Re: Re: Re: Implicition declarations of functions and bugs

["David Mosberger-Tang" <David.Mosberger@acm.org>]

Samuel,

You're missing the point of the check-implicit-pointer-functions
script.  Its purposes is not to "grep for warnings" but instead to
look for pairs of warnings that are *guaranteed* to cause crashes on
64-bit machines.  gcc -Wall normally spits out tons of spurious
warnings for 64-bit dirty code, but most of those warnings are just
noise.  The problem with gcc-4.0 warnings is that you can't
distinguish between harmless implicit function declarations and ones
that need to be flagged.  Example:

$ cat t.c
char *
foo (char *str)
{
  return strdup (str);
}

enum e_t { a, b };

enum e_t
bar (char *str)
{
  return strlen (str);
}
$ gcc-3.3 -c -g -O -Wall t.c
t.c: In function `foo':
t.c:4: warning: implicit declaration of function `strdup'
t.c:4: warning: return makes pointer from integer without a cast
t.c: In function `bar':
t.c:12: warning: implicit declaration of function `strlen'
$ gcc-4.0 -c -g -O -Wall t.c
t.c: In function 'foo':
t.c:4: warning: implicit declaration of function 'strdup'
t.c:4: warning: incompatible implicit declaration of built-in function 'strdup'
t.c: In function 'bar':
t.c:12: warning: implicit declaration of function 'strlen'
t.c:12: warning: incompatible implicit declaration of built-in function 'strlen'

As you can see, the gcc-4.0 warnings are inferior because they
provides no way to distinguish the "foo" vs. "bar" case, even though
they are qualitatively different.

  --david

On 3/10/06, Samuel Thibault <samuel.thibault@ens-lyon.org> wrote:
> David Mosberger-Tang, le Fri 10 Mar 2006 17:06:22 -0700, a écrit :
> > I'm inclined to treat this as a gcc-4 bug.
>
> It is not.
>
> > $ cat t.c
> > char *
> > foo (char *str)
> > {
> >   return strdup(str);
> > }
> > $ gcc-3.3 -c -g -O -Wall t.c
> > t.c: In function `foo':
> > t.c:4: warning: implicit declaration of function `strdup'
> > t.c:4: warning: return makes pointer from integer without a cast
>
> Because strdup() here gets an implicit
> int strdup(int str)
> declaration, hence the warnings.
>
> > gcc-4.0 -c -g -O -Wall t.c
> > t.c: In function 'foo':
> > t.c:4: warning: implicit declaration of function 'strdup'
> > t.c:4: warning: incompatible implicit declaration of built-in function 'strdup'
>
> Same story, except that gcc has additionnal knowledge of which prototype
> the strdup function should have: char *strdup(const char *str); . And
> char * and int are not compatible types.
>
> > The gcc-3.3 warnings makes perfect sense.  The gcc-4.0 warnings are
> > useless.
>
> gcc-4.0 warnings are actually just more precise: not only there is a
> missing declaration, but gcc has strong conviction that the implicit
> prototype is really wrong (strdup() should really take char * and return
> char *, not int).
>
> > There is no hint on how the "implicit declaration of built-in function
> > `strdup'" is incompatible.
>
> Implicit declarations are in the form
> int foo(int bar, int baz, etc.)
> and the built-in strdup function has
> char *strdup(const char *str)
> as prototype. This is incompatible. gcc could even say "int is
> incompatible with char*", but I guess gcc people consider this as too
> verbose.
>
> This is a warning and not an error, because using one's own strdup()
> function (that would take ints) is perfectly legal. gcc-4.0 emits the
> warning to let the programmer know that he should disambiguate this by
> either #including the usual C header, or by giving the prototype of his
> own strdup() function.
>
> Anyway, such warnings deserve grepping, since they are evidence of a
> potential binary break.
>
> Regards,
> Samuel
>


--
Mosberger Consulting LLC, http://www.mosberger-consulting.com/