (branch blocks are statement-free by definition).
315 let _ = cond;
316 } else {
317 if let Some(f) = follow {
318 self.if_follows.push(f);
319 }
320 let save_exp = self.expect_next.get();
321 // follow=None keeps the inherited expectation: the
322 // flow leaving this If's arms continues to whatever
323 // follows the If itself.
324 self.expect_next.set(follow.or(save_exp));
325 let then_stmt = self.conv(*then_r);
326 let else_stmt = self.conv(*else_r);
327 self.expect_next.set(save_exp);
328 if follow.is_some() {
329 self.if_follows.pop();
330 }
331 let if_stmt = Stmt::If {
332 cond,
333 then_stmt: Box::new(then_stmt),
334 else_stmt: if else_stmt.is_empty_block() { None } else { Some(Box::new(else_stmt)) },
335 };
336 head.push(if_stmt);
337 }
338 if head.len() == 1 {
339 head.pop().unwrap()
340 } else {
341 Stmt::Block(head)
342 }
343 }
344 Region::Loop { header, body, members, exits } => {
345 let label = self.next_label();
346 self.loops.push(LoopCtx {
347 header,
348 exits: exits.iter().copied().collect(),
349 label: label.clone(),
350 });
351 let body_stmt = self.conv(*body);
352 let ctx = self.loops.pop().unwrap();
353 let _ = members;
354 let mut st = self.classify_loop(header, body_stmt, ctx);
355 if crate::dbg_flag!("JCDC_DBG_LOOP") {
356 let kind = match &st {
357 Stmt::While { .. } => "While".to_string(),
358 Stmt::DoWhile { .. } => "DoWhile".to_string(),
359 Stmt::For { .. } => "For".to_string(),
360 Stmt::Block(b) => format!("Block{}[{}]", b.len(), b.iter().map(|x| match x {
361 Stmt::DoWhile { .. } => "DoWhile",
362 Stmt::While { .. } => "While",
363 Stmt::Break(_) => "Break",
364 Stmt::Return(_) => "Return",
365 _ => "_",
366 }).collect::<Vec<_>>().join(",")),
367 Stmt::Labeled { .. } => "Labeled".to_string(),
368 _ => "Other".to_string(),
369 };
370 eprintln!("CLASSIFY-OUT header={} -> {}", header, kind);
371 }
372 if self.used_labels.contains(&label) {
373 st = Stmt::Labeled { label, body: Box::new(st) };
374 }
375 st
376 }
377 Region::Switch { block, selector, cases, default, follow } => {
378 self.cur_block = block;
379 let sw_label = self.next_label();
380 self.switches.push(SwitchCtx {
381 follow,
382 label: sw_label.clone(),
383 loops_depth: self.loops.len(),
384 });
385 let mut case_groups = Vec::new();
386 for (vals, r) in cases {
387 let s = self.conv(r);
388 case_groups.push(crate::ir::stmt::CaseGroup {
389 labels: vals,
390 string_labels: vec![],
391 enum_labels: vec![],
392 raw_labels: vec![],
393 guard: None,
394 body: stmt_to_vec(s),
395 });
396 }
397 let default_stmt = default.map(|d| Box::new(self.conv(*d)));
398 self.switches.pop();
399 let mut head = self.block_stmts_no_term(block);
400 let mut sw = Stmt::Switch {
401 selector,
402 cases: case_groups,
403 default: default_stmt,
404 on_string: false,
405 };
406 if self.used_labels.contains(&sw_label) {
407 sw = Stmt::Labeled { label: sw_label, body: Box::new(sw) };
408 }
409 head.push(sw);
410 if head.len() == 1 {
411 head.pop().unwrap()
412 } else {
413 Stmt::Block(head)
414 }
415 }
416 Region::Try { group_idx, body, catches } => {
417 let mut body_stmt = self.conv(*body);
418 // Natural exits of the try body / handlers jump to the block
419 // right after the try span; that is structured fallthrough.
420 let gend = self.groups_end(group_idx);
421 let try_follow = gend.and_then(|e| self.cfg.block_at(e));
422 strip_trailing_goto(&mut body_stmt, try_follow);
423 let mut catch_stmts = Vec::new();
424 for (tys, _hblock, r) in catches {
425 let mut s = self.conv(*r);
426 strip_trailing_goto(&mut s, try_follow);
427 catch_stmts.push(Catch {
428 exc: tys,
429 var: u32::MAX, // assigned by the method pipeline (handler store)
430 var_name: None,
431 body: Box::new(s),
432 });
433 }
434 Stmt::Try { body: Box::new(body_stmt), catches: catch_stmts, finally: None }
435 }
436 Region::Goto { target } => {
437 if crate::dbg_flag!("JCDC_DBG_GOTO") {
438 eprintln!("conv Goto target={} cur_block={} loops={:?} switches={} if_follows={:?}",
439 target, self.cur_block,
440 self.loops.iter().map(|l| (l.header, l.exits.len())).collect::<Vec<_>>(),
441 self.switches.len(), self.if_follows);
442 }
443 // A jump to an enclosing if's follow that TERMINATES
444 // (return/throw) can be inlined: the copy ends this path
445 // exactly like the jump would, without needing a label.
446 // Same for a jump to a LOOP EXIT that heads a pure
447 // terminator chain (stmt-bearing fall-through links
448 // ending in return/throw): resolve_goto's loop-exit arm
449 // would turn it into a bare `break`, silently dropping
450 // the chain's statements between the loop end and the
451 // shared return (jdk26 Bits.reserveMemory phase-1:
452 // `if (interrupted) <goto interrupt-block>; ` rendered
453 // as else-break and the Thread.currentThread()
454 // .interrupt() vanished — interruption lost). The
455 // inlined copy keeps the abrupt ending (no fall-out
456 // duplication); a genuine fall-out exit still breaks.
457 //
458 let inline_terminator = self.if_follows.contains(&target)
459 && matches!(
460 self.results[target].term,
461 Term::Return(_) | Term::Throw(_)
462 )
463 && !self.tail_chain_writes_final(target);
464 if inline_terminator {
465 // Copy the WHOLE terminator chain: the target's own
466 // statements plus every fall-through link down to
467 // the final return/throw (a single-block chain is
468 // the original if-follow case).
469 let mut v: Vec<Stmt> = Vec::new();
470 let mut t = target;
471 let mut guard = 0;
472 loop {
473 v.extend(self.results[t].stmts.clone());
474 match &self.results[t].term {
475 Term::Return(e) => {
476 v.push(Stmt::Return(e.clone()));
477 break;
478 }
479 Term::Throw(e) => {
480 v.push(Stmt::Throw(e.clone()));
481 break;
482 }
483 Term::Fallthrough | Term::Goto
484 if self.cfg.blocks[t].succ.len() == 1 && guard < 8 =>
485 {
486 t = self.cfg.blocks[t].succ[0];
487 guard += 1;
488 }
489 _ => break,
490 }
491 }
492 return if v.len() == 1 {
493 v.into_iter().next().unwrap()
494 } else {
495 Stmt::Block(v)
496 };
497 }
498 if let Some(j) = self.resolve_goto(target) {
499 match j {
500 Jump::Break(lbl) => {
501 if let Some(l) = &lbl {
502 self.used_labels.insert(l.clone());
503 }
504 Stmt::Break(lbl)
505 }
506 Jump::Continue(lbl) => {
507 if let Some(l) = &lbl {
508 self.used_labels.insert(l.clone());
509 }
510 Stmt::Continue(lbl)
511 }
512 Jump::ContinueVia(t, lbl) => {
513 if let Some(l) = &lbl {
514 self.used_labels.insert(l.clone());
515 }
516 let mut cv = if self.prev_copy.get() == Some(t) {
517 // Statements already emitted by the
518 // immediately preceding CopyStmts.
519 Vec::new()
520 } else {
521 self.results[t].stmts.clone()
522 };
523 cv.push(Stmt::Continue(lbl));
524 if cv.len() == 1 {
525 cv.into_iter().next().unwrap()
526 } else {
527 Stmt::Block(cv)
528 }
529 }
530 Jump::RawGoto(t) => {
531 if crate::dbg_flag!("JCDC_DBG_GOTO") {
532 eprintln!("RAWGOTO t={} copied_tails={:?} if_follows={:?} last={}", t, self.copied_tails, self.if_follows, self.goto_is_last);
533 }
534 // A Goto region is always the last part of its
535 // walk, so the jump either falls through to the
536 // enclosing merge or must be inlined.
537 let term_copy = matches!(
538 self.results[t].term,
539 Term::Return(_) | Term::Throw(_)
540 ) && !self.tail_chain_writes_final(t)
541 && !self.is_retry_loop_header(t);
542 if term_copy {
543 let mut cv = self.results[t].stmts.clone();
544 match &self.results[t].term {
545 Term::Return(e) => cv.push(Stmt::Return(e.clone())),
546 Term::Throw(e) => cv.push(Stmt::Throw(e.clone())),
547 _ => {}
548 }
549 if cv.len() == 1 {
550 cv.into_iter().next().unwrap()
551 } else {
552 Stmt::Block(cv)
553 }
554 } else if self
555 .loops
556 .iter()
557 .any(|l| l.exits.contains(&t) || l.header == t)
558 && (self.goto_is_last || self.if_follows.contains(&t))
559 {
560 // The target is an ENCLOSING LOOP's exit:
561 // inside a loop body, falling out of an if
562 // continues the LOOP — the if-follow
563 // elision would silently drop the jump
564 // (jdk11 ObjectInputStream.readSerialData's
565 // finally-copy retry loop lost its `break`,
566 // the finally's while(true) never completed
567 // normally, and the tail went unreachable —
568 // 无法访问的语句 x2 trees). Materialize the
569 // break (innermost enclosing loop).
570 Stmt::Break(None)
571 } else if !matches!(
572 self.results[t].term,
573 Term::Return(_) | Term::Throw(_)
574 ) && !self.results[t].stmts.is_empty()
575 && !self.copied_tails.contains(&t)
576 && self.prev_copy.get() != Some(t)
577 && self
578 .expect_next
579 .get()
580 .map(|nx| {
581 // The next emission is a copy that
582 // STARTS PAST t (at a successor of t
583 // already in copied_tails): eliding
584 // would silently skip t's own
585 // statements, and no emitted copy
586 // covers them. (sj17 receive:
587 // Goto{16=SET}, expect=Some(17),
588 // 17 copied, 16 not.) If t itself is
589 // in copied_tails, an emitted full
590 // copy already includes t's stmts —
591 // inlining would double-emit (w26
592 // receive: Goto{14}, 14 copied).
593 let mut x = t;
594 for _ in 0..8 {
595 if x == nx {
596 return x != t;
597 }
598 if self.copied_tails.contains(&x) {
599 return false;
600 }
601 match self.cfg.blocks[x].succ.first() {
602 Some(&n) => x = n,
603 None => return false,
604 }
605 }
606 false
607 })
608 .unwrap_or(false)
609 && (self.if_follows.iter().any(|f| {
610 self.cfg.blocks[t].succ.contains(f)
611 }))
612 {
613 // Statement-bearing target FIRST: eliding
614 // (goto_is_last / expect / copy-tail) would
615 // silently skip t's own statements when the
616 // next emission is a copy of t's SUCCESSOR
617 // rather than t itself (jdk17
618 // DatagramChannelImpl.receive: Goto{16=
619 // `sender = sourceSocketAddress()`} elided
620 // via reaches_copy_tail(16->17) into the
621 // arm's Basic(17) copy — the n==0&&isOpen
622 // path returned a null sender). The inline
623 // below emits t's statements and then falls
624 // through to the same continuation.
625 let cv = self.results[t].stmts.clone();
626 if cv.len() == 1 {
627 cv.into_iter().next().unwrap()
628 } else {
629 Stmt::Block(cv)
630 }
631 } else if self.goto_is_last
632 || self.expect_next.get() == Some(t)
633 || self.reaches_copy_tail(t)
634 {
635 // Natural fallthrough reaches the same
636 // continuation the jump targeted.
637 Stmt::Block(vec![])
638 } else if !matches!(
639 self.results[t].term,
640 Term::Return(_) | Term::Throw(_)
641 ) && self.prev_copy.get() != Some(t)
642 && (self.if_follows.iter().any(|f| {
643 self.cfg.blocks[t].succ.contains(f)
644 }))
645 {
646 // The jump re-enters an already-structured
647 // block whose flow ends at an enclosing
648 // if-follow: inline its statements; the
649 // natural continuation matches the jump's.
650 let cv = self.results[t].stmts.clone();
651 if cv.len() == 1 {
652 cv.into_iter().next().unwrap()
653 } else {
654 Stmt::Block(cv)
655 }
656 } else {
657 Stmt::Goto(t as u32)
658 }
659 }
660 }
661 } else {
662 Stmt::Block(vec![])
663 }
664 }
665 }
666 }
667
668 /// Statements of a block including its terminal (return/throw), but NOT
669 /// the structural terminals (cond/switch/goto — handled by regions).
670 fn block_stmts(&mut self, block: usize) -> Stmt {
671 let mut v = self.block_stmts_no_term(block);
672 match &self.results[block].term {
673 Term::Return(e) => v.push(Stmt::Return(e.clone())),
674 Term::Throw(e) => v.push(Stmt::Throw(e.clone())),
675 Term::Jsr | Term::Ret => v.push(Stmt::Comment("// jsr/ret not supported".into())),
676 _ => {}
677 }
678 if v.len() == 1 {
679 v.pop().unwrap()
680 } else {
681 Stmt::Block(v)
682 }
683 }
684
685 fn block_stmts_no_term(&self, block: usize) -> Vec<Stmt> {
686 // Monitor markers are kept; synchronized reconstruction happens later.
687 self.results[block].stmts.clone()
688 }
689
690 /// Resolve a goto target against the scope stack.
691 fn resolve_goto(&mut self, target: usize) -> Option<Jump> {
692 // continue: innermost loop whose header == target
693 if let Some(i) = (0..self.loops.len()).rev().find(|&i| self.loops[i].header == target) {
694 let depth = self.loops.len() - 1 - i;
695 return Some(if depth == 0 {
696 Jump::Continue(None)
697 } else {
698 Jump::Continue(Some(self.loops[i].label.clone()))
699 });
700 }
701 // break out of a loop whose exit set contains the target; pick the
702 // OUTERMOST such loop — an exit of nested loops (e.g. `break outer`)
703 // belongs to every enclosing loop's exit set, and breaking only the
704 // innermost would wrongly re-enter the outer loop.
705 if let Some(li) = (0..self.loops.len()).find(|&i| self.loops[i].exits.contains(&target)) {
706 // RESTART-STUB exit: the target chains through statement-free
707 // stubs back into an enclosing-or-self loop HEADER — the jump
708 // re-iterates THAT loop, it does not break out of anything
709 // (jdk11/17/26 ConcurrentLinkedQueue.poll's p==q arm `goto
710 // restartFromHead`: exit 12 is the `goto 0` stub; rendering
711 // break+break fell to the post-loop tail and returned the
712 // STALE item instead of restarting from head). The broad
713 // walk-side continue-stub precedence was reverted once
714 // (Pattern.clazz: switch-case goto stubs must stay
715 // RawGoto-elided so the emitter re-appends case breaks) —
716 // this form is scoped to genuine loop exits with no open
717 // switch on the stack.
718 if self.switches.is_empty() {
719 let mut x = target;
720 let mut guard = 0;
721 let header_hit = loop {
722 if let Some(hi) = self.loops.iter().position(|l| l.header == x) {
723 if hi <= li {
724 break Some(hi);
725 }
726 }
727 if guard >= 8 {
728 break None;
729 }
730 let r = &self.results[x];
731 let stub = r.stmts.is_empty()
732 && r.out_stack.is_empty()
733 && matches!(r.term, Term::Fallthrough | Term::Goto)
734 && self.cfg.blocks[x].succ.len() == 1;
735 if !stub {
736 break None;
737 }
738 x = self.cfg.blocks[x].succ[0];
739 guard += 1;
740 };
741 if let Some(hi) = header_hit {
742 let depth = self.loops.len() - 1 - hi;
743 return Some(if depth == 0 {
744 Jump::Continue(None)
745 } else {
746 Jump::Continue(Some(self.loops[hi].label.clone()))
747 });
748 }
749 }
750 // switches opened after that loop intercept a plain `break`
751 let crosses_switch = !self.switches.is_empty();
752 return Some(if li + 1 == self.loops.len() && !crosses_switch {
753 Jump::Break(None)
754 } else {
755 Jump::Break(Some(self.loops[li].label.clone()))
756 });
757 }
758 if let Some(si) = (0..self.switches.len()).rev().find(|&i| self.switches[i].follow == Some(target)) {
759 return Some(if si + 1 == self.switches.len() && self.loops.len() == self.switches[si].loops_depth {
760 Jump::Break(None)
761 } else {
762 Jump::Break(Some(self.switches[si].label.clone()))
763 });
764 }
765 // Natural flow to the enclosing if's merge point: no statement.
766 // Only the INNERMOST open follow qualifies — eliding a jump to an
767 // outer follow would silently skip the statements between (e.g. a
768 // shared throw at the end of an else-branch).
769 if self.if_follows.last() == Some(&target) {
770 return None;
771 }
772 // Inside a loop, a jump to an already-structured block that flows
773 // back into the loop is a CONTINUE of the innermost such loop: the
774 // target's statements were emitted earlier in this body, and the
775 // natural continuation after this region is the loop bottom (an
776 // if-follow or a fall-out exit), never the target. Eliding it as
777 // fallthrough (the RawGoto `goto_is_last` arm) silently dropped the
778 // iteration: jdk11/17 InstantPrinterParser.format's second loop
779 // copy emitted `if (i >= fractionalDigits) {}` with no way out --
780 // the digit append + back edge vanished (semantically an infinite
781 // loop; javac: 缺少返回语句 on the method tail).
782 // A target that is a loop EXIT keeps its RawGoto resolution (the
783 // conv arm materializes the break): Pattern.clazz's switch-case
784 // `break` targets sit on the loop-bottom path AND name exits —
785 // continuing there made every switch exit abrupt and the post-
786 // switch tail unreachable (无法访问的语句 x3 trees).
787 let is_any_exit = self.loops.iter().any(|l| l.exits.contains(&target));
788 if !is_any_exit && self.switches.is_empty() {
789 // STUB-TARGET continue: a statement-free confluence stub that
790 // can reach an enclosing header re-iterates that loop (its own
791 // flow was emitted earlier in this body copy; IPP's can_reach
792 // targets t=43/18/96/89/114/21/7 are all stubs).
793 // HEADER-ADJACENT statement-bearing target: the jump re-runs a
794 // shared tail whose statements were NOT emitted on this path —
795 // inline them and continue (jdk11 BigInteger.nextProbablePrime
796 // sieve arms: Goto{result=result.add(TWO)} flowing straight to
797 // the header; a bare Continue skipped the advance and left
798 // while(true) as the method's only completion: 缺少返回语句).
799 // Statement-bearing targets with a BRANCHING or statement
800 // bearing chain to the header keep RawGoto resolution (huc
801 // getInputStream0's clone/getValue tails flow through shared
802 // forward code — continuing there would skip it).
803 let mut via_label: Option<Option<String>> = None;
804 if self.results[target].stmts.is_empty() {
805 if let Some(i) = (0..self.loops.len())
806 .rev()
807 .find(|&i| {
808 crate::structure::can_reach_cfg(self.cfg, target, self.loops[i].header, 4096)
809 })
810 {
811 let depth = self.loops.len() - 1 - i;
812 via_label = Some(if depth == 0 {
813 None
814 } else {
815 Some(self.loops[i].label.clone())
816 });
817 }
818 } else {
819 let mut x = target;
820 for _ in 0..8 {
821 if let Some(hi) = self.loops.iter().position(|l| l.header == x) {
822 let depth = self.loops.len() - 1 - hi;
823 via_label = Some(if depth == 0 {
824 None
825 } else {
826 Some(self.loops[hi].label.clone())
827 });
828 break;
829 }
830 if !self.results[x].stmts.is_empty() && x != target {
831 break;
832 }
833 if self.cfg.blocks[x].succ.len() != 1 {
834 break;
835 }
836 x = self.cfg.blocks[x].succ[0];
837 }
838 }
839 if let Some(lbl) = via_label {
840 return Some(if self.results[target].stmts.is_empty() {
841 Jump::Continue(lbl)
842 } else {
843 Jump::ContinueVia(target, lbl)
844 });
845 }
846 }
847 // Inside a loop, a forward escape that cannot reach back into the
848 // loop is a break of the innermost loop.
849 if !self.loops.is_empty()
850 && !crate::structure::can_reach_cfg(
851 self.cfg,
852 target,
853 self.loops[self.loops.len() - 1].header,
854 4096,
855 )
856 {
857 return Some(Jump::Break(None));
858 }
859 // Unstructured jump.
860 Some(Jump::RawGoto(target))
861 }
862
863 /// True if `a` post-dominates `b` (every path from b exits through a).
864 /// True when a jump to `t` is equivalent to natural fallthrough: `t`
865 /// (or the end of a chain of statement-free blocks starting at `t`) is
866 /// a copied shared tail or an open if-follow that the enclosing flow
867 /// will emit next anyway.
868 /// True when `t` is the head of a javac retry loop: some exception
869 /// handler protecting `t` has no normal preds and flows straight back
870 /// to `t` (`catch (InterruptedException e) { ..; goto head }`). A Goto
871 /// to such a header re-executes the protected body — inlining the
872 /// body's terminator as a "shared tail copy" drops it out of its try
873 /// (jdk26 Future.exceptionNow: the IE catch got `get(); throw ISE`
874 /// unprotected — 未报告的异常错误 InterruptedException).
875 fn is_retry_loop_header(&self, t: usize) -> bool {
876 self.cfg.exc_edges.iter().any(|e| {
877 e.from == t
878 && e.to != t
879 && self.cfg.blocks[e.to].pred.is_empty()
880 && self.cfg.blocks[e.to].succ.len() == 1
881 && self.cfg.blocks[e.to].succ[0] == t
882 })
883 }
884
885 fn reaches_copy_tail(&self, t: usize) -> bool {
886 let mut x = t;
887 for _ in 0..8 {
888 if self.copied_tails.contains(&x) || self.if_follows.contains(&x) {
889 return true;
890 }
891 if !self.results[x].stmts.is_empty() {
892 return false;
893 }
894 match &self.results[x].term {
895 Term::Fallthrough | Term::Goto => {
896 match self.cfg.blocks[x].succ.first() {
897 Some(&n) => x = n,
898 None => return false,
899 }
900 }
901 _ => return false,
902 }
903 }
904 false
905 }
906
907#[allow(dead_code)]
908 fn postdominates(&self, a: usize, b: usize, universe: &HashSet<usize>) -> bool {
909 // BFS from b avoiding a; if no exit block is reachable, a post-dominates.
910 let mut seen = HashSet::new();
911 let mut q = std::collections::VecDeque::new();
912 q.push_back(b);
913 seen.insert(b);
914 while let Some(x) = q.pop_front() {
915 for &s in &self.cfg.blocks[x].succ {
916 if s == a {
917 continue;
918 }
919 if !universe.contains(&s) {
920 continue;
921 }
922 if seen.insert(s) {
923 q.push_back(s);
924 }
925 }
926 if self.cfg.blocks[x].succ.is_empty() && x != a && x != b {
927 // reached an exit without passing through a
928 return false;
929 }
930 }
931 // also treat blocks leaving the universe as exits
932 true
933 }
934
935 /// Classify a loop region into while / do-while / infinite and assemble
936 /// the final statement.
937 fn classify_loop(&mut self, header: usize, body_stmt: Stmt, ctx: LoopCtx) -> Stmt {
938 let term = self.results[header].term.clone();
939 let succs = self.cfg.blocks[header].succ.clone();
940
941 match term {
942 Term::Cond { cond } if succs.len() == 2 => {
943 let taken = succs[1];
944 let fall = succs[0];
945 if taken == header && ctx.exits.contains(&fall) {
946 // Self-loop: the header block holds both the body and
947 // the trailing conditional back edge → do-while. The
948 // fallthrough side is the loop exit. (When the fall
949 // side is INTERIOR the self-edge is an in-body retry
950 // jump of a loop with no normal completion at all —
951 // rotating it into a do-while bottom invents an exit
952 // path: jdk11/17/26 DSAParameterGenerator.generatePandQ's
953 // `if (!resultQ.isProbablePrime(..)) goto <Q-gen>`
954 // self-retry falls through into the P-search interior;
955 // the fabricated do-while fell out to the method end —
956 // 缺少返回语句. Those shapes take the interior-diamond
957 // while(true) arm below.)
958 // A compound (multi-test) trailing condition folds first:
959 // `if(c1)continue; .. if(cN)continue; else exit;` becomes
960 // `do{body}while(c1||..||cN); exit`.
961 if let Some((stmts, c, exit)) = extract_compound_do_while(&body_stmt) {
962 // An extracted exit that is the loop's OWN unlabeled
963 // break must be dropped: the do-while condition false
964 // already exits, and re-emitting the break OUTSIDE the
965 // loop is a stray `break;` (compile error). SESE's
966 // explicit `Goto{header}` -> continue makes self-loop
967 // bodies hit this shape (`if (c) continue; else break;`).
968 // return/throw/labeled-break exits stay (real fall-out).
969 if matches!(exit, Stmt::Break(None)) {
970 return Stmt::DoWhile {
971 body: Box::new(Stmt::Block(stmts)),
972 cond: c,
973 };
974 }
975 return Stmt::Block(vec![
976 Stmt::DoWhile { body: Box::new(Stmt::Block(stmts)), cond: c },
977 exit,
978 ]);
979 }
980 let (stmts, c) = match extract_trailing_do_while(&body_stmt) {
981 Some(x) => x,
982 None => (stmt_to_vec(body_stmt), cond.clone()),
983 };
984 return Stmt::DoWhile {
985 body: Box::new(Stmt::Block(stmts)),
986 cond: c,
987 };
988 }
989 if fall == header && ctx.exits.contains(&taken) {
990 // Inverted self-loop: `while (!(c)) body` — the taken
991 // side exits, the fallthrough loops back.
992 let inner2 = strip_trailing_continue(body_stmt);
993 return Stmt::While {
994 cond: negate(cond.clone()),
995 body: Box::new(inner2),
996 };
997 }
998 let taken_is_exit = ctx.exits.contains(&taken);
999 if crate::dbg_flag!("JCDC_DBG_LOOP") {
1000 eprintln!("CLASSIFY2 header={} taken={} fall={} exits={:?}", header, taken, fall, ctx.exits);
1001 }
1002 if !taken_is_exit && !ctx.exits.contains(&fall) {
1003 // Multi-block loop condition. The header's fallthrough
1004 // may be a pure condition-continuation test block whose
1005 // taken side is the loop exit and whose fall side rejoins
1006 // the body (`taken`). javac compiles `while (cH || !cB)`
1007 // exactly this way:
1008 // H: if cH goto body; B: if cB goto exit; body: ..; goto H
1009 // Fold B into the while condition. Without this the body
1010 // walk re-emits H and B as empty `if`s inside a
1011 // `while(true)`, dropping the exit and leaving an infinite
1012 // loop plus an unreachable trailing statement.
1013 if self.results[fall].stmts.is_empty() {
1014 if let Term::Cond { cond: cb } = &self.results[fall].term {
1015 let bs = self.cfg.blocks[fall].succ.clone();
1016 if bs.len() == 2 && ctx.exits.contains(&bs[1]) && bs[0] == taken {
1017 let combined = Expr::Bin {
1018 op: BinOp::LogOr,
1019 l: Box::new(cond.clone()),
1020 r: Box::new(negate(cb.clone())),
1021 ty: None,
1022 };
1023 // The body is the converted region
1024 // remainder, NOT the taken block's raw
1025 // statements: when `taken` is wrapped in
1026 // regions (a try/catch group around the
1027 // body call), the raw-stmt body silently
1028 // drops them (jdk11 SocketChannelImpl
1029 // .implCloseSelectableChannel: the
1030 // wait()-loop's try/catch(Interrupted
1031 // Exception) region vanished — bare
1032 // wait() — 未报告的异常错误). The
1033 // leading guard If is the condition being
1034 // folded here; strip it and keep the rest.
1035 let mut parts = stmt_to_vec(body_stmt);
1036 if matches!(parts.first(), Some(Stmt::If { .. })) {
1037 parts.remove(0);
1038 }
1039 let body = if parts.is_empty() {
1040 Stmt::Block(self.results[taken].stmts.clone())
1041 } else {
1042 let rest = if parts.len() == 1 {
1043 parts.pop().unwrap()
1044 } else {
1045 Stmt::Block(parts)
1046 };
1047 strip_trailing_continue(rest)
1048 };
1049 if crate::dbg_flag!("JCDC_DBG_LOOP") {
1050 eprintln!(
1051 "CLASSIFY-CONDCHAIN header={} B={} body={} exits={:?}",
1052 header, fall, taken, ctx.exits
1053 );
1054 }
1055 return Stmt::While { cond: combined, body: Box::new(body) };
1056 }
1057 }
1058 }
1059 // Neither successor leaves the loop: the header's Cond
1060 // is an interior diamond (e.g. a ternary inside an
1061 // infinite `for(;;)` whose back edge targets the header
1062 // top), not an exit test. Keep it in the body.
1063 return Stmt::While {
1064 cond: Expr::Const(ConstVal::Int(1)),
1065 body: Box::new(body_stmt),
1066 };
1067 }
1068 // The body walk re-emitted the header's own If region; unwrap
1069 // it so the loop reads `while (C) { else-part }`.
1070 let (inner, exit_stmts) = split_leading_if(&body_stmt, &cond);
1071 // A top test the structurer wrapped in a TRY (the header
1072 // block sits inside its handler's protected range — jdk26
1073 // Bits.reserveMemory's backoff retry loop tests
1074 // tryReserveOrClean(..) throws InterruptedException under
1075 // its own catch(IE) retry handler): rotating the test into
1076 // `while (C)` evaluates the call OUTSIDE every handler —
1077 // 未报告的异常错误InterruptedException. split_leading_if
1078 // deliberately cannot reach through the Try (the rotated
1079 // cond would still be unprotected), so keep the test where
1080 // the bytecode has it and stay on while(true). The earlier
1081 // implCloseSelectableChannel guard (raw-stmt bodies losing
1082 // the leading If's statements) does not apply: the If lives
1083 // intact inside the Try.
1084
1085 if crate::dbg_flag!("JCDC_DBG_LOOP") {
1086 eprintln!(
1087 "CLASSIFY header={} taken_is_exit={} exit_stmts={}",
1088 header,
1089 taken_is_exit,
1090 exit_stmts.len()
1091 );
1092 }
1093 let inner = strip_trailing_continue(inner);
1094 let while_cond = if taken_is_exit { negate(cond.clone()) } else { cond.clone() };
1095 let plain_exit = exit_stmts.is_empty()
1096 || matches!(exit_stmts.as_slice(), [Stmt::Break(_)])
1097 || matches!(exit_stmts.as_slice(), [Stmt::Block(b)] if b.is_empty());
1098 if taken_is_exit && plain_exit {
1099 match try_protected_dowhile(inner, &cond) {
1100 Ok(dw) => dw,
1101 Err(inner) => Stmt::While { cond: while_cond, body: Box::new(inner) },
1102 }
1103 } else if taken_is_exit {
1104 // Exit branch runs statements before leaving:
1105 // while (true) { if (C_exit) { stmts; break; } body }
1106 // `cond` IS the exit test here (the taken side exits —
1107 // negate(cond) inverted the guard: jdk17 CHM
1108 // comparableClassFor emitted `if (i < len) return null`
1109 // inside while(true), inverting the loop). The break is
1110 // only appended when the exit statements do not already
1111 // terminate (return null + break = unreachable stmt).
1112 let mut ex = exit_stmts;
1113 let already_term = matches!(
1114 ex.last(),
1115 Some(Stmt::Return(_)) | Some(Stmt::Throw(_)) | Some(Stmt::Break(_))
1116 );
1117 if !already_term {
1118 ex.push(Stmt::Break(None));
1119 }
1120 let guard = Stmt::If {
1121 cond: cond.clone(),
1122 then_stmt: Box::new(Stmt::Block(ex)),
1123 else_stmt: None,
1124 };
1125 let mut v = vec![guard];
1126 v.extend(stmt_to_vec(inner));
1127 Stmt::While { cond: Expr::Const(ConstVal::Int(1)), body: Box::new(Stmt::Block(v)) }
1128 } else {
1129 // Unusual orientation: the fallthrough side exits.
1130 if exit_stmts.is_empty()
1131 || matches!(exit_stmts.as_slice(), [Stmt::Break(_)])
1132 || matches!(exit_stmts.as_slice(), [Stmt::Block(b)] if b.is_empty())
1133 {
1134 // then-side was the exit scaffolding; body is `inner`
1135 match try_protected_dowhile(inner, &cond) {
1136 Ok(dw) => dw,
1137 Err(inner) => Stmt::While { cond: while_cond, body: Box::new(inner) },
1138 }
1139 } else {
1140 // then-side carries the body ending with the back
1141 // edge; fallthrough exits.
1142 let body2 = strip_trailing_continue(stmts_to_stmt(exit_stmts));
1143 Stmt::While { cond: while_cond, body: Box::new(body2) }
1144 }
1145 }
1146 }
1147 Term::Goto | Term::Fallthrough => {
1148 // Possibly do-while: the body ends with a conditional jump
1149 // back to the header (`if (c) continue;` or inverted), or a
1150 // COMPOUND run of such tests ending in the loop exit.
1151 if let Some((stmts, c, exit)) = extract_compound_do_while(&body_stmt) {
1152 // Same guard as the self-loop compound path above: an
1153 // extracted exit that is the loop's OWN unlabeled break
1154 // must be dropped — the do-while condition false already
1155 // exits, and a `break;` re-emitted OUTSIDE the loop is a
1156 // stray (it also makes stmt_terminates treat the sequence
1157 // as terminated, so prune_unreachable drops the real
1158 // tail: jdk26 Resolver.bind `return this` — 缺少返回语句).
1159 // SESE's rotated do-whiles hit this shape: the tail test
1160 // block becomes `if (c) continue; else break;`.
1161 // return/throw/labeled-break exits stay (real fall-out).
1162 if matches!(exit, Stmt::Break(None)) {
1163 return Stmt::DoWhile {
1164 body: Box::new(Stmt::Block(stmts)),
1165 cond: c,
1166 };
1167 }
1168 return Stmt::Block(vec![
1169 Stmt::DoWhile { body: Box::new(Stmt::Block(stmts)), cond: c },
1170 exit,
1171 ]);
1172 }
1173 if let Some((stmts, c)) = extract_trailing_do_while(&body_stmt) {
1174 return Stmt::DoWhile { body: Box::new(Stmt::Block(stmts)), cond: c };
1175 }
1176 Stmt::While {
1177 cond: Expr::Const(ConstVal::Int(1)),
1178 body: Box::new(body_stmt),
1179 }
1180 }
1181 _ => Stmt::While {
1182 cond: Expr::Const(ConstVal::Int(1)),
1183 body: Box::new(body_stmt),
1184 },
1185 }
1186 }
1187
1188 /// True if `target` is inside the loop currently being classified
1189 /// (approximated: target can reach the header again).
1190#[allow(dead_code)]
1191 fn loop_contains(&self, header: usize, target: usize) -> bool {
1192 can_reach(self.cfg, target, header, 8192)
1193 }
1194
1195#[allow(dead_code)]
1196 fn loop_exit_contains(&self, header: usize, target: usize) -> bool {
1197 // The loop currently being classified is the innermost on the stack
1198 // BEFORE it was popped; ctx.exits was consumed. Recompute cheaply:
1199 // target is an exit if it is not reachable-back-to-header within the
1200 // loop members — approximated by checking the parent scope recorded
1201 // in loops stack is empty here; use cfg: header's own successors that
1202 // leave... For the header itself, an exit is any successor that is
1203 // not dominated by header OR is not in the natural loop. Simple
1204 // approximation: a successor s is "in loop" if s can reach header
1205 // again via normal edges.
1206 can_reach(self.cfg, target, header, 4096).not()
1207 }
1208}
1209
1210#[allow(dead_code)]
1211trait NotExt {
1212 fn not(self) -> bool;
1213}
1214impl NotExt for bool {
1215 fn not(self) -> bool {
1216 !self
1217 }
1218}
1219
1220#[allow(dead_code)]
1221fn can_reach(cfg: &Cfg, from: usize, to: usize, budget: usize) -> bool {
1222 if from == to {
1223 return true;
1224 }
1225 let mut seen = HashSet::new();
1226 let mut q = std::collections::VecDeque::new();
1227 q.push_back(from);
1228 seen.insert(from);
1229 let mut n = 0;
1230 while let Some(b) = q.pop_front() {
1231 n += 1;
1232 if n > budget {
1233 return false;
1234 }
1235 for &s in &cfg.blocks[b].succ {
1236 if s == to {
1237 return true;
1238 }
1239 if seen.insert(s) {
1240 q.push_back(s);
1241 }
1242 }
1243 }
1244 false
1245}
1246
1247/// If the body starts with the loop header's own `If` (same condition
1248/// expression), return (else-branch content, then-branch content).
1249/// Otherwise return (body, []).
1250fn stmts_to_stmt(v: Vec<Stmt>) -> Stmt {
1251 match v.len() {
1252 0 => Stmt::Block(vec![]),
1253 1 => v.into_iter().next().unwrap(),
1254 _ => Stmt::Block(v),
1255 }
1256}
1257
1258/// When the loop's top test lives inside a Try in `inner` (the header
1259/// block sits in its handler's protected range — jdk26
1260/// Bits.reserveMemory's backoff retry loop tests
1261/// `tryReserveOrClean(..) throws InterruptedException` under its own
1262/// catch(IE) handler), the `while (C)` rotation would evaluate C
1263/// OUTSIDE every handler — 未报告的异常错误InterruptedException.
1264/// Rewrite as a do-while instead: inject `break` into the embedded
1265/// If's EMPTY arm (the exit side — the exit blocks live outside the
1266/// loop region as its continuation siblings, so the empty arm is the
1267/// fall-out) and loop unconditionally. The sibling tail then stays
1268/// reachable as the do-while's natural completion. Returns None when
1269/// the test is not try-wrapped (plain rotation applies) or when the
1270/// shape is ambiguous (both If arms materialized).
1271fn try_protected_dowhile(inner: Stmt, cond: &Expr) -> Result<Stmt, Stmt> {
1272 // The exit side of the embedded test: an Empty region conversion
1273 // (empty block / missing else) or an ALREADY-materialized plain
1274 // `break` (the merged-span shape walks the sleep region as the
1275 // then-arm and breaks on the exit side directly).
1276 fn is_empty(s: &Stmt) -> bool {
1277 match s {
1278 Stmt::Block(v) => v.is_empty() || (v.len() == 1 && is_empty(&v[0])),
1279 Stmt::Break(None) => false,
1280 _ => false,
1281 }
1282 }
1283 fn inject(s: &mut Stmt, cond: &Expr) -> bool {
1284 match s {
1285 Stmt::If { cond: c, then_stmt, else_stmt } if c == cond => {
1286 let then_empty = is_empty(then_stmt);
1287 let else_empty =
1288 else_stmt.as_ref().map(|e| is_empty(e)).unwrap_or(true);
1289 if then_empty && !else_empty {
1290 *then_stmt = Box::new(Stmt::Break(None));
1291 true
1292 } else if else_empty && !then_empty {
1293 *else_stmt = Some(Box::new(Stmt::Break(None)));
1294 true
1295 } else {
1296 false
1297 }
1298 }
1299 Stmt::Try { body, .. } | Stmt::TryWithResources { body, .. } => {
1300 inject(body, cond)
1301 }
1302 Stmt::Block(v) => v.first_mut().map(|f| inject(f, cond)).unwrap_or(false),
1303 Stmt::Labeled { body, .. } | Stmt::Synchronized { body, .. } => inject(body, cond),
1304 _ => false,
1305 }
1306 }
1307 let mut b = inner;
1308 if inject(&mut b, cond) {
1309 Ok(Stmt::DoWhile {
1310 body: Box::new(b),
1311 cond: Expr::Const(ConstVal::Int(1)),
1312 })
1313 } else {
1314 Err(b)
1315 }
1316}
1317
1318fn split_leading_if(body: &Stmt, cond: &Expr) -> (Stmt, Vec<Stmt>) {
1319 let items = match body {
1320 Stmt::Block(v) => v,
1321 other => {
1322 return match other {
1323 Stmt::If { cond: c, then_stmt, else_stmt } if c == cond => (
1324 else_stmt.as_ref().map(|e| (**e).clone()).unwrap_or(Stmt::Block(vec![])),
1325 stmt_to_vec((**then_stmt).clone()),
1326 ),
1327 _ => (other.clone(), vec![]),
1328 };
1329 }
1330 };
1331 if let Some(first) = items.first() {
1332 if let Stmt::If { cond: c, then_stmt, else_stmt } = first {
1333 if c == cond {
1334 let inner = if items.len() == 1 {
1335 else_stmt.as_ref().map(|e| (**e).clone()).unwrap_or(Stmt::Block(vec![]))
1336 } else {
1337 let mut rest = vec![else_stmt
1338 .as_ref()
1339 .map(|e| (**e).clone())
1340 .unwrap_or(Stmt::Block(vec![]))];
1341 rest.extend(items[1..].iter().cloned());
1342 Stmt::Block(rest)
1343 };
1344 return (inner, stmt_to_vec((**then_stmt).clone()));
1345 }
1346 }
1347 }
1348 (body.clone(), vec![])
1349}
1350
1351/// Remove a trailing `continue;` (loop back-edge) from a statement list.
1352fn strip_trailing_continue(s: Stmt) -> Stmt {
1353 match s {
1354 Stmt::Block(mut v) => {
1355 while matches!(v.last(), Some(Stmt::Continue(None))) {
1356 v.pop();
1357 }
1358 if v.len() == 1 && matches!(v[0], Stmt::Block(_)) {
1359 let inner = match v.pop().unwrap() {
1360 Stmt::Block(inner) => inner,
1361 _ => unreachable!(),
1362 };
1363 return strip_trailing_continue(Stmt::Block(inner));
1364 }
1365 Stmt::Block(v)
1366 }
1367 Stmt::Continue(None) => Stmt::Block(vec![]),
1368 other => other,
1369 }
1370}
1371
1372pub fn negate(e: Expr) -> Expr {
1373 match e {
1374 Expr::Un { op: UnOp::Not, e } => *e,
1375 // De Morgan for the short-circuit operators: !(A && B) =
1376 // !A || !B, !(A || B) = !A && !B — recursing through negate
1377 // keeps comparison operands cleanly inverted. An op-only swap
1378 // (the historical invert() LogAnd<->LogOr mapping) silently
1379 // changed the condition's meaning: Arrays.equals' folded
1380 // `a == null || a2 == null` inverted to `a == null && a2 ==
1381 // null`, NPE-ing the a2==null path.
1382 Expr::Bin { op: op @ (crate::ir::expr::BinOp::LogAnd | crate::ir::expr::BinOp::LogOr), l, r, ty } => {
1383 let swapped = if matches!(op, crate::ir::expr::BinOp::LogAnd) {
1384 crate::ir::expr::BinOp::LogOr
1385 } else {
1386 crate::ir::expr::BinOp::LogAnd
1387 };
1388 Expr::Bin {
1389 op: swapped,
1390 l: Box::new(negate(*l)),
1391 r: Box::new(negate(*r)),
1392 ty,
1393 }
1394 }
1395 Expr::Bin { op, l, r, ty } => {
1396 if let Some(inv) = op.invert() {
1397 Expr::Bin { op: inv, l, r, ty }
1398 } else {
1399 Expr::Un { op: UnOp::Not, e: Box::new(Expr::Bin { op, l, r, ty }) }
1400 }
1401 }
1402 other => Expr::Un { op: UnOp::Not, e: Box::new(other) },
1403 }
1404}
1405
1406/// True for a bare `continue;` (or a block wrapping just one).
1407fn is_continue_stmt(s: &Stmt) -> bool {
1408 matches!(s, Stmt::Continue(None))
1409 || matches!(s, Stmt::Block(v) if v.len() == 1 && matches!(v[0], Stmt::Continue(None)))
1410}
1411
1412/// Detect a COMPOUND trailing do-while condition: the body ends with a run of
1413/// `if (c1) continue;` ... `if (cN) continue; else <exit>;`. javac emits this
1414/// for `do { body } while (c1 || ... || cN);` followed by the loop exit when
1415/// the condition is a short-circuit `||` whose tests each jump back to the
1416/// body head (e.g. `Random.internalNextInt`: `while (r < origin || r >=
1417/// bound)`). Fold the run into a single do-while whose condition is the `||`
1418/// of the tests, returning the body statements, that condition, and the exit
1419/// action (`<exit>`) to emit right after the loop.
1420fn extract_compound_do_while(body: &Stmt) -> Option<(Vec<Stmt>, Expr, Stmt)> {
1421 // Flatten nested straight-line blocks: the loop body may group the first
1422 // condition test(s) inside a nested Block (the self-loop header's region)
1423 // while later tests are siblings. Splicing pure sequence blocks is safe.
1424 fn flatten_seq(s: &Stmt) -> Vec<Stmt> {
1425 match s {
1426 Stmt::Block(v) => {
1427 let mut out = Vec::new();
1428 for x in v {
1429 out.extend(flatten_seq(x));
1430 }
1431 out
1432 }
1433 other => vec![other.clone()],
1434 }
1435 }
1436 let mut stmts = flatten_seq(body);
1437 // Unnest the rendered if/else-if CHAIN spine: javac's bottom-tested
1438 // `if (c1) continue; if (c2) continue; ... else exit;` disjunct run
1439 // converts to ONE nested If whose else arms hold the remaining
1440 // tests — opaque to the backwards scan below, which would fold only
1441 // c1 and push the whole c2..cN chain out of the loop as the "exit"
1442 // (jdk26 ML_DSA.signInternal: `do { expandMask.. } while
1443 // (vectorNormBound(z,..));` with the c2/hint chain — and its
1444 // continue/break — stranded after the loop, continue 在 loop 外部
1445 // ×2). Splitting each continue-then If back into siblings restores
1446 // the flat run; semantics are identical (the then arm is a bare
1447 // continue with no fallthrough, and the LAST else is preserved as
1448 // the final statement's else = the loop's exit route).
1449 {
1450 let mut flat: Vec<Stmt> = Vec::new();
1451 for st in stmts.into_iter() {
1452 let mut cur = st;
1453 loop {
1454 match cur {
1455 Stmt::If { cond, then_stmt, else_stmt: Some(e) }
1456 if is_continue_stmt(&then_stmt) =>
1457 {
1458 flat.push(Stmt::If {
1459 cond,
1460 then_stmt,
1461 else_stmt: None,
1462 });
1463 cur = *e;
1464 }
1465 other => {
1466 flat.push(other);
1467 break;
1468 }
1469 }
1470 }
1471 }
1472 stmts = flat;
1473 }
1474 let n = stmts.len();
1475 if n < 2 {
1476 return None;
1477 }
1478 // Last statement: `if (c_last) continue; else <exit>` — the exit is the
1479 // loop's fall-out (a return/throw/break), NOT another continue.
1480 let (c_last, exit) = match &stmts[n - 1] {
1481 Stmt::If { cond, then_stmt, else_stmt: Some(e) }
1482 if is_continue_stmt(then_stmt) && !is_continue_stmt(e) =>
1483 {
1484 (cond.clone(), (**e).clone())
1485 }
1486 _ => return None,
1487 };
1488 // Walk backwards over preceding `if (ci) continue;` (no else) tests.
1489 let mut conds = vec![c_last];
1490 let mut i = n - 1;
1491 while i >= 1 {
1492 match &stmts[i - 1] {
1493 Stmt::If { cond, then_stmt, else_stmt: None } if is_continue_stmt(then_stmt) => {
1494 conds.push(cond.clone());
1495 i -= 1;
1496 }
1497 _ => break,
1498 }
1499 }
1500 let body_stmts: Vec<Stmt> = stmts[..i].to_vec();
1501 // cond = c1 || c2 || ... || c_last (conds currently reversed).
1502 let mut it = conds.into_iter().rev();
1503 let mut cond = it.next().unwrap();
1504 for c in it {
1505 cond = Expr::Bin { op: BinOp::LogOr, l: Box::new(c), r: Box::new(cond), ty: None };
1506 }
1507 Some((body_stmts, cond, exit))
1508}
1509
1510/// Detect a trailing do-while condition at the end of a loop body:
1511/// `if (c) continue;` → cond c, or `if (c) {} else continue;` → cond !c.
1512fn extract_trailing_do_while(body: &Stmt) -> Option<(Vec<Stmt>, Expr)> {
1513 let mut stmts = match body {
1514 Stmt::Block(v) => v.clone(),
1515 other => vec![other.clone()],
1516 };
1517 if stmts.is_empty() {
1518 return None;
1519 }
1520 let n = stmts.len() - 1;
1521 if let Stmt::If { cond, then_stmt, else_stmt } = &stmts[n] {
1522 let then_is_cont = |s: &Stmt| {
1523 matches!(s, Stmt::Continue(None))
1524 || matches!(s, Stmt::Block(v) if v.len() == 1 && matches!(v[0], Stmt::Continue(None)))
1525 };
1526 if then_is_cont(then_stmt) && else_stmt.is_none() {
1527 let c = cond.clone();
1528 stmts.pop();
1529 return Some((stmts, c));
1530 }
1531 if else_stmt.as_ref().map(|e| then_is_cont(e)).unwrap_or(false)
1532 && (matches!(&**then_stmt, Stmt::Block(v) if v.is_empty()))
1533 {
1534 let c = negate(cond.clone());
1535 stmts.pop();
1536 return Some((stmts, c));
1537 }
1538 }
1539 None
1540}
1541
1542/// If the body ends with `if (c) continue;` (Continue(None), no else),
1543/// strip it and return (body without trailing if, c).
1544#[allow(dead_code)]
1545fn extract_trailing_continue_if(body: &Stmt) -> Option<(Vec<Stmt>, Expr)> {
1546 let mut stmts = match body {
1547 Stmt::Block(v) => v.clone(),
1548 other => vec![other.clone()],
1549 };
1550 if stmts.is_empty() {
1551 return None;
1552 }
1553 let n = stmts.len() - 1;
1554 // Direct trailing `if (c) continue;`.
1555 let is_direct = match &stmts[n] {
1556 Stmt::If { then_stmt, else_stmt: None, .. } => {
1557 matches!(&**then_stmt, Stmt::Continue(None))
1558 || matches!(&**then_stmt, Stmt::Block(tb) if tb.len() == 1 && matches!(tb[0], Stmt::Continue(None)))
1559 }
1560 _ => false,
1561 };
1562 if is_direct {
1563 if let Stmt::If { cond, .. } = stmts.pop().unwrap() {
1564 return Some((stmts, cond));
1565 }
1566 }
1567 // Descend into a trailing block.
1568 if matches!(stmts.last(), Some(Stmt::Block(_))) {
1569 let last = stmts.pop().unwrap();
1570 if let Some((inner, c)) = extract_trailing_continue_if(&last) {
1571 if !inner.is_empty() {
1572 stmts.push(Stmt::Block(inner));
1573 }
1574 return Some((stmts, c));
1575 } else {
1576 stmts.push(last);
1577 }
1578 }
1579 None
1580}
1581
1582/// Separate the first Basic block's statements from the rest of a body.
1583/// (Bodies built by `walk` start with the header's statements.)
1584#[allow(dead_code)]
1585fn split_first_block(body: Stmt) -> (Stmt, Option<usize>) {
1586 // The body is whatever the walk produced; we keep it intact — header
1587 // statements were already woven in by conv(Basic). Return as-is.
1588 (body, None)
1589}
1590
1591fn stmt_to_vec(s: Stmt) -> Vec<Stmt> {
1592 match s {
1593 Stmt::Block(v) => v,
1594 other => vec![other],
1595 }
1596}