Re: Systemtap performance impact on N800

["Bruno Abinader" <bruno dot abinader at openbossa dot org>]

Hi Frank,
Thank you for the valuable feedback. Some comments below:

On Tue, May 13, 2008 at 9:56 AM, Frank Ch. Eigler <fche@redhat.com> wrote:
> "Bruno Abinader" <bruno.abinader@openbossa.org> writes:
>
>  > We at INdT created a document called "Systemtap performance impact on
>  > N800", with performance comparisons between Kprobes, Systemtap and
>  > Oprofile usage. Feel free to read and ask us questions about our
>  > results!
>
>  Thanks for running these tests.  One important thing to watch out for
>  is that you're comparing apples to apples.
>
>  For example, the stap_gettimeofday.stp file in section 2.2.3 asks to
>  probe both function entry and return, whereas the kprobe code listed
>  only probes function entry.  For another, the stap_tapsets.stp test in
>  section 3.2 asks to trace some data for every probe hit, but I see no
>  corresponding kprobe source to confirm that equivalent output is being
>  generated there.

I've looked into the k-009.c code (from section 2.2.3) and noticed
that kretprobes implementation was missing, so I've fixed it (inserted
 kretprobes code into it) and now the Kprobes-only implementation has
probes on begin and end of do_gettimeofday function. It is interesting
to notice that with this modification, kprobes results took a higher
value, from 3 to 5 seconds, approaching to actual systemtap results.

As for section 3.2, there is really no kprobe source. We made some
confusion at the time of writing the document and put it there, but
the only results were made for these scenarios:
- default kernel
- kprobes (not used)
- systemtap (with kprobes)
- oprofile

Thanks!

[]s

>
>  - FChE

-- 
Bruno de Oliveira Abinader
Mobile Linux Software (MLS) / Instituto Nokia de Tecnologia (INdT)
Tel: +55 (92) 2126-1068