Re: [PATCH v3 2/6] xen/riscv: introduce asm/types.h header file

[Jan Beulich <jbeulich@xxxxxxxx>]

On 11.01.2023 11:22, Xenia Ragiadakou wrote:
> 
> On 1/11/23 11:08, Jan Beulich wrote:
>> On 11.01.2023 09:47, Oleksii wrote:
>>> On Tue, 2023-01-10 at 17:58 +0100, Jan Beulich wrote:
>>>> On 10.01.2023 16:17, Oleksii Kurochko wrote:
>>>>> Signed-off-by: Oleksii Kurochko <oleksii.kurochko@xxxxxxxxx>
>>>>> ---
>>>>> Changes in V3:
>>>>>      - Nothing changed
>>>>> ---
>>>>> Changes in V2:
>>>>>      - Remove unneeded now types from <asm/types.h>
>>>>
>>>> I guess you went a little too far: Types used in common code, even if
>>>> you
>>> It looks then I didn't understand which one of types are needed.
>>>
>>> In "[PATCH v1 2/8] xen/riscv: introduce asm/types.h header file" all
>>> types were introduced as most of them are used in <xen/types.h>:
>>> __{u|s}{8|16|32|64}. Thereby it looks like the following types in
>>> <asm/types.h> should be present from the start:
>>>    typedef __signed__ char __s8;
>>>    typedef unsigned char __u8;
>>>
>>>    typedef __signed__ short __s16;
>>>    typedef unsigned short __u16;
>>>
>>>    typedef __signed__ int __s32;
>>>    typedef unsigned int __u32;
>>>
>>>    #if defined(__GNUC__) && !defined(__STRICT_ANSI__)
>>>    #if defined(CONFIG_RISCV_32)
>>>      typedef __signed__ long long __s64;
>>>      typedef unsigned long long __u64;
>>>    #elif defined (CONFIG_RISCV_64)
>>>      typedef __signed__ long __s64;
>>>      typedef unsigned long __u64;
>>>    #endif
>>>    #endif
>>>
>>>   Then it turns out that there is no any sense in:
>>>    typedef signed char s8;
>>>    typedef unsigned char u8;
>>>
>>>    typedef signed short s16;
>>>    typedef unsigned short u16;
>>>
>>>    typedef signed int s32;
>>>    typedef unsigned int u32;
>>>
>>>    typedef signed long long s64;
>>>    typedef unsigned long long u64;
>>>      OR
>>>    typedef signed long s64;
>>>    typedef unsigned long u64;
>>> As I understand instead of them should be used: {u|s}int<N>_t.
>>
>> Hmm, the situation is worse than I thought (recalled) it was: You're
>> right, xen/types.h actually uses __{u,s}<N>. So I'm sorry for mis-
>> guiding you; we'll need to do more cleanup first for asm/types.h to
>> become smaller.
> 
> What is the reason for not declaring __{u,s}<N> directly in xen/types.h 
> as type alias to {u,s}<N>?
> 
> IIUC __{u,s}<N> and {u,s}<N> are needed by code ported from Linux while 
> Xen code should use {u|s}int<N>_t instead, right?

Yes. Hence in the long run only Linux files should get to see __{u,s}<N>
and {u,s}<N>, perhaps from a new linux-types.h.

Jan