Fix up some unusual rounding modes.

r~

Richard Henderson (2):
  fpu: Handle all rounding modes in partsN_uncanon_normal
  fpu: Handle all rounding modes in partsN_round_to_int_normal

 fpu/softfloat-parts.c.inc | 5 +++++
 1 file changed, 5 insertions(+)

-- 
2.43.0

On Mon, 8 Jun 2026 at 20:03, Richard Henderson
<richard.henderson@linaro.org> wrote:
>
> Fix up some unusual rounding modes.
>
> r~
>
> Richard Henderson (2):
>   fpu: Handle all rounding modes in partsN_uncanon_normal
>   fpu: Handle all rounding modes in partsN_round_to_int_normal
>

I'm going to apply this to target-arm.next since it's
a prerequisite for the FEAT_FP8 series.

thanks
-- PMM

On Mon, 8 Jun 2026 at 20:03, Richard Henderson
<richard.henderson@linaro.org> wrote:
>
> Fix up some unusual rounding modes.
>
> r~
>
> Richard Henderson (2):
>   fpu: Handle all rounding modes in partsN_uncanon_normal
>   fpu: Handle all rounding modes in partsN_round_to_int_normal
>
>  fpu/softfloat-parts.c.inc | 5 +++++
>  1 file changed, 5 insertions(+)

roundAndPackFloatx80() is the other function that doesn't
implement all the rounding modes. But I guess we can assume
that 80-bit floats are not going to ever need them. (It will
assert if called with anything other than nearest_even,
ties_away, to_zero, round_up or round_down, so we'll easily
spot the cause if we ever need the functionality.)

thanks
-- PMM