RFR: JDK-8299388: java/util/regex/NegativeArraySize.java fails on Alpine and sometimes Windows [v2]

Matthias Baesken mbaesken at openjdk.org
Thu Dec 29 12:16:51 UTC 2022


On Thu, 29 Dec 2022 10:26:11 GMT, Matthias Baesken <mbaesken at openjdk.org> wrote:

>> The test java/util/regex/NegativeArraySize.java seems to have high memory requirements, and these requirements lead to some errors.
>> On Alpine Linux we run regularly into this error when executing the test:
>> result: Failed. Unexpected exit from test [exit code: 137]
>> This seems to be OOM related.
>> Probably we should avoid running the test on Alpine.
>> 
>> On Windows the test usually works, but seems to depend as well on the memory situation of the machine.
>> Once we got this error recently :
>> OpenJDK 64-Bit Server VM warning: INFO: os::commit_memory(0x00000006c0000000, 5368709120, 0) failed; error='The paging file is too small for this operation to complete' (DOS error/errno=1455)
>> OpenJDK 64-Bit Server VM warning: INFO: os::commit_memory(0x00000006c0000000, 5368709120, 0) failed; error='The paging file is too small for this operation to complete' (DOS error/errno=1455)
>> result: Failed. Unexpected exit from test [exit code: 1]
>> 
>> The hs_err file generated showed :
>> 
>> 
>> # There is insufficient memory for the Java Runtime Environment to continue.
>> # Native memory allocation (mmap) failed to map 5368709120 bytes for G1 virtual space
>> # Possible reasons:
>> # The system is out of physical RAM or swap space
>> # The process is running with CompressedOops enabled, and the Java Heap may be blocking the growth of the native heap
>> 
>> 
>> So it looks like having 5g maxMemory as a requirement is not sufficient for the test (the reported mmap value is already slightly above 5g).
>
> Matthias Baesken has updated the pull request incrementally with one additional commit since the last revision:
> 
>   Remove double whitespace

Hi Martin, thanks for the review !

> Looks reasonable. Let's hear what other people think.

Indeed we might discuss better ways of making the test more reliable.  However it is not so easy to predict / handle all potential error situations of the test that are caused by various memory limitations.

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

PR: https://git.openjdk.org/jdk/pull/11796


More information about the core-libs-dev mailing list