diff --git a/llvm/test/CodeGen/LoongArch/ir-instruction/atomicrmw.ll b/llvm/test/CodeGen/LoongArch/ir-instruction/atomicrmw.ll
--- a/llvm/test/CodeGen/LoongArch/ir-instruction/atomicrmw.ll
+++ b/llvm/test/CodeGen/LoongArch/ir-instruction/atomicrmw.ll
@@ -50,6 +50,92 @@
   ret i8 %1
 }
 
+define i8 @atomicrmw_xchg_0_i8_acquire(ptr %a) nounwind {
+; LA32-LABEL: atomicrmw_xchg_0_i8_acquire:
+; LA32:       # %bb.0:
+; LA32-NEXT:    slli.w $a1, $a0, 3
+; LA32-NEXT:    ori $a2, $zero, 255
+; LA32-NEXT:    sll.w $a2, $a2, $a1
+; LA32-NEXT:    bstrins.w $a0, $zero, 1, 0
+; LA32-NEXT:  .LBB1_1: # =>This Inner Loop Header: Depth=1
+; LA32-NEXT:    dbar 0
+; LA32-NEXT:    ll.w $a3, $a0, 0
+; LA32-NEXT:    addi.w $a4, $zero, 0
+; LA32-NEXT:    xor $a4, $a3, $a4
+; LA32-NEXT:    and $a4, $a4, $a2
+; LA32-NEXT:    xor $a4, $a3, $a4
+; LA32-NEXT:    sc.w $a4, $a0, 0
+; LA32-NEXT:    beqz $a4, .LBB1_1
+; LA32-NEXT:  # %bb.2:
+; LA32-NEXT:    srl.w $a0, $a3, $a1
+; LA32-NEXT:    ret
+;
+; LA64-LABEL: atomicrmw_xchg_0_i8_acquire:
+; LA64:       # %bb.0:
+; LA64-NEXT:    slli.d $a1, $a0, 3
+; LA64-NEXT:    ori $a2, $zero, 255
+; LA64-NEXT:    sll.w $a2, $a2, $a1
+; LA64-NEXT:    addi.w $a2, $a2, 0
+; LA64-NEXT:    bstrins.d $a0, $zero, 1, 0
+; LA64-NEXT:  .LBB1_1: # =>This Inner Loop Header: Depth=1
+; LA64-NEXT:    dbar 0
+; LA64-NEXT:    ll.w $a3, $a0, 0
+; LA64-NEXT:    addi.w $a4, $zero, 0
+; LA64-NEXT:    xor $a4, $a3, $a4
+; LA64-NEXT:    and $a4, $a4, $a2
+; LA64-NEXT:    xor $a4, $a3, $a4
+; LA64-NEXT:    sc.w $a4, $a0, 0
+; LA64-NEXT:    beqz $a4, .LBB1_1
+; LA64-NEXT:  # %bb.2:
+; LA64-NEXT:    srl.w $a0, $a3, $a1
+; LA64-NEXT:    ret
+  %1 = atomicrmw xchg ptr %a, i8 0 acquire
+  ret i8 %1
+}
+
+define i8 @atomicrmw_xchg_minus_1_i8_acquire(ptr %a) nounwind {
+; LA32-LABEL: atomicrmw_xchg_minus_1_i8_acquire:
+; LA32:       # %bb.0:
+; LA32-NEXT:    slli.w $a1, $a0, 3
+; LA32-NEXT:    ori $a2, $zero, 255
+; LA32-NEXT:    sll.w $a2, $a2, $a1
+; LA32-NEXT:    bstrins.w $a0, $zero, 1, 0
+; LA32-NEXT:  .LBB2_1: # =>This Inner Loop Header: Depth=1
+; LA32-NEXT:    dbar 0
+; LA32-NEXT:    ll.w $a3, $a0, 0
+; LA32-NEXT:    addi.w $a4, $a2, 0
+; LA32-NEXT:    xor $a4, $a3, $a4
+; LA32-NEXT:    and $a4, $a4, $a2
+; LA32-NEXT:    xor $a4, $a3, $a4
+; LA32-NEXT:    sc.w $a4, $a0, 0
+; LA32-NEXT:    beqz $a4, .LBB2_1
+; LA32-NEXT:  # %bb.2:
+; LA32-NEXT:    srl.w $a0, $a3, $a1
+; LA32-NEXT:    ret
+;
+; LA64-LABEL: atomicrmw_xchg_minus_1_i8_acquire:
+; LA64:       # %bb.0:
+; LA64-NEXT:    slli.d $a1, $a0, 3
+; LA64-NEXT:    ori $a2, $zero, 255
+; LA64-NEXT:    sll.w $a2, $a2, $a1
+; LA64-NEXT:    addi.w $a2, $a2, 0
+; LA64-NEXT:    bstrins.d $a0, $zero, 1, 0
+; LA64-NEXT:  .LBB2_1: # =>This Inner Loop Header: Depth=1
+; LA64-NEXT:    dbar 0
+; LA64-NEXT:    ll.w $a3, $a0, 0
+; LA64-NEXT:    addi.w $a4, $a2, 0
+; LA64-NEXT:    xor $a4, $a3, $a4
+; LA64-NEXT:    and $a4, $a4, $a2
+; LA64-NEXT:    xor $a4, $a3, $a4
+; LA64-NEXT:    sc.w $a4, $a0, 0
+; LA64-NEXT:    beqz $a4, .LBB2_1
+; LA64-NEXT:  # %bb.2:
+; LA64-NEXT:    srl.w $a0, $a3, $a1
+; LA64-NEXT:    ret
+  %1 = atomicrmw xchg ptr %a, i8 -1 acquire
+  ret i8 %1
+}
+
 define i16 @atomicrmw_xchg_i16_acquire(ptr %a, i16 %b) nounwind {
 ; LA32-LABEL: atomicrmw_xchg_i16_acquire:
 ; LA32:       # %bb.0:
@@ -60,7 +146,7 @@
 ; LA32-NEXT:    bstrpick.w $a1, $a1, 15, 0
 ; LA32-NEXT:    sll.w $a1, $a1, $a3
 ; LA32-NEXT:    bstrins.w $a0, $zero, 1, 0
-; LA32-NEXT:  .LBB1_1: # =>This Inner Loop Header: Depth=1
+; LA32-NEXT:  .LBB3_1: # =>This Inner Loop Header: Depth=1
 ; LA32-NEXT:    dbar 0
 ; LA32-NEXT:    ll.w $a4, $a0, 0
 ; LA32-NEXT:    addi.w $a5, $a1, 0
@@ -68,7 +154,7 @@
 ; LA32-NEXT:    and $a5, $a5, $a2
 ; LA32-NEXT:    xor $a5, $a4, $a5
 ; LA32-NEXT:    sc.w $a5, $a0, 0
-; LA32-NEXT:    beqz $a5, .LBB1_1
+; LA32-NEXT:    beqz $a5, .LBB3_1
 ; LA32-NEXT:  # %bb.2:
 ; LA32-NEXT:    srl.w $a0, $a4, $a3
 ; LA32-NEXT:    ret
@@ -84,7 +170,7 @@
 ; LA64-NEXT:    sll.w $a1, $a1, $a3
 ; LA64-NEXT:    addi.w $a1, $a1, 0
 ; LA64-NEXT:    bstrins.d $a0, $zero, 1, 0
-; LA64-NEXT:  .LBB1_1: # =>This Inner Loop Header: Depth=1
+; LA64-NEXT:  .LBB3_1: # =>This Inner Loop Header: Depth=1
 ; LA64-NEXT:    dbar 0
 ; LA64-NEXT:    ll.w $a4, $a0, 0
 ; LA64-NEXT:    addi.w $a5, $a1, 0
@@ -92,7 +178,7 @@
 ; LA64-NEXT:    and $a5, $a5, $a2
 ; LA64-NEXT:    xor $a5, $a4, $a5
 ; LA64-NEXT:    sc.w $a5, $a0, 0
-; LA64-NEXT:    beqz $a5, .LBB1_1
+; LA64-NEXT:    beqz $a5, .LBB3_1
 ; LA64-NEXT:  # %bb.2:
 ; LA64-NEXT:    srl.w $a0, $a4, $a3
 ; LA64-NEXT:    ret
@@ -100,15 +186,105 @@
   ret i16 %1
 }
 
+define i16 @atomicrmw_xchg_0_i16_acquire(ptr %a) nounwind {
+; LA32-LABEL: atomicrmw_xchg_0_i16_acquire:
+; LA32:       # %bb.0:
+; LA32-NEXT:    lu12i.w $a1, 15
+; LA32-NEXT:    ori $a1, $a1, 4095
+; LA32-NEXT:    slli.w $a2, $a0, 3
+; LA32-NEXT:    sll.w $a1, $a1, $a2
+; LA32-NEXT:    bstrins.w $a0, $zero, 1, 0
+; LA32-NEXT:  .LBB4_1: # =>This Inner Loop Header: Depth=1
+; LA32-NEXT:    dbar 0
+; LA32-NEXT:    ll.w $a3, $a0, 0
+; LA32-NEXT:    addi.w $a4, $zero, 0
+; LA32-NEXT:    xor $a4, $a3, $a4
+; LA32-NEXT:    and $a4, $a4, $a1
+; LA32-NEXT:    xor $a4, $a3, $a4
+; LA32-NEXT:    sc.w $a4, $a0, 0
+; LA32-NEXT:    beqz $a4, .LBB4_1
+; LA32-NEXT:  # %bb.2:
+; LA32-NEXT:    srl.w $a0, $a3, $a2
+; LA32-NEXT:    ret
+;
+; LA64-LABEL: atomicrmw_xchg_0_i16_acquire:
+; LA64:       # %bb.0:
+; LA64-NEXT:    lu12i.w $a1, 15
+; LA64-NEXT:    ori $a1, $a1, 4095
+; LA64-NEXT:    slli.d $a2, $a0, 3
+; LA64-NEXT:    sll.w $a1, $a1, $a2
+; LA64-NEXT:    addi.w $a1, $a1, 0
+; LA64-NEXT:    bstrins.d $a0, $zero, 1, 0
+; LA64-NEXT:  .LBB4_1: # =>This Inner Loop Header: Depth=1
+; LA64-NEXT:    dbar 0
+; LA64-NEXT:    ll.w $a3, $a0, 0
+; LA64-NEXT:    addi.w $a4, $zero, 0
+; LA64-NEXT:    xor $a4, $a3, $a4
+; LA64-NEXT:    and $a4, $a4, $a1
+; LA64-NEXT:    xor $a4, $a3, $a4
+; LA64-NEXT:    sc.w $a4, $a0, 0
+; LA64-NEXT:    beqz $a4, .LBB4_1
+; LA64-NEXT:  # %bb.2:
+; LA64-NEXT:    srl.w $a0, $a3, $a2
+; LA64-NEXT:    ret
+  %1 = atomicrmw xchg ptr %a, i16 0 acquire
+  ret i16 %1
+}
+
+define i16 @atomicrmw_xchg_minus_1_i16_acquire(ptr %a) nounwind {
+; LA32-LABEL: atomicrmw_xchg_minus_1_i16_acquire:
+; LA32:       # %bb.0:
+; LA32-NEXT:    lu12i.w $a1, 15
+; LA32-NEXT:    ori $a1, $a1, 4095
+; LA32-NEXT:    slli.w $a2, $a0, 3
+; LA32-NEXT:    sll.w $a1, $a1, $a2
+; LA32-NEXT:    bstrins.w $a0, $zero, 1, 0
+; LA32-NEXT:  .LBB5_1: # =>This Inner Loop Header: Depth=1
+; LA32-NEXT:    dbar 0
+; LA32-NEXT:    ll.w $a3, $a0, 0
+; LA32-NEXT:    addi.w $a4, $a1, 0
+; LA32-NEXT:    xor $a4, $a3, $a4
+; LA32-NEXT:    and $a4, $a4, $a1
+; LA32-NEXT:    xor $a4, $a3, $a4
+; LA32-NEXT:    sc.w $a4, $a0, 0
+; LA32-NEXT:    beqz $a4, .LBB5_1
+; LA32-NEXT:  # %bb.2:
+; LA32-NEXT:    srl.w $a0, $a3, $a2
+; LA32-NEXT:    ret
+;
+; LA64-LABEL: atomicrmw_xchg_minus_1_i16_acquire:
+; LA64:       # %bb.0:
+; LA64-NEXT:    lu12i.w $a1, 15
+; LA64-NEXT:    ori $a1, $a1, 4095
+; LA64-NEXT:    slli.d $a2, $a0, 3
+; LA64-NEXT:    sll.w $a1, $a1, $a2
+; LA64-NEXT:    addi.w $a1, $a1, 0
+; LA64-NEXT:    bstrins.d $a0, $zero, 1, 0
+; LA64-NEXT:  .LBB5_1: # =>This Inner Loop Header: Depth=1
+; LA64-NEXT:    dbar 0
+; LA64-NEXT:    ll.w $a3, $a0, 0
+; LA64-NEXT:    addi.w $a4, $a1, 0
+; LA64-NEXT:    xor $a4, $a3, $a4
+; LA64-NEXT:    and $a4, $a4, $a1
+; LA64-NEXT:    xor $a4, $a3, $a4
+; LA64-NEXT:    sc.w $a4, $a0, 0
+; LA64-NEXT:    beqz $a4, .LBB5_1
+; LA64-NEXT:  # %bb.2:
+; LA64-NEXT:    srl.w $a0, $a3, $a2
+; LA64-NEXT:    ret
+  %1 = atomicrmw xchg ptr %a, i16 -1 acquire
+  ret i16 %1
+}
+
 define i32 @atomicrmw_xchg_i32_acquire(ptr %a, i32 %b) nounwind {
 ; LA32-LABEL: atomicrmw_xchg_i32_acquire:
 ; LA32:       # %bb.0:
-; LA32-NEXT:  .LBB2_1: # =>This Inner Loop Header: Depth=1
+; LA32-NEXT:  .LBB6_1: # =>This Inner Loop Header: Depth=1
 ; LA32-NEXT:    dbar 0
 ; LA32-NEXT:    ll.w $a2, $a0, 0
 ; LA32-NEXT:    move $a3, $a1
 ; LA32-NEXT:    sc.w $a3, $a0, 0
-; LA32-NEXT:    beqz $a3, .LBB2_1
+; LA32-NEXT:    beqz $a3, .LBB6_1
 ; LA32-NEXT:  # %bb.2:
 ; LA32-NEXT:    move $a0, $a2
 ; LA32-NEXT:    ret
@@ -151,7 +327,7 @@
 ; LA32-NEXT:    andi $a1, $a1, 255
 ; LA32-NEXT:    sll.w $a1, $a1, $a2
 ; LA32-NEXT:    bstrins.w $a0, $zero, 1, 0
-; LA32-NEXT:  .LBB4_1: # =>This Inner Loop Header: Depth=1
+; LA32-NEXT:  .LBB8_1: # =>This Inner Loop Header: Depth=1
 ; LA32-NEXT:    dbar 0
 ; LA32-NEXT:    ll.w $a4, $a0, 0
 ; LA32-NEXT:    add.w $a5, $a4, $a1
@@ -159,7 +335,7 @@
 ; LA32-NEXT:    and $a5, $a5, $a3
 ; LA32-NEXT:    xor $a5, $a4, $a5
 ; LA32-NEXT:    sc.w $a5, $a0, 0
-; LA32-NEXT:    beqz $a5, .LBB4_1
+; LA32-NEXT:    beqz $a5, .LBB8_1
 ; LA32-NEXT:  # %bb.2:
 ; LA32-NEXT:    srl.w $a0, $a4, $a2
 ; LA32-NEXT:    ret
@@ -174,7 +350,7 @@
 ; LA64-NEXT:    sll.w $a1, $a1, $a2
 ; LA64-NEXT:    addi.w $a1, $a1, 0
 ; LA64-NEXT:    bstrins.d $a0, $zero, 1, 0
-; LA64-NEXT:  .LBB4_1: # =>This Inner Loop Header: Depth=1
+; LA64-NEXT:  .LBB8_1: # =>This Inner Loop Header: Depth=1
 ; LA64-NEXT:    dbar 0
 ; LA64-NEXT:    ll.w $a4, $a0, 0
 ; LA64-NEXT:    add.w $a5, $a4, $a1
@@ -182,7 +358,7 @@
 ; LA64-NEXT:    and $a5, $a5, $a3
 ; LA64-NEXT:    xor $a5, $a4, $a5
 ; LA64-NEXT:    sc.w $a5, $a0, 0
-; LA64-NEXT:    beqz $a5, .LBB4_1
+; LA64-NEXT:    beqz $a5, .LBB8_1
 ; LA64-NEXT:  # %bb.2:
 ; LA64-NEXT:    srl.w $a0, $a4, $a2
 ; LA64-NEXT:    ret
@@ -200,7 +376,7 @@
 ; LA32-NEXT:    bstrpick.w $a1, $a1, 15, 0
 ; LA32-NEXT:    sll.w $a1, $a1, $a3
 ; LA32-NEXT:    bstrins.w $a0, $zero, 1, 0
-; LA32-NEXT:  .LBB5_1: # =>This Inner Loop Header: Depth=1
+; LA32-NEXT:  .LBB9_1: # =>This Inner Loop Header: Depth=1
 ; LA32-NEXT:    dbar 0
 ; LA32-NEXT:    ll.w $a4, $a0, 0
 ; LA32-NEXT:    add.w $a5, $a4, $a1
@@ -208,7 +384,7 @@
 ; LA32-NEXT:    and $a5, $a5, $a2
 ; LA32-NEXT:    xor $a5, $a4, $a5
 ; LA32-NEXT:    sc.w $a5, $a0, 0
-; LA32-NEXT:    beqz $a5, .LBB5_1
+; LA32-NEXT:    beqz $a5, .LBB9_1
 ; LA32-NEXT:  # %bb.2:
 ; LA32-NEXT:    srl.w $a0, $a4, $a3
 ; LA32-NEXT:    ret
@@ -224,7 +400,7 @@
 ; LA64-NEXT:    sll.w $a1, $a1, $a3
 ; LA64-NEXT:    addi.w $a1, $a1, 0
 ; LA64-NEXT:    bstrins.d $a0, $zero, 1, 0
-; LA64-NEXT:  .LBB5_1: # =>This Inner Loop Header: Depth=1
+; LA64-NEXT:  .LBB9_1: # =>This Inner Loop Header: Depth=1
 ; LA64-NEXT:    dbar 0
 ; LA64-NEXT:    ll.w $a4, $a0, 0
 ; LA64-NEXT:    add.w $a5, $a4, $a1
@@ -232,7 +408,7 @@
 ; LA64-NEXT:    and $a5, $a5, $a2
 ; LA64-NEXT:    xor $a5, $a4, $a5
 ; LA64-NEXT:    sc.w $a5, $a0, 0
-; LA64-NEXT:    beqz $a5, .LBB5_1
+; LA64-NEXT:    beqz $a5, .LBB9_1
 ; LA64-NEXT:  # %bb.2:
 ; LA64-NEXT:    srl.w $a0, $a4, $a3
 ; LA64-NEXT:    ret
@@ -243,12 +419,12 @@
 define i32 @atomicrmw_add_i32_acquire(ptr %a, i32 %b) nounwind {
 ; LA32-LABEL: atomicrmw_add_i32_acquire:
 ; LA32:       # %bb.0:
-; LA32-NEXT:  .LBB6_1: # =>This Inner Loop Header: Depth=1
+; LA32-NEXT:  .LBB10_1: # =>This Inner Loop Header: Depth=1
 ; LA32-NEXT:    dbar 0
 ; LA32-NEXT:    ll.w $a2, $a0, 0
 ; LA32-NEXT:    add.w $a3, $a2, $a1
 ; LA32-NEXT:    sc.w $a3, $a0, 0
-; LA32-NEXT:    beqz $a3, .LBB6_1
+; LA32-NEXT:    beqz $a3, .LBB10_1
 ; LA32-NEXT:  # %bb.2:
 ; LA32-NEXT:    move $a0, $a2
 ; LA32-NEXT:    ret
@@ -291,7 +467,7 @@
 ; LA32-NEXT:    andi $a1, $a1, 255
 ; LA32-NEXT:    sll.w $a1, $a1, $a2
 ; LA32-NEXT:    bstrins.w $a0, $zero, 1, 0
-; LA32-NEXT:  .LBB8_1: # =>This Inner Loop Header: Depth=1
+; LA32-NEXT:  .LBB12_1: # =>This Inner Loop Header: Depth=1
 ; LA32-NEXT:    dbar 0
 ; LA32-NEXT:    ll.w $a4, $a0, 0
 ; LA32-NEXT:    sub.w $a5, $a4, $a1
@@ -299,7 +475,7 @@
 ; LA32-NEXT:    and $a5, $a5, $a3
 ; LA32-NEXT:    xor $a5, $a4, $a5
 ; LA32-NEXT:    sc.w $a5, $a0, 0
-; LA32-NEXT:    beqz $a5, .LBB8_1
+; LA32-NEXT:    beqz $a5, .LBB12_1
 ; LA32-NEXT:  # %bb.2:
 ; LA32-NEXT:    srl.w $a0, $a4, $a2
 ; LA32-NEXT:    ret
@@ -314,7 +490,7 @@
 ; LA64-NEXT:    sll.w $a1, $a1, $a2
 ; LA64-NEXT:    addi.w $a1, $a1, 0
 ; LA64-NEXT:    bstrins.d $a0, $zero, 1, 0
-; LA64-NEXT:  .LBB8_1: # =>This Inner Loop Header: Depth=1
+; LA64-NEXT:  .LBB12_1: # =>This Inner Loop Header: Depth=1
 ; LA64-NEXT:    dbar 0
 ; LA64-NEXT:    ll.w $a4, $a0, 0
 ; LA64-NEXT:    sub.w $a5, $a4, $a1
@@ -322,7 +498,7 @@
 ; LA64-NEXT:    and $a5, $a5, $a3
 ; LA64-NEXT:    xor $a5, $a4, $a5
 ; LA64-NEXT:    sc.w $a5, $a0, 0
-; LA64-NEXT:    beqz $a5, .LBB8_1
+; LA64-NEXT:    beqz $a5, .LBB12_1
 ; LA64-NEXT:  # %bb.2:
 ; LA64-NEXT:    srl.w $a0, $a4, $a2
 ; LA64-NEXT:    ret
@@ -340,7 +516,7 @@
 ; LA32-NEXT:    bstrpick.w $a1, $a1, 15, 0
 ; LA32-NEXT:    sll.w $a1, $a1, $a3
 ; LA32-NEXT:    bstrins.w $a0, $zero, 1, 0
-; LA32-NEXT:  .LBB9_1: # =>This Inner Loop Header: Depth=1
+; LA32-NEXT:  .LBB13_1: # =>This Inner Loop Header: Depth=1
 ; LA32-NEXT:    dbar 0
 ; LA32-NEXT:    ll.w $a4, $a0, 0
 ; LA32-NEXT:    sub.w $a5, $a4, $a1
@@ -348,7 +524,7 @@
 ; LA32-NEXT:    and $a5, $a5, $a2
 ; LA32-NEXT:    xor $a5, $a4, $a5
 ; LA32-NEXT:    sc.w $a5, $a0, 0
-; LA32-NEXT:    beqz $a5, .LBB9_1
+; LA32-NEXT:    beqz $a5, .LBB13_1
 ; LA32-NEXT:  # %bb.2:
 ; LA32-NEXT:    srl.w $a0, $a4, $a3
 ; LA32-NEXT:    ret
@@ -364,7 +540,7 @@
 ; LA64-NEXT:    sll.w $a1, $a1, $a3
 ; LA64-NEXT:    addi.w $a1, $a1, 0
 ; LA64-NEXT:    bstrins.d $a0, $zero, 1, 0
-; LA64-NEXT:  .LBB9_1: # =>This Inner Loop Header: Depth=1
+; LA64-NEXT:  .LBB13_1: # =>This Inner Loop Header: Depth=1
 ; LA64-NEXT:    dbar 0
 ; LA64-NEXT:    ll.w $a4, $a0, 0
 ; LA64-NEXT:    sub.w $a5, $a4, $a1
@@ -372,7 +548,7 @@
 ; LA64-NEXT:    and $a5, $a5, $a2
 ; LA64-NEXT:    xor $a5, $a4, $a5
 ; LA64-NEXT:    sc.w $a5, $a0, 0
-; LA64-NEXT:    beqz $a5, .LBB9_1
+; LA64-NEXT:    beqz $a5, .LBB13_1
 ; LA64-NEXT:  # %bb.2:
 ; LA64-NEXT:    srl.w $a0, $a4, $a3
 ; LA64-NEXT:    ret
@@ -383,12 +559,12 @@
 define i32 @atomicrmw_sub_i32_acquire(ptr %a, i32 %b) nounwind {
 ; LA32-LABEL: atomicrmw_sub_i32_acquire:
 ; LA32:       # %bb.0:
-; LA32-NEXT:  .LBB10_1: # =>This Inner Loop Header: Depth=1
+; LA32-NEXT:  .LBB14_1: # =>This Inner Loop Header: Depth=1
 ; LA32-NEXT:    dbar 0
 ; LA32-NEXT:    ll.w $a2, $a0, 0
 ; LA32-NEXT:    sub.w $a3, $a2, $a1
 ; LA32-NEXT:    sc.w $a3, $a0, 0
-; LA32-NEXT:    beqz $a3, .LBB10_1
+; LA32-NEXT:    beqz $a3, .LBB14_1
 ; LA32-NEXT:  # %bb.2:
 ; LA32-NEXT:    move $a0, $a2
 ; LA32-NEXT:    ret
@@ -433,7 +609,7 @@
 ; LA32-NEXT:    andi $a1, $a1, 255
 ; LA32-NEXT:    sll.w $a1, $a1, $a2
 ; LA32-NEXT:    bstrins.w $a0, $zero, 1, 0
-; LA32-NEXT:  .LBB12_1: # =>This Inner Loop Header: Depth=1
+; LA32-NEXT:  .LBB16_1: # =>This Inner Loop Header: Depth=1
 ; LA32-NEXT:    dbar 0
 ; LA32-NEXT:    ll.w $a4, $a0, 0
 ; LA32-NEXT:    and $a5, $a4, $a1
@@ -442,7 +618,7 @@
 ; LA32-NEXT:    and $a5, $a5, $a3
 ; LA32-NEXT:    xor $a5, $a4, $a5
 ; LA32-NEXT:    sc.w $a5, $a0, 0
-; LA32-NEXT:    beqz $a5, .LBB12_1
+; LA32-NEXT:    beqz $a5, .LBB16_1
 ; LA32-NEXT:  # %bb.2:
 ; LA32-NEXT:    srl.w $a0, $a4, $a2
 ; LA32-NEXT:    ret
@@ -457,7 +633,7 @@
 ; LA64-NEXT:    sll.w $a1, $a1, $a2
 ; LA64-NEXT:    addi.w $a1, $a1, 0
 ; LA64-NEXT:    bstrins.d $a0, $zero, 1, 0
-; LA64-NEXT:  .LBB12_1: # =>This Inner Loop Header: Depth=1
+; LA64-NEXT:  .LBB16_1: # =>This Inner Loop Header: Depth=1
 ; LA64-NEXT:    dbar 0
 ; LA64-NEXT:    ll.w $a4, $a0, 0
 ; LA64-NEXT:    and $a5, $a4, $a1
@@ -466,7 +642,7 @@
 ; LA64-NEXT:    and $a5, $a5, $a3
 ; LA64-NEXT:    xor $a5, $a4, $a5
 ; LA64-NEXT:    sc.w $a5, $a0, 0
-; LA64-NEXT:    beqz $a5, .LBB12_1
+; LA64-NEXT:    beqz $a5, .LBB16_1
 ; LA64-NEXT:  # %bb.2:
 ; LA64-NEXT:    srl.w $a0, $a4, $a2
 ; LA64-NEXT:    ret
@@ -484,7 +660,7 @@
 ; LA32-NEXT:    bstrpick.w $a1, $a1, 15, 0
 ; LA32-NEXT:    sll.w $a1, $a1, $a3
 ; LA32-NEXT:    bstrins.w $a0, $zero, 1, 0
-; LA32-NEXT:  .LBB13_1: # =>This Inner Loop Header: Depth=1
+; LA32-NEXT:  .LBB17_1: # =>This Inner Loop Header: Depth=1
 ; LA32-NEXT:    dbar 0
 ; LA32-NEXT:    ll.w $a4, $a0, 0
 ; LA32-NEXT:    and $a5, $a4, $a1
@@ -493,7 +669,7 @@
 ; LA32-NEXT:    and $a5, $a5, $a2
 ; LA32-NEXT:    xor $a5, $a4, $a5
 ; LA32-NEXT:    sc.w $a5, $a0, 0
-; LA32-NEXT:    beqz $a5, .LBB13_1
+; LA32-NEXT:    beqz $a5, .LBB17_1
 ; LA32-NEXT:  # %bb.2:
 ; LA32-NEXT:    srl.w $a0, $a4, $a3
 ; LA32-NEXT:    ret
@@ -509,7 +685,7 @@
 ; LA64-NEXT:    sll.w $a1, $a1, $a3
 ; LA64-NEXT:    addi.w $a1, $a1, 0
 ; LA64-NEXT:    bstrins.d $a0, $zero, 1, 0
-; LA64-NEXT:  .LBB13_1: # =>This Inner Loop Header: Depth=1
+; LA64-NEXT:  .LBB17_1: # =>This Inner Loop Header: Depth=1
 ; LA64-NEXT:    dbar 0
 ; LA64-NEXT:    ll.w $a4, $a0, 0
 ; LA64-NEXT:    and $a5, $a4, $a1
@@ -518,7 +694,7 @@
 ; LA64-NEXT:    and $a5, $a5, $a2
 ; LA64-NEXT:    xor $a5, $a4, $a5
 ; LA64-NEXT:    sc.w $a5, $a0, 0
-; LA64-NEXT:    beqz $a5, .LBB13_1
+; LA64-NEXT:    beqz $a5, .LBB17_1
 ; LA64-NEXT:  # %bb.2:
 ; LA64-NEXT:    srl.w $a0, $a4, $a3
 ; LA64-NEXT:    ret
@@ -529,26 +705,26 @@
 define i32 @atomicrmw_nand_i32_acquire(ptr %a, i32 %b) nounwind {
 ; LA32-LABEL: atomicrmw_nand_i32_acquire:
 ; LA32:       # %bb.0:
-; LA32-NEXT:  .LBB14_1: # =>This Inner Loop Header: Depth=1
+; LA32-NEXT:  .LBB18_1: # =>This Inner Loop Header: Depth=1
 ; LA32-NEXT:    dbar 0
 ; LA32-NEXT:    ll.w $a2, $a0, 0
 ; LA32-NEXT:    and $a3, $a2, $a1
 ; LA32-NEXT:    nor $a3, $a3, $zero
 ; LA32-NEXT:    sc.w $a3, $a0, 0
-; LA32-NEXT:    beqz $a3, .LBB14_1
+; LA32-NEXT:    beqz $a3, .LBB18_1
 ; LA32-NEXT:  # %bb.2:
 ; LA32-NEXT:    move $a0, $a2
 ; LA32-NEXT:    ret
 ;
 ; LA64-LABEL: atomicrmw_nand_i32_acquire:
 ; LA64:       # %bb.0:
-; LA64-NEXT:  .LBB14_1: # =>This Inner Loop Header: Depth=1
+; LA64-NEXT:  .LBB18_1: # =>This Inner Loop Header: Depth=1
 ; LA64-NEXT:    dbar 0
 ; LA64-NEXT:    ll.w $a2, $a0, 0
 ; LA64-NEXT:    and $a3, $a2, $a1
 ; LA64-NEXT:    nor $a3, $a3, $zero
 ; LA64-NEXT:    sc.w $a3, $a0, 0
-; LA64-NEXT:    beqz $a3, .LBB14_1
+; LA64-NEXT:    beqz $a3, .LBB18_1
 ; LA64-NEXT:  # %bb.2:
 ; LA64-NEXT:    move $a0, $a2
 ; LA64-NEXT:    ret
@@ -569,13 +745,13 @@
 ;
 ; LA64-LABEL: atomicrmw_nand_i64_acquire:
 ; LA64:       # %bb.0:
-; LA64-NEXT:  .LBB15_1: # =>This Inner Loop Header: Depth=1
+; LA64-NEXT:  .LBB19_1: # =>This Inner Loop Header: Depth=1
 ; LA64-NEXT:    dbar 0
 ; LA64-NEXT:    ll.d $a2, $a0, 0
 ; LA64-NEXT:    and $a3, $a2, $a1
 ; LA64-NEXT:    nor $a3, $a3, $zero
 ; LA64-NEXT:    sc.d $a3, $a0, 0
-; LA64-NEXT:    beqz $a3, .LBB15_1
+; LA64-NEXT:    beqz $a3, .LBB19_1
 ; LA64-NEXT:  # %bb.2:
 ; LA64-NEXT:    move $a0, $a2
 ; LA64-NEXT:    ret
@@ -593,12 +769,12 @@
 ; LA32-NEXT:    sll.w $a1, $a1, $a2
 ; LA32-NEXT:    orn $a1, $a1, $a3
 ; LA32-NEXT:    bstrins.w $a0, $zero, 1, 0
-; LA32-NEXT:  .LBB16_1: # =>This Inner Loop Header: Depth=1
+; LA32-NEXT:  .LBB20_1: # =>This Inner Loop Header: Depth=1
 ; LA32-NEXT:    dbar 0
 ; LA32-NEXT:    ll.w $a3, $a0, 0
 ; LA32-NEXT:    and $a4, $a3, $a1
 ; LA32-NEXT:    sc.w $a4, $a0, 0
-; LA32-NEXT:    beqz $a4, .LBB16_1
+; LA32-NEXT:    beqz $a4, .LBB20_1
 ; LA32-NEXT:  # %bb.2:
 ; LA32-NEXT:    srl.w $a0, $a3, $a2
 ; LA32-NEXT:    ret
@@ -630,12 +806,12 @@
 ; LA32-NEXT:    sll.w $a1, $a1, $a3
 ; LA32-NEXT:    orn $a1, $a1, $a2
 ; LA32-NEXT:    bstrins.w $a0, $zero, 1, 0
-; LA32-NEXT:  .LBB17_1: # =>This Inner Loop Header: Depth=1
+; LA32-NEXT:  .LBB21_1: # =>This Inner Loop Header: Depth=1
 ; LA32-NEXT:    dbar 0
 ; LA32-NEXT:    ll.w $a2, $a0, 0
 ; LA32-NEXT:    and $a4, $a2, $a1
 ; LA32-NEXT:    sc.w $a4, $a0, 0
-; LA32-NEXT:    beqz $a4, .LBB17_1
+; LA32-NEXT:    beqz $a4, .LBB21_1
 ; LA32-NEXT:  # %bb.2:
 ; LA32-NEXT:    srl.w $a0, $a2, $a3
 ; LA32-NEXT:    ret
@@ -660,12 +836,12 @@
 define i32 @atomicrmw_and_i32_acquire(ptr %a, i32 %b) nounwind {
 ; LA32-LABEL: atomicrmw_and_i32_acquire:
 ; LA32:       # %bb.0:
-; LA32-NEXT:  .LBB18_1: # =>This Inner Loop Header: Depth=1
+; LA32-NEXT:  .LBB22_1: # =>This Inner Loop Header: Depth=1
 ; LA32-NEXT:    dbar 0
 ; LA32-NEXT:    ll.w $a2, $a0, 0
 ; LA32-NEXT:    and $a3, $a2, $a1
 ; LA32-NEXT:    sc.w $a3, $a0, 0
-; LA32-NEXT:    beqz $a3, .LBB18_1
+; LA32-NEXT:    beqz $a3, .LBB22_1
 ; LA32-NEXT:  # %bb.2:
 ; LA32-NEXT:    move $a0, $a2
 ; LA32-NEXT:    ret
@@ -706,12 +882,12 @@
 ; LA32-NEXT:    slli.w $a2, $a0, 3
 ; LA32-NEXT:    sll.w $a1, $a1, $a2
 ; LA32-NEXT:    bstrins.w $a0, $zero, 1, 0
-; LA32-NEXT:  .LBB20_1: # =>This Inner Loop Header: Depth=1
+; LA32-NEXT:  .LBB24_1: # =>This Inner Loop Header: Depth=1
 ; LA32-NEXT:    dbar 0
 ; LA32-NEXT:    ll.w $a3, $a0, 0
 ; LA32-NEXT:    or $a4, $a3, $a1
 ; LA32-NEXT:    sc.w $a4, $a0, 0
-; LA32-NEXT:    beqz $a4, .LBB20_1
+; LA32-NEXT:    beqz $a4, .LBB24_1
 ; LA32-NEXT:  # %bb.2:
 ; LA32-NEXT:    srl.w $a0, $a3, $a2
 ; LA32-NEXT:    ret
@@ -736,12 +912,12 @@
 ; LA32-NEXT:    slli.w $a2, $a0, 3
 ; LA32-NEXT:    sll.w $a1, $a1, $a2
 ; LA32-NEXT:    bstrins.w $a0, $zero, 1, 0
-; LA32-NEXT:  .LBB21_1: # =>This Inner Loop Header: Depth=1
+; LA32-NEXT:  .LBB25_1: # =>This Inner Loop Header: Depth=1
 ; LA32-NEXT:    dbar 0
 ; LA32-NEXT:    ll.w $a3, $a0, 0
 ; LA32-NEXT:    or $a4, $a3, $a1
 ; LA32-NEXT:    sc.w $a4, $a0, 0
-; LA32-NEXT:    beqz $a4, .LBB21_1
+; LA32-NEXT:    beqz $a4, .LBB25_1
 ; LA32-NEXT:  # %bb.2:
 ; LA32-NEXT:    srl.w $a0, $a3, $a2
 ; LA32-NEXT:    ret
@@ -762,12 +938,12 @@
 define i32 @atomicrmw_or_i32_acquire(ptr %a, i32 %b) nounwind {
 ; LA32-LABEL: atomicrmw_or_i32_acquire:
 ; LA32:       # %bb.0:
-; LA32-NEXT:  .LBB22_1: # =>This Inner Loop Header: Depth=1
+; LA32-NEXT:  .LBB26_1: # =>This Inner Loop Header: Depth=1
 ; LA32-NEXT:    dbar 0
 ; LA32-NEXT:    ll.w $a2, $a0, 0
 ; LA32-NEXT:    or $a3, $a2, $a1
 ; LA32-NEXT:    sc.w $a3, $a0, 0
-; LA32-NEXT:    beqz $a3, .LBB22_1
+; LA32-NEXT:    beqz $a3, .LBB26_1
 ; LA32-NEXT:  # %bb.2:
 ; LA32-NEXT:    move $a0, $a2
 ; LA32-NEXT:    ret
@@ -808,12 +984,12 @@
 ; LA32-NEXT:    slli.w $a2, $a0, 3
 ; LA32-NEXT:    sll.w $a1, $a1, $a2
 ; LA32-NEXT:    bstrins.w $a0, $zero, 1, 0
-; LA32-NEXT:  .LBB24_1: # =>This Inner Loop Header: Depth=1
+; LA32-NEXT:  .LBB28_1: # =>This Inner Loop Header: Depth=1
 ; LA32-NEXT:    dbar 0
 ; LA32-NEXT:    ll.w $a3, $a0, 0
 ; LA32-NEXT:    xor $a4, $a3, $a1
 ; LA32-NEXT:    sc.w $a4, $a0, 0
-; LA32-NEXT:    beqz $a4, .LBB24_1
+; LA32-NEXT:    beqz $a4, .LBB28_1
 ; LA32-NEXT:  # %bb.2:
 ; LA32-NEXT:    srl.w $a0, $a3, $a2
 ; LA32-NEXT:    ret
@@ -838,12 +1014,12 @@
 ; LA32-NEXT:    slli.w $a2, $a0, 3
 ; LA32-NEXT:    sll.w $a1, $a1, $a2
 ; LA32-NEXT:    bstrins.w $a0, $zero, 1, 0
-; LA32-NEXT:  .LBB25_1: # =>This Inner Loop Header: Depth=1
+; LA32-NEXT:  .LBB29_1: # =>This Inner Loop Header: Depth=1
 ; LA32-NEXT:    dbar 0
 ; LA32-NEXT:    ll.w $a3, $a0, 0
 ; LA32-NEXT:    xor $a4, $a3, $a1
 ; LA32-NEXT:    sc.w $a4, $a0, 0
-; LA32-NEXT:    beqz $a4, .LBB25_1
+; LA32-NEXT:    beqz $a4, .LBB29_1
 ; LA32-NEXT:  # %bb.2:
 ; LA32-NEXT:    srl.w $a0, $a3, $a2
 ; LA32-NEXT:    ret
@@ -864,12 +1040,12 @@
 define i32 @atomicrmw_xor_i32_acquire(ptr %a, i32 %b) nounwind {
 ; LA32-LABEL: atomicrmw_xor_i32_acquire:
 ; LA32:       # %bb.0:
-; LA32-NEXT:  .LBB26_1: # =>This Inner Loop Header: Depth=1
+; LA32-NEXT:  .LBB30_1: # =>This Inner Loop Header: Depth=1
 ; LA32-NEXT:    dbar 0
 ; LA32-NEXT:    ll.w $a2, $a0, 0
 ; LA32-NEXT:    xor $a3, $a2, $a1
 ; LA32-NEXT:    sc.w $a3, $a0, 0
-; LA32-NEXT:    beqz $a3, .LBB26_1
+; LA32-NEXT:    beqz $a3, .LBB30_1
 ; LA32-NEXT:  # %bb.2:
 ; LA32-NEXT:    move $a0, $a2
 ; LA32-NEXT:    ret
@@ -912,14 +1088,14 @@
 ; LA32-NEXT:    andi $a1, $a1, 255
 ; LA32-NEXT:    sll.w $a1, $a1, $a2
 ; LA32-NEXT:    bstrins.w $a0, $zero, 1, 0
-; LA32-NEXT:  .LBB28_1: # =>This Inner Loop Header: Depth=1
+; LA32-NEXT:  .LBB32_1: # =>This Inner Loop Header: Depth=1
 ; LA32-NEXT:    ll.w $a4, $a0, 0
 ; LA32-NEXT:    addi.w $a5, $a1, 0
 ; LA32-NEXT:    xor $a5, $a4, $a5
 ; LA32-NEXT:    and $a5, $a5, $a3
 ; LA32-NEXT:    xor $a5, $a4, $a5
 ; LA32-NEXT:    sc.w $a5, $a0, 0
-; LA32-NEXT:    beqz $a5, .LBB28_1
+; LA32-NEXT:    beqz $a5, .LBB32_1
 ; LA32-NEXT:  # %bb.2:
 ; LA32-NEXT:    srl.w $a0, $a4, $a2
 ; LA32-NEXT:    ret
@@ -934,14 +1110,14 @@
 ; LA64-NEXT:    sll.w $a1, $a1, $a2
 ; LA64-NEXT:    addi.w $a1, $a1, 0
 ; LA64-NEXT:    bstrins.d $a0, $zero, 1, 0
-; LA64-NEXT:  .LBB28_1: # =>This Inner Loop Header: Depth=1
+; LA64-NEXT:  .LBB32_1: # =>This Inner Loop Header: Depth=1
 ; LA64-NEXT:    ll.w $a4, $a0, 0
 ; LA64-NEXT:    addi.w $a5, $a1, 0
 ; LA64-NEXT:    xor $a5, $a4, $a5
 ; LA64-NEXT:    and $a5, $a5, $a3
 ; LA64-NEXT:    xor $a5, $a4, $a5
 ; LA64-NEXT:    sc.w $a5, $a0, 0
-; LA64-NEXT:    beqz $a5, .LBB28_1
+; LA64-NEXT:    beqz $a5, .LBB32_1
 ; LA64-NEXT:  # %bb.2:
 ; LA64-NEXT:    srl.w $a0, $a4, $a2
 ; LA64-NEXT:    ret
@@ -949,6 +1125,88 @@
   ret i8 %1
 }
 
+define i8 @atomicrmw_xchg_0_i8_monotonic(ptr %a) nounwind {
+; LA32-LABEL: atomicrmw_xchg_0_i8_monotonic:
+; LA32:       # %bb.0:
+; LA32-NEXT:    slli.w $a1, $a0, 3
+; LA32-NEXT:    ori $a2, $zero, 255
+; LA32-NEXT:    sll.w $a2, $a2, $a1
+; LA32-NEXT:    bstrins.w $a0, $zero, 1, 0
+; LA32-NEXT:  .LBB33_1: # =>This Inner Loop Header: Depth=1
+; LA32-NEXT:    ll.w $a3, $a0, 0
+; LA32-NEXT:    addi.w $a4, $zero, 0
+; LA32-NEXT:    xor $a4, $a3, $a4
+; LA32-NEXT:    and $a4, $a4, $a2
+; LA32-NEXT:    xor $a4, $a3, $a4
+; LA32-NEXT:    sc.w $a4, $a0, 0
+; LA32-NEXT:    beqz $a4, .LBB33_1
+; LA32-NEXT:  # %bb.2:
+; LA32-NEXT:    srl.w $a0, $a3, $a1
+; LA32-NEXT:    ret
+;
+; LA64-LABEL: atomicrmw_xchg_0_i8_monotonic:
+; LA64:       # %bb.0:
+; LA64-NEXT:    slli.d $a1, $a0, 3
+; LA64-NEXT:    ori $a2, $zero, 255
+; LA64-NEXT:    sll.w $a2, $a2, $a1
+; LA64-NEXT:    addi.w $a2, $a2, 0
+; LA64-NEXT:    bstrins.d $a0, $zero, 1, 0
+; LA64-NEXT:  .LBB33_1: # =>This Inner Loop Header: Depth=1
+; LA64-NEXT:    ll.w $a3, $a0, 0
+; LA64-NEXT:    addi.w $a4, $zero, 0
+; LA64-NEXT:    xor $a4, $a3, $a4
+; LA64-NEXT:    and $a4, $a4, $a2
+; LA64-NEXT:    xor $a4, $a3, $a4
+; LA64-NEXT:    sc.w $a4, $a0, 0
+; LA64-NEXT:    beqz $a4, .LBB33_1
+; LA64-NEXT:  # %bb.2:
+; LA64-NEXT:    srl.w $a0, $a3, $a1
+; LA64-NEXT:    ret
+  %1 = atomicrmw xchg ptr %a, i8 0 monotonic
+  ret i8 %1
+}
+
+define i8 @atomicrmw_xchg_minus_1_i8_monotonic(ptr %a) nounwind {
+; LA32-LABEL: atomicrmw_xchg_minus_1_i8_monotonic:
+; LA32:       # %bb.0:
+; LA32-NEXT:    slli.w $a1, $a0, 3
+; LA32-NEXT:    ori $a2, $zero, 255
+; LA32-NEXT:    sll.w $a2, $a2, $a1
+; LA32-NEXT:    bstrins.w $a0, $zero, 1, 0
+; LA32-NEXT:  .LBB34_1: # =>This Inner Loop Header: Depth=1
+; LA32-NEXT:    ll.w $a3, $a0, 0
+; LA32-NEXT:    addi.w $a4, $a2, 0
+; LA32-NEXT:    xor $a4, $a3, $a4
+; LA32-NEXT:    and $a4, $a4, $a2
+; LA32-NEXT:    xor $a4, $a3, $a4
+; LA32-NEXT:    sc.w $a4, $a0, 0
+; LA32-NEXT:    beqz $a4, .LBB34_1
+; LA32-NEXT:  # %bb.2:
+; LA32-NEXT:    srl.w $a0, $a3, $a1
+; LA32-NEXT:    ret
+;
+; LA64-LABEL: atomicrmw_xchg_minus_1_i8_monotonic:
+; LA64:       # %bb.0:
+; LA64-NEXT:    slli.d $a1, $a0, 3
+; LA64-NEXT:    ori $a2, $zero, 255
+; LA64-NEXT:    sll.w $a2, $a2, $a1
+; LA64-NEXT:    addi.w $a2, $a2, 0
+; LA64-NEXT:    bstrins.d $a0, $zero, 1, 0
+; LA64-NEXT:  .LBB34_1: # =>This Inner Loop Header: Depth=1
+; LA64-NEXT:    ll.w $a3, $a0, 0
+; LA64-NEXT:    addi.w $a4, $a2, 0
+; LA64-NEXT:    xor $a4, $a3, $a4
+; LA64-NEXT:    and $a4, $a4, $a2
+; LA64-NEXT:    xor $a4, $a3, $a4
+; LA64-NEXT:    sc.w $a4, $a0, 0
+; LA64-NEXT:    beqz $a4, .LBB34_1
+; LA64-NEXT:  # %bb.2:
+; LA64-NEXT:    srl.w $a0, $a3, $a1
+; LA64-NEXT:    ret
+  %1 = atomicrmw xchg ptr %a, i8 -1 monotonic
+  ret i8 %1
+}
+
 define i16 @atomicrmw_xchg_i16_monotonic(ptr %a, i16 %b) nounwind {
 ; LA32-LABEL: atomicrmw_xchg_i16_monotonic:
 ; LA32:       # %bb.0:
@@ -959,14 +1217,14 @@
 ; LA32-NEXT:    bstrpick.w $a1, $a1, 15, 0
 ; LA32-NEXT:    sll.w $a1, $a1, $a3
 ; LA32-NEXT:    bstrins.w $a0, $zero, 1, 0
-; LA32-NEXT:  .LBB29_1: # =>This Inner Loop Header: Depth=1
+; LA32-NEXT:  .LBB35_1: # =>This Inner Loop Header: Depth=1
 ; LA32-NEXT:    ll.w $a4, $a0, 0
 ; LA32-NEXT:    addi.w $a5, $a1, 0
 ; LA32-NEXT:    xor $a5, $a4, $a5
 ; LA32-NEXT:    and $a5, $a5, $a2
 ; LA32-NEXT:    xor $a5, $a4, $a5
 ; LA32-NEXT:    sc.w $a5, $a0, 0
-; LA32-NEXT:    beqz $a5, .LBB29_1
+; LA32-NEXT:    beqz $a5, .LBB35_1
 ; LA32-NEXT:  # %bb.2:
 ; LA32-NEXT:    srl.w $a0, $a4, $a3
 ; LA32-NEXT:    ret
@@ -982,14 +1240,14 @@
 ; LA64-NEXT:    sll.w $a1, $a1, $a3
 ; LA64-NEXT:    addi.w $a1, $a1, 0
 ; LA64-NEXT:    bstrins.d $a0, $zero, 1, 0
-; LA64-NEXT:  .LBB29_1: # =>This Inner Loop Header: Depth=1
+; LA64-NEXT:  .LBB35_1: # =>This Inner Loop Header: Depth=1
 ; LA64-NEXT:    ll.w $a4, $a0, 0
 ; LA64-NEXT:    addi.w $a5, $a1, 0
 ; LA64-NEXT:    xor $a5, $a4, $a5
 ; LA64-NEXT:    and $a5, $a5, $a2
 ; LA64-NEXT:    xor $a5, $a4, $a5
 ; LA64-NEXT:    sc.w $a5, $a0, 0
-; LA64-NEXT:    beqz $a5, .LBB29_1
+; LA64-NEXT:    beqz $a5, .LBB35_1
 ; LA64-NEXT:  # %bb.2:
 ; LA64-NEXT:    srl.w $a0, $a4, $a3
 ; LA64-NEXT:    ret
@@ -997,14 +1255,100 @@
   ret i16 %1
 }
 
+define i16 @atomicrmw_xchg_0_i16_monotonic(ptr %a) nounwind {
+; LA32-LABEL: atomicrmw_xchg_0_i16_monotonic:
+; LA32:       # %bb.0:
+; LA32-NEXT:    lu12i.w $a1, 15
+; LA32-NEXT:    ori $a1, $a1, 4095
+; LA32-NEXT:    slli.w $a2, $a0, 3
+; LA32-NEXT:    sll.w $a1, $a1, $a2
+; LA32-NEXT:    bstrins.w $a0, $zero, 1, 0
+; LA32-NEXT:  .LBB36_1: # =>This Inner Loop Header: Depth=1
+; LA32-NEXT:    ll.w $a3, $a0, 0
+; LA32-NEXT:    addi.w $a4, $zero, 0
+; LA32-NEXT:    xor $a4, $a3, $a4
+; LA32-NEXT:    and $a4, $a4, $a1
+; LA32-NEXT:    xor $a4, $a3, $a4
+; LA32-NEXT:    sc.w $a4, $a0, 0
+; LA32-NEXT:    beqz $a4, .LBB36_1
+; LA32-NEXT:  # %bb.2:
+; LA32-NEXT:    srl.w $a0, $a3, $a2
+; LA32-NEXT:    ret
+;
+; LA64-LABEL: atomicrmw_xchg_0_i16_monotonic:
+; LA64:       # %bb.0:
+; LA64-NEXT:    lu12i.w $a1, 15
+; LA64-NEXT:    ori $a1, $a1, 4095
+; LA64-NEXT:    slli.d $a2, $a0, 3
+; LA64-NEXT:    sll.w $a1, $a1, $a2
+; LA64-NEXT:    addi.w $a1, $a1, 0
+; LA64-NEXT:    bstrins.d $a0, $zero, 1, 0
+; LA64-NEXT:  .LBB36_1: # =>This Inner Loop Header: Depth=1
+; LA64-NEXT:    ll.w $a3, $a0, 0
+; LA64-NEXT:    addi.w $a4, $zero, 0
+; LA64-NEXT:    xor $a4, $a3, $a4
+; LA64-NEXT:    and $a4, $a4, $a1
+; LA64-NEXT:    xor $a4, $a3, $a4
+; LA64-NEXT:    sc.w $a4, $a0, 0
+; LA64-NEXT:    beqz $a4, .LBB36_1
+; LA64-NEXT:  # %bb.2:
+; LA64-NEXT:    srl.w $a0, $a3, $a2
+; LA64-NEXT:    ret
+  %1 = atomicrmw xchg ptr %a, i16 0 monotonic
+  ret i16 %1
+}
+
+define i16 @atomicrmw_xchg_minus_1_i16_monotonic(ptr %a) nounwind {
+; LA32-LABEL: atomicrmw_xchg_minus_1_i16_monotonic:
+; LA32:       # %bb.0:
+; LA32-NEXT:    lu12i.w $a1, 15
+; LA32-NEXT:    ori $a1, $a1, 4095
+; LA32-NEXT:    slli.w $a2, $a0, 3
+; LA32-NEXT:    sll.w $a1, $a1, $a2
+; LA32-NEXT:    bstrins.w $a0, $zero, 1, 0
+; LA32-NEXT:  .LBB37_1: # =>This Inner Loop Header: Depth=1
+; LA32-NEXT:    ll.w $a3, $a0, 0
+; LA32-NEXT:    addi.w $a4, $a1, 0
+; LA32-NEXT:    xor $a4, $a3, $a4
+; LA32-NEXT:    and $a4, $a4, $a1
+; LA32-NEXT:    xor $a4, $a3, $a4
+; LA32-NEXT:    sc.w $a4, $a0, 0
+; LA32-NEXT:    beqz $a4, .LBB37_1
+; LA32-NEXT:  # %bb.2:
+; LA32-NEXT:    srl.w $a0, $a3, $a2
+; LA32-NEXT:    ret
+;
+; LA64-LABEL: atomicrmw_xchg_minus_1_i16_monotonic:
+; LA64:       # %bb.0:
+; LA64-NEXT:    lu12i.w $a1, 15
+; LA64-NEXT:    ori $a1, $a1, 4095
+; LA64-NEXT:    slli.d $a2, $a0, 3
+; LA64-NEXT:    sll.w $a1, $a1, $a2
+; LA64-NEXT:    addi.w $a1, $a1, 0
+; LA64-NEXT:    bstrins.d $a0, $zero, 1, 0
+; LA64-NEXT:  .LBB37_1: # =>This Inner Loop Header: Depth=1
+; LA64-NEXT:    ll.w $a3, $a0, 0
+; LA64-NEXT:    addi.w $a4, $a1, 0
+; LA64-NEXT:    xor $a4, $a3, $a4
+; LA64-NEXT:    and $a4, $a4, $a1
+; LA64-NEXT:    xor $a4, $a3, $a4
+; LA64-NEXT:    sc.w $a4, $a0, 0
+; LA64-NEXT:    beqz $a4, .LBB37_1
+; LA64-NEXT:  # %bb.2:
+; LA64-NEXT:    srl.w $a0, $a3, $a2
+; LA64-NEXT:    ret
+  %1 = atomicrmw xchg ptr %a, i16 -1 monotonic
+  ret i16 %1
+}
+
 define i32 @atomicrmw_xchg_i32_monotonic(ptr %a, i32 %b) nounwind {
 ; LA32-LABEL: atomicrmw_xchg_i32_monotonic:
 ; LA32:       # %bb.0:
-; LA32-NEXT:  .LBB30_1: # =>This Inner Loop Header: Depth=1
+; LA32-NEXT:  .LBB38_1: # =>This Inner Loop Header: Depth=1
 ; LA32-NEXT:    ll.w $a2, $a0, 0
 ; LA32-NEXT:    move $a3, $a1
 ; LA32-NEXT:    sc.w $a3, $a0, 0
-; LA32-NEXT:    beqz $a3, .LBB30_1
+; LA32-NEXT:    beqz $a3, .LBB38_1
 ; LA32-NEXT:  # %bb.2:
 ; LA32-NEXT:    move $a0, $a2
 ; LA32-NEXT:    ret
@@ -1047,14 +1391,14 @@
 ; LA32-NEXT:    andi $a1, $a1, 255
 ; LA32-NEXT:    sll.w $a1, $a1, $a2
 ; LA32-NEXT:    bstrins.w $a0, $zero, 1, 0
-; LA32-NEXT:  .LBB32_1: # =>This Inner Loop Header: Depth=1
+; LA32-NEXT:  .LBB40_1: # =>This Inner Loop Header: Depth=1
 ; LA32-NEXT:    ll.w $a4, $a0, 0
 ; LA32-NEXT:    add.w $a5, $a4, $a1
 ; LA32-NEXT:    xor $a5, $a4, $a5
 ; LA32-NEXT:    and $a5, $a5, $a3
 ; LA32-NEXT:    xor $a5, $a4, $a5
 ; LA32-NEXT:    sc.w $a5, $a0, 0
-; LA32-NEXT:    beqz $a5, .LBB32_1
+; LA32-NEXT:    beqz $a5, .LBB40_1
 ; LA32-NEXT:  # %bb.2:
 ; LA32-NEXT:    srl.w $a0, $a4, $a2
 ; LA32-NEXT:    ret
@@ -1069,14 +1413,14 @@
 ; LA64-NEXT:    sll.w $a1, $a1, $a2
 ; LA64-NEXT:    addi.w $a1, $a1, 0
 ; LA64-NEXT:    bstrins.d $a0, $zero, 1, 0
-; LA64-NEXT:  .LBB32_1: # =>This Inner Loop Header: Depth=1
+; LA64-NEXT:  .LBB40_1: # =>This Inner Loop Header: Depth=1
 ; LA64-NEXT:    ll.w $a4, $a0, 0
 ; LA64-NEXT:    add.w $a5, $a4, $a1
 ; LA64-NEXT:    xor $a5, $a4, $a5
 ; LA64-NEXT:    and $a5, $a5, $a3
 ; LA64-NEXT:    xor $a5, $a4, $a5
 ; LA64-NEXT:    sc.w $a5, $a0, 0
-; LA64-NEXT:    beqz $a5, .LBB32_1
+; LA64-NEXT:    beqz $a5, .LBB40_1
 ; LA64-NEXT:  # %bb.2:
 ; LA64-NEXT:    srl.w $a0, $a4, $a2
 ; LA64-NEXT:    ret
@@ -1094,14 +1438,14 @@
 ; LA32-NEXT:    bstrpick.w $a1, $a1, 15, 0
 ; LA32-NEXT:    sll.w $a1, $a1, $a3
 ; LA32-NEXT:    bstrins.w $a0, $zero, 1, 0
-; LA32-NEXT:  .LBB33_1: # =>This Inner Loop Header: Depth=1
+; LA32-NEXT:  .LBB41_1: # =>This Inner Loop Header: Depth=1
 ; LA32-NEXT:    ll.w $a4, $a0, 0
 ; LA32-NEXT:    add.w $a5, $a4, $a1
 ; LA32-NEXT:    xor $a5, $a4, $a5
 ; LA32-NEXT:    and $a5, $a5, $a2
 ; LA32-NEXT:    xor $a5, $a4, $a5
 ; LA32-NEXT:    sc.w $a5, $a0, 0
-; LA32-NEXT:    beqz $a5, .LBB33_1
+; LA32-NEXT:    beqz $a5, .LBB41_1
 ; LA32-NEXT:  # %bb.2:
 ; LA32-NEXT:    srl.w $a0, $a4, $a3
 ; LA32-NEXT:    ret
@@ -1117,14 +1461,14 @@
 ; LA64-NEXT:    sll.w $a1, $a1, $a3
 ; LA64-NEXT:    addi.w $a1, $a1, 0
 ; LA64-NEXT:    bstrins.d $a0, $zero, 1, 0
-; LA64-NEXT:  .LBB33_1: # =>This Inner Loop Header: Depth=1
+; LA64-NEXT:  .LBB41_1: # =>This Inner Loop Header: Depth=1
 ; LA64-NEXT:    ll.w $a4, $a0, 0
 ; LA64-NEXT:    add.w $a5, $a4, $a1
 ; LA64-NEXT:    xor $a5, $a4, $a5
 ; LA64-NEXT:    and $a5, $a5, $a2
 ; LA64-NEXT:    xor $a5, $a4, $a5
 ; LA64-NEXT:    sc.w $a5, $a0, 0
-; LA64-NEXT:    beqz $a5, .LBB33_1
+; LA64-NEXT:    beqz $a5, .LBB41_1
 ; LA64-NEXT:  # %bb.2:
 ; LA64-NEXT:    srl.w $a0, $a4, $a3
 ; LA64-NEXT:    ret
@@ -1135,11 +1479,11 @@
 define i32 @atomicrmw_add_i32_monotonic(ptr %a, i32 %b) nounwind {
 ; LA32-LABEL: atomicrmw_add_i32_monotonic:
 ; LA32:       # %bb.0:
-; LA32-NEXT:  .LBB34_1: # =>This Inner Loop Header: Depth=1
+; LA32-NEXT:  .LBB42_1: # =>This Inner Loop Header: Depth=1
 ; LA32-NEXT:    ll.w $a2, $a0, 0
 ; LA32-NEXT:    add.w $a3, $a2, $a1
 ; LA32-NEXT:    sc.w $a3, $a0, 0
-; LA32-NEXT:    beqz $a3, .LBB34_1
+; LA32-NEXT:    beqz $a3, .LBB42_1
 ; LA32-NEXT:  # %bb.2:
 ; LA32-NEXT:    move $a0, $a2
 ; LA32-NEXT:    ret
@@ -1182,14 +1526,14 @@
 ; LA32-NEXT:    andi $a1, $a1, 255
 ; LA32-NEXT:    sll.w $a1, $a1, $a2
 ; LA32-NEXT:    bstrins.w $a0, $zero, 1, 0
-; LA32-NEXT:  .LBB36_1: # =>This Inner Loop Header: Depth=1
+; LA32-NEXT:  .LBB44_1: # =>This Inner Loop Header: Depth=1
 ; LA32-NEXT:    ll.w $a4, $a0, 0
 ; LA32-NEXT:    sub.w $a5, $a4, $a1
 ; LA32-NEXT:    xor $a5, $a4, $a5
 ; LA32-NEXT:    and $a5, $a5, $a3
 ; LA32-NEXT:    xor $a5, $a4, $a5
 ; LA32-NEXT:    sc.w $a5, $a0, 0
-; LA32-NEXT:    beqz $a5, .LBB36_1
+; LA32-NEXT:    beqz $a5, .LBB44_1
 ; LA32-NEXT:  # %bb.2:
 ; LA32-NEXT:    srl.w $a0, $a4, $a2
 ; LA32-NEXT:    ret
@@ -1204,14 +1548,14 @@
 ; LA64-NEXT:    sll.w $a1, $a1, $a2
 ; LA64-NEXT:    addi.w $a1, $a1, 0
 ; LA64-NEXT:    bstrins.d $a0, $zero, 1, 0
-; LA64-NEXT:  .LBB36_1: # =>This Inner Loop Header: Depth=1
+; LA64-NEXT:  .LBB44_1: # =>This Inner Loop Header: Depth=1
 ; LA64-NEXT:    ll.w $a4, $a0, 0
 ; LA64-NEXT:    sub.w $a5, $a4, $a1
 ; LA64-NEXT:    xor $a5, $a4, $a5
 ; LA64-NEXT:    and $a5, $a5, $a3
 ; LA64-NEXT:    xor $a5, $a4, $a5
 ; LA64-NEXT:    sc.w $a5, $a0, 0
-; LA64-NEXT:    beqz $a5, .LBB36_1
+; LA64-NEXT:    beqz $a5, .LBB44_1
 ; LA64-NEXT:  # %bb.2:
 ; LA64-NEXT:    srl.w $a0, $a4, $a2
 ; LA64-NEXT:    ret
@@ -1229,14 +1573,14 @@
 ; LA32-NEXT:    bstrpick.w $a1, $a1, 15, 0
 ; LA32-NEXT:    sll.w $a1, $a1, $a3
 ; LA32-NEXT:    bstrins.w $a0, $zero, 1, 0
-; LA32-NEXT:  .LBB37_1: # =>This Inner Loop Header: Depth=1
+; LA32-NEXT:  .LBB45_1: # =>This Inner Loop Header: Depth=1
 ; LA32-NEXT:    ll.w $a4, $a0, 0
 ; LA32-NEXT:    sub.w $a5, $a4, $a1
 ; LA32-NEXT:    xor $a5, $a4, $a5
 ; LA32-NEXT:    and $a5, $a5, $a2
 ; LA32-NEXT:    xor $a5, $a4, $a5
 ; LA32-NEXT:    sc.w $a5, $a0, 0
-; LA32-NEXT:    beqz $a5, .LBB37_1
+; LA32-NEXT:    beqz $a5, .LBB45_1
 ; LA32-NEXT:  # %bb.2:
 ; LA32-NEXT:    srl.w $a0, $a4, $a3
 ; LA32-NEXT:    ret
@@ -1252,14 +1596,14 @@
 ; LA64-NEXT:    sll.w $a1, $a1, $a3
 ; LA64-NEXT:    addi.w $a1, $a1, 0
 ; LA64-NEXT:    bstrins.d $a0, $zero, 1, 0
-; LA64-NEXT:  .LBB37_1: # =>This Inner Loop Header: Depth=1
+; LA64-NEXT:  .LBB45_1: # =>This Inner Loop Header: Depth=1
 ; LA64-NEXT:    ll.w $a4, $a0, 0
 ; LA64-NEXT:    sub.w $a5, $a4, $a1
 ; LA64-NEXT:    xor $a5, $a4, $a5
 ; LA64-NEXT:    and $a5, $a5, $a2
 ; LA64-NEXT:    xor $a5, $a4, $a5
 ; LA64-NEXT:    sc.w $a5, $a0, 0
-; LA64-NEXT:    beqz $a5, .LBB37_1
+; LA64-NEXT:    beqz $a5, .LBB45_1
 ; LA64-NEXT:  # %bb.2:
 ; LA64-NEXT:    srl.w $a0, $a4, $a3
 ; LA64-NEXT:    ret
@@ -1270,11 +1614,11 @@
 define i32 @atomicrmw_sub_i32_monotonic(ptr %a, i32 %b) nounwind {
 ; LA32-LABEL: atomicrmw_sub_i32_monotonic:
 ; LA32:       # %bb.0:
-; LA32-NEXT:  .LBB38_1: # =>This Inner Loop Header: Depth=1
+; LA32-NEXT:  .LBB46_1: # =>This Inner Loop Header: Depth=1
 ; LA32-NEXT:    ll.w $a2, $a0, 0
 ; LA32-NEXT:    sub.w $a3, $a2, $a1
 ; LA32-NEXT:    sc.w $a3, $a0, 0
-; LA32-NEXT:    beqz $a3, .LBB38_1
+; LA32-NEXT:    beqz $a3, .LBB46_1
 ; LA32-NEXT:  # %bb.2:
 ; LA32-NEXT:    move $a0, $a2
 ; LA32-NEXT:    ret
@@ -1319,7 +1663,7 @@
 ; LA32-NEXT:    andi $a1, $a1, 255
 ; LA32-NEXT:    sll.w $a1, $a1, $a2
 ; LA32-NEXT:    bstrins.w $a0, $zero, 1, 0
-; LA32-NEXT:  .LBB40_1: # =>This Inner Loop Header: Depth=1
+; LA32-NEXT:  .LBB48_1: # =>This Inner Loop Header: Depth=1
 ; LA32-NEXT:    ll.w $a4, $a0, 0
 ; LA32-NEXT:    and $a5, $a4, $a1
 ; LA32-NEXT:    nor $a5, $a5, $zero
@@ -1327,7 +1671,7 @@
 ; LA32-NEXT:    and $a5, $a5, $a3
 ; LA32-NEXT:    xor $a5, $a4, $a5
 ; LA32-NEXT:    sc.w $a5, $a0, 0
-; LA32-NEXT:    beqz $a5, .LBB40_1
+; LA32-NEXT:    beqz $a5, .LBB48_1
 ; LA32-NEXT:  # %bb.2:
 ; LA32-NEXT:    srl.w $a0, $a4, $a2
 ; LA32-NEXT:    ret
@@ -1342,7 +1686,7 @@
 ; LA64-NEXT:    sll.w $a1, $a1, $a2
 ; LA64-NEXT:    addi.w $a1, $a1, 0
 ; LA64-NEXT:    bstrins.d $a0, $zero, 1, 0
-; LA64-NEXT:  .LBB40_1: # =>This Inner Loop Header: Depth=1
+; LA64-NEXT:  .LBB48_1: # =>This Inner Loop Header: Depth=1
 ; LA64-NEXT:    ll.w $a4, $a0, 0
 ; LA64-NEXT:    and $a5, $a4, $a1
 ; LA64-NEXT:    nor $a5, $a5, $zero
@@ -1350,7 +1694,7 @@
 ; LA64-NEXT:    and $a5, $a5, $a3
 ; LA64-NEXT:    xor $a5, $a4, $a5
 ; LA64-NEXT:    sc.w $a5, $a0, 0
-; LA64-NEXT:    beqz $a5, .LBB40_1
+; LA64-NEXT:    beqz $a5, .LBB48_1
 ; LA64-NEXT:  # %bb.2:
 ; LA64-NEXT:    srl.w $a0, $a4, $a2
 ; LA64-NEXT:    ret
@@ -1368,7 +1712,7 @@
 ; LA32-NEXT:    bstrpick.w $a1, $a1, 15, 0
 ; LA32-NEXT:    sll.w $a1, $a1, $a3
 ; LA32-NEXT:    bstrins.w $a0, $zero, 1, 0
-; LA32-NEXT:  .LBB41_1: # =>This Inner Loop Header: Depth=1
+; LA32-NEXT:  .LBB49_1: # =>This Inner Loop Header: Depth=1
 ; LA32-NEXT:    ll.w $a4, $a0, 0
 ; LA32-NEXT:    and $a5, $a4, $a1
 ; LA32-NEXT:    nor $a5, $a5, $zero
@@ -1376,7 +1720,7 @@
 ; LA32-NEXT:    and $a5, $a5, $a2
 ; LA32-NEXT:    xor $a5, $a4, $a5
 ; LA32-NEXT:    sc.w $a5, $a0, 0
-; LA32-NEXT:    beqz $a5, .LBB41_1
+; LA32-NEXT:    beqz $a5, .LBB49_1
 ; LA32-NEXT:  # %bb.2:
 ; LA32-NEXT:    srl.w $a0, $a4, $a3
 ; LA32-NEXT:    ret
@@ -1392,7 +1736,7 @@
 ; LA64-NEXT:    sll.w $a1, $a1, $a3
 ; LA64-NEXT:    addi.w $a1, $a1, 0
 ; LA64-NEXT:    bstrins.d $a0, $zero, 1, 0
-; LA64-NEXT:  .LBB41_1: # =>This Inner Loop Header: Depth=1
+; LA64-NEXT:  .LBB49_1: # =>This Inner Loop Header: Depth=1
 ; LA64-NEXT:    ll.w $a4, $a0, 0
 ; LA64-NEXT:    and $a5, $a4, $a1
 ; LA64-NEXT:    nor $a5, $a5, $zero
@@ -1400,7 +1744,7 @@
 ; LA64-NEXT:    and $a5, $a5, $a2
 ; LA64-NEXT:    xor $a5, $a4, $a5
 ; LA64-NEXT:    sc.w $a5, $a0, 0
-; LA64-NEXT:    beqz $a5, .LBB41_1
+; LA64-NEXT:    beqz $a5, .LBB49_1
 ; LA64-NEXT:  # %bb.2:
 ; LA64-NEXT:    srl.w $a0, $a4, $a3
 ; LA64-NEXT:    ret
@@ -1411,24 +1755,24 @@
 define i32 @atomicrmw_nand_i32_monotonic(ptr %a, i32 %b) nounwind {
 ; LA32-LABEL: atomicrmw_nand_i32_monotonic:
 ; LA32:       # %bb.0:
-; LA32-NEXT:  .LBB42_1: # =>This Inner Loop Header: Depth=1
+; LA32-NEXT:  .LBB50_1: # =>This Inner Loop Header: Depth=1
 ; LA32-NEXT:    ll.w $a2, $a0, 0
 ; LA32-NEXT:    and $a3, $a2, $a1
 ; LA32-NEXT:    nor $a3, $a3, $zero
 ; LA32-NEXT:    sc.w $a3, $a0, 0
-; LA32-NEXT:    beqz $a3, .LBB42_1
+; LA32-NEXT:    beqz $a3, .LBB50_1
 ; LA32-NEXT:  # %bb.2:
 ; LA32-NEXT:    move $a0, $a2
 ; LA32-NEXT:    ret
 ;
 ; LA64-LABEL: atomicrmw_nand_i32_monotonic:
 ; LA64:       # %bb.0:
-; LA64-NEXT:  .LBB42_1: # =>This Inner Loop Header: Depth=1
+; LA64-NEXT:  .LBB50_1: # =>This Inner Loop Header: Depth=1
 ; LA64-NEXT:    ll.w $a2, $a0, 0
 ; LA64-NEXT:    and $a3, $a2, $a1
 ; LA64-NEXT:    nor $a3, $a3, $zero
 ; LA64-NEXT:    sc.w $a3, $a0, 0
-; LA64-NEXT:    beqz $a3, .LBB42_1
+; LA64-NEXT:    beqz $a3, .LBB50_1
 ; LA64-NEXT:  # %bb.2:
 ; LA64-NEXT:    move $a0, $a2
 ; LA64-NEXT:    ret
@@ -1449,12 +1793,12 @@
 ;
 ; LA64-LABEL: atomicrmw_nand_i64_monotonic:
 ; LA64:       # %bb.0:
-; LA64-NEXT:  .LBB43_1: # =>This Inner Loop Header: Depth=1
+; LA64-NEXT:  .LBB51_1: # =>This Inner Loop Header: Depth=1
 ; LA64-NEXT:    ll.d $a2, $a0, 0
 ; LA64-NEXT:    and $a3, $a2, $a1
 ; LA64-NEXT:    nor $a3, $a3, $zero
 ; LA64-NEXT:    sc.d $a3, $a0, 0
-; LA64-NEXT:    beqz $a3, .LBB43_1
+; LA64-NEXT:    beqz $a3, .LBB51_1
 ; LA64-NEXT:  # %bb.2:
 ; LA64-NEXT:    move $a0, $a2
 ; LA64-NEXT:    ret
@@ -1472,11 +1816,11 @@
 ; LA32-NEXT:    sll.w $a1, $a1, $a2
 ; LA32-NEXT:    orn $a1, $a1, $a3
 ; LA32-NEXT:    bstrins.w $a0, $zero, 1, 0
-; LA32-NEXT:  .LBB44_1: # =>This Inner Loop Header: Depth=1
+; LA32-NEXT:  .LBB52_1: # =>This Inner Loop Header: Depth=1
 ; LA32-NEXT:    ll.w $a3, $a0, 0
 ; LA32-NEXT:    and $a4, $a3, $a1
 ; LA32-NEXT:    sc.w $a4, $a0, 0
-; LA32-NEXT:    beqz $a4, .LBB44_1
+; LA32-NEXT:    beqz $a4, .LBB52_1
 ; LA32-NEXT:  # %bb.2:
 ; LA32-NEXT:    srl.w $a0, $a3, $a2
 ; LA32-NEXT:    ret
@@ -1508,11 +1852,11 @@
 ; LA32-NEXT:    sll.w $a1, $a1, $a3
 ; LA32-NEXT:    orn $a1, $a1, $a2
 ; LA32-NEXT:    bstrins.w $a0, $zero, 1, 0
-; LA32-NEXT:  .LBB45_1: # =>This Inner Loop Header: Depth=1
+; LA32-NEXT:  .LBB53_1: # =>This Inner Loop Header: Depth=1
 ; LA32-NEXT:    ll.w $a2, $a0, 0
 ; LA32-NEXT:    and $a4, $a2, $a1
 ; LA32-NEXT:    sc.w $a4, $a0, 0
-; LA32-NEXT:    beqz $a4, .LBB45_1
+; LA32-NEXT:    beqz $a4, .LBB53_1
 ; LA32-NEXT:  # %bb.2:
 ; LA32-NEXT:    srl.w $a0, $a2, $a3
 ; LA32-NEXT:    ret
@@ -1537,11 +1881,11 @@
 define i32 @atomicrmw_and_i32_monotonic(ptr %a, i32 %b) nounwind {
 ; LA32-LABEL: atomicrmw_and_i32_monotonic:
 ; LA32:       # %bb.0:
-; LA32-NEXT:  .LBB46_1: # =>This Inner Loop Header: Depth=1
+; LA32-NEXT:  .LBB54_1: # =>This Inner Loop Header: Depth=1
 ; LA32-NEXT:    ll.w $a2, $a0, 0
 ; LA32-NEXT:    and $a3, $a2, $a1
 ; LA32-NEXT:    sc.w $a3, $a0, 0
-; LA32-NEXT:    beqz $a3, .LBB46_1
+; LA32-NEXT:    beqz $a3, .LBB54_1
 ; LA32-NEXT:  # %bb.2:
 ; LA32-NEXT:    move $a0, $a2
 ; LA32-NEXT:    ret
@@ -1582,11 +1926,11 @@
 ; LA32-NEXT:    slli.w $a2, $a0, 3
 ; LA32-NEXT:    sll.w $a1, $a1, $a2
 ; LA32-NEXT:    bstrins.w $a0, $zero, 1, 0
-; LA32-NEXT:  .LBB48_1: # =>This Inner Loop Header: Depth=1
+; LA32-NEXT:  .LBB56_1: # =>This Inner Loop Header: Depth=1
 ; LA32-NEXT:    ll.w $a3, $a0, 0
 ; LA32-NEXT:    or $a4, $a3, $a1
 ; LA32-NEXT:    sc.w $a4, $a0, 0
-; LA32-NEXT:    beqz $a4, .LBB48_1
+; LA32-NEXT:    beqz $a4, .LBB56_1
 ; LA32-NEXT:  # %bb.2:
 ; LA32-NEXT:    srl.w $a0, $a3, $a2
 ; LA32-NEXT:    ret
@@ -1611,11 +1955,11 @@
 ; LA32-NEXT:    slli.w $a2, $a0, 3
 ; LA32-NEXT:    sll.w $a1, $a1, $a2
 ; LA32-NEXT:    bstrins.w $a0, $zero, 1, 0
-; LA32-NEXT:  .LBB49_1: # =>This Inner Loop Header: Depth=1
+; LA32-NEXT:  .LBB57_1: # =>This Inner Loop Header: Depth=1
 ; LA32-NEXT:    ll.w $a3, $a0, 0
 ; LA32-NEXT:    or $a4, $a3, $a1
 ; LA32-NEXT:    sc.w $a4, $a0, 0
-; LA32-NEXT:    beqz $a4, .LBB49_1
+; LA32-NEXT:    beqz $a4, .LBB57_1
 ; LA32-NEXT:  # %bb.2:
 ; LA32-NEXT:    srl.w $a0, $a3, $a2
 ; LA32-NEXT:    ret
@@ -1636,11 +1980,11 @@
 define i32 @atomicrmw_or_i32_monotonic(ptr %a, i32 %b) nounwind {
 ; LA32-LABEL: atomicrmw_or_i32_monotonic:
 ; LA32:       # %bb.0:
-; LA32-NEXT:  .LBB50_1: # =>This Inner Loop Header: Depth=1
+; LA32-NEXT:  .LBB58_1: # =>This Inner Loop Header: Depth=1
 ; LA32-NEXT:    ll.w $a2, $a0, 0
 ; LA32-NEXT:    or $a3, $a2, $a1
 ; LA32-NEXT:    sc.w $a3, $a0, 0
-; LA32-NEXT:    beqz $a3, .LBB50_1
+; LA32-NEXT:    beqz $a3, .LBB58_1
 ; LA32-NEXT:  # %bb.2:
 ; LA32-NEXT:    move $a0, $a2
 ; LA32-NEXT:    ret
@@ -1681,11 +2025,11 @@
 ; LA32-NEXT:    slli.w $a2, $a0, 3
 ; LA32-NEXT:    sll.w $a1, $a1, $a2
 ; LA32-NEXT:    bstrins.w $a0, $zero, 1, 0
-; LA32-NEXT:  .LBB52_1: # =>This Inner Loop Header: Depth=1
+; LA32-NEXT:  .LBB60_1: # =>This Inner Loop Header: Depth=1
 ; LA32-NEXT:    ll.w $a3, $a0, 0
 ; LA32-NEXT:    xor $a4, $a3, $a1
 ; LA32-NEXT:    sc.w $a4, $a0, 0
-; LA32-NEXT:    beqz $a4, .LBB52_1
+; LA32-NEXT:    beqz $a4, .LBB60_1
 ; LA32-NEXT:  # %bb.2:
 ; LA32-NEXT:    srl.w $a0, $a3, $a2
 ; LA32-NEXT:    ret
@@ -1710,11 +2054,11 @@
 ; LA32-NEXT:    slli.w $a2, $a0, 3
 ; LA32-NEXT:    sll.w $a1, $a1, $a2
 ; LA32-NEXT:    bstrins.w $a0, $zero, 1, 0
-; LA32-NEXT:  .LBB53_1: # =>This Inner Loop Header: Depth=1
+; LA32-NEXT:  .LBB61_1: # =>This Inner Loop Header: Depth=1
 ; LA32-NEXT:    ll.w $a3, $a0, 0
 ; LA32-NEXT:    xor $a4, $a3, $a1
 ; LA32-NEXT:    sc.w $a4, $a0, 0
-; LA32-NEXT:    beqz $a4, .LBB53_1
+; LA32-NEXT:    beqz $a4, .LBB61_1
 ; LA32-NEXT:  # %bb.2:
 ; LA32-NEXT:    srl.w $a0, $a3, $a2
 ; LA32-NEXT:    ret
@@ -1735,11 +2079,11 @@
 define i32 @atomicrmw_xor_i32_monotonic(ptr %a, i32 %b) nounwind {
 ; LA32-LABEL: atomicrmw_xor_i32_monotonic:
 ; LA32:       # %bb.0:
-; LA32-NEXT:  .LBB54_1: # =>This Inner Loop Header: Depth=1
+; LA32-NEXT:  .LBB62_1: # =>This Inner Loop Header: Depth=1
 ; LA32-NEXT:    ll.w $a2, $a0, 0
 ; LA32-NEXT:    xor $a3, $a2, $a1
 ; LA32-NEXT:    sc.w $a3, $a0, 0
-; LA32-NEXT:    beqz $a3, .LBB54_1
+; LA32-NEXT:    beqz $a3, .LBB62_1
 ; LA32-NEXT:  # %bb.2:
 ; LA32-NEXT:    move $a0, $a2
 ; LA32-NEXT:    ret