On Sat, Sep 26, 2020 at 12:32 AM Jonathan Neuschäfer
<j.neuschaefer@xxxxxxx> wrote:
> On Fri, Sep 25, 2020 at 12:29:45PM +0300, Andy Shevchenko wrote:
> > On Thu, Sep 24, 2020 at 10:26 PM Jonathan Neuschäfer
> > <j.neuschaefer@xxxxxxx> wrote:
...
> > > + dev_alert(&poweroff_restart_client->dev,
> > > + "Failed to power off (err = %d)\n", res);
> >
> > alert? This needs to be explained.
>
> I copied the dev_alert from drivers/mfd/rn5t618.c.
>
> Upon reconsideration, I'm not sure what the correct log level would be,
> but _warn seems enough, or maybe _err is better
It's up to you to decide, but crit/alert and similar has to be
justified (commit message / comment in the code).
...
> > > + /*
> > > + * NOTE: The lower half of the reset value is not sent, because sending
> > > + * it causes an error
> >
> > Why? Any root cause? Perhaps you need to send 0xffff ?
>
> Unknown, because I don't have the EC firmware for analysis. The vendor
> kernel sends 0xff00 and gets an error.
>
> Sending 0xffff doesn't help.
Maybe a slightly elaborated comment that it's copied from the vendor
kernel (which is?).
> > > + */
...
> > > + dev_info(ec->dev,
> > > + "Netronix embedded controller version %04x detected.\n",
> > > + version);
> >
> > This info level may confuse users if followed by an error path.
>
> Right. I suppose printing incompatible versions is still useful, so how
> about something like the following?
>
>
> /* Bail out if we encounter an unknown firmware version */
> switch (version) {
> case 0xd726: /* found in Kobo Aura */
> dev_info(ec->dev,
> "Netronix embedded controller version %04x detected.\n",
> version);
> break;
> default:
> dev_err(ec->dev,
> "Netronix embedded controller version %04x is not supported.\n",
> version);
> return -ENODEV;
> }
This is better.
...
> > > + WARN_ON(poweroff_restart_client);
> >
> > WARN_ON? All these alerts, WARNs, BUGs must be explained. Screaming to
> > the user is not good if it wasn't justified.
>
> poweroff_restart_client being already set is not a situation that should
> happen (and would indicate a bug in this driver, AFAICT), but I guess
> the log message could be better in that case...
As per above.
--
With Best Regards,
Andy Shevchenko
