Re: [PATCH 3/6] x86/MPX: suppress base/index swapping in Intel mode for bndmk, bndldx, and bndstx

["H.J. Lu" <hjl.tools@gmail.com>]

On Tue, Oct 8, 2013 at 9:00 AM, Jan Beulich <JBeulich@suse.com> wrote:
>>>> On 08.10.13 at 17:33, "H.J. Lu" <hjl.tools@gmail.com> wrote:
>> On Tue, Oct 8, 2013 at 8:22 AM, Jan Beulich <JBeulich@suse.com> wrote:
>>>>>> On 08.10.13 at 17:16, "H.J. Lu" <hjl.tools@gmail.com> wrote:
>>>> On Tue, Oct 8, 2013 at 7:42 AM, Jan Beulich <JBeulich@suse.com> wrote:
>>>>> bndmk, bndldx, and bndstx assign special meaning to base and index
>>>>> registers, and hence silently swapping the registers should be
>>>>> suppressed.
>>>>>
>>>>> gas/
>>>>> 2013-10-08  Jan Beulich <jbeulich@suse.com>
>>>>>
>>>>>         * tc-i386.c (i386_intel_simplify_register): Suppress base/index
>>>>>         swapping for bndmk, bndldx, and bndstx.
>>>>>
>>>>> --- 2013-10-07/gas/config/tc-i386-intel.c
>>>>> +++ 2013-10-07/gas/config/tc-i386-intel.c
>>>>> @@ -291,6 +291,8 @@ i386_intel_simplify_register (expression
>>>>>    else if (!intel_state.index)
>>>>>      {
>>>>>        if (intel_state.in_scale
>>>>> +          || current_templates->start->base_opcode == 0xf30f1b /* bndmk */
>>>>> +          || (current_templates->start->base_opcode & ~1) == 0x0f1a /*
>> bnd{ld,st}x */
>>>>>           || i386_regtab[reg_num].reg_type.bitfield.baseindex)
>>>>>         intel_state.index = i386_regtab + reg_num;
>>>>>        else
>>>>>
>>>>>
>>>>>
>>>>
>>>> We need a testcase for this.
>>>
>>> Which is included in patch 1!
>>
>> Does that mean I got "make check" failure with patch 1 applied?
>> A patch shouldn't introduce a "make check" failure and a testcase
>> should be together with the change.
>
> Both 0/6 and 1/6 mentioned this quite clearly. And no, with how
> badly the MPX tests were written (referring to other badly written
> ones would at best be a lame excuse), I don't think it's appropriate
> for you to ask that I now go back and disentangle all the various
> changes to those test cases. You shouldn't have approved/
> committed such non-extensible test cases in the first place.
>

I prefer a testcase together with the corresponding change,
instead of a jumbo testcase patch.  I also don't agree every
MPX change you proposed.  If it makes it easier to write
testcases, you can use a separate testcase file for each
change.

Thanks.


-- 
H.J.