RFR[9]: 8179084: HotSpot VM fails to start when AggressiveHeap is set

Erik Helin erik.helin at oracle.com
Wed Apr 26 16:42:14 UTC 2017


Hi Kim,

thanks for taking care of this bug!

On 04/25/2017 07:10 AM, Kim Barrett wrote:
> Please review this change to eliminate a VM startup failure when the
> -XX:+AggressiveHeap option is used without a preceeding
> -XX:ParallelGCThreads=<n> option.
> 
> In a debug build that configuration fails with:
> 
> #  assert(_initial_active_processor_count > 0) failed: Initial active processor count not set yet.
> 
> In a release build that configuration fails with java failing to
> start, with a message of:
> The Parallel GC can not be combined with -XX:ParallelGCThreads=0
> 
> So even the simplest of tests would have detected this problem.
> Unfortunately, there are no tests in the jdk forest involving
> -XX:+AggressiveHeap. This change adds a minimal test.
> 
> CR:
> https://bugs.openjdk.java.net/browse/JDK-8179084
> 
> Webrev:
> http://cr.openjdk.java.net/~kbarrett/8179084/hotspot.00/

The patch looks good, Reviewed. Great that you added a regression test!

> Testing:
> JPRT, including new test.
> Locally executed new test with fastdebug and release builds:
> - without change in fastdebug, verified the expected assert
> - iwhtout change in release, verified the failure from the bug report
> - with change in either, verified pass

Thanks for the thorough testing,
Erik



More information about the hotspot-gc-dev mailing list