[Intel-gfx] [PATCH i-g-t 2/2] tests/gem_pwrite_snooped: fix const cast warning

Ville Syrjälä ville.syrjala at linux.intel.com
Mon Sep 7 08:18:36 PDT 2015


On Mon, Sep 07, 2015 at 03:49:58PM +0100, Thomas Wood wrote:
> On 7 September 2015 at 13:53, Ville Syrjälä
> <ville.syrjala at linux.intel.com> wrote:
> > On Mon, Sep 07, 2015 at 12:34:12PM +0100, Thomas Wood wrote:
> >> Signed-off-by: Thomas Wood <thomas.wood at intel.com>
> >> ---
> >>  tests/gem_pwrite_snooped.c | 4 ++--
> >>  1 file changed, 2 insertions(+), 2 deletions(-)
> >>
> >> diff --git a/tests/gem_pwrite_snooped.c b/tests/gem_pwrite_snooped.c
> >> index d3f6223..29142c3 100644
> >> --- a/tests/gem_pwrite_snooped.c
> >> +++ b/tests/gem_pwrite_snooped.c
> >> @@ -76,14 +76,14 @@ static void blit(drm_intel_bo *dst, drm_intel_bo *src,
> >>       intel_batchbuffer_free(batch);
> >>  }
> >>
> >> -static void *memchr_inv(const void *s, int c, size_t n)
> >> +static void const *memchr_inv(const void *s, int c, size_t n)
> >
> > Oh, and I'll just note that I wrote it the way I did orignally so that
> > it matches memchr(). But I suppose matching memchr() exactly is not all
> > that important.
> 
> It may actually be more useful to remove const from the function
> parameter rather than add it to the return type.

The function doesn't change the data, so passing in const definitely
makes sense. I guess it could also use __attribute__((pure))

> Other than adding or
> removing the const annotations, there doesn't seem to be any other way
> of avoiding the warning from -Wcast-qual in this case.

Would be nice if we could have const and non-const versions of the
function, so that the constness of the returned type depends on what
gets passed in. But this is C and not C++ so we can't have that. And 
even in C++ you would then have to make the function argument to
non-const as well, which loses the information that it doesn't change
the data.

> 
> 
> >
> >>  {
> >>       const unsigned char *us = s;
> >>       unsigned char uc = c;
> >>
> >>       while (n--) {
> >>               if (*us != uc)
> >> -                     return (void *) us;
> >> +                     return (void const *) us;
> >>               us++;
> >>       }
> >>
> >> --
> >> 1.9.1
> >>
> >> _______________________________________________
> >> Intel-gfx mailing list
> >> Intel-gfx at lists.freedesktop.org
> >> http://lists.freedesktop.org/mailman/listinfo/intel-gfx
> >
> > --
> > Ville Syrjälä
> > Intel OTC

-- 
Ville Syrjälä
Intel OTC


More information about the Intel-gfx mailing list