RFR: 8306767: Concurrent repacking of extra data in MethodData is potentially unsafe [v16]

[Emanuel Peter]

On Mon, 15 Jan 2024 12:29:20 GMT, Emanuel Peter <epeter at openjdk.org> wrote:

>> src/hotspot/share/ci/ciMethodData.cpp line 135:
>> 
>>> 133: 
>>> 134:   // Lock to read ProfileData, and ensure lock is not unintentionally broken by a safepoint
>>> 135:   MutexLocker ml(mdo->extra_data_lock(), Mutex::_no_safepoint_check_flag);
>> 
>> Is there anyway to have MutexLocker take care of verifying that the there's no safepoint? It would be nice to replace:
>> 
>> 
>> MutexLocker ml();
>> NoSafepointVerifier no_safepoint;
>> 
>> 
>> by: 
>> 
>> 
>> MutexLocker ml();
>> 
>> 
>> only.
>
> @rwestrel I introduced a `NoSafepointMutexLocker`, which is composed of both a `ConditionalMutexLocker` and a `NoSafepointVerifier`.

But in this specific instance I keep them separate, because I need to pass the `NoSafepointVerifier` down, so that I can pass it to the `PauseNoSafepointVerifier`.

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

PR Review Comment: https://git.openjdk.org/jdk/pull/16840#discussion_r1452320637