[jdk20] RFR: 8298061: vmTestbase/nsk/sysdict/vm/stress/btree/btree012/btree012.java failed with "fatal error: refcount has gone to zero" [v2]

[David Holmes]

On Mon, 19 Dec 2022 22:18:07 GMT, Coleen Phillimore <coleenp at openjdk.org> wrote:

>> Please review this limited fix to make symbol refcount not go to zero for the placeholder supername field. This supername field is to detect ClassCircularityError for parallel capable class loaders without deadlocking.
>> ie. t1 : Class A (lock object for A) -> super B (wait for lock object for B)
>> t2: Class B (lock object for B) -> super A (wait for lock object for A).
>> 
>> We get the supername field out of the placeholder table to do the check for super class CCE but the reference count in the placeholder table for this name can be one. The fix is not to decrement and increment if the name is the same.
>> 
>> Retesting with tier1-4 in progress.
>
> Coleen Phillimore has updated the pull request incrementally with one additional commit since the last revision:
> 
>   Revert change in clear_supername() that doesn't strictly fix the bug.

Seems reasonable. Thanks.

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

Marked as reviewed by dholmes (Reviewer).

PR: https://git.openjdk.org/jdk20/pull/58