RFR for JDK-8030089: java/util/zip/ZipFile/FinalizeZipFile.java intermittently fails with fastdebug builds

Alan Bateman Alan.Bateman at oracle.com
Wed Jan 8 12:51:18 UTC 2014

On 08/01/2014 11:01, Tristan Yan wrote:
> Hi Alan
> Maybe my understanding is wrong. Are you saying even in agentvm mode, 
> there will be still different VM for every test. If the answer is yes, 
> when should we use othervm mode?
The purpose of -agentvm is to speed up the test execution by eliding the 
VM startup that otherwise dominates the execution of many short running 
tests. Tests are still executed sequentially, it's just that there isn't 
a VM started up for each test. With concurrency then the tests don't 
execute concurrently in the same VM, instead there is a pool of 
independent VMs that run concurrently. We still need othervm for some 
tests and for varied reasons. One reason is tests that can't restore 
things or clean up,  another reason is tests that require special VM 
options. We moved to -agentvm a few years ago but we still periodically 
find issues where a test does something that causes a test that runs 
subsequently to have a problem.

Anyway, for this test then my point was that it probably doesn't need to 
run in othervm mode, just changing it to await indefinitely should be 
fine (as it's await will timeout would require a long timeout when 
running with a fastdebug build).


More information about the core-libs-dev mailing list