FW: Defect in GCC 4.4.0 bfin Port : Wrong code generated while accessing multiple arrays with different loop index variables within loop (-Os option)

[Bernd Schmidt]

> Consider the following test case:
> 
> Test Case Reference:
> --------------------
> 
> int siVect[40] ;
> int siCoeff[40] ;
> int siSumofDotProduct  ;
> int siIndex1, siIndex2 ;
> 
> vTestMultipleArrayAccessWithDifferentLoopIndex()
> {
>         for (siIndex1=0 ; siIndex1<40 ; siIndex1++)
>         {
>             siSumofDotProduct += siVect[siIndex1] * siCoeff[siIndex2]; 
>             siIndex2++;
>         }
> }

This is not a complete testcase; it needs additional code to call this
function and test the result.  When I added this, it produced the same
result with -O0, -O2 and -Os, hence I don't know...

> 
> Assembly generated by bfin port (GCC 4.4.0 --target=bfin-elf) (-Os):
> --------------------------------------------------------------------
>         P1 = 41 (X);
>         LSETUP (.L3, .L6) LC1 = P1;  <---- (1)
>         jump.s .L2;                  <---- (2)
> .L3:
>         R2 = [P0++];
>         R3 = [P5++];
>         R2 *= R3;
>         R0 += 1;
>         R1 = R1 + R2;
> .L2:
>         R2 = P2;
> .L6:
>         P2 += 1;
> 
> 
> Here, marked instruction (2) "jump.s" is wrongly generated along with
> hardware loop.

... what exactly you believe is wrong with this.


Bernd
-- 
This footer brought to you by insane German lawmakers.
Analog Devices GmbH      Wilhelm-Wagenfeld-Str. 6      80807 Muenchen
Sitz der Gesellschaft Muenchen, Registergericht Muenchen HRB 40368
Geschaeftsfuehrer Thomas Wessel, William A. Martin, Margaret Seif