JDK-8258101 : compiler/blackhole tests fail with -Xcomp
  • Type: Bug
  • Component: hotspot
  • Sub-Component: compiler
  • Affected Version: 16,17
  • Priority: P3
  • Status: Closed
  • Resolution: Duplicate
  • Submitted: 2020-12-11
  • Updated: 2021-01-06
  • Resolved: 2021-01-06
The Version table provides details related to the release that this issue/RFE will be addressed.

Unresolved : Release in which this issue/RFE will be addressed.
Resolved: Release in which this issue/RFE has been resolved.
Fixed : Release in which this issue/RFE has been fixed. The release containing this fix may be available for download as an Early Access Release or a General Availability Release.

To download the current JDK release, click here.
JDK 16
16Resolved
Related Reports
Duplicate :  
JDK-8258558 - Revert changes for JDK-8252505 and related issues
Relates :  
JDK-8257803 - Add -Xbatch to compiler/blackhole tests
Relates :  
JDK-8252505 - C1/C2 compiler support for blackholes
Sub Tasks
JDK-8258111 :  
Problemlist compiler/blackhole tests for -Xcomp until JDK-8258101 is fixed - Resolved
Description
Even after JDK-8257803, the compiler/blackhole tests still fail:

----------System.out:(1/88)----------
CompileCommand: blackhole compiler/blackhole/BlackholeTarget.bh_* bool blackhole = true
----------System.err:(15/1076)----------
java.lang.IllegalStateException: Should not have been entered: bh_s_byte_0
	at compiler.blackhole.BlackholeTarget.shouldNotBeEntered(BlackholeTarget.java:51)
	at compiler.blackhole.BlackholeStaticTest.runTries(BlackholeStaticTest.java:118)
	at compiler.blackhole.BlackholeStaticTest.main(BlackholeStaticTest.java:80)
	at java.base/jdk.internal.reflect.NativeMethodAccessorImpl.invoke0(Native Method)
	at java.base/jdk.internal.reflect.NativeMethodAccessorImpl.invoke(NativeMethodAccessorImpl.java:78)
	at java.base/jdk.internal.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43)
	at java.base/java.lang.reflect.Method.invoke(Method.java:567)
	at com.sun.javatest.regtest.agent.MainWrapper$MainThread.run(MainWrapper.java:127)
	at java.base/java.lang.Thread.run(Thread.java:831)

-Dtest.java.opts='-Xcomp -XX:+CreateCoredumpOnCrash -ea -esa -XX:CompileThreshold=100 -XX:+UnlockExperimentalVMOptions -server -XX:-TieredCompilation' \\
Comments
JDK-8258558 reverts blackhole support, so once that lands, this would not longer be an issue.
05-01-2021
Right, I've set this to P2 because we run with -Xcomp at several tiers in our CI's. Lowering priority now that these tests are problem listed.
14-12-2020
The impact of the issue on our CI testing makes these a high priority, at least until tests have been ProblemListed. Thanks for ProblemListing them.
14-12-2020
JDK-8258111 problem-lists these for Xcomp.
11-12-2020
The reason why Xcomp is failing eludes me so far. In -Xcomp mode, bh_* methods get inlined, even though compiler oracle instructs to blackhole it. It might be due to ciMethod thinking the method is unloaded. -Xcomp: 2487 1041 b 1 compiler.blackhole.BlackholeStaticTest$$Lambda$1/0x0000000801000a18::run (4 bytes) @ 0 compiler.blackhole.BlackholeStaticTest::test_boolean_0 (18 bytes) inline @ 8 compiler.blackhole.BlackholeTarget::bh_s_boolean_0 (6 bytes) inline @ 2 compiler.blackhole.BlackholeTarget::registerEntered (62 bytes) callee is too large default: 200 29 % b 1 compiler.blackhole.BlackholeStaticTest::test_boolean_0 @ 2 (18 bytes) @ 8 compiler.blackhole.BlackholeTarget::bh_s_boolean_0 (6 bytes) intrinsic I think I'll just quarantine these for -Xcomp before I have a chance to look at it more thoroughly.
11-12-2020
I am on vacation right now, and would be able to look at these failures in a few weeks. Maybe sooner. Meanwhile, are isolated test bugs like these, run in non-default configurations like -Xcomp, really P2?
11-12-2020
Looks like these tests mustn't run under -Xcomp.
11-12-2020
ILW = Test fails (test bug), compiler/blackhole tests at higher tiers, no workaround = MHH = P2
11-12-2020