On 28.11.2023 14:16, Juergen Gross wrote:
> On 24.11.23 19:59, Alejandro Vallejo wrote:
>> On 20/11/2023 11:38, Juergen Gross wrote:
>>> --- a/xen/include/xen/spinlock.h
>>> +++ b/xen/include/xen/spinlock.h
>> > [snip]
>>> @@ -182,8 +191,10 @@ typedef struct spinlock {
>>> #endif
>>> } spinlock_t;
>>> +typedef spinlock_t rspinlock_t;
>>> #define spin_lock_init(l) (*(l) = (spinlock_t)SPIN_LOCK_UNLOCKED)
>>> +#define rspin_lock_init(l) (*(l) = (rspinlock_t)SPIN_LOCK_UNLOCKED)
>>
>> nit: Both variants of [r]spin_lock_init(l) could be inline functions
>> rather than macros.
>
> I was following the spin_lock_init() example, and I guess that is following
> the Linux kernel example.
>
> I don't mind either way, but maybe other maintainers have a preference?
While switching to inline functions would be nice, the new item should
imo be in line with the existing one (i.e. be a macro as long as that
other one also is a macro). And converting what has been there shouldn't
be a requirement for this series to land.
Jan
