[IR] add shuffle queries for identity extend/extract 

This was one of the potential follow-ups suggested in D48236, 
and these will be used to make matching the patterns in PR38691 cleaner:
https://bugs.llvm.org/show_bug.cgi?id=38691

About the vocabulary: in the DAG, these would be concat_vector with an 
undef operand or extract_subvector. Alternate names are discussed in the
review, but I think these are familiar/good enough to proceed. Once we
have uses of them in code, we might adjust if there are better options.

https://reviews.llvm.org/D51392

llvm-svn: 341075
diff --git a/llvm/include/llvm/IR/Instructions.h b/llvm/include/llvm/IR/Instructions.h
index 140a6f20..dea7fcd 100644
--- a/llvm/include/llvm/IR/Instructions.h
+++ b/llvm/include/llvm/IR/Instructions.h
@@ -2456,7 +2456,7 @@
   }
 
   /// Return true if this shuffle returns a vector with a different number of
-  /// elements than its source elements.
+  /// elements than its source vectors.
   /// Example: shufflevector <4 x n> A, <4 x n> B, <1,2>
   bool changesLength() const {
     unsigned NumSourceElts = Op<0>()->getType()->getVectorNumElements();
@@ -2497,15 +2497,22 @@
     return isIdentityMask(MaskAsInts);
   }
 
-  /// Return true if this shuffle mask chooses elements from exactly one source
+  /// Return true if this shuffle chooses elements from exactly one source
   /// vector without lane crossings and does not change the number of elements
   /// from its input vectors.
   /// Example: shufflevector <4 x n> A, <4 x n> B, <4,undef,6,undef>
-  /// TODO: Optionally allow length-changing shuffles.
   bool isIdentity() const {
     return !changesLength() && isIdentityMask(getShuffleMask());
   }
 
+  /// Return true if this shuffle lengthens exactly one source vector with
+  /// undefs in the high elements.
+  bool isIdentityWithPadding() const;
+
+  /// Return true if this shuffle extracts the first N elements of exactly one
+  /// source vector.
+  bool isIdentityWithExtract() const;
+
   /// Return true if this shuffle mask chooses elements from its source vectors
   /// without lane crossings. A shuffle using this mask would be
   /// equivalent to a vector select with a constant condition operand.
diff --git a/llvm/lib/IR/Instructions.cpp b/llvm/lib/IR/Instructions.cpp
index 1a2752d..2bf9f0b 100644
--- a/llvm/lib/IR/Instructions.cpp
+++ b/llvm/lib/IR/Instructions.cpp
@@ -1660,17 +1660,17 @@
   }
 }
 
-bool ShuffleVectorInst::isSingleSourceMask(ArrayRef<int> Mask) {
+static bool isSingleSourceMaskImpl(ArrayRef<int> Mask, int NumOpElts) {
   assert(!Mask.empty() && "Shuffle mask must contain elements");
   bool UsesLHS = false;
   bool UsesRHS = false;
-  for (int i = 0, NumElts = Mask.size(); i < NumElts; ++i) {
+  for (int i = 0, NumMaskElts = Mask.size(); i < NumMaskElts; ++i) {
     if (Mask[i] == -1)
       continue;
-    assert(Mask[i] >= 0 && Mask[i] < (NumElts * 2) &&
+    assert(Mask[i] >= 0 && Mask[i] < (NumOpElts * 2) &&
            "Out-of-bounds shuffle mask element");
-    UsesLHS |= (Mask[i] < NumElts);
-    UsesRHS |= (Mask[i] >= NumElts);
+    UsesLHS |= (Mask[i] < NumOpElts);
+    UsesRHS |= (Mask[i] >= NumOpElts);
     if (UsesLHS && UsesRHS)
       return false;
   }
@@ -1678,18 +1678,30 @@
   return true;
 }
 
-bool ShuffleVectorInst::isIdentityMask(ArrayRef<int> Mask) {
-  if (!isSingleSourceMask(Mask))
+bool ShuffleVectorInst::isSingleSourceMask(ArrayRef<int> Mask) {
+  // We don't have vector operand size information, so assume operands are the
+  // same size as the mask.
+  return isSingleSourceMaskImpl(Mask, Mask.size());
+}
+
+static bool isIdentityMaskImpl(ArrayRef<int> Mask, int NumOpElts) {
+  if (!isSingleSourceMaskImpl(Mask, NumOpElts))
     return false;
-  for (int i = 0, NumElts = Mask.size(); i < NumElts; ++i) {
+  for (int i = 0, NumMaskElts = Mask.size(); i < NumMaskElts; ++i) {
     if (Mask[i] == -1)
       continue;
-    if (Mask[i] != i && Mask[i] != (NumElts + i))
+    if (Mask[i] != i && Mask[i] != (NumOpElts + i))
       return false;
   }
   return true;
 }
 
+bool ShuffleVectorInst::isIdentityMask(ArrayRef<int> Mask) {
+  // We don't have vector operand size information, so assume operands are the
+  // same size as the mask.
+  return isIdentityMaskImpl(Mask, Mask.size());
+}
+
 bool ShuffleVectorInst::isReverseMask(ArrayRef<int> Mask) {
   if (!isSingleSourceMask(Mask))
     return false;
@@ -1761,6 +1773,33 @@
   return true;
 }
 
+bool ShuffleVectorInst::isIdentityWithPadding() const {
+  int NumOpElts = Op<0>()->getType()->getVectorNumElements();
+  int NumMaskElts = getType()->getVectorNumElements();
+  if (NumMaskElts <= NumOpElts)
+    return false;
+
+  // The first part of the mask must choose elements from exactly 1 source op.
+  ArrayRef<int> Mask = getShuffleMask();
+  if (!isIdentityMaskImpl(Mask, NumOpElts))
+    return false;
+
+  // All extending must be with undef elements.
+  for (int i = NumOpElts; i < NumMaskElts; ++i)
+    if (Mask[i] != -1)
+      return false;
+
+  return true;
+}
+
+bool ShuffleVectorInst::isIdentityWithExtract() const {
+  int NumOpElts = Op<0>()->getType()->getVectorNumElements();
+  int NumMaskElts = getType()->getVectorNumElements();
+  if (NumMaskElts >= NumOpElts)
+    return false;
+
+  return isIdentityMaskImpl(getShuffleMask(), NumOpElts);
+}
 
 //===----------------------------------------------------------------------===//
 //                             InsertValueInst Class
diff --git a/llvm/unittests/IR/InstructionsTest.cpp b/llvm/unittests/IR/InstructionsTest.cpp
index 6bac596..d153f99 100644
--- a/llvm/unittests/IR/InstructionsTest.cpp
+++ b/llvm/unittests/IR/InstructionsTest.cpp
@@ -837,6 +837,78 @@
 
   EXPECT_TRUE(ShuffleVectorInst::isTransposeMask(ConstantVector::get({C1, C5, C3, C7})));
   EXPECT_TRUE(ShuffleVectorInst::isTransposeMask(ConstantVector::get({C1, C3})));
+
+  // Identity with undef elts.
+  ShuffleVectorInst *Id1 = new ShuffleVectorInst(ConstantVector::get({C0, C1, C3, C3}),
+                                                 ConstantVector::get({C0, C1, C2, C3}),
+                                                 ConstantVector::get({C0, C1, CU, CU}));
+  EXPECT_TRUE(Id1->isIdentity());
+  EXPECT_FALSE(Id1->isIdentityWithPadding());
+  EXPECT_FALSE(Id1->isIdentityWithExtract());
+  delete Id1;
+
+  // Result has less elements than operands.
+  ShuffleVectorInst *Id2 = new ShuffleVectorInst(ConstantVector::get({C0, C1, C2, C3}),
+                                                 ConstantVector::get({C0, C1, C2, C3}),
+                                                 ConstantVector::get({C0, C1, C2}));
+  EXPECT_FALSE(Id2->isIdentity());
+  EXPECT_FALSE(Id2->isIdentityWithPadding());
+  EXPECT_TRUE(Id2->isIdentityWithExtract());
+  delete Id2;
+
+  // Result has less elements than operands; choose from Op1.
+  ShuffleVectorInst *Id3 = new ShuffleVectorInst(ConstantVector::get({C0, C1, C2, C3}),
+                                                 ConstantVector::get({C0, C1, C2, C3}),
+                                                 ConstantVector::get({C4, CU, C6}));
+  EXPECT_FALSE(Id3->isIdentity());
+  EXPECT_FALSE(Id3->isIdentityWithPadding());
+  EXPECT_TRUE(Id3->isIdentityWithExtract());
+  delete Id3;
+
+  // Result has less elements than operands; choose from Op0 and Op1 is not identity.
+  ShuffleVectorInst *Id4 = new ShuffleVectorInst(ConstantVector::get({C0, C1, C2, C3}),
+                                                 ConstantVector::get({C0, C1, C2, C3}),
+                                                 ConstantVector::get({C4, C1, C6}));
+  EXPECT_FALSE(Id4->isIdentity());
+  EXPECT_FALSE(Id4->isIdentityWithPadding());
+  EXPECT_FALSE(Id4->isIdentityWithExtract());
+  delete Id4;
+
+  // Result has more elements than operands, and extra elements are undef.
+  ShuffleVectorInst *Id5 = new ShuffleVectorInst(ConstantVector::get({C0, C1, C2, C3}),
+                                                 ConstantVector::get({C0, C1, C2, C3}),
+                                                 ConstantVector::get({CU, C1, C2, C3, CU, CU}));
+  EXPECT_FALSE(Id5->isIdentity());
+  EXPECT_TRUE(Id5->isIdentityWithPadding());
+  EXPECT_FALSE(Id5->isIdentityWithExtract());
+  delete Id5;
+
+  // Result has more elements than operands, and extra elements are undef; choose from Op1.
+  ShuffleVectorInst *Id6 = new ShuffleVectorInst(ConstantVector::get({C0, C1, C2, C3}),
+                                                 ConstantVector::get({C0, C1, C2, C3}),
+                                                 ConstantVector::get({C4, C5, C6, CU, CU, CU}));
+  EXPECT_FALSE(Id6->isIdentity());
+  EXPECT_TRUE(Id6->isIdentityWithPadding());
+  EXPECT_FALSE(Id6->isIdentityWithExtract());
+  delete Id6;
+  
+  // Result has more elements than operands, but extra elements are not undef.
+  ShuffleVectorInst *Id7 = new ShuffleVectorInst(ConstantVector::get({C0, C1, C2, C3}),
+                                                 ConstantVector::get({C0, C1, C2, C3}),
+                                                 ConstantVector::get({C0, C1, C2, C3, CU, C1}));
+  EXPECT_FALSE(Id7->isIdentity());
+  EXPECT_FALSE(Id7->isIdentityWithPadding());
+  EXPECT_FALSE(Id7->isIdentityWithExtract());
+  delete Id7;
+  
+  // Result has more elements than operands; choose from Op0 and Op1 is not identity.
+  ShuffleVectorInst *Id8 = new ShuffleVectorInst(ConstantVector::get({C0, C1, C2, C3}),
+                                                 ConstantVector::get({C0, C1, C2, C3}),
+                                                 ConstantVector::get({C4, CU, C2, C3, CU, CU}));
+  EXPECT_FALSE(Id8->isIdentity());
+  EXPECT_FALSE(Id8->isIdentityWithPadding());
+  EXPECT_FALSE(Id8->isIdentityWithExtract());
+  delete Id8;
 }
 
 TEST(InstructionsTest, SkipDebug) {