RFR: 8151556: Use the PreservedMarks* classes for the G1 preserved mark stacks

Tony Printezis tprintezis at twitter.com
Wed Apr 20 13:45:17 UTC 2016


Thanks Thomas for looking at it.

I had completely missed the "To-space exhausted" message (was looking for promotion / evacuation failure):

[1.194s][info][gc           ] GC(3) To-space exhausted

I kinda prefer “To-space exhausted” I have to say (maybe, it’s not always promotion to old that fails). I’d vote to leave it as is. If we want consistency, maybe we change the “Promotion failed” messages to “To-space exhausted” (on a separate CR, though)?

Tony

On April 19, 2016 at 12:34:50 PM, Thomas Schatzl (thomas.schatzl at oracle.com) wrote:

Hi,  

On Wed, 2016-04-13 at 09:06 -0400, Tony Printezis wrote:  
> Hi all,  
>  
> First, thanks to Thomas for pushing the patch for 8152312. Here’s a  
> new webrev with the refactored G1 changes:  
>  
> http://cr.openjdk.java.net/~tonyp/8151556/webrev.1/  
>  

looks good to me. Testing (jprt, vm.gc testlist) looks fine.  

> Related question: Does G1 report on the GC log that an evacuation  
> failure has occurred, the same way all the other GCs report promotion  
> failures?  
>  
> [6.319s][info][gc,promotion ] Promotion failed  
>  
> Should I add an extra message as part of this change?  

G1 reports promotion failure by indicating this in the GC kind message  
(like "Pause Young (to-space-exhausted)").  

It would be nice to add this message to get consistent output for all  
gcs though.  

Thanks,  
Thomas  

-----

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

@TonyPrintezis
tprintezis at twitter.com

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.openjdk.java.net/pipermail/hotspot-gc-dev/attachments/20160420/f04eb764/attachment-0001.html>


More information about the hotspot-gc-dev mailing list