diff --git a/docs/sql-migration-guide.md b/docs/sql-migration-guide.md
index 73a4d19a1b15..9825e2ff4893 100644
--- a/docs/sql-migration-guide.md
+++ b/docs/sql-migration-guide.md
@@ -256,6 +256,8 @@ license: |
 
   - Since Spark 3.0, day-time interval strings are converted to intervals with respect to the `from` and `to` bounds. If an input string does not match to the pattern defined by specified bounds, the `ParseException` exception is thrown. For example, `interval '2 10:20' hour to minute` raises the exception because the expected format is `[+|-]h[h]:[m]m`. In Spark version 2.4, the `from` bound was not taken into account, and the `to` bound was used to truncate the resulted interval. For instance, the day-time interval string from the showed example is converted to `interval 10 hours 20 minutes`. To restore the behavior before Spark 3.0, you can set `spark.sql.legacy.fromDayTimeString.enabled` to `true`.
 
+  - Since Spark 3.0, the `date_add` and `date_sub` functions only accepts int, smallint, tinyint as the 2nd argument, fractional and string types are not valid anymore, e.g. `date_add(cast('1964-05-23' as date), '12.34')` will cause `AnalysisException`. In Spark version 2.4 and earlier, if the 2nd argument is fractional or string value, it will be coerced to int value, and the result will be a date value of `1964-06-04`.
+
 ## Upgrading from Spark SQL 2.4 to 2.4.1
 
   - The value of `spark.executor.heartbeatInterval`, when specified without units like "30" rather than "30s", was