Index: llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp
===================================================================
--- llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp
+++ llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp
@@ -9724,6 +9724,17 @@
     return DAG.getNode(ISD::BITREVERSE, DL, VT, BSwap);
   }
 
+  // fold (bswap(srl (bswap c), 8*x)) -> (shl c, 8*x)
+  if (N0->getOpcode() == ISD::SRL && N0.hasOneUse()) {
+    auto *ShAmt = dyn_cast<ConstantSDNode>(N0.getOperand(1));
+    if (ShAmt && ShAmt->getZExtValue() % 8 == 0) {
+      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));
+    }
+  }
+
   // fold (bswap shl(x,c)) -> (zext(bswap(trunc(shl(x,sub(c,bw/2))))))
   // iff x >= bw/2 (i.e. lower half is known zero)
   unsigned BW = VT.getScalarSizeInBits();
Index: llvm/test/CodeGen/RISCV/bswap-srli-bswap.ll
===================================================================
--- llvm/test/CodeGen/RISCV/bswap-srli-bswap.ll
+++ llvm/test/CodeGen/RISCV/bswap-srli-bswap.ll
@@ -69,28 +69,22 @@
 define i16 @test_bswap_srli_8_bswap_i16(i16 %a) nounwind {
 ; RV32I-LABEL: test_bswap_srli_8_bswap_i16:
 ; RV32I:       # %bb.0:
-; RV32I-NEXT:    andi a0, a0, 255
 ; RV32I-NEXT:    slli a0, a0, 8
 ; RV32I-NEXT:    ret
 ;
 ; RV64I-LABEL: test_bswap_srli_8_bswap_i16:
 ; RV64I:       # %bb.0:
-; RV64I-NEXT:    andi a0, a0, 255
 ; RV64I-NEXT:    slli a0, a0, 8
 ; RV64I-NEXT:    ret
 ;
 ; RV32ZB-LABEL: test_bswap_srli_8_bswap_i16:
 ; RV32ZB:       # %bb.0:
-; RV32ZB-NEXT:    andi a0, a0, 255
-; RV32ZB-NEXT:    rev8 a0, a0
-; RV32ZB-NEXT:    srli a0, a0, 16
+; RV32ZB-NEXT:    slli a0, a0, 8
 ; RV32ZB-NEXT:    ret
 ;
 ; RV64ZB-LABEL: test_bswap_srli_8_bswap_i16:
 ; RV64ZB:       # %bb.0:
-; RV64ZB-NEXT:    andi a0, a0, 255
-; RV64ZB-NEXT:    rev8 a0, a0
-; RV64ZB-NEXT:    srli a0, a0, 48
+; RV64ZB-NEXT:    slli a0, a0, 8
 ; RV64ZB-NEXT:    ret
     %1 = call i16 @llvm.bswap.i16(i16 %a)
     %2 = lshr i16 %1, 8
@@ -101,65 +95,22 @@
 define i32 @test_bswap_srli_8_bswap_i32(i32 %a) nounwind {
 ; RV32I-LABEL: test_bswap_srli_8_bswap_i32:
 ; RV32I:       # %bb.0:
-; RV32I-NEXT:    srli a1, a0, 8
-; RV32I-NEXT:    lui a2, 16
-; RV32I-NEXT:    addi a2, a2, -256
-; RV32I-NEXT:    and a1, a1, a2
-; RV32I-NEXT:    srli a2, a0, 24
-; RV32I-NEXT:    or a1, a1, a2
-; RV32I-NEXT:    slli a2, a0, 8
-; RV32I-NEXT:    lui a3, 4080
-; RV32I-NEXT:    and a2, a2, a3
-; RV32I-NEXT:    slli a0, a0, 24
-; RV32I-NEXT:    or a0, a0, a2
-; RV32I-NEXT:    or a0, a0, a1
-; RV32I-NEXT:    srli a1, a0, 8
-; RV32I-NEXT:    slli a1, a1, 24
-; RV32I-NEXT:    and a2, a0, a3
-; RV32I-NEXT:    or a1, a1, a2
-; RV32I-NEXT:    srli a0, a0, 16
-; RV32I-NEXT:    andi a0, a0, -256
-; RV32I-NEXT:    or a0, a0, a1
+; RV32I-NEXT:    slli a0, a0, 8
 ; RV32I-NEXT:    ret
 ;
 ; RV64I-LABEL: test_bswap_srli_8_bswap_i32:
 ; RV64I:       # %bb.0:
-; RV64I-NEXT:    srliw a1, a0, 8
-; RV64I-NEXT:    lui a2, 16
-; RV64I-NEXT:    addiw a2, a2, -256
-; RV64I-NEXT:    and a1, a1, a2
-; RV64I-NEXT:    srliw a2, a0, 24
-; RV64I-NEXT:    or a1, a1, a2
-; RV64I-NEXT:    slli a2, a0, 8
-; RV64I-NEXT:    lui a3, 4080
-; RV64I-NEXT:    and a2, a2, a3
-; RV64I-NEXT:    slliw a0, a0, 24
-; RV64I-NEXT:    or a0, a0, a2
-; RV64I-NEXT:    or a0, a0, a1
-; RV64I-NEXT:    slli a1, a0, 32
-; RV64I-NEXT:    srli a1, a1, 32
-; RV64I-NEXT:    srli a1, a1, 8
-; RV64I-NEXT:    and a2, a0, a3
-; RV64I-NEXT:    slliw a1, a1, 24
-; RV64I-NEXT:    or a1, a1, a2
-; RV64I-NEXT:    srliw a0, a0, 16
-; RV64I-NEXT:    andi a0, a0, -256
-; RV64I-NEXT:    or a0, a0, a1
+; RV64I-NEXT:    slliw a0, a0, 8
 ; RV64I-NEXT:    ret
 ;
 ; RV32ZB-LABEL: test_bswap_srli_8_bswap_i32:
 ; RV32ZB:       # %bb.0:
-; RV32ZB-NEXT:    rev8 a0, a0
-; RV32ZB-NEXT:    srli a0, a0, 8
-; RV32ZB-NEXT:    rev8 a0, a0
+; RV32ZB-NEXT:    slli a0, a0, 8
 ; RV32ZB-NEXT:    ret
 ;
 ; RV64ZB-LABEL: test_bswap_srli_8_bswap_i32:
 ; RV64ZB:       # %bb.0:
-; RV64ZB-NEXT:    rev8 a0, a0
-; RV64ZB-NEXT:    srli a0, a0, 40
-; RV64ZB-NEXT:    rev8 a0, a0
-; RV64ZB-NEXT:    srli a0, a0, 32
+; RV64ZB-NEXT:    slliw a0, a0, 8
 ; RV64ZB-NEXT:    ret
     %1 = call i32 @llvm.bswap.i32(i32 %a)
     %2 = lshr i32 %1, 8
@@ -170,61 +121,22 @@
 define i32 @test_bswap_srli_16_bswap_i32(i32 %a) nounwind {
 ; RV32I-LABEL: test_bswap_srli_16_bswap_i32:
 ; RV32I:       # %bb.0:
-; RV32I-NEXT:    srli a1, a0, 8
-; RV32I-NEXT:    lui a2, 16
-; RV32I-NEXT:    addi a2, a2, -256
-; RV32I-NEXT:    and a1, a1, a2
-; RV32I-NEXT:    srli a2, a0, 24
-; RV32I-NEXT:    or a1, a1, a2
-; RV32I-NEXT:    slli a2, a0, 8
-; RV32I-NEXT:    lui a3, 4080
-; RV32I-NEXT:    and a2, a2, a3
-; RV32I-NEXT:    slli a0, a0, 24
-; RV32I-NEXT:    or a0, a0, a2
-; RV32I-NEXT:    or a0, a0, a1
-; RV32I-NEXT:    srli a1, a0, 16
-; RV32I-NEXT:    slli a1, a1, 24
-; RV32I-NEXT:    srli a0, a0, 24
 ; RV32I-NEXT:    slli a0, a0, 16
-; RV32I-NEXT:    or a0, a1, a0
 ; RV32I-NEXT:    ret
 ;
 ; RV64I-LABEL: test_bswap_srli_16_bswap_i32:
 ; RV64I:       # %bb.0:
-; RV64I-NEXT:    srliw a1, a0, 8
-; RV64I-NEXT:    lui a2, 16
-; RV64I-NEXT:    addiw a2, a2, -256
-; RV64I-NEXT:    and a1, a1, a2
-; RV64I-NEXT:    srliw a2, a0, 24
-; RV64I-NEXT:    or a1, a1, a2
-; RV64I-NEXT:    slli a2, a0, 8
-; RV64I-NEXT:    lui a3, 4080
-; RV64I-NEXT:    and a2, a2, a3
-; RV64I-NEXT:    slli a0, a0, 24
-; RV64I-NEXT:    or a0, a0, a2
-; RV64I-NEXT:    or a0, a0, a1
-; RV64I-NEXT:    slli a0, a0, 32
-; RV64I-NEXT:    srli a0, a0, 32
-; RV64I-NEXT:    srli a1, a0, 16
-; RV64I-NEXT:    srliw a0, a0, 24
-; RV64I-NEXT:    slli a0, a0, 16
-; RV64I-NEXT:    slliw a1, a1, 24
-; RV64I-NEXT:    or a0, a1, a0
+; RV64I-NEXT:    slliw a0, a0, 16
 ; RV64I-NEXT:    ret
 ;
 ; RV32ZB-LABEL: test_bswap_srli_16_bswap_i32:
 ; RV32ZB:       # %bb.0:
-; RV32ZB-NEXT:    rev8 a0, a0
-; RV32ZB-NEXT:    srli a0, a0, 16
-; RV32ZB-NEXT:    rev8 a0, a0
+; RV32ZB-NEXT:    slli a0, a0, 16
 ; RV32ZB-NEXT:    ret
 ;
 ; RV64ZB-LABEL: test_bswap_srli_16_bswap_i32:
 ; RV64ZB:       # %bb.0:
-; RV64ZB-NEXT:    rev8 a0, a0
-; RV64ZB-NEXT:    srli a0, a0, 48
-; RV64ZB-NEXT:    rev8 a0, a0
-; RV64ZB-NEXT:    srli a0, a0, 32
+; RV64ZB-NEXT:    slliw a0, a0, 16
 ; RV64ZB-NEXT:    ret
     %1 = call i32 @llvm.bswap.i32(i32 %a)
     %2 = lshr i32 %1, 16
@@ -245,15 +157,12 @@
 ;
 ; RV32ZB-LABEL: test_bswap_srli_24_bswap_i32:
 ; RV32ZB:       # %bb.0:
-; RV32ZB-NEXT:    andi a0, a0, 255
-; RV32ZB-NEXT:    rev8 a0, a0
+; RV32ZB-NEXT:    slli a0, a0, 24
 ; RV32ZB-NEXT:    ret
 ;
 ; RV64ZB-LABEL: test_bswap_srli_24_bswap_i32:
 ; RV64ZB:       # %bb.0:
-; RV64ZB-NEXT:    andi a0, a0, 255
-; RV64ZB-NEXT:    rev8 a0, a0
-; RV64ZB-NEXT:    srli a0, a0, 32
+; RV64ZB-NEXT:    slliw a0, a0, 24
 ; RV64ZB-NEXT:    ret
     %1 = call i32 @llvm.bswap.i32(i32 %a)
     %2 = lshr i32 %1, 24
@@ -264,76 +173,24 @@
 define i64 @test_bswap_srli_48_bswap_i64(i64 %a) nounwind {
 ; RV32I-LABEL: test_bswap_srli_48_bswap_i64:
 ; RV32I:       # %bb.0:
-; RV32I-NEXT:    srli a1, a0, 8
-; RV32I-NEXT:    lui a2, 16
-; RV32I-NEXT:    addi a2, a2, -256
-; RV32I-NEXT:    and a1, a1, a2
-; RV32I-NEXT:    srli a2, a0, 24
-; RV32I-NEXT:    or a1, a1, a2
-; RV32I-NEXT:    slli a2, a0, 8
-; RV32I-NEXT:    lui a3, 4080
-; RV32I-NEXT:    and a2, a2, a3
-; RV32I-NEXT:    slli a0, a0, 24
-; RV32I-NEXT:    or a0, a0, a2
-; RV32I-NEXT:    or a0, a0, a1
-; RV32I-NEXT:    srli a1, a0, 16
-; RV32I-NEXT:    slli a1, a1, 24
-; RV32I-NEXT:    srli a0, a0, 24
-; RV32I-NEXT:    slli a0, a0, 16
-; RV32I-NEXT:    or a1, a1, a0
+; RV32I-NEXT:    slli a1, a0, 16
 ; RV32I-NEXT:    li a0, 0
 ; RV32I-NEXT:    ret
 ;
 ; RV64I-LABEL: test_bswap_srli_48_bswap_i64:
 ; RV64I:       # %bb.0:
-; RV64I-NEXT:    srli a1, a0, 24
-; RV64I-NEXT:    lui a2, 4080
-; RV64I-NEXT:    and a1, a1, a2
-; RV64I-NEXT:    srli a2, a0, 8
-; RV64I-NEXT:    li a3, 255
-; RV64I-NEXT:    slli a4, a3, 24
-; RV64I-NEXT:    and a2, a2, a4
-; RV64I-NEXT:    or a1, a2, a1
-; RV64I-NEXT:    srli a2, a0, 40
-; RV64I-NEXT:    lui a4, 16
-; RV64I-NEXT:    addiw a4, a4, -256
-; RV64I-NEXT:    and a2, a2, a4
-; RV64I-NEXT:    srli a4, a0, 56
-; RV64I-NEXT:    or a2, a2, a4
-; RV64I-NEXT:    or a1, a1, a2
-; RV64I-NEXT:    slli a2, a0, 24
-; RV64I-NEXT:    slli a4, a3, 40
-; RV64I-NEXT:    and a2, a2, a4
-; RV64I-NEXT:    srliw a4, a0, 24
-; RV64I-NEXT:    slli a4, a4, 32
-; RV64I-NEXT:    or a2, a2, a4
-; RV64I-NEXT:    slli a4, a0, 40
-; RV64I-NEXT:    slli a3, a3, 48
-; RV64I-NEXT:    and a3, a4, a3
-; RV64I-NEXT:    slli a0, a0, 56
-; RV64I-NEXT:    or a0, a0, a3
-; RV64I-NEXT:    or a0, a0, a2
-; RV64I-NEXT:    or a0, a0, a1
-; RV64I-NEXT:    srli a1, a0, 48
-; RV64I-NEXT:    slli a1, a1, 56
-; RV64I-NEXT:    srli a0, a0, 56
 ; RV64I-NEXT:    slli a0, a0, 48
-; RV64I-NEXT:    or a0, a1, a0
 ; RV64I-NEXT:    ret
 ;
 ; RV32ZB-LABEL: test_bswap_srli_48_bswap_i64:
 ; RV32ZB:       # %bb.0:
-; RV32ZB-NEXT:    rev8 a0, a0
-; RV32ZB-NEXT:    srli a0, a0, 16
-; RV32ZB-NEXT:    rev8 a1, a0
+; RV32ZB-NEXT:    slli a1, a0, 16
 ; RV32ZB-NEXT:    li a0, 0
 ; RV32ZB-NEXT:    ret
 ;
 ; RV64ZB-LABEL: test_bswap_srli_48_bswap_i64:
 ; RV64ZB:       # %bb.0:
-; RV64ZB-NEXT:    rev8 a0, a0
-; RV64ZB-NEXT:    srli a0, a0, 48
-; RV64ZB-NEXT:    rev8 a0, a0
+; RV64ZB-NEXT:    slli a0, a0, 48
 ; RV64ZB-NEXT:    ret
     %1 = call i64 @llvm.bswap.i64(i64 %a)
     %2 = lshr i64 %1, 48
Index: llvm/test/CodeGen/X86/combine-bswap.ll
===================================================================
--- llvm/test/CodeGen/X86/combine-bswap.ll
+++ llvm/test/CodeGen/X86/combine-bswap.ll
@@ -38,22 +38,18 @@
   ret i32 %c
 }
 
-; TODO: fold (bswap(srl (bswap c), x)) -> (shl c, x)
+; fold (bswap(srl (bswap c), x)) -> (shl c, x)
 define i32 @test_bswap_srli_8_bswap_i32(i32 %a) nounwind {
 ; X86-LABEL: test_bswap_srli_8_bswap_i32:
 ; X86:       # %bb.0:
 ; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax
-; X86-NEXT:    bswapl %eax
-; X86-NEXT:    shrl $8, %eax
-; X86-NEXT:    bswapl %eax
+; X86-NEXT:    shll $8, %eax
 ; X86-NEXT:    retl
 ;
 ; X64-LABEL: test_bswap_srli_8_bswap_i32:
 ; X64:       # %bb.0:
 ; X64-NEXT:    movl %edi, %eax
-; X64-NEXT:    bswapl %eax
-; X64-NEXT:    shrl $8, %eax
-; X64-NEXT:    bswapl %eax
+; X64-NEXT:    shll $8, %eax
 ; X64-NEXT:    retq
     %1 = call i32 @llvm.bswap.i32(i32 %a)
     %2 = lshr i32 %1, 8