Snippets Groups Projects

[SPARK-18207][SQL] Fix a compilation error due to HashExpression.doGenCode

Kazuaki Ishizaki authored 8 years ago

This PR avoids a compilation error due to more than 64KB Java byte code size. This error occur since  generate java code for computing a hash value for a row is too big. This PR fixes this compilation error by splitting a big code chunk into multiple methods by calling `CodegenContext.splitExpression` at `HashExpression.doGenCode`

The test case requires a calculation of hash code for a row that includes 1000 String fields. `HashExpression.doGenCode` generate a lot of Java code for this computation into one function. As a result, the size of the corresponding Java bytecode is more than 64 KB.

Generated code without this PR
````java
/* 027 */   public UnsafeRow apply(InternalRow i) {
/* 028 */     boolean isNull = false;
/* 029 */
/* 030 */     int value1 = 42;
/* 031 */
/* 032 */     boolean isNull2 = i.isNullAt(0);
/* 033 */     UTF8String value2 = isNull2 ? null : (i.getUTF8String(0));
/* 034 */     if (!isNull2) {
/* 035 */       value1 = org.apache.spark.unsafe.hash.Murmur3_x86_32.hashUnsafeBytes(value2.getBaseObject(), value2.getBaseOffset(), value2.numBytes(), value1);
/* 036 */     }
/* 037 */
/* 038 */
/* 039 */     boolean isNull3 = i.isNullAt(1);
/* 040 */     UTF8String value3 = isNull3 ? null : (i.getUTF8String(1));
/* 041 */     if (!isNull3) {
/* 042 */       value1 = org.apache.spark.unsafe.hash.Murmur3_x86_32.hashUnsafeBytes(value3.getBaseObject(), value3.getBaseOffset(), value3.numBytes(), value1);
/* 043 */     }
/* 044 */
/* 045 */
...
/* 7024 */
/* 7025 */     boolean isNull1001 = i.isNullAt(999);
/* 7026 */     UTF8String value1001 = isNull1001 ? null : (i.getUTF8String(999));
/* 7027 */     if (!isNull1001) {
/* 7028 */       value1 = org.apache.spark.unsafe.hash.Murmur3_x86_32.hashUnsafeBytes(value1001.getBaseObject(), value1001.getBaseOffset(), value1001.numBytes(), value1);
/* 7029 */     }
/* 7030 */
/* 7031 */
/* 7032 */     boolean isNull1002 = i.isNullAt(1000);
/* 7033 */     UTF8String value1002 = isNull1002 ? null : (i.getUTF8String(1000));
/* 7034 */     if (!isNull1002) {
/* 7035 */       value1 = org.apache.spark.unsafe.hash.Murmur3_x86_32.hashUnsafeBytes(value1002.getBaseObject(), value1002.getBaseOffset(), value1002.numBytes(), value1);
/* 7036 */     }
````

Generated code with this PR
````java
/* 3807 */   private void apply_249(InternalRow i) {
/* 3808 */
/* 3809 */     boolean isNull998 = i.isNullAt(996);
/* 3810 */     UTF8String value998 = isNull998 ? null : (i.getUTF8String(996));
/* 3811 */     if (!isNull998) {
/* 3812 */       value1 = org.apache.spark.unsafe.hash.Murmur3_x86_32.hashUnsafeBytes(value998.getBaseObject(), value998.getBaseOffset(), value998.numBytes(), value1);
/* 3813 */     }
/* 3814 */
/* 3815 */     boolean isNull999 = i.isNullAt(997);
/* 3816 */     UTF8String value999 = isNull999 ? null : (i.getUTF8String(997));
/* 3817 */     if (!isNull999) {
/* 3818 */       value1 = org.apache.spark.unsafe.hash.Murmur3_x86_32.hashUnsafeBytes(value999.getBaseObject(), value999.getBaseOffset(), value999.numBytes(), value1);
/* 3819 */     }
/* 3820 */
/* 3821 */     boolean isNull1000 = i.isNullAt(998);
/* 3822 */     UTF8String value1000 = isNull1000 ? null : (i.getUTF8String(998));
/* 3823 */     if (!isNull1000) {
/* 3824 */       value1 = org.apache.spark.unsafe.hash.Murmur3_x86_32.hashUnsafeBytes(value1000.getBaseObject(), value1000.getBaseOffset(), value1000.numBytes(), value1);
/* 3825 */     }
/* 3826 */
/* 3827 */     boolean isNull1001 = i.isNullAt(999);
/* 3828 */     UTF8String value1001 = isNull1001 ? null : (i.getUTF8String(999));
/* 3829 */     if (!isNull1001) {
/* 3830 */       value1 = org.apache.spark.unsafe.hash.Murmur3_x86_32.hashUnsafeBytes(value1001.getBaseObject(), value1001.getBaseOffset(), value1001.numBytes(), value1);
/* 3831 */     }
/* 3832 */
/* 3833 */   }
/* 3834 */
...
/* 4532 */   private void apply_0(InternalRow i) {
/* 4533 */
/* 4534 */     boolean isNull2 = i.isNullAt(0);
/* 4535 */     UTF8String value2 = isNull2 ? null : (i.getUTF8String(0));
/* 4536 */     if (!isNull2) {
/* 4537 */       value1 = org.apache.spark.unsafe.hash.Murmur3_x86_32.hashUnsafeBytes(value2.getBaseObject(), value2.getBaseOffset(), value2.numBytes(), value1);
/* 4538 */     }
/* 4539 */
/* 4540 */     boolean isNull3 = i.isNullAt(1);
/* 4541 */     UTF8String value3 = isNull3 ? null : (i.getUTF8String(1));
/* 4542 */     if (!isNull3) {
/* 4543 */       value1 = org.apache.spark.unsafe.hash.Murmur3_x86_32.hashUnsafeBytes(value3.getBaseObject(), value3.getBaseOffset(), value3.numBytes(), value1);
/* 4544 */     }
/* 4545 */
/* 4546 */     boolean isNull4 = i.isNullAt(2);
/* 4547 */     UTF8String value4 = isNull4 ? null : (i.getUTF8String(2));
/* 4548 */     if (!isNull4) {
/* 4549 */       value1 = org.apache.spark.unsafe.hash.Murmur3_x86_32.hashUnsafeBytes(value4.getBaseObject(), value4.getBaseOffset(), value4.numBytes(), value1);
/* 4550 */     }
/* 4551 */
/* 4552 */     boolean isNull5 = i.isNullAt(3);
/* 4553 */     UTF8String value5 = isNull5 ? null : (i.getUTF8String(3));
/* 4554 */     if (!isNull5) {
/* 4555 */       value1 = org.apache.spark.unsafe.hash.Murmur3_x86_32.hashUnsafeBytes(value5.getBaseObject(), value5.getBaseOffset(), value5.numBytes(), value1);
/* 4556 */     }
/* 4557 */
/* 4558 */   }
...
/* 7344 */   public UnsafeRow apply(InternalRow i) {
/* 7345 */     boolean isNull = false;
/* 7346 */
/* 7347 */     value1 = 42;
/* 7348 */     apply_0(i);
/* 7349 */     apply_1(i);
...
/* 7596 */     apply_248(i);
/* 7597 */     apply_249(i);
/* 7598 */     apply_250(i);
/* 7599 */     apply_251(i);
...
````

Add a new test in `DataFrameSuite`

Author: Kazuaki Ishizaki <ishizaki@jp.ibm.com>

Closes #15745 from kiszk/SPARK-18207.

(cherry picked from commit 47731e18)
Signed-off-by: Herman van Hovell <hvanhovell@databricks.com>

ee400f67

ee400f67 8 years ago

Name	Last commit	Last update
.github
R
assembly
bin
build
common
conf
core
data
dev
docs
examples
external
graphx
launcher
licenses
mesos
mllib-local
mllib
project
python
repl
sbin
sql
streaming
tools
yarn
.gitattributes
.gitignore
.travis.yml
CONTRIBUTING.md
LICENSE
NOTICE
README.md
appveyor.yml
pom.xml
scalastyle-config.xml

Apache Spark

Spark is a fast and general cluster computing system for Big Data. It provides high-level APIs in Scala, Java, Python, and R, and an optimized engine that supports general computation graphs for data analysis. It also supports a rich set of higher-level tools including Spark SQL for SQL and DataFrames, MLlib for machine learning, GraphX for graph processing, and Spark Streaming for stream processing.

http://spark.apache.org/

Online Documentation

You can find the latest Spark documentation, including a programming guide, on the project web page and project wiki. This README file only contains basic setup instructions.

Building Spark

Spark is built using Apache Maven. To build Spark and its example programs, run:

build/mvn -DskipTests clean package

(You do not need to do this if you downloaded a pre-built package.)

You can build Spark using more than one thread by using the -T option with Maven, see "Parallel builds in Maven 3". More detailed documentation is available from the project site, at "Building Spark". For developing Spark using an IDE, see Eclipse and IntelliJ.

Interactive Scala Shell

The easiest way to start using Spark is through the Scala shell:

./bin/spark-shell

Try the following command, which should return 1000:

scala> sc.parallelize(1 to 1000).count()

Interactive Python Shell

Alternatively, if you prefer Python, you can use the Python shell:

./bin/pyspark

And run the following command, which should also return 1000:

>>> sc.parallelize(range(1000)).count()

Example Programs

Spark also comes with several sample programs in the examples directory. To run one of them, use ./bin/run-example <class> [params]. For example:

./bin/run-example SparkPi

will run the Pi example locally.

You can set the MASTER environment variable when running examples to submit examples to a cluster. This can be a mesos:// or spark:// URL, "yarn" to run on YARN, and "local" to run locally with one thread, or "local[N]" to run locally with N threads. You can also use an abbreviated class name if the class is in the examples package. For instance:

MASTER=spark://host:7077 ./bin/run-example SparkPi

Many of the example programs print usage help if no params are given.

Running Tests

Testing first requires building Spark. Once Spark is built, tests can be run using:

./dev/run-tests

Please see the guidance on how to run tests for a module, or individual tests.

A Note About Hadoop Versions

Spark uses the Hadoop core library to talk to HDFS and other Hadoop-supported storage systems. Because the protocols have changed in different versions of Hadoop, you must build Spark against the same version that your cluster runs.

Please refer to the build documentation at "Specifying the Hadoop Version" for detailed guidance on building for a particular distribution of Hadoop, including building for particular Hive and Hive Thriftserver distributions.

Configuration

Please refer to the Configuration Guide in the online documentation for an overview on how to configure Spark.

## Contributing

Please review the Contribution to Spark wiki for information on how to get started contributing to the project.