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

[Rainer Orth]

David Miller <davem@davemloft.net> writes:

> From: Eric Botcazou <ebotcazou@adacore.com>
> Date: Sat, 3 Mar 2012 00:00:07 +0100
>
>>> I see no value in blindly allowing relocations to overflow when we
>>> know that when it occurs something is very wrong.  Failing and
>>> emitting an error is the right thing for 'ld' to do.
>> 
>> The consensus for years has been to silently accept this, so turning
>> the tables in 2012 doesn't make much sense I think.  It's just debug
>> info after all.
>
> Fair enough but I think this is a quality of implementation issue and
> that all overflows like this are hard errors because it's losing
> information.
>
> Anyways, Rainer can check in his change if he wants, I still disagree
> but I won't stand in the way.

Thanks.  I've applied the patch to mainline, but will probably keep it
in my local tree for 2.22.

I think it's pointless for ld to error out for one of 3 relocs: this
just obscures real problems in gcc testsuite results.  An alternative
would be to skip 64-bit -gstabs* testing, of course.

	Rainer

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