Re: A patch for http://sourceware.cygnus.com/ml/binutils/2000-04/msg00373.html

[Franz Sirl <Franz dot Sirl-kernel at lauterbach dot com>]

At 15:45 17.04.00, Alan Modra wrote:
>On Mon, 17 Apr 2000, Franz Sirl wrote:
>
> > Am Sun, 16 Apr 2000 schrieb H . J . Lu:
> > >The old gld${EMULATION_NAME}_place_orphan assumes it only processes
> > >SEC_ALLOC sections. When we allow ~SEC_ALLOC sections, we have to
> > >check it. This patch seems to fix
> > >
> > >http://sourceware.cygnus.com/ml/binutils/2000-04/msg00373.html
>
>I thought I had run the testsuite, but obviously not.  I'll try not to be
>so lax again..
>
> > this patch solves the vmlinux relocation errors on PPC I got, but if I 
> compile
> > the kernel with -gdwarf-2, it seems the debug sections no longer get 
> merged now:
>
>The following patch hopefully Does The Right Thing.  I've incorporated
>H.J.'s fix along with a few of my own, and I'll commit it after hearing
>back from you.  (This one does pass "make check" on x86!)
>
>--
>Linuxcare.  Support for the Revolution.
>
>ld/ChangeLog
>         * emultempl/elf32.em (struct orphan_save): Add section field.
>         (gld${EMULATION_NAME}_place_orphan): Use above to keep sections in
>         better order, and place first orphan section as we did before the
>         2000-04-12 patch.  Ignore ~SEC_ALLOC sections when choosing place.
>         Don't call make_bfd_section here, let wild_doit do the job for us.
>         Don't build a statement list when we'll only throw it away.

Yes, this one works nice for me on PPC too. All sections are in perfect 
order now and the debug sections get merged.

Thanks,
Franz.