Re: [PATCH 2/6] soc: qcom: llcc: Refactor llcc driver to support multiple configuration

[Mukesh Ojha]

On 8/10/2023 5:52 PM, Bryan O'Donoghue wrote:
> On 10/08/2023 07:11, Komal Bajaj wrote:
> > +    if (!cfgs || cfgs->num_config != DEF_NUM_CFG) {
> > +        ret = -EINVAL;
> > +        goto err;
> > +    }
> > +    cfg = &cfgs->llcc_config[DEF_NUM_CFG - 1];
>
> This is a bit of a redundant check.
>
> You add in the check for num_config != 1, then deref llc_config[0] but 
> in patch #4 you get an index and check that index against num_config

I would take this blame on me to suggest this..,but i was trying to
avoid the hard-coding initially done for [1], now, num_config[2]
converted to ARRAY_SIZE(), i find no harm in checking
cfgs->num_config >  DEF_NUM_CFG
since, anyways it will move to different function in #4.

[1]
https://lore.kernel.org/lkml/39b4bafd-410f-cae8-13ae-e18d751b51a2@xxxxxxxxxxx/

[2]
.num_cfgs	= 1,

-Mukesh
> I'm not seeing how at this point in your series, how num_config could be 
> anything other than 1.
>
> I'd do away with the DEF_NUM_CFG define in this code/series completely.
>
> num_config should encode all the necessary detail we need, DEF_NUM_CFG 
> just adds noise.
>
> ---
> bod