Re: [PATCH] Support backtrace () on IA-64, make it work even with -fomit-frame-pointer on AMD64

[Jakub Jelinek <jakub at redhat dot com>]

On Thu, Oct 09, 2003 at 04:04:42PM +0200, Andreas Schwab wrote:
> Jakub Jelinek <jakub@redhat.com> writes:
> 
> > On Thu, Oct 09, 2003 at 02:47:53PM +0200, Andreas Schwab wrote:
> >> Jakub Jelinek <jakub@redhat.com> writes:
> >> 
> >> > Sorry, I only compiled it outside of glibc and there it worked.
> >> > Here is a better patch (built on amd64, passed make check, checked output
> >> > of a few sample proglets):
> >> 
> >> Doesn't seem to work on ia64, though.
> >
> > Works for me just fine (gcc-3_3-rhl-branch libgcc_s.so.1),
> > with various CFLAGS used for backtrace-tst (-O2, -O0, with or without
> > -Wl,--export-dynamic, etc.).
> > What kind of libgcc you're using?
> 
> Just the one that comes with gcc 3.3.1.

Diffing gcc-3_3{,-rhl}-branch, I think just
2003-05-16  Jakub Jelinek  <jakub@redhat.com>

        * config/ia64/unwind-ia64.c (uw_update_reg_address): Handle
        .save XX, r0.
might be the difference which matters.
This patch is on the trunk, gcc-3_3-rhl-branch and gcc-3_2-rhl8-branch, but not on
gcc-3_3-branch.
Might ask Mark if he allows that for 3.3.2...
Certainly, without that not just backtrace, but lots of other things (like
NPTL) don't work properly.

	Jakub