Re: [Cocci] Using cocci to mangle function arguments

[Julia Lawall <julia.lawall@inria.fr>]

On Sun, 17 May 2020, Thomas Adam wrote:

> Hi all,
>
> I have a situation where I'm trying to use Coccinelle to both rename one
> function to another, and at the same time modify the number of arguments.
>
> I have a function called:
>
>     func_old(ERR, "func_name", "%s message", charstring);
>
> Which is therefore variadic.  In terms of argument ordering func_old() takes:
>
>   1.  log-level type
>   2.  the calling function
>   3.  A variadic string
>
> I'm wanting to replace func_old() with func_new() such that it looks like
> this:
>
>     func_new("%s: %s message", __func__, charstring);
>
> Hence, func_new() reduces the number of arguments to just one -- a format
> string, and variadic arguments.
>
> My question is how would I go about trying to get coccinelle to help me
> translate this?  Is this even possible?  I have tried:
>
> @@
> expression O1, O2, O3
> @@
>
> - func_old(O1, O2, O3, ...);
>
> But I don't know how to convert what would be O2 to '__func__'.  When calling
> func_new(), there is no explicit O2 parameter from func_old(), it should be
> part of the format string, hence:
>
> + func_new("%s: ...", __func__, O3);
>
> I suspect I might be stretching coccinelle's abilities in trying to craft new
> parameters, but I thought I'd ask.

I'm not sure to completely understand the problem.  Is it that O1 is
always an explicit string and you want to add %s: at the beginning of that
string?

If you require O1 to be an explicit string, it would be better to declare
it as:

constant char[] O1;

For creating the new string, you can use python.  Check
demos/pythontococci.cocci.  Basically, you need to play with the data you
are given in the python code to create the string that you want.
According to the example, when you use it in the final SmPL rule that
creates the new code, you will call it an identifier, even though it is a
string.  But it doesn't matter.

There is a more clean way to do this with some make_expr function, but
there doens't seem to be a convenient demo for it.

julia


>
> Apologies if this is convoluted.  If I can help answer any additional
> questions, or if something's not clear, let me know.
>
> TIA!
>
> Thomas
> _______________________________________________
> Cocci mailing list
> Cocci@systeme.lip6.fr
> https://systeme.lip6.fr/mailman/listinfo/cocci
>
_______________________________________________
Cocci mailing list
Cocci@systeme.lip6.fr
https://systeme.lip6.fr/mailman/listinfo/cocci