harmony-commits mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Ilya Berezhniuk (JIRA)" <j...@apache.org>
Subject [jira] Commented: (HARMONY-4388) [drlvm][jvmti] DRLVM crashes in jvmti_start_single_step_in_virtual_method(), file jvmti_step.cpp, line 504: assertion "vtable"
Date Wed, 12 Sep 2007 12:52:32 GMT

    [ https://issues.apache.org/jira/browse/HARMONY-4388?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#action_12526775
] 

Ilya Berezhniuk commented on HARMONY-4388:
------------------------------------------

I've checked this test with debug info in class file - the test passes in most cases.
Sometimes the test fires an assertion in vm/jitrino/src/jet/rt.cpp:273: "method_get_num_handlers(method)
== 0" on debug VM - I'm going to file a bug for this. But reported problem is not reproducible
when EventDebuggee014.java is compiled with debug info.

Looks like the only change required to fix this bug is adding debug information to class files.
I'm going to provide patch soon.

> [drlvm][jvmti] DRLVM crashes in jvmti_start_single_step_in_virtual_method(), file jvmti_step.cpp,
line 504: assertion "vtable"
> ------------------------------------------------------------------------------------------------------------------------------
>
>                 Key: HARMONY-4388
>                 URL: https://issues.apache.org/jira/browse/HARMONY-4388
>             Project: Harmony
>          Issue Type: Bug
>          Components: DRLVM
>         Environment: Windows/x86, harmony-jdk-r553376
>            Reporter: Ivan Popov
>
> After applying patch HARMONY-2889 for JDWP agent the following JDWP stress test:
>     org.apache.harmony.test.stress.jpda.jdwp.scenario.EVENT014.EventTest014
> crashes DRLVM in function jvmti_start_single_step_in_virtual_method(), file jvmti_step.cpp,
line 504: assertion "vtable".
> Stack trace is the following:
> >	harmonyvm.dll!_assert(const char * expr=0x007fe9cc, const char * filename=0x007fe970,
unsigned int lineno=504)  Line 295	C
>  	harmonyvm.dll!jvmti_start_single_step_in_virtual_method(DebugUtilsTI * ti=0x016150a0,
VMBreakPoint * bp=0x03e9fa94, void * data=0x038d5950)  Line 504 + 0x1a	C++
>  	harmonyvm.dll!jvmti_process_jit_single_step_event(TIEnv * unused_env=0x00000000, VMBreakPoint
* bp=0x03e9fa94, void * data=0x038d5950)  Line 589 + 0x11	C++
>  	harmonyvm.dll!VMBreakPoints::process_native_breakpoint()  Line 673 + 0x25	C++
>  	harmonyvm.dll!process_native_breakpoint_event()  Line 1329	C++
>  	harmonyvm.dll!asm_process_native_breakpoint_event()  Line 1341	C++
>  	12737a4c()	
>  	harmonyvm.dll!vm_invoke_native_array_stub(unsigned int * args=0x03e9fd14, int sz=1,
void * f=0x0188c5d0)  Line 77	C++
>  	harmonyvm.dll!JIT_execute_method_default(void * jit=0x00000000, _jmethodID * methodID=0x02f9a148,
jvalue * return_value=0x00000000, jvalue * args=0x03e9fef4)  Line 200	C++
>  	em.dll!DrlEMImpl::executeMethod(_jmethodID * meth=0x02f9a148, jvalue * return_value=0x00000000,
jvalue * args=0x03e9fef4)  Line 509 + 0x16	C++
>  	em.dll!ExecuteMethod(_jmethodID * meth=0x02f9a148, jvalue * return_value=0x00000000,
jvalue * args=0x03e9fef4)  Line 43 + 0x21	C++
>  	harmonyvm.dll!vm_execute_java_method_array(_jmethodID * method=0x02f9a148, jvalue *
result=0x00000000, jvalue * args=0x03e9fef4)  Line 60 + 0x1b	C++
>  	harmonyvm.dll!Objects_To_Finalize::do_finalization(int quantity=0)  Line 519 + 0xf
C++
>  	harmonyvm.dll!vm_do_finalization(int quantity=0)  Line 631	C++
>  	harmonyvm.dll!finalizer_thread_func(void * * args=0x035f02a0)  Line 241 + 0x7	C++
>  	hythr.dll!thread_start_proc(void * arg=0x00427320)  Line 729 + 0x9	C
>  	hythr.dll!_threadstartex(void * ptd=0x00427418)  Line 241 + 0xd	C
>  	kernel32.dll!77e66063() 	
> This crash happens only in JIT mode. If debugg VM is started in -Xint mode (using option
"-Djpda.settings.debuggeeVMExtraOptions=-Xint"), test successfully passes. I believe this
is bug in implementation of JVMTI single step support in JIT mode.

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.


Mime
View raw message