RFR: 8273487: Zero: Handle "zero" variant in runtime tests

Aleksey Shipilev shade at openjdk.java.net
Wed Sep 8 13:25:27 UTC 2021

On Wed, 8 Sep 2021 10:57:33 GMT, Aleksey Shipilev <shade at openjdk.org> wrote:

> JDK-8179317 rewritten runtime shell tests to Java. There is a little omission in VM variant selection, which makes Zero fail some of the tier1 tests, like:
> $ CONF=linux-x86_64-zero-fastdebug make exploded-test TEST=runtime/StackGap/TestStackGap.java
> java.lang.Error: TESTBUG: unsupported vm variant
> at jdk.test.lib.Platform.variant(Platform.java:368)
> at jdk.test.lib.Platform.jvmLibDir(Platform.java:357)
> at jdk.test.lib.process.ProcessTools.addJvmLib(ProcessTools.java:585)
> at jdk.test.lib.process.ProcessTools.createNativeTestProcessBuilder(ProcessTools.java:575) 
> Additional testing:
>  - [x] Linux x86_64 Zero affected tests (StackGap, StackGuardPages, TestTLS) now run

Huh, it turns out there is the awkwardness in the build system that puts product `libjvm.so` to `server` for Zero. See [JDK-8273494](https://bugs.openjdk.java.net/browse/JDK-8273494). This was not the cause for gtest `libjvm.so`, which is in `zero` folder for Zero. So I had to make it more awkward until the build system fix is here. See new commit.


PR: https://git.openjdk.java.net/jdk/pull/5413

More information about the core-libs-dev mailing list