Messages in this thread | | | Subject | Re: [PATCH v2 3/5] thermal: devfreq_cooling: add new registration functions with Energy Model | From | Lukasz Luba <> | Date | Wed, 2 Dec 2020 11:54:00 +0000 |
| |
On 12/2/20 11:49 AM, Ionela Voinescu wrote: > On Wednesday 02 Dec 2020 at 11:14:02 (+0000), Lukasz Luba wrote: >> Hi Ionela, >> >> On 12/2/20 10:24 AM, Ionela Voinescu wrote: >>> Hi Lukasz, >>> >>> On Wednesday 18 Nov 2020 at 12:03:56 (+0000), Lukasz Luba wrote: >> >> [snip] >> >>>> + struct device_node *np = NULL; >> >> [snip] >> >>>> + >>>> + if (dev->of_node) >>>> + np = of_node_get(dev->of_node); >>>> + >>> >>> Should np be checked before use? I'm not sure if it's better to do the >>> assign first and then the check on np before use. It depends on the >>> consequences of passing a NULL node pointer later on. >> >> The np is actually dev->of_node (or left NULL, as set at the begging). >> The only meaning of the line above is to increment the counter and then >> decrement if CONFIG_OF_DYNAMIC was used. >> The devfreq_cooling_register() has np = NULL and the registration can >> handle it, so we should be OK here as well. >> > > Yes, I just wanted to make sure later registration can handle np = NULL, > or whether we need to bail out. > > In this case, you can drop both ifs - for (dev->of_node) before get and > for np before put below, as of_node_get/of_node_put can handle NULL > pointers themselves.
Right. I agree, I will resend this patch with that small change. Thank you for having a look at it.
Lukasz
> > Thanks, > Ionela. >
| |