RFR: 8343782: G1: Use one G1CardSet instance for multiple old gen regions [v12]

[Ivan Walulya]

On Wed, 11 Feb 2026 12:42:36 GMT, cxzl25 <duke at openjdk.org> wrote:

>> @luben, could you please file an issue at https://bugreport.java.com/  or send an email to hotspot-gc-dev at openjdk.org  so we can follow up on this outside of the current PR? When filing, please include any relevant hs_err logs or logs with verification enabled to help with diagnosing the issue. Additionally, could you test with JDK 26 to confirm whether the issue persists in later versions?
>
> @walulyai Following your advice, I began to attempt validation testing on JDK 26, and ultimately found that this bug was resolved by the ticket JDK-8370807. However, it has not been backported to JDK 25, and its type is Enhancement.
> 
> Do I still need to send an email to report this issue?
> 
> ---
> 8370807: G1: Improve region attribute table method naming
> https://bugs.openjdk.org/browse/JDK-8370807
> Fix Version/s: 26
> Resolved In Build:b22
> https://github.com/openjdk/jdk/pull/28038
> 
> ---
> I suspect that `r->has_pinned_objects()` is the key fix.
> 
> <img width="1557" height="357" alt="Image" src="https://github.com/user-attachments/assets/53f7afaf-7c3a-4b40-b537-75b278daed12" />
> 
> ---
> 
> I have cherry‑picked commit 17fd801b onto 86cec4ea, and it has been verified in the production environment that this fixes the zstd‑jni data corruption issue.
> 
> https://github.com/cxzl25/jdk/commits/test_g1/

@cxzl25, thank you for the investigation. There is no need to send the email at this time; we will look into how to backport the fix to 25.

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

PR Comment: https://git.openjdk.org/jdk/pull/22015#issuecomment-3884658022