diff --git a/runtime/ecma_string.cpp b/runtime/ecma_string.cpp
index 8b6d085c3b5d36c84a27ff253cd544ca3146a78c..b784b485f2f205083b031101eca984349e0c82dd 100644
--- a/runtime/ecma_string.cpp
+++ b/runtime/ecma_string.cpp
@@ -411,11 +411,12 @@ uint32_t EcmaString::ComputeHashcodeUtf8(const uint8_t *utf8Data, size_t utf8Len
     if (canBeCompress) {
         hash = ComputeHashForUtf8(utf8Data);
     } else {
-        auto utf16Len = base::utf_helper::Utf8ToUtf16Size(utf8Data, utf8Len);
+        // utf8len >= utf16len for equal strings
+        auto utf16Len = utf8Len;
         PandaVector<uint16_t> tmpBuffer(utf16Len);
         [[maybe_unused]] auto len =
             base::utf_helper::ConvertRegionUtf8ToUtf16(utf8Data, tmpBuffer.data(), utf8Len, utf16Len, 0);
-        ASSERT(len == utf16Len);
+        ASSERT(len == base::utf_helper::Utf8ToUtf16Size(utf8Data, utf8Len));
         hash = ComputeHashForData(tmpBuffer.data(), utf16Len);
     }
     return hash;