Your recent ELF-related changes cause cris-axis-elf ld regressions when assembling for a.out

[Hans-Peter Nilsson]

(Apologies to Joseph, who may see this twice because I forgot to CC binutils@)

My autotester complains about these regressions for
--target=cris-axis-elf, indicating by cvs diff -D '2007-06-29
15:06:22 UTC' -D '2007-06-29 16:29:50 UTC' that they're due to
some of your changes:

Running /h/hp/binutils/cvs_latest/src/ld/testsuite/ld-cris/cris.exp ...
FAIL: ld-cris/def2
FAIL: ld-cris/undef2
FAIL: ld-cris/warn1
FAIL: ld-cris/warn3

ld.log tells a similar story for all of them:
/n/asic_slask/hp/autobinutest/ld/../gas/as-new  --em=crisaout  -o tmpdir/dump0.o /h/hp/binutils/cvs_latest/src/ld/testsuite/ld-cris/start1.s
child killed: segmentation violation
failed with: <child killed: segmentation violation>
FAIL: ld-cris/def2

To wit, this is when generating somewhat trivial a.out objects.

brgds, H-P