RFR: 8298377: JfrVframeStream causes deadlocks in ZGC

David Holmes dholmes at openjdk.org
Mon Dec 12 01:19:52 UTC 2022


On Thu, 8 Dec 2022 11:23:57 GMT, Stefan Karlsson <stefank at openjdk.org> wrote:

> The JfrVFrameStream is used while generating stack traces for events. One of the events are the ZPage allocation event. This event is sometimes sent when ZGC is relocating. The current implementation of JfrVFrameStream uses WalkContinuation::include, which causes JFR to walk the continuation and perform GC barriers. This is problematic, since ZGC has a requirement that we never perform load barriers while running the relocation code. If we do, we might end up performing other reloctions from the the relocation code, and in some cases that causes dead locks.
> 
> I propose that JFR doesn't walk the continuations when sending events. An alternative could be to limit this to ZGC, but I'd like to get some feedback around that from JFR / Loom devs.
> 
> We've been testing this patch in the Generational ZGC repository.

It doesn't seem reasonable to me to truncate VT stacks in all JFR events just because there is an issue with one ZGC event.

Any events triggered from the GC internals could potentially try to do things with Java objects that are not permitted given the current state of the GC. This seems like a general problem that the JFR event mechanism needs to resolve.

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

PR: https://git.openjdk.org/jdk/pull/11586


More information about the hotspot-dev mailing list