PPC32 and state of the project

[Steve Munroe]

Ulrich Drepper writes:

> Kevin.Hendricks wrote:
>
> > Why not save the thread id register when doing a setjmp/longjmp pair?
> > It should never change across a setjmp, longjmp pair should it?
>
> Correct.  A thread is a thread.  If you use longjmp or swapcontext
> mustn't change this.  Whatever is used as the thread register, it must
> always be preserved.

Just to clarify. __longjmp and setcontext/swapcontext should not change the
thread register. (i.e. preserved as in not touched)?

At least this matches the implementation of sysdeps/i386/__longjmp.S and
sysdeps/powerpc/powerpc64/__longjmp.S. However
sysdeps/powerpc/powerpc32/__longjmp.S does restore r2 so that needs to
change.