Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

[SPARK-22500][SQL] Fix 64KB JVM bytecode limit problem with cast #19730

Closed
wants to merge 6 commits into from
Closed

[SPARK-22500][SQL] Fix 64KB JVM bytecode limit problem with cast #19730

Show file tree
Hide file tree
Changes from 3 commits
Commits
Show all changes
6 commits
Select commit Hold shift + click to select a range
9013671
initial commit
kiszk Nov 12, 2017
94dfde2
address review comment
kiszk Nov 13, 2017
39ffaa9
make changes smaller
kiszk Nov 13, 2017
83fef40
address review comment
kiszk Nov 17, 2017
c46ef5c
address review comments
kiszk Nov 21, 2017
574691f
address review comment
kiszk Nov 21, 2017
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
10 changes: 8 additions & 2 deletions sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/Cast.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1039,13 +1039,19 @@ case class Cast(child: Expression, dataType: DataType, timeZoneId: Option[String
}
}
"""
}.mkString("\n")
}
val fieldsEvalCodes = if (ctx.INPUT_ROW != null && ctx.currentVars == null) {
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

shouldn't be ctx.currentVars != null?

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

If ctx.currentVars != null, we need to use mkString("\n").

ctx.splitExpressions(fieldsEvalCode, "castStruct",
("InternalRow", ctx.INPUT_ROW) :: (rowClass, result) :: ("InternalRow", tmpRow) :: Nil)
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

how about inner struct? We also need to pass in the ctx.INPUT_ROW?

Copy link
Member Author

@kiszk kiszk Nov 16, 2017 •
edited
Loading

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Inner struct case in the following code and existing cases in CastSuite works well. Tomorrow, I will add similar test case with many fields.

    val struct = Literal.create(
      Row(
        UTF8String.fromString("123.4"),
        Seq("456", "true", "78.9"),
        Row(7)),
      StructType(Seq(
        StructField("i", StringType, nullable = true),
        StructField("a",
          ArrayType(StringType, containsNull = false), nullable = true),
        StructField("s",
          StructType(Seq(
            StructField("i", IntegerType, nullable = true)))))))

    val ret = cast(struct, StructType(Seq(
      StructField("d", DoubleType, nullable = true),
      StructField("a",
        ArrayType(IntegerType, containsNull = true), nullable = true),
      StructField("s",
        StructType(Seq(
          StructField("l", LongType, nullable = true)))))))

    assert(ret.resolved === true)
    checkEvaluation(ret, Row(123.4, Seq(456, null, 78), Row(7L)))

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I mean, we don't need to pass in ctx.INPUT_ROW to the split functions.

} else {
fieldsEvalCode.mkString("\n")
}

(c, evPrim, evNull) =>
s"""
final $rowClass $result = new $rowClass(${fieldsCasts.length});
final InternalRow $tmpRow = $c;
$fieldsEvalCode
$fieldsEvalCodes
$evPrim = $result;
"""
}
Expand Down
13 changes: 13 additions & 0 deletions sql/catalyst/src/test/scala/org/apache/spark/sql/catalyst/expressions/CastSuite.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -827,4 +827,17 @@ class CastSuite extends SparkFunSuite with ExpressionEvalHelper {

checkEvaluation(cast(Literal.create(input, from), to), input)
}

test("SPARK-22500: cast for struct should not generate codes beyond 64KB") {
val N = 1000
val from = new StructType(
(1 to N).map(i => StructField(s"s$i", StringType)).toArray)
val to = new StructType(
(1 to N).map(i => StructField(s"i$i", IntegerType)).toArray)

val input = Row.fromSeq((1 to N).map(i => i.toString))
val output = Row.fromSeq((1 to N))

checkEvaluation(cast(Literal.create(input, from), to), output)
}
}