RFR XS: 7092140 Test: TimedAcquireLeak.java fails on SE-E due to -XX:-UsePerfData (original) (raw)

Chris Hegarty chris.hegarty at oracle.com
Wed Mar 7 11:23:32 UTC 2012

I ok with this as is.

I wonder if we should added a note about the divergence, just for future maintenance and re-syncs?

-Chris.

On 07/03/2012 06:55, David Holmes wrote:

webrev:

http://cr.openjdk.java.net/~dholmes/7092140/webrev/ inlined: *** 146,156 **** final String classToCheckForLeaks = Job.classToCheckForLeaks(); final String uniqueID = String.valueOf(new Random().nextInt(Integer.MAXVALUE)); final String[] jobCmd = { ! java, "-Xmx8m", "-classpath", System.getProperty("test.classes", "."), childClassName, uniqueID }; final Process p = new ProcessBuilder(jobCmd).start(); --- 146,156 ---- final String classToCheckForLeaks = Job.classToCheckForLeaks(); final String uniqueID = String.valueOf(new Random().nextInt(Integer.MAXVALUE)); final String[] jobCmd = { ! java, "-Xmx8m", "-XX:+UsePerfData", "-classpath", System.getProperty("test.classes", "."), childClassName, uniqueID }; final Process p = new ProcessBuilder(jobCmd).start(); Trivial fix to add -XX:+UsePerfData on the exec of the VM being checked for leaks. Chris: I've discussed this one with Doug and as we're adding a hotspot specific flag here we don't plan to sync this up with Doug's cvs. As Doug put it: "I think that source divergence is a small price to pay for getting rid of this issue, so feel free. " Thanks, David

More information about the core-libs-dev mailing list