[PATCH] D120648: [RISCV] DAGcombine fold (bswap(srl (bswap c), x)) -> (shl c, x)

[Chenbing.Zheng via Phabricator via llvm-commits]

Chenbing.Zheng updated this revision to Diff 411788.
Chenbing.Zheng retitled this revision from "[RISCV] DAG combine bswap(srl (bswap t), x) to shl t, x" to "[RISCV] DAGcombine  fold (bswap(srl (bswap c), x)) -> (shl c, x)".
Chenbing.Zheng added a reviewer: RKSimon.
Chenbing.Zheng added a comment.
Herald added a subscriber: ecnelises.

address comments


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D120648/new/

https://reviews.llvm.org/D120648

Files:
  llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp
  llvm/test/CodeGen/RISCV/bswap-bitreverse.ll


Index: llvm/test/CodeGen/RISCV/bswap-bitreverse.ll
===================================================================
--- llvm/test/CodeGen/RISCV/bswap-bitreverse.ll
+++ llvm/test/CodeGen/RISCV/bswap-bitreverse.ll
@@ -848,18 +848,14 @@
 ;
 ; RV32ZBKB-LABEL: test_bswap_bitreverse_i16:
 ; RV32ZBKB:       # %bb.0:
-; RV32ZBKB-NEXT:    rev8 a0, a0
-; RV32ZBKB-NEXT:    srli a0, a0, 16
-; RV32ZBKB-NEXT:    rev8 a0, a0
+; RV32ZBKB-NEXT:    slli a0, a0, 16
 ; RV32ZBKB-NEXT:    brev8 a0, a0
 ; RV32ZBKB-NEXT:    srli a0, a0, 16
 ; RV32ZBKB-NEXT:    ret
 ;
 ; RV64ZBKB-LABEL: test_bswap_bitreverse_i16:
 ; RV64ZBKB:       # %bb.0:
-; RV64ZBKB-NEXT:    rev8 a0, a0
-; RV64ZBKB-NEXT:    srli a0, a0, 48
-; RV64ZBKB-NEXT:    rev8 a0, a0
+; RV64ZBKB-NEXT:    slli a0, a0, 48
 ; RV64ZBKB-NEXT:    brev8 a0, a0
 ; RV64ZBKB-NEXT:    srli a0, a0, 48
 ; RV64ZBKB-NEXT:    ret
@@ -977,9 +973,7 @@
 ;
 ; RV64ZBKB-LABEL: test_bswap_bitreverse_i32:
 ; RV64ZBKB:       # %bb.0:
-; RV64ZBKB-NEXT:    rev8 a0, a0
-; RV64ZBKB-NEXT:    srli a0, a0, 32
-; RV64ZBKB-NEXT:    rev8 a0, a0
+; RV64ZBKB-NEXT:    slli a0, a0, 32
 ; RV64ZBKB-NEXT:    brev8 a0, a0
 ; RV64ZBKB-NEXT:    srli a0, a0, 32
 ; RV64ZBKB-NEXT:    ret
@@ -1238,18 +1232,14 @@
 ;
 ; RV32ZBKB-LABEL: test_bitreverse_bswap_i16:
 ; RV32ZBKB:       # %bb.0:
-; RV32ZBKB-NEXT:    rev8 a0, a0
-; RV32ZBKB-NEXT:    srli a0, a0, 16
-; RV32ZBKB-NEXT:    rev8 a0, a0
+; RV32ZBKB-NEXT:    slli a0, a0, 16
 ; RV32ZBKB-NEXT:    brev8 a0, a0
 ; RV32ZBKB-NEXT:    srli a0, a0, 16
 ; RV32ZBKB-NEXT:    ret
 ;
 ; RV64ZBKB-LABEL: test_bitreverse_bswap_i16:
 ; RV64ZBKB:       # %bb.0:
-; RV64ZBKB-NEXT:    rev8 a0, a0
-; RV64ZBKB-NEXT:    srli a0, a0, 48
-; RV64ZBKB-NEXT:    rev8 a0, a0
+; RV64ZBKB-NEXT:    slli a0, a0, 48
 ; RV64ZBKB-NEXT:    brev8 a0, a0
 ; RV64ZBKB-NEXT:    srli a0, a0, 48
 ; RV64ZBKB-NEXT:    ret
@@ -1367,9 +1357,7 @@
 ;
 ; RV64ZBKB-LABEL: test_bitreverse_bswap_i32:
 ; RV64ZBKB:       # %bb.0:
-; RV64ZBKB-NEXT:    rev8 a0, a0
-; RV64ZBKB-NEXT:    srli a0, a0, 32
-; RV64ZBKB-NEXT:    rev8 a0, a0
+; RV64ZBKB-NEXT:    slli a0, a0, 32
 ; RV64ZBKB-NEXT:    brev8 a0, a0
 ; RV64ZBKB-NEXT:    srli a0, a0, 32
 ; RV64ZBKB-NEXT:    ret
Index: llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp
===================================================================
--- llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp
+++ llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp
@@ -9650,6 +9650,14 @@
   if (N0.getOpcode() == ISD::BSWAP)
     return N0.getOperand(0);
 
+  // fold (bswap(srl (bswap c), x)) -> (shl c, x)
+  if (N0->getOpcode() == ISD::SRL && N0.hasOneUse()) {
+    SDValue BSwap = N0->getOperand(0);
+    if (BSwap->getOpcode() == ISD::BSWAP && BSwap.hasOneUse())
+      return DAG.getNode(ISD::SHL, DL, VT, BSwap->getOperand(0),
+                         N0->getOperand(1));
+  }
+
   // Canonicalize bswap(bitreverse(x)) -> bitreverse(bswap(x)). If bitreverse
   // isn't supported, it will be expanded to bswap followed by a manual reversal
   // of bits in each byte. By placing bswaps before bitreverse, we can remove


-------------- next part --------------
A non-text attachment was scrubbed...
Name: D120648.411788.patch
Type: text/x-patch
Size: 3064 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20220228/9c86f90d/attachment.bin>