| 1 | use rustc_hash::{FxBuildHasher, FxHashMap, FxHashSet}; |
| 2 | |
| 3 | use indexmap::IndexMap; |
| 4 | use react_compiler_diagnostics::{CompilerDiagnostic, CompilerDiagnosticDetail, ErrorCategory}; |
| 5 | use react_compiler_hir::environment::Environment; |
| 6 | use react_compiler_hir::visitors; |
| 7 | use react_compiler_hir::*; |
| 8 | |
| 9 | // ============================================================================= |
| 10 | // SSABuilder |
| 11 | // ============================================================================= |
| 12 | |
| 13 | struct IncompletePhi { |
| 14 | old_place: Place, |
| 15 | new_place: Place, |
| 16 | } |
| 17 | |
| 18 | struct State { |
| 19 | defs: FxHashMap<IdentifierId, IdentifierId>, |
| 20 | incomplete_phis: Vec<IncompletePhi>, |
| 21 | } |
| 22 | |
| 23 | struct SSABuilder { |
| 24 | states: FxHashMap<BlockId, State>, |
| 25 | current: Option<BlockId>, |
| 26 | unsealed_preds: FxHashMap<BlockId, u32>, |
| 27 | block_preds: FxHashMap<BlockId, Vec<BlockId>>, |
| 28 | unknown: FxHashSet<IdentifierId>, |
| 29 | context: FxHashSet<IdentifierId>, |
| 30 | pending_phis: FxHashMap<BlockId, Vec<Phi>>, |
| 31 | processed_functions: Vec<FunctionId>, |
| 32 | } |
| 33 | |
| 34 | impl SSABuilder { |
| 35 | fn new(blocks: &IndexMap<BlockId, BasicBlock, FxBuildHasher>) -> Self { |
| 36 | let mut block_preds = FxHashMap::default(); |
| 37 | for (id, block) in blocks { |
| 38 | block_preds.insert(*id, block.preds.iter().copied().collect()); |
| 39 | } |
| 40 | SSABuilder { |
| 41 | states: FxHashMap::default(), |
| 42 | current: None, |
| 43 | unsealed_preds: FxHashMap::default(), |
| 44 | block_preds, |
| 45 | unknown: FxHashSet::default(), |
| 46 | context: FxHashSet::default(), |
| 47 | pending_phis: FxHashMap::default(), |
| 48 | processed_functions: Vec::new(), |
| 49 | } |
| 50 | } |
| 51 | |
| 52 | fn define_function(&mut self, func: &HirFunction) { |
| 53 | for (id, block) in &func.body.blocks { |
| 54 | self.block_preds |
| 55 | .insert(*id, block.preds.iter().copied().collect()); |
| 56 | } |
| 57 | } |
| 58 | |
| 59 | fn state_mut(&mut self) -> &mut State { |
| 60 | let current = self |
| 61 | .current |
| 62 | .expect("we need to be in a block to access state!"); |
| 63 | self.states |
| 64 | .get_mut(¤t) |
| 65 | .expect("state not found for current block") |
| 66 | } |
| 67 | |
| 68 | fn make_id(&mut self, old_id: IdentifierId, env: &mut Environment) -> IdentifierId { |
| 69 | let new_id = env.next_identifier_id(); |
| 70 | let old = &env.identifiers[old_id.0 as usize]; |
| 71 | let declaration_id = old.declaration_id; |
| 72 | let name = old.name.clone(); |
| 73 | let loc = old.loc; |
| 74 | let new_ident = &mut env.identifiers[new_id.0 as usize]; |
| 75 | new_ident.declaration_id = declaration_id; |
| 76 | new_ident.name = name; |
| 77 | new_ident.loc = loc; |
| 78 | new_id |
| 79 | } |
| 80 | |
| 81 | fn define_place( |
| 82 | &mut self, |
| 83 | old_place: &Place, |
| 84 | env: &mut Environment, |
| 85 | ) -> Result<Place, CompilerDiagnostic> { |
| 86 | let old_id = old_place.identifier; |
| 87 | |
| 88 | if self.unknown.contains(&old_id) { |
| 89 | let ident = &env.identifiers[old_id.0 as usize]; |
| 90 | let name = match &ident.name { |
| 91 | Some(name) => format!("{}${}", name.value(), old_id.0), |
| 92 | None => format!("${}", old_id.0), |
| 93 | }; |
| 94 | return Err(CompilerDiagnostic::new( |
| 95 | ErrorCategory::Todo, |
| 96 | "[hoisting] EnterSSA: Expected identifier to be defined before being used", |
| 97 | Some(format!("Identifier {} is undefined", name)), |
| 98 | ) |
| 99 | .with_detail(CompilerDiagnosticDetail::Error { |
| 100 | loc: old_place.loc, |
| 101 | message: None, |
| 102 | identifier_name: None, |
| 103 | })); |
| 104 | } |
| 105 | |
| 106 | // Do not redefine context references. |
| 107 | if self.context.contains(&old_id) { |
| 108 | return Ok(self.get_place(old_place, env)); |
| 109 | } |
| 110 | |
| 111 | let new_id = self.make_id(old_id, env); |
| 112 | self.state_mut().defs.insert(old_id, new_id); |
| 113 | Ok(Place { |
| 114 | identifier: new_id, |
| 115 | effect: old_place.effect, |
| 116 | reactive: old_place.reactive, |
| 117 | loc: old_place.loc, |
| 118 | }) |
| 119 | } |
| 120 | |
| 121 | #[allow(dead_code)] |
| 122 | fn define_context( |
| 123 | &mut self, |
| 124 | old_place: &Place, |
| 125 | env: &mut Environment, |
| 126 | ) -> Result<Place, CompilerDiagnostic> { |
| 127 | let old_id = old_place.identifier; |
| 128 | let new_place = self.define_place(old_place, env)?; |
| 129 | self.context.insert(old_id); |
| 130 | Ok(new_place) |
| 131 | } |
| 132 | |
| 133 | /// A function's context places capture a *binding*, not a value: the |
| 134 | /// variable is only read when the function is later called, so a context |
| 135 | /// place may reference a binding that is declared after the function |
| 136 | /// expression itself (eg `const colgroup = useMemo(() => <colgroup>...)`, |
| 137 | /// where the JSX tag name resolves to the variable being assigned). Unmark |
| 138 | /// such identifiers so the later declaration doesn't error; if the function |
| 139 | /// body actually *reads* the variable before it is defined, visiting the |
| 140 | /// body re-marks it and the hoisting bailout in define_place still applies. |
| 141 | fn unmark_unknown(&mut self, id: IdentifierId) { |
| 142 | self.unknown.remove(&id); |
| 143 | } |
| 144 | |
| 145 | fn get_place(&mut self, old_place: &Place, env: &mut Environment) -> Place { |
| 146 | let current_id = self.current.expect("must be in a block"); |
| 147 | let new_id = self.get_id_at(old_place, current_id, env); |
| 148 | Place { |
| 149 | identifier: new_id, |
| 150 | effect: old_place.effect, |
| 151 | reactive: old_place.reactive, |
| 152 | loc: old_place.loc, |
| 153 | } |
| 154 | } |
| 155 | |
| 156 | fn get_id_at( |
| 157 | &mut self, |
| 158 | old_place: &Place, |
| 159 | block_id: BlockId, |
| 160 | env: &mut Environment, |
| 161 | ) -> IdentifierId { |
| 162 | if let Some(state) = self.states.get(&block_id) { |
| 163 | if let Some(&new_id) = state.defs.get(&old_place.identifier) { |
| 164 | return new_id; |
| 165 | } |
| 166 | } |
| 167 | |
| 168 | let preds = self.block_preds.get(&block_id).cloned().unwrap_or_default(); |
| 169 | |
| 170 | if preds.is_empty() { |
| 171 | self.unknown.insert(old_place.identifier); |
| 172 | return old_place.identifier; |
| 173 | } |
| 174 | |
| 175 | let unsealed = self.unsealed_preds.get(&block_id).copied().unwrap_or(0); |
| 176 | if unsealed > 0 { |
| 177 | let new_id = self.make_id(old_place.identifier, env); |
| 178 | let new_place = Place { |
| 179 | identifier: new_id, |
| 180 | effect: old_place.effect, |
| 181 | reactive: old_place.reactive, |
| 182 | loc: old_place.loc, |
| 183 | }; |
| 184 | let state = self.states.get_mut(&block_id).unwrap(); |
| 185 | state.incomplete_phis.push(IncompletePhi { |
| 186 | old_place: old_place.clone(), |
| 187 | new_place, |
| 188 | }); |
| 189 | state.defs.insert(old_place.identifier, new_id); |
| 190 | return new_id; |
| 191 | } |
| 192 | |
| 193 | if preds.len() == 1 { |
| 194 | let pred = preds[0]; |
| 195 | let new_id = self.get_id_at(old_place, pred, env); |
| 196 | self.states |
| 197 | .get_mut(&block_id) |
| 198 | .unwrap() |
| 199 | .defs |
| 200 | .insert(old_place.identifier, new_id); |
| 201 | return new_id; |
| 202 | } |
| 203 | |
| 204 | let new_id = self.make_id(old_place.identifier, env); |
| 205 | self.states |
| 206 | .get_mut(&block_id) |
| 207 | .unwrap() |
| 208 | .defs |
| 209 | .insert(old_place.identifier, new_id); |
| 210 | let new_place = Place { |
| 211 | identifier: new_id, |
| 212 | effect: old_place.effect, |
| 213 | reactive: old_place.reactive, |
| 214 | loc: old_place.loc, |
| 215 | }; |
| 216 | self.add_phi(block_id, old_place, &new_place, env); |
| 217 | new_id |
| 218 | } |
| 219 | |
| 220 | fn add_phi( |
| 221 | &mut self, |
| 222 | block_id: BlockId, |
| 223 | old_place: &Place, |
| 224 | new_place: &Place, |
| 225 | env: &mut Environment, |
| 226 | ) { |
| 227 | let preds = self.block_preds.get(&block_id).cloned().unwrap_or_default(); |
| 228 | |
| 229 | let mut pred_defs: IndexMap<BlockId, Place, FxBuildHasher> = IndexMap::default(); |
| 230 | for pred_block_id in &preds { |
| 231 | let pred_id = self.get_id_at(old_place, *pred_block_id, env); |
| 232 | pred_defs.insert( |
| 233 | *pred_block_id, |
| 234 | Place { |
| 235 | identifier: pred_id, |
| 236 | effect: old_place.effect, |
| 237 | reactive: old_place.reactive, |
| 238 | loc: old_place.loc, |
| 239 | }, |
| 240 | ); |
| 241 | } |
| 242 | |
| 243 | let phi = Phi { |
| 244 | place: new_place.clone(), |
| 245 | operands: pred_defs, |
| 246 | }; |
| 247 | |
| 248 | self.pending_phis.entry(block_id).or_default().push(phi); |
| 249 | } |
| 250 | |
| 251 | fn fix_incomplete_phis(&mut self, block_id: BlockId, env: &mut Environment) { |
| 252 | let incomplete_phis: Vec<IncompletePhi> = self |
| 253 | .states |
| 254 | .get_mut(&block_id) |
| 255 | .unwrap() |
| 256 | .incomplete_phis |
| 257 | .drain(..) |
| 258 | .collect(); |
| 259 | for phi in &incomplete_phis { |
| 260 | self.add_phi(block_id, &phi.old_place, &phi.new_place, env); |
| 261 | } |
| 262 | } |
| 263 | |
| 264 | fn start_block(&mut self, block_id: BlockId) { |
| 265 | self.current = Some(block_id); |
| 266 | self.states.insert( |
| 267 | block_id, |
| 268 | State { |
| 269 | defs: FxHashMap::default(), |
| 270 | incomplete_phis: Vec::new(), |
| 271 | }, |
| 272 | ); |
| 273 | } |
| 274 | } |
| 275 | |
| 276 | // ============================================================================= |
| 277 | // Public entry point |
| 278 | // ============================================================================= |
| 279 | |
| 280 | pub fn enter_ssa(func: &mut HirFunction, env: &mut Environment) -> Result<(), CompilerDiagnostic> { |
| 281 | let mut builder = SSABuilder::new(&func.body.blocks); |
| 282 | let root_entry = func.body.entry; |
| 283 | enter_ssa_impl(func, &mut builder, env, root_entry)?; |
| 284 | |
| 285 | // Apply all pending phis to the actual blocks |
| 286 | apply_pending_phis(func, env, &mut builder); |
| 287 | |
| 288 | Ok(()) |
| 289 | } |
| 290 | |
| 291 | fn apply_pending_phis(func: &mut HirFunction, env: &mut Environment, builder: &mut SSABuilder) { |
| 292 | for (block_id, block) in func.body.blocks.iter_mut() { |
| 293 | if let Some(phis) = builder.pending_phis.remove(block_id) { |
| 294 | block.phis.extend(phis); |
| 295 | } |
| 296 | } |
| 297 | for fid in &builder.processed_functions.clone() { |
| 298 | let inner_func = &mut env.functions[fid.0 as usize]; |
| 299 | for (block_id, block) in inner_func.body.blocks.iter_mut() { |
| 300 | if let Some(phis) = builder.pending_phis.remove(block_id) { |
| 301 | block.phis.extend(phis); |
| 302 | } |
| 303 | } |
| 304 | } |
| 305 | } |
| 306 | |
| 307 | fn enter_ssa_impl( |
| 308 | func: &mut HirFunction, |
| 309 | builder: &mut SSABuilder, |
| 310 | env: &mut Environment, |
| 311 | root_entry: BlockId, |
| 312 | ) -> Result<(), CompilerDiagnostic> { |
| 313 | let mut visited_blocks: FxHashSet<BlockId> = FxHashSet::default(); |
| 314 | let block_ids: Vec<BlockId> = func.body.blocks.keys().copied().collect(); |
| 315 | |
| 316 | for block_id in &block_ids { |
| 317 | let block_id = *block_id; |
| 318 | |
| 319 | if visited_blocks.contains(&block_id) { |
| 320 | return Err(CompilerDiagnostic::new( |
| 321 | ErrorCategory::Invariant, |
| 322 | format!("found a cycle! visiting bb{} again", block_id.0), |
| 323 | None, |
| 324 | )); |
| 325 | } |
| 326 | |
| 327 | visited_blocks.insert(block_id); |
| 328 | builder.start_block(block_id); |
| 329 | |
| 330 | // Handle params at the root entry |
| 331 | if block_id == root_entry { |
| 332 | if !func.context.is_empty() { |
| 333 | return Err(CompilerDiagnostic::new( |
| 334 | ErrorCategory::Invariant, |
| 335 | "Expected function context to be empty for outer function declarations", |
| 336 | None, |
| 337 | )); |
| 338 | } |
| 339 | let params = std::mem::take(&mut func.params); |
| 340 | let mut new_params = Vec::with_capacity(params.len()); |
| 341 | for param in params { |
| 342 | new_params.push(match param { |
| 343 | ParamPattern::Place(p) => ParamPattern::Place(builder.define_place(&p, env)?), |
| 344 | ParamPattern::Spread(s) => ParamPattern::Spread(SpreadPattern { |
| 345 | place: builder.define_place(&s.place, env)?, |
| 346 | }), |
| 347 | }); |
| 348 | } |
| 349 | func.params = new_params; |
| 350 | } |
| 351 | |
| 352 | // Process instructions |
| 353 | let instruction_ids: Vec<InstructionId> = func |
| 354 | .body |
| 355 | .blocks |
| 356 | .get(&block_id) |
| 357 | .unwrap() |
| 358 | .instructions |
| 359 | .clone(); |
| 360 | |
| 361 | for instr_id in &instruction_ids { |
| 362 | let instr_idx = instr_id.0 as usize; |
| 363 | let instr = &mut func.instructions[instr_idx]; |
| 364 | |
| 365 | // For FunctionExpression/ObjectMethod, we need to handle context |
| 366 | // mapping specially because env.functions is borrowed by the closure. |
| 367 | // First, check if this is a FunctionExpression/ObjectMethod and handle |
| 368 | // context mapping separately. |
| 369 | let func_expr_id = match &instr.value { |
| 370 | InstructionValue::FunctionExpression { lowered_func, .. } |
| 371 | | InstructionValue::ObjectMethod { lowered_func, .. } => Some(lowered_func.func), |
| 372 | _ => None, |
| 373 | }; |
| 374 | |
| 375 | // Map context places for function expressions before other operands |
| 376 | if let Some(fid) = func_expr_id { |
| 377 | let context = std::mem::take(&mut env.functions[fid.0 as usize].context); |
| 378 | env.functions[fid.0 as usize].context = context |
| 379 | .into_iter() |
| 380 | .map(|place| builder.get_place(&place, env)) |
| 381 | .collect(); |
| 382 | } |
| 383 | |
| 384 | // Map non-context operands |
| 385 | visitors::for_each_instruction_value_operand_mut(&mut instr.value, &mut |place| { |
| 386 | *place = builder.get_place(place, env); |
| 387 | }); |
| 388 | |
| 389 | // Map lvalues (skip DeclareContext/StoreContext — context variables |
| 390 | // don't participate in SSA renaming) |
| 391 | let instr = &mut func.instructions[instr_idx]; |
| 392 | let mut lvalue_err: Option<CompilerDiagnostic> = None; |
| 393 | visitors::for_each_instruction_lvalue_mut(instr, &mut |place| { |
| 394 | if lvalue_err.is_none() { |
| 395 | match builder.define_place(place, env) { |
| 396 | Ok(new_place) => *place = new_place, |
| 397 | Err(e) => lvalue_err = Some(e), |
| 398 | } |
| 399 | } |
| 400 | }); |
| 401 | if let Some(e) = lvalue_err { |
| 402 | return Err(e); |
| 403 | } |
| 404 | |
| 405 | // Handle inner function SSA |
| 406 | if let Some(fid) = func_expr_id { |
| 407 | let context_ids: Vec<IdentifierId> = env.functions[fid.0 as usize] |
| 408 | .context |
| 409 | .iter() |
| 410 | .map(|place| place.identifier) |
| 411 | .collect(); |
| 412 | for id in context_ids { |
| 413 | builder.unmark_unknown(id); |
| 414 | } |
| 415 | builder.processed_functions.push(fid); |
| 416 | let inner_func = &mut env.functions[fid.0 as usize]; |
| 417 | let inner_entry = inner_func.body.entry; |
| 418 | let entry_block = inner_func.body.blocks.get_mut(&inner_entry).unwrap(); |
| 419 | |
| 420 | if !entry_block.preds.is_empty() { |
| 421 | return Err(CompilerDiagnostic::new( |
| 422 | ErrorCategory::Invariant, |
| 423 | "Expected function expression entry block to have zero predecessors", |
| 424 | None, |
| 425 | )); |
| 426 | } |
| 427 | entry_block.preds.insert(block_id); |
| 428 | |
| 429 | builder.define_function(inner_func); |
| 430 | |
| 431 | let saved_current = builder.current; |
| 432 | |
| 433 | // Map inner function params |
| 434 | let inner_params = std::mem::take(&mut env.functions[fid.0 as usize].params); |
| 435 | let mut new_inner_params = Vec::with_capacity(inner_params.len()); |
| 436 | for param in inner_params { |
| 437 | new_inner_params.push(match param { |
| 438 | ParamPattern::Place(p) => { |
| 439 | ParamPattern::Place(builder.define_place(&p, env)?) |
| 440 | } |
| 441 | ParamPattern::Spread(s) => ParamPattern::Spread(SpreadPattern { |
| 442 | place: builder.define_place(&s.place, env)?, |
| 443 | }), |
| 444 | }); |
| 445 | } |
| 446 | env.functions[fid.0 as usize].params = new_inner_params; |
| 447 | |
| 448 | // Take the inner function out of the arena to process it |
| 449 | let mut inner_func = |
| 450 | std::mem::replace(&mut env.functions[fid.0 as usize], placeholder_function()); |
| 451 | |
| 452 | enter_ssa_impl(&mut inner_func, builder, env, root_entry)?; |
| 453 | |
| 454 | // Put it back |
| 455 | env.functions[fid.0 as usize] = inner_func; |
| 456 | |
| 457 | builder.current = saved_current; |
| 458 | |
| 459 | // Clear entry preds |
| 460 | env.functions[fid.0 as usize] |
| 461 | .body |
| 462 | .blocks |
| 463 | .get_mut(&inner_entry) |
| 464 | .unwrap() |
| 465 | .preds |
| 466 | .clear(); |
| 467 | builder.block_preds.insert(inner_entry, Vec::new()); |
| 468 | } |
| 469 | } |
| 470 | |
| 471 | // Map terminal operands |
| 472 | let terminal = &mut func.body.blocks.get_mut(&block_id).unwrap().terminal; |
| 473 | visitors::for_each_terminal_operand_mut(terminal, &mut |place| { |
| 474 | *place = builder.get_place(place, env); |
| 475 | }); |
| 476 | |
| 477 | // Handle successors |
| 478 | let terminal_ref = &func.body.blocks.get(&block_id).unwrap().terminal; |
| 479 | let successors = visitors::each_terminal_successor(terminal_ref); |
| 480 | for output_id in successors { |
| 481 | let output_preds_len = builder |
| 482 | .block_preds |
| 483 | .get(&output_id) |
| 484 | .map(|p| p.len() as u32) |
| 485 | .unwrap_or(0); |
| 486 | |
| 487 | let count = if builder.unsealed_preds.contains_key(&output_id) { |
| 488 | builder.unsealed_preds[&output_id] - 1 |
| 489 | } else { |
| 490 | output_preds_len - 1 |
| 491 | }; |
| 492 | builder.unsealed_preds.insert(output_id, count); |
| 493 | |
| 494 | if count == 0 && visited_blocks.contains(&output_id) { |
| 495 | builder.fix_incomplete_phis(output_id, env); |
| 496 | } |
| 497 | } |
| 498 | } |
| 499 | |
| 500 | Ok(()) |
| 501 | } |
| 502 | |
| 503 | /// Create a placeholder HirFunction for temporarily swapping an inner function |
| 504 | /// out of `env.functions` via `std::mem::replace`. The placeholder is never |
| 505 | /// read — the real function is swapped back immediately after processing. |
| 506 | pub fn placeholder_function() -> HirFunction { |
| 507 | HirFunction { |
| 508 | loc: None, |
| 509 | id: None, |
| 510 | name_hint: None, |
| 511 | fn_type: ReactFunctionType::Other, |
| 512 | params: Vec::new(), |
| 513 | return_type_annotation: None, |
| 514 | returns: Place { |
| 515 | identifier: IdentifierId(0), |
| 516 | effect: Effect::Unknown, |
| 517 | reactive: false, |
| 518 | loc: None, |
| 519 | }, |
| 520 | context: Vec::new(), |
| 521 | body: HIR { |
| 522 | entry: BlockId(0), |
| 523 | blocks: IndexMap::default(), |
| 524 | }, |
| 525 | instructions: Vec::new(), |
| 526 | generator: false, |
| 527 | is_async: false, |
| 528 | directives: Vec::new(), |
| 529 | aliasing_effects: None, |
| 530 | } |
| 531 | } |