RFR: 8348402: PerfDataManager stalls shutdown for 1ms [v3]

Aleksey Shipilev shade at openjdk.org
Fri Jan 31 08:05:54 UTC 2025


On Fri, 31 Jan 2025 04:38:17 GMT, David Holmes <dholmes at openjdk.org> wrote:

>> Aleksey Shipilev has updated the pull request with a new target base due to a merge or a rebase. The incremental webrev excludes the unrelated changes brought in by the merge/rebase. The pull request contains six additional commits since the last revision:
>> 
>>  - Merge branch 'master' into JDK-8348402-perfdatamanager-stall
>>  - Also add a blurb in usage doc
>>  - Only protect one dodgy place
>>  - Merge branch 'master' into JDK-8348402-perfdatamanager-stall
>>  - Fix minimal build
>>  - Use GlobalCounter to coordinate deletions
>
> src/hotspot/share/runtime/objectMonitor.cpp line 946:
> 
>> 944:     // Note that the counter is not protected by a lock or updated by atomics.
>> 945:     // That is by design - we trade "lossy" counters which are exposed to
>> 946:     // races during updates for a lower probe effect.
> 
> I would suggest restoring the first three deleted lines as the counter update is still racy/lossy.

Restored.

> src/hotspot/share/runtime/objectMonitor.cpp line 1075:
> 
>> 1073:     // Note that the counter is not protected by a lock or updated by atomics.
>> 1074:     // That is by design - we trade "lossy" counters which are exposed to
>> 1075:     // races during updates for a lower probe effect.
> 
> I would suggest restoring the first three deleted lines as the counter update is still racy/lossy.

Restored.

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

PR Review Comment: https://git.openjdk.org/jdk/pull/23293#discussion_r1936823572
PR Review Comment: https://git.openjdk.org/jdk/pull/23293#discussion_r1936822604


More information about the hotspot-runtime-dev mailing list