This is the mail archive of the
binutils@sources.redhat.com
mailing list for the binutils project.
Re: MN10300 gas: reject invalid uses of `epsw'
On Feb 12, 2002, Eric Christopher <echristo@redhat.com> wrote:
> Should add a comment to that effect then.
Err... I had already added a comment:
+/* We abuse the `value' field, that would be otherwise unused, to
+ encode the architecture on which (access to) the register was
+ introduced. */
> As far as not recognizing pc and epsw, I'm rather ambivalent about
> it. If it's mentioned one way or the other in the docs that's one
> thing.
It couldn't possibly be mentioned in the AM30 docs, because pc was not
directly accessible and epsw didn't even exist, so they were just
parsed as symbol names. On AM33, they started being handled as
register names. This patch causes the exact expected behavior for
each version of the architecture to be obtained.
My only thought is that we might print warnings should a register name
be used while assembling for an architecture that didn't support it,
and handle it as a symbol. Unfortunately, after some hacking around,
I couldn't find a good way to do that in a way that wouldn't get us
to report multiple warnings for the same occurrence of the symbol, so
I gave up.
May I go ahead and check the patch in as-is, and then wait for
inspiration on some way to introduce the warning? :-)
--
Alexandre Oliva Enjoy Guarana', see http://www.ic.unicamp.br/~oliva/
Red Hat GCC Developer aoliva@{cygnus.com, redhat.com}
CS PhD student at IC-Unicamp oliva@{lsd.ic.unicamp.br, gnu.org}
Free Software Evangelist Professional serial bug killer