On Thu, Apr 18, 2019 at 04:45:00PM -0400, Jean-Francois Dagenais wrote:
>
> > On Apr 18, 2019, at 16:13, Guenter Roeck <linux@xxxxxxxxxxxx> wrote:
> >
> >> + if (IS_ERR(data->cooling_dev)) {
> >> + err = PTR_ERR(data->cooling_dev);
> >> + dev_err(&client->dev,
> >> + "Failed to register as cooling device (%d)\n", err);
> >
> > As mentioned in my other mail, this message adds zero value. Please drop.
>
> How does one distinguish the different failures which can occur within a probe
> function then. I know it is only useful for system integrators while debugging
> DTS for example. But I find my self always having to insert these kind of messages
> at each "return" statements of a failing probe implementation. I just think it's
> nice for developers which can use the info to quickly troubleshoot the problem.
>
> I won't fight this though. You are maintainer of this subsys so...
>
Yes, I know there are peple who like a lot of messages. I don't - it happend
to me too often that the actual important message(s) get lost in the noise.
Anyway, see below.
> >
> >> + return err;
> >
> > Sorry, I won't accept this.
>
> You mean, you won't accept v3, but a v4 without the dev_err?
>
I won't accept v3 due to the error return, which may be considered a
regression for those who don't care about using the driver with the
thermal subsystem.
If you insist on an extra message, I'll accept a dev_warn.
> Perhaps also, we should wait a v4 which would use your new devm_register... ?
Your call, but there is no guarantee that this series will ever be
accepted. The thermal maintainers may object to the new devm function.
Guenter
