Re: [RFC PATCH 12/16] clk: hifiberry-dacpro: add ACPI support

[Stephen Boyd <sboyd@xxxxxxxxxx>]

Quoting Pierre-Louis Bossart (2020-04-22 02:54:38)
> 
> 
> On 4/22/20 4:32 AM, Stephen Boyd wrote:
> > Quoting Pierre-Louis Bossart (2020-04-09 12:58:37)
> >> On ACPI platforms the of_ functions are irrelevant, conditionally
> >> compile them out and add devm_clk_hw_register_clkdev() call instead.
> >>
> >> Signed-off-by: Pierre-Louis Bossart <pierre-louis.bossart@xxxxxxxxxxxxxxx>
> >> ---
> >>   drivers/clk/clk-hifiberry-dacpro.c | 7 +++++++
> >>   1 file changed, 7 insertions(+)
> >>
> >> diff --git a/drivers/clk/clk-hifiberry-dacpro.c b/drivers/clk/clk-hifiberry-dacpro.c
> >> index bf0616c959da..d01a90fed51b 100644
> >> --- a/drivers/clk/clk-hifiberry-dacpro.c
> >> +++ b/drivers/clk/clk-hifiberry-dacpro.c
> >> @@ -114,15 +114,22 @@ static int clk_hifiberry_dacpro_probe(struct platform_device *pdev)
> >>                  return ret;
> >>          }
> >>   
> >> +#ifndef CONFIG_ACPI
> > 
> > Use if (!IS_ENABLED(CONFIG_ACPI)) instead?
> 
> git grep CONFIG_ACPI shows most of the kernel code uses #if(n)def 
> CONFIG_ACPI. It's equivalent, it's a boolean.

It's not equivalent. It is a pre-processor directive vs. an if statement
that evaluates to 0 or 1 and lets the compiler see both sides of the
code to check types.