More problems with MIPS gas relocations

[Richard Sandiford]

Thanks Eric, comiitted.

cgd@broadcom.com writes:

> (I've not submitted a patch for mine, at this point, and probably
> won't for a while.  It's a set of options ~nobody is using, and i'm
> Really Afraid of changing that chunk of code...)

FWIW, I hacked a sandbox gas so that it aborts in that OBJ_ELF code if value
!= 0 && !fixP->fx_pcrel.  Just to see how often the condition triggers on
non empic code.  I've been using it for a while with no problems.  MIPS
GNU/Linux built fine with the abort there.  GCC showed no regressions.  And
it definitely fixes the test case....

Definitely seems worth submitting.

Richard