2011/6/27 Georg-Johann Lay <avr@gjlay.de>:
> Denis Chertykov wrote:
>> 2011/6/26 Georg-Johann Lay <avr@gjlay.de>:
>>> Denis Chertykov schrieb:
>>>> 2011/6/24 Richard Henderson <rth@redhat.com>:
>>>>
>>>>> On 06/23/2011 01:15 PM, Denis Chertykov wrote:
>>>>>
>>>>>>>  text    data     bss     dec     hex filename
>>>>>>> 10032      25       0   10057    2749 bld-avr-orig/gcc/z.o
>>>>>>>  5816      25       0    5841    16d1 bld-avr-new/gcc/z.o
>>>>>> Richard, can you send me this z.c file ?
>>>>>> Right now I'm notice that new code is worse.
>>>>> That's gcc.c-torture/compile/950612-1.c.
>>>> I have founded that postreload optimizations can't handle results of
>>>> new L_R_A code.
>>>> I think that it's can be handled by CSE (postreload).
>>> Did you try to add constraint alternative to *addhi3?
>>> Like "*!d,d,n" or even "*!r,r,n"
>>>
>>> I saw some code improvement with that alternative.
>>
>> I'm trying:
>>
>> (define_insn "*addhi3"
>>   [(set (match_operand:HI 0 "register_operand" "=r,!w,!w,d,r,r,!d")
>>       (plus:HI
>>        (match_operand:HI 1 "register_operand" "%0,0,0,0,0,0,!r")
>>        (match_operand:HI 2 "nonmemory_operand" "r,I,J,i,P,N,!ri")))]
>>   ""
>>   "@
>>       add %A0,%A2\;adc %B0,%B2
>>       adiw %A0,%2
>>       sbiw %A0,%n2
>>       subi %A0,lo8(-(%2))\;sbci %B0,hi8(-(%2))
>>       sec\;adc %A0,__zero_reg__\;adc %B0,__zero_reg__
>>       sec\;sbc %A0,__zero_reg__\;sbc %B0,__zero_reg__
>>         #"
>>   [(set_attr "length" "2,1,1,2,3,3,4")
>>    (set_attr "cc" "set_n,set_czn,set_czn,set_czn,set_n,set_n,set_n")])
>>
>
> That split will split always:
>
>> ;; Special split three addressing addhi3
>> ;; to make postreload optimization possible
>> (define_split ; addhi3 !d,!r,!ri
>>   [(set (match_operand:HI 0 "d_register_operand" "")
>>       (plus:HI (match_operand:HI 1 "register_operand" "")
>>                (match_operand:HI 2 "nonmemory_operand" "")))]
>>   "reload_completed"
>     && REGNO(operands[0]) != REGNO(operands[1])"
>>   [(set (match_dup 0) (match_dup 2))
>>    (set (match_dup 0) (plus:HI (match_dup 0) (match_dup 1)))]
>>   "")
> Maybe it can also restrict to const_int_operand in #2 and then it's
> best to
>    (set (match_dup 0)
>         (match_dup 1))
>    (set (match_dup 0)
>         (plus:HI (match_dup 0)
>                  (match_dup 2)))

Thanks for suggestions.

>> The main problem for me is that the new addressing mode produce a
>> worse code in many tests.
>
> You have an example source?

In attachment.

Denis.