[jdk21u-dev] RFR: 8351500: G1: NUMA migrations cause crashes in region allocation

[Thomas Stuefe]

On Thu, 13 Mar 2025 16:47:52 GMT, Thomas Stuefe <stuefe at openjdk.org> wrote:

> This pull request contains a backport of commit [37ec7962](https://github.com/openjdk/jdk/commit/37ec796255ae857588a5c7e0d572407dd81cbec9) from the [openjdk/jdk](https://git.openjdk.org/jdk) repository.
> 
> The commit being backported was authored by Thomas Stuefe on 13 Mar 2025 and was reviewed by Roman Kennke, Stefan Johansson and Thomas Schatzl.
> 
> ---
> 
> Please consider this patch for backporting. It fixes a G1 crash we see at a customer with a very large NUMA installation.
> 
> The backported patch was in itself not sufficient. For JDK 21 we also must fix up `G1Allocator::attempt_allocation_force`, see commit [7ee2df2ac9c14599a274250dd8ff58c7e6979611](https://github.com/openjdk/jdk21u-dev/pull/1488/commits/7ee2df2ac9c14599a274250dd8ff58c7e6979611) to use the NUMA node index handed in instead of querying libnuma on its own. 
> 
> Testing: I tested the patch with a change mimicking tons of NUMA node migrations. Without patch we get crashes/asserts, with patch everything works (I ran a modified version of TestJNICriticalStressTest.java, since that exercises the crashing paths).

After discussions with my colleagues, it was decided not to do this as a critical patch, but as a normal patch for the JDK 21 July update. Therefore, I resurrect this PR and will send the one targeting the critical patch repo (https://github.com/openjdk/jdk21u/pull/461). 

Sorry for all this confusion.

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

PR Comment: https://git.openjdk.org/jdk21u-dev/pull/1488#issuecomment-2757044326