diff --git a/libs/libcommon/include/common/arithmeticOverflow.h b/libs/libcommon/include/common/arithmeticOverflow.h
index 38a1482e1c4..b80b98b218c 100644
--- a/libs/libcommon/include/common/arithmeticOverflow.h
+++ b/libs/libcommon/include/common/arithmeticOverflow.h
@@ -1,4 +1,5 @@
 #pragma once
+#include "types.h"
 
 namespace common
 {
@@ -99,11 +100,11 @@ inline bool mulOverflow(__int128 x, __int128 y, __int128 & res)
     if (!x || !y)
         return false;
 
-    unsigned __int128 a = (x > 0) ? x : -x;
-    unsigned __int128 b = (y > 0) ? y : -y;
-    return (a * b) / b != a;
+    return res / x != y; /// whether overflow int128
 }
 
+/// Int256 doesn't use the complement representation to express negative values, but uses an extra bit to express the sign flag,
+/// the actual range of Int256 is from -(2^256 - 1) to 2^256 - 1, so 2^255 ~ 2^256-1 do not overflow Int256.
 template <>
 inline bool mulOverflow(Int256 x, Int256 y, Int256 & res)
 {
diff --git a/libs/libcommon/src/tests/CMakeLists.txt b/libs/libcommon/src/tests/CMakeLists.txt
index ceb88c12502..0b434473a2b 100644
--- a/libs/libcommon/src/tests/CMakeLists.txt
+++ b/libs/libcommon/src/tests/CMakeLists.txt
@@ -23,6 +23,7 @@ add_executable (gtests_libcommon
     gtest_mem_utils.cpp
     gtest_crc64.cpp
     gtest_logger.cpp
+    gtest_arithmetic_overflow.cpp
     )
 target_link_libraries (gtests_libcommon gtest_main common)
 add_check(gtests_libcommon)
diff --git a/libs/libcommon/src/tests/gtest_arithmetic_overflow.cpp b/libs/libcommon/src/tests/gtest_arithmetic_overflow.cpp
new file mode 100644
index 00000000000..c3bc3b243bc
--- /dev/null
+++ b/libs/libcommon/src/tests/gtest_arithmetic_overflow.cpp
@@ -0,0 +1,53 @@
+// Copyright 2022 PingCAP, Ltd.
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+#include <common/arithmeticOverflow.h>
+#include <common/types.h>
+#include <gtest/gtest.h>
+TEST(OVERFLOW_Suite, SimpleTest)
+{
+    /// mul int128
+    __int128 res128;
+    bool is_overflow;
+    /// 2^126
+    static constexpr __int128 int_126 = __int128(__int128(1) << 126);
+
+    /// 2^126 << 0 = 2^126
+    is_overflow = common::mulOverflow(int_126, __int128(1), res128);
+    ASSERT_EQ(is_overflow, false);
+
+    /// 2^126 << 1 = 2^127
+    is_overflow = common::mulOverflow(int_126, __int128(2), res128);
+    ASSERT_EQ(is_overflow, true);
+
+    /// 2^126 << 2 = 2^128
+    is_overflow = common::mulOverflow(int_126, __int128(4), res128);
+    ASSERT_EQ(is_overflow, true);
+
+    /// mul int256
+    Int256 res256;
+    /// 2^254
+    static constexpr Int256 int_254 = Int256((Int256(0x1) << 254));
+    /// 2^254 << 0 = 2^254
+    is_overflow = common::mulOverflow(int_254, Int256(1), res256);
+    ASSERT_EQ(is_overflow, false);
+
+    /// 2^254 << 1 = 2^255
+    is_overflow = common::mulOverflow(int_254, Int256(2), res256);
+    ASSERT_EQ(is_overflow, false); /// because the sign flag is processed by an extra bit, excluding from 256 bits of Int256.
+
+    /// 2^254 << 2 = 2^256
+    is_overflow = common::mulOverflow(int_254, Int256(4), res256);
+    ASSERT_EQ(is_overflow, true);
+}
diff --git a/tests/tidb-ci/fullstack-test-dt/expr_push_down.test b/tests/tidb-ci/fullstack-test-dt/expr_push_down.test
index 60d2fe0fc68..a3ee5bb1d16 100644
--- a/tests/tidb-ci/fullstack-test-dt/expr_push_down.test
+++ b/tests/tidb-ci/fullstack-test-dt/expr_push_down.test
@@ -96,3 +96,25 @@ mysql> use test; set @@tidb_isolation_read_engines='tiflash,tidb'; set @@tidb_al
 | NULL                 | NULL | NULL | NULL | NULL                 | NULL | NULL       |        2 |
 | 平凯xingchen公司     |      | 平   |      | 平凯xingchen公司     | NULL | NULL       |        1 |
 +----------------------+------+------+------+----------------------+------+------------+----------+
+
+## test overflow int128, uint128 or not.
+mysql> drop table if exists test.t;
+mysql> CREATE TABLE test.t (v1 decimal(20,20),v2 decimal(30,0));
+mysql> insert into test.t values (0.00000000000000000000 , 2585910611040796672),(0.00000000000000000000 , -1901644942657191936), (0.00000000000000000000 , -11901644942657191936),(0.00000000000000000000 , 25859106110407966722),(0.00000000000000000000 , 2585912),(0.00000000000000000000 , -190);
+mysql> alter table test.t set tiflash replica 1;
+
+mysql> analyze table test.t;
+
+func> wait_table test t
+
+mysql> use test; set @@tidb_isolation_read_engines='tiflash,tidb'; set @@tidb_enforce_mpp=1; select v1,v2,v1>v2,v1>=v2, v1<v2,v1<=v2 from test.t;
++------------------------+-----------------------+-------+--------+-------+--------+
+| v1                     | v2                    | v1>v2 | v1>=v2 | v1<v2 | v1<=v2 |
++------------------------+-----------------------+-------+--------+-------+--------+
+| 0.00000000000000000000 |   2585910611040796672 |     0 |      0 |     1 |      1 |
+| 0.00000000000000000000 |  -1901644942657191936 |     1 |      1 |     0 |      0 |
+| 0.00000000000000000000 | -11901644942657191936 |     1 |      1 |     0 |      0 |
+| 0.00000000000000000000 |  25859106110407966722 |     0 |      0 |     1 |      1 |
+| 0.00000000000000000000 |               2585912 |     0 |      0 |     1 |      1 |
+| 0.00000000000000000000 |                  -190 |     1 |      1 |     0 |      0 |
++------------------------+-----------------------+-------+--------+-------+--------+