Re: [PATCH 03/20] Staging: ipack/bridges/tpci200: provide newcallbacks to tpci200

[Jens Taprogge]

 On Wed, Sep 12, 2012 at 02:59:26PM +0300, Dan Carpenter wrote:
> On Wed, Sep 12, 2012 at 01:21:17PM +0200, Samuel Iglesias Gonsálvez wrote:
> > On Wed, 2012-09-12 at 14:13 +0300, Dan Carpenter wrote:
> > > On Wed, Sep 12, 2012 at 11:28:33AM +0200, Jens Taprogge wrote:
> > > > On Tue, Sep 11, 2012 at 11:47:02AM +0300, Dan Carpenter wrote:
> > > > > > +static int tpci200_get_clockrate(struct ipack_device *dev)
> > > > > > +{
> > > > > > +	struct tpci200_board *tpci200 = check_slot(dev);
> > > > > > +	__le16 __iomem *addr;
> > > > > 
> > > > > The point of the underscores in the __le16 is that you don't want to
> > > > > pollute user space headers in glibc with a bunch of kernel typedefs.
> > > > > It is not needed here.  (Or if it is, then we would need to replace
> > > > > the u16 uses as well).
> > > > 
> > > > I was under the impression that "__le16" is used to indicate the
> > > > byteorder of the pointed to memory.  As far as I can see that
> > > > information is lost when we use u16.  Am I missing something?
> > > > 
> > > 
> > > Use the no-underscore version unless it's inside a header which is
> > > exported to userspace.
> > > 
> > > 	le16 __iomem *addr;
> > > 
> > 
> > But it is not declared in linux/types, it is?
> > 
> > I have found only this typedef in a quick search:
> > 
> > http://lxr.linux.no/#linux+v3.5.3/fs/ntfs/types.h#L28
> > 
> > Should we define them in ipack.h header file or they are defined in
> > other place?
> > 
> 
> Oh crap!  You're right.  That's embarrassing.  I'm just totally
> wrong here.  Sorry for that.
> 
> regards,
> dan carpenter

Sam,

can you please add the endianess-aware versions back?  Patch attached.

One of the earlier uses was not correct.

Best Regards,
Jens