Re: [PATCH v1] firmware loader: introduce module parameter tocustomize fw search path

[Kevin Easton]

Quoting Al Viro <viro@xxxxxxxxxxxxxxxxxx>:

> On Thu, Oct 25, 2012 at 08:38:25PM -0700, Linus Torvalds wrote:
>
> > It's valid to cast a non-const pointer to a const one. It's the
> > *other* way around that is invalid.
> >
> > So marking fw_path[] as having 'const char *' elements just means that
> > we won't be changing those elements through the fw_path[] array
> > (correct: we only read them). The fact that one of those same pointers
> > is then also available through a non-const pointer variable means that
> > they can change through *that* pointer, but that doesn't change the
> > fact that fw_path[] itself contains const pointers.
> >
> > Remember: in C, a "const pointer" does *not* mean that the thing it
> > points to cannot change. It only means that it cannot change through
> > *that* pointer.
>
> It's a bit trickier, unfortunately - pointer to pointer to const char
> and pointer to pointer to char do not mix.  Just for fun, try to constify
> envp and argv arguments of call_usermodehelper()...

That's because if it _was_ allowed, you could use it to silently  
launder the const away:

const char *c = "rodata";
char *x;
const char **y;

y = &x;
*y = c;

/* We now have (const char) values accessible through a (char *) pointer x */



--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/