On Sat, 14 May 2011, Jim Cromie wrote:
> >> drivers/tty/tty_io.c: unregister_chrdev_region(MKDEV(driver->major,
> >> driver->minor_start),
> >> drivers/tty/tty_io.c: register_chrdev_region(MKDEV(TTYAUX_MAJOR,
> >> 0), 1, "/dev/tty") < 0)
> >> drivers/tty/tty_io.c: register_chrdev_region(MKDEV(TTYAUX_MAJOR,
> >> 1), 1, "/dev/console") < 0)
> >>
> >> Ive tried this rule to fix it:
> >>
> >> @ mdev_inline @
> >> dev_t devid;
> >> expression ct, name;
> >> expression major, minor;
> >> position p;
> >> @@
> >>
> >> {
> >> @p
> >> ...
> >> - register_chrdev_region(MKDEV(major, minor), ct, name)
> >> ...
> >> }
> >>
> >> @p
> >> + dev_t devid = MKDEV(major, minor);
> >>
> >> + register_chrdev_ids(devid, ct, name)
> >
> > I'm not sure what you mean here. Are the last three lines a separate
> > rule?
>
>
> I hadnt intended it, but accept that it might be necessary.
> @p was supposed to mark the beginning of the inner {scope}
> where its legal to insert a local variable decl with the first +line
>
> The "rule" (sarcastic air-quotes)
> had the ... between the beginning of scope and the -line,
> but didnt capture it for reuse.
>
> >
> > Also, a position variable has to be attached to something, which is the
> > token that comes just before it.
>
> does '{' ie 'open new scope' count ?
Yes, anything except ... <... ...> <+... ...+>
> > Finally, register_chrdev_ids(devid, ct, name) is an expression, not a
> > statement (missing semicolon).
>
> That was intended. Some of the uses are embedded inside
> if (-here-) so wouldnt be a complete statement.
Actually, what I cited was the + code. If you have eg:
+ foo();
+ bar()
that won't make a legal program, because a statement has to be followed by
another statement.
julia
> > If you make a second rule
> >
> > @@
> > position rule1.p; // assuming the first was called rule1
> > @@
> >
> > {@p
> > + dev_t devid = MKDEV(major, minor);
> > + register_chrdev_ids(devid, ct, name);
> > ...
> > }
> >
> > then it should put the declaration at the top of the innermost pair of
> > braces.
>
> ok, that seems to confirm my question: {@p is meaningful.
> (and better written on same line - clearer intent)
>
> > julia
> >
>
> thanks again. I'll go try this now,
> _______________________________________________
Cocci mailing list
[email protected]
http://lists.diku.dk/mailman/listinfo/cocci
(Web access from inside DIKUs LAN only)
