1 //===- ConstantRange.cpp - ConstantRange implementation -------------------===// 2 // 3 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions. 4 // See https://llvm.org/LICENSE.txt for license information. 5 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception 6 // 7 //===----------------------------------------------------------------------===// 8 // 9 // Represent a range of possible values that may occur when the program is run 10 // for an integral value. This keeps track of a lower and upper bound for the 11 // constant, which MAY wrap around the end of the numeric range. To do this, it 12 // keeps track of a [lower, upper) bound, which specifies an interval just like 13 // STL iterators. When used with boolean values, the following are important 14 // ranges (other integral ranges use min/max values for special range values): 15 // 16 // [F, F) = {} = Empty set 17 // [T, F) = {T} 18 // [F, T) = {F} 19 // [T, T) = {F, T} = Full set 20 // 21 //===----------------------------------------------------------------------===// 22 23 #include "llvm/ADT/APInt.h" 24 #include "llvm/Config/llvm-config.h" 25 #include "llvm/IR/ConstantRange.h" 26 #include "llvm/IR/Constants.h" 27 #include "llvm/IR/InstrTypes.h" 28 #include "llvm/IR/Instruction.h" 29 #include "llvm/IR/Metadata.h" 30 #include "llvm/IR/Operator.h" 31 #include "llvm/Support/Compiler.h" 32 #include "llvm/Support/Debug.h" 33 #include "llvm/Support/ErrorHandling.h" 34 #include "llvm/Support/KnownBits.h" 35 #include "llvm/Support/raw_ostream.h" 36 #include <algorithm> 37 #include <cassert> 38 #include <cstdint> 39 40 using namespace llvm; 41 42 ConstantRange::ConstantRange(uint32_t BitWidth, bool Full) 43 : Lower(Full ? APInt::getMaxValue(BitWidth) : APInt::getMinValue(BitWidth)), 44 Upper(Lower) {} 45 46 ConstantRange::ConstantRange(APInt V) 47 : Lower(std::move(V)), Upper(Lower + 1) {} 48 49 ConstantRange::ConstantRange(APInt L, APInt U) 50 : Lower(std::move(L)), Upper(std::move(U)) { 51 assert(Lower.getBitWidth() == Upper.getBitWidth() && 52 "ConstantRange with unequal bit widths"); 53 assert((Lower != Upper || (Lower.isMaxValue() || Lower.isMinValue())) && 54 "Lower == Upper, but they aren't min or max value!"); 55 } 56 57 ConstantRange ConstantRange::fromKnownBits(const KnownBits &Known, 58 bool IsSigned) { 59 assert(!Known.hasConflict() && "Expected valid KnownBits"); 60 61 if (Known.isUnknown()) 62 return getFull(Known.getBitWidth()); 63 64 // For unsigned ranges, or signed ranges with known sign bit, create a simple 65 // range between the smallest and largest possible value. 66 if (!IsSigned || Known.isNegative() || Known.isNonNegative()) 67 return ConstantRange(Known.One, ~Known.Zero + 1); 68 69 // If we don't know the sign bit, pick the lower bound as a negative number 70 // and the upper bound as a non-negative one. 71 APInt Lower = Known.One, Upper = ~Known.Zero; 72 Lower.setSignBit(); 73 Upper.clearSignBit(); 74 return ConstantRange(Lower, Upper + 1); 75 } 76 77 ConstantRange ConstantRange::makeAllowedICmpRegion(CmpInst::Predicate Pred, 78 const ConstantRange &CR) { 79 if (CR.isEmptySet()) 80 return CR; 81 82 uint32_t W = CR.getBitWidth(); 83 switch (Pred) { 84 default: 85 llvm_unreachable("Invalid ICmp predicate to makeAllowedICmpRegion()"); 86 case CmpInst::ICMP_EQ: 87 return CR; 88 case CmpInst::ICMP_NE: 89 if (CR.isSingleElement()) 90 return ConstantRange(CR.getUpper(), CR.getLower()); 91 return getFull(W); 92 case CmpInst::ICMP_ULT: { 93 APInt UMax(CR.getUnsignedMax()); 94 if (UMax.isMinValue()) 95 return getEmpty(W); 96 return ConstantRange(APInt::getMinValue(W), std::move(UMax)); 97 } 98 case CmpInst::ICMP_SLT: { 99 APInt SMax(CR.getSignedMax()); 100 if (SMax.isMinSignedValue()) 101 return getEmpty(W); 102 return ConstantRange(APInt::getSignedMinValue(W), std::move(SMax)); 103 } 104 case CmpInst::ICMP_ULE: 105 return getNonEmpty(APInt::getMinValue(W), CR.getUnsignedMax() + 1); 106 case CmpInst::ICMP_SLE: 107 return getNonEmpty(APInt::getSignedMinValue(W), CR.getSignedMax() + 1); 108 case CmpInst::ICMP_UGT: { 109 APInt UMin(CR.getUnsignedMin()); 110 if (UMin.isMaxValue()) 111 return getEmpty(W); 112 return ConstantRange(std::move(UMin) + 1, APInt::getNullValue(W)); 113 } 114 case CmpInst::ICMP_SGT: { 115 APInt SMin(CR.getSignedMin()); 116 if (SMin.isMaxSignedValue()) 117 return getEmpty(W); 118 return ConstantRange(std::move(SMin) + 1, APInt::getSignedMinValue(W)); 119 } 120 case CmpInst::ICMP_UGE: 121 return getNonEmpty(CR.getUnsignedMin(), APInt::getNullValue(W)); 122 case CmpInst::ICMP_SGE: 123 return getNonEmpty(CR.getSignedMin(), APInt::getSignedMinValue(W)); 124 } 125 } 126 127 ConstantRange ConstantRange::makeSatisfyingICmpRegion(CmpInst::Predicate Pred, 128 const ConstantRange &CR) { 129 // Follows from De-Morgan's laws: 130 // 131 // ~(~A union ~B) == A intersect B. 132 // 133 return makeAllowedICmpRegion(CmpInst::getInversePredicate(Pred), CR) 134 .inverse(); 135 } 136 137 ConstantRange ConstantRange::makeExactICmpRegion(CmpInst::Predicate Pred, 138 const APInt &C) { 139 // Computes the exact range that is equal to both the constant ranges returned 140 // by makeAllowedICmpRegion and makeSatisfyingICmpRegion. This is always true 141 // when RHS is a singleton such as an APInt and so the assert is valid. 142 // However for non-singleton RHS, for example ult [2,5) makeAllowedICmpRegion 143 // returns [0,4) but makeSatisfyICmpRegion returns [0,2). 144 // 145 assert(makeAllowedICmpRegion(Pred, C) == makeSatisfyingICmpRegion(Pred, C)); 146 return makeAllowedICmpRegion(Pred, C); 147 } 148 149 bool ConstantRange::getEquivalentICmp(CmpInst::Predicate &Pred, 150 APInt &RHS) const { 151 bool Success = false; 152 153 if (isFullSet() || isEmptySet()) { 154 Pred = isEmptySet() ? CmpInst::ICMP_ULT : CmpInst::ICMP_UGE; 155 RHS = APInt(getBitWidth(), 0); 156 Success = true; 157 } else if (auto *OnlyElt = getSingleElement()) { 158 Pred = CmpInst::ICMP_EQ; 159 RHS = *OnlyElt; 160 Success = true; 161 } else if (auto *OnlyMissingElt = getSingleMissingElement()) { 162 Pred = CmpInst::ICMP_NE; 163 RHS = *OnlyMissingElt; 164 Success = true; 165 } else if (getLower().isMinSignedValue() || getLower().isMinValue()) { 166 Pred = 167 getLower().isMinSignedValue() ? CmpInst::ICMP_SLT : CmpInst::ICMP_ULT; 168 RHS = getUpper(); 169 Success = true; 170 } else if (getUpper().isMinSignedValue() || getUpper().isMinValue()) { 171 Pred = 172 getUpper().isMinSignedValue() ? CmpInst::ICMP_SGE : CmpInst::ICMP_UGE; 173 RHS = getLower(); 174 Success = true; 175 } 176 177 assert((!Success || ConstantRange::makeExactICmpRegion(Pred, RHS) == *this) && 178 "Bad result!"); 179 180 return Success; 181 } 182 183 /// Exact mul nuw region for single element RHS. 184 static ConstantRange makeExactMulNUWRegion(const APInt &V) { 185 unsigned BitWidth = V.getBitWidth(); 186 if (V == 0) 187 return ConstantRange::getFull(V.getBitWidth()); 188 189 return ConstantRange::getNonEmpty( 190 APIntOps::RoundingUDiv(APInt::getMinValue(BitWidth), V, 191 APInt::Rounding::UP), 192 APIntOps::RoundingUDiv(APInt::getMaxValue(BitWidth), V, 193 APInt::Rounding::DOWN) + 1); 194 } 195 196 /// Exact mul nsw region for single element RHS. 197 static ConstantRange makeExactMulNSWRegion(const APInt &V) { 198 // Handle special case for 0, -1 and 1. See the last for reason why we 199 // specialize -1 and 1. 200 unsigned BitWidth = V.getBitWidth(); 201 if (V == 0 || V.isOneValue()) 202 return ConstantRange::getFull(BitWidth); 203 204 APInt MinValue = APInt::getSignedMinValue(BitWidth); 205 APInt MaxValue = APInt::getSignedMaxValue(BitWidth); 206 // e.g. Returning [-127, 127], represented as [-127, -128). 207 if (V.isAllOnesValue()) 208 return ConstantRange(-MaxValue, MinValue); 209 210 APInt Lower, Upper; 211 if (V.isNegative()) { 212 Lower = APIntOps::RoundingSDiv(MaxValue, V, APInt::Rounding::UP); 213 Upper = APIntOps::RoundingSDiv(MinValue, V, APInt::Rounding::DOWN); 214 } else { 215 Lower = APIntOps::RoundingSDiv(MinValue, V, APInt::Rounding::UP); 216 Upper = APIntOps::RoundingSDiv(MaxValue, V, APInt::Rounding::DOWN); 217 } 218 // ConstantRange ctor take a half inclusive interval [Lower, Upper + 1). 219 // Upper + 1 is guaranteed not to overflow, because |divisor| > 1. 0, -1, 220 // and 1 are already handled as special cases. 221 return ConstantRange(Lower, Upper + 1); 222 } 223 224 ConstantRange 225 ConstantRange::makeGuaranteedNoWrapRegion(Instruction::BinaryOps BinOp, 226 const ConstantRange &Other, 227 unsigned NoWrapKind) { 228 using OBO = OverflowingBinaryOperator; 229 230 assert(Instruction::isBinaryOp(BinOp) && "Binary operators only!"); 231 232 assert((NoWrapKind == OBO::NoSignedWrap || 233 NoWrapKind == OBO::NoUnsignedWrap) && 234 "NoWrapKind invalid!"); 235 236 bool Unsigned = NoWrapKind == OBO::NoUnsignedWrap; 237 unsigned BitWidth = Other.getBitWidth(); 238 239 switch (BinOp) { 240 default: 241 llvm_unreachable("Unsupported binary op"); 242 243 case Instruction::Add: { 244 if (Unsigned) 245 return getNonEmpty(APInt::getNullValue(BitWidth), 246 -Other.getUnsignedMax()); 247 248 APInt SignedMinVal = APInt::getSignedMinValue(BitWidth); 249 APInt SMin = Other.getSignedMin(), SMax = Other.getSignedMax(); 250 return getNonEmpty( 251 SMin.isNegative() ? SignedMinVal - SMin : SignedMinVal, 252 SMax.isStrictlyPositive() ? SignedMinVal - SMax : SignedMinVal); 253 } 254 255 case Instruction::Sub: { 256 if (Unsigned) 257 return getNonEmpty(Other.getUnsignedMax(), APInt::getMinValue(BitWidth)); 258 259 APInt SignedMinVal = APInt::getSignedMinValue(BitWidth); 260 APInt SMin = Other.getSignedMin(), SMax = Other.getSignedMax(); 261 return getNonEmpty( 262 SMax.isStrictlyPositive() ? SignedMinVal + SMax : SignedMinVal, 263 SMin.isNegative() ? SignedMinVal + SMin : SignedMinVal); 264 } 265 266 case Instruction::Mul: 267 if (Unsigned) 268 return makeExactMulNUWRegion(Other.getUnsignedMax()); 269 270 return makeExactMulNSWRegion(Other.getSignedMin()) 271 .intersectWith(makeExactMulNSWRegion(Other.getSignedMax())); 272 } 273 } 274 275 ConstantRange ConstantRange::makeExactNoWrapRegion(Instruction::BinaryOps BinOp, 276 const APInt &Other, 277 unsigned NoWrapKind) { 278 // makeGuaranteedNoWrapRegion() is exact for single-element ranges, as 279 // "for all" and "for any" coincide in this case. 280 return makeGuaranteedNoWrapRegion(BinOp, ConstantRange(Other), NoWrapKind); 281 } 282 283 bool ConstantRange::isFullSet() const { 284 return Lower == Upper && Lower.isMaxValue(); 285 } 286 287 bool ConstantRange::isEmptySet() const { 288 return Lower == Upper && Lower.isMinValue(); 289 } 290 291 bool ConstantRange::isWrappedSet() const { 292 return Lower.ugt(Upper) && !Upper.isNullValue(); 293 } 294 295 bool ConstantRange::isUpperWrapped() const { 296 return Lower.ugt(Upper); 297 } 298 299 bool ConstantRange::isSignWrappedSet() const { 300 return Lower.sgt(Upper) && !Upper.isMinSignedValue(); 301 } 302 303 bool ConstantRange::isUpperSignWrapped() const { 304 return Lower.sgt(Upper); 305 } 306 307 bool 308 ConstantRange::isSizeStrictlySmallerThan(const ConstantRange &Other) const { 309 assert(getBitWidth() == Other.getBitWidth()); 310 if (isFullSet()) 311 return false; 312 if (Other.isFullSet()) 313 return true; 314 return (Upper - Lower).ult(Other.Upper - Other.Lower); 315 } 316 317 bool 318 ConstantRange::isSizeLargerThan(uint64_t MaxSize) const { 319 assert(MaxSize && "MaxSize can't be 0."); 320 // If this a full set, we need special handling to avoid needing an extra bit 321 // to represent the size. 322 if (isFullSet()) 323 return APInt::getMaxValue(getBitWidth()).ugt(MaxSize - 1); 324 325 return (Upper - Lower).ugt(MaxSize); 326 } 327 328 bool ConstantRange::isAllNegative() const { 329 // Empty set is all negative, full set is not. 330 if (isEmptySet()) 331 return true; 332 if (isFullSet()) 333 return false; 334 335 return !isUpperSignWrapped() && !Upper.isStrictlyPositive(); 336 } 337 338 bool ConstantRange::isAllNonNegative() const { 339 // Empty and full set are automatically treated correctly. 340 return !isSignWrappedSet() && Lower.isNonNegative(); 341 } 342 343 APInt ConstantRange::getUnsignedMax() const { 344 if (isFullSet() || isUpperWrapped()) 345 return APInt::getMaxValue(getBitWidth()); 346 return getUpper() - 1; 347 } 348 349 APInt ConstantRange::getUnsignedMin() const { 350 if (isFullSet() || isWrappedSet()) 351 return APInt::getMinValue(getBitWidth()); 352 return getLower(); 353 } 354 355 APInt ConstantRange::getSignedMax() const { 356 if (isFullSet() || isUpperSignWrapped()) 357 return APInt::getSignedMaxValue(getBitWidth()); 358 return getUpper() - 1; 359 } 360 361 APInt ConstantRange::getSignedMin() const { 362 if (isFullSet() || isSignWrappedSet()) 363 return APInt::getSignedMinValue(getBitWidth()); 364 return getLower(); 365 } 366 367 bool ConstantRange::contains(const APInt &V) const { 368 if (Lower == Upper) 369 return isFullSet(); 370 371 if (!isUpperWrapped()) 372 return Lower.ule(V) && V.ult(Upper); 373 return Lower.ule(V) || V.ult(Upper); 374 } 375 376 bool ConstantRange::contains(const ConstantRange &Other) const { 377 if (isFullSet() || Other.isEmptySet()) return true; 378 if (isEmptySet() || Other.isFullSet()) return false; 379 380 if (!isUpperWrapped()) { 381 if (Other.isUpperWrapped()) 382 return false; 383 384 return Lower.ule(Other.getLower()) && Other.getUpper().ule(Upper); 385 } 386 387 if (!Other.isUpperWrapped()) 388 return Other.getUpper().ule(Upper) || 389 Lower.ule(Other.getLower()); 390 391 return Other.getUpper().ule(Upper) && Lower.ule(Other.getLower()); 392 } 393 394 ConstantRange ConstantRange::subtract(const APInt &Val) const { 395 assert(Val.getBitWidth() == getBitWidth() && "Wrong bit width"); 396 // If the set is empty or full, don't modify the endpoints. 397 if (Lower == Upper) 398 return *this; 399 return ConstantRange(Lower - Val, Upper - Val); 400 } 401 402 ConstantRange ConstantRange::difference(const ConstantRange &CR) const { 403 return intersectWith(CR.inverse()); 404 } 405 406 static ConstantRange getPreferredRange( 407 const ConstantRange &CR1, const ConstantRange &CR2, 408 ConstantRange::PreferredRangeType Type) { 409 if (Type == ConstantRange::Unsigned) { 410 if (!CR1.isWrappedSet() && CR2.isWrappedSet()) 411 return CR1; 412 if (CR1.isWrappedSet() && !CR2.isWrappedSet()) 413 return CR2; 414 } else if (Type == ConstantRange::Signed) { 415 if (!CR1.isSignWrappedSet() && CR2.isSignWrappedSet()) 416 return CR1; 417 if (CR1.isSignWrappedSet() && !CR2.isSignWrappedSet()) 418 return CR2; 419 } 420 421 if (CR1.isSizeStrictlySmallerThan(CR2)) 422 return CR1; 423 return CR2; 424 } 425 426 ConstantRange ConstantRange::intersectWith(const ConstantRange &CR, 427 PreferredRangeType Type) const { 428 assert(getBitWidth() == CR.getBitWidth() && 429 "ConstantRange types don't agree!"); 430 431 // Handle common cases. 432 if ( isEmptySet() || CR.isFullSet()) return *this; 433 if (CR.isEmptySet() || isFullSet()) return CR; 434 435 if (!isUpperWrapped() && CR.isUpperWrapped()) 436 return CR.intersectWith(*this, Type); 437 438 if (!isUpperWrapped() && !CR.isUpperWrapped()) { 439 if (Lower.ult(CR.Lower)) { 440 // L---U : this 441 // L---U : CR 442 if (Upper.ule(CR.Lower)) 443 return getEmpty(); 444 445 // L---U : this 446 // L---U : CR 447 if (Upper.ult(CR.Upper)) 448 return ConstantRange(CR.Lower, Upper); 449 450 // L-------U : this 451 // L---U : CR 452 return CR; 453 } 454 // L---U : this 455 // L-------U : CR 456 if (Upper.ult(CR.Upper)) 457 return *this; 458 459 // L-----U : this 460 // L-----U : CR 461 if (Lower.ult(CR.Upper)) 462 return ConstantRange(Lower, CR.Upper); 463 464 // L---U : this 465 // L---U : CR 466 return getEmpty(); 467 } 468 469 if (isUpperWrapped() && !CR.isUpperWrapped()) { 470 if (CR.Lower.ult(Upper)) { 471 // ------U L--- : this 472 // L--U : CR 473 if (CR.Upper.ult(Upper)) 474 return CR; 475 476 // ------U L--- : this 477 // L------U : CR 478 if (CR.Upper.ule(Lower)) 479 return ConstantRange(CR.Lower, Upper); 480 481 // ------U L--- : this 482 // L----------U : CR 483 return getPreferredRange(*this, CR, Type); 484 } 485 if (CR.Lower.ult(Lower)) { 486 // --U L---- : this 487 // L--U : CR 488 if (CR.Upper.ule(Lower)) 489 return getEmpty(); 490 491 // --U L---- : this 492 // L------U : CR 493 return ConstantRange(Lower, CR.Upper); 494 } 495 496 // --U L------ : this 497 // L--U : CR 498 return CR; 499 } 500 501 if (CR.Upper.ult(Upper)) { 502 // ------U L-- : this 503 // --U L------ : CR 504 if (CR.Lower.ult(Upper)) 505 return getPreferredRange(*this, CR, Type); 506 507 // ----U L-- : this 508 // --U L---- : CR 509 if (CR.Lower.ult(Lower)) 510 return ConstantRange(Lower, CR.Upper); 511 512 // ----U L---- : this 513 // --U L-- : CR 514 return CR; 515 } 516 if (CR.Upper.ule(Lower)) { 517 // --U L-- : this 518 // ----U L---- : CR 519 if (CR.Lower.ult(Lower)) 520 return *this; 521 522 // --U L---- : this 523 // ----U L-- : CR 524 return ConstantRange(CR.Lower, Upper); 525 } 526 527 // --U L------ : this 528 // ------U L-- : CR 529 return getPreferredRange(*this, CR, Type); 530 } 531 532 ConstantRange ConstantRange::unionWith(const ConstantRange &CR, 533 PreferredRangeType Type) const { 534 assert(getBitWidth() == CR.getBitWidth() && 535 "ConstantRange types don't agree!"); 536 537 if ( isFullSet() || CR.isEmptySet()) return *this; 538 if (CR.isFullSet() || isEmptySet()) return CR; 539 540 if (!isUpperWrapped() && CR.isUpperWrapped()) 541 return CR.unionWith(*this, Type); 542 543 if (!isUpperWrapped() && !CR.isUpperWrapped()) { 544 // L---U and L---U : this 545 // L---U L---U : CR 546 // result in one of 547 // L---------U 548 // -----U L----- 549 if (CR.Upper.ult(Lower) || Upper.ult(CR.Lower)) 550 return getPreferredRange( 551 ConstantRange(Lower, CR.Upper), ConstantRange(CR.Lower, Upper), Type); 552 553 APInt L = CR.Lower.ult(Lower) ? CR.Lower : Lower; 554 APInt U = (CR.Upper - 1).ugt(Upper - 1) ? CR.Upper : Upper; 555 556 if (L.isNullValue() && U.isNullValue()) 557 return getFull(); 558 559 return ConstantRange(std::move(L), std::move(U)); 560 } 561 562 if (!CR.isUpperWrapped()) { 563 // ------U L----- and ------U L----- : this 564 // L--U L--U : CR 565 if (CR.Upper.ule(Upper) || CR.Lower.uge(Lower)) 566 return *this; 567 568 // ------U L----- : this 569 // L---------U : CR 570 if (CR.Lower.ule(Upper) && Lower.ule(CR.Upper)) 571 return getFull(); 572 573 // ----U L---- : this 574 // L---U : CR 575 // results in one of 576 // ----------U L---- 577 // ----U L---------- 578 if (Upper.ult(CR.Lower) && CR.Upper.ult(Lower)) 579 return getPreferredRange( 580 ConstantRange(Lower, CR.Upper), ConstantRange(CR.Lower, Upper), Type); 581 582 // ----U L----- : this 583 // L----U : CR 584 if (Upper.ult(CR.Lower) && Lower.ule(CR.Upper)) 585 return ConstantRange(CR.Lower, Upper); 586 587 // ------U L---- : this 588 // L-----U : CR 589 assert(CR.Lower.ule(Upper) && CR.Upper.ult(Lower) && 590 "ConstantRange::unionWith missed a case with one range wrapped"); 591 return ConstantRange(Lower, CR.Upper); 592 } 593 594 // ------U L---- and ------U L---- : this 595 // -U L----------- and ------------U L : CR 596 if (CR.Lower.ule(Upper) || Lower.ule(CR.Upper)) 597 return getFull(); 598 599 APInt L = CR.Lower.ult(Lower) ? CR.Lower : Lower; 600 APInt U = CR.Upper.ugt(Upper) ? CR.Upper : Upper; 601 602 return ConstantRange(std::move(L), std::move(U)); 603 } 604 605 ConstantRange ConstantRange::castOp(Instruction::CastOps CastOp, 606 uint32_t ResultBitWidth) const { 607 switch (CastOp) { 608 default: 609 llvm_unreachable("unsupported cast type"); 610 case Instruction::Trunc: 611 return truncate(ResultBitWidth); 612 case Instruction::SExt: 613 return signExtend(ResultBitWidth); 614 case Instruction::ZExt: 615 return zeroExtend(ResultBitWidth); 616 case Instruction::BitCast: 617 return *this; 618 case Instruction::FPToUI: 619 case Instruction::FPToSI: 620 if (getBitWidth() == ResultBitWidth) 621 return *this; 622 else 623 return getFull(); 624 case Instruction::UIToFP: { 625 // TODO: use input range if available 626 auto BW = getBitWidth(); 627 APInt Min = APInt::getMinValue(BW).zextOrSelf(ResultBitWidth); 628 APInt Max = APInt::getMaxValue(BW).zextOrSelf(ResultBitWidth); 629 return ConstantRange(std::move(Min), std::move(Max)); 630 } 631 case Instruction::SIToFP: { 632 // TODO: use input range if available 633 auto BW = getBitWidth(); 634 APInt SMin = APInt::getSignedMinValue(BW).sextOrSelf(ResultBitWidth); 635 APInt SMax = APInt::getSignedMaxValue(BW).sextOrSelf(ResultBitWidth); 636 return ConstantRange(std::move(SMin), std::move(SMax)); 637 } 638 case Instruction::FPTrunc: 639 case Instruction::FPExt: 640 case Instruction::IntToPtr: 641 case Instruction::PtrToInt: 642 case Instruction::AddrSpaceCast: 643 // Conservatively return getFull set. 644 return getFull(); 645 }; 646 } 647 648 ConstantRange ConstantRange::zeroExtend(uint32_t DstTySize) const { 649 if (isEmptySet()) return getEmpty(DstTySize); 650 651 unsigned SrcTySize = getBitWidth(); 652 assert(SrcTySize < DstTySize && "Not a value extension"); 653 if (isFullSet() || isUpperWrapped()) { 654 // Change into [0, 1 << src bit width) 655 APInt LowerExt(DstTySize, 0); 656 if (!Upper) // special case: [X, 0) -- not really wrapping around 657 LowerExt = Lower.zext(DstTySize); 658 return ConstantRange(std::move(LowerExt), 659 APInt::getOneBitSet(DstTySize, SrcTySize)); 660 } 661 662 return ConstantRange(Lower.zext(DstTySize), Upper.zext(DstTySize)); 663 } 664 665 ConstantRange ConstantRange::signExtend(uint32_t DstTySize) const { 666 if (isEmptySet()) return getEmpty(DstTySize); 667 668 unsigned SrcTySize = getBitWidth(); 669 assert(SrcTySize < DstTySize && "Not a value extension"); 670 671 // special case: [X, INT_MIN) -- not really wrapping around 672 if (Upper.isMinSignedValue()) 673 return ConstantRange(Lower.sext(DstTySize), Upper.zext(DstTySize)); 674 675 if (isFullSet() || isSignWrappedSet()) { 676 return ConstantRange(APInt::getHighBitsSet(DstTySize,DstTySize-SrcTySize+1), 677 APInt::getLowBitsSet(DstTySize, SrcTySize-1) + 1); 678 } 679 680 return ConstantRange(Lower.sext(DstTySize), Upper.sext(DstTySize)); 681 } 682 683 ConstantRange ConstantRange::truncate(uint32_t DstTySize) const { 684 assert(getBitWidth() > DstTySize && "Not a value truncation"); 685 if (isEmptySet()) 686 return getEmpty(DstTySize); 687 if (isFullSet()) 688 return getFull(DstTySize); 689 690 APInt LowerDiv(Lower), UpperDiv(Upper); 691 ConstantRange Union(DstTySize, /*isFullSet=*/false); 692 693 // Analyze wrapped sets in their two parts: [0, Upper) \/ [Lower, MaxValue] 694 // We use the non-wrapped set code to analyze the [Lower, MaxValue) part, and 695 // then we do the union with [MaxValue, Upper) 696 if (isUpperWrapped()) { 697 // If Upper is greater than or equal to MaxValue(DstTy), it covers the whole 698 // truncated range. 699 if (Upper.getActiveBits() > DstTySize || 700 Upper.countTrailingOnes() == DstTySize) 701 return getFull(DstTySize); 702 703 Union = ConstantRange(APInt::getMaxValue(DstTySize),Upper.trunc(DstTySize)); 704 UpperDiv.setAllBits(); 705 706 // Union covers the MaxValue case, so return if the remaining range is just 707 // MaxValue(DstTy). 708 if (LowerDiv == UpperDiv) 709 return Union; 710 } 711 712 // Chop off the most significant bits that are past the destination bitwidth. 713 if (LowerDiv.getActiveBits() > DstTySize) { 714 // Mask to just the signficant bits and subtract from LowerDiv/UpperDiv. 715 APInt Adjust = LowerDiv & APInt::getBitsSetFrom(getBitWidth(), DstTySize); 716 LowerDiv -= Adjust; 717 UpperDiv -= Adjust; 718 } 719 720 unsigned UpperDivWidth = UpperDiv.getActiveBits(); 721 if (UpperDivWidth <= DstTySize) 722 return ConstantRange(LowerDiv.trunc(DstTySize), 723 UpperDiv.trunc(DstTySize)).unionWith(Union); 724 725 // The truncated value wraps around. Check if we can do better than fullset. 726 if (UpperDivWidth == DstTySize + 1) { 727 // Clear the MSB so that UpperDiv wraps around. 728 UpperDiv.clearBit(DstTySize); 729 if (UpperDiv.ult(LowerDiv)) 730 return ConstantRange(LowerDiv.trunc(DstTySize), 731 UpperDiv.trunc(DstTySize)).unionWith(Union); 732 } 733 734 return getFull(DstTySize); 735 } 736 737 ConstantRange ConstantRange::zextOrTrunc(uint32_t DstTySize) const { 738 unsigned SrcTySize = getBitWidth(); 739 if (SrcTySize > DstTySize) 740 return truncate(DstTySize); 741 if (SrcTySize < DstTySize) 742 return zeroExtend(DstTySize); 743 return *this; 744 } 745 746 ConstantRange ConstantRange::sextOrTrunc(uint32_t DstTySize) const { 747 unsigned SrcTySize = getBitWidth(); 748 if (SrcTySize > DstTySize) 749 return truncate(DstTySize); 750 if (SrcTySize < DstTySize) 751 return signExtend(DstTySize); 752 return *this; 753 } 754 755 ConstantRange ConstantRange::binaryOp(Instruction::BinaryOps BinOp, 756 const ConstantRange &Other) const { 757 assert(Instruction::isBinaryOp(BinOp) && "Binary operators only!"); 758 759 switch (BinOp) { 760 case Instruction::Add: 761 return add(Other); 762 case Instruction::Sub: 763 return sub(Other); 764 case Instruction::Mul: 765 return multiply(Other); 766 case Instruction::UDiv: 767 return udiv(Other); 768 case Instruction::SDiv: 769 return sdiv(Other); 770 case Instruction::URem: 771 return urem(Other); 772 case Instruction::SRem: 773 return srem(Other); 774 case Instruction::Shl: 775 return shl(Other); 776 case Instruction::LShr: 777 return lshr(Other); 778 case Instruction::AShr: 779 return ashr(Other); 780 case Instruction::And: 781 return binaryAnd(Other); 782 case Instruction::Or: 783 return binaryOr(Other); 784 // Note: floating point operations applied to abstract ranges are just 785 // ideal integer operations with a lossy representation 786 case Instruction::FAdd: 787 return add(Other); 788 case Instruction::FSub: 789 return sub(Other); 790 case Instruction::FMul: 791 return multiply(Other); 792 default: 793 // Conservatively return getFull set. 794 return getFull(); 795 } 796 } 797 798 ConstantRange 799 ConstantRange::add(const ConstantRange &Other) const { 800 if (isEmptySet() || Other.isEmptySet()) 801 return getEmpty(); 802 if (isFullSet() || Other.isFullSet()) 803 return getFull(); 804 805 APInt NewLower = getLower() + Other.getLower(); 806 APInt NewUpper = getUpper() + Other.getUpper() - 1; 807 if (NewLower == NewUpper) 808 return getFull(); 809 810 ConstantRange X = ConstantRange(std::move(NewLower), std::move(NewUpper)); 811 if (X.isSizeStrictlySmallerThan(*this) || 812 X.isSizeStrictlySmallerThan(Other)) 813 // We've wrapped, therefore, full set. 814 return getFull(); 815 return X; 816 } 817 818 ConstantRange ConstantRange::addWithNoSignedWrap(const APInt &Other) const { 819 // Calculate the subset of this range such that "X + Other" is 820 // guaranteed not to wrap (overflow) for all X in this subset. 821 auto NSWRange = ConstantRange::makeExactNoWrapRegion( 822 BinaryOperator::Add, Other, OverflowingBinaryOperator::NoSignedWrap); 823 auto NSWConstrainedRange = intersectWith(NSWRange); 824 825 return NSWConstrainedRange.add(ConstantRange(Other)); 826 } 827 828 ConstantRange 829 ConstantRange::sub(const ConstantRange &Other) const { 830 if (isEmptySet() || Other.isEmptySet()) 831 return getEmpty(); 832 if (isFullSet() || Other.isFullSet()) 833 return getFull(); 834 835 APInt NewLower = getLower() - Other.getUpper() + 1; 836 APInt NewUpper = getUpper() - Other.getLower(); 837 if (NewLower == NewUpper) 838 return getFull(); 839 840 ConstantRange X = ConstantRange(std::move(NewLower), std::move(NewUpper)); 841 if (X.isSizeStrictlySmallerThan(*this) || 842 X.isSizeStrictlySmallerThan(Other)) 843 // We've wrapped, therefore, full set. 844 return getFull(); 845 return X; 846 } 847 848 ConstantRange 849 ConstantRange::multiply(const ConstantRange &Other) const { 850 // TODO: If either operand is a single element and the multiply is known to 851 // be non-wrapping, round the result min and max value to the appropriate 852 // multiple of that element. If wrapping is possible, at least adjust the 853 // range according to the greatest power-of-two factor of the single element. 854 855 if (isEmptySet() || Other.isEmptySet()) 856 return getEmpty(); 857 858 // Multiplication is signedness-independent. However different ranges can be 859 // obtained depending on how the input ranges are treated. These different 860 // ranges are all conservatively correct, but one might be better than the 861 // other. We calculate two ranges; one treating the inputs as unsigned 862 // and the other signed, then return the smallest of these ranges. 863 864 // Unsigned range first. 865 APInt this_min = getUnsignedMin().zext(getBitWidth() * 2); 866 APInt this_max = getUnsignedMax().zext(getBitWidth() * 2); 867 APInt Other_min = Other.getUnsignedMin().zext(getBitWidth() * 2); 868 APInt Other_max = Other.getUnsignedMax().zext(getBitWidth() * 2); 869 870 ConstantRange Result_zext = ConstantRange(this_min * Other_min, 871 this_max * Other_max + 1); 872 ConstantRange UR = Result_zext.truncate(getBitWidth()); 873 874 // If the unsigned range doesn't wrap, and isn't negative then it's a range 875 // from one positive number to another which is as good as we can generate. 876 // In this case, skip the extra work of generating signed ranges which aren't 877 // going to be better than this range. 878 if (!UR.isUpperWrapped() && 879 (UR.getUpper().isNonNegative() || UR.getUpper().isMinSignedValue())) 880 return UR; 881 882 // Now the signed range. Because we could be dealing with negative numbers 883 // here, the lower bound is the smallest of the cartesian product of the 884 // lower and upper ranges; for example: 885 // [-1,4) * [-2,3) = min(-1*-2, -1*2, 3*-2, 3*2) = -6. 886 // Similarly for the upper bound, swapping min for max. 887 888 this_min = getSignedMin().sext(getBitWidth() * 2); 889 this_max = getSignedMax().sext(getBitWidth() * 2); 890 Other_min = Other.getSignedMin().sext(getBitWidth() * 2); 891 Other_max = Other.getSignedMax().sext(getBitWidth() * 2); 892 893 auto L = {this_min * Other_min, this_min * Other_max, 894 this_max * Other_min, this_max * Other_max}; 895 auto Compare = [](const APInt &A, const APInt &B) { return A.slt(B); }; 896 ConstantRange Result_sext(std::min(L, Compare), std::max(L, Compare) + 1); 897 ConstantRange SR = Result_sext.truncate(getBitWidth()); 898 899 return UR.isSizeStrictlySmallerThan(SR) ? UR : SR; 900 } 901 902 ConstantRange 903 ConstantRange::smax(const ConstantRange &Other) const { 904 // X smax Y is: range(smax(X_smin, Y_smin), 905 // smax(X_smax, Y_smax)) 906 if (isEmptySet() || Other.isEmptySet()) 907 return getEmpty(); 908 APInt NewL = APIntOps::smax(getSignedMin(), Other.getSignedMin()); 909 APInt NewU = APIntOps::smax(getSignedMax(), Other.getSignedMax()) + 1; 910 return getNonEmpty(std::move(NewL), std::move(NewU)); 911 } 912 913 ConstantRange 914 ConstantRange::umax(const ConstantRange &Other) const { 915 // X umax Y is: range(umax(X_umin, Y_umin), 916 // umax(X_umax, Y_umax)) 917 if (isEmptySet() || Other.isEmptySet()) 918 return getEmpty(); 919 APInt NewL = APIntOps::umax(getUnsignedMin(), Other.getUnsignedMin()); 920 APInt NewU = APIntOps::umax(getUnsignedMax(), Other.getUnsignedMax()) + 1; 921 return getNonEmpty(std::move(NewL), std::move(NewU)); 922 } 923 924 ConstantRange 925 ConstantRange::smin(const ConstantRange &Other) const { 926 // X smin Y is: range(smin(X_smin, Y_smin), 927 // smin(X_smax, Y_smax)) 928 if (isEmptySet() || Other.isEmptySet()) 929 return getEmpty(); 930 APInt NewL = APIntOps::smin(getSignedMin(), Other.getSignedMin()); 931 APInt NewU = APIntOps::smin(getSignedMax(), Other.getSignedMax()) + 1; 932 return getNonEmpty(std::move(NewL), std::move(NewU)); 933 } 934 935 ConstantRange 936 ConstantRange::umin(const ConstantRange &Other) const { 937 // X umin Y is: range(umin(X_umin, Y_umin), 938 // umin(X_umax, Y_umax)) 939 if (isEmptySet() || Other.isEmptySet()) 940 return getEmpty(); 941 APInt NewL = APIntOps::umin(getUnsignedMin(), Other.getUnsignedMin()); 942 APInt NewU = APIntOps::umin(getUnsignedMax(), Other.getUnsignedMax()) + 1; 943 return getNonEmpty(std::move(NewL), std::move(NewU)); 944 } 945 946 ConstantRange 947 ConstantRange::udiv(const ConstantRange &RHS) const { 948 if (isEmptySet() || RHS.isEmptySet() || RHS.getUnsignedMax().isNullValue()) 949 return getEmpty(); 950 951 APInt Lower = getUnsignedMin().udiv(RHS.getUnsignedMax()); 952 953 APInt RHS_umin = RHS.getUnsignedMin(); 954 if (RHS_umin.isNullValue()) { 955 // We want the lowest value in RHS excluding zero. Usually that would be 1 956 // except for a range in the form of [X, 1) in which case it would be X. 957 if (RHS.getUpper() == 1) 958 RHS_umin = RHS.getLower(); 959 else 960 RHS_umin = 1; 961 } 962 963 APInt Upper = getUnsignedMax().udiv(RHS_umin) + 1; 964 return getNonEmpty(std::move(Lower), std::move(Upper)); 965 } 966 967 ConstantRange ConstantRange::sdiv(const ConstantRange &RHS) const { 968 // We split up the LHS and RHS into positive and negative components 969 // and then also compute the positive and negative components of the result 970 // separately by combining division results with the appropriate signs. 971 APInt Zero = APInt::getNullValue(getBitWidth()); 972 APInt SignedMin = APInt::getSignedMinValue(getBitWidth()); 973 ConstantRange PosFilter(APInt(getBitWidth(), 1), SignedMin); 974 ConstantRange NegFilter(SignedMin, Zero); 975 ConstantRange PosL = intersectWith(PosFilter); 976 ConstantRange NegL = intersectWith(NegFilter); 977 ConstantRange PosR = RHS.intersectWith(PosFilter); 978 ConstantRange NegR = RHS.intersectWith(NegFilter); 979 980 ConstantRange PosRes = getEmpty(); 981 if (!PosL.isEmptySet() && !PosR.isEmptySet()) 982 // pos / pos = pos. 983 PosRes = ConstantRange(PosL.Lower.sdiv(PosR.Upper - 1), 984 (PosL.Upper - 1).sdiv(PosR.Lower) + 1); 985 986 if (!NegL.isEmptySet() && !NegR.isEmptySet()) { 987 // neg / neg = pos. 988 // 989 // We need to deal with one tricky case here: SignedMin / -1 is UB on the 990 // IR level, so we'll want to exclude this case when calculating bounds. 991 // (For APInts the operation is well-defined and yields SignedMin.) We 992 // handle this by dropping either SignedMin from the LHS or -1 from the RHS. 993 APInt Lo = (NegL.Upper - 1).sdiv(NegR.Lower); 994 if (NegL.Lower.isMinSignedValue() && NegR.Upper.isNullValue()) { 995 // Remove -1 from the LHS. Skip if it's the only element, as this would 996 // leave us with an empty set. 997 if (!NegR.Lower.isAllOnesValue()) { 998 APInt AdjNegRUpper; 999 if (RHS.Lower.isAllOnesValue()) 1000 // Negative part of [-1, X] without -1 is [SignedMin, X]. 1001 AdjNegRUpper = RHS.Upper; 1002 else 1003 // [X, -1] without -1 is [X, -2]. 1004 AdjNegRUpper = NegR.Upper - 1; 1005 1006 PosRes = PosRes.unionWith( 1007 ConstantRange(Lo, NegL.Lower.sdiv(AdjNegRUpper - 1) + 1)); 1008 } 1009 1010 // Remove SignedMin from the RHS. Skip if it's the only element, as this 1011 // would leave us with an empty set. 1012 if (NegL.Upper != SignedMin + 1) { 1013 APInt AdjNegLLower; 1014 if (Upper == SignedMin + 1) 1015 // Negative part of [X, SignedMin] without SignedMin is [X, -1]. 1016 AdjNegLLower = Lower; 1017 else 1018 // [SignedMin, X] without SignedMin is [SignedMin + 1, X]. 1019 AdjNegLLower = NegL.Lower + 1; 1020 1021 PosRes = PosRes.unionWith( 1022 ConstantRange(std::move(Lo), 1023 AdjNegLLower.sdiv(NegR.Upper - 1) + 1)); 1024 } 1025 } else { 1026 PosRes = PosRes.unionWith( 1027 ConstantRange(std::move(Lo), NegL.Lower.sdiv(NegR.Upper - 1) + 1)); 1028 } 1029 } 1030 1031 ConstantRange NegRes = getEmpty(); 1032 if (!PosL.isEmptySet() && !NegR.isEmptySet()) 1033 // pos / neg = neg. 1034 NegRes = ConstantRange((PosL.Upper - 1).sdiv(NegR.Upper - 1), 1035 PosL.Lower.sdiv(NegR.Lower) + 1); 1036 1037 if (!NegL.isEmptySet() && !PosR.isEmptySet()) 1038 // neg / pos = neg. 1039 NegRes = NegRes.unionWith( 1040 ConstantRange(NegL.Lower.sdiv(PosR.Lower), 1041 (NegL.Upper - 1).sdiv(PosR.Upper - 1) + 1)); 1042 1043 // Prefer a non-wrapping signed range here. 1044 ConstantRange Res = NegRes.unionWith(PosRes, PreferredRangeType::Signed); 1045 1046 // Preserve the zero that we dropped when splitting the LHS by sign. 1047 if (contains(Zero) && (!PosR.isEmptySet() || !NegR.isEmptySet())) 1048 Res = Res.unionWith(ConstantRange(Zero)); 1049 return Res; 1050 } 1051 1052 ConstantRange ConstantRange::urem(const ConstantRange &RHS) const { 1053 if (isEmptySet() || RHS.isEmptySet() || RHS.getUnsignedMax().isNullValue()) 1054 return getEmpty(); 1055 1056 // L % R for L < R is L. 1057 if (getUnsignedMax().ult(RHS.getUnsignedMin())) 1058 return *this; 1059 1060 // L % R is <= L and < R. 1061 APInt Upper = APIntOps::umin(getUnsignedMax(), RHS.getUnsignedMax() - 1) + 1; 1062 return getNonEmpty(APInt::getNullValue(getBitWidth()), std::move(Upper)); 1063 } 1064 1065 ConstantRange ConstantRange::srem(const ConstantRange &RHS) const { 1066 if (isEmptySet() || RHS.isEmptySet()) 1067 return getEmpty(); 1068 1069 ConstantRange AbsRHS = RHS.abs(); 1070 APInt MinAbsRHS = AbsRHS.getUnsignedMin(); 1071 APInt MaxAbsRHS = AbsRHS.getUnsignedMax(); 1072 1073 // Modulus by zero is UB. 1074 if (MaxAbsRHS.isNullValue()) 1075 return getEmpty(); 1076 1077 if (MinAbsRHS.isNullValue()) 1078 ++MinAbsRHS; 1079 1080 APInt MinLHS = getSignedMin(), MaxLHS = getSignedMax(); 1081 1082 if (MinLHS.isNonNegative()) { 1083 // L % R for L < R is L. 1084 if (MaxLHS.ult(MinAbsRHS)) 1085 return *this; 1086 1087 // L % R is <= L and < R. 1088 APInt Upper = APIntOps::umin(MaxLHS, MaxAbsRHS - 1) + 1; 1089 return ConstantRange(APInt::getNullValue(getBitWidth()), std::move(Upper)); 1090 } 1091 1092 // Same basic logic as above, but the result is negative. 1093 if (MaxLHS.isNegative()) { 1094 if (MinLHS.ugt(-MinAbsRHS)) 1095 return *this; 1096 1097 APInt Lower = APIntOps::umax(MinLHS, -MaxAbsRHS + 1); 1098 return ConstantRange(std::move(Lower), APInt(getBitWidth(), 1)); 1099 } 1100 1101 // LHS range crosses zero. 1102 APInt Lower = APIntOps::umax(MinLHS, -MaxAbsRHS + 1); 1103 APInt Upper = APIntOps::umin(MaxLHS, MaxAbsRHS - 1) + 1; 1104 return ConstantRange(std::move(Lower), std::move(Upper)); 1105 } 1106 1107 ConstantRange 1108 ConstantRange::binaryAnd(const ConstantRange &Other) const { 1109 if (isEmptySet() || Other.isEmptySet()) 1110 return getEmpty(); 1111 1112 // TODO: replace this with something less conservative 1113 1114 APInt umin = APIntOps::umin(Other.getUnsignedMax(), getUnsignedMax()); 1115 return getNonEmpty(APInt::getNullValue(getBitWidth()), std::move(umin) + 1); 1116 } 1117 1118 ConstantRange 1119 ConstantRange::binaryOr(const ConstantRange &Other) const { 1120 if (isEmptySet() || Other.isEmptySet()) 1121 return getEmpty(); 1122 1123 // TODO: replace this with something less conservative 1124 1125 APInt umax = APIntOps::umax(getUnsignedMin(), Other.getUnsignedMin()); 1126 return getNonEmpty(std::move(umax), APInt::getNullValue(getBitWidth())); 1127 } 1128 1129 ConstantRange 1130 ConstantRange::shl(const ConstantRange &Other) const { 1131 if (isEmptySet() || Other.isEmptySet()) 1132 return getEmpty(); 1133 1134 APInt max = getUnsignedMax(); 1135 APInt Other_umax = Other.getUnsignedMax(); 1136 1137 // If we are shifting by maximum amount of 1138 // zero return return the original range. 1139 if (Other_umax.isNullValue()) 1140 return *this; 1141 // there's overflow! 1142 if (Other_umax.ugt(max.countLeadingZeros())) 1143 return getFull(); 1144 1145 // FIXME: implement the other tricky cases 1146 1147 APInt min = getUnsignedMin(); 1148 min <<= Other.getUnsignedMin(); 1149 max <<= Other_umax; 1150 1151 return ConstantRange(std::move(min), std::move(max) + 1); 1152 } 1153 1154 ConstantRange 1155 ConstantRange::lshr(const ConstantRange &Other) const { 1156 if (isEmptySet() || Other.isEmptySet()) 1157 return getEmpty(); 1158 1159 APInt max = getUnsignedMax().lshr(Other.getUnsignedMin()) + 1; 1160 APInt min = getUnsignedMin().lshr(Other.getUnsignedMax()); 1161 return getNonEmpty(std::move(min), std::move(max)); 1162 } 1163 1164 ConstantRange 1165 ConstantRange::ashr(const ConstantRange &Other) const { 1166 if (isEmptySet() || Other.isEmptySet()) 1167 return getEmpty(); 1168 1169 // May straddle zero, so handle both positive and negative cases. 1170 // 'PosMax' is the upper bound of the result of the ashr 1171 // operation, when Upper of the LHS of ashr is a non-negative. 1172 // number. Since ashr of a non-negative number will result in a 1173 // smaller number, the Upper value of LHS is shifted right with 1174 // the minimum value of 'Other' instead of the maximum value. 1175 APInt PosMax = getSignedMax().ashr(Other.getUnsignedMin()) + 1; 1176 1177 // 'PosMin' is the lower bound of the result of the ashr 1178 // operation, when Lower of the LHS is a non-negative number. 1179 // Since ashr of a non-negative number will result in a smaller 1180 // number, the Lower value of LHS is shifted right with the 1181 // maximum value of 'Other'. 1182 APInt PosMin = getSignedMin().ashr(Other.getUnsignedMax()); 1183 1184 // 'NegMax' is the upper bound of the result of the ashr 1185 // operation, when Upper of the LHS of ashr is a negative number. 1186 // Since 'ashr' of a negative number will result in a bigger 1187 // number, the Upper value of LHS is shifted right with the 1188 // maximum value of 'Other'. 1189 APInt NegMax = getSignedMax().ashr(Other.getUnsignedMax()) + 1; 1190 1191 // 'NegMin' is the lower bound of the result of the ashr 1192 // operation, when Lower of the LHS of ashr is a negative number. 1193 // Since 'ashr' of a negative number will result in a bigger 1194 // number, the Lower value of LHS is shifted right with the 1195 // minimum value of 'Other'. 1196 APInt NegMin = getSignedMin().ashr(Other.getUnsignedMin()); 1197 1198 APInt max, min; 1199 if (getSignedMin().isNonNegative()) { 1200 // Upper and Lower of LHS are non-negative. 1201 min = PosMin; 1202 max = PosMax; 1203 } else if (getSignedMax().isNegative()) { 1204 // Upper and Lower of LHS are negative. 1205 min = NegMin; 1206 max = NegMax; 1207 } else { 1208 // Upper is non-negative and Lower is negative. 1209 min = NegMin; 1210 max = PosMax; 1211 } 1212 return getNonEmpty(std::move(min), std::move(max)); 1213 } 1214 1215 ConstantRange ConstantRange::uadd_sat(const ConstantRange &Other) const { 1216 if (isEmptySet() || Other.isEmptySet()) 1217 return getEmpty(); 1218 1219 APInt NewL = getUnsignedMin().uadd_sat(Other.getUnsignedMin()); 1220 APInt NewU = getUnsignedMax().uadd_sat(Other.getUnsignedMax()) + 1; 1221 return getNonEmpty(std::move(NewL), std::move(NewU)); 1222 } 1223 1224 ConstantRange ConstantRange::sadd_sat(const ConstantRange &Other) const { 1225 if (isEmptySet() || Other.isEmptySet()) 1226 return getEmpty(); 1227 1228 APInt NewL = getSignedMin().sadd_sat(Other.getSignedMin()); 1229 APInt NewU = getSignedMax().sadd_sat(Other.getSignedMax()) + 1; 1230 return getNonEmpty(std::move(NewL), std::move(NewU)); 1231 } 1232 1233 ConstantRange ConstantRange::usub_sat(const ConstantRange &Other) const { 1234 if (isEmptySet() || Other.isEmptySet()) 1235 return getEmpty(); 1236 1237 APInt NewL = getUnsignedMin().usub_sat(Other.getUnsignedMax()); 1238 APInt NewU = getUnsignedMax().usub_sat(Other.getUnsignedMin()) + 1; 1239 return getNonEmpty(std::move(NewL), std::move(NewU)); 1240 } 1241 1242 ConstantRange ConstantRange::ssub_sat(const ConstantRange &Other) const { 1243 if (isEmptySet() || Other.isEmptySet()) 1244 return getEmpty(); 1245 1246 APInt NewL = getSignedMin().ssub_sat(Other.getSignedMax()); 1247 APInt NewU = getSignedMax().ssub_sat(Other.getSignedMin()) + 1; 1248 return getNonEmpty(std::move(NewL), std::move(NewU)); 1249 } 1250 1251 ConstantRange ConstantRange::inverse() const { 1252 if (isFullSet()) 1253 return getEmpty(); 1254 if (isEmptySet()) 1255 return getFull(); 1256 return ConstantRange(Upper, Lower); 1257 } 1258 1259 ConstantRange ConstantRange::abs() const { 1260 if (isEmptySet()) 1261 return getEmpty(); 1262 1263 if (isSignWrappedSet()) { 1264 APInt Lo; 1265 // Check whether the range crosses zero. 1266 if (Upper.isStrictlyPositive() || !Lower.isStrictlyPositive()) 1267 Lo = APInt::getNullValue(getBitWidth()); 1268 else 1269 Lo = APIntOps::umin(Lower, -Upper + 1); 1270 1271 // SignedMin is included in the result range. 1272 return ConstantRange(Lo, APInt::getSignedMinValue(getBitWidth()) + 1); 1273 } 1274 1275 APInt SMin = getSignedMin(), SMax = getSignedMax(); 1276 1277 // All non-negative. 1278 if (SMin.isNonNegative()) 1279 return *this; 1280 1281 // All negative. 1282 if (SMax.isNegative()) 1283 return ConstantRange(-SMax, -SMin + 1); 1284 1285 // Range crosses zero. 1286 return ConstantRange(APInt::getNullValue(getBitWidth()), 1287 APIntOps::umax(-SMin, SMax) + 1); 1288 } 1289 1290 ConstantRange::OverflowResult ConstantRange::unsignedAddMayOverflow( 1291 const ConstantRange &Other) const { 1292 if (isEmptySet() || Other.isEmptySet()) 1293 return OverflowResult::MayOverflow; 1294 1295 APInt Min = getUnsignedMin(), Max = getUnsignedMax(); 1296 APInt OtherMin = Other.getUnsignedMin(), OtherMax = Other.getUnsignedMax(); 1297 1298 // a u+ b overflows high iff a u> ~b. 1299 if (Min.ugt(~OtherMin)) 1300 return OverflowResult::AlwaysOverflowsHigh; 1301 if (Max.ugt(~OtherMax)) 1302 return OverflowResult::MayOverflow; 1303 return OverflowResult::NeverOverflows; 1304 } 1305 1306 ConstantRange::OverflowResult ConstantRange::signedAddMayOverflow( 1307 const ConstantRange &Other) const { 1308 if (isEmptySet() || Other.isEmptySet()) 1309 return OverflowResult::MayOverflow; 1310 1311 APInt Min = getSignedMin(), Max = getSignedMax(); 1312 APInt OtherMin = Other.getSignedMin(), OtherMax = Other.getSignedMax(); 1313 1314 APInt SignedMin = APInt::getSignedMinValue(getBitWidth()); 1315 APInt SignedMax = APInt::getSignedMaxValue(getBitWidth()); 1316 1317 // a s+ b overflows high iff a s>=0 && b s>= 0 && a s> smax - b. 1318 // a s+ b overflows low iff a s< 0 && b s< 0 && a s< smin - b. 1319 if (Min.isNonNegative() && OtherMin.isNonNegative() && 1320 Min.sgt(SignedMax - OtherMin)) 1321 return OverflowResult::AlwaysOverflowsHigh; 1322 if (Max.isNegative() && OtherMax.isNegative() && 1323 Max.slt(SignedMin - OtherMax)) 1324 return OverflowResult::AlwaysOverflowsLow; 1325 1326 if (Max.isNonNegative() && OtherMax.isNonNegative() && 1327 Max.sgt(SignedMax - OtherMax)) 1328 return OverflowResult::MayOverflow; 1329 if (Min.isNegative() && OtherMin.isNegative() && 1330 Min.slt(SignedMin - OtherMin)) 1331 return OverflowResult::MayOverflow; 1332 1333 return OverflowResult::NeverOverflows; 1334 } 1335 1336 ConstantRange::OverflowResult ConstantRange::unsignedSubMayOverflow( 1337 const ConstantRange &Other) const { 1338 if (isEmptySet() || Other.isEmptySet()) 1339 return OverflowResult::MayOverflow; 1340 1341 APInt Min = getUnsignedMin(), Max = getUnsignedMax(); 1342 APInt OtherMin = Other.getUnsignedMin(), OtherMax = Other.getUnsignedMax(); 1343 1344 // a u- b overflows low iff a u< b. 1345 if (Max.ult(OtherMin)) 1346 return OverflowResult::AlwaysOverflowsLow; 1347 if (Min.ult(OtherMax)) 1348 return OverflowResult::MayOverflow; 1349 return OverflowResult::NeverOverflows; 1350 } 1351 1352 ConstantRange::OverflowResult ConstantRange::signedSubMayOverflow( 1353 const ConstantRange &Other) const { 1354 if (isEmptySet() || Other.isEmptySet()) 1355 return OverflowResult::MayOverflow; 1356 1357 APInt Min = getSignedMin(), Max = getSignedMax(); 1358 APInt OtherMin = Other.getSignedMin(), OtherMax = Other.getSignedMax(); 1359 1360 APInt SignedMin = APInt::getSignedMinValue(getBitWidth()); 1361 APInt SignedMax = APInt::getSignedMaxValue(getBitWidth()); 1362 1363 // a s- b overflows high iff a s>=0 && b s< 0 && a s> smax + b. 1364 // a s- b overflows low iff a s< 0 && b s>= 0 && a s< smin + b. 1365 if (Min.isNonNegative() && OtherMax.isNegative() && 1366 Min.sgt(SignedMax + OtherMax)) 1367 return OverflowResult::AlwaysOverflowsHigh; 1368 if (Max.isNegative() && OtherMin.isNonNegative() && 1369 Max.slt(SignedMin + OtherMin)) 1370 return OverflowResult::AlwaysOverflowsLow; 1371 1372 if (Max.isNonNegative() && OtherMin.isNegative() && 1373 Max.sgt(SignedMax + OtherMin)) 1374 return OverflowResult::MayOverflow; 1375 if (Min.isNegative() && OtherMax.isNonNegative() && 1376 Min.slt(SignedMin + OtherMax)) 1377 return OverflowResult::MayOverflow; 1378 1379 return OverflowResult::NeverOverflows; 1380 } 1381 1382 ConstantRange::OverflowResult ConstantRange::unsignedMulMayOverflow( 1383 const ConstantRange &Other) const { 1384 if (isEmptySet() || Other.isEmptySet()) 1385 return OverflowResult::MayOverflow; 1386 1387 APInt Min = getUnsignedMin(), Max = getUnsignedMax(); 1388 APInt OtherMin = Other.getUnsignedMin(), OtherMax = Other.getUnsignedMax(); 1389 bool Overflow; 1390 1391 (void) Min.umul_ov(OtherMin, Overflow); 1392 if (Overflow) 1393 return OverflowResult::AlwaysOverflowsHigh; 1394 1395 (void) Max.umul_ov(OtherMax, Overflow); 1396 if (Overflow) 1397 return OverflowResult::MayOverflow; 1398 1399 return OverflowResult::NeverOverflows; 1400 } 1401 1402 void ConstantRange::print(raw_ostream &OS) const { 1403 if (isFullSet()) 1404 OS << "full-set"; 1405 else if (isEmptySet()) 1406 OS << "empty-set"; 1407 else 1408 OS << "[" << Lower << "," << Upper << ")"; 1409 } 1410 1411 #if !defined(NDEBUG) || defined(LLVM_ENABLE_DUMP) 1412 LLVM_DUMP_METHOD void ConstantRange::dump() const { 1413 print(dbgs()); 1414 } 1415 #endif 1416 1417 ConstantRange llvm::getConstantRangeFromMetadata(const MDNode &Ranges) { 1418 const unsigned NumRanges = Ranges.getNumOperands() / 2; 1419 assert(NumRanges >= 1 && "Must have at least one range!"); 1420 assert(Ranges.getNumOperands() % 2 == 0 && "Must be a sequence of pairs"); 1421 1422 auto *FirstLow = mdconst::extract<ConstantInt>(Ranges.getOperand(0)); 1423 auto *FirstHigh = mdconst::extract<ConstantInt>(Ranges.getOperand(1)); 1424 1425 ConstantRange CR(FirstLow->getValue(), FirstHigh->getValue()); 1426 1427 for (unsigned i = 1; i < NumRanges; ++i) { 1428 auto *Low = mdconst::extract<ConstantInt>(Ranges.getOperand(2 * i + 0)); 1429 auto *High = mdconst::extract<ConstantInt>(Ranges.getOperand(2 * i + 1)); 1430 1431 // Note: unionWith will potentially create a range that contains values not 1432 // contained in any of the original N ranges. 1433 CR = CR.unionWith(ConstantRange(Low->getValue(), High->getValue())); 1434 } 1435 1436 return CR; 1437 } 1438