Re: [head tinderbox] failure on arm/arm

On Sun, Nov 12, 2006 at 03:42:30PM +0100, Giorgos Keramidas wrote:
> On 2006-11-12 17:00, Ruslan Ermilov <ru_at_freebsd.org> wrote:
> > > /src/lib/libelf/elf_rand.c: In function `elf_rand':
> > > /src/lib/libelf/elf_rand.c:47: warning: cast increases required alignment of target type
> > > *** Error code 1
> > >
> > > Stop in /src/lib/libelf.
> > > *** Error code 1
> >
> > This looks like a GCC bug to me.  The following code snippet,
> > when compiled on FreeBSD/arm, causes a -Wcast-align warning
> > which doesn't look right:
> >
> > %%%
> > $ cat a.c
> > struct foo {
> > 	char x;
> > };
> >
> > struct foo *
> > bubu(char *s)
> > {
> >
> > 	return (struct foo *)s;
> > }
> > $ cc -c -Wcast-align a.c
> > a.c: In function `bubu':
> > a.c:9: warning: cast increases required alignment of target type
> > %%%
> >
> > (None of other supported architecutes see the issue here.)
> 
> You can't cast any random (char *) pointer to a pointer of a type which
> is (potentially) larger than 1 byte.  It's the same sort of warning you
> will get if you try to:
> 
>     char ch[] = "\x00\x00\x00\x00";
>     char *p = &(ch[0]);
>     unsigned long *lptr = (unsigned long *)p;
> 
> You cannot guarantee that `ch' is stored in an address that is properly
> aligned for (unsigned long), and this is what GCC warns about here.
> 
No, your example I perfectly understand but it is completely different.
Note that the first (and only) member in my structure is "char", so it
doesn't need to be more than sizeof(char) aligned.

> On 2006-11-12 15:27, Stefan Farfeleder <stefan_at_fafoe.narf.at> wrote:
> > What is sizeof(struct foo)?  If it's > 1 it makes sense.
> 
> Exactly :)
> 
Still doesn't make much sense to me.  If all structure members are chars
(like is the case with "struct ar_hdr" from <ar.h> which GCC complains
about, and in my example, the required alignment shouldn't be more than
sizeof(char).  What am I missing?


Cheers,
-- 
Ruslan Ermilov
ru_at_FreeBSD.org
FreeBSD committer