RFR 8146991: Introduce per-worker preserved mark stacks in ParallelGC (original) (raw)

Tony Printezis tprintezis at twitter.com
Fri Mar 4 00:15:12 UTC 2016


Thomas (and all),

Updated webrev with the changes below:

http://cr.openjdk.java.net/~tonyp/8146991/webrev.1/

Tony

On March 3, 2016 at 7:02:03 PM, Tony Printezis (tprintezis at twitter.com) wrote:

Thomas.

Thanks for looking at it. See inline...

On March 3, 2016 at 9:35:26 AM, Thomas Schatzl (thomas.schatzl at oracle.com) wrote:

Hi Tony, 

On Tue, 2016-03-01 at 13:22 -0500, Tony Printezis wrote: 

Hi all, 

These are the per-worker preserved mark stack changes for ParallelGC:  http://cr.openjdk.java.net/~tonyp/8146991/webrev.0/  They are similar to the ParNew changes and rely on some classes I  introduced then. 

 

Done.

 

Done.

 

I can accumulate the size incrementally, so the output should be done at the end of the method then?

Like PreservedMarks::get() returned the number of preserved marks, and  the method just added them together. 

 

I’ve always done it this way, ++i / i++ as a statement just looks weird...

Quick question on this: I moved: 

logtrace(gc, ergo)("Restoring " SIZEFORMAT " marks", totalsize());  to the shared code so that the message is logged irrespective of GC  (for example, I don’t think we were logging this for ParNew or  DefNew). Is “ergo” the right category here though? 

 

Done. I’ll post a new webrev shortly…

Tony

Otherwise it looks good to me. I will run it through jprt and some test  list. 

Thanks,  Thomas 


Tony Printezis | JVM/GC Engineer / VM Team | Twitter

@TonyPrintezis tprintezis at twitter.com


Tony Printezis | JVM/GC Engineer / VM Team | Twitter

@TonyPrintezis tprintezis at twitter.com

-------------- next part -------------- An HTML attachment was scrubbed... URL: <https://mail.openjdk.org/pipermail/hotspot-gc-dev/attachments/20160303/a0e820e4/attachment.htm>



More information about the hotspot-gc-dev mailing list