Re: [Qemu-devel] [PATCH] fix gcc4 compile warnings

["andrzej zaborowski" <balrogg@gmail.com>]

On 30/11/2007, Andre Przywara <andre.przywara@amd.com> wrote:
>  > These casts are not the right way to get rid of the warnings, as are
>  > some of the casts in other files in qemu_put_* and qemu_get_*
>  > arguments. In this case the warnings are true positives and the bugs
>  > causing the warnings have to be addressed instead of just the
>  > warnings.
>
> Are you sure of that? Most of the fixes are like this:
>  >> -    qemu_put_be32s(f, &s->count_shift);
>  >> +    qemu_put_be32s(f, (uint32_t *)&s->count_shift);
> qemu_put_be32s is (QEMUFile *f, const uint32_t *pv), but after all the
> 2nd argument is only a _pointer_ to an unsigned variable, the size is
> the same (thanks to the C99 explicit types).

count_shift is an int so it is the machine word size, not necesarily
32-bit AFAIK. Otherwise I guess gcc wouldn't warn. You can possibly
use:

qemu_put_be32(f, s->count_shift);

Or point qemu_put_be32_s() to a int32_t/uint32_t variable.

>
> What solution do you prefer for the opaque types? I have used the simple:
>  >> -    void *args[MAX_ARGS];
>  >> +    intptr_t args[MAX_ARGS];
> A more portable and clean solution would be this:
> -    void *args[MAX_ARGS];
> +    union
> +    {
> +        void* ptr;
> +        int i;
> +    } args[MAX_ARGS];
> If you prefer this, I can change the patch accordingly.

I'm not sure why you get a warning here and I'm unable to run a build
at the moment. A void * should be able to store some (unknown size)
integer regardless of the platform.

Regards