RFR (S) : 8239557 : [TESTBUG] VeryEarlyAssertTest.java validating "END." marker at lastline is not always true

Fairoz Matte fairoz.matte at oracle.com
Mon Feb 24 04:18:52 UTC 2020


Hi,

Please review a small change to correct the testcase runtime/ErrorHandling/VeryEarlyAssertTest.java
JBS - https://bugs.openjdk.java.net/browse/JDK-8239557 
Webrev - http://cr.openjdk.java.net/~fmatte/8239557/webrev.00/ 

Background: 
Testcase assume that "END." Marker is always at the last line of the hs_error file.
Crash mechanism is triggered by static initialization of libjvm.so. This happens before any of the actual VM initialization.
During error reporting it may end up with errors, and additional logs may get appended after END marker.
This issue is only observed on internal testing (Mach5).

Thanks,
Fairoz


More information about the hotspot-runtime-dev mailing list