Re: [Qemu-devel] [PATCH for-4.1] tcg: Fix constant folding of INDEX_op_extract2_i32

[Alex Bennée]

Richard Henderson <address@hidden> writes:

> On a 64-bit host, discard any replications of the 32-bit
> sign bit when performing the shift and merge.
>
> Fixes: https://bugs.launchpad.net/bugs/1834496
> Signed-off-by: Richard Henderson <address@hidden>

Reviewed-by: Alex Bennée <address@hidden>
Tested-by: Alex Bennée <address@hidden>

> ---
>  tcg/optimize.c | 4 ++--
>  1 file changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/tcg/optimize.c b/tcg/optimize.c
> index d7c71a6085..d2424de4af 100644
> --- a/tcg/optimize.c
> +++ b/tcg/optimize.c
> @@ -1213,8 +1213,8 @@ void tcg_optimize(TCGContext *s)
>                  if (opc == INDEX_op_extract2_i64) {
>                      tmp = (v1 >> op->args[3]) | (v2 << (64 - op->args[3]));
>                  } else {
> -                    tmp = (v1 >> op->args[3]) | (v2 << (32 - op->args[3]));
> -                    tmp = (int32_t)tmp;
> +                    tmp = (int32_t)(((uint32_t)v1 >> op->args[3]) |
> +                                    ((uint32_t)v2 << (32 - op->args[3])));
>                  }
>                  tcg_opt_gen_movi(s, op, op->args[0], tmp);
>                  break;


--
Alex Bennée