[Mspgcc-users] GCC behaviour

[Joerg Huebler]

Hello,

I'm trying to adapt the FreeRTOS (www.freertos.org) to the MSP430(F449) and 
found
some strange behaviour of the compiler in the file 
"list.c" function "pvListGetOwnerOfNextEntry".

Original Sourcecode is:

volatile void *pvListGetOwnerOfNextEntry( volatile xList *pxList )
{
        if( !listLIST_IS_EMPTY( pxList ) )
        {
                /* Increment the index to the next item and return the item, 
ensuring
                we don't return the marker used at the end of the list. */
                pxList->pxIndex = pxList->pxIndex->pxNext;
#ifdef __GNUrC__
                if( (pxList->pxIndex - pxList->pxHead) == NULL )
#else
                if( pxList->pxIndex == pxList->pxHead )
#endif
                {
                        pxList->pxIndex = pxList->pxIndex->pxNext;
                }
                return pxList->pxIndex->pvOwner;
        }
        return NULL;
}



and that's what the compiler generated:


#ifdef __GNUrC__
                if( (pxList->pxIndex - pxList->pxHead) == NULL )
#else
                if( pxList->pxIndex == pxList->pxHead )
    1438:       1e 4d 02 00     mov     2(r13), r14     ;
    143c:       2f 4d           mov     @r13,   r15     ;
    143e:       05 24           jz      $+12            ;abs 0x144a
#endif

I'm missing a comparision between 143c and 134e ?
(To get a working result just remove the "r" from __GNUrC__)

Is it really a bug, do I have to modify arguments of the command line or
anything else ?


I'm using msp430-gcc 3.2.3 (binaries downloaded from www.mikrocontroller.net)
Win98/Cygwin. For looking to the entire sources I used they are attached.

JoergJ. Hübler
GEMAC mbH
Zwickauer Str. 227 
09116 Chemnitz
Tel. +49 371 3377 350
Fax. +49 371 3377 272
mailto: hueb...@gemac-chemnitz.de



--
GEMAC Virencheck: OK
Virus checked by G DATA AntiVirusKit
Version: AVK 12.0.535 from 22.08.2003

FreeRTOS.tar.bz2
Description: Binary data