dtrace discuss - Jul 2009 - "args may not be referenced.... unstable.." error with sched:::change-pri

One of our Java RTS users has encountered a problem running the DTrace scripts
that we supply:

http://forums.sun.com/thread.jspa?threadID=5399648

The error pertains to the sched provider''s change-pri probe:

?... in action list: args[ ] may not be referenced because probe description
sched:::change-pri matches an unstable set of probes?

I''ve done some research and understand what the error means, but I
don''t understand why it is being seen in this case, nor why we have not
seen it ourselves at any time. Are there any recent DTrace patches that might
have affected this?

Thanks,
David Holmes
Java RTS Engineering Group
-- 
This message posted from opensolaris.org

57 views but no replies ... has no-one any idea what might be going wrong here?

David Holmes
-- 
This message posted from opensolaris.org

I''m unable to reproduce your problem. What bits are you running? Can I
have
access to that system?

Adam

On Jul 27, 2009, at 10:56 PM, David Holmes wrote:
> One of our Java RTS users has encountered a problem running the  
> DTrace scripts that we supply:
>
> http://forums.sun.com/thread.jspa?threadID=5399648
>
> The error pertains to the sched provider''s change-pri probe:
>
> ?... in action list: args[ ] may not be referenced because probe  
> description sched:::change-pri matches an unstable set of probes?
>
> I''ve done some research and understand what the error means, but I
> don''t understand why it is being seen in this case, nor why we
have
> not seen it ourselves at any time. Are there any recent DTrace  
> patches that might have affected this?
>
> Thanks,
> David Holmes
> Java RTS Engineering Group
> -- 
> This message posted from opensolaris.org
> _______________________________________________
> dtrace-discuss mailing list
> dtrace-discuss at opensolaris.org

--
Adam Leventhal, Fishworks                        http://blogs.sun.com/ahl

> I''m unable to reproduce your problem. What bits are
> you running? Can I  have access to that system?
> 
> Adam
Adam,

I can''t reproduce it either, only the user on the forum has seen the
problem - on Solaris 10 as I understand it. Hence my query as to what may make
this behave differently on different systems.

Thanks,
David Holmes
-- 
This message posted from opensolaris.org

> > I''m unable to reproduce your problem. What bits are
> > you running? Can I  have access to that system?
> 
> I can''t reproduce it either, only the user on the forum has seen
the problem -
> on Solaris 10 as I understand it. Hence my query as to what may make this
> behave differently on different systems.
It''s possible that there was some bug in a Solaris 10 update, but
there''s
nothing that should make this happen. Please find out exactly what version
the user is running along with any patches. Further, it would be useful to
see the output of ''dtrace -lv -n sched:::change-pri''.

Adam

-- 
Adam Leventhal, Fishworks                     http://blogs.sun.com/ahl

A relevant piece of information that I overlooked in the original Java RTS forum
posting is that dtrace is being launched via pfexec. I just received additional
info that if the user logs in as root and doesn''t use pfexec then it
works fine.

So I guess this may just be a permissions issue, though the failure mode is
somewhat unexpected.

David
-- 
This message posted from opensolaris.org

The problem is not having the dtrace_kernel permission.

I don''t know if this failure mode is a reasonable one for a missing
permission, but that''s for you guys to decide.

Thanks,
David Holmes
-- 
This message posted from opensolaris.org