Re: [PATCH 1/2] tcg: avoid integer overflow

[Peter Maydell <peter.maydell@linaro.org>]

On Mon, 16 Mar 2020 at 13:15, Yifei Jiang <jiangyifei@huawei.com> wrote:
>
> This fixes coverity issues 75234842, etc.,:

Where does this issue number come from, by the way?
It's not from the online Coverity Scan we use which
is the issue ID we usually cite for coverity stuff.

>     2221    tcg_gen_andi_i64(t, t, dup_const(vece, 1));
> CID 75234842: (OVERFLOW_BEFORE_WIDEN)
>     2222. overflow_before_widen: Potentially overflowing expression "1 << nbit" with type "int" (32 bits, signed) is evaluated using 32-bit arithmetic, and then used in a context that expects an expression of type "int64_t" (64 bits, signed).
>     2222    tcg_gen_muli_i64(t, t, (1 << nbit) - 1);

Again, you need to apply a more critical eye to the Coverity
suggestions. For instance:

> diff --git a/tcg/tcg-op-gvec.c b/tcg/tcg-op-gvec.c
> index 327d9588e0..3aeb049a46 100644
> --- a/tcg/tcg-op-gvec.c
> +++ b/tcg/tcg-op-gvec.c
> @@ -2219,7 +2219,7 @@ static void gen_absv_mask(TCGv_i64 d, TCGv_i64 b, unsigned vece)
>      /* Create -1 for each negative element.  */
>      tcg_gen_shri_i64(t, b, nbit - 1);
>      tcg_gen_andi_i64(t, t, dup_const(vece, 1));
> -    tcg_gen_muli_i64(t, t, (1 << nbit) - 1);
> +    tcg_gen_muli_i64(t, t, ((int64_t)1 << nbit) - 1);

In this function nbit can only be 8 or 16, so this shift
can never overflow.

I haven't checked whether any of the others are valid.

thanks
-- PMM