This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: Anyone knows about a testcase which was fixed by Wed Dec 24 18:05:13 1997 SPARC change?
- To: rth at cygnus dot com
- Subject: Re: Anyone knows about a testcase which was fixed by Wed Dec 24 18:05:13 1997 SPARC change?
- From: Mike Stump <mrs at windriver dot com>
- Date: Mon, 16 Oct 2000 21:56:17 -0700 (PDT)
- Cc: gcc at gcc dot gnu dot org, jakub at redhat dot com
> Date: Mon, 16 Oct 2000 17:55:48 -0700
> From: Richard Henderson <rth@cygnus.com>
> You're not listening: we reload the *entire* register window
> for the target function in the process of "longjmping" back.
> That takes care of the pic register and everything else that
> must be restored.
I'm sorry, I'm just muddied the waters. I agree with most of what has
been said already. I don't think reloading the PIC register was
necessary in the egcs (and now the gcc) tree in the non flat case.
There are some permutations of the code that would have required the
reloading of the pic register in the non flat case, but I don't know
if those permutations ever existed in the old FSF tree or the ACT
tree. In the flat case:
;; ??? Doesn't work with -mflat.
(define_expand "nonlocal_goto"
That kinda sums it up. The change in question might be enough to make
it limp along, but I don't know. My original comment was speculation
(this is why I used the word imagine) about why kenner might have done
it. Sorry I didn't label it better.