Re: [PATCH 2/2] byteorder: eliminate pointer bytorder api

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

On Tue, 2008-05-20 at 14:17 -0700, David Miller wrote:
> From: Harvey Harrison <harvey.harrison@gmail.com>
> Date: Tue, 20 May 2008 12:24:09 -0700
> 
> > Not a great api, should be using cpu_to_etc and deref the pointer yourself.
> > cpu_to_le16p
> > cpu_to_le32p
> > cpu_to_le64p
> > cpu_to_be16p
> > cpu_to_be32p
> > cpu_to_be64p
> > 
> > Replaced by the aligned get_/put_ helpers
> > le16_to_cpup
> > le32_to_cpup
> > le64_to_cpup
> > be16_to_cpup
> > be32_to_cpup
> > be64_to_cpup
> > 
> > Also add const to the get/put helpers and use them in the access_ok case for
> > unaligned access.
> > 
> > Signed-off-by: Harvey Harrison <harvey.harrison@gmail.com>
> 
> But what you're doing in the first patch is killing performance for some
> cases.
> 
> The reason we use the cpu_to_*p() interfaces is to get the other-endian
> load and store instructions some processors have.
> 
> What you're doing is undoing all of that work we've done to take
> advantage of such things.

Obviously I missed that part, my apologies.  Would it be acceptable if,
taking the possibly arch-specific parts, moved the [endian]_to_cpup
name over to get_[endian]

ie:

le16_to_cpup -> get_le16

I'd leave cpu_to_le16p alone (it's not used that much anyway).

To make it similar to the get_unaligned_le16 helpers?

Thanks for having a look,

Harvey