Re: [PATCH 2/3] xen/x86: Drop useless non-Kconfig CONFIG_* variables

[Andrew Cooper <andrew.cooper3@xxxxxxxxxx>]

On 22/05/2024 7:03 am, Jan Beulich wrote:
> On 21.05.2024 19:15, Andrew Cooper wrote:
>> These are all either completely unused, or do nothing useful.
>>
>> Signed-off-by: Andrew Cooper <andrew.cooper3@xxxxxxxxxx>
> Not an objection, i.e. you're fine to commit as is with Stefano's R-b, yet
> still a question:
>
>> @@ -30,11 +29,8 @@
>>  /* Intel P4 currently has largest cache line (L2 line size is 128 bytes). */
>>  #define CONFIG_X86_L1_CACHE_SHIFT 7
>>  
>> -#define CONFIG_ACPI_SRAT 1
>>  #define CONFIG_ACPI_CSTATE 1
>>  
>> -#define CONFIG_WATCHDOG 1
> I wonder if this wouldn't make sense to become a proper Kconfig setting,
> thus ...
>
>> --- a/xen/include/xen/watchdog.h
>> +++ b/xen/include/xen/watchdog.h
>> @@ -9,8 +9,6 @@
>>  
>>  #include <xen/types.h>
>>  
>> -#ifdef CONFIG_WATCHDOG
>> -
>>  /* Try to set up a watchdog. */
>>  int watchdog_setup(void);
>>  
>> @@ -23,13 +21,4 @@ void watchdog_disable(void);
>>  /* Is the watchdog currently enabled. */
>>  bool watchdog_enabled(void);
>>  
>> -#else
>> -
>> -#define watchdog_setup() ((void)0)
>> -#define watchdog_enable() ((void)0)
>> -#define watchdog_disable() ((void)0)
>> -#define watchdog_enabled() ((void)0)
>> -
>> -#endif
> ... assigning purpose to these stubs.

Actually I need to keep CONFIG_WATCHDOG to not break the build on other
architectures.

watchdog_{en,dis}able() are called from common/{debugtrace,keyhandler}.c.

I'll leave it in place for now, and we can Kconfig it up later.

~Andrew