[PATCH] thermal: sun8i: Be loud when probe fails

Ondřej Jirman megous at megous.com
Wed Jul 8 09:44:41 EDT 2020


On Wed, Jul 08, 2020 at 03:36:54PM +0200, Maxime Ripard wrote:
> On Wed, Jul 08, 2020 at 03:29:24PM +0200, Ondřej Jirman wrote:
> > Hello Maxime,
> > 
> > On Wed, Jul 08, 2020 at 02:25:42PM +0200, Maxime Ripard wrote:
> > > Hi,
> > > 
> > > On Wed, Jul 08, 2020 at 12:55:27PM +0200, Ondrej Jirman wrote:
> > > > I noticed several mobile Linux distributions failing to enable the
> > > > thermal regulation correctly, because the kernel is silent
> > > > when thermal driver fails to probe. Add enough error reporting
> > > > to debug issues and warn users in case thermal sensor is failing
> > > > to probe.
> > > > 
> > > > Failing to notify users means, that SoC can easily overheat under
> > > > load.
> > > > 
> > > > Signed-off-by: Ondrej Jirman <megous at megous.com>
> > > > ---
> > > >  drivers/thermal/sun8i_thermal.c | 55 ++++++++++++++++++++++++++-------
> > > >  1 file changed, 43 insertions(+), 12 deletions(-)
> > > > 
> > > > diff --git a/drivers/thermal/sun8i_thermal.c b/drivers/thermal/sun8i_thermal.c
> > > > index 74d73be16496..9065e79ae743 100644
> > > > --- a/drivers/thermal/sun8i_thermal.c
> > > > +++ b/drivers/thermal/sun8i_thermal.c
> > > > @@ -287,8 +287,12 @@ static int sun8i_ths_calibrate(struct ths_device *tmdev)
> > > >  
> > > >  	calcell = devm_nvmem_cell_get(dev, "calibration");
> > > >  	if (IS_ERR(calcell)) {
> > > > +		dev_err(dev, "Failed to get calibration nvmem cell (%ld)\n",
> > > > +			PTR_ERR(calcell));
> > > > +
> > > >  		if (PTR_ERR(calcell) == -EPROBE_DEFER)
> > > >  			return -EPROBE_DEFER;
> > > > +
> > > 
> > > The rest of the patch makes sense, but we should probably put the error
> > > message after the EPROBE_DEFER return so that we don't print any extra
> > > noise that isn't necessarily useful
> > 
> > I thought about that, but in this case this would have helped, see my other
> > e-mail. Though lack of "probe success" message may be enough for me, to
> > debug the issue, I'm not sure the user will notice that a message is missing, while
> > he'll surely notice if there's a flood of repeated EPROBE_DEFER messages.
> 
> Yeah, but on the other hand, we regularly have people that come up and
> ask if a "legitimate" EPROBE_DEFER error message (as in: the driver
> wasn't there on the first attempt but was there on the second) is a
> cause of concern or not.

That's why I also added a success message, to distinguish this case. 

> > And people run several distros for 3-4 months without anyone noticing any
> > issues and that thermal regulation doesn't work. So it seems that lack of a
> > success message is not enough.
> 
> I understand what the issue is, but do you really expect phone users to
> monitor the kernel logs every time they boot their phone to see if the
> thermal throttling is enabled?

Not phone users, but people making their own kernels/distributions. Those people
monitor dmesg, and out of 4 distros or more nobody noticed there was an issue
(despite the complaints of overheating by their users).

So I thought some warning may be in order, so that distro people more easily
notice they have misconfigured the kernel or sometging.

End users really don't care about dmesg.

regards,
	o.

> If anything, it looks like a distro problem, and the notification /
> policy to deal with that should be implemented in userspace.
> 
> > Other solution may be to select CONFIG_NVMEM_SUNXI_SID if this driver
> > is enabled. That may get rid of this error scenario of waiting infinitely
> > for calibration data with EPROBE_DEFER. And other potential EPROBE_DEFER sources
> > will probably be quite visible even without this driver telling the user.
> > So this message may not be necessary in that case.
> 
> That would only partially solve your issue. If the nvmem driver doesn't
> load for some reason, you would end up in a similar situation.
> 
> Maxime



More information about the linux-arm-kernel mailing list