1 use crate::config::Config; 2 use crate::cranelift_arbitrary::CraneliftArbitrary; 3 use anyhow::Result; 4 use arbitrary::{Arbitrary, Unstructured}; 5 use cranelift::codegen::data_value::DataValue; 6 use cranelift::codegen::ir::immediates::Offset32; 7 use cranelift::codegen::ir::instructions::{InstructionFormat, ResolvedConstraint}; 8 use cranelift::codegen::ir::stackslot::StackSize; 9 10 use cranelift::codegen::ir::{ 11 types::*, AtomicRmwOp, Block, ConstantData, Endianness, ExternalName, FuncRef, Function, 12 LibCall, Opcode, SigRef, Signature, StackSlot, Type, UserExternalName, UserFuncName, Value, 13 }; 14 use cranelift::codegen::isa::CallConv; 15 use cranelift::frontend::{FunctionBuilder, FunctionBuilderContext, Switch, Variable}; 16 use cranelift::prelude::isa::OwnedTargetIsa; 17 use cranelift::prelude::{ 18 EntityRef, ExtFuncData, FloatCC, InstBuilder, IntCC, JumpTableData, MemFlags, StackSlotData, 19 StackSlotKind, 20 }; 21 use once_cell::sync::Lazy; 22 use std::collections::HashMap; 23 use std::ops::RangeInclusive; 24 use target_lexicon::{Architecture, Triple}; 25 26 type BlockSignature = Vec<Type>; 27 28 fn insert_opcode( 29 fgen: &mut FunctionGenerator, 30 builder: &mut FunctionBuilder, 31 opcode: Opcode, 32 args: &[Type], 33 rets: &[Type], 34 ) -> Result<()> { 35 let mut vals = Vec::with_capacity(args.len()); 36 for &arg in args.into_iter() { 37 let var = fgen.get_variable_of_type(arg)?; 38 let val = builder.use_var(var); 39 vals.push(val); 40 } 41 42 // Some opcodes require us to look at their input arguments to determine the 43 // controlling type. This is not the general case, but we can neatly check this 44 // using `requires_typevar_operand`. 45 let ctrl_type = if opcode.constraints().requires_typevar_operand() { 46 args.first() 47 } else { 48 rets.first() 49 } 50 .copied() 51 .unwrap_or(INVALID); 52 53 // Choose the appropriate instruction format for this opcode 54 let (inst, dfg) = match opcode.format() { 55 InstructionFormat::NullAry => builder.ins().NullAry(opcode, ctrl_type), 56 InstructionFormat::Unary => builder.ins().Unary(opcode, ctrl_type, vals[0]), 57 InstructionFormat::Binary => builder.ins().Binary(opcode, ctrl_type, vals[0], vals[1]), 58 InstructionFormat::Ternary => builder 59 .ins() 60 .Ternary(opcode, ctrl_type, vals[0], vals[1], vals[2]), 61 _ => unimplemented!(), 62 }; 63 let results = dfg.inst_results(inst).to_vec(); 64 65 for (val, &ty) in results.into_iter().zip(rets) { 66 let var = fgen.get_variable_of_type(ty)?; 67 builder.def_var(var, val); 68 } 69 Ok(()) 70 } 71 72 fn insert_call_to_function( 73 fgen: &mut FunctionGenerator, 74 builder: &mut FunctionBuilder, 75 call_opcode: Opcode, 76 sig: &Signature, 77 sig_ref: SigRef, 78 func_ref: FuncRef, 79 ) -> Result<()> { 80 let actuals = fgen.generate_values_for_signature( 81 builder, 82 sig.params.iter().map(|abi_param| abi_param.value_type), 83 )?; 84 85 let addr_ty = fgen.isa.pointer_type(); 86 let call = match call_opcode { 87 Opcode::Call => builder.ins().call(func_ref, &actuals), 88 Opcode::ReturnCall => builder.ins().return_call(func_ref, &actuals), 89 Opcode::CallIndirect => { 90 let addr = builder.ins().func_addr(addr_ty, func_ref); 91 builder.ins().call_indirect(sig_ref, addr, &actuals) 92 } 93 Opcode::ReturnCallIndirect => { 94 let addr = builder.ins().func_addr(addr_ty, func_ref); 95 builder.ins().return_call_indirect(sig_ref, addr, &actuals) 96 } 97 _ => unreachable!(), 98 }; 99 100 // Assign the return values to random variables 101 let ret_values = builder.inst_results(call).to_vec(); 102 let ret_types = sig.returns.iter().map(|p| p.value_type); 103 for (ty, val) in ret_types.zip(ret_values) { 104 let var = fgen.get_variable_of_type(ty)?; 105 builder.def_var(var, val); 106 } 107 108 Ok(()) 109 } 110 111 fn insert_call( 112 fgen: &mut FunctionGenerator, 113 builder: &mut FunctionBuilder, 114 opcode: Opcode, 115 _args: &[Type], 116 _rets: &[Type], 117 ) -> Result<()> { 118 assert!(matches!(opcode, Opcode::Call | Opcode::CallIndirect)); 119 let (sig, sig_ref, func_ref) = fgen.u.choose(&fgen.resources.func_refs)?.clone(); 120 121 insert_call_to_function(fgen, builder, opcode, &sig, sig_ref, func_ref) 122 } 123 124 fn insert_stack_load( 125 fgen: &mut FunctionGenerator, 126 builder: &mut FunctionBuilder, 127 _opcode: Opcode, 128 _args: &[Type], 129 rets: &[Type], 130 ) -> Result<()> { 131 let typevar = rets[0]; 132 let type_size = typevar.bytes(); 133 let (slot, slot_size) = fgen.stack_slot_with_size(type_size)?; 134 let offset = fgen.u.int_in_range(0..=(slot_size - type_size))? as i32; 135 136 let val = builder.ins().stack_load(typevar, slot, offset); 137 let var = fgen.get_variable_of_type(typevar)?; 138 builder.def_var(var, val); 139 140 Ok(()) 141 } 142 143 fn insert_stack_store( 144 fgen: &mut FunctionGenerator, 145 builder: &mut FunctionBuilder, 146 _opcode: Opcode, 147 args: &[Type], 148 _rets: &[Type], 149 ) -> Result<()> { 150 let typevar = args[0]; 151 let type_size = typevar.bytes(); 152 let (slot, slot_size) = fgen.stack_slot_with_size(type_size)?; 153 let offset = fgen.u.int_in_range(0..=(slot_size - type_size))? as i32; 154 155 let arg0 = fgen.get_variable_of_type(typevar)?; 156 let arg0 = builder.use_var(arg0); 157 158 builder.ins().stack_store(arg0, slot, offset); 159 Ok(()) 160 } 161 162 fn insert_cmp( 163 fgen: &mut FunctionGenerator, 164 builder: &mut FunctionBuilder, 165 opcode: Opcode, 166 args: &[Type], 167 rets: &[Type], 168 ) -> Result<()> { 169 let lhs = fgen.get_variable_of_type(args[0])?; 170 let lhs = builder.use_var(lhs); 171 172 let rhs = fgen.get_variable_of_type(args[1])?; 173 let rhs = builder.use_var(rhs); 174 175 let res = if opcode == Opcode::Fcmp { 176 let cc = *fgen.u.choose(FloatCC::all())?; 177 178 // We filter out condition codes that aren't supported by the target at 179 // this point after randomly choosing one, instead of randomly choosing a 180 // supported one, to avoid invalidating the corpus when these get implemented. 181 let unimplemented_cc = match (fgen.isa.triple().architecture, cc) { 182 // Some FloatCC's are not implemented on AArch64, see: 183 // https://github.com/bytecodealliance/wasmtime/issues/4850 184 (Architecture::Aarch64(_), FloatCC::OrderedNotEqual) => true, 185 (Architecture::Aarch64(_), FloatCC::UnorderedOrEqual) => true, 186 (Architecture::Aarch64(_), FloatCC::UnorderedOrLessThan) => true, 187 (Architecture::Aarch64(_), FloatCC::UnorderedOrLessThanOrEqual) => true, 188 (Architecture::Aarch64(_), FloatCC::UnorderedOrGreaterThan) => true, 189 (Architecture::Aarch64(_), FloatCC::UnorderedOrGreaterThanOrEqual) => true, 190 191 // These are not implemented on x86_64, for vectors. 192 (Architecture::X86_64, FloatCC::UnorderedOrEqual | FloatCC::OrderedNotEqual) => { 193 args[0].is_vector() 194 } 195 _ => false, 196 }; 197 if unimplemented_cc { 198 return Err(arbitrary::Error::IncorrectFormat.into()); 199 } 200 201 builder.ins().fcmp(cc, lhs, rhs) 202 } else { 203 let cc = *fgen.u.choose(IntCC::all())?; 204 builder.ins().icmp(cc, lhs, rhs) 205 }; 206 207 let var = fgen.get_variable_of_type(rets[0])?; 208 builder.def_var(var, res); 209 210 Ok(()) 211 } 212 213 fn insert_const( 214 fgen: &mut FunctionGenerator, 215 builder: &mut FunctionBuilder, 216 _opcode: Opcode, 217 _args: &[Type], 218 rets: &[Type], 219 ) -> Result<()> { 220 let typevar = rets[0]; 221 let var = fgen.get_variable_of_type(typevar)?; 222 let val = fgen.generate_const(builder, typevar)?; 223 builder.def_var(var, val); 224 Ok(()) 225 } 226 227 fn insert_bitcast( 228 fgen: &mut FunctionGenerator, 229 builder: &mut FunctionBuilder, 230 args: &[Type], 231 rets: &[Type], 232 ) -> Result<()> { 233 let from_var = fgen.get_variable_of_type(args[0])?; 234 let from_val = builder.use_var(from_var); 235 236 let to_var = fgen.get_variable_of_type(rets[0])?; 237 238 // TODO: We can generate little/big endian flags here. 239 let mut memflags = MemFlags::new(); 240 241 // When bitcasting between vectors of different lane counts, we need to 242 // specify the endianness. 243 if args[0].lane_count() != rets[0].lane_count() { 244 memflags.set_endianness(Endianness::Little); 245 } 246 247 let res = builder.ins().bitcast(rets[0], memflags, from_val); 248 builder.def_var(to_var, res); 249 Ok(()) 250 } 251 252 fn insert_load_store( 253 fgen: &mut FunctionGenerator, 254 builder: &mut FunctionBuilder, 255 opcode: Opcode, 256 args: &[Type], 257 rets: &[Type], 258 ) -> Result<()> { 259 if opcode == Opcode::Bitcast { 260 return insert_bitcast(fgen, builder, args, rets); 261 } 262 263 let ctrl_type = *rets.first().or(args.first()).unwrap(); 264 let type_size = ctrl_type.bytes(); 265 266 let is_atomic = [Opcode::AtomicLoad, Opcode::AtomicStore].contains(&opcode); 267 let (address, flags, offset) = 268 fgen.generate_address_and_memflags(builder, type_size, is_atomic)?; 269 270 // The variable being loaded or stored into 271 let var = fgen.get_variable_of_type(ctrl_type)?; 272 273 match opcode.format() { 274 InstructionFormat::LoadNoOffset => { 275 let (inst, dfg) = builder 276 .ins() 277 .LoadNoOffset(opcode, ctrl_type, flags, address); 278 279 let new_val = dfg.first_result(inst); 280 builder.def_var(var, new_val); 281 } 282 InstructionFormat::StoreNoOffset => { 283 let val = builder.use_var(var); 284 285 builder 286 .ins() 287 .StoreNoOffset(opcode, ctrl_type, flags, val, address); 288 } 289 InstructionFormat::Store => { 290 let val = builder.use_var(var); 291 292 builder 293 .ins() 294 .Store(opcode, ctrl_type, flags, offset, val, address); 295 } 296 InstructionFormat::Load => { 297 let (inst, dfg) = builder 298 .ins() 299 .Load(opcode, ctrl_type, flags, offset, address); 300 301 let new_val = dfg.first_result(inst); 302 builder.def_var(var, new_val); 303 } 304 _ => unimplemented!(), 305 } 306 307 Ok(()) 308 } 309 310 fn insert_atomic_rmw( 311 fgen: &mut FunctionGenerator, 312 builder: &mut FunctionBuilder, 313 _: Opcode, 314 _: &[Type], 315 rets: &[Type], 316 ) -> Result<()> { 317 let ctrl_type = *rets.first().unwrap(); 318 let type_size = ctrl_type.bytes(); 319 320 let rmw_op = *fgen.u.choose(AtomicRmwOp::all())?; 321 322 let (address, flags, offset) = fgen.generate_address_and_memflags(builder, type_size, true)?; 323 324 // AtomicRMW does not directly support offsets, so add the offset to the address separately. 325 let address = builder.ins().iadd_imm(address, i64::from(offset)); 326 327 // Load and store target variables 328 let source_var = fgen.get_variable_of_type(ctrl_type)?; 329 let target_var = fgen.get_variable_of_type(ctrl_type)?; 330 331 let source_val = builder.use_var(source_var); 332 let new_val = builder 333 .ins() 334 .atomic_rmw(ctrl_type, flags, rmw_op, address, source_val); 335 336 builder.def_var(target_var, new_val); 337 Ok(()) 338 } 339 340 fn insert_atomic_cas( 341 fgen: &mut FunctionGenerator, 342 builder: &mut FunctionBuilder, 343 _: Opcode, 344 _: &[Type], 345 rets: &[Type], 346 ) -> Result<()> { 347 let ctrl_type = *rets.first().unwrap(); 348 let type_size = ctrl_type.bytes(); 349 350 let (address, flags, offset) = fgen.generate_address_and_memflags(builder, type_size, true)?; 351 352 // AtomicCas does not directly support offsets, so add the offset to the address separately. 353 let address = builder.ins().iadd_imm(address, i64::from(offset)); 354 355 // Source and Target variables 356 let expected_var = fgen.get_variable_of_type(ctrl_type)?; 357 let store_var = fgen.get_variable_of_type(ctrl_type)?; 358 let loaded_var = fgen.get_variable_of_type(ctrl_type)?; 359 360 let expected_val = builder.use_var(expected_var); 361 let store_val = builder.use_var(store_var); 362 let new_val = builder 363 .ins() 364 .atomic_cas(flags, address, expected_val, store_val); 365 366 builder.def_var(loaded_var, new_val); 367 Ok(()) 368 } 369 370 fn insert_shuffle( 371 fgen: &mut FunctionGenerator, 372 builder: &mut FunctionBuilder, 373 opcode: Opcode, 374 _: &[Type], 375 rets: &[Type], 376 ) -> Result<()> { 377 let ctrl_type = *rets.first().unwrap(); 378 379 let lhs = builder.use_var(fgen.get_variable_of_type(ctrl_type)?); 380 let rhs = builder.use_var(fgen.get_variable_of_type(ctrl_type)?); 381 382 let mask = { 383 let mut lanes = [0u8; 16]; 384 for lane in lanes.iter_mut() { 385 *lane = fgen.u.int_in_range(0..=31)?; 386 } 387 let lanes = ConstantData::from(lanes.as_ref()); 388 builder.func.dfg.immediates.push(lanes) 389 }; 390 391 // This function is called for any `InstructionFormat::Shuffle`. Which today is just 392 // `shuffle`, but lets assert that, just to be sure we don't accidentally insert 393 // something else. 394 assert_eq!(opcode, Opcode::Shuffle); 395 let res = builder.ins().shuffle(lhs, rhs, mask); 396 397 let target_var = fgen.get_variable_of_type(ctrl_type)?; 398 builder.def_var(target_var, res); 399 400 Ok(()) 401 } 402 403 fn insert_ins_ext_lane( 404 fgen: &mut FunctionGenerator, 405 builder: &mut FunctionBuilder, 406 opcode: Opcode, 407 args: &[Type], 408 rets: &[Type], 409 ) -> Result<()> { 410 let vector_type = *args.first().unwrap(); 411 let ret_type = *rets.first().unwrap(); 412 413 let lhs = builder.use_var(fgen.get_variable_of_type(vector_type)?); 414 let max_lane = (vector_type.lane_count() as u8) - 1; 415 let lane = fgen.u.int_in_range(0..=max_lane)?; 416 417 let res = match opcode { 418 Opcode::Insertlane => { 419 let rhs = builder.use_var(fgen.get_variable_of_type(args[1])?); 420 builder.ins().insertlane(lhs, rhs, lane) 421 } 422 Opcode::Extractlane => builder.ins().extractlane(lhs, lane), 423 _ => todo!(), 424 }; 425 426 let target_var = fgen.get_variable_of_type(ret_type)?; 427 builder.def_var(target_var, res); 428 429 Ok(()) 430 } 431 432 type OpcodeInserter = fn( 433 fgen: &mut FunctionGenerator, 434 builder: &mut FunctionBuilder, 435 Opcode, 436 &[Type], 437 &[Type], 438 ) -> Result<()>; 439 440 macro_rules! exceptions { 441 ($op:expr, $args:expr, $rets:expr, $(($($cases:pat),*)),* $(,)?) => { 442 match ($op, $args, $rets) { 443 $( ($($cases,)* ..) => return false, )* 444 _ => true, 445 } 446 } 447 } 448 449 /// Returns true if we believe this `OpcodeSignature` should compile correctly 450 /// for the given target triple. We currently have a range of known issues 451 /// with specific lowerings on specific backends, and we don't want to get 452 /// fuzz bug reports for those. Over time our goal is to eliminate all of these 453 /// exceptions. 454 fn valid_for_target(triple: &Triple, op: Opcode, args: &[Type], rets: &[Type]) -> bool { 455 // Rule out invalid combinations that we don't yet have a good way of rejecting with the 456 // instruction DSL type constraints. 457 match op { 458 Opcode::FcvtToUintSat | Opcode::FcvtToSintSat => { 459 assert_eq!(args.len(), 1); 460 assert_eq!(rets.len(), 1); 461 462 let arg = args[0]; 463 let ret = rets[0]; 464 465 // Vector arguments must produce vector results, and scalar arguments must produce 466 // scalar results. 467 if arg.is_vector() != ret.is_vector() { 468 return false; 469 } 470 471 if arg.is_vector() && ret.is_vector() { 472 // Vector conversions must have the same number of lanes, and the lanes must be the 473 // same bit-width. 474 if arg.lane_count() != ret.lane_count() { 475 return false; 476 } 477 478 if arg.lane_of().bits() != ret.lane_of().bits() { 479 return false; 480 } 481 } 482 } 483 484 Opcode::Bitcast => { 485 assert_eq!(args.len(), 1); 486 assert_eq!(rets.len(), 1); 487 488 let arg = args[0]; 489 let ret = rets[0]; 490 491 // The opcode generator still allows bitcasts between different sized types, but these 492 // are rejected in the verifier. 493 if arg.bits() != ret.bits() { 494 return false; 495 } 496 } 497 498 _ => {} 499 } 500 501 match triple.architecture { 502 Architecture::X86_64 => { 503 exceptions!( 504 op, 505 args, 506 rets, 507 (Opcode::UmulOverflow | Opcode::SmulOverflow, &[I128, I128]), 508 (Opcode::Imul, &[I8X16, I8X16]), 509 // https://github.com/bytecodealliance/wasmtime/issues/5468 510 (Opcode::Smulhi | Opcode::Umulhi, &[I8, I8]), 511 // https://github.com/bytecodealliance/wasmtime/issues/4756 512 (Opcode::Udiv | Opcode::Sdiv, &[I128, I128]), 513 // https://github.com/bytecodealliance/wasmtime/issues/5474 514 (Opcode::Urem | Opcode::Srem, &[I128, I128]), 515 // https://github.com/bytecodealliance/wasmtime/issues/5466 516 (Opcode::Iabs, &[I128]), 517 // https://github.com/bytecodealliance/wasmtime/issues/3370 518 ( 519 Opcode::Smin | Opcode::Umin | Opcode::Smax | Opcode::Umax, 520 &[I128, I128] 521 ), 522 // https://github.com/bytecodealliance/wasmtime/issues/5107 523 (Opcode::Cls, &[I8], &[I8]), 524 (Opcode::Cls, &[I16], &[I16]), 525 (Opcode::Cls, &[I32], &[I32]), 526 (Opcode::Cls, &[I64], &[I64]), 527 (Opcode::Cls, &[I128], &[I128]), 528 // https://github.com/bytecodealliance/wasmtime/issues/4897 529 // https://github.com/bytecodealliance/wasmtime/issues/4899 530 ( 531 Opcode::FcvtToUint 532 | Opcode::FcvtToUintSat 533 | Opcode::FcvtToSint 534 | Opcode::FcvtToSintSat, 535 &[F32 | F64], 536 &[I8 | I16 | I128] 537 ), 538 (Opcode::FcvtToUint | Opcode::FcvtToSint, &[F32X4], &[I32X4]), 539 ( 540 Opcode::FcvtToUint 541 | Opcode::FcvtToUintSat 542 | Opcode::FcvtToSint 543 | Opcode::FcvtToSintSat, 544 &[F64X2], 545 &[I64X2] 546 ), 547 // https://github.com/bytecodealliance/wasmtime/issues/4900 548 (Opcode::FcvtFromUint, &[I128], &[F32 | F64]), 549 // This has a lowering, but only when preceded by `uwiden_low`. 550 (Opcode::FcvtFromUint, &[I64X2], &[F64X2]), 551 // https://github.com/bytecodealliance/wasmtime/issues/4900 552 (Opcode::FcvtFromSint, &[I128], &[F32 | F64]), 553 (Opcode::FcvtFromSint, &[I64X2], &[F64X2]), 554 ( 555 Opcode::Umulhi | Opcode::Smulhi, 556 &([I8X16, I8X16] | [I16X8, I16X8] | [I32X4, I32X4] | [I64X2, I64X2]) 557 ), 558 ( 559 Opcode::UaddSat | Opcode::SaddSat | Opcode::UsubSat | Opcode::SsubSat, 560 &([I32X4, I32X4] | [I64X2, I64X2]) 561 ), 562 (Opcode::Fcopysign, &([F32X4, F32X4] | [F64X2, F64X2])), 563 (Opcode::Popcnt, &([I8X16] | [I16X8] | [I32X4] | [I64X2])), 564 ( 565 Opcode::Umax | Opcode::Smax | Opcode::Umin | Opcode::Smin, 566 &[I64X2, I64X2] 567 ), 568 // https://github.com/bytecodealliance/wasmtime/issues/6104 569 (Opcode::Bitcast, &[I128], &[_]), 570 (Opcode::Bitcast, &[_], &[I128]), 571 (Opcode::Uunarrow), 572 (Opcode::Snarrow | Opcode::Unarrow, &[I64X2, I64X2]), 573 (Opcode::SqmulRoundSat, &[I32X4, I32X4]), 574 // This Icmp is not implemented: #5529 575 (Opcode::Icmp, &[I64X2, I64X2]), 576 // IaddPairwise is implemented, but only for some types, and with some preceding ops. 577 (Opcode::IaddPairwise), 578 // Nothing wrong with this select. But we have an isle rule that can optimize it 579 // into a `min`/`max` instructions, which we don't have implemented yet. 580 (Opcode::Select, &[_, I128, I128]), 581 // These stack accesses can cause segfaults if they are merged into an SSE instruction. 582 // See: #5922 583 ( 584 Opcode::StackStore, 585 &[I8X16 | I16X8 | I32X4 | I64X2 | F32X4 | F64X2] 586 ), 587 ( 588 Opcode::StackLoad, 589 &[], 590 &[I8X16 | I16X8 | I32X4 | I64X2 | F32X4 | F64X2] 591 ), 592 ) 593 } 594 595 Architecture::Aarch64(_) => { 596 exceptions!( 597 op, 598 args, 599 rets, 600 (Opcode::UmulOverflow | Opcode::SmulOverflow, &[I128, I128]), 601 // https://github.com/bytecodealliance/wasmtime/issues/4864 602 (Opcode::Udiv | Opcode::Sdiv, &[I128, I128]), 603 // https://github.com/bytecodealliance/wasmtime/issues/5472 604 (Opcode::Urem | Opcode::Srem, &[I128, I128]), 605 // https://github.com/bytecodealliance/wasmtime/issues/5467 606 (Opcode::Iabs, &[I128]), 607 // https://github.com/bytecodealliance/wasmtime/issues/4313 608 ( 609 Opcode::Smin | Opcode::Umin | Opcode::Smax | Opcode::Umax, 610 &[I128, I128] 611 ), 612 // https://github.com/bytecodealliance/wasmtime/issues/4870 613 (Opcode::Bnot, &[F32 | F64]), 614 ( 615 Opcode::Band 616 | Opcode::Bor 617 | Opcode::Bxor 618 | Opcode::BandNot 619 | Opcode::BorNot 620 | Opcode::BxorNot, 621 &([F32, F32] | [F64, F64]) 622 ), 623 // https://github.com/bytecodealliance/wasmtime/issues/5198 624 (Opcode::Bitselect, &[I128, I128, I128]), 625 // https://github.com/bytecodealliance/wasmtime/issues/4934 626 ( 627 Opcode::FcvtToUint 628 | Opcode::FcvtToUintSat 629 | Opcode::FcvtToSint 630 | Opcode::FcvtToSintSat, 631 &[F32 | F64] 632 ), 633 // https://github.com/bytecodealliance/wasmtime/issues/4933 634 ( 635 Opcode::FcvtFromUint | Opcode::FcvtFromSint, 636 &[I128], 637 &[F32 | F64] 638 ), 639 ( 640 Opcode::Umulhi | Opcode::Smulhi, 641 &([I8X16, I8X16] | [I16X8, I16X8] | [I32X4, I32X4] | [I64X2, I64X2]) 642 ), 643 (Opcode::Popcnt, &[I16X8 | I32X4 | I64X2]), 644 // Nothing wrong with this select. But we have an isle rule that can optimize it 645 // into a `min`/`max` instructions, which we don't have implemented yet. 646 (Opcode::Select, &[I8, I128, I128]), 647 // https://github.com/bytecodealliance/wasmtime/issues/6104 648 (Opcode::Bitcast, &[I128], &[_]), 649 (Opcode::Bitcast, &[_], &[I128]), 650 ) 651 } 652 653 Architecture::S390x => { 654 exceptions!( 655 op, 656 args, 657 rets, 658 (Opcode::UaddOverflow | Opcode::SaddOverflow), 659 (Opcode::UsubOverflow | Opcode::SsubOverflow), 660 (Opcode::UmulOverflow | Opcode::SmulOverflow), 661 ( 662 Opcode::Udiv | Opcode::Sdiv | Opcode::Urem | Opcode::Srem, 663 &[I128, I128] 664 ), 665 (Opcode::Bnot, &[F32 | F64]), 666 ( 667 Opcode::Band 668 | Opcode::Bor 669 | Opcode::Bxor 670 | Opcode::BandNot 671 | Opcode::BorNot 672 | Opcode::BxorNot, 673 &([F32, F32] | [F64, F64]) 674 ), 675 ( 676 Opcode::FcvtToUint 677 | Opcode::FcvtToUintSat 678 | Opcode::FcvtToSint 679 | Opcode::FcvtToSintSat, 680 &[F32 | F64], 681 &[I128] 682 ), 683 ( 684 Opcode::FcvtFromUint | Opcode::FcvtFromSint, 685 &[I128], 686 &[F32 | F64] 687 ), 688 (Opcode::SsubSat | Opcode::SaddSat, &[I64X2, I64X2]), 689 // https://github.com/bytecodealliance/wasmtime/issues/6104 690 (Opcode::Bitcast, &[I128], &[_]), 691 (Opcode::Bitcast, &[_], &[I128]), 692 ) 693 } 694 695 Architecture::Riscv64(_) => { 696 // RISC-V Does not support SIMD at all 697 let is_simd = args.iter().chain(rets).any(|t| t.is_vector()); 698 if is_simd { 699 return false; 700 } 701 702 exceptions!( 703 op, 704 args, 705 rets, 706 // TODO 707 (Opcode::UaddOverflow | Opcode::SaddOverflow), 708 (Opcode::UsubOverflow | Opcode::SsubOverflow), 709 (Opcode::UmulOverflow | Opcode::SmulOverflow), 710 // TODO 711 ( 712 Opcode::Udiv | Opcode::Sdiv | Opcode::Urem | Opcode::Srem, 713 &[I128, I128] 714 ), 715 // TODO 716 (Opcode::Iabs, &[I128]), 717 // TODO 718 (Opcode::Bitselect, &[I128, I128, I128]), 719 // https://github.com/bytecodealliance/wasmtime/issues/5528 720 ( 721 Opcode::FcvtToUint | Opcode::FcvtToSint, 722 [F32 | F64], 723 &[I128] 724 ), 725 ( 726 Opcode::FcvtToUintSat | Opcode::FcvtToSintSat, 727 &[F32 | F64], 728 &[I8 | I16 | I128] 729 ), 730 // https://github.com/bytecodealliance/wasmtime/issues/5528 731 ( 732 Opcode::FcvtFromUint | Opcode::FcvtFromSint, 733 &[I128], 734 &[F32 | F64] 735 ), 736 // https://github.com/bytecodealliance/wasmtime/issues/6104 737 (Opcode::Bitcast, &[I128], &[_]), 738 (Opcode::Bitcast, &[_], &[I128]), 739 // TODO 740 (Opcode::SelectSpectreGuard, &[_, _, _], &[F32]), 741 (Opcode::SelectSpectreGuard, &[_, _, _], &[F64]), 742 ) 743 } 744 745 _ => true, 746 } 747 } 748 749 type OpcodeSignature = (Opcode, Vec<Type>, Vec<Type>); 750 751 static OPCODE_SIGNATURES: Lazy<Vec<OpcodeSignature>> = Lazy::new(|| { 752 let types = &[ 753 I8, I16, I32, I64, I128, // Scalar Integers 754 F32, F64, // Scalar Floats 755 I8X16, I16X8, I32X4, I64X2, // SIMD Integers 756 F32X4, F64X2, // SIMD Floats 757 ]; 758 759 Opcode::all() 760 .iter() 761 .filter(|op| { 762 match op { 763 // Control flow opcodes should not be generated through `generate_instructions`. 764 Opcode::BrTable 765 | Opcode::Brif 766 | Opcode::Jump 767 | Opcode::Return 768 | Opcode::ReturnCall 769 | Opcode::ReturnCallIndirect => false, 770 771 // Constants are generated outside of `generate_instructions` 772 Opcode::Iconst => false, 773 774 // TODO: extract_vector raises exceptions during return type generation becuase it 775 // uses dynamic vectors. 776 Opcode::ExtractVector => false, 777 778 _ => true, 779 } 780 }) 781 .flat_map(|op| { 782 let constraints = op.constraints(); 783 784 let ctrl_types = if let Some(ctrls) = constraints.ctrl_typeset() { 785 Vec::from_iter(types.iter().copied().filter(|ty| ctrls.contains(*ty))) 786 } else { 787 vec![INVALID] 788 }; 789 790 ctrl_types.into_iter().flat_map(move |ctrl_type| { 791 let rets = Vec::from_iter( 792 (0..constraints.num_fixed_results()) 793 .map(|i| constraints.result_type(i, ctrl_type)), 794 ); 795 796 // Cols is a vector whose length will match `num_fixed_value_arguments`, and whose 797 // elements will be vectors of types that are valid for that fixed argument 798 // position. 799 let mut cols = vec![]; 800 801 for i in 0..constraints.num_fixed_value_arguments() { 802 match constraints.value_argument_constraint(i, ctrl_type) { 803 ResolvedConstraint::Bound(ty) => cols.push(Vec::from([ty])), 804 ResolvedConstraint::Free(tys) => cols.push(Vec::from_iter( 805 types.iter().copied().filter(|ty| tys.contains(*ty)), 806 )), 807 } 808 } 809 810 // Generate the cartesian product of cols to produce a vector of argument lists, 811 // argss. The argss vector is seeded with the empty argument list, so there's an 812 // initial value to be extended in the loop below. 813 let mut argss = vec![vec![]]; 814 let mut cols = cols.as_slice(); 815 while let Some((col, rest)) = cols.split_last() { 816 cols = rest; 817 818 let mut next = vec![]; 819 for current in argss.iter() { 820 // Extend the front of each argument candidate with every type in `col`. 821 for ty in col { 822 let mut args = vec![*ty]; 823 args.extend_from_slice(¤t); 824 next.push(args); 825 } 826 } 827 828 let _ = std::mem::replace(&mut argss, next); 829 } 830 831 argss.into_iter().map(move |args| (*op, args, rets.clone())) 832 }) 833 }) 834 .filter(|(op, args, rets)| { 835 // These op/signature combinations need to be vetted 836 exceptions!( 837 op, 838 args.as_slice(), 839 rets.as_slice(), 840 (Opcode::Debugtrap), 841 (Opcode::Trap), 842 (Opcode::Trapz), 843 (Opcode::ResumableTrap), 844 (Opcode::Trapnz), 845 (Opcode::ResumableTrapnz), 846 (Opcode::CallIndirect, &[I32]), 847 (Opcode::FuncAddr), 848 (Opcode::X86Pshufb), 849 (Opcode::AvgRound), 850 (Opcode::Uload8x8), 851 (Opcode::Sload8x8), 852 (Opcode::Uload16x4), 853 (Opcode::Sload16x4), 854 (Opcode::Uload32x2), 855 (Opcode::Sload32x2), 856 (Opcode::StackAddr), 857 (Opcode::DynamicStackLoad), 858 (Opcode::DynamicStackStore), 859 (Opcode::DynamicStackAddr), 860 (Opcode::GlobalValue), 861 (Opcode::SymbolValue), 862 (Opcode::TlsValue), 863 (Opcode::GetPinnedReg), 864 (Opcode::SetPinnedReg), 865 (Opcode::GetFramePointer), 866 (Opcode::GetStackPointer), 867 (Opcode::GetReturnAddress), 868 (Opcode::TableAddr), 869 (Opcode::Null), 870 (Opcode::X86Blendv), 871 (Opcode::VallTrue), 872 (Opcode::IcmpImm), 873 (Opcode::X86Pmulhrsw), 874 (Opcode::IaddImm), 875 (Opcode::ImulImm), 876 (Opcode::UdivImm), 877 (Opcode::SdivImm), 878 (Opcode::UremImm), 879 (Opcode::SremImm), 880 (Opcode::IrsubImm), 881 (Opcode::IaddCin), 882 (Opcode::IaddCarry), 883 (Opcode::UaddOverflowTrap), 884 (Opcode::IsubBin), 885 (Opcode::IsubBorrow), 886 (Opcode::BandImm), 887 (Opcode::BorImm), 888 (Opcode::BxorImm), 889 (Opcode::RotlImm), 890 (Opcode::RotrImm), 891 (Opcode::IshlImm), 892 (Opcode::UshrImm), 893 (Opcode::SshrImm), 894 (Opcode::IsNull), 895 (Opcode::IsInvalid), 896 (Opcode::ScalarToVector), 897 (Opcode::X86Pmaddubsw), 898 (Opcode::X86Cvtt2dq), 899 (Opcode::Bitselect, &[F32, F32, F32], &[F32]), 900 (Opcode::Bitselect, &[F64, F64, F64], &[F64]), 901 (Opcode::Bitselect, &[F32X4, F32X4, F32X4], &[F32X4]), 902 (Opcode::Bitselect, &[F64X2, F64X2, F64X2], &[F64X2]), 903 (Opcode::VanyTrue, &[F32X4], &[I8]), 904 (Opcode::VanyTrue, &[F64X2], &[I8]), 905 (Opcode::VhighBits, &[F32X4], &[I8]), 906 (Opcode::VhighBits, &[F64X2], &[I8]), 907 (Opcode::VhighBits, &[I8X16], &[I16]), 908 (Opcode::VhighBits, &[I16X8], &[I16]), 909 (Opcode::VhighBits, &[I32X4], &[I16]), 910 (Opcode::VhighBits, &[I64X2], &[I16]), 911 (Opcode::VhighBits, &[F32X4], &[I16]), 912 (Opcode::VhighBits, &[F64X2], &[I16]), 913 (Opcode::VhighBits, &[I8X16], &[I32]), 914 (Opcode::VhighBits, &[I16X8], &[I32]), 915 (Opcode::VhighBits, &[I32X4], &[I32]), 916 (Opcode::VhighBits, &[I64X2], &[I32]), 917 (Opcode::VhighBits, &[F32X4], &[I32]), 918 (Opcode::VhighBits, &[F64X2], &[I32]), 919 (Opcode::VhighBits, &[I8X16], &[I64]), 920 (Opcode::VhighBits, &[I16X8], &[I64]), 921 (Opcode::VhighBits, &[I32X4], &[I64]), 922 (Opcode::VhighBits, &[I64X2], &[I64]), 923 (Opcode::VhighBits, &[F32X4], &[I64]), 924 (Opcode::VhighBits, &[F64X2], &[I64]), 925 (Opcode::VhighBits, &[I8X16], &[I128]), 926 (Opcode::VhighBits, &[I16X8], &[I128]), 927 (Opcode::VhighBits, &[I32X4], &[I128]), 928 (Opcode::VhighBits, &[I64X2], &[I128]), 929 (Opcode::VhighBits, &[F32X4], &[I128]), 930 (Opcode::VhighBits, &[F64X2], &[I128]), 931 (Opcode::VhighBits, &[I8X16], &[I8X16]), 932 (Opcode::VhighBits, &[I16X8], &[I8X16]), 933 (Opcode::VhighBits, &[I32X4], &[I8X16]), 934 (Opcode::VhighBits, &[I64X2], &[I8X16]), 935 (Opcode::VhighBits, &[F32X4], &[I8X16]), 936 (Opcode::VhighBits, &[F64X2], &[I8X16]), 937 (Opcode::VhighBits, &[I8X16], &[I16X8]), 938 (Opcode::VhighBits, &[I16X8], &[I16X8]), 939 (Opcode::VhighBits, &[I32X4], &[I16X8]), 940 (Opcode::VhighBits, &[I64X2], &[I16X8]), 941 (Opcode::VhighBits, &[F32X4], &[I16X8]), 942 (Opcode::VhighBits, &[F64X2], &[I16X8]), 943 (Opcode::VhighBits, &[I8X16], &[I32X4]), 944 (Opcode::VhighBits, &[I16X8], &[I32X4]), 945 (Opcode::VhighBits, &[I32X4], &[I32X4]), 946 (Opcode::VhighBits, &[I64X2], &[I32X4]), 947 (Opcode::VhighBits, &[F32X4], &[I32X4]), 948 (Opcode::VhighBits, &[F64X2], &[I32X4]), 949 (Opcode::VhighBits, &[I8X16], &[I64X2]), 950 (Opcode::VhighBits, &[I16X8], &[I64X2]), 951 (Opcode::VhighBits, &[I32X4], &[I64X2]), 952 (Opcode::VhighBits, &[I64X2], &[I64X2]), 953 (Opcode::VhighBits, &[F32X4], &[I64X2]), 954 (Opcode::VhighBits, &[F64X2], &[I64X2]), 955 (Opcode::Ineg, &[I8X16], &[I8X16]), 956 (Opcode::Ineg, &[I16X8], &[I16X8]), 957 (Opcode::Ineg, &[I32X4], &[I32X4]), 958 (Opcode::Ineg, &[I64X2], &[I64X2]), 959 (Opcode::Umulhi, &[I128, I128], &[I128]), 960 (Opcode::Smulhi, &[I128, I128], &[I128]), 961 // https://github.com/bytecodealliance/wasmtime/issues/6073 962 (Opcode::Iconcat, &[I32, I32], &[I64]), 963 (Opcode::Iconcat, &[I16, I16], &[I32]), 964 (Opcode::Iconcat, &[I8, I8], &[I16]), 965 // https://github.com/bytecodealliance/wasmtime/issues/6073 966 (Opcode::Isplit, &[I64], &[I32, I32]), 967 (Opcode::Isplit, &[I32], &[I16, I16]), 968 (Opcode::Isplit, &[I16], &[I8, I8]), 969 (Opcode::Rotl, &[I8X16, I8], &[I8X16]), 970 (Opcode::Rotl, &[I8X16, I16], &[I8X16]), 971 (Opcode::Rotl, &[I8X16, I32], &[I8X16]), 972 (Opcode::Rotl, &[I8X16, I64], &[I8X16]), 973 (Opcode::Rotl, &[I8X16, I128], &[I8X16]), 974 (Opcode::Rotl, &[I16X8, I8], &[I16X8]), 975 (Opcode::Rotl, &[I16X8, I16], &[I16X8]), 976 (Opcode::Rotl, &[I16X8, I32], &[I16X8]), 977 (Opcode::Rotl, &[I16X8, I64], &[I16X8]), 978 (Opcode::Rotl, &[I16X8, I128], &[I16X8]), 979 (Opcode::Rotl, &[I32X4, I8], &[I32X4]), 980 (Opcode::Rotl, &[I32X4, I16], &[I32X4]), 981 (Opcode::Rotl, &[I32X4, I32], &[I32X4]), 982 (Opcode::Rotl, &[I32X4, I64], &[I32X4]), 983 (Opcode::Rotl, &[I32X4, I128], &[I32X4]), 984 (Opcode::Rotl, &[I64X2, I8], &[I64X2]), 985 (Opcode::Rotl, &[I64X2, I16], &[I64X2]), 986 (Opcode::Rotl, &[I64X2, I32], &[I64X2]), 987 (Opcode::Rotl, &[I64X2, I64], &[I64X2]), 988 (Opcode::Rotl, &[I64X2, I128], &[I64X2]), 989 (Opcode::Rotr, &[I8X16, I8], &[I8X16]), 990 (Opcode::Rotr, &[I8X16, I16], &[I8X16]), 991 (Opcode::Rotr, &[I8X16, I32], &[I8X16]), 992 (Opcode::Rotr, &[I8X16, I64], &[I8X16]), 993 (Opcode::Rotr, &[I8X16, I128], &[I8X16]), 994 (Opcode::Rotr, &[I16X8, I8], &[I16X8]), 995 (Opcode::Rotr, &[I16X8, I16], &[I16X8]), 996 (Opcode::Rotr, &[I16X8, I32], &[I16X8]), 997 (Opcode::Rotr, &[I16X8, I64], &[I16X8]), 998 (Opcode::Rotr, &[I16X8, I128], &[I16X8]), 999 (Opcode::Rotr, &[I32X4, I8], &[I32X4]), 1000 (Opcode::Rotr, &[I32X4, I16], &[I32X4]), 1001 (Opcode::Rotr, &[I32X4, I32], &[I32X4]), 1002 (Opcode::Rotr, &[I32X4, I64], &[I32X4]), 1003 (Opcode::Rotr, &[I32X4, I128], &[I32X4]), 1004 (Opcode::Rotr, &[I64X2, I8], &[I64X2]), 1005 (Opcode::Rotr, &[I64X2, I16], &[I64X2]), 1006 (Opcode::Rotr, &[I64X2, I32], &[I64X2]), 1007 (Opcode::Rotr, &[I64X2, I64], &[I64X2]), 1008 (Opcode::Rotr, &[I64X2, I128], &[I64X2]), 1009 (Opcode::Ishl, &[I8X16, I8], &[I8X16]), 1010 (Opcode::Ishl, &[I8X16, I16], &[I8X16]), 1011 (Opcode::Ishl, &[I8X16, I32], &[I8X16]), 1012 (Opcode::Ishl, &[I8X16, I64], &[I8X16]), 1013 (Opcode::Ishl, &[I8X16, I128], &[I8X16]), 1014 (Opcode::Ishl, &[I16X8, I8], &[I16X8]), 1015 (Opcode::Ishl, &[I16X8, I16], &[I16X8]), 1016 (Opcode::Ishl, &[I16X8, I32], &[I16X8]), 1017 (Opcode::Ishl, &[I16X8, I64], &[I16X8]), 1018 (Opcode::Ishl, &[I16X8, I128], &[I16X8]), 1019 (Opcode::Ishl, &[I32X4, I8], &[I32X4]), 1020 (Opcode::Ishl, &[I32X4, I16], &[I32X4]), 1021 (Opcode::Ishl, &[I32X4, I32], &[I32X4]), 1022 (Opcode::Ishl, &[I32X4, I64], &[I32X4]), 1023 (Opcode::Ishl, &[I32X4, I128], &[I32X4]), 1024 (Opcode::Ishl, &[I64X2, I8], &[I64X2]), 1025 (Opcode::Ishl, &[I64X2, I16], &[I64X2]), 1026 (Opcode::Ishl, &[I64X2, I32], &[I64X2]), 1027 (Opcode::Ishl, &[I64X2, I64], &[I64X2]), 1028 (Opcode::Ishl, &[I64X2, I128], &[I64X2]), 1029 (Opcode::Ushr, &[I8X16, I8], &[I8X16]), 1030 (Opcode::Ushr, &[I8X16, I16], &[I8X16]), 1031 (Opcode::Ushr, &[I8X16, I32], &[I8X16]), 1032 (Opcode::Ushr, &[I8X16, I64], &[I8X16]), 1033 (Opcode::Ushr, &[I8X16, I128], &[I8X16]), 1034 (Opcode::Ushr, &[I16X8, I8], &[I16X8]), 1035 (Opcode::Ushr, &[I16X8, I16], &[I16X8]), 1036 (Opcode::Ushr, &[I16X8, I32], &[I16X8]), 1037 (Opcode::Ushr, &[I16X8, I64], &[I16X8]), 1038 (Opcode::Ushr, &[I16X8, I128], &[I16X8]), 1039 (Opcode::Ushr, &[I32X4, I8], &[I32X4]), 1040 (Opcode::Ushr, &[I32X4, I16], &[I32X4]), 1041 (Opcode::Ushr, &[I32X4, I32], &[I32X4]), 1042 (Opcode::Ushr, &[I32X4, I64], &[I32X4]), 1043 (Opcode::Ushr, &[I32X4, I128], &[I32X4]), 1044 (Opcode::Ushr, &[I64X2, I8], &[I64X2]), 1045 (Opcode::Ushr, &[I64X2, I16], &[I64X2]), 1046 (Opcode::Ushr, &[I64X2, I32], &[I64X2]), 1047 (Opcode::Ushr, &[I64X2, I64], &[I64X2]), 1048 (Opcode::Ushr, &[I64X2, I128], &[I64X2]), 1049 (Opcode::Sshr, &[I8X16, I8], &[I8X16]), 1050 (Opcode::Sshr, &[I8X16, I16], &[I8X16]), 1051 (Opcode::Sshr, &[I8X16, I32], &[I8X16]), 1052 (Opcode::Sshr, &[I8X16, I64], &[I8X16]), 1053 (Opcode::Sshr, &[I8X16, I128], &[I8X16]), 1054 (Opcode::Sshr, &[I16X8, I8], &[I16X8]), 1055 (Opcode::Sshr, &[I16X8, I16], &[I16X8]), 1056 (Opcode::Sshr, &[I16X8, I32], &[I16X8]), 1057 (Opcode::Sshr, &[I16X8, I64], &[I16X8]), 1058 (Opcode::Sshr, &[I16X8, I128], &[I16X8]), 1059 (Opcode::Sshr, &[I32X4, I8], &[I32X4]), 1060 (Opcode::Sshr, &[I32X4, I16], &[I32X4]), 1061 (Opcode::Sshr, &[I32X4, I32], &[I32X4]), 1062 (Opcode::Sshr, &[I32X4, I64], &[I32X4]), 1063 (Opcode::Sshr, &[I32X4, I128], &[I32X4]), 1064 (Opcode::Sshr, &[I64X2, I8], &[I64X2]), 1065 (Opcode::Sshr, &[I64X2, I16], &[I64X2]), 1066 (Opcode::Sshr, &[I64X2, I32], &[I64X2]), 1067 (Opcode::Sshr, &[I64X2, I64], &[I64X2]), 1068 (Opcode::Sshr, &[I64X2, I128], &[I64X2]), 1069 (Opcode::Fmin, &[F32X4, F32X4], &[F32X4]), 1070 (Opcode::Fmin, &[F64X2, F64X2], &[F64X2]), 1071 (Opcode::Fmax, &[F32X4, F32X4], &[F32X4]), 1072 (Opcode::Fmax, &[F64X2, F64X2], &[F64X2]), 1073 (Opcode::FcvtToUintSat, &[F32X4], &[I8]), 1074 (Opcode::FcvtToUintSat, &[F64X2], &[I8]), 1075 (Opcode::FcvtToUintSat, &[F32X4], &[I16]), 1076 (Opcode::FcvtToUintSat, &[F64X2], &[I16]), 1077 (Opcode::FcvtToUintSat, &[F32X4], &[I32]), 1078 (Opcode::FcvtToUintSat, &[F64X2], &[I32]), 1079 (Opcode::FcvtToUintSat, &[F32X4], &[I64]), 1080 (Opcode::FcvtToUintSat, &[F64X2], &[I64]), 1081 (Opcode::FcvtToUintSat, &[F32X4], &[I128]), 1082 (Opcode::FcvtToUintSat, &[F64X2], &[I128]), 1083 (Opcode::FcvtToUintSat, &[F32], &[I8X16]), 1084 (Opcode::FcvtToUintSat, &[F64], &[I8X16]), 1085 (Opcode::FcvtToUintSat, &[F32X4], &[I8X16]), 1086 (Opcode::FcvtToUintSat, &[F64X2], &[I8X16]), 1087 (Opcode::FcvtToUintSat, &[F32], &[I16X8]), 1088 (Opcode::FcvtToUintSat, &[F64], &[I16X8]), 1089 (Opcode::FcvtToUintSat, &[F32X4], &[I16X8]), 1090 (Opcode::FcvtToUintSat, &[F64X2], &[I16X8]), 1091 (Opcode::FcvtToUintSat, &[F32], &[I32X4]), 1092 (Opcode::FcvtToUintSat, &[F64], &[I32X4]), 1093 (Opcode::FcvtToUintSat, &[F64X2], &[I32X4]), 1094 (Opcode::FcvtToUintSat, &[F32], &[I64X2]), 1095 (Opcode::FcvtToUintSat, &[F64], &[I64X2]), 1096 (Opcode::FcvtToUintSat, &[F32X4], &[I64X2]), 1097 (Opcode::FcvtToSintSat, &[F32X4], &[I8]), 1098 (Opcode::FcvtToSintSat, &[F64X2], &[I8]), 1099 (Opcode::FcvtToSintSat, &[F32X4], &[I16]), 1100 (Opcode::FcvtToSintSat, &[F64X2], &[I16]), 1101 (Opcode::FcvtToSintSat, &[F32X4], &[I32]), 1102 (Opcode::FcvtToSintSat, &[F64X2], &[I32]), 1103 (Opcode::FcvtToSintSat, &[F32X4], &[I64]), 1104 (Opcode::FcvtToSintSat, &[F64X2], &[I64]), 1105 (Opcode::FcvtToSintSat, &[F32X4], &[I128]), 1106 (Opcode::FcvtToSintSat, &[F64X2], &[I128]), 1107 (Opcode::FcvtToSintSat, &[F32], &[I8X16]), 1108 (Opcode::FcvtToSintSat, &[F64], &[I8X16]), 1109 (Opcode::FcvtToSintSat, &[F32X4], &[I8X16]), 1110 (Opcode::FcvtToSintSat, &[F64X2], &[I8X16]), 1111 (Opcode::FcvtToSintSat, &[F32], &[I16X8]), 1112 (Opcode::FcvtToSintSat, &[F64], &[I16X8]), 1113 (Opcode::FcvtToSintSat, &[F32X4], &[I16X8]), 1114 (Opcode::FcvtToSintSat, &[F64X2], &[I16X8]), 1115 (Opcode::FcvtToSintSat, &[F32], &[I32X4]), 1116 (Opcode::FcvtToSintSat, &[F64], &[I32X4]), 1117 (Opcode::FcvtToSintSat, &[F64X2], &[I32X4]), 1118 (Opcode::FcvtToSintSat, &[F32], &[I64X2]), 1119 (Opcode::FcvtToSintSat, &[F64], &[I64X2]), 1120 (Opcode::FcvtToSintSat, &[F32X4], &[I64X2]), 1121 (Opcode::FcvtFromUint, &[I8X16], &[F32]), 1122 (Opcode::FcvtFromUint, &[I16X8], &[F32]), 1123 (Opcode::FcvtFromUint, &[I32X4], &[F32]), 1124 (Opcode::FcvtFromUint, &[I64X2], &[F32]), 1125 (Opcode::FcvtFromUint, &[I8X16], &[F64]), 1126 (Opcode::FcvtFromUint, &[I16X8], &[F64]), 1127 (Opcode::FcvtFromUint, &[I32X4], &[F64]), 1128 (Opcode::FcvtFromUint, &[I64X2], &[F64]), 1129 (Opcode::FcvtFromUint, &[I8], &[F32X4]), 1130 (Opcode::FcvtFromUint, &[I16], &[F32X4]), 1131 (Opcode::FcvtFromUint, &[I32], &[F32X4]), 1132 (Opcode::FcvtFromUint, &[I64], &[F32X4]), 1133 (Opcode::FcvtFromUint, &[I128], &[F32X4]), 1134 (Opcode::FcvtFromUint, &[I8X16], &[F32X4]), 1135 (Opcode::FcvtFromUint, &[I16X8], &[F32X4]), 1136 (Opcode::FcvtFromUint, &[I64X2], &[F32X4]), 1137 (Opcode::FcvtFromUint, &[I8], &[F64X2]), 1138 (Opcode::FcvtFromUint, &[I16], &[F64X2]), 1139 (Opcode::FcvtFromUint, &[I32], &[F64X2]), 1140 (Opcode::FcvtFromUint, &[I64], &[F64X2]), 1141 (Opcode::FcvtFromUint, &[I128], &[F64X2]), 1142 (Opcode::FcvtFromUint, &[I8X16], &[F64X2]), 1143 (Opcode::FcvtFromUint, &[I16X8], &[F64X2]), 1144 (Opcode::FcvtFromUint, &[I32X4], &[F64X2]), 1145 (Opcode::FcvtFromSint, &[I8X16], &[F32]), 1146 (Opcode::FcvtFromSint, &[I16X8], &[F32]), 1147 (Opcode::FcvtFromSint, &[I32X4], &[F32]), 1148 (Opcode::FcvtFromSint, &[I64X2], &[F32]), 1149 (Opcode::FcvtFromSint, &[I8X16], &[F64]), 1150 (Opcode::FcvtFromSint, &[I16X8], &[F64]), 1151 (Opcode::FcvtFromSint, &[I32X4], &[F64]), 1152 (Opcode::FcvtFromSint, &[I64X2], &[F64]), 1153 (Opcode::FcvtFromSint, &[I8], &[F32X4]), 1154 (Opcode::FcvtFromSint, &[I16], &[F32X4]), 1155 (Opcode::FcvtFromSint, &[I32], &[F32X4]), 1156 (Opcode::FcvtFromSint, &[I64], &[F32X4]), 1157 (Opcode::FcvtFromSint, &[I128], &[F32X4]), 1158 (Opcode::FcvtFromSint, &[I8X16], &[F32X4]), 1159 (Opcode::FcvtFromSint, &[I16X8], &[F32X4]), 1160 (Opcode::FcvtFromSint, &[I64X2], &[F32X4]), 1161 (Opcode::FcvtFromSint, &[I8], &[F64X2]), 1162 (Opcode::FcvtFromSint, &[I16], &[F64X2]), 1163 (Opcode::FcvtFromSint, &[I32], &[F64X2]), 1164 (Opcode::FcvtFromSint, &[I64], &[F64X2]), 1165 (Opcode::FcvtFromSint, &[I128], &[F64X2]), 1166 (Opcode::FcvtFromSint, &[I8X16], &[F64X2]), 1167 (Opcode::FcvtFromSint, &[I16X8], &[F64X2]), 1168 (Opcode::FcvtFromSint, &[I32X4], &[F64X2]), 1169 ) 1170 }) 1171 .collect() 1172 }); 1173 1174 fn inserter_for_format(fmt: InstructionFormat) -> OpcodeInserter { 1175 match fmt { 1176 InstructionFormat::AtomicCas => insert_atomic_cas, 1177 InstructionFormat::AtomicRmw => insert_atomic_rmw, 1178 InstructionFormat::Binary => insert_opcode, 1179 InstructionFormat::BinaryImm64 => todo!(), 1180 InstructionFormat::BinaryImm8 => insert_ins_ext_lane, 1181 InstructionFormat::Call => insert_call, 1182 InstructionFormat::CallIndirect => insert_call, 1183 InstructionFormat::CondTrap => todo!(), 1184 InstructionFormat::DynamicStackLoad => todo!(), 1185 InstructionFormat::DynamicStackStore => todo!(), 1186 InstructionFormat::FloatCompare => insert_cmp, 1187 InstructionFormat::FuncAddr => todo!(), 1188 InstructionFormat::IntAddTrap => todo!(), 1189 InstructionFormat::IntCompare => insert_cmp, 1190 InstructionFormat::IntCompareImm => todo!(), 1191 InstructionFormat::Load => insert_load_store, 1192 InstructionFormat::LoadNoOffset => insert_load_store, 1193 InstructionFormat::NullAry => insert_opcode, 1194 InstructionFormat::Shuffle => insert_shuffle, 1195 InstructionFormat::StackLoad => insert_stack_load, 1196 InstructionFormat::StackStore => insert_stack_store, 1197 InstructionFormat::Store => insert_load_store, 1198 InstructionFormat::StoreNoOffset => insert_load_store, 1199 InstructionFormat::TableAddr => todo!(), 1200 InstructionFormat::Ternary => insert_opcode, 1201 InstructionFormat::TernaryImm8 => insert_ins_ext_lane, 1202 InstructionFormat::Trap => todo!(), 1203 InstructionFormat::Unary => insert_opcode, 1204 InstructionFormat::UnaryConst => insert_const, 1205 InstructionFormat::UnaryGlobalValue => todo!(), 1206 InstructionFormat::UnaryIeee32 => insert_const, 1207 InstructionFormat::UnaryIeee64 => insert_const, 1208 InstructionFormat::UnaryImm => insert_const, 1209 1210 InstructionFormat::BranchTable 1211 | InstructionFormat::Brif 1212 | InstructionFormat::Jump 1213 | InstructionFormat::MultiAry => { 1214 panic!( 1215 "Control-flow instructions should be handled by 'insert_terminator': {:?}", 1216 fmt 1217 ) 1218 } 1219 } 1220 } 1221 1222 pub struct FunctionGenerator<'r, 'data> 1223 where 1224 'data: 'r, 1225 { 1226 u: &'r mut Unstructured<'data>, 1227 config: &'r Config, 1228 resources: Resources, 1229 isa: OwnedTargetIsa, 1230 name: UserFuncName, 1231 signature: Signature, 1232 } 1233 1234 #[derive(Debug, Clone)] 1235 enum BlockTerminator { 1236 Return, 1237 Jump(Block), 1238 Br(Block, Block), 1239 BrTable(Block, Vec<Block>), 1240 Switch(Type, Block, HashMap<u128, Block>), 1241 TailCall(FuncRef), 1242 TailCallIndirect(FuncRef), 1243 } 1244 1245 #[derive(Debug, Clone)] 1246 enum BlockTerminatorKind { 1247 Return, 1248 Jump, 1249 Br, 1250 BrTable, 1251 Switch, 1252 TailCall, 1253 TailCallIndirect, 1254 } 1255 1256 #[derive(Default)] 1257 struct Resources { 1258 vars: HashMap<Type, Vec<Variable>>, 1259 blocks: Vec<(Block, BlockSignature)>, 1260 blocks_without_params: Vec<Block>, 1261 block_terminators: Vec<BlockTerminator>, 1262 func_refs: Vec<(Signature, SigRef, FuncRef)>, 1263 stack_slots: Vec<(StackSlot, StackSize)>, 1264 usercalls: Vec<(UserExternalName, Signature)>, 1265 libcalls: Vec<LibCall>, 1266 } 1267 1268 impl Resources { 1269 /// Partitions blocks at `block`. Only blocks that can be targeted by branches are considered. 1270 /// 1271 /// The first slice includes all blocks up to and including `block`. 1272 /// The second slice includes all remaining blocks. 1273 fn partition_target_blocks( 1274 &self, 1275 block: Block, 1276 ) -> (&[(Block, BlockSignature)], &[(Block, BlockSignature)]) { 1277 // Blocks are stored in-order and have no gaps, this means that we can simply index them by 1278 // their number. We also need to exclude the entry block since it isn't a valid target. 1279 let target_blocks = &self.blocks[1..]; 1280 target_blocks.split_at(block.as_u32() as usize) 1281 } 1282 1283 /// Returns blocks forward of `block`. Only blocks that can be targeted by branches are considered. 1284 fn forward_blocks(&self, block: Block) -> &[(Block, BlockSignature)] { 1285 let (_, forward_blocks) = self.partition_target_blocks(block); 1286 forward_blocks 1287 } 1288 1289 /// Generates a slice of `blocks_without_params` ahead of `block` 1290 fn forward_blocks_without_params(&self, block: Block) -> &[Block] { 1291 let partition_point = self.blocks_without_params.partition_point(|b| *b <= block); 1292 &self.blocks_without_params[partition_point..] 1293 } 1294 1295 /// Generates an iterator of all valid tail call targets. This includes all functions with both 1296 /// the `tail` calling convention and the same return values as the caller. 1297 fn tail_call_targets<'a>( 1298 &'a self, 1299 caller_sig: &'a Signature, 1300 ) -> impl Iterator<Item = &'a (Signature, SigRef, FuncRef)> { 1301 self.func_refs.iter().filter(|(sig, _, _)| { 1302 sig.call_conv == CallConv::Tail && sig.returns == caller_sig.returns 1303 }) 1304 } 1305 } 1306 1307 impl<'r, 'data> FunctionGenerator<'r, 'data> 1308 where 1309 'data: 'r, 1310 { 1311 pub fn new( 1312 u: &'r mut Unstructured<'data>, 1313 config: &'r Config, 1314 isa: OwnedTargetIsa, 1315 name: UserFuncName, 1316 signature: Signature, 1317 usercalls: Vec<(UserExternalName, Signature)>, 1318 libcalls: Vec<LibCall>, 1319 ) -> Self { 1320 Self { 1321 u, 1322 config, 1323 resources: Resources { 1324 usercalls, 1325 libcalls, 1326 ..Resources::default() 1327 }, 1328 isa, 1329 name, 1330 signature, 1331 } 1332 } 1333 1334 /// Generates a random value for config `param` 1335 fn param(&mut self, param: &RangeInclusive<usize>) -> Result<usize> { 1336 Ok(self.u.int_in_range(param.clone())?) 1337 } 1338 1339 fn system_callconv(&mut self) -> CallConv { 1340 // TODO: This currently only runs on linux, so this is the only choice 1341 // We should improve this once we generate flags and targets 1342 CallConv::SystemV 1343 } 1344 1345 /// Finds a stack slot with size of at least n bytes 1346 fn stack_slot_with_size(&mut self, n: u32) -> Result<(StackSlot, StackSize)> { 1347 let first = self 1348 .resources 1349 .stack_slots 1350 .partition_point(|&(_slot, size)| size < n); 1351 Ok(*self.u.choose(&self.resources.stack_slots[first..])?) 1352 } 1353 1354 /// Generates an address that should allow for a store or a load. 1355 /// 1356 /// Addresses aren't generated like other values. They are never stored in variables so that 1357 /// we don't run the risk of returning them from a function, which would make the fuzzer 1358 /// complain since they are different from the interpreter to the backend. 1359 /// 1360 /// `min_size`: Controls the amount of space that the address should have. 1361 /// 1362 /// `aligned`: When passed as true, the resulting address is guaranteed to be aligned 1363 /// on an 8 byte boundary. 1364 /// 1365 /// Returns a valid address and the maximum possible offset that still respects `min_size`. 1366 fn generate_load_store_address( 1367 &mut self, 1368 builder: &mut FunctionBuilder, 1369 min_size: u32, 1370 aligned: bool, 1371 ) -> Result<(Value, u32)> { 1372 // TODO: Currently our only source of addresses is stack_addr, but we 1373 // should add global_value, symbol_value eventually 1374 let (addr, available_size) = { 1375 let (ss, slot_size) = self.stack_slot_with_size(min_size)?; 1376 1377 // stack_slot_with_size guarantees that slot_size >= min_size 1378 let max_offset = slot_size - min_size; 1379 let offset = if aligned { 1380 self.u.int_in_range(0..=max_offset / min_size)? * min_size 1381 } else { 1382 self.u.int_in_range(0..=max_offset)? 1383 }; 1384 1385 let base_addr = builder.ins().stack_addr(I64, ss, offset as i32); 1386 let available_size = slot_size.saturating_sub(offset); 1387 (base_addr, available_size) 1388 }; 1389 1390 // TODO: Insert a bunch of amode opcodes here to modify the address! 1391 1392 // Now that we have an address and a size, we just choose a random offset to return to the 1393 // caller. Preserving min_size bytes. 1394 let max_offset = available_size.saturating_sub(min_size); 1395 Ok((addr, max_offset)) 1396 } 1397 1398 // Generates an address and memflags for a load or store. 1399 fn generate_address_and_memflags( 1400 &mut self, 1401 builder: &mut FunctionBuilder, 1402 min_size: u32, 1403 is_atomic: bool, 1404 ) -> Result<(Value, MemFlags, Offset32)> { 1405 // Should we generate an aligned address 1406 // Some backends have issues with unaligned atomics. 1407 // AArch64: https://github.com/bytecodealliance/wasmtime/issues/5483 1408 // RISCV: https://github.com/bytecodealliance/wasmtime/issues/5882 1409 let requires_aligned_atomics = matches!( 1410 self.isa.triple().architecture, 1411 Architecture::Aarch64(_) | Architecture::Riscv64(_) 1412 ); 1413 let aligned = if is_atomic && requires_aligned_atomics { 1414 true 1415 } else if min_size > 8 { 1416 // TODO: We currently can't guarantee that a stack_slot will be aligned on a 16 byte 1417 // boundary. We don't have a way to specify alignment when creating stack slots, and 1418 // cranelift only guarantees 8 byte alignment between stack slots. 1419 // See: https://github.com/bytecodealliance/wasmtime/issues/5922#issuecomment-1457926624 1420 false 1421 } else { 1422 bool::arbitrary(self.u)? 1423 }; 1424 1425 let mut flags = MemFlags::new(); 1426 // Even if we picked an aligned address, we can always generate unaligned memflags 1427 if aligned && bool::arbitrary(self.u)? { 1428 flags.set_aligned(); 1429 } 1430 // If the address is aligned, then we know it won't trap 1431 if aligned && bool::arbitrary(self.u)? { 1432 flags.set_notrap(); 1433 } 1434 1435 let (address, max_offset) = self.generate_load_store_address(builder, min_size, aligned)?; 1436 1437 // Pick an offset to pass into the load/store. 1438 let offset = if aligned { 1439 0 1440 } else { 1441 self.u.int_in_range(0..=max_offset)? as i32 1442 } 1443 .into(); 1444 1445 Ok((address, flags, offset)) 1446 } 1447 1448 /// Get a variable of type `ty` from the current function 1449 fn get_variable_of_type(&mut self, ty: Type) -> Result<Variable> { 1450 let opts = self.resources.vars.get(&ty).map_or(&[][..], Vec::as_slice); 1451 let var = self.u.choose(opts)?; 1452 Ok(*var) 1453 } 1454 1455 /// Generates an instruction(`iconst`/`fconst`/etc...) to introduce a constant value 1456 fn generate_const(&mut self, builder: &mut FunctionBuilder, ty: Type) -> Result<Value> { 1457 Ok(match self.u.datavalue(ty)? { 1458 DataValue::I8(i) => builder.ins().iconst(ty, i as i64), 1459 DataValue::I16(i) => builder.ins().iconst(ty, i as i64), 1460 DataValue::I32(i) => builder.ins().iconst(ty, i as i64), 1461 DataValue::I64(i) => builder.ins().iconst(ty, i as i64), 1462 DataValue::I128(i) => { 1463 let hi = builder.ins().iconst(I64, (i >> 64) as i64); 1464 let lo = builder.ins().iconst(I64, i as i64); 1465 builder.ins().iconcat(lo, hi) 1466 } 1467 DataValue::F32(f) => builder.ins().f32const(f), 1468 DataValue::F64(f) => builder.ins().f64const(f), 1469 DataValue::V128(bytes) => { 1470 let data = bytes.to_vec().into(); 1471 let handle = builder.func.dfg.constants.insert(data); 1472 builder.ins().vconst(ty, handle) 1473 } 1474 _ => unimplemented!(), 1475 }) 1476 } 1477 1478 /// Chooses a random block which can be targeted by a jump / branch. 1479 /// This means any block that is not the first block. 1480 fn generate_target_block(&mut self, source_block: Block) -> Result<Block> { 1481 // We try to mostly generate forward branches to avoid generating an excessive amount of 1482 // infinite loops. But they are still important, so give them a small chance of existing. 1483 let (backwards_blocks, forward_blocks) = 1484 self.resources.partition_target_blocks(source_block); 1485 let ratio = self.config.backwards_branch_ratio; 1486 let block_targets = if !backwards_blocks.is_empty() && self.u.ratio(ratio.0, ratio.1)? { 1487 backwards_blocks 1488 } else { 1489 forward_blocks 1490 }; 1491 assert!(!block_targets.is_empty()); 1492 1493 let (block, _) = self.u.choose(block_targets)?.clone(); 1494 Ok(block) 1495 } 1496 1497 fn generate_values_for_block( 1498 &mut self, 1499 builder: &mut FunctionBuilder, 1500 block: Block, 1501 ) -> Result<Vec<Value>> { 1502 let (_, sig) = self.resources.blocks[block.as_u32() as usize].clone(); 1503 self.generate_values_for_signature(builder, sig.iter().copied()) 1504 } 1505 1506 fn generate_values_for_signature<I: Iterator<Item = Type>>( 1507 &mut self, 1508 builder: &mut FunctionBuilder, 1509 signature: I, 1510 ) -> Result<Vec<Value>> { 1511 signature 1512 .map(|ty| { 1513 let var = self.get_variable_of_type(ty)?; 1514 let val = builder.use_var(var); 1515 Ok(val) 1516 }) 1517 .collect() 1518 } 1519 1520 /// The terminator that we need to insert has already been picked ahead of time 1521 /// we just need to build the instructions for it 1522 fn insert_terminator( 1523 &mut self, 1524 builder: &mut FunctionBuilder, 1525 source_block: Block, 1526 ) -> Result<()> { 1527 let terminator = self.resources.block_terminators[source_block.as_u32() as usize].clone(); 1528 1529 match terminator { 1530 BlockTerminator::Return => { 1531 let types: Vec<Type> = { 1532 let rets = &builder.func.signature.returns; 1533 rets.iter().map(|p| p.value_type).collect() 1534 }; 1535 let vals = self.generate_values_for_signature(builder, types.into_iter())?; 1536 1537 builder.ins().return_(&vals[..]); 1538 } 1539 BlockTerminator::Jump(target) => { 1540 let args = self.generate_values_for_block(builder, target)?; 1541 builder.ins().jump(target, &args[..]); 1542 } 1543 BlockTerminator::Br(left, right) => { 1544 let left_args = self.generate_values_for_block(builder, left)?; 1545 let right_args = self.generate_values_for_block(builder, right)?; 1546 1547 let condbr_types = [I8, I16, I32, I64, I128]; 1548 let _type = *self.u.choose(&condbr_types[..])?; 1549 let val = builder.use_var(self.get_variable_of_type(_type)?); 1550 builder 1551 .ins() 1552 .brif(val, left, &left_args[..], right, &right_args[..]); 1553 } 1554 BlockTerminator::BrTable(default, targets) => { 1555 // Create jump tables on demand 1556 let mut jt = Vec::with_capacity(targets.len()); 1557 for block in targets { 1558 let args = self.generate_values_for_block(builder, block)?; 1559 jt.push(builder.func.dfg.block_call(block, &args)) 1560 } 1561 1562 let args = self.generate_values_for_block(builder, default)?; 1563 let jt_data = JumpTableData::new(builder.func.dfg.block_call(default, &args), &jt); 1564 let jt = builder.create_jump_table(jt_data); 1565 1566 // br_table only supports I32 1567 let val = builder.use_var(self.get_variable_of_type(I32)?); 1568 1569 builder.ins().br_table(val, jt); 1570 } 1571 BlockTerminator::Switch(_type, default, entries) => { 1572 let mut switch = Switch::new(); 1573 for (&entry, &block) in entries.iter() { 1574 switch.set_entry(entry, block); 1575 } 1576 1577 let switch_val = builder.use_var(self.get_variable_of_type(_type)?); 1578 1579 switch.emit(builder, switch_val, default); 1580 } 1581 BlockTerminator::TailCall(target) | BlockTerminator::TailCallIndirect(target) => { 1582 let (sig, sig_ref, func_ref) = self 1583 .resources 1584 .func_refs 1585 .iter() 1586 .find(|(_, _, f)| *f == target) 1587 .expect("Failed to find previously selected function") 1588 .clone(); 1589 1590 let opcode = match terminator { 1591 BlockTerminator::TailCall(_) => Opcode::ReturnCall, 1592 BlockTerminator::TailCallIndirect(_) => Opcode::ReturnCallIndirect, 1593 _ => unreachable!(), 1594 }; 1595 1596 insert_call_to_function(self, builder, opcode, &sig, sig_ref, func_ref)?; 1597 } 1598 } 1599 1600 Ok(()) 1601 } 1602 1603 /// Fills the current block with random instructions 1604 fn generate_instructions(&mut self, builder: &mut FunctionBuilder) -> Result<()> { 1605 for _ in 0..self.param(&self.config.instructions_per_block)? { 1606 let (op, args, rets) = self.u.choose(&OPCODE_SIGNATURES)?; 1607 1608 // We filter out instructions that aren't supported by the target at this point instead 1609 // of building a single vector of valid instructions at the beginning of function 1610 // generation, to avoid invalidating the corpus when instructions are enabled/disabled. 1611 if !valid_for_target(&self.isa.triple(), *op, &args, &rets) { 1612 return Err(arbitrary::Error::IncorrectFormat.into()); 1613 } 1614 1615 let inserter = inserter_for_format(op.format()); 1616 inserter(self, builder, *op, &args, &rets)?; 1617 } 1618 1619 Ok(()) 1620 } 1621 1622 fn generate_funcrefs(&mut self, builder: &mut FunctionBuilder) -> Result<()> { 1623 let usercalls: Vec<(ExternalName, Signature)> = self 1624 .resources 1625 .usercalls 1626 .iter() 1627 .map(|(name, signature)| { 1628 let user_func_ref = builder.func.declare_imported_user_function(name.clone()); 1629 let name = ExternalName::User(user_func_ref); 1630 (name, signature.clone()) 1631 }) 1632 .collect(); 1633 1634 let lib_callconv = self.system_callconv(); 1635 let libcalls: Vec<(ExternalName, Signature)> = self 1636 .resources 1637 .libcalls 1638 .iter() 1639 .map(|libcall| { 1640 let pointer_type = Type::int_with_byte_size( 1641 self.isa.triple().pointer_width().unwrap().bytes().into(), 1642 ) 1643 .unwrap(); 1644 let signature = libcall.signature(lib_callconv, pointer_type); 1645 let name = ExternalName::LibCall(*libcall); 1646 (name, signature) 1647 }) 1648 .collect(); 1649 1650 for (name, signature) in usercalls.into_iter().chain(libcalls) { 1651 let sig_ref = builder.import_signature(signature.clone()); 1652 let func_ref = builder.import_function(ExtFuncData { 1653 name, 1654 signature: sig_ref, 1655 colocated: self.u.arbitrary()?, 1656 }); 1657 1658 self.resources 1659 .func_refs 1660 .push((signature, sig_ref, func_ref)); 1661 } 1662 1663 Ok(()) 1664 } 1665 1666 fn generate_stack_slots(&mut self, builder: &mut FunctionBuilder) -> Result<()> { 1667 for _ in 0..self.param(&self.config.static_stack_slots_per_function)? { 1668 let bytes = self.param(&self.config.static_stack_slot_size)? as u32; 1669 let ss_data = StackSlotData::new(StackSlotKind::ExplicitSlot, bytes); 1670 let slot = builder.create_sized_stack_slot(ss_data); 1671 self.resources.stack_slots.push((slot, bytes)); 1672 } 1673 1674 self.resources 1675 .stack_slots 1676 .sort_unstable_by_key(|&(_slot, bytes)| bytes); 1677 1678 Ok(()) 1679 } 1680 1681 /// Zero initializes the stack slot by inserting `stack_store`'s. 1682 fn initialize_stack_slots(&mut self, builder: &mut FunctionBuilder) -> Result<()> { 1683 let i8_zero = builder.ins().iconst(I8, 0); 1684 let i16_zero = builder.ins().iconst(I16, 0); 1685 let i32_zero = builder.ins().iconst(I32, 0); 1686 let i64_zero = builder.ins().iconst(I64, 0); 1687 let i128_zero = builder.ins().uextend(I128, i64_zero); 1688 1689 for &(slot, init_size) in self.resources.stack_slots.iter() { 1690 let mut size = init_size; 1691 1692 // Insert the largest available store for the remaining size. 1693 while size != 0 { 1694 let offset = (init_size - size) as i32; 1695 let (val, filled) = match size { 1696 sz if sz / 16 > 0 => (i128_zero, 16), 1697 sz if sz / 8 > 0 => (i64_zero, 8), 1698 sz if sz / 4 > 0 => (i32_zero, 4), 1699 sz if sz / 2 > 0 => (i16_zero, 2), 1700 _ => (i8_zero, 1), 1701 }; 1702 builder.ins().stack_store(val, slot, offset); 1703 size -= filled; 1704 } 1705 } 1706 Ok(()) 1707 } 1708 1709 /// Creates a random amount of blocks in this function 1710 fn generate_blocks(&mut self, builder: &mut FunctionBuilder) -> Result<()> { 1711 let extra_block_count = self.param(&self.config.blocks_per_function)?; 1712 1713 // We must always have at least one block, so we generate the "extra" blocks and add 1 for 1714 // the entry block. 1715 let block_count = 1 + extra_block_count; 1716 1717 // Blocks need to be sorted in ascending order 1718 self.resources.blocks = (0..block_count) 1719 .map(|i| { 1720 let is_entry = i == 0; 1721 let block = builder.create_block(); 1722 1723 // Optionally mark blocks that are not the entry block as cold 1724 if !is_entry { 1725 if bool::arbitrary(self.u)? { 1726 builder.set_cold_block(block); 1727 } 1728 } 1729 1730 // The first block has to have the function signature, but for the rest of them we generate 1731 // a random signature; 1732 if is_entry { 1733 builder.append_block_params_for_function_params(block); 1734 Ok(( 1735 block, 1736 self.signature.params.iter().map(|a| a.value_type).collect(), 1737 )) 1738 } else { 1739 let sig = self.generate_block_signature()?; 1740 sig.iter().for_each(|ty| { 1741 builder.append_block_param(block, *ty); 1742 }); 1743 Ok((block, sig)) 1744 } 1745 }) 1746 .collect::<Result<Vec<_>>>()?; 1747 1748 // Valid blocks for jump tables have to have no parameters in the signature, and must also 1749 // not be the first block. 1750 self.resources.blocks_without_params = self.resources.blocks[1..] 1751 .iter() 1752 .filter(|(_, sig)| sig.len() == 0) 1753 .map(|(b, _)| *b) 1754 .collect(); 1755 1756 // Compute the block CFG 1757 // 1758 // cranelift-frontend requires us to never generate unreachable blocks 1759 // To ensure this property we start by constructing a main "spine" of blocks. So block1 can 1760 // always jump to block2, and block2 can always jump to block3, etc... 1761 // 1762 // That is not a very interesting CFG, so we introduce variations on that, but always 1763 // ensuring that the property of pointing to the next block is maintained whatever the 1764 // branching mechanism we use. 1765 let blocks = self.resources.blocks.clone(); 1766 self.resources.block_terminators = blocks 1767 .iter() 1768 .map(|&(block, _)| { 1769 let next_block = Block::with_number(block.as_u32() + 1).unwrap(); 1770 let forward_blocks = self.resources.forward_blocks(block); 1771 let paramless_targets = self.resources.forward_blocks_without_params(block); 1772 let has_paramless_targets = !paramless_targets.is_empty(); 1773 let next_block_is_paramless = paramless_targets.contains(&next_block); 1774 1775 let mut valid_terminators = vec![]; 1776 1777 if forward_blocks.is_empty() { 1778 // Return is only valid on the last block. 1779 valid_terminators.push(BlockTerminatorKind::Return); 1780 } else { 1781 // If we have more than one block we can allow terminators that target blocks. 1782 // TODO: We could add some kind of BrReturn here, to explore edges where we 1783 // exit in the middle of the function 1784 valid_terminators.extend_from_slice(&[ 1785 BlockTerminatorKind::Jump, 1786 BlockTerminatorKind::Br, 1787 BlockTerminatorKind::BrTable, 1788 ]); 1789 } 1790 1791 // As the Switch interface only allows targeting blocks without params we need 1792 // to ensure that the next block has no params, since that one is guaranteed to be 1793 // picked in either case. 1794 if has_paramless_targets && next_block_is_paramless { 1795 valid_terminators.push(BlockTerminatorKind::Switch); 1796 } 1797 1798 // Tail Calls are a block terminator, so we should insert them as any other block 1799 // terminator. We should ensure that we can select at least one target before considering 1800 // them as candidate instructions. 1801 let has_tail_callees = self 1802 .resources 1803 .tail_call_targets(&self.signature) 1804 .next() 1805 .is_some(); 1806 let is_tail_caller = self.signature.call_conv == CallConv::Tail; 1807 1808 let supports_tail_calls = match self.isa.triple().architecture { 1809 Architecture::Aarch64(_) | Architecture::Riscv64(_) => true, 1810 // TODO: x64 currently requires frame pointers for tail calls. 1811 Architecture::X86_64 => self.isa.flags().preserve_frame_pointers(), 1812 // TODO: Other platforms do not support tail calls yet. 1813 _ => false, 1814 }; 1815 1816 if is_tail_caller && has_tail_callees && supports_tail_calls { 1817 valid_terminators.extend([ 1818 BlockTerminatorKind::TailCall, 1819 BlockTerminatorKind::TailCallIndirect, 1820 ]); 1821 } 1822 1823 let terminator = self.u.choose(&valid_terminators)?; 1824 1825 // Choose block targets for the terminators that we picked above 1826 Ok(match terminator { 1827 BlockTerminatorKind::Return => BlockTerminator::Return, 1828 BlockTerminatorKind::Jump => BlockTerminator::Jump(next_block), 1829 BlockTerminatorKind::Br => { 1830 BlockTerminator::Br(next_block, self.generate_target_block(block)?) 1831 } 1832 // TODO: Allow generating backwards branches here 1833 BlockTerminatorKind::BrTable => { 1834 // Make the default the next block, and then we don't have to worry 1835 // that we can reach it via the targets 1836 let default = next_block; 1837 1838 let target_count = self.param(&self.config.jump_table_entries)?; 1839 let targets = Result::from_iter( 1840 (0..target_count).map(|_| self.generate_target_block(block)), 1841 )?; 1842 1843 BlockTerminator::BrTable(default, targets) 1844 } 1845 BlockTerminatorKind::Switch => { 1846 // Make the default the next block, and then we don't have to worry 1847 // that we can reach it via the entries below 1848 let default_block = next_block; 1849 1850 let _type = *self.u.choose(&[I8, I16, I32, I64, I128][..])?; 1851 1852 // Build this into a HashMap since we cannot have duplicate entries. 1853 let mut entries = HashMap::new(); 1854 for _ in 0..self.param(&self.config.switch_cases)? { 1855 // The Switch API only allows for entries that are addressable by the index type 1856 // so we need to limit the range of values that we generate. 1857 let (ty_min, ty_max) = _type.bounds(false); 1858 let range_start = self.u.int_in_range(ty_min..=ty_max)?; 1859 1860 // We can either insert a contiguous range of blocks or a individual block 1861 // This is done because the Switch API specializes contiguous ranges. 1862 let range_size = if bool::arbitrary(self.u)? { 1863 1 1864 } else { 1865 self.param(&self.config.switch_max_range_size)? 1866 } as u128; 1867 1868 // Build the switch entries 1869 for i in 0..range_size { 1870 let index = range_start.wrapping_add(i) % ty_max; 1871 let block = *self 1872 .u 1873 .choose(self.resources.forward_blocks_without_params(block))?; 1874 1875 entries.insert(index, block); 1876 } 1877 } 1878 1879 BlockTerminator::Switch(_type, default_block, entries) 1880 } 1881 BlockTerminatorKind::TailCall => { 1882 let targets = self 1883 .resources 1884 .tail_call_targets(&self.signature) 1885 .collect::<Vec<_>>(); 1886 let (_, _, funcref) = *self.u.choose(&targets[..])?; 1887 BlockTerminator::TailCall(*funcref) 1888 } 1889 BlockTerminatorKind::TailCallIndirect => { 1890 let targets = self 1891 .resources 1892 .tail_call_targets(&self.signature) 1893 .collect::<Vec<_>>(); 1894 let (_, _, funcref) = *self.u.choose(&targets[..])?; 1895 BlockTerminator::TailCallIndirect(*funcref) 1896 } 1897 }) 1898 }) 1899 .collect::<Result<_>>()?; 1900 1901 Ok(()) 1902 } 1903 1904 fn generate_block_signature(&mut self) -> Result<BlockSignature> { 1905 let param_count = self.param(&self.config.block_signature_params)?; 1906 1907 let mut params = Vec::with_capacity(param_count); 1908 for _ in 0..param_count { 1909 params.push(self.u._type(self.isa.triple().architecture)?); 1910 } 1911 Ok(params) 1912 } 1913 1914 fn build_variable_pool(&mut self, builder: &mut FunctionBuilder) -> Result<()> { 1915 let block = builder.current_block().unwrap(); 1916 1917 // Define variables for the function signature 1918 let mut vars: Vec<_> = builder 1919 .func 1920 .signature 1921 .params 1922 .iter() 1923 .map(|param| param.value_type) 1924 .zip(builder.block_params(block).iter().copied()) 1925 .collect(); 1926 1927 // Create a pool of vars that are going to be used in this function 1928 for _ in 0..self.param(&self.config.vars_per_function)? { 1929 let ty = self.u._type(self.isa.triple().architecture)?; 1930 let value = self.generate_const(builder, ty)?; 1931 vars.push((ty, value)); 1932 } 1933 1934 for (id, (ty, value)) in vars.into_iter().enumerate() { 1935 let var = Variable::new(id); 1936 builder.declare_var(var, ty); 1937 builder.def_var(var, value); 1938 self.resources 1939 .vars 1940 .entry(ty) 1941 .or_insert_with(Vec::new) 1942 .push(var); 1943 } 1944 1945 Ok(()) 1946 } 1947 1948 /// We generate a function in multiple stages: 1949 /// 1950 /// * First we generate a random number of empty blocks 1951 /// * Then we generate a random pool of variables to be used throughout the function 1952 /// * We then visit each block and generate random instructions 1953 /// 1954 /// Because we generate all blocks and variables up front we already know everything that 1955 /// we need when generating instructions (i.e. jump targets / variables) 1956 pub fn generate(mut self) -> Result<Function> { 1957 let mut fn_builder_ctx = FunctionBuilderContext::new(); 1958 let mut func = Function::with_name_signature(self.name.clone(), self.signature.clone()); 1959 1960 let mut builder = FunctionBuilder::new(&mut func, &mut fn_builder_ctx); 1961 1962 // Build the function references before generating the block CFG since we store 1963 // function references in the CFG. 1964 self.generate_funcrefs(&mut builder)?; 1965 self.generate_blocks(&mut builder)?; 1966 1967 // Function preamble 1968 self.generate_stack_slots(&mut builder)?; 1969 1970 // Main instruction generation loop 1971 for (block, block_sig) in self.resources.blocks.clone().into_iter() { 1972 let is_block0 = block.as_u32() == 0; 1973 builder.switch_to_block(block); 1974 1975 if is_block0 { 1976 // The first block is special because we must create variables both for the 1977 // block signature and for the variable pool. Additionally, we must also define 1978 // initial values for all variables that are not the function signature. 1979 self.build_variable_pool(&mut builder)?; 1980 1981 // Stack slots have random bytes at the beginning of the function 1982 // initialize them to a constant value so that execution stays predictable. 1983 self.initialize_stack_slots(&mut builder)?; 1984 } else { 1985 // Define variables for the block params 1986 for (i, ty) in block_sig.iter().enumerate() { 1987 let var = self.get_variable_of_type(*ty)?; 1988 let block_param = builder.block_params(block)[i]; 1989 builder.def_var(var, block_param); 1990 } 1991 } 1992 1993 // Generate block instructions 1994 self.generate_instructions(&mut builder)?; 1995 1996 // Insert a terminator to safely exit the block 1997 self.insert_terminator(&mut builder, block)?; 1998 } 1999 2000 builder.seal_all_blocks(); 2001 builder.finalize(); 2002 2003 Ok(func) 2004 } 2005 } 2006