You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@harmony.apache.org by Xiao-Feng Li <xi...@gmail.com> on 2007/04/25 01:15:19 UTC
Re: [jira] Created: (HARMONY-3746) [drlvm][gc_gen] DRLVM failed on SPECjbb2005 with GCv5.
Judging based on the stack trace and what you reported, I guess it is
not a GC bug. The wrapper_func is only used for Java thread. I will
check it anyway.
Thanks,
xiaofeng
On 4/24/07, Sergey Kuksenko (JIRA) <ji...@apache.org> wrote:
> [drlvm][gc_gen] DRLVM failed on SPECjbb2005 with GCv5.
> ------------------------------------------------------
>
> Key: HARMONY-3746
> URL: https://issues.apache.org/jira/browse/HARMONY-3746
> Project: Harmony
> Issue Type: Bug
> Reporter: Sergey Kuksenko
> Priority: Blocker
>
>
> DRLVM failed on SPECjbb2005 with GCv5 in server mode.
> When we tunrning back gc_cc everything is ok.
>
> Used DRLVM parametters: -Xms1536m -Xmx1536m -Xem:server5
> server5 - is a copy of the default server.emconf modified for gc_gen.
> Platform: Win32.
> Failire occures on 4 warehouses.
> Call stack is
> harmonyvm.dll!run_java_detach(_jobject * java_thread=0x00070023) Line 167 + 0xd C++
> harmonyvm.dll!vm_detach(_jobject * java_thread=0x66ac1a68) Line 280 C++
> harmonyvm.dll!jthread_detach(_jobject * java_thread=0x66ac1a68) Line 293 + 0x6 C
> harmonyvm.dll!wrapper_proc(void * arg=0x66abfb10) Line 105 + 0x6 C
> hythr.dll!thread_start_proc(void * arg=0x008281b0) Line 714 C
> hythr.dll!_threadstartex(void * ptd=0x003ad088) Line 241 + 0x6 C
>
> Also I wish to mention that on svn rev:531370 (12 April) SPECjbb2005 passed on DRLVM + GCv5.
>
>
>
> --
> This message is automatically generated by JIRA.
> -
> You can reply to this email to add a comment to the issue online.
>
>
--
http://xiao-feng.blogspot.com
Re: [jira] Created: (HARMONY-3746) [drlvm][gc_gen] DRLVM failed on SPECjbb2005 with GCv5.
Posted by Xiao-Feng Li <xi...@gmail.com>.
I also had the same curiosity. GC is checking if it is a root set
update problem. Probably some JIT guy could also check if there is any
issue with the root set enumeration in server mode. It is actually not
reproducible locally in single processor and dual-core processor. It
would be great if some people can reproduce and give us more
information. Thanks.
Sergey, would you tell us the environment (esp. processor number) you
are using that hit this bug?
Thanks,
xiaofeng
On 4/25/07, Evgueni Brevnov <ev...@gmail.com> wrote:
> I wonder how local pointer (_jobject * java_thread) could change? Or
> it is just corrupted stack?
>
> Evgueni
>
> On 4/25/07, Xiao-Feng Li <xi...@gmail.com> wrote:
> > Judging based on the stack trace and what you reported, I guess it is
> > not a GC bug. The wrapper_func is only used for Java thread. I will
> > check it anyway.
> >
> > Thanks,
> > xiaofeng
> >
> > On 4/24/07, Sergey Kuksenko (JIRA) <ji...@apache.org> wrote:
> > > [drlvm][gc_gen] DRLVM failed on SPECjbb2005 with GCv5.
> > > ------------------------------------------------------
> > >
> > > Key: HARMONY-3746
> > > URL: https://issues.apache.org/jira/browse/HARMONY-3746
> > > Project: Harmony
> > > Issue Type: Bug
> > > Reporter: Sergey Kuksenko
> > > Priority: Blocker
> > >
> > >
> > > DRLVM failed on SPECjbb2005 with GCv5 in server mode.
> > > When we tunrning back gc_cc everything is ok.
> > >
> > > Used DRLVM parametters: -Xms1536m -Xmx1536m -Xem:server5
> > > server5 - is a copy of the default server.emconf modified for gc_gen.
> > > Platform: Win32.
> > > Failire occures on 4 warehouses.
> > > Call stack is
> > > harmonyvm.dll!run_java_detach(_jobject * java_thread=0x00070023) Line 167 + 0xd C++
> > > harmonyvm.dll!vm_detach(_jobject * java_thread=0x66ac1a68) Line 280 C++
> > > harmonyvm.dll!jthread_detach(_jobject * java_thread=0x66ac1a68) Line 293 + 0x6 C
> > > harmonyvm.dll!wrapper_proc(void * arg=0x66abfb10) Line 105 + 0x6 C
> > > hythr.dll!thread_start_proc(void * arg=0x008281b0) Line 714 C
> > > hythr.dll!_threadstartex(void * ptd=0x003ad088) Line 241 + 0x6 C
> > >
> > > Also I wish to mention that on svn rev:531370 (12 April) SPECjbb2005 passed on DRLVM + GCv5.
> > >
> > >
> > >
> > > --
> > > This message is automatically generated by JIRA.
> > > -
> > > You can reply to this email to add a comment to the issue online.
> > >
> > >
> >
> >
> > --
> > http://xiao-feng.blogspot.com
> >
>
--
http://xiao-feng.blogspot.com
Re: [jira] Created: (HARMONY-3746) [drlvm][gc_gen] DRLVM failed on SPECjbb2005 with GCv5.
Posted by Evgueni Brevnov <ev...@gmail.com>.
I wonder how local pointer (_jobject * java_thread) could change? Or
it is just corrupted stack?
Evgueni
On 4/25/07, Xiao-Feng Li <xi...@gmail.com> wrote:
> Judging based on the stack trace and what you reported, I guess it is
> not a GC bug. The wrapper_func is only used for Java thread. I will
> check it anyway.
>
> Thanks,
> xiaofeng
>
> On 4/24/07, Sergey Kuksenko (JIRA) <ji...@apache.org> wrote:
> > [drlvm][gc_gen] DRLVM failed on SPECjbb2005 with GCv5.
> > ------------------------------------------------------
> >
> > Key: HARMONY-3746
> > URL: https://issues.apache.org/jira/browse/HARMONY-3746
> > Project: Harmony
> > Issue Type: Bug
> > Reporter: Sergey Kuksenko
> > Priority: Blocker
> >
> >
> > DRLVM failed on SPECjbb2005 with GCv5 in server mode.
> > When we tunrning back gc_cc everything is ok.
> >
> > Used DRLVM parametters: -Xms1536m -Xmx1536m -Xem:server5
> > server5 - is a copy of the default server.emconf modified for gc_gen.
> > Platform: Win32.
> > Failire occures on 4 warehouses.
> > Call stack is
> > harmonyvm.dll!run_java_detach(_jobject * java_thread=0x00070023) Line 167 + 0xd C++
> > harmonyvm.dll!vm_detach(_jobject * java_thread=0x66ac1a68) Line 280 C++
> > harmonyvm.dll!jthread_detach(_jobject * java_thread=0x66ac1a68) Line 293 + 0x6 C
> > harmonyvm.dll!wrapper_proc(void * arg=0x66abfb10) Line 105 + 0x6 C
> > hythr.dll!thread_start_proc(void * arg=0x008281b0) Line 714 C
> > hythr.dll!_threadstartex(void * ptd=0x003ad088) Line 241 + 0x6 C
> >
> > Also I wish to mention that on svn rev:531370 (12 April) SPECjbb2005 passed on DRLVM + GCv5.
> >
> >
> >
> > --
> > This message is automatically generated by JIRA.
> > -
> > You can reply to this email to add a comment to the issue online.
> >
> >
>
>
> --
> http://xiao-feng.blogspot.com
>