change log for rtems (2011-07-01)

Joel Sherrill joel.sherrill at OARcorp.com
Sun Jul 3 18:53:58 UTC 2011


On 07/03/2011 01:07 PM, Sebastian Huber wrote:
> On 03/07/11 19:46, Joel Sherrill wrote:
>> On 07/03/2011 12:49 PM, Sebastian Huber wrote:
>>> On 01/07/11 21:11, rtems-vc at rtems.org wrote:
>>>> diff -u rtems/cpukit/libcsupport/src/malloc_initialize.c:1.14
>>>> rtems/cpukit/libcsupport/src/malloc_initialize.c:1.15
>>>> --- rtems/cpukit/libcsupport/src/malloc_initialize.c:1.14    Wed Aug
>>>> 25 09:30:01 2010
>>>> +++ rtems/cpukit/libcsupport/src/malloc_initialize.c    Fri Jul  1
>>>> 13:23:09 2011
>>>> @@ -45,7 +45,7 @@
>>>>     {
>>>>       /*
>>>>        *  If configured, initialize the statistics support
>>>> -  */
>>>> +   */
>>>>       if ( rtems_malloc_statistics_helpers != NULL ) {
>>>>         (*rtems_malloc_statistics_helpers->initialize)();
>>>>       }
>>>> @@ -59,11 +59,13 @@
>>>>        *  Initialize the optional sbrk support for extending the heap
>>>>        */
>>>>       if ( rtems_malloc_sbrk_helpers != NULL ) {
>>>> -    heap_begin = (*rtems_malloc_sbrk_helpers->initialize)(
>>>> +    void *new_heap_begin = (*rtems_malloc_sbrk_helpers->initialize)(
>>>>           heap_begin,
>>>>           sbrk_amount
>>>>         );
>>>> -    heap_size  = (uintptr_t) sbrk_amount;
>>>> +
>>>> +    heap_size -= (uintptr_t) new_heap_begin - (uintptr_t) heap_begin;
>>>> +    heap_begin = new_heap_begin;
>>> With this change libtests/malloc04 fails.
>> On what BSP?  It passed on psim which failed everything before it
>> was reverted.
> I think this passed only by accident.  In the test we have heap_begin ==
> NULL == 0x0.  If the memory starts not at 0x0 the problem becomes obvious.
>
Without a working test case I can tinker with, I can't do much.

AFAIK BSP sbrk() is working for those BSPs that have it
and all BSPs which don't have sbrk() support are working.

If you can find a fix, please commit it.  This only impacts the head.
>>> _______________________________________________
>>> rtems-vc mailing list
>>> rtems-vc at rtems.org
>>> http://www.rtems.org/mailman/listinfo/rtems-vc




More information about the vc mailing list