Re: [PATCH] staging: fbtft: replace __ATTR() with DEVICE_ATTR()

[Greg Kroah-Hartman <gregkh@linuxfoundation.org>]

On Thu, Dec 14, 2017 at 11:51:17AM +0530, Aishwarya Pant wrote:
> On Wed, Dec 13, 2017 at 12:50:00PM +0100, Greg Kroah-Hartman wrote:
> > On Mon, Dec 11, 2017 at 03:24:30PM +0530, Aishwarya Pant wrote:
> > > This is a clean-up patch which replaces the uses of raw __ATTR(...)
> > > macro with the more conventional DEVICE_ATTR(...) for defining device
> > > attributes.
> > > 
> > > Done using coccinelle-
> > > 
> > > @r@
> > > identifier foo, n;
> > > @@
> > > 
> > > struct device_attribute foo =  __ATTR(n, ...);
> > > 
> > > @script:python p@
> > > id;
> > > foo << r.foo;
> > > n << r.n;
> > > @@
> > > 
> > > // standardise the variable name to dev_attr_{name}
> > > coccinelle.id = "dev_attr_" + n
> > > 
> > > @@
> > > identifier r.foo;
> > > declarer name DEVICE_ATTR;
> > > @@
> > > 
> > > //change definition
> > > - struct device_attribute foo = __ATTR
> > > + DEVICE_ATTR
> > > 	(...);
> > > 
> > > @depends on r@
> > > identifier r.foo, p.id;
> > > @@
> > > 
> > > // replace usages everywhere
> > > - foo
> > > + id
> > > 
> > > Signed-off-by: Aishwarya Pant <aishpant@gmail.com>
> > > ---
> > >  drivers/staging/fbtft/fbtft-sysfs.c | 7 +++----
> > >  1 file changed, 3 insertions(+), 4 deletions(-)
> > > 
> > > diff --git a/drivers/staging/fbtft/fbtft-sysfs.c b/drivers/staging/fbtft/fbtft-sysfs.c
> > > index 712096659aa0..506d604d01bb 100644
> > > --- a/drivers/staging/fbtft/fbtft-sysfs.c
> > > +++ b/drivers/staging/fbtft/fbtft-sysfs.c
> > > @@ -197,19 +197,18 @@ static ssize_t show_debug(struct device *device,
> > >  	return snprintf(buf, PAGE_SIZE, "%lu\n", par->debug);
> > >  }
> > >  
> > > -static struct device_attribute debug_device_attr =
> > > -	__ATTR(debug, 0660, show_debug, store_debug);
> > > +static DEVICE_ATTR(debug, 0660, show_debug, store_debug);
> > 
> > This should be DEVICE_ATTR_RW(), right?  0660 makes no sense...
> 
> If it doesn't make sense here, I can replace it with DEVICE_ATTR_RW.
> Though, 0660 has more open permissions than 0644.

Unless there is a really good reason for keeping an "odd" sysfs file
mode, use the default permissions.  As this is just a debug flag, 0644
makes a lot more sense, don't you think?

Actually, just drop this whole attribute.  Random "turn on debugging"
config files do not even belong in sysfs.  We have a kernel-wide
infrastructure for all of that already, no single driver, or tiny
subsystem, should have their own custom way of doing this.

thanks,

greg k-h