RFR: 8196083: Avoid locking in OopStorage::release
coleen.phillimore at oracle.com
coleen.phillimore at oracle.com
Thu Feb 8 16:10:29 UTC 2018
On 2/8/18 7:50 AM, coleen.phillimore at oracle.com wrote:
>
>
> On 2/8/18 1:37 AM, Kim Barrett wrote:
>>> On Feb 7, 2018, at 9:24 PM, Kim Barrett <kim.barrett at oracle.com> wrote:
>>>
>>>> On Feb 7, 2018, at 4:19 PM, coleen.phillimore at oracle.com wrote:
>>>> I think this logging should be debug mode because I don't know if
>>>> you want this with -Xlog (default is log everything with info mode,
>>>> I think), except this one:
>>>>
>>>> log_info(oopstorage, ref)("%s: failed allocation", name());
>>> If anything, I'm tempted to make that one a warning. Odds are pretty
>>> good that it will lead to some error or abort somewhere up the call
>>> chain.
>>>
>>> I think the oopstorage,blocks log_info's are reasonable.
>>> Other than the one you mention, oopstorage,ref log_info's might be a
>>> bit much, particularly as we start making more use of oopstorage. But
>>> I'd like to consider changes to the logging levels here as a separate
>>> issue.
Yes, we can discuss this as a separate issue. I just happened to
notice it while reading through your change. The lock free release
change looks good.
Thanks,
Coleen
>> "java -Xlog -version” produces over 1100 lines of output. It’s not
>> obvious the
>> existing info output from oopstorage would make a noticeable difference.
>>
>
> Yes, that is why I would like this logging to not be info.
> Thanks,
> Coleen
More information about the hotspot-dev
mailing list