[AArch64] Support commuted operands in performFlagSettingCombine (llvm#162496)

Fixes llvm#122624.

Assisted-by: gpt-5-codex

Author: c-rhodes

llvm/include/llvm/CodeGen/SelectionDAG.h

@@ -1850,9 +1850,11 @@ class SelectionDAG {
   /// Get the specified node if it's already available, or else return NULL.
   LLVM_ABI SDNode *getNodeIfExists(unsigned Opcode, SDVTList VTList,
                                    ArrayRef<SDValue> Ops,
-                                   const SDNodeFlags Flags);
+                                   const SDNodeFlags Flags,
+                                   bool AllowCommute = false);
   LLVM_ABI SDNode *getNodeIfExists(unsigned Opcode, SDVTList VTList,
-                                   ArrayRef<SDValue> Ops);
+                                   ArrayRef<SDValue> Ops,
+                                   bool AllowCommute = false);
 
   /// Check if a node exists without modifying its flags.
   LLVM_ABI bool doesNodeExist(unsigned Opcode, SDVTList VTList,

llvm/lib/CodeGen/SelectionDAG/SelectionDAG.cpp

@@ -11844,25 +11844,38 @@ SDValue SelectionDAG::getTargetInsertSubreg(int SRIdx, const SDLoc &DL, EVT VT,
 /// getNodeIfExists - Get the specified node if it's already available, or
 /// else return NULL.
 SDNode *SelectionDAG::getNodeIfExists(unsigned Opcode, SDVTList VTList,
-                                      ArrayRef<SDValue> Ops) {
+                                      ArrayRef<SDValue> Ops,
+                                      bool AllowCommute) {
   SDNodeFlags Flags;
   if (Inserter)
     Flags = Inserter->getFlags();
-  return getNodeIfExists(Opcode, VTList, Ops, Flags);
+  return getNodeIfExists(Opcode, VTList, Ops, Flags, AllowCommute);
 }
 
 SDNode *SelectionDAG::getNodeIfExists(unsigned Opcode, SDVTList VTList,
                                       ArrayRef<SDValue> Ops,
-                                      const SDNodeFlags Flags) {
-  if (VTList.VTs[VTList.NumVTs - 1] != MVT::Glue) {
+                                      const SDNodeFlags Flags,
+                                      bool AllowCommute) {
+  if (VTList.VTs[VTList.NumVTs - 1] == MVT::Glue)
+    return nullptr;
+
+  auto Lookup = [&](ArrayRef<SDValue> LookupOps) -> SDNode * {
     FoldingSetNodeID ID;
-    AddNodeIDNode(ID, Opcode, VTList, Ops);
+    AddNodeIDNode(ID, Opcode, VTList, LookupOps);
     void *IP = nullptr;
-    if (SDNode *E = FindNodeOrInsertPos(ID, SDLoc(), IP)) {
+    if (SDNode *E = FindNodeOrInsertPos(ID, IP)) {
       E->intersectFlagsWith(Flags);
       return E;
     }
-  }
+    return nullptr;
+  };
+
+  if (SDNode *Existing = Lookup(Ops))
+    return Existing;
+
+  if (AllowCommute && TLI->isCommutativeBinOp(Opcode))
+    return Lookup({Ops[1], Ops[0]});
+
   return nullptr;
 }
 

llvm/lib/Target/AArch64/AArch64ISelLowering.cpp

@@ -26196,9 +26196,10 @@ static SDValue performFlagSettingCombine(SDNode *N,
     return DCI.CombineTo(N, Res, SDValue(N, 1));
   }
 
-  // Combine identical generic nodes into this node, re-using the result.
+  // Combine equivalent generic nodes into this node, re-using the result.
   if (SDNode *Generic = DCI.DAG.getNodeIfExists(
-          GenericOpcode, DCI.DAG.getVTList(VT), {LHS, RHS}))
+          GenericOpcode, DCI.DAG.getVTList(VT), {LHS, RHS},
+          /*AllowCommute=*/true))
     DCI.CombineTo(Generic, SDValue(N, 0));
 
   return SDValue();

llvm/test/CodeGen/AArch64/adds_cmn.ll

@@ -22,10 +22,8 @@ entry:
 define { i32, i32 } @adds_cmn_c(i32 noundef %x, i32 noundef %y) {
 ; CHECK-LABEL: adds_cmn_c:
 ; CHECK:       // %bb.0: // %entry
-; CHECK-NEXT:    cmn w0, w1
-; CHECK-NEXT:    add w1, w1, w0
-; CHECK-NEXT:    cset w8, lo
-; CHECK-NEXT:    mov w0, w8
+; CHECK-NEXT:    adds w1, w0, w1
+; CHECK-NEXT:    cset w0, lo
 ; CHECK-NEXT:    ret
 entry:
   %0 = tail call { i32, i1 } @llvm.uadd.with.overflow.i32(i32 %x, i32 %y)

llvm/test/CodeGen/AArch64/sat-add.ll

@@ -290,8 +290,7 @@ define i32 @unsigned_sat_variable_i32_using_cmp_sum(i32 %x, i32 %y) {
 define i32 @unsigned_sat_variable_i32_using_cmp_notval(i32 %x, i32 %y) {
 ; CHECK-LABEL: unsigned_sat_variable_i32_using_cmp_notval:
 ; CHECK:       // %bb.0:
-; CHECK-NEXT:    add w8, w0, w1
-; CHECK-NEXT:    cmn w1, w0
+; CHECK-NEXT:    adds w8, w1, w0
 ; CHECK-NEXT:    csinv w0, w8, wzr, lo
 ; CHECK-NEXT:    ret
   %noty = xor i32 %y, -1
@@ -331,8 +330,7 @@ define i64 @unsigned_sat_variable_i64_using_cmp_sum(i64 %x, i64 %y) {
 define i64 @unsigned_sat_variable_i64_using_cmp_notval(i64 %x, i64 %y) {
 ; CHECK-LABEL: unsigned_sat_variable_i64_using_cmp_notval:
 ; CHECK:       // %bb.0:
-; CHECK-NEXT:    add x8, x0, x1
-; CHECK-NEXT:    cmn x1, x0
+; CHECK-NEXT:    adds x8, x1, x0
 ; CHECK-NEXT:    csinv x0, x8, xzr, lo
 ; CHECK-NEXT:    ret
   %noty = xor i64 %y, -1