Thread Local Handshake in JVMTI functions

Yasumasa Suenaga suenaga at oss.nttdata.com
Thu Apr 9 06:52:43 UTC 2020 

On 2020/04/09 15:44, serguei.spitsyn at oracle.com wrote:
> On 4/8/20 23:36, Yasumasa Suenaga wrote:
>> Thanks David and Serguei!
>>
>> I created 3 subtasks under JDK-8201641, of course I will send review request in each them.
>>
>>>   - GetOneCurrentContendedMonitor => GetCurrentContendedMonitor
>>
>> `GetCurrentContendedMonitor` is JVMTI function name, and also it exists as public member of JvmtiEnv class.
>> So I want to different name for HandshakeClosure - for example, GetCurrentContendedMonitorClosure.
>>
>> Do you have any idea for it?
> 
> It is better to follow the same pattern as we already have in other places (until a decision is made about a different rule).
> So, the GetCurrentContendedMonitorClosure should work for now.
> It is interesting if there will be any other suggestions though.

Currently, most of classes which extend HandshakeClosure have "Closure" as suffix in below:

```
$ grep -nrw HandshakeClosure * | grep -w class
share/gc/shenandoah/shenandoahUnload.cpp:161:class ShenandoahUnloadRendezvousClosure : public HandshakeClosure {
share/gc/z/zHeap.cpp:333:class ZRendezvousClosure : public HandshakeClosure {
share/gc/z/zMark.cpp:422:class ZMarkFlushAndFreeStacksClosure : public HandshakeClosure {
share/prims/whitebox.cpp:2046:  class TraceSelfClosure : public HandshakeClosure {
share/runtime/biasedLocking.cpp:505:class RevokeOneBias : public HandshakeClosure {
share/runtime/deoptimization.cpp:808:class DeoptimizeMarkedClosure : public HandshakeClosure {
share/runtime/handshake.hpp:44:class HandshakeClosure : public ThreadClosure {
share/runtime/sweeper.cpp:200:class NMethodMarkingClosure : public HandshakeClosure {
share/runtime/thread.cpp:528:class InstallAsyncExceptionClosure : public HandshakeClosure {
share/runtime/vmThread.cpp:391:class HandshakeALotClosure : public HandshakeClosure {
```

I think `GetCurrentContendedMonitorClosure` is not so bad :)


Thanks,

Yasumasa


> Thanks,
> Serguei
> 
>>
>> Yasumasa
>>
>>
>> On 2020/04/09 15:19, serguei.spitsyn at oracle.com wrote:
>>> Hi Yasumasa,
>>>
>>> I'm okay with using sub-tasks to do it incrementally.
>>>
>>> This needs to be removed with your fix:
>>>    src/hotspot/share/runtime/vmOperations.hpp: template(GetCurrentContendedMonitor)            \
>>>
>>> Also, I agree with comments from David below:
>>>   - GetOneCurrentContendedMonitor => GetCurrentContendedMonitor
>>>   - Thread::current() =>  calling_thread
>>>
>>> I hope, you will update the copyright comments in jvmtiEnvBase.?pp.
>>>
>>> Are you going to re-post this as an RFR with a correct sub-task number?
>>>
>>> Thanks,
>>> Serguei
>>>
>>>
>>> On 4/8/20 22:52, David Holmes wrote:
>>>> Hi Yasumasa,
>>>>
>>>> On 9/04/2020 3:08 pm, Yasumasa Suenaga wrote:
>>>>> Hi,
>>>>>
>>>>> JDK-8240918 has been pushed, so I made a patch for GetCurrentContendedMonitor(). How about this?
>>>>>
>>>>> http://cr.openjdk.java.net/~ysuenaga/jvmti-thread-local-handshake/2/
>>>>
>>>> Generally looks okay. A couple of comments:
>>>>
>>>> src/hotspot/share/prims/jvmtiEnv.cpp
>>>>
>>>> Please update the comment just before the change:
>>>>
>>>>    // thread. All other usage needs to use a vm-safepoint-op for safety.
>>>>
>>>> ---
>>>>
>>>>  GetOneCurrentContendedMonitor
>>>>
>>>> I don't think you need to add One into the name here - the current contended monitor is inherently a singleton.
>>>>
>>>> ---
>>>>
>>>> src/hotspot/share/prims/jvmtiEnvBase.cpp
>>>>
>>>> 655           Thread::current() == java_thread->active_handshaker() ||
>>>>
>>>> The calling_thread parameter is now the current thread. So this can also be changed:
>>>>
>>>> 679     Handle     hobj(Thread::current(), obj);
>>>>
>>>>
>>>>
>>>>>>>>>>>> An observation, it seems to me that calling_thread is not used when this is not a VMOperation.
>>>>>
>>>>> calling_thread is used for creating JNI reference:
>>>>>
>>>>> ```
>>>>> *monitor_ptr = jni_reference(calling_thread, hobj);
>>>>> ```
>>>>>
>>>>>
>>>>> If it is ok, I will make patches for other JVMTI functions.
>>>>> However the patch might be bigger, so I want to separate as below. What do you think?
>>>>> (I think we can file them as subtask under JDK-8201641)
>>>>>
>>>>>
>>>>> * Monitor operation
>>>>>      - VM_GetOwnedMonitorInfo
>>>>>      - VM_GetCurrentContendedMonitor
>>>>>
>>>>> * Frame operation
>>>>>      - VM_UpdateForPopTopFrame
>>>>>      - VM_SetFramePop
>>>>>
>>>>> * Thread operation
>>>>>      - VM_GetFrameCount
>>>>>      - VM_GetFrameLocation
>>>>>      - VM_GetThreadListStackTraces
>>>>>      - VM_GetCurrentLocation
>>>>>      - VM_EnterInterpOnlyMode
>>>>
>>>> Creating subtasks with separate RFRs is fine by me. It may be good to slowly introduce these changes so that we get some testing coverage to iron out any initial bugs with the approach.
>>>>
>>>> Thanks,
>>>> David
>>>> -----
>>>>
>>>>>
>>>>> Thanks,
>>>>>
>>>>> Yasumasa
>>>>>
>>>>>
>>>>> On 2020/04/01 21:29, Yasumasa Suenaga wrote:
>>>>>> Thanks David!
>>>>>>
>>>>>> If JDK-8201641 is not assigned when JDK-8240918 is resolved, I will start to work for JDK-8201641.
>>>>>> (It would be large patch...)
>>>>>>
>>>>>>
>>>>>> Cheers,
>>>>>>
>>>>>> Yasumasa
>>>>>>
>>>>>>
>>>>>> On 2020/04/01 19:05, David Holmes wrote:
>>>>>>> On 1/04/2020 11:02 am, Yasumasa Suenaga wrote:
>>>>>>>> Thanks Dan and Serguei!
>>>>>>>>
>>>>>>>> I added a comment for this to JDK-8201641.
>>>>>>>>
>>>>>>>> David, can you share Bug ID for thread-to-thread handshake?
>>>>>>>> I want to record it to JDK-8201641 as a blocker.
>>>>>>>
>>>>>>> https://bugs.openjdk.java.net/browse/JDK-8240918
>>>>>>>
>>>>>>> I heard the RFR could be as soon as tomorrow :)
>>>>>>>
>>>>>>> Cheers,
>>>>>>> David
>>>>>>>
>>>>>>>>
>>>>>>>> Yasumasa
>>>>>>>>
>>>>>>>>
>>>>>>>> On 2020/04/01 1:59, serguei.spitsyn at oracle.com wrote:
>>>>>>>>> Hi Yasumasa,
>>>>>>>>>
>>>>>>>>> Yes, this works needs to be done.
>>>>>>>>> I'll take look at you webrev.
>>>>>>>>>
>>>>>>>>> Thanks,
>>>>>>>>> Serguei
>>>>>>>>>
>>>>>>>>> On 3/31/20 07:41, Daniel D. Daugherty wrote:
>>>>>>>>>> Add Robbin to this thread...
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>> This reminded of the following RFE that Robbin filed:
>>>>>>>>>>
>>>>>>>>>>     JDK-8201641 JVMTI: GetThreadListStackTraces should use Thread-Local Handshakes
>>>>>>>>>> https://bugs.openjdk.java.net/browse/JDK-8201641
>>>>>>>>>>
>>>>>>>>>> We could update 8201641 to include everything that Yasumasa-san is requesting.
>>>>>>>>>> Would be a good place to track it...
>>>>>>>>>>
>>>>>>>>>> Dan
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>> On 3/31/20 7:40 AM, Yasumasa Suenaga wrote:
>>>>>>>>>>> Hi David,
>>>>>>>>>>>
>>>>>>>>>>> On 2020/03/31 19:16, David Holmes wrote:
>>>>>>>>>>>> Hi Yasumasa,
>>>>>>>>>>>>
>>>>>>>>>>>> On 31/03/2020 8:06 pm, Yasumasa Suenaga wrote:
>>>>>>>>>>>>> Hi all,
>>>>>>>>>>>>>
>>>>>>>>>>>>> Many JVMTI functions uses VM Operation to get information. However some of them need to stop only one thread - they don't need to stop all threads.
>>>>>>>>>>>>> So I think we can use Thread Local Handshake as this webrev. It is example for GetOneCurrentContendedMonitor().
>>>>>>>>>>>>
>>>>>>>>>>>> True, but at the moment handshakes involve the VMThread. There is work being done to support direct thread-to-thread handshakes and once that is done this kind of conversion should be more easily done. It might be worth waiting for that.
>>>>>>>>>>>
>>>>>>>>>>> Thanks, I will be back to this topic when thread-to-thread handshake is done.
>>>>>>>>>>> I wondered at first why VMThread involves handshake. Its improvement is welcome for me ;)
>>>>>>>>>>>
>>>>>>>>>>>
>>>>>>>>>>> Cheers,
>>>>>>>>>>>
>>>>>>>>>>> Yasumasa
>>>>>>>>>>>
>>>>>>>>>>>
>>>>>>>>>>>>> http://cr.openjdk.java.net/~ysuenaga/jvmti-thread-local-handshake/
>>>>>>>>>>>>
>>>>>>>>>>>> An observation, it seems to me that calling_thread is not used when this is not a VMOperation.
>>>>>>>>>>>>
>>>>>>>>>>>> Cheers,
>>>>>>>>>>>> David
>>>>>>>>>>>>
>>>>>>>>>>>>> Also I think we can replace following VM Operations to Thread Local Handshake:
>>>>>>>>>>>>>
>>>>>>>>>>>>> class VM_GetCurrentLocation
>>>>>>>>>>>>> class VM_EnterInterpOnlyMode
>>>>>>>>>>>>> class VM_UpdateForPopTopFrame
>>>>>>>>>>>>> class VM_SetFramePop
>>>>>>>>>>>>> class VM_GetOwnedMonitorInfo
>>>>>>>>>>>>> class VM_GetCurrentContendedMonitor
>>>>>>>>>>>>> class VM_GetFrameCount
>>>>>>>>>>>>> class VM_GetFrameLocation
>>>>>>>>>>>>>
>>>>>>>>>>>>> What do you think?
>>>>>>>>>>>>> It it is acceptable, I will file it to JBS and send review request.
>>>>>>>>>>>>>
>>>>>>>>>>>>>
>>>>>>>>>>>>> Thanks,
>>>>>>>>>>>>>
>>>>>>>>>>>>> Yasumasa
>>>>>>>>>>
>>>>>>>>>
>>>
>

More information about the serviceability-dev mailing list