[InstCombine] Handle const select arm in foldSelectCtlzToCttz()

The select arm that takes the ctlz result can also instead be a
constant with the bit width (as this is what the ctlz evaluates to
for a==0).

This avoids a regression when strengthening the
simplifyWithOpReplaced() fold.

Proof: https://alive2.llvm.org/ce/z/DMRL5A

diff --git a/llvm/lib/Transforms/InstCombine/InstCombineSelect.cpp b/llvm/lib/Transforms/InstCombine/InstCombineSelect.cpp
index 85892b0..350c037 100644 (file)
--- a/llvm/lib/Transforms/InstCombine/InstCombineSelect.cpp
+++ b/llvm/lib/Transforms/InstCombine/InstCombineSelect.cpp
@@ -1060,6 +1060,8 @@ static Value *foldAbsDiff(ICmpInst *Cmp, Value *TVal, Value *FVal,
 /// \code
 ///   int a = ctlz(x & -x);
 //    x ? 31 - a : a;
+//    // or
+//    x ? 31 - a : 32;
 /// \code
 ///
 /// into:
@@ -1074,15 +1076,19 @@ static Instruction *foldSelectCtlzToCttz(ICmpInst *ICI, Value *TrueVal,
   if (ICI->getPredicate() == ICmpInst::ICMP_NE)
     std::swap(TrueVal, FalseVal);
 
+  Value *Ctlz;
   if (!match(FalseVal,
-             m_Xor(m_Deferred(TrueVal), m_SpecificInt(BitWidth - 1))))
+             m_Xor(m_Value(Ctlz), m_SpecificInt(BitWidth - 1))))
     return nullptr;
 
-  if (!match(TrueVal, m_Intrinsic<Intrinsic::ctlz>()))
+  if (!match(Ctlz, m_Intrinsic<Intrinsic::ctlz>()))
+    return nullptr;
+
+  if (TrueVal != Ctlz && !match(TrueVal, m_SpecificInt(BitWidth)))
     return nullptr;
 
   Value *X = ICI->getOperand(0);
-  auto *II = cast<IntrinsicInst>(TrueVal);
+  auto *II = cast<IntrinsicInst>(Ctlz);
   if (!match(II->getOperand(0), m_c_And(m_Specific(X), m_Neg(m_Specific(X)))))
     return nullptr;
 

diff --git a/llvm/test/Transforms/InstCombine/select-ctlz-to-cttz.ll b/llvm/test/Transforms/InstCombine/select-ctlz-to-cttz.ll
index 7bd6a20..f0b36c8 100644 (file)
--- a/llvm/test/Transforms/InstCombine/select-ctlz-to-cttz.ll
+++ b/llvm/test/Transforms/InstCombine/select-ctlz-to-cttz.ll
@@ -44,12 +44,7 @@ define i32 @select_clz_to_ctz_preserve_flag(i32 %a) {
 
 define i32 @select_clz_to_ctz_constant_for_zero(i32 %a) {
 ; CHECK-LABEL: @select_clz_to_ctz_constant_for_zero(
-; CHECK-NEXT:    [[SUB:%.*]] = sub i32 0, [[A:%.*]]
-; CHECK-NEXT:    [[AND:%.*]] = and i32 [[SUB]], [[A]]
-; CHECK-NEXT:    [[LZ:%.*]] = tail call i32 @llvm.ctlz.i32(i32 [[AND]], i1 false), !range [[RNG0]]
-; CHECK-NEXT:    [[TOBOOL:%.*]] = icmp eq i32 [[A]], 0
-; CHECK-NEXT:    [[SUB1:%.*]] = xor i32 [[LZ]], 31
-; CHECK-NEXT:    [[COND:%.*]] = select i1 [[TOBOOL]], i32 32, i32 [[SUB1]]
+; CHECK-NEXT:    [[COND:%.*]] = call i32 @llvm.cttz.i32(i32 [[A:%.*]], i1 false), !range [[RNG0]]
 ; CHECK-NEXT:    ret i32 [[COND]]
 ;
   %sub = sub i32 0, %a