[RISCV] Custom legalize BITREVERSE with Zbkb.

[craig.topper]

Summary

With Zbkb, a bitreverse can be split into a rev8 and a brev8.

[craig.topper]

llvm/test/CodeGen/RISCV/bswap-bitreverse.ll

850

We need some more DAG combines to improve this.

[VincentWu]

llvm/test/CodeGen/RISCV/bswap-bitreverse.ll

850

We need some more DAG combines to improve this.

Will you continue to update this patch? or put it later

if put it later. what about add an // TODO: comment in the proper place.

[craig.topper]

Add FIXMEs to the test cases that need more work.

[craig.topper]

llvm/test/CodeGen/RISCV/bswap-bitreverse.ll

850

I'll do it another patch. It's likely going to need changes in a different file.

[alextsao1999]

llvm/lib/Target/RISCV/RISCVISelLowering.cpp

285

zbkp -> zbkb?

[craig.topper]

Fix comment mistake

[VincentWu]

LGTM

llvm/lib/Target/RISCV/RISCVISelLowering.h

 //===-- RISCVISelLowering.h - RISCV DAG Lowering Interface ------*- C++ -*-===//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 //===----------------------------------------------------------------------===//
 //
enum NodeType : unsigned {
   // result being sign extended to 64 bit. These saturate out of range inputs.
   // Used for FP_TO_S/UINT and FP_TO_S/UINT_SAT lowering. Rounding mode
   // is passed as a TargetConstant operand using the RISCVFPRndMode enum.
   FCVT_W_RV64,
   FCVT_WU_RV64,
   // READ_CYCLE_WIDE - A read of the 64-bit cycle CSR on a 32-bit target
   // (returns (Lo, Hi)). It takes a chain operand.
   READ_CYCLE_WIDE,
+  // Reverse bits in each byte.
+  BREV8,
   // Generalized Reverse and Generalized Or-Combine - directly matching the
   // semantics of the named RISC-V instructions. Lowered as custom nodes as
   // TableGen chokes when faced with commutative permutations in deeply-nested
   // DAGs. Each node takes an input operand and a control operand and outputs a
   // bit-manipulated version of input. All operands are i32 or XLenVT.
   GREV,
   GREVW,
   GORC,

llvm/lib/Target/RISCV/RISCVISelLowering.cpp

 //===-- RISCVISelLowering.cpp - RISCV DAG Lowering Implementation  --------===//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 //===----------------------------------------------------------------------===//
 //
if (Subtarget.hasStdExtZbp()) {
     }
   } else {
     // With Zbb we have an XLen rev8 instruction, but not GREVI. So we'll
     // pattern match it directly in isel.
     setOperationAction(ISD::BSWAP, XLenVT,
                        (Subtarget.hasStdExtZbb() || Subtarget.hasStdExtZbkb())
                            ? Legal
                            : Expand);
+    // Zbkp can use rev8+brev8 to implement bitreverse.
+    setOperationAction(ISD::BITREVERSE, XLenVT,
+                       Subtarget.hasStdExtZbkb() ? Custom : Expand);
   }
 
   if (Subtarget.hasStdExtZbb()) {
     setOperationAction(ISD::SMIN, XLenVT, Legal);
     setOperationAction(ISD::SMAX, XLenVT, Legal);
     setOperationAction(ISD::UMIN, XLenVT, Legal);
     setOperationAction(ISD::UMAX, XLenVT, Legal);
 
SDValue RISCVTargetLowering::LowerOperation(SDValue Op,
   case ISD::INTRINSIC_WO_CHAIN:
     return LowerINTRINSIC_WO_CHAIN(Op, DAG);
   case ISD::INTRINSIC_W_CHAIN:
     return LowerINTRINSIC_W_CHAIN(Op, DAG);
   case ISD::INTRINSIC_VOID:
     return LowerINTRINSIC_VOID(Op, DAG);
   case ISD::BSWAP:
   case ISD::BITREVERSE: {
-    // Convert BSWAP/BITREVERSE to GREVI to enable GREVI combinining.
-    assert(Subtarget.hasStdExtZbp() && "Unexpected custom legalisation");
     MVT VT = Op.getSimpleValueType();
     SDLoc DL(Op);
+    if (Subtarget.hasStdExtZbp()) {
+      // Convert BSWAP/BITREVERSE to GREVI to enable GREVI combinining.
       // Start with the maximum immediate value which is the bitwidth - 1.
       unsigned Imm = VT.getSizeInBits() - 1;
       // If this is BSWAP rather than BITREVERSE, clear the lower 3 bits.
       if (Op.getOpcode() == ISD::BSWAP)
         Imm &= ~0x7U;
       return DAG.getNode(RISCVISD::GREV, DL, VT, Op.getOperand(0),
                          DAG.getConstant(Imm, DL, VT));
     }
+    assert(Subtarget.hasStdExtZbkb() && "Unexpected custom legalization");
+    assert(Op.getOpcode() == ISD::BITREVERSE && "Unexpected opcode");
+    // Expand bitreverse to a bswap(rev8) followed by brev8.
+    SDValue BSwap = DAG.getNode(ISD::BSWAP, DL, VT, Op.getOperand(0));
+    return DAG.getNode(RISCVISD::BREV8, DL, VT, BSwap);
+  }
   case ISD::FSHL:
   case ISD::FSHR: {
     MVT VT = Op.getSimpleValueType();
     assert(VT == Subtarget.getXLenVT() && "Unexpected custom legalization");
     SDLoc DL(Op);
     // FSL/FSR take a log2(XLen)+1 bit shift amount but XLenVT FSHL/FSHR only
     // use log(XLen) bits. Mask the shift amount accordingly to prevent
     // accidentally setting the extra bit.
#define NODE_NAME_CASE(NODE)                                                   \
   NODE_NAME_CASE(FMV_X_ANYEXTW_RV64)
   NODE_NAME_CASE(FCVT_X)
   NODE_NAME_CASE(FCVT_XU)
   NODE_NAME_CASE(FCVT_W_RV64)
   NODE_NAME_CASE(FCVT_WU_RV64)
   NODE_NAME_CASE(STRICT_FCVT_W_RV64)
   NODE_NAME_CASE(STRICT_FCVT_WU_RV64)
   NODE_NAME_CASE(READ_CYCLE_WIDE)
+  NODE_NAME_CASE(BREV8)
   NODE_NAME_CASE(GREV)
   NODE_NAME_CASE(GREVW)
   NODE_NAME_CASE(GORC)
   NODE_NAME_CASE(GORCW)
   NODE_NAME_CASE(SHFL)
   NODE_NAME_CASE(SHFLW)
   NODE_NAME_CASE(UNSHFL)
   NODE_NAME_CASE(UNSHFLW)

llvm/lib/Target/RISCV/RISCVInstrInfoZb.td

 def riscv_clzw   : SDNode<"RISCVISD::CLZW",   SDT_RISCVIntUnaryOpW>;
 def riscv_ctzw   : SDNode<"RISCVISD::CTZW",   SDT_RISCVIntUnaryOpW>;
 def riscv_rolw   : SDNode<"RISCVISD::ROLW",   SDT_RISCVIntBinOpW>;
 def riscv_rorw   : SDNode<"RISCVISD::RORW",   SDT_RISCVIntBinOpW>;
 def riscv_fslw   : SDNode<"RISCVISD::FSLW",   SDT_RISCVIntShiftDOpW>;
 def riscv_fsrw   : SDNode<"RISCVISD::FSRW",   SDT_RISCVIntShiftDOpW>;
 def riscv_fsl    : SDNode<"RISCVISD::FSL",    SDTIntShiftDOp>;
 def riscv_fsr    : SDNode<"RISCVISD::FSR",    SDTIntShiftDOp>;
+def riscv_brev8  : SDNode<"RISCVISD::BREV8",  SDTIntUnaryOp>;
 def riscv_grev   : SDNode<"RISCVISD::GREV",   SDTIntBinOp>;
 def riscv_grevw  : SDNode<"RISCVISD::GREVW",  SDT_RISCVIntBinOpW>;
 def riscv_gorc   : SDNode<"RISCVISD::GORC",   SDTIntBinOp>;
 def riscv_gorcw  : SDNode<"RISCVISD::GORCW",  SDT_RISCVIntBinOpW>;
 def riscv_shfl   : SDNode<"RISCVISD::SHFL",   SDTIntBinOp>;
 def riscv_shflw  : SDNode<"RISCVISD::SHFLW",  SDT_RISCVIntBinOpW>;
 def riscv_unshfl : SDNode<"RISCVISD::UNSHFL", SDTIntBinOp>;
 def riscv_unshflw: SDNode<"RISCVISD::UNSHFLW",SDT_RISCVIntBinOpW>;

 
 let Predicates = [HasStdExtZbf] in
 def : PatGprGpr<riscv_bfp, BFP>;
 
 let Predicates = [HasStdExtZbf, IsRV64] in
 def : PatGprGpr<riscv_bfpw, BFPW>;
 
 let Predicates = [HasStdExtZbkb] in {
+def : PatGpr<riscv_brev8, BREV8>;
 def : PatGpr<int_riscv_brev8, BREV8>;
 } // Predicates = [HasStdExtZbkb]
 
 let Predicates = [HasStdExtZbkb, IsRV32] in {
 def : PatGpr<int_riscv_zip, ZIP_RV32>;
 def : PatGpr<int_riscv_unzip, UNZIP_RV32>;
 } // Predicates = [HasStdExtZbkb, IsRV32]
 
 let Predicates = [HasStdExtZbkx] in {
 def : PatGprGpr<int_riscv_xperm4, XPERMN>;
 def : PatGprGpr<int_riscv_xperm8, XPERMB>;
 }

llvm/test/CodeGen/RISCV/bswap-bitreverse.ll

 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
 ; RUN: llc -mtriple=riscv32 -verify-machineinstrs < %s \
 ; RUN:   | FileCheck %s -check-prefix=RV32I
 ; RUN: llc -mtriple=riscv64 -verify-machineinstrs < %s \
 ; RUN:   | FileCheck %s -check-prefix=RV64I
 ; RUN: llc -mtriple=riscv32 -mattr=+zbb -verify-machineinstrs < %s \
-; RUN:   | FileCheck %s -check-prefixes=RV32ZB
+; RUN:   | FileCheck %s -check-prefixes=RV32ZB,RV32ZBB
 ; RUN: llc -mtriple=riscv64 -mattr=+zbb -verify-machineinstrs < %s \
-; RUN:   | FileCheck %s -check-prefixes=RV64ZB
+; RUN:   | FileCheck %s -check-prefixes=RV64ZB,RV64ZBB
 ; RUN: llc -mtriple=riscv32 -mattr=+zbkb -verify-machineinstrs < %s \
-; RUN:   | FileCheck %s -check-prefixes=RV32ZB
+; RUN:   | FileCheck %s -check-prefixes=RV32ZB,RV32ZBKB
 ; RUN: llc -mtriple=riscv64 -mattr=+zbkb -verify-machineinstrs < %s \
-; RUN:   | FileCheck %s -check-prefixes=RV64ZB
+; RUN:   | FileCheck %s -check-prefixes=RV64ZB,RV64ZBKB
 
 declare i16 @llvm.bswap.i16(i16)
 declare i32 @llvm.bswap.i32(i32)
 declare i64 @llvm.bswap.i64(i64)
 declare i8 @llvm.bitreverse.i8(i8)
 declare i16 @llvm.bitreverse.i16(i16)
 declare i32 @llvm.bitreverse.i32(i32)
 declare i64 @llvm.bitreverse.i64(i64)

 ; RV64I-NEXT:    or a0, a0, a1
 ; RV64I-NEXT:    andi a1, a0, 85
 ; RV64I-NEXT:    slli a1, a1, 1
 ; RV64I-NEXT:    srli a0, a0, 1
 ; RV64I-NEXT:    andi a0, a0, 85
 ; RV64I-NEXT:    or a0, a0, a1
 ; RV64I-NEXT:    ret
 ;
-; RV32ZB-LABEL: test_bitreverse_i8:
-; RV32ZB:       # %bb.0:
-; RV32ZB-NEXT:    andi a1, a0, 15
-; RV32ZB-NEXT:    slli a1, a1, 4
-; RV32ZB-NEXT:    slli a0, a0, 24
-; RV32ZB-NEXT:    srli a0, a0, 28
-; RV32ZB-NEXT:    or a0, a0, a1
-; RV32ZB-NEXT:    andi a1, a0, 51
-; RV32ZB-NEXT:    slli a1, a1, 2
-; RV32ZB-NEXT:    srli a0, a0, 2
-; RV32ZB-NEXT:    andi a0, a0, 51
-; RV32ZB-NEXT:    or a0, a0, a1
-; RV32ZB-NEXT:    andi a1, a0, 85
-; RV32ZB-NEXT:    slli a1, a1, 1
-; RV32ZB-NEXT:    srli a0, a0, 1
-; RV32ZB-NEXT:    andi a0, a0, 85
-; RV32ZB-NEXT:    or a0, a0, a1
-; RV32ZB-NEXT:    ret
-;
-; RV64ZB-LABEL: test_bitreverse_i8:
-; RV64ZB:       # %bb.0:
-; RV64ZB-NEXT:    andi a1, a0, 15
-; RV64ZB-NEXT:    slli a1, a1, 4
-; RV64ZB-NEXT:    slli a0, a0, 56
-; RV64ZB-NEXT:    srli a0, a0, 60
-; RV64ZB-NEXT:    or a0, a0, a1
-; RV64ZB-NEXT:    andi a1, a0, 51
-; RV64ZB-NEXT:    slli a1, a1, 2
-; RV64ZB-NEXT:    srli a0, a0, 2
-; RV64ZB-NEXT:    andi a0, a0, 51
-; RV64ZB-NEXT:    or a0, a0, a1
-; RV64ZB-NEXT:    andi a1, a0, 85
-; RV64ZB-NEXT:    slli a1, a1, 1
-; RV64ZB-NEXT:    srli a0, a0, 1
-; RV64ZB-NEXT:    andi a0, a0, 85
-; RV64ZB-NEXT:    or a0, a0, a1
-; RV64ZB-NEXT:    ret
+; RV32ZBB-LABEL: test_bitreverse_i8:
+; RV32ZBB:       # %bb.0:
+; RV32ZBB-NEXT:    andi a1, a0, 15
+; RV32ZBB-NEXT:    slli a1, a1, 4
+; RV32ZBB-NEXT:    slli a0, a0, 24
+; RV32ZBB-NEXT:    srli a0, a0, 28
+; RV32ZBB-NEXT:    or a0, a0, a1
+; RV32ZBB-NEXT:    andi a1, a0, 51
+; RV32ZBB-NEXT:    slli a1, a1, 2
+; RV32ZBB-NEXT:    srli a0, a0, 2
+; RV32ZBB-NEXT:    andi a0, a0, 51
+; RV32ZBB-NEXT:    or a0, a0, a1
+; RV32ZBB-NEXT:    andi a1, a0, 85
+; RV32ZBB-NEXT:    slli a1, a1, 1
+; RV32ZBB-NEXT:    srli a0, a0, 1
+; RV32ZBB-NEXT:    andi a0, a0, 85
+; RV32ZBB-NEXT:    or a0, a0, a1
+; RV32ZBB-NEXT:    ret
+;
+; RV64ZBB-LABEL: test_bitreverse_i8:
+; RV64ZBB:       # %bb.0:
+; RV64ZBB-NEXT:    andi a1, a0, 15
+; RV64ZBB-NEXT:    slli a1, a1, 4
+; RV64ZBB-NEXT:    slli a0, a0, 56
+; RV64ZBB-NEXT:    srli a0, a0, 60
+; RV64ZBB-NEXT:    or a0, a0, a1
+; RV64ZBB-NEXT:    andi a1, a0, 51
+; RV64ZBB-NEXT:    slli a1, a1, 2
+; RV64ZBB-NEXT:    srli a0, a0, 2
+; RV64ZBB-NEXT:    andi a0, a0, 51
+; RV64ZBB-NEXT:    or a0, a0, a1
+; RV64ZBB-NEXT:    andi a1, a0, 85
+; RV64ZBB-NEXT:    slli a1, a1, 1
+; RV64ZBB-NEXT:    srli a0, a0, 1
+; RV64ZBB-NEXT:    andi a0, a0, 85
+; RV64ZBB-NEXT:    or a0, a0, a1
+; RV64ZBB-NEXT:    ret
+;
+; RV32ZBKB-LABEL: test_bitreverse_i8:
+; RV32ZBKB:       # %bb.0:
+; RV32ZBKB-NEXT:    rev8 a0, a0
+; RV32ZBKB-NEXT:    brev8 a0, a0
+; RV32ZBKB-NEXT:    srli a0, a0, 24
+; RV32ZBKB-NEXT:    ret
+;
+; RV64ZBKB-LABEL: test_bitreverse_i8:
+; RV64ZBKB:       # %bb.0:
+; RV64ZBKB-NEXT:    rev8 a0, a0
+; RV64ZBKB-NEXT:    brev8 a0, a0
+; RV64ZBKB-NEXT:    srli a0, a0, 56
+; RV64ZBKB-NEXT:    ret
   %tmp = call i8 @llvm.bitreverse.i8(i8 %a)
   ret i8 %tmp
 }
 
 define i16 @test_bitreverse_i16(i16 %a) nounwind {
 ; RV32I-LABEL: test_bitreverse_i16:
 ; RV32I:       # %bb.0:
 ; RV32I-NEXT:    slli a1, a0, 8

 ; RV64I-NEXT:    lui a2, 5
 ; RV64I-NEXT:    addiw a2, a2, 1365
 ; RV64I-NEXT:    and a1, a1, a2
 ; RV64I-NEXT:    and a0, a0, a2
 ; RV64I-NEXT:    slli a0, a0, 1
 ; RV64I-NEXT:    or a0, a1, a0
 ; RV64I-NEXT:    ret
 ;
-; RV32ZB-LABEL: test_bitreverse_i16:
-; RV32ZB:       # %bb.0:
-; RV32ZB-NEXT:    rev8 a0, a0
-; RV32ZB-NEXT:    srli a1, a0, 12
-; RV32ZB-NEXT:    lui a2, 15
-; RV32ZB-NEXT:    addi a2, a2, 240
-; RV32ZB-NEXT:    and a1, a1, a2
-; RV32ZB-NEXT:    srli a0, a0, 20
-; RV32ZB-NEXT:    andi a0, a0, -241
-; RV32ZB-NEXT:    or a0, a0, a1
-; RV32ZB-NEXT:    srli a1, a0, 2
-; RV32ZB-NEXT:    lui a2, 3
-; RV32ZB-NEXT:    addi a2, a2, 819
-; RV32ZB-NEXT:    and a1, a1, a2
-; RV32ZB-NEXT:    and a0, a0, a2
-; RV32ZB-NEXT:    slli a0, a0, 2
-; RV32ZB-NEXT:    or a0, a1, a0
-; RV32ZB-NEXT:    srli a1, a0, 1
-; RV32ZB-NEXT:    lui a2, 5
-; RV32ZB-NEXT:    addi a2, a2, 1365
-; RV32ZB-NEXT:    and a1, a1, a2
-; RV32ZB-NEXT:    and a0, a0, a2
-; RV32ZB-NEXT:    slli a0, a0, 1
-; RV32ZB-NEXT:    or a0, a1, a0
-; RV32ZB-NEXT:    ret
-;
-; RV64ZB-LABEL: test_bitreverse_i16:
-; RV64ZB:       # %bb.0:
-; RV64ZB-NEXT:    rev8 a0, a0
-; RV64ZB-NEXT:    srli a1, a0, 44
-; RV64ZB-NEXT:    lui a2, 15
-; RV64ZB-NEXT:    addiw a2, a2, 240
-; RV64ZB-NEXT:    and a1, a1, a2
-; RV64ZB-NEXT:    srli a0, a0, 52
-; RV64ZB-NEXT:    andi a0, a0, -241
-; RV64ZB-NEXT:    or a0, a0, a1
-; RV64ZB-NEXT:    srli a1, a0, 2
-; RV64ZB-NEXT:    lui a2, 3
-; RV64ZB-NEXT:    addiw a2, a2, 819
-; RV64ZB-NEXT:    and a1, a1, a2
-; RV64ZB-NEXT:    and a0, a0, a2
-; RV64ZB-NEXT:    slli a0, a0, 2
-; RV64ZB-NEXT:    or a0, a1, a0
-; RV64ZB-NEXT:    srli a1, a0, 1
-; RV64ZB-NEXT:    lui a2, 5
-; RV64ZB-NEXT:    addiw a2, a2, 1365
-; RV64ZB-NEXT:    and a1, a1, a2
-; RV64ZB-NEXT:    and a0, a0, a2
-; RV64ZB-NEXT:    slli a0, a0, 1
-; RV64ZB-NEXT:    or a0, a1, a0
-; RV64ZB-NEXT:    ret
+; RV32ZBB-LABEL: test_bitreverse_i16:
+; RV32ZBB:       # %bb.0:
+; RV32ZBB-NEXT:    rev8 a0, a0
+; RV32ZBB-NEXT:    srli a1, a0, 12
+; RV32ZBB-NEXT:    lui a2, 15
+; RV32ZBB-NEXT:    addi a2, a2, 240
+; RV32ZBB-NEXT:    and a1, a1, a2
+; RV32ZBB-NEXT:    srli a0, a0, 20
+; RV32ZBB-NEXT:    andi a0, a0, -241
+; RV32ZBB-NEXT:    or a0, a0, a1
+; RV32ZBB-NEXT:    srli a1, a0, 2
+; RV32ZBB-NEXT:    lui a2, 3
+; RV32ZBB-NEXT:    addi a2, a2, 819
+; RV32ZBB-NEXT:    and a1, a1, a2
+; RV32ZBB-NEXT:    and a0, a0, a2
+; RV32ZBB-NEXT:    slli a0, a0, 2
+; RV32ZBB-NEXT:    or a0, a1, a0
+; RV32ZBB-NEXT:    srli a1, a0, 1
+; RV32ZBB-NEXT:    lui a2, 5
+; RV32ZBB-NEXT:    addi a2, a2, 1365
+; RV32ZBB-NEXT:    and a1, a1, a2
+; RV32ZBB-NEXT:    and a0, a0, a2
+; RV32ZBB-NEXT:    slli a0, a0, 1
+; RV32ZBB-NEXT:    or a0, a1, a0
+; RV32ZBB-NEXT:    ret
+;
+; RV64ZBB-LABEL: test_bitreverse_i16:
+; RV64ZBB:       # %bb.0:
+; RV64ZBB-NEXT:    rev8 a0, a0
+; RV64ZBB-NEXT:    srli a1, a0, 44
+; RV64ZBB-NEXT:    lui a2, 15
+; RV64ZBB-NEXT:    addiw a2, a2, 240
+; RV64ZBB-NEXT:    and a1, a1, a2
+; RV64ZBB-NEXT:    srli a0, a0, 52
+; RV64ZBB-NEXT:    andi a0, a0, -241
+; RV64ZBB-NEXT:    or a0, a0, a1
+; RV64ZBB-NEXT:    srli a1, a0, 2
+; RV64ZBB-NEXT:    lui a2, 3
+; RV64ZBB-NEXT:    addiw a2, a2, 819
+; RV64ZBB-NEXT:    and a1, a1, a2
+; RV64ZBB-NEXT:    and a0, a0, a2
+; RV64ZBB-NEXT:    slli a0, a0, 2
+; RV64ZBB-NEXT:    or a0, a1, a0
+; RV64ZBB-NEXT:    srli a1, a0, 1
+; RV64ZBB-NEXT:    lui a2, 5
+; RV64ZBB-NEXT:    addiw a2, a2, 1365
+; RV64ZBB-NEXT:    and a1, a1, a2
+; RV64ZBB-NEXT:    and a0, a0, a2
+; RV64ZBB-NEXT:    slli a0, a0, 1
+; RV64ZBB-NEXT:    or a0, a1, a0
+; RV64ZBB-NEXT:    ret
+;
+; RV32ZBKB-LABEL: test_bitreverse_i16:
+; RV32ZBKB:       # %bb.0:
+; RV32ZBKB-NEXT:    rev8 a0, a0
+; RV32ZBKB-NEXT:    brev8 a0, a0
+; RV32ZBKB-NEXT:    srli a0, a0, 16
+; RV32ZBKB-NEXT:    ret
+;
+; RV64ZBKB-LABEL: test_bitreverse_i16:
+; RV64ZBKB:       # %bb.0:
+; RV64ZBKB-NEXT:    rev8 a0, a0
+; RV64ZBKB-NEXT:    brev8 a0, a0
+; RV64ZBKB-NEXT:    srli a0, a0, 48
+; RV64ZBKB-NEXT:    ret
   %tmp = call i16 @llvm.bitreverse.i16(i16 %a)
   ret i16 %tmp
 }
 
 define i32 @test_bitreverse_i32(i32 %a) nounwind {
 ; RV32I-LABEL: test_bitreverse_i32:
 ; RV32I:       # %bb.0:
 ; RV32I-NEXT:    srli a1, a0, 8

 ; RV64I-NEXT:    lui a2, 349525
 ; RV64I-NEXT:    addiw a2, a2, 1365
 ; RV64I-NEXT:    and a1, a1, a2
 ; RV64I-NEXT:    and a0, a0, a2
 ; RV64I-NEXT:    slliw a0, a0, 1
 ; RV64I-NEXT:    or a0, a1, a0
 ; RV64I-NEXT:    ret
 ;
-; RV32ZB-LABEL: test_bitreverse_i32:
-; RV32ZB:       # %bb.0:
-; RV32ZB-NEXT:    rev8 a0, a0
-; RV32ZB-NEXT:    srli a1, a0, 4
-; RV32ZB-NEXT:    lui a2, 61681
-; RV32ZB-NEXT:    addi a2, a2, -241
-; RV32ZB-NEXT:    and a1, a1, a2
-; RV32ZB-NEXT:    and a0, a0, a2
-; RV32ZB-NEXT:    slli a0, a0, 4
-; RV32ZB-NEXT:    or a0, a1, a0
-; RV32ZB-NEXT:    srli a1, a0, 2
-; RV32ZB-NEXT:    lui a2, 209715
-; RV32ZB-NEXT:    addi a2, a2, 819
-; RV32ZB-NEXT:    and a1, a1, a2
-; RV32ZB-NEXT:    and a0, a0, a2
-; RV32ZB-NEXT:    slli a0, a0, 2
-; RV32ZB-NEXT:    or a0, a1, a0
-; RV32ZB-NEXT:    srli a1, a0, 1
-; RV32ZB-NEXT:    lui a2, 349525
-; RV32ZB-NEXT:    addi a2, a2, 1365
-; RV32ZB-NEXT:    and a1, a1, a2
-; RV32ZB-NEXT:    and a0, a0, a2
-; RV32ZB-NEXT:    slli a0, a0, 1
-; RV32ZB-NEXT:    or a0, a1, a0
-; RV32ZB-NEXT:    ret
-;
-; RV64ZB-LABEL: test_bitreverse_i32:
-; RV64ZB:       # %bb.0:
-; RV64ZB-NEXT:    rev8 a0, a0
-; RV64ZB-NEXT:    srli a1, a0, 36
-; RV64ZB-NEXT:    lui a2, 61681
-; RV64ZB-NEXT:    addiw a2, a2, -241
-; RV64ZB-NEXT:    and a1, a1, a2
-; RV64ZB-NEXT:    srli a0, a0, 28
-; RV64ZB-NEXT:    lui a2, 986895
-; RV64ZB-NEXT:    addiw a2, a2, 240
-; RV64ZB-NEXT:    and a0, a0, a2
-; RV64ZB-NEXT:    sext.w a0, a0
-; RV64ZB-NEXT:    or a0, a1, a0
-; RV64ZB-NEXT:    srli a1, a0, 2
-; RV64ZB-NEXT:    lui a2, 209715
-; RV64ZB-NEXT:    addiw a2, a2, 819
-; RV64ZB-NEXT:    and a1, a1, a2
-; RV64ZB-NEXT:    and a0, a0, a2
-; RV64ZB-NEXT:    slliw a0, a0, 2
-; RV64ZB-NEXT:    or a0, a1, a0
-; RV64ZB-NEXT:    srli a1, a0, 1
-; RV64ZB-NEXT:    lui a2, 349525
-; RV64ZB-NEXT:    addiw a2, a2, 1365
-; RV64ZB-NEXT:    and a1, a1, a2
-; RV64ZB-NEXT:    and a0, a0, a2
-; RV64ZB-NEXT:    slliw a0, a0, 1
-; RV64ZB-NEXT:    or a0, a1, a0
-; RV64ZB-NEXT:    ret
+; RV32ZBB-LABEL: test_bitreverse_i32:
+; RV32ZBB:       # %bb.0:
+; RV32ZBB-NEXT:    rev8 a0, a0
+; RV32ZBB-NEXT:    srli a1, a0, 4
+; RV32ZBB-NEXT:    lui a2, 61681
+; RV32ZBB-NEXT:    addi a2, a2, -241
+; RV32ZBB-NEXT:    and a1, a1, a2
+; RV32ZBB-NEXT:    and a0, a0, a2
+; RV32ZBB-NEXT:    slli a0, a0, 4
+; RV32ZBB-NEXT:    or a0, a1, a0
+; RV32ZBB-NEXT:    srli a1, a0, 2
+; RV32ZBB-NEXT:    lui a2, 209715
+; RV32ZBB-NEXT:    addi a2, a2, 819
+; RV32ZBB-NEXT:    and a1, a1, a2
+; RV32ZBB-NEXT:    and a0, a0, a2
+; RV32ZBB-NEXT:    slli a0, a0, 2
+; RV32ZBB-NEXT:    or a0, a1, a0
+; RV32ZBB-NEXT:    srli a1, a0, 1
+; RV32ZBB-NEXT:    lui a2, 349525
+; RV32ZBB-NEXT:    addi a2, a2, 1365
+; RV32ZBB-NEXT:    and a1, a1, a2
+; RV32ZBB-NEXT:    and a0, a0, a2
+; RV32ZBB-NEXT:    slli a0, a0, 1
+; RV32ZBB-NEXT:    or a0, a1, a0
+; RV32ZBB-NEXT:    ret
+;
+; RV64ZBB-LABEL: test_bitreverse_i32:
+; RV64ZBB:       # %bb.0:
+; RV64ZBB-NEXT:    rev8 a0, a0
+; RV64ZBB-NEXT:    srli a1, a0, 36
+; RV64ZBB-NEXT:    lui a2, 61681
+; RV64ZBB-NEXT:    addiw a2, a2, -241
+; RV64ZBB-NEXT:    and a1, a1, a2
+; RV64ZBB-NEXT:    srli a0, a0, 28
+; RV64ZBB-NEXT:    lui a2, 986895
+; RV64ZBB-NEXT:    addiw a2, a2, 240
+; RV64ZBB-NEXT:    and a0, a0, a2
+; RV64ZBB-NEXT:    sext.w a0, a0
+; RV64ZBB-NEXT:    or a0, a1, a0
+; RV64ZBB-NEXT:    srli a1, a0, 2
+; RV64ZBB-NEXT:    lui a2, 209715
+; RV64ZBB-NEXT:    addiw a2, a2, 819
+; RV64ZBB-NEXT:    and a1, a1, a2
+; RV64ZBB-NEXT:    and a0, a0, a2
+; RV64ZBB-NEXT:    slliw a0, a0, 2
+; RV64ZBB-NEXT:    or a0, a1, a0
+; RV64ZBB-NEXT:    srli a1, a0, 1
+; RV64ZBB-NEXT:    lui a2, 349525
+; RV64ZBB-NEXT:    addiw a2, a2, 1365
+; RV64ZBB-NEXT:    and a1, a1, a2
+; RV64ZBB-NEXT:    and a0, a0, a2
+; RV64ZBB-NEXT:    slliw a0, a0, 1
+; RV64ZBB-NEXT:    or a0, a1, a0
+; RV64ZBB-NEXT:    ret
+;
+; RV32ZBKB-LABEL: test_bitreverse_i32:
+; RV32ZBKB:       # %bb.0:
+; RV32ZBKB-NEXT:    rev8 a0, a0
+; RV32ZBKB-NEXT:    brev8 a0, a0
+; RV32ZBKB-NEXT:    ret
+;
+; RV64ZBKB-LABEL: test_bitreverse_i32:
+; RV64ZBKB:       # %bb.0:
+; RV64ZBKB-NEXT:    rev8 a0, a0
+; RV64ZBKB-NEXT:    brev8 a0, a0
+; RV64ZBKB-NEXT:    srli a0, a0, 32
+; RV64ZBKB-NEXT:    ret
   %tmp = call i32 @llvm.bitreverse.i32(i32 %a)
   ret i32 %tmp
 }
 
 define i64 @test_bitreverse_i64(i64 %a) nounwind {
 ; RV32I-LABEL: test_bitreverse_i64:
 ; RV32I:       # %bb.0:
 ; RV32I-NEXT:    srli a2, a1, 8

 ; RV64I-NEXT:    or a0, a1, a0
 ; RV64I-NEXT:    srli a1, a0, 1
 ; RV64I-NEXT:    and a1, a1, a2
 ; RV64I-NEXT:    and a0, a0, a2
 ; RV64I-NEXT:    slli a0, a0, 1
 ; RV64I-NEXT:    or a0, a1, a0
 ; RV64I-NEXT:    ret
 ;
-; RV32ZB-LABEL: test_bitreverse_i64:
-; RV32ZB:       # %bb.0:
-; RV32ZB-NEXT:    rev8 a1, a1
-; RV32ZB-NEXT:    srli a2, a1, 4
-; RV32ZB-NEXT:    lui a3, 61681
-; RV32ZB-NEXT:    addi a3, a3, -241
-; RV32ZB-NEXT:    and a2, a2, a3
-; RV32ZB-NEXT:    and a1, a1, a3
-; RV32ZB-NEXT:    slli a1, a1, 4
-; RV32ZB-NEXT:    or a1, a2, a1
-; RV32ZB-NEXT:    srli a2, a1, 2
-; RV32ZB-NEXT:    lui a4, 209715
-; RV32ZB-NEXT:    addi a4, a4, 819
-; RV32ZB-NEXT:    and a2, a2, a4
-; RV32ZB-NEXT:    and a1, a1, a4
-; RV32ZB-NEXT:    slli a1, a1, 2
-; RV32ZB-NEXT:    or a1, a2, a1
-; RV32ZB-NEXT:    srli a2, a1, 1
-; RV32ZB-NEXT:    lui a5, 349525
-; RV32ZB-NEXT:    addi a5, a5, 1365
-; RV32ZB-NEXT:    and a2, a2, a5
-; RV32ZB-NEXT:    and a1, a1, a5
-; RV32ZB-NEXT:    slli a1, a1, 1
-; RV32ZB-NEXT:    or a2, a2, a1
-; RV32ZB-NEXT:    rev8 a0, a0
-; RV32ZB-NEXT:    srli a1, a0, 4
-; RV32ZB-NEXT:    and a1, a1, a3
-; RV32ZB-NEXT:    and a0, a0, a3
-; RV32ZB-NEXT:    slli a0, a0, 4
-; RV32ZB-NEXT:    or a0, a1, a0
-; RV32ZB-NEXT:    srli a1, a0, 2
-; RV32ZB-NEXT:    and a1, a1, a4
-; RV32ZB-NEXT:    and a0, a0, a4
-; RV32ZB-NEXT:    slli a0, a0, 2
-; RV32ZB-NEXT:    or a0, a1, a0
-; RV32ZB-NEXT:    srli a1, a0, 1
-; RV32ZB-NEXT:    and a1, a1, a5
-; RV32ZB-NEXT:    and a0, a0, a5
-; RV32ZB-NEXT:    slli a0, a0, 1
-; RV32ZB-NEXT:    or a1, a1, a0
-; RV32ZB-NEXT:    mv a0, a2
-; RV32ZB-NEXT:    ret
-;
-; RV64ZB-LABEL: test_bitreverse_i64:
-; RV64ZB:       # %bb.0:
-; RV64ZB-NEXT:    lui a1, %hi(.LCPI6_0)
-; RV64ZB-NEXT:    ld a1, %lo(.LCPI6_0)(a1)
-; RV64ZB-NEXT:    rev8 a0, a0
-; RV64ZB-NEXT:    srli a2, a0, 4
-; RV64ZB-NEXT:    and a2, a2, a1
-; RV64ZB-NEXT:    and a0, a0, a1
-; RV64ZB-NEXT:    lui a1, %hi(.LCPI6_1)
-; RV64ZB-NEXT:    ld a1, %lo(.LCPI6_1)(a1)
-; RV64ZB-NEXT:    slli a0, a0, 4
-; RV64ZB-NEXT:    or a0, a2, a0
-; RV64ZB-NEXT:    srli a2, a0, 2
-; RV64ZB-NEXT:    and a2, a2, a1
-; RV64ZB-NEXT:    and a0, a0, a1
-; RV64ZB-NEXT:    lui a1, %hi(.LCPI6_2)
-; RV64ZB-NEXT:    ld a1, %lo(.LCPI6_2)(a1)
-; RV64ZB-NEXT:    slli a0, a0, 2
-; RV64ZB-NEXT:    or a0, a2, a0
-; RV64ZB-NEXT:    srli a2, a0, 1
-; RV64ZB-NEXT:    and a2, a2, a1
-; RV64ZB-NEXT:    and a0, a0, a1
-; RV64ZB-NEXT:    slli a0, a0, 1
-; RV64ZB-NEXT:    or a0, a2, a0
-; RV64ZB-NEXT:    ret
+; RV32ZBB-LABEL: test_bitreverse_i64:
+; RV32ZBB:       # %bb.0:
+; RV32ZBB-NEXT:    rev8 a1, a1
+; RV32ZBB-NEXT:    srli a2, a1, 4
+; RV32ZBB-NEXT:    lui a3, 61681
+; RV32ZBB-NEXT:    addi a3, a3, -241
+; RV32ZBB-NEXT:    and a2, a2, a3
+; RV32ZBB-NEXT:    and a1, a1, a3
+; RV32ZBB-NEXT:    slli a1, a1, 4
+; RV32ZBB-NEXT:    or a1, a2, a1
+; RV32ZBB-NEXT:    srli a2, a1, 2
+; RV32ZBB-NEXT:    lui a4, 209715
+; RV32ZBB-NEXT:    addi a4, a4, 819
+; RV32ZBB-NEXT:    and a2, a2, a4
+; RV32ZBB-NEXT:    and a1, a1, a4
+; RV32ZBB-NEXT:    slli a1, a1, 2
+; RV32ZBB-NEXT:    or a1, a2, a1
+; RV32ZBB-NEXT:    srli a2, a1, 1
+; RV32ZBB-NEXT:    lui a5, 349525
+; RV32ZBB-NEXT:    addi a5, a5, 1365
+; RV32ZBB-NEXT:    and a2, a2, a5
+; RV32ZBB-NEXT:    and a1, a1, a5
+; RV32ZBB-NEXT:    slli a1, a1, 1
+; RV32ZBB-NEXT:    or a2, a2, a1
+; RV32ZBB-NEXT:    rev8 a0, a0
+; RV32ZBB-NEXT:    srli a1, a0, 4
+; RV32ZBB-NEXT:    and a1, a1, a3
+; RV32ZBB-NEXT:    and a0, a0, a3
+; RV32ZBB-NEXT:    slli a0, a0, 4
+; RV32ZBB-NEXT:    or a0, a1, a0
+; RV32ZBB-NEXT:    srli a1, a0, 2
+; RV32ZBB-NEXT:    and a1, a1, a4
+; RV32ZBB-NEXT:    and a0, a0, a4
+; RV32ZBB-NEXT:    slli a0, a0, 2
+; RV32ZBB-NEXT:    or a0, a1, a0
+; RV32ZBB-NEXT:    srli a1, a0, 1
+; RV32ZBB-NEXT:    and a1, a1, a5
+; RV32ZBB-NEXT:    and a0, a0, a5
+; RV32ZBB-NEXT:    slli a0, a0, 1
+; RV32ZBB-NEXT:    or a1, a1, a0
+; RV32ZBB-NEXT:    mv a0, a2
+; RV32ZBB-NEXT:    ret
+;
+; RV64ZBB-LABEL: test_bitreverse_i64:
+; RV64ZBB:       # %bb.0:
+; RV64ZBB-NEXT:    lui a1, %hi(.LCPI6_0)
+; RV64ZBB-NEXT:    ld a1, %lo(.LCPI6_0)(a1)
+; RV64ZBB-NEXT:    rev8 a0, a0
+; RV64ZBB-NEXT:    srli a2, a0, 4
+; RV64ZBB-NEXT:    and a2, a2, a1
+; RV64ZBB-NEXT:    and a0, a0, a1
+; RV64ZBB-NEXT:    lui a1, %hi(.LCPI6_1)
+; RV64ZBB-NEXT:    ld a1, %lo(.LCPI6_1)(a1)
+; RV64ZBB-NEXT:    slli a0, a0, 4
+; RV64ZBB-NEXT:    or a0, a2, a0
+; RV64ZBB-NEXT:    srli a2, a0, 2
+; RV64ZBB-NEXT:    and a2, a2, a1
+; RV64ZBB-NEXT:    and a0, a0, a1
+; RV64ZBB-NEXT:    lui a1, %hi(.LCPI6_2)
+; RV64ZBB-NEXT:    ld a1, %lo(.LCPI6_2)(a1)
+; RV64ZBB-NEXT:    slli a0, a0, 2
+; RV64ZBB-NEXT:    or a0, a2, a0
+; RV64ZBB-NEXT:    srli a2, a0, 1
+; RV64ZBB-NEXT:    and a2, a2, a1
+; RV64ZBB-NEXT:    and a0, a0, a1
+; RV64ZBB-NEXT:    slli a0, a0, 1
+; RV64ZBB-NEXT:    or a0, a2, a0
+; RV64ZBB-NEXT:    ret
+;
+; RV32ZBKB-LABEL: test_bitreverse_i64:
+; RV32ZBKB:       # %bb.0:
+; RV32ZBKB-NEXT:    rev8 a1, a1
+; RV32ZBKB-NEXT:    brev8 a2, a1
+; RV32ZBKB-NEXT:    rev8 a0, a0
+; RV32ZBKB-NEXT:    brev8 a1, a0
+; RV32ZBKB-NEXT:    mv a0, a2
+; RV32ZBKB-NEXT:    ret
+;
+; RV64ZBKB-LABEL: test_bitreverse_i64:
+; RV64ZBKB:       # %bb.0:
+; RV64ZBKB-NEXT:    rev8 a0, a0
+; RV64ZBKB-NEXT:    brev8 a0, a0
+; RV64ZBKB-NEXT:    ret
   %tmp = call i64 @llvm.bitreverse.i64(i64 %a)
   ret i64 %tmp
 }
 
 define i16 @test_bswap_bitreverse_i16(i16 %a) nounwind {
 ; RV32I-LABEL: test_bswap_bitreverse_i16:
 ; RV32I:       # %bb.0:
 ; RV32I-NEXT:    srli a1, a0, 4

 ; RV64I-NEXT:    lui a2, 5
 ; RV64I-NEXT:    addiw a2, a2, 1365
 ; RV64I-NEXT:    and a1, a1, a2
 ; RV64I-NEXT:    and a0, a0, a2
 ; RV64I-NEXT:    slli a0, a0, 1
 ; RV64I-NEXT:    or a0, a1, a0
 ; RV64I-NEXT:    ret
 ;
-; RV32ZB-LABEL: test_bswap_bitreverse_i16:
-; RV32ZB:       # %bb.0:
-; RV32ZB-NEXT:    srli a1, a0, 4
-; RV32ZB-NEXT:    lui a2, 1
-; RV32ZB-NEXT:    addi a2, a2, -241
-; RV32ZB-NEXT:    and a1, a1, a2
-; RV32ZB-NEXT:    and a0, a0, a2
-; RV32ZB-NEXT:    slli a0, a0, 4
-; RV32ZB-NEXT:    or a0, a1, a0
-; RV32ZB-NEXT:    srli a1, a0, 2
-; RV32ZB-NEXT:    lui a2, 3
-; RV32ZB-NEXT:    addi a2, a2, 819
-; RV32ZB-NEXT:    and a1, a1, a2
-; RV32ZB-NEXT:    and a0, a0, a2
-; RV32ZB-NEXT:    slli a0, a0, 2
-; RV32ZB-NEXT:    or a0, a1, a0
-; RV32ZB-NEXT:    srli a1, a0, 1
-; RV32ZB-NEXT:    lui a2, 5
-; RV32ZB-NEXT:    addi a2, a2, 1365
-; RV32ZB-NEXT:    and a1, a1, a2
-; RV32ZB-NEXT:    and a0, a0, a2
-; RV32ZB-NEXT:    slli a0, a0, 1
-; RV32ZB-NEXT:    or a0, a1, a0
-; RV32ZB-NEXT:    ret
-;
-; RV64ZB-LABEL: test_bswap_bitreverse_i16:
-; RV64ZB:       # %bb.0:
-; RV64ZB-NEXT:    srli a1, a0, 4
-; RV64ZB-NEXT:    lui a2, 1
-; RV64ZB-NEXT:    addiw a2, a2, -241
-; RV64ZB-NEXT:    and a1, a1, a2
-; RV64ZB-NEXT:    and a0, a0, a2
-; RV64ZB-NEXT:    slli a0, a0, 4
-; RV64ZB-NEXT:    or a0, a1, a0
-; RV64ZB-NEXT:    srli a1, a0, 2
-; RV64ZB-NEXT:    lui a2, 3
-; RV64ZB-NEXT:    addiw a2, a2, 819
-; RV64ZB-NEXT:    and a1, a1, a2
-; RV64ZB-NEXT:    and a0, a0, a2
-; RV64ZB-NEXT:    slli a0, a0, 2
-; RV64ZB-NEXT:    or a0, a1, a0
-; RV64ZB-NEXT:    srli a1, a0, 1
-; RV64ZB-NEXT:    lui a2, 5
-; RV64ZB-NEXT:    addiw a2, a2, 1365
-; RV64ZB-NEXT:    and a1, a1, a2
-; RV64ZB-NEXT:    and a0, a0, a2
-; RV64ZB-NEXT:    slli a0, a0, 1
-; RV64ZB-NEXT:    or a0, a1, a0
-; RV64ZB-NEXT:    ret
+; RV32ZBB-LABEL: test_bswap_bitreverse_i16:
+; RV32ZBB:       # %bb.0:
+; RV32ZBB-NEXT:    srli a1, a0, 4
+; RV32ZBB-NEXT:    lui a2, 1
+; RV32ZBB-NEXT:    addi a2, a2, -241
+; RV32ZBB-NEXT:    and a1, a1, a2
+; RV32ZBB-NEXT:    and a0, a0, a2
+; RV32ZBB-NEXT:    slli a0, a0, 4
+; RV32ZBB-NEXT:    or a0, a1, a0
+; RV32ZBB-NEXT:    srli a1, a0, 2
+; RV32ZBB-NEXT:    lui a2, 3
+; RV32ZBB-NEXT:    addi a2, a2, 819
+; RV32ZBB-NEXT:    and a1, a1, a2
+; RV32ZBB-NEXT:    and a0, a0, a2
+; RV32ZBB-NEXT:    slli a0, a0, 2
+; RV32ZBB-NEXT:    or a0, a1, a0
+; RV32ZBB-NEXT:    srli a1, a0, 1
+; RV32ZBB-NEXT:    lui a2, 5
+; RV32ZBB-NEXT:    addi a2, a2, 1365
+; RV32ZBB-NEXT:    and a1, a1, a2
+; RV32ZBB-NEXT:    and a0, a0, a2
+; RV32ZBB-NEXT:    slli a0, a0, 1
+; RV32ZBB-NEXT:    or a0, a1, a0
+; RV32ZBB-NEXT:    ret
+;
+; RV64ZBB-LABEL: test_bswap_bitreverse_i16:
+; RV64ZBB:       # %bb.0:
+; RV64ZBB-NEXT:    srli a1, a0, 4
+; RV64ZBB-NEXT:    lui a2, 1
+; RV64ZBB-NEXT:    addiw a2, a2, -241
+; RV64ZBB-NEXT:    and a1, a1, a2
+; RV64ZBB-NEXT:    and a0, a0, a2
+; RV64ZBB-NEXT:    slli a0, a0, 4
+; RV64ZBB-NEXT:    or a0, a1, a0
+; RV64ZBB-NEXT:    srli a1, a0, 2
+; RV64ZBB-NEXT:    lui a2, 3
+; RV64ZBB-NEXT:    addiw a2, a2, 819
+; RV64ZBB-NEXT:    and a1, a1, a2
+; RV64ZBB-NEXT:    and a0, a0, a2
+; RV64ZBB-NEXT:    slli a0, a0, 2
+; RV64ZBB-NEXT:    or a0, a1, a0
+; RV64ZBB-NEXT:    srli a1, a0, 1
+; RV64ZBB-NEXT:    lui a2, 5
+; RV64ZBB-NEXT:    addiw a2, a2, 1365
+; RV64ZBB-NEXT:    and a1, a1, a2
+; RV64ZBB-NEXT:    and a0, a0, a2
+; RV64ZBB-NEXT:    slli a0, a0, 1
+; RV64ZBB-NEXT:    or a0, a1, a0
+; RV64ZBB-NEXT:    ret
+;
+; 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:    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:    brev8 a0, a0
+; RV64ZBKB-NEXT:    srli a0, a0, 48
+; RV64ZBKB-NEXT:    ret
   %tmp = call i16 @llvm.bswap.i16(i16 %a)
   %tmp2 = call i16 @llvm.bitreverse.i16(i16 %tmp)
   ret i16 %tmp2
 }
 
 define i32 @test_bswap_bitreverse_i32(i32 %a) nounwind {
 ; RV32I-LABEL: test_bswap_bitreverse_i32:
 ; RV32I:       # %bb.0:

 ; RV64I-NEXT:    lui a2, 349525
 ; RV64I-NEXT:    addiw a2, a2, 1365
 ; RV64I-NEXT:    and a1, a1, a2
 ; RV64I-NEXT:    and a0, a0, a2
 ; RV64I-NEXT:    slliw a0, a0, 1
 ; RV64I-NEXT:    or a0, a1, a0
 ; RV64I-NEXT:    ret
 ;
-; RV32ZB-LABEL: test_bswap_bitreverse_i32:
-; RV32ZB:       # %bb.0:
-; RV32ZB-NEXT:    srli a1, a0, 4
-; RV32ZB-NEXT:    lui a2, 61681
-; RV32ZB-NEXT:    addi a2, a2, -241
-; RV32ZB-NEXT:    and a1, a1, a2
-; RV32ZB-NEXT:    and a0, a0, a2
-; RV32ZB-NEXT:    slli a0, a0, 4
-; RV32ZB-NEXT:    or a0, a1, a0
-; RV32ZB-NEXT:    srli a1, a0, 2
-; RV32ZB-NEXT:    lui a2, 209715
-; RV32ZB-NEXT:    addi a2, a2, 819
-; RV32ZB-NEXT:    and a1, a1, a2
-; RV32ZB-NEXT:    and a0, a0, a2
-; RV32ZB-NEXT:    slli a0, a0, 2
-; RV32ZB-NEXT:    or a0, a1, a0
-; RV32ZB-NEXT:    srli a1, a0, 1
-; RV32ZB-NEXT:    lui a2, 349525
-; RV32ZB-NEXT:    addi a2, a2, 1365
-; RV32ZB-NEXT:    and a1, a1, a2
-; RV32ZB-NEXT:    and a0, a0, a2
-; RV32ZB-NEXT:    slli a0, a0, 1
-; RV32ZB-NEXT:    or a0, a1, a0
-; RV32ZB-NEXT:    ret
-;
-; RV64ZB-LABEL: test_bswap_bitreverse_i32:
-; RV64ZB:       # %bb.0:
-; RV64ZB-NEXT:    srli a1, a0, 4
-; RV64ZB-NEXT:    lui a2, 61681
-; RV64ZB-NEXT:    addiw a2, a2, -241
-; RV64ZB-NEXT:    and a1, a1, a2
-; RV64ZB-NEXT:    and a0, a0, a2
-; RV64ZB-NEXT:    slliw a0, a0, 4
-; RV64ZB-NEXT:    or a0, a1, a0
-; RV64ZB-NEXT:    srli a1, a0, 2
-; RV64ZB-NEXT:    lui a2, 209715
-; RV64ZB-NEXT:    addiw a2, a2, 819
-; RV64ZB-NEXT:    and a1, a1, a2
-; RV64ZB-NEXT:    and a0, a0, a2
-; RV64ZB-NEXT:    slliw a0, a0, 2
-; RV64ZB-NEXT:    or a0, a1, a0
-; RV64ZB-NEXT:    srli a1, a0, 1
-; RV64ZB-NEXT:    lui a2, 349525
-; RV64ZB-NEXT:    addiw a2, a2, 1365
-; RV64ZB-NEXT:    and a1, a1, a2
-; RV64ZB-NEXT:    and a0, a0, a2
-; RV64ZB-NEXT:    slliw a0, a0, 1
-; RV64ZB-NEXT:    or a0, a1, a0
-; RV64ZB-NEXT:    ret
+; RV32ZBB-LABEL: test_bswap_bitreverse_i32:
+; RV32ZBB:       # %bb.0:
+; RV32ZBB-NEXT:    srli a1, a0, 4
+; RV32ZBB-NEXT:    lui a2, 61681
+; RV32ZBB-NEXT:    addi a2, a2, -241
+; RV32ZBB-NEXT:    and a1, a1, a2
+; RV32ZBB-NEXT:    and a0, a0, a2
+; RV32ZBB-NEXT:    slli a0, a0, 4
+; RV32ZBB-NEXT:    or a0, a1, a0
+; RV32ZBB-NEXT:    srli a1, a0, 2
+; RV32ZBB-NEXT:    lui a2, 209715
+; RV32ZBB-NEXT:    addi a2, a2, 819
+; RV32ZBB-NEXT:    and a1, a1, a2
+; RV32ZBB-NEXT:    and a0, a0, a2
+; RV32ZBB-NEXT:    slli a0, a0, 2
+; RV32ZBB-NEXT:    or a0, a1, a0
+; RV32ZBB-NEXT:    srli a1, a0, 1
+; RV32ZBB-NEXT:    lui a2, 349525
+; RV32ZBB-NEXT:    addi a2, a2, 1365
+; RV32ZBB-NEXT:    and a1, a1, a2
+; RV32ZBB-NEXT:    and a0, a0, a2
+; RV32ZBB-NEXT:    slli a0, a0, 1
+; RV32ZBB-NEXT:    or a0, a1, a0
+; RV32ZBB-NEXT:    ret
+;
+; RV64ZBB-LABEL: test_bswap_bitreverse_i32:
+; RV64ZBB:       # %bb.0:
+; RV64ZBB-NEXT:    srli a1, a0, 4
+; RV64ZBB-NEXT:    lui a2, 61681
+; RV64ZBB-NEXT:    addiw a2, a2, -241
+; RV64ZBB-NEXT:    and a1, a1, a2
+; RV64ZBB-NEXT:    and a0, a0, a2
+; RV64ZBB-NEXT:    slliw a0, a0, 4
+; RV64ZBB-NEXT:    or a0, a1, a0
+; RV64ZBB-NEXT:    srli a1, a0, 2
+; RV64ZBB-NEXT:    lui a2, 209715
+; RV64ZBB-NEXT:    addiw a2, a2, 819
+; RV64ZBB-NEXT:    and a1, a1, a2
+; RV64ZBB-NEXT:    and a0, a0, a2
+; RV64ZBB-NEXT:    slliw a0, a0, 2
+; RV64ZBB-NEXT:    or a0, a1, a0
+; RV64ZBB-NEXT:    srli a1, a0, 1
+; RV64ZBB-NEXT:    lui a2, 349525
+; RV64ZBB-NEXT:    addiw a2, a2, 1365
+; RV64ZBB-NEXT:    and a1, a1, a2
+; RV64ZBB-NEXT:    and a0, a0, a2
+; RV64ZBB-NEXT:    slliw a0, a0, 1
+; RV64ZBB-NEXT:    or a0, a1, a0
+; RV64ZBB-NEXT:    ret
+;
+; RV32ZBKB-LABEL: test_bswap_bitreverse_i32:
+; RV32ZBKB:       # %bb.0:
+; RV32ZBKB-NEXT:    brev8 a0, a0
+; RV32ZBKB-NEXT:    ret
+;
+; 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:    brev8 a0, a0
+; RV64ZBKB-NEXT:    srli a0, a0, 32
+; RV64ZBKB-NEXT:    ret
   %tmp = call i32 @llvm.bswap.i32(i32 %a)
   %tmp2 = call i32 @llvm.bitreverse.i32(i32 %tmp)
   ret i32 %tmp2
 }
 
 define i64 @test_bswap_bitreverse_i64(i64 %a) nounwind {
 ; RV32I-LABEL: test_bswap_bitreverse_i64:
 ; RV32I:       # %bb.0:

 ; RV64I-NEXT:    or a0, a2, a0
 ; RV64I-NEXT:    srli a2, a0, 1
 ; RV64I-NEXT:    and a2, a2, a1
 ; RV64I-NEXT:    and a0, a0, a1
 ; RV64I-NEXT:    slli a0, a0, 1
 ; RV64I-NEXT:    or a0, a2, a0
 ; RV64I-NEXT:    ret
 ;
-; RV32ZB-LABEL: test_bswap_bitreverse_i64:
-; RV32ZB:       # %bb.0:
-; RV32ZB-NEXT:    srli a2, a0, 4
-; RV32ZB-NEXT:    lui a3, 61681
-; RV32ZB-NEXT:    addi a3, a3, -241
-; RV32ZB-NEXT:    and a2, a2, a3
-; RV32ZB-NEXT:    and a0, a0, a3
-; RV32ZB-NEXT:    slli a0, a0, 4
-; RV32ZB-NEXT:    or a0, a2, a0
-; RV32ZB-NEXT:    srli a2, a0, 2
-; RV32ZB-NEXT:    lui a4, 209715
-; RV32ZB-NEXT:    addi a4, a4, 819
-; RV32ZB-NEXT:    and a2, a2, a4
-; RV32ZB-NEXT:    and a0, a0, a4
-; RV32ZB-NEXT:    slli a0, a0, 2
-; RV32ZB-NEXT:    or a0, a2, a0
-; RV32ZB-NEXT:    srli a2, a0, 1
-; RV32ZB-NEXT:    lui a5, 349525
-; RV32ZB-NEXT:    addi a5, a5, 1365
-; RV32ZB-NEXT:    and a2, a2, a5
-; RV32ZB-NEXT:    and a0, a0, a5
-; RV32ZB-NEXT:    slli a0, a0, 1
-; RV32ZB-NEXT:    or a0, a2, a0
-; RV32ZB-NEXT:    srli a2, a1, 4
-; RV32ZB-NEXT:    and a2, a2, a3
-; RV32ZB-NEXT:    and a1, a1, a3
-; RV32ZB-NEXT:    slli a1, a1, 4
-; RV32ZB-NEXT:    or a1, a2, a1
-; RV32ZB-NEXT:    srli a2, a1, 2
-; RV32ZB-NEXT:    and a2, a2, a4
-; RV32ZB-NEXT:    and a1, a1, a4
-; RV32ZB-NEXT:    slli a1, a1, 2
-; RV32ZB-NEXT:    or a1, a2, a1
-; RV32ZB-NEXT:    srli a2, a1, 1
-; RV32ZB-NEXT:    and a2, a2, a5
-; RV32ZB-NEXT:    and a1, a1, a5
-; RV32ZB-NEXT:    slli a1, a1, 1
-; RV32ZB-NEXT:    or a1, a2, a1
-; RV32ZB-NEXT:    ret
-;
-; RV64ZB-LABEL: test_bswap_bitreverse_i64:
-; RV64ZB:       # %bb.0:
-; RV64ZB-NEXT:    lui a1, %hi(.LCPI9_0)
-; RV64ZB-NEXT:    ld a1, %lo(.LCPI9_0)(a1)
-; RV64ZB-NEXT:    srli a2, a0, 4
-; RV64ZB-NEXT:    and a2, a2, a1
-; RV64ZB-NEXT:    and a0, a0, a1
-; RV64ZB-NEXT:    lui a1, %hi(.LCPI9_1)
-; RV64ZB-NEXT:    ld a1, %lo(.LCPI9_1)(a1)
-; RV64ZB-NEXT:    slli a0, a0, 4
-; RV64ZB-NEXT:    or a0, a2, a0
-; RV64ZB-NEXT:    srli a2, a0, 2
-; RV64ZB-NEXT:    and a2, a2, a1
-; RV64ZB-NEXT:    and a0, a0, a1
-; RV64ZB-NEXT:    lui a1, %hi(.LCPI9_2)
-; RV64ZB-NEXT:    ld a1, %lo(.LCPI9_2)(a1)
-; RV64ZB-NEXT:    slli a0, a0, 2
-; RV64ZB-NEXT:    or a0, a2, a0
-; RV64ZB-NEXT:    srli a2, a0, 1
-; RV64ZB-NEXT:    and a2, a2, a1
-; RV64ZB-NEXT:    and a0, a0, a1
-; RV64ZB-NEXT:    slli a0, a0, 1
-; RV64ZB-NEXT:    or a0, a2, a0
-; RV64ZB-NEXT:    ret
+; RV32ZBB-LABEL: test_bswap_bitreverse_i64:
+; RV32ZBB:       # %bb.0:
+; RV32ZBB-NEXT:    srli a2, a0, 4
+; RV32ZBB-NEXT:    lui a3, 61681
+; RV32ZBB-NEXT:    addi a3, a3, -241
+; RV32ZBB-NEXT:    and a2, a2, a3
+; RV32ZBB-NEXT:    and a0, a0, a3
+; RV32ZBB-NEXT:    slli a0, a0, 4
+; RV32ZBB-NEXT:    or a0, a2, a0
+; RV32ZBB-NEXT:    srli a2, a0, 2
+; RV32ZBB-NEXT:    lui a4, 209715
+; RV32ZBB-NEXT:    addi a4, a4, 819
+; RV32ZBB-NEXT:    and a2, a2, a4
+; RV32ZBB-NEXT:    and a0, a0, a4
+; RV32ZBB-NEXT:    slli a0, a0, 2
+; RV32ZBB-NEXT:    or a0, a2, a0
+; RV32ZBB-NEXT:    srli a2, a0, 1
+; RV32ZBB-NEXT:    lui a5, 349525
+; RV32ZBB-NEXT:    addi a5, a5, 1365
+; RV32ZBB-NEXT:    and a2, a2, a5
+; RV32ZBB-NEXT:    and a0, a0, a5
+; RV32ZBB-NEXT:    slli a0, a0, 1
+; RV32ZBB-NEXT:    or a0, a2, a0
+; RV32ZBB-NEXT:    srli a2, a1, 4
+; RV32ZBB-NEXT:    and a2, a2, a3
+; RV32ZBB-NEXT:    and a1, a1, a3
+; RV32ZBB-NEXT:    slli a1, a1, 4
+; RV32ZBB-NEXT:    or a1, a2, a1
+; RV32ZBB-NEXT:    srli a2, a1, 2
+; RV32ZBB-NEXT:    and a2, a2, a4
+; RV32ZBB-NEXT:    and a1, a1, a4
+; RV32ZBB-NEXT:    slli a1, a1, 2
+; RV32ZBB-NEXT:    or a1, a2, a1
+; RV32ZBB-NEXT:    srli a2, a1, 1
+; RV32ZBB-NEXT:    and a2, a2, a5
+; RV32ZBB-NEXT:    and a1, a1, a5
+; RV32ZBB-NEXT:    slli a1, a1, 1
+; RV32ZBB-NEXT:    or a1, a2, a1
+; RV32ZBB-NEXT:    ret
+;
+; RV64ZBB-LABEL: test_bswap_bitreverse_i64:
+; RV64ZBB:       # %bb.0:
+; RV64ZBB-NEXT:    lui a1, %hi(.LCPI9_0)
+; RV64ZBB-NEXT:    ld a1, %lo(.LCPI9_0)(a1)
+; RV64ZBB-NEXT:    srli a2, a0, 4
+; RV64ZBB-NEXT:    and a2, a2, a1
+; RV64ZBB-NEXT:    and a0, a0, a1
+; RV64ZBB-NEXT:    lui a1, %hi(.LCPI9_1)
+; RV64ZBB-NEXT:    ld a1, %lo(.LCPI9_1)(a1)
+; RV64ZBB-NEXT:    slli a0, a0, 4
+; RV64ZBB-NEXT:    or a0, a2, a0
+; RV64ZBB-NEXT:    srli a2, a0, 2
+; RV64ZBB-NEXT:    and a2, a2, a1
+; RV64ZBB-NEXT:    and a0, a0, a1
+; RV64ZBB-NEXT:    lui a1, %hi(.LCPI9_2)
+; RV64ZBB-NEXT:    ld a1, %lo(.LCPI9_2)(a1)
+; RV64ZBB-NEXT:    slli a0, a0, 2
+; RV64ZBB-NEXT:    or a0, a2, a0
+; RV64ZBB-NEXT:    srli a2, a0, 1
+; RV64ZBB-NEXT:    and a2, a2, a1
+; RV64ZBB-NEXT:    and a0, a0, a1
+; RV64ZBB-NEXT:    slli a0, a0, 1
+; RV64ZBB-NEXT:    or a0, a2, a0
+; RV64ZBB-NEXT:    ret
+;
+; RV32ZBKB-LABEL: test_bswap_bitreverse_i64:
+; RV32ZBKB:       # %bb.0:
+; RV32ZBKB-NEXT:    brev8 a0, a0
+; RV32ZBKB-NEXT:    brev8 a1, a1
+; RV32ZBKB-NEXT:    ret
+;
+; RV64ZBKB-LABEL: test_bswap_bitreverse_i64:
+; RV64ZBKB:       # %bb.0:
+; RV64ZBKB-NEXT:    brev8 a0, a0
+; RV64ZBKB-NEXT:    ret
   %tmp = call i64 @llvm.bswap.i64(i64 %a)
   %tmp2 = call i64 @llvm.bitreverse.i64(i64 %tmp)
   ret i64 %tmp2
 }
 
 define i16 @test_bitreverse_bswap_i16(i16 %a) nounwind {
 ; RV32I-LABEL: test_bitreverse_bswap_i16:
 ; RV32I:       # %bb.0:

 ; RV64I-NEXT:    lui a2, 5
 ; RV64I-NEXT:    addiw a2, a2, 1365
 ; RV64I-NEXT:    and a1, a1, a2
 ; RV64I-NEXT:    and a0, a0, a2
 ; RV64I-NEXT:    slli a0, a0, 1
 ; RV64I-NEXT:    or a0, a1, a0
 ; RV64I-NEXT:    ret
 ;
-; RV32ZB-LABEL: test_bitreverse_bswap_i16:
-; RV32ZB:       # %bb.0:
-; RV32ZB-NEXT:    srli a1, a0, 4
-; RV32ZB-NEXT:    lui a2, 1
-; RV32ZB-NEXT:    addi a2, a2, -241
-; RV32ZB-NEXT:    and a1, a1, a2
-; RV32ZB-NEXT:    and a0, a0, a2
-; RV32ZB-NEXT:    slli a0, a0, 4
-; RV32ZB-NEXT:    or a0, a1, a0
-; RV32ZB-NEXT:    srli a1, a0, 2
-; RV32ZB-NEXT:    lui a2, 3
-; RV32ZB-NEXT:    addi a2, a2, 819
-; RV32ZB-NEXT:    and a1, a1, a2
-; RV32ZB-NEXT:    and a0, a0, a2
-; RV32ZB-NEXT:    slli a0, a0, 2
-; RV32ZB-NEXT:    or a0, a1, a0
-; RV32ZB-NEXT:    srli a1, a0, 1
-; RV32ZB-NEXT:    lui a2, 5
-; RV32ZB-NEXT:    addi a2, a2, 1365
-; RV32ZB-NEXT:    and a1, a1, a2
-; RV32ZB-NEXT:    and a0, a0, a2
-; RV32ZB-NEXT:    slli a0, a0, 1
-; RV32ZB-NEXT:    or a0, a1, a0
-; RV32ZB-NEXT:    ret
-;
-; RV64ZB-LABEL: test_bitreverse_bswap_i16:
-; RV64ZB:       # %bb.0:
-; RV64ZB-NEXT:    srli a1, a0, 4
-; RV64ZB-NEXT:    lui a2, 1
-; RV64ZB-NEXT:    addiw a2, a2, -241
-; RV64ZB-NEXT:    and a1, a1, a2
-; RV64ZB-NEXT:    and a0, a0, a2
-; RV64ZB-NEXT:    slli a0, a0, 4
-; RV64ZB-NEXT:    or a0, a1, a0
-; RV64ZB-NEXT:    srli a1, a0, 2
-; RV64ZB-NEXT:    lui a2, 3
-; RV64ZB-NEXT:    addiw a2, a2, 819
-; RV64ZB-NEXT:    and a1, a1, a2
-; RV64ZB-NEXT:    and a0, a0, a2
-; RV64ZB-NEXT:    slli a0, a0, 2
-; RV64ZB-NEXT:    or a0, a1, a0
-; RV64ZB-NEXT:    srli a1, a0, 1
-; RV64ZB-NEXT:    lui a2, 5
-; RV64ZB-NEXT:    addiw a2, a2, 1365
-; RV64ZB-NEXT:    and a1, a1, a2
-; RV64ZB-NEXT:    and a0, a0, a2
-; RV64ZB-NEXT:    slli a0, a0, 1
-; RV64ZB-NEXT:    or a0, a1, a0
-; RV64ZB-NEXT:    ret
+; RV32ZBB-LABEL: test_bitreverse_bswap_i16:
+; RV32ZBB:       # %bb.0:
+; RV32ZBB-NEXT:    srli a1, a0, 4
+; RV32ZBB-NEXT:    lui a2, 1
+; RV32ZBB-NEXT:    addi a2, a2, -241
+; RV32ZBB-NEXT:    and a1, a1, a2
+; RV32ZBB-NEXT:    and a0, a0, a2
+; RV32ZBB-NEXT:    slli a0, a0, 4
+; RV32ZBB-NEXT:    or a0, a1, a0
+; RV32ZBB-NEXT:    srli a1, a0, 2
+; RV32ZBB-NEXT:    lui a2, 3
+; RV32ZBB-NEXT:    addi a2, a2, 819
+; RV32ZBB-NEXT:    and a1, a1, a2
+; RV32ZBB-NEXT:    and a0, a0, a2
+; RV32ZBB-NEXT:    slli a0, a0, 2
+; RV32ZBB-NEXT:    or a0, a1, a0
+; RV32ZBB-NEXT:    srli a1, a0, 1
+; RV32ZBB-NEXT:    lui a2, 5
+; RV32ZBB-NEXT:    addi a2, a2, 1365
+; RV32ZBB-NEXT:    and a1, a1, a2
+; RV32ZBB-NEXT:    and a0, a0, a2
+; RV32ZBB-NEXT:    slli a0, a0, 1
+; RV32ZBB-NEXT:    or a0, a1, a0
+; RV32ZBB-NEXT:    ret
+;
+; RV64ZBB-LABEL: test_bitreverse_bswap_i16:
+; RV64ZBB:       # %bb.0:
+; RV64ZBB-NEXT:    srli a1, a0, 4
+; RV64ZBB-NEXT:    lui a2, 1
+; RV64ZBB-NEXT:    addiw a2, a2, -241
+; RV64ZBB-NEXT:    and a1, a1, a2
+; RV64ZBB-NEXT:    and a0, a0, a2
+; RV64ZBB-NEXT:    slli a0, a0, 4
+; RV64ZBB-NEXT:    or a0, a1, a0
+; RV64ZBB-NEXT:    srli a1, a0, 2
+; RV64ZBB-NEXT:    lui a2, 3
+; RV64ZBB-NEXT:    addiw a2, a2, 819
+; RV64ZBB-NEXT:    and a1, a1, a2
+; RV64ZBB-NEXT:    and a0, a0, a2
+; RV64ZBB-NEXT:    slli a0, a0, 2
+; RV64ZBB-NEXT:    or a0, a1, a0
+; RV64ZBB-NEXT:    srli a1, a0, 1
+; RV64ZBB-NEXT:    lui a2, 5
+; RV64ZBB-NEXT:    addiw a2, a2, 1365
+; RV64ZBB-NEXT:    and a1, a1, a2
+; RV64ZBB-NEXT:    and a0, a0, a2
+; RV64ZBB-NEXT:    slli a0, a0, 1
+; RV64ZBB-NEXT:    or a0, a1, a0
+; RV64ZBB-NEXT:    ret
+;
+; 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:    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:    brev8 a0, a0
+; RV64ZBKB-NEXT:    srli a0, a0, 48
+; RV64ZBKB-NEXT:    ret
   %tmp = call i16 @llvm.bitreverse.i16(i16 %a)
   %tmp2 = call i16 @llvm.bswap.i16(i16 %tmp)
   ret i16 %tmp2
 }
 
 define i32 @test_bitreverse_bswap_i32(i32 %a) nounwind {
 ; RV32I-LABEL: test_bitreverse_bswap_i32:
 ; RV32I:       # %bb.0:

 ; RV64I-NEXT:    lui a2, 349525
 ; RV64I-NEXT:    addiw a2, a2, 1365
 ; RV64I-NEXT:    and a1, a1, a2
 ; RV64I-NEXT:    and a0, a0, a2
 ; RV64I-NEXT:    slliw a0, a0, 1
 ; RV64I-NEXT:    or a0, a1, a0
 ; RV64I-NEXT:    ret
 ;
-; RV32ZB-LABEL: test_bitreverse_bswap_i32:
-; RV32ZB:       # %bb.0:
-; RV32ZB-NEXT:    srli a1, a0, 4
-; RV32ZB-NEXT:    lui a2, 61681
-; RV32ZB-NEXT:    addi a2, a2, -241
-; RV32ZB-NEXT:    and a1, a1, a2
-; RV32ZB-NEXT:    and a0, a0, a2
-; RV32ZB-NEXT:    slli a0, a0, 4
-; RV32ZB-NEXT:    or a0, a1, a0
-; RV32ZB-NEXT:    srli a1, a0, 2
-; RV32ZB-NEXT:    lui a2, 209715
-; RV32ZB-NEXT:    addi a2, a2, 819
-; RV32ZB-NEXT:    and a1, a1, a2
-; RV32ZB-NEXT:    and a0, a0, a2
-; RV32ZB-NEXT:    slli a0, a0, 2
-; RV32ZB-NEXT:    or a0, a1, a0
-; RV32ZB-NEXT:    srli a1, a0, 1
-; RV32ZB-NEXT:    lui a2, 349525
-; RV32ZB-NEXT:    addi a2, a2, 1365
-; RV32ZB-NEXT:    and a1, a1, a2
-; RV32ZB-NEXT:    and a0, a0, a2
-; RV32ZB-NEXT:    slli a0, a0, 1
-; RV32ZB-NEXT:    or a0, a1, a0
-; RV32ZB-NEXT:    ret
-;
-; RV64ZB-LABEL: test_bitreverse_bswap_i32:
-; RV64ZB:       # %bb.0:
-; RV64ZB-NEXT:    srli a1, a0, 4
-; RV64ZB-NEXT:    lui a2, 61681
-; RV64ZB-NEXT:    addiw a2, a2, -241
-; RV64ZB-NEXT:    and a1, a1, a2
-; RV64ZB-NEXT:    and a0, a0, a2
-; RV64ZB-NEXT:    slliw a0, a0, 4
-; RV64ZB-NEXT:    or a0, a1, a0
-; RV64ZB-NEXT:    srli a1, a0, 2
-; RV64ZB-NEXT:    lui a2, 209715
-; RV64ZB-NEXT:    addiw a2, a2, 819
-; RV64ZB-NEXT:    and a1, a1, a2
-; RV64ZB-NEXT:    and a0, a0, a2
-; RV64ZB-NEXT:    slliw a0, a0, 2
-; RV64ZB-NEXT:    or a0, a1, a0
-; RV64ZB-NEXT:    srli a1, a0, 1
-; RV64ZB-NEXT:    lui a2, 349525
-; RV64ZB-NEXT:    addiw a2, a2, 1365
-; RV64ZB-NEXT:    and a1, a1, a2
-; RV64ZB-NEXT:    and a0, a0, a2
-; RV64ZB-NEXT:    slliw a0, a0, 1
-; RV64ZB-NEXT:    or a0, a1, a0
-; RV64ZB-NEXT:    ret
+; RV32ZBB-LABEL: test_bitreverse_bswap_i32:
+; RV32ZBB:       # %bb.0:
+; RV32ZBB-NEXT:    srli a1, a0, 4
+; RV32ZBB-NEXT:    lui a2, 61681
+; RV32ZBB-NEXT:    addi a2, a2, -241
+; RV32ZBB-NEXT:    and a1, a1, a2
+; RV32ZBB-NEXT:    and a0, a0, a2
+; RV32ZBB-NEXT:    slli a0, a0, 4
+; RV32ZBB-NEXT:    or a0, a1, a0
+; RV32ZBB-NEXT:    srli a1, a0, 2
+; RV32ZBB-NEXT:    lui a2, 209715
+; RV32ZBB-NEXT:    addi a2, a2, 819
+; RV32ZBB-NEXT:    and a1, a1, a2
+; RV32ZBB-NEXT:    and a0, a0, a2
+; RV32ZBB-NEXT:    slli a0, a0, 2
+; RV32ZBB-NEXT:    or a0, a1, a0
+; RV32ZBB-NEXT:    srli a1, a0, 1
+; RV32ZBB-NEXT:    lui a2, 349525
+; RV32ZBB-NEXT:    addi a2, a2, 1365
+; RV32ZBB-NEXT:    and a1, a1, a2
+; RV32ZBB-NEXT:    and a0, a0, a2
+; RV32ZBB-NEXT:    slli a0, a0, 1
+; RV32ZBB-NEXT:    or a0, a1, a0
+; RV32ZBB-NEXT:    ret
+;
+; RV64ZBB-LABEL: test_bitreverse_bswap_i32:
+; RV64ZBB:       # %bb.0:
+; RV64ZBB-NEXT:    srli a1, a0, 4
+; RV64ZBB-NEXT:    lui a2, 61681
+; RV64ZBB-NEXT:    addiw a2, a2, -241
+; RV64ZBB-NEXT:    and a1, a1, a2
+; RV64ZBB-NEXT:    and a0, a0, a2
+; RV64ZBB-NEXT:    slliw a0, a0, 4
+; RV64ZBB-NEXT:    or a0, a1, a0
+; RV64ZBB-NEXT:    srli a1, a0, 2
+; RV64ZBB-NEXT:    lui a2, 209715
+; RV64ZBB-NEXT:    addiw a2, a2, 819
+; RV64ZBB-NEXT:    and a1, a1, a2
+; RV64ZBB-NEXT:    and a0, a0, a2
+; RV64ZBB-NEXT:    slliw a0, a0, 2
+; RV64ZBB-NEXT:    or a0, a1, a0
+; RV64ZBB-NEXT:    srli a1, a0, 1
+; RV64ZBB-NEXT:    lui a2, 349525
+; RV64ZBB-NEXT:    addiw a2, a2, 1365
+; RV64ZBB-NEXT:    and a1, a1, a2
+; RV64ZBB-NEXT:    and a0, a0, a2
+; RV64ZBB-NEXT:    slliw a0, a0, 1
+; RV64ZBB-NEXT:    or a0, a1, a0
+; RV64ZBB-NEXT:    ret
+;
+; RV32ZBKB-LABEL: test_bitreverse_bswap_i32:
+; RV32ZBKB:       # %bb.0:
+; RV32ZBKB-NEXT:    brev8 a0, a0
+; RV32ZBKB-NEXT:    ret
+;
+; 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:    brev8 a0, a0
+; RV64ZBKB-NEXT:    srli a0, a0, 32
+; RV64ZBKB-NEXT:    ret
   %tmp = call i32 @llvm.bitreverse.i32(i32 %a)
   %tmp2 = call i32 @llvm.bswap.i32(i32 %tmp)
   ret i32 %tmp2
 }
 
 define i64 @test_bitreverse_bswap_i64(i64 %a) nounwind {
 ; RV32I-LABEL: test_bitreverse_bswap_i64:
 ; RV32I:       # %bb.0:

 ; RV64I-NEXT:    or a0, a2, a0
 ; RV64I-NEXT:    srli a2, a0, 1
 ; RV64I-NEXT:    and a2, a2, a1
 ; RV64I-NEXT:    and a0, a0, a1
 ; RV64I-NEXT:    slli a0, a0, 1
 ; RV64I-NEXT:    or a0, a2, a0
 ; RV64I-NEXT:    ret
 ;
-; RV32ZB-LABEL: test_bitreverse_bswap_i64:
-; RV32ZB:       # %bb.0:
-; RV32ZB-NEXT:    srli a2, a0, 4
-; RV32ZB-NEXT:    lui a3, 61681
-; RV32ZB-NEXT:    addi a3, a3, -241
-; RV32ZB-NEXT:    and a2, a2, a3
-; RV32ZB-NEXT:    and a0, a0, a3
-; RV32ZB-NEXT:    slli a0, a0, 4
-; RV32ZB-NEXT:    or a0, a2, a0
-; RV32ZB-NEXT:    srli a2, a0, 2
-; RV32ZB-NEXT:    lui a4, 209715
-; RV32ZB-NEXT:    addi a4, a4, 819
-; RV32ZB-NEXT:    and a2, a2, a4
-; RV32ZB-NEXT:    and a0, a0, a4
-; RV32ZB-NEXT:    slli a0, a0, 2
-; RV32ZB-NEXT:    or a0, a2, a0
-; RV32ZB-NEXT:    srli a2, a0, 1
-; RV32ZB-NEXT:    lui a5, 349525
-; RV32ZB-NEXT:    addi a5, a5, 1365
-; RV32ZB-NEXT:    and a2, a2, a5
-; RV32ZB-NEXT:    and a0, a0, a5
-; RV32ZB-NEXT:    slli a0, a0, 1
-; RV32ZB-NEXT:    or a0, a2, a0
-; RV32ZB-NEXT:    srli a2, a1, 4
-; RV32ZB-NEXT:    and a2, a2, a3
-; RV32ZB-NEXT:    and a1, a1, a3
-; RV32ZB-NEXT:    slli a1, a1, 4
-; RV32ZB-NEXT:    or a1, a2, a1
-; RV32ZB-NEXT:    srli a2, a1, 2
-; RV32ZB-NEXT:    and a2, a2, a4
-; RV32ZB-NEXT:    and a1, a1, a4
-; RV32ZB-NEXT:    slli a1, a1, 2
-; RV32ZB-NEXT:    or a1, a2, a1
-; RV32ZB-NEXT:    srli a2, a1, 1
-; RV32ZB-NEXT:    and a2, a2, a5
-; RV32ZB-NEXT:    and a1, a1, a5
-; RV32ZB-NEXT:    slli a1, a1, 1
-; RV32ZB-NEXT:    or a1, a2, a1
-; RV32ZB-NEXT:    ret
-;
-; RV64ZB-LABEL: test_bitreverse_bswap_i64:
-; RV64ZB:       # %bb.0:
-; RV64ZB-NEXT:    lui a1, %hi(.LCPI12_0)
-; RV64ZB-NEXT:    ld a1, %lo(.LCPI12_0)(a1)
-; RV64ZB-NEXT:    srli a2, a0, 4
-; RV64ZB-NEXT:    and a2, a2, a1
-; RV64ZB-NEXT:    and a0, a0, a1
-; RV64ZB-NEXT:    lui a1, %hi(.LCPI12_1)
-; RV64ZB-NEXT:    ld a1, %lo(.LCPI12_1)(a1)
-; RV64ZB-NEXT:    slli a0, a0, 4
-; RV64ZB-NEXT:    or a0, a2, a0
-; RV64ZB-NEXT:    srli a2, a0, 2
-; RV64ZB-NEXT:    and a2, a2, a1
-; RV64ZB-NEXT:    and a0, a0, a1
-; RV64ZB-NEXT:    lui a1, %hi(.LCPI12_2)
-; RV64ZB-NEXT:    ld a1, %lo(.LCPI12_2)(a1)
-; RV64ZB-NEXT:    slli a0, a0, 2
-; RV64ZB-NEXT:    or a0, a2, a0
-; RV64ZB-NEXT:    srli a2, a0, 1
-; RV64ZB-NEXT:    and a2, a2, a1
-; RV64ZB-NEXT:    and a0, a0, a1
-; RV64ZB-NEXT:    slli a0, a0, 1
-; RV64ZB-NEXT:    or a0, a2, a0
-; RV64ZB-NEXT:    ret
+; RV32ZBB-LABEL: test_bitreverse_bswap_i64:
+; RV32ZBB:       # %bb.0:
+; RV32ZBB-NEXT:    srli a2, a0, 4
+; RV32ZBB-NEXT:    lui a3, 61681
+; RV32ZBB-NEXT:    addi a3, a3, -241
+; RV32ZBB-NEXT:    and a2, a2, a3
+; RV32ZBB-NEXT:    and a0, a0, a3
+; RV32ZBB-NEXT:    slli a0, a0, 4
+; RV32ZBB-NEXT:    or a0, a2, a0
+; RV32ZBB-NEXT:    srli a2, a0, 2
+; RV32ZBB-NEXT:    lui a4, 209715
+; RV32ZBB-NEXT:    addi a4, a4, 819
+; RV32ZBB-NEXT:    and a2, a2, a4
+; RV32ZBB-NEXT:    and a0, a0, a4
+; RV32ZBB-NEXT:    slli a0, a0, 2
+; RV32ZBB-NEXT:    or a0, a2, a0
+; RV32ZBB-NEXT:    srli a2, a0, 1
+; RV32ZBB-NEXT:    lui a5, 349525
+; RV32ZBB-NEXT:    addi a5, a5, 1365
+; RV32ZBB-NEXT:    and a2, a2, a5
+; RV32ZBB-NEXT:    and a0, a0, a5
+; RV32ZBB-NEXT:    slli a0, a0, 1
+; RV32ZBB-NEXT:    or a0, a2, a0
+; RV32ZBB-NEXT:    srli a2, a1, 4
+; RV32ZBB-NEXT:    and a2, a2, a3
+; RV32ZBB-NEXT:    and a1, a1, a3
+; RV32ZBB-NEXT:    slli a1, a1, 4
+; RV32ZBB-NEXT:    or a1, a2, a1
+; RV32ZBB-NEXT:    srli a2, a1, 2
+; RV32ZBB-NEXT:    and a2, a2, a4
+; RV32ZBB-NEXT:    and a1, a1, a4
+; RV32ZBB-NEXT:    slli a1, a1, 2
+; RV32ZBB-NEXT:    or a1, a2, a1
+; RV32ZBB-NEXT:    srli a2, a1, 1
+; RV32ZBB-NEXT:    and a2, a2, a5
+; RV32ZBB-NEXT:    and a1, a1, a5
+; RV32ZBB-NEXT:    slli a1, a1, 1
+; RV32ZBB-NEXT:    or a1, a2, a1
+; RV32ZBB-NEXT:    ret
+;
+; RV64ZBB-LABEL: test_bitreverse_bswap_i64:
+; RV64ZBB:       # %bb.0:
+; RV64ZBB-NEXT:    lui a1, %hi(.LCPI12_0)
+; RV64ZBB-NEXT:    ld a1, %lo(.LCPI12_0)(a1)
+; RV64ZBB-NEXT:    srli a2, a0, 4
+; RV64ZBB-NEXT:    and a2, a2, a1
+; RV64ZBB-NEXT:    and a0, a0, a1
+; RV64ZBB-NEXT:    lui a1, %hi(.LCPI12_1)
+; RV64ZBB-NEXT:    ld a1, %lo(.LCPI12_1)(a1)
+; RV64ZBB-NEXT:    slli a0, a0, 4
+; RV64ZBB-NEXT:    or a0, a2, a0
+; RV64ZBB-NEXT:    srli a2, a0, 2
+; RV64ZBB-NEXT:    and a2, a2, a1
+; RV64ZBB-NEXT:    and a0, a0, a1
+; RV64ZBB-NEXT:    lui a1, %hi(.LCPI12_2)
+; RV64ZBB-NEXT:    ld a1, %lo(.LCPI12_2)(a1)
+; RV64ZBB-NEXT:    slli a0, a0, 2
+; RV64ZBB-NEXT:    or a0, a2, a0
+; RV64ZBB-NEXT:    srli a2, a0, 1
+; RV64ZBB-NEXT:    and a2, a2, a1
+; RV64ZBB-NEXT:    and a0, a0, a1
+; RV64ZBB-NEXT:    slli a0, a0, 1
+; RV64ZBB-NEXT:    or a0, a2, a0
+; RV64ZBB-NEXT:    ret
+;
+; RV32ZBKB-LABEL: test_bitreverse_bswap_i64:
+; RV32ZBKB:       # %bb.0:
+; RV32ZBKB-NEXT:    brev8 a0, a0
+; RV32ZBKB-NEXT:    brev8 a1, a1
+; RV32ZBKB-NEXT:    ret
+;
+; RV64ZBKB-LABEL: test_bitreverse_bswap_i64:
+; RV64ZBKB:       # %bb.0:
+; RV64ZBKB-NEXT:    brev8 a0, a0
+; RV64ZBKB-NEXT:    ret
   %tmp = call i64 @llvm.bitreverse.i64(i64 %a)
   %tmp2 = call i64 @llvm.bswap.i64(i64 %tmp)
   ret i64 %tmp2
 }