[DA] Delinearization of fixed-size multi-dimensional arrays

[bmahjour]

Summary

Currently the dependence analysis in LLVM is unable to compute accurate dependence vectors for multi-dimensional fixed size arrays. For example:

#define N 1024
#define M 2048
void foo(int a[N][M]) {                                                                                                                                                                                                                                                                                                                                                                                                               
  for (long i = 0; i < N-1; ++i)                                                                                                                                                                                      
    for (long j = 2; j < M; ++j)                                                                                                                                                                                      
      a[i][j] = a[i+1][j-2];

gives the following output for the dependence between the load of a[i+1][j-2] and the store into a[i][j]:

da analyze - anti [< >]!

While the direction vectors are correct, no dependence distances are computed, as we expect:

da analyze - anti [1 -2]!

This is mainly because the delinearization algorithm in scalar evolution relies on parametric terms to be present in the access functions. In the case of fixed size arrays such parametric terms are not present, but we can use the indexes from GEP instructions to recover the subscripts for each dimension of the arrays.

It appears that https://reviews.llvm.org/D35430 removed the capability to look at GEP instructions early in 2018. The justification for that change appears to be related to the concern over subscripts overlapping into next array dimensions in languages like C/C++. Please note that https://reviews.llvm.org/D62610 added a debug option to address this concern.

Removing support for fixed-size array delinearization resulted in over-pissimization of DependenceAnalysis and reduction of test coverage for both DA and LoopInterchange.

I've also noticed that polly tries to recover fixed-size array subscripts for dependence testing as well. In this patch I add support for fixed-size array delinearization (using the same mechanism that polly uses) under the option introduced in D62610. Two follow up revisions succeed this patch:

1. In D73995 I've refactored the polly code and moved it to scalar-evolution to make it reuseable by both polly and DA.
2. In D73998 I've renamed the option to reflect the fact that it controls more than just the delinearization validity checks.

Quite a few DA and loop interchange test cases improve as a result of this change, and it makes writing tests for new and existing transformations easier.

[bmahjour]

ping

[bmahjour]

ping

[bmahjour]

Could people on the review list please take a look at this patch? I would also appreciate suggestions for who should be a reviewer instead (or in addition) to the current list of reviewers, if there are any.

[fhahn]

I think it would be good to split this up into the 3 distinct parts you mention in the description:

1. re-introduce the checks
2. rename option
3. improve logic

llvm/lib/Analysis/DependenceAnalysis.cpp

120

What is the rational for renaming the option? Enabling the option is unsafe and the new name makes it sound a lot more harmless than the original one. Also I think we should retain the wording in the description stating why it is unsafe.

llvm/test/Analysis/DependenceAnalysis/Separability.ll

3

Not sure if it's the best idea to enable this option for a bunch of tests, as we miss test coverage for the code path actually enabled by default. I think the runs with -da-assume-inrange-subscripts should be an addition, rather than replacing the existing one.

[bmahjour]

llvm/lib/Analysis/DependenceAnalysis.cpp

120

I renamed it because the option is now being used to control more than just the delinearization validity checks. Now it also controls whether fixed size array delinearization happens or not. I can add the words of caution from the old description back.

llvm/test/Analysis/DependenceAnalysis/Separability.ll

3

Unfortunately that will cause dual maintenance. IMHO the loss of coverage suffered as a result of disabling delinearization for the old tests are more concerning than missing coverage on inaccurate results of the current default. Having said that I see your point in that we may want to make sure the current default results don't get any more inaccurate than they already are.

I really don't see a point in dual maintaining the loop interchange tests though, given that the transform is off by default and there is no point in testing that it fails due to data dependence when there is no real data dependence. Agreed?

[bmahjour]

@fhahn As per your request I've reduced this patch to just adding support for delinearization of fixed-size arrays. Also restored the default tests and added new ones for when the option is enabled. Two more patches are created to 1. do the refactoring (D73995) 2. rename the option (D73998).

[bmahjour]

ping

[Meinersbur]

llvm/include/llvm/Analysis/DependenceAnalysis.h

932–933

Could you add some more details that it tries to derive the dimensions from GEP and tryDelinearizeParametricSize tries to do delinearization? I mean, that's the reason why these functions are separate.

llvm/lib/Analysis/DependenceAnalysis.cpp

3311–3314

I think this has been split into D73995.

3360–3361

Please add a comment why this is bailing out here.

We could also check for GEP's inrange modifier and require it unless DisableDelinearizationChecks is set.

3391–3392

[style] LLVM style does not use almost-always-auto

3470

[style] Don't evaluate size() in every iteration

llvm/test/Analysis/DependenceAnalysis/Separability.ll

3

I second @fhahn's remark. IMHO regression tests for something unsafe does not happen (here: delinearization without proven safety) are even more important.
Regarding maintenance: I too wish that we had something more maintainable than FileCheck.

llvm/test/Transforms/LoopInterchange/currentLimitation.ll

1–3

I think we should continue to also check that interchange does not happen without -da-disable-delinearization-checks. Could you add a separate RUN line with another -check-prefix=? This also applies to other test cases.

[fhahn]

From the description

Removing support for fixed-size array delinearization resulted in over-pissimization of DependenceAnalysis and reduction of test coverage for both DA and LoopInterchange.

I'm not sure it over-pessimized DA. The change was required for correctness.

llvm/test/Analysis/DependenceAnalysis/Separability.ll

3

I think it might also be beneficial to just extract the interesting cases into a separate test file, rather than adding just a few additional lines to the existing tests.

[bmahjour]

llvm/lib/Analysis/DependenceAnalysis.cpp

3360–3361

We could also check for GEP's inrange modifier and require it unless DisableDelinearizationChecks is set.

We could, however I'd suggest we consider that as a separate patch. There are some peculiarities with GEP's inrange that I need to understand better. In particular, it's not clear to me why the syntax allows inrange to appear before multiple indexes, but the getInRangeIndex() API only allows a single index to be retrieved.

If I understand this conversation https://reviews.llvm.org/D22793?id=65626#inline-194586 correctly, then we can only try to delinearize GEPs when
SrcGEP->getInRangeIndex().getValue() == SrcGEP->getNumIndices(). Is that true?

[Meinersbur]

llvm/lib/Analysis/DependenceAnalysis.cpp

3360–3361

We could, however I'd suggest we consider that as a separate patch.

Agreed. Just something to keep in mind when renaming -da-disable-delinearization-checks that might not totally disable non-linear GEPs unless set.

the getInRangeIndex() API only allows a single index to be retrieved.

Yes, and it contradicts the LLVM reference for inrange: https://llvm.org/docs/LangRef.html#id231 . It also seems to be allowed for constant expressions currently.

<result> = getelementptr inbounds <ty>, <ty>* <ptrval>{, [inrange] <ty> <idx>}*
... if the load or store would access memory outside of the bounds of the element selected by the index marked as inrange.

For reference, here is the inline comment from @pcc:

No more than one. The rationale is that inrange on a given index will be at least as restrictive as inrange on any earlier index, so there's no point in allowing more than one. I'll document this in the langref.

This seems not to have been documented. However, there are other motivations than @pcc had in mind (vtables) and some may allow non-inrange indices after an inrange one. Furthermore, I think it is problematic if we'd implicitly interpret all subscripts after an inrange subscript as well.

[bmahjour]

Addressed all review comments.

[Meinersbur]

LGTM

llvm/include/llvm/Analysis/DependenceAnalysis.h

930

Is checkSubscript supposed to be public now?

[bmahjour]

llvm/include/llvm/Analysis/DependenceAnalysis.h

930

No, it is still private. The private keyword above was redundant, which is why I removed it.

[bmahjour]

fix issues found by clang-tidy and clang-format.

[Meinersbur]

llvm/include/llvm/Analysis/DependenceAnalysis.h

930

OK, I see.

llvm/include/llvm/Analysis/DependenceAnalysis.h

const SCEV *addToCoefficient(const SCEV *Expr,
                                  const Loop *TargetLoop,
                                  const SCEV *Value)  const;
 
     /// updateDirection - Update direction vector entry
     /// based on the current constraint.
     void updateDirection(Dependence::DVEntry &Level,
                          const Constraint &CurConstraint) const;
 
+    /// Given a linear access function, tries to recover subscripts
+    /// for each dimension of the array element access.
     bool tryDelinearize(Instruction *Src, Instruction *Dst,
                         SmallVectorImpl<Subscript> &Pair);
 
-  private:
+    /// Tries to delinearize access function for a fixed size multi-dimensional
+    /// array. Returns true upon success and false otherwise.
+    bool tryDelinearizeFixedSize(Instruction *Src, Instruction *Dst,
+                                 const SCEV *SrcAccessFn, const SCEV *DstAccessFn,
+                                 SmallVectorImpl<const SCEV *> &SrcSubscripts,
+                                 SmallVectorImpl<const SCEV *> &DstSubscripts);
+
+    /// Tries to delinearize access function for a multi-dimensional array with
+    /// symbolic runtime sizes.
+    /// Returns true upon success and false otherwise.
+    bool tryDelinearizeParametricSize(Instruction *Src, Instruction *Dst,
+                                      const SCEV *SrcAccessFn, const SCEV *DstAccessFn,
+                                      SmallVectorImpl<const SCEV *> &SrcSubscripts,
+                                      SmallVectorImpl<const SCEV *> &DstSubscripts);
+
     /// checkSubscript - Helper function for checkSrcSubscript and
     /// checkDstSubscript to avoid duplicate code
     bool checkSubscript(const SCEV *Expr, const Loop *LoopNest,
                         SmallBitVector &Loops, bool IsSrc);
   }; // class DependenceInfo
 
   /// AnalysisPass to compute dependence information in a function
   class DependenceAnalysis : public AnalysisInfoMixin<DependenceAnalysis> {

llvm/lib/Analysis/DependenceAnalysis.cpp

Delinearize("da-delinearize", cl::init(true), cl::Hidden, cl::ZeroOrMore,
                 cl::desc("Try to delinearize array references."));
 static cl::opt<bool> DisableDelinearizationChecks(
     "da-disable-delinearization-checks", cl::init(false), cl::Hidden,
     cl::ZeroOrMore,
     cl::desc(
         "Disable checks that try to statically verify validity of "
         "delinearized subscripts. Enabling this option may result in incorrect "
         "dependence vectors for languages that allow the subscript of one "
         "dimension to underflow or overflow into another dimension."));
 
 //===----------------------------------------------------------------------===//
 // basics
 
 DependenceAnalysis::Result
 DependenceAnalysis::run(Function &F, FunctionAnalysisManager &FAM) {
   auto &AA = FAM.getResult<AAManager>(F);
   auto &SE = FAM.getResult<ScalarEvolutionAnalysis>(F);

 /// this function flattens the nested recurrences into separate recurrences
 /// for each loop level.
 bool DependenceInfo::tryDelinearize(Instruction *Src, Instruction *Dst,
                                     SmallVectorImpl<Subscript> &Pair) {
   assert(isLoadOrStore(Src) && "instruction is not load or store");
   assert(isLoadOrStore(Dst) && "instruction is not load or store");
   Value *SrcPtr = getLoadStorePointerOperand(Src);
   Value *DstPtr = getLoadStorePointerOperand(Dst);
-
   Loop *SrcLoop = LI->getLoopFor(Src->getParent());
   Loop *DstLoop = LI->getLoopFor(Dst->getParent());
+  const SCEV *SrcAccessFn = SE->getSCEVAtScope(SrcPtr, SrcLoop);
+  const SCEV *DstAccessFn = SE->getSCEVAtScope(DstPtr, DstLoop);
+  const SCEVUnknown *SrcBase =
+      dyn_cast<SCEVUnknown>(SE->getPointerBase(SrcAccessFn));
+  const SCEVUnknown *DstBase =
+      dyn_cast<SCEVUnknown>(SE->getPointerBase(DstAccessFn));
+
+  if (!SrcBase || !DstBase || SrcBase != DstBase)
+    return false;
+
+  SmallVector<const SCEV *, 4> SrcSubscripts, DstSubscripts;
+
+  if (!tryDelinearizeFixedSize(Src, Dst, SrcAccessFn, DstAccessFn,
+                               SrcSubscripts, DstSubscripts) &&
+      !tryDelinearizeParametricSize(Src, Dst, SrcAccessFn, DstAccessFn,
+                                    SrcSubscripts, DstSubscripts))
+    return false;
+
+  int size = SrcSubscripts.size();
+  LLVM_DEBUG({
+    dbgs() << "\nSrcSubscripts: ";
+    for (int i = 0; i < size; i++)
+      dbgs() << *SrcSubscripts[i];
+    dbgs() << "\nDstSubscripts: ";
+    for (int i = 0; i < size; i++)
+      dbgs() << *DstSubscripts[i];
+  });
+
+  // The delinearization transforms a single-subscript MIV dependence test into
+  // a multi-subscript SIV dependence test that is easier to compute. So we
+  // resize Pair to contain as many pairs of subscripts as the delinearization
+  // has found, and then initialize the pairs following the delinearization.
+  Pair.resize(size);
+  for (int i = 0; i < size; ++i) {
+    Pair[i].Src = SrcSubscripts[i];
+    Pair[i].Dst = DstSubscripts[i];
+    unifySubscriptType(&Pair[i]);
+  }
+
+  return true;
+}
+
+static void
+getIndexExpressionsFromGEP(const GetElementPtrInst *GEP,
+                           SmallVectorImpl<const SCEV *> &Subscripts,
+                           SmallVectorImpl<int> &Sizes, ScalarEvolution &SE) {
+  assert(GEP && "getIndexExpressionsFromGEP called with a null GEP");
+  Type *Ty = GEP->getPointerOperandType();
+  bool DroppedFirstDim = false;
+  for (unsigned i = 1; i < GEP->getNumOperands(); i++) {
+    const SCEV *Expr = SE.getSCEV(GEP->getOperand(i));
+    if (i == 1) {
+      if (auto *PtrTy = dyn_cast<PointerType>(Ty)) {
+        Ty = PtrTy->getElementType();
+      } else if (auto *ArrayTy = dyn_cast<ArrayType>(Ty)) {
+        Ty = ArrayTy->getElementType();
+      } else {
+        Subscripts.clear();
+        Sizes.clear();
+        break;
+      }
+      if (auto *Const = dyn_cast<SCEVConstant>(Expr))
+        if (Const->getValue()->isZero()) {
+          DroppedFirstDim = true;
+          continue;
+        }
+      Subscripts.push_back(Expr);
+      continue;
+    }
+
+    auto *ArrayTy = dyn_cast<ArrayType>(Ty);
+    if (!ArrayTy) {
+      Subscripts.clear();
+      Sizes.clear();
+      break;
+    }
+
+    Subscripts.push_back(Expr);
+    if (!(DroppedFirstDim && i == 2))
+      Sizes.push_back(ArrayTy->getNumElements());
+
+    Ty = ArrayTy->getElementType();
+  }
+  return;
+}
+
+bool DependenceInfo::tryDelinearizeFixedSize(
+    Instruction *Src, Instruction *Dst, const SCEV *SrcAccessFn,
+    const SCEV *DstAccessFn, SmallVectorImpl<const SCEV *> &SrcSubscripts,
+    SmallVectorImpl<const SCEV *> &DstSubscripts) {
 
-  // Below code mimics the code in Delinearization.cpp
-  const SCEV *SrcAccessFn =
-    SE->getSCEVAtScope(SrcPtr, SrcLoop);
-  const SCEV *DstAccessFn =
-    SE->getSCEVAtScope(DstPtr, DstLoop);
+  if (!DisableDelinearizationChecks)
+    return false;
 
+  Value *SrcPtr = getLoadStorePointerOperand(Src);
+  Value *DstPtr = getLoadStorePointerOperand(Dst);
   const SCEVUnknown *SrcBase =
       dyn_cast<SCEVUnknown>(SE->getPointerBase(SrcAccessFn));
   const SCEVUnknown *DstBase =
       dyn_cast<SCEVUnknown>(SE->getPointerBase(DstAccessFn));
+  assert(SrcBase && DstBase && SrcBase == DstBase &&
+         "expected src and dst scev unknowns to be equal");
 
-  if (!SrcBase || !DstBase || SrcBase != DstBase)
-    return false;
+  // Check the simple case where the array dimensions are fixed size.
+  auto *SrcGEP = dyn_cast<GetElementPtrInst>(SrcPtr);
+  auto *DstGEP = dyn_cast<GetElementPtrInst>(DstPtr);
+  if (!SrcGEP || !DstGEP)
+    return false;
+
+  SmallVector<int, 4> SrcSizes, DstSizes;
+  getIndexExpressionsFromGEP(SrcGEP, SrcSubscripts, SrcSizes, *SE);
+  getIndexExpressionsFromGEP(DstGEP, DstSubscripts, DstSizes, *SE);
+
+  // Check that the two size arrays are non-empty and equal in length and
+  // value.
+  if (SrcSizes.empty() || SrcSubscripts.size() <= 1 || SrcSizes.size() != DstSizes.size() ||
+      !std::equal(SrcSizes.begin(), SrcSizes.end(), DstSizes.begin())) {
+    SrcSubscripts.clear();
+    DstSubscripts.clear();
+    return false;
+  }
+
+  auto *SrcBasePtr = SrcGEP->getOperand(0);
+  auto *DstBasePtr = DstGEP->getOperand(0);
+  while (auto *PCast = dyn_cast<BitCastInst>(SrcBasePtr))
+    SrcBasePtr = PCast->getOperand(0);
+  while (auto *PCast = dyn_cast<BitCastInst>(DstBasePtr))
+    DstBasePtr = PCast->getOperand(0);
+
+  // Check that for identical base pointers we do not miss index offsets
+  // that have been added before this GEP is applied.
+  if (SrcBasePtr == SrcBase->getValue() &&
+      DstBasePtr == DstBase->getValue()) {
+    assert(SrcSubscripts.size() == DstSubscripts.size() &&
+           SrcSubscripts.size() == SrcSizes.size() + 1 &&
+           "Expected equal number of entries in the list of sizes and "
+           "subscripts.");
+    LLVM_DEBUG({
+      dbgs() << "Delinearized subscripts of fixed-size array\n"
+             << "SrcGEP:" << *SrcGEP << "\n"
+             << "DstGEP:" << *DstGEP << "\n";
+    });
+    return true;
+  }
+
+  SrcSubscripts.clear();
+  DstSubscripts.clear();
+}
+
+bool DependenceInfo::tryDelinearizeParametricSize(
+    Instruction *Src, Instruction *Dst, const SCEV *SrcAccessFn,
+    const SCEV *DstAccessFn, SmallVectorImpl<const SCEV *> &SrcSubscripts,
+    SmallVectorImpl<const SCEV *> &DstSubscripts) {
+
+  Value *SrcPtr = getLoadStorePointerOperand(Src);
+  Value *DstPtr = getLoadStorePointerOperand(Dst);
+  const SCEVUnknown *SrcBase =
+      dyn_cast<SCEVUnknown>(SE->getPointerBase(SrcAccessFn));
+  const SCEVUnknown *DstBase =
+      dyn_cast<SCEVUnknown>(SE->getPointerBase(DstAccessFn));
+  assert(SrcBase && DstBase && SrcBase == DstBase &&
+         "expected src and dst scev unknowns to be equal");
 
   const SCEV *ElementSize = SE->getElementSize(Src);
   if (ElementSize != SE->getElementSize(Dst))
     return false;
 
   const SCEV *SrcSCEV = SE->getMinusSCEV(SrcAccessFn, SrcBase);
   const SCEV *DstSCEV = SE->getMinusSCEV(DstAccessFn, DstBase);
 
   const SCEVAddRecExpr *SrcAR = dyn_cast<SCEVAddRecExpr>(SrcSCEV);
   const SCEVAddRecExpr *DstAR = dyn_cast<SCEVAddRecExpr>(DstSCEV);
   if (!SrcAR || !DstAR || !SrcAR->isAffine() || !DstAR->isAffine())
     return false;
 
   // First step: collect parametric terms in both array references.
   SmallVector<const SCEV *, 4> Terms;
   SE->collectParametricTerms(SrcAR, Terms);
   SE->collectParametricTerms(DstAR, Terms);
 
   // Second step: find subscript sizes.
   SmallVector<const SCEV *, 4> Sizes;
   SE->findArrayDimensions(Terms, Sizes, ElementSize);
 
   // Third step: compute the access functions for each subscript.
-  SmallVector<const SCEV *, 4> SrcSubscripts, DstSubscripts;
   SE->computeAccessFunctions(SrcAR, SrcSubscripts, Sizes);
   SE->computeAccessFunctions(DstAR, DstSubscripts, Sizes);
 
   // Fail when there is only a subscript: that's a linearized access function.
   if (SrcSubscripts.size() < 2 || DstSubscripts.size() < 2 ||
       SrcSubscripts.size() != DstSubscripts.size())
     return false;
 
-  int size = SrcSubscripts.size();
-
   // Statically check that the array bounds are in-range. The first subscript we
   // don't have a size for and it cannot overflow into another subscript, so is
   // always safe. The others need to be 0 <= subscript[i] < bound, for both src
   // and dst.
   // FIXME: It may be better to record these sizes and add them as constraints
   // to the dependency checks.
   if (!DisableDelinearizationChecks)
-    for (int i = 1; i < size; ++i) {
+    for (size_t i = 1; i < SrcSubscripts.size(); ++i) {
       if (!isKnownNonNegative(SrcSubscripts[i], SrcPtr))
         return false;
 
       if (!isKnownLessThan(SrcSubscripts[i], Sizes[i - 1]))
         return false;
 
       if (!isKnownNonNegative(DstSubscripts[i], DstPtr))
         return false;
 
       if (!isKnownLessThan(DstSubscripts[i], Sizes[i - 1]))
         return false;
     }
 
-  LLVM_DEBUG({
-    dbgs() << "\nSrcSubscripts: ";
-    for (int i = 0; i < size; i++)
-      dbgs() << *SrcSubscripts[i];
-    dbgs() << "\nDstSubscripts: ";
-    for (int i = 0; i < size; i++)
-      dbgs() << *DstSubscripts[i];
-  });
-
-  // The delinearization transforms a single-subscript MIV dependence test into
-  // a multi-subscript SIV dependence test that is easier to compute. So we
-  // resize Pair to contain as many pairs of subscripts as the delinearization
-  // has found, and then initialize the pairs following the delinearization.
-  Pair.resize(size);
-  for (int i = 0; i < size; ++i) {
-    Pair[i].Src = SrcSubscripts[i];
-    Pair[i].Dst = DstSubscripts[i];
-    unifySubscriptType(&Pair[i]);
-  }
-
   return true;
 }
 
 //===----------------------------------------------------------------------===//
 
 #ifndef NDEBUG
 // For debugging purposes, dump a small bit vector to dbgs().
 static void dumpSmallBitVector(SmallBitVector &BV) {

llvm/test/Analysis/DependenceAnalysis/Invariant.ll

 ; RUN: opt < %s -disable-output "-passes=print<da>" -aa-pipeline=basic-aa 2>&1 \
 ; RUN: | FileCheck %s
 ; RUN: opt < %s -analyze -basicaa -da | FileCheck %s
 
+; RUN: opt < %s -disable-output "-passes=print<da>" -aa-pipeline=basic-aa 2>&1 \
+; RUN: -da-disable-delinearization-checks | FileCheck %s --check-prefix=IN-RANGE
+
 ; Test for a bug, which caused an assert when an invalid
 ; SCEVAddRecExpr is created in addToCoefficient.
 
 ; CHECK-LABEL: foo
 ; CHECK: da analyze - consistent input [S 0]!
 ; CHECK: da analyze - input [* *|<]!
 ; CHECK: da analyze - none!
 
+; IN-RANGE-LABEL: foo
+; IN-RANGE: da analyze - consistent input [S 0]!
+; IN-RANGE: da analyze - input [* 0|<]!
+; IN-RANGE: da analyze - none!
+
 define float @foo(float %g, [40 x float]* %rr) nounwind {
 entry:
   br label %for.cond1.preheader
 
 for.cond1.preheader:
   %i.04 = phi i32 [ 0, %entry ], [ %add10, %for.inc9 ]
   %res.03 = phi float [ 0.000000e+00, %entry ], [ %add.res.1, %for.inc9 ]
   br label %for.body3

llvm/test/Analysis/DependenceAnalysis/NonCanonicalizedSubscript.ll

 ; RUN: opt < %s -disable-output "-passes=print<da>" -aa-pipeline=basic-aa 2>&1 \
 ; RUN: | FileCheck %s -check-prefix=DELIN
 ; RUN: opt < %s -analyze -basicaa -da | FileCheck %s -check-prefix=DELIN
 
+; RUN: opt < %s -disable-output "-passes=print<da>" -aa-pipeline=basic-aa 2>&1 \
+; RUN: -da-disable-delinearization-checks | FileCheck %s -check-prefix=IN-RANGE
+
 target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128-n8:16:32:64-S128"
 target triple = "x86_64-apple-macosx10.6.0"
 
 ; for (int i = 0; i < 100; ++i) {
 ;   int t0 = a[i][i];
 ;   int t1 = t0 + 1;
 ;   a[i][5] = t1;
 ; }

 @a = global [10004 x [10004 x i32]] zeroinitializer, align 16
 
 ; Function Attrs: nounwind uwtable
 define void @coupled_miv_type_mismatch(i32 %n) #0 {
 ; DELIN-LABEL: 'Dependence Analysis' for function 'coupled_miv_type_mismatch'
 entry:
   br label %for.cond
 
+; IN-RANGE: da analyze - input [0 *]!
+; IN-RANGE: da analyze - anti [1 *]!
+; IN-RANGE: da analyze - none!
+
 ; DELIN: da analyze - input [* *]!
 ; DELIN: da analyze - anti [* *|<]!
 ; DELIN: da analyze - none!
 for.cond:                                         ; preds = %for.inc11, %entry
   %indvars.iv11 = phi i64 [ %indvars.iv.next12, %for.inc11 ], [ 1, %entry ]
   %exitcond14 = icmp ne i64 %indvars.iv11, 10000
   br i1 %exitcond14, label %for.cond1.preheader, label %for.end13
 

llvm/test/Analysis/DependenceAnalysis/Preliminary.ll

 ; RUN: opt < %s -disable-output "-passes=print<da>" -aa-pipeline=basic-aa 2>&1 \
 ; RUN: | FileCheck %s
 ; RUN: opt < %s -analyze -basicaa -da | FileCheck %s
 
+; RUN: opt < %s -disable-output "-passes=print<da>" -aa-pipeline=basic-aa 2>&1 \
+; RUN: -da-disable-delinearization-checks | FileCheck %s --check-prefix=IN-RANGE
+
 target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128-n8:16:32:64-S128"
 target triple = "x86_64-apple-macosx10.6.0"
 
 
 ;;int p0(int n, int *A, int *B) {
 ;;  A[0] = n;
 ;;  return B[1];
 

 ;;      for (long int k = 0; k < n; k++) {
 ;;        *B++ = A[i + 3][j + 2][k + 1];
 
 define void @p2(i64 %n, [100 x [100 x i64]]* %A, i64* %B) nounwind uwtable ssp {
 entry:
   %cmp10 = icmp sgt i64 %n, 0
   br i1 %cmp10, label %for.cond1.preheader.preheader, label %for.end26
 
+; IN-RANGE-LABEL: p2
+; IN-RANGE: da analyze - none!
+; IN-RANGE: da analyze - flow [-3 -2]!
+; IN-RANGE: da analyze - confused!
+; IN-RANGE: da analyze - none!
+; IN-RANGE: da analyze - confused!
+; IN-RANGE: da analyze - output [* * *]!
+
 ; CHECK-LABEL: p2
 ; CHECK: da analyze - output [* * *]!
 ; CHECK: da analyze - flow [* *|<]!
 ; CHECK: da analyze - confused!
 ; CHECK: da analyze - input [* * *]!
 ; CHECK: da analyze - confused!
 ; CHECK: da analyze - output [* * *]!
-
 for.cond1.preheader.preheader:                    ; preds = %entry
   br label %for.cond1.preheader
 
 for.cond1.preheader:                              ; preds = %for.cond1.preheader.preheader, %for.inc24
   %B.addr.012 = phi i64* [ %B.addr.1.lcssa, %for.inc24 ], [ %B, %for.cond1.preheader.preheader ]
   %i.011 = phi i64 [ %inc25, %for.inc24 ], [ 0, %for.cond1.preheader.preheader ]
   %cmp26 = icmp sgt i64 %n, 0
   br i1 %cmp26, label %for.cond4.preheader.preheader, label %for.inc24

llvm/test/Analysis/DependenceAnalysis/Propagating.ll

 ; RUN: opt < %s -disable-output "-passes=print<da>" -aa-pipeline=basic-aa 2>&1 \
 ; RUN: | FileCheck %s
 ; RUN: opt < %s -analyze -basicaa -da | FileCheck %s
 
+; RUN: opt < %s -disable-output "-passes=print<da>" -aa-pipeline=basic-aa 2>&1 \
+; RUN: -da-disable-delinearization-checks | FileCheck %s --check-prefix=IN-RANGE
+
 target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128-n8:16:32:64-S128"
 target triple = "x86_64-apple-macosx10.6.0"
 
 
 ;;  for (long int i = 0; i < 100; i++)
 ;;    for (long int j = 0; j < 100; j++) {
 ;;      A[i + 1][i + j] = i;
 ;;      *B++ = A[i][i + j];
 
 define void @prop0([100 x i32]* %A, i32* %B, i32 %n) nounwind uwtable ssp {
 entry:
   br label %for.cond1.preheader
 
 ; CHECK-LABEL: prop0
 ; CHECK: da analyze - none!
 ; CHECK: da analyze - flow [< >]!
 ; CHECK: da analyze - confused!
 ; CHECK: da analyze - none!
 ; CHECK: da analyze - confused!
 ; CHECK: da analyze - none!
 
+; IN-RANGE-LABEL: prop0
+; IN-RANGE: da analyze - none!
+; IN-RANGE: da analyze - consistent flow [1 -1]!
+; IN-RANGE: da analyze - confused!
+; IN-RANGE: da analyze - none!
+; IN-RANGE: da analyze - confused!
+; IN-RANGE: da analyze - none!
+
 for.cond1.preheader:                              ; preds = %entry, %for.inc9
   %B.addr.04 = phi i32* [ %B, %entry ], [ %scevgep, %for.inc9 ]
   %i.03 = phi i64 [ 0, %entry ], [ %inc10, %for.inc9 ]
   br label %for.body3
 
 for.body3:                                        ; preds = %for.cond1.preheader, %for.body3
   %j.02 = phi i64 [ 0, %for.cond1.preheader ], [ %inc, %for.body3 ]
   %B.addr.11 = phi i32* [ %B.addr.04, %for.cond1.preheader ], [ %incdec.ptr, %for.body3 ]

 ;;      for (long int k = 0; k < 100; k++) {
 ;;        A[j - i][i + 1][j + k] = i;
 ;;        *B++ = A[j - i][i][j + k];
 
 define void @prop1([100 x [100 x i32]]* %A, i32* %B, i32 %n) nounwind uwtable ssp {
 entry:
   br label %for.cond1.preheader
 
+; IN-RANGE-LABEL: prop1
+; IN-RANGE: da analyze - none!
+; IN-RANGE: da analyze - consistent flow [1 1 -1]!
+; IN-RANGE: da analyze - confused!
+; IN-RANGE: da analyze - none!
+; IN-RANGE: da analyze - confused!
+; IN-RANGE: da analyze - none!
+
 ; CHECK-LABEL: prop1
 ; CHECK: da analyze - output [* * *]!
 ; CHECK: da analyze - flow [<> <> *]!
 ; CHECK: da analyze - confused!
 ; CHECK: da analyze - input [* * *]!
 ; CHECK: da analyze - confused!
 ; CHECK: da analyze - none!
 

 ;;    for (long int j = 0; j < 100; j++) {
 ;;      A[i - 1][2*i] = i;
 ;;      *B++ = A[i][i + j + 110];
 
 define void @prop2([100 x i32]* %A, i32* %B, i32 %n) nounwind uwtable ssp {
 entry:
   br label %for.cond1.preheader
 
+; IN-RANGE-LABEL: prop2
+; IN-RANGE: da analyze - consistent output [0 S]!
+; IN-RANGE: da analyze - none!
+; IN-RANGE: da analyze - confused!
+; IN-RANGE: da analyze - none!
+; IN-RANGE: da analyze - confused!
+; IN-RANGE: da analyze - none!
+
 ; CHECK-LABEL: prop2
 ; CHECK: da analyze - consistent output [0 S]!
 ; CHECK: da analyze - flow [> *]!
 ; CHECK: da analyze - confused!
 ; CHECK: da analyze - none!
 ; CHECK: da analyze - confused!
 ; CHECK: da analyze - none!
 

 ;;    for (long int j = 0; j < 100; j++) {
 ;;      A[i][2*j + i] = i;
 ;;      *B++ = A[i][2*j - i + 5];
 
 define void @prop3([100 x i32]* %A, i32* %B, i32 %n) nounwind uwtable ssp {
 entry:
   br label %for.cond1.preheader
 
+; IN-RANGE-LABEL: prop3
+; IN-RANGE: da analyze - none!
+; IN-RANGE: da analyze - none!
+; IN-RANGE: da analyze - confused!
+; IN-RANGE: da analyze - none!
+; IN-RANGE: da analyze - confused!
+; IN-RANGE: da analyze - none!
+
 ; CHECK-LABEL: prop3
 ; CHECK: da analyze - output [* *]!
 ; CHECK: da analyze - flow [<> *]!
 ; CHECK: da analyze - confused!
 ; CHECK: da analyze - input [* *]!
 ; CHECK: da analyze - confused!
 ; CHECK: da analyze - none!
 

 ;;    for (long int j = 0; j < 100; j++) {
 ;;      A[i + 2][2*i + j + 1] = i;
 ;;      *B++ = A[i][2*i + j];
 
 define void @prop4([100 x i32]* %A, i32* %B, i32 %n) nounwind uwtable ssp {
 entry:
   br label %for.cond1.preheader
 
+; IN-RANGE-LABEL: prop4
+; IN-RANGE: da analyze - none!
+; IN-RANGE: da analyze - consistent flow [2 -3]!
+; IN-RANGE: da analyze - confused!
+; IN-RANGE: da analyze - none!
+; IN-RANGE: da analyze - confused!
+; IN-RANGE: da analyze - none!
+
 ; CHECK-LABEL: prop4
 ; CHECK: da analyze - none!
 ; CHECK: da analyze - flow [< <>]!
 ; CHECK: da analyze - confused!
 ; CHECK: da analyze - none!
 ; CHECK: da analyze - confused!
 ; CHECK: da analyze - none!
 

 ;;    for (long int j = 0; j < 100; j++) {
 ;;      A[3*i - 18][22 - i][2*i + j] = i;
 ;;      *B++ = A[i][i][3*i + j];
 
 define void @prop5([100 x [100 x i32]]* %A, i32* %B, i32 %n) nounwind uwtable ssp {
 entry:
   br label %for.cond1.preheader
 
+; IN-RANGE-LABEL: prop5
+; IN-RANGE: da analyze - none!
+; IN-RANGE: da analyze - flow [< -16] splitable!
+; IN-RANGE: da analyze - confused!
+; IN-RANGE: da analyze - none!
+; IN-RANGE: da analyze - confused!
+; IN-RANGE: da analyze - none!
+
 ; CHECK-LABEL: prop5
 ; CHECK: da analyze - none!
 ; CHECK: da analyze - flow [* *|<]!
 ; CHECK: da analyze - confused!
 ; CHECK: da analyze - none!
 ; CHECK: da analyze - confused!
 ; CHECK: da analyze - none!
 

 ;;    for (long int j = 0; j < 100; j++) {
 ;;      A[i + 1][4*i + j + 2] = i;
 ;;      *B++ = A[2*i][8*i + j];
 
 define void @prop6([100 x i32]* %A, i32* %B, i32 %n) nounwind uwtable ssp {
 entry:
   br label %for.cond1.preheader
 
+; IN-RANGE-LABEL: prop6
+; IN-RANGE: da analyze - none!
+; IN-RANGE: da analyze - flow [=> -2]!
+; IN-RANGE: da analyze - confused!
+; IN-RANGE: da analyze - none!
+; IN-RANGE: da analyze - confused!
+; IN-RANGE: da analyze - none!
+
 ; CHECK-LABEL: prop6
 ; CHECK: da analyze - none!
 ; CHECK: da analyze - flow [=> <>]!
 ; CHECK: da analyze - confused!
 ; CHECK: da analyze - none!
 ; CHECK: da analyze - confused!
 ; CHECK: da analyze - none!
 

 ;;    for (long int j = 0; j < 100; j++) {
 ;;      A[2*i + 4][-5*i + j + 2] = i;
 ;;      *B++ = A[-2*i + 20][5*i + j];
 
 define void @prop7([100 x i32]* %A, i32* %B, i32 %n) nounwind uwtable ssp {
 entry:
   br label %for.cond1.preheader
 
+; IN-RANGE-LABEL: prop7
+; IN-RANGE: da analyze - none!
+; IN-RANGE: da analyze - flow [* -38] splitable!
+; IN-RANGE: da analyze - split level = 1, iteration = 4!
+; IN-RANGE: da analyze - confused!
+; IN-RANGE: da analyze - none!
+; IN-RANGE: da analyze - confused!
+; IN-RANGE: da analyze - none!
+
 ; CHECK-LABEL: prop7
 ; CHECK: da analyze - none!
 ; CHECK: da analyze - flow [* <>]!
 ; CHECK: da analyze - confused!
 ; CHECK: da analyze - none!
 ; CHECK: da analyze - confused!
 ; CHECK: da analyze - none!
 

 ;;    for (long int j = 0; j < 100; j++) {
 ;;      A[4][j + 2] = i;
 ;;      *B++ = A[-2*i + 4][5*i + j];
 
 define void @prop8([100 x i32]* %A, i32* %B, i32 %n) nounwind uwtable ssp {
 entry:
   br label %for.cond1.preheader
 
+; IN-RANGE-LABEL: prop8
+; IN-RANGE: da analyze - consistent output [S 0]!
+; IN-RANGE: da analyze - flow [p=> 2]!
+; IN-RANGE: da analyze - confused!
+; IN-RANGE: da analyze - none!
+; IN-RANGE: da analyze - confused!
+; IN-RANGE: da analyze - none!
+
 ; CHECK-LABEL: prop8
 ; CHECK: da analyze - consistent output [S 0]!
 ; CHECK: da analyze - flow [=> <]!
 ; CHECK: da analyze - confused!
 ; CHECK: da analyze - none!
 ; CHECK: da analyze - confused!
 ; CHECK: da analyze - none!
 

 ;;    for (long int j = 0; j < 100; j++) {
 ;;      A[2*i + 4][5*i + j + 2] = i;
 ;;      *B++ = A[4][j];
 
 define void @prop9([100 x i32]* %A, i32* %B, i32 %n) nounwind uwtable ssp {
 entry:
   br label %for.cond1.preheader
 
+; IN-RANGE-LABEL: prop9
+; IN-RANGE: da analyze - none!
+; IN-RANGE: da analyze - flow [p<= 2]!
+; IN-RANGE: da analyze - confused!
+; IN-RANGE: da analyze - consistent input [S 0]!
+; IN-RANGE: da analyze - confused!
+; IN-RANGE: da analyze - none!
+
 ; CHECK-LABEL: prop9
 ; CHECK: da analyze - none!
 ; CHECK: da analyze - flow [<= <]!
 ; CHECK: da analyze - confused!
 ; CHECK: da analyze - consistent input [S 0]!
 ; CHECK: da analyze - confused!
 ; CHECK: da analyze - none!
 

llvm/test/Analysis/DependenceAnalysis/Separability.ll

 ; RUN: opt < %s -disable-output "-passes=print<da>" -aa-pipeline=basic-aa 2>&1 \
 ; RUN: | FileCheck %s
 ; RUN: opt < %s -analyze -basicaa -da | FileCheck %s
 
+; RUN: opt < %s -disable-output "-passes=print<da>" -aa-pipeline=basic-aa 2>&1 \
+; RUN: -da-disable-delinearization-checks | FileCheck %s --check-prefix=IN-RANGE
+
 target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128-n8:16:32:64-S128"
 target triple = "x86_64-apple-macosx10.6.0"
 
 
 ;;  for (long int i = 0; i < 50; i++)
 ;;    for (long int j = 0; j < 50; j++)
 ;;      for (long int k = 0; k < 50; k++)
 ;;        for (long int l = 0; l < 50; l++) {
 ;;          A[n][i][j + k] = i;
 ;;          *B++ = A[10][i + 10][2*j - l];
 
 define void @sep0([100 x [100 x i32]]* %A, i32* %B, i32 %n) nounwind uwtable ssp {
 entry:
   br label %for.cond1.preheader
 
+; IN-RANGE: da analyze - output [0 * * S]!
+; IN-RANGE: da analyze - flow [-10 * * *]!
+; IN-RANGE: da analyze - confused!
+; IN-RANGE: da analyze - input [0 * S *]!
+; IN-RANGE: da analyze - confused!
+; IN-RANGE: da analyze - none!
+
 ; CHECK: da analyze - output [= * * S]!
 ; CHECK: da analyze - flow [* * * *|<]!
 ; CHECK: da analyze - confused!
 ; CHECK: da analyze - input [* * S *]!
 ; CHECK: da analyze - confused!
 ; CHECK: da analyze - none!
 
 for.cond1.preheader:                              ; preds = %entry, %for.inc22

 ;;        for (long int l = 0; l < 50; l++) {
 ;;          A[i][i][j + k] = i;
 ;;          *B++ = A[10][i + 10][2*j - l];
 
 define void @sep1([100 x [100 x i32]]* %A, i32* %B, i32 %n) nounwind uwtable ssp {
 entry:
   br label %for.cond1.preheader
 
+; IN-RANGE: da analyze - output [0 * * S]!
+; IN-RANGE: da analyze - flow [> * * *]!
+; IN-RANGE: da analyze - confused!
+; IN-RANGE: da analyze - input [0 * S *]!
+; IN-RANGE: da analyze - confused!
+; IN-RANGE: da analyze - none!
+
 ; CHECK: da analyze - output [= * * S]!
 ; CHECK: da analyze - flow [* * * *|<]!
 ; CHECK: da analyze - confused!
 ; CHECK: da analyze - input [* * S *]!
 ; CHECK: da analyze - confused!
 ; CHECK: da analyze - none!
 
 for.cond1.preheader:                              ; preds = %entry, %for.inc22

 ;;        for (long int l = 0; l < 50; l++) {
 ;;          A[i][i][i + k][l] = i;
 ;;          *B++ = A[10][i + 10][j + k][l + 10];
 
 define void @sep2([100 x [100 x [100 x i32]]]* %A, i32* %B, i32 %n) nounwind uwtable ssp {
 entry:
   br label %for.cond1.preheader
 
+; IN-RANGE: da analyze - consistent output [0 S 0 0]!
+; IN-RANGE: da analyze - flow [> * * -10]!
+; IN-RANGE: da analyze - confused!
+; IN-RANGE: da analyze - input [0 * * 0]!
+; IN-RANGE: da analyze - confused!
+; IN-RANGE: da analyze - none!
+
 ; CHECK: da analyze - output [= S = =]!
 ; CHECK: da analyze - flow [* * * <>]!
 ; CHECK: da analyze - confused!
 ; CHECK: da analyze - input [= * * *]!
 ; CHECK: da analyze - confused!
 ; CHECK: da analyze - none!
 
 for.cond1.preheader:                              ; preds = %entry, %for.inc26

 ;;        for (long int l = 0; l < 50; l++) {
 ;;          A[i][i][i + k][l + k] = i;
 ;;          *B++ = A[10][i + 10][j + k][l + 10];
 
 define void @sep3([100 x [100 x [100 x i32]]]* %A, i32* %B, i32 %n) nounwind uwtable ssp {
 entry:
   br label %for.cond1.preheader
 
+; IN-RANGE: da analyze - consistent output [0 S 0 0]!
+; IN-RANGE: da analyze - flow [> * * *]!
+; IN-RANGE: da analyze - confused!
+; IN-RANGE: da analyze - input [0 * * 0]!
+; IN-RANGE: da analyze - confused!
+; IN-RANGE: da analyze - none!
+
 ; CHECK: da analyze - output [= S = =]!
 ; CHECK: da analyze - flow [* * * *|<]!
 ; CHECK: da analyze - confused!
 ; CHECK: da analyze - input [= * * *]!
 ; CHECK: da analyze - confused!
 ; CHECK: da analyze - none!
 
 for.cond1.preheader:                              ; preds = %entry, %for.inc27

llvm/test/Analysis/DependenceAnalysis/SimpleSIVNoValidityCheckFixedSize.ll

+; RUN: opt < %s -disable-output -passes="print<da>"                            \
+; RUN: -da-disable-delinearization-checks 2>&1 | FileCheck %s
+; RUN: opt < %s -da -analyze -da-disable-delinearization-checks | FileCheck %s
+
+; CHECK-LABEL: t1
+; CHECK: da analyze - none!
+; CHECK: da analyze - consistent anti [1 -2]!
+; CHECK: da analyze - none!
+
+;; #define N 1024
+;; #define M 2048
+;; void t1(int a[N][M]) {
+;;   for (int i = 0; i < N-1; ++i)
+;;     for (int j = 2; j < M; ++j)
+;;       a[i][j] = a[i+1][j-2];
+;; }
+
+define void @t1([2048 x i32]* %a) {
+entry:
+  br label %for.body
+
+for.body:                                         ; preds = %entry, %for.inc11
+  %indvars.iv4 = phi i64 [ 0, %entry ], [ %indvars.iv.next5, %for.inc11 ]
+  br label %for.body4
+
+for.body4:                                        ; preds = %for.body, %for.body4
+  %indvars.iv = phi i64 [ 2, %for.body ], [ %indvars.iv.next, %for.body4 ]
+  %0 = add nuw nsw i64 %indvars.iv4, 1
+  %1 = add nsw i64 %indvars.iv, -2
+  %arrayidx6 = getelementptr inbounds [2048 x i32], [2048 x i32]* %a, i64 %0, i64 %1
+  %2 = load i32, i32* %arrayidx6, align 4
+  %arrayidx10 = getelementptr inbounds [2048 x i32], [2048 x i32]* %a, i64 %indvars.iv4, i64 %indvars.iv
+  store i32 %2, i32* %arrayidx10, align 4
+  %indvars.iv.next = add nuw nsw i64 %indvars.iv, 1
+  %exitcond = icmp ne i64 %indvars.iv.next, 2048
+  br i1 %exitcond, label %for.body4, label %for.inc11
+
+for.inc11:                                        ; preds = %for.body4
+  %indvars.iv.next5 = add nuw nsw i64 %indvars.iv4, 1
+  %exitcond7 = icmp ne i64 %indvars.iv.next5, 1023
+  br i1 %exitcond7, label %for.body, label %for.end13
+
+for.end13:                                        ; preds = %for.inc11
+  ret void
+}
+
+
+; CHECK-LABEL: t2
+; CHECK: da analyze - none!
+; CHECK: da analyze - consistent anti [1 -2 0 -3 2]!
+; CHECK: da analyze - none!
+
+;; #define N 1024
+;; #define M 2048
+;; void t2(int a[][N][N][N][M]) {
+;;   for (int i1 = 0; i1 < N-1; ++i1)
+;;     for (int i2 = 2; i2 < N; ++i2)
+;;       for (int i3 = 0; i3 < N; ++i3)
+;;         for (int i4 = 3; i4 < N; ++i4)
+;;           for (int i5 = 0; i5 < M-2; ++i5)
+;;             a[i1][i2][i3][i4][i5] = a[i1+1][i2-2][i3][i4-3][i5+2];
+;; }
+
+define void @t2([1024 x [1024 x [1024 x [2048 x i32]]]]* %a) {
+entry:
+  br label %for.body
+
+for.body:                                         ; preds = %entry, %for.inc46
+  %indvars.iv18 = phi i64 [ 0, %entry ], [ %indvars.iv.next19, %for.inc46 ]
+  br label %for.body4
+
+for.body4:                                        ; preds = %for.body, %for.inc43
+  %indvars.iv14 = phi i64 [ 2, %for.body ], [ %indvars.iv.next15, %for.inc43 ]
+  br label %for.body8
+
+for.body8:                                        ; preds = %for.body4, %for.inc40
+  %indvars.iv11 = phi i64 [ 0, %for.body4 ], [ %indvars.iv.next12, %for.inc40 ]
+  br label %for.body12
+
+for.body12:                                       ; preds = %for.body8, %for.inc37
+  %indvars.iv7 = phi i64 [ 3, %for.body8 ], [ %indvars.iv.next8, %for.inc37 ]
+  br label %for.body16
+
+for.body16:                                       ; preds = %for.body12, %for.body16
+  %indvars.iv = phi i64 [ 0, %for.body12 ], [ %indvars.iv.next, %for.body16 ]
+  %0 = add nuw nsw i64 %indvars.iv18, 1
+  %1 = add nsw i64 %indvars.iv14, -2
+  %2 = add nsw i64 %indvars.iv7, -3
+  %3 = add nuw nsw i64 %indvars.iv, 2
+  %arrayidx26 = getelementptr inbounds [1024 x [1024 x [1024 x [2048 x i32]]]], [1024 x [1024 x [1024 x [2048 x i32]]]]* %a, i64 %0, i64 %1, i64 %indvars.iv11, i64 %2, i64 %3
+  %4 = load i32, i32* %arrayidx26, align 4
+  %arrayidx36 = getelementptr inbounds [1024 x [1024 x [1024 x [2048 x i32]]]], [1024 x [1024 x [1024 x [2048 x i32]]]]* %a, i64 %indvars.iv18, i64 %indvars.iv14, i64 %indvars.iv11, i64 %indvars.iv7, i64 %indvars.iv
+  store i32 %4, i32* %arrayidx36, align 4
+  %indvars.iv.next = add nuw nsw i64 %indvars.iv, 1
+  %exitcond = icmp ne i64 %indvars.iv.next, 2046
+  br i1 %exitcond, label %for.body16, label %for.inc37
+
+for.inc37:                                        ; preds = %for.body16
+  %indvars.iv.next8 = add nuw nsw i64 %indvars.iv7, 1
+  %exitcond10 = icmp ne i64 %indvars.iv.next8, 1024
+  br i1 %exitcond10, label %for.body12, label %for.inc40
+
+for.inc40:                                        ; preds = %for.inc37
+  %indvars.iv.next12 = add nuw nsw i64 %indvars.iv11, 1
+  %exitcond13 = icmp ne i64 %indvars.iv.next12, 1024
+  br i1 %exitcond13, label %for.body8, label %for.inc43
+
+for.inc43:                                        ; preds = %for.inc40
+  %indvars.iv.next15 = add nuw nsw i64 %indvars.iv14, 1
+  %exitcond17 = icmp ne i64 %indvars.iv.next15, 1024
+  br i1 %exitcond17, label %for.body4, label %for.inc46
+
+for.inc46:                                        ; preds = %for.inc43
+  %indvars.iv.next19 = add nuw nsw i64 %indvars.iv18, 1
+  %exitcond21 = icmp ne i64 %indvars.iv.next19, 1023
+  br i1 %exitcond21, label %for.body, label %for.end48
+
+for.end48:                                        ; preds = %for.inc46
+  ret void
+}

llvm/test/Transforms/LoopInterchange/currentLimitation.ll

 ; RUN: opt < %s -basicaa -loop-interchange -pass-remarks-missed='loop-interchange' \
-; RUN:   -pass-remarks-output=%t -verify-loop-info -verify-dom-info -S | FileCheck -check-prefix=IR %s
+; RUN:   -da-disable-delinearization-checks -pass-remarks-output=%t -verify-loop-info   \
+; RUN:   -verify-dom-info -S | FileCheck -check-prefix=IR %s
 ; RUN: FileCheck --input-file=%t %s
 
 target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128"
 target triple = "x86_64-unknown-linux-gnu"
  
 @A = common global [100 x [100 x i32]] zeroinitializer
 @B = common global [100 x [100 x [100 x i32]]] zeroinitializer
 @C = common global [100 x [100 x i64]] zeroinitializer
  
 ;;--------------------------------------Test case 01------------------------------------
 ;; [FIXME] This loop though valid is currently not interchanged due to the limitation that we cannot split the inner loop latch due to multiple use of inner induction
 ;; variable.(used to increment the loop counter and to access A[j+1][i+1]
 ;;  for(int i=0;i<N-1;i++)
 ;;    for(int j=1;j<N-1;j++)
 ;;      A[j+1][i+1] = A[j+1][i+1] + k;
 
-; FIXME: Currently fails because of DA changes.
 ; IR-LABEL: @interchange_01
 ; IR-NOT: split
 
-; CHECK:      Name:            Dependence
+; CHECK:      Name:            UnsupportedInsBetweenInduction
 ; CHECK-NEXT: Function:        interchange_01
 
 define void @interchange_01(i32 %k, i32 %N) {
  entry:
    %sub = add nsw i32 %N, -1
    %cmp26 = icmp sgt i32 %N, 1
    br i1 %cmp26, label %for.cond1.preheader.lr.ph, label %for.end17
  

llvm/test/Transforms/LoopInterchange/loop-interchange-optimization-remarks.ll

 ; Test optimization remarks generated by the LoopInterchange pass.
 ;
 ; RUN: opt < %s -basicaa -loop-interchange -verify-dom-info -verify-loop-info \
 ; RUN:     -pass-remarks-output=%t -pass-remarks-missed='loop-interchange' \
-; RUN:     -pass-remarks='loop-interchange' -S
+; RUN:     -pass-remarks='loop-interchange' -da-disable-delinearization-checks -S
 ; RUN: cat %t |  FileCheck %s
 
 @A = common global [100 x [100 x i32]] zeroinitializer
 @B = common global [100 x [100 x i32]] zeroinitializer
 @C = common global [100 x i32] zeroinitializer
 
 ;;---------------------------------------Test case 01---------------------------------
 ;; Loops interchange is not profitable.
for.inc17:
   br i1 %exitcond38, label %for.end19, label %for.body3.lr.ph
 
 for.end19:
   ret void
 }
 
 ; CHECK: --- !Missed
 ; CHECK-NEXT: Pass:            loop-interchange
-; CHECK-NEXT: Name:            Dependence
+; CHECK-NEXT: Name:            InterchangeNotProfitable
 ; CHECK-NEXT: Function:        test01
 ; CHECK-NEXT: Args:
-; CHECK-NEXT:   - String:          Cannot interchange loops due to dependences.
+; CHECK-NEXT:   - String:          'Interchanging loops is too costly (cost='
+; CHECK-NEXT:   - Cost:            '2'
+; CHECK-NEXT:   - String:          ', threshold='
+; CHECK-NEXT:   - Threshold:       '0'
+; CHECK-NEXT:   - String:          ') and it does not improve parallelism.'
 ; CHECK-NEXT: ...
 
 ;;--------------------------------------Test case 02------------------------------------
 ;; [FIXME] This loop though valid is currently not interchanged due to the
 ;; limitation that we cannot split the inner loop latch due to multiple use of inner induction
 ;; variable.(used to increment the loop counter and to access A[j+1][i+1]
 ;;  for(int i=0;i<N-1;i++)
 ;;    for(int j=1;j<N-1;j++)
for.body4:
    br i1 %exitcond, label %for.cond.loopexit, label %for.body4
 
  for.end17:
    ret void
 }
 
 ; CHECK: --- !Missed
 ; CHECK-NEXT: Pass:            loop-interchange
-; CHECK-NEXT: Name:            Dependence
+; CHECK-NEXT: Name:            UnsupportedInsBetweenInduction
 ; CHECK-NEXT: Function:        test02
 ; CHECK-NEXT: Args:
-; CHECK-NEXT:   - String:          Cannot interchange loops due to dependences.
+; CHECK-NEXT:   - String:          Found unsupported instruction between induction variable increment and branch.
 ; CHECK-NEXT: ...
 
 ;;-----------------------------------Test case 03-------------------------------
 ;; Test to make sure we can handle output dependencies.
 ;;
 ;;  for (int i = 0; i < 2; ++i)
 ;;    for(int j = 0; j < 3; ++j) {
 ;;      A[j][i] = i;
for.body4:                                        ; preds = %for.body4, %for.cond1.preheader
   %arrayidx10 = getelementptr inbounds [3 x [3 x i32]], [3 x [3 x i32]]* @A10, i64 0, i64 %indvars.iv, i64 %indvars.iv.next27
   %tmp1 = trunc i64 %indvars.iv to i32
   store i32 %tmp1, i32* %arrayidx10, align 4
   %indvars.iv.next = add nuw nsw i64 %indvars.iv, 1
   %exitcond = icmp ne i64 %indvars.iv.next, 3
   br i1 %exitcond, label %for.body4, label %for.cond.loopexit
 }
 
-; CHECK: --- !Missed
+; CHECK: --- !Passed
 ; CHECK-NEXT: Pass:            loop-interchange
-; CHECK-NEXT: Name:            Dependence
+; CHECK-NEXT: Name:            Interchanged
 ; CHECK-NEXT: Function:        test03
 ; CHECK-NEXT: Args:
-; CHECK-NEXT:   - String:          Cannot interchange loops due to dependences.
+; CHECK-NEXT:  - String:          Loop interchanged with enclosing loop.
 ; CHECK-NEXT: ...
 
 ;;--------------------------------------Test case 04-------------------------------------
 ;; Loops not tightly nested are not interchanged
 ;;  for(int j=0;j<N;j++) {
 ;;    B[j] = j+k;
 ;;    for(int i=0;i<N;i++)
 ;;      A[j][i] = A[j][i]+B[j];
for.inc15:
   br i1 %exitcond36, label %for.end17, label %for.body
 
 for.end17:
   ret void
 }
 
 ; CHECK: --- !Missed
 ; CHECK-NEXT: Pass:            loop-interchange
-; CHECK-NEXT: Name:            Dependence
+; CHECK-NEXT: Name:            NotTightlyNested
 ; CHECK-NEXT: Function:        test04
 ; CHECK-NEXT: Args:
-; CHECK-NEXT:   - String:          Cannot interchange loops due to dependences.
+; CHECK-NEXT:  - String:          Cannot interchange loops because they are not tightly nested.
 ; CHECK-NEXT: ...

llvm/test/Transforms/LoopInterchange/profitability.ll

 ; RUN: opt < %s -loop-interchange -pass-remarks-output=%t -verify-dom-info -verify-loop-info \
-; RUN:     -pass-remarks=loop-interchange -pass-remarks-missed=loop-interchange
+; RUN:     -pass-remarks=loop-interchange -pass-remarks-missed=loop-interchange -da-disable-delinearization-checks
 ; RUN: FileCheck -input-file %t %s
 
 ;; We test profitability model in these test cases.
 
 target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128"
 target triple = "x86_64-unknown-linux-gnu"
 
 @A = common global [100 x [100 x i32]] zeroinitializer
 @B = common global [100 x [100 x i32]] zeroinitializer
 
 ;;---------------------------------------Test case 01---------------------------------
 ;; Loops interchange will result in code vectorization and hence profitable. Check for interchange.
 ;;   for(int i=1;i<100;i++)
 ;;     for(int j=1;j<100;j++)
 ;;       A[j][i] = A[j - 1][i] + B[j][i];
-;; FIXME: DA misses this case after D35430
 
-; CHECK:      Name:            Dependence
+; CHECK:      Name:            Interchanged
 ; CHECK-NEXT: Function:        interchange_01
 define void @interchange_01() {
 entry:
   br label %for2.preheader
 
 for2.preheader:
   %i30 = phi i64 [ 1, %entry ], [ %i.next31, %for1.inc14 ]
   br label %for2