[U-Boot] [PATCH 4/5 v4] gen: Add ACE acceleration to hash

[Kim Phillips]

On Tue, 12 Mar 2013 16:40:38 -0700
Simon Glass <sjg at chromium.org> wrote:

> Hi Kim,
> 
> On Tue, Mar 12, 2013 at 12:32 PM, Kim Phillips
> <kim.phillips at freescale.com> wrote:
> > On Mon, 11 Mar 2013 17:53:37 -0700
> > Simon Glass <sjg at chromium.org> wrote:
> >
> >> On Mon, Mar 11, 2013 at 5:44 PM, Kim Phillips <kim.phillips at freescale.com>wrote:
> >> > On Thu, 7 Mar 2013 19:11:16 -0800
> >> > Simon Glass <sjg at chromium.org> wrote:
> >> > > OK so let's look at adding the hash_register() idea. But not in this
> >> > > series. That should come later in a revision of the hash.c
> >> > > infrastructure, since it needs to adjust sha1, sha255 and crc32.
> >> >
> >> > I don't understand: why not s/ace/hw/g in common/ and include/ on
> >> > this patchseries, then move straight to the device model at some
> >> > later point?  It's a compromise, but it works fine for the time
> >> > being - other vendors can add their hash support without having to
> >> > touch common code, code size is not affected, etc.
> >>
> >> Fine with me. The effect is the same - this is just a rename. Should not be
> >> done in the ace.h file though, only in the naming of the functions called
> >> from hash.c, right?
> >
> > the ace_sha_hash_digest() declaration should be removed from
> > ace_sha.h (it's only needed by the driver, which is ok without it
> > being there).  ACE_SHA_TYPE_SHA* definitions belong in the driver
> > too - they're ACE h/w-specific.  Rename the filename ace_sha.h to
> > hw_sha.h, and all remaining ACE references contained in it.
> 
> Maybe I misunderstand - are you saying that that all the ACE symbols
> in the driver should be renamed to hw? That doesn't make a lot of
> sense to me - why is this needed?

no, only the entry points to the driver should be renamed - they are
currently called ace_sha{1,256}().  The include/ace_sha.h file as it
is currently can be made completely h/w agnostic.

Kim