> >> Attached is a script I want to use to backport drivers. However, the
> >> second (unnamed) rule matches for each occurence of 'struct
> >> net_device_ops'. Is there a way to disable a rule after a desired amount
> >> of matches (in my case once)?
> >
> > Here is a somewhat clunky solution:
> >
> > @initialize:python@
> > @@
> >
> > first_ops = 0
> >
> > @r@
> > identifier OPS;
> > position p;
> > @@
> >
> > struct net_device_ops OPS@p = { ... };
> >
> > @script:python depends on r@
> > @@
> >
> > first_ops = 0
> >
> > @script:python@
> > p << r.p;
> > @@
> >
> > ln = int(p[0].line)
> > if first_ops == 0 or ln < first_ops:
> > first_ops = ln
> >
> > @script:python@
> > p << r.p;
> > @@
> >
> > ln = int(p[0].line)
> > if not(first_ops == ln):
> > cocci.include_match(False)
> >
> > @r1 exists@
> > expression ndevexp, e1, e2;
> > identifier func;
> > @@
> > func(...) {
> > <+...
> > - ndevexp->min_mtu = e1;
> > - ndevexp->max_mtu = e2;
> > ...+>
> > }
> >
> > @r2@
> > expression r1.e1,r1.e2;
> > identifier r.OPS;
> > @@
> > + static int __change_mtu(struct net_device *ndev, int new_mtu)
> > + {
> > + if (new_mtu < e1 || new_mtu > e2)
> > + return -EINVAL;
> > + ndev->mtu = new_mtu;
> > + }
> > +
> > struct net_device_ops OPS = {
> > ...
> > };
> >
> > @depends on r2@
> > identifier OPS;
> > @@
> >
> > struct net_device_ops OPS = {
> > + .ndo_change_mtu = __change_mtu,
> > ...
> > };
> >
> > -------------------------
>
> Big thanks. Have not tried it yet, but do you want me to add your
> Signed-off-by given that you mostly solved the problem.
OK.
>
> > Basically, the idea is to find the line number of the start of each
> > net_device_ops structure. The first script:python rule initializes
> > first_ops for the current file. The second one passes through the
> > position information for each net_device_ops structure and finds the
> > smallest line number. The third python rule discards all matches that
> > don't have the name of the structure on the collected smallest line.
>
> Yeah. Was wondering how to tackle that as I read in the documentation
> that the order in which matches are processed is not predictable.
>
> > Rule
> > r1 then matches the two assignments, rule r2 puts the new function on top
> > of the earliest detected structure (determined by the inherited value of
> > OPS), and then the last rule added the field to the structure.
>
> Understand. Thanks again.
>
> > I noticed that there is a problem though when the structure is define
> > within a function. Then the added function goes in that structure too.
> > The affected file is: drivers/tty/n_gsm.c. Maybe you odn't want o
> > backport that one?
>
> It is not (yet) in the backports copy-list so we are good for now.
Great. I guess one could solve the issue by more clunky position
manipulations to detect when a declaration is inside some function
definition and when it is at top level.
julia
_______________________________________________
Cocci mailing list
[email protected]
https://systeme.lip6.fr/mailman/listinfo/cocci
