[PATCH] D59471: [InstCombine] Fold add nuw + uadd.with.overflow

[Dan Robertson via Phabricator via llvm-commits]

dlrobertson created this revision.
dlrobertson added reviewers: spatel, nikic.
Herald added subscribers: llvm-commits, hiraditya.
Herald added a project: LLVM.

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.


https://reviews.llvm.org/D59471

Files:
  llvm/lib/Transforms/InstCombine/InstCombineCalls.cpp
  llvm/test/Transforms/InstCombine/uadd-with-overflow.ll


Index: llvm/test/Transforms/InstCombine/uadd-with-overflow.ll
===================================================================
--- llvm/test/Transforms/InstCombine/uadd-with-overflow.ll
+++ llvm/test/Transforms/InstCombine/uadd-with-overflow.ll
@@ -9,9 +9,8 @@
 
 define { i32, i1 } @simple_fold(i32) {
 ; CHECK-LABEL: @simple_fold(
-; CHECK-NEXT:    [[TMP2:%.*]] = add nuw i32 [[TMP0:%.*]], 7
-; CHECK-NEXT:    [[TMP3:%.*]] = tail call { i32, i1 } @llvm.uadd.with.overflow.i32(i32 [[TMP2]], i32 13)
-; CHECK-NEXT:    ret { i32, i1 } [[TMP3]]
+; CHECK-NEXT:    [[TMP2:%.*]] = call { i32, i1 } @llvm.uadd.with.overflow.i32(i32 [[TMP0:%.*]], i32 20)
+; CHECK-NEXT:    ret { i32, i1 } [[TMP2]]
 ;
   %2 = add nuw i32 %0, 7
   %3 = tail call { i32, i1 } @llvm.uadd.with.overflow.i32(i32 %2, i32 13)
@@ -20,9 +19,8 @@
 
 define { i8, i1 } @fold_on_constant_add_no_overflow(i8) {
 ; CHECK-LABEL: @fold_on_constant_add_no_overflow(
-; CHECK-NEXT:    [[TMP2:%.*]] = add nuw i8 [[TMP0:%.*]], -56
-; CHECK-NEXT:    [[TMP3:%.*]] = tail call { i8, i1 } @llvm.uadd.with.overflow.i8(i8 [[TMP2]], i8 55)
-; CHECK-NEXT:    ret { i8, i1 } [[TMP3]]
+; CHECK-NEXT:    [[TMP2:%.*]] = call { i8, i1 } @llvm.uadd.with.overflow.i8(i8 [[TMP0:%.*]], i8 -1)
+; CHECK-NEXT:    ret { i8, i1 } [[TMP2]]
 ;
   %2 = add nuw i8 %0, 200
   %3 = tail call { i8, i1 } @llvm.uadd.with.overflow.i8(i8 %2, i8 55)
@@ -42,9 +40,8 @@
 
 define { <2 x i32>, <2 x i1> } @fold_simple_splat_constant(<2 x i32>) {
 ; CHECK-LABEL: @fold_simple_splat_constant(
-; CHECK-NEXT:    [[TMP2:%.*]] = add nuw <2 x i32> [[TMP0:%.*]], <i32 12, i32 12>
-; CHECK-NEXT:    [[TMP3:%.*]] = tail call { <2 x i32>, <2 x i1> } @llvm.uadd.with.overflow.v2i32(<2 x i32> [[TMP2]], <2 x i32> <i32 30, i32 30>)
-; CHECK-NEXT:    ret { <2 x i32>, <2 x i1> } [[TMP3]]
+; CHECK-NEXT:    [[TMP2:%.*]] = call { <2 x i32>, <2 x i1> } @llvm.uadd.with.overflow.v2i32(<2 x i32> [[TMP0:%.*]], <2 x i32> <i32 42, i32 42>)
+; CHECK-NEXT:    ret { <2 x i32>, <2 x i1> } [[TMP2]]
 ;
   %2 = add nuw <2 x i32> %0, <i32 12, i32 12>
   %3 = tail call { <2 x i32>, <2 x i1> } @llvm.uadd.with.overflow.v2i32(<2 x i32> %2, <2 x i32> <i32 30, i32 30>)
@@ -75,9 +72,8 @@
 
 define { i32, i1 } @fold_nuwnsw(i32) {
 ; CHECK-LABEL: @fold_nuwnsw(
-; CHECK-NEXT:    [[TMP2:%.*]] = add nuw nsw i32 [[TMP0:%.*]], 12
-; CHECK-NEXT:    [[TMP3:%.*]] = tail call { i32, i1 } @llvm.uadd.with.overflow.i32(i32 [[TMP2]], i32 30)
-; CHECK-NEXT:    ret { i32, i1 } [[TMP3]]
+; CHECK-NEXT:    [[TMP2:%.*]] = call { i32, i1 } @llvm.uadd.with.overflow.i32(i32 [[TMP0:%.*]], i32 42)
+; CHECK-NEXT:    ret { i32, i1 } [[TMP2]]
 ;
   %2 = add nuw nsw i32 %0, 12
   %3 = tail call { i32, i1 } @llvm.uadd.with.overflow.i32(i32 %2, i32 30)
Index: llvm/lib/Transforms/InstCombine/InstCombineCalls.cpp
===================================================================
--- llvm/lib/Transforms/InstCombine/InstCombineCalls.cpp
+++ llvm/lib/Transforms/InstCombine/InstCombineCalls.cpp
@@ -2054,7 +2054,30 @@
 
     break;
   }
-  case Intrinsic::uadd_with_overflow:
+  case Intrinsic::uadd_with_overflow: {
+    if (Instruction *I = canonicalizeConstantArg0ToArg1(CI))
+      return I;
+    if (Instruction *I = foldIntrinsicWithOverflowCommon(II))
+      return I;
+
+    // Given 2 constant operands whose sum does not overflow:
+    // uaddo (X +nuw C0), C1 -> uaddo X, C0 + C1
+    Value *X;
+    const APInt *C0, *C1;
+    Value *Arg0 = II->getArgOperand(0);
+    Value *Arg1 = II->getArgOperand(1);
+    if (match(Arg0, m_NUWAdd(m_Value(X), m_APInt(C0))) &&
+        match(Arg1, m_APInt(C1))) {
+      bool Overflow;
+      APInt NewC = C1->uadd_ov(*C0, Overflow);
+      if (!Overflow)
+        return replaceInstUsesWith(
+            *II, Builder.CreateBinaryIntrinsic(
+                     Intrinsic::uadd_with_overflow, X,
+                     ConstantInt::get(Arg1->getType(), NewC)));
+    }
+    break;
+  }
   case Intrinsic::umul_with_overflow:
   case Intrinsic::smul_with_overflow:
     if (Instruction *I = canonicalizeConstantArg0ToArg1(CI))


-------------- next part --------------
A non-text attachment was scrubbed...
Name: D59471.191021.patch
Type: text/x-patch
Size: 4047 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20190317/820fca7b/attachment.bin>