[jdk17u-dev] RFR: 8369992: JFR: Disable Placeholder-, LoaderConstraints- and ProtectionDomainCacheTableStatistics in default.jfc

[Goetz Lindenmaier]

On Tue, 28 Oct 2025 09:03:08 GMT, Matthias Baesken <mbaesken at openjdk.org> wrote:

>> I backport this for parity with 17.0.18-oracle.
>> 
>> First, there was the attempt to delete these events, see backport of https://bugs.openjdk.org/browse/JDK-8291733: Remove JFR events that expose hashtable
>> 
>> @eghalin pointed out to me that Oracle reverts this (See [JDK-8369975](https://bugs.openjdk.org/browse/JDK-8369975) [BACKOUT BACKPORT] Remove JFR events that expose hashtable) and instead disables the events, 
>> see also comment in https://github.com/openjdk/jdk17u-dev/pull/4047 
>> 
>> This now implements the backport according to that comment.
>> 
>> Thanks again, Eric, for informing about this.
>
> The JBS issue https://bugs.openjdk.org/browse/JDK-8369992 says the events 'add no value' and 'may cause performance regressions'.
> So I wonder why we keep them in profile.jfc , at least in this commit  profile.jfc was touched too
> https://github.com/openjdk/jdk/commit/b7d2bde8e37817e827e685bd675b075c1f29b4c2
> 
> (in jdk head they are gone completely).

Hi @MBaesken 

JDK-8291733 removed the feature in 20.  Therefore the events 
are gone in head, obviously.
Also, there was the attempt to backport the removal.
But Oracle then decided that this is too intrusive for a 
update release.
In this comment https://github.com/openjdk/jdk17u-dev/pull/4047#issuecomment-3408937611
in the PR where I attempted to backport the removal to 17,
Eric advises me to do it exactly like I did in this change.
In some mail to me he also mentioned that they keep it in profile.jfc
so that people have the well known experience here.

Best, Goetz.

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

PR Comment: https://git.openjdk.org/jdk17u-dev/pull/4108#issuecomment-3455538002