RFR(S/M): 8246476: remove AsyncDeflateIdleMonitors option and the safepoint based deflation mechanism

Patricio Chilano patricio.chilano.mateo at oracle.com
Mon Jul 13 16:05:09 UTC 2020


Hi Dan,

Changes look good to me!

In synchronizer.cpp we have this comment about ObjectMonitor lifecycle:

// Inflation unlinks monitors from om_list_globals._free_list or a 
per-thread
// free list and associates them with objects. Deflation -- which occurs at
// STW-time or asynchronously -- disassociates idle monitors from objects.
// Such scavenged monitors are returned to the om_list_globals._free_list.

With all the older code removed, are there still cases where we do 
deflations at safepoint?

Thanks!
Patricio
On 7/13/20 10:47 AM, Daniel D. Daugherty wrote:
> Hi David,
>
> Thanks for the review!
>
> I need a second review folks... any takers?
>
> Dan
>
>
> On 7/12/20 10:57 PM, David Holmes wrote:
>> Hi Dan,
>>
>> This all looks good to me.
>>
>> Thanks,
>> David
>> -----
>>
>> On 8/07/2020 5:51 pm, David Holmes wrote:
>>> On 8/07/2020 7:21 am, Daniel D. Daugherty wrote:
>>>> Ping! Any takers??? Code deletion should be really appealing here!!
>>>
>>> Sorry Dan didn't get to it before vacation. But if you can wait till 
>>> Monday ...
>>>
>>> Cheers,
>>> David
>>>
>>>> Dan
>>>>
>>>>
>>>> On 7/6/20 12:35 PM, Daniel D. Daugherty wrote:
>>>>> Greetings,
>>>>>
>>>>> It's time to remove the AsyncDeflateIdleMonitors option from 
>>>>> JDK16. We can
>>>>> also get rid of the safepoint based deflation mechanism since 
>>>>> turning off
>>>>> async deflation (-XX:-AsyncDeflateIdleMonitors) was the only way 
>>>>> left to
>>>>> use it.
>>>>>
>>>>> This is marked as an "S/M" review because the number of 
>>>>> touched/deleted
>>>>> lines makes it a Medium review, but the number of touched/changed 
>>>>> lines
>>>>> (outside of the deletions) makes it a Small review. It's actually 
>>>>> a pretty
>>>>> fast read... :-)
>>>>>
>>>>> Here's the bug ID:
>>>>>
>>>>>     JDK-8246476 remove AsyncDeflateIdleMonitors option and the 
>>>>> safepoint
>>>>>                 based deflation mechanism
>>>>>     https://bugs.openjdk.java.net/browse/JDK-8246476
>>>>>
>>>>> Here's the webrev URL:
>>>>>
>>>>> http://cr.openjdk.java.net/~dcubed/8246476-webrev/0_for_jdk16/
>>>>>
>>>>> The webrev is baselined on Thomas S's fix for 8248650 which is 
>>>>> jdk-16+4
>>>>> plus a dozen or so changesets.
>>>>>
>>>>> This change has been tested with Mach5 Tier[1-3],4,5,6,7,8 and 
>>>>> there are
>>>>> no regressions (and very few known failures). My inflation stress 
>>>>> testing
>>>>> is still in process. I had to restart that testing after a 
>>>>> thunderstorm
>>>>> related power failure took down my servers in Florida. Sigh...
>>>>>
>>>>> Thanks, in advance, for any comments, questions, or suggestions.
>>>>>
>>>>> Dan
>>>>
>



More information about the hotspot-runtime-dev mailing list