RFR: 8356152: String.concat can throw StringIndexOutOfBoundsException [v4]

[Raffaello Giulietti]

On Mon, 5 May 2025 17:11:46 GMT, Andrey Turbanov <aturbanov at openjdk.org> wrote:

>> test/jdk/java/lang/String/concat/HugeConcatTest.java line 38:
>> 
>>> 36: import static org.junit.Assert.assertThrows;
>>> 37: 
>>> 38: public class HugeConcatTest {
>> 
>> I've tried to execute the test with reverting changes in `StringConcatHelper` - it doesn't fail.
>> Added logging and actually OOM happened due to heap size:
>> 
>> Reall got OOM: java.lang.OutOfMemoryError: Overflow: String length out of range
>> Reall got OOM: java.lang.OutOfMemoryError: Overflow: String length out of range
>> Reall got OOM: java.lang.OutOfMemoryError: Java heap space
>> Reall got OOM: java.lang.OutOfMemoryError: Java heap space
>> Reall got OOM: java.lang.OutOfMemoryError: Java heap space
>> Reall got OOM: java.lang.OutOfMemoryError: Java heap space
>> Reall got OOM: java.lang.OutOfMemoryError: Overflow: String length out of range
>> Reall got OOM: java.lang.OutOfMemoryError: Overflow: String length out of range
>> ----------System.err:(13/854)----------
>> STARTED    HugeConcatTest::testConcat_UTF16_UTF16 'testConcat_UTF16_UTF16()'
>> SUCCESSFUL HugeConcatTest::testConcat_UTF16_UTF16 'testConcat_UTF16_UTF16()' [9ms]
>> STARTED    HugeConcatTest::testConcat_Latin1_UTF16 'testConcat_Latin1_UTF16()'
>> SUCCESSFUL HugeConcatTest::testConcat_Latin1_UTF16 'testConcat_Latin1_UTF16()' [657ms]
>> STARTED    HugeConcatTest::testConcat_UTF16_Latin1 'testConcat_UTF16_Latin1()'
>> SUCCESSFUL HugeConcatTest::testConcat_UTF16_Latin1 'testConcat_UTF16_Latin1()' [52ms]
>> STARTED    HugeConcatTest::testConcat_Latin1_Latin1 'testConcat_Latin1_Latin1()'
>> SUCCESSFUL HugeConcatTest::testConcat_Latin1_Latin1 'testConcat_Latin1_Latin1()' [1ms]
>> 
>> 
>> I think we should validate error message too. Not only fact that it was thrown.
>
> _Really_ fails with Xmx7G

Yes, the required heap size was not enough for the reverted variant to throw a non-OOME exception/error.

Just out of curiosity, is JDK-8356152 a real-world issue you stumbled upon, or did you encounter it during some testing of yours?

-------------

PR Review Comment: https://git.openjdk.org/jdk/pull/25038#discussion_r2073899613