RFR (S) 8049304: race between VM_Exit and _sync_FutileWakeups->inc()
Daniel D. Daugherty
daniel.daugherty at oracle.com
Fri Sep 4 13:03:10 UTC 2015
On 9/3/15 5:45 PM, Kim Barrett wrote:
> On Sep 3, 2015, at 11:50 AM, Daniel D. Daugherty <daniel.daugherty at oracle.com> wrote:
>> Kim,
>>
>> Wow... Not at all where I was expecting this review to go. Normally
>> I continue a review cycle until I can negotiate agreement amongst
>> all the reviewers, but I can see that's just not going to happen
>> with this review.
>>
>> Short version:
>> I'll take some of the feedback from this review and I'll make
>> some minor changes, but I have no plans to change the code to
>> address most of these comments.
>>
>> Slightly longer version:
>> The fix that I have in hand resolves the issues that I was
>> trying to fix. It also resolves the issue described in
>> JDK-8129978. Additional changes will require a new bug or,
>> if you wish, you can reopen this bug:
>>
>> JDK-8129978 SIGSEGV when parsing command line options
>> https://bugs.openjdk.java.net/browse/JDK-8129978
>>
>> and use it to make additional changes. I will be on vacation
>> next week and David Holmes will be on vacation for a month
>> starting next week. Any further work in this area should
>> wait until David returns since he has strong opinions on
>> this code.
> For the record, I'm OK with this change having been pushed.
Thanks! I was hoping that would be the case...
> While I'm
> at least at present still in disagreement with Dan about the predicate
> being used to decide whether to destroy the PerfData, this change is
> an improvement on the status quo, and that predicate can be fixed
> separately if needed.
>
> I have some additional responses to Dan, but might not finish them
> before he goes on vacation, so wanted to get this reply out before
> then. Have a great vacation.
Thanks!That's the plan...
Dan
More information about the serviceability-dev
mailing list