This is the mail archive of the
binutils@sourceware.org
mailing list for the binutils project.
Re: [PATCH] ia64: equate improvements
- From: "Jan Beulich" <JBeulich at novell dot com>
- To: "James Wilson" <wilson at specifix dot com>
- Cc: <binutils at sourceware dot org>
- Date: Wed, 12 Oct 2005 09:53:12 +0200
- Subject: Re: [PATCH] ia64: equate improvements
- References: <433D7B25.76F0.0078.0@novell.com> <1129054479.23389.5.camel@aretha.corp.specifix.com>
>>> James E Wilson <wilson@specifix.com> 11.10.05 20:14:39 >>>
>On Fri, 2005-09-30 at 08:51, Jan Beulich wrote:
>> * config/tc-ia64.c (dot_reg_val): Use expression_and_evaluate.
>> (dot_pred_rel): Likewise.
>> (parse_operand): Likewise.
>> (ia64_unrecognized_line): Likewise.
>> (md_operand): Likewise.
>> * gas/ia64/forward.[sd]: New.
>> * gas/ia64/ia64.exp: Run new test.
>
>OK.
>
>You changed all expression calls in tc-ia64.c to
expression_and_evaluate
>except for one. This missed one is in ia64_parse_name, and is used
for
>expressions inside link operators such as @gprel(expr). Is there a
>reason why you left this one alone? It looks to me like it should
have
>been converted also in order to be consistent. OK if you change that
>one too.
I wasn't really decided about that one. Since it wants O_symbol, there
no strong need to try to evaluate the expression (this is only needed
for seeing whether a complex expression actually can be evaluated to
O_constant or O_register). Changing it might catch some wrong uses early
(O_symbol which in fact can be resolved to O_constant or O_register),
but I'm certain these wrong uses can get caught later and not all wrong
cases would be caught here, so not changing the call seems to result in
more consistent behavior.
>By the way, your previous ChangeLog entry has some typos. "thrid"
>instead of "third" occurs twice.
Thanks for noticing.
Jan