[LoongArch] Optimize (and (a & ~((2^^X - 1) << Y)) to (bstrins a, zero, X+Y-1, Y)

[SixWeining]

Summary

Inspired by D158384.

llvm/lib/Target/LoongArch/LoongArchInstrInfo.td

def AlslSlliImmI0 : SDNodeXForm<imm, [{
   case 5:  I0 = 2; break;
   case 9:  I0 = 3; break;
   default: I0 = 4; break;
   }
   return CurDAG->getTargetConstant(I0, SDLoc(N),
                                    N->getValueType(0));
 }]>;
 
+// Check if (and r, imm) can be optimized to (BSTRINS r, R0, msb, lsb),
+// in which imm = ~((2^^(msb-lsb+1) - 1) << lsb).
+def BstrinsImm : PatLeaf<(imm), [{
+  if (!N->hasOneUse())
+    return false;
+  uint64_t Imm = N->getZExtValue();
+  // andi can be used instead if Imm <= 0xfff.
+  if (Imm <= 0xfff)
+    return false;
+  unsigned MaskIdx, MaskLen;
+  return N->getValueType(0).getSizeInBits() == 32
+             ? llvm::isShiftedMask_32(~Imm, MaskIdx, MaskLen)
+             : llvm::isShiftedMask_64(~Imm, MaskIdx, MaskLen);
+}]>;
+
+def BstrinsMsb: SDNodeXForm<imm, [{
+  uint64_t Imm = N->getZExtValue();
+  unsigned MaskIdx, MaskLen;
+  N->getValueType(0).getSizeInBits() == 32
+      ? llvm::isShiftedMask_32(~Imm, MaskIdx, MaskLen)
+      : llvm::isShiftedMask_64(~Imm, MaskIdx, MaskLen);
+  return CurDAG->getTargetConstant(MaskIdx + MaskLen - 1, SDLoc(N),
+                                   N->getValueType(0));
+}]>;
+
+def BstrinsLsb: SDNodeXForm<imm, [{
+  uint64_t Imm = N->getZExtValue();
+  unsigned MaskIdx, MaskLen;
+  N->getValueType(0).getSizeInBits() == 32
+      ? llvm::isShiftedMask_32(~Imm, MaskIdx, MaskLen)
+      : llvm::isShiftedMask_64(~Imm, MaskIdx, MaskLen);
+  return CurDAG->getTargetConstant(MaskIdx, SDLoc(N), N->getValueType(0));
+}]>;
+
 //===----------------------------------------------------------------------===//
 // Instruction Formats
 //===----------------------------------------------------------------------===//
 
 include "LoongArchInstrFormats.td"
 include "LoongArchFloatInstrFormats.td"
 include "LoongArchLSXInstrFormats.td"
 include "LoongArchLASXInstrFormats.td"

 def : PatGprImm<or, ORI, uimm12>;
 def : PatGprGpr<xor, XOR>;
 def : PatGprImm<xor, XORI, uimm12>;
 def : Pat<(not GPR:$rj), (NOR GPR:$rj, R0)>;
 def : Pat<(not (or GPR:$rj, GPR:$rk)), (NOR GPR:$rj, GPR:$rk)>;
 def : Pat<(or GPR:$rj, (not GPR:$rk)), (ORN GPR:$rj, GPR:$rk)>;
 def : Pat<(and GPR:$rj, (not GPR:$rk)), (ANDN GPR:$rj, GPR:$rk)>;
 
+let Predicates = [IsLA32] in {
+def : Pat<(and GPR:$rj, BstrinsImm:$imm),
+          (BSTRINS_W GPR:$rj, R0, (BstrinsMsb BstrinsImm:$imm),
+                     (BstrinsLsb BstrinsImm:$imm))>;
+} // Predicates = [IsLA32]
+
+let Predicates = [IsLA64] in {
+def : Pat<(and GPR:$rj, BstrinsImm:$imm),
+          (BSTRINS_D GPR:$rj, R0, (BstrinsMsb BstrinsImm:$imm),
+                     (BstrinsLsb BstrinsImm:$imm))>;
+} // Predicates = [IsLA64]
+
 /// Traps
 
 // We lower `trap` to `amswap.w rd:$r0, rk:$r1, rj:$r0`, as this is guaranteed
 // to trap with an INE (non-existent on LA32, explicitly documented to INE on
 // LA64). And the resulting signal is different from `debugtrap` like on some
 // other existing ports so programs/porters might have an easier time.
 def PseudoUNIMP : Pseudo<(outs), (ins), [(trap)]>,
                   PseudoInstExpansion<(AMSWAP_W R0, R1, R0)>;

llvm/test/CodeGen/LoongArch/alloca.ll

 define void @simple_alloca(i32 %n) nounwind {
 ; LA32-LABEL: simple_alloca:
 ; LA32:       # %bb.0:
 ; LA32-NEXT:    addi.w $sp, $sp, -16
 ; LA32-NEXT:    st.w $ra, $sp, 12 # 4-byte Folded Spill
 ; LA32-NEXT:    st.w $fp, $sp, 8 # 4-byte Folded Spill
 ; LA32-NEXT:    addi.w $fp, $sp, 16
 ; LA32-NEXT:    addi.w $a0, $a0, 15
-; LA32-NEXT:    addi.w $a1, $zero, -16
-; LA32-NEXT:    and $a0, $a0, $a1
+; LA32-NEXT:    bstrins.w $a0, $zero, 3, 0
 ; LA32-NEXT:    sub.w $a0, $sp, $a0
 ; LA32-NEXT:    move $sp, $a0
 ; LA32-NEXT:    bl %plt(notdead)
 ; LA32-NEXT:    addi.w $sp, $fp, -16
 ; LA32-NEXT:    ld.w $fp, $sp, 8 # 4-byte Folded Reload
 ; LA32-NEXT:    ld.w $ra, $sp, 12 # 4-byte Folded Reload
 ; LA32-NEXT:    addi.w $sp, $sp, 16
 ; LA32-NEXT:    ret

 define void @scoped_alloca(i32 %n) nounwind {
 ; LA32-LABEL: scoped_alloca:
 ; LA32:       # %bb.0:
 ; LA32-NEXT:    addi.w $sp, $sp, -16
 ; LA32-NEXT:    st.w $ra, $sp, 12 # 4-byte Folded Spill
 ; LA32-NEXT:    st.w $fp, $sp, 8 # 4-byte Folded Spill
 ; LA32-NEXT:    st.w $s0, $sp, 4 # 4-byte Folded Spill
 ; LA32-NEXT:    addi.w $fp, $sp, 16
-; LA32-NEXT:    addi.w $a0, $a0, 15
-; LA32-NEXT:    addi.w $a1, $zero, -16
-; LA32-NEXT:    and $a0, $a0, $a1
 ; LA32-NEXT:    move $s0, $sp
+; LA32-NEXT:    addi.w $a0, $a0, 15
+; LA32-NEXT:    bstrins.w $a0, $zero, 3, 0
 ; LA32-NEXT:    sub.w $a0, $sp, $a0
 ; LA32-NEXT:    move $sp, $a0
 ; LA32-NEXT:    bl %plt(notdead)
 ; LA32-NEXT:    move $sp, $s0
 ; LA32-NEXT:    addi.w $sp, $fp, -16
 ; LA32-NEXT:    ld.w $s0, $sp, 4 # 4-byte Folded Reload
 ; LA32-NEXT:    ld.w $fp, $sp, 8 # 4-byte Folded Reload
 ; LA32-NEXT:    ld.w $ra, $sp, 12 # 4-byte Folded Reload

 define void @alloca_callframe(i32 %n) nounwind {
 ; LA32-LABEL: alloca_callframe:
 ; LA32:       # %bb.0:
 ; LA32-NEXT:    addi.w $sp, $sp, -16
 ; LA32-NEXT:    st.w $ra, $sp, 12 # 4-byte Folded Spill
 ; LA32-NEXT:    st.w $fp, $sp, 8 # 4-byte Folded Spill
 ; LA32-NEXT:    addi.w $fp, $sp, 16
 ; LA32-NEXT:    addi.w $a0, $a0, 15
-; LA32-NEXT:    addi.w $a1, $zero, -16
-; LA32-NEXT:    and $a0, $a0, $a1
+; LA32-NEXT:    bstrins.w $a0, $zero, 3, 0
 ; LA32-NEXT:    sub.w $a0, $sp, $a0
 ; LA32-NEXT:    move $sp, $a0
 ; LA32-NEXT:    addi.w $sp, $sp, -16
 ; LA32-NEXT:    ori $a1, $zero, 12
 ; LA32-NEXT:    st.w $a1, $sp, 12
 ; LA32-NEXT:    ori $a1, $zero, 11
 ; LA32-NEXT:    st.w $a1, $sp, 8
 ; LA32-NEXT:    ori $a1, $zero, 10

llvm/test/CodeGen/LoongArch/atomicrmw-uinc-udec-wrap.ll

 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
 ; RUN: llc --mtriple=loongarch64 < %s | FileCheck --check-prefix=LA64 %s
 
 define i8 @atomicrmw_uinc_wrap_i8(ptr %ptr, i8 %val) {
 ; LA64-LABEL: atomicrmw_uinc_wrap_i8:
 ; LA64:       # %bb.0:
-; LA64-NEXT:    addi.w $a2, $zero, -4
-; LA64-NEXT:    and $a2, $a0, $a2
-; LA64-NEXT:    ld.w $a3, $a2, 0
-; LA64-NEXT:    slli.d $a0, $a0, 3
+; LA64-NEXT:    slli.d $a3, $a0, 3
+; LA64-NEXT:    bstrins.d $a0, $zero, 1, 0
+; LA64-NEXT:    ld.w $a2, $a0, 0
 ; LA64-NEXT:    ori $a4, $zero, 255
-; LA64-NEXT:    sll.w $a4, $a4, $a0
-; LA64-NEXT:    andi $a0, $a0, 24
+; LA64-NEXT:    sll.w $a4, $a4, $a3
+; LA64-NEXT:    andi $a3, $a3, 24
 ; LA64-NEXT:    nor $a4, $a4, $zero
 ; LA64-NEXT:    andi $a1, $a1, 255
 ; LA64-NEXT:    .p2align 4, , 16
 ; LA64-NEXT:  .LBB0_1: # %atomicrmw.start
 ; LA64-NEXT:    # =>This Loop Header: Depth=1
 ; LA64-NEXT:    # Child Loop BB0_3 Depth 2
-; LA64-NEXT:    srl.w $a5, $a3, $a0
+; LA64-NEXT:    srl.w $a5, $a2, $a3
 ; LA64-NEXT:    andi $a6, $a5, 255
 ; LA64-NEXT:    sltu $a6, $a6, $a1
 ; LA64-NEXT:    addi.d $a5, $a5, 1
 ; LA64-NEXT:    xori $a6, $a6, 1
 ; LA64-NEXT:    masknez $a5, $a5, $a6
 ; LA64-NEXT:    andi $a5, $a5, 255
-; LA64-NEXT:    sll.w $a5, $a5, $a0
-; LA64-NEXT:    and $a6, $a3, $a4
+; LA64-NEXT:    sll.w $a5, $a5, $a3
+; LA64-NEXT:    and $a6, $a2, $a4
 ; LA64-NEXT:    or $a6, $a6, $a5
 ; LA64-NEXT:  .LBB0_3: # %atomicrmw.start
 ; LA64-NEXT:    # Parent Loop BB0_1 Depth=1
 ; LA64-NEXT:    # => This Inner Loop Header: Depth=2
-; LA64-NEXT:    ll.w $a5, $a2, 0
-; LA64-NEXT:    bne $a5, $a3, .LBB0_5
+; LA64-NEXT:    ll.w $a5, $a0, 0
+; LA64-NEXT:    bne $a5, $a2, .LBB0_5
 ; LA64-NEXT:  # %bb.4: # %atomicrmw.start
 ; LA64-NEXT:    # in Loop: Header=BB0_3 Depth=2
 ; LA64-NEXT:    dbar 0
 ; LA64-NEXT:    move $a7, $a6
-; LA64-NEXT:    sc.w $a7, $a2, 0
+; LA64-NEXT:    sc.w $a7, $a0, 0
 ; LA64-NEXT:    beqz $a7, .LBB0_3
 ; LA64-NEXT:    b .LBB0_6
 ; LA64-NEXT:  .LBB0_5: # %atomicrmw.start
 ; LA64-NEXT:    # in Loop: Header=BB0_1 Depth=1
 ; LA64-NEXT:    dbar 1792
 ; LA64-NEXT:  .LBB0_6: # %atomicrmw.start
 ; LA64-NEXT:    # in Loop: Header=BB0_1 Depth=1
-; LA64-NEXT:    addi.w $a6, $a3, 0
-; LA64-NEXT:    move $a3, $a5
+; LA64-NEXT:    addi.w $a6, $a2, 0
+; LA64-NEXT:    move $a2, $a5
 ; LA64-NEXT:    bne $a5, $a6, .LBB0_1
 ; LA64-NEXT:  # %bb.2: # %atomicrmw.end
-; LA64-NEXT:    srl.w $a0, $a5, $a0
+; LA64-NEXT:    srl.w $a0, $a5, $a3
 ; LA64-NEXT:    ret
   %result = atomicrmw uinc_wrap ptr %ptr, i8 %val seq_cst
   ret i8 %result
 }
 
 define i16 @atomicrmw_uinc_wrap_i16(ptr %ptr, i16 %val) {
 ; LA64-LABEL: atomicrmw_uinc_wrap_i16:
 ; LA64:       # %bb.0:
-; LA64-NEXT:    addi.w $a2, $zero, -4
-; LA64-NEXT:    and $a2, $a0, $a2
-; LA64-NEXT:    ld.w $a3, $a2, 0
-; LA64-NEXT:    slli.d $a0, $a0, 3
+; LA64-NEXT:    slli.d $a3, $a0, 3
+; LA64-NEXT:    bstrins.d $a0, $zero, 1, 0
+; LA64-NEXT:    ld.w $a2, $a0, 0
 ; LA64-NEXT:    lu12i.w $a4, 15
 ; LA64-NEXT:    ori $a4, $a4, 4095
-; LA64-NEXT:    sll.w $a4, $a4, $a0
-; LA64-NEXT:    andi $a0, $a0, 24
+; LA64-NEXT:    sll.w $a4, $a4, $a3
+; LA64-NEXT:    andi $a3, $a3, 24
 ; LA64-NEXT:    nor $a4, $a4, $zero
 ; LA64-NEXT:    bstrpick.d $a1, $a1, 15, 0
 ; LA64-NEXT:    .p2align 4, , 16
 ; LA64-NEXT:  .LBB1_1: # %atomicrmw.start
 ; LA64-NEXT:    # =>This Loop Header: Depth=1
 ; LA64-NEXT:    # Child Loop BB1_3 Depth 2
-; LA64-NEXT:    srl.w $a5, $a3, $a0
+; LA64-NEXT:    srl.w $a5, $a2, $a3
 ; LA64-NEXT:    bstrpick.d $a6, $a5, 15, 0
 ; LA64-NEXT:    sltu $a6, $a6, $a1
 ; LA64-NEXT:    addi.d $a5, $a5, 1
 ; LA64-NEXT:    xori $a6, $a6, 1
 ; LA64-NEXT:    masknez $a5, $a5, $a6
 ; LA64-NEXT:    bstrpick.d $a5, $a5, 15, 0
-; LA64-NEXT:    sll.w $a5, $a5, $a0
-; LA64-NEXT:    and $a6, $a3, $a4
+; LA64-NEXT:    sll.w $a5, $a5, $a3
+; LA64-NEXT:    and $a6, $a2, $a4
 ; LA64-NEXT:    or $a6, $a6, $a5
 ; LA64-NEXT:  .LBB1_3: # %atomicrmw.start
 ; LA64-NEXT:    # Parent Loop BB1_1 Depth=1
 ; LA64-NEXT:    # => This Inner Loop Header: Depth=2
-; LA64-NEXT:    ll.w $a5, $a2, 0
-; LA64-NEXT:    bne $a5, $a3, .LBB1_5
+; LA64-NEXT:    ll.w $a5, $a0, 0
+; LA64-NEXT:    bne $a5, $a2, .LBB1_5
 ; LA64-NEXT:  # %bb.4: # %atomicrmw.start
 ; LA64-NEXT:    # in Loop: Header=BB1_3 Depth=2
 ; LA64-NEXT:    dbar 0
 ; LA64-NEXT:    move $a7, $a6
-; LA64-NEXT:    sc.w $a7, $a2, 0
+; LA64-NEXT:    sc.w $a7, $a0, 0
 ; LA64-NEXT:    beqz $a7, .LBB1_3
 ; LA64-NEXT:    b .LBB1_6
 ; LA64-NEXT:  .LBB1_5: # %atomicrmw.start
 ; LA64-NEXT:    # in Loop: Header=BB1_1 Depth=1
 ; LA64-NEXT:    dbar 1792
 ; LA64-NEXT:  .LBB1_6: # %atomicrmw.start
 ; LA64-NEXT:    # in Loop: Header=BB1_1 Depth=1
-; LA64-NEXT:    addi.w $a6, $a3, 0
-; LA64-NEXT:    move $a3, $a5
+; LA64-NEXT:    addi.w $a6, $a2, 0
+; LA64-NEXT:    move $a2, $a5
 ; LA64-NEXT:    bne $a5, $a6, .LBB1_1
 ; LA64-NEXT:  # %bb.2: # %atomicrmw.end
-; LA64-NEXT:    srl.w $a0, $a5, $a0
+; LA64-NEXT:    srl.w $a0, $a5, $a3
 ; LA64-NEXT:    ret
   %result = atomicrmw uinc_wrap ptr %ptr, i16 %val seq_cst
   ret i16 %result
 }
 
 define i32 @atomicrmw_uinc_wrap_i32(ptr %ptr, i32 %val) {
 ; LA64-LABEL: atomicrmw_uinc_wrap_i32:
 ; LA64:       # %bb.0:

 ; LA64-NEXT:    ret
   %result = atomicrmw uinc_wrap ptr %ptr, i64 %val seq_cst
   ret i64 %result
 }
 
 define i8 @atomicrmw_udec_wrap_i8(ptr %ptr, i8 %val) {
 ; LA64-LABEL: atomicrmw_udec_wrap_i8:
 ; LA64:       # %bb.0:
-; LA64-NEXT:    addi.w $a2, $zero, -4
-; LA64-NEXT:    and $a2, $a0, $a2
-; LA64-NEXT:    ld.w $a3, $a2, 0
-; LA64-NEXT:    slli.d $a0, $a0, 3
+; LA64-NEXT:    slli.d $a3, $a0, 3
+; LA64-NEXT:    bstrins.d $a0, $zero, 1, 0
+; LA64-NEXT:    ld.w $a2, $a0, 0
 ; LA64-NEXT:    ori $a4, $zero, 255
-; LA64-NEXT:    sll.w $a4, $a4, $a0
-; LA64-NEXT:    andi $a0, $a0, 24
+; LA64-NEXT:    sll.w $a4, $a4, $a3
+; LA64-NEXT:    andi $a3, $a3, 24
 ; LA64-NEXT:    nor $a4, $a4, $zero
 ; LA64-NEXT:    andi $a5, $a1, 255
 ; LA64-NEXT:    .p2align 4, , 16
 ; LA64-NEXT:  .LBB4_1: # %atomicrmw.start
 ; LA64-NEXT:    # =>This Loop Header: Depth=1
 ; LA64-NEXT:    # Child Loop BB4_3 Depth 2
-; LA64-NEXT:    srl.w $a6, $a3, $a0
+; LA64-NEXT:    srl.w $a6, $a2, $a3
 ; LA64-NEXT:    andi $a7, $a6, 255
 ; LA64-NEXT:    sltu $t0, $a5, $a7
 ; LA64-NEXT:    addi.d $a6, $a6, -1
 ; LA64-NEXT:    masknez $a6, $a6, $t0
 ; LA64-NEXT:    maskeqz $t0, $a1, $t0
 ; LA64-NEXT:    or $a6, $t0, $a6
 ; LA64-NEXT:    sltui $a7, $a7, 1
 ; LA64-NEXT:    masknez $a6, $a6, $a7
 ; LA64-NEXT:    maskeqz $a7, $a1, $a7
 ; LA64-NEXT:    or $a6, $a7, $a6
 ; LA64-NEXT:    andi $a6, $a6, 255
-; LA64-NEXT:    sll.w $a6, $a6, $a0
-; LA64-NEXT:    and $a7, $a3, $a4
+; LA64-NEXT:    sll.w $a6, $a6, $a3
+; LA64-NEXT:    and $a7, $a2, $a4
 ; LA64-NEXT:    or $a7, $a7, $a6
 ; LA64-NEXT:  .LBB4_3: # %atomicrmw.start
 ; LA64-NEXT:    # Parent Loop BB4_1 Depth=1
 ; LA64-NEXT:    # => This Inner Loop Header: Depth=2
-; LA64-NEXT:    ll.w $a6, $a2, 0
-; LA64-NEXT:    bne $a6, $a3, .LBB4_5
+; LA64-NEXT:    ll.w $a6, $a0, 0
+; LA64-NEXT:    bne $a6, $a2, .LBB4_5
 ; LA64-NEXT:  # %bb.4: # %atomicrmw.start
 ; LA64-NEXT:    # in Loop: Header=BB4_3 Depth=2
 ; LA64-NEXT:    dbar 0
 ; LA64-NEXT:    move $t0, $a7
-; LA64-NEXT:    sc.w $t0, $a2, 0
+; LA64-NEXT:    sc.w $t0, $a0, 0
 ; LA64-NEXT:    beqz $t0, .LBB4_3
 ; LA64-NEXT:    b .LBB4_6
 ; LA64-NEXT:  .LBB4_5: # %atomicrmw.start
 ; LA64-NEXT:    # in Loop: Header=BB4_1 Depth=1
 ; LA64-NEXT:    dbar 1792
 ; LA64-NEXT:  .LBB4_6: # %atomicrmw.start
 ; LA64-NEXT:    # in Loop: Header=BB4_1 Depth=1
-; LA64-NEXT:    addi.w $a7, $a3, 0
-; LA64-NEXT:    move $a3, $a6
+; LA64-NEXT:    addi.w $a7, $a2, 0
+; LA64-NEXT:    move $a2, $a6
 ; LA64-NEXT:    bne $a6, $a7, .LBB4_1
 ; LA64-NEXT:  # %bb.2: # %atomicrmw.end
-; LA64-NEXT:    srl.w $a0, $a6, $a0
+; LA64-NEXT:    srl.w $a0, $a6, $a3
 ; LA64-NEXT:    ret
   %result = atomicrmw udec_wrap ptr %ptr, i8 %val seq_cst
   ret i8 %result
 }
 
 define i16 @atomicrmw_udec_wrap_i16(ptr %ptr, i16 %val) {
 ; LA64-LABEL: atomicrmw_udec_wrap_i16:
 ; LA64:       # %bb.0:
-; LA64-NEXT:    addi.w $a2, $zero, -4
-; LA64-NEXT:    and $a2, $a0, $a2
-; LA64-NEXT:    ld.w $a3, $a2, 0
-; LA64-NEXT:    slli.d $a0, $a0, 3
+; LA64-NEXT:    slli.d $a3, $a0, 3
+; LA64-NEXT:    bstrins.d $a0, $zero, 1, 0
+; LA64-NEXT:    ld.w $a2, $a0, 0
 ; LA64-NEXT:    lu12i.w $a4, 15
 ; LA64-NEXT:    ori $a4, $a4, 4095
-; LA64-NEXT:    sll.w $a4, $a4, $a0
-; LA64-NEXT:    andi $a0, $a0, 24
+; LA64-NEXT:    sll.w $a4, $a4, $a3
+; LA64-NEXT:    andi $a3, $a3, 24
 ; LA64-NEXT:    nor $a4, $a4, $zero
 ; LA64-NEXT:    bstrpick.d $a5, $a1, 15, 0
 ; LA64-NEXT:    .p2align 4, , 16
 ; LA64-NEXT:  .LBB5_1: # %atomicrmw.start
 ; LA64-NEXT:    # =>This Loop Header: Depth=1
 ; LA64-NEXT:    # Child Loop BB5_3 Depth 2
-; LA64-NEXT:    srl.w $a6, $a3, $a0
+; LA64-NEXT:    srl.w $a6, $a2, $a3
 ; LA64-NEXT:    bstrpick.d $a7, $a6, 15, 0
 ; LA64-NEXT:    sltu $t0, $a5, $a7
 ; LA64-NEXT:    addi.d $a6, $a6, -1
 ; LA64-NEXT:    masknez $a6, $a6, $t0
 ; LA64-NEXT:    maskeqz $t0, $a1, $t0
 ; LA64-NEXT:    or $a6, $t0, $a6
 ; LA64-NEXT:    sltui $a7, $a7, 1
 ; LA64-NEXT:    masknez $a6, $a6, $a7
 ; LA64-NEXT:    maskeqz $a7, $a1, $a7
 ; LA64-NEXT:    or $a6, $a7, $a6
 ; LA64-NEXT:    bstrpick.d $a6, $a6, 15, 0
-; LA64-NEXT:    sll.w $a6, $a6, $a0
-; LA64-NEXT:    and $a7, $a3, $a4
+; LA64-NEXT:    sll.w $a6, $a6, $a3
+; LA64-NEXT:    and $a7, $a2, $a4
 ; LA64-NEXT:    or $a7, $a7, $a6
 ; LA64-NEXT:  .LBB5_3: # %atomicrmw.start
 ; LA64-NEXT:    # Parent Loop BB5_1 Depth=1
 ; LA64-NEXT:    # => This Inner Loop Header: Depth=2
-; LA64-NEXT:    ll.w $a6, $a2, 0
-; LA64-NEXT:    bne $a6, $a3, .LBB5_5
+; LA64-NEXT:    ll.w $a6, $a0, 0
+; LA64-NEXT:    bne $a6, $a2, .LBB5_5
 ; LA64-NEXT:  # %bb.4: # %atomicrmw.start
 ; LA64-NEXT:    # in Loop: Header=BB5_3 Depth=2
 ; LA64-NEXT:    dbar 0
 ; LA64-NEXT:    move $t0, $a7
-; LA64-NEXT:    sc.w $t0, $a2, 0
+; LA64-NEXT:    sc.w $t0, $a0, 0
 ; LA64-NEXT:    beqz $t0, .LBB5_3
 ; LA64-NEXT:    b .LBB5_6
 ; LA64-NEXT:  .LBB5_5: # %atomicrmw.start
 ; LA64-NEXT:    # in Loop: Header=BB5_1 Depth=1
 ; LA64-NEXT:    dbar 1792
 ; LA64-NEXT:  .LBB5_6: # %atomicrmw.start
 ; LA64-NEXT:    # in Loop: Header=BB5_1 Depth=1
-; LA64-NEXT:    addi.w $a7, $a3, 0
-; LA64-NEXT:    move $a3, $a6
+; LA64-NEXT:    addi.w $a7, $a2, 0
+; LA64-NEXT:    move $a2, $a6
 ; LA64-NEXT:    bne $a6, $a7, .LBB5_1
 ; LA64-NEXT:  # %bb.2: # %atomicrmw.end
-; LA64-NEXT:    srl.w $a0, $a6, $a0
+; LA64-NEXT:    srl.w $a0, $a6, $a3
 ; LA64-NEXT:    ret
   %result = atomicrmw udec_wrap ptr %ptr, i16 %val seq_cst
   ret i16 %result
 }
 
 define i32 @atomicrmw_udec_wrap_i32(ptr %ptr, i32 %val) {
 ; LA64-LABEL: atomicrmw_udec_wrap_i32:
 ; LA64:       # %bb.0:

llvm/test/CodeGen/LoongArch/ir-instruction/and.ll

 ; LA64:       # %bb.0:
 ; LA64-NEXT:    lu12i.w $a1, 255
 ; LA64-NEXT:    and $a0, $a0, $a1
 ; LA64-NEXT:    ret
   %b = and i64 %a, 1044480
   ret i64 %b
 }
 
-;; This case is not optimized to `bstrpick + slli`,
-;; since the immediate -2048 can be composed via
-;; a single `addi.w $rx, $zero, -2048`.
 define i64 @and_i64_minus_2048(i64 %a) {
 ; LA32-LABEL: and_i64_minus_2048:
 ; LA32:       # %bb.0:
-; LA32-NEXT:    addi.w $a2, $zero, -2048
-; LA32-NEXT:    and $a0, $a0, $a2
+; LA32-NEXT:    bstrins.w $a0, $zero, 10, 0
 ; LA32-NEXT:    ret
 ;
 ; LA64-LABEL: and_i64_minus_2048:
 ; LA64:       # %bb.0:
-; LA64-NEXT:    addi.w $a1, $zero, -2048
-; LA64-NEXT:    and $a0, $a0, $a1
+; LA64-NEXT:    bstrins.d $a0, $zero, 10, 0
 ; LA64-NEXT:    ret
   %b = and i64 %a, -2048
   ret i64 %b
 }
 
 ;; This case is not optimized to `bstrpick + slli`,
 ;; since the immediate 0xfff0 has more than 2 uses.
 define i64 @and_i64_0xfff0_multiple_times(i64 %a, i64 %b, i64 %c) {
; LA64-NEXT:    ret
   %e = and i64 %b, 65520
   %f = and i64 %c, 65520
   %g = sub i64 %d, %e
   %h = mul i64 %e, %f
   %i = xor i64 %g, %h
   ret i64 %i
 }
 
-;; TODO: this can be codegened to bstrins.[wd] $a0, $zero, 23, 16.
 define i64 @and_i64_0xffffffffff00ffff(i64 %a) {
 ; LA32-LABEL: and_i64_0xffffffffff00ffff:
 ; LA32:       # %bb.0:
-; LA32-NEXT:    lu12i.w $a2, -4081
-; LA32-NEXT:    ori $a2, $a2, 4095
-; LA32-NEXT:    and $a0, $a0, $a2
+; LA32-NEXT:    bstrins.w $a0, $zero, 23, 16
 ; LA32-NEXT:    ret
 ;
 ; LA64-LABEL: and_i64_0xffffffffff00ffff:
 ; LA64:       # %bb.0:
-; LA64-NEXT:    lu12i.w $a1, -4081
-; LA64-NEXT:    ori $a1, $a1, 4095
-; LA64-NEXT:    and $a0, $a0, $a1
+; LA64-NEXT:    bstrins.d $a0, $zero, 23, 16
 ; LA64-NEXT:    ret
   %b = and i64 %a, 18446744073692839935
   ret i64 %b
 }
 
 define i32 @and_add_lsr(i32 %x, i32 %y) {
 ; LA32-LABEL: and_add_lsr:
 ; LA32:       # %bb.0:

llvm/test/CodeGen/LoongArch/ir-instruction/atomic-cmpxchg.ll

 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
 ; RUN: llc --mtriple=loongarch64 < %s | FileCheck %s --check-prefix=LA64
 
 define void @cmpxchg_i8_acquire_acquire(ptr %ptr, i8 %cmp, i8 %val) nounwind {
 ; LA64-LABEL: cmpxchg_i8_acquire_acquire:
 ; LA64:       # %bb.0:
-; LA64-NEXT:    addi.w $a3, $zero, -4
-; LA64-NEXT:    and $a3, $a0, $a3
-; LA64-NEXT:    slli.d $a0, $a0, 3
 ; LA64-NEXT:    andi $a1, $a1, 255
-; LA64-NEXT:    sll.w $a1, $a1, $a0
+; LA64-NEXT:    slli.d $a3, $a0, 3
+; LA64-NEXT:    sll.w $a1, $a1, $a3
 ; LA64-NEXT:    andi $a2, $a2, 255
-; LA64-NEXT:    sll.w $a2, $a2, $a0
+; LA64-NEXT:    sll.w $a2, $a2, $a3
 ; LA64-NEXT:    ori $a4, $zero, 255
-; LA64-NEXT:    sll.w $a0, $a4, $a0
-; LA64-NEXT:    addi.w $a0, $a0, 0
+; LA64-NEXT:    sll.w $a3, $a4, $a3
+; LA64-NEXT:    addi.w $a3, $a3, 0
 ; LA64-NEXT:    addi.w $a2, $a2, 0
 ; LA64-NEXT:    addi.w $a1, $a1, 0
+; LA64-NEXT:    bstrins.d $a0, $zero, 1, 0
 ; LA64-NEXT:  .LBB0_1: # =>This Inner Loop Header: Depth=1
-; LA64-NEXT:    ll.w $a4, $a3, 0
-; LA64-NEXT:    and $a5, $a4, $a0
+; LA64-NEXT:    ll.w $a4, $a0, 0
+; LA64-NEXT:    and $a5, $a4, $a3
 ; LA64-NEXT:    bne $a5, $a1, .LBB0_3
 ; LA64-NEXT:  # %bb.2: # in Loop: Header=BB0_1 Depth=1
 ; LA64-NEXT:    dbar 0
-; LA64-NEXT:    andn $a5, $a4, $a0
+; LA64-NEXT:    andn $a5, $a4, $a3
 ; LA64-NEXT:    or $a5, $a5, $a2
-; LA64-NEXT:    sc.w $a5, $a3, 0
+; LA64-NEXT:    sc.w $a5, $a0, 0
 ; LA64-NEXT:    beqz $a5, .LBB0_1
 ; LA64-NEXT:    b .LBB0_4
 ; LA64-NEXT:  .LBB0_3:
 ; LA64-NEXT:    dbar 1792
 ; LA64-NEXT:  .LBB0_4:
 ; LA64-NEXT:    ret
   %res = cmpxchg ptr %ptr, i8 %cmp, i8 %val acquire acquire
   ret void
 }
 
 define void @cmpxchg_i16_acquire_acquire(ptr %ptr, i16 %cmp, i16 %val) nounwind {
 ; LA64-LABEL: cmpxchg_i16_acquire_acquire:
 ; LA64:       # %bb.0:
-; LA64-NEXT:    addi.w $a3, $zero, -4
-; LA64-NEXT:    and $a3, $a0, $a3
-; LA64-NEXT:    slli.d $a0, $a0, 3
 ; LA64-NEXT:    bstrpick.d $a1, $a1, 15, 0
-; LA64-NEXT:    sll.w $a1, $a1, $a0
+; LA64-NEXT:    slli.d $a3, $a0, 3
+; LA64-NEXT:    sll.w $a1, $a1, $a3
 ; LA64-NEXT:    bstrpick.d $a2, $a2, 15, 0
-; LA64-NEXT:    sll.w $a2, $a2, $a0
+; LA64-NEXT:    sll.w $a2, $a2, $a3
 ; LA64-NEXT:    lu12i.w $a4, 15
 ; LA64-NEXT:    ori $a4, $a4, 4095
-; LA64-NEXT:    sll.w $a0, $a4, $a0
-; LA64-NEXT:    addi.w $a0, $a0, 0
+; LA64-NEXT:    sll.w $a3, $a4, $a3
+; LA64-NEXT:    addi.w $a3, $a3, 0
 ; LA64-NEXT:    addi.w $a2, $a2, 0
 ; 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:    ll.w $a4, $a3, 0
-; LA64-NEXT:    and $a5, $a4, $a0
+; LA64-NEXT:    ll.w $a4, $a0, 0
+; LA64-NEXT:    and $a5, $a4, $a3
 ; LA64-NEXT:    bne $a5, $a1, .LBB1_3
 ; LA64-NEXT:  # %bb.2: # in Loop: Header=BB1_1 Depth=1
 ; LA64-NEXT:    dbar 0
-; LA64-NEXT:    andn $a5, $a4, $a0
+; LA64-NEXT:    andn $a5, $a4, $a3
 ; LA64-NEXT:    or $a5, $a5, $a2
-; LA64-NEXT:    sc.w $a5, $a3, 0
+; LA64-NEXT:    sc.w $a5, $a0, 0
 ; LA64-NEXT:    beqz $a5, .LBB1_1
 ; LA64-NEXT:    b .LBB1_4
 ; LA64-NEXT:  .LBB1_3:
 ; LA64-NEXT:    dbar 1792
 ; LA64-NEXT:  .LBB1_4:
 ; LA64-NEXT:    ret
   %res = cmpxchg ptr %ptr, i16 %cmp, i16 %val acquire acquire
   ret void

 ; LA64-NEXT:    ret
   %res = cmpxchg ptr %ptr, i64 %cmp, i64 %val acquire acquire
   ret void
 }
 
 define i8 @cmpxchg_i8_acquire_acquire_reti8(ptr %ptr, i8 %cmp, i8 %val) nounwind {
 ; LA64-LABEL: cmpxchg_i8_acquire_acquire_reti8:
 ; LA64:       # %bb.0:
-; LA64-NEXT:    addi.w $a3, $zero, -4
-; LA64-NEXT:    and $a3, $a0, $a3
-; LA64-NEXT:    slli.d $a0, $a0, 3
+; LA64-NEXT:    slli.d $a3, $a0, 3
 ; LA64-NEXT:    ori $a4, $zero, 255
-; LA64-NEXT:    sll.w $a4, $a4, $a0
-; LA64-NEXT:    addi.w $a4, $a4, 0
+; LA64-NEXT:    sll.w $a4, $a4, $a3
 ; LA64-NEXT:    andi $a2, $a2, 255
-; LA64-NEXT:    sll.w $a2, $a2, $a0
+; LA64-NEXT:    addi.w $a4, $a4, 0
+; LA64-NEXT:    sll.w $a2, $a2, $a3
 ; LA64-NEXT:    addi.w $a2, $a2, 0
 ; LA64-NEXT:    andi $a1, $a1, 255
-; LA64-NEXT:    sll.w $a1, $a1, $a0
+; LA64-NEXT:    sll.w $a1, $a1, $a3
 ; 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:    ll.w $a5, $a3, 0
+; LA64-NEXT:    ll.w $a5, $a0, 0
 ; LA64-NEXT:    and $a6, $a5, $a4
 ; LA64-NEXT:    bne $a6, $a1, .LBB4_3
 ; LA64-NEXT:  # %bb.2: # in Loop: Header=BB4_1 Depth=1
 ; LA64-NEXT:    dbar 0
 ; LA64-NEXT:    andn $a6, $a5, $a4
 ; LA64-NEXT:    or $a6, $a6, $a2
-; LA64-NEXT:    sc.w $a6, $a3, 0
+; LA64-NEXT:    sc.w $a6, $a0, 0
 ; LA64-NEXT:    beqz $a6, .LBB4_1
 ; LA64-NEXT:    b .LBB4_4
 ; LA64-NEXT:  .LBB4_3:
 ; LA64-NEXT:    dbar 1792
 ; LA64-NEXT:  .LBB4_4:
-; LA64-NEXT:    srl.w $a0, $a5, $a0
+; LA64-NEXT:    srl.w $a0, $a5, $a3
 ; LA64-NEXT:    ret
   %tmp = cmpxchg ptr %ptr, i8 %cmp, i8 %val acquire acquire
   %res = extractvalue { i8, i1 } %tmp, 0
   ret i8 %res
 }
 
 define i16 @cmpxchg_i16_acquire_acquire_reti16(ptr %ptr, i16 %cmp, i16 %val) nounwind {
 ; LA64-LABEL: cmpxchg_i16_acquire_acquire_reti16:
 ; LA64:       # %bb.0:
-; LA64-NEXT:    addi.w $a3, $zero, -4
-; LA64-NEXT:    and $a3, $a0, $a3
-; LA64-NEXT:    slli.d $a0, $a0, 3
-; LA64-NEXT:    lu12i.w $a4, 15
-; LA64-NEXT:    ori $a4, $a4, 4095
-; LA64-NEXT:    sll.w $a4, $a4, $a0
-; LA64-NEXT:    addi.w $a4, $a4, 0
+; LA64-NEXT:    lu12i.w $a3, 15
+; LA64-NEXT:    ori $a3, $a3, 4095
+; LA64-NEXT:    slli.d $a4, $a0, 3
+; LA64-NEXT:    sll.w $a3, $a3, $a4
 ; LA64-NEXT:    bstrpick.d $a2, $a2, 15, 0
-; LA64-NEXT:    sll.w $a2, $a2, $a0
+; LA64-NEXT:    addi.w $a3, $a3, 0
+; LA64-NEXT:    sll.w $a2, $a2, $a4
 ; LA64-NEXT:    addi.w $a2, $a2, 0
 ; LA64-NEXT:    bstrpick.d $a1, $a1, 15, 0
-; LA64-NEXT:    sll.w $a1, $a1, $a0
+; LA64-NEXT:    sll.w $a1, $a1, $a4
 ; 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:    ll.w $a5, $a3, 0
-; LA64-NEXT:    and $a6, $a5, $a4
+; LA64-NEXT:    ll.w $a5, $a0, 0
+; LA64-NEXT:    and $a6, $a5, $a3
 ; LA64-NEXT:    bne $a6, $a1, .LBB5_3
 ; LA64-NEXT:  # %bb.2: # in Loop: Header=BB5_1 Depth=1
 ; LA64-NEXT:    dbar 0
-; LA64-NEXT:    andn $a6, $a5, $a4
+; LA64-NEXT:    andn $a6, $a5, $a3
 ; LA64-NEXT:    or $a6, $a6, $a2
-; LA64-NEXT:    sc.w $a6, $a3, 0
+; LA64-NEXT:    sc.w $a6, $a0, 0
 ; LA64-NEXT:    beqz $a6, .LBB5_1
 ; LA64-NEXT:    b .LBB5_4
 ; LA64-NEXT:  .LBB5_3:
 ; LA64-NEXT:    dbar 1792
 ; LA64-NEXT:  .LBB5_4:
-; LA64-NEXT:    srl.w $a0, $a5, $a0
+; LA64-NEXT:    srl.w $a0, $a5, $a4
 ; LA64-NEXT:    ret
   %tmp = cmpxchg ptr %ptr, i16 %cmp, i16 %val acquire acquire
   %res = extractvalue { i16, i1 } %tmp, 0
   ret i16 %res
 }
 
 define i32 @cmpxchg_i32_acquire_acquire_reti32(ptr %ptr, i32 %cmp, i32 %val) nounwind {
 ; LA64-LABEL: cmpxchg_i32_acquire_acquire_reti32:
; LA64-NEXT:    ret
   %tmp = cmpxchg ptr %ptr, i64 %cmp, i64 %val acquire acquire
   %res = extractvalue { i64, i1 } %tmp, 0
   ret i64 %res
 }
 
 define i1 @cmpxchg_i8_acquire_acquire_reti1(ptr %ptr, i8 %cmp, i8 %val) nounwind {
 ; LA64-LABEL: cmpxchg_i8_acquire_acquire_reti1:
 ; LA64:       # %bb.0:
-; LA64-NEXT:    addi.w $a3, $zero, -4
-; LA64-NEXT:    and $a3, $a0, $a3
-; LA64-NEXT:    slli.d $a0, $a0, 3
-; LA64-NEXT:    ori $a4, $zero, 255
-; LA64-NEXT:    sll.w $a4, $a4, $a0
 ; LA64-NEXT:    andi $a1, $a1, 255
-; LA64-NEXT:    sll.w $a1, $a1, $a0
+; LA64-NEXT:    slli.d $a3, $a0, 3
+; LA64-NEXT:    ori $a4, $zero, 255
+; LA64-NEXT:    sll.w $a4, $a4, $a3
+; LA64-NEXT:    sll.w $a1, $a1, $a3
 ; LA64-NEXT:    andi $a2, $a2, 255
-; LA64-NEXT:    sll.w $a0, $a2, $a0
-; LA64-NEXT:    addi.w $a0, $a0, 0
+; LA64-NEXT:    sll.w $a2, $a2, $a3
+; LA64-NEXT:    addi.w $a2, $a2, 0
 ; LA64-NEXT:    addi.w $a1, $a1, 0
-; LA64-NEXT:    addi.w $a2, $a4, 0
+; LA64-NEXT:    bstrins.d $a0, $zero, 1, 0
+; LA64-NEXT:    addi.w $a3, $a4, 0
 ; LA64-NEXT:  .LBB8_1: # =>This Inner Loop Header: Depth=1
-; LA64-NEXT:    ll.w $a5, $a3, 0
-; LA64-NEXT:    and $a6, $a5, $a2
+; LA64-NEXT:    ll.w $a5, $a0, 0
+; LA64-NEXT:    and $a6, $a5, $a3
 ; LA64-NEXT:    bne $a6, $a1, .LBB8_3
 ; LA64-NEXT:  # %bb.2: # in Loop: Header=BB8_1 Depth=1
 ; LA64-NEXT:    dbar 0
-; LA64-NEXT:    andn $a6, $a5, $a2
-; LA64-NEXT:    or $a6, $a6, $a0
-; LA64-NEXT:    sc.w $a6, $a3, 0
+; LA64-NEXT:    andn $a6, $a5, $a3
+; LA64-NEXT:    or $a6, $a6, $a2
+; LA64-NEXT:    sc.w $a6, $a0, 0
 ; LA64-NEXT:    beqz $a6, .LBB8_1
 ; LA64-NEXT:    b .LBB8_4
 ; LA64-NEXT:  .LBB8_3:
 ; LA64-NEXT:    dbar 1792
 ; LA64-NEXT:  .LBB8_4:
 ; LA64-NEXT:    and $a0, $a5, $a4
 ; LA64-NEXT:    addi.w $a0, $a0, 0
 ; LA64-NEXT:    xor $a0, $a1, $a0
 ; LA64-NEXT:    sltui $a0, $a0, 1
 ; LA64-NEXT:    ret
   %tmp = cmpxchg ptr %ptr, i8 %cmp, i8 %val acquire acquire
   %res = extractvalue { i8, i1 } %tmp, 1
   ret i1 %res
 }
 
 define i1 @cmpxchg_i16_acquire_acquire_reti1(ptr %ptr, i16 %cmp, i16 %val) nounwind {
 ; LA64-LABEL: cmpxchg_i16_acquire_acquire_reti1:
 ; LA64:       # %bb.0:
-; LA64-NEXT:    addi.w $a3, $zero, -4
-; LA64-NEXT:    and $a3, $a0, $a3
-; LA64-NEXT:    slli.d $a0, $a0, 3
-; LA64-NEXT:    lu12i.w $a4, 15
-; LA64-NEXT:    ori $a4, $a4, 4095
-; LA64-NEXT:    sll.w $a4, $a4, $a0
 ; LA64-NEXT:    bstrpick.d $a1, $a1, 15, 0
-; LA64-NEXT:    sll.w $a1, $a1, $a0
+; LA64-NEXT:    lu12i.w $a3, 15
+; LA64-NEXT:    ori $a3, $a3, 4095
+; LA64-NEXT:    slli.d $a4, $a0, 3
+; LA64-NEXT:    sll.w $a3, $a3, $a4
+; LA64-NEXT:    sll.w $a1, $a1, $a4
 ; LA64-NEXT:    bstrpick.d $a2, $a2, 15, 0
-; LA64-NEXT:    sll.w $a0, $a2, $a0
-; LA64-NEXT:    addi.w $a0, $a0, 0
+; LA64-NEXT:    sll.w $a2, $a2, $a4
+; LA64-NEXT:    addi.w $a2, $a2, 0
 ; LA64-NEXT:    addi.w $a1, $a1, 0
-; LA64-NEXT:    addi.w $a2, $a4, 0
+; LA64-NEXT:    bstrins.d $a0, $zero, 1, 0
+; LA64-NEXT:    addi.w $a4, $a3, 0
 ; LA64-NEXT:  .LBB9_1: # =>This Inner Loop Header: Depth=1
-; LA64-NEXT:    ll.w $a5, $a3, 0
-; LA64-NEXT:    and $a6, $a5, $a2
+; LA64-NEXT:    ll.w $a5, $a0, 0
+; LA64-NEXT:    and $a6, $a5, $a4
 ; LA64-NEXT:    bne $a6, $a1, .LBB9_3
 ; LA64-NEXT:  # %bb.2: # in Loop: Header=BB9_1 Depth=1
 ; LA64-NEXT:    dbar 0
-; LA64-NEXT:    andn $a6, $a5, $a2
-; LA64-NEXT:    or $a6, $a6, $a0
-; LA64-NEXT:    sc.w $a6, $a3, 0
+; LA64-NEXT:    andn $a6, $a5, $a4
+; LA64-NEXT:    or $a6, $a6, $a2
+; LA64-NEXT:    sc.w $a6, $a0, 0
 ; LA64-NEXT:    beqz $a6, .LBB9_1
 ; LA64-NEXT:    b .LBB9_4
 ; LA64-NEXT:  .LBB9_3:
 ; LA64-NEXT:    dbar 1792
 ; LA64-NEXT:  .LBB9_4:
-; LA64-NEXT:    and $a0, $a5, $a4
+; LA64-NEXT:    and $a0, $a5, $a3
 ; LA64-NEXT:    addi.w $a0, $a0, 0
 ; LA64-NEXT:    xor $a0, $a1, $a0
 ; LA64-NEXT:    sltui $a0, $a0, 1
 ; LA64-NEXT:    ret
   %tmp = cmpxchg ptr %ptr, i16 %cmp, i16 %val acquire acquire
   %res = extractvalue { i16, i1 } %tmp, 1
   ret i1 %res
 }

llvm/test/CodeGen/LoongArch/ir-instruction/atomicrmw-minmax.ll

 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
 ; RUN: llc --mtriple=loongarch64 --verify-machineinstrs < %s | \
 ; RUN:   FileCheck %s --check-prefix=LA64
 
 ;; TODO: Testing for LA32 architecture will be added later
 
 define i8 @atomicrmw_umax_i8_acquire(ptr %a, i8 %b) nounwind {
 ; LA64-LABEL: atomicrmw_umax_i8_acquire:
 ; LA64:       # %bb.0:
-; LA64-NEXT:    addi.w $a2, $zero, -4
-; LA64-NEXT:    and $a2, $a0, $a2
-; LA64-NEXT:    slli.d $a0, $a0, 3
+; LA64-NEXT:    slli.d $a2, $a0, 3
 ; LA64-NEXT:    ori $a3, $zero, 255
-; LA64-NEXT:    sll.w $a3, $a3, $a0
+; LA64-NEXT:    sll.w $a3, $a3, $a2
 ; LA64-NEXT:    addi.w $a3, $a3, 0
 ; LA64-NEXT:    andi $a1, $a1, 255
-; LA64-NEXT:    sll.w $a1, $a1, $a0
+; LA64-NEXT:    sll.w $a1, $a1, $a2
 ; LA64-NEXT:    addi.w $a1, $a1, 0
+; LA64-NEXT:    bstrins.d $a0, $zero, 1, 0
 ; LA64-NEXT:  .LBB0_1: # =>This Inner Loop Header: Depth=1
 ; LA64-NEXT:    dbar 0
-; LA64-NEXT:    ll.w $a4, $a2, 0
+; LA64-NEXT:    ll.w $a4, $a0, 0
 ; LA64-NEXT:    and $a6, $a4, $a3
 ; LA64-NEXT:    move $a5, $a4
 ; LA64-NEXT:    bgeu $a6, $a1, .LBB0_3
 ; LA64-NEXT:  # %bb.2: # in Loop: Header=BB0_1 Depth=1
 ; LA64-NEXT:    xor $a5, $a4, $a1
 ; LA64-NEXT:    and $a5, $a5, $a3
 ; LA64-NEXT:    xor $a5, $a4, $a5
 ; LA64-NEXT:  .LBB0_3: # in Loop: Header=BB0_1 Depth=1
-; LA64-NEXT:    sc.w $a5, $a2, 0
+; LA64-NEXT:    sc.w $a5, $a0, 0
 ; LA64-NEXT:    beqz $a5, .LBB0_1
 ; LA64-NEXT:  # %bb.4:
 ; LA64-NEXT:    dbar 1792
 ; LA64-NEXT:  # %bb.5:
-; LA64-NEXT:    srl.w $a0, $a4, $a0
+; LA64-NEXT:    srl.w $a0, $a4, $a2
 ; LA64-NEXT:    ret
   %1 = atomicrmw umax ptr %a, i8 %b acquire
   ret i8 %1
 }
 
 define i16 @atomicrmw_umax_i16_acquire(ptr %a, i16 %b) nounwind {
 ; LA64-LABEL: atomicrmw_umax_i16_acquire:
 ; LA64:       # %bb.0:
-; LA64-NEXT:    addi.w $a2, $zero, -4
-; LA64-NEXT:    and $a2, $a0, $a2
-; LA64-NEXT:    slli.d $a0, $a0, 3
-; LA64-NEXT:    lu12i.w $a3, 15
-; LA64-NEXT:    ori $a3, $a3, 4095
-; LA64-NEXT:    sll.w $a3, $a3, $a0
-; LA64-NEXT:    addi.w $a3, $a3, 0
+; LA64-NEXT:    lu12i.w $a2, 15
+; LA64-NEXT:    ori $a2, $a2, 4095
+; LA64-NEXT:    slli.d $a3, $a0, 3
+; LA64-NEXT:    sll.w $a2, $a2, $a3
+; LA64-NEXT:    addi.w $a2, $a2, 0
 ; LA64-NEXT:    bstrpick.d $a1, $a1, 15, 0
-; LA64-NEXT:    sll.w $a1, $a1, $a0
+; 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:    dbar 0
-; LA64-NEXT:    ll.w $a4, $a2, 0
-; LA64-NEXT:    and $a6, $a4, $a3
+; LA64-NEXT:    ll.w $a4, $a0, 0
+; LA64-NEXT:    and $a6, $a4, $a2
 ; LA64-NEXT:    move $a5, $a4
 ; LA64-NEXT:    bgeu $a6, $a1, .LBB1_3
 ; LA64-NEXT:  # %bb.2: # in Loop: Header=BB1_1 Depth=1
 ; LA64-NEXT:    xor $a5, $a4, $a1
-; LA64-NEXT:    and $a5, $a5, $a3
+; LA64-NEXT:    and $a5, $a5, $a2
 ; LA64-NEXT:    xor $a5, $a4, $a5
 ; LA64-NEXT:  .LBB1_3: # in Loop: Header=BB1_1 Depth=1
-; LA64-NEXT:    sc.w $a5, $a2, 0
+; LA64-NEXT:    sc.w $a5, $a0, 0
 ; LA64-NEXT:    beqz $a5, .LBB1_1
 ; LA64-NEXT:  # %bb.4:
 ; LA64-NEXT:    dbar 1792
 ; LA64-NEXT:  # %bb.5:
-; LA64-NEXT:    srl.w $a0, $a4, $a0
+; LA64-NEXT:    srl.w $a0, $a4, $a3
 ; LA64-NEXT:    ret
   %1 = atomicrmw umax ptr %a, i16 %b acquire
   ret i16 %1
 }
 
 define i32 @atomicrmw_umax_i32_acquire(ptr %a, i32 %b) nounwind {
 ; LA64-LABEL: atomicrmw_umax_i32_acquire:
 ; LA64:       # %bb.0:

 ; LA64-NEXT:    ret
   %1 = atomicrmw umax ptr %a, i64 %b acquire
   ret i64 %1
 }
 
 define i8 @atomicrmw_umin_i8_acquire(ptr %a, i8 %b) nounwind {
 ; LA64-LABEL: atomicrmw_umin_i8_acquire:
 ; LA64:       # %bb.0:
-; LA64-NEXT:    addi.w $a2, $zero, -4
-; LA64-NEXT:    and $a2, $a0, $a2
-; LA64-NEXT:    slli.d $a0, $a0, 3
+; LA64-NEXT:    slli.d $a2, $a0, 3
 ; LA64-NEXT:    ori $a3, $zero, 255
-; LA64-NEXT:    sll.w $a3, $a3, $a0
+; LA64-NEXT:    sll.w $a3, $a3, $a2
 ; LA64-NEXT:    addi.w $a3, $a3, 0
 ; LA64-NEXT:    andi $a1, $a1, 255
-; LA64-NEXT:    sll.w $a1, $a1, $a0
+; 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 $a4, $a2, 0
+; LA64-NEXT:    ll.w $a4, $a0, 0
 ; LA64-NEXT:    and $a6, $a4, $a3
 ; LA64-NEXT:    move $a5, $a4
 ; LA64-NEXT:    bgeu $a1, $a6, .LBB4_3
 ; LA64-NEXT:  # %bb.2: # in Loop: Header=BB4_1 Depth=1
 ; LA64-NEXT:    xor $a5, $a4, $a1
 ; LA64-NEXT:    and $a5, $a5, $a3
 ; LA64-NEXT:    xor $a5, $a4, $a5
 ; LA64-NEXT:  .LBB4_3: # in Loop: Header=BB4_1 Depth=1
-; LA64-NEXT:    sc.w $a5, $a2, 0
+; LA64-NEXT:    sc.w $a5, $a0, 0
 ; LA64-NEXT:    beqz $a5, .LBB4_1
 ; LA64-NEXT:  # %bb.4:
 ; LA64-NEXT:    dbar 1792
 ; LA64-NEXT:  # %bb.5:
-; LA64-NEXT:    srl.w $a0, $a4, $a0
+; LA64-NEXT:    srl.w $a0, $a4, $a2
 ; LA64-NEXT:    ret
   %1 = atomicrmw umin ptr %a, i8 %b acquire
   ret i8 %1
 }
 
 define i16 @atomicrmw_umin_i16_acquire(ptr %a, i16 %b) nounwind {
 ; LA64-LABEL: atomicrmw_umin_i16_acquire:
 ; LA64:       # %bb.0:
-; LA64-NEXT:    addi.w $a2, $zero, -4
-; LA64-NEXT:    and $a2, $a0, $a2
-; LA64-NEXT:    slli.d $a0, $a0, 3
-; LA64-NEXT:    lu12i.w $a3, 15
-; LA64-NEXT:    ori $a3, $a3, 4095
-; LA64-NEXT:    sll.w $a3, $a3, $a0
-; LA64-NEXT:    addi.w $a3, $a3, 0
+; LA64-NEXT:    lu12i.w $a2, 15
+; LA64-NEXT:    ori $a2, $a2, 4095
+; LA64-NEXT:    slli.d $a3, $a0, 3
+; LA64-NEXT:    sll.w $a2, $a2, $a3
+; LA64-NEXT:    addi.w $a2, $a2, 0
 ; LA64-NEXT:    bstrpick.d $a1, $a1, 15, 0
-; LA64-NEXT:    sll.w $a1, $a1, $a0
+; 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:    dbar 0
-; LA64-NEXT:    ll.w $a4, $a2, 0
-; LA64-NEXT:    and $a6, $a4, $a3
+; LA64-NEXT:    ll.w $a4, $a0, 0
+; LA64-NEXT:    and $a6, $a4, $a2
 ; LA64-NEXT:    move $a5, $a4
 ; LA64-NEXT:    bgeu $a1, $a6, .LBB5_3
 ; LA64-NEXT:  # %bb.2: # in Loop: Header=BB5_1 Depth=1
 ; LA64-NEXT:    xor $a5, $a4, $a1
-; LA64-NEXT:    and $a5, $a5, $a3
+; LA64-NEXT:    and $a5, $a5, $a2
 ; LA64-NEXT:    xor $a5, $a4, $a5
 ; LA64-NEXT:  .LBB5_3: # in Loop: Header=BB5_1 Depth=1
-; LA64-NEXT:    sc.w $a5, $a2, 0
+; LA64-NEXT:    sc.w $a5, $a0, 0
 ; LA64-NEXT:    beqz $a5, .LBB5_1
 ; LA64-NEXT:  # %bb.4:
 ; LA64-NEXT:    dbar 1792
 ; LA64-NEXT:  # %bb.5:
-; LA64-NEXT:    srl.w $a0, $a4, $a0
+; LA64-NEXT:    srl.w $a0, $a4, $a3
 ; LA64-NEXT:    ret
   %1 = atomicrmw umin ptr %a, i16 %b acquire
   ret i16 %1
 }
 
 define i32 @atomicrmw_umin_i32_acquire(ptr %a, i32 %b) nounwind {
 ; LA64-LABEL: atomicrmw_umin_i32_acquire:
 ; LA64:       # %bb.0:

 ; LA64-NEXT:    ret
   %1 = atomicrmw umin ptr %a, i64 %b acquire
   ret i64 %1
 }
 
 define i8 @atomicrmw_max_i8_acquire(ptr %a, i8 %b) nounwind {
 ; LA64-LABEL: atomicrmw_max_i8_acquire:
 ; LA64:       # %bb.0:
-; LA64-NEXT:    addi.w $a2, $zero, -4
-; LA64-NEXT:    and $a2, $a0, $a2
-; LA64-NEXT:    slli.d $a0, $a0, 3
+; LA64-NEXT:    slli.d $a2, $a0, 3
 ; LA64-NEXT:    ori $a3, $zero, 255
-; LA64-NEXT:    sll.w $a3, $a3, $a0
+; LA64-NEXT:    sll.w $a3, $a3, $a2
 ; LA64-NEXT:    addi.w $a3, $a3, 0
 ; LA64-NEXT:    ext.w.b $a1, $a1
-; LA64-NEXT:    sll.w $a1, $a1, $a0
+; LA64-NEXT:    sll.w $a1, $a1, $a2
 ; LA64-NEXT:    addi.w $a1, $a1, 0
-; LA64-NEXT:    andi $a4, $a0, 24
+; LA64-NEXT:    bstrins.d $a0, $zero, 1, 0
+; LA64-NEXT:    andi $a4, $a2, 24
 ; LA64-NEXT:    xori $a4, $a4, 56
 ; LA64-NEXT:  .LBB8_1: # =>This Inner Loop Header: Depth=1
 ; LA64-NEXT:    dbar 0
-; LA64-NEXT:    ll.w $a5, $a2, 0
+; LA64-NEXT:    ll.w $a5, $a0, 0
 ; LA64-NEXT:    and $a7, $a5, $a3
 ; LA64-NEXT:    move $a6, $a5
 ; LA64-NEXT:    sll.w $a7, $a7, $a4
 ; LA64-NEXT:    sra.w $a7, $a7, $a4
 ; LA64-NEXT:    bge $a7, $a1, .LBB8_3
 ; LA64-NEXT:  # %bb.2: # in Loop: Header=BB8_1 Depth=1
 ; LA64-NEXT:    xor $a6, $a5, $a1
 ; LA64-NEXT:    and $a6, $a6, $a3
 ; LA64-NEXT:    xor $a6, $a5, $a6
 ; LA64-NEXT:  .LBB8_3: # in Loop: Header=BB8_1 Depth=1
-; LA64-NEXT:    sc.w $a6, $a2, 0
+; LA64-NEXT:    sc.w $a6, $a0, 0
 ; LA64-NEXT:    beqz $a6, .LBB8_1
 ; LA64-NEXT:  # %bb.4:
 ; LA64-NEXT:    dbar 1792
 ; LA64-NEXT:  # %bb.5:
-; LA64-NEXT:    srl.w $a0, $a5, $a0
+; LA64-NEXT:    srl.w $a0, $a5, $a2
 ; LA64-NEXT:    ret
   %1 = atomicrmw max ptr %a, i8 %b acquire
   ret i8 %1
 }
 
 define i16 @atomicrmw_max_i16_acquire(ptr %a, i16 %b) nounwind {
 ; LA64-LABEL: atomicrmw_max_i16_acquire:
 ; LA64:       # %bb.0:
-; LA64-NEXT:    addi.w $a2, $zero, -4
-; LA64-NEXT:    and $a2, $a0, $a2
-; LA64-NEXT:    slli.d $a0, $a0, 3
-; LA64-NEXT:    andi $a3, $a0, 24
+; LA64-NEXT:    slli.d $a2, $a0, 3
+; LA64-NEXT:    andi $a3, $a2, 24
 ; LA64-NEXT:    ori $a4, $zero, 48
 ; LA64-NEXT:    sub.d $a3, $a4, $a3
 ; LA64-NEXT:    lu12i.w $a4, 15
 ; LA64-NEXT:    ori $a4, $a4, 4095
-; LA64-NEXT:    sll.w $a4, $a4, $a0
+; LA64-NEXT:    sll.w $a4, $a4, $a2
 ; LA64-NEXT:    addi.w $a4, $a4, 0
 ; LA64-NEXT:    ext.w.h $a1, $a1
-; LA64-NEXT:    sll.w $a1, $a1, $a0
+; LA64-NEXT:    sll.w $a1, $a1, $a2
 ; 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:    dbar 0
-; LA64-NEXT:    ll.w $a5, $a2, 0
+; LA64-NEXT:    ll.w $a5, $a0, 0
 ; LA64-NEXT:    and $a7, $a5, $a4
 ; LA64-NEXT:    move $a6, $a5
 ; LA64-NEXT:    sll.w $a7, $a7, $a3
 ; LA64-NEXT:    sra.w $a7, $a7, $a3
 ; LA64-NEXT:    bge $a7, $a1, .LBB9_3
 ; LA64-NEXT:  # %bb.2: # in Loop: Header=BB9_1 Depth=1
 ; LA64-NEXT:    xor $a6, $a5, $a1
 ; LA64-NEXT:    and $a6, $a6, $a4
 ; LA64-NEXT:    xor $a6, $a5, $a6
 ; LA64-NEXT:  .LBB9_3: # in Loop: Header=BB9_1 Depth=1
-; LA64-NEXT:    sc.w $a6, $a2, 0
+; LA64-NEXT:    sc.w $a6, $a0, 0
 ; LA64-NEXT:    beqz $a6, .LBB9_1
 ; LA64-NEXT:  # %bb.4:
 ; LA64-NEXT:    dbar 1792
 ; LA64-NEXT:  # %bb.5:
-; LA64-NEXT:    srl.w $a0, $a5, $a0
+; LA64-NEXT:    srl.w $a0, $a5, $a2
 ; LA64-NEXT:    ret
   %1 = atomicrmw max ptr %a, i16 %b acquire
   ret i16 %1
 }
 
 define i32 @atomicrmw_max_i32_acquire(ptr %a, i32 %b) nounwind {
 ; LA64-LABEL: atomicrmw_max_i32_acquire:
 ; LA64:       # %bb.0:

 ; LA64-NEXT:    ret
   %1 = atomicrmw max ptr %a, i64 %b acquire
   ret i64 %1
 }
 
 define i8 @atomicrmw_min_i8_acquire(ptr %a, i8 %b) nounwind {
 ; LA64-LABEL: atomicrmw_min_i8_acquire:
 ; LA64:       # %bb.0:
-; LA64-NEXT:    addi.w $a2, $zero, -4
-; LA64-NEXT:    and $a2, $a0, $a2
-; LA64-NEXT:    slli.d $a0, $a0, 3
+; LA64-NEXT:    slli.d $a2, $a0, 3
 ; LA64-NEXT:    ori $a3, $zero, 255
-; LA64-NEXT:    sll.w $a3, $a3, $a0
+; LA64-NEXT:    sll.w $a3, $a3, $a2
 ; LA64-NEXT:    addi.w $a3, $a3, 0
 ; LA64-NEXT:    ext.w.b $a1, $a1
-; LA64-NEXT:    sll.w $a1, $a1, $a0
+; LA64-NEXT:    sll.w $a1, $a1, $a2
 ; LA64-NEXT:    addi.w $a1, $a1, 0
-; LA64-NEXT:    andi $a4, $a0, 24
+; LA64-NEXT:    bstrins.d $a0, $zero, 1, 0
+; LA64-NEXT:    andi $a4, $a2, 24
 ; LA64-NEXT:    xori $a4, $a4, 56
 ; LA64-NEXT:  .LBB12_1: # =>This Inner Loop Header: Depth=1
 ; LA64-NEXT:    dbar 0
-; LA64-NEXT:    ll.w $a5, $a2, 0
+; LA64-NEXT:    ll.w $a5, $a0, 0
 ; LA64-NEXT:    and $a7, $a5, $a3
 ; LA64-NEXT:    move $a6, $a5
 ; LA64-NEXT:    sll.w $a7, $a7, $a4
 ; LA64-NEXT:    sra.w $a7, $a7, $a4
 ; LA64-NEXT:    bge $a1, $a7, .LBB12_3
 ; LA64-NEXT:  # %bb.2: # in Loop: Header=BB12_1 Depth=1
 ; LA64-NEXT:    xor $a6, $a5, $a1
 ; LA64-NEXT:    and $a6, $a6, $a3
 ; LA64-NEXT:    xor $a6, $a5, $a6
 ; LA64-NEXT:  .LBB12_3: # in Loop: Header=BB12_1 Depth=1
-; LA64-NEXT:    sc.w $a6, $a2, 0
+; LA64-NEXT:    sc.w $a6, $a0, 0
 ; LA64-NEXT:    beqz $a6, .LBB12_1
 ; LA64-NEXT:  # %bb.4:
 ; LA64-NEXT:    dbar 1792
 ; LA64-NEXT:  # %bb.5:
-; LA64-NEXT:    srl.w $a0, $a5, $a0
+; LA64-NEXT:    srl.w $a0, $a5, $a2
 ; LA64-NEXT:    ret
   %1 = atomicrmw min ptr %a, i8 %b acquire
   ret i8 %1
 }
 
 define i16 @atomicrmw_min_i16_acquire(ptr %a, i16 %b) nounwind {
 ; LA64-LABEL: atomicrmw_min_i16_acquire:
 ; LA64:       # %bb.0:
-; LA64-NEXT:    addi.w $a2, $zero, -4
-; LA64-NEXT:    and $a2, $a0, $a2
-; LA64-NEXT:    slli.d $a0, $a0, 3
-; LA64-NEXT:    andi $a3, $a0, 24
+; LA64-NEXT:    slli.d $a2, $a0, 3
+; LA64-NEXT:    andi $a3, $a2, 24
 ; LA64-NEXT:    ori $a4, $zero, 48
 ; LA64-NEXT:    sub.d $a3, $a4, $a3
 ; LA64-NEXT:    lu12i.w $a4, 15
 ; LA64-NEXT:    ori $a4, $a4, 4095
-; LA64-NEXT:    sll.w $a4, $a4, $a0
+; LA64-NEXT:    sll.w $a4, $a4, $a2
 ; LA64-NEXT:    addi.w $a4, $a4, 0
 ; LA64-NEXT:    ext.w.h $a1, $a1
-; LA64-NEXT:    sll.w $a1, $a1, $a0
+; LA64-NEXT:    sll.w $a1, $a1, $a2
 ; 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:    dbar 0
-; LA64-NEXT:    ll.w $a5, $a2, 0
+; LA64-NEXT:    ll.w $a5, $a0, 0
 ; LA64-NEXT:    and $a7, $a5, $a4
 ; LA64-NEXT:    move $a6, $a5
 ; LA64-NEXT:    sll.w $a7, $a7, $a3
 ; LA64-NEXT:    sra.w $a7, $a7, $a3
 ; LA64-NEXT:    bge $a1, $a7, .LBB13_3
 ; LA64-NEXT:  # %bb.2: # in Loop: Header=BB13_1 Depth=1
 ; LA64-NEXT:    xor $a6, $a5, $a1
 ; LA64-NEXT:    and $a6, $a6, $a4
 ; LA64-NEXT:    xor $a6, $a5, $a6
 ; LA64-NEXT:  .LBB13_3: # in Loop: Header=BB13_1 Depth=1
-; LA64-NEXT:    sc.w $a6, $a2, 0
+; LA64-NEXT:    sc.w $a6, $a0, 0
 ; LA64-NEXT:    beqz $a6, .LBB13_1
 ; LA64-NEXT:  # %bb.4:
 ; LA64-NEXT:    dbar 1792
 ; LA64-NEXT:  # %bb.5:
-; LA64-NEXT:    srl.w $a0, $a5, $a0
+; LA64-NEXT:    srl.w $a0, $a5, $a2
 ; LA64-NEXT:    ret
   %1 = atomicrmw min ptr %a, i16 %b acquire
   ret i16 %1
 }
 
 define i32 @atomicrmw_min_i32_acquire(ptr %a, i32 %b) nounwind {
 ; LA64-LABEL: atomicrmw_min_i32_acquire:
 ; LA64:       # %bb.0:

llvm/test/CodeGen/LoongArch/ir-instruction/atomicrmw.ll

 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
 ; RUN: llc --mtriple=loongarch32 < %s | FileCheck %s --check-prefix=LA32
 ; RUN: llc --mtriple=loongarch64 < %s | FileCheck %s --check-prefix=LA64
 
 define i8 @atomicrmw_xchg_i8_acquire(ptr %a, i8 %b) nounwind {
 ; LA32-LABEL: atomicrmw_xchg_i8_acquire:
 ; LA32:       # %bb.0:
-; LA32-NEXT:    addi.w $a2, $zero, -4
-; LA32-NEXT:    and $a2, $a0, $a2
-; LA32-NEXT:    slli.w $a0, $a0, 3
+; LA32-NEXT:    slli.w $a2, $a0, 3
 ; LA32-NEXT:    ori $a3, $zero, 255
-; LA32-NEXT:    sll.w $a3, $a3, $a0
+; LA32-NEXT:    sll.w $a3, $a3, $a2
 ; LA32-NEXT:    andi $a1, $a1, 255
-; LA32-NEXT:    sll.w $a1, $a1, $a0
+; LA32-NEXT:    sll.w $a1, $a1, $a2
+; LA32-NEXT:    bstrins.w $a0, $zero, 1, 0
 ; LA32-NEXT:  .LBB0_1: # =>This Inner Loop Header: Depth=1
 ; LA32-NEXT:    dbar 0
-; LA32-NEXT:    ll.w $a4, $a2, 0
+; 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, $a2, 0
+; LA32-NEXT:    sc.w $a5, $a0, 0
 ; LA32-NEXT:    beqz $a5, .LBB0_1
 ; LA32-NEXT:  # %bb.2:
-; LA32-NEXT:    srl.w $a0, $a4, $a0
+; LA32-NEXT:    srl.w $a0, $a4, $a2
 ; LA32-NEXT:    ret
 ;
 ; LA64-LABEL: atomicrmw_xchg_i8_acquire:
 ; LA64:       # %bb.0:
-; LA64-NEXT:    addi.w $a2, $zero, -4
-; LA64-NEXT:    and $a2, $a0, $a2
-; LA64-NEXT:    slli.d $a0, $a0, 3
+; LA64-NEXT:    slli.d $a2, $a0, 3
 ; LA64-NEXT:    ori $a3, $zero, 255
-; LA64-NEXT:    sll.w $a3, $a3, $a0
+; LA64-NEXT:    sll.w $a3, $a3, $a2
 ; LA64-NEXT:    addi.w $a3, $a3, 0
 ; LA64-NEXT:    andi $a1, $a1, 255
-; LA64-NEXT:    sll.w $a1, $a1, $a0
+; LA64-NEXT:    sll.w $a1, $a1, $a2
 ; LA64-NEXT:    addi.w $a1, $a1, 0
+; LA64-NEXT:    bstrins.d $a0, $zero, 1, 0
 ; LA64-NEXT:  .LBB0_1: # =>This Inner Loop Header: Depth=1
 ; LA64-NEXT:    dbar 0
-; LA64-NEXT:    ll.w $a4, $a2, 0
+; 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, $a2, 0
+; LA64-NEXT:    sc.w $a5, $a0, 0
 ; LA64-NEXT:    beqz $a5, .LBB0_1
 ; LA64-NEXT:  # %bb.2:
-; LA64-NEXT:    srl.w $a0, $a4, $a0
+; LA64-NEXT:    srl.w $a0, $a4, $a2
 ; LA64-NEXT:    ret
   %1 = atomicrmw xchg ptr %a, i8 %b acquire
   ret i8 %1
 }
 
 define i16 @atomicrmw_xchg_i16_acquire(ptr %a, i16 %b) nounwind {
 ; LA32-LABEL: atomicrmw_xchg_i16_acquire:
 ; LA32:       # %bb.0:
-; LA32-NEXT:    addi.w $a2, $zero, -4
-; LA32-NEXT:    and $a2, $a0, $a2
-; LA32-NEXT:    slli.w $a0, $a0, 3
-; LA32-NEXT:    lu12i.w $a3, 15
-; LA32-NEXT:    ori $a3, $a3, 4095
-; LA32-NEXT:    sll.w $a3, $a3, $a0
+; LA32-NEXT:    lu12i.w $a2, 15
+; LA32-NEXT:    ori $a2, $a2, 4095
+; LA32-NEXT:    slli.w $a3, $a0, 3
+; LA32-NEXT:    sll.w $a2, $a2, $a3
 ; LA32-NEXT:    bstrpick.w $a1, $a1, 15, 0
-; LA32-NEXT:    sll.w $a1, $a1, $a0
+; 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:    dbar 0
-; LA32-NEXT:    ll.w $a4, $a2, 0
+; 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:    and $a5, $a5, $a2
 ; LA32-NEXT:    xor $a5, $a4, $a5
-; LA32-NEXT:    sc.w $a5, $a2, 0
+; LA32-NEXT:    sc.w $a5, $a0, 0
 ; LA32-NEXT:    beqz $a5, .LBB1_1
 ; LA32-NEXT:  # %bb.2:
-; LA32-NEXT:    srl.w $a0, $a4, $a0
+; LA32-NEXT:    srl.w $a0, $a4, $a3
 ; LA32-NEXT:    ret
 ;
 ; LA64-LABEL: atomicrmw_xchg_i16_acquire:
 ; LA64:       # %bb.0:
-; LA64-NEXT:    addi.w $a2, $zero, -4
-; LA64-NEXT:    and $a2, $a0, $a2
-; LA64-NEXT:    slli.d $a0, $a0, 3
-; LA64-NEXT:    lu12i.w $a3, 15
-; LA64-NEXT:    ori $a3, $a3, 4095
-; LA64-NEXT:    sll.w $a3, $a3, $a0
-; LA64-NEXT:    addi.w $a3, $a3, 0
+; LA64-NEXT:    lu12i.w $a2, 15
+; LA64-NEXT:    ori $a2, $a2, 4095
+; LA64-NEXT:    slli.d $a3, $a0, 3
+; LA64-NEXT:    sll.w $a2, $a2, $a3
+; LA64-NEXT:    addi.w $a2, $a2, 0
 ; LA64-NEXT:    bstrpick.d $a1, $a1, 15, 0
-; LA64-NEXT:    sll.w $a1, $a1, $a0
+; 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:    dbar 0
-; LA64-NEXT:    ll.w $a4, $a2, 0
+; 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:    and $a5, $a5, $a2
 ; LA64-NEXT:    xor $a5, $a4, $a5
-; LA64-NEXT:    sc.w $a5, $a2, 0
+; LA64-NEXT:    sc.w $a5, $a0, 0
 ; LA64-NEXT:    beqz $a5, .LBB1_1
 ; LA64-NEXT:  # %bb.2:
-; LA64-NEXT:    srl.w $a0, $a4, $a0
+; LA64-NEXT:    srl.w $a0, $a4, $a3
 ; LA64-NEXT:    ret
   %1 = atomicrmw xchg ptr %a, i16 %b acquire
   ret i16 %1
 }
 
 define i32 @atomicrmw_xchg_i32_acquire(ptr %a, i32 %b) nounwind {
 ; LA32-LABEL: atomicrmw_xchg_i32_acquire:
 ; LA32:       # %bb.0:

 ; LA64-NEXT:    ret
   %1 = atomicrmw xchg ptr %a, i64 %b acquire
   ret i64 %1
 }
 
 define i8 @atomicrmw_add_i8_acquire(ptr %a, i8 %b) nounwind {
 ; LA32-LABEL: atomicrmw_add_i8_acquire:
 ; LA32:       # %bb.0:
-; LA32-NEXT:    addi.w $a2, $zero, -4
-; LA32-NEXT:    and $a2, $a0, $a2
-; LA32-NEXT:    slli.w $a0, $a0, 3
+; LA32-NEXT:    slli.w $a2, $a0, 3
 ; LA32-NEXT:    ori $a3, $zero, 255
-; LA32-NEXT:    sll.w $a3, $a3, $a0
+; LA32-NEXT:    sll.w $a3, $a3, $a2
 ; LA32-NEXT:    andi $a1, $a1, 255
-; LA32-NEXT:    sll.w $a1, $a1, $a0
+; 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 $a4, $a2, 0
+; 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, $a2, 0
+; LA32-NEXT:    sc.w $a5, $a0, 0
 ; LA32-NEXT:    beqz $a5, .LBB4_1
 ; LA32-NEXT:  # %bb.2:
-; LA32-NEXT:    srl.w $a0, $a4, $a0
+; LA32-NEXT:    srl.w $a0, $a4, $a2
 ; LA32-NEXT:    ret
 ;
 ; LA64-LABEL: atomicrmw_add_i8_acquire:
 ; LA64:       # %bb.0:
-; LA64-NEXT:    addi.w $a2, $zero, -4
-; LA64-NEXT:    and $a2, $a0, $a2
-; LA64-NEXT:    slli.d $a0, $a0, 3
+; LA64-NEXT:    slli.d $a2, $a0, 3
 ; LA64-NEXT:    ori $a3, $zero, 255
-; LA64-NEXT:    sll.w $a3, $a3, $a0
+; LA64-NEXT:    sll.w $a3, $a3, $a2
 ; LA64-NEXT:    addi.w $a3, $a3, 0
 ; LA64-NEXT:    andi $a1, $a1, 255
-; LA64-NEXT:    sll.w $a1, $a1, $a0
+; 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 $a4, $a2, 0
+; 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, $a2, 0
+; LA64-NEXT:    sc.w $a5, $a0, 0
 ; LA64-NEXT:    beqz $a5, .LBB4_1
 ; LA64-NEXT:  # %bb.2:
-; LA64-NEXT:    srl.w $a0, $a4, $a0
+; LA64-NEXT:    srl.w $a0, $a4, $a2
 ; LA64-NEXT:    ret
   %1 = atomicrmw add ptr %a, i8 %b acquire
   ret i8 %1
 }
 
 define i16 @atomicrmw_add_i16_acquire(ptr %a, i16 %b) nounwind {
 ; LA32-LABEL: atomicrmw_add_i16_acquire:
 ; LA32:       # %bb.0:
-; LA32-NEXT:    addi.w $a2, $zero, -4
-; LA32-NEXT:    and $a2, $a0, $a2
-; LA32-NEXT:    slli.w $a0, $a0, 3
-; LA32-NEXT:    lu12i.w $a3, 15
-; LA32-NEXT:    ori $a3, $a3, 4095
-; LA32-NEXT:    sll.w $a3, $a3, $a0
+; LA32-NEXT:    lu12i.w $a2, 15
+; LA32-NEXT:    ori $a2, $a2, 4095
+; LA32-NEXT:    slli.w $a3, $a0, 3
+; LA32-NEXT:    sll.w $a2, $a2, $a3
 ; LA32-NEXT:    bstrpick.w $a1, $a1, 15, 0
-; LA32-NEXT:    sll.w $a1, $a1, $a0
+; 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:    dbar 0
-; LA32-NEXT:    ll.w $a4, $a2, 0
+; 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:    and $a5, $a5, $a2
 ; LA32-NEXT:    xor $a5, $a4, $a5
-; LA32-NEXT:    sc.w $a5, $a2, 0
+; LA32-NEXT:    sc.w $a5, $a0, 0
 ; LA32-NEXT:    beqz $a5, .LBB5_1
 ; LA32-NEXT:  # %bb.2:
-; LA32-NEXT:    srl.w $a0, $a4, $a0
+; LA32-NEXT:    srl.w $a0, $a4, $a3
 ; LA32-NEXT:    ret
 ;
 ; LA64-LABEL: atomicrmw_add_i16_acquire:
 ; LA64:       # %bb.0:
-; LA64-NEXT:    addi.w $a2, $zero, -4
-; LA64-NEXT:    and $a2, $a0, $a2
-; LA64-NEXT:    slli.d $a0, $a0, 3
-; LA64-NEXT:    lu12i.w $a3, 15
-; LA64-NEXT:    ori $a3, $a3, 4095
-; LA64-NEXT:    sll.w $a3, $a3, $a0
-; LA64-NEXT:    addi.w $a3, $a3, 0
+; LA64-NEXT:    lu12i.w $a2, 15
+; LA64-NEXT:    ori $a2, $a2, 4095
+; LA64-NEXT:    slli.d $a3, $a0, 3
+; LA64-NEXT:    sll.w $a2, $a2, $a3
+; LA64-NEXT:    addi.w $a2, $a2, 0
 ; LA64-NEXT:    bstrpick.d $a1, $a1, 15, 0
-; LA64-NEXT:    sll.w $a1, $a1, $a0
+; 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:    dbar 0
-; LA64-NEXT:    ll.w $a4, $a2, 0
+; 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:    and $a5, $a5, $a2
 ; LA64-NEXT:    xor $a5, $a4, $a5
-; LA64-NEXT:    sc.w $a5, $a2, 0
+; LA64-NEXT:    sc.w $a5, $a0, 0
 ; LA64-NEXT:    beqz $a5, .LBB5_1
 ; LA64-NEXT:  # %bb.2:
-; LA64-NEXT:    srl.w $a0, $a4, $a0
+; LA64-NEXT:    srl.w $a0, $a4, $a3
 ; LA64-NEXT:    ret
   %1 = atomicrmw add ptr %a, i16 %b acquire
   ret i16 %1
 }
 
 define i32 @atomicrmw_add_i32_acquire(ptr %a, i32 %b) nounwind {
 ; LA32-LABEL: atomicrmw_add_i32_acquire:
 ; LA32:       # %bb.0:

 ; LA64-NEXT:    ret
   %1 = atomicrmw add ptr %a, i64 %b acquire
   ret i64 %1
 }
 
 define i8 @atomicrmw_sub_i8_acquire(ptr %a, i8 %b) nounwind {
 ; LA32-LABEL: atomicrmw_sub_i8_acquire:
 ; LA32:       # %bb.0:
-; LA32-NEXT:    addi.w $a2, $zero, -4
-; LA32-NEXT:    and $a2, $a0, $a2
-; LA32-NEXT:    slli.w $a0, $a0, 3
+; LA32-NEXT:    slli.w $a2, $a0, 3
 ; LA32-NEXT:    ori $a3, $zero, 255
-; LA32-NEXT:    sll.w $a3, $a3, $a0
+; LA32-NEXT:    sll.w $a3, $a3, $a2
 ; LA32-NEXT:    andi $a1, $a1, 255
-; LA32-NEXT:    sll.w $a1, $a1, $a0
+; 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:    dbar 0
-; LA32-NEXT:    ll.w $a4, $a2, 0
+; 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, $a2, 0
+; LA32-NEXT:    sc.w $a5, $a0, 0
 ; LA32-NEXT:    beqz $a5, .LBB8_1
 ; LA32-NEXT:  # %bb.2:
-; LA32-NEXT:    srl.w $a0, $a4, $a0
+; LA32-NEXT:    srl.w $a0, $a4, $a2
 ; LA32-NEXT:    ret
 ;
 ; LA64-LABEL: atomicrmw_sub_i8_acquire:
 ; LA64:       # %bb.0:
-; LA64-NEXT:    addi.w $a2, $zero, -4
-; LA64-NEXT:    and $a2, $a0, $a2
-; LA64-NEXT:    slli.d $a0, $a0, 3
+; LA64-NEXT:    slli.d $a2, $a0, 3
 ; LA64-NEXT:    ori $a3, $zero, 255
-; LA64-NEXT:    sll.w $a3, $a3, $a0
+; LA64-NEXT:    sll.w $a3, $a3, $a2
 ; LA64-NEXT:    addi.w $a3, $a3, 0
 ; LA64-NEXT:    andi $a1, $a1, 255
-; LA64-NEXT:    sll.w $a1, $a1, $a0
+; 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:    dbar 0
-; LA64-NEXT:    ll.w $a4, $a2, 0
+; 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, $a2, 0
+; LA64-NEXT:    sc.w $a5, $a0, 0
 ; LA64-NEXT:    beqz $a5, .LBB8_1
 ; LA64-NEXT:  # %bb.2:
-; LA64-NEXT:    srl.w $a0, $a4, $a0
+; LA64-NEXT:    srl.w $a0, $a4, $a2
 ; LA64-NEXT:    ret
   %1 = atomicrmw sub ptr %a, i8 %b acquire
   ret i8 %1
 }
 
 define i16 @atomicrmw_sub_i16_acquire(ptr %a, i16 %b) nounwind {
 ; LA32-LABEL: atomicrmw_sub_i16_acquire:
 ; LA32:       # %bb.0:
-; LA32-NEXT:    addi.w $a2, $zero, -4
-; LA32-NEXT:    and $a2, $a0, $a2
-; LA32-NEXT:    slli.w $a0, $a0, 3
-; LA32-NEXT:    lu12i.w $a3, 15
-; LA32-NEXT:    ori $a3, $a3, 4095
-; LA32-NEXT:    sll.w $a3, $a3, $a0
+; LA32-NEXT:    lu12i.w $a2, 15
+; LA32-NEXT:    ori $a2, $a2, 4095
+; LA32-NEXT:    slli.w $a3, $a0, 3
+; LA32-NEXT:    sll.w $a2, $a2, $a3
 ; LA32-NEXT:    bstrpick.w $a1, $a1, 15, 0
-; LA32-NEXT:    sll.w $a1, $a1, $a0
+; 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:    dbar 0
-; LA32-NEXT:    ll.w $a4, $a2, 0
+; 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:    and $a5, $a5, $a2
 ; LA32-NEXT:    xor $a5, $a4, $a5
-; LA32-NEXT:    sc.w $a5, $a2, 0
+; LA32-NEXT:    sc.w $a5, $a0, 0
 ; LA32-NEXT:    beqz $a5, .LBB9_1
 ; LA32-NEXT:  # %bb.2:
-; LA32-NEXT:    srl.w $a0, $a4, $a0
+; LA32-NEXT:    srl.w $a0, $a4, $a3
 ; LA32-NEXT:    ret
 ;
 ; LA64-LABEL: atomicrmw_sub_i16_acquire:
 ; LA64:       # %bb.0:
-; LA64-NEXT:    addi.w $a2, $zero, -4
-; LA64-NEXT:    and $a2, $a0, $a2
-; LA64-NEXT:    slli.d $a0, $a0, 3
-; LA64-NEXT:    lu12i.w $a3, 15
-; LA64-NEXT:    ori $a3, $a3, 4095
-; LA64-NEXT:    sll.w $a3, $a3, $a0
-; LA64-NEXT:    addi.w $a3, $a3, 0
+; LA64-NEXT:    lu12i.w $a2, 15
+; LA64-NEXT:    ori $a2, $a2, 4095
+; LA64-NEXT:    slli.d $a3, $a0, 3
+; LA64-NEXT:    sll.w $a2, $a2, $a3
+; LA64-NEXT:    addi.w $a2, $a2, 0
 ; LA64-NEXT:    bstrpick.d $a1, $a1, 15, 0
-; LA64-NEXT:    sll.w $a1, $a1, $a0
+; 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:    dbar 0
-; LA64-NEXT:    ll.w $a4, $a2, 0
+; 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:    and $a5, $a5, $a2
 ; LA64-NEXT:    xor $a5, $a4, $a5
-; LA64-NEXT:    sc.w $a5, $a2, 0
+; LA64-NEXT:    sc.w $a5, $a0, 0
 ; LA64-NEXT:    beqz $a5, .LBB9_1
 ; LA64-NEXT:  # %bb.2:
-; LA64-NEXT:    srl.w $a0, $a4, $a0
+; LA64-NEXT:    srl.w $a0, $a4, $a3
 ; LA64-NEXT:    ret
   %1 = atomicrmw sub ptr %a, i16 %b acquire
   ret i16 %1
 }
 
 define i32 @atomicrmw_sub_i32_acquire(ptr %a, i32 %b) nounwind {
 ; LA32-LABEL: atomicrmw_sub_i32_acquire:
 ; LA32:       # %bb.0:

 ; LA64-NEXT:    ret
   %1 = atomicrmw sub ptr %a, i64 %b acquire
   ret i64 %1
 }
 
 define i8 @atomicrmw_nand_i8_acquire(ptr %a, i8 %b) nounwind {
 ; LA32-LABEL: atomicrmw_nand_i8_acquire:
 ; LA32:       # %bb.0:
-; LA32-NEXT:    addi.w $a2, $zero, -4
-; LA32-NEXT:    and $a2, $a0, $a2
-; LA32-NEXT:    slli.w $a0, $a0, 3
+; LA32-NEXT:    slli.w $a2, $a0, 3
 ; LA32-NEXT:    ori $a3, $zero, 255
-; LA32-NEXT:    sll.w $a3, $a3, $a0
+; LA32-NEXT:    sll.w $a3, $a3, $a2
 ; LA32-NEXT:    andi $a1, $a1, 255
-; LA32-NEXT:    sll.w $a1, $a1, $a0
+; 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:    dbar 0
-; LA32-NEXT:    ll.w $a4, $a2, 0
+; LA32-NEXT:    ll.w $a4, $a0, 0
 ; LA32-NEXT:    and $a5, $a4, $a1
 ; LA32-NEXT:    nor $a5, $a5, $zero
 ; LA32-NEXT:    xor $a5, $a4, $a5
 ; LA32-NEXT:    and $a5, $a5, $a3
 ; LA32-NEXT:    xor $a5, $a4, $a5
-; LA32-NEXT:    sc.w $a5, $a2, 0
+; LA32-NEXT:    sc.w $a5, $a0, 0
 ; LA32-NEXT:    beqz $a5, .LBB12_1
 ; LA32-NEXT:  # %bb.2:
-; LA32-NEXT:    srl.w $a0, $a4, $a0
+; LA32-NEXT:    srl.w $a0, $a4, $a2
 ; LA32-NEXT:    ret
 ;
 ; LA64-LABEL: atomicrmw_nand_i8_acquire:
 ; LA64:       # %bb.0:
-; LA64-NEXT:    addi.w $a2, $zero, -4
-; LA64-NEXT:    and $a2, $a0, $a2
-; LA64-NEXT:    slli.d $a0, $a0, 3
+; LA64-NEXT:    slli.d $a2, $a0, 3
 ; LA64-NEXT:    ori $a3, $zero, 255
-; LA64-NEXT:    sll.w $a3, $a3, $a0
+; LA64-NEXT:    sll.w $a3, $a3, $a2
 ; LA64-NEXT:    addi.w $a3, $a3, 0
 ; LA64-NEXT:    andi $a1, $a1, 255
-; LA64-NEXT:    sll.w $a1, $a1, $a0
+; 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:    dbar 0
-; LA64-NEXT:    ll.w $a4, $a2, 0
+; LA64-NEXT:    ll.w $a4, $a0, 0
 ; LA64-NEXT:    and $a5, $a4, $a1
 ; LA64-NEXT:    nor $a5, $a5, $zero
 ; LA64-NEXT:    xor $a5, $a4, $a5
 ; LA64-NEXT:    and $a5, $a5, $a3
 ; LA64-NEXT:    xor $a5, $a4, $a5
-; LA64-NEXT:    sc.w $a5, $a2, 0
+; LA64-NEXT:    sc.w $a5, $a0, 0
 ; LA64-NEXT:    beqz $a5, .LBB12_1
 ; LA64-NEXT:  # %bb.2:
-; LA64-NEXT:    srl.w $a0, $a4, $a0
+; LA64-NEXT:    srl.w $a0, $a4, $a2
 ; LA64-NEXT:    ret
   %1 = atomicrmw nand ptr %a, i8 %b acquire
   ret i8 %1
 }
 
 define i16 @atomicrmw_nand_i16_acquire(ptr %a, i16 %b) nounwind {
 ; LA32-LABEL: atomicrmw_nand_i16_acquire:
 ; LA32:       # %bb.0:
-; LA32-NEXT:    addi.w $a2, $zero, -4
-; LA32-NEXT:    and $a2, $a0, $a2
-; LA32-NEXT:    slli.w $a0, $a0, 3
-; LA32-NEXT:    lu12i.w $a3, 15
-; LA32-NEXT:    ori $a3, $a3, 4095
-; LA32-NEXT:    sll.w $a3, $a3, $a0
+; LA32-NEXT:    lu12i.w $a2, 15
+; LA32-NEXT:    ori $a2, $a2, 4095
+; LA32-NEXT:    slli.w $a3, $a0, 3
+; LA32-NEXT:    sll.w $a2, $a2, $a3
 ; LA32-NEXT:    bstrpick.w $a1, $a1, 15, 0
-; LA32-NEXT:    sll.w $a1, $a1, $a0
+; 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:    dbar 0
-; LA32-NEXT:    ll.w $a4, $a2, 0
+; LA32-NEXT:    ll.w $a4, $a0, 0
 ; LA32-NEXT:    and $a5, $a4, $a1
 ; LA32-NEXT:    nor $a5, $a5, $zero
 ; LA32-NEXT:    xor $a5, $a4, $a5
-; LA32-NEXT:    and $a5, $a5, $a3
+; LA32-NEXT:    and $a5, $a5, $a2
 ; LA32-NEXT:    xor $a5, $a4, $a5
-; LA32-NEXT:    sc.w $a5, $a2, 0
+; LA32-NEXT:    sc.w $a5, $a0, 0
 ; LA32-NEXT:    beqz $a5, .LBB13_1
 ; LA32-NEXT:  # %bb.2:
-; LA32-NEXT:    srl.w $a0, $a4, $a0
+; LA32-NEXT:    srl.w $a0, $a4, $a3
 ; LA32-NEXT:    ret
 ;
 ; LA64-LABEL: atomicrmw_nand_i16_acquire:
 ; LA64:       # %bb.0:
-; LA64-NEXT:    addi.w $a2, $zero, -4
-; LA64-NEXT:    and $a2, $a0, $a2
-; LA64-NEXT:    slli.d $a0, $a0, 3
-; LA64-NEXT:    lu12i.w $a3, 15
-; LA64-NEXT:    ori $a3, $a3, 4095
-; LA64-NEXT:    sll.w $a3, $a3, $a0
-; LA64-NEXT:    addi.w $a3, $a3, 0
+; LA64-NEXT:    lu12i.w $a2, 15
+; LA64-NEXT:    ori $a2, $a2, 4095
+; LA64-NEXT:    slli.d $a3, $a0, 3
+; LA64-NEXT:    sll.w $a2, $a2, $a3
+; LA64-NEXT:    addi.w $a2, $a2, 0
 ; LA64-NEXT:    bstrpick.d $a1, $a1, 15, 0
-; LA64-NEXT:    sll.w $a1, $a1, $a0
+; 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:    dbar 0
-; LA64-NEXT:    ll.w $a4, $a2, 0
+; LA64-NEXT:    ll.w $a4, $a0, 0
 ; LA64-NEXT:    and $a5, $a4, $a1
 ; LA64-NEXT:    nor $a5, $a5, $zero
 ; LA64-NEXT:    xor $a5, $a4, $a5
-; LA64-NEXT:    and $a5, $a5, $a3
+; LA64-NEXT:    and $a5, $a5, $a2
 ; LA64-NEXT:    xor $a5, $a4, $a5
-; LA64-NEXT:    sc.w $a5, $a2, 0
+; LA64-NEXT:    sc.w $a5, $a0, 0
 ; LA64-NEXT:    beqz $a5, .LBB13_1
 ; LA64-NEXT:  # %bb.2:
-; LA64-NEXT:    srl.w $a0, $a4, $a0
+; LA64-NEXT:    srl.w $a0, $a4, $a3
 ; LA64-NEXT:    ret
   %1 = atomicrmw nand ptr %a, i16 %b acquire
   ret i16 %1
 }
 
 define i32 @atomicrmw_nand_i32_acquire(ptr %a, i32 %b) nounwind {
 ; LA32-LABEL: atomicrmw_nand_i32_acquire:
 ; LA32:       # %bb.0:

 ; LA32-LABEL: atomicrmw_and_i8_acquire:
 ; LA32:       # %bb.0:
 ; LA32-NEXT:    slli.w $a2, $a0, 3
 ; LA32-NEXT:    ori $a3, $zero, 255
 ; LA32-NEXT:    sll.w $a3, $a3, $a2
 ; LA32-NEXT:    andi $a1, $a1, 255
 ; LA32-NEXT:    sll.w $a1, $a1, $a2
 ; LA32-NEXT:    orn $a1, $a1, $a3
-; LA32-NEXT:    addi.w $a3, $zero, -4
-; LA32-NEXT:    and $a0, $a0, $a3
+; LA32-NEXT:    bstrins.w $a0, $zero, 1, 0
 ; LA32-NEXT:  .LBB16_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:  # %bb.2:
 ; LA32-NEXT:    srl.w $a0, $a3, $a2
 ; LA32-NEXT:    ret
 ;
 ; LA64-LABEL: atomicrmw_and_i8_acquire:
 ; LA64:       # %bb.0:
 ; LA64-NEXT:    slli.d $a2, $a0, 3
 ; LA64-NEXT:    ori $a3, $zero, 255
 ; LA64-NEXT:    sll.w $a3, $a3, $a2
 ; LA64-NEXT:    andi $a1, $a1, 255
 ; LA64-NEXT:    sll.w $a1, $a1, $a2
 ; LA64-NEXT:    orn $a1, $a1, $a3
-; LA64-NEXT:    addi.w $a3, $zero, -4
-; LA64-NEXT:    and $a0, $a0, $a3
+; LA64-NEXT:    bstrins.d $a0, $zero, 1, 0
 ; LA64-NEXT:    amand_db.w $a3, $a1, $a0
 ; LA64-NEXT:    srl.w $a0, $a3, $a2
 ; LA64-NEXT:    ret
   %1 = atomicrmw and ptr %a, i8 %b acquire
   ret i8 %1
 }
 
 define i16 @atomicrmw_and_i16_acquire(ptr %a, i16 %b) nounwind {
 ; LA32-LABEL: atomicrmw_and_i16_acquire:
 ; LA32:       # %bb.0:
 ; LA32-NEXT:    lu12i.w $a2, 15
 ; LA32-NEXT:    ori $a2, $a2, 4095
 ; LA32-NEXT:    slli.w $a3, $a0, 3
 ; LA32-NEXT:    sll.w $a2, $a2, $a3
 ; LA32-NEXT:    bstrpick.w $a1, $a1, 15, 0
 ; LA32-NEXT:    sll.w $a1, $a1, $a3
 ; LA32-NEXT:    orn $a1, $a1, $a2
-; LA32-NEXT:    addi.w $a2, $zero, -4
-; LA32-NEXT:    and $a0, $a0, $a2
+; LA32-NEXT:    bstrins.w $a0, $zero, 1, 0
 ; LA32-NEXT:  .LBB17_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:  # %bb.2:
 ; LA32-NEXT:    srl.w $a0, $a2, $a3
 ; LA32-NEXT:    ret
 ;
 ; LA64-LABEL: atomicrmw_and_i16_acquire:
 ; LA64:       # %bb.0:
 ; LA64-NEXT:    lu12i.w $a2, 15
 ; LA64-NEXT:    ori $a2, $a2, 4095
 ; LA64-NEXT:    slli.d $a3, $a0, 3
 ; LA64-NEXT:    sll.w $a2, $a2, $a3
 ; LA64-NEXT:    bstrpick.d $a1, $a1, 15, 0
 ; LA64-NEXT:    sll.w $a1, $a1, $a3
 ; LA64-NEXT:    orn $a1, $a1, $a2
-; LA64-NEXT:    addi.w $a2, $zero, -4
-; LA64-NEXT:    and $a0, $a0, $a2
+; LA64-NEXT:    bstrins.d $a0, $zero, 1, 0
 ; LA64-NEXT:    amand_db.w $a2, $a1, $a0
 ; LA64-NEXT:    srl.w $a0, $a2, $a3
 ; LA64-NEXT:    ret
   %1 = atomicrmw and ptr %a, i16 %b acquire
   ret i16 %1
 }
 
 define i32 @atomicrmw_and_i32_acquire(ptr %a, i32 %b) nounwind {

 ; LA64-NEXT:    ret
   %1 = atomicrmw and ptr %a, i64 %b acquire
   ret i64 %1
 }
 
 define i8 @atomicrmw_or_i8_acquire(ptr %a, i8 %b) nounwind {
 ; LA32-LABEL: atomicrmw_or_i8_acquire:
 ; LA32:       # %bb.0:
-; LA32-NEXT:    addi.w $a2, $zero, -4
-; LA32-NEXT:    and $a2, $a0, $a2
-; LA32-NEXT:    slli.w $a0, $a0, 3
 ; LA32-NEXT:    andi $a1, $a1, 255
-; LA32-NEXT:    sll.w $a1, $a1, $a0
+; 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:    dbar 0
-; LA32-NEXT:    ll.w $a3, $a2, 0
+; LA32-NEXT:    ll.w $a3, $a0, 0
 ; LA32-NEXT:    or $a4, $a3, $a1
-; LA32-NEXT:    sc.w $a4, $a2, 0
+; LA32-NEXT:    sc.w $a4, $a0, 0
 ; LA32-NEXT:    beqz $a4, .LBB20_1
 ; LA32-NEXT:  # %bb.2:
-; LA32-NEXT:    srl.w $a0, $a3, $a0
+; LA32-NEXT:    srl.w $a0, $a3, $a2
 ; LA32-NEXT:    ret
 ;
 ; LA64-LABEL: atomicrmw_or_i8_acquire:
 ; LA64:       # %bb.0:
-; LA64-NEXT:    addi.w $a2, $zero, -4
-; LA64-NEXT:    and $a2, $a0, $a2
-; LA64-NEXT:    slli.d $a0, $a0, 3
 ; LA64-NEXT:    andi $a1, $a1, 255
-; LA64-NEXT:    sll.w $a1, $a1, $a0
-; LA64-NEXT:    amor_db.w $a3, $a1, $a2
-; LA64-NEXT:    srl.w $a0, $a3, $a0
+; LA64-NEXT:    slli.d $a2, $a0, 3
+; LA64-NEXT:    sll.w $a1, $a1, $a2
+; LA64-NEXT:    bstrins.d $a0, $zero, 1, 0
+; LA64-NEXT:    amor_db.w $a3, $a1, $a0
+; LA64-NEXT:    srl.w $a0, $a3, $a2
 ; LA64-NEXT:    ret
   %1 = atomicrmw or ptr %a, i8 %b acquire
   ret i8 %1
 }
 
 define i16 @atomicrmw_or_i16_acquire(ptr %a, i16 %b) nounwind {
 ; LA32-LABEL: atomicrmw_or_i16_acquire:
 ; LA32:       # %bb.0:
-; LA32-NEXT:    addi.w $a2, $zero, -4
-; LA32-NEXT:    and $a2, $a0, $a2
-; LA32-NEXT:    slli.w $a0, $a0, 3
 ; LA32-NEXT:    bstrpick.w $a1, $a1, 15, 0
-; LA32-NEXT:    sll.w $a1, $a1, $a0
+; 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:    dbar 0
-; LA32-NEXT:    ll.w $a3, $a2, 0
+; LA32-NEXT:    ll.w $a3, $a0, 0
 ; LA32-NEXT:    or $a4, $a3, $a1
-; LA32-NEXT:    sc.w $a4, $a2, 0
+; LA32-NEXT:    sc.w $a4, $a0, 0
 ; LA32-NEXT:    beqz $a4, .LBB21_1
 ; LA32-NEXT:  # %bb.2:
-; LA32-NEXT:    srl.w $a0, $a3, $a0
+; LA32-NEXT:    srl.w $a0, $a3, $a2
 ; LA32-NEXT:    ret
 ;
 ; LA64-LABEL: atomicrmw_or_i16_acquire:
 ; LA64:       # %bb.0:
-; LA64-NEXT:    addi.w $a2, $zero, -4
-; LA64-NEXT:    and $a2, $a0, $a2
-; LA64-NEXT:    slli.d $a0, $a0, 3
 ; LA64-NEXT:    bstrpick.d $a1, $a1, 15, 0
-; LA64-NEXT:    sll.w $a1, $a1, $a0
-; LA64-NEXT:    amor_db.w $a3, $a1, $a2
-; LA64-NEXT:    srl.w $a0, $a3, $a0
+; LA64-NEXT:    slli.d $a2, $a0, 3
+; LA64-NEXT:    sll.w $a1, $a1, $a2
+; LA64-NEXT:    bstrins.d $a0, $zero, 1, 0
+; LA64-NEXT:    amor_db.w $a3, $a1, $a0
+; LA64-NEXT:    srl.w $a0, $a3, $a2
 ; LA64-NEXT:    ret
   %1 = atomicrmw or ptr %a, i16 %b acquire
   ret i16 %1
 }
 
 define i32 @atomicrmw_or_i32_acquire(ptr %a, i32 %b) nounwind {
 ; LA32-LABEL: atomicrmw_or_i32_acquire:
 ; LA32:       # %bb.0:

 ; LA64-NEXT:    ret
   %1 = atomicrmw or ptr %a, i64 %b acquire
   ret i64 %1
 }
 
 define i8 @atomicrmw_xor_i8_acquire(ptr %a, i8 %b) nounwind {
 ; LA32-LABEL: atomicrmw_xor_i8_acquire:
 ; LA32:       # %bb.0:
-; LA32-NEXT:    addi.w $a2, $zero, -4
-; LA32-NEXT:    and $a2, $a0, $a2
-; LA32-NEXT:    slli.w $a0, $a0, 3
 ; LA32-NEXT:    andi $a1, $a1, 255
-; LA32-NEXT:    sll.w $a1, $a1, $a0
+; 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:    dbar 0
-; LA32-NEXT:    ll.w $a3, $a2, 0
+; LA32-NEXT:    ll.w $a3, $a0, 0
 ; LA32-NEXT:    xor $a4, $a3, $a1
-; LA32-NEXT:    sc.w $a4, $a2, 0
+; LA32-NEXT:    sc.w $a4, $a0, 0
 ; LA32-NEXT:    beqz $a4, .LBB24_1
 ; LA32-NEXT:  # %bb.2:
-; LA32-NEXT:    srl.w $a0, $a3, $a0
+; LA32-NEXT:    srl.w $a0, $a3, $a2
 ; LA32-NEXT:    ret
 ;
 ; LA64-LABEL: atomicrmw_xor_i8_acquire:
 ; LA64:       # %bb.0:
-; LA64-NEXT:    addi.w $a2, $zero, -4
-; LA64-NEXT:    and $a2, $a0, $a2
-; LA64-NEXT:    slli.d $a0, $a0, 3
 ; LA64-NEXT:    andi $a1, $a1, 255
-; LA64-NEXT:    sll.w $a1, $a1, $a0
-; LA64-NEXT:    amxor_db.w $a3, $a1, $a2
-; LA64-NEXT:    srl.w $a0, $a3, $a0
+; LA64-NEXT:    slli.d $a2, $a0, 3
+; LA64-NEXT:    sll.w $a1, $a1, $a2
+; LA64-NEXT:    bstrins.d $a0, $zero, 1, 0
+; LA64-NEXT:    amxor_db.w $a3, $a1, $a0
+; LA64-NEXT:    srl.w $a0, $a3, $a2
 ; LA64-NEXT:    ret
   %1 = atomicrmw xor ptr %a, i8 %b acquire
   ret i8 %1
 }
 
 define i16 @atomicrmw_xor_i16_acquire(ptr %a, i16 %b) nounwind {
 ; LA32-LABEL: atomicrmw_xor_i16_acquire:
 ; LA32:       # %bb.0:
-; LA32-NEXT:    addi.w $a2, $zero, -4
-; LA32-NEXT:    and $a2, $a0, $a2
-; LA32-NEXT:    slli.w $a0, $a0, 3
 ; LA32-NEXT:    bstrpick.w $a1, $a1, 15, 0
-; LA32-NEXT:    sll.w $a1, $a1, $a0
+; 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:    dbar 0
-; LA32-NEXT:    ll.w $a3, $a2, 0
+; LA32-NEXT:    ll.w $a3, $a0, 0
 ; LA32-NEXT:    xor $a4, $a3, $a1
-; LA32-NEXT:    sc.w $a4, $a2, 0
+; LA32-NEXT:    sc.w $a4, $a0, 0
 ; LA32-NEXT:    beqz $a4, .LBB25_1
 ; LA32-NEXT:  # %bb.2:
-; LA32-NEXT:    srl.w $a0, $a3, $a0
+; LA32-NEXT:    srl.w $a0, $a3, $a2
 ; LA32-NEXT:    ret
 ;
 ; LA64-LABEL: atomicrmw_xor_i16_acquire:
 ; LA64:       # %bb.0:
-; LA64-NEXT:    addi.w $a2, $zero, -4
-; LA64-NEXT:    and $a2, $a0, $a2
-; LA64-NEXT:    slli.d $a0, $a0, 3
 ; LA64-NEXT:    bstrpick.d $a1, $a1, 15, 0
-; LA64-NEXT:    sll.w $a1, $a1, $a0
-; LA64-NEXT:    amxor_db.w $a3, $a1, $a2
-; LA64-NEXT:    srl.w $a0, $a3, $a0
+; LA64-NEXT:    slli.d $a2, $a0, 3
+; LA64-NEXT:    sll.w $a1, $a1, $a2
+; LA64-NEXT:    bstrins.d $a0, $zero, 1, 0
+; LA64-NEXT:    amxor_db.w $a3, $a1, $a0
+; LA64-NEXT:    srl.w $a0, $a3, $a2
 ; LA64-NEXT:    ret
   %1 = atomicrmw xor ptr %a, i16 %b acquire
   ret i16 %1
 }
 
 define i32 @atomicrmw_xor_i32_acquire(ptr %a, i32 %b) nounwind {
 ; LA32-LABEL: atomicrmw_xor_i32_acquire:
 ; LA32:       # %bb.0:

 ; LA64-NEXT:    ret
   %1 = atomicrmw xor ptr %a, i64 %b acquire
   ret i64 %1
 }
 
 define i8 @atomicrmw_xchg_i8_monotonic(ptr %a, i8 %b) nounwind {
 ; LA32-LABEL: atomicrmw_xchg_i8_monotonic:
 ; LA32:       # %bb.0:
-; LA32-NEXT:    addi.w $a2, $zero, -4
-; LA32-NEXT:    and $a2, $a0, $a2
-; LA32-NEXT:    slli.w $a0, $a0, 3
+; LA32-NEXT:    slli.w $a2, $a0, 3
 ; LA32-NEXT:    ori $a3, $zero, 255
-; LA32-NEXT:    sll.w $a3, $a3, $a0
+; LA32-NEXT:    sll.w $a3, $a3, $a2
 ; LA32-NEXT:    andi $a1, $a1, 255
-; LA32-NEXT:    sll.w $a1, $a1, $a0
+; 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:    ll.w $a4, $a2, 0
+; 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, $a2, 0
+; LA32-NEXT:    sc.w $a5, $a0, 0
 ; LA32-NEXT:    beqz $a5, .LBB28_1
 ; LA32-NEXT:  # %bb.2:
-; LA32-NEXT:    srl.w $a0, $a4, $a0
+; LA32-NEXT:    srl.w $a0, $a4, $a2
 ; LA32-NEXT:    ret
 ;
 ; LA64-LABEL: atomicrmw_xchg_i8_monotonic:
 ; LA64:       # %bb.0:
-; LA64-NEXT:    addi.w $a2, $zero, -4
-; LA64-NEXT:    and $a2, $a0, $a2
-; LA64-NEXT:    slli.d $a0, $a0, 3
+; LA64-NEXT:    slli.d $a2, $a0, 3
 ; LA64-NEXT:    ori $a3, $zero, 255
-; LA64-NEXT:    sll.w $a3, $a3, $a0
+; LA64-NEXT:    sll.w $a3, $a3, $a2
 ; LA64-NEXT:    addi.w $a3, $a3, 0
 ; LA64-NEXT:    andi $a1, $a1, 255
-; LA64-NEXT:    sll.w $a1, $a1, $a0
+; 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:    ll.w $a4, $a2, 0
+; 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, $a2, 0
+; LA64-NEXT:    sc.w $a5, $a0, 0
 ; LA64-NEXT:    beqz $a5, .LBB28_1
 ; LA64-NEXT:  # %bb.2:
-; LA64-NEXT:    srl.w $a0, $a4, $a0
+; LA64-NEXT:    srl.w $a0, $a4, $a2
 ; LA64-NEXT:    ret
   %1 = atomicrmw xchg ptr %a, i8 %b monotonic
   ret i8 %1
 }
 
 define i16 @atomicrmw_xchg_i16_monotonic(ptr %a, i16 %b) nounwind {
 ; LA32-LABEL: atomicrmw_xchg_i16_monotonic:
 ; LA32:       # %bb.0:
-; LA32-NEXT:    addi.w $a2, $zero, -4
-; LA32-NEXT:    and $a2, $a0, $a2
-; LA32-NEXT:    slli.w $a0, $a0, 3
-; LA32-NEXT:    lu12i.w $a3, 15
-; LA32-NEXT:    ori $a3, $a3, 4095
-; LA32-NEXT:    sll.w $a3, $a3, $a0
+; LA32-NEXT:    lu12i.w $a2, 15
+; LA32-NEXT:    ori $a2, $a2, 4095
+; LA32-NEXT:    slli.w $a3, $a0, 3
+; LA32-NEXT:    sll.w $a2, $a2, $a3
 ; LA32-NEXT:    bstrpick.w $a1, $a1, 15, 0
-; LA32-NEXT:    sll.w $a1, $a1, $a0
+; 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:    ll.w $a4, $a2, 0
+; 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:    and $a5, $a5, $a2
 ; LA32-NEXT:    xor $a5, $a4, $a5
-; LA32-NEXT:    sc.w $a5, $a2, 0
+; LA32-NEXT:    sc.w $a5, $a0, 0
 ; LA32-NEXT:    beqz $a5, .LBB29_1
 ; LA32-NEXT:  # %bb.2:
-; LA32-NEXT:    srl.w $a0, $a4, $a0
+; LA32-NEXT:    srl.w $a0, $a4, $a3
 ; LA32-NEXT:    ret
 ;
 ; LA64-LABEL: atomicrmw_xchg_i16_monotonic:
 ; LA64:       # %bb.0:
-; LA64-NEXT:    addi.w $a2, $zero, -4
-; LA64-NEXT:    and $a2, $a0, $a2
-; LA64-NEXT:    slli.d $a0, $a0, 3
-; LA64-NEXT:    lu12i.w $a3, 15
-; LA64-NEXT:    ori $a3, $a3, 4095
-; LA64-NEXT:    sll.w $a3, $a3, $a0
-; LA64-NEXT:    addi.w $a3, $a3, 0
+; LA64-NEXT:    lu12i.w $a2, 15
+; LA64-NEXT:    ori $a2, $a2, 4095
+; LA64-NEXT:    slli.d $a3, $a0, 3
+; LA64-NEXT:    sll.w $a2, $a2, $a3
+; LA64-NEXT:    addi.w $a2, $a2, 0
 ; LA64-NEXT:    bstrpick.d $a1, $a1, 15, 0
-; LA64-NEXT:    sll.w $a1, $a1, $a0
+; 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:    ll.w $a4, $a2, 0
+; 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:    and $a5, $a5, $a2
 ; LA64-NEXT:    xor $a5, $a4, $a5
-; LA64-NEXT:    sc.w $a5, $a2, 0
+; LA64-NEXT:    sc.w $a5, $a0, 0
 ; LA64-NEXT:    beqz $a5, .LBB29_1
 ; LA64-NEXT:  # %bb.2:
-; LA64-NEXT:    srl.w $a0, $a4, $a0
+; LA64-NEXT:    srl.w $a0, $a4, $a3
 ; LA64-NEXT:    ret
   %1 = atomicrmw xchg ptr %a, i16 %b monotonic
   ret i16 %1
 }
 
 define i32 @atomicrmw_xchg_i32_monotonic(ptr %a, i32 %b) nounwind {
 ; LA32-LABEL: atomicrmw_xchg_i32_monotonic:
 ; LA32:       # %bb.0:

 ; LA64-NEXT:    ret
   %1 = atomicrmw xchg ptr %a, i64 %b monotonic
   ret i64 %1
 }
 
 define i8 @atomicrmw_add_i8_monotonic(ptr %a, i8 %b) nounwind {
 ; LA32-LABEL: atomicrmw_add_i8_monotonic:
 ; LA32:       # %bb.0:
-; LA32-NEXT:    addi.w $a2, $zero, -4
-; LA32-NEXT:    and $a2, $a0, $a2
-; LA32-NEXT:    slli.w $a0, $a0, 3
+; LA32-NEXT:    slli.w $a2, $a0, 3
 ; LA32-NEXT:    ori $a3, $zero, 255
-; LA32-NEXT:    sll.w $a3, $a3, $a0
+; LA32-NEXT:    sll.w $a3, $a3, $a2
 ; LA32-NEXT:    andi $a1, $a1, 255
-; LA32-NEXT:    sll.w $a1, $a1, $a0
+; 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:    ll.w $a4, $a2, 0
+; 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, $a2, 0
+; LA32-NEXT:    sc.w $a5, $a0, 0
 ; LA32-NEXT:    beqz $a5, .LBB32_1
 ; LA32-NEXT:  # %bb.2:
-; LA32-NEXT:    srl.w $a0, $a4, $a0
+; LA32-NEXT:    srl.w $a0, $a4, $a2
 ; LA32-NEXT:    ret
 ;
 ; LA64-LABEL: atomicrmw_add_i8_monotonic:
 ; LA64:       # %bb.0:
-; LA64-NEXT:    addi.w $a2, $zero, -4
-; LA64-NEXT:    and $a2, $a0, $a2
-; LA64-NEXT:    slli.d $a0, $a0, 3
+; LA64-NEXT:    slli.d $a2, $a0, 3
 ; LA64-NEXT:    ori $a3, $zero, 255
-; LA64-NEXT:    sll.w $a3, $a3, $a0
+; LA64-NEXT:    sll.w $a3, $a3, $a2
 ; LA64-NEXT:    addi.w $a3, $a3, 0
 ; LA64-NEXT:    andi $a1, $a1, 255
-; LA64-NEXT:    sll.w $a1, $a1, $a0
+; 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:    ll.w $a4, $a2, 0
+; 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, $a2, 0
+; LA64-NEXT:    sc.w $a5, $a0, 0
 ; LA64-NEXT:    beqz $a5, .LBB32_1
 ; LA64-NEXT:  # %bb.2:
-; LA64-NEXT:    srl.w $a0, $a4, $a0
+; LA64-NEXT:    srl.w $a0, $a4, $a2
 ; LA64-NEXT:    ret
   %1 = atomicrmw add ptr %a, i8 %b monotonic
   ret i8 %1
 }
 
 define i16 @atomicrmw_add_i16_monotonic(ptr %a, i16 %b) nounwind {
 ; LA32-LABEL: atomicrmw_add_i16_monotonic:
 ; LA32:       # %bb.0:
-; LA32-NEXT:    addi.w $a2, $zero, -4
-; LA32-NEXT:    and $a2, $a0, $a2
-; LA32-NEXT:    slli.w $a0, $a0, 3
-; LA32-NEXT:    lu12i.w $a3, 15
-; LA32-NEXT:    ori $a3, $a3, 4095
-; LA32-NEXT:    sll.w $a3, $a3, $a0
+; LA32-NEXT:    lu12i.w $a2, 15
+; LA32-NEXT:    ori $a2, $a2, 4095
+; LA32-NEXT:    slli.w $a3, $a0, 3
+; LA32-NEXT:    sll.w $a2, $a2, $a3
 ; LA32-NEXT:    bstrpick.w $a1, $a1, 15, 0
-; LA32-NEXT:    sll.w $a1, $a1, $a0
+; 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:    ll.w $a4, $a2, 0
+; 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:    and $a5, $a5, $a2
 ; LA32-NEXT:    xor $a5, $a4, $a5
-; LA32-NEXT:    sc.w $a5, $a2, 0
+; LA32-NEXT:    sc.w $a5, $a0, 0
 ; LA32-NEXT:    beqz $a5, .LBB33_1
 ; LA32-NEXT:  # %bb.2:
-; LA32-NEXT:    srl.w $a0, $a4, $a0
+; LA32-NEXT:    srl.w $a0, $a4, $a3
 ; LA32-NEXT:    ret
 ;
 ; LA64-LABEL: atomicrmw_add_i16_monotonic:
 ; LA64:       # %bb.0:
-; LA64-NEXT:    addi.w $a2, $zero, -4
-; LA64-NEXT:    and $a2, $a0, $a2
-; LA64-NEXT:    slli.d $a0, $a0, 3
-; LA64-NEXT:    lu12i.w $a3, 15
-; LA64-NEXT:    ori $a3, $a3, 4095
-; LA64-NEXT:    sll.w $a3, $a3, $a0
-; LA64-NEXT:    addi.w $a3, $a3, 0
+; LA64-NEXT:    lu12i.w $a2, 15
+; LA64-NEXT:    ori $a2, $a2, 4095
+; LA64-NEXT:    slli.d $a3, $a0, 3
+; LA64-NEXT:    sll.w $a2, $a2, $a3
+; LA64-NEXT:    addi.w $a2, $a2, 0
 ; LA64-NEXT:    bstrpick.d $a1, $a1, 15, 0
-; LA64-NEXT:    sll.w $a1, $a1, $a0
+; 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:    ll.w $a4, $a2, 0
+; 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:    and $a5, $a5, $a2
 ; LA64-NEXT:    xor $a5, $a4, $a5
-; LA64-NEXT:    sc.w $a5, $a2, 0
+; LA64-NEXT:    sc.w $a5, $a0, 0
 ; LA64-NEXT:    beqz $a5, .LBB33_1
 ; LA64-NEXT:  # %bb.2:
-; LA64-NEXT:    srl.w $a0, $a4, $a0
+; LA64-NEXT:    srl.w $a0, $a4, $a3
 ; LA64-NEXT:    ret
   %1 = atomicrmw add ptr %a, i16 %b monotonic
   ret i16 %1
 }
 
 define i32 @atomicrmw_add_i32_monotonic(ptr %a, i32 %b) nounwind {
 ; LA32-LABEL: atomicrmw_add_i32_monotonic:
 ; LA32:       # %bb.0:

 ; LA64-NEXT:    ret
   %1 = atomicrmw add ptr %a, i64 %b monotonic
   ret i64 %1
 }
 
 define i8 @atomicrmw_sub_i8_monotonic(ptr %a, i8 %b) nounwind {
 ; LA32-LABEL: atomicrmw_sub_i8_monotonic:
 ; LA32:       # %bb.0:
-; LA32-NEXT:    addi.w $a2, $zero, -4
-; LA32-NEXT:    and $a2, $a0, $a2
-; LA32-NEXT:    slli.w $a0, $a0, 3
+; LA32-NEXT:    slli.w $a2, $a0, 3
 ; LA32-NEXT:    ori $a3, $zero, 255
-; LA32-NEXT:    sll.w $a3, $a3, $a0
+; LA32-NEXT:    sll.w $a3, $a3, $a2
 ; LA32-NEXT:    andi $a1, $a1, 255
-; LA32-NEXT:    sll.w $a1, $a1, $a0
+; 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 $a4, $a2, 0
+; 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, $a2, 0
+; LA32-NEXT:    sc.w $a5, $a0, 0
 ; LA32-NEXT:    beqz $a5, .LBB36_1
 ; LA32-NEXT:  # %bb.2:
-; LA32-NEXT:    srl.w $a0, $a4, $a0
+; LA32-NEXT:    srl.w $a0, $a4, $a2
 ; LA32-NEXT:    ret
 ;
 ; LA64-LABEL: atomicrmw_sub_i8_monotonic:
 ; LA64:       # %bb.0:
-; LA64-NEXT:    addi.w $a2, $zero, -4
-; LA64-NEXT:    and $a2, $a0, $a2
-; LA64-NEXT:    slli.d $a0, $a0, 3
+; LA64-NEXT:    slli.d $a2, $a0, 3
 ; LA64-NEXT:    ori $a3, $zero, 255
-; LA64-NEXT:    sll.w $a3, $a3, $a0
+; LA64-NEXT:    sll.w $a3, $a3, $a2
 ; LA64-NEXT:    addi.w $a3, $a3, 0
 ; LA64-NEXT:    andi $a1, $a1, 255
-; LA64-NEXT:    sll.w $a1, $a1, $a0
+; 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 $a4, $a2, 0
+; 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, $a2, 0
+; LA64-NEXT:    sc.w $a5, $a0, 0
 ; LA64-NEXT:    beqz $a5, .LBB36_1
 ; LA64-NEXT:  # %bb.2:
-; LA64-NEXT:    srl.w $a0, $a4, $a0
+; LA64-NEXT:    srl.w $a0, $a4, $a2
 ; LA64-NEXT:    ret
   %1 = atomicrmw sub ptr %a, i8 %b monotonic
   ret i8 %1
 }
 
 define i16 @atomicrmw_sub_i16_monotonic(ptr %a, i16 %b) nounwind {
 ; LA32-LABEL: atomicrmw_sub_i16_monotonic:
 ; LA32:       # %bb.0:
-; LA32-NEXT:    addi.w $a2, $zero, -4
-; LA32-NEXT:    and $a2, $a0, $a2
-; LA32-NEXT:    slli.w $a0, $a0, 3
-; LA32-NEXT:    lu12i.w $a3, 15
-; LA32-NEXT:    ori $a3, $a3, 4095
-; LA32-NEXT:    sll.w $a3, $a3, $a0
+; LA32-NEXT:    lu12i.w $a2, 15
+; LA32-NEXT:    ori $a2, $a2, 4095
+; LA32-NEXT:    slli.w $a3, $a0, 3
+; LA32-NEXT:    sll.w $a2, $a2, $a3
 ; LA32-NEXT:    bstrpick.w $a1, $a1, 15, 0
-; LA32-NEXT:    sll.w $a1, $a1, $a0
+; 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:    ll.w $a4, $a2, 0
+; 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:    and $a5, $a5, $a2
 ; LA32-NEXT:    xor $a5, $a4, $a5
-; LA32-NEXT:    sc.w $a5, $a2, 0
+; LA32-NEXT:    sc.w $a5, $a0, 0
 ; LA32-NEXT:    beqz $a5, .LBB37_1
 ; LA32-NEXT:  # %bb.2:
-; LA32-NEXT:    srl.w $a0, $a4, $a0
+; LA32-NEXT:    srl.w $a0, $a4, $a3
 ; LA32-NEXT:    ret
 ;
 ; LA64-LABEL: atomicrmw_sub_i16_monotonic:
 ; LA64:       # %bb.0:
-; LA64-NEXT:    addi.w $a2, $zero, -4
-; LA64-NEXT:    and $a2, $a0, $a2
-; LA64-NEXT:    slli.d $a0, $a0, 3
-; LA64-NEXT:    lu12i.w $a3, 15
-; LA64-NEXT:    ori $a3, $a3, 4095
-; LA64-NEXT:    sll.w $a3, $a3, $a0
-; LA64-NEXT:    addi.w $a3, $a3, 0
+; LA64-NEXT:    lu12i.w $a2, 15
+; LA64-NEXT:    ori $a2, $a2, 4095
+; LA64-NEXT:    slli.d $a3, $a0, 3
+; LA64-NEXT:    sll.w $a2, $a2, $a3
+; LA64-NEXT:    addi.w $a2, $a2, 0
 ; LA64-NEXT:    bstrpick.d $a1, $a1, 15, 0
-; LA64-NEXT:    sll.w $a1, $a1, $a0
+; 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:    ll.w $a4, $a2, 0
+; 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:    and $a5, $a5, $a2
 ; LA64-NEXT:    xor $a5, $a4, $a5
-; LA64-NEXT:    sc.w $a5, $a2, 0
+; LA64-NEXT:    sc.w $a5, $a0, 0
 ; LA64-NEXT:    beqz $a5, .LBB37_1
 ; LA64-NEXT:  # %bb.2:
-; LA64-NEXT:    srl.w $a0, $a4, $a0
+; LA64-NEXT:    srl.w $a0, $a4, $a3
 ; LA64-NEXT:    ret
   %1 = atomicrmw sub ptr %a, i16 %b monotonic
   ret i16 %1
 }
 
 define i32 @atomicrmw_sub_i32_monotonic(ptr %a, i32 %b) nounwind {
 ; LA32-LABEL: atomicrmw_sub_i32_monotonic:
 ; LA32:       # %bb.0:

 ; LA64-NEXT:    ret
   %1 = atomicrmw sub ptr %a, i64 %b monotonic
   ret i64 %1
 }
 
 define i8 @atomicrmw_nand_i8_monotonic(ptr %a, i8 %b) nounwind {
 ; LA32-LABEL: atomicrmw_nand_i8_monotonic:
 ; LA32:       # %bb.0:
-; LA32-NEXT:    addi.w $a2, $zero, -4
-; LA32-NEXT:    and $a2, $a0, $a2
-; LA32-NEXT:    slli.w $a0, $a0, 3
+; LA32-NEXT:    slli.w $a2, $a0, 3
 ; LA32-NEXT:    ori $a3, $zero, 255
-; LA32-NEXT:    sll.w $a3, $a3, $a0
+; LA32-NEXT:    sll.w $a3, $a3, $a2
 ; LA32-NEXT:    andi $a1, $a1, 255
-; LA32-NEXT:    sll.w $a1, $a1, $a0
+; 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:    ll.w $a4, $a2, 0
+; LA32-NEXT:    ll.w $a4, $a0, 0
 ; LA32-NEXT:    and $a5, $a4, $a1
 ; LA32-NEXT:    nor $a5, $a5, $zero
 ; LA32-NEXT:    xor $a5, $a4, $a5
 ; LA32-NEXT:    and $a5, $a5, $a3
 ; LA32-NEXT:    xor $a5, $a4, $a5
-; LA32-NEXT:    sc.w $a5, $a2, 0
+; LA32-NEXT:    sc.w $a5, $a0, 0
 ; LA32-NEXT:    beqz $a5, .LBB40_1
 ; LA32-NEXT:  # %bb.2:
-; LA32-NEXT:    srl.w $a0, $a4, $a0
+; LA32-NEXT:    srl.w $a0, $a4, $a2
 ; LA32-NEXT:    ret
 ;
 ; LA64-LABEL: atomicrmw_nand_i8_monotonic:
 ; LA64:       # %bb.0:
-; LA64-NEXT:    addi.w $a2, $zero, -4
-; LA64-NEXT:    and $a2, $a0, $a2
-; LA64-NEXT:    slli.d $a0, $a0, 3
+; LA64-NEXT:    slli.d $a2, $a0, 3
 ; LA64-NEXT:    ori $a3, $zero, 255
-; LA64-NEXT:    sll.w $a3, $a3, $a0
+; LA64-NEXT:    sll.w $a3, $a3, $a2
 ; LA64-NEXT:    addi.w $a3, $a3, 0
 ; LA64-NEXT:    andi $a1, $a1, 255
-; LA64-NEXT:    sll.w $a1, $a1, $a0
+; 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:    ll.w $a4, $a2, 0
+; LA64-NEXT:    ll.w $a4, $a0, 0
 ; LA64-NEXT:    and $a5, $a4, $a1
 ; LA64-NEXT:    nor $a5, $a5, $zero
 ; LA64-NEXT:    xor $a5, $a4, $a5
 ; LA64-NEXT:    and $a5, $a5, $a3
 ; LA64-NEXT:    xor $a5, $a4, $a5
-; LA64-NEXT:    sc.w $a5, $a2, 0
+; LA64-NEXT:    sc.w $a5, $a0, 0
 ; LA64-NEXT:    beqz $a5, .LBB40_1
 ; LA64-NEXT:  # %bb.2:
-; LA64-NEXT:    srl.w $a0, $a4, $a0
+; LA64-NEXT:    srl.w $a0, $a4, $a2
 ; LA64-NEXT:    ret
   %1 = atomicrmw nand ptr %a, i8 %b monotonic
   ret i8 %1
 }
 
 define i16 @atomicrmw_nand_i16_monotonic(ptr %a, i16 %b) nounwind {
 ; LA32-LABEL: atomicrmw_nand_i16_monotonic:
 ; LA32:       # %bb.0:
-; LA32-NEXT:    addi.w $a2, $zero, -4
-; LA32-NEXT:    and $a2, $a0, $a2
-; LA32-NEXT:    slli.w $a0, $a0, 3
-; LA32-NEXT:    lu12i.w $a3, 15
-; LA32-NEXT:    ori $a3, $a3, 4095
-; LA32-NEXT:    sll.w $a3, $a3, $a0
+; LA32-NEXT:    lu12i.w $a2, 15
+; LA32-NEXT:    ori $a2, $a2, 4095
+; LA32-NEXT:    slli.w $a3, $a0, 3
+; LA32-NEXT:    sll.w $a2, $a2, $a3
 ; LA32-NEXT:    bstrpick.w $a1, $a1, 15, 0
-; LA32-NEXT:    sll.w $a1, $a1, $a0
+; 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:    ll.w $a4, $a2, 0
+; LA32-NEXT:    ll.w $a4, $a0, 0
 ; LA32-NEXT:    and $a5, $a4, $a1
 ; LA32-NEXT:    nor $a5, $a5, $zero
 ; LA32-NEXT:    xor $a5, $a4, $a5
-; LA32-NEXT:    and $a5, $a5, $a3
+; LA32-NEXT:    and $a5, $a5, $a2
 ; LA32-NEXT:    xor $a5, $a4, $a5
-; LA32-NEXT:    sc.w $a5, $a2, 0
+; LA32-NEXT:    sc.w $a5, $a0, 0
 ; LA32-NEXT:    beqz $a5, .LBB41_1
 ; LA32-NEXT:  # %bb.2:
-; LA32-NEXT:    srl.w $a0, $a4, $a0
+; LA32-NEXT:    srl.w $a0, $a4, $a3
 ; LA32-NEXT:    ret
 ;
 ; LA64-LABEL: atomicrmw_nand_i16_monotonic:
 ; LA64:       # %bb.0:
-; LA64-NEXT:    addi.w $a2, $zero, -4
-; LA64-NEXT:    and $a2, $a0, $a2
-; LA64-NEXT:    slli.d $a0, $a0, 3
-; LA64-NEXT:    lu12i.w $a3, 15
-; LA64-NEXT:    ori $a3, $a3, 4095
-; LA64-NEXT:    sll.w $a3, $a3, $a0
-; LA64-NEXT:    addi.w $a3, $a3, 0
+; LA64-NEXT:    lu12i.w $a2, 15
+; LA64-NEXT:    ori $a2, $a2, 4095
+; LA64-NEXT:    slli.d $a3, $a0, 3
+; LA64-NEXT:    sll.w $a2, $a2, $a3
+; LA64-NEXT:    addi.w $a2, $a2, 0
 ; LA64-NEXT:    bstrpick.d $a1, $a1, 15, 0
-; LA64-NEXT:    sll.w $a1, $a1, $a0
+; 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:    ll.w $a4, $a2, 0
+; LA64-NEXT:    ll.w $a4, $a0, 0
 ; LA64-NEXT:    and $a5, $a4, $a1
 ; LA64-NEXT:    nor $a5, $a5, $zero
 ; LA64-NEXT:    xor $a5, $a4, $a5
-; LA64-NEXT:    and $a5, $a5, $a3
+; LA64-NEXT:    and $a5, $a5, $a2
 ; LA64-NEXT:    xor $a5, $a4, $a5
-; LA64-NEXT:    sc.w $a5, $a2, 0
+; LA64-NEXT:    sc.w $a5, $a0, 0
 ; LA64-NEXT:    beqz $a5, .LBB41_1
 ; LA64-NEXT:  # %bb.2:
-; LA64-NEXT:    srl.w $a0, $a4, $a0
+; LA64-NEXT:    srl.w $a0, $a4, $a3
 ; LA64-NEXT:    ret
   %1 = atomicrmw nand ptr %a, i16 %b monotonic
   ret i16 %1
 }
 
 define i32 @atomicrmw_nand_i32_monotonic(ptr %a, i32 %b) nounwind {
 ; LA32-LABEL: atomicrmw_nand_i32_monotonic:
 ; LA32:       # %bb.0:

 ; LA32-LABEL: atomicrmw_and_i8_monotonic:
 ; LA32:       # %bb.0:
 ; LA32-NEXT:    slli.w $a2, $a0, 3
 ; LA32-NEXT:    ori $a3, $zero, 255
 ; LA32-NEXT:    sll.w $a3, $a3, $a2
 ; LA32-NEXT:    andi $a1, $a1, 255
 ; LA32-NEXT:    sll.w $a1, $a1, $a2
 ; LA32-NEXT:    orn $a1, $a1, $a3
-; LA32-NEXT:    addi.w $a3, $zero, -4
-; LA32-NEXT:    and $a0, $a0, $a3
+; LA32-NEXT:    bstrins.w $a0, $zero, 1, 0
 ; LA32-NEXT:  .LBB44_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:  # %bb.2:
 ; LA32-NEXT:    srl.w $a0, $a3, $a2
 ; LA32-NEXT:    ret
 ;
 ; LA64-LABEL: atomicrmw_and_i8_monotonic:
 ; LA64:       # %bb.0:
 ; LA64-NEXT:    slli.d $a2, $a0, 3
 ; LA64-NEXT:    ori $a3, $zero, 255
 ; LA64-NEXT:    sll.w $a3, $a3, $a2
 ; LA64-NEXT:    andi $a1, $a1, 255
 ; LA64-NEXT:    sll.w $a1, $a1, $a2
 ; LA64-NEXT:    orn $a1, $a1, $a3
-; LA64-NEXT:    addi.w $a3, $zero, -4
-; LA64-NEXT:    and $a0, $a0, $a3
+; LA64-NEXT:    bstrins.d $a0, $zero, 1, 0
 ; LA64-NEXT:    amand_db.w $a3, $a1, $a0
 ; LA64-NEXT:    srl.w $a0, $a3, $a2
 ; LA64-NEXT:    ret
   %1 = atomicrmw and ptr %a, i8 %b monotonic
   ret i8 %1
 }
 
 define i16 @atomicrmw_and_i16_monotonic(ptr %a, i16 %b) nounwind {
 ; LA32-LABEL: atomicrmw_and_i16_monotonic:
 ; LA32:       # %bb.0:
 ; LA32-NEXT:    lu12i.w $a2, 15
 ; LA32-NEXT:    ori $a2, $a2, 4095
 ; LA32-NEXT:    slli.w $a3, $a0, 3
 ; LA32-NEXT:    sll.w $a2, $a2, $a3
 ; LA32-NEXT:    bstrpick.w $a1, $a1, 15, 0
 ; LA32-NEXT:    sll.w $a1, $a1, $a3
 ; LA32-NEXT:    orn $a1, $a1, $a2
-; LA32-NEXT:    addi.w $a2, $zero, -4
-; LA32-NEXT:    and $a0, $a0, $a2
+; LA32-NEXT:    bstrins.w $a0, $zero, 1, 0
 ; LA32-NEXT:  .LBB45_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:  # %bb.2:
 ; LA32-NEXT:    srl.w $a0, $a2, $a3
 ; LA32-NEXT:    ret
 ;
 ; LA64-LABEL: atomicrmw_and_i16_monotonic:
 ; LA64:       # %bb.0:
 ; LA64-NEXT:    lu12i.w $a2, 15
 ; LA64-NEXT:    ori $a2, $a2, 4095
 ; LA64-NEXT:    slli.d $a3, $a0, 3
 ; LA64-NEXT:    sll.w $a2, $a2, $a3
 ; LA64-NEXT:    bstrpick.d $a1, $a1, 15, 0
 ; LA64-NEXT:    sll.w $a1, $a1, $a3
 ; LA64-NEXT:    orn $a1, $a1, $a2
-; LA64-NEXT:    addi.w $a2, $zero, -4
-; LA64-NEXT:    and $a0, $a0, $a2
+; LA64-NEXT:    bstrins.d $a0, $zero, 1, 0
 ; LA64-NEXT:    amand_db.w $a2, $a1, $a0
 ; LA64-NEXT:    srl.w $a0, $a2, $a3
 ; LA64-NEXT:    ret
   %1 = atomicrmw and ptr %a, i16 %b monotonic
   ret i16 %1
 }
 
 define i32 @atomicrmw_and_i32_monotonic(ptr %a, i32 %b) nounwind {

 ; LA64-NEXT:    ret
   %1 = atomicrmw and ptr %a, i64 %b monotonic
   ret i64 %1
 }
 
 define i8 @atomicrmw_or_i8_monotonic(ptr %a, i8 %b) nounwind {
 ; LA32-LABEL: atomicrmw_or_i8_monotonic:
 ; LA32:       # %bb.0:
-; LA32-NEXT:    addi.w $a2, $zero, -4
-; LA32-NEXT:    and $a2, $a0, $a2
-; LA32-NEXT:    slli.w $a0, $a0, 3
 ; LA32-NEXT:    andi $a1, $a1, 255
-; LA32-NEXT:    sll.w $a1, $a1, $a0
+; 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:    ll.w $a3, $a2, 0
+; LA32-NEXT:    ll.w $a3, $a0, 0
 ; LA32-NEXT:    or $a4, $a3, $a1
-; LA32-NEXT:    sc.w $a4, $a2, 0
+; LA32-NEXT:    sc.w $a4, $a0, 0
 ; LA32-NEXT:    beqz $a4, .LBB48_1
 ; LA32-NEXT:  # %bb.2:
-; LA32-NEXT:    srl.w $a0, $a3, $a0
+; LA32-NEXT:    srl.w $a0, $a3, $a2
 ; LA32-NEXT:    ret
 ;
 ; LA64-LABEL: atomicrmw_or_i8_monotonic:
 ; LA64:       # %bb.0:
-; LA64-NEXT:    addi.w $a2, $zero, -4
-; LA64-NEXT:    and $a2, $a0, $a2
-; LA64-NEXT:    slli.d $a0, $a0, 3
 ; LA64-NEXT:    andi $a1, $a1, 255
-; LA64-NEXT:    sll.w $a1, $a1, $a0
-; LA64-NEXT:    amor_db.w $a3, $a1, $a2
-; LA64-NEXT:    srl.w $a0, $a3, $a0
+; LA64-NEXT:    slli.d $a2, $a0, 3
+; LA64-NEXT:    sll.w $a1, $a1, $a2
+; LA64-NEXT:    bstrins.d $a0, $zero, 1, 0
+; LA64-NEXT:    amor_db.w $a3, $a1, $a0
+; LA64-NEXT:    srl.w $a0, $a3, $a2
 ; LA64-NEXT:    ret
   %1 = atomicrmw or ptr %a, i8 %b monotonic
   ret i8 %1
 }
 
 define i16 @atomicrmw_or_i16_monotonic(ptr %a, i16 %b) nounwind {
 ; LA32-LABEL: atomicrmw_or_i16_monotonic:
 ; LA32:       # %bb.0:
-; LA32-NEXT:    addi.w $a2, $zero, -4
-; LA32-NEXT:    and $a2, $a0, $a2
-; LA32-NEXT:    slli.w $a0, $a0, 3
 ; LA32-NEXT:    bstrpick.w $a1, $a1, 15, 0
-; LA32-NEXT:    sll.w $a1, $a1, $a0
+; 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:    ll.w $a3, $a2, 0
+; LA32-NEXT:    ll.w $a3, $a0, 0
 ; LA32-NEXT:    or $a4, $a3, $a1
-; LA32-NEXT:    sc.w $a4, $a2, 0
+; LA32-NEXT:    sc.w $a4, $a0, 0
 ; LA32-NEXT:    beqz $a4, .LBB49_1
 ; LA32-NEXT:  # %bb.2:
-; LA32-NEXT:    srl.w $a0, $a3, $a0
+; LA32-NEXT:    srl.w $a0, $a3, $a2
 ; LA32-NEXT:    ret
 ;
 ; LA64-LABEL: atomicrmw_or_i16_monotonic:
 ; LA64:       # %bb.0:
-; LA64-NEXT:    addi.w $a2, $zero, -4
-; LA64-NEXT:    and $a2, $a0, $a2
-; LA64-NEXT:    slli.d $a0, $a0, 3
 ; LA64-NEXT:    bstrpick.d $a1, $a1, 15, 0
-; LA64-NEXT:    sll.w $a1, $a1, $a0
-; LA64-NEXT:    amor_db.w $a3, $a1, $a2
-; LA64-NEXT:    srl.w $a0, $a3, $a0
+; LA64-NEXT:    slli.d $a2, $a0, 3
+; LA64-NEXT:    sll.w $a1, $a1, $a2
+; LA64-NEXT:    bstrins.d $a0, $zero, 1, 0
+; LA64-NEXT:    amor_db.w $a3, $a1, $a0
+; LA64-NEXT:    srl.w $a0, $a3, $a2
 ; LA64-NEXT:    ret
   %1 = atomicrmw or ptr %a, i16 %b monotonic
   ret i16 %1
 }
 
 define i32 @atomicrmw_or_i32_monotonic(ptr %a, i32 %b) nounwind {
 ; LA32-LABEL: atomicrmw_or_i32_monotonic:
 ; LA32:       # %bb.0:

 ; LA64-NEXT:    ret
   %1 = atomicrmw or ptr %a, i64 %b monotonic
   ret i64 %1
 }
 
 define i8 @atomicrmw_xor_i8_monotonic(ptr %a, i8 %b) nounwind {
 ; LA32-LABEL: atomicrmw_xor_i8_monotonic:
 ; LA32:       # %bb.0:
-; LA32-NEXT:    addi.w $a2, $zero, -4
-; LA32-NEXT:    and $a2, $a0, $a2
-; LA32-NEXT:    slli.w $a0, $a0, 3
 ; LA32-NEXT:    andi $a1, $a1, 255
-; LA32-NEXT:    sll.w $a1, $a1, $a0
+; 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:    ll.w $a3, $a2, 0
+; LA32-NEXT:    ll.w $a3, $a0, 0
 ; LA32-NEXT:    xor $a4, $a3, $a1
-; LA32-NEXT:    sc.w $a4, $a2, 0
+; LA32-NEXT:    sc.w $a4, $a0, 0
 ; LA32-NEXT:    beqz $a4, .LBB52_1
 ; LA32-NEXT:  # %bb.2:
-; LA32-NEXT:    srl.w $a0, $a3, $a0
+; LA32-NEXT:    srl.w $a0, $a3, $a2
 ; LA32-NEXT:    ret
 ;
 ; LA64-LABEL: atomicrmw_xor_i8_monotonic:
 ; LA64:       # %bb.0:
-; LA64-NEXT:    addi.w $a2, $zero, -4
-; LA64-NEXT:    and $a2, $a0, $a2
-; LA64-NEXT:    slli.d $a0, $a0, 3
 ; LA64-NEXT:    andi $a1, $a1, 255
-; LA64-NEXT:    sll.w $a1, $a1, $a0
-; LA64-NEXT:    amxor_db.w $a3, $a1, $a2
-; LA64-NEXT:    srl.w $a0, $a3, $a0
+; LA64-NEXT:    slli.d $a2, $a0, 3
+; LA64-NEXT:    sll.w $a1, $a1, $a2
+; LA64-NEXT:    bstrins.d $a0, $zero, 1, 0
+; LA64-NEXT:    amxor_db.w $a3, $a1, $a0
+; LA64-NEXT:    srl.w $a0, $a3, $a2
 ; LA64-NEXT:    ret
   %1 = atomicrmw xor ptr %a, i8 %b monotonic
   ret i8 %1
 }
 
 define i16 @atomicrmw_xor_i16_monotonic(ptr %a, i16 %b) nounwind {
 ; LA32-LABEL: atomicrmw_xor_i16_monotonic:
 ; LA32:       # %bb.0:
-; LA32-NEXT:    addi.w $a2, $zero, -4
-; LA32-NEXT:    and $a2, $a0, $a2
-; LA32-NEXT:    slli.w $a0, $a0, 3
 ; LA32-NEXT:    bstrpick.w $a1, $a1, 15, 0
-; LA32-NEXT:    sll.w $a1, $a1, $a0
+; 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:    ll.w $a3, $a2, 0
+; LA32-NEXT:    ll.w $a3, $a0, 0
 ; LA32-NEXT:    xor $a4, $a3, $a1
-; LA32-NEXT:    sc.w $a4, $a2, 0
+; LA32-NEXT:    sc.w $a4, $a0, 0
 ; LA32-NEXT:    beqz $a4, .LBB53_1
 ; LA32-NEXT:  # %bb.2:
-; LA32-NEXT:    srl.w $a0, $a3, $a0
+; LA32-NEXT:    srl.w $a0, $a3, $a2
 ; LA32-NEXT:    ret
 ;
 ; LA64-LABEL: atomicrmw_xor_i16_monotonic:
 ; LA64:       # %bb.0:
-; LA64-NEXT:    addi.w $a2, $zero, -4
-; LA64-NEXT:    and $a2, $a0, $a2
-; LA64-NEXT:    slli.d $a0, $a0, 3
 ; LA64-NEXT:    bstrpick.d $a1, $a1, 15, 0
-; LA64-NEXT:    sll.w $a1, $a1, $a0
-; LA64-NEXT:    amxor_db.w $a3, $a1, $a2
-; LA64-NEXT:    srl.w $a0, $a3, $a0
+; LA64-NEXT:    slli.d $a2, $a0, 3
+; LA64-NEXT:    sll.w $a1, $a1, $a2
+; LA64-NEXT:    bstrins.d $a0, $zero, 1, 0
+; LA64-NEXT:    amxor_db.w $a3, $a1, $a0
+; LA64-NEXT:    srl.w $a0, $a3, $a2
 ; LA64-NEXT:    ret
   %1 = atomicrmw xor ptr %a, i16 %b monotonic
   ret i16 %1
 }
 
 define i32 @atomicrmw_xor_i32_monotonic(ptr %a, i32 %b) nounwind {
 ; LA32-LABEL: atomicrmw_xor_i32_monotonic:
 ; LA32:       # %bb.0:

llvm/test/CodeGen/LoongArch/stack-realignment-with-variable-sized-objects.ll

 ; LA32-NEXT:    .cfi_offset 1, -4
 ; LA32-NEXT:    .cfi_offset 22, -8
 ; LA32-NEXT:    .cfi_offset 31, -12
 ; LA32-NEXT:    addi.w $fp, $sp, 64
 ; LA32-NEXT:    .cfi_def_cfa 22, 0
 ; LA32-NEXT:    bstrins.w $sp, $zero, 5, 0
 ; LA32-NEXT:    move $s8, $sp
 ; LA32-NEXT:    addi.w $a0, $a0, 15
-; LA32-NEXT:    addi.w $a1, $zero, -16
-; LA32-NEXT:    and $a0, $a0, $a1
+; LA32-NEXT:    bstrins.w $a0, $zero, 3, 0
 ; LA32-NEXT:    sub.w $a0, $sp, $a0
 ; LA32-NEXT:    move $sp, $a0
 ; LA32-NEXT:    addi.w $a1, $s8, 0
 ; LA32-NEXT:    bl %plt(callee)
 ; LA32-NEXT:    addi.w $sp, $fp, -64
 ; LA32-NEXT:    ld.w $s8, $sp, 52 # 4-byte Folded Reload
 ; LA32-NEXT:    ld.w $fp, $sp, 56 # 4-byte Folded Reload
 ; LA32-NEXT:    ld.w $ra, $sp, 60 # 4-byte Folded Reload

llvm/test/CodeGen/LoongArch/vararg.ll

 ; LA64-FPELIM-NEXT:    st.d $a7, $fp, 56
 ; LA64-FPELIM-NEXT:    st.d $a6, $fp, 48
 ; LA64-FPELIM-NEXT:    st.d $a5, $fp, 40
 ; LA64-FPELIM-NEXT:    st.d $a4, $fp, 32
 ; LA64-FPELIM-NEXT:    st.d $a3, $fp, 24
 ; LA64-FPELIM-NEXT:    st.d $a2, $fp, 16
 ; LA64-FPELIM-NEXT:    addi.d $a0, $fp, 16
 ; LA64-FPELIM-NEXT:    st.d $a0, $fp, -32
+; LA64-FPELIM-NEXT:    st.d $a1, $fp, 8
 ; LA64-FPELIM-NEXT:    addi.d $a0, $a1, 15
-; LA64-FPELIM-NEXT:    addi.w $a1, $zero, -16
-; LA64-FPELIM-NEXT:    and $a0, $a0, $a1
-; LA64-FPELIM-NEXT:    st.d $s0, $fp, 8
+; LA64-FPELIM-NEXT:    bstrins.d $a0, $zero, 3, 0
 ; LA64-FPELIM-NEXT:    sub.d $a0, $sp, $a0
 ; LA64-FPELIM-NEXT:    move $sp, $a0
 ; LA64-FPELIM-NEXT:    bl %plt(notdead)
 ; LA64-FPELIM-NEXT:    move $a0, $s0
 ; LA64-FPELIM-NEXT:    addi.d $sp, $fp, -32
 ; LA64-FPELIM-NEXT:    ld.d $s0, $sp, 8 # 8-byte Folded Reload
 ; LA64-FPELIM-NEXT:    ld.d $fp, $sp, 16 # 8-byte Folded Reload
 ; LA64-FPELIM-NEXT:    ld.d $ra, $sp, 24 # 8-byte Folded Reload

 ; LA64-WITHFP-NEXT:    st.d $a7, $fp, 56
 ; LA64-WITHFP-NEXT:    st.d $a6, $fp, 48
 ; LA64-WITHFP-NEXT:    st.d $a5, $fp, 40
 ; LA64-WITHFP-NEXT:    st.d $a4, $fp, 32
 ; LA64-WITHFP-NEXT:    st.d $a3, $fp, 24
 ; LA64-WITHFP-NEXT:    st.d $a2, $fp, 16
 ; LA64-WITHFP-NEXT:    addi.d $a0, $fp, 16
 ; LA64-WITHFP-NEXT:    st.d $a0, $fp, -32
+; LA64-WITHFP-NEXT:    st.d $a1, $fp, 8
 ; LA64-WITHFP-NEXT:    addi.d $a0, $a1, 15
-; LA64-WITHFP-NEXT:    addi.w $a1, $zero, -16
-; LA64-WITHFP-NEXT:    and $a0, $a0, $a1
-; LA64-WITHFP-NEXT:    st.d $s0, $fp, 8
+; LA64-WITHFP-NEXT:    bstrins.d $a0, $zero, 3, 0
 ; LA64-WITHFP-NEXT:    sub.d $a0, $sp, $a0
 ; LA64-WITHFP-NEXT:    move $sp, $a0
 ; LA64-WITHFP-NEXT:    bl %plt(notdead)
 ; LA64-WITHFP-NEXT:    move $a0, $s0
 ; LA64-WITHFP-NEXT:    addi.d $sp, $fp, -32
 ; LA64-WITHFP-NEXT:    ld.d $s0, $sp, 8 # 8-byte Folded Reload
 ; LA64-WITHFP-NEXT:    ld.d $fp, $sp, 16 # 8-byte Folded Reload
 ; LA64-WITHFP-NEXT:    ld.d $ra, $sp, 24 # 8-byte Folded Reload