harmony-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Xiao-Feng Li" <xiaofeng...@gmail.com>
Subject Re: [drlvm][classloader] classloader.StressLoader hangs on my 2-cpu Linux box. Does anyone else see this problem?
Date Thu, 26 Apr 2007 11:24:51 GMT
On 4/25/07, Naveen Neelakantam <neelakan@uiuc.edu> wrote:
> Has there been progress on this issue?
>
> My CC setup is hung running classloader.StressLoader in interpreter
> (on a 2-cpu RHEL4 box).  It actually has been hung for weeks now, but
> I only just noticed it.

Then you were using GCv4.1 when hit this problem? There was a switch
to GCv5 in last weekend. With GCv5, the testing never met StressLoader
hangs so far. Once it was suspected to be hanging in interpreter mode,
but finally it proved to be just too long time to finish. (We can
observe the processor is busy during the long time running.)

Thanks. -xiaofeng

> Naveen
>
> On Mar 28, 2007, at 7:37 AM, Elena Semukhina wrote:
>
> > On 3/27/07, Weldon Washburn <weldonwjw@gmail.com> wrote:
> >>
> >> Good work.
> >>
> >> The comments on 3495 says this is intermittant hang.  Does it
> >> happend 10%
> >> of
> >> the time?  80%??  How many CPUs does your system have?  Can you
> >> try it on
> >> a
> >> 4-way?
> >
> >
> > The test hangs rarely. Yesterday it hung after ~400 iterations,
> > today it has
> > run 850 times and is still running on linux. On 4-way windows it
> > has hung on
> > 799th iteration.
> >
> > Elena
> >
> > On 3/27/07, Elena Semukhina <elena.semukhina@gmail.com> wrote:
> >> >
> >> > Weldon,
> >> >
> >> > today I saw the hang of classloader.StressLoader but only on linux
> >> x86_64
> >> > (OPT) when running it in the loop. The hang is very similar to
> >> what I
> >> see
> >> > for gc.NPE. I reported
> >> https://issues.apache.org/jira/browse/HARMONY-3495to
> >> > track the issue.
> >> >
> >> > Elena
> >> >
> >> >
> >> > On 3/26/07, Weldon Washburn <weldonwjw@gmail.com> wrote:
> >> > >
> >> > > Elena,
> >> > > Thanks!  It looks like I will keep StressLoader.java on my local
> >> > excludes
> >> > > list for now.
> >> > >
> >> > >
> >> > >
> >> > > On 3/26/07, Elena Semukhina <elena.semukhina@gmail.com> wrote:
> >> > > >
> >> > > > Weldon,
> >> > > >
> >> > > > today I ran the test iteratively on my SUSE9 linux x86 2 CPU
> >> machine
> >> > in
> >> > > > default mode and it successfully completed about 2000 times.
> >> The
> >> same
> >> > > > result
> >> > > > is for linux x86_64.
> >> > > >
> >> > > > Thanks,
> >> > > > Elena
> >> > > >
> >> > > > On 3/24/07, Weldon Washburn <weldonwjw@gmail.com> wrote:
> >> > > > >
> >> > > > > I would submit a JIRA if I had a specific bug to report.
 The
> >> > > situation
> >> > > > is
> >> > > > > that the StressLoader test hangs in file LogLoader.java
at
> >> the
> >> > > following
> >> > > > > line:
> >> > > > >
> >> > > > > InputStream is = ClassLoader.getSystemResourceAsStream(path);
> >> > > > >
> >> > > > > It seems to be aggrevated by all the threads that execute
the
> >> above
> >> > > > code.
> >> > > > >
> >> > > > > For now, I plan to put this test on my local Linux
> >> excludes list
> >> and
> >> > > go
> >> > > > > back
> >> > > > > to committing patches.
> >> > > > >
> >> > > > > --
> >> > > > > Weldon Washburn
> >> > > > > Intel Enterprise Solutions Software Division
> >> > > > >
> >> > > >
> >> > >
> >> > >
> >> > >
> >> > > --
> >> > > Weldon Washburn
> >> > > Intel Enterprise Solutions Software Division
> >> > >
> >> >
> >>
> >>
> >>
> >> --
> >> Weldon Washburn
> >> Intel Enterprise Solutions Software Division
> >>
>
>


-- 
http://xiao-feng.blogspot.com

Mime
View raw message