Re: [PATCH 14/17] target/mips: Declare gen_msa/_branch() in 'translate.h'

[Philippe Mathieu-Daudé]

Hi Richard,

On 12/9/20 1:03 AM, Richard Henderson wrote:
> On 12/8/20 6:01 PM, Richard Henderson wrote:
>> On 12/8/20 5:56 PM, Richard Henderson wrote:
>>> On 12/7/20 6:36 PM, Philippe Mathieu-Daudé wrote:
>>>> Make gen_msa() and gen_msa_branch() public declarations
>>>> so we can keep calling them once extracted from the big
>>>> translate.c in the next commit.
>>>>
>>>> Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
>>>> ---
>>>>  target/mips/translate.h | 2 ++
>>>>  target/mips/translate.c | 4 ++--
>>>>  2 files changed, 4 insertions(+), 2 deletions(-)
>>>
>>> Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
>>
>> Actually, I think this should be dropped, and two other patches rearranged.
> 
> Actually, nevermind, you already get the right result in the end; there's no
> point re-rearranging.

I'm interested in looking at your idea to see if I can follow it
for the next conversions after the MSA ASE. The criteria I'm using
is (in this order):

- keep bisectability working
- keep patches trivial enough to review
- avoid moving things twice

In a previous version I tried to directly pass from

static void gen_msa(DisasContext *ctx) ...

to:

static bool trans_MSA(DisasContext *ctx, arg_MSA *a) ...

without declaring the intermediate 'void gen_msa(DisasContext)'
in "translate.h" (this patch). The result was less trivial to
review, so I went back to using an intermediate patch for
simplicity.

Is that what you were thinking about?

Thanks,

Phil.