[Savannah-register-public] [task #4192] Submission of pDI-Tools

[Gerardo García Peña]

Follow-up Comment #2, task #4192 (project administration):

> Hi,
> I'm evaluating the project you submitted for approval in
> Savannah.

> Note that we do not want to encourage software primarily aimed
> at cracking purposes. Would you mind telling us what is the
> primary purpose of the software?

NO! It's is  not for cracking purposes! It is one of its possible uses
because it is very powerful. And, in spite of we like it or not, it can be
used for that purpose.

Its primary target is super computing: instrumentation of code at runtime. It
has been coded for making performance evaluation on scientific programs on
multis and clusters. Its first mission would be tracing MPI communications
and show them graphically with all details.  I will try to make it debut at
Marenostrum [1]. Marenostrum is a supercomputer made of IBM PowerPC Server
Blades running GNU/Linux. It is part of BSC [2] (Barcelona Supercomputer
Center).

I have coded pDI-Tools for CEPBA [3] (CEPBA will be probably assimilated by
BSC in the near future). They have a dynamic instrumentation program called
DI-Tools, but it is not so powerful as pDI-Tools. I have write pDI-Tools from
zero and all copyright is mine, and I have decided to put it under a LGPL
license.

The original DI-Tools only run on Irix with MIPS and it is not portable,
while pDI-Tools run on a lot of a platforms and it is very portable.

pDI-Tools is designed as a replacement for DI-Tools. DI-Tools is the base of
a lot of programs of CEPBA, for instance PARAVER [4] and DIMEMAS [5]. Both
are graphical interpreters of the data collected by an instrumentation
program called OMPItrace, which is built on top DI-Tools. Making a free
pDI-Tools can motivate CEPBA to free its software in the future or can be the
base of future free instrumentation software.

Currently there is only one Open Source Dynamic Instrumentation API: Dyninst
[6]. Dyninst is NOT FREE. And it is currently the only option. The most
similar free software I have found is Elfsh [7] (that can be also used for
cracking  ;) ). It is possible to write a instrumentation software on Elfsh,
but this is not the primary objective of Elfsh. Elfsh is more oriented to
debug ELF structures
and playing with them on runtime, so it can do more powerful modifications
but it is also more difficult to port to another archs (it only runs on Sparc
32 (¿64?) and i386, while pDI-Tools run on i386, PPC, PPC64, SPARC32, SPARC64,
MIPS N32 & MIPS 64).

So I think that it is necesary to fill this hole at free software giving a
tool such pDI-Tools.

In the other hand pDI-Tools can be used to make execution drive simulations.
For instance with pDI-Tools you can substitute some code with your own code.
Your new code can intercept some critical funcs as open, read, write, etc...
and make them fail sometimes. Doing this you can check the robustness of
your
program in a very unfavorable scenario. So it is not only useful for
supercomputer labs.

[1] Marenostrum TOP500: http://top500.org/sublist/System.php?id=7119
[2] Barcelona Supercomputer Center: http://www.bsc.org.es/
[3] CEPBA: http://www.cepba.upc.es/
[4] PARAVER: http://www.cepba.upc.edu/paraver/
[5] DIMEMAS: http://www.cepba.upc.edu/dimemas/
[6] Dyninst: http://www.dyninst.org/
[7] Elfsh: http://elfsh.segfault.net/

> Please include a (perhaps temporary) URL pointing to
> the source code.
>
> We wish to review your source code, even if it is not
> functional, to catch potential legal issues early.

If you want I can send you a .tar.gz with all code and docs, but remember
that this version is too preliminar to show it to the public. It is a very
early version and it is not very stable.

If you only want to see the source code you can find it at:
    http://kung-foo.dhs.org/killabyte/pdi/

the principal sources are in
    http://kung-foo.dhs.org/killabyte/pdi/src/link

I am worried about legal problems so I wrote all from scratch without looking
to any program except gdb and glibc.

> For example, to release your program properly under the GPL you
> must include a copyright notice and permission-to-copy
> statements at the beginning of every file of source code. This
> is explained in
> http://www.gnu.org/licenses/gpl-howto.html.
> Our review would help catch potential omissions such as these.

Thanks, I will read it, but already I included the legal notice  ;)

My best regards,
    Gerardo García Peña


    _______________________________________________________

Reply to this item at:

  <http://savannah.gnu.org/task/?func=detailitem&item_id=4192>

_______________________________________________
  Message sent via/by Savannah
  http://savannah.gnu.org/