public inbox for systemtap@sourceware.org
 help / color / mirror / Atom feed
* scheduler.stp scheduler.migrate having problems on 2.6.20-1
@ 2007-05-21 21:02 Nathan DeBardeleben
  2007-05-21 21:36 ` Frank Ch. Eigler
  0 siblings, 1 reply; 2+ messages in thread
From: Nathan DeBardeleben @ 2007-05-21 21:02 UTC (permalink / raw)
  To: systemtap

While it seems to find the pull_task kernel function, I seem to be 
unable to access the internal variables from that function.  I was 
hoping to use this to look at how the kernel moves around processes in a 
4 processor dual core (8 CPU) per node cluster I'm running on.  I'm 
looking to experiment with whether the kernel really is giving processes 
the memory that is closest to which CPU it has chosen to run a process 
on.  This may not be the right way (would love some comments) but 
regardless it doesn't seem to be working on my system:

> Pass 1: parsed user script and 54 library script(s) in 
> 160usr/10sys/177real ms.
> semantic error: failed to retrieve location attribute for local 'p' 
> (dieoffset: 0x2ae510): identifier '$p' at 
> /usr/local/share/systemtap/tapset/scheduler.stp:93:12
> semantic error: failed to retrieve location attribute for local 'p' 
> (dieoffset: 0x2ae510): identifier '$p' at 
> /usr/local/share/systemtap/tapset/scheduler.stp:94:16
> semantic error: failed to retrieve location attribute for local 
> 'this_cpu' (dieoffset: 0x2ae501): identifier '$this_cpu' at 
> /usr/local/share/systemtap/tapset/scheduler.stp:95:14
> Pass 2: analyzed script: 2 probe(s), 0 function(s), 0 embed(s), 0 
> global(s) in 240usr/80sys/311real ms.
> Pass 2: analysis failed.  Try again with more '-v' (verbose) options.
Here's the system:

> [root@kraken1 tapset]# rpm -qa | grep kernel
> kernel-devel-2.6.20-1.2316.fc5
> kernel-2.6.20-1.2316.fc5
> kernel-debuginfo-2.6.20-1.2316.fc5
> [root@kraken1 tapset]# uname -a
> Linux kraken1 2.6.20-1.2316.fc5 #1 SMP Fri Apr 27 19:19:10 EDT 2007 
> x86_64 x86_64 x86_64 GNU/Linux
> [root@kraken1 tapset]# 
Maybe this got inlined?

-- 
-- Nathan
Correspondence
---------------------------------------------------------------------
Nathan DeBardeleben, Ph.D.
Los Alamos National Laboratory
Parallel Tools Team
High Performance Computing Environments (HPC-4)
phone: 505-667-3428
email: ndebard@lanl.gov
---------------------------------------------------------------------

^ permalink raw reply	[flat|nested] 2+ messages in thread

* Re: scheduler.stp scheduler.migrate having problems on 2.6.20-1
  2007-05-21 21:02 scheduler.stp scheduler.migrate having problems on 2.6.20-1 Nathan DeBardeleben
@ 2007-05-21 21:36 ` Frank Ch. Eigler
  0 siblings, 0 replies; 2+ messages in thread
From: Frank Ch. Eigler @ 2007-05-21 21:36 UTC (permalink / raw)
  To: Nathan DeBardeleben; +Cc: systemtap

Nathan DeBardeleben <ndebard@lanl.gov> writes:

> While it seems to find the pull_task kernel function, I seem to be
> unable to access the internal variables from that function.  [...]

Yes, yet another victim of bug #1155.

> I'm looking to experiment with whether the kernel really is giving
> processes the memory that is closest to which CPU it has chosen to
> run a process on.

Neat.

You might try using a statement probe instead, choosing a source
file:line somewhere near the beginning of the pull_task function.

- FChE

^ permalink raw reply	[flat|nested] 2+ messages in thread

end of thread, other threads:[~2007-05-21 21:36 UTC | newest]

Thread overview: 2+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2007-05-21 21:02 scheduler.stp scheduler.migrate having problems on 2.6.20-1 Nathan DeBardeleben
2007-05-21 21:36 ` Frank Ch. Eigler

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).