RFR: 8326521: JFR: CompilerPhase event test fails on windows 32 bit
Erik Gahlin
egahlin at openjdk.org
Wed Mar 20 10:12:21 UTC 2024
On Tue, 19 Mar 2024 13:50:52 GMT, Robert Toyonaga <duke at openjdk.org> wrote:
> This test failure is a problem for the Adoptium CI. The reason that this test fails on 32 bit Windows is that Hotspot only uses the C1 compiler in this configuration by design. If the system is Windows and not 64 bit, [NeverActAsServerClassMachine](https://github.com/openjdk/jdk/blob/master/src/hotspot/share/compiler/compilerDefinitions.cpp#L569) will be set. This results in setting the compilation mode to be [quick_only](https://github.com/openjdk/jdk/blob/master/src/hotspot/share/compiler/compilerDefinitions.cpp#L191), which results in [constraining to C1 compilation](https://github.com/openjdk/jdk/blob/master/src/hotspot/share/compiler/compilerDefinitions.inline.hpp#L53).
>
> The CompilerPhase JFR events are only emitted from C2 code in hotspot. So although the test succeeds in compiling the method it intends to (with C1), it isn't able to generate the JFR events it expects, and so fails.
>
> ----------System.out:(4/182)----------
> CompileCommand: compileonly jdk/jfr/event/compiler/TestCompilerPhase.dummyMethod bool compileonly = true
> 1 compiler directives added
> WB error: invalid compilation level 4
> ----------System.err:(16/1050)----------
> java.lang.RuntimeException: No events: expected false, was true
> at jdk.test.lib.Asserts.fail(Asserts.java:594)
> at jdk.test.lib.Asserts.assertFalse(Asserts.java:461)
> at jdk.test.lib.jfr.Events.hasEvents(Events.java:161)
> at jdk.jfr.event.compiler.TestCompilerPhase.main(TestCompilerPhase.java:76)
> at java.base/jdk.internal.reflect.NativeMethodAccessorImpl.invoke0(Native Method)
> at java.base/jdk.internal.reflect.NativeMethodAccessorImpl.invoke(NativeMethodAccessorImpl.java:77)
> at java.base/jdk.internal.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43)
> at java.base/java.lang.reflect.Method.invoke(Method.java:568)
> at com.sun.javatest.regtest.agent.MainWrapper$MainTask.run(MainWrapper.java:138)
> at java.base/java.lang.Thread.run(Thread.java:840)
>
> JavaTest Message: Test threw exception: java.lang.RuntimeException: No events: expected false, was true
> JavaTest Message: shutting down test
>
> STATUS:Failed.`main' threw exception: java.lang.RuntimeException: No events: expected false, was true
>
> This PR prevents `NeverActAsServerClassMachine` from being set during the test, so that it isn't restricted to C1.
Looks reasonable.
-------------
Marked as reviewed by egahlin (Reviewer).
PR Review: https://git.openjdk.org/jdk/pull/18379#pullrequestreview-1948441648
More information about the hotspot-jfr-dev
mailing list