[Mesa-dev] [RFC 3/7] nir: coverity unitialized pointer read

[Matt Turner]

On Wed, May 18, 2016 at 8:54 AM, Rob Clark <robdclark at gmail.com> wrote:
> From: Rob Clark <robclark at freedesktop.org>
>
> Not sure how coverity arrives at the conclusion that we can read comp[j]
> unitialized (around line 204), other than not being aware that ncomp is
> greater than 1 so it won't underflow in the 'if (tex->is_array)' case.
> ---
>  src/compiler/nir/nir_lower_tex.c | 6 ++++++
>  1 file changed, 6 insertions(+)
>
> diff --git a/src/compiler/nir/nir_lower_tex.c b/src/compiler/nir/nir_lower_tex.c
> index a080475..c05d48b 100644
> --- a/src/compiler/nir/nir_lower_tex.c
> +++ b/src/compiler/nir/nir_lower_tex.c
> @@ -177,6 +177,12 @@ saturate_src(nir_builder *b, nir_tex_instr *tex, unsigned sat_mask)
>        /* split src into components: */
>        nir_ssa_def *comp[4];
>
> +      /* NOTE: coord_components won't be >4 or <1 but coverity doesn't
> +       * know this:
> +       */

I'd drop the comment. git blame will allow us to figure out why the
assume() is there if needed.

> +      assume(tex->coord_components < ARRAY_SIZE(comp));
> +      assume(tex->coord_components >= 1);

I think the second one is sufficient, since part of the path involves
ncomp-- I think that it believes coord_components can be zero so
subtracting 1 will produce UINT_MAX.

With the comment and the first assume() dropped,

Reviewed-by: Matt Turner <mattst88 at gmail.com>