Re: [PATCH v2 13/22] tcg/i386: Split out constraint sets to tcg-target-con-set.h

[Richard Henderson]

On 1/19/21 5:27 AM, Peter Maydell wrote:
> On Fri, 15 Jan 2021 at 21:20, Richard Henderson
> <richard.henderson@linaro.org> wrote:
>>
>> This exports the constraint sets from tcg_target_op_def to
>> a place we will be able to manipulate more in future.
>>
>> Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
>> ---
>>  tcg/i386/tcg-target-con-set.h |  54 ++++++++++
>>  tcg/i386/tcg-target.h         |   1 +
>>  tcg/tcg.c                     | 122 +++++++++++++++++++++
>>  tcg/i386/tcg-target.c.inc     | 194 ++++++++++++----------------------
>>  4 files changed, 244 insertions(+), 127 deletions(-)
>>  create mode 100644 tcg/i386/tcg-target-con-set.h
> 
>> +#define C_O2_I1(O1, O2, I1)             C_PFX3(c_o2_i1_, O1, O2, I1),
>> +#define C_O2_I2(O1, O2, I1, I2)         C_PFX4(c_o2_i2_, O1, O2, I1, I2),
>> +#define C_O2_I3(O1, O2, I1, I2, I3)     C_PFX5(c_o2_i3_, O1, O2, I1, I2, 
>> I3),
>> +#define C_O2_I4(O1, O2, I1, I2, I3, I4) \
>> +    C_PFX6(c_o2_i4_, O1, O2, I1, I2, I3, I4),
> 
> Personally this is the kind of code where I would follow
> CODING_STYLE.rst's suggestion of "If wrapping the line at 80
> columns is obviously less readable and more awkward, prefer not
> to wrap it; better to have an 85 character line than one which
> is awkwardly wrapped.". The parallelism between the lines
> is much easier to see without the linebreak.

The maximum within this section is column 92.
Still ok with that?

> I know the macro magic should make it impossible, but maybe
> we should have an assert that we definitely have a valid
> array index here ?

Ok.

>>      case INDEX_op_andc_i32:
>>      case INDEX_op_andc_i64:
>> -        {
>> -            static const TCGTargetOpDef andc
>> -                = { .args_ct_str = { "r", "r", "rI" } };
>> -            return &andc;
>> -        }
>> -        break;
>> +        return C_O1_I2(r, 0, rI);
> 
> Old constraint was r r rI; new one is r 0 rI  ?

Oops, good catch.  And of course the error would have worked in testing.


r~