[patch] RTEMS pthread.h changes

Corinna Vinschen vinschen@redhat.com
Wed Dec 8 23:24:00 GMT 2010

On Dec  8 14:54, Ralf Corsepius wrote:
> On 12/08/2010 11:16 AM, Corinna Vinschen wrote:
> >On Dec  8 11:02, Corinna Vinschen wrote:
> >>Hi Ralf,
> >>
> >>On Dec  7 17:09, Ralf Corsepius wrote:
> >>>Hi,
> >>>
> >>>I am submitting the patch below on behalf of Joel, who currently is
> >>>busy with other obligations.
> >>>
> >>>This patch adds a couple of POSIX pthread functions RTEMS is
> >>>supporting and adds a new field "guardsize" to pthread_attr_t RTEMS
> >>>internally applies to implement these functions.
> >>>
> >>>The latter is an ABI change and therefore might not be suitable for
> >>>other OSes. May-be it should be enclosed into #ifndef __CYGWIN__ or
> >>>#ifdef __rtems__ defines?
> >>
> >>Cygwin has its own pthread.h, so Cygwin doesn't matter here.  However,
> >>since it's an ABI change I'd suggest to enclose it in __rtems__
> >>brackets.  It should only be enabled on a case by case basis.
> OK, with me.
> Just to prevent miscommunications: I was referring to the guardsize
> part in sys/types.h. The pthread.h part only extends the API, but
> doesn't introduce ABI changes.

Yes, that's how I understood it.

> >...unless we can be sure that RTEMS and Cygwin are the only newlib-based
> >systems which actually implement pthreads.
> The ABI change is inside of the patch to sys/types.h :)

Nevertheless, for systems which don't implement pthreads it doesn't
quite matter how you change the definition of pthread_attr_t.

> Updated patch enclosed below (Now with hopefully correct
> ChangeLog entries)



Corinna Vinschen
Cygwin Project Co-Leader
Red Hat

More information about the Newlib mailing list