On Tue, 25 Jul 2023 06:28:16 GMT, Thomas Stuefe <stu...@openjdk.org> wrote:
> Trivial patch. > > The test just fires `jcmd System.trim_native_heap`, but without preparing any > special memory allocation conditions. Typically, we end up firing the command > when VM initialization is done, and therefore end up trimming a lot of > memory. But once in a blue moon concurrent allocations may be ongoing, so we > may not trim anything, or even see increased RSS. > > The patch just loosens the success condition to succeed if the trim happened, > irregardless of how much memory had been trimmed. This pull request has now been integrated. Changeset: 117f42db Author: Thomas Stuefe <stu...@openjdk.org> URL: https://git.openjdk.org/jdk/commit/117f42dbe9a78bcf43bdf3873d5d86a19a9092d3 Stats: 1 line in 1 file changed: 0 ins; 0 del; 1 mod 8312625: Test serviceability/dcmd/vm/TrimLibcHeapTest.java failed: RSS use increased Reviewed-by: kevinw, dholmes ------------- PR: https://git.openjdk.org/jdk/pull/15009
