On 9/15/25 13:31, Denis Sergeev wrote:
> In gen_mcrfs() the FPSCR nibble mask is computed as:
> `~((0xF << shift) & FP_EX_CLEAR_BITS)`
>
> Here, 0xF is of type int, so the left shift is performed in
> 32-bit signed arithmetic. For bfa=0 we get shift=28,
> and (0xF << 28) = 0xF0000000, which is not representable as a 32-bit
> signed int. Static analyzers flag this as a potential integer
> overflow.
>
> Found by Linux Verification Center (linuxtesting.org) with SVACE.
>
> Signed-off-by: Denis Sergeev <zeff@altlinux.org>
> ---
> target/ppc/translate/fp-impl.c.inc | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/target/ppc/translate/fp-impl.c.inc b/target/ppc/translate/fp-impl.c.inc
> index a66b83398b..d01c0b5aba 100644
> --- a/target/ppc/translate/fp-impl.c.inc
> +++ b/target/ppc/translate/fp-impl.c.inc
> @@ -479,7 +479,7 @@ static void gen_mcrfs(DisasContext *ctx)
> tcg_gen_extu_tl_i64(tnew_fpscr, cpu_fpscr);
> /* Only the exception bits (including FX) should be cleared if read */
> tcg_gen_andi_i64(tnew_fpscr, tnew_fpscr,
> - ~((0xF << shift) & FP_EX_CLEAR_BITS));
> + ~(MAKE_64BIT_MASK(shift, 4) & FP_EX_CLEAR_BITS));
> /* FEX and VX need to be updated, so don't set fpscr directly */
> tmask = tcg_constant_i32(1 << nibble);
> gen_helper_store_fpscr(tcg_env, tnew_fpscr, tmask);
Reviewed-by: Chinmay Rath <rathc@linux.ibm.com>
Thanks,
Chinmay