This is the mail archive of the
binutils@sourceware.org
mailing list for the binutils project.
Re: [patch][rfc] Complex Relocations gas and bfd
- From: Alan Modra <amodra at bigpond dot net dot au>
- To: Dave Brolley <brolley at redhat dot com>
- Cc: binutils at sources dot redhat dot com
- Date: Sat, 20 Jan 2007 15:54:24 +1030
- Subject: Re: [patch][rfc] Complex Relocations gas and bfd
- References: <45B1029E.1000302@redhat.com>
On Fri, Jan 19, 2007 at 12:40:46PM -0500, Dave Brolley wrote:
> >>On Fri, Jan 12, 2007 at 06:07:56PM -0500, Dave Brolley wrote:
> >>> + #define R_RELC 0xff /* Complex relocation type
> >>code (see ELFXX_R_TYPE()) */
> >>
> >>$ egrep 'RELOC.*(255|0xff)' include/elf/*.h
> >>include/elf/arm.h: RELOC_NUMBER (R_ARM_RBASE, 255)
> >>include/elf/ppc.h: RELOC_NUMBER (R_PPC_TOC16, 255)
> >>include/elf/sh.h: RELOC_NUMBER (R_SH_64_PCREL, 255)
>
> If you're trying to tell me that 0xff|255 is not a good value then I can
> change that as well.
Yes, that's exactly what I was (cryptically) saying. Existing object
files for these targets may already be using 255. The following
appear to be available according to my scan over include/elf/*.h: 142,
143, 152, 155, 156, 157, 159, 197, 198, 199, 202, 203, 204, 205, 206,
207, 208, 209, 210, 211, 212, 213, 214, 215, 217, 218, 225, 226.
Perhaps it might be better to define the reloc on a target by target
basis?
--
Alan Modra
IBM OzLabs - Linux Technology Centre