This is the mail archive of the
binutils@sourceware.org
mailing list for the binutils project.
Re: [PATCH] x86: don't omit disambiguating suffixes from "fi*"
>>> On 26.11.17 at 17:33, <hjl.tools@gmail.com> wrote:
> On Fri, Nov 24, 2017 at 6:37 AM, H.J. Lu <hjl.tools@gmail.com> wrote:
>> On Thu, Nov 23, 2017 at 7:46 AM, H.J. Lu <hjl.tools@gmail.com> wrote:
>>> On Thu, Nov 23, 2017 at 7:43 AM, Jan Beulich <JBeulich@suse.com> wrote:
>>>>>>> On 23.11.17 at 14:41, <hjl.tools@gmail.com> wrote:
>>>>> On Mon, Nov 20, 2017 at 5:07 AM, Jan Beulich <JBeulich@suse.com> wrote:
>>>>>> "fi*" typically come in two (loads/stores: three) flavors, distinguished
>>>>>> by the suffix. Don't omit the 's' one when disassembling, and for
>>>>>> consistency also add the missing 's' ones for the testsuite inputs, even
>>>>>> if this will become a requirement only later.
>>>>>
>>>>> If we want to disallow them without 's' nor 'l' in the future, we need an
>>>>> assembler option to allow them. Before we do that, please keep
>>>>> the test without 's' nor 'l' for now. We can move them to a new test
>>>>> later. OK with this change.
>>>>
>>>> Just so I understand you right: You want the testsuite *.s files to be
>>>> left alone, i.e. as offered ...
>>>>
>>>>>> ---
>>>>>> I'd be fine with dropping the testsuite input changes here, moving them
>>>>>> to the patch where they will become a requirement (still WIP).
>>>>
>>>> ... here?
>>>
>>> Yes. But it is OK to add the new 's'/'l' tests without removing existing
> ones.
>>>
>>
>> I got:
>>
>> FAIL: i386 SSE without AVX equivalent
>> FAIL: x86-64 SSE without AVX equivalent
>> FAIL: x86-64 (ILP32) SSE without AVX equivalent
>>
>> on Linux/x86-64. Please fix them.
>
> This is what I checked in.
Thanks for taking care of this, H.J.! I'm sorry for the breakage
I did introduce.
Jan