Fix

Author: kalxr

llvm/lib/Target/NVPTX/NVPTXISelDAGToDAG.cpp

@@ -2478,22 +2478,32 @@ bool NVPTXDAGToDAGISel::SelectDirectAddr(SDValue N, SDValue &Address) {
   return false;
 }
 
-// symbol+offset
-bool NVPTXDAGToDAGISel::SelectADDRsi_imp(
-    SDNode *OpNode, SDValue Addr, SDValue &Base, SDValue &Offset, MVT mvt) {
+bool NVPTXDAGToDAGISel::FindRootAddressAndTotalOffset(
+    SDValue Addr, SDValue &Base, uint64_t &AccumulatedOffset) {
   if (isAddLike(Addr)) {
     if (ConstantSDNode *CN = dyn_cast<ConstantSDNode>(Addr.getOperand(1))) {
       SDValue base = Addr.getOperand(0);
-      if (SelectDirectAddr(base, Base)) {
-        Offset = CurDAG->getTargetConstant(CN->getZExtValue(), SDLoc(OpNode),
-                                           mvt);
+      AccumulatedOffset += CN->getZExtValue();
+      if (SelectDirectAddr(base, Base))
         return true;
-      }
+      return FindRootAddressAndTotalOffset(base, Base, AccumulatedOffset);
     }
   }
   return false;
 }
 
+// symbol+offset
+bool NVPTXDAGToDAGISel::SelectADDRsi_imp(SDNode *OpNode, SDValue Addr,
+                                         SDValue &Base, SDValue &Offset,
+                                         MVT mvt) {
+  uint64_t AccumulatedOffset = 0;
+  if (FindRootAddressAndTotalOffset(Addr, Base, AccumulatedOffset)) {
+    Offset = CurDAG->getTargetConstant(AccumulatedOffset, SDLoc(OpNode), mvt);
+    return true;
+  }
+  return false;
+}
+
 // symbol+offset
 bool NVPTXDAGToDAGISel::SelectADDRsi(SDNode *OpNode, SDValue Addr,
                                      SDValue &Base, SDValue &Offset) {

llvm/lib/Target/NVPTX/NVPTXISelDAGToDAG.h

@@ -97,6 +97,8 @@ class LLVM_LIBRARY_VISIBILITY NVPTXDAGToDAGISel : public SelectionDAGISel {
   void SelectCpAsyncBulkTensorPrefetchCommon(SDNode *N, bool IsIm2Col = false);
   void SelectCpAsyncBulkTensorReduceCommon(SDNode *N, unsigned RedOp,
                                            bool IsIm2Col = false);
+  bool FindRootAddressAndTotalOffset(SDValue Addr, SDValue &Base,
+                                     uint64_t &AccumulatedOffset);
 
   inline SDValue getI32Imm(unsigned Imm, const SDLoc &DL) {
     return CurDAG->getTargetConstant(Imm, DL, MVT::i32);

llvm/test/CodeGen/NVPTX/param-add.ll

@@ -0,0 +1,44 @@
+; RUN: llc < %s -march=nvptx64 --debug-counter=dagcombine=0 | FileCheck %s
+; RUN: %if ptxas %{ llc < %s -march=nvptx64 | %ptxas-verify %}
+
+%struct.8float = type <{ [8 x float] }>
+
+declare i32 @callee(%struct.8float %a)
+
+define i32 @test(%struct.8float alignstack(32) %data) {
+  ;CHECK-NOT: add.
+  ;CHECK-DAG: ld.param.u8 %r{{.*}}, [test_param_0];
+  ;CHECK-DAG: ld.param.u8 %r{{.*}}, [test_param_0+1];
+  ;CHECK-DAG: ld.param.u8 %r{{.*}}, [test_param_0+2];
+  ;CHECK-DAG: ld.param.u8 %r{{.*}}, [test_param_0+3];
+  ;CHECK-DAG: ld.param.u8 %r{{.*}}, [test_param_0+4];
+  ;CHECK-DAG: ld.param.u8 %r{{.*}}, [test_param_0+5];
+  ;CHECK-DAG: ld.param.u8 %r{{.*}}, [test_param_0+6];
+  ;CHECK-DAG: ld.param.u8 %r{{.*}}, [test_param_0+7];
+  ;CHECK-DAG: ld.param.u8 %r{{.*}}, [test_param_0+8];
+  ;CHECK-DAG: ld.param.u8 %r{{.*}}, [test_param_0+9];
+  ;CHECK-DAG: ld.param.u8 %r{{.*}}, [test_param_0+10];
+  ;CHECK-DAG: ld.param.u8 %r{{.*}}, [test_param_0+11];
+  ;CHECK-DAG: ld.param.u8 %r{{.*}}, [test_param_0+12];
+  ;CHECK-DAG: ld.param.u8 %r{{.*}}, [test_param_0+13];
+  ;CHECK-DAG: ld.param.u8 %r{{.*}}, [test_param_0+14];
+  ;CHECK-DAG: ld.param.u8 %r{{.*}}, [test_param_0+15];
+  ;CHECK-DAG: ld.param.u8 %r{{.*}}, [test_param_0+16];
+  ;CHECK-DAG: ld.param.u8 %r{{.*}}, [test_param_0+17];
+  ;CHECK-DAG: ld.param.u8 %r{{.*}}, [test_param_0+18];
+  ;CHECK-DAG: ld.param.u8 %r{{.*}}, [test_param_0+19];
+  ;CHECK-DAG: ld.param.u8 %r{{.*}}, [test_param_0+20];
+  ;CHECK-DAG: ld.param.u8 %r{{.*}}, [test_param_0+21];
+  ;CHECK-DAG: ld.param.u8 %r{{.*}}, [test_param_0+22];
+  ;CHECK-DAG: ld.param.u8 %r{{.*}}, [test_param_0+23];
+  ;CHECK-DAG: ld.param.u8 %r{{.*}}, [test_param_0+24];
+  ;CHECK-DAG: ld.param.u8 %r{{.*}}, [test_param_0+26];
+  ;CHECK-DAG: ld.param.u8 %r{{.*}}, [test_param_0+27];
+  ;CHECK-DAG: ld.param.u8 %r{{.*}}, [test_param_0+28];
+  ;CHECK-DAG: ld.param.u8 %r{{.*}}, [test_param_0+29];
+  ;CHECK-DAG: ld.param.u8 %r{{.*}}, [test_param_0+30];
+  ;CHECK-DAG: ld.param.u8 %r{{.*}}, [test_param_0+31];
+
+  %1 = call i32 @callee(%struct.8float %data)
+  ret i32 %1
+}