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

[Ian Lance Taylor]

Eric Botcazou <ebotcazou@adacore.com> writes:

>> Since 64-bit -gstabs won't actually work in practice, it seems
>> preferable that the tests not run or, if run, fail.
>
> That would have been a valid reasoning a decade ago.  But we have accepted this 
> for at least a decade so let's choose the easiest path and spend our time on 
> more interesting things than discussing the handling of STABS on SPARC 64-bit.

I'm not trying to discuss how 64-bit STABS should work.

I'm just suggesting that we should not ignore a relocation overflow
simply so that a test for 64-bit STABS will pass rather than fail.  As
far as I can see, ignoring a relocation overflow could lead to silently
generating an invalid program.  It does not seem like a good choice to
introduce the possibility of silently generating an invalid program in
order to let a test pass, particularly so given that the test is for a
feature that doesn't work anyhow.

Ian