[PATCH] D62294: [DAGCombine] (x - C) - y -> (x - y) - C fold

Phabricator via Phabricator via llvm-commits llvm-commits at lists.llvm.org
Tue May 28 10:52:10 PDT 2019


This revision was automatically updated to reflect the committed changes.
Closed by commit rL361856: [DAGCombine] (x - C) - y  ->  (x - y) - C  fold (authored by lebedevri, committed by ).

Changed prior to commit:
  https://reviews.llvm.org/D62294?vs=201702&id=201712#toc

Repository:
  rL LLVM

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D62294/new/

https://reviews.llvm.org/D62294

Files:
  llvm/trunk/lib/CodeGen/SelectionDAG/DAGCombiner.cpp
  llvm/trunk/test/CodeGen/AArch64/sink-addsub-of-const.ll
  llvm/trunk/test/CodeGen/X86/sink-addsub-of-const.ll


Index: llvm/trunk/lib/CodeGen/SelectionDAG/DAGCombiner.cpp
===================================================================
--- llvm/trunk/lib/CodeGen/SelectionDAG/DAGCombiner.cpp
+++ llvm/trunk/lib/CodeGen/SelectionDAG/DAGCombiner.cpp
@@ -2957,6 +2957,13 @@
     SDValue Sub = DAG.getNode(ISD::SUB, DL, VT, N0, N1.getOperand(0));
     return DAG.getNode(ISD::SUB, DL, VT, Sub, N1.getOperand(1));
   }
+  // (x - C) - y  ->  (x - y) - C
+  // This is necessary because SUB(X,C) -> ADD(X,-C) doesn't work for vectors.
+  if (N0.hasOneUse() && N0.getOpcode() == ISD::SUB &&
+      isConstantOrConstantVector(N0.getOperand(1))) {
+    SDValue Sub = DAG.getNode(ISD::SUB, DL, VT, N0.getOperand(0), N1);
+    return DAG.getNode(ISD::SUB, DL, VT, Sub, N0.getOperand(1));
+  }
 
   // If the target's bool is represented as 0/-1, prefer to make this 'add 0/-1'
   // rather than 'sub 0/1' (the sext should get folded).
Index: llvm/trunk/test/CodeGen/AArch64/sink-addsub-of-const.ll
===================================================================
--- llvm/trunk/test/CodeGen/AArch64/sink-addsub-of-const.ll
+++ llvm/trunk/test/CodeGen/AArch64/sink-addsub-of-const.ll
@@ -314,8 +314,8 @@
 ; CHECK-NEXT:    adrp x8, .LCPI20_0
 ; CHECK-NEXT:    ldr q3, [x8, :lo12:.LCPI20_0]
 ; CHECK-NEXT:    sub v0.4s, v0.4s, v1.4s
-; CHECK-NEXT:    sub v0.4s, v0.4s, v3.4s
 ; CHECK-NEXT:    sub v0.4s, v0.4s, v2.4s
+; CHECK-NEXT:    sub v0.4s, v0.4s, v3.4s
 ; CHECK-NEXT:    ret
   %t0 = sub <4 x i32> %a, %b
   %t1 = sub <4 x i32> %t0, <i32 42, i32 24, i32 undef, i32 46>
Index: llvm/trunk/test/CodeGen/X86/sink-addsub-of-const.ll
===================================================================
--- llvm/trunk/test/CodeGen/X86/sink-addsub-of-const.ll
+++ llvm/trunk/test/CodeGen/X86/sink-addsub-of-const.ll
@@ -460,15 +460,15 @@
 ; X32-LABEL: vec_sink_sub_of_const_to_sub:
 ; X32:       # %bb.0:
 ; X32-NEXT:    psubd %xmm1, %xmm0
-; X32-NEXT:    psubd {{\.LCPI.*}}, %xmm0
 ; X32-NEXT:    psubd %xmm2, %xmm0
+; X32-NEXT:    psubd {{\.LCPI.*}}, %xmm0
 ; X32-NEXT:    retl
 ;
 ; X64-LABEL: vec_sink_sub_of_const_to_sub:
 ; X64:       # %bb.0:
 ; X64-NEXT:    psubd %xmm1, %xmm0
-; X64-NEXT:    psubd {{.*}}(%rip), %xmm0
 ; X64-NEXT:    psubd %xmm2, %xmm0
+; X64-NEXT:    psubd {{.*}}(%rip), %xmm0
 ; X64-NEXT:    retq
   %t0 = sub <4 x i32> %a, %b
   %t1 = sub <4 x i32> %t0, <i32 42, i32 24, i32 undef, i32 46>


-------------- next part --------------
A non-text attachment was scrubbed...
Name: D62294.201712.patch
Type: text/x-patch
Size: 2397 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20190528/9e158cff/attachment.bin>


More information about the llvm-commits mailing list