RFR: 8231214: PrintInlining should avoid intermixing of messages

[Tobias Hartmann]

Hi Jie,

On 23.09.19 05:51, Jie Fu wrote:
> After digging into the code, I had to admit that PrintInlining is much more complicated than I had
> thought.
> The inlining messages are dumped at many sites line by line, which is quite different with
> CIPrintCompileQueue.
> It makes no sense to add ttyLock at each site.

It's very unfortunate that the inlining code is so distributed and can't be easily fixed. In
general, the workaround for avoiding intermixing of these inlining statements is to run with a
single compiler thread.

> I'm sorry to waste your time.
> Could this issue be closed as won't fix?

That's fine with me. Just go ahead, add a comment and close the issue.

Best regards,
Tobias
> On 2019/9/20 下午4:32, Tobias Hartmann wrote:
>> Hi Jie,
>>
>> print_inlining_stream()->print() is also called at other places. Shouldn't we lock there as well?
>>
>> Best regards,
>> Tobias
>>
>> On 19.09.19 07:54, Jie Fu wrote:
>>> Hi all,
>>>
>>> JBS:    https://bugs.openjdk.java.net/browse/JDK-8231214
>>> Webrev: http://cr.openjdk.java.net/~jiefu/8231214/webrev.00/
>>>
>>> When analyzing the inlining status of the compilation, the output of PrintInlining was found to be
>>> intermixed.
>>> It might be better to dump the inlining messages together in one block.
>>>
>>> Testing:
>>>   - make test TEST="tier1" CONF=release   on Linux/x64
>>>   - make test TEST="tier1" CONF=fastdebug on Linux/x64
>>>
>>> Could you please review it and give me some advice?
>>>
>>> Thanks a lot.
>>> Best regards,
>>> Jie
>>>
>>>
>