Re: [PATCH 2/8] i2c: i2c-qcom-geni: Signify successful driver probe

[Lee Jones]

On Wed, 05 Jun 2019, Johan Hovold wrote:

> On Wed, Jun 05, 2019 at 08:16:25AM +0100, Lee Jones wrote:
> > On Tue, 04 Jun 2019, Bjorn Andersson wrote:
> > 
> > > On Tue 04 Jun 03:44 PDT 2019, Lee Jones wrote:
> > > 
> > > > The Qualcomm Geni I2C driver currently probes silently which can be
> > > > confusing when debugging potential issues.  Add a low level (INFO)
> > > > print when each I2C controller is successfully initially set-up.
> > > > 
> > > > Signed-off-by: Lee Jones <lee.jones@xxxxxxxxxx>
> > > > ---
> > > >  drivers/i2c/busses/i2c-qcom-geni.c | 2 ++
> > > >  1 file changed, 2 insertions(+)
> > > > 
> > > > diff --git a/drivers/i2c/busses/i2c-qcom-geni.c b/drivers/i2c/busses/i2c-qcom-geni.c
> > > > index 0fa93b448e8d..e27466d77767 100644
> > > > --- a/drivers/i2c/busses/i2c-qcom-geni.c
> > > > +++ b/drivers/i2c/busses/i2c-qcom-geni.c
> > > > @@ -598,6 +598,8 @@ static int geni_i2c_probe(struct platform_device *pdev)
> > > >  		return ret;
> > > >  	}
> > > >  
> > > > +	dev_info(&pdev->dev, "Geni-I2C adaptor successfully added\n");
> > > > +
> > > 
> > > I would prefer that we do not add such prints, as it would be to accept
> > > the downstream behaviour of spamming the log to the point where no one
> > > will ever look through it.
> > 
> > We should be able to find a middle ground.  Spamming the log with all
> > sorts of device specific information/debug is obviously not
> > constructive, but a single liner to advertise that an important
> > device/controller has been successfully initialised is more helpful
> > than it is hinderous.
> > 
> > This print was added due to the silent initialisation costing me
> > several hours of debugging ACPI device/driver code (albeit learning a
> > lot about ACPI as I go) just to find out that it was already doing the
> > right thing - just very quietly.
> 
> No, we don't add noise like this to the logs just because it may be
> useful while debugging. Even one-liners add up.

One line per device is should not cause an issue.

Problems occur when developers try to print all kinds of device
specifics to the boot log.  A simple, single line for such an
important device/controller has more benefits than drawbacks.

> There are plenty of options for debugging already ranging from adding a
> temporary dev_info() to the probe function in question to using dynamic
> debugging to have driver core log every successful probe.

This is what I ended up doing.  It was time consuming to parse though
a log of that size when you have no paging or keyboard.

> And in this case you say the driver was in fact already bound; that can
> easily be verified through sysfs too in case things aren't behaving the
> way you expect.

Not in a non-booting system with no keyboard you can't. ;)

-- 
Lee Jones [æçæ]
Linaro Services Technical Lead
Linaro.org â Open source software for ARM SoCs
Follow Linaro: Facebook | Twitter | Blog