synth_core/wasm_decoder.rs
1//! WASM Binary Decoder - Converts wasmparser operators to WasmOp sequences
2//!
3//! This module bridges the gap between parsed WASM binaries and any backend.
4//! It extracts function bodies and converts wasmparser operators to our internal WasmOp format.
5
6use crate::wasm_op::WasmOp;
7use anyhow::{Context, Result};
8use std::collections::HashMap;
9use wasmparser::{ExternalKind, Parser, Payload};
10
11/// Kind of a WASM import
12#[derive(Debug, Clone, PartialEq, Eq)]
13pub enum ImportKind {
14 /// Imported function with type index
15 Function(u32),
16 /// Imported memory
17 Memory,
18 /// Imported table
19 Table,
20 /// Imported global
21 Global,
22}
23
24/// A WASM import entry with full metadata
25#[derive(Debug, Clone)]
26pub struct ImportEntry {
27 /// Module name (e.g., "wasi:cli/stdout" or "env")
28 pub module: String,
29 /// Field name (e.g., "write" or "memory")
30 pub name: String,
31 /// Import kind and associated data
32 pub kind: ImportKind,
33 /// Index of this import within its kind (e.g., function import index)
34 pub index: u32,
35}
36
37/// WASM linear memory specification
38#[derive(Debug, Clone)]
39pub struct WasmMemory {
40 /// Memory index
41 pub index: u32,
42 /// Initial size in pages (64KB each)
43 pub initial_pages: u32,
44 /// Maximum size in pages (if specified)
45 pub max_pages: Option<u32>,
46 /// Whether memory is shared (requires threads proposal)
47 pub shared: bool,
48}
49
50/// A captured constant global initializer (#649). Only INTEGER `t.const` init
51/// exprs are captured: `f32.const`/`f64.const` inits deliberately decode to
52/// `None` — float-typed global ACCESS is the GI-FPU-001 (#369) loud-skip lane,
53/// and fabricating a bit-pattern here must not quietly unskip it. Non-const
54/// init exprs (e.g. `global.get` of an import) are not statically known and
55/// also decode to `None`.
56#[derive(Debug, Clone, Copy, PartialEq, Eq)]
57pub enum GlobalInit {
58 /// A leading `i32.const` initializer.
59 I32(i32),
60 /// A leading `i64.const` initializer — BOTH words must reach the emitted
61 /// global slot (#649: `init_i32`-shaped capture silently zeroed these).
62 I64(i64),
63}
64
65/// A WASM global's declaration — its initial value and mutability (#237).
66/// Needed so the native-pointer ABI can recognize a global whose initializer is
67/// a linear-memory address (e.g. `$__stack_pointer = 65536`) and make it
68/// `__synth_wasm_data`-relative, rather than reading it from an R9 globals table
69/// the self-contained drop-in object can't rely on.
70#[derive(Debug, Clone)]
71pub struct WasmGlobal {
72 /// Global index (defined globals; imported globals are not counted here).
73 pub index: u32,
74 /// The captured constant initializer (#237/#649): `i32.const` or
75 /// `i64.const`. Float/non-const init exprs decode to `None` — see
76 /// [`GlobalInit`].
77 pub init: Option<GlobalInit>,
78 /// Whether the global is mutable.
79 pub mutable: bool,
80 /// #643: byte width of the global's storage slot, from its declared value
81 /// type — 4 for i32/f32, 8 for i64/f64, 16 for v128. The globals table is
82 /// laid out by SUMMING these widths (not `index * 4`): an i64 global needs
83 /// room for both words, and every later global's offset shifts with it.
84 pub slot_bytes: u32,
85}
86
87impl WasmMemory {
88 /// Get initial size in bytes
89 pub fn initial_bytes(&self) -> u32 {
90 self.initial_pages * 65536
91 }
92
93 /// Get maximum size in bytes (or initial if not specified)
94 pub fn max_bytes(&self) -> u32 {
95 self.max_pages.unwrap_or(self.initial_pages) * 65536
96 }
97}
98
99/// #642: one element segment's statically-decoded shape — see
100/// [`DecodedModule::elem_segments`].
101#[derive(Debug, Clone, PartialEq, Eq)]
102pub struct ElemSegmentInfo {
103 /// #650: the table this ACTIVE segment initializes (0 for the pre-#650
104 /// single-table form). Meaningless when `offset` is `None`.
105 pub table_index: u32,
106 /// Const i32 offset of an ACTIVE segment into its table; `None` =
107 /// placement not statically verifiable (passive/declared segment or a
108 /// non-const offset expression).
109 pub offset: Option<u32>,
110 /// The segment's function indices in slot order; `None` = contents not
111 /// statically verifiable (an entry was not a plain `ref.func`).
112 pub funcs: Option<Vec<u32>>,
113}
114
115/// #642/#650: one table's `call_indirect` guard inputs — see
116/// [`CallIndirectGuards`] for the layout contract and soundness argument.
117#[derive(Debug, Clone, Default, PartialEq, Eq)]
118pub struct TableGuards {
119 /// Compile-time size of this table (entries); `None` = no sound bound
120 /// known (an imported table with growable limits).
121 pub table_size: Option<u32>,
122 /// #650: byte offset of this table's base within the contiguous R11
123 /// region — `sum(size(0..N)) * 4`, a compile-time constant. `None` when
124 /// any PRECEDING table's size is unknown (the base is then not a
125 /// compile-time constant and the lowering declines).
126 pub base_byte_offset: Option<u32>,
127 /// Per expected-type index: `None` = closed-world type property VERIFIED
128 /// against THIS table; `Some(reason)` = not verifiable (the lowering
129 /// declines).
130 pub type_reject: Vec<Option<String>>,
131 /// #664: whether this table's image contains at least one uninitialized
132 /// (null funcref) slot. WASM Core §4.4.8 requires a `call_indirect`
133 /// reaching a null slot to TRAP — the closed-world type check verifies
134 /// the INITIALIZED slots only, and the lowering must emit a runtime
135 /// null check (pointer == 0 → trap) before the indirect branch when
136 /// this is set. `false` for a fully-initialized table keeps today's
137 /// exact dispatch bytes (no null check) BY CONSTRUCTION. Only
138 /// meaningful when the type verdict is `None` (verified); reject paths
139 /// decline before it is consulted.
140 pub has_null_slots: bool,
141 /// #676: this table's image is statically known but HETEROGENEOUS — its
142 /// initialized slots span at least two distinct STRUCTURAL signature
143 /// classes, so no expected type's closed world can hold
144 /// (`type_reject[t]` is `Some` for every `t`) — yet the mismatch trap
145 /// (WASM Core §4.4.8) IS dischargeable at runtime: the type-id sidecar
146 /// (see [`CallIndirectGuards`]) carries each slot's structural class id,
147 /// and the dispatch compares the indexed slot's id against the expected
148 /// type's class id (a compile-time immediate), trapping on inequality.
149 /// When set (and [`CallIndirectGuards::type_ids_byte_offset`] is known),
150 /// the lowering emits that runtime check INSTEAD of declining. `false`
151 /// keeps the pre-#676 behavior: verified tables dispatch unchecked
152 /// (byte-identical), unverifiable tables decline.
153 pub runtime_type_check: bool,
154}
155
156/// #642/#650: everything the `call_indirect` lowering needs to emit its
157/// guards — computed once per module by
158/// [`DecodedModule::call_indirect_guards`] and threaded to the instruction
159/// selector via `CompileConfig`.
160///
161/// ## The R11 multi-table layout contract (#650)
162///
163/// The runtime/harness links every funcref table as ONE contiguous region of
164/// raw 4-byte code pointers based at R11, in declaration order (imported
165/// tables first): table 0 at `R11 + 0`, table N at
166/// `R11 + sum(size(0..N)) * 4`. The offsets are compile-time constants
167/// because tables are provably fixed-size (`table.grow`/`table.set` are
168/// unsupported ops whose functions loud-skip at decode — #642). A
169/// single-table module degenerates to the pre-#650 contract (table 0 at
170/// R11, offset 0) BY CONSTRUCTION, keeping its emitted bytes identical.
171///
172/// WASM Core §4.4.8 requires `call_indirect` to trap when `index >=
173/// table.size` and when the callee's type does not match the instruction's
174/// expected type. The region stores no size fields and no type ids, so, per
175/// table:
176/// - the BOUNDS check is emitted at runtime against THAT table's
177/// compile-time `table_size` immediate (sound: fixed-size, see above), and
178/// - the TYPE check is discharged at COMPILE time: for expected type `t`,
179/// `tables[n].type_reject[t]` is `None` only when every INITIALIZED slot
180/// of table `n` verifiably holds a function whose signature structurally
181/// equals type `t` (the closed-world property — no runtime mismatch is
182/// then possible). Otherwise it holds the reason, and the lowering
183/// declines LOUDLY rather than emit an unchecked indirect branch, and
184/// - a NULL (uninitialized) slot traps at RUNTIME (#664): the layout
185/// contract requires the runtime/harness to link every uninitialized
186/// slot as a ZERO word (null funcref has no code address; 0 is never a
187/// valid function pointer in the region), and when `has_null_slots` is
188/// set the dispatch emits a null check on the loaded pointer
189/// (`CMP #0` → trap) between the bounds guard and the indirect branch.
190/// A fully-initialized table (`has_null_slots == false`) keeps the
191/// pre-#664 dispatch bytes identical BY CONSTRUCTION, and
192/// - a HETEROGENEOUS table (mixed signatures — the closed-world property
193/// cannot hold for ANY expected type) is dispatched through a runtime
194/// type check against the **type-id sidecar** (#676): a parallel `u32`
195/// array the layout contract places at `R11 + type_ids_byte_offset`
196/// (immediately after the LAST table's pointer words, i.e. at
197/// `sum(size(0..num_tables)) * 4`), mirroring the pointer region slot
198/// for slot — table N's type-ids start at
199/// `R11 + type_ids_byte_offset + base_byte_offset(N)`. Each word is the
200/// slot's STRUCTURAL signature class id: structurally-equal function
201/// types share one dense id (1-based, first-occurrence order over the
202/// type section); id **0 is reserved for null slots**, so the type
203/// compare (expected ids are always >= 1) subsumes the #664 null trap
204/// in the same `CMP`. The dispatch loads `type_id[idx]`, compares it
205/// against the expected type's class id (compile-time immediate) and
206/// traps (`UDF`) on mismatch — WASM Core §4.4.8's runtime type check —
207/// before the pointer load and `BLX`. The sidecar words are emitted
208/// into the relocatable object as the `.synth.table_type_ids` section
209/// (non-ALLOC metadata, like `.meld_import_table`): the runtime/harness
210/// that links the pointer region copies them to
211/// `R11 + type_ids_byte_offset` verbatim — it never re-derives ids. A
212/// module with NO heterogeneous table emits no sidecar and no runtime
213/// type check anywhere: homogeneous dispatch bytes stay identical BY
214/// CONSTRUCTION (the #650 offset-0 / #664 `null_check: false` trick).
215///
216/// pre-#664 dispatch bytes identical BY CONSTRUCTION.
217///
218/// ## Companion: the self-contained SRAM layout contract (#687)
219///
220/// The R11 register above is ALSO the linear-memory base register, whose
221/// placement inside SRAM is governed by the self-contained image's stack
222/// layout: `--stack-layout=high` (default) keeps linmem at the SRAM start
223/// with the stack growing down from the top; `--stack-layout=low` reserves
224/// the stack at the SRAM BOTTOM and shifts linmem/globals (and the optimized
225/// path's absolute `0x2000_0100` base) up by the stack size, so an overflow
226/// BusFaults below SRAM instead of silently corrupting them. The full layout
227/// tables live on `build_multi_func_cortex_m_elf` in `synth-cli` (the builder
228/// that owns the addresses). Relocatable/host-linked objects are NOT covered
229/// — their linker script owns the layout, and the flag is refused there.
230#[derive(Debug, Clone, Default, PartialEq, Eq)]
231pub struct CallIndirectGuards {
232 /// Per-table guard inputs, indexed by table index (imports first). The
233 /// default (empty — no module context) DECLINES every `call_indirect`.
234 pub tables: Vec<TableGuards>,
235 /// #676: byte offset of the type-id sidecar within the R11 region — the
236 /// total pointer-region size, `sum(size(0..num_tables)) * 4`. `Some`
237 /// only when a sidecar exists: at least one table is heterogeneous
238 /// (see [`TableGuards::runtime_type_check`]) AND every table's size is
239 /// compile-time known (otherwise the sidecar base is not a constant and
240 /// heterogeneous dispatches keep declining). `None` = no sidecar.
241 pub type_ids_byte_offset: Option<u32>,
242 /// #676: the sidecar image — one `u32` structural class id per slot
243 /// across ALL tables in region order (0 = null slot). Emitted into the
244 /// object as `.synth.table_type_ids`; empty exactly when
245 /// `type_ids_byte_offset` is `None`. A table whose image is not
246 /// statically known contributes ZERO words (it declines at the
247 /// lowering, and 0 never equals an expected class id, so even a rogue
248 /// dispatch would trap, not branch).
249 pub type_ids_image: Vec<u32>,
250 /// #676: per module type index, that type's structural class id
251 /// (1-based, dense; structurally-equal duplicate types share an id).
252 /// The expected-type immediate the dispatch compares against. Empty
253 /// when `type_ids_byte_offset` is `None` (no sidecar — never consulted).
254 pub type_class_ids: Vec<u32>,
255}
256
257impl CallIndirectGuards {
258 /// Single-table (table 0 at R11 offset 0) guards — the pre-#650 shape,
259 /// used by tests and single-table call sites.
260 pub fn single_table(table_size: Option<u32>, type_reject: Vec<Option<String>>) -> Self {
261 Self {
262 tables: vec![TableGuards {
263 table_size,
264 base_byte_offset: Some(0),
265 type_reject,
266 has_null_slots: false,
267 runtime_type_check: false,
268 }],
269 ..Self::default()
270 }
271 }
272}
273
274/// Decoded WASM module with functions and memory
275#[derive(Debug, Clone)]
276pub struct DecodedModule {
277 /// Decoded functions
278 pub functions: Vec<FunctionOps>,
279 /// Linear memories
280 pub memories: Vec<WasmMemory>,
281 /// Data segments (offset, data) for memory initialization.
282 ///
283 /// MEMORY 0 ONLY — the legacy single-memory field every existing consumer
284 /// reads; its shape and contents are unchanged by multi-memory (#406).
285 /// Segments targeting memory > 0 live in
286 /// [`Self::extra_memory_data_segments`].
287 pub data_segments: Vec<(u32, Vec<u8>)>,
288 /// VCR-MEM-002 phase 1 (#406): active const-offset data segments on
289 /// NON-DEFAULT memories, as `(memory_index, offset, bytes)` with
290 /// `memory_index > 0`. Previously these were silently dropped (memory k
291 /// shipped uninitialized while its loads compiled). Declaration order.
292 pub extra_memory_data_segments: Vec<(u32, u32, Vec<u8>)>,
293 /// VCR-MEM-002 phase 1 (#406): `Some(reason)` when the module contains a
294 /// multi-memory shape decode cannot lower (e.g. an active data segment on
295 /// memory > 0 with a non-constant offset). The multi-memory compile path
296 /// must decline LOUDLY with this reason; single-memory modules never set
297 /// it.
298 pub multi_memory_decline: Option<String>,
299 /// Import entries (module name, field name, kind)
300 pub imports: Vec<ImportEntry>,
301 /// Number of imported functions (for distinguishing import calls from local calls)
302 pub num_imported_funcs: u32,
303 /// AAPCS integer-argument count per function, indexed by the *full* WASM
304 /// function index (imported functions first, then locally-defined ones).
305 /// Used by the backend to marshal call arguments into R0–R3 (issue #195).
306 /// Counts every parameter as one slot (i64/f64 over-counted — see the
307 /// backend's `set_func_arg_counts` scope note).
308 pub func_arg_counts: Vec<u32>,
309 /// AAPCS integer-argument count per *function type*, indexed by type index.
310 /// Used by `call_indirect`, whose callee arg count comes from the static
311 /// type index (issue #195).
312 pub type_arg_counts: Vec<u32>,
313 /// #311: whether each *function* (full index, imports first) returns i64 —
314 /// the call lowering must tag the result as a register PAIR (r0:r1) or the
315 /// hi half is invisible to liveness and the next constant clobbers it.
316 pub func_ret_i64: Vec<bool>,
317 /// #311: whether each *function type* returns i64 (for `call_indirect`).
318 pub type_ret_i64: Vec<bool>,
319 /// #359: declared parameter widths per *function* (full index, imports
320 /// first): `func_params_i64[f][k]` is true when param `k` is i64/f64. The
321 /// AAPCS stack-argument path needs the declared widths — op-stream inference
322 /// can't see an unused i64 param that still shifts the incoming-stack layout.
323 pub func_params_i64: Vec<Vec<bool>>,
324 /// GI-FPU-002 (#619/#369): declared f32-param mask per *function* (full
325 /// index, imports first): `func_params_f32[f][k]` is true when param `k` is
326 /// f32. The direct selector homes hard-float f32 args in S0..S15 (AAPCS-VFP),
327 /// which op-stream inference cannot recover for a pure-passthrough f32 param.
328 pub func_params_f32: Vec<Vec<bool>>,
329 /// GI-FPU-002 phase 2 (#369): declared f64-param mask per *function*
330 /// (full index, imports first): `func_params_f64[f][k]` is true when param
331 /// `k` is f64. Hard-float targets decline such functions loudly (the
332 /// legacy width inference treats the param as an i64 CORE pair — wrong
333 /// registers under AAPCS-VFP). Distinct from `func_params_i64`, which
334 /// deliberately lumps i64 and f64 for frame-layout purposes.
335 pub func_params_f64: Vec<Vec<bool>>,
336 /// GI-FPU-002 phase 2 (#719/#369): whether each *function* (full index,
337 /// imports first) returns f32. The direct selector's epilogue homes an f32
338 /// result in S0 (AAPCS-VFP); when the result value transited a core register
339 /// (e.g. it came from a call that returned f32 as an integer-tagged R0), the
340 /// epilogue must loudly decline rather than emit the integer R0 return (a
341 /// silent miscompile — the caller reads S0). Op-stream inference cannot see a
342 /// pure-passthrough f32 return, so it is carried from the declared signature.
343 pub func_ret_f32: Vec<bool>,
344 /// GI-FPU-002 phase 2 (#719/#369): whether each *function* returns f64 (D0
345 /// under AAPCS-VFP). Same epilogue-soundness role as `func_ret_f32`.
346 pub func_ret_f64: Vec<bool>,
347 /// GI-FPU-002 phase 2 (#719/#369): whether each *function type* returns
348 /// f32 / f64 — the `call_indirect` analogue of `func_ret_f32`/`func_ret_f64`
349 /// (the selector loudly declines an indirect call whose static type returns
350 /// a float this increment does not marshal, rather than tag S0/D0 as R0).
351 pub type_ret_f32: Vec<bool>,
352 /// See [`Self::type_ret_f32`].
353 pub type_ret_f64: Vec<bool>,
354 /// Defined globals with their initializers (#237). Empty if the module has
355 /// no global section. Used by the native-pointer ABI to make a global whose
356 /// initializer is a linear-memory address (e.g. `$__stack_pointer`)
357 /// self-contained rather than table-relative.
358 pub globals: Vec<WasmGlobal>,
359 /// Function indices that populate any table via an element segment (#275).
360 /// These are the possible `call_indirect` targets — a function reached only
361 /// through the table is invisible to direct-`call` reachability, so the
362 /// whole-graph closure must treat every table entry as reachable once any
363 /// reachable function performs a `call_indirect`. Empty for modules with no
364 /// element section (every leaf/direct-call module), keeping output identical.
365 pub elem_func_indices: Vec<u32>,
366 /// #642: compile-time size (in entries) of table 0 — `table_sizes[0]`,
367 /// kept as a convenience accessor. See [`Self::table_sizes`].
368 pub table_size: Option<u32>,
369 /// #650: compile-time size (in entries) per table, indexed by table index
370 /// (imported tables first, then the table section, in declaration order).
371 /// A DEFINED table's size is exact: `table.grow`/`table.set` are
372 /// unsupported ops (their functions loud-skip at decode), so nothing
373 /// synth compiles can resize or retype a table. An imported table only
374 /// yields a sound bound when its limits pin the size (`max == initial`);
375 /// otherwise its entry is `None` and the `call_indirect` lowering
376 /// declines (for that table AND for any later table, whose base offset
377 /// within the contiguous R11 region is then unknown).
378 pub table_sizes: Vec<Option<u32>>,
379 /// #642: per element segment, everything the closed-world `call_indirect`
380 /// type check needs. `offset` is the const i32 placement of an ACTIVE
381 /// segment into table `table_index` (`None` = passive/declared/non-const
382 /// offset — statically unverifiable placement); `funcs` are the segment's
383 /// function indices in slot order (`None` = an entry was not a plain
384 /// `ref.func`, e.g. `ref.null` — statically unverifiable contents).
385 pub elem_segments: Vec<ElemSegmentInfo>,
386 /// #642: type index per function, indexed by the FULL function index
387 /// (imports first, then locally-defined ones).
388 pub func_type_indices: Vec<u32>,
389 /// #642: canonical structural signature per type index (params/results
390 /// rendered as a string) — used for the closed-world `call_indirect` type
391 /// check, which must compare SIGNATURES, not raw type indices (a module
392 /// may carry structurally-identical duplicate types).
393 pub type_signatures: Vec<String>,
394 /// VCR-PERF-002 Phase 1 (#494): proven invariants from loom's `wsc.facts`
395 /// custom section, keyed by `(function index, value id)` — see
396 /// `docs/design/wsc-facts-encoding.md` (schema v1) and
397 /// [`crate::wsc_facts::parse_wsc_facts`]. FAIL-SAFE by contract (loom#231
398 /// Q4): a missing/unparseable section or unknown version yields the empty
399 /// vec, unknown fact kinds are skipped — never a decode error. Phase 1 is
400 /// ingestion only: NO codegen path consumes these yet, so emitted bytes
401 /// are unchanged whether or not a module carries the section.
402 pub wsc_facts: Vec<crate::wsc_facts::WscFact>,
403}
404
405impl DecodedModule {
406 /// #642/#650: compute the `call_indirect` guard inputs — per table, the
407 /// compile-time size for the runtime bounds check, the base byte offset
408 /// within the contiguous R11 region, and the per-expected-type
409 /// closed-world verdict that discharges the type check at compile time.
410 /// See [`CallIndirectGuards`] for the layout contract and soundness
411 /// argument.
412 pub fn call_indirect_guards(&self) -> CallIndirectGuards {
413 let n_types = self.type_signatures.len();
414
415 // A segment whose PLACEMENT is not statically attributable
416 // (passive/declared segment, non-const offset, or a table index the
417 // module does not declare) poisons EVERY table: `table.init` (itself
418 // an unsupported op) or a computed offset could land its entries
419 // anywhere, so no table's image is verifiable.
420 let global_poison: Option<&'static str> = self
421 .elem_segments
422 .iter()
423 .any(|seg| seg.offset.is_none() || seg.table_index as usize >= self.table_sizes.len())
424 .then_some(
425 "element segment is not statically verifiable (passive/declared \
426 segment, non-const offset, out-of-range table, or non-`ref.func` \
427 entry)",
428 );
429
430 // #676: structural signature classes — structurally-equal types share
431 // one dense 1-based id (first-occurrence order over the type section);
432 // id 0 is reserved for null slots. These feed the type-id sidecar and
433 // the expected-type compare immediate of the runtime type check.
434 let mut class_of_sig: std::collections::HashMap<&str, u32> =
435 std::collections::HashMap::new();
436 let mut type_class_ids: Vec<u32> = Vec::with_capacity(n_types);
437 for sig in &self.type_signatures {
438 let next = class_of_sig.len() as u32 + 1;
439 type_class_ids.push(*class_of_sig.entry(sig.as_str()).or_insert(next));
440 }
441
442 let mut tables = Vec::with_capacity(self.table_sizes.len());
443 // #676: per table, the slot class ids (None = image not statically
444 // known) — concatenated into the sidecar image below.
445 let mut per_table_slot_ids: Vec<Option<Vec<u32>>> =
446 Vec::with_capacity(self.table_sizes.len());
447 // Running word offset of the next table's base within the R11 region;
448 // `None` once a table of unknown size is passed (every later base is
449 // then not a compile-time constant).
450 let mut base_words: Option<u32> = Some(0);
451 for (n, &size) in self.table_sizes.iter().enumerate() {
452 let base_byte_offset = base_words.and_then(|w| w.checked_mul(4));
453 let (type_reject, has_null_slots, slot_class_ids) =
454 self.table_type_reject(n as u32, size, global_poison, n_types, &type_class_ids);
455 // #676: heterogeneous = the image is statically known and its
456 // INITIALIZED slots span >= 2 distinct structural classes (null
457 // slots — id 0 — don't count; a sparse homogeneous table stays
458 // on the #664 verified-plus-null-check path, bytes identical).
459 let runtime_type_check = slot_class_ids.as_ref().is_some_and(|ids| {
460 let mut distinct: Vec<u32> = ids.iter().copied().filter(|&c| c != 0).collect();
461 distinct.sort_unstable();
462 distinct.dedup();
463 distinct.len() >= 2
464 });
465 per_table_slot_ids.push(slot_class_ids);
466 tables.push(TableGuards {
467 table_size: size,
468 base_byte_offset,
469 type_reject,
470 has_null_slots,
471 runtime_type_check,
472 });
473 base_words = match (base_words, size) {
474 (Some(w), Some(s)) => w.checked_add(s),
475 _ => None,
476 };
477 }
478
479 // #676: the sidecar exists only when some table actually needs the
480 // runtime check AND the whole pointer region's size is compile-time
481 // known (`base_words` survived every table) — otherwise the sidecar
482 // base is not a constant and heterogeneous dispatches keep declining
483 // (their `runtime_type_check` flag is cleared so the lowering sees a
484 // plain reject).
485 let any_hetero = tables.iter().any(|t| t.runtime_type_check);
486 let type_ids_byte_offset = base_words
487 .filter(|_| any_hetero)
488 .and_then(|w| w.checked_mul(4));
489 let type_ids_image = if type_ids_byte_offset.is_some() {
490 self.table_sizes
491 .iter()
492 .zip(&per_table_slot_ids)
493 .flat_map(|(&size, ids)| match ids {
494 Some(ids) => ids.clone(),
495 // Image not statically known: zero words (id 0 never
496 // matches an expected class id >= 1 — trap, not branch).
497 None => vec![0u32; size.unwrap_or(0) as usize],
498 })
499 .collect()
500 } else {
501 for t in &mut tables {
502 t.runtime_type_check = false;
503 }
504 Vec::new()
505 };
506 CallIndirectGuards {
507 tables,
508 type_ids_byte_offset,
509 type_ids_image,
510 type_class_ids: if type_ids_byte_offset.is_some() {
511 type_class_ids
512 } else {
513 Vec::new()
514 },
515 }
516 }
517
518 /// #642/#650: the closed-world type verdicts for ONE table — `None` per
519 /// expected type when every INITIALIZED slot of table `n` verifiably
520 /// holds a function of that exact structural signature; `Some(reason)`
521 /// otherwise. The second component is `has_null_slots` (#664): whether
522 /// the table image left any slot uninitialized — a `call_indirect`
523 /// reaching one must TRAP at runtime (null check on the loaded pointer),
524 /// which the lowering emits only when this is set. Reject paths return
525 /// `false` (the verdict declines before the flag is consulted). The
526 /// third component (#676) is the table's slot class ids — per slot, the
527 /// structural signature class of the initializing function (0 for a
528 /// null slot) — `Some` exactly when the table image is statically
529 /// known; it feeds the type-id sidecar and the heterogeneity verdict.
530 fn table_type_reject(
531 &self,
532 n: u32,
533 size: Option<u32>,
534 global_poison: Option<&str>,
535 n_types: usize,
536 type_class_ids: &[u32],
537 ) -> (Vec<Option<String>>, bool, Option<Vec<u32>>) {
538 let reject_all = |reason: String| (vec![Some(reason); n_types], false, None);
539
540 if let Some(reason) = global_poison {
541 return reject_all(reason.to_string());
542 }
543 let Some(size) = size else {
544 return reject_all(format!(
545 "table {n} has no compile-time-fixed size (imported table with \
546 growable limits)"
547 ));
548 };
549
550 // Reconstruct the table image: slot -> initializing function index.
551 let mut slots: Vec<Option<u32>> = vec![None; size as usize];
552 for seg in self.elem_segments.iter().filter(|s| s.table_index == n) {
553 let (Some(off), Some(funcs)) = (seg.offset, seg.funcs.as_ref()) else {
554 // Placement is known (global_poison ruled `offset: None` out),
555 // so this is an unverifiable CONTENTS case — it poisons only
556 // the table it targets.
557 return reject_all(format!(
558 "element segment targeting table {n} is not statically \
559 verifiable (non-`ref.func` entry)"
560 ));
561 };
562 for (k, &f) in funcs.iter().enumerate() {
563 let Some(slot) = slots.get_mut(off as usize + k) else {
564 return reject_all(format!(
565 "element segment (offset {off}, {} entries) writes past \
566 table {n}'s declared size {size}",
567 funcs.len()
568 ));
569 };
570 *slot = Some(f);
571 }
572 }
573 // #664: an uninitialized slot is a null funcref — calling it must
574 // trap (WASM Core §4.4.8). It no longer poisons the closed world
575 // (pre-#664 it rejected EVERY type): the layout contract requires
576 // null slots to be linked as ZERO words, so the lowering discharges
577 // the trap at RUNTIME with a null check on the loaded pointer. The
578 // type check below therefore covers the INITIALIZED slots only —
579 // a null slot can never produce a live callee of the wrong type,
580 // because the null check traps before the branch.
581 let has_null_slots = slots.iter().any(|s| s.is_none());
582
583 let rejects = (0..n_types)
584 .map(|t| {
585 for f in slots.iter().flatten() {
586 let Some(&fty) = self.func_type_indices.get(*f as usize) else {
587 return Some(format!(
588 "table {n} entry references function {f} with no known type"
589 ));
590 };
591 if self.type_signatures.get(fty as usize) != self.type_signatures.get(t) {
592 return Some(format!(
593 "table {n} entry (function {f}, type {fty}) has a different \
594 signature than expected type {t}"
595 ));
596 }
597 }
598 None
599 })
600 .collect();
601 // #676: per-slot structural class ids (0 = null). `None` as soon as
602 // any initializing function's type is unknown — the image is then
603 // not statically classifiable and the table can neither verify nor
604 // carry the runtime check (the rejects above already name it).
605 let slot_class_ids: Option<Vec<u32>> = slots
606 .iter()
607 .map(|s| match s {
608 None => Some(0u32),
609 Some(f) => self
610 .func_type_indices
611 .get(*f as usize)
612 .and_then(|&fty| type_class_ids.get(fty as usize).copied()),
613 })
614 .collect();
615 (rejects, has_null_slots, slot_class_ids)
616 }
617}
618
619/// Decode a WASM binary and extract functions, memory, and data segments
620pub fn decode_wasm_module(wasm_bytes: &[u8]) -> Result<DecodedModule> {
621 let mut functions = Vec::new();
622 let mut memories = Vec::new();
623 let mut data_segments = Vec::new();
624 // VCR-MEM-002 phase 1 (#406): (memory_index, offset, bytes) for active
625 // const-offset data segments on memory > 0, and the first decode-level
626 // reason multi-memory lowering must be declined (if any).
627 let mut extra_memory_data_segments: Vec<(u32, u32, Vec<u8>)> = Vec::new();
628 let mut multi_memory_decline: Option<String> = None;
629 let mut globals: Vec<WasmGlobal> = Vec::new();
630 let mut imports = Vec::new();
631 let mut func_index = 0u32;
632 let mut num_imported_funcs = 0u32;
633 let mut export_names: HashMap<u32, String> = HashMap::new();
634 // #195: per-type AAPCS arg count (indexed by type index) and per-function
635 // arg count (indexed by full function index: imports first, then locals).
636 let mut type_arg_counts: Vec<u32> = Vec::new();
637 let mut func_arg_counts: Vec<u32> = Vec::new();
638 let mut type_ret_i64: Vec<bool> = Vec::new();
639 let mut func_ret_i64: Vec<bool> = Vec::new();
640 // GI-FPU-002 phase 2 (#719/#369): per-type / per-function f32/f64 return
641 // flags, so the direct selector's epilogue can loudly decline an f32/f64
642 // result that reaches it in a core register (never a silent R0 return).
643 let mut type_ret_f32: Vec<bool> = Vec::new();
644 let mut func_ret_f32: Vec<bool> = Vec::new();
645 let mut type_ret_f64: Vec<bool> = Vec::new();
646 let mut func_ret_f64: Vec<bool> = Vec::new();
647 // #359: declared param widths per type / per function (full index).
648 let mut type_params_i64: Vec<Vec<bool>> = Vec::new();
649 let mut func_params_i64: Vec<Vec<bool>> = Vec::new();
650 // GI-FPU-002 (#619/#369): per-type / per-function declared f32-param mask,
651 // so the direct selector can home hard-float (AAPCS-VFP) f32 args in S0..S15
652 // instead of the core-register (R0..R3) integer path. Independent of
653 // `params_i64` (which lumps f64 with i64): an f32 param is neither.
654 let mut type_params_f32: Vec<Vec<bool>> = Vec::new();
655 let mut func_params_f32: Vec<Vec<bool>> = Vec::new();
656 // GI-FPU-002 phase 2 (#369): per-type / per-function f64-param mask —
657 // hard-float targets decline f64 params loudly (D-register homing is a
658 // later increment; the legacy i64-pair treatment reads wrong registers).
659 let mut type_params_f64: Vec<Vec<bool>> = Vec::new();
660 let mut func_params_f64: Vec<Vec<bool>> = Vec::new();
661 // #509: (param_count, result_count) per type index, for FuncType blocktypes.
662 let mut type_block_arity: Vec<(u8, u8)> = Vec::new();
663 let mut elem_func_indices: Vec<u32> = Vec::new();
664 // #642/#650: call_indirect guard inputs — per-table fixed sizes (imports
665 // first, then the table section, in declaration order), per-segment
666 // static shapes, per-function type index, per-type canonical signature.
667 let mut table_sizes: Vec<Option<u32>> = Vec::new();
668 let mut elem_segments: Vec<ElemSegmentInfo> = Vec::new();
669 let mut func_type_indices: Vec<u32> = Vec::new();
670 let mut type_signatures: Vec<String> = Vec::new();
671 // #394 Tier-1.x: function index → developer-facing name from the wasm
672 // `name` custom section (function-names subsection). Applied to
673 // `FunctionOps.debug_name` after the parse loop — the custom section
674 // conventionally trails the code section, so the entries are not yet
675 // available when each `CodeSectionEntry` is decoded.
676 let mut name_section_names: HashMap<u32, String> = HashMap::new();
677 // VCR-PERF-002 Phase 1 (#494): facts from loom's `wsc.facts` custom
678 // section. `None` until (and unless) the first such section is seen —
679 // duplicates are ignored (one prover, one section; encoding doc rule).
680 let mut wsc_facts: Option<Vec<crate::wsc_facts::WscFact>> = None;
681 // GI-FPU-001 (#369): f32/f64-typed globals in the FULL global index space
682 // (imports first, then defined). `global.get`/`global.set` decode fine
683 // (they are type-agnostic ops), but there is no float lowering: the
684 // f32.const/f64.const initializer is silently dropped (`init_i32: None`
685 // → slot zeroed), so a read returns 0.0 instead of the init — a silent
686 // wrong value. Functions touching a float global loud-skip instead.
687 let mut num_imported_globals = 0u32;
688 let mut float_globals: std::collections::HashSet<u32> = std::collections::HashSet::new();
689 // #680: v128-typed globals (same index space) — a SIMD access has no
690 // lowering on any target, so touching one must loud-skip the function.
691 let mut v128_globals: std::collections::HashSet<u32> = std::collections::HashSet::new();
692 // #680: per-type "params/results contain v128" and its per-defined-function
693 // projection — a v128 param/result is expressible with ZERO SIMD-proposal
694 // operators in the body (`local.get 0` passthrough), so the operator-level
695 // catch alone would miss it.
696 let mut type_has_v128: Vec<bool> = Vec::new();
697 let mut func_sig_has_v128: Vec<bool> = Vec::new();
698
699 for payload in Parser::new(0).parse_all(wasm_bytes) {
700 let payload = payload.context("Failed to parse WASM payload")?;
701
702 match payload {
703 Payload::TypeSection(reader) => {
704 // Record the parameter count of each function type so calls can
705 // marshal the right number of arguments (issue #195).
706 for rec_group in reader {
707 let rec_group = rec_group.context("Failed to parse type")?;
708 for sub_ty in rec_group.types() {
709 // #509: blocktype arity per type index (saturated u8 —
710 // >255 params/results is far beyond anything the
711 // selector supports anyway, and the selector declines
712 // rather than trusting a saturated count).
713 type_block_arity.push(match &sub_ty.composite_type.inner {
714 wasmparser::CompositeInnerType::Func(f) => (
715 u8::try_from(f.params().len()).unwrap_or(u8::MAX),
716 u8::try_from(f.results().len()).unwrap_or(u8::MAX),
717 ),
718 _ => (u8::MAX, u8::MAX),
719 });
720 let (count, ret_i64, params_i64) = match &sub_ty.composite_type.inner {
721 wasmparser::CompositeInnerType::Func(func_ty) => (
722 func_ty.params().len() as u32,
723 func_ty
724 .results()
725 .first()
726 .is_some_and(|t| *t == wasmparser::ValType::I64),
727 // #359: i64/f64 params occupy 8 bytes / a register
728 // pair under AAPCS. f32/f64 are not in scope for the
729 // stack-arg path (refused), but mark both 64-bit
730 // float and i64 so the guard catches them.
731 func_ty
732 .params()
733 .iter()
734 .map(|t| {
735 matches!(
736 t,
737 wasmparser::ValType::I64 | wasmparser::ValType::F64
738 )
739 })
740 .collect::<Vec<bool>>(),
741 ),
742 _ => (0, false, Vec::new()),
743 };
744 // GI-FPU-002: declared f32-param mask for this type.
745 let params_f32 = match &sub_ty.composite_type.inner {
746 wasmparser::CompositeInnerType::Func(func_ty) => func_ty
747 .params()
748 .iter()
749 .map(|t| matches!(t, wasmparser::ValType::F32))
750 .collect::<Vec<bool>>(),
751 _ => Vec::new(),
752 };
753 // GI-FPU-002 phase 2: declared f64-param mask.
754 let params_f64 = match &sub_ty.composite_type.inner {
755 wasmparser::CompositeInnerType::Func(func_ty) => func_ty
756 .params()
757 .iter()
758 .map(|t| matches!(t, wasmparser::ValType::F64))
759 .collect::<Vec<bool>>(),
760 _ => Vec::new(),
761 };
762 // GI-FPU-002 phase 2: f32/f64 return flags for this type.
763 let (ret_f32, ret_f64) = match &sub_ty.composite_type.inner {
764 wasmparser::CompositeInnerType::Func(func_ty) => (
765 func_ty
766 .results()
767 .first()
768 .is_some_and(|t| *t == wasmparser::ValType::F32),
769 func_ty
770 .results()
771 .first()
772 .is_some_and(|t| *t == wasmparser::ValType::F64),
773 ),
774 _ => (false, false),
775 };
776 type_arg_counts.push(count);
777 type_ret_i64.push(ret_i64);
778 type_ret_f32.push(ret_f32);
779 type_ret_f64.push(ret_f64);
780 type_params_i64.push(params_i64);
781 type_params_f32.push(params_f32);
782 type_params_f64.push(params_f64);
783 // #680: v128 anywhere in the signature.
784 type_has_v128.push(match &sub_ty.composite_type.inner {
785 wasmparser::CompositeInnerType::Func(f) => f
786 .params()
787 .iter()
788 .chain(f.results())
789 .any(|t| *t == wasmparser::ValType::V128),
790 _ => false,
791 });
792 // #642: canonical structural signature for the
793 // closed-world call_indirect type check (compares
794 // SIGNATURES so duplicate types stay interchangeable).
795 type_signatures.push(match &sub_ty.composite_type.inner {
796 wasmparser::CompositeInnerType::Func(f) => {
797 format!("{:?}->{:?}", f.params(), f.results())
798 }
799 other => format!("non-func:{other:?}"),
800 });
801 }
802 }
803 }
804 Payload::ImportSection(reader) => {
805 // wasmparser 0.221+ groups imports (the "compact imports"
806 // proposal): the section reader yields `Imports` groups, each of
807 // which may expand to several `Import`s. `into_imports()`
808 // flattens groups back to individual `Import`s (preserving the
809 // module/name/ty fields), keeping the per-import loop intact.
810 for import in reader.into_imports() {
811 let import = import.context("Failed to parse import")?;
812 let (kind, idx) = match import.ty {
813 wasmparser::TypeRef::Func(type_idx) => {
814 let idx = num_imported_funcs;
815 num_imported_funcs += 1;
816 // Record the imported function's arg count at its
817 // full function index (imports come first).
818 func_type_indices.push(type_idx); // #642
819 func_arg_counts
820 .push(type_arg_counts.get(type_idx as usize).copied().unwrap_or(0));
821 func_ret_i64.push(
822 type_ret_i64
823 .get(type_idx as usize)
824 .copied()
825 .unwrap_or(false),
826 );
827 func_ret_f32.push(
828 type_ret_f32
829 .get(type_idx as usize)
830 .copied()
831 .unwrap_or(false),
832 );
833 func_ret_f64.push(
834 type_ret_f64
835 .get(type_idx as usize)
836 .copied()
837 .unwrap_or(false),
838 );
839 func_params_i64.push(
840 type_params_i64
841 .get(type_idx as usize)
842 .cloned()
843 .unwrap_or_default(),
844 );
845 func_params_f32.push(
846 type_params_f32
847 .get(type_idx as usize)
848 .cloned()
849 .unwrap_or_default(),
850 );
851 func_params_f64.push(
852 type_params_f64
853 .get(type_idx as usize)
854 .cloned()
855 .unwrap_or_default(),
856 );
857 (ImportKind::Function(type_idx), idx)
858 }
859 wasmparser::TypeRef::Memory(_) => (ImportKind::Memory, 0),
860 wasmparser::TypeRef::Table(t) => {
861 // #642: an imported table only yields a SOUND
862 // compile-time bound when its limits pin the size
863 // exactly (max == initial) — a growable import
864 // could be larger at runtime, and a bounds guard
865 // against `initial` would trap spec-valid calls.
866 // #650: imported tables take the leading table
867 // indices, in declaration order.
868 table_sizes.push(match (u32::try_from(t.initial), t.maximum) {
869 (Ok(init), Some(max)) if u64::from(init) == max => Some(init),
870 _ => None,
871 });
872 (ImportKind::Table, 0)
873 }
874 wasmparser::TypeRef::Global(g) => {
875 // GI-FPU-001 (#369): imported globals come first in
876 // the global index space — record float-typed ones
877 // so accesses loud-skip their function.
878 if matches!(
879 g.content_type,
880 wasmparser::ValType::F32 | wasmparser::ValType::F64
881 ) {
882 float_globals.insert(num_imported_globals);
883 }
884 // #680: v128-typed imported globals — same lane.
885 if g.content_type == wasmparser::ValType::V128 {
886 v128_globals.insert(num_imported_globals);
887 }
888 num_imported_globals += 1;
889 (ImportKind::Global, 0)
890 }
891 _ => continue,
892 };
893 imports.push(ImportEntry {
894 module: import.module.to_string(),
895 name: import.name.to_string(),
896 kind,
897 index: idx,
898 });
899 }
900 }
901 Payload::FunctionSection(reader) => {
902 // Each entry gives the type index of a locally-defined function,
903 // in order. Their full function indices follow the imports, so
904 // appending to `func_arg_counts` keeps it indexed by full index
905 // (issue #195).
906 for ty in reader {
907 let type_idx = ty.context("Failed to parse function type index")?;
908 func_type_indices.push(type_idx); // #642
909 func_arg_counts
910 .push(type_arg_counts.get(type_idx as usize).copied().unwrap_or(0));
911 func_ret_i64.push(
912 type_ret_i64
913 .get(type_idx as usize)
914 .copied()
915 .unwrap_or(false),
916 );
917 func_ret_f32.push(
918 type_ret_f32
919 .get(type_idx as usize)
920 .copied()
921 .unwrap_or(false),
922 );
923 func_ret_f64.push(
924 type_ret_f64
925 .get(type_idx as usize)
926 .copied()
927 .unwrap_or(false),
928 );
929 func_params_i64.push(
930 type_params_i64
931 .get(type_idx as usize)
932 .cloned()
933 .unwrap_or_default(),
934 );
935 func_params_f32.push(
936 type_params_f32
937 .get(type_idx as usize)
938 .cloned()
939 .unwrap_or_default(),
940 );
941 func_params_f64.push(
942 type_params_f64
943 .get(type_idx as usize)
944 .cloned()
945 .unwrap_or_default(),
946 );
947 // #680: defined-function order matches code-entry order.
948 func_sig_has_v128.push(
949 type_has_v128
950 .get(type_idx as usize)
951 .copied()
952 .unwrap_or(false),
953 );
954 }
955 }
956 Payload::TableSection(reader) => {
957 // #642: a DEFINED table's compile-time size is exact — its
958 // initial size is its permanent size, because nothing synth
959 // compiles can resize it (`table.grow` is an unsupported op
960 // whose function loud-skips at decode). #650: EVERY table is
961 // recorded — the contiguous R11 region places table N at
962 // byte offset `sum(size(0..N)) * 4`.
963 for table in reader {
964 let table = table.context("Failed to parse table")?;
965 table_sizes.push(u32::try_from(table.ty.initial).ok());
966 }
967 }
968 Payload::MemorySection(reader) => {
969 for (idx, memory) in reader.into_iter().enumerate() {
970 let mem = memory.context("Failed to parse memory")?;
971 memories.push(WasmMemory {
972 index: idx as u32,
973 initial_pages: mem.initial as u32,
974 max_pages: mem.maximum.map(|m| m as u32),
975 shared: mem.shared,
976 });
977 }
978 }
979 Payload::GlobalSection(reader) => {
980 // #237/#649: capture each defined global's constant initializer
981 // + mutability. The init is a const expr; we decode a leading
982 // `i32.const` (the `$__stack_pointer`/data-layout shape) or
983 // `i64.const` (#649: capturing only i32 silently ZEROED every
984 // nonzero i64 init). f32/f64 inits stay `None` on purpose —
985 // float global access is the GI-FPU-001 (#369) loud-skip lane —
986 // as do non-const init exprs (`global.get` of an import).
987 for (idx, global) in reader.into_iter().enumerate() {
988 let global = global.context("Failed to parse global")?;
989 let mut ops = global.init_expr.get_operators_reader();
990 let init = match ops.read() {
991 Ok(wasmparser::Operator::I32Const { value }) => {
992 Some(GlobalInit::I32(value))
993 }
994 Ok(wasmparser::Operator::I64Const { value }) => {
995 Some(GlobalInit::I64(value))
996 }
997 _ => None,
998 };
999 // #643: record the slot width from the DECLARED value type.
1000 // i64/f64 globals occupy 8 bytes (a register pair on the
1001 // 32-bit targets), v128 sixteen; laying every global out at
1002 // `index * 4` silently dropped the high word of every i64.
1003 let slot_bytes = match global.ty.content_type {
1004 wasmparser::ValType::I64 | wasmparser::ValType::F64 => 8,
1005 wasmparser::ValType::V128 => 16,
1006 _ => 4,
1007 };
1008 // GI-FPU-001 (#369): a float-typed global's initializer is
1009 // NOT captured (`init_i32` only decodes `i32.const`), so
1010 // its slot would be silently zeroed — record it so any
1011 // function accessing it loud-skips instead of reading a
1012 // silently-wrong 0.0.
1013 if matches!(
1014 global.ty.content_type,
1015 wasmparser::ValType::F32 | wasmparser::ValType::F64
1016 ) {
1017 float_globals.insert(num_imported_globals + idx as u32);
1018 }
1019 // #680: a v128-typed global's `v128.const` initializer is
1020 // not captured either (slot zeroed) and an access moves 4
1021 // of the 16 bytes — record it so accesses loud-skip.
1022 if global.ty.content_type == wasmparser::ValType::V128 {
1023 v128_globals.insert(num_imported_globals + idx as u32);
1024 }
1025 globals.push(WasmGlobal {
1026 index: idx as u32,
1027 init,
1028 mutable: global.ty.mutable,
1029 slot_bytes,
1030 });
1031 }
1032 }
1033 Payload::DataSection(reader) => {
1034 for data in reader {
1035 let data = data.context("Failed to parse data segment")?;
1036 if let wasmparser::DataKind::Active {
1037 memory_index,
1038 offset_expr,
1039 } = data.kind
1040 {
1041 let mut ops = offset_expr.get_operators_reader();
1042 let const_off = match ops.read() {
1043 Ok(wasmparser::Operator::I32Const { value }) => Some(value as u32),
1044 _ => None,
1045 };
1046 if memory_index == 0 {
1047 // Memory-0 behavior unchanged (frozen): a const-
1048 // offset segment is captured, anything else keeps
1049 // the legacy drop.
1050 if let Some(off) = const_off {
1051 data_segments.push((off, data.data.to_vec()));
1052 }
1053 } else if let Some(off) = const_off {
1054 // VCR-MEM-002 phase 1 (#406): capture non-default-
1055 // memory segments — previously they were silently
1056 // DROPPED (memory k's init data never shipped).
1057 extra_memory_data_segments.push((
1058 memory_index,
1059 off,
1060 data.data.to_vec(),
1061 ));
1062 } else {
1063 // A non-const offset on a non-default memory cannot
1064 // be placed at compile time — record it so the
1065 // multi-memory compile path declines LOUDLY instead
1066 // of shipping memory k uninitialized.
1067 multi_memory_decline.get_or_insert(format!(
1068 "active data segment on memory {memory_index} has a \
1069 non-constant offset expression — cannot be placed \
1070 at compile time (multi-memory phase 1, #406)"
1071 ));
1072 }
1073 }
1074 }
1075 }
1076 Payload::ElementSection(reader) => {
1077 // #275: collect every function index that initializes a table.
1078 // These are the `call_indirect` targets the direct-call closure
1079 // cannot see; `reachable_from_exports` unions them in when a
1080 // reachable function does a `call_indirect`. Both element forms
1081 // are handled: a flat function-index list, and the const-expr
1082 // form whose `ref.func` entries name the functions.
1083 for elem in reader {
1084 let elem = elem.context("Failed to parse element segment")?;
1085 // #642/#650: the segment's static placement — a const i32
1086 // offset of an ACTIVE segment into its target table (any
1087 // table index: the R11 region is contiguous, #650);
1088 // anything else is unverifiable and poisons the
1089 // closed-world type check.
1090 let (seg_table, seg_offset): (u32, Option<u32>) = match &elem.kind {
1091 wasmparser::ElementKind::Active {
1092 table_index,
1093 offset_expr,
1094 } => {
1095 let mut ops = offset_expr.get_operators_reader();
1096 let off = match ops.read() {
1097 Ok(wasmparser::Operator::I32Const { value }) => {
1098 u32::try_from(value).ok()
1099 }
1100 _ => None,
1101 };
1102 (table_index.unwrap_or(0), off)
1103 }
1104 _ => (0, None),
1105 };
1106 let mut seg_funcs: Option<Vec<u32>> = Some(Vec::new());
1107 match elem.items {
1108 wasmparser::ElementItems::Functions(funcs) => {
1109 for f in funcs {
1110 let f = f.context("Failed to parse element func index")?;
1111 elem_func_indices.push(f);
1112 if let Some(v) = seg_funcs.as_mut() {
1113 v.push(f);
1114 }
1115 }
1116 }
1117 wasmparser::ElementItems::Expressions(_, exprs) => {
1118 for expr in exprs {
1119 let expr = expr.context("Failed to parse element expr")?;
1120 // #642: an entry is verifiable only when it is
1121 // a single plain `ref.func` (reader yields the
1122 // op + the implicit `end`). `ref.null` or any
1123 // computed entry poisons the segment.
1124 let mut entry_func: Option<u32> = None;
1125 let mut plain = true;
1126 for (k, op) in expr.get_operators_reader().into_iter().enumerate() {
1127 match (k, op.context("Failed to parse element op")?) {
1128 (0, wasmparser::Operator::RefFunc { function_index }) => {
1129 elem_func_indices.push(function_index);
1130 entry_func = Some(function_index);
1131 }
1132 (_, wasmparser::Operator::End) => {}
1133 (_, wasmparser::Operator::RefFunc { function_index }) => {
1134 // Keep the pre-#642 reachability
1135 // behaviour: every ref.func seen
1136 // anywhere is a possible target.
1137 elem_func_indices.push(function_index);
1138 plain = false;
1139 }
1140 _ => plain = false,
1141 }
1142 }
1143 match (plain, entry_func, seg_funcs.as_mut()) {
1144 (true, Some(f), Some(v)) => v.push(f),
1145 _ => seg_funcs = None,
1146 }
1147 }
1148 }
1149 }
1150 elem_segments.push(ElemSegmentInfo {
1151 table_index: seg_table,
1152 offset: seg_offset,
1153 funcs: seg_funcs,
1154 });
1155 }
1156 }
1157 Payload::ExportSection(exports) => {
1158 for export in exports {
1159 let export = export.context("Failed to parse export")?;
1160 if export.kind == ExternalKind::Func {
1161 export_names.insert(export.index, export.name.to_string());
1162 }
1163 }
1164 }
1165 Payload::CodeSectionEntry(body) => {
1166 let (ops, op_offsets, block_arity, mut unsupported) =
1167 decode_function_body(&body, &type_block_arity, &float_globals, &v128_globals)?;
1168 // #680: a v128 param/result reaches the body only through
1169 // type-agnostic ops (a `local.get 0` passthrough compiles to
1170 // a 4-byte `mov`), so flag the SIGNATURE even when the body
1171 // contains no SIMD-proposal operator.
1172 if unsupported.is_none()
1173 && func_sig_has_v128
1174 .get(func_index as usize)
1175 .copied()
1176 .unwrap_or(false)
1177 {
1178 unsupported = Some(
1179 "signature has a v128 param/result — no SIMD lowering \
1180 for this target (#680)"
1181 .to_string(),
1182 );
1183 }
1184 let actual_index = num_imported_funcs + func_index;
1185 let export_name = export_names.get(&actual_index).cloned();
1186
1187 functions.push(FunctionOps {
1188 index: actual_index,
1189 export_name,
1190 debug_name: None, // filled from the `name` section after the loop
1191 ops,
1192 op_offsets,
1193 unsupported,
1194 block_arity,
1195 });
1196 func_index += 1;
1197 }
1198 Payload::CustomSection(c) => {
1199 // #394 Tier-1.x: the wasm `name` custom section.
1200 if let wasmparser::KnownCustom::Name(reader) = c.as_known() {
1201 parse_name_section_func_names(reader, &mut name_section_names);
1202 }
1203 // VCR-PERF-002 Phase 1 (#494): loom's `wsc.facts` section.
1204 // `parse_wsc_facts` is TOTAL (fail-safe skew, loom#231 Q4):
1205 // any malformed payload decodes to the empty fact list WITH a
1206 // stderr diagnostic, never an error — facts are optional
1207 // accelerators and must not be able to change a compilation
1208 // outcome. First section wins.
1209 if c.name() == crate::wsc_facts::WSC_FACTS_SECTION_NAME && wsc_facts.is_none() {
1210 let parsed = crate::wsc_facts::parse_wsc_facts(c.data());
1211 if let Some(reason) = &parsed.section_ignored {
1212 eprintln!(
1213 "warning: ignoring unparseable `wsc.facts` custom section \
1214 ({reason}) — facts are optional accelerators, compilation \
1215 is unaffected (#494 fail-safe skew rule)"
1216 );
1217 } else if parsed.records_skipped > 0 {
1218 eprintln!(
1219 "warning: skipped {} unknown/undecodable `wsc.facts` \
1220 record(s) (likely a newer loom emitter); {} known fact(s) \
1221 kept, compilation is unaffected (#494 fail-safe skew rule)",
1222 parsed.records_skipped,
1223 parsed.facts.len()
1224 );
1225 }
1226 wsc_facts = Some(parsed.facts);
1227 }
1228 }
1229 _ => {}
1230 }
1231 }
1232
1233 apply_name_section(&mut functions, &name_section_names);
1234
1235 Ok(DecodedModule {
1236 functions,
1237 memories,
1238 data_segments,
1239 extra_memory_data_segments,
1240 multi_memory_decline,
1241 imports,
1242 num_imported_funcs,
1243 func_arg_counts,
1244 type_arg_counts,
1245 func_ret_i64,
1246 type_ret_i64,
1247 func_params_i64,
1248 func_params_f32,
1249 func_params_f64,
1250 func_ret_f32,
1251 func_ret_f64,
1252 type_ret_f32,
1253 type_ret_f64,
1254 globals,
1255 elem_func_indices,
1256 table_size: table_sizes.first().copied().flatten(),
1257 table_sizes,
1258 elem_segments,
1259 func_type_indices,
1260 type_signatures,
1261 wsc_facts: wsc_facts.unwrap_or_default(),
1262 })
1263}
1264
1265/// Parse the function-names subsection of a wasm `name` custom section into
1266/// `out` (function index → developer-facing name, e.g.
1267/// `core::panicking::panic_fmt::h...`). Best-effort by design: the section is
1268/// DEBUG METADATA only, so a malformed entry is skipped rather than failing the
1269/// compile — no codegen path depends on it (#394 Tier-1.x).
1270fn parse_name_section_func_names(
1271 reader: wasmparser::NameSectionReader<'_>,
1272 out: &mut HashMap<u32, String>,
1273) {
1274 for subsection in reader.into_iter().flatten() {
1275 if let wasmparser::Name::Function(map) = subsection {
1276 for naming in map.into_iter().flatten() {
1277 out.insert(naming.index, naming.name.to_string());
1278 }
1279 }
1280 }
1281}
1282
1283/// Fill each function's `debug_name` from the `name`-section map (keyed by the
1284/// FULL function index, imports first — the same index space `FunctionOps.index`
1285/// uses). A function without an entry keeps `None` (⇒ `func_N` downstream).
1286fn apply_name_section(functions: &mut [FunctionOps], names: &HashMap<u32, String>) {
1287 if names.is_empty() {
1288 return;
1289 }
1290 for f in functions {
1291 f.debug_name = names.get(&f.index).cloned();
1292 }
1293}
1294
1295/// Decode a WASM binary and extract all function bodies as WasmOp sequences
1296pub fn decode_wasm_functions(wasm_bytes: &[u8]) -> Result<Vec<FunctionOps>> {
1297 let mut functions = Vec::new();
1298 let mut func_index = 0u32;
1299 let mut num_imported_funcs = 0u32;
1300 let mut export_names: HashMap<u32, String> = HashMap::new();
1301 let mut name_section_names: HashMap<u32, String> = HashMap::new();
1302 // #509: (param_count, result_count) per type index, for FuncType blocktypes.
1303 let mut type_block_arity: Vec<(u8, u8)> = Vec::new();
1304 // GI-FPU-001 (#369): float-typed globals (full index space, imports first)
1305 // whose accesses must loud-skip — see `decode_wasm_module`.
1306 let mut num_imported_globals = 0u32;
1307 let mut float_globals: std::collections::HashSet<u32> = std::collections::HashSet::new();
1308 // #680: v128-typed globals + v128 params/results — see `decode_wasm_module`.
1309 let mut v128_globals: std::collections::HashSet<u32> = std::collections::HashSet::new();
1310 let mut type_has_v128: Vec<bool> = Vec::new();
1311 let mut func_sig_has_v128: Vec<bool> = Vec::new();
1312
1313 for payload in Parser::new(0).parse_all(wasm_bytes) {
1314 let payload = payload.context("Failed to parse WASM payload")?;
1315
1316 match payload {
1317 Payload::TypeSection(reader) => {
1318 // #509: the blocktype-arity side-table needs the type section
1319 // for `BlockType::FuncType(i)` lookups (the wasm binary format
1320 // places types before code, so the table is complete before any
1321 // `CodeSectionEntry` is decoded).
1322 for rec_group in reader {
1323 let rec_group = rec_group.context("Failed to parse type")?;
1324 for sub_ty in rec_group.types() {
1325 type_block_arity.push(match &sub_ty.composite_type.inner {
1326 wasmparser::CompositeInnerType::Func(f) => (
1327 u8::try_from(f.params().len()).unwrap_or(u8::MAX),
1328 u8::try_from(f.results().len()).unwrap_or(u8::MAX),
1329 ),
1330 _ => (u8::MAX, u8::MAX),
1331 });
1332 // #680: v128 anywhere in the signature.
1333 type_has_v128.push(match &sub_ty.composite_type.inner {
1334 wasmparser::CompositeInnerType::Func(f) => f
1335 .params()
1336 .iter()
1337 .chain(f.results())
1338 .any(|t| *t == wasmparser::ValType::V128),
1339 _ => false,
1340 });
1341 }
1342 }
1343 }
1344 Payload::ImportSection(imports) => {
1345 // wasmparser 0.221+ compact-imports grouping — flatten groups
1346 // to individual imports (see the ImportSection handler above).
1347 for import in imports.into_imports() {
1348 let import = import.context("Failed to parse import")?;
1349 match import.ty {
1350 wasmparser::TypeRef::Func(_) => num_imported_funcs += 1,
1351 wasmparser::TypeRef::Global(g) => {
1352 // GI-FPU-001 (#369): see `decode_wasm_module` —
1353 // float-typed global accesses must loud-skip.
1354 if matches!(
1355 g.content_type,
1356 wasmparser::ValType::F32 | wasmparser::ValType::F64
1357 ) {
1358 float_globals.insert(num_imported_globals);
1359 }
1360 // #680: v128-typed imported globals — same lane.
1361 if g.content_type == wasmparser::ValType::V128 {
1362 v128_globals.insert(num_imported_globals);
1363 }
1364 num_imported_globals += 1;
1365 }
1366 _ => {}
1367 }
1368 }
1369 }
1370 Payload::FunctionSection(reader) => {
1371 // #680: defined-function type indices, in order — the per-
1372 // function v128-signature flag (`decode_wasm_module` gets this
1373 // from its existing FunctionSection handling).
1374 for ty in reader {
1375 let type_idx = ty.context("Failed to parse function type index")?;
1376 func_sig_has_v128.push(
1377 type_has_v128
1378 .get(type_idx as usize)
1379 .copied()
1380 .unwrap_or(false),
1381 );
1382 }
1383 }
1384 Payload::GlobalSection(reader) => {
1385 // GI-FPU-001 (#369): record f32/f64-typed defined globals so
1386 // `decode_function_body` flags accesses (their initializer is
1387 // dropped on this path too — same silent-zero hazard).
1388 for (idx, global) in reader.into_iter().enumerate() {
1389 let global = global.context("Failed to parse global")?;
1390 if matches!(
1391 global.ty.content_type,
1392 wasmparser::ValType::F32 | wasmparser::ValType::F64
1393 ) {
1394 float_globals.insert(num_imported_globals + idx as u32);
1395 }
1396 // #680: v128-typed defined globals — same lane.
1397 if global.ty.content_type == wasmparser::ValType::V128 {
1398 v128_globals.insert(num_imported_globals + idx as u32);
1399 }
1400 }
1401 }
1402 Payload::ExportSection(exports) => {
1403 for export in exports {
1404 let export = export.context("Failed to parse export")?;
1405 if export.kind == ExternalKind::Func {
1406 export_names.insert(export.index, export.name.to_string());
1407 }
1408 }
1409 }
1410 Payload::CodeSectionEntry(body) => {
1411 let (ops, op_offsets, block_arity, mut unsupported) =
1412 decode_function_body(&body, &type_block_arity, &float_globals, &v128_globals)?;
1413 // #680: v128 param/result — see `decode_wasm_module`.
1414 if unsupported.is_none()
1415 && func_sig_has_v128
1416 .get(func_index as usize)
1417 .copied()
1418 .unwrap_or(false)
1419 {
1420 unsupported = Some(
1421 "signature has a v128 param/result — no SIMD lowering \
1422 for this target (#680)"
1423 .to_string(),
1424 );
1425 }
1426 let actual_index = num_imported_funcs + func_index;
1427 let export_name = export_names.get(&actual_index).cloned();
1428
1429 functions.push(FunctionOps {
1430 index: actual_index,
1431 export_name,
1432 debug_name: None, // filled from the `name` section after the loop
1433 ops,
1434 op_offsets,
1435 unsupported,
1436 block_arity,
1437 });
1438 func_index += 1;
1439 }
1440 Payload::CustomSection(c) => {
1441 // #394 Tier-1.x: the wasm `name` custom section.
1442 if let wasmparser::KnownCustom::Name(reader) = c.as_known() {
1443 parse_name_section_func_names(reader, &mut name_section_names);
1444 }
1445 }
1446 _ => {}
1447 }
1448 }
1449
1450 apply_name_section(&mut functions, &name_section_names);
1451
1452 Ok(functions)
1453}
1454
1455/// Decoded function with its WasmOp sequence
1456#[derive(Debug, Clone)]
1457pub struct FunctionOps {
1458 /// Function index in the module (includes imported functions)
1459 pub index: u32,
1460 /// Export name if this function is exported
1461 pub export_name: Option<String>,
1462 /// #394 Tier-1.x: the function's developer-facing name from the wasm `name`
1463 /// custom section (function-names subsection), e.g.
1464 /// `core::panicking::panic_fmt::h6651313c3e2c6c2f` — present for INTERNAL
1465 /// (non-exported) functions too, unlike `export_name`. DEBUG METADATA only:
1466 /// consumed by the `--debug-line` `DW_TAG_subprogram` emit (name priority:
1467 /// name-section > export name > `func_N`); no codegen or symbol-table path
1468 /// reads it, so emitted `.text`/`.symtab` are unchanged (frozen-safe).
1469 /// `None` when the module has no `name` section or no entry for this index.
1470 pub debug_name: Option<String>,
1471 /// The WASM operations in this function body
1472 pub ops: Vec<WasmOp>,
1473 /// VCR-DBG-001 step 1 (#394): module-relative wasm byte offset of each op in
1474 /// `ops` (same index → same op). This is the address space DWARF-for-wasm
1475 /// `.debug_line` keys on, so it is the bridge from synth's op-index
1476 /// `source_line` to the input wasm's DWARF (wasm-offset → source). PURELY
1477 /// ADDITIVE metadata: no codegen path reads it, so emitted `.text` is
1478 /// unchanged and the frozen fixtures stay bit-identical. Empty until consumed
1479 /// by the DWARF emitter (Tier 1).
1480 pub op_offsets: Vec<u32>,
1481 /// `Some(reason)` when the body contained a value-affecting operator the
1482 /// decoder cannot lower (e.g. scalar f32/f64 — #369, bulk-memory
1483 /// memory.copy/fill). Such an op would otherwise be silently *dropped*
1484 /// (`convert_operator` → `None`), leaving the operand stack wrong and the
1485 /// function a silent miscompile. The compile path LOUD-SKIPS a flagged
1486 /// function (diagnostic + symbol absent → link error names it) instead —
1487 /// the #180/#185 "unsupported op must Err, never silently continue"
1488 /// contract. `None` once every op decoded or was intentionally ignorable
1489 /// (Nop).
1490 pub unsupported: Option<String>,
1491 /// #509: blocktype arity side-table — `(param_count, result_count)` of the
1492 /// k-th `Block`/`Loop`/`If` op in `ops`, in order of appearance.
1493 /// ORDINAL-keyed, not op-index-keyed, on purpose: the backend may rewrite
1494 /// the op stream before selection (e.g. the #539 `i32.const 0; memory.grow`
1495 /// → `memory.size` fold), which shifts op indices but never adds/removes
1496 /// control ops, so the ordinal stays aligned. `BlockType::Empty → (0,0)`,
1497 /// `ValType → (0,1)`, `FuncType(i) →` counts from the type section
1498 /// (saturated to u8; an unresolvable type index records `(u8::MAX,
1499 /// u8::MAX)` so the selector declines loudly instead of miscompiling).
1500 /// This is what lets the direct selector land a value carried by
1501 /// `br`/`br_if`/`br_table` in the target block's designated result
1502 /// register instead of dropping it — `WasmOp::Block/Loop/If` stay bare
1503 /// unit variants (zero ripple through the backends' match sites), and an
1504 /// empty table (hand-built op streams in unit tests) keeps the legacy
1505 /// void-block lowering.
1506 pub block_arity: Vec<(u8, u8)>,
1507}
1508
1509/// #509: `(param_count, result_count)` of a wasm blocktype, for the
1510/// [`FunctionOps::block_arity`] side-table. `type_block_arity` is the type
1511/// section's per-type-index counts (needed for the `FuncType` form); a missing
1512/// entry saturates to `(u8::MAX, u8::MAX)` so downstream declines loudly.
1513fn blocktype_arity(bt: &wasmparser::BlockType, type_block_arity: &[(u8, u8)]) -> (u8, u8) {
1514 match bt {
1515 wasmparser::BlockType::Empty => (0, 0),
1516 wasmparser::BlockType::Type(_) => (0, 1),
1517 wasmparser::BlockType::FuncType(i) => type_block_arity
1518 .get(*i as usize)
1519 .copied()
1520 .unwrap_or((u8::MAX, u8::MAX)),
1521 }
1522}
1523
1524/// The per-function payload [`decode_function_body`] extracts: `(ops,
1525/// op_offsets, block_arity, unsupported)` — see the matching
1526/// [`FunctionOps`] fields for each component's contract.
1527type DecodedBody = (Vec<WasmOp>, Vec<u32>, Vec<(u8, u8)>, Option<String>);
1528
1529/// Decode a single function body to WasmOp sequence.
1530///
1531/// Returns the ops plus `Some(reason)` if any operator was a value-affecting
1532/// op the decoder cannot lower (so the function must be loud-skipped, #369 —
1533/// not silently miscompiled by dropping the op).
1534fn decode_function_body(
1535 body: &wasmparser::FunctionBody,
1536 type_block_arity: &[(u8, u8)],
1537 float_globals: &std::collections::HashSet<u32>,
1538 v128_globals: &std::collections::HashSet<u32>,
1539) -> Result<DecodedBody> {
1540 let mut ops = Vec::new();
1541 // VCR-DBG-001 step 1: parallel to `ops` — the module-relative wasm byte
1542 // offset of each emitted op (the DWARF-for-wasm address space). Captured via
1543 // the offset-aware reader; pushed only when an op is pushed, so indices stay
1544 // aligned with `ops`. Additive metadata, no codegen consumer ⇒ frozen-safe.
1545 let mut op_offsets = Vec::new();
1546 // #509: ordinal blocktype-arity side-table — one entry per Block/Loop/If in
1547 // `ops` order (see `FunctionOps::block_arity`).
1548 let mut block_arity: Vec<(u8, u8)> = Vec::new();
1549 let mut unsupported: Option<String> = None;
1550
1551 // #680: a v128-typed LOCAL is expressible with zero SIMD-proposal
1552 // operators (`local.get`/`local.set`/`local.tee` are type-agnostic), but
1553 // every selector lowers those as 4-byte (or 8-byte i64) register moves —
1554 // silently truncating the 16-byte value. Flag the declaration up front.
1555 for local in body.get_locals_reader()? {
1556 let (count, ty) = local.context("Failed to read local declaration")?;
1557 if unsupported.is_none() && count > 0 && ty == wasmparser::ValType::V128 {
1558 unsupported = Some(
1559 "declares a v128-typed local — no SIMD lowering for this \
1560 target, accesses would silently truncate the 16-byte value \
1561 (#680)"
1562 .to_string(),
1563 );
1564 }
1565 }
1566
1567 let ops_reader = body.get_operators_reader()?;
1568 for item in ops_reader.into_iter_with_offsets() {
1569 let (op, offset) = item.context("Failed to read operator")?;
1570
1571 // #680: SIMD (v128) category-level honesty guard. Some SIMD ops decode
1572 // into `WasmOp` v128 variants that only a dead, never-wired Helium/MVE
1573 // prototype can select (`has_helium` is set by tests alone), so on
1574 // every real target they were silently dropped at selection —
1575 // `i32x4.add` compiled to an operand passthrough and `v128.store`
1576 // left memory unwritten. Catch the ENTIRE SIMD + relaxed-SIMD
1577 // proposal space here (macro-generated from wasmparser's operator
1578 // table — no hand-kept list to fall out of date) and route the
1579 // function through the same loud-skip/honest-bail lane as scalar
1580 // floats (GI-FPU-001). Targets with real SIMD hardware (Helium/MVE on
1581 // cortex-m55) can lift this once a lowering is actually wired.
1582 if unsupported.is_none() && is_simd_operator(&op) {
1583 unsupported = Some(format!(
1584 "{op:?}: no SIMD lowering for this target — the op would be \
1585 silently dropped to a no-op (WASM SIMD proposal, #680)"
1586 ));
1587 }
1588
1589 if let Some(wasm_op) = convert_operator(&op) {
1590 // #509: capture the blocktype arity BEFORE the enum flattens it away
1591 // (`WasmOp::Block/Loop/If` are unit variants by design).
1592 if let wasmparser::Operator::Block { blockty }
1593 | wasmparser::Operator::Loop { blockty }
1594 | wasmparser::Operator::If { blockty } = &op
1595 {
1596 block_arity.push(blocktype_arity(blockty, type_block_arity));
1597 }
1598 // GI-FPU-001 (#369): `global.get`/`global.set` decode fine (the
1599 // ops are type-agnostic), but an f32/f64-typed global has no
1600 // float lowering — its const initializer is dropped (slot zeroed),
1601 // so a read returns a silently-wrong 0.0. Flag the function for
1602 // the same loud-skip the scalar float ops get.
1603 if unsupported.is_none()
1604 && let WasmOp::GlobalGet(i) | WasmOp::GlobalSet(i) = &wasm_op
1605 && float_globals.contains(i)
1606 {
1607 unsupported = Some(format!(
1608 "{wasm_op:?} on an f32/f64-typed global — float globals \
1609 have no lowering, the initializer would be silently \
1610 zeroed (GI-FPU-001)"
1611 ));
1612 }
1613 // #680: same hazard for v128-typed globals — `global.get`/
1614 // `global.set` decode fine, but there is no SIMD lowering: the
1615 // access would move 4 of the 16 bytes and the `v128.const`
1616 // initializer is never captured (slot zeroed).
1617 if unsupported.is_none()
1618 && let WasmOp::GlobalGet(i) | WasmOp::GlobalSet(i) = &wasm_op
1619 && v128_globals.contains(i)
1620 {
1621 unsupported = Some(format!(
1622 "{wasm_op:?} on a v128-typed global — no SIMD lowering \
1623 for this target (#680)"
1624 ));
1625 }
1626 // VCR-MEM-002 phase 1 (#406): a load/store whose memarg targets a
1627 // NON-DEFAULT memory must carry its index — dropping it silently
1628 // aliased every memory onto the one R11 base (a store to memory
1629 // `$b` clobbered memory `$a`). memidx 0 stays the bare variant, so
1630 // single-memory streams are bit-identical by construction.
1631 let wasm_op = match memarg_memory_index(&op) {
1632 Some(mem) if mem > 0 => WasmOp::MultiMemory {
1633 memory: mem,
1634 op: Box::new(wasm_op),
1635 },
1636 _ => wasm_op,
1637 };
1638 ops.push(wasm_op);
1639 op_offsets.push(offset as u32);
1640 } else if unsupported.is_none() && !is_intentionally_ignored(&op) {
1641 // #406 phase 1: bulk-memory ops on a non-default memory (including
1642 // the cross-memory `memory.copy` dst_mem != src_mem form) have no
1643 // lowering yet — name the decline precisely instead of the generic
1644 // dropped-op message.
1645 unsupported = match &op {
1646 wasmparser::Operator::MemoryCopy { dst_mem, src_mem }
1647 if *dst_mem != 0 || *src_mem != 0 =>
1648 {
1649 Some(format!(
1650 "memory.copy dst_mem={dst_mem} src_mem={src_mem}: \
1651 cross-/non-default-memory memory.copy is not lowered \
1652 in multi-memory phase 1 (#406) — only memory-0 \
1653 memory.copy is supported"
1654 ))
1655 }
1656 wasmparser::Operator::MemoryFill { mem } if *mem != 0 => Some(format!(
1657 "memory.fill mem={mem}: non-default-memory memory.fill is \
1658 not lowered in multi-memory phase 1 (#406) — only \
1659 memory-0 memory.fill is supported"
1660 )),
1661 // The op was DROPPED by `convert_operator` (`_ => None`) and is
1662 // not an intentional no-op (Nop) — record it so the function is
1663 // loud-skipped rather than silently miscompiled (#369).
1664 _ => Some(format!("{op:?}")),
1665 };
1666 }
1667 }
1668
1669 Ok((ops, op_offsets, block_arity, unsupported))
1670}
1671
1672/// Operators that `convert_operator` returns `None` for *on purpose* — they
1673/// carry no value-affecting semantics for our backend, so dropping them is
1674/// correct (NOT a silent miscompile). Everything else that decodes to `None`
1675/// is an unsupported op that must loud-skip its function (#369).
1676///
1677/// #665: `Unreachable` is NOT on this list — it traps (WASM §4.4.5), so it
1678/// decodes to `WasmOp::Unreachable` and every backend lowers it to a trap
1679/// instruction (or loud-declines). Only `Nop` is genuinely ignorable.
1680fn is_intentionally_ignored(op: &wasmparser::Operator) -> bool {
1681 use wasmparser::Operator::*;
1682 matches!(op, Nop)
1683}
1684
1685/// #680: is `op` from the WASM SIMD or relaxed-SIMD proposal?
1686///
1687/// CATEGORY-LEVEL by construction: the match is macro-generated from
1688/// `wasmparser::for_each_operator!`'s own proposal markers (`@simd` /
1689/// `@relaxed_simd`), so it covers the entire SIMD operator space of the
1690/// pinned wasmparser — there is no hand-kept op list that a new lane op,
1691/// load/store variant, or relaxed-SIMD instruction can silently fall out of.
1692/// Used to loud-skip functions with SIMD ops: no target has a SIMD lowering
1693/// wired today (the Helium/MVE selector arms are gated on a `has_helium`
1694/// flag only tests set), so a decoded v128 `WasmOp` was silently dropped at
1695/// selection — the #554-class miscompile this predicate closes.
1696fn is_simd_operator(op: &wasmparser::Operator) -> bool {
1697 macro_rules! define_match_operator {
1698 ($( @$proposal:ident $op:ident $({ $($arg:ident: $argty:ty),* })? => $visit:ident ($($ann:tt)*))*) => {
1699 match op {
1700 $(
1701 wasmparser::Operator::$op { .. } => {
1702 define_match_operator!(impl_one @$proposal)
1703 }
1704 )*
1705 // `Operator` is non-exhaustive; an operator outside the
1706 // pinned wasmparser's own table cannot be produced by it.
1707 _ => false,
1708 }
1709 };
1710 (impl_one @simd) => { true };
1711 (impl_one @relaxed_simd) => { true };
1712 (impl_one @$proposal:ident) => { false };
1713 }
1714 wasmparser::for_each_operator!(define_match_operator)
1715}
1716
1717/// VCR-MEM-002 phase 1 (#406): the `memarg.memory` index of a load/store
1718/// operator that [`convert_operator`] lowers, `None` for every other op.
1719///
1720/// MIRROR PIN: this list must cover exactly the memarg-carrying arms of
1721/// `convert_operator` (every `{ memarg }` load/store it returns `Some` for).
1722/// A memarg op missing HERE but lowered THERE would silently drop a non-zero
1723/// memory index again — the pre-#406 aliasing bug. Ops `convert_operator`
1724/// drops (`_ => None`) loud-skip their function regardless, so they need no
1725/// entry. `memory.size`/`grow`/`copy`/`fill` carry their indices in their own
1726/// `WasmOp` variants / decode-time declines, not via this helper.
1727fn memarg_memory_index(op: &wasmparser::Operator) -> Option<u32> {
1728 use wasmparser::Operator::*;
1729 match op {
1730 I32Load { memarg }
1731 | I32Store { memarg }
1732 | I64Load { memarg }
1733 | I64Store { memarg }
1734 | I32Load8S { memarg }
1735 | I32Load8U { memarg }
1736 | I32Load16S { memarg }
1737 | I32Load16U { memarg }
1738 | I32Store8 { memarg }
1739 | I32Store16 { memarg }
1740 | I64Load8S { memarg }
1741 | I64Load8U { memarg }
1742 | I64Load16S { memarg }
1743 | I64Load16U { memarg }
1744 | I64Load32S { memarg }
1745 | I64Load32U { memarg }
1746 | I64Store8 { memarg }
1747 | I64Store16 { memarg }
1748 | I64Store32 { memarg }
1749 | F32Load { memarg }
1750 | F32Store { memarg }
1751 | F64Load { memarg }
1752 | F64Store { memarg }
1753 | V128Load { memarg }
1754 | V128Store { memarg } => Some(memarg.memory),
1755 _ => None,
1756 }
1757}
1758
1759/// Convert a wasmparser Operator to our WasmOp enum
1760fn convert_operator(op: &wasmparser::Operator) -> Option<WasmOp> {
1761 use wasmparser::Operator::*;
1762
1763 match op {
1764 // Constants
1765 I32Const { value } => Some(WasmOp::I32Const(*value)),
1766
1767 // i32 Arithmetic
1768 I32Add => Some(WasmOp::I32Add),
1769 I32Sub => Some(WasmOp::I32Sub),
1770 I32Mul => Some(WasmOp::I32Mul),
1771 I32DivS => Some(WasmOp::I32DivS),
1772 I32DivU => Some(WasmOp::I32DivU),
1773 I32RemS => Some(WasmOp::I32RemS),
1774 I32RemU => Some(WasmOp::I32RemU),
1775
1776 // i64 Constants
1777 I64Const { value } => Some(WasmOp::I64Const(*value)),
1778
1779 // i64 Arithmetic
1780 I64Add => Some(WasmOp::I64Add),
1781 I64Sub => Some(WasmOp::I64Sub),
1782 I64Mul => Some(WasmOp::I64Mul),
1783 I64DivS => Some(WasmOp::I64DivS),
1784 I64DivU => Some(WasmOp::I64DivU),
1785 I64RemS => Some(WasmOp::I64RemS),
1786 I64RemU => Some(WasmOp::I64RemU),
1787
1788 // i64 Bitwise
1789 I64And => Some(WasmOp::I64And),
1790 I64Or => Some(WasmOp::I64Or),
1791 I64Xor => Some(WasmOp::I64Xor),
1792 I64Shl => Some(WasmOp::I64Shl),
1793 I64ShrS => Some(WasmOp::I64ShrS),
1794 I64ShrU => Some(WasmOp::I64ShrU),
1795 I64Rotl => Some(WasmOp::I64Rotl),
1796 I64Rotr => Some(WasmOp::I64Rotr),
1797 I64Clz => Some(WasmOp::I64Clz),
1798 I64Ctz => Some(WasmOp::I64Ctz),
1799 I64Popcnt => Some(WasmOp::I64Popcnt),
1800 I64Extend8S => Some(WasmOp::I64Extend8S),
1801 I64Extend16S => Some(WasmOp::I64Extend16S),
1802 I64Extend32S => Some(WasmOp::I64Extend32S),
1803 // i32<->i64 width conversions. Previously UNMAPPED → silently dropped,
1804 // which left an i32 value as a 64-bit operand with a garbage high half
1805 // (harmless when a following `i64.shl 32` discards it, but a latent
1806 // miscompile for extend-then-arithmetic, and it breaks width-correct
1807 // register allocation). (#204)
1808 I64ExtendI32U => Some(WasmOp::I64ExtendI32U),
1809 I64ExtendI32S => Some(WasmOp::I64ExtendI32S),
1810 I32WrapI64 => Some(WasmOp::I32WrapI64),
1811
1812 // i64 Comparison
1813 I64Eqz => Some(WasmOp::I64Eqz),
1814 I64Eq => Some(WasmOp::I64Eq),
1815 I64Ne => Some(WasmOp::I64Ne),
1816 I64LtS => Some(WasmOp::I64LtS),
1817 I64LtU => Some(WasmOp::I64LtU),
1818 I64LeS => Some(WasmOp::I64LeS),
1819 I64LeU => Some(WasmOp::I64LeU),
1820 I64GtS => Some(WasmOp::I64GtS),
1821 I64GtU => Some(WasmOp::I64GtU),
1822 I64GeS => Some(WasmOp::I64GeS),
1823 I64GeU => Some(WasmOp::I64GeU),
1824
1825 // Bitwise
1826 I32And => Some(WasmOp::I32And),
1827 I32Or => Some(WasmOp::I32Or),
1828 I32Xor => Some(WasmOp::I32Xor),
1829 I32Shl => Some(WasmOp::I32Shl),
1830 I32ShrS => Some(WasmOp::I32ShrS),
1831 I32ShrU => Some(WasmOp::I32ShrU),
1832 I32Rotl => Some(WasmOp::I32Rotl),
1833 I32Rotr => Some(WasmOp::I32Rotr),
1834 I32Clz => Some(WasmOp::I32Clz),
1835 I32Ctz => Some(WasmOp::I32Ctz),
1836 I32Popcnt => Some(WasmOp::I32Popcnt),
1837 I32Extend8S => Some(WasmOp::I32Extend8S),
1838 I32Extend16S => Some(WasmOp::I32Extend16S),
1839
1840 // Comparison
1841 I32Eqz => Some(WasmOp::I32Eqz),
1842 I32Eq => Some(WasmOp::I32Eq),
1843 I32Ne => Some(WasmOp::I32Ne),
1844 I32LtS => Some(WasmOp::I32LtS),
1845 I32LtU => Some(WasmOp::I32LtU),
1846 I32LeS => Some(WasmOp::I32LeS),
1847 I32LeU => Some(WasmOp::I32LeU),
1848 I32GtS => Some(WasmOp::I32GtS),
1849 I32GtU => Some(WasmOp::I32GtU),
1850 I32GeS => Some(WasmOp::I32GeS),
1851 I32GeU => Some(WasmOp::I32GeU),
1852
1853 // Memory
1854 I32Load { memarg } => Some(WasmOp::I32Load {
1855 offset: memarg.offset as u32,
1856 align: memarg.align as u32,
1857 }),
1858 I32Store { memarg } => Some(WasmOp::I32Store {
1859 offset: memarg.offset as u32,
1860 align: memarg.align as u32,
1861 }),
1862 // #372: full-width i64 load/store. The selector already lowers these to
1863 // a lo/hi i32 register-pair access (`generate_i64_load/store_with_bounds_check`,
1864 // reusing the #171 pair regalloc) — only the decoder arm was missing, so
1865 // `i64.load`/`i64.store` fell through `_ => None` and (since v0.11.46)
1866 // loud-skipped their function. The narrow forms (I64Load8.. / I64Store32)
1867 // were already decoded below.
1868 I64Load { memarg } => Some(WasmOp::I64Load {
1869 offset: memarg.offset as u32,
1870 align: memarg.align as u32,
1871 }),
1872 I64Store { memarg } => Some(WasmOp::I64Store {
1873 offset: memarg.offset as u32,
1874 align: memarg.align as u32,
1875 }),
1876
1877 // Sub-word loads (i32)
1878 I32Load8S { memarg } => Some(WasmOp::I32Load8S {
1879 offset: memarg.offset as u32,
1880 align: memarg.align as u32,
1881 }),
1882 I32Load8U { memarg } => Some(WasmOp::I32Load8U {
1883 offset: memarg.offset as u32,
1884 align: memarg.align as u32,
1885 }),
1886 I32Load16S { memarg } => Some(WasmOp::I32Load16S {
1887 offset: memarg.offset as u32,
1888 align: memarg.align as u32,
1889 }),
1890 I32Load16U { memarg } => Some(WasmOp::I32Load16U {
1891 offset: memarg.offset as u32,
1892 align: memarg.align as u32,
1893 }),
1894
1895 // Sub-word stores (i32)
1896 I32Store8 { memarg } => Some(WasmOp::I32Store8 {
1897 offset: memarg.offset as u32,
1898 align: memarg.align as u32,
1899 }),
1900 I32Store16 { memarg } => Some(WasmOp::I32Store16 {
1901 offset: memarg.offset as u32,
1902 align: memarg.align as u32,
1903 }),
1904
1905 // Local/Global
1906 LocalGet { local_index } => Some(WasmOp::LocalGet(*local_index)),
1907 LocalSet { local_index } => Some(WasmOp::LocalSet(*local_index)),
1908 LocalTee { local_index } => Some(WasmOp::LocalTee(*local_index)),
1909 GlobalGet { global_index } => Some(WasmOp::GlobalGet(*global_index)),
1910 GlobalSet { global_index } => Some(WasmOp::GlobalSet(*global_index)),
1911
1912 // Control flow
1913 Block { .. } => Some(WasmOp::Block),
1914 Loop { .. } => Some(WasmOp::Loop),
1915 Br { relative_depth } => Some(WasmOp::Br(*relative_depth)),
1916 BrIf { relative_depth } => Some(WasmOp::BrIf(*relative_depth)),
1917 // br_table: indexed multi-way branch. Previously UNMAPPED → silently
1918 // dropped, so the selector never emitted the index dispatch and control
1919 // fell straight into the first table arm — every br_table behaved as if
1920 // it always took target 0 (gale's binary-sem WAKE path never fired). The
1921 // jump-table relative depths + default depth are preserved in order.
1922 BrTable { targets } => {
1923 let default = targets.default();
1924 let tgts: Vec<u32> = targets.targets().filter_map(Result::ok).collect();
1925 Some(WasmOp::BrTable {
1926 targets: tgts,
1927 default,
1928 })
1929 }
1930 Return => Some(WasmOp::Return),
1931 Call { function_index } => Some(WasmOp::Call(*function_index)),
1932 CallIndirect {
1933 type_index,
1934 table_index,
1935 ..
1936 } => Some(WasmOp::CallIndirect {
1937 type_index: *type_index,
1938 table_index: *table_index,
1939 }),
1940
1941 // End is needed for control flow pattern matching
1942 End => Some(WasmOp::End),
1943
1944 // #665: `unreachable` MUST reach the backends — WASM Core §4.4.5
1945 // requires it to trap unconditionally. It was previously dropped here
1946 // (treated like Nop), so every backend compiled it to a no-op and
1947 // control FELL THROUGH panic!/abort/unreachable-default guards with
1948 // undefined register state. The selector arms (ARM: UDF #0, RV32:
1949 // ebreak) already existed; they just never received the op.
1950 Unreachable => Some(WasmOp::Unreachable),
1951
1952 // Nop - skip (genuinely no semantics)
1953 Nop => None,
1954
1955 // Drop is needed for br_if pattern matching
1956 Drop => Some(WasmOp::Drop),
1957
1958 // Select
1959 Select => Some(WasmOp::Select),
1960
1961 // If/Else - simplified handling
1962 If { .. } => Some(WasmOp::If),
1963 Else => Some(WasmOp::Else),
1964
1965 // i64 sub-word loads
1966 I64Load8S { memarg } => Some(WasmOp::I64Load8S {
1967 offset: memarg.offset as u32,
1968 align: memarg.align as u32,
1969 }),
1970 I64Load8U { memarg } => Some(WasmOp::I64Load8U {
1971 offset: memarg.offset as u32,
1972 align: memarg.align as u32,
1973 }),
1974 I64Load16S { memarg } => Some(WasmOp::I64Load16S {
1975 offset: memarg.offset as u32,
1976 align: memarg.align as u32,
1977 }),
1978 I64Load16U { memarg } => Some(WasmOp::I64Load16U {
1979 offset: memarg.offset as u32,
1980 align: memarg.align as u32,
1981 }),
1982 I64Load32S { memarg } => Some(WasmOp::I64Load32S {
1983 offset: memarg.offset as u32,
1984 align: memarg.align as u32,
1985 }),
1986 I64Load32U { memarg } => Some(WasmOp::I64Load32U {
1987 offset: memarg.offset as u32,
1988 align: memarg.align as u32,
1989 }),
1990
1991 // i64 sub-word stores
1992 I64Store8 { memarg } => Some(WasmOp::I64Store8 {
1993 offset: memarg.offset as u32,
1994 align: memarg.align as u32,
1995 }),
1996 I64Store16 { memarg } => Some(WasmOp::I64Store16 {
1997 offset: memarg.offset as u32,
1998 align: memarg.align as u32,
1999 }),
2000 I64Store32 { memarg } => Some(WasmOp::I64Store32 {
2001 offset: memarg.offset as u32,
2002 align: memarg.align as u32,
2003 }),
2004
2005 // Memory management
2006 MemorySize { mem, .. } => Some(WasmOp::MemorySize(*mem)),
2007 MemoryGrow { mem, .. } => Some(WasmOp::MemoryGrow(*mem)),
2008
2009 // Bulk memory (#374). The backend supports a single linear memory
2010 // (memory 0); any non-zero memory index falls through to `_ => None` and
2011 // loud-skips the function (GI-FPU-001 honesty contract) rather than
2012 // miscompiling a multi-memory copy. memory.copy reads dst/src memories;
2013 // memory.fill one. The selector lowers these to a bounds-checked byte
2014 // loop (see select_with_stack).
2015 MemoryCopy {
2016 dst_mem: 0,
2017 src_mem: 0,
2018 } => Some(WasmOp::MemoryCopy),
2019 MemoryFill { mem: 0 } => Some(WasmOp::MemoryFill),
2020
2021 // ========================================================================
2022 // v128 SIMD operations (WASM SIMD proposal, 0xFD prefix)
2023 // ========================================================================
2024 V128Const { value } => {
2025 let mut bytes = [0u8; 16];
2026 bytes.copy_from_slice(value.bytes());
2027 Some(WasmOp::V128Const(bytes))
2028 }
2029 V128Load { memarg } => Some(WasmOp::V128Load {
2030 offset: memarg.offset as u32,
2031 align: memarg.align as u32,
2032 }),
2033 V128Store { memarg } => Some(WasmOp::V128Store {
2034 offset: memarg.offset as u32,
2035 align: memarg.align as u32,
2036 }),
2037
2038 // v128 bitwise
2039 V128And => Some(WasmOp::V128And),
2040 V128Or => Some(WasmOp::V128Or),
2041 V128Xor => Some(WasmOp::V128Xor),
2042 V128Not => Some(WasmOp::V128Not),
2043 V128AndNot => Some(WasmOp::V128AndNot),
2044
2045 // i8x16
2046 I8x16Add => Some(WasmOp::I8x16Add),
2047 I8x16Sub => Some(WasmOp::I8x16Sub),
2048 I8x16Neg => Some(WasmOp::I8x16Neg),
2049 I8x16Eq => Some(WasmOp::I8x16Eq),
2050 I8x16Ne => Some(WasmOp::I8x16Ne),
2051 I8x16LtS => Some(WasmOp::I8x16LtS),
2052 I8x16LtU => Some(WasmOp::I8x16LtU),
2053 I8x16GtS => Some(WasmOp::I8x16GtS),
2054 I8x16GtU => Some(WasmOp::I8x16GtU),
2055 I8x16LeS => Some(WasmOp::I8x16LeS),
2056 I8x16LeU => Some(WasmOp::I8x16LeU),
2057 I8x16GeS => Some(WasmOp::I8x16GeS),
2058 I8x16GeU => Some(WasmOp::I8x16GeU),
2059 I8x16Splat => Some(WasmOp::I8x16Splat),
2060 I8x16ExtractLaneS { lane } => Some(WasmOp::I8x16ExtractLaneS(*lane)),
2061 I8x16ExtractLaneU { lane } => Some(WasmOp::I8x16ExtractLaneU(*lane)),
2062 I8x16ReplaceLane { lane } => Some(WasmOp::I8x16ReplaceLane(*lane)),
2063 I8x16Shuffle { lanes } => Some(WasmOp::I8x16Shuffle(*lanes)),
2064 I8x16Swizzle => Some(WasmOp::I8x16Swizzle),
2065
2066 // i16x8
2067 I16x8Add => Some(WasmOp::I16x8Add),
2068 I16x8Sub => Some(WasmOp::I16x8Sub),
2069 I16x8Mul => Some(WasmOp::I16x8Mul),
2070 I16x8Neg => Some(WasmOp::I16x8Neg),
2071 I16x8Eq => Some(WasmOp::I16x8Eq),
2072 I16x8Ne => Some(WasmOp::I16x8Ne),
2073 I16x8LtS => Some(WasmOp::I16x8LtS),
2074 I16x8LtU => Some(WasmOp::I16x8LtU),
2075 I16x8GtS => Some(WasmOp::I16x8GtS),
2076 I16x8GtU => Some(WasmOp::I16x8GtU),
2077 I16x8LeS => Some(WasmOp::I16x8LeS),
2078 I16x8LeU => Some(WasmOp::I16x8LeU),
2079 I16x8GeS => Some(WasmOp::I16x8GeS),
2080 I16x8GeU => Some(WasmOp::I16x8GeU),
2081 I16x8Splat => Some(WasmOp::I16x8Splat),
2082 I16x8ExtractLaneS { lane } => Some(WasmOp::I16x8ExtractLaneS(*lane)),
2083 I16x8ExtractLaneU { lane } => Some(WasmOp::I16x8ExtractLaneU(*lane)),
2084 I16x8ReplaceLane { lane } => Some(WasmOp::I16x8ReplaceLane(*lane)),
2085
2086 // i32x4
2087 I32x4Add => Some(WasmOp::I32x4Add),
2088 I32x4Sub => Some(WasmOp::I32x4Sub),
2089 I32x4Mul => Some(WasmOp::I32x4Mul),
2090 I32x4Neg => Some(WasmOp::I32x4Neg),
2091 I32x4Eq => Some(WasmOp::I32x4Eq),
2092 I32x4Ne => Some(WasmOp::I32x4Ne),
2093 I32x4LtS => Some(WasmOp::I32x4LtS),
2094 I32x4LtU => Some(WasmOp::I32x4LtU),
2095 I32x4GtS => Some(WasmOp::I32x4GtS),
2096 I32x4GtU => Some(WasmOp::I32x4GtU),
2097 I32x4LeS => Some(WasmOp::I32x4LeS),
2098 I32x4LeU => Some(WasmOp::I32x4LeU),
2099 I32x4GeS => Some(WasmOp::I32x4GeS),
2100 I32x4GeU => Some(WasmOp::I32x4GeU),
2101 I32x4Splat => Some(WasmOp::I32x4Splat),
2102 I32x4ExtractLane { lane } => Some(WasmOp::I32x4ExtractLane(*lane)),
2103 I32x4ReplaceLane { lane } => Some(WasmOp::I32x4ReplaceLane(*lane)),
2104
2105 // i64x2
2106 I64x2Add => Some(WasmOp::I64x2Add),
2107 I64x2Sub => Some(WasmOp::I64x2Sub),
2108 I64x2Mul => Some(WasmOp::I64x2Mul),
2109 I64x2Neg => Some(WasmOp::I64x2Neg),
2110 I64x2Eq => Some(WasmOp::I64x2Eq),
2111 I64x2Ne => Some(WasmOp::I64x2Ne),
2112 I64x2LtS => Some(WasmOp::I64x2LtS),
2113 I64x2GtS => Some(WasmOp::I64x2GtS),
2114 I64x2LeS => Some(WasmOp::I64x2LeS),
2115 I64x2GeS => Some(WasmOp::I64x2GeS),
2116 I64x2Splat => Some(WasmOp::I64x2Splat),
2117 I64x2ExtractLane { lane } => Some(WasmOp::I64x2ExtractLane(*lane)),
2118 I64x2ReplaceLane { lane } => Some(WasmOp::I64x2ReplaceLane(*lane)),
2119
2120 // === Scalar f32 (GI-FPU-002 phase 1, #619/#369) ===
2121 // Un-dropped so the FPU targets (cortex-m4f/m7/m7dp) can route these to
2122 // the VFP selector arms in `select_with_stack`. The FPU gate lives at
2123 // the selector/validate layer (`requires_fpu()` + `set_target`): on a
2124 // non-FPU target (m0/m3/r5) these still honest-reject. f64 stays dropped
2125 // (phase 2 — M7DP D-registers). Only the wired scope is un-dropped; the
2126 // rest of the scalar f32 surface (abs/neg/sqrt/min/max/…) still falls to
2127 // `_ => None` and loud-skips its function until phase 1b wires it.
2128 F32Add => Some(WasmOp::F32Add),
2129 F32Sub => Some(WasmOp::F32Sub),
2130 F32Mul => Some(WasmOp::F32Mul),
2131 F32Div => Some(WasmOp::F32Div),
2132 F32Eq => Some(WasmOp::F32Eq),
2133 F32Ne => Some(WasmOp::F32Ne),
2134 F32Lt => Some(WasmOp::F32Lt),
2135 F32Le => Some(WasmOp::F32Le),
2136 F32Gt => Some(WasmOp::F32Gt),
2137 F32Ge => Some(WasmOp::F32Ge),
2138 F32Const { value } => Some(WasmOp::F32Const(f32::from_bits(value.bits()))),
2139 // #708 (phase 1b): `f32.load` un-dropped. The selector lowers it as the
2140 // proven `i32.load` address sequence (`[R11,idx]`→absolute-base rewrite +
2141 // bounds guard) into a core register, then a bit-exact `VMOV Sd,Rd`
2142 // (reinterpret) — a VLDR loads the same 4 bytes, so the bit pattern is
2143 // identical.
2144 F32Load { memarg } => Some(WasmOp::F32Load {
2145 offset: memarg.offset as u32,
2146 align: memarg.align as u32,
2147 }),
2148 // #719 (phase 1b): `f32.store` — the VFP-store twin of `f32.load`. The
2149 // selector moves the S-register value into a core register (`VMOV Rn,Sn`,
2150 // a reinterpret) and reuses the PROVEN `i32.store` address path; a VSTR
2151 // would write the same 4 bytes, so the stored word is bit-exact. (falcon
2152 // has 10 f32.store functions, #719.)
2153 F32Store { memarg } => Some(WasmOp::F32Store {
2154 offset: memarg.offset as u32,
2155 align: memarg.align as u32,
2156 }),
2157 // #719 (phase 1b): scalar f32 sign-family math — `VABS.F32` / `VNEG.F32`
2158 // and the `copysign` sign-bit splice. Pure single-precision VFP, no
2159 // numeric approximation; bit-exact across ±0.0 / NaN-sign / ±inf.
2160 F32Abs => Some(WasmOp::F32Abs),
2161 F32Neg => Some(WasmOp::F32Neg),
2162 F32Copysign => Some(WasmOp::F32Copysign),
2163 // #708 (phase 1b): the f32<->i32 bit-casts. Pure `VMOV` between a core
2164 // register and a single-precision S-register — no numeric conversion.
2165 F32ReinterpretI32 => Some(WasmOp::F32ReinterpretI32),
2166 I32ReinterpretF32 => Some(WasmOp::I32ReinterpretF32),
2167 F32ConvertI32S => Some(WasmOp::F32ConvertI32S),
2168 F32ConvertI32U => Some(WasmOp::F32ConvertI32U),
2169 I32TruncF32S => Some(WasmOp::I32TruncF32S),
2170 I32TruncF32U => Some(WasmOp::I32TruncF32U),
2171
2172 // === Scalar f64 (GI-FPU-002 phase 2, #369) ===
2173 // Un-dropped for the DOUBLE-precision FPU target (cortex-m7dp D0..D15);
2174 // the capability gate lives in `select_with_stack`'s preamble — any
2175 // f64 op on m4f/m7 (single-precision) or m0/m3/r5 (no FPU) still
2176 // honest-rejects its function. Exactly this set is lowered by
2177 // `try_lower_f64` + the `F64Load`/`F64Store` selector arms; the rest
2178 // of the f64 surface (min/max/copysign/rounding/i64<->f64/…) stays at
2179 // `_ => None` (loud-skip) until a later increment wires it.
2180 F64Const { value } => Some(WasmOp::F64Const(f64::from_bits(value.bits()))),
2181 F64PromoteF32 => Some(WasmOp::F64PromoteF32),
2182 F64Add => Some(WasmOp::F64Add),
2183 F64Sub => Some(WasmOp::F64Sub),
2184 F64Mul => Some(WasmOp::F64Mul),
2185 F64Div => Some(WasmOp::F64Div),
2186 F64Abs => Some(WasmOp::F64Abs),
2187 F64Neg => Some(WasmOp::F64Neg),
2188 F64Sqrt => Some(WasmOp::F64Sqrt),
2189 F64Eq => Some(WasmOp::F64Eq),
2190 F64Ne => Some(WasmOp::F64Ne),
2191 F64Lt => Some(WasmOp::F64Lt),
2192 F64Le => Some(WasmOp::F64Le),
2193 F64Gt => Some(WasmOp::F64Gt),
2194 F64Ge => Some(WasmOp::F64Ge),
2195 F64Load { memarg } => Some(WasmOp::F64Load {
2196 offset: memarg.offset as u32,
2197 align: memarg.align as u32,
2198 }),
2199 F64Store { memarg } => Some(WasmOp::F64Store {
2200 offset: memarg.offset as u32,
2201 align: memarg.align as u32,
2202 }),
2203 // GI-FPU-002 phase 3 (#369): the f64 op tail — rounding via single
2204 // VRINT{P,M,Z,N}.F64 (FPv5), min/max via VMINNM/VMAXNM + the
2205 // NaN-propagating fix-up, copysign via the VABS/conditional-VNEG
2206 // splice, f32.demote_f64 / i32<->f64 conversions via VCVT
2207 // (i32.trunc_f64_* carries the #709 trap-on-out-of-range domain
2208 // guard). Still m7dp-only (the selector preamble honest-rejects any
2209 // f64 op elsewhere). Remaining dropped f64 surface: the i64<->f64
2210 // conversions and reinterprets (need lowered i64 pair plumbing).
2211 F64Ceil => Some(WasmOp::F64Ceil),
2212 F64Floor => Some(WasmOp::F64Floor),
2213 F64Trunc => Some(WasmOp::F64Trunc),
2214 F64Nearest => Some(WasmOp::F64Nearest),
2215 F64Min => Some(WasmOp::F64Min),
2216 F64Max => Some(WasmOp::F64Max),
2217 F64Copysign => Some(WasmOp::F64Copysign),
2218 F32DemoteF64 => Some(WasmOp::F32DemoteF64),
2219 F64ConvertI32S => Some(WasmOp::F64ConvertI32S),
2220 F64ConvertI32U => Some(WasmOp::F64ConvertI32U),
2221 I32TruncF64S => Some(WasmOp::I32TruncF64S),
2222 I32TruncF64U => Some(WasmOp::I32TruncF64U),
2223
2224 // f32x4
2225 F32x4Add => Some(WasmOp::F32x4Add),
2226 F32x4Sub => Some(WasmOp::F32x4Sub),
2227 F32x4Mul => Some(WasmOp::F32x4Mul),
2228 F32x4Div => Some(WasmOp::F32x4Div),
2229 F32x4Abs => Some(WasmOp::F32x4Abs),
2230 F32x4Neg => Some(WasmOp::F32x4Neg),
2231 F32x4Sqrt => Some(WasmOp::F32x4Sqrt),
2232 F32x4Eq => Some(WasmOp::F32x4Eq),
2233 F32x4Ne => Some(WasmOp::F32x4Ne),
2234 F32x4Lt => Some(WasmOp::F32x4Lt),
2235 F32x4Le => Some(WasmOp::F32x4Le),
2236 F32x4Gt => Some(WasmOp::F32x4Gt),
2237 F32x4Ge => Some(WasmOp::F32x4Ge),
2238 F32x4Splat => Some(WasmOp::F32x4Splat),
2239 F32x4ExtractLane { lane } => Some(WasmOp::F32x4ExtractLane(*lane)),
2240 F32x4ReplaceLane { lane } => Some(WasmOp::F32x4ReplaceLane(*lane)),
2241
2242 // Other operators not yet supported
2243 _ => None,
2244 }
2245}
2246
2247#[cfg(test)]
2248mod tests {
2249 use super::*;
2250
2251 #[test]
2252 fn test_decode_simple_add() {
2253 let wat = r#"
2254 (module
2255 (func (export "add") (param i32 i32) (result i32)
2256 local.get 0
2257 local.get 1
2258 i32.add
2259 )
2260 )
2261 "#;
2262
2263 let wasm = wat::parse_str(wat).expect("Failed to parse WAT");
2264 let functions = decode_wasm_functions(&wasm).expect("Failed to decode");
2265
2266 assert_eq!(functions.len(), 1);
2267 assert_eq!(functions[0].index, 0);
2268 assert_eq!(functions[0].export_name, Some("add".to_string()));
2269 assert_eq!(
2270 functions[0].ops,
2271 vec![
2272 WasmOp::LocalGet(0),
2273 WasmOp::LocalGet(1),
2274 WasmOp::I32Add,
2275 WasmOp::End
2276 ]
2277 );
2278 }
2279
2280 /// #204 regression: `i64.extend_i32_u`, `i64.extend_i32_s` and
2281 /// `i32.wrap_i64` must DECODE (they were previously unmapped → silently
2282 /// dropped by `convert_operator`, leaving an i32 value as a 64-bit operand
2283 /// with a garbage high half — the root cause of gale's miscompiled
2284 /// `(new_count << 32)` pack). The decoder must surface all three.
2285 #[test]
2286 fn test_decode_i64_i32_width_conversions() {
2287 let wat = r#"
2288 (module
2289 (func (export "conv") (param i32 i64) (result i32)
2290 local.get 0
2291 i64.extend_i32_u
2292 local.get 0
2293 i64.extend_i32_s
2294 i64.add
2295 local.get 1
2296 i64.add
2297 i32.wrap_i64
2298 )
2299 )
2300 "#;
2301 let wasm = wat::parse_str(wat).expect("parse");
2302 let functions = decode_wasm_functions(&wasm).expect("decode");
2303 let ops = &functions[0].ops;
2304 assert!(
2305 ops.contains(&WasmOp::I64ExtendI32U),
2306 "i64.extend_i32_u must decode (not be dropped): {ops:?}"
2307 );
2308 assert!(
2309 ops.contains(&WasmOp::I64ExtendI32S),
2310 "i64.extend_i32_s must decode (not be dropped): {ops:?}"
2311 );
2312 assert!(
2313 ops.contains(&WasmOp::I32WrapI64),
2314 "i32.wrap_i64 must decode (not be dropped): {ops:?}"
2315 );
2316 }
2317
2318 /// #204 WAKE-path regression: `br_table` must DECODE (it was unmapped in
2319 /// `convert_operator` → silently dropped, so the selector emitted no index
2320 /// dispatch and every `br_table` fell through to target 0 — gale's binary
2321 /// semaphore never took its WAKE branch). Targets + default are preserved.
2322 #[test]
2323 fn test_decode_br_table() {
2324 let wat = r#"
2325 (module
2326 (func (export "bt") (param i32) (result i32)
2327 (block (block (block
2328 local.get 0
2329 br_table 2 0 1 2)
2330 i32.const 30 return)
2331 i32.const 20 return)
2332 i32.const 10))
2333 "#;
2334 let wasm = wat::parse_str(wat).expect("parse");
2335 let functions = decode_wasm_functions(&wasm).expect("decode");
2336 let bt = functions[0]
2337 .ops
2338 .iter()
2339 .find_map(|o| match o {
2340 WasmOp::BrTable { targets, default } => Some((targets.clone(), *default)),
2341 _ => None,
2342 })
2343 .expect("br_table must decode (not be dropped)");
2344 assert_eq!(bt.0, vec![2, 0, 1], "br_table targets preserved in order");
2345 assert_eq!(bt.1, 2, "br_table default preserved");
2346 }
2347
2348 #[test]
2349 fn test_decode_arithmetic() {
2350 let wat = r#"
2351 (module
2352 (func (export "calc") (result i32)
2353 i32.const 5
2354 i32.const 3
2355 i32.mul
2356 i32.const 2
2357 i32.add
2358 )
2359 )
2360 "#;
2361
2362 let wasm = wat::parse_str(wat).expect("Failed to parse WAT");
2363 let functions = decode_wasm_functions(&wasm).expect("Failed to decode");
2364
2365 assert_eq!(functions.len(), 1);
2366 assert_eq!(functions[0].export_name, Some("calc".to_string()));
2367 assert_eq!(
2368 functions[0].ops,
2369 vec![
2370 WasmOp::I32Const(5),
2371 WasmOp::I32Const(3),
2372 WasmOp::I32Mul,
2373 WasmOp::I32Const(2),
2374 WasmOp::I32Add,
2375 WasmOp::End,
2376 ]
2377 );
2378 }
2379
2380 #[test]
2381 fn test_decode_multi_function_module() {
2382 let wat = r#"
2383 (module
2384 (func $helper)
2385 (func (export "add") (param i32 i32) (result i32)
2386 local.get 0
2387 local.get 1
2388 i32.add
2389 )
2390 (func (export "sub") (param i32 i32) (result i32)
2391 local.get 0
2392 local.get 1
2393 i32.sub
2394 )
2395 )
2396 "#;
2397
2398 let wasm = wat::parse_str(wat).expect("Failed to parse WAT");
2399 let functions = decode_wasm_functions(&wasm).expect("Failed to decode");
2400
2401 assert_eq!(functions.len(), 3);
2402 assert_eq!(functions[0].index, 0);
2403 assert_eq!(functions[0].export_name, None);
2404 assert_eq!(functions[1].index, 1);
2405 assert_eq!(functions[1].export_name, Some("add".to_string()));
2406 assert_eq!(functions[2].index, 2);
2407 assert_eq!(functions[2].export_name, Some("sub".to_string()));
2408 }
2409
2410 #[test]
2411 fn test_decode_module_with_imports() {
2412 let wat = r#"
2413 (module
2414 (import "env" "log" (func $log (param i32)))
2415 (import "env" "memory" (memory 1))
2416 (func (export "run") (param i32)
2417 local.get 0
2418 call 0
2419 )
2420 )
2421 "#;
2422
2423 let wasm = wat::parse_str(wat).expect("Failed to parse WAT");
2424 let module = decode_wasm_module(&wasm).expect("Failed to decode");
2425
2426 // Should have 2 imports (1 func, 1 memory)
2427 assert_eq!(module.imports.len(), 2);
2428 assert_eq!(module.num_imported_funcs, 1);
2429
2430 // First import is the function
2431 assert_eq!(module.imports[0].module, "env");
2432 assert_eq!(module.imports[0].name, "log");
2433 assert!(matches!(module.imports[0].kind, ImportKind::Function(_)));
2434
2435 // Second import is memory
2436 assert_eq!(module.imports[1].module, "env");
2437 assert_eq!(module.imports[1].name, "memory");
2438 assert_eq!(module.imports[1].kind, ImportKind::Memory);
2439
2440 // Should have 1 local function (index 1, because import is index 0)
2441 assert_eq!(module.functions.len(), 1);
2442 assert_eq!(module.functions[0].index, 1);
2443 assert_eq!(module.functions[0].export_name, Some("run".to_string()));
2444 }
2445
2446 #[test]
2447 fn test_find_function_by_export_name() {
2448 let wat = r#"
2449 (module
2450 (func $helper)
2451 (func (export "add") (param i32 i32) (result i32)
2452 local.get 0
2453 local.get 1
2454 i32.add
2455 )
2456 )
2457 "#;
2458
2459 let wasm = wat::parse_str(wat).expect("Failed to parse WAT");
2460 let functions = decode_wasm_functions(&wasm).expect("Failed to decode");
2461
2462 let add_func = functions
2463 .iter()
2464 .find(|f| f.export_name.as_deref() == Some("add"))
2465 .expect("Should find 'add' function");
2466
2467 assert_eq!(add_func.index, 1);
2468 assert!(add_func.ops.contains(&WasmOp::I32Add));
2469 }
2470
2471 #[test]
2472 fn test_decode_subword_loads() {
2473 let wat = r#"
2474 (module
2475 (memory 1)
2476 (func (export "test") (param i32) (result i32)
2477 local.get 0
2478 i32.load8_u
2479 )
2480 )
2481 "#;
2482
2483 let wasm = wat::parse_str(wat).expect("Failed to parse WAT");
2484 let functions = decode_wasm_functions(&wasm).expect("Failed to decode");
2485
2486 assert_eq!(functions.len(), 1);
2487 assert!(functions[0].ops.contains(&WasmOp::I32Load8U {
2488 offset: 0,
2489 align: 0,
2490 }));
2491 }
2492
2493 #[test]
2494 fn test_decode_subword_stores() {
2495 let wat = r#"
2496 (module
2497 (memory 1)
2498 (func (export "test") (param i32 i32)
2499 local.get 0
2500 local.get 1
2501 i32.store8
2502 )
2503 )
2504 "#;
2505
2506 let wasm = wat::parse_str(wat).expect("Failed to parse WAT");
2507 let functions = decode_wasm_functions(&wasm).expect("Failed to decode");
2508
2509 assert_eq!(functions.len(), 1);
2510 assert!(functions[0].ops.contains(&WasmOp::I32Store8 {
2511 offset: 0,
2512 align: 0,
2513 }));
2514 }
2515
2516 #[test]
2517 fn test_decode_memory_size_grow() {
2518 let wat = r#"
2519 (module
2520 (memory 1)
2521 (func (export "test") (result i32)
2522 memory.size
2523 )
2524 )
2525 "#;
2526
2527 let wasm = wat::parse_str(wat).expect("Failed to parse WAT");
2528 let functions = decode_wasm_functions(&wasm).expect("Failed to decode");
2529
2530 assert_eq!(functions.len(), 1);
2531 assert!(functions[0].ops.contains(&WasmOp::MemorySize(0)));
2532 }
2533
2534 #[test]
2535 fn test_decode_memory_grow() {
2536 let wat = r#"
2537 (module
2538 (memory 1)
2539 (func (export "test") (param i32) (result i32)
2540 local.get 0
2541 memory.grow
2542 )
2543 )
2544 "#;
2545
2546 let wasm = wat::parse_str(wat).expect("Failed to parse WAT");
2547 let functions = decode_wasm_functions(&wasm).expect("Failed to decode");
2548
2549 assert_eq!(functions.len(), 1);
2550 assert!(functions[0].ops.contains(&WasmOp::MemoryGrow(0)));
2551 }
2552
2553 #[test]
2554 fn test_decode_bulk_memory_374() {
2555 // #374: memory.copy / memory.fill on the single linear memory decode to
2556 // the new WasmOp variants (was `_ => None` -> loud-skip).
2557 let wat = r#"
2558 (module
2559 (memory 1)
2560 (func (export "cpy") (param i32 i32 i32)
2561 local.get 0 local.get 1 local.get 2 memory.copy)
2562 (func (export "fil") (param i32 i32 i32)
2563 local.get 0 local.get 1 local.get 2 memory.fill)
2564 )
2565 "#;
2566 let wasm = wat::parse_str(wat).expect("Failed to parse WAT");
2567 let functions = decode_wasm_functions(&wasm).expect("Failed to decode");
2568 assert_eq!(functions.len(), 2);
2569 assert!(functions[0].ops.contains(&WasmOp::MemoryCopy));
2570 assert!(functions[1].ops.contains(&WasmOp::MemoryFill));
2571 // Neither function is flagged unsupported (they now lower).
2572 assert!(functions[0].unsupported.is_none());
2573 assert!(functions[1].unsupported.is_none());
2574 }
2575
2576 #[test]
2577 fn test_decode_i64_subword_loads() {
2578 let wat = r#"
2579 (module
2580 (memory 1)
2581 (func (export "test") (param i32) (result i64)
2582 local.get 0
2583 i64.load8_s
2584 )
2585 )
2586 "#;
2587
2588 let wasm = wat::parse_str(wat).expect("Failed to parse WAT");
2589 let functions = decode_wasm_functions(&wasm).expect("Failed to decode");
2590
2591 assert_eq!(functions.len(), 1);
2592 assert!(functions[0].ops.contains(&WasmOp::I64Load8S {
2593 offset: 0,
2594 align: 0,
2595 }));
2596 }
2597
2598 #[test]
2599 fn test_decode_all_subword_memory_ops() {
2600 // Test that all sub-word operations are decoded from WAT
2601 let wat = r#"
2602 (module
2603 (memory 1)
2604 (func (export "test") (param i32)
2605 ;; i32 sub-word loads
2606 local.get 0
2607 i32.load8_s
2608 drop
2609 local.get 0
2610 i32.load8_u
2611 drop
2612 local.get 0
2613 i32.load16_s
2614 drop
2615 local.get 0
2616 i32.load16_u
2617 drop
2618
2619 ;; i32 sub-word stores
2620 local.get 0
2621 i32.const 42
2622 i32.store8
2623 local.get 0
2624 i32.const 42
2625 i32.store16
2626
2627 ;; i64 sub-word loads
2628 local.get 0
2629 i64.load8_s
2630 drop
2631 local.get 0
2632 i64.load8_u
2633 drop
2634 local.get 0
2635 i64.load16_s
2636 drop
2637 local.get 0
2638 i64.load16_u
2639 drop
2640 local.get 0
2641 i64.load32_s
2642 drop
2643 local.get 0
2644 i64.load32_u
2645 drop
2646
2647 ;; i64 sub-word stores
2648 local.get 0
2649 i64.const 42
2650 i64.store8
2651 local.get 0
2652 i64.const 42
2653 i64.store16
2654 local.get 0
2655 i64.const 42
2656 i64.store32
2657 )
2658 )
2659 "#;
2660
2661 let wasm = wat::parse_str(wat).expect("Failed to parse WAT");
2662 let functions = decode_wasm_functions(&wasm).expect("Failed to decode");
2663
2664 assert_eq!(functions.len(), 1);
2665 let ops = &functions[0].ops;
2666
2667 // Verify i32 sub-word ops are present
2668 assert!(ops.iter().any(|o| matches!(o, WasmOp::I32Load8S { .. })));
2669 assert!(ops.iter().any(|o| matches!(o, WasmOp::I32Load8U { .. })));
2670 assert!(ops.iter().any(|o| matches!(o, WasmOp::I32Load16S { .. })));
2671 assert!(ops.iter().any(|o| matches!(o, WasmOp::I32Load16U { .. })));
2672 assert!(ops.iter().any(|o| matches!(o, WasmOp::I32Store8 { .. })));
2673 assert!(ops.iter().any(|o| matches!(o, WasmOp::I32Store16 { .. })));
2674
2675 // Verify i64 sub-word ops are present
2676 assert!(ops.iter().any(|o| matches!(o, WasmOp::I64Load8S { .. })));
2677 assert!(ops.iter().any(|o| matches!(o, WasmOp::I64Load8U { .. })));
2678 assert!(ops.iter().any(|o| matches!(o, WasmOp::I64Load16S { .. })));
2679 assert!(ops.iter().any(|o| matches!(o, WasmOp::I64Load16U { .. })));
2680 assert!(ops.iter().any(|o| matches!(o, WasmOp::I64Load32S { .. })));
2681 assert!(ops.iter().any(|o| matches!(o, WasmOp::I64Load32U { .. })));
2682 assert!(ops.iter().any(|o| matches!(o, WasmOp::I64Store8 { .. })));
2683 assert!(ops.iter().any(|o| matches!(o, WasmOp::I64Store16 { .. })));
2684 assert!(ops.iter().any(|o| matches!(o, WasmOp::I64Store32 { .. })));
2685 }
2686
2687 #[test]
2688 fn test_decode_simd_i32x4_add() {
2689 let wat = r#"
2690 (module
2691 (func (export "add_v128") (param v128 v128) (result v128)
2692 local.get 0
2693 local.get 1
2694 i32x4.add
2695 )
2696 )
2697 "#;
2698
2699 let wasm = wat::parse_str(wat).expect("Failed to parse WAT with SIMD");
2700 let functions = decode_wasm_functions(&wasm).expect("Failed to decode");
2701
2702 assert_eq!(functions.len(), 1);
2703 assert!(
2704 functions[0].ops.contains(&WasmOp::I32x4Add),
2705 "Should decode i32x4.add: {:?}",
2706 functions[0].ops
2707 );
2708 }
2709
2710 #[test]
2711 fn test_decode_simd_v128_const() {
2712 let wat = r#"
2713 (module
2714 (func (export "const_v128") (result v128)
2715 v128.const i32x4 1 2 3 4
2716 )
2717 )
2718 "#;
2719
2720 let wasm = wat::parse_str(wat).expect("Failed to parse WAT with SIMD");
2721 let functions = decode_wasm_functions(&wasm).expect("Failed to decode");
2722
2723 assert_eq!(functions.len(), 1);
2724 assert!(
2725 functions[0]
2726 .ops
2727 .iter()
2728 .any(|o| matches!(o, WasmOp::V128Const(_))),
2729 "Should decode v128.const: {:?}",
2730 functions[0].ops
2731 );
2732 }
2733
2734 #[test]
2735 fn test_decode_simd_v128_load_store() {
2736 let wat = r#"
2737 (module
2738 (memory 1)
2739 (func (export "load_store") (param i32)
2740 local.get 0
2741 v128.load
2742 local.get 0
2743 v128.store
2744 )
2745 )
2746 "#;
2747
2748 let wasm = wat::parse_str(wat).expect("Failed to parse WAT with SIMD");
2749 let functions = decode_wasm_functions(&wasm).expect("Failed to decode");
2750
2751 assert_eq!(functions.len(), 1);
2752 let ops = &functions[0].ops;
2753 assert!(
2754 ops.iter().any(|o| matches!(o, WasmOp::V128Load { .. })),
2755 "Should decode v128.load"
2756 );
2757 assert!(
2758 ops.iter().any(|o| matches!(o, WasmOp::V128Store { .. })),
2759 "Should decode v128.store"
2760 );
2761 }
2762
2763 #[test]
2764 fn test_decode_simd_bitwise_ops() {
2765 let wat = r#"
2766 (module
2767 (func (export "bitwise") (param v128 v128) (result v128)
2768 local.get 0
2769 local.get 1
2770 v128.and
2771 )
2772 )
2773 "#;
2774
2775 let wasm = wat::parse_str(wat).expect("Failed to parse WAT with SIMD");
2776 let functions = decode_wasm_functions(&wasm).expect("Failed to decode");
2777
2778 assert_eq!(functions.len(), 1);
2779 assert!(functions[0].ops.contains(&WasmOp::V128And));
2780 }
2781
2782 #[test]
2783 fn test_decode_simd_splat() {
2784 let wat = r#"
2785 (module
2786 (func (export "splat") (param i32) (result v128)
2787 local.get 0
2788 i32x4.splat
2789 )
2790 )
2791 "#;
2792
2793 let wasm = wat::parse_str(wat).expect("Failed to parse WAT with SIMD");
2794 let functions = decode_wasm_functions(&wasm).expect("Failed to decode");
2795
2796 assert_eq!(functions.len(), 1);
2797 assert!(functions[0].ops.contains(&WasmOp::I32x4Splat));
2798 }
2799
2800 #[test]
2801 fn test_decode_simd_extract_lane() {
2802 let wat = r#"
2803 (module
2804 (func (export "extract") (param v128) (result i32)
2805 local.get 0
2806 i32x4.extract_lane 2
2807 )
2808 )
2809 "#;
2810
2811 let wasm = wat::parse_str(wat).expect("Failed to parse WAT with SIMD");
2812 let functions = decode_wasm_functions(&wasm).expect("Failed to decode");
2813
2814 assert_eq!(functions.len(), 1);
2815 assert!(
2816 functions[0].ops.contains(&WasmOp::I32x4ExtractLane(2)),
2817 "Should decode i32x4.extract_lane 2"
2818 );
2819 }
2820
2821 #[test]
2822 fn test_decode_simd_f32x4_arithmetic() {
2823 let wat = r#"
2824 (module
2825 (func (export "f32x4_add") (param v128 v128) (result v128)
2826 local.get 0
2827 local.get 1
2828 f32x4.add
2829 )
2830 )
2831 "#;
2832
2833 let wasm = wat::parse_str(wat).expect("Failed to parse WAT with SIMD");
2834 let functions = decode_wasm_functions(&wasm).expect("Failed to decode");
2835
2836 assert_eq!(functions.len(), 1);
2837 assert!(functions[0].ops.contains(&WasmOp::F32x4Add));
2838 }
2839
2840 #[test]
2841 fn test_369_scalar_float_op_flags_function_unsupported_not_dropped() {
2842 // GI-FPU-002 (#619): the in-scope scalar f32 ops (add/sub/mul/div,
2843 // comparisons, i32.trunc_f32_s/u, f32.convert_i32_s/u, f32.const) are
2844 // now DECODED (routed to the VFP selector on FPU targets), so `f32.add`
2845 // is no longer flagged — and since phase 2 (#369) so is the lowered
2846 // f64 subset (`f64.add` here; the m7dp-only capability gate lives in
2847 // the selector). Since phase 3 the f64 op TAIL (`f64.min` here) is
2848 // decoded too; an out-of-scope scalar float op (`i64.trunc_f64_s` —
2849 // the i64<->f64 conversions need lowered pair plumbing) is STILL
2850 // flagged (loud-skip), never silently dropped — the #369 honesty
2851 // contract holds for the not-yet-lowered surface. A pure-integer
2852 // function stays clean.
2853 let wat = r#"
2854 (module
2855 (func (export "fadd") (param f32 f32) (result f32)
2856 local.get 0 local.get 1 f32.add)
2857 (func (export "dadd") (param f64 f64) (result f64)
2858 local.get 0 local.get 1 f64.add)
2859 (func (export "dmin") (param f64 f64) (result f64)
2860 local.get 0 local.get 1 f64.min)
2861 (func (export "dtrunc64") (param f64) (result i64)
2862 local.get 0 i64.trunc_f64_s)
2863 (func (export "iadd") (param i32 i32) (result i32)
2864 local.get 0 local.get 1 i32.add))
2865 "#;
2866 let wasm = wat::parse_str(wat).expect("parse");
2867 let functions = decode_wasm_functions(&wasm).expect("decode");
2868 let fadd = functions
2869 .iter()
2870 .find(|f| f.export_name.as_deref() == Some("fadd"))
2871 .unwrap();
2872 let dadd = functions
2873 .iter()
2874 .find(|f| f.export_name.as_deref() == Some("dadd"))
2875 .unwrap();
2876 let dmin = functions
2877 .iter()
2878 .find(|f| f.export_name.as_deref() == Some("dmin"))
2879 .unwrap();
2880 let dtrunc64 = functions
2881 .iter()
2882 .find(|f| f.export_name.as_deref() == Some("dtrunc64"))
2883 .unwrap();
2884 let iadd = functions
2885 .iter()
2886 .find(|f| f.export_name.as_deref() == Some("iadd"))
2887 .unwrap();
2888 // In-scope f32 op: now decoded (reachable), not flagged.
2889 assert!(
2890 fadd.unsupported.is_none(),
2891 "GI-FPU-002: f32.add must now decode (not be flagged), got {:?}",
2892 fadd.unsupported
2893 );
2894 assert!(
2895 fadd.ops.contains(&WasmOp::F32Add),
2896 "f32.add must decode to WasmOp::F32Add: {:?}",
2897 fadd.ops
2898 );
2899 // In-scope f64 op (phase 2, #369): now decoded, not flagged.
2900 assert!(
2901 dadd.unsupported.is_none(),
2902 "GI-FPU-002 phase 2: f64.add must now decode (not be flagged), got {:?}",
2903 dadd.unsupported
2904 );
2905 assert!(
2906 dadd.ops.contains(&WasmOp::F64Add),
2907 "f64.add must decode to WasmOp::F64Add: {:?}",
2908 dadd.ops
2909 );
2910 // In-scope f64 tail op (phase 3, #369): now decoded, not flagged.
2911 assert!(
2912 dmin.unsupported.is_none(),
2913 "GI-FPU-002 phase 3: f64.min must now decode (not be flagged), got {:?}",
2914 dmin.unsupported
2915 );
2916 assert!(
2917 dmin.ops.contains(&WasmOp::F64Min),
2918 "f64.min must decode to WasmOp::F64Min: {:?}",
2919 dmin.ops
2920 );
2921 // Out-of-scope scalar double op: still flagged, never dropped.
2922 assert!(
2923 dtrunc64.unsupported.is_some(),
2924 "i64.trunc_f64_s must still flag the function unsupported (out of scope), got {:?}",
2925 dtrunc64.unsupported
2926 );
2927 assert!(
2928 iadd.unsupported.is_none(),
2929 "a pure-integer function must NOT be flagged: {:?}",
2930 iadd.unsupported
2931 );
2932 }
2933
2934 #[test]
2935 fn test_369_float_global_access_flags_function_unsupported() {
2936 // GI-FPU-001 (#369): `global.get`/`global.set` on an f32/f64-typed
2937 // global decode fine (the ops are type-agnostic), but the float
2938 // initializer is dropped (`init_i32: None` -> slot zeroed), so a read
2939 // returned a silently-wrong 0.0 instead of the init (verified: the
2940 // 2.5f bit pattern 0x40200000 was absent from the output ELF). The
2941 // access must flag the function for the loud-skip path. Accesses to
2942 // integer globals stay clean.
2943 let wat = r#"
2944 (module
2945 (global $fg f32 (f32.const 2.5))
2946 (global $dg (mut f64) (f64.const 1.5))
2947 (global $ig (mut i32) (i32.const 7))
2948 (func (export "fget") (result f32) global.get $fg)
2949 (func (export "dset") (param f64) local.get 0 global.set $dg)
2950 (func (export "iget") (result i32) global.get $ig))
2951 "#;
2952 let wasm = wat::parse_str(wat).expect("parse");
2953
2954 // Both decode entry points must flag (the CLI compiles through both:
2955 // decode_wasm_module on the all-exports/module paths,
2956 // decode_wasm_functions on the single-function path).
2957 let module = decode_wasm_module(&wasm).expect("decode module");
2958 for functions in [
2959 &module.functions,
2960 &decode_wasm_functions(&wasm).expect("decode fns"),
2961 ] {
2962 let by_name = |n: &str| {
2963 functions
2964 .iter()
2965 .find(|f| f.export_name.as_deref() == Some(n))
2966 .unwrap()
2967 };
2968 let fget = by_name("fget");
2969 assert!(
2970 fget.unsupported.is_some(),
2971 "global.get of an f32 global must flag the function (loud-skip), got {:?}",
2972 fget.unsupported
2973 );
2974 let reason = fget.unsupported.as_deref().unwrap();
2975 assert!(
2976 reason.contains("GlobalGet") && reason.contains("GI-FPU-001"),
2977 "diagnostic should name the op and GI-FPU-001: {reason:?}"
2978 );
2979 let dset = by_name("dset");
2980 assert!(
2981 dset.unsupported
2982 .as_deref()
2983 .is_some_and(|r| r.contains("GlobalSet")),
2984 "global.set of an f64 global must flag the function, got {:?}",
2985 dset.unsupported
2986 );
2987 assert!(
2988 by_name("iget").unsupported.is_none(),
2989 "an i32 global access must NOT be flagged: {:?}",
2990 by_name("iget").unsupported
2991 );
2992 }
2993 }
2994
2995 #[test]
2996 fn test_369_imported_float_global_shifts_index_space() {
2997 // GI-FPU-001 (#369): imported globals come FIRST in the global index
2998 // space. An imported f64 global at index 0 must be flagged, and the
2999 // defined i32 global at index 1 must NOT be mistaken for it.
3000 let wat = r#"
3001 (module
3002 (import "env" "fg" (global f64))
3003 (global $ig i32 (i32.const 3))
3004 (func (export "fget") (result f64) global.get 0)
3005 (func (export "iget") (result i32) global.get 1))
3006 "#;
3007 let wasm = wat::parse_str(wat).expect("parse");
3008 let functions = decode_wasm_functions(&wasm).expect("decode");
3009 let by_name = |n: &str| {
3010 functions
3011 .iter()
3012 .find(|f| f.export_name.as_deref() == Some(n))
3013 .unwrap()
3014 };
3015 assert!(
3016 by_name("fget")
3017 .unsupported
3018 .as_deref()
3019 .is_some_and(|r| r.contains("GI-FPU-001")),
3020 "imported f64 global access must flag: {:?}",
3021 by_name("fget").unsupported
3022 );
3023 assert!(
3024 by_name("iget").unsupported.is_none(),
3025 "defined i32 global at shifted index 1 must NOT flag: {:?}",
3026 by_name("iget").unsupported
3027 );
3028 }
3029
3030 #[test]
3031 fn test_680_simd_ops_flag_function_unsupported_not_dropped() {
3032 // #680: SIMD (v128) ops decode into WasmOp variants no production
3033 // target can select (`has_helium` is test-only), so they were silently
3034 // dropped at selection — `i32x4.add` compiled to an operand
3035 // passthrough (`mov r0,r1`) and shipped a wrong result. The issue's
3036 // exact reproducer must flag the function; the scalar sibling must
3037 // stay compilable (non-vacuity).
3038 let wat = r#"
3039 (module
3040 (memory 1)
3041 (func (export "vadd") (param i32 i32) (result i32)
3042 (i32x4.extract_lane 2
3043 (i32x4.add (i32x4.splat (local.get 0))
3044 (i32x4.splat (local.get 1)))))
3045 (func (export "vstore") (param i32 i32) (result i32)
3046 (v128.store (i32.const 0) (i32x4.splat (local.get 0)))
3047 (i32.load (i32.const 0)))
3048 (func (export "iadd") (param i32 i32) (result i32)
3049 local.get 0 local.get 1 i32.add))
3050 "#;
3051 let wasm = wat::parse_str(wat).expect("parse");
3052
3053 // Both decode entry points must flag (the CLI compiles through both).
3054 let module = decode_wasm_module(&wasm).expect("decode module");
3055 for functions in [
3056 &module.functions,
3057 &decode_wasm_functions(&wasm).expect("decode fns"),
3058 ] {
3059 let by_name = |n: &str| {
3060 functions
3061 .iter()
3062 .find(|f| f.export_name.as_deref() == Some(n))
3063 .unwrap()
3064 };
3065 for name in ["vadd", "vstore"] {
3066 let reason = by_name(name).unsupported.as_deref();
3067 assert!(
3068 reason.is_some(),
3069 "{name}: v128 ops must flag the function (loud-skip), got None"
3070 );
3071 let reason = reason.unwrap();
3072 assert!(
3073 reason.contains("no SIMD lowering for this target") && reason.contains("#680"),
3074 "{name}: diagnostic must name the target gap and #680: {reason:?}"
3075 );
3076 }
3077 // The reason names the FIRST SIMD op hit (splat in both bodies).
3078 assert!(
3079 by_name("vadd")
3080 .unsupported
3081 .as_deref()
3082 .unwrap()
3083 .contains("I32x4Splat"),
3084 "diagnostic should name the op: {:?}",
3085 by_name("vadd").unsupported
3086 );
3087 assert!(
3088 by_name("iadd").unsupported.is_none(),
3089 "a scalar function in the same module must NOT be flagged: {:?}",
3090 by_name("iadd").unsupported
3091 );
3092 }
3093 }
3094
3095 #[test]
3096 fn test_680_v128_local_and_signature_flag_function() {
3097 // #680: v128 VALUES are expressible with ZERO SIMD-proposal operators
3098 // in the body — a v128-typed local or a v128 param/result is reached
3099 // through type-agnostic `local.get`/`local.set`, which the selectors
3100 // lower as 4-byte moves (silent 16-byte truncation). Both must flag.
3101 let wat = r#"
3102 (module
3103 (func (export "vlocal") (result i32) (local v128)
3104 i32.const 7)
3105 (func (export "vpass") (param v128) (result v128)
3106 local.get 0)
3107 (func (export "scalar") (param i32) (result i32)
3108 local.get 0))
3109 "#;
3110 let wasm = wat::parse_str(wat).expect("parse");
3111 let module = decode_wasm_module(&wasm).expect("decode module");
3112 for functions in [
3113 &module.functions,
3114 &decode_wasm_functions(&wasm).expect("decode fns"),
3115 ] {
3116 let by_name = |n: &str| {
3117 functions
3118 .iter()
3119 .find(|f| f.export_name.as_deref() == Some(n))
3120 .unwrap()
3121 };
3122 assert!(
3123 by_name("vlocal")
3124 .unsupported
3125 .as_deref()
3126 .is_some_and(|r| r.contains("v128-typed local") && r.contains("#680")),
3127 "a v128-typed local declaration must flag: {:?}",
3128 by_name("vlocal").unsupported
3129 );
3130 assert!(
3131 by_name("vpass")
3132 .unsupported
3133 .as_deref()
3134 .is_some_and(|r| r.contains("v128 param/result") && r.contains("#680")),
3135 "a v128 param/result signature must flag (op-free body!): {:?}",
3136 by_name("vpass").unsupported
3137 );
3138 assert!(
3139 by_name("scalar").unsupported.is_none(),
3140 "a scalar function must NOT be flagged: {:?}",
3141 by_name("scalar").unsupported
3142 );
3143 }
3144 }
3145
3146 #[test]
3147 fn test_680_v128_global_access_flags_function() {
3148 // #680: `global.get`/`global.set` on a v128-typed global decode fine
3149 // (type-agnostic ops), but the access would move 4 of the 16 bytes and
3150 // the `v128.const` initializer is never captured. Same lane as the
3151 // float globals (#648/GI-FPU-001); imported globals shift the index
3152 // space (imports first). The i32-global sibling stays compilable.
3153 let wat = r#"
3154 (module
3155 (import "env" "vg" (global v128))
3156 (global $ig (mut i32) (i32.const 7))
3157 (global $dg (mut v128) (v128.const i32x4 1 2 3 4))
3158 (func (export "vget") global.get 0 drop)
3159 (func (export "iget") (result i32) global.get $ig))
3160 "#;
3161 let wasm = wat::parse_str(wat).expect("parse");
3162 let module = decode_wasm_module(&wasm).expect("decode module");
3163 for functions in [
3164 &module.functions,
3165 &decode_wasm_functions(&wasm).expect("decode fns"),
3166 ] {
3167 let by_name = |n: &str| {
3168 functions
3169 .iter()
3170 .find(|f| f.export_name.as_deref() == Some(n))
3171 .unwrap()
3172 };
3173 let reason = by_name("vget").unsupported.as_deref();
3174 assert!(
3175 reason.is_some_and(|r| r.contains("GlobalGet")
3176 && r.contains("v128-typed global")
3177 && r.contains("#680")),
3178 "global.get of an imported v128 global must flag: {reason:?}"
3179 );
3180 assert!(
3181 by_name("iget").unsupported.is_none(),
3182 "an i32 global access must NOT be flagged: {:?}",
3183 by_name("iget").unsupported
3184 );
3185 }
3186 }
3187
3188 #[test]
3189 fn test_decode_simd_multiple_ops() {
3190 let wat = r#"
3191 (module
3192 (func (export "simd_ops") (param v128 v128 v128) (result v128)
3193 ;; (a + b) * c
3194 local.get 0
3195 local.get 1
3196 i32x4.add
3197 local.get 2
3198 i32x4.mul
3199 )
3200 )
3201 "#;
3202
3203 let wasm = wat::parse_str(wat).expect("Failed to parse WAT with SIMD");
3204 let functions = decode_wasm_functions(&wasm).expect("Failed to decode");
3205
3206 assert_eq!(functions.len(), 1);
3207 let ops = &functions[0].ops;
3208 assert!(ops.contains(&WasmOp::I32x4Add));
3209 assert!(ops.contains(&WasmOp::I32x4Mul));
3210 }
3211
3212 /// VCR-DBG-001 step 1 (#394): the decoder records a module-relative wasm byte
3213 /// offset per emitted op — the DWARF-for-wasm address space that bridges
3214 /// synth's op-index `source_line` to the input wasm's `.debug_line`. Purely
3215 /// additive metadata (no codegen consumer ⇒ frozen fixtures byte-identical,
3216 /// verified separately); this test pins the structural invariants.
3217 #[test]
3218 fn test_decode_records_aligned_increasing_op_offsets_dbg001() {
3219 let wat = r#"
3220 (module
3221 (func (export "f") (param i32 i32) (result i32)
3222 local.get 0
3223 local.get 1
3224 i32.add
3225 i32.const 7
3226 i32.mul))
3227 "#;
3228 let wasm = wat::parse_str(wat).expect("parse WAT");
3229 let functions = decode_wasm_functions(&wasm).expect("decode");
3230 let f = &functions[0];
3231
3232 // One offset per emitted op, index-aligned with `ops`.
3233 assert_eq!(
3234 f.op_offsets.len(),
3235 f.ops.len(),
3236 "op_offsets must be parallel to ops"
3237 );
3238 assert!(!f.op_offsets.is_empty());
3239
3240 // Byte offsets are strictly increasing through the body (each op consumes
3241 // at least one byte) and module-relative (well past the header).
3242 assert!(
3243 f.op_offsets.windows(2).all(|w| w[1] > w[0]),
3244 "wasm byte offsets must strictly increase: {:?}",
3245 f.op_offsets
3246 );
3247 assert!(
3248 f.op_offsets[0] >= 8,
3249 "module-relative offset is past the 8-byte wasm header"
3250 );
3251 }
3252
3253 /// #237: the decoder captures a global's `i32.const` initializer + mutability,
3254 /// so the native-pointer ABI can recognize the stack-pointer global.
3255 #[test]
3256 fn test_decode_captures_global_initializer() {
3257 let wat = r#"
3258 (module
3259 (memory 2)
3260 (global $__stack_pointer (mut i32) (i32.const 65536))
3261 (global $immutable_const i32 (i32.const 7))
3262 (func (export "f") (result i32) global.get 0)
3263 )
3264 "#;
3265 let wasm = wat::parse_str(wat).expect("Failed to parse WAT");
3266 let module = decode_wasm_module(&wasm).expect("Failed to decode");
3267
3268 assert_eq!(module.globals.len(), 2, "both globals captured");
3269 let sp = &module.globals[0];
3270 assert_eq!(sp.index, 0);
3271 assert_eq!(
3272 sp.init,
3273 Some(GlobalInit::I32(65536)),
3274 "stack-pointer init captured"
3275 );
3276 assert!(sp.mutable, "stack pointer is mutable");
3277 let c = &module.globals[1];
3278 assert_eq!(c.init, Some(GlobalInit::I32(7)));
3279 assert!(!c.mutable, "second global is immutable");
3280 assert_eq!(sp.slot_bytes, 4, "i32 global occupies one 4-byte slot");
3281 assert_eq!(c.slot_bytes, 4);
3282 }
3283
3284 /// #643: the decoder records the DECLARED slot width per global — an i64
3285 /// (or f64) global occupies 8 bytes, so the globals-table layout can give
3286 /// it room for both words and shift every later global's offset.
3287 #[test]
3288 fn test_decode_records_global_slot_widths_643() {
3289 let wat = r#"
3290 (module
3291 (global $c (mut i64) (i64.const 0))
3292 (global $k (mut i32) (i32.const 0))
3293 (global $f (mut f64) (f64.const 0))
3294 (func (export "f") (result i32) global.get 1)
3295 )
3296 "#;
3297 let wasm = wat::parse_str(wat).expect("Failed to parse WAT");
3298 let module = decode_wasm_module(&wasm).expect("Failed to decode");
3299
3300 assert_eq!(module.globals.len(), 3);
3301 assert_eq!(module.globals[0].slot_bytes, 8, "i64 global is 8 bytes");
3302 assert_eq!(module.globals[1].slot_bytes, 4, "i32 global is 4 bytes");
3303 assert_eq!(module.globals[2].slot_bytes, 8, "f64 global is 8 bytes");
3304 }
3305
3306 /// #649: a nonzero `i64.const` initializer is captured as BOTH words — the
3307 /// `init_i32`-shaped capture dropped it to `None` and every consumer's
3308 /// `unwrap_or(0)` silently ZEROED the global. f32/f64 inits stay `None`
3309 /// (GI-FPU-001/#369 loud-skip lane — never fabricate a float bit-pattern).
3310 #[test]
3311 fn test_decode_captures_i64_global_initializer_649() {
3312 let wat = r#"
3313 (module
3314 (global $g (mut i64) (i64.const 0x123456789ABCDEF0))
3315 (global $n (mut i64) (i64.const -1))
3316 (global $f (mut f64) (f64.const 1.5))
3317 (global $h (mut f32) (f32.const 2.5))
3318 (func (export "f") (result i32) i32.const 0)
3319 )
3320 "#;
3321 let wasm = wat::parse_str(wat).expect("Failed to parse WAT");
3322 let module = decode_wasm_module(&wasm).expect("Failed to decode");
3323
3324 assert_eq!(module.globals.len(), 4);
3325 assert_eq!(
3326 module.globals[0].init,
3327 Some(GlobalInit::I64(0x123456789ABCDEF0u64 as i64)),
3328 "nonzero i64 init captured with both words"
3329 );
3330 assert_eq!(module.globals[1].init, Some(GlobalInit::I64(-1)));
3331 assert_eq!(
3332 module.globals[2].init, None,
3333 "f64 init is NOT captured (GI-FPU-001 loud-skip lane)"
3334 );
3335 assert_eq!(
3336 module.globals[3].init, None,
3337 "f32 init is NOT captured (GI-FPU-001 loud-skip lane)"
3338 );
3339 }
3340
3341 /// #509: the decoder records `(param_count, result_count)` for every
3342 /// `Block`/`Loop`/`If`, ordinal-keyed in op order, covering all three
3343 /// blocktype encodings: `Empty → (0,0)`, `ValType → (0,1)`, and
3344 /// `FuncType(i) →` counts from the type section (here a multi-result
3345 /// block, which wat encodes as a functype blocktype).
3346 #[test]
3347 fn test_decode_records_block_arity_side_table_509() {
3348 let wat = r#"
3349 (module
3350 (func (export "f") (param i32) (result i32)
3351 (block (result i32)
3352 (block (nop))
3353 (local.get 0)
3354 (if (result i32)
3355 (then (i32.const 1))
3356 (else (i32.const 2)))))
3357 (func (export "g") (result i32)
3358 (block (result i32 i32)
3359 (i32.const 1) (i32.const 2))
3360 i32.add)
3361 (func (export "h") (param i32) (result i32)
3362 (local.get 0)
3363 (loop (param i32) (result i32))))
3364 "#;
3365 let wasm = wat::parse_str(wat).expect("parse WAT");
3366
3367 // Both decode entry points must produce the same side-table.
3368 for functions in [
3369 decode_wasm_functions(&wasm).expect("decode"),
3370 decode_wasm_module(&wasm).expect("decode").functions,
3371 ] {
3372 // f: Block(result i32), Block(void), If(result i32) — in op order.
3373 assert_eq!(
3374 functions[0].block_arity,
3375 vec![(0, 1), (0, 0), (0, 1)],
3376 "f: ValType/Empty/ValType blocktypes"
3377 );
3378 // g: one multi-result block via a FuncType blocktype.
3379 assert_eq!(
3380 functions[1].block_arity,
3381 vec![(0, 2)],
3382 "g: functype blocktype result count from the type section"
3383 );
3384 // h: a parameterized loop — the input arity is what a br to the
3385 // header would carry (the #509 loud-decline discriminator).
3386 assert_eq!(
3387 functions[2].block_arity,
3388 vec![(1, 1)],
3389 "h: loop params captured"
3390 );
3391 }
3392 }
3393
3394 /// #642: the decoder captures table 0's compile-time size, per-segment
3395 /// element shapes and per-function type indices, and the closed-world
3396 /// verdict VERIFIES a fully-covered homogeneous table.
3397 #[test]
3398 fn test_call_indirect_guards_closed_world_verified_642() {
3399 // The #642 repro shape: 3-entry table, fully covered, one signature.
3400 let wat = r#"
3401 (module
3402 (type $bin (func (param i32 i32) (result i32)))
3403 (table 3 funcref)
3404 (elem (i32.const 0) $add $sub $mul)
3405 (func $add (param i32 i32) (result i32)
3406 (i32.add (local.get 0) (local.get 1)))
3407 (func $sub (param i32 i32) (result i32)
3408 (i32.sub (local.get 0) (local.get 1)))
3409 (func $mul (param i32 i32) (result i32)
3410 (i32.mul (local.get 0) (local.get 1)))
3411 (func (export "f") (param i32 i32) (result i32)
3412 (call_indirect (type $bin)
3413 (local.get 0) (i32.const 10) (local.get 1)))
3414 )
3415 "#;
3416 let wasm = wat::parse_str(wat).expect("parse");
3417 let module = decode_wasm_module(&wasm).expect("decode");
3418
3419 assert_eq!(module.table_size, Some(3), "table section min size");
3420 assert_eq!(module.table_sizes, vec![Some(3)], "#650 per-table sizes");
3421 assert_eq!(
3422 module.elem_segments,
3423 vec![ElemSegmentInfo {
3424 table_index: 0,
3425 offset: Some(0),
3426 funcs: Some(vec![0, 1, 2]),
3427 }]
3428 );
3429 // 2 type-section entries ($bin + the export's (i32 i32)->i32 dedups
3430 // to one in practice, but don't assume — just check func 0..2 share
3431 // a signature with type 0).
3432 assert_eq!(module.func_type_indices.len(), 4);
3433
3434 let guards = module.call_indirect_guards();
3435 assert_eq!(guards.tables.len(), 1);
3436 assert_eq!(guards.tables[0].table_size, Some(3));
3437 assert_eq!(
3438 guards.tables[0].base_byte_offset,
3439 Some(0),
3440 "#650: a single-table module keeps table 0 at R11 offset 0 by construction"
3441 );
3442 // Type index 0 ($bin) must be VERIFIED: every table entry has its
3443 // exact signature.
3444 assert_eq!(
3445 guards.tables[0].type_reject.first(),
3446 Some(&None),
3447 "closed-world type check must verify the homogeneous table: {:?}",
3448 guards.tables[0].type_reject
3449 );
3450 assert!(
3451 !guards.tables[0].has_null_slots,
3452 "#664: a fully-initialized table must NOT request the runtime \
3453 null check (dispatch bytes stay identical by construction)"
3454 );
3455 }
3456
3457 /// #642: a heterogeneous table (an entry whose signature differs from the
3458 /// expected type) must REJECT that expected type — the raw code-pointer
3459 /// table cannot be runtime-type-checked, so the lowering has to decline.
3460 #[test]
3461 fn test_call_indirect_guards_heterogeneous_table_rejects_642() {
3462 let wat = r#"
3463 (module
3464 (type $bin (func (param i32 i32) (result i32)))
3465 (type $un (func (param i32) (result i32)))
3466 (table 2 funcref)
3467 (elem (i32.const 0) $add $neg)
3468 (func $add (type $bin)
3469 (i32.add (local.get 0) (local.get 1)))
3470 (func $neg (type $un)
3471 (i32.sub (i32.const 0) (local.get 0)))
3472 (func (export "f") (param i32 i32) (result i32)
3473 (call_indirect (type $bin)
3474 (local.get 0) (i32.const 10) (local.get 1)))
3475 )
3476 "#;
3477 let wasm = wat::parse_str(wat).expect("parse");
3478 let module = decode_wasm_module(&wasm).expect("decode");
3479 let guards = module.call_indirect_guards();
3480 assert_eq!(guards.tables[0].table_size, Some(2));
3481 // BOTH expected types must be rejected: the table holds one function
3482 // of each signature, so neither type's closed world holds.
3483 assert!(
3484 guards.tables[0].type_reject[0].is_some() && guards.tables[0].type_reject[1].is_some(),
3485 "heterogeneous table must reject every expected type: {:?}",
3486 guards.tables[0].type_reject
3487 );
3488 // #676: ... but the image is statically known, so the mismatch trap
3489 // is dischargeable at RUNTIME via the type-id sidecar.
3490 assert!(
3491 guards.tables[0].runtime_type_check,
3492 "heterogeneous-but-known table must offer the runtime check (#676)"
3493 );
3494 assert_eq!(
3495 guards.type_ids_byte_offset,
3496 Some(8),
3497 "sidecar sits after the 2-slot pointer region"
3498 );
3499 assert_eq!(
3500 guards.type_ids_image,
3501 vec![1, 2],
3502 "slot 0 = $bin (class 1), slot 1 = $un (class 2)"
3503 );
3504 assert_eq!(guards.type_class_ids, vec![1, 2]);
3505 }
3506
3507 /// #664 (relaxes the #642 all-reject): an uninitialized table slot (elem
3508 /// covers less than the declared size) is a null funcref — calling it
3509 /// must trap, which is now discharged at RUNTIME (null check on the
3510 /// zero-linked pointer), so the closed-world verdict verifies the
3511 /// INITIALIZED slots and sets `has_null_slots` for the lowering.
3512 #[test]
3513 fn test_call_indirect_guards_null_slot_verifies_with_flag_664() {
3514 let wat = r#"
3515 (module
3516 (type $s (func (result i32)))
3517 (table 3 funcref)
3518 (elem (i32.const 0) $f0 $f1)
3519 (func $f0 (result i32) (i32.const 10))
3520 (func $f1 (result i32) (i32.const 11))
3521 (func (export "run") (param i32) (result i32)
3522 (call_indirect (type $s) (local.get 0)))
3523 )
3524 "#;
3525 let wasm = wat::parse_str(wat).expect("parse");
3526 let module = decode_wasm_module(&wasm).expect("decode");
3527 let guards = module.call_indirect_guards();
3528 assert_eq!(guards.tables[0].table_size, Some(3));
3529 assert_eq!(
3530 guards.tables[0].type_reject.first(),
3531 Some(&None),
3532 "initialized slots are homogeneous in $s — the verdict must \
3533 verify despite the null slot (#664): {:?}",
3534 guards.tables[0].type_reject
3535 );
3536 assert!(
3537 guards.tables[0].has_null_slots,
3538 "slot 2 is uninitialized — the lowering must emit the runtime \
3539 null check (#664)"
3540 );
3541 }
3542
3543 /// #664: the falcon shape — a SPARSE table (only slots 1 and 3 of 4
3544 /// initialized, by two separate segments) verifies with the null flag;
3545 /// a sparse table whose INITIALIZED slots are heterogeneous still
3546 /// rejects (the runtime null check cannot discharge a TYPE mismatch).
3547 #[test]
3548 fn test_call_indirect_guards_sparse_table_664() {
3549 let wat = r#"
3550 (module
3551 (type $t (func (param i32) (result i32)))
3552 (table 4 4 funcref)
3553 (func $f1 (type $t) (i32.add (local.get 0) (i32.const 100)))
3554 (func $f3 (type $t) (i32.sub (i32.const 1000) (local.get 0)))
3555 (elem (i32.const 1) $f1)
3556 (elem (i32.const 3) $f3)
3557 (func (export "via") (param i32 i32) (result i32)
3558 (call_indirect (type $t) (local.get 0) (local.get 1)))
3559 )
3560 "#;
3561 let wasm = wat::parse_str(wat).expect("parse");
3562 let module = decode_wasm_module(&wasm).expect("decode");
3563 let guards = module.call_indirect_guards();
3564 assert_eq!(guards.tables[0].table_size, Some(4));
3565 assert_eq!(
3566 guards.tables[0].type_reject.first(),
3567 Some(&None),
3568 "slots 1,3 are homogeneous in $t — verified: {:?}",
3569 guards.tables[0].type_reject
3570 );
3571 assert!(guards.tables[0].has_null_slots, "slots 0,2 are null");
3572
3573 // Heterogeneous INITIALIZED slots in a sparse table: still rejected.
3574 let wat = r#"
3575 (module
3576 (type $t (func (param i32) (result i32)))
3577 (type $u (func (param i32 i32) (result i32)))
3578 (table 4 4 funcref)
3579 (func $f1 (type $t) (local.get 0))
3580 (func $f3 (type $u) (i32.add (local.get 0) (local.get 1)))
3581 (elem (i32.const 1) $f1)
3582 (elem (i32.const 3) $f3)
3583 (func (export "via") (param i32 i32) (result i32)
3584 (call_indirect (type $t) (local.get 0) (local.get 1)))
3585 )
3586 "#;
3587 let wasm = wat::parse_str(wat).expect("parse");
3588 let module = decode_wasm_module(&wasm).expect("decode");
3589 let guards = module.call_indirect_guards();
3590 assert!(
3591 guards.tables[0].type_reject[0].is_some() && guards.tables[0].type_reject[1].is_some(),
3592 "a heterogeneous sparse table must still reject every type: {:?}",
3593 guards.tables[0].type_reject
3594 );
3595 // #676: the sparse-heterogeneous case is now dischargeable at
3596 // runtime too — null slots take the reserved class id 0, so ONE
3597 // sidecar compare covers both the type mismatch and the null trap.
3598 assert!(guards.tables[0].runtime_type_check, "#676 runtime check");
3599 assert_eq!(guards.type_ids_byte_offset, Some(16), "4 pointer slots");
3600 assert_eq!(
3601 guards.type_ids_image,
3602 vec![0, 1, 0, 2],
3603 "nulls at 0/2 carry the reserved id 0; $t slot 1 = class 1, \
3604 $u slot 3 = class 2"
3605 );
3606 }
3607
3608 /// #676: the heterogeneous type-id sidecar — structural duplicate types
3609 /// share one class id (the meld 31-decls/25-distinct shape), null slots
3610 /// take the reserved id 0, and the sidecar base is the total pointer
3611 /// region size. A module with NO heterogeneous table gets NO sidecar
3612 /// (empty image, `None` offset) — homogeneous modules stay untouched.
3613 #[test]
3614 fn test_call_indirect_guards_heterogeneous_sidecar_676() {
3615 let wat = r#"
3616 (module
3617 (type $bin (func (param i32 i32) (result i32)))
3618 (type $un (func (param i32) (result i32)))
3619 (type $bin2 (func (param i32 i32) (result i32)))
3620 (table 5 5 funcref)
3621 (func $add (type $bin) (i32.add (local.get 0) (local.get 1)))
3622 (func $neg (type $un) (i32.sub (i32.const 0) (local.get 0)))
3623 (func $sub (type $bin2) (i32.sub (local.get 0) (local.get 1)))
3624 (elem (i32.const 0) func $add $neg $sub)
3625 (func (export "via2") (param i32 i32) (result i32)
3626 (call_indirect (type $bin)
3627 (local.get 0) (i32.const 3) (local.get 1)))
3628 (func (export "via1") (param i32 i32) (result i32)
3629 (call_indirect (type $un) (local.get 0) (local.get 1)))
3630 )
3631 "#;
3632 let wasm = wat::parse_str(wat).expect("parse");
3633 let module = decode_wasm_module(&wasm).expect("decode");
3634 let guards = module.call_indirect_guards();
3635 assert!(guards.tables[0].runtime_type_check);
3636 assert_eq!(
3637 guards.type_class_ids,
3638 vec![1, 2, 1],
3639 "$bin2 is a structural duplicate of $bin — one class id (#676)"
3640 );
3641 assert_eq!(
3642 guards.type_ids_image,
3643 vec![1, 2, 1, 0, 0],
3644 "slots: $add(bin)=1, $neg(un)=2, $sub(bin2 ≡ bin)=1, null, null"
3645 );
3646 assert_eq!(
3647 guards.type_ids_byte_offset,
3648 Some(20),
3649 "sidecar starts after the 5 pointer words"
3650 );
3651
3652 // Homogeneous module → NO sidecar, no runtime check anywhere.
3653 let wat = r#"
3654 (module
3655 (type $t (func (param i32) (result i32)))
3656 (table 2 2 funcref)
3657 (func $f0 (type $t) (local.get 0))
3658 (func $f1 (type $t) (i32.const 7))
3659 (elem (i32.const 0) func $f0 $f1)
3660 (func (export "via") (param i32 i32) (result i32)
3661 (call_indirect (type $t) (local.get 0) (local.get 1)))
3662 )
3663 "#;
3664 let wasm = wat::parse_str(wat).expect("parse");
3665 let module = decode_wasm_module(&wasm).expect("decode");
3666 let guards = module.call_indirect_guards();
3667 assert!(!guards.tables[0].runtime_type_check);
3668 assert_eq!(guards.type_ids_byte_offset, None, "no heterogeneous table");
3669 assert!(guards.type_ids_image.is_empty());
3670 assert!(guards.type_class_ids.is_empty());
3671 }
3672
3673 /// #642: no table at all → no compile-time bound → table_size None and
3674 /// every type rejected (the lowering declines).
3675 #[test]
3676 fn test_call_indirect_guards_no_table_642() {
3677 let wat = r#"
3678 (module
3679 (func (export "f") (param i32) (result i32) (local.get 0))
3680 )
3681 "#;
3682 let wasm = wat::parse_str(wat).expect("parse");
3683 let module = decode_wasm_module(&wasm).expect("decode");
3684 assert_eq!(module.table_size, None);
3685 assert!(module.table_sizes.is_empty(), "#650: no tables declared");
3686 let guards = module.call_indirect_guards();
3687 assert!(
3688 guards.tables.is_empty(),
3689 "no table → no guard entry → every call_indirect declines"
3690 );
3691 }
3692
3693 /// #642: duplicate-but-structurally-identical types stay interchangeable —
3694 /// the closed-world check compares SIGNATURES, not type indices.
3695 #[test]
3696 fn test_call_indirect_guards_duplicate_types_verified_642() {
3697 let wat = r#"
3698 (module
3699 (type $a (func (result i32)))
3700 (type $b (func (result i32)))
3701 (table 1 funcref)
3702 (elem (i32.const 0) $f)
3703 (func $f (type $a) (i32.const 7))
3704 (func (export "run") (param i32) (result i32)
3705 (call_indirect (type $b) (local.get 0)))
3706 )
3707 "#;
3708 let wasm = wat::parse_str(wat).expect("parse");
3709 let module = decode_wasm_module(&wasm).expect("decode");
3710 let guards = module.call_indirect_guards();
3711 // $f has type $a; the call expects $b — structurally identical, so
3712 // BOTH type indices must verify. (A third type — the export's
3713 // (i32)->i32 — is correctly rejected: different signature.)
3714 assert_eq!(
3715 &guards.tables[0].type_reject[0..2],
3716 &[None, None],
3717 "structural signature comparison must accept duplicate types: {:?}",
3718 guards.tables[0].type_reject
3719 );
3720 assert!(
3721 guards.tables[0].type_reject[2].is_some(),
3722 "the structurally-different third type must still be rejected"
3723 );
3724 }
3725
3726 /// #650: TWO tables become a contiguous R11 region — table 0 at offset 0
3727 /// (byte-identical single-table degeneration), table 1 at
3728 /// `size(table 0) * 4`. Each table gets its OWN size, base offset, and
3729 /// per-type closed-world verdicts (segments only poison the table they
3730 /// target).
3731 #[test]
3732 fn test_call_indirect_guards_multi_table_650() {
3733 // The #650 repro shape: overlapping indices, distinct functions —
3734 // table0[1] != table1[1] (the aliasing canary).
3735 let wat = r#"
3736 (module
3737 (type $t (func (param i32) (result i32)))
3738 (type $u (func (param i32 i32) (result i32)))
3739 (table $t0 3 3 funcref)
3740 (table $t1 2 2 funcref)
3741 (func $a0 (type $t) (i32.add (local.get 0) (i32.const 100)))
3742 (func $a1 (type $t) (i32.add (local.get 0) (i32.const 200)))
3743 (func $a2 (type $t) (i32.add (local.get 0) (i32.const 300)))
3744 (func $b0 (type $u) (i32.add (local.get 0) (local.get 1)))
3745 (func $b1 (type $u) (i32.sub (local.get 0) (local.get 1)))
3746 (elem (table $t0) (i32.const 0) func $a0 $a1 $a2)
3747 (elem (table $t1) (i32.const 0) func $b0 $b1)
3748 (func (export "f") (param i32 i32) (result i32)
3749 (call_indirect $t1 (type $u)
3750 (local.get 0) (i32.const 10) (local.get 1)))
3751 )
3752 "#;
3753 let wasm = wat::parse_str(wat).expect("parse");
3754 let module = decode_wasm_module(&wasm).expect("decode");
3755 assert_eq!(module.table_sizes, vec![Some(3), Some(2)]);
3756 assert_eq!(module.table_size, Some(3), "compat accessor = table 0");
3757 assert_eq!(
3758 module.elem_segments[0].table_index, 0,
3759 "segment 0 targets table 0"
3760 );
3761 assert_eq!(
3762 module.elem_segments[1],
3763 ElemSegmentInfo {
3764 table_index: 1,
3765 offset: Some(0),
3766 funcs: Some(vec![3, 4]),
3767 },
3768 "segment 1 is statically attributed to table 1 (#650)"
3769 );
3770
3771 let guards = module.call_indirect_guards();
3772 assert_eq!(guards.tables.len(), 2);
3773 assert_eq!(guards.tables[0].table_size, Some(3));
3774 assert_eq!(guards.tables[0].base_byte_offset, Some(0));
3775 assert_eq!(guards.tables[1].table_size, Some(2));
3776 assert_eq!(
3777 guards.tables[1].base_byte_offset,
3778 Some(12),
3779 "table 1 base = size(table 0) * 4 within the contiguous R11 region"
3780 );
3781 // Table 0 is homogeneous in $t (type 0); table 1 in $u (type 1) —
3782 // each verifies ITS type and rejects the other's.
3783 assert_eq!(guards.tables[0].type_reject[0], None, "table 0 vs $t");
3784 assert!(guards.tables[0].type_reject[1].is_some(), "table 0 vs $u");
3785 assert!(guards.tables[1].type_reject[0].is_some(), "table 1 vs $t");
3786 assert_eq!(guards.tables[1].type_reject[1], None, "table 1 vs $u");
3787 }
3788
3789 /// #650: an unknown-size table (growable import) declines ITSELF and
3790 /// makes every LATER table's base offset non-constant — but a table
3791 /// BEFORE it is unaffected.
3792 #[test]
3793 fn test_call_indirect_guards_unknown_size_poisons_later_bases_650() {
3794 let wat = r#"
3795 (module
3796 (type $t (func (result i32)))
3797 (import "env" "tbl" (table 4 funcref))
3798 (table $d 1 1 funcref)
3799 (func $f (type $t) (i32.const 7))
3800 (elem (table $d) (i32.const 0) func $f)
3801 (func (export "run") (param i32) (result i32)
3802 (call_indirect $d (type $t) (local.get 0)))
3803 )
3804 "#;
3805 let wasm = wat::parse_str(wat).expect("parse");
3806 let module = decode_wasm_module(&wasm).expect("decode");
3807 assert_eq!(
3808 module.table_sizes,
3809 vec![None, Some(1)],
3810 "growable import (no max) has no sound compile-time size"
3811 );
3812 let guards = module.call_indirect_guards();
3813 assert_eq!(guards.tables[0].base_byte_offset, Some(0));
3814 assert!(
3815 guards.tables[0].type_reject.iter().all(|r| r.is_some()),
3816 "unknown-size table rejects every type"
3817 );
3818 assert_eq!(
3819 guards.tables[1].base_byte_offset, None,
3820 "a later table's base is not a compile-time constant when a \
3821 preceding table's size is unknown (#650)"
3822 );
3823 assert_eq!(guards.tables[1].table_size, Some(1));
3824 }
3825}