This is the mail archive of the
binutils@sourceware.org
mailing list for the binutils project.
Re: [PATCH] gas: require an operand to .startof.()/.sizeof.()
- From: "Jan Beulich" <JBeulich at suse dot com>
- To: "Nick Clifton" <nickc at redhat dot com>
- Cc: <binutils at sourceware dot org>
- Date: Wed, 22 Feb 2017 10:12:25 -0700
- Subject: Re: [PATCH] gas: require an operand to .startof.()/.sizeof.()
- Authentication-results: sourceware.org; auth=none
- References: <87o9xu2nmq.fsf@redhat.com>
>>> On 22.02.17 at 17:31, <nickc@redhat.com> wrote:
>> gas/
>> 2017-02-21 Jan Beulich <jbeulich@suse.com>
>>
>> * expr.c (operand): Handle missing operand to .startof.() and
>> .sizeof.().
>> * testsuite/gas/all/err-sizeof.s: New.
>
> The cris architecture is not the only one affected by this new test.
> The following are also affected:
Hmm, I had thought powerpc would be a very strange exception,
rather than the rule.
> Given the large number of targets affected, maybe the test itself needs
> to be reconsidered ? Or maybe split into two tests, one general one,
> and one for targets which do support the difference of two symbols in an
> expression ?
Thing is - the test doesn't use the difference of symbols. There
ought to be a syntax error long before that expression would
need evaluating. I'll need to go look again, but perhaps best is
going to be to remove the problematic line altogether?
Jan