[PATCH] D98091: [RISCV] Make use of DAG.getNeutralElement in lowerVECREDUCE to avoid repeating the same list of constants. NFC

[Craig Topper via Phabricator via llvm-commits]

craig.topper updated this revision to Diff 328828.
craig.topper added a comment.

Rename function


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D98091/new/

https://reviews.llvm.org/D98091

Files:
  llvm/lib/Target/RISCV/RISCVISelLowering.cpp


Index: llvm/lib/Target/RISCV/RISCVISelLowering.cpp
===================================================================
--- llvm/lib/Target/RISCV/RISCVISelLowering.cpp
+++ llvm/lib/Target/RISCV/RISCVISelLowering.cpp
@@ -2387,27 +2387,26 @@
       RISCV::RVVBitsPerBlock / VT.getVectorElementType().getSizeInBits());
 }
 
-static std::pair<unsigned, uint64_t>
-getRVVReductionOpAndIdentityVal(unsigned ISDOpcode, unsigned EltSizeBits) {
+static unsigned getRVVReductionOp(unsigned ISDOpcode) {
   switch (ISDOpcode) {
   default:
     llvm_unreachable("Unhandled reduction");
   case ISD::VECREDUCE_ADD:
-    return {RISCVISD::VECREDUCE_ADD, 0};
+    return RISCVISD::VECREDUCE_ADD;
   case ISD::VECREDUCE_UMAX:
-    return {RISCVISD::VECREDUCE_UMAX, 0};
+    return RISCVISD::VECREDUCE_UMAX;
   case ISD::VECREDUCE_SMAX:
-    return {RISCVISD::VECREDUCE_SMAX, minIntN(EltSizeBits)};
+    return RISCVISD::VECREDUCE_SMAX;
   case ISD::VECREDUCE_UMIN:
-    return {RISCVISD::VECREDUCE_UMIN, maxUIntN(EltSizeBits)};
+    return RISCVISD::VECREDUCE_UMIN;
   case ISD::VECREDUCE_SMIN:
-    return {RISCVISD::VECREDUCE_SMIN, maxIntN(EltSizeBits)};
+    return RISCVISD::VECREDUCE_SMIN;
   case ISD::VECREDUCE_AND:
-    return {RISCVISD::VECREDUCE_AND, -1};
+    return RISCVISD::VECREDUCE_AND;
   case ISD::VECREDUCE_OR:
-    return {RISCVISD::VECREDUCE_OR, 0};
+    return RISCVISD::VECREDUCE_OR;
   case ISD::VECREDUCE_XOR:
-    return {RISCVISD::VECREDUCE_XOR, 0};
+    return RISCVISD::VECREDUCE_XOR;
   }
 }
 
@@ -2422,13 +2421,11 @@
          "Unexpected vector-reduce lowering");
   MVT VecVT = Op.getOperand(0).getSimpleValueType();
   MVT VecEltVT = VecVT.getVectorElementType();
-  unsigned RVVOpcode;
-  uint64_t IdentityVal;
-  std::tie(RVVOpcode, IdentityVal) =
-      getRVVReductionOpAndIdentityVal(Op.getOpcode(), VecEltVT.getSizeInBits());
+  unsigned RVVOpcode = getRVVReductionOp(Op.getOpcode());
   MVT M1VT = getLMUL1VT(VecVT);
-  SDValue IdentitySplat =
-      DAG.getSplatVector(M1VT, DL, DAG.getConstant(IdentityVal, DL, VecEltVT));
+  SDValue NeutralElem = DAG.getNeutralElement(
+      ISD::getVecReduceBaseOpcode(Op.getOpcode()), DL, VecEltVT, SDNodeFlags());
+  SDValue IdentitySplat = DAG.getSplatVector(M1VT, DL, NeutralElem);
   SDValue Reduction =
       DAG.getNode(RVVOpcode, DL, M1VT, Op.getOperand(0), IdentitySplat);
   SDValue Elt0 = DAG.getNode(ISD::EXTRACT_VECTOR_ELT, DL, VecEltVT, Reduction,


-------------- next part --------------
A non-text attachment was scrubbed...
Name: D98091.328828.patch
Type: text/x-patch
Size: 2427 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20210307/4ac37f46/attachment.bin>