8248362: JVMTI frame operations should use Thread-Local Handshake

[David Holmes]

On 22/07/2020 6:19 pm, Yasumasa Suenaga wrote:
> On 2020/07/22 16:57, David Holmes wrote:
>> Hi Yasumasa,
>>
>> On 22/07/2020 5:39 pm, Yasumasa Suenaga wrote:
>>> Hi all,
>>>
>>> I'm working for fixing JDK-8248362, but I saw some errors on submit 
>>> repo.
>>> Someone can share the details of 
>>> mach5-one-ysuenaga-JDK-8248362-20200722-0550-12850261 ?
>>>
>>> I wonder why build task of linux-x64 was failed because I can do it 
>>> on my Fedora 32 box.
>>
>> [2020-07-22T06:21:49,141Z] 
>> ./open/src/hotspot/share/prims/jvmtiThreadState.cpp:222:45: error: no 
>> member named 'active_handshaker' in 'JavaThread'
>> [2020-07-22T06:21:49,142Z]             current_thread == 
>> get_thread()->active_handshaker(),
>> [2020-07-22T06:21:49,142Z]                               ~~~~~~~~~~~~  ^
> 
> Thanks David!
> This statement is in guarantee(), so it seems to be failed to build for 
> production VM.
> 
> guarantee() call has been introduced in JDK-6471769, originally it was 
> assert() call.
> Can we replace guarantee() to assert() at this point? or are there 
> methods to detect the call is happened in direct handshake without 
> active_handshaker()?

I would replace with assert. There's no non-debug query for the handshaker.

David
-----

> 
> Thanks,
> 
> Yasumasa
> 
> 
>> David
>> -----
>>
>>>
>>> Thanks,
>>>
>>> Yasumasa