[InstCombine] Fold add nuw + uadd.with.overflow

Fold add nuw and uadd.with.overflow with constants if the
addition does not overflow.

Part of https://bugs.llvm.org/show_bug.cgi?id=38146.

Patch by Dan Robertson.

Differential Revision: https://reviews.llvm.org/D59471

llvm-svn: 356584

diff --git a/llvm/lib/Transforms/InstCombine/InstCombineCalls.cpp b/llvm/lib/Transforms/InstCombine/InstCombineCalls.cpp
index df80221..359f617 100644 (file)
--- a/llvm/lib/Transforms/InstCombine/InstCombineCalls.cpp
+++ b/llvm/lib/Transforms/InstCombine/InstCombineCalls.cpp
@@ -2063,6 +2063,7 @@ Instruction *InstCombiner::visitCallInst(CallInst &CI) {
       return &CI;
     break;
   }
+  case Intrinsic::uadd_with_overflow:
   case Intrinsic::sadd_with_overflow: {
     if (Instruction *I = canonicalizeConstantArg0ToArg1(CI))
       return I;
@@ -2070,25 +2071,27 @@ Instruction *InstCombiner::visitCallInst(CallInst &CI) {
       return I;
 
     // Given 2 constant operands whose sum does not overflow:
+    // uaddo (X +nuw C0), C1 -> uaddo X, C0 + C1
     // saddo (X +nsw C0), C1 -> saddo X, C0 + C1
     Value *X;
     const APInt *C0, *C1;
     Value *Arg0 = II->getArgOperand(0);
     Value *Arg1 = II->getArgOperand(1);
-    if (match(Arg0, m_NSWAdd(m_Value(X), m_APInt(C0))) &&
-        match(Arg1, m_APInt(C1))) {
+    bool IsSigned = II->getIntrinsicID() == Intrinsic::sadd_with_overflow;
+    bool HasNWAdd = IsSigned ? match(Arg0, m_NSWAdd(m_Value(X), m_APInt(C0)))
+                             : match(Arg0, m_NUWAdd(m_Value(X), m_APInt(C0)));
+    if (HasNWAdd && match(Arg1, m_APInt(C1))) {
       bool Overflow;
-      APInt NewC = C1->sadd_ov(*C0, Overflow);
+      APInt NewC =
+          IsSigned ? C1->sadd_ov(*C0, Overflow) : C1->uadd_ov(*C0, Overflow);
       if (!Overflow)
         return replaceInstUsesWith(
             *II, Builder.CreateBinaryIntrinsic(
-                     Intrinsic::sadd_with_overflow, X,
+                     II->getIntrinsicID(), X,
                      ConstantInt::get(Arg1->getType(), NewC)));
     }
-
     break;
   }
-  case Intrinsic::uadd_with_overflow:
   case Intrinsic::umul_with_overflow:
   case Intrinsic::smul_with_overflow:
     if (Instruction *I = canonicalizeConstantArg0ToArg1(CI))

diff --git a/llvm/test/Transforms/InstCombine/uadd-with-overflow.ll b/llvm/test/Transforms/InstCombine/uadd-with-overflow.ll
index c9776f4..b306cb6 100644 (file)
--- a/llvm/test/Transforms/InstCombine/uadd-with-overflow.ll
+++ b/llvm/test/Transforms/InstCombine/uadd-with-overflow.ll
@@ -11,9 +11,8 @@ declare { i8, i1 } @llvm.uadd.with.overflow.i8(i8, i8)
 
 define { i32, i1 } @simple_fold(i32 %x) {
 ; CHECK-LABEL: @simple_fold(
-; CHECK-NEXT:    [[A:%.*]] = add nuw i32 [[X:%.*]], 7
-; CHECK-NEXT:    [[B:%.*]] = tail call { i32, i1 } @llvm.uadd.with.overflow.i32(i32 [[A]], i32 13)
-; CHECK-NEXT:    ret { i32, i1 } [[B]]
+; CHECK-NEXT:    [[TMP1:%.*]] = call { i32, i1 } @llvm.uadd.with.overflow.i32(i32 [[X:%.*]], i32 20)
+; CHECK-NEXT:    ret { i32, i1 } [[TMP1]]
 ;
   %a = add nuw i32 %x, 7
   %b = tail call { i32, i1 } @llvm.uadd.with.overflow.i32(i32 %a, i32 13)
@@ -22,9 +21,8 @@ define { i32, i1 } @simple_fold(i32 %x) {
 
 define { i8, i1 } @fold_on_constant_add_no_overflow(i8 %x) {
 ; CHECK-LABEL: @fold_on_constant_add_no_overflow(
-; CHECK-NEXT:    [[A:%.*]] = add nuw i8 [[X:%.*]], -56
-; CHECK-NEXT:    [[B:%.*]] = tail call { i8, i1 } @llvm.uadd.with.overflow.i8(i8 [[A]], i8 55)
-; CHECK-NEXT:    ret { i8, i1 } [[B]]
+; CHECK-NEXT:    [[TMP1:%.*]] = call { i8, i1 } @llvm.uadd.with.overflow.i8(i8 [[X:%.*]], i8 -1)
+; CHECK-NEXT:    ret { i8, i1 } [[TMP1]]
 ;
   %a = add nuw i8 %x, 200
   %b = tail call { i8, i1 } @llvm.uadd.with.overflow.i8(i8 %a, i8 55)
@@ -65,9 +63,8 @@ define { <2 x i8>, <2 x i1> } @no_fold_vector_overflow(<2 x i8> %x) {
 
 define { <2 x i32>, <2 x i1> } @fold_simple_splat_constant(<2 x i32> %x) {
 ; CHECK-LABEL: @fold_simple_splat_constant(
-; CHECK-NEXT:    [[A:%.*]] = add nuw <2 x i32> [[X:%.*]], <i32 12, i32 12>
-; CHECK-NEXT:    [[B:%.*]] = tail call { <2 x i32>, <2 x i1> } @llvm.uadd.with.overflow.v2i32(<2 x i32> [[A]], <2 x i32> <i32 30, i32 30>)
-; CHECK-NEXT:    ret { <2 x i32>, <2 x i1> } [[B]]
+; CHECK-NEXT:    [[TMP1:%.*]] = call { <2 x i32>, <2 x i1> } @llvm.uadd.with.overflow.v2i32(<2 x i32> [[X:%.*]], <2 x i32> <i32 42, i32 42>)
+; CHECK-NEXT:    ret { <2 x i32>, <2 x i1> } [[TMP1]]
 ;
   %a = add nuw <2 x i32> %x, <i32 12, i32 12>
   %b = tail call { <2 x i32>, <2 x i1> } @llvm.uadd.with.overflow.v2i32(<2 x i32> %a, <2 x i32> <i32 30, i32 30>)
@@ -98,9 +95,8 @@ define { <2 x i32>, <2 x i1> } @no_fold_splat_not_constant(<2 x i32> %x, <2 x i3
 
 define { i32, i1 } @fold_nuwnsw(i32 %x) {
 ; CHECK-LABEL: @fold_nuwnsw(
-; CHECK-NEXT:    [[A:%.*]] = add nuw nsw i32 [[X:%.*]], 12
-; CHECK-NEXT:    [[B:%.*]] = tail call { i32, i1 } @llvm.uadd.with.overflow.i32(i32 [[A]], i32 30)
-; CHECK-NEXT:    ret { i32, i1 } [[B]]
+; CHECK-NEXT:    [[TMP1:%.*]] = call { i32, i1 } @llvm.uadd.with.overflow.i32(i32 [[X:%.*]], i32 42)
+; CHECK-NEXT:    ret { i32, i1 } [[TMP1]]
 ;
   %a = add nuw nsw i32 %x, 12
   %b = tail call { i32, i1 } @llvm.uadd.with.overflow.i32(i32 %a, i32 30)