Re: don't bother users with unimportant messages.

[Russell King <rmk+lkml@arm.linux.org.uk>]

On Sun, Feb 19, 2006 at 03:29:16AM -0500, Dave Jones wrote:
> On Sun, Feb 19, 2006 at 08:15:23AM +0000, Russell King wrote:
>  > On Sat, Feb 18, 2006 at 08:09:10PM -0500, Dave Jones wrote:
>  > > When users see these printed to the console, they think
>  > > something is wrong.  As it's just informational and something
>  > > that only developers care about, lower the printk level.
>  > 
>  > If you're getting complaints about this, wouldn't it be better to
>  > forward them here so that they can be fixed up?
> 
> w83627hf, and probably other drivers from drivers/hwmon/

I don't see it - w83627hf registers w83627hf_driver via i2c_isa_add_driver,
which does not have any probe, remove or shutdown methods.  Moreover,
i2c_isa_add_driver doesn't set any of these methods.  So how can:

        if ((drv->bus->probe && drv->probe) ||
            (drv->bus->remove && drv->remove) ||
            (drv->bus->shutdown && drv->shutdown)) {

be true?  Local distro modifications maybe?

> 
>  > The thing about this particular message is that if you see it, the
>  > driver will _not_ work properly, so it's actually more than a
>  > "debugging" message.  It's telling you why driver FOO doesn't work.
> 
> I'm pretty certain this driver _was_ working fine before this change.

If driver->bus->{probe,remove,shutdown} are set, the driver model will
_not_ call driver->{probe,remove,shutdown} itself.  Hence, if both
contain pointers then it's highly likely the driver will not work.
Not reporting these instances will lead to users complaining "my
driver doesn't work" and then a headache for someone to work out
why - since nothing obvious will have changed.

-- 
Russell King
 Linux kernel    2.6 ARM Linux   - http://www.arm.linux.org.uk/
 maintainer of:  2.6 Serial core