Bug 860698
| Summary: | java.lang.VerifyError: JVMVRFY012 stack shape inconsistent | ||
|---|---|---|---|
| Product: | [JBoss] JBoss Enterprise BRMS Platform 5 | Reporter: | Geoffrey De Smet <gdesmet> |
| Component: | BRE (Expert, Fusion) | Assignee: | Mario Fusco <mfusco> |
| Status: | VERIFIED --- | QA Contact: | |
| Severity: | unspecified | Docs Contact: | |
| Priority: | unspecified | ||
| Version: | BRMS 5.3.0.GA | CC: | ajuricic, alazarot, cbasquez, kverlaen, mfusco, rwagner |
| Target Milestone: | ER4 | ||
| Target Release: | One Off Releases | ||
| Hardware: | Unspecified | ||
| OS: | Unspecified | ||
| Whiteboard: | |||
| Fixed In Version: | Doc Type: | Bug Fix | |
| Doc Text: |
When a declaration double dereferences within the Business Rules Engine, and the value of the method returned by the second dereferencing is a primitive type, the bytecode of the eval class using it is not correctly generated. A java.lang.VerifyError occurs when trying to instance a class from the incorrectly generated bytecode. This has been resolved by fixing the bytecode generator to allow it to work even in these specific primitive dereferencing instances. It is now possible to have an eval class do a double dereferencing, which returns a primitive type without any malformed bytecodes.
|
Story Points: | --- |
| Clone Of: | Environment: | ||
| Last Closed: | Type: | Bug | |
| Regression: | --- | Mount Type: | --- |
| Documentation: | --- | CRM: | |
| Verified Versions: | Category: | --- | |
| oVirt Team: | --- | RHEL 7.3 requirements from Atomic Host: | |
| Cloudforms Team: | --- | Target Upstream Version: | |
| Embargoed: | |||
| Bug Depends On: | |||
| Bug Blocks: | 873846 | ||
This bug is manifested in a mixed community / product environment. Please see the associated support case for customer interactions. Thanks, Rick Can we have a reproducer, please? Also, from the stack trace it would appear that the bug occurs with IBM JVM. Which version? Which platform? Hi Lukas, Fair requests. We've requested a reproducer, we'll ask for the JVM information as well. Thanks, Rick Talked to the customer, they are running Oracle JDK 1.6 on Windows. Thanks, Rick I have been asked to clarify this bug report: It's a bug in Drools Expert (found by a customer using Planner, but that's irrelevant). It's a regression from drools community 5.3.0.Final and this bug prevented the customer from migrating from community to BRMS 5.3.0.Final. BRMS 5.3.1.Final likely fixes it already, thanks to Mario's fixes. But without a reproduce recipe or more information (which I do not have), it's impossible to tell. Geoffrey, Rick, Can we ask the customer to test with the current snapshot of BRMS.5.3.1? Until we know if the problem is fixed or not, there isn't much we can do on this ticket. If it is not fixed, then we need a way to reproduce and/or more information. The above stack trace does not tell us much other than the problem is happening on an "eval" invocation. Morteza confirmed the bug is still present in BRMS 5.3.1-ER3. "He used the 5.1.0.BRMS.ER3 core/compiler/knowledge-api and mvel2-2.1.3.Final". Getting a similar exception when using Oracle JDK 1.6.0.31:
Exception in thread "main" java.lang.VerifyError: (class: gov/ssa/asa/rules/Rule_vhrGroupingByLocationEval0Invoker, method: evaluate signature: (Lorg/drools/spi/Tuple;[Lorg/drools/rule/Declaration;Lorg/drools/WorkingMemory;Ljava/lang/Object;)Z) Expecting to find integer on stack
at java.lang.Class.getDeclaredConstructors0(Native Method)
at java.lang.Class.privateGetDeclaredConstructors(Class.java:2389)
at java.lang.Class.getConstructor0(Class.java:2699)
at java.lang.Class.newInstance0(Class.java:326)
at java.lang.Class.newInstance(Class.java:308)
at org.drools.rule.JavaDialectRuntimeData.wire(JavaDialectRuntimeData.java:413)
at org.drools.rule.JavaDialectRuntimeData.wire(JavaDialectRuntimeData.java:400)
at org.drools.rule.JavaDialectRuntimeData.onBeforeExecute(JavaDialectRuntimeData.java:241)
at org.drools.rule.DialectRuntimeRegistry.onBeforeExecute(DialectRuntimeRegistry.java:134)
at org.drools.compiler.PackageBuilder.reloadAll(PackageBuilder.java:1025)
at org.drools.compiler.PackageBuilder.addPackage(PackageBuilder.java:976)
at org.drools.compiler.PackageBuilder.addPackageFromDrl(PackageBuilder.java:405)
at org.drools.compiler.PackageBuilder.addPackageFromDrl(PackageBuilder.java:380)
at org.drools.planner.config.score.director.ScoreDirectorFactoryConfig.buildRuleBase(ScoreDirectorFactoryConfig.java:245)
at org.drools.planner.config.score.director.ScoreDirectorFactoryConfig.buildDroolsScoreDirectorFactory(ScoreDirectorFactoryConfig.java:225)
at org.drools.planner.config.score.director.ScoreDirectorFactoryConfig.buildScoreDirectorFactory(ScoreDirectorFactoryConfig.java:157)
at org.drools.planner.config.solver.SolverConfig.buildSolver(SolverConfig.java:142)
at gov.ssa.asa.service.impl.EngineServiceImpl.configurePlanner(EngineServiceImpl.java:153)
at gov.ssa.asa.service.impl.EngineServiceImpl.runEngine(EngineServiceImpl.java:74)
at gov.ssa.asa.main.EngineMain.main(EngineMain.java:44)
This seems to be a JIT problem when compiling the invoker, but that is all we can infer from the stack trace. We will investigate, but without being able to reproduce the problem, it will be hard to fix this one. Mario Fusco <mario.fusco> updated the status of jira JBRULES-3657 to Resolved Hi Mario, Can you provided release notes content for this issue to be included in the BRMS 5.3.1 release notes. Thanks Lee Thanks for the updated information. Editing the doc text for Release Notes purposes. Thanks again, guys. - Doug Verified on 5.3.1.BRMS-ER4 Has it been fixed on brms 5.3.1. GA deployable ? I have one customer getting the same error described in comment #11 on brms 5.3.1. Mario Fusco <mario.fusco> made a comment on jira JBRULES-3657 This seems to be a duplicate of this bug: https://issues.jboss.org/browse/JBRULES-3283 If so I backported the fix to the 5.3.x branch only 2 days ago. Included in roll-up #1: BZ-873846 Verified on 5.3.1.BRMS roll-up #1 |
A customer gets this error with Drools from BRMS 5.3.0.GA, but not with drools community 5.3.0.Final (so it's regression): Exception in thread "main" java.lang.VerifyError: JVMVRFY012 stack shape inconsistent; class=.../rules/Rule_myrule01Eval0Invoker, method=evaluate(Lorg/drools/spi/Tuple;[Lorg/drools/rule/Declaration;Lorg/drools/WorkingMemory;Ljava/lang/Object;)Z, pc=122 at java.lang.J9VMInternals.verifyImpl(Native Method) at java.lang.J9VMInternals.verify(J9VMInternals.java:72) at java.lang.J9VMInternals.initialize(J9VMInternals.java:134) at java.lang.J9VMInternals.newInstanceImpl(Native Method) at java.lang.Class.newInstance(Class.java:1345) at org.drools.rule.JavaDialectRuntimeData.wire(JavaDialectRuntimeData.java:413) On master this problem is fixed, so it's likely, but not certain that BRMS 5.3.1 will fix it. Currently I have no reproduce recipe. Contact is Prabakar Singaram (RH consultant).