Hello Jan, I'm not sure to understand your problem, but here is a nice feature I use to get expression types. I think this example doesn't do what you want but I guess it shows how to get types from expressions:
@r1@ expression ep; identifier f, fld; type t; t e; @@ [email protected] = f @script:python@ ep << r1.ep; fld << r1.fld; t << r1.t; @@ print t, ep, fld There are two important parts: @r1@ ... type t; t e; @@ which just tells coccinelle that you are interested in e of type t, but you don't want to match the declaration. Then the '@e' on the line: [email protected] = f makes e match the expression that comes before '.fld', and as the type of that expression is t, you can use t on your rules (to print, import it in other rules, ...). Hope it helps... On Mon, Sep 14, 2015 at 10:29 AM, Jan Moskyto Matejka <[email protected]> wrote: > Hello, > > I'm trying to match all expressions of a given type. I want to check > whether all accesses to any part of some struct are locked. > > In the hypothetical case (which doesn't happen), every access to that > struct would create a local variable of the wanted type and I could then > match it with something like this: > > @@ > identifier i; > @@ > > ... > my_specific_type *var; > ... > var->i > ... > > Unfortunately, this structure may be also referenced from another > structure, like this: > > struct A { int foo; int bar; }; > struct B { int foo; struct A *bar }; > > struct B *b; > ... > b->bar->foo; > > I would like to write something like: > > @@ > expression struct A* E; > @@ > > ... > +E > ... > > and match every occurence of not only variables of the given type, but > also an expression. > > I was first trying to find some hint in the doc, found > expression struct *E; > which is almost what I want but still too broad. > > Does there exist some way to > A) achieve the behaviour I want by config > or > B) implement it? > > If A, could you please give me a hint? > If B, I wanted to implement it but got totally lost in the code. Could > you please give me a pointer, where to begin? I'd like to contribute the > result then. > > Best regards and thanks > MQ > _______________________________________________ > Cocci mailing list > [email protected] > https://systeme.lip6.fr/mailman/listinfo/cocci -- Peter _______________________________________________ Cocci mailing list [email protected] https://systeme.lip6.fr/mailman/listinfo/cocci
