PATCH: PR gas/10704: as segfault in memory lookup intel syntax

[H.J. Lu]

On Tue, Oct 6, 2009 at 9:45 PM, Alan Modra <amodra@bigpond.net.au> wrote:
> On Tue, Oct 06, 2009 at 08:44:42PM -0700, H.J. Lu wrote:
>> On Tue, Oct 6, 2009 at 1:40 AM, Jan Beulich <JBeulich@novell.com> wrote:
>> > Basically, all
>> > uses outside of parse_operands() appear to need immediate
>> > resolution.
>
> I came to the same conclusion.
>
>> This patch fixes the testcase. Does it look OK?
>
> Not without taking care of Jan's comment above.  You will need
> something like your patch though.
>
>> +  for (i = 0; i < num_operands; i++)
>
> Fails for IA64_OPND_SOF and IA64_OPND_SOF since operand 0 isn't
> yet parsed.
>
> Here's a merge of your patch with what I already had for the other
> easy to fix uses of parse_operand.  I stopped when I hit the hard part
> that you've tackled..
>
>        * config/tc-is64.c (parse_operand): Use expression rather than
>        expression_and_evalute.
>        (parse_operand_and_eval): New function.  Replace all uses of
>        parse_operand outside of parse_operands with this function.
>        (parse_operans_maybe_eval): New function.  Replace uses of
>        parse_operand in parse_operands, except for the dummy, with
>        this function.
>

Can you check it in if it passes all testcases? I can test the new binutils
on gcc, glibc and kernel tomorrow.

Thanks.


-- 
H.J.