RFR: 8293167: Memory leak in JfrThreadSampler if stackdepth is larger than default (64)

Jaroslav Bachorik jbachorik at openjdk.org
Fri Sep 2 16:12:52 UTC 2022


On Thu, 1 Sep 2022 20:40:08 GMT, Markus Grönlund <mgronlun at openjdk.org> wrote:

> Greetings,
> 
> please help review this fix that address a memory leak in the JfrThreadSampler when using a stackdepth larger than default (64).
> 
> Please see the JIRA issue for details.
> 
> Testing: jdk_jfr, stress
> 
> Thanks
> Markus

Looking good.
I could find only one small error in a comment :)

src/hotspot/share/jfr/recorder/checkpoint/types/traceid/jfrTraceIdLoadBarrier.cpp line 37:

> 35: // The queue instances are used by the load barrier to enqueue tagged Klass'es.
> 36: static JfrTraceIdKlassQueue* _klass_queue = nullptr; // Generic for all Java threads.
> 37: static JfrTraceIdKlassQueue* _sampler_klass_queue = nullptr; // Specialized for the Jfr Thread Sampler using use a larger buffer size.

Suggestion:

static JfrTraceIdKlassQueue* _sampler_klass_queue = nullptr; // Specialized for the Jfr Thread Sampler using a larger buffer size.

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

Marked as reviewed by jbachorik (Reviewer).

PR: https://git.openjdk.org/jdk/pull/10129


More information about the hotspot-jfr-dev mailing list