This is the mail archive of the binutils@sourceware.org mailing list for the binutils project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: Avoid 64-bit Solaris/SPARC -gstabs link failures (PR ld/12152)


David Miller <davem@davemloft.net> writes:

> From: Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
> Date: Thu, 01 Mar 2012 20:05:51 +0100
>
>> It seems this is a known issue and R_SPARC_32 relocs are already handled
>> in bfd/elfxx-sparc.c (_bfd_sparc_elf_relocate_section).  The following
>> patch handles R_SPARC_UA32 in the same way and makes all those failures
>> go away.
>
> Every time I've done significant work on the sparc BFD elf backend
> I've looked at this piece of code and scratched my head.
>
> I'd like to know why it's OK to just let the relocation overflow in
> these cases, doesn't it make the debugging information unusable and
> incorrect?

Honestly, I don't know.  When I looked at the OpenSolaris ld sources, I
couldn't even find where the R_SPARC_{, UA}32 relocs are handled ;-(

What I do find is that using Solaris ld with the exact same objects that
cause gld to error out gets me an executable.  I doubt many people are
using stabs these days (outside of the gcc and gdb testsuites, that is :-)

I'll ask the Solaris linker engineers.

	Rainer

-- 
-----------------------------------------------------------------------------
Rainer Orth, Center for Biotechnology, Bielefeld University


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]