RFR: TestFullGCCount still not quite right

[Roman Kennke]

Oh man. Yes, makes sense.

So in a sense, we went full-circle because it means flag!=false would cover it because it's not the same as flag==true. Right?

Roman


Am 30. November 2018 22:13:31 MEZ schrieb Zhengyu Gu <zgu at redhat.com>:
>By default, ExplicitGCInvokesConcurrent flag is true for Shenandoah, 
>therefore, we need to filter it out when the flag is not specified.
>
>diff -r 8b751872d64a test/hotspot/jtreg/gc/TestFullGCCount.java
>--- a/test/hotspot/jtreg/gc/TestFullGCCount.java        Fri Nov 30 
>16:04:44 2018 -0500
>+++ b/test/hotspot/jtreg/gc/TestFullGCCount.java        Fri Nov 30 
>16:10:37 2018 -0500
>@@ -26,7 +26,7 @@
>   * @bug 7072527
>   * @summary CMS: JMM GC counters overcount in some cases
>   * @requires !(vm.gc == "ConcMarkSweep" & 
>vm.opt.ExplicitGCInvokesConcurrent == true)
>- * @requires !(vm.gc == "Shenandoah"    & 
>vm.opt.ExplicitGCInvokesConcurrent == true)
>+ * @requires !(vm.gc == "Shenandoah"    & 
>(vm.opt.ExplicitGCInvokesConcurrent == true | 
>vm.opt.ExplicitGCInvokesConcurrent == null))
>   * @modules java.management
>   * @run main/othervm -Xlog:gc TestFullGCCount
>   */
>
>
>Test:
>   make CONF=linux-x86_64-server-fastdebug run-test 
>TEST=gc/TestFullGCCount.java 
>JTREG="VM_OPTIONS=-XX:+UnlockExperimentalVMOptions
>-XX:+UseShenandoahGC"
>
>Should not run.
>
>and
>
>make CONF=linux-x86_64-server-fastdebug run-test 
>TEST=gc/TestFullGCCount.java 
>JTREG="VM_OPTIONS=-XX:+UnlockExperimentalVMOptions -XX:+UseShenandoahGC
>
>-XX:-ExplicitGCInvokesConcurrent"
>
>Should run.
>
>Thanks,
>
>-Zhengyu