Re: Finalizing fix committed for "[PATCH] sim: make sure to include strsignal prototype"

[Hans-Peter Nilsson <hp@bitrange.com>]

On Sun, 25 Mar 2012, Mike Frysinger wrote:
> On Sunday 25 March 2012 03:40:20 Mike Frysinger wrote:
> > the BFD64 logic in gas' expr.c is fairly light.  i'm not even sure it makes
> > sense to key off of BFD64 since this code doesn't use bfd for number
> > processing.  it's just deciding whether to automatically upgrade bignum's
> > into int64's.  i'll post a patch to just delete the BFD64 logic there and
> > see what people have to say.
>
> ok, digging a bit more, that isn't going to work.  the gas parsing logic is
> tied too closely to the bfd_vma type which is directly tied to the
> aforementioned configure flag.
>
> since the fact.s test already tests both values (full & truncated), and in
> reality the compiled code is the same for both (since gas does the
> truncation), i've commented out the test that is larger than 32bits.

For cris-elf there's a gas test I added (a loong time ago,
before 1.1) that I never pursued, one with IIRC a temporarily
overflowing expression; one of the temporaries truncated on
32-bit-bfd.  I xfailed the test (gas/cris/shexpr-1) which makes
it xpass on 64-bit-bfd. :/  Mea culpa.  Maybe the same bug maybe
slightly different, but the same 32-vs-64-bit-bfd basic reason.

It'd be nice to have gas PR's for whatever you can blame on gas
from the bfin sim tests.

brgds, H-P