[jdk11u-dev] RFR: 8270306: [TESTBUG] Add smoke test for JVMCI bootstrapping

[Severin Gehwolf]

On Mon, 12 Jul 2021 10:35:36 GMT, Michal Karm Babacek <github.com+691097+Karm at openjdk.org> wrote:

>> Test fails with the undermentioned when fix for 8226533 is reverted (ed48e5cb).
>> 
>> 
>> ----------System.err:(67/7089)----------
>>  stdout: [Bootstrapping JVMCI.Thread[JVMCI CompilerThread0,9,system]: Compilation of org.graalvm.compiler.phases.common.util.HashSetNodeEventListener.changed(Graph$NodeEvent, Node) failed: org.graalvm.compiler.graph.GraalGraphError: java.lang.InternalError: Effectively static method org.graalvm.compiler.graph.Node.int org.graalvm.compiler.graph.Node.hashCode() should be handled in Java code
>>     at node: 113|LoadMethod
>>     at exact jdk.internal.vm.compiler.collections.EconomicMapImpl.getHashIndex(Object):int
>>     ...
>> 
>> 
>> The test passes with the fix in place.
>> 
>> The test requires JVMCI implementation compiler to be present,
>> i.e. it makes sense only in versions before https://openjdk.java.net/jeps/410.
>
> test/hotspot/jtreg/compiler/jvmci/TestBootstrapJVMCIOption.java line 28:
> 
>> 26:  * @bug 8226533
>> 27:  * @summary Ensure that -XX:+BootstrapJVMCI does not trigger an assertion
>> 28:  * @requires vm.jvmci
> 
> @jerboaa Hello, does this `@requires` make sense? Should I put something like `vm.graal.enabled` in there too?

Yes, `vm.graal.enabled` would make sense for this test.

-------------

PR: https://git.openjdk.java.net/jdk11u-dev/pull/122