Re: [PATCH] io_uring: parenthesize io_ring_head_to_buf() expansion

On 5/14/26 8:43 AM, Caleb Sander Mateos wrote:
> On Thu, May 14, 2026 at 7:25 AM Jens Axboe <[email protected]> wrote:
>>
>> On 5/14/26 8:22 AM, Caleb Sander Mateos wrote:
>>> On Thu, May 14, 2026 at 1:35?AM Yi Xie <[email protected]> wrote:
>>>>
>>>> Wrap the io_ring_head_to_buf() macro value in an extra pair of parentheses
>>>> so it is safe when composed into larger expressions, and to satisfy
>>>> scripts/checkpatch.pl.
>>>>
>>>> Signed-off-by: Yi Xie <[email protected]>
>>>> ---
>>>>  io_uring/kbuf.c | 2 +-
>>>>  1 file changed, 1 insertion(+), 1 deletion(-)
>>>>
>>>> diff --git a/io_uring/kbuf.c b/io_uring/kbuf.c
>>>> index 63061aa1cab9..dd54e43e9ddf 100644
>>>> --- a/io_uring/kbuf.c
>>>> +++ b/io_uring/kbuf.c
>>>> @@ -21,7 +21,7 @@
>>>>  #define MAX_BIDS_PER_BGID (1 << 16)
>>>>
>>>>  /* Mapped buffer ring, return io_uring_buf from head */
>>>> -#define io_ring_head_to_buf(br, head, mask)    &(br)->bufs[(head) & (mask)]
>>>> +#define io_ring_head_to_buf(br, head, mask)    (&(br)->bufs[(head) & (mask)])
>>>
>>> Is there a reason this can't just be an inline function?
>>
>> No reason at all. But also don't see a strong reason why it can't just
>> be a define. And generally I don't like cleanups like this, but this one
>> at least made sense to me.
> 
> A macro can certainly work, but as this patch shows, it's tricky to
> remember all the parentheses. An inline function also results in
> better compiler error messages since the arguments are strongly typed.
> And not applicable in this case, but if an argument is used multiple
> times, a function ensures it's only evaluated once. I would generally
> only reach for a macro when something can't be expressed as an inline
> function.

I do know the benefits of a function over a macro :-)

I just don't think it'll buy us anything in this case.

-- 
Jens Axboe