RFR: 8366716: Move SmapsParser from runtime/os/TestTracePageSizes.java into testlib
jonghoonpark
duke at openjdk.org
Mon Sep 15 12:24:49 UTC 2025
On Mon, 15 Sep 2025 06:46:43 GMT, Stefan Johansson <sjohanss at openjdk.org> wrote:
>>> Is this the correct way to verify it?
>>
>> I think @kstefanj , who wrote the RFE and probably intended to implement it, might have ideas on how to verify this.
>
>> Is this the correct way to verify it?
>
> The problem we had in testing ([JDK-8366980](https://bugs.openjdk.org/browse/JDK-8366980)) was that we saw failures when running with `-XX:-UseCompressedOops` so that should be explicitly verified (`-XX:+UseCompressedOops` is the default). Adding it to the process launcher in the code works for testing, but to test without altering the code you can also pass it in. If you run the tests through make:
>
> make test TEST=gc/TestTransparentHugePagesHeap.java JTREG="JAVA_OPTIONS=-XX:-UseCompressedOops"
>
>
> A couple of other comments without doing a proper review is that the new smaps classes should be located along side `HugePageConfiguration` under `lib/os/linux` and I think we want to think about how to structure the classes. Maybe something like this:
>
> public class Smaps {
> public static class Parser { ... }
> public static class Range { ... }
> }
@kstefanj
I've also tested it using the method that doesn't require altering the code, and it passed successfully.
Now, I'll move the path and refactor the code as you reviewed.
-------------
PR Comment: https://git.openjdk.org/jdk/pull/27273#issuecomment-3291891500
More information about the hotspot-dev
mailing list