diff --git a/src/mapleall/maple_be/include/cg/aarch64/aarch64_cgfunc.h b/src/mapleall/maple_be/include/cg/aarch64/aarch64_cgfunc.h
index 1f4410d0f2feb172b7e1ebc321a3b0b651cfb551..17c6d21c1d57f3d14df7f05bd34756a31add847c 100644
--- a/src/mapleall/maple_be/include/cg/aarch64/aarch64_cgfunc.h
+++ b/src/mapleall/maple_be/include/cg/aarch64/aarch64_cgfunc.h
@@ -435,6 +435,7 @@ class AArch64CGFunc : public CGFunc {
 
   MOperator PickStInsn(uint32 bitSize, PrimType primType, AArch64isa::MemoryOrdering memOrd = AArch64isa::kMoNone);
   MOperator PickLdInsn(uint32 bitSize, PrimType primType, AArch64isa::MemoryOrdering memOrd = AArch64isa::kMoNone);
+  MOperator PickExtInsn(PrimType dtype, PrimType stype);
 
   bool CheckIfSplitOffsetWithAdd(const AArch64MemOperand &memOpnd, uint32 bitLen);
   AArch64MemOperand &SplitOffsetWithAddInstruction(const AArch64MemOperand &memOpnd, uint32 bitLen,
@@ -566,6 +567,10 @@ class AArch64CGFunc : public CGFunc {
     return (o.IsRegister() ? static_cast<RegOperand&>(o) : SelectCopy(o, oty, oty));
   }
 
+  RegOperand &LoadIntoRegister(Operand &o, PrimType dty, PrimType sty) {
+    return (o.IsRegister() ? static_cast<RegOperand&>(o) : SelectCopy(o, sty, dty));
+  }
+
   void CreateCallStructParamPassByStack(int32 symSize, MIRSymbol *sym, RegOperand *addrOpnd, int32 baseOffset);
   void CreateCallStructParamPassByReg(AArch64reg reg, MemOperand &memOpnd, AArch64ListOperand &srcOpnds,
                                       fpParamState state);
diff --git a/src/mapleall/maple_be/src/cg/aarch64/aarch64_cgfunc.cpp b/src/mapleall/maple_be/src/cg/aarch64/aarch64_cgfunc.cpp
index ae4ddad5a48eb3f1e50fdda787fc0d28b53602d2..82f2ce38dca97459cfd4e097cc278ce41e560cf8 100644
--- a/src/mapleall/maple_be/src/cg/aarch64/aarch64_cgfunc.cpp
+++ b/src/mapleall/maple_be/src/cg/aarch64/aarch64_cgfunc.cpp
@@ -74,6 +74,26 @@ MOperator stFs[kFloatByteSizeDimension] = { MOP_sstr, MOP_dstr };
 MOperator ldFsAcq[kFloatByteSizeDimension] = { MOP_undef, MOP_undef };
 MOperator stFsRel[kFloatByteSizeDimension] = { MOP_undef, MOP_undef };
 
+/* extended to unsigned ints */
+MOperator uextIs[kIntByteSizeDimension][kIntByteSizeDimension] = {
+  /*  u8         u16          u32          u64      */
+  { MOP_undef, MOP_xuxtb32, MOP_xuxtb32, MOP_xuxtb32},  /* u8/i8   */
+  { MOP_undef, MOP_undef,   MOP_xuxth32, MOP_xuxth32},  /* u16/i16 */
+  { MOP_undef, MOP_undef,   MOP_xuxtw64, MOP_xuxtw64},  /* u32/i32 */
+  { MOP_undef, MOP_undef,   MOP_undef,   MOP_undef}     /* u64/u64 */
+};
+
+/* extended to signed ints */
+MOperator extIs[kIntByteSizeDimension][kIntByteSizeDimension] = {
+  /*  i8         i16          i32          i64      */
+  { MOP_undef, MOP_xsxtb32, MOP_xsxtb32, MOP_xsxtb64},  /* u8/i8   */
+  { MOP_undef, MOP_undef,   MOP_xsxth32, MOP_xsxth64},  /* u16/i16 */
+  { MOP_undef, MOP_undef,   MOP_undef,   MOP_xsxtw64},  /* u32/i32 */
+  { MOP_undef, MOP_undef,   MOP_undef,   MOP_undef}     /* u64/u64 */
+};
+
+/* extended to signed ints */
+
 MOperator PickLdStInsn(bool isLoad, uint32 bitSize, PrimType primType, AArch64isa::MemoryOrdering memOrd) {
   ASSERT(__builtin_popcount(static_cast<uint32>(memOrd)) <= 1, "must be kMoNone or kMoAcquire");
   ASSERT(primType != PTY_ptr, "should have been lowered");
@@ -126,6 +146,24 @@ MOperator AArch64CGFunc::PickStInsn(uint32 bitSize, PrimType primType, AArch64is
   return PickLdStInsn(false, bitSize, primType, memOrd);
 }
 
+MOperator AArch64CGFunc::PickExtInsn(PrimType dtype, PrimType stype) {
+  int32 sBitSize = GetPrimTypeBitSize(stype);
+  int32 dBitSize = GetPrimTypeBitSize(dtype);
+  /* __builtin_ffs(x) returns: 0 -> 0, 1 -> 1, 2 -> 2, 4 -> 3, 8 -> 4 */
+  if (IsPrimitiveInteger(stype) && IsPrimitiveInteger(dtype)) {
+    MOperator(*table)[kIntByteSizeDimension];
+    table = IsUnsignedInteger(dtype) ? uextIs : extIs;
+    /* __builtin_ffs(x) returns: 8 -> 4, 16 -> 5, 32 -> 6, 64 -> 7 */
+    uint32 row = static_cast<uint32>(__builtin_ffs(static_cast<int32>(sBitSize))) - 4;
+    ASSERT(row <= 3, "wrong bitSize");
+    uint32 col = static_cast<uint32>(__builtin_ffs(static_cast<int32>(dBitSize))) - 4;
+    ASSERT(col <= 3, "wrong bitSize");
+    return table[row][col];
+  }
+  CHECK_FATAL(0, "extend not primitive integer");
+  return MOP_undef;
+}
+
 MOperator AArch64CGFunc::PickMovInsn(PrimType primType) {
   switch (primType) {
     case PTY_u8:
@@ -429,11 +467,21 @@ void AArch64CGFunc::SelectCopyMemOpnd(Operand &dest, PrimType dtype, uint32 dsiz
   }
   Insn *insn = nullptr;
   uint32 ssize = src.GetSize();
+  PrimType regTy = PTY_void;
+  RegOperand *loadReg;
+  MOperator mop = MOP_undef;
   if (IsPrimitiveFloat(stype)) {
     CHECK_FATAL(dsize == ssize, "dsize %u expect equals ssize %u", dtype, ssize);
     insn = &GetCG()->BuildInstruction<AArch64Insn>(PickLdInsn(ssize, stype), dest, src);
   } else {
-    insn = &GetCG()->BuildInstruction<AArch64Insn>(PickLdInsn(ssize, stype), dest, src);
+    mop = PickExtInsn(dtype, stype);
+    if (ssize == (GetPrimTypeSize(dtype) * kBitsPerByte) || mop == MOP_undef) {
+      insn = &GetCG()->BuildInstruction<AArch64Insn>(PickLdInsn(ssize, stype), dest, src);
+    } else {
+      regTy = dsize == k64BitSize ? dtype : PTY_i32;
+      loadReg = &CreateRegisterOperandOfType(regTy);
+      insn = &GetCG()->BuildInstruction<AArch64Insn>(PickLdInsn(ssize, stype), *loadReg, src);
+    }
   }
 
   if (GetCG()->GenerateVerboseCG()) {
@@ -453,6 +501,10 @@ void AArch64CGFunc::SelectCopyMemOpnd(Operand &dest, PrimType dtype, uint32 dsiz
   }
 
   GetCurBB()->AppendInsn(*insn);
+
+  if (regTy != PTY_void && mop != MOP_undef) {
+    GetCurBB()->AppendInsn(GetCG()->BuildInstruction<AArch64Insn>(mop, dest, *loadReg));
+  }
 }
 
 bool AArch64CGFunc::IsImmediateValueInRange(MOperator mOp, int64 immVal, bool is64Bits,
@@ -1533,6 +1585,10 @@ Operand *AArch64CGFunc::SelectDread(const BaseNode &parent, DreadNode &expr) {
       IsImmediateOffsetOutOfRange(*static_cast<AArch64MemOperand*>(memOpnd), dataSize)) {
     return &SplitOffsetWithAddInstruction(*static_cast<AArch64MemOperand*>(memOpnd), dataSize);
   }
+  if (symType != expr.GetPrimType()) {
+    RegOperand *opnd = &LoadIntoRegister(*memOpnd, expr.GetPrimType(), symType);
+    return opnd;
+  }
   return memOpnd;
 }
 
diff --git a/src/mapleall/maple_be/src/cg/aarch64/aarch64_operand.cpp b/src/mapleall/maple_be/src/cg/aarch64/aarch64_operand.cpp
index 1d2c84df475f3093525527391cba27d632078c6c..2681590c7f223f3bf3c36de74f7188edceb9f3a2 100644
--- a/src/mapleall/maple_be/src/cg/aarch64/aarch64_operand.cpp
+++ b/src/mapleall/maple_be/src/cg/aarch64/aarch64_operand.cpp
@@ -175,11 +175,7 @@ bool StImmOperand::Less(const Operand &right) const{
 void StImmOperand::Emit(Emitter &emitter, const OpndProp *opndProp) const {
   CHECK_FATAL(opndProp != nullptr, "opndProp is nullptr in  StImmOperand::Emit");
   if (static_cast<const AArch64OpndProp*>(opndProp)->IsLiteralLow12()) {
-    emitter.Emit("#:lo12:" + GetName());
-    if (offset != 0) {
-      emitter.Emit("+" + std::to_string(offset));
-    }
-    return;
+    emitter.Emit("#:lo12:");
   }
   if (CGOptions::IsPIC() && (symbol->GetStorageClass() == kScGlobal || symbol->GetStorageClass() == kScExtern)) {
     emitter.Emit(":got:" + GetName());