[PATCH 2/2] ld: Don't evaluate unneeded PROVIDE expressions.

[Hans-Peter Nilsson]

> From: Andrew Burgess <andrew.burgess@embecosm.com>
> Date: Mon, 19 Jan 2015 12:06:20 +0100

> * Alan Modra <amodra@gmail.com> [2015-01-12 11:08:53 +1030]:
> 
> > > ld/testsuite/ChangeLog:
> > >
> > >     * ld-scripts/provide-4.d: New file.
> > >     * ld-scripts/provide-4-map.d: New file.
> > >     * ld-scripts/provide-4.t: New file.
> > >     * ld-scripts/provide.exp: Run the provide-4.d test.
> >
> > This part isn't OK yet.  You need to make sure new tests pass on more
> > than just Linux x86.
> 
> Alan,
> 
> Thanks for taking the time to review my patch, and I'm sorry that the
> change wasn't up to standard.
> 
> I've updated the expected results, and tested against the same set of
> builds you suggest, I'm now seeing passes on all but 3 builds, and
> these 3 failures are, I believe, not related to my changes.

I'd guess the testsuite adjustments were not double-checked that
they actually do pass on a 32-bit host, as these FAILs are
introduced by your patches, for targets cris-axis-linux-gnu,
arm-unknown-eabi, cris-axis-elf *on a 32-bit host* (so if
nothing else use 'CC=gcc -m32' on your x86_64-unknown-linux-gnu
host as my autotester does and double-check that you get a
'32-bit bfd_vma'):

Running /tmp/hpautotest-binutils/bsrc/src/ld/testsuite/ld-scripts/provide.exp ...
FAIL: ld-scripts/provide-4 (map file check)
FAIL: ld-scripts/provide-5 (map file check)

(Please fix.)

brgds, H-P