Messages in this thread | | | Date | Fri, 17 May 2024 11:32:45 -0700 | Subject | Re: [PATCH 1/7] hwspinlock: Introduce refcount | From | Chris Lew <> |
| |
On 5/17/2024 1:58 AM, Bryan O'Donoghue wrote: > On 17/05/2024 00:58, Chris Lew wrote: >> + unsigned int refcnt; > > Why int and not refcount_t ? > > Have you an argument for or against use of one over another ? >
I wanted to avoid the warning if you try to do a refcount_inc on 0. In this case, 0 means the the hwlock is unused but the hwlock should persist while waiting for another request. It seemed like refcount_t expected the associated object to be released once the count hit 0.
Also the count here is serialized by hwspinlock_tree_lock so the need for the atomicity provided by refcount_t was unneeded. Using unsigned int here seemed simpler.
> --- > bod
| |