Re: architecture paper draft

[Richard J Moore <richardj_moore at uk dot ibm dot com>]

- -
Richard J Moore
IBM Advanced Linux Response Team - Linux Technology Centre
MOBEX: 264807; Mobile (+44) (0)7739-875237
Office: (+44) (0)1962-817072

"Frank Ch. Eigler" <fche@redhat.com> wrote on 09/02/2005 23:34:19:

> Hi -
>
>
> > 1) You have a box labelled "probe kernel object". I know that the
kprobes
> > patch that's been accepted by Linux and Andrew only supports kernel
object
> > probing, but the full patch allows for system-wide probing i.e.
includes
> > user-space.  [...]
>
> The diagram is misleading you.  By "kernel object", I am referring to
> the ".ko" file that results from translating/compiling the input script.
> Whether that script relates to kernel or user space is independent
> of this.

Thanks, I understand.


>
> (For what it's worth, I'm still hoping for greater expertise to
> make itself known on the subject of use of copy_from_user and its
> kin from arbitrary kprobes contexts, to provide safety guarantees.)
>
> > 2) I think there is an additional final step in the flow: "automated
> > reduction/analysis". There will be occasions where post-tracing data
> > reduction has to be automated and tools or even extensions to systemTAP
> > will be built to meet this need.
>
> Do you envision specific systemtap code for this?  What would it do?
> Or would it be a separate third-party program?
>
Could be either. The degree to which data should be processed at the point
it's gathered is a moot point and dependent on circumstance. I can envisage
have the option of performing statistical calculations after data gathering
has ended.
>
> - FChE
> [attachment "att6ibdp.dat" deleted by Richard J Moore/UK/IBM]