Re: [PATCH v9 4/5] counter: Add Renesas RZ/G2L MTU3a counter driver

[William Breathitt Gray <william.gray@linaro.org>]

[-- Attachment #1: Type: text/plain, Size: 1114 bytes --]

On Fri, Dec 16, 2022 at 05:00:30PM +0000, Biju Das wrote:
> > Hello Biju,
> > 
> > Do you need to take the ch->lock before checking ch->is_busy to ensure it
> > does not change?
> 
> priv->count_is_enabled[count->id]-> true means channel is held by counter.
> So pwm won't be able to change the state ch->is_busy.
> 
> priv->count_is_enabled[count->id]-> false and if there is contention for ch->busy
> whoever is first calling rz_mtu3_request_channel() will get the channel.
> among pwm_request and counter_enable.
> 
> So I think it is safe here. Please correct me if I am missing something.
> 
> static inline bool rz_mtu3_request_channel(struct rz_mtu3_channel *ch)
> {
> 	bool is_idle;
> 
> 	mutex_lock(&ch->lock);
> 	is_idle = !ch->is_busy;
> 	if (is_idle)
> 		ch->is_busy = true;
> 	mutex_unlock(&ch->lock);
> 
> 	return is_idle;
> }

Okay seems safe then. If the respective count_is_enabled will only be
true when the respective channel is held by the counter, is there a need
to check ch->is_busy, or would checking count_is_enabled alone suffice?

William Breathitt Gray

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 228 bytes --]