[TargetLowering] fix isConstTrueVal to account for build vector truncation

Build vectors have magical truncation powers, so we have things like this:

v4i1 = BUILD_VECTOR Constant:i32<1>, Constant:i32<1>, Constant:i32<1>, Constant:i32<1>
v4i16 = BUILD_VECTOR Constant:i32<1>, Constant:i32<1>, Constant:i32<1>, Constant:i32<1>

If we don't truncate the splat node returned by getConstantSplatNode(), then we won't find
truth when ZeroOrNegativeOneBooleanContent is the rule.

Differential Revision: https://reviews.llvm.org/D32505

llvm-svn: 301408

diff --git a/llvm/lib/CodeGen/SelectionDAG/TargetLowering.cpp b/llvm/lib/CodeGen/SelectionDAG/TargetLowering.cpp
index 18b238d..136dec8 100644 (file)
--- a/llvm/lib/CodeGen/SelectionDAG/TargetLowering.cpp
+++ b/llvm/lib/CodeGen/SelectionDAG/TargetLowering.cpp
@@ -1371,27 +1371,31 @@ bool TargetLowering::isConstTrueVal(const SDNode *N) const {
   if (!N)
     return false;
 
-  const ConstantSDNode *CN = dyn_cast<ConstantSDNode>(N);
-  if (!CN) {
-    const BuildVectorSDNode *BV = dyn_cast<BuildVectorSDNode>(N);
-    if (!BV)
-      return false;
-
-    // Only interested in constant splats, we don't care about undef
-    // elements in identifying boolean constants and getConstantSplatNode
-    // returns NULL if all ops are undef;
-    CN = BV->getConstantSplatNode();
+  APInt CVal;
+  if (auto *CN = dyn_cast<ConstantSDNode>(N)) {
+    CVal = CN->getAPIntValue();
+  } else if (auto *BV = dyn_cast<BuildVectorSDNode>(N)) {
+    auto *CN = BV->getConstantSplatNode();
     if (!CN)
       return false;
+
+    // If this is a truncating build vector, truncate the splat value.
+    // Otherwise, we may fail to match the expected values below.
+    unsigned BVEltWidth = BV->getValueType(0).getScalarSizeInBits();
+    CVal = CN->getAPIntValue();
+    if (BVEltWidth < CVal.getBitWidth())
+      CVal = CVal.trunc(BVEltWidth);
+  } else {
+    return false;
   }
 
   switch (getBooleanContents(N->getValueType(0))) {
   case UndefinedBooleanContent:
-    return CN->getAPIntValue()[0];
+    return CVal[0];
   case ZeroOrOneBooleanContent:
-    return CN->isOne();
+    return CVal == 1;
   case ZeroOrNegativeOneBooleanContent:
-    return CN->isAllOnesValue();
+    return CVal.isAllOnesValue();
   }
 
   llvm_unreachable("Invalid boolean contents");

diff --git a/llvm/test/CodeGen/X86/setcc-combine.ll b/llvm/test/CodeGen/X86/setcc-combine.ll
index a7c5dec..38205c6 100644 (file)
--- a/llvm/test/CodeGen/X86/setcc-combine.ll
+++ b/llvm/test/CodeGen/X86/setcc-combine.ll
@@ -5,11 +5,10 @@ define i32 @test_eq_1(<4 x i32> %A, <4 x i32> %B) {
 ; CHECK-LABEL: test_eq_1:
 ; CHECK:       # BB#0:
 ; CHECK-NEXT:    pcmpgtd %xmm0, %xmm1
-; CHECK-NEXT:    pxor {{.*}}(%rip), %xmm1
-; CHECK-NEXT:    pshufd {{.*#+}} xmm0 = xmm1[1,1,2,3]
+; CHECK-NEXT:    pcmpeqd %xmm0, %xmm0
+; CHECK-NEXT:    pxor %xmm1, %xmm0
+; CHECK-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[1,1,2,3]
 ; CHECK-NEXT:    movd %xmm0, %eax
-; CHECK-NEXT:    andl $1, %eax
-; CHECK-NEXT:    negl %eax
 ; CHECK-NEXT:    retq
   %cmp = icmp slt <4 x i32> %A, %B
   %sext = sext <4 x i1> %cmp to <4 x i32>
@@ -51,11 +50,10 @@ define i32 @test_ge_1(<4 x i32> %A, <4 x i32> %B) {
 ; CHECK-LABEL: test_ge_1:
 ; CHECK:       # BB#0:
 ; CHECK-NEXT:    pcmpgtd %xmm0, %xmm1
-; CHECK-NEXT:    pxor {{.*}}(%rip), %xmm1
-; CHECK-NEXT:    pshufd {{.*#+}} xmm0 = xmm1[1,1,2,3]
+; CHECK-NEXT:    pcmpeqd %xmm0, %xmm0
+; CHECK-NEXT:    pxor %xmm1, %xmm0
+; CHECK-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[1,1,2,3]
 ; CHECK-NEXT:    movd %xmm0, %eax
-; CHECK-NEXT:    andl $1, %eax
-; CHECK-NEXT:    negl %eax
 ; CHECK-NEXT:    retq
   %cmp = icmp slt <4 x i32> %A, %B
   %sext = sext <4 x i1> %cmp to <4 x i32>
@@ -97,11 +95,10 @@ define i32 @test_eq_2(<4 x i32> %A, <4 x i32> %B) {
 ; CHECK-LABEL: test_eq_2:
 ; CHECK:       # BB#0:
 ; CHECK-NEXT:    pcmpgtd %xmm1, %xmm0
-; CHECK-NEXT:    pxor {{.*}}(%rip), %xmm0
-; CHECK-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[1,1,2,3]
+; CHECK-NEXT:    pcmpeqd %xmm1, %xmm1
+; CHECK-NEXT:    pxor %xmm0, %xmm1
+; CHECK-NEXT:    pshufd {{.*#+}} xmm0 = xmm1[1,1,2,3]
 ; CHECK-NEXT:    movd %xmm0, %eax
-; CHECK-NEXT:    andl $1, %eax
-; CHECK-NEXT:    negl %eax
 ; CHECK-NEXT:    retq
   %cmp = icmp slt <4 x i32> %B, %A
   %sext = sext <4 x i1> %cmp to <4 x i32>
@@ -130,11 +127,10 @@ define i32 @test_le_2(<4 x i32> %A, <4 x i32> %B) {
 ; CHECK-LABEL: test_le_2:
 ; CHECK:       # BB#0:
 ; CHECK-NEXT:    pcmpgtd %xmm1, %xmm0
-; CHECK-NEXT:    pxor {{.*}}(%rip), %xmm0
-; CHECK-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[1,1,2,3]
+; CHECK-NEXT:    pcmpeqd %xmm1, %xmm1
+; CHECK-NEXT:    pxor %xmm0, %xmm1
+; CHECK-NEXT:    pshufd {{.*#+}} xmm0 = xmm1[1,1,2,3]
 ; CHECK-NEXT:    movd %xmm0, %eax
-; CHECK-NEXT:    andl $1, %eax
-; CHECK-NEXT:    negl %eax
 ; CHECK-NEXT:    retq
   %cmp = icmp slt <4 x i32> %B, %A
   %sext = sext <4 x i1> %cmp to <4 x i32>