On Thu, 20 May 2010, Arnaud Lacombe wrote:
> Hi,
>
> On Thu, May 20, 2010 at 3:14 PM, Julia Lawall <[email protected]> wrote:
> > On Thu, 20 May 2010, Arnaud Lacombe wrote:
> >
> >> Hi,
> >>
> >> I'm running into some [old] code using K&R C function definition:
> >>
> >> void
> >> rtalloc(ro)
> >> register struct route *ro;
> >> {
> >> ... function body ...
> >> }
> >>
> >> Which is making spatch fails with:
> >>
> >> bad: void
> >> BAD:!!!!! rtalloc(ro)
> >> ERROR-RECOV: found sync '}' at line 56
> >> parsing pass2: try again
> >> ERROR-RECOV: found sync '}' at line 56
> >> parse error
> >> = File "rt.c", line 27, column 0, charpos = 501
> >> around = '{', whole content = {
> >> badcount: 30
> >> bad: register struct route *ro;
> >> BAD:!!!!! {
> >> [...]
> >> SEMANTIC:parameter name omitted, but I continue
> >> TYPEDEF CONFLICT:prt
> >> TODO:typedef now used as an identifier
> >>
> >>
> >> While the Graal would be to have spatch be capable to convert K&R to
> >> ANSI C, would it be feasible to teach it not to fail on such
> >> declaration ?
> >
> > There seems to be a bit more to triggering the error than the bit of code
> > shown above.
> you're right, I reduced the failure to the following:
>
> void
> f0(a)
> register struct s0 *a;
> {
> }
>
> void
> f1(b)
> register struct s1 *b;
> {
> struct s2 *a;
> f(a->xdst);
> }
>
> which returns:
>
> init_defs_builtins: /data/opt/coccinelle/share/coccinelle/standard.h
>
> PARSING: route.c
> (ONCE) CPP-MACRO: found toplevel macro noptvirg: f0
> ERROR-RECOV: found sync col 0 at line 3
> parsing pass2: try again
> ERROR-RECOV: found sync col 0 at line 3
> parse error
> = File "route.c", line 2, column 3, charpos = 8
> around = 'a', whole content = f0(a)
> badcount: 1
> bad: void
> BAD:!!!!! f0(a)
> ERROR-RECOV: found sync '}' at line 5
> parsing pass2: try again
> ERROR-RECOV: found sync '}' at line 5
> parse error
> = File "route.c", line 4, column 0, charpos = 34
> around = '{', whole content = {
> badcount: 2
> bad: register struct s0 *a;
> BAD:!!!!! {
> bad: }
> (ONCE) CPP-TYPEDEF: promoting: b
>
> f1() is still parsed correctly.
You don't have an example that provokes the TODO?
The heuristic seems to be that K&R is recognized if the type list is
indented. If you put a space in front of register in the definition of
f0, then everything is ok. I will see if the indentation is necessary,
but I don't know if it will be possible. As long as there is no code you
care about in f0, it doesn't matter, unless you also hit the TODO. It
will just go an parse the rest of the file.
julia
_______________________________________________
Cocci mailing list
[email protected]
http://lists.diku.dk/mailman/listinfo/cocci
(Web access from inside DIKUs LAN only)
