RFR: 8334026: Provide a diagnostic PrintMemoryMapAtExit switch on Linux [v2]
Thomas Stuefe
stuefe at openjdk.org
Thu Jun 20 08:35:25 UTC 2024
On Tue, 18 Jun 2024 09:23:43 GMT, David Holmes <dholmes at openjdk.org> wrote:
>> It is needed because in this function we call Thread::name() which for java threads returns an RA-allocated string. This issue popped up now because we are called without a ResourceMark at VM exit. I have not investigated closer since I think the Resourcemark makes sense here: The function does not export any RA-allocated memory but may be called in a loop many times.
>
> Okay. I was trying to find where the existing callers have their RM and it is not obvious.
Existing caller is DCmd::parse_and_execute, where we have an RM that guards every individual Dcmd execution. So, this had been a small leak of sorts, since we accumulate thread names for all java threads printed.
-------------
PR Review Comment: https://git.openjdk.org/jdk/pull/19660#discussion_r1647192920
More information about the hotspot-runtime-dev
mailing list