Re: [RFC PATCH v4 2/2] clk: Use devm_add in managed functions

[Marc Gonzalez <marc.w.gonzalez@free.fr>]

On 27/02/2020 14:36, Geert Uytterhoeven wrote:

> Hi Marc,
> 
> Thanks for your patch!
> 
> On Wed, Feb 26, 2020 at 4:55 PM Marc Gonzalez <marc.w.gonzalez@free.fr> wrote:
>> Using the helper produces simpler code, and smaller object size.
>> E.g. with gcc-arm-9.2-2019.12-x86_64-aarch64-none-linux-gnu:
>>
>>     text           data     bss     dec     hex filename
>> -   1708             80       0    1788     6fc drivers/clk/clk-devres.o
>> +   1524             80       0    1604     644 drivers/clk/clk-devres.o
> 
> And the size reduction could have been even more ;-)

I'll see what I can do! ;-)

I have another patch with even smaller object code, but it requires
C11 to be well-defined (memcmp the whole struct, which requires zeros
in the padding holes).


>> --- a/drivers/clk/clk-devres.c
>> +++ b/drivers/clk/clk-devres.c
> 
>> @@ -55,25 +51,17 @@ static void devm_clk_bulk_release(struct device *dev, void *res)
>>  static int __devm_clk_bulk_get(struct device *dev, int num_clks,
>>                                struct clk_bulk_data *clks, bool optional)
>>  {
>> -       struct clk_bulk_devres *devres;
>>         int ret;
>>
>> -       devres = devres_alloc(devm_clk_bulk_release,
>> -                             sizeof(*devres), GFP_KERNEL);
>> -       if (!devres)
>> -               return -ENOMEM;
>> -
>>         if (optional)
>>                 ret = clk_bulk_get_optional(dev, num_clks, clks);
>>         else
>>                 ret = clk_bulk_get(dev, num_clks, clks);
>> -       if (!ret) {
>> -               devres->clks = clks;
>> -               devres->num_clks = num_clks;
>> -               devres_add(dev, devres);
>> -       } else {
>> -               devres_free(devres);
>> -       }
>> +
>> +       if (ret)
>> +               return ret;
>> +
>> +       ret = devm_vadd(dev, my_clk_bulk_put, clk_bulk_args, num_clks, clks);
>>
>>         return ret;
> 
> return devm_vadd(...);

If you think that makes it look better, I'll make the change!


>> @@ -128,30 +109,22 @@ static int devm_clk_match(struct device *dev, void *res, void *data)
>>
>>  void devm_clk_put(struct device *dev, struct clk *clk)
>>  {
>> -       int ret;
>> -
>> -       ret = devres_release(dev, devm_clk_release, devm_clk_match, clk);
>> -
>> -       WARN_ON(ret);
>> +       WARN_ON(devres_release(dev, my_clk_put, devm_clk_match, clk));
> 
> Getting rid of "ret" is an unrelated change, which actually increases
> kernel size, as the WARN_ON() parameter is stringified for the warning
> message.

Weird... Are you sure about that? I built the preprocessed file,
and it didn't appear to be so.

#ifndef WARN_ON
#define WARN_ON(condition) ({						\
	int __ret_warn_on = !!(condition);				\
	if (unlikely(__ret_warn_on))					\
		__WARN();						\
	unlikely(__ret_warn_on);					\
})
#endif

Maybe you were thinking of i915's WARN_ON?

#define WARN_ON(x) WARN((x), "%s", "WARN_ON(" __stringify(x) ")")

Regards.