[DAGCombiner][RISCV] CSE zext nneg and sext. (#82597)

If we have a sext and a zext nneg with the same types and operand
we should combine them into the sext. We can't go the other way
because the nneg flag may only be valid in the context of the uses
of the zext nneg.

2 files changed, 30 insertions, 46 deletions

diff --git a/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp b/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp
index 89ef648..ed43dd7 100644
--- a/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp
+++ b/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp
@@ -13997,6 +13997,13 @@ SDValue DAGCombiner::visitZERO_EXTEND(SDNode *N) {
   if (SDValue Res = tryToFoldExtendSelectLoad(N, TLI, DAG, Level))
     return Res;
 
+  // CSE zext nneg with sext if the zext is not free.
+  if (N->getFlags().hasNonNeg() && !TLI.isZExtFree(N0.getValueType(), VT)) {
+    SDNode *CSENode = DAG.getNodeIfExists(ISD::SIGN_EXTEND, N->getVTList(), N0);
+    if (CSENode)
+      return SDValue(CSENode, 0);
+  }
+
   return SDValue();
 }
 
diff --git a/llvm/test/CodeGen/RISCV/sext-zext-trunc.ll b/llvm/test/CodeGen/RISCV/sext-zext-trunc.ll
index 09516d9..87f2a63 100644
--- a/llvm/test/CodeGen/RISCV/sext-zext-trunc.ll
+++ b/llvm/test/CodeGen/RISCV/sext-zext-trunc.ll
@@ -882,11 +882,10 @@ define void @load_zext_nneg_sext_cse(ptr %p) nounwind {
 ; RV32I-NEXT:    addi sp, sp, -16
 ; RV32I-NEXT:    sw ra, 12(sp) # 4-byte Folded Spill
 ; RV32I-NEXT:    sw s0, 8(sp) # 4-byte Folded Spill
-; RV32I-NEXT:    lhu s0, 0(a0)
-; RV32I-NEXT:    slli a0, s0, 16
-; RV32I-NEXT:    bltz a0, .LBB50_2
+; RV32I-NEXT:    lh s0, 0(a0)
+; RV32I-NEXT:    bltz s0, .LBB50_2
 ; RV32I-NEXT:  # %bb.1: # %bb1
-; RV32I-NEXT:    srai a0, a0, 16
+; RV32I-NEXT:    mv a0, s0
 ; RV32I-NEXT:    call bar_i16
 ; RV32I-NEXT:    mv a0, s0
 ; RV32I-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
@@ -899,48 +898,26 @@ define void @load_zext_nneg_sext_cse(ptr %p) nounwind {
 ; RV32I-NEXT:    addi sp, sp, 16
 ; RV32I-NEXT:    ret
 ;
-; RV64I-LABEL: load_zext_nneg_sext_cse:
-; RV64I:       # %bb.0:
-; RV64I-NEXT:    addi sp, sp, -16
-; RV64I-NEXT:    sd ra, 8(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    sd s0, 0(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    lhu s0, 0(a0)
-; RV64I-NEXT:    slli a0, s0, 48
-; RV64I-NEXT:    bltz a0, .LBB50_2
-; RV64I-NEXT:  # %bb.1: # %bb1
-; RV64I-NEXT:    srai a0, a0, 48
-; RV64I-NEXT:    call bar_i16
-; RV64I-NEXT:    mv a0, s0
-; RV64I-NEXT:    ld ra, 8(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s0, 0(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    addi sp, sp, 16
-; RV64I-NEXT:    tail bar_i32
-; RV64I-NEXT:  .LBB50_2: # %bb2
-; RV64I-NEXT:    ld ra, 8(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    ld s0, 0(sp) # 8-byte Folded Reload
-; RV64I-NEXT:    addi sp, sp, 16
-; RV64I-NEXT:    ret
-;
-; RV64ZBB-LABEL: load_zext_nneg_sext_cse:
-; RV64ZBB:       # %bb.0:
-; RV64ZBB-NEXT:    addi sp, sp, -16
-; RV64ZBB-NEXT:    sd ra, 8(sp) # 8-byte Folded Spill
-; RV64ZBB-NEXT:    sd s0, 0(sp) # 8-byte Folded Spill
-; RV64ZBB-NEXT:    lhu s0, 0(a0)
-; RV64ZBB-NEXT:    sext.h a0, s0
-; RV64ZBB-NEXT:    bltz a0, .LBB50_2
-; RV64ZBB-NEXT:  # %bb.1: # %bb1
-; RV64ZBB-NEXT:    call bar_i16
-; RV64ZBB-NEXT:    mv a0, s0
-; RV64ZBB-NEXT:    ld ra, 8(sp) # 8-byte Folded Reload
-; RV64ZBB-NEXT:    ld s0, 0(sp) # 8-byte Folded Reload
-; RV64ZBB-NEXT:    addi sp, sp, 16
-; RV64ZBB-NEXT:    tail bar_i32
-; RV64ZBB-NEXT:  .LBB50_2: # %bb2
-; RV64ZBB-NEXT:    ld ra, 8(sp) # 8-byte Folded Reload
-; RV64ZBB-NEXT:    ld s0, 0(sp) # 8-byte Folded Reload
-; RV64ZBB-NEXT:    addi sp, sp, 16
-; RV64ZBB-NEXT:    ret
+; RV64-LABEL: load_zext_nneg_sext_cse:
+; RV64:       # %bb.0:
+; RV64-NEXT:    addi sp, sp, -16
+; RV64-NEXT:    sd ra, 8(sp) # 8-byte Folded Spill
+; RV64-NEXT:    sd s0, 0(sp) # 8-byte Folded Spill
+; RV64-NEXT:    lh s0, 0(a0)
+; RV64-NEXT:    bltz s0, .LBB50_2
+; RV64-NEXT:  # %bb.1: # %bb1
+; RV64-NEXT:    mv a0, s0
+; RV64-NEXT:    call bar_i16
+; RV64-NEXT:    mv a0, s0
+; RV64-NEXT:    ld ra, 8(sp) # 8-byte Folded Reload
+; RV64-NEXT:    ld s0, 0(sp) # 8-byte Folded Reload
+; RV64-NEXT:    addi sp, sp, 16
+; RV64-NEXT:    tail bar_i32
+; RV64-NEXT:  .LBB50_2: # %bb2
+; RV64-NEXT:    ld ra, 8(sp) # 8-byte Folded Reload
+; RV64-NEXT:    ld s0, 0(sp) # 8-byte Folded Reload
+; RV64-NEXT:    addi sp, sp, 16
+; RV64-NEXT:    ret
   %load = load i16, ptr %p
   %zext = zext nneg i16 %load to i32
   %cmp = icmp sgt i16 %load, -1