[PATCH 0/3] RISC-V: alias insn adjustments

[Jan Beulich]

On 13.09.2022 18:58, Tsukasa OI wrote:
> On 2022/09/14 1:11, Jan Beulich wrote:
>> On 13.09.2022 16:54, Tsukasa OI wrote:
>>> On 2022/09/13 21:59, Jan Beulich via Binutils wrote:
>>>> 1: re-arrange opcode table for consistent alias handling
>>>> 2: drop stray INSN_ALIAS flags
>>>> 3: add alias for SLLI.UW
>>>>
>>>> In particular the 3rd patch may want considering to be at least slightly
>>>> RFC.
>>>>
>>>> Jan
>>>>
>>>
>>> +1 to PATCH 1-2/3 except we can just add -M no-aliases to existing tests
>>> zbkb-32.d and fence-tso.d just like my submission (they don't test any
>>> aliases):
>>
>> I'd consider that wrong - disassembly should work correctly in both modes,
>> so both modes would better be tested. After all it was just one of the
>> two modes which didn't work.
>>
>> One might further consider generalization, forcing most (all?) tests to be
>> done both ways.
> 
> Good point and I'm convinced.  Then, how about using like "-noalias"
> instead of "-na"?  I like the names that can be understood without
> looking the source code and we rarely use acronyms on RISC-V GAS tests.

Personally I prefer shorter names, but if arch maintainers agree with you,
I'd surely adjust.

Jan