Re: [PATCH] misc drivers/net endianness noise

[Harvey Harrison <harvey.harrison@gmail.com>]

On Wed, 2008-05-21 at 01:55 +0100, Al Viro wrote:
> On Tue, May 20, 2008 at 05:48:00PM -0700, Harvey Harrison wrote:
> > I was contemplating an api like:
> > 
> > void put_le16(u16 val, __le16 *ptr)
> > {
> > 	*ptr = cpu_to_le16(val);
> > }
> > 
> > which would allow the above to become:
> > 
> > 	if (catc->is_f5u011)
> > 		put_be16(skb->len, tx_buf);
> > 	else
> > 		put_le16(skb->len, tx_buf);
> > 
> > Thoughts?
> 
> Do not grow API too much.  Mental savings on recognizing what's done are
> offset by need to remember more helper functions...

Fair point, I was hoping to get the following family:

get_le16
put_le16

get_unaligned_le16
put_unaligned_le16

And get rid of the le16_to_cpup api.

Then the it's explicit when alignment is an issue, and the endianess is
also explicit.  That and I think it just looks nicer than the existing
*p versions.

Just a thought,

Harvey