sys/types.h is broken for _POSIX_THREADS

[Ralf Corsepius]

On 02/24/2010 11:50 AM, Renato Caldas wrote:
> On Wed, Feb 24, 2010 at 10:30 AM, Ralf Corsepius
> <ralf.corsepius@rtems.org>  wrote:
>> On 02/24/2010 11:03 AM, Renato Caldas wrote:
>>>
>>> On Wed, Feb 24, 2010 at 1:49 AM, Joel Sherrill
>>> <joel.sherrill@oarcorp.com>    wrote:
>>>>
>>>> On 02/23/2010 06:19 PM, Renato Caldas wrote:
>>
>>> Any help is really appreciated. In the meantime, I guess I can test
>>> with prebuilt RPMs..
>>
>> Check how we build the RPMS. The way you are trying to build is not how
>> newlib is supposed to be built and is substantially different from how we
>> build the toolchains (the nominal way).
>
> I've noticed you place newlib inside gcc and build them together.
Correct - This is what is commonly known as "one-tree-style"-building.

> I
> didn't do it because it didn't seem to make sense (as you said, its
> not how newlib is supposed to be built).
We do so, because gcc and newlib are closely tied together and mutually 
depend on each other.

> Is there a reason for building them that way?
Yes. Somewhat oversimplied, GCC needs to know some 
defines/types/prototypes/headers newlib provides and newlib needs to be 
supplied with defines/types/prototypes/headers and characteristics GCC 
provides (e.g. multilibs).

I.e. building newlib standalone for RTEMS can only work if newlib's 
ABI/API and GCC's ABI/API doesn't change between 2 build-iterations.

Ralf