Frama-C-discuss mailing list archives

This page gathers the archives of the old Frama-C-discuss archives, that was hosted by Inria's gforge before its demise at the end of 2020. To search for mails newer than September 2020, please visit the page of the new mailing list on Renater.

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[Frama-c-discuss] Need help on C-code slicing

  • Subject: [Frama-c-discuss] Need help on C-code slicing
  • From: pascal.cuoq at gmail.com (Pascal Cuoq)
  • Date: Thu, 9 Jun 2011 15:45:54 +0200
  • In-reply-to: <BANLkTi=eKHMSg8Sdq88OwqqO890tjdg+Cw@mail.gmail.com>
  • References: <BANLkTi=eKHMSg8Sdq88OwqqO890tjdg+Cw@mail.gmail.com>
> I compiled my sources with gcc -save-temps
> and run frama-c-gui - val myfile.i providing only the i file I'm interested
> in.
>
> I get a "Degeneration occurred"? banner in the window title bar.. and seem
> to be unable to get any usable result from this point.

Here are a couple of suggestions to go further:

- if you use the batch version frama-c instead of the GUI version frama-c-gui,
you will get a partial analysis log that can provide indices.

- you are in the same situation as Sali Sene in the discussion thread
"[Frama-c-discuss] Value analysis aborted" at the bottom of
http://lists.gforge.inria.fr/pipermail/frama-c-discuss/2011-May/thread.html
You are in a situation that I did my best to prevent, and it's a bug
that this happened.

- the situation is that the program writes at a completely unknown
address. The reason why this shouldn't happen is that it should be
impossible to have a completely unknown value for the evaluation
of any expression. The origin of your problem is the first completely
unknown value during the analysis of the program (once you have
a completely unknown value, it is normal that it contaminates
any computations that involve it). The "completely unknown" value
is denoted as "{‌{ANYTHING}‌}". In the case of Sali Sene, it had
something to do with function pointers, but it was not easy to
reproduce on a simple case for the precise reason that I tried
to make it impossible to happen. Therefore, no fix is available
yet, but there is still hope.

- I am very much looking forward to your reduced examples
where you show "{‌{ANYTHING}‌}" appearing during the analysis.
If you can produce these reduced (or even, if the code's license
allows, unreduced) examples, I will probably be able to fix
the issue.
Frama_C_dump_each() is your friend: it shows the entire
state at the time of the call. An {‌{ANYTHING}‌} anywhere
in this state is abnormal behavior.

Pascal