RFR: 8326521: JFR: CompilerPhase event test fails on windows 32 bit
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/compil...) 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/compil...), which results in [constraining to C1 compilation](https://github.com/openjdk/jdk/blob/master/src/hotspot/share/compiler/compil...). 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. ------------- Commit messages: - Prevent NeverActAsServerClassMachine since we require C2 Changes: https://git.openjdk.org/jdk/pull/18379/files Webrev: https://webrevs.openjdk.org/?repo=jdk&pr=18379&range=00 Issue: https://bugs.openjdk.org/browse/JDK-8326521 Stats: 1 line in 1 file changed: 1 ins; 0 del; 0 mod Patch: https://git.openjdk.org/jdk/pull/18379.diff Fetch: git fetch https://git.openjdk.org/jdk.git pull/18379/head:pull/18379 PR: https://git.openjdk.org/jdk/pull/18379
On Tue, 19 Mar 2024 13:50:52 GMT, Robert Toyonaga <duke@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/compil...) 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/compil...), which results in [constraining to C1 compilation](https://github.com/openjdk/jdk/blob/master/src/hotspot/share/compiler/compil...).
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
On Tue, 19 Mar 2024 13:50:52 GMT, Robert Toyonaga <duke@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/compil...) 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/compil...), which results in [constraining to C1 compilation](https://github.com/openjdk/jdk/blob/master/src/hotspot/share/compiler/compil...).
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.
This pull request has now been integrated. Changeset: 96530bcc Author: Robert Toyonaga <rtoyonag@redhat.com> Committer: Erik Gahlin <egahlin@openjdk.org> URL: https://git.openjdk.org/jdk/commit/96530bcc07514c3eda40fd6ffa74f197fe541dea Stats: 1 line in 1 file changed: 1 ins; 0 del; 0 mod 8326521: JFR: CompilerPhase event test fails on windows 32 bit Reviewed-by: egahlin ------------- PR: https://git.openjdk.org/jdk/pull/18379
participants (2)
-
Erik Gahlin
-
Robert Toyonaga