Re: [PATCH 10/10] x86: Delete mach_apic.h

[Andrew Cooper <andrew.cooper3@xxxxxxxxxx>]

On 11/11/2024 9:13 am, Jan Beulich wrote:
> On 08.11.2024 20:58, Andrew Cooper wrote:
>> All useful content has been moved elsewhere.
>>
>> Clean up the dregs, and remove the entire mach-generic include path.
> What's "dregs" here?

Oh.  "The leftovers".

Surprisingly (to me at least), it's not slang.  It is most common used
to refer to what remains in the cup when you've drunk a coffee/tea/etc.

>
>> --- a/xen/arch/x86/include/asm/mach-generic/mach_apic.h
>> +++ /dev/null
>> @@ -1,24 +0,0 @@
>> -#ifndef __ASM_MACH_APIC_H
>> -#define __ASM_MACH_APIC_H
>> -
>> -#include <asm/apic.h>
>> -#include <asm/io_apic.h>
>> -#include <asm/genapic.h>
>> -#include <asm/smp.h>
>> -
>> -static inline void enable_apic_mode(void)
>> -{
>> -    /* Not needed for modern ES7000 which boot in Virtual Wire mode. */
>> -    /*es7000_sw_apic();*/
>> -}
>> -
>> -#define apicid_to_node(apicid) ((int)apicid_to_node[(u32)apicid])
>> -
>> -extern u32 bios_cpu_apicid[];
> For these two, isn't there some Misra aspect? They're both unused, and Misra
> generally dislikes unused stuff.

Eclair won't complain about either, I don't think.

The macro will just disappear after preprocessing, whereas we've
explicitly deviated Rule 8.6 because of how we use dead-code-elimination
to reduce ifdefary.

>  Whereas ...
>
>> -static inline int multi_timer_check(int apic, int irq)
>> -{
>> -    return 0;
>> -}
> ... this and the other inline function further up clearly have their sole
> users removed here, so are fine without any further text in the description.
>
> Preferably with the description amended / clarified:
> Reviewed-by: Jan Beulich <jbeulich@xxxxxxxx>

Lemme have another go.

~Andrew