| 1 | //! Bench-facing capture helpers. |
| 2 | //! |
| 3 | //! This module exposes a stable-ish API for collecting intermediate artifacts |
| 4 | //! from the compiler pipeline so the external `afs-tests` runner can assert on |
| 5 | //! more than just final program output. |
| 6 | |
| 7 | use std::collections::{BTreeMap, BTreeSet}; |
| 8 | use std::fmt::Write; |
| 9 | use std::fs; |
| 10 | use std::path::{Path, PathBuf}; |
| 11 | use std::process::Command; |
| 12 | use std::sync::atomic::{AtomicU64, Ordering}; |
| 13 | use std::time::SystemTime; |
| 14 | |
| 15 | use crate::codegen::mir::{ |
| 16 | ArmCond, ConstPoolEntry, MBlockId, MachineFunction, MachineInst, MachineOperand, PhysReg, |
| 17 | RegClass, |
| 18 | }; |
| 19 | use crate::codegen::{emit, isel, linearscan}; |
| 20 | use crate::driver::OptLevel; |
| 21 | use crate::ir::{lower, printer as ir_printer, verify}; |
| 22 | use crate::lexer::{detect_source_form, tokenize, SourceForm, Token}; |
| 23 | use crate::opt::pipeline::OptLevel as IrOptLevel; |
| 24 | use crate::opt::{build_i128_pipeline, build_pipeline}; |
| 25 | use crate::parser::Parser; |
| 26 | use crate::sema::{resolve, validate}; |
| 27 | |
| 28 | static CAPTURE_COUNTER: AtomicU64 = AtomicU64::new(0); |
| 29 | |
| 30 | /// Capturable compiler stages for the bench. |
| 31 | #[derive(Debug, Clone, Copy, PartialEq, Eq, PartialOrd, Ord, Hash)] |
| 32 | pub enum Stage { |
| 33 | Preprocess, |
| 34 | Tokens, |
| 35 | Ast, |
| 36 | Sema, |
| 37 | Ir, |
| 38 | OptIr, |
| 39 | Mir, |
| 40 | Regalloc, |
| 41 | Asm, |
| 42 | Obj, |
| 43 | Run, |
| 44 | } |
| 45 | |
| 46 | impl Stage { |
| 47 | pub const ALL: [Stage; 11] = [ |
| 48 | Stage::Preprocess, |
| 49 | Stage::Tokens, |
| 50 | Stage::Ast, |
| 51 | Stage::Sema, |
| 52 | Stage::Ir, |
| 53 | Stage::OptIr, |
| 54 | Stage::Mir, |
| 55 | Stage::Regalloc, |
| 56 | Stage::Asm, |
| 57 | Stage::Obj, |
| 58 | Stage::Run, |
| 59 | ]; |
| 60 | |
| 61 | pub fn parse(name: &str) -> Option<Self> { |
| 62 | match name.trim().to_ascii_lowercase().as_str() { |
| 63 | "preprocess" => Some(Self::Preprocess), |
| 64 | "tokens" => Some(Self::Tokens), |
| 65 | "ast" => Some(Self::Ast), |
| 66 | "sema" => Some(Self::Sema), |
| 67 | "ir" => Some(Self::Ir), |
| 68 | "optir" => Some(Self::OptIr), |
| 69 | "mir" => Some(Self::Mir), |
| 70 | "regalloc" => Some(Self::Regalloc), |
| 71 | "asm" => Some(Self::Asm), |
| 72 | "obj" => Some(Self::Obj), |
| 73 | "run" => Some(Self::Run), |
| 74 | _ => None, |
| 75 | } |
| 76 | } |
| 77 | |
| 78 | pub fn as_str(&self) -> &'static str { |
| 79 | match self { |
| 80 | Self::Preprocess => "preprocess", |
| 81 | Self::Tokens => "tokens", |
| 82 | Self::Ast => "ast", |
| 83 | Self::Sema => "sema", |
| 84 | Self::Ir => "ir", |
| 85 | Self::OptIr => "optir", |
| 86 | Self::Mir => "mir", |
| 87 | Self::Regalloc => "regalloc", |
| 88 | Self::Asm => "asm", |
| 89 | Self::Obj => "obj", |
| 90 | Self::Run => "run", |
| 91 | } |
| 92 | } |
| 93 | } |
| 94 | |
| 95 | /// Failure point while trying to capture compiler stages. |
| 96 | #[derive(Debug, Clone, Copy, PartialEq, Eq)] |
| 97 | pub enum FailureStage { |
| 98 | Preprocess, |
| 99 | Lexer, |
| 100 | Parser, |
| 101 | Sema, |
| 102 | Ir, |
| 103 | Obj, |
| 104 | Run, |
| 105 | } |
| 106 | |
| 107 | impl FailureStage { |
| 108 | pub fn parse(name: &str) -> Option<Self> { |
| 109 | match name.trim().to_ascii_lowercase().as_str() { |
| 110 | "preprocess" => Some(Self::Preprocess), |
| 111 | "lexer" | "tokens" => Some(Self::Lexer), |
| 112 | "parser" | "parse" | "ast" => Some(Self::Parser), |
| 113 | "sema" => Some(Self::Sema), |
| 114 | "ir" | "optir" => Some(Self::Ir), |
| 115 | "obj" | "asm" => Some(Self::Obj), |
| 116 | "run" => Some(Self::Run), |
| 117 | _ => None, |
| 118 | } |
| 119 | } |
| 120 | |
| 121 | pub fn as_str(&self) -> &'static str { |
| 122 | match self { |
| 123 | Self::Preprocess => "preprocess", |
| 124 | Self::Lexer => "lexer", |
| 125 | Self::Parser => "parser", |
| 126 | Self::Sema => "sema", |
| 127 | Self::Ir => "ir", |
| 128 | Self::Obj => "obj", |
| 129 | Self::Run => "run", |
| 130 | } |
| 131 | } |
| 132 | } |
| 133 | |
| 134 | /// A stage capture request for one input source file. |
| 135 | #[derive(Debug, Clone)] |
| 136 | pub struct CaptureRequest { |
| 137 | pub input: PathBuf, |
| 138 | pub requested: BTreeSet<Stage>, |
| 139 | pub opt_level: OptLevel, |
| 140 | } |
| 141 | |
| 142 | impl CaptureRequest { |
| 143 | pub fn new(input: impl Into<PathBuf>) -> Self { |
| 144 | Self { |
| 145 | input: input.into(), |
| 146 | requested: BTreeSet::new(), |
| 147 | opt_level: OptLevel::O0, |
| 148 | } |
| 149 | } |
| 150 | |
| 151 | pub fn with_stage(mut self, stage: Stage) -> Self { |
| 152 | self.requested.insert(stage); |
| 153 | self |
| 154 | } |
| 155 | |
| 156 | pub fn with_all_stages(mut self) -> Self { |
| 157 | self.requested.extend(Stage::ALL); |
| 158 | self |
| 159 | } |
| 160 | |
| 161 | pub fn with_opt_level(mut self, opt_level: OptLevel) -> Self { |
| 162 | self.opt_level = opt_level; |
| 163 | self |
| 164 | } |
| 165 | } |
| 166 | |
| 167 | /// Captured data for a single run. |
| 168 | #[derive(Debug, Clone)] |
| 169 | pub struct CaptureResult { |
| 170 | pub input: PathBuf, |
| 171 | pub opt_level: OptLevel, |
| 172 | pub stages: BTreeMap<Stage, CapturedStage>, |
| 173 | } |
| 174 | |
| 175 | impl CaptureResult { |
| 176 | pub fn get(&self, stage: Stage) -> Option<&CapturedStage> { |
| 177 | self.stages.get(&stage) |
| 178 | } |
| 179 | } |
| 180 | |
| 181 | /// Failed stage capture with partial artifacts preserved. |
| 182 | #[derive(Debug, Clone)] |
| 183 | pub struct CaptureFailure { |
| 184 | pub input: PathBuf, |
| 185 | pub opt_level: OptLevel, |
| 186 | pub stage: FailureStage, |
| 187 | pub detail: String, |
| 188 | pub stages: BTreeMap<Stage, CapturedStage>, |
| 189 | } |
| 190 | |
| 191 | impl CaptureFailure { |
| 192 | pub fn partial_result(&self) -> CaptureResult { |
| 193 | CaptureResult { |
| 194 | input: self.input.clone(), |
| 195 | opt_level: self.opt_level, |
| 196 | stages: self.stages.clone(), |
| 197 | } |
| 198 | } |
| 199 | } |
| 200 | |
| 201 | impl std::fmt::Display for CaptureFailure { |
| 202 | fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result { |
| 203 | write!(f, "{}: {}", self.stage.as_str(), self.detail) |
| 204 | } |
| 205 | } |
| 206 | |
| 207 | impl std::error::Error for CaptureFailure {} |
| 208 | |
| 209 | /// Captured artifact content. |
| 210 | #[derive(Debug, Clone)] |
| 211 | pub enum CapturedStage { |
| 212 | Text(String), |
| 213 | Run(RunCapture), |
| 214 | } |
| 215 | |
| 216 | impl CapturedStage { |
| 217 | pub fn as_text(&self) -> Option<&str> { |
| 218 | match self { |
| 219 | Self::Text(text) => Some(text), |
| 220 | Self::Run(_) => None, |
| 221 | } |
| 222 | } |
| 223 | |
| 224 | pub fn as_run(&self) -> Option<&RunCapture> { |
| 225 | match self { |
| 226 | Self::Text(_) => None, |
| 227 | Self::Run(run) => Some(run), |
| 228 | } |
| 229 | } |
| 230 | } |
| 231 | |
| 232 | /// Final executable output. |
| 233 | #[derive(Debug, Clone)] |
| 234 | pub struct RunCapture { |
| 235 | pub exit_code: i32, |
| 236 | pub stdout: String, |
| 237 | pub stderr: String, |
| 238 | pub files: BTreeMap<String, Vec<u8>>, |
| 239 | } |
| 240 | |
| 241 | /// Capture the requested stages for one source file. |
| 242 | pub fn capture_from_path(request: &CaptureRequest) -> Result<CaptureResult, CaptureFailure> { |
| 243 | let mut stages = BTreeMap::new(); |
| 244 | let wants = |stage| request.requested.contains(&stage); |
| 245 | let needs_backend = request.requested.iter().any(|stage| { |
| 246 | matches!( |
| 247 | stage, |
| 248 | Stage::Mir | Stage::Regalloc | Stage::Asm | Stage::Obj | Stage::Run |
| 249 | ) |
| 250 | }); |
| 251 | let input = request.input.clone(); |
| 252 | let source_form = detect_source_form(&input.to_string_lossy()); |
| 253 | |
| 254 | let source = fs::read_to_string(&input).map_err(|e| CaptureFailure { |
| 255 | input: input.clone(), |
| 256 | opt_level: request.opt_level, |
| 257 | stage: FailureStage::Preprocess, |
| 258 | detail: format!("cannot read '{}': {}", input.display(), e), |
| 259 | stages: stages.clone(), |
| 260 | })?; |
| 261 | |
| 262 | let pp_config = crate::preprocess::PreprocConfig { |
| 263 | filename: input.to_string_lossy().into_owned(), |
| 264 | fixed_form: matches!(source_form, SourceForm::FixedForm), |
| 265 | ..crate::preprocess::PreprocConfig::default() |
| 266 | }; |
| 267 | let pp_result = |
| 268 | crate::preprocess::preprocess(&source, &pp_config).map_err(|e| CaptureFailure { |
| 269 | input: input.clone(), |
| 270 | opt_level: request.opt_level, |
| 271 | stage: FailureStage::Preprocess, |
| 272 | detail: e.to_string(), |
| 273 | stages: stages.clone(), |
| 274 | })?; |
| 275 | let preprocessed = pp_result.text; |
| 276 | if wants(Stage::Preprocess) { |
| 277 | stages.insert(Stage::Preprocess, CapturedStage::Text(preprocessed.clone())); |
| 278 | } |
| 279 | |
| 280 | let tokens = tokenize(&preprocessed, 0, source_form).map_err(|e| CaptureFailure { |
| 281 | input: input.clone(), |
| 282 | opt_level: request.opt_level, |
| 283 | stage: FailureStage::Lexer, |
| 284 | detail: format!( |
| 285 | "{}:{}: lexer error: {}", |
| 286 | input.display(), |
| 287 | e.span.start.line, |
| 288 | e.msg |
| 289 | ), |
| 290 | stages: stages.clone(), |
| 291 | })?; |
| 292 | if wants(Stage::Tokens) { |
| 293 | stages.insert(Stage::Tokens, CapturedStage::Text(format_tokens(&tokens))); |
| 294 | } |
| 295 | |
| 296 | let mut parser = Parser::new(&tokens); |
| 297 | let units = parser.parse_file().map_err(|e| CaptureFailure { |
| 298 | input: input.clone(), |
| 299 | opt_level: request.opt_level, |
| 300 | stage: FailureStage::Parser, |
| 301 | detail: format!( |
| 302 | "{}:{}:{}: parse error: {}", |
| 303 | input.display(), |
| 304 | e.span.start.line, |
| 305 | e.span.start.col, |
| 306 | e.msg |
| 307 | ), |
| 308 | stages: stages.clone(), |
| 309 | })?; |
| 310 | if wants(Stage::Ast) { |
| 311 | stages.insert(Stage::Ast, CapturedStage::Text(format!("{:#?}", units))); |
| 312 | } |
| 313 | |
| 314 | let rr = resolve::resolve_file(&units, &[]).map_err(|e| CaptureFailure { |
| 315 | input: input.clone(), |
| 316 | opt_level: request.opt_level, |
| 317 | stage: FailureStage::Sema, |
| 318 | detail: format!("{}:{}: {}", input.display(), e.span.start.line, e.msg), |
| 319 | stages: stages.clone(), |
| 320 | })?; |
| 321 | let st = rr.st; |
| 322 | let type_layouts = rr.type_layouts; |
| 323 | let diags = validate::validate_file(&units, &st); |
| 324 | if wants(Stage::Sema) { |
| 325 | stages.insert( |
| 326 | Stage::Sema, |
| 327 | CapturedStage::Text(format_sema_snapshot(&st, &type_layouts, &diags)), |
| 328 | ); |
| 329 | } |
| 330 | let sema_errors: Vec<_> = diags |
| 331 | .iter() |
| 332 | .filter(|d| d.kind == validate::DiagKind::Error) |
| 333 | .collect(); |
| 334 | if !sema_errors.is_empty() { |
| 335 | return Err(CaptureFailure { |
| 336 | input: input.clone(), |
| 337 | opt_level: request.opt_level, |
| 338 | stage: FailureStage::Sema, |
| 339 | detail: format_diagnostics(&input, &sema_errors), |
| 340 | stages, |
| 341 | }); |
| 342 | } |
| 343 | |
| 344 | let (ir_module, _module_globals) = lower::lower_file( |
| 345 | &units, |
| 346 | &st, |
| 347 | &type_layouts, |
| 348 | std::collections::HashMap::new(), |
| 349 | std::collections::HashMap::new(), |
| 350 | std::collections::HashMap::new(), |
| 351 | std::collections::HashMap::new(), |
| 352 | ); |
| 353 | let ir_errors = verify::verify_module(&ir_module); |
| 354 | if !ir_errors.is_empty() { |
| 355 | let msg = ir_errors |
| 356 | .iter() |
| 357 | .map(|e| e.to_string()) |
| 358 | .collect::<Vec<_>>() |
| 359 | .join("\n"); |
| 360 | return Err(CaptureFailure { |
| 361 | input: input.clone(), |
| 362 | opt_level: request.opt_level, |
| 363 | stage: FailureStage::Ir, |
| 364 | detail: format!("internal error: IR verification failed:\n{}", msg), |
| 365 | stages, |
| 366 | }); |
| 367 | } |
| 368 | |
| 369 | let ir_text = ir_printer::print_module(&ir_module); |
| 370 | if wants(Stage::Ir) { |
| 371 | stages.insert(Stage::Ir, CapturedStage::Text(ir_text.clone())); |
| 372 | } |
| 373 | let module_has_i128 = ir_module.contains_i128(); |
| 374 | let needs_optimized_pipeline = request.requested.iter().any(|stage| { |
| 375 | matches!( |
| 376 | stage, |
| 377 | Stage::OptIr | Stage::Mir | Stage::Regalloc | Stage::Asm | Stage::Obj | Stage::Run |
| 378 | ) |
| 379 | }) && request.opt_level != OptLevel::O0; |
| 380 | |
| 381 | let optimized_module = if needs_optimized_pipeline { |
| 382 | let mut optimized = ir_module.clone(); |
| 383 | let ir_opt = match request.opt_level { |
| 384 | OptLevel::O0 => IrOptLevel::O0, |
| 385 | OptLevel::O1 => IrOptLevel::O1, |
| 386 | OptLevel::O2 => IrOptLevel::O2, |
| 387 | OptLevel::O3 => IrOptLevel::O3, |
| 388 | OptLevel::Os => IrOptLevel::Os, |
| 389 | OptLevel::Ofast => IrOptLevel::Ofast, |
| 390 | }; |
| 391 | let pm = if ir_module.contains_i128_outside_globals() && request.opt_level != OptLevel::O0 { |
| 392 | build_i128_pipeline(ir_opt).ok_or_else(|| CaptureFailure { |
| 393 | input: input.clone(), |
| 394 | opt_level: request.opt_level, |
| 395 | stage: FailureStage::Ir, |
| 396 | detail: format!( |
| 397 | "integer(16) / i128 optimization at {} is not yet supported; capture raw IR or use a supported opt level", |
| 398 | request.opt_level.as_flag() |
| 399 | ), |
| 400 | stages: stages.clone(), |
| 401 | })? |
| 402 | } else { |
| 403 | build_pipeline(ir_opt) |
| 404 | }; |
| 405 | pm.run(&mut optimized); |
| 406 | Some(optimized) |
| 407 | } else { |
| 408 | None |
| 409 | }; |
| 410 | |
| 411 | if wants(Stage::OptIr) { |
| 412 | let opt_ir_text = if let Some(module) = optimized_module.as_ref() { |
| 413 | ir_printer::print_module(module) |
| 414 | } else { |
| 415 | ir_text.clone() |
| 416 | }; |
| 417 | stages.insert(Stage::OptIr, CapturedStage::Text(opt_ir_text)); |
| 418 | } |
| 419 | |
| 420 | if !needs_backend { |
| 421 | return Ok(CaptureResult { |
| 422 | input, |
| 423 | opt_level: request.opt_level, |
| 424 | stages, |
| 425 | }); |
| 426 | } |
| 427 | |
| 428 | let backend_ir = optimized_module.as_ref().unwrap_or(&ir_module); |
| 429 | if module_has_i128 && !backend_ir.i128_backend_o0_supported() { |
| 430 | return Err(CaptureFailure { |
| 431 | input: input.clone(), |
| 432 | opt_level: request.opt_level, |
| 433 | stage: FailureStage::Ir, |
| 434 | detail: |
| 435 | "backend does not yet support integer(16) / i128 codegen; capture IR at O0 for now" |
| 436 | .into(), |
| 437 | stages, |
| 438 | }); |
| 439 | } |
| 440 | let machine_funcs = isel::select_module(backend_ir); |
| 441 | if wants(Stage::Mir) { |
| 442 | stages.insert( |
| 443 | Stage::Mir, |
| 444 | CapturedStage::Text(format_machine_functions(&machine_funcs)), |
| 445 | ); |
| 446 | } |
| 447 | |
| 448 | let mut allocated = machine_funcs.clone(); |
| 449 | // Backend peephole at O2+ (must run BEFORE regalloc — it |
| 450 | // operates on vregs). |
| 451 | if request.opt_level >= OptLevel::O2 { |
| 452 | for mf in &mut allocated { |
| 453 | crate::codegen::peephole::run_peephole(mf); |
| 454 | } |
| 455 | } |
| 456 | for mf in &mut allocated { |
| 457 | let liveness = crate::codegen::liveness::compute_liveness(mf); |
| 458 | let result = linearscan::linear_scan(mf); |
| 459 | linearscan::apply_allocation(mf, &result, &liveness); |
| 460 | // Post-allocation passes — must mirror the driver pipeline so |
| 461 | // captured asm matches the binary the user actually ships. |
| 462 | // parallelize_call_arg_moves in particular fixes a w0/w1 |
| 463 | // clobber pattern visible at high register pressure. |
| 464 | linearscan::parallelize_entry_arg_moves(mf); |
| 465 | linearscan::parallelize_call_arg_moves(mf); |
| 466 | linearscan::insert_split_bridges(mf, &result.split_records); |
| 467 | linearscan::insert_callee_saves(mf, &result.callee_saved_used); |
| 468 | linearscan::coalesce_moves(mf); |
| 469 | if request.opt_level >= OptLevel::O1 { |
| 470 | crate::codegen::tailcall::tail_call_opt(mf); |
| 471 | } |
| 472 | crate::codegen::relax_branches::relax_branches(mf); |
| 473 | } |
| 474 | if wants(Stage::Regalloc) { |
| 475 | stages.insert( |
| 476 | Stage::Regalloc, |
| 477 | CapturedStage::Text(format_machine_functions(&allocated)), |
| 478 | ); |
| 479 | } |
| 480 | |
| 481 | let asm_text = emit_module_asm(backend_ir, &allocated); |
| 482 | if wants(Stage::Asm) { |
| 483 | stages.insert(Stage::Asm, CapturedStage::Text(asm_text.clone())); |
| 484 | } |
| 485 | |
| 486 | if wants(Stage::Obj) || wants(Stage::Run) { |
| 487 | let temp_root = next_temp_root("afs_tests_capture"); |
| 488 | let asm_path = temp_root.join("input.s"); |
| 489 | let obj_path = temp_root.join("output.o"); |
| 490 | let bin_path = temp_root.join("output"); |
| 491 | |
| 492 | fs::create_dir_all(&temp_root).map_err(|e| CaptureFailure { |
| 493 | input: input.clone(), |
| 494 | opt_level: request.opt_level, |
| 495 | stage: FailureStage::Obj, |
| 496 | detail: format!("cannot create temp dir '{}': {}", temp_root.display(), e), |
| 497 | stages: stages.clone(), |
| 498 | })?; |
| 499 | fs::write(&asm_path, &asm_text).map_err(|e| CaptureFailure { |
| 500 | input: input.clone(), |
| 501 | opt_level: request.opt_level, |
| 502 | stage: FailureStage::Obj, |
| 503 | detail: format!("cannot write temp assembly '{}': {}", asm_path.display(), e), |
| 504 | stages: stages.clone(), |
| 505 | })?; |
| 506 | |
| 507 | assemble_with_system(&asm_path, &obj_path).map_err(|detail| CaptureFailure { |
| 508 | input: input.clone(), |
| 509 | opt_level: request.opt_level, |
| 510 | stage: FailureStage::Obj, |
| 511 | detail, |
| 512 | stages: stages.clone(), |
| 513 | })?; |
| 514 | |
| 515 | if wants(Stage::Obj) { |
| 516 | let snapshot = object_snapshot(&obj_path).map_err(|detail| CaptureFailure { |
| 517 | input: input.clone(), |
| 518 | opt_level: request.opt_level, |
| 519 | stage: FailureStage::Obj, |
| 520 | detail, |
| 521 | stages: stages.clone(), |
| 522 | })?; |
| 523 | stages.insert(Stage::Obj, CapturedStage::Text(snapshot)); |
| 524 | } |
| 525 | |
| 526 | if wants(Stage::Run) { |
| 527 | link_with_runtime(&obj_path, &bin_path).map_err(|detail| CaptureFailure { |
| 528 | input: input.clone(), |
| 529 | opt_level: request.opt_level, |
| 530 | stage: FailureStage::Run, |
| 531 | detail, |
| 532 | stages: stages.clone(), |
| 533 | })?; |
| 534 | let sandbox = temp_root.join("run_sandbox"); |
| 535 | fs::create_dir_all(&sandbox).map_err(|e| CaptureFailure { |
| 536 | input: input.clone(), |
| 537 | opt_level: request.opt_level, |
| 538 | stage: FailureStage::Run, |
| 539 | detail: format!("cannot create run sandbox '{}': {}", sandbox.display(), e), |
| 540 | stages: stages.clone(), |
| 541 | })?; |
| 542 | let output = Command::new(&bin_path) |
| 543 | .current_dir(&sandbox) |
| 544 | .output() |
| 545 | .map_err(|e| CaptureFailure { |
| 546 | input: input.clone(), |
| 547 | opt_level: request.opt_level, |
| 548 | stage: FailureStage::Run, |
| 549 | detail: format!("cannot run '{}': {}", bin_path.display(), e), |
| 550 | stages: stages.clone(), |
| 551 | })?; |
| 552 | let files = snapshot_sandbox_files(&sandbox).map_err(|detail| CaptureFailure { |
| 553 | input: input.clone(), |
| 554 | opt_level: request.opt_level, |
| 555 | stage: FailureStage::Run, |
| 556 | detail, |
| 557 | stages: stages.clone(), |
| 558 | })?; |
| 559 | stages.insert( |
| 560 | Stage::Run, |
| 561 | CapturedStage::Run(RunCapture { |
| 562 | exit_code: output.status.code().unwrap_or(-1), |
| 563 | stdout: String::from_utf8_lossy(&output.stdout).into_owned(), |
| 564 | stderr: String::from_utf8_lossy(&output.stderr).into_owned(), |
| 565 | files, |
| 566 | }), |
| 567 | ); |
| 568 | } |
| 569 | |
| 570 | let _ = fs::remove_dir_all(&temp_root); |
| 571 | } |
| 572 | |
| 573 | Ok(CaptureResult { |
| 574 | input, |
| 575 | opt_level: request.opt_level, |
| 576 | stages, |
| 577 | }) |
| 578 | } |
| 579 | |
| 580 | fn collect_sandbox_files( |
| 581 | root: &Path, |
| 582 | dir: &Path, |
| 583 | out: &mut BTreeMap<String, Vec<u8>>, |
| 584 | ) -> std::io::Result<()> { |
| 585 | for entry in fs::read_dir(dir)? { |
| 586 | let entry = entry?; |
| 587 | let path = entry.path(); |
| 588 | if entry.file_type()?.is_dir() { |
| 589 | collect_sandbox_files(root, &path, out)?; |
| 590 | } else { |
| 591 | let rel = path.strip_prefix(root).unwrap(); |
| 592 | out.insert(rel.to_string_lossy().replace('\\', "/"), fs::read(&path)?); |
| 593 | } |
| 594 | } |
| 595 | Ok(()) |
| 596 | } |
| 597 | |
| 598 | fn snapshot_sandbox_files(sandbox: &Path) -> Result<BTreeMap<String, Vec<u8>>, String> { |
| 599 | let mut files = BTreeMap::new(); |
| 600 | collect_sandbox_files(sandbox, sandbox, &mut files) |
| 601 | .map_err(|e| format!("cannot snapshot sandbox '{}': {}", sandbox.display(), e))?; |
| 602 | Ok(files) |
| 603 | } |
| 604 | |
| 605 | fn format_tokens(tokens: &[Token]) -> String { |
| 606 | let mut out = String::new(); |
| 607 | for token in tokens { |
| 608 | let _ = writeln!( |
| 609 | out, |
| 610 | "{}:{}-{}:{}\t{}\t{:?}", |
| 611 | token.span.start.line, |
| 612 | token.span.start.col, |
| 613 | token.span.end.line, |
| 614 | token.span.end.col, |
| 615 | token.kind, |
| 616 | token.text |
| 617 | ); |
| 618 | } |
| 619 | out |
| 620 | } |
| 621 | |
| 622 | fn format_sema_snapshot( |
| 623 | st: &crate::sema::symtab::SymbolTable, |
| 624 | type_layouts: &crate::sema::type_layout::TypeLayoutRegistry, |
| 625 | diags: &[validate::Diagnostic], |
| 626 | ) -> String { |
| 627 | let mut out = String::new(); |
| 628 | if diags.is_empty() { |
| 629 | let _ = writeln!(out, "diagnostics: none"); |
| 630 | } else { |
| 631 | let _ = writeln!(out, "diagnostics:"); |
| 632 | for diag in diags { |
| 633 | let _ = writeln!(out, " {}", diag); |
| 634 | } |
| 635 | } |
| 636 | let _ = writeln!(out, "\nsymbol_table:\n{:#?}", st); |
| 637 | let _ = writeln!(out, "\ntype_layouts:\n{:#?}", type_layouts); |
| 638 | out |
| 639 | } |
| 640 | |
| 641 | fn format_diagnostics(path: &Path, diags: &[&validate::Diagnostic]) -> String { |
| 642 | diags |
| 643 | .iter() |
| 644 | .map(|diag| { |
| 645 | format!( |
| 646 | "{}:{}:{}: {}", |
| 647 | path.display(), |
| 648 | diag.span.start.line, |
| 649 | diag.span.start.col, |
| 650 | diag.msg |
| 651 | ) |
| 652 | }) |
| 653 | .collect::<Vec<_>>() |
| 654 | .join("\n") |
| 655 | } |
| 656 | |
| 657 | fn format_machine_functions(machine_funcs: &[MachineFunction]) -> String { |
| 658 | let mut out = String::new(); |
| 659 | for (index, mf) in machine_funcs.iter().enumerate() { |
| 660 | if index > 0 { |
| 661 | out.push('\n'); |
| 662 | } |
| 663 | |
| 664 | let _ = writeln!(out, "function {}:", mf.name); |
| 665 | let _ = writeln!(out, " frame_size: {}", mf.frame.size); |
| 666 | |
| 667 | if mf.frame.locals.is_empty() { |
| 668 | let _ = writeln!(out, " locals: none"); |
| 669 | } else { |
| 670 | let locals = mf |
| 671 | .frame |
| 672 | .locals |
| 673 | .iter() |
| 674 | .map(|slot| format!("[fp{:+}]:{}", slot.offset, slot.size)) |
| 675 | .collect::<Vec<_>>() |
| 676 | .join(", "); |
| 677 | let _ = writeln!(out, " locals: {}", locals); |
| 678 | } |
| 679 | |
| 680 | if mf.vregs.is_empty() { |
| 681 | let _ = writeln!(out, " vregs: none"); |
| 682 | } else { |
| 683 | let vregs = mf |
| 684 | .vregs |
| 685 | .iter() |
| 686 | .map(|vreg| format!("v{}:{}", vreg.id.0, format_reg_class(vreg.class))) |
| 687 | .collect::<Vec<_>>() |
| 688 | .join(", "); |
| 689 | let _ = writeln!(out, " vregs: {}", vregs); |
| 690 | } |
| 691 | |
| 692 | for block in &mf.blocks { |
| 693 | let _ = writeln!(out, " block {}:", block.label); |
| 694 | if block.insts.is_empty() { |
| 695 | let _ = writeln!(out, " <empty>"); |
| 696 | } else { |
| 697 | for inst in &block.insts { |
| 698 | let _ = writeln!(out, " {}", format_machine_inst(mf, inst)); |
| 699 | } |
| 700 | } |
| 701 | } |
| 702 | |
| 703 | if mf.const_pool.is_empty() { |
| 704 | let _ = writeln!(out, " const_pool: none"); |
| 705 | } else { |
| 706 | let _ = writeln!(out, " const_pool:"); |
| 707 | for (pool_index, entry) in mf.const_pool.iter().enumerate() { |
| 708 | let _ = writeln!( |
| 709 | out, |
| 710 | " [{}] {}", |
| 711 | pool_index, |
| 712 | format_const_pool_entry(entry) |
| 713 | ); |
| 714 | } |
| 715 | } |
| 716 | } |
| 717 | out |
| 718 | } |
| 719 | |
| 720 | fn format_machine_inst(mf: &MachineFunction, inst: &MachineInst) -> String { |
| 721 | let opcode = format!("{:?}", inst.opcode).to_ascii_lowercase(); |
| 722 | let operands = inst |
| 723 | .operands |
| 724 | .iter() |
| 725 | .map(|operand| format_machine_operand(mf, operand)) |
| 726 | .collect::<Vec<_>>(); |
| 727 | if operands.is_empty() { |
| 728 | opcode |
| 729 | } else { |
| 730 | format!("{} {}", opcode, operands.join(", ")) |
| 731 | } |
| 732 | } |
| 733 | |
| 734 | fn format_machine_operand(mf: &MachineFunction, operand: &MachineOperand) -> String { |
| 735 | match operand { |
| 736 | MachineOperand::VReg(vreg) => format!("v{}", vreg.0), |
| 737 | MachineOperand::PhysReg(reg) => format_phys_reg(*reg), |
| 738 | MachineOperand::Imm(value) => { |
| 739 | if *value == -1 { |
| 740 | "#frame-16".to_string() |
| 741 | } else { |
| 742 | format!("#{}", value) |
| 743 | } |
| 744 | } |
| 745 | MachineOperand::FrameSlot(offset) => format!("[fp{:+}]", offset), |
| 746 | MachineOperand::Cond(cond) => format_cond(*cond).to_string(), |
| 747 | MachineOperand::BlockRef(block_id) => block_label(mf, *block_id), |
| 748 | MachineOperand::Extern(name) => format!("extern {}", name), |
| 749 | MachineOperand::ConstPool(index) => format!("constpool[{}]", index), |
| 750 | MachineOperand::Shift(bits) => format!("lsl #{}", bits), |
| 751 | MachineOperand::GlobalLabel(name) => format!("global:{}", name), |
| 752 | } |
| 753 | } |
| 754 | |
| 755 | fn format_phys_reg(reg: PhysReg) -> String { |
| 756 | match reg { |
| 757 | PhysReg::Gp(num) => format!("x{}", num), |
| 758 | PhysReg::Gp32(num) => format!("w{}", num), |
| 759 | PhysReg::Fp(num) => format!("d{}", num), |
| 760 | PhysReg::Fp32(num) => format!("s{}", num), |
| 761 | PhysReg::Sp => "sp".to_string(), |
| 762 | PhysReg::Xzr => "xzr".to_string(), |
| 763 | PhysReg::Wzr => "wzr".to_string(), |
| 764 | } |
| 765 | } |
| 766 | |
| 767 | fn format_reg_class(class: RegClass) -> &'static str { |
| 768 | match class { |
| 769 | RegClass::Gp64 => "gp64", |
| 770 | RegClass::Gp32 => "gp32", |
| 771 | RegClass::Fp64 => "fp64", |
| 772 | RegClass::Fp32 => "fp32", |
| 773 | RegClass::V128 => "v128", |
| 774 | } |
| 775 | } |
| 776 | |
| 777 | fn format_cond(cond: ArmCond) -> &'static str { |
| 778 | match cond { |
| 779 | ArmCond::Eq => "eq", |
| 780 | ArmCond::Ne => "ne", |
| 781 | ArmCond::Hs => "hs", |
| 782 | ArmCond::Lo => "lo", |
| 783 | ArmCond::Mi => "mi", |
| 784 | ArmCond::Pl => "pl", |
| 785 | ArmCond::Hi => "hi", |
| 786 | ArmCond::Ls => "ls", |
| 787 | ArmCond::Ge => "ge", |
| 788 | ArmCond::Lt => "lt", |
| 789 | ArmCond::Gt => "gt", |
| 790 | ArmCond::Le => "le", |
| 791 | } |
| 792 | } |
| 793 | |
| 794 | fn block_label(mf: &MachineFunction, block_id: MBlockId) -> String { |
| 795 | mf.blocks |
| 796 | .iter() |
| 797 | .find(|block| block.id == block_id) |
| 798 | .map(|block| block.label.clone()) |
| 799 | .unwrap_or_else(|| format!("block#{}", block_id.0)) |
| 800 | } |
| 801 | |
| 802 | fn format_const_pool_entry(entry: &ConstPoolEntry) -> String { |
| 803 | match entry { |
| 804 | ConstPoolEntry::F32(value) => format!("f32 {}", value), |
| 805 | ConstPoolEntry::F64(value) => format!("f64 {}", value), |
| 806 | ConstPoolEntry::I64(value) => format!("i64 {}", value), |
| 807 | ConstPoolEntry::Bytes(bytes) => format!("bytes {:?}", escape_const_bytes(bytes)), |
| 808 | } |
| 809 | } |
| 810 | |
| 811 | fn escape_const_bytes(bytes: &[u8]) -> String { |
| 812 | let mut out = String::new(); |
| 813 | for &byte in bytes { |
| 814 | match byte { |
| 815 | b'\\' => out.push_str("\\\\"), |
| 816 | b'"' => out.push_str("\\\""), |
| 817 | b'\n' => out.push_str("\\n"), |
| 818 | b'\t' => out.push_str("\\t"), |
| 819 | b if b.is_ascii_graphic() || b == b' ' => out.push(b as char), |
| 820 | other => { |
| 821 | let _ = write!(out, "\\x{:02x}", other); |
| 822 | } |
| 823 | } |
| 824 | } |
| 825 | out |
| 826 | } |
| 827 | |
| 828 | fn emit_module_asm(module: &crate::ir::inst::Module, allocated: &[MachineFunction]) -> String { |
| 829 | let mut asm_text = String::new(); |
| 830 | asm_text.push_str(".section __TEXT,__text,regular,pure_instructions\n"); |
| 831 | for mf in allocated { |
| 832 | asm_text.push_str(".section __TEXT,__text,regular,pure_instructions\n"); |
| 833 | asm_text.push_str(&emit::emit_function(mf)); |
| 834 | asm_text.push('\n'); |
| 835 | } |
| 836 | |
| 837 | if !module.globals.is_empty() { |
| 838 | asm_text.push_str(&emit::emit_globals(&module.globals)); |
| 839 | asm_text.push('\n'); |
| 840 | } |
| 841 | |
| 842 | if let Some(user_func) = main_wrapper_target(allocated) { |
| 843 | if user_func != "main" { |
| 844 | let _ = write!( |
| 845 | asm_text, |
| 846 | "\n.section __TEXT,__text,regular,pure_instructions\n\ |
| 847 | .globl _main\n\ |
| 848 | .p2align 2\n\ |
| 849 | _main:\n\ |
| 850 | stp x29, x30, [sp, #-16]!\n\ |
| 851 | mov x29, sp\n\ |
| 852 | bl _afs_program_init\n\ |
| 853 | bl _{}\n\ |
| 854 | bl _afs_program_finalize\n\ |
| 855 | mov x0, #0\n\ |
| 856 | ldp x29, x30, [sp], #16\n\ |
| 857 | ret\n", |
| 858 | user_func |
| 859 | ); |
| 860 | } |
| 861 | } |
| 862 | |
| 863 | asm_text |
| 864 | } |
| 865 | |
| 866 | fn main_wrapper_target(allocated: &[MachineFunction]) -> Option<&str> { |
| 867 | allocated |
| 868 | .iter() |
| 869 | .find(|func| func.name.starts_with("__prog_")) |
| 870 | .or_else(|| allocated.iter().find(|func| func.name != "main")) |
| 871 | .map(|func| func.name.as_str()) |
| 872 | } |
| 873 | |
| 874 | fn next_temp_root(prefix: &str) -> PathBuf { |
| 875 | let id = CAPTURE_COUNTER.fetch_add(1, Ordering::Relaxed); |
| 876 | std::env::temp_dir().join(format!("{}_{}_{}", prefix, std::process::id(), id)) |
| 877 | } |
| 878 | |
| 879 | fn assemble_with_system(asm_path: &Path, obj_path: &Path) -> Result<(), String> { |
| 880 | let output = Command::new("as") |
| 881 | .args([ |
| 882 | "-o", |
| 883 | obj_path.to_str().unwrap_or("output.o"), |
| 884 | asm_path.to_str().unwrap_or("input.s"), |
| 885 | ]) |
| 886 | .output() |
| 887 | .map_err(|e| format!("cannot run assembler: {}", e))?; |
| 888 | if output.status.success() { |
| 889 | Ok(()) |
| 890 | } else { |
| 891 | Err(format!( |
| 892 | "assembler failed:\n{}", |
| 893 | String::from_utf8_lossy(&output.stderr) |
| 894 | )) |
| 895 | } |
| 896 | } |
| 897 | |
| 898 | fn link_with_runtime(obj: &Path, output: &Path) -> Result<(), String> { |
| 899 | let rt_path = find_runtime_lib()?; |
| 900 | let sdk = Command::new("xcrun") |
| 901 | .args(["--show-sdk-path"]) |
| 902 | .output() |
| 903 | .map_err(|e| format!("cannot run xcrun: {}", e))?; |
| 904 | if !sdk.status.success() { |
| 905 | return Err(format!( |
| 906 | "xcrun failed:\n{}", |
| 907 | String::from_utf8_lossy(&sdk.stderr) |
| 908 | )); |
| 909 | } |
| 910 | let sysroot = String::from_utf8_lossy(&sdk.stdout).trim().to_string(); |
| 911 | |
| 912 | let ld = Command::new("ld") |
| 913 | .args([ |
| 914 | obj.to_str().unwrap(), |
| 915 | &rt_path, |
| 916 | "-lSystem", |
| 917 | "-syslibroot", |
| 918 | &sysroot, |
| 919 | "-e", |
| 920 | "_main", |
| 921 | "-o", |
| 922 | output.to_str().unwrap(), |
| 923 | ]) |
| 924 | .output() |
| 925 | .map_err(|e| format!("cannot run linker: {}", e))?; |
| 926 | if ld.status.success() { |
| 927 | Ok(()) |
| 928 | } else { |
| 929 | Err(format!( |
| 930 | "linker failed:\n{}", |
| 931 | String::from_utf8_lossy(&ld.stderr) |
| 932 | )) |
| 933 | } |
| 934 | } |
| 935 | |
| 936 | fn find_runtime_lib() -> Result<String, String> { |
| 937 | if let Some(workspace_root) = find_workspace_root() { |
| 938 | maybe_refresh_runtime_lib(&workspace_root)?; |
| 939 | for candidate in [ |
| 940 | workspace_root.join("target/debug/libarmfortas_rt.a"), |
| 941 | workspace_root.join("target/release/libarmfortas_rt.a"), |
| 942 | ] { |
| 943 | if candidate.exists() { |
| 944 | return Ok(candidate.to_string_lossy().into_owned()); |
| 945 | } |
| 946 | } |
| 947 | } |
| 948 | |
| 949 | if let Ok(exe) = std::env::current_exe() { |
| 950 | if let Some(dir) = exe.parent() { |
| 951 | let candidate = dir.join("libarmfortas_rt.a"); |
| 952 | if candidate.exists() { |
| 953 | return Ok(candidate.to_string_lossy().into_owned()); |
| 954 | } |
| 955 | } |
| 956 | } |
| 957 | |
| 958 | Err("cannot find libarmfortas_rt.a — build with 'cargo build -p armfortas-rt'".into()) |
| 959 | } |
| 960 | |
| 961 | fn maybe_refresh_runtime_lib(workspace_root: &Path) -> Result<(), String> { |
| 962 | let runtime_dir = workspace_root.join("runtime"); |
| 963 | if !runtime_dir.join("Cargo.toml").exists() { |
| 964 | return Ok(()); |
| 965 | } |
| 966 | |
| 967 | let Some(source_mtime) = newest_mtime(&runtime_dir) else { |
| 968 | return Ok(()); |
| 969 | }; |
| 970 | let debug_archive = workspace_root.join("target/debug/libarmfortas_rt.a"); |
| 971 | let archive_mtime = fs::metadata(&debug_archive) |
| 972 | .ok() |
| 973 | .and_then(|meta| meta.modified().ok()); |
| 974 | |
| 975 | if archive_mtime.is_some_and(|mtime| mtime >= source_mtime) { |
| 976 | return Ok(()); |
| 977 | } |
| 978 | |
| 979 | let cargo = std::env::var("CARGO").unwrap_or_else(|_| "cargo".into()); |
| 980 | let output = Command::new(cargo) |
| 981 | .current_dir(workspace_root) |
| 982 | .args(["build", "-p", "armfortas-rt"]) |
| 983 | .output() |
| 984 | .map_err(|e| format!("cannot rebuild libarmfortas_rt.a: {}", e))?; |
| 985 | if output.status.success() { |
| 986 | Ok(()) |
| 987 | } else { |
| 988 | Err(format!( |
| 989 | "cannot rebuild libarmfortas_rt.a:\n{}", |
| 990 | String::from_utf8_lossy(&output.stderr) |
| 991 | )) |
| 992 | } |
| 993 | } |
| 994 | |
| 995 | fn newest_mtime(path: &Path) -> Option<SystemTime> { |
| 996 | let meta = fs::metadata(path).ok()?; |
| 997 | let mut newest = meta.modified().ok()?; |
| 998 | if meta.is_dir() { |
| 999 | for entry in fs::read_dir(path).ok()? { |
| 1000 | let entry = entry.ok()?; |
| 1001 | let child = newest_mtime(&entry.path())?; |
| 1002 | if child > newest { |
| 1003 | newest = child; |
| 1004 | } |
| 1005 | } |
| 1006 | } |
| 1007 | Some(newest) |
| 1008 | } |
| 1009 | |
| 1010 | fn find_workspace_root() -> Option<PathBuf> { |
| 1011 | let mut bases = Vec::new(); |
| 1012 | if let Ok(cwd) = std::env::current_dir() { |
| 1013 | bases.push(cwd); |
| 1014 | } |
| 1015 | if let Ok(exe) = std::env::current_exe() { |
| 1016 | if let Some(dir) = exe.parent() { |
| 1017 | bases.push(dir.to_path_buf()); |
| 1018 | } |
| 1019 | } |
| 1020 | |
| 1021 | for base in bases { |
| 1022 | for ancestor in base.ancestors() { |
| 1023 | if ancestor.join("Cargo.toml").exists() && ancestor.join("runtime/Cargo.toml").exists() |
| 1024 | { |
| 1025 | return Some(ancestor.to_path_buf()); |
| 1026 | } |
| 1027 | } |
| 1028 | } |
| 1029 | None |
| 1030 | } |
| 1031 | |
| 1032 | fn object_snapshot(path: &Path) -> Result<String, String> { |
| 1033 | let text = normalize_tool_output(&tool_output("otool", &["-t", path.to_str().unwrap()])?); |
| 1034 | let load = normalize_tool_output(&tool_output("otool", &["-l", path.to_str().unwrap()])?); |
| 1035 | let relocs = normalize_tool_output(&tool_output("otool", &["-rv", path.to_str().unwrap()])?); |
| 1036 | let symbols = normalize_tool_output(&tool_output("nm", &["-m", path.to_str().unwrap()])?); |
| 1037 | |
| 1038 | Ok(format!( |
| 1039 | "== text ==\n{}\n\n== load_commands ==\n{}\n\n== relocations ==\n{}\n\n== symbols ==\n{}", |
| 1040 | text, load, relocs, symbols |
| 1041 | )) |
| 1042 | } |
| 1043 | |
| 1044 | fn tool_output(tool: &str, args: &[&str]) -> Result<String, String> { |
| 1045 | let output = Command::new(tool) |
| 1046 | .args(args) |
| 1047 | .output() |
| 1048 | .map_err(|e| format!("cannot run {}: {}", tool, e))?; |
| 1049 | if output.status.success() { |
| 1050 | Ok(String::from_utf8_lossy(&output.stdout).into_owned()) |
| 1051 | } else { |
| 1052 | Err(format!( |
| 1053 | "{} failed:\n{}", |
| 1054 | tool, |
| 1055 | String::from_utf8_lossy(&output.stderr) |
| 1056 | )) |
| 1057 | } |
| 1058 | } |
| 1059 | |
| 1060 | fn normalize_tool_output(text: &str) -> String { |
| 1061 | text.lines() |
| 1062 | .filter(|line| !line.trim_end().ends_with(".o:")) |
| 1063 | .map(str::trim_end) |
| 1064 | .collect::<Vec<_>>() |
| 1065 | .join("\n") |
| 1066 | } |
| 1067 |