G1: Lightweight remembered set logging

Thomas Schatzl thomas.schatzl at oracle.com
Fri Apr 29 08:28:45 UTC 2022


Hi,

On 28.04.22 15:41, Tianqi Xia wrote:
> Hi,
> 
> I tried the master branch, and was very glad to find both the rset memory
> footprint and the cost of logging significantly decreased.
> 

Happy to hear :)

> Later I came across this issue:
> https://bugs.openjdk.java.net/browse/JDK-8224839 in the bug database, it is
> a little bit outdated, but it still seems to hold its value to have a
> brief, maybe one-line summary of rset info on every GC cycle.
> 
> I tried to implement it myself:
> https://github.com/tqxia/jdk/commit/85a0e227b15973685af36722df654c43232e709f
> 
> And here is a sample output:
> [1.783s][info][gc,remset   ] GC(0) Before GC RS summary: Total per region
> rem sets sizes = 608384 Max = 8784 wasted = 0
> ...
> [1.991s][info][gc,remset   ] GC(0) After GC RS summary: Total per region
> rem sets sizes = 607368 Max = 8112 wasted = 0
> 
> May I have some suggestions on this?

Make a pull request (and sign the OCA if not done so yet)? The output 
seems okay to me.

Thanks,
   Thomas



More information about the hotspot-gc-dev mailing list