Index: llvm/lib/CodeGen/SelectionDAG/TargetLowering.cpp
===================================================================
--- llvm/lib/CodeGen/SelectionDAG/TargetLowering.cpp
+++ llvm/lib/CodeGen/SelectionDAG/TargetLowering.cpp
@@ -1544,6 +1544,16 @@
     Known.Zero = Known2.Zero.reverseBits();
     break;
   }
+  case ISD::BSWAP: {
+    SDValue Src = Op.getOperand(0);
+    APInt DemandedSrcBits = DemandedBits.byteSwap();
+    if (SimplifyDemandedBits(Src, DemandedSrcBits, DemandedElts, Known2, TLO,
+                             Depth + 1))
+      return true;
+    Known.One = Known2.One.byteSwap();
+    Known.Zero = Known2.Zero.byteSwap();
+    break;
+  }
   case ISD::SIGN_EXTEND_INREG: {
     SDValue Op0 = Op.getOperand(0);
     EVT ExVT = cast<VTSDNode>(Op.getOperand(1))->getVT();
Index: llvm/test/CodeGen/AArch64/arm64-rev.ll
===================================================================
--- llvm/test/CodeGen/AArch64/arm64-rev.ll
+++ llvm/test/CodeGen/AArch64/arm64-rev.ll
@@ -39,8 +39,8 @@
 define i32 @test_rev_w_srl16(i16 %a) {
 ; CHECK-LABEL: test_rev_w_srl16:
 ; CHECK:       // %bb.0: // %entry
-; CHECK-NEXT:    and w8, w0, #0xffff
-; CHECK-NEXT:    rev16 w0, w8
+; CHECK-NEXT:    rev w8, w0
+; CHECK-NEXT:    lsr w0, w8, #16
 ; CHECK-NEXT:    ret
 ;
 ; FALLBACK-LABEL: test_rev_w_srl16:
@@ -60,7 +60,8 @@
 ; CHECK-LABEL: test_rev_w_srl16_load:
 ; CHECK:       // %bb.0: // %entry
 ; CHECK-NEXT:    ldrh w8, [x0]
-; CHECK-NEXT:    rev16 w0, w8
+; CHECK-NEXT:    rev w8, w8
+; CHECK-NEXT:    lsr w0, w8, #16
 ; CHECK-NEXT:    ret
 ;
 ; FALLBACK-LABEL: test_rev_w_srl16_load:
@@ -106,8 +107,9 @@
 define i64 @test_rev_x_srl32(i32 %a) {
 ; CHECK-LABEL: test_rev_x_srl32:
 ; CHECK:       // %bb.0: // %entry
-; CHECK-NEXT:    mov w8, w0
-; CHECK-NEXT:    rev32 x0, x8
+; CHECK-NEXT:    // kill: def $w0 killed $w0 def $x0
+; CHECK-NEXT:    rev x8, x0
+; CHECK-NEXT:    lsr x0, x8, #32
 ; CHECK-NEXT:    ret
 ;
 ; FALLBACK-LABEL: test_rev_x_srl32:
@@ -128,7 +130,8 @@
 ; CHECK-LABEL: test_rev_x_srl32_load:
 ; CHECK:       // %bb.0: // %entry
 ; CHECK-NEXT:    ldr w8, [x0]
-; CHECK-NEXT:    rev32 x0, x8
+; CHECK-NEXT:    rev x8, x8
+; CHECK-NEXT:    lsr x0, x8, #32
 ; CHECK-NEXT:    ret
 ;
 ; FALLBACK-LABEL: test_rev_x_srl32_load:
Index: llvm/test/CodeGen/AMDGPU/bswap.ll
===================================================================
--- llvm/test/CodeGen/AMDGPU/bswap.ll
+++ llvm/test/CodeGen/AMDGPU/bswap.ll
@@ -732,7 +732,6 @@
 ; VI-LABEL: missing_truncate_promote_bswap:
 ; VI:       ; %bb.0: ; %bb
 ; VI-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
-; VI-NEXT:    v_and_b32_e32 v0, 0xffff, v0
 ; VI-NEXT:    v_alignbit_b32 v1, v0, v0, 8
 ; VI-NEXT:    v_alignbit_b32 v0, v0, v0, 24
 ; VI-NEXT:    s_mov_b32 s4, 0xff00ff
Index: llvm/test/CodeGen/X86/combine-bswap.ll
===================================================================
--- llvm/test/CodeGen/X86/combine-bswap.ll
+++ llvm/test/CodeGen/X86/combine-bswap.ll
@@ -40,8 +40,7 @@
 define i32 @test_demandedbits_bswap(i32 %a0) nounwind {
 ; X86-LABEL: test_demandedbits_bswap:
 ; X86:       # %bb.0:
-; X86-NEXT:    movl $-16777216, %eax # imm = 0xFF000000
-; X86-NEXT:    orl {{[0-9]+}}(%esp), %eax
+; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax
 ; X86-NEXT:    bswapl %eax
 ; X86-NEXT:    andl $-65536, %eax # imm = 0xFFFF0000
 ; X86-NEXT:    retl
@@ -49,7 +48,6 @@
 ; X64-LABEL: test_demandedbits_bswap:
 ; X64:       # %bb.0:
 ; X64-NEXT:    movl %edi, %eax
-; X64-NEXT:    orl $-16777216, %eax # imm = 0xFF000000
 ; X64-NEXT:    bswapl %eax
 ; X64-NEXT:    andl $-65536, %eax # imm = 0xFFFF0000
 ; X64-NEXT:    retq