Re: [PRELIMINARY]: Patch to add bfd support for IBM s390

[DJBARROW at de dot ibm dot com]

 Martin Schwidefsky is responsible for the BFD & Binutils
port he will be able to fill you in with these details & whether the
BFD_RELOC entries are valid or not.

The breakpoint.c kludge I did as far as I remember is probably unwanted, I
just found that the
hardware breakpoints in 4-18 were in absolutely terrible shape staying in
when not wanted &
disappearing for no reason, has this code improved ?, my clueless kludge
improved the situation a tiny
bit in the test cases I was playing with.
I haven't had a chance to see whether the 5.0 code has improved
substantially in my opinion
this code needs a lot of attention by someone who knows it, if this code
has improved recently
please leave this out.

The readline addition to  configure.in is because our compiler by default
uses unsigned characters
& results in some conditions in readline not behaving as expected without
this addition.

I also had to add this ugly kludge in trad-core.c as the configure script
was leaving
TRAD_HEADER defined i.e
#define TRAD_HEADER /* Nothing */
 as something didn't completely undef it & attempted to include
thus attempted to include fresh air, I didn't know the best way to fix
this.
--- gdb-5.0/bfd/trad-core.c   Mon Feb 21 13:01:25 2000
+++ gdb-5.0-s390/bfd/trad-core.c   Wed Jul 26 00:09:16 2000
@@ -30,6 +30,9 @@

 #include <sys/user.h>        /* After a.out.h  */

+#ifdef __s390__
+#undef TRAD_HEADER /* problem with configure script */
+#endif
 #ifdef TRAD_HEADER
 #include TRAD_HEADER
 #endif

D.J. Barrow Linux for S/390 kernel developer
eMail: djbarrow@de.ibm.com,barrow_dj@yahoo.com
Phone: +49-(0)7031-16-2583
IBM Germany Lab, Schönaicherstr. 220, 71032 Böblingen


Geoff Keating <geoffk@cygnus.com> on 25.08.2000 00:07:33

Please respond to Geoff Keating <geoffk@cygnus.com>

To:   msnyder@redhat.com
cc:   binutils@sourceware.cygnus.com, bfd-local@cygnus.com,
      nickc@cygnus.com, Denis Joseph Barrow/Germany/Contr/IBM@IBMDE, Boas
      Betzler/Germany/IBM@IBMDE, Gregory Burke/Poughkeepsie/IBM@IBMUS
Subject:  Re: [PRELIMINARY]: Patch to add bfd support for IBM s390




> Date: Thu, 24 Aug 2000 12:55:24 -0700
> From: Michael Snyder <msnyder@redhat.com>

> +/* additional s390/elf relocations */
> +  BFD_RELOC_390_8,
> +  BFD_RELOC_390_12,
> +  BFD_RELOC_390_16,
> +  BFD_RELOC_390_GOT12,
> +  BFD_RELOC_390_GOT32,
> +  BFD_RELOC_390_PLT32,
> +  BFD_RELOC_390_COPY,
> +  BFD_RELOC_390_GLOB_DAT,
> +  BFD_RELOC_390_JMP_SLOT,
> +  BFD_RELOC_390_RELATIVE,
> +  BFD_RELOC_390_GOTOFF,
> +  BFD_RELOC_390_GOTPC,
> +  BFD_RELOC_390_GOT16,
> +  BFD_RELOC_390_PC16DBL,
> +  BFD_RELOC_390_PLT16DBL,

Most of it seemed OK, although I didn't look at the s390-specific
files too closely.

I have some comments about this bit though:

- Comment.  Comments are full sentences, they start with an uppercase
  letter and end with a full stop and two spaces.  I know that
  this isn't always done, but it should be.  It also looks like best
  practise is to put a comment above each reloc saying what it does.

- What's the difference between BFD_RELOC_390_8 and BFD_RELOC_8?
  I suspect they're the same, in which case BFD_RELOC_8 should be
  used.  Likewise, although there is no BFD_RELOC_12, perhaps there
  should be.  Likewise, there's a BFD_RELOC_32_GOT_PCREL and
  BFD_RELOC_32_GOTOFF, one of which might be the same as
  BFD_RELOC_390_GOT32.

  ...and perhaps if there was a BFD_RELOC_COPY we wouldn't need each
  port to define one... although that's now starting to look like hard
  work.  At least use the relocs that are already there, though.

--
- Geoffrey Keating <geoffk@cygnus.com>