Re: [PATCH 3/3] bus: mhi: replace snprintf in show functions with sysfs_emit

[Joe Perches <joe@perches.com>]

On Sat, 2021-10-16 at 17:07 +0200, Greg KH wrote:
> On Sat, Oct 16, 2021 at 03:24:17AM -0700, Joe Perches wrote:
> > On Sat, 2021-10-16 at 09:37 +0200, Greg KH wrote:
> > > On Sat, Oct 16, 2021 at 12:27:34PM +0530, Manivannan Sadhasivam wrote:
> > > > From: Qing Wang <wangqing@vivo.com>
> > > > coccicheck complains about the use of snprintf() in sysfs show functions.
> > []
> > > > diff --git a/drivers/bus/mhi/core/init.c b/drivers/bus/mhi/core/init.c
> > []
> > > > @@ -94,7 +94,7 @@ static ssize_t serial_number_show(struct device *dev,
> > > >  	struct mhi_device *mhi_dev = to_mhi_device(dev);
> > > >  	struct mhi_controller *mhi_cntrl = mhi_dev->mhi_cntrl;
> > > >  
> > > > -	return snprintf(buf, PAGE_SIZE, "Serial Number: %u\n",
> > > > +	return sysfs_emit(buf, "Serial Number: %u\n",
> > > >  			mhi_cntrl->serial_number);
> > > 
> > > The text "Serial Number: " should not be in here, right?  It's obvious
> > > this is a serial number, that's what the documentation and file name
> > > says.  Userspace should not have to parse sysfs files.
> > 
> > sysfs is ABI right?  Parsing or not, it's what's already there.
> 
> If no tools rely on this, and we can change it, we should at least try.
> 
> We can not change ABI if something breaks.  If nothing relies on it,
> then it is fine to do so.

That's a quite bad way to think of an ABI.

All that does is tempt fate as you don't know if something already
uses it until someone complains and by that time something else may
be written to depend on the new behavior.