On 2020/7/23 04:30, Richard Sandiford wrote:
>
> I now realise the reason is that the starting mode is too wide.
> I think we should fix that by doing:
>
>FOR_EACH_MODE_IN_CLASS (new_mode_iter, MODE_INT)
> {
>…
>
> and then add:
>
>if (maybe_lt (GET_MODE_SIZE (new_mo
Richard Sandiford writes:
> luoxhu writes:
>> Hi,
>>
>> On 2020/7/22 19:05, Richard Sandiford wrote:
>>> This wasn't really what I meant. Using subregs is fine, but I was
>>> thinking of:
>>>
>>>/* Also try a wider mode if the necessary punning is either not
>>> desirable or not po
gument to simplify_gen_subreg above). It shouldn't be necessary
to use smallest_int_mode_for_size to find a different mode.
That's also why the multiple_p is the way round it is above. The idea
is that the shift amount must be a multiple of the size of the outer mode
(here new_mode) in or
_for_size (shift)" to get rhs_subreg,
otherwise it will return NULL if new_mode is equal to store_mode.
Lastly, move the BITS_PER_UNIT after multiple_p as it still need generate
shift_seq for other circumstances. :)
[PATCH v3] dse: Remove partial load after full store for high part
access[P
