[SPARK-21330][SQL] Bad partitioning does not allow to read a JDBC table with extreme values on the partition column

## What changes were proposed in this pull request?

An overflow of the difference of bounds on the partitioning column leads to no data being read. This
patch checks for this overflow.

## How was this patch tested?

New unit test.

Author: Andrew Ray <[email protected]>

Closes #18800 from aray/SPARK-21330.

Author: aray

sql/core/src/main/scala/org/apache/spark/sql/execution/datasources/jdbc/JDBCRelation.scala

@@ -64,7 +64,8 @@ private[sql] object JDBCRelation extends Logging {
       s"bound. Lower bound: $lowerBound; Upper bound: $upperBound")
 
     val numPartitions =
-      if ((upperBound - lowerBound) >= partitioning.numPartitions) {
+      if ((upperBound - lowerBound) >= partitioning.numPartitions || /* check for overflow */
+          (upperBound - lowerBound) < 0) {
         partitioning.numPartitions
       } else {
         logWarning("The number of partitions is reduced because the specified number of " +

sql/core/src/test/scala/org/apache/spark/sql/jdbc/JDBCSuite.scala

@@ -96,6 +96,15 @@ class JDBCSuite extends SparkFunSuite
         |         partitionColumn 'THEID', lowerBound '1', upperBound '4', numPartitions '3')
        """.stripMargin.replaceAll("\n", " "))
 
+    sql(
+      s"""
+        |CREATE OR REPLACE TEMPORARY VIEW partsoverflow
+        |USING org.apache.spark.sql.jdbc
+        |OPTIONS (url '$url', dbtable 'TEST.PEOPLE', user 'testUser', password 'testPass',
+        |         partitionColumn 'THEID', lowerBound '-9223372036854775808',
+        |         upperBound '9223372036854775807', numPartitions '3')
+       """.stripMargin.replaceAll("\n", " "))
+
     conn.prepareStatement("create table test.inttypes (a INT, b BOOLEAN, c TINYINT, "
       + "d SMALLINT, e BIGINT)").executeUpdate()
     conn.prepareStatement("insert into test.inttypes values (1, false, 3, 4, 1234567890123)"
@@ -376,6 +385,12 @@ class JDBCSuite extends SparkFunSuite
     assert(ids(2) === 3)
   }
 
+  test("overflow of partition bound difference does not give negative stride") {
+    val df = sql("SELECT * FROM partsoverflow")
+    checkNumPartitions(df, expectedNumPartitions = 3)
+    assert(df.collect().length == 3)
+  }
+
   test("Register JDBC query with renamed fields") {
     // Regression test for bug SPARK-7345
     sql(