RFR(S) 8077400: Unnecessary and incorrect "Code Cache Roots" G1 log entry

Thomas Schatzl thomas.schatzl at oracle.com
Fri Apr 10 12:22:48 UTC 2015


Hi Mikael,

On Fri, 2015-04-10 at 14:14 +0200, Mikael Gerdin wrote:
> Hi all,
> 
> When "JDK-8027962 Per-phase timing measurements for strong roots 
> processing" was ported to the G1RootProcessor change it still contained 
> a left over reference to "Code Cache Roots". Code Cache scanning is 
> already done and measured lower in the call chain in G1.
> The current "Code Cache Roots" measurement is actually the sum of the 
> UpdateRS, ScanRS and Code Root Scan times.
> 
> I also took the liberty to remove the comment about the complement of 
> the collection set, since it should be more clear now that the call to 
> oops_into_collection_set_do is in a separate function with a fairly 
> descriptive name.
> 
> I also updated the TestGCLogMessages test to not expect the Code Cache 
> Roots log entry.
> 
> Bug: https://bugs.openjdk.java.net/browse/JDK-8077400
> Webrev: http://cr.openjdk.java.net/~mgerdin/8077400/webrev.0


  looks good.

Thomas




More information about the hotspot-gc-dev mailing list