Change RTEMS GCC Thread model to POSIX.

[Joel Sherrill]

On 8/12/2013 12:13 PM, Ralf Corsepius wrote:
> On 08/12/2013 03:41 PM, Gedare Bloom wrote:
>> Is anything lost by going to POSIX threads in GCC?
> RTEMS has been using native RTEMS threads in GCC.
>
> The purpose of this had been:
> - Avoiding to have POSIX support hard-wired into RTEMS.
That was true but almost all of the code that was linked in initially is
left out now. The biggest culprit was POSIX signals and that added
about 3k to the code space on SPARC.
> - Code Size (RTEMS POSIX threads are wrappers around RTEMS threads).
No. Cllassic API tasks and POSIX API threads are both direct instances of
Score Thread. They are equivalent but initialized to different settings.
> Because of these reasons, similar proposals had been rejected before.
The main reason was that we didn't want POSIX threads to always have to 
be on.

Most of the reasons we made this decision no longer exist. Plus
if we move to per function sections and some type of constructor type
scheme for initialization, linked pthread applications will be smaller
than Classic API equivalents today.
> Ralf
>
>
> _______________________________________________
> rtems-devel mailing list
> rtems-devel at rtems.org
> http://www.rtems.org/mailman/listinfo/rtems-devel


-- 
Joel Sherrill, Ph.D.             Director of Research & Development
joel.sherrill at OARcorp.com        On-Line Applications Research
Ask me about RTEMS: a free RTOS  Huntsville AL 35805
Support Available                (256) 722-9985