[11u] RFR: 8208677: Move inner metaspace cleaning out of class unloading

[Ekaterina Vergizova]

Hi Goetz,
I agree that Oracle probably fixed 8251945 in a different way, although JDK-8210422 has an unresolved backport in 11-pool.
But as sometimes discussed (e.g. [1]), it would be preferable to backport changes that were tested in the parent repos rather than to make manual unsafe changes.
That's why I suggested that approach. If somebody can suggest an alternative safe fix it would be great.

>From my side, I'm ready to provide the full list of suggested backports. All of them passed through extensive testing including full regression on multiple platforms.

Thanks,
Ekaterina

[1] http://mail.openjdk.java.net/pipermail/jdk8u-dev/2020-January/010997.html


-----Original Message-----
From: Lindenmaier, Goetz <goetz.lindenmaier at sap.com> 
Sent: Thursday, October 1, 2020 2:45 PM
To: Ekaterina Vergizova <katya at azul.com>; jdk-updates-dev at openjdk.java.net
Subject: RE: [11u] RFR: 8208677: Move inner metaspace cleaning out of class unloading

Hi Ekaterina,

I understand you want to downport 5 changes to fix https://bugs.openjdk.java.net/browse/JDK-8251945,
starting with this one. 

8251945 is fixed in 11.0.10-oracle. But Oracle did not mark these five changes as downported, as they usually would do.
So I assume they applied a tailored fix to 11 we unfortunately can not see, but they did not downport these 5 changes. 

This particular change here seem quite risky to me. 
It changes the behavior of the VM by scheduling tasks differently and changing safe points.

Isn't it possible to fix the issue more defensively?

Best regards,
  Goetz.



> -----Original Message-----
> From: jdk-updates-dev <jdk-updates-dev-retn at openjdk.java.net> On 
> Behalf Of Ekaterina Vergizova
> Sent: Wednesday, September 23, 2020 11:10 PM
> To: jdk-updates-dev at openjdk.java.net
> Subject: [11u] RFR: 8208677: Move inner metaspace cleaning out of 
> class unloading
> 
> Hello,
> I would like to backport 8208677 to 11u as a prerequisite for 8210422.
> 8210422 fixes the crash described in 8251945 that is already fixed in 
> 11.0.10- oracle.
> The full list of suggested prerequisites is specified in [1].
> 
> JBS: https://bugs.openjdk.java.net/browse/JDK-8208677
> Original patch: http://hg.openjdk.java.net/jdk/jdk/rev/aa3bfacc912c
> Webrev for 11u:
> https://cr.openjdk.java.net/~apetushkov/8208677/webrev.00/
> 
> The patch applies almost cleanly except for the change in 
> src/hotspot/share/code/nmethod.cpp.
> It is already included in 11u by 8220718.
> 
> Tested with tier1; the full list of backports specified in [1] is 
> tested with full regression.
> 
> Thanks,
> Ekaterina
> 
> [1] https://bugs.openjdk.java.net/browse/JDK-
> 8251945?focusedCommentId=14363860&page=com.atlassian.jira.plugin.sys
> tem.issuetabpanels%3Acomment-tabpanel#comment-14363860