RFR: 8343298: Improve stability of runtime/cds/DeterministicDump.java test [v2]

[Thomas Schatzl]

On Mon, 4 Nov 2024 12:15:54 GMT, Aleksey Shipilev <shade at openjdk.org> wrote:

>> Thomas Schatzl has updated the pull request incrementally with one additional commit since the last revision:
>> 
>>   * shade review comments
>
> test/hotspot/jtreg/runtime/cds/DeterministicDump.java line 56:
> 
>> 54:         baseArgs.add("-Xmx128M");
>> 55:         baseArgs.add("-Xms128M");
>> 56:         baseArgs.add("-Xmn128M");
> 
> Is it legal/useful to set `Xmn` = `Xmx`?

The GC will just ignore you. It is useful to set it to a high value since it avoids any GC until the CDS dumps happen, removing randomness.

I changed the code to have a slightly smaller `-Xmn` than `-Xms`, and removed all other options. This is sufficient to avoid any GCs until CDS dump time; I do not want to dive into how to make G1 GCs as deterministic as possible here if there are no GCs at all.

(The options are renmants from earlier attempts to make the CDS dump as deterministic as possible in the presence of GCs. After not succeeding (but making issues rare), checking with @iklam showed that it is not the point to make GC deterministic, but it is fine to just pass the test).

> test/hotspot/jtreg/runtime/cds/DeterministicDump.java line 57:
> 
>> 55:         baseArgs.add("-Xms128M");
>> 56:         baseArgs.add("-Xmn128M");
>> 57:         baseArgs.add("-XX:ParallelGCThreads=1");
> 
> `-XX:ConcGCThreads=1` as well, maybe?

See above.

> test/hotspot/jtreg/runtime/cds/DeterministicDump.java line 59:
> 
>> 57:         baseArgs.add("-XX:ParallelGCThreads=1");
>> 58:         baseArgs.add("-XX:-ResizeTLAB");
>> 59:         baseArgs.add("-XX:-ResizePLAB");
> 
> I am guessing we do not want to `-XX:-UseTLAB` summarily, so we get at least the same sized LABs all the time?

See above.

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

PR Review Comment: https://git.openjdk.org/jdk/pull/21871#discussion_r1827805603
PR Review Comment: https://git.openjdk.org/jdk/pull/21871#discussion_r1827805922
PR Review Comment: https://git.openjdk.org/jdk/pull/21871#discussion_r1827806058