newlib malloc/memalign
Joel Sherrill
joel.sherrill at OARcorp.com
Tue Sep 30 15:29:07 UTC 2003
Chuck Tuffli wrote:
> On Mon, Sep 29, 2003 at 04:52:53PM -0500, Joel Sherrill wrote:
>
>>Chuck Tuffli wrote:
>>
>>
>>>I saw a posting* in March that talks about switching from using RTEM's
>>>malloc to the one in newlib. I'm interested in using the memalign
>>>function and was wondering what would I need to do to implement this
>>>support.
>>>
>>>*
>>>http://www.rtems.com/rtems/maillistArchives/rtems-users/2003/march/msg00050.html
>>
>>The first step would be to disable the malloc code in RTEMS,
>>provide the malloc mutex used by newlib (see the gxx_thread_wrappers
>>in RTEMS), and arrange for newlib to use the memory formerly
>>given to our own malloc.
>
>
> Ok, I'm probably being dense,
I must be dense this week because that seems to be a common comment. :)
> but I don't see a mutex being taken
> in the malloc code in cpukit/libcsupport/src/malloc.c (unless the
> mutex is part of rtems_region_get_segment()).
The Region manager uses an internal RTEMS mutex for allocation. So
it isn't explicit in the RTEMS malloc code.
> I'm also not finding
> the gxx_thread_wrappers you refered to in either newlib or RTEMS.
> Where am I going wrong?
cpukit/libcsupport/src/gxx_wrappers.c is the set of routines used
by gcc run-time support functions to be thread-safe.
newlib's malloc implementation newlib/libc/stdlib/mallocr.c expect
MALLOC_LOCK and MALLOC_UNLOCK to be defined. There are other macros
in there to define as well.
> This is building off the CVS tag rtems-ss-20030703, newlib 1.11 and
> gcc 3.2.3
>
>
>>>--
>>>Chuck Tuffli
>>>Agilent Technologies
>>
>>
>>--
>>Joel Sherrill, Ph.D. Director of Research & Development
>>joel at OARcorp.com On-Line Applications Research
>>Ask me about RTEMS: a free RTOS Huntsville AL 35805
>>Support Available (256) 722-9985
>
>
--
Joel Sherrill, Ph.D. Director of Research & Development
joel at OARcorp.com On-Line Applications Research
Ask me about RTEMS: a free RTOS Huntsville AL 35805
Support Available (256) 722-9985
More information about the users
mailing list