Re: [PATCHv3 02/16] drivers: thermal: introduce device tree parser

[Joe Perches <joe@perches.com>]

On Wed, 2013-09-18 at 15:44 -0400, Eduardo Valentin wrote:
> Hello Joe,

Hello Eduardo.

>  18-09-2013 15:11, Joe Perches wrote:
> > On Wed, 2013-09-18 at 15:02 -0400, Eduardo Valentin wrote:
> >> This patch introduces a device tree bindings for
> >> describing the hardware thermal behavior and limits.
> >> Also a parser to read and interpret the data and feed
> >> it in the thermal framework is presented.
> > 
> > trivial notes:
[]
> > If readings are within some non zero noise level,
> > perhaps stable should be returned.
> 
> Yes, there should be some sort of threshold for temperature trend. But I
> am not sure this is the right place to implement this. This type of
> feature is in my TODO list, but I am planing to get it done within the
> core code of the thermal framework.

No worries, one day...  Just wanted to bring it to attention.

> >> +static struct __thermal_zone *
> >> +thermal_of_build_thermal_zone(struct device_node *np)
> >> +{
[]
> >> +	tz = kzalloc(sizeof(*tz), GFP_KERNEL);
> >> +	if (!tz) {
> >> +		pr_err("not enough memory for thermal of zone\n");
> > 
> > Unnecessary OOM message.
> > All allocs without GFP_NOWARN get a dump_stack()

[]

> Hmmm.. I am pretty sure you have a good point. But to me seams to be
> still a common practice to have drivers outputing error messages when
> allocation fails. A simple git grep -A 4 kzalloc for instance, shows
> that there are still quite a considerable amount of occurrences of such
> practice.

True though a considerable number have been removed.
I think these should all be deleted eventually and
there's no good reason to add more.