On 2025-03-17 17:33:58 [+0100], Petr Pavlu wrote:
> >> I'd understand changing module::refcnt from atomic_t to refcount_t, but
> >> it isn't clear to me from the above description what using rcuref_t
> >> actually gains. Could you please explain why you think it is more
> >> appropriate over refcou
On 3/10/25 22:24, Sebastian Andrzej Siewior wrote:
> On 2025-03-10 15:28:23 [+0100], Petr Pavlu wrote:
>> On 3/9/25 13:19, Sebastian Andrzej Siewior wrote:
>>> By using rcuref_t the atomic_inc_not_zero() and atomic_dec_if_positive()
>>> can disappear. By design rcuref_t does not allow decrementing
On 3/9/25 13:19, Sebastian Andrzej Siewior wrote:
> By using rcuref_t the atomic_inc_not_zero() and atomic_dec_if_positive()
> can disappear. By design rcuref_t does not allow decrementing past the
> "0" reference and increment once it has been released. It will spill
> warnings if there are more p
On 2025-03-10 15:28:23 [+0100], Petr Pavlu wrote:
> On 3/9/25 13:19, Sebastian Andrzej Siewior wrote:
> > By using rcuref_t the atomic_inc_not_zero() and atomic_dec_if_positive()
> > can disappear. By design rcuref_t does not allow decrementing past the
> > "0" reference and increment once it has b
