On 31 January 2011 16:53, Ulrich Weigand <ulrich.weig...@de.ibm.com> wrote:
> Ira Rosen <ira.ro...@linaro.org> wrote:
>
>> (define_insn "neon_vzip<mode>_internal"
>>   [(set (match_operand:VDQW 0 "s_register_operand" "=w")
>>        (unspec:VDQW [(match_operand:VDQW 1 "s_register_operand" "0")]
>>                     UNSPEC_VZIP1))
>>    (set (match_operand:VDQW 2 "s_register_operand" "=w")
>>         (unspec:VDQW [(match_operand:VDQW 3 "s_register_operand" "2")]
>>                     UNSPEC_VZIP2))]
>>   "TARGET_NEON"
>>   "vzip.<V_sz_elem>\t%<V_reg>0, %<V_reg>2"
>>   [(set (attr "neon_type")
>>       (if_then_else (ne (symbol_ref "<Is_d_reg>") (const_int 0))
>>                     (const_string "neon_bp_simple")
>>                     (const_string "neon_bp_3cycle")))]
>> )
>>
>> Is there a way to properly mark the dependence?
>
> I guess you need to make sure the UNSPEC_VZIP forms both explicitly
> depend on both input operands.  This can be a bit tricky due to
> various reload constraints on how matching constraints and/or
> match_dup can be used, but I think something along the following
> lines should work:
>
>   [(set (match_operand:VDQW 0 "s_register_operand" "=w")
>        (unspec:VDQW [(match_operand:VDQW 1 "s_register_operand" "0")
>                      (match_operand:VDQW 2 "s_register_operand" "w")]
>                     UNSPEC_VZIP1))
>    (set (match_operand:VDQW 3 "s_register_operand" "=2")
>         (unspec:VDQW [(match_dup 1) (match_dup 2)]
>                     UNSPEC_VZIP2))]
>
>

Thanks a lot! It seems to work. It fixed the problem and I am now
testing the patch on the rest of the vectorizer testsuite.

Thanks,
Ira

>
> Mit freundlichen Gruessen / Best Regards
>
> Ulrich Weigand
>
> --
>  Dr. Ulrich Weigand | Phone: +49-7031/16-3727
>  STSM, GNU compiler and toolchain for Linux on System z and Cell/B.E.
>  IBM Deutschland Research & Development GmbH
>  Vorsitzender des Aufsichtsrats: Martin Jetter | Geschäftsführung: Dirk
> Wittkopp
>  Sitz der Gesellschaft: Böblingen | Registergericht: Amtsgericht
> Stuttgart, HRB 243294
>
>

_______________________________________________
linaro-toolchain mailing list
linaro-toolchain@lists.linaro.org
http://lists.linaro.org/mailman/listinfo/linaro-toolchain

Reply via email to