[PATCH, aarch64] Fix pr69305 -- addti miscompilation
Richard Henderson
rth@redhat.com
Wed Jan 27 18:18:00 GMT 2016
On 01/27/2016 09:31 AM, James Greenhalgh wrote:
>> +(define_insn "*add<mode>3_compare1_cconly"
>
> I don't understand the naming scheme, it got me a wee bit confused with
> add<mode>3_compare0 and friends, where the 0 indicates a comparison with
> zero...
Ah, well, I didn't get that 0 indicated comparison with zero.
I thought it was simply zero-th out of a list of alternatives.
Certainly quite a lot of other patterns are like that.
Indeed, the add.*carryin patterns that I delete in this very
patch were marked [0-3].
>> +;; Note that a single add with carry is matched by cinc,
>> +;; and the adc_reg and csel types are matched into the same
>> +;; pipelines by existing cores.
>
> I can't see us remembering to update this comment on pipeline models
> were it to ever become false. Maybe just drop it?
Fair enough; I'll just mention cinc without the scheduling note.
r~
More information about the Gcc-patches
mailing list