On Sat, 15 Aug 2020 at 02:31, Richard Henderson
<richard.hender...@linaro.org> wrote:
>
> We want to ensure that access is checked by the time we ask
> for a specific fp/vector register. We want to ensure that
> we do not emit two lots of code to raise an exception.
>
> But sometimes it's difficult to cleanly organize the code
> such that we never pass through sve_check_access exactly once.
> Allow multiple calls so long as the result is true, that is,
> no exception to be raised.
>
> Signed-off-by: Richard Henderson <richard.hender...@linaro.org>
> ---
> target/arm/translate.h | 1 +
> target/arm/translate-a64.c | 27 ++++++++++++++++-----------
> 2 files changed, 17 insertions(+), 11 deletions(-)
Reviewed-by: Peter Maydell <peter.mayd...@linaro.org>
> diff --git a/target/arm/translate.h b/target/arm/translate.h
> index 16f2699ad7..ad7972eb22 100644
> --- a/target/arm/translate.h
> +++ b/target/arm/translate.h
> @@ -64,6 +64,7 @@ typedef struct DisasContext {
> * that it is set at the point where we actually touch the FP regs.
> */
> bool fp_access_checked;
> + bool sve_access_checked;
Is there anywhere it's worthwhile to put in an equivalent
of assert_fp_access_checked() for sve_access_checked, or is
there no point that's both (a) common to SVE accesses and
(b) not common to generic FP accesses ? One could put it
in pred_full_reg_offset() I suppose but I dunno if that
really gains us much. The existing fp_access_checked will
catch "forgot the check entirely" which seems more likely
as a bug than "put in the FP check when we wanted SVE".
thanks
-- PMM
