Index: llvm/lib/Target/X86/X86InstrCompiler.td
===================================================================
--- llvm/lib/Target/X86/X86InstrCompiler.td
+++ llvm/lib/Target/X86/X86InstrCompiler.td
@@ -896,15 +896,15 @@
 multiclass ATOMIC_LOGIC_OP_RM<bits<8> Opc8, string s> {
   let Defs = [EFLAGS], mayLoad = 1, mayStore = 1, isCodeGenOnly = 1,
       SchedRW = [WriteBitTestSetRegRMW]  in {
-    def 16rm : Ii8<Opc8, MRMDestMem, (outs), (ins i16mem:$src1, GR16:$src2),
+    def 16rm : I<Opc8, MRMDestMem, (outs), (ins i16mem:$src1, GR16:$src2),
                   !strconcat(s, "{w}\t{$src2, $src1|$src1, $src2}"),
                   [(set EFLAGS, (!cast<SDNode>("x86_rm_" # s) addr:$src1, GR16:$src2))]>,
                OpSize16, TB, LOCK;
-    def 32rm : Ii8<Opc8, MRMDestMem, (outs), (ins i32mem:$src1, GR32:$src2),
+    def 32rm : I<Opc8, MRMDestMem, (outs), (ins i32mem:$src1, GR32:$src2),
                   !strconcat(s, "{l}\t{$src2, $src1|$src1, $src2}"),
                   [(set EFLAGS, (!cast<SDNode>("x86_rm_" # s) addr:$src1, GR32:$src2))]>,
                OpSize32, TB, LOCK;
-    def 64rm : RIi8<Opc8, MRMDestMem, (outs), (ins i64mem:$src1, GR64:$src2),
+    def 64rm : RI<Opc8, MRMDestMem, (outs), (ins i64mem:$src1, GR64:$src2),
                    !strconcat(s, "{q}\t{$src2, $src1|$src1, $src2}"),
                    [(set EFLAGS, (!cast<SDNode>("x86_rm_" # s) addr:$src1, GR64:$src2))]>,
                TB, LOCK;
Index: llvm/test/CodeGen/X86/atomic-rm-bit-test-64.ll
===================================================================
--- llvm/test/CodeGen/X86/atomic-rm-bit-test-64.ll
+++ llvm/test/CodeGen/X86/atomic-rm-bit-test-64.ll
@@ -1,17 +1,17 @@
 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
-; RUN: llc -mtriple=x86_64-unknown-linux-gnu < %s | FileCheck %s
+; RUN: llc -mtriple=x86_64-unknown-linux-gnu -show-mc-encoding < %s | FileCheck %s
 
 define i64 @atomic_shl1_xor_64_gpr_val(ptr %v, i64 %c) nounwind {
 ; CHECK-LABEL: atomic_shl1_xor_64_gpr_val:
 ; CHECK:       # %bb.0: # %entry
-; CHECK-NEXT:    movq %rsi, %rcx
-; CHECK-NEXT:    andl $63, %ecx
-; CHECK-NEXT:    xorl %eax, %eax
-; CHECK-NEXT:    lock btcq %rcx, (%rdi)
-; CHECK-NEXT:    setb %al
+; CHECK-NEXT:    movq %rsi, %rcx # encoding: [0x48,0x89,0xf1]
+; CHECK-NEXT:    andl $63, %ecx # encoding: [0x83,0xe1,0x3f]
+; CHECK-NEXT:    xorl %eax, %eax # encoding: [0x31,0xc0]
+; CHECK-NEXT:    lock btcq %rcx, (%rdi) # encoding: [0xf0,0x48,0x0f,0xbb,0x0f]
+; CHECK-NEXT:    setb %al # encoding: [0x0f,0x92,0xc0]
 ; CHECK-NEXT:    # kill: def $cl killed $cl killed $rcx
-; CHECK-NEXT:    shlq %cl, %rax
-; CHECK-NEXT:    retq
+; CHECK-NEXT:    shlq %cl, %rax # encoding: [0x48,0xd3,0xe0]
+; CHECK-NEXT:    retq # encoding: [0xc3]
 entry:
   %shl = shl nuw i64 1, %c
   %0 = atomicrmw xor ptr %v, i64 %shl monotonic, align 8
@@ -22,21 +22,22 @@
 define i64 @atomic_shl2_xor_64_gpr_val(ptr %v, i64 %c) nounwind {
 ; CHECK-LABEL: atomic_shl2_xor_64_gpr_val:
 ; CHECK:       # %bb.0: # %entry
-; CHECK-NEXT:    movq %rsi, %rcx
-; CHECK-NEXT:    movl $2, %edx
+; CHECK-NEXT:    movq %rsi, %rcx # encoding: [0x48,0x89,0xf1]
+; CHECK-NEXT:    movl $2, %edx # encoding: [0xba,0x02,0x00,0x00,0x00]
 ; CHECK-NEXT:    # kill: def $cl killed $cl killed $rcx
-; CHECK-NEXT:    shlq %cl, %rdx
-; CHECK-NEXT:    movq (%rdi), %rax
+; CHECK-NEXT:    shlq %cl, %rdx # encoding: [0x48,0xd3,0xe2]
+; CHECK-NEXT:    movq (%rdi), %rax # encoding: [0x48,0x8b,0x07]
 ; CHECK-NEXT:    .p2align 4, 0x90
 ; CHECK-NEXT:  .LBB1_1: # %atomicrmw.start
 ; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
-; CHECK-NEXT:    movq %rax, %rcx
-; CHECK-NEXT:    xorq %rdx, %rcx
-; CHECK-NEXT:    lock cmpxchgq %rcx, (%rdi)
-; CHECK-NEXT:    jne .LBB1_1
+; CHECK-NEXT:    movq %rax, %rcx # encoding: [0x48,0x89,0xc1]
+; CHECK-NEXT:    xorq %rdx, %rcx # encoding: [0x48,0x31,0xd1]
+; CHECK-NEXT:    lock cmpxchgq %rcx, (%rdi) # encoding: [0xf0,0x48,0x0f,0xb1,0x0f]
+; CHECK-NEXT:    jne .LBB1_1 # encoding: [0x75,A]
+; CHECK-NEXT:    # fixup A - offset: 1, value: .LBB1_1-1, kind: FK_PCRel_1
 ; CHECK-NEXT:  # %bb.2: # %atomicrmw.end
-; CHECK-NEXT:    andq %rdx, %rax
-; CHECK-NEXT:    retq
+; CHECK-NEXT:    andq %rdx, %rax # encoding: [0x48,0x21,0xd0]
+; CHECK-NEXT:    retq # encoding: [0xc3]
 entry:
   %shl = shl i64 2, %c
   %0 = atomicrmw xor ptr %v, i64 %shl monotonic, align 8
@@ -47,24 +48,25 @@
 define i64 @atomic_shl1_neq_xor_64_gpr_val(ptr %v, i64 %c) nounwind {
 ; CHECK-LABEL: atomic_shl1_neq_xor_64_gpr_val:
 ; CHECK:       # %bb.0: # %entry
-; CHECK-NEXT:    movq %rsi, %rcx
-; CHECK-NEXT:    movl $1, %edx
-; CHECK-NEXT:    shlq %cl, %rdx
-; CHECK-NEXT:    movq (%rdi), %rax
+; CHECK-NEXT:    movq %rsi, %rcx # encoding: [0x48,0x89,0xf1]
+; CHECK-NEXT:    movl $1, %edx # encoding: [0xba,0x01,0x00,0x00,0x00]
+; CHECK-NEXT:    shlq %cl, %rdx # encoding: [0x48,0xd3,0xe2]
+; CHECK-NEXT:    movq (%rdi), %rax # encoding: [0x48,0x8b,0x07]
 ; CHECK-NEXT:    .p2align 4, 0x90
 ; CHECK-NEXT:  .LBB2_1: # %atomicrmw.start
 ; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
-; CHECK-NEXT:    movq %rax, %rsi
-; CHECK-NEXT:    xorq %rdx, %rsi
-; CHECK-NEXT:    lock cmpxchgq %rsi, (%rdi)
-; CHECK-NEXT:    jne .LBB2_1
+; CHECK-NEXT:    movq %rax, %rsi # encoding: [0x48,0x89,0xc6]
+; CHECK-NEXT:    xorq %rdx, %rsi # encoding: [0x48,0x31,0xd6]
+; CHECK-NEXT:    lock cmpxchgq %rsi, (%rdi) # encoding: [0xf0,0x48,0x0f,0xb1,0x37]
+; CHECK-NEXT:    jne .LBB2_1 # encoding: [0x75,A]
+; CHECK-NEXT:    # fixup A - offset: 1, value: .LBB2_1-1, kind: FK_PCRel_1
 ; CHECK-NEXT:  # %bb.2: # %atomicrmw.end
-; CHECK-NEXT:    incb %cl
-; CHECK-NEXT:    movl $1, %edx
+; CHECK-NEXT:    incb %cl # encoding: [0xfe,0xc1]
+; CHECK-NEXT:    movl $1, %edx # encoding: [0xba,0x01,0x00,0x00,0x00]
 ; CHECK-NEXT:    # kill: def $cl killed $cl killed $rcx
-; CHECK-NEXT:    shlq %cl, %rdx
-; CHECK-NEXT:    andq %rdx, %rax
-; CHECK-NEXT:    retq
+; CHECK-NEXT:    shlq %cl, %rdx # encoding: [0x48,0xd3,0xe2]
+; CHECK-NEXT:    andq %rdx, %rax # encoding: [0x48,0x21,0xd0]
+; CHECK-NEXT:    retq # encoding: [0xc3]
 entry:
   %shl = shl nuw i64 1, %c
   %0 = atomicrmw xor ptr %v, i64 %shl monotonic, align 8
@@ -77,14 +79,14 @@
 define i64 @atomic_shl1_small_mask_xor_64_gpr_val(ptr %v, i64 %c) nounwind {
 ; CHECK-LABEL: atomic_shl1_small_mask_xor_64_gpr_val:
 ; CHECK:       # %bb.0: # %entry
-; CHECK-NEXT:    movq %rsi, %rcx
-; CHECK-NEXT:    andl $31, %ecx
-; CHECK-NEXT:    xorl %eax, %eax
-; CHECK-NEXT:    lock btcq %rcx, (%rdi)
-; CHECK-NEXT:    setb %al
+; CHECK-NEXT:    movq %rsi, %rcx # encoding: [0x48,0x89,0xf1]
+; CHECK-NEXT:    andl $31, %ecx # encoding: [0x83,0xe1,0x1f]
+; CHECK-NEXT:    xorl %eax, %eax # encoding: [0x31,0xc0]
+; CHECK-NEXT:    lock btcq %rcx, (%rdi) # encoding: [0xf0,0x48,0x0f,0xbb,0x0f]
+; CHECK-NEXT:    setb %al # encoding: [0x0f,0x92,0xc0]
 ; CHECK-NEXT:    # kill: def $cl killed $cl killed $rcx
-; CHECK-NEXT:    shlq %cl, %rax
-; CHECK-NEXT:    retq
+; CHECK-NEXT:    shlq %cl, %rax # encoding: [0x48,0xd3,0xe0]
+; CHECK-NEXT:    retq # encoding: [0xc3]
 entry:
   %rem = and i64 %c, 31
   %shl = shl nuw nsw i64 1, %rem
@@ -96,14 +98,14 @@
 define i64 @atomic_shl1_mask0_xor_64_gpr_val(ptr %v, i64 %c) nounwind {
 ; CHECK-LABEL: atomic_shl1_mask0_xor_64_gpr_val:
 ; CHECK:       # %bb.0: # %entry
-; CHECK-NEXT:    movq %rsi, %rcx
-; CHECK-NEXT:    andl $63, %ecx
-; CHECK-NEXT:    xorl %eax, %eax
-; CHECK-NEXT:    lock btcq %rcx, (%rdi)
-; CHECK-NEXT:    setb %al
+; CHECK-NEXT:    movq %rsi, %rcx # encoding: [0x48,0x89,0xf1]
+; CHECK-NEXT:    andl $63, %ecx # encoding: [0x83,0xe1,0x3f]
+; CHECK-NEXT:    xorl %eax, %eax # encoding: [0x31,0xc0]
+; CHECK-NEXT:    lock btcq %rcx, (%rdi) # encoding: [0xf0,0x48,0x0f,0xbb,0x0f]
+; CHECK-NEXT:    setb %al # encoding: [0x0f,0x92,0xc0]
 ; CHECK-NEXT:    # kill: def $cl killed $cl killed $rcx
-; CHECK-NEXT:    shlq %cl, %rax
-; CHECK-NEXT:    retq
+; CHECK-NEXT:    shlq %cl, %rax # encoding: [0x48,0xd3,0xe0]
+; CHECK-NEXT:    retq # encoding: [0xc3]
 entry:
   %rem = and i64 %c, 63
   %shl = shl nuw i64 1, %rem
@@ -116,14 +118,14 @@
 define i64 @atomic_shl1_mask1_xor_64_gpr_val(ptr %v, i64 %c) nounwind {
 ; CHECK-LABEL: atomic_shl1_mask1_xor_64_gpr_val:
 ; CHECK:       # %bb.0: # %entry
-; CHECK-NEXT:    movq %rsi, %rcx
-; CHECK-NEXT:    andl $63, %ecx
-; CHECK-NEXT:    xorl %eax, %eax
-; CHECK-NEXT:    lock btcq %rcx, (%rdi)
-; CHECK-NEXT:    setb %al
+; CHECK-NEXT:    movq %rsi, %rcx # encoding: [0x48,0x89,0xf1]
+; CHECK-NEXT:    andl $63, %ecx # encoding: [0x83,0xe1,0x3f]
+; CHECK-NEXT:    xorl %eax, %eax # encoding: [0x31,0xc0]
+; CHECK-NEXT:    lock btcq %rcx, (%rdi) # encoding: [0xf0,0x48,0x0f,0xbb,0x0f]
+; CHECK-NEXT:    setb %al # encoding: [0x0f,0x92,0xc0]
 ; CHECK-NEXT:    # kill: def $cl killed $cl killed $rcx
-; CHECK-NEXT:    shlq %cl, %rax
-; CHECK-NEXT:    retq
+; CHECK-NEXT:    shlq %cl, %rax # encoding: [0x48,0xd3,0xe0]
+; CHECK-NEXT:    retq # encoding: [0xc3]
 entry:
   %shl = shl nuw i64 1, %c
   %0 = atomicrmw xor ptr %v, i64 %shl monotonic, align 8
@@ -136,14 +138,14 @@
 define i64 @atomic_shl1_mask01_xor_64_gpr_val(ptr %v, i64 %c) nounwind {
 ; CHECK-LABEL: atomic_shl1_mask01_xor_64_gpr_val:
 ; CHECK:       # %bb.0: # %entry
-; CHECK-NEXT:    movq %rsi, %rcx
-; CHECK-NEXT:    andl $63, %ecx
-; CHECK-NEXT:    xorl %eax, %eax
-; CHECK-NEXT:    lock btcq %rcx, (%rdi)
-; CHECK-NEXT:    setb %al
+; CHECK-NEXT:    movq %rsi, %rcx # encoding: [0x48,0x89,0xf1]
+; CHECK-NEXT:    andl $63, %ecx # encoding: [0x83,0xe1,0x3f]
+; CHECK-NEXT:    xorl %eax, %eax # encoding: [0x31,0xc0]
+; CHECK-NEXT:    lock btcq %rcx, (%rdi) # encoding: [0xf0,0x48,0x0f,0xbb,0x0f]
+; CHECK-NEXT:    setb %al # encoding: [0x0f,0x92,0xc0]
 ; CHECK-NEXT:    # kill: def $cl killed $cl killed $rcx
-; CHECK-NEXT:    shlq %cl, %rax
-; CHECK-NEXT:    retq
+; CHECK-NEXT:    shlq %cl, %rax # encoding: [0x48,0xd3,0xe0]
+; CHECK-NEXT:    retq # encoding: [0xc3]
 entry:
   %rem = and i64 %c, 63
   %shl = shl nuw i64 1, %rem
@@ -155,20 +157,21 @@
 define i64 @atomic_blsi_xor_64_gpr_val(ptr %v, i64 %c) nounwind {
 ; CHECK-LABEL: atomic_blsi_xor_64_gpr_val:
 ; CHECK:       # %bb.0: # %entry
-; CHECK-NEXT:    movq %rsi, %rcx
-; CHECK-NEXT:    negq %rcx
-; CHECK-NEXT:    andq %rsi, %rcx
-; CHECK-NEXT:    movq (%rdi), %rax
+; CHECK-NEXT:    movq %rsi, %rcx # encoding: [0x48,0x89,0xf1]
+; CHECK-NEXT:    negq %rcx # encoding: [0x48,0xf7,0xd9]
+; CHECK-NEXT:    andq %rsi, %rcx # encoding: [0x48,0x21,0xf1]
+; CHECK-NEXT:    movq (%rdi), %rax # encoding: [0x48,0x8b,0x07]
 ; CHECK-NEXT:    .p2align 4, 0x90
 ; CHECK-NEXT:  .LBB7_1: # %atomicrmw.start
 ; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
-; CHECK-NEXT:    movq %rax, %rdx
-; CHECK-NEXT:    xorq %rcx, %rdx
-; CHECK-NEXT:    lock cmpxchgq %rdx, (%rdi)
-; CHECK-NEXT:    jne .LBB7_1
+; CHECK-NEXT:    movq %rax, %rdx # encoding: [0x48,0x89,0xc2]
+; CHECK-NEXT:    xorq %rcx, %rdx # encoding: [0x48,0x31,0xca]
+; CHECK-NEXT:    lock cmpxchgq %rdx, (%rdi) # encoding: [0xf0,0x48,0x0f,0xb1,0x17]
+; CHECK-NEXT:    jne .LBB7_1 # encoding: [0x75,A]
+; CHECK-NEXT:    # fixup A - offset: 1, value: .LBB7_1-1, kind: FK_PCRel_1
 ; CHECK-NEXT:  # %bb.2: # %atomicrmw.end
-; CHECK-NEXT:    andq %rcx, %rax
-; CHECK-NEXT:    retq
+; CHECK-NEXT:    andq %rcx, %rax # encoding: [0x48,0x21,0xc8]
+; CHECK-NEXT:    retq # encoding: [0xc3]
 entry:
   %sub = sub i64 0, %c
   %and = and i64 %sub, %c
@@ -180,23 +183,24 @@
 define i64 @atomic_shl1_xor_64_gpr_valz(ptr %v, i64 %c) nounwind {
 ; CHECK-LABEL: atomic_shl1_xor_64_gpr_valz:
 ; CHECK:       # %bb.0: # %entry
-; CHECK-NEXT:    movq %rsi, %rcx
-; CHECK-NEXT:    movl $1, %edx
-; CHECK-NEXT:    shlq %cl, %rdx
-; CHECK-NEXT:    movq (%rdi), %rax
+; CHECK-NEXT:    movq %rsi, %rcx # encoding: [0x48,0x89,0xf1]
+; CHECK-NEXT:    movl $1, %edx # encoding: [0xba,0x01,0x00,0x00,0x00]
+; CHECK-NEXT:    shlq %cl, %rdx # encoding: [0x48,0xd3,0xe2]
+; CHECK-NEXT:    movq (%rdi), %rax # encoding: [0x48,0x8b,0x07]
 ; CHECK-NEXT:    .p2align 4, 0x90
 ; CHECK-NEXT:  .LBB8_1: # %atomicrmw.start
 ; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
-; CHECK-NEXT:    movq %rax, %rsi
-; CHECK-NEXT:    xorq %rdx, %rsi
-; CHECK-NEXT:    lock cmpxchgq %rsi, (%rdi)
-; CHECK-NEXT:    jne .LBB8_1
+; CHECK-NEXT:    movq %rax, %rsi # encoding: [0x48,0x89,0xc6]
+; CHECK-NEXT:    xorq %rdx, %rsi # encoding: [0x48,0x31,0xd6]
+; CHECK-NEXT:    lock cmpxchgq %rsi, (%rdi) # encoding: [0xf0,0x48,0x0f,0xb1,0x37]
+; CHECK-NEXT:    jne .LBB8_1 # encoding: [0x75,A]
+; CHECK-NEXT:    # fixup A - offset: 1, value: .LBB8_1-1, kind: FK_PCRel_1
 ; CHECK-NEXT:  # %bb.2: # %atomicrmw.end
-; CHECK-NEXT:    xorl %edx, %edx
-; CHECK-NEXT:    btq %rcx, %rax
-; CHECK-NEXT:    setae %dl
-; CHECK-NEXT:    movq %rdx, %rax
-; CHECK-NEXT:    retq
+; CHECK-NEXT:    xorl %edx, %edx # encoding: [0x31,0xd2]
+; CHECK-NEXT:    btq %rcx, %rax # encoding: [0x48,0x0f,0xa3,0xc8]
+; CHECK-NEXT:    setae %dl # encoding: [0x0f,0x93,0xc2]
+; CHECK-NEXT:    movq %rdx, %rax # encoding: [0x48,0x89,0xd0]
+; CHECK-NEXT:    retq # encoding: [0xc3]
 entry:
   %shl = shl nuw i64 1, %c
   %0 = atomicrmw xor ptr %v, i64 %shl monotonic, align 8
@@ -209,24 +213,25 @@
 define i64 @atomic_shl2_xor_64_gpr_valz(ptr %v, i64 %c) nounwind {
 ; CHECK-LABEL: atomic_shl2_xor_64_gpr_valz:
 ; CHECK:       # %bb.0: # %entry
-; CHECK-NEXT:    movq %rsi, %rcx
-; CHECK-NEXT:    movl $2, %edx
+; CHECK-NEXT:    movq %rsi, %rcx # encoding: [0x48,0x89,0xf1]
+; CHECK-NEXT:    movl $2, %edx # encoding: [0xba,0x02,0x00,0x00,0x00]
 ; CHECK-NEXT:    # kill: def $cl killed $cl killed $rcx
-; CHECK-NEXT:    shlq %cl, %rdx
-; CHECK-NEXT:    movq (%rdi), %rax
+; CHECK-NEXT:    shlq %cl, %rdx # encoding: [0x48,0xd3,0xe2]
+; CHECK-NEXT:    movq (%rdi), %rax # encoding: [0x48,0x8b,0x07]
 ; CHECK-NEXT:    .p2align 4, 0x90
 ; CHECK-NEXT:  .LBB9_1: # %atomicrmw.start
 ; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
-; CHECK-NEXT:    movq %rax, %rcx
-; CHECK-NEXT:    xorq %rdx, %rcx
-; CHECK-NEXT:    lock cmpxchgq %rcx, (%rdi)
-; CHECK-NEXT:    jne .LBB9_1
+; CHECK-NEXT:    movq %rax, %rcx # encoding: [0x48,0x89,0xc1]
+; CHECK-NEXT:    xorq %rdx, %rcx # encoding: [0x48,0x31,0xd1]
+; CHECK-NEXT:    lock cmpxchgq %rcx, (%rdi) # encoding: [0xf0,0x48,0x0f,0xb1,0x0f]
+; CHECK-NEXT:    jne .LBB9_1 # encoding: [0x75,A]
+; CHECK-NEXT:    # fixup A - offset: 1, value: .LBB9_1-1, kind: FK_PCRel_1
 ; CHECK-NEXT:  # %bb.2: # %atomicrmw.end
-; CHECK-NEXT:    xorl %ecx, %ecx
-; CHECK-NEXT:    testq %rax, %rdx
-; CHECK-NEXT:    sete %cl
-; CHECK-NEXT:    movq %rcx, %rax
-; CHECK-NEXT:    retq
+; CHECK-NEXT:    xorl %ecx, %ecx # encoding: [0x31,0xc9]
+; CHECK-NEXT:    testq %rax, %rdx # encoding: [0x48,0x85,0xc2]
+; CHECK-NEXT:    sete %cl # encoding: [0x0f,0x94,0xc1]
+; CHECK-NEXT:    movq %rcx, %rax # encoding: [0x48,0x89,0xc8]
+; CHECK-NEXT:    retq # encoding: [0xc3]
 entry:
   %shl = shl i64 2, %c
   %0 = atomicrmw xor ptr %v, i64 %shl monotonic, align 8
@@ -239,25 +244,26 @@
 define i64 @atomic_shl1_neq_xor_64_gpr_valz(ptr %v, i64 %c) nounwind {
 ; CHECK-LABEL: atomic_shl1_neq_xor_64_gpr_valz:
 ; CHECK:       # %bb.0: # %entry
-; CHECK-NEXT:    movq %rsi, %rcx
-; CHECK-NEXT:    movl $1, %edx
-; CHECK-NEXT:    shlq %cl, %rdx
-; CHECK-NEXT:    movq (%rdi), %rax
+; CHECK-NEXT:    movq %rsi, %rcx # encoding: [0x48,0x89,0xf1]
+; CHECK-NEXT:    movl $1, %edx # encoding: [0xba,0x01,0x00,0x00,0x00]
+; CHECK-NEXT:    shlq %cl, %rdx # encoding: [0x48,0xd3,0xe2]
+; CHECK-NEXT:    movq (%rdi), %rax # encoding: [0x48,0x8b,0x07]
 ; CHECK-NEXT:    .p2align 4, 0x90
 ; CHECK-NEXT:  .LBB10_1: # %atomicrmw.start
 ; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
-; CHECK-NEXT:    movq %rax, %rsi
-; CHECK-NEXT:    xorq %rdx, %rsi
-; CHECK-NEXT:    lock cmpxchgq %rsi, (%rdi)
-; CHECK-NEXT:    jne .LBB10_1
+; CHECK-NEXT:    movq %rax, %rsi # encoding: [0x48,0x89,0xc6]
+; CHECK-NEXT:    xorq %rdx, %rsi # encoding: [0x48,0x31,0xd6]
+; CHECK-NEXT:    lock cmpxchgq %rsi, (%rdi) # encoding: [0xf0,0x48,0x0f,0xb1,0x37]
+; CHECK-NEXT:    jne .LBB10_1 # encoding: [0x75,A]
+; CHECK-NEXT:    # fixup A - offset: 1, value: .LBB10_1-1, kind: FK_PCRel_1
 ; CHECK-NEXT:  # %bb.2: # %atomicrmw.end
-; CHECK-NEXT:    incb %cl
-; CHECK-NEXT:    movzbl %cl, %edx
-; CHECK-NEXT:    xorl %ecx, %ecx
-; CHECK-NEXT:    btq %rdx, %rax
-; CHECK-NEXT:    setae %cl
-; CHECK-NEXT:    movq %rcx, %rax
-; CHECK-NEXT:    retq
+; CHECK-NEXT:    incb %cl # encoding: [0xfe,0xc1]
+; CHECK-NEXT:    movzbl %cl, %edx # encoding: [0x0f,0xb6,0xd1]
+; CHECK-NEXT:    xorl %ecx, %ecx # encoding: [0x31,0xc9]
+; CHECK-NEXT:    btq %rdx, %rax # encoding: [0x48,0x0f,0xa3,0xd0]
+; CHECK-NEXT:    setae %cl # encoding: [0x0f,0x93,0xc1]
+; CHECK-NEXT:    movq %rcx, %rax # encoding: [0x48,0x89,0xc8]
+; CHECK-NEXT:    retq # encoding: [0xc3]
 entry:
   %shl = shl nuw i64 1, %c
   %0 = atomicrmw xor ptr %v, i64 %shl monotonic, align 8
@@ -271,24 +277,25 @@
 define i64 @atomic_shl1_small_mask_xor_64_gpr_valz(ptr %v, i64 %c) nounwind {
 ; CHECK-LABEL: atomic_shl1_small_mask_xor_64_gpr_valz:
 ; CHECK:       # %bb.0: # %entry
-; CHECK-NEXT:    movq %rsi, %rcx
-; CHECK-NEXT:    andl $31, %ecx
-; CHECK-NEXT:    movl $1, %edx
-; CHECK-NEXT:    shlq %cl, %rdx
-; CHECK-NEXT:    movq (%rdi), %rax
+; CHECK-NEXT:    movq %rsi, %rcx # encoding: [0x48,0x89,0xf1]
+; CHECK-NEXT:    andl $31, %ecx # encoding: [0x83,0xe1,0x1f]
+; CHECK-NEXT:    movl $1, %edx # encoding: [0xba,0x01,0x00,0x00,0x00]
+; CHECK-NEXT:    shlq %cl, %rdx # encoding: [0x48,0xd3,0xe2]
+; CHECK-NEXT:    movq (%rdi), %rax # encoding: [0x48,0x8b,0x07]
 ; CHECK-NEXT:    .p2align 4, 0x90
 ; CHECK-NEXT:  .LBB11_1: # %atomicrmw.start
 ; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
-; CHECK-NEXT:    movq %rax, %rsi
-; CHECK-NEXT:    xorq %rdx, %rsi
-; CHECK-NEXT:    lock cmpxchgq %rsi, (%rdi)
-; CHECK-NEXT:    jne .LBB11_1
+; CHECK-NEXT:    movq %rax, %rsi # encoding: [0x48,0x89,0xc6]
+; CHECK-NEXT:    xorq %rdx, %rsi # encoding: [0x48,0x31,0xd6]
+; CHECK-NEXT:    lock cmpxchgq %rsi, (%rdi) # encoding: [0xf0,0x48,0x0f,0xb1,0x37]
+; CHECK-NEXT:    jne .LBB11_1 # encoding: [0x75,A]
+; CHECK-NEXT:    # fixup A - offset: 1, value: .LBB11_1-1, kind: FK_PCRel_1
 ; CHECK-NEXT:  # %bb.2: # %atomicrmw.end
-; CHECK-NEXT:    xorl %edx, %edx
-; CHECK-NEXT:    btl %ecx, %eax
-; CHECK-NEXT:    setae %dl
-; CHECK-NEXT:    movq %rdx, %rax
-; CHECK-NEXT:    retq
+; CHECK-NEXT:    xorl %edx, %edx # encoding: [0x31,0xd2]
+; CHECK-NEXT:    btl %ecx, %eax # encoding: [0x0f,0xa3,0xc8]
+; CHECK-NEXT:    setae %dl # encoding: [0x0f,0x93,0xc2]
+; CHECK-NEXT:    movq %rdx, %rax # encoding: [0x48,0x89,0xd0]
+; CHECK-NEXT:    retq # encoding: [0xc3]
 entry:
   %rem = and i64 %c, 31
   %shl = shl nuw nsw i64 1, %rem
@@ -302,23 +309,24 @@
 define i64 @atomic_shl1_mask0_xor_64_gpr_valz(ptr %v, i64 %c) nounwind {
 ; CHECK-LABEL: atomic_shl1_mask0_xor_64_gpr_valz:
 ; CHECK:       # %bb.0: # %entry
-; CHECK-NEXT:    movq %rsi, %rcx
-; CHECK-NEXT:    movl $1, %edx
-; CHECK-NEXT:    shlq %cl, %rdx
-; CHECK-NEXT:    movq (%rdi), %rax
+; CHECK-NEXT:    movq %rsi, %rcx # encoding: [0x48,0x89,0xf1]
+; CHECK-NEXT:    movl $1, %edx # encoding: [0xba,0x01,0x00,0x00,0x00]
+; CHECK-NEXT:    shlq %cl, %rdx # encoding: [0x48,0xd3,0xe2]
+; CHECK-NEXT:    movq (%rdi), %rax # encoding: [0x48,0x8b,0x07]
 ; CHECK-NEXT:    .p2align 4, 0x90
 ; CHECK-NEXT:  .LBB12_1: # %atomicrmw.start
 ; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
-; CHECK-NEXT:    movq %rax, %rsi
-; CHECK-NEXT:    xorq %rdx, %rsi
-; CHECK-NEXT:    lock cmpxchgq %rsi, (%rdi)
-; CHECK-NEXT:    jne .LBB12_1
+; CHECK-NEXT:    movq %rax, %rsi # encoding: [0x48,0x89,0xc6]
+; CHECK-NEXT:    xorq %rdx, %rsi # encoding: [0x48,0x31,0xd6]
+; CHECK-NEXT:    lock cmpxchgq %rsi, (%rdi) # encoding: [0xf0,0x48,0x0f,0xb1,0x37]
+; CHECK-NEXT:    jne .LBB12_1 # encoding: [0x75,A]
+; CHECK-NEXT:    # fixup A - offset: 1, value: .LBB12_1-1, kind: FK_PCRel_1
 ; CHECK-NEXT:  # %bb.2: # %atomicrmw.end
-; CHECK-NEXT:    xorl %edx, %edx
-; CHECK-NEXT:    btq %rcx, %rax
-; CHECK-NEXT:    setae %dl
-; CHECK-NEXT:    movq %rdx, %rax
-; CHECK-NEXT:    retq
+; CHECK-NEXT:    xorl %edx, %edx # encoding: [0x31,0xd2]
+; CHECK-NEXT:    btq %rcx, %rax # encoding: [0x48,0x0f,0xa3,0xc8]
+; CHECK-NEXT:    setae %dl # encoding: [0x0f,0x93,0xc2]
+; CHECK-NEXT:    movq %rdx, %rax # encoding: [0x48,0x89,0xd0]
+; CHECK-NEXT:    retq # encoding: [0xc3]
 entry:
   %rem = and i64 %c, 63
   %shl = shl nuw i64 1, %rem
@@ -332,23 +340,24 @@
 define i64 @atomic_shl1_mask1_xor_64_gpr_valz(ptr %v, i64 %c) nounwind {
 ; CHECK-LABEL: atomic_shl1_mask1_xor_64_gpr_valz:
 ; CHECK:       # %bb.0: # %entry
-; CHECK-NEXT:    movq %rsi, %rcx
-; CHECK-NEXT:    movl $1, %edx
-; CHECK-NEXT:    shlq %cl, %rdx
-; CHECK-NEXT:    movq (%rdi), %rax
+; CHECK-NEXT:    movq %rsi, %rcx # encoding: [0x48,0x89,0xf1]
+; CHECK-NEXT:    movl $1, %edx # encoding: [0xba,0x01,0x00,0x00,0x00]
+; CHECK-NEXT:    shlq %cl, %rdx # encoding: [0x48,0xd3,0xe2]
+; CHECK-NEXT:    movq (%rdi), %rax # encoding: [0x48,0x8b,0x07]
 ; CHECK-NEXT:    .p2align 4, 0x90
 ; CHECK-NEXT:  .LBB13_1: # %atomicrmw.start
 ; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
-; CHECK-NEXT:    movq %rax, %rsi
-; CHECK-NEXT:    xorq %rdx, %rsi
-; CHECK-NEXT:    lock cmpxchgq %rsi, (%rdi)
-; CHECK-NEXT:    jne .LBB13_1
+; CHECK-NEXT:    movq %rax, %rsi # encoding: [0x48,0x89,0xc6]
+; CHECK-NEXT:    xorq %rdx, %rsi # encoding: [0x48,0x31,0xd6]
+; CHECK-NEXT:    lock cmpxchgq %rsi, (%rdi) # encoding: [0xf0,0x48,0x0f,0xb1,0x37]
+; CHECK-NEXT:    jne .LBB13_1 # encoding: [0x75,A]
+; CHECK-NEXT:    # fixup A - offset: 1, value: .LBB13_1-1, kind: FK_PCRel_1
 ; CHECK-NEXT:  # %bb.2: # %atomicrmw.end
-; CHECK-NEXT:    xorl %edx, %edx
-; CHECK-NEXT:    btq %rcx, %rax
-; CHECK-NEXT:    setae %dl
-; CHECK-NEXT:    movq %rdx, %rax
-; CHECK-NEXT:    retq
+; CHECK-NEXT:    xorl %edx, %edx # encoding: [0x31,0xd2]
+; CHECK-NEXT:    btq %rcx, %rax # encoding: [0x48,0x0f,0xa3,0xc8]
+; CHECK-NEXT:    setae %dl # encoding: [0x0f,0x93,0xc2]
+; CHECK-NEXT:    movq %rdx, %rax # encoding: [0x48,0x89,0xd0]
+; CHECK-NEXT:    retq # encoding: [0xc3]
 entry:
   %shl = shl nuw i64 1, %c
   %0 = atomicrmw xor ptr %v, i64 %shl monotonic, align 8
@@ -362,24 +371,25 @@
 define i64 @atomic_shl1_mask01_xor_64_gpr_valz(ptr %v, i64 %c) nounwind {
 ; CHECK-LABEL: atomic_shl1_mask01_xor_64_gpr_valz:
 ; CHECK:       # %bb.0: # %entry
-; CHECK-NEXT:    movq %rsi, %rcx
-; CHECK-NEXT:    andl $63, %ecx
-; CHECK-NEXT:    movl $1, %edx
-; CHECK-NEXT:    shlq %cl, %rdx
-; CHECK-NEXT:    movq (%rdi), %rax
+; CHECK-NEXT:    movq %rsi, %rcx # encoding: [0x48,0x89,0xf1]
+; CHECK-NEXT:    andl $63, %ecx # encoding: [0x83,0xe1,0x3f]
+; CHECK-NEXT:    movl $1, %edx # encoding: [0xba,0x01,0x00,0x00,0x00]
+; CHECK-NEXT:    shlq %cl, %rdx # encoding: [0x48,0xd3,0xe2]
+; CHECK-NEXT:    movq (%rdi), %rax # encoding: [0x48,0x8b,0x07]
 ; CHECK-NEXT:    .p2align 4, 0x90
 ; CHECK-NEXT:  .LBB14_1: # %atomicrmw.start
 ; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
-; CHECK-NEXT:    movq %rax, %rsi
-; CHECK-NEXT:    xorq %rdx, %rsi
-; CHECK-NEXT:    lock cmpxchgq %rsi, (%rdi)
-; CHECK-NEXT:    jne .LBB14_1
+; CHECK-NEXT:    movq %rax, %rsi # encoding: [0x48,0x89,0xc6]
+; CHECK-NEXT:    xorq %rdx, %rsi # encoding: [0x48,0x31,0xd6]
+; CHECK-NEXT:    lock cmpxchgq %rsi, (%rdi) # encoding: [0xf0,0x48,0x0f,0xb1,0x37]
+; CHECK-NEXT:    jne .LBB14_1 # encoding: [0x75,A]
+; CHECK-NEXT:    # fixup A - offset: 1, value: .LBB14_1-1, kind: FK_PCRel_1
 ; CHECK-NEXT:  # %bb.2: # %atomicrmw.end
-; CHECK-NEXT:    xorl %edx, %edx
-; CHECK-NEXT:    btq %rcx, %rax
-; CHECK-NEXT:    setae %dl
-; CHECK-NEXT:    movq %rdx, %rax
-; CHECK-NEXT:    retq
+; CHECK-NEXT:    xorl %edx, %edx # encoding: [0x31,0xd2]
+; CHECK-NEXT:    btq %rcx, %rax # encoding: [0x48,0x0f,0xa3,0xc8]
+; CHECK-NEXT:    setae %dl # encoding: [0x0f,0x93,0xc2]
+; CHECK-NEXT:    movq %rdx, %rax # encoding: [0x48,0x89,0xd0]
+; CHECK-NEXT:    retq # encoding: [0xc3]
 entry:
   %rem = and i64 %c, 63
   %shl = shl nuw i64 1, %rem
@@ -393,23 +403,24 @@
 define i64 @atomic_blsi_xor_64_gpr_valz(ptr %v, i64 %c) nounwind {
 ; CHECK-LABEL: atomic_blsi_xor_64_gpr_valz:
 ; CHECK:       # %bb.0: # %entry
-; CHECK-NEXT:    movq %rsi, %rdx
-; CHECK-NEXT:    negq %rdx
-; CHECK-NEXT:    andq %rsi, %rdx
-; CHECK-NEXT:    movq (%rdi), %rax
+; CHECK-NEXT:    movq %rsi, %rdx # encoding: [0x48,0x89,0xf2]
+; CHECK-NEXT:    negq %rdx # encoding: [0x48,0xf7,0xda]
+; CHECK-NEXT:    andq %rsi, %rdx # encoding: [0x48,0x21,0xf2]
+; CHECK-NEXT:    movq (%rdi), %rax # encoding: [0x48,0x8b,0x07]
 ; CHECK-NEXT:    .p2align 4, 0x90
 ; CHECK-NEXT:  .LBB15_1: # %atomicrmw.start
 ; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
-; CHECK-NEXT:    movq %rax, %rcx
-; CHECK-NEXT:    xorq %rdx, %rcx
-; CHECK-NEXT:    lock cmpxchgq %rcx, (%rdi)
-; CHECK-NEXT:    jne .LBB15_1
+; CHECK-NEXT:    movq %rax, %rcx # encoding: [0x48,0x89,0xc1]
+; CHECK-NEXT:    xorq %rdx, %rcx # encoding: [0x48,0x31,0xd1]
+; CHECK-NEXT:    lock cmpxchgq %rcx, (%rdi) # encoding: [0xf0,0x48,0x0f,0xb1,0x0f]
+; CHECK-NEXT:    jne .LBB15_1 # encoding: [0x75,A]
+; CHECK-NEXT:    # fixup A - offset: 1, value: .LBB15_1-1, kind: FK_PCRel_1
 ; CHECK-NEXT:  # %bb.2: # %atomicrmw.end
-; CHECK-NEXT:    xorl %ecx, %ecx
-; CHECK-NEXT:    testq %rdx, %rax
-; CHECK-NEXT:    sete %cl
-; CHECK-NEXT:    movq %rcx, %rax
-; CHECK-NEXT:    retq
+; CHECK-NEXT:    xorl %ecx, %ecx # encoding: [0x31,0xc9]
+; CHECK-NEXT:    testq %rdx, %rax # encoding: [0x48,0x85,0xd0]
+; CHECK-NEXT:    sete %cl # encoding: [0x0f,0x94,0xc1]
+; CHECK-NEXT:    movq %rcx, %rax # encoding: [0x48,0x89,0xc8]
+; CHECK-NEXT:    retq # encoding: [0xc3]
 entry:
   %sub = sub i64 0, %c
   %and = and i64 %sub, %c
@@ -423,23 +434,24 @@
 define i64 @atomic_shl1_xor_64_gpr_valnz(ptr %v, i64 %c) nounwind {
 ; CHECK-LABEL: atomic_shl1_xor_64_gpr_valnz:
 ; CHECK:       # %bb.0: # %entry
-; CHECK-NEXT:    movq %rsi, %rcx
-; CHECK-NEXT:    movl $1, %edx
-; CHECK-NEXT:    shlq %cl, %rdx
-; CHECK-NEXT:    movq (%rdi), %rax
+; CHECK-NEXT:    movq %rsi, %rcx # encoding: [0x48,0x89,0xf1]
+; CHECK-NEXT:    movl $1, %edx # encoding: [0xba,0x01,0x00,0x00,0x00]
+; CHECK-NEXT:    shlq %cl, %rdx # encoding: [0x48,0xd3,0xe2]
+; CHECK-NEXT:    movq (%rdi), %rax # encoding: [0x48,0x8b,0x07]
 ; CHECK-NEXT:    .p2align 4, 0x90
 ; CHECK-NEXT:  .LBB16_1: # %atomicrmw.start
 ; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
-; CHECK-NEXT:    movq %rax, %rsi
-; CHECK-NEXT:    xorq %rdx, %rsi
-; CHECK-NEXT:    lock cmpxchgq %rsi, (%rdi)
-; CHECK-NEXT:    jne .LBB16_1
+; CHECK-NEXT:    movq %rax, %rsi # encoding: [0x48,0x89,0xc6]
+; CHECK-NEXT:    xorq %rdx, %rsi # encoding: [0x48,0x31,0xd6]
+; CHECK-NEXT:    lock cmpxchgq %rsi, (%rdi) # encoding: [0xf0,0x48,0x0f,0xb1,0x37]
+; CHECK-NEXT:    jne .LBB16_1 # encoding: [0x75,A]
+; CHECK-NEXT:    # fixup A - offset: 1, value: .LBB16_1-1, kind: FK_PCRel_1
 ; CHECK-NEXT:  # %bb.2: # %atomicrmw.end
-; CHECK-NEXT:    xorl %edx, %edx
-; CHECK-NEXT:    btq %rcx, %rax
-; CHECK-NEXT:    setb %dl
-; CHECK-NEXT:    movq %rdx, %rax
-; CHECK-NEXT:    retq
+; CHECK-NEXT:    xorl %edx, %edx # encoding: [0x31,0xd2]
+; CHECK-NEXT:    btq %rcx, %rax # encoding: [0x48,0x0f,0xa3,0xc8]
+; CHECK-NEXT:    setb %dl # encoding: [0x0f,0x92,0xc2]
+; CHECK-NEXT:    movq %rdx, %rax # encoding: [0x48,0x89,0xd0]
+; CHECK-NEXT:    retq # encoding: [0xc3]
 entry:
   %shl = shl nuw i64 1, %c
   %0 = atomicrmw xor ptr %v, i64 %shl monotonic, align 8
@@ -451,24 +463,25 @@
 define i64 @atomic_shl2_xor_64_gpr_valnz(ptr %v, i64 %c) nounwind {
 ; CHECK-LABEL: atomic_shl2_xor_64_gpr_valnz:
 ; CHECK:       # %bb.0: # %entry
-; CHECK-NEXT:    movq %rsi, %rcx
-; CHECK-NEXT:    movl $2, %edx
+; CHECK-NEXT:    movq %rsi, %rcx # encoding: [0x48,0x89,0xf1]
+; CHECK-NEXT:    movl $2, %edx # encoding: [0xba,0x02,0x00,0x00,0x00]
 ; CHECK-NEXT:    # kill: def $cl killed $cl killed $rcx
-; CHECK-NEXT:    shlq %cl, %rdx
-; CHECK-NEXT:    movq (%rdi), %rax
+; CHECK-NEXT:    shlq %cl, %rdx # encoding: [0x48,0xd3,0xe2]
+; CHECK-NEXT:    movq (%rdi), %rax # encoding: [0x48,0x8b,0x07]
 ; CHECK-NEXT:    .p2align 4, 0x90
 ; CHECK-NEXT:  .LBB17_1: # %atomicrmw.start
 ; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
-; CHECK-NEXT:    movq %rax, %rcx
-; CHECK-NEXT:    xorq %rdx, %rcx
-; CHECK-NEXT:    lock cmpxchgq %rcx, (%rdi)
-; CHECK-NEXT:    jne .LBB17_1
+; CHECK-NEXT:    movq %rax, %rcx # encoding: [0x48,0x89,0xc1]
+; CHECK-NEXT:    xorq %rdx, %rcx # encoding: [0x48,0x31,0xd1]
+; CHECK-NEXT:    lock cmpxchgq %rcx, (%rdi) # encoding: [0xf0,0x48,0x0f,0xb1,0x0f]
+; CHECK-NEXT:    jne .LBB17_1 # encoding: [0x75,A]
+; CHECK-NEXT:    # fixup A - offset: 1, value: .LBB17_1-1, kind: FK_PCRel_1
 ; CHECK-NEXT:  # %bb.2: # %atomicrmw.end
-; CHECK-NEXT:    xorl %ecx, %ecx
-; CHECK-NEXT:    testq %rdx, %rax
-; CHECK-NEXT:    setne %cl
-; CHECK-NEXT:    movq %rcx, %rax
-; CHECK-NEXT:    retq
+; CHECK-NEXT:    xorl %ecx, %ecx # encoding: [0x31,0xc9]
+; CHECK-NEXT:    testq %rdx, %rax # encoding: [0x48,0x85,0xd0]
+; CHECK-NEXT:    setne %cl # encoding: [0x0f,0x95,0xc1]
+; CHECK-NEXT:    movq %rcx, %rax # encoding: [0x48,0x89,0xc8]
+; CHECK-NEXT:    retq # encoding: [0xc3]
 entry:
   %shl = shl i64 2, %c
   %0 = atomicrmw xor ptr %v, i64 %shl monotonic, align 8
@@ -481,25 +494,26 @@
 define i64 @atomic_shl1_neq_xor_64_gpr_valnz(ptr %v, i64 %c) nounwind {
 ; CHECK-LABEL: atomic_shl1_neq_xor_64_gpr_valnz:
 ; CHECK:       # %bb.0: # %entry
-; CHECK-NEXT:    movq %rsi, %rcx
-; CHECK-NEXT:    movl $1, %edx
-; CHECK-NEXT:    shlq %cl, %rdx
-; CHECK-NEXT:    movq (%rdi), %rax
+; CHECK-NEXT:    movq %rsi, %rcx # encoding: [0x48,0x89,0xf1]
+; CHECK-NEXT:    movl $1, %edx # encoding: [0xba,0x01,0x00,0x00,0x00]
+; CHECK-NEXT:    shlq %cl, %rdx # encoding: [0x48,0xd3,0xe2]
+; CHECK-NEXT:    movq (%rdi), %rax # encoding: [0x48,0x8b,0x07]
 ; CHECK-NEXT:    .p2align 4, 0x90
 ; CHECK-NEXT:  .LBB18_1: # %atomicrmw.start
 ; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
-; CHECK-NEXT:    movq %rax, %rsi
-; CHECK-NEXT:    xorq %rdx, %rsi
-; CHECK-NEXT:    lock cmpxchgq %rsi, (%rdi)
-; CHECK-NEXT:    jne .LBB18_1
+; CHECK-NEXT:    movq %rax, %rsi # encoding: [0x48,0x89,0xc6]
+; CHECK-NEXT:    xorq %rdx, %rsi # encoding: [0x48,0x31,0xd6]
+; CHECK-NEXT:    lock cmpxchgq %rsi, (%rdi) # encoding: [0xf0,0x48,0x0f,0xb1,0x37]
+; CHECK-NEXT:    jne .LBB18_1 # encoding: [0x75,A]
+; CHECK-NEXT:    # fixup A - offset: 1, value: .LBB18_1-1, kind: FK_PCRel_1
 ; CHECK-NEXT:  # %bb.2: # %atomicrmw.end
-; CHECK-NEXT:    incb %cl
-; CHECK-NEXT:    movzbl %cl, %edx
-; CHECK-NEXT:    xorl %ecx, %ecx
-; CHECK-NEXT:    btq %rdx, %rax
-; CHECK-NEXT:    setb %cl
-; CHECK-NEXT:    movq %rcx, %rax
-; CHECK-NEXT:    retq
+; CHECK-NEXT:    incb %cl # encoding: [0xfe,0xc1]
+; CHECK-NEXT:    movzbl %cl, %edx # encoding: [0x0f,0xb6,0xd1]
+; CHECK-NEXT:    xorl %ecx, %ecx # encoding: [0x31,0xc9]
+; CHECK-NEXT:    btq %rdx, %rax # encoding: [0x48,0x0f,0xa3,0xd0]
+; CHECK-NEXT:    setb %cl # encoding: [0x0f,0x92,0xc1]
+; CHECK-NEXT:    movq %rcx, %rax # encoding: [0x48,0x89,0xc8]
+; CHECK-NEXT:    retq # encoding: [0xc3]
 entry:
   %shl = shl nuw i64 1, %c
   %0 = atomicrmw xor ptr %v, i64 %shl monotonic, align 8
@@ -512,24 +526,25 @@
 define i64 @atomic_shl1_small_mask_xor_64_gpr_valnz(ptr %v, i64 %c) nounwind {
 ; CHECK-LABEL: atomic_shl1_small_mask_xor_64_gpr_valnz:
 ; CHECK:       # %bb.0: # %entry
-; CHECK-NEXT:    movq %rsi, %rcx
-; CHECK-NEXT:    andl $31, %ecx
-; CHECK-NEXT:    movl $1, %edx
-; CHECK-NEXT:    shlq %cl, %rdx
-; CHECK-NEXT:    movq (%rdi), %rax
+; CHECK-NEXT:    movq %rsi, %rcx # encoding: [0x48,0x89,0xf1]
+; CHECK-NEXT:    andl $31, %ecx # encoding: [0x83,0xe1,0x1f]
+; CHECK-NEXT:    movl $1, %edx # encoding: [0xba,0x01,0x00,0x00,0x00]
+; CHECK-NEXT:    shlq %cl, %rdx # encoding: [0x48,0xd3,0xe2]
+; CHECK-NEXT:    movq (%rdi), %rax # encoding: [0x48,0x8b,0x07]
 ; CHECK-NEXT:    .p2align 4, 0x90
 ; CHECK-NEXT:  .LBB19_1: # %atomicrmw.start
 ; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
-; CHECK-NEXT:    movq %rax, %rsi
-; CHECK-NEXT:    xorq %rdx, %rsi
-; CHECK-NEXT:    lock cmpxchgq %rsi, (%rdi)
-; CHECK-NEXT:    jne .LBB19_1
+; CHECK-NEXT:    movq %rax, %rsi # encoding: [0x48,0x89,0xc6]
+; CHECK-NEXT:    xorq %rdx, %rsi # encoding: [0x48,0x31,0xd6]
+; CHECK-NEXT:    lock cmpxchgq %rsi, (%rdi) # encoding: [0xf0,0x48,0x0f,0xb1,0x37]
+; CHECK-NEXT:    jne .LBB19_1 # encoding: [0x75,A]
+; CHECK-NEXT:    # fixup A - offset: 1, value: .LBB19_1-1, kind: FK_PCRel_1
 ; CHECK-NEXT:  # %bb.2: # %atomicrmw.end
-; CHECK-NEXT:    xorl %edx, %edx
-; CHECK-NEXT:    btl %ecx, %eax
-; CHECK-NEXT:    setb %dl
-; CHECK-NEXT:    movq %rdx, %rax
-; CHECK-NEXT:    retq
+; CHECK-NEXT:    xorl %edx, %edx # encoding: [0x31,0xd2]
+; CHECK-NEXT:    btl %ecx, %eax # encoding: [0x0f,0xa3,0xc8]
+; CHECK-NEXT:    setb %dl # encoding: [0x0f,0x92,0xc2]
+; CHECK-NEXT:    movq %rdx, %rax # encoding: [0x48,0x89,0xd0]
+; CHECK-NEXT:    retq # encoding: [0xc3]
 entry:
   %rem = and i64 %c, 31
   %shl = shl nuw nsw i64 1, %rem
@@ -542,23 +557,24 @@
 define i64 @atomic_shl1_mask0_xor_64_gpr_valnz(ptr %v, i64 %c) nounwind {
 ; CHECK-LABEL: atomic_shl1_mask0_xor_64_gpr_valnz:
 ; CHECK:       # %bb.0: # %entry
-; CHECK-NEXT:    movq %rsi, %rcx
-; CHECK-NEXT:    movl $1, %edx
-; CHECK-NEXT:    shlq %cl, %rdx
-; CHECK-NEXT:    movq (%rdi), %rax
+; CHECK-NEXT:    movq %rsi, %rcx # encoding: [0x48,0x89,0xf1]
+; CHECK-NEXT:    movl $1, %edx # encoding: [0xba,0x01,0x00,0x00,0x00]
+; CHECK-NEXT:    shlq %cl, %rdx # encoding: [0x48,0xd3,0xe2]
+; CHECK-NEXT:    movq (%rdi), %rax # encoding: [0x48,0x8b,0x07]
 ; CHECK-NEXT:    .p2align 4, 0x90
 ; CHECK-NEXT:  .LBB20_1: # %atomicrmw.start
 ; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
-; CHECK-NEXT:    movq %rax, %rsi
-; CHECK-NEXT:    xorq %rdx, %rsi
-; CHECK-NEXT:    lock cmpxchgq %rsi, (%rdi)
-; CHECK-NEXT:    jne .LBB20_1
+; CHECK-NEXT:    movq %rax, %rsi # encoding: [0x48,0x89,0xc6]
+; CHECK-NEXT:    xorq %rdx, %rsi # encoding: [0x48,0x31,0xd6]
+; CHECK-NEXT:    lock cmpxchgq %rsi, (%rdi) # encoding: [0xf0,0x48,0x0f,0xb1,0x37]
+; CHECK-NEXT:    jne .LBB20_1 # encoding: [0x75,A]
+; CHECK-NEXT:    # fixup A - offset: 1, value: .LBB20_1-1, kind: FK_PCRel_1
 ; CHECK-NEXT:  # %bb.2: # %atomicrmw.end
-; CHECK-NEXT:    xorl %edx, %edx
-; CHECK-NEXT:    btq %rcx, %rax
-; CHECK-NEXT:    setb %dl
-; CHECK-NEXT:    movq %rdx, %rax
-; CHECK-NEXT:    retq
+; CHECK-NEXT:    xorl %edx, %edx # encoding: [0x31,0xd2]
+; CHECK-NEXT:    btq %rcx, %rax # encoding: [0x48,0x0f,0xa3,0xc8]
+; CHECK-NEXT:    setb %dl # encoding: [0x0f,0x92,0xc2]
+; CHECK-NEXT:    movq %rdx, %rax # encoding: [0x48,0x89,0xd0]
+; CHECK-NEXT:    retq # encoding: [0xc3]
 entry:
   %rem = and i64 %c, 63
   %shl = shl nuw i64 1, %rem
@@ -571,23 +587,24 @@
 define i64 @atomic_shl1_mask1_xor_64_gpr_valnz(ptr %v, i64 %c) nounwind {
 ; CHECK-LABEL: atomic_shl1_mask1_xor_64_gpr_valnz:
 ; CHECK:       # %bb.0: # %entry
-; CHECK-NEXT:    movq %rsi, %rcx
-; CHECK-NEXT:    movl $1, %edx
-; CHECK-NEXT:    shlq %cl, %rdx
-; CHECK-NEXT:    movq (%rdi), %rax
+; CHECK-NEXT:    movq %rsi, %rcx # encoding: [0x48,0x89,0xf1]
+; CHECK-NEXT:    movl $1, %edx # encoding: [0xba,0x01,0x00,0x00,0x00]
+; CHECK-NEXT:    shlq %cl, %rdx # encoding: [0x48,0xd3,0xe2]
+; CHECK-NEXT:    movq (%rdi), %rax # encoding: [0x48,0x8b,0x07]
 ; CHECK-NEXT:    .p2align 4, 0x90
 ; CHECK-NEXT:  .LBB21_1: # %atomicrmw.start
 ; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
-; CHECK-NEXT:    movq %rax, %rsi
-; CHECK-NEXT:    xorq %rdx, %rsi
-; CHECK-NEXT:    lock cmpxchgq %rsi, (%rdi)
-; CHECK-NEXT:    jne .LBB21_1
+; CHECK-NEXT:    movq %rax, %rsi # encoding: [0x48,0x89,0xc6]
+; CHECK-NEXT:    xorq %rdx, %rsi # encoding: [0x48,0x31,0xd6]
+; CHECK-NEXT:    lock cmpxchgq %rsi, (%rdi) # encoding: [0xf0,0x48,0x0f,0xb1,0x37]
+; CHECK-NEXT:    jne .LBB21_1 # encoding: [0x75,A]
+; CHECK-NEXT:    # fixup A - offset: 1, value: .LBB21_1-1, kind: FK_PCRel_1
 ; CHECK-NEXT:  # %bb.2: # %atomicrmw.end
-; CHECK-NEXT:    xorl %edx, %edx
-; CHECK-NEXT:    btq %rcx, %rax
-; CHECK-NEXT:    setb %dl
-; CHECK-NEXT:    movq %rdx, %rax
-; CHECK-NEXT:    retq
+; CHECK-NEXT:    xorl %edx, %edx # encoding: [0x31,0xd2]
+; CHECK-NEXT:    btq %rcx, %rax # encoding: [0x48,0x0f,0xa3,0xc8]
+; CHECK-NEXT:    setb %dl # encoding: [0x0f,0x92,0xc2]
+; CHECK-NEXT:    movq %rdx, %rax # encoding: [0x48,0x89,0xd0]
+; CHECK-NEXT:    retq # encoding: [0xc3]
 entry:
   %shl = shl nuw i64 1, %c
   %0 = atomicrmw xor ptr %v, i64 %shl monotonic, align 8
@@ -600,24 +617,25 @@
 define i64 @atomic_shl1_mask01_xor_64_gpr_valnz(ptr %v, i64 %c) nounwind {
 ; CHECK-LABEL: atomic_shl1_mask01_xor_64_gpr_valnz:
 ; CHECK:       # %bb.0: # %entry
-; CHECK-NEXT:    movq %rsi, %rcx
-; CHECK-NEXT:    andl $63, %ecx
-; CHECK-NEXT:    movl $1, %edx
-; CHECK-NEXT:    shlq %cl, %rdx
-; CHECK-NEXT:    movq (%rdi), %rax
+; CHECK-NEXT:    movq %rsi, %rcx # encoding: [0x48,0x89,0xf1]
+; CHECK-NEXT:    andl $63, %ecx # encoding: [0x83,0xe1,0x3f]
+; CHECK-NEXT:    movl $1, %edx # encoding: [0xba,0x01,0x00,0x00,0x00]
+; CHECK-NEXT:    shlq %cl, %rdx # encoding: [0x48,0xd3,0xe2]
+; CHECK-NEXT:    movq (%rdi), %rax # encoding: [0x48,0x8b,0x07]
 ; CHECK-NEXT:    .p2align 4, 0x90
 ; CHECK-NEXT:  .LBB22_1: # %atomicrmw.start
 ; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
-; CHECK-NEXT:    movq %rax, %rsi
-; CHECK-NEXT:    xorq %rdx, %rsi
-; CHECK-NEXT:    lock cmpxchgq %rsi, (%rdi)
-; CHECK-NEXT:    jne .LBB22_1
+; CHECK-NEXT:    movq %rax, %rsi # encoding: [0x48,0x89,0xc6]
+; CHECK-NEXT:    xorq %rdx, %rsi # encoding: [0x48,0x31,0xd6]
+; CHECK-NEXT:    lock cmpxchgq %rsi, (%rdi) # encoding: [0xf0,0x48,0x0f,0xb1,0x37]
+; CHECK-NEXT:    jne .LBB22_1 # encoding: [0x75,A]
+; CHECK-NEXT:    # fixup A - offset: 1, value: .LBB22_1-1, kind: FK_PCRel_1
 ; CHECK-NEXT:  # %bb.2: # %atomicrmw.end
-; CHECK-NEXT:    xorl %edx, %edx
-; CHECK-NEXT:    btq %rcx, %rax
-; CHECK-NEXT:    setb %dl
-; CHECK-NEXT:    movq %rdx, %rax
-; CHECK-NEXT:    retq
+; CHECK-NEXT:    xorl %edx, %edx # encoding: [0x31,0xd2]
+; CHECK-NEXT:    btq %rcx, %rax # encoding: [0x48,0x0f,0xa3,0xc8]
+; CHECK-NEXT:    setb %dl # encoding: [0x0f,0x92,0xc2]
+; CHECK-NEXT:    movq %rdx, %rax # encoding: [0x48,0x89,0xd0]
+; CHECK-NEXT:    retq # encoding: [0xc3]
 entry:
   %rem = and i64 %c, 63
   %shl = shl nuw i64 1, %rem
@@ -630,23 +648,24 @@
 define i64 @atomic_blsi_xor_64_gpr_valnz(ptr %v, i64 %c) nounwind {
 ; CHECK-LABEL: atomic_blsi_xor_64_gpr_valnz:
 ; CHECK:       # %bb.0: # %entry
-; CHECK-NEXT:    movq %rsi, %rdx
-; CHECK-NEXT:    negq %rdx
-; CHECK-NEXT:    andq %rsi, %rdx
-; CHECK-NEXT:    movq (%rdi), %rax
+; CHECK-NEXT:    movq %rsi, %rdx # encoding: [0x48,0x89,0xf2]
+; CHECK-NEXT:    negq %rdx # encoding: [0x48,0xf7,0xda]
+; CHECK-NEXT:    andq %rsi, %rdx # encoding: [0x48,0x21,0xf2]
+; CHECK-NEXT:    movq (%rdi), %rax # encoding: [0x48,0x8b,0x07]
 ; CHECK-NEXT:    .p2align 4, 0x90
 ; CHECK-NEXT:  .LBB23_1: # %atomicrmw.start
 ; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
-; CHECK-NEXT:    movq %rax, %rcx
-; CHECK-NEXT:    xorq %rdx, %rcx
-; CHECK-NEXT:    lock cmpxchgq %rcx, (%rdi)
-; CHECK-NEXT:    jne .LBB23_1
+; CHECK-NEXT:    movq %rax, %rcx # encoding: [0x48,0x89,0xc1]
+; CHECK-NEXT:    xorq %rdx, %rcx # encoding: [0x48,0x31,0xd1]
+; CHECK-NEXT:    lock cmpxchgq %rcx, (%rdi) # encoding: [0xf0,0x48,0x0f,0xb1,0x0f]
+; CHECK-NEXT:    jne .LBB23_1 # encoding: [0x75,A]
+; CHECK-NEXT:    # fixup A - offset: 1, value: .LBB23_1-1, kind: FK_PCRel_1
 ; CHECK-NEXT:  # %bb.2: # %atomicrmw.end
-; CHECK-NEXT:    xorl %ecx, %ecx
-; CHECK-NEXT:    testq %rdx, %rax
-; CHECK-NEXT:    setne %cl
-; CHECK-NEXT:    movq %rcx, %rax
-; CHECK-NEXT:    retq
+; CHECK-NEXT:    xorl %ecx, %ecx # encoding: [0x31,0xc9]
+; CHECK-NEXT:    testq %rdx, %rax # encoding: [0x48,0x85,0xd0]
+; CHECK-NEXT:    setne %cl # encoding: [0x0f,0x95,0xc1]
+; CHECK-NEXT:    movq %rcx, %rax # encoding: [0x48,0x89,0xc8]
+; CHECK-NEXT:    retq # encoding: [0xc3]
 entry:
   %sub = sub i64 0, %c
   %and = and i64 %sub, %c
@@ -660,14 +679,14 @@
 define i64 @atomic_shl1_and_64_gpr_val(ptr %v, i64 %c) nounwind {
 ; CHECK-LABEL: atomic_shl1_and_64_gpr_val:
 ; CHECK:       # %bb.0: # %entry
-; CHECK-NEXT:    movq %rsi, %rcx
-; CHECK-NEXT:    andl $63, %ecx
-; CHECK-NEXT:    xorl %eax, %eax
-; CHECK-NEXT:    lock btrq %rcx, (%rdi)
-; CHECK-NEXT:    setb %al
+; CHECK-NEXT:    movq %rsi, %rcx # encoding: [0x48,0x89,0xf1]
+; CHECK-NEXT:    andl $63, %ecx # encoding: [0x83,0xe1,0x3f]
+; CHECK-NEXT:    xorl %eax, %eax # encoding: [0x31,0xc0]
+; CHECK-NEXT:    lock btrq %rcx, (%rdi) # encoding: [0xf0,0x48,0x0f,0xb3,0x0f]
+; CHECK-NEXT:    setb %al # encoding: [0x0f,0x92,0xc0]
 ; CHECK-NEXT:    # kill: def $cl killed $cl killed $rcx
-; CHECK-NEXT:    shlq %cl, %rax
-; CHECK-NEXT:    retq
+; CHECK-NEXT:    shlq %cl, %rax # encoding: [0x48,0xd3,0xe0]
+; CHECK-NEXT:    retq # encoding: [0xc3]
 entry:
   %shl = shl nuw i64 1, %c
   %not = sub i64 -1, %shl
@@ -679,23 +698,24 @@
 define i64 @atomic_shl2_and_64_gpr_val(ptr %v, i64 %c) nounwind {
 ; CHECK-LABEL: atomic_shl2_and_64_gpr_val:
 ; CHECK:       # %bb.0: # %entry
-; CHECK-NEXT:    movq %rsi, %rcx
-; CHECK-NEXT:    movl $2, %edx
+; CHECK-NEXT:    movq %rsi, %rcx # encoding: [0x48,0x89,0xf1]
+; CHECK-NEXT:    movl $2, %edx # encoding: [0xba,0x02,0x00,0x00,0x00]
 ; CHECK-NEXT:    # kill: def $cl killed $cl killed $rcx
-; CHECK-NEXT:    shlq %cl, %rdx
-; CHECK-NEXT:    movq %rdx, %rcx
-; CHECK-NEXT:    notq %rcx
-; CHECK-NEXT:    movq (%rdi), %rax
+; CHECK-NEXT:    shlq %cl, %rdx # encoding: [0x48,0xd3,0xe2]
+; CHECK-NEXT:    movq %rdx, %rcx # encoding: [0x48,0x89,0xd1]
+; CHECK-NEXT:    notq %rcx # encoding: [0x48,0xf7,0xd1]
+; CHECK-NEXT:    movq (%rdi), %rax # encoding: [0x48,0x8b,0x07]
 ; CHECK-NEXT:    .p2align 4, 0x90
 ; CHECK-NEXT:  .LBB25_1: # %atomicrmw.start
 ; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
-; CHECK-NEXT:    movq %rax, %rsi
-; CHECK-NEXT:    andq %rcx, %rsi
-; CHECK-NEXT:    lock cmpxchgq %rsi, (%rdi)
-; CHECK-NEXT:    jne .LBB25_1
+; CHECK-NEXT:    movq %rax, %rsi # encoding: [0x48,0x89,0xc6]
+; CHECK-NEXT:    andq %rcx, %rsi # encoding: [0x48,0x21,0xce]
+; CHECK-NEXT:    lock cmpxchgq %rsi, (%rdi) # encoding: [0xf0,0x48,0x0f,0xb1,0x37]
+; CHECK-NEXT:    jne .LBB25_1 # encoding: [0x75,A]
+; CHECK-NEXT:    # fixup A - offset: 1, value: .LBB25_1-1, kind: FK_PCRel_1
 ; CHECK-NEXT:  # %bb.2: # %atomicrmw.end
-; CHECK-NEXT:    andq %rdx, %rax
-; CHECK-NEXT:    retq
+; CHECK-NEXT:    andq %rdx, %rax # encoding: [0x48,0x21,0xd0]
+; CHECK-NEXT:    retq # encoding: [0xc3]
 entry:
   %shl = shl i64 2, %c
   %not = xor i64 %shl, -1
@@ -707,24 +727,25 @@
 define i64 @atomic_shl1_neq_and_64_gpr_val(ptr %v, i64 %c) nounwind {
 ; CHECK-LABEL: atomic_shl1_neq_and_64_gpr_val:
 ; CHECK:       # %bb.0: # %entry
-; CHECK-NEXT:    movq %rsi, %rcx
-; CHECK-NEXT:    movq $-2, %rdx
-; CHECK-NEXT:    rolq %cl, %rdx
-; CHECK-NEXT:    movq (%rdi), %rax
+; CHECK-NEXT:    movq %rsi, %rcx # encoding: [0x48,0x89,0xf1]
+; CHECK-NEXT:    movq $-2, %rdx # encoding: [0x48,0xc7,0xc2,0xfe,0xff,0xff,0xff]
+; CHECK-NEXT:    rolq %cl, %rdx # encoding: [0x48,0xd3,0xc2]
+; CHECK-NEXT:    movq (%rdi), %rax # encoding: [0x48,0x8b,0x07]
 ; CHECK-NEXT:    .p2align 4, 0x90
 ; CHECK-NEXT:  .LBB26_1: # %atomicrmw.start
 ; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
-; CHECK-NEXT:    movq %rax, %rsi
-; CHECK-NEXT:    andq %rdx, %rsi
-; CHECK-NEXT:    lock cmpxchgq %rsi, (%rdi)
-; CHECK-NEXT:    jne .LBB26_1
+; CHECK-NEXT:    movq %rax, %rsi # encoding: [0x48,0x89,0xc6]
+; CHECK-NEXT:    andq %rdx, %rsi # encoding: [0x48,0x21,0xd6]
+; CHECK-NEXT:    lock cmpxchgq %rsi, (%rdi) # encoding: [0xf0,0x48,0x0f,0xb1,0x37]
+; CHECK-NEXT:    jne .LBB26_1 # encoding: [0x75,A]
+; CHECK-NEXT:    # fixup A - offset: 1, value: .LBB26_1-1, kind: FK_PCRel_1
 ; CHECK-NEXT:  # %bb.2: # %atomicrmw.end
-; CHECK-NEXT:    incb %cl
-; CHECK-NEXT:    movl $1, %edx
+; CHECK-NEXT:    incb %cl # encoding: [0xfe,0xc1]
+; CHECK-NEXT:    movl $1, %edx # encoding: [0xba,0x01,0x00,0x00,0x00]
 ; CHECK-NEXT:    # kill: def $cl killed $cl killed $rcx
-; CHECK-NEXT:    shlq %cl, %rdx
-; CHECK-NEXT:    andq %rdx, %rax
-; CHECK-NEXT:    retq
+; CHECK-NEXT:    shlq %cl, %rdx # encoding: [0x48,0xd3,0xe2]
+; CHECK-NEXT:    andq %rdx, %rax # encoding: [0x48,0x21,0xd0]
+; CHECK-NEXT:    retq # encoding: [0xc3]
 entry:
   %shl = shl nuw i64 1, %c
   %not = xor i64 %shl, -1
@@ -738,14 +759,14 @@
 define i64 @atomic_shl1_small_mask_and_64_gpr_val(ptr %v, i64 %c) nounwind {
 ; CHECK-LABEL: atomic_shl1_small_mask_and_64_gpr_val:
 ; CHECK:       # %bb.0: # %entry
-; CHECK-NEXT:    movq %rsi, %rcx
-; CHECK-NEXT:    andl $31, %ecx
-; CHECK-NEXT:    xorl %eax, %eax
-; CHECK-NEXT:    lock btrq %rcx, (%rdi)
-; CHECK-NEXT:    setb %al
+; CHECK-NEXT:    movq %rsi, %rcx # encoding: [0x48,0x89,0xf1]
+; CHECK-NEXT:    andl $31, %ecx # encoding: [0x83,0xe1,0x1f]
+; CHECK-NEXT:    xorl %eax, %eax # encoding: [0x31,0xc0]
+; CHECK-NEXT:    lock btrq %rcx, (%rdi) # encoding: [0xf0,0x48,0x0f,0xb3,0x0f]
+; CHECK-NEXT:    setb %al # encoding: [0x0f,0x92,0xc0]
 ; CHECK-NEXT:    # kill: def $cl killed $cl killed $rcx
-; CHECK-NEXT:    shlq %cl, %rax
-; CHECK-NEXT:    retq
+; CHECK-NEXT:    shlq %cl, %rax # encoding: [0x48,0xd3,0xe0]
+; CHECK-NEXT:    retq # encoding: [0xc3]
 entry:
   %rem = and i64 %c, 31
   %shl = shl nuw nsw i64 1, %rem
@@ -758,14 +779,14 @@
 define i64 @atomic_shl1_mask0_and_64_gpr_val(ptr %v, i64 %c) nounwind {
 ; CHECK-LABEL: atomic_shl1_mask0_and_64_gpr_val:
 ; CHECK:       # %bb.0: # %entry
-; CHECK-NEXT:    movq %rsi, %rcx
-; CHECK-NEXT:    andl $63, %ecx
-; CHECK-NEXT:    xorl %eax, %eax
-; CHECK-NEXT:    lock btrq %rcx, (%rdi)
-; CHECK-NEXT:    setb %al
+; CHECK-NEXT:    movq %rsi, %rcx # encoding: [0x48,0x89,0xf1]
+; CHECK-NEXT:    andl $63, %ecx # encoding: [0x83,0xe1,0x3f]
+; CHECK-NEXT:    xorl %eax, %eax # encoding: [0x31,0xc0]
+; CHECK-NEXT:    lock btrq %rcx, (%rdi) # encoding: [0xf0,0x48,0x0f,0xb3,0x0f]
+; CHECK-NEXT:    setb %al # encoding: [0x0f,0x92,0xc0]
 ; CHECK-NEXT:    # kill: def $cl killed $cl killed $rcx
-; CHECK-NEXT:    shlq %cl, %rax
-; CHECK-NEXT:    retq
+; CHECK-NEXT:    shlq %cl, %rax # encoding: [0x48,0xd3,0xe0]
+; CHECK-NEXT:    retq # encoding: [0xc3]
 entry:
   %rem = and i64 %c, 63
   %shl = shl nuw i64 1, %rem
@@ -779,14 +800,14 @@
 define i64 @atomic_shl1_mask1_and_64_gpr_val(ptr %v, i64 %c) nounwind {
 ; CHECK-LABEL: atomic_shl1_mask1_and_64_gpr_val:
 ; CHECK:       # %bb.0: # %entry
-; CHECK-NEXT:    movq %rsi, %rcx
-; CHECK-NEXT:    andl $63, %ecx
-; CHECK-NEXT:    xorl %eax, %eax
-; CHECK-NEXT:    lock btrq %rcx, (%rdi)
-; CHECK-NEXT:    setb %al
+; CHECK-NEXT:    movq %rsi, %rcx # encoding: [0x48,0x89,0xf1]
+; CHECK-NEXT:    andl $63, %ecx # encoding: [0x83,0xe1,0x3f]
+; CHECK-NEXT:    xorl %eax, %eax # encoding: [0x31,0xc0]
+; CHECK-NEXT:    lock btrq %rcx, (%rdi) # encoding: [0xf0,0x48,0x0f,0xb3,0x0f]
+; CHECK-NEXT:    setb %al # encoding: [0x0f,0x92,0xc0]
 ; CHECK-NEXT:    # kill: def $cl killed $cl killed $rcx
-; CHECK-NEXT:    shlq %cl, %rax
-; CHECK-NEXT:    retq
+; CHECK-NEXT:    shlq %cl, %rax # encoding: [0x48,0xd3,0xe0]
+; CHECK-NEXT:    retq # encoding: [0xc3]
 entry:
   %shl = shl nuw i64 1, %c
   %not = xor i64 %shl, -1
@@ -800,14 +821,14 @@
 define i64 @atomic_shl1_mask01_and_64_gpr_val(ptr %v, i64 %c) nounwind {
 ; CHECK-LABEL: atomic_shl1_mask01_and_64_gpr_val:
 ; CHECK:       # %bb.0: # %entry
-; CHECK-NEXT:    movq %rsi, %rcx
-; CHECK-NEXT:    andl $63, %ecx
-; CHECK-NEXT:    xorl %eax, %eax
-; CHECK-NEXT:    lock btrq %rcx, (%rdi)
-; CHECK-NEXT:    setb %al
+; CHECK-NEXT:    movq %rsi, %rcx # encoding: [0x48,0x89,0xf1]
+; CHECK-NEXT:    andl $63, %ecx # encoding: [0x83,0xe1,0x3f]
+; CHECK-NEXT:    xorl %eax, %eax # encoding: [0x31,0xc0]
+; CHECK-NEXT:    lock btrq %rcx, (%rdi) # encoding: [0xf0,0x48,0x0f,0xb3,0x0f]
+; CHECK-NEXT:    setb %al # encoding: [0x0f,0x92,0xc0]
 ; CHECK-NEXT:    # kill: def $cl killed $cl killed $rcx
-; CHECK-NEXT:    shlq %cl, %rax
-; CHECK-NEXT:    retq
+; CHECK-NEXT:    shlq %cl, %rax # encoding: [0x48,0xd3,0xe0]
+; CHECK-NEXT:    retq # encoding: [0xc3]
 entry:
   %rem = and i64 %c, 63
   %shl = shl nuw i64 1, %rem
@@ -820,23 +841,24 @@
 define i64 @atomic_blsi_and_64_gpr_val(ptr %v, i64 %c) nounwind {
 ; CHECK-LABEL: atomic_blsi_and_64_gpr_val:
 ; CHECK:       # %bb.0: # %entry
-; CHECK-NEXT:    movq %rsi, %rcx
-; CHECK-NEXT:    negq %rcx
-; CHECK-NEXT:    andq %rsi, %rcx
-; CHECK-NEXT:    movq %rcx, %rdx
-; CHECK-NEXT:    notq %rdx
-; CHECK-NEXT:    movq (%rdi), %rax
+; CHECK-NEXT:    movq %rsi, %rcx # encoding: [0x48,0x89,0xf1]
+; CHECK-NEXT:    negq %rcx # encoding: [0x48,0xf7,0xd9]
+; CHECK-NEXT:    andq %rsi, %rcx # encoding: [0x48,0x21,0xf1]
+; CHECK-NEXT:    movq %rcx, %rdx # encoding: [0x48,0x89,0xca]
+; CHECK-NEXT:    notq %rdx # encoding: [0x48,0xf7,0xd2]
+; CHECK-NEXT:    movq (%rdi), %rax # encoding: [0x48,0x8b,0x07]
 ; CHECK-NEXT:    .p2align 4, 0x90
 ; CHECK-NEXT:  .LBB31_1: # %atomicrmw.start
 ; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
-; CHECK-NEXT:    movq %rax, %rsi
-; CHECK-NEXT:    andq %rdx, %rsi
-; CHECK-NEXT:    lock cmpxchgq %rsi, (%rdi)
-; CHECK-NEXT:    jne .LBB31_1
+; CHECK-NEXT:    movq %rax, %rsi # encoding: [0x48,0x89,0xc6]
+; CHECK-NEXT:    andq %rdx, %rsi # encoding: [0x48,0x21,0xd6]
+; CHECK-NEXT:    lock cmpxchgq %rsi, (%rdi) # encoding: [0xf0,0x48,0x0f,0xb1,0x37]
+; CHECK-NEXT:    jne .LBB31_1 # encoding: [0x75,A]
+; CHECK-NEXT:    # fixup A - offset: 1, value: .LBB31_1-1, kind: FK_PCRel_1
 ; CHECK-NEXT:  # %bb.2: # %atomicrmw.end
-; CHECK-NEXT:    andq %rax, %rcx
-; CHECK-NEXT:    movq %rcx, %rax
-; CHECK-NEXT:    retq
+; CHECK-NEXT:    andq %rax, %rcx # encoding: [0x48,0x21,0xc1]
+; CHECK-NEXT:    movq %rcx, %rax # encoding: [0x48,0x89,0xc8]
+; CHECK-NEXT:    retq # encoding: [0xc3]
 entry:
   %sub = sub i64 0, %c
   %and = and i64 %sub, %c
@@ -849,23 +871,24 @@
 define i64 @atomic_shl1_and_64_gpr_valnz(ptr %v, i64 %c) nounwind {
 ; CHECK-LABEL: atomic_shl1_and_64_gpr_valnz:
 ; CHECK:       # %bb.0: # %entry
-; CHECK-NEXT:    movq %rsi, %rcx
-; CHECK-NEXT:    movq $-2, %rdx
-; CHECK-NEXT:    rolq %cl, %rdx
-; CHECK-NEXT:    movq (%rdi), %rax
+; CHECK-NEXT:    movq %rsi, %rcx # encoding: [0x48,0x89,0xf1]
+; CHECK-NEXT:    movq $-2, %rdx # encoding: [0x48,0xc7,0xc2,0xfe,0xff,0xff,0xff]
+; CHECK-NEXT:    rolq %cl, %rdx # encoding: [0x48,0xd3,0xc2]
+; CHECK-NEXT:    movq (%rdi), %rax # encoding: [0x48,0x8b,0x07]
 ; CHECK-NEXT:    .p2align 4, 0x90
 ; CHECK-NEXT:  .LBB32_1: # %atomicrmw.start
 ; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
-; CHECK-NEXT:    movq %rax, %rsi
-; CHECK-NEXT:    andq %rdx, %rsi
-; CHECK-NEXT:    lock cmpxchgq %rsi, (%rdi)
-; CHECK-NEXT:    jne .LBB32_1
+; CHECK-NEXT:    movq %rax, %rsi # encoding: [0x48,0x89,0xc6]
+; CHECK-NEXT:    andq %rdx, %rsi # encoding: [0x48,0x21,0xd6]
+; CHECK-NEXT:    lock cmpxchgq %rsi, (%rdi) # encoding: [0xf0,0x48,0x0f,0xb1,0x37]
+; CHECK-NEXT:    jne .LBB32_1 # encoding: [0x75,A]
+; CHECK-NEXT:    # fixup A - offset: 1, value: .LBB32_1-1, kind: FK_PCRel_1
 ; CHECK-NEXT:  # %bb.2: # %atomicrmw.end
-; CHECK-NEXT:    xorl %edx, %edx
-; CHECK-NEXT:    btq %rcx, %rax
-; CHECK-NEXT:    setb %dl
-; CHECK-NEXT:    movq %rdx, %rax
-; CHECK-NEXT:    retq
+; CHECK-NEXT:    xorl %edx, %edx # encoding: [0x31,0xd2]
+; CHECK-NEXT:    btq %rcx, %rax # encoding: [0x48,0x0f,0xa3,0xc8]
+; CHECK-NEXT:    setb %dl # encoding: [0x0f,0x92,0xc2]
+; CHECK-NEXT:    movq %rdx, %rax # encoding: [0x48,0x89,0xd0]
+; CHECK-NEXT:    retq # encoding: [0xc3]
 entry:
   %shl = shl nuw i64 1, %c
   %not = xor i64 %shl, -1
@@ -878,26 +901,27 @@
 define i64 @atomic_shl2_and_64_gpr_valnz(ptr %v, i64 %c) nounwind {
 ; CHECK-LABEL: atomic_shl2_and_64_gpr_valnz:
 ; CHECK:       # %bb.0: # %entry
-; CHECK-NEXT:    movq %rsi, %rcx
-; CHECK-NEXT:    movl $2, %edx
+; CHECK-NEXT:    movq %rsi, %rcx # encoding: [0x48,0x89,0xf1]
+; CHECK-NEXT:    movl $2, %edx # encoding: [0xba,0x02,0x00,0x00,0x00]
 ; CHECK-NEXT:    # kill: def $cl killed $cl killed $rcx
-; CHECK-NEXT:    shlq %cl, %rdx
-; CHECK-NEXT:    movq %rdx, %rcx
-; CHECK-NEXT:    notq %rcx
-; CHECK-NEXT:    movq (%rdi), %rax
+; CHECK-NEXT:    shlq %cl, %rdx # encoding: [0x48,0xd3,0xe2]
+; CHECK-NEXT:    movq %rdx, %rcx # encoding: [0x48,0x89,0xd1]
+; CHECK-NEXT:    notq %rcx # encoding: [0x48,0xf7,0xd1]
+; CHECK-NEXT:    movq (%rdi), %rax # encoding: [0x48,0x8b,0x07]
 ; CHECK-NEXT:    .p2align 4, 0x90
 ; CHECK-NEXT:  .LBB33_1: # %atomicrmw.start
 ; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
-; CHECK-NEXT:    movq %rax, %rsi
-; CHECK-NEXT:    andq %rcx, %rsi
-; CHECK-NEXT:    lock cmpxchgq %rsi, (%rdi)
-; CHECK-NEXT:    jne .LBB33_1
+; CHECK-NEXT:    movq %rax, %rsi # encoding: [0x48,0x89,0xc6]
+; CHECK-NEXT:    andq %rcx, %rsi # encoding: [0x48,0x21,0xce]
+; CHECK-NEXT:    lock cmpxchgq %rsi, (%rdi) # encoding: [0xf0,0x48,0x0f,0xb1,0x37]
+; CHECK-NEXT:    jne .LBB33_1 # encoding: [0x75,A]
+; CHECK-NEXT:    # fixup A - offset: 1, value: .LBB33_1-1, kind: FK_PCRel_1
 ; CHECK-NEXT:  # %bb.2: # %atomicrmw.end
-; CHECK-NEXT:    xorl %ecx, %ecx
-; CHECK-NEXT:    testq %rax, %rdx
-; CHECK-NEXT:    setne %cl
-; CHECK-NEXT:    movq %rcx, %rax
-; CHECK-NEXT:    retq
+; CHECK-NEXT:    xorl %ecx, %ecx # encoding: [0x31,0xc9]
+; CHECK-NEXT:    testq %rax, %rdx # encoding: [0x48,0x85,0xc2]
+; CHECK-NEXT:    setne %cl # encoding: [0x0f,0x95,0xc1]
+; CHECK-NEXT:    movq %rcx, %rax # encoding: [0x48,0x89,0xc8]
+; CHECK-NEXT:    retq # encoding: [0xc3]
 entry:
   %shl = shl i64 2, %c
   %not = xor i64 %shl, -1
@@ -911,25 +935,26 @@
 define i64 @atomic_shl1_neq_and_64_gpr_valnz(ptr %v, i64 %c) nounwind {
 ; CHECK-LABEL: atomic_shl1_neq_and_64_gpr_valnz:
 ; CHECK:       # %bb.0: # %entry
-; CHECK-NEXT:    movq %rsi, %rcx
-; CHECK-NEXT:    movq $-2, %rdx
-; CHECK-NEXT:    rolq %cl, %rdx
-; CHECK-NEXT:    movq (%rdi), %rax
+; CHECK-NEXT:    movq %rsi, %rcx # encoding: [0x48,0x89,0xf1]
+; CHECK-NEXT:    movq $-2, %rdx # encoding: [0x48,0xc7,0xc2,0xfe,0xff,0xff,0xff]
+; CHECK-NEXT:    rolq %cl, %rdx # encoding: [0x48,0xd3,0xc2]
+; CHECK-NEXT:    movq (%rdi), %rax # encoding: [0x48,0x8b,0x07]
 ; CHECK-NEXT:    .p2align 4, 0x90
 ; CHECK-NEXT:  .LBB34_1: # %atomicrmw.start
 ; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
-; CHECK-NEXT:    movq %rax, %rsi
-; CHECK-NEXT:    andq %rdx, %rsi
-; CHECK-NEXT:    lock cmpxchgq %rsi, (%rdi)
-; CHECK-NEXT:    jne .LBB34_1
+; CHECK-NEXT:    movq %rax, %rsi # encoding: [0x48,0x89,0xc6]
+; CHECK-NEXT:    andq %rdx, %rsi # encoding: [0x48,0x21,0xd6]
+; CHECK-NEXT:    lock cmpxchgq %rsi, (%rdi) # encoding: [0xf0,0x48,0x0f,0xb1,0x37]
+; CHECK-NEXT:    jne .LBB34_1 # encoding: [0x75,A]
+; CHECK-NEXT:    # fixup A - offset: 1, value: .LBB34_1-1, kind: FK_PCRel_1
 ; CHECK-NEXT:  # %bb.2: # %atomicrmw.end
-; CHECK-NEXT:    incb %cl
-; CHECK-NEXT:    movzbl %cl, %edx
-; CHECK-NEXT:    xorl %ecx, %ecx
-; CHECK-NEXT:    btq %rdx, %rax
-; CHECK-NEXT:    setb %cl
-; CHECK-NEXT:    movq %rcx, %rax
-; CHECK-NEXT:    retq
+; CHECK-NEXT:    incb %cl # encoding: [0xfe,0xc1]
+; CHECK-NEXT:    movzbl %cl, %edx # encoding: [0x0f,0xb6,0xd1]
+; CHECK-NEXT:    xorl %ecx, %ecx # encoding: [0x31,0xc9]
+; CHECK-NEXT:    btq %rdx, %rax # encoding: [0x48,0x0f,0xa3,0xd0]
+; CHECK-NEXT:    setb %cl # encoding: [0x0f,0x92,0xc1]
+; CHECK-NEXT:    movq %rcx, %rax # encoding: [0x48,0x89,0xc8]
+; CHECK-NEXT:    retq # encoding: [0xc3]
 entry:
   %shl = shl nuw i64 1, %c
   %not = xor i64 %shl, -1
@@ -943,24 +968,25 @@
 define i64 @atomic_shl1_small_mask_and_64_gpr_valnz(ptr %v, i64 %c) nounwind {
 ; CHECK-LABEL: atomic_shl1_small_mask_and_64_gpr_valnz:
 ; CHECK:       # %bb.0: # %entry
-; CHECK-NEXT:    movq %rsi, %rcx
-; CHECK-NEXT:    andl $31, %ecx
-; CHECK-NEXT:    movq $-2, %rdx
-; CHECK-NEXT:    rolq %cl, %rdx
-; CHECK-NEXT:    movq (%rdi), %rax
+; CHECK-NEXT:    movq %rsi, %rcx # encoding: [0x48,0x89,0xf1]
+; CHECK-NEXT:    andl $31, %ecx # encoding: [0x83,0xe1,0x1f]
+; CHECK-NEXT:    movq $-2, %rdx # encoding: [0x48,0xc7,0xc2,0xfe,0xff,0xff,0xff]
+; CHECK-NEXT:    rolq %cl, %rdx # encoding: [0x48,0xd3,0xc2]
+; CHECK-NEXT:    movq (%rdi), %rax # encoding: [0x48,0x8b,0x07]
 ; CHECK-NEXT:    .p2align 4, 0x90
 ; CHECK-NEXT:  .LBB35_1: # %atomicrmw.start
 ; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
-; CHECK-NEXT:    movq %rax, %rsi
-; CHECK-NEXT:    andq %rdx, %rsi
-; CHECK-NEXT:    lock cmpxchgq %rsi, (%rdi)
-; CHECK-NEXT:    jne .LBB35_1
+; CHECK-NEXT:    movq %rax, %rsi # encoding: [0x48,0x89,0xc6]
+; CHECK-NEXT:    andq %rdx, %rsi # encoding: [0x48,0x21,0xd6]
+; CHECK-NEXT:    lock cmpxchgq %rsi, (%rdi) # encoding: [0xf0,0x48,0x0f,0xb1,0x37]
+; CHECK-NEXT:    jne .LBB35_1 # encoding: [0x75,A]
+; CHECK-NEXT:    # fixup A - offset: 1, value: .LBB35_1-1, kind: FK_PCRel_1
 ; CHECK-NEXT:  # %bb.2: # %atomicrmw.end
-; CHECK-NEXT:    xorl %edx, %edx
-; CHECK-NEXT:    btl %ecx, %eax
-; CHECK-NEXT:    setb %dl
-; CHECK-NEXT:    movq %rdx, %rax
-; CHECK-NEXT:    retq
+; CHECK-NEXT:    xorl %edx, %edx # encoding: [0x31,0xd2]
+; CHECK-NEXT:    btl %ecx, %eax # encoding: [0x0f,0xa3,0xc8]
+; CHECK-NEXT:    setb %dl # encoding: [0x0f,0x92,0xc2]
+; CHECK-NEXT:    movq %rdx, %rax # encoding: [0x48,0x89,0xd0]
+; CHECK-NEXT:    retq # encoding: [0xc3]
 entry:
   %rem = and i64 %c, 31
   %shl = shl nuw nsw i64 1, %rem
@@ -974,23 +1000,24 @@
 define i64 @atomic_shl1_mask0_and_64_gpr_valnz(ptr %v, i64 %c) nounwind {
 ; CHECK-LABEL: atomic_shl1_mask0_and_64_gpr_valnz:
 ; CHECK:       # %bb.0: # %entry
-; CHECK-NEXT:    movq %rsi, %rcx
-; CHECK-NEXT:    movq $-2, %rdx
-; CHECK-NEXT:    rolq %cl, %rdx
-; CHECK-NEXT:    movq (%rdi), %rax
+; CHECK-NEXT:    movq %rsi, %rcx # encoding: [0x48,0x89,0xf1]
+; CHECK-NEXT:    movq $-2, %rdx # encoding: [0x48,0xc7,0xc2,0xfe,0xff,0xff,0xff]
+; CHECK-NEXT:    rolq %cl, %rdx # encoding: [0x48,0xd3,0xc2]
+; CHECK-NEXT:    movq (%rdi), %rax # encoding: [0x48,0x8b,0x07]
 ; CHECK-NEXT:    .p2align 4, 0x90
 ; CHECK-NEXT:  .LBB36_1: # %atomicrmw.start
 ; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
-; CHECK-NEXT:    movq %rax, %rsi
-; CHECK-NEXT:    andq %rdx, %rsi
-; CHECK-NEXT:    lock cmpxchgq %rsi, (%rdi)
-; CHECK-NEXT:    jne .LBB36_1
+; CHECK-NEXT:    movq %rax, %rsi # encoding: [0x48,0x89,0xc6]
+; CHECK-NEXT:    andq %rdx, %rsi # encoding: [0x48,0x21,0xd6]
+; CHECK-NEXT:    lock cmpxchgq %rsi, (%rdi) # encoding: [0xf0,0x48,0x0f,0xb1,0x37]
+; CHECK-NEXT:    jne .LBB36_1 # encoding: [0x75,A]
+; CHECK-NEXT:    # fixup A - offset: 1, value: .LBB36_1-1, kind: FK_PCRel_1
 ; CHECK-NEXT:  # %bb.2: # %atomicrmw.end
-; CHECK-NEXT:    xorl %edx, %edx
-; CHECK-NEXT:    btq %rcx, %rax
-; CHECK-NEXT:    setb %dl
-; CHECK-NEXT:    movq %rdx, %rax
-; CHECK-NEXT:    retq
+; CHECK-NEXT:    xorl %edx, %edx # encoding: [0x31,0xd2]
+; CHECK-NEXT:    btq %rcx, %rax # encoding: [0x48,0x0f,0xa3,0xc8]
+; CHECK-NEXT:    setb %dl # encoding: [0x0f,0x92,0xc2]
+; CHECK-NEXT:    movq %rdx, %rax # encoding: [0x48,0x89,0xd0]
+; CHECK-NEXT:    retq # encoding: [0xc3]
 entry:
   %rem = and i64 %c, 63
   %shl = shl nuw i64 1, %rem
@@ -1004,23 +1031,24 @@
 define i64 @atomic_shl1_mask1_and_64_gpr_valnz(ptr %v, i64 %c) nounwind {
 ; CHECK-LABEL: atomic_shl1_mask1_and_64_gpr_valnz:
 ; CHECK:       # %bb.0: # %entry
-; CHECK-NEXT:    movq %rsi, %rcx
-; CHECK-NEXT:    movq $-2, %rdx
-; CHECK-NEXT:    rolq %cl, %rdx
-; CHECK-NEXT:    movq (%rdi), %rax
+; CHECK-NEXT:    movq %rsi, %rcx # encoding: [0x48,0x89,0xf1]
+; CHECK-NEXT:    movq $-2, %rdx # encoding: [0x48,0xc7,0xc2,0xfe,0xff,0xff,0xff]
+; CHECK-NEXT:    rolq %cl, %rdx # encoding: [0x48,0xd3,0xc2]
+; CHECK-NEXT:    movq (%rdi), %rax # encoding: [0x48,0x8b,0x07]
 ; CHECK-NEXT:    .p2align 4, 0x90
 ; CHECK-NEXT:  .LBB37_1: # %atomicrmw.start
 ; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
-; CHECK-NEXT:    movq %rax, %rsi
-; CHECK-NEXT:    andq %rdx, %rsi
-; CHECK-NEXT:    lock cmpxchgq %rsi, (%rdi)
-; CHECK-NEXT:    jne .LBB37_1
+; CHECK-NEXT:    movq %rax, %rsi # encoding: [0x48,0x89,0xc6]
+; CHECK-NEXT:    andq %rdx, %rsi # encoding: [0x48,0x21,0xd6]
+; CHECK-NEXT:    lock cmpxchgq %rsi, (%rdi) # encoding: [0xf0,0x48,0x0f,0xb1,0x37]
+; CHECK-NEXT:    jne .LBB37_1 # encoding: [0x75,A]
+; CHECK-NEXT:    # fixup A - offset: 1, value: .LBB37_1-1, kind: FK_PCRel_1
 ; CHECK-NEXT:  # %bb.2: # %atomicrmw.end
-; CHECK-NEXT:    xorl %edx, %edx
-; CHECK-NEXT:    btq %rcx, %rax
-; CHECK-NEXT:    setb %dl
-; CHECK-NEXT:    movq %rdx, %rax
-; CHECK-NEXT:    retq
+; CHECK-NEXT:    xorl %edx, %edx # encoding: [0x31,0xd2]
+; CHECK-NEXT:    btq %rcx, %rax # encoding: [0x48,0x0f,0xa3,0xc8]
+; CHECK-NEXT:    setb %dl # encoding: [0x0f,0x92,0xc2]
+; CHECK-NEXT:    movq %rdx, %rax # encoding: [0x48,0x89,0xd0]
+; CHECK-NEXT:    retq # encoding: [0xc3]
 entry:
   %shl = shl nuw i64 1, %c
   %not = xor i64 %shl, -1
@@ -1034,26 +1062,27 @@
 define i64 @atomic_shl1_mask01_and_64_gpr_valnz(ptr %v, i64 %c) nounwind {
 ; CHECK-LABEL: atomic_shl1_mask01_and_64_gpr_valnz:
 ; CHECK:       # %bb.0: # %entry
-; CHECK-NEXT:    movq %rsi, %rcx
-; CHECK-NEXT:    movl %ecx, %edx
-; CHECK-NEXT:    andl $63, %edx
-; CHECK-NEXT:    movq $-2, %rsi
+; CHECK-NEXT:    movq %rsi, %rcx # encoding: [0x48,0x89,0xf1]
+; CHECK-NEXT:    movl %ecx, %edx # encoding: [0x89,0xca]
+; CHECK-NEXT:    andl $63, %edx # encoding: [0x83,0xe2,0x3f]
+; CHECK-NEXT:    movq $-2, %rsi # encoding: [0x48,0xc7,0xc6,0xfe,0xff,0xff,0xff]
 ; CHECK-NEXT:    # kill: def $cl killed $cl killed $rcx
-; CHECK-NEXT:    rolq %cl, %rsi
-; CHECK-NEXT:    movq (%rdi), %rax
+; CHECK-NEXT:    rolq %cl, %rsi # encoding: [0x48,0xd3,0xc6]
+; CHECK-NEXT:    movq (%rdi), %rax # encoding: [0x48,0x8b,0x07]
 ; CHECK-NEXT:    .p2align 4, 0x90
 ; CHECK-NEXT:  .LBB38_1: # %atomicrmw.start
 ; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
-; CHECK-NEXT:    movq %rax, %rcx
-; CHECK-NEXT:    andq %rsi, %rcx
-; CHECK-NEXT:    lock cmpxchgq %rcx, (%rdi)
-; CHECK-NEXT:    jne .LBB38_1
+; CHECK-NEXT:    movq %rax, %rcx # encoding: [0x48,0x89,0xc1]
+; CHECK-NEXT:    andq %rsi, %rcx # encoding: [0x48,0x21,0xf1]
+; CHECK-NEXT:    lock cmpxchgq %rcx, (%rdi) # encoding: [0xf0,0x48,0x0f,0xb1,0x0f]
+; CHECK-NEXT:    jne .LBB38_1 # encoding: [0x75,A]
+; CHECK-NEXT:    # fixup A - offset: 1, value: .LBB38_1-1, kind: FK_PCRel_1
 ; CHECK-NEXT:  # %bb.2: # %atomicrmw.end
-; CHECK-NEXT:    xorl %ecx, %ecx
-; CHECK-NEXT:    btq %rdx, %rax
-; CHECK-NEXT:    setb %cl
-; CHECK-NEXT:    movq %rcx, %rax
-; CHECK-NEXT:    retq
+; CHECK-NEXT:    xorl %ecx, %ecx # encoding: [0x31,0xc9]
+; CHECK-NEXT:    btq %rdx, %rax # encoding: [0x48,0x0f,0xa3,0xd0]
+; CHECK-NEXT:    setb %cl # encoding: [0x0f,0x92,0xc1]
+; CHECK-NEXT:    movq %rcx, %rax # encoding: [0x48,0x89,0xc8]
+; CHECK-NEXT:    retq # encoding: [0xc3]
 entry:
   %rem = and i64 %c, 63
   %shl = shl nuw i64 1, %rem
@@ -1067,25 +1096,26 @@
 define i64 @atomic_blsi_and_64_gpr_valnz(ptr %v, i64 %c) nounwind {
 ; CHECK-LABEL: atomic_blsi_and_64_gpr_valnz:
 ; CHECK:       # %bb.0: # %entry
-; CHECK-NEXT:    movq %rsi, %rdx
-; CHECK-NEXT:    negq %rdx
-; CHECK-NEXT:    andq %rsi, %rdx
-; CHECK-NEXT:    movq %rdx, %rcx
-; CHECK-NEXT:    notq %rcx
-; CHECK-NEXT:    movq (%rdi), %rax
+; CHECK-NEXT:    movq %rsi, %rdx # encoding: [0x48,0x89,0xf2]
+; CHECK-NEXT:    negq %rdx # encoding: [0x48,0xf7,0xda]
+; CHECK-NEXT:    andq %rsi, %rdx # encoding: [0x48,0x21,0xf2]
+; CHECK-NEXT:    movq %rdx, %rcx # encoding: [0x48,0x89,0xd1]
+; CHECK-NEXT:    notq %rcx # encoding: [0x48,0xf7,0xd1]
+; CHECK-NEXT:    movq (%rdi), %rax # encoding: [0x48,0x8b,0x07]
 ; CHECK-NEXT:    .p2align 4, 0x90
 ; CHECK-NEXT:  .LBB39_1: # %atomicrmw.start
 ; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
-; CHECK-NEXT:    movq %rax, %rsi
-; CHECK-NEXT:    andq %rcx, %rsi
-; CHECK-NEXT:    lock cmpxchgq %rsi, (%rdi)
-; CHECK-NEXT:    jne .LBB39_1
+; CHECK-NEXT:    movq %rax, %rsi # encoding: [0x48,0x89,0xc6]
+; CHECK-NEXT:    andq %rcx, %rsi # encoding: [0x48,0x21,0xce]
+; CHECK-NEXT:    lock cmpxchgq %rsi, (%rdi) # encoding: [0xf0,0x48,0x0f,0xb1,0x37]
+; CHECK-NEXT:    jne .LBB39_1 # encoding: [0x75,A]
+; CHECK-NEXT:    # fixup A - offset: 1, value: .LBB39_1-1, kind: FK_PCRel_1
 ; CHECK-NEXT:  # %bb.2: # %atomicrmw.end
-; CHECK-NEXT:    xorl %ecx, %ecx
-; CHECK-NEXT:    testq %rdx, %rax
-; CHECK-NEXT:    setne %cl
-; CHECK-NEXT:    movq %rcx, %rax
-; CHECK-NEXT:    retq
+; CHECK-NEXT:    xorl %ecx, %ecx # encoding: [0x31,0xc9]
+; CHECK-NEXT:    testq %rdx, %rax # encoding: [0x48,0x85,0xd0]
+; CHECK-NEXT:    setne %cl # encoding: [0x0f,0x95,0xc1]
+; CHECK-NEXT:    movq %rcx, %rax # encoding: [0x48,0x89,0xc8]
+; CHECK-NEXT:    retq # encoding: [0xc3]
 entry:
   %sub = sub i64 0, %c
   %and = and i64 %sub, %c
@@ -1100,16 +1130,17 @@
 define i64 @atomic_shl1_and_64_gpr_brnz(ptr %v, i64 %c) nounwind {
 ; CHECK-LABEL: atomic_shl1_and_64_gpr_brnz:
 ; CHECK:       # %bb.0: # %entry
-; CHECK-NEXT:    movl %esi, %eax
-; CHECK-NEXT:    andl $63, %eax
-; CHECK-NEXT:    lock btrq %rax, (%rdi)
-; CHECK-NEXT:    jae .LBB40_1
+; CHECK-NEXT:    movl %esi, %eax # encoding: [0x89,0xf0]
+; CHECK-NEXT:    andl $63, %eax # encoding: [0x83,0xe0,0x3f]
+; CHECK-NEXT:    lock btrq %rax, (%rdi) # encoding: [0xf0,0x48,0x0f,0xb3,0x07]
+; CHECK-NEXT:    jae .LBB40_1 # encoding: [0x73,A]
+; CHECK-NEXT:    # fixup A - offset: 1, value: .LBB40_1-1, kind: FK_PCRel_1
 ; CHECK-NEXT:  # %bb.2: # %if.then
-; CHECK-NEXT:    movq (%rdi,%rsi,8), %rax
-; CHECK-NEXT:    retq
+; CHECK-NEXT:    movq (%rdi,%rsi,8), %rax # encoding: [0x48,0x8b,0x04,0xf7]
+; CHECK-NEXT:    retq # encoding: [0xc3]
 ; CHECK-NEXT:  .LBB40_1:
-; CHECK-NEXT:    movl $123, %eax
-; CHECK-NEXT:    retq
+; CHECK-NEXT:    movl $123, %eax # encoding: [0xb8,0x7b,0x00,0x00,0x00]
+; CHECK-NEXT:    retq # encoding: [0xc3]
 entry:
   %shl = shl nuw i64 1, %c
   %not = xor i64 %shl, -1
@@ -1131,28 +1162,30 @@
 define i64 @atomic_shl2_and_64_gpr_brnz(ptr %v, i64 %c) nounwind {
 ; CHECK-LABEL: atomic_shl2_and_64_gpr_brnz:
 ; CHECK:       # %bb.0: # %entry
-; CHECK-NEXT:    movq %rsi, %rcx
-; CHECK-NEXT:    movl $2, %edx
-; CHECK-NEXT:    shlq %cl, %rdx
-; CHECK-NEXT:    movq %rdx, %rsi
-; CHECK-NEXT:    notq %rsi
-; CHECK-NEXT:    movq (%rdi), %rax
+; CHECK-NEXT:    movq %rsi, %rcx # encoding: [0x48,0x89,0xf1]
+; CHECK-NEXT:    movl $2, %edx # encoding: [0xba,0x02,0x00,0x00,0x00]
+; CHECK-NEXT:    shlq %cl, %rdx # encoding: [0x48,0xd3,0xe2]
+; CHECK-NEXT:    movq %rdx, %rsi # encoding: [0x48,0x89,0xd6]
+; CHECK-NEXT:    notq %rsi # encoding: [0x48,0xf7,0xd6]
+; CHECK-NEXT:    movq (%rdi), %rax # encoding: [0x48,0x8b,0x07]
 ; CHECK-NEXT:    .p2align 4, 0x90
 ; CHECK-NEXT:  .LBB41_1: # %atomicrmw.start
 ; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
-; CHECK-NEXT:    movq %rax, %r8
-; CHECK-NEXT:    andq %rsi, %r8
-; CHECK-NEXT:    lock cmpxchgq %r8, (%rdi)
-; CHECK-NEXT:    jne .LBB41_1
+; CHECK-NEXT:    movq %rax, %r8 # encoding: [0x49,0x89,0xc0]
+; CHECK-NEXT:    andq %rsi, %r8 # encoding: [0x49,0x21,0xf0]
+; CHECK-NEXT:    lock cmpxchgq %r8, (%rdi) # encoding: [0xf0,0x4c,0x0f,0xb1,0x07]
+; CHECK-NEXT:    jne .LBB41_1 # encoding: [0x75,A]
+; CHECK-NEXT:    # fixup A - offset: 1, value: .LBB41_1-1, kind: FK_PCRel_1
 ; CHECK-NEXT:  # %bb.2: # %atomicrmw.end
-; CHECK-NEXT:    testq %rdx, %rax
-; CHECK-NEXT:    je .LBB41_3
+; CHECK-NEXT:    testq %rdx, %rax # encoding: [0x48,0x85,0xd0]
+; CHECK-NEXT:    je .LBB41_3 # encoding: [0x74,A]
+; CHECK-NEXT:    # fixup A - offset: 1, value: .LBB41_3-1, kind: FK_PCRel_1
 ; CHECK-NEXT:  # %bb.4: # %if.then
-; CHECK-NEXT:    movq (%rdi,%rcx,8), %rax
-; CHECK-NEXT:    retq
+; CHECK-NEXT:    movq (%rdi,%rcx,8), %rax # encoding: [0x48,0x8b,0x04,0xcf]
+; CHECK-NEXT:    retq # encoding: [0xc3]
 ; CHECK-NEXT:  .LBB41_3:
-; CHECK-NEXT:    movl $123, %eax
-; CHECK-NEXT:    retq
+; CHECK-NEXT:    movl $123, %eax # encoding: [0xb8,0x7b,0x00,0x00,0x00]
+; CHECK-NEXT:    retq # encoding: [0xc3]
 entry:
   %shl = shl i64 2, %c
   %not = xor i64 %shl, -1
@@ -1174,28 +1207,30 @@
 define i64 @atomic_shl1_neq_and_64_gpr_brnz(ptr %v, i64 %c) nounwind {
 ; CHECK-LABEL: atomic_shl1_neq_and_64_gpr_brnz:
 ; CHECK:       # %bb.0: # %entry
-; CHECK-NEXT:    movq %rsi, %rcx
-; CHECK-NEXT:    movq $-2, %rdx
-; CHECK-NEXT:    rolq %cl, %rdx
-; CHECK-NEXT:    movq (%rdi), %rax
+; CHECK-NEXT:    movq %rsi, %rcx # encoding: [0x48,0x89,0xf1]
+; CHECK-NEXT:    movq $-2, %rdx # encoding: [0x48,0xc7,0xc2,0xfe,0xff,0xff,0xff]
+; CHECK-NEXT:    rolq %cl, %rdx # encoding: [0x48,0xd3,0xc2]
+; CHECK-NEXT:    movq (%rdi), %rax # encoding: [0x48,0x8b,0x07]
 ; CHECK-NEXT:    .p2align 4, 0x90
 ; CHECK-NEXT:  .LBB42_1: # %atomicrmw.start
 ; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
-; CHECK-NEXT:    movq %rax, %rsi
-; CHECK-NEXT:    andq %rdx, %rsi
-; CHECK-NEXT:    lock cmpxchgq %rsi, (%rdi)
-; CHECK-NEXT:    jne .LBB42_1
+; CHECK-NEXT:    movq %rax, %rsi # encoding: [0x48,0x89,0xc6]
+; CHECK-NEXT:    andq %rdx, %rsi # encoding: [0x48,0x21,0xd6]
+; CHECK-NEXT:    lock cmpxchgq %rsi, (%rdi) # encoding: [0xf0,0x48,0x0f,0xb1,0x37]
+; CHECK-NEXT:    jne .LBB42_1 # encoding: [0x75,A]
+; CHECK-NEXT:    # fixup A - offset: 1, value: .LBB42_1-1, kind: FK_PCRel_1
 ; CHECK-NEXT:  # %bb.2: # %atomicrmw.end
-; CHECK-NEXT:    leal 1(%rcx), %edx
-; CHECK-NEXT:    movzbl %dl, %edx
-; CHECK-NEXT:    btq %rdx, %rax
-; CHECK-NEXT:    jae .LBB42_3
+; CHECK-NEXT:    leal 1(%rcx), %edx # encoding: [0x8d,0x51,0x01]
+; CHECK-NEXT:    movzbl %dl, %edx # encoding: [0x0f,0xb6,0xd2]
+; CHECK-NEXT:    btq %rdx, %rax # encoding: [0x48,0x0f,0xa3,0xd0]
+; CHECK-NEXT:    jae .LBB42_3 # encoding: [0x73,A]
+; CHECK-NEXT:    # fixup A - offset: 1, value: .LBB42_3-1, kind: FK_PCRel_1
 ; CHECK-NEXT:  # %bb.4: # %if.then
-; CHECK-NEXT:    movq (%rdi,%rcx,8), %rax
-; CHECK-NEXT:    retq
+; CHECK-NEXT:    movq (%rdi,%rcx,8), %rax # encoding: [0x48,0x8b,0x04,0xcf]
+; CHECK-NEXT:    retq # encoding: [0xc3]
 ; CHECK-NEXT:  .LBB42_3:
-; CHECK-NEXT:    movl $123, %eax
-; CHECK-NEXT:    retq
+; CHECK-NEXT:    movl $123, %eax # encoding: [0xb8,0x7b,0x00,0x00,0x00]
+; CHECK-NEXT:    retq # encoding: [0xc3]
 entry:
   %shl = shl nuw i64 1, %c
   %not = xor i64 %shl, -1
@@ -1219,15 +1254,16 @@
 define i64 @atomic_shl1_small_mask_and_64_gpr_brnz(ptr %v, i64 %c) nounwind {
 ; CHECK-LABEL: atomic_shl1_small_mask_and_64_gpr_brnz:
 ; CHECK:       # %bb.0: # %entry
-; CHECK-NEXT:    andl $31, %esi
-; CHECK-NEXT:    lock btrq %rsi, (%rdi)
-; CHECK-NEXT:    jae .LBB43_1
+; CHECK-NEXT:    andl $31, %esi # encoding: [0x83,0xe6,0x1f]
+; CHECK-NEXT:    lock btrq %rsi, (%rdi) # encoding: [0xf0,0x48,0x0f,0xb3,0x37]
+; CHECK-NEXT:    jae .LBB43_1 # encoding: [0x73,A]
+; CHECK-NEXT:    # fixup A - offset: 1, value: .LBB43_1-1, kind: FK_PCRel_1
 ; CHECK-NEXT:  # %bb.2: # %if.then
-; CHECK-NEXT:    movq (%rdi,%rsi,8), %rax
-; CHECK-NEXT:    retq
+; CHECK-NEXT:    movq (%rdi,%rsi,8), %rax # encoding: [0x48,0x8b,0x04,0xf7]
+; CHECK-NEXT:    retq # encoding: [0xc3]
 ; CHECK-NEXT:  .LBB43_1:
-; CHECK-NEXT:    movl $123, %eax
-; CHECK-NEXT:    retq
+; CHECK-NEXT:    movl $123, %eax # encoding: [0xb8,0x7b,0x00,0x00,0x00]
+; CHECK-NEXT:    retq # encoding: [0xc3]
 entry:
   %rem = and i64 %c, 31
   %shl = shl nuw nsw i64 1, %rem
@@ -1250,16 +1286,17 @@
 define i64 @atomic_shl1_mask0_and_64_gpr_brnz(ptr %v, i64 %c) nounwind {
 ; CHECK-LABEL: atomic_shl1_mask0_and_64_gpr_brnz:
 ; CHECK:       # %bb.0: # %entry
-; CHECK-NEXT:    movl %esi, %eax
-; CHECK-NEXT:    andl $63, %eax
-; CHECK-NEXT:    lock btrq %rax, (%rdi)
-; CHECK-NEXT:    jae .LBB44_1
+; CHECK-NEXT:    movl %esi, %eax # encoding: [0x89,0xf0]
+; CHECK-NEXT:    andl $63, %eax # encoding: [0x83,0xe0,0x3f]
+; CHECK-NEXT:    lock btrq %rax, (%rdi) # encoding: [0xf0,0x48,0x0f,0xb3,0x07]
+; CHECK-NEXT:    jae .LBB44_1 # encoding: [0x73,A]
+; CHECK-NEXT:    # fixup A - offset: 1, value: .LBB44_1-1, kind: FK_PCRel_1
 ; CHECK-NEXT:  # %bb.2: # %if.then
-; CHECK-NEXT:    movq (%rdi,%rsi,8), %rax
-; CHECK-NEXT:    retq
+; CHECK-NEXT:    movq (%rdi,%rsi,8), %rax # encoding: [0x48,0x8b,0x04,0xf7]
+; CHECK-NEXT:    retq # encoding: [0xc3]
 ; CHECK-NEXT:  .LBB44_1:
-; CHECK-NEXT:    movl $123, %eax
-; CHECK-NEXT:    retq
+; CHECK-NEXT:    movl $123, %eax # encoding: [0xb8,0x7b,0x00,0x00,0x00]
+; CHECK-NEXT:    retq # encoding: [0xc3]
 entry:
   %rem = and i64 %c, 63
   %shl = shl nuw i64 1, %rem
@@ -1283,16 +1320,17 @@
 define i64 @atomic_shl1_mask1_and_64_gpr_brnz(ptr %v, i64 %c) nounwind {
 ; CHECK-LABEL: atomic_shl1_mask1_and_64_gpr_brnz:
 ; CHECK:       # %bb.0: # %entry
-; CHECK-NEXT:    movl %esi, %eax
-; CHECK-NEXT:    andl $63, %eax
-; CHECK-NEXT:    lock btrq %rax, (%rdi)
-; CHECK-NEXT:    jae .LBB45_1
+; CHECK-NEXT:    movl %esi, %eax # encoding: [0x89,0xf0]
+; CHECK-NEXT:    andl $63, %eax # encoding: [0x83,0xe0,0x3f]
+; CHECK-NEXT:    lock btrq %rax, (%rdi) # encoding: [0xf0,0x48,0x0f,0xb3,0x07]
+; CHECK-NEXT:    jae .LBB45_1 # encoding: [0x73,A]
+; CHECK-NEXT:    # fixup A - offset: 1, value: .LBB45_1-1, kind: FK_PCRel_1
 ; CHECK-NEXT:  # %bb.2: # %if.then
-; CHECK-NEXT:    movq (%rdi,%rsi,8), %rax
-; CHECK-NEXT:    retq
+; CHECK-NEXT:    movq (%rdi,%rsi,8), %rax # encoding: [0x48,0x8b,0x04,0xf7]
+; CHECK-NEXT:    retq # encoding: [0xc3]
 ; CHECK-NEXT:  .LBB45_1:
-; CHECK-NEXT:    movl $123, %eax
-; CHECK-NEXT:    retq
+; CHECK-NEXT:    movl $123, %eax # encoding: [0xb8,0x7b,0x00,0x00,0x00]
+; CHECK-NEXT:    retq # encoding: [0xc3]
 entry:
   %shl = shl nuw i64 1, %c
   %not = xor i64 %shl, -1
@@ -1316,16 +1354,17 @@
 define i64 @atomic_shl1_mask01_and_64_gpr_brnz(ptr %v, i64 %c) nounwind {
 ; CHECK-LABEL: atomic_shl1_mask01_and_64_gpr_brnz:
 ; CHECK:       # %bb.0: # %entry
-; CHECK-NEXT:    movl %esi, %eax
-; CHECK-NEXT:    andl $63, %eax
-; CHECK-NEXT:    lock btrq %rax, (%rdi)
-; CHECK-NEXT:    jae .LBB46_1
+; CHECK-NEXT:    movl %esi, %eax # encoding: [0x89,0xf0]
+; CHECK-NEXT:    andl $63, %eax # encoding: [0x83,0xe0,0x3f]
+; CHECK-NEXT:    lock btrq %rax, (%rdi) # encoding: [0xf0,0x48,0x0f,0xb3,0x07]
+; CHECK-NEXT:    jae .LBB46_1 # encoding: [0x73,A]
+; CHECK-NEXT:    # fixup A - offset: 1, value: .LBB46_1-1, kind: FK_PCRel_1
 ; CHECK-NEXT:  # %bb.2: # %if.then
-; CHECK-NEXT:    movq (%rdi,%rsi,8), %rax
-; CHECK-NEXT:    retq
+; CHECK-NEXT:    movq (%rdi,%rsi,8), %rax # encoding: [0x48,0x8b,0x04,0xf7]
+; CHECK-NEXT:    retq # encoding: [0xc3]
 ; CHECK-NEXT:  .LBB46_1:
-; CHECK-NEXT:    movl $123, %eax
-; CHECK-NEXT:    retq
+; CHECK-NEXT:    movl $123, %eax # encoding: [0xb8,0x7b,0x00,0x00,0x00]
+; CHECK-NEXT:    retq # encoding: [0xc3]
 entry:
   %rem = and i64 %c, 63
   %shl = shl nuw i64 1, %rem
@@ -1348,28 +1387,30 @@
 define i64 @atomic_blsi_and_64_gpr_brnz(ptr %v, i64 %c) nounwind {
 ; CHECK-LABEL: atomic_blsi_and_64_gpr_brnz:
 ; CHECK:       # %bb.0: # %entry
-; CHECK-NEXT:    movq %rsi, %rcx
-; CHECK-NEXT:    negq %rcx
-; CHECK-NEXT:    andq %rsi, %rcx
-; CHECK-NEXT:    movq %rcx, %rdx
-; CHECK-NEXT:    notq %rdx
-; CHECK-NEXT:    movq (%rdi), %rax
+; CHECK-NEXT:    movq %rsi, %rcx # encoding: [0x48,0x89,0xf1]
+; CHECK-NEXT:    negq %rcx # encoding: [0x48,0xf7,0xd9]
+; CHECK-NEXT:    andq %rsi, %rcx # encoding: [0x48,0x21,0xf1]
+; CHECK-NEXT:    movq %rcx, %rdx # encoding: [0x48,0x89,0xca]
+; CHECK-NEXT:    notq %rdx # encoding: [0x48,0xf7,0xd2]
+; CHECK-NEXT:    movq (%rdi), %rax # encoding: [0x48,0x8b,0x07]
 ; CHECK-NEXT:    .p2align 4, 0x90
 ; CHECK-NEXT:  .LBB47_1: # %atomicrmw.start
 ; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
-; CHECK-NEXT:    movq %rax, %r8
-; CHECK-NEXT:    andq %rdx, %r8
-; CHECK-NEXT:    lock cmpxchgq %r8, (%rdi)
-; CHECK-NEXT:    jne .LBB47_1
+; CHECK-NEXT:    movq %rax, %r8 # encoding: [0x49,0x89,0xc0]
+; CHECK-NEXT:    andq %rdx, %r8 # encoding: [0x49,0x21,0xd0]
+; CHECK-NEXT:    lock cmpxchgq %r8, (%rdi) # encoding: [0xf0,0x4c,0x0f,0xb1,0x07]
+; CHECK-NEXT:    jne .LBB47_1 # encoding: [0x75,A]
+; CHECK-NEXT:    # fixup A - offset: 1, value: .LBB47_1-1, kind: FK_PCRel_1
 ; CHECK-NEXT:  # %bb.2: # %atomicrmw.end
-; CHECK-NEXT:    testq %rcx, %rax
-; CHECK-NEXT:    je .LBB47_3
+; CHECK-NEXT:    testq %rcx, %rax # encoding: [0x48,0x85,0xc8]
+; CHECK-NEXT:    je .LBB47_3 # encoding: [0x74,A]
+; CHECK-NEXT:    # fixup A - offset: 1, value: .LBB47_3-1, kind: FK_PCRel_1
 ; CHECK-NEXT:  # %bb.4: # %if.then
-; CHECK-NEXT:    movq (%rdi,%rsi,8), %rax
-; CHECK-NEXT:    retq
+; CHECK-NEXT:    movq (%rdi,%rsi,8), %rax # encoding: [0x48,0x8b,0x04,0xf7]
+; CHECK-NEXT:    retq # encoding: [0xc3]
 ; CHECK-NEXT:  .LBB47_3:
-; CHECK-NEXT:    movl $123, %eax
-; CHECK-NEXT:    retq
+; CHECK-NEXT:    movl $123, %eax # encoding: [0xb8,0x7b,0x00,0x00,0x00]
+; CHECK-NEXT:    retq # encoding: [0xc3]
 entry:
   %sub = sub i64 0, %c
   %and = and i64 %sub, %c
@@ -1392,17 +1433,18 @@
 define i64 @atomic_shl1_xor_64_const_br(ptr %v) nounwind {
 ; CHECK-LABEL: atomic_shl1_xor_64_const_br:
 ; CHECK:       # %bb.0: # %entry
-; CHECK-NEXT:    xorl %eax, %eax
-; CHECK-NEXT:    lock btcq $4, (%rdi)
-; CHECK-NEXT:    setb %al
-; CHECK-NEXT:    shlq $4, %rax
-; CHECK-NEXT:    je .LBB48_1
+; CHECK-NEXT:    xorl %eax, %eax # encoding: [0x31,0xc0]
+; CHECK-NEXT:    lock btcq $4, (%rdi) # encoding: [0xf0,0x48,0x0f,0xba,0x3f,0x04]
+; CHECK-NEXT:    setb %al # encoding: [0x0f,0x92,0xc0]
+; CHECK-NEXT:    shlq $4, %rax # encoding: [0x48,0xc1,0xe0,0x04]
+; CHECK-NEXT:    je .LBB48_1 # encoding: [0x74,A]
+; CHECK-NEXT:    # fixup A - offset: 1, value: .LBB48_1-1, kind: FK_PCRel_1
 ; CHECK-NEXT:  # %bb.2: # %if.then
-; CHECK-NEXT:    movq 32(%rdi), %rax
-; CHECK-NEXT:    retq
+; CHECK-NEXT:    movq 32(%rdi), %rax # encoding: [0x48,0x8b,0x47,0x20]
+; CHECK-NEXT:    retq # encoding: [0xc3]
 ; CHECK-NEXT:  .LBB48_1:
-; CHECK-NEXT:    movl $123, %eax
-; CHECK-NEXT:    retq
+; CHECK-NEXT:    movl $123, %eax # encoding: [0xb8,0x7b,0x00,0x00,0x00]
+; CHECK-NEXT:    retq # encoding: [0xc3]
 entry:
   %0 = atomicrmw xor ptr %v, i64 16 monotonic, align 8
   %and = and i64 %0, 16
@@ -1422,23 +1464,25 @@
 define i64 @atomic_shl1_neq_xor_64_const_br(ptr %v) nounwind {
 ; CHECK-LABEL: atomic_shl1_neq_xor_64_const_br:
 ; CHECK:       # %bb.0: # %entry
-; CHECK-NEXT:    movq (%rdi), %rax
+; CHECK-NEXT:    movq (%rdi), %rax # encoding: [0x48,0x8b,0x07]
 ; CHECK-NEXT:    .p2align 4, 0x90
 ; CHECK-NEXT:  .LBB49_1: # %atomicrmw.start
 ; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
-; CHECK-NEXT:    movq %rax, %rcx
-; CHECK-NEXT:    xorq $16, %rcx
-; CHECK-NEXT:    lock cmpxchgq %rcx, (%rdi)
-; CHECK-NEXT:    jne .LBB49_1
+; CHECK-NEXT:    movq %rax, %rcx # encoding: [0x48,0x89,0xc1]
+; CHECK-NEXT:    xorq $16, %rcx # encoding: [0x48,0x83,0xf1,0x10]
+; CHECK-NEXT:    lock cmpxchgq %rcx, (%rdi) # encoding: [0xf0,0x48,0x0f,0xb1,0x0f]
+; CHECK-NEXT:    jne .LBB49_1 # encoding: [0x75,A]
+; CHECK-NEXT:    # fixup A - offset: 1, value: .LBB49_1-1, kind: FK_PCRel_1
 ; CHECK-NEXT:  # %bb.2: # %atomicrmw.end
-; CHECK-NEXT:    movl $123, %ecx
-; CHECK-NEXT:    testb $32, %al
-; CHECK-NEXT:    je .LBB49_4
+; CHECK-NEXT:    movl $123, %ecx # encoding: [0xb9,0x7b,0x00,0x00,0x00]
+; CHECK-NEXT:    testb $32, %al # encoding: [0xa8,0x20]
+; CHECK-NEXT:    je .LBB49_4 # encoding: [0x74,A]
+; CHECK-NEXT:    # fixup A - offset: 1, value: .LBB49_4-1, kind: FK_PCRel_1
 ; CHECK-NEXT:  # %bb.3: # %if.then
-; CHECK-NEXT:    movq 32(%rdi), %rcx
+; CHECK-NEXT:    movq 32(%rdi), %rcx # encoding: [0x48,0x8b,0x4f,0x20]
 ; CHECK-NEXT:  .LBB49_4: # %return
-; CHECK-NEXT:    movq %rcx, %rax
-; CHECK-NEXT:    retq
+; CHECK-NEXT:    movq %rcx, %rax # encoding: [0x48,0x89,0xc8]
+; CHECK-NEXT:    retq # encoding: [0xc3]
 entry:
   %0 = atomicrmw xor ptr %v, i64 16 monotonic, align 8
   %and = and i64 %0, 32
@@ -1458,17 +1502,18 @@
 define i64 @atomic_shl1_xor_64_const_brz(ptr %v) nounwind {
 ; CHECK-LABEL: atomic_shl1_xor_64_const_brz:
 ; CHECK:       # %bb.0: # %entry
-; CHECK-NEXT:    xorl %eax, %eax
-; CHECK-NEXT:    lock btcq $4, (%rdi)
-; CHECK-NEXT:    setb %al
-; CHECK-NEXT:    shlq $4, %rax
-; CHECK-NEXT:    movl $123, %eax
-; CHECK-NEXT:    je .LBB50_1
+; CHECK-NEXT:    xorl %eax, %eax # encoding: [0x31,0xc0]
+; CHECK-NEXT:    lock btcq $4, (%rdi) # encoding: [0xf0,0x48,0x0f,0xba,0x3f,0x04]
+; CHECK-NEXT:    setb %al # encoding: [0x0f,0x92,0xc0]
+; CHECK-NEXT:    shlq $4, %rax # encoding: [0x48,0xc1,0xe0,0x04]
+; CHECK-NEXT:    movl $123, %eax # encoding: [0xb8,0x7b,0x00,0x00,0x00]
+; CHECK-NEXT:    je .LBB50_1 # encoding: [0x74,A]
+; CHECK-NEXT:    # fixup A - offset: 1, value: .LBB50_1-1, kind: FK_PCRel_1
 ; CHECK-NEXT:  # %bb.2: # %return
-; CHECK-NEXT:    retq
+; CHECK-NEXT:    retq # encoding: [0xc3]
 ; CHECK-NEXT:  .LBB50_1: # %if.then
-; CHECK-NEXT:    movq 32(%rdi), %rax
-; CHECK-NEXT:    retq
+; CHECK-NEXT:    movq 32(%rdi), %rax # encoding: [0x48,0x8b,0x47,0x20]
+; CHECK-NEXT:    retq # encoding: [0xc3]
 entry:
   %0 = atomicrmw xor ptr %v, i64 16 monotonic, align 8
   %and = and i64 16, %0
@@ -1488,23 +1533,25 @@
 define i64 @atomic_shl1_neq_xor_64_const_brz(ptr %v) nounwind {
 ; CHECK-LABEL: atomic_shl1_neq_xor_64_const_brz:
 ; CHECK:       # %bb.0: # %entry
-; CHECK-NEXT:    movq (%rdi), %rax
+; CHECK-NEXT:    movq (%rdi), %rax # encoding: [0x48,0x8b,0x07]
 ; CHECK-NEXT:    .p2align 4, 0x90
 ; CHECK-NEXT:  .LBB51_1: # %atomicrmw.start
 ; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
-; CHECK-NEXT:    movq %rax, %rcx
-; CHECK-NEXT:    xorq $16, %rcx
-; CHECK-NEXT:    lock cmpxchgq %rcx, (%rdi)
-; CHECK-NEXT:    jne .LBB51_1
+; CHECK-NEXT:    movq %rax, %rcx # encoding: [0x48,0x89,0xc1]
+; CHECK-NEXT:    xorq $16, %rcx # encoding: [0x48,0x83,0xf1,0x10]
+; CHECK-NEXT:    lock cmpxchgq %rcx, (%rdi) # encoding: [0xf0,0x48,0x0f,0xb1,0x0f]
+; CHECK-NEXT:    jne .LBB51_1 # encoding: [0x75,A]
+; CHECK-NEXT:    # fixup A - offset: 1, value: .LBB51_1-1, kind: FK_PCRel_1
 ; CHECK-NEXT:  # %bb.2: # %atomicrmw.end
-; CHECK-NEXT:    movl $123, %ecx
-; CHECK-NEXT:    testb $32, %al
-; CHECK-NEXT:    jne .LBB51_4
+; CHECK-NEXT:    movl $123, %ecx # encoding: [0xb9,0x7b,0x00,0x00,0x00]
+; CHECK-NEXT:    testb $32, %al # encoding: [0xa8,0x20]
+; CHECK-NEXT:    jne .LBB51_4 # encoding: [0x75,A]
+; CHECK-NEXT:    # fixup A - offset: 1, value: .LBB51_4-1, kind: FK_PCRel_1
 ; CHECK-NEXT:  # %bb.3: # %if.then
-; CHECK-NEXT:    movq 32(%rdi), %rcx
+; CHECK-NEXT:    movq 32(%rdi), %rcx # encoding: [0x48,0x8b,0x4f,0x20]
 ; CHECK-NEXT:  .LBB51_4: # %return
-; CHECK-NEXT:    movq %rcx, %rax
-; CHECK-NEXT:    retq
+; CHECK-NEXT:    movq %rcx, %rax # encoding: [0x48,0x89,0xc8]
+; CHECK-NEXT:    retq # encoding: [0xc3]
 entry:
   %0 = atomicrmw xor ptr %v, i64 16 monotonic, align 8
   %and = and i64 %0, 32
@@ -1524,17 +1571,18 @@
 define i64 @atomic_shl1_xor_64_const_brnz(ptr %v) nounwind {
 ; CHECK-LABEL: atomic_shl1_xor_64_const_brnz:
 ; CHECK:       # %bb.0: # %entry
-; CHECK-NEXT:    xorl %eax, %eax
-; CHECK-NEXT:    lock btcq $4, (%rdi)
-; CHECK-NEXT:    setb %al
-; CHECK-NEXT:    shlq $4, %rax
-; CHECK-NEXT:    je .LBB52_1
+; CHECK-NEXT:    xorl %eax, %eax # encoding: [0x31,0xc0]
+; CHECK-NEXT:    lock btcq $4, (%rdi) # encoding: [0xf0,0x48,0x0f,0xba,0x3f,0x04]
+; CHECK-NEXT:    setb %al # encoding: [0x0f,0x92,0xc0]
+; CHECK-NEXT:    shlq $4, %rax # encoding: [0x48,0xc1,0xe0,0x04]
+; CHECK-NEXT:    je .LBB52_1 # encoding: [0x74,A]
+; CHECK-NEXT:    # fixup A - offset: 1, value: .LBB52_1-1, kind: FK_PCRel_1
 ; CHECK-NEXT:  # %bb.2: # %if.then
-; CHECK-NEXT:    movq 32(%rdi), %rax
-; CHECK-NEXT:    retq
+; CHECK-NEXT:    movq 32(%rdi), %rax # encoding: [0x48,0x8b,0x47,0x20]
+; CHECK-NEXT:    retq # encoding: [0xc3]
 ; CHECK-NEXT:  .LBB52_1:
-; CHECK-NEXT:    movl $123, %eax
-; CHECK-NEXT:    retq
+; CHECK-NEXT:    movl $123, %eax # encoding: [0xb8,0x7b,0x00,0x00,0x00]
+; CHECK-NEXT:    retq # encoding: [0xc3]
 entry:
   %0 = atomicrmw xor ptr %v, i64 16 monotonic, align 8
   %and = and i64 %0, 16
@@ -1554,23 +1602,25 @@
 define i64 @atomic_shl1_neq_xor_64_const_brnz(ptr %v) nounwind {
 ; CHECK-LABEL: atomic_shl1_neq_xor_64_const_brnz:
 ; CHECK:       # %bb.0: # %entry
-; CHECK-NEXT:    movq (%rdi), %rax
+; CHECK-NEXT:    movq (%rdi), %rax # encoding: [0x48,0x8b,0x07]
 ; CHECK-NEXT:    .p2align 4, 0x90
 ; CHECK-NEXT:  .LBB53_1: # %atomicrmw.start
 ; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
-; CHECK-NEXT:    movq %rax, %rcx
-; CHECK-NEXT:    xorq $16, %rcx
-; CHECK-NEXT:    lock cmpxchgq %rcx, (%rdi)
-; CHECK-NEXT:    jne .LBB53_1
+; CHECK-NEXT:    movq %rax, %rcx # encoding: [0x48,0x89,0xc1]
+; CHECK-NEXT:    xorq $16, %rcx # encoding: [0x48,0x83,0xf1,0x10]
+; CHECK-NEXT:    lock cmpxchgq %rcx, (%rdi) # encoding: [0xf0,0x48,0x0f,0xb1,0x0f]
+; CHECK-NEXT:    jne .LBB53_1 # encoding: [0x75,A]
+; CHECK-NEXT:    # fixup A - offset: 1, value: .LBB53_1-1, kind: FK_PCRel_1
 ; CHECK-NEXT:  # %bb.2: # %atomicrmw.end
-; CHECK-NEXT:    movl $123, %ecx
-; CHECK-NEXT:    testb $32, %al
-; CHECK-NEXT:    je .LBB53_4
+; CHECK-NEXT:    movl $123, %ecx # encoding: [0xb9,0x7b,0x00,0x00,0x00]
+; CHECK-NEXT:    testb $32, %al # encoding: [0xa8,0x20]
+; CHECK-NEXT:    je .LBB53_4 # encoding: [0x74,A]
+; CHECK-NEXT:    # fixup A - offset: 1, value: .LBB53_4-1, kind: FK_PCRel_1
 ; CHECK-NEXT:  # %bb.3: # %if.then
-; CHECK-NEXT:    movq 32(%rdi), %rcx
+; CHECK-NEXT:    movq 32(%rdi), %rcx # encoding: [0x48,0x8b,0x4f,0x20]
 ; CHECK-NEXT:  .LBB53_4: # %return
-; CHECK-NEXT:    movq %rcx, %rax
-; CHECK-NEXT:    retq
+; CHECK-NEXT:    movq %rcx, %rax # encoding: [0x48,0x89,0xc8]
+; CHECK-NEXT:    retq # encoding: [0xc3]
 entry:
   %0 = atomicrmw xor ptr %v, i64 16 monotonic, align 8
   %and = and i64 %0, 32
@@ -1590,19 +1640,37 @@
 define i64 @atomic_and_with_not_arg(ptr %v, i64 %c) nounwind {
 ; CHECK-LABEL: atomic_and_with_not_arg:
 ; CHECK:       # %bb.0: # %entry
-; CHECK-NEXT:    movq $-1, %rcx
-; CHECK-NEXT:    movq (%rdi), %rax
+; CHECK-NEXT:    movq $-1, %rcx # encoding: [0x48,0xc7,0xc1,0xff,0xff,0xff,0xff]
+; CHECK-NEXT:    movq (%rdi), %rax # encoding: [0x48,0x8b,0x07]
 ; CHECK-NEXT:    .p2align 4, 0x90
 ; CHECK-NEXT:  .LBB54_1: # %atomicrmw.start
 ; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
-; CHECK-NEXT:    movq %rax, %rdx
-; CHECK-NEXT:    orq %rcx, %rdx
-; CHECK-NEXT:    lock cmpxchgq %rdx, (%rdi)
-; CHECK-NEXT:    jne .LBB54_1
+; CHECK-NEXT:    movq %rax, %rdx # encoding: [0x48,0x89,0xc2]
+; CHECK-NEXT:    orq %rcx, %rdx # encoding: [0x48,0x09,0xca]
+; CHECK-NEXT:    lock cmpxchgq %rdx, (%rdi) # encoding: [0xf0,0x48,0x0f,0xb1,0x17]
+; CHECK-NEXT:    jne .LBB54_1 # encoding: [0x75,A]
+; CHECK-NEXT:    # fixup A - offset: 1, value: .LBB54_1-1, kind: FK_PCRel_1
 ; CHECK-NEXT:  # %bb.2: # %atomicrmw.end
-; CHECK-NEXT:    retq
+; CHECK-NEXT:    retq # encoding: [0xc3]
   entry:
   %0 = xor i64 0, -1
   %1 = atomicrmw or ptr %v, i64 %0 monotonic, align 8
   ret i64 %1
 }
+
+@a = external dso_local global { { i64 } }
+
+define i32 @atomic_global() nounwind {
+; CHECK-LABEL: atomic_global:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    xorl %eax, %eax # encoding: [0x31,0xc0]
+; CHECK-NEXT:    lock btsq %rax, a(%rip) # encoding: [0xf0,0x48,0x0f,0xab,0x05,A,A,A,A]
+; CHECK-NEXT:    # fixup A - offset: 5, value: a-4, kind: reloc_riprel_4byte
+; CHECK-NEXT:    xorl %eax, %eax # encoding: [0x31,0xc0]
+; CHECK-NEXT:    retq # encoding: [0xc3]
+entry:
+  %shl.i = shl i64 1, 0
+  %0 = atomicrmw or ptr @a, i64 %shl.i monotonic, align 8
+  %and.i = and i64 %shl.i, %0
+  ret i32 0
+}