Re: [edk2-devel] [PATCH V2 2/3] MdePkg/Base.h: Introduce various alignment-related macros

["Marvin Häuser" <mhaeuser@posteo.de>]

Ping? :)

On 16.08.21 15:10, Marvin Häuser wrote:
> Hey Ray,
>
> On 16/08/2021 11:42, Ni, Ray wrote:
>> Marvin,
>> So lucky to have you in the edk2 project looking into these 
>> fundamentals!
>
> Thank you. :)
>
>> +  #define ALIGNOF(TYPE) OFFSET_OF (struct { CHAR8 C; TYPE A; }, A)
>>
>> 1. Does struct{} inside a macro conform to C standard? How is the 
>> compatibility with different compilers?
>
> This should work, yes. The C standard defines offsetof as such:
>
> "The macros are [...]
>
>         offsetof(type, member-designator)
>
> which expands to an integer constant expression that has type size_t, 
> the value of
> which is the offset in bytes, to the structure member (designated by 
> member-designator),
> from the beginning of its structure (designated by type). The type and 
> member designator
> shall be such that given
>
>         static type t;
>
> then the expression &(t.member-designator) evaluates to an address 
> constant. [...]" [1]
>
> If we plug in t:
>
>         static struct { CHAR8 C; TYPE A; } t;
>
> we get a valid static storage duration variable declaration that 
> satisfies the the last condition because:
>
> "An address constant is [...], a pointer to an lvalue designating an 
> object of static
> storage duration, or [...]" [2]
>
> It worked with all compilers I tinkered with at https://godbolt.org/
> I sadly do not have access to any of the compilers where this may be 
> used effectively (RVCT, EBC).
>
>> +#define IS_POW2(Value)  ((Value) != 0U && ((Value) & ((Value) - 1U)) ==
>> +0U)
>>
>> 2. Good to me. I learned this trick when implementing the MtrrLib.
>>
>> +#define ALIGN_VALUE_ADDEND(Value, Alignment)  (((Alignment) - (Value))
>> +& ((Alignment) - 1U))
>>
>> 3. Is any other open source project using the same macro for the addend?
>> This is actually a general question to all new macros.
>> I would like the macros look familiar to developers from other open 
>> source projects.
>
> Good question, I never really saw it. I only came up with it because 
> for the new PE loader, we may align the PE memory within an 
> underaligned buffer, and for that we need the addend. I initially used 
> to align up and then subtract, but I saw this could be simplified with 
> ALIGN_VALUE_ADDEND, which was used in ALIGN_VALUE anyway. If you have 
> a better name, I'll change it.
>
> Best regards,
> Marvin
>
>
> [1] ISO/IEC 9899:2011, 7.19, 3.
>
> [2] ISO/IEC 9899:2011, 6.6, 9.
>
>>
>>
>> 
>>
>>
>